A Chandra HETG Observation of the Quasar H 1821+643 and Its Surrounding Cluster

Abstract

We present the high-resolution X-ray spectrum of the low-redshift quasar H 1821+643 and its surrounding hot cluster observed with the Chandra High Energy Transmission Grating Spectrometer (HETGS). An iron emission line attributed to the quasar at ~6.43 keV (rest frame) is clearly resolved, with an equivalent width of ~100 eV. Although we cannot rule out contributions to the line from a putative torus, the diskline model provides an acceptable fit to this iron line. We also detect a weak emission feature at ~6.9 keV (rest frame). We suggest that both lines could originate in an accretion disk comprised of a highly ionized optically thin atmosphere sitting atop a mostly neutral disk. We search for absorption features from a warm/hot component of the intergalactic medium along the ~1.5Gpc/h line of sight to the quasar. No absorption features are detected at or above the 3 sigma level while a total of six OVI intervening absorption systems have been detected with HST and FUSE. Based on the lack of OVII and OVIII absorption lines and by assuming collisionally ionization, we constrain the gas temperature of a typical OVI absorber to 10^5 < T < 10^6 K, which is consistent with the results from hydrodynamic simulations of the intergalactic medium. The zeroth order image reveals the extended emission from the surrounding cluster. We have been able to separate the moderate CCD X-ray spectrum of the surrounding cluster from the central quasar and find that this is a hot cluster with a temperature of ~10 keV and a metal abundance of ~0.3 Zo. We also independently obtain the redshift of the cluster, which is consistent with the optical results. We estimate that the cluster makes negligible contributions to the 6.9 keV iron K line flux.

Full text

Page 1

arXiv:astro-ph/0109389v1 21 Sep 2001
A Chandra HETG Observation of the Quasar H 1821+643 and Its
Surrounding Cluster
Taotao Fang, David S. Davis, Julia C. Lee, Herman L. Marshall, Greg L. Bryan, AND
Claude R.Canizares
Department of Physics and Center for Space Research
Massachusetts Institute of Technology
NE80-6081, 77 Massachusetts Avenue, Cambridge, MA 02139
fangt@space.mit.edu
ABSTRACT
We present the high-resolution X-ray spectrum of the low-redshift quasar
H 1821+643 and its surrounding hot cluster observed with the Chandra High
Energy Transmission Grating Spectrometer (HETGS). An iron emission line at-
tributed to the quasar at ∼ 6.43 keV (rest frame) is clearly resolved, with an
equivalent width of ∼ 100 eV. Although we cannot rule out contributions to the
line from a putative torus, the diskline model provides an acceptable fit to this
iron line. We also detect a weak emission feature at ∼ 6.9 keV (rest frame).
We suggest that both lines could originate in an accretion disk comprised of a
highly ionized optically thin atmosphere sitting atop a mostly neutral disk. We
search for absorption features from a warm/hot component of the intergalactic
medium along the ∼ 1.5h−1Gpc line of sight to the quasar. No absorption fea-
tures are detected at or above the 3σ level while a total of six O VI intervening
absorption systems have been detected with HST and FUSE. Based on the lack
of O VII and O VIII absorption lines and by assuming collisionally ionization, we
constrain the gas temperature of a typical O VI absorber to 105< T < 106K,
which is consistent with the results from hydrodynamic simulations of the inter-
galactic medium. The zeroth order image reveals the extended emission from the
surrounding cluster. We have been able to separate the moderate CCD X-ray
spectrum of the surrounding cluster from the central quasar and find that this is
a hot cluster with a temperature of ∼ 10 keV and a metal abundance of ∼ 0.3Z⊙.
We also independently obtain the redshift of the cluster, which is consistent with
the optical results. We estimate that the cluster makes negligible contributions
to the 6.9 keV iron K line flux.

Page 2

– 2 –
Subject headings: intergalactic medium — quasars: absorption lines — quasars:
individual (H 1821+643) — quasars: emission lines — cosmology: observations
— X-rays: galaxies:clusters
1.Introduction
H 1821+643 is one of the most luminous quasars (mv= 14.1) at low redshift (z = 0.297).
It was discovered as a serendipitous X-ray source detected with the Einstein Observatory
(Pravdo & Marshall 1984) and has been studied extensively (e.g., Kolman et al. 1991; Blun-
dell & Lacy 1995; Saxton et al. 1997). It has been classified as a radio-quiet quasar ac-
cording to its radio luminosity and nuclear [O III] luminosity (Lacy, Rawlings, & Hill 1992).
H 1821+643 is located in a cluster of galaxies with Abell richness class ≥ 2 (Schneider et al.
1992).
Previous observations have shown an iron Kα emission line in the X-ray spectrum of
H 1821+643. Ginga (Kii et al. 1991; Kolman et al. 1991) observed an iron line at 6.6±0.3 keV
(rest frame of the quasar). The subsequent ASCA observation improved the measurement,
and gave an observed line center energy of 5.07 ± 0.04 keV, or 6.58 ± 0.05 keV in the
quasar rest frame (Yamashita et al. 1997), and a line width of ∼ 160 eV. The authors
attributed this to a broadened fluorescent iron line from a highly-ionized accretion disk. A
particular problem with these observations is that although extened X-ray emission from the
cluster that surrounds H 1821+643 has been detected with ROSAT (Saxton et al. 1997; Hall,
Ellingson, & Green 1997), none of these observations has been able to obtain the spectrum of
the cluster and therefore separate the contribution to the quasar iron line from the cluster. In
a ROSAT PSPC observation Saxton et al. (1997) concluded that the cluster could contribute
10 − 70% to the observed Fe emission line with Ginga. Yamashita et al. (1997) placed an
upper limit to the cluster contribution at < 7% by assuming a metal abundance of < 0.4Z⊙
and a thermal temperature of kT < 10 keV.
In this paper we present a high-resolution X-ray spectrum of H 1821+643 obtained
with the Chandra High Energy Transimission Grating Spectrometer (HETGS). We clearly
resolve an emission line from neutral or low-ionized iron. Using the high spatial resolution
of Chandra we have been able to resolve the extended emission from the surrounding cluster
and for the first time obtain its X-ray spectrum. This enables us to determine the cluster
contribution to the central quasar and make a more accurate line diagnostic.
Another motivation of this observation is to detect possible X-ray absorption features
introduced by the warm/hot intergalactic medium in the local universe. At a moderate

