Chandra and XMM–Newton observations of the low-luminosity X-ray

Mon. Not. R. Astron. Soc. 394, 1597–1604 (2009) doi:10.1111/j.1365-2966.2009.14438.x

Chandra and XMM–Newton observations of the low-luminosity X-raypulsators SAX J1324.4−6200 and SAX J1452.8−5949

R. Kaur,1� Rudy Wijnands,2 Alessandro Patruno,2 Vincenzo Testa,3 GianLuca Israel,3

Nathalie Degenaar,2 Biswajit Paul4 and Brijesh Kumar1

1Aryabhatta Research Institute of Observational Sciences, Manora Peak, Nainital 263 129, India2Astronomical Institute ‘Anton Pannekoek’, University of Amsterdam, Kruislaan 403, 1098 SJ Amsterdam, the Netherlands3INAF – Osservatorio Astronomico di Roma, via Frascati 33, 00040 Monte Porzio Catone, Italy4Raman Research Institute, C. V. Raman Avenue, Sadashivanagar, Bangalore 560 080, India

Accepted 2008 December 19. Received 2008 November 30; in original form 2008 September 7

ABSTRACTWe present results from our Chandra and XMM–Newton observations of two low-luminosityX-ray pulsators SAX J1324.4−6200 and SAX J1452.8−5949 which have spin periods of 172and 437 s, respectively. The XMM–Newton spectra for both sources can be fitted well with asimple power-law model of photon index, � ∼ 1.0. A blackbody model can equally well fit thespectra with a temperature, kT ∼ 2 keV, for both sources. During our XMM–Newton observa-tions, SAX J1324.4−6200 is detected with coherent X-ray pulsations at a period of 172.86 ±0.02 s while no pulsations with a pulse fraction greater than 18 per cent (at 95 per cent confi-dence level) in 0.2–12 keV energy band are detected in SAX J1452.8−5949. The spin period ofSAX J1324.4−6200 is found to be increasing on a time-scale of P = (6.34±0.08)×10−9 s s−1

which would suggest that the accretor is a neutron star and not a white dwarf. Using subarcsecspatial resolution of the Chandra telescope, possible counterparts are seen for both sourcesin the near-infrared images obtained with the son of infrared spectrometer and array camera(SOFI) instrument on the New Technology Telescope. The X-ray and near-infrared propertiesof SAX J1324.4−6200 suggest it to be a persistent high-mass accreting X-ray pulsar at adistance ≤8 kpc. We identify the near-infrared counterpart of SAX J1452.8−5949 to be alate-type main-sequence star at a distance ≤10 kpc, thus ruling out SAX J1452.8−5949 to bea high-mass X-ray binary. However, with the present X-ray and near-infrared observations,we cannot make any further conclusive conclusion about the nature of SAX J1452.8−5949.

Key words: binaries: close – stars: neutron – pulsars: individual: SAX J1324.4−6200, SAXJ1452.8−5949 – X-rays: binaries.

1 IN T RO D U C T I O N

In the past 10 years, using the Advanced Satellite for Cosmologyand Astrophysics (ASCA), BeppoSAX and the Rossi X-ray TimingExplorer (RXTE) satellites, a population of faint (LX � 1036 erg s−1)X-ray sources has been found in our Galaxy which harbour a slowpulsating source with periods ranging from several seconds to over athousand seconds. These ‘slow’ pulsators are found to harbour a va-riety of source types like anomalous X-ray pulsars (isolated slowlyrotating neutron stars; e.g. Torii et al. 1998), accreting magnetizedwhite dwarfs (i.e. intermediate polars (IPs), AM Her type systems;e.g. Misaki et al. 1996) and neutron stars accreting from a high-masscompanion star (i.e. mostly as Be/X-ray transients; e.g. Hulleman,in ’t Zand & Heise 1998). Despite the successes in determining

�E-mail: [email protected]

the nature of these slow pulsators, there remains a group of persis-tent systems whose nature still has not been determined. The X-rayproperties of these sources suggest that most of them are neutronstars accreting from a high-mass companion star; however, accret-ing white dwarf cannot be excluded. Furthermore, in some cases,the possibility of a system in which a neutron star accretes from alow-mass companion star also cannot be excluded (Lin et al. 2002).More observations at all wavelengths (i.e. X-ray or near-infrared)could help to unveil the nature of these pulsators. In this paper, wepresent the results from our Chandra, XMM–Newton and New Tech-nology Telescope (NTT) observations of two faint X-ray pulsatorsSAX J1324.4−6200 (hereafter SAX13; l = 306.◦79, b = 0.◦60) andSAX J1452.8−5949 (hereafetr SAX14; l = 317.◦65, b = −0.◦46).

SAX13, which has a pulse period of ∼170 s, was discov-ered serendipitously from observations of the low-mass X-raybinary (LMXB) XB 1323−619 performed on 1997 August 22 withBeppoSAX (Angelini et al. 1998). The X-ray (1.8–10 keV) spectrum

C© 2009 The Authors. Journal compilation C© 2009 RASDownloaded from https://academic.oup.com/mnras/article-abstract/394/3/1597/1070925by gueston 31 January 2018

1598 R. Kaur et al.

Table 1. Log of the X-ray observations of SAX13 and SAX14.

