On the significance of He-like plasma diagnostics
for the study of photoionized media
Anabela C. Gonçalves 1,2
Olivier Godet 3, Anne-Marie Dumont 1 ,Suzy Collin 1
1 Paris
Observatory (LUTh), France
Observatory (CAAUL), Portugal
3 Leicester University, UK
2 Lisbon Astronomical
X-ray Grating Spectroscopy meeting
Cambridge, July 11-14, 2007
Astrophysical plasmas
Plasmas




Ubiquitous, extremely common, represent 99% of known matter
“Artificial” plasmas: nuclear fusion, spatial propulsion, neon signs, TV
“Natural” plasmas: auroras, solar wind, stellar coronae, nebulae, in AGN
X-ray plasmas cover a wide range in properties: T, nH, NH, x, etc.
X-ray gas can get ionized by
T (K)
controled fusion
inertial
confinement
 Collisions (~ keV)
• X-ray bright starburst regions
nebulae
solar winds
thunder
WA
• Stellar coronae
• SNRs
neon
lights
 Photoionization (~ eV)
propeler
aurorae
flames
• X-ray binaries
• AGN: BLR, NRL, Warm Absorber
particles/m3
solar core
Plasma Diagnostics in the X-rays
Astrophysical plasma diagnostics
Brickhouse et al. (2000)
 Rely on spectroscopy measurements and/or
simulations => compute line-ratios
 Conspicuous lines, close in l/Energy =>
minimize calibration errors
 High-quality, medium-resolution spectra
(R ≥ 300) => needed to resolve the lines
Gu et al. (2006)
Diagnostics in the X-rays
 X-band very rich: many lines from
each ion, many ions from each element
 Different ions probe different gas conditions:
H-like, He-like in high-Z and low-Z elements
 WA conditions: CV, NVI, OVII (well resolved,
close in E range, easy to use as
diagnostics)
 High enough resolution spectra achieved
He-like lines as plasma diagnostics
G(T) = (F+I) / R
R(n) = F / I
Porquet & Dubau (2000)
He-like triplet lines
 n=2→1 transitions produce 3 important lines
• Resonance line (R or w)
• Intercombination line (I or x+y)
• Forbidden line (F or z)
 G and R diagnostics (Gabriel & Jordan 1969)
G(T) = (F+I)/R = (z+x+y)/w => temperature T
• R line mainly excited by photons, also by
recombinations and by collisions (depend on T)
• F and I lines mainly excited by cascades from
upper levels and by recombination
R(n) = F/I = z/(x+y) => density n
• Collisions (depend on n) depopulate the F level
• Whereas the I level get more populated
 Triplet lines appear different in collisional and
photoionized plasmas
(w)
metastable
(x+y)
(z)
G and R applications
G(T) = (F+I) / R
R(n) = F / I
Applications and Complications
 G and R used in solar plasma studies (Freeman & Jones 1970; McKenzie et al.
1978,1982)
 Then applied to collisional, extra-solar plasmas (Ness et al. 2001; Porquet et al. 2001)
 Attempts at photoionized extragalactic plasmas (hybrid/photoionized: Porquet &
Dubau 2000; collisional/photoionized: Bautista & Kallman 2001)
 Later, Porquet et al. (2001) added a “photoexcitation” term to their computations
 Even when photoexcitation and its effects on the resonant lines taken into account
(e.g.
Sako
2000; Kinkhabwala 2002), or when using performing codes (e.g.
Cloudy,
XSTAR)
Problem:
“Photoionization conditions” = spectrum
due to recombination,
4 K) compatible
Problem:
Radiation
field
assumed
(BB
T~10
with OB
Problem: Thermal
eq. not
consistently
computedexcitation,
with ionization
eq. stars
radiative
cascades,
and collisional
only
Incidence
onnot
Visible-UV
lines,
but none
resonant
lines excitation
Transfer
eq.
solved, not
adequate
to on
model
moderately
thick media
No photoionization/photoexcitation
involved
=>=>
Not
applicable
inin
the
X-ray
absorber/emitter
inin
AGN!
=>
Not
applicable
in
the
X-ray
absorber/emitter
in
AGN!
