Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/QEDT/Papers/tokyo94.ps Äàòà èçìåíåíèÿ: Sat Jul 16 00:01:33 1994 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 23:46:37 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: http astrokuban.info astrokuban

X­RAYS AND HIGH REDSHIFT QUASARS
Martin Elvis
Harvard­Smithsonian Center for Astrophysics, Cambridge, MA 02138, USA
(internet: elvis cfa.harvard.edu)
ABSTRACT
X­ray observations of high redshift (z¸3) quasars have yielded a wealth of sur­
prises, both in their emission spectra, and in their absorption properties. The same
observations can place limits on a hot diffuse IGM (the ``X­ray Gunn­Peterson test''),
and on physical conditions in damped Lyman­ff absorbers. This paper reviews the
current status of the ROSAT observations, and looks forward to the diagnostics that
will be provided by ASCA and other spectroscopy missions, such as AXAF.
1. The Moving Frontier
At the Nagoya meeting on `Frontiers of X­ray Astronomy' in 1991 I showed the
X­ray frontier for quasars in several observable dimensions, e.g. ff OX , redshift (Elvis
1991), and speculated on how our ideas of what was true for quasars in general may
be biased by our limited observing capabilities.
Since then we have pursued a strategy with ROSAT that explores the whole pa­
rameter space of quasars ­ high redshift, X­ray quiet (Fiore et al 1993, Mathur et
al 1994), optically­quiet (Elvis et al 1994a, Mathur 1994), IR­quiet and radio­silent
(Aldcroft et al 1994 in preparation), high signal­to­noise (Fiore et al 1994), and low
Galactic NH (Laor et al 1994). Figure 1 shows the 1991 frontier projected into (ff OX ,
z) space. Our new ROSAT PSPC points are superposed, showing how far we have
reached beyond our knowledge of 3 years ago.
Figure 1: Observed regions in (ff OX ,z) space. Stars show our ROSAT points,
filled squares the IPC spectral observations, defining the 1991 `frontier'­ approx­
imated by the dashed line.

Our strategy has successfully found surprises. Here I describe only our results on
high redshift quasars and note where ASCA will do better, and where it can not.
2. A ROSAT Program
We observed 14 high z quasars between z=2.85 and z=4.11. yielding 25 \Gamma 1000
PSPC counts per quasar. Only radio­loud quasars, which are relatively X­ray bright,
gave enough counts to allow well­constrained spectral fits. Radio­quiet quasars give
only an X­ray color, which nevertheless turned out to be valuable.
An altered perspective is needed when thinking about ROSAT observations of
high redshift quasars. First, in the quasar frame ROSAT observes 0.5\Gamma10 keV. This
is the same rest energy range as Ginga observes for z¸0 quasars, thus enabling simple
comparisons to be made. Next, optical evolution studies show that the typical quasar
at z=3 is ¸50 times more luminous than at z=0: L \Lambda
Q (z = 3) ¸ 50 L \Lambda
Q (z = 0) (Boyle
et al 1990). By coincidence the quasars which are observable with ROSAT at z=3 are
also ¸50 times more luminous than those Ginga could observe, so we are investigating
the same relative position on the quasar luminosity function at both epochs. These
luminosities are large, LX = 10 47\Gamma48 erg s \Gamma1 , compared with 10 46 erg s \Gamma1 for 3C273.
In black hole models these luminosities require central masses of ¸10 10 M fi in order
not to exceed the Eddington limit, i.e. of order 1/10 the Milky Way mass. Finally,
z=3 corresponds to a look back time of ¸75% the age of the
Universe(\Omega 0
=0). This
leaves plenty of room for evolution of both the quasars and their environment.
3. Absorption: Radio­Loud High z Quasars
Low energy cut­offs are seen in 3 out of 4 ROSAT spectra of z¸3 quasars and
so are common (Elvis et al 1994b), but are not universal, so a diffuse intergalactic
medium (IGM) is not responsible.
