Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/QEDT/Papers/0014lett_revised.ps Äàòà èçìåíåíèÿ: Fri Aug 26 22:24:21 1994 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 01:51:10 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: m 97

AN ASCA GIS SPECTRUM OF S5 0014+813 AT Z=3.384
Martin Elvis 1 , M. Matsuoka 2 , A. Siemiginowska 1 , F. Fiore 1;3 ,
T. Mihara 2 and W. Brinkmann 4
1 Harvard­Smithsonian Center for Astrophysics
60 Garden Street Cambridge MA 02138
e­mail: elvis@cfa.harvard.edu
2 RIKEN, the Institute of Physical and Chemical Research
Wako, Saitama, Japan
3 Osservatorio Astronomico di Roma
via dell'Osservatorio 5, Monteporzio­Catone (RM), I00040 Italy
4 Max Planck Institut f¨ur Extraterrestrische Physik
D­85740 Garching b. M¨unchen, Germany
Ap.J. Letters ASCA Special Issue, November 1994, in press

Abstract
ASCA has detected the z=3.384 quasar S5 0014+813 up to energies of 34
keV in the quasar rest frame using the two GIS instruments. The combined X­ray
spectrum has a signal­to­noise of over 50 oe and is consistent with a single power­law
of energy slope 0.63\Sigma0.03 over the 0.8­8 keV (observed) energy range. The spectrum
is also well fit with a simple thermal bremsstrahlung model of kT=40\Sigma4 keV (in the
quasar frame), which raises obvious possibilities for contributions to the diffuse X­ray
background.
A maximum solid angle
of\Omega d =2ú=0.4 (90% confidence) can be placed on the
strength of a Compton reflection component above the energy of the Fe­K edge. The
Fe­K 6.4 keV fluorescence line has a rest frame equivalent width !120 eV (90%
confidence) at its redshifted energy of 1.46 keV. The weakness of these features
differentiates this high luminosity, high redshift quasar from the majority of Seyfert
galaxies using its X­ray spectrum alone. The GIS slope is consistent with the slope
derived by the ROSAT PSPC. The normalization at 1 keV in the ASCA observation
is however a factor 30 % to 40 % higher than in the ROSAT observation, suggesting
a significant increase in the 1 keV (observed) flux over the 31.5 months between the
two observations (7.2 months, rest frame).
Subject Headings: galaxies: active, quasars: general, quasars: individual:
S5 0014+813, diffuse radiation.
1

1. Introduction
X­ray observations of AGN are heavily biased toward X­ray loud objects of
low redshift and moderate luminosity (Elvis 1991). Moreover these X­ray spectra
tend to be taken at low energies, less than ¸3 keV for Einstein (Shastri et al 1993)
and ROSAT spectra (Walter & Fink 1993, Fiore et al 1994, Laor et al 1994), and less
than ¸10 keV for Ginga spectra (Williams et al 1992, Lawson et al 1992). Naturally
we would like to know the form of AGN spectra at higher energies, and how the
many features seen in the low redshift, low luminosity AGN extrapolate to extreme
examples of the AGN population.
ROSAT PSPC spectra have allowed some determination of spectral slopes for
quasars at high redshift and luminosity (Elvis et al 1994a, Bechtold et al 1994) in the
1­10 keV emitted range. Unfortunately the common low energy cut­off discovered
with the PSPC and due possibly to absorption, leads to poorly determined slopes
(Elvis et al 1994a).
ASCA (Tanaka et al 1994) now provides the sensitivity to explore higher
energies for the brightest quasars at z¸3. The ASCA Gas Imaging Spectrometer
(GIS, Makishima et al 1994) instruments, in particular, are able to determine spectral
slopes over the 3­8 keV and 8­30 keV emitted energy ranges to good accuracy. These
energy ranges are optimal for setting constraints on Fe­K fluorescence emission and
on Compton reflection components (Lightman & White 1988, Guilbert & Rees 1988)
which begin above 8 keV in the rest frame.
Accordingly we observed S5 0014+813 with ASCA. This is a z=3.384 quasar
(K¨uhr et al 1983), and is among the highest X­ray flux quasars known with z¸3.
Indeed it is the only z¸ 3 quasar to have a 2­10 keV spectrum reported by EXOSAT
(Lawson et al 1992).
2. ASCA GIS observations of S5 0014+813
The quasar was observed on 29 October 1993 for a total good exposure time of
31.5 ksec in both GIS2 and GIS3. We applied the standard conservative data filters: a
minimum elevation angle of 15 degrees above the Earth's limb, and a minimum cut­off
rigidity of 8 GeV/c. We also excluded data acquired in the first 60 seconds following
each passage through the SAA. The source was readily apparent in the center of the
GIS field of view and the source centroid is 1.5 arcmin from the optical position, an
offset typical of ASCA data at present. 2904 and 3244 counts were recorded within a
6 arcmin radius circle centered on the quasar in GIS2 and GIS3 respectively.
The background within the source extraction region was estimated in two ways:
(1) The backgrounds were taken from annuli of inner and outer radii between 10
arcmin and 17 arcmin.
2

