Discovery of X-ray Pulsations from the HMXB Source AXJ1749.1-2733

0.46 MB·English

by  D. I. Karasev

Mon.Not.R.Astron.Soc.000,1–6(2007) Printed2February2008 (MNLATEXstylefilev2.2)
Discovery of X-ray Pulsations from the HMXB Source
AXJ1749.1-2733
D. I. Karasev1⋆, S. S. Tsygankov1, A. A. Lutovinov1
8 1Space Research Institute, Profsoyuznaya Str. 84/32, Moscow 117997, Russia
0
0
2

n Accepted ....Received....
a
J

1 ABSTRACT
2

WearereportingadiscoveryofX-raypulsationsfromthesourceAXJ1749.1-2733
h] with the period of ∼132 s based on the XMM-Newton data obtained in March 2007.
p The observed pulse profile has a double-peaked structure with the pulse fraction of
- about25−30%inthe3-10keVenergyband.Wehavealsofoundthataperiodicitywith
o
practically the same period has been detected from the source by the IBIS telescope
r
t onboard the INTEGRAL observatory during an outburst on Sept. 9, 2003 in the 20-
s
50 keV energy band. Due to the double-peaked pulse profile, there is an additional
a
[ peak on both periodograms of nearly ∼ 66 s, therefore we have also investigated the
possibility that the last value is the true pulse period. The source spectrum obtained

1 by the XMM-Newton observatory in the soft energy band is being heavily absorbed
v (NH ≃2×1023 cm−2) due to a strong intrinsic absorptionin the binary system that
7
leads to the conclusion that AX J1749.1-2733 is a new transient X-ray pulsar in the
4
high mass X-ray binary system.
2
3
Key words: X-ray:binaries – (stars:)X-ray pulsars:individual – AX J1749.1-2733
.
1
0
8
0
: 1 INTRODUCTION tory–RA=17h49m10s,Dec=-27◦.33′14′′(J2000)theuncer-
v ′′
tainty of about 55 , Sakanoet al. (2002). Subsequent anal-
i The source AX J1749.1-2733 was discovered on Sept. 19,
X ysis showed that the source had been significantly detected
1996 by the ASCA space observatory (Sakanoet al. 2002).
byINTEGRALfromtimetotimein2004-2007,butthetyp-
ar The object had not been detected during previous observa- ical value of its flux was about 5−8 mCrab. Nevertheless,
tionsofthisregion in1995evenwithahigherexposureand
based on these data, Zurita Heras et al. (2007) suggested
wassignificantlydetectedagainonlyinSeptember1997and the presence in the system of the ∼185-day orbital period.
March 1998, which reflects its transient nature. Due to the
faintnessofthesourceemission(1.5−6×10−12 ergcm−2s−1 Observations of AX J1749.1-2733 performed in soft X-rays
bytheSWIFTobservatoryin2007, allowed Kong(2007)to
in the 0.7−10 keV energy band),it was practically impos- ′′
improveitslocalizationto3.8 andrevealedastrongabsorp-
sible to obtain good restrictions to its spectral parameters, tion (NH ≃1.9×1023 cm−2) in thesource spectrum.
but some evidence for the strong absorption in the system
In this work, we reanalyzed the INTEGRAL data ob-
was mentioned (Sakanoet al. 2002).
tainedduringtheoutburstinSeptember2003andanalyzed
On Sept. 9, 2003 a short (with a duration of ∼ 1 day)
the XMM-Newton data obtained in March 2007. In both
outburst from the source was observed by the INTEGRAL
datasets,strongperiodicmodulationsoftheX-rayfluxwith
observatoryinhardX-rayswiththepeakfluxof∼40mCrab the period of ∼132 s were found. The preliminary results
inthe20-60keVenergyband(Grebenev2004).Detailedin-
were reported byKarasev et al. (2007).
