Polycyclic Aromatic Hydrocarbons and Astrophysics
NATO ASI Series
Advanced Science Institutes Series
A series presentmg the results of activities sponsored by the NATO Science Committee,
which aims at the dissemination of advanced scientific and technological knowledge,
with a view to strengthening links between scientific communities.
The series is published by an international board of publishers in conjunction with the
NATO Scientific Affairs Division
A Life Sciences
B Physics
Plenum Publishing Corporation
London and New York
C Mathematical
and Physical Sciences
D. Reidel Publishing Company
Dordrecht, Boston, Lancaster and Tokyo
D Behavioural and Social Sciences
E Engineering and
Materials Sciences
Martinus Nijhoff Publishers
The Hague, Boston and Lancaster
F Computer and Systems Sciences
G Ecological Sciences
Springer-Verlag
Berlin, Heidelberg, New York and Tokyo
Series C: Mathematical and Physical Sciences Vol. 191
Polycyclic Aromatic Hydrocarbons
and Astrophysics
edited by
A. Leger
and
L. d'Hendecourt
Groupe de Physique des Solides de l'Ecole Normale Superieure,
Universite Paris VII, France
and
N. Boccara
C. E. N., Saclay, France
D. Reidel Publishing Company
Dordrecht / Boston / Lancaster / Tokyo
Published in cooperation with NATO Scientific Affairs Division
Proceedings of the NATO Advanced Research and CNRS Workshop on
Polycyclic Aromatic Hydrocarbons and Astrophysics
Les Houches, France
February 17-22, 1986
Library of Congress CataloginQ in Publication Data
Polycyclic aromatic hydrocarbons and astrophysics.
(NATO ASI series. Series C, Mathematical and physical sciences; v. 191)
"Published in cooperation with NATO Scientific Affairs Division."
"Proceedings of the NATO Advanced Research and CNRS Workshop on Polycyclic
Aromatic Hydrocarbons and Astrophysics, Les Houches, France, February 17 - 22,
1986"-Verso t.p.
Includes index.
1. Interstellar matter-Congresses. 2. Polycyclic aromatic
hydrocarbons-Congresses. I. Leger, A. (Alain), 1943. II. Hendecourt, L. (Louis)
d',1953. III. Boccara, N. (Nino) IV. NATO Advanced Research and CNRS
Workshop on Polycyclic Aromatic Hydrocarbons and Astrophysics (1986: Les Houches,
Haute-Savoie, France) V. Series: NATO ASI series.
Series C, Mathematical and physical sciences; vol. 191.
QB790.P65 1986
523.1.'12
86-22022
ISBN-13: 978-94-010-8619-6
001: 10. 1007/978-94-009-4776-4
e-ISBN-13: 978-94-009-4776-4
Published by D. Reidel Publishing Company
P.O. Box 17, 3300 AA Dordrecht, Holland
Sold and distributed in the USA and Canada
by Kluwer Academic Publishers,
101 Philip Drive, Asslnippi Park, Norwell, MA 02061, USA
In all other countries, sold and distributed
by Kluwer Academic Publishers Group,
P.O. Box 322, 3300 AH Dordrecht, Holland
D. Reidel Publishing Company is a member of the Kluwer Academic Publishers Group
All Rights Reserved
© 1987 by D. Reidel Publishing Company, Dordrecht, Holland.
Softcover reprint of the hardcover 1st edition 1987
No part of the material protected by this copyright notice may be reproduced or utilized
in any form or by any means, electronic or mechanical, including photocopying', recording
or by any information storage and retrieval system, without written permission from the
copyright owner.
L E S
CENTRE
HOUCHES
DE
PHYSIQUE
operated by the
Universit~
Counc:il Hembers : M.
Scientifique et M~dicale de Grenoble
Tanche,
president, D.
Bloch, vice-president,
P. Averbuch, R. Balian, N. Boccara, R. Carr~, C. DeWitt, J.P. Hansen,
S. Haroche, J.C.
Lacoume, R. Maynard, A. Neveu, P. Papon, Y. Rocard,
R. Romestain, R. Stora, D. Thoulouze, N. Vinh Mau, G. Weill.
D:irect:or
Nino Boccara, CEN-Saclay
POLYCYCLIC AROMATIC HYDROCARBONS AND ASTROPHYSICS
February 17 - 22, 1986
Sess:ion Organ:izat::ion CommJ.'t:t:ee. : A. Leger, L. d'Hendecourt, A. Beswick,
J. Friedel, S. Leach, A. Omont, N. Panagia, J.L. Puget, W. SchmidL
TABLE OF CONTENTS
Preface
Organizing Committee and List of Participants
xi
xiii
Section 1 : INTRODUCTION
Objectives of the Workshop
by L. d'Hendecourt
Photons, Molecules and Solids in Interstellar and Circumstellar Regions: an Introduction for Non-Astronomers
by M. Jura
I
3
Section 2 : PHYSICS AND CHEMISTRY OF GRAPHITE, CARBONACEOUS
GRAINS AND CLUSTERS
On the Electronic Structure of Graphite
by P. Joyes
Various Kinds of Solid Carbon: Structure and Optical
Properties
by A. Marchand
Gas/Carbonaceous Surface Interactions
by A. Thorny, P. Wehrer
15
31
55
VUV to FIR Laboratory Observations on Submicron Amorphous
Carbon Particles
by E. Bussoletti, L. Colangeli, A. Borghesi
63
Spectroscopy of Matrix-Isolated Carbon Molecules in the
UV, VIS, and IR Spectral Range
by W. Kratschmer, K. Nachtigall
75
Remarkable Periodicities in the Mass Spectra of Carbon
Aggregates
by P. Joyes
85
Reactions of Thermal Hydrogen Atoms and Energetic
Hydrogen and Oxygen Ions with Pyrolytic Graphite
by V. Philipps, E. Vietzke
95
viii
TABLE OF CONTENTS
Section 3 : PHYSICS AND CHEMISTRY OF POLYCYCLIC AROMATIC
HYDROCARBONS (PAD's)
Photophysics, Electronic Spectroscopy and Relaxation
of Molecular Ions and Radicals with Special Reference to
Polycyclic Aromatic Hydrocarbone
by S. Leach
Fluorescence Lineshapes of Polyatomic Molecules - Spectroscopy without Eigenstates
by S. Mukamel, K. Shan, Yi Jing Yan
Structure and Chemistry of PAH's
by W. Schmidt
99
129
149
Synthesis and Spectroscopy of Tribenzo(a,g,m)Coronene,
a New, Exceptionally Stable, Fully Benzenoid PAH
by S. Obenland, W. Schmidt
165
Hot Carbon Atoms as a Potential Source for Polycyclic
Aromatic Hydrocarbons
by K. Rossler
173
Section 4 : CARBONACEOUS MATERIALS AND ASTROPHYSICS
Carbon Components of Interstellar Dust
by J.M. Greenberg, M.S. De Groot, G.P. Van der Zwet
177
Molecular Origin of the 216 nm Interstellar Hump
by G.P. Van der Zwet, M.S. De Groot, F. Baas,
J.M. Greenberg
183
Chains and Grains in Interstellar Space
by H.W. Kroto
197
Mid Infrared Excess and Ultraviolet Extinction
by P. Cox, A. Leene
207
High Spectral Resolution Observation of the 3.3 pm
Emission Band and Comparison with Laboratory-Synthesized
Quenched Carbonaceous Composite (QCC)
by T. Onaka, A. Sakata, S. Wada Y. Nakada,
A.T. Tokunaga, K. Sellgren, R.G. Smith, D.L. DePoy
DISCUSSION I : Carbon in the Interstellar Medium
Chairman: P.G. Martin
213
215
ix
TABLE OF CONTENTS
Section 5 : POLYCYCLIC AROMATIC HYDROCARBONS AND ASTROPHYSICS
Identification of PAH's in Astronomical IR Spectra
- Implications
by A. Leger, L. d'Hendecourt
223
The IR Emission Features : Emission from PAH Molecules
and Amorphous Carbon Particles
by L.J. Allamandola, A.G.G.M. Tielens, J.R. Barker
255
The Hydrogen Coverage of Interstellar PAH's
by A.G.G.M. Tielens, L.J. Allamandola, J.R. Barker,
M. Cohen
273
New Observations of Infrared Astronomical Bands: lRAS-LRS
and 3 ~m Ground-Based Spectra
by M. de Muizon, L.B. d'Hendecourt, T.R. Geballe
287
Distribution of PAH in the Galaxy Derived from the IRAS
Data
by J.L. Puget
303
Infrared Features in Extragalactic Objects
by P.F. Roche
307
Very Small Grains in Spiral Galaxies
by S.K. Ghosh, S. Drapatz
317
lRAS Observations of a "Typical" Dark Cloud
by R.J. Laureijs, G. Chlewicki, F.O. Clark
323
Coal Tar as a Laboratory Analog of an Interstellar PAH
Mixture
by T.J. Wdowiak
327
Hydrogenated Amorphous Carbon (a:C-H) in the Planetary
Nebula NGC 7027
by J.H. Goebel
329
Visual and Infrared Fluorescence from L1780
by G. Chlewicki, R.J. Laureijs
DISCUSSION II : Interpretation of IR Observations
ChaIrman : L. Allamandola
335
339
Section 6 : DIFFUSE INTERSTELLAR BANDS
Possible Carriers of the Diffuse Interstellar Bands
by G. Van der Zwet
DISCUSSION III : The Diffuse Interstellar Bands. Are they
Carried by PAH's ?
Chairman : M. Jura
351
367
x
TABLE OF CONTENTS
Section 7
PHYSICS AND CHEMISTRY OF PAD t S IN THE INTERSTELLAR
MEDIUM
Physic.s and Chemistry of Interstellar Polyc.yc.lic. Aromatic.
Molec.ules
by A. Omont
371
Formation, Destruc.tion and Exc.itation of Carbon Grains
and PAR Molec.ules
by W.W. Duley
373
Polyaromatic. Hydroc.arbons and the Condensation of
Carbon in Stellar Winds
by R. Keller
387
Subject Index
399
PREFACE
The near Infra-Red emission of the Interstellar Medium is a very
puzzling subject. In the brightest regions, where spectroscopic observations are possible from the ground, several bands (3.3 - 3.4 - 6.2 7.7 - 8.6 - 11.3 ~m) have been observed since 1973. The absence of
satisfying explanation was so obvious that they were called "Unidentified IR Emission Bands".
The puzzle still increased when were known the first results of the
general IR sky survey made by the satellite IRAS. On a large scale, the
near IR emission of the Interstellar medium was expected to be very
small but it was observed to be about one third of the total IR emission
for our own galaxy ..•
The situation has moved in 1984 when it was suggested that a class of
stable organic molecules, the Polycyclic Aromatic Hydrocarbons (PAH's)
could be at the origin of this near IR emission. Initially based on the
required refractory character of particules that should be heated to
high temperature without subliming, this hypothesis leads to a suggestive spectroscopic similarity with the observed astronomical bands.
This hypothesis is attractive and it has many implications, for instance, the PAHs would be the most abundant organic molecules in the
universe. However, many points have to be clarified and the different
consequences of this suggestion should be explored.
To reach such a goal, advanced enlightening from Chemistry, Solid State
and Molecular Physics are needed. On the other hand, the astrophysical
conditions provide a unique situation for fundamental studies in these
disciplines: highly isolated molecules, species unstable under laboratory conditions (radicals), long time scales ...
Such conditions motivated the organization of this LnterdLscLplLnary
meetLng between chemists, solid state and molecular physicists and
astrophysicists. All the participants have tried to play the game of
interdisciplinary exchanges, paying special attention to be accessible
to scientists not in their own field and not using technical concepts or
vocabulary without prior introduction.
This was not always easy, for instance, we had discussions to decide
whether'distinct appellations in two disciplines were describing the
same physical phenomenum or not. The net result of this effort was quite
positive and stimulating by all accounts.
xi
PREFACE
xii
The meeting location in the Centre de Physique des Houches was extremely
favorable. Its wonderful scenery and quiteness was a permanent call for
thinking. We thank the Les Houches direction and staff for their welcome
and making things easy and pleasant.
We are very grateful to the CNRS and NATO Scientific Division that
supported the meeting and made it possible. We also thank Laure Anne
Nemirouvsky and Micheline Picarda for secretary work under difficult
conditions.
It is our hope that the proceedings of the meeting will reflect the
exciting atmosphere of the different talks and discussions.
Alain Leger
Louis d'Hendecourt
Nino Boccara
Paris, July 16, 1986
WORKSHOP ORGANIZING COMMITTEE
A. Leger (Paris)
L. d'Hendecourt (Paris)
A. Beswick (Orsay)
J. Friedel (Orsay)
S. Leach (Orsay)
A. Omont (Grenoble)
N. Pana9ia (Bologna)
J.L. Puget (Paris)
W. Schmidt (Ahrensburg)
LIST OF PARTICIPANTS
Allamandola 1.
NASA - Ames Research Center, MS 245-6, Moffett Field,
CA 94035, USA
Baas F.
Laboratory Astrophysics, Huygens Laboratorium,
Wassenaarseweg 78, 2300 RA Leiden, The Netherlands
Beswick A.
LURE, t'niversite Paris-Sud, 91405 Orsay Cedex, France
Bussoletti E.
Dipartimento di Fisica, Universita degli studi di
Leece, 1-73100 Leece, Italy
Colangeli
Dipartimento di Fisica, Universita degli studi di
Leece, 1-73100 Leece, Italy
Cox P.
Max-Planck-Institut fur Radioastronomie, Auf dem Hugel
69, D-5300 Bonn, FRG
Duley W.W,
Department of Physics, York University, 4700 Keele
Street, Downsview, Toronto, Canada M3J IP3
Garwin
Department of Solid State Physics, ETR, Ch-8093 Zurich,
Switzerland
Ghosh S.K.
Max-Planck-Institut fur Extraterrestrische Physik,
D-8046 Garching bei Munchen, FRG
Greenberg M.
Lab. Astrophysics, Huygens Labor., Wassenaarseweg 78,
2300 R.I\ Leiden, The Netherlands
De Groot M.S.
Lab. Astrophysics, Huygens Labor., Wassenaarseweg 78,
2300 RA Leiden, The Netherlands
xiii
LIST OF PARTICIPANTS
xiv
d'Hendecourt L.
Groupe de Physique des Solides de l'Ecole Normale
Superieure, Universite Paris VII, Tour 23, 2 place
Jussieu, 75251 Paris Cedex OS, France
Jortner J.
Dept. of Chemistry, Tel Aviv University, Ramat Aviv,
Tel Aviv 69978, Israel
Joyes P.
Physique des Solides, Bat. 510, Universite Paris-Sud,
91405 Orsay, France
Jura M.
Dept. of Astronomy, UCLA, Los Angeles, CA 90024, USA
Keller R.
Inst. for Astronomy ans Astrophysics, Technische Univ.
Berlin, Hardenberg Str., Berlin, FRG
Kratschmer W.
Max-Planck-Institut fiir Kernphysik, Postfach 103980,
D-6900 Heidelberg I, FRG
Kroto H.W.
School of Chemistry & Molecular Sciences, Univ. of
Sussex, Brighton BN1 9QJ, UK
Laureijs R.J.
Lab. for Space Research and Kapteyn Astronomical
Institut, PO Box 800, 9700 AV Groningen, The Netherlands
Leach S.
Laboratoire de Photophysique Moleculaire, Bat. 213,
Universite Paris-Sud, 91405 Orsay Cedex, France
Lee T.J.
Royal Observatory, Blackford Hill, Edinburgh EH9 3HJ,
Scotland, UK
Leene A.
Kapteyn Astronomical Inst.,
Groningen" The Netherlands
Leger A.
Groupe de Physique des Solides de l'Ecole Normale
Superieure, Universite Paris VII, Tour 23, 2 pl~ce
Jussieu, 75251 Paris Cedex OS, France
Maier J.P.
Inst. fur Physikalische Chemie, University of Basel,
Klingelbergstrasse 80, CH-4056 Basel, Switzerland
Marchand A.
CNRS, Centre Paul Pascal, Domaine Universitaire, 33405
Talence Cedex, France
Martin P.G.
C.I.T.A., University of Toronto,
M5S 1A7, Canada
De Muizon M.
Sterrewacht Leiden, Postbus 9513, 2300 RA Leiden, The
Netherlands (also at Observatoire de Paris, Meudon,
France)
PO
Box
800,
Toronto,
9700 AV
Ontario
LIST OF PARTICIPANTS
xv
Mukamel S.
Department of Chemistry, University of Rochester,
Rochester, N.Y. 14627, USA
Obenland S.
Biochemical Institute, Sieker Landstrasse 19, D-2070
Ahrensburg, FRG
Omont A.
Astrophysique, Observatoire de Grenoble, CERMO, BP 68,
38402 Saint-Martin-d'Heres Cedex, France
Onaka T.
Astronomical Institute, "Anton Pannekoek", University
of Amsterdam, Roetersstraat 15, 1018 WB Amsterdam, The
Netherlands
Phillips
Inst. fiir Chemie, Kernforschungsanlege Julich, GmbH,
Postfach 1913, D-S170 Julich 1, FRG
Puget J.L.
Radioastronomie, Laboratoire de Physique, E.N.S.,
24 rue Lhomond, 75005 Paris, France
Roche P.F.
Dept. of Physics & Astronomy, University College
London, Gower Street, London WC1E 6BT, UK
Rossler K.
Inst. fiir Chemie, Kernforschungsanlege Julich, GmbH,
Postfach 1913, D-5170 Julich 1, FRG
Schmidt W.
Biochemical Institute, Sieker Landstrasse 19, D-2070
Ahrensburg, FRG
Siegmann H.
Dept. of Solid State Physics, ETH, CH-8093 Ziirich,
Switzerland
Thomy A.
Laboratoire Maurice
Villers-Nancy, France
Tielens A.G.
NASA, Ames Research Center, MS 245-6, Moffett Field,
CA 94035, USA
Vala M.
Dept. of Chemistry, University of Florida, Gainesville,
FL 32611, USA
Van Der Zwet G.
Lab. Astrofysica, Leiden University, PO Box 9504, 2300
RA Leiden, The Netherlands
Wdowiak T.J.
Dept. of Physics, University of Alabama at Birmingham,
Birmingham, AL 35294, USA
Lefort,
CNRS,
BP 104,
54600
INTF.oDUCTION
OBJECTIVES OF THE WORKSHOP
L.
d'Hendecourt
Groupe de Physique des Solides de l'Ecole Normale Superieure,
Universite Paris VII, Tour 23,
2 place Jussieu,
75251 Paris Cedex OS, France
ABSTRACT.
Al though we do expect that this Iyorkshop will bring more
questions than solutions, we try here to define schematically what are
the main topics that could possibly be covered at this meeting.
1.
VALIDITY OF THE PAH'S HYPOTHESIS
The identification of the set of the main emission lines present in many
astronomical spectra of very different objects with the fundamental
vibrations of aromatic molecules is very suggestive of the presence of
such molecules in the interstellar medium. By validity of the PAH's
hypothesis, we consider the following items:
a) Are these molecules isolated in space or aggregated onto larger
grains ?
b) Is their deduced abundance, in the objects where the emission
lines are observed, correct and constant from source to source ?
c) Considering their high stability against photodissociation can
we finally assume the presence of these molecules in the diffuse
interstellar medium with a defined abundance respective to the cosmic
abundance of carbon ?
d) If such an assumption is correct, do we have to consider the
PAH's molecules as a new component of the interstellar medium affecting
not only some observations (i.e. the so-called interstellar extinction
curve) but also the physics and the chemistry of the interstellar gas ?
2.
IDENTIFICATION OF PAH'S IN THE INFRARED
Medium resolution infrared spectroscopy only of such complicated
molecules does not allow an unequivocal identification of a precise
molecule. On the other hand, it seems reasonable to assume that interstella.r PAH' s are in fact a collection of different, presumably the most
stable, molecules. From laboratory spectra and by comparison with
astronomical spectra, can we select a mixture of PAH's whose collective
spectrum is representative of the true interstellar one ?
A. Leger et al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 1-2.
© 1987 by D. Reidel Publishing Company.
2
L. D'HENDECOURT
Two bands seem to be of importance for the determination of an
'average size' of the carrier molecule: the band at 3.3 ~m and the one
at 11.3 which pertain to the same molecular subgroup (- CH stretch and
bend respectively). Has the intensity ratio of these two bands been
measured in many objects ? Can we improve the statistics of this
particular data? Equally important is the region 11 - 14 ~m which can
be considered as the 'fingerprints' of the distribution of the CH groups
at the periphery of the molecule. Do we have evidence for this in
astronomical spectra ? What about the far infrared ?
3.
SPECTROSCOPIC PROPERTIES OF ISOLATED MOLECULES
PAH's are large molecules which provide high densities for electronic
and vibrational states. In space, these molecules are collision free:
the relaxation of the adsorbed energy (UV or visible photon) will be
governed by intramolecular processes. What are the main relaxation
channels ? Is the energy re-emitted in the visible by luminescence
phenomena and/or in the infrared? Naturally, complicated problems are
expected there because we should also worry about the spectroscopic
properties of the ions and radicals of these PAH's as in some cases,
these ions and radicals are suspected to be the actual form of the PAH's
in the interstellar environment.
4.
INFLUENCE OF THE PAH'S ON THE INTERSTELLAR EXTINCTION CURVE
PAH's have many transitions in the UV and the visible regions. They have
already been proposed to be the carrier of the famous diffuse interstellar bands although no precise identification has been done. Because
optical transitions are much more specific of a given molecule than
infrared transitions, this problem of the diffuse bands needs a solution
to the precise nature and structure of the molecule(s) carrying these
bands. Are they partially dehydrogenated and ionized?
PAH's do have also strong transitions in the ultraviolet region,
especially around 2000
What could be their influence on the notorious
2200
absorption band in the interstellar medium.
A
5.
A.
POSSIBLE EXPERIMENTS
Finally, it is already obvious that many questions will get an answer
only if laboratory experiments are performed. Because of the nature of
the interstellar environment, one must produce isolated molecules and
study their spectroscopy. Ions and radicals have to be produced. Matrix
isolation techniques provide a straightforward way to isolate molecules
ions and radicals in a cryogenic matrix. In such matrices, spectroscopy
from the far UV to the far IR can be performed. However, the understanding of the conversion of electronic to vibrational energy in
isolated molecules as well as the luminescence phenomena, can only be
dealt with more sophisticated experiments involving supercold supersonic
free jet expansion. Are these experiments useful and feasible ?
PHOTONS, MOLECULES AND SOLIDS IN INTERSTELLAR AND CIRCUMSTELLAR REGIONS:
AN INTRODUCTION FOR NON ASTRONOMERS
M. Jura
Astronomy Department
UCLA
Los Angeles CA 90024
USA
ABSTRACT. We discuss both the physical conditions in the interstellar
medium, and the nature of the mass lost from stars that is injected into
the interstellar environment. Likely sources of PAH's are mass-losing
carbon-rich red giants and carbon-rich planetary nebulae. These objects
account for about half of the mass and most of the carbon injected into
the interstellar medium.
I.
INTRODUCTION
The interstellar medium is a complex, constantly evolving region with a
wide range of physical conditions. Because of the acquisition of data
during the past 20 years from radio to X-ray wavelengths, there now
exists a moderately good understanding of the overall structure of this
material. This review aims to very briefly describe current models to
non astronomers. Much more detailed reviews are of course available;
see, for example, Spitzer (1978).
The interstellar medium is very far from thermodynamic equilibrium.
While it is still possible to use a temperature parameter to approximate
the microscopic velocity distribution of the atoms and molecules as a
Maxwellian, the gas temperature generally only refers to the mean
kinetic energy. The state of ionization and the molecular composition
are all controlled by non-equilibrium processes. However, in many but
not all circumstances, a steady state does prevail.
Each region of the interstellar medium is evolving. It is not
uncommon for an element of interstellar matter to be engulfed in a
supernova explosion with the consequence there there is impulsive
heating of the gas (Cox and Smith 1974, McKee and Ostriker 1977). It
seems likely that these supernova explosions play a key role in
controlling the general structure, the energy density and the
composition of the interstellar medium.
II.
ENERGY DENSITY OR PRESSURE
A fundamental feature of the interstellar medium is that even though
3
A. Leger et al. (eds.) , Polycyclic Aromatic Hydrocarbons and Astrophysics, 3-14.
© 1987 by D. Reidel Publishing Company.
M.JURA
4
there is a wide range of temperatures and pressures, the typical !nergy
density of the important constituents is on the order of 1 eV cm- •
That is, although the gas may be as cool as 10 oK or as hot as 10 6 oK,
the typical value of the gas pressure, nT, is .. 3 10 3 cm- 3 oK or, about
1 eV cm- 3 (Jenkins, Jura and Loewenstein 1983). A number of years ago,
it was suggested that the gas was strictly at constant pressure (Field,
Goldsmith and Habing 1969). However, more sophisticated observations
and models show that while the matter is driven towards uniform
pressure, there are large and important excursions. Important effects
that lead to enhanced densities and pressures in the gas are shock waves
resulting from supernova explosions, winds from hot stars and other
processes (McCray and Snow 1979), and the existence of clouds of
sufficiently large mass that self-gravitation is important so that there
is compression of the gas.
An important consequence of the current analysis of the
interstellar medium is that much if not most of the volume of the region
is filled with very hot gas at ~ 10 6 oK, but a density of only about
0.003 cm- 3 (see Table 1 below). Most of the mass of the interstellar
matter is contained within clouds with densities larger than 10 cm- 3 and
temperatures lower than 100 oK.
Not only does the gas have an energy density of about 1 eV cm- 3 ,
but so do other important constituents as well. It is relatively
straightforward to measure the radiation field in the neighborhood of
the sun. In the ultraviolet for 5 eV < E < 13.6 eV, Draine (1978)
parameterizes the radiation field as:
photons cm- Z s-1 sr- 1 eV- 1
This radiation corresponds to an energy density of 0.06 eV cm- 3 •
However, as shown, for example, by Mathis, Mezger and Panagia 1983, the
bulk of the radiative energy in the solar neighborhood is emitted at
optical and near infrared wavelengths. They find that between 912 A and
8 ~m, the mean energy density is 0.5 eV cm- 3 • The radiation field near
the sun seems to be representative of most regions in the interstellar
medium (Jura 1974), but there are locations either near stars where the
radiation energy dellsity is particularly high or inside dust clouds
where the energy density is unusually low.
The energy density of cosmic rays is close to 1 eV cm- 3 (see, for
example, Leger, Jura and Omont 1985). This conclusion is reached both
from measurements of the cosmic ray flux at the earth and, indirectly,
by studying the ionization in the interstellar matter and the production
of Y rays that result from interactions between cosmic rays and cold
matter (Cesarsky and Volk 1978). The ionizat.ion of cold matter is
mesaured by a variety of means (see Spitzer and Jenkins 1975) including,
for example, the mill.imeter study of emission from molecular ions such
as HCO+. The energy density of cosmic rays seems to be relatively
constant in the solar neighborhood.
Finally, the magnetic field energy is also characteristically on
the order of 3 10- 6 Gauss (Heiles 1976, Troland and HeHes 1982) which
corresponds to an energy density, BZ/811, of close to 1 eV cm- 3 •
PHOTONS, MOLECULES AND SOLIDS IN INTERSTELLAR AND CIRCUMSTELLAR REGIONS
5
Measurement of the magnetic field in the Milky Way is difficult, but it
does seem to be close to this value at least in ionized regions where
pulsar dispersion measures can be performed and in some clouds where
Zeeman splitting of the 21 cm line of hydrogen is observed.
It should also be recognized that while the magnetic field, cosmic
ray energy density, radiation field of optical and ultraviolet photons
and gas thermal pressure are all roughly comparable, the theory to
explain this rough equality has not yet been described. It may even be
a coincidence, but this seems unlikely.
II.
0
MORPHOLOGICAL STRUCTURES
Hydrogen is by far the most abundant constituent of the medium, and
one can describe the gas by whether the hydrogen is either mainly
ionized or mainly neutral. Traditionally (see Spitzer 1978), the
interstellar medium has been divided into H I and H II regions, refering
to neutral and ionized, hydrogen, respectively. Hot stars with
Teff ) 35,000 OK photo-ionize hydrogen and produce prominent and easily
detectable volumes of ionized gas (or Stromgren S?heres) such as the
Orion Nebula. The boundary between regions of ionized gas and neutral
gas is usually sharp on an astronomical distance scale. An H II region
might have a diameter of 10 pc (1 pc = 3.085 10 18 cm) while the boundary
between the ionized and neutral gas might have a thickness of 0.01 pc.
The ionized gas has a typical temperature of 10,000 OK, while the
neutral gas might have a temperature of 100 OK. Within the H I regions,
the Lyman continuum opacity is sufficiently large that only photons with
energies smaller than 13.6 eV penetrate. The only exception to this is
high energy X-rays and Y rays.
This traditional view of the interstellar medium has been
substantially modified during the past ten years. In Table 1, we sketch
some of the parameters of many of the main structures now thought to
exist in the interstellar matter. The general picture is there exist
dense clouds embedded within a hot, low density intercloud medium.
However, as mentioned above, the pressure within the interstellar medium
is not uniform.
The difference structures listed in Table 1 are thought to be at
least in part a consequence of multiple supernova explosions in the
interstellar medium. Not only do supernovae inject large amounts of
kinetic energy into localized regions of the interstellar medium, but
the shocks resulting from the supernova explosions may actually
propagate to very large distances. As a result, the overall structure
of clouds and inter cloud gas in the medium may be largely controlled by
supernova explosions. After a shock of )200 km s-1 passes through the
gas, it is heated to over 10 6 OK and it takes a long time () 10 7 years)
to cool down to its initial temperature. As a result, large volumes of
the interstellar medium may be at this very high temperature. However,
lower speed shocks only lead to a relatively brief rise in the gas
temperature because at lower shock speeds, radiative cooling behind the
shock occurs rapidly. Consequently, dense condensations result from
"snowplowing" behind these relatively slow shocks, and this may be the
origin of many interstellar clouds. As noted above, most of the gas is
contained within relatively cool (T < 100 OK) clouds.
6
M.JURA
Table 1 -- Structures in the Interstellar Medium
Name
Density
(cm- 3 )
Temperature
(OK)
Diagnostics
Hot, intercloud
gas
3 10- 3
10 6
Soft X-ray emission, 0 VI
absorption in the UV
Warm, inter cloud
gas
10- 1
10 4
If ionized
optical
emission lines, pulsar
dispersion measures
If neutral-- 21 cm
observations of H I
Diffuse cloud
30
100
Optical, UV absorption
lines, 21 cm emission and
absorption
10
Optical, dark patches
IR emission, molecular
emission
Dark cloud
Note: There are also well differentiated special regions in the
interstellar medium such as supernova remnants and H II regions.
The neutral gas displays a variety of structures. Some fraction
of this material is relatively wam (10,000 oK > T > 1000 OK) and of low
density (Dickey, Salpeter and Terzian 1979). Most, however, appear to
have temperatures less than 100 oK.
As described below, there is a good correlation between gas and
dust in the interstellar medium. Clouds with relatively small amounts
of dust which are therefore nearly transparent to optical and
ultraviolet photons are denoted " diffuse". Otherwise, i f there is a
large amount of dust, a cloud is described as "dark". Very small
diffuse clouds are mainly composed of atomic hydrogen. However,
molecular hydrogen can be synthesized on the surfaces of grains, and
when there is enough of it, the H2 is "self-shielding" in the sense that
it protects itself against destruction by ultraviolet photons
(Hollenbach, Werner and Salpeter 1971, Shull and Beckwith 1982). That
is, the H2 in the outer portion of a cloud absorbs the ultraviolet
photons and protects the molecular hydrogen in the inner portion of the
cloud. As a result, H2 can survive in large quantities in some diffuse
clouds and in dark clouds it is thought that nearly all the hydrogen is
molecular.
Within diffuse clouds, the physical conditions are such that aside
from hydrogen, most of the material is atomic. These atoms are subject
to the general stellar radiation field, and if the ionization potential
is less than 13.6 eV, the species is mostly ionized (> 99%) in the gas.
as, for example, is the case for carbon and iron. If the atom's
ionization potential is larger than 13.6 eV, such as the case for oxygen
and nitrogen. then the atoms are mostly neutral. In dark clouds where
there is a large amount of dust to shield out the ultraviolet. atoms and
PHOTONS, MOLECULES AND SOLIDS IN INTERSTELLAR AND CIRCUMSTELLAR REGIONS
molecules are mainly neutral except for the relatively small
concentrations of molecular ions such as HCO+ that result from cosmic
rays and X-rays.
IV.
ABUNDANCES AND DUST
Although hydrogen is the major constituent of the interstellar medium,
the other elements playa vital role in controlling the gas temperature,
and, of course, the molecular composition. It seems that at least
within about 1 kpc of the sun, the abundances in the interstella~ medium
are uniform to within a factor of 2 (Jura 1982)1 and are quite similar
to those in the sun. In fact, observations of 2CH+/13/ CH+, chosen to
be insensitive to problems such as systematic depletion onto grains,
indicate that the abundances may be homogeneous to within 20% (Hawkins,
Jura and Meyer 1985; Hawkins and Jura 1986). The most recent
determinations of solar abundances are listed by Grevesse (1984) and
Breneman and Stone (1985).
Solid dust grains contain about half of all the material heavier
than helium within the interstellar medium (Aannestad and Purcell 1973).
This is inferred both from the presence of interstellar dust and the
observed lack of many refractory elements in the gas phase within
interstellar clouds (Spitzer and Jenkins 1975). The dust to gas ratio
appears to be uniform to better than about a factor of two (Bohlin,
Savage and Drake 1978).
Analysis of the nature of grains depends upon the interstellar
extinction, the inferred scattering properties of grains in reflection
nebulae and the diffuse galactic light, the interstellar polarization
and the infrared emission. A detailed description is given by Greenberg
in this workshop.
V.
INTERSTELLAR MOLECULES
Nearly 50 years ago, the molecules CH, CH+ and CN were first discovered
in diffuse clouds in the interstellar medium by their optical absorption
lines. The abundances of these carbon bearing molecules in diffuse
clouds is < 10- 4 that of hydrogen. Subsequent to this work, there was
little additional information about interstellar molecules until the
advance of radio and millimeter technology during the past 20 years. At
the moment, there are 68 known or suspected identifications of
interstellar molecules (see Snyder 1985). H2 is special; it has been
discovered by its ultraviolet absorption lines to be very widespread in
the interstellar medium (Spitzer and Jenkins 1975).
With modern techniques, it is possible to make confident detections
at optical wavelengths of molecules with gas phase abundances ~10-10 of
that of hydrogen, as for example shown by the detection of 13CH+ toward
~ Oph by Hawkins, Jura and Meyer 1985.
Therefore, if PAR's have optical
absorption lines, and if specific PAH's truly have abundances near 10- 7
of hydrogen in diffuse clouds, it may be possible to detect them.
Known interstellar molecules, perhaps unlike PAR's, are
concentrated in dark clouds. This is because the rates of formation are
relatively rapid and there is little destruction because ultraviolet
radiation is shielded by dust. It is thought that in dark clouds, most
M.1VRA
8
matter (except noble gases) is contained either in grains or in
molecules.
Models for understanding the presence of these different species
mainly by gas phase chemistry have grown quite elaborate, and, on, the
whole, seem reasonably successful (Graedel, Langer and Frerking 1982,
Prasad and Huntress 1980). The following conclusions can be drawn from
observations and analysis of these molecules:
1.) H2 is extremely abundant and is synthesized on the surfaces of
grains. However, the concentrations of most other molecules can be
explained by gas-phase synthesis. Certainly some gas phase reactions
most proceed as is shown by the widespread presence of ionic molecules
such as HCO+.
2.)
Molecules are mainly destroyed by absorption of ultraviolet
photons. Therefore, dark clouds are regions where molecules are
particularly abundant. As discussed by Leger and Puget (1984), these
arguments may not pertain to PAR's.
3.) Except in shock waves and a few special regions, most of the gas is
quite cold (T < 100 OK). Therefore, gas phase chemical reactions
without significant activation energy barriers are most important.
These include ion-molecule reactions, charge-exchange reactions,
dissoci~tive recombination, reactions with some radicals and radiative
association. Inside a dark cloud, ionization of molecules is
precipitated mainly by cosmic rays and therefore the chemical
composition of a cloud is the result of a complex interplay among dust,
ultraviolet radiation, gas-phase abundances and cosmic rays.
4.) Besides hydrogen and helium, which, of course, is chemically inert,
the most abundant elements in the gas phase in the interstellar medium
are carbon, nitrogen and oxygen. Therefore, it is not surprising that
most interstellar molecules are composed of these species. A few
molecules are known to contain sulfur and silicon.
5.) Rather complex species can be synthesized in the interstellar
medium. The molecule with the largest number of atoms that is known to
exist in the interstellar medium is HC11N with 13 atoms.
VI.
SOURCES OF INTERSTELLAR MATTER
Interstellar matter undergoes continuous evolution. For example, grains
grow mantles, collide and coalesce, and are also eroded and destroyed by
various processes (see, for example, Seab and Shull 1983, 1985).
Therefore, there is not a simple correspondence between the sources of
interstellar matter and the grains that we observe. Nevertheless, it is
important for understanding the origin and evolution of PAH's to
consider how grains form and evolve.
The major sources of interstellar matter are mass-losing stars.
The stars that eject large amounts of mass into the interstellar medium
are red giants, planetary nebulae, hot stars, and supernovae. Probably
most interstellar matter results from mass loss from cool, red giant
stars (Zuckerman 1980, Olofsson 1985). These stars eject up to
10-~ Mo yr- 1 at a characteristic speed of -15 km s-l. The. atmospheres
of these. stars are sufficiently cool (Teff - 3000 OK) that as the matter
flows away from the star and cools even further by adiabatic expansion
and radiative loses, solid grains condense out of the flow. In fact, it
PHOTONS, MOLECULES AND SOLIDS IN INTERSTELLAR AND CIRCUMSTELLAR REGIONS
9
seems likely that the grains are not only a result of the flow, but, in
addition, radiation pressure on the dust expels the matter to infinity
(see Jura 1986a).
It appears that in the outflows from a cool red giant, most of the
matter that is sufficiently refractory that it could condense into
solids at, say, T • 1000 oK, actually does so (Jura 1984, Knapp 1985,
Jura 1986b). That is, red giant stars are very efficient factories for
the manufacture of small grains. Although the surface layers of red
giants often display unusual abudances as the result of nucleosynthesis
in the star's interior and subsequent dredge-up of the newly generated
elements, this effect is usually not too pronounced except in rare
objects such as R CrB stars. The dust to gas ratios in the outflows
from red giants are usually close to the value, approximately 1% by
mass, in the interstellar medium (Knapp 1985, Jura 1986).
Aside from hydrogen and helium, carbon and oxygen are the most
abundant constituents in the atmospheres of red giant stars. Because CO
is so very stable, this molecule is the most common constituent besides
hydrogen in the outflow from the star. (This is fortunate because with
current technology, CO is easily studied at millimeter wavelengths and
therefore we are able to derive a great deal of information about the
mass loss from red giants from such radio observations.) Stars are
oxygen rich if [0) ) [C) so that there are free oxygen atoms to form
other molecules besides CO. If [C) > [0), then there is excess free
carbon beyond that contained in CO, and there is then a very rich carbon
chemistry in the outflow.
As is shown by millimeter wavelenght observations, the carbon-rich
stars, such as the very well studied object IRC +10216 display a large
number (about 20, see Morris 1985) carbon-bearing molecules ranging from
CO to RC11N. Besides having a rich chemistry, these stars lose a large
amount of mass; i t appears that IRC +10216 has already ejected close to
a solar mass of material back into the interstellar medium. Therefore,
it is imaginable though by no means proven that these stars are a major
source of PAR's. While most stars in the sky are oxygen-rich, it turns
out, for reasons that are not yet fully understood, that about half of
the mass ejected into the interstellar medium is produced by carbon-rich
stars and about half by oxygen-rich stars (Knapp and Morris 1985).
Therefore, well over half of the carbon injected into the'interstellar
medium comes from carbon stars. Much of this carbon is contained within
grains and CO but much could also be contained within PAR's. It should
be noted that only a relatively few molecules including CO, SiO, OR ,
H20, H2S and HCN have been detected in the outflows from oxygen-rich
stars.
Red giants are cool stars where nuclear reactions in the interiors
provide the source of the observed luminosity. Eventually, the star
uses up all its nuclear fuel and unless it is more massive than 1.4
times the mass of the sun, it becomes a white dwarf, the final stage in
its evolution. As the star becomes a white dwarf, it shrinks in radius
and its surface temperature rises dramatically so that the star now
emits high energy (E > 13.6 eV) photons. Consequently, the cool matter
that was ejected when the stars was a red giant is photo-ionized and
heated (see Kwok 1982). This hot gas is detected optically through
emission lines such as Ra as a planetary nebula.
10
M.JURA
The progenitors of planetary nebulae are red giant stars, and as
with red giants,
planetaries therefore are either oxygen-rich or
carbon rich (see, for example, Zuckerman and Aller 1986). It turns out
that the emission lines associated with PAR's are only seen in
carbon-rich planetary nebulae, consistent with the view that the
infrared features are carried by carbon-rich substances (Barlow 1983).
The detection of PAR's in planetary nebulae raises an important
question. Are these species manufactured in the red giant stars as they
lose mass or are the PAH's present in the outflows produced only as the
gas is shocked as the ionization front moves through the cold material?
Future work will help establish the correct picture.
The size distribution of circumstellar grains is not very well
established (Jura 1985). Criteria that can be used to infer grain sizes
include the circumstellar extinction curves, the amount of circumstellar
scattering and its resulting polarization as a function of wavelength,
and the heating of the gas by supersonic streaming of the grains through
the envelopes. Also, the shape of the silicate emission features in
oxygen-rich stars have been used to constrain the size of circumstellar
grains (Papoular and Pegourie 1983), but this procedure is somewhat
uncertain because the intrinsic shapes of the emission features and the
temperature distribution of the circumstellar grains are not well known.
At the moment, it is certainly possible to imagine that there are very
small circumstellar grains which may in fact be PAR's; however, there is
no direct evidene that outflows from red giants do in fact contain
PAR's.
In contrast to what we know about red giants, it seems well
established that carbon rich planetary nebulae do in fact eject PAR's
(or at least the carrier of the features such as that at 3.3 ~m) into
the interstellar medium. Red giant stars eject about 0.3 Mo (one solar
mass) yr- 1 into the interstellar medium (Knapp and Morris 1985). If
planetary nebulae have an average mass of 0.2 }fo, then their overall
rate of mass loss into the interstellar medium is comparable to that
from red giants (Cahn and Wyatt 1976). Even if red giants do not
produce PAR's, the contribution from planetary nebulae could be
substantial.
Not all sources of interstellar matter produce PAR's. Except in
rare case, mass loss from hot stars probably does not produce dust.
However, the amount of mass ejected from these hot stars is
significantly smaller than the mass ejected from cool stars (Abbott
1982). Also, supernovae return newly synthesized elements into the
interstellar medium, but the rate of injection of matter appears to be
about an order of magnitude smaller than the 0.3 }fa yr- 1 quoted above
for red giants (Trimble 1983). It is not yet clear whether supernovae
produce grains much less whether they produce PAR's.
In conclusion, it seems quite possible that PAR's are synthesized
in the outflows from carbon-rich red giants and planetary nebulae. Their
fate after injection into the interstellar medium is described by Omont
elsewhere in this workshop.
This work has been partly supported by the NSF and NASA. I have had
many useful conversations and correspondence with Alain Leger and Kris
Sellgren.
PHOTONS, MOLECULES AND SOLIDS IN INTERSTELLAR AND CIRCUMSTELLAR REGIONS
11
REFERENCES
Aannestad, P. A., and Purcell, E. M. 1973, Ann. Rev. Astr. and Ap., 13,
133.
Abbott, D. C. 1982, Ap. J., 263, 723.
Barlow, M. J. 1983, IAifSymp. No. 103: Planetary Nebulae, ed. D. R.
Flower (Dordrecht: Reidel), p. 105.
Bohlin, R. C., Savage, B. D., and Drake, J. F. 1978, Ap. J., 224, 132.
Breneman, H. H., and Stone, E. C. 1985, Ap. J. (Lettersr:-299, L57.
Cahn, J. H., and Wyatt, S. P. 1976,~, 210, 508.
Cesarsky, C. J., and Volk, H. J. 1978, Astr. and Ap., 70, 367.
Cox, D. P., and Smith, B. W. 1974, Ap. J. (Letters), 189, LI05.
Dickey, J. M., Salpeter, E. E., and Terzian, Y. 1979,~, 228, 465.
Draine, B. T. 1978, Ap. J. Suppl., 36, 595.
Field, G. B., Goldsmith, D. W., and Habing, H. J. 1969, Ap. J.
(Letters), 155, L49.
-----Graedel, T. E., Langer, W. D., and Frerking, M. A. 1982, Ap. J. Suppl.,
48, 321.
Grevesse, N. 1984, Phys. Scripta, T8, 49.
Hawkins, I., and Jura, M. 1986, in preparation.
Hawkins, I., Jura, M., and Meyer, D. M. 1985, Ap. J. (Letters), 294,
L131.
Heiles, C. 1976, Ann. Rev. Astr. and Ap., 14, 1.
Herbst, E., and Klemperer, W. 1973, ~, 185, 505.
Hollenbach, D. J., Werner, M. W., and Salpeter, E. E. 1971, Ap. J., 163,
165.
Jenkins, E. B., Jura, M., and Loewenstein, M. 1983, Ap. J., 270, 88.
Jura, M. 1974, Ap. J., 191, 375.
-----Jura, M. 1982, "irlMvances in Ultraviolet Astronomy, Four Years of IUE
Res earch, Y. Kondo and J. Mead, eds., NASA.
Jura, M. 1984, ~, 286, 630.
Jura, M. 1985, in Inter-Relationshipos Among Circumstellar,
Interstellar, and Inter-Planetary Dust, ed. J. Nuth and R. E.
Stencel, NASA.
Jura, M. 1986a, Irish Astr. J., in press.
Jura, M. 1986b, ~, in press.
Knapp, G. R. 1985, Ap. J., 293, 273.
Knapp, G. R., and Morris, M. 1985, Ap. J., 292, 640.
Kowk, S. 1982, ~, 258, 280.
-----Leger, A., Jura, M., and Omont, A. 1985, Astr. and Ap., 144, 147.
Leger, A., and Puget, J.-L. 1984, Astr. and Ap., 137, L5.
Mathis, J., Mezger, P., and Panagia, N. 1983, Astr. and Ap., 128, 212.
McCray, R., and Snow, T. P. 1979, Ann. Rev. Astr. and Ap., 17, 213.
McKee, C. F., and Ostriker, J. P. 1977,~, 218, 148.
Morris, M. 1985, in Mass Loss from Red Giants, M. Morris and B.
Zuckerman, eds. (Dordrecht: Reidel).
Morris, M., and Zuckerman, B. 1985, Mass Loss from Red Giants
(Dordrecht: Reidel).
Olofsson, H. 1985, in Workshop on Submillimeter Astronomy, ed. P. A.
Shaver, European Southern Observatory.
Papoular, R., and Pegonrie, B. 1983, Astr. and Ap., 128, 335.
Prasad, S. S., and Huntress, W. T. 1980, Ap. J. Suppl., 43, 1.
MoJURA
12
Seab, G., and Shull, J. M. 1983, Ap. J., 275, 652.
Seab, G., and Shull, J. M. 1985, iUIirter-Relationships Among
Circumstellar, Interstellar and Interplanetary Dust, ed. J. Nuth and
R. E. Stencel, NASA.
Shull, J. M., and Beckwith, S. 1982, Ann. Rev. Astr. and Ap., 20, 163.
Snyder, L. 1985, in Workshop on Submillimeter Astronomy, ed. P., A.
Shaver, European Southern Observatory.
Spitzer, L. 1978, Physical Processes in the Interstellar Medium (J.
Wiley: New York).
Spitzer, L., and Jenkins, E. B. 1975, Ann. Rev. Astr. and Ap., 13, 133.
Trimble, V. 1983, Rev. Mod. Phys., 55, 511.
Troland, T. R., and Reiles, C. 1982, ~, 252, 179.
Zuckerman, B. 1980, Ann. Rev. Astr. and Ap., 18, 263.
Zuckerman, B., and Aller, L. R. 1986, Ap. J., in press.
DISCUSSION
Omont: When you take into account the possible lifetime of PAR's in the
interstellar medium, do you think that the amount of PAR's injected by
red giants and planetary nebulae could account for the amount of
interstellar PAR's?
Answer: If PAR's are in fact pervasive throughout the diffuse
interstellar medium, it is unlikely that they are only produced in
stellar sources. Interstellar grains are thought to be destroyed on a
time scale short compared to their synthesis time in evolving stars, and
PAR's are probably not any more durable than grains.
Leach: 1.) What are the prospects for higher spatial resolution for
observations in the 3 -12 \1m region? 2.) Degradation of cosmic ray
energy absorbed by species in dark clouds should produce a distribution
of electrons having at least some quite high energy components. These
could play specific roles in excitation and ionization processes. 3.)
Optical determination of species having an abundance of 10- 10 with
respect to hydrogen is perhaps restricted to diatomics and some small
polyatomic species. In the far ultraviolet, PAR's would have very broad
absorption features and so would be difficult to detect in this spectral
region. Some Rydberg feat rues would have narrow structures, but the
higher members would be relatively weak in absorption from the ground
state.
Answer: 1.) Infrared instrumentation is constantOly improving; detector
arrays at 10 \1m will greatly increase our ability to perform spatial
mapping. 2.) Prasad and Tarafder (1983, Ap. J., 267, 603). have noted
that the secondary electrons ejected from
ray ionization will
lead to collisional excitaion of the Lyman and Werner bands of R2. This
could be a significant source of ultraviolet photons within dark clouds.
3.) At the Lick Observatory, we have used spectral resolution of 0.05 A
to study interstellar molecules. Of course if the line is intrinsically
broader than this, our sensitivity to column density will be decreased.
cosmrc
Roessler: 1.) Row much of a serendipity approach is the assignment of
IR spectra to the precise class of PAR's, especially coronene? 2.)
Comment: Carbon species with some eV kinetic energy (10-100 km s-l)
PHOTONS, MOLECULES AND SOLIDS IN INTERSTELLAR AND CIRCUMSTELLAR REGIONS
13
when moving into another gas or dust domain, may undergo precise hot
reactions with hydrocarbons to form precursors of PAR's: acetylene and
its derivatives.
Answer: 1.) See the presentation by Leger.
Leger: If the abundance of PAH's in carbon mass-losing stars was the
same as where they are observed (in emission), one would expect an
optical depth at 3.3 ~m of 1% for a visual optical depth of 25. I do
not think that the present measurements are able to detect such an
absorption in the IR.
Answer: I agree.
Abouaf: As there are positive ions in the interstellar medium, there are
also electrons, do they playa role in the reactions?
Answer: Electrons are important in collisional excitation and
ionization in hot regions. They are of course central in radiative
recombination of atoms and dissociative recombination of positively
charged molecules. Electronic excitation of molecules is usually not
too important with a few exceptions such as CN, which has a large dipole
moment and is found in diffuse clouds where the relative concentration
of electrons is considerably higher in dark clouds (see, for example,
Meyer and Jura 1985,~, 297, 119).
Kroto: Will IRC+I0216 evolve into a planetary?
Answer: Yes. Iben and Renzini (1983, Ann. Rev. Astr. and Ap., 21,
271) have suggested that this may occur within the next ten years.
Siegmann: Is it correct that PAR's are not detected in oxygen-rich
stars, but are detected in carbon rich stars? If this is true, it would
speak for a synthesis of PAR's in circumstellar regions, because one
observes in combustion of coal that PAR's are formed only in oxygen
deficient combustion.
Answer: Moderately complex carbon molecules have been detected in
IRC+I0216, but as yet nothing is known about species as complicated as a
PAR.
Ghosh: You showed the CO contours of IRC+I0216. What was the beam size
for that observation Is the source resolved?
Answer: I showed unpublished data obtained at Bell Labs in the J = 1-0
rotational line of CO using a telescope with a 90" beam. Since the
object is 8' in diamter, it is well resolved.
Wdowiak: Comment: Ted Snow has suggested that current ultraviolet
measurements would not reveal an absorption feature equivalent to the
4430 A diffuse intestellar band which is the strongest of the visible
diffuse interstellar bands.
Roche: Copmment: NGC 7027 has been mapped in the 11. 3~m band and the
[S IV] emission line, and the data clearly show that the 11.3 ~m
emission peaks outside the ionized region, probably in a thin shell just
outside the R II region.
M.JURA
14
DuleY: What is the detection limit for molecules in the VUV?
TIf-=iil" n?
Is it
Answer: It should be close to that value with the old Copernicus
satellite data; IUE is not so powerful. With the Righ Resolution
Spectrograph on the Space Telescope, it should be possible to do
better.
Greenberg: 1.) According to Roche, the IR emission occurs at the outer
shell of the ionization region. Could this indicate that the PAR's (or
in any case the emitters) are produced by shock breakup of already
present dust? 2.) In the Pleiades the dust is moving with respect to
the stars at ~ 30 km s-l. Could this provide a physical basis for
making the PAR's locally rather than their being already present in the
interstellar medium?
Answer: 1.) This is a reasonable but not certain interpretation of the
data of NGC 7027 obtained by Roche and Aitken. 2.) PAR's could be
manufactured by grain-grain collisions in shocks.
ON THE ELECTRONIC STRUCTURE OF GRAPHITE
Pierre Joyes
Laboratoire de Physique des Solides,
Batiment SID, Universite de Paris-Sud
91405 Orsay Cedex (France)
ABSTRACT. We develop the simple tight binding description of the
graphite band structure for TI and 0 electrons and we compare it with
recent calculations. A discussion of the origin of the optical absorption peaks and of the effect of dielectronic correlations on electronic
properties is also presented.
I. INTRODUCTION. GENERALITIES ON BAND STRUCTURES.
Let us first recall some general properties of band structures in periodic media.
I. According to the Bloch theorem the wave functions in a periodic
potential can be written under the form :
-+
1Jik(r)
=
e
ih
-+
uk:(r)
(I)
where k: is a wave vector of the reci~rocal space which can be use~ as
an index for the wave functions, uk:(r) is a periodic function of r.
-+
2. The vectors K of the reciprocal space defined by
.-+ .....
e~K.R = I
(2)
R
where
is any vector of the direct lattice, form the reciprocal lattice
(R.L.). Let us see on some examples how they playa particular role in
electronic problems.
-+
As the function uk:(r) is periodic, one may write :
-+
1Jik(r)
(3)
-+
KER.L.
where
~
are constants. Let us now consider a typical matrix element
15
A. Leger et al. leds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 15-30.
© 1987 by D. Reidel Publishing Company.
16
P. JOYES
(4)
where Hint is a space independent operator.
This kind of matrix element appears, for example, in the expression of
the optical transition probability. By using the development 3 in
formula 4, it is easy to show that this matrix element is non zero only
if
(5)
that is, only
lattice.
.
~f
+
+
kl and k2 are separated by a vector of the reciprocal
Once a given point of R.L+ is chosen as an origin,it can be shown that
the dispersion curves E(k) (E(k) is the eigenvalue corres£onding to
the eigenfunction Wk(~» can all be drawn, by convenient K translations,
in one region of the reciprocal lattice called the first Brilloin zone.
This zone can be defined as the region of the reciprocal space which
is closer to the origin than to any other point of the reciprocal lattice. It has a more or less complicated polyhedral shape.
In the first Brilloin zone repre~entation of the E(k) curves
allowed transitions 5 appear as K = 0 vertical transitions.
the
Another characteristic of band structures is that, in general, when k
tends to the first Brilloin zone values,the surfaces of equal energy
tend to be orthogonal to the sides of this first Brilloin zone. This is
one of the differences between Bloch electrons and free electrons
which are described by plane waves and for which surfaces of equal
energy are spheres centered on the origin.
II. PLANAR GRAPHITE.
11.1 Direct and reciprocal lattices.
Graphite is known as a highly anisotropic medium which consists of
slightly coupled graphite planes. In these planes carbon atoms are Sp2
hybridized. A first step in the study of graphite is to neglect interplane coupling and to examine only one graphitic plane.
The direct lattice appears on figure I. There are two non equivalent
atoms per unit cell which is the rhombus drawn on the figure.
The f.irst points of the reciprocal lattice and the first Brillon zone
are given figure 2.
17
ON THE ELECI'RONIC STRUCI1JRE OF GRAPHITE
Figure 1
The direct lattice for planar graphite 1~1=lbl=I~1
41t
ra
Figure 2
First Brilloin zone for planar graphite.
o
1.42 A.
P.JOYES
18
11.2 Tight binding study of
~
electrons.
In Sp2 hybridization, the orbitals are separated into a orbitals, which
are linear combinations of 2s, 2px and 2py atomic orbitals, and ~ orbital which is the 2pz component (Oz axis is perpendicular to the graphite
plane).
In the tight binding model, the bulk ~ levels are linear combination of
the ~ atomic orbitals lij>, where i refer to a cell and j = 1 or 2 to
one of the two non-equivalent atoms in this cell. We therefore write:
-+-
"'1'...
1<: ()
r
The
~
=
-+-k
E a 1J
.. liJ·>
ij
(6)
hamiltonian can be written
H=E
Elij><ijl +Yo E
lij><i'j'l
p ij
ij and i'j' neighbours
(7)
where Ep is the atomic 2p energy and Yo (Yo < 0) a hopping integral
between first neighbours.
We see on figure I that atoms I, V and VI are of j = I type and atoms
II, II, IV of type j = 2. The equations which couple the aij coefficients
of formula 6 are, by using 7
k
k
k
k
(E~(k) - Ep)a I = Yo (all + alII + a IV )
(8)
k
Now, the tight binding form of the Bloch theorem I shows that all a ..
1J
with the same j are simply related by a phase factor :
k
alII
k
= all
e
ik(b - ;)
(9)
k
k
By using relations 9 and similar formulas for avkand a VI fhe system 8
is reduced into a second order system (between a I and
solution is non zero only if
E (k)
'If
=
-+-
-+-
-+
-+-
-+-
all) whose
(10) I
-+
-+-
-+-
E ± Y (3 + 2cos k (b - a) + 2cos k (a - c) + 2cos k (c - b»
p
0
2'
19
ON THE ELECTRONIC STRUCTURE OF GRAPHITE
The E
iT
(k)
curves are given figure 3 for various
of equal energy (E (k)
iT
=
figure 4.
k directions.
constant) in the k ,k plane are given
x
y
>.
D'I
L-
eu
c:
UJ
M
r
K
Figure 3
Variation of
figure 2.
iT
The lines
k along
energies with
the
rKM
o
Figure 4
Line of equal (E (k) - E )/y
iT
P
0
(formula 10)
path of
P.JOYES
20
For a solid with N atoms, the Born Von Karman conditions give the N
possible k values (there is one IT state per atom).
11.3 Tight binding study of cr electrons.
There are three cr wave functions per site. The hamiltonian of the
problem can be written (see figure 5) :
H = EO L I iJ><iJI +
cr iJ
+
where
I).
8
L
iJ,i J
I).
L
iJ,J'~J
I iJ><iJ' I
(J 1)
liJ><iJJI
(8) are negative intrasite (intersite) hopping integrals with
E - E
s
P
I).
3
and
E + 2E
P
EO = s
0-
3
(Es is the 2s energy). In the carbon case we have
8 > 23
X
Figure 5
Schematic representation of the cr orbitals.
It can be shown (1) that cr and
IT
eigenenergies are simply related by
1
± (~1).2 + S2 + I).S£)2
4
(12)
ON THE ELECTRONIC STRUCTURE OF GRAPHITE
with
and
21
-+-
£ = (E (k) - E )/y
1T
p
0
(13)
E (k) is given by 10.
1T
As there is one 1T state per atom, formula 12 gives two cr states per
atom, the other cr eigenenergies are :
E
cr
=
EO - b, ±
cr
B
(14)
Each of the two energies given by 14 appear N/2 times in aN-atom
graphite plane. The cr band presents a large gap bet~een its bonding
and antibonding parts. In the case of carbon where X
the limits of
>j
the cr bonding part are given by using the minus sign in 12 and letting
£ = + 3 (which gives Eg + 2b, + B) and £
3 (which gives
b, + S).
The antibonding band limits are EO + 2b, - Band Eg - b, - B
(plus sign in 12 and £ = ± 3).
cr
Eg -
III. THE 3-DlMENSIONAL GRAPHITE.
111.1 Direct and reciprocal lattices.
The direct lattice is shown on figure 6.
z
Figure 6
Direct lattice of graphite, Co
o
6.74 A
c
Two successive graphitic planes (separated by distance ~) are shifted
2
from one another. The elementary cell contains 4 atoms. The first
Brilloin zone is given figure 7.
22
P. JOYES
Figure 7
First Brilloin zone of graphite.
111.2 n electronic structure.
Instead of the second order problem of chapter II, we must now solve
a fourth order problem. Many interplane terms can be included in the
hamiltonian. The first step is to take into account a YI hopping term
(lyll < IYol) between the A and A' atoms of figure 6 which are exactly
superimposed. The region near the HKH line of the Brilloin zone (see
figure 7) is of particular interest since in the 2-dimensional description the n bondin~+and antibonding bands join together at the K point.
By letting KII = I k II - k(HKH) I the dispersion curves in this region are
given by
E
p
± Y j cos(k
z
Co
Co
3
2
2
j
--) ± (Yj cos(k --) + - y a K )2
2
z
2
4
0
II
(15)
We see (figure 8) that in the neighbourhood of the HKH line we must
consider four n energies instead of the two values given by the 2-dimensional treatment. At the H point we obtain the 2-dimensional behaviour.
When we include other hopping integrals in the hamiltonian the calculation becomes rapidly more intricate. However, some general features may
still be understood. For example, when one considers the matrix element
Y2 between B atoms of figure 6 separated by one period Co (i.e. separated by 2 graphitic planes) ( IY 2 1 < Iyjl) it appears, in En(t) , a dependence on :
23
ON THE ELECTRONIC STRUCTURE OF GRAPHITE
(bl
Figure S
variation of energy in the HKH neighbourhood.
a) Variation along HKH.
b) For one k z value, variation with
(formula 15).
Xu
c
{as we obtained, in 15, a dependence on YI cos (k z 20 ) from the coupling
Co
of planes separated by - ) .
2
This new k z dependence is responsible for an oscillation of the central
curve of figure Sa, with a maximum larger than EF (the Fermi level) at
point K and a minimum smaller than EF at point H. It can then be understood (2) that the free carriers near H are electrons whereas near K
they are holes. This has a great importance for all the transport
properties.
111.3 Comparison with recent calculations.
The
Korringa-Kohn-~ostoker
technique developed by Tatar and Rabii (TR)
(3) gives precise results in which many of the aspects seen before can
be recognized.
The general shapes of the E(k) curves are similar. The three dimensional
effect along the HKH line shows that, as expected, the degeneracy
disappears at point K, not at point H.
A difference between TR and Tight Binding results appears in a bands.
None of them is strictly k independent as would predict formula 14.
From the TR results it is possible to deduce (3), by identification,
some of the TI tight binding parameters encountered above
Yo
= -
2.92 eV
YI
= -
0,27 eV
Y2
=
0,022 eV
We observe that, as predicted by 10, the total width of the
about 61Y o I.
TI
band is
24
P. JOYES
A feature which is important for the optical absorption is the presence
of two peaks in the n.density of states. Their energies are symmetrical
with respect to EF and distant from EF by about Yo. One peak is in the
occupied part and the other in the empty part of the n density of states
so that electronic transitions can occur.
It is noteworthy that, when we calculate the density of state of finite
planar clusters which are fragments of the graphite plane, we obtain a
curve similar to the TR result, exhibiting in particular the two peaks.
The difference between TR results and figure 9 near E ~ E is due to
the presence of border atoms in the last case which have fess than
three neighbours.
Ep
Figure 9
Density of state of a graphite plane fragment with 96
atoms (6 x 6 rings along 2 perpendicular directions)
calculated by using hamiltonian 7. The discrete. structure
has been smoothed (see (18».
111.4 Comparison with experiment.
The experimental studies devoted to graphite are very numerous. Much
of them are directly related to electronic properties. Let us mention
a few examples.
.....
The angle resolved photoemission technique (4,5) provides the E(k)
curves, the agreement with TR results is satisfactory. In the De-Haas
van Alphen effect, the diamagnetic moment which appears when a graphite
sample is submitted to a magnetic field ~(directed along to) is measured. This moment exhibits oscillations when}t varies, with a period:
ON THE ELECTRONIC STRUCTURE OF GRAPHITE
25
ll(1) = ~
1:1
(16)
Vtctft:
F
where~ is an extremum of the cross-section of the Fermi surface
EF ) with a k z = constant plane. We have seen in chapter III. 2 that
(E::
.
the structure of the Fermi surface versus k z was complex, as a consequence in graphite various periods are observed which corresponds,
through 16, to various extrema of Jt F •
IV. ABSORPTION PROPERTIES.
The ~ level structure is also important for the study of the plasmon
energy due to ~ carriers. The plasmon frequency is given by :
4~n .. e
w;. = __
1""J__
2
1J
m
where n .. is an effective density of electrons defined by writing that
1J
dli
along direction i the derivative of the current
is given by
dt
where A. is the electric field along j.
J
The n 1J
.. values can be deduced from the TR band structure results.
They obtain (3)
;\w
a
0,46 eV
and
;\w
c
0,04 eV with
and
A value of 0.4 eV has been observed by Philipp (7). This point has also
been discussed by Draine et al. (8).
The reflectivity of graphite to light at normal incidence has been
measured. The reflectance curve shows a maximum at about 5 eV (9).
This peak is attributed (2,10,11) to the transition between the states
of maximum density in the ~ band from occupied to empty levels (see
chapter III.3) which satisfy the vertical transition requirement
(formula 5). A maximum of £2 (the imaginary part of the dielectric function) appears at this energy (10,11).
26
P.JOYES
Another maximum appears in the Im(~) energy loss function at about 7 eV
(9). Its position has been confirm~d by a recent calculation (12). It
is in good agreement with the electron energy loss measurements 6.8 eV
(13) or 7.2 eV (14). We can also mention the electron energy loss measurements of Fink et al. (15) on amorphous carbon film prepared by plasma
decomposition of benzene where a n plasmon peak appears at high annealing temperatures.
Physically the maxima of £2 and of
Im(~) = £2/(£~ +
£2)
are not of the
same nature. The second one appears in general at the plasmon 9xcitation
frequency for which £1 ~ 0 and £2 is small. Moreover, it is analytically
evident that maxima of £2 and of £2/(£1 +1£2) cannot occur at the same
time. There is another maximum of the Im(-) curve at about 25 eV which
£
is interpreted (9) as due to the excitation of the "cr + rr" plasmon.
Let us mention that according to Bohren and Huffmann (ref. 16 p. 467)
the strong 5.7 eV absorption by interstellar dust could be attributed
to graphite plasmon. It must be recalled that the strong visible absorptio~ by metallic cluster (for example Ag clusters with radius of about
50 A) is also attributed to plasmons, more precisely to small spheres
plasmons which occurs at £1 ~ - 2. The displacement of the plasmon
energy with size and shape has been studied (ref. 16, p. 373).
V. INFLUENCE OF DIELECTRONIC CORRELATIONS.
The effect of dielectronic correlations, which is important for diamond
(17), should also be discussed for graphite.
If we limit ourselves to intraatomic correlations we must add to the
hamiltonian 10 a term
U L: n ijt niH
ij
(17)
where nijt is :he number of t electrons on site ij and U is the intraatomic correlat~on energy (U > 0).
Various treatments can be applied. Among them, the variational Gutzwiller
technique is well adapted to study the relatively high U value which
appears for n electrons :
U
~
5.5 eV (I8).
The Gutzwiller treatment has first been proposed for an "s" band in a
cubic crystal. When the band is half-filled (Nt = N~ = N/2) the total
electronic energy per atom E can be deduced from the one electron energy EH by :
27
ON THE ELECTRONIC STRUCTURE OF GRAPHITE
H
U
U2
(18)
E = E + - + ---4
64EH
It is possible to see that one effect of the dielectronic correlation
is to narrow the electronic bands. Let us consider the excitation of an
electron. In a one-electron description, the initial total electronic
energy is NE? • Similarly in the final state this energy is
1
NE~ = NE~ + ~E/where ~E is the energy change of the excited electron.
By using 18, we obtain the energy excitation ~E' when correlation are
taken into account as :
~E'
l1E +
X
64N
[_1_ _ _1_]
E:
E~
~E'
(19)
We see, from 19, that the correction to ~E increases with ~E ; in other
words the width of the band is reduced. Though a precise calculation
has not yet been made, a similar effect can be expected for K electrons
in graphite.
The Gutzwiller method has been applied to the study of the magnetism
of K electrons in polyenes (18). It can be seen that, for some geometries,6-atom or 8-atom polyenes are magnetic (Nt ~ N+ in the fundamental state) ; some other shapes (as benzene) are non magnetic.
Another variational technique which allows one to take into account
spin waves, that is, to include the possibility of partial antiferromagnetism, has also been extended to aggregates (19).
The study of the ionization energy of relatively large polyenes (with
N ~ 34) (18) has also been achieved with a satisfactory agreement
with experiment.
CONCLUSION
This study of graphite has shown that some questions were well understood. However other aspects as the strong 5.7 eV absorption seem to
be controversal. Experimental and theoretical studies are still needed
in these fields.
REFERENCES
(I) Friedel J., Lannoo M., J. de Physique 34 (1973) 115.
(2) Haering R.R., Mrozowski S., Progress in Semiconductors (1960)
John Wiley & Sons, vol. 5, p. 273.
(3) Tatar R.C., Rabiis, Phys. Rev. B25 (1982) 4126.
P.J~YES
28
(4) Law A.R., Barry J.J •• Hughes H.P., Phys. Rev. B28 (1983) 5332.
(5) Marchand D., Fretigny C., Lagues M., Batallan F., Simon Ch.,
Rosennan I., Pinchaux R., Phys. Rev. B30 (1984) 4788
(6) Mc Clure J.W., Phys. Rev. 108 (1957) 612.
(7) Philipp H.R., Phys. Rev.
B~
(1977) 2896.
(8) Draine B.T., Hyung Mok Lee, The Astro. Journal, 285 (1984) 89.
(9) Taft E.A., Philipp H.R., Phys. Rev. B, 138 (1965) A197.
(10) Bassani F., Pastori-Parkavicini G., Nuovo Cimento 50B (1967) 96.
(II) Johnson L.G., Dresselhaus G., Phys. Rev. B,
(12) Chen N.X., Rabii S., Phys. Rev. B,
1l
Z
(1973) 2275.
(1985) 8242.
(13) Creuzburg M., Z.Phys. 191 (I9bb) 211.
(14) Killat U., J. Phys. C,
Z (1974)
2396.
(15) Fink J., Muller-Heizerling Th., Pfluger J., Scheerer B.,
Dischler B., Koidl P., Bubenzer A., Sah R.E., Phys. Rev. B, 30
(1984) 4713.
(16) Bohren C.F., Huffman D.R., Absorption and Scattering of Light by
Small Particles, John Wiley & Sons (1983).
(17) Horsch S., Horsch P., Fulde P., Phys. Rev. B29 (1984) 1870.
(18) Joyes P., Phys. Rev. B28 (1984) 4006.
(19) Joyes P., Phys. Rev. B32 (1985) 7356.
Question (A. Leger)
o
Is there a simple interpretation of the 2200 A absorption in terms of
the graphite band structure ?
Answer
One may interpret it by a one electron ~ + ~* transition which corresponds to the maximum of E2 which is about at 5 eV. It can also be due
to the excitation of a plasmon which, for the bulk material, corresponds
to a maximum of Im(1/E) at about tiwl> "v 7 eV. In finite media the plasmon
occurs at an energy ~F less than ~p (more precisely when E 1 (wF) "v - 2,
by using the E1 curve of ref. (9) one obtains ~ "v 5.5 eV).
The fact that we consider finite media is of particular importance since
the coupling of light and plasmon is better on small particles or rough
surfaces. For example photoyield enhancement have been observed in small
silver aggregates (Schmidt-Ott et al., Phys. Rev. Let. 45 (1980) 1284).
Theoretical study of this phenomenon have been publishe~(Inglesfield
J.E., Surf. Sci. 156 (1985) 830).
ON THE ELECfRONIC STRUcruRE OF GRAPHITE
29
Question (S. Leach)
The graphite particles in the interstellar medium are considered to be
electrically charged. What effect would this have on the optical properties you described for graphite ?
Answer
The effect of the charges will first depend on their localization. For
very small (N ~ 50-100) aggregates they may be localized on a central
atom even if the element is metallic in its bulk state (this phenomenon
in Hg2+ clusters is studied by C. Brechignac et al., Chem. Phys. Let. 0
188 (Y985) 174). When the size of the cluster is larger (radius R.t ·100 A)
the bulk behaviour appears and, for metallic clusters the charges are then
localized at the surface. Deformation and lattice parameter variations
can be induced by these surface charges as it is observed for larger
systems as dropplets. Moreover the charges create a perturbative potential
which displace the electronic bands.
Question (Tramer)
Do you expect important differences in optical properties between a
two-dimensional (simple plane) carbon cluster and three-dimensional
graphitic crystal ?
Answer
One must distinguish two regions.
(1) In the major part of the Brilloin zone the E(k) curves obtained
from the. two-dimensional descriptions are similar. The positions of
the density of states peaks are almost the same. In the KHM plane there
only occurs a splitting of the levels due to the fact that there are
four atoms (instead of two) in the unit cell.
(2) Near the HKH line, around which is centered the Fermi surface,
the difference between the two and three-dimensional representation
is important. In particular, in the second case there appears, at 0° K,
free carriers which would not exist in the first case.
Question (F. Pauzat)
Model implies limitation of carbon chain. If, as it seems, number of
atoms (if large enough) have no influence or little, how can the model
help to determine the structure of PAB.
On the other hand, how the assumption of curving the surface influences the
results, considering it is not really possible to curve this surface
while taking into account only hexagonal elementary structures.
30
P. JOYES
Answer
The electronic properties in N-atom fragments of the graphite plane
tend rather rapidly to the graphite two dimensional properties. For
example, the total electronic energy per atom is 1.33[Yol for N = 6
(benzene), 1.476 (N = 96), 1.57 for the graphite plane.
The Born Von &arman conditions are geometrically impossible in two or
three dimensions. The allowed curvature realized in regular polyhedra
cannot be studied by using the Born Von Karman conditions. Indeed,they
imply the existence of NI (N2) cells in each direction and therefore
of(N I x N2 ) cells which is not realized in regular polyhedra.
VlUUOOS KINDS OF SOLID CARBON:
And~
STRtrC"l'ORE AND OP'l'ICAL PROPERl'IES
MARCHAND
de Bordeaux I
Centre Paul Pascal (CNRS)
Oomaine Universitaire 33405 Talence, France
Universit~
ABSTRACT. Graphitic carbons are considered from two complementary
points of view:
as imperfect forms of graphite and also as very
large aromatic molecules.
The various existing materials are
presented
and
a
short
overview
of the carbonization and
graphitization processes is given, with an emphasis on the defects
Which may arise from the finite size of the crystallites, from the
increased distance between the carbon layers, and from their
non-planarity.
The known data concerning the optical spectra of
carbon materials are then reviewed, and suggestions are presented
for the spectra of very small graphitic or aromatic grains.
1.
INTRODOCTION
The crystalline forms of carbon (diamond where the C atoms are sp3
hybridized and graphite corresponding to· Sp~) are well known. I t is
less common knowledge that crystal forms corresponding to the sp
hybridization have been described: these are the "carbynes" (1).
But some alJthors still consider their existence as doubtful.
However, as neither diamond nor carbynes seem to qualify as
important components of interstellar dust, we will be interested in
sp~ hybridized carbon only.
But since it is
evident
that
interstellar dust grains cannot be perfect graphite crystals (i.e.
large size particles), we must focus .our interest on "graphitic
carbons":
solid carbon particles constitlJted of condensed aromatic
rings, with a more or less "graphite-like" structure.
These graphitic carbons are a class of solids of unusual
diversity,
including both industrial products manufactured by
millions of tons and laboratory samples. Since all of them have a
graphite-like structtJre, they may be considered as more or less
imperfect forms of graphite, containing various kinds or various
concentrations of structural defects. Their electronic energy
31
A. Leger et al. (eds.) , Polycyclic Aromatic Hydrocarbons and Astrophysics, 3/-54.
© 1987 by D. Reidel Publishing Company.
A.MARCHAND
32
levels structure and the vibration modes of their lattices are
consequently more or less similar to those of graphite: the
reslJlting optical properties can be studied as modifications of the
graphite properties.
This is what we will do in the following
pages, and this is the reason why it is necessary first to review
briefly the crystal and electronic structl.lreS of graphite.
But the graphitic carbons may also be considered as extremely large
aromatic molecules, and their electronic structlJre and optical
properties can be extrapolated from those of smaller condensed rings
hydrocarbons.
This is the reason why we will also begin with a
short review of the electronic structure of aromatic molecules.
1.1.
Graphite
Fi911re 1 presents the well known crystal lattice of hexagonal
graphite. Two layer planes are shown and it must be remembered that
there are two non-equivalent sites for carbon atoms:
each A atom
has close neighbours directly above and below itself in the adjacent
planes. The B sites are different Since they are located above and
below the centers of the hexagons of the neighbouring planes. This
situation is a cons~Jence of the relative poSitions of adjacent
planes as shown in Fi911re 21 each plane is shifted relative to the
next one by 1.42 Aalong the AS direction, but the shifts are
alternately
forward
and
backward,
so
that the successive
translations of the planes can be represented by the "stacking
sequence" ASABABA.
We will see later that the "rhomboedral"
modification is characterized by successive shifts which are always
forward, so that the "stacking sequence" is ABCABCABCA.
The Brillouin zone of hexagonal graphite is shown in reciprocal
space on Figure 3.
There are just enough 'Tt electrons to fill
completely one zone, and that should reslJlt in a full valence band
and an empty conduction band. But there is no energy gap between
the bands. Actually they are even overlapping along the Hrc:H zone
corner, so that there are some electrons in the conduction band and
some holes in the valence band in the vicinity of HKH (it must also
be remembered that all HKH corners are equivalent). The detailed
electronic energy levels strucbJre around HKH was worked out by
Slonczewski and WeiSS (13) and is illustrated by Fig1lre 4.
The conduction and valence bands have the same energy E3 all
HKH, but there is also another conduction band at energy
The central part of
E.> E3 and another valence band at E2< E 3 .
Figure 4 shows the variation of the "1"C electron energy in the
conduction band (E , and E3 ) and in the valence band (E~ and E3)
a.long HKH, as a f1.lnction of the wave vector ~. The left side of
Figure 4 shows, in the neighbo1lrhood of K (~= 0), the energy of the
valence band holes as a function of the wave vector 0- perpendicular
to fOOL The right Side shows, in the vicinity of H (~=1: 0.5) the
energy of the conduction band electrons as a function of (J.
along
VARIOUS KINDS OF SOLID CARBON
33
z
y
Figure
Figure
2:
1:
Graphite crystal lattice
Relative positions of graphitic layers
HOLES
ELECTRONS
HOLES
/
Figure
3:
/
Brillouin zone of graphite
34
A.MARCHAND
Figure 4: Slonczewski-Weiss model (13):
energy near the BKH zone corner
variations of the Tt
bands
~~\,
v, \
y,'
:
'" \
3,
,
V I
r
\
1
1
1
\
\
A
't"
:
:
:
:
I
A'
,
I
"
I'Y2
V'
'1'
'V
I '"
\
"
'
,
"
,,
~ A3\~,
:} ~:
~/
B~.
A
B
A
A --,,--
Figure
5:
B
A
B
Interaction parameters between C atoms (s.W.
Graph it..
Figure
6:
Density of states in the
model)
GraphitEI'
~
band of graphite
VARIOUS KINDS OF SOLID CARBON
35
'it
The parameters
are a measure of the various interactions
between neighbouring C atoms in the lattice, as illustrated by
Figure 5: the interaction between first neighbours A and B in the
same layer is represented by 1., and the interaction between first
neighbours A in adjacent layers is measl..lred by '61 ' Fi91.1re 4 shows
that parameter Yt is responsible for the overlap of the valence and
conduction bandS.
A general pictl.1re of the energy levels in the whole "It band
(and not only in the vicinity of HKR) is given on Fi91Jre 6, Which
shows the density of states N(E) as a function of the ~ electrons
energy: the energy difference between the two maxima of the density
of states was formerly estimated at about
5
eV
and
the
Slonczewski-weiss model gives now the value 2ta'
1.2. Aromatic Molecules
A considerable amount of theoretical work has been devoted to the
determination of the ~ energy levels of these large molecules:
there are minor differences in the detailed results from various
authors, but the general picture is quite clear and widely accepted.
As an example (19), Figure 7a shows the energy levels of a molecule
2.3 x 2.9 run conta.ining 60 C atoms. The energy scale is in ~ units
(~ is the resonance integral between two neighbours:
it can be
identified
with the 'fa parameter of graphite, and its value
estimated at 2.5-3 eV). The number of levels at each energy value
is represented by the length of the corresponding vertical segment.
The lower (negative) energy levels are fully occupied and the upper
(positive) levels are elr,pty at 0 K: this situation is similar to
the valence and conduction bands of graphite. The total width of
the bands is 6 ~ and the average separation of the levels is
6 ~ /60 tV 0.25-0.30 eV, Which is 1III.Ich larger than kT.
Two other features of this energy structure are noteworthy.
l) The most highly degenerate levels are exactly located at E =± ~:
it is a general characteristics of the electronic structure of all
these molecules, Whatever their size, so that the energy difference
between these levels is always 2 ~ .
2) There is a large gap ( 1 . 04 ~ wide) between the top of the
occupied levels ("valence band") and the lowest empty levels (bottom
of the "conduction band").
Figure 7b shows similarly the computed electroniC structure
(19) for a larger molecule, 4.6 x 4.6 run, containing 200 atoms. In
this case the levels are very closely paCked, and a better
presentation consists in drawing a denSity of states histogram,
where the number of levels in each O. 2 ~ interval is shown as a
function of energy.
The "valence" and "conduction bands" are
clearly visible: their total width (6~ ) is the same as in the
36
A.MARCHAND
smaller
but the average level
separation
is
now
eV, the same order of magnitude as kT at
density of states is again located at E =1'(3,
but the energy gap between the bands is reduced to 0.6(3.
61'>
molecule,
/200rv 0.07-0.09
1000 Je::.
The maximum
This trend goes on as the size of the molecules increases: the
energy
separation of levels is smaller and smaller so that
continuous energy bands are formed, the position of the denSity of
states peaks is always tf ' and the gap decreases progressively to
zero.
2.
A SHORT PRESENTATION OF THE GRAPHITIC CARBON
~TERIALS
With the exception of a few graphite depoSits, the only natural
forms of graphitic carbons are the various kinds of coals. However
they are not relevant in the present context because their purity is
usually quite low ( they contain many other elements than carbon)
and their structures are generally quite removed from that of
graphite (high concentrations of aliphatic chains and small aromatic
units).
We must then focus our attention on the artificial carbonaceous
solids manufactured by laboratories and industries.
These are
nearly always products of the pyrolysis and thermal treatment of
organic sl.lbstances.
They may be listed according to the nature of
their organiC precursor and the value of their heat-treatment
temperat1lre (H. T . T . for short).
Solid or liqtJid precursors: their pyrolysis or heat-treatment
yields cokes (more or less graphitizable by high temperature
treatment), as well as fibers or glass-like carbons Which are
usually non-graphitizable (although their crystal structure can be
improved by heat-treatment).
Gaseous
precursors:
their pyrolysis at lower
temperatures
yields thin carbon films, fibers, pyrolytic carbons
( "pyrocarbons") or ca.rbon blacks.
At much higher temperatures
(T~ 2000·C) the pyrolysis products are pyrocarbons with a structure
qtlite close to that of graphite.
(T~ 1000·C)
Now interstellar dust grains are
certainly
very
small
particles, While in these solid carbon materials one dimension at
least is usually "macroscopic":
fibers for instance consist of
filaments a few micrometers or tens of micrometers in diameter, but
their length is at least 10 em. Thin films and carbon blacks are
the only materials with sufficiently "microscopiC" dimenSions that
can be comparable to those of interstellar dust grains. All carbon
materials of course may be subjected to grinding and reduced to
microscopic grains.
But grinding does not amount only to a
reduction in dimensions: it results in an alteration of the
37
VARIOUS KINDS OF SOLID CARBON
structure by introducing various defects,
specific products of the grinding process.
some
of
them
being
We will then examine
the structlAral characteristics of all the
carbonaceous solids, with a particular emphasis on the manner they
differ from graphite and on the various types of crystal defects
which are present. But we will have a particular interest in ground
graphite, carbon blacks, and thin films, and we will look for
correlations between structural defects and the optical properties
of the corresponding materials.
It is convenient, at the beginning of this study, to describe the
transformations
which
take
place all along the process of
graphitization of a carbon:
the progressive ordering of its
structure
with
higher
and higher heat-treatment, which may
eventually lead to the quasi-perfect graphite crystal.
3.
THE BUILDING OF GRAPHITIC STRUCTURES
Only recently did we aCqtJire a reliable and general (although
still quite incomplete) model of the way graphitic structlJreS are
built. We will sl.lJlllla.rize it briefly, without giving any description
of the many studies which led to OlJr present understanding.
Temperature is the main driving force for these processes, although
other factors (time, pressure, etc ..... ) must also be considered.
The building blocks of all carbon materials are small groups of
condensed aromatic rings.
- BELOW 1000-1100o C (CARBONIZATION RANGE) the carbon materials can
still be considered as molecular solids, with a large proportion of
heteroatoms and non-aromatic C-C bonds.
Their
structure
is
characterized (2) by "structural units" (S.U.), which are molecules
constituted of less than 10-12 condensed aromatic rings: their size
is in the order of 9-10 Angstrom units and they are stacked by two
or three. There is no general organization of these units, except a
possible
preferred
orientation,
usually local but sometimes
extending more or less through the whole solid.
The effect of
increased
heat
treatment
temperatures (B. T. T.) is mainly a
progressive "purification" due to the loss of heteroatoms.
carbons manufactured from a gas phase in this temperature range
have the same dharacteristics.
- IN THE 1000-1600· C RANGE (PRE-GRAPHITIZATION) there are very few
heteroatoms left and most of them are prObably bound laterally to
the aromatic molecules.
The S. u.
are thus able to
orient
themselves parallel to one another and to build columnar structures
like stacks of plates, but these stacks are bent and very irregular
becalJSe they are constituted of units of slightly different shapes
A.MARCHAND
38
and diameters. The stack height is also limited since there are
still some disoriented units, but the number of aromatic layers
associated in such parallel stacking increases with H.T.T.
as the
proportion of disoriented units decreases.
This is shown on
Fi~lre 8, which illustrates the graphitization of carbon films (3).
- GRAPHITIZATION OCCURS BETWEEN l6000 AND 3000·C IN TWO STEPS:
a
two-dimensional
growth
from
1600P C
to
2000-2l00·C, and a
three-dimensional organization from 2100 0 C upwards.
The so-called
''hard
carbons"
or noh-graphitizing carbons deviate from the
graphitizing behavior at this stage:
they do not acquire the
complete three-dimensional lattice of graphite even at the highest
treatment temperatures.
The two-dimensional growth of the aromatic systems results from
the gradual removal of the various defects (heteroatoms, sp· bonds,
twists and dislocations) located at the periphery of the structural
units.
This allows the coalescence of the columnar aSSOCiations,
which then constitute very large "wrinkled" parallel sheets of
aromatic carbons (Fig 8c). The continuing elimination of defects
progressively rubs out the S.U. boundaries, and very extended rigid
two-dimensional parallel graphitic layers (for which the word
"graphenes" has been suggested) are obtained
at
2000-2100° C
(Fig 8d).
Pyrolytic carbons deposited near 2l00o C also
show
this
two-dimensional structure built from stacked parallel graphenes,
randomly shifted or rotated relative to one another.
The final step consists in ordering this stacking into the
AB1.BAB... sequence of hexagonal graphite. There are hardly
any defect left in the solid to prevent the necessary shifts of the
graphenes relative to one another, PROVIDED ALL GRAPHEMES ~CKS
HAVE THE SAME ORIENTATION: non-graphitizing carbons usually do not
meet this condition because of either a remaining mtcroporosity
(where the graphitic layers are aligned parallel to the pore walls),
or too small grain sizes (e.g. carbon blacks).
re~llar
4.
THE WJUOUS KINDS OF STRUCTURAL DEFECTS
From this description ~f the building process of graphitic
structures and from Fi~lre 8, it is possible to review the defects
which may characterize imperfect crystals in carbonaceous solids.
4.1. Finite size of the graphite crystals ("crystallites")
Well graphitized materials have crystallite sizes which may extend
to a few micrometers, but lower graphitization stages (lower R.T.T.)
correspond to smaller and smaller crystallites (3-5 nm for some
carbon blacks), and the structural units (s.u.) of the carbonized
39
VARIOUS KINDS OF SOLID CARBON
200 aloms
20
.
-0.6 (3
10
-.;
~ Or--r-------r------,---~~T_~--~~----~------~~
c.o
...u
..a
~ 10
z
60 aloms
.1.04 (3
5
•
E/(3
Figure 7:
ref. 19)
Tr energy levels of two large
aromatic
molecules
(after
rigid
plane
layer.
s.u.
1Goo
HTT(OC)
Figure 81 Progressive building of graphitic layers with
from the structural units (s.U.) to the graphite crystal
R.T.T.:
40
A. MARCHAND
organic materials (H.T.T . .(lOOOoC), no larger than 1 nm, are
large aromatic moleclAles rather than graphitic crystallites.
really
Since a carbon atom with sp~ hybridization, if located at the
periphery of a crystallite, cannot be linked to three other carbon
atoms, we may assume that crystallites (or aromatic molecules) are
bounded by C-H groups, or C atoms with dangling or double bonds, or
C atoms carrying non-bonding electron pairs.
The smaller the
crysta.llites, the larger the proportion of CH or unsaturated C
atoms.
Corresponding modifications of the electron band structure must
be expected, particularly in the neighbourhood of the Brillouin zone
corner HI<H (Fig. 3 and 4) because the boundary C atoms with
dangling bonds act as traps for the electrons and remove a number of
them from the 1t' valence band, thus lowering the Fermi level.
For very small crystallite sizes the energy level separation in
the bands becomes large enough to be comparable to JeT (0.03-0. 10 eV
between room temperature and 1000·C) and the levels can no longer be
considered as constituting continuous bands.
This happens for
crystallites containing about a hundred C atoms or less I
their
electronic structure is similar to that of large aromatic molecules
(see 1.2.).
4.2. Increased distance
(doo~)
between the graphitic layers
In imperfectly graphitized materials the d.o & distance is always
larger than the perfect graphite crystal value of 0.3354 nm and the
distance increases with increaSing imperfection. It means that the
attractive Van der Waals interaction between the layers becomes
progressively weaker ana weaker, until eventually each graphitic
layer must be conSidered as an independent two-dimensional molecule
(it would be called a "graphene" if its size was infinite).
IncreaSing doo& values thus correspond to the transformation
from the three-dimensional graphitic structure to a two-dimenSional
organization. Each layer becomes independent from its neighbours
and any geometriC correlation of orientation or position between
adjacent layers disappears: the layers can be rotated or shifted
arbitrarily
relative to one another.
This is the so-called
"turbostratic structure".
Simultaneously carbon atoms or various iJnpuJ;"ities (particularly
hydrogen) may be able to intercalate between the layers which are no
longer in close interaction. It has been shown theoretically that a
preferential site for these interstitial atoms is above or below the
center of an aromatic ring, with which they are linked by a kind of
triple bond (sp hybridization of interstitial C).
41
V ARIOUS KINDS OF SOLID CARBON
A particular feature of the progressive loss of the graphite
three-dimensional order is the possible creation of "rhomboedra1
defects" in the layer stacking sequence. The rhomboedra1 graphite
structure differs from the hexagonal one by the sequence of
translations of each layer relative to its neighbour:
instead of
the normal (hexagonal) sequence ABP.BABA, the rhomboedral stacking is
lUlCABCABCA (Fig 9).
When interactions between layers are weakened, rhomboedral
stacking faults are likely to happen more and more often. They grow
particularly frequent when d oo 2. exceeds 0.338 nm. Some methods of
graphite grinding (with a rotating blade) have been shown to
generate up to 40% of rhomboedra1 sequences in graphite.
The
rhomboedral order corresponds to an electronic band structure quite
different from the hexagonal one and reslJ1ts in
spectacular
modifications of some electronic properties (e.g. diamagnetism).
Rhomboedra1 defects of course eventually merge into the general
three-dimensional
disorder
of turbostratic carbons when d.o~
increases further.
The progressive decrease of interactions between neighbolAring
layers
must
be
translated in terms of a decrease of the
tt parameters (Fig. 5) in the electronic band model. Figure 4
shows clearly that the vanishing of
reslJ1ts in the disappearance
of the overlap of the valence and conduction bands, and that the two
conduction bands (E. and E 3 ) on one hand and the two valence bands
(E2, and E) ) on the other hand merge When
decreases to zero.
Moreover the A parameter (in Fig.
4) is a consequence of the
existence of two non-equivalent Sites A and B (Fig. 1 and 2) in the
three-dimensional
structure of graphite:
this parameter also
vanishes in the two-dimensional "graphene".
t2.
t.
4. 3. Non planar carbon layers
Figure 8 shows that in imperfectly graphitized carbons the aromatiC
structural units (S.U.) coalesce into extended carbon layers. But
these layers are not real planes (true "graphenes" ) at
the
beginning:
in the l600-2000·C heat-treatment range they are rather
similar to wrinkled sheets of paper. The curved or folded regions
corresponding to these wrinkles probably contain a high percentage
of s~ hybridized (diamond-like) C atoms, as well as interstitial
atoms.
Moreover all of these may act as electron traps and
consequently lower the Fermi level in the 11: valence band.
4.4. Summary
After this brief review and from What we know of the electronic
structure, we can sUJl'lllarize the different types of defects Which are
likely to exist in imperfectly graphitized materials.
42
A. MARCHAND
A
B
A
B
A
B
• •• •• ••
•• •• •• ••
• •• •• ••
• • •• •• • •
• •• • • • •
•• •• •• ••
•• • • • • • •
•• •• •• ••
•• •• •• •
•• •• •• •• •
• • • • •• • •
•• •• •• • •
9:
B
C
A
B
C
Rhomboedral
Hexagonal
Fiqure
A
Hexagonal layer stacking and rhomboedra1 layer stacking
I.OO....----,---,---...,----,---,----,----,-----,.---,----r---,---,---,
.90
.80
.70
.60
w
u
:i
....
.50
o
w
~ .40
w
Ir
.30
.20
.10
o
26
WAVE ENERGY,.,
Fiqure 10:
Transverse reflectance spectrum of graphite (6)
VARIOUS KINDS OF SOLID CARBON
43
decreased interactions between layers ( 1. -. 0)
vanishing overlap between the bands (t~ --+ 0)
possibility of an energy gap at the zone corner
possibility of rhomboedral stacking sequences
presence of heteroatoms (H mainly)
presence of spl (diamond-like) carbon C atoms
- possibility of sp hybridized interstitial C
- Fermi level lowered below the Brillouin zone corner
- discrete energy levels instead of energy bands
-
5.
OPTICAL PROPERTIES OF GRAPHITE AND CARBONS
can now describe the optical properties of graphite and their
modifications
in
the
other forms of solid carbons:
these
modifications can be discussed in terms of the various types of
defects.
We
5.1. Graphite
It must be pointed out first that nearly all the measurements were
performed
with the electric field of the incident radiation
perpendicular to the c-axis of the graphite lattice: graphite is a
highly anisotropic uniaxial crystal, but we know very little about
the anisotropy of its optical properties.
5.1.1. Visible and Ultra-Violet. The behaviour of graphite in the
visible and UV regions of the spectrum is known from studies of the
electron energy loss, reflectance, and absorption by thin flakes of
single crystals.
The electron energy loss and reflectance data (4,6,7,8),
illustrated by Figure 10, show that the imaginary part of the
dielectric constant presents a strong peak near 5 eV and another one
in the 12-15 eV zone. They seem to be in reasonable agreement with
ab initio calculations of the optical spectrum by Chen
and
Rabii (9).
The high energy peak is interpreted as a 0- -+ cr*
transition betwen the occupied and empty levels of the 0- band.
The low energy peak (5 ev) was shown (4) to correspond to a sharp
maximum of the extinction coefficient k at 260 nm. This maximwn is
also ~Jite prominent in the absorption spectra of thin flakes of
single crystals which were published by Ergun and coworkers (10, 11)
for the UV and visible region (200-600 nm)(Fig.11}.
Since 5 eV was then considered to be approximately the energy
difference between the maxima of the density of states in the
-n= valence and conduction bands of graphite (Fig 6), Humphreys-<Men
suggested that the 260 nm absorption peak originates from this
'It -+1t*transition.
This interpretation was confirmed by the later
work of Johnson and Dresselhaus (12) on the electronic structure and
optical properties of graphite: they show that the theoretical
A,MARCHAND
44
energy difference calculated from the Slonczewski-Weiss model (13)
is indeed 2
(Fig 6 and 7b), but that, although the best generally
admitted value for '60 is about 3 ev, when interactions from more
distant neighbours are included the value ~= 2.3 eV is consistent
with the optical data. Hence a transition at 4.6 eV (270 run).
to
5.1.2. Infra-Red. The investigations in the infra-red domain include
both reflectance (6,14,15) and absorption (16) studies.
The absorption results (Fig 12) show an absorption maximum near
1.5 tlm (0.8 eV). This is consistent with a transition between the 'It
valence and conduction bands near the J( point of the Brillouin zone
edge, where the bands overlap in the Slonczewski-Weiss model
(Fig 13): 0.8 eV is the expected value of 2 '(I .
The study by Nemanich, Lucovsky and Solin ( 14) of
the
reflectivity of HOPG (highly oriented pyrolytic graphite) extends
from 500 to 4000 em -I (2.5 to 20 Ilm).
Both polarizat ions of the
electric field direction E (parallel and perpendicular to the
c-axis) were investigated (Fig 14). The E.L C curve shows a regular
decrease of the reflectance with increasing frequency (as seen also
in the reslJlts of Taft and Philipp - extreme left side of figure
10).
Two peaks are visible in Fi~Jre 14: they have been aSSigned
to the fr~Jencies of the only zone center optic modes of graphite
(Fig 15) which are infra-red active, namely 1588 em -I for mode EllA.
and 868 em-I for mode A 2 L1.'
But it is important to point out that these modes are infra-red
active only in the three-dimensional structure of graphite, because
of the non-equivalence of the atomic sites A and B in the lattice.
