Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~pgreen/Papers/htxs_97.ps Äàòà èçìåíåíèÿ: Sat Dec 21 23:25:03 1996 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 00:33:05 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: viking 2

X­ray Spectroscopy of QSO Absorbers
Paul J. Green & Smita Mathur
Harvard­Smithsonian Center for Astrophysics, 60 Garden St., Cambridge, MA 02138
Received ; accepted
2 pgreen@cfa.harvard.edu

-- 2 --
ABSTRACT
The optical/ultraviolet (OUV) emission of QSOs has not proved a ready key
to unlock the physics of the broad line region immediately surrounding the central
engine of QSOs. X­ray emission suffers from a similar barrier to our understanding
­ many regions with widely ranging conditions may contribute, each of which may
vary with a different response lag. The study of intrinsic absorption in QSOs may
surmount some of these problems, if sensitive UV and X­ray data are combined.
Recent results from the ROSAT All Sky Survey, and from deep ROSAT
pointings reveal that broad absorption line quasars (BALQSOs) are weak in the
soft X­ray bandpass (ff ox ? 1:8) in comparison to QSOs with normal optical/UV
spectra (hff ox i = 1:4). The ubiquitous association of soft X­ray absorption
with UV absorption suggests that the absorbers in each regime are physically
associated. UV/X­ray studies show great promise for unifying many different
kinds of QSOs through their often uniquely­related absorbers. We argue that since
absorbed QSOs are faint in soft X­rays, a High Throughput X­ray Spectroscopy
(HTXS) mission of wide bandpass and high sensitivity is critical to constraining
the physics of QSO absorption line clouds, which are in turn one of the best hopes
we have for probing the region immediately surrounding the AGN engine.

-- 3 --
1. QSO Emission Lines and Outstanding Mysteries
Most of the ¸ 10; 000 QSOs to date were discovered either via their prominent optical
and ultraviolet (OUV) emission lines, or from their distinct colors in these bandpasses.
Continuum emission has also been studied in detail, but the answers to some surprisingly
fundamental questions remain at large: 1) Do SEDs directly determine emission line
strengths/profiles? 2) What is the balance of gravity and p rad in broad line production?
3) What causes the Baldwin Effect (the observed nonlinear scaling of emission line with
continuum flux) ? Part of the reason for the persistence of such enigmas is probably that
optical/UV spectra of QSOs are remarkably similar, despite the great variety of QSO
SEDs. The overall similarity of QSO emission line spectra had been taken as evidence of
fairly uniform, robust physical conditions in the BELR, which encouraged the assumption
that clouds in the BELR inhabit a narrow swath of parameter space (in density, size,
and ionization parameter). Early photoionization pioneers such as Mushotsky & Ferland
(1984) ran models on a single cloud. Refinements using cloud ensembles showed a reduced
dependence of total line emission on intrinsic QSO SEDs (Binette et al. 1989). Details of
individual clouds or even clouds in a single `zone'' can be lost in the mix, and correlations
between continuum shape and observed line parameters diluted. Baldwin et al. (1995)
reiterate that averaging of emission from clouds with a wide variety of properties (but
uniformly large columns) results in QSO line spectra robustly consistent with those
observed.
Are QSOs thus a P 2 C 2 E (process too complicated too explain; Rushdie 1990) ? Recent
observational progress is beginning to approach the putative supermassive black hole in
nearby AGN empirically, for example through H 2 O megamasers (Greenhill et al. 1995),
and gravitationally broadened/shifted Fe Kff (Fabian et al. 1995). Are OUV spectra, the
first, most detailed, and most common data available for QSOs, going to prove irrelevant