(2) Background events were extracted from the same region as the source events in
blank sky background event files, provided by the ASCA GOF (the files being a
superposition of 15 blank sky fields observed during the ASCA PV phase (May 93 ­
Mar 94) with a total exposure time of 350 ksec), for the same values of cut­off rigidity
(? 8GeV/c) used for the source events.
The former background was significantly smaller than the latter (by a factor
¸ 0:75) as expected due to the vignetting of the XRT. In the following we present
the results obtained with the second background subtraction method. We point out
however that the shape of the spectra obtained with the two methods agree to within
¸ 3%. The total background is ¸ 40% of the source counts, leading to a total of
about 2200 net counts and a 40 oe spectrum of S5 0014+813 in each GIS (Table 1).
The ASCA point spread function (PSF) in the GIS puts ¸ 10% of the flux of a point
source with a spectrum similar to that of S5 0014+813 outside the 6 arcmin radius
used here. All reported fluxes are corrected for this. Fluxes are not corrected for
dead time effects. The highest energy at which the quasar is detected at the 3 sigma
level is 7.7 keV (bin width of 0.3 keV), which corresponds to 34 keV in the emitted
frame.
A complication is the presence of two sources in the ROSAT PSPC image
within 6 arcmin of the quasar. In the ROSAT band the count rate of these sources
represents 90% of that of the quasar. The brighter source (80% of S5 0014+813), five
arcminutes from the quasar, is probably identified with a V=8.8 K0 star (HD1165,
SAO 44, PSPC offset = 9 arcsec). The fainter source (10% of S5 0014+813), only
1.5 arcmin from the quasar is unidentified . It has a PSPC hardness ratio (R= H­
S/H+S = 0.76\Sigma0.36, where S is the 0.1­0.4 keV band and H is the 0.4­2.4 keV band)
similar to the quasar, suggesting that it is outside the substantial Galactic NH (14.4\Theta
10 20 atoms cm \Gamma2 , Stark et al 1989) in this direction. HD1165 is detected in the ASCA
SIS0 image of the field (Elvis et al 1994b, in preparation). Its count rate is about
10% of that of the quasar in the 0.8­2 keV band, and is only 2% in the 2­8 keV band.
We shall assume the contribution from these two sources is negligible, but this caveat
should be borne in mind.
3. Spectral Fits
In all spectral fits we used the response matrices ``g2v3 1 4c'' and ``g3v3 1 4c''
(1994 April 20) provided by the ASCA GOF. Note that, because of the small count
rates, statistical errors dominate over systematic errors from the two instrumental
responses. The results of all fits are given in Table 2.
A single power­law fit gave no significant evidence of excess absorption in
either GIS so we fixed NH to the Galactic value. The GIS are not sensitive to
column densities smaller than 10 21 atoms cm \Gamma2 in any case. For a power­law the two
instruments agree well (to 20% in normalization, 2% in slope), and both give good ü 2 ,
so we made a simultaneous joint fit (letting the normalization in the two instruments
3