They are inactive in a Single two-dimensional layer ("graphene").
These two fr~Jencies ar.e then expected to disappear when the
interaction between layers vanishes ( 'tl --+ 0) and should be absent
from the spectra of very poorly graphitized materials.
Similarly
the 1.5 I'm absorption maximum (0.8 eV = 2 '(I) should shift to longer
wave-lengths and disappear when ~I--+ O.
5.2. Coals
The reslJlts plJblished by Ergun, Mccartney, and Walline (10) on the
absorption of ultra-thin sections of coal vitrinites between 200 and
600 run (2-6 ev) show (Fig 16) a general variation of: the extinction
coefficient very similar to that of graphite, but the extinction
coefficient tends to decrease with decreaSing rank of the coal.
Moreover the characteristic graphite peak near 250 run is replaced by
a much weaker sholJlder, so weak indeed that its exact position is
quite difficlJlt to determine.
Since that peak is interpreted in
terms of valence to conduction '1(' band transition ( see above), it
might be concluded that the electronic structure of these coals
cannot be represented in an energy band scheme, and that we are
dealing in this case with smaller-size molecules. It is well known
45
VARIOUS KINDS OF SOLID CARBON
20,-----,-----r-----r-----r----,r--·
1.9
18
17
~'
6
U
0:
-1!
~ 1.5
8
z
o
~
u
I
1.4
z
\;
!oiJ
l
1.3
1.2
I
i
---j
II
I
10
WAVE ENERGY.
_,000
I
4,000
~V.
[
3,000
WA .... E LENGTH, Antstroms
2,000
Figure 11: Extinction
coefficient
of
graphite
obtained
transmission measurements with ultrathin flakes (10,11)
WAVELENGTH,
from
~
5040
e
I
,O~.~25~--~0~50~----~~----~~----~~----~~------~1.75
WAVE ENERGY. ev
Figure 12: Reciprocal transmittance of ultrathin flakes
in the infra-red spectrum (16)
of
graphite
A.MARCHAND
46
H
K
H
wove
Figure 13: Transitions between the valence and
near
the
K
point
of
the
Brillouin
E,- EJ,-v Ei- E z r.J2
1
'(I
conduction
zone
of
(15)
highl Yoriented
pyrolyLic graphite
(0)
0,8
E.lC
0.6
.. 0.40.2
<II
u
c:
0
u
EIIC
<II
<::
<II
a:
0
0,25
0.20
2000
850
Figure 14:
900
3000
E.lC
1525
0)5
1600
Frequency (cm-1)
Reflectance of H.O.P.G.
in the infra-red (14)
bands
graphite:
~
VARIOUS KINDS OF SOLID CARBON
47
~
r+hl
.:±b
I
I
I
~
E,u
A2u
868 cm-1
Figure 15:
1588
cm-1
Infra-red active phonon modes of graphite
E
~180
~
160
140
1
2
3
4
5
6
1
8
9
Graphite
Anthracite
Semi-anthracite
Low-volatile bituminous
Medium -volatile bituminous
High-volatile A bituminous
High-volatile C bituminous
Sub-bituminous B
Lignite
u
1
60
40
20
0
2000
Figure 16: Ultra-violet
vitrinites (10 )
3000
4000
Wavelength
absorption
5000
by
6000
thin
a
slices
of
coal
A.MARCHAND
48
however that aromatic molecules, whatever their number of condensed
rings, do absorb light quite strongly in the 5 eV region, and the
weaker shoulder in the coal spectra may be only an effect of
dilution of such molecules in a non-absorbing medium. It is then
possible that very small graphite crystallites (or very large
aromatic molecules) still have a strong absorption band in the 5 eV
or 250 nm region. Such a possibility also seems very likely from
other experimental evidence and from theoretical considerations (see
the levels of aromatic molecules in Figure 7 for instance, where the
energy difference between the density of states maxima is 21! ' very
close to 5 eV).
5.3.
Ground Graphite
The infra-red absorption of ground graphite was studied by Friedel
and Carlson (17), who found strong and wide bands at 1360 and
1587 em-I, and weak ones at 830 and 2200 em-I.
The 1587 em-I
frequency is obviously the E,~ infra-red active vibration observed
by Nemanich et al. (see above), and the weak bands may originate
from small molecular fragments (aromatic CH at 830 em-I? cac or
C SN at 2200 em-I?) created by the grinding process.
But the
1360 em-I strong absorption is more interesting because it is a well
known Raman frequency of :imperfectly graphitized carbons. There is
no doubt that in a finite size graphite lattice the selection rules
are somewhat relaxed: instead of being limited to the zone center
modes, the observed frequencies may include those of high denSity of
states of phonons. The 1350 em-I zone (7.4 ""m) characterizes the
Raman spectrum of small graphite crystallites and an inverse
correlation has been established between the intenSity of this Raman
band and the crystallite size. For very finely ground graphite, it
should be expected that only the 1350 em-' band should remain, since
all "normal" infra-red modes become inactive when the interactions
between layers vanish (see above)
It was Observed also that a decrease of the crystallite size is
associated with a gradual (up to 20 em-I) sh~ft of the main
1580 em-I band towards higher freqlJencies (A decreasing from 6.3 to
6.2 rm).
5.4.
Chars and Carbon Blacks
Infra-red spectra from charred organic material and carbon blacks
are characterized by a continuous absorption which is generally
increasing with increasing wave-number. It fits with the visible
and ultra-violet trend shown on figure 16. In addition to this
background, various peaks may be observed, which characterize the
chemical groups and bonds present in the material (partic1llarly on
its surface because of the strong absorption of the incident
radiation).
One should remember that the solids are constituted of
structural aromatic .mits (S.U.) of 10-12 condensed rings, with
numerous interstitial C atoms and heteroatoms (mainly H) on their
VARIOUS KINDS OF SOLID CARBON
49
periphery. Trigonal bonding (Spt. hybridization) is still the rule,
but a fraction of diamond-like (Spl) structures must be expected,
especially at the boundaries between units.
Figure 17 shows an example of this kind of spectra, with a
number of strong "aromatic CB bending" peaks in the 700-900 em-I
(ll-14fUD) area, CH a and CHI aliphatic bands near 1400 em-I (7rm),
the characteristic aromatic C-c peak at 1600 em-I (6.25 fU"), an the
aromatic C-B stretching frequencies near 3000 em-I (3.3
We may
observe incidentally that, should the graphite infra-red modes at
868 and 1588 em-' still be active, they would be completely lost in
the "aromatic background" of the 800-900 em-I and 1600 em-I domains.
rm)'
5.5. "Amorphous" Carbon Films
Carbon films or "amorphous carbon films" is a name which deSignates
a great variety of materials, obtained by quite different methods
(thermal vaporization of graphite or carbons, CVD or plasma-assisted
CVD of gaseous hydrocarbons at various temperatures or pressures,
ion beam sputtering, etc .... ). Moreover these films may have been
subsequently heat-treated at various temperatures (H.T.T.), and they
are not really amorphous, even in the "as-deposited" state.
They
are constituted of aromatic structural units (s.u.) which may have a
preferred orientation parallel to the substrate surface, and the
aroma ~ic planes in the same unit are roughly parallel to each other
("turbostratic structure"). From their manufacturing processes as
well as from their structure, these films are the carbon variety
which is most likely to resemble interstellar dust.
Because of important possible uses (infra-red transparency,
electrical insulation, hardness, and chemical resistance to acids
and bases), they have been subjected recently to numerous optical
studies. We will summarize here the main findings.
5.5.1. Infra-Red. The infra-red spectra are similar to those of
other "low-temperature" carbon materials such as chars and carbon
blacks
( see
5 . 4.
above) .
The
high-frequency
region
(2700-3300 em-') of the spectra of benzene plasma-deposited films
was very carefully analyzed (20) in order to examine the relative
contributions of aromatic and aliphatic C-B: it was concluded that
two-thirds of the carbon atoms in such "as-deposited" films are
sp~ - hybridized (diamond-like) and 2% are sp (triple-bonded), while
the
sp'l
(trigonal)
fraction
increases
with
subsequent
heat-treatment.
There does not seem to be any evidence in the published spectra
of an absorption line in the 1350 em-' (7.41m) zone, which would
correspond to the "small crystallite" band observed in the Raman
spectra of imperfectly graphitized materials and in the infra-red
spectrum of ground graphite (17). It shol)ld be pointed out however
that this region of the spectrum does not seem to have drawn much
A.MARCHAND
50
attention from anyone
Another characteristic feature of the infra-red spectra of
these carbon films is the existence of an energy gap between the
occupied and empty 11:' levels, as could be expected from the
theoretical study of large aroDlatic molecules (see Fig. 7 ) . An
absorption threshold was observed at 0.2 eV for evaporated carbon
films (21), and the optical gap of benzene plasma~eposited films
decreases from about I or 2 eV in the as~eposited state to zero by
heat-treatment at 600· C (18).
5.5.2. u~tra-Violet. The energy loss function or the imaginary part
£&of the die~ectric constant was obtained from electron energy loss
studies of plasma~eposi ted carbon films (18), and Figure 18 shows
its variations up to 30 eV. The same kind of curve was found from a
Kramers-JCronig analysis of the reflectance spectrwn of graphite
crystals ( see Fig. 10) by Taft and Philipp (6). TWo peaks are
visible on Figure 18, Which are interpreted as r-. U-*transitions
bet:ween the valence and conduction tr bands (hig1:) energy peak) and
"1t ... 1('" transitions between the '1'C valence and conduction bands ( low
energy peak). In graphite, the 1t ..... '1l.1P' transi.tion corresponds to an
energy difference of about 5 eV (0.25 tun) as pointed out: earlier,
but it is clear from Figure 18 that in some carbon films (#= 1 and
2), the peak is located at a slightly higher energy (6-6.5 ev).
FUrther studies of these same films (18) have shown that the peak
poSition moves slightly wi.th heat-treatment,
its
wave-length
shifting up eventually to the 250 nm value of graphite. Theoretical
considerations (12) had already suggested, and this result thus
confirms that, although all a~tic structures, including graphite
and all forms of graphitic carbons, show a strong absorption at
approximately
(1.e.
about 6 eV or 0.21-0.22 flm), the
absorption maximum· exact position is a fUDction of the interactions
between more distant neighbouring carbon atoms.
2'1.'
6.
CONCLUSION: THE OP'l'ICAL SPECTRUM OF A SINGLE STRUC'l'ORAL UNIT ?
All the known carbon materials have dime~ions much larger than the
hypothetiC interstellar dust grain of 5 A diameter containing about
50 C atoms. However this grain dimension is about the same as that
of the a.J:Omatic structural units (see Fig. 8) Which are the building
blocks (10-12 condensed aromatic rings and size of 8-10 1) of all
materials in the "carbonization range" (H.T.T.
1000 C). These
S. U. unfort'lnately have not been isolated for a study of their
optical properties, but: it would probably be worthWhile to thinJc
about a way of obtaining their spectra.
Anyway we may try and
j.magine what these cO'lld look like.
The expected defects in the graphitic structure would be:
a) the small size of the structure and consequently
- discrete electronic energy levels
- an energy gap betwen the occupied and empty levels
51
VARIOUS KINDS OF SOLID CARBON
oram.
CH
t
err
CH3
oram. CH
4000
Figure 17:
500·(: (22)
CH
ora mot .
..
wavenumber
Infra-red spectrum of a naphtalene pitch
heat-treated
at
...
w
ENERGY leVI
Figure 18: Plasma-deposi ted carbon films I
imaginary part of the
dielectric
function.
samples
1,2,3,4 were deposited at room
temperature with increasing kinetic energy of the ions impinging on
the sUbstrate (18)
52
A.MARCHAND
b) vanishing interactions between carbon layers, resulting in
- near-zero values for A and all '(t parameters except
- two-dimensional behaviour of the graphitic lattice
e
c) a large proportion of spl - hybridized C atoms
d) the possible presence of numerous H atoms at the periphery
t
These spectra would be probably very much like those of carbon
films depoSited at relatively low temperatures from an hydrocarbon
plasma (including the contributions from a sizeable fraction of
hydrogen).
They wOI~ld show an electronic absorption background generally
increasing with the wave number, with a threshold around 1 ev,
corresponding to the gap between the valence and
conduction
Tt' bands.
The electronic spectrum at higher energies wOI~ld be
characterized by two strong absorption peaks:
around 12 eV (0--. cr~ transition)
around 6 eV (Tt-+1t* transition), with possible shifts
to slightly longer wave-lengths if interlayer interactions are not
qlJite negligible.
The infra-red phonon spectrum would be completely absent, or
possibly reduced to a Single absorption band at 1350 em-I (7.4 ~m)
since all the normal lattice vibration modes wOIJld be infra-red
inactive, because of the two-di.mensional character of the aromatic
structure (see infra-red spectrum of graphite).
The infra-red "molecular" spectrum would be
aromatic molecule:
that
of
a
large
- Aromatic CH bending absorption bands in the 11-14~ area
Aromatic absorption band at 1600 em-' (6.25 tun)
CH absorption frequencies in the 2700-3300 em-I area
(3 - 3.6 fom) where separate contriblJtions from aromatic and
aliphatic CH might be analyzed.
probably no ali.phatic CH'l. or CHJ bands in the 7 ,.m region
because of the low probability of 2 H atoms being
bonded to the same carbon.
Who can deny that such a spectrum wOIJld bear a very strong
resemblance to that observed from interstellar dust or polycyclic
aromatic molecllies ?
REFERENCES
(1) R. HE tMANN , J. fCLEIMM, N. SALANSI<Y
-~I~'
147-55 (1984)
(2) A. OBERLIN - carbon, 17, 7-20 (1979)
==
(3) A. OBERLIN, J. C..oMA, J.N. ROU'ZAfJD -
Proc. tntem. Carbon COnf.
Bordeallx 1984 - J. Chim. Phys. 81, 701-10 (1985)
VARIOUS KINDS OF SOLID CARBON
53
(4) S.P.F. HUMPHREYS-oWEN and L.A. GILBERT - Indust. Carbon and
Graphite Conf. London (1957) - Soc. of Chern. Ind. 1958 p.37-41
(5) J.T. McCARTNEY and S. ERGUN
-~,
37, 272 (1958)
(6) E.A. TAFT and H.R. PHILIPP - Phys. Rev., A 138, 197 (1965)
(7) O. K. GREENAWAY, G. HARBEKE, F. BASSANI, E. TOSA'I'l'I Phys. Rev., 178, 1340 (1969)
( 8) R. KLUCJ<ER, M. SKIBOWSKI, W. STEINMANN Phys. Stat. So1.(B), 65, 703 (1974)
(9) N.X. CHEN and S. BABII - Phys. Rev. B,
(10) S. ERGUN, J.T. MCCARTNEY, R.E. WALLINE
~,
8242-3 (1985)
-~, ~
109 (1961)
(11) S. ERGUN and J.T. McCARTNEY - Proc. 5th ConE. on Carbon 1961
Vol. 2, p. 167 (pergamon Press 1963)
==
(12) L.G. JOHNSON and G. ORESSELHAUS - Phys. Rev. B, 7, 2275 (1973)
(13) J.C. SLONCZEWSKI and P".R. WEISS - Phys. Rev., 109, 272 (1958)
(14) R.J. NEMANICH, G. LUCOVSKY, S.A. SOLIN Solid State Com., 23, 117-20 (1977)
(15) A. MISU, E. MENDEZ, M.S. ORESSELHAUS J. Phys. Soc. Japan,:%1..: 199-207 (1979)
(16) J.B. YASINSKI and S. ERGUN - carbon, 2, 355 (1965)
( 17) R. A. FRIEDEL and G. L. CARLSON Journal Phys. Chern., 75, 1149-51 (1971)
(18) J. FINK, T. MULLER-HEI~ERLING, J. PFLUGER, B. SCHEERER,
B. OISCHLER, P. KOIOL, A. BUB~R, R.E. SAH Phys. Rev. B, 30, 4713-18 (1984)
=
( 19) J. P. CHAUSSE - These de 3eme eyc le - Bordeaux 1966
(20) B. oISCHLER , A. BUB~, P. KOIOL Solid state Comm., 48, 105-8 (1983)
(21) C. LEVY-MANNHEIM and J. HERING Comptes-Rendus Ac. Sc. PariS, B 263, 1033-6 (1966)
(22) R.A. GREINKE and I.C. LEWIS - carbon, 22, 305 (1984)
54
A. MARCHAND
DISCUSSION
S. MUkamel
Has the density of vibrationaL modes as a function of
cLuster size been caLcuLated?
Answer: As far I understand it, this caLcuLation is very difficuLt.
Some efforts were made, but nothing reaLLy vaLid has been pubLished.
L. Allamandola: Since the organization of carbon structure from the
disordered domains ("amorphous carbon") to the graphitic, ordered phase,
which occurs at about 1000 - 1500° C, is a chemicaL or kinetic process,
time is important. How long does it take to go from the stached phase
to the earLiest "graphitic" phase? This is important since it is quite
likely that the particles of carbon produced in mass-losing, carbon
rich, stars are really amorphous carbon particles and in order to
convert this to graphite requires that the particle be heated to more
than 1000° C. From what I understand, this requires severaL hours.
Answer: The necessary time of course is Longer when the temperature
is Lower. TypicaLLy a complete graphitization may occur in half an hour
at 2800° C, but you may need days or months at 2000° C, and centuries
at 1500° C. But deveLopment of two-dimensionaL extended structures
("graphenes" or wrinkled carbon sheets) is obtained in 1 - 3 hours at
2000° C.
KPoto
What are the edges of graphite? In particular, do the sp2
dangLing bonds not take up H, OH, etc ••• if not in vacuo?
Answer: Yes, of course, a Lot of the "dangLing bonds" at the edges of
crystaLLites are used to fix H atoms. In other cases the C atoms act as
eLectron traps and there are electron pairs LocaLized on these atoms.
A. Leger: Can we hope to get a laboratory spectrum of a carbon structuraL unit in the future?
Answer: It does not look impossibLe. But I do not see cLearly yet
what the experimental set-up should be.
GAS/CARBONACEOUS SURFACE INTERACTIONS
A. Thorny
CNRS, Laboratoire 11aurice Letort associe a. I 'Uni versi te de
Nancy I, B.P. 104, 54600 Villers-les-Nancy, France
P. Wehrer
Laboratoire de Cinetique Heterogene, Universite de Nancy I,
B.P. 239, 54506 Vandoeuvre-les-Nancy, France
ABSTRACT. Typical results about physisorption and superficial reactivity
on non porous carbonaceous solids are summarized. On physisorption an
overview is given of the results obtained at thermodynamical equilibrium on the (0001) graphite face. On reactivity the studies here
considered have been carried out at low pressure (below 10- 2 Torr) and
high temperature (mainly above 1100K). Many experimental results among
those which are summarized are due to M. Letort, X. Duval and their
collaborators at the 'Ecole Nationale Superieure des Industries Chimiques, INPL - Laboratoire de Cinetique et Catalyse Heterogene de 1 'Universite de Nancy I - Laboratoire Maurice Letort du CNRS'.
1. PHYSISORPTION OF GASES ON GRAPHITE
1.1. The outstanding properties of graphite as adsorbent for the study
of 2D phase transitions (two-dimensional phase transitions occurring in monomolecular layers).
The interesting adsorbing properties of graphite are due to its thermal
stability and low chemical reactivity. The constituting layers are
reactive only at their edges, and their surface often presents, especially in natural graphites, large and uniform patches without defects
(with a mean diameter of the order of 10 3 ~). This is why the (0001)
crystal face of graphite constitutes a 'vesse~ of prime quality for
the study of 2D phase transitions in physisorbed films, being in addition very uniform energeticaJ.ly (shallow potential wells) so that the
·mobility of adsorbed molecules and their interactions are very slightly
hindered. Finally, graphite is suitable for studies of the adsorbed
films by a number of different techniques (volumetry, calorimetry, electron diffractions and spectroscopies, X-ray and neutron scattering, ellipsometry •• ) owing to a variety of available materials: graphitized
carbon blacks, powders of natural graphite exfoliated from intercalation compounds of very homogeneous surface and high specific area
(more than 50m 2 .g- 1 ), foils of compressed exfoliated graphites of type
Grafoil (Union Carbide) or Papyex (Le Carbone Lorraine) commonly used
55
A. Leger et al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 55-62.
©
1987 by D. Reidel Publishing Company.
56
A. THOMY AND P. WEHRER
in neutron and X-ray diffraction experiments, single crystals ; graphites of HOPG type (highly oriented pyrolytic graphite), or ZWX (HOPG
expanded in a volume of suitable geometry (cf {1-6}).
1. 2. Adsorbates
A great number of adsorbates have been studied, especially : He, Ne,
Ar, Kr, Xe, H2, 02, N2, NO, CO, C02, H20, NH 3 , CH 4 , C2H2' C2H4, C2H6'
C6 H6 , C6H12 , CF 4 , SF 6 ,····
1.3. The different steps of formation of an adsorbed monolayer
Generally as the equilibrium pressure of the 3D gas increases at constant temperature, the film passes through the following steps on the
uniform part of the surface :
- 2D gas in which the 'adsorbate-adsorbate' interactions are negligible (linear part OM of the isotherms, fig. 1) ;
- 2D gas in which the 'adsorbate-adsorbate' interactions become
more and more important (part I1N curved towards the e axis, fig. 1) ;
- 2D condensation if the temperature is low enough (NP or NQ
parts, fig. 1). In this transition which is of first order, the 2D gas
coexists with a 2D condensed phase 'liquid' or 'solid'.
The 2D condensation may be followed by other transitions bringing
the film to denser 2D states (ex. : RP part, fig. 1).
Finally the monolayer reaches a limit compacity (monolayer completion) which may generally be considered as self-determined. Its density
and structure are then approximately the same as on a surface of uniform adsorption potential ('adsorbate-adsorbent' interactions equal to
the average graphite potential at any point of the surface).
Remark : after the first monolayer completion other layers may form
with increasing the 3D gas pressure. An unlimited number of layers
corresponds to a perfect wetting of the surface by the adsorbate, while
a limited number of layers corresponds to an imperfect wetting (cf.{7})
In some cases (e.g. with C02 below 104K {B}) even one compact monolayer
cannot be formed (no wetting).
1.4. Different types of 2D phase transitions
Since the early se·venties, hundreds of papers have been devoted to physisorbed films on graphite, which has been the preferred adsorbent for
the most complete studies using a variety of techniques (cf. {2-6},
{9-11}). The different kinds of transitions which have been obersed are:
- successive 2D phase transitions of the type 'gas-liquid-solid'
including 2D triple and critical points (cf. fig. 1 - ex. : with xenon);
- 'commensurate-incommensurate' transitions, corresponding to a
change from a structure closely related to that of the substrate, to a
quasi-self-determined structure (ex. : with krypton) ;
- 2D polymorphism (e.g. molecules lying on the surface and then
standing up) in cases of non spherical or non globular adsorbate molecules (ex. : with oxygen).
Some cases of continuous melting (without 2D triple point) have
GAS/CARBONACEOUS SURFACE INTERACTIONS
57
also been observed (ex. : with C2H5).
When 'gas-liquid-solid' transitions with triple point (t) and
critical point (c) occur as in films of rare gases or globular molecules (ex. : CH 4 ) the following temperature ratios are obtained:
These values are in agreement with theories and simulations
assuming energetically uniform surfaces (without potential wells). Of
course, many other adsorbents than graphite have been used, especially
1~1ellar halides of divalent metals MX 2 (M
metal, X halides) extensively studied by Y. Larher and co-workers (cf {2,11}), but graphite
has still offered the largest collection of various 2D phases, including the most elaborate and detailed descriptions.
=
=
Size effects : the adsorbed film properties described above apply to
surfaces constituted of quasi-infinite uniform patches. When their
mean diameter is smaller than about 1000 1\, the phase properties are
depending on their size. The coexistence conditions may then be cons iderablymodified. As a consequence, the isotherm steps may be tilted
and even disappear as for partially graphitized carbons. In this case,
the edges of the uniform patches become important and contribute to
the smoothing of the isotherms step (cf {11,12}).
1.5. Theories
Numerous new theoretical developments have been devoted to 2D phases
which have mainly been applied to films adsorbed on graphite (cf {10,
11, S}). ThE; most frequently quoted authors are:
- Elgin and Goodstein ; Koster1itz and Thouless ; Toxvaert,
Halper'in and Nelson (on 2D melting and triple point) ;
- Domany, Schick and Walker (on critical behaviour) ;
- Berker and Ostlun~ (app1. of the renormalization group theory)
-, Villain, Shiba (on 'solid-solid' transitions) ;
- Monson, Steele and Henderson (on the effect of the substrate
relief on film properties).
1.5. The time of adsorption of an isolated admolecule on the (0001)
face of graphite
The time T during which an isolated molecule remains adsorbed on a
uniform surface (case of a 2D ideal gas) is expressed by :
=
T
= Toexp(-
Uo/kT) with To(sec) = 1i/kT
= 4.8.1O- 11 /T
{13}
U
minimal value of the adsorption potential relative to the 'adsorb~te-substrate' couple considered (cf {14}).
In the following table are given the Uo/k mean value and corresponding T for some 'adsorbate-graphite (0001)'
A. THOMY AND P. WEHRER
58
Adsorbates
He
240
T
(sec)
Ne
Ar
490
1110
after {l4,1}
10 10
at 10K
The adsorption potential of simple adsorbates on isolated polyaromatic hydrocarbons (PAH) could be estimated using the calculation
methods and data supplied in {11+}. The corresponding time of adsorption
T may be deduced as well as the probability at a given temperature for
a physisorbed molecule to explore all the sites of such a particle and
find a reactive one if it exists, where to chemisorb.
e
B
1
B
G-S
o
---~
p
Figure 1 : Example of 2D diagram with triple point (t) and critical
point (c). P : pressure of the 3D gas in equilibrium with the film.
S = surface coverage taken equal to 1 at the monolayer completion.
In thick line, adsorption isotherms corresponding from the left to the
right to increasing temperatures. In dashed line, boundaries of phase
coexistence domains. G = 2D gas ; L = 2D liquid ; S = 2D solid.
2. GAS-CARBONACEOUS SOLIDS REACTIVITY AT LOW. PRESSURE (p < 10- 2 Torr)
AND HIGH TEMPERATURE (1100<T<2500K.)
Reactions of carbons with gases have given rise to numerous studies
{IS}. Those we are going to summarize are a small part of them, nevertheless they present a great interest in so far as they enable the
approach of the elementary chemical acts.
2.1. Experimental
Most of the results were obtained by a dynamical method : constant gas
flow rate in an enclosure at room temperature with vacuum conditions
better than 10- 7 Torr.
Samples were generally pyrolytic graphitic
or non graphitic
carbon filaments or rods, electrically heated. The reaction products
GAS/CARBONACEOUS SURFACE INTERACTIONS
59
were analysed by mass spectrometry.
2.2.
Gaseou~
reactant
They are : H2 , H· (atomic hydrogen), F2 , F·, N2 , N·, 02' H2 0, CO 2 , NO,
N20 , N0 2 , 52, HZ5, 502' NH3' N2H4, N3H ; saturated hydrocarbons (CH 4 ,
C2 H6, C3Ha, ... ) unsaturated hydrocarbons (C2H2' C2H4, C3H4, C4H2 .. )
aromatic hydrocarbons (benzene, naphtalene, anthracene, coronene),
halocarbons (C 2 C1 4 , C4C14)'
2.3. General results
The following steps of a gas/carbonaceous solid reaction may be considered : gaseous reactant --> physisorption --> migration --> chemisorption --> dissociation --> formation and desorption of the reaction
products.
The different types of reactions are :
- gasification reactions if the carbon balance on the substrate
is negative ;
- carbon deposition reactions if the carbon balance is positive
- catalytic reactions if the carbon tialance is zero.
Gaseous reactants giving rise to gasification are
H· --> CH;, CH 4 , C2H6 (reaction products) {16-18}
02' H20, CO 2 , NO, N20, N02 --> mainly CO {19,20}
52' H25, 502 --> mainly CS 2 {21}
NH 3 , N2H4 , N3H --> HCN {22}
F2 , F· --> CF4 {23}
H2' N2 , N· do not react in the conditions considered.
Carbon deposition reactions occur with all hydrocarbons except CH 4
which does not appreciably decompose in the experimental conditions
considered. The general scheme of a deposition reaction may be written
as following: C H --> C + H2 + £C H (cf {24,25}).
p q
t
n m
Catalytic reactions are rare. They occur only as a parallel process to a gasification or deposition reaction.
2.4. Kinetic features and reaction mechanism.
The specificity of carbon behaviour between 1100 and 2500K is due to
the fact that its intrinsic reactivity and consequently its 'surface
state' are depending on the temperature (T) and pressure (P). To a
given carbon type and (T,P) couple corresponds a characteristic stationary rate which is determined by the stationary state reached by
the surface. The setting up of the surface stationary state is not
instantaneous : gasification or deposit of the equivalent of several
atomic layers is necessary. The reaction rate variation during the
transient regimes (change from a stationary state to an other) shows
that the Garbon surface becomes l~ss active with increasing T or decreasing P, and more active with decreasing T or increasing P. That is
60
A. THOMY AND P. WEHRER
why under the (T,P) conditions considered, the reaction rate exhibits
one maximum or, sometimes two (fig. 2), the position of which depends
on the nature and pressure of the gaseous reactant.
m
1
I
O
/~~~-'\
\
(c)
..
~
..
T(K)
~--~--~~~--~--~--~~-~~----~
1100
1500
1900
2200
Figure 2. Dependence of the stationary reaction rate on temperature.
Full lines correspond to gasification reactions : 02/C (a curve, after
{19}) and S2/C (b curve, after {21}). Dashed line corresponds to a
deposit reaction: C3H4 (allene)/C (c curve, after {29}). Stationary
pressure P
10- 4 Torr in the three cases. As on fig. 3 the reaction
rate of collision efficiency g is rapported to its maximum value ~.
=
The very strong increase of the reaction rate (g) at the lowest
temperatures (before the maximum) is mainly due to the desorption of
chemisorbed species which are considered as playing an inhibitory role.
The interpretation of the decrease of g (after the maximum) proposed
by X. Duval {19} is still admitted : competition between two effects
on the one hand, creation of reactive sites due to the attack of the
surface by the gas, on the other hand, decreasing of the number of
reactive sites by thermal healing process this second effect becoming
more and more important with increasing temperature. Nevertheless some
results show that chemisorption phenomena and modifications of the surface area should also be taken into account {21,26}.
2.5. Maximum values of the reaction rates
For a given type of carbon, the highest maximum values are obtained
with 02, N02, 52, C3H4, C4 H2 and probably with H- too (collision
efficiency g ~ 10- 3 ) ; then come H20, NH3, NO, C2H4, C2H2 (g ~ 10- 4 )
and at last CO 2 , N20, C2H6 (g ~ 10- 5 - 10- 6 ).
2.6. Effect of the substrate structure
Whatever the type of carbon is, the variation of the stationary reaction rate with temperature is qualitatively the same, but the values
obtained on non graphitic carbons are higher than on graphite by a
factor up to ten. In the case of graphite only a part of the edges or
defects of the constituting basal planes are really reactive as it has
been confirmed by studies on single crystals {23,27}.
2.7. The case of atomic hydrogen
GAS/CARBONACEOUS SURFACE INTERACTIONS
With atomic hydrogen a maximum exists but definitely below 1100K
(fig. 3), and stationary states are reached quasi-instantaneously. This
time the surface coverage by chemisorbed hydrogen is considered as the
determining factor, variations of the carbon surface state by thermal
healing process being negligible below 1100K. Too low temperatures are
unfavourable to the mobility and desorption of chemisorbed species,
while at too high temperatures the surface coverage by chemisorbed
species necessary to get 3D gaseous products (CH3 , CH 4 , C2H6) comes
to be too low. The optima gasification conditions correspond to the
maximum of the g vs T curve (fig. 3) {16-1S, 2S}.
Z/Zm
m
Figure 3 : Dependence of the reaction rate on temperature in the case
of atomic hydrogen (after {IS}). In this case, the maximum rate reaction is located under 1100K and is not due to changes in the surface
state as in the cases considered fig. 3.
References
1. Avgu1 N.N. and Kiselev A.V., 'Physical adsorption of gases and
vapours of graphitized carbon blacks', in : Chemistry and Physics
of Carbons, Vol. 6, Ed. P.L. Walker (Dekker, New-York) 1970.
2. Thomy A., ~uval X. and Regnier J. : 'Two-dimensional phase transitions as displayed by the adsorption isotherms on graphite and other
lamellar solids', Surface Sci. Reports, l, 1 (1981)
3. Dash J.G., Films on Solid Surfaces (New-York, Academic,1970)
4. Bienfait M. and Suzanne J., Eds. Colloque Internat. CNRS, 'Phases
bidimensionnelles adsorbees', Marseille, J. Physique, Paris ~~ (1977
5. Dash J.G. and Ruvalds J. Eds., Phase Transitions in Surface rilms,
Summer School, Erice (New-York, London Plenum, 19S0)
6. Sinha S.K. Ed., Ordering in Two Dimensions, Lake Geneva, Wisconsin
(North-Holland, Amsterdam, 1990)
7. Bienfait M., Surface Sci. ~§~, 411 (1985)
8. Terlain A. and Larher Y., Surface Sci. 125, 304 (1983)
9. Bienfait M. 'Two-dimensional phase tran~rtions of simple molecules
adsorbed on graphite', in : Current Topics in Materials Science,
Ed. E. Kaldis, (North-Hoiland, Amsterdam) ~, 361 (1979)
10.Vilehes O.E., 'Phase transitions in momolecular layer films physisprbed on crystalline surfaceS', Ann. Rev. Phys. Chern. ~!,463 (1980)
62
A. TImMY AND P. WEHRER
11. Thorny A. and Duval X. 'Developments in the study of two-dimensional
phase transitions during the last decade' in : 'Adsorption at the
Gas-Solid and Liquid-Solid Interface', Eds. J. Rouquerol and
K.S.W. Sing, (Elsevier, 1982)
12. Ross S. and Olivier J.P. : On Physical AdsOrption (Wiley-Interscience, New-York, 1964)
13. De Boer : The Dynamical Character of Adsorption (Clarendon Press.
Oxford, 1968)
14. Steele W.A. 'The interaction of gases with solid surfaces', in :
Enc clo edia of Ph sical Chemist
and Chemical Ph sics. vol. 3
Ed. D.H. Everett Pergamon, Oxford, 1973
15. See Chemistry and Physics of Carbons. Ed. P.L. Walker (Dekker.
New-York) and Carbon.
16. Coulon M. and Bonnetain L. J. Chim. Physique 71, 711 (1974)
17. Wood B.J. and Wise H.J. Phys. Chern. 7~. 1348 rI969)
18. Vietzke E•• Flaskamp P. and Philipps-V.J. Nuclear Mat. 111 and 11~
763 (1982). ibid. 1~~ and 1~~. 545 (1984)
--19. Duval X. 'Cinetique-aes reactions du carbone a haute temperature'
in : Les Carbones, Ed. A. Pacault (Masson, 1965)
20. Boulangier F •• Duval X., Letort M. in 'Froc. of the 1957 Carbon
Conf.' (Pergamon Press. Oxford). p. 1967
21. Wehrer P. and Duval X•• Carbon 1~. 241 (1980)
22. Sauvageot R•• Wehrer P. and Duval X. J. Chim. Physique ~1, 65 (1984)
23. Rosner D.E. and Strakey J.P. J. Phys. Chern. 77. 690 (1973)
24. Wehrer A., Wehrer P. and Duval X. Carbon f!,-247 (1983)
25. Wehrer A., Wehrer P. and Duval X. Carbon ii. 551 (1984)
26. Wehrer A. and Wehrer P. Unpublished results
27. Olander D.R •• Siekhaus W•• Jones R. and Schwarz J.A. J. Chern. Phys.
21, 408 (1972)
28. Mc Carrol B. and Mc Kee D.W., Carbon ~, 301 (1971)
29. Wehrer A•• Wehrer P. and Duval X. Bull. Soc. Chim. France,
434
(1980)
I.
Addendum
Experiments on the graphite-atomic hydrogen reactivity at very low temperatures carried out by Bar-Nun et al. (a) are worth mentioning because
they have been made in the frame of interstellar molecules studies. According to Bar-Nun et al., under hydrogen pressures of several torrs
and a microwave discharge, hydrocarbons (mostly CH4) would form down to
about 10K.
Mention may also be made of low energy electron diffraction studies performed around 10K on molecular hydrogen films physisorbed on the (0001)
graphite crystal face (b, c).
(a) Bar-Nun A., Litman M. and Rappaport M.L. Astron. Astrophys.
(1980)
(b) Seguin J.L. and Suzanne J. Surf. Sci. ll§. L241 (1982)
(c) Cui J. and Fain S.C. Faraday Discussion-SO (1985).
~~,
197
VUV TO FIR LABORATORY OBSERVATIONS ON SUBMICRON AMORPHOUS
CARBON PARTICLES
E. Bussoletti, L. Colangeli, A. Borghesi
Dipartimento di Fisica, Universita di Lecce
Via per Arnesano
73100 Lecce I
ABSTRACT. We present here a general overview of our experimental results
obtained on submicron amorphous carbon particles. Morphological and spectroscopic analysis represent a useful tool to test the particles producedin the laboratory as possible candidates to simulate the interstellar dust. Amorphous carbon grains have been produced and spectroscopically analized in the wavelength range from 1000 A to 330 !-1m following the
principal goal of obtaining homogeneous and reproduceable data under carefully controlled experimental ambient conditions. Present results and
related speculations suggest that submicron amorphous carbon grains with
radii smaller or equal to 40 A may account for the observed interstellar
hump; a mixture of amorphous carbon grains and polycyclic aromatic hydrocarbons is able to produce the entire family of the so called unidentified interstellar emission bands.
1.
INTRODUCTION
The interpretation of the interstellar extinction curve is one of the
main sources of information about the physical and chemical properties
of solid particles that permeate the interstellar space (see, for instance, the reviews by Huffman, 1977; Salpeter, 1977; Greenberg, 1978;
Allamandola, 1984; Willner, 1984).
Laboratory analysis and spectroscopy represent then a useful tool
to test different substances as possible candidate able to simulate the
observations.
However it is quite customary to utilize inhomogeneous experimental
informations without any critical "a priori" check of their internal consistency. Data obtained in different ambient conditions, from apparently
similar materials regardless of the production methods and/or raw materials, are used to simulate the astronomical observations.
To overcome this problem, our group is performing a systematic laboratory work since five years having the following goals: a) to work in
controlled experimental conditions; b) to obtain homogeneous data; c) to
verify that the results are reproduceable when the same ambient conditions apply.
63
A. ueger et aI. (em.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 63-74.
© 1987 by D. Reidel Publishing Company.
64
E. BUSSOLETTI ET AL.
With this aim, submicron amorphous carbon particles have beenopro
duced and spectroscopically analized in the wavelength range 1000 A 330 ~m (Borghesi et al., 1983; 1985; 1986; Colangeli et al., 1986).
We present here a general overview of these results.
2.
EXPERIMENTAL
Our approach consists
a) production of
b) morphological
c) spectroscopic
2.1
of three phases :
dust samples;
and structural analysis;
measurements.
Production
Amorphous carbon particles are obtained by means of two methods :
- striking an arc between two amorphous carbon electrodes in a controlled Ar atmosphere at different low pressures, ~ 1 Torr (AC);
- burning hydrocarbons (benzene and xylene) in air at room pressure
(BE/XY) •
2.2
Morphological and structural analysis
High magnification Transmission Electron Microscopy (TEN) of the samples
provides the following informations :
a) the grains appear as spheroids in chain-like structures (Fig.1a,
1b) ;
b) typical average grain radii, <a> , depend upon ambient pressure,
P, i.e.
P = 760 Torr
----->
<a>= 100
+
150 A
P =
1 Torr
--->
<a>= 40
.,.
50 A
c) size distributions are mainly skew (Fig. 2).
2.3
Spectroscopic observations
They have been performed between 1000 'A and 330 ~m by using the following instruments for the different wavelength ranges :
1000 A
3000 A PULS-synchrotron light facility at the "Adone"
storage ring in Frascati, Rome;
2000 A
2.6 ~m Perkin Elmer 330 spectrophotometer;
2.5 ~m
50 ~m Perkin Elmer 683 spectrophotometer;
30 ~m
330 ~m Bruker FTIR IFS 113 interferometer.
Complete details on the experimental procedures are reported in previous papers (Borghesi et al., 1983; Co1angeli et al., 1986) and therefore will not be repeated here.
3.
EXPERIMENTAL RESULTS
3.1
VUV-Visible observations
LABORATORY OBSERVATIONS ON SUBMICRON AMORPHOUS CARBON PARTICLES
65
The extensive work performed at PULS facility has
shown the following main evidences in the amorphous carbon extinction curve (see Fig.3):
a) a pronounced hump in the range 2300 - 2400 A;
b) a shoulder at about 1800
c) a second peak at about 1500
d) a very steep increase below 1500
e) a continuous trend",>--l.4
beyond 3000 A.
The peaks position shifts towards shorter A with decreasing the
grain radii. In addition, all the features tend to smear out progressively for particles collected near and near\to the arc.
The main peak at ~ 2350
is interpreted as due to a surface plasmon mode of surface electrons in the ground state. This interpretation
is confirmed by the absence of any absorption band in the lower energy
range of the electronic transitions.
The origin of the other structures is less easy to identify because
no other evidences concerning amorphous carbon have been, so far, reported in the literature. We tentatively suggest that the steep increase at
~ < 1500 A may be due to electronic transitions to excited states. The
other features (i.e. the secondary peak and the shoulder) may be explained with a quantum size effect produced by the density of states (DOS)
at the surface of the grains (see for instance Bassani et al., 1985).
The observed shift towards shorter wavelengths is a peculiar characteristic of small particle absorption (Peremboom et al., 1981; Charle
and Schulze, 1985). A similar behaviour has been already noticed for graphite grains (see, for example, Gilra, 1971; 1972).
Finally the progressive smearing of the structures which is observed in the extinction of the samples collected at the shortest distances
from the arc may be attributed to a higher coverage rate of the LiF substrate, on which grains are collected, which tends to reduce the effect
of surface DOS.
A;
A;
A;
A
3.2
Infrared observations
The amorphous carbon extinction spectrum at these wavelengths has a general continuum trend which follows a ).-1.0 ±0,1
law in agreement with
the observations by Koike et al. (1980), performed on amorphous carbon
samples obtained with production methods similar to ours.
Weak bands are detected in the 3 - 14 ~m range superimposed to
the continuum, which may be related to the so-called unidentified infrared emission (UIR) bands.
These bands are detected in a wide variety of galactic and extragalactic regions rich in UV flux (Aitken, 1981; Allamandola, 1984; Willner,
1984) and they occur generally in emission respectively at 3.3, 3.4, 3.5,
6.2, 7.7, 8.6 and 11.3 ~m. Their nature and origin are still object of
a wide debate. Initially UIR bands were attributed to a solid state origin; presently it has been also suggested that they may be consistent
with an hypotesis of emission from highly vibrationally excited polycyclic aromatic hydrocarbons (PAHs) (Allamandola, 1984; Leger and Puget,
1984; Allamandola et al., 1985; Cohen et al., 1985a,b).
Absorption spectra of our AC samples taken at room temperatures and
at three different higher temperatures (100, 260 and 400 C) show six
E. BUSSOLETfI ET AL.
66
bands which fit quite well, in wavelength, the UIR bands observed in space: 3.4, 3.51, 5.78, 6.29, 6.85, 11.3 ~m (see Fig. 4a,b,c). The other
three astronomical bands at 3.28, 7.7 and 8.6
~m are apparently lacking in our spectra.
By observing Fig. 4 (a,b,c) we note that some of the bands are sensible to the temperature, that is:
a) the bands at 3.39, 3.42, 3.51 and 6.85 ~m decrease in intensity
as the temperature increases and tend to disappear completely at
400 C;
b) the 6.29 ~m band increases in intensity;
c) the bands at 5.78, 7.3 and 11.3 ~m remain mainly stable.
According to Duley and Williams (1981), we attribute our bands to
functional groups bonded to chemically active sites. They are located on
the periphery of carbon grains and are due to randomly oriented graphitic platelets present in the AC particles. This interpretation is also
consistent with the temperature dependence observed in our bands. In fact
it seems possible to interpret the bands,which have a similar behaviour
when the temperature increases, as due to different modes of the same
functional groups adsorbed onto the surface of the grains.
In Table I we summarize our results and some possible interpretations. The data for the quenched carbonaceous composite (QCC), sintetized in form of a carbon film by Sakata et al. (1984). and the preliminary results for some PAHs (coronene and crysene) presented by Leger and
Puget (1984) and by Allamandola et al. (1985), are also given for comparison. We agree. however, about the fact that data from these PAHs must
be considered as only indicative for a "class" of possible candidate compounds.
By comparing our observations with the bands expected from PAHs -see
Table 1- we note some sort of complementarity: AC spectra are lacking
some bands present in PAHs and viceversa.
Since UIR bands are observed in Planetary Nebulae, according to
Allamandola et al. (1985) and Crawford et al. (1985). it is likely that
PAHs are formed in the carbon rich outflow from these sources. They may
be then the leftover condensation nuclei which have not been incorporated into solid carbon grains.
We expect, therefore, that around these objects there may exists a
mixture of both PAHs and AC grains, whose emission is able to produce
the entire family of UIR bands. The solid particles should be responsible for one portion of bands while a collection of PAHs should be responsible for the remaining part. Both sources contribute, evidently, to
the emission spectrum of common bands.
We point out that this speculation appears particularly attractive
because it is able to solve many problems by exclusively adhering to one
Qypotesis or to the other and, in addition, it is able, with the same
basic material. to account also for the observed interstellar hump at
2200
At present we have reported transmission spectra; in order to better compare the properties of small amorphous carbon grains with astronomical observations, emission measurements from these particles are
planned in the near future.
A.
13.24
11.96
11.4
11.4
11.9
(.) seen in NGC 7027 (Bregman et al. 1983)
11.3
6.2
8.85
11.3
6.25
5.2
3.3
8.6
6.94
7.27
6.25
(lJm)
12.2
7.8
8.2
8.75
10.6
11.4
6.5
6.9
6.2
3.3
CORONENE CHRYSENE
600 C C24 H12
C18H12
7.6
7.3
5.78
6.29
3.29
3.39
3.42
3.48
T mb
a
QCC
7.7
6.85
7.3
6.9
7.27
11.3
*
5.78
6.29
3.39
3.42
3.51
AC
Tamb 400 C
5.62
6.29
3.4
3.51
3.28
UIR
BANDS
x
~
~
~
=
C
~
~
0
~
'"~
d
~
g
~
o
out of plane aromatic CH bending
(one adjacent H atom)
out of plane aromatic CH bending
(two adjacent H atoms)
out of plane aromatic CH bending
(three adjacent H atoms)
~
~
p
~
asymmetriC deformation in -CH3
g
CH rocking in -CHOf
~
symmetriC deforma. of CH in -CH 3 >
CN stretching in -NH2
~
z
in-plane aromatic CH bending
~
3!
CO stretching in -CHO
~
skeletal in plane C=C vibrationl ~
-NH 2 deformation
~
asymmetriC stretching in -CH 3
symmetric stretching in -CH 3
CH stretching of -CH
possible
interpretation
of UIR bands
Table I. Infrared bands measured in laboratory for carbonaceous materials
and the UIR bands.
68
E. BUSSOLETTI ET AL.
Figure 1. TEM photographs of typical amorphous carbon samples produced
by arc striking. (a) chain-like clusters of grainsj(b} a single grain.
LABORATORY OBSERVATIONS ON SUBMICRON AMORPHOUS CARBON PARTICLES
.
AG
..
69
..
10
40
---
20
r
100
dIll
300
~
dIAl
SOO
700
Figure 2. Typical size distributions obtained by measuring the diameter
d, of about 1000 single grains in TEM images of AC and BE grains.
I.......
C
c
0.0
0.0
O.O~_~_---J._ _. . . L - _ - '
1000
3000
Figure 3. UV spectra obtained on AC amorphous carbon grains collected
on different distances from the arc : 5 cm (a), 7 em (b), 10 em (e).
E. BUSSOLETTI ET AL.
70
3.33
3.57
7.69
UJ1
12.50"",
~
I
Z
0
:0.95
I
5."
...iz
..
6.67
5.18
.... ....
I
n.30t"ft
I
C
.:
0.
I
3A2
0.85
a
3000
1700
1500
WIAVIN~. cm-'
900 100
Figure 4(a,b). Infrared weak bands detected in transmission measurements on AC at room temperature (a) and at 260 C (b).
LABORATORY OBSERVATIONS ON SUBMICRON AMORPHOUS CARBON PARTICLES
11.30
I""
z
o
iii
~O;
~
til
Z
C
•...
c
~--~----~ +--~-+--+--+--~#+---~~~~
1500 1300
WAV.NUMMa, cm-1
900 100
Figure 4c. Infrared weak bands detected in transmission measurements
on AC at the temperature of 400 C.
71
72
E. BUSSOLETII ET AL.
QUESTIONS
Duley:
Have you an explanation why you see the 11.3 ~m CH bending
mode but no 3.28 ~m CH streching mode in spectra of your samples?
Bussoletti
spectra is
CH is very
ficient to
: I can only point out that the 11.3 ~m band in our AC
very weak. It seems, therefore, that the functional group
poor in our ambient conditions and probably it is not sufproduce any detectable 3.28 ~m band in transmission.
Garwin: Aerosol research clearly shows that PAHs formation critically
depends on the ventilation of the flame. When you burn xylene in ai~
without very precise knowledge and control of the combustion conditions,
a totally unknown and irreproducible result is produced. In addition,
coagulation occurs in the rather haphazard conditions of diffusion collection of the particles near the arc source. Aerosol technique offer a
powerful 1 set of tools for accurately classifying and collecting particles under your production conditions.
Bussoletti : This question seems to indicate that probably I have not
been sufficiently clear in explaining our different production methods
and consequent results. Burning of hydrocarbons is not done by means of
arc, which is instead used in an Argon atmosphere. As I have already
mentioned all our results are very well reproducible. Actually the morphological data and the spectroscopic ones that I have shown represent
average values which have been obtained by repeating several times each
step of the experimental procedure. I may however agree that aerosol
tectmiques may help somehow our work even though they do not seem better
than present procedures.
Marchand: I would like to comment on the shape, size, and general
aspects of your carbon grains, as seen with TEM. They look extremely
similar to carbon blacks, and I would not be surprised if they have a
similar graphitic structure. Anyway, I think a very careful study with
electron microdiffraction techniques is necessary before they can really
be called 'amorphous'. I would not be very surprised either if some
industrial carbon blacks turn out to have the same UV, visible, and IR
spectra.
LABORATORY OBSERVATIONS ON SUB MICRON AMORPHOUS CARBON PARTICLES
73
REFERENCES
Aitken, D.K. : 1981, in Infrared Astronomy, IAU Symposium No. 96, eds.
C.G. Wynn-Williams and D.P. Cruikshank (Dordrecht : Reidel),
p.207
Allamandola, L.J. : 1984, in Galactic and extragalactic infrared spectroscopy, XVI ESLAB Symposium, eds. M.K. Kessler, J.P. Phillips and T.D. Guyenne (Dordrecht : Reidel)
Allamandola, L.J., Tielens, A.G.G.M., Barker, J.R. : 1985, Astrophys. J.,
290, L25
Bassani, F~~=Bourg, M., Cocchini, F. : 1985, II Nuovo Cimento,5D, 419
Borghesi, A., Bussoletti, E., Colangeli, L., Minafra, A., Rubin1, F. :
1983, Infrared Phys., 23, 85
Borghesi, A., Bussoletti, E., Cor=angeli, L.
1985, Astron. Astrophys.,
~~~, 225
Borghesi, A., Bussoletti, E., Colangeli, L.
1986, 'Amorphous carbon and
the IR unidentified bands', Astrophys. J.,submitted
Bregman, J.D., Dinerstein, H.L., Goebel, J.H., Lesler, D.F., Witteborn,
F.C., Rank, D.M., : 1983, Astrophys. J., ~~~, 666
Charle, K.P., Schulze, W. : 1985, 'Optical properties-of metal clusters
containing some ten to ten thousand atoms', preprint
Cohen, M., Allamandola, L.J., Tielens, A.G.G.M., Bregman, J., Simpson,
J., Witteborn, F.C., Wooden, D., Rank, D. : 1985a, Astrophys.
~ ., preprint
Cohen, M., Tielens, A.G.G.M., Allamandola, L.J. : 1985b 'A new emission
feature in lRAS spectra and the polycyclic aromatic hydrocarbons spectrum', preprint
Colangeli, L., Capozzi, V., Bussoletti, E., Minafra, A. : 1986,'Extinction spectra of amorphous carbon submicron grains in the UVVisible range', in preparation
Crawford, N.K., Tielens, A.G.G.M., Al lamandol a , L.J. : 1985, Astrophys.,
J., ~~::!, L45
Duley, W.W., Williams, D.A. : 1981, M.N.R.A.S., ~~~, 269
Gilra, D.P. : 1971, Nature, ~~~, 237
Gilra, D.P. : 1972, 'Collective excitations and dust particles in space'
in The scientific results from the Orbiting Astronomical Observatory OAO-2, NASA SP-310, p.295
Greenberg, J.M. : 1978, 'Interstellar Dust' in Cosmic Dust, ed. J.A.M.
McDonnell, (Wiley, Chichester), p.187
Huffman, D.R. : 1977, Adv. Phys., ~~, 129
Koike, C., Hasegawa, H., Manabe, A:-: 1980, Astrophys. Space Sci., ~~,
495
Leger, A., Puget, J.L. : 1984 , Astron. Astrophys., ~::!~, L5
Peremboom, J.A.A.J., Wyder, P., Meier, F. : 1981, Physics Reports,78.
173
74
E. BUSSOLETTI ET AL.
Sakata. A•• Wada, S •• Tanabe. T •• Onaka. T. : 1984. Astrophys. J •• 287.
151
Salpeter. E.E. : 1977. Ann. Rev. Astron. Astrophys •• 15. 26
Willner. S.P. : 1984. in Galactic and extragalactic infrared spectroscopy. XVI ESLAB Symposium, eds. M.K. Kessler. J.P. Phillips and
T.D. Guyenne (Dordrecht : Reidel)
SPECTROSCOPY OF MATRIX-ISOLATED CARBON MOLECULES IN THE UV, VIS, AND IR
SPECTRAL RANGE.
W. Kratschmer, K. Nachtigall
Max-Planck-Institut fur Kernphysik
P.O.Box 103980
6900 Heidelberg
W-Germany
ABSTRACT. We investigated the UV-VIS and IR spectra of mixtures of
larger carbon molecules, Cn (n ranging from 2 to about 9), in matrices
of solid argon. In the UV-VIS domain, several strong absorptions (oscillator strengths of about 0.3) were observed. Each feature seems to
originate from a specific carbon molecule. In an earlier attempt to
interpret the spectra, we assumed that the individual bands are produced by the electronic E -) Eu transitions within the linear carbon
chain molecules C4' C5' e~c. up to C9' According to this assignment, an
absorption band detected at around 450 nm originates from the C7 molecule. Although distorted by matrix effects, this and a few weaker bands
show wavelength positions rather close to some of the diffuse interstellar lines.
We found the IR spectra difficult to interpret in terms of linear
carbon molecules alone. Our previous assignments of the UV-VIS absorptions to linear molecules thus have to be regarded as highly preliminary. Weak IR-features were found at the positions at which aromatic
carbon ring molecules should absorb; otherwise no striking coincidences
with interstellar IR-features were noticed.
1. INTRODUCTION
Carbon in the form of small grains and larger molecules seems to be
responsible for a variety of interstellar features in the UV, VIS, and
IR. In, particular, linear carbon molecules Cn with n= 5, 7, or 9
(Douglas, 1977) or polycyclic hydrocarbons, probably in an ionized
state (Leger and d'Hendecourt, 1985; van der Zwet and Allamandola,
1985) have been suggested as the carriers of the diffuse interstellar
bands. Since a test of the hypothesis of Douglas was in close range of
our experimental capabilities, we started to study the spectral features of matrix isolated carbon molecules.
The technique to produce larger carbon molecules in matrix isolated form is simple: Initially, the lighter molecules of carbon vapour
(i.e. the species C, C2' and C3) are trapped in an inert matrix (e.g.
solid argon). Upon thermal annealing, the lighter molecules start to
75
A. Leger et al. (etis.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 75-83.
© 1987 by D. Reidel Publishing Company.
76
w.
KRATSCHMER AND K. NACHTIGALL
diffuse through the matrix and to react with each other, e.g. according
to Cm+C n = Cm+n . One finally obtains a mixture of matrix isolated heavy
carbon molecules, i.e. a kind of "carbon-soup". The key problem is to
identify the individual molecules present in that "soup" from the
signature of their spectral features.
In a series of important papers, Weltner and co-workers have
already almost exhausted the field of VIS and IR spectroscopy on matrix
isolated carbon molecules ranging from the species C3 up to C9 (Weltner
et al., 1964; Weltner and McLeod, 1964; Weltner and McLeod, 1966;
Thompson et al., 1971; Graham et al., 1976). We provided what was left
out, namely the UV-VIS spectra of these molecules (Kratschmer et al.,
1985). We found that a particular molecule (probably C7) exhibits a
strong absorption at 447 nm, i.e. close to the position of the most
intense diffuse interstellar band. A line at about 450 nm has already
been observed by Wdowiak (1980) in a matrix-isolation study of the
spectra of molecules produced by burning discharges in methane-argon
mixtures.
Earlier theoretical work (Spitzer and Clementi, 1959) suggests
that the Cn molecules form linear chains as long as n is not too large
(n < 10). More recent and more refined MO-calculations indicate that
for Cn (n > 4) non-linear ground-state configurations may be more
likely (see e.g. Koutecky and Paccioni, 1984). Mass spectroscopy of
free carbon clusters indicates linear structures for the even n Cn
species (up to C24 ) while the atomic arrangement of the odd n Cn
clusters remains unclear (see, e.g. Rohlfing et al.. 1984). The IR
investigations of Thompson et al. (1971) suggested linear structures
for C4' C5' C6' and C9' however. these authors already noticed features
they could not assign to linear molecules. In our recent IR- work on
intensity correlations between features we reach similar conclusions.
2. EXPERIMENTAL
The experimental set-up for the UV-VIS studies has been described
previously (Kratschmer et al., 1985). By a carbon evaporator (resistively heated carbon rods) carbon molecules were produced and deposited along with the matrix gas (mainly argon) onto a cryogenically
cooled window (initial window temperature about 10-15 K). The matrix
deposition rate (in the order of l~m/min) and the carbon evaporation
were adjusted such that carbon concentrations between .1 and 1 mol%
with respect to the matrix were achieved. The spectrometer beam passed
through the cold window (sapphire for the UV-VIS and KBr for the IR).
Two essential similar matrix isolation set-ups were used. one attached
to an IR, the other to an UV-VIS spectrometer. We can not yet perform
IR and UV-VIS measurements on the same matrix-isolation samples. Usually, initial spectra were taken immediately after sample deposition.
Then the matrix was warmed up until changes in the absorption features
could be noticed, re-cooled again to the initial temperature, and the
next spectrum recorded. Three to ten annealing steps were recorded
until the matrix started to sublime. Strong annealing effects in argon
matrices usually occur at temperatures above 30 K.
SPECTROSCOPY OF MATRIX-ISOLATED CARBON MOLECULES
77
3. RESULTS AND DISCUSSION
3.1. UV-VIS Spectra
Since the spectral data and their interpretation have already been
published (Kratschmer et a1., 1985) we merely want to describe the
major findings:
Several features were observed to grow or decay upon annealing.
The decaying features are the known bands of C2 at 238 nm and of C3 at
410 nm. The C3 band is a E -) nu transition of low (10- 3 ) oscillator strength (We1tner an§ McLeod, 1964). The growing features are
broad and exhibit vibrational sub-structures. Most intense are the
features centered at 247, 311, 394, and 447 nm. Each of the individual
features (at least up to the 447 nm band) grows in a characteristic
fashion, indicating that each absorption originates from a particular
molecule. The bands are strong (we estimated the oscillator strength of
the 247 nm band to be about 0.3), i.e. the features belong to allowed
electronic transitions. A low resolution spectrum of a strongly annealed matrix-sample is shown in Fig. 1.
1.4
Carbon Molecules en
12
WAVELENGTH (nml
Figure 1. The spectrum of carbon molecules present in the argon matrix
after strong thermal annealing (to about 35 K). The spectral resolution
is 2 nm. On a background produced by scattering, a number of absorption
bands can be noticed. Each band seems to originate from a specific Cn
molecule. We tentatively assigned the features to the molecules C4 to
e9·
78
W. KRATSCHMER AND K. NACHTIGALL
Our assignments of features to molecules is based on the following
arguments:
The molecular composition of carbon vapour is roughly 20% C. 5%
C2 • and 75% C3 (see. e.g. Zavitsanos and Carlson. 1973). Under the
reasonable assumption that atomic carbon diffuses relatively easily
through the matrix. the first heavier molecule that should form upon
annealing is C4 . Two narrow bands. located at 470 and 520 nm are
claimed to belong to C4 which probably is a linear molecule (Graham et
al .• 1976). These two lines appear in our spectra as well (see Fig. 2)
and were found to grow correlated with the 247 nm band. We thus identified the 247 nm band as due to C4 . The strength of the 247 nm band
suggests that it corresponds to a ~g -) ~u transition from the ground
state u orbital to the adjacent u orbital. For the linear C3
molecule. the transition of this type occurs at 178 nm (Kok Wai Chang
and Graham. 1982). In Fig. 1 which shows the entire spectrum of the
carbon-molecule-mixture on a compressed wavelength scale. it can be
seen that the features are roughly equidistant in wavelength spacing.
The C3 feature at 178 nm (which we could not observe since it is beyond
the range of our spectrometer) would fit perfectly into this pattern.
Because C3 and probably also C4 are linear molecules. the uniformity of
the wavelength spacing suggests that the other molecules are linear as
well (see e.g. the discussion of Platt. 1961. on conjugated chain
molecules). We thus extrapolate from the C3 -C4 "baseline" and conclude
that the 311 nm band belongs to C5' the 394 nm band to C6' and the 447
nm band to C7 • etc •• These are the assignments depicted in Fig. 1.
At first glance it appears. that the UV-VIS spectra of the carbon
molecules can be easily interpreted in terms of linear molecules.
However. as will be shown below. the IR data suggest that the carbon
molecules in our matrices form not only linear but also other types of
structures. Since we cannot yet correlate the individual UV-VIS features with the particular IR bands of the different carrier molecules.
the assignments given in Fig. 1 should be regarded as preliminary.
3.2. The 447 nm Feature
Investigating the carbon-molecule spectra in different matrices. we
observed that in a N2 matrix the wavelength of the ~g -) Uu transition
in C3 agrees well with the gas phase value (405 nm). The 447nm feature
in the argon matrix appears at 443 nm in the N2 matrix. Provided the
matrix shift of this feature in N2 is similarly small. the absorption
should occur at 443 nm for the free molecule. i.e. very close to the
position of the strongest diffuse interstellar line (442.8 nm). A
direct comparison of one of our laboratory spectra (argon-matrix) with
that of the diffuse interstellar lines (Herbig. 1975) is shown in Fig.
2. Even though the lines in the laboratory spectrum are strongly broadened by matrix effects. the pattern of features of both spectra in the
range 440 to 560 nm appears to be quite similar. One immediately realizes from Fig. 2 how valuable the spectroscopy of carbon molecules in
the gas phase (or at least in a less distorting matrix) would be as the
ultimate test of the Douglas hypothesis.
SPECTROSCOPY OF MATRIX-ISOLATED CARBON MOLECULES
79
Laboratory Spectrum
40
500
5 0
600
60
700
WAVELENGTH I nml
Figure 2. Comparison of the laboratory spectrum of carbon molecules in
an argon matrix (resolution 0.1 nm) and the spectrum of the diffuse
interstellar bands taken from the compilation of Herbig (1975). The two
narrower bands at 470 and 520 nm in the laboratory spectrum probably
belong to the linear C4 molecule (Graham et al., 1976). According to
Douglas (1977), C4 and the other carbon molecules lighter than the
carrier of the 443 nm absorption should not contribute to the diffuse
interstellar bands.
3.3. IR Spectra
Some of the IR spectra obtained after matrix-annealing are shown in
Fig. 3. Our data generally agree very well with those obtained by
Thompson et al., (1971). The initially strongest band is the C3 feature
at 2039 em-I. A few other bands are present as well. These have been
assigned by the mixed-isotope studies of Thompson et al., to the linear
species C4 (at 2164 cm- 1 ), C5 (at 1952 and 1544 em-I), C6 (at 1997 and
1197 cm- 1 ), and C9 (at 2138, 1893, and 1447 em-I). At later states of
annealing, additional features grow which, according to Thompson et
al., belong to pure carbon molecules as well but could not be assigned
to specific carbon species. Linear C7 does not appear in the list of
molecules identified by these authors. We detected a few features which
were not reported by Thompson et al., especially in the region around
2200 em-I. We believe that these are produced by pure carbon molecules
as well.
Remarkable is the strong growth of the 1997 cm- 1 absorption (the
C6 of Thompson et al.) which, during the initial states of annealing is
distinctly correlated with the decrease of the 2039 em-I C3 band. Of
the UV-VIS bands, the 247 nm feature, assigned by us to linear C4
80
W, KRATSCHMER AND K, NACHTIGALL
rather than to C6 • seems to exhibit a similar growing behaviour.
Whether these two bands are in fact related (i.e. whether a contradiction in the assignments exists) has to be checked more carefully by
future UV-VIS and IR studies.
WAVELENGTH
7.5
0.5
6,5
(~m)
5,5
4,5
C molecules in Argon matrix
0.4
02
~
I'"
z
0.1
OD ........J'.......-.-'''"''"'""',,·..--.i
~ o.~
~
~
0.1
0,'
02
0.1
at 13K
WAVENUMBER (em-I)
Figure 3. The IR spectra of argon matrix-isolated carbon molecules at
different degrees of thermal annealing (resolution between 1 and
2 cm- 1 ). The scattering background has been subtracted from the spectrum.
The initially strongest feature is the C3 band at 2039 cm- 1 • Upon
annealing, this band decreases whereas other bands increase in intensity indicating the formation of molecular species larger than C3. Already in the initial state (bottom s~ectrum) other molecules besides C3
are present. The line at 2140 cm- comes from the almost unavoidable
contamination by CO. Most of the other lines are known to originate
from pure carbon molecules (Thompson et al., 1971).
Since we could not perform measurements on isotopically exchanged
molecules, we looked for intensity correlations between the IR-bands.
We found that the growing behaviour of the 1997 cm- 1 line does not at
all correlate with that of the band at 1197 cm- 1 , even though both
bands, according to Thompson et al., should belong to the same molecule, namely linear C6. We thus feel that at least this particular
assignment of Thompson et al. has to be revised.
81
SPECTROSCOPY OF MATRIX-ISOLATED CARBON MOLECULES
In all the matrix work performed, a very close intensity correlation between the lines at 1804 and 1844 cm- 1 was found and we are
convinced that both lines belong to the same molecule. Using reasonable
force constants, the positions and the energy spacing of these two
lines are difficult to understand by a linear configuration. Cyclic
(however not fully symmetric) and other structures are more suited to
explain this kind of line pattern.
After stronger annealing, weaker features appear in the vicinity
of 6.2 ~m wavelength (1610 cm- 1 ). Whether these are related to polycyclic carbon structures is not yet clear. The most intense bands grow
between 4.5 and 6 ~m and none of these bands obviously coincides with
any of the unidentified interstellar IR emissions presently known.
Even though a large fraction of the molecules obtained after
matrix annealing probably are non-linear, our data nevertheless indicate that some linear structures may be formed as well. By inspection
of Fig. 3 one notices that features tend to pile-up in the 2200-2300
cm- 1 domain. This suggests the presence of larger linear molecules
which, under the assumption of reasonable force constants, should all
possess IR-active modes at such comparatively short wavelengths.
4. CONCLUSIONS
It appears that the carbon molecules produced in our matrix annealing experiments exhibit linear as well as non-linear structures and
thus an assignment of spectral features to specific molecules within
this "carbon-soup" is rather difficult. Consequently, the carrier molecule of the 447 nm feature cannot yet be identified with certainty. As
far as the Douglas hypothesis is concerned it can be stated that a
particular carbon molecule in fact seems to exhibit a strong feature at
about 440 nm. Whether this molecule is one of the carriers of the
diffuse interstellar bands remains to be investigated.
REFERENCES:
Douglas A.E., 1977: Nature
~~2'
130
Graham W.R.M., K.I. Dismuke, W. Weltner, 1976: Astrophys. J.
Herbig, G.H., 1975: Astrophys. J.
~~~,
301
~2!'
129
Kok Wai Chang, W.R.M. Graham, 1982: J. Chem. Phys.
ZZ'
4300
Koutecky J., G. Pacchioni, 1984: Ber. Bunsenges. Physik. Chern.
Kratschmer W., N. Sorg, D.R. Huffman, 1985: Surface Science
Leger A.,L. d'Hendecourt, 1985: Astron. Astrophys. ~~~, 81
!!,
~~~,
233
814
W. KRATSCHMER AND K. NACHTIGALL
82
Platt J.R., 1961: Encyclopedia of Physics
(Springer, Berlin) p. 173
~Z,
Part 2, Ed. S.Flugge
Rohlfing E.A., D.M. Cox, A. Kaldor, 1984: J. Chern. Phys.
Spitzer K.S., E. Clementi, 1959: J. Am. Chern. Soc.
~J,
~J
(7), 3322
4477
Thompson K.R., R.t. DeKock, W. Weltner, 1971: J. Am. Chern. Soc.
4688
2~'
van der Zwet G.P., L.J. Allamandola, 1985: Astron. Astrophys. 1~2' 76
Weltner W., P.N. Walsh, C.L.Angell, 1964: J. Chem. Phys. ~~, 1299
Weltner W. , D. McLeod, 1964: J. Chem. Phys. ~~, 1305
Weltner W. , D. McLeod, 1966: J. Chern. Phys.
Wdowiak T.J., 1980: Astrophys. J.
~~~,
4.~,
3096
L55
Zavitsanos P.D., G.A. Carlson, 1973: J. Chem. Phys.
~2,
2966
SPECTROSCOPY OF MATRIX-ISOLATED CARBON MOLECULES
83
DISCUSSION
D. DuZey: During diffusion of Cn molecules in a matrix is it not
more likely that non-linear molecules will be created ?
Answer: Our IR data indicate that at lea~t a fraction of the molecules produced may be non-linear. However, one observed after matrix
annealing a pile up of features at around 2300 em-I. This can be understood in terms of linear molecules, since the vibrational levels for
longer and longer chains converge at about 2300 em-I. We probably have
produced a mixture of linear - or somehow bended - chain molecules and
something like ring-molecules.
K. RoessZer : I) Comment : Similar spectra and annealing behaviour were
observed upon C+-ion implantation into alkaline halides at 5 K and
annealing at room temperature (K. Rizsler, A. Manzanares, Report Jul1924 (1984».
2) Question : Do you assume diffusion of large C3 or C4
units in 30 - 40 K Argon matrices, or may the molecules be formed by
the intermediary of single carbon atoms ?
Answer: The opinion is that the carbon molecules diffuse through the
matrix when the matrix is thermally annealed. If two carbon species
come into contact, a chemical reaction takes place by which large
molecules may form. To my knowledge, the details of the reaction
mechanisms have not yet been worked out.
Lou AZZamandoZa:
You mention that the correlated 1804/1844 cm- I
bands may be due to a cyclic species. Do you see any bands between say
1300 - and 1600-1700 em-I which is where C-C stretches in rings lie?
Answer: There are features at 1280. 1420 and 1600 cm- I • Eventhough
they appear after stronger thermal matrix annealing, these lines stay
weak compared to other features. We so far did not look for correlations in between thes.e lines. Our data seem to indicate that the
carbon-ring-molecules of the PAR type, if present at all, are not the
major species formed in our matrix-isolation experiments.
REMARKABLE PERIODICITIES IN THE MASS SPECTRA OF CARBON AGGREGATES
Pierre Joyes
Laboratoire de Physique des Solides
Universite de Paris-Sud
Batiment 510
91405 Orsay Cedex (France)
+
-
2+
ABSTRACT. Mass spectra of carbon en' en ,en aggregates produced by
various techniques show three different n regions. For n ~ 10, an alternation behaviour appears with aggregates more abundant for a given parity of n. For 10 ~ n ~ 30, the mass spectra exhibit a modulo-4 periodicity.
For n ~ 40, an alternation behaviour appears again and some species as
C60 are very intense. These results which are related to the stabilities
of the aggregates are discussed.
I. INTRODUCTION
One may choose between various ways for producing aggregates. Firstly,
it is possible to start from a compact (liquid or solid) phase and
try to divide it into small fragments. One may also
start from a
supersaturated vapor and initiate a clustering process which, when it
can be stopped, provides clusters of a given size. These two modes of
formation will be examinated in chapters III and IV. We will first
analy~e in chapter II the formation of clusters in the thermal equilibrium between a solid and its vapor.
II. VAPORIZATION
At temperatures which can be reached by classical ways of heating the
carbon vapor contains small en aggregates. This fact is rather surprising for such a refractory material (large enthalpy of vaporization
~Ho)' Noble or transition metals, where ~Ho is smaller, do not exhibit
in their vapors clusters larger than dimers for the same range of
temperatures.
Thermodynamical models (I) have shown that this carbon property was
related to the Cn dissociation energies Dn'
all the materials which are solid at ambient temperatures Dn increases with n but for carbon this increase is more rapid than for others.
This can already be seen on the value of ~Ho/D2 which is 1.19 for carbon
whereas it is larger than 2 for noble or transition metals. As a consequence, though it is not easy to vaporize carbon, aggregates of this
element easily appear in its vapor.
Fo~
85
A. Uger et al. (eds.) , Polycyclic Aromatic Hydrocarbons and Astrophysics, 85-93.
© 1987 by D. Reidel Publishing. Company.
P. JOYES
86
In the experimental procedure (2), the vapor inside a heated cavity
(2000° K < T < 3000° K) is allowed to effuse from it through a
small hole. The molecular species are post-ionized and, by using (estimated) ionization cross-sections, it is possible to plot the Pn partial
pressures versus liT (fig. I).
~
"a. -9
;
-10
-11
-12
Figure I
Partial pressures versus temperature from ref. (2).
In a simple scheme, one may say that the slopes of the fig. I curves
give
nllHo - Dn
The values obtained for (Dn - Dn-I) (with
table I.
n
2
D - D
n-I
n
6.20
3
7.58
4
4.59
5
7.28
7.37 eV) are given
~Ha
(eV)
Table I. Increase of Cn binding energy.
We observe that, when passing from Cn-I to Cn , the dissociation energy
increase is larger when n is odd.
87
REMARKABLE PERIODICITIES IN THE MASS SPECTRA OF CARBON AGGREGATES
In chapter V we will show that this even-odd alternations of Dn can be
understood. The case of other molecules, such as HCnN, NaSnNa, etc ••• ,
will also be examined.
III. FRAGMENTATION OF A SOLID PHASE
When a solid target is submitted to a bombardment with primary ions of
about 10 kV, there occurs a secondary emission of various monoatomic
and polyatomic particles in various charge states.
The mass spectrum recorded in these experiments (Secondary Ion Mass
Spectrometry : SIMS) is given fig. 2.
4
6
8
10
12
14
n
Figure 2
SIMS normalized intensities.
One observes that, for n ~ 10, odd (even) n species are more abundant
for ~ (en).
The spark source technique (high frequency discharge between graphite
electrodes) also gives interesting results (fig. 3). The even-odd
alternation appears. Similar spectra are obtained when the target is
submitted to a laser irradiation (7,8) (LIMS method). The results are
shown figure 4.
By looking at figure 3 and figure 4, one may notice that, for n
it appears a modulo-4 periodicity : C~ with :
~
10,
n = 11, 15, 19
are more abundant.
The behaviour of c~+ ions has also been investigated by using the spark
source technique (9) (figure 5).
We see t h at ~2+.~ons exhibit the same modulo-4 periodicity as C: ions.
88
P. JOYES
Figure 3
Spark source relative intensities, from (6).
100
c;
E
·c
~
4i
10
'-
:e::
·iii
c:
<II
;S
1
n
Figure 4
LIMS resul ts, from (7).
REMARKABLE PERIODICITIES IN THE MASS SPECTRA OF CARBON AGGREGATES
89
-2
-3
-7UL--+---~--~--~~
5
10
15 n 20
Figure 5
2 + ~ntens~t~es,
.
..
f rom (9) •
Cn
IV. CLUSTERING IN THE GAS PHASE
The general process is the following. The vapor is rapidly cooled by
adiabatic expansion through a nozzle, supersaturation sets in and
condensation starts. For alkalin metals it is possible to obtain vapors
in the range 10-1000 Torr, the subsequent expansion in the vacuum
(10- 5 Torr) is supersonic. The ionization potentials of Nan (n ~ 10)
(10) or Hgn (n ~ 15) (11) clusters have been measured in beams produced
by this method.
When the vapor pressures accessible by classical ways of heating are
too low other methods are necessary. The metallic vapor can be mixed
with a cold helium gas (THe ~ 100 0 K, PHe ~ 10 Torr). As it takes the
He temperature, the metallic vapor is supersaturated and cluster growth
occurs. The clusters are then transported by the gas through a nozzle
in a chamber with low pressure. Interesting results have been obtained
for Sb n (n ~ 300) (12), Sn (n ~ 10) (13), Agn (n ~ 30) (14).
In the works mentioned above the metallic vapor was produced by an oven,
a modified version of the technique has been applied to carbon where the
vapor is produced by laser irradiation (15). The recorded mass spectrum
(after post-ionization) shows clusters with n as large as about 150.
Several interesting features can be noticed
1) The modu10-4 periodicity of fig. 3, 4, 5 also
sities for n = 11, 15, 19, 23).
a~pears
(larger inten-
2) For clusters with n ~ 40, a new modulo-2 periodicity is observed
(with even species more intense).
P. JOYES
90
3) ~ong this last group, C;O and C;O have a particularly large intens~ty.
4) Between clusters of group I and 2, no clusters are detected.
The experiments have been continued in two directions
a) Bloomfield et al. (16) have isolate~ and fragmented C60. The
daughter ~ fragments present a similar behaviour as the original ones
(modulo-2 periodicity for n ~ 10, modulo-4 periodicity for 10 ~ n ~ 25).
Other fragmentation experiments of Cd clusters are described in (17).
b) The experimental conditions have been varied (19). The PRe pressure at the time of the laser irradiation has been increased and the
thermalisation of the cluster has been maximized. The result is that it
is possible to make the cto intensity completely dominant (more than
50 % of the total large cluster abundance is accounted for by C60 ).
V. DISCUSSION OF THE RESULTS
V.I Small clusters.
The alternation of the binding energy Dn have been explained by Pitzer
and Clementi (19). In this theory, the molecules are supposed to be
linear (sp hybridization) and the 4n valence electrons are shared
between three kinds of levels :
I. (n-I)a bondin b levels (corresponding to the (n-I) bonds), occupied
by 2{n-l) electrons.
2. two a terminal levels occupied by 4 electrons.
3. there remain 2n-2 electrons which occupy doubly degenerate n levels.
When 2n-2 is a mUltiple of 4 this n "band" is closed and a maximum
of stability is expected. This occurs for odd n in agreement with
experiment.
The same simple scheme can be applied to other cases.
The detection of HCnN molecules in various interstellar molecular
clouds has been made (observation of their rotational spectrum (20».
Only odd n values appear (n = 7, 9, II). Let us examine how this can
be explained.
The number of valence electrons is :
4n + I + 3
= 4n
+ 4.
If the molecule is, as before, supposed to be linear, the electrons
occupy two kinds of levels :
I. (n+l)a bonding levels where 2{n+l) electrons take place.
2. The remaining 2n+2 electrons occupy n levels and the n'~and'is closed
for odd n values, leading to a maximum of stability for this n parity.
Let us mention that the laboratory synthesis of these molecules (21)
confirm the larger stability of odd n species since only molecules with
n - 3, 5, 7 have been synthetized.
REMARKABLE PERIODICITIES IN THE MASS SPECTRA OF CARBON AGGREGATES
91
Seeded supersonic beam experiments (15) only produce KCnK molecules
with even n. By applying the same model we see that the number of
valence electrons is 4n + 2 ; 2n + 2 electrons occupy (n+l)o bonding
levels and 2n electrons are remaining for the 'IT "band" which is closed
when n is even.
The molecules : Et 3 SiCn SiEt 3
have been synthetized only
for even n values (n = 4, 6 ; Et means an Ethyl group) (21). In these
molecules the Si atoms are Sp3 hJbridized and completely saturated,
then the carbon atoms participate in (n+l)a bonds. The 2(n+l) electrons
needed for the bonds are given by carbon (2n) and silicon (2) atoms.
There remains 2n electrons which occupy the 'IT "band" which is closed
(and more stable) for even n.
Let us mention that the conclusions of this simple model have been
verified by more precise calculations for Cn (22).
V.2 Clusters with n
~
10.
One may think that, for n ~ 10, Cn aggregates adopt the Sp2 hybridization. Then, maxima of stab~lity would appear when the number of atoms
is such that it is possible to form an integer number of 6-atom rings.
As each new ring is formed by borrowing two atoms from the preceding
stable structure the number of atoms increase between two successive
shapes is 4. We then obtain 10, 14, 18, 22.
It is difficult to go further and to give the shapes of the clusters.
One possibility could be that, as the number of C atoms increases,
the cluster shape tends towards the Archimedean solids described by
Kroto (18), Haymet (23, 24) and Stankevitch (25) which, indeed, are
made up of loops of 6 or 5 atoms.
If the preceding conjecture is correct a supplementary atom is needed
to obtain the observed numbers: II, IS, 19, 23. It could be a positive
ion situated at a "central" position of the structure which would hold
the neutral part of the molecule by an ion-induced dipole interaction.
The possibility that an ion can be placed at the central position of
regular polyhedra has already been proposed (18, 23). Moreover, the
same model explain the observed maxima of stability of cri and C~+ ions
(figure 5), only the charge of the central element would differ between
the two cases.
3+
It is also noteworthy that the modulo-6 periodicity observed for Gen
and S~+ ions can be explained by similar arguments (8-atom buildingblock) (26).
REFERENCES
(I) Verhaegen G., Stafford F.E., Goldfinger P., Ackerman M.,
Trans. Far. Soc., 58 (1962) 1926.
(2) Palmer H.B., Shelef M., Chemistry and Physics of Carbon, Ed.
P.L. Walker Jr. (1968) Marcel Dekker, p. 85.
P. JOYES
92
(3) Gupta S.K., Gincerich K.A., J. Chern. Phys., 71 (1979) 3072.
(4) Gupta S.K., Gingerich K.A., J. Chern. Phys., 72 (1980) 2795.
(5) Leleyter M., Joyes P., J. de Physique,
~
(1975) 343.
(6) Dornenburg E., Hintenberger H., Z. Naturforsch., 14a (1959) 765.
(7) Furstenau N., Hillenkamp F., Nitsche R., Int. J. Mass Spectro.
and Ion Phys., II (1979) 85.
(8) Cornides I. , Int. J. Mass Spectra and Ion Phys., 45 (1982) 219.
(9) Cornides 1., Morvay L., Mass Spectrocopy II (1983) 81.
(10) Herrmann A. , Schumacher E., Waste L. , J. Chern. Phys., 68 (1978)
2327.
(11) Cabaud B., Hoareau A., M€linon P., J. Phys. D.
~
(1980) 1831.
(12) Sattler K., Muhlbach J., pfau P., Recknagel E., Phys. Lett. 87A
(1982) 415.
(13) Martin T.P., Surf. Sci. 156 (1985) 584.
(14) Hoareau A., M€linon P., Cabaud B., J. Phys.
(1985) 1731.
D~
(15) Rohlfing E.A., Cox D.M., Kaldor A., J. Chern. Phys.
~
(1984) 3322.
(16) Bloomfield L.A., Geusic M.E., Freeman R.R., Brown W.L., Chern. Phys.
Let. ~ (1985) 33.
(17) Geusic M.E., Mc Ilrath T.J., Jarrold M.F., Bloomfield L.A., Freeman
R.R., Brown W.L., J. Chern. Phys. 84 (1986) 2421.
(18) Kroto H.W., Heath J.R., O'Brien S.C., Curl R.F., Smalley R.E.,
Nature 318 (1985) 162.
(19) pitzer K.S., Clementi E., J. Am. Chern. Soc.
~
(1959) 4477.
(20) Bell M.P., Feldman P.A., Kwok S., Matthews M.E., Nature, 295 (1982)
389.
(21) Kirby C. , Kroto H.W., Walton D.R.M., J. Mol. Spectra 83 (1980) 261.
(22) Joyes P. , Leleyter M., J. Phys. 45 (1984) 1681.
(23) Haymet A. D. J. , Chern. Phys. Lett. 122 (1985) 421.
(24) Haymet A.D.J., J. Am. Chern. Soc. 108 (1986) 319.
(25) Stankevich I.V., Nikerov M.V., Bochvar D.A., Russian Chern. Rev.
53 (1984) 640.
(26) Van de Walle J., Joyes P., Phys. Rev. B32 (1985) 8381.
question (K. Roessler)
Comment : Source selection of even-odd structures may come from the
microscopie chenistry underlyinp, the formation : enereetic carbon atoms
(some eV kinetic energy from the laser excitation or the ion implantation
procedure) insert into C-H or C-C bonds (ct. ~oster session of. v.. ~oss­
ler). In ~eneral, a bridBe structure such as ~.-,¥ or ~-9 is preceding.
'C'
REMARKABLE PERIODICITIES IN THE MASS SPECTRA OF CARBON AGGREGATES
93
This intermediate may lead to elimination of unsaturated compounds
(e.g. acetylene) or to incorporation and elongation of cycles. The
decision depends on the stability of the new compound to be formed.
Thus even chain structures may be prefered due to their higher
stabilization energy.
Answer
The formation mechanism has not yet been studied for all the methods
of aggregate production mentioned above. In a simple scheme, one may
say that the emission mechanism is responsible for the general envelope
of the mass-spectrum intensities. The odd-even alternations which are
present in all the results are certainly due to the aggregate stabilities. As you Stl8Sest, the insertion of hot C atoms can playa role in
the various fo~ation mechanisms.
Reactions of Thermal Hydrogen Atoms and Energetic Hydrogen and Oxygen
Ions with Pyrolytic Graphite
V. Philipps, E. Vietzke
Institut fUr Chemie 1 (Nuklearchemie)
Kernforschungsanlage JUlich GmbH, 5170 JUlich
Federal Republic of Germany
The formation of volatile hydrocarbons by bombarding the surface of
pyrolytic graphite with thermal and energetic hydrogen atoms has been
studied above room temperature using a special massspectrometric
technique. In the reaction of thermal atomic hydrogen with pyrolytic
graphite the formation of the radical CH 3 and C2 Hx-species
(predominantly C2H4) as well as C3 H has been observed. The reaction
probability drastically increases w~en using either energetic hydrogen
ions or a bombardment with any kind of energetic ions concomitant to
the exposure to the thermal hydrogen atoms. Bombardment of the graphite
with energetic oxygen ions results in the formation of CO and CO 2 with
an overall reaction rate near one. Energetic hydrogen ions are stored
in the graphite within the range of their penetration until a critical
saturation level of H/C of 0.5 is achieved.
1. Introduction
Since graphite is used in fusion devices as a first wall material its
interaction with any kind of hydrogen and oxygen species is of
important interest for controlling the impurity levels in such machines.
Extensive investigations have been done on the formation of various
hydrocarbons by bombarding graphite with different kinds of hydrogen
species [1,2,3,4,5] and also on the interaction of oxygen ions with
graphite [6]. Furthermore, the retention of thermal and energetic
hydrogen atoms as well as energetic oxygen ions on graphite have
been investigated in recent years in order to clarify the hydrogen
and oxygen inventory in fusion machines [7,8,9]. In a next step,
"diamond-like" carbon films have been deposited on the whole first wall
of fusion machines in order to clad the complete wall with a low Z
material [10]. These films contain about 30 % hydrogen and their
physical properties differ significantly from that of pure graphite.
The purpose of this contribution is to summarize some main features of
the work on graphite which may be relevant to the discussion on
possible reactions on grains in interstellar media.
95
A. Leger et al. (ed.,.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 95-97.
© 1987 by D. Reidel Publishing Company.
96
V. PHILIPPS AND E. VIETZKE
2. Hydrogen Graphite Surface Reactions
The chemical erosion of carbon by hydrogen strongly depends on the
energy of the impinging particle. Two typical cases can be distinguished,
namely, the reaction of thermal atomic hydrogen (E < 1 eV) and that
of energetic ions (E > 100 eV).
2.1. Thermal Hydrogen Atoms
Results on carbon erosion by thermal hydrogen atoms and hydrogen plasma
are inconsistent, especially with regard to the reaction probability
[1,3]. An average trend can be observed in all the experiments that
methane production from the reaction of hydrogen atoms on graehite
takes place only below 800 K with a reaction rate of some 10- CH 3 per
incident hydrogen atom. It is generally accepted that the reaction takes
place on the edges of the carbon surface (free bonds) and proceeds
stepwise until CH 3 is formed which then desorbs. Our results have
shown that in this case a CH 3-radical leaves the graphite surface [3].
2.2. Energetic Hydrogen Ions
The results on the chemical erosion of graphite by energetic ions are
very consistent and independent of the method used [2,3,5]. A quite
sharp maximum around 800 K is observed with a methane production rate
of some 10- 2 CH~ per incident ion for energies between 5 keV and
30 keV. The appearance of the methane follows very closely the H2
reemission. Therefore it is assumed that hydrogen atoms migrating from
the bulk to the surface react with surface atoms rather than the
incident ion. In this case the CH~ molecule is formed [4].
2.3. Simultaneous Bombardment W'ith Thermal Atomic Hydrogen And
Energetic Ions
A simultaneous bombardment of the graphite with atomic hydrogen and a
small fraction of any kind of energetic particles results in a
similar formation of hydrocarbons as in the case of energetic hydrogen
ions alone (synergistic effect). In contrast to the case of hydrogen
ions, the main reaction product is again the CH3-radical [3].
3. Retention of Thermal and Energetic Atomic Hydrogen in Graphite
Thermal atomic hydrogen is trapped on the surface of p~rolytic graphite
with a decreasing trapping probability. A ~uasi saturation of the
amount trapped is reached near 4x10 15 H/cm [8,9]. It is commonly
accepted that the atomic hydrogen is trapped on the surface of the
graphite (no solution in the bulk). The trapped hydrogen desorbs as H2
during heating of the graphite in two desorption processes at about
1000 and 1450 K. Within the first desorption peak of H2, a certain
amount of CH 3 • C2Hx and C3Hx species also desorbs [8].
THERMAL HYDROGEN ATOMS, ENERGETIC HYDROGEN AND OXYGEN IONS
97
Energetic Hydrogen is stored within the graphite within its range of
penetration until a level of H/C of 0.5 is reached [7J. Further
implantation of hydrogen ions results in the release of hydrogen
molecules. The hydrogen saturated graphite layer shows significantly
different physical properties when compared with pure graphite [llJ.
\
4. Reaction of Energetic Oxygen Ions with Graphite
When graphite is bombarded with energetic oxygen ions, the incident
oxygen reacts completely to form volatile CO and CO 2 , This occurs
after completion of a transient storage of the impinging oxygen
in the near surface range of penetration [6J.
5. "Diamond-like" Dense Carbon Films
Diamond-like dense carbon films are prepared by depositing in a glow
discharge of hydrocarbon species (CH 4 , C6H6 etc.) ionized hydrocarbons
on a substrate with sufficient energy to break the original structure
of the molecule [10J. The films obtained are hard, transparent and have
a high electrical resistivity. They contain a large amount of hydrogen
bonded to the graphite. The carbon atoms are preferentially bonded by
sp3-bonds. CH, CH 2 and CH 3 type bonding of the hydrogen has been
found by infrared absorption measurements [12J.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
M. Balooch and D.R. Olander, J.Chem.Phys. 63, 4772 (1975)
S.K. Erents, C.M. Braganza and G.M. Mc-CraCKen, J.Nucl.Mat. 63,
399 (1976) and 75, 220 (1978)
E. Vietzke, K. rTaskamp and V. Philipps, J.Nucl.Mat. 111+112,
763 (1982)
E. Vietzke, K. Flaskamp, V. Philipps, J.Nucl.Mat. 128+129,
545 (1984)
J. Roth, J. Bohdansky and K.L. Wilson, J.Nucl.Mat. 111+112,
775 (1982)
E. Vietzke, K. Flaskamp, V. Philipps, T. Tanabe, to be published
W.R. Wampler, D.M. Brice, C.W. Magee, J.Nucl.Mat. 102,304 (1981)
V. Philipps, E. Vietzke, K. Flaskamp, to be publisned
P.C. Stangeby, O. Auciello, A.A. Haasz, B.L. Doyle, J.Nucl.Mat.
122/123, 1592 (1984)
1. SchlUter, E. Graffmann, L. Konen, F. Walbroek, G. Waidmann,
I. Winter and the Textor Team, 12th European Conference of
Controlled Fusion and Plasma Physics, Budapest 2-6 Sept. 1985,
Hungary
W.R. Wampler, App.Phys.Lett. 41 (4), 335 (1982)
B. Dischler, A. Bubenzer, R. K<Oidle, Solid State Communications
48. No.2, 105 (1983)
PHOTOPHYSICS,
ELECTRONIC SPECTROSCOPY AND RElAXATION OF
MOLECULAR IONS AND RADICALS WITH SPECIAL REFERENCE TO POLYCYCLIC
AROMATIC HYDROCARBONS.
Sydney LEACH
D~partement d'Astrophyslque Fondamentale,
Observatolre de Parls-Meudon,
92190 - MEUDON France.
and
Laboratoire de Photophyslque Mol~culalre du C. N. R. S. "
Unlverslt~ de Paris-Sud
91405 - ORSAY, France.
ABSTRACT
Structural and dynamic aspects of molecular Ions and radicals relevant to the
possible role of polycyclic aromatic hydrocarbons (PAH) In the Interstellar
medium are discussed. Processes of formation of radicals and Ions are
reviewed, with emphasis, for Ions, on the role of superexclted states.
Differences between open and closed shell hydrocarbon species, of
relevance to their optical spectroscopy and relaxation properties, are
discussed. Trends In the first Ionization potentials of PAWs are considered In
relation to graphite as a limiting case. Techniques for studying the electronic
spectra of molecular Ions and radicals are reviewed and general problems In
the analysis of their vlbronlc structure are discussed. Matrix effects on
electronic spectra and the effects of a number of parameters on rotational
band contours of gas phase species are considered. Intramolecular
non radiative transition studies are evoked for both radicals and Ions and this
Is followed by discussions on experimental and theoretical aspects of the
Intra- and Inter- molecular chemistry of these species. Finally, some
properties of doubly-charged molecular Ions are presented In connection with
their possible formation and destruction In the case of polycyclic aromatic
hydrocarbons In the Interstellar medium.
I. - INTRODUCTION
Structural and dynamic Information on molecular Ions and radicals are not
only of Intrinsic Interest but are also vital for Interpreting physico-chemical
processes operative In astrophysics and aeronomy, especially planetary
atmospheres and Ionospheres, comets, cool stars and the Interstellar
medium. The latter Is particularly relevant to the present workshop on the
role of polycyclic aromatic hydrocarbons In astrophysics. The present review
" Laboratolre
assocl~
6
l'Unlverslt~
de Paris-Sud.
99
A. Leger et al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 99-127.
© 1987 by D. Reidel Publishing Company.
s. LEACH
100
will discuss work on the photophyslcs, the electronic spectroscopy and the
Intramolecular relaxation of molecular Ions and radicals. Results and
properties of potential importance to the subject of this workshop will be
emphasized.
Laboratory studies of gas phase molecular Ions and radicals Involve the
following themes : (I) physics (and chemistry) of their formation, (II)
spectroscopy and structure, (III) Intramolecular nonradlatlve transitions, (Iv)
Intramolecular chemistry, (v) collisionally Induced physics, (vI) collisionally
Induced chemistry. Recent reviews of Interest to this field are as follows: a)
spectroscopy of free radicals (1-6)
b) spectroscopy and relaxation of
molecular Ions (7-12) and their applications to astrophysical problems
(13,14).
2. - RADICAL AND ION FORMATION PROCESSES
Neutral free radicals are usually formed by unlmolecular dissociation
processes or as the result of a bimolecular chemical reaction. Bond cleavage
to form radicals requires excitation energies of between 3 and 10 eV, with
some exceptions outside these limits. The physics of molecular Ion formation
concerns direct Ionization and autolonlzation processes and so mainly
Involves excitation energies greater than 6 eV. Both types of species are
usually formed by Interaction between neutral molecules and particles of
sufficient energy : photons, electrons, fast atoms or molecules, atoms or
molecules In metastable excited states, charged atoms or molecules. The
following lists molecular excitation and relaxation channels leading to the
formation of Ions and radicals by photon Impact. Analoguous channels exist
for other types of excitation. Typical rates for these processes are Indicated
In parentheses.
Primary al""a :
Ulreel Ionization (~ 1015a-l)
Aa ... hv ... AS+ ... .-
(auperexolted) alale excllallon (~ 1015,-1)
.... A+8
Dlrecl dlaaoclallon (~ 1015a-l)
... A+ ... 8 + .-
Dla,oclallve Ionization (~ 10 14,-1)
8econdary ateps :
ABo
.. AS + hVf
... AS'" .... -
Aulolonlzallon
\ Eleclronlc (1015-108 a- 1)
I
Vibrational 110 16_10 12,-1)
Eleotronlc (e 10 14.- 1)
... A+8
Predlssoclatlon
... A+ ... B + .-
Dlaaoclallve AUlolonlzallon (e 10 14,-1)
Vlbrallonal (e 1012a -l)
Secondary and Tertiary stepa :
Ion fluor ••oence <c 108.- 1)
.. A+ ...
e
Ion dlaaoclatlon (e 10 14,-1)
101
ON MOLECULAR IONS, RADICALS AND PAH's
In addition there exist analoguous processes Involving the formation,
dissociation and fluorescence of doubly-charged molecular Ions (12,14),
In the primary processes listed above some particular attention must be
paid to superexclted states and their evolution. In certain energy regions, a
molecule can be excited to electronic states of the neutral species lying
above the lowest Ionization potential. As seen above, these neutral
superexclted states can decay by various processes, amongst which is
autolonlzation which Involves coupling with the Ionization continuum (11), A
measure of the relative Importance of the nonlonlc relaxation channels of
superexclted states Is given by the photolonlzatlon efficiency YI which Is
defined as the number of Ions produced per photon Initially absorbed,
In general, It Is found that YI for molecules does not reach unity until
several electron volts above the Ionization threshold. This Implies that
nonlonlc channels have decay rates comparable (within a factor of 100) to
autolonlzatlon rates over this energy region. There Is little systematic detailed
work on the competitive processes for molecular species other than H2 (15).
