An Optical Counterpart Candidate for the Isolated Neutron Star RBS1774
S. Zane, R. Mignani, R. Turolla, A. Treves, F. Haberl, C. Motch, L. Zampieri, M. Cropper
aa r X i v : . [ a s t r o - ph ] A p r Send offprint requests to: [email protected]
Preprint typeset using L A TEX style emulateapj v. 12/01/06
AN OPTICAL COUNTERPART CANDIDATE FOR THE ISOLATED NEUTRON STAR RBS1774
S. Zane , R. P. Mignani , R. Turolla , A. Treves , F. Haberl , C. Motch , L. Zampieri , M. Cropper Send offprint requests to: [email protected]
ABSTRACTMultiwavelength studies of the seven identified X-ray dim isolated neutron stars (XDINSs) offera unique opportunity to investigate their surface thermal and magnetic structure and the matter-radiation interaction in presence of strong gravitational and magnetic fields. As a part of an ongoingcampaign aimed at a complete identification and spectral characterization of XDINSs in the opticalband, we performed deep imaging with the ESO
Very Large Telescope ( VLT ) of the field of the XDINSRBS1774 (1RXS J214303.7 +065419). The recently upgraded
FORS1 instrument mounted on the
VLT provided the very first detection of a candidate optical counterpart in the B band. The identificationis based on a very good positional coincidence with the X-ray source (chance probability ∼ × − ).The source has B=27.4 ± . σ confidence level), and the optical flux exceeds the extrapolation ofthe X-ray blackbody at optical wavelengths by a factor ∼
35 ( ±
20 at 3 σ confidence level). This isbarely compatible with thermal emission from the neutron star surface, unless the source distanceis d ≈ Subject headings: star: individual (RBS1774) — stars: neutron — X-rays: stars — ultraviolet: stars INTRODUCTION
One of the most intriguing results of the ROSATAll Sky Survey has been the detection of seven close-by neutron stars (NSs), with particular character-istics (XDINSs in the following, see Haberl 2007;Van Kerkwijk & Kaplan 2007, for recent reviews). Thesesources stand apart with respect to other known classesof isolated NSs detected at X-ray energies. Their X-rayspectrum is close to a blackbody, and no evidence of radioemission has been reported so far despite deep searches(e.g. Kondratiev et al. 2008). They are likely to be en-dowed with relatively strong magnetic fields, B ≈ –10 G, as inferred from X-ray timing measurements andobservations of broad spectral lines (equivalent width ≈ Mullard Space Science Laboratory, University College LondonHolmbury St Mary, Dorking, Surrey, RH5 6NT, UK Department of Physics, University of Padova, via Marzolo 8,I-35131 Padova, Italy Universit´a degli Studi dell’Insubria, Dipartimento di Fisica eMatematica, via Valleggio 11, 22100 Como, Italy Max-Planck Institut f¨ur extraterrestrische Physik, Giessen-bachstrasse, 85748 Garching, Germany Observatoire Astronomique, 11, rue de l’Universite, F-67000Strasbourg, France INAF-Astronomical Observatory of Padova, Vicolodell’Osservatorio 5, I-35122 Padova, Italy The detection of pulsed emission from two sources has beenclaimed at very low frequencies (Malofeev et al. 2005, 2006) but is,so far, unconfirmed.