to be free to vary). This gave an energy slope of 0.63\Sigma0.05, consistent with the
ROSAT PSPC slope of 0.8\Sigma0.2 (assuming Galactic NH ). Adding the PSPC spectra
simultaneously gives the same results. Allowing an extra absorption component at
the quasar redshift gives a 90% upper limit to the intrinsic absorption is 2.6\Theta 10 22
atoms cm \Gamma2 . Figures 1 and 2 show the fit and residuals to all three data sets.
A thermal bremsstrahlung spectrum fitted the joint GIS data just as well as a
power­law. Such a fit is constrained on both sides and gives a rest frame temperature
of 39\Sigma4 keV. This is a temperature of some interest.
More complex spectra are not required. To investigate the presence of
curvature we fitted a power­law to the PSPC­GIS data above and below 2 keV (8.8
keV in the quasar frame) separately. No slope change is seen greater than 0.2 (1oe).
A complete Compton reflection model fit to the GIS data using a cold flat disk for
the reflector gave a 90% upper limit (for two interesting parameters)
of\Omega d /2ú=0.4 to
the solid angle subtended by the disk. (Adding the PSPC data tightened this limit
slightly to 0.3.) None of the fits is improved by an Fe­K fluorescence line at 6.4 or 6.7
keV (emitted, 1.46 keV, 1.53 keV observed respectively, \Deltaü 2 !1). A 90% upper limit
to the equivalent width for an intrinsically narrow line at 1.46 keV is 27 eV (observed
frame, 120 eV emitted). The 90% limit for a narrow line at 1.53 keV is EW=31 eV
(EW=140 eV, quasar frame).
The mean GIS normalization for a power­law fit, when corrected for the PSF
loss, is a factor 1.4 above that of the ROSAT data and suggests variability in the 31.5
months between the two observations (7.2 months in the quasar frame). If this light
travel time radius corresponds to 10 Schwarzschild radii, then M! 4 \Theta 10 11 M fi . The
EXOSAT 1 keV normalization is close to the PSPC value and the EXOSAT slope
(0.9\Sigma0.4) is fully consistent with both the ASCA GIS and PSPC values (see Bechtold
et al 1994).
The flux of S5 0014+813 in the 1­8 keV observed band is 3:2 \Theta 10 \Gamma12 ergs s \Gamma1
cm \Gamma2 , which corresponds to a 4­30 keV luminosity of 1:1 \Theta 10 48 ergs s \Gamma1 , about 2.5
times greater than the inferred ROSAT rest frame luminosity of 4.4\Theta10 47 ergs s \Gamma1 ,
(the GIS 2­10 keV luminosity is 6:2 \Theta 10 47 ergs s \Gamma1 ) for H 0 =50 km s \Gamma1 Mpc \Gamma1 , and
\Omega\Gammand The GIS X­ray luminosity requires a central black hole mass ¸10 10 M fi in order
not to violate the Eddington limit, which does not conflict with the variability limit.
If the bolometric luminosity could be used then the two limits would become very
close.
4

