Table Of Content

Draftversion January15,2013 PreprinttypesetusingLATEXstyleemulateapjv.5/2/11 HOT THERMAL X-RAY EMISSION FROM THE BE STAR HD119682. J.M. Torrejoń1, N.S. Schulz2, M.A. Nowak2, P. Testa3 and J.J. Rodes1 1Instituto UniversitariodeF´ısicaAplicadaalasCienciasylasTecnolog´ıas,UniversidaddeAlicante,E03080Alicante,Spain;[email protected] 2MITKavliInstituteforAstrophysicsandSpaceResearch,CambridgeMA02139,USAand 3HarvardSmithsonianCenter forAstrophysics,CambridgeMA02138,USA Draft version January 15, 2013 ABSTRACT 3 1 We present an analysis of a series of four consecutive Chandra high resolution transmission grat- 0 ings observations, amounting to a total of 150 ks, of the Be X-ray source HD119682 (= 1WGA 2 J1346.5−6255), a member of the new class of γ Cas analogs. The Chandra lightcurve shows significant brightness variations on timescales of hours. However, the spectral distribution appears rather n stable within each observation and during the whole campaign. A detailed analysis is not able to a detect any coherent pulsation up to a frequency of 0.05 Hz. The Chandra HETG spectrum seems J to be devoid of any strong emission line, including Fe Kα fluorescence. The continuum is well de- 4 scribed with the addition of two collisionally ionized plasmas of temperatures kT ≈ 15 keV and 0.2 1 keV respectively, by the apec model. Models using photoionized plasma components (mekal) or non thermal components (powerlaw)give poorer fits, giving support to the pure thermal scenario. These ] E twocomponentsareabsorbedbya singlecolumnwithN =(0.20+0.15)×1022 cm−2 compatiblewith H −0.03 H the interstellar value. We conclude that HD119682 can be regarded as a pole-on γ Cas analog. . Subject headings: stars: individual (HD119682) - X-rays: binaries h p o- 1. INTRODUCTION sourcefromthebackgroundSupernovaRemnantG309.2- r HD119682(=1WGAJ1346.5−6255)isaBestarwhich 00.6. These authors find that the XMM-Newton t PN+MOS data can be satisfactorily described by the s presents copious X-ray emission. A member of the open a cluster NGC 5281, located at a distance of ∼ 1.3 kpc, addition of two optically thin thermal plasmas, using [ this Be star displays an X-ray luminosity in the 0.5–10 the mekal model, with kT1 = 1.7±0.3 keV and kT2 = 1 keV energy band of the order of LX ∼ 1032 ergs−1, low 13.0+−22..64 keV respectively, while the fit to the Chandra 3v wrahyenbincaormiepsa(rLedXw∼it1h03t6h−e37tyeprigcsa−l1l)u,mtwinoosoitrideesrsofofBmeaXg-- seprsecttermumpe,reaxtutrraecsteodf kfrTo1m=th0e.9S53+−0A0..32C44ISkeCVCaDndchkipT,2d≥eli1v4- 1 nitude lower than persistent Be X-ray binaries (Reig & keV respectively. Alternatively, the hotter thermal com- 9 Roche,1999)andatleastone orderofmagnitude higher ponentcanbedescribedbyapowerlaw. Thisdegeneracy 2 thantheexpectedX-rayemissionfromisolatedOBstars hasalsobeenfoundinγ Cas(Smithetal. 2004,Lopesde 1. (Berghöfer et al. 1997). With a mass of M =18±1M⊙ Oliveira et al. 2011) and HD110432 (Lopes de Oliveira and an age of 4±1 Myr, HD119682 seems to be a blue 2007, Torrejón et al. 2012). S07 present also hints of 0 straggler (Marco et al. 2009), the second (out of the 7 a possible pulsation with P ≃ 1500 s which could be 3 known) gamma Cas analogs that shares this property. caused by the spin of a NS or a WD. 1 Rakowski et al. (2006; R06), using the superb spa- The origin of the X-ray emission in γ Cas analogs is : v tialresolutionofChandra aswellasHαimagesfromthe still a mystery. Two hypotheses have been put forward: i Clay6.5mTelescopeintheMagellanObservatoryatLas X Campanas, unambiguously identified the X-ray source • TheX-raybinaryscenarioinwhichaWhiteDwarf r 1WGA J1346.5−6255 with the OB star HD119682. Us- (WD) accretes matter from the Be donor (Haberl, a ingXMM-Newton PN+MOSdata,theseauthorsshowed 1995; Kubo et al., 1998; Owens et al., 1999 for the that the X-ray spectrum is well described by the addi- caseofγ Cas;Torrejón&Orr,2001forHD110432). tion of two optically thin (raymond) thermal