vestigations of the source behavior and its characteristics
duringtheoutburstallowed Grebenev & Sunyaev(2007)to
classify AX J1749.1-2733 as a fast X-ray transient. During
the outburst, the source position in hard X-rays was re- 2 OBSERVATIONS AND DATA ANALYSIS
◦ ′ ′′
solved as RA=17h 49m 07s, Dec=-27.3238 (J2000) with
the uncertainty of 1.8′ (Sguera et al. 2006), that is coinci- The source AX J1749.1-2733 was observed with the XMM-
dentwiththesourcepositionrevealedbytheASCAobserva- Newton observatory on March 31, 2007. Our analysis was
based in general on the data from the EPIC-PN camera,
butthedatafromtheMOStelescopeswereusedalsoforthe
⋆ E-mail:[email protected] testoftheconsistencyoftheusedspectralmodels.Alldata
2 D. I. Karasev, S. S. Tsygankov, A. A. Lutovinov
Figure 2. Periodograms of background subtracted emissions of
AXJ1749.1-2733(a)andAXJ1749.2-2725(b)obtainedwiththe
INTEGRAL/IBISdatainthe20-50keVenergyband.
Table 1. ListofAXJ1749.1-2733 pulseperiods
Figure 1.Periodogramsofabackgroundsubtractedemissionof
AX J1749.1-2733 (a) and background region (b) obtained with Date Observatory PulsePeriods
the XMM-Newton observatory (the PN instrument) in the 6-10
keVenergyband. Sept9,2003 INTEGRAL 131.54±0.02,65.77±0.01
March31,2007 XMM-Newton 131.95±0.24,66.05±0.15
3 TIMING ANALYSIS
were processed with the Science Analysis System (SAS)1.
Weprovidedastandardmethodforthefiltrationofthepro- Fig.1a shows a periodogram of a background subtracted
ton flares produced as a result of the interaction between emission of AX J1749.1-2733 obtained from the XMM-
soft protons in the Earth’s magnetosphere with the tele- Newton/PN data in the 6-10 keV energy band. The sig-
scope. The source spectrum and lightcurve were extracted nificant excess (∼ 6σ) of the χ2-distribution at ∼132 s can
′′
fromthecirclewiththeradiusof14 aroundthesource;the be clearly seen. Apart from them, there are several other
background spectrum and lightcurve were extracted from a peaks of a lower intensity, which are mostly prominent at
nearbycircle region with thesame radius. Atotal exposure the periods of ∼66 s (the significance is ∼3σ) and ∼198 s,
of XMM-Newton/PN observations was approximately 6 ks. whichcorrespondstoahalfandoneandahalfofthe∼132-s
As mentioned above, the source AX J1749.1-2733 had period,respectively.Toclarifythattheseexcessesarenotar-
been significantly detected by the IBIS telescope several tificial and connected with the source AX J1749.1-2733, we
times during the four years of observation, but only once havealsobuiltaperiodogramforemissionsfromthenearby
was its intensity enough to detect the pulsations. The to- background region. No signals with the periods of 66, 132
tal exposure of the INTEGRAL/IBIS observations during or 198 s were detected on this periodogram (Fig.1b). This
this outburst on Sept 9, 2003 was about 50 ks. The INTE- fact indicates that the origin of the detected periodicity is
GRAL/IBISdataforthetiminganalysiswereprocessedus- connectedwiththesourceAXJ1749.1-2733. Thepulsations
ingthestandardsoftwareOSA6.0,distributedbytheINTE- inthesourceemissionwascheckedalsoinhardX-rays(>20
GRALScienceDataCenter,Versoix,Switzerland2,andthe keV) using the data of the IBIS telescope onboard the IN-
softwaredesignedandsupportedbytheNationalInstituteof TEGRAL observatory. In Fig.2a, the periodogram of the
Astrophysics in Palermo, Italy3 (Segreto & Ferrigno 2007). 20-50keVsourceemissionisshownneartheperiodsof∼66,
Forthespectralanalysis,weusedthesoftwaredevelopedat ∼132, and ∼198 s. Similarly to the XMM-Newton results,
theSpaceResearchInstituteRAS,Moscow,Russia(descrip- the signal can be clearly seen again, but the significances
tion of its main features can be found in Revnivtsevet al. of the ∼66 and ∼ 132 s peaks in this case are roughly the
(2004) and Krivonos et al. (2007)). The final timing and same(∼15−16σ).Thelistofpossiblepulseperiodsnear66
spectralanalysiswereprovidedwiththeFTOOLSpackage4. and 132 s measured with different instruments is presented
inTable1.Errors(correspondingto1σ)weredeterminedby
the bootstrap method from the analysis of a large number
of simulated light curves (see e.g. Tsygankov & Lutovinov
(2005) for details).