Object Telescope/instrument Date Obs. ID Total observation(UT) span (ks)

SAX13 Chandra/ACIS-I 2007 November28 9012 1.1SAX14 Chandra/ACIS-I 2007 December 30 9014 1.0SAX13 XMM–Newton/EPIC 2008 January11 0511010201 19.3SAX14 XMM–Newton/EPIC 2008 February 07 0511010501 6.9

of the source could be fitted with either an absorbed power law withphoton index � = 1.0 ± 0.4 and hydrogen column density NH =7.8+2.7

−1.1 × 1022 cm−2 or with a blackbody model with a temperaturekT = 2.4 ± 0.4 keV and NH = 4+3

−2 × 1022 cm−2 (Angelini et al.1998). SAX13 was also detected in the observations performedon XB 1323−619 using ASCA on 1994 August 4 (Angelini et al.1998). In addition, pointed ASCA observations (of 187 ks) wereperformed with SAX13 on 2000 February 2, and the source wasdetected with a pulse period of ∼171.2 s (Lin et al. 2002). Lin et al.(2002) found a possible orbital period of the system of 27 ± 1 h andsuggested that the system could be a LMXB pulsar. Recently, dur-ing short observations performed using Swift on 2007 December 30,SAX13 was detected with a pulse period of 172.8 s (Mereghetti, Ro-mano & Sidoli 2008) which would imply a spin-down over the last10 years of P =∼ 6 × 10−9 s s−1. Mereghetti et al. (2008) identi-fied a possible Two-Micron All-Sky Survey (2MASS) near-infraredcounterpart in the Swift error circle of SAX13 with a K-band mag-nitude of 14.39 ± 0.08 and suggested that the source could be apersistent Be accreting X-ray pulsar.

SAX14 was discovered using BeppoSAX with a spin period of∼ 437 s (Oosterbroek et al. 1999). The X-ray spectra of the sourcecould be fitted well with an absorbed power-law model with aphoton index, � = 1.4 ± 0.6. Oosterbroek et al. (1999) suggestedthat SAX14 could be an accreting Be X-ray pulsar at a distance of6–12 kpc with the luminosity of ∼ 1034 erg s−1.

2 X -RAY OBSERVATIONS

Our X-ray observations of SAX13 and SAX14 were carried outusing the European Photon Imaging Camera (EPIC) aboard theXMM–Newton satellite and advanced CCD Imaging Spectrometer(ACIS) aboard the Chandra satellite.

2.1 XMM–Newton

SAX13 and SAX14 were observed for ∼19 ks on 2008 January11 (Obs. ID 0511010201) and ∼7 ks on 2008 February 7 (Obs.ID 0511010501), respectively. Both the EPIC–metal-oxide-silicon(MOS) and EPIC-pn cameras (Struder et al. 2001; Turner et al.2001) were operated in the full-frame mode and with the mediumfilter. The observation details are summarized in Table 1. TheEPIC observation data files were processed using the XMM Sci-ence Analysis System (SAS, version 7.1.0).1 Investigation of thefull-field count rate of SAX13 revealed ∼3 ks of high count ratebackground particle flaring and is removed from the data. Wedid not see any significant background flaring in SAX14 obser-vation. We used the task EDETECT CHAIN to find the exact posi-tion of X-ray source in the combined images of EPIC–MOS and

1 See http://xmm2.esac.esa.int/sas/.

EPIC-pn instruments. In the BeppoSAX error circle of SAX13(Angelini et al. 1998), we detected only one source in the XMM–Newton observations at a position of RA = 13h24m26.s64 andDec. = −62◦01′18.′′48 with an error circle of radius 2.0 arcsec(90 per cent confidence; J2000; all coordinates in the paperare for J2000 epoch). In the BeppoSAX error circle of SAX14(Oosterbroek et al. 1999), only one source is detected in the XMM–Newton observations at a position of RA = 14h52m52.s80 andDec. = −59◦49′08.′′04 with an error circle of radius 2.0 arcsec (90per cent confidence). The error circle on the position of the X-raysource is adopted as a quadratic sum of the bore sight error of theXMM–Newton telescope2 and the statistical error given by the taskEDETECT CHAIN.

2.2 Chandra

SAX13 and SAX14 were observed for ∼1 ks each on 2007November 28 (Obs. ID 9012) and 2007 December 30 (Obs. ID9014), respectively. The details of the observations are given in Ta-ble 1. The data were obtained with the ACIS-I CCDs operating in theFAINT mode for both sources. We processed the event 2 files usingthe standard software packages CIAO 4.03 and CALDB 3.4.2.4 The taskWAVDETECT was used to find the exact position of X-ray source in theimage. One source is detected in the XMM–Newton error circle ofSAX13 with a total of 87 counts at a position of RA = 13h24m26.s70and Dec. = −62◦01′19.′′49 with the error circle of radius0.65 arcsec (Fig. 1). In the Chandra image of SAX14, one sourceis detected (Fig. 2) in the XMM–Newton error circle of SAX14,albeit with only four counts at a position of RA = 14h52m52.s70and Dec. = −59◦49′8.′′07 with the error circle of radius 0.95 arc-sec. The average background count is ∼1 count for a similar areain the full Chandra image of SAX14, i.e. indicates that indeed wehave detected SAX14. The errors on the positions of SAX13 andSAX14 are obtained as a quadratic sum of the bore sight error ofthe Chandra telescope,5 1σ WAVEDETECT errors and a contributionthat depends on the number of detected counts (van den Berg et al.2004; especially important for SAX14).