4. Discussion and Conclusions
We have detected a z=3.384 quasar up to 34 keV in the quasar frame with the
ASCA GIS. With a S/N of ¸ 50oe the GIS measurements of high redshift quasars
currently give us, paradoxically, our best measurements of quasar spectra above
10keV 1 and approach the quality of spectra for bright Seyfert galaxies from Ginga.
A thermal bremsstrahlung fit gives a temperature of 39\Sigma4 keV (quasar frame).
Similar temperatures were reported to two other high z quasars by Serlemitsos et al
(1994), unfortunately without quoted uncertainties. The closeness of these three
temperatures to the X­ray background temperature (40\Sigma5keV, Marshall et al 1980)
is remarkable. Superficially it suggests that AGN produce the X­ray background.
Radio­loud, z=3 quasars such as those observed with ASCA, cannot themselves
produce the XRB, being too rare and redshifted down to kT¸10 keV. However OSSE
also give a temperature of ¸45 keV (Johnson et al 1994) for low z AGN. That AGN
from opposite ends of the luminosity, redshift and radio­loudness scales give the same
temperature suggests a remarkably uniform process at work, such as the Klein­Nishina
cross­section, and may allow the background spectrum to be produced by AGN.
A power law fit gives a slope ff E =0.63\Sigma0.03 and needs no additional compo­
nents down to quite tight limits: a Lightman & White (1988) Compton reflection
component is not required and the 90% limit to the solid angle subtended by the
disk
is\Omega d =2ú = 0:4; a narrow 6.4 keV (rest frame) iron line is not required by the
fit either, and the 90 % limit on its rest frame equivalent width is 120 eV. This
is strikingly similar to the unusual Seyfert galaxy NGC4151 (Yaqoob et al. 1993),
and differentiates S5 0014+813 from typical low redshift Seyfert galaxies which have
strong iron K­lines (EW=100­300 eV), and a strong Compton
hump(\Omega d =2ú ú 1,
Nandra & Pounds 1994 and references therein). Williams et al (1992) found these
features to be weak in high luminosity, radio­loud AGN. S5 0014+813 strengthens this
trend. Zdziarski, —
Zycki, & Krolik (1993) use a strong reflection component in AGN
to integrate over redshift to produce the X­ray background. Not seeing a reflection
hump in S5 0014+813 detracts from this model.
ASCA has demonstrated an ability to produce good spectra of quite faint
quasars. This should be exploited for more high redshift quasars, particularly radio­
quiet quasars and quasars at redshifts more likely to dominate the X­ray background.
We also note that even in the lower (rest frame) energy range of 2­10 keV there are only
about 20 good S/N quasar spectra, all from Ginga (Williams et al 1992). Here ASCA
is ideally placed to obtain high quality spectra. Even short ASCA observations of
bright low redshift quasars will dramatically improve our basic knowledge of quasars.
1 The only other quasar with a high S/N spectrum above 10 keV (emitted) was 3C273 (Turner et
al 1990). However, an observation of a transient source of equal or greater strength to 3C273 at 60
keV only 15 arcmin from 3C273 (GRS1227+025, Bassani et al 1991) puts in doubt the reliability of
the large beam Ginga measurement.
5

We thank Smita Mathur for identifying the K0 star. We thank Keith Gendreau
and the staff of the ASCA GOF at GSFC for their assistance with the data reduction.
The spectral analysis was performed using XSPEC, and the data was extracted
using XSELECT, both packages were provided by the ASCA GOF. This work was
supported by NASA grants NAGW­2201 (LTSA) and NAG5­2563 (ASCA).
REFERENCES
Bassani L., et al, 1991, in Proceedings of the 22nd International Cosmic Ray
Conference (Dublin), 1, 173
Bechtold, J., Elvis, M., Fiore, M., Kuhn, O., Cutri, R.M., McDowell, J.C., Rieke, M.,
Siemiginowska, A., & Wilkes, B.J., 1994, AJ, in press
Elvis, M. 1991, ``Frontiers of X­ray Astronomy'' eds. Y.Tanaka, K.Koyama
[Tokyo:Universal Academic Press] p. 567
Elvis, M. Fiore, Wilkes, B.J., McDowell, J.C.M., & Bechtold J. 1994a, ApJ, 422, 60
Elvis, M., et al 1994b, in preparation
Fiore, F., Elvis, E., Siemiginowska, A., Wilkes, B.J., & McDowell, J.C. 1994, ApJ, in
press (August 20)
Guilbert P.W., & Rees M.J, 1988, MNRAS, 233, 475
Johnson W.N., et al., 1994, Proc. 2nd Compton Symposium
K¨uhr H., Witzel A., Pauliny­Toth I.I.K., & Nauber U., 1981, A& AS, 45, 367.
Lawson A.J., Turner M.J.L., Williams O.R., Stewart G.C. & Saxton R.D. 1992,
MNRAS, 259, 743.
Laor, A., Fiore, F. Elvis, E., Wilkes, B.J., & McDowell, J.C. 1994, ApJ, in press
Lightman A.P., and White T.R., 1988, ApJ, 335, 57.
Makishima, K. et al 1994, PASJ, in press
Marshall F.E., et al, 1980, ApJ, 235, 4
Mushotzky R.F., Done C. & Pounds K.A., 1993, Ann. Rev. Astr. & Ap., 31, 717.
Nandra, K., & Pounds, K.A. 1992, Nature, 356, 215
Nandra, K., & Pounds, K.A. 1994, MNRAS, in press.
Serlemitsos P., Yaqoob T., Ricker G., Woo J., Kunieda H., Terashima Y., Iwasawa
K., 1994, PASJ, in press.
Stark A.A., Gammie C.F., Wilson R.W., Bally J., Linke R., Heiles C., & Hurwitz,
M. 1989, ApJS, 79, 77.
Shastri, P., Wilkes, B.J., Elvis, M., & McDowell, J.C. 1993, ApJ, 410, 29
Tanaka, Y., Holt, S.S., & Inoue, H. 1994, PASJ, 46. L37
Turner, M.J.L., et al., 1990, MNRAS, 244, 310
Turner, T.J., Nandra, K., George, I.A., Fabian, A.C., & Pounds, K.A. 1993b, ApJ,
in press
Williams, O.R., et al. 1992, ApJ, 389, 157
Walter, R. & Fink, H.H. 1993, A&A, 274, 105
Yaqoob T., Warwick R.S., Makino F., Otani C., Sokoloski J.L., Bond I.A., &
Yamauchi M., 1993, MNRAS, 262, 435.
Zdziarski A.A., —
Zycki P.T., & Krolik J.H., 1993, Ap.J.Letters, 414, L81
6