Detailed multiwavelength studies of XDINSs are fun-damental for tracking their evolutionary history, andfor shedding light on their thermal and magnetic sur-face properties. While the XDINSs have similar spectralproperties in the X-rays, in the optical the paucity ofmulti-band observations prevents a clear spectral char-acterization. For the XDINSs with a certified coun-terpart (see e.g. Kaplan 2008, for a recent review)the optical emission lies typically a factor ∼
10, ormore, above the extrapolation of the X-ray blackbodyinto the optical/UV band. However, while the opti-cal flux closely follows a Rayleigh-Jeans distribution inRX J1856.5-3754, possible deviations from a λ − be-haviour have been reported for RX J0720.4-3125 andRX J1605.3+3249 (Kaplan et al. 2003a; Motch et al.2003, 2005; Zane et al. 2006). Thus, whether the opti-cal emission from XDINSs is produced by regions of thestar surface at a lower temperature (e.g. Pons et al. 2002)or by other mechanisms, such as non-thermal emissionfrom particles in the star magnetosphere or reprocessingof the surface radiation by an optically thin (to X-rays)hydrogen layer surrounding the star (Motch et al. 2003;Zane et al. 2004; Ho et al. 2007), is still under debate.One of the XDINSs which so far eluded optical iden-tification is RBS1774 (1RXS J214303.7 +065419). Thissource was firstly identified in a pointed ROSAT/PSPCobservation by Zampieri et al. (2001); accurate spectraland timing information were then obtained with XMM-Newton by Zane et al. (2005). The EPIC-PN count rate(background corrected) is ∼ . ∼ × − erg cm − s − . The EPIC-PN spectrum isvery soft and well fitted by an absorbed blackbody with kT ∼
104 eV and N H ∼ . × cm − . There is evi-dence for a spectral absorption feature at ∼ . σ confidence level) with a Zane S. et al.pulsed fraction of ∼
4% in semiamplitude (Zane et al.2005; Cropper et al. 2007).The first optical follow-ups with the
NTT and withthe
VLT revealed no optical counterpart within the
XMM-Newton error circle, down to limiting magnitudesof R ∼ ∼ . VLT , Blanco and Magellan telescopes were recently re-ported by Rea et al. (2007), exploiting the subarcsec po-sition obtained through a DDT
Chandra observation.Again, no plausible optical and/or infrared counterpartfor RBS1774 was detected down to r’ ∼ ∼ L = 0.02 mJykpc at 1.4 GHz. Very recently Kondratiev et al. (2008)placed more stringent upper limits on the radio luminos-ity of RBS1774, L . Hz ∼ .
005 and ∼ . for pulsed and bursty emission respectively, which arethe most stringent limits obtained to date from radioobservations of XDINSs.In this paper, we present the first detection of a candi-date optical counterpart to RBS1774, obtained with the VLT . The observations and data analysis are describedin §
2, while discussion and conclusions follow in § §
4, respectively. THE NEW
VLT
OBSERVATIONS
Observations description
We performed deep optical imaging of the RBS1774field with
FORS1 (FOcal Reducer Spectrograph), amulti-mode instrument for imaging and long-slit/multi-object spectroscopy mounted at the
VLT
Kueyen tele-scope (Paranal Observatory). The instrument has beenrecently upgraded with the installation of a new detectorwhich is the mosaic of two 2k ×
4k E2V CCDs, optimizedfor the blue range. Due to vignetting, the effective skycoverage of the two CCD chips is smaller than the pro-jected detector field of view, and larger for the upperchip (dubbed “Norma”). Observations were carried outin Service Mode on July 11th and 21st 2007.
FORS1 was set up in its default standard resolution mode, witha 2 × . ′′
25. The tele-scope pointing was set in order to position our target onthe “Norma” CCD chip, and, thanks to its large effectivesky coverage (7 × λ = 429 nm; ∆ λ = 88 nm), for a total integrationtime of 8850 and 2950 s in the first and second night, re-spectively. The observations were collected in dark time,with an average seeing of ∼ . ′′ ∼ . ′′ ∼ . − .
37 and ∼ .
18 on the first and secondnight, respectively .
Data reduction
The usual reduction steps (bias subtraction, flat-fielding) were applied to the data through the ESO
FORS1 data reduction pipeline using calibration framesacquired as part of the FORS1 calibration plan. Then,single reduced images were aligned and coadded to filterout cosmic rays using the IRAF task imcombine . Thephotometric calibration was performed through the ob-servation of standard stars from the fields PG 1323 − −
239 (Landolt 1992) at the begin-ning of the night. This yielded nominal extinction andcolor-corrected zero points of 28 . ± .
04 and 28 . ± . FORS1 ( k B = 0 .