It is important to note that in a near vicinity to the
1 http://xmm.esac.esa.int/sas/
2 http://isdc.unige.ch studiedsource thereis located anothertransient X-raypul-
3 http://www.pa.iasf.cnr.it/∼ferrigno/INTEGRALsoftware.html sar AX J1749.2-2725 with the pulse period of ∼ 220 s
4 http://heasarc.gsfc.nasa.gov/lheasoft (Torii et al. 1998). Due to a small angular distance (∼8′)
Discovery of X-ray Pulsations from the HMXB Source AXJ1749.1-2733 3
Figure4.(a)PulseprofilesofAXJ1749.1-2733obtainedwiththe
INTEGRAL/ISGRI data and folded with two different periods:
131.54 s (thick line) and 65.77 s (thin line). The second one has
beenplottedtwicetodemonstrateadiscrepancybetweenprofiles
onthetimescaleof131.54s.(b)Thesame,buttheperiods131.95
s and 66.05 s have been obtained with the XMM-Newton/PN
data.
observatories and folded with the period of ∼132 s are pre-
sented for several energy bands: 3-6, 6-10, 20-30, 30-40 and
40-60 keV; due to slightly different values of the pulse pe-
riod,thedataofbothobservatorieswereconvolvedwiththe
correspondingperiods(seeTable1).Thesourcepulseprofile
builded in such a way has a clear double-peaked structure
at all energies that account for several peaks in the peri-
odograms with intervals divisible by ∼66 s.
Figure3.TheAXJ1749.1-2733pulseprofilesinthe3-6,6-10,20- To clarify which of pulse periods is true, we used the
30, 30-40 and40-60 keV energy bands obtained withthe XMM- following procedure: the source pulse profile folded with
Newton/PN (two upper panels) and INTEGRAL/IBIS (three the period of 131.54 s and the doubled source pulse profile
bottompanels)instrumentsforthepulseperiodsof131.95sand folded with the period of 65.77 s obtained from the INTE-
131.54s,respectively. Errorscorrespondto1σ. GRAL/IBIS data in the 20-60 keV energy band were plot-
ted in the same figure and compared between themselves
(Fig.4a). The test clearly demonstrates that there are no
significant differences between the pulse profiles obtained
between these sources, there is a possibility of a “photon-
for these two periods, which might be due to a relatively
drifting” from one source to another in the processing of
faintemissionfromthesourceandlargeuncertaintiesinthe
thedatareduction oftheIBIStelescope (asitsangularres-
olution is only 12′). To check a possible influence of such observed count rates. To formalize this test, weused a sim-
an effect, we have done the same timing analysis for AX ple criteria in the form of χ2 = Pki=1 (Xiσ−i2µi)2, where Xi
J1749.2-2725. Fig.2b shows the resulting periodogram for is the count rate in the ith bin of the 132 s folded light
this source. It can be seen that significant peaks are de- curve, µi is the count rate in the ith bin of the 66 s folded
tectedneitherat ∼132snorat ∼220s(notethatnopeak light curve, which is being used here as a model; and σi is
near 220 s can also be seen on the AXJ1749.1-2733 peri- thecorrespondenceuncertainty.Asummationisbeingdone
odogram, Fig. 2a). Most likely,thepulsarAXJ1749.2-2725 only on the half of bins (which are statistically indepen-
was in the “switched-off” state during the INTEGRAL ob- dent) of the 132 s folded light curve (in our case, k = 9).
servations(Grebenev & Sunyaev2007)andassuch,didnot The application of this test to the INTEGRAL/IBIS data
affect our results. leads to the value of 6.8 for 9 d.o.f., which means that the
Thus,thepulsationsinthesourceAXJ1749.1-2733can differenceisinsignificant.Thisresultreflectsthepractically
beclearlyestablished.However,thequestionaboutthetrue equal significance of the 66-s and 132-s peaks at the corre-
valueofitsperiodisstillrelevant.InFig.3,thesourcepulse spondence periodogram (Fig.2a). The same procedure was
phase curves obtained by XMM-Newton and INTEGRAL used for the XMM-Newton data obtained in the 6-10 keV
4 D. I. Karasev, S. S. Tsygankov, A. A. Lutovinov
Figure5.Pulsefractiondependenceontheenergyforthepulse
periodof131.95and131.54sfortheXMM-Newton(circles)and
INTEGRAL(squares)observatories,respectively.
