Parsec-Scale Bipolar X-ray Shocks Produced by Powerful Jets from the Neutron Star Circinus X-1
P. H. Sell, S. Heinz, D. E. Calvelo, V. Tudose, P. Soleri, R. P. Fender, P. G. Jonker, N. S. Schulz, W. N. Brandt, M. A. Nowak, R. Wijnands, M. van der Klis, P. Casella
aa r X i v : . [ a s t r o - ph . H E ] A ug Draft version September 5, 2018
Preprint typeset using L A TEX style emulateapj v. 11/10/09
PARSEC-SCALE BIPOLAR X-RAY SHOCKS PRODUCED BY POWERFUL JETS FROM THE NEUTRONSTAR CIRCINUS X-1
P. H. Sell , S. Heinz , D. E. Calvelo , V. Tudose , P. Soleri , R. P. Fender , P. G. Jonker , N. S. Schulz ,W. N. Brandt , M. A. Nowak , R. Wijnands , M. van der Klis , P. Casella Draft version September 5, 2018
ABSTRACTWe report the discovery of multi-scale X-ray jets from the accreting neutron star X-ray binary,Circinus X-1. The bipolar outflows show wide opening angles and are spatially coincident with theradio jets seen in new high-resolution radio images of the region. The morphology of the emissionregions suggests that the jets from Circinus X-1 are running into a terminal shock with the interstellarmedium, as is seen in powerful radio galaxies. This and other observations indicate that the jets havea wide opening angle, suggesting that the jets are either not very well collimated or precessing. Weinterpret the spectra from the shocks as cooled synchrotron emission and derive a cooling age of ∼ × erg s − . P jet . × erg s − ,making this one of a few microquasars with a direct measurement of its jet power and the only knownmicroquasar that exhibits stationary large-scale X-ray emission. Subject headings:
ISM: jets and outflows, X-rays: binaries, X-rays: individual (Circinus X-1) INTRODUCTION
Circinus X-1 is an unusual X-ray binary (XRB).Stewart et al. (1993) first noticed that Circinus X-1 re-sides inside of a parsec-scale radio nebula inflated bycurved jets. More-recent radio observations have ob-served jet emission on multiple scales (Tudose et al.2008, and references therein). Additionally, at the peakof the 40+ year X-ray light curve (see Figure 1 ofParkinson et al. 2003), Fender et al. (2004) detected ev-idence of superluminal motion from ultra-relativistic jetswithin a few arcseconds of the XRB, which would makethis the only known accreting neutron star with an ultra-relativistic jet. With its resolved, parsec-scale radio jets,Circinus X-1 has become an important stepping stonein our understanding of microquasars over the past fewdecades.Circinus X-1 has been extensively studied at X-ray Department of Astronomy, University of Wisconsin-Madison, Madison, WI 53706, USA School of Physics and Astronomy, University of Southamp-ton, Southampton SO17 1BJ, UK Netherlands Institute for Radio Astronomy, Postbus 2, 7990AA Dwingeloo, The Netherlands Astronomical Institute of the Romanian Academy, Cutitulde Argint 5, RO-040557 Bucharest, Romania Research Center for Atomic Physics and Astrophysics,Atomistilor 405, RO-077125 Bucharest, Romania Kapteyn Astronomical Institute, University of Groningen,P.O. Box 800, 9700 AV Groningen, The Netherlands SRON, Netherlands Institute for Space Research, Sorbon-nelaan 2, 3584 CA, Utrecht, The Netherlands Department of Astrophysics, IMAPP, Radboud UniversityNijmegen, P.O. Box 9010, NL-6500 GL Nijmegen, the Nether-lands Harvard–Smithsonian Center for Astrophysics, 60 GardenStreet, Cambridge, MA 02138, USA Kavli Institute for Astrophysics and Space Research, Mas-sachusetts Institute of Technology, Cambridge, MA 02139, USA Department of Astronomy and Astrophysics, The Pennsyl-vania State University, 525 Davey Laboratory, University Park,PA 16802, USA Astronomical Institute Anton Pannekoek, University ofAmsterdam, Science Park 904, 1098XH Amsterdam, TheNetherlands wavelengths. Type-I X-ray bursts, originally observedby Tennant et al. (1986) and seen again by Linares et al.(2010) during a recent flare, now firmly establish that thecompact object is an accreting neutron star. Multiple
