Spectral and timing characterization of the X-ray source 1RXS J194211.9+255552

1.3 MB·

by  A. D'Aì

Mon.Not.R.Astron.Soc.000,1–5() Printed14January2016 (MNLATEXstylefilev2.2)
Spectral and timing characterization of the X-ray source
1RXS J194211.9+255552
A. D’A`ı1,(cid:63) G.Cusumano1, V. La Parola1, A. Segreto1
1 INAF - Istituto di Astrofisica Spaziale e Fisica Cosmica, Via U. La Malfa 153, I-90146 Palermo, Italy
6
1
0 14January2016
2

n
a ABSTRACT
J We report on the first spectral and timing characterization of the transient X-ray

3 source 1RXS J194211.9+255552 using all available data from the Swift X-ray satel-
1 lite.Weused10yearsofhardX-raydatafromtheBurstAlertTelescope,tocharacter-
ize its long-term behaviour and to search for long periodicities, finding evidence for a

] periodic modulation at 166.5±0.5 days, that we interpret as the orbital period of the
E
source. The folded light curve reveals that the X-ray emission is mostly concentrated
H
inarestrictedphase-intervalandweproposetoassociate1RXSJ194211.9+255552 to
. theclassoftheBeX-raybinaries.Thisisalsosupportedbytheresultsofthespectral
h
p analysis, where we used the BAT data and three pointed Swift/XRT observations to
- characterize the X-ray broadband spectral shape. We found mild spectral variability
o in soft X-rays that can be accounted for by a varying local neutral absorber, while
r
the intrinsic emission is consistent with a hard power-law multiplied by a high-energy
t
s exponential cut-off as typically observed in this class of systems.
a
[
Keywords: X-rays:binaries–X-rays:individual:1RXSJ194211.9+255552.Facility:

1 Swift
v
6
4
2
1 INTRODUCTION We present here the first complete spectral and timing
3
0 study in the X-ray domain of this source using all available
The X-ray source 1RXS J194211.9+255552 (J194211 here-
. data collected by Swift.
1 after)wasfirstreportedintheROSATAll-SkyFaintSource
0 Catalogue1, at RA=295.54959 and Dec.=25.93125 with a
6 positional uncertainty of 18 arcsec and at a flux level of
1 (1.8±0.7)×10−13 erg cm−2 s−1. Probably due to its rela- 2 DATA REDUCTION
:
v tive faintness, the source has been thereafter poorly stud- ThesurveydatacollectedwithSwift/BATbetweenDecem-
i ied. At the end of December 2011, J194211 was clearly ber2004andJune2014wereretrievedfromtheHEASARC
X
detected during the INTEGRAL Galactic Plane Scan- publicarchive2 andprocessedusingasoftwarededicatedto
r ning in soft X-rays at 26±7 mCrab (Chenevez et al. the analysis of the data from coded mask telescopes (Seg-
a
2011). A follow-up observation with Swift/XRT allowed reto et al. 2010). The source is detected with a significance
for the first determination of the X-ray spectrum, con- of 5.8 standard deviations in the 15–150 keV energy band,
sistent with an absorbed (Galactic equivalent absorption but the signal is optimized in the 15–45 keV energy band
column 1.3±0.6×1022 cm−2) power-law of photon-index whereweobtainedasignificanceof7.0standarddeviations.