255 for the “Norma”chip). Since the zero point computed by the
FORS1 pipeline is in units of e − /s, while the flux on the imageis measured in ADU/s, we corrected the computed zeropoint by applying the detector electrons–to–ADU con-version factor ( GAIN=0.45 ). According to the ParanalObservatory sky monitor and weather report, sky con-ditions were photometric on the first night and at thebeginning of the second night but they were then af-fected by the presence of thin, variable cirri. This meansthat the photometry of the second night is affected by arandom, unknown uncertainty. We have tried to quan-tify this uncertainty by comparing the photometry ofa number of reference field stars observed in the twonights. The star detection was performed using the
SEx-tractor program and magnitudes were computed throughcustomized aperture photometry (
SExtractor parameter
MAG AUTO ). In all sources detected at the flux level ex-pected for our target (B ≥ ∼ . ± . Astrometry
The astrometry on the
FORS1 image was computed us-ing as a reference the coordinates of stars selected fromthe
GSC-2 (version 3.2; Lasker et al. 2008). All starsfrom the
UCAC-2 (Zacharias et al. 2004) are saturatedin our images. Approximately 90
GSC-2 objects are iden-tified in the
FORS1 “Norma” field of view. After fil-tering out extended objects, stellar-like objects that aresaturated or too faint to be used as reliable astromet-ric calibrators, objects falling close to the chip edges,and outliers, we performed our astrometric calibrationusing 20
GSC-2 reference stars, evenly distributed in theinstrument field of view. Their pixel coordinates weremeasured by Gaussian fitting their intensity profiles withthe GAIA (Graphical Astronomy and Image Analysis)tool while the fit to the celestial reference frame wasperformed using the Starlink package ASTROM . Thiscode is based on higher order polynomials, which accountfor unmodelled CCD distorsions. The rms of the astro- n Optical Counterpart... 3metric solution was determined as ≈ . ′′
11, per coordi-nate. Following Caraveo et al. (1998), we estimatedthe overall uncertainty of our astrometry by adding inquadrature the rms of the astrometric fit and the accu-racy with which we can register our field on the
GSC-2 reference frame. This is estimated as √ × σ GSC / √ N s ,where the √ σ GSC is the mean (radial) error of the
GSC-2 coordinates(0 . ′′
3; Lasker et al. 2008) and N s is the number of starsused for the astrometric calibration. The uncertainty onthe reference stars centroids is well below 0.1 pixel andhas been neglected. We also added in quadrature theradial uncertainty on the tie of the GSC-2 to the ICRF(0 . ′′
15; Lasker et al. 2008). Thus, the overall radial accu-racy of our astrometry is 0 . ′′
24 (1 σ ). Results
Fig. 1 shows a section of the co-added
FORS1
B-bandimage of the RBS1774 field. Fig. 1 (right) shows a zoomwith the
Chandra position of RBS1774 overlaied. Thecoordinates of RBS1774 were measured with high pre-cision using
Chandra
HRC observations by Rea et al.(2007): α ( J h m s , δ ( J ◦ . ′′
53 with a nominal accuracy of 0 . ′′ Chandra observation was recentlyreprocessed (December 2007) using updated calibrationdata products, and revised coordinates were obtained, α ( J h m s , δ ( J ◦
54’ 17 . ′′ . ′′ FORS1 spatialresolution, rules out the possibility of blending of fieldsobject. We computed its magnitude through aperturephotometry, by using a customized aperture and back-ground region, following the same procedure describedin the previous sections. The background was computedin an annular region centered on the target, in order tohave the most reliable estimate close to the position ofour source.The measured value, corrected for the atmospheric ex-tinction using the trended
FORS1 extinction coefficients,is B=27.4 ± ∼ σ ) where the quotederror (1 σ c.l.) is purely statistical and does not includethe much smaller errors on the zero point and on theatmospheric extinction correction. No other object isdetected within or close to the Chandra position downto a 3 σ limit of B ∼ .