Representative data on the domain of competitive non Ionic decay are given In
Table I as the difference between ET, the threshold potential for YI = 1
(accuracy 0.2 - 0.5 eV) and El the adiabatic first Ionization potential
(accuracy 0.01 eV).
TABLE I - Domain of competitive non Ionic decay processes of superexclted
8tates In some molecular species,
Molecule
=
El =
Adiabatic
first Ionization
potential
(eVl
ET
Threshold
potential
forYI=l
(eVl
(ErEll =
Domain of
competitive
non Ionic decay
(eVl
15.43
18,23
2.80
15.58
20.66
5.08
12.07
18.09
6.02
14.01
19.37
5.36
NO
9.26
17.71
8.45
H20
12.62
20.66
8.04
C02
13. 77
19.37
5.60
NH3
10.15
17.71
7.56
CH4
12.62
15.49
2.87
C2 H2
11.40
17.71
6.31
C2 H4
10.51
16. 75
6.24
C2 H6
11.56
C6 H6
9.26
It:
18
16.53
It:
6.4
7.27
102
S. LEACH
The value of ET - E 1 does not seem to correlate smoothly with the size of
the target molecule. The extent of the domain of noncompetitive decay Is
therefore specific to the Internal dynamics of electrons and nuclei of
Individual species.
3. - SOME PROPERTIES OF OPEN AND CLOSED SHELL HYDROCARBON
SPECIES
3. 1 . - Orbital and state energies and Intervals
Many molecular Ions and radicals are open-shell species. The energy
separation AE(D, - DO> between the electronic ground state (doublet state
DO> and the first excited state of the same multiplicity (doublet state 0, > will
often be less than In the nearest related closed shell species. This Is
particularly true for hydrocarbons. As an example we list In Table II the
energy of the first excited electronic state relative to the ground state of same
multiplicity for some open shell cationic species and their parent neutral
closed shell molecules of a) small hydrocarbons, b) linear acenes and c)
perlcondensed systems.
TABLE II
FIRST ELECTRONIC TRANSITION ENERGIES BETllEEN STATES OF SAME SPIN MANIFOLD
FOR CORRESPONDING NEUTRAL (N) AND CATIONIC (I) HYDROCARBONS.
TABLE IIa
Species
Acetylene
Ethylene
Sma 11 Hydrocarbons
Trans i t ion
~
C2 "2
C2 "4
N
N
5.23
A2r+ _ i 2n
g
u
4.94
AlB
A2B
Ethane
C2 "6
N
C4 "2
N
u< IE)
2g
-
Ig
C6 "6
N
AlB
A2E
g
-1 +
2u
2g
g
4.96
1.87
7.69
Ig
_ i 2E
A - X r
A2n
u
Benzene
- ilA
_ i 2B
2u
A - ilA
A2A
Diacetylene
Iran.ieion Enersx/eV
-I •
AlA - X r
u
g
I. 51
4.22
g
_ i[2 n
2.45
g
- ilA
_ i 2E
Ig
Ig
4.72
2.25
103
ON MOLECULAR IONS, RADICALS AND PAH's
TAHE JIb
Ca tacondensed Po lycyc 1 ic Aromatic Hydrocarbons.
Sp,,'cies
Formula
Napht ha len~
(X)
Transi tion
AIS
3u
- Xl A
g
3.99
lu
_ X2 A
u
0.73
- XIA
3.31
A2B
Anthracene
AIS
N
2u
;;?S
Tetracene
cxx:co
Pentacene
TAIlLE Ile
AIS
N
(Platt notation for Neutral 51
Perylene
1.12 Benzoperylene
Ovalene
Pyrene
Anthanthrene
lu
- XIA
_xa
u
g
2
u
2~
I. 10
2.64
1.4(1
2.13
I. 29
Pericondensed Polycyclic Aromatic Hydrocarbons :
Species
Coronene
g
_ X2A
2u
A2A
2&
- i(IA
2u
A2 S
P.
_ X2 S
3&
AlB
CXXO
Transi tion Ener8;~/eV
Formula
&5
N
@
N
@
N
W
0
6£?
- So
transi tions).
Transition EnerSl/eV
Transition
IL
-
a
1.55
ILb - IA
3.05
- i 2A2
N
0.67
ILa - IA
A2E
N
2.82
'A;2s 2& - li2Au
A2 S1
N
IA
Ig
_ i 2E
I~
- IA
A2B
_i
3g
2.90
2u
2.67
2s
2g
ILb -
IA
A2 S
_ i 2s
51
A2A
2g
u
I. 29
0.60
3.34
3g
0.75
2.86
50
_ i 2A
u
I. 16
s.
104
LEACH
Analogous data for some free radicals and their related closed shell parents
are given In Table III.
TABLE Ill. FIRST ELECTRONIC TRANSITION ENERGIES BETWEEN STATES OF SAME SPIN MANIFOLD
FOR SOME CORRESPONDING CLOSED SHELL PARENT (P) AND OPEN SHELL RADICAL (R)
HYDROCARBONS.
Species
formula
Transition
Transition Energy/eV
1,3 Cyclopentadiene
CSH6
A- X
4.82
Cyclopentadienyl
CSHS
A2A"2 _ i 2E"I
3.67
Toluene
C6 HSCH 3
AlB
I
4.65
Benzyl
C6 HSCH2
-2
-2
A A2 - X B2
2.73
Benzene
C6 H6
AlB
Phenyl
C6 H5
A2B
I
2u
I
- ilA
- ilA
_i
2A
I
Ig
4.72
2.34
Data for the open shell species and their parent closed shell molecules
are taken from references 16-19, ( Ions), 20-23, (closed shell neutrals),
20,24,2S (neutral radicals). It should be stressed that In these tables the
first excited electronic state does not necessarily correspond to an orbltallyallowed transition from the ground state. However, for all open shell species
given In the above tables, the first allowed transition stili lies below that of
the corresponding parent closed shell molecule.
The D1 state In open shell Ions and radicals will tend to lie below the first
dissociation limit, whereas In the corresponding closed shell parent, the first
excited singlet state S1 often has an energy close to or greater than that of
the lowest dissociation threshold, and therefore Is more subject to
predlssoclatlon. Furthermore, the lower excitation energy of the open-shell
species brings the first electronic transition to a more easily accessible
spectral region from the viewpoint of excitation (at least In laboratory
circumstances) and detection of photon emission (Fig. 1).
Another property of open shell species Is particularly relevant to the
applicability of radlatlonless transition theories, and provides a simplification
with respect to closed-shell species. This derives from the fact that In
closed-shell systems, the excited S1 state can be coupled to vlbronlc levels
of at least JWg lower-lying electronic states, namely, the first excited triplet
state T1 and the electronic ground state SQ. However, In the open-shell
species of concern to us, the vertical excitation energy with respect to the DO
state of the lowest-lying quartet state Q1 will be greater than that of D1, so
that coupling leading to radlatlonless transitions will occur between D1 and
vlbronlc levels of only Q!lJt lower-lying state, I. e. the electronic ground state
DO (Fig. 1).
105
ON MOLECULAR IONS, RADICALS AND PAH's
CLOSED SHELL
(Even electron species)
OPEN SHELL
(Odd electron species)
Q,
---
==T,
,-I
I
I
I
AB
Figure 1, Comparison between level manifolds and coupling for open-shell
and closed shell molecular species.
That Q1 will generally be higher than 01 In hydrocarbons of Interest to us
can be seen from the following discussion.
On a monoelectronlc
configurational basis, and neglecting configuration Interaction, excitation
from the DO to the 01 state In an open-shell system involves electron
promotion to the lower unfilled a orbital (3(a) or rr orbital (3(rr) In the simple
orbital schemes shown In Figs. 2a and 2b respectively. However, excitation
to the lowest quartet state Q1 would require more energy since It corresponds
to electron promotion to the lowest unoccupied orbital y.
oRBITAL
-L-
¥ -ct
~(cr)...Lcc ...tl.-
-------~
b
---
~(;n)..ut..
0(
li-
-1L
.L- -
-
-
.H1!
~
-L
-L-
-- -~
...L.1.
~
- -- ------ - --)s-
C;
~
~(n)-1-
0(1.L
Do
~
L-
D,
..1L
-L-
Q,
Figure 2.
Molecular orbital
energy patterns for doublet
(D> and quartet (Q) levels of
open-sheil species. a Is 10west filled orbital, {3 Is lowest
unfilled orbital and y Is lowest
unoccupied orbital In unexcited
species.
S. LEACH
106
Figure 2c does not correspond to the hydrocarbons listed above but Is
given here to complete the discussion. In this case, since the l3(rr) orbital
contains only one electron In the unexcited configuration, the states 01 and
01 will correspond to the same excited electronic configuration. From Hund's
rule we then expect 01 to lie below 01' One further general point Is worth
noting In cases 2a and 2b. If the bonding properties of the 13 and y orbitals
are very different, It Is then possible for the equilibrium nuclear configuration
qeq (01) to differ from that of 01 so that, If the molecular orbital energies
depended markedly on nuclear configurations, the potential energy of 01 at
Its qeq could be less than that of 01 at qeq (01)' However, vertical
transitions 01 t- DO would stili tend to lie at lower energies than the
corresponding (forbidden) 01 t- DO, so that the small Franck-Condon factors
between 01 and 01 In the region of the potential energy minimum of 01
would limit or prevent efficient Intramolecular coupling between these two
states (Fig. 3).
E!q).----------.,.---------,
q
Figure 3. Schematic representation of doublet (DO, 01) and lowest quartet
(01) potential energies E (q) as a function of an arbitrary nuclear coordinate
q.
3. 2. - First Ionization potentials of polycyclic aromatic hydrocarbons
Large polycyclic aromatic hydrocarbons have been shown to Ionize somewhat
like two-dimensional sheets or linear fragments of graphite, according to
particular structures. On the basis of semi-empirical calculations of Ionization
energies (26, 27), the first Ionization potentials of PAWs tend to approach
the sum of the graphite work function and the energy necessary to charge a
single-plate capacitor whose size and shape Is that of the molecule. The
appropriate value of the work function to be used In such calculations is
structure dependent and Is a matter of discussion (27, 28). The first
ON MOLECULAR TONS, RADICALS AND PAH's
107
Ionization potentials of PAH molecules, examined as a function of number of
rr electrons, and organized Into different point-group symmetries, makes
more clear the trends of the I. P. 's towards limiting values (29). The most
recent published data on these I. P. 's (19) confirm the earlier observation
(on a more limited number of PAH's) that of all the PAH molecules, the
slowest approach to the graphite limit (solid graphite work function = 4. S9
eV) Is for the DSh and D6h species, I. e. the most symmetrical molecules.
An Increased rate of approach to a limiting value Is observed with Increasing
linearity of the PAH series. This behaviour Indicates that the limiting values
for linear PAH's would correspond to that of a wire of graphite (27), and that
a higher limiting work function would be appropriate for the species such as
coronene, ovalene, hexabenzocoronene etc ... , which relate closer to two
dimensional. graphite sheets.
Deviations from the symmetry classification trends are observed for the
D2h species perylene, pyrene and ovalene, whose Ionization energies are
higher than expected. These deviations have been rationalized In terms of
tendency to radial or linear structure (29).
of. ELECTRONIC SPECTROSCOPY: TECHNIQUES
The spectroscopy of molecular Ions and radicals Is a basic source of
Information on their geometrical structure, Internal dynamics and electronic
configurations In various states of excitation. Classical spectroscopic
techniques successful with stable neutral species are of limited value for
molecular Ions and radicals. This Is due to the difficulty of creating densities
of these transient species that are sufficiently high and of sufficient duration
for spectroscopic measurements to be made by traditional methods. For
example, the usual maximum molecular Ion densities are .. 108 cm- S but can
be higher In some plasmas. The processes that limit the densities of specific
Ions (radicals) are (I) lon-electron (radical-radical> recombination, (II)
lon-molecule (radical-molecule> reactions and (III) spontaneous dissociation
of excited electronic states of the desired transient species. Carefully
designed Ion and radical sources are therefore necessary for spectroscopic
studies.
Although high densities of transient species are difficult to achieve, high
resolution emission spectra of many gas phase molecular Ions and radicals
have been obtained with electron Impact and discharge sources (SO).
Spectral emission of free radicals can Indeed be observed with a number of
other techniques, mainly used on small species. These Include radical
fragment fluorescence of vacuum ultraviolet excited molecules, high
temperature furnaces and flames, and electron-Ion recombination (e. g. HS+
+ e- ... HS- ... HS + h"'F) (6). Spectral congestion can be reduced using low
temperature matrix techniques (S, 8, 9) and, more recently, supersonic beam
methods (8). In the latter, Jet expansion cooling enables the gaseous neutral
species to achieve low vibrational (SO - 100 K) and very low rotational (1 5K' temperatures. The neutral molecules can be subsequently (photo)
dissociated to form free radicals (S1> or Ionized e. g. by controlled electron
Impact (S2). The rotational and vibrational temperatures of free radicals
formed In this way will depend on details of the dissociation dynamics. For
!O8
s. LEACH
parent Ions formed In supersonic beams by electron Impact there will be little
change In rotational temperature with respect to the cooled gas phase neutral
; changes In vibrational temperature on Ionization depend on the relevant
Franck-Condon vibrational factors In the Ionization process.
Both small and large free radicals, as well as molecular Ions In massselected beams (33), Ion traps (9) and In plasmas produced by Penning
Ionization (8) or by time-gated electron Impact (34) have been excited by
lasers (6). The resulting fluorescence excitation spectra provide Information
on the electronic excited state, whereas the dispersed fluorescence Informs
on the lower (ground) state (34). Photodlssoclatlon (7,35,36) and chargeexchange (37) spectroscopy of Ion beams, monitoring fragment Ions or
parent Ion attenuation respectively, give Doppler-free high resolution (e. g.
50 MHz) electronic spectra. Thermal velocities of the Ions become
Insignificant through kinematic compression ; the Ion levels are Dopplertuned to fit the coaxial laser beam frequency. With fast Ion beams (35) scans
of 100 cm- 1 or more can be achieved.
Ion and electron spectroscopies Inform about Internal energy states,
usually limited by energy resolution to electronic and vibrational levels, of
molecular Ions formed by. Inelastic collisions of neutral molecules with
photons (11) or electrons (38). Attempts made to observe electronic
absorption spectra of molecular Ions by flash discharge or pulsed electron
excitation (39) have had limited results. More successful has been the
recent use of matrix Isolation methods (40). Absorption spectra of free
radicals have been extensively studied using flash photolysis and flash
discharge techniques as well as by photochemical modulation methods In the
gas phase (6), and by various matrix Isolation devices for low temperature
work (3).
There have been many recent advances In the microwave and Infrared
spectroscopy of small molecular Ions and radicals (9,41). Although relatively
few species have been observed, they Include many of astrophysical Interest
In connection with the Identification of Interstellar and cometary molecules. In
the present review, we are mainly concerned with the electronic spectra of
molecular Ions and radicals, but It should be stressed that detailed
spectroscopic Information obtained In anyone spectral region can be of use
for estimating molecular properties and predicting spectra to be observed In
other spectral regions. This Is particularly Important for astrophysics, where
even low resolution or approximate energy prediction can help narrow the
bounds of observational search. Quantum-chemical calculations of energy
levels and oscillator strengths are also Important In this respect.
5. - ANALYSIS OF VIBRATIONAL STRUCTURE IN THE ELECTRONIC SPECTRA
OF RADICALS AND IONS
Analysis of the vibrational structure exhibited In electronic spectra of
polyatomlc molecular Ions and radicals Is more difficult than for stable neutral
species. In the latter, the ready existence of Infrared and Raman spectra
provides basic data on ground state vibrational modes and frequencies which
ON MOLECULAR IONS, RADICALS AND PAH's
109
can be used to assign features In the electronic spectra. For molecular ions
and radicals, the corresponding vibrational spectra, although increasingly
available (3,9,41) are stili In limited supply. When this Information Is
lacking, assignment of vibrational frequency Intervals to vibrational modes In
the electronic spectra of polyatomlc radicals (42-47> and Ions (48) can be
carried out by explicit use of the quasl-lsodynamlc molecule (0. I. M.)
method (49). In this method, a molecule Is chosen which Is expected to
have a force field similar to that of the radical or Ionic species considered, at
least In their respective ground states. There can then exist a good
correlation between the modes and corresponding frequencies of the quaslIsodynamlc molecule and the radical or Ion. Thus If the modes and
frequencies of the Isodynamic molecule, generally the parent molecule, are
known from infrared and Raman spectra, the vibrational frequencies observed
In the electronic (or vibrational> spectra of the Ion or radical can be
assigned to specific vibrational modes. Isotope and other substitutional
effects are used, where possible, to give further consistency and as a check
on the assignments.
As an example, let us consider the case of the polyfluorobenzene
cations. Here the O. I. M. method takes the form of first correlating the
modes and frequencies of C6FnH6-n (n = 0 - 6) neutral fluorobenzenes, and
then correlating the neutral frequencies with the observed ion frequencies.
Such correlations are Justifiable for the following reasons : 1) although the
force field changes with successive fluorine substitutions, the dominant effect
In determining this field Is that of the carbon ring ; 2) going from the neutral
parent to the Ion corresponds to removal of only one of six TT electrons on the
carbon ring, Implying that the ring force field will change little, so that the
neutral and Ion species will be quasl-lsodynamlc.
Indeed the Internal dynamics may be mOTe similar In the case of Ions to
that of the parent neutral molecule (e. g. C6F4H2+ cf. C6F4H2) than
between a radical and Its parent (e. g. C6H5-CH2 cf. C6H5CH3)' Mass
changes are zero for the Ionization case, so that one has to be concerned
only with possible modifications of force constants due to changes In bonding
character on electron ejection. This can be established, to a certain extent,
from application of the Franck-Condon principle In examining the vibrational
structure exhibited by photoelectron spectra and Rydberg bands of the neutral
parent species. Another check can be carried out by determining expected
bond length and force constant changes from bond order calculations
(45,46) .
For small polyatomlc radicals and Ions, e. g. trlatomlcs and tetratomlcs,
the small number of vibrational modes makes It usually possible to assign the
observed vibrational Intervals by simple Inspection, without the formal O. I. M.
approach. But the latter becomes very useful for large species, as has been
shown In the case of analysis of the benzyl radicals and substituted benzyl
radicals (42-47) (benzyl Is the prototype conjugated radical> and polyhalobenzene cation (48) electronic spectra. The quality of the O. I. M. approach
Is Illustrated In Tables IV and V which gives the percentage shift of the radical
or Ion frequencies from those of the parent 0.1. M. species.
110
S. LEACH
Deviations from Q. I. M. expectations occur for particular vibrational modes
when the radical or Ion electronic state exhibits vlbronlc coupling effects
Involving these vibrations. This has been used to reveal such coupling
effects, e. g. two-mode vlbronlc Interaction between neighboring excited
electronic states of benzyl-type radicals (4S~47), and Jahn-Teller coupling
In the ground electronic states of the hexafluorobenzene and 1,3, 5-symtrlfluorobenzene cations (48) and In related polyhalobenzene cations (5053). Jahn-Teller coupling effects are Illustrated In table V for the two species
CSFS+ and 1,3, 5-CSF3H3+ whose vS mode frequencies show marked
deviations from the general behaviour of the other polyfluorobenzene Ions of
lesser symmetry for which Jahn-Teller effects cannot occur. When JahnTeller effects In the D3h and DSh Ions are removed by deperturbatlon
procedures, the Q. I. M. behaviour of the deperturbed frequencies falls In line
with that of the other polyfluorobenzene species. We note that polycyclic
aromatic hydrocarbon cations having axes of symmetry of order 3 or more
(e. g. coronene cation) will have Jahn-Teller ground states. In these cases
we expect the vibrational Intervals In the ground state to exhibit Irregular
behaviour. One would not expect simple quasi-harmonic relations or
progressions of spectral bands Involving Jahn-Teller vibrational modes
TABLE IV.
QUASI-ISODYNAMIC MOLECULE BEHAVIOUR : SOME CORRESPONDING GROUND STATE
VIBRATIONAL MODE FREQUENCIES IN TOLUENE MOLECULE (M) AND BENZYL RADICAL
(R) AND RELATIVE PERCENTAGE SHIFT S (%)
Mode
S~etrl
Toluene
~
100(VM - VR)/VM·
Benzyl
S(%)
al
1003
983
+ 2.0
6a
al
514
522
- 1.6
7a
al
1212
1269
- 4.7
8a
al
1605
1606
+ O. I
9a
al
1176
1181
+ 0.4
12
al
785
814
+ 3.7
19a
ar
1494
1430
- 4.3
6b
bl
620
616
- 0.1
8b
bl
1586
1546
- 2.5
9b
bl
1154
1156
+
bl
1080
1089
+ 0.8
+
4.7
2.1
15
0.2
18b
bl
344
360
lOa
a2
843
861
+
16a
a2
405
393
- 2.0
16b
b2
464
430
- 7.3
111
ON MOLECULAR IONS, RADICALS AND PAH's
TABLE V.
QUASI-ISODYNAMIC MOLECULE BEHAVIOUR: GROUND STATE MODE 6 VIBRATIONAL FREQUENCIES
FOR SOME POLYfLUOROBENZENE MOLECULES (M) AND THEIR CORRESPONDING CATIONS
(1)
AND
RELATIVE PERCENTAGE SHIFT S (%) • 100(VM - Vr)/vW
Molecular
Molecule
"6(M)
S %
"6(1)
Point Group
Cs
1,2,4-C 6F3H3
a 441
402
+
8.8
b 503
484
+
3.8
474
457
+
3.6
C2v
C6F5 H
C2v
"1,2,3,5-C 6F4H2
458
427
+
4.6
C2v
1,2,3,4-C 6F4H2
459
442
+
3.7
D2h
1,2,4,S-C 6 F4H2
485
488
- 0.6
D3h
1,3,5-C6 F3H3
500
558
[480]
- 11.6
C6 1'6
443
D6h
494
(415)
[+ 4.0]
-
11.5
[+ 6.3]
6. - MEDIUM EFFECTS ON ELECTRONIC AND VIBRATIONAL TRANSITIONS OF
IONS AND RADICALS
Interstellar Ions and radicals exist either In the gas phase or embedded In
Interstellar grains. In general, the Interaction energy of a neutral guest
species In a matrix will be two or more orders of magnitude lower than the
bond energies In the guest. For molecular Ions In the condensed phase the
strength of Interaction can be considerably higher than for neutral radical or
molecule guest species. The magnitude of changes In the optical transition
frequencies In going from the gas to the matrix phase will depend on whether
the solute-solvent Interaction Is modified significantly between the lower and
upper states of the solute transition. Most spectral studies of radicals and
Ions In matrices have used rare-gas matrices (8,40,54), hydrocarbon
matrices (55) or halocarbon matrices (56). Neon matrix and some
hydrocarbon matrix (e. g. cyclohexane) spectra often give rise to absorption
and/or fluorescence vlbronlc bands (first electronic transition) which are
quite sharp, being usually less than 2 cm- 1 In width at low temperatures (6 K
Neon ; 77 K or 4 K cyclohexane). The same transitions In argon, krypton or
xenon or In glassy organic matrices give much broader vlbronlc bands often
over 100 cm- 1 In width. The electronic transition gas-matrix shifts can be
hundreds of cm- 1 In magnitude for matrices other than Neon. In the latter
the electronic and vibrational frequency shifts are usually small enough for
the resulting spectra to represent well the spectroscopic properties of the
Isolated guest species. Nevertheless, some significant spectral changes can
S. LEACH
112
occur due to host lattice symmetry or Interaction effects e. g. the appearance
of forbidden vlbronlc transitions (55) or modification of Jahn-Teller
Interactions (57).
The ground state vibrational fundamental frequencies of diatomic
molecules, Including many diatomic radicals, In the gas phase and In Inert
solid matrices have been complied (58) and a detailed comparison made.
For polyatomlc radicals (42-47) and Ions (8) vibrational frequency shifts are
small, generally less than 1 % In neon and/or hydrocarbon matrices at low
temperatures.
7.- ROTATIONAL BAND CONTOURS OF GAS PHASE SPECIES
In relation to the above discussion on medium effects It Is of Interest to
consider effects of temperature on gas phase rotational band contours of
vlbronlc bands of large molecular Ions and radicals. This Is of particular
relevance to possible aSSignments of the diffuse Interstellar bands to
polycyclic aromatic hydrocarbon species (59,60). The electronic spectra of
such large species would have a high density of overlapping rovlbronlc lines,
giving rise to characteristic band contours at the low temperatures of the
ISM. The exact shape of the contour and the bandwidth would depend on the
particular band carrier.
The rotational contour of a vlbronlc band, for a molecule considered as a
rigid rotor, depends on the rotational line strength part of the total matrix
element for the tranSition, and thus on the particular selection rules for the
transition under consideration. The detailed shape of the contour Is also a
function of the following seven parameters : the rotational temperature T, the
rotational constants A, Band C In the Initial lower or upper electronic state
of the transition, which determine the level Intervals, and the variations AA,
AB and AC of these constants In going from one electronic state to the other
In the transition. Other factors which may be of Importance when degenerate
levels are concerned are Corlolls coupling coefficients and, for degenerate
electronic states, spin-orbit Interactions. Level occupancy will depend on the
degeneracy of the Initial rovlbronlc state, Including nuclear spin degeneracies.
In the most general case of an asymmetric rotor, the selection rules for a
particular vlbronlc transition will essentially be governed by the direction of
the transition moment with respect to the molecular framework. If the moment
Is directed along the a, b or c principal axes, the resulting band contours
will be type A, type B, or type C, respectively ; hybrid band contours will
result If the vlbronlc transition moment has components In more than one axis
direction. The AJ
0, .t 1 selection rule will be valid for all types of bands
but other selection rules will vary according to the symmetries of the
rotational levels and will differ for the different band types (25).
=
113
ON MOLECULAR IONS, RADICALS AND PAH's
~'
'"'r.'
······~~~·~·:·:';:~/~
...
·ro
11<""
Figure 4, Benzyl absorption bands : Observed (a) and calculated Cb)
rotational contours of the A 1 type A and 6aJ type B bands. Peak frequencies
are given relative to the principal peak for each band (after ref. 61>.
"I
1~1r--------'r-----------~
o
16K LW-l.o em-I
&0
Figure 5. Low temperature,
low resolution calculated rotational
contours
of
benzyl
bands. T = 16 K, IInewldth LW
1.0 cm- 1 (after ref. 61>.
=
60
nL-__
10
~
____
~
____
~~
0.6
cm.. 1
__
~
-10
114
S. LEACH
Figure 4 shows a portion of the observed and calculated contours of benzyl
radical vlbronlc absorption bands In the visible region : I) the A 1 type A band
whose observed principal peak Is at 22329.77 cm- 1 ; II) the 6aJ type B band
whose observed principal peak Is at 22435.53 cm- 1 (61). The rotational
temperature Is 293 K and the effective spectral resolution 0.05 cm- 1 . Most
astronomical spectra are taken at rather lower resolution. Figure 5 shows the
rotational band contours of these bands calculated for a rotational
temperature of 16 K and a rotational IInewldth of 1 cm- 1 . The calculated
contours are In good agreement with those observed In laser-Induced
fluorescence excitation Jet spectra of the benzyl radical (31). The contours
are very different from the room-temperature contours. This demonstrates
that one cannot use room temperature contours for comparison (59,62) with
Interstellar absorption spectra corresponding to low temperature conditions.
However, If low temperature spectra are not available, then the molecular
parameters determined from a higher temperature spectrum can be used to
simulate a low temperature using a suitable band contour calculation
programme.
IOOC---------...,......-a-,::----------;---b:-,
t~O.O
TR=300 K
t"=-0.8
TR=300 K
50
c
d
C;-0.8
TR=25 K
50
o
-20.0
-10.0
10.0 .20.0
em":"
.10.0
o
10.0
a
2A2" - S(2E" transition of 1,3, 5-C6F3H3+' Rotational contours
Figure 6.
calculated, with neglect of spin-orbit Interaction, for two values of the
effective Corio lis coefficient (" and the rotational temperature TR : a, c for
the O:and b, d for the 6?~8.. bands. For further details see text and (57).
Figure 6 (57) shows slmu1ated rotational contours for certain emission
bands of the 2 A"2 - S(2E" transition of the 1,3, 5-C6F3H3+ Ion, for two
different values of the effective Corlolls coefficient (" and of the rotational
temperature TR' A triangle line froflle of FWHM = 0.1 cm- 1 was used In
contou.rs a and b, and 0.4 cm- In contours c and d. It Is clear that the
contours are sensitive not only to the rotational temperature but also to
a
115
ON MOLECULAR IONS, RADICALS AND PAH's
Corlolls Interactions, Polycyclic aromatic hydrocarbons of high enough
symmetry will exhibit Corlolls coupling effects In degenerate electronic states
and/'or degenerate vibrational levels,
8. - INTRAMOLECULAR NONRADIATIVE TRANSITIONS
Two main types of (nondlssoclatlve) Intramolecular nonradlatlve transitions
are very actively studied at the present time. These are electronic
non radiative transitions (ENRT> , which Involve the coupling between two or
more zero-order electronic states, and vibrational non radiative transitions
(VNRT> (usually known by the acronym IVR = Intramolecular vibrational
redistribution) In which zero-order vibrational levels are coupled together.
The theory of radlatlonless transitions has been extensively developed over
the past few years (63,64). Recent studies have shown how ENRT and VNRT
can affect each other. The role of rotational levels and coupling, In particular
via Corlolls coupling eff",cts on these processes Is a subject of active
Investigation (65,66). Most radlatlonless transition studies have been on
neutral species. ENRT and VNRT have been studied explicitly for only a few
Ions (7, 12) and free radicals. For the latter, ENRT studies have been
mainly restricted to Interpretation of spectral perturbations (6,46), whereas
for molecular Ions (as for closed shell neutrals), electronic nonradlatlve
transitions have been Investigated not only through high resolution
spectroscopy (34) but also by quantum yield and lifetime measurements (1014, 34). Interelectronic coupling effects on lifetimes have also been studied
for some radicals, among which are 602 (67), N02 (68) and benzyl (69).
Vibrational non radiative tansltlons are assumed to occur very rapidly In
the excited states of polyatomlc Ions e. g. In the quasi-equilibrium theory of
mass spectra (70). However, explicit study of VNRT In Ions Is limited to the
case of C6F6+ (71).
The fluorescence quantum yield QF and lifetime Tm are related to the
electronic non radiative knr and radiative kr relaxation rates of an excited
electronic state by the expressions kr
.F Tm -1 and knr
(l-.F) Tm -1.
More complex relations exist when the decay rate Is multlexponentlal (72).
=
8.
t. - Meaaurement
=
of excited atate re/llJ(atlon parameteflJ
Measurement of excited state lifetimes have been made for free radicals using
laser excitation techniques (67-69) but quantum yield determinations are rare
because of the difficulty of knowing the radical concentration and photon
absorption factors. In molecular Ions both .F and Tm can be measured with
relative ease, using coincidence counting techniques which obviate the need
for knowing Ion concentrations. Radlatlonless transitions will therefore be
discussed here mainly for molecular Ions.
Table VI lists a number of relevant coincidence methods and Includes
those used to study dissociative channels. Fluorescence lifetimes and
quantum yields averaged over occupied vibrational levels of a particular
excited state are measured by PIFCO. and of specific vibrational levels, with
an energy resolution of 30 - 100 meV, by PEFCO and T-PEFCO. The PIFCO
technique can measure very low fluorescence quantum yields (e. g. 802+ : 6
x 10- 5 (78» and Is particularly useful for showing whether an Ion fluoresces,
116
S. LEACH
TABLE VI.
COINCIDENCE TECHNIQUES USED FOR STUDYING INTRAMOLECULAR RELAXATION PROCESSES
IN MOLECULAR CATIONS.
Acronym
PIFCO
Coincidence between
(ref. 73)
!hoto!on-!luorescence Photon
PEFCO
(refs. 74,75)
T-PEFCO
(ref. 76)
PEPICO
(ref. 77)
PhotoElectron-Fluorescence
Photon
Threshold PhotoElectronfluorescence photon
!hoto!lectron-!hoto!on
T-PEPICO (ref. 77)
PIPICO
(ref. 117 )
Threshold-PhotoElectronPhotoion !hotolon-!hoto!on
Channel studied
Fluorescence of mass
aelected ion
Fluorescence of energy
aelected ion
Fluorescence of energy
ae lec ted ion
Fragmentation of energy
selected ion
Fragmentation of energy
selected ion
Fragmentation of doubly
charged ion
as a preliminary to spectroscopic stUdies. The use of a pulsed synchrotron
radiation source enables threshold photoelectrons to be determined by hlghthroughput tlme-of-fllght methods. The narrow pulse makes It possible to
measure lifetimes as short as 1 ns (76). The threshold photoelectron
techniques enable autolonlzatlon processes to be studied (11).
8.2- Radlatlon/ess transitions : the statlstlca/ limit case
Intramolecular electronic nonradlatlve transitions fall Into a number of cases
which reflect Inc.reaslng density of Interacting levels (63). Full theoretical
discussion of the whole range of cases for molecular Ions Is given elsewhere
( 11 , 72). A brief presentation will be given here only for the statistical limit
case, which would apply to large species such as polycyclic aromatic
hydrocarbons.
The optically excited zero-order state IS> (e. g. In Fig. 1, D1 for open
shell and 81 for closed shell species) Is coupled nonradlatlvely to vibrational
states { Ii>} of a lower electronic level. For Ions the matrix element vsl
Involved In electronic nonradlatlve transitions will generally Involve only the
nuclear kinetic energy operator but, often for closed shell species, and In
general where the transition Involves spin multiplicity changes, there will be
contributions from spin-orbit op-erators. In the statistical limit, the density of
( Ii)},
»10 5 states/ cm- 1 , so that the ( I i ) } states form a quaslcontinuum. Electronic nonradlatlve transitions are then virtually Irreversible
so that the Fermi-Wentzel Golden rule radlatlonless rate Is given by the
expression knr
(2ul11) €t vsi 2 when vs.l varies slowly enough to be
represented by Its average value. The decay of the IS) state Is then
monoexponentlal.
The relaxation of several molecular Ions has been studied within the
statistical limit context. For the benzene Ion (79), knr was found to be
greater than 8 x 10 10 s-l for the ~2E29 state and greater than 5 x 10 12 s-l
el
=
ON MOLECULAR IONS, RADICALS AND PAH's
117
for e2A2u' These excited states decay by Interelectronic e, 8w.r"> X
coupling and by Isomerlsatlon to Dewar Benzene and/or benzvalene cations,
Particularly Interesting are the polyfluorobenzene cations whose excited state
electronic (8,71,75,81) and Intramolecular vibrational (71,75,81) relaxations have been studied In detail. Vibrational mode selective effects In ENRT
have been observed (71,75). The high vibrational levels In the ground state
of benzenold Ions, which are the final states of ENRT In these species, might
be Involved In Isomerization processes In some cases. Otherwise the other
most probable collision less relaxation process would be Infrared radiative
emission. This appears to have a rate of the order of 10 s-l for the benzene
Ion ground state containing 2.4 - 2.7 eV vibrational energy (82). Indeed a
study of several polyatomlc Ions by different techniques (83,84) Indicates
that the collision less relaxation of 1-2 eV of Internal energy occurs at rates of
the order of 1 - 100 s-l. Stili higher rates have been suggested for Infrared
radiative relaxation of vibrational levels at 3. 5 eV above the zero-vibration
level of the ground state of chloroacetylene cations (72,85),
9. - INTRAMOLECULAR CHEMISTRY
Particular examples of radlatlonless transitions occur In Intramolecular
chemistry, I. e. Isomerization and unlmolecular fragmentation. The theoretical approach to these problems has mainly been cast In a statistical
framework, e. g. In the quasi equilibrium theory of mass spectra (70) and In
Its modern refinements (86). As mentioned earlier, In such approaches It Is
assumed that molecular Ions In excited electronic states rapidly convert their
energy Into vibrational energy of the ground electronic state. This vibrational
energy Is more or less randomized before Isomerization and/or Herzberg
case II (20) vibrational predlssoclatlon takes place. For these chemical
changes to occur, sufficient vibrational energy must be made available and,
furthermore, the Intramolecular dynamics must lead to the new Isomeric or
dissociative nuclear configuration at a rate faster than any other competitive
process tending to deactivate the newly attained high vibrational levels of the
ground state. Similar processes can obviously occur for excited neutral
species such as free radicals. The dissociative processes tending to
deactivate high vibrational levels of the ground state are energy transfer via
collisions and energy loss through rovlbratlonal radiative emission. The rates
of the latter process have been presented earlier.
With respect to the rates of the processes discussed here, physical
conditions In the Interstellar medium are virtually collision-free. Thus the
stability of large molecules, radicals or Ions of the PAH type, either on
formation or on subsequent photon or particle excitation, will depend critically
on the competition between the rate of reaching the fragmentation region of
the potential energy hypersurface and the rate of Infrared radiative decay
bringing the Internal energy content of the species below that necessary for
fragmentation.
Recent theoretical work In the area of unlmolecular dissociation has paid
attention to the role of the rotational degrees of freedom (87) and to general
questions Involving Intramolecular vibrational energy flow (88,89). There Is
also renewed Interest 'In Isomerization (90) and In energy partitioning In
unlmolecular dissociation to several products (91,92). The latter process
118
S. LEACH
has become easier to study through the Increased availability of high energy
photon sources such as synchrotron radiation.
Some recent theoretical studies have gone beyond the purely statistical
approach, for example In studies of the Interaction (nonadlabatlc coupling)
between potential energy surfaces (93,94), carrying out specific semiclassical trajectory calculations on the energy surfaces Including nonadlabatlc
Interactions between surfaces (95) and quantum mechanical approaches
(96,97). The time dependent process of dissociation can also be described
In terms of time-correlation functions Instead of rate constants. Spectroscopic line or band shapes provide the basic data for this approach since the
correlation functions can be obtained from their Fourier transforms (98).
This is an active area of research on intramoiecular relaxation processes
(99,100) .
Experimental work in the area of Intramolecular ion chemistry relies on a
wide range of mass spectrometric techniques and methods of molecular Ion
preparation (10,101). The most specific data on Intramolecular Ion
chemistry comes from the use of state selection techniques, mainly using
coincidence methods, as well as the measurement of product Internal and
kinetic energies.
Much recent work on unlmolecular reactions on small species has
concentrated on state-to-state studies. This requires a knowledge of the
Initial excited state -Its total energy, Internal energy, (ro) vlbronlc symmetry,
etc. .. - and a means of determining the nature of the fragment products,
their Internal energy states and corresponding symmetries, and the kinetic
energy release In the dissociative process. This Information, coupled if
possible with adequate theoretical descriptions, can provide detailed
knowledge of the Intramolecular dynamics leading to dissociation. If to these
scalar properties are added vector properties (fragment angular distributions,
rotational alignment, orientation) deriving from polarization and other studies
( 102, 103), a complete state-to-state study becomes theoretically possible.
The Information required, methods and their limitations for obtaining the data
have been described and compared for neutral molecules, singly-charged
and doubly-charged Ions (104).
to. -
COLLIS/ONALLY INDUCED PHYSICS AND CHEMISTRY
Collisional studies involving molecular Ions and radicals cover a very wide
range of techniques and objectives. For Ions, the techniques Include the use
of Ion traps, flow tube methods, molecular beams, the whole paraphernalia
of coincidence techniques and particle and/or photon detection devices
(7, 101). Relatively recent developments In lon-molecule Inelastic and elastic
reaction studies Include the use of chemical (105) and Penning ionization,
Improved state-selection methods (106) and the application of collisional
activation mass spectrometries (107). Of speciai interest to astrophysics is
the development of techniques for the measurement at low temperatures of
lon-molecule reaction rates (101, 108-111). Binary lon-molecule reactions
are of direct astrophysical interest since such reactions are considered to be
Important in cosmochemical schemes (112). Ternary (three-body) lonmolecule reactions, although common in laboratory pl'asmas, would be rare
astrophysical events. However they are of considerable Interest In connection
ON MOLECULAR IONS, RADICALS AND PAH's
119
with estimations of the rates of binary radiative association reactions whose
Importance In cosmochemistry Is under discussion (11 S, 114) ,
The detection of free radicals Is the key to the study of reactions In which
they are Involved, The methods Include the whole range of optical, Infrared,
microwave and magnetic resonance spectroscopies as well as mass
spectrometry. Sometimes the radical to be detected Is converted Into another
species whose detection Is more readily accomplished. There also exist a
number of specific chemical or physical means of detection of radicals,
which are mainly nonspectroscoplc In nature. Detection methods for gas
phase radicals have been recently reviewed (115).
11. - DOUBLY CHARGED MOLECULAR IONS
Doubly-charged molecular Ions are not usually considered as taking part In
astrophysical processes. This Is due to Insufficient laboratory-based
Information on these species as well as to the difficulty In distinguishing
between processes originating In doubly and In singly-charged molecular
Ions. One must also consider whether astrophysical conditions exist which
are propitious for the formation of non-negligible amounts of doubly-charged
cations. The properties of these species have been discussed In the context
of their possible formation and detection In astrophysical conditions (14).
Some of these properties and related laboratory studies will be recalled here
In relation to their possible significance to polycyclic hydrocarbons In the
Interstellar medium.
The formation of doubly-charged molecular cations requires energies of
the order of 20-40 eV with respect to the neutral ground state. It Is worth
noting that for many molecules the ratio of double to single Ionization
potential Is =2. 8 (116). Laboratory methods of formation are by electron
Impact, Ion Impact, double charge transfer (e. g. H+ + AB .... H- + AB++),
charge stripping and by photon Impact, (117, 118). Energy deposition Is well
defined In photon Impact In contrast to most of the other techniques.
The lowest dissociation limit lies ~ the double-Ionization threshold,
due to Coulomb repulsion forces (Fig. 7). The consequent Instability can
lead to rapid dissociation. However, for efficient potential barriers, decay of
metastable state, by tunneling, can be slower. In principle, relatively longlived metastable excited states can fluorescence and have been shown to do
so for N2++ (119) and NO++ (120). Metastable doubly-charged molecular
cations (lifetime greater than ... 1 I'S) have been detected by electronImpact mass spectrometry (121) and from mass spectrometry with soft X-ray
excitation (122) and VUV photon excitation (116,123).
Electron-Impact mass spectra and Ionization energies of some polyphenyls (124) and condensed-ring aromatic and heterocyclic compounds
(29) have been measured.
The condensed-ring mass spectra are
characterized by few fragment Ions, the prominence of multiply Ionized
parent, and few metastable transitions. For families of related compounds,
there Is a tendency for the relative abundance of the doubly-charged parent
Ion to Increase, and Its appearance potential to decrease, with an Increasing
total number of ". electrons. The percentage ratio of doubly-charged to
singly-charged parent Ions of polycyclic aromatic hydrocarbons Increases
rapidly, and quasi-linearly, with Increasing number of ". electrons. The
s. LEACH
120
quasi-linear relation Is perturbed by structural effects (29), but It Is of
Interest that for the largest PAH studied In this context, hexabenzocoronane,
which has 42 Tr electrons, the ratio of doubly to singly-charged Ions Is quite
high, about 60 ~.
Figure 7 Illustrates the fact that the doubly-charged cation, formed by
removal of two valence electrons, may have- a different equilibrium structure
from that of the parent neutral molecule, so that a vertical transition might
lead directly to the dissociative part of the doubly-charged Ion potential
surface. The propensity for dissociation by direct or tunneling processes has
been used, via photolon-photolon coincidence (PIPICO) measurements
(Table VI), as a specific tool for spectroscopic and dissociative relaxation
studies on doubly-charged molecular Ions (104,117,118,125-128>.
E
r(A-Bl
Figure 7. Schematic representation of potential ener~y surfaces of a neutral
(AB), singly Ionized (AB+) and doubly Ionized (AB +) molecule.
In addition to AB++ ..... A+ + B dissociation, a second fragmentation
pathway 15 also possible, namely AB++ ..... A++ + B. Both types of reaction
have been observed but dissociation to two singly charged products 15
expected to be more probable than to one doubly-charged fragment, at
least for the lower electronic states OT AB++ (117).
Doubly-charged molecular Ions could be formed In the Interstellar medium
by Interaction of molecules with high energy photons and/or cosmic ray
particles. Ultraviolet radiation emitted by a star embedded In a dense cloud 15
absorbed In the Immediate neighbourhood of the star by the H atom Ionization
continuum at " < 912 A (13.6 eV). However, the H atom Ionization cross
section at " < 100 A (124 eV) Is sufficiently small for photons having these
energies to penetrate great depths of Interstellar space. These soft X-rays, If
121
ON MOLECULAR IONS, RADICALS AND PAH's
of suffIcIent flux, as well as the low energy components Of cosmIc rays, could
playa role In creatIng doubly-charged molecular Ions In the laM, as they do
In the formatIon of multIply-charged atoms (129). Another possIbilIty for the
formatIon of doubly-charged molecular Ions Is by the sequentIal charge
strIppIng process AB -f AB+ -f AB++. ThIs has relevance to polycyclic
aromatIc hydrocarbons In that for the larger members of these specIes, the
IonIzatIon energIes AB -f AB+ and AB+ -f AB++ can each be less than 13.6
eV. Table VII gIves the first (f+) and second (f++) IonIzatIon potentIals of a
number of PAH's as well as the dIfference 1++ - 1+ whIch represents the
energy needed for formatIon of a doubly-charged PAH Ion from a slnglycharged PAH. The data show that the first IonIzatIon potentials are all less
than 13.6 eV and that for the larger PAH's the dIfference 1++ -1+ Is also less
than 13.6 eV.
lAHLI:: VB.
IOSlZATION ENERCIE~ (eV) FOR fORMATION Of' SINCLY
AND DOUBLY CllARCEU POLYCYCLIC
AROHATlC HYDROCARBON CATIONS.
Specie. X
Benzene
Naphthalene
Anthracene
Phenanthrene
1
+
• X"X
+
1++ • X .x++
1++ - 1+ • X· .. x++
C6 H6
9.24
26.4
11.16
CIOIIa
8.15
22.1
14.55
C14"10
1.41
21.1
13.63
C14"10
1.86
23.1
15.24
Pyrene
C 16"10
1.41
24.0
16.59
letracene
C 18"12
1.04
22.14
15.10
Tetraphene
C I8 H12
1.41
22.03
14.56
23.33
15.73
Chry.ene
C18"12
1.60
Triphenylene
C 18"12
7.89
24.10
16.21
Per),lehe
C20HI2
1.00
20.0
13.00
Pent.cene
C22 HI4
6.14
19.6
12.86
C22 "14
7.38
20.8
13.42
C22"14
1.54
21.5
13.96
C24"12
1.36
21.0
13.64
12.05
12.14
1,2,5,6 Dibenl.athracene
PiceDe
CorOhene
1,2,8.9 Dibenzopentacene
C30 H22
6.95
19.0
Ovaleoe
C)2"14
6.86
19.6
Dec.eye lene
C36"18
7.27
20.1
12.83
Kexabenzocoronene
C42"18
1.05
19.6
12.55
It has been postulated that PAH's In the laM could exist largely In slnglyIonized form ; ion percentages of 50 % or more have been estimated
122
S.LEACH
( 59, 60). The Ionization cross sections AB ... AB+ and AB+ ... AB++ should be
of the same order of magnitude, It: 100 Mb (14) (they could be somewhat
less Just above threshold because of the possibility of competing nonlonlz8tlon
processes, as discussed In section 2). Although the values will depend
strongly on particular regions of the ISM, the Interval between photon
absorption In the ISM will be taken here as being between 3 x 10-4 and 3
years, and the Interval between H or H2 collisions with 8 PAH species to be
of the order of 3 x 10-2 year or greater. Thus It appears possible for doublycharged PAH's to be formed by a sequential two-photon process. As
mentioned earlier, doubly-charged molecular Ions are fragile and easily
subject to dissociation via direct or tunneling processes. Unit probability of
dissociation via tunneling would exist within the collision Interval of 3 x 10-2
year for a tunneling rate of the order of O. 1 s-l. If direct dissociation of the
newly formed doubly-charged PAH Ion does not occur, the doubly charged
PAH cation would be trapped In a potential well, losing Its rovlbratlonal
energy by Infrared radiative processes In competition with dissociation by
tunneling. Fragmentation by a tunneling process, even at very low tunneling
rates (e. g. In the vibration less state), could provide significant mechanisms
for destruction of these species on cosmic tlmescales. There exists very little
laboratory work of relevance to the fragmentation processes and v!elds of the
doubly charged cations of polycyclic aromatic hydrocarbons. However, a
PIPICO study of the prototype cyclic aromatic hydrocarbon, benzene (130),
Indicates that the mean rate for the dissociation process C6H6++ ... C5H3+ +
CH3+ Is 5 x 10-6 s-l. A considerable extension of such studies, In particular
to larger PAH's, Is timely.
References
1. -
G. Herzberg, The Spectra and Structure of Simple Free Radicals,
Cornell University Press (1971>.
2.- A. Carrington, Microwave Spectroscopy of Free Radicals. Academic
Press, London, (1974).
3.- M. E. Jacox, Rev. Chem. Intermediates, 2, 1 (1978).
4.- G. W. Hills, In Magnetic Resonance Rev., In press.
5.- E. Hirota, J. Phys. Chem., 87, 3375 (1983).
6.- O. A. Ramsay, Vibrational Spectra and Structure. (J. A. Ourlg, ed.),
14, 69 (1985).
7.- S. Leach (ed.) Molecular Ion Studies, A special Issue of J. Chlm.
Phys., 77, 585-777 (1980).
8.- T. A. Millar and V. E. Bondybey (ads.), Molecular Ions: Spectroscopy,
Structure and Chemistry. North-Holland, Amsterdam, (1983).
9.- J. Berkowitz and K. O. Groenveld (eds.), Molecular Ions : Geometric
and Electronic Structures, Plenum, N. Y. (1983).
10.- M. T. Bowers (ed. ), Ga~ Phase Ion Chemistry. vol. 3 : Ions and Light,
Academic Press, N. Y. "( 1984).
11. - S. Leach In Photophyslcs and Photochemistry In the Vacuum Ultraviolet,
(S. McGlynn, G. Findley and R. Huebner eds.), Reidel, Oordrecht
(1985), p. 297.
12.- S. Leach, J. Molec. Structure, 141, 43 (1986).
ON MOLECULAR IONS, RADICALS AND PAH's
123
13. - S. Leach In Photophyslcs and Photochemistry above 6 eV, (F. Lahmanl
ed.), Elsevier, Amsterdam, (1985), p. 113.
14. - S. Leach In Molecular Astrophysics : State of the Art and Future
Directions (G. H. F. Dlercksen, W. F. Huebner and P. W. Langhoff
eds.), Reidel, Dordrecht, (1985), p. 353.
15. - R. S. Berry and S. Leach, Adv. Electron. Electron Phys., 57, 1
(1981) .
16. - D. W. Turner, C. Baker, A. D. Baker and C. A. Brundle, Molecular
Photoelectron Spectroscopy, Wiley, London (1970).
17. - K. Kimura, S. Katsumata, Y. Achlba, T. Yamazaki and S. Iwata,
Handbook of Hel Photoelectron Spectra of Fundamental Organic
Molecules, Japan Scientific Societies Press, Tokyo (1981).
18. - J. W. Rabalais, Principles of Ultraviolet Photoelectron Spectroscopy,
Wiley, N.Y. (1977).
19. - E. Hellbronner and J. P. Maler In Electron Spectroscopy : Theory,
Techniques and Applications (C. A. Brundle and A. D. Baker eds.),
vol. 1 (1977), p. 205 ; A. Boschl, E. Clar and W. Schmidt, J.
Chem. Phys., 60, 4406 (1974).
20. - G.
Herzberg,
Electronic Spectra of Polyatomlc Molecules,
Van
Nostrand, Princeton (1966),
21. - H. Okabe, Photochemistry of Small Molecules, Wiley, N. Y. (1978).
22. - E. Clar, Polycyclic Hydrocarbons, Academic, London, vol. 2 (1964).
23. - J. Kahane-Palllous and S. Leach, J. Chlm. Phys., 55, 439
(1958) .
24. - R. Engleman Jr. and D. A. Ramsay, Can. J. Phys., 48, 964 (1970>.
25. - C. Cossart-Magos and S. Leach, J. Chem. Phys., 56, 1534 (1972).
26.- F.T. Smith, J. Chem. Phys., 34,793 (1961).
27. - M. E. Wacks, J. Chem. Phys., 41, 1661 (1964).
28. - A. S. Becker and W. E. Wentworth, J. Am. Chem. Soc., 85, 2210
(1963) .
29.- E.J. Gallegos, J. Phys. Chem., 72,3452 (1968).
30. - D. Cossart, J. Chlm. Phys., 76, 1045 (1979) ; 78, 703, 711 (1981).
31. - M. Heaven, L. DIMauro and T. A. Miller, Chem. Phys. LeU., 95, 347
(1983) ; P. Chen, S.D. Colson, W.A. Chupka and J.A. Berson, J.
Phys. Chem., 90, 2319 (1986).
32. - J. Rostas and R. P. Tuckett, J. Molec. Spectrosc., 98, 77
(1982) .
33. - A. D. Brown, P. D. Godfrey, D. C. McGilvery and J. G. Crofts, Chern.
Phys. Lett., 84, 437 (1981).
34. - M. A. Johnson, R. N. Zare, J, Rostas and S. Leach, J. Chem. Phys.,
80, 2407 (1984).
35. - M. Larzllll4)re, M. Carre, M. L. Gaillard, J. Rostas, M. Horanl and M.
Velghe, J. Chlm. Phys., 77, 689 (1980) ; S. Abed, M. Broyer, M.
Carre, M. L. Gaillard and M. Larzllll4)re, Chem. Phys., 74, 97
(1983) .
36. - C. P. Edwards, C. S. MacLean and P. J. Sarre, Molec. Phys., 52, 1453
(1983) .
37.-A. Carrington, D.R.J. MlivertonandP.J. Sarre, Molec. Phys., 35,
1505 (1978).
124
S. LEACH
38. - J. H. D. Eland, Photoelectron Spectroscopy, 2nd Edition, Butterworths,
London, (1984).
39. - G. Herzberg, Quart. Rev. Chern. Soc., 25, 201 (1971>.
40. - V. E. Bondybey, J. H. English and T. A. Miller, J. Chern. Phys., 81,
455 (1980) ; S. Leutwyler, J. P. Maler and U. Splttel, Molec. Phys.,
51, 437 (1984).
41. - Proc. 17 th Int. Symp. Free Radicals, National Bureau of Standards
Publication (1986), In press.
42. - L. Grajcar and S. Leach, J. Chlm. Phys., 61, 1523 (1964).
43. - S. Leach and E. Mlglrdlcyan, C. A. Acad. ScI. Paris, 264, 156
(1967> .
44. - S. Leach, A. Lopez-Camplllo, .1. Lopez-Delgado and M. C. TomasMagos, J. Phys., PaMs, 28, C3-147 (1967).
45. - L. Watmann-Grajcar, J. Chlm. Phys., 66, 1023 (1969).
46. - C. Cossart-Magos and S. Leach, J. Chern. Phys., 64, 4006
(1976) .
47. - A. Despr~s, S. Leach, V. Lejeune and E. Mlglrdlcyan, Radlat. Phys.
Chern., 15,441 (1980).
48. - C. Cossart-Magos, D. Cossart and S. Leach, Molec. Phys., 37, 793
(1979) .
49.- S. Leach, J. Phys., Paris, 28, C3-134 (1967).
50. - C. Cossart-Magos, D. Cossart and S. Leach, Chern. Phys., 41, 345
(1979) .
51. - C. Cossart-Magos, D. Cossart and S. Leach, Chern. Phys., 41, 363
(1979) .
52. - C. Cossart-Magos and S. Leach, Chern. Phys., 48, 329 (1980).
53. - C. Cossart-Magos and S. Leach, Chern. Phys., 48, 349 (1980).
54. - R. Rossetti and L. E. Brus, Rev. ScI. Instr., 51, 467 (1980).
55. - S. Leach, Mem. Soc. Roy. ScI. LI~ge 5~me stlrle, 20, 179 (1970)
S. Leach, Pure and Appl. Chern., 27, 457 (1971>.
56. - T. Shlda, H. Kubodera and Y. Egawa, Chern. Phys. Lett., 79, 179
(1981) .
57. - D. Klapsteln, S. Leutwyler, J. P. Maler, C. Cossart-Magos, D. Cossart
and S. Leach, Molec. Phys., 51, 413 (1984).
58. - M. E. Jacox, J. Molec. Spectrosc., 113, 286 (1985).
59. - G. P. van der Zwet and L. J. Allamandola, Astron. Astrophys., 146, 76
(1985) .
60. - A. Ltlger and L. d'Hendecourt, Astron. Astrophys., 146, 81
(1985) .
61. - C. Cossart-Magos and W. Goetz, J. Molec. Spectrosc., 115, 366
(1986) .
62. - A. C. Danks and D. L. Lambert, Mon. Not. Roy. Astron. Soc., 174,
571 (1976).
63.- K.F. Freed, Top. Appl. Phys., 15, 23 (1976).
64. - S. H. Lin (ed.), Radlatlonless Transitions, Academic Press, N. Y.
(1980) .
65. - C. S. Parmenter, Faraday Disc. Chern. Soc., 75, 7 (1983).
66. - E. Rledle and H. J. Neusser, J. Chern. Phys., 80, 4686 (1984).
67. - A. Hodgson, J. Chern. Soc. Faraday Trans. (2) 81, 1445
(1985) .
ON MOLECULAR IONS, RADICALS AND PAH's
125
68. - D. L. Monts, B, Soep and A. N. Zare, J. Molec. Spectrosc., 77, 402
(1979) ; G. Persch, H.J. Vedder and W. Demtr6der, Chern. Phys.,
105, 471 (1986).
69. - C. Cossart-Magos and S. Leach, unpublished work.
70. - H. M. Rosenstock, Adv. Mass. Spectrom., 4, 523 (1968).
71. - G. Dujardln and S. Leach, Faraday Discuss. Chern. Soc., 75, 23
(1983) .
72. - S. Leach, G. Dujardln and G. Ta'ieb, J. Chlm. Phys., 77, 705
(1980) .
73. - G. Dujardln, S. Leach and G. Ta"ieb, Chern. Phys., 46, 407 (1980).
74. - J. P. Maler and F. Thommen, Chern. Phys., 51, 319 (1980).
75. - D. Wlnkoun, D. Chapoulard, G. Dujardln and S. Leach, Can. J.
Phys., 62, 1361 <1984>76. - G. Dujardln, S. Leach, O. Dutult, T. Govers and P. M. Guyon, J.
Chern. Phys., 79, 644 (1983).
77.- T. Baer, Adv. Chern. Phys., 64,111 <1986>78. - G. Dujardln and S. Leach, J. Chern. Phys., 75, 2521 (1981>.
79. - O. Braltbart, E. Castellucci, G. Dujardln and S. Leach, J. Phys.
Chern., 87, 4799 (1983).
80. - J. P. Maler and F. Thommen, Chern. Phys., 57, 319 (1981>.
81. - G. Dujardln and S. Leach, J. Chern. Phys., 79, 658 (1983).
82. - R. C. Dunbar, Chern. Phys. Lett., 125, 543 (1986).
83. - P. N. T. van Velzen and W. J. van der Hart, Chern. Phys., 61, 325
<1981> .
84.- A.C. Dunbar, J. Phys. Chern., 87, 3105 (1983).
85. - G. Dujardln, S. Leach, G. Ta'feb, J. P. Maler and W. M. Gelbart, J.
Chern. Phys., 73, 4987 (1980).
86. - C. lifshitz, Int. Rev. Phys.· Chern. (1986) In press.
87. - M. Shapiro and A. Bersohn, Ann. Rev. Phys. Chern., 33, 409 (1982).
88. - M. Quack In Energy Storage and Redistribution In Molecules, (J.
Hinze, ed.), Plenum, N. Y. (1983) p. 493.
89. - See contributions to Faraday Discuss. Chern. Soc., n° 75,
Intramolecular klnettcs, (1983).
90.- B.J. Berne, Chern. Phys. Lett., 107, 131 (1984).
91. - T. Baer, A. E. De Prlsto and J. J. Hermans, J. Chern. Phys., 76, 5917
(1982) .
92. - G. Dujardln, T. Govers, S. Leach and D. Wlnkoun, In Photophyslc8
and Photochemistry above 6 eV, ( F. Lahmanl, ed. ) , Elsevier,
Amsterdam (1985) p. 163.
93. J. C. Lorquet,. D. Dehareng, C. Sannen and G. Raseev, J. Chlm.
Phys., 77 , 719 (1980).
94. H. KOppel, W. Domke and L. S. Cederbaum, Adv. Chern. Phys., 57,
59 <1984>95. E. M. Goldfield, P. L. Houston and G. S. Ezra, J. Chern. Phys., 84,
3120 (1986).
96. W. M. Gelbart, Ann. Rev. Phys. Chern., 28, 323 (1977).
97.,- P. Brumer and M. Shapiro, Adv. Chern. Phys., 60, 371 (1985)
G. G. Ballnt-Kurtl and M. Shapiro, Adv. Chern. Phys., 60, 403
(1985),
98. E. J. Heller, Accounts Chern. Res., 14, 368 (1981).
126
99. 100. 101 . 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114.115. 116. 117. 118. 119. 120. 121. 122. -
s. LEACH
A. J. Lorquet, J. C. Lorquet, J. Delwlche and M. J. Hubln-Franskln, J.
Chem. Phys., 76, 4692 (1982),
J. E. Reutt, L. S. Wang, J. E. Pollard, D. J. Trevor, Y. T. Lee and
O. A. Shirley, J. Chem. Phys., 84, 3022 (1986).
M. T. Bowers ( ed. ), Gas Phase Ion Chemistry vols.
and 2,
Academic Press, N. Y. (1979).
J. P. Simons, J. Phys. Chem., 88, 1287 (1984).
C. H. Greene and R. N. Zare, Ann. Rev. Phys. Chem., 33"
119
(1982) .
G. Dujardln, L. Hellner, S. Leach and D. Wlnkoun, Proc. 17th Int.
Symp. Free Radicals, National Bureau of Standards Publication'
( 1986), In press.
A. G. Harrison, Chemical Ionization Mass Spectrometry, C. R. C.
Press, Boca Raton (1983).
T. R. Govers, P. M. Guyon, T. Baer, K. Cole, H. Frohlich and M.
Lavoll.e, Chem. Phys., 87, 373 (1984).
F. W. Mclafferty, Interpretation of Mass Spectra, 3rd Edition, Unlv.
ScI. Books, Calif. (1980).
B. R. Rowe, G. Dupeyrat, J. B. Marquette and P. Gaucherel, J.
Chem. Phys., 80, 4915 (1984).
H. BOhrlnger and F. Arnold, Int. J. Mass. Spectrom. Ion Phys., 49,
61 <1983>S. E. Barlow, G. H. Ounn and M. Schauer, Phys. Rev. Lett., 52, 902
<1984>O. Smith, N. G. Adams and E. Alge, Chem. Phys. Lett., 105, 317
<1984>W. W. Duley and D. A. Williams, Interstellar Chemistry, Academic,
N. Y. <1984>E. Herbst, N. G. Adams and O. Smith, Ap. J., 269, 329
(1983) .
O.R. Bates, J. Chem. Phys., 81, 298 (1984) ; Ibid, 83, 4448
(1985) .
W. Hack, Int. Rev. Phys. Chem., 4, 165 (1985).
B. P. Tisa and J. H. O. Eland, Int. J. Mass. Spectrom. Ion Phys., 38,
143 (1980),
G. Dujardln, S. Leach, O. Dutult, P. M. Guyon and M. Rlchard-Vlard,
Chem. Phys., 88, 339 (1984).
G. Dujardln, S. Leach and D. Wlnkoun In Photophyslcs and
Photochemistry above 6 eV, (F. Lahmanl, ed.), Elsevier, Amsterdam
(1985) p. 63.
D. Cossart, F. Launay, J. M. Robbe and G. Gandara, J. Molec.
Spectrosc., 113, 142 (1985).
M. J. Besnard, L. Hellner, Y. Mallnovlch and G. DuJardln, J. Chem.
Phys., (1986) In press.
H. M. Rosenstock, K. Oraxl, B. W. Steiner and J. T. Herron, J. Phys.
Chem. Ref. Oata, 8, 1, (1977).
R. J. Brunt, F. W. Powell, R. G. Hirsch and W. D. Whitehead, J.
Chem. Phys., 57, 3120 (1972),
ON MOLECULAR IONS, RADICALS AND PAH's
127
123. - T, Masuoka and J. A. A. Samson, J, Chem. Phys" 74, 1093 (1981> ;
Ibid, 75, 4946 (T981)
T. Masuoka, Zeit. f. Physik. 0, (1986) In
press.
124. - E', J, Gallegos, J, Phys. Chem"
71, 1647 (1967).
125. - G. Ou/ardln, O. Wlnkoun and S. Leach, Phys. Rev., A31, 3027
<1985>126. - G. Ou/ardln and O. Wlnkoun, J. Chem. Phys., 83, 6222 (1985).
127. - G. Du/ardln, L. Hellner, O. Wlnkoun and M. J. Besnard, Chem.
Phys., 105, 291 (1986).
128. - J. H. O. Eland, F. S. Wort, P. Lablanqule and I. Nenner, Zeit. f.
Physik 0, (1986) In press.
129. - W. O. Watson and A. B. Kunz, Ap. J., 201, 165 (1975).
130. - P. J. Richardson, J. H. O. Eland and P. Lablanqule, Org. Mass.
Spectrom. (1986), In press.
DISCUSSION
K. Roessler
Are there simpLe ruLes to determine whether an unknown
spectrum comes from a neutraL, a radicaL or an ionized state?
Answer '. No !
M.S. de Groot: It has been suggested that the diffuse bands find
their origin in Light absorption by molecuLar species in space. Why do
you think this is probabLy not true?
Answer: My objection is not to moLecular species in generaL but to
PAH's in particuLar. If PAH's were responsibLe, even as a considerabLe
mixture, the relative narrowness of the visible diffuse absorption
bands, leads me to expect considerabLe broad, but banded, structures
in the near and far U.V. The waveLength-intensity behaviour oj the
continuum, including the Mont-Blanc type of 2200 A feature, in the near
and far U.V. extinction curve do not fit my expectations of PAH's.
FLUORESCENCE LINESHAPES OF POLYATOHIC MOLECULES SPECTROSCOPY WITHOUT EIGENSTATES
Shaul Mukamel; Kaiyu Shan and Yi Uing Yan
Department of Chemistry
University of Rochester
Rochester, N. Y. '4627
Abstract
Traditional spectroscopy of isolated molecules focuses on
individual molecular eigenstates (their positions and dipole
strengths). Macroscopic lineshapes on the other hand
contain collective coarse-grained information which is the
average of many eigenstates. A reduced correlation-function
formulation which allows a microscopic calculation of
spectra without having to consider individual eigenstates is
then used. Spectra of large polyatomic molecules may be
treated using both types of approaches. However, as the
molecular size increases, the macroscopic approaches become
more applicable and much more efficient. In this article we
develop a Green function correlation function approach which
provides a reduced description of molecular lineshapes.
Effects of intramolecular vibrational redistribution (IVR)
and intramolecular dephasing are readily accounted for.
Application is made to the dispersed fluorescence of
ultracold Anthracene in a supersonic beam.
I.
IIITRODUCTIOIi
The calculation of molecular Fluorescence and Raman spectra
in large anharmonic molecules is one of the fundamental
problems in molecular dynamics and spectroscopy. Recent
tCamille and Henry-Dreyfus Teacher Scholar
129
A. Uger et Ill. (etb.), Polycyclic Aromatic HydrOCllrbons and Astrophysics, 129-148.
© 1987 by D. Reuul Publuhing Company.
130
S. MUKAMEL ET AL.
experiments, particularly involving ultracold molecules in
supersonic beams are yielding accurate and detailed
information (both time-resolved and frequency-resolved)
[1-5]. This creates the need for the development of
appropriate theoretical tools which could be used to extract
dynamical information from these spectra.
The calculation of spectral lineshapes (and any other
response function) in macroscopic systems is usually made
using correlation function methods which are based on a
reduced description [6-8]. This is the case for pressure
broadening in the gas phase, lineshapes in liquids and solid
matrices, etc. One never attempts to calculate the exact
eigenstates of the macroscopic system. The reason is twofold: first, such a calculation is extremely difficult due
to the enormous number of degrees of freedom involved, and
second, the experimental broadened lineshapes contain highly
averaged information and do not reveal properties of
individual eigenstates. The calculation of individual
eigenstates of macroscopic systems is therefore neither
feasible nor desirable.
The analysis of spectra of isolated molecules on the
other hand is traditionally made in terms of properties of
individual molecular eigenstates (level positions and dipole
matrix elements) [9]. Such an approach is appropriate for
small or intermediate size molecules but for large molecules
(10 atoms or more) it is impractical. The spectra show
intramolecular line broadening in which information on
individual eigenstates is highly averaged. This state of
affairs is very similar to the behavior of macroscopic
systems and it is obvious that methods and techniques
developed for the latter may be adopted towards the
treatment of intramolecular line broadening of large
isolated polyatomic molecules [10-12]. As a simple
demonstration of the usefulness of the macroscopic concepts
we recall that in macroscopic systems we usually consider
the density of modes ~ unit volume as a fundamental
dynamical quantity whereas in molecules we usually look at
the total density of states. It is clear that for many
spectroscopic and dynamical observables the former quantity
is more relevant. When the molecule is large enough its
exact size is not so important as far as the behavior of a
single bond is concerned. This is an intramolecular
"thermodynamic limit".
In this article we develop a stochastic model which is
most suitable for the calculation of emission lineshapes of
large isolated polyatomic molecules as well as molecular
clusters and molecules in condensed phases [10,13]. The
photon emission process is illustrated in Fig. 1.
FLUORESCENCE LINESHAPES OF POLY ATOMIC MOLECULES
131
ylVR
--~------------~
O)eg
The level scheme for molecular fluorescence. The
incident light with frequency W excites the doorway state Id>.
The emitted w,,: photon can Origrnate either from the doorway
state or from a lower lying statelb> which is accessible
via intramoiecular vibrational redistribution (IVR).
Fig. 1
A molecule initially in the vibronic state la> and energy £
a
absorbs a photon wL and emits a photon ws ' and ends up in
the vibronic state Ic> with energy £.
Ib> and Id> denote
a manifold of vibronic states belOngrnj to an excited
electronic state Ie>, whereas la> and c> are vibronic
states belonging to the ground electronic state Ig>. For
isolated molecules (e.g. in supersonic beams or in the gas
phase at low pressure), the steady state rate of emission of
::(::~:::s is :~ve; b:(::le!Krame::::::se"ble:g6;::::::~:::
a,c
b
wL-w ba
(1)
iY b/2
Here £i is the energy of state Ii> and Y is its inverse
lifetime, wij
£i - £j' pea) is the equilibrium population
of la>, and ~i3 1s the dipole matrix element between states
Ii> and Ij>. The absorption line shape is given by
Z
°O(wL)
L
a,b
pea) h.lbal
2
Y/2
2
2
(wL-W ba ) +(Y b/2)
(2)
132
S. MUKAMEL ET AL.
Eqs. (1) and (2) involve multiple summations over
molecular eigenstates. These summations can be easily
carried out for small molecules with a few relevant levels.
However for large polyatomic molecules they become
intractable. This is the motivation for developing a
reduced approximate description which may eliminate the
necessity to perform these summations. The stochastic model
presented here [13-17] provides a convenient means to
achieve that goal. It is based on partitioning the degrees
of freedom into "system" modes which are directly coupled to
the radiation field and "bath" modes which are treated in an
approximate way. The model was first developed for
polyatomic molecules in a solvent [13]. In that case the
"system" modes are all the molecular degrees of freedom and
the bath consists of the solvent molecules. However, under
certain conditions the model applies also to large isolated
molecules whereby we adopt a reduced description in which we
consider explicitly only a few relevant vibrational modes
which are directly coupled to the optical transition, and
all the rest are treated as a bath [18,19].
II.
The Stochastic Langevin Model
In this section we present our general stochastic model for
the emission line shapes of polyatomic molecules. We
consider a molecule with two electronic states, the ground
state Ig> and an excited state Ie>. The molecular
Hamiltonian is partitioned as follows.
H
Ig> Hg <gl
+
Ie> [H e
eg (t) - (i/2)Y] <el
+ w
(3)
Here Y is the inverse lifetime of the electronic excited
state and w
is the electronic energy gap (0-0 transition)
between thee~wo states. For an isolated small molecule, H
and H contain all the vibrational degrees of freedom and g
w
i~ independent on time. We, however, adopt a different
p~~titioning. Hand H will contain only the optically
active modes. III theerest are treated as a thermal bath
which exerts a random force on the optically active degrees
of freedom. The main assumption of the present model is
that this effect can be described by assuming that Weg
becomes a stochastic function of time, i.e.,
(4a)
w
eg (t)
where W is the mean electronic energy gap. ow (t) is
taken t5 g be a Gaussian Markovian process with ze~§ mean
<ow eg (t»
0,
and' its correlation function is
(4b)
FLUORESCENCE L1NESHAPES OF POLY ATOMIC MOLECULES
133
11 2 exp(-At)
<ow eg (t) ow eg (0»
( 4c)
Here <•.. > denotes averaging over the stochastic variabl~~.
11 is the amplitude of the stochastic fluctuations, and A
is their time scale (correlation time). The states la>,
Ic>, etc. (Fig. 1) are the vibronic eigenstates of Hg
Hg
11>
£i
11>
i = a, c, ...
(Sa)
whereas the states Ib>, Id> are the vibronic eigenstates of
He' 1.e.,
(;;;eg + He)
11>
£i
11>
i = b, d,
(Sb)
The choice (Eqs.(4» of ow (t) is based on the
assumption that the bath couple§gmainly to the electronic
degrees of freedom, so that the ground state and the excited
state manifolds are being stochastically modulated with
respect to each other but no modulation occurs for
frequencies of levels belonging to the same electronic
manifold. This is often a realistic assumption. The
Gaussian nature of ow (t) can often be justified using the
centraal limit theoreffi¥ The absorber is coupled to the
applied radiation field by the electronic dipole operator
which couples vibronic states belonging to different
electronic states, i.e.
V
=
L
a,b,
c,d
[~abla><bl + ~ad la><dl
+
~cblc><bl
(6)
+
where the summation runs over the entire manifolds of ground
and electronically excited states.
Our solution for the absorption and the emission
lineshapes is given using the following auxiliary functions
[18] :
s=O,1, •••
(7)
where
K- 2
with
K
[exp(-AT) - 1
+
AT]
Alt:.
(7a)
(7b)
In addition we introduce the operators:
T( t)
T( t)
V exp(-iH t) V
e
V exp(-iH t) V
g
(8a)
(8b)
s. MUKAMEL ET AL.
134
and
v pg exp(-iH g t) V
T(t)
(8c)
Here p is the equilibrium cononical density matrix of the
Ig> st§te, i.e.
p
g
= exp(-aH)
g
with S = (kT)-1.
-s
jdl
Is!
K
K
-s
(9)
exp[iWl-Y1/2] J S (l) T(l),
(10a)
exp[iwl-Y,/2] J s (.) T(.)
(10b)
exp[iwl-Y1/2] J S (l) T(l)
(10c)
""
-i-
(w)
exp(-aH)
g
We further define
-i~
( s)
I Tr
jdl
Is!
and
=
-s
""
-i~
jdl
is!
Using these quantities we can write the emission spectrum
as[ 18]:
-21m
+
L
b,d
L
s-O
( 11)
The absorption line shape is given by
O(w L )
-21m
L p(a)K~~)(Ea+WL)
a
(12 )
FLUORESCENCE LINESHAPES OF POLY ATOMIC MOLECULES
135
III. The fast and the slow modulation limits; The
distinction between Raman and fluorescence components
We are now in a position to analyze our general expression
for the emission. First, it will be useful to explore the
partitioning of I(wL,wS ) (Eq.(ll» into a Raman and a
fluorescence component denoted IR(wL,wS ) and IF(wL,wS )
respectively. Such a distinction can De made, when the
emission spectrum consists of relatively narrow lines
centered around wL-w =w ,where Ic> and la> are two ground
state vibronic level~, g~d a much broader emission which
does not vary considerably as we tune w. When this
situation holds, we denote the former ltnes as Raman and the
latter broad emission as fluorescence.
The only term in Eq.(ll) which can contribute to a
Raman type of emission is
isA
Ws - wL + wca
1
------
2
wS-wL+wca+iSA
(wS-wL+w ca ) +(SA)
2
Typically A » (Ya + Yc ). Therefore the imaginary part of
Eq.(13) with SaO will result in narrow resonances at
wL-wS=w
whereas the terms with s>O will be much broader.
ca
(0)
We thus conclude that the Kca term is responsible for
the Raman components whereas all the other terms in Eq.(ll)
contribute to the fluorescence. A more detailed analysis of
this point was made elsewhere[13]. We thus have:
I(wL,wS )
IR(wL,w S )
+
IF(wL,wS )
(14a)
where the Raman component is
21T
L
( 14b)
a,c
We shall now consider some limiting cases. The fast
modulation (homogeneous) limit is obtained when the
correlation time of the bath fluctuations is very fast
compared with their magnitude, i.e., K »1. In this case,
the exp(-AT) in the r.h.s. of Eq.(7a) vanishes very raoidly
A
A
and may be tgnored. We then get geT) - fT, where,f In the limit Eqs.(10) assume the form
K(O)(w)
L Vlb><bIV
b
L
vlc><clv
c
w - e:
c
+
if
~
2
IA.
( 15a)
( 15b)
FLUORESCENCE LINESHAPES OF POLY ATOMIC MOLECULES
136
and
~ P(a)
a
V/a><a/V
W - £a+ ir
( 15c)
where r = Y/2 + r. In this case, only the s=O terms
contribute to Eq.(11) and the absorption line shape becomes
a sum of Lorentzian terms corresponding to homogeneous
dephasing.
I(wL,wS )
~ llab llbc llcd llda P (a)
a,c
b,d
x
W + if
W
ba -
x
x
L
(21TO(W
ac
+ W -
L
wda - wL- if
2r
wS )-
wbd
+
iY
] }
+
[
wbc
- Ws + if
wcd + Ws + if
(16)
Note that in the absence of broadening (r=O) Eq.(16)
reduces to the Kramers-Heisenberg form (Eq.(1). The slow
modulation (static, inhomogeneous) limit is obtained when
K «1.
In this case, we can make a Taylor expansion of
E~.~7a) (short-time approximation) resulting in g(,) =
6 ,/2. We then have
( 17a)
1
12;6
2
2
exp(-w /26 )
( 17b)
where 00 and I are given by Eqs.(2) and (1). As K is
varied, the abgorption line shape thus changes continuously
from a Gaussian to a Lorentzian.
IV.
~ission
Lineshapes ot Harmonic Molecules
In general the evaluation of Eq.(8), for example T(t), still
requires a summation over eigenstates i.e.,
T(t)
~
b
Vlb> exp(-icbt) <blv
(18)
137
FLUORESCENCE LINESHAPES OF POLY ATOMIC MOLECULES
However for harmonic molecules this summation can be carried
out formally. The expressions derived here for the matrix
elements of T(t) may be used to calculate the matrix elements
-
of T(t) as well by simply changing D to
-~
-1
D.
~
to
~
-1
and exchanging w' and w". T(t) may be evaluated using T(;)
by SUbstituting~T=t-iS: It is therefore sufficient to
consider only T(t). Let us consider now a general harmonic
molecule with the hamiltonian
H
I g> Hg <gl
N
1
H =- ~
g
2 j=1
N
1
H =- ~
e
2 j=l
with
N
w=
0
1. ~
2 j=l
+
w"j (pl!2
J
w! (p'.
J
2
J
Ie> [H
+
112)
qj
+
12)
qj
+
e
+
w -(i/2)Y] <el
eg
Wo
(w I - w")
j
j
( 19a)
(19b)
( 19c)
( 19d)
We adopt the common spectroscopic notation. whereby we label
ground-state quantities by a double prime and excited-state
quantities by a single prime. The dimensionless momentum
and coordinate corresponding to the j'th excited state
normal mode are denoted by plj and qlj respectively:
(20a)
p'
j
q'
(20b)
j
pI
and
QI
being the conjugate momentum and the normal
c~ordina~e. respectively. of the j'th mode of the excited
state. A similar transformation between Pj. qj. and Pj. Qj
is defined by changing all I indexes in Eq.(20j
to ". w' (w") and m are the frequency and the mass of
the j'thjmodJ. The ~uantity Wo accounts for the zero point
energy. and w
is the fundamental (0-0) transition
frequency. W~gshall now introduce a vector notation and
define the N component column vectors q' and qR whose
components are q; and qj, j-1 •••• N, respectively. The
normal modes q' ::ind q" ::ire not necessarily the same. and
most generally they may be related by the transformation
S. MUKAMEL ET AL.
138
q'
S
q" + D
(21)
where S 1s the Oush1nsky transformation matrix. D is an N
component vector whose components 0 denote the
dimensionless displacements of the Jquilibrium positions
between the two electronic states. In this section we shall
introduce an additional simplifying assumption, namely, that
the electronic dipole operator is weakly dependent on the
nuclear coordinates, so that ~ij becomes simply the FranckCondon factor, i.e., ~ij = ~eg <ilj>. This is the Condon
approximation. The eigenstates la> and Ic> will be denoted
in this section by In> and I~, respectively i.e,
N
la>
In> ~
IT
j-l
Inj >
N
Ic>
(22a)
(22b)
The final result is considerably simplified when the
normal modes in the ground and the excited electronic states
are identical, i.e., when the Oushlnsky transformation
matrix is diagonal
( Wi'/ Wj,,)1/2..,u
We then have:
TIIIl(t)
I~egl
2
1j
GIIIl(t)
(23 )
(24a)
where
N
(24b)
.G
(t)
IT
mjnj
j-1
For simplifying the notation, we shall, hereafter, consider
a single mode and omi t all the j subscripts from W " W ",
0 1 , n j • mj • The total Green function can then be Jalcu1ated
u~ing Eq.~24b).
For a single mode, we get[18]
GIIIl(t)
(25a)
where
FLUORESCENCE LINESHAPES OF POLY ATOMIC MOLECULES
139
(25b)
k
*
\L
k=O
(2k)1 n
[y(t)]k Hm.n- 2k[Af(t)D/a(t)]
mnk
kl
Here
w.
1jJ(t)
2
4w'w"
w+
(wjw.)
-
(26a)
(26b)
w_ exp(-iw't)
2w. - 2w_ exp(-iw't)
1/2
(26c)
]
2A2 + iO 4 - 1) sin w't
n 2 i04 1) sin w't
y( t)
=
w'
(26d)
± w"
(26e)
(w·lw.,)'/2
A
2k
=
L
(-1 ) q
qaO
{ .'
(26f)
m
n
(2k-q)(q)
Hj are the Hermite polynomials.
(m) =
i
and k*
2 exp(-2iw't)]
w- - w+ exp(-iw't)
a( t)
nmnk
-
w"[ 1 - exe(-iw't)]
f( t)
w±
[1
(25c)
(26g)
We have further defined
when m <: i
(27)
i! (:-i) I
when m < i
is the integer part of (m+n)/2.
For supercooled
molecules at zero temperature we need only n - O.
then assumes the fprm:
Wmo(t)
(m! 2m)-1/2 am(t) Hm[Af(t)D/a(t)]
Eq.(25c)
(28)
S. MUKAMEL ET AL.
140
where
w+w_[l - exp(-2iw't)J
B( t)
2
2
w+ - w_ exp(-2iw't)
1/2
}
(29 )
When the frequencies w' and w" are the same (w=w'=w"), we
have A = 1 and B(t) = O. Eq.(29a) then reduces to the form:
(m!2 m)-1/2[2Df(t)J m
m -1/2 m
m
(m!2 )
D [exp(-iwt) - 1J .
(30 )
The most general expression for G (t) including the
Dushinsky rotation is given elsew~re[18J.
Within the Condon approximation, the absorption
lineshape at temperature T (Eq.(2» is[18J
o(w L)
IUeg l 2 I Pen) 1m Gnn(wL - Weg + En)
n
In the absence of Dushinsky mixing, then Eq.(24b)
holds. In this case we need only consider a single mode.
oCt)
[~T(t)J
-1/2
2
exp[D fT(t)J
(32a)
with
~T(t)
1
4w'w"
(w"C+ A- + w' C_A+ )(w'C+ A + w" C- A+)
(3lb)
and
w"
fTC t)
CA
w"C +A- +w'C - A+
(32c)
where
C
±
±
exp( -iw' t)
A
(n + 1) ± n
n
[expOlw"/kT) - 1 J- 1
±
exp(iw"t)
(33a)
C33b)
(33c)
141
FLUORESCENCE LINESHAPES OF POLY ATOMIC MOLECULES
We shall now consider two limiting cases of Eq.(32). At
zero temperature aCt) becomes identical to Eq. (25b),
Eqs.(32b) and (32c) reduce to Eqs.(26a) and (26b),
respectively, and a(t) = GOO (t).
When w = w' = w", l/IT(t) = 1 , fT(t) = -1/2 C_A_ and we
get
exp(- ~02C_A_)
aCt)
+
V.
=
exp {~02 (n
+
1)[exp(-iwt)-1J
10 2 n[exp(iwt) - 1J }
2
(34 )
The Role or Intramolecular Vibrational Redistribution
So far we have considered purely harmonic molecules. The
previous expressions apply to small molecules or to large
polyatomic molecules with small amounts of vibrational
energy. As the degree of excitation is increased the
anharmonicities start to playa major role by inducing
relaxation and line broadening. Anharmonic effects may be
incorporated systematically in a perturbative way, starting
with the previous expresslons[19J. A simpler procedure for
incorporating the effects of IVR processes on the emission
lineshapes of supercooled molecules is provided by
introducing an intramolecular vibrational redistribution
IVR
(IVR) rate Ybd whereby the doorway state Id> relaxes to
state Ib>. Once the system is in the Ib> state, then the
bath modes become hot since they absorb the excess
vibrational energy wdb • As a result the emission from the
Ib> state can be represented by our stochastic model whereby
the parameters ~ and A (Eq.(4» now depend on the amount of
vibrational energy in the bath. For the doorway state Id>,
~-O and there is no broadening.
We expect that the
stochastic fluctuations amplitude ~ for emission from a
given Ib> state will increase as the available energy of the
bath w
increases. The total emission thus consists of a
progreg~ion of narrow lines originating from the doorway
state and a series of broad emission lines corresponding to
the redistributed emission. Our expression for the emission
in this case is[19J:
S. MUKAMEL ET AL.
142
+
I
IVR
Ybd
I
IVR
Ybd
b
where
Yd
Y
+
b
/Y
-(0)
Kbb
(£b-wS)
J
(35 )
(35a)
and
K~~)(w)
-i
f
0
d, exp(iw, - Y,/2) Jb b )(,)
Gbb (.r)
(35b)
Here J(b) is given by Eq.(7) but the bath parameters 6 and
o
A are taken to be dependent on Ib>. We have applied
Eqs.(35) towards the calculation of the emission spectra of
ultracold Anthracene[19J. A comprehensive supersonic beam
study of this molecule was conducted recently by Zewail and
coworkers[4J. Our results are displayed in Figures 2-5.
Anthracene has 17 modes which are directly coupled to the
electronic transition. The parameters of these modes used
on our calculation are summarized in Table I. Fig. 2
-1
demonstrates that for excess vibrational energy of 1168 cm
IVR processes already play an important role in the emission
since the purely harmonic emission (bottom figure) clearly
fails to reproduce the broad background as shown in the
experimental spectrum (top figure). The middle figure which
introduced IVR via our stochastic model reproduces the
experimental spectrum quantitatively. This effect becomes
more dramalfc as the excess vibrational energy is increased
to 1792 cm
as shown in Fig. 3. Both figures 2 and 3 were
calculated in the fast modulation (Lorentzian) limit. The
effect of K is demonstrated in Fig. 4. A Lorentzian (K-~),
intermediate (K s 1) and a Gaussian (KsO) lineshapes with the
same full width at half maximum are compared. Finally Fig.
5 shows the emission spectrum of vibrationally hot
anthracene at three temperature~~R This calculation was made
in the harmonic limit setting Y =0. The present
calculations demonstrate the simplicity in which this model
can be used to interpret the spectra of large polyatomic
molecules. The amount of computational effort does not
increase dramatically as the molecular size increases. This
is in sharp contrast to conventional expressions which are
based on summations over eigenstates, which are intractable
for large molecules.
FLUORESCENCE LINESHAPES OF POLY ATOMIC MOLECULES
1255
1165
\
1000
143
1255+'408
1165+390
1165+'408
'\.
\~55+390
°
'\.
-1000
-2000
Figure 2.
The 7~ dispersed fluorescence of ultracold
Anthracene in a s~~ersonic beam. The available vibrational
energy is 1168 cm . The parameters of the optically active
modes are given in Table I. The top figure is the
experimental spectrum[4]. The bottom figure is the emission
IVR
in the harmonic approximation (y d ~O) using the first term
in Eq.(35). The calculation cle~rly fails to reproduce the
broad redistributed emission. The middle figure was
calculated with IVR. (Eqs.(35» with Eq.(25». Only one
Ib> state (the ground vibrational state Ib>~IO» was used.
y~~R IY=5.
The relaxed emission was calcula~Td in the
fast modulation limit (Eq.(16» with r 35 cm •
E
s. MUKAMEL ET AL.
144
1792
~
I--~~~---+----~~~~--~~~~~~~
2000
1000
a
-1000
!lgure 3.
The same as Fig. 2 but for the combin~~ion
5~ 12~ dIspersed fluorescence spectrum with 1792 cm
-1
IVR
of excess vibrational energy. r-75 cm • Ybd /Y = 40.
FLUORESCENCE L1NESHAPES OF POLY ATOMIC MOLECULES
145
1515
Figure 4.
The same as Efg. 2 but for the 6~ d1spersed
of'excess v1brat1onal energy.
fluorescence w1th 1380 cm
y~~R/Y_30.
The calculations demonstrate the variation of
the emission w1th the stochast1c parameter K. K-~, 1 and 0
correspond respectJvely to Lorentz1an, 1nterme~fate, and
Gauss1an broaden1ng. The K-OO curve has r-65cm • In the
other calculat10n ~ and A were chosen such that the
stochast1c 11neshape has the same full w1dth at half
maximum.
146
S. MUKAMEL ET AL.
T
= 1000
K
T = 600.K
T=
I
2000
OK
-2000
-1
(em)
-"1000
Figure 5.
The total emission spectrum of Anthracene at
finite temperatures, taking the electronically excited state
to be in thermal equilibrium at three different
temperatures. The calculation was made using Eqs.if') and
(32) in the absence of lVR. The linewidth Y-l0 cm •
147
FLUORESCENCE LlNESHAPES OF POLYATOMIC MOLECULES
Acknowledgements
The support of the National Science Foundation, the Office
of Naval Research, the U.S. Army Research pffice, and the
Petroleum Research Fund, administered by the American
Chemical Society, is gratefully acknowledged. We wish to
thank Prof. A. Zewail for most illuminating discussions.
Table I.
Frequencies and Dimensionless Displacements of Anthracene
Mode
w"(cm
12
11
10
9
8
7
6
5
4
11
1Q
2
§
I
2
4
3
-1
)
-1
w'(cm)
D
390
624
753
1012
1165
1263
1408
1486
1566
385
583
755
1019
1163
1168
1380
1420
1501
0.79
0.21
0.24
0.20
0.666
0.69
1.35
0.20
0.80
391
524
912
1100
1184
1382
1576
1643
232
473
889
1046
1183
1409
1514
1635
0.33
0.15
0.20
0.15
0.24
0.20
0.35
0.63
148
S. MUKAMEL ET AL.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
R.R. Langridge-Smith, D.V. Brumbaugh, C.A. Hoynam, and
D.H. Levy, J. Phys. Chern. 85, 3742 (1981).
C.S. Parmenter, J. Phys. Chern. 86, 1735 (1982).
See papers in Farad. Discuss. Chem. Soc. 75, (1983).
W.R. Lambert,P.M. Felker, J.A. Syage and A.H. Zewail,
J. Chern. Phys.,81, 2195 (1984); W.R. Lambert,
P.M. Felker and A.H. Zewail, J. Chem. Phys., 81, 2209
(1984); 2217 (1984).
A. Amirav, U. Even and J. Jortner, J. Chern. Phys. 71,
2319 (1979).
R. Kubo, Adv. Chem. Phys. 15, 101 (1969).
R.G. Breene, Jr. Theories of Spectral Line Shapes,
Wiley, New York (1981).
S. Mukamel, Physics Reports 93, 1 (1982).
G. Herzberg, Molecular Spectra and Molecular Structure,
Van Nostrand (New York) (1966).
S. Mukarnel, J. Phys. Chem. 89, 1077 (1985).
S. Mukamel, J. Chem. Phys. 82, 2867 (1985).
R. Islarnpour and S. Mukarnel, J. Chem. Phys. 80 5487
(1984); S. Mukamel, A. Pandey and J. Sue, Chem. Phys.
Lett., 105, 134 (1984).
S. Mukamel, J. Chem. Phys., 82, 5398 (1985); J. Sue,
Y. Yan and S. Mukarne 1, J. Chem. Phys. (in press).
N. Bloembergen, E.M. Purcell and R.V. Pound, Phys. Rev.
73, 679 (1948).
P.W. Anderson and P.R. WeiSS, Rev. Mod. Phys. 25, 269
1953 ).
Y. Toyozawa, J. Phys. Soc. Jap. 41, 400 (1976);
A. Kotani and Y. Toyozawa, ibid., 41, 1699 (1976).
T. Takagahara, E. Hanarnura and R. Kubo, J. Phys. Soc.
Jpn. 43, 802, 811 (1977); 44, 728, 742 (1978).
Y. Yan and S. Mukamel, J. Chern. Phys. (in press).
K. Shan, Y. Yan and S. Mukamel, J. Chem. Phys.
(in press).
STRUCTURE AND CHEMISTRY OF PAHs
w.
Schmidt
Biochemical Institute
Sieker Landstrasse 19
2070 Ahrensburg
W. Germany
ABSTRACT. The composition of some typical PAH mixtures formed
by incomplete combustion and pyrolysis of organic matter is reviewed
in terms of (a) nature of precursors, (b) thermal history, (c) transport,
and (d) destruction. Although compact peri-condensed PAHs with low
HlC ratio, such as coronene and ovalene~ are favoured in high-temperature processes, thermodynamic stability is not a prime factor in determining the PAH composition. PAHs which are photochemically or
kinetically labile may well be formed in substantial amounts but are readily degraded thereafter.
Polycyclic aromatic hydrocarbons (PAHs) are formed during incomplete combustion and pyrolysis of almost any kind of organic material.
They are continuously released into our environment and are therefore
present in coal extracts, coal tar, carbon black, crude and processed
petroleum, soot, soil, effluents from coal and wood burning, automobil
exhaust, cigarette smoke, waste incineration and even fried food.
Because of the adverse effects on human health, considerable effort has
been devoted to developing reliable methods for PAH monitoring, mainly by Grimmer and his group at Ahrensburg (1)0
In terms of composition, PAH mixtures (naturally occurring or anthropogenic) range from simple to exceedingly comple~ thus reflecting
the thermal history and the nature of the precursors from which they are
formed. During transport from the emission source to the site of collection, the composition of PAH mixtures may also be altered due to
fractionation and/or selective degradation. Some rare PAH minerals
consist of a single PAH in almost pure form, whereas coal tar is estim149
A. Leger et al. (etis.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 149-164.
© 1987 by D. Reidel Publishing Company.
W. SCHMIDT
150
ated to contain some 30000 PAHs (including S-, 0- and N-analogs and
their alkyl homologs), of which only a small number has been isolated
and structurally identified. Combustion effluents from fossil fuels occupy an intermediate position.
Due to the lack of observational in-situ techniques, our understanding of the processes leading to PAH formation during combustion and
pyrolysis is still in its infancy. Model calculations, such as those reported by Keller at this Workshop, are therefore of great importance.
In what follows, I shall review the composition of some typical PAH
mixtures found on earth in terms of (a) nature of precursors~ (b) thermal
history, (c) transport, and (d) destruction. Hopefully, these case studies provide some hints as to the composition of the "PAH soup" (Leger,
at this Workshop) believed to be responsible for the interstellar IR
emission and UV extinction. I shall begin with the simplest cases and
will then move on to more complex PAH mixtures.
1. PAH minerals
In association with cinnabar in mercury ores, several rare PAH
minerals have been found at distinct locations. In 1844, Boedeker (2)
noted the occurrence of two organic compounds, later named pyrene
and fluoranthene, during the "roasting" of mercury ore in Idrija, Yugoslavia. While the structure of pyrene 1 was deduced already in 1887, it
was not until 1929 that fluoranthene was assigned the correct formula 1:
Two other PAH minerals, named idrialite and curtisite, have been
found in mercury mines in Czechoslovakia and California. The first of
these has already been studied in 1833 by Dumas (3). Using modern
chromatographic techniques (GC and HPLC), Blumer (4) and Wise (5)
showed that they are of similar composition, containing some 200 PAHs.
Due to the lack of the reference compounds, only about 20 out of these
could be structurally identified. The most abundant ones are chrysene
l, picene~, fulminene Q, 1.2, 5. 6-dibenzanthracene.§, 3.4,8. 9-dibenzotetraphene 1., 2.3-benzopicene jl and the difluorene derivative~:
roB! #i oxx9§
1
0J5C? ~
(C(x)::;o.!!
STRUcrURE AND CHEMISTRY OF PAH's
151
The most intriguing PAH mineral is pendletonite from San Benito
County, California, which has been identified (4, 6) as almost pure (99%)
coronene 10, the remainder being 1-methylcoronene 11. To my know-
@
10
@
eH
3
11
~
12
ledge, only a few grams of this precious mineral have been collected so
far. It should be noted that coronene which is available from Aldrich
and Fluka is inferior in purity to pendletonite. A related mineral, carpathite& occurs in Trans-Carpathia; judSing from its high melting point
of 430 C (pure coronene melts at 435 - 8 C), this seems to be similar
to pendletonite.
Asbestos minerals like crocidolite, amosite and chrysotil have also
been shown to contain PAHs, prime among them the carcinogenic 3.4benzopyrene 12.
