Elihu Boldt

Although x-ray astronomy certainly flourished during the initial 15 years of its history, research on the CXB (cosmic x-ray background) remained comparatively dormant. During this early period experiments concentrated on well-isolated bright sources, particularly compact objects of high astrophysical interest such as neutron stars in binary stellar systems. It remained for the first two High-Energy Astronomy Observatories (HEAO-1 and HEAO-2) to significantly remedy the relatively poor observational situation that existed before concerning the CXB. The all-sky survey carried out over a broad band of photon energies (from 0.1 keV to over 0.5 MeV) with the HEAO-1 mission involved newly developed experiments especially designed to unambiguously distinguish the x-ray sky background from that due to other causes. The grazing incidence focusing x-ray telescope flown on the HEAO-2 mission, usually known as the Einstein observatory, brought the power of focusing optics to x-ray astronomy. For soft x-rays (leq 3 keV), the imaging detectors at the focus of this telescope were used to resolve a substantial portion of the background into discrete faint sources.

Basic measured characteristics of the cosmic x-ray background are reviewed here, including recent results from autocorrelation studies of surface brightness fluctuations examined with the HEAO-1 and the Einstein observatory (HEAO-2). Prospects for addressing some key outstanding issues with future experiments are discussed as regards possible weak large-scale anisotropies (e.g., dipole) and spectral tests to help discriminate among candidate scenarios for the sources.


The x-ray sky (at photon energies greater than about 3 keV) is dominated by a remarkably isotropic extragalactic CXB exhibiting an optically thin thermal-type spectrum characterized by a photon energy of 40 keV corresponding to a temperature T = 46 x 107 K, much as the sky in the microwave band is dominated by an isotropic thermal CMB (cosmic microwave background), albeit characteristic of a blackbody at 3 K. Discrete x-ray sources resolvable with the imaging telescope of the Einstein observatory account for about 20% of the CXB(evaluated at 3 keV); they number about 10 per square degree. The upper limit to arcminute scale fluctuations in the apparent surface brightness of the unresolved background observed with this telescope implies that most of this residual CXB is either diffuse (e.g., due to a hot intergalactic plasma) or arises from a discrete source population having a number density on the sky exceeding one per square arcminute, much more than can be accounted for by quasars. An upper limit to the autocorrelation function for surface brightness fluctuations on scales geq 5 arcminutes has also been derived from Einstein observatory data; it sets an upper bound on the correlation length for the possible dominant sources of the residual CXB that is only 0.3% of the characteristic distance scale (c / H0) associated with the Hubble constant (H0) for the expansion of the universe (c = speed of light; 1 / H0 = 10-20 billion years).

The broadband (3-50 keV) all-sky study of the background carried out with the gas proportional chambers of the HEAO-1 A2 experiment resolved only the brightest foreground sources; in total these resolved sources account for about 1% of the CXB (at 3 keV). Fluctuations in the surface brightness of the CXB on scales geq 3° observed with this experiment are consistent with random variations in the expected number of unresolved foreground sources. The HEAO-1-derived upper limit to the autocorrelation function for these CXB surface brightness fluctuations on scales geq 3° sets an upper bound on the correlation length for x-radiating rich clusters of galaxies which is consistent with the correlation found in the optical. For those unresolved AGN (active galactic nuclei) within the present epoch (i.e., at redshifts z leq 0.1) contributing to the foreground fluctuations in surface brightness of the CXB the HEAO-1 limit implies an upper bound on their correlation length = 0.6%(c / H0).


If the proper frame of the CXB were at rest relative to the proper frame of the CMB, then our own velocity (v) relative to this universal system would induce a CXB dipole anisotropy (i.e., Compton-Getting effect) in the direction of the CMB apex having an amplitude (alpha + 3)v/c that varies fromapprox 0.4% at leq 1O keV (where the effective CXB energy spectral index alpha approx 0.4) to approx 0.6% at geq 100 keV (where the effective alpha approx 1.6). However, fluctuations in CXB surface brightness observed with the HEAO-1 amount to geq 0.3% for bands in the sky which are leq 1 sr(i.e., leq 3000 square degrees) in solid angle. Within the uncertainty imposed by such fluctuations, the weak large-scale anisotropy of the CXB is consistent with that implied by the CMB. If we could resolve out foreground sources at a level corresponding to geq 10 per square degree (i.e., comparable to the point source sensitivity level of the Einstein Observatory or better) over the whole sky, then large-scale surface brightness fluctuations in the residual CXB might well be suppressed by at least an order of magnitude, thereby permitting a sufficiently precise determination of the CXB Compton-Getting dipole anisotropy. Possible structure of the CXB on intermediate angular scales (e.g., such as might be associated with extragalactic objects like the ``great attractor'' or even with our galaxy) could, however, constitute a fundamental complication that would still have to be addressed. At geq 100 keV the best available all-sky data are from the HEAO-1 A4 scintillation counter experiment. The limitation in measuring weak large-scale CXB anisotropies with this experiment arises from variable radioactivity induced in the scintillators by the radiation environment of the HEAO-1 orbit, in spite of the precautions employed. Future high-energy studies of the CXB will have to avoid or further minimize this complication.