Page 3

– 3 –
redshift, the sight line toward H 1821+643 traverses a distance of nearly 1.5h−1Gpc (we
use a Hubble constant of H0 = 100h km s−1Mpc−1and h = 0.5). Numerous absorption
systems (such as H, C, N, O, Si, etc.) have been discovered in this sight line in optical and
UV bands (e.g., Tripp, Lu, & Savage 1998). Especially in recent observations, the Hubble
Space Telescope (HST) and the Far Ultraviolet Spectroscopic Explorer (FUSE) have detected
a number of O VI absorption lines (see, e.g., Tripp & Savage 2000; Oegerle et al. 2000)
which are not clearly associated with any galactic system. Numerical simulations (Cen &
Ostriker 1999a; Dav´ e et al. 2001; Cen et al. 2001; Fang & Bryan 2001) predict that at least
some of this may be in the form of a moderately warm/hot (∼ 105< T < 107K) component
of the intergalactic medium (IGM) and the baryonic gas contained in the O VI lines is about
20 − 30% of this warm/hot IGM (WHIM). The remaining 70 − 80% WHIM gas is in the
form of hot gas which may be detected by O VII and O VIII absorption lines in the X-ray
band. H 1821+643 is one of the best candidates to observe for this purpose, given that it
is one of the brightest X-ray quasars and has a relatvely simple spectral shape in the soft
X-ray band.
This paper is arranged as follows: §2 is data analysis, §3 describes the continuum
spectrum of H 1821+643, §4 is devoted to emission line diagnostics, in §5 we discuss the
X-ray constraints on the O VI absorbers, and study an absorption system at z ≈ 0.1214. We
discuss the emission from the surrounding cluster in §6. The last section is the summary.
2.Data Analysis
H 1821+643 was observed with the Chandra High Energy Transmission Grating Spec-
trometer (HETGS; Canizares et al. 2001) on 2001 February 9. The total exposure time was
100 ksec. The HETGS produced a zeroth order image at the aim-point on the focal plane
detector, the ACIS-S array, with higher order spectra dispersed to either side (for ACIS-S,
see Garmire et al. 2001). The telescope pointing direction was offset 20
to move the zeroth order off a detector node boundary. The Science Instrument Module
(SIM) was moved toward the read-out row by about 3 mm to get better ACIS energy res-
olution (for detailed instrument setups, see the Chandra Proposers’ Observatory Guide, or
POG1). Figure 1 displays the zeroth-order image in a 2
the image for visual clarity. The zeroth-order count rate within a circle of 10
0.23 ± 0.03 counts s−1(error is quoted at 1σ level) and its light curve shows no significant
variation during the observation. This count rate gives a pileup fraction of ∼ 20% (see POG)
′′along +Y in order
′× 2
′region. We have smoothed
′′radius is
1See Chandra Proposers’ Observatory Guild (POG) at http://asc.harvard.edu.