3 INFRARED O BSERVATI ONSA N D DATA A NA LY S I S

Near-infrared observations in the J, H, Ks wavebands were per-formed in 2001 February with the 3.58 m European South-ern Observatory–NTT (ESO–NTT) telescope equipped with the

2 See http://xmm2.esac.esa.int/docs/documents/CAL-TN-0018.pdf.3 Chandra Interactive Analysis of Observations (CIAO), http://cxc.harvard.edu/ciao/.4 Chandra Calibration Data base (CALDB), http://cxc.harvard.edu/caldb/.5 See http://cxc.harvard.edu/cal/ASPECT/celmon.

C© 2009 The Authors. Journal compilation C© 2009 RAS, MNRAS 394, 1597–1604Downloaded from https://academic.oup.com/mnras/article-abstract/394/3/1597/1070925by gueston 31 January 2018

SAX J1324.4−6200 and SAX J1452.8−5949 1599

Figure 1. Left-hand panel: the Ks waveband image of SAX13 taken by SOFI instrument with the NTT. The XMM–Newton error circle of radius 2.0 arcsec(in black colour) and the Chandra error circle of radius 0.65 arcsec (in white colour) of SAX13 are also plotted. Right-hand panel: Chandra ACIS-I image ofSAX13 with the XMM–Newton error circle of radius 2.0 arcsec.

Figure 2. Left-hand panel: the Ks waveband image of SAX14 taken by SOFI instrument with the NTT. The XMM–Newton error circle of radius 2.0 arcsec(in black colour) and the Chandra error circle of radius 0.65 arcsec (in white colour) of SAX14 are also plotted. Right-hand panel: Chandra ACIS-I image ofSAX14 with the XMM–Newton error circle of radius 2.0 arcsec.

near-infrared imager and spectrograph SOFI. The instrument wasset up in large imaging mode with a pixel scale of 0.29 arcsec anda field of view of 5 × 5 arcmin2 when SAX13 was observed and insmall imaging mode with a pixel scale of 0.14 arcsec and a field ofview of 2.5 × 2.5 arcmin2 when SAX14 was observed (both in dou-ble ‘correlated’ readout mode). In both cases, the seeing conditionswere very good, ranging from 0.58 to 0.66 arcsec between Ks and Jbands. Images were acquired with the auto-jitter sequence, obtain-ing several dithered frames for each filter, that were then re-alignedand co-added together. Table 2 summarizes the observations. Theacquisition mode is the usual one for near-infrared arrays: a numberof single frames (number of detector integrations (NDIT)) having

Table 2. Log of the 3.58 m ESO–NTT near-infrared observations of SAX13 and SAX14. NDIT represents the number of single frames, having exposure timesof DIT seconds and are used to generate an output image having exposure time equal to one DIT. The number of output frames are represented by Nframes.

Source Date J H Ks FWHM (arcsec)(UT) DIT NDIT Nframes DIT NDIT Nframes DIT NDIT Nframes

SAXJ1324.4−6200 2001 Februray 06 3 5 8 3 5 8 3 5 12 0.6SAXJ1452.8−5949 2001 February 08 3 5 10 3 5 12 3 5 16 0.6

Note. FWHM – full width at half-maximium.

exposure times of Detector Integrator Time (DIT) seconds are ac-quired and then co-added aboard generating an output image havingexposure time equal to one DIT, but the signal-to-noise ratio (S/N)statistics of NDIT×DIT seconds. All the output frames (Nframes)are then processed and co-added offline, generating the final scienceimage.

Images were reduced using the package ECLIPSE developed byESO to handle near-infrared observations. The pipeline executesthe following steps: first, a master ‘sky’ image is obtained bymedian stacking all the frames for one observation and for eachfilter. This ‘sky’ image is subtracted to every single image gener-ating sky-subtracted frames. This step also subtracts bias and dark


1600 R. Kaur et al.

Table 3. J, H and Ks magnitudes of stars falling in the XMM–Newton error circle of SAX13 and SAX14. The stars listed here are alsomarked in the near-infrared ESO–NTT images shown in Figs 1 and 2.

Star RA Dec. J H Ks

(hh:mm:ss) (◦′′′) (mag) (mag) (mag)

Stars in XMM–Newton error circle of SAX13C1 13:24:26.71 −62:01:19.59 19.57 ± 0.11 16.61 ± 0.09 14.97 ± 0.11C2 13:24:26.56 −62:02:19.38 18.76 ± 0.10 16.48 ± 0.09 15.45 ± 0.11C3 13:24:26.72 −62:01:17.71 18.37 ± 0.10 17.00 ± 0.09 16.62 ± 0.12

Stars in XMM–Newton error circle of SAX14S1 14:52:52.72 −59:49:07.92 18.59 ± 0.12 17.75 ± 0.12 17.93 ± 0.12S2 14:52:52.70 −59:49:10.14 16.92 ± 0.11 15.50 ± 0.10 15.06 ± 0.10

current contributions. The images are then flat-fielded with flat-fieldimages obtaining on the dome. In the case of SOFI, also a ‘screen’correction image applied to correct for residual illumination gradi-ents in the dome. At the end, the pre-reduced images are alignedand average stacked, with a proper σ -clipping selection, obtainingthe science-ready final images. Source detection and photometryhave been performed with the DAOPHOT package within IRAF.6 In par-ticular, a second-order variable point spread function (PSF) modelhas been computed for our images to minimize positional biasesdue to the (small) PSF variation over the field. The output magni-tudes have been aperture corrected and calibrated with the 2MASScatalogue (Skrutskie et al. 2006). The same 2MASS has been usedalso to obtain an astrometric calibrations of the frames. The inter-active software SKYCAT-GAIA has been used, selecting the portion ofcatalogue of the area and fitting a solution to the image. The posi-tions obtained in this way have (absolute) uncertainties of about0.2 arcsec, and relative uncertainties far below a fraction of apixel.