plasmas The low potential well of the WD, compared with with kT = 1.07 ± 0.16 keV and kT = 10.4+2.3 keV thatofaNS,accountsnaturallyforthelowerlumi- 1 2 −1.6 nosity and the thermal nature of the X-ray emis- respectively. Based on the similarities shared between sion. This explanation faces several problems. On HD119682 and γ Cas (early type optical counterpart, one hand, the X-ray emission shows no signof any thermal nature of the X-ray emission, logLX/Lbol = −5.4, etc) R06 classify the source as a γ Cas analog, orbital modulation or spin pulse. On the other hand, the progenitor of the putative WD should and argue against other X-ray production mechanisms havebeenstillmoremassivethantheearlyBcom- likecollidingwinds(withanunseencompanion)oremis- sion from a magnetically confined wind (given the very panion (M ∼ 20M⊙). From the evolutionary per- spective, it is challenging to produce a WD from high temperature of the hot plasma). such a massive star; naturally, they would explode Safi-Harb et al. (2007; S07), in a multifrequecy as a supernova leaving behind a NS, as is the case analysis, confirm the identification between 1WGA J1346.5−6255 and HD119682 and characterize the op- forthe BeX-raybinaries. InthecaseofHD119682 no SN explosionseems to have occurred(Marco et tical counterpart (B0.5Ve), definitely unrelating this al. 2009) which argues against the NS companion. 2 Torrejón et al. and,consequently,the spectrais notaffectedbypileup1. TABLE 1 For the lightcurve extractions we add data from all dis- Journalof Chandra HETG observations. persed orders (m = −3,−2,−1,1,2,3) from HEG and ObsID Date texp Ratea MEG. High dispersion orders(m=±3,±2)constitute a (dd-mm-yyyy) (ks) (c/s) 25 % of the total count rate. The analysis was performed with the Interac- tive Spectral Interpretation System (ISIS) v 1.6.1-24 8929 17-12-200800:20:10 28.9 0.039 10835 19-12-200809:32:42 29.7 0.031 (Houck & Denicola 2000). The photoelectric absorp- 10834 20-12-200815:45:04 59.3 0.032 tion has been treated with the tbnew absorption model 10836 21-12-200815:47:03 29.9 0.035 (Wilms et al. 2012), which includes the most up to date a Zerothorder photoelectric cross sections. • The standalone star scenario in which the X-rays are produced by the interaction of the star’s mag- 3. TIMINGANALYSIS netic field with the circumstellar disk (Robinson The four observations analyzed in this work span five et al. 2002, Smith & Robinson 2003, Smith et differentdayswithuninterruptedcoveragesrangingfrom al. 2004). In this scenario, magneto rotational 30 ks to 60 ks. In total they represent 150 ks worth of instabilities in the inner part of the circumstel- data. This allows us to search for coherent pulsations lar disk around the Be star (close to the Keple- over a wide range of frequencies with unprecedented de- rian co-rotation radius) enhance a seed magnetic tail. In the left panel of Fig. 1 we present the Chandra fieldthroughadiskdynamomechanism(Balbus& lightcurve binned to 2000 s bins. As can be seen, the Hawley,1998). Thedifferentialrotationofthedisk, source brightness is variable on timescales shorter than stretch and sever the entangled field lines acceler- thedurationofasingleobservation(ObsID).Duringthe ating particles which, upon impact on the star’s whole campaign, however, the source remains quite sta- surface, heat the plasma up to MK temperatures. ble. In order to search for coherent pulsations we have The finding that the X-ray emission seems to be analyzedthe lightcurvewithamaximumtimeresolution produced close to the surface of the Be star sup- of 10 s. Subsequently we have created lightcurves with ports this scenario (Smith et al. 2012, Torrejón et largertimebinsof100,250and500storeducetheerror al. 2012). The details of this mechanism, though, bars. In Fig. 1, right panel, we show, as an example, are far from clear. Why other early type Be stars the normalized Fast Fourier Transform periodogram of withlargecircumstellardisksdonotproduceX-ray