At energies above 30 MeV, detection of cosmic Y-rays becomes significantly easier. The detection principle is based on conversion of the primary photon to an electron-positron pair, and on subsequent measurements of the tracks of secondary electrons with tracking detectors and their energy with a total-absorption calorimeter. This technique, originally developed for particle accelerator experiments, has a great potential for cosmic Y-ray studies. It allows reconstruction of the arrival direction and energy of primary Y-rays on an event-by-event basis. The energy resolution is basically determined by fluctuations in the electromagnetic cascade that develops in the calorimeter, as well as, especially at higher energies, by the absorbing capability of the calorimeter. The best energy resolution is achieved at GeV energies because of relatively small fluctuations and, at the same time, due to high efficiency of the total confinement of the cascade products in the calorimeter. The energy resolution could be as good as a few percent, although it would require quite massive calorimeter. Below several 100 MeV, the energy of the electron and positron can be determined also by analysing the multiple Coulomb scattering angles in converters. However, generally this is not considered as a prime priority in current designs of Y-ray telescopes. In fact, the multiple Coulomb scattering of electrons significantly limits the accuracy of determination of the arrival direction of primary photons. Therefore, as a basic element, in high energy Y-ray telescopes a multi-layer tracker with thin converters is used. The thickness and number of converters is determined from the trade-off between the overall efficiency of photon conversion and minimisation of the Coulomb scattering effect. Actually, there is another effect that limits the angular resolution. It is induced by the "invisible" recoil momentum at pair production that prevents full (unambiguous) reconstruction of kinematics of the process, and thus significantly limits the angular resolution of telescopes, especially at sub-GeV energies. The uncertainty in the angular resolution induced by this intrinsic effect is estimated
.
Angular resolution is one of the key parameters characterising performance of high energy detectors. In addition to accurate determination of the size and location of a Y-ray source, good angular resolution improves, to a certain extent, the minimum detectable flux (sensitivity) from point sources through reduction of Y-ray backgrounds – both of local and astronomical (diffuse galactic and extragalactic) origins. The background caused by charged cosmic rays can be removed with very high efficiency using an active anti-coincidence shield consisting of thin scintillation counters. The shield is "transparent" for Y-rays but provides an effective veto against charged particles. The flux sensitivity of a Y-ray telescope is determined by the residual background rate and the effective detection area – the physical area of the detector multiplied by the (energy-dependent) detection efficiency. In the space-based experiments the physical area of telescopes is limited, and cannot significantly exceed several m2.
The first meaningful observational results of Y-ray astronomy appeared in the 1970s, basically due to two successful space missions called SAS-2 (Fichtel et al., 1975) and COS B (e.g. Bignami and Hermsen, 1983). Four point sources were detected by SAS-2, including the Crab and Vela pulsars, as well as the X-ray binary source Cyg X-3, although the identification of the excess emission with this object remains rather controversial. The origin of the fourth source which later was called Geminga (Bignami et al., 1983), remained a mystery over almost 20 years until it was identified with a X-ray pulsar.
The COS-B mission increased the number of Y-ray sources to 25, although most of these sources were not identified. An undisputed success of COS-B was a detailed study of the spatial distribution of the galactic diffuse Y-ray emission and discovery of several "hot spots" in some active star formation regions and giant molecular clouds like the Orion complex. Also, COS-B discovered the first extragalactic Y-ray source which was promptly identified with the quasar 3C 273 – the first representative of the most populous high-energy Y-ray source population associated with blazars, as revealed by EGRET a decade later.
EGRET, as a part of the Compton GRO mission, provided a deep study of the high-energy Y-ray sky during nine very successful and exciting years, from 1991 to 2000. It is often said that EGRET brought gamma-ray astronomy to maturity. This instrument with an effective energy threshold around 50 MeV and FoV of about 0.5 sr, had best performance at energies around 1 GeV: (i) effective detection area ~ 0.1 m2, (ii) angular resolution ~ 1.5°, (iii) energy resolution ~ 10%, (iv) minimum detectable photon flux
or energy flux
from a point source during a 1 year all-sky survey.