The simplest interpretation is that these cut­offs are due to photoelectric absorp­
tion. If so, the column densities implied are ¸ 10 21\Gamma22 atoms cm \Gamma2 . The higher values
apply if the absorption is close to the quasar, the lower values if the absorption is local
to us. These values also assume solar abundances, which may well not apply. Abun­
dances for intervening absorbers are likely to be lower (Pei, Fall & Bechtold 1991)
implying larger NH by up to a factor 4. Abundances in the quasar may be enhanced
by factors ¸ 3 \Gamma 14 (Hamman & Ferland 1992), decreasing NH by the same factors.
A site within the quasar broad line region seems unlikely, as does absorption in
a jet (Elvis et al. 1994), so a new absorption site is required. The ROSAT spectra
give no X­ray redshift so this could be anywhere on the long path between us and
the quasar. This path length is more than 20 times greater than that toward the
previously studied quasars at zŸ0.4 (from Einstein (Wilkes & Elvis 1987) and Ginga
(Williams et al 1993)), suggesting that intervening explanations are likely. Probabil­
ity arguments show that the population of ``damped Lyman­ff'' absorbers provide a
plausible location. 1 These systems have column densities of 10 19 ­10 22 atoms cm \Gamma2
(Wolfe 1988) and are often considered to be the disks of protogalaxies. For the quasars
1 `damped' refers to being on the `damping wing' of the `curve­of­growth, i.e. the relation
between line width and column density (Spitzer 1978). The damping wing is caused by the
`natural' line width due to Heisenberg Uncertainty Principle broadening. This simple point
can be hard to track down in textbooks.

we observe they would be located at z!2 in order for the damped Lyman­ff absorp­
tion line not to be detectable. The probability argument is good, if no evolution in
damped Lyman­ff density occurs for z!2. However, such evolution is expected since
protogalaxies are rare today. Other explanations should not be dismissed therefore.
An intrinsic origin, i.e. one physically related to the quasar, must be considered.
To find a new site we probably need to look on a larger scale than for low redshift
absorbers. The host galaxy or its environment are obvious locations to consider.
At z?0.7 radio­loud quasars tend to be found in rich galaxy environments (Yee
and Green 1984, Ellingson, Yee and Green 1991). Half of all z¸3 radio­loud quasars
are highly compact (¸ 100pc) and show low frequency radio cut­offs in the GHz range
(the `GPS' or ``GigaHertz Peaked Sources'', O'Dea 1990). These symptoms strongly
suggest a confining medium (leading to either free­free absorption or synchrotron self­
absorption in the radio). If high z clusters are similar to low z clusters then many
of them will contain cooling flows (Edge & Stewart 1991) many of these will exhibit
X­ray absorption (White et al 1991, Allen, these proceedings). Thus in the X­ray
absorption of high z quasars we may be detecting the intracluster medium at z¸3.
The key input needed to answer the ambiguities in the PSPC data is an X­ray
redshift. It is unlikely that ASCA can give this because the oxygen edge is the strongest
feature to search for, and this is rapidly redshifted out of the SIS band. Nevertheless
it is worth attempting (Yaqoob et al these proceedings). A direct redshift requires
fair spectral resolution (E/\DeltaE –20) over the 0.1­0.6 keV band. AXAF and XMM will
provide this. In the meantime indirect approaches are promising (Stewart et al, these
proceedings), e.g. if the absorption is limited to radio­loud objects, esp. the GPS,
then it must be intrinsic.
4. Absorption: X­Ray Limits on the Intergalactic Medium (IGM)
Not all z=3 quasars show cut­offs. One such has known damped Lyman­alpha
absorbers. The X­ray column limits the ionization state and, given a background
ionizing flux, the density and size of the absorber. The present limit indicates low
ionization and is unrestrictive on size (!25 Mpc, Elvis et al 1994b). However an
achievable limit of 1/10 the column density would give a size limit comparable with a
galaxy (!25 kpc) so that this method is a highly promising tool (Elvis et al 1994b).
The absence of absorption toward some high z quasars can also be used to put
limits on the possible cosmological
density,\Omega IGM , of a hot diffuse IGM: an `X­ray
Gunn­Peterson test' (Shapiro and Bahcall 1982) using both edge and line, rather than
just line, opacity in the soft X­rays. The K­edges of oxygen, neon, silicon and the L­
edge of iron produce the absorption, which is spread out by the redshift of the source.