Chandra gratings observations of the point source haverevealed variable X-ray P-Cygni line profiles that wereinterpreted as high-velocity outflows (Brandt & Schulz2000; Schulz & Brandt 2002). Iaria et al. (2008) haveclaimed possible detection of a highly-inclined precess-ing jet (similar to SS433, e.g., Lopez et al. 2006) throughDoppler-shifted X-ray line emission. However, this in-terpretation is not unique as the emission is also con-sistent with simple orbital motion of the neutron star.Moreover, Schulz et al. (2008) also analyzed that partic-ular data set and, with systematic uncertainties included,find no shifts. Finally, in the longest gratings observa-tion of the point source (zero-order, 50 ks) taken in 2005,Heinz et al. (2007) discovered faint, diffuse X-ray emis-sion roughly coincident with the arcminute-scale radiojets. However, the observation had insufficient signal–to–noise to constrain the nature of the emission spec-troscopically. Similar diffuse emission was subsequentlyseen in a 50 ks HRC-I observation (Soleri et al. 2009),but no spectral information was available with this in-strument.In this Letter, we report on a follow-up deep
Chandra imaging observation of the diffuse X-ray emission foundby Heinz et al. (2007). We outline our observations inSection 2 and present an initial analysis of the emissionin Section 3. Section 4 discusses the physical implica-tions. Finally, we summarize our results in Section 5.Throughout this Letter, we assume a distance of 7.8 kpc(Jonker et al. 2007), which is in the middle of a widerange of distance estimates (4.1–11.8 kpc; Iaria et al.2005). OBSERVATIONS
We observed Circinus X-1 on the S3 chip of the Ad-vanced CCD Imaging Spectrometer (ACIS) aboard the
Chandra X-ray Observatory on 2009 May 1, in a contin-
Fig. 1.—
Reduced event file showing the point source and thebipolar “caps” created by the outflow. Overlaid are contours fromthe lower resolution radio image (logarithmically spaced from 1.9to 25 mJy beam − ; beam size: 18 . ′′ × . ′′
5) from Tudose et al.(2006). uous 99 ks exposure. Data were taken in timed exposuremode and telemetered in Faint mode. Data reductionand analysis were completed using CIAO and Sherpaversions 4.2, XSpec 12.5.1, and ACIS Extract version2010-02-26 (AE; Broos et al. 2010). Compared to pre-vious gratings and HRC-I observations, this observationproved about an order of magnitude more sensitive todiffuse flux not only because of the long exposure andthe lack of gratings, but also because the contaminatingemission from the point source was at an exceptionallylow level (Section 3.1).Radio comparison data were constructed from twoseparate observations. Standard calibrations using theMiriad software (Sault et al. 1995) were applied to bothsets of observations. A low angular resolution radio im-age was derived from data taken on 2001 August 3 dur-ing an 11 hr run at 1.4 GHz with the Australia Tele-scope Compact Array (ATCA) in 1.5A array configura-tion. This observation and the resulting images were pre-sented in Tudose et al. (2006) and are shown as contoursin Figure 1.A higher angular resolution radio image (Figure 2)was created from ∼
80 hr of observations spread from2009 December 30 to 2010 January 8, using the ATCA-Compact Array Broadband Backend in 6A configurationat 5.5 GHz (D. E. Calvelo et al., in preparation). ANALYSIS
The reduced event file of the X-ray observation isshown in Figure 1 with a contour overlay of the low-resolution radio image of the large-scale radio nebula ofCircinus X-1. This image clearly shows the existence oftwo bright emission regions ∼ ′′ from the point source.In order to highlight the morphology of the diffuseemission closer to the point source and to allow detailedcomparison with the high-resolution radio image, we cre-ated an exposure-corrected, background-subtracted im-
48 46 44 42 15:20:40 38 36 34RA (J2000)4020-57:10:004020 D ec ( J ) Inner Emission Caps
Fig. 2.—
Exposure-corrected and azimuthally-smoothed,background-subtracted X-ray image. Overlaid are contours fromthe higher resolution radio image (levels: -1, 2, 4, 8, 16, 32 timesrms noise of ∼ µ Jy beam − ; beam size: 3 . ′′ × . ′′
21; D. E.Calvelo et al., in preparation). The point source, inner emission,and the X-ray caps described in Sections 3.1–3.3 are all evidenthere and well-matched to the radio contours. age (Figure 2). Background surface brightness profileswere constructed by azimuthally averaging over nestedpartial annuli in the NE and SW quadrants (visuallychosen to exclude the excess diffuse emission) and thensubtracted from the exposure-corrected image.Finally, we note that on the largest scales, we findroughly axisymmetric soft emission from what we in-terpret as the dust-scattering halo, centered on the bi-nary, best visible in the adaptively-smoothed image inFigure 3. An in-depth discussion of the dust-scatteringhalo will be presented in a follow-up paper.Image analysis and X-ray–radio comparisons are pre-sented below, ordered from smaller to larger scales. Allwell-determined background/foreground point sourceswere masked and excluded from our analysis.