It is an open question why only specific types of PAHs, sometimes
a single one, occur in these minerals. One may think that this is due
to a transport phenomenon. Organic compounds derived from microorganisms were pyrolized at elevated temperatures, giving complex
assemblages of PAHs which are subsequently fractionated by crystallization or hydrothermal transport during migration to the surface of the
mercury ore. This hypothesis explains the sole occurrence of pyrene
and fluoranthene in Idrija; being isomers, these two compounds have
very similar boiling points and would hence co-deposit in the ore. The
abundance of the zig-zag annelated PAHs ~ - j!, with widely differing
boiling pOints, in idrialite and curtisite is less readily explained; in
particular, we note the absence of triphenylene and of any peri-condensed system such as pyrene and fluoranthene.
2. Soils and sediments
Between 10 and 15 unsubstituted PAHs were found in similar proportions in soils and young marine sediments at widely scattered locations
throughout the world. These include phenanthrene, anthracene, chrysene, fluoranthene, pyrene, perylene ~ 1.2-benzopyrene 14, 3.4-benzopyrene, 1.12-benzoperylene 15, anthanthrene 16 and coronene:
152
W. SCHMIDT
Alkylated PAHs are also present, albeit at lower concentration. The
uniformity of the PAH profile argues against a biochemical origin, i. e.
from micro-organisms, and points towards a pyrolytic source, presumably from forest and prairie fires (7) and volcanos. The PAHs formed
in this manner are bound to soot -particles (which protects them against
photo-oxidation), are dispersed by prevailing winds throughout our globe,
and are finally deposited by rain or snow-fall.
The perylene case is more involved. This PAH is abundant in peat
and certain, but not all, recent marine sediments. It has been suggested (7 - 9) that it originates from a pigment called erythroapin 17 which
is contained in certain insects, fungi and marine organisms. In the presence of reducing agents, such as carotenoids, this pigment is converted into perylene which is stable enough to persist for millions of years.
Where oxidative conditions prevail, the pigment will be degraded and no
perylene is found.
OH 0
'C~Hgl
~g
OH 0
17_
CHJ
Blumer (7) also reported 011 the occurrence of hexahydro-bisanthene
18 in extracts of fossil sea lilies deposited in the Jura mountains of Switzerland. This unusual compound is thought to be the geochemical transformation product of pigment 19 which is synthesized' by these organisms:
OH 0 OH
OH@OH
OH@OH
19
-
OH 0 OH
Only traces of other PAHs were detected in the extracts.
3. Meteorites
PAHs found in meteorites are believed to be synthesized by FischerTropsch-type reactions in the solar nebulae at about 1 OOOOC The reactions involve CO, H2 and CH4 and presumably take place at the surface
of dust grains.
The PAH composition is reminiscent of that encountered in soil and
marine sediments, but the PAH concentrations are much higher. For
example, one meteorite contained typically 1 mg of fluorene 20 per gram
of sample:
STRUCTURE AND CHEMISTRY OF PAH's
153
Analysis by GC, either alone or in conjunction with MS, revealed the
presence of the following PAHs (11,12): Biphenyl, acenaphthene, phenanthrene, anthracene, fluoranthene, pyrene, plus unidentified PAHs of
mass 228 (chrysene, triphenylene, tetraphene?) and 252 (1.2- or 3.4benzopyrene). The principal ones are phenanthrene, fluoranthene and
pyrene, as is the case with environmental PAH mixtures formed by pyrolysis or combustion. Coronene and high~r annelated PAHs are certainly present in the meteorites but could not be identified because the analytical procedure was inadequate. The occurrence of several methyl homologs reveals that the temperature, at which these PAHs were formed,
was actually lower than suggested above.
In addition to the above compounds, . several N-PAHs, ketones and
qUinones, which are also prominent in environmental samples, have
been identified.
4. Carbon black
Carbon black is used as a reinforcing material in the tyre industry
and as printing ink. It is formed by burning methane, low-boiling alkanes or mineral oil under carefully controlled conditions so as to minimize
PAH formation. Transmission electron microscopy shows it to be amorphous carbon of clustered or branched-chain morphology (13). The carbon content, as determined by elemental analysis, is 99.6% or greater;
for comparison, soot has of the order of 98.8 % while a typical pAH such
as coronene hs 96.0 % carbon.
Due to the high temperatures involved in its formation. (typically
1500 0 C), methyl homologs are practically absent in carbon black, and
the PAH composition is therefore fairly simple. The predominant PAHs
are pyrene, fluoranthene, 3.4-benzopyrene, anthanthrene, 1.12-benzoperylene and coronene. Cyclopentapyrene 21, a most unusual and not
very stable compound, was isolated from carbon black for the first time
(14); its synthesis' followed shortly thereafter.
A great many higher-molecular weight PAHs are present but are
difficult to remove by solvent extraction because they are tightly bound
to the surface of the carbon particles. This is related to the fact that
154
W. SCHMIDT
the PAH content of carbon black is well below that required for monolayer coverage. A preliminary study (15), using CH2Cl2 for extraction
and reversed-phase HPLC for separation, showed the presence of some
40 PAHs in the range 302 to 448 Daltons, as determined by MS. Since
the authors (15) had none of these reference compounds in hand, their
structural identifications remained speculative. However, from their
mass numbers one can infer one important aspect: Compact peri-condensed PAHs with low H/C ratio predominate in the high-molecular
weight range.
5. Coal hydrogenation products
During world war IT, pressure hydrogenation of coal was carried
out on an industrial scale in Germany to produce liquid fuel. This required MoS2 as catalyst and took place at 400 - 700 0 C and 400 - 600 atmospheres. Several reports (16-19) refer to the formation of a yelloworange, high-melting organic compound which interfered with the processing of the fuel by blocking the tubes. This was later identified as
coronene, and there are strong reasons to believe - vide infra - that
the material collected at that time is a major source of commercially
available coronene still today.
Through the courtesy of BASF AG, Ludwigshafen, the author obtained a large batch of this material. The UV spectrum showed the presence of several impurities, one of which was readily identified (16) as
1.12-benzoperylene. This was removed by slow fractional sublimation
at O. 1 mbar on account of its higher volatility. Coronene was collected
at about 270 °C. On increasing the temperature further on to 450 0 C, a
small quantity of a yellow material, approximately 1 % by weight based
on coronene, was obtained.
Direct-inlet MS of this residue revealed the presence of the following masses (major peaks are underlined):
374 C30H14
398 C32H14
424 C34H16
448 C36H16
474 C38H18
476 C38H20
498 C40H18
500 C40H20
528 C42H24
If we consider six-membered rings only, then mass 398 is unique in the
sense that there is only one possibility, namely ovalene ~ (just as mass
STRUcrURE AND CHEMISTRY OF PAH's
155
300 corresponds to coronene and nothing else). The presence of ovalene in the residue as well as in the crude COl'onene was confirmed by
comparing the UV maxima with those of an authentic specimen.
In an attempt to identify the other peaks, the residue was separated
by slow gradient sublimation into four fractions whose UV~ fluorescence
excitation and emission spectra were separately measured.
The first fraction was almost pure C30H14 of mass 374. Although
this mass is not unique, we feel confident that this hitherto unknown
compound of mp 345 0 C is naphthocoronene 22. Our assignment is based
@
22
C30 H14
374
@
23
C32 H14
398
~ ~
24
C34 H16
424
25
C36 H16
448
on a comparison of the measured a-, p- and ~-band positions with those
calculated by Huckel theory (20). The second fraction, approximately
30 % by weight~ is ovalene 23~ as already discussed above. The third
fraction, not obtained free of ovalene~ was readily identified as tribenzoperopyrene 24 by comparison with an authentic standard. The fourth
fraction, again contaminated with some 24, turned out to be benzonaphthobisanthene 25; this new hydrocarbon has very recently been synthesized by J. C. Fetzer at Chevron Research Co., Californi~ and was
placed at our disposal for direct comparison. The remaining compounds
in the residue could not be structurally identified because their concentrations were too small for spectral characterization.
That coronene and compounds 22 - 25 withstand the harsh conditions
of the hydrogenation process, is not fortuitous. As a general rule, compact peri-condensed PAHs with low H!C ratio have a large number of
Clar sextets and a large number of KekuM structures; consequently,
they lack reactive carbon atoms (as, for example, the 9,10 positions
in anthracene or phenanthrene) which would accept hydrogen. Once a
PAH has been partially hydrogenated, it can be broken up into smaller
fragments.
All commercial samples of coronene which we have examined clearly show the characteristic p-band of ovalene at 451 nm in benzene solution. No ovalene was detected in a synthetic sample provided by Professor Claro
156
W. SCHMIDT
6. Pyrolysis of pure PAHs
Lang et ale have stuaied the pyrolysis of several PAHs (benzene/
pyrene, naphthalene, 1-methyl- and 2-methylnaphthalene, fluorene,
acenaphthene, phenanthrene) in the temperature range 700 - 800 °c on
a clay or 8i02 support .. With naphthalene as substrate (21), the primary process is C-H bond rupture; the resulting 1- or 2-naphthyl radical attacks another naphthalene molecule to give, with the eliminati..
on of an H atom (fate unknown), the three isomeric dinaphthyls 26 - 28.
While 2, 2'-dinaphthyI28 is unable to undergo further condensation reactions and. hence builds up in the reaction mixture, the other isomers
give perylene and the benzofluoranthenes 29 and 30, respectively:
gs
J2?
26
!
gg
exyB 28
!
/~
£29
&
30
no
reaction
Once formed, these dimers undergo further pyrosynthetic reactions
with naphthalene to give the following trimers:
32
No evidence for C -C rupture was found, as is consistent with the higher
binding energy of the aromatic C-C vs. the C-H bond.
Using a different set of conditions (450 0 C, 20 hrs reaction time,
no catalyst, HPLC for analysis), Zander et ale recently studied the
pyrolysis of perylene (22). The product pattern follows similar lines
as above, with 35 being the main product; next came diperylenyls such
as!§, followed by traces of quaterrylene 37. Interestingly, small amounts of splitting products (presumably 1, 8-dimethylphenanthrene W
and secondary products (methyl- and dimethylperylenes, 1.12-benzoperylene) were also detected:
STRUCTURE AND CHEMISTRY OF PAH's
157
36
37
The origin of these splitting products is not clear. One may speculate that
the hydrogen released in the above dimerizations is taken up by perylene
with the formation of hydroaromatic rings which are subsequently cleaved.
The cleavage products (CH2, C21l4?) may insert into C-H bonds of the substrate, giving methylperylene, or may enter into Diels-Alder reactions to
give 1.12-benzoperylene.
Such cleavages and even skeletal isomerizations have occasionally been
observed in AICl3 catalyzed condensations (23). They seem to proceed
through the intermediacy of dihydro- or tetrahydroaromatic species.
7. Combustion and pyrolysis effluents
Effluents formed by incomplete combustion and pyrolysis or organiC
matter show a fairly constant PAH profile which is, in a first approximation, independent of the specific fuel type, viz. brown-coal, hard-coal,
gasoline, Diesel fuel and wood; even polystyrene and polyvinylchloride give
rise, upon pyrolysis, to roughly the same PAH profile as the other materials (24). The total amount of PAHs emitted, per kg of burned or pyrolized
material, depends on the nature of the latter, however. Due to the fairly
high temperatures involved, alkyl substituted PAHs are less important.
There are three distinct PAHs which are characteristic for the nature
of the precursor and which therefore allow the emission source to be specified: Picene and its alkyl homo logs which are characteristic of brown-coal,
cyclopentapyrene which abounds in gasoline exhaust, and a PAH of mass
300 - structure unknown - which occurs in both hard-coal effluents and
gasoline exhaust. In the case of brown-coal, the picenes originate from
triterpenoids with a- and f3-amyri'ne structures (25). For the other two
PAHs, the precursor is not known.
The prominent PAHs formed by combustion and pyrolysifl are phenanthrene/anthracene (both of mass 178), fluoranthene/pyrene (mass 202) and
158
W. SCHMIDT
chrysene/triphenylene/tetraphene (mass 228). The benzofluorantheneS/benzopyrenes (mass 252) and 1. 12-benzoperylene/2. 3-0-phenylenepyrene 39 (mass 276) are less abundant by a factor of typically four.
There is then a similar decline in going to coronene (mass 300) and the
dibenzofluoranthenes/dibenzopyrenes (mass 302). Higher annelated
PAHs are undoubtedly present but are not amenable to analysis by coupled GC/MS. Direct-inlet MS revealed the presence of PAHs with molecular weights as high as 500, possibly 600 (26)0
In a recen.t study (27) of hard-coal combustion effluents, specific
attention was devoted to PAHs of mass 302. This study provides an
opportunity to examine whether the relative abundancies are gouverned
by thermodynamic factors, i. e. stability. The isomers were separated
by thin-layer chromatography on RP-18 plates and were subsequently
identified by UV and fluorescence analysis. Abundancies, as estimated
by GC, decrease from left to right:
1
£
40
45
~
41
§v
42 c \ 4 3
c\44
~ cgm ~ ~
46
47
48
49
This is clearly not the order expected on the basis of stability, for dibenzopyrene 48 with four Clar sextets should then be the most abundant
one. We also note that several compounds with reactive anthracene
reSidues, such as 43 and 45, survive under the harsh conditions prevailing in thp. combustion zone.
That stability is not the decisive factor, is also clemonstrated by
the abundance of cyclopentapyrene in combustion systems. Cyclopentapyrene is thermally stable - in one laboratory synthesis it is formed
by a pyrolytic ring closure at 850 0 C - but photochemically labile to
give dimers; in ambient air, it is readily degraded by reaction with
STRUCTURE AND CHEMISTRY OF PAH's
159
NOx, 03 or OH radicals.
It does not, therefore, accumulate in our environment as do, for example, chrysene, fluoranthene, pyrene etc.
Depending on the residence time in the atmosphere, the composition
of PAH mixtures may change from complex to simple. There are two
main mechanisms, the first being photo-oxidation - either in the gas
phase or in the adsorbed state - which eli~inates PAHs with elongated
acene chains. A well-known case is that 01 anthracene which is first
converted into an endo-peroxide which subsequently rearranges to give
anthraquinone:
ceo
,
©o
The second mechanism involves reactive species like NOx, 03 or OH
radicals. The primary products - nitro-PAHs, dialdehydes and phenols - are not overly stable and may be further degraded or oxidized to
give ketones or quinones.
The factors gouverning both types of reactions are well understood
on a molecular level. In terms of H"uckel theory, rates are the higher
the smaller the loss in 1T'-electron energy in going from the starting PAH
to the primary product. With most PAHs, both mechanisms act in unison
to eliminate (with decreasing ease) tetracene, cyclopentapyrene, anthracene, tetraphene, 3.4-benzopyrene and anthanthrene. Under daylight
conditions and in the adsorbed state, their atmospheric lifetimes are estimated to range from 1 min to typically 1 day (28). Phenanthrene, chrysene, fluoranthene (including some of its benzologs), 1.2-benzopyrene
and coronene are the most stable compounds, with lifetimes of many
weeks; they may travel with lJrevailing winds o.ver 10000 km and more,
and are therefore the only PAHs found in soil remote of industrial activity.
s.
Fossil fuels
Petroleum is notorious for its extreme compositional complexity.
It is rich in methyl- and isopropyl-substituted naphthalenes and phenan-
threnes, as typified below by eudalene (50), cadalene (51) and retene (52):
.
CH].
I-Pr~
~
50
CH)
~
CH3~
i-Pr
51
---
~
i-Pr
52
~---
%
This has been interpreted in terms of low-temperature (100-150 0 C) degradation of biological material, viz. terpenoids, steroids and abietic
acid. Hydroaromatic structures, such as -tetraline and its homolog~
are also prominent. On the other hand, unsubstituted PAHs are rare be-
160
W. SCHMIDT
cause the temperature was not high enough to eliminate alkyl sidechains.
Coal is nowadays considered to be a complex mixture of giant molecules with an average molecular weight of about 3 000. IR analysis (29)
shows that the carbon atoms are distributed among the various group
types as follows:
2%
C-OH
aliphatic C H
5%
aliphatic CH2
7%
aliphatic CH3 12 %
aromatic CH 18%
aromatic C
53 %
The amount of carbon involved in carbonyl groups is not precisely known.
For illustration, a typical aromatic molecule like 53 contains the following carbon types:
1 aliphatic C H ( 5.5%)
1 aliphatic C H2 ( 5.5%)
2 aliphatic C H3 (11.1 %)
7 aromatic CH (38.9%)
(38.9%)
7 aromatic C
Sulfur, oxygen and nitrogen are also present in coal, either as heteroatoms in aromatic nuclei, or as bridging groups between such nuclei.
The figures given above show of course considerable variation depending on the maturity and thermal evolution of the coal type considered.
CONCLUSIONS
The case studies reviewed above may be summed up as follows:
1. Pyrolysis of organic matter gives complex assemblages of PAHs
whose composition reflects primarily the structure of the precursors.
2. The thermodynamic stability of the resulting PAHs (as measured by
heats of formation or estimated by HUckel, resonance or Clar theory)
plays a secondary role in determining the composition.
3. Depending on the thermal history (low- or medium-temperature pyrolysiS), alkyl side-chains and hydroaromatic rings are predominant
or absent in the PAHs.
4. Transport phenomena {fractional distillation and/or crystallization,
selective destruction} can give rise to deceptively simple PAH com-
STRUCTURE AND CHEMISTRY OF PAH's
161
positions.
5. In pyrosynthetic reactions, compact PAHs with low H/C ratio are
favoured over non-compact ones; presumably the latter provide
active binding sites for the attachment of smaller units in DielsAlder and cycloaddition reactions.
6. The carbon skeleton of pAHs is stable towards isomerization and
cleavage. Small PAHs grow into larger ones through dimerization
with elimination of hydrogen.
These simple rules of thumb cover the "normal" chemistry of PAHs;
they do not include highly excited electronic states, doubly ionized states
and the behaviour in shock waves.
REFERENCES
(1)
G. Grimmer (Ed.): Environmental Carcinogens: Polycyclic Aromatic Hydrocarbons, CRC Press, Boca Raton, Florida, 1983.
(2)
C. Boedeker, Liebigs Ann. Chern. 52, 100 (1844).
(3)
J. Dumas, Liebigs Ann. Chern. ,2., 5 (1833).
(4)
M. Blumer, Chern. Geology
(5)
SoA. Wise, Org. Geochem., in print, and personal communication,
1985.
(6)
J. Murdoch and T. A. Geissman, Amer. Mineral. 52, 611 (1967).
(7)
M. Blumer, Sci. Amer. 234, 34 (1976).
(8)
Z. Aizenshtat, Geochim. Cosmochim. Acta 37, 559 (1973).
(9)
W. L. Orr and J. R. Grady, Geochim. Cosmochim. Acta 31, 1201
(1967).
~
245 (1975).
(10) M. H. Studier, R. Hayatsuand E. Anders, Geochim. Cosmochim.
Acta ~ 189 (1972).
(11) B. Po Basile, B. So Middleditch and J. Oro, Org. Geochem• .§, 211
(1984)" and earlier work cited therein.
(12) K. L. Pering and C. Ponnamperuma, Science 173, 237 (1971).
(13) A.I. Medalia, D. Rivin and D. R. Sanders, Sci. Total Envir. 31, 1
(1983)
(14) A. Gold, Anal. Chern. 11, 1469 (1975).
(15) PoA. Peaden, M.L. Lee, Y. Hirata and M. Novotny, Anal o Chern.
52, 2268 (1980).
162
w. SCHMIDT
(16) 1.. Boente, Brennstoff-Chemie, ~ 210 (1955).
(17) H. Fromherz, L. Thaler and G. Wolf, Zeitschr. Elektrochem. 49,
387- (1943).
(18) M. Orchin and J. Feldman, J. Org. Chem. 1], 609 (1953).
(19} E. Clar, Polycyclic Hydrocarbons, Vol. II, Academic Press~ New
York, 1964, and personal communication.
(20) E. Clar~ J.M. Robertson, R. Schlogl and W. Schmidt, J. Amer.
Chem. Soc. 103, 1320 (1981).
(21) K. F. Lang, He Buffleb and J. Kalowy, Chem. Ber. 90, 2888 (1957).
(22) M. Zander, J. Haase and H. Dreeskamp, Erdal und Kohle, Erdgas,
Petrochem. 35, 65 (1982).
(23) G. P. Bliimer, K. D. Gundermann and M. Zander, Chern. Ber. 109,
1991 (1976).
(24) G. Grimmer and A. Hildebrandt, unpublished results.
(25) Vo Jarolim, K. Heino, Fo Hemmert and F. Sorm, Collo Czech. Chem.
Commun. §.Q, 873 (1965).
(26) G. Grimmer, J. Jacob, G. Dettbarn and K. W. Naujack, Fresenius
Zeitschr. Anal. Chem. 322, 595 (1985).
(27) Wo Schmidt, G. Grimmer~ J. Jacob, G. Dettbarn and K. W. Naujack,
Fresenius Zeitschr. Anal. Chem., In print.
(28) Lo Blau and H. Glisten, 6th Internat. Symp. on "Polynuclear Aromatic Hydrocarbons, Phys. and BioI. Chem~', Columbus, Ohio, 1981.
(29) He H. Oelert, Nachr. Chem. Techn. 19, 165 (1971).
DISCUSSION
N. N. : What are the prospects for high-temperature GC to enable highermolecular weight species to be separated without an added solvent?
Answer: GC of PAHs with molecular weights up to 302 Daltons is nowadays a routine matter with commerCial instruments. PAHs up to 400
Daltons can be chromatographed at somewhat higher temperature (350
rather than 270 °C, as in routine operation), but at the cost of resolution
and column lifetime. Using a short (5.5 m) capillary column, K. Grob
(Chromatographia 1., 94 (1974» has extended the range to 532 Daltons,
at poor resolution of course. I think that packed micro-capillary columns
for HPLC will be available in 1 or 2 years time; this technique offers resolution comparable to capillary GC and has no limitations as to molecular size and thermal stability of the compounds to be analyzed.
STRUcrURE AND CHEMISTRY OF PAH's
163
E. Evleth: PAHs have distinctive C-H out-of-plane vibrations in the
10 - 13 mp regwn whose intensities may be proportional to the number
and type of C-H bonds. Are quantitative KBr pellet IR spectra in this
spectral region of aid in structural characterization of the materials?
Answer: The spectral ranges of solo, duo, trio and quartet C-H vibrations overlap partly so that unambiguous structural information cannot be deduced, although the spectra are useful in allowing certain
structures (out of a limited number of alternatives) to be rejected from
the outset. Solo vibrations are in general more intense than duo vibrations; for example, in ovalene (2 solo and 12 duo H atoms) and dicoronylene (4 solo and 16 duo), both bands assume roughly the same integrated intensity. However, counterexamples are known. For more information on this topic see: Ref. (20); S. Obenland, at this Workshop;
M. Zander, Erdal und Kohle, Erdgas, Petrochem. 15, 362 (1962);
W. Hendel, Z. H. Khan and W. Schmidt, Tetrahedron~, 1127 (1986).
Ellinger: Which vibrational frequencies of the positive ions can be extracted from the PE spectra, if any? Do they fit with the unidentified
IR bands?
Answer: The first PE band always consists of a short (3 - 4 members)
vibrational progression in 1400:!: 400 cm- 1 ; within experimental error,
this interval is independent of size, shape and symmetry of the PAH.
In analogy to the p-band of the UV spectra, this mode is interpreted as
an aromatic C-C vibration in which successive C-C bonds show alternate stretching and shortening - analogous to the change-over from one
Kekul~ structure of benzene to the other.
The vibrational interval seen
in the second PE band varies from 500 - 1000 cm- 1 and corresponds to
a skeletal breathing mode. According to the selection rules, only totally-symmetric vibrations can be excited in the PE spectra; prospects for
detecting out-of-plane C-H vibrations in the PE spectra are thus meagre.
The modest resolution aChievable (200 - 400 cm -I for the first band) is
also a serious drawback.
L. d'Hendecourt: On the PE spectra, what are the diffuse bands beyond
10.5 eV? Do they show some photodissociation of the C-H bonds?
Answer: The diffuse PE bands above 10.5 eV are due to ionization from
C-H and C-C IS -bonds. These 6-orbitals are more strongly bonding than
the 1T-orbitals and hence give rise, by virtue of the Franck-Condon prinCiple, to broad bands. Moreover, a PAH has more 6- than 1I'-orbitals
(coronene: 1211', 30 (5 C-C, 12 6 C-H). For large PAHs, both effects
result in a merging of the d-bands iote a quasi-continuum. Whether
ther~ is photodissociation for the C-H bonds, cannot be irtferred from
164
w. SCHMIDT
the PE spectra. This is the domain of photoelectron/ion-coincidence
spectroscopy. To my knowledge, no such measurements have yet been
done for PAHs of interest.
SYNTHESIS AND SPECTROSCOPY OF TRIBENZO(a,g,m)CORONENE,
A NEW, EXCEPTIONALLY STABLE, FULLY BENZENOID PAH
S. Obenland and W. Schmidt
Biochemical Institute
Sieker Landstrasse 19
2070 Ahrensburg
W. Germany
ABSTRACT. Two independent syntheses for the hitherto unknown tribenzo(a,g,m)coronene, a member of the "fully benzenoid" PAH family
are reported. This exceptionally stable hydrocarbon melts at 537 - 9b'C
and can be sublimed, without decomposition, in an open test tube. The
UV, fluorescence, phosphorescence, triplet-triplet absorption and IR
spectra are reported and discussed in connection with the trigonal symmetry of the molecule. The T1 lifetime at 77 K in dioxane, 10.6 sec,
is the highest so far recorded for an organic compound.
Due to their thermodynamiC and photochemical stability, polycyclic
aromatic hydrocarbons (PAHs) are ubiquitous in our environment, being
formed during incomplete combustion and high-temperature treatment
of any kind of organic material. Very recently, they have been implicated (1,2) as probably the most abundant class of organic compounds present in the interstellar medium.
As the collection of interstellar dust particles for subsequent analysis is not possible in the foreseeable future, all speculations concerning the nature of the PAH or PAHs present in space are indirect, resting upon stability considerations and comparison of the astrophysical UV
extinction and m emission data with laboratory spectra.
Among the huge PAH variety, the "fully benzenoid" ones (3) are particulary stable and are thence attractive candidates (4) to account for
the astrophysical data. We report here on the synthesis and spectroscopic properties of tribenzo(a,g,m)coronene, a distinguished (5,6) member of this particular class. In addition to being fully benzenoid, this
molecule has D3h symmetry which might result in unusual optical properties, e. g. a very long triplet lifetime.
Tribenzo(a,g,m)coronene was synthesized in two independent ways.
The first, conventional, synthesis (Fig. 1, above) utilizes the known
165
A. Uger et al. (eds.), Polycyclic Aromatic Hydrocarbons and Aslrophysics, 165-172:
<C> 1987 by D. Reidel Publishing Company.
166
S. OBENLAND AND W. SCHMIDT
~
--
o ..
0
2
1
=
'=
--- ~ -- ~
=3
000 0
o .
0
0
4
=
!
~
~
CH2Cl
~
CH2C1
6
=
--
<
7
i
-- < -- ~
8
l
Fig. 1. Two independent syntheses for tribenzo(a,g,m)coronene
~.
diene.1. for which an improved synthesis (5 steps from naphthalene) was
worked out. Diels-Alder reaction of! with p-benzoquinone in nitrobenzene afforded the dark-red quinone ~ (mp 270 - 6 0 C, 43 % yield) which
was reduced with Al/cyclohexanol and then cyclized with Cu at 400 0 C
to give the new perylene homologue l (pale-yellow needles, mp 459 461 oC9 47% from ~). Addition of maleic anhydride in the presence of
iodine and chloranil yielded the anhydride! (87 % yield) which was decarboxylated with Ba(OH}2 at 400 0 C to give.§. in 59 % yield. The overall yield of this laborious 10-step synthesis, based on starting naphthalene, was 0.3 %.
An elegant and more convenient access to §. is provided (Fig. 1, below) by Wittig condensation of the known 5, 8-bischloromethyl tetraline .§
with 1-naphthaldehyde in DMF9 The resulting bis-olefine 1 (pale-yellow
needles, mp 180 - 2 oC, 78 % yield) underwent a remarkably smooth
three-fold dehydro-photocyclization, for which there is scarce precedent
in the literature, to give the perylene derivative §. in 46 % yield. Conditions for irradiation, in cyclohexane solution, were as follows: 500 W
167
SYNTHESIS AND SPECTROSCOPY OF TRIBENZO (a,g,m)CORONENE
halogen lamp through a Durane filter in the presence of 0.1 equivalent
of iodine for 3 days, then 125 W mercury lamp through Durane for 5 h.
Perylene ~ was dehydrogenated in 50 % yield with Pd/C at 310 0 C to
give ~ which was converted into g as described above. A serious drawback of this reaction sequence is the limited solubility of bis-olefine 7
in cyclohexane; consequently, this synthesis can be carried through on
a small scale only
0
Tribenzo(a,g,m)coronene ~) crystallizes in straw-yellow, fine
needles which do not dissolve in warm conc H2S04 (70 0 C, 1 day) and
which melt at 537- 9 0 C without decomposition to a yellow liquid. In a
test tube, which is open to the atmosphere, g sublimes after prior melting without residue and condenses at the cold walls. UV spectra taken
before and after this experiment showed no noticeable difference.
In our opinion, this remarkable stability is due to the fact that 5 is
fully benzenoid and possesses at the same time a three-fold symmetry
axis. Therefore,.§. has the highest HUckel IT-electron energy (51. 984 P),
the highest Hess-Schaad resonance energy (1.937 M, the highest HOMOI
LUMO gap (1.035 M and the largest KekuM structure count (104) within
the C36H18 family and presumably within all C36 PAHs o
This stability is also borne out by the mass spectrum (70 eV, 300
OC) which is largely devoid of fragmentation. At m/ e II: 400, a peak
with 3 % intensity relative to the base peak at m/ e • 450 is observed;
this could possibly arise from the loss of C4H2 from M+, that is, to the
removal of one of the peripheral benzo rings. However, as the. purity
of our compound (970/0, 99%?) is difficult to assess, we are reluctant
to attach too much weight to this finding. In this connection we note that
the related molecule hexabenzo(a,d,g,j,m,p)coronene, which is not fully
benzenoid, undergoes extensive fragmentation (70 eV, 435 OC) to give
the following ions (structure of the last fragment uncertain):
mle :;;
intensity =
*
C48 H24
600
100%
-
-C6 H2
-cW cF
-C6 H2
o
0
C42 H22
526
41%
C36 H20
452
25%
To our knowledge, this is the only case of a PAH which shows clear evidence for skeletal fragmentation upon electron impact.
In line with the high molecular symmetry, the IR spectrum of .2
(Fig. 2) is relatively Simple, with the "duo" out-of plane CH vibrations
S. OBENLAND AND W. SCHMIDT
168
1614
1245
1480
1396
688
1270
811
duo
quartet
1600
Fig. 2.
1400
1200
1000
IR spectrum of tribenzo(a,g,m)coronene
800
~
748
9 [crn1
in KBr.
coming at 811 cm- I , and the "quartet" vibrations at 748 cm- l • The
weak C -H stretching band is split into a doublet appearing at 3040 and
3075 cm- 1• Clearly, ~ is a poor candidate for the astronomical species
because the CH out-of-plane bands are in the wrong place.
The presence of a three-fold symmetry axis manifests itself also in
a relatively sparse photoelectron spectrum (Fig. 3). Band assignments,
within point group Daru arE' based on the four-parameter lfuckel model
described in Ref. 7 and upon intensity considerations. The predicted
HOMO degeneracy is clearly evidenced by the Jahn-Teller splitting seen
in the 0-1 vibrational component of the first band. All bands out to 10.1
eVare due to 1T-ionizations; the onset of ,,-ionization is at about 11 eV.
The IFl value of 7.04 eV is the highest found so far in a C36 PAH.
The unique structural (Dah symmetry) and electronic (fully benz enoid) features present in.2. have interesting consequences for the UV and
phosphorescence spectra. It has earlier been shown (8) that the energy. of the p-band bears a rough linear correlation with 1F1, whereas the
energies of the a- and fl-bands correlate both linearly with IP, the mean
of !P1 and 1P2. IPt can, in principle, assume any value between 5.8
eV (the value estimated for an infinite graphitic plane) and 9.24 eV (the
value found in benzene). IF depends, to a good approximation, only on
169
SYNTHESIS AND SPECfROSCOPY OF TRIBENZO (a,g,m)CORONENE
3e'9
10.1
30 2u
1b2~·8
9.62
6
Fig. 3.
8
12
10
11.
16
IP leV)
Photoelectron spectrum of tribenzo(a,g,m)coronene W in the
gas phase at 330 oC, with MO assignments in D3h symmetry.
the size of the molecule; it is lowered with an increa.sing number of
carbon atoms. The ex- and {3-bands in the UV spectrum are hence expected and found (Fig. 4) at about the same position as in other PAHs
of the same size, namely at 427 and 325 nm, respectively, in dioxane,
and their wavelength difference nicely fits the 100 nm rule (9). On the
other hand, because of 1P1 • IP, the p-band occurs at the shortest wavelength possible for a PAH of this size, namely at 357.5 om, in dioxane.
There is satisfactory mirror symmetry between absorption and
fluorescence in Fig. 4. From this we are confident that the weak absorption feature at 427 om with f z 80 represents the true 0-0 origin of
the a-band. The weakness of the 0-0 transition stems from the fact
that it is strictly forbidden by symmetry, as is the case with other D3h
and DSh molecules like triphenylene, coronene, hexa-peri-benzocoronelle etc. In a more polar solvent like 1,2, 4-trichlorobenzene, the 0-0
170
S. OBENLAND AND W. SCHMIDT
EPS
200000
~
325
560
41.3
160000
120000
221.5
80000
ijOOOO
o
uv
200
250
300
350
ijOO
~
Fig. 4.
ij50
500
550
600
650
(nmJ
UV, fluorescence and phosphorescence spectra of tribenzo(a,g,m}coronene <ID in dioxane. Insert shows weak (X-band on
an expanded scale. Phosphorescence spectrum taken at 77 K.
band is red-shifted to 430 nm, and its E-value has doubled to 160. The
red shift for the p- and (3-bands amounts to 4 nm; however, their intensities show little change relative to dioxane.
In alternant PAHs, the energy difference between the p-band and the
phosphorescence band is a monotonously decreasing function of molecular size; for PARs with 36 carbon atoms, this energy difference is of
the order of 1.0 eV. The energy of the phosphorescence band (Tl state)
has, in turn, a bearing on the Tl half-lifetime 1"1/2; the bigger this
Tl-SO gap, the bigger is Tl/2 (10). Since, as shown above, the p-band
in §. is extremely blue-shifted, the same should hold for the phosphorescence whiCh, furthermore, should be extremely long-lived.
Both these expectations are impressively borne out by experiment.
The 0-0 transition of the phosphorescence spectrum is observed at
513.5 nm in dioxane (Fig. 4), and at 519.5 nm in 1,2, 4-trichlorobenzene, each at 77 K. The Tl half-lifetime '[1/2, measured in dioxane at 77
K, is 10. 6 sec. This is the highest lifetime recorded so far for an or-
171
SYNTHESIS AND SPECTROSCOPY OF TRIBENZO (a.g.m)CORONENE
1-T
1.0
0.8
0.6
o.y
0.2
0.0
Fig. 5.
350
YOO
Y50
500
550
A [nm]
600
650
700
750
Triplet-triplet absorption spectrum of tribenzo(a,g,m)coronene
(2) in degassed 1,2, 4-trichlorobenzene at 20 °C. Spectrum
given as (1 -transmission) vs. wavelength.
ganic compound, surpassing even the already high values of benzene and
triphenylene which are 6.2 and 8.4 sec, respectively, under the same
conditions. In 1, 2, 4-trichlorobenzene, the lifetime is reduced to 5.5
sec due to the heavy-atom effect of the solvent.
The weakness of the 0-0 phosphorescence band in dioxane is again
due to symmetry; the T1- So transition is not only spin- but also symmetry-forbidden. For this transition to occur at all, the molecule must
be vibrationally distorted from D3h symmetry, that is, radiative transitions from the vibrationally cool T1 state to the So state can only take
place to vibrationally excited levels of the latter. In a polar solvent
such as 1,2, 4-trichlorobenzene, this requirement is partly relaxed so
that the 0-0 band is the dominant one.
If PAHs like g would occur in the interstellar medium, they might
be detectable by virtue of their intense phosphorescence emission, pro-
vided a near-by star is available for excitation (11).
In Fig. 5 is shown the flashlamp-excited T1-Tx absorption spectrum of a 1.15.10- 5 molar solution of i in degassed 1,2, 4-trichlorobenzene at 20 0 C. The spectrum is relatively diffuse and covers almost
the entire visible regiOn, thereby preventing a theoretical analysis. The
r1 halt-lifetime is 0.053 msec.
172
S. OBENLAND AND W. SCHMIDT
If the PAH hypothesis of the interstellar dust (1, 2) is correct, then
the fully benzenoid compounds, as typified by g, occupy a central position in this fascinating research area. As the data presented in this abstract clearly demonstrate, compounds belonging to this family provide
the ultimate in stability which organic chemistry has to offer.
References and footnotes
(1) A. L~ger and J. Puget, Astron. Astrophys. 137, L5 (1984); J. L.
Puget, A. L~ger and F. Boulanger, Astron. Astrophys. 142, Lt9
(1985); A. Leger and L. d'Hendecourt, Astron. Astrophys. 146, 81
(1985); F. x. Desert, F. Boulanger, A. Leger, J. L. Puget and K.
Sellgren, Astron. Astrophys., submitted.
(2) G. P. van der Zwet and L. J. Allamandola, Astron. Astrophys. 146,
76 (1985); L. J. Allamandola, A. G. G. M. Tielens and J. R. Barker,
Astrophys. J. 290, L25 (1985); M. K. Crawford, A. G. G. M. Tielens
and L.J. Allamandola, Astrophys. J. 293, L45 (1985).
(3) E. Clar, Polycyclic Hydrocarbons, Academic Press, New York,
1964; E. Clar, The Aromatic Sextet, Wiley, London, 1972.
(4) W. Hendel, Z. H. Khan and W. Schmidt, Tetrahedron
~
1127 (1986).
(5) It has been claimed to be formed during catalytic hydrocracking of
mineral oil, see G. W. Hendricks, E. C. Attane, J. W. Wilson, U. S.
Pat. 3, 619407 (1971).
(6) For unsuccessful attempts aimed at synthesis see C. T. Ironside,
Ph. D. TheSis, University of Glasgow, 1959; A. McCallum, Ph. D.
Thesis, University of Glasgow, 1963.
(7) E. Clar, J.M. Robertson, R. SchIegl and W. Schmidt, J. Amer.
Chem. Soc. 103, 1320 (1981).
(8) W. Schmidt, J. Chem. Phys. 66, 828 (1977).
(9) Clar has earlier shown (3) that the wavelengths of the Ot- and f3bands are in the ratio of 1.3. According to own work this ratio
varies slightly with molecular size. The 100 nm rule due to S.
Obenland, Ph. D. thesis, University of Munich, 1984, takes this
size effect into account.
(19) W. Siebrand, J. Chem. Phys. ~ 4055 (1966); ibid. 46, 440 (1967);
ibid. 47, 2411 (1967).
(11) L. B. d'Hendecourt, A. Leger, G. Olofsson and W. Schmidt, Astron.
Astrophys., in press.
HOT CARBON ATOMS AS A POTENTIAL SOURCE FOR POLYCYCLIC AROMATIC
HYDROCARBONS
K. Rossler
Institut fUr Chemie 1 (Nuk1earchemie)
Kernforschungsan1age JU1ich GmbH
D-5170 JU1ich, FRG
ABSTRACT. Amorphous hydrogenated carbon layers (a-C:H) are among the
candidates for the species generating the interstellar optical features
assigned to PAH's. A mechanism for their formation is discussed,
including the interaction of energetic (hot) carbon atoms with hydrocarbons. the formation of unsaturated bonds (acetylene. ethylene and
derivatives) and their oligomerization or cyc1o-o1igomerization to
larger molecules resembling the PAH's. The reactions may be initiated
by impact of hydrogen onto graphitic or carbonaceous material or by that
of energetic particles onto any kind of condensed hydrocarbons.
1. I NTRODUCTI ON
Amorphous hydrogenated carbon layers (a-C:H) contain up to 50 mole %
hydrogen and vary in their consistence from polymer-like hydrocarbons
to hard "diamond like carbon". cf. (1). Their infrared spectra resemble
those observed for the "PAH's" in space and include an interesting shift
from a 3.4 vm to a 3.3 vm feature upon thermal annealing of freshly
formed films. accompanied by partial hydrogen elimination and hardening.
indicating the transition from Sp3 to spz bonding (2-4). Furthermore,
the optical properties of the a-C:H's in the UV. measured by electron
energy loss spectroscopy (1.5). seem to bear some.significance for the
interpretation of the DIB's. especially the 2200 A hump. The amorphous
films are often produced by plasma deposition from hydrocarbon vapour.
i.e. by implantation of C+, CH+. CH2+' CzH+. etc. with kinetic energies
in the keY-region. Another method could be hydrogen implantation into
carbon films. cf. e.g. (6). A hot chemistry mechanism is postulated. including the reactions of accelerated carbon atoms. either primaries or
secondaries. created by knock-on processes with hydrocarbons. In solids.
carbon atoms have to be considered only, since ions are neutralized by
fast exchange processes before undergoing chemical reactions.
2. HOT CARBON ATOM REACTIONS WITH HYDROCARBONS
In order to perform hot reactions. e.g. endothermic processes and atom
173
A. Leger et al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 173-176.
© 1987 by D. Reidel Publishing Company.
IN
K.ROSSLER
molecule interactions, the atoms must possess kinetic energies of at
least several eV. The energies of plasma deposited species depend on the
BIAS-voltage (several 100 V to some kV) and are in general in the range
between 400 and 1000 eV. Hot carbon atoms in space are found in stellar
winds and flares, cosmic rays, in the radiation belts of planets and
their satellites and in fast moving gas clouds. A velocity of 10 to 20
km S-l corresponds for a carbon atom to a kinetic energy of 6-24 eV.
Much more important than the primary species are those created by knockon in carbon containing substances. General reviews on hot atom chemistry
are given in (7-11). Hot carbon atoms are known to undergo insertion reactions into the C-H bonds of hydrocarbons yielding acetylene, ethylene.
and derivatives (7,10,12-16), cf. Fig. 1. Hot llC forms in frozen methane
28 % C2 H2 and 28 % C2 H4 , not including a higher boiling fraction of 23 %
which may be composed by oligomerization products of unsaturated molecules (12). Thus, every energetic motion of a carbon atom in a matrix containing C-H bonds results to a high degree in the formation of unsaturated carbon bonds. In aromatic substrates addition to the rings and formation of cycloheptatriene was observed (17,18).
3. FORMATION OF SECONDARY HOT CARBON ATOMS
Any kind of energetic particle penetrating a carbonaceous solid produces
secondary hot carbon atoms by collisions. These are in general more numerous than the' primary ones. Their energies are lower than those of the
latter, but still sufficient for hot reactions (11,19). A computer simulation of collision cascades of energetic H, He and C in graphite and
frozen CH 4 as model systems for carbon compounds with low and high Hcontent, resp., was performed with the code MARLOWE (19,20). The ranges
of primaries, the number of secondaries and their energy distribution
were calculated. Fig. 2 represents the projection of a typical cascade
of a 1 keY carbon in polycrystalline graphite. Many secondaries are
formed in a dense array with appreciable energies. Fig. 3 represents the
projection of a collision cascade of a 100 eV carbon in polycrystalline
CH 4 (77 K). Fig. 4 shows the energy distribution of secondaries created
by 10 2 , 10 3 and 10 4 eV carbon primaries in frozen CH 4 • Table 1 lists the
average number of all hot carbon atoms (primaries, secondaries, tertiaries, etc.) per cascade in some typical systems. The overlap of several
cascades yields a high local concentration of unsaturated bonds and molecules. This has been confirmed by mass spectrometry of the compounds
sputtered by keY protons from frozen CH 4 , which showed high yields of
unsaturated species (21).
4. INTERACTION OF UNSATURATED SPECIES
Unsaturated bonds in the a-C:H's and the hydrocarbons in space can themselves be responsible for the observed IR-features. Beyond that it has
to be considered that acetylene can easily cyclo-oligomerize to compounds
such as benzene, cyclo-octatetraene, etc., at ambient temperature by the
intermediacy of transition metal complexes and at about 400 0 C without
catalysts (22.23). The energies deposited in the collision cascade may be
sufficient to induce cyclo-oligomerization processes of adjacent unsatu-
HOT CARBON ATOMS AS A POTENTIAL SOURCE FOR PAH's
[~: f .
CH. -
[ H-
~ ~
-
H
f-
[n' . (H. - . [ H - ~ =(H, ] .*H'
175
15
H( • CH • 2H'
--+
projectl!d onto the
(0011 ptone
H,(
=(H,
[~: f
.
Fig.1. Insertion of hot carbon into
C-H bonds of hydrocarbons
30
O~"""fI~
0..-,
...-...
..... - . - • •'1
(/(H~ (s)
10
Ed·S-SoV
!10' oV I
rShlS
(100) oxi$,
1 -
r'$.2. Collision cascade of a 1 keY
C in polycryst. graphite
10
projected onto the
(OO!) pt...
Ilo'ovi
e"
1:25: S
.t. . . ......
0 ............ ",..,..,.,
!wevi
r 5:2
o
"
0,..,..... .......'"
.............v
O~~-r-r-r~~~~~--r
o
~
10
1 5 11 50 101
energy of secondary projed; los, oV-
Fig.4. Energy distribution of secon- Fig.3. Collision cascade of a 100 eV
daries in C/CH~ (77 K)
C in polycryst. CH~ (77 K)
TABLE I Total number of hot atoms per cascade in some systems
prim. energy',
eV
CIC (graphite)
e/CH,(s)
He/CH,(s)
H/CH,(s)
10'
4-5
3-4
1-2
1
10'
30-40
20- 25
10-15
ca. 6
10'
200 - 250
130 -150
ca. 50
ca. 25
L
176
K. ROSSLER
rated species even at low temperatures. Radiolysis, photolysis and special bridging reactions of hot carbons (14,17,18) may help in the formation of cyclic and polycyclic aromatic compounds.
5. SPECIFIC ROLE OF HOT REACTIONS
The fact that the quality of a-C:H's, and particular the amount of Sp3
or Sp2 hybridization depends very much on the energy of plasma deposition,
the BIAS energy (I), allows the conclusion that the concentration of
defects and secondary atoms governs the chemical form of the products.
Hot atom reactions may, besides many other pathways to PAH's a~d related
compounds in space, specifically contribute to the build-up of hard
hydrogenated carbon layers with preferential Sp2 bonding. It seems
interesting for future research to correlate sensitive interstellar IRfeatures, such as the 3.4 to 3.3 ~m shift, to regions of higher or lower
velocity of gas clouds and, thus, stronger or weaker contribution of hot
reactions. This may result in a better understanding of the "tuning" of
infrared f~atures in different regions of space.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
9.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
ll.
22.
23.
J.C. Angus, P. Koidl, S. Domitz, in Plasma Deposition of Thin
Films, C.R.C.-Press, 1986, in press.
~schler, A. Bubenzer, P. KOidl, A~el.Phts.Lett. 42, 636 (1983).
B. Dischler, A. Bubenzer, P. Koidl, T ,n So id Films-r16, 241 (1984).
M.P. Nadler, T.M. Donnavan, A.K. Green, ,61d. 116, 24r-[1984).
J. Fink, et al., Phys.Rev. B 30,4713 (1984). V. Philips, E. Vietike, this-TSsue.
G. Stocklin, Chemie hei6er Atome, Verlag Chemie, Weinheim 1969.
T. Tominaga, E. fachikawa, Modern Hot Atom Chemistry, Springer
Verlag, Berlin 1981.
K. Rossler, H.-J. Jung, B. Nebeling, Adv.Space Res. 4, 83 (1984)
K. Rossler, in The Atmospheres of Saturn and Tltan, ~SA-SP 241,
175 (1985).
K. Rossler, Rad.Effects 1986, in press
G. Stocklin, et al., J.Phys.Chem. 67, 1735 (1963)
G. Stocklin, A.P. Wolf. J.Am.Chem.SOc. 85, 229 (1963).
J. Dubrin, C. MacKay, R. Wolfgang. J.AmJrhem.Soc. 86, 959 (1964).
M. Marshall, C. MacKay, R. Wolfgang, J.Am.Chem.Soc:-86, 4741 (1964).
J. Dubrin, C. MacKay, R. Wolfgang. J.Chem.Phys. 41, 3Z67 (1964).
R.M. Lemmon, Acc.Chem.Res. 6, 65 (1973).
-R.M. Lemmon. W.R. Erwln, ScT. American 232(1), 72 (1975).
K. Rossler, G. Eich, in prosertles and Interactions of Inter~laneta~ Dust, Reidel, Dor recht 1985, 357.
:T. Roblnson, J.M. Torrens, Phis.Rev. B 9, 5008 (1974).
A.E. de Vries, et al., Nature 3 I, 40 (~4).
W. Reppe, M.v. Kutepow,~g~ Angew.Chem.lnt.Ed. 8, 727 (1969).
D. Neville Jones (ed.), Comprehenslve Organ,c Chemlstry, Pergamon
Oxford 1979. 1, 1208
CARBON COMPONENTS OF INTERSTELLAR DUST
J. Mayo Greenberg, M.S. de Groot and G.P. van der Zwet
Laboratory Astrophysics,
P.O.
Box 9504,
2300 RA
The Netherlands
Leiden,
1. INTRODUCTION
In this meeting a new carbon component of interstellar dust provides the
focus of attention. The PAH's are the incentive for reconsidering many
of the earlier suggestions of such carbon components as graphite,
carbonaceous particles, amorphous carbon particles, and photochemically
produced organic refractory materials to see whether or not there are
some connections among them. We shall consider these components only in
connection with various absorption (extinction) features of the interstellar medi um. Thus we restrict ourselves to solid particles even
though we realize that the overlap with large molecules may become fuzzy
in a certain size range.
2. GRAPHITE
Graphite was the first candidate suggested for the 220 nm hump (1). One
of the basiC difficulties with this material has been the fact that the
shape and position of the hump depend strongly on the size and shape of
the particles whereas the observed hump is highly uniform in structure
(2). Another problem has been the fact that, although the hump and the
visual extinction are remarkably well correlated, the sources of these
two dust components are independent in the sense that the visual
extinction particles are basically created in the interstellar medium
while the graphi te is (presumed to be) formed in stellar atmospheres
(3). We shall not discuss here the question of how graphite carbon can
form in the first place. However if interstellar dust provides the basic
components of the protosolar nebula, as is indicated by the evidence of
their presence in comets, then there should be some remnarts in primitive meteorites. The lack of well-crystalyzed carbon in meteorites (4)
would seem to preclude graphite as a substantial component of the interstellar medium. Certainly not as much as the - 32% of all carbon
required in the form of graphi te to g1 ve the 220 nm hump (2, 5). There
is also a problem of replenishing the graphite at a rate suf.f'icient to
coUnter Its erosion and destruction rate (6, 7). This problem also
177
A. Uger et al. (ells.), Polycyclic Aromolic Hydrocarbons and Astrophysics, 177-181.
© 1987 by D. Reidel Publishing Company.
178
exists for
sections.
J. M. GREENBERG ET AL.
the QCC's and amorphous carbon as discussed in the next
3. QUENCHED CARBONACEOUS COMPOSITE (QCC)
The suggestion has been made by Sakata et al. (5) that a carbonaceous
material containing conjugated double bonds produces a satisfactory
match to the 220 nm hump. As they point out, there seems to be evidence
that the QCC they synthesized by plasma discharge in a gas, consisting
of hydrogen and carbon, contains at least two components: fine graphite
grains and hydrocarbons having conjugated double bonds, as well as such
molecules as polyynes which they removed by long time solution in
methanol. The suggested source of these particles is, as for graphite,
from stellar atmospheres which again raises the question of how to
account for their tight correlation with the visual extinction. The
visual extinction and the hump can not be both gi ven by QCC's for a
number of reasons: (1) The QCC's are strongly absorbing in the visual
and therefore are eliminated as providing the interstellar polarization,
(2) The amount of QCC required to give the 220 nm hump uses - 80% (6) of
the available carbon in space which does not leave either enough to give
the visual extinction as an independent component or enough to account
for gas phase molecular species. However, as we shall see, there is some
indication that something like the QCC may be produced in space and
satisfy the above criteria.---4. AMORPHOUS CARBON
Amorphous carbon, or soot, has been produced and studied by a number of
workers (9, 10). The absorption measurements for small particles
produced by various methods have been used to give not only the shape of
the extinction curve but also - and this is quite important - the
strength of the absorption per unit mass. Disregarding the fact that the
hump shape is not precisely correct, which could perhaps be remedied by
some change in physical or chemical structure - the basic problem is
that all the materials of this type appear to require more than the
cosmic abundance of carbon to give the qump (6).
5. ULTRAVIOLET PROCESSED GRAIN MANTLES
A purely interstellar process leading to the presence of a carbon
component follows from ultraviolet photo processing of grain mantle ices
(11, 12). However the question of whether the organic refractories so
produced contribute to the absorption features - whether visible or
ultraviolet - of interstellar dust has not yet been fully explored. A
factor in favor of some interstellar produced organic particles
contributing to the 220 nm hump is the degree of correlation between the
visual extinction and the 220 nm hump (2, 13). In other words we might
expect a visual-hump correlation if the "visual particles" can be the
CARBON COMPONENTS OF INTERSTELLAR DUST
179
source of the "220 nm particles".
-----H-owever, we must at the outset point out that the organic
refractory mantles as such can not be responsible for the ultraviolet
absorption feature. This is because the mean particle sizes producing
the visual extinction relative to 220 nm are such that 2~a/A > 1 and the
extinction in the ultraviolet in at the saturation level. As a consequence, any extra absorption in the mantle material would either be
unnoticeable (if it is superposed on an already significant continuum
absorptivity) or could even lead to a depressed extinction.
The con.tribution of the organic refractories to the 220 nm hump, if
true, would have to be as very small particles according to the
extinction law constraints (13~n other words there would have to be
some sort of small organic particles which are relatively strong
absorbers at 220 nm relative to the near and far ultraviolet. We know
that the organic refractory mantles can not be strongly absorbing in the
visual because, otherwise, the core-mantle particles would violate the
low absorptivity (m" < 0.05) condition demanded by the Circular polarization reversal occuring at the maximum of the linear polarization (4).
I t is not known what the near ultraviolet absorpti vi ty of the interstellar organic refractory is but we shall assume, based on laboratory
produced samples which are at most partially evolved photochemically,
that it is perhaps not much greater than, say - 0.5; Le. any extra
absorpti vi ty could be observed at least in small particles.
From our other paper in this meeting (15), we have proposed that
certain linear unsaturated carbon chain molecules have extremely strong
absorption in the 200 nm region. We have, in matrices, produced these
absorptions by ultraviolet photolysis of a number of starting organic
molecules - either hydrocarbons or those containing oxygen or nitrogen.
The stable end products appear to be linear chain molecules.
Although the absorption strength per cartJ.9naatom of these molecules
is quite a bit larger than the 0c - a x 10 1 cm 2 for small graphite
particles (for molecules containing 6 carbon atoms, 6.4 x 10 17 ~ 0c ~
20 x 10- 17 cm 2 ) the peak wavelength and the widths are not at all a good
match. On the other hand, when put in a matrix, as an appropriate
mi xture of various polyacetylenes and cyano-polyacetylenes, the peaks
shift and the widths can be made to match the 220 nm hump. Thus one
requires a matrix and a generic mixture of these linear molecules to
provide the observed ultraviolet absorption feature. The next question
is how might such particles be made in space.
One possibility which suggests itself is that small QCC particles
have some of their internal molecular structure modified by interstellar
ultraviolet photolysis so that they contain a large enough fraction of
strongly absorbing molecules to bring the normal aO% cosmic abundance
requirement for the hump strength down to some more acceptable value.
There still remains the problem of their replenishment as well as how to
justify the correlation of the hump and visual extinction since they
arise from two different sources of particles.
Another possibility, which goes in the direction of answering this
objection, is that the small carbon containing particles in interstellar
space are produced by the core-mantle (visual extinction) particles. The
assumption here is that some of the organic refractory mantles are
J. M. GREENBERG ET AL.
180
ei ther broken off by some grain destruction process, such as shocks, or
ejected by grain explosions, and appear in space initially as very small
carbonaceous particles which do not produce the hump. Subsequently,
because of their very small si ze, the effecti veness of penetrating
ultraviolet photons in photolyzing the hydrogen away is enhanced
relative to that of the mantle of the core mantle grains where the cage
effect is more important. Thus a sufficiently large mass fraction (just
as was suggested for the QCC's) may be converted into chain molecules in
the small parti cles so that they become eff ici ent "hump" parti cles. The
critical proviso must be that this conversion to linear molecules is not
as efficient in the mantles, otherwise, the extinction effects of these
large particles would distort the hump shape (16).
As a cyclic phenomena the 220 nm particles are accreted along with
the icy mantles in the molecular cloud phase (12) and as a result of
oxydation and other photochemically induced chemical reactions in the
solid which is oxygen rich they are incorporated into the mantle
material as non 220 nm absorbing matter. The observed 220 nm decrease in
strength in molecular clouds (17) is consistent with this as well as the
fact that even i f small organic refractory particles are ejected from
mantles as a result of explosions, there is not enough ultraviolet to
convert them to "hump particles" before they are reincorporated in the
mantles (18). Regeneration of the hump particles then occurs when the
particles reappear in the diffuse cloud phase. The mass of such
parti cles need only be a small fraction, - 5%, of the molecular cloud
mantles (as little as 2% of the total carbon) depending on their 220 nm
absorptivity, and what is more important they only have to survive
during their diffuse cloud phase. Since, their production rate is
proportional to the the amount of visual extinction particle material
the hump/E( B-V) correlation becomes plausi ble. Note that the absorbing
molecules must be incorporated into solid particles so that their
absorptions-are broadened and shifted by matrix effects. The minimum
carbon consumption of - 2% is based on the assumption that all the
molecules in the matrix are strong absorbers but there are enOi:igh of
them in an individual particle to provide the required matrix effect.
Finally, we note that a substantial presence of triply bonded
carbon (H-C"'C) is excluded for the grains which provide the 3.4 \.lm
feature along with the visual extinction because of the clear absence of
a narrow 3.0 \.lm absorption feature towards the galactic center (19, 20).
REFERENCES
( 1)
(2)
(3)
(4)
(5)
(6)
(7)
T. P. Stecher and B. Donn, Astrophys. J. 142, 1681 (1965).
G.C. Chlewicki, Ph. D. Thesis, Leiden (1985).
S.J. Czyzak and J.J. Santiago, Astrophys. and Sp. Sci. ~, 443
(1973).
J.A. Nuth, Nature, 318,116 (1985).
S.S. Hong and J.M. Greenberg, Astron. Astrophys. 88, 189 (1980).
J.M. Greenberg, in "Light on Dark Matter", ed. F.P. Israel
(Reidel, Dordrecht), 177 (1986).
B.T. Draine and E.E. Sal peter , Astrophys. J. 231, 438 (1979).
CARBON COMPONENTS OF INTERSTELLAR DUST
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
181
A. Sakata, S. Wada, Y. Okutsu, H. Shintani and Y. Nakada, Nature,
301, 493 (1983).
W.W. Duley, I. Najdowski, Astrophys. Sp. SCi., 95, 187 (1983).
A. Borghesi, E. Bussoletti, L. Colangeli, Astron. Astrophys. 142,
225 (1985).
J.M. Greenberg, in "Molecules in the Galactic Environment" eds.
M.A. Gordon, L.E. Snyder (J. Wiley and Sons), 93 (1973).
J.M. Greenberg, in "Submillimetre Wave Astronomy", eds. T.P.
Phillips and J. Beckman (Cambridge U. Press), 261 (1982).
J.M. Greenberg and G.C. Chlewicki, Astrophys. J. 272, 563 (1983).
P.G. Martin, Astrophys.J. 187, 461 (1974).
G.P. van der Zwet, M.S. de Groot, F. Baas, J.M. Greenberg, these
proceedings (1986).
J.M. Greenberg, in "Stars and Stellar Systems, Vol. VII: Nebulae
and Interstellar Matter", eds. B.M. Middlehurst and L.H. Aller (U.
of Chicago Press), 221 (1968).
R.C. Bless and B.D. Savage, Astrophys.J. 171, 293 (1972).
B.C. Bohlin, B.D. Savage, Astrophys. J. 244, 109 (1981).
B.D. Savage, Astrophys. J. 199, 92 (1975~
J.M. Greenberg, in "Les Spectres des Molecules Simples an
Laboratoire et en Astrophysique, (U. de Liege Inst. de'Astrophysique), 555 (1977).
W. Schutte and J.M. Greenberg, in "Light on Dark Matter"', ed. F.P.
Israel (Reidel, Dordrecht), 229 (1986).
I. Butchart, A.D. McFadzean, D.C.B. Whittet, T.R. Geballe and J.M.
Greenberg, Astron. Astrophys. 154, L5 (1986).
MOLECULAR ORJGIN OF THE 216 NM INTERSTELLAR HUMP
G.P. van der Zwet, M.S. de Groot, F. Baas and J.M. Greenberg
Laboratory Astrophysics, Huygens Laboratorium, Rijksuniversiteit
Leiden, POB 9504, 2300 RA Leiden, The Netherlands
ABSTRACT.
Vacuum ultraviolet photochemical experiments have been
performed in an argon matrix, at low temperature (- 12 K). It appears
that starting from a variety of simple organic molecules photo-products
are formed which nearly always show strong light absorption in the 200235 nm range. Infrared and ultraviolet spectra indicate that the
absorbing species are polyacetylene and cyano-polyacetylene type
molecules. Such molecules are found in space. The polyacetylenes and
cyano-polyacetylenes have extremely strong light absorption bands in the
ul traviolet; in a matrix the shift in band position is qui te large. It
is suggested that such molecules on grains could give rise to the 216 nm
interstellar extinction hump.
1. INTRODUCTION
The discovery by Stecher [1 J in 1965 of a prominent maximum in the
interstellar extinction at 216 nm (4.6 lim -1) has gi ven birth to many
theories about its origin. Much work has been done on models of grains
[2-5J in which a graphitic component gives rise to the 216 nm "hump".
Such models require a delicate balance of grains of different size as
well as different composition. Carbonaceous grains [6J and amorphous
carbon grains [7J have also been proposed as possible sources of
extinction in the interstellar medium near 220 nm. Another suggestion
concerns small grains of magnesium silicate composition [8,9J. Here the
light absorbing species (at 222.5 nm) is 0 2- in low coordination sites.
Furthermore, i t has been put forward [10J that hollow bacterial grains
in space would lead to an extinction nearly identical to that observed.
Unfortunately independent evidence for the above models is lacking.
It is known, however, that under conditions prevailing in interstellar
space photochemistry occurs, in particular on grains [llJ. Through
photochemistry a variety of organic molecules can be formed from simple
ingredients like H2 0, CO, NH 3" CH4' H2S, etc. In fact a great many
molecules and radicals with up to thirteen atoms per molecule have been
observed in space [12,13J. Such photochemistry takes place by irradiation with ultraviolet light down to 91.2 nm [14J, at low temperature
183
A. Leger et aI. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 183-195.
© 1987 by D. Reidel Publishing Company.
G. P. VAN DER ZWETETAL.
184
(- 10 K on grains) and at very low pressure. Photochemical experiments
with vacuum ultraviolet light (100-200 nm) have shown [15J that with
such hi gh energy Ii ght quanta many reactions are possi ble and that
highly unsaturated organic molecules are easily formed. Many unsaturated
organic molecules with conjugated bonds show strong light absorption
between 200 and 300 nm. Hence, if such molecules are common in interstellar space this should manifest itself in an increased extinction.
Light absorption by unsaturated linear carbon chain molecules has
earlier been suggested [16J to be a possible origin of the 216 nm
interstellar extinction hump. The photochemical experiments described in
this paper were started with the intention to elucidate the possi ble
role of molecules in space with regard to the observed extinction curve
in the ultraviolet part of the spectrum.
The 216 nm hump requires a considerable extinction in relation to
the abundance of carbon. As pointed out by Greenberg [17 J, if 32% of
available carbon in the form of graphite is used, a mass extinction
coefficient of 4 x 10 5 cm 2;g is needed. This yields for the crosssection per carbon atom: a = 8 x 10-18cm 2. The width of the hump implies
an oscillator strength exceeding one. Clearly, if molecules in space
gi ve rise to the 216 nm hump they must show a very strong absorption
near this wavelength.
2. EXPERIMENTAL
Sample materials were deposited on a sapphire plate, housed in a vacuum
chamber and cooled to about 12 K by an Air Products displex closed cycle
refrigeration system. The sapphire plate could be rotated inside the
vacuum system. After deposition the samples were irradiated with light
from a microwave powered hydrogen lamp, with maximum ligh intensity at
160 nm. Usually this was followed by irradiation with a high pressure
mercury arc (Oriel, model 6141). A magnesium fluoride window was mounted
between the hydrogen lamp compartment and the main vacuum system.
Ultraviolet absorption spectra were recorded in transmission using a
deuteri urn lamp as a light source and a motor-driven monochromator
(Jarrel-Ash, model 82-410). During irradiation with hydrogen light Fcentres were generated in the sample material leading to a continuous,
constant light absorption between 200 and 300 nm. The continuous
absorption was eliminated by comparing the lamp output curve (without
sample) at reduced height with the actually observed transmission curve.
From the ratio of intensities the absorbance as a function of wavelength
was obtained.
Infrared measurements were taken in reflection, using a similar
low-temperature set-up and a Digilab FTS-15BD spectrometer.
s.
RESULTS
We searched for product molecules which show strong light absorption
near 216 nm. Such molecules must have conjugated bonds and thus contain
several carbon atoms. On the other hand too many carbon atoms in an
185
MOLECULAR ORIGIN OF THE 216 nm INTERSTELLAR HUMP
unsaturated molecule leads to several absorption bands including bands
well above 216 nm, contrary to our requirement. With thi s in mind we
selected a number of starting compounds with 6 to 8 carbon atoms per
molecule and a few outside this range.
0+
106
hv
'0:('
0+
OB
I
hv
0.6
Ul
IJJ
<>
<>
z
Z
<t
CD
<t
CD
:3
g; 0.4
0.4
V1
V1
CD
<t
CD
<t
0.2
0.2
200
Figure 1
Photolysis of benzene in argon (1:10000)
at 12 K. Irradiation: hydrogen light 3
hours.
The
absorbance 1010g (I III is
plotted as a function of wave~ength.
Assignment of photo-products as indicated:
dlmethylenecyclobutene (1).
benzvalene
(2). benzyne (3) and fulvene (4).
Figure 2
Photolysis of benzene in argon (1 :10000)
at 12 K. Irradiation: hydrogen light 20
hours. The spectrum of deposited benzene.
before irradiation. is also gi ven (x 5).
Figures 1, 2 and 3 show spectra obtained when benzene (C 6H6 ) in an
argon matrix was photolyzed at 12 K with hydrogen light (Amax= 160 nm).
Essentially this is photolYSiS of isolated molecules. The photochemistry
of benzene has been studied in great detail [18-21 J and many products
have been identified. However, no information was found on benzene in a
matrix irradiated at wavelengths as short as 160 nm. We used literature
data [22-29J to assign a number of peaks of figure 1 to specific
molecules, as indicated. The peak around 220 nm, for which no aSSignment
is given, is of particular interest. Prolonged irradiation with hydrogen
light made it grow while the neighbouring peak nearly vanished, see
figure 2. Similarly, the double peak at 246 nm increased while the peak
to the left of it decreased. Further irradiation with mercury light
diminished the height of the 246 nm peak whereas the 220 nm peak
remained unaffected (or grew a little). This is shown for a separate
experiment in figure 3. It is known that ful vene, which gives rise to
the 246 nm peak, is not photochemically stable [25J. Thus it appears
that photolysis of benzene in argon at about 12 K can lead to a
photochemically stable compound which shows strong light absorption
around 220 nm. It may be noted at this pOint that there was ample light
G. P. VAN DER ZWET ET AL.
186
to cause photochemical action at 220 nm. The intensity of the hydrogen
lamp at this wavelength is at least one quarter of that at its maximum
at 160 nm. The light absorption cross-section (at 220 nm) of the stable
compound could be roughly estimated, using the measured absorption of
deposited benzene (figure 2) and known extinction coefficients of
benzene and ful vene, and assuming that one benzene mOle.:ul-e can gi ve
only one product molecule. The resu~t is Cimol = 3. a x 10 1 cm 2 , which
yields per carbon atom Ciat = 6.4 x 10 17 cm 2 . This is about eight times
higher than the value mentioned in the introduction for graphite. The
width of the 220 nm peak in figures 1-3 implies that the oscillator
strength f exceeds one. The 220 nm peak could also be generated by
photolysis of cyclohexane (C6H12)' in which case other peaks due to
ful vene, et.c. were less dominant. Clearly, during photolysis abstraction
of hydrogen taken place.
I
I
0+
hv
I
w
'"z«
w
'"«z
III
a:
0
U)
III
«
III
g§ 08
1.5
U)
III
«
1.0
0.5
200
240
280
WAVELENGTH Inml - -
Flsure 3
Photolysis of benzene in argon (1 :10000)
at 12 K. Irradiation: hydrogen light 20.5
hours. merctry light 17.25 hours.
Fll\W'e _
Photolysis of toluene in argon (1 :1000) at
12 K. Irradiation: hydrogen light 8 hours.
mercury light 40.25 hours.
Results for toluene (C7Ha) are given in figure 4. Again hydrogen
light and mercury light were used. The mercury light removes most of the
absorption due to non-photostable material and leaves a "clean" band due
to the stable compound. Compared to the benzene case the absorption
maximum has shifted to a somewhat longer wavelength (226 nm). Apart from
this shift the results are very similar: in both cases a stable, more or
less single peak is obtained.
The spectrum obtained after photolysis of ortho-xylene (CaH10) is
gi ven in figure 5. Photolysis of meta-xylene, para-xylene or ethyl
benzene (all CaHl0) produced spectra identical to that of figure 5. A
MOLECULAR ORIGIN OF THE 216 nm INTERSTELLAR HUMP
187
product is formed which shows strong light absorption around 220 nm, but
unlike the spectra of figures 1-3 a clear structure is present. The
somewhat broader band at 257 nm is probably due to substi tuted ful vene.
It decreased by irradiation with mercury light. Four peaks are clearly
visible in figure 5, the position of a fifth one is also indicated. The
distance between the_ peaks, sta!'ting from _the long wave~ength side, was
found to be: 1788 cm T, 2298 cm 1, 1784 cm 1 and 2131 cm 1. It points to
carbon-carbon stretch vibrations in a molt¥cule with triple bonds [30].
The peak absorption cross-section at 222 nm in figure 5 was found to be
about the same as that corresponding to the peak at 220 nm in figures 1-
3.
9
H3
CH3-CH-CH
0.8
9H3
z 9- CH3
CH3
112
+
.
h~
Ul
..
U
Ul
U
Z
Z
<II
<II
0::
Vl
<II
<II
..
~0.8
~ 0.4
.
0.4
0.2
200
Fl..... e 5
PhotolysIs
of
ertho-xylene
in argon
(1 :1000) at 12 K. IrradIation: hydrogen
light 10 hours. mercury light 9.25 hOlrs.
200
Fl..... e 6
PhotolysIs or iso-octane in argon (1 :1000)
at 12K. Irradiation: hydrogen light 43.,5
hours. mercury light 21 hours.
Figure 6 shows the result obtained by photolysis of iso-octane
(CSH18 ). This is an alkane, completely different from the aromatic
starting materials mentioned above. Surprisingly the sharp absorption
maxima of figure 6 coincide with those of figurf! 5. Clearly the same
photo-product is formed, but in the case of figure 6 additional products
are produced which gi ve rise to a broad absorption around 219 nm
together with an absorption tail above 240 nm. Post-irradiation with
mercury light raised the absorption around 219 nm while reducing the
tail. Ultimately irradiation became ineffective due to deposition of air
on the sample.
The starting molecules discussed up to now contained 6-8 carbon
atoms and it proved interesting to use molecules outside this range. The
G. P. VAN DER ZWET ET AL.
188
result of photolysis of cyclopentene (C5Ha) is given in figure 7. There
is strong absorption around 221 nm which, considering the number of
irregularly spaced peaks, must be due to a mixture of compounds.
Apparently with less than 6 carbon atoms no predominantly single product
is formed. Alternatively a number of experiments were carried out with
larger molecules. The result for mesitylene (C gH12 ) is given in figure
a.
It
appears
that
intermediate products
(possibly substituted
benzvalene and - ful vene) are formed which are stable enough to resist
further attack by the hydrogen light. In this case higher energy light
quanta are required. Photolysis of biphenyl (C 1 zH10) led to a trace of a
peak at 220 nm, identical to the peak formed with benzene (figures 1-3).
O+hV
..
UJ
'-'
..
'"..
z
UJ
'-'
z
'"a:s: 08
'"
.
~ 08
o
V1
04
04
200
240
320
WAVELENGTH I n m l - -
FI_e 7
Photolysis
of
cyclopentene
in argon
(1 :1000) at 12 K. Irradiation: hydrogen
light 22.5 hours. mercury light 28 hours.
200
240
280
320
WAVELENGTH I n m l - - -
FIgure 8
Photolysis of mesitylene in argon (1 :1000)
at 12 K. Irradiation: hYdrogen light 25.7
hours. mercury light 11.2 hours.
Since nitrogen and oxygen are abundant in space it appeared
interesting to use molecules containing hetero-atoms. The spectrum of
photolyzed 2,6-dimethyl pyridine (C 7HgN) is shown in figure g. There is
strong absorption with structure around 214 nm. Compared to the spectrum
of the photo-product of the related xylenes (figure 5) the average
absorption position has shifted to shorter wavelength. I t is possible
that the spectrum of figure 9 is due to more than one photo-product, but
then they all show absorption in the same wavelength range. Thus 1 t
appears that nitrogen containing molecules can also give stable photoproducts with strong light absorption near the posi tion of the
interstellar extinction maximum.
189
MOLECULAR ORIGIN OF THE 216 nm INTERSTELLAR HUMP
Changing to oxygen containing compounds the result of photolysis of
meta-cresol (C 7HSO) is given in figure 10. The peak at 220 nm appears to
be the same as that due to the stable photo-product from benzene
(figures 1-3). In addi tion there are many other, smaller peaks resulting
from a variety of photo-produced molecules. Apparently during photolysis
CO can be split off and the overall reaction then leads to a C6 photostable material. Similarly during irradiation of benzaldehyde (CJH60) a
CO molecule is removed and the same peak at 220 nm is observe . This
indicates that, contrary to nitrogen, oxygen is not easily incorporated
in the ultimate photo-product.
H,C
o
N
CH,
+
NoH
y.hV
h~
CH,
UJ
w
'-'
z
'-'
z
CD
0::
«
«
CD
~0.8
~ 0.4
OJ)
«
CD
CD
«
0.4
0.2
200
240
280
WAVELENGTH Inml--
320
1'11\111"e 9
Photolysis of 2,'-dimethYI pyridine in
argon (1.1000) at 12 K. Irradiation.
hydrogen light 10.5 hours, mercury light
1 ~ hours.
240
280
320
WAVELENGTH Inml - -
F1l\111"e 10
Photolysis of meta-cresol 1n argon (1.500)
at 12 K. Irradiat1on: hydrogen light ~8
hours, mer cur y light 27 hours.
4. INFRARED MEASUREMENTS
Infrared spectra were taken during photolysis of benzene, ortho-xylene
and 2,6-dimethyl pyridine. For all three starting materials the aromatic
C-H stretch bands in the range 2900-3100 cm- 1 and also the C-H bending
bands between 670-770 cm- 1 disappeared in the course of the experiment.
New C-H bands emerged around 3315-3320 cm- 1 , pointing to the formation
of triple Carbon-carbon bonds (by removal of hydrogen). Similarly the
aromatic C=C ring stretch absorptions around 1400-1600 cm- 1 vanished
during photolysis. New bands appeared in the range 1700-2300 cm- 1 ,
indicating the formation of triple bonds [30J. Furthermore a m.mber of
190
G. P. VAN DER ZWET ET AL.
new bands were observed between 1400-600 cm- 1 .
On the basis of literature data some of the photo-products could be
identified. With benzene as a starting compound, after a short time of
photolysis, infrared absorption bands were observed due to ful vene [31 ]
(around 616 cm- 1 , 771 cm- 1 , 926 cm- 1 , 1079 cm- 1 and 1343 cm- 1 ) and
possibly due to benzvalene [32] (737 cm- 1 , 744 cm- 1 , very weak) and
dimethylenecyclobutene [33J (791 cm-', very weak). Prolonged irradiation
led to the appearance of bands due to triacetylene [34, 35] (3313 cm- 1 ,
1231 cm- 1 and 621 cm- 1 ).
Photolysis of ortho-xylene and 2,6-dimethyl
pyridine produced an infrared "fingerprint" qualitatively the same as
that obtained with benzene, except that the bands were shifted somewhat.
Hence, the same type of photoproduct is ultimately formed with all three
starting molecules. It pOints in the direction of polyacetylenes and
cyano-polyacetylenes, depending on atoms present in the starting
molecule. Addi tional bands in the 1700-2300 cm- 1 region indicated the
presence of carbon molecules [36, 37] and possibly partially hydrogenated polyacetylenes or radicals. A small amount of olefinic hydrogen may
have escaped infrared detection.
5. DISCUSSION
It turns out that vacuum ultraviolet photolysis in an argon matrix at
low temperature of several relatively simple organic molecules leads to
photo-products which show strong light absorption around 200-235 nm, the
spectral range where the interstellar extinction curve shows a
pronounced maximum. The photo-products appear to be stable to further
irradiation by ultraviolet light. As was shown photolysis of different
starting compounds may lead to the same end-product. For example, on
photolysis benzene, cyclohexane, meta-cresol and benzaldehyde all gave
rise to the peak at 220 nm. In the case of meta-cresol and benzaldehyde
CO is removed. Similarly the characteristic band structure of figure 4
appeared during photolysis of three different xylenes, ethyl benzene and
iso-octane. Apparently absorption of high-energy light quanta gives rise
to the formation of highly unsaturated product molecules, depending on
the number of carbon atoms in the starting molecule and more or less
independent of which compound contains those carbon atoms.
Stable photo-products can also be formed with nitrogen containing
compounds. This was demonstrated for one particular case (figure 9).
Nitrogen can form triple bonds and, hence, it can easily be incorporated
in highly unsaturated molecules. For oxygen containing molecules the
si tuation is different. Photolysis of two such molecules (meta-cresol,
benzaldehyde) appeared to lead to removal of oxygen.
The question remains which molecules are produced during photolysis. Infrared measurements indicate that polyacetylene (or cyano-polyacetylene) chain molecules are present. Starting from benzene (C6H6) the
photo-product triacetylene (H-(C.C)~-H) is observed. Similarly, from
ortho-xylene (CSH 10 ) and 2,6-dimethy1. pyridine (C 7H9N) one obtains the
product-molecules tetraacetylene (H-(C=C)4-H) and cyano-triacetylene (H(ceC)3-CN), respectively. The electronic absorption spectra in the
ultraviolet of a number of pol yacetyl enes are known [3S, 39]. In
MOLECULAR ORIGIN OF THE 216 om INTERSTELLAR HUMP
191
particular the extremely intense B bands [38J are of great interest
here. For triacetylene the B band (around 182 nm, in the gas phase)
shows little structure, similar to the 220 nm peak in figures 1-3. On
the contrary for tetraacetylene the B band system (from 207 nm downwards, in the gas phase) shows a great deal of structure, almost
identical to the spectrum of figure 4. A special feature of the B bands
is the large band shift (to longer wavelength.) when going from the gas to the solution phase. For tetraacetylene, in the gas phase and in npentane solution, the shift is about 15 nm. Similarly for pentaacetylene
it is of the order of 23 nm. The conclusion then lies near to accept
that the peak at 220 nm in figures 1-3 is caused by triacetylene and
that the spectrum of figure 4 is due to tetraacetylene. Apparently the
absorption bands are shifted considerably, to the observed posi tions, in
an argon matrix. Of course this requires experimental verification, but
our observations strongly support this view. For example it is then
understandable why so many different starting compounds all yield the
same photo-product. In line with these oonsiderations photolysis of
toluene (figure 4), containing an odd number of carbon atoms, is thought
to yield the linear molecule CH3-(C=C)3-H or possibly H2 C=C=C=C=C=C=CH 2 .
The molecule CH3-(C=C)2-H has Deen observed in space L40J. Clearly, i f
nitrogen is present it will occupy an end-position in the product
molecule. In the case of figure 7 (cyclopentene) the situation is
somewhat obscure. Possi bly a C5 compound is formed, together with
radicals.
Unquestionably with molecules in space under the influence of hard
ultraviolet irradiation photochemistry will take place. It i~ not
unlikely then that photo-products are generated which are similar to
those observed in the laboratory. Such photo-products can be made from a
variety of starting molecules. As pointed out in this paper polyacetylene and cyano-polyacetylene type molecules are formed. The fact that
many such molecules have been observed in space [12, 13J confirms this
view. Probably most of the polyacetylene and cyano-polyacetylene
molecules are contained on amorphous grains, implying that the light
absorption bands are (matrix) shifted and that the indi vidual band
structure is washed out. As shown in our experiments in an argon matrix
the shift can be quite large, to well above 200 nm for triacetylene.
Thus such molecules can give rise to the 216 nm hump in the interstellar
medium, the width being determined by the different molecules present
and by the grain composition. The polyacetylene and cyano-polyacetylene
molecules are continuously formed in the interstellar medium, hence
their life-time need not be infini tely long. The abundance of elements
and the dynamics of events in space probably limit their ultimate
molecular weight.
A search was made in the visible and ultraviolet part of the spectrum for absorptions due to carbon molecules [36, 41 J. Even though infrared measurements indicated the presence of carbon molecules no such absorptions were found, except that possi bly C4 contributed to the 246 nm
"ful vene" double peak of figures 1-3. A few very weak bands were
observed in the visible; they may have arisen from radicals or ions.
Certainly, i f large quanti ties of unsaturated molecules are present
in space, on grains or in the gas phase, this should come out in other
G. P. VANDER ZWETET AL.
192
observations. Due to the ultraviolet flux molecules may be ionized or
radicals may be formed. For di-, tri-, and tetraacetylene the ionization
energy is about 9-10 eV [42J. Emission spectra and laser excitation
spectra of the cations in the gas phase in the visi ble part of the
spectrum have been reported [42, 43J. Unfortunately the observed line
wavelengths do not agree with published data [44J on the diffuse
interstellar band positions. Hence these controversial bands cannot be
ascribed to such cations. On the other hand it takes less energy (- 125
kcal./mole or 5.4 eV) to remove a hydrogen atom from a polyacetylene
molecule and form a neutral radical. Absorption of light by these
radicals could possibly contribute to the diffuse bands. Since the total
extinction due to the diffuse bands is only a small fraction of that due
to the 216 nm hump only a small amount of such radicals, relative to the
neutral molecules, would be needed. Another feature that could t urn up
in observations is the infrared triple bond C-H stretch absorption
around 3300-3340 cm- 1 (2.99 - 3.03 ).1m). In fact it could be part of the
infrared absorption found in this region [45J.
Undoubtedly many more experiments will have to be done. This is
part of the program in our laboratory.
Discussions with Dr. G. Chlewicki, Drs. W.A. Schutte and Drs.
R.J.A. Grim are gratefully acknowledged.
6. REFERENCES
[1J
[2J
[3J
[4J
[5J
[6J
[7J
[8J
[9J
[10J
[11J
[12J
[13J
[1'4J
[15J
[16J
T.P. Stecher, Astrophys. J. 142,1683 (1965).
T.P. Stecher and B. Donn, Astrophys. J. 142, 1681 (1965).
J.S. MathiS, W. Rumpl and K.H. Nordsieck, Astrophys. J. 217, 425
(1977) .
J.M. Greenberg, in Cosmic Dust, ed. J.A.M. McDonnell (New York,
Wiley, 1978), p. , 87.
J.M. Greenberg, in Submillimetre Wave Astronomy, eds. J.P.
Phillips and J. Beckman (Cambridge Univ. Press, 1982), p. 261.
A. Sakata, S. Wada, Y. Okutsu, H. Shintani and Y. Nakada, Nature
301, 493 (1983).
-A. Borghesi, E. Bussoletti and L. Colangeli, Astron. Astrophys.
1112, 225 (1 985) .
S. MacLean, W.W. Duley and T.J. Millar, Astrophys. J. Lett. 256,
L61 (1982).
W.W. Duley and 1. Najdowsky, Astrophys. Space Science 95, 187
(1983) .
F. Hoyle, N.C. Wickramasinghe and S. AI-Mufti, Astrophys. Space
Science 111, 65 (1985).
W. Hagen, L.J. Allamandola and J.M. Greenberg, Astrophys. Space
Science 65, 215 (1979).
A.P.C. Mann and D.A. Williams, Nature 283, 721 (1980).
M.B. Bell and H.E. Matthews, Astrophys. J. Lett. 291, L63 (1985).
J.S. Mathis, P.G. Mezger and N. Panagia, Astron. Astrophys. 128,
212 (1983).
J.R. McNesby and H. Okabe, Adv. Photochemistry 3, 157 (1964).
F. Hoyle and N.C. Wickramasinghe, Nature 266, 241 (1977).
MOLECULAR ORIGIN OF THE 216 nm INTERSTELLAR HUMP
193
[17]
J.M. Greenberg, IRAS Symposium Light on Dark Matter, Noordwijk,
Holland, June 1985.
[18J D. Bryce-Smith, Pure Appl. Chern. 16, 47 (1968).
[19J O.L. Chapman, Pure Appl. Chern. 40, 511 (1974).
[20J L. Kaplan and K.E. Wilzbach, J. Am. Chern. Soc. 89, 1030 (1967).
[21 J L. Kaplan, S.P. Walch and K.E. Wilzbach, J. Am. Chern. Soc. 90,
5646 (1968).
[22J D.W.T. Griffith, J.E. Kent and M.F. O'Dwyer, J. Mol. Spectr. 58,
427 (1975).
[23J P. J. Domaille, J. E. Kent and M. F. 0' Dwyer, Chern. Phys. 6, 66
(1974).
[24J P.J. Harman, J.E. Kent, M.F. O'Dwyer and M.H. Smith, Aust. J.
Chern. 32, 2579 (1979).
[25J ~Kent, P.J. Harman and M.F. O'Dwyer, J. Phys. Chern. 85, 2726
(1981).
[26J P.J. Harman, J.E. Kent, M.F. O'Dwyer and D.W.T. Griffith, J. Phys.
Chern. 85, 2731 (1981).
[27J D.W.T. Griffith, J.E. Kent and M.F. O'Dwyer, Aust. J. Chern. 25,
241 (1972).
[28J M.E. Schafer and R.S. Berry, J. Am. Chern. Soc. 87, 4497 (1965).
[29J R.S. Berry, G.N. Spokes and M. Stiles, J. Am. Chern. Soc. 84, 3570
(1962).
[30J The Aldrich Library of Infrared Spectra, Aldrich Chemical Company,
Inc., 940 W. St. Paul Ave., Milwaukee, Wisconsin, USA.
[31J H. Schaltegger, M. Neuenschwander and D. Meuche, Helv. Chim. Acta.
"8, 955 (1965).
[32J D.W.T. Griffith, J.E. Kent and M.F. O'Dwyer, Aust. J. Chern. 28,
1397 (1975).
[33J D.W.T. Griffith, J.E. Kent and M.F. O'Dwyer, Aust. J. Chern. 25,
241 (1972).
[34J E. Bjarnov, D.H. Christensen and O.F. Nielsen, Spectrochim. Acta
30A, 1255 (1974).
[35J D.J. Huckall and J.V. Shepherd, Spectroscopy Letters 7, 381
(1974).
[36J W. Weltner, Jr. and D. McLeod, Jr., J. Chern. Phys. 45, 3096
[37J
[38J
[39J
[40J
[41J
[42J
[43J
[44J
[45J
( 1 966)
•
K.R. Thompson, R.L. DeKock and W. Weltner, Jr., J. Am. Chern. Soc.
93,4688 (1971).
E. Kloster-Jensen, H.J. Haink and H. Christen, Helv. Chim. Acta
57, 1731 (1974).
H.J. Haink and M. Jungen, Chern. Phys. Letters 61, 319 (1979).
C.H. Walmsley, P.R. Jewell, L.E. Snyder and G. Winnewisser,
Astron. Astrophys. 134, L11 (1984).
W. Kratschmer, N. Sorg and D.R. Huffman, Surface Science 156, 814
( 1 985) .
M. Allan, E. Kloster-Jensen and J.P. Maier, Chern. Phys. 7, 11
(1976).
D. Klapstein, R. Kuhn, J.P. Maier, M. Ochsner and W. Zambach, J.
Phys. Chern. 88, 5176 (1984).
G.H. Herbig, Astrophys. J. 196, 129 (1975).
I. Butchart, A.D. McFadzean, D.C.B. Whittet, T.R. Geballe and J.M.
Greenberg, Astron. Astrophys. 154, L5 (1986).
G. P. VAN DER ZWETET AL.
194
DISCUSSION
Lou AZZamandoZa:
Could all the weaker features be photobleached away
leaving only the 2200 A bump ? In other words is the species you prepared in these experiments which produced the 2200 Aabsorption more
photostable than any of the other products prepared ?
A)
Answer: In our experiments we used a hydrogen lamp (maximum at 1680
followed by irradiation with a mercury lamp. Not only were most of the
weaker features bleached away by the mercury light but it also gave
rise to some additional formations of the stable compound, indicating
the presence of an intermediate species. After about 30 hours cf irradiation further processing turned out to be impossible by condensation
of air on the sample surface.
K. R8ssZep:
Would cyclooctatetracen not be a good starting material
for your kind of new molecules ? It contains less hydrogen than most of
the other precursors.
Answer: Up to now we have only photolyzed a limited number of carefully choosen molecules, with the objective to explore the field in
different directions. There are many more candidate molecules containing C, N, Si and other elements. Cyclooctatetracen would be an excellent starting material to produce the C8 compounds.
Pieppe Cox:
The fact that the molecular species which produce the
2200
extinction hump do not show emission bands in the near infrared
may be independently confirmed by results obtained by A. Leene and
myself: using the IRAS data in the direction of stars showing uv
extinction excesses we tried to find a correlation between the strength
of the UV hump and the strength of the mid-infrared excess (the I(12~)/
1(100 ~) ratio), supposedly due to the presence of the infrared emission bands; but, we did not find any sort of correlation.
A
Answer: The above observation seems to confirm our idea that the
2200 X extinction hump and the infrared emission lines are caused by
different (molecular) species.
s.
Leach: I) Many of the bands observed after photolysis could be due
to transient species such as tra~ped radicals e.g. O-xylyl radical,
which should absorb in the 3200 A region, on photolysis of O-xylene.
Warm-up experiments should be done. 2) Did you look for fluorescence of
parent species ? Internal conversion, involving energy transfer to the
matrix, could be in competition with photodissociation. 3) In these
experiments, was the matrix annealed? After annealing, benzene in particular, would give very sharp absorption lines. Your spectra show relatively broad bands. 4) If aromatic ions were formed, as you suggest as
carriers for the visible diffuse interstellar bands, these should show
many other absorption features in the near ultraviolet. Did you look for
them ? Photo-induced ion-electron recombination experiments and warm-up
experiments would be useful as tests for transient trapped ions.
Answer: I) In addition to the photochemically stable compounds other
products were invariably formed, depending on which starting material
was used. We have not attempted to assign all peaks in the spectra.
MOLECULAR ORIGIN OF THE 216 nm INTERSTELLAR HUMP
195
2) The search for fluorescence and phosphorescence of the stable species
is on our list. High emission by these compounds may also have implications for the interstellar extinction curve.
3) No warm-up experiments were performed, but possibly the mercury
light induced some annealing. The appearance of the sample matrix
changed somewhat during irradiation with mercury light. The width of 0
narrow absorption lines was limited by our monochromator to about 15 A.
A.
4) We searched for absorption lines up to about 7000
Only a few
extremely small peaks were found in the visible part of the spectrum.
Apparently little aromatic material was left in the mixture. The observed small peaks did not correspond to diffuse bands. Experiments are
in progress to ionize the obtained stable species by irradiation with
light at 1200
in the presence of an electron acceptor.
A
CHAINS AND GRAINS IN INTERSTELLAR SPACE
H W Kroto
University of Sussex
Falmer
Brighton BN! 9QJ
UK
ABSTRACT. The intriguing abundance of long linear carbon chain
molecules in dark clouds such as TMC1 and in circumstellar shells such
as that surrounding IRC+I0216 are still not understood. Recent
radioastronomy studies are beginning to confirm that more saturated
analogues of the carbon chain species are not abundant. Recent
laboratory studies indicate that when carbon vapour nucleates to form
particles, linear chains and spheroidal shell molecules also form
probably as intermediates. The results appear to have consequences as
far as the formation, structures and spectroscopic properties of
interstellar grains are concerned, in particular those ejected from
cool carbon stars.
The polyynes which were discovered in interstellar space as a result of
laboratory experiments initiated, originally, for quite a different
purpose. They are long molecules consisting of chains of carbon atoms
linked by alternating single and triple bonds .
. . . C.C-C.C-C.C ...
Such long linear molecules present some most interesting problems in
molecular dynamics and the study of these problems, in the particular
case of cyanobutadiyne H-C.C-C.C-C.N (HC5N) coincided with the breakthrough in the detection of interstellar molecules by radioastronomy.
The observation and analysis of the microwave spectrum of this species
(1) seemed particularly significant when connected with the roughly
simultaneous realization that the previous member of the family,
cyanoethyne H-C.C-C.N (HC3N), was a relatively abundant interstellar
species. (2) This abundance and the laboratory radio frequency
stimulated a radioastronomy programme to search for HC5N in space.
The molecule was subsequently detected using the NRC 46 m telescope in
Algonquin Park in Canada. (3)
At the time the detection of HC5N was very exciting as it had six
(heavy) atoms (C, Nor 0), two more than any other molecule previously
197
A. Uger et al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 197-206.
© 1987 by D. Reidel Publishing Company.
1%
H.W.KROTO
detected. A semiquantitative view of the interstellar chemical
situation at the time indicated that small molecules with one or two
heavy atoms tended to be fairly abundant and that (after two) each
successive heavy atom tended to reduce the abundance by a factor of
about 10. This rough rule seemed to make sense in the light of some
vague statistical reasoning based on the apparent molecular composition
of the interstellar medium. Some doubts. however. about the
applicability of this rule to the polyyne family began to creep in.
Indeed searches for these types of molecule indicated that there were
clouds such as TMCI with much higher HC3N and HC5N abundances than
expected.
The detection of HC5N. together with its high abundance. clearly
promised the possibility of detecting the next polyyne. HC7N. and urged
us accordingly to attempt its synthesis and spectroscopic analysis
which resulted in its detection. (4) Not only was HC7N detected but
again the abundance was very high suggesting. now rather persistently.
that perhaps there was something special about the chemistry giving
rise to these species. The next step. the quest for HCgN. was obvious.
though the route to detection much less so. Takeshi Oka discovered a
simple empirical technique which enabled the radio lines of HCgN to be
predicted with quite remarkable accuracy by extrapolating from the
known values of HCnN with n = 1. 3. 5 and 7 which resulted in the
detection of HCgN. (5) The ratios of HCnN species with n = 3. 5. 7. 9
turned out to be 10:5.0:1.2:0.32. respectively.
It is clear that even longer species can now be sought using the
extrapolation technique and indeed a detection of HCI1N has been
reported. (6)
Ever since the detection of CH. CH+ and CN in the very early days the
problem of how such molecules could form and survive in the ISM has
been a major field for study. The detections since 1968 however have
injected the study of interstellar chemistry with a new lease of life.
The three main processes which have been postulated are: gas-phase.
ion-molecule reactions. grain surface catalysed processes and
circumstellar shell formation followed by ejection into the ISM. Of
course all three may be important and the balance is certainly not at
all clear at present.
Both gas phase ion molecule reactions and grain surface catalysis are.
however. presented with severe problems now that substantial
concentrations of long chain compounds have been observed. It is
certainly not clear that ion-molecule reactions can build up such
chains in preference to. say. branched species. especially as branched
ions are generally expected to be more stable than non-branched ones.
Indeed. if there are analogous branched hydrocarbons and related
species in commensurate numbers with the Cn chain molecules. the clouds
must contain significantly more molecules than ever considered
possible. The questions that must be answered are: Are the polyynes a
CHAINS AND GRAINS IN INTERSTELLAR SPACE
199
special case in that interstellar chemistry is geared specifically
towards producing them (as opposed to other feasible large species) or
are they in fact just the tip of an iceberg? Are there vast hordes of
molecules, unknown and as yet undetectable due to lack of ~ensitivity
and other problems? Apart from the possibility of branched carbon
chain species there are also numerous linear chain family members which
possess various degrees of feasibility.
It is not at all clear that the chains are formed on grains as it is
very difficult for them to desorb at the low temperatures that exist in
clouds such as TMC1 (c. 10-30 K).
As a consequence of these observations our recent radioastronomy
experiments have aimed at determining the relative abundances of
molecules related to various polyynes. For instance the molecules
CH2=CHC.N and HC.CCH=NH are related to H-C.C-CEN by the addition of two
H atoms and CH2=CHC.C-CaN is related in a similar way to HC.C-C.C-C.N.
Our radio measurements indicate that species with extra H atoms have
significantly lower abundances than have the parent polyynes. (7,8)
This is a somewhat surprising result in light of the fact that the
ratio C/H = 10- 4 .
Our radio data suggest that the answer may lie elsewhere, for instance,
one should consider the formation of molecules in the envelopes of cool
stars. The cool star, IRC+10216, which has a high carbon to oxygen
ratio, has now been shown to be pumping out molecules, in particular
the carbon chain compounds. In addition, it seems to be pumping out
grains and it may be that in these stars, grains and chains are formed
at roughly the same time.
Suffice it to say that interstellar studies have shown that some very
long molecules exist in space. They may be very long indeed and their
relationship with grains is far from understood and in fact, it is only
now that a possible relationship can even be contemplated. The long
chains may be an intermediate form of carbon between the well-known
small species consisting of one, two or three carbon atoms and
particles with high carbon content such as soot. Douglas has suggested
that the chains may be responsible for the Diffuse Interstellar Lines.
(9)
The main conclusion reached after several years of trying to explain
the origins of the long carbon chains we had detected in the ISM was
that some new laboratory experiments were called for. In particular
studies of efficient chain synthesis routes and the mechanisms of
carbon particle nucleation were necessary. New spectroscopic studies
of, carbonaceous material might also yield clues about interstellar
spectra in the infra red and visible - particularly with regard to the
carrier of the Diffuse Interstellar Lines.