-- 4 --
to our understanding of their intrinsic physics? We here offer that absorption intrinsic to
QSOs will be a crucial key to physics in the broad line region (BLR) nearest the central
engine of QSOs. The study of QSOs showng intrinsic absorption in their OUV spectra is
just beginning to provide a wealth of information, particularly in conjunction with sensitive
X­ray spectra.
2. UV Spectra of BALQSOs
From 10 to 15% of optically­selected QSO spectra show broad absorption line (BAL)
troughs. BALs suggest a line of sight passing through highly ionized, high column density
(NH ¸ 10 20 cm \Gamma2 ) absorbers, flowing outward from the nuclear region at speeds up to
0.1 ­ 0.2c. Low BAL cloud covering factors and the absence of emission lines at the high
velocities observed in BALQSOs, along with the similarity of emission­line and continuum
properties of BAL and non­BALQSOs (Hammann, Korista, & Morris 1993, hereafter HKM;
Weymann et al. 1991, hereafter WMFH) suggest that all radio quiet (RQ) QSOs (which
in turn comprise ¸ 90% of all QSOs) have BAL clouds. Far from being exotic, since
BAL clouds probably inhabit a region just outside the BELR (r BALR – r BELR ¸ 0:01pc),
BALQSOs offer a special probe of gas dynamics in typical QSOs.
Analysis of UV BALs yields column densities of a few ions (e.g., OVI–1035, CIV–1549,
NV–1240), but little information on the ionization state, and so even less on the total
column of the absorber. Furthermore, OUV data (N intr
H ¸ 10 19 to 10 20 ) predict very little
soft X­ray absorption (Ü !! 1).

-- 5 --
3. BALQSOs in SOFT X­rays
Green et al. (1995) recently demonstrated that, when compared to normal RQ QSOs,
BALQSOs are weak in the soft X­ray bandpass: Their sample of 36 BALQSOs was chosen
from a large uniformly­selected QSO sample (the LBQS), as observed during the ROSAT
All­Sky Survey (RASS). Although the short (¸ 600 sec) exposure times of the RASS meant
that the upper limits (for 35 of the 36 QSOs) were not very sensitive, by stacking the X­ray
data, they were able to show that their uniform BALQSO sample was X­ray quiet at the
99.5% significance level compared to carefully chosen comparison RQ QSO samples.
Pursuing this line of study with deep pointed observations from the ROSAT PSPC,
Green & Mathur (1996; hereafter GM96) confirmed that BALQSOs are weak in the soft
X­ray bandpass in comparison to RQ QSOs with normal OUV spectra. While a comparison
sample of 10 similar RQ QSOs (from Laor et al. 1994) without BALs yielded a sample
mean of ff ox = 1:45 \Sigma 0:08, nine out of twelve reputed BAL QSOs were not detected by
ROSAT deep pointings, yielding ff ox ? 1:8. Of three remaining BALs in GM96, one was
too distant for an interesting lower limit to ff ox . One was detected, yielding ff ox = 1:98,
and N intr
H ¸ 10 23 cm \Gamma2 . Only one QSO, PG 1416--129, appeared to be X­ray bright, with a
value of ff ox = 1:4 typical of non­BAL QSOs. However, a new HST spectrum of this object
reveals some surprises (Green et al. 1997, in preparation) which bring it into line with the
hypothesis that BALQSOs are all weak in the soft X­ray bandpass.
If indeed the central continuum source of BALQSOs is similar to that of other QSOs,
as argued above, the instrinsic absorbing columns required to explain the observed soft
X­ray deficit must be at least 100 times higher than those inferred from the UV data alone
(N intr
H ? 2 \Theta 10 22 cm \Gamma2 ). In contrast, the non­BAL sample of PSPC observations shows no
evidence at all for absorption.

-- 6 --
1200 1400 1600 1800 2000 2200
0
2
4
6
8
10
Wavelength (angstroms)
NOAO/IRAF V2.10.4EXPORT pgreen@etre Tue 14:59:21 24­Sep­96
[0226m1024_zeq0]: Writeiraf INDEF ap:1 beam:1
Figure 1. The rest frame spectrum of a typical BAL QSO 0226--1024 (WMFH) overlaid
with a composite spectrum (Francis et al. 1991) consisting mostly of unabsorbed radio­
quiet QSOs. The intrinsic continuum and the emission lines of the unabsorbed spectrum are
consistent with the unabsorbed part of the BAL QSO spectrum.

-- 7 --
Figure 2. Histogram of countrates (per 600sec) for 1000 random subsamples of 36 QSOs from
a list of 363 definite (z ? 1:3) non­BAL QSOs, for comparison to the true BAL sample of 36
QSOs. All samples are taken from the ROSAT All Sky Survey (RASS) observations of Large
Bright Quasar Survey (LBQS) QSOs (Green et al. 1995). The number of stacked counts for
the BAL sample is indicated, with errors (square root of the area­normalized background
counts) shown as dotted lines to either side. Only 5 in 1000 trials generates equal or fewer
counts than the BAL sample. The sample of 36 stacked BAL QSOs thus appears to be X­ray
quiet at the 99.5% level.