the lightcurveusing250sbins,using the Lomb-Scargle emission at all is not explained. technique, for each ObsID. The upper plot corresponds to all four intervals together taking into account the du- In any case, to understand the true nature of these ration of the time gaps between each observation. Safi- systems will have importantconsequences in our picture Harbetal. (2007)suggestedthepossibilityofa∼1500s of the structure and evolutionof early type stars and/or pulsationinthelightcurveofHD119682. InourlastOb- how binary systems evolve. sID(10386)thereisindeedapeakatP =1443±17s(and In this work we present an analysis of a series of four alsootherat2350±50s). Thesignificanceofthispeak,as consecutiveChandra HighEnergyTransmissionGratings givenbytheFalseAlarmProbability(FAP),islow. FAP (HETG) observations of the Be star HD119682 allow- is defined as the probability that no periodic component ing us to study the spectrum at the highest resolution is present in the data with such a period within the cur- availabletoday. Thefourobservationsamountto atotal rent frequency search parameters. In order to compute of ∼ 150 ks, spread over five days, allowing a sensitive this significance we have run 103 Monte Carlo simula- search for pulsations in a wide range of frequencies. tions of the randomized Y-axis values of the lightcurve 2. OBSERVATIONS and performed the corresponding periodograms. Then we search for peaks higher than that in the original We have processed and analyzed the Chandra data periodogram at any frequency. We have found higher from the ObsIDs 8929, 10835, 10834 and 10836 avail- peaks in 20±3 % (two sigma standard error) of the pe- able at the public archive (PI C.E. Rakowski). In Ta- riodograms. Therefore, the False Alarm Probability is ble 1 we show the journal of observations. The Chan- ratherhigh,∼20%and,consequently,thesignificanceof dra High Energy Transmission Gratings spectrometer the peak low. Similar results are obtained irrespectively (HETG; Canizareset al. 2005)observedthe source dur- ofthesizeofthetimebin. Howeverneitherofthesepeaks ing 150 ks, splitted in four intervals. Both HETG in- arepresentinthepreviousdatasets. Althoughsomelow struments, the High Energy Gratings (HEG; 0.7-8 keV) significancepeakscanbeseenineachperiodogram,none and Medium Energy Gratings (MEG; 0.4-8 keV) were ofthemremainstable. Inconsequence,inthewholedata used for the analysis. After filtering the data, the 0.5 set we do not find evidence of any significant period up to 7.5 keV spectral range showed enough signal-to-noise to a frequency of 0.05 mHz2). ratio for the scientific analysis. Both the spectra as well In order to study the spectral variations of the X- as the response files (arf and rmf) were extracted us- ray source we have extracted the Chandra lightcurves ing the CIAO software (v 4.4). First dispersion orders in the wavelength intervals 3–6 keV (H), 2–3 keV (M) (m = ±1) were fitted simultaneously. The extracted and 0.5–2 keV (S), and have created the correspond- spectra were binned to match the resolution of HEG (0.012˚AFWHM) and MEG(0.023˚A FWHM) andhave 1 See The Chandra ABC Guide to Pileup, v.2.2, aminimumS/Nratioof7. Thepeakcountrateinasin- http://cxc.harvard.edu/ciao/download/doc/pileup-abc.pdf gle (MEG) gratings arm spectrum is ≈ 0.008 cts/sec/˚A 2 TheNyquistfrequencycorrespondingtothe10stimebin. Chandra observations of HD119682 3 1 . 0 8929 10835 10834 10836 5 s 0 c/ 0. 0 0 5×104 105 1.5×105 Time (s) Fig. 1.