As can be seen in the Ks waveband image taken with NTT,the position of three point sources labeled as ‘C1’, ‘C2’, ‘C3’ areconsistent with the XMM–Newton error circle of SAX13 (Fig. 1).The position of two point sources labelled as ‘S1’ and ‘S2’ isconsistent with the XMM–Newton error circle of SAX14 (Fig. 2).The Chandra error circle for both the sources is also shown: theposition of star ‘C1’ falls in the Chandra error circle of SAX13 andthat of ‘S1’ in the Chandra error circle of SAX14. Therefore weconsider ‘C1’ as the near-infrared counterpart of SAX13 and ‘S1’that of SAX14. We have listed the positions (in right ascension anddeclination) and magnitudes in J, H, Ks bands for the near-infraredcounterparts of SAX13 and SAX14 in Table 3. For the sake ofcompleteness, the positions and magnitudes of the other stars fallingin the XMM–Newton error circles of SAX13 and SAX14 are alsolisted in Table 3.

4 DATA A NA LY SIS AND RESULTS

4.1 Timing Analysis

The XMM–Newton EPIC–MOS and pn data were used for the tim-ing analysis for both SAX13 and SAX14. The X-ray events wereextracted in a circular region of radius 30 arcsec centred on theposition of the target in EPIC–MOS images for both SAX13 and

6 Image Reduction and Analysis Facility (IRAF) is distributed by the NationalOptical Astronomy Observatories, which are operated by the Association ofUniversities for Research in Astronomy, Inc., under cooperative agreementwith the National Science Foundation.

SAX14 while a circular region of radius 15 and 30 arcsec was usedto extract X-ray source events from EPIC-pn images for SAX13 andSAX14, respectively. We were forced to take smaller circular regionin EPIC-pn image of SAX13 as it was located close to a CCD gap.The background X-ray events were extracted with a similar circularregion on the same CCD in a source-free region for both EPIC–MOSand pn images of SAX13 and SAX14. The times of the events weretransformed to barycentric times using the SAS tool BARYCOR usingthe Chandra position of the source and the JPL-DE405 ephemeris.The events were rebinned with a time resolution of 0.2 s and thebackground is subtracted to generate the final light curves.

We reduced ≈16 ks of pointed observations from the EPIC–MOS and the EPIC-pn CCD cameras for SAX13, and detected thesource with a background-corrected count rate of 0.16 counts s−1

and 0.28 counts s−1, respectively. We searched for the spin pe-riod of SAX13 in the combined XMM–Newton EPIC–MOS and pndata using POWSPEC and refined it further using EFSEARCH in FTOOLS7

software package. Using this technique, we got the pulse periodof SAX13 to be 172.57 ± 0.22 s (1σ confidence level). To fur-ther refine the pulse period of SAX13, we folded the light curvesin five chunks of ≈3000 s, taking the above period as the bestavailable spin period. The pulse profiles were then cross-correlatedwith a template profile obtained by folding the entire light curve.The cross-correlation returns the time of arrivals (TOAs) of eachpulse profile with the fiducial point for the measurement fixed to bethe peak of the template profile. The technique applied to measurethe TOAs and their statistical uncertainties closely resemble the ra-dio pulsar technique (e.g. Taylor 1992). The pulsations were thenphase connected with a linear fit to the TOAs to obtain a more pre-cise pulsar spin period. Our measured spin period is Ps = 172.86 ±0.02 s (referred at MJD 54460; The error is calculated with68 per cent confidence level) and is consistent with the previousmeasurements of Mereghetti et al. (2008) that reported a spin pe-riod of 172.84 ± 0.1 s at MJD 54464.2. The detection of a spin-period derivative using a phase coherent technique is preventedgiven the short baseline of the observation. We fitted all the previ-ous and our new measured value of the spin period (Angelini et al.1998; Lin et al. 2002; Mereghetti et al. 2008) with a linear relationPs(t) = P0+Pst , where Ps is the spin-period derivative of the pulsarand P0 is the spin period at the time t = 0. The value obtained fromour fit is Ps = (6.34 ± 0.08) × 10−9s s−1 with a χ 2 of 4.46 for threedegrees of freedom (d.o.f.).

We folded the entire XMM–Newton EPIC-pn light curve in onesingle profile to increase the S/N and measure the harmonic content

7 http://heasarc.gsfc.nasa.gov/docs/software/ftools/ftools_menu.html


SAX J1324.4−6200 and SAX J1452.8−5949 1601

0 0.5 1 1.5 2

0.2

0.4

0.6

0.8

1

1.2

Phase

Nor

mal

ized

cou

nts

s

Figure 3. The XMM–Newton EPIC-pn 0.2–12 keV background subtractedpulse profile of SAX13. Two cycles are shown for clarity.

and the fractional amplitude of the pulsation. The pulse profile isfitted well with a single sinusoid with a significance larger than 20σ

and has a fractional amplitude of (52 ± 4) per cent in 0.2–12 keVenergy band. Fig. 3 shows the pulse profile of SAX13 in the energyrange 0.2–12 keV.