The current list of high-energy Y-ray sources released by the EGRET team in the form of the 3rd EGRET Catalog (Hartman et al., 1999) consists of 271 sources detected above 100 MeV. The catalog includes 66 high-confidence and 27 lower confidence identifications with AGN, five pulsars, a nearby dwarf galaxy – the Large Magellanic Cloud, a nearby radiogalaxy – Centaurus A, as well as a single Solar flare detected in 1991. Finally, the catalog contains 170 sources not yet firmly identified with known objects. After the release of the 3rd EGRET Catalog, a number of papers have been published suggesting possible identifications of many Y-ray sources with individual objects representing several source populations – pulsars, supernova remnants, microquasars, molecular clouds, plerions, clusters of galaxies, etc. The distribution of high energy Y-ray sources from the the 3rd EGRET Catalog in galactic coordinates is shown in Fig. 1.1.
Along with discrete sources, Y-rays of diffuse origin, i.e. photons produced by interactions of cosmic rays with ambient gas and photon fields, are expected. Actually, diffuse emission from the galactic disk dominates over the contribution of resolved sources. After removal of all, identified and unidentified, objects the diffuse emission appears with several interesting spatial and spectral features which reflect distributions of cosmic rays and the interstellar gas in the galactic disk. The fluxes of diffuse gamma-radiation from the inner Galaxy detected by EGRET between 20 MeV and 10 GeV are shown in Fig. 4.20, together with fluxes at lower energies measured by COMPTEL and by OSSE.
Diffuse high energy Y-ray emission is observed also at large galactic latitudes. The bulk of the detected flux is believed to be of extragalactic origin. Almost surely, a significant fraction of this component comes from superposition of faint unresolved Y-ray sources, first of all from blazers, and possibly also from more extended structures like galaxy clusters. There may also be significant contributions from electromagnetic cascades in the intergalactic medium triggered by interactions of very high energy Y-rays from discrete sources with extragalactic photon fields.In either case, the diffuse extragalactic radiation detected by EGRET up to 100 GeV shown in Fig. 1.3 contains unique cosmological information.
Below we briefly discuss three major high-energy source populations detected by EGRET.
GeV blazars
Almost all AGN detected by EGRET belong to the blazar population -objects characterised by nonthermal continuum emission with high radio and optical polarizations and short timescale variability observed at all wavelengths. In addition, EGRET has found evidence of Y-ray emission from Centaurus A, a nearby radiogalaxy also detected by COMPTEL at MeV energies.
The main contributors to the EGRET list of blazars (see e.g. Mukherjee et al., 1997) reported with a high degree of confidence (at least 4a detection for high galactic latitudes, and 5a detection for |b| < 10°) are the so-called flat-spectrum radio quasars (FSRQs). Sixteen objects are identified with BL Lac objects which are characterised by stronger polarization and weaker optical lines than FSRQs. BL Lac objects are relatively closer and have lower luminosities than FSRQs.
A significant fraction of GeV blazars, in particular 3C 273 and 3C 279, exhibit apparent superluminal motion (Jorstad et al., 2001) detected by VLBI radio observations. Many EGRET blazars show variability on timescales of months. For many EGRET blazars, the study of short timescale variability is limited by the Y-ray photon statistics. Nevertheless, short flares on timescales less than 10 h have been detected from very strong objects like PKS 1622-297 and 3C 279.
The spectra of EGRET blazars are well-described by a simple power-law over the energy region from 30 MeV to 10 GeV, with an average photon index ~ 2.2. There is no evidence for spectral cutoffs, at least at energies below 10 GeV. Also, there is no apparent correlation between the photon indices and the source redshift (distance), despite a very strong luminosity-redshift correlation, especially at low z. In Fig. 2.4 the spectra of two EGRET blazars representing the FSRQ (3C 279 at z=0.538) and BL Lac (1219+285 at z=0.102) AGN populations are shown. Remarkably, despite almost 3 order of magnitude difference in apparent Y-ray luminosities, the spectra of these object are quite similar.
The strongest GeV Blazars do not show TeV emission. Only three EGRET blazars have been detected at TeV energies – Mkn 421, Mkn 501, and PKS 2155-304, all three being low-redshift X-ray selected BL Lac objects. At the same time all these three objects are weak GeV Y-ray emitters. Actually this GeV-TeV anti-correlation agrees with expectation.
