An Overview of the Performance and Scientific Results from the Chandra X‐Ray Observatory

Hughes Danbury Optical Systems (HDOS; Danbury, Connecticut) precision figured and superpolished the four‐mirror pair grazing‐incidence X‐ray optics out of Zerodur blanks from Schott Glaswerke (Mainz, Germany). Optical Coating Laboratory Incorporated (OCLI; Santa Rosa, California) coated the optics with iridium, chosen for high X‐ray reflectivity and chemical stability. The Eastman Kodak Company (Rochester, New York) aligned and assembled the mirrors into the 10 m focal length HRMA (Fig. 2). The forward contamination cover houses 16 radioactive sources used for verifying transfer of the flux scale from the ground to orbit (Elsner et al. 1994, 1998, 2000).

Zoom In Zoom Out Reset image size

Aft of the HRMA are two OTGs: the Low‐Energy Transmission Grating (LETG) and the High‐Energy Transmission Grating (HETG). Positioning mechanisms are used to insert either OTG into the converging beam where they disperse the X‐radiation onto the focal plane. Figure 3 shows the gratings mounted behind the HRMA in their retracted position.

Zoom In Zoom Out Reset image size

2.6.2.1. Low‐Energy Transmission Grating (LETG)

2.6.2.2. High‐Energy Transmission Grating (HETG)

The Integrated SIM (§ 2.4) houses Chandra's two focal plane science instruments: the (microchannel plate) High‐Resolution Camera (HRC) and the Advanced CCD Imaging Spectrometer (ACIS). Each instrument provides both a so‐called (as all the detectors are imagers) imaging detector (I) and a spectroscopy detector (S), the latter designed especially to serve as a readout for the photons dispersed by the transmission gratings.

2.6.3.1. High‐Resolution Camera

SAO designed and fabricated the HRC (Murray et al. 2000) shown in Figure 4. Made of a single 10 cm² microchannel plate, the HRC‐I provides high‐resolution imaging over a 31 arcmin square field of view. Comprising three rectangular segments (3 cm × 10 cm each) mounted end to end along the OTG dispersion direction, the HRC‐S serves as the primary readout detector for the LETG. Both detectors are coated with a cesium‐iodide photocathode and covered with aluminized‐polyimide UV/ion shields.

Zoom In Zoom Out Reset image size

2.6.3.2. Advanced CCD Imaging Spectrometer

Pennsylvania State University (PSU; University Park, Pennsylvania) and MIT designed and fabricated the ACIS (Fig. 5) with CCDs produced by MIT Lincoln Laboratory (Lexington, Massachusetts). Some subsystems and systems integration was provided by Lockheed Martin Astronautics (Littleton, Colorado). Made of a 2 × 2 array of large‐format, front‐illuminated (FI), 2.5 cm², CCDs, ACIS‐I provides high‐resolution spectrometric imaging over a 17 arcmin square field of view. ACIS‐S, a 6 × 1 array of four FI CCDs and two back‐illuminated (BI) CCDs mounted along the OTG dispersion direction, both serves as the primary readout detector for the HETG and, using the one BI CCD which can be placed at the aim point of the telescope, also provides high‐resolution spectrometric imaging extending to lower energies but over a smaller (8 arcmin square) field than ACIS‐I. Both ACIS detectors are covered with aluminized‐polyimide optical blocking filters.

Zoom In Zoom Out Reset image size

The HRC‐I on‐orbit counting rate is about 250 counts s⁻¹ (∼2 counts s⁻¹ cm⁻²). These are mostly cosmic‐ray events and are detected in the anticoincidence shield. The onboard veto function has been activated and reduces the valid event rate to about 50 counts s⁻¹.

In the case of the HRC‐S, enabling the anticoincidence shield does not reduce the background as expected. The problem appears to be a timing error in the electronics that cannot be changed from the ground. The HRC‐S trigger pulses arrive earlier than the anticoincidence signals and are not held long enough to trigger a coincidence, resulting in telemetry saturation. To cope, a "spectroscopic region" has been defined which is a strip of the detector, 9.6 mm wide, centered on the nominal spectroscopy aim point. This region is about 1/2 of the total HRC‐S area and reduces the valid count rate to about 120 counts s⁻¹, well below the telemetry limit.

An (on‐ground) event screening algorithm has been developed to help in identifying non–X‐ray events and further reduces the background for data from both HRC‐I and HRC‐S. For HRC‐I, e.g., this results in a background rate of about 0.8 counts arcsec⁻² for a 10⁵ s integration. For HRC‐S, this approach reduces the background to ∼13 counts in a 0.06 Å spectral "slice" per 10⁵ s integration.