If gas-phase or surface-catalysed reactions are responsible for the
polyynes it should be possible to prove this in the laboratory. On the
200
H. W. KROTO
other hand the detection of carbon chains in circumstellar shells
suggests that carbonaceous particles are intimately associated with the
processes which produce the chains. Perhaps the chains can provide a
key to unlocking some of the numerous problems associated with
interstellar grains themselves.
At Easter 1984 I visited Rice University (in Houston. Texas) and saw
the elegant work being carried out by Rick Smalley's group. Smalley
and his colleagues have developed a powerful technique for producing
and studying cold metal clusters by using a focused pulse of intense
laser light to vaporise material from a solid surface. (10) The
clusters are entrained in a pulse of helium gas and then expanded
through a supersonic nozzle. skimmed to produce a beam and interrogated
by a second laser which produces cluster ions whose constituents can be
determined by mass spectrometry. The apparatus seemed ideally suited
to the study of carbon aggregates and so I returned to Rice last
September (1985) to follow up this approach in a collaborative study of
the interstellar problem with Bob Curl and Rick Smalley together with
graduate students. Jim Heath and Sean O'Brien.
Almost immediately we were able to show that the same polyynes that we
had detected in space are readily produced in the vapourisation process
and that much longer linear chains with as many as 20 or more C atoms
were also produced. (11)
During the experiments a bizarre and. at the time. quite bewildering
effect was observed in that the signal for one particular cluster. that
for CSO. (which had always been prominent) became. unaccountably.
overwhelmingly stronger than the rest. So much so that under certain
conditions it was some 30-40 times stronger than the nearest
neighbours, accounting for over 50% of the observed cluster population
between C40 and C100. At first sight it appeared that the C40-C100
species consisted of sheets of hexagons blown away from the graphite
surface as essentially intact hexagonally disposed arrays of C atoms
somewhat like fragments of flat chicken wire mesh. The prominence of
the CSO peak was difficult to rationalise on this basis as we could
find no obvious reason why a SO atom sheet should show so much more
stability than any other.
After much conjecture we came to the conclusion that stability and
unreactivity .,ight be achievable i f the sheet had somehow rolled into a
closed spheroidal shell and so eliminated its reactive edges. Such a
structure would be much like the geodesic domes of Buckminster Fuller
and in fact his ideas guided much of our reasoning and played an
important part in the quest for a solution to the structure of CSO'
We came to the conclusion that the structure of the CSO molecule is a
truncated icosahedron. (12)
This beautiful shape has SO vertices. 12
pentagonal and 20 hexagonal faces and satisfies the tetravalency
condition of carbon in a truly elegant manner.
CHAINS AND GRAINS IN INTERSTELLAR SPACE
20\
There is a parallel here with Kekul~. who proposed that the properties
of the compound benzene. (CeHe). could be rationalised if the C atoms
were arranged in a ring. in that here we propose the next step - to
the surface of a sphere.
We subsequently discovered that Daedalus (David Jones) in the New
Scientist (13) in a remarkably imaginative article proposed that hollow
molecules might be made if a graphite sheet could be curled up by
introducing into a graphite array of hexagons some pentagon inducing
impurity atoms. We also discovered that Bochvar and Gal'pern had
discussed the properties of a hypothetical molecule with exactly this
structure in 1973. (14)
Perhaps the most remarkable aspect of the Ceo work lies in the
discovery that such an object forms spontaneously. We believe such
structures actually form naturally (though imperfectly) but they tend
to have smaller spheroidal structures inside in onion-like
configurations. We propose that such structures explain why soot
particles are made up of concentric spheroidal shells of hexagonal
graphitic sheets consisting mainly of C atoms.
The results have a direct astrophysical aspect in that they also imply
that such $tructures are likely to be ejected (together with chains)
from stars into interstellar space. Although the Ceo molecule
itself may not be the major species it now seems likely that the
circumstellar grains are indeed formed at the same time as the chains
and their formation and morphology are likely to be governed by the
phenomenon that we have discovered - that carbon spontaneously clusters
in the gas phase into spheroidal structures. The inertness of soot and
carbonaceous particles in general is also explained as the results
suggest that the reason for the spheroidal structure is a spontaneous
drive towards the elimination of the reactive edges of the carbon sheet
as it forms in the gas phase.
As far as interstellar problems are concerned it is likely that in the
circumstellar shells of carbon stars, processes not dissimilar to those
in these experiments occur and the solid particles which are known to
be ejected from such objects are likely to have spheroidal structures
as a consequence of the same mechanism that forms Ceo itself.
How
prevalent the t-icosahedral Ceo molecule is likely to be in interstellar space still remains to be determined.
It appears to be a
survivor of the rapid clustering process which produces spheroidal
structures of random sizes. Some structures are fortunate in that as
the shell builds uP. from small carbon species. they have the correct
configuration to close exactly.
The majority however do not. leaving
a reactive overlapping edge which can progress on to form further outer
shells in a complex 2-dimensional spiral. (15)
We have seen from our measurements that the larger carbon particles are
relatively sensitive to photoionisation laser power as they readily
shake down to produce smaller daughter fragments (n<100). On the other
2m
H.W.KROTO
hand C60 appears to be astonishingly resistant. (16) This behaviour is
unique as no other cluster species has shown anything like this level
of stability. Indeed most other clusters dissociate at relatively
modest photon flux. From these observations we can draw a number of
conclusions that are likely to be significant as far as the structure
and properties of carbonaceous interstellar grains are concerned.
First of all the results suggest that chains and carbonaceous grains
will generally tend to form at the same time and in those regions (such
as TMC1 and IRC+10216) where they are both detected, the grains are
likely to have spheroidal graphitic structures. We know that chains
and grains are forming in the shell surrounding IRC+10216 and other
similar objects from where they are continually being ejected into the
interstellar medium. Once in the general interstellar medium they are
subjected to the ambient stellar flux and occasional shock waves which
are the main agents of molecule destruction. On the basis of these
arguments we can draw some interesting conclusions about the
astrophysical significance of C60 buckminsterfullerene.
After ejection the outstanding photostability that C60 displays should
enable it to withstand all but the most severe radiative (and, we
believe, shock-wave) conditions.
We know that the resilience of CO is
the main reason for its abundance and the fact that it is so much more
widespread than other molecules, extending significantly further into
unshielded (by grains) regions of space than any other molecule so far
detected. The present observations indicate that C60 might survive in
the general interstellar medium (probably as the ion C60+) protected by
its unique ability to survive processes so drastic that, most if not
all, other known molecules are destroyed.
As C60 has no dipole moment it cannot be detected by radioastronomy
which is the most specific of interstellar analytic techniques. It has
however four ir active characteristic fundamental vibrational
frequencies. It is likely that these will be somewhat similar to those
frequencies of corannulene and coronene which do not involve C-H bonds
and some of the frequencies of the unidentified Infra Red emission
features. The general spheroidal carbonaceous structures, which we
propose as possible forms for carbonaceous particles such as soot and
interstellar grains (15) are also likely to possess similar
characteristic frequencies and as they will contain significant numbers
of H atoms in general they will also possess the C-H vibrational
modes. It has been shown by Duley and Williams (17) that the
correspondence between the UIR features and polycyclic aromatic
hydrocarbon (PAH) frequencies is good and Leger and Puget (18) have
shown that coronene also shows a good fit. Although it is not clear
just how the emission spectrum will correspond with the absorption
spectrum, it is likely that the fundamental frequencies detected in
absorption will be important. These studies present relatively
convincing evidence that the UIR features are associated with molecular
material in which carbon rings are present. It is likely that the IR
emission of soot will give rise to soaewhat similar features. CSO
CHAINS AND GRAINS IN INTERSTELLAR SPACE
203
should be even more extensively distributed and its emission excited by
pumping processes so violent that other species are dissociated.
We have also seen that under certain circumstances it is possible to
generate complexes of the form C60X in which we believe the atom X
resides inside the carbon shell. (19) During the carbon grain
formation process it is likely that many of the atomic and molecular
constituents (other than carbon and hydrogen) which are present in the
ISM are almost certain to become entangled in the resulting three
dimensional structures either molecularly bound or adsorbed/absorbed.
It is well known that graphite itself forms intercalation compounds
with numerous atoms and molecules, and that soot and carbon black
particles do so in a similar fashion. Such species are likely to show
electronic spectra which are characteristic of atoms in ligated
situations and thus broadened by the pseudo molecular environment. The
resulting lines would lie at precisely determinable characteristic
frequencies and so if these types of complexes or C60 and its bare
colleagues are responsible for any interstellar features such as, for
instance, the Diffuse Interstellar Bands, the 2170A uv band or the
Unassigned Infra Red emission features, the assignment should prove
amenable to verification in due course.
ACKNOWLEDGEMENTS
It is a pleasure to acknowledge my co-workers in the work described
here: Anthony Alexander, Colin Kirby, Lorne Avery, Norm Broten,
John MacLeod, Takeshi Oka, with whom the polyyne studies were carried
out; Don McNaughton, Osman Osman, Lesley Little and David Walton with
whom the studies of partially hydrogenated carbon chain species are
being studied, and Jim Heath, Sean O'Brien, Bob Curl and Rick Smalley
with whom the graphite vapourisation studies are in progress.
2M
H.W.KROTO
References
1
A Alexander, H W Kroto and D R M Walton, J.Mol.Spectrosc.,
1976, 62, 175.
2
BE Turner, Ap.J., 1971, 163, L35.
3
L W Avery, J M Broten, J M MacLeod, T Oka artd H W Kroto,
Ap.J., 1976, 205, L173.
4
H W Kroto, C Kirby, D R M Walton, L W Avery, N W Broten,
J M Macleod and T Oka, Ap.J., 1978,219, L133.
5
N W Broten, T Oka, L W Avery, J M Macleod and H W Kroto,
Ap.J., 1978, 223, L105.
6
M B Bell, S Kwok, P A Feldman and H E Matthews, Nature, 1982,
295, 389.
7
8
H W Kroto, D McNaughton, L T Little, N Matthews, M.N.R.A.S.,
1985, 213, 753.
H W Kroto and L T Little; to be published,.
9
A E Douglas, Nature, 1977, 269, 130.
10
R E Smalley, Laser Chem., 1983, 2, 167.
11
J R Heath, Q L Zhang, S C O'Brien, R F Curl, H W Kroto and
R E Smalley. To be published.
12
H W Kroto, J R Heath, S C O'Brien, R F Curl and R E Smalley.
Nature, 1985, 318, 162.
13
DE H Jones, New Scientist, 19663 Nov., 245 (Ariadne column).
14
D A Bochvar and G G Gal'pern, Dokl.Akad.Nauk SSSR., 1973, 203,
610.
15
Q L Zhang, S C O'Brien, J R Heath, Y Liu, R F Curl, H W Kroto
16
Y Liu, S C O'Brien, Q L Zhang, J R Heath, R F Curl,
H W Kroto and R E Smalley. To be published.
17
W W Duley and D A Williams, M.N.R.A.S., 1981,196, 269.
18
A Leger and J L Puget, Astron. Astrophys., 1984, 137, L5.
19
J R Heath, S C O'Brien, Q L Zhang, Y Liu, R F Curl, H W Kroto,
F K Tittel and R E Smalley, J.Am.Chem.Soc., 1985, 107, 7779.
and R E Smalley, J.Phys.Chem., 1986, 90, 525.
CHAINS AND GRAINS IN INTERSTELLAR SPACE
205
DISCUSSION
A. Leger: 1) In connection with astrophysical conditions, have you
tried to introduce H2 in the system ?
2) If you start building planar structures (hexagons) how can you
bend the structure and close it even if it is energetically more
stable ? The kinetics seem to be difficult to explain.
Answer: 1) A careful study of the formation of large cluster population
in the presence of hydrogen (H? or H) has not been carried out as yet.
This is a very important experiment especially for C nucleation in
circumstellar (or any other interstellar) environment.
2) Whatever the difficulty in explanation, the data are inequivocal: C60 is a very special object and our explanation is not only
elegant but has yet to be shot down.
d 'Hendecourt : How specific is this experiment ? Can the conditions of
the experiment applied to astrophysical cases (grain nucleation in
circumstellar shells) ?
Answer : We have obtained widely different results as the clustering
conditions are varied. In particular, they are slightly dependent on the
He pressure and the timing of the vaporization laser pulse relative to
the arrival of the He pulse over the target graphite surface. The
situation appears to be very complicated but three observations can be
made. If the vaporization laser is fired before the He pulse only small
clusters are detected; during the pulse both small and large (C60
prominent): after the pulse only large clusters are seen. And all
variations between.
We have deliberated long over these observations, and have not yet
reached a good rationalization. I suspect that at low pressure wall
effects are important.
S. Hukamel : What is the relative stability (energy) of the football
component with other structures
Does it compensate for the loss of
entropy ?
Answer : I am not sure that entropy is as important at second sight as
it appears at first sight. We must ask what is the relative entropy
difference between a 'perfect' hexagonal sheet and a closed shell with
pentagonal 'defects' ? There is a paper in press (JACS) on this stability which confirms our conjecture that the closed Buckminsterfullerence
for C40 - CIOO are more stable than graphitic sheets of similar size.
H.C. Siegmann: The atability of the C60La-cluster: Can it be explained
by assuming that La is located in the center of the soccer-ball ?
Answer : The data indicate that C60La or CnLa (n, even) are more stable
than the bare Cn clusters (C60 aC70 perhaps excepted). If the clusters
we see were 'say' flat plates with La attached one might expect C60Lan
with n > I etc. to be as detectable as C60La. The data suggest ~ and
only one very special site and the centre of a sphere fits such an
observation well if not better than anything else we have thought of.
H.W.KROTO
Lou Allsmsndols : 1) For this very symmetric molecule, there must be
only a few IR allowed transitions. Have you done the group theoretical
analysis to determine how many of the 180 vibrations are IR active - and
if so - do you have any idea where they are ?
2) Could the observation that when you vaporize the graphite after
the He pulse has passed shows primarily this smaller chain-like
clusters while vaporization before produces the larger clusters be a
consequence of the inherent stabilities of these two forms ? Stein has
some theoretical results which indicate that when C atoms are to produce
clusters/molecules above ~ 3000° C linear chains are favored, while
below, PAHs are.
Answer : 1) Again this problem has been addressed by Bell Labs theoreticians (in press) who find that only 4 vibrations are allowed by symmetry. I think that soon there will -be reliable estimates of these
frequencies. They should be fairly susceptible to standard free field
calculation.
2) The conditions are actually the reverse. I hesitate to present
an explanation for these observations even though my colleagues and I
have deliberated for many hours over them.
d'Hendecourt : After the experiment, were you able to 'save' these
clusters in order to study them with various methods (X-ray crystallography, visible/IR spectroscopy) ?
Answer : We are of course trying to collect viable quantities of the
products of vaporization, C60 in particular, but at present there are
non-negligible technical problems associated with this which we hope
soan to overcame.
F.J. KdowjSK : What is the repetition rate and can you use gases such as
Argon instead of helium so the species can be matrix isolated for
optical spectroscopy ?
Answer: Rate: 10 Hz. Up to 50 % Ar has been mixed with the He and
without upsetting the conditions under which C60 predominates. Above
50 % and up to 100 % the production of C60 is diminished. I suspect that
this experiment also should be redone in more detail.
Framer : Have you any data or predictions about C60 electronic spectra ?
Answer : A group at Bells Labs have a paper in press on the optical
spectrum which indicates that in the UV-visible only one transition is
allowed and it is expected to have a very large oscillator strength.
Greenberg : Have you looked into the ultraviolet absorption properties
of C60 ? Its form seems to suggest a spherised conducting shell of
!!!!S size so that it would make a nice candidate for the 2200 A band.
Answer : We are attempting to observe the RZPI spectrum using the whole
range of accessible laser dyes. I suapect that carbonaceous interstellar
matter could be spheroidal shells and the UV abso~ption should be
considered as a candidate. This is just one of the many experiments
which we are hoping to carry out on C60 and its relatives.
MID INFRARED EXCESS AND ULTRAVIOLET EXTINCTION
Pierre cox 1 and Arnaud Leene 2
1. Max Planck Institut fOr Radioastronomie, Bonn 5300, FRG
2. Kapteyn Astronomical Institute, PO Box 800, 9700 AV
Groningen, The Netherlands
1. Abstract
Correlations between the mid infrared excess and the UV extinction
parameters are investigated, using a sample of 50 stars reported to
have peculiar extinction properties. No correlation is found between
the mid infrared excess and the 2175
extinction bump, suggesting that
the carriers of the UV bump and the mid infrared emission are different.
R
2. Introduction
The 2175 R bump in the interstellar extinction curve has received considerable attention since its discovery by Stecher (1965). A variety of
explanations for this feature have been proposed, but none of the possible carriers accounts satisfactorily for all the observed properties
of the bump (see Savage, 1975 for an excellent review of this problem).
We reconsider in this paper earlier suggestions made by Donn (1968) who
proposed that a family of stable ring molecules might account for the
2175 Rbump present in the interstellar extinction curve.
Polycyclic Aromatic Molecules (PAH' s) are the most probable carriers of the set of emission features in the infrared, the former
unidentified infrared emission bands (Leger and Puget 1984, Allamandola
et al. 1985) and have been proposed to be the origin of the diffuse interstellar absorption bands in the visible (Leger and d'Hendecourt
1985, van der Zwet and Allamandola 1985). Such complex molecular
species, which are amongst the most abundant molecules after H and CO
(Leger and Puget, 1984) are also known to have strong absorptl~n bands
in the UV (Clar, 1964).
207
A. Uger et al. (eds.J, Polycyclic Aromatic Hydrocarbons and Astrophysics, 207-211.
© 1987 by D. Reidel Publishing Company.
P. COX AND A. LEENE
208
Recent laboratory work on synthesized carbonaceous composites,
which contain a significant fraction of mass in organic compounds
(mainly in the form of aromatic polymers), has shown that these substances do clearly exhibit an ultraviolet absorption bump at about 2200
~ and that their infrared spectra present most of the infrared emission
bands (Sakata et a1. 1983, 1985, Borghesi et a1. 1985, Wdowiak 1986).
A link between the carriers of the infrared emission bands and the carriers of the UV bump is thus strongly suggested from laboratory experiments, although the exact nature of the carbonaceous compounds is as
yet unclear. Further support for such a link is given by a weak but
significant correlation between the strengths of the UV bump and the
diffuse interstellar band at 4430 i in our galaxy (Seab and Snow,
1984) .
If present, the infrared emission bands dominate the emission in
the mid infrared (Puget et a1. 1985) and should significantly contribute to the intensity measured in the IRAS 12 ~ band. This would give
an excess emission in the mid infrared (over what one would expect from
the thermal radiation of a distribution of normal grains), which has
indeed been observed towards high latitude diffuse HI clouds (Boulanger
et al. 1985) and reflection nebulae (Leene, 1986), where the mid infrared excess represents 20% (or more) of the 100 ~m emission. Providing no other mechanism gives rise to the mid infrared emission, the
strength of the infrared emission bands can thus be measured by the ratio of the intensities at 12 and 100 ~, the "mid infrared excess".
The wealth of data released by the IRAS satellite provides an
unique opportunity to investigate the mid infrared excess in given
directions and to look for possible correlations with either the
strength of the UV bump or the strength of the diffuse i.nterstellar
bands. Preliminary results of such an investigation are presented here,
where the UV bump and mid infrared excess are compared.
3. Program stars and data reduction
The program stars were selected from the catalog of ultraviolet interstellar extinction excesses (Savage et a1. 1985) derived from the Astronomical Netherlands Satellite (ANS) results. In this catalog the excess in the UV extinction bump is reported with respect to a straight
baseline drawn between a normalized 11 A plot defined by the ANS photometric bands at 1800 and 2500 ~ and refered to as E(Bump). Stars
were selected according to the following criteria: great distances CD ~
1 kpc for lines of sight avoiding the galactic plane and D ~ 3 kpc for
lines of sight in the galactiC plane) and well defined extinction
excesses. Stars showing signs of nebulosity, indication of UV variability, lying in a cluster or being identified as double were systematically rejected. The final sample contains a homogenous set of 85
stars.
209
MID INFRARED EXCESS AND ULTRAVIOLET EXTINCTION
The mid infrared excess was derived in the direction of these
stars from the brightnesses as measured on the IRAS SKYFLUX maps. The
evaluation of the 12 and 100 ~ brightness from the IRAS data, requires
a careful determination of the zodiacal emission, especially in the
direction of regions where the emission is low leveled and diffusely
distributed. Results are based upon a geometrical model of the zodiacal
emission, whose characteristics are outlined elsewhere (co~land Leene,
198621 The accuracy of this model is typically 0.3 MJy ster
and 1 MJy
ster
at 12 ~ and 100 ~, respectively.
4. Correlation results and suggestions
Figure 1 shows the plot of the mid infrared excess against the UV extinction bump normalized by the colour excess, E(Bump)/E(B-V), for the
stars for which the best accuracy in the evaluation of the ratio 12/100
~ was achieved i.e. 0 :0 0.005. This limited Figure 1 to 45~ of the
original sample. Figure 1 shows no sign of a correlation betwe the
normalized UV bump and the mid infrared excess.
Figure 2 shows for completeness the plot of the mid infrared excess against the normalized extinction excess in the far UV rise as
measured at 1550 R. E(1550-V)/E(B-V), for the same stars as in figure
1. No correlation is obvious from this diagram either.
From the scatter diagrams presented in Figures 1 and 2 it is apparent that the mid infrared excess is not clearly correlated with the
strength of the UV extinction bump nor with the far UV extinction rise.
These non correlations, based on a sample of stars probing the difflJse
interstellar medium at quite different locations and hence different
physical conditions. are somewhat puzzling. but suggest that the carriers of the UV bump and the carriers of the 12 ~ emission are different.
It is interesting in this context to note that the Red Rectangle
(HD 44179) which shows all of the emission bands in its infrared spectrum (Russell et al. 1978) has an ultraviolet extinction curve with a
weak (or even absent) UV bump (~itko et al. 1982).
Testing Donn's earlier suggestions
sible for the UV extinction bump should
boratory work (careful identification
further correlation studies between the
more sensitive Quantity describing the
sion bands than the mid infrared excess.
that PAH's may be held responawait either more extensive laof carbonaceous compounds) or
UV extinction parameters and a
strength of the infrared emis-
Finally it is remarkable, ~hat for a large spread in the values of
the UV extinction excesses, the mid infrared excess varies so Ii ttle:
\,(12 ~)/IV<100 ~) ranges from 0.015 to 0.04, with a mean value of
P. COX AND A. LEENE
210
I
O.OS~
T
I
..... . .
.....
~
tTYPICAL ERROR
oI-
-
....
I
I
o
I
I
1
2
EIBUMP)/E(B-V)
3
Fig 1: Plot of the mid infrared excess against the normalized UV extinction bump.
0.05 t-
I
I
I
I
I (12fl)
I (1OOfl)
0-
I
I
I
I
. .. ,.'.....:•....
I
I
I
-
... .
ITYPICAL ERROR
0
I
I
I
2
I
I
I
4
6
E(1500A-V)/E(B-V)
I
I
8
I
10
Fig 2: Plot of the mid infrared excess against the normalized far UV
rise as measured at 1550
R.
MID INFRARED EXCESS AND ULTRAVIOLET EXTINCTION
211
0.025. This value, which is reported more or less identical for such
widely different classes of objects as galaxies, reflection nebulae and
the diffuse interstellar medium, may give some indications as to the
size distribution from normal grains down to PAH's. Since the mid infrared excess is a measure of the relative populations of PAH's (accounting for the bulk of the 12 IJII1 emission) and the normal grains
(responsible for the 100 ~m emission) (Puget et al., 1985).
5. Acknowledgements
We wish to thank M. Walmsley for a careful reading of the manuscript.
6. References
- Allamandola, L.J., Tielens, A.G.G.M., Barker, J .R. (1985) Astrophys. J. Letters 290, L25
- Borghesi, A., Bussoletti, E., Glangelli, L. (1985) Astron. Astrophys. 153, 1
- Boulanger, F., Baud, B., van Albada, G.D. (1985) Astron. Astrophys. 144, L9
- Clar, ~(1964) Polycyclic Hydrocarbons, Academic Press, London
- Cox, P., Leene, A. (1986) in preparation
- Donn, B. (1968) Astrophys. J. Letters 152, L129
- Leene, A. (1986) Astron. Astrophys. 15q:-295
- Leger, A•• Puget, J.L. (1984) Astron:-Astrophys. 137, L5
- Leger, A., d'Hendecourt, L.B. (1985) Astron. Astrophys. 146,81
- Puget, J.L., Leger, A., Boulanger, F. (1985) Astron.Astrophys
142, 119
- RiiSsell, R.W., Soifer, B.T., Willner, S.P. (1978) Astrophys. J.
220, 568
- Sakata, A., Wada, S., Okutsu, Y., Shintani, H., Nakada, Y. (1983)
Nature, 301, 10
- Sakata,
Wada, S., Tanabe, T., Onaka, T. (1984) Astrophys. J.
Letters 287, L51
- Savage, BJD~ (1975) Astrophys. J. 199, 92
- Savage, B.D., Massa, D., Meade, M-.-,-Wesselius, P.R. (1985) Astrophys. Supple Series 59, 397
- Seab, C.G., Snow, T.o. Jr. (1984) Astrophys. J. 277, 200
- Sitko, M.L., Savage, B.D., Meade, M.R. (1982) Astrophys. J. 246,
176
- Stecher, T.P. (1965) Astrophys. J. 142, 1683
- Van der Zwet, G. P ., Allamandola', D. ( 1985) As tron. As trophys.
146, 76
- Wdowiak, T.J. (1986) see these proceedings
r.
HIGH SPECTRAL RESOLUTION OBSERVATION OF THE 3.3~m EMISSION BAND AND
COMPARISON WITH LABORATORY-SYNTHESIZED QUENCHED CARBONACEOUS COMPOSITE
(QCC)
3 S4 wada3, Y. Nakada 2 , A. T. Tokunaga,
4
T. Ona k a 1,2'4A. Sakata,
4
K. Sellgren , R. G. Smith , and D. L. DePoy
1.Astronomical Institute, Univ. of Amsterdam, The Netherlands
2.Dept. of Astronomy, Fac. of Science, Univ. of Tokyo, Japan
3.Laboratories of Applied Science and Chemistry,
Univ. of Electro-Communications, Japan
4.Institute for Astronomy, Univ. of Hawaii, U.S.A.
ABSTRACT. New high spectral resolution(A/~A=1400) observations by a
cooled grating array spectrometer are reported of the 3~m emission
features of NGC7027, BD+303639, and HD44179. These data provide an
accurate measurement of the feature's center and width for the first time.
The spectrum of laboratory-synthesized quenched carbonaceous composite
(QCC) taken by the same spectrometer is also reported together with the
comparison with the observed spectra.
High spectral resolution spectra of NGC7027 and BD+303639 were obtained
on the NASA 3m Infrared Telescope Facility(IRTF) on Mauna Kea, Hawaii in
1985 June using a cooled grating array spectrometer(Tokunaga and Smith
1986). The spectra were taken over the wavelength range 3.27-3.31~m
with a spectral resolution of A/~A=1400. aAqr was observed as a standard
star at the same time to remove the atmospheric absorption features. The
measured spectra are shown in Fig. 1. The strong 3.3~m emission feature
is clearly separated from the Pfa (3.297~m) line compared to the published
high resolution spectra(A/~A=400; Geball et al. 1985). HD44179 was also
observed with the same spectrometer on the IRTF in the wavelength range
3.20-3.54~m in 1986 January.
The spectrum clearly shows the presence of
10.0
z
-'"
a 11.0
E
u
z : 10.'
.....
: 11.0
z:
.....
~ 11.0
)(
::J
io!
1.1
---.--,......-..,.--"'"t
1.0 ......- . . . - -......
NGC7027
li
E
!:!
z:
.....
I.'
BO+303639
1.0
'"• 4.'
•
z:
.....
1.0
)( I.'
:II:
::J
io!
l.1
Fig. 1. High spectral resolution spectra of NGC7027 and BD+303639
213
A. Uger et aI. (eds.), Polycyclic Aromatic HytIroaJrbons and Astrophysics. 213-214.
«:> 1987 by D. ReUkl Publi.rhing CompiUlY.
214
T. ONAKA ET AL.
four new emission features at 3.340,
3.397, 3.458, and 3.523~m(the waveQCC
lengths are preliminary).
1&1
The spectrum of laboratory-U
Z
a:
synthesized carbonaceous composite
ID
IE
(QCC) was also taken by the same
CI
en
spectrometer (Fig. 2). QCC is a
ID
a:
carbonaceous material produced
through a quenching process of
methane plasma(Sakata et al. 1983,
1984a). It has been proposed as a
.1~--L---~--~--4---~~
likely carbonaceous material in the
S.tl 1.1'1 I.n I.n I.ID 1.1\ 1.lt
'KIeRON)
interstellar space because QCC model
explains the interstellar extinction Fig. 2. Spectrum of QCC
220nm hump better than graphite
model(Sakata et al. 1983; Onaka et al. 1986). A recent study of carbonaceous material in meteorites also suggests that graphite is not a major
component of interstellar material (Nuth 1985).
Infrared measurement of QCC has shown that QCC has several features
in near infrared region(3-13~m), which are in good agreement with the
observed unidentified emission bands (Sakata et al. 1984a, b). These
features are attributed to the volatile components of QCC. Fig. 2 shows
that the center of 3.3~m band of QCC agrees well with the observed
feature. The infrared, visual, NMR, and mass spectroscopy suggest that
the volatile components of QCC consist of polyyne-ene solids, PARs, and
saturated hydrocarbons.
••
••
..,
One of the authors would like to thank the Netherlands Organization
for Pure Research ZWO for financial support of his stay in University of
Amsterdam and the Leids Kerkhoven-Bosscha Fonds for support of travel
expenses to the work shop.
REFERENCES
Geball, T.R., Lacy, J.H., Persson, S.E., McGregor, P.J., and Soifer, B.T.
1985, Ap. J., 292, 500.
Nuth, J.A., 1985, Natupe, 318, 166.
Onaka, T., Nakada, Y., Tanab~, T., Sakata, A., and Wada, S., 1986, Ap.
Space Sci., U!#.) 411.
Sakata, A., Wada, S., Okutsu, Y., Shintani, H., and Nakada, Y., 1983,
Natupe, 301, 493.
Sakata, A., Wada, S., Tanab~, T., and Onaka, T., 1984a, in Laboratory
and Observational Infrared Spectra of Interstellar Dust, eds. R. D.
Wolstencroft and J. M. Greenberg, Roy. Obs. Ed., ISSN 0309-099X, 128.
Sakata, A., Wada, S., Tanabe, T., and Onaka, T. 1984b, Ap. J. (Letters),
287, LS1.
Tokunaga, A.T. and Smith, R.G., 1986, Ap. Space Sci., U'9,. 471.
DISCUSSIO~
I:
CARBON IN THE INTERSTELLAR MEDIUM
P. G. Martin
Canadian Institute for Theoretical Astrophysics
University of Toronto
Toronto
Ontario
Canada M5S lA1
ABSTRACT. The first open discussion, on which this paper is based,
concerned the major forms in which carbon is found in the interstellar medium. Preliminary topics included the different approaches
used to describe the physics of small particles and large molecules,
and the size distribution of interstellar grains. Then the amount
of C required in various forms was discussed. in the context of gasphase C depletion. In particular, implications of spectral features
at 2200 A and 3.4 ~m in the extinction curve were addressed. The
discussion was rounded out by consideration of the possible interstellar and/or circumstellar origins of PAH's and carbon grains.
1.
INTRODUCTION
This paper is a summary of the first open discussion of the workshop. The purpose of that discussion was to focus on issues relating to the papers that had been presented in the preceding three
sessions, and to anticipate questions that might be addressed in
more detail later in the workshop. Since everyone had questions
that they wanted addressed during the meeting. this first discussion
was potentially open-ended. The conference organizers suggested a
primary focus on the interstellar extinction curve. In preparing
for and steering the discussion I attempted to broaden the topic
somewhat to a consideration of the major forms in which carbon is
found in the (diffuse) interstellar medium, as evidenced by both
extinction and emission observations, with a view to illuminating
possible interrelationships between interstellar grains and the hypothesized PAH's. This summary contains many paraphrased comments
made by the participants (many attributions are indicated in parentheses), based on notes taken by L. d'Hendecourt. The framework has
been altered slightly from that followed during the session, to juxtapose related points. To make this summary more coherent I have
a180 included some introductory remarks, and where it seemed useful
215
A. Leger e/ al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 215-222.
© 1Q1I7 by D. Reidel Publishing Company.
DISCUSSION I
216
I have added cross-references to talks both earlier and later in the
workshop (see ... ). and references to the literature.
2.
SMALL GRAINS AND LARGE MOLECULES
The optical interstellar extinction is produced by solid particles
of characteristic dimension (radius a) 0.1 ~m. Smaller particles
are required to support the observed continued rise of extinction
into the ultraviolet. In contrast to these grains. some of which
might be carbon rich. one thinks of a PAH as a large molecule. In
distinguishing between small particles and large molecules. where
does one draw the line? Why? Does it matter?
One practical concern is the use of Mie theory to calculate the
electromagnetic cross-section of a small particle. All (!) that is
required is specification of the complex refractive index (or equivalently. the dielectric function). These are measured in the laboratory for bulk samples of the material. but it is not appropriate
to use these particular constants for small particles. since the
finite size of the particle and the existence of the surface can
dramatically change the electronic and vibrational energy structure (L6ger). A specific illustration of this is provided by the
discussion of graphite and other carbon solids earlier in the workshop (see Joyes and Marchand). An estimate is that one has to worry
about usinf the bulk optical constants for sizes (radii) below about
30 A (- 10 atoms for spheres. but only - 500 for planar systems for
which one has additional concerns). One alternative approach is to
make direct laboratory measurements of extinction by small particles
(Duley. Bussoletti). These measurements usually emphasize absorption (and hence are useful indicators of the spectral dependence of
emissivity). but for any modelling involving scattering it is still
necessary to rely on a Mie calculation (Martin).
Another topical illustration of the transition from a large
molecule to a solid is provided by PAH's and amorphous carbon (the
hydrogenated form, HAC
a-C:H. is one example). Crudely speaking.
PAH's have distinct spectral features in the infrared and possibly
a low-lying vibrational continuum, whereas small amorphous carbon
particles have a strong vibrational continuum with respect to which
any spectral features have low contrast. PAH's have discrete transitions in the optical and ultraviolet (see Jortner). whereas amorphous carbons have continuous and relatively featureless electronic
spectra. with a band gap when sufficiently hydrogenated (see Duley).
In a paper later in the workshop. Duley discussed the transition
from a solid state to a quantum mechanical description in small particles, and showed that there was continuity in the results of the
two approaches in the region of overlap.
There is a geometrical distinction between the small particles
and large molecules. with potentially interesting effects. While
PAH's are specifically two-dimensional. other large molecules might
in fact be clusters or alternatively. small particles (Bussoletti).
For small particles, the surface to volume ratio can be a major
=
DISCUSSION I
217
consideration in evaluating different processes (Greenberg): for
example, molecule formation on grain surfaces depends on the area,
whereas electromagnetic absorption depends more nearly on the volume. For PAH's, volume has little meaning, and even specification
of the surface is complicated by defects (Marchand).
There is no fundamental difference (or point of demarcation between) between a small particle and a large molecule: it is just a
matter of choosing the appropriate computational tools to treat the
physics of the system in question (L6ger, Marchand).
3.
THE SIZE DISTRIBUTION OF INTERSTELLAR GRAINS
In Greenberg's presentation the evidence for (at least) three populations of grains was discussed. Each of these would have a characteristic (and different) size, and in all probability there would
be a range in sizes for each component. Mathis, Rump1, and Nordsieck (1977) actually derived a size distribution self-consistently
for their graphite + silicate grain model, namely n(a} oc a-a.s. This
is an interesting power law, in that the mass is dominated by larger
grains whereas the area is dominated by the smaller ones. A warning
about indiscriminate use of this particular form of the size distribution was made by Duley. who noted that one of the major components, graphite, might not be present in the large amount postulated. Marchand raised the possibility of cloud to cloud variations
in the size distibution, since the size of a particle is the result
of competing formation (growth) and destruction processes. There is
evidence for changes in the extinction curve in denser clouds, supporting grain growth either by accretion or coagulation (Martin).
Greenberg noted that a power-law size distribution of the above form
could be produced by grinding processes, but that the grain evolution is more complicated. In dense clouds, grains grow icy mantles
by accretion and some mantle processing begins. Subsequently, on
entering the diffuse cloud phase, these mantles are processed and
eroded (see e.g., Greenberg, van de Bu1t, and Allamandola 1984).
Whatever the evolution actually is, the resulting shape of the extinction curve in the diffuse clouds is empirically very uniform.
including the far ultraviolet.
Suppose, following the discussion below (§4) , that a few 10's
of percent of the C is in grains and somewhat less, several percent.
is in PAH's. This is crudely consistent with a prediction based
on extrapolating the grain power-law size distribution down to PAH
sizes (see also Duley). The question arises whether this is a coincidence or there is an evolutionary relationship, and then whether
there are planar PAH's or simply smaller particles. L6ger suggested
collisions between graphite grains as a source of PAH's. However.
Allamandola stated that there was little evidence for graphite in
the interstellar medium. Instead, carbon condensation in carbon
star outflows would tend to form PAH's and amorphous carbons rather
than graphite planes of large extent (see also Keller). The PAH's
would stick to grain surfaces, providing a mantle, which might later
DISCUSSION I
218
be removed. Duley wondered why all PAH's don't stick to grains in
the interstellar medium, and Tielens responded that it must be a
dynamical equilibrium leaving a steady state gas-phase population.
Puget (see presentation) observes that in dense molecular cloud
cores the relative 12-~m flux is down, which suggests that PAH's
are accreted in such an environment. See also §7.
4.
THE AMOUNT OF C IN VARIOUS FORMS
Ultraviolet absorption line measurements of the gas phase abundance
of C (and its ions) in the interstellar medium yield values lower
than the cosmic C/H abundance. With the assumption that the abundance is really the same throughout the interstellar medium, the
conclusion is drawn that gas-phase carbon is 'depleted' and must
exist in some other form, such as molecules or solids. There are
several problems with C depletion studies. First, the cosmic abundance of C by number relative to H is somewhat uncertain: Cameron
(1982) gives 4.2x10- 4 , while other authors quote as low as 3.7xlO- 4 •
Second, the C/H ratio might not be strictly constant throughout the
Galaxy. When small amounts of depletion are involved, these factors can make a significant quantitative difference in some detailed arguments about the form of depleted C. Third, the actual
measurements are mostly of neutral C, for which there are abundant
detectable lines; however, the dominant stage of ionization in the
observed diffuse clouds will be C+ which is not usually measured,
and so there is a significant and somewhat uncertain ionization correction factor in arriving at the total C abundance (Jura). Future
progress in this important question of depletion will come with the
widespread detection of weak lines of C+ (e.g., at 2325 A), which
would be possible with the Hubble space telescope. Most of the C is
in the gas phase (Tielens). An average diffuse cloud C abundance is
2.6 x 10-4 (see Whittet 1984 for a review), with lower values in more
dense clouds.
What is the form in which the depleted C exists? CO is the
most abundant small molecule observed containing C (e.g., Table 1.6
in Duley and Williams 1984). For some dark clouds eO/Htot .., is as
large as 10-4 , but in diffuse clouds the ratio is much lower, 10-6 •
Thus the depleted C must be in dust grains, or in large molecules
like PAH's.
What are the e reqUirements of models of carbon grains and
PAH's? Silicates by themselves are not abundant enough to produce the observed amount of continuum extinction for a given H column density (Av/lVH) , and so some e is also needed in grains (see
Greenberg). The type of C material is not settled. The Mathis,
Rumpl, and Nordsieck (1977) model of interstellar extinction has
a major component of graphite and would require more than half the
cosmic carbon. Duley's (1985a) analysis of e depletion relative
to Ar depletion suggests little residual depletion in the most diffuse regions, and e accretion in grain mantles in more dens. clouds.
DISCUSSION I
219
There are alternative compositions proposed for the mantle too. Duley favours amorphous carbon (or HAC) while Greenberg and collaborators prefer organic refractories ('yellow stuff') produced from icy
mantles exposed to UV radiation and cosmic rays (see e.g., Greenberg, van de Bult, and Allamandola 1984). Either mantle model can
be made consistent with the observed depletion. There might even be
some relationship between these forms, as nth generation organic refractories might be highly carbonized and HAC-like (Greenberg, later
in meeting). Laboratory measurements of the optical constants of
the 'yellow stuff' are needed for models of continuum extinction
and polarization in the visible. This is being worked on; it is not
very absorbing, and though the imaginary part of the refractive index does go up to the blue, it is less than 0.06 (Greenberg). Other
diagnostics, depending on the grain model, might be the optical polarization and the extinction and albedo in the far ultraviolet.
Spectral features in the extinction curve provide useful constraints too. If the 2200-A extinction bump were attributed to
graphite (§6), about 30~ of the cosmic C is required in small particles (see Greenberg). The absorption feature at 3.4 ~m provides
another example (§6). Duley and Williams (1983) suggest this arises
from HAC material; the observed strength of the feature would require a large fraction of the cosmic C to be in such form.
Finally, what are the requirements for PAH's? To explain the
near-infrared spectral emission features in reflection nebulae,
about 3~ of C in PAH's would be needed (see L6ger). Even more (10~)
is required in models of the widespread infrared cirrus detected
by the broad 12-~m band of lRAS (see Puget and Leene). These are
to be contrasted with the relative C content in the largest observed interstellar molecules (e.g., 10-9 in HCllN). If there were
that many large PAH molecules there would be a huge surface area,
which might result in too large a formation rate (R) for 82 (Duley). Tielens questioned this high rate, wondering if there was an
activation barrier. Tielens suggested that an absorption feature at
7.7 ~m in Wa3A might be from a PAR; d'Rendecourt stated it was the
CB. bending mode, but Tielens countered that the observed feature
was not sharp enough. The predicted 3.3-~m absorption (aromatic CR
stretch) from PAH's is too weak to be seen in the Galactic centre
spectrum (aee L6ger and §6).
6.
ORIGIN OF THE 2200-A EXTINCTION BUMP
This ia one of the few spectral features in the extinction curve,
and .v.r aince it was discov.red it has be.n attribut.d to amall
graphite particles. If the identification is corr.ct. 301 of the
cosmic C must b. in these graphite grains. Thia ia a large enough
fraction to be pro~d (in principl.) by sensitive studiea of d.pletion. i .•.• because the strength of the bump (relative to visual extinction. say) varies fr~m star to star. th.re is an opportunity to
DISCUSSION I
220
look for the expected correlation between bump strength and depletion. Millar (1979) actually found a weak anti-correlation, but because C might also be incorporated in grain mantles, the interpretation is ambiguous (Whittet 1984, Greenberg). This is worth pursuing
when better observations of C depletion are available.
Environments different than the Galactic interstellar medium
might provide clues about this feature. For example, in the Large
Magellanic Cloud (LMC) the metallicity is lower than in the Galaxy,
and C is more underabundant than 0; there the relative bump strength
is lower, while the far-ultraviolet extinction is larger (e.g.,
Clayton and Martin 1986). The same trend is continued in the SMC.
Circumstellar environments offer other possibilities. The Hdeficient C-rich (relative to a) R Cor Bor stars have been examined,
but must be interpreted with care (Tielens); they are too cool to
excite the near-infrared emission features of PAH's (L'ger).
HD 44179 (inside the Red Rectangle nebula) is C rich, and suffers lots of circumstellar extinction, but does not have a 2200-A
extinction bump. It does have prominent near-infrared emission features of the'type attributed to PAH's; the spectrum of coronene has
absorption bands shortward of 3000 A (see Jortner), and this star is
certainly hot enough to excite these emission features (L'ger). Duley (1986b) demonstrated that the continuum extinction, which rises
smoothly into the far ultraviolet, could be caused by amorphous carbon with a high fraction of tetrahedral (as opposed to trigonal)
bonding, and that similar HAC material absorbing the stellar ultraviolet radiation could produce the broad red luminescence feature in
the nebular spectrum.
IRC +10216, a cool luminous carbon star shrouded by dust, has
a featureless infrared spectrum, except for an 11-pm feature attributed to solid SiC (Allamandola). The optical depth in the near
infrared can suppress the contrast of any features (Martin). The
continuum emissivity of the dust seems more consistent with amorphous carbon than with graphite. Many C-bearing molecules, including long-chain hydrocarbons, are detected.
One got the impression at several points throughout the meeting that a graphite origin of this feature was falling into disfavour. Alternatives were presented by de Groot (irradiated aromatic molecules) and Duley (silicates), but a consensus was neither
sought nor reached. A critique is found in Draine (1984).
6.
THE 3.4-J.'m ABSORPTION FEATURE
There is an absorption feature at 3.4 J.'m in the spectrum of IRS7 in
the Galactic centre. This absorption is thought to arise in diffuse
cloud material along the line of sight, because there is no strong
ice band as might be seen in molecular cloud material (s.e the review by Allamandola 1084). Duley and Williams (1083) suggest that
this is a CH stretch in chemisorbed aliphatic CH2 or CHa groups on
HAC material. Laboratory spectra of HAC solids show this feature,
and little evidence for absorption at 3.3 pm by aromatic CR groups.
DISCUSSION 1
221
The strength of the observed astronomical feature would require a
large fraction of the cosmic C to be in such form, though the exact
amount depends on the adopted band strength. The alternative mantle material, the refractory organic residue produced on processing
ice mantles, should also have a high C (and H) content. Because of
C abundance constraints (§4) it might not be possible to have large
amounts of both HAC and organic refractories (as mentioned before,
there might be a close evolutionary relationship between the two, so
that they would not be distinct alternatives), and so it is interesting to examine the laboratory spectrum of such material. There
is a 3.4-~ absorption in the wing of the (usually stronger) waterice band in laboratory spectra (Greenberg), whose origin is controversial (Allamandola 1984); it could be due to saturated hydrocarbons (again CH2 and CHs groups) in the residue of a material initially containing a hydrocarbon like CH4. Other diagnostic spectral
differences between processed ices and HAC might derive from the
differing 0 content - which leads to C02, H2CO, etc. (d'Hendecourt
1984) - in the two materials.
The aromatic CH stretch is intrinsically weak (Allamandola).
Consequently, the 3.3-~m absorption predicted for the PAH's is too
weak to be seen in the Galactic centre spectrum (see L6ger), despite
the large abundances required for PAH's to explain infrared emission observations (see §4). These separate abundance arguments seem
to imply comparable amounts of HAC material and PAH material in the
interstellar medium. Again, is this a coincidence, or is there an
evolutionary relationship? Or can the infrared emission observations be reproduced with HAC material without resorting to PAH's at
all (see Duley)?
7.
INTERSTELLAR OR CIRCUMSTELLAR ORIGIN
It is important to assess formation and destruction channels for
grains and PAH's to understand their origin and equilibrium abundance. Dust is observed in the outflows from cool supergiants, either silicate dust or carbon dust, depending on whether the abundance of 0 exceeds C or not. Carbon is also ejected as CO. About
half the mass ejected into the interstellar medium is from carbon
(rich) stars with large C/O, and so it is not unreasonable to have
a few 10's of percent of C locked up in carbon solids (or PAH's)
in the injected material (Jura). The timescale for cycling interstellar material into stars and back again is about 109 y. However,
graina are 'destroyed' on a timescale 10 times shorter (~ 108 y),
and so grains must reform in the interstellar medium (Greenberg).
Their composition would then reflect the growth processes in the
cold gas. Tielens remarked that the destruction timescale depends
on our understanding of grains in shocks, and of shock recurrence
rates, both of which are rather uncertain. Furthermore, since we
do see 8i largely depleted in silicates, either allieate. are not
destroyed that rapidly, or must be re-formed in the interstellar
medium (is this possible?). In response to a question by Kroto,
222
DISCUSSION I
Martin mentioned that some evidence for shock destruction came from
a lowering of depletion in high velocity gas clouds, thought to have
been accelerated by shocks.
Jura wondered how PAH's survived shocks and cosmic ray hits.
Omont has found that they are not destroyed in low velocity shocks,
but that in strong shocks they don't survive (lifetime ~ 108 y).
Tielens added that if PAR's are charged then thermal sputtering
is significant. Duley stated that combustion might be a rapid destroyer of PAH's; Allamandola remarked that there was a big activation energy, and that even the energy available when a photon
strikes might not be enough; the combustion rate would not have to
be too high, since just one catastrophic event in '" 109 Y would have
significant implications for a theory involving a circumstellar origin. Since temperatures are high in shocks, the combustion reaction might be most important there (Duley). Omont agreed and added
that he did not believe that carbon stars could supply enough PAR's.
In interstellar shocks grain-grain collisions and sputtering might
produce PAH's while destroying grains (Jura). Given a very large
C+ + PAH reaction cross-section, L6ger questioned whether there
might also be interstellar growth of PAH's: although 0 has a larger
abundance, C is ionized and so is favoured. This is discussed in
the paper by Omont.
REFERENCES
Allamandola, L. J. 1984. In Galactic and Extragalactic Infrared Spectroscopy, eds. M. F. Kessler and J. P. Phillips, Dordrecht: Reidel, p. 5.
Cameron, A. G. W. 1982. In Essays in Nuclear Astrophysics, eds. C. A.
Barnes, D. D. Clayton and D. N. Schramm, Cambridge: Cambridge
Univ. Press, p. 23.
Clayton, G. C., and Martin, P. G. 1985. Ap. J., 288, 558.
d'Hendecourt, L. B. 1984. Ph. D. Thesis, Leiden University.
Draine, B. T. 1984. Ap. J. (Letters), 2'1'1, L71.
Duley, W. W. 1985a. Ap. J., 291, 296.
Duley, W. W. 1985b. M. N. R. A. S., 215, 259.
Duley, W. W., and Williams, D. A. 1983. M. N. R. A. S.• 205. 67P.
Duley, W. W., and Williams, D. A. 1984. Interstellar Chemistry. London: Academic Press.
Greenberg, J. M., van de Bult, C. E. P. M., and Allamandola. L. J.
1984. J. Phys. Chem.. 81, 4243.
Mathis, J. S., Rumpl. W'., and Nordsieck, K. H. 1977. Ap. J.. 211,
425.
Millar. T. J. 1979. M. N. R. A. S., 189, 509.
Whittet. D. C. B. 1984. M. N. R. A. S., 210, 279.
IDENTIFICATION OF PARs IN ASTRONOMICAL IR SPECTRA- IMPLICATIONS
A. Leger and L. d'Hendecourt
Groupe de Physique des Solides de l'E.N.S.
Universite Paris VII - Tour 23
2, place Jussieu
75251 PARIS CEDEX 05
FRANCE
ABSTRACT. We propose a consistent explanation of the so-called "unidentified" IR emission bands of the Interstellar Medium observed at 3.3 6.2 - 7.7 - 8.6 - 11.3 ~m. Following Sellgren (1984), we consider the
transient heating of Very Small Grains (~ 50 atoms) to a peak temperature of ~ 1000 K by the absorption of a single UV photon and estimate
the subsequent IR emission.
The necessary stability of these clusters against sublimation
suggests that they are graphitic and specifically Polycyclic Aromatic
Hydrocarbon (PAH) molecules. We estimate the emission spectrum of a
typical PAR, coronene: C24H12, and find a suggestive spectroscopic
agreement of the main spectral features with the astronomical ones. This
explanation of the observed emission bands can account for their excitation mechanism which was not hitherto explained.
New IR absorption spectra of large PAH molecules are presented with
the cross section of the different vibrational modes. The corresponding
emissions are calculated and it is shown that the agreement found for
coronene is not casual: a family of PAR molecules can explain the astronomical observations. We can even specify that compact PAHs give a
better agreement than non-compact ones.
Different implications of this identification are reviewed such as the
abundance of these molecules. It is found that they could contain
several percent of the cosmic carbon,which raises them to the level of
the most abundant organic molecules in the Interstellar Medium. We also
mention several open problems.
1. TOOLS FOR STUDYING THE INTERSTELLAR MATTER
The Interstellar Matter (1M) is made of 99% in mass by H and He gas and
1% by heavy element compounds. Although the latter are in a minority,
they are essential for the optical properties of the medium because
they can condense and make solids or large molecules which have optical
wide bands whereas atom ions or small molecules (H, H+, He, H2 •.. ) only
have narrow lines and cannot give a continuous absorption in a large
range of wavelengths.
223
A. Uger el al. (eds.), Polycyclic Aromatic Hydrocarbons turd Astrophysics, 223-254.
© 1987 by D. Reidel Publishing Company.
224
A. LEGER AND L. D'HENDECOURT
Basically, we have two ways of getting information on the 1M: the
absorption of the star light when it travels accross and the emission
of photons by the 1M itself when it is heated. Let us briefly review
the features which are observed and the information which is deduced.
1.1. Absorption
Star spectra exhibit very specific features that allow the identification of the star type even if its light has been somewhat modified when
travelling. We can then compare the spectra from two stars of the same
type, one behind a large column density of 1M, the other close to us
and deduce the specific extinction (absorption + scattering) by the 1M.
The corresponding extinction curve is analysed by M. Greenberg in
this Workshop, so let us only summarize its main features in the UVVisible and IR ranges:
- A general up slope from long to short wavelengths. Comparing it with
the scattering theory, a size is inferred for the solid particles present
in the 1M (a distribution from r < 50 A to 0.2, or 1 ~m, according to
the regions);
- A hudge and broad absorption feature at 2200
which has been tentatively attributed to graphitic material;
- More than 40 very well defined absorption bands in the Visible, called
"Diffuse Interstellar Bands". Their origin is completely unknown and
the corresponding information disregarded (see G. van der Zwet, in this
Workshop);
- Features in the Visible due to small molecules as Hz, CH, CH+, CN (see
M. Jura, in this Workshop);
- Several bands in the IR observed only in dense interstellar regions
(3.1 - 4.62 - 4.67 - 4.9 - 6.0 - 6.8 pm .•• ) which are attributed to
ices (H20, CO ••• );
- Two bands at 9.6 and 18 pm ubiquitously observed in both dense and
diffuse regions. They are attributed to silicates.
A,
1.2. Emission
The 1M is heated and emits resolved bands in the IR mainly in two cases:
- when in dust shells close to stars. The main observed feature is at
9.7 pm and is also attributed to silicate vibrational modes. This material is therefore observed both in absorption and in emission.
- when dust is irradiated by star light, even if the distance is large.
Several bands are observed. The main ones are at 3.28 - 6.2 - 7.7 8.6 - 11.3 pm. They are observed in a wide variety of astronomicalobjects: reflection nebulae(an interstellar cloud close to a star, Fig. 1
bottom), bipolar nebulae(matter probably ejected from a star, Fig. 8
middle), planetary nebulae (matter around a very hot star) and whole
regions of active galaxies (Fig. 1 top). Although the luminosity of
these objects varies by 7 orders of magnitude (IO~ to lOll solar luminosities) the positions of the features are always the same.
These last bands have been discovered since 1973 and were one of
the longest-standing puzzle in IR as~ronomy (see Allamartdola 1984,
Willner ]984). They were called "Unidentified IR Emission Bands".
225
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
2
,
6
8
10
12
,"""'--'-"T,--'-"'-"""',--'--'
-14.5 r.,.,"""'--'--'-,""""--'-"',"""'--'--'--
1
:3
~
M 82
-15 -
Ll
'"
A
!
(\
I~I~ (lJ.i ~,
V
.::~u -" l j \ ; '
\
~'-.. -, l ,"~l, "..,
,
{\~
V
,
I
'
I
'" ::
,
"I
L
1/1
-8 '-
~
~o
~~
-9
1.-".
•
'
L-' ',,:"-'"
'j
\J
I
8.2
,
7.7
B.8
,
11.3
,
I'm
-
,
10
Figure I. Mid-infrared spectra of: (I) the central part of the active
galaxy M82, adapted from Willner et al. (1977) and Gillett et al. (197~.
The Emission Bands at 3.3 - 6.2 - 7.7 - 8.6 -11.3 ~m are dominating
the emission of this region which luminosity is a large fraction of the
whole galaxy luminosity. The additional emissions at 4.05 - 7.0 - 12.8
~m are identified as ionized gas lines, H+, Ar+, Ne+ respectively.
(2) The reflection nebula NGC 2023 adapted from Sellgren et al. (1986).
For spectra (I) and (2), the continuous line between measured points
is only for clarity. The luminosity of the first object is over 10 6
times larger than that of the second, the Emission Bands are at the
same position, pointing to an universal process.
The aim of this paper is to show that Polycyclic Aromatic Hydrocarbons
(PARs) are likely their carriers. If this interpretat~on was correct,
we would have significantly progressed in our understanding of the
spectroscopic message from the IM.
226
A. LEGER AND L. D'HENDECOURT
2. ORIGIN OF THE EMISSION BANDS
Any explanation have to account for the following points:
- the five main bands always appear together;
- the regions where they are observed have Interstellar Matter and
strong UV-Visible irradiation;
- they emit a substancial fraction of the whole energy radiated by
these regions.
2.1. Equilibrium temperature of interstellar dust
Let us consider the case of a reflection nebula: a star of luminosity L
is illuminating Interstellar matter located at distance d. Typical
values are Ll = 310 3 Lo, where Lo is the sun luminosity, d 1 = 0.6 ly
'" 6 10 17 cm.
An interstellar dust grain mainly exchanges energy with its surroundings by absorption and emission of photons. If it is large enough,
its temperature fluctuations are negligible and its mean temperature is
determined by balancing its absorption and emission power: P b = P •
For a spherical grain with radius a:
a s
em
1Ta 2 Q(A UV ' a) 4>.1n
P
em
where QU, a) is the emissivity or absorptivity of the sphere at wavelength A, 4>in the star flux, a the Stephan constant. For typical interstellar gra1n material, one finds:
T
eq
(2.1 )
It is important to notice that such a rain emits in the Far-IR
iAmax '" 50 ~m) and definitely not in the range [3-12 ~m • For instance,
there is no way to obtain realistic parameters that would give a
significant emission at 3.3 ~m.
We conclude that dust at equilibrium temperature cannot explain the
observed Emission Bands in reflection nebulae.
2.2. Tentative explanations of the Emission Bands in the past
Before J984, several propositions have been made for explaining the
Bands:
- Allamandola et al (1979) suggested that simple molecules or
radicals (CH4, NH3, H20 ••• ) frozen in grain mantles could be excited and
emit with their vibrational modes. But the ubiquity of the observed
spectra would imply an unlikely uniformity in the composition of the
mantles.
In addition, the observed yield of conversion of UV photon to IR
photon would have implied a quantum yield about unity. Now, for an
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
227
excited radical in a solid, the non-radiative transfer of the energy
to the matrix is many orders of magnitude faster than the radiative
decay because vibrational modes have a long radiative life time. Most
of the UV energy is then transferred to the bulk of the grain that will
re-radiate it in the far IR because of its low equilibrium temperature,
not at the wavelengths of the observed bands.
- Duley and Williams (1981) showed that radicals bound at the
periphery of graphitic grains would produce very interesting bands.
Specifically, CH groups would have modes at 3.3 and 11.3 Mm, fitting
nicely two of the observed bands. Unfortunately, to produce the others,
they had to invoke groups such as NHz that would also produce unobserved
bands (2.9 Mm for instance). Moreover, the problem of the excitation
was still unsolved because of the competition between radiative and
non-radiative decays for a vibrationally excited radical bound to a
solid.
2.3. The idea of Emission during Temperature Fluctuations
Sellgren et al. (1983) measuring the IR spectrum of reflection nebula
noticed that in addition to the prominent emission bands, there was a
continuum that could not be explained by scattered light from the illuminating star (Fig. 2). This continuum could be fitted by a diluted
black body emission whose temperature (T ~ 1500 K) was far too high to
be an equilibrium temperature of dust in the nebula (see equation 2.1).
3,28pm
T
NGC 7023
3Q"W20"N
roomJy
Figure 2. The reflection nebula
NGC 7023 near Infra-Red spectrum adapted from Sellgren, 1984. The drawn
background is the emission of a diluted black body at 1500 K. This temperature is
much higher than the
equilibrium temperature of dust in
the nebula. The 3.28 pm emission band
is prominent in this spectrum.
rO~O~~~~2~D~~~3D~~~~4D~~~7.
>'(I'm)
In 1984, K. Sellgren suggested that such temperatures could be
achieved during the thermal impulse reached after the absorption of a
UV photon by a grain if it is small enough. The temperature increase,
!J.T, is given by:
A. LEGER AND L. D'HENDECOURT
228
if one approximates the N atoms solid specific heat by its high temperature limit, kB being the Boltzmann constant. With hvUV = 10 eV, one
gets ~T = 10 3 K for N = 50 atoms.
The temperature evolution of such a Very Small Grain located in
a typical reflection nebula is reported in Fig. 3a. The Very Small
Grain emits most of its energy when cooling from the high temperature
peaks. The color temperature of its emission is then high whereas its
mean temperature is low as dominated by the long periods between two
absorptions. This last temperature is similar to that of bigger grains
where the fluctuations are negligible (Fig. 3b and equation 2.1).
If they had a smooth emittivity, such Very Small Grains could
explain the nebula emission background but the Bands would still be
unexplained.
T(K)
10-
N1h
(0)
0
T'~:lt:::::=======-__..
(b)
0
Figure 3. (a) Temperature versus time for
a very small grain in a reflection nebula.
The spikes are due to the discrete absorptions of UV photon. The intensity of the
photon flux changes the interval between
two absorptions but not the peak value of
T. This value is determined by the photon
energy and the specific heat of the grain.
(b) Same curve for a big grain where
fluctuations are negligible.
t
2.4. Nature of these Very Small Grains
In 1984, Leger and Puget pointed out that in order to resist to sublimation, these grains must be refractory. They computed the erosion rate
for typical interstellar materials taking into account single and double
photon processes. They found that i~es and silicates are too volatile
whereas graphite can stand such high temperature excursions. Considering
the large cosmic abundance of carbon, they concluded that graphite-like
Very Small Grains are good candidates for the quantum heating by UV
photon.
Hydrogen is very abundant in the astronomical environment. In
addition, the carbon atoms at the periphery of the graphitic planes have
unsatisfied bonds. It is therefore very likely that hydrogen atoms are
bound to the peripheric carbon atoms as in organic molecules.
This gives a new argument to the suggestion of Duley and Williams
(198]) to account for the 3.3 and 11.3 ~m bands with CH modes with the
possibility of explaining the excitation mechanism because the whole
system is heated. However, a key feature to explain is the simultaneous
occurrence of the five Bands. With that aim, let us consider the whole
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
229
IR spectra of a hydrogenated graphitic cluster.
2.5. Optical properties of hydrogenated graphitic clusters
To calculate the absorption of a sphere of graphite surrounded by CH
groups, the first idea is to use bulk graphite optical constants and CH
oscillators. Such an electromagnetic calculation (Mie calculation) gives
a disappointing result: besides the 3.28 and 11.3 ~m modes, the spectrum
is dominated by a strong continuum due to the graphite; the other spectral features are minute and cannot account for the observed Bands.
However, the use of bulk optical constants for a small cluster may
be uncorrect. The optical absorption of a system in the ]-15 ~m range
is governed by its electronic and lattice excitations, so let us compare
them for bulk graphite and for a graphitic cluster.
The electron band structure of a graphitic plane is reported in
Fig. 4a. It exhibits a full band (rr) and an empty one (rr*) which are
~ 10 eV broad and touch at the Brillouin zone points (Dresselhaus and
Dresselhaus, 1981 and Joyes at this Workshop). Therefore, there is a
continuum of allowed transitions from 0 to 20 eV. The presence of these
interband electronic transitions in the IR is responsible for the
absorption continuum and the high IR dielectric constant of graphite
that screens the vibrational modes.
The electron band structure of a finite piece of graphite can be
crudely approximated by applying boundary conditions to the infinite
solid (see Ashcrof and Mermin, 1976). The electronic states are still
on the same curves but they are in a finite number and equally spaced
in k space (Fig. 4b). The interband transitions do not form a continuum
anymore, they are discrete and their lowest energy is of a few eV for a
50 atom clus ter.
The optical constants of a graphitic cluster are then completely
different from those of bulk graphite in the range of interest (3-15 ~m,
E
(<<1)
.<:
~
'II'
a
I
I
k
n:tJ
E
(bj
~
a
2'11'/L
(c)
_ ......._ _ _ _
h~
k
Figure 4. Band structure sketch
of : (a) bulk graphite, all the
values of k are allowed;
(b) a graphitic cluster, only
discrete values are allowed. L is
the linear size of the cluster;
(c) Corresponding optical absorption cross sections. For photon
energies less than hVmin, the
cluster exhibits no electronic
absorption whereas the bulk sample
does absorb.
A. LEGER ANO L. O'HENOECOURT
230
0.4-0.08 eV) because the cluster has no electronic transition at these
energies (Fig. 4c). The strong electronic absorption continuum vanishes
and the lattice modes appear as they are no more screened by electrons. So,
we expect more spectroscopic features in the absorption curve than
indicated by the first Mie calculation. But before going further, we
need a more precise model for our hydrogenated graphitic clusters.
2.6. Small grains or large molecules?
o
A 50 carbon atoms graphite "sphere" would have a radius of '\, 5 A and
could contain onll two layers of aromatic planes because the interplane
distance is 3.35 A (Fig. 5). As the grains are subjected to high temperature heatings, the planes would split because the interplane binding
a)
b)
HHHH
H
~
H
H
H
H
H
H
H
H
H
H
Figure 5. a) What a 50 atom graphite "sphere" would look like.
b) Proposed structure of hydrogenated graphitic cluster.
of carbon atoms is much weaker than the intraplane binding (0.06 eV and
7.5 eV per atom respectively). Such hydrogenated pieces of graphitic
plane are known in organic chemistry as large Polycyclic Aromat~c Hydrocarbon (PAR) molecules (Clar, 1964). Figure 6 shows some of them.
We suggest that the Interstellar Very Small Grains proposed by
K. Sellgren are large PAR molecules'. These species exhibit a very good
stability against heating and UV photolysis (Turro, 1978) and can stand
the expected hot events and direct photolysis in the Interstellar
Medium.
With this molecular model we can now estimate the emission spectrum of a cluster when it has absorbed an UV photon.
2.7. Emission spectrum of a PAH molecule: Coronene
The energy diagram of a closed shell isolated molecule having absorbed
an UV photon is reported in Figure 7 in the one electron approximation.
The absorbed photon makes a transition between the ground singlet state
So to an upper singlet state Si. The molecule being isolated cannot
231
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
exchange energy with its surroundings but by radiation.
co:o
co
naphlalene
lelracene
chrysene
perylene
Figure 6. Some PAH molecules.
Conventially, H atoms are not represented, they are at the periphery and
saturate the free carbon bonds.
caranene
benzo_coronene
ovalene
circo- biphenyl
internal
- ...-
I
converSion
-"'-""I,,~---+
"
• l IR
---')emi$$ion
--1
/ "
fl uoresc.
So
absorption
'"
t
)
.t T1
Figure 7. Energy level diagram of a large closed shell molecule. Si and
Ti are the singlet and triplet states of the molecule.
When there is no fluorescence de-excitation nor intersystem
crossing (= transfer to a triplet state Ti), the fast (t - 10- 12 _ 10- 8 s)
non-radiative processes transfer the energy to the vibrationally excited
electronic ground state So (Birks, 1970; Jortner, in this Workshop). The
energy exchange between vibrational modes (Internal Vibrational Randomization, Parmenter, 1982) is fast enough to reach a regime where their
232
A. LEGER AND L. D'HENDECOURT
population can be described by a vibrational temperature after
t = 10- 11 _ 10- 8 s. The validity of this approximation is due to the long
time (t ~ 10- 1 s before the first IR photon is emitted.
If the energy is transferred to a vibrationally excited triplet
state as suggested by Allamandola et al. (1985), there is also a thermalization of the vibrational modes in the triplet state. If the molecules are ionized, as it is likely in many astronomical objects, the
states are doublets and quartets and a similar thermalization of the
vibrational excitations is expected.
The IR emission of the molecule can be described as that of a
system at an equilibrium vibrational temperature because it has enough
time between two photon emissions to redistribute energy between its
vibrational levels. The hypothesis of thermal physics is fulfilled.
Besides fluorescence or phosphorescence, this IR photon emission
is the only energy release process for an isolated molecule. This point
is crucial as it overcomes the problem of non-radiative de-excitation
that Allamandola et al. (1979) were faced with.
The emission intensity IA of a physical object is, at wavelength
and equilibrium temperature T:
I
A
=
B (T) £emis(T)
A
A
(2.3)
where BA(T) is the black-body emission intensity and £~mis the emissivity of the object at that temperature.
The second thermodynamic law implies that the sample emissivity
is equal to its absortivity at the same temperature, AA(T), (Kirshhoff
law, see Reif, 1965). This quantity can be measured in an absorption
experiment at temperature T, where an incident intensity J o is transmitted as J A:
Jo - JA
J
(2.4)
o
where N (mol.cm- z ) is the column density of the molecule assembly and
0A (cm z mol-l) the cross section per molecule. In the optically thin
limit, NO A « 1, one reads:
Unfortunately, IR absorption spectra of large PARs are not available at high temperature. Room temperature of molecules as pyrene
(Cl6HlO), perylene (CZOHl2) and coronene (C24Hl2) have been published
and their comparison can give an idea of what can be expected for larger
molecules (Sadtler spectra, 1959).
Figure 8 shows the emission of heated coronene calculated from
relation (2.5), approximating the absorption at high temperature by
the spectrum at room temperature published by Sadtler (1959). Although
the interstellar PARs are likely a complex mixture of molecules, the
main emission bands of coronene already give a suggestive fit to the
observed features.
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
233
"
~o'"
N
I
E
u
3
LL'"
10'"
10HD44179
'::I....
N
IE
u
:3
2
3
4
5
6
7
8 9 10
12 14
A (", m)
Figure 8_ (I) Observed mid-IR spectrum of the central part of the active
galaxy M82, adapted from Willner et al. (1977) and Gillett et al. (1975).
(2) Observed spectrum of the bipolar nebula "the Red Rectangle" around
the star HD 44179 adapted from Russel et al. (1978). (3) Calculated
emission of coronene heated to 600 K using the absorption spectrum from
Sadtler Standard Spectra (1959). However, some of the bands being saturated in this absorption spectrum, the temperature and the intensity
of the peaks are not well determined. Better data are in Figure 12.
A. LEGER AND L. D'HENDECOURT
234
The 3.3 ~m (3030 cm- l )
band has already
been discussed and corresponds to an aromatic C-H stretch. The 5.2 ~m
(1920 cm- l ) peak of coronene seems to be weak or absent in astronomical
spectra. The 6. 2 ~m (I 61 0 cm- l ) absorption is "highly characteristic" of
C:C stretch in an aromatic ring (Bellamy, 1966, p. 69); its presence in
most astronomical spectra and the absence of 6.7 ~m (1500 cm- l ) absorption indicates that most aromatic rings are in a compact arrangement.
The 7.65 ~m (1310 cm- l ) feature of coronene is present in astronomical
spectra. It is not found in small aromatic compounds and may be specific
to larger species. The 8.85 ~m (1130 cm- l ) band is an in-plane aromatic
C-H bend whereas the 11.9 ~m (840 cm- l ) is an out-of-plane aromatic C-H
bend. The exact position of this last mode depends upon the number of
adjacent R which are on a ring (Fig. 9): solo (11.0-11.6 ~m, 910-860
cm- I ) , duo (11.6-12.5 ~m, 860-800 cm- l ) or trio (12.4-13.3 ~m, 810-750
cm- l ) (Clar et al., 1981). The observed spectra have a peak at 11.3 ~m
(885 cm- l ) and a broad structure up to 13.5 ~m (Tielens et al., de
Muizon et al., in this Workshop). This indicates the presence of different hydrogen sites with possibly a predominance of solo.
In conclusion, from the point of view of the analytic IR spectroscopy, the observation of the 3.3, 6.2 and 11.3 ~m bands is highly characteristic of PAH compounds.
solo -
X:11.0 -11.6I'm
duo -
11.6-12.5
trio -
12.4 -13.3
Figure 9. Different sites for
hydrogen atoms and the corresponding wavelengths of the out
of-plane bending mode yCH.
3. NEW IR LABORATORY DATA
3.1. More IR Absorption Spectra of large PARs
The absorption IR spectra of several large PARs have being recently
measured with emphasis on the intensity of the bands (Leger, d'Rendecourt,
Schmidt, 1986). Typical spectra are reported in Fig. 10 and the integrated intensities of the main bands are given in Table I.
235
DENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
tlX.
e. U
t ••
t.72
circum-diphenyl
ovalene
t .• e
~ I.&lt
~ to 4
Z
Z
""
'"~ to 2'
~
.'"
'"~ 1.2
~
'"
10 II
5 ••
'.2
2.5
5.8
9.2
'12.5
f"
?igure 10. IR spectra of two large PARs.
A. (\.lm)
1
Jnits for Ai
10- 2 5 cm 3
X
3.3
6.2
H- 1
c- 1
7.7 (±.3)
C- 1
8.8
11.3 ± 11.9
H- 1
H- 1
@
Coronene
C24 Hl2
1.2
.54
2.25
~
Dicoronene
0. BH2O
1.7
1.04
2.32
1.0
50 {IIO/H solo
34/H duo
Me thy 1 coronene
1.4
.90
2.32
1.6
38
1.5
.45
1.13
1.1
.59
2.0
1.6
33
Typical compact PAR
1.4
.70
2.0
1.2
42
f i/!O-G
I. 45
.21
(C- 1 )
.38
(C- 1 )
.18
(H- 1 )
3.4
(rl)
@
CH
3
@
®
.90
47
C25Hl~
Ovalene
C32Hl~
Circobiphenyl
C3BH20
(H- 1 )
.89
cr.
43 {160/H solo
23/H duo
Table I. Integrated absorption cross sections (~=
llAi) for different
compact PAHs as measured in the laboratory. The average values can be
used for astrophysical calculations on "typical" compact PARs. The resulting osci llator strengths, fi = I. 13 10 20 (Ai/1 cm 3 )(AJ I \.lID) -~ are also given.
236
A. LEGER AND L. D'HENDECOURT
It can be noticed that :
- The integrated cross sections of the different bands are not too
scattered from one compact species to another. The average value can
then be used for calculations in Astrophysics on typical large compact
PAHs.
- The cross sections per solo hydrogen (11.3 ~m) are quite larger
than per duo hydrogen (11.9 ~m) in the two measured species that exhibit
both modes. If this is confirmed with more molecules, it may be important to explain that the 11.3 pm band is dominating in the astronomical
spectra.
3.2. Calculated Emission Spectra: effect of Temperature
The emission spectrum (I A) of a molecule, at a temperature T, depends
on its absorption spectrum (OA) and on the temperature through the Planck
function BA(T) (relation 2.5).
The effect of T is illustrated in Fig. 11. The relative intensities
of the bands at 3.3 and 11.9 pm clearly depends on this emission temperature.
n"r
T= 600 K
."
1.,,,-
Figure 11. Effect of the temperature
(T) on the emission spectrum of a
PAH, circobiphenyl, calculated from
its absorption spectrum (Fig. 10).
T= 850 K
T = 1500 K
,;'
10
It
14
3.3. Calculated Emission for Compact PAHs
The, calculated emissions of several large compact PAHs are reported in
Fig. 12 and are compared with an astronomical observed spectrum. The fit
has only one adjustable parameter : the emission temperature. The simi"LaPity between the speatra indiaates that the "Unidentified" IR Emission
Bands aan be expLained by a mixture of aompaat PAHs. Coronene is therefore not the only "miracle" molecule that can account for the observations.
237
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
3.3
10
I
coronene
11. 3
6.2 7.7 8.6
I
@
!\
:.\ :A\
:
: i
V i.
:
•
I
t
\., ....../:.J
(900 K)
~
1
J
10
c
>-
L
c:
c i rcobi phenyl
~8~Kl
L
.D
L
0
'-.c
lL..
;<.
1
10
dicoronene
1'~Kl
1
10
:\
j \
.....
ovolene
(850 K)
@
rtlv~\
f
:
.::.r
............
1
2
4
6
8
A (.urn)
10
12
14
Figure 12. Emission spectra of several compact PARs calculated from
their absorption spectra measured in the laboratory at room temperature
(Leger et al., 1986) using relation (2.5). The observed spectrum of the
reflection nebula NGC 2023 is reported for comparison (dotted line).
The only free parameter in the calculation is the mean emission temperature (in brackets) which is adjusted to reproduce the observed ratio
of the two CH bands at 3.3 and 11.3 ~m. For clarity, the bands attributed to C-C modes are filled whearas only the contour of the C-H bands
is drawn.
238
A. LEGER AND L. D'HENDECOURT
Figure 13. Emission spectra of non-compact PAHs with same conditions
as in Fig. 12. DBP is diphenylbenzoperylene. The fit with astronomical
bands is poor.
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
239
The agreement for the 3.3 and 6.2 ~m bands is quite good with any
of the PAR spectra. This confirms that these bands are characteristic
of large PARs (see also Fig. 13) and supports their identification in
the Interstellar Medium.
A band is observed in all the calculated spectra at 5.2 ~m. This
region of the spectrum has not yet been spectroscopically well observed.
We strongly reaommend in the future to look for this feature in astronorrriaal objeats.
In the 6.5-8.5 ~m range, the main bands of the compact PARs appear
within the nebula 7.7 ~m broad feature. This region is important as it
permits to discriminate between different aromatic compounds. As already
mentioned, it seems possible to explain the observed broad band at 7.7
~m by a mixture of compact PARs.
The main spectroscopic disagreement still concerns the 11-13 ~m
region. It probably indicates a larger abundance of solo R atoms (11.3
~m) in the molecules present in Space than in those which have been
studied here and where most R atoms are duo (II.9 ~m).
3.4. Calculated Emission for Non-compact PARs
Similar emission spectra are reported in Fig. 13 for non-compact PARs.
A mode is present in these spectra, about 6.7 ~m (1500 cm- I ) that
is characteristic of protuberant aromatia rings, as opposed to rings in
compact arrangements. The latter have only a mode at 6.2 ~m (1600 cm- 1 )
(Fig. 12; Roussel, 1984). On the contrary, the astronomical spectra
exhibi t a
dip
in the range 6.5-7. 2 ~m when they have not the line
at 7.0 ~m which has a different origin (see also Fig. 8). This lack of
fit indicates that non-compact PARs, if present, are minute constituents
of the Interstellar Medium.
We conclude that tJe aan make the identifiaation of PABs in the as-
tronomiaal IR speatra more aaaurate: the dorrrinating speaies are large
aompaat moleaules as opposed to moleaules tJith protuberant ayales. This
conclusion agrees with the higher thermodynamic stability of the former
species (Clar, 1964).
Fig. 14 reports spectra of two PARs with non aromatic radicals.
The comparison with" astronomical bands indicates that the presence of
alkane groups is possible and can probably explain the observed bands
about 3.4 ~m (see de Muizon et al., in this Workshop). On the other
hand, large amount of C = 0 groups, for instance, would give structures
at 5.9 ~m (1700 em-I) which are not present in the observed spectra.
3.5. Emission Temperature
Equation 2.5 relates the integrated absorption in a band, Ai = ai~\i'
to its integrated emission intensity, Ei = I\i ~\i, using the Planck
function, B\(T). The mean emission temperature Tem can be deduced from
the ratio of two bands, of a same group, that occurs at different wavelengths. Considering the vC-R (3.3 ~m) and yC-R (11.3 ~m) modes gives:
E II . 3
AII . 3
BII . 3 (Tem)
(3. I)
A3 • 3
E3 • 3 = B3 • 3 (Tem)
A. LEGER AND L. D'HENDECOURT
240
Figure 14. Spectra of PAHs with
radicals in the Interstellar Medium.
The presence of methyl groups seems
possible but that of abundant aldehyde groups excluded.
2
6
8
m
~
~ ("Mm)
3.3
6.2
7.7
11.3
AAi/ A3.3].lm
I
1.0
2.9
30 (97)
EA/E 3 . 3 ].lm (NGC 2023)
I
4.6
9.8
4.2
b,c
(Red Rect.)
I
4.3
8. I
2.0
c
(M 82)
]
4.1
17.6
1.8
d
(NGC 2023)
12%
11%
(Red Rect. )
9%
8%
(M 82)
9%
3%
(NGC 2023)
T
em
780 K (II90 K)
A. (].lm)
1.
---
~
-
-
(Red Rect.)
-
(M 82)
a
980 K
1020 K
Table 2. Mean emission temperature (Tem) and hydrogen coverage (xH)
deduced from comparison between astronomical observations and laboratory
absorption of PARs. AAi are the integrated absorption cross sections for
a typical PAH (Table I). The values in brackets are deduced from data on
solo H in dicoronene and ovalene. EAi are the observed integrated band
intensities. (a) (c): Values are from Leger, d'Hendecourt and Schmidt (1986)
Cohen et al. (1986); (b) (d): Values are deduced from spectra by Sellgren
et a1. (I 986), Willner et a1. (I 977).
241
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECfRA
Table 2 gives the corresponding emission temperatures for different
astronomical objects using the absorption of a typical compact PAH as
reported in Table 1. The dependence of Tem upon the absorption cross
sections points out the importance of determining, in the future,
whether the value for solo H is larger than for duo H.
3.6. Dehydrogenation of PAHs in the Interstellar Medium
A quick look at spectra in Fig. 12 indicates that C=C modes are more
intense, relatively to C-H modes, in the spectrum of the nebula than in
the PAH spectra. This favors the idea of partial dehydrogenation of PAH~
in Space.
Quantitatively, the H coverage, xH = (H present)/(sites for H),
can be deduced from the intensities of Hand C bands if one assumes that
dehydrogenation does not ~pset the oscillator strength of the remaining
modes:
(3.2)
where T has already been determined from the ratio of two C-R modes.
This hydrogen coverage is reported in table 2 for three astronomical
objects. This estimate points to vey.y strong dehydrogenation of PAHs in
space (xH
~
10%).
Theoretical studies of PAH dehydrogenation by photons (Desert and
Leger, 1986; Tielens et al., in this Workshop) favors either no dehydrogenation (large molecules) or complete dehydrogenation (small ones).
We suggest that this value of hydrogen coverage mainly reflects the
fraction of hydrogenated species versus total number of species rather
than partial coverage of a given species.
4. IMPLICATIONS OF THIS IDENTIFICATION
Let us first determine some properties of, isolated PARs that are useful
for their study. We shall then derive different implications of their
identification as their size, their abundance in Space and the diffe~
rent fields in Astrophysics where they can playa role.
4.1. Specific heat of PARs
The specific heat of a PAH depends on: (a) Nt, its total number of
atoms, (b) NU/N C' its relative number of hydrogen and carbon atoms and
(c) the spec1fic molecule considered.
Taking into account points (a) and (b), the dependence on point (c)
is probably weak within a category of compounds (e.g. compact PARs).
The relative number of R versus C atoms is rather constant in the
serie : coronene (.50), ovalene (.44), circobiphenyl (.53). So, we
derive the specific heat for hydrogenated compact PARs from data on
coronene. For dehydrogenated molecules, we shall use the values for
graphitic planes, an infinite dehydrogenated PAH.
242
A. LEGER AND L. D'HENDECOURT
~
3
•
"Eo
III
4
5
l
<JLm}
, ,, , ,
6
10
•
..
20 50
.... 10
o 8
~ 6
..0
§
4
2
c 0
30 )()
2000
~
1000
la o
Figure 15. Energy distribution of all the vibrational modes of coronene
(optically active + inactive) from the force field model by Cyvin (1982)
and Cyvin et al. (1984). The arrows correspond to the most intense IR
absorption bands.
3
graphi Ie
~
N
I
.
2
-Z
"-
u 1
2000
T(K)
Figure 16. Specific heat of a compact PAR with Nt atoms. The curve
derived for coronene applies to hydrogenated species and that for graphite to dehydrogenated species. The crude approximation (4.2) is also
reported (d Q~ h cel line).
A free molecule with Nt atoms has 3Nt - 6 vibrational modes. If
the frequency Vi of each mode is known, its vibrational specific heat
is given by an Einstein multi-frequencies model:
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
243
c
(4. I)
where k8i = hVi. This approximation is in fact quite good as it depends
only on the validity of the harmonic one.
Fortunately, the frequencies of all the modes in coronene have been
calculated by Cyvin (1982) and Cyvin et al. (1984), using a force field
model for condensed aromatics. The quality of such a semi-empirical
model is indicated by the agreement with the measured active modes (IR
and Raman) which are much more numerous than the free parameters of the
model. Their resulting energy distribution is reported in Fig. IS and
the specific heat in Fig. 16.
We consider this curve as representative for the specific heat of
hydrogenated PAHs in Space. A crude approximation for it is:
=
C(T)
I
3(Nt-2) k (T/1300 K), for T
3(N t -2) k
, for T
<
1300 K
> 1300
K
(4.2)
For completely dehydrogenated PAHs 3 the curve for graphite in Fig.
16 should be used. It is derived from Krumhansl and Brooks (1953).
4.2. Cooling time for a PAH in Space
In the Interstellar Medium, the main cooling process for a hot PAH is
IR radiation besides fluorescence or phosphorescence in the Visible. Let
us evaluate this cooling time.
One can show, using equation (2.5), that the radiative power, P,
emitted by a molecule at temperature T, is : P = ! PAdA, with:
This temperature decreases according to:
P dt = -C(T) dT,
and a cooling time T can be defined as:
T '" C(T)T
P
The emitted power is:
P
= 41T
I
.
1
B(A.,T). A~ N
1
1
C
where B(A,T) is the Planck function, Ai D o.c ~Ai the integrated cross
section per carbon of the ith band and Nc tfie number of carbon atoms
in the molecule. Using values of Table 1 for Ai and of Fig. 16 for C(T),
with NH/NC - 0.5, one finds the molecule cooling time T:
244
A. LEGER AND L. O'HENDECOURT
T
1000 K
->-
T
3.8 s
T
1500 K
->-
T
2.4 s
(4.3)
It is independent of the molecule size as both the emitted power and the
specific heat are proportional to the number of atoms.
The cooling time should be compared to the radiative life time, to,
of an excited vibrational mode of the molecule:
me
1.50
in CGS units,
\/ f
v
where A is the Einstein emission coefficient and
is in em-I. The
oscillator strength, f, is related to the integrated cross section AA as:
= 1.13
f
10 20 (AA/I cm 3)(A/I ~m)-2
From Table I, a typical value for an aromatic C-H stretch
f = 1.45 10- 6 , leading to a radiative life time:
to
=
(3.3
~m)
is
(4.4)
0.12 s
The radiative life time of this mode is therefore significantly
shorter than the cooling time of the whole moleaule. This can be under-
stood as followed. The emission from the vC-H (3.3 ~m) modes of the
molecules is efficient (45% of the total power when T = 1500 K) but
these modes are 1/9 th of the total number of modes and, at T = 1500 K,
they are excited only in a fraction: n~.3/n~.3 ~ e- hV / kT = 5.6 10- 2 .
Each radiative emission from these modes takes out hv = 3000 cm-) from
the molecule. They occur at rate tol per excited mode. One finds :
to
hv
T .. kT 9a/3
e
-hv/kT
where a .. C(T)/C(oo) and /3(T) is the fraction of the total emitted power
that goes through the emission of the considered modes. For vC-H in
PARs :
t
o
/T ..
4.6 10- 2
which is small and is in agreement with (4.3) and (4.4).
The mean time between two emissions of IR photon for a molecule at
1000 K or between the UV absorption and the first IR emission is
tl ~ T <hv1R>/hV UV ' or :
~
tl
~
0.1 s
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
245
4.3. Size of Interstellar PAR Molecules
4.3.1. Evaluation from the mean emission color temperature. Considering
the situation in a reflection nebula, the illuminating staI has a temperature T* ~ 2000~ K and a spectrum with a maximum at 1500 A. PAR having
a strong absorpt~on around 2000-2200 A, we assume that the mean energy
of absorbed photon is:
<hv b >
a s
=6
eV (A
= 2100
A)
The comparison between laboratory data and the spectrum of NGC 2023
indicates a mean emission temperature Tem ~ 800 K (table 2). This value
is an average. The actual process is a heating up to T~eak by the absorption of the UV photon and subsequent emissions, at var~able temperatures,
during the cooling. Assuming that the average Tem is close to the temperature at which the molecule has lost half of its energy, one reads:
T
C(T)dT
1
= 2"
Io
Peak
C(T)dT,
and the approximation (4.2) for Tpeak < 1300 K gives:
Tem
~
0.71 T k
pea
(4.5)
The relation between the energy of absorbed UV photon and the peak
temperature,
hv
=
f:
peak C(T)dT,
gives, using (4.2):
hv
abs
Fo~
atoms:
~ (Nt - 2) k T2
the
pea
~e~ection
(4.6)
k/867 K
nebula
envi~nment~
one deduces a mean
numbe~
of
(4.7)
This derivation is an approximation for a more complete calculation
(Puget, Leger and Boulanger, 1985) that includes summations over:
(i) the incident photon spectrum, (ii) the molecule temperature during
cooling and (iii) the molecule size distribution.
It should be noted that the PAHs reported in Fig. 12 are relevant
for molecules in reflection nebulae as far as the number of atoms is
concerned.
246
A. LEGER AND L. D'HENDECOURT
4.3.2. Evaluation from Stability against Photo-thermodissociation. When
the energy hv of an UV photon has been absorbed by the molecule, the
probability for ejecting one atom can be calculated by the theory of
unimolecular processes (Desert and Leger, 1986). A preliminary result
for reflection nebulae is that for :
Nt > 25 atoms
(4.8)
the molecule is destroyed slower than it can be reconstructed by the
available mechanisms.
Conditions (4.7) and (4.8) indicate a size distribution for PARs,
in the reflection nebula environment, that starts above 25 atoms and has
a mean value of 50 atoms.
4.4. The most abundant known organic molecules in gas phase
The evaluation of the abundance of PAHs in the Interstellar Medium
results directly from the interpretation of its Near IR and Far IR emissions as illustrated in Fig. 17.
Figure 17. Sketch of the Interstellar
Matter. Grains and PAH molecules are
irradiated by the same incident stellar flux. Grains re-emit in Far IR and
PAHs in Near IR.
Grains and molecules absorb the incoming stellar flux according to
their respective abundance Ni and specific cross section 0i. As they
re-emit in separate domains of wavelength, their emission fluxes ~i can
be measured separately. One has:
~NIR
abs
<opAH> • NpAH
abs
<Ogr > • Ngr
247
IDENTIFICA TlON OF PAH's IN ASTRONOMICAL IR SPECTRA
The average absorption cross sections <oabs> can be estimated from
molecular and solid state physics in different radiation fields (UVVisible). Then, the relative abundance of PAHs results from the ratio
<PNIRI<PFIR •
The IRAS mission gives an opportunity for estimating this ratio in
many astronomical objects. Typical values are reported in Table 3 (see
also Puget et al., and Ghosh and Drapatz in this Workshop).
<PNIR
<PFIR
c
MpAH
MC
cosmic
SpAH
-S-graph
A2200
PAH
A2200
ra h
Reflection
Nebula
15%
3%
1.9
16%
Diffuse
Interstellar
Medium
30%
6%
3.8
33%
100%
20%
R. Cor. Austr.*
13
110%
Table 3. Ratios of Near IR to Far IR emissions of different objects and
deduced fraction of the cosmic carbon involved in PAHs. The relative
geometrical surface (S) of these molecules to that of graphite grains in
current models emphasizes their possible r6le in the Interstellar Mediu~
Their expected contribution to the 2200 A feature of the Extinction
Curve is also reported. A2 200 is the integrated cross section of the
species x around 2200
x
* A specific molecular cloud, see Leene (1986).
A.
Inspection of Table 3 indicates that:
PAHs are the most
abundant organic molecules known at that date in the IM. The abundance
(I) Involving several percent of the cosmic carbon,
of other detected interstellar molecules is reported in table 4 (Duley
and Williams, 1984). PAHs are in. third position, far above molecules
which are easily observed in radioastronomy. It is remarkable that such
abundant species have remained undetected for a long time. This is due
to their lack of simple signature in the radiowavelength domain which
used to be the royal, if not unique, way of detecting interstellar
molecules.
(2) PAR geometrical surface is comparable or larger than that of classical grains. This points to their possible role in interstellar chemistry
as catalyst (Omont, 1986).
(3) The contribution of PARs to the 2200 A bump of the Extinction Curve
is important. However, in most regions, these molecules alone seem not
to be able to explain the feature intensity.
A. LEGER AND L. D'HENDECOURT
248
Molecule
0.5
co
PAHs
Table 4. Abundance of different interstellar
molecules expressed in number of atoms
included versus total number of hydrogen
nuclei.
HCN, H2CO
4.5. Are PAHs observable in Absorption?
If PAHs are as abundant as we claim, a question immediately rises: why
do not we observe them in absorption ? Typical absorption cross sections
are:
3.3 ].lm,
o
2.4 10- 20 cm 2
Y C-H, I I. 9 ].lm,
o
o
1.2 10- 19 cm 2
\) C-H,
\)Si-O,
9.7 ].lm,
1.4 10- 16 cm 2
If a cloud, with optical depth in the Visual
Tv ~ Ay = 2.5, is in
front of a source its optical depth in the PAH bands w1ll be:
T = NCH 0, with NCH ~ 4 10- 2 Nc '
Nc = 4 10-4 NH and NH = 2 10 21 Av. Then:
T3 • 3 ].lm = 10- 2
T IL3 ].lm = 5 10
-2
These values are small and are quite difficult to observe. It is our
hope that better signal/noise in the future will permit their detection.
However, there is no conflict between the presence of PAHs at the level
we have inferred and the present IR observations in absorption.
It should be noticed that the easily detected silicate absorption
at 9.7 ].lm corresponds to the Si-O group that has an exceptionally high
specific cross section.
A feature around 3.4 ]Jm is observed towards the Galactic Center
(Butchard et al., 1986). It is attributed to alkane CH stretching modes
(Bellamy, 1966). The observation of a 3.4 ]Jm band in absorption and a
3.3 ]Jm band in emission is curious at first glance. However, it can be
explained if the saturated compounds are more abundant but incorporated
into grains, whereas the aromatic ones are less abundant but are free
flyer molecules. Such a situation would lead to a strong selection
effect for the emission: as already explained, the latter are the only
ones that can emit at short wavelengths.
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECfRA
249
5. CONCLUSION AND OPEN QUESTIONS
The identification of PARs in astronomical IR spectra as presented in
this paper was not obtained by looking for the best spectrum in a big IR
Atlas. The presence of large PAHs was inferred from discussing the possible composition of interstellar Very Small Grains and the spectroscopic
agreement was obtained afterwards. as a check.
The possibility of explaining at once the five main bands by a single family of molecules is also an argument in favour of this explanation because the bands are always observed to occur together.
However, it is not a one to one spectroscopic identification with
a given molecule but an identification with a family of molecules which
have characteristic bands. Some points are still not clear such as the
precise interpretation of the 11.3 ~m band.
The content of this molecular family has to be made more precise in
the future. Advances in that direction have been presented in the present
paper: compact PAHs give a better spectroscopical agreement than noncompact ones. Coronone is no more the only molecule that fits the observations.
In this explanation, the fact that the molecules are free flyers is
essential in order to give the high conversion yield from UV to Near IR.
It is needed to explain that 30% of the interstellar radiation is emitted
by this process.
This identification of an important component of the Interstellar
Medium. which was ignored before. has numerous implications on the phy-
sics of the medium (IR emission, Extinction Curve) and on its chemistry.
It also gives new ideas for explaining the mystery of the Diffuse Interstellar Bands in the Visual.
Several problems remain however, some of them are:
5.1. Ionization of PAHs
In regions as Reflection Nebulae most of the PAHs should be ionized
(Omont, 1986). What is the IR spectrum of ionized PAH is therefore an
important question. Would it give such good a fit to the observed bands
as the neutral PARs ?
The extracted electron of a PAR+ is one de localized TI electron out
of tens. One expects then only a small change in the vibration force
constants and in the frequencies. But is it as small as 1% which is the
aacuraay of the agreement on the 3.28 and 6.20 ~ bands? Laboratory
data are needed to answer that question.
5.2. Dehydrogenation of PAHs in Space
This point has to be investigated seriously. The initial arguments were:
(1) the observation of solo H modes whereas in the laboratory species
many H are in duo position ; (2) the comparison of C-H and C-C mode
intensities in the observed and in the calculated spectra points to a
strong deficiency in H ; (3) the binding energies of Hand C atoms to
the molecule are quite different (4.8 eV and> 7.5 eV) and allow dehydrogenation.
250
A. LEGER AND L. D'HENDECOURT
The study of photo-thermodissociation (Desert and Leger, 1986,
Tielens et al., in this Workshop) tends to invalidate argument (I) as
dehydrogenation seems to be either weak (large molecule) or almost
completed (small ones). The range of size for partially dehydrogenated
species is rather narrow and it seems difficult to explain the dominating presence of the solo H band this way.
A dehydrogenated PAH is a small fragment of graphite and there is
a severe need for laboratory data on IR spectra of carbon clusters specially with the new suggestions of spheroidal particles (Kroto,' in this
Workshop). Obtaining laboratory spectra of partially and totally dehydrogenated PAH would be a significative advance in the field.
ACKNOWLEDGEMENTS
The authors want to thank W. Schmidt, E. Clar, J.C. Roussel and
Setton for important contributions.
DISCUSSION
s.
LEACH: I) Did you calculate the shift of emission bands (profile
change) with respect to that of coronene which was studied at quite different a temperature ?
2) How many oscillators did you use in your calculation of IR
emission of coronene ? Population of vibrational levels after internal
conversion could induce new IR transitions in addition to those you have
considered.
3) If PAH are ionized once and with an odd number of electron, there is
no more triplet state. The ion fundamental state will be a doublet, Do,
and so for the first excited state DI. The first accessible quartet will
be higher in energy than D1 • Internal conversion processes will be different than for an even number of electrons PAR.
Answer: 1) and 2) We completely agree that it is the absorption at the
emission temperature (~ 1000 K) that should be used. The best approach
would probably be to measure it in the laboratory at that temperature.
In principle, this is possible and we are thinking of experiments for
that. Our feeling is that this should modify more the low energy modes
than the high energy ones, at least in absence of Fermi resonances.
3) We also agree that the electronic configuration of PAH+ is quite
different from that of PAHo. However, in our calculations of the IR
emission, we have only considered the internal conversion route that
leads to a vibrationnally excited state of the fundamental electronic
state (So or Do for an ion). This should be valid whatever is the ionization state of the species. The routes where the IR emission takes place
from electronic excited states are those that depend on the exact species
and its ionization.
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
251
w.w. DULEY: What is the role of the triplet state in storing energy
and in phosphorescence of these molecules ?
Answer: As mentioned above by S. Leach, we should probably speak of
states with multiplicity different from that of the fundamental because
PAHs in space are possibly ionized and without triplet state.