Γ=0.64 and an unabsorbed flux in the 1–10 keV range of Fig.1showsthesignificancemapoftheBATskyregion
7.88×10−11 erg cm−2 s−1 (Sidoli et al. 2011). The source around J194211 in this energy band. We extracted a back-
positional error lead to the identification of two possible groundsubtractedlightcurveinthe15–45keVband,binned
near infra-red counterparts, and Sidoli et al. (2011) pro- at15-daysresolutiontocheckthelong-termsourcevariabil-
posedtheassociationwiththebrighterone,the2MASSstar ity in hard X-rays (Fig. 2). The light curve clearly shows
19421116+2556056, whose optical spectrum (1159-0412038 twodifferentfluxlevels,withasignificanthigherfluxinthe
in the USNO-B1.0 catalogue) is consistent with an early- morerecentyears.Tohaveanestimateofthetimewhenthis
typestar(Masettietal.2012)pointingtothepossiblehigh- enhancedactivitybegan,webuiltthesourcesignal-to-noise
mass X-ray binary nature of the X-ray source. ratio (SNR) as a function of time; we started calculating
the source significance using the entire 10-year time-span;
(cid:63) E-mail:[email protected]
1 http://www.xray.mpe.mpg.de/rosat/survey/rass-fsc/ 2 http://heasarc.gsfc.nasa.gov/docs/archive.html
(cid:13)c RAS
2
Figure 3. 2MASS sky region around J194221. Black circle: our
refined Swift/XRT position. Green circles: the two proposed
infra-redcounterpartsasreportedinSidolietal.(2011).
TheSwift/XRTdatawereprocessedusingtheftools
package with standard procedures (xrtpipeline v.0.12.4),
Figure 1. 15–45 keV Swift/BAT significance map centred on filteringandscreeningcriteria,withstandardgradefiltering
1RXSJ194211.9+255552. 0-12. The source is clearly detected in the three visits and
noothercontaminatingsourceispresentintheSwift/XRT
field of view (see left panel of Fig. 1). To better constrain
the source position, we stacked the three images and we
04 obtained a refined source position at RA=19:42:11.22 and
1 2
xel) x Duseicn.g=th+e25o:n5l6in:0e5.S2w(i1f.9t/aXrcRsTecbeurrioldraptro9d0u%ctcsontofiodle3n,ctehaletveall)-
pi
c/ lows also astrometric and enhanced position determination
e 0
s/s (Goadetal.2007;Evansetal.2014).Thisnewpositionand
ct error results consistent with early estimates (Sidoli et al.
e (
at 2011), but the smaller uncertainty allows us to better sup-
r 2
− port the association of the possible infra-red counterpart
10 with the 2MASS object 19421116+2556056, as shown in
N 5 Fig.3.
S/ 0 A first inspection of the light curves showed mild vari-
5 ability, with averaged count rates per snapshot varying up
− 2006 2008 2010 2012 2014 to a factor of 2. The maximum count rate is ∼0.53 c/s in
year
Obs.ID 00032228001, close to the threshold indicated for
Figure 2.Long-term15–45keVSwift/BATlightcurve(upper a pile-up check4. We verified according to the suggested
panel)andsignal-to-noiseratio(lowerpanel).Bin-timeis15days. pipeline5 that for this Obs.ID a circular region with 4 pix-
MagentaboxindicatesthetimewindowoftheactiveX-raystate.
elsradius(centredatthesourceposition)mustbeexcluded
to avoid spectral pile-up distortions. Therefore, the source
spectrum was extracted froman annular regionof 4and 40
pixels internal and external radius, respectively, centred on
we then progressively removed one week of data from the
the source coordinates as previously determined. A back-
start of the BAT monitoring and obtained the plot of the
ground spectrum was extracted from an annular region of
sourcesignificanceasafunctionofthetimewindow.Weob-
inner and outer radius 90 and 150 pixels, respectively. No
tained the highest significance using the time window from
pile-up correction was found necessary for the other two
2010.6to2014.5(magentaboxinFig.2).TheaverageBAT
Obs.ID, and we used a circular regions of 40 pixels radius
rate before 2010.6 has only an upper limit of 3.6× 10−6
toextractthesourcespectrumandthesameannularregion
counts s−1 pixel−1, that is compatible with zero, whereas
for the background spectrum.