Even with a reduction in high voltage and gain, there is a degradation of image quality in HRC‐I for the higher amplitude events due to added electronic noise during event processing. Fortunately, there is enough information in the telemetry stream to correct for this effect. A revised ground system event processing algorithm has been developed which identifies distorted events and flags them as such. The added electronic noise, a systematic effect, can be partially compensated for and removed. Algorithms for making these corrections have been added to the data processing system. Using the plate focus data as a test case, the measured encircled energy near the focus was improved from about 40% within 1^'' diameter to greater than 60%.

A second image artifact is the appearance of "ghost" images near strong sources. These ghost images are due to events where the readout amplifiers are saturated leading to fairly large miscalculation of the event position. The number of such misplaced events is 1% or less of the properly placed events. Screening the data for events where the amplifiers are all near their peak value effectively eliminates these "ghosts."

Because the HRC‐S gain is lower than for HRC‐I, electronic saturation effects are less significant. As a result, the imaging performance appears to be excellent once the corrections described above for HRC‐I have been applied.

The count rates from the sources observed are about what was expected. For LMC X‐1, Ar Lac, and Cas A the HRC rates are close to the prelaunch predictions. There have been no measurable changes to the HRC efficiency since launch.

The HRC was miswired so that the time of the event associated with the jth trigger is that of the previous (j - 1) trigger. If the data from all triggers were routinely telemetered, the miswiring would not be problematical and could be dealt with by simply reassigning the time tag which is nominally accurate to 16 μs. Since the problem has been discovered, new operating modes have been defined which allow one to telemeter all data whenever the total counting rate is moderate to low, albeit at the price of higher background. For very bright sources the counting rate is so high that information associated with certain triggers is never telemetered. In this case, the principal reason for dropping events is that the onboard, first‐in‐first‐out (FIFO) buffer fills as the source is introducing events at a rate faster than the telemetry readout. Events are dropped until readout commences freeing one or more slots in the FIFO. This situation can also be dealt with (Tennant et al. 2001a), and time resolution of the order of a millisecond can be achieved even under these conditions. More details of the HRC and its performance may be found in Murray et al. (2000), Kenter et al. (2000), and Kraft et al. (2000).

The background experienced by the ACIS CCDs shows occasional flares depending upon the orientation of the observatory's orbit with respect to Earth's magnetosphere. The frequency and magnitude of the flares are much more pronounced in the BI CCDs than in the FI CCDs, consistent with the suggestion that the flaring events are caused by low‐energy protons that enter the observatory through reflection off the very smooth iridium‐coated mirrors that comprise the HRMA. The gate structure of the FI CCDs absorbs most of the protons for the majority of the flares.

The region of the Hubble Deep Field–North (HDF‐N) was observed using the ACIS‐I array, for 970 ks in a series of 12 pointings spanning a period of 15 months, and thus provides, after removing sources, an excellent representation of the background and its spectrum. The background (0.5–10.0 keV) was observed to be constant to within about 10% for eight of the pointings. The first three pointings were made with the focal plane temperature at −110°C, and the background was about 50% higher than for the background 1 yr later taken at −120°C. A reduction in background was expected at the lower temperature. One pointing, ObsId 2344, experienced a highly variable background with a flare increasing the background by a factor of between 2 and 3 for about 20 ks, depending upon the energy band.

The spectrum is shown in Figure 10. The prominent lines in the background spectrum are from cosmic‐ray–induced fluorescence of the gold‐coated collimator, the nickel‐coated substrate of the collimator, the silicon in the CCDs, and aluminum used in various places in the housing and the filter coating. In general the background produced by the fluorescent lines is only about 2.6% of the background not found in the lines in the soft (0.5–2.0 keV) band and 13.5% of the flux in the hard (2.0–10.0 keV) band. Brandt et al. (2001a) have noted that by selecting certain ACIS event grades it is possible to suppress the background by another 36% in the soft band (0.5–2.0 keV) and by 28% in the hard band (2.0–8.0 keV), while only reducing the source counts by 12% and 14%, respectively.

Zoom In Zoom Out Reset image size

The ACIS FI CCDs originally approached the theoretical limit for energy resolution at almost all energies, while the BI CCDs were of somewhat lesser quality in this regard. Subsequent to launch and orbital activation, the energy resolution of the FI CCDs has become a function of the row number, being nearer prelaunch values close to the frame store region and progressively degraded toward the farthest row (Fig. 11). The points are for the FI data and the curves for the BI data. These data were taken at −120°C. Note that these curves are representative of the variation with row number but do not account for an added row‐dependent gain variation which increases the energy resolution by an additional 15%–20% for the larger row numbers.