Such states could play a role if routes that arrive on them are
important. In that case, the IR emission occurs from a system with a
vibrational energy reduced from that it would have been in the ground
state.
The residual energy can be emitted as a phosphorescence photon in
the Visual. This is being proposed to explain the red emission of the
Red Rectangle nebula and could account for similar emissions in many
astronomical objects (d'Hendecourt and Leger, in this Workshop and to
be published).
A.G. TIELENS : Comments: Although we disagree with the details of
Leger's calculation'of the size of interstellar PAHs, the basic idea is
that the 11.3/3.3 ~m intensity ratio is very sensitive to the size of
the emitting molecules. We have calculated the infrared emission expected from a highly vibrationally excited PAH molecule using quantum
statistical theory (Allamandola, Tielens and Barker, 1985, Ap. J. Lett.
290, L25; 1986, in preparation). The results confirm that the 11.3/3.3
~m intensity ratio is indeed very sensitive to the size of the emitting
molecules as discussed by Leger. Comparing these calculations with the
observed interstellar 11.3/3.3 ~m ratio implies that interstellar PARs
have about 50 carbon atoms. This is very close to the result obtained by
Leger in his more simple minded calculatio"
Answer : I am pleased to see that "a simple minded" calculation gives
the same result as a complicated one. In addition, I raise objections
to your calculation. It applies to the specific case of neutral chrysene
for which you claimed that the main route of internal conversion leads
to the lower triplet state. I can see two difficulties:
(i) Chrysene is not a good example of interstellar PAHs because it has
strong modes in the 6.5-7.2 ~m region, in conflict with the observ~tions
(section 3.4).
(ii) As mentioned by S. Leach, in the bright regions where the presently
available spectra have been taken, PARs are ionized once and therefore
have no triplet state.
E. BUSSOLETTI :
I) Does the absence of a band (UIR) at 5.2 pm, which
is actually present in coronene, means the need of some sort of "PARs
soup" more than a simple class of molecules to justify part of the
observed UrR bands ?
2) Which are the physical grounds by which you do not believe to the
"complementarity" between PAR and amorphous carbon grains in the sense
that they may coexist and, even, PAR may be preliminary to amorphous
carbon grain formation !
252
A. LEGER AND L. D'HENDECOURT
Answer: 1) We have never said that coronene is responsible for the
UIR bands. We do think they come from a "PAR soup" which composition
has to be determined.
2) We do not reject the complementarity between PARs and amorphous carbon clusters specially as we conclude to the presence of dehydrogenated
PARs which are graphitic clusters. The species present in Space have
clearly less hydrogen than typical compact PARs.
But we think we should try in the future to be more specific and
determine what are the species we have in an experiment, what are their
structures and their specific IR spectra.
A. kMRCHAND : I am a little bothered by the fact that molecules much
smaller than 25 atoms are unstable under UV irradiation. How can larger
molecules be formed if smaller ones are destroyed ? I thought that small
molecules were growing into large ones through addition of new C atoms
or groups of C atoms: but such a process apparently is not possible.
Then, where do the PAHs come from ?
Answer: Our present picture is that, under irradiation, the addition
of new C atoms or ions can compete with the destruction processes only
for molecules above a minimum size. We suggest (Desert et al., 1986)
that the replenishment comes from the large part of the size distribution. Collisions in shocks between graphitic grains may be the clue for
the production of smaller species, the smallest ones disappearing and
the intermediate ones surviving.
However, in absence of hard UV irradiation and at higher gas density, as in envelopes of carbon stars, the growth from carbon atoms is
possible explaining the formation of graphitic carbon grains,and possibly PAHs,and their injection into the Interstellar Medium.
IDENTIFICATION OF PAH's IN ASTRONOMICAL IR SPECTRA
253
REFERENCES
- Allamandola L.J., Greenberg J.M. and Norman C.A., 1979: Astron. &
Astrophys. 11, 66
- Allamandola~.J., 1984: in Galactic and Extragalactic IR Spectroscopy,
eds Kessler M.F. and Phillips J.P., Reidel, Holland
- Allamandola L.J., Tielens A.G. and Barker J.R., 1985: Astroph. J.
Lett. 290, L25
- Ashcroft N.W. and Mermin N.D., 1976: Solid State Physics, ed. Holt,
Rinehart and Winston
- Bellamy L.J., 1966: IR spectra of Complex Molecules, Wiley
- Birks J.B., 1970: Photophysics of Aromatic Molecules, Wileylnterscience
- Butchard l., McFadzean A.D., Whittet D.C.B., Geballe T.R. and
Greenberg J.M., 1986: Astron. & Astrophys. Lett. 154, L5
- Clar E., 1964: Polycyclic Hydrocarbons, Academic Press
- Cohen M., et al., 1986: Astrophys. J., in press
- Combes M. et al., 1986: Nature 321, 266
- Cyvin S.J., 1982: J. of Mol. Struct. 79, 423
- Cyvin B.N., Brunvoll J. and Cyvin S.J., 1984: spectrosc. Let. 17 (9),
559
~ Desert F.X., Boulanger F., Leger A., Puget J.L. and Sellgren K., 1986:
Astron. & Astrophys., in press
- Desert F.X. and Leger A., 1986: in preparation
Dresselhaus M.S. and Dresselhaus G., 1981: Adv. in Phys. 30, 139
- Duley W.W. and Williams D.A., 1981: M.N.R.A.S. 196, 269
- Duley W.W. and Williams D.A., 1984: Interstellar Chemistry, Acad. Pres~
- Gillett F.C., Kleinmann D.E., Wright E.L. and Capps R.W., 1975: AstroPhys. J. Lett. ~, L65
- Huffman D.R., 1977: Adv. in Phys. 26, 129
- Krumhansl J. and Brooks H., 1953: J: Chern. Phys. 21, 1663
- Leene A., 1986: Astr. Astroph. _ill, 295
- Leger A. and Puget J.L., 1984: Astron. & Astrophys. Let. 137, L5
- Leger A., d'Hendecourt L. and Schmidt W., 1986: in preparation
- Omont A., 1986: Astron. & Astrophys., in press
- Puget J.L., Leger A. and Boulanger F., 1985: Astron. & Astrophys. Let.
142, Ll9
- Parmenter C.S., 1982: J. Phys. Chern. 86, 1735
- Reif F., 1965: Fundamentals of Statistical and Thermal Physics, Mc
Graw Hill
- Roussel J.C., 1984: private communication
- Russell R.W., Soifer B.T. and Willner S.P., 1978: Astroph. J. 220,
568
- Sadtler Standard Spectra, 1959, Midget edition
- Sellgren K., Werner M.W. and Dinerstein H.L., 1983: Astroph. J. Lett.
271, L13
254
A. LEGER AND L. D'HENDECOURT
- Sellgren K., 1984: Astroph. J. 277, 623
- Sellgren K., Allamandola L.J., Bregman J.D., Werner M.W. and Wooden
D.H., 1985: Ap. J. 299, 416
- Turro N.J., 1978: Modern MoZecuZar Photoahemistry, Menlo Park, Ca.:
Benjamin/Cummings
- Willner S.P., Soifer B.T., Russell R.W., Joyce R.R. and Gillett F.C.,
1977: Astroph. J. Lett. 217, L121
- Willner S.P., 1984: Same book as Allamandola, 1984.
THE IR EMISSION FEATURES:
CARBON PARTICLES
EMISSION FROM PAR MOLECULES AND AMORPHOUS
L. J. Allamandola and A. G. G. M. Tielens
Space Science Division, MS 245-6
NASA/Ames Research Center
Moffett Field, CA 94035
U. S .A.
J. R. Barker
Department of Atmospheric and Oceanic Science
Space Research Building
University of Michigan
Ann Arbor, MI 48109-2143
U.S.A.
ABSTRACT. PARs can have several forms in the interstellar medium. To
assess the importance of each requires the availability of a collection
of high quality, complete mid-IR interstellar emission spectra, a collection of laboratory spectra of PAR samples prepared under realistic
conditions and a firm understanding of the microscopic emission mechanism. Given what we currently know about PARs, the spectroscopic data
suggests that there are at least two components which contribute to the
interstellar emission spectrum: free molecule sized PARs producing the
narrow features and amorphous carbon particles (which are primarily made
up of an irregular "lattice" of PARs) contributing to the broad underlying components. An exact treatment of the IR fluorescence from highly
vibrationally excited large molecules shows that species containing
between 20 and 30 carbon atoms are responsible for the narrow features
although the spectra match more closely with the spectra of amorphous
carbon particles. Since little is known about the spectroscopic properties of free PARs and PAR clusters, much laboratory work is called
for in conjunction with an observational program which focuses on the
spatial characteristics of the spectra. In this way the distribution
and evolution of carbon from molecule to particle can be traced.
1.
INTRODUCTION
To assess the importance of PARs in astrophysics, one must determine the
amount of carbon tied up in these species and understand how they are
distributed among the various forms possible. For example, they can be
neutral or charged (both positively and negatively); and they can exist
as free species, loosely bound in clusters, or tightly bound in amor255
A. Uger el fII. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 255-271.
@ 1987 by D. Reidel PublUhing Company.
256
L. J. ALLAMANDOLA ET AL.
phous carbon
particles. To
4000
3000 2500 2000
1500
1000
gain this knowledge requires
NGC 7027
7.7!J.
11.3!J.
f
detailed infor6.2!J.
3.3!J.
,
10- 15
I ...
mation in three
r~.. 8.6!J.
I
•. ..I
f
:
""~.,•.
areas which is
not presently
"':of"
J
! \ !.I
,
available. The
....,
~ I ..
first is ·the
N=l.
/I ,f :f¥
I
requirement of a
E
collection of at
I.)
least 10 to 20
~
good quality,
U:<1(J"16
high signal-tonoise, moderate
,
spectral and
I
I
I I
I
I
B,),
[Arml [5 Illl [Ne nl
spatial resoluHell Pf,), 8Q
tion, complete
mid-IR. interstellar emission
2
3
4
5
6
7 8 9 10 12 14
spectra. The
second requirement is the
Figure 1. The 2-14 micron spectrum of NGC 7027
availability of a
shows the emission features superimposed on an
collection of
intense long wavelength "continuum". (Russell
laboratory specet a1., 1977).
tra of PARs and
PAR clusters
taken under conditions which mimic, as closely as possible, the
conditions which determine the properties of the interstellar material.
The third requirement is a firm understanding of the details of the
microscopic excitation-emission mechanism. Given the space limitation,
rather than discuss any of these points in detail we will present the
overall problem and we will summarize current knowledge and recent
results. The state of the observations and theories up to 1983 are
reviewed in Willner (1984) and Allamandola (1984).
/.J
(cm- 1 )
n
~
2.
. (
:"':'-\
~
tII I~
j II
THE SPECTROSCOPIC PROBLEM
2.1. Observational Spectroscopy
Good examples of nearly complete mid-IR spectra of objects which have
the emission bands are shown in figures 1,2, and 3 a,b. These figures
convincingly illustrate that the spectral variations from one object to
the other are significant. Although the 3.3/3.4, 6.2, "7.7", 8.6 and
11.3 micron bands are present in all of these spectra, important
differences are obvious. The 3.3 and 6.2 micron bands are remarkably
similar in all objects, having a full width at half-height that is
probably characteristic of non-radiative vibrational energy redistri-
257
THE IR EMISSION FEATURES
FREQUENCY (cm- 1)
3000
140
2800
HD44179
120
>~
>
~
iii 100
zw
~
z
80
60
(a)
3.2
3.3
3.4
3.5
3.6
WAVELENGTH (11m)
250
....
I
~ 200
N
I
E
u
~
.....
....
0 100
..<:
LL
0
5
(b)
7
9
WAVELENGTH (11m)
11
13
Figure 2. The 3-13 micron spectrum of HD 44179
shows the emission features superimposed on a
flat "continuum" (3-4 micron spectrum - Geballe
et al., 1985, 5-13 micron spectrum, Cohen et al.,
1986).
bution times
within individual
molecules
(Allamandola,
Tielens, and
Barker, 1985,
1986 - hereafter
referred to as
ATBa,b) rather
than overlap of
narrower lines
from different
PARs. There are
clear variations
however, in peak
wavelength and
profile of the
very broad
feature generally
referred to as
the 7.7 micron
band. In NGC
7027 and Orion
this peaks near
7.7 microns while
in HD 44179 the
maximum is near 8
microns. The
11.3 micron band
also shows some
variation, being
sharply peaked in
NGC 7027 and at
position 4 in
Orion but broader
in HD 44179.
Aitken and Roche
have spectroscopically studied
many objects in
the 8 to 13
micron region.
Perusal of their
spectra show
similar behavior
for both the 8.6
and 11.3 micron
features.
258
L. J. ALLAMANDOLA ET AL.
FREQUENCY (cm- 1)
30~3_1rOO__~3~~~__-=29~OO~____ :2=8~OO~
e
~
N
~ 20
ORION
EMISSION
;>
~
~
~
10
30r--.1~5TOO~-r~__~1~0~OO~~____~
E
~
!'Ie 20
ORION
EMISSION
u
~
::;:
~
10
IL'
00
~
CHRYSENE
-;
..!!!
(c)
....
J:
0W
-I
«
u
j:
0-
0
+
i
i
c
PYRENE
PYRENE
(e)
0
§
000
00
CORONENE
(g)
3.2
3.3
3.4
3.5
000
00
CORONENE
(h)
3.6
6
WAVELENGTH (pm)
7
8
9
10
11
12
Figure 3. The 3-13 micron emission spectrum from the Orion Bar compared
with the absorption spectra of the PARs chrysene, pyrene and coronene
suspended in KBr pellets. The "continuum" emission for the Orion Bar is
intermediate between that of NGC 7027 and ED 44179. (Orion, Bregman et
al., 1986; Chrysene, Cyvin et al., 1982a; Pyrene, Cyvin et al., 1979;
Coronene, Bakke et al., 1979, Cyvin et al., 1982b).
There is also evidence for the presence of broad emission plateaux
underneath these emission features. In the 3 micron region there is a
broad component extending from about 3.1 to 3.6 microns (Geballe et al.,
1985). An equally important plateau is the very broad, roughly
THE IR EMISSION FEATURES
259
triangularly shaped hump under the 6.2 and 7.7 micron bands which peaks
around 8 microns. The 11.3 micron band has also been shown to be part
of a broader plateau which extends from about 11 to 13.5 microns (see
Tielens et ale elsewhere in this volume and references therein) which is
prominent in NGC 7027, evident in Orion, but nearly absent in HD 44179.
The behavior of the 8 micron plateau parallels that of the 12 micron
plateau; it dominates the mid-IR emission from NGC 7027, seems to be
comparable in importance to the bands in Orion at position 4, and it is
absent in HD 44179. This broad component is also evident to a lesser
degree in the spectra of other objects; see for example the spectra of
the reflection nebulae NGC 7023 and NGC 2023 (Sellgren et al., 1985) and
several of the planetary nebulae spectra reported in Cohen et al.,
(1986). This behavior shows that there ar~ two different components
contributing to the mid-IR emission, one producing the bands and the
other producing the broader features.
Apart from the overall resemblance of the UIR band spectrum to the
spectrum expected from PAB-like species, (see Section 2 below, and paper
by Leger and references therein, elsewhere in this volume) the results
of recent, related observations tend to favor an aromatic origin as
well. Cohen et al., (1986) have shown that the fraction of total IR
luminosity radiated by the 7.7 micron feature in planetary nebulae is
strongly correlated with the nebular C/O ratio. Because the carriers
must be produced in these nebulae under harsh conditions, they must be
extremely stable and carbon rich, two characteristics completely
consistent with the aromatic hydrocarbon hypothesis. Cohen et ale also
show that, while there is variation among the UIR band intensities, they
are correlated, implying that a single class of chemical species is
responsible.
These are the highlights of what we feel are some of the most
important aspects of recent infrared observations concerning the PAB
problem.
2.2. Laboratory Spectroscopy and the Interstellar Spectra
Although the UIR band spectrum resembles what one might expect from a
mixture of PAHs, it does not match in details such as frequency, band
profile or relative intensities predicted from the absorption spectra of
any known PAB or their mixtures. In Figure 3, the emission spectrum
from position 4 in Orion is compared with a schematic version of the
absorption spectra of three PABs: chrysene (C 18H12 ), pyrene (C 16H10 ) and
coronene (C 24Hl2 ). Leger (this volume), shows a similar comparison
between the em ssion spectrum from the reflection nebula NGC 7023 and
the absorption spectra of several larger PABs suspended in KBr pellets.
Similar rather suggestive, but uncompelling, comparisons between the
interstellar emission spectra with the emission spectrum expected from
the PABs coronene and chrysene can be found in L~ger and Puget (1984,
hereafter LP) and ATBa,b. These emission spectra are discussed further
in the next section.
Because of the suggestive match, the assumption has been made that
PARs in some form or combination are the carriers of the interstellar
spectra. Only when detailed laboratory spectra for the various PABs
260
L. 1. ALLAMANDOLA ET AL.
become available can precise conclusions be drawn regarding their
respective importance. Keeping this qualification in mind, the
following general remarks concerning band assignments apply to virtually
all PAHs.
As illustrated in Figure 3, the 3.29 micron band is highly characteristic of an aromatic system (Duley and Williams, 1981; Bellamy,
1958), which shows a dominant band, corresponding to a C-H stretch, in
addition to a number of weaker bands between 3.1 and 3.6 microns which
are overtone and combination bands involving lower frequency fundamentals. (Bellamy, 1958, Cyvin, et al. 1982, Herzberg, 1968).
Figure 3 also shows the 6.2 micron emission feature (which
corresponds to a C-C stretching vibration in PARs) which is as
characteristic of polycyclic aromatic species as is the 3.3 micron band
(Leger and Puget, 1984; Bellamy, 1958).
Perusal of Figure 3 also shows that the 5 to 10 micron region in
PAR spectra is richest in IR active vibrations and the largest density
of bands occurs in the 7.2 to 8.5 micron range (ATBa,b, Bellamy, 1958).
Unlike the 3.3 and 6.2 micron bands which consistently occur at nearly
the same wavelength, independent of the molecule, the precise position
of these C-C stretching bands depends on the particular molecular
structure. Thus the infrared spectrum of a mixture of PARs could
produce a broad band, possibly with substructure, in this region. Of
course the precise peak position and profile would vary somewhat
depending on the particular PAR mixture responsible.
The small shoulder at 8.6 microns on the "7.7" micron feature which
often appears in the interstellar spectra is assigned to the in-plane
aromatic C-R bending mode in PARs (LP; Bellamy, 1958). As shown in
Figure 3, PARs show several bands close to this position.
The 11.3 micron feature is assigned to the out-of-plane C-H bending
vibration (Duley and Williams, 1981; Bellamy, 1958). Because this
frequency is so highly characteristic for aromatic species with edge
rings which contain only non-adjacent peripheral hydrogen atoms
(Bellamy, 1958), Duley and Williams postulated that the aromatic
containing material they believed responsible, amorphous carbon
particles, was only partially hydrogenated. Figure 3 shows that fully
hydrogenated PARs which contain more than one H atom per edge ring
possess several strong bands in the 11-15 micron range. The discovery
of the 11-13 micron interstellar emission plateau not only relieves some
of the difficulty associated with understanding partial hydrogenation in
exceedingly H rich environments, but also shows that edge rings of PARs
responsible for the interstellar emission can have non-adjacent as well
as 2 or 3 adjacent peripheral R atoms, but not 4 or 5. (Tielens et al.
and references therein, this volume, and Cohen, Tielens and Allamandola,
1985).
The spectra shown in Figure 3 c-h serve to illustrate several
additional points. For a free, highly symmetric PAR (e.g., coronene)
with an inversion center, the infrared spectrum will appear simple and
the Raman and IR active vibrational modes will be mutually exclusive due
to -the high molecular symmetry. However, if the vibrational force field
is not so symmetric (as is the case in free, less symmetric PARs, or
PARs in clusters or amorphous carbon particles) the IR and Raman spectra
261
THE IR EMISSION FEATURES
MICRONS.~m
10
B.3
7.14
I.) AUTO SOOT-RAMAN
6.25
5.55
Figure 4. Comparison of the 5 to 10
micron Raman spectrum of auto soot (a
form of amorphous carbon) with the
emission from Orion (soot spectrum,
adapted from Rosen and Novakov, 1984;
Orion, Bregman et al., 1984, 1986).
Taken frOID ATBa.
can be very similar. For example,
although coronene has 66 C-C modes
while chrysene has only 48, the IR
spectrum of chrysene is far richer in
the C-C stretching region (5-10 microns) because the molecule is less
....
symmetric. It is for these reasons
~ 10
that we legitimately could compare
the Raman spectrum of soot in the 6 to
8 micron range with emission from
Orion (Figure 4, ATBa), and thus pOint
1000
1200
1400
1600
1BOO
out the striking similarity between
WAVENUMBERS, cm- 1
the vibrational spectra of a mixture
of PARs (amorphous carbon is made up
of PAR subunits cross linked in an
irregular fashion, see, e.g., Oberlin, 1984, and Marchand, elsewhere i ..
this volume) and the interstellar emission bands (ATBa). For comparison, Figure 5 shows the IR absorption spectrum of a different mixture of
aromatic hydrocarbons, known as a char (Mortera and Low, 1983). Note
that the char spectrum also shows structure in the 3 micron region which
is similar to that shown in many of the interstellar emission spectra
(i.e., a dominant 3.3 micron band and a broader, weaker component
starting at about 3.15 microns, extending to about 3.65 microns, and
peaking near 3.41 microns). Figure 6 shows that spectra of individual
PARs also have similar structure. As mentioned above, the weak
absorptions by PARs in this region are due to overtones and combinations
of 5-10 micron fundamentals. The spectra shown in Figures 4,5 and 6 are
of PARs in various solids where the perturbations within the solid
broaden the individual bands causing them to overlap and produce a broad
component. Free PABs will show individual bands whose positions and
intensities are determined by the molecular structure of each PAR. The
interstellar emission component is due to the overlap of individual
emission bands which arise from different PARs. In addition to the weak
contribution from overtone and combination bands, highly vibrationally
excited molecules emit from vibrational levels greater than V=I.
Emission from these higher levels, shifted for anharmonicity,
contributes significantly to the 3.4 micron plateau and many of the
other components of the interstellar spectrum (see Section B, Barker et
al., 1986 and ATBa,b). The recent discovery of specific bands in this
plateau (de Muizon. this volume) supports this explanation of the
plateau in terms of overlapping individual lines.
As shown above in Figure~ 4 through 6, the infrared signature of
amorphous carbon soots and chars resembles the UIR band spectrum rather
L. J. ALLAMANDOLA ET AL.
262
WAVELENGTH (pm)
3.2 3.3 3.4 3.53.6
5
6
7 8 910
16
Figure 5. Infrared
absorption spectra
of chars at 400 and
480°C. Note the
similarity between
the 480°C char (a
form of amorphous
carbon) absorption
spectrum, the Raman
spectrum of soot
and the long wavelength continuum in
NGC 7027. (Char
spectra from
Mortera and Low,
1983) •
closely. Weak,
broad features at
3200
3000
2800
2000
1000
roughly 6.2, 7.7,
FREQUENCY (cm- 1 )
and 11. 3 microns
are also evident in
the extinction
curve of amorphous carbon particles (Koike, Hasegawa and Manabe, 1980,
Borghesi et al., 1983). Sakata et al., (1984) have also suggested that
the absorption spectrum of the hydrocarbon residue deposited by a methane plasma resembles the interstellar spectrum. The similarities and
differences between the interstellar IR emission spectrum and spectra
associated with various forms of amorphous carbon are discussed in
detail in the papers by Duley, Bussoletti et al., Goebel, and Roessler,
elsewhere in this volume. Since amorphous carbon is primarily. composed
of randomly oriented clusters of PARs cros's-linked and interconnected by
saturated and unsaturated hydrocarbon chains, the overall spectral emissivity of small amorphous carbon particles will resemble PAR spectra,
with the individual bands blurred out due to the solid state effects
which produce the line shifting, broadening and intensity changes discussed by Allamandola (1984). Thus, the infrared spectroscopic properties of small amorphous carbon particles will be largely determined by
the properties of the PARs of which they are made. As the particles get
larger, bands will overlap, producing broad features that possibly
retain some substructure indicative of the individual PARs. For still
larger particles, bulk propert1es dominate and the broad components will
appear as substructure on a strong continuum which follows a I/A law
producing the extinction curve reported by Koike et al. (1980) and
Borghesi et al. (1983).
In concluding this section on the spectroscopic aspect of the
problem it is important to stress two points:
1) Given what is currently known about the spectroscopic properties
of PARs and PAR-like species, free PAHs seem promising candidates to
account for the narrower emission components.
2) Having said this, there presently seems to be a better match
THE IR EMISSION FEATURES
263
A (,..lm)
3.2
3.4
3.6
3.8
CHRYSENE
between the observed IR emission bands
and laboratory amorphous carbon particle
spectra than with a mixture of the free
PAR spectra which we have available.
Thus, on the basis of spectroscopic
evidence alone, we would conclude that
amorphous carbon particles are
responsible. However, as described
below (and in the article by Leger, this
volume), a quantitative assessment of
the emission band intensity, in
particular that at 3 microns, forces us
to conclude that the narrow bands are
due to IR fluorescence from molecule
sized species. Although this point had
been previously raised by a few
researchers (see review by Allamandola,
1984) the importance of this aspect of
the emission problem was largely
unrecognized until quite recently.
3.
CORONENE
3200
3000
2800
A (cm- 1)
2600
THE EMISSION MECHANISM
Recently attention was redirected toward
the importance of the microscopic
details of the emission mechanism.
Sellgren, Werner and Dinerstein (1983)
discovered the 3.3 micron emission band
in reflection nebulae along with a
continuum from about 1.25 to 5 microns
with a constant color temperature,
independent of position in the nebula.
The color temperature of the emission
was found to be relatively constant
across the nebulae even though the
pumping photon flux decreased, thus
eliminating the possibility of a pumping
mechanism in thermal equilibrium.
Assuming the emission originated from
particles, Sellgren (1984) proposed that
thermal fluctuations in very small (lOA)
grains, which were caused by absorption
of individual UV photons, could account
for the observations. She also noted
that the species responsible must be
very stable, as two photons would
Figure 6. Infrared absorption spectra in the 3.3 micron region of the
PARs chrysene, pyrene and coronene, suspended in KBr. Spectra courtesy
of Drs. Cyvin and Klaeboe, University of Trondheim, Norway.
264
L. J. ALLAMANDOLA ET AL.
occasionally be absorbed by these species raising their temperature to
over 2000 o K. (The two photon event timescale is only about 100 7years,
while the lifetime of a reflection nebula is on the order of 10 years.
If two photon events could destroy the carrier, it would deplete the
material in the emission zone.) Although previous results implied the
emission is molecular in origin, (in particular the intensity at 3
microns relative to that of the longer wavelength bands,) the observation of a constant color temperature across the nebulae unambiquiously
pointed out that the emission process is not at thermal equilibrium.
The importance of this non thermal-equilibrium, fluorescence process in
determining the radiation balance of the interstellar medium also seems
responsible for the IR cirrus (discovered by lRAS) and for the very
extended 10 micron excess emission associated with reflection nebulae
(discovered by Castellaz et al., 1986).
Our philosophy in developing a model to describe the emission process was to do as thorough a calculation as possible on a well understood PAR (chrysene, CI8 RI2 ), for which most of the intramolecular
relaxation processes were well known and which had well characterized
vibrational modes. Although the intersystem crossing and internal
conversion rates available for chrysene (Birks 1970) were measured in
condensed phases and may be different for free gaseous species, the
model calculation involved a minimum of speculation.
Figure 7 shows a highly schematic version of the overall excitation
- emission process. (See the articles by Jortner and Mukamel elsewhere
in this volume for a detailed discussion of these processes.) Prior to
the absorption of a photon, any neutral PAR is in the lowest singlet
electronic state, So. Photons of many energies are incident on the
molecule, and of those absorbed, some can ionize the molecule and others
can excite it to higher singlet states (Sl,S2' ••• ). The ~~~orbed energy
will quickly redistribute itself within tne molecule «10
s) via
internal conversion (IC) (Sf+-Si) and intersystem crossing (ISC)
(Tf+-S i ) processes. For chrysene, the strongest absorption corresponds
to the S3+-So) transition which peaks near 2675A. Nearly 90% of these
excited molecules undergo rapid ISC to high vibrational levels of the
T state. Because T is lower than SI (this is true for virtually all
ngutral PARs) and thg direct radiative transition (phosphorescence) from
T to S occurs very slowly (on the order of several seconds) the energy
o
0
is trapped in the lowest triplet state, TO~1 Thus, a few nanoseconds
af~rr the absorption of a 2675A (37,400 cm
) photon, about 17400
cm
is trapped in the triplet state as vibrational energy. The
remainder is "stored" by the molecule which is now in the triplet
electronic state. This highly vibrationally excited molecule can lose
its vibrational energy only by radiating IR photons, primarily by the
transitions (Vn-1+-V n ) at the fundamental vibrational frequencies of the
T state. This process is known as infrared fluorescence (IRF). IRF
l~fetimes are on the order of a tenth of a second while the phosphorescent lifetime (S +- T ) for chrysene is longer than three seconds.
Thus, IRF is the major dgactivation route, although a small fraction of
the excited population will relax via phosphorescent emission of red and
near IR photons.
If the excited PAR is ionized (as we believe to be the case in the
265
THE IR EMISSION FEATURES
V3
- - - - V2
----V,
S2
INTERSYSTEM
CROSSING
~~~~(~~-....
V,
_+-+-____
'<......_
S,
UV
PHOTON
ABSORBED
_______ V3
QIR
PHOTONS
- - - - - V2
EMITTED
--- V,
To
RED PHOTONS
EMITTED
- - _ J J V3
- - - - V2
----V,
So
UV PUMPED IR FLUORESCENCE
PHOTOPHYSICS FOR NEUTRALS
Figure 7. Schematic energy level diagram for a neutral PAR, showing the
various radiative and nonradiative excitation and relaxation channels
possible. Taken from ATBb.
L. J. ALLAMANDOLA ET AL.
266
FREQUENCY (cm- 1)
45 4000
2000
1500
VIBRATIONAL
ENERGY CONTENT
35
.. 30
:;
:rl
030,000 cm- 1
020,000 cm- 1
• 10,000 cm- 1
1
750
~
CHRYSENE
~ 25
~
~ 20
w
C!
15
10
5
WAVELENGTH (/.Lm)
Figure 8. The IR fluorescence spectrum from chrysene as a function of
vibrational energy content. Taken from ATBb.
emitting zone), the energy level diagram will be quite different,
involving doublet (D) and quartet (Q) states rather than singlets and
triplets, with the ground electronic state being a doublet, Do. "In the
case of an ion, the lowest excited doublet state, Dl , is always lower
than Q and trapping in the Q state, which cannot radiatively relax to
the gr8und state, D , is not ~mportant. (see the articles by Leach, this
volume, for a compl~te description of this situation.) Thus, for free
molecular ions nearly all of the energy is quickly converted to
vibrational energy in the ground electronic state, Do' where IRF is the
only deactivation route.
The infrared emission spectrum predicted for chrysene is plotted in
Figure 8 as a function of vibrational energy content. From Figures 3
and 8 several conclusions may be reached. Firstly, the IRF spectrum
varies dramatically as the vibrational energy is changed and it does not
necessarily match the intensity distribution observed in absorption.
Se£?ndly, no significant IRF emission is observed from the 3000
cm
modes (3.3 microns), unless the molecule contains a significant
amoudt of vibrational energy per vibrational mode. Thirdly, the
in!insities of the 3000 cm
modes (3.3 microns) relative to the 1000
cm
modes (10 microns) are directly related to the internal energy of
the molecule.
THE IR EMISSION FEATURES
267
One of the many important consequences of this treatment is that
the intensity ratio of the 11.3 to 3.3 micron bands provides a measure
of the vibrational energy content of the emitting species if its size is
known; conversely, if the energy content of the molecules is known, the
size of the emitting species can be estimated. Thus, if one knows the
spectrum of the exciting UV field, the 11.3/3.3 intensity ratio indicates the number of carbon atoms in the smallest emitting species (the
most intense emitters). Conversely, if one can confine the size of the
most intense emitting PAR on the basis of spectroscopic constraints, one
can determine the part of the incident radiation field that is most
important in pumping the IR bands.
To make this estimate note that for each C-H bond one out-of-plane
bending mode and one stretchi~¥ mode will be present; thus, the number
of modes emitting nea!1885 cm
(11.3 micron) equals the number of modes
emitting near 3000 cm
(3.3 microns), regardless of the actual number
of C-H bonds present in each molecule. Vibrational assignments for
benzene (C 6H6 ), azulene (C 10 H8 ), anthracene (C I4 H10 )' chrysene (C 18 HI2 ),
and perylene (C 20 HI2 ) were used to predict the re ative IRF emiss on
intensities due to one C-H stretch mode and due to one C-H out-of-plane
mode, assuming they have equal integrated absorption coefficients, as is
approximately true in the gas for free molecular benzene (Bishop and
Cheung, 1982), napthalene and anthracene (Niki, 1986).
As an initially vibrationally excited molecule relaxes by infrared
emission, it must sequentially emit many infrared photons and undergo an
energy cascade before it reaches equilibrium with the low temperatures
of the ISM. Thus, the observed interstellar emission bands are due to
emission from molecules in all stages of relaxation subsequent to excitation to some initial energy. For comparison with the emission bands,
the calculated decay rates of IRF emission were used to average the
calculated intensities over the entire range of energy up to the initial
excitation level. Ratios of IRF intensities (averaged over the cascade)
were calculated for several initial excitation energies. The calculated
ratios of IRF intensities are plotted in Figure 9 as a function of the
number of carbon atoms in each molecule for several values of initial
internal energy.
The observed 11.3/3.3 intensity ratio in Orion is near unity and
the ratios observed in other objects range up to about 4 or 5 (Cohen et
al., 1986). If_fhe average emitting species was initially excited to
about 80,000 cm
(1250A) and all of that energy was converted to
vibrational energy, it must have 20 to 80 carbon atoms to explain the
observed intensity ratios ranging from 1 to 5. At lower excitation
energies, even smaller species can explain the observed intensity
ratios, but the maximum size is limited by thf maximum energy of the
photons available "(about 80,000 - 100,000 cm ).
Visual inspection of published IR absorption spectra for PARs
suspended in KBr shows that the integrated absorption strength of the
bands in the 11-13 micron range is generally about three to five times
more intense than the bands at 3.3 microns. This differs from the near
unity values (used in the calculation) observed in the gas phase for
benzene, napthalene and anthracene (Niki, 1986). However, if the 11-13
micron modes in the larger emitting PARs consisting of several fused
L. J. ALLAMANDOLA ET AL.
268
...
I
E
(.)
o
In
o
~
...:::::
I
"11.3/3.3" INTENSITY RATIO
VIBRATIONAL ENERGY CONTENT
40,000 cm- 1
60,000 cm- 1
80,000 cm- 1
100,000 cm- 1
E
(.)
In
CO
~
10
CARBON ATOM NUMBER
100
Figure 9. The 11.3 micron/3.3 micron intensity ratio plotted as a
function of carbon atom number and vibrational energy content. The
relative fluorescence intensities includes integration over the
vibrational cascade. Taken from ATBb.
rings has an integrated absorption coefficient three times larger than
that of the 3.3 micron mode, the maximum molecular size consistent with
the observed intensity ratio range is reduced by a factor of two to
three, indicating that the band carriers are species containing about 20
to 30 carbon atoms. This conclusion is in agreement with the number deduced from individual band profiles analyzed by accounting for emission
from higher vibrational levels, which are shifted by anharmonicity.
Space limitations preclude more than a passing reference to this
important point. A detailed discussion is presented in Barker et al.,
1986 and ATBb.
In concluding this section on the detailed emission mechanism, we
wish to point out that while one can model the emis~ion phenomenon using
a thermal model approximation, such approximations are of severely limited validity (ATBa,b; Barker, Allamandola and Tielens, 1986). We
stress that the molecular approach permits one to address questions
regarding photostability, hydrogen coverage, reactivity, photoisomerization and deuterium fractionati~n in a completely self-consistent,
non-approximate manner in addition to being able to explain the observed
emission ratio and account for band profile behavior.
4.
THE MULTICOMPONENT NATURE OF THE EMISSION SPECTRUM.
A "dilemma" is presented by 1) the spectroscopic case which suggests
amorphous carbon particles in addition to free PARs as the carrier of
the emission, and 2) the energetics which point to molecule sized
species containing about 20 carbon atoms. This dilemma is resolved by
THE IR EMISSION FEATURES
269
realizing that in nearly all cases the aperture sizes used to measure
the interstellar spectra encompass a large fraction of the entire
emission object. Thus, in many cases, we are collecting photons
originating from a zone which includes cold, dense, neutral material as
well as highly excited, ionized material close to the pumping star.
The high excitation planetary nebula NGC 7027, which occupies only
a few arcseconds of the sky, typifies this situation. Because it is a
very intense infrare4"pbject, several groups have mapped this object at
different wavelengths. On the basis of broad band maps Bentley (1982)
concludes that two separate components are responsible for the IR
emission. One which carried the features, the other responsible for
the continuum as measured at 10 microns. Aitken and Roche (1983) have
carried out a very thorough 8-13 micron spectral mapping program of
this source which is very informative. Their spectra show interesting
spectral variations in the features, not only in going from the ionized
to the neutral zone, but also within each zone as well. Most recently,
Goebel (1986, and this volume) has shown that there is a remarkable
resemblance between the spectrum of hydrogenated amorphous carbon (HAC)
and the underlying continuum (not the features) in NGC 7027 and suggested that HAC particles, warmed by the incident photons, were
primarily responsible for the broad underlying continuum. High spatial
and spectral resolution IR maps of NGC 7027 shows that the emission at
8.2 microns (presumed to be a tracer of the "7.7" micron feature) peaks
in a very different location within the nebula than the continuum
emission (due to the broad component) as measured at .10 microns
(Goebel, 1986). Taken together these results show that emission from
both amorphous carbon like particles and free PAHs contribute to the
overall infrared emission from NGC 7027 (and presumably in other
objects which have sufficient flux to heat small particles).
Our principle conclusion is that progress in this field requires
studies which focus on the spatial characteristics of the spectra from
these components. It is only in this way that the distribution and
evolution of carbon from molecule to particle can be traced. Thus, at
this stage of the interstellar IR emission band problem, spectroscopic
studies on extended objects are called for. Under the right conditions
(where confusion from particle emission can be eliminated) it may be
possible to limit the particular PAHs responsible for the features to a
handful.
ACKNOWLEDGEMENTS: We are indebted to Professor Cyvin of Trondheim
University, Norway for kindly sending us original spectra and pointing
out some important spectroscopic properties of condensed aromatics. J.
R. Barker acknowledges partial support from the U. S. Department of
Energy, Office of Basic Energy Sciences.
5.
REFERENCE S
Aitken, D. K. and Roche, P. F. (1983), Mon. Not. R. Ast., Soc. 202,
1233
Aitken, D. K. and Roche, P. F. (1982), Mon. Not. R. Astr. Soc. 200, 217
Aitken, D. K., Roche, P. F., Spenser, P. M. and Jones, B. (1979) Ap. J.
270
L. J. ALLAMANDOLA ET AL.
233, 925.
Allamandola, L. J. (1984), "Absorption and Emission Characteristics of
Interstellar Dust" in Galactic and Extragalactic Infrared
Spectroscopy, eds. Kessler, M. F. and Philips, J. P. (D. Reidel
Publishing Co., Dordrecht), 5.
Allamandola, L. J., Tielens, A. G. G. M. and Barker, J. R. (1985), Ap.
J, Letters, 290 L25 (ATBa).
Allamandola, L. J~Tielens, A. G. G. M. and Barker, J. R. (1986), Ap.
J., submitted. (ATBb).
Barker, J. R., Allamandola, L. J., and Tielens, A. G. G. M. (1986), Ap.
J. (Letters) submitted.
Bakke, A., Cyvin, B. N., Whitmer, J. C., Cyvin, S. J., Gustavsen, J.
E., and Klaeboe, P. (1979), Z. Naturforsch 34a, 579.
Bellamy, L. J. 1958, The Infrared Spectra of Comp!;x Organic Molecules,
(John Wiley and Sons, 2nd ed. New York)
Bentley, A. F. (1982) Astron. J., 87, 1810.
Birks, J. B. (1970), Photophysics Of Aromatic Molecules (London:
Wiley-Interscience)
Bishop, D. M. and Cheung, L. M. (1982), J. Phys. Chem. Ref. Data, ll,
119.
Borghesi, A., Bussoletti, E., Colangeli, L., Minafra, A. and Rubini,
F., (1983), Infrared Physics 23, 321.
Bregman, J., Allamandola, L. J., Simpson, J, Tielens, A, and Witteborn,
F. (1984) NASA/ASP Symposium, Airborne Astronomy, NASA/Ames
Research Center (NASA CP 2353).
Bregman, J. et al., (1986) in preparation.
Castellaz, M., Sellgren, K. and Werner, M., (1986), Ap. J. (Letters)
submitted.
Cohen, M., Tielens, A. G. G. M., and Allamandola, L. J. (1985) Ap. J.
(Letters), 299, L93.
Cohen, M., Allamandola, L. J., Tielens, A. G. G. M., Bregman, J.,
Simpson, J. P., Witteborn, F. C., Wooden, D. and Rank, D. (1985)
Ap. J., in press.
Cyvin, S. J., Cyvin, B. N., Brunvoll, J., Whitmer, J. C" Klaeboe, P.,
and Gustavsen, J. E. (1979), Z. Naturforsch, 34a, 876.
Cyvin, S. J., Cyvin, B. N., Brunvoll J., Whitmer, ~C., and Klaeboe,
P. (1982b), Z. Naturforsch, 379, 1359.
Cyvin, B. N., Klaeboe, P., Whitme~J. C. and Cyvin, S. J. (1982a), Z.
Naturforsch, 37a, 251.
Duley, W. W. and Williams, D. A. (1981) Mon. Not. R. Astro. Soc., 196,
269.
Geballe, T. R., Lacy, J. H., Persson, S. E., McGregor, P. J., and
Soifer, B. T. (1985), Ap. J. 292, 500.
Goebel, J. (1986), B. A. A. S., l7:-Abstract 52.06, 908.
Herzberg, G. H., (1968), Infrare~and Raman Spectra of Polyatomic
Molecules, (D. van Nostrand Co., Princeton).
Koike, C., Hasegawa, H., and Manabe, A., (1980), Astrophys. Space Sci.
67, 495
L~ger~A., and Puget, J. L. (1984), Astro. Ap. 137, L5. (LP).
Mortera, C., and Low, M. J. D. (1983), Carbon, 21, 283.
Niki, H. (1986), (private communication).
THE IR EMISSION FEATURES
271
Oberlin, A. (1984), Carbon 22, 521.
Robin, M. (1975), Higher ExCited States of Polyatomic Molecules II
(Academic Press, N.Y.)
Rosen, H., and Novakov, T., (1978), Atmospheric Environment 12, 923.
Russell, R.W., Soifer, B. T. and Willner, S. P. (1977) Ap. J.-rLetters)
217,1149.
Sakata:-A., Wada, S., Tanabe, T., and Onaka, T. (1984), Ap.
J.(Letters), 287, L51.
Sellgren, K. (1984) Ap. J., 277, 623.
Sellgren, K., Allamandola, L~., Bregman, J. D., Werner, M. W. and
Wooden, D. H., (1985) Ap. J. 299, 416.
Sellgren, K., Werner, M. W., and Dinerstein, H. L. (1983), Ap. J.
(Letters) 217, L149.
Willner, S. P. (1984), "Observed Spectral Features of Dust" in Galactic
and Extra alactic Infrared S ectrosco
eds. Kessler, M. F. and
Phillips, J. P. D. Reidel Publish ng Co., Dordrecht). 37.
THE HYDROGEN COVERAGE OF INTERSTELLAR PAHs
A. G. G. M. Tielens and L. J. Allamandola
Space Science Division, MS:245-6
NASA-Ames Research Center
Moffett Field, California 94035
J. R. Barker
Department of Atmospheric and Oceanic Science,
Space Research Building
University of Michigan
Ann Arbor, Michigan 48109-2143
M. Cohen
Radio Astronomy Laboratory
University of California at Berkeley
Berkeley, California 94720
ABSTRACT. We have calculated the rate at which the CH bond in
interstellar PAHs ruptures due to the absorption of a UV photon. The
results show that small PAHs (~25 carbon atoms) are expected to be
partially dehydrogenated in regions with intense UV fields, while large
PAHs (~25 carbon atoms) are expected to be completely hydrogenated in
those regions. Because estimates of the carbon content of interstellar
PAHs lie in the range of 20-50 carbon atoms, dehydrogenation is probably
not very important.
Because of the absence of other emission features besides the
11.3pm feature in ground-based 8-13pm spectra, it has been suggested
that interstellar PAHs are partially dehydrogenated. However, IRAS 822pm spectra of most sources that show strong 7.7 and 11.2pm emission
features also show a plateau of emission extending from about 11.3 to
13pm. Like the 11.3pm feature, we attribute this new feature to the CH
out-of-plane bending mode in PAHs. This new feature shows that
interstellar PAHs are not as dehydrogenated as estimated from groundbased 8-13pm spectra. It also constrains the molecular structure of
interstellar PAHs. In particular, it seems that very condensed PAHs,
such as coronene and circumcoronene, dominate the interstellar PAH
mixture as expected from stability arguments.
273
A. Uger et Ill. (eds.). Polycyclic Aromatic Hydrocarbons and Astrophysics. 273-286.
© 1987 by D. Rritkl Publishing COmptl"y.
274
1.
A. G. G. M. T1ELENS ET AL.
INTRODUCTION
Emission fe~tures at 3.3, 6.2, 1.1, 8.6 and 11.3~m have been discovered
in a wide variety of objects all characterized as having copious amounts
of UV photons (see the reviews of Aitken (1981) and Willner (1984) and
references therein). Spectroscopic analysis of the emission spectrum
implicates Polycyclic Aromatic Hydrocarbons (hereafter PAHs) as the
carriers (Duly and Williams 1981; Leger and Puget 1984; Allamandola,
Tielens and Barker 1985, 1986 herafter ATBa and b). In this
interpretation the 6.2 and 1.1~m features are due to the CC stretching
vibrations, while the 3.3, 8.6 and 11.3~m features are due to the CH
stretching and bending vibrations.
The 11-15~m spectral region, the region of the CH out-of-plane
deformation modes in PAHs, can be particularly indicative of the
molecular structure of the interstellar PAHs. For an increasing number
of adjacent H atoms on an aromatic ring, the position of this
vibrational mode shifts to longer wavelengths (Bellamy 1960)·. The
presence of only one interstellar emission band at 11.3~m ground based
8-13~m spectra has-been taken to imply that interstellar PAHs ar~
partially dehydrogenated (Duly and Williams 1981). That is,
interstellar PAHs are thought to have only isolated H atoms on their
aromatic rings. It has been suggested that this partial dehydrogenation
of interstellar PAHs results from the large UV flux in the emission
regions (Leger and Puget 1984).
We have theoretically investigated the hydrogen loss of
interstellar PAHs due to UV photon absorption. From these calculations,
it is concluded that the typical interstellar PAH with more than 25
carbon atoms is completely covered with hydrogen (section II). In order
to reconcile this apparent difference between (ground-based)
observations and theory, we have examined the 8-22~m LRS spectra
obtained by the IRAS satellite of all the sources which show the
unidentified emission features. These data show the presence of a
plateau of emission between 11.3 and 13~m, which has not been recognized
in ground-based studies due to telluric absorption (section III). The
implications of this emission plateau for the derived hydrogen coverage
of interstellar PAHs is discussed in section IV.
2.
HYDROGEN LOSS
When a PAH molecule absorbs an UV photon, internal conversion will
quickly transform the electronic excitation into vibrational excitation
in a lower electronic state. Generally, this internal conversion
process is so fast that direct H loss (e.g., through absorption into a
dissociative electronic state) is unimportant. The resulting highly
vibrationally excited PAH molecule can, however, still lose a hydrogen
atom when the number of vibrational quanta located in one CH bond is
sufficient to cause rupture (e.g., through excitation into the
vibrational continuum of the CH bond). In this section we will
calculate the probability for an interstellar PAH to lose a H atom in
this way.
THE HYDROGEN COVERAGE OF INTERSTELLAR PAH's
...'",
275
W
~
~
10000
~u..
1000
:I:
U
a:
o
u..
100
IZ
~
en
1
o
(.)
.1
Z
w
~
ow
a:
.01
~
.001
...J
~ .0001~~~~__~~__~~~~~__~~~~~~~
~
40000 50000 60000 70000 80000 90000 100,000 110,000
~
fig. 1
ENERGY, cm- 1
The rate of CH bond rupture as a function of the vibrational
energy content 9f PAHs calculated using quantum RRK theory
(Barker 1983).
Because little is known about the unimolecular reactions of PAHs,
we have used the simple quantum RRK statistical theory (see Barker 1983
and references therein), which requires less detailed knowledge of the
molecular species and reaction parameters than the RRKM statistical
theory. The details of this calculation are discu~sed elsewhere
(ATBb). The results of the calculation for several PAH molecules are
shown in figure 1.
for the same internal energy, small molecules lose hydrogen at a
higher rate than larger molecules. This is due to the fact that smaller
molecules have fewer vibrational modes in which to distribute the energy
and, thus, have a higher average excitation per vibrational mode. The
probability that enough energy is localized in one CH bond to cause
rupture is, therefore, higher. The calculated rate for CH bond rupture
has to be compared to that of other loss channels for the vibrational
excitation energy. The most important one is infrared fluorescence
emis~ion which typically has a rate of about 10
sec-. Comparison with the calculated results for chrysene (C 18H12),
for example, shows that hydrogen loss will dominate the fluorescent
relaxation for this ~lecule when it has a total vibrational energy·
content of 80000 cm- or more (~10eV; cf., figure 1).
References:
Notes:
1.0
4.0x10- 1
interstellar radiation field (lOS cm- 2 sec- 1; Habing 1968).
The hydrogen loss rate.
The rehydrogenation rate.
The fractional hydrogen coverage of chrysene (C 1SH 12 ; fuv = 0.5) and
circumcoronene (C54H18; fuv = 0.0).
Tielens and Hollenbach 1985b;
Castelaz et al. 1986.
b)
c)
d)
1)
3)
2) Ellis and Werner 1986;
The intensity of the UV field (6-13.6eV) normalized to the average diffuse
3
2
Ref.
a)
10- 5
3x10 3
104
NGC7023
Reflection
Nebula
1.Sxl0- S
1.0
2x10- 4 2.Sxl0- 1
6x10- 4
1.2xl0S
2x10 5
NGC7027
Planetary
Nebula
3.1x10- 1
10- 4
1.0
f d
H
CS4H 18
Rc
H
[sec -1 J C18H12
b
[sec- 1]
Ruv
2.2x10- 4
0
Ga
4.4xl0 4
(cm- 3 )
nH
lOS
Type
Orion Bar H II Region
Region
Table I: A comparison of the hydrogen loss rate and the
rehydrogenation rate in different interstellar regions.
r>
t:4
i
::l
~
?>
P
p
-.J
0'.
N
277
THE HYDROGEN COVERAGE OF INTERSTELLAR PAH's
The degree of dehydrogenation of interstellar chrysene molecules
can now be estimated (assuming no other energy relaxation channels such
as electronic fluorescence or ionization), by comparing the rate at
which these molecules absorb 10eV photons, RUY ' with their collision
rate with H atoms, RH (i.e., the maximum rehydrogenation rate
possible). These rates are given by
(1)
and
(2)
where 0uv is the UV photon absorption cross section, Nuv is the flux of
UV photons with sufficient energy to cause H loss (e.g., hv~10eV for
chrysene), <ov>H is the collision rate coefficient and n H is the density
of H atoms. The UV ab1orption cross section of PAHs for energies
greater than 50000 cm- (2000Aj 6eV) is not known. We will assume that
it is equal to the UV absorPY6on ~ross section associated with the
n-->n* transition, about 10cm for chrysene (Birks 1970). This is
perhaps on the low side. The
absorption cross section for coronene,
for example, is about 5 x 10- 1 (Birks 1970). Introducing Go as the
intensity of the UV field in terms of the a~erag~ diff~se interstellar
UV radiation field between 6 and 13.6eV (10 cm- sec- ; Habing 1968)
and fuv as the fraction of these photons between 13.6eV and the
threshold for H loss by a particular PAH (i.e., 10eV for chrysene)
yields for the H loss rate
gv
(3)
The collision rate coefficient is about 2 x 10- 10 cm 3 /sec, given a
geometrical cross section of 20A for chrysene and a mean H velocity of
km/sec at the gas temperatures of interest (~100K). This leads to a
rehydrogenation rate of
( 4)
Actually, the collision rate coefficient for interstellar PAHs may be
somewhat larger than this. Because of their low ionization potential,
most interstellar PAHs will be singly ionized (ATBa). The rate
coefficient for collisions of neutral H atoms with ionized PAHs is
probably comparable to the Langevin rate (~10-Y cm 3 /sec). In our
calculations we will assume that the rehydrogenation rate for
interstellar PAHs is given by
(5)
Table I lists the UV flux and density in three typical emission
regions derived from observations of the 01 (63~m) and ell (158~m) fine
structure lines. Note that these atomic cooling lines originate from
A. G. G. M. TIELENS ET AL.
278
the region where most of the FUV emission is absorbed (Tielens and
Hollenbach 1985a), which is also the region where most of the emission
in the infrared features originates. Assuming fuv is 0.5 for chrysene,
Ruv and RH have been calculated for these regions using these densities
and UV fields (table I). The rate for H loss in all three interstellar
regions is about a factor of 2 larger than the rehydrogenation rate. In
view of the uncertainties in the UV absorption and the collision cross
section assumed above, no firm conclusion can be drawn from this
comparison. It seems, however, possible that chrysene is partially
dehydrogenated in the interstellar medium.
Assuming that the hydrogen loss rate for partially dehydrogenated
chrysene is equal to that of completely hydrogenated chrysene, the
fractional hydrogen coverage of chrysene, fH' can be calculated. That
is,
(6)
Thus, chrysene might lose a considerable fraction of its hydrogen atoms
(c.f., table I). These results can be generalized to other PAHs by
using the internal energy threshold (i.e., the excition energy for which
the H loss rate equals the IR fluorescent rate), appropriate for the
specific molecule under consideration. Clearly, small PAH molecules,
with a threshold less than 13.6eV, may lose a considerable fraction of
their hydrogen atoms.
Large PAH molecules will, however, be completely hydrogenated. For
PAH molecules substantially larger than the ones shown in figure 1, the
calculated rate constant for CH bond rupture is always much less than
the cooling rate due to infrared fluorescence for interstellar
vibrational excitation energies less than 13.6eV (~110000 cm- 1 ; the H
atom ionization limit), where we have made the reasonable assumption
that the IR fluorescent rate for large PAils is similar to that of small
PAHs. Thus, outside of ionized gas regions, where more energetic UV
photons are available, such large PAH molecules will be completely
hydrogenated. Note that two photon processes (the absorption of a
second UV photon before the first one has been completely radiated away)
can cause hydrogen loss for large PAHs. The probability for the
occurance of a two photon process is, however, small. For example, fgr
the Orion bar this probability is calculated to be only about 4 x 10- ,
assuming a IR radiative lifetime of 0.1 sec. Even if all two photon
pro~esses lead to CH bond rupture, the hydrogen loss rate is only 2 x
10- tl sec- 1 . This is much less than the rehydrogenation rate (c.f.,
table I) and, thus, large PAHs will be completely hydrogenated.
Because of the steepness of the CII bond rupture rate with internal
vibrational excitation energy, the transition from small, dehydrogenated
PAHs to large, completely hydrogenated PAils is very sharp. From figure
1 it is estimated that this transition occurs for molecules containing
about 25 carbon atoms {e.g., coronene (C241112»' Estimates of the
number of carbon atoms in interstellar PAHs range from 20 to 50 (Leger
and Puget 1984; ATBa,b). It seems, thus, that the majority of the
interstellar PAHs should be fully hydrogenated.
\l,
279
THE HYDROGEN COVERAGE OF INTERSTELLAR PAH's
f'
2,O~
0
....
.... 1,5
x
;-,~-,--.~
'1
HD44179
4.0!Y
~O 3.5
HEN 1044
.... 3.0
x
X 1,0
X 2.5
:>
:>
...I
it 2.0
~
U..5.
1.5
·---~~I~~-~.
1,4 ~'
P18
1.2f
N
~ 1'O~
....
x
.8t
it
.4
I
o
1
06114+1745
....
N
I
o
x
X
~ .6 t
~
--.
:>
it
.4
.2
2.0
20319+3958
N
'7 1.0
o
....
~ 1.5
x
X
X .5
:>
:: 1.0
...I
U.
U.
.5
o
8 10 12 14 16 18 20 22
WAVELENGTH, microns
Fig. 2
17199-3446
16 18 20 22
The LRS spectra of six representative IRAS sources are
shown. The short wavelength data span the full useful range
from 7.67 to 13.45pm. The long wavelength data are shown
from 13.72 to 22.55pm (see text). The 7.7pm feature
dominates these spectra. Note also the inflection at 8.7pm
which is sometimes visible. The previously known 11.3pm band
shows a long wavelength p~ateau extending to about 13.0pm.
Ordinates are flux in W/m Ipm.
3. IRAS SPECTRA
Ground-based spectral studies of the CH bending modes in interstellar
PAHs are limited to the 8-13pm atmospheric window with considerable
280
A. G. G. M. T1ELENS ET AL.
uncertainty at the edges, because of atmospheric absorptions. Recently,
the Infrared Astronomical Satellite (IRAS) has obtained the first
complete interstellar infrared spectra in the 8-22pm region, which
contains these important vibrational modes of PAHs. This section
examines a set of spectra obtained by this satellite. The details of
the analysis are presented elsewhere (Cohen, Tielens and Allamandola
1985) .
3.1
The LRS Spectra.
The IRAS low-resolution spectrometer (LRS) has obtained short-wavelength
(8-13.5pm) and long-wavelength (1'-22.5pm) spectra, each with a
resolution (A/~A) ranging from 10 at the shortest to 40 at the longest
wavelength (De Huizon and Habing 1985). From the LRS data base we have
examined all the spectra characterized as having the l'.3pm band in
emission and selected a sample of 20 which show evidence for an emission
feature between 11.3 and 15pm. Objects with strong ionic emission lines
have been disregarded because of possible confusion (i.e., the Ne II
12.8pm line). Inspection of the NGS Palomar Observatory and SRC
southern sky photographs indicates that the vast majority of our sample
of 20 objects are associated with either very red, or reflection,
nebulae.
Figure 2 presents six typical spectra from this sample. Some of
these spectra are of well-known sources: the reflection nebulae, P18 and
HD 44179, and the WC 10 nucleus of a planetary nebula, Hen 1044 (=He 2113). Spectra of three sources newly found by IRAS (06114+1745, 171993446, and 20319+3958) are also included. To construct these spectra we
utilized the full useful range of the short-wavelength LRS spectra and
displayed the long-wavelength spectra starting at 13.72pm. Because of
the much lower spectral resolution of the long-wavelength IRAS detector
we have not used both long- and short-wavelength data in the region of
overlap ("-'3.5pm). However, the data in the long-wavelength spectra
reveal the same structure as the short wavelength spectra, albeit at a
degraded resolution. In fact the upturn shortward·of 15pm in the longwavelength spectra of some sources (e.g., P18) is due to the new
emission feature at this low resolution.
Besides the 1'.3pm feature all of the selected spectra show good
evidence for the presence of the well-known 7.7 and 8.7pm emission
features. The new feature has the appearance of an emission plateau,
ending abruptly at about 13.0pm. This plateau feature is a common
characteristic of many of the sources showing the ".3pm feature, but
not of spectra of other objects. The spectrum of 17199-3446 is
particularly striking because there is essentially no flux at all
between the emission features at 7.7 and 11.3pm, and again beyond the
plateau until 15.3pm. Note that the spike in the emission plateau in
this source is present in only one of the three IRAS spectra of this
source and, in fact, is probably not real (de Huizon, private
communication). Although all spectra in our sample show the same
pattern of emission between 11 and 15pm, there exist real spectral
variations from source to source (e.g., while evident, the new feature
281
THE HYDROGEN COVERAGE OF INTERSTELLAR PAH's
PAHS
1
2.0
2
N
~
3
1.5
I
4
o
X
X
:3
u..
5
1.0
'.QrH
H~H
H-<Q;H
H ,H
H::@H
H
~
5
.5
4
3
2
1
8
10
12
14
16
18
WAVELENGTH, microns
Fig. 3
<Q>H
H
@H
H~H
20
22
The average spectrum of the sources in our sample (see
text). The horizontal bars in the 11-15~m range indicate
the absorption range for isolated, 2, 3. 4 and 5 adjacent
hydrogens on an aromatic ring.
is weak in HD 44179).
The average of the 20 spectra (Fig. 3) shows the new emission
plateau between 11.3 and 13.0~m. Note that the peak wavelength of the
secondary feature at 12.7~m is largely determined by the brightest
source (HD 44179) and is therefore presumably not associated with Ne II
emission.
3.2
Interpretation
In line with the interpretation of the other emission features in these
spectra (e.g., 7.7,8.7, and 11.3~m), we attribute the newly discovered
emission plateau between 11.3 and 13.0~m to emission by interstellar
PAHs. The ground-based and airborne 3-13~m spectra of several of the
objects shown in Figure 2 are dominated by emission features at 3.3,
6.2, 7.7, 8.7, and 11.3~m. Although the new IRAS sources in our sample
have not yet been studied shortward of 7.7~m, the steep rise toward
7.7~m and the presence of the 11.3~m feature strongly suggest the
282
-. •...
A. G. G. M. TIELENS ET AL.
SERIESI
O@~.
SERIES20
SERIES30
SERIES40
Fig. 4.
**e ••...
* e •
'* _
•...
Homologous series of hexagonally symmetric PAH molecules. In
each series except 2, all of the molecules have tbe same
corner structure. Molecules in series 2 have three different
corner structures (Stein and Brown 1986). Note that the
molecules in series 1 have only isolated and two adjacent
hydrogens on their aromatic rings.
presence of these emission bands in their spectra also. Other
interpretations of the emission plateau between 11.3 and 13.0~m have
been considered and rejected (Cohen, Tielens, and Allamandola 1985).
The absorption frequencies of the CH out-of-plane bending mode are
highly characteristic for the number of adjacent H atoms on an aromatic
ring (Bellamy 1960). Due to strong coupling between the bending
vibrations of adjacent H atoms, the peak frequency of this mode shifts
to longer wavelengths when the number of adjacent H atoms increases.
This is illustrated in Figure 3 where the range of absorption
frequencies for different numbers of adjacent H atoms is indicated by
horizontal bars. Large PAHs generally have several absorption
bands in this wavelength region, because different rings can have
different numbers of adjacent H atoms depending on their location within
the molecule. For example, the condensed 10 ring molecule ovalene
(C 32 H14 ) has aromatic rings with one and with two adjacent H atoms. Its
spectrum shows, therefore, two absorption bands (at 11.43 and 11.92pm;
Sadtler spectrum 28159). In contrast, in the linear five-ring system
pentacene (C 22 H14 ) the end rings have four adjacent hydrogens, while the
other rings have isolated nonadjacent H atoms. Its spectrum shows
strong bands at about 11.05 and 13.6pm (Sadtler spectrum 15223). Thus,
THE HYDROGEN COVERAGE OF INTERSTELLAR PAH's
283
the 11-15~m wavelength region places important constraints on the
molecular structure.
It should be stressed that only PAHs with isolated H atoms can give
rise to an 11.3~m feature; hence, the suggestion that interstellar
aromatic hydrocarbons are only partially hydrogenated (Duly and Williams
1981). The theoretical estimates of the degree of dehydrogenation of
interstellar PAHs (see section II) and the discovery of a plateau of
emission longward of 11.3~m forces a reconsideration of partial
dehydrogenation. The IRAS spectra show that the interstellar PAHs
responsible for the emission features are predominantly made up of
aromatic rings with isolated, two, and, possibly, three adjacent H
atoms. PAHs with four or five adjacent H atoms are excluded. This
important clue strongly suggests that pericondensed PAHs such as ovalene
(C 32 H14 ) and circumcoronene (C5~H18)' which have only isolated and two
adjacent H atoms, dominate the lnterstellar PAH mixture.
Finally, the presence of spectral variations among the sources in
this sample should be emphasized. Unfortunately, the signal-to-noise in
most sources is insufficient to judge the reality of the minor
substructure on this newly discovered emission plateau. Yet, some
sources differ clearly from the average, The Red Rectangle (HD44119) is
one such case (c.f., figure 2). The simplicity of its 11-15~m spectrum
parallels that of its 3~m and 5-8~m spectrum (Cohen et al. 1986). The
weakness of the 11.3-13.0~m emission plateau in this source is
presumably related to its benign conditions, which allows smaller PAHs
to survive and dominate the infrared spectrum (Cohen, Tielens and
Allamandola 1985; Cohen et al. 1986).
IV.
DISCUSSION
In building up graphite sheets starting from benzene, different edge
structures can be recognized. Figure 4 shows these different edge
structures possible in various series of hexagonally symmetric PAHs
(Stein and Brown 1986). Disregarding the first member of each series
(Benzene), these series are characterized by a different number of
adjacent peripheral hydrogen atoms. The IRAS observations suggest that
the members of series one dominate the interstellar mixture, with
perhaps a small contribution of series three. Series two or four,
however, do not seem to contribute substantially to the interstellar
mixture. It is our contention that this reflects directly the carbon
condensation process in circumstellar shells.
Interstellar PAHs are probably formed in the carbon-rich outflows
from planetary nebulae and carbon Miras. Figure 5 illustrates the first
steps in the high temperature carbon polymerization route which is
thermodynamically most favorable (Stein 1918). Note that this carbon
condensation route goes through coronene and, eventually,
circumcoronene (i.e., series one in figure 4). Within this scheme, the
completely condensed PAHs, in which all aromatic rings are connected to
at least two others (e.g., pyrene and coronene), are even more stable
relative to the others. Essentially, this is because this structure
permits electron delocalization to a larger extent than the less
284
Fig. 5.
A. G. G. M. TIELENS ET AL.
First six members of the most thermodynamically favorable
high-temperature carbon condensation route (Stein 1978).
Successive members of this series are formed by adding two or
four carbon atoms at a time, the number required to complete
an additional fused aromatic ring. Note that C2 and C4H2
refer only to the number and type of atoms added in each step
and do not refer to specific chemical species or mechanisms.
condensed forms. These PAHs are, therefore, expected to dominate the
mixture of PAHs initially injected into the interstellar medium by
carbon stars and planetary nebulae and to survive the longest once in
the interstellar medium. For example, the distribution of PAHs injected
into the interstellar medium by carbon-rich objects will be further
modified by processes in the interstellar medium, such as
photodissociation, photoisomerization and sputtering in shocks
(Crawford, Tielens, and Allamandola 1985). Again, because of their
stabil ity, these processes will favor the 'Jery compact PAHs such as
pyrene, coronene, ovalene, and circumcoronene. Of course, other
geometries will also be present but to a lesser degree, in particular in
the harsher environments. In regions with benign conditions, such as
the Red Rectangle, one might, however, expect to see emission from these
less stable PAHs (as well as from smaller PAHs).
The dominance of these condensed PAHs in the interstellar mixture,
as evidenced by the 11-15~m spectra (e.g., the CH out-of-plane
deformation modes) may, therefore, merely reflect the role thermodynamic
stability plays in the formation and destruction processes of
interstellar PAHs.
THE HYDROGEN COVERAGE OF INTERSTELLAR PAH's
285
Acknowledgments: We thank Dr. Marie de Muizon for helpful discussions
on the IRAS LRS spectra. J. R. Barker acknowledges partial support from
the U.S. Department of Energy, Office of Basic Energy Sciences.
References
Aitken, D.K., 1981, in IAU Symp. No. 96, Infrared Astronomy, eds. C.G.
Wynn Williams and D.P. Cruikshank, (Reidel, Dordrecht), p.207.
Allarnandola, L.J., Tielens, A.G.G.M., and Barker, J.R., 1985, Ap.J.
Letters, 290, L25 (ATBa).
=:--:-_---:::---=_-:-::-=:-' 1986 in preparation (ATBb).
Barker, J.R., 1983, Chern. Phys., 77, 301.
Bellamy, L.J., 1960, "The Infrared Spectra of Complex Molecules",
(Methuen, London).
Birks, J.B., 1970 "Phetophysics of Aromatic Molecules" (London: Wiley
and Sons).
Castelaz, M.W., et a1. 1986, in preparation.
Cohen, M., Allamandola, L.J, Tielens, A.G.G.M., Bregman, J.D., Simpson,
J., Witteborn, F.C., Wooden, D., and Rank, D.M., 1986, Ap.J., 302,
737.
Cohen, M., Tielens, A.G.G.M., and Allamandola, L.J., 1985, Ap.J.
Letters, 299, L93.
Crawford, M.K., Tielens, A.G.G.M., and Allamandola, L.J., 1985, Ap.J.
Letters, 293, L45.
De Muizon, M. and Habing, H.J., 1985, Proc. of VIII European Regional
lAU Meeting, Toulouse, France, Lecture Notes in Phys., in press.
Duly, W.W. and Williams, D.A., 1981, M.N.R.A.S., 196, 269.
Ellis, B., and Werner, M.W., 1986, Ap.J., submitted.
Habing, H.J., 1968, BUll. Astr. Inst., Netherlands, 19,421.
Leger, A., and Puget, J.L., 1984, Astr. Ap., 13'1, L5.
Stein, S.E., and Brown, R.L., 1986, preprint.
Stein, S.E., 1978, J. Chern. Phys., 82, 566.
Tielens, A.G.G.M., and Hollenbach, D., 1985a, Ap.J., 291, 722.
:-:-:-:-::---::-:::----:-=0-:-' 1985b, Ap.J., 291, 747.
Willner, S. P., 1984, in "Galactic and Extragalactic Infra-Red
Spectroscopy", eds. M.F. Kessler and J.P. Phillips, (Reidel,
Dordrecht), p37.
286
A. G. G. M. TIELENS ET AL.
DISCUSSION
d'Hendecourc : What about dehydrogenation by direct UV photolysis of a
C-H bond on a PAH molecule ?
Answer : I expect that large PAH molecules will not have isolated electronic states. So that internal conversion will be much faster than
direct electronic dissociation.
Puley : Are there any sources that have been observed that show only the
11. 3 pm peak ?
Answer: All the sources in our lRAS sample (cf. Cohen, Tielens and
Allamandola, 1985, Ap. J. Letters, 299, L93) show evidence for the
emission plateau between 11.3 and 13 p~he lRAS sources (classified as
having an 11.3 pm emission feature) which are not included in our
sample, have too noisy a spectrum or there is evidence for contamination
by nebular line emission (e.g. Nell) to answer this question.
K. Roessler: Comment: The problems of temperature for H2-(or H)
elimination become less stringent, if you would assume a kinetic energy
of a few eV for the reactents C, C2, C4H2. etc. The so-called hot
reactions of insertion procede in general with H-elimination, without
demanding any temperature of the substrate itself. See paper K. Roessler
(poster).
A. Harchand : I do not unde'rstand very well how you can state that it is
not possible for a large PAH molecule to loose H atoms, since when such
molecules are heated in the laboratory they very easily loose hydrogen
at 400 - 500 0 C, a temperature much lower than the assumed vibrational
temperature of 1000 K.
Answer : The large PAHs cannot loose H atoms on the short timescale
available. In contrast to the laboratory experiments, the large interstellar PARs are out of thermodynamical equilibrium. The excess vibrational energy of the UV-pumped PAH can decay through several channels,
including IR photon emission and H loss. For large PAHs, the former
decay channel is favored in this competition. For small PAHs, it is the
other way around. Note that the concept of a vibrational temperature is
not very meaningfull for this highly NLTE situation. This hampers any
direct comparison with equilibrium laboratory experiments.
NEW OBSERVATIONS OF INFRARED ASTRONOMICAL BANDS:
IRAS-LRS AND 314m GROUND-BASED SPECTRA
M. de MUIZON
Sterrewacht Leiden, Postbus 9513,2300 RA Leiden, The Netherlands
and Observatoire de Paris, Section de Meudon, France.
L.B. d'HENDECOURT
Groupe de Physique des Solides de L'E.N.S., Universite Paris 7, France.
T.R. GEBALLE
Foundation for Astronomical Research in the Netherlands (ASTRON)
and United Kingdom Infrared Telescope, Hawaii, USA.
ABSTRACT. We present infrared spectral data of a sample of fourteen IRAS
sources. The observations include the IRAS Low Resolution Spectra, and groundbased 314m spectroscopy. The sources have been selected because they have strong
emission features at 7.7, 8.6 and B.3#4m in their IRAS-LRS spectra. We have detected in most of them the well-known features at 3.3 and 3.4#4m, together. with a
plateau at 3.4-3.6#4m. Additionally, in two sources observed at higher resolution,
new emission features at 3.46, 3.51 and 3.56#4m are detected. Together with the
3.40#4m feature they arise on top of a broad plateau which clearly emerges above
the 314m continuum. Wings are present on the edges of the 3.3#4m feature. The
IRAS spectra also reveal an emission plateau above the continuum longward of
11.5#4m. Possible interpretations for the new features are suggested and a preliminary analysis of these data is presented in the context of the Polycyclic Aromatic
Hydrocarbons hypothesis.
1. INTRODUCTION
Infrared emission features at 3.3, 3.4, 6.2, 7.7, 8.6 and B.3#4m have been detected
in a variety of objects, including planetary nebulae, reflection nebulae, HII regions
and extragalactic objects (see reviews by Aitken 1981 and Willner 1984). Although
attributed to dust particles, the carriers of this family of bands had never been
convincingly identified (see review by Allamandola 1984). The situation changed
when Leger and Puget (1984) showed that the wavelengths of the main absorption
bands in the infrared spectrum of a particular molecule (coronene) approximately
matched those of the emission features in the spectrum of NGC 7027 and HD44179
(the Red Rectangle). Each of the features was identified with a particular fundamental vibrational mode of a molecular species. Allamandola et al. (1985) proposed
287
A. Leger et al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 287-302.
© 1987 by D. Reidel Publishing Company.
288
M. DE MUIZON ET AL.
a similar explanation and described the excitation mechanism. Both groups of authors proposed that Polycyclic Aromatic Hydrocarbons (PAHs) could be responsible
for producing the "unidentified" infrared emission features observed in the interstellar medium. Although the basic P AH identification is quite promising, several
questions remain open. Among them is the variation in the intensity ratio of the 3.3
to 1l.3fJ.m features from object to object, which exists even though the two bands
come from fundamental vibrations of the same molecular sub-group (C-H stretching and bending, respectively). Other questions are related to the interpretation of
the 3.4fJ.m feature and additional features such as the feature at 3.53fJ.m, detected'
in two objects only (Blades and Whittet 1980; Allen et al. 1982), the plateau of
emission extending from 3.4 to 3.6fJ.m, observed in a few objects by Geballe et al.
(1985), and the new features reported in this paper.
In order to extend the available sample of spectroscopic data on infrared emission
features for further studies, we have used the IRAS database of Low Resolution
Spectra (LRS). We have selected from it a number of sources having strong emission
features at 7.7, 8.6 and 11.3fJ.m, and we have searched for the counterparts of these
features in the 3fJ.m region from ground-based telescopes. In this paper, we present
low (CVF) and moderate (grating) spectral resolution observations from 3.0 to
3.8fJ.m, obtained at UKIRT (Mauna Kea) or at ESO (La Silla), for a subset of our
selected IRAS-LRS sources: grating spectra have been obtained for two sources of
the sample, and CVF spectra for fourteen sources. A first analysis of these new
data is presented in the context of the P AH hypothesis.
2.
OBSERVATIONS
Several sets of observations are reported here: the IRAS Low Resolution (LRS)
spectra (7.7-22.5J.1.m); ground-based CVF (Circular Variable Filter) 3fJ.m spectra
and ground-based grating 3fJ.m spectra obtained at UKIRT, the United Kingdom
Infrared 3.75m Telescope (Mauna Kea, Hawaii), in Autumn 1985, and ground-based
CVF 3fJ.m spectra obtained at the ESO (European Southern Observatory, la Silla,
Chile) 1m telescope in August 1985. The sample presented here is composed of
fourteen IRAS sources. For each of them, we present IRAS-LRS and CVF 3.03.8fJ.m spectral data. The 3fJ.m CVF spectra were obtained at UKIRT for all the
sources except the three at declination lower than -45 0 for which they were obtained
at the ESO 1m telescope. Additionally, for two of the northern sources, IRAS
21282+5050 and 03035+5819, we have also obtained spectra at higher resolution,
with a seven-channel (InSb detector) cooled grating spectrometer (Wade 1983). All
the 3J.1.m spectra (UKIRT and ESO) were flux-calibrated by observing standard
stars, measured close in airmass to the objects, in order to remove atmospheric
absorption features. Flux calibration is thought to be accurate to ±20%.
The LRS spectra were obtained in 1983 by the Low Resolution Spectrometer (Wilde-
289
NEW OBSERVATIONS OF INFRARED ASTRONOMICAL BANDS
man et al. 1983), onboard IRAS, the Infrared Astronomical Satellite. Two wavelength ranges, 7.7-13.5ILm and 11-22.5ILm, were recorded simultaneously, with respective fields of view of 6'x5' and 6'x7 .5'. However, the instrument is basically
an objective prism and therefore good quality spectra are obtained only for sources
less than 20 to 30 arcsec in diameter. In each wavelength band, the resolving power
increases with wavelength and varies from 10 to 40. The LRS data have been calibrated using the survey calibration (IRAS explanatory Supplement 1985). The
absolute flux calibration of the LRS spectra is thought to be accurate to ±30%.
All spectra are the average of several individual spectra (generally 3 to 6). A log
of all the observations presented in this paper, IRAS and ground-based, is given in
Table 1.
Table 1: Loa or OllBrYlltl one
Spectral
Re... oll.ltion
WavelenBth
Coverage
Telescope
D1ameter
7.7-22.5...,
SOCII
6'x5'
~8ourcesl
S30"
12.q"
(1/61)
lRAS-LRS
Spectra (1983)
10 to qO
UKIRT
CVF Spectra
(Sept_bar 1985)
100
3.0-3.8111
3.7S.
UKIRT
Grati nl Spectra
(Sept_bar 1985)
QOO
3.05-3.85...,
3.7511
3.0-3.8...,
1.
ESo-1M
70
Beam
S1ze
Sc .... ces
Obolarved
All 1 q so .... ces
or this s_ph
03035+5819
OSOq4-0325
06572-07q2
2002Q+3330
20319+3958
22308+5812
03260+3111
06303+1021
'8Q'6-oQ20
20293+3952
21282+5050
5"
03035+5819
21282+5050
15"
12063-6259
16362-Q8Q5
12389-61 Q7
3. RESULTS
We present in Figure 1 the spectra obtained for the two sources IRAS 21282+5050
and 03035+5819 (AFGL 437), those for which we have also obtained grating 3ILm
spectra. The LRS spectra (Figure la) of the two sources have similar shapes shortward of 12ILm and three features in common at 7.7, 8.6 and 1l.3ILm. Due to the
short wavelength cut-off of the LRS, only the long wavelength slope of the 7. 7ILm
feature is evident, with the 8.6lLmfeature appearing on it as a shoulder. From 12 to
23ILm the slopes of the two spectra differ significantly. In 21282+5050, the emission
feature at 11.31Lm is narrower than usual in other IRAS-LRS sources (de Muizon
and Habing 1985). The weak and narrow emission feature at 12.61Lm appears to be
real, since it is present on all the individual spectra. There is no possible confusion
with the [NeIll 12.81Lm emission line. It is clear that the long wavelength slope
290
M. DE MUIZON ET AL.
(a)
21282+5050
i-1.~
~~
2.5
,
~~
2
E
~
=
~
I
=
~
03035+5819
f.~
~
><
I
:>
~
0.5
10
12
I.
16
18
20
22
8
10
12
WAVELENGTH ( pm )
I.
16
18
22
20
WAVELENGTH ( pm )
8
;~
(b)
6
"I~
!
"I~
!
•
~~
3
,
2
g
x
><
~
~
----------0
3.2
3.6
3.'
WAVELENGTH ( pm )
0
3.8
3.2
3
3.8
3.6
3.'
WAVELENGTH ( pm )
6
,
"!~
(c)
21282+5050
5
•
~
~
~
1.5
I
'0
~ 0.5
~
I
0
a
~
1 1 1! ! !
3
~
;
"I'
E
3
3.2
3.'
3.6
0
3.8
3
3.2
WAVELENGTH ( pm )
3.'
3.6
3.8
WAVELENGTH ( pm )
2.5
;~
a
(d)
2
;
03035+5819
a
0.8
"I'
~.
II
~
Ji 1.5
;0
g
~><
0.6
~ 0.4
~
~
0.5
3'
3.2
3.6
3.'
WAVELENGTH ( pm )
3.8
0.2
3
3.2
3.6
3.'
WAVELENGTH ( pm )
3.8
Figure 1. Infrared spectra of the two IRAS sources 21282+5050 and 03035+5819. (a) IRASLRS spectra: thick line is band 1 (7.7-13.5I'm)j thin line is band 2 (1l-22.5I'm). In the overlapping range, the resolution is A:l40 in band 1 and JO;110 in band 2. (b) UKIRT 31'm CVF spectra.
Resolution is A:llOO. (c) UKIRT 31'm grating spectra. Resolution is A:l400. (d) Blow-up of (c).
"Flux· is Flux density in units of 1O-lOWcm- 2pm- 1 for the IRAS spectra, and 1O-lTWcm-2IAm-1
for the 3IAm spectra.
291
NEW OBSERVATIONS OF INFRARED ASTRONOMICAL BANDS
of the 11.3ftm feature does not reach the continuum. Therefore, there is a significant plateau of emission from 11.5 to about 15ftm. Photometry with the IRAS
Survey Instrument (Table 2) shows that the infrared emission peaks in the short
wavelength bands, and decreases sharply from 25 to l00ftm. For 03035+5819, the
situation is different. The shapes of the features at 7.7,8.6 and 11.3ftm are more
similar to what is observed in other objects (e.g. the Red Rectangle). Again there
is an excess between 11.5 and 13ftm, similar to the plateau reported by Cohen et
al. (1985), together with two possible weak emission features at 12.4 and 12.8ftm.
The first is seen on each individual spectrum. However, the second, which occurs
at the position of the [NeIIjline, is present in only two out of three individual spectra. These two features should not be given much credence until high resolution
and more sensitive spectra are obtained in this wavelength range. The slope of the
spectrum of 03035+5819 is positive and consistent with the IRAS Survey data.
Table 2: lRAS broad band data
IRAS Inband Survey fluxes Band
(W .m- 2 )
Band
Band
Band
21282+5050
1
2
3
4
( 12)J111)
( 25)J111)
( 60)JIII)
( 1 OO)JIII)
6.45
3.70
8.40
1.51
10- 12
10- 12
10- 13
10- 13
03035+5819
3.92
1. 98
2.68
1. 31
10- 12
10- 12
10- 11
10- 11
In the 3ftm CVF spectra of both sources (Figure Ib), three emission fea~ures appear
clearly. They are the strong and narrow feature at 3.3ftm, the weaker and narrow
one at 3.40ftm, and the plateau extending from 3.35 to 3.6ftm on which the 3.40ftm
feature is superimposed. In earlier papers, the 3.40ftm feature and the plateau were
not distinguished from one another and were collectively referred to as the 3.4ftm
feature. Following Geballe et al.(1985), we make the distinction between the narrow
3.40ftm feature and the plateau.
The most significant new result of the 3ftm spectroscopy of these two sources
is the detection of three new emission features having peaks at 3.460±0.OO5ftm,
3.515±0.OO5ftm and 3.565±0.010ftm (corresponding respectively to the frequencies
2890±5, 2845±5 and 2805± 10 em -1). All of these features are clearly seen in the
grating spectra (Figure lc,d); some of them are marginally evident in the CVF spectra (Figure Ib). All three are present in the spectrum of 21282+5050. Only the
first two are in that of 03035+5819. These two also appear weakly in the spectrum
of HD44179 (the Red Rectangle) by Geballe et al. (1985), although not mentioned
by them. The feature at 3.56ftm is only seen in 21282+5050 and is rather weak;
however, its presence in that source was confirmed in additional grating spectra obtained in November 1985 (de Muizon et al. 1986). Along with the 3.40ftm feature,
the new features are superimposed on the plateau emission. The width of each new
feature is 1'::I0.03ftm (FWHM). Their peak intensities are all less than that of the
M. DE MUIZON ET AL.
292
•
Table 3: Par aaeters of the observed Infrared Emission Featlres in
sOirees 21282+5050 and 03035+5819.
311m CVF data 9 12.4"
311m Grating data 9 5"
21282+5050
03035+5819
21282+5050
3.3I1m:
Amlak
F M
\/peak
Fobs
3.28±0.01
0.053±0.008
3049±15
3.8±0.4
3.28±0.01
0.056±0.008
3049±15
1 • 3±0. 2
3. 4O llm:
Apeak
vpeak
Fobs
3.39±0.01
2952±15
0.45±0.05
Plateau
3.4-3.6lJM:
3.44±0.06
Apeak
2907±50
\/peak
1.8±0.2
Fobs
3. 3I1m+wings:
3.292±0.005
~~k 0.045±0.005
3038±5
vpeak
2.4±0.2
Fobs
3. 290±0.005
0.045±0.005
3039±5
0.65±0.07
3.39±0.01
2952±15
0.19±0.02
3.40Ilm:
Amlak
F M
\/peak
Fobs
3.399±0.005
0.035±0.005
2942±5
0.22±0.02
3. 395±0. 005
0.031±0.005
2945±5
0.08±0.01
3.46±0.06
2890±50
0.40±0.04
3. 46lJM:
Amlak
F M
vpeak
Fobs
3.461±0.005
0.035±0.008
2889±5
0.09±0.01
3.459±0.005
0.035±0.008
2891±5
0.05±0.01
3.51I1m:
Amlak
F M
\/peak
Fobs
3.516±0.005
0.022±0.005
2844±5
0.045±0.01
3.514±0.008
0.026±0.005
2846±8
0.03±0.005
3.56lJM:
Amlak
F M
vpeak
Fobs
3.565±0.001
0.035±0.010
2805±10
0.06±0.01
IRAS-LRS data 9 6'x5'
21282+5050
03035+5819
7.7lJM:
Apeak
Fobs
7. 7±0. 15
132±30
7. ?to.15
190±40
8.6lJM:
Apeak
Fobs
8.8±0.1
1H2
8.6±0.1
19±4
11.3lJM:
Apeak
Fobs
' 1 • 4±0. 1
40±5
1'.4±0.1
28±3
Plateau
l'.5-'5)l1D:
54±10
Fobs
*
Units: A in
)lID;
03035+5819
19±4
FWHM in
)lID;
Plateau
3.4-3.6lJM:
3.45±0.05
Amlak
0.20±0.04
F M
2900±50
\/peak
1 • 1±O. 1
Fobs
3.45±0.05
0.20±0.04
2900±50
0.18±0.02
v in cm- 1 ; Fobs in 10- 18 wcm- 2 •
NEW OBSERV AnONS OF INFRARED ASTRONOMICAL BANDS
293
3.40p,m feature and decrease with increasing wavelength. In the two sources, the
features at 3.40 and 3.46p,m both present a shoulder on their long wavelength edge,
possibly indicating the presence of blended features. The detection and strength
of all the features in these two sources were confirmed by the additional November
1985 grating spectra. The grating spectra of both sources also reveal details not
seen at lower resolution in the shape of the 3.3p,m feature which is resolved. Its
width is 0.045±0.005p,m (FWHM). In both sources, two wings are apparent on the
edges of the 3.3p,m feature, displaced from its centre by 0.05p,m (46 cm- l ). They
are particularly prominent in the case of 03035+5819. Parameters of the various
features in the 3p,m and 10p,m spectra of sources 21282+5050 and 03035+5819 are
summarised in Table 3.
Figure 2 shows the IRAS-LRS spectra and 3p,m-CVF spectra of four of the sources
observed at UKIRT. For all these sources the LRS spectra show the long wavelength
slope of the 7.7 p,mfeature, the 8.6p,m feature as a shoulder, and the 11.3p,mfeature.
For the first two sources, 22308+5812 (SI38) and 18416-0420 (AFGL 2243), the
ionic fine-structure lines of [Nell] 12.8p,m and [SIll] 18.7 p,m are prominent, and
are characteristic of medium excitation HII regions. Indeed, these two sources also
show the Ph (3.74p,m) hydrogen recombination line in their 3p,mspectra. The 3p,m
spectra of these four sources all have a strong 3.3p,m emission feature and a plateau
of emission from 3.35 to 3.6p,m. The feature at 3.40p,m is clear only in 065720742 (Parsamyan 18). It may be present, although weak, on top of the plateau
in the other sources. The presence of wings on the edges of the 3.3p,m features is
conspicuous in these four sources. The plateau of emission between 11.5 and 13p,m
is also conspicuous in the LRS spectra of 03260+3111 (NGC 1333) and 06572-0742.
We suspect it is also present in 22308+5812 and 18416-0420 but the [Nell] 12.8p,m
line superimposed on it makes it less obvious. The feature intensities deduced from
the CVF and IRAS-LRS spectra are given in Table 4 for the fourteen sources of the
sample studied.
4. ANALYSIS AND DISCUSSION
4.1 The New Features in Sources IRAS 21282+5050 and 03035+5819
The major significance of these observations is the discovery of three new emission
features at 3.46, 3.51 and 3.56p,m, which appear together with the previously known
emission features at 3.3 and 3.4p,m, and the 3.4-3.6p,m plateau. The widths of
the new features as well as the absence in these spectra of lines such as [Nell]
12.8p,m, [SIll] 18.7p,m and Ph (3.74p,m) indicate that the new features are not
atomic emission lines. It seems most likely that they are related to the "family"
which includes the 3.3, 3.4, 6.2, 7.7, 8.6 and 11.3p,m features and the plateaux
at 3.4-3.6p,m and 12p,m. This interpretation is assumed for the remainder of this
paper.
Figure 2. Left: IRAS-LRS spectra (see Fig. 1a caption). Right: UKIRT 31'm CVF spectra.
"Flux· is Flux density in units of 1O-16W cm- 2 I'm- 1 for the IRAS spectra, and lO-lTW cm- 2 I'm- 1
for the 31'm spectra.
295
NEW OBSERV AnONS OF INFRARED ASTRONOMICAL BANDS
Table 4: Feature Intensities* (in ...it of'10-11l v.- 2 ). deduced frau
CYF and lRAS-LRS spectra. far the 14 sources of the s_ple.
3. 4011m
SOURCE
3.31J11l
03035+5819
AFGL 437
1.3
Yes
03260+3111
NOC 1333 #3
1.7
05044-0325
FIRSSE 67
PLATEAU/
LINES
3.4-3.6I1m
8.61J11l
Yes
190
19
28
19
Yes
Yes
730
28
70
35
.22
?
?
230
9
25
18
06303+1021
HDE 259431
1 .2
No
No
44
4
6
4
06572-0742
PARSAMYAN 18
FIRSSE 192
1.0
Yes
Yes
195
8
30
12
12063-6259
HE2-77
1.8
Yes
?
344
8
46
42
12389-6147
3.5
Yes
?
217
29
23
17
16362-4845
RCW 108
3.6
Yes
?
1490
50
259
110
18416-0420
AFGL 2243
4.2
Yes
Yes
470
23
57
63
20024+3330
<0.5
No
No
142
12
27
8
20293+3952
0.4
?
?
180
8
25
16
20319+3958
4.8
Yes
Yes
314
19
51
25
21282+5050
3.8
Yes
Yes
132
11
40
54
22308+5812
S138
1• 1
Yes
Yes
236
12
45
23
*
11.311m
PLATEAU
.5-131J11l
7.71J11l
11
Errors are of the order of 20% for lines in the LRS spectra and of
10% for the 31J11l data.
For the 3.401Jlll feature and the plateau at 3.4-3.61J11l:
Yes = detected. No = not detected. ? = doubtful
296
M. DE MUIZON ET AL.
Although the present data provide information on the wavelengths and shapes of
the new features, no specific identification can be made at present. Even for the
previously known family members no definite assignment exists. Leger and Puget
(1984) noticed that the set of emission bands at 3.3, 6.2, 7.7, 8.6 and 1l.3Jlm is
not only present in the absorption spectrum of one particular molecule (coronene
C 24 H 12 ) but that each of these lines, always observed as a set, can be identified with
a fundamental vibration frequency: C-H stretch, C=C stretch, C=C skeletal mode,
C=H bending in-plane and C=H bending out-of-plane respectively. They suggested
that a mixture of various molecules of this type -Polycyclic Aromatic Hydrocarbons or PAHs- are responsible for these emission features. The advantage of the
PAH interpretation is that, with the minimum of assumptions, it can explain simultaneously the position of the lines, their respective intensities and their excitation
mechanism. Allamandola et al. (1985) supported this suggestion by showing the
good agreement between the wavelengths of the observed IR bands and the Raman
spectrum of automobile soot (a mixture of PAHs and small carbon particles). The
identification of the 3.3 and 1l.3Jlm features with aromatic C-H stretching and
bending had been proposed also by Duley and Williams (1981).
Based on the presently available data there appears to be some hierarchy among the
3Jlm emission features. The 3.3Jlm feature is present whenever any other features
are present. For the 3.4Jlm feature to be present only the 3.3Jlm feature needs
to be present. For the plateau to be present, both the 3.3 and 3.4Jlm features
must be present and for the 3.46, 3.51 and 3.56Jlm features to be present all of the
above (with the possible exception of the plateau -see the spectrum of HD44179
in Geballe et al. 1985) are required. This hierarchy may be understandable in
terms of a range of stability and/or complexity of the molecules responsible for
producing the various features. The typical size of the hypothetical PAHs was
estimated to be 20 to 50 carbon atoms (Leger and Puget 1984; Allamandola et
al. 1985). Absorption spectra of some small PAIls, such as pyrene (C 16 H lO ),
perylene (C 2o H 12 ) and coronene (C 24 H12)' have been published (Sadtler spectra
1959) and are promising: the spectrum of coronene matches rather well, although
not perfectly, the observed spectrum of the Red Rectangle (Leger and Puget 1984).
There is probably a certain mixture of PAHs in the ISM and they may well be in
physical conditions very different from what we can presently test in the laboratory
(e.g. ionized or partially dehydrogenated). However this mixture must be somewhat
limited since only a limited number of P AHs are stable enough to survive the rough
conditions of the ISM (Reed and Tennent 1971; Leger and d'Hendecourt 1985).
Another argument in favour of the presence of a mixture of PAHs is also suggested
by the i2Jlmplateau, which is observed in many IRAS-LRS spectra having the 7.7,
8.6 and 11.3Jlm features. As pointed out by Leger and Puget (1984), the exact
wavelength position of the out-of-plane bending mode, responsible for the 1l.3Jlm
feature, depends on the number of adjacent H-atoms on a ring. The more H-
297
NEW OBSERV AnONS OF INFRARED ASTRONOMICAL BANDS
atoms that are adjacent, the longer the wavelength of the bending mode becomes
(Bellamy 1966j Clar et al. 1981). The predominance of the 1l.3J.tm feature points
to molecules with a majority of solo H (no adjacent CH)j but the emission observed
from 11.5 to about 15J.tm might indicate the presence of molecules with duo and
trio H (one or two adjacent CH).
Table 5: C-H Stretching FrequenCies in various Molecular
SUbgroups (Bell., 1966).
Wa vel ength
Radical
Alkanes (saturated)
C-H in CH 3
C-H in CH 2
C-H in CH
29621 and 2862 2
2926 1 and 2853 2
2890
3.38, 3.49
3.42, 3.51
3.46
Aranatic
-C-H
3030
3.3
().lID)
lsymmetric mode
2Assymmetric mode
The conspicuous presence of the plateau at 3.4 - 3.6J.tm and of the lines at 3.40,
3.46, 3.51 and 3.56J.tm gives a new insight to the possible nature of the molecules
responsible for the emission. The precise position of the C-H stretching vibration
frequency depends on the nature of the molecules considered. In an unsaturated
hydrocarbon such as an aromatic, the mode occurs at ~ 3030 cm- 1 (3.3J.tm). In a
saturated hydrocarbon such as an alkane, the C-H vibration lies in the frequency
interval 2962-2853 cm- 1 , i.e. 3.38-3.51J.tm (see Table 5, taken from Bellamy 1966).
These frequencies fall in the range reported in this paper for the new lines. We thus
suggest that these lines could be due to PAHs, on which are attached molecular
subgroups such as -CHa or -C2HS. Infrared spectra of methyl- and ethyl-coronene
have been obtained in CsI matrices (Leger and d'Hendecourt 1986). They all show
a series of features spanning the range 3.37-3.55J.tm, together with the strong 3.3J.tm
feature which originates from the normal aromatic C-H stretch. However, none of
these series coincide exactly with the complete set of observed features, so that
a definite precise identification is not yet possible. Similar but smaller molecules
such as toluene (CsHsCHa) and xylene (CaHs(CHah) are easily ionized and photodissociated in a strong UV field, producing the ion C 7 Hi (CaHsCHi) whose
infrared spectrum is not yet known (Dunbar 1973a,bj Omont 1986). Such free radicals and ions can be produced in the laboratory via matrix isolation techniques
(Bass and Broida 1960j Bondybey and Miller 1983). These techniques have already been used for the interpretation of astronomical infrared spectra (Lacy et al.
1984j d'Hendecourt et al. 1986). Such laboratory studies on PAHs should bring a
significant input to the identification of the new features.
M. DE MUIZON ET AL.
298
Finally, the wings on the 3.3JLm feature might provide another observational constraint. One possible explanation for these wings could be looked for in Fermi
resonances between two vibrational levels (e.g. a fundamental and an-harmonics)
very close in energy. The coupling between these two levels creates a doublet around
the fundamental (Herzberg 1945; Silverstein et al. 1974) as what is observed for
example in CO 2 , CHaOH and CsHs (Herzberg 1945). However, a Fermi resonance
is very difficult to recognise even in a well studied molecule, and it is not specific of
a particular structure of the molecule.
4.2 Intensity Ratio of the 11.3 to the 3.3JLm Features and the Size of the
Emitting Molecules.
In the following we investigate the possibility of using the combined 3JLm and 10JLm
spectra of our sample of objects in order to deduce the average temperature of
the emitting molecules and estimate their average size (Le.: the number of carbon
atoms per molecule). We adopt the approach to the problem as described by Leger
and Puget (1984).
The emission of a species in one of the bands can be computed from its absorption
spectrum and is a function of the temperature T. The intensity emitted in one band
can be expressed as:
I>. = B>.(T) . N . u>.