7. As pointed out in the pre-vious sections, the
FORS1 coordinates are tied to theICRF within 0 . ′′
15 so that we exclude an hypotheticalshift with respect to the
Chandra coordinates which arealso tied to the ICRF . Thus, given its very good coin-cidence with the X-ray position, we regard this object asa viable candidate counterpart to RBS1774. http://cxc.harvard.edu/cal/ASPECT/celmon/ DISCUSSION
In this paper, we report the first detection of a likelyoptical counterpart for RBS1774. Standardly, opticalidentifications of isolated neutron stars are robustly con-firmed either by the detection of optical pulsations orby the measurement of a significant proper motion. Inabsence of such information we can base the optical iden-tification only on the positional coincidence between thecoordinates of our candidate counterpart and those ofRBS1774, as measured in the X-rays. In order to quan-tify the statistical significance, we estimated the chancecoincidence probability that an unrelated field objectmight fall within the
Chandra error circle. This can becomputed as 1 − exp( − πµr ) (see, e.g., Severgnini et al.2005), where µ is the measured object density in the FORS1 “Norma” field of view and r is the registered ra-dius of the Chandra error circle (0 . ′′
65, accounting for theuncertainty of the
FORS1 astrometry while the uncer-tainty due to the unknown RBS1774 proper motion isnegligible given the small time span between the
Chan-dra and
VLT epochs). The measured density of stellarobjects in the field with magnitude 27.2 ≤ B ≤ ∼ P ∼ × − which shows thatour association is robust. Thus, we are confident thatwe have identified a very likely optical counterpart toRBS1774.The flux of our candidate counterpart in the B band,after correcting for interstellar extinction, is F B =(2 . ± . × − keV/cm /s (the error is at 1 σ c.l.). Tocompute this value, we used the column density derivedfrom the best fit to the XMM-Newton spectrum ( N H =3 . × cm − , Cropper et al. 2007), with the A V derived according to the relation of Predehl & Schmitt(1995). The interstellar extinction in the B -bandhas been computed using the extinction coefficients ofFitzpatrick (1999). When compared to the extrapola-tion in the B band of the blackbody which best-fits the XMM-Newton spectrum, F B,x = 7 . × − keV/cm /s(Cropper et al. 2007), this gives an optical excess of35 ±
20 (at 3 σ c.l.), where all the uncertainties on themagnitude–to–flux conversion are negligible. The resultis shown in Fig. 2, where the optical/IR upper limits ofRea et al. (2007) are also overplotted. The optical excessis larger than that typically observed in other XDINSs,and this may cast some doubts on the association ofthe newly detected source with RBS1774. On the otherhand, if, as we suggest, the association is real, it can beused to infer physical constraints on the mechanism thatis responsible for the optical emission.The first scenario we consider is one in which the op-tical emission originates from a cooler fraction of theneutron star surface, which emits as a blackbody attemperature T o (Braje & Romani 2002; Pons et al. 2002; If we compute the chance probability by considering all objectsbrighter than B=27.7, we get a very similar result, P = 2 . × − . The Predehl & Schmitt (1995) relation is affected by uncer-tainties for close objects, due to the problems of modelling the ISMat small distance from the Sun where microstructures weight more.We checked that, when using the relations of Bohlin et al. (1978),as in Rea et al. (2007), and of Paresce (1984) extinction correc-tions are consistent within 0.04 magnitudes, well below the purestatistical error on the source count rate.