in the period 2010.6–2014.5 the average rate increases to
The XRT ancillary response file generated with
(3.3±0.4)× 10−5 counts s−1 pixel−1. xrtmkarf accounts for PSF and vignetting correction; we
DuringthisactiveX-raystate,thelightcurvesuggested
usedthespectralredistributionmatrixv013availableinthe
a pattern of recurrent peaks (smoothed interpolated green
Swift calibration database. The spectral analysis was per-
line in Fig. 2), that we investigated performing a temporal formedusingxspecv.12.5,aftergroupingthespectrumwith
analysis on the data (see Sect.3). a minimum of 20 counts per channel to allow the use of χ2
J194221 was observed during three visits (Obs.ID
00032228001/02/03) with Swift/XRT in Photon Counting
mode at the end of December 2011, for a total collecting 3 http://www.swift.ac.uk/user_objects/
time of 6455 s. In Table 1, we show the log containing the 4 seee.g.http://www.swift.ac.uk/analysis/xrt/pileup.php
details about the three Swift/XRT pointings. 5 http://www.swift.ac.uk/analysis/xrt/pileup.php
(cid:13)c RAS,MNRAS000,1–5
Study on 1RXS J194211.9+255552 3
Obs.ID Tstart Tstop Exposure Rate
UTC UTC s countss−1
00032228001 2011-12-2106:09:04 2011-12-2107:37:02 1955 0.5261
00032228002 2011-12-2404:40:04 2011-12-2408:03:56 2063 0.4602
00032228003 2011-12-2600:03:19 2011-12-2603:25:57 2437 0.2641
Table 1.LogofthepointedSwift/XRTobservationsof1RXSJ194211.9+255552.
statistics. The source events arrival times were corrected to weevaluatedthecorrespondingsignificanceofthefeatureat
the SSB using the task barycorr6. P in units of Gaussian standard deviations (∼6.0σ). The
0
lightcurveprofile(Fig.4,panelc)foldedatP withT
0 epoch
= 55828.91015625 MJD, is characterized by a a peak of in-
tensity ∼4 times higher than the profile-averaged flux and
3 TIMING ANALYSIS
by a phase-span of ∼1/8 of the total period, corresponding
We performed a timing analysis of the Swift/BAT data to a duration of ∼20 days.
searching for long term periodic modulations in the 15–45 ThethreeSwift/XRTobservationsfallatphases0.524,
keVenergyrangeduringtheenhancedactivityperiodfrom 0.543 and 0.553 and, according to the folded profile, are in
2010.6 to 2014.5. We applied a folding algorithm to the phasewiththeemissionpeak.Wealsosearchedforthepres-
barycentred arrival times searching in the 1–1000days time enceofshorterperiodicsignalsintheSwift/XRTdata,per-
rangewithastepofP2/(N∆T),wherePisthetrialperiod, forming a timing analysis on the arrival time of the source
N = 16 is the number of profile phase bins and ∆T=124 events extracted from each of the 3 Swift/XRT observa-
Msisthedatatimespan.Theaveragerateineachphasebin tions. In order to avoid systematics caused by the read-out
was evaluated by weighting the rates by the inverse square time in PC mode (characterized by a time resolution bin of
of the corresponding statistical errors (see Cusumano et al. δT =2.5073 s), these arrival times were first randomized
xrt
2010). within δT . Then we performed a folding search in the
xrt
Figure 4 (a) shows the periodogram with several features range [δT : 500] s, but no significant feature above the
xrt
emerging. The most prominent one is at P0=166.5±0.5 d level of noise was detected.
(χ2=172),wheretheperiodanditserror∆P arethecen-
0
troidofthepeakandthestandarddeviationobtainedfrom
aGaussianfittotheχ2 featureatP ;othersignificantfea-
0 4 SPECTRAL ANALYSIS
tures at higher (lower) periods corresponding to multiples
(sub-multiples) of P are also clearly detected in the peri- We studied the high-energy BAT spectrum collecting the
0
odogram. events only after 2010.6 to optimize the SNR, but keeping
The long-term variability of the source causes the distribu- all the available band (15–150 keV energy range) to bet-
tionofχ2 todeviatesharplybothinaverageandinfluctua- ter constrain the cut-off energy of the spectrum. We pro-
tionamplitudefromthebehaviourexpectedforawhitenoise duced a phase-averaged and a phase-selected spectrum in
signaldominatedbystatisticalvariations.Asaconsequence the phase-interval 0.3–0.7 that corresponds to the peak of
the χ2 statistics cannot be applied, and for an estimation thefoldedprofileofFig.4.Bothspectrawereaccumulatedin
of the significance of this feature, we applied the following eight energy channels and analysed using the BAT redistri-
procedure (e.g. see also D’A`ı et al. 2011): butionmatrixavailableintheSwiftcalibrationdatabase7.