Page 4

– 4 –
and prevents us from analyzing the zeroth-order spectrum of the nucleus. We also label the
position of the star K1-16. The central star of the planetary nebula K1-16 was detected with
the ROSAT PSPC at about 1.
Although this star would only significantly contribute to the ultrasoft emission between 0.07
and 0.17 keV, a roll angle of 46◦was applied to keep the spectra of the two sources from
overlapping.
′3 north and 0.
′5 west of H 1821+643 (Kolman et al. 1991).
Spectral extractions and reductions were performed with the standard pipeline for the
Chandra HETGS provided by the Chandra X-ray Center (CXC)2. We used a combination of
Chandra Interactive Analysis of Observations (CIAO) V2.1 and custom routines in IDL. The
standard screening criteria were applied to the data. We selected photon events with ASCA
grades 0, 2, 3, 4, 6 and excluded those with energies above 10 keV. The HETGS consists of
two different grating assemblies, the High Energy Grating (HEG) and the Medium Energy
Grating (MEG), and provides nearly constant spectral resolution (∆λ = 0.012˚ A for HEG
and ∆λ = 0.023˚ A for MEG) through the entire bandpass (HEG: 0.8-10 keV, MEG: 0.4-8
keV). Extraction windows of 10
applied for HEG and MEG, respectively. The moderate energy resolution of the ACIS-S is
used to separate the overlapping orders of the dispersed spectrum. We added the plus and
minus sides to obtain the first order spectra of both grating assemblies (see Figure 2). The
effective area (ARF) for this observation was obtained using the CIAO tools.
′′.5 and 7
′′.8 widths in the cross-dispersion direction were
3. Continuum
The spectrum in Figure 2 shows a smooth continuum with an emission feature at ∼ 5
keV. In general, we find that a single power law with photoelectric absorption fits the entire
bandpass of both MEG and HEG spectra very well. The MEG and HEG spectra were
binned to a constant resolution of 0.02˚ A. We initially performed a simultaneous fit for HEG
(0.9—8 keV) and MEG (0.6—8 keV) with a single power law plus photoelectric absorption
at z = 0. We ignored the MEG spectrum below 0.6 keV due to ∼ 30 − 50% systematic
uncertainties of spectral fluxes at the oxygen K edge region. The fitting parameters are
shown in Table 1. In two fits, we fixed the absorption column density at the Galactic value
(NG
the ratio (data divided by model). There is some weak evidence for excess emission below
1 keV. When we allow the absorption column density to vary, the best-fit column density
NH is lower than 1.9 × 1019cm−2at 90% confidence level, so we infer that there is excess
H= 3.8 × 1020cm−2; Lockman & Savage 1995). Figure 3 shows the best fit spectra and
2See http://asc.harvard.edu/.

Page 5

– 5 –
emission in the soft X-ray band. The free-NH model is a better fit to the data because
the C-statistic drops significantly (∆C = 82) while the number of the degree of freedom
drops imperceptibily. However, the soft excess might also result from residual calibration
uncertainties in the MEG effective area (∼ 30% for 0.5 < E < 0.8 keV).
Since the systematic uncertainties of the HETGS spectral fluxes are lower above 2
keV (less than 10%) and an accurate measurement of the continuum level is important in
identifying the emission feature, we fit instead the HEG and MEG spectra between 2 and
7 keV. Figure 2 shows that a single power law fits both MEG and HEG spectra very well
in this band. Since this energy band is largely insensitive to NH, we fix the column density
at the Galactic value. The best-fit photon index (Γ = 1.761+0.047
previous result (Γ = 1.75 ± 0.03) reported by ASCA (Yamashita et al. 1997) at better than
the 1σ level. It is also consistent with indices of other radio-quiet quasars (Reeves & Turner
2000). The flux (2—10 keV, observer frame) and luminosity (2—10 keV, quasar frame) are
∼ 1.2 × 10−11ergs cm−2s−1and ∼ 5 × 1045ergs s−1, respectively. These values are also
consistent with those reported from ASCA (Yamashita et al. 1997), EXOSAT (Warwick,
Barstow, & Yaqoob 1989), Ginga (Kii et al. 1991; Kolman et al. 1991) and BBXRT (Yaqoob
et al. 1993). ASCA observation reported a flux of ∼ 1.75 × 10−11ergs cm−2s−1and a
luminosity of ∼ 7 × 1045ergs s−1in the corresponding energy bands.
−0.052)3is consistent with the
4.Emission Lines
4.1. Spectral Fitting
Figure 3 shows that the Emission feature at ∼ 5 keV can be well fitted with the Gaussian
parameters detailed in Table 2. We also fit the HEG and MEG spectra separately to check
for consistency, and we list the results from fitting the HEG data only in Table 2. Given
that the respective HEG and MEG resolutions are ∼ 40 eV and 80 eV FWHM (full width
at half-maximum) at the observed line energy (POG), the line is clearly resolved by both
grating assemblies. Table 2 shows that the Gaussian fit significantly improves the C-statistic
by ∆C = 40.2, indicating a detection significance of nearly 6σ for three degrees of freedom
(namely, Gaussian line flux, line width and the center energy). To investigate whether or
not the line is the neutral Kα line, we also fit the spectra by fixing the line center energy at
6.4 keV. The C-statistic is higher by 1.28 with one less degree of freedom, which means a
neutral iron Kα line can be ruled out at 68% level but not at 90% level. Figure 4 shows the
3Unless mentioned, all the errors are quoted at 90% confidence level.