Figure Captions
Figure 1: ASCA GIS + PSPC X­ray spectra of S5 0014+813. The solid line
shows the best fit single power­law model fitted from 0.2 to 8 keV. Energies are in
the observed frame.
Figure 2: Residuals after subtracting the best fitting 0.2­8 keV power law
from the ASCA GIS and PSPC spectra of S5 0014+813. Energies are in the observed
frame.
7

Table 1: ASCA Observations of S5 0014+813
Instrument Counts exposure (s) background (cts) net ct/s
GIS2 2904 31635 800 0.066
GIS3 3244 31500 860 0.075
Table 2: GIS ASCA Power­law spectral fits for S5 0014+813
Instruments energy range a N b
H ff E , or kT c norm. d ü 2 (dof)
[\Omega d /2ú]
Power Law
GIS2 0.8­8 0.22\Sigma0.12 0.69\Sigma0.14 6.1\Sigma1.0 78.1 (58)
GIS2 0.8­8 0.144FIX 0.62\Sigma0.05 5.5\Sigma0.3 79.1 (59)
GIS3 0.8­8 0.23\Sigma0.12 0.72\Sigma0.13 7.3\Sigma1.2 49.2 (67)
GIS3 0.8­8 0.144FIX 0.63\Sigma0.05 6.5\Sigma0.3 50.4 (68)
GIS2+3 0.8­8 0.22\Sigma0.09 0.70\Sigma0.09 -- 127.5 (127)
GIS2+3 0.8­8 0.144FIX 0.63\Sigma0.03 -- 129.5 (128)
PSPC+GIS2+3 0.2­8 0.144FIX 0.63\Sigma0.05 -- 149.8 (147)
0+2.0 e
PSPC+GIS2+3 0.2­2 0.144FIX 0.67\Sigma0.17 -- 65.0 (55)
GIS2+3 2­8 0.144FIX 0.68\Sigma0.05 -- 81.9 (88)
Thermal Bremsstrahlung
PSPC+GIS2+3 0.2­8 0.144FIX 38.9 +4:1
\Gamma3:5 149.6 (148)
GIS2+3 0.8­8 0.144FIX 39.6 +4:3
\Gamma3:6 125.6 (128)
Reflection
PSPC+GIS2+3 0.8­8 0.144FIX 0.63\Sigma0.03 -- 149.8 (147)
[0.0 + 0.3]
GIS2+3 0.2­8 0.144FIX 0.62\Sigma0.03 -- 129.5 (127)
[0.0 + 0.4]
a in keV; b in units of 10 22 atoms cm \Gamma2 ; c in keV, in quasar rest frame, for thermal
Bremsstrahlung fits; d in units of 10 \Gamma4 keV s \Gamma1 cm \Gamma2 keV \Gamma1 ; e column at z=3.38.
8