(1)
where B>. is the Planck function intensity, N is the column density of the emitting
molecules and u>. is the infrared cross-section per molecule integrated over the band
considered. We have :
(2)
where Nc is the total number of C-atoms in the molecule and A>. is the infrared
cross-section of the band per C-atom. Quantitative infrared absorbance spectra of
various PAH molecules including large PAHs such as coronene (Nc=24), ovalene
(Nc=36) and hexabenzocoronene (Nc=48) have been measured in CsI matrices
(Leger and d'Hendecourt 1986). The infrared cross-sections per C-atom, for the
bands at 3.3 and 11-13JLm, are found not to vary much froin one molecule to another.
Therefore we have adopted in our calculations the values of A>. for coronene at 3.3
and 11.9JLm.
The average temperature < T > of the cooling molecule can be derived from the
observed intensity ratio of the 11.3 to 3.3JLm lines through the equation:
(3)
Indeed the bands at 3.3 and 11.3JLm originate in two modes of the same molecular
subgroup (aromatic C-H stretching and bending respectively); therefore, provided
their relative oscillator strength is not too variable with the surrounding, one expects
299
NEW OBSERVATIONS OF INFRARED ASTRONOMICAL BANDS
Table 6: Ratios of 1'.3}a to 3.3.. Featlre8
for the 111 SOlrces of the SIDple.
SOURCE
I(11.31J11l)
I~ 3.3Ilm~
<T>a
(K)
Nl.IIlber of
C atoms b
03035+5819
AFGL 437
21 ±7
565 ±30
56 ±6
03260+3111
NGC 1333 113
41 ±12
500 -+~o0
71 ±7
05044-0325
FIRSSE 67
113 ±35
430 ±15
97 ±7
06303+1021
HOE 259431
5 ±1
800 ±50
28 ±4
06572-0742
PARSAMYAN 18
FIHSSE 192
30 ±10
530 +40
-25
64 +6
-9
12063-6259
HE2-77
25 ±8
+40
550 -15
59 +4
-8
12389-6147
7 ±2
+60
725 -45
34 ±5
16362-4845
HCW 108
72 ±20
460 ±20
85 ±7
18416-0420
AFGL 2243
14 ±4
620 ±40
47 ±6
20024+3330
>52
20293+3952
63 ±20
20319+3958
11
<480
>78
465 -+~O0
83 +7
-10
±3
650 +50
-30
42 ±5
21282+5050
1 1 ±3
650 +50
-30
42 ±5
22308+5812
S138
39 ±10
+25
510 -20
68 ±6
a Average temperature of PAHs
b Assuming a flux of UV photons of average
energy hv • 6 eV
300
M. DE MUIZON ET AL.
their ratio to be dependent on the temperature rather than on a particular molecule.
Figure 3 shows the dependence of the intensity ratio I(1l.3ILm}/I(3.3ILm) versus
temperature. The observed intensity ratios for our sample of fourteen IRAS sources,
and the corresponding average temperatures of the molecules are given in Table 6.
Although the errors on the intensity ratios are quite high (l':::! 30%), the deduced
average temperatures span at least the range 400 to 8ooK. The average temperature
is related to the peak temperature reached by a molecule just after absorption of
a UV photon. The peak temperature depends on: the energy of the incoming
photon, the specific heat of the molecules, and the efficiency of the transfer of
electronic energy to vibrational energy. For our calculations, we adopt the specific
heat given by Leger (this volume). We assume that the efficiency of the conversion
of electronic energy to vibrational energy is equal to 1. We have then:
rTpeak
hl/uv =
10
C(T)dT
(4)
where C(T) is the specific heat of a molecule. From the peak temperature the
cooling of the molecule as a function of time is computed by taking into account
the emission of infrared photons in all the bands present in the spectrum of the
molecule. The ratio of the energy emitted in the 1l.91Lm band to the energy emitted
in the 3.31Lm one during the cooling can be computed and an average temperature
deduced to be used in Equation (3). We obtain < T > l':::! ~ Tpe<>" . The last
unknown in Equation (4) is the energy of the incident UV photon. In absence of
astronomical data allowing a precise determination of the UV flux distribution in
our sources, we estimate the number of C-atoms per molecule for different values
of the average energy per UV photon. The dependence of No as a function of
temperature, for hl/uv = 4, 6, 8 and lOeV is shown in Figure 4, and the values
of N c , in the case hl/uv = 6eV, are reported in Table 6 for each source. There
appears to be a rather wide range of the number of C-atoms per molecule among
the observed sources. However, two restrictions should be applied to this analysis.
Firstly, the variation in the intensity ratio is a function of the size of the molecules
but also of the spectral energy distribution of the incident UV flux. More data
on the UV-visible energy distribution are required. Secondly, some problems may
occur when ratioing the intensities of two bands, which have been measured with
totally different instruments and different beam sizes, especially in sources where
the emission is spatially extended. The beam sizes for the 3ILm observations are
12.4 and 15 arcsec. In all cases, the IRAS beam (6'x5' for the 7.7-13ILm channel)
include the whole source. We emphasise that the sources selected in our sample
have high quality LRS spectra. This ensures that they are not more extended than
20 to 30 arcsec. Nevertheless, ratios larger than l':::! 50 in Table 6 should be taken
with care. Spectral mapping of these sources is in progress and will allow a more
accurate determination of the 1l.3ILm/3.3ILm bands intensity ratios.
NEW OBSERVATIONS OF INFRARED ASTRONOMICAL BANDS
301
N(e)
150
100
'i- 10'
50
lOeV
,
8eV
8eV
...................L..J.........................'-'-.&....I....L.I....L.........~
200
<400
600
800
<T> (K)
1000
1200
4eV
800
Figure 3. Dependence of the intensity ratio
I(11.3~m)/I(3.3~m) versus average temperature.
1000
1200
1<400
T..... (IC)
Figure 4. Dependence of the number of carbon atoms per molecule as a function of the
peak temperature reached by the molecule,
for different values of the average energy per
UV photon.
REFERENCES:
Aitken, D.K. 1981, in IAU Symposium 96, Infrared Astronomy,
ed. C.G. Wynn-Williams and D.P. Cruishank (Dordrecht: Reidel), p.207.
Allamandola, L.J. 1984, in Galactic and Extragalactic Infrared Spectroscopy,
ed. M.F. Kessler and J.P. Phillips (Dordrecht: Reidel), p.5.
Allamandola, L.J., Tielens, A.G.G.M., and Barker, J.R. 1985,
Ap. J. (Letters), 290, L25.
Allen, D.A., Baines, D.W.T., Blades, J.C., and Whittet, D.C.B. 1982,
M.N.R.A.S., 199,1017.
Bass, A.M., and Broida, M.P. 1960, in Formation and Trapping of Free Radicals,
(New-York: Academic Press).
Bellamy, L.J. 1966, in The Infrared Spectra of Complex Molecules,
(New-York: Wiley).
Blades, J.C., and Whittet, D.C.B. 1980, M.N.R.A.S., 191, 701.
302
M. DE MUIZON ET AL.
Bondybey, V.E., and Miller, T.A. 1983, in Molecular Ions: Spectroscopy, Structure
and Chemistry, ed. T.A. Miller and V.E. Bondybey (North Holland) p.125.
Clar, E., Robertson, J.M., Schlogl, R., and Schmidt, W. 1981,
J. Am. Chern. Soc., 103, 1320.
Cohen, M., Tielens, A.G.G.M., and Allamandola, L.J. 1985,
Ap. J. (Letters), 299, L93.
Duley, W.W., and Williams, D.A. 1981, M.N.R.A.S., 196,269.
Dunbar, W.C. 1973a, J. Am. Chem. Soc., 95,472.
Dunbar, W.C. 1973b, J. Am. Chern. Soc., 95, 6192.
Geballe, T.R., Lacy, J.H., Persson, S.E., Mc Gregor, P.J., and Soifer, B.T. 1985,
Ap. J., 292, 500.
Hendecourt, L.B. d', Allamandola, L.J., Grim, R.J.A., and Greenberg, J.M. 1986,
Astr. Ap., in press.
Hendecourt, L.B. d', and Leger, A. 1985, Fifth Conference on Matrix Isolation
Spectroscopy, July 8-12, 1985, Fontrevraud, France, p.154.
Herzberg, G. 1945, in Molecular Spectra and Molecular Structure: II Infrared and
Raman Spectra of Polyatomic Molecules (Van Nostrand).
Lacy, J.H., Baas, F., Allamandola, L.J., Persson, S.E., Mc Gregor, P.J.,
Lonsdale, C.J., Geballe, T.R., and van de Bult, C.E.P.M. 1984, Ap. J., '19, 256.
Leger, A., and Puget, J.L .. 1984, Astr. Ap. (Letters), 13'1, L5.
Leger, A., and d'Hendecourt, L.B. 1986, in preparation.
Muizon, M. de, and Habing, H.J. 1985, in Nearby Molecular Clouds,
proc. VlIIth IAU European Regional Astronomy Meeting, ed. G. Serra
(Springer-Verlag), Lecture Notes in Physics vol. 23'1, p.130.
Muizon, M. de, Geballe, T.R., d'Hendecourt, L.B., and Baas, F. 1986,
Ap. J. (Letters), in press.
Omont, A. 1986, Astr. Ap., in press.
Reed, R.I., and Tennent, A. 1971, Org. Mass Spectr., 5, 619.
Sadtler Standard Spectra 1959, Midget edition.
Silverstein, R.M., Busler, C.G., and Morrill, T.C. 1974, in Spectrometric
Identification of Organic Compounds, (New-York: Wiley).
Wade, R. 1983, SPIE Proc., 445, 47.
Wildeman, K.J., Beintema, D.A., and Wesselius, P.R. 1983,
J. British Interplanet. Soc., 36, 21.
Willner, S.P. 1984, in Galactic and Extragalactic Infrared Spectroscopy,
ed M.F. Kessler and J.P. Phillips, (Dordrecht: Reidel), p.37.
DISCUSSION
TieZens : In comparing ground-based 3
~ observations with IRAS LRS
10 ~ spectra, the disparity in beam sizes should be kept in mind. This
may in fact explain why you sometimes find a 11.3/3.3 ~m ratio which is
much larger than in our ground-based studies (Cohen et al., 1986, Ap.
J. in press) •
DISTRIBUTION OF PAR IN THE GALAXY DERIVED FROM THE IRAS DATA
J.L. Puget
Laboratoire de Physique ENS
24, Rue Lhomond 75231 Paris
(x)
ABSTRACT: The lRAS data at 12~ for extended galactic sources are very
likely to be due to the emission of PAR's. They can be used to get information on their distribution. The most striking result is their
low abundance in HII regions. There is no clear evidence for a galactic
gradient. The presence of PAR's in molecular material is clearly shown
by the lRAS data on the molecular filament in the p-Ophiuchi cloud.
The excess of emission seen in the mid infra-red in the extended
emission from our galaxy Pagot et al (1985) could be interpreted either
as emission from late type stars with extended circumstellar envelopes
(Cox et al 1986) or as emission from very small grains transiently
heated by single photons (Puget, Leger, Boulanger 1985). This mechanism
is the most likely explanation and its association with the mid infrared bands (3,28~ , 6.2~, 7.7~, 8,8~, 11,3~) for the interstellar gas
illuminated by nearly B stars is well established (Sellgren 1984, Leger
and Puget 1984, Allamandola et al 1985).
The measurement of the spectrum of the emission up to 13~ in two
reflection nebulae (NGC 2023, NGC 7023) Sellgren et al 1986) at about
l' from the exciting star allows a comparison to be made between the
energy in the bands which dominates this emission in Sellgren's data
and the flux in the 12~ IRAS band. After integrating over the lRAS
filter (~7-15~) and correction for the different spatial resolution one
finds that most of the "12~" IRAS flux can be due to the bands (part of
the 7.7~, 8.8~ and 11.3~) with an uncertainty of about 30%.
In the following we make the assumption that the extended galactic
emission seen by IRAS in its 12~ filter is dominated by the bands and
is due to reradiation by PAR's. Such an hypothesis although well supported by the argument presented above and by the finding that the mid
infra-red emission of star-burst galaxies like M82 is dominated by the
bands need to be demonstrated by a measurement of the spectrum of the
e~tended emission.
(x) also DEMIRM Observatoire de Meudon
92190 MEUDON
303
A. Uger et Ill. (ells.), Polycyclic Aro1lllltic Hydrocarbons and Astrophysics, 303-306.
© 1987 by D. Reidel Publishi"g Compa"y.
J. L. PUGET
304
The IRAS survey provides us with a sky survey in four bands
centered at 12~, 25~, 60~ and 100~. It thus allows a study of the ratio
of the emission in the bands and at long wavelength (60~ and 100~)
likely to be dominated by the radiation of standard interstellar dust
and thus gives information on the abundance of PAR in various astrophysical environments and also give constraints on the excitation mechanism.
A longitude profile of the galactic emission along the galactic equator
can be constructed for each of the four bands after substracting the
emission from interplanatery dust which dominates in the two shortest
wavelengths bands. The colours of emission can then be studied comparing the solar vicinity, the central regions of our galaxy and the anticenter direction. The findings are summarized in table 1.
Table
central regions
(molecular ring)
60/100
0.42
12/(60 + 100)
0.22 (b
=
±3)
Solar vicinity
0.37
0.25
anticenter
0.30
0.29
The opacity of dust throughout the galactic disc is not negligible in
the 12~ band. After correction for this effect, it is found that the
12/(60 + 100) color is roughly constant on large scale. This is confirmed
by the color of the emission of gas which lies a few hundred parsees
above and below the molecular ring and is seen with negligible absorption. The results are taken from Perault et al 1986. They suggest that
there is no gradient in the relative abundance of the PAR's and normal
grains. The energy density of the interstellar radiation field decreases by more than an order of magnitude between galactic radii 5 and
12 kpC. Standard dust models (see for ex. Mathis et al 1983) and Draine
and Anderson 1985) predict that this decrease in the radiation field
intensity should lead to a steep decrease of the 60~/100~ ratio by a
factor of 5 which is not observed. The implication is that a significant
fraction of the emission in the 60~ band is due to fluctuations of temperature and not radiation by large grains in equilibrium as noticed by
Draine and Anderson 1985. Altogether about one third of the energy
radiated by our galaxy in the mid and far infra-red is due to temperature fluctuations of very small particles. This is a surprisingly large
fraction which says that the same fraction of the absorbing part of the
interstellar extinction in the visible and UV is due to these particles.
The absence of any obvious signature of this absorption other
than the 2200
bump which has been associated with PAR (Donn 1968) is
surprising.
To get a more detailed information on these questions one can
look at one of the nearest molecular complex, the one associated with
the star
RHO Ophinicus. This region contains a dense molecular core,
IDolecular f~laments~ reflection nebulae and HII regions around B stars.
The finding concerning the [12/(60 + 100)] color ratio is summarized
in table 2
A
305
DISTRIBUTION OF PAH IN THE GALAXY DERIVED FROM THE IRAS DATA
Molecular filament
Reflection nebular
(HD 147 889)
HII region (as co)
12
60 - 100
0.35
0.12
<
0.05
These results are taken from a detailed analysis of this question by
Ryter, Puget and Perault 1986. The very low .12~ emission in HII
regions is an agreement with the results by Gosh et al (these proceedings). It shows that PAH are distroyed in HII regions either by
interaction with ionized gas or by the ionizing uv radiation. It should
be noted that for a Bl star like a Sco the ionizing flux is less than
10% of the total flux.
The rather high value for the molecular filament shows that PAH's
are quite abundant in molecula materiel and is consistent with a gas
phase origin of PAH's.
In the direction of the dense molecular core the ratio decreases
but the 12 ~ absorption through the core is likely to be large (~ 50
magnitude of visual extinction). At 60 and 100~ we see the emission of
both faces of the cloud but at 12~ we see only this face looking towards
us.
The general trend emerging from these results is that the abundance
of PAH is larger in molecular material than in more diffuse gas. The
molecular filament is heated mostly by the nearby B stars so no definite
conclusions can be drawn for the excitation mechanism. It should be
remembered than B stars are. the dominant energy source for the infrared emission from our galaxy (Puget 1985) and thus one should look
well isolated molecular clouds heated by the diffuse radiation field
to get an answer on the relative contribution of UV visible and near
infra-red (1~) to the excitation of the very small particles
REFERENCES:
Allamandola L.J., Tielens A.G.M and Barker J.R. 1985, Astrophys.
J. Letters 290 L 25
Cox P., Krugel E. and Mezger P.G. 1986, Astron. and Astrophys. 155 380
Dann B. 1968, Astrophys. J. Letters 152 L 129.
Draine B.T. and Anderson N. 1985, Astrophys. J. 292
494.
Leger A. and Puget J.L. 1984, Astron. and Astrophys. 137 L5.
Mathis J.S., Mezger P.G. and Panagia N. 1983, Astron and Astrophys.
128
212.
Pajot F., Boisse P., Gispert R., Lamarre J.M., Puget J.L. and Serra G.
1986, Astron and Astrophys. 157 393.
Perault M., Boulanger F., Falgarone E. and Puget J.L. 1986, in preparation.
J. L. PUGET
306
Puget J.L. 1985, in Birth and infraray of Stars ed. R. Lucas, A. Omont
and R. Stora, Elsevier Science Pub.
Puget J.L., Leger A. and Boulanger F. 1985, Astron and Astrophys. 142
L19
Ryter C., Puget J.L. and Perault M., 1986, in preparation.
Sellgren K.
1984, Astrophys. J. 277
623
Sellgren,K.,Allamandola L.J., Breyman J.D., Werner N.W. and Wooder D.H.
1986, Astrophys. J. in press.
DISCUSSION
S .K. Ghosh
I have a remark to make. The ] 2 t'm deficiency you find
in the ionized regions, has in fact been established by us (our Poster
paper) for a large sample (328) of H II regions.
Infrared features in extragalactic objects.
P.F. Roche
Department of Physics & Ast.ronomy,
University college London,
Gower st.reet, London NClE 6B'l'.
ABSTRACT. Infrared spectroscopy provides a powerful tool for
invest.igating t.he properties of dust in a variety of ast.rophysical
sources. Observat.ions of galaxy nuclei in the lo-,.un at.mospheric window
have revealed t.hat active and non-active nuclei have dist.inctly
different. spect.ral properties. The lat.t.er have spect.ra dominat.ed by t.he
family of unident.ified infrared emission bands Whilst. the active nuclei
show lit.tle evidence of spectral structure t.hat could be attributed to
emission from dust grains. This separation in properties may indicate
that. the grains responsible for t.he infrared band emission are destroyed
by t.he hard phot.on flux emitted by the active nuclei.
1.
Introduction
In discussing t.he infrared spectral properties of galaxies, a very
useful distinction can be made bet.ween those galaxies t.hat show evidence
of a strong non-thermal source in their nuclei and those that. are
apparently comprised of normal stars. The former, termed 'active'
galaxies, have non-thermal energy distributions extending to high
energies in t.he far-uv and X-ray regions, produced by a source of
lumiosity ot.her than nuclear processing in stars. They include galaxies
classed as Quasars and seyferts Which show evidence of high excitation
and high velocity gas and probably derive most. of t.heir energy from
accretion by massive compact objects in t.heir nuclei. The' normal'
galaxies have spectra that can be eJq)lained ent.irely in terms of st.ellar
populations although many of t.he most. luminous examples cont.ain a large
number of very hot young stars. Many galaxies emit most of their
luminosit.y in the infrared (~-lOOO,.un) region of their spectrum, either
as synchrotron radiation from the most luminous active nuclei, or as
thermal reradiation of short wavelength photons absorbed by dust grains.
~pect.roscopy allows us to probe the emission mechani811 and, Where the
emission is t.hermal radiation from dust, investigate the chemistry and
excitation of the grains.
307
A. Uger et af. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 307-316.
© 1987 by D. Reidel Publishing Company.
308
P. F. ROCHE
The IRIaS survey has allowed a systematic survey of the infrared
properties of galaxies. Amongst the results that have emerged is
confirmation of conclusions from ground-based surveys (e.g. Rieke &
Lebofsky 1978, Scoville et al 1983) that galaxies that show little
evidence of gas, dust, star-formation or non-thermal activity in their
optical properties are relatively weak in the IR. Normal elliptical
galaxies show little evidence of excess infrared emission over the
extrapolated stellar photospheres after allowing for a contribution from
circumstellar shells around late type stars. Those ellipticals that are
bright in the IR are usually peculiar and often show evidence of strong
radio emission and/or active nuclei. On the other hand, spiral galaxies
generally show infrared emission from the galactic discs Which probably
arises mainly from spiral arm HII regions similar to the HII region
complexes seen in the Galaxy. This emission has a characteristic colour
temperature of about 30 K, again similar to typical temperatures. in
Galactic HII regions.
The most luminous 'normal' galaxies in the infrared are those that
show evidence of vigorous star formation in their nuclear regions.
These objects, termed 'starburst' galaxies, contain nuclear HII regions
photoionized by hot stars Which produce high infrared fluxes from dust
grains in and close to the ionized regions. The efficiency of star
formation in the starburst nuclei can be very high, producing infrared
luminosities of up to 1012 La in the central hundreds of parsecs (up to
=2 kpc) of the galaxies, and implying that up to 1010 stars have been
.formed recently in the central regions of the most extreme examp: 1S. To
investigate the nature of the in.frared emission, we can compare the
Observed properties of the galaxy nuclei with known sources in the
Galaxy. However, a direct comparison is difficult because even the
small (typically 5 arcsec) beams used by infrared spectrometers
correspond to spatial scales o.f 100-1000pc in external galaxies,
depending on their distance. The same instruments have beam sizes ~a pc
in Galactic sources, so that only the bright warm cores of HII regions
are usually measured in the Galaxy. Normal galaxies that do not have
luminous nucle~HII regions, and all but the brightest extragalactic
disc HII regions are too faint for 8-13~ spectroscopy with existing
instruments.
spectra at 8-13~ of more than 50 galaxy nuclei have been secured
and many have been published (e.g. Aitken & Roche 1985, Roche & Aitken
1985a and references therein). The division of the galaxies into two
classes, active or starburst, is reflected in their 8-13~ spectra. The
starburst galaxies generally have spectra dominated by the well known,
but. as yet unidentified, infrared (UIR) emission features at 11.25, 8.65
and 7. 7jAD1 seen in a variety of Galactic sources together with strong
ionic line emission from the 12.8jAD1 [NeII] fin_structure line. By
contrast, these features are seen only rarely in the active nuclei Which
instead have 8-13~ spectra well represented by a power law with, in
some cases, evidence for silicate absoxpt:ion.
309
INFRARED FEATURES IN EXTRAGALACfIC OBJECfS
u
+
IONIZATION
FRONT
8
10
').
Fig 1.
2.
11
12
13
The different 8-13j.111l spectral properties found in three regions
of the Orion nebula.
Dust in the Galaxy
To underst.and the observed spectral characteristics of
extragalactic objects, we can ccxapare them with sources in our Galaxy
which are spatially resolved and whose evolutionary status and
environment are reasonably well underst.ood. Of the Galactic HII regions
that have been studied in the infrared, the Orion nebula has received
the most. attention. A host of different phenomena has been unearthed,
but the most prominent. identifiable regions by their mid-infrared
spectral properties (fig 1) are I
1. The region of hot dust near the Trapezium heated by the exciting
stars of the nebula. This shows a prominent silicate emiSSion band
(Forrest et al 1975) which is characteristic of warm dust in
molecular clouds and the int.erstellar medium and circumstellar
shells around oxygen rich stars.
2. BNICL a region of star fo~tion where the nascent. stars have not
yet emerged from the molecular cloud in which they were formed, the
spectrum is dominated by a deep silicate absorption band produced by
cold molecular cloud _terial bet.ween the ionized nebula and the
young stars. (Gillett & Forrest 1973)
3. The Orion bar, an ionization front at the south-eastern edge of
P. F. ROCHE
310
the nebula where the UIR bands peak (Becklin et al 1976, Aitken et
al 1979).
The mid-IR emission from the whole Orion nebula may be represented
a weighted sum of these three components, together with narrow
emission lines from ionized gas and shock-excited molecules. However,
the contribution from each dust component to the total emission is not
completely clear and, in particular, the amount of low surface
brightness emission in the UIR bands from the edges of the HII region
could be quite large. HII regions in different evolutionary stages will
also have different fractions of these components. This should be borne
in mind when comparing HII regions in the Galactic disc with extragalactic objects. In addition to the observations of the Orion bar
(Aitken et al 1979), spatial studies of NGC 7027, a high excitation
planetary nebula, clearly show that the 11.3~ emission feature arises
from just outside the ionized region (Aitken & Roche 1983). These
results suggest that the grains responsible for the UIR band emission
are destroyed in high excitation regions, but can oo-exist with neutral
and, perhaps, low-excitation ionized gas.
by
The most natural object to compare with the nuclei of external
galaxies is the centre of our own Galaxy. In this case, observations
are hampered by the large extinction due to the dust lying in the
interstellar medium between the sun and the Galactic centre. This
produces an extinction of ~30 mag in the visible and a deep silicate
absorption feature at 9.7~. The very compact non-thermal radio source
right at the centre of the Galaxy, Sgr A-, is not a strong mid-IR
emitter, but rather the flux in the central parsec is dominated by a
number of compact lUI regions. However, it is likely that much of the
radiation emitted from the central parsec is absorbed by dust several pc
away. Indeed, most of I:he cenl:ral 4 pc of I:he Galaxy has a very low
density and is transparenl: to the radiation emitted from the cenl:re,
which insl:ead emerges in the IR on being absorbed by a ring of dusty
material around the Galacl:ic centre (see Gatley 1984). After correction
for the interstellar absorption, the 8-l3~ spectrum shows a prominent
Silicate emiSSion band, I:ogether with emiSSion in the l2.8~ [Nell]
fine-structure line (e.g. Roche & Aitken 1985b). Here, I:he 10~
emission is dominated by warm silicate grains, but observations by
Gatley (1983) suggest that the 3.3~ UIR band is present at a relatively
low flux level throughout the central couple of parsecs of the Galaxy,
so that the emiSSion from the UIR bands could be significant in a large
beam; again, the spectrometer beams sample only the central core whereas
observations of exl:ragalacl:ic nuclei cover a volume I:ypically ~104
larger.
3.
Starburst galaxies.
Sl:arbursl: galaxies show evidence of sl:rong HII region acl:ivil:y in
their nuclei. In the opl:ical, the spectra show brighl: ionic line
emission with excil:ation indical:ive of photoionization by hot stars.
311
INFRARED FEATURES IN EXTRAGALACTIC OBJECTS
II
tlit
1I I I
.
t I"
3C 273
NGC
• t •••• " ,
,
NGC
".111 '1Itft'I'"
+
..
.:
l til I
.s
'\
I I I11 tit II
l
'- -_~t I~I
Pig 2b
••
,
".
t t'"
•••••••
I
MCG
253
4151
I
'
-5-.23-16
tf
f! t!
NGC
1808
•\
,t
","
lc7:t~G:-t~_', : -;" : ~;-j ~o--.::'llJ.ull~: ~:':~
t_! t;:-tl_l1-,-;:':1t;:-tt_1t-:-:-,;tf;:-"'_I
Fig 2a
I
"f'
tltll
! j
NGC
t
.'
", , ••• '_'
t '.
u
,t···t·
..
ttt ,tt
5253
~
----'
A
spectra of 4 statburst galaxies. NGC 5253 is unusual in
showing a strong 10.5~ [SIV] line and no evidence of the UIR
bands or t.he 12.8~ [Nell] emission seen in typical HII region
galaxies
Spectra of the quasar 3C 273, NGC 4151 (seyfert. 1.5),
1CG-5-23-16 (seyfert. 2) and NGC 4418, an lRAS-selected galaxy
with an extraordinarily deep silicate absorption band.
8-13~
The widths of the emission lines are generally <250 laas-1 , IlUch SlIIIlller
than those seen in active galaxies, and presUlllilbly arise as a result of
galactic rotation and turbulent motions in the star-forming regions.
The presence of dust in the starburst nuclei is often apparent from
reddening of the emission lines emitted by the ionized gas and the
presence of dust lanes running across the bright emission knots seen in
optical photographs. As is the case in Galactic HII regions, we would
expect the starburst galaxy nuclei to be powerful emitters in the
infrared.
The starburst galaxies are typified by the archetypes NGC 253 and
M82, whose spectra show strong emission in the narrow unidentified
emission features and the [Nell] line together with B<m8 silicate
absorption (Gillett et al 1975). Observations in the 3-8~ region have
shown that the other ID8IIIbers of the family of UIR bands are also present
(Willner et al 1977). some 90' of the nuclear HII region galaxies show
remukably silllilar spectra (fig 2a), over a range in 10,... l\8inosity of
5 x 107 to 1010 I.e. The _ _ ured 10~ flux in these galaxies is
312
P. F. ROCHE
usually aperture dependent, implying that the emitting regions are
extended on a scale of a few hundred parsecS. Spatial studies of the
8-13~ emission have been carried out over the central disc of M82, and
they show that although there may some differences in detail from
position to position, the spectra are remarkably similar throughout the
whole 300pc extent of the bright IR emission (Jones & RodriguezEspinosa 1984). The emission in the narrow bands can account for a few
percent of the total luminosity of these galaxies. The equivalent
widths of the emission features in the starburst galaxies are generally
higher than in Galactic sources and dominate the spectra to an extent
that the silicate emission signature seen in Galactic HII regions is
swamped, if present at all. Although the apparent minimum near 10~ in
the 8-13~ spectra of starburst galaxies is largely a result of the
emission structure at 7.7, 8.6 and 11. 3~, there is probably a
contribution from absorption by cool silicate grains in the galaxy
nuclei. The relationship between the equivalent width of the UIR
emission features and the depth of the apparent minimum in the 8-13~
spectra has been investigated in relatively unObscured Galactic sources.
Applying that correlation to the galaxy spectra often indicates that the
galaxies have deeper minima, implying that there is a silicate
absorption component (Aitken & Roche 1984). M82 has been studied in
detail by Rieke et al (1980) who find that the hydrogen Brackett
recombination line ratios indicate an extinction of Ay =25 mag, which
corresponds to the deep Silicate absorption band inferred from the
8-13j.1111 spectrum. The origin and excitation of the features are
discussed at length in this volume, but by analogy with HII regions in
our Galaxy (Aitken et al 1979), it is likely that the grains are excited
by blue photons escaping from the ionized regions in the nuclear HII
region complexes. With the vigorous star formation in the nuclei of
these galaxies, it is p~le that much of the central region is bathed
in radiation with A> 912 ~ If non-equilibrium emission from very
small grains is invoked as the excitation mechanism (Bellgren 1984),
these non-ionizing photons will be efficient in producing the
unidentified bands.
Recently, observations in the 2~ region have shown that many of
the starburst galaxies also have powerful emission in the vibrational
bands of molecular hydrogen and the line ratios indicate that the ~ is
shock excited. The ratio of H2 emission to far-IR emission is similar
to that seen in Orion, providing further evidence that the IR emission
mechanism is similar to that seen in Galactic HII regions (Joseph 1986).
TWo of the HII region galaxies, ROC 5253 (Aitken et al 1982; fig
2a) and II ZW 40 (unpublished data) have quite different spectra from
the remainder. They display the 10.52~ [SIV] line indicating higher
excitation gas than the other starburst nuclei, and the narrow emission
features are not detected. The high excitation, implying very hot and
hence young stars, together with the compact size at 10~ suggests that
the IR emitting regions in these two nuclei are very young. It DIlly be
that we are seeing a very early stage in a starburst nucleus where the
IR emission is dominated by one central region, as opposed to the more
INFRARED FEATURES IN EXTRAGALACTIC OBJECTS
313
extended emission in the other galaxies with nuclear HII regions. we
might then expect the dust emisSion to be dominated by heating in the
central core rather than diffuse emission outside the ionized region.
The fraction of staxburst galaxy nuclei dominated by the UIR bands
is very high (90" c.f. "'30" of Galactic planetary nebulae), suggesting
that the emission in these galaxies is not strongly dependent upon
parameters such as the details of the staxburst mechanism or the
environment provided by the host galaxy. It appears that the only
requirements to produce a typical mid-IR starburst spectrum dominated by
the UIR bands are a vigorous nuclear starburst and the associated gas
and dust. The IR emission may be related to the IR cirrus detected by
1MB in our Galaxy, but very efficiently excited by the abundant soft tJV
photons escaping fX'Olll the star forming regions in the starburst nuclei.
4.
Active galalc:ies.
The signature of the active nucleus in a galaxy can be lIIiUlifest in
several different ways, not all of which will be evident in any one
object. The source of luminosity is generally thought to be accretion
of material by a massive compoct object at the centre of the host
galaxy. The observational classification of a galaxy as possessing an
active nucleus rests on the followingl variability in the flux of line
andVor continuum emission, ionized gas showing high excitation species
such as HeII, [l"eVII] and [NeV] which cannot be produced by normal
population I stars but suggest photoionization by a power-law continuum
(P..,a: vG), high velocities (up to =-104 kms-l ) in the emission lines,
radio maps Which often show jets extending from a compact nucleus, and
strong X-ray emission. Par a review of the properties of active nuclei
see Hazard & Mitton (~979).
The active ga~axies encompass a huge range in luminosity from the
quasars to the low luminosity seyfert 2 galaxies, and generally have
compact IR sources centred on the optical active nuclei. The 8-~31A1D
spectra of almost all the active nuclei are broadly similar with little
evidence of spectral structure that could be attributed to dust emission
features (fig 2b). The gsos and seyfert ~ galaxies in our sample mostly
have very smooth ~01AlD spectra, and only 2 out of 13 show significant
departures from a power law fit to their 8-13JA111 spectra. The two
exceptions are NGC 7469 and 111m 231, both of which are rather atypical
seyfert 1 galaxies. Spatial studies of ROC 7469 have been carried out
by CUtri et al (H84) who found that the 3. 3JA111 UIR band is extended over
a region "'2.5 kpc in size. It is liJc:ely that the 3.31A1D feature, and
presUlllllb1y the other members of the UIR family, arise in an extended
star fODlllltion region near the centre of the galaxy. and are not
directly associated with the seyfert nucleus. several of the lower
luminosity seyfert 2 nuclei show evidence for silicate absorption, but
again the absence of the UIR emission bands and fine-structure ioniC
line emission that are so prominent in the HII region galaxies is
remarkable.
P. F. ROCHE
314
The mechanism responsible for the IR continuum in active galaxies
has been rather controversial. However, variability detected throughout
the IR and millimetre regions in 3C 273 (Clegg et al 1983) has
convincingly demonstrated a non-thermal synchrotron origin for the
emission in that object, and presumably other radio-loud luminous active
nuclei. In less powerful nuclei, the picture is not as clear, and it is
likely that both thermal and non-thermal processes are important.
With a low-luminosity active nucleus, contamination from HII
regions in the central regions of the galaxy may be a problem. This can
be seen clearly in the nucleus of the nearby barred spiral NGC 1365
where the 8-13j.1l11 spectrum of the seyfert nucleus is smooth, and similar
to other active nuclei, but the emission from two optically bright HII
regions only ==8 arcsecs (==1 kpc) away is strongly featured and typical
of starburst galaxies (fig 3). If this galaxy were located at a greater
distance, these separate components would not be resolvable, and the
contributions from the HII regions within the central few hundred
parsecs would dominate the 10j.llll emission. Clearly, high spatial
resolution is important, and the characteristic mid-IR signature of a
low-luminosity active nucleus may not be detectable in distant galaxies.
NGG 1365
jjitllljjljllttlttllllllljll
NUCLEUS (xlO)
8
Fig 3.
9
10
').
11
12
13
spectra of the seyfert nucleus and HI! regions near the centre
of NGC 1365.
There is no evidence of the silicate emission feature which is
almost ubiquitous in M giants and HI! regions in our Galaxy. If the
10j.llll flux is produced by thermal emission from dust in the active
nuclei, the heating mechanism or grain population must be such that the
silicate grains do not emit strongly, although the approximately power
law IR flux distribution requires emission from grains over a large
temperature range. The remarkable absence of the narrow emission
features in the active nuclei can be explained in terms of excitation of
small grains. The fact that the 11.25j.1l11 emission feature peaks outside
the ionized regions in Orion and NGC 7027 suggests that the grains
INFRARED FEATURES IN EXTRAGALACTIC OBJECTS
315
producing it are destroyed by energetic photons. The active nuclei are
characterised by strong non-thermal emission extending into the Far-UV
and X-ray regions and t:his hard photon continuum will quickly destroy
small grains by evaporation. However, because the grain absorption
cross-section falls with increasing energy, larger grains, Which require
more energetiC photons to beat them above the critical evaporation
temperature, will survive. A consequence of this is t.hat the regions
around the act.ive nuclei should have a cut off in the grain size
distribution. In the active galaxies, a substantial amount of the
luminosity is emitted by a Single compact source, as opposed to the
distributed HII regions that form the starburst nuclei, so that too will
affect the radiation field emerging from the nuclei and available to
excite dust grains in the central regions of the galaxies.
The lower luminosity active nuclei with 8-13pm spect.ra well fit by
power laws tend to have steeper spectral indices than their more
luminous peers. This often shows as a break in the shape of the IR
spectrum between 3-10pm, and is presumably produced by emission from a
warm dust component. overall, the 10pm spectra of active galaxies
suggest two main components, namely the non-thermal energy source from
the active nucleus and radiation from dust in the nuclear region. In
the highest luminosity objects, the non-thermal source dominates with
dust emission becoming more important with decreasing luminOSity.
IRAS galaxies
The lRAS satellite has identified several thousand IR-bright
galaxies with no a priori selection effects. A start has been made in
investigating Whether the lRAS-selected galaxies show up any differences
in their 8-l3pm spectral properties. Because they are luminous in the
IR, we might expect many of the IRAS galaxies to be dusty and, from the
small saD\Ple available (6 galaxies), i t appears that a substantial
number may have deeper silicate absorption than optically-selected
galaxies (in preparation). Indeed, some galaxies may be so heavily
Obscured that the activity giving rise to the powerful IR emission is
not discernable in the visible. The 8-13pm spectrum of the most extreme
of these galaxies, NGC 4418, is shown in figure 2b Where the depth of
the silicate absorptd-on feature T9.7 =-7 corresponds to ~ 100. Optical
spectra (3000-10000 A) Obtained at the IlAT reveal little Sign of the
luminous IR source but convey the impression of a rather unremarkable
galaxy. In this galaxy, the IR luminosity is 7 x 1010 Le and all this
energy lies behind a very large column of obscuring dust. Although the
exact reasons for the difference between the active and non-active
nuclei in the 8-13J.1l11 region may not yet be fully understood, we can use
the Observed properties to investigate the underlying emission source in
highly obscured galaxies. From the non~etection of ionic and dust
emission structure in the 8-13pm spectrum, together with the relatively
blue XRAS colours and compact size of the XR emitting source, it is
likely that ROC 4418 harbours an extremely heavily Obscured Seyfert
nucleus (Roche et al 1986).
316
P. F. ROCHE
Referencesr
Aitken, O.K. & Roche, P.P., 1983. M.N.R.A.S. , 202, 1233.
Aitken, O.K. & Roche, P.P., 1984. M.N.R.A.S. , 208, 751Aitken, O.K. & Roche, P.P., 1985. M.N.R.A.S., 213, 777.
Aitken, O.K., Roche, P.P., Allen, M.C. & PhiUips, M.M. , 1982.
M.N.R.A.S., 199, 3LP.
Aitken, O.K., Roche, P.P., Spenser, P.M. & Jones, B., 1979.
Astr. AstrophyS. 76, 60.
BecJtlin, E.E., Beckwith, S., Gatley, I., Matthews, K., Neugebauer, G.,
Sarazin, C. & Nemer, M.W., 1976. Ap. J., 207, 770.
Clegg, P.E., Gear, W.K., Ade, P.A.R., Robson, E.I., Smith, M.G., Molt,
I.G., Radostitz, J.V., Glaccum, W., Harper, O.A. & IDw, P.J.,
Ap.J., 273, 58.
CUtri, R.M., Rudy, R.J., Rieke, G.B., TOkunaga, A.T. & Willner, S.P.,
1984. Ap.J., 280, 521.
Forrest, W.J., Gillett, F.C. & stein, W.A., 1975. Ap.J., 195, 423.
Gatley, I.G., 1984. in Galactic and EXtragalactic Infrared spectroscopy,
p 351. XVI £SlAB Symp, ads M.F. Kessler & J.P. PhiUips.
Gatley, I.G., 1984. in Proc. WOrkshop on Infrared Spectra of Interstellar
Dust, p 118, ads R.O. WOlstonecraft & J.M. Greenberg.
Gillett, F.C. & Forrest, W.J., 1973. Ap. J., 179, 483.
Gillett, F.C., Kleinmann, O.E., Wright, E.L. & capps, R.W., 1975.
Ap.J., 198, L65.
Hazard, C. & Mitton,S., 1979. "Active galactic nuclei", cambridge.
Jones, B. & Rodriguez-Espinosa, J.M., 1984. Ap. J., 285, 580.
Joseph, R.O., 1986. Proc. 1st lRAS conference "Light on Dark Matter"
Rieke, G.B., Lebofsky, M.J., Thompson, R.I., Low, P.J. & TOkunaga, A.T.,
1980. Ap. J., 238, 24.
Rieke, G.B. & Lebofsky, M.J., 1978. Ap. J., 220, La7.
Roche, P.P. & Aitken, O.K., 1985. M.N.R.A.S., 213, 789.
Roche, P.P. & Aitken, O.K., 1985. M.N.R.A.S., 215, 425.
Roche, P.P., Aitken, O.K., s.ith, C.B. & James, S.O., 1986. M.N.R.A.S.,
218, 19P.
SCOville, N.Z., BecJtlin, E.E., Young, J.S. & capps, R.W., 1983. Ap.J.,
271, 512.
Sellgren, K., 1984. Ap.J., 277, 623
Willner, S.P., SOifer, B.T., Russell, R.W., Joyce, R.R. & Gillett, P.C.,
1977. Ap. J., 217, L121.
Discussion
Duley:
Do you have an estimate of the electron temperature in those
objects where the UIR features are weak or missing ?
Answer : II Zw 40 (and probably NGe 5253) is certainly metal poor, and
probably has a somewhat higher Te than the 'standard' 104 K, of the
order 13,000 K or so. I think that the values of Te ~n both the broadline and narrow-line regions of Seyferts are generally thought to be
about 104 K.
VERY SMALL GRAINS IN SPIRAL GALAXIES
S.K.Ghosh 1 ,2 & S.Drapatz 1
1 Max-Planck-Institut fiir extraterrestrische Physik
D-8046, Garching bei Miinchen, West Germany
2Tata Institute of Fundamental Research
Homi Bhabha Road, Bombay-400005, India
ABSTRACT. The existence of a hot (Td ~ 2000 K) dust component in normal
spiral galaxies has been inferred from a statistical study of their mid- and farinfrared emission. Evidences have been presented for this dust emission to be of
diffuse interstellar rather than circumstellar origin, in agreement with the picture
of transient heating of Very Small Grains (size ~ 10 A) by UV photons of the
interstellar radiation field, i.e. the mid- and far-infrared emissions are correlated.
1. INFRARED CHARACTERISTICS OF SPIRAL GALAXIES
1.1 Spiral galaxies :
Based on the four band !RAS survey experiment (Neugebauer et al 1984)
measurements, two parameters Q: (a measure of FIR/MIR emission) and"'f have
been defined as :
+ PlOO )
I
( P25 + P60
Q: = og10
P
12
"'f
= I og10 (
P60 + PlOO)
P.
25
where P). is the power received in the ~ micron band.
A comparison of the Q: and "'f distributions have been made (see e.g. Q: in
Fig.l) for the two samples of (i) spiral galaxies (sample size = 215) selected from
the ESO/Uppsala Catalogue (Lauberts 1982) and (ii) Galactic HII regions (sample
size = 328) selected from the Bonn and Parkes radio continuum surveys (Altenhoff
et al1979, Haynes et al1979) which have been detected in all the four !RAS bands.
Sources of both these samples have been demonstrated to be effectively smaller
than the mAS beams.
317
A. Leger et al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 317-322.
© 1987 by D. Reidel Publishing Company.
S. K. GHOSH AND S. DRAPATZ
318
From Fig.1 one notices that the a distributions are very narrow (halfwidth ""a
factor 3) and (FIR/MIR)HIl "" 5(FIR/MIR).piral. "" 20. The'Y distributions are
similar. This implies that in spiral galaxies in general, there exists a component
of dust hotter than the average dust temperature in HI! regions.
0.3
...
en
~
80.2
..Eo.1
....co
en
SPIRAL GALAXIES
f1-GALACTIC HII
::
REGIONS
r"
co
I
I
.......
I
r,I
~ ..
!
c
,
'"1
,
I
,
L,
0
-to
-0.5
Tfort' I~H/'"
Fig.l The distribution
1401
0/ a, which
1001
is a measure
80K
25
3.0
ALPHA-
0/ FIR to MIR emission ratio.
1.2 Our Galaxy :
The longitude distribution of the 12, 25, 60 and 100 I'm Galactic Emission
(GE) of the galactic plane (Ibl ~ 5°) has been computed from the IRAS Low
Resolution All Sky Images (LORASI) (Beichman et al 1985) by subtracting the
Zodiacal Emission (ZE). The ZE has been estimated locally by taking cuts across
the galactic plane and fitting orthogonal Legendre polynomials to the off-the plane
(Ibl ~ 6°) data points with appropriate latitude and longitude binning (~b =
10 , ~l = 10°). Resulting GE at 12 and 60 I'm are shown in the Figs.2 & 3
respectively.
The average values of a and 'Y for the Galaxy have been computed to be 0.68
± 0.08 {i.e. (FIR/MIR) = 4.8} and 0.99 ± 0.08 respectively which lie at the peaks
of the a and 'Y distributions for the sample of spiral galaxies.
This shows that our Galaxy is a typical spiral and hence in the next section
it has been studied further to obtain general properties of spirals. Specifically the
sources with (FIR/MIR) 5 5 have been looked for.
The LORASI results are in good agreement with previous AFGL surveys of
the MIR emission of the Galaxy and imply a luminosity LGalas,,{MIR} .... 109 Le.
319
VERY SMALL GRAINS IN SPIRAL GALAXIES
3.
f
:
r'
~
"I
Ll
~2
.a:-
1
ri
~1
...
I,
~
2
r,'
".
~
--
&D,....ID
-SALam: EMISSIDII
""IISS·,
--ZlmACAL EMISSIIIIIZE
"i"..
...........f' 12 pm BAlD
III
i!h
I II
e
i-
-
...
8
Fig.9 The galactic longitude
distribution 0/ the
60 p.m emission.
Fig.1! The galactic longitude
distribution 0/ the
11! p.m emission.
2. HOT DUST IN "DISCRETE" SOURCES IN OUR GALAXY
2.1 Young objects :
(i) HIT regions : They have Fm/Mm ;?: IS, which means that they do not
have sufficient hot dust to explain the MIR emission of the Galaxy (see Fig.l).
(ii) Protostars : The fact that the protostar IRS-l BS (Beichman et al1984)
has a value of Fm/Mm ::5 1 (although the MIR emission mechanism is unclear),
makes such objects attractive for explaining the Galactic Mm emission.
However, using the mass and Mm luminosity of mS-l BS, the contribution
of such protostars in the Galaxy has been estimated to be :
107(!~;) )Le '" 10-2 LG,al,lI:II(MIR)
Lprotost,m(MIR) '"
assuming an universal IMFi the star formation rate '"" 10 Mellr-1 and.,. '" lifetime
in 'IRS1-BS like' phase,"" 105 yrs.
(iii) Energetic Molecular Outflow sources: The source L1551-mS 5 (Emerson
et al1984) has (Fm/MIR) '" 4. But, the estimated total MIR luminosity of all
the EMO's in the Galaxy is:
LEMo(MIR) '"" (
5
X
10'
"
)Le = (
5
X
10- 5
"
)LGfl'flzlI(MIR)
from the observed number density of EMO's in the solar neighbourhood (Lada
1985), and assuming LlS51-mSS to be a typical EMO. To get LEMo(MIR) '""
320
S. K. GHOSH AND S. DRAPATZ
LGal(MIR), one needs "1 (detection pfficiency of EMO's within 0.5kpc) ::; 10- 4
which is unrealistic.
(iv) T-Tauri stars: These objects have (FIR/MIR) "" 1.5, but, byextrapolating from the observed number of HII regions (i.e. O/B stars) and assuming an
universal IMF, one gets the T-Tauri contribution to be:
where f ,.., 0.1 - 0.5 is the detection efficiency of radio continuum surveys for HII
regions, TTTS ,.., 105 yrs & THII "" 106 yrs.
(v) Globules: They have (FIR/MIR) ?:10 and hence are not efficient contributors to the Galactic MIR emission.
2.2 Evolved objects :
2.2.1 In our Galaxy :
Contribution of all discrete sources in the Galaxy, along certain lines of sight
(e.g. 90° ::; III ::; 100°, WII ::; 5°) where there is no source confusion (as evident
from the differential source counts), to the MIR brightness is :
(obtained from the IRAS Point Source Catalog and LORASI). This shows that
Galactic MIR emission is not mainly from discrete sources (evolved stars) and
most of the LGal(MIR) originates in the diffuse interstellar medium.
In other galaxies :
Most of the FIR emission of spirals are energized by young objects (T ,.., 106
yrs). If the MIR emission originates from evolved objects like the circumstellar
shells of asymptotic giant branch stars or OH/IR stars (T ?: 108 yrs) then the ratio
(FIR/MIR) must be sensitive to the past star formation rate. However, a statistical
test on a sample of Interacting/ Merging galaxies ( having star bursts as a result of
interaction, i.e. change in the recent SFR) show the same (FIR/MIR) distribution
as normal spiral galaxies (compare Figs.1 & 4). The sample of interacting galaxies
(sample size = 86) was selected from the Morphological Catalogue of Galaxies.
Thus, the MIR emission of spirals does not mainly originate from the evolved
stars and the FIR and MIR emissions are correlated.
2.2~2
321
VERY SMALL GRAINS IN SPIRAL GALAXIES
~~---------------------------,
II1ERACTlIIG GALAXIES
(MC &)
--
Ii
OJ
~
&&.
Tfur £ (A" 1/~ 1401(
8UK
Fig.-I The distribution 01 a, lor the sample 01 interacting galaries.
3. HOT DUST IN DIFFUSE INTERSTELLAR MEDIUM
We conclude from our statistical comparisons and discussions above, that,
there is no substantial contribution of the "discrete" sources to the Mm emission
of our Galaxy and spiral galaxies in general and the emission has to come from
the diffuse interstellar medium.
The remaining possible source of Mm emission is the Very Small Grains
(VSG), which are discussed in the recent literature ( Leger && Puget 1984, Puget
et aI1985). They are of typical size ,... 10 A. and emit mid-infrared radiation as a
result of transient heating by energetic photons of the interstellar radiation field
(ISRF).
The heating of the VSG's by the ISRF is energetically feasible as :
LGtll(ISRF) ,...10LGtll(MIR)
A major contribution to the LGtll(ISRF) is from the OIB stars.
The observation that :
is consistent with the predicted evaporation of the VSG'. in all regions (Puget et
al1985).
S. K. GHOSH AND S. DRAPATZ
322
The VSG picture is also consistent with the observed insensitivity of the
(FIR/MIR) ratio to the possible non-universal initial mass function (IMF) in the
star burst galaxies (Rieke et al 1980, Kronberg et aI1985).
A more detailed version of the present study is in preparation.
REFERENCES:
Altenhoff, W.J. et al.: 1979, Astr. Ap.(Suppl.), 35,23.
Beichman, C.A. et al.: 1984, Ap.J.(Letters), 278, L45.
Beichman, C.A. et al.: 1985, [RAS Catalogs and Atlases Explanatory Supplement.
Emerson, J.P. et al.: 1984, Ap.J.(Letters), 278, L49.
Haynes, R.F. et al.: 1979, Aust.J.Phys.Ap.(Suppl.) No. 48.
Kronberg, K.K. et al.: 1985, Ap.J., 291, 693.
Lada, C.J.: 1985, Ann. Rev. Astr. Ap., 23,267.
Lauberts, A.: 1982, The ESO/Uppsala Survey 0/ the ESO(B) Atlas.
Leger, A. and Puget, J.L.: 1984, Astr. Ap., 137, L5.
Neugebauer, G. et al.: 1984, Ap.J.(Letters), 278, L1.
Puget, J.L. et al.: 1985, Astr. Ap., 142, L19.
Rieke, G.H. et al.: 1980, Ap.J., 238, 24.
IRAS OBSERVATIONS OF A 'TYPICAL' DARK CLOUD
Rene J. Laureijsl, Grzegorz Chlewicki 1 and Frank O. Clark 1 ,2
1
Laboratory for Space Research and Kapteyn Astronomical
2nstitute, PO Box 800, 9700 AV Groningen, The Netherlands.
Department of Physics and Astronomy, University of Kentucky,
Lexington, Kentucky, U.S.A.
ABSTRACT. We discuss the implications of the IRAS observations of a
regular isolated diffuse cloud. The dependence of infrared radiation on
the optical depth is different at short wavelengths (12 and 25 ~m) and
in the far-IR (60 and 100 ~m). Radiation at both 12 and 25 ~m appears
to be due to nonequilibrium emission from the same population of small
particles. The far-IR radiation can be explained by steady-state thermal
emission from interstellar grains.
1. Introduction
The detection of emission from the general diffuse medium at wavelengths from 12 to 100 ~m was among the most surprising early results
derived from IRAS data. The short wavelength radiation (12, 25 ~m) has
been attributed to transient heating of small particles (large molecules). Specific constraints on the nature of the particles and the
emission process can be derived from the analysis of individual clouds.
We present the IRAS observations of an apparently diffuse cloud selected
for a preliminary study on the basis of its regular shape and high
quality of the data.
2. Observational material and data reduction
The object analyzed in this study was selected on the basis of IRAS
maps of the Chamaeleon area. In the ESO/SERC J-plate, the cloud can be
seen as a gark object a%ainst the background of diffuse scattered light
at 1 - 300 and b ~ -17. The data base consisted of IRAS HCON3 scans
(IRAS Explanatory Supplement, 1985). The maps were corrected for detector sensitivity effects by two-dimensional Fourier filtering. The
zodiacal light contribution was estimated by masking bright sources and
fitting a cosecant law. The average of weak extended emission surrounding the cloud was adopted as the background level.
The maps obtained at 12 and 100 11m have been shown in Fig. ',. The
12 ~ data have been smoothed to the resolution of 100 ~ detectors.
323
A. Uger el al. (eds.), Polycyclic ArOnuJlic Hydrocarbons and Astrophysics, 323-325.
© 1987 by D. Reidel Publishing Company.
R. J. LAUREUS ETAL.
324
"/,.·0' () •.
0:,@
.~~::::.~
'''.
..
,
-' .....
Fig. 1. IRAS maps at 12
~m
(left) and 100
~m
(right).
3. Results
The striking regularity of the brightness distribution within the
cloud can be used to construct radial intensity profiles by averaging in
concentric rings. The results of this procedure are presented in Fig. 2,
which also contains a pixel-to-pixel comparison of radiation at various
wavelengths. The main features revealed by the brightness distribution
analysis in Fig. 2 are summariZed in the following list:
(1) The 60/100 ~ ratio remains constant over the entire radial extent
of the cloud. (2) The ratio of 12 and 25 ~m intensities does not vary
strongly but after improved background correction the data indicates a
slight decrease towards the centre of the cloud. The ratio of 12 and 25
~m fluxes integrated over each band is 0.3. (3) Linear regression for §~
and 100 ~ data implies a colour temperature of 27K assuming a A
emissivity law. (4) The 12/100 ~m ratio peaks at an intermediate radial
position in the cloud and declines towards the centre.
4. Discussion
The 60/100 ~ colour temperature obtained for our object and
reported previously in several other studies (Low et al., 1985; de Vries
and Le Poole, 1985) cannot be explained by current grain models, which
predict equilibrium particle temperatures below 20K. Our observations do
not support the hypothesis which attributes the excess emission at 60 ~
to small temperature fluctuations in graphite particles with sizes below
0.005 ~ (Draine and Anderson,1985). The slow decline of the 60/100 ~m
temperature wi th optical depth suggested by the observations is more
consistent wi th the assumption of equilibrium thermal emission from an
as yet unidentified population of "hot" (27K) particles.
The nearly constant 12/25 ~ ratio favours the association of the
12 ~ emission wi th the PAH' s rather than the 9. 7 ~m Si-O stretch in
IRAS OBSERVATIONS OF A "TYPICAL" DARK CLOUD
325
silicates, since low-energy vibrations in aromatic molecules could contribute significantly to the radiation in the 25 ~m band.
Rad lU5 (arcmln)
. '."
/i,!j?i
10 1
)(
)(
)(
)(
)(
)(
)(
100 \l1li
:'-.-
)(
."
.
)(
)(
)(
)(
+
60
+
+
+ +
+
\l1li
)(
)(
+
)(
+
)(
+
.'
25 IJIII
0
12 \l1li
..
0
......
0
0
0
o
)(
+
)(
+
- .. +
00 0
)(
.+~
&t+
)(
"...
)(
112 vs. 1 100
....
+
0
)(
)(
~
.'
0
•
+
Fig. 2. Left: radial brightness distribution. Right: pixel-to-pixel carS
relations between individual IRAS bands. Intensities in units of 10
Jy/sr.
REFERENCES
Draine B.T., and Anderson, N., 1985, Ap. J., 292, 494.
IRAS Explanatory Supplement, 1985, eds. C. A.-seichman et al.
Low, F.J., et al., 1984, Ap. J. (Letters), 278, L19.
de Vries, C.P., and Le Poole, R.S., 1985, Astr. Ap., 145, L7.
COAL TAR AS A LABORATORY ANALOG OF AN INTERSTELLAR PAH MIXTURE
T. J. Wdowiak
Physics Department
University of Alabama at Birmingham
Birmingham, Alabama 35294
USA
ABSTRACT. Spectral characteristics are described for coal tar prepared
by coking coal in the absence of air at 1400K. Coal tar films at 300K
exhibit broadband luminescence similar to the 5500A-7500A emission
"bwnp" of the Red Rectangle/HD44179, and a DIB-like absorption band
near 4430A. The infrared absorption spectrwn includes features at
wavelengths of most of the observed infrared emission bands. The
luminescence of coal tar has the character of excited dimer emission
suggesting such a mechanism may be occurring in the Red Rectangle.
SPECTROSCOPY. Stimulated by the suggestion of IAger and Puget
(1984) that polycyclic aromatic hydrocarbons (PAHs) are a constituent
of the interstellar mediwn, it was proposed that the broad band visible
emission of interstellar dust such as that of the Red Rectangle/HD44179
could be due to excited dimer (excimer) emission of PAHs clwnped together on the surfaces or interiors of grain mantles (Wdowiak 1985).
Coal tar prepared by coking coal in the absence of air at 1400K (Alabama By-Products/Drummond Coal Co.) has been examined as a laboratory
analog mixture representative of what might exist in the interstellar
mediwn. To remove opaque material the bulk fluid sample was diluted in
acetone and filtered through coarse filter paper. Luminescence spectra
of the cleaned sample diluted in ethyl alcohol (1) and as a film on
quartz (2) were determined by excitation at 3070A using a Perkin-Elmer
650-40 spectrofluorimeter having a R928 photomultiplier. The spectra
are shown in Fig. 1. above that of the Red Rectangle (Schmidt et al.
1980). The structured blue emission in ethyl alcohol shifts over 1500A
redward into the broad featureless band of the film in the character of
excimer emission (See Stevens 1962). The absorption spectrwn of a film
on glass (vs. glass) between 4000A and 7000A was obtained with a Cary
double beam spectrophotometer. It increases in a continuous fashion
toward shorter wave lengths and is marked with a distinct band near
wavelength of the strongest diffuse interstellar band (DIB) at 4430A.
The IR spectrwn shows features in absorption at most wavelengths for
which emission has been observed from various nebulae including the Red
Rectangle. It is similar to that of materials proposed by others.
1.
327
A. Leger et al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 327-328.
© 1987 by D. Reidel Publishing Company.
T. J. WDOWIAK
328
2
z
o
j
c
4000
5000
6000
A
4000
A
5000
Figure 1. Luminescence of 3070A excited coal tar in ethyl alcohol (1)
and as a film on quartz (2) corrected for R928 PMT sensitivity shown
with the spectrum of the Red Rectangle, and the absorption spectrum of
coal tar on glass versus glass.
4
Figure 2.
5
6
78910
1520
MICRONS
Infrared transmission spectrum of coal tar film on KBr.
2. ACKNOWLEDGEMENTS. I wish to thank J. Cleghorn for supplying the
coal tar and Herb Cheung for assistance with luminescence spectra.
This work was supported by NASA Grant NAGW-749.
3.
REFERENCES.
Leger, A. and Puget, J.L. 1984, Astron. Astrophys., 1~~, L5.
Schmidt, G.D., Cohen, M., and Margon, B. 1980, Ap. J.,
Stevens, B. 1962, Spectrochim. Acta,
!~,
~~2'
L133.
439.
Wdowiak, T.J. 1985 in Proc. Com. Inter-relationshi~Amo~ Circumstellar, Interstellar, and Interplanetaryt,. J. NUth
and R. Stencel (NASA).
HYDROGENATED AMORPHOUS CARBON (a:C-H) in the Planetary Nebula NGC 7027
John H. Goebel
NASA Ames Research Center
Moffett Field, California 94035
ABSTRACT. A spectroscopic identification of the infrared continuum
radiation is proposed for the planetary nebula NGC 7027. Hydrogenated
amorphous carbon (a:C-H) is shown to account for the undulating spectrum
between 5 and 15~m. The unidentified infrared emission bands lie at
the peaks in the a:C-H spectrum, pointing to their association with a
carbon polymorph, possibly a:C-H or polycyclic aromatic hydrocarbon
molecules (PARs). Except for atomic emission lines, all the infrared
emission from NGC 7027 comes from one or another polymorph of carbon.
INTRODUCTION
For over ten years, the origin of the thermal infrared (5-15~m) continuum radiation from NGC 7027 has remained a puzzle (Gillett, Forrest,
and Merrill 1973). Various attempts have been made to model the
continuum radiation with known astrophysica11y relevant dust species,
notably SiC, silicates, and graphite (Aiken and Roche 1982). Over a
limited wavelength interval, such as the 8-14~m window, reasonable
agreement has been shown with such heuristic concoctions (Aiken et a1.
1979). To date no convincing spectroscopic identification has been
proposed. The subject of this note is to propose that hydrogenated
amorphous carbon (a:C-H), a material measured in the laboratory, can
account for the thermal continuum's spectral energy distribution in the
carbon rich planetary nebula NGC 7027.
It is also shown that the unidentified infrared emission bands
(UIRs) are intimately related to a:C-H, and therefore arise from C-H and
Cac bonds exclusively. The implication of this is that the existence of
PARs (polycyclic aromatic hydrocarbons (Leget and Puget 1984, and
Allamandola, Tielens and Barker 1985» is supported while the ideas
concerning surface functional groups on grains causing the UIRs (Duley
and Williams 1981) are weakened.
DISCUSSION
329
A. Uger Itt aI. (eds.), PolycycJit: Aromatic Hydrocflrbons and Astrophysics, 329-334.
<C> 1987 by D. lUidd Publi.rhing Company.
330
J. H. GOEBEL
Recent studies of the properties of hard carbon films have yielded
voluminous measurements of their properties in an effort to fine tune
certain optical parameters, most notably the index of refraction nand
the visible absorption coefficient, for a wide variety of technological
applications. It happens, that one research group (Dischler et al.
1983a, b) has measured the mass absorption coefficient k in the thermal
infrared for films of (a:C-H) made from a plasma deposition of low
pressure benzene gas. The emission spectrum of NGC 7027 is compared
with the absorption coefficient of a:C-H in Figure 1. In the 4 to'
13 ~m spectral region, the similarity is striking enough to warrant
serious consideration. Neither unhydrogenated amorphous carbon, not
graphite compare as well to the NGC 7027 spectrum (cf., Knoll and Geiger
1984) •
Several questions arise from the comparison. Among the most
important are:
a)
Is it chemically reasonable to have a:C-H in the planetary
nebula NGC 7027?
b)
How much a:C-H is required to account for the infrared
emission from NGC 70277
c)
How is the a:C-H emission excited?
d)
Are the UIRs related in any way to a:C-H? How are they
excited?
e)
Are the known formation mechanisms for a:C-H reasonable in a
planetary nebula?
f)
Does a:C-H evolve into graphite? Do PARs evolve into a:C-H or
vise versa? What evolution of a:C-H occurs within the planetary nebula
phase and what occurs in dark clouds and the diffuse interstellar
medium?
Space does not permit a full discussion of each of these questions.
What L believe to be reasonable initial answers to these questions are
given below.
a)
The C/O > 1 in NGC 7027 data indicates a carbon rich chemistry, in which case carbon dust condensation is expected. The apparent
absence of a normal SiC band in NGC 7027, which is found in many other
carbon rich objects, is remarkable.
b)
Using the measured mass absorption coefficient (Sah et al.
1985) and the average dust optical depth at 10~m in NGC 7027, one
calculates-8 x 10 E-6 M of a:C-H, assuming a normal C/H abundance
ratio. Clearly not a pr~blem.
c)
A thermal emission mechanism with a temperature in the 200-300
K range would reproduce the undulating continuum of NGC 7027 nicely. As
most of the dust is at a temperature of 90 K (which is determined from
far infrared observations), it is not readily apparent that the stellar
UV field will produce the heating required for a:C-H excitation. Only
infrared imagery with narrow bandpass filters can produce the kind of
data necessary to answer the question fully.
d)
The UIRs lie on the a:C-H pedestals. For the major UIR bands
at 3.3-3.4, 6.2, 6.9, 7.7-8.6, and 11.3~m this occurs with a one to one
correspondence. The conclusion that the UIRs are due to C-H and C-C
bonds is unavoidable. It does not follow that surface functional groups
HYDROGENATED AMORPHOUS CARBON IN THE PLANETARY NEBULA NGC 7027
331
on a:C-H are the origin of the UIR bands. The relatively narrow UIRs
could be from smaller a:C-H particles (PAHs) or from molecule sized
domains in larger a:C-H grains.
A plasma excitation process is another interesting possibility. Planetaries have highly ionized gases like S IV (40 ev). The
existence of plasma and confining magnetic fields is apparent from radio
images. However, the details of formation and excitation have not yet
been worked out. Recent laboratory studies of the plasma emission from
a:C-H both in its diamond like form and the polymer form show that
luminescence is an important deexcitation mechanism for the gases as
well as the bulk materials formed. Duley (1985) has demonstrated that
the Red Rectangle's red visible emission feature can be described by the
a:C-H polymeric form. In addition to the broad red feature, several of
the sharp emissions can be ascribed to the gas plasma based upon
laboratory spectroscopy (Park, Bodart, and Feldman 1985). Infrared
studies of the polymeric forms by Dischler et al. (1985) show narrow
absorption features similar to the 3.2 to 3.5 p.m and 6.2 p.m bands in NGC
7027. A red visible emission feature is possibly present in NGC 7027 as
well.
Hence, a study of the red emission and the narrow features
could produce a correlation which supports their origin from a plasma
excitation mechanism of the polymeric forms of a:C-H. Similar comments
would then apply to the case of reflection nebulae. In many ways, these
arguments also support the concept of ionized PAHs proposed by
Allamandola, Tielens, and Barker (1985). The spatial regions of UIR
emissions would be expected to be correlated with regions of higher
electron density rather than simply density enhancements or clumpiness.
The quality or excitation level of the plasma must necessarily have a
very low threshold as there is no perceptable atomic emission from the
Red Rectangle.
e)
Allamandola et al. (1985) argue that carbon grains are made
from PAHs. Duley and Williams (1986) argue that PAHs are made from
hydrogenated amorphous carbon. Probably both are right.
f)
Laboratory studies indicate a clear evolutionary path for
a:C-H to become graphitic carbon upon heating in excess of 600 C.
Heating of as formed a:C-H in the laboratory will effuse hydrogen from
the bulk and cause a reordering of the bonding between the Cs and Hs to
take place. This is reflected in a change in the absorption spectrum.
For example, the 3.4 p.m band d'ominates the 3.3 p.m band in as formed
a:C-H. but heating to 500 K causes the 3.3p.m band to dominate (Dischler
et al. 1983a). Further heating will give rise to hydrogen effusion
until little H is left and a graphitic structure remains. Heating is
observed in the laboratory to cause the appearance of a pi plasmon
resonance in the UV, i.e., the 2200 A feature (Fink et al. 1984). Duley
(1984) has shown that there is no 2200 A feature in the HAC particles he
has produced in the laboratory, in contrast to the well known interstellar absorption feature. So it appears to be possible to evolve the
higher ordered forms of carbon from the hydrogenated forms.
Far infrared spectrophotometry of NGC 7027 (McCarthy, Forrest,
and Houck 1978) gives a spectral index of n - -2 which is associated
with graphitic carbon, not amorphous carbon (n - -1). The far infrared
J_ H. GOEBEL
332
2000
I
1000
~I
.....
~
!
NGC 7027 and Hydrogenated Amorphous Carbon
--------------------------------,
I
I
I
I
500
~
I"
II-"
I
i: :r
Ie
IX
Ix
II--~,"
1\
, C>
!
!:E
II
\
~ :: LJL~'"'t=:"Pho",_:'b,"___
1
2
I
10;
5
10
20
Wave length (urn)
50
I~'
I~
Ie
Ie..
I
I
I
100
Figure 1. The comparison of the infrared emission spectrum of the
planetary nebula NGC 7027 with the absorption coefficient of Hydrogenated Amorphous Carbon. The NGC 7027 data are taken from Russell et
ale 1977 between 4 and 8 JLm, Gillett et ale 1973 between 8 and 14 JLm,
McCarthy et al. 1978, and Moseley et al. 1986. The a:C-H spectrum is
taken from Bubenzer et al. 1983 and is measured in a 300K sample. The
bands have not been weighted by a Planck Function.
333
HYDROGENATED AMORPHOUS CARBON IN THE PLANETARY NEBULA NGC 7027
emitting material also displays a lower emission temperature, 90 K, and
therefore one suspects that this material is fully converted to the
graphitic form. In fact, the bulk of the infrared energy coming from
NGC 7027 is in the far infrared and coming from the graphitic component.
The emission coming from the PAHs and a:C-H is relatively minor energetically, but happens to dominate the mid infrared spectrum. Hence,
one observes at least three carbon polymorphs in the nebula NGC 7027,
but their relative spatial distributions are not yet known. It will be
extremely interesting to see detailed spatial maps in the future when
such instrumentation becomes available.
Small molecule like subunits of a:C-H can have a spatial localization
phenomenon in their electronic bands due to hydrogen bonding (Bredas and
Street 1985). These molecule like regions could then behave like PARs
without being separated from the grain. This would only be true if the
PAH emissions arise in the electronic band itself which is a question
that needs to be addressed. It seems entirely plausable that the UIRs
could result from a vibronic deexcitation process if the appropriate
time scales can be shown to apply.
Some of the carbon dust is already graphitic in NGC 7027, so an
evolution of carbon polymorphs is taking place within the nebula. As
just remarked, laboratory studies indicate that as a:C-H is heated,
hydrogen is effused giving rise to more C=C bonding (Dischler et al.
1983a). Above 500 C, a pi plasmon is observed which is weak or absent
at lower temperatures (Fink et al. 1984). So a:C-H present in dark
clouds will show little or no 2200 A feature. As it goes into the
diffuse clouds, a:C-H will be subjected to heating and UV processing
causing the evolution to graphitic carbon and the well known 2200 A
feature. The stars like HD 29647 are the answer to this problem. It is
situated nicely behind a small dark cloud in Taurus, and displays little
or no 2200 A feature, yet i t has an E(B-V) = 1. The dust in that cloud
is relatively unprocessed, as is indicated by the observation of the
3.1 f-tm ice band (Goebel 1983).
REFERENCES
Aiken, D. K., and Roche, P. F., 1982, M.N.R.A.S. 200, 217.
Aiken, D. K., Roche, P. F., Spenser, P. M., and Jones, B., 1979, Ap. J.
233, 925.
A11amando1a, L. J., Tie1ens, A. G. G. M., and Barker, J. R., 1985, Ap.
J. Letters 290, L25.
Bredas, J. L., and Street, G. B., 1985, J. Phys. C .18, L651.
Bubenzer, A., Disch1er, B., and Koidl, P., 1983, J. Appl.
4590.
Phys.~,
Dischler, B., Bubenzer, A., and Koid1, P., 1983a, Sol. State Comms. 48,
J. H. GOEBEL
334
105.
Dischler, B., Bubenzer, A., and Koidl, P., 1983b, Appl. Phys. Lett. 42,
636.
Dischler, B., Sah, R. E., Koidl, P., Fluhr, W., and Wokaun, A., 1985,
Prqc. 7th Internat. Symp. on Plasma Chemistry, ed. Timmermaus, C. J.,
(Eindhoven) •
Duley, W. W., and Williams, D. A., 1981, M.N.R.A.S. 196, 269.
Duley, W. W., 1985, M.N.R.A.S.
~,
259.
Duley, W. W., and Williams, D. A., 1986, M.N.R.A.S. 219, 859.
Fink, J., Muller-Heizerling, J. Pfluger, Scheerer, B., Dischler, B.,
Koidl, P., Bubenzer, A., and Sah, R. E., 1984, Phys. Rev. B30, 4713.
Gillett, F. C., Forrest, W. J., and Merrill, K. M., 1973, Ap. J. 183,
87.
Goebel, J. H., 1983, Ap. J., 268, L41.
Leget, A. and Puget, J. L., 1984, Astron. ,and Astrophys. 137, L5.
McCarthy, J. F., Forrest, W. J., and Houck, J. R., 1978, Ap. J.
109.
~,
Moseley, S. H., Silverberg, R., and Glaccum, W. 1986, private
communication.
Park, C. S., Bodart, J. R., and Feldman, B. J., 1986, preprint.
Russell, R. W., Soifer, B. T., and Willner, S. P., 1977, Ap. J. 217,
L149.
Sah, R. E., Dischler, B., Bubenzer, A. and Koidl, P., 1985, Appl. Phys.
Lett. 46, 739.
VISUAL AND INFRARED FLUORESCENCE FROM L1780
Grzegorz Chlewicki and Rene J. Laureijs
Laboratory for Space Research and Kapteyn Astronomical Institute, P.O. Box 800, 9700 AV, Groningen, The Netherlands
ABSTRACT. The paper presents IRAS survey observations of a high latitude
galactic cloud with moderate extinction, L1780 (A B ~ 3m). The data are
consistent wi th the hypothesis that the red excess observed in the
visual spectrum of the cloud is due to fluorescence from the same
species which gives rise to the 12 ~m emission observed by IRAS.
1. Introduction
The existence of polycyclic aromatic hydrocarbons in the interstellar medium appears to be one of the most promising explanations for the
12 ~m emission observed by IRAS (Puget et a1., 1985). This explanation
suggests that optical emission from diffuse clouds should be equally
common as 12 ~ emission, since aromatic molecules typically have high
visual fluorescence yields. Broad-band emission suggestive of molecular
origin has so far been reported for only one object - a peculiar nebula
known as the Red Rectangle (Schmidt et a1., 1980). A unique spectrum
obtained by Mattila (1979) for an interstellar cloud, L1780, shows an
excess of red light with spectral characteristics similar to the feature
observed in the Red Rectangle. The analysis of the IRAS observations of
the cloud presented in this paper has been carried out in order to test
whether the PAH' s can simultaneously explain the visual and infrared in
an 'ordinary' interstellar cloud.
2. Observational data
The emission from L1780 has be ell registered by IRAS in all four
bands (12, 25, 60, 100 ~m). The reductions applied to the data are
described in detail elsewhere (Laureijs et a1., this volume). Optical
analysis of L1780 and. the surrounding clouds indicates a distance r ~
100 ~ (1_= 359 0 , b ~ 36 0 ), a peak extinction AB = 3m, and a density nH
= 10
cm 3 (Mattila, 1&79). At the distance of 100 pc, the angular
diameter of the cloud (0 .7) corresponds to approximately 1 pc. Preliminary IRAS results (no destriping) are presented in Fig. 1. The distribution of infrared radiation broadly follows the extinction map derived
from star counts by Mattila (1979).
335
A. Leger et aI. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 335-337.
© 1987 by D. Reidel Publishing Company.
336
G. CHLEWICKI AND R. J. LAUREIJS
15"39"
as
15"36"
as
15"39"
as
15"36"
as
Fig. 1. IRAS maps of L1780: 12 11m (left); 100 11m (right). The eastern
part of the region has not been covered by available scans.
The visual spectrum of L1780 obtained by Mattila (1979) is compared
with the spectrum of the Red Rectangle (Schmidt et al., 1980) in Fig. 2
on the following page. The increase towards longer wavelengths significantly exceeds the likely errors, and the feature appears to peak at
-6500A as in the Red Rectangle.
3. Discussion
The blue portion of the L1780 spectrum can be accounted for by diffuse galactic light scattered by dust grains in the cloud. The intensity
of scattered light is expected to remain nearly constant for A > 6000A.
The hypothesis that the excess of red light is due to fluorescence
(phosphorescence) from molecules either in the gas phase or in grain
mantles can be tested by estimating the required quantum efficiency (qR)
of the process. Assuming spherical symmetry for the cloud and subtrac~­
ing a constant component due to scattering from the red sp~~trum, ~T
arrive at the integrated intensity for the red feature of 8*10
erg s
= 0.2 L • Table I contains the values of q , whIch for molecules in
grain ma~tles have been based on the assumpt~on of spherical pa~f~cles
with a radius of 0.15 11m and a normal gas-to-dust ratio, n
- 10
n.
For gas phase speCies, we have assumed that the mOieculesgS9Psist2of ~O
carbon atoms and have a UV absorption cross-section of 10
cm; the
total content of carbon is taken to be 1% of the cosmic abundance.
In aromatic molecules, the most likely source of the red emission
Is T]-SO phosphorescence following intersystem crossing from an excited
singlet state. For a typical photon of -4 eV j the energy is distributed
337
VISUAL AND INFRARED FLUORESCENCE FROM Ll780
I
A.IAI
3000
Fig. 2. The comparison of the spectrum of L1780 from Mattila and the Red
Rectangle data obtained by Schmidt et al ••
Table I
Fluorescence yields derived from the L1780 emission
Excitation
threshold
3.5 eV (UV)
2 eV (UV+Vis)
Grain
mantles
0.03
0.01
Free
molecules
-1
0.4
approximately equally between visual and infrared emission (IRAS 12 \.lm
band). Therefore..i5 the C.9~p~,iS9~ of intensity observed in the 12 ).lffi
band~5I12 = ~~191
~fgcm
s sr ,with that of the red feature, I vis =
4*10
ergcm s sr ,provides an alternative method of estimating qR'
The agreement of this estimate (0.45) with the number listed in Table I
supports the attribution of visual and infrared emission to PAH's.
REFERENCES
Mattila, K., 1979, Astron. Ap., 78, 253.
Puget, J.L •• L~ger, A., Boulanger, F., 1985, Astron. Ap., 142, L19.
Schmidt, G.D., Cohen, M., Margon, B., 1980, Ap. J. (Lett.), 239, L133.
DISCUSSION II:
INTERPRETATION OF IR OBSERVATIONS
Discussion
Chairman:
Lou Allamandola
NASA Ames Research Center
MS 245-6
Moffett Field, CA 94035
U.S.A.
The problem of understanding the UIR phenomenon really consists of two
parts. The first is the identification of the carrier of the spectrum
and the second is the elucidation of the microscopic emission mechanism. The first part of this discussion will focus on the former the
second on the latter. The eventual solution of these two questions
will require a two pronged attack, one focusing on very specific observations the other on laboratory experiments on relevant systems.
Regarding this last point it is important to constantly bear in mind
that at present we are comparing measured absorption spectra of neutral
molecules often in condensed, or microcrystalline form, with emission
features which may well arise from a mixture of things - many of which
may be in ionized and radical form. We will also discuss what kinds of
observations and experiments are needed to provide this information.
To address the first question requires that we have to agree on
what the UIR spectrum is. Table 1 (Allamandola, Tielens and Barker,
1986) summarizes what we believe make up the spectroscopic part of the
urR band problem. The last two entries regarding the 6000A to 2 micron
continuum and the 1-5 micron continuum are somewhat controversial and
probably not worth discussing in detail. They are included because
there is evidence in some objects that this "continuum" is associated
with the UIR phenomenon. So, with these as the observational "facts",
along with the assignments proposed within the framework of the PAH
hypothesis, one embarrasing question we should also ask ourselves is what evidence is there in absorption?
With this introduction, the following discussion took place.
P. Brechignac:
You should ask "What are the carriers of the UIR
bands?", rather than "What is the carrier?". You may have several
species or kinds of molecules.
A. Leger:
r agree, 3.3, 6.2, 8.6 and 11.3 are very characteristic of PAHs in general, a mixture of the right ones should be able
to reproduce the observed spectrum.
339
et af. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 339-350.
© 1987 by D. Reidel Publishing Company.
A.
L~ger
A
ASSIGNMENT
70-200
1315-1250
3.4
I
13.21-3.65
2940 I
13115-2740
1950-740
880 I
Overlap of many aromatic C-H out-of-plane bending modes for non-adjacent as well as doubly
and triply adjacent peripheral H-atoms
Aromatic C-H out-of-plane bend for nonadjacent, peripheral H atoms
Aromatic C-H in-plane bend
"300"
Overlap of C-H stretching modes, shifted by
anharmonic effects, with overtones and
combinations of fundamentals in the 1670-1250
cm-l (6-8 micron) region,?
THE BROAD COMPONENTS
"157"
885
11.3
11O.;:13.s1
30
1150
8.7
Blending of several strong aromatic C-C
stretching bands
Aromatic C-C stretch
30
1615
6.2
7.6- 8.0
Aromatic C-H stretch (v=1 -- v=O)
THE AROMATIC HYDROCARBON BANDS
FWHg l
(cm )
30
)
3040
11_1
(cm
3.29
(Microns)
TABLE 1 EMISSION COMPONENTS
(ij
oz
12
8
Cl
'"~
A
Overlap of many aromatic C-H out-of-plane
bending modes for non-adjacent as well as
doubly and triply adjacent peripheral Hatoms
3.46 a
3.39-3.41
2890
2950-2935
Aromatic CH stretch (high v), aliphatic CH
stretch, overtone/combination ba~~ involving
fundamentals in the 1810-1050 cm
(5.52-9.52
micron) range
Aromatic CH stretch (v=2 -. v=l)
Overtone and/or combination invo!ying
fundamentals in the 1810-1050 cm
(5.52-9.52 micron) range
2995
3.34
"20"
ASSIGNMENT·
Blending of many weak aromatic C-C
stretching bands
THE MINOR FEATURES
"160"
"400"
)
Overtone and/or combination invo!ying
fundamentals in the 1810-1050 cm
(5.52-9.52 micron) range
125.0-1110
11810-1050
(cm
FWH!!1
3085
~i3.51
)
I
880 I
I950-740
V-I
(cm
3.24
110.5
I5.52-8-99.52 I
(Microns)
TABLE 1 EMISSION COMPONENTS (CONTD)
~
~
.:::
CIl
~
I;j
A
30*
1470-1450
6.8-6.9
*
a
II
Continuum
Quasi-continuum formed by overlapping
overtone and combination bands
Electronic transitions between low-lying
levels in ionized and complexed PAHs
Aromatic C-C stretch, aliphatic CH
deformation
C-C stretch; overtone of 885
cm
(11.3 micron) band; Carbonyl C=O
stretch
Ar~fatic
Aromatic CH stretch (high v), aldehydic CH
stretch, overtone and combinatiog1band involving
fundamentals in the 1810-1050 cm
(5.52-9.52
micron) range
Aromatic CH stretch (high v), aliphatic CH
stretch, overtone/combination bagy involving
fundamentals in the 1810-1050 cm
(5.5.2-9.52
micron) range
ASSIGNMENT
From Bregman et al., 1983
Quotation marks indicate a value estimated from published spectra
From de Muizon et.al., 1986
The upper number indicates the peak position, the lower numbers indicate the high and low
frequency limits of the band.
M~-IR
Red-Near IR Continuum
40*
FWH!!1
(cm )
1785-1755
2810
3.56 a
)
5.6-5.7
2850
V-I
(cm
3.51 a
(Microns)
TABLE 1 ( CONTD)
~
'"'"
~
t:I
~
DISCUSSION II
343
S. Leach:
You must be very careful when you compare an absorption spectrum to an emission spectrum. You can't make a point concerning the identification too strongly.
H. Kroto:
Under what circumstances are the bands seen in
emission? (Allamandola, this was answered by a short "review" of what
appears in the papers by Puget, Leger, Jura and Allamandola, elsewhere
in this publication).
F. Tramer:
Although this is not my area of expertise, I believe
you should be careful. If there are many different molecules, radicals
and molecular ions, won't you obtain something like a continuum?
S. Leach:
I don't believe a continuum would result, broadening
yes, but not a continuum. If you have a class of very similar molecules such as the PAHs are, they will all have IR bands in very
specific frequency ranges.
W. Duley:
RACs (hydrogenated amorphous carbon particles) are a
big collection of different molecular subunits and we have seen on
several occasions here that its spectrum can provide a rather nice
match.
J. Jortner:
With all of the vibrational energy in the molecules you should have some contribution from say V=10 or V=9. If this
occurs, you should see effects due to anharmonicity which will result
in broadening and shifting of the bands considerably.
J. Puget:
It does not make sense to rely too heavily on a
detailed spectroscopic comparison, but you must consider the excitation
mechanism simultaneously.
A. Leger:
(Shows the Transparancy of the figure in his article
elsewhere in this publication in which the IR absorption spectra of a
few large PAHs suspended in KBr - pellets are compared to the emission
spectra from Reflection Nebulae.) This comparison shows very well that
the absorption spectra of a few molecules can fit the emission
spectrum. Furthermore, broadening must not be too much of a problem as
the 6.2 micron band shows that the emission features matches well what
is expected from the absorption spectra.
W. Duley:
The 6.2 band has long been known to be highly
characteristic of aromatic systems.
M. Jura:
I noticed nearly all of the laboratory measured PAH
spectra we have seen so far shows something at about 5.2 microns. Has
a feature ever been seen at this position in any astronomical object?
L. Allamandola:
Generally, the region between 5 and 5.5 microns has
not been covered with astronomical observations. This is one of the
observations we have proposed to do. If granted KAO telescope time,
344
DISCUSSION II
we'll have the answer within the year.
M. Greenberg:
What is the chance of seeing any of these bands in
absorption, say towards something like the galactic center?
A. Leger:
The infrared absorption strength of PARs are very
weak. For example, with Av of 25, T (3.3 microns) should be only about
1%.
M. Greenberg:
(Shows several transparancies of spectra of several
protostars such as BN and W33A in the 2.5 - 4 micron range taken using
the UKIRT cooled grating spectrometer). Note that in the spectrum of
BN there may be an absorption at about 3.3 microns which is on the
order of several percent. Note however that this is not so clear in
the spectra from these other objects.
P. Martin:
Your spectrum of BN doesn't seem to show any
absorption at 3.4 microns while it is evident in spectra toward the
galactic center. Is that correct?
M. Greenberg:
(Shows several more transparancies of spectra of
protostars and one of the galactic center). Sometimes there is a very
weak, very broad depression near 3.4 microns in the protostar data, but
it does not seem to consistently span the same frequency range. It
overlaps a lot, but the extremes seem to vary. At this point I'd have
to say that it isn't really clear whether there's a band there or not
toward the protostars. Towards the galactic center therp. is no
question. The band is clear and strong.
A. Leger to
W. Duley:
Do you have any absorption spectra of HAC?
W. Duley:
K. Roessler has shown us some spectra of HAC at
various stages of hydrogenation and it shows the full range of
features. (The spectrum of this form of amorphous carbon is presented
in the papers by Roessler, spectra of other forms of amorphous carbon
are presented in the three separate papers by Onaka, Bussoletti and
Allamandola.
A. Tielens:
We have spectra of deeply embedded protostars which
show a weak, broad band centered about 7.8 microns. If this is interpreted as arising from PARs and related material, we deduce an abundance which is consistent with that deduced from the emission features.
L. Allamandola:
John Goebel has very recently made infrared photographs of NGC 7027 using an IR array camera which clearly shows that
the emission spectrum from this object varies considerably from one
region to the other on a few arcsecond scale. The shoulder of the 7.7
micron feature (measured at 8.2 microns), for example, peaks in a
different location than the broad plateau centered at about 12 microns.
This clearly shows that in this high excitation object, more than one
345
DISCUSSION II
component contributes to the spectrum and that if one measures the
spectrum of the entire object, all components contribute. Furthermore,
Goebel has also shown (see the short paper, elsewhere in this volume)
that the underlying emission baseline in NGC 7027 matches exceedingly
well, with what is expected on the basis of a simple comparison of the
absorption "spectrum" of hydrogenated amorphous carbon that Dr.
Roessler has shown us several times during this meeting. Thus, it is
important to both spatially and spectrally resolve the IR emissions to
minimize (eliminate?) the confusion introduced by several component
spectra.
In concluding the first part of this discussion, a few moments
were spent in considering the types of observations which would be
needed to further progress in this field. They stressed the necessity
to obtain good quality, moderate resolution spectra of say 10-20 well
chosen objects. This will provide confidence in the spectroscopic
details of the problem and give an indication of how the spectrum can
vary from object to object.
A "well-chosen" object is one that is:
a)
b)
spatially extended, with a well characterized
geometry and exciting star
devoid of high excitation (no ions in emitting zone).
Spatially resolved spectroscopic studies can then be used to trace the
distribution of the emitting species from higher denSity, lower flux
regions to lower density, high flux regions and understand how its
physical nature changes as it undergoes this transition.
J. Puget:
Another important observation would be to look for
the emission features in some embedded objects as this would give us
information about the excitation within a cloud.
L. Allamandola:
Now lets shift the discussion to the emission
mechanism. Recall that while many pumping mechanisms can excite the
emission, the environment in most of the objects (not all) strongly
points to ultraviolet (and blue) photons pumping the emission (see
mechanisms described in the papers by Leger, Puget and Allamandola and
references therein). To further constrain this, Pat Roche has asked me
to stress that (1) the emission is seen almost exclusively only in
association with early type stars, stars which emit a large fraction of
their energy with photons in the 1000-2000A range, but not in
association with later type stars which are cooler and emit primarily
in the visible and at longer wavelengths; and (2) it is seen in both
neutral and ionized regions. With these as rough "boundary conditions"
on the environment from which the UIR bands are seen, let's now focus
on the nature of the carriers we've been discussing at this meeting.
While the match is not perfect, the UIR emission band spectrum
resembles various types of amorphous carbon at least as well as the
match with several individual PAR spectra. (See papers by Roessler,
Leger, Goebel, Duley, Bussoletti, and Allamandola) this shouldn't be
DISCUSSION II
346
too surprising as amorphous carbon is made up largely of PAH building
blocks. To account for the observed emission intensity at 3 microns
however forces us to small systems which are on a molecular scale
(Sellgren, 1984, Ap. J. 277, 623, Leger, Allamandola). The coupling
between individual PAH structures in an amorphous carbon particle to
the phonon bath will presumably be so strong that the energy from an
individual UV photon will be distributed to the low lying phonon modes
on a time scale much shorter than the IR radiation time (see papers by
Jortner) and so rather than emission at 3 microns, emission at longer
wavelengths would be expected. Thus, we are faced with a "dilemma" on the one hand the spectra point to particles - on the other the
energetics point to molecules.
Several suggestions have been made which may provide a way to get
around this dilemma. Since at this meeting there are so many experts
in related fields it would perhaps be best to focus the discussion
somewhat by mentioning these suggestions and hearing what people think.
Duley (paper elsewhere in this publication) has suggested that one
can treat emission from hydrogenated amorphous carbon (HAC) as an
electronic transition in a semiconductor. In this case, the C-R
stretching vibration at 3 microns would have the intensity of an
electronic transition rather than a vibrational transition. Since
allowed electronic transitions are generally several orders of
magnitude stronger than allowed vibrational transitions, the 3.3 to
11.3 micron intensity ratio problem would be resolved.
At Ames we have been considering that loosely bound clusters of
PARs might provide a solution. Since the attractive force between two
PAH molecules is on the order of several tenths of an ev (for coronene,
the calculated potential is .3 - .4 ev), PAH clusters made up of
several PAHs should be stable entities Figure 1. If a photon is
absorbed in one of the PAH molecules, provided it is loosely coupled to
the others in the cluster so that low frequency nonradiative relaxation
uv.,
PHOTON
Figure 1. Schematic of a PAH cluster, in which the vibrational energy
from a UV photon or exciton is "localized" in one of the PAHs.
347
DISCUSSION II
is quenched, the energy may remain localized and the individual PAR can
emit in the IR similar to the way a free molecule would. Admittedly
there are problems since non-radiative relaxation is generally very
fast in solids. The question to ask is, "Would a PAH cluster behave
more like a particle or molecule regarding this process?" Fortunately,
this type of process can now be tested in laboratory experiments and I
hope that Professor Jortner and his colleagues, as well as others with
molecular beam capabilities will carry out these types of measurements.
Dr. Tramer was quick to make a similar suggestion after hearing of
all this for the first time at this meeting. He wonders if PAR
"fragments", loosely attached to grain surfaces as shown in Figure 2
might be sufficiently loosely coupled from the phonon bath of the grain
to effectively emit at the higher frequencies.
((r~,))
,~
"!:.-
GRAIN
Figure 2.
surface.
Schematic of a vibrationally excited PAR attached to a grain
With this summary of the points people have asked to be brought up,
let's begin the discussion.
M. Jura:
Regarding the question of UV photons, I don't
believe CRL 618 shows the emission bands, yet it is a B star and
therefore produces a lot of photons in the 1000-2000A range.
L. d'Hendecourt,
P. Roche and
A. Omont:
It seems that you always need UV to pump features,
but there are also several good examples of objects
which are carbon rich and which are exposed to UV
but which do not show the features.
W. Duley:
In the visible spectroscopy of these molecules, you
see clearly that C-C and C-H stretching modes are excited yet there is
some doubt expressed about the 3 micron (C-H stretch) intensity in
emission having comparable strengths.
348
DISCUSSION II
L. Allamandola:
I have some difficulty with this since the oscillator strength f, which is directly related to the absorption or
emission strength for any particular transition, is defined as
inversely proportional to the mass of the oscillating charge which
generates the oscillating dipole and produces the electromagnetic
radiation at that frequency. Thus, since the 3.3 micron band is
due to a C-H stretching vibration, the oscillator strength fCH~
(m /~)f, where m is the electron mass, ~ is the proton mass and
f e is th~ oscillato? strength for purely electronic transitions which
i~ what is measured in the visible. Thus, unless there is some sort of
charge localization on the CH, I don't see how an electronic transition
involving the CH vibration could produce such a strong band at the CH
vibrational fundamental frequency.
L. d'Hendecourt:
Apart from this question of intrinsic oscillator
strength, hasn't a lot of experimental work been done in cryogenic
(10K) matrices which shows that for vibrationally excited species in a
solid, non-radiative relaxation to the phonon bath is the dominant
relaxation process and IR emission from vibrationally excited
fundamentals is simply not seen from solids?
P. Boissel:
IR fluorescence from vibrationally excite21CO is
known to occur in matrices because it's fr~~uency of 2140 cm
is so
much higher than the phonon modes «100 cm ) and you have a very large
energy gap. If the energy gap is large, non-radiative relaxation is
generally low. SF 6 on the other hand has many more levels, some much
lower in frequency and closer to the phonon frequencies. Since the
energy gap is so small in this case, the vibrations in SF 6 can couple
more efficiently to the phonon bath. The measured non-radiative
relaxation times for the different modes in SF 6 in a matrix at 10K lie
in the microsecond range. I believe this would also be the case for
PAR units in a solid.
P. Brechignac:
You should also consider relaxation via loss of
rotational quanta.
S. Mukamel:
In such a large system the rotational constant is
small. In order to lose vibrational energy via rotational relaxation
would require a many rotational quanta transition. Such high quantum
number transitions are generally not very efficient non-radiative
relaxation channels in solids.
S. Leach:
It Is conceivable that a PAR unit could act as an
impurity trapping site either in or on a, solid or loose cluster. It
would have to be weakly coupled vibrationally and strongly coupled
electronically. Of course the more order in the grain, the more
efficient the non-radiative processes would be.
J. Jortner:
Yes, such a system is possible. For example, one
could imagine in HACs that each subunit is sufficiently decoupled from
the rest to permit relaxation via IR emission.
DISCUSSION II
349
T. Wdowiak:
This is an interesting idea because these types of
clusters of PARs will form excimers (when they absorb a UV photon)
which are known to emit in the red. This may be relevant for the broad
emission seen in the Red Rectangle. I wonder if this can also explain
the 1 to 5 micron continuum emission seen in several reflection
nebulae?
A. Omont:
Yes, but do you reallX3think it would be possible to
preserve the storage of energy during 10
seconds in a PAR on a
surface?
S. Leach:
It is really a question of how decoupled the
vibrations in the PAR are from the lattice vibrations.
M. Greenberg:
This picture might have difficulties with abundance.
If you covered a grain surface with PARs you would use up all of the
carbon.
S. Leach:
It is worth mentioning that one way to study the
relaxation channels in large highly vibrationally excited molecules
would be by selective ionization from high vibrational levels in a two
photon experiment. By selectively varying the state ionized one can
gain insight into the various relaxation pathways which occur in the
high vibrational levels of the excited electronic state.
This is what
one needs to know concerning the disposal of energy once a PAR absorbs
a photon.
L. Allamandola:
Before we conclude this discussion, it would be
useful to review the feasible laboratory experiments which are needed
to further progress in this field.
Perhaps the foremost experiment to be done is to measure the
UV-Vis to IR conversion efficiency as a function of wavelength for
free molecules and clusters of molecules. While it would clearly be
most desirable to have these done in jet experiments of the type
described to us by Dr. Jortner, where the isolation and cold conditions
most closely approximate those in the interstellar medium, it would
also be useful to carry out these measurements in bulb type (static gas
cell) experiments and compare the results between the two techniques.
The reason for this two pronged approach is that expanding jet
experiments require a much higher degree of technical support than bulb
experiments and therefore can be done reliably in only a few institutions, institutions whose interests often lie elsewhere. Bulb experiments have been successful in this regard. For example, Barker, Rossi
and Pladziewicz Chem. Phys. Lett. 90, 99 (1982), have measured emission
at 3.3 and 4 microns from laser excited azulene (C 10R8 , a bicyclic
aromatic hydrocarbon) in a long gas cell and found that the intensity
at 4 microns is about 5 times weaker than that at 3.3 microns, a result
consistent with the ratio of the 3.3 micron line to continuum ratio
measured by Sellgren (1994), Ap. J. 277, 623 and Sellgren et al.,
(1985) Ap. 299, 416. In addition to quantifying the process, the
results of these experiments will be important in testing the various
350
DISCUSSION II
treatments (thermal vs. non-thermal) which have been proposed to
account for emission from highly vibrationally excited large molecules
(see the discussion session lead by A. Beswick and papers by Leger,
Puget and Allamandola elsewhere in this volume). Once a reasonable
understanding starts to emerge, it would be desirable to extend the
theory and measurements to larger molecules and ions. This introduces
far more complexity, but we have seen at this meeting that we must
consider species which increase in size up to the particle limit.