Page 6

– 6 –
best fit and 68%, 90% and 99% joint confidence regions for the line center energy and the
line flux.
The iron Kα line feature in H 1821+643 was measured with Ginga observations (Kii
et al. 1991) at 6.6 ± 0.3 keV (rest frame of the quasar) and with ASCA at 6.58 ± 0.05 keV
(Yamashita et al. 1997). Chandra observations give a rest-frame line energy of 6.435±0.041
keV (6.435+0.197
was excluded at 90% confidence but is within 3σ of the Chandra observation.
−0.106keV at 3σ level). This shows that the line energy from the ASCA observation
Even after the addition of this iron Kα line, there are still residuals present at energies
> 6.4 keV (quasar frame) in both HEG and MEG spectra. We add a second Gaussian line
component, with the energy and the line width as free parameters. The spectral fit results
for this second component are also listed in Table 2. The C-statistic decreases by ∆C = 9.7,
which corresponds to a detection significance of 2.3σ. Fitting with the HEG data only fails
to constrain the line width; however, the simultaneous spectral fit of HEG and MEG gives
a line width of ∼ 61 eV.
We have used the extracted cluster profile and spectrum to estimate the amount of flux
that the cluster could have contributed to both emission lines in the grating spectra (see
§6). We conclude that the hot cluster should have no contribution to the neutral iron line at
∼ 6.4 keV (rest frame). Using the profile derived in §6 we find that the cluster can contribute
at most 3% flux to the 6.9 keV line.
4.2. The 6.4 KeV Line
The observed line energy of ∼ 6.4 keV implies that the fluorescent iron line is from
neutral and/or low ionization states of iron, although the 90% confidence errors cannot rule
out contribution from ionization states up to Fe XIX (see, e.g., Mewe, Gronenschild, & van
den Oord 1985). The line center energy is consistent with the energy of Fe III Kα1(6.434
keV) or Fe IV Kα2(6.435 keV) (see, e.g., Kaastra & Mewe 1993); however, we cannot rule
out Fe I Kα emission from an outflow with a velocity of v ∼ 2000 km s−1.
Such a fluorescent iron line implies the existence of cold (T < 106K) reprocessing
material (see, e.g., Fabian et al. 2000) illuminated by X-ray continuum. An incident X-ray
photon is either Compton scattered by electrons in the cold gas, or photoelectrically absorbed
followed by fluorescent line emission or Auger de-excitation. This photoelectric absorption
would produce iron K edges at relatively higher energies. Iron K edges were not detected in
the HEG and MEG spectra of H 1821+643. In order to estimate the significance of the iron
K edge, we include a multiplicative edge model to the power law plus the Gaussian model.

Page 7

– 7 –
The edge energy is fixed at 7.1 keV (quasar frame) because the K edges of most neutral and
low ionization states of iron (up to Fe V) are around this energy (Verner & Yakovlev 1995).
The 90% upper limit to the optical depth of the edge is τ ∼ 6.2 × 10−2, which corresponds
to a 90% upper limit of the hydrogen equivalent column density at ∼ 6 × 1022cm−2. Here
we assume solar abundance (Anders & Grevesse 1989) and use the photoionization cross
sections from Verner & Yakovlev (1995).
Several models have been proposed to explain the origin of the reprocessing material
responsible for the Fe line. Three possible scenarios are that the iron line is reflected (1) from
material in the disk close to the black hole which is subject to relativistic and transverse
Doppler effects (diskline model, Fabian et al. 1989), (2) from a molecular torus (Antonucci
1993), (3) or from the broad line region (BLR) clouds (Yaqoob et al. 2001). We find that
the diskline model gives an acceptable fit to the iron emission line. However, we cannot
rule out the torus model if the reflection material has a small Thomson optical depth. A
BLR origin for the Fe line requires a strong flux variation on timescales of ∼ months. Here
we discuss each model.
diskline model: The diskline model provides acceptable fits to both HEG and MEG
data. Thus far, this has been the most commonly used model for explaining the broad Fe
Kα line observed in many Seyfert 1 galaxies (Nandra et al. 1997) and quasars (Reeves &
Turner 2000). In this model, X-rays are reprocessed in a cold or highly ionized accretion disk
rotating around a Schwarzschild black hole with a characteristic radius of rg= GM/c2(see,
e.g., Fabian et al. 1989), where M is the mass of the central black hole. The intrinsically
narrow line (line width < 1 eV) is broadened by the relativistic motion of the disk ranging
from a few hundred eV to ∼ keV (see Fabian et al. 2000 for a review). We fix the inner and
outer disk radii at 6rgand 1000rgrespectively, where 6rgis the last stable orbit radius of
a Schwarzschild black hole with mass M. The radial power law index of the emissivity is
fixed at α = −2. Three parameters, the line centroid energy, the line intensity and the disk
inclination angle (θ), are allowed to vary. The spectral fit results are shown in Table 2. The C-
statistic of the diskline model is comparable to that of a simple Gaussian fit, implying both
models are acceptable. The line centroid energy is ∼ 6.49 keV, comparable to within the 90%
confidence error of the best-fit Gaussian value reported earlier. The diskline model predicts
a skewed, double-horned line profile in which the blueward side drops sharply. However, our
Chandra observation is not able to detect this level of structure.
Iron K emission line seen with ASCA shows an EW of 170±50 eV at 6.58 keV (Yamashita
et al. 1997). One of their explanations for the line strength and position is that the line is
reprocessed from an accretion disk in which iron is ionized up to the He-like states (∼
6.68 keV) in the innermost part, and redshifted to the observed 6.58 keV energy by the