The Point Source
We caught Circinus X-1 at the lowest observed flux todate ( F . − ≃ × − erg cm − s − ). Even so, thesource is still significantly piled up with a pileup fractionof ∼ An Arcsecond Jet
Because the point source was so faint, the point spreadfunction (PSF) from the XRB only dominates the in-ner ∼ ′′ . This enabled us to search for jet emission onscales of a fraction of a parsec, much smaller than whathad been possible in Heinz et al. (2007) and Soleri et al.(2009). We found two sources within ∼ ′′ of the pointsource, which are both evident in Figure 2:1. In the W-NW direction (between position angles0 ◦ and +45 ◦ , measured counterclockwise from dueW), we find a clear surface brightness enhancementseparated from the XRB by ∼ ′′ and extending outto ∼ ′′ . The emission appears to be resolved and Fig. 3.—
Adaptively smoothed three-color image ∼ . ′ × . ′ is coincident with a source in the high-resolutionradio image. No IR counterpart is found in ei-ther 24 µ m MIPSGAL or 3.6–8 µ m GLIMPSE Spitzer images. Given that the source is extendedand aligned with the large-scale radio jet and withthe X-ray caps reported below, we interpret thissource as genuine jet emission.2. In the SE direction (at position angle 230 ◦ ), wedetect a bright, unresolved point source at ∼ ′′ from the XRB. This emission is coincident with apoint-like radio source. It is spatially coincidentwith a very bright IR point source, with no opti-cal counterpart detected in an archival 5 minuteHST/WFPC I exposure.From the FIRST (Faint Images of the Radio Skyat Twenty-Centimeters) log N –log S , we estimatethat the probability of finding a background radiopoint source of the same or larger flux within 13arcseconds from Circinus X-1 is ∼ http://mipsgal.ipac.caltech.edu/ these sources are beyond the scope of this Letter andwill be presented in a follow-up paper. Extended Diffuse Emission: X-ray Caps
The most obvious features in Figures 1-3 are the twodiffuse emission regions between ∼ ′′ and 50 ′′ fromCircinus X-1 in the NW and SE directions. The po-sitions of both regions are consistent with the diffuseemission tentatively reported in Heinz et al. (2007) andSoleri et al. (2009). Based on their morphological ap-pearance (NW cap appears slightly concave) and theirplacement away from the binary along the jet axis, wewill refer to both regions as “caps” throughout the restof this Letter. However, the geometry of these regionscould be complicated by projection effects, and other in-terpretations are possible.The caps have distinctly different X-ray colors fromboth the much redder large-scale diffuse background(dominated by the dust-scattering halo) and the dis-tinctly bluer PSF, as can be seen from the color image inFigure 3. The two caps have very similar surface bright-nesses ( ∼ × − erg cm − s − arcsec − ) and appearto be similar in their angular extent (NW: − ◦ –65 ◦ ; SE:185 ◦ –245 ◦ ). However, they are asymmetric in their ra-dial extent (NW: ∼ ′′ –50 ′′ , SE: ∼ ′′ ). The innerNW jet emission (Section 3.2) has similar position andopening angles to the NW cap.The correspondence between the high-resolution radiocontours ( ≥ σ ) and the X-ray cap emission in Figure 2 isstriking. While the X-ray image has significantly higherangular resolution than the radio, it is clear that the X-ray caps align closely with the extended jet emission.The diffuse emission to the SE of the XRB in the radioimage shows a sharp surface brightness drop at the posi-tion of the cap and what has been interpreted as a bendin the jet direction at this position, trailing to the south(referred to as “knot B” in Tudose et al. 2006). We seeno clear corresponding surface brightness enhancementto this trail in the X-rays. However, our sensitivity iscompromised by the fact that the read streak runs rightalong this feature.We used partial annuli regions to extract the cap spec-tra. Because the background is dominated by the dust-scattering halo, the surface brightness of which dependson the angular distance to Circinus X-1, we carefullyselected background partial annuli at the same radial ex-tent as the source regions. We recover a total of 6240source counts over 2860 expected background counts inboth cap regions combined (0.5–9.5 keV). The resultingspectra are shown in Figure 4.Initial extractions of each of the caps indicated thatthe spectral parameters are statistically indistinguish-able. Therefore, we jointly fit the spectra. We restrictedour spectral analysis to 0.5–9.5 keV. Quoted uncertain-ties correspond to 90% confidence intervals. The spectraare well fit by an absorbed powerlaw with a neutral hy-drogen column of N H = 2 . +0 . − . × cm − and a pho-ton index, Γ = 1 . +0 . − . ( χ = 243 . . × − erg cm − s − (NW) and 2 . × − erg cm − s − (SE).The caps can also be fit by a thermal (APEC) model( χ = 237 . -5 -4 -3 -2 c oun t s s - k e V - χ Fig. 4.—