i)wefittedtheperiodogrambetween50and250days(char- Wefittedtogetherthetwospectrawithasimplepower-law,
acterized by a noise level consistent with the noise level at and we obtained a satisfactory fit (χ2/dof=15/14) with a
P ), excluding opportune intervals around the peaked fea- common photon-index (Γ=3.3±0.3) and free to vary nor-
0
tures, with a linear fit that describes well the trend of the malizations.Wethenadoptedabremsstrahlungmodelhav-
χ2 values. The best fit function was then subtracted from ing an exponential high-energy cut-off and we obtained a
the χ2 to obtain a new, flattened, periodogram (hereafter, χ2/dof=7/14 with a plasma temperature of 13+4 keV, in-
−3
z); the new z value at P is 148.2. dicativeofanexponentialdecayofthespectrumathighen-
0
ii) We then built the histogram of these z values, and fit- ergies. These results imply a factor ∼2 higher flux (20–100
ted the positive tail of the histogram distribution (beyond keVenergyband)atthepeakofthefoldedprofile(1.7±0.3
z=10) with an exponential function, and we evaluated the × 10−11 erg cm−2 s−1) with respect to the phase-averaged
integral, Σ, between 148.2 and infinity (normalized to the flux emission (7.1±1.4 × 10−12 erg cm−2 s−1), but unde-
total area below the distribution). The area below the his- tected variability in the spectral shape.
togramwasevaluatedsummingthecontributionofeachsin- WestudiedthepointedSwift/XRTspectrainthe0.5–
gle bin from its left boundary up to z=10 and integrating 10 keV range, while we used the 0.3–0.7 phase-selected
thebest-fitexponentialfunctionbeyondz=10.Σrepresents Swift/BAT spectrum to provide a coverage of the hard X-
theprobabilityofchanceoccurrenceofzgreaterthan148.2. ray emission above 15 keV. The BAT spectrum is averaged
iii) From the probability of chance occurrence (2.2×10−9), for many orbital periods and its use in a common fit with
the Swift/XRT spectra is helpful in giving an estimate of
6 http://http://heasarc.gsfc.nasa.gov/ftools/caldb/help/
barycorr.html 7 http://swift.gsfc.nasa.gov/docs/heasarc/caldb/swift/
(cid:13)c RAS,MNRAS000,1–5
4
200 a P0 Parameter Best-fittingvalue
2P0 Commonfittingvalues
150 3P0 Γ 0.4±0.2
Ecut (keV) 18+−410
χ2 100 P−40 P−30 P−20 4P0 Efold (keV) Obs.ID7.031+−32.7
50 NH (1022 cm−2) 0.8±0.5
Fa 10±1.3
0.5−10keV
Obs.ID02
0
1 10 100 1000
Period (day) NH (1022 cm−2) 1.7±0.6
Fa 9.0±0.6
b 0.5−10keV
0 Obs.ID03
0
1
ues NH (1022 cm−2) 3.7±1
χval2 c F0a.5−10keV 6.6±0.6
Frequency of 10 FCBaAT BAT(28.4.6±+−022.0.40)×10−2
15−100keV
χ2 /dof 93/86
1 a Inunitsof10−11 ergcm−2 s−1
−20 0 20 40 Table 2. Best-fitting spectral parameters for the combined
χ2
c Swift/XRT and Swif/BAT spectra adopting a model of an ab-
c sorbedpower-lawwithahigh-energycut-off.Weleftthecolumn
0−4 densityoftheSwift/XRTdata-setsfreetovary,whilethecolumn
1
densityoftheBATspectrumistiedtotheObs.ID01value.