Page 8

– 8 –
strong gravitation of the central black hole. We suspect that the ASCA data was unable
to distinguish between the 6.4 and 6.9 keV lines, given its resolution (the ASCA resolution
at the Fe K energies is ∼ 160 eV FWHM, respectively 4 and 2 times worse when compared
with the HEG and MEG). The Chandra data clearly show an emission line from neutral or
low ionization states.
Torus: Recent Chandra HETGS and XMM-Newton observations have shown that nar-
row iron K emission at ∼ 6.4 keV is prevalent in many active galactic nuclei (AGNs) and
quasars (e.g., Reeves et al. 2001 for Mkn 205, Pounds et al. 2001 for Mkn 509 and Yaqoob
et al. 2001 for NGC 5548). A common phenomenon is that these AGNs and quasars showed
a broader Fe K emission in their ASCA spectra. For Mkn 205 and Mkn 509 the narrow
component is attributed to reflection from the putative torus based on the ∼ 50 − 80 eV
measured EWs. According to Krolik, Madau, & Zycki (1994), the the characteristic EW of
an iron Kα line from the torus could be ∼ 100 eV if the reflection material has a typical
Thomson optical depth τT= 0.5 − 1. The EW in H 1821+643 is ∼ 110 eV.
BLR clouds: The line width of the 6.4 keV line in H 1821+643 is σ = 106.9+58.3
eV, which corresponds to 1.171+0.639
a narrower line (σ ≈ 4525 km s−1FWHM) in NGC 5548 with the Chandra HETG. They
attributed the broadening of the Fe Kα line to the bulk velocity of the emitting gas, which
may be located in the broad line region. Based on their model, we calculate the line intensity
and EW of the 6.4 keV line in H 1821+643, using the upper limit of the hydrogen column
density from the iron K-edge constraint. We obtain the 90% upper limit of the line intensity
and EW of ∼ 5 × 10−6photons cm−2s−1and ∼ 25 eV, respectively. Both values fall short
of the observed values by a significant factor of ∼ 5. Yaqoob et al. (2001) encountered
the same problem and they attributed this to the possible variation of the continuum level
shortly before the Chandra observation. For H 1821+643, on a central black hole mass
of ∼ 109M⊙ (Kolman et al. 1991), the BLR regions which emit the iron K line should
be located at a distance of 50.1+69.7
reconcile the observed EW with that expected from the BLR, the X-ray flux of H 1821+643
would have to have varied by a factor of ∼ 5 in a timescale of ∼ 2 months. ASCA and
Chandra observations showed that H 1821+643 experienced only a ∼ 30% flux variation in
a timescale of a few years. So we consider a fivefold variation to be unlikely and conclude
that the emission from BLR is not the main source of the observed iron K line.
−37.8
−0.414× 104km s−1FWHM. Yaqoob et al. (2001) observed
−29.1light days from the central X-ray source. In order to

Page 9

– 9 –
4.3.The 6.9 keV Line
There are several possibilities for the origin of the Fe XXVI emission line detected at
6.9 keV. These include environments (e.g. narrow line regions, or hot accretion disk) which
are hot enough, or have high enough ionization to produce H-like Fe. However, it is of
interest to consider a scenario which can accommodate both the Fe Kα and Fe XXVI lines
seen in the Chandra HETGS spectrum for this source, since this could provide us with the
unique prospect for probing in detail the condition and structure of the accretion disk in
H 1821+643.
Irradiation by X-rays can photoionize the surface layers of the accretion disk (e.g. Ross
& Fabian 1993; Ross, Fabian, & Young 1999). The iron line(s) produced by the illuminated
matter and the associated reflection spectrum depend largely on the ionization parameter ξ
(e.g. Matt, Fabian, & Ross 1993, 1996, for the case of a constant density structure accretion
disk). There are also other dependences, e.g. the spectral index of the incident X-rays (the
maximum gas temperature is indirectly related to Γ), and the strength of the illumination
(characterized by the gravity parameter A, respectively for high and low illumination, A ?
0.1, and A ≫ 1). Recent calculations of ionized reflection spectra which include hydrostatic
equilibrium effects have been presented by Nayakshin, Kazanas, & Kallman (2000) and
Ballantyne, Iwasawa, & Fabian (2001).
In the context of these models, one may envisage the accretion disk of H 1821+643 to be
comprised of a highly ionized optically thin atmosphere sitting atop a mostly neutral disk.
The Thomson depth of the upper Compton heated layer should be such (e.g. τT? 1) that
most of the X-rays propagating through the disk can reach the bottom colder layers to pro-
duce the observed cold reflection spectrum, while also showing ionized features. (Nayakshin,
Kazanas, & Kallman 2000 and Ballantyne, Iwasawa, & Fabian 2001 report narrow transi-
tions from hot to cold material in the illuminated layers.) It is curious that a He-like Fe line
at ∼ 6.7 keV is not seen. However, it is possible that the upper layer has an ionization state
to support only H-like Fe.
5. Absorption
5.1. Constraints on Absorption Column Densities
H 1821+643 is one of the brightest X-ray quasars at low redshift, therefore it provides
a unique opportunity to probe the so-called “missing baryons” along this particular line-of-
sight (see Fukugita, Hogan & Peebles 1998 and Cen & Ostriker 1999a for discussion about

