APPLIED PHYSICS LETTERS
VOLUME 77, NUMBER 8
21 AUGUST 2000
Influence of oxygen on band alignment at the organicÕaluminum interface
R. I. R. Blyth,a) S. A. Sardar, F. P. Netzer, and M. G. Ramsey
Institut für Experimentalphysik, Karl-Franzens Universität Graz, Universitätsplatz 5, A-8010 Graz, Austria
共Received 24 April 2000; accepted for publication 22 June 2000兲
The presence of adventitious oxygen is inevitable when organic/metal interfaces are formed by
evaporation in high vacuum (10⫺6 mbar.). In this letter, we highlight the importance of this oxygen
for band alignment, and hence, performance, in organic-based devices. The influence of controlled
amounts of oxygen on band alignment in benzene/aluminum model cathode interfaces has been
studied using ultraviolet photoemission in ultrahigh vacuum. We show that even small amounts of
oxygen significantly lower the aluminum work function with concomitant improvement in band
alignment. © 2000 American Institute of Physics. 关S0003-6951共00兲02534-1兴
energy with respect to the metal Fermi level. In this letter we
investigate the affect of coadsorbed oxygen on band alignment at model phenyl-aluminum interfaces, using ultraviolet
共UV兲 photoemission to determine the HOMO binding energy. Benzene was used as a model for larger oligophenyls,
as it can be readily dosed at high purity in UHV, and the
details of its interaction with clean Al is known, both theoretically and experimentally.5 Benzene can also be considered a more general model for the adsorption of aromatic
molecules, with, for example, the interaction of thiophenes
with aluminum being of essentially the same form.6
The experiments were performed with a Vacuum Generators ADES 400 spectrometer, base pressure 10⫺10 mbar,
using He I radiation (h ␯ ⫽21.2 eV), giving an overall energy
resolution of ⬃100 meV. An Al共111兲 single crystal was used
as the substrate, cleaned using sputter-anneal cycles. The
oxygen was dosed at room temperature, while the benzene
was dosed at 90 K, with photoemission spectra recorded at
the same temperature, as it desorbs from Al共111兲 at 145 K.5
The interaction of benzene with clean Al共111兲 is very weak,
being essentially electrostatic.5 In the photoemission spectra
of a monolayer of benzene on Al共111兲,5 there are no relative
shifts of orbital binding energies on adsorption. This is unlike the case for adsorption on transition metals,7 where a
much stronger bond occurs, involving ␲ donation, giving
rise to the so-called ␲ stabilization, i.e., a relative shift to
higher binding energy of the HOMO ␲ orbitals. However, in
the case of benzene/Al共111兲 a bond occurs without charge
transfer, where, despite the weak interaction, an interface
dipole is formed.5
Figure 1 shows the effect of oxygen exposure on the
work function of Al共111兲, determined from the secondary
electron cutoff of the photoemission spectrum. Note that the
actual work function can differ by up to 1 eV from the literature value of 4.4 eV depending on the oxygen coverage.
The adsorption of oxygen on Al共111兲 has been studied in
detail elsewhere.8 The details are rather complex, but can be
summarized as follows. Initial oxygen deposition on Al共111兲
does not result in the formation of aluminum oxide, but
rather of a chemisorbed phase. Oxide formation only begins
at a dose of 50 L, equivalent to a coverage of 20% of the
surface area. Thereafter clean Al, islands of chemisorbed
oxygen, and oxide islands coexist until the surface is cov-
Band alignment at the organic-inorganic interfaces in devices based on electroactive organic materials is clearly critical to device performance.1 At the electron injecting interface the optimal alignment is one which has the lowest
unoccupied molecular orbital 共LUMO兲 of the organic material close to the Fermi level of the metallic contact. This in
turn implies a high binding energy, relative to the metal
Fermi level, of the highest occupied orbital 共HOMO兲. Based
on the assumption of vacuum level alignment at such
interfaces,1 the electron injecting contact is often chosen to
be a nominally low work function metal, while the hole injecting contact is usually the transparent conductor indium
tin oxide 共ITO兲. It is becoming recognized that, in fact,
vacuum level alignment does not occur,2 and that the actual
band alignment depends on the interface dipole induced by
the particular interaction between molecule and contact
material.2 When considering the choice of contact material,
literature values for the metal work function are generally
quoted. Since work function is, of course, a surface property,
the literature values are based on atomically clean surfaces
measured in ultra high vacuum 关共UHV兲, i.e., ⬍10⫺9 mbar].
Aluminum is one of the most widely used cathodes, chosen
because it is regarded as having a low work function relative
to that of ITO. However, the work function of clean aluminum is 4.4 eV, while that of ITO ranges from 4.1 to 4.9 eV
depending on the surface cleaning technique used.3 Thus, by
the earlier criterion it should actually be a poor cathode in
many circumstances. Since most device preparation takes
place in vacua of the order 10⫺6 mbar, the metal interfaces
will not be atomically clean, and this will affect both the
work function and the surface chemistry.
In an earlier letter4 it was shown that the onset voltage
for luminescence of a sexiphenyl-based light emitting diode
depended on the evaporation rate of the aluminum contact.