el−1
x
nts s pi−1 5×10−5 tnoifitchaendtasttaee(preendiuncgedinχt2h=eS3w.0iffotr/B88AdTofe)n,ebrgeycaruasnegoef.tThheesrige--
ou sultingspectrumwasmuchbetterfitted(reducedχ2=1.15,
c
for 87 dof) with a cut-off power-law model (cutoffpl in
XSPEC) having a best-fitting photon index Γ=-0.4±0.3
0 and a cut-off energy E =6.8+1.6 keV. Adopting a model
cut −1.2
withapower-lawmultipliedbyahigh-energycut-offwithan
0 0.5 1 1.5 2 e-foldingenergy(Exp[(E −E)/E ]),highecutmodel),
Phase cut fold
we obtained a reduced χ2 of 1.08 for 86 dof, that pro-
Figure 4. a: Periodogram of Swift-BAT (15–45keV) data for
vides,however,onlyamarginalimprovement(F-testchance
J194221. b: Distribution of the z values derived from the χ2
improvement 4.4%) with respect to the cutoffpl model
periodogram. The positive tail beyond z=10 is modelled with
(Fig.5).InTable2,weshowthebest-fittingspectralparam-
an exponential function. c: Light curve folded at a period
P0=166.5days,with16phasebins. etersanderror-barscalculatedat∆χ2=2.7(90%confidence
level) for this final model.
5 DISCUSSION
the broadband spectral shape assuming negligible spectral
variability between the pointed and the long-term spectra. In this paper we presented the first study on the timing
ThethreeSwift/XRTspectrashowedslightlydifferent and spectral properties of the X-ray source J1944221 ex-
spectral shapes and fluxes. We noted that the spectra had ploiting the data recorded by the X-ray instruments on
thelargestdiscrepancyinthesoftestpartoftheX-rayspec- board of Swift. The source showed significant and steady
trum,andwetookitasapossibleindicationofavaryinglo- higher accretion rates since 2011, and it is clearly resolved
calcolumndensity(N ).Allowingfordifferentvaluesofthe in hard X-rays up to ∼45 keV. The timing analysis on the
H
neutral absorption for each observation and introducing a Swift/BAT survey revealed a periodic modulation with a
multiplicativeconstanttoaccountforthedifferentfluxes,we period of P =166.5±0.5 days, that we associate to the or-
0
obtainedasatisfactoryaccountofalltheSwift/XRTspec- bital period of an X-ray binary system. The profile of the
tra,consistentwithahardpower-lawofbest-fittingphoton- light curve folded at P shows a peak in emission lasting ∼
0
index Γ=0.4. When the BAT spectrum is added to the fit, 1/8oftheperiod.Thefoldedprofileisconsistentwithasce-
thesimplepower-lawmodeldidnolongerprovideagoodfit nario of enhanced accretion from a compact object close to
(cid:13)c RAS,MNRAS000,1–5
Study on 1RXS J194211.9+255552 5
REFERENCES
0.01
Chenevez J., Fiocchi M., Bazzano A., Tarana A., Sidoli
V)−1 L., Sguera Bird A. J., Drave S. Kuulkers E., 2011, The
e
m s k−2−1 10−3 CAussturmonaonmoeGr’.s, LTaelePgarraomla,V38.1,6R,o1mano P., Segreto A., Ver-
ns c cellone S., Chincarini G., 2010, MNRAS, 406, L16
o
hot Cusumano G., Segreto A., La Parola V., Masetti N., D’A`ı
P
V ( A., Tagliaferri G., 2013, MNRAS, 436, L74
e 10−4
k
D’A`ıA.,LaParolaV.,CusumanoG.,SegretoA.,Romano
P., Vercellone S., Robba N. R., 2011, A&A, 529, A30