Page 10

– 10 –
the missing baryons problem). The path length to H1821+643 is almost 1h−1Gpc. The low
Galactic hydrogen column density (∼ 4 × 1020cm−2) and relatively simple spectral shape
reduce the spectral complexity typically seen in Seyfert galaxies and other active galaxy
nuclei, making it easy to identify any intervening absorption. Furthermore, recent observa-
tions with HST and FUSE indicate that there are at least six intervening O VI absorption
systems (Oegerle et al. 2000; see Table 3). Highly ionized ions, such as O VI, are important
ion species in diagnosing the physical conditions of the intergalactic medium (IGM). The
production of O VI ions requires photons or electrons with energy > 114 eV, so O VI can
be produced by photoionization from the background radiation, or collision ionization in a
hot gas with temperatures above 105K. However, current observations of O VI absorption
lines cannot definitively constrain the ionization mechanism (Tripp & Savage 2000). X-ray
absorption lines from He-like or H-like ions, such as O VII (0.574 keV) and O VIII (0.654
keV) can give direct evidence of a hot, collisionally ionized gas with temperature ranging
from 5 × 105K to 107K.
Identifying absorption features requires a careful measurement of the continuum. To
achieve this, we take the MEG data (HEG does not have enough effective area below 0.8
keV) and initially fit the 0.4—4 keV spectrum with a power law and the Galactic absorption.
We then fit the residual with a five-order polynomial to obtain an accurate characterization
of the continuum. The five-order polynomial will account for spectral features larger than
5˚ A (such as calibration uncertainties) but will preserve narrow line features. The data are
binned at 0.02˚ A and we calculated the χ (the signal-to-noise ratio) of each bin. Figure 5
shows the 16 − 28˚ A spectrum. The bottom panel of each plot gives χ, and the two dotted
lines within the χ plot correspond to ±3σ level.
An absorption line feature should at least have a signal-to-noise ratio of < −3σ level to
be identified. Figure 5 shows that each bin in 16—28˚ A has χ > −3, which indicates that
no absorption line is identified in this observation. Interestingly, we find an emission feature
at around 21˚ A with χ ∼ 3.5. There is no known emission feature in this wavelength in the
rest frame of the quasar. Assuming a Poisson distribution for each bin, the probability for
observing one bin with χ ≥ 3.5 in one observation of H 1821+643 is ∼ 50%. So the line may
be the result of a statistical fluctuation.
Given the six intervening O VI absorption systems in the UV spectra of H 1821+643
and assuming that X-ray absorption ions coexist with O VI, we calculate the upper limits of
O VII and O VIII column densities, and place constraints on the physical conditions of the
absorption systems. The 3σ upper limits of the O VII and O VIII column densities for all
six intervening systems are estimated based on the quasar spectral flux, the MEG resolution
and effective area. We also take into account the instrumental line response function (see

Page 11

– 11 –
Fang et al. 2001). The results are shown in Table 3. We take a constant velocity dispersion
of b = 100 km s−1. Hydrodynamic simulations (Fang, Bryan & Canizares 2001) predict that
the the mean O VII and O VIII Doppler b-parameters are around ∼ 50 km s−1and the
turbulence in the got gas can easily increase b-parameter to ∼ 100 km s−1. Since the MEG
has very little collection area below 0.4 keV, we estimate the upper limit of O VII column
density only with the z = 0.1214 system. The O VI column densities were adopted from
various references (Tripp, Savage, & Jenkins 2000; Oegerle et al. 2000; Savage, Tripp, & Lu
1998). For those observations which had column densities of the O VI λλ 1031.93˚ A,1037.62˚ A
doublet, we adopt the average of the two values.
5.2.A Sample Study: z ≈ 0.1214 System
Our limits on O VII and O VIII absorption constrains the physical condition of the
O VI absorber if it is collisionally ionized, as recent hydrodynamic simulations predict (Fang
& Bryan 2001; Cen et al. 2001), but not if it is photoionized. We take the z ≈ 0.1214
absorption system as an example. The O VI absorption (z = 0.12137; EW = 98 ± 21 m˚ A)
was detected by (Oegerle et al. 2000) with FUSE. They also detected an H I Lyβ absorption
line at z = 0.12129, with an EW of 137±23 m˚ A and a column density of 3.2±1.6×1014cm−2.
Collisionally Ionized Absorber:
produced in hot gas by collisional ionization, the ratio between O VI and O VII or O VIII
column densities can place tight constraints on the gas temperature. Figure 6 displays
log(NOV I/NOV II) and log(NOV I/NOV III). The atomic data was adopted from Mazzotta et
al. (1998). The 3σ upper limits on O VII or O VIII column densities are 8.81 × 1015cm−2
and 2.27×1016cm−2respectively, which give a lower limit of log(NOV I/NOV II) > −2.27 and
log(NOV I/NOV III) > −2.36. Figure 6 shows that the gas temperature must be lower than
106K to satisfy the O VI-O VII ratio and lower than 2 × 106K to satisfy the O VI-O VIII
ratio. This gives an upper limit of the temperature of the z ≈ 0.1214 absorption system at
T < 106K. We can safely place a lower limit on the temperature of T > 105K according to
the collisional ionization fraction of O VI (Mazzotta et al. 1998).
We find that if the O VI absorption lines are
The temperature range, 105< T < 106K, is in accord with cosmological simulations
that predict most of these O VI absorbers are located in the filaments that connect the viri-
alized regions and have been shock-heated to temperatures about 105K. Recently large-scale
cosmological hydrodynamic simulations have shown that a large amount of the intergalac-
tic medium (30% ∼ 50%) lies in regions with overdensity 5 < δ < 200 and temperature
105< T < 107K (Warm-hot intergalactic medium, or “WHIM”, see, e.g., Cen & Ostriker
1999a; Dav´ e et al. 2001; Fang, Bryan & Canizares 2001). Here δ = ρb/?ρb?−1 and ?ρb? is the