Fig. 2.4 Gamma-ray spectra of blazars 3C 279 and 1219+285 as representatives of FSRQ and BL Lac source populations.
The strong EGRET blazars are located at large distances, and therefore TeV Y-rays emitted by these objects suffer severe intergalactic absorption due to interactions with extragalactic diffuse infrared radiation.Moreover, there are other reasons for the TeV-GeV anti-correlation associated, for example, with essentially different conditions for particle acceleration, as well as for the production and absorption of Y-rays in the jets of radio-loud quasars and BL Lac objects. The significantly higher densities of infrared and optical photons in quasars not only provide effective Y-ray production in these objects through inverse Compton scattering, but also limit the maximum energy of accelerated electrons due to the same process. As a result, one may expect a strong shift of both synchrotron and inverse Compton peaks in the spectral energy distributions of powerful blazars towards lower frequencies. Such a tendency of "becoming redder" with increasing source bolometric luminosity is indeed observed,although the picture could be, of course, more complex and sophisticated compared to this simple interpretation. It is important to note in this regard that the two brightest TeV blazars, Mkn 421 and Mkn 501, have sub-luminal parsec-scale jets, in contrast to the apparently superluminal jets of majority of GeV blazars detected by EGRET (Edwards and Piner, 2002).
Unfortunately, the low TeV source statistics and low GeV photon statistics do not allow any detailed quantitative studies of the links between the TeV and GeV blazars. This issue could be properly addressed only after having more information about both source populations, especially at the intermediate energies around 100 GeV. Such information will be available in the foreseeable future with GLAST (the Gamma-ray Large Area Space Telescope) and the new generation of 100 GeV threshold Cherenkov telescope arrays.
GeV pulsars
Pulsars – single neutron stars powered by fast rotation – are effective high energy Y-ray emitters. Two prominent representatives of this source population, the Crab and Vela pulsars, were the first astronomical objects discovered in high energy Y-rays. Together with the Geminga pulsar, they are the brightest persistent Y-ray sources on the GeV sky. At least six EGRET sources (with another 3 possible candidates) are identified with pulsars. The multiwavelength spectral energy distributions of these objects are shown in Fig. 2.1. Only in the case of the Crab pulsar does the luminosity peak at sub-MeV energies. For the other EGRET pulsars the dominant power is released in the Y-ray band; in the case of PSR B1951+32 – beyond 10 GeV. Although the Y-ray spectra of all pulsars are very hard with photon index < 2, at higher energies one expects significant steepening or a spectral cutoff. In some cases this can be directly seen in the highest energy bins of EGRET data, or is implied from upper limits obtained at TeV energies. A break around 10 GeV in the Vela spectrum is clearly seen in Fig. 2.5.
Fig. 2.5 High energy spectrum of the Vela pulsar. Heavy error bars – EGRET data. Dotted line – prediction of the outer gap model (Romani, 1996), dashed line – prediction of the polar cap model (Daugherty and Harding, 1996). The error bars shown at the model curves are those expected from the 1 year survey of GLAST.
Pulsars are characterised by their so-called light curves - indicators of the time structure of emission of these astronomical clocks. The light curves of all EGRET pulsars show a double-peak structure. The Crab light curve above 100 MeV is rather similar to the light curves seen at other wavelengths, in both the pulse shape and phase. At the same time, the light curve of Vela does not resemble the radio light curve. At present, these differences, as well as some other peculiarities of GeV pulsars do not have a convincing theoretical interpretation. The current two basic concepts based on the polar cap and outer gap models, can explain certain, but not all, features of individual pulsars. Therefore perhaps these two models should be taken as only a first step in the direction of development of a self-consistent theory of nonthermal radiation of radio pulsars.
Fig. 2.6 Light curves of four 7-ray pulsars at energies above 100 MeV and 5 GeV.