Dickey J. M., Lockman F. J., 1990, ARAA, 28, 215
2
Evans P. A., Osborne J. P., Beardmore A. P., Page K. L.,
χ 0
WillingaleR.,MountfordC.J.,PaganiC.,BurrowsD.N.,
−2
Kennea J. A., Perri M., Tagliaferri G., Gehrels N., 2014,
10 100
Energy (keV) ApJS, 210, 8
Figure 5. Data, best-fitting unfolded model (upper panel) and Goad M. R., Tyler L. G., Beardmore A. P., Evans P. A.,
residualsinunitsofσ(lowerpanel)forthecombinedSwift/XRT Rosen S. R., Osborne J. P., Starling R. L. C., Marshall
and Swift/BAT spectra, adopting the model of Table 2. Black, F. E., Yershov V., Burrows D. N., Gehrels N., Roming
red, green and blue color for the Swift/XRT Obs.ID 01, 02, 03 P. W. A., Moretti A., Capalbi M., Hill J. E., Kennea J.,
andSwift/BATdata,respectively. Koch S., vanden Berk D., 2007, A&A, 476, 1401
La Parola V., Cusumano G., Segreto A., D’A`ı A., Masetti
N., D’Elia V., 2013, ApJL, 775, L24
La Parola V., Segreto A., Cusumano G., Masetti N., D’A`ı
A., Melandri A., 2014, MNRAS, 445, L119
Masetti N., Landi R., Parisi P., Bazzano A., Bird A. J.,
the periastron passage, as typical for the class of Be X-ray
2012, The Astronomer’s Telegram, 4209, 1
binarysystems.OurinterpretationagreeswithMasettietal.
Segreto A., Cusumano G., Ferrigno C., La Parola V.,
(2012), that on the base of the presence of strong H-alpha
ManganoV.,MineoT.,RomanoP.,2010,A&A,510,A47
line in the optical spectrum, first proposed a high-mass X-
Sidoli L., Fiocchi M., Bird A. J., Drave S. P., Bazzano A.,
ray identification for the J194221.
PersiP.,TaranaA.,SgueraV.,ChenevezJ.,KuulkersE.,
Weanalysedthebroad-bandX-rayspectrumofJ194221us-
2011, The Astronomer’s Telegram, 3818, 1
ingtheXRTpointedobservationdatainthesoftX-rayband
and the BAT survey data. The spectrum can be well mod-
elled with a hard (Γ=0.4) absorbed power-law with an en-
ergycut-offE at∼18keVandafoldingenergyE ∼7.3
cut fold
keV. We found evidence of a possible varying local neutral
absorber in the Swift/XRT spectra. Among the three ex-
aminedobservations,thelowestcolumndensityvalueiscon-
sistentwiththeGalacticvalueinthedirectionofthesource
(1.09×1022 cm−2 Dickey&Lockman1990),whilethehigh-
estvalueindicatemoderatelocalabsorption.ThehardX-ray
spectrum, the high-energy folding and cut-off energies, and
the variations in the equivalent hydrogen column observed
inclose(withinfewdays)observationsattheputativeperi-
astron passage are all typical spectral characteristics of the
BeX-raybinaryclassandfurthersupportouridentification
ofthesourceasanactiveBeX-raybinary(seealsoforsim-
ilar interpretations La Parola et al. 2013, 2014; Cusumano
et al. 2013).
ACKNOWLEDGEMENTS
Thisresearchhasmadeuseofdataand/orsoftwareprovided
by the High Energy Astrophysics Science Archive Research
Center (HEASARC), which is a service of the Astrophysics
ScienceDivi-sionatNASA/GSFCandtheHighEnergyAs-
trophysicsDivisionoftheSmithsonianAstrophysicalObser-
vatory.
This work has been supported by ASI grant I/011/07/0.
(cid:13)c RAS,MNRAS000,1–5