Zane S. et al.Kaplan et al. 2003b; Tr¨umper et al. 2004) . In thiscase, the ratio between the optical and X-ray fluxes scalesas ≈ r o T o /r x T x ≡ f , where r o is the radial size of thecold region (which of course can not exceed the maxi-mum value of the neutron star radius), r x and T x arethe blackbody radius and temperature as inferred fromthe X-ray spectrum, T x = 104 eV, r x = 2( d/
300 pc) km(Cropper et al. 2007). By making an assumption on r o we can obtain the lower limit on T o that corresponds tovalues of f between the central value and the 3 σ lowerlimit (15 ≤ f ≤ XMM-Newton spectrum, it must be R ≪
1, where R = (cid:18) r o r x (cid:19) (cid:18) T o T x (cid:19) R /T o . /T o t / (exp( t ) − dt R /T x . /T x t / (exp( t ) − dt . (1)We repeated this calculation by varying the distance be-tween 200 and 500 pc and for r o = 20 ,
15 or 10 km.Results show that this scenario is only barely compat-ible with our data and requires very small distances: d ∼ −
400 pc are only allowed within 3 σ and for neu-tron star radii as large as r o = 20 km, while if r o = 15 kmthe only allowed combinations require f = 15 −
22 and d ∼ −
250 pc. On the other hand, the lower thedistance the smaller is the size of the X-ray emitting re-gion and the more difficult is to explain the small pulsedfraction observed in the X-rays (which is ∼
4% in semi-amplitude). As we tested by assuming simple polar capmodelling and blackbody emission, if d = 300 pc the al-lowed parameter space is already confined to a very smallregion where the inclination angles of the line of sight andof the magnetic dipole axis with respect to the star spinaxis are both . ◦ . If d = 200 pc, the allowed region iseven smaller and the only possible configurations requirethe star to be either a perfectly aligned rotator or viewedalmost along its spin axis.As discussed by Motch et al. (2003), Zane et al. (2004)and Ho et al. (2007), spectral models consisting of bareneutron stars surrounded by thin atmospheres may pre-dict very different amount of optical excess. However,such models are affected by our poor knowledge of theproperties of the condensate surface, and whether theycan produce optical excesses as large as that measuredhere is still an open issue.An alternative interpretation is that the optical emis-sion is non thermal, probably of magnetospheric origin(Pacini & Salvati 1983). As it can be seen from Fig. 2,a power law spectral component E − α matching the B-flux and with index α ∼ − . Note, however, that when applied to other XDINSs this sce-nario often requires large neutron star radii. We do not use the distance determination by Posselt et al.(2007), since their model is currently under revision for the case ofRBS1774 (Posselt, private communication). E ∼ × erg/s. If the case of RBS1774 is analogous, and ifwe take L opt / ˙ E ≈ − for the optical emission efficiency(as measured in old rotation-powered neutron stars byMignani et al. 2004), this yields an optical luminosity of ∼ × erg/s. In order to reproduce the observedflux of our candidate counterpart, after accounting forthe assumed interstellar extinction, RBS1774 should beat an unrealistically small distance of ∼
25 pc. Thus, forthe assumed value of ˙ E , a purely rotation-powered opti-cal emission is only compatible with a scenario in whichour source is at least a two orders of magnitude moreefficient optical emitter (see below). Similar conclusionscan be reached if we consider the case in which the ob-served optical emission of RBS1774 is due to a compos-ite mechanism consisting of both thermal emission fromthe neutron star surface and a rotation-powered magne-tospheric emission (as in the middle-aged neutron starsPSR B0656+14 and Geminga, e.g. Kargaltsev & Pavlov2007).A further, intriguing possibility is that the opticalemission is non-thermal and powered by a mechanismdifferent from rotation. Interestingly, at least in the IRdomain, hints have been found for an increase of the low-energy emission efficiency with the dipole magnetic fieldstrength (Mignani et al. 2007c). Indeed, in magnetarsthe efficiency is at least two orders of magnitude largerthan in rotation-powered neutron stars and the spectrashow a typical (and unexplained) flattening toward theinfrared (Israel et al. 2003). It is interesting to note themagnetic field strength of RBS1774 as inferred from theabsorption feature in the X-ray spectrum is the high-est among all XDINSs ( B ≈ G), larger than theQED critical limit and comparable with that of magne-tars. The optical emission of RBS1774 could thus bepowered by a magnetar-like process. If the spectrum ofRBS1774 shows a similar magnetar-like turn over, thenthe flattening is appearing blueward of the IR band. CONCLUSIONS
We report here the deepest optical observations sofar of the RBS1774 field, performed with the upgraded
FORS1 instrument at the
VLT
Kueyen telescope. Basedon the positional coincidence, we have identified a likelycandidate counterpart. RBS1774 would then be thefifth XDINS detected in the optical band, the secondby the
VLT . For the most reasonable distance range tothe X-ray source, the measured brightness of the can-didate counterpart (B ∼ .