—Left: Chandra lightcurve, inthe2-25˚Awavelength range, binnedin2000sintervals. Alllightcurves have anarbitrarytime offsetand,therefore,thetimegapsarenottoscale. Right: Powerspectrumcorrespondingtothe250sbinnedChandra HETGlightcurve. The1500spulsation(ν≃6.7×10−4 Hz)isonlydetected inObsID10836butisnotdetected neitherintherestoftheObsIDsnorinthe combinedligthcurve. ing hardness ratios versus source intensity plots. In Fig. 1 2, we show the variation of the soft/high hardness ratio against the source intensity (0.5–6 keV) using 7000 s time bins. As can be seen, the sourcedoes not show any 8 spectral correlation with brightness. The average val- 0. ues of the hardness ratios are 0.43±0.07 (ObsID 8929), 0.48±0.07(ObsID10835),0.53±0.06(ObsID10834)and ) H 6 0.50±0.06(ObsID10836). Theyare,therefore,compat- + 0. ible within errors. The differences between single values S of the hardness ratio are not significant either. Individ- )/( ual errors range between 0.15 and 0.20. In conclusion, H 4 while changes in brightness from bin to bin are, overall, − 0. S significantthe hardness ratios are consistent with a con- ( stant value. The source appears rather monochromatic 2 on timescales from hours to days. 0. 4. SPECTRALANALYSIS According to the previous analysis, the spectra of 0 HD119682 is fairly stable in a time scale of days de- 0.02 0.03 0.04 0.05 spite the variationsin the brightness. Thereforewe have Counts/s combinedthe fourobservationsintoasinglespectrumto increase the signal to noise ratio (S/N). The resulting spectrum is shown in Fig. 3. Fig.2.—Chandra hardness-intensityplotusing7000stimebins. In order to describe the continuum, models consist- DifferentcolorsrepresentthedifferentObsIDs,withthesamecolor ing of one component do not give satisfactory results. coding as in Fig. 1. While changes in brightness from bin to bin For example, an absorbed powerlaw gives a reduced chi are, in general, significant the hardness ratios are consistent with square χ2 = 1.71 for 88 degrees of freedom (dof) while aconstant value. r a single mekal gives χ2 = 1.64 (88 dof). The spectrum The high sensitivity of the Chandra gratings allow us to r shows always a pronounced excess below 1 keV which constrainthistemperaturewithunprecedentedaccuracy. can only be modeled when a second component to the Although there are few energy bins below 1 keV, the as- model is added. The best fit is attained with the ad- sociated errors (blue crosses) are very small. The ob- dition of two optically thin collisionally ionized plasmas tained plasma temperature kT = 0.20+0.03 keV is fully 2 −0.02 (apec). The parameters are shown in the fourth col- compatible with the temperature of the cool plasma in umn of Table 2. The low energy range is fitted by a HD110432derivedfromhighresolutionspectra(Torrejón plasma with kT = 0.20+0.03 keV and solar abundance. et al. 2012) and significantly lower than that obtained −0.02 4 Torrejón et al. The parameters are shown in the third column of Ta- TABLE 2 ble 2. However, in terms of χ2 the two apec model is Model parametersfor Chandra HETG data. significantly better. Replacing the collisionally ionized plasmas (apec) by photoionized plasmas (mekal), gives Component Parameter Value slightly worse fits. For example, the two mekal model PL+APEC 2APEC gives χ2 = 1.38 (86 dof) while powerlaw + mekal gives r NHa 0.15+−00..1053 0.20+−00..1053 mχ2rod=el1co.4m9p(o8s6eddboyf).twoInapcoencccluosmiopno,nethnetsfiusllfyavtohreerdmbayl powerlaw Norm 0.00026+−00..00000054 ... the present data set. Γ 1.36+−00..1051 ... The most conspicuous characteristic of the spectrum is the lack of any discernible emission line. In particu- apec1 Norm ... 0.0018+−00..00000056 lar, no Fe Kα line is seen and even the flux at 6 keV is kT (keV) ... 15.7+−45..74 below the 7 σ level. There is flux up to 7 keV at the 3σ level; however, there is certainly not an Fe emission line apec2 Norm 0.0003+−00..00000031 0.0005+−00..00000012 evenat1σ. Finally,theresidualspresentsomestructure kT (keV) 0.21+−00..0023 0.20+−00..0032 between 1 and 2 keV that could be due to the presence of emission lines. Indeed, this is the line rich region for Fluxb 1.78 1.81 highly ionized plasmas. The HETG spectra of bright γ χ2 (dof) 1.48(86) 1.35(86) Cas analogs display several strong emission lines in this r region(Smithetal. 