Zoom In Zoom Out Reset image size

For a number of reasons, we believe that the damage was caused by low‐energy protons, encountered during radiation belt passages and reflecting off the X‐ray telescope onto the focal plane. Subsequent to the discovery of the degradation, operational procedures were changed so that the ACIS instrument is not left at the focal position during radiation belt passages. Since this procedure was initiated, no further degradation in performance has been encountered. The BI CCDs were not impacted, consistent with the proton damage scenario, as it is far more difficult for low‐energy protons to deposit their energy in the buried channels (where damage is most detrimental to performance) of the BI devices, as these channels are near the gates and the gates face in the direction opposite to the HRMA. The energy resolution for the two BI CCDs remains at their prelaunch values.

The aspect solution of the spacecraft typically provides absolute sky coordinates to about 1^''. It is usual for observations exceeding 20 ks to detect several stars in X‐rays that have accurate positions (typically 04 for USNO‐A2.0 stars and 015 for Tycho‐2 Catalog stars). Using the stellar positions as a local reference, the X‐ray sources can be positioned to about 02–04. One example of using such positions is the determination of the positions of the 1000 X‐ray sources in the Orion Trapezium region (Fig. 13) where the stellar and X‐ray positions differ by about 03 rms (Feigelson et al. 2001).

Zoom In Zoom Out Reset image size

Precise positioning with Chandra was critical for the unique identification with Sgr A* (Baganoff et al. 2001a) in the extremely crowded region of the Galactic center (Fig. 14). The mass of the black hole at Sgr A* has been determined from stellar motion to be 2.6 × 10⁶ M_⊙ (Genzel et al. 2000). The X‐ray source associated with Sgr A* is very faint compared to other galactic nuclei, emitting only about 2 × 10³³ ergs s⁻¹. A bright flare was detected from this source on 2000 October 27, where the flux increased by over an order of magnitude for about 10 ks and then rapidly dipped on a timescale of 600 s (Baganoff et al. 2001b).

Zoom In Zoom Out Reset image size

Another example of Chandra's ability to provide high‐contrast images of features of low surface brightness is exemplified by the now classic image (Fig. 15) of the Crab Nebula and its pulsar (Weisskopf et al. 2000; Tennant et al. 2001a) which shows the intricate structure produced by the pulsar wind and the synchrotron torus.

Zoom In Zoom Out Reset image size

In cases in which there is sufficient signal, the precisely measured point response function of the HRMA permits image deconvolution. A good example is given by the observation of the recent SNR SN 1987A in the LMC. The image is shown in Figure 16, where a Lucy‐Richardson algorithm has been used to deconvolve the telescope PSF (Burrows et al. 2000).

Zoom In Zoom Out Reset image size

One of the most striking examples of the power of high‐resolution X‐ray imaging is in the spectacular Chandra images of globular clusters. Figure 17 is the moderately deep exposure (70 ks) ACIS‐I image recently published by Grindlay et al. (2001b) of 47 Tuc. The top panel of this "true" color X‐ray image (composed of red/green/blue images derived from counts recorded in soft [0.5–1.2 keV], medium [1.2–2 keV], and hard [2–6 keV] bands) shows the central 25 × 2^' of the cluster, or approximately central three core radii. The enlargement at the bottom of the figure is ∼30 arcsec square, and thus the central ∼0.7 core radius portion of the cluster. Some 108 sources are detected in the field excluding the central core, with L_X>10³⁰ ergs s⁻¹. Another greater than 100 sources are likely present in the central core.

Zoom In Zoom Out Reset image size

The image, associated spectra, and measured time variability reveal more about the binary content and stellar, as well as dynamical, evolution of a globular cluster than achieved with all previous X‐ray observations of globulars combined (and even, arguably, many HST observations as well). All 16 of the millisecond pulsars (MSPs) recently located from precise pulse timing (Freire et al. 2001) are detected (Fig. 17, circles). Their X‐ray spectral properties (colors) versus radio pulsation spin‐down measures show them to obey a significantly different L_X‐Ė relation than for MSPs in the field (Grindlay et al. 2001a), and 50–100 MSPs are likely detected in the cluster.

The second most abundant X‐ray source population in 47 Tuc is that of the long‐sought accreting white dwarfs, or cataclysmic variables (CVs), marked as squares in Figure 17 for those already identified in deep HST images. Many other (primarily blue or whitish color) candidates are present, so that perhaps a third of the Chandra sources are CVs. A third significant population of X‐ray binaries was discovered (and unanticipated) in this image: main‐sequence star binaries (detached), or so‐called BY Draconis stars (the brightest few of which, detected as flaring sources, are marked with triangles).