Not
applicable
the
X-ray
absorber/emitter
AGN!
TITAN photoionization code
TITAN code (Dumont et al. 00; Collin et al. 04; Gonçalves et al 07)
 A stationary, photoionization-transfer code developed at Paris Observatory (LUTh)
 Code optimized for optically thick media (NH ~> 1022 cm-2), but also thin media
 Computes the exact transfer for ~4000 lines (same as Cloudy) and the continuum
 Atomic data: H, He, C, N, O, Ne, Mg, Si, S, Fe (UTAs), good He-like description
 Assumes a 1D plane-parallel geometry: slab of gas illuminated on one side by an
irradiating X-ray source (flux and SED continuum)
 Self-consisting ionization and thermal eq. computation, provides gas structure in
Temperature, density, pressure, ionization
 Gives the spectra in transmission, plus emission and reflection in multiple directions
 Modes include constant Density, Gaseous Pressure or Total Pressure
 Deal with thermal instability, computing models for the hot and cold stable solutions
TITAN photoionization code
Computes the transfer of lines and continuum
 No escape probability approximation, but throughout calculations (ALI method)
Gonçalves et al. 2006a
Multi-angle spectra
 “normal direction” + 5 cones
(18°, 40°, 60°, 77°, 87°)
 computes the transmitted,
reflected and emitted flux
Gonçalves et al. 2006a
OVIII l 18.97
● Chandra data
TITAN model

Can account for P Cyg-like profiles
 Can simulate the expected spectrum
in function of the line-of-sight
He-like line-ratios dependency
G(T) = (F+I) / R
R(n) = F / I
G and R depend on a lot of factors…
 Optical thickness
• For a very large NH, G reaches a constant value
• Degeneracy: same G value for a small NH and small x, or for a
large NH and large x
• Multiple ion features must be used to disentangle the possibilities
Coupé et al. (2004)
• G varies strongly with NH for a given ionization parameter x
(= L/nH2.D)
• G is very sensitive to microturbulence => NH deduced from G in
turbulent gas would be larger
• R does not depend on microturbulence and abundances
 Exact Transfer vs. Escape Probability (EP)
• Large NH: interactions with other ions causing
photon destruction must be taken into account
• Not properly done by EP => exact transfer
photoionization code needed => TITAN
Godet et al. (2004)
 Microturbulence
Dumont et al. (2003)
He-like line-ratios dependency
G(T) = (F+I) / R
R(n) = F / I
G and R depend on a lot of factors…
 Plasma equilibrium conditions (constant Ptot vs. constant Density)
• Single T for whole medium currently assumed, but T varies along the WA
• Stratification of the WA best explained by gas in constant Total Pressure (e.g. NGC 3783,
Gonçalves et al. 06)
• He-like region T differs from that of the H-like region
• G is different in constant density and constant pressure models
Seyfert 2
F
R
I
He-like line-ratios dependency
G(T) = (F+I) / R
R(n) = F / I
G and R depend on a lot of factors…
Gonçalves & Godet (2007)
 Orientation effects
• Know the flux angular-dependence =>
G and R in function of the line-of-sight
• G is extremely sensitive to the l.o.s.
while R does not depend on orientation
• Relative contribution of Reflection,
Emission and Absorption to the
“observed” spectrum depends on
medium size, geometry…
• Type 1 and 2 AGN assume different
geometry => G and R don’t convey the
same information in Sey1 and Sey2!
Some assumptions
Obscuring torus opening angle ~45°
Accretion disk optically thick
Observer located at infinity
Orientation effects: Sey 1 vs. Sey 2
G(T) = (F+I) / R
R(n) = F / I
Orientation effects: Seyfert 1 vs. Seyfert 2
 Seyfert 1 with WA (50% of Sey 1)
• Absorbing material on the l.o.s.
reprocesses the primary spectrum
• Also contribution from Emission, the
proportion depends on geometry of
the whole gas
• Example => half contribution from
Emission, half from Absorption
 Seyfert 1 without WA (50% of Sey1)
• No absorbing material on the l.o.s.
• The primary source is thus visible
• There is contribution from Emission
from material not on l.o.s.