For an assumed thermal history the optical depth of this absorption can be calculated,
with some other assumptions.
This test is restrictive (Aldcroft et al 1994) for a somewhat enriched IGM at
temperatures intermediate between those ruled out by COBE (Mather et al 1993), and
the cold IGM ruled out by the traditional Lyman­alpha Gunn­Peterson test (Gunn
and Peterson 1965, Giallongo et al 1992), i.e. 10 5\Gamma6 K.
The PSPC data for three z¸3 quasars limit an IGM of solar abundance quite well
(\Omega IGM !0.2 for T IGM ! 5 \Theta 10 6 , for
0.2!\Omega !1.0) (figure 2a), although a more
plausible abundance of 1/10 solar is not better constrained than by the Lyman­ff

Figure 2: (a) ROSAT limits
to\Omega IGM versus the Temperature of the IGM . (b)
Potential limits for E/\DeltaE=200, A=2000 cm 2 , t=10 5 s. Solar abundances, heavy
line, 1/10 solar, medium line. Regions to top and left are disallowed.
test (assuming an isothermal thermal history.) Photoionization of the IGM by the
integrated quasar light, J š21 (Bajtlik et al 1988), has a large effect (Aldcroft et al
1994), making early estimates too optimistic (Shapiro & Bahcall 1982, Rees & Sciama
1967). Even for a J š21 1/10 the nominal value, the limits are an order of magnitude
above the limits
on\Omega from big bang nucleosynthesis (BBN, Kolb & Turner 1990).
Somewhat surprisingly simulations show that the improved energy resolution of
the SIS detectors on ASCA do not improve these limits significantly. At high z this
is because all spectral features are spread by a factor four due to redshifting. At
low redshift one should gain by being able to discriminate individual absorption lines.
The SIS do not do this both because their spectral resolution is not high enough, and
because the main edges are due to redshifted Oxygen, which lies at the extreme low
energy end of the SIS bandpass.
A resolution of E/\DeltaE–100 is needed to cleanly resolve lines. The low energy
gratings on AXAF (Brinkman et al 1987) easily have this. On 3C273 a 10 5 s exposure
will bring the X­ray Gunn­Peterson limits close to the BBN limits (Aldcroft et al
1994). Since all quasar observations contribute to an IGM limit we can expect that
over the course of the AXAF lifetime sufficient exposure will accumulate to bring the
sensitivity of the test into an interesting regime (Figure 2b, Cen & Ostriker 1993).
5. Emission: X­ray loudness (a OX vs. z, L)
A basic characteristic of quasars is the scale­free nature of their emission. From the
lowest luminosity AGN (e.g. M81) to the highest redshift, most luminous quasars, the
quasar continuum and emission lines scale almost linearly with luminosity, regardless
of redshift (e.g. Netzer 1990; Blandford 1990). This is true within both the radio­loud
and the radio­quiet classes of quasar. There are small deviations from this pattern,
e.g. the Baldwin effect (Baldwin 1977), and the slowly decreasing X­ray loudness for
higher luminosity quasars (Worrall et al. 1987). Even these, however, are continuous
changes in the sense that there is no characteristic redshift or luminosity at which
they occur. This is somewhat surprising because the Eddington luminosity provides

Figure 3: X­ray loudness (ff OX ) versus optical luminosity (L opt ) for (a) radio­
quiet quasars, (b) steep spectrum radio­loud quasars.
a natural break point in the standard black hole model for AGN.
Breaks in the scaling laws for quasars may now be beginning to appear (Elvis et
al 1994, Bechtold et al 1994 a, b):
(1) Steep spectrum radio­loud quasars do not continue to follow the well­established
correlation (e.g. Worrall & Wilkes 1990) of ff ox increasing with L opt at high redshift 2 ;
they are too X­ray bright (Figure 3b). Radio­quiet quasars instead continue to follow
the correlation of ff ox increasing with increasing L opt seen at lower redshift (Figure 3a).