Both the poor source statistics of the Y-ray pulsar population and poor photon statistics of individual Y-ray pulsars do not allow conclusive tests in favour of, or against, the polar cap and outer gap models. In Fig. 2.5 the energy spectra predicted by these two models for the Vela pulsar are shown. While at energies below several GeV these models predict similar Y-ray spectra, both being in good agreement with EGRET measurements, at energies above 10 GeV the theoretical spectra are dramatically different. Unfortunately, large statistical uncertainties in the EGRET data in this crucial energy region do not allow us to give preference to either of these models (note that the last spectral point in Fig. 2.5 is based on only 4 detected photons). GLAST (Thompson, 2001) and future low-energy threshold ground-based instruments like 5@5 will be able to discriminate between these two models. Moreover, the statistics of detected Y-rays should be sufficient to probe the light curves at different energies. The EGRET data show evidence of change of structure of light curves with energy. The light curves of four EGRET pulsars with reasonable statistics in two energy bands, above 100 MeV and above 5 GeV, are shown in Fig. 2.6. The multi-GeV light curves are dominated by one of the two pulses seen at lower energies (Thompson, 2001). There is little doubt how crucial will be for the pulsar physics a confirmation of this trend with much better photon statistics, and an extension of these studies to even higher energies. Also, high photon statistics is a necessary condition for probes of possible irregular and regular (i .e. as a function of phase) time-variations of Y-ray spectra. While GLAST can provide adequate detection rates above 100 MeV, it might run out of > 10 GeV photons at given narrow phase intervals. Studies of the light curves at different energies, and the energy spectra at different phase intervals can be effectively covered by sub-10 GeV threshold Cherenkov telescope arrays.
The high detector sensitivity is an important factor allowing effective searches for periodic signals from Y-ray sources without relying on observations at other energy band. Because pulsars are not perfect clocks (their periods increase with time due to rotational losses), it is important to accumulate photon statistics adequate for search for periodic signals during rather short periods (say less than 1 day), thus any change of a signal’s phase can be ignored. The superior sensitivities of GLAST and 5@5 should allow to increase significantly the number of Y-ray pulsars, as well as to reveal the pulsed emission component from a number of unidentified EGRET sources, if they have indeed pulsar origin.
Unidentified EGRET sources
Almost 2/3 of sources from the 3rd EGRET Catalog are not yet identified with known astrophysical objects. Although a number of potential identifications have been suggested, the origin of the major fraction of the high energy Y-ray sources remains a mystery. Although it is quite possible that a part of these sources constitute a new class of astrophysical objects that shine mainly in Y-rays, the poor angular resolution of EGRET is likely to be the major reason for such a large fraction of unidentified objects. The main hope of solving the puzzle of unidentified EGRET sources is the GLAST mission, and perhaps also future low-energy threshold ground based Y-ray detectors. On the other hand, statistical studies of the properties of these objects, as well as continuation of multiwavelength probes of environments surrounding unidentified Y-ray sources ("source by source analysis") is an important approach to be continued during the next several years – before GLAST comes on line. The multiwavelength studies of unidentified EGRET sources will have also an important impact on preparation of the observation programs of GLAST and ground-based detectors.
The statistical studies allow differentiation of characteristics of unidentified sources that might give a hint for a possible links of the EGRET sources to certain source populations. Such a study (Gehrels et al., 2000) shows that indeed the stable (time-independent) EGRET sources are grouped in two populations when characterised by the spectral indices and the logN-logS distributions. Namely, it is found that brighter sources with harder energy spectra are concentrated at low galactic latitudes (|b| < 5°), while fainter and softer ones are located at medium galactic latitudes (5° < |b| < 30°).
A rather smooth longitude profile of about 50 low-latitude sources sets a limit to their distance of a few kpc (Grenier, 2001). Most of these sources do not show signs of variability. Although it is not possible to link these objects to a single class of galactic sources, their positions seem to correlate with objects which are believed to be tracers of active star formations regions in our Galaxy – HII regions, pulsars, SNRs, OB associations. It has been argued (Romero et al., 1999) that 22 of these sources can be associated with SNRs and 26 with OB association, with ten sources coincident with both SNRs and OB association – regions called SNOBs (Montmerle, 1979). Using the known distances of counterparts, the luminosities of Y-ray sources are estimated between 1034 — 1035 erg/s, in a good agreement with the luminosity estimates inferred from the global spatial distribution (Grenier, 2001). There is a very large dispersion in photon indices of these sources ranging from 1.7 to 3.1. At first glance, this contradicts the assumption that all these objects belong the same source population. However, such a dispersion can be readily explained, provided that Y-rays are produced by interactions of cosmic rays with molecular clouds, and that the particle accelerator is separated from the target/cloud.