Page 12

– 12 –
mean baryon density of the universe. In this temperature range, most of the gas should be
collisionally ionized. Fang & Bryan (2001) and (Cen et al. 2001) showed that by assuming
collisional ionization in WHIM gas, hydrodynamic simulation can reasonably reproduce the
observed distribution of O VI absorption lines. Furthermore, Fang & Bryan (2001) and
Cen et al. (2001) showed that collisional ionization dominates for O VI absorption lines with
EW > 35 − 40 m˚ A.
Photoionized Absorber: If the O VI is produced by photoionization in a low den-
sity gas, corresponding column densities of O VII and O VIII are too low to be detected
with any current X-ray telescope.We used CLOUDY (Ferland 2001) to calculate the
ionization structure in a slab of gas illuminated by a background radiation from Haardt
& Madau (1996), in which we adopt a mean specific intensity at the Lyman limit of
Jν = 2 × 10−23ergs s−1Hz−1sr−1. Three metallicity models are considered here: two with
a constant metallicity (Z = 0.1Z⊙and 1Z⊙, where Z⊙is the solar abundance) and one in
which metallicity is a function of baryon density. In the last model, we adopt the numerical
simulation data from Cen & Ostriker (1999b). The CLOUDY calculation stops when the
neutral hydrogen column density reaches logNHI= 14.51 (Oegerle et al. 2000). To obtain
the O VI, O VII and O VIII column densities under different physical conditions we adjust
the ionization parameter U = nγ/nH, where nγ is the density of H I ionizing photons and
nHis the hydrogen number density.
We find that none of the three models can give detectable O VII and O VIII column
densities in the X-ray band. The results are shown in Figure 7. The top, middle and
bottom panels stand for models with Z = 0.1Z⊙, Z = 1Z⊙and Z = Z(ρb) (ρbis the baryon
density, see Cen & Ostriker 1999b), respectively. For each model we calculate the O VI,
O VII and O VIII column densities. The observed O VI column density is shown as the
horizontal dashed line in each panel and the vertical dashed line in each panel indicates the
corresponding ionization parameters (or hydrogen number density). These vertical lines also
give corresponding O VII and O VIII column densities. None of these O VII and O VIII
column densities can be detected with current and planned X-ray telescopes (for X-ray
telescope detection limits, see Fang & Canizares 2000). For instance, in the Z = 0.1Z⊙model,
the expected O VII and O VIII column densities are only ∼ 1014cm−2and < 3.2×1012cm−2,
respectively. Given such a low density system at z ≈ 0.1214, only when its metallicity reaches
solar abundance could O VII absorption line be detected with current X-ray instruments
(When −5.4 < log(nH) < −6.4; see Figure 7). Recent hydrodynamic simulations showed
that the typical metallicity of the gas that produces O VI absorption lines is 0.1−0.3 Z⊙(Cen
et al. 2001). So we conclude that any X-ray absorption lines that are discovered by current
telescopes are unlikely to come from a low-density gas photoionized by the background
radiation.

Page 13

– 13 –
6.The Surrounding Cluster
Optical studies revealed that H 1821+643 lies in a rich cluster of galaxies (Schneider et
al. 1992; Lacy, Rawlings, & Hill 1992). Optical spectroscopy of six member galaxies showed
that the redshift of the cluster, 0.299 ± 0.002, is consistent with the redshift of the quasar
(Schneider et al. 1992). They also determined the velocity dispersion of the cluster to be
1050±320 km s−1which indicates that this cluster is quite massive. This is consistent with
the cluster richness, R>2, determined by Lacy, Rawlings, & Hill (1992).
Figure 1 shows the zeroth order image, and from the contours it appears that there is
extended emission around H 1821+643. To quantify this we first corrected the image for
the known detector irregularities and for telescope vignetting. We then extracted a profile
in radial bins with each bin being 1
along with the model fits. The initial model was a combination of the Chandra point spread
function (PSF) and a β-model (Cavaliere & Fusco-Femiano 1978) for the extended emission
from the cluster. This proved to be a rather poor fit to the X-ray profile, leaving excess
emission between 5 and 15
extra emission and found an acceptable fit to the profile. Formally, the reduced χ2is 2.51
(343.6/137 dof) but the largest contribution is from the quasar where a modest amount of
instrumental pileup is not well modeled by the Chandra PSF. Evaluating the model between
5
′′wide. The profile from 2
′′to 160
′′is shown in Figure 8
′′. We then added a Gaussian component to account for this
′′and 160
′′results in χ2
ν= 1.71.
The parameters for the cluster profile are similar to those found for other clusters, a core
radius of 17.6+0.17
−0.17
a σ value of 6.54+0.14
−0.13
similar to the scale for a giant elliptical galaxy.
′′(0.1 Mpc) and a β = 0.74+0.05
′′. At the distance of the cluster this corresponds to 37.3 kpc, which is
−0.03. The additional Gaussian component has
The ACIS-S3 spectral data for the cluster was extracted from 3
what we refer to as the global spectrum (Figure 9). By excluding a 3
the quasar we have eliminated essentially all the flux from a point source from our analysis
(the Chandra Proposers’ Observatory Guide, Figure 6.3). We used the XSPEC mekal
model with Galactic absoption to fit the cluster spectrum and the fit parameters are given in
Table 4. In addition to the temperature and abundance we also allowed the redshift to be a
free parameter in fitting the global spectrum. From the X-ray data we find that the redshift
of the hot gas is 0.303 with a lower limit of 0.299 and an upper limit of 0.309 (90% confidence
limits) consistent with the redshift of the quasar and that of the optical galaxies. We also had
sufficient counts to extract the spectrum in radial rings and the parameters for these fits are
listed in Table 4. We find the overall temperature and metal abundance (kT = 10.8+1.0
and Z = 0.35 ± 0.08 Z⊙) are consistent with those of a typical hot cluster. The luminosity
′′to 100
′′radius circle around
′′and this is
−0.9keV