(2) Radio­loud quasars become strongly X­ray absorbed between z=0.5 and z=3,
whereas radio­quiet quasars probably do not (section 6, Bechtold et al 1994a, b).
These characteristic luminosities and/or redshifts can be used to limit the physics of
the emission for steep spectrum radio­loud quasars. For example, since LX ceases to
be proportional to L opt it is likely that the source of the X­ray power ceases to be
proportional to the central mass and accretion rate. A second source, e.g. the extrac­
tion of black hole spin energy (Blandford 1990) may become important. If we identify
the break luminosity as where the spin­derived luminosity becomes comparable to
accretion luminosity (Bechtold et al 1994b) then, since the two luminosities depend
differently on central mass (Blandford 1990) they define a critical mass, ¸10 9\Gamma10 M fi ,
assuming an Eddington limited source. The efficiency with which spin energy is con­
verted into X­rays also comes out of this analysis and is small, ¸1% .
6. Emission: X­ray Spectral Evolution
Radio­quiet quasars are fainter in X­rays than radio­loud quasars for a given optical
luminosity, a situation that is made worse at high luminosities by the breaking of the
scaling law described in the preceding section. This makes it hard to obtain high
signal­to­noise detections of radio­quiet quasars at high redshifts. We have found that
even a simple hardness ratio gives useful information on their X­ray properties, even
though we have as few as 25 counts in individual detections (Bechtold et al 1994).
Figure 4 is a plot of the hardness ratio R , for the quasars in our sample. (R=(H­
S)/(H+S), where H = PI channels 11­40, 0.11­0.40 keV, S = PI channels 41­245, 0.41­
2 ff OX = \Gammalog(f opt (2500) š A=fX (2keV))=2:605. Larger ff OX equals more X­ray quiet.

­1 ­0.5 0 0.5 1 1.5
Figure 4: X­ray hardness ratio, R, from the ROSAT PSPC for high z quasars.
Ginga predicted ranges of R for radio­quiet (RQ) and radio­loud (RL) are shown.
2.48 keV, in the observed frame). We also plot the R of four intermediate redshift,
radio­quiet quasars (1:5 !z! 2:2) from the PSPC field of Q0130­403 (Bechtold et al
1994b). Note the systematic difference between the radio­loud and radio­quiet quasars.
The difference in R between low and high redshift radio­loud quasars, can be
interpreted in different ways. The two most simple and extreme possibilities are:
(A) a difference in emitted power law index, in the sense that the high­redshift quasars
are flatter than the low redshift ones, or
(B) a difference in absorbing column density, in the sense that the high z quasars have
a larger absorbing column than at low z.
ASCA observations can easily determine whether radio­quiet quasars change their
slopes radically at z¸2, and so will quickly remove many of the ambiguities from the
present situation. The results of Stewart et al (these proceedings) are suggestive.
7. Conclusions: Scaling Laws and High Redshift Quasars
Expanding the frontier over which we know quasars' X­ray spectra has led us to
two ways in which the scaling laws for quasars break down at high luminosities and/or
redshifts. These breakdowns occur for radio­loud quasars, which become strongly X­
ray absorbed between z=0.5 and z=3, whereas radio­quiet quasars probably do not;
also steep spectrum radio­loud quasars do not follow the well­known correlation of ff ox
increasing with L opt at high redshift­ they are too X­ray bright; Radio­quiet quasars
instead continue to follow the correlation of ff ox increasing with increasing L opt seen
at lower redshift. These breaks in the scaling laws may be related to the origin of the
quasar power output from either accretion or spin energy.
We have also made new calculations for the X­ray `Gunn­Peterson' test for an

IGM, and made the first observational use of this test.
I thank my fellow team members involved in this work­ Tom Aldcroft, Jill Bechtold,
Fabrizio Fiore, Smita Mathur, Jonathan McDowell, Aneta Siemiginowska, and Belinda
Wilkes. This work was supported in part by NASA grant NAGW­2201 (LTSA).
8. References
Aldcroft T.L., Elvis M., McDowell J. & Fiore R., 1994, Ap.J., submitted.
Baldwin, J. 1977, MNRAS, 178, 67P.