Top: spectrum of the two large-scale caps and best-fitabsorbed power-law fit (red: NW cap; green: SE cap). Bottom:deviations from the best-fit model. with N H = 1 . +0 . − . × cm − . This is consistent withthe fact that there is no obvious line emission present inthe spectra, the APEC fit requires a low abundance of Z = 0 . +0 . − . and a high temperature of kT = 6 . +2 . − . keV. When forced to solar abundance, the fit requireseven higher temperatures. The 0.5–9.5 keV cap fluxesare 3 . × − erg cm − s − (NW) and 2 . × − ergcm − s − (SE). DISCUSSION
The limb-brightened morphology of the X-ray image(which has significantly higher angular resolution thaneither radio image) suggests that the emission from thetwo caps originates at a shock in the outflow from Circi-nus X-1, seen in projection. This suggestion is supportedby a sharp drop in surface brightness in radio and X-rayjust outside of the caps and by the fact that the SE ra-dio jet seems to be changing direction at the location ofthe X-ray cap. The fact that the caps appear inside thelarge-scale radio nebula would be a result of foreshort-ening, since the jet axis is likely inclined with respectto the line of sight (though the actual inclination is un-known). The outflow itself appears largely X-ray dark(except for the inner NW diffuse radio–X-ray feature),similar to the X-ray cavities observed in many clusterswith central radio galaxies.The caps span projected half–opening angles of 35 ◦ (NW) and 30 ◦ (SE). This implies that either:1. the outflow is, in fact, much wider than what wouldtypically be considered a jet and might thus be bet-ter characterized as a non-thermal wind, or2. the jets are precessing with a fairly wide open-ing angle, as is the case for SS433 (Margon 1984),and/or3. the jet or precession cone axis is very close to theline of sight, causing significant foreshortening ofan intrinsically narrow jet.The wide opening angle of the inner NW jet would beconsistent with (1) and (3), but to be consistent with (2),it would require a precession period of the jet that is short compared to the travel time through the X-ray emissionregion to blend the jet emission (as seen in the arcsecondradio jets of SS433 Hjellming & Johnston 1981).The claim of a highly relativistic jet (Γ jet ∼ θ . ◦ ;Fender et al. 2004) coupled with these new observationswould require that such a jet has a precession conethat is within 5 ◦ of the line–of–sight such that the jetsometimes points very close to the line–of–sight. Sucha geometry would imply a physical scale of ∼
20 pcfrom cap to cap. For comparison, this is an order ofmagnitude larger than the projected distance betweenthe hotspots of XTE J1550–564 (Corbel et al. 2002)and H1743–322 (Corbel et al. 2005) but would still fitcomfortably within the radio nebula surrounding SS433(Dubner et al. 1998).Below, we will discuss two possible interpretations ofthe radiative origin of the cap emission.
Cooled Synchrotron Model
The most likely interpretation of the spectra is syn-chrotron emission. In this case, the emission arisesfrom the reaccelerated jet particles entering the termi-nal shock, as is observed in the hot spots of many FR IIradio galaxies (e.g., Meisenheimer et al. 1989). The X-ray emission in this case should be co-spatial with theradio, which appears consistent with the radio contoursin Figure 2.The radio synchrotron spectrum of the large-scale neb-ula has a typical spectral index of α R = 0 .
53 (with F ν ∝ ν − α R ) (Tudose et al. 2006), consistent with thestandard power-law slope of first-order Fermi accelera-tion. For the canonical model of continuously injectedpower-law electrons one would expect a spectral indexof α X = α R + 0 .
53 = 1 .
03 above the cooling break atfrequency ν b , perfectly consistent with the observed X-ray spectral index of α X = 0 .