SAX14 was detected with a background subtracted count rate of0.05 counts s−1. We did not detect any pulsations in the combinedEPIC–MOS and pn lightcurve of SAX14 using POWSPEC in FTOOLS

software. We repeated the same procedure as described above to tryto detect the pulses using TOAs for SAX14 using the best availablespin period from Oosterbroek et al. (1999). Again no significantpulse profiles have been detected in the energy band 0.2–12 keVwith an upper limit on the pulsed fraction amplitude (Vaughan et al.1994) of 18 per cent (at 95 per cent confidence level). To furtherinvestigate the presence of pulsations, we divided the EPIC lightcurve into soft (0.2–4 keV) and hard (4–12 keV) energy bands withapproximately equal count rate. No significant pulse profiles weredetected in both soft and hard energy bands of EPIC data of SAX14with an upper limit on the pulse fractional amplitude of 25 and22 per cent (at 95 per cent confidence level), respectively.

4.2 Spectral Analysis

The XMM–Newton EPIC–MOS and pn data are used for spectralanalysis for SAX13 and SAX14. We used the same extraction region

Table 4. Spectral Parameters for SAX13 and SAX14 for XMM–Newton EPIC observations.

SAX13 SAX14

Parameter Spectral parameters of model (absorption + power law)NH × 1022 (cm−2) 5.81 ± 0.65 1.22 ± 0.69Photon index 1.01 ± 0.14 0.83 ± 0.28Reduced χ2/d.o.f. 0.95/383 0.6/19Observed flux 2–10 keV (erg cm−2 s−1) (4.5 ± 1.3) × 10−12 (5.4 ± 2.5) × 10−13

Unabsorbed flux 2–10 keV (erg cm−2 s−1) (6.0 ± 1.4) × 10−12 (5.8 ± 2.9) × 10−13

Parameter Spectral parameters of model (absorption + blackbody)NH × 1022 (cm−2) 3.07 ± 0.42 <0.54 (90 per cent confidence)Blackbody temperature (keV) 2.24 ± 0.14 1.82 ± 0.24Reduced χ2/dof 1.0/383 0.7/20Observed flux 2–10 keV (erg cm−2 s−1) (4.4 ± 0.2) × 10−12 (4.9 ± 0.4)× 10−13

Unabsorbed flux 2–10 keV (erg cm−2 s−1) (5.0 ± 0.3) × 10−12 (5.0 ± 0.4) × 10−13

for EPIC data as were used for the timing analysis reported inSection 4.1. SAS tool XMMSELECT was used to extract both source andbackground spectra. We used XSPEC (version 11.0.0) for our spectralanalysis. The resulting spectra were rebinned to have minimum of20 counts per bin.

We fitted both power-law and blackbody models to the SAX13EPIC–MOS and pn X-ray spectra together. Both models could fitthe data with a reduced χ 2 of ∼1.0 for 385 d.o.f. The full values ofthe obtained fit parameters are given in Table 4. For the power-lawmodel, the spectrum could be fitted with a photon index, � = 1.0,and an NH of 5.8 × 1022 cm−2 while using a blackbody model, weobtained a temperature, kT = 2.2 keV, and an NH of 3.1 × 1022 cm−2.The observed flux for both models in the 2–10 keV energy band is∼5 × 10−12 erg s−1 cm−2. With the present data, we are unable todistinguish between the power-law and blackbody model fit. Thepower-law fit to SAX13 XMM–Newton EPIC–MOS and pn spectrais shown in Fig. 4. No significant Fe 6.4 keV emission line is seenin the combined XMM–Newton EPIC–MOS and pn spectrum ofSAX13 (Fig. 4). To find an upper limit on the Fe 6.4 keV emissionline, we fixed the line center at the 6.4 keV in the spectrum andfitted a Gaussian to the line. The upper limit on the equivalent widthof the 6.4 keV emission line is determined to be 118 eV (90 per centconfidence limit).

Similarly, we fitted both power-law and blackbody models to theSAX14 spectra. Both models provide acceptable fits with reducedχ 2 in the range of 0.6–0.7 for 20 d.o.f. From the power-law model,we obtained a photon index, � = 0.8, and an NH of 1.2 × 1022 cm−2

(shown in Fig. 5) while using the blackbody model we obtaineda temperature, kT = 1.8 keV, and an NH = 0.5 × 1022 cm−2. Theobserved flux in 2–10 keV energy band (for both the models) is∼5 × 10−13 erg s−1 cm−2. The spectral parameters of SAX14 arelisted in Table 4. No significant Fe 6.4 keV emission line is seenin the SAX14 spectra (Fig. 5). However, an upper limit on iron6.4 keV emission line in the combined XMM–Newton EPIC–MOSand pn spectrum of SAX14 is only 436 eV (at 90 per cent confidencelevel).

5 D ISCUSSION

We carried out observations with the Chandra and XMM–Newtonsatellites to investigate the nature of the accreting X-ray pulsatorsSAX13 and SAX14. In addition, we obtained near-infrared imag-ing observations using the ESO–NTT to search for near-infraredcounterparts for these pulsators.