The observations of other globular clusters with Chandra are beginning to be published, with a recent example being the nearby core‐collapsed cluster NGC 6397 (Grindlay et al. 2001c). This cluster shows a dramatic contrast with 47 Tuc: although almost as abundant in CVs, it is nearly devoid of MSPs. Grindlay et al. (2001c) note that this suggests fundamental differences in the relative neutron star versus white dwarf content, as well as compact binary formation history. Clearly, the high‐resolution X‐ray view made possible with Chandra is opening a new era in understanding these oldest, and dynamically most interesting, stellar systems.

In addition to mapping the structure of extended sources and the diffuse emission in galaxies, the high angular resolution permits studies of ensembles of discrete sources, which would otherwise be impossible owing to source confusion. A beautiful example comes from the observations of the center of M31 (Fig. 18) performed by Garcia et al. (2000; see also M. R. Garcia et al 2002, in preparation). The image shows what used to be considered as emission associated with the black hole at the center of the galaxy now resolved into five distinct objects. A most interesting consequence is that the emission from the region surrounding the central black hole is unexpectedly faint relative to the mass of the central black hole, as with the Milky Way.

Zoom In Zoom Out Reset image size

M81 (NGC 3031) is an Sab spiral at a distance of approximately 3.6 Mpc. The galaxy was observed by Tennant et al. (2001b) with the S3 chip of the ACIS‐S instrument on Chandra for 50 ks (Fig. 19). Prior to this observation, the galaxy had been observed with both the Einstein and ROSAT observatories. Nine sources were detected with Einstein by Fabbiano (1988), of which five were in the spiral arms. A total of 26 sources were detected with ROSAT by Immler & Wang (2001). The Chandra observation detected the bright nucleus at a (0.2–8.0 keV) luminosity of 4 × 10⁴⁰ ergs s⁻¹. In addition, 96 other sources were detected: 81 with signal‐to‐noise ratio S/N≥3.5 and 16 with 3.0≤S/N≤3.5. Here S/N = 3 is one false detection; 3.5 is 0.1. Based on a canonical spectrum, the (0.2–8.0 keV) luminosity of the sources, including the nucleus, ranges from 3 × 10³⁶ to 4 × 10⁴⁰ ergs s⁻¹.

Zoom In Zoom Out Reset image size

There were 41 sources in the bulge of the galaxy, and, excluding the nucleus, these had a total luminosity of 1.6 × 10³⁹ ergs s⁻¹. One of these sources had a soft spectrum (T≃70 eV) and an observed luminosity in excess of 2 × 10³⁸ ergs s⁻¹, about the Eddington limit for a canonical neutron star. In addition, the bulge shows 0.8 × 10³⁹ ergs s⁻¹ of unresolved emission which follows the starlight and is not completely consistent with the extrapolation of the log N–log S curve for the galaxy, indicating diffuse emission.

There were 56 sources in the disk that were within the ACIS‐S3 field of view. A total of 21 were within ±400 pc of the spiral arms. These 21 include four of five of the soft sources and seven of the 10 brightest sources. Five of the 21 sources are near SNRs. Interestingly, 35 disk sources are not in the spiral arms and are typically fainter than those that are, leading Tennant et al. (2001b) to speculate that these are perhaps high‐mass X‐ray binaries or black hole candidates. There were no associations with any of the three known globular clusters in the S3 viewing field.

The sheer number of X‐ray sources detected by Chandra in a typical nearby galaxy makes studies of the global properties of the X‐ray source populations possible. For instance, it has been suggested by Sarazin, Irwin, & Bregman (2001), and supported in a theoretical framework by Wu et al. (2001), that log N–log S distributions can be used as a distance indicator for giant elliptical galaxies. Wu et al. (2001) further show that log N–log S distributions can be used as a probe of recent star formation and of the dynamical history of spiral galaxies. This is only a sample of the use of X‐ray data to place constraints on galaxy evolution.