It is now well established that the apparently thermal-type spectrum of the CXB is significantly different from the power law x-ray spectrum characteristic of typical bright AGN(active galactic nuclei). This has led to the notion that perhaps there is an evolution of AGN whereby those at the largest redshifts have x-ray spectra that differ from the canonical spectrum observed for those within the present epoch. However, such spectral evolution would not be necessary if most of the CXB were due to a hot IGM (intergalactic medium). Although this hot IGM would have to be of such magnitude as to dominate the baryonic matter content of the universe, cooling by the CMB would restrict it to redshifts within z = 6. Given this redshift constraint, an acceptable IGM model for the observed CXB spectrum would have to demand that the perturbation permissible from canonical AGN contributions be less than currently estimated.


Including the contribution of foreground extragalactic sources unresolved with the HEAO-1, the total 3-100 keV CXB spectrum observed may be nicely described by optically thin thermal bremsstrahlung radiation corresponding to a hot plasma at T = 46 x 107 K (i.e., kT = 40 keV, where k is the Boltzmann constant). A good thermal fit characterized by kT = 40 keV was first established for the band 3-50 keV with gas proportional chamber data from the HEAO-1 A2 experiment. Data from the scintillation counters of the HEAO-1 A4 experiment were then used to determine that a thermal fit was similarly valid up to 100 keV. The temperature associated with this CXB spectrum is an order of magnitude higher than that for the x-radiating thermal plasmas associated with rich clusters of galaxies. Furthermore, the thermal form of the 3-100 keV CXB spectrum is apparently distinct from the nonthermal power law x-ray spectra characteristic of the brightest AGN observed with the HEAO-1 over the same band. Subtracting the estimated contributions of sources making up such known extragalactic populations (i.e., a foreground amounting to about 40% of the CXB, at 3 keV) yields a residual CXB energy spectrum that is remarkably well described by a simple exponential function characterized by an e-folding energy of 23 keV; this spectrum is significantly flatter below about 10 keV than what is expected from a thin thermal plasma. Hence the spectral paradox posed by the total CXB appears to become appreciably more so by our attempting to isolate a residual CXB. With this procedure, however, we have sharpened our picture of the very particular sort of spectrum required for major sources of the CXB.

The principal portion of the subtracted foreground discussed above arises from AGN with canonical power law spectra characterized by an energy spectral index alpha approx 0.7. To exhibit the spectral consequences of subtracting various different estimated amounts of such AGN foreground from the CXB measured with the HEAO-1 A2 experiment, the residual energy spectrum (3-50 keV) corresponding to each assumed foreground level has been fitted with the functional form

IE ident dI / dE propto E-alpha0 exp(-E / B),

where E is the photon energy and alpha0 and B are parameters determined from the spectral fit. In doing so the simplifying assumption is made that all foreground AGN have canonical power law spectra (i.e., propto E-0.7) over this band. A graphical representation of alpha0 and B thereby obtained for the residual CXB is displayed in Fig. 1 as a function of the AGN foreground level at 3 keV. For zero foreground, we recover alpha0 = 0.29 and B = 40 keV corresponding to the thermal bremsstrahlung spectrum characterized by kT = 40 keV that describes the total CXB. We note that for an AGN contribution exceeding about 30% (at 3 keV), alpha0 < 0.2 and B < 30 keV; this limit on alpha0 would imply that the candidate "thin thermal" sources of the residual CXB have kT > 200 keV in their proper frame and, coupled with the limit on B, that they are located at redshifts z = [(kT / B)-1] > 6, beyond the highest-redshift quasars as well as beyond a possible hot x-radiating intergalactic plasma.

Figure 1

Figure 1. Parameters alpha0 and B characterizing the indicated spectral fit to the residual CXB (3-50 keV) as a function of the assumed percentage of AGN foreground (at 3 keV).

If due to discrete objects, the residual CXB spectrum could readily arise from the Comptonized thermal emission characteristic of extremely compact sources whose accretion powered luminosity is mainly in x-radiation, at the maximum permissible level. These compact x-ray sources of the residual CXB could very well be high-redshift objects that represent an early stage in the evolution of AGN and not a new ad hoc population; this sort of spectral evolution would be inherent to the underlying physical processes involved. Unlike canonical AGN (e.g., those at low redshifts), their radiation in the IR, optical, and UV would be relatively small. In summary, AGN spectral evolution is not only an attractive simple solution to the severe spectral paradox associated with a residual CXB but could provide us with strong evidence that redshift is indeed a direct measure of the ``arrow of time.''