2004,Fig. 2;,forγ Cas;Torrejónet a Inunitsof×1022 cm−2 al. 2012,Fig. 7and8,forHD110432). StronglinesofNe b Unabsorbed0.5–8keVfluxinunitsof×10−12 ergs−1 cm−2 x Lyα (1.02 keV) and Si xiv Lyα (2.00 keV) as well as weakerlinesofFexxiv (1.17keV)andMg xiiLyα(1.47 keV) might be expected in this area. However, none of 4 these lines show up in the spectrum of HD119682 with − 0 EW upper limits of the order of 1 eV. 1 V × The high sensitivity of Chandra HETG at low ener- e 5 gies also allows us to study the absorption column with k s/ −4 unprecedented detail. The analysis of brighter mem- / 0 2 1 bers of this class have shown that the different spectral m × componentsareaffectedby differentabsorptioncolumns c 2 s −4 (Smith et al. 2004, for γ Cas; LO07 and Torrejón on 10 kT=0.20 keV et al. 2012, for HD110432). Here we have tried the t same strategy. The values of the resulting columns turn o 5 kT=15.7 keV h 0− out to be identical, within the errors, for all models. P 1 Therefore a single photoelectric absorption satisfacto- × 5 rilydescribes the spectrum ofHD119682. The measured 5 NH =(0.20+−00..1053)×1022 cm−2 is very low, in agreement with R06 and S07. χ 0 5. DISCUSSION 5 The spectrum of HD119682 differs from other γ Cas − analogs observed at high resolution mainly in that it 0.5 1 2 5 shows no discernible emission lines. Particularly strik- ing is the fact that no Fe complex is observed. Despite Energy (keV) this, the continuum spectral parameters of HD119682 Fig.3.— The ChandraHETG unfolded spectrum of HD119682 are very similar to those of γ Cas (Smith et al. 2004, showingtherelativecontributionsofthelowandhightemperature Fig. 2) and HD110432 (Torrejón et al. 2012). Consis- plasmasrespectively. Alsoshown(inred)isthebestfitmodel. tently with these previousworks,the correctdescription of the spectrum is attained with the addition of several by R06 (kT = 1.07± 0.16 keV; Raymond-Smith) and thermal components. In the case of HD119682 only two S07 (kT = 1.7±0.3 keV; mekal). The absorption col- are needed with characteristic temperatures in the ∼ 2 umn, however, remains essentially compatible (see be- MK and ∼ 150 MK range. Both components are mod- low). Therefore, the differences in kT could be due to ified by a single absorption column NX ∼ 0.20×1022 intrinsic long term variability of the source. This vari- H cm−2. This column would imply an interstellar ex- ability in the cool plasma temperature seems also to be present in γ Cas (Lopes de Oliveira et al. 2010) while cess E(B −V) = NHX/6.83× 1021 = 0.29+−00..0031 (Ryter the hot plasma temperature appears to be more stable. 19963) in good agreement with the interstellar medium The high energy range requires a second, much hot- (ISM) absorption towards NGC 5281 (Moffat & Vogt ter apec model with kT = 15.7+4.7 keV. This tempera- 1973, E(B − V) = 0.26 ± 0.03; Marco et al. 2009, −5.4 tureiscompatiblewithinthe errorswiththatderivedby E(B−V)=0.28±0.02). R06 and S07. Alternatively, in agreement with previous works on γ Cas candidates, the high energy range can 3 The errors represent the propagated uncertainty in the mea- be fit with a powerlaw with a photon index Γ = 1.36. suredNHX Chandra observations of HD119682 5 HD119682 has been observedseveraltimes both in X- is absorbed by material with a column density compati- rays (XMM-Newton August 2001 and Chandra Decem- ble with the circumstelar disk. If this is also correct for ber 2004, R06, S07; December 2008, this work) and in HD119682, then the NX ≈ NISM implies that the sys- H H the Hα light (June 2003 and February 2004; R06, S07) temmustbeobservedathighinclination. Thiswouldbe showing always a strong emission. Although the time consistent with the single peaked emission lines seen in coverageisstillpoor,the sourceappearstobepersistent the optical spectrum of this object (S07). in both energy bands. Assuming that the circumstellar R06andS07reporthintsofFeKαemission. However, envelope(andhencetheHαemission)wasalsopresentat no sign of such emission line (nor any other) is present the epoch of our Chandra observation, the X-ray source here. Thus, the fluorescence emission from