99. From extrapolation ofthe radio and X-ray spectra, we estimate the break fre-quency from uncooled to cooled synchrotron emission tobe ν b ∼ × Hz, which is uncertain by about anorder of magnitude when uncertainties in the radio andX-ray spectral indices are used primarily because the X-ray and radio power-law slopes are extrapolated over avery large range of frequencies.Taking the emission regions to be spherical to low-est order, we estimate the equipartition magnetic fieldstrength to be B eq ∼ µ G, which gives a minimum to-tal internal energy for each emission region of ∼ × erg. From the estimate of the break frequency, we esti-mate a cooling time, τ cool ∼ lower limiton the total jet power of P cir > × erg s − , needed just to put the synchrotron emitting particles in place(not including any plasma currently in the jet or the en-ergy needed to inflate the large-scale radio nebula). De-parting from equipartition can only increase this number.This limit is independent of the assumed geometry andsize of the emission regions.Conversely, we can place a rough upper limit on theoutflow power from the fact that the large-scale radionebula must be at least as old as τ cool . Using the samearguments as presented in Heinz (2002) and Tudose et al.(2006) for the expansion of the radio nebula we find that P cir . × erg s − , though this limit is not as robustsince the magnetic field could well be out of equiparti-tion and the nebula could be significantly elongated alongthe line–of–sight, both of which would increase the totalpower.These limits on P cir confirm that the accreting com-pact object in this system is driving powerful jets intothe interstellar medium (ISM). Taking the high 42 yearaverage luminosity of Circinus X-1 at about 80% ofthe Eddington luminosity for a 1.4 M ⊙ neutron star(Parkinson et al. 2003) at face value would imply thatthe long-term average jet power is between 2% and 10%of the current average radiative power.Given that the average long-term accretion rate cannotgreatly exceed the Eddington rate, we can derive a robustlower limit on the efficiency with which accretion poweris converted into outflow power of η jet = P cir / ˙ mc > . ˙ m Edd h ˙ m i (see also Heinz et al. 2007 and Soleri et al.2009). Thermal Model
Given the equally good fit we achieved with a ther-mal emission model, it is worth discussing the alterna-tive scenario that the X-ray caps are the shocked ISMpushed ahead of the expanding radio outflow (similar tothe shells around many X-ray cavities in galaxy clustersand around the microquasar Cygnus X-1; Gallo et al.2005).In the thermal model the X-rays should be located fur-ther out from the binary than the radio emission. How-ever, the high-resolution radio data indicate that the ra-dio and the X-ray emission are co-spatial. This requiresa very low inclination angle in order to align the tworegions by projection (which is consistent with the incli-nation required by the ultra-relativistic flow claimed inFender et al. 2004).From the best-fit temperature of 6.6 keV, we directly infer a shock velocity of ∼ . × km s − . From theprojected distance, we can infer an estimated travel timeof ∼ / sin( θ ) years out to the observed shock positionsfor each of the caps. Again approximating the emissionregion as spherical, we estimate the mass and the total(thermal plus kinetic) energy of the shocked material foreach of the caps to be ∼ × g and ∼ × erg,respectively. Combined with the shock travel age, thisgives a minimum outflow power for both caps combinedof P cir > × × sin ( θ ) erg − s − , just to supplythe X-ray emitting material (not including the putativepower required to inflate the large-scale radio nebula).Unless the inclination angle is very small, this power isexceedingly large. Even for inclinations as low as thoseinferred from the relativistic jet claimed in Fender et al.(2004), the minimum average power would still be P cir > × erg s − .While we cannot completely rule out a thermal ori-gin, the very high power required by the thermal modelcombined with the perfect spectral correspondence to aclassic broken synchrotron spectrum and the spatial cor-respondence with the radio suggest that the synchrotronmodel is the more natural choice. SUMMARY
We have presented an initial analysis of a deep
Chan-dra imaging observation of Circinus X-1. We detect twodiffuse X-ray caps that are likely the terminal shocks ofpowerful jets running into the ISM. In addition, we findan arcsecond-scale outflow between the XRB and one ofthe X-ray caps, coincident with the radio jet. These dis-coveries make Circinus X-1 the first microquasar withboth an X-ray jet and two stationary X-ray shocks, andone of only a handful of microquasars with a direct esti-mate of the jet power.S.H. and P.S. acknowledge support from NASA grantGO9-0056X.imaging observation of Circinus X-1. We detect twodiffuse X-ray caps that are likely the terminal shocks ofpowerful jets running into the ISM. In addition, we findan arcsecond-scale outflow between the XRB and one ofthe X-ray caps, coincident with the radio jet. These dis-coveries make Circinus X-1 the first microquasar withboth an X-ray jet and two stationary X-ray shocks, andone of only a handful of microquasars with a direct esti-mate of the jet power.S.H. and P.S. acknowledge support from NASA grantGO9-0056X.