Figure6.BroadbandenergyspectrumofAXJ1749.1-2733inthe
energy band.Wefolded them again with thecorresponding
3-100keVenergyband;XMM-Newton(crosses)andINTEGRAL
periods(seeTable1)andfoundthatthedifferencebetween
(circles)datawerecombinedusingtherelativenormalization.
the pulse profiles in this energy band is more significant
(Fig.4b). Itsvalue(22 for 9 d.o.f.) corresponds totheprob-
abilityof∼0.01thatthisdifferenceiscausal.Notethatthe bined spectral analysis is possible only assuming the un-
suggested procedure is formally equivalent to the compari- changed shape of the source spectrum. However, as it was
son of the odd and even 66 s segments on the 132 s folded shownbyGrebenev & Sunyaev(2007),thesourcespectrum
light curve.Thus, the useof theXMM-Newton data brings was changing significantly even duringtheoutburst,all the
ustotheconclusionthat∼132sistheprobablepulseperiod more, it can be different in the quiescent state. Therefore,
of the source AX J1749.1-2733. hereweareprovidingtheresultsofthespectralanalysissep-
During the XMM-Newton observations, the pulse pe- arately for the XMM-Newton data and only for the rough
riod was slightly longer than during the INTEGRAL ones. estimation of thegeneral spectral shape and its parameters
Due to a relatively high uncertainty of the XMM-Newton ina wideenergy bandwheretheXMM-Newton and INTE-
period measurements, this difference is not very significant GRAL data were combined using relative normalization.
(∼2σ),butif wesuggest that itbeconnectedwith thereal Thesourcespectrumobtainedwith XMM-Newton/PN
deceleration of the neutron star rotation, the formal P˙/P canbewelldescribedbyasimplepowerlawwiththephoton
willbeabout10−3 yr−1,whichcouldbeconsistentwiththe index Γ = 1.17+−00..2036, the photoelectric absorption NH =
typical rates of the pulse period change for X-ray pulsars 21.1+2.7 ×1022 cm−2 and the unabsorbed source flux of ∼
−1.3
(see e.g. Bildsten et al. (1997); Lutovinovet al. (1994)). 1.8×10−11 erg cm−2 s−1 in the 3−10 keV energy band
Foralltheenergybands(Fig.3),wecalculatedthepulse (note that the use of the XMM-Newton/MOS data gives
fraction, which is defined as PF = (Imax−Imin)/(Imax+ approximately the same results).
Imin),whereIminandImaxarebackground-correctedcount The source spectrum in the soft X-ray band (2−10
rates at the pulse profile minimum and maximum. Fig.5 keV)hadpreviouslybeenmeasuredwiththeXRTtelescope
shows the energy dependence of this value for both obser- oftheSWIFTobservatory(Romano et al.2007;Kong2007)
vatories. It is clear that the pulse fraction is relatively con- andtheASCAobservatory(Sakano et al.2002).Itwasalso
stant in theenergy band from ∼3 to ∼40 keVat thelevel approximatedbythepowerlawmodelwiththephotoelectric
of ∼ 25−30% and slightly increases up to ∼ 50% in hard absorption.ThevaluesofNH complywiththeresultsofour
X-rays (> 40 keV), which is typical for X-ray pulsars (see analysis, but thevaluesof Γ are different,which can be the
e.g. Tsygankov et al. (2007)). result of the spectral shape variations similarly as it was
found by Grebenev & Sunyaev (2007) for the outburst. On
the other hand, it is necessary to note that uncertainties
4 SPECTRAL ANALYSIS of the photon index measurements with the SWIFT and
ASCAobservatoriesareverylargeand,nominallyspeaking,
It is important to note that the INTEGRAL and XMM- ourresults fall within thecorresponding error boxes.