1602 R. Kaur et al.

10

0.01

0.1

Nor

mal

ized

cou

nts

s1

keV

1

011 525.0

5

0

5

χ

Energy (keV)

Figure 4. The XMM–Newton EPIC–MOS and pn spectra of SAX13 fittedwith an absorbed power-law model. The pn data points are represented byopen triangles, the MOS1 data points are represented by open circles andthe MOS2 data points are represented by filled circle.

10

0.01

2 10 3

5 10 3

0.02

Nor

mal

ized

cou

nts

s1

keV

1

1052

2

1

0

1

Energy (keV)

χ

Figure 5. The XMM–Newton EPIC–MOS and pn spectra of SAX14 fittedwith an absorbed power-law model. The pn data points are represented byopen triangles, the MOS1 data points are represented by open circles andMOS2 are represented by filled circle.

5.1 SAX J1324.4−6200

During our XMM–Newton observation, SAX13 is detected witha pulse period of 172.85 ± 0.02 s and with a pulse fraction of(52 ± 4) per cent. The spin-period history of SAX13 clearly showsa linear increase in spin period from 170.35 to 172.86 s over thelast 14 years (Table 5) and by using it we calculated the spin-periodderivative, P = (6.34 ± 0.08) × 10−9s s−1. The XMM–NewtonX-ray spectra of SAX13 fit well with both power-law and blackbodymodels with photon index, � = 1.0, and temperature, kT = 2.2 keV,respectively. No significant Fe 6.4 keV emission line with an equiv-alent width ≥118 eV (at 90 per cent confidence level) is detected inthe X-ray spectrum of SAX13. Our ESO–NTT observations clearlyshowed that the near-infrared counterpart identified by Mereghettiet al. (2008) in the 2MASS data is a combination of three starsmarked as C1, C2 and C3 in Fig. 1. However, our Chandra observa-tions of SAX13 identified star ‘C1’ as the most likely near-infraredcounterpart of SAX13, with a magnitude in Ks waveband = 14.97 ±0.11 mag.

The spin period of high-mass X-ray binary (HMXB) pulsarsranges from a few to a few hundred seconds and display several spin-

Table 5. Spin-period history of SAX13.

Telescope Date Spin period References(UT) (s)

ASCA 1994 August 04 170.35 ± 0.48 aBeppoSAX 1997 August 22 170.84 ± 0.04 aASCA 2000 February 02 171.25 ± 0.01 bSwift 2007 December 30 172.84 ± 0.1 cXMM–Newton 2008 January 11 172.86 ± 0.02 d

References: (a) Angelini et al. 1998; (b) Lin et al. 2002; (c) Mereghettiet al. 2008; (d) this paper.

up/down trends on time-scales ranging from days to years (Bildstenet al. 1997). For example, 4U 1907+09, a persistent supergiant X-raybinary with Ps = 440 s, is spinning down for more than 15 years withP = 7.3 × 10−7s s−1 (Baykal et al. 2001). Among LMXB pulsars,GX 1+4 has the longest spin period of ∼141 s. LMXB pulsars showa wide range of spin period derivatives ranging from ∼10−8 to ∼10−11s s−1 (Chakrabarty et al. 1997; Ferrigno et al. 2007). IPs arealso slow pulsators with spin period of a few hundred seconds andusually show spin-up phases with a typical spin-period derivative ∼− 7 × 10−11s s−1 (Patterson 1994). However, there are a very fewIPs which are found to be spinning down and the fastest spinningdown IP PQ Gem is observed with P = 1.1 × 10−10s s−1 (Mason1997). Since in SAX13 the observed spin torque is comparablewith measures of spin torques in high B field accreting neutronstars, the presence of a neutron star is favoured. Lin et al. (2002)identified a possible orbital period of SAX13 to be ∼27 h. Ourpresent observation span is not long enough to confirm this period.However, if we consider the orbital period to be real, then SAX13could be a LMXB pulsar or an IP. On the other hand, this orbitalperiod would be too small for a HMXB pulsar.

Accreting X-ray pulsars with high magnetic field typically dis-play hard spectra while pulsars with low magnetic field strengthdisplay soft X-ray spectra with photon index, � ≥ 2.0 (Bildstenet al. 1997). IPs also display hard X-ray spectra similar to HMXBpulsars but usually they show strong iron emission lines in their X-ray spectra (e.g. Norton, Watson & King 1991; Muno et al. 2004).The absence of Fe emission lines in the SAX13 X-ray spectra furthermakes it very unlikely to be an IP.

Using the observed NH = 5.8 × 1022 cm−2 (Table 4), the extinc-tion towards SAX13 in J, H and Ks wavebands is calculated to beAJ = 9.01 mag, AH = 5.56 mag and AK = 3.77 mag (Predehl &Schmitt 1995; Fitzpatrick 1999). Here, we assume that there is nolocal contribution of the absorbing matter from the X-ray sourceand the companion star would experience the same NH as observedby the X-ray source (we note that this, not necessary, could betrue in which case our conclusions about the possible nature of thecompanion star could be significantly affected.).