4) is barely compatible withpurely thermal emission from the neutron star surface,while, assuming a value of ˙ E similar to that of otherXDINSs, rotation-powered emission from the magneto-sphere (eventually in combination with a thermal com-ponent) requires RBS1774 to have a very large opticalefficiency ( ∼ d = 400 pc). If the optical emission is poweredby a different process, the likely high magnetic field ofRBS1774 tantalizingly suggests magnetar-like magneto-spheric emission as a viable option. New multi-band ob-servations, especially in the near-UV and in the near-IR,are required to characterize the counterpart spectrumand to assess the contribution of possibly different spec-tral components. At the same time, high resolution op-tical astrometry of the candidate counterpart with then Optical Counterpart... 5refurbished HST could yield the first direct estimate ofthe RBS1774 parallax and distance, crucial to build theneutron star surface thermal map, and to provide confir-mation of the optical counterpart through proper motionmeasurements.SZ acknowledges support from a STFC (ex-PPARC) AF. RM acknowledges STFC for support through aRolling Grant. RT acknowledges financial support fromASI-INAF through grant AAE TH-058. LZ acknowl-edges financial support from ASI-INAF through grantI/023/05/0.
REFERENCESBohlin, R.C., Savage, B. D., Drake, J. F. 1978, ApJ, 224, 132Braje, T.M., Romani, R.W. 2002, ApJ, 580, 1043Caraveo, P.A., Lattanzi, M.G., Massone, G., et al. 1998, A&A, 329,L1Cropper, M. et al. 2004, MNRAS, 351, 1099Cropper, M. et al. 2007, Ap&SS, 308, 161Fitzpatrick, E.L., 1999, PASP, 111, 63Haberl, F. 2007, Ap&SS, 308, 181Heyl, J.S. & Kulkarni, S.R. 1998, ApJ, 506, L61Ho, W.C.G., et al. 2007, MNRAS, 375, 821Israel, G.L. et al. 2003, ApJ, 589, L93Kargaltsev, O.Y., Pavlov, G.G. 2007, Ap&SS, 308, 287Kaplan, D.L., Kulkarni, S., Van Kerkwijk, M.H. 2003a, ApJ, 588,L33Kaplan, D.L., et al. 2003b, ApJ, 590, 1008Kaplan, D.L., van Kerkwijk, M.H. 2005a, ApJ, 628, L45Kaplan, D.L., van Kerkwijk, M.H. 2005b, ApJ, 635, L65Kaplan, D.L. 2008, proceedings of ”40 Years of Pulsars: MillisecondPulsars, Magnetars, and More”, August 12-17, 2007, McGillUniversity, Montreal, Canada, AIP, 983, 331Kondratiev, V.I. et al. 2008, submitted to ApJLandolt, A.U. 1992, AJ, 103, 340Lasker, B.M. et al. 2008, submitted to AJMalofeev, V.M. et al. 2005, Astr. Rep., 49, 242Malofeev, V.M. et al. 2006, ATel
Zane S. et al.
Fig. 1.—
Left. B -band image (40 ′′ × ′′ ) of the RBS1774 field obtained with FORS1 at the
VLT
Kueyen telescope. The squarecorresponds to the 6 ′′ × ′′ zoom shown in the right-hand panel. The intensity scale has been re-adjusted for an easier view of the brighterobjects in the field. Right. ′′ × ′′ zoom of the same field. The circles correspond to the original (Rea et al. 2007) and revised (thiswork, § Chandra position of RBS1774, and are drawn at the 90% (red) and 99% (blue) confidence level. Their size (0 . ′′
65 and 0 . ′′ FORS1 image. The object detected at the centre of theerror circles (B= 27 . ± .
2) is our candidate counterpart to RBS1774.
Fig. 2.—
Multi-band spectrum of RBS1774. Diamonds represent the
XMM-Newton spectrum (Zane et al. 2005). The solid line showsthe unabsorbed blackbody that best fits the X-ray data, while arrows represent the 5 σ upper limits reported by Rea et al. (2007). Thenew VLT measurement is shown as a cross. Dotted and dashed lines represent two powerlaw components matching the B-flux and withslope α = 0 , ..