-- 8 --
4. Soft X­ray Absorption and Spectral Energy Distributions
For comparison of an observed soft X­ray count rate (or its upper limit) to that expected
from a typical RQ QSO, GM96 first derived an optical luminosity derived from the observed
magnitude. From the optical luminosity, an intrinsic soft X­ray flux is then predicted using
the mean slope ff ox of a hypothetical power law connecting rest­frame 2500 š A and 2 keV
as derived for non­BAL RQ QSOs in the ROSAT PSPC (Green et al. 1995). The expected
PSPC count rate corresponding to this flux is derived assuming a power­law model with
ff E = 1:5 (the mean PSPC ff E for radio­quiet quasars; Laor et al. 1994) and the Galactic
column density. If the expected PSPC count rate from PIMMS was greater than observed
(as is the case for all BALQSOs), the column density of the absorber in the spectral model
was increased until the observed count rate/upper limit was reached. This resulted in a
lower limit to the `observed' intrinsic column N intr
H , on the assumption that the absorber
is cold gas with solar metallicity Z = Z fi . Many of these assumptions could be bypassed,
greatly increasing the reliability of the conclusions, if high signal­to­noise ratio (SNR), wide
bandpass X­ray spectroscopy could be obtained. The implications of currently required
assumptions are briefly outlined below:
Although Z = Z fi is assumed, line ratios in high­z (non­BAL) QSOs suggest
2 ! Z=Z fi ! 10 (Hamann & Ferland 1993), and estimates of N intr
H are inversely proportional
to the assumed metallicity. However, unless AGN have abundances 100 times solar
(Turnshek 1988), column densities at least an order of magnitude higher for BALQSOs
than is typical for RQ QSOs are still required.
If instead of neutral gas, U ¸ 0:1 is assumed (based on photoionization models of the
OUV data; HKM), then our N intr
H estimates would rise by about an order of magnitude.
Even higher ionization (U –1) are likely to be required.
Perhaps the most critical assumption is that BALQSOs harbor the same intrinsic

-- 9 --
(unabsorbed) spectral energy distribution (SED) as normal RQ QSOs. So far, PHL 5200
provides the only direct evidence for this. An ASCA spectrum of PHL5200 shows a best­fit
spectral slope ff E = 0:6 +0:9
\Gamma0:6 is normal for an RQ QSO. The resulting ff ox = 1:5, is also normal
for an RQ QSO. Excess absorption is required at 99% confidence: N intr
H = 0:2 \Gamma 4 \Theta 10 23 Z fi
Z
cm \Gamma2 at the source (Mathur, Elvis, & Singh 1996). This indeed confirms that, although
BALQSOs are X­ray quiet, it is strong absorption in the BAL region that makes them
appear faint to low energy X­ray experiments. Since the required intrinsic absorbing column
density is two to three orders of magnitude larger than that derived from UV estimates of
column densities in BALQSOs, BAL systems are much more highly ionized than previously
thought. Unfortunately, PHL 5200 is one of less than a handful of QSOs bright enough for
this sort of analysis with ASCA.
Since BALs span a wide range in velocity, most models of BAL cloud outflows require
either an ensemble of clouds along the line of sight, or winds blown off the surface of the
accretion disk (deKool & Begelman 1995; Murray & Chiang 1995). Most BALQSO models
reproduce columns as inferred from OUV data alone (N ¸ 10 20 cm \Gamma2 ). However, entrained
disk wind models produce column densities of 10 23 \Gamma 10 24 cm \Gamma2 , consistent with the N intr
H
so far derived from ROSAT PSPC observations.
If the BALQSO phenomenon is indeed only a viewing angle effect, then hard X­ray
emission (?10 keV) should show fluxes and slopes typical for normal RQ QSOs. The results
for PHL 5200 look promising, and we have been awarded XTE time to further pursue hard
X­ray studies of BALQSOs LBQS 2212­1759 and PG1700+518.