With the given dereddened near-infrared fluxes and using a black-body model, an estimate of a distance is made for different type ofstars (main-sequence stars, supergiants and giants), for a given tem-perature and a radius (Cox 2000). With these calculations, we foundthat any late-type main-sequence star (M through A type) would lieat a distance ≤1.4 kpc. Thus, it is possible that the system could bean IP at a distance of ≤1.4 kpc and in that case the correspondingX-ray luminosity of the system would be ≤1033 erg s−1. However,it is very unlikely to find a LMXB pulsar at such a low X-ray lu-minosity. Also it is not very likely that a supergiant is the infraredcounterpart of SAX13 as it would lie outside the Galaxy for thegiven magnitudes. A main-sequence early-type star (e.g. B type)


SAX J1324.4−6200 and SAX J1452.8−5949 1603

1 525.0

12

10

8

T = 20000K

T = 3000K

λ (μm)

log (

λFλ)

erg

s1 c

m2

J H K

SAX J1324.4 6200

1 525.0

16

14

12

10

T = 20000K

T = 3000K

λ (μm)

log

(λ F

λ) e

rg s

1 c

m2

J H K

SAX J1452.8 5949

Figure 6. Observed (open circles) and extinction-free (filled circles) spectral energy densities of near-infrared counterpart of SAX1324.4−6200 (left-handpanel) and SAX1452.8−5949 (right-hand panel). The solid and dash-dotted lines represent blackbody model spectral energy density of the star at temperature,T = 20 000 and 3000 K, respectively, and are shifted to match the extinction-free spectral energy density of the star at near-infrared H waveband.

with temperature of 11 000–30 000 K and the radius of 3.0–7.5 Rwould lie at a distance ≤8 kpc, indicating SAX13 could be anaccreting neutron star HMXB pulsar. However, for the given fluxdensities, a late-type giant would also be well within the Galaxyat a similar distance and in that case SAX13 could be a symbioticX-ray binary pulsar (Iben & Tutukov 1996; Corbet et al. 2008). Thelack of long-term X-ray flux variations in SAX13 indicates that it isunlikely for it to be a symbiotic X-ray binary pulsar. However, withthe limited X-ray observations of SAX13, we cannot completelyrule out SAX13 to be a symbiotic X-ray binary pulsar.

To further confirm the near-infrared counterpart to be a high-massstar, we approximated the blackbody model to the extinction-freenear-infrared fluxes of SAX13 (shown as open circles in Fig. 6).Blackbody curves for temperature T = 20 000 and 3000 K areshown in Fig. 6 as a solid and a dash-dotted line, respectively,and are shifted to match the spectral flux density of SAX13 at Hwaveband. The solid line clearly indicates that SAX13 is an early-type main-sequence star of temperature ≥20 000 K. For the sake ofcompleteness, the observed fluxes of SAX13 are also shown in thesame figure as filled circles.

Based on the above arguments, it is therefore possible that SAX13is an accreting neutron star HMXB pulsar at a distance of 1.5–8.0 kpc. For the observed X-ray flux of SAX13 (Table 4), the lumi-nosity of the system would be 1.2 × 1033–3.4 × 1034 erg s−1. Mostof the HMXB pulsars (especially Be X-ray pulsars) are transient innature and have been observed at luminosities varying from 1033

to 1038 erg s−1. However, SAX13 has been persistent for the past14 years. Thus it is possible that SAX13 is a persistent Be X-ray bi-nary pulsar, similar to X-Persei (Reig & Roche 1999; La Palombara& Mereghetti 2007). These systems might be part of an unusualclass of accreting neutron stars with high-mass companions whichhave long orbital periods (>30 d) and low eccentricities (e.g. Pfahlet al. 2002). Such long orbital period indicates that tidal circular-ization cannot yet have occurred after the supernova explosion thatcreated the neutron star. Therefore, these low eccentricities mustbe primordial and the supernova explosion could not have beenaccompanied by a kick to the neutron star (Pfahl et al. 2002).

The present X-ray and near-infrared observations of SAX13 sug-gest it to be an accreting HMXB pulsar. However, the detection ofthe Brackett series emission in the near-infrared spectrum of SAX13would further help us to confirm the Be nature of its infrared coun-terpart.

5.2 SAX 1452.8−5949

We did not detect any pulsations in our XMM–Newton EPIC–MOSand pn data with a pulse fractional amplitude greater than 18 percent in 0.2–12 keV energy band in SAX14. The XMM–NewtonX-ray spectra of SAX14 is well fitted with a power-law model ofphoton index, � = 0.83, (Table 4) with no significant detection ofFe 6.4 keV emission line with an equivalent width ≥436 eV. Wedetected a faint star in the ESO–NTT images of SAX14, consistentwith the Chandra error circle, shown in Fig. 2 with a magnitude inKs waveband = 17.93 ± 0.12 mag. Therefore we identify ‘S1’ asthe most likely near-infrared counterpart of SAX14 (Fig. 2).

Both an accreting pulsar and a white dwarf are consistent withthe inferred X-ray spectral parameters of SAX14, thus it is difficultto distinguish between a neutron star or an accreting white dwarffrom the X-ray spectral information. Non-detection of a strong Fe6.4 keV emission line in the X-ray spectra indicates that it is unlikelythat the system is an IP. The spectral parameters from our presentXMM–Newton observations of SAX14 are comparable with the onereported by Oosterbroek et al. (1999).