These contacts were evaporated in a vacuum of 2
⫻10⫺6 mbar, and as a result oxygen was incorporated at the
sexiphenyl/aluminum interface.4 The devices with the slower
evaporation rate had the highest concentration of oxygen at
this interface, and had a much lower onset voltage, implying
a more optimal band alignment, i.e., a high HOMO binding
a兲
Author to whom correspondence should be addressed; electronic mail:
blyth@kfunigraz.ac.at
0003-6951/2000/77(8)/1212/3/$17.00
1212
© 2000 American Institute of Physics
Appl. Phys. Lett., Vol. 77, No. 8, 21 August 2000
FIG. 1. Work function of Al共111兲 as a function of oxygen exposure, in
langmuirs 共L兲, where 1L⫽10⫺6 mbar s.
ered, which requires 1000 L of oxygen. At this point chemisorbed and oxide areas coexist, with an amorphous oxide
formed at higher exposures. Aside from the clean surface,
benzene was deposited onto Al共111兲 dosed with 150, 1000,
and 2000 L of oxygen. These correspond to the mixed, the
monolayer, and the amorphous oxide phases, respectively.
Photoemission spectra of monolayer benzene coverages
on the oxygen predosed surfaces showed no relative orbital
binding energy shifts, implying the same form of bonding as
was seen for benzene on clean Al共111兲,5 i.e., electrostatic.
Spectra of the HOMO, the 1e 1g ␲ orbitals, of condensed
films of benzene on the different substrates are shown in Fig.
2. The HOMO binding energy clearly increases as the oxygen predose is increased, which in turn implies that the
LUMO is closer to the metal Fermi level, reducing the electron injection barrier. Note that biggest jump in the HOMO
binding energy occurs between clean Al and a predose of
150 L of oxygen. The oxygen coverage at this dose is much
less than a monolayer with chemisorbed oxygen predominat-
FIG. 2. UV photoemission spectra of the HOMO of thin films of benzene on
clean and oxygen exposed Al共111兲. The energy scale is relative to the substrate Fermi level.
Blyth et al.
1213
FIG. 3. Variation of the HOMO binding energy of thin films of benzene, on
clean and oxygen-exposed Al共111兲, with substrate work function.
ing over oxide.8 The sexiphenyl devices with the lowest onset voltage showed an oxygen/Al ratio of 0.2 at the metal/
organic interface, with no oxide seen in Al Auger spectra.4 It
is this oxygen induced shift, rather than the evaporation rate
per se, that accounts for the improved device performance
when evaporating the Al slowly in an oxygen-containing environment. Note that the relative effect is likely to be greater
in sexiphenyl since its band gap is significantly smaller than
that of benzene.9
The HOMO binding energy versus the substrate work
function is plotted in Fig. 3. The band alignment at organicmetal interfaces depends on the precise nature of the
interaction.10 However, since the benzene/Al共111兲 interaction is of the same, electrostatic, form regardless of oxygen
coverage, it might be expected that there would be a simple
linear relationship in this case. Figure 3 shows that this is
apparently not so, with in particular the 150 L dose producing a high binding energy. However, this substrate is a mixed
phase, and close inspection of Fig. 2 shows that the HOMO
width is broader on this substrate than on the others. This
implies that there are several local HOMO energies coexisting, and that the linear relationship may be preserved on a
local scale. It is, however, further evidence that the simplistic
assumption of such a relationship can be hazardous, even in
cases where it might be expected to hold.
It should not be forgotten that the major component of
the residual gas in a high vacuum (10⫺6 mbar) system is
water, an oxygen containing molecule which is if anything
actually more reactive to aluminum than oxygen itself.11 The
results presented earlier suggest that the control of the
oxygen/water partial pressures during contact evaporation is
extremely important. Not only can this be used to optimize
band alignment for device performance, but the reproducibility of this performance will clearly be dependent on the reproducibility of the level of oxygen at the metal/organic interface. Here we have shown that oxygen improves the band
alignment at an aluminum/organic interface. However, we
caution that this should not be regarded as a general case—
for example oxygen at the thiophene/copper interface has the
opposite effect.10
This work was supported by the Austrian Science Foundation through the SFB:Electroactive material.
1214
1
Blyth et al.
Appl. Phys. Lett., Vol. 77, No. 8, 21 August 2000
W. R. Salaneck, S. Stafström, and J.-L. Bredas, Conjugated Polymer Surfaces and Interfaces 共Cambridge University Press, Cambridge, 1996兲.
2
H. Ishii, K. Sugiyama, E. Ito, and K. Seki, Adv. Mater. 11, 972
共1999兲.
3
K. Sugiyama, H. Ishii, Y. Ouchi, and K. Seki, J. Appl. Phys. 87, 295
共2000兲.
4
N. Koch, A. Pogantsch, E. J. W. List, R. I. R. Blyth, M. G. Ramsey, F. P.
Netzer, and G. Leising, Appl. Phys. Lett. 74, 2909 共1999兲.
5
R. Duschek, F. Mittendorfer, R. I. R. Blyth, F. P. Netzer, J. Hafner, and
M. G. Ramsey, Chem. Phys. Lett. 318, 43 共2000兲.
6
R. I. R. Blyth, F. Mittendorfer, R. Duschek, F. P. Netzer, J. Hafner, and
M. G. Ramsey 共unpublished兲.
7
F. P. Netzer, Langmuir 7, 2544 共1991兲.
8
H. Brune, J. Wintterlin, J. Trost, G. Ertl, J. Wiechers, and R. J. Behm, J.
Chem. Phys. 99, 2128 共1993兲.
9
M. G. Ramsey, D. Steinmüller, M. Schatzmayr, M. Kiskinova, and F. P.
Netzer, Chem. Phys. 177, 349 共1993兲.
10
G. Koller, R. I. R. Blyth, S. A. Sardar, F. P. Netzer, and M. G. Ramsey,
Appl. Phys. Lett. 76, 927 共2000兲.
11
J. E. Crowell, J. G. Chen, D. M. Hercules, and J. T. Yates, Jr., J. Chem.
Phys. 86, 5804 共1987兲.