Pulsator-like spectra from Ultraluminous X-ray sources and the search for more ultraluminous pulsars

0.33 MB·

by  F. Pintore

DRAFTVERSIONJANUARY16,2017
PreprinttypesetusingLATEXstyleemulateapjv.5/2/11
PULSATOR-LIKESPECTRAFROMULTRALUMINOUSX-RAYSOURCESANDTHESEARCHFORMORE
ULTRALUMINOUSPULSARS
F.PINTORE1,L.ZAMPIERI2,L.STELLA3,A.WOLTER4,S.MEREGHETTI1,G.L.ISRAEL3
(Received;Revised;Accepted)
DraftversionJanuary16,2017
ABSTRACT
Ultraluminous X-ray sources (ULXs) are a population of extragalactic objects whose luminosity exceeds
the Eddington limit for a 10 M black hole (BH). Their properties have been widely interpreted in terms of
(cid:12)
7
accretingstellar-massorintermediate-massBHs.Howeveratleastthreeneutronstars(NSs)havebeenrecently
1
identified in ULXs through the discovery of periodic pulsations. Motivated by these findings we studied the
0
spectralproperties ofasample ofbright ULXsusinga simplecontinuummodel whichwas extensivelyused
2
tofittheX-rayspectraofaccretingmagneticNSsintheGalaxy. Wefoundthatsuchamodel,consistingofa

n power-lawwithahigh-energyexponentialcut-off,fitsverywellmostoftheULXspectraanalyzedhere,ata
a levelcomparabletothatofmodelsinvolvinganaccretingBH.Onthesegroundsalonewesuggestