Chandra Discovery of a 100 kiloparsec X-Ray Jet in PKS 0637–752

The quasar PKS 0637-752, the first celestial X-ray target of the Chandra X-Ray Observatory, has revealed asymmetric X-ray structure extending from 3'' to 12'' west of the quasar, coincident with the inner portion of the jet previously detected in a 4.8 GHz radio image (Tingay et al. 1998). At a redshift of z = 0.651, the jet is the largest (≳100 kpc in the plane of the sky) and most luminous (~1044.6 ergs s-1) of the few so far detected in X-rays. This Letter presents a high-resolution X-ray image of the jet, from 42 ks of data when PKS 0637-752 was on-axis and ACIS-S was near the optimum focus. For the inner portion of the radio jet, the X-ray morphology closely matches that of new Australian Telescope Compact Array radio images at 4.8 and 8.6 GHz. Observations of the parsec-scale core using the very long baseline interferometry space observatory program mission show structure aligned with the X-ray jet, placing important constraints on the X-ray source models. Hubble Space Telescope images show that there are three small knots coincident with the peak radio and X-ray emission. Two of these are resolved, which we use to estimate the sizes of the X-ray and radio knots. The outer portion of the radio jet and a radio component to the east show no X-ray emission to a limit of about 100 times lower flux. The X-ray emission is difficult to explain with models that successfully account for extranuclear X-ray/radio structures in other active galaxies. We think the most plausible is a synchrotron self-Compton model, but this would imply extreme departures from the conventional minimum energy and/or homogeneity assumptions. We also rule out synchrotron or thermal bremsstrahlung models for the jet X-rays, unless multicomponent or ad hoc geometries are invoked.


INTRODUCTION
PKS 0637Ϫ752 was the first celestial X-ray target of the Chandra X-ray Observatory (Weisskopf et al. 2000).As a moderate strength point source it was used to locate the optical axis and focus of the X-ray mirror assembly.Surprisingly, even the first short, out-of-focus image clearly revealed an X-ray jet, coincident with the radio jet reported by Tingay et al. (1998).This Letter addresses the X-ray jet, which appears as a extension from 3Љ to 11Љ .5 west of the quasar, with brighter condensations from 7Љ .5 to 9Љ .5.This corresponds 11 to an extension in the plane of the sky from ∼30 to ∼100 kpc from the nucleus, with an X-ray luminosity of ∼ ergs s Ϫ1 : the largest 44 4.2 # 10 and most luminous X-ray jet discovered to date.In a companion paper (Chartas et al. 2000) we examine the X-ray spectra of the core and jet in more detail by adding substantial additional data where the detector was slightly out of focus.
PKS 0637Ϫ752 was identified with a pointlike object by Hunstead (1971), based on an accurate radio position, and a redshift of was measured by Savage, Browne, & z p 0.651 Bolton (1976).The HEAO 1 all-sky survey suggested it as a 2-10 keV X-ray source (Wood et al. 1984).Definitive X-ray identification was made in the 0.3-3.5 keV band with the Einstein Observatory (Elvis & Fabbiano 1984), and numerous X-ray observations have since been made (see Yaqoob et al. 1998 and references therein); however, none approached the subarcsecond image quality of Chandra (Jerius et al. 2000) and so could not resolve the jet.The source is gamma-ray quiet (Fichtel et al. 1994), with a 2 j upper limit of photons Ϫ8 4 # 10 cm Ϫ2 s Ϫ1 for gamma rays above 100 MeV.
Very long baseline interferometry (VLBI) space observatory program (VSOP) observations were rescheduled to overlap the Chandra observations to investigate links between the known parsec-scale jet (Tingay et al. 1998) and the X-ray emission from the quasar core.During the VSOP observations we also obtained Australian Telescope Compact Array (ATCA) radio images with comparable resolution to the Chandra images.We discuss the radio data in § 2.2.Hubble Space Telescope (HST) Wide Field Planetary Camera 2 (WFPC2) images were obtained fortuitously on 1999 May 30 and are discussed in § 2.3.

X-Ray
Six Chandra observations, totaling 42 ks when the target was on-axis and the detector was within 0.25 mm of focus,  .3 4, 8, 16, 32, 64, 128, 256, 512, 1024, and mJy beam Ϫ1 , where the 2048 # 2.5 restoring beam is circular with an FWHM of 0Љ .84.Bottom: ATCA 4.8 GHz radio contours at 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, and 2048 # 3 mJy beam Ϫ1 , the fractional polarization in false color; the direction of the electric vector is given by the red lines whose lengths are proportional to the fractional polarization.The bend in the radio jet, the change in radio polarization fraction and direction, and the termination of the X-ray emission are all coincident, 9Љ .5-11Љ.5 west of the quasar.The X-ray histogram gives counts in a 0Љ .2bin, normalized to unity at 69 counts s Ϫ1 bin Ϫ1 .The triangles are 8.6 GHz flux density per beam, normalized to unity at 0.04 Jy.The dashed Gaussian curves mark the positions of knots detected in the HST optical images, but we do not actually measure the profiles.The total flux for these three knots is 0.2, 0.28, and 0.094 mJy, respectively, east to west.
were used to create an X-ray image.We use the full energy band of the back side-illuminated S3 CCD chip, ∼0.3-10 keV.
The point-spread function (PSF) of the resultant core image has a half-power radius of 0Љ .42 and a 90% encircled energy radius of 1Љ .55.The image quality of the quasar core is broadened slightly by event pileup, indicated by overrepresentation of the high grades as well as the high counting rates of ∼0.4 s Ϫ1 .In the direction perpendicular to the jet, the core fits a Gaussian profile with an rms of 0Љ .37.The jet is faint enough that pileup is not significant at any point.
The X-ray and radio images are shown in Figure 1.The X-ray jet, which corresponds to the inner portion of the western radio jet before the bend to the northwest, separates into two regions: within 7Љ of the core the X-ray flux is somewhat faint and maintains a position angle (P.A.) of Ϫ82Њ; at about 7Љ from the core the jet bends distinctly to the south to a P.A. of Ϫ86Њ, and the flux is about 3 times greater.The increased brightening may be dominated by three pointlike knots, as detected with HST and shown as the plus signs.In the brighter parts of the jet, a fit to the cross-jet profile is consistent with the core PSF; thus, the jet is unresolved in this direction, with an intrinsic width less than 0Љ .3FWHM.
We detect a net 1205 X-ray counts from the entire inner, western jet, defined as a rectangle from 3Љ to 11Љ .5 west and from 0Љ .5 south to 2Љ north of the quasar.For the power-law energy index 12 of (Chartas et al. 2000), this a p 0.85 ‫ע‬ 0.08 corresponds to a measured 2-10 keV flux of ergs Ϫ13 1.2 # 10 cm Ϫ2 s Ϫ1 , or a spectral normalization at 1 keV of 5.9 # ergs cm Ϫ2 s Ϫ1 keV Ϫ1 .That is, the flux density is 25 nJy Figure 2 plots the X-ray and radio surface brightness as a 12 We use .
Ϫa f ∝ n n function of the distance west of the quasar core, integrated ‫1ע‬Љ perpendicular to the jet.There are three distinct, partially resolved radio peaks or knots in the brighter part of the inner western jet, which we designate by their direction and distance from the core using the notation WK7.8, WK8.9, and WK9.7.
The enhanced X-ray emission is closely associated with those knots but not in exact detail.In particular, the X-ray flux falls off more steeply past 9Љ .5.The ratio of the flux densities at 1 keV to 8.6 GHz varies by no more than a factor of 2 from the value in the region ∼4Љ-9Љ .5 west of the quasar Ϫ7 1.3 # 10 core.

Radio
PKS 0637Ϫ752 is the subject of ongoing VSOP (Hirabayashi et al. 1998) observations to monitor its parsec-scale evolution (Tingay et al. 2000), with one of the VSOP observations being rescheduled to overlap the first Chandra observation.ATCA observations were scheduled in parallel with the VSOP observations to provide 1Љ and 2Љ resolution radio images at 8.6 and 4.8 GHz, respectively, to complement the Chandra images.
These ATCA images confirmed the remarkable coincidence of the radio and X-ray jets.Subsequent observations were made to improve image sensitivity and to determine the jet polarization.The inner west radio jet is optically thin with a spectral index of 0.81, and the polarization is 10%-20%.The E-vectors are perpendicular to the jet where X-rays are detected, but as the X-ray flux decreases near the bend in the radio jet, the polarization P.A. begins to change so that the E-vectors become parallel to the jet's center line for the remainder of the radio jet.
We have reanalyzed our VLBI observations (Lovell et al. 2000;Chartas et al. 2000) to search for the presence of more compact components within the radio knots.We find that the knots are indeed resolved at 0Љ .05resolution with no more than 5 mJy, about 14% remaining at this resolution.This suggests that the radio knots are most likely low surface brightness "hot spots." A rotation measure image was constructed from the polar-ization data at the two frequencies.Faraday rotation was detected in the quasar core ( rad m Ϫ2 ) but not in the RM p 80 jet, with an upper limit of ‫03ע‬ rad m Ϫ2 .The absence of significant Faraday rotation implies that the intrinsic magnetic field in the jet is perpendicular to the observed E-vectors in Figure 2.That is, the magnetic field is longitudinal where the Xray emission is strong and perpendicular to the jet where we do not detect X-ray emission.
The VSOP observations show that PKS 0637Ϫ752 displays apparent superluminal motion in its parsec-scale jet, co-aligned with the inner portion of the arcsecond-scale radio jet (Lovell et al. 2000).Six data sets covering 1995-1999 (four from the US Naval Observatory's geodetic VLBI database [Piner & Kingham 1998], two from VSOP) allow the motion of three parsec-scale features to be measured.Linear least-squares fits to the separation of these features from the VLBI core versus time yield proper motions of , , and 0.41 ‫ע‬ 0.03 0.29 ‫ע‬ 0.05 mas yr Ϫ1 from the outermost component inward 0.36 ‫ע‬ 0.09 (Lovell et al. 2000).These proper motions and associated errors are just consistent with all three components moving at the weighted average proper motion of mas yr Ϫ1 , 0.36 ‫ע‬ 0.02 which corresponds to an apparent speed of .Since 17.8c ‫ע‬ 1.0c the apparent transverse speed is given by b p b sin v/(1 Ϫ obs (Rees 1966), this apparent speed places limits on the b cos v) bulk Lorentz factor in the VLBI jet and the angle of the VLBI jet to the line of sight of and , respectively.G 1 17.8 v !6Њ .4Since the parsec-scale and arcsecond-scale radio jets are very well aligned, unless the jet goes through a large bend in a plane perpendicular to the plane of sky the actual length of the Xray jet would be at least 940 kpc.

Optical
An optical image of PKS 0637Ϫ752 was obtained with the HST WFPC2 using the F702W filter.Three observations were combined in order to eliminate cosmic rays, giving a total exposure time of 2100 s.Three distinct knots are detected within the radio image contours 7Љ .7,8Љ .8, and 9Љ .6 west of the core (Fig. 1).These have fluxes 0.2, 0.28, and 0.094 mJy, respectively, at an effective frequency Hz (6969 A ˚).The quasar image is saturated, so it is difficult to detect sources within about 2Љ of the quasar core.We estimate that the knots can be located relative to the quasar to an accuracy of about 0Љ .07.The HST image also shows the presence of a group of faint galaxies, ∼100 kpc in radius, surrounding the quasar.

DISCUSSION
We will discuss the X-ray emission specifically from the knots 7Љ .8 and 8Љ .9west of the nucleus, which contain most of the luminosity of the inner western jet and which are coincident with radio knots and the optical emission detected by HST.We consider in turn synchrotron, thermal bremsstrahlung, and inverse Compton for the X-ray emission mechanism.

Synchrotron
The strong polarization shows that the radio emission arises from synchrotron radiation.Standard models assume a powerlaw population of relativistic electrons with a density , where g is the Lorentz factor.The radio spectral Ϫm n(g) p n g 0 index is observed to be 0.81, so that a p (m Ϫ 1)/2 m p . A natural hypothesis is that the X-rays are also synchro-2.62tron radiation from the same population of electrons.However, the optical flux falls a factor of 10 below such a continuous spectrum and therefore rules out such a simple model.The absence of optical emission at the level of 5.6 mJy, which would be required, implies that the high-energy cutoff to the electron spectrum be such as to cause the radio spectrum to steepen at Hz.For the X-rays to result from synchrotron 12 n ! 3 # 10 emission there would have to be an independent population of electrons with a similar index in the energy region m p 2.7 emitting the X-rays but that flattened at lower g to avoid overproducing the optical.There then would be no apparent reason for the existing spatial correlation of the radio and X-ray emission.

Thermal Bremsstrahlung
Although the data allow a fit to a thermal spectrum providing kT is at least 4 keV (Chartas et al. 2000), thermal bremsstrahlung is not a viable origin for the X-ray emission unless a contrived geometry is invoked.Taking an upper limit size of 0Љ .4 for the diameter of a cylindrical jet, it would require an electron density of cm Ϫ3 to produce the measured lu-n p 2 e minosity.But the upper limit to the rotation measure places a limit /(HL).Even if the magnetic field were as Ϫ5 n !3.7 # 10 e low as 1 mG, a path length of only pc through such a L p 20 thermal plasma would exceed the rotation measure limit.

Synchrotron Self-Compton
The obvious remaining mechanism is inverse Compton, which can successfully explain the Chandra observations of hot spots in 3C 295 (Harris et al. 2000).Because we have measured the X-ray spectral index to be nearly the same as the 4.8-8.6GHz radio index, it is natural to assume that both arise from the same population of relativistic electrons.In inverse Compton scattering scenarios the synchrotron radio emission is typically produced by the higher energy electrons while the X-rays are produced by lower energy electrons.It is then natural to have high-energy and low-energy cutoffs such that the optical emission is no larger than observed.In particular, we expect electrons of to scatter radio seed photons at 4 g ∼ 10 ∼10 GHz up to the Chandra X-ray energy range.The secondorder scattering would already be limited by the Klein-Nishina cross section, and we would have no "Compton Catastrophe." To estimate the expected X-ray flux, we first apply standard synchrotron theory (e.g., Miley 1980) to estimate the magnetic field H and particle density n 0 , giving a minimum total energy.For example, if WK7.8 is a sphere of 0Љ .3diameter, uniformly filled with magnetic field and electrons and with no proton component, then mG and the total minimum energy H Ӎ 320 in that sphere is ergs.We have assumed the 57 U Ӎ 1.5 # 10 min radio spectrum extends from 10 MHz up to 500 GHz.In this situation, the predicted X-ray flux is ∼300 times less than observed.Such a model would satisfactorily explain the lack of X-ray emission beyond the point where the radio jet bends toward the northwest.The radio synchrotron photons would have an energy density ergs cm Ϫ3 in the knot if we assume the radio spec-Ϫ11 8 # 10 trum extends up to 500 GHz.By comparison, the minimum energy magnetic field calculated above would have an energy density ergs cm Ϫ3 .In order that there 2 Ϫ9 H /(8p) p 3.8 # 10 be 300 times more relativistic electrons, and to predict the same GHz radio flux, we must assume the magnetic field is ∼6 mG.
To balance the radio-to-X-ray flux ratios from the entire inner western jet to within a factor of 2 as is observed then requires the apparently smooth jet to be composed primarily of many radio knots, with the particle and field densities delicately balanced to produce the X-rays.Such a gross departure from equipartition, increasing the total energy by a factor of ∼1000, would pose significant problems for models of the particle acceleration and the jet confinement.
We can make any inverse Compton scenario more realistic by assuming that the magnetic field strength need not be constant throughout the volume.The exact distribution of field strength with volume would be nonunique.We must assume that the relativistic particles throughout the entire volume produce the X-rays, while a small fraction of the jet containing high magnetic fields provides essentially all of the radio emission.Such inhomogeneity and/or departure from equipartion may be plausible when we note the bend in the radio jet, the dramatic change in radio polarization, and the decrease in the X-ray flux, all coincident in the region 9Љ .5-11Љ.5 west of the nucleus.These may be indicative of shock activity causing particle acceleration.

Alternative Photon Sources
Alternate inverse Compton models could use some source of unseen photons.The equivalent luminosity of these photons in WK7.8 would have to be ergs s Ϫ1 .If the core is 45 4 # 10 the source of this luminosity, it must radiate at an unreasonable value of ergs s Ϫ1 , considering that the knot subtends 49 2 # 10 a solid angle of only sr.We cannot rule out that a Ϫ4 1.9 # 10 Doppler-boosted beam is shining on the knots but out of our direct line of sight (Perez-Fournon 1985).This beam would presumably be optical/IR emission.However, if this could happen then when we did fall in such a beam we would infer optical luminosities of տ10 49 ergs s Ϫ1 for such a source.Such objects are not observed.

Relativistic Beaming
We might invoke relativistic beaming of the jet and knots to reconcile the observed X-ray flux with the apparent equipartition magnetic field.The conventional formula (Jones, O'Dell, & Stein 1974) for the ratio of synchrotron self-Compton (SSC) X-ray-to-radio flux is multiplied by a factor where is the beaming factor of ] matter moving with bulk Lorentz factor G at an angle v to the line of sight (see Madejski & Schwartz 1983).We can explain the absence of X-rays from a putative eastern jet by Doppler suppression if , and then a value would reconcile G p 8 d p 0.3 a minimum energy radio source with the observed X-rays being produced by SSC.This would require the X-ray jet to be at an angle of 53Њ to our line of sight.Since the VLBI jet must be at an angle less than about 6Њ, and since the VLBI jet and 100 kpc jet appear to be in the same direction, the required bending would have to occur about an axis that is very nearly in the plane of the sky, an unlikely coincidence.Even with such a coincidence, we note that the apparent radio luminosity of ergs s Ϫ1 would actually be a factor 43 Ϫ(2ϩa) 3 # 10 d p 25 higher in the rest frame, and for similar sources beamed toward us we would infer a radio luminosity of ergs s Ϫ1 .46 2 # 10 This exceeds observed blazar radio luminosities.

CONCLUSIONS
Discovery of the largest and most luminous X-ray jet in the very first celestial X-ray target of Chandra has dramatically proven the power of this observatory.It immediately shows the value of the two-dimensional angular resolution improvement of a factor of 100 over the best previous missions.It is likely that many further X-ray jets will be detected in extragalactic radio sources.The X-rays place difficult constraints on the physical conditions, eliminating standard scenarios, and will surely have important astrophysical consequences, e.g., in terms of understanding regions of particle acceleration, inhomogeneities of magnetic field structures, and/or extreme departures from equipartition conditions.While it seems that inverse Compton emission from the same electrons producing the synchrotron radio emission is the most plausible source of the X-ray emission, we have noted difficulties with the specific scenarios we have considered.

Fig. 2 .
Fig.2.-Radio and X-ray profiles of the large-scale jet in PKS 0637Ϫ752.The X-ray histogram gives counts in a 0Љ .2bin, normalized to unity at 69 counts s Ϫ1 bin Ϫ1 .The triangles are 8.6 GHz flux density per beam, normalized to unity at 0.04 Jy.The dashed Gaussian curves mark the positions of knots detected in the HST optical images, but we do not actually measure the profiles.The total flux for these three knots is 0.2, 0.28, and 0.094 mJy, respectively, east to west.
effective frequency.The knot at 7Љ .7 is not resolved, while the other two are about 0Љ .3 in diameter.