Newton data were obtained not only in different epochs A rough estimation of the interstellar absorption
but also in different source states — INTEGRAL had ob- in the direction to the source using the NH map
served the source during the outburst in September 2003 (Dickey& Lockman 1990) gives the value of NH ≃ 1.63×
and XMM-Newton had observed the source in March 2007 1022 cm−2,whichismorethanoneorderofmagnitudelower
most likely in a quiescentstate (seebelow). Thus, thecom- thanthevalueobtainedfromourspectralanalysis.Basedon
Discovery of X-ray Pulsations from the HMXB Source AXJ1749.1-2733 5
these measurements and previous results of the ASCA and tions were performed ∼3.8 later. As the 2003 outburst du-
SWIFT observatories, we can undoubtedly claim that AX ration was about 1 day (Grebenev & Sunyaev 2007), it is
J1749.1-2733 is a heavily absorbed source, which is proba- veryunlikely,forthesuggestedorbitalparameters,thatthe
bly connected with the massive counterpart in the binary XMM-Newton observations were performed also during an
system (likely a supergiant or a giant). outburst.TakingintoaccountthemeasuredratiooftheIN-
The formal approximation of the XMM-Newton and TEGRAL and XMM-Newton fluxes(6 times), it is possible
INTEGRAL data simultaneously with the ”standard” pul- to combine theobtained results to suppose that theorbital
sars model (Whiteet al. 1983) modified by the photoelec- periodisslightlylonger,∼185.5days.However,thissuppo-
tric absorption gives the photon index of Γ = 1.03+0.18, sition requires an additional study.
−0.24
the cutoff energy Ecut = 7.1+−00..69 keV, the folding en- Thisquestionaswellastheoneaboutthenatureofthe
ergy Ecut = 19.8+−12..96 keV, the absorption column NH = system will be addressed in detail in a separate paper with
(20.2+1.0) × 1022 cm−2 with the corresponding value of dataoftheAXJ1749.2-2733specialobservationsperformed
−1.9
the reduced χ2=0.94 (Fig.6). Note, that the obtained val- with the Russian-Turkish Telescope (RTT-150) in August
ues and the spectrum shape are typical for X-ray pulsars 2007 (Karasev at al., 2008, in preparation).
(see e.g. Filippova et al. (2005)) despite the roughness and
a certain randomness of such a combination of the XMM-
NewtonandINTEGRALdata.Thebroadband3−100keV
6 ANKNOWLEGEMENTS
X-ray fluxes corresponding to the XMM-Newton and IN-
TEGRAL epochs of observations differ approximately by We would like to thank E.M.Churazov for developing the
six times: ∼4.9×10−11 ergscm−2s−1 versus∼2.8×10−10 IBIS data analysis algorithms, providing the software and
ergs cm−2s−1, respectively. The last value corresponds to the discussion of the timing analysis results. We extend
the results of Grebenev & Sunyaev (2007) obtained from a our gratitude to S.A.Grebenev and R.A.Sunyaev for the
combined analysis of the INTEGRAL/JEM-X and INTE- discussion of the results and for the useful remarks. We
GRAL/IBIS data. would also like to pay tribute to the very useful comments
of an anonymous referee, which allowed us to significantly
improve the paper. We used data from the archive of the
GoddardSpaceFlight Center(NASA),theIntegralScience
5 SUMMARY
Data Centre (Versois, Switzerland), and the Russian Sci-
We are reporting a discovery of X-ray pulsations from AX ence Data Center for INTEGRAL (Moscow, Russia). This
J1749.1-2733 bytheINTEGRALand XMM-Newton obser- work was supported by the Russian Foundation for Basic
vatorieswiththeperiodsof131.54sinSeptember2003and Research (Project No. 07-02-01051), the Presidium of the
131.95 s in March 2007, respectively. The pulse profile has Russian Academyof Sciences(TheOrigin andEvolution of
a double-peaked shape in a wide energy band (3−60 keV) StarsandGalaxiesProgram),andtheProgramofthePres-
andthepulsefractionslightlyincreaseswiththeenergyfrom ident of the Russian Federation for the Support of Scien-
∼25−30%to∼50%.Thesourcespectrumcanbedescribed tifcSchools(Project No.NSh-1100.2006.2). A.A.Lutovinov
by the powerlaw model with the photoelectric absorption, would like to thank the Russian Science Support Founda-
the value of which is much higher than the interstellar one tion as well. AL and ST thank International Space Science
in the direction to the source. Thus, we can conclude that Institute (ISSI, Bern, Swiss) for the hospitality and partial
AX J1749.1-2733 must be a new transient X-ray pulsar in support.