Is the pronounced difference between the spectrum of the residual CXB and that of foreground sources compelling evidence for AGN spectral evolution? To explore this question, we consider the alternate possibility that there exists an as-yet unknown broadband x-ray spectral form for AGN which is essentially independent of cosmological epoch and can be understood to account for the puzzling spectrum of the entire CXB by suitably integrating the redshifted contributions of all such sources. This universal spectrum would have to be significantly flatter than the canonical (alpha = 0.7) power law over a substantial portion of the x-ray band. In particular, as shown in Fig. 2, the redshifted spectra of such principal sources of the CXB must superpose to a composite spectrum characterized by alpha approx 0.4 over the band 3-10 keV and alpha approx 0.7 over the band 10-20 keV. We already know, though, that the spectra for essentially all present-epoch AGN are well described by alpha = 0.7 over the band 3-10 keV. If the principal sources of the CXB are indeed to be representative of a universal AGN spectral form, they should exhibit spectra which, in the reference frame of emission, matches the canonical one over the 3-10-keV band. To ensure that this spectral component manifests itself to the observer mainly below 3 keV, however, most of the CXB would have to arise from sources of redshifts z > 2 [ie., E (observed) = E (emitted) / (1+z)]. In fact, our knowledge of the CXB spectrum below 3 keV is still relatively uncertain, and the possibility of a universal AGN spectrum cannot as yet be ruled out.

Figure 2

Figure 2. The ratio (R) as a function of energy of the counts observed for the total CXB to that predicted by convolving, with the detector response function, power law spectra characterized by spectral indices alpha = 0.4 and alpha = 0.7. Different symbols distinguish data obtained with the HED-1 and HED-3 high-energy detectors of the HEAO-1 A2 experiment. These units were xenon-filled multianode proportional chambers. Statistical errors are shown when larger than the size of the symbols.


If the CXB is mainly due to AGN (i.e., not diffuse and not due to other discrete objects such as star-forming galaxies), then these sources must undergo substantial evolution in their luminosity whether or not spectral evolution is involved. As such, much of the CXB would arise from AGN at z > 2. Hence we have the possibility of a redshift test for the spectral evolution of these sources relative to present-epoch AGN. However, we note that future observations of such faint sources to be made possible by the powerful high-resolution imaging x-ray telescope of the AXAF (Advanced X-RAY Astrophysics Facility) will be restricted to photon energies less than 10 keV. Furthermore, if most of the CXB does indeed arise from faint discrete x-ray sources, then observations with AXAF must necessarily yield a significant sample of sources that exhibit a spectral index alpha leq 0.4 (over the 3-10-keV band) just to be compatible with the known CXB spectrum (see Fig. 2), regardless of AGN spectral evolution. However, if these same sources exhibit steeper spectra (corresponding to alpha approx 0.7) below 3 keV this could constitute real evidence for a universal AGN spectrum (i.e., absence of spectral evolution). On the other hand, if the flat spectra of high-redshift sources of the CXB are found to persist well below 3 keV, this would suggest that spectral evolution is a fundamental aspect of the AGN phenomenon.

Finally, should we fail to find a population of point sources that dominate the CXB, then we can use ``fast'' moderate-resolution x-ray optics for arcminute mapping of the weak surface brightness variations that could be characteristic of a thermal background that is intrinsically diffuse. The study of such variations might provide us with a direct indication of the large-scale gravitational fluctuations that trace the distribution of all matter (i.e., dark as well as visible).

Additional Reading

  1. Barcons, X. and Fabian, A. (1989).The small-scale autocorrelation functions of the x-ray background. Monthly Notices Roy. Astron. Soc. 237 119
  2. Boldt, E. (1987). The cosmic x-ray background. Phys. Rep. 146 215.
  3. Leiter, D. and Boldt, E. (1982). Spectral evolution of active galactic nuclei. Ap. J. 260 1.
  4. Marshall, F., Boldt, E., Holt, S., Miller, R., Mushotzky, R., Rose, L., Rothschild, R., and Serlemitsos, P. (1980). The diffuse x-ray background spectrum from 3 to 50 keV. Ap. J. 235 4.
  5. Persic, M., DeZotti, G., Boldt, E., Marshall, F., Danese, L., Franceschini, A., and Palumbo, G. (1989). The autocorrelation properties of fluctuations in the cosmic x-ray background. Ap. J. (Lett.) 336 L47
  6. Schwartz, D. and Tucker, W. (1988). Production of the diffuse x-ray background spectrum by active galactic nuclei. Ap. J. 332 157.
  7. Tucker, W.(1984). The Star Splitters (The High Energy Astronomy Observatories). NASA, SP-466, Washington, DC.
  8. See also Active Galaxies and Quasistellar Objects, X-Ray Emission; Clusters of Galaxies, X-Ray Observations; X-Ray Astronomy; Space Missions.