The remarkable Seyfert 2 galaxy Circinus has been observed by Sambruna et al. (2001), Bauer et al. (2001), and Smith & Wilson (2001). The spectrum of the nuclear region shows a wealth of emission lines including lines of Ne, Mg, Si, S, Ar, Ca, and Fe and a very prominent Fe K line at 6.4 keV (Sambruna et al. 2001). The emission appears to be the reprocessed radiation from the obscured central source and originates within 60 pc of the object. In addition to the very detailed spectrum of the nuclear region, 16 point sources were detected (Bauer et al. 2001; Smith & Wilson 2001), several of which exhibit emission lines and are very luminous supernova candidates. One bright object has a luminosity of 3.4 × 10³⁹ ergs s⁻¹ and a strong iron emission line at 6.9 keV with an equivalent width of 1.6 keV. An eclipsing X‐ray binary was discovered with a period of 7.5 hr and a 0.5–10.0 keV luminosity of 3.7 × 10³⁹ ergs s⁻¹ assuming isotropic emission. If not local or beamed, such a high luminosity implies that the X‐ray–emitting object must have a mass substantially greater than that of a neutron star and is thus likely to be a black hole with a mass of more than 25 M_⊙. These "ultraluminous" sources appear to be quite common in nearby galaxies (see also Blanton, Sarazin, & Irwin 2001; Sarazin, Irwin, & Bregman 2000, 2001; Angelini, Lowenstein, & Mushotzky 2001; Fabbiano, Zezas, & Murray 2001). There appear to be too many of such objects to give much credence to the idea that they are all more local (or more distant) than one might think. King et al. (2001) argue that the ultraluminous sources might be beamed and discuss a link with microquasars.

Another unique application of the excellent imaging properties of Chandra is the study of gravitational lenses, where the image separation is usually only about an arcsecond. By comparing different images it is possible to measure differential time delays of temporal changes and obtain an estimate of the Hubble constant. More than a dozen lensed systems have thus far been observed, and one is shown in Figure 20. The four lensed images of the quasar are clearly resolved, and a rapid flare was seen in one of two images (Morgan et al. 2001). Several lens systems for which the expected delay is only hours are under study. The large magnification of some of the lenses allows the study of objects at very high redshifts that would otherwise not be detectable, and the most distant (z = 1.4) X‐ray jet has recently been detected in one of the images of Q0957+561 (Chartas et al. 2002).

Zoom In Zoom Out Reset image size

Chandra observations frequently exhibit structures with characteristic angular scales of a few arcseconds in clusters of galaxies which previously were believed to be simple systems. Two of the more important types of results involve investigations of the interactions between radio sources and the hot cluster gas in some clusters and the existence and implications of cold fronts in others.

The Hydra A radio galaxy (3C 218) at a redshift of z = 0.052 is associated with a relatively poor cluster of galaxies. This cluster was observed during the orbital verification and activation phase of the observatory; these very early images contained large areas of low X‐ray surface brightness, indicating low‐density regions or cavities in the intracluster gas. These cavities and other aspects of the X‐ray emission now have been studied by a number of authors (McNamara et al. 2000; David et al. 2001; Nulsen et al. 2001). Similar but more dramatic cavities are found in the Perseus Cluster (Fabian et al. 2000); this image is shown in Figure 21.

Zoom In Zoom Out Reset image size

Briefly, the Hydra A cavities are found to coincide with the emission lobes of the radio source. The overall temperature of the cluster gas increases from ∼3 keV in the central 10 kpc to ∼4 keV at a radius of 200 kpc and then gradually decreases to ∼3 keV toward the radial limit of the Chandra observation at about 300 kpc. However, the cavities are surrounded by bright rims of enhanced X‐ray emission which are cooler than the cluster gas away from the cavities at comparable radii; this shows that the cavities are not created by expanding radio lobes which shock the surrounding medium, as expected in some earlier models of these sources (e.g., Heinz, Reynolds, & Begelman 1998).

The cavities probably are in local pressure equilibrium with their surroundings. The energy input required to maintain the cavities is comparable to the current radio emission and also to that required to substantially inhibit a cooling flow in the inner part of the cluster.

The Hydra A Cluster contains an unresolved X‐ray source coincident with the radio core. The point‐source X‐ray spectra are highly absorbed, which indicates that the source must be contained within the high column density material found toward the nucleus of the galaxy in VLBI radio observations by Taylor (1996). This limits the size of the X‐ray–emitting region to ≤24 pc (McNamara et al. 2000).

The angular resolution also enables more quantitative studies of the cluster merger process. The early images of the cluster A2142 (Markevitch et al. 2000) show two sharp, bow‐shaped shocklike surface brightness features; the surface brightness is discontinuous on a scale smaller than 5^''–10^''. However, a detailed investigation shows that the pressure is continuous across the boundary and that the temperature transition has the opposite sign to that expected from a shock. The most likely explanation is that these edges delineate the dense subcluster cores which have survived a merger and the associated ram pressure stripping by the surrounding gas.