• Example => half contribution from
Emission, half from primary continuum
Seyfert 1
He-like line-ratios dependency
G(T) = (F+I) / R
R(n) = F / I
Orientation effects: Seyfert 1 vs. Seyfert 2
 Seyfert 2s
• Edge-on observer detects some of the
reprocessed primary spectrum => flux
from large opening angles (≥ 45°)
Seyfert 1
• Observed spectrum: contribution from
Emission and Reflection components
• Different scenarios are possible: 50%
reflection and 50% emission; or 90%
reflection and 10% emission, etc.
Seyfert 2
• G is systematically lower in Sey 2, as
the R line is comparatively higher
• This is true for both constant density
and constant Ptot models
• R is not affected by orientation much
Some comments
G(T) = (F+I) / R
R(n) = F / I
More complications than applications?
 G and R are particularly tricky to use in the case of thick, photoionized, stratified
media such as the Warm Absorber in AGN
• Because a transfer-photoionization code is needed
• Because the WA is stratified and a single T is not enough to describe the whole medium
• Because Seyfert 1 and Seyfert 2 have different geometry and convey different information
Things we have noticed
 Constant Pressure plasma tend to have higher G and R values
 In general, Seyfert 2 have higher G than Seyfert 1
 R is less affected by orientation effects (because of the resonant line)
 Degeneracy: same G may correspond to a high-ionization, constant density medium
in Seyfert 1, or to a low-ionization, constant total pressure medium in Seyfert 2
Unless you know the WA geometry and physics… be careful!
 Taking into account multiple spectral features could help disentangle options
Additional slides
Anabela C. Gonçalves
Paris Observatory (LUTh), France
LUTh seminar
Meudon, January the 18th, 2007
He-like line-ratios dependency
G(T) = (F+I) / R
R(n) = F / I
G and R depend on a lot of factors…
 Plasma equilibrium conditions (constant Ptot vs. constant Density)
• Single T for whole medium currently assumed, but T varies along the WA
• Stratification of the WA best explained by gas in constant Total Pressure (e.g. NGC 3783,
Gonçalves et al. 06)
• He-like region T differs from that of the H-like region
• G is different in constant density and constant pressure models
Constant Density
Constant Total Pressure
NH=3 1023 cm-2, nH=107 cm-3, x=1000 (U=13.2, Ux= 1.8)
He-like ions
H-like ions
Gonçalves & Godet (2007)
He-like ions
H-like ions
High-, medium-, low-ionization WA
Temperature profiles
G line-ratios, medium-ionization WA
G line-ratios
R line-ratios, medium-ionization WA
R line-ratios
He-like ions atomic model
15 levels
He-like line-ratios dependency
G(T) = (F+I) / R
R(n) = F / I
G and R depend on a lot of factors…
 Orientation effects
• G and R do not convey the same information in type 1 or type 2 AGN
• G is extremely sensitive to the observation angle, while R does not depend on
orientation
• G is systematically lower in the case of Seyfert 2s, both in constant density or
constant pressure models
TITAN grids of models
Need for a database of theoretical results
 Code has a real potential: unique in dealing with thick media, exact transfer,
thermal instabilities and proper total pressure equilibrium computations
 TITAN models compute the exact transfer for ~4000 lines and the continuum
=> long computation times (~30h for constant Ptot model)
 TITAN allows for the modeling of regions in total pressure equilibrium, solves
the thermal instabilities => complex models, check for convergence
 Several domains of applicability: physical parameters can vary over a large
range => needs quick, first-order estimation of the physical parameters prior
to complete modeling
 To compare TITAN physical modeling with other tools, to model and to
simulate X-ray data in XSPEC => need for table FITS models
Opening TITAN to the community
Interoperability with XSPEC
 Grid of models converted into XSPEC model tools => easily applicable by a
larger astrophysical community
 Scientific applications: theoretical modeling of Active Galactic Nuclei (AGN),
X-ray binaries, Ultraluminous X-ray sources (ULXs), comparison of models
 Observational applications: interpretation of high-quality X-ray data from
Chandra, XMM-Newton, Suzaku, …
 Instrumental applications: preparation of future X-ray missions (Con-X,
Simbol-X, XEUS,…), data simulation