Page 14

– 14 –
between 2 and 10 keV (quasar frame) is 2.54 × 1045ergs s−1, which is roughly consistent
with temperature-luminosity relationship obtained with ASCA (Horner et al. 2000). We also
find that the temperature is cooler in inner regions between 3 and 11.5
temperature is not consistent with that of a typical elliptical galaxy (Davis & White 1996).
′′; however, the cooler
We estimate that the cluster makes negligible contribution to the iron 6.9 keV emission
line detected in the grating spectrum. The strongest emission line from a hot cluster plasma
in that line region is Fe XXVI Kα line. From the 3 to 11
Kα line flux is ∼ 5.2 × 10−7photons cm−2s−1. We also estimate that for the part of the
cluster hidden by the quasar the fraction of the cluster flux should be at most 6.56% from
the surface brightness profile (Figure 8). By comparing the line flux from the grating spectra
in Table 2, we find that the cluster can contribute at most ∼ 3% to the 6.9 keV line flux.
′′.5 region, we obtain the Fe XXVI
7.Summary
In this paper we study the spectral features of a low-redshift quasar H 1821+643 with
Chandra HETGS. Our main conclusions can be summarized as follows:
1. An emission feature at ∼ 5 kev was detected at nearly 6σ level and has been identified
as a fluorescent iron line. The rest-frame line-center energy (6.435±0.041 keV) implies
this line is from neutral or low ionization states of iron, although we cannot rule out
contribution from ionization states up to Fe XIX. An emission feature at around 6.9
keV (rest frame of the quasar) was marginally detected at 2.3σ level and is identified
as a possible H-like Fe emission line.
2. The diskline model provides an acceptable fit to the 6.4 keV iron line; however, a
putative torus could also make contributions to the EW of the emission line. The
model in which X-rays are reflected from the BLR clouds requires the X-ray flux from
H 1821+643 to vary at a significant level (a factor of at least 5) on a timescale of ∼ 2
months, which seems unlikely. In order to accommodate both the Fe Kα and Fe XXVI
lines, we suggest that both lines could originate in an accretion disk comprised of a
highly ionized optically thin atmosphere sitting atop a mostly neutral disk.
3. No absorption features were detected at or above the 3σ level. A total of six O VI
intervening absorption systems have been detected with HST and FUSE. We place
3σ upper limits on O VII and O VIII column densities at the corresponding redshifts,
which have typical values of ∼ 1016cm−2.

Page 15

– 15 –
4. We focus on the z ≈ 0.1214 absorption system and constrain its physical conditions
by combining UV and X-ray observations. Assuming collisional ionization, we con-
strain the gas temperature at 105< T < 106K, which is consistent the results from
hydrodynamic simulations. However, if the O VI absorbers are photo-ionized by the
background radiation, no detectable O VII or O VIII lines are expected.
5. We obtain the surface brightness profile and spectra of the cluster that surrounds
H 1821+643. The X-ray spectra reveal that this is a typical hot cluster with a tem-
perature of kT ∼ 10 keV and a metal abundance of ∼ 0.3Z⊙. We also obtain the
redshift of the cluster, which is consistent with results from optical measurements. We
estimate that the cluster makes negligible contributions to the 6.9 keV iron K line flux.
An additional Gaussian component is found in fitting the surface brightness profile.
This component has a width of that similar to the scale for a giant elliptical galaxy.
TF thanks the MIT/CXC team for its support. This work is supported in part by con-
tracts NAS 8-38249 and SAO SV1-61010. Support for GLB was provided by NASA through
Hubble Fellowship grant HF-01104.01-98A from the Space Telescope Science Institute, which
is operated by the Association of Universities for Research in Astronomy, Inc., under NASA
contract NAS 6-26555.