Taking NH = 1.22 × 1022 cm−2 (Table 4), the dereddened mag-nitudes of SAX14 are calculated to be J = 16.66 ± 0.12 mag, H =16.45 ± 0.12 mag and Ks = 17.15 ± 0.12 mag. We calculated thedistance of the possible infrared counterparts with the same methodused for SAX13 (see Section 5.1). We could rule out a supergiant,giant or an O/B-type star as the infrared counterpart of SAX14 as,given the observed flux densities, it would lie outside the Galaxy.This rules out the possibility for SAX14 to be a HMXB. Only alate spectral-type main-sequence star (M through A type) can bein the Galaxy, at a distance ≤10 kpc. A blackbody approximationto the extinction-free near-infrared fluxes of SAX14 indicates thatSAX14 would have temperature <20 000 K (Fig. 6). However, incase, a part of the X-ray absorption density is local to the X-raysource, we would expect less extinction in near-infrared wavebandsas compared to the extinctions calculated above, and in that case, thenear-infrared counterpart of SAX14 would tend to be a low-massstar. With the above arguments, we can say that, if SAX14 is in abinary system it must have a low-mass companion (LMXB or IP),regardless if there are pulsations or not.

The non-detection of pulsations in SAX14 is in contrast with theprevious detection made by Oosterbroek et al. (1999) who detectedpulsations with a fractional amplitude of 75 ± 25 per cent (at


1604 R. Kaur et al.

90 per cent confidence level). To explain this discrepancy, we areleft with two possibilities: the fractional amplitude of the pulsationsdecreased or the pulsations observed by Oosterbroek et al. (1999)were spurious (as the detection significance of pulsations is lessthan 3σ ). In the later case, SAX14 can be any non-pulsating sourcein the Galaxy, like LMXB, accreting white dwarf, BY Dra, RS CVnor active star, etc.

If, on the other hand, the detection of pulsations by Oosterbroeket al. (1999) was real, then the pulse amplitude was reduced bya substantial amount. This behaviour has been observed in otherLMXBs also where the pulsed fraction decreased on the time-scaleof days (Her X-1; Ramsay et al. 2002), hours (GX 1+4; Naik,Paul & Callanan 2005) and minutes (4U 1907+09; in ’t Zand,Strohmayer & Baykal 1997). Given the large value for the upperlimit we obtained, the possibility that SAX14 is a slow-pulsatingLMXB or an IP is still open.

AC K N OW L E D G M E N T S

We would like to thank anonymous referee for the constructivecomments which helped us to improve the paper. We would also liketo thank Nanda Rea for the help with XMM–Newton data reduction.One of the authors RK would like to thank ‘Astronomical InstituteAnton Pannekoek’ of University of Amsterdam where much of thiswork was done for the kind hospitality provided during the stay.

REFERENCES

Angelini L., Church M. J., Parmar A. N., Balucinska-Church M., Mineo T.,1998, A&A, 339, 41

Baykal A., Inam C., Alpar M. A., in ’t Zand J., Tod S., 2001, MNRAS, 327,1269

Bildsten L. et al., 1997, ApJS, 113, 367Chakrabarty D. et al., 1997, ApJ, 474, 414Corbet R. H. D., Sokoloski J. L., Mukai K., Markwardt C. B., Tueller J.,

2008, ApJ, 675, 1424

Cox A. N. ed., 2000, Allen’s Astrophysical Quantities, 4th edn. AIP, NewYork, p. 388

Ferrigno C., Segreto A., Santangelo A., Wilms J., Kreykenbohm I., DenisM., Staubert R., 2007, A&A, 462, 995

Fitzpatrick E. L., 1999, PASP, 111, 63Hulleman F., in ’t Zand J. J. M., Heise J., 1998, A&A, 337, L25Iben I., Tutukov A. V., 1996, ApJS, 105, 145in ’t Zand J., Strohmayer T., Baykal A., 1997, ApJ, 479, L47La Palombara N., Mereghetti S., 2007, A&A, 474, 137Lin X. B., Church M. J., Nagase F., Balucinska-Church M., 2002, MNRAS,

337, 1245Mason K. O., 1997, MNRAS, 285, 493Mereghetti S., Romano P., Sidoli L., 2008, A&A, 483, 249Misaki K., Terashima Y., Kamata Y., Ishida M., Kunieda H., Tawara Y.,

1996, ApJ, 470, L53Muno M. P. et al., 2004, ApJ, 613, 1179Naik S., Paul B., Callanan P. J., 2005, ApJ, 618, 866Norton A. J., Watson M. G., King A. R., 1991, Lecture Notes in Physics,

385, 155Oosterbroek T. et al., 1999, A&A, 351, L33Patterson J., 1994, PASJ, 106, 209Pfahl E., Rappaport S., Podsiadlowski P., Spruit H., 2002, ApJ, 574, 364Predehl P., Schmitt J. H. M. M., 1995, A&A, 293, 889Ramsay G., Zane S., Jimenez-Garate M. A., den Herder J., Hailey C. J.,

2002, MNRAS, 337, 1185Reig P., Roche P., 1999, MNRAS, 306, 100Skrutskie M. F. et al., 2006, AJ, 131, 1163Struder L. et al., 2001, A&A, 365, L18Taylor J. H., 1992, Phil. Trans. R. Soc., 341, 117Torii K., Kinugasa K., Katayama K., Tsunemi H., Yamauchi S., 1998, ApJ,

503, 843Turner M. J. L. et al., 2001, A&A, 365, L27van den Berg M., Tagliaferri G., Belloni T., Verbunt F., 2004, A&A, 418,

509Vaughan B. A. et al., 1994, ApJ, 435, 362

This paper has been typeset from a TEX/LATEX file prepared by the author.


Documents

Chandra and XMM–Newton observations of the low-luminosity X-ray