neutral Fe was not obscured by the disk material. The same would seemstobevariableinthelongterm. Ontheotherhand be true for the observations analyzed in previous works the0.5-8.5keVX-rayluminosityofthesourcehasvaried (R06,S07). Thecircumstellardisk(confinedtotheequa- withinamaximumfactorof∼2inaperiodofsevenyears torial plane) does not intercept the X-ray emission. In (R06, S07, this work). One of the possible sites for the the framework of the accreting binary system, the com- Kα reprocessing of the X-rays is the circumstelar disk. pact object (the WD) should not be deeply embedded Smith et al. (2012)find a direct correlationbetween the into the circumstellar disk; yet it should be accreting absorption column of the hard emission and the EW(Fe enough material to power persistently the X-ray source Kα) (as well as with the reddening) for γ Cas (Table 5 up to a non negligible LX ∼ 3.8×1032 ergs−1. This ofthatreference)implyingthatmost,perhapsall,ofthe would require a very special tuning of the astrophysical X-rayemissionis producedbehind the circumstellardisk parameters of the system. Yet, even though rare, this material. These authors also deduced an inclination of situation can happen: 4U2206+54 is a low luminosity i=[43±1]oforγ Cas. InthecaseofHD119682,however, (L ≈1035−36 ergs−1)HMXBwithaNSaccretingfrom the absorption column remains always compatible with X the wind of a massive O9.5V star while the X-ray spec- theISMvaluedespitethechangesintheX-raybrightness trum presents very few (if any) local absorption (Reig of the source and variable Fe Kα emission because the et al. 2012). Contrary to the case of 4U2206+54, how- circumstellardisk,confinedtotheequatorialplane,never ever,nopulsationshavebeenfoundforHD119682orany intercepts the X-ray emission. otherγ Casanalogsofar. Incontrast,within the frame- HD119682 would thus appear as a pole-on γ Cas, an work of the standalone active star, the single absorption importantbenchmark to explorethe phenomenonwhere column, compatible with the ISM value, can be simply theplasma,anditsvariations,couldbedirectlyobserved understood if the system is seen at very high inclina- without any intervening circumstellar material. tion angles. It has been recently found that the hard The authors acknowledge the constructive criticism X-raycomponentofthe multi-thermalemissionofγ Cas of an anonymous referee. This work has been sup- (Smithetal. 2012)andHD110432(Torrejónetal. 2012) ported by the Spanish Ministerio de Ciencia e Inno- isverylikelyproducednearthesurfaceoftheBestarand vacioń (MICINN) through the grants AYA2010-15431 and AIB2010DE-00057. REFERENCES BalbusS.A.,HawleyJ.F.,1998, RvMP70,1 Rakowski,C.E.,Schulz,N.S.,Wolk,S.J.,Testa,P.2006,ApJ575, Canizares,C.,Davis,J.E.,Dewey,D.etal.2005,PASP117,1144 435(R06) Haberl,F.1995,A&A296,685 Reig,P.,Roche,P.1999,MNRAS306,100 HouckJ.C.,DenicolaL.A.,2000,inASPConf.Ser: Reig,P.,Torrejón,J.M.,Blay,P.2012, MNRAS425,595 AstronomicalDataAnalysisSoftwareandSystemsIX,9,591 Ryter,Ch.E.1996,ApSS236,285 KuboS.,MurakamiT.IshidaM.,CorbetR.H.D.1998,PASJ50, RobinsonR.D.,SmithM.2000, ApJ540,474 417 RobinsonR.D.,SmithM.,HenryG.W.2002,ApJ575,435 LopesdeOliveiraR.,MotchC.,SmithM.,NegueruelaI,Torrejón Safi-HarbS.,Ribo´M.,ButtY.etal.,2007, ApJ659,407(S07) J.M.,2007,A&A474,983(LO07) SmithM.,CohenD.H.,FengGuM.RobinsonR.D.,Evans N.R., LopesdeOliveiraR.,SmithM.,MotchC.,2010,A&A512,A22 SchranP.G.2004,ApJ600,972 LopesdeOliveiraR.,MotchC.,2011ApJ731,L6 SmithM.,BalonaL.,2006,ApJ640,491 Marco,A.,Negueruela,I.Ribo´,M.,Lorenzo,J.,Motch,C.2009, Smith,M.,Lopes deOliveira,R.,Motch,C.etal.,2012,A&A AdSpR44(3), 348 540,A53 Moffat,A.F.J.,Vogt,N.1973,A&AS10,135., Torrejón,J.M.,Schulz,N.S.,Nowak,M.A.,2012ApJ750,75 MotchC.,LopesdeOliveiraR.,NegueruelaI.,HaberlF., Janot-Pacheco E.,2007ASPC361,117 Wilms,J.,Juett,J.,Schulz,N.S.,Nowak,M.2012,inpreparation OwensA.,OosterbroekT,ParmarA.N.,Stu¨weJ.A.,HaberlF., http://pulsar.sternwarte.uni-erlangen.de/wilms/research/tbabs/ 1999, A&A348,170