These experiments can be useful in determining what that limit is for
the properties we are interested in.
L. d'Hendecourt:
Are cold jets actually necessary?
of a photon, the molecules will be hot anyway.
After absorption
J. Jortner:
The question is one of collisions. In some cases,
when conditions in the bulb are choosen carefully, results similar to
jet experiment results have been obtained. To the best of my knowledge, IR emission from molecules in jets has not yet been measured.
L. Allamandola:
Parallel to these experiments, and equally
important, is the necessity to measure the spectroscopic properties of
free molecules in the ultraviolet, visible and infrared; in the ground
and, when possible, excited states. It has not come up here, but the
fact is that nearly all PAR molecular spectra taken to date have been
for microcrystals (large ordered "clusters") suspended either in KBr
pellets or in solution. There will surely be solid state effects which
effect the spectra in all three regions. To truly test the PAR
hypothesis and learn about the fundamental spectroscopic properties of
free PARs, one must study the spectra of isolated species. The two
techniques which have been proven for this are matrix isolation
spectroscopy and spectroscopy in high temperature gas cells. The later
technique suffers from the introduction of hot bands which introduce
transitions not appropriate to cold, free molecules in the interstellar
medium. It will, however, provide useful data for other regions.
Again, once some experience has been gained, it will be important to
extend these studies to ions (both cations and anions) and larger
molecules.
A third set of experiments is needed to better characterize the
physical properties of free PARs, both in the molecular and ionic form.
These include the determination of first and second ionization
potential, unimolecular reaction rates such as bond rupture and
photoisomerization (again calling for jet experiments) and reactions
with atoms such as C, 0, Nand H at low temperatures.
Lastly, it is important to realize that similar experiments should
be done on particles of various types of "amorphous" carbon. While
some of these experiments will be difficult, the questions are similar
and equally important since PABs, amorphous carbon (and graphite 1!1)
are intimately related and intermixed in the Cosmos. Unravelling their
respective importance requires that we understand their respective
properties.
POSSIBLE CARRIERS OF THE DIFFUSE INTERSTELLAR BANDS
Gerard van der Zwet
Laboratory Astrophysics, Huygens Laboratorium, Rijksuniversiteit
Leiden, POB 9504, 2300 RA Leiden, The Netherlands
ABSTRACT. Possible carriers of the Diffuse Interstellar Bands (DIB's)
are considered, with emphasis on carbon containing molecules. It is
shown that the observations indicate a molecular carrier of the DIB's
rather than impurities embedded in grains.
1. INTRODUCTION
The DIB's form a series of approximately 50 visible absorptions extending from 4430 A into the near IR. After their interstellar origin was
recognized (Merrill, 1934), the identification of the carrier of the
DIB's has become a classic problem in modern astronomy. For a historical
introduction and a review of the observations in this field the reader
is referred to the paper of Grzegorz Chlewicki in these proceedings. In
addition, several excellent reviews can be found in the literature
(Herbig, 1975; Smith et al., 1977).
In section 2 of this paper a review is given of the suggestions
which have been made until now to explain the DIB's. Section 3 shows how
observational results can be used to obtain information concerning the
general nature of the DIB carriers. Finally, in section 4 a specific
class of compounds is conSidered, i.e. molecules consisting mainly of
carbon, which are likely candidates for the DIB's.
2.REVIEW OF PROPOSED CARRIERS
It is possible that the following list of suggested carriers for the
DIB's is not complete. There is a vast amount of literature on this
subject and some papers may have escaped attention, especially those of
the earlier days. But it should give the reader a fairly good idea of
the lines along which people have been thinking about this problem
durlng the last decades.
Two general classes of possible explanations for the DIB's can be
distinguished: impuri ties embedded in grains and gas phase molecules. A
general survey of the merits and drawbacks of both models is postponed
351
A. Uger el al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 351-365.
© 1987 by D. Reidel Publishing Company.
352
G. V AN DER ZWET
until the next section. Here, the discussion is limited to specific
comments on the individual suggestions.
2.1 • Impur i ti es embedded in grai ns
The solid state hypothesis has drawn most attention because of the
following reasons:
- the generally good correlation of the equivalent widths of the DIB's
with E(B-V), the colour excess (recent studies however, indicate that
the degree of correlation varies from one line to another (Wu et al.,
1981; Chlewicki et al., 1986»,
- the lack of rotational fine structure in the lines, at least up to a
resolution of 0.05 A (Snell and VandenBout, 1981),
- no simple molecUle seems to be able to account for the lines.
Let us now turn to the list of candidates:
1. Solid oxygen (McKellar, 1955).
This involves a double electronic transition of a complex of two oxygen
molecUles and shows a coincidence in wavelength with three DIB's at
4760, 5780 and 6284 A. However, the visible spectrum of solid molecular
oxygen shows many more absorption bands, which are not observed in
space.
2. F-centers in alkali hydrides (Johnson, 1955).
(F-centers are electrons bound at negative ion vacancies in solid
crystals). The experimental linewidths are a few hundred A, much broader
than any of the DIB's.
3. Metastable H20 in or on the surface of grains (Herbig, 1963).
This suggestion is supposed to account for the 4430 A band only. A
severe drawback is that absorption is required from an excited
(metastable) state of H2 •
4. Plasma oscillations in small metallic particles (Unsold, 1964).
In this model, each DIB originates from a grain of a particUlar kind and
with a specific size (25 A < a < 100 A). The width of the lines depends
on the electrical conductivity of the material. The model is purely
hypothetical and doesn't provide a match with any of the DIB's.
5. Calcium atoms in hydrocarbon matrices (Stoeckly and Dressler, 1964;
DUley, 1968).
The principal idea is that the 4226 A absorption of the Ca atom may
shift to 4430 A and broaden significantly if the Ca is embedded in a
hydrocarbon matrix. Similarly, the 5780 and 5795 A DIB's may be ascribed
to sodium atoms. The problem with the calcium is that the band is too
broad and only at approximately the right wavelength in a solid benzene
matrix.
6. Platt particles, Polycyclic Aromatic Hydrocarbons (PAH's; Platt,
1956; Donn, 1968).
In 1956 Platt put forward the idea that solid particles smaller than
10 A, formed by random accretion from the interstellar gas, can explain
the observed interstellar extinction and polarization. These particles
will have unfilled energy bands and will therefore act like free
radicals and absorb throughout the visible region of the spectrum.
Subsequently Donn, in 1968, proposed that PAH's are the corresponding
Platt particles and that radical side chains on the PAH's may be
POSSIBLE CARRIERS OF THE DIFFUSE INTERSTELLAR BANDS
353
responsible for the DIB's. This idea was not developed further at that
time, but we'll come back to it later (section 4).
7. Pure electronic transitions associated with impurities and defects
(Wu,1972).
Wu argued that pure electronic transitions (0-0 transitions, i.e.
without any change in vibrational quantum numbers) of impurities and
defects can give rise to narrow visible transitions. The impurities and
grain materials were not specified.
8. Polyatomic molecUlar ions adsorbed on grain surfaces (DUley and
McCullogh, 1976).
MolecUles adsorbed on chemically active grain surface sites may act as
molecUlar ions and show strong visible transitions. A specific example
is benzene adsorbed on magnesium, calcium and iron protosilicates, which
shows a broad absorption between 4400 and 4500 A.
9. Transition metal ions embedded in oxides.
a) Fe 2+ and Fe 3 + ions in silicates (Manning, 1970).
This suggestion is based on the Fe 3+ absorption at 4400 A, observed in a
mlllber of terrestrial silicates.
b) Intrinsic iron ions in iron oxides (Hufmann, 1970).
This model provides a rough match of the 4430, 4760 and 4890 A DIB 's,
but predicts a strong DIB at 4160 A, which is not observed.
c) Transition metal ions in magnesium oxide (DUley 1979, 1981, 1982).
Electronic transitions of ions in defective sites have oscillator
strengths that are two to three orders of magnitude larger than in 4
perfec~+crystals. Therefore, the abundance of ions such as Cr 3+, Mn +
and Ni
embedded in MgO grains is sufficient to account for most of the
DIB's. In his 1982 paper, DUley proposed that 28 of the DIB's may be
ascribed to three vibronic systems with forbidden origins. One cannot
put much against this model, but up to date it lacks experimental
verification.
10. Free radical species in (photolyzed) dirty ices and organic residues
(van der Zwet, 1986).
This is part of the research program in the Laboratory Astrophysics
group in Leiden. The ices are prepared in two different ways: either via
slowly depositing mixtures of simple gases (e.g. H20, CO, CH 30H, CH4 and
NH3) on a 10 K substrate and simUltaneous VUV photolysis witn a
microwave powered hydrogen discharge lamp, or via passing the gas
mixtures through a microwave discharge tube and condensing the products
on the 10 K substrate'. Visible and infrared spectra were taken of the
ices and their organic residues which appear after warm up to room
temperature (Agarwal et al., 1985). The visible spectra of the residues
show continuous absorption increasing toward shorter wavelengths (van
IJzendoorn, 1985). The spectra of the ices generally show broad absorption bands which do not match the DIB's, except for two features in
oxygen rich mixtures. These are due to absorption of 02 dimers (see 1).
2.2.Gas phase molecules
The, two main problems with molecUles as possible carriers for the DIB's
are:
- the absence of rotational fine structure in the DIB's, which made
354
G. VANDERZWET
people consider line broadening mechanisms such as autoionization
(preionization) and predissociation. This creates an additional problem,
because if the species disappears upon absorption of a photon, very
efficient molecule formation mechanisms are needed in order to maintain
the required concentration of the species,
- no molecule seems stable enough to survive the harsh interstellar
radiation field.
For these reasons the list of proposed molecular carriers is rather
short:
1. Preionization of atomic ions: H-, 0-, C-, N- (Herzberg, 1955;
Rudkj6bing, 1969).
Because of their low ionization energies and high oscillator strengths
of the tranSitions, atomic ions such as H-, 0-, C- and N- were
considered possible carriers of the DIB's. Based on quantummechanical
calculations the 4430, 4760, 4890 and 6180 A bands may be ascri bed to
autoionization of hydrogen atoms, the 5780 and 5797 A bands to oxygen.
2. Predissociation of polyatomic ions or free radicals (Herzberg 1964,
1965, 1967, 197".
Herzberg suggested that predissociation of molecules such as CH4' NH4'
H30 or CH 5 might account for the DIB's. However, experiments aimed
toward producing these species in the laboratory were unsuccesful.
Later~ Ensberg et al. (1975) succeeded in obtaining the visible spectrum
of CH4' on the basis of which CH4 could be rejected as a possible
carrier of the DIB's.
3. Magnesium tetrabenzporphin (MgTBP; Johnson, 1972).
In this case all the lines are assigned to one partiCularly stable
molecule, MgTBP. However, in order to account for the DIB-spectrum,
Johnson had to invoke absorption from vi brationally exci ted states of
MgTBP for some of the lines. Moreover, it was necessary to attach two
axial pyridine ligands (or other similar molecules) to the central
magnesil.ll1 atom. As Johnson remarked himself, the ligand bonds are not
very stable and he concluded : "Consequently if the MgTBP molecule is to
survive in space, one might have to hypothesize that these molecules
predominantly eXist in a paraffin matrix under similar conditions to the
laboratory simulation experiments."
4. Carbon containing molecules.
This covers some of the more recent suggestiOns, and we'll come back to
it in detail in section 4.
3.0BSERVATIONAL CONSTRAINTS ON THE GENERAL NATURE OF THE ABSORBING
SPECIES - GRAINS VERSUS MOLECULES
We will now consider the two groups of possible carriers, discuss the
general properties of the absorption features expected for each group
and see how this fits in with the observational results.
3. 1.Impuri ties embedded in grains
a) The wi dths of the 1 ines •
Solid state absorption bands are generally broad, unless crystalline
POSSIBLE CARRIERS OF THE DIFFUSE INTERSTELLAR BANDS
355
materials are involved. There is recent observational and theoretical
evidence that most of the interstellar grain materials are amorphous
rather than crystalline (Rowan-Robinson and Harris, 1983; Gail and
Sedlmayr, 1984). Therefore, it ~~pears very unlikely that the narrow
DIB's with linewidths of - 1 em
are due to solid state absorptions. On
the other hand, some of the broader DIB's correlate extremely well with
narrow features (Chlewicki et al., 1986), thereby making a solid state
origin for the broader bands unlikely as well.
b) The prof iles of the lines.
For impurities embedded in the "classical" grains responsible for the
visual extinction (a - 0.1 ~), scattering theory predicts strongly
asymmetric extinction profiles with a steep edge and an apparent
"emission wing" (van de Hulst, 1949; Greenberg and Hong, 1976; Chlewicki
et al., 1986). Although a number of DIB's show a slight asymmetry in a
similar sense, this effect is much less pronounced than for impurities
in classical grains. Furthermore, some of the lines do not correlate
very well with the visual exctinction (Wu et al., 1981; Chlewicki et
al., 1986), and the DIB's for which the polarization has been measured
(4430, 5780 and 6284 A; Wampler, 1966; A'Hearn, 1972; Martin and Angel,
1974; Martin and Angel, 1975), do not show any polarization effect,
while the grains responsible for the visual extinction do. So the large
(0.1 ~) grains can be excluded as possible carriers for the DIB's.
Small grains (a - 0.01 ~) which are required for the extinction in
the far UV (Greenberg and Chlewicki, 1983), can be excluded as well as
DIB-carriers: scattering theory predicts symmetriC profiles or slightly
asymmetriC profiles with a steeper edge on the long wavelength side of
the banj, depending on whether one uses the Clausius-Mosotti or the
Purcell-Shapiro theory (Purcell and Shapiro, 1977) to calculate the
profiles (Chlewicki et al., 1986). Moreover, the 4430, 5780 and 6284 A
bands do not seem to show any correlation with the far UV extinction (Wu
et al., 1981; Nandy et al., 1982; Witt et al., 1983; Seab and Snow,
1984) •
It turns out that the only way to account for the DIB's with
impurities embedded in grains as far as the profiles are concerned, is
to adopt a separate population of grains with a range of sizes
(0.02 < a < O. 10 ~m) which do not correlate with any of the major grain
populations required for the observed exctinction (Chlewicki et al.,
1986).
c) The peak wavelengths of impurity absorptions depend on the chemical
composition and temperature of the grains. The constant peak wavelengths
of the DIB's require an essentially constant chemical composition and
temperature toward every object, if the DIB's arise from impurities in
grains •
From the above it is clear that the constraints, which the
observations impose on the nature of possible DIB carrying grains, are
quite stringent.
3.2.Gas phase molecules
a) Prof iles •
The absence of rotational fine structure in the DIB's implies that if
G. VAN DER ZWET
356
the DIB's arise from gas phase molecules, the number of atoms in these
molecules must be sufficiently large, so that the rotational structure
remains unresolved (Danks and Lambert, 1976). This will result in socalled rovibronic band contours of which a few examples are shown \n
figure 1. Note that these spectra were obtained at room temperature and
therefore are not directly applicable to the physical situation in the
interstellar medium. However, in general a variety of profiles is
observed, depending upon the particular type of transition (A, B or Cband; Hollas, 1973), and for the narrow lines the profiles are Slightly
asymmetriC, in the same sense as for the DIB's. This asymmetry arises
from the difference between the rotational constants in the ground and
excited electronic state involved in the transition (Steinfeld, 1976).
(a)~
~O=O
--!"-
~
f
(b)
p-BENZOQUINONE
:1L
-1
CO
~
~
NAPHTHALENE
(e)
CO
~
~
~
INDENE
(d)
,
-10
---:!l
I
o
,
cm- 1 10
CO
II.ZULENE
Figure 1: Rotational band contours measured for some planar, cyclic
molecules. All contours are presented on a common frequency scale which
increases to the right. Actual band positions are a) p-benzoquinone-4764
and 4761 A; b) naphthalene-3081 and 3123 A; c) indene-2880 A; and d)
azulene-3478 A. These have been reproduced with permission from Ross,
Adv. Chem. Phys. 20 (1971), 341.
POSSIBLE CARRIERS OF THE DIFFUSE INTERSTELLAR BANDS
357
b) Widths.
The linewidths of electronic transitions of large gas phase molecules
can be as narrow as - 1 cm- 1 under interstellar conditions (van der Zwet
and Allamandola, 1985). However, line broadening mechanisms such as
internal conversion (IC) may be important. IC is the process in which
interaction between the particular level excited and nearby levels of a
lower electronic state resul ts in radiationless decay and a lifetime
broadening of the absorption. The rate of IC varies greatly from one
level to another and may be as fast as 10- 13 s (Byrne and Ross, 1971),
resulting in a width of - 30 cm- 1 , which is comparable to those of some
of the broader DIB's. The broadening is Lorentzian, so the lines tend to
lose their "intrinsic" asymmetry. It was Douglas (1977), who fi~rst
suggested IC as the mechanism which may account for the observed widths
of the DIB's. It is worth mentioning that the broader DIB's tend to lie
at shorter wavelengths, as was noted by Herbig (1975). This is
consistent with the idea of IC, because the density of states within a
molecule grows very rapidly with energy, thus increasing the probability
of IC.
c) The peak wavelengths are obviously constant for gas phase molecules
under a variety of physical conditions.
d) The absence of polarization is also easy to explain for molecules,
because there's no efficient alignment mechanism known for molecules in
the interstellar medium.
The conclusion is therefore, that the observational constraints are
quite easily fulfilled within the framework of the molecular hypothesis.
4.CARBON CONTAINING MOLECULES
In this section a specific class of compounds is considered, namely
molecules containing (primarily) carbon, which are attractive candidates
for the solution of the longstanding DIB mystery. The choice for carbon
molecules seems, in addi tion to the arguments presented below for each
individual case, natur.al because of two reasons: first, carbon is an
abundant element and readily available in interstellar space, and
second, carbon bonds tend to be quite stable and thus favorable under
interstellar conditions (VUV radiation).
Several possibilities will be discussed now:
4.1.Carbon chains
These were proposed by Douglas (1977). The observation of the
cyanoacetylenes, HCnN (n=1,3,5 ••• 11) in some dense molecular clouds
(Mann and Williams, 1980; Bell and Matthews, 1985) made Douglas suggest
that long chain acetylenes are also present in space and may be even
more abundant than the corresponding cyanoacetylenes. Furthermore, he
hypothesized that in the diffuse medium these chains are devoid of
hydrogen and exist as pure carbon chains (C n ).
Recently, Kr~tschmer et al. (1985) obtained UV and visible spectra
of carbon molecules, produced by diffusi ve coagul ation of condensed
carbon vapor in an argon matrix. They observed strong absorptions
358
G. VAN DER ZWET
(f - 0.2 - 0.3), which they ascribed to electronic transitions in carbon
chain molecules. However, the interpretation of the corresponding IR
spectra presents some difficulties as to the exact nature of the
absorbing species (Krtitschmer, this volume).
The two main attractions of the carbon chains model are:
- it may explain the fact that no DIB's are observed shortward of
4430 A and that fewer lines are observed in the near infrared: the
Shorter chains will absorb in the blue and near UV and may not be stable
against UV radiation and dissociate. The longer chains will be stable
and absorb more to the red, but the abundance will decrease after a
certain chain length, resulting in a long wavelength end of the spectrum
as well,
- the visible transitions are strong (at least in the experiments of
Kl'1itschmer et al.).
Drawbacks of the carbon chains hypothesis are the fact that the
cyanoacetylenes are only observed in a nLlDber of dense molecular clouds
and not in the diffuse medium , and the questionable stability of carbon
chains against VUV radiation.
4. 2. Pol ycycl1c Aromatic Hydrocarbons (PAH's)
These molecules were put forward by several people as possible carriers
of the DIB's (Leger and d'Hendecourt, 1985; van der Zwet and Allamandola, 1985; Crawford et aI., 1985), after i t appeared that the infrared
spectra of PAH's reasonably match the Unidentified InfraRed (UIR)
features, which are observed in a wide variety of objects in the Galaxy
and in Extragalactic regions, submitted to a high UV flux (Leger and
puget, 1984; All am an dol a et aI., 1985).
PAH's are attractive candidates for explaining the DIB's because of
the following reasons:
- the carbon skeleton of PAH's is extremely stable with respect to
photodissociation, due to delocalization of the TI electrons,
- the abundance of PAH's deri ved from the regions where the UIR bands
are observed is quite high (see below),
- they show visible transitions, particularly under interstellar
conditions, when they may be singly positively ionized (see below) and
partially dehydrogenated: the first electronic transition of a PAH
cation generally lies in the visible or near IR (Crawford et al., 1985),
which also holds for the partially dehydrogenated radical species.
The fraction of the PAH's that is ionized can be estimated as
follows: the first ionization potential of PAH's is 6-8 eV (Gallegos,
1968; Clar and Schmidt, 1977, 1978, and references therein). (The second
ionization potential is much higher, about 20 ev.) Let's take coronene
(C24H12) for which I - 7.3 eV as a typical example. As a crude (I)
approximation it is assLlDed that every absorbed photon with an higher
energy than I ionizes the molecule. First the nLlDber of ionizing photons
is calculated by integrating the interstellar photon flux F(E) between
7.3 and 13.6 eV, the ionization potential of hydrogen. The expression
for F(E) is taken from Draine (1978). This gives for the number of
ionizing photons:
POSSIBLE CARRIERS OF THE DIFFUSE INTERSTELLAR BANDS
F
359
}3.6
= J
7
3
Assuming an average cross section of 10- 16 cm 2 for coronene (Clar,
1964), yields the following value for the ionization rate:
The dominant neutralization channel shoUld be radiative recombination.
An upper limit to this rate is given by the electron-ion collision
rate, nea, where a is the recombination coefficient which is taken to be
6 x 10-7 (T/300)-0.5 cm 3s- 1
(Prasad and Huntress, 1980),
and ne the electron number density. In the diffuse medium the degree of
ionization ~£ the gas is determined by the carbon content, therefore
ne - 3 x 10
n H. Taking nH and T equal to 10 cm- 3 and 100 K
respectively, yields
Rrec - 3.1 x 10-9 s-l
From the calcUlated values for the ionization and recombination rate it
follows that approximately 80% of the coronene is ionized in the
interstellar medium.
The column density of PAH's estimated from the UIR bands with
respect to hydrogen is - 2 x 10-7 (Allamandola et al., 1985). Suppose
that the abundance of PAH's in the diffuse medium is the same and that
the complete DIB spectrum is accounted for by 10 ionized PAH molecUles.
The column density of a PAH cation which agcounts for one particUlar DIB
wi th respect to hydrogen is then - 2 x 10- • In general, the number of
absorbers needed for one DIB can be calcUlated using the well known
relation:
(Spitzer, 1978),
where WA is the equi valent width of the line, e 2 /mc 2 the clasSical
radius of the electron, AO the peak wavelength, NA the column densi ty of
absorbers and f the oscillator strength. Using
(Spitzer, 1918).
one finds
360
G. VAN DER ZWET
for a strong DIB. If f - 0.1 ~a typical value for a strong electronic
transition), then NA/NH - 10- , similar to the estimated column density
for an ionized PAH molecule in the diffuse medium. So if the abundance
of PAH's calculated from the UIR bands is extrapolated to the general
diffuse medium, the number of ionized species is sufficient to account
for the DIB' s.
A number of problems remain though: we have assumed that PAH's are
stable molecules in the diffuse interstellar medium. Although PAH's are
not very susceptible to photodissociation, chemical reactions between
PAH's and atoms and ions in the interstellar gas may provide a variety
of destruction channels (Duley and Williams, 1986).
The second question one might ask is why only about 50 DIB's are
observed: a surprisingly small number if PAH's account for the DIB's, in
view of the many possible species. The answer may be that there is a
selection of the most stable molecules in the interstellar medium. In
general the most stable PAH's have the most "condensed" configuration
(Stein, 1978), and among the PAH's there's a subclass called "superaromatics" which are extremely stable (Clar, 1972). Another possi bili ty
is that many bands are too weak to be detected, for instance because of
their broadness.
Finally, perhaps the most seriOUS problem is provided by the UV
spectra of PAH's: they generally show strong absorption between 3000 and
4000 A, with e: > 10 5 1Mole- 1cm- 1 ( 0 ) 10- 1b cm 2 , Clar 1964), while there
are no DIB's observed shortward of 4430 A. On the other hand, the bands
may be difficult to detect in the UV because of their broadn.ess, the
contamination by stellar lines in the spectrum, and because of problems
associated with setting the continuum level in the UV. Furthermore,
there are not many data available on the UV spectra of PAH cations.
4.3.Highly unsaturated organic molecules
The products of the VUV photolysis of simple organic molecules
containing 6-8 heavy atoms (mostly carbon, but other atoms such as
nitrogen and probably silicon are also possible). have been proposed as
a possible origin of the 2200 A hump (van der Zwet et al •• these
proceedings). These product molecules are photostable and all show
strong absorption around 2200 A. Although the molecules have not been
identified yet. the infrared spectra indicate that hydrogen atoms are
lost from the starting molecules during photolYSiS, so that the products
must be highly unsaturated.
If these molecules indeed account for the hump. this will have
consequences for the DIB's as well, because the DIB's and the hump are
both seen in the (same) diffuse medium. Like for the PAH's. the positive
ions will show visible transiti~ns. The equivalent width of the hump per
unit colour excess is 450 Amag- (Nandyet al., 1975). Using an
oscillator strength f - 3. th~ column density of absorbers for the hump
relative to hydrogen is - 10- • Assuming an ionization potential of - 10
eV (compare benzene: 9.24 eV; acetylene: 11.4 eV (eRC Handbook of
Ghemistry and PhysiCS», and a cross section of 10- 17 cm 2 for these
molecules (they do not absorb very strongly in the far UV if they are
POSSIBLE CARRIERS OF THE DIFFUSE INTERSTELLAR BANDS
361
responsible for the hump), a similar calculation as for the PAH's
results in - 10% of the molecules being ionized. Therefore, the column
density of ionized species with respect to that of hydrogen is - 10- 7 ,
sufficient to account for the DIB's. Thus, although we do not yet know
what the visible spectra of these ions look like, a simple calculation
shows they should be considered as possible candidates for the DIB's.
~.~.Other
carbon containing molecules?
There is yet a fourth possibility, that another kind of carbon species
forms the source of the DIB's. Carbon chemistry is extremely rich (in
fact, we have seen some beautiful examples during this workshop), and
all molecules which may exist under interstellar conditions should be
considered.
5.EPILOGUE
The question: what gives rise to the DIB's remains unanswered, despite
all efforts. The observations strongly favor gas phase molecules as the
origin of the DIB's, although impurities embedded in grains cannot be
excluded completely.
Carbon containing molecules are likely candidates in view of the
abundance of carbon and the stability of carbon based species. Various
possibilities exist and a lot of laboratory work needs to be done in
order to test the various hypotheses. This may look like a hopeless task
because of the seemingly unlimited amount of possible molecules. On the
other hand, if we don't try we'll never learn the ultimate answer about
one of the most baffling problems of today's astronomy.
6. ACKNOWLEDGEMENT
The support of the "Stlchting voor Fundamenteel Onderzoek der Materie
(F.O.M.)" is gratefully acknowledged.
7. REFERENCES
Agarwal, V.K., Schutte, W., Greenberg, J.M., FerriS, J.P., Briggs, R.,
Connor, S., van de SUIt, C.E.P.M., Baas, F.: 1985, Origins of Life
16, 21
Bell, M.B., Matthews, H.E.: 1985, Astrophys. J. Letters 291, L63
A'Hearn, M.F.: 1972, Astron. J. 77, 302
Allamandola, L.J., Tlelens, A.G.G.M., Barker, J.R.: 1985, Astrophys. J.
Letters 290, L25
Byrne, J.P., Ross, I.G.: 1971, Australian J. Chern. 2ll, 1107
Chlewlcki, G., van der Zwet, G.P., van IJzendoorn, L.J., Greenberg,
J.M., Alvarez, P.P.: 1986, Astrophysical J. (in press)
Clar, E.: 196~, Polycyclic Hydrocarbons, AcademiC Press, London
Clar, E.: 1972, The Aromatic Sextet, J. WlleyandSons, Ltd., London
362
Go VAN DER ZWET
Clar, E., Schmidt, W.: 1977, Tetrahedron 33, 2093
Clar, E., Schmidt, W.: 1978, Tetrahedron 3-, 3219
Crawford, M.K., Tielens, A.G.G.M., Allamandola, L.J.: 1985, Astrophys.
J. Letters 293, L45
Danks, A.C., Lambert, D.L.: 1976, Monthly Notices Roy. Astron. Soc. 17',
571
Donn, B.: 1968, Astrophys. J. Letters 152, L129
Douglas, A.E.: 1977, Nature 269, 130
Draine, B.T.: 1978, AstrOPhys. J. Suppl. Ser. 36, 595
Duley, w.w.: 1968, Nature 218, 153
Duley, W.W.: 1979, AS"trOj)hys. J. 227,824
Duley, W.W.: 1981, Astrophys. Space Sci. 78, 145
Duley, W.W.: 1982, Astrophys. Space Sci. 88, 501
Duley, W.W., McCullogh , J.D.: 1977, Astrophys. J. 211, L145
Duley, W.W., Williams, D.A.: 1986, Monthly Notices ROy. Astron. Soc.
219, 859.
Ens berg , E.S., Jefferts, K.B.: 1975, Astrophys. J. Letters 195, L89
Gail, H.P .. Sedlmayr, E.: 1984, Astron. Astrophys. 132, 163
Gallegos, E.J.: 1968, J. Phys. Chem. 72, 3452
Greenberg, J.M., Chlewicki, G.: 1983, Astrophys. J. 272, 563
Greenberg. J.M., Hong, S-S.: 1976, Astrophys. Space SC1. 39, 31
Herbig, G.H.: 1963, Astrophys. J. 137, 200
Herbig, G.H.: 1975, Astrophys. J. 196,129
Herzberg, G.: 1955, Mem. Soc. R. Sci. Liege, 4eme Ser., 15, 291
Herzberg, G.: 1964, Publ. Roy. Obs. Edinburgh 4, 67
Her zber g, G.: 1965 , J. Opt. Soc. Am. 55, 229
Herzberg, G.: 1967, IAU Symposium no. 31, ed. H. van Woerden, Academic
Press, London and New York, 91
Her zberg, G.: 1971, Highlights in Astronomy 2, 415
Hollas, J.M.: 1973, Chapter 2 in Molecular Spectroscopy: A Specialized
Periodical Report Volume 1, The Chemical Society, London, 63
Huffman, D.R.: 1970, Astrophys. J. 161, 1157
Johnson, F.M.: 1955, Radiation Laboratory Progress Report, 30
Johnson, F.M.: 1972, Mem. Soc. R. Sci. Liege, 6eme Ser., 3, 91
Kr§tschmer, W., Sorg, N., Huffman, D.R.: 1985, Surface Science 156, 814
Leger, A., d'Hendecourt, L.: 1985, Astron. Astrophys. 146, 81
Leger, A., Puget, J.L.: 1984, Astron. Astrophys. 137, L5
Mann, A.P.C., Williams, D.A.: 1980, Nature 283, 721
Manning, P.G.: 1970, Nature 226, 829
Martin, P.G., Angel, Ji.R:P7: 1974, Astrophys. J. 188, 517
Martin, P.G., Angel, J.R.P.: 1975, Astrophys. J. 195, 379
McKellar, A., Welsh, H.L., Stephenson, F.C.: 1955, Astron. J. 50, 170
Merrill, P.W.: 1934, Publ. Astron. Soc. Pac. _6, 206
Nandy, K., Morgan, D.H., Houziaux, L.: 1982, Astrophys. Space Sci. 85,
159
Nandy, K., Thompson, G.I., Jamar, C., Monfils, A., Wilson, R.: 1975,
Astron. Astrophys. 4_, 195
Prasad, S.S., Huntress, W.'1°., Jr.: 1980, Astrophys. J. Suppl. Sar. '3,
Purcell, E.M., Shapiro, P.R.: 1977, Astr0phys. J. 214, 92
Rowan-Robinson, M., HarriS, S.: 1983, Monthly Notices Roy. Astron. Soc.
202, 767
POSSIBLE CARRIERS OF THE DIFFUSE INTERSTELLAR BANDS
363
Rudkj6bing, M.: 1969, Astrophys. Space Sci. 3, 102
Seab, C.G., Snow, T.P.: 1984, Astrophys. J. 277, 200
Smith, W.H., Snow, T.P., York, D.G.: 1977, Astrophys. J. 218,124
Snell, R.L., Vanden Bout, P.A.: 1981, Astrophys. J. 244, 844
Spitzer, L., Jr.: 1918, Physical Processes in the Interstellar Medillll,
J. Wiley and Sons, Ltd., New York
Stein, S.E.: 1918, J. Phys. Chern. 82, 566
Steinfeld, J.I.: 1918, Molecules and Radiation, MIT Press, Cambridge,
Mass., p 121-128
Stoeckly, R., Dressler, K.: 1964, Astrophys. J. 139, 240
Uns6ld, A.: 1963, Z. Astrophys. 56, 221
Uns6ld, A.: 1964, Pub. Roy. Cbs. Edinburgh 4,35
van de Hulst, H.C.: 1949, Rech. Astr. Cbs. Utrecht 11, part 2
van IJzendoorn, L.J.: 1985, Ph.D. Thesis, University of Leiden, The
Netherlands
van der Zwet, G.P.: 1986, Ph.D. TheSiS, University of Leiden, The
Netherl ands
van der Zwet, G.P.: 1986, Allamandola, L.J.: 1985, Astron. Astrophys.
146, 16
Wampler, E.J.: 1966, Astrophys. J. P'II, 921
Witt, A.N., Bohlin, R.C., Stecher, T.P.: 1983, Astrophys. J. 267, L47
Wu, C-C.: 1912, Astrophys. J. 178, 681
Wu, C-C., York, D.G., Snow, T.P.: 1981, Astron. J. 86, 155
364
G.VANDERZWET
DISCUSSION FOLLOWING VAN DER ZWET
Wdowiak: A comment on the speaker's placing our experiment in the
category of linear molecular candidates for the DIB's: our 1980
experiments (Astrophys. J. 241, L55) while inspired by the hypothesis of
Douglas (1977, Nature 269, 130) and the use of methane as a reagent
suggested by the reactions of Mitchell and Huntress (1979, Nature 278,
722), is really not wedded to a particular hypothesis. It was an attempt
to produce DIB candidates in the laboratory through the production of a
carbon containing free radical "soup" using the inert gas matrix
isolation technique. Candidates for 10 DIB's were suggested.
Cox: Just a comment: to test your unsaturated organic molecules as
being at the origin of the 2200 A hump, it would be nice to have some
polarization measurements through this feature.
van der Zwet: Indeed. As far as I know there has been only one
attempt to measure linear polarization across the hump, by Gehrels
(1974, Astron. J. 79, 590). His results, limited to two stars only,
indicate that the carrieres) of the hump is (are) not aligned. However,
the quality of the data is not very high, and further observations would
certainly be useful.
d'Hendecourt: About the last solution (highly unsaturated organic
molecules) you proposed: don't you expect that, because these molecules
produce the 2200 A hump, there will be a very good correlation between
the 2200 A hump and the DIB's?
van der Zwet: Not necessarily. The DIB's are likely due to ionic
speCies, in contrast to the hump which is probably caused by absorption
of neutral molecules, as our experiments indicate. The fraction of the
molecules that is ionized (- 10~) depends on the local radiation field.
Therefore, one does not really expect a tight correlation between the
hump and the DIB's.
d'Hendecourt: You suppose that the DIB's are due to ions? There is
some degree of correlation between the 5780 A band and E(B-V), even for
high E(B-V) values. If the extinction is high, shouldn't one expect the
molecules to become neutral and the correlation to break down?
Jura: To restate d'Hendecourt's question, some lines of sight show
so much extinction that it is difficult to imagine that the bands are
carried by ionized species. One would expect most of the molecules to be
neutral.
van der Zwet: Strictly speaking, the reddening is a measure of the
total amount of dust in the line of Sight. In other words: a high E(B-V)
value may simply mean a large path length. What one should really look
at is the correlation between the number density and the DIB's, which is
not easy. In fact, there are some indications that the DIB's are weaker
in dense clouds: in the paper of Snow and Cohen (1974, Astrophys. J.
194, 133), a deficiency with respect to color excess in the lines of
Sight to stars lying behind dense interstellar clouds is reported for
the 4430, 5780 and 5797 A bands. A similar effect is observed for the
5191 A band in Cyg OB2 #12 (Chlewicki et al.: 1986, Astrophys. J., in
press) •
The ionization equilibrium should also be taken into acocunt: if
the molecules giving rise to the DIB's are 100~ ionized, then an
increase in the density does not necessarily result in weaker DIB's.
POSSIBLE CARRIERS OF THE DIFFUSE INTERSTELLAR BANDS
365
Roche: What is the status of the DIB's in the Magellanic Clouds?
van-der Zwet: The Magellanic Clouds are characterized by UV
extinction shortward of 2000 A which is considerably higher than in the
Galaxy. Nandy et al. (1982, Astrophys. Space Sci. 85, 159) have studied
the 4430 A DIB for a few stars in the Large Magellanic Cloud and didn't
find any link between the far UV extinction and the strength of the
4430 A band. However, I would like to remark that:
the number of objects they looked at is limited, and
the reddening of these objects is quite lOW, so that the uncertainties in the central depths of the 4430 A band are substantial.
DISCUSSION II I:
THE DIFFUSE INTERSTELLAR BN{DS -- ARE THEY CARRIED BY PAH'S?
M. Jura, Chairman
Astronomy Department
UCLA
Los Angeles CA 90024
USA
Jura: Yesterday, we heard from van der Zwet arguments that the diffuse
interstellar bands are carried by PAH's. Today, I would like to express
some reservations about this idea. First, with the excellent CCD
optical obsrevations of the diffuse interstellar bands presented here by
the Leiden group, it seems that there is a moderately good correlation
between the equivalent width of the feature at 5780 A and the amount of
dust measured by E(B-V). That is, while there is a scatter by about a
factor of 2 in this ratio, the results from Leiden indicate that for
reddened stars with E(B-V) up to 3.0, WA(5780 A)/E(B-V) - 200 rnA/mag.
In a very different sample of stars, those with E(B-V) < 0.1 mag, a
factor of 30 less dust, both Meyer (1983, Ap. J. Letters, 266, LSI) and
Federman, Kumar and van den Bout (1984, ~~, 282, 485) find that
WA(5780 A)/E(B-V) = 300 rnA/mag. Therefore, over a very wide range of
physical conditions, there is probably better than a factor of two
correlation between the amount of dust and the concentration of the
carrier of the diffuse band. This is 'empirical evidence against the
idea that the bands are carried either by a netural or ionized PAR,
because we would might expect a fairly wide variation in the amount of
any relatively minor molecule within the interstellar medium.
Greenberg: In our data, within a particular stellar association, there
is absolutely no correlation between E(B-V) and WA(5780 A).
Jura: That is true at the factor of 2 level. However, for a range of
extinction of over a factor of 30, there is a correlation to better than
a factor of 2.
Leger: Is it not a question of distance; the stars studied by the
Leiden group are simply further away.
Jura: This is only part of the story, there is not a factor of 30 in
distance but perhaps only a factor of 5.
367
A. Uger et aI. (etls.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 367-370.
© 1987 by D. Reidel Publishing Company.
368
DISCUSSION III
Jortner: What about negative ions?
Jura: In diffuse clouds, most PAR's will not have a negative charge (see
the paper by Omont).
Allamandola: There are positive ions such as sodium in the diffuse
clouds.
Jura: Because of the wide range in the amount of extinction, there is a
wide range of radiation fields in the different clouds. In view of the
calculations by Omont, it seems most unlikely that the PAR's would be
mostly ionized everywhere.
Omont: I will discuss later the ionization processes of PAH's. The main
uncertainty is the photoionization cross section. In diffuse clouds,
the amounts of neutral and ionized PAR's are comparable.
d'Rendecourt: In reflection nebulae, the radiation field is so high
that most of the PAR's are probably ionized.
Jortner: There is information for the photoionization cross section for
some PAR's. The rate of photoionization depends upon the rate of
absorption of photons and the yield of photoelectrons.
Leach: An ionization yield of 1 is not reached easily. For big
molecules, the yield may be smaller than the smaller hydrocarbons.
Greenberg: In our observations, we tried to select stars in which the
line of sight is really in the diffuse medium. The local extinction is
not high in these clouds. Jura's argument about substantially varying
radiation fields does not really hold. The radiation field is not
reduced by more than a factor of 10.
d'Rendecourt: The close correlation between WA(5780 A) and E(B-V) may be
used to discriminate about the state of ionization of PAR's.
Leach: Doubly ionized PAR's may undergo Coulomb cracking or Coulomb
explosions.
Jortner:
Duley:
What is the 2nd ionization potention of PAR's?
What about cosmic rays; they should also be considered.
Jura: To me what is certainly a clue to the nature of the carriers, is
that wherever you look, it seems that the amount of this material is
within a factor of 2 of the amount of dust. Another reason to doubt
that the diffuse interstellar bands are produced by carbon-bearing
gas-phase molecules such as PAR's is that with the relatively crude data
that are available, they do not appear in the spectra of mass-losing
carbon stars. For example, Cohen and Schmidt (1982, Ap. J., 259, 693)
display an optical spectrum of IRC+10216 which, despite its low
resolution, does not show any evidence for the broad interstellar
DISCUSSION III
369
feature at 6284 A. This feature should have been observed considering
the very large amount of dust around this star if the carrier of the
band is mixed in with the out flowing material.
Allamandola:
What about the 2200 A in this star?
Jura: There is too much dust; it is impossible to measure.
Kroto:
Is it possible that 6284 A is lost in the noise?
Jura: Most of these features are probably real rather than noise.
Leger: Molecules such as CO/H2 vary a lot because they are very
sensitive to photodissociation. If this is not the case for PAH's, this
argument does not hold.
Jura: There are other processes to destroy PAH's such as accretion onto
grains and shock waves.
Leger: The 12 ~m emission in IRAS which is probably due to PAH's shows
a very constant 12/100 ~m flux ratio.
Jura: That result only refers to PAR's as a class; it does not
discriminate among specific molecules.
Tramer: You have to face not only the 0-0 transition but the other
transitions as well. You should find a vibrational progression.
Jortner: The 0-0 is very strong and probably very dominant.
worry too much about the vibrational progression.
Don't
Leach: The sudden ending of the spectrum at 4428 A is an argument
against large molecules. The diffuse interstellar bands are narrow, and
any aromatic should have broader structures in the near ultraviolet and
ultraviolet. We don't see that.
Jortner: Tetracene has a very narrow band at 0-0; the other transitions
are very broad.
Leger: There is stronger absorption in the interstellar extinction
curve in the ultraviolet.
d'Hendecourt:
ultravi.olet.
How difficult is it to look for diffuse bands in the near
Jura: Herbig looked there in spectra taken from the ground and never saw
anything.
Wdowiak: We have performed experiments which can reproduce diffuse bands
in the laboratory.
DISCUSSION III
370
van der Zwet: Most of your bands except that at 4430
A are
02 lines.
Wdowiak: I do not believe it.
Leach: Do these features persist if you warm up your sample?
Wdowiak: Yes'
Jortner: The 4428 A feature is very intense and it is very peculier.
For PAR's the 0-0 transition is typically a very sharp a transition with
f = 10-4 • The p transitions are stronger. I don't think 4428 is a p
transition because we don't see any vibrational structure. The So - S2
transitions are very broad so we won't see any thing in the ultraviolet;
it would be smeared out.
Leach: Forget PAR's. There is one diatomic, SiC, whose spectrum has
never been obtained in the lab.
Jortner:
Why should it be broad?
Jura: SiC could be very abundant in some regions.
Leger:
There should be some rotational structure.
van der Zwet: If the 4430 A feature is due to So - S2, it would imply a
very large molecule.
Leger: We need a transition with a large f value, and an a transition
won't work because of the abundance constraint.
Martin: I am curious about the broad and narrow features; they appear
to go together. We need species to produce both.
d'Rendecourt: If we have an ion, the electronic transitions are red
shifted and much stronger than in the neutrals.
Jortner: Yes, definitely.
Leger: Do you agree that the PAR's ions have higher f values for their
transitons?
Jortner: Yes
PHYSICS AND CHEMISTRY OF INTERSTELLAR POLYCYCLIC AROMATIC MOLECULES
Alain OMONT
Groupe d'Astrophysique (U.A. CNRS 708)
Observatoire de Grenoble. Universite de Grenoble I
B.P. 68 - F38402 Saint Martin d'Heres Cedex
ABSTRACT. The properties and the behaviour of polycyclic aromatic
molecules (PAH, mainly hydrocarbons with 20-100 carbon atoms) are
discussed in different interstellar environments. Their charge is
regulated by the same mechanisms as that of interstellar grains.
However, it is mainly limited in practice to a single elementary
charge, positive or negative ; several states of charge very often
coexist. In molecular clouds their large polarizability and
photodetachrnent are important in determining their charge. In dense
clouds they can play an important role in the ionization by bearing
a significant fraction of the negative charge and as a sink for
molecular and especially atomic positive ions. Their temperature is
not well defined when their internal energy Ei is small, because the
low density of energy levels prevents an efficient energy
redistribution; however, the latter is achieved when Ei ~0.1 eV.
They remain in their ground vibrational level most of the time
between high temperature spikes following absorption of UV or
visible photons, ion recombination or other reactions with gas
particles. The rate of photolysis of different atoms is estimated by
the theory of molecular reactions and their lifetime in interstellar
radiation fields is evaluated. Direct photolysis of H atoms can be
important. Their periphery is probably often not completely saturated
by H atoms ; they therefore can contain radical sites, and possibly
hetero-atoms and non hexagonal cycles. Physisorption on the lattice
surface is probably unimportant ; chemisorption there and hence of
sticking of gas particles is uncertain. The main reactions with the
gas are accretion of atoms on peripheral radical sites, and reactions
with positive ions. C+ can lead both to condensation or to sputtering.
The generation, growth and destruction mechanisms are discussed.
PAH's with less than ~ 20 carbon atoms are photolysed by UV radiation.
Larger ones are mainly destro~ed in shocks and in the hot gas on
time scales of about a few 10 years comparable to those of grains.
One possible generation mechanism is by cleavage of carbon grains
from grain-grain collisions in moderate shocks. The possibility of
synthesis from small carbon molecules is also discussed. Growth by
ac'cretion of gas particles, mainly C+, can be relatively fast, if
371
A. Ug" et al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 371 -372.
© 1987 by D. Reidel Publishing Company.
A.OMONT
condensation is more important than sputtering in C+ reactions.
Accretion onto grains in molecular clouds is slowed by the negative
charges ; it proceeds at a rate comparable to that of grain
coagulation. Desorption of accreted PAH's is probably efficient in
moderate shocks.
To be published in Astronomy and Astrophysics (1986).
FORMATION, DESTRUCTION AND EXCITATION OF CARBON GRAINS AND PAR MOLECULES
W.W. Duley
Physics Department
York University
Toronto, Ontario
Canada M3J IP3
ABSTRACT. The nature of grains in diffuse clouds is reviewed with
particular emphasis on very small grains. An attempt is made to correlate the properties of very small grains and large molecules of the
PAR type. It is found that the transition from particle to molecule
likely occurs for systems containing about 100 carbon atoms and involves no discontinuity in physical parameters. Possible formation and
destruction routes for PAR under interstellar conditions are also
discussed. It appears that the lifetime of PAR in clouds will be
limited by photo-oxidation type reactions. The most plausible formation route for PAR seems to be via grain disruption in interstellar
shocks. This implies that gaseous PAR molecules may exist only in
localized regions of high excitation. Hydrogenated amorphous carbon
(aC:H or HAC) grains are the dominant carbon condensate in interstellar
clouds and their excitation can lead to both broad band and narrow-line
luminescent emission in interstellar sources. In addition, IR absorption by HAC grains provides a good simulation of the 3.4~m feature seen
in spectra of the galactic centre. The relation between PAR molecules
and HAC is discussed in an evolutionary model of dust grains.
1.
INTRODUCTION
The space density of interstellar dust in our galaxy is ~3Mapc-2. With
a formation rate from sources such as planetary nebulae, red giant outflows, novae and supernovae of about 1.3 x 10-9 Me pc- 2 yr- l , (Dwek and
Scalo 1980) this implies that ~2.5 x 109 yr would be required to
'refill' our galaxy with dust to the observed density in the absence of
competing destruction mechanisms. But dust is continuously destroyed
in the interstellar medium (ISM); for example, the loss rate of dust
due to star formation alone is estimated to be some 3 x 10-9 Me pc-2
yr-l (Dwek and Scalo 1980). Dust can also be destroyed during cloudcloud collisions as well as in the blast-waves from supernova explosions (Martin 1978)'. In general, then, the destruction rate of IS
dust is considerably larger than its formation rate if one considers
only formation in stellar or nebulae sources. This implies that a
373
A. Uger et aI. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 373-386.
@ 1987 by D. Reidel Publishing Company.
W.W.DULEY
374
mechanism must exist for the growth of IS grains in interstellar clouds.
Such heterogeneous condensation occurs when atoms and ions from
the gas in IS clouds collide with and stick to pre existing dust grains.
Such a reaction for an element X can be written
x+
grain ~> X - grain
(1)
where k is the rate constant = SxTIa2vx where Sx = sticking efficiency,
a = grain radius and Vx = thermal speed of X in gas. For large grains
with a ~ O.l~m, k = 10-6 cm3 sec-l while for small grains with a ~
O.Ol~m, k = 10-8 cm 3 sec-I. This implies a timescale, T, for grain
growth via condensation of carbon
T
~
yr
--;;n--
where n = density of hydrogen nuclei. With n = 10 2 in diffuse clouds,
T ~ 3 x 108 yr a timescale that is comparable to cloud-cloud collision times (Martin 1978). Thus the mass of IS material in dust may
increase in interstellar clouds with the result that the grain forming
elements will be observed to be depleted from the gas in such clouds.
Such depletions are commonly observed (Cowie and Songaila 1986).
2.
INTERSTELLAR DEPLETIONS AND DUST COMPOSITION
The depletion of an element X in the IS gas arises when X is incorporated in dust grains. This depletion can be due to the superposition of
two effects. Firstly, X can be depleted by its inclusion in dust cores
ie. in the nuclei that act as condensation centres for heterogeneous
accretion. Secondly, heterogeneous condensation on these cores leads
to further depletion of X. The logarithmic depletion of X is defined
as
log
t
Nx ] - log
~
0
[:~J
(2)
H
(3)
where Nx , etc., are column densities while ng is the density of grains
in the ISM and Dx(t=O) is the depletion prior to the onset of dynamic
accretion in the ISM (or after the last shock). The first term in
equation 3 measures the dynamic accretion after time t. A summary of
Dx(t) for the grain forming elements taken from the compilation of
Cowie and Songaila (1986) is given in table I.
FORMATION, DESTRUCfION AND EXCITATION OF CARBON GRAINS AND PAH MOLECULES
TABLE I.
Element
C
N
°
Mg
Si
Fe
375
Depletions, -Dx of grain forming elements
Cold
Hot
0.59 ± 0.06
1.35
2.05 ± 0.06
0.27 ± 0.04
0.75
1.43 ± 0.05
Average
0.29 ± 0.6
0.09 ± 0.4
0.21 ± 0.3
Table I shows that only the refractory elements such as Mg, Siand
Fe are strongly depleted in diffuse clouds while N is likely undepleted
and C and
may be only slightly depleted. Since Mg, etc tend to combine with oxygen to form refractory oxides and silicates what little
oxygen depletion there is, is likely due to the formation of these
solids. In fact, a comparison of oxygen and Mg, Si, Fe column densities in dust using the data in table I shows that the observed
depletion is compatible with the existence of Mg and Fe silicates of
nominal composition Mg2Si03 ~ MgSi0 3 and Fe 2Si0 4 ~ FeSi0 3 • Support
for the existence of silicates of this type come from a variety of
observations (see for example Aitken et al. 1979). A recent laboratory
study also shows that absorption in small silicate particles provides
an excellent fit to the 217.5 nm extinction bump (Steel and Duley 1986),
suggesting that silicates are abundant in diffuse IS clouds; These
silicate particles constitute the refractory cores on which subsequent
dynamic accretion occurs in the ISM.
In a recent analysis of the depletion of Ar and other elements
Duley (l985a) showed that, along some lines of sight, carbon and Ar are
both undepleted while background depletions of Mg, Si and Fe are observed. The subsequent dynamic completion of C and Ar correlate well but
the efficiency with which C atoms are retained by dust grains is less
than that for Ar atoms. This implies that reactions between C and
other atoms (eg H, 0, N) may inhibit the accretion of C by dust and
that carbon may not accrete efficiently until the chemical activity of
the ambient gas is reduced by the conversion of H atoms to Hz and
to
CO. When carbo.n does condense, however, it is on silicate cores. This
produces core-mantle particles where the mantle material is likely a
hydrogenated form of amorphous carbon. Such material is called a C:H
or HAC. The role played by such core-mantle particles in the ISM and
their relation to PAR will be discussed in subsequent sections
°
°
°
3.
LARGE MOLECULES AND SMALL GRAINS
fhe interface region between the regime of large molecules and small
particles is an interesting one from both a physico-chemical and
astronomical point of view. The behaviour of such systems has not been
W. W. DULEY
376
extensively studied either theoretically or in the laboratory although
the largest PAH molecules that are easily available to experimentalists
(eg. hexa benzocoronene, etc) have sizes that place them in this range.
The largest known interstellar molecule HelIN is long enough to assume
some of the characteristics of a small particle.
Figure I shows a plot of the product (N x RA) for known interstellar carbon molecules and grains where N is the number of carbon
atoms in a molecule/particle present with a relative abundance (RA)
compared to hydrogen. Points are shown for the cyanopolyynes as well
as for grains of ~ 0.02~m and 0.2~m diameter. One can see from this
0
"-
C -
'x"
~
+
CD
z
...,
§-10
10
Figure 1
'SIZE' ( II·)
10
103
Log (N x RA) for carbon molecules and dust
figure that the interval between the largest known molecule and the
smallest grains spans less than a decade in size - but this is a size
range in which we have little information concerning the presence or
absence of particles in the ISM.
The possible presence of IS molecules with sizes in the range
between 10 and 10 2
This suggests that larger molecules may indeed
be more resistant to dissolution under IS conditions. Grains, of
course, have lifetimes of >10 8 yr under diffuse cloud conditions.
To estimate the lifetime of PAH molecules against photodissociation and dissociative recombination with electrons, a simple RRKM
calculation has been performed. The probability for decomposition P on
release of an energy jhv is given by
X.
P
=
[j - m + s - I)! j!
[j + s - 1] I [j - m]!
(4)
where hv = vibrational bond energy, mhv = bond dissociation energy and
$ = 3N - 6.
I f i t is assumed that hv = 0.5 eV, j = 20 (10 eV input
energy) and mhv = 4 eV as appropriate for CH bonds, then it is a simple
matter to show that a lifetime of 10Byr for such a molecule under
diffuse cloud conditions implies that N > 60. However, dehydrogenation
FORMATION, DESTRUCfION AND EXCITATION OF CARBON GRAINS AND PAH MOLECULES
377
is unlikely to be a major destruction route for molecules in diffuse
clouds because of the abundance of gaseous H atoms. A potentially more
serious destruction route could involve the dissociation of a C-C bond
within a PAR ring. The energy required for such bond breaking is ~ 11
eV (Benson 1965) while hv ~ 0.2 eV. Taking jhv = 13 eV so that j = 65
with m = 55 one obtains, N ~ 7 is such a molecule is to be stable under
diffuse cloud conditions for ~ 10 8 yr. This suggests that even small
rings will be stable against VUV dissociation over lifetimes comparable
to that of grains in IS clouds.
4.
PAR AND DUST-FORMATION AND DESTRUCTION
In this section I discuss some chemical processes that may influence
the abundance of PAR molecules and carbon dust in interstellar clouds.
We first consider possible chemical routes that could lead to the formation of PAR molecules in IS clouds. To facilitate calculation benzene
will be adopted as a prototype for 'PAR'. It should be remembered,
however, that actual IS PAR are predicted to contain many more than one
ring.
4.1
Ion-Molecule Reactions
The following reaction scheme would be a straightforward initial step
in the formation of PAR in IS clouds. The initiating reaction involves
c+ addition to acetylene
(5)
followed by
(6)
(7)
and then
(8)
with a final reaction involving methane
(9)
to yield the benzene cation. If one makes the optimistic assumption
that such a reaction would proceed in diffuse clouds where
n(C+) ~ 10-~n and that n(C 2 H2 ) = 10-8 n with all rate constants taken to
be 10- 9 cm 3 sec- 1 then the timescale for formation of benzene to
n(C 6 H6 ) = 10- 7 n is T ~ 10 7 n- 1 yr. Such a result also neglects competition due to dissociative recombination with electrons, ego
C H + + e-> products
(9a)
W. W. DULEY
378
When such competing routes are taken into account, > 10 8 yr. for
benzene formation under diffuse cloud conditions.
4.2
Surface Reactions
To examine the possible role of surface reactions we can use the
simplified reaction scheme
x + grain ->X - grain
(10)
y + X - grain ->PAH (gas) + grain
(11)
Rate constants for each reaction would be k = OV ~ 3 X 10-s cm g sec- 1
To make such a reaction as fast as possible assume that the ratelimiting reaction involves the arrival of C or c+ at the grain surface
and that n(C) = 10- 4 • With ng = 10-12n • the timescale for the formation of benzene to the n(CsHs) = 10-'n level is , ~ lOS n- 1 yr in the
absence of competing reactions and under the assumption that only
reactions with Y limit the concentration of X-grain centres. Both of
these assumptions are unrealistic under IS cloud conditions (Duley and
Williams 1984a) making it unlikely that grain reactions of this sort
will be important sources of PAH molecules in diffuse clouds.
In darker clouds where C2H2 and C2H can increase in abundance
(Mitchell Ginzberg and Kuntz. 1978) the grain reaction
(12)
may be of potential significance. Such reactions could be initiated
by cosmic ray impacts on grains. A study of the possible role of
acetylenic reactions on grains in the formation of IS PAH has been
published by Floyd. Prince and Duley (1974). It is of interest that
led in this work to the production of PAH molecules also yielded the
simplest aromatic molecule cyclopropenium. CgH g . The cyclopropenylidene radical. CgH2 has subsequently been widely detected in the ISM
(Thaddeus. Urtilek and Gottlieb 1985)
4.3
Formation in Shocks
In shocks CsHs will be formed in reactions such as
(13)
(14)
organic molecules will be destroyed in combustion reactions
CgH4 + 0 -> CO + products
(15)
CsHs + 0 -> CO + products
(16)
As gas temperatures of
~
3000 K (as appropriate for shocks in diffuse
FORMATION, DESTRUCfION AND EXCITATION OF CARBON GRAINS AND PAH MOLECULES
379
clouds) all rate constants will be k ~ lOll cm 3 sec- l (Mitchell and
Deveau 1983) and combustion reactions limit n(CsRs) ~ 10- 10 n; a negligible amoun t.
In dense objects where atomic oxygen abundances are reduced one
~ight expect that shocks could lead to the synthesis of large PAR
simply by C insertion reactions
+
XII
C -
XC
+
R
(17)
with k ~ 10-11cm 3 sec-I. The timescale for formation of a molecule
with NC carbon atoms is then
t
~
n
sec
(18)
c
If nc ~ 10-sn while Nc = 30, t = 1010n- l yr. Taking n = 104 cm-3, then
t = lOs yr which should be compared to a typical shock cooling time of
ca. 103 yr • As a result, it is unlikely that large PAR molecules can be
synthesized during shocks in either diffuse or dense clouds.
4.4 Grain-Grain Disruption
The formation of large molecules via grain disruption in diffuse cloud
shocks has been discussed recently by Duley and Williams (1984b). Such
molecules are liberated in the cool post-shock gas as carbon grains
undergo grain-grain collisions. When applied to the formation of PAR,
the following equation predicts the column density of a molecule with
~c carbon atoms
NL
(19)
where V is the shock speed, fng is the fraction of carbon locked in
grains before the shock, y is the fraction returned per shock and S is
the destruction rate of the gas phase molecule.
With no destruction of the final product
NL
(20)
where L is the size of the cloud or integrated shocked region. Table
II summarizes values of NL for the two cases taking Nc = 30. These are
appreciable column densities suggesting that grain-grain collisions
may be a major source of large carbon molecules in diffuse clouds. The
product of such collisions may include carbon clusters as well as
linear and planar carbon molecules (Kroto, this volume).
380
W. W. DULEY
Table II.
PAH column densities, NL from grain-grain
collisions in shocks. Nc ; 30
fnCO
y
S(sec l )
v(km secL(pc)
NL(cm- z )
4.5
Destruction
10-4 n
0.01
10- 11
50
No-Destruction
> 20
1
lOll n
Hot Atom Reactions
ROssler (1986) has discussed a possible role for hot-atom chemistry in
the generation of IS PAH. Hot carbon atoms will be present to a
limited extent in diffuse clouds as the result of photolytic decomposition of CO and other molecules. They can also be formed in ion-molecule reactions (Adams, Smith and Millar 1984). Excess energies from
these routes appear to be limited to less than 2 eV in most cases of
interest. Larger excess energies may, however, be attained under shock
conditions or during cosmic ray impact.
4.6
Destruction Mechanisms
The reaction of graphite and other forms of carbon solid with Hand 0
atoms is well documented. Bar-Nun (1975) and Bar-Nun et al. (1980)
have shown that Hand 0 each react with powdered graphite at cryogenic
temperatures. With H, the primary reaction product is methane, while
CO and COz are seen to evolve from graphite exposed to atomic oxygen.
These conclusions are supported by experimental data obtained for bulk
carbon at higher temperature (cf. Vietzke, Flaskamp and Philipps (1982).
This suggests that the lifetime of carbon grains in diffuse clouds may
be limited by reaction with the abundant 0 and H present in these
objects. Duley and Williams (1986) have discussed the effect that such
reactions can have on the inhibition of carbon accretion in diffuse
clouds and conclude that carbon grains should be under abundant in
regions where H atoms are present.
It seems likely that these considerations should also apply to the
destruction of IS PAH with the overall reaction rate being enhanced by
the temperature fluctuations that accompany VUV absorption by these
molecules. Figure 2 shows a photo-oxidation scheme for PAH that
rapidly leads to the dissolution of the aromatic structure. Here thermal spikes ~T ~ 103K permit the system to tunnel through the activation
barrier for liberation of CO or HCO. Such spikes are largest for small
molecules and would accompany IR emission (Leger and Puget 1984) so
that regions of strong UIR emission would signal regions of PAH destruction. A schematic of the reaction channel for such photo-oxidation
reactions is given in figure 3.
FORMATION, DESTRUCTION AND EXCITATION OF CARBON GRAINS AND PAH MOLECULES
381
Figure 2. Photooxidation
of gaseous PAR
c;, -OH
Figure 3. An exothermic reaction to
yield CO from surface OH occurs by
surmounting the energy barrier
during a temperature excursion
c;,..,
H.
co t,.J
DH
The activation energy for breaking of a C-C bond can be taken to
be ~ 20% of the bond energy. For PAR this is roughly 2 eV. UIR
emission would suggest that an excited PAR molecule reaches an internal
temperature T ~ 900 K (kT = 0.075 eV) and that this excitation persists
for ~ 10- 3 sec. With a vibrational frequency of 10 12 Hz, the probability factor for surmounting the activation barrier associated with CO
formation would be
P ~ 10- 3
2.5
X
X
10 12 x exp[-2/0.075]
10- 3
or about 0.25% per UV absorption. In a diffuse cloud a molecule with a
cross-sectional area cr = 10- 15 cm 2 will absorb VUV photons at a rate
~ 10-8 sec- I (Duley and Williams 1984a). If P ~ 10- 3 per event, then
this implies a lifetime of only lOll sec against photo-oxidation reactions. In practice, the rate-limiting step in this reaction will
likely be the initial reaction with 0 atoms or O-bearing molecules to
yield phenolic surface groups. Nevertheless, it appears that PAR lifetimes in diffuse clouds may be severely limited by reaction mediated by
the UV excitation that leads to IR emission. PAR molecules, are likely,
therefore, to be seen only where they are formed; probably in regions
of shock-induced grain disruption.
5
HAC STRUCTURE AND PROPERTIES
An analysis of the depletion of interstellar carbon in diffuse clouds
shows that the dynamic accretion of carbon by silicate/oxide cores
W. W. DULEY
382
leads to the formation of carbon core-mantle grains. Carbon solids
formed by vapor deposition from a carbon plasma or atomic/molecular gas
have been extensively studied (Miyazawa et al 1984, Duley 1984, Savvides
1985). The resulting solid is an amorphous carbon with combined trigonal (ie graphite-like) and tetrahedral (ie diamond-like) bonding.
Such solids can be characterized by their bandgap energy Eg • Solids
with primarily trigonal bonding have Eg small «< 1 eV. Savvides
(1985) prepared amorphous carbon films by argon ion bombardment of
graphite that had ~ 75% tetrahedral structure and only 25% of trigonal
structure. For this material Eg was as large as 0.74 eV.
Carbon solids prepared in a similar way in a hydrogen rich atmosphere can have much larger bandgap energies. For example Watanabe and
Inoue (1983) observed Eg ~ 3 eV in films prepared by plasma decamposition of CH 4 • The involvement of hydrogen in tetrahedral carbon bonds
is evident through the observation of an IR absorption band due to
aliphatic CH at 3.4 ~m (Bubenzer et al. 1983, Discheler et al 1985).
When such solids are heated, driving off this hydrogen, Eg tends to
decrease together with the intensity of the 3.4 ~m IR feature. At the
same time, a new absorption band at 3.28 ~m appears due to aromatic CR.
The decrease in Eg with heating has been studied by Smith (1984).
Figure 4 is a schematic representation of the structure of these hydrogenated amorphous carbon (HAC or a-C:H) solids.
Figure 4.
Structure of HAC solid
The wide bandgap of some HAC solids means that edge luminescence
can be excited with visible and UV radiation. This luminescence has
been observed by a variety of groups (Street 1980, Dunstan and Boulitrop 1984, Lin and Feldman 1982, Watanabe et al. 1982) and consists of
a broad emission band extending over an energy range EL < Eg • Watanabe
et al. (1982) find that EL ~ 1.7 - 1.9 eV for HAC with Eg in the range
between 2.6 and 2.0 eV, respectively. The similarity of this emission
to that seen in the spectrum of the Red Rectangle has been noted by
Duley (1985b) who argues that the dust in this object is HAC-like. It
is of interest that a complete set of VIR features are observed in the
Red Rectangle spectrum.
,
Lin and Feldman (1982) have observed structure in this broad-band
emission from HAC which they interpret as due to the excitation of IR
vibrational modes. This excitation will lead to IR emission by
FORMATION, DESTRUCTION AND EXCITATION OF CARBON GRAINS AND PAH MOLECULES
383
functional groups at the HAC surface and therefore to discrete IR
luminescence from HAC at -CH, -CC and other frequencies. It should be
noted that this emission is not thermal in nature and therefore does not
require that HAC grains be heated to temperatures in the lOoK range.
Instead, this emission would be a true luminescence involving localized
states within the bandgap as observed in several other amorphous semiconductors (Huang et al. 1984). In the ISM this emission would originate from carbon-silicate core mantle particles. A schematic representation of the energy levels involved in IR luminescent emission from HAC
is shown in figure 5. Since these particles will be relatively large,
COIIDIICnDN
Figure 5.
HAC luminescence spectrum
(sizes 0.01 - 0.2 ~m) their absorption efficiency is comparable to
their geometrical cross-section. This, together with a high efficiency
for conversion of UV-VIS photons to band-gap luminescence in HAC
(Watanabe, Inoue and Atoji 1983), means that HAC luminescence should be
widespread in the ISM. The recent observation of extended red emission
in many nebulae (Sel~gren, Werner and Dinerstein 1983, Witt and Schild
1985) is likely related to the present of HAC in these objects. It can
be predicted that the diffuse galactic light component should also have
a weak discrete emission feature in the 600-700 nm spectral region
analogous to that observed in the Red Rectangle if IS HAC dust in
diffuse clouds has Eg > 2 eV.
6.
RELATION BETWEEN HAC AND PAll
Amorphous carbon and graphitic solids can be considered in many instances to be a collection of· dehydrogenated PAll molecules. For
example. Mrozowski (1950) has shown how the electrical behaviour of
calcined cokes can be understood as a progression from large PAll to
small graphitic crystallites. Partially hydrogenated amorphous carbons
such as HAC will also exhibit many similarities to a collection of PAll
molecules as noted by Duley and Williams (1981). It is therefore not
surprising that the luminescence of HAC should display both molecular
(CH, CC emission) and solid-state )broad band IR and NIR emission)
character. The disruption of such dust in ~ocks will result in the
liberation of PAll and PAll fragments (Duley and Williams 1984b) which
384
W. W. DULEY
HAC MORPHOLOGY
/H,C
"if
- : / , - - SILICATE,
DIFFUSE CLOUD
\OXIDE H
!J'H
H /
'c
H........
r~t.
~
CH CH
DARK CLOUD
HAC
----H- -H
Eg ",2ev
/?~HAC
H H
CH
EMISSION OBJECT\
Eg ,,-1ev
\
~- C,~.+MOLECULES
-
/
Figure 6.
rSM.
~
\
POST SHOCK
A heirarchal model for HAC evolution in the
PAR derives from HAC in shocks
may also emit at IR wavelengths. Thus HAC and PAR will be related,
with broad IR and UIR emission features deriving from luminescence and
normal thermal dust emission while sharp UIR emission derives from PAR
liberated from HAC in shocks. Since PAR is less resistant than HAC to
chemical reaction under ISM conditions, the emission from PAR should be
relatively localized to regions where PAR is being formed. HAC
emission should be more widespread, as it would occur from a component
of the dust that provides a significant fraction of the extinction in
IS clouds.
An attempt to summarize the relation between HAC and PAR is given
in figure 6. Silicate and oxide grains accrete carbon in the form of
HAC in denser diffuse clouds. This material exhibits strong absorption
at 3.38, 3.41 and 3.48 ~m (Watanabe et al. 1982) as seen for example in
absorption spectra of IRS 7 (Jones, Hyland and Allen 1983). These
features are very characteristic of HAC, Dischler et al. (1985) and are
due to aliphatic CH. The Bandgap of such a solid is anticipated to be
Eg ~ 2 eV. It will luminesce on exposure to UV-VIS light.
FORMATION, DESTRUCfION AND EXCITATION OF CARBON GRAINS AND PAH MOLECULES
385
When HAC that has been in a dark cloud is exposed to an enhanced
radiation field and/or a shock wave, then the bond structure evolves
toward a trigonal (graphitic) configuration with a loss of aliphatic
CH. Such grains, as seen in emission, will show aromatic CH features
(3.28 ~m, 11.3 ~m) as well as those due to residual aliphatic CH. Eg
will diminish and broad band emission will shift from visible wavelengths to the NIR, or even into the middle IR for dust that has been
subjected to heattng to temperatures in excess of 600K. Gas phase PAll
will be liberated from shocked HAC and can also yield 3.28 ~m emission.
Prolonged exposure to high velocity shocks will result in the loss of
all HAC surface layers on silicate dust.
ACKNOWLEDGMENTS
This research has been supported by grants from the NSERCC.
REFERENCES
Adams, N.G., Smith, D. and Millar, T.J. 1984. M.N.R.A.S. ~11, 857.
Aitken, D.K., Roche, P.F., Spenser, P.M. and Jones, B. 1979~-Ap.J. ~JJ,
925.
Bar-Nun, A. 1975. Ap.J. 197, 341.
Bar-Nun, A., Litman, M.,=~~d Rappaport, M.L. 1980. Astr. Ap. ~~, 197.
Benson, S.W. 1965. J. Chern. Education 42, 502.
Bubenzer, A., Dischler, B., Brandt, C. ~~d Koidl, P. 1983. J. Appl.
Phys. 2~' 4590.
Cowie, L.L~~ and Songaila, A. 1986. Ann. Rev. Astr. and Ap. (in press).
Dischler, B., Sah, R.E., Bubenzer, A., Koidl, P. 1985. Solid State
Commun. (in press).
Duley, W.W. 1984. Ap.J. ~~1, 694.
Duley, W.W. 1985a. Ap.J.-Z97, 296.
Duley, W.W. 1985b. M.N.R.A~S. ~12, 259.
Duley, W.W. and Williams, D.A. 1981. M.N.R.A.S. 196, 269.
Duley, W.W. and Williams, D.A. 1984a. "Interstell~~ Chemistry"
Academic Press, London.
Duley, W.W. and Williams, D.A. 1984b. M.N.R.A.S. 211, 97.
Duley, W.W. and Williams, D.A. 1986. M.N.R.A.S. (i~=press).
Dunstan, D.J. and Boulitrop, F. 1984. Phys. Rev. B ~Q, 5945.
Dwek, E. and Scala, J.M. 1980. Ap.J. ~12, 193.
Floyd, C.R., Prince, R.H. and Duley, W~W~ 1974. J.R. Astr. Soc.
Canada ~Z, 299.
Huang, F-S., Chang, H., Chen, J-R. and Liu, Y-C. 1984. Jap. J. Appl.
Phys. n, 6.
Jones, T.J~, Hyland, A.R. and Allen, D.A. 1983. M.N.R.A.S. ~Q~, 187.
Leger, A. and Puget, J.L. 1984. Astr. Ap. 111, L5.
--Lin, S-H. and Feldman, B.J. 1982. Phys. Rev. Lett. 48, 829.
Martin, P.G. 1978. "Cosmic Dust" Oxford Univ. Press:'=
Mitchell, G.F. and Deveau, T.J. 1983. Ap.J. ~§g, 646.
3~
W.W.DULEY
Mitchell, G.F., Ginsberg, J.L. and Kuntz, P.J. 1978. Ap.J. Supp1. ~~,
39.
Miyazawa, T., Misawa, S., Yoshida, S. and Gonda, S. 1984. J. Appl.
Phys. 55, 188.
Mrozowski~=S. 1950. Phys. Rev. 11, 838.
Rossler, K. 1986. Rad. Effects ~In press).
Savvides, N. 1985. J. App1. Phys. 2§, 518.
Se11gren, K., Werner, M.W., and Dinerstein, H.L. 1983. Ap.J. ~Z1, L13.
Smith, F.W. 1984. J. App1. Phys. 22, 764.
Steel, T.M. and Duley, W.W. 1986. -Xp.J. (submitted).
Street, R.A. 1980. Adv. Phys. 30, 593.
Thaddeus, P., Vrti1ek, J.M. and=Gott1ieb, C.A. 1985. Ap.J. ~22, L63.
Vietzke, E., F1askamp, K. and Philipps. 1982. J. Nuc1. Mat.-!!~/~~~,
763.
Watanabe, I., Hasegawa, S. and Kurata, Y. 1982. Jap. J. App1. Phys.
~1, 856.
Watanabe, I. and Inoue, M. 1983. Jap. J. App1. Phys. ~£, L176.
Watanabe, I., Inoue, M. and Atoji, T. 1983. J. Non-Cryst. Solids.
59/60, 377.
Witt~ A~N. and Schild, R.E. 1985. Ap.J. ~2~, 225.
DISCUSSION
S. Leach:
A
J) Is the OH- in silicates interpretation of the 2200
extinction hump based on another interpretation of the laboratory experiments previously used to argue in favour of 0 2- in silicates ?
2) If the OH- interpretation is correct, there should be a
fairly strong OH- absorption in the IR. Would there be enough carriers
for ohservation of such a feature ?
Answer: J) Yes. It appears that the Mg-silicate analog of the 0 2absorption in MgO is the OH- centre, although this needs to be confirmed with further experiments.
2) It might be possible to detect such a feature, but we need
experimental data on the position, intensity and width of this IR feature in amorphous Mg-silicate particles.
POLYAROOATIC HYDROCARBONS AND TIlE CONDENSATIOO OF CARBON IN STEUAR
WINDS
Rudolf Keller
Technische Universitat Berlin
Institut flir Astronomie und Astrophysik PN8-1
Hardenbergstr.36
D-1000 Berlin 12
Gennany
ABSTRACT. The condensation of carbon in cool stellar winds is described
by two processes: the fonnation and the growth of carbon particles. In
contrast to the assumptions of conventional droplet theory of
condensation, these particles were found to be PAH's. Their fonnation
and growth is described in tenns of chemical reactions. The derived
temperature of effective condensation fits the observations. Structure
and abundances of the molecules and particles at the end of the
condensation process are discussed.
1. In troduc tion
During the last years the astronomical environments, where dust is
produced, have been extensively studied, both observationally and by
theoretical models, but the physics and chemistry of the dust fonnation
process itself is still poorly understood. It is the aim of this paper
to outline in detail the fonnation of carbon grains from gases. To avoid
the complications by a strong UV-radiation field, I consider the case of
a massive stellar wind (M )10- 6 M,/y) from an evolved cool (Teff <3000K)
carbon star, e.g. the famous IRC+10216.
2.Shortcomings of the conventional droplet theory of condensation
Modelling of the radiation transport in several C-star circumstellar
shells gave by comparison to the observed spectra a temperature of 750K
to 1000K for the inner edge of the dust shell (Rowan-Robinson and
Harris,1983). This is in striking disagreement with the droplet theory,
which predicts the temperature of effective condensation, where the
carbon grains fonn, to be never less than 1300K (Gail and Sedlmayr,
1985) •
Theoretically there is a mistake too: the droplet theory considers
the condensation process as fonnation and growth of roughly spherical
387
A. Uger el al. (eds.), Polycyclic Aromatic Hydrocarbons and Astrophysics, 387-397.
© 1987 by D. Reidel Publishing Company.
R. KELLER
388
particles. Their heat of formation is calculated as the product of
surface and specific surface energy. In contrast the graphite crystal is
not isotropic, but consists of layers with the binding energy between
neighboring layers being quite small. Hence it seems necessary to
replace the spherical particles with roughly circular pieces cut from an
infinite monoatomic graphitic layer. This way one finds, that by the use
of the droplet theory the heats of formation are underestimated and
hence the effective condensation temperature is overestimated.
Now, having arrived at substantially lower temperatures the circular
pieces are not thermodynamically stable: the free ~-bonds of the edge
atoms catch H-atoms and PAH's are formed.
Thus one has to consider carbon condensation via PAH's, and
therefore I give some remarks on general non-homogenous nucleat~n
theory.
3.General nonhomogenous nucleation theory
Figure 1 shows the equilibrium densities of carbon molecules in a
supercooled gas as a function of their size ( = number of C-atoms/
molecule), considering only the most stable ones in each size class.
Because of the supercooling, very big molecules have the tendency to
grow and show an increase of equilibrium density with the size (I will
refer to such particles as "grains".). For small molecules this tendency
is reversed. Therefore the equilibrium density has a pronounced minimum,
the "nucleation barrier" (see also S.Stein,1978). To form a carbon
particle means to go through a chain of chemical reactions, starting
from the most abundant molecule, acetylene, passing through the
nucleation barrier and ending up with a grain.
Formally:
Aj + B j - A j • 1 + Cj i=1,2, •• N-l
with Aj carbon-molecule, Al = acetylene and AN = grain
and Bi growth substance
and Ci byproduct
With the two assumptions,
that firstly acetylene, all growth substances and all byproducts are
in thermal equilibrium
and that secondly the chain of reactions has reached a steady state,
one derives for the densities n:
-k.I
ft A'I
nB;
fi Ai _, n Bi _,
-n Ai.,
) +
-n A.
)
I
=0
with the equilibrium densities ft. Hence the quantity J:
J
=
k
r\. n B. (nA./ft
I
I
I
A I
n A•
1+1
/ft A .
1+1
)
is independent of i. J gives the rate, at which Al is consumed and AN is
formed, i.e. the grain formation rate.
POLY AROMA TIC HYDROCARBONS AND THE CONDENSATION OF CARBON IN STELLAR WINDS
389
Figure 1
Equilibrium densities of carbon molecules
log P (dyn/cm 2 )
_5~----------------------------~~~L-------~
-10~--~~------~~----~~------------~~
-201~------~~~--------~~~~--------~
2
4
6
8
10
12
14
16
number of carbon atoms per molecule
18
20
The equilibrium densities of the polyacetylenes are nearly independent
of the temperature.
Continous (broken) lines represent growth reactions by the addition of
one (two) molecule(s) of acetylene. Dotted lines represent growth
reactions, Which include internal rearrangements.
390
R. KELLER
By summing up the former equations with respect to i, one obtains:
N-1
J ~
L
1/(k j fiA nil)
I
I
== n A /fiA - n A/ftA == 1
1
1
N
N
because acetylene is in TE and ~A -+,(0. Hence the grain formation rate
is:
N
J
== ( [ l/(k j RAo n Sol »-1,.,. min(k j RAo n B,>
I
I
I
Since ilAo has a pronollllced minimum, the sun is dominated by one term.
In principal there are many different reaction chains, but in
reality it is sufficient to consider only the most efficient one:
J
==
max
reac tion-chains
min
steps
Since J depends linear on the equilibriun density at the nucleation
barrier, it decreases drastically with the temperature (a factor of 10
by 10K). Thus the onset of effective condensation is sharply defined.
4.The condensation of carbon
The physical and chemical conditions in the condensation zone in massive
stellar winds are:
a) temperature: 750K (T (950K (see below)
b) total gas pressure: 10-6 (P< 10-3 dyn/cm 2 (estimated from a massloss
rate 10- 6 (M( 1O- 4 M",/yrj chosen for the example: P==10- 3 dyn/cm 2 )
c) chemical composition: [H)) [C])[O) )[N] (chosen for the example:
t( == ([C]-[O])/[H]==.OOl)
d) most abundant molecules: Hz, H (H can be excluded, if TE holds, see
below), CO (containes nearly all oxygen), CzH2 (containes nearly all
remaining carbonj only carbon of that form may condense).
Under these conditions polyacetylenes and PAH's are by far the most
abundant carbon moleculesj excited or ionised states, radicals or carbon
molecules containing also nitrogen and/or oxygen can be neglected.
The following consideration leads to a rule for determining the most
abundant species among the PAH's.
In the formula for the equilibriun density of a PAH based on the
densities of molecular hydrogen and acetylene:
p (ZjH
Zj
== p (ZH
j
pi-H2i exp( - 6H/RT + /jS/R)
2
one can approximate the entropies by their translational contributions:
b,S
leading to:
POLY AROMATIC HYDROCARBONS AND THE CONDENSATION OF CARBON IN STELLAR WINDS
p
~r~
.: exp(S Ir /R)
p. exp( -S Ir /R)
!!!
( (p -exp( _str
£i
391
/R»i exp( - ~H/R)
10- 18
For fixed size N = 2i, only the two last factors in the equation for the
equilibrium density are variing:
The enthalpy factor favoures molecules with a maximum of binding
energy; consequently C-atoms should be arranged exclusively in hexagons
for stabilisation by resonance and for the absence of stress.
The entropy factor tends to minimize the number of H-atoms, i.e. the
number of places at the edge of the molecule. Consequently the molecule
should be as compact as possible and it should contain many pentagons
and squares,since this reduces the number of edge positions. This is
demonstrated in the following sequence:
From calculations of many equilibrium densities, it was found, that the
competition between enthalpy and entropy leads to the following
compromise for the configuration of the most abundant molecules:
They contain as many pentagons as possible, but squares and attached
pentagons are excluded, since the enormous stress and the negative
resonance energy would increase the enthalpy too much.
From equilibrium data an upper limit of the effective condensation
temperature may be estimated by the following assumptions:
a) The condensation time is less than 10Oyr.
b) More than 1% of the carbon is condensed.
c) The grains contain less than 10eC-atoms each.
This yields a lower limit of J:
J/n
C2H 2
)3'1()-20 S-l
The reaction constant is bounded by the collision frequency:
k-n
With
J
(2 H2
=k
(10- 4 s- 1
n(
H
R
2 2
(R is to be taken at- the nucleation barrier) one derives from both
inequalities a lower limit for &:
g/nC..H
C"
2
= J/ (k
n~ H ) .) 3 -10- 16
2 2
a value,which requires T(950K (c.f.Fig.1). This result is independent
of the specific mechanism of the growth reaction.
392
R.KELLER
Figure 2
The chemical reactions
For simplicity H-atoms are omitted, but radical sites are indicated by a
dot. Hexagons represent aromatic rings, double and triple bars represent
double and triple bonds. E.g. acetylene is indicated by a triple bar.
1. In case of TE: continous lines show the most efficient chain of
reactions for the formation of grains. The heavy lines indicate the
ratedetermining steps at the different temperatures. The numbers at
these steps give a rate estimate for T = 850K in terms of
log(k j ne. fIA In(~) s-1.
I
I
["2
2. In case of non-TE: broken lines represent the most important
reactions. The arrows show, whether a reaction is mainly constructive or
destructive. For naphthalene and bigger molecules the important
reactions are the same as in case of TE.
POLY AROMATIC HYDROCARBONS AND THE CONDENSATION OF CARBON IN STELLAR WINDS
393
5.Formation and growth of grains
For investigating formation and growth of grains, the reaction, by which
a molecule of acetylene is added to a PAH, needs to be studied in
detail. Because both partners are nonradical molecules, the direct
addition is nearly completely suppressed by the high energy of
activation (-1.5-2.0 eV). It is more efficient to produce a radical
site on the PAH by abstracting a H-atom first and attaching there the
acetylene in a second step.
H
>-<
H
H
H
-~
•
H
'H
Figure J. The growth reaction of ~AH's. (An
leads to the formation of a pentagon.)
H
-Qanalogou~ ~rocess
The ring closure with the side chain has to surmount a considerable
energy of activation too, but its rate is high, as it is a monomolecular
reaction. Therefore the entire growth process is controlled by the
second step. Neglecting the energy of activation of this radical
reaction, the rate of growth is given by: R = Rop with Ro being the
collision rate of acetylene with an edge site of the PAH and p being the
chance, that the H-atom at this site is lost. The hydrogen abstraction
is comparitively fast and therefore in equilibrium.
With this rate R and some wellknown rates for the formation of
carbon chains (Koike and Morinaga,1981, Tanzawa and Gardiner,1980, Frank
and Just,1980), a reactionscheme from acetylene to big PAH's can be
constructed (Fig.2). At 850K the slowest steps on this pathway are the
formation of the second and the third aromatic ring. They determine a
grain formation rate of:
J/n
C2H2
~
10-2-0
s-1
This is below the lower limit of J as calculated above. Therefore an
effective condensation can take place only at temperatures below 850K.
From the rate R one can also derive the growth velocity of grains: a
PAH with N C-atoms has approximately M edge sites with:
M=
JbN
Hence the growth of the PAH is described by:
N = 2R nc 2H2 .JON
or
.IN =$)R"n C2H2
At 850K and below the growth is handicapped by the very small number of
radical sites and
$
-1/(l00yr)
R.KELLER
394
These two estimates for formation and growth under TE conditions imply:
only very few grains can form at T >8S0K and those formed at T <8S0K
cannot grow. Hence the condensation process needs additional supply by
non-TE effects.
6.The non-TE effect
The non-TE effects under consideration depend on the formation of the
small molecules.
a) N2,OO,CN,C 3 ,C 2H: these molecules form at T )2S00K and P >10~dyn/cm2
in the stellar atmosphere; radiative and three-body collisions are
effective to produce and maintain thermal equilibrium.
b) H2 : with respect to the thermal equilibrium of atomic and molecular
hydrogen the track from the upper stellar atmosphere to the
condensation zone can be divided in two parts:
Upstream from a point in the wind at lS00K (P-1 dyn/cm 2 ) TE
requires a roughly constant fraction of several percent of atomic
hydrogen. By three-body collisions the equilibrium value is
reached; the molecular hydrogen is possibly reduced by shockwaves.
Downstream from this point the TE-value for H/H2 drops rapidly
reaching 10-6 in the condensation zone, but there are no
effective reactions for a further decrease of the actual value
for H/H2:
aa) The density is to small for three-body collisions.
bb) The radiative recombination of hydrogen is very ineffective,
because it requires a quadrupol transition between vibrational
states.
cc) The catalytic reactions:
C2 H) + H C 2H2 + H -
C2H2 + H2
C 2H + H2
H + C2H2 -- C2H3 + ny
H + C2H - C2H2 + hv-
don't reduce the fraction of the atomic hydrogen efficiently
either.
Reaching the condensation zone the wind has consequently several percent
of atomic hydrogen. Since the hydrogen abstraction is very fast, the
number of radical sites on PAR's exceeds its TE-value by the same factor
as atomic hydrogen does (-10 5 ). Thus the rate R, and hence the growth
velocity are increased by 10 5 • The increase of the grain formation rate
is smaller, because the atomic hydrogen also has the tendency to
destruct the first and the second aromatic ring (c.f.Fig.2).
Combining this condensation mechanism with model calculations of
stellar winds one finds, that a substantial fraction of the acetylene is
condensing to giant aromatic molecules consisting of millions of Catoms. The condensation process stops due to:
a) the dilution of the gas in the wind
b) the carbon consumption
c) the conversion of the atomic hydrogen into molecular hydrogen.
POLY AROMATIC HYDROCARBONS AND THE CONDENSATION OF CARBON IN STELLAR WINDS
395
The latter process is catalysed by the grains the following way:
-<>H
H
-:2 )
-<>•
-<>H
H
-4
H
Figure 4. lbe formation of Hz catalysed by fAH's.
As Lllb:e are always more non-radical sites than radical sites on the
PAH's, the second step is ratedetermining. Hence the formation of H
shows a close analogy to the growth reaction depicted in Fig.3.
The consumptions of atomic hydrogen and of acetylene follow the
differential equations:
n,s being the number of radical sites on PAH's per unit volume.
Elimination of n,s yields:
1\/n H = m TIc.>< /nc
,,"2
H
2 2
with m = 2k,lk z
Neglecting the temperature dependence of m, since both reactions are
exothermal radical reactions, this equation can be integrated:
m
n H = c n C21i2
The density of acetylene is expressed by the fraction f of condensable
carbon, Which is actually condensed:
f
=
n condo /(2n Cil2 + n cond.)
The integration constant is determined by the conditions before
condensation takes place:
n H/n H2 = (nH/nH2~ and f
=0
one obtains: n/n H2 = (n/n H2 )0 (i-f)'"
Thus by the increasing degree of condensation the atomic hydrogen and
hence all other radicals are removed and the condensation slows down.
The arguments presented lead to the conclusion, that the
condensation process finally results in:
a) A small number of very big PAH's, containing a considerable amount
of the condensable carbon is produced.
b) The PAH's formed in the late stages of condensation exceed by number
the big PAH's, but they cannot grow and hence the amount of the
carbon condensed in such molecules remains small. Among them
acenaphthylene should be the most abundant, because it is located
in a local minimum of free energy (c.f. Fig.i) and its next growing
step is thermodynamically unfavourable.
c) Most of the atomic hydrogen is converted into H2 •
396
R. KELLER
Three-dimensional grains can form only by adhesive collisio~s of big
PAH's. The binding energy of such a compound may be estimated from the
surface energy of graphite (Abrahamson,1973) or from the heats of
sublimation of PAH-crystals. Assuming realistic vibration frequencies
the free energy can be calculated. From that one derives, that the
formation of dimers and of stacks of PAH's is thermodynamically
favourable, only if the size of the PAH's involved exceeds N C-atoms
with: N - T/8K - 100.
It is not yet clear, whether enough PAH's of sufficient size are
produced in stellar winds to collide and form three-dimensional grains
during the expansion the wind, or if the "carbon-grains", the absorption
of which is observed, are just giant PAH's.
References
H.P.Gail and E.Sedlmayr 1985, Astron.Astoph.,~,183
M.Rowan-Robinson and S.Harris 1983, MNRAS,202,797
S.Stein 1978, J.Phys.Chem.,82,566
--T.Koike and K.Morinaga 1981, Bull.Chem.Soc.Japan,54,530
T.Tanzawaand W.C.Gardiner 1980, Comb.Flame,39,24I
T.Tanzawa and W.C.Gardiner 1980, J.Phys.Chem.,84,236
-P.Frank and Th.Just 1980, Comb.Flame,38,231
J.Abrahamson 1973, Carbon,11,337
-Questions
L. Allamandola:
In your calculation, you used the "surface tension" of graphite to
determine whether or not the small PAH molecules you make can stack
together at the temperature in the region. What is the value you used?
Calculations of the well-depth for benzene-pyrene-coronene dimers give 1
to 7 kcal/mol. Would this type of number affect your particle growth by
PAH clustering mechanism?
Answer:
From surface tension of graphite one derives 1.5 kcal/mol per C-atom.
For PAH stacking energies I assumed 2/3 of this value for bulk graphite,
about 5 times more than your values. If I have overestimated the binding
energy of PAH dimers at the critical size (100 C-atoms) too, then the
formation of threedimensional grains would become still more unprobable.
M.Jura:
1.There is H~ in emission and possibly ionised gas around IRC+l0216.
Therefore there could be some UV penetrating into the condensation
zone.
POLY AROMATIC HYDROCARBONS AND THE CONDENSATION OF CARBON IN STELLAR WINDS
397
2.A product of C2H2 photodissociation, C2H, has been observed to be
quite weak in the outflow of IRC+10216. Therefore it seems as though
most C2H2 does not survive the condensation process.
Answer:
1. The optical depth of the dustshell of IRC+10216 in the UV is quite
large and the number of UV-photons penetrating it is negligeable.
2. Assuming C/O = 2, one derives from the molecular abundances in the
outer shell of IRC+10216 (Lafont et al.,1982), that 70% of the initial
acetylene is condensed. The order of magnitude of this degree of
condensation can be reproduced by the theory presented.
S.Lafont, R.Lucas and A.Dmont 1982, Astron.Astoph.,106,201
E.Evleth:
For the pressures (10- 3 dyn/cm 2 ) and the temperatures (800K) used in your
model how long are the molecules used in your model subject to these
conditions in a typical real system (hours,months)? This can be referred
to the residence time for the reaction.
Answer:
The temperature gradient is given by adiabatic cooling or by the
dilution of the radiation field, the velocity of the wind in the
condensation zone is comparable to the sound velocity. Thus one finds,
that the gas spends several months to several years at a temperature
between 800K and 900K.
SUBJECT INDEX
Absorbate 56
Absorption spectra 108, 235, 248
Abundances 7, 210, 219, 303
Amorphous carbon 31, 49, 173, 255, 345
Archimedean solids 91
Aromatic molecules 35
Auto-ionization 106
Band structure 15
Benzyl 113
Carbon aggregate 85
Carbon blacks 36, 48
Carbon films 51
Carbon molecules 75
Carbon particules 63, 215, 255
Carbon star 197, 217
Carbon vapor 75
Catacondensed 103
Chain molecules 75, 197
Chars 48, 261
Chemical reaction 95
Chemistry 117, 173, 346, 371, 373, 387
Chrysene 264
Clusters 89, 90, 197, 216, 229, 346
Coals 44, 327
Color temperature 227, 245
Compact PAR's 236
Condensation 387
Conduction band 32, 35
Cooling time 243
Coronene 230
Critical point 56
Cross-section 235
Defects 38
Dehydrogenation 241, 249, 273
Density of states 35
Depletion 220, 221
Destruction 373
Diamond 97
Diffuse interstellar bands 75, 112, 351, 367
399
400
Double charged ions 119
Duo hydrogen 234, 340
Electron correlation 26
Electronic structure 15
Emission mechanism 230, 263, 339
Emission temperature 239, 299
Emission spectra 107, 213, 236
Extinction 45, 177, 183, 207, 219
Extragalactic objects 307
Fluorescence 115, 129, 170, 264, 335
Formation 373
Free radicals 108
Fully benzenoid 165
Galaxy 307, 317
Gap 32
Gas absorption 56
Grain collision 217
Grain size 216, 230
Graphenes 38
Graphite 15, 31, 55, 95, 178, 218
Graphitic materials 36, 229
Graphitization 37
Grinding 41
Holes 32
Hot atom 173
Hydrogen 15, 95
Hydrogen coverage 241, 273
Hydrogenated amorphous carbon 219, 269, 329, 344, 373
H II regions 303
Identification in astronomical spectra 223
Infrared emission 223, 255, 307, 339
Infrared excess 207
Ionization potential 106
Internal conversion 115, 231
Interstellar matter 7, 8, 177, 183, 215
Interstellar molecules 7
Intramolecular vibration redistribution 115, 129, 231
Ions 85, 95, 99
Ion-molecule reaction 118
Ionization of PAH's 249, 371
lRAS 273, 287, 303
Isomerization 117
Jet expansion experiment 107
Kinetic 59
SUBJECT INDEX
SUBJECT INDEX
Laboratory spectra 63, 183, 234, 258
Layer planes 32, 33
Life-time 115
Line shape 129
Mantles 178, 217
Mass spectra 85
Matrix isolation 75, 183
Molecular cloud 303
Non-compact PAH's 238
Non radiative transitions 115, 232
Odd-even alternation 85
Open shell 102
optical absorption 25, 43
Organic refractory 177, 219
Ovalene 237
Overtone 260
Oxygene 95
Pericondensed 103
Phase transition 56
Phonons 47
Phosphorescence 170
Photo-ionization efficiency 101
Photo-thermodissociation 246
Photon electron spectrum 169
Physisorption 55
Pi-electrons 18
Polyacetylene 183
Polyynes 199
Quartet state 104
Quenched carbonaceous composite 178, 213
Radicals 99
Raman spectra 49
Reactivity 58
Reciprocal lattice 16
Reflectance 42
Shocks 219, 371
Sigma band 43
Size effect 57
Size of interstellar PAH's 245, 298
Solid carbon 31
Solo hydrogen 234, 340
Soot 261
Source of interstellar matter 8
401
402
Specific heat 241
Spectroscopy 63, 75, 165, 234
Spheroidal molecules 197
Stacking 42
Statistical limit 116
Stellar winds 387
Structural units 38, 48
Surface interaction 55
Synthesis 165
Temperature fluctuation 227, 323
Temperature of dust 224
Three dimensional 21, 38
Tight binding 18
Tribenzocoronene 165
Trio hydrogen 234, 341
Turbostatic 40
Two dimensional 16, 38, 57
Ultraviolet 3, 15, 207
Unidentified IR emission bands 224
Unsaturated species 174
Valence band 32, 35
Vaporization 85
Very small grains 228, 317, 323
Vibrational structure 108
SUBJECT INDEX