The image of the cluster A3667 (Vikhlinin, Markevitch, & Murray 2001a, 2001b) contains a well‐defined subcluster. The pressure jump across the boundary of the subcluster is approximately a factor of 2, which indicates that the subcluster relative velocity is about equal to the sound speed of the surrounding medium. The transition width in this case is 35 or less, smaller than the Coulomb mean free path of electrons and protons on either side of the front. A model of the magnetic field (Vikhlinin et al. 2001a) needed to suppress the particle diffusion requires a field strength of order 10 μG. These and similar studies enabled by the good angular resolution of the observatory should lead to significant improvements in our understanding of cluster merger phenomena.

The first sounding rocket flight that detected the brightest X‐ray source in the sky, other than the Sun, also detected a general background of X‐radiation (Giacconi et al. 1962). The nature of the background radiation has been a puzzle for nearly 40 yr, although the lack of distortion of the spectrum of the cosmic microwave background places a strong upper limit to the possibility of a truly diffuse component (less than 3%; Mather et al. 1990). Observations with ROSAT at energies below 2 keV made a major step in resolving a significant fraction (70%–80%) into discrete objects (Hasinger et al. 1998) and found that the sources reside mainly in active galactic nuclei (AGNs) at redshifts from 0.1 to 3.5. ASCA satellite observations extended the search for sources in the 2–10 keV band, resolving about 30% into mainly AGNs (Ueda et al. 1998). Observations with BeppoSAX have continued these studies. Currently two 1 Ms exposures have been accomplished with Chandra: the Chandra Deep Field–North (Fig. 22; Brandt et al. 2001b) and Chandra Deep Field–South (Giacconi et al. 2001). These surveys extend the study of the background to flux levels more than an order of magnitude fainter in the 0.5–2.0 keV band and resolve over 90% of the background into a variety of discrete sources. The largest uncertainty in establishing the fraction is now in the knowledge of the total level of the background itself.

Zoom In Zoom Out Reset image size

The spectrum of the X‐ray background has been called a "spectral paradox" by Boldt (1987) because the majority of the bright AGNs are found to have a photon index larger than that for the background itself (1.7 vs. 1.4). One of the purposes of the Chandra Deep Surveys has been to explore the spectra of the sources. Since it must be the faint sources that modify the spectrum from the average of the bright AGN sample, the faint source spectrum was estimated by adding the spectra of individual sources detected in the surveys. Garmire et al. (2001) found the spectral index to decrease from 1.8 to 1.0 as the flux of the sources decreased from 1 × 10^-14 to 2 × 10^-15 ergs cm⁻² s⁻¹. Similar results are found in the data from the southern field. The spectrum of all of the sources has a slope very near to 1.4 (Tozzi et al. 2001; Garmire et al. 2001), consistent with the wide field‐of‐view measurements made with detectors that could not resolve the fainter sources.

In each of the Chandra Surveys about 350 sources were detected. The flux levels attained in the soft and hard bands were 3 × 10^-17 and 2 × 10^-16 ergs cm⁻² s⁻¹, respectively. The highest redshift detected (so far) is a QSO at 5.2. A highly obscured type 2 QSO at z = 3.4 has been reported in the Chandra Deep Field–South by Norman et al. (2001).

Since these fields were selected not to have any nearby galaxies in them, the nearest galaxies detected are at redshifts of ∼0.08. In one of these galaxies, located in the HDF‐N, the X‐ray emission appears to be nonnuclear, based on the offset from the center of the optical light distribution, perhaps implying on the basis of the high X‐ray luminosity that intermediate‐mass black hole candidates are present (Hornschemeier et al. 2001b).

The log N–log S function has a change in slope at just over 1 × 10^-14 ergs cm⁻² s⁻¹ for the 2–10 keV band and at ∼5 × 10^-15 ergs cm⁻² s⁻¹ for the 0.5–2.0 keV band. A nominal slope of −1.5 would be expected for a Euclidean geometry populated with a uniform distribution of sources. In the hard band, the slope flattens to −1 at about 1.4 × 10^-15 ergs cm⁻² s⁻¹ and then becomes even flatter at lower fluxes (Fig. 23). In the soft band the slope flattens from −1.5 to −0.65 down to 1 × 10^-16 ergs cm⁻² s⁻¹ and then becomes even flatter for fainter fluxes (see Fig. 24). Eventually, these curves should change to steeper slopes once the population of galaxies is reached. Normal galaxies will not contribute very much to the total X‐ray background flux, since they are so faint, but will ultimately become the major contributor at the lowest flux levels (Hornschemeier et al. 2001a).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The majority of the X‐ray sources beyond a redshift of 0.5 are AGNs or QSOs. Barger et al. (2001) found, on the basis of the luminosities of the hard‐band X‐ray sources, that the accretion rate onto black holes grows linearly with redshift to a redshift of 1 and then flattens with only a slight increase out to a redshift of 3. The volume density of black hole accretion is found to increase as (1 + z)³.