the high-mass X-ray binary system with a strong intrinsic
absorption likeIGR J16465-4507 (Lutovinovet al. 2005) or
AX J1841.0-0536/IGR J18410-0535) (Halpern et al. 2004). REFERENCES
It is interesting to compare the AX J1749.1-2733 posi-
tion with the position of different classes of binary sources Bildsten, L., Chakrabarty D., Chiu J., 1997, Astrophys
on theCorbet Porb−Ps diagram (Corbet 1986).According Journal, 113, 367
to the relation between the orbital period of ∼ 185 days Corbet R., 1986, MNRAS,220, 1047
suggest by Zurita Heras et al. (2007) and the source pulse Dickey J., Lockman F., 1990, Ann.Rev.Ast.Astr. 28, 215
period of 132 s, this source can belong to the class of Be- Filippova E.V., Tsygankov S.S., Lutovinov A.A., Sunyaev
systems with the orbital eccentricity of ∼ 0.3 (see eq.[1] of R.A.,2005, Astron.Letters, 31, 729
Corbet (1986)).But themeasured valueof theintrinsic ab- GrebenevS.A.,2004,Reportatthe5thINTEGRALWork-
sorption in the binary system is repeatedly high and larger shop,Munich, February 16-20, 2004
than it is usually observed in Be-systems and is typical for Grebenev S.A., Sunyaev R.A., 2007, Astron.Letters, 33,
supergiant systems. 175
Note that the INTEGRAL and XMM-Newton obser- Halpern J., Gotthelf E., Helfand D., 2004, Astron. Tele-
vations are separated in time by an approximately inte- gram, 289, 1
ger of the suggested orbital periods (∼ 7). Therefore we Karasev D., Tsygankov S., Lutovinov A., Churazov E.,
checkedthepossibility that theXMM-Newton observations SunyaevR., 2007, Astron. Telegram, 1245, 1
might have been performed also during an outburst. Using Kong A., 2007, Astron. Telegram, 1039, 1
the moment of the 2003 outburst maximum MJD52891.65 Krivonos,R.,Revnivtsev,M.,Lutovinovetal.,2007,A&A,
(Grebenev & Sunyaev2007)andtheorbitalperiod valueof 475, 775
185 days, we determined the moment of the 2007 outburst LebrunF.,LerayJ.P.,LavocatP.,etal.,2003,A&A,411,
maximum as MJD 54186.65. The XMM-Newton observa- L141
6 D. I. Karasev, S. S. Tsygankov, A. A. Lutovinov
Lutovinov A., Grebenev S., Pavlinsky M., Sunyaev R.,
1994, Astron. Lett., 20, 538
Lutovinov A., Revnivtsev M., Gilfanov M., et al., 2005,
A&A,444, 821
Remillard R.A.,Smith D.A.,2002, IAUCirc. 7880
Revnivtsev M., Sunyaev R., Varshalovich D., et al., 2004,
Astron.Lett., 30, 382
Romano P., Mangano V., 2007, Astron.Telegram, 1040, 1
SakanoM.,KoyamaK.,MaedaY.,2002, Astrophys.Jour-
nal, 138, 19
Segreto A.,Ferrigno C., 2007, arXiv:astro-ph/0709.4132
Sguera V., Bazzano A., Bird A. et al., 2006, Astrophys.
Journal, 646, 452
Tsygankov S., LutovinovA, 2005, Astron.Letters, 31, 88
Tsygankov S., Lutovinov A., Churasov E., Sunyaev R.,
2007, Astron. Letters, 33, 368
Torii K., Kunugasa K., Katayama K., 1998, The Astro-
physical Journal, 508, 854
White N., Swank J., Holt S., Astrophis. Journal, 270, 711
WinklerC.,CourvoisierT.J.-L.,DiCoccoG.,etal.,A&A,
411, L1
Zurita Heras J.A., Chaty S., Rodriguez J., 2007, Astron.
Telegram, 1035, 1