-- 10 --
5. Are the UV And X­Ray Absorbers The Same Material?
Intrinsic ``associated'' absorption has been a well known but little understood
phenomenon in ultraviolet studies of both low and high luminosity active galactic nuclei
(AGN) and in X­ray observations of the former. The possibility that both UV and X­ray
absorption is produced by the same gas is an attractive one, but attempts to reconcile the
two had until recently been unsuccessful (e.g. Kolman et al 1993).
ROSAT observations of the quasar 3C351 (z=0.371) have revealed the presence of a
`warm absorber' in its X­ray spectrum showing a K­edge due to OVII or OVIII (Fiore
et al 1993). 3C351 also shows unusually strong, associated, high ionization absorption
lines in the HST ultraviolet spectrum (Bahcall et al 1993) including OVI ––1031; 1037.
Through detailed photoionization modeling, Mathur (1994) and Mathur et al. (1995) found
an excellent match between the X­ray and UV absorber properties. The combination of
both datasets allows strong constraints to be placed upon the physical conditions of the
absorber. In 3C351, the absorber appears to be well­described by material quite similar to
that seen in BALQSOs; highly ionized, outflowing, low density material situated outside
the broad emission line region (BELR).
Since the discovery of a joint X­ray/UV absorber in an `X­ray quiet' quasar 3C351,
consistent properties of UV and X­ray absorbers have been noted in a `red' quasar 3C212
(Mathur 1994), and in the variable Seyfert galaxy NGC5548 (Mathur et al 1995). This
opens up the possibility that the wide range of associated absorbers could form a continuum
of properties deep in the active nucleus.

-- 11 --
6. BAL Variability
Absorption line variability has been measured both in narrow associated absorbers, and
in BALs (Koratkar et al. 1996; Barlow et al. 1992). About 15% of BALs show variability
in the residual absorption intensity (normalized flux within the line), at the level of Ÿ 20%
(Barlow 1994; Smith & Penston 1988). The resulting change in column is measured to be
\DeltaN ? 10 14 cm \Gamma2 . Is there a concomitant change in absorption in the soft X­ray regime?
A demonstration of correlated variability between BALs and soft X­ray flux in true
BALQSOs would provide strong evidence that the UV and soft X­ray absorbers are
indeed physically associated. Since BALs show long stable states between variability, an
event found in a campaign of ground­based optical monitoring could provide a target of
opportunity for followup X­ray spectroscopy. Unfortunately, such studies present a daunting
task for the current generation of X­ray telescopes, given the weak soft X­ray fluxes of
BALQSOs. However, with the larger effective areas proposed for high throughput X­ray
spectroscopy missions, UV/X­ray variability studies could not only confirm the physical
identity of UV/X­ray absorbers, but also determine whether BAL variability is a result of
a true change in column (e.g., due to motion of the absorber out of the line of sight), or
in ionization. A variable photoionizing continuum might cause a change in the ionization
levels, and thus a redistribution of the fractional abundances of the ions, resulting in the
strengthening or weakening of the absorption lines.
Limits on n e and the response time scale of the BAL changes indicate that
0:1 Ÿ r BAL Ÿ100pc from the ionizing source, so that the inner regions nearest the massive
black hole can be studied using BALs. The great advantage of studying absorption rather
than emission variability is that, unlike in the BELR, the BAL variation mechanism travels
in step with the continuum emission, rendering unnecessary the prohibitively complicated
(and observationally expensive) deconvolution of a temporal response function.

-- 12 --
7. The Potential of HTXS as a PROBE of the BALR
Some very achievable goals for HTXS study of absorbed QSOs are outlined above: 1)
Verify that intrinsic SEDs are consistent for BAL and RQ QSOs. 2) Constrain the column
density (NH ), ionization parameter (U) and the ion abundances of the absorbing clouds. 3)
Describe the location and dynamics of the absorbing gas.
One of the fundamental such questions still eluding an answer is why BALs are found
only among radio­quiet QSOs. Only in the hard X­ray regime do BAL, non­BAL, RQ and
RL QSOs all boast spectra that are both bright and distinct. High throughput spectroscopy
in this bandpass would go a long way toward establishing a connection among the different
kinds of QSOs and related absorbers: BALQSOs ! high ionization associated absorbers
! high + low ionization absorbers ! MgII absorbers ! red quasars. Thus HTXS offers
promise for the still­elusive unification of diverse types of AGN. An observational mapping
of the apparent diversity of AGN represents a deprojection onto a more physical basis set
that must consist of at least such fundamental physical properties as 1) black hole mass and
accretion rate, 2) orientation of accretion disk and jet normal to the line of sight and the
plane of the host galaxy 3) abundances, densities, and ionization of intervening absorbers.
8. CONCLUSION
A marked soft X­ray deficit is a defining characteristic of BALQSOs, caused by
absorption by high column density, ionized material along the line of sight. Such absorption,
especially in tandem with UV absorption, can be used to nail down the intrinsic column
and the ionization state of clouds deep inside the QSO nucleus. These in turn will greatly
illuminate continuing debate surrounding such questions as the intrinsic SEDs of QSOs, the
balance of gravity and p rad in the broad line region, production, and the effect of changes in
luminosity and ionization on clouds. A HTXS mission of wide bandpass and high sensitivity