Actually, the large Y-ray luminosities up to 1035 erg/s require a combination of a powerful accelerator (e.g. a relatively young SNR shell or a pulsar or a microquasar) coupled with a nearby dense gas target. On the other hand, the accelerators hardly can operate effectively in dense environments. Therefore, both the dispersion of photon indices and large luminosities provide an argument that Y-rays are produced in systems like SNRs interacting with dense molecular clouds. Motivated by this argument, Torres et al. (2003) looked at the positional coincidences between unidentified EGRET sources and the regions around SNRs, namely, "enlarging" the size of the SNR by a half degree. They found approximately 30 coincidences of this kind. Interestingly, the associations of unidentified EGRET sources with SNRs are strongest for remnants close to molecular clouds. However, it should be noticed that no supernova remnant has yet been firmly detected in high energy Y-rays. The probability for chance alignment is quite high, exceeding 0.1 per cent (Grenier, 2001).
Representatives of some other galactic source populations like colliding winds in Wolf-Rayet binaries (Benaglia and Romero, 2002), plerions (Roberts et al., 2001) and microquasars (Parades et al., 2000; Kaufman Bernado et al., 2002) have been suggested as possible counterparts of low-latitude EGRET sources. In particular, recent searches of X-ray regions by ASCA in the fields containing bright sources of GeV emission, resulted in discovery of several hard X-ray sources, presumably pulsar-driven nebulae, positionally coincident with unidentified sources of GeV Y-ray emission (Roberts et al., 2001).
Although the EGRET observations did not reveal Y-ray fluxes from the most prominent microquasars like GRS 1915+105, galactic X-ray binaries with relativistic jets have been proposed as new potential candidates. This hypothesis recently received interesting support based on the positional coincidence of the microquasar LS 5039 with an unidentified GeV source (Paredes et al., 2000). If confirmed, this would perhaps require reconsideration of a general sceptical view (see however, Mori et al., 1997 and Vestrand et al., 1997) that EGRET detections of GeV Y-rays from directions of the X-ray binaries Cyg X-3 (presumably a microquasar, or even microblazar) and Cen X-3 are results of random coincidence. Unlike shell type SNRs, Giant Molecular Clouds (GMCs) and plerions, galactic jet sources are variable objects on timescales down to 1 day or less. Therefore at present they seem to offer one of a few possibilities for interpretation of variable low-latitude unidentified y-ray sources.
Whereas there is no shortage in candidates as counterparts for low-latitude unidentified EGRET sources, away from the galactic plane the situation is reversed. The error boxes of many of unidentified high-latitude EGRET sources are "empty". Approximately half of the « 130 sources above |b| = 2.5° are variable. The large scale height of variable sources, as well as their spectral shapes indicate that this subset of high latitude sources is likely to consist largely of blazars. On the other hand, the distribution of the persistent sources closely follows the curved lane of a local (within a few 100 pc) structure at medium galactic latitudes called the Gould Belt (Gehrels et al., 2000, Grenier, 2001). Like in the galactic plane, several type of objects could be counterparts of EGRET sources. Among such sources are massive stars in OB associations with highly supersonic winds which can supply relativistic particles, e.g. through terminal shocks, for further production of Y-rays at interactions with dense ambient gas or photon fields. Pulsars born in the Gould Belt during the last 3 million years (Grenier and Perrot, 1999) are currently discussed as another promising candidate. Harding and Zhang (2001) have recently noticed that off-beam Y-ray pulsars in the Gould Belt, i.e. those viewed at large angles to the neutron star magnetic pole, can qualitatively match both the detected Y-ray fluxes and the number of EGRET sources at medium galactic latitudes. If so, GLAST and 5@5 type ground-based instruments will be able, as argued above, to detect Y-ray pulsations from most of these sources.