The moderately rapid response for targets of opportunity has made possible the study of the afterglows of gamma‐ray bursts with the observatory. The afterglow of GRB 991216 showed the first X‐ray iron line profile indicating a very high velocity (∼0.1c) in the ejected material (Piro et al. 2000). These authors also reported a recombination edge from hydrogenic ions of iron at a redshift of 1. This observation supports a hypernova interpretation, or delayed gamma‐ray burst following a supernova (Vietri & Stella 1998; Meszaros & Rees 2001). A second gamma‐ray burst, GRB 000926, revealed an unusual X‐ray light curve that implied that the burst expanded into a dense medium (n∼10⁴ cm⁻³) and that the fireball was only moderately collimated initially, which then slowed down and became nonrelativistic after 5 days (Piro et al. 2001).

The spectra of stellar coronae obtained with Chandra contain a large number of interesting emission‐line features that serve as diagnostic tools for temperatures, densities, and emission measures. Figure 25 shows the LETGS (defined here as the LETG used with HRC‐S) spectra (5–175 Å) of Capella and Procyon, two different coronal sources. There are a large number of lines from very many different elements and also a strong temperature dependence. In Capella there are many lines around 15 Å from Fe xvii, while these lines are weak in the Procyon spectrum. Conversely, the Fe ix line at 171 Å is very prominent in the Procyon spectrum, indicating a cooler corona.

Apart from a chemical and temperature analysis, one can derive densities from many density‐sensitive lines. In the wavelength range around 100 Å of the Capella spectrum, density‐sensitive lines of highly ionized Fe ions appear, and in the short‐wavelength region (between 6 and 45 Å) of both spectra, temperature‐ and density‐sensitive lines are present. The latter originate from the He‐like ions Si xiii, Mg xi, Ne ix, O vii, N vi, and C v. For these ions the resonance line (r) 1s^2 1 S₀–1s2p ¹ P₁, the intercombination line (i) 1s^2 1 S₀–1s2p ³ P_{1, 2}, and the forbidden line (f) 1s^2 1 S₀–1s2s ³ S₁ are resolved.

From the resonance lines one can obtain an estimate of the temperature—an estimate as the flux may arise from two different regions on the stellar surface. In the He‐like ions, the ratio between the intercombination line and the forbidden line is strongly density dependent, while the ratio between the sum of the intercombination line and forbidden line and the resonance line is temperature dependent. In a low‐density plasma the forbidden line is stronger than the intercombination line, but for increasing density, the 1s2s ³ S₁ (the upper level of the forbidden transition) will be depopulated by collisions in favor of the 1s2p ³ P_{1, 2} (the upper level of the intercombination line). The resonance line intensity is comparable to the sum of the intensities of the two other lines and increases at higher temperatures. Table 1 shows the temperature for the coronae of Capella and Procyon, based on the ratio of the sum of the intercombination and forbidden lines to the resonance line and also on the ratio between lines of succeeding stages of ionization. From this table we notice that the coronae of both stars have a multitemperature structure and that the temperature of the corona of Capella is higher and extends to higher ionization stages. Table 2 shows the densities derived from the appropriate line ratios of the He‐like ions.

The X‐ray output of the bright Galactic X‐ray binaries is generally dominated by continuum from an optically thick accretion disk, but grating spectra are revealing a rich variety of absorption and emission features that carry new information about material in and around the source (e.g., Brandt & Schulz 2000; Paerels et al. 2001; Cottam et al. 2001; Marshall, Canizares, & Schulz 2002; Schulz et al. 2001). Doppler structure is often seen in the emission and/or absorption lines, with velocities ranging from hundreds of kilometers per second (e.g., for 4U 1822−37 where the lines are attributed to recombination in an X‐ray–illuminated bulge where the accretion stream hits the disk; Cottam et al. 2001) up to 0.26c for emission lines in relativistic jets of the binary SS 433 (Marshall et al. 2002). One interesting example is the mysterious binary Circinus X‐1, thought to contain a neutron star that at times radiates beyond its Eddington limit (Brandt & Schulz 2000). The HETG spectra reveal lines from H‐like and/or He‐like Ne, Mg, Si, S, and Fe. The lines exhibit broad (±2000 km s⁻¹) P Cygni profiles, with blueshifted absorption flanking redshifted emission. Examples are shown in Figure 26. Brandt & Schulz (2000) interpret these features as the X‐ray signatures of a wind being driven off the accretion disk, making Cir X‐1 the X‐ray analog of a broad absorption line quasar.