Bajtlik S., Duncan R. C., Ostriker J. P., 1988, Ap.J.,327, 570.
Bechtold, J., Elvis, M., Fiore, M., Kuhn, O., Cutri, R.M., McDowell, J.C., Rieke,
M., Siemiginowska, A., & Wilkes, B.J., 1994a, A.J., in press.
Bechtold, J., Elvis, M., Fiore, M., Kuhn, O., Cutri, R.M., McDowell, J.C., Rieke,
M., Siemiginowska, A., & Wilkes, B.J., 1994b, A.J., submitted.
Blandford, R. D., 1990, in ``Active Galactic Nuclei'', Saas­Fee Advanced Course
20 [Berlin: Springer­Verlag] eds. T.J.­L. Courvoisier and M. Mayor.
Boyle B.J., Fong R., Shanks T., and Peterson B.A. 1990, MNRAS, 243, 1.
Edge A.C., & Stewart G.C., 1991, MNRAS, 252, 414.
Ellingson E., Yee H.K.C. and Green R.F., 1991, Ap.J.s, 76, 455.
Elvis M., 1991, ``Frontiers of X­ray Astronomy'',
Elvis, M., Fiore, F., Mathur S. & Wilkes, B., 1994a, Ap.J., 425, 103.
Elvis, M., Fiore, F., Wilkes, B., McDowell, J., & Bechtold, J. 1994b, Ap.J.,422,60.
Fiore F., Elvis M., Mathur S., Wilkes B.J., and McDowell, 1993, ApJ, 415, 129.
Fiore, F., Elvis, E., Siemiginowska, A., Wilkes, B.J., & McDowell, J.C., 1994, ApJ,
submitted.
Giallongo, E., Cristiani, S., & Trevese, D. 1992, Ap.J.lett, 398, L9.
Gunn J.E. and Peterson B. A., 1965, Ap.J., 142, 1633.
Hamman F. & Ferland G., 1992, Ap.J., 391, L53.
Kolb, E.W. & Turner,M.S. 1990, The Early Universe, (New York:Addison­Wesley)
Laor, A., Fiore, F. Elvis, E., Wilkes, B.J., & McDowell, J.C. 1994, Ap.J., submitted
Brinkman A.C., van Rooijen J.J., Bleeker J.A.M., Dijkstra J.H., Heise J., De Korte
P.A.J., Mewe R., and Paerels F., 1987, Ap. Letters & Comm., 26, 73.
Mather, J. C., et al., 1993, COBE preprint no. 93­01
Mathur, S., 1994, Ap.J., submitted.
Mathur, S., Wilkes, B., & Elvis, M. 1994, Ap.J., in press.
Netzer, H. 1990, in ``Active Galactic Nuclei'', Saas­Fe Advanced Course 20 [Berlin:
Springer­Verlag] eds. T.J.­L. Courvoisier and M. Mayor.
O'Dea C.P., 1990, MNRAS, 245, 20P.
Pei Y., Fall M., Bechtold J., 1991, Ap.J., 378, 6.
Shapiro P.R. and Bahcall J.N., 1980, Ap.J., 241, 1.
Spitzer L., 1978, `Physical Processes in the Interstellar Medium, [Wiley­Interscience].
White D.A., Fabian A.C., Johnstone R.M., Mushotzky R.F., and Arnaud K.A.,
1991, MNRAS, 252, 72.
Wilkes B.J. and Elvis M., 1987, ApJ, 323, 243.
Wilkes B.J., Elvis M., Fiore F., McDowell J.C., Tananbaum H. and Lawrence A.
1992, ApJ (Lett), 393, L1.
Williams O.R. et al., 1992, ApJ, 389, 157.
Wolfe A.M., 1988, in ``QSO Absorption Lines'', eds. J.C. Blades, D. Turnshek,
and C.A. Norman [Cambridge:CUP], p.297.
Worrall D.M., and Wilkes B.J., 1990, Ap.J., 360, 396.
Worrall, D.M., Giommi, P., Tananbaum, H. & Zamorani G., 1987, ApJ, 313, 596.
Worrall D.M., and Wilkes B.J., 1990, Ap.J., 360, 396.