-- 13 --
is critical to constraining the physics of QSO Broad Absorption Line clouds, which are in
turn one of the best hopes we have for understanding the region immediately surrounding
the AGN engine.
9. Acknowledgements
Thanks to Craig Foltz for providing QSO spectra. PJG acknowledges support provided
by NASA through Grant NAG5­1253, and through GO­06528.01­95A awarded by the
Space Telescope Science Institute, which is operated by the Association of Universities for
Research in Astronomy, Inc., under NASA contract NAS5­26555. SM acknowledges support
from NASA grant NAGW­4490 (LTSA).

-- 14 --
Figure 3. ASCA X­ray spectra of PHL 5200 from Mathur, Elvis & Singh 1996. This shows
the ASCA SIS and GIS spectra (upper and lower data points, respectively) of the BAL QSO
PHL 5200. The ordinate is in counts/keV. The best fit model shown as histograms consists
of an absorbed power law of slope ff E = 0:6. Although this slope is normal for a RQ QSO,
excess absorption is required at 99% confidence: N intr
H = 0:2 \Gamma 4 \Theta 10 23 Z fi
Z
cm \Gamma2 at the
source. This confirms that, although BALQSOs are X­ray quiet, it is strong absorption in
the BAL region that makes them appear faint to low energy X­ray experiments.

-- 15 --
REFERENCES
Bahcall, J. N. 1993, ApJS, 87, 1
Barlow, T. A. et al. 1992, ApJ, 397, 81
Barlow, T. A. 1994, PASP, 106, 548
Binette, L., Prieto, A., Szusziewicz, E., & Zheng, W. 1989, ApJ., 343, 135
DeKool, M. & Begelman, M. C. 1995, ApJ, 455, 448
Fabian, A. C. et al. 1995, MNRAS, 277, 11
Fiore, F., Elvis, M., Mathur, S., Wilkes, B. J., & McDowell, J. C. 1993, ApJ, 415, 129
Francis, P. J., Hewett, P., Foltz, C., Chaffee, F., Weymann, R. J., & Morris, S. 1991, ApJ,
373, 465
Green, P. J., et al. 1995, ApJ, 450, 51
Green, P. J. & Mathur, S. 1996, ApJ, 462, 637
Greenhill, L. J., Jiang, D. R., Moran, J. M., Reid, M. J., Lo, K. Y., & Claussen, M. J. 1995,
ApJ, 440, 619
Hamann, F. & Ferland, G. J. 1993, ApJ, 418, 11
Hamann, W. R., Korista, K. T., & Morris, S. L. 1993, ApJ, 415, 541 (HKM)
Hines, D. C., & Wills, B. J. 1995, ApJ Letters, 448, L69
Kolman, M. et al. 1993, ApJ, 403, 592
Koratkar, A. et al. 1996, ApJ, 470, 378
Laor, A., Bahcall, J. N., Jannuzi, B. T., Schneider, D. P., Green, R. F. 1995, ApJS, 99, 1
Mathur, S. 1994, ApJ, 431, L75
Mathur, S., Elvis, M., & Singh, K. P. 1995, ApJ Letters, 455, L9

-- 16 --
Murray, N., & Chiang, J. 1995, ApJ Letters, 454, L105
Mushotzky, R. F., & Ferland, G. J. 1984, ApJ, 278, 558
Rushdie, S. 1990, in Haroun & the Sea of Stories (New York: Viking and Granta)
Smith, L. J. & Penston, M. V. 1988, MNRAS, 235, 551
Turnshek, D. A. 1988, in QSO Absorption Lines: Probing the Universe, ed. S. C. Blades,
D. A. Turnshek, and C. A. Norman (Cambridge: Cambridge University Press), 17
Weymann, R. J. et al. 1991, ApJ, 373, 23 (WMFH)
This manuscript was prepared with the AAS L A T E X macros v4.0.