Zoom In Zoom Out Reset image size

Line ratios are being used to constrain physical properties of the emitting regions, and time variability of intensities and Doppler structures helps to locate the source of the lines in the binary system. The HETG spectrum of the ultracompact binary 4U 1626−67 shows unexpectedly strong photoelectric absorption edges of Ne and O, most probably from cool, metal‐rich material local to the source (Schulz et al. 2001). The anomalous abundances led the authors to suggest that the mass donor in this binary is the chemically fractionated core of a C‐O‐Ne or O‐Ne‐Mg white dwarf.

High‐resolution spectra of AGNs, especially Seyfert galaxies, are providing extraordinary new details about the physical and dynamical properties of material surrounding the active nucleus. In the case of Seyfert 1 galaxies, whose signal is dominated by the bright X‐ray continuum from the central engine, the partially ionized circumsource material introduces a prominent pattern of absorption lines and edges. These are the "warm absorbers" originally discovered in low‐resolution spectra by Reynolds (1997) and George et al. (1998) but now revealed in much greater detail.

For example, the LETG spectrum of NGC 5548 shown in Figure 27 exhibits three dozen absorption lines, plus a few in emission (Kaastra et al. 2000), and the HETGS (defined here as the HETG used with ACIS‐S) spectrum of NGC 3783 (Kaspi et al. 2000, 2001) has over five dozen lines. In both cases, there is evidence for bulk motions of several hundred kilometers per second. For NGC 3783, detailed modeling of an absorber with two ionization components does a remarkably good job of reproducing the observed line strengths (Kaspi et al. 2001). The HETG spectrum of MCG −6‐30‐15 (Fig. 28) also has dozens of absorption lines from a wide range of ionization states (Lee et al. 2001). In addition, an absorption feature at 0.704 keV (17.61 Å) is well fitted by the neutral Fe L3 absorption edge (and associated resonant structure) with a column density equal to the amount of line‐of‐sight dust deduced from earlier reddening studies. (Alternatively, Branduardi‐Raymond et al. 2001 attribute this feature to relativistically broadened O viii emission, based on their XMM‐Newton RGS spectrum.) The HETG also has been used to (marginally) resolve a narrow component of the Fe Kα line in NGC 5548 (Yaqoob 2001).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

For Seyfert 2 galaxies the strong continuum from the central engine is not seen directly, so the surrounding photoionized regions are seen in emission. The HETG spectra of Mrk 3 (Sako et al. 2000) and NGC 4151 (classified as Seyfert 1.5 but observed when the continuum was exceptionally faint; Ogle et al. 2000) bristle with emission lines, whose ratios provide diagnostics of the conditions in the emitting clouds (Fig. 28). There are clear signatures of photoionization, such as the relatively strong forbidden line from He‐like ions and narrow features from free‐bound radiative recombination. Other diagnostics suggested the presence of a smaller amount of collisionally excited plasma (Ogle et al. 2000), although more recent observations and modeling indicate that photoexcitation together with the photoionization and recombination can explain the line ratios very well (Sako et al. 2000).

Many AGN spectra show primarily a strong, power‐law continuum with little or no evidence for any absorption or emission lines. Examples include the narrow‐line Seyfert 1 galaxy Ton S 180 (Turner et al. 2001), BL Lac objects, and radio‐loud quasars (Fang et al. 2001).

Since the X‐ray output of a young SNR is dominated by a moderate number of strong emission lines, the dispersed spectrum resembles a spectroheliogram, showing multiple images of the remnant in the light of individual lines. Such is the case for 1E 0102−72, an SNR in the SMC estimated to be ∼1000 yr old (Fig. 29; Canizares et al. 2001; Flanagan et al. 2001). The monochromatic images are dominated by emission from the shocked stellar ejecta. For a given element, the images for the He‐like resonance are systematically smaller than those for the H‐like Lyα line, which is graphic evidence for the progression of the so‐called reverse shock backward into the expanding ejecta (also seen in the ACIS image by Gaetz et al. 2000). Doppler shifts of approximately ±2000 km s⁻¹ are measured by comparing the strongest monochromatic images from the plus and minus orders. The velocities appear asymmetric, suggesting that the shock‐heated ejecta fill a toroidal region inclined to the line of sight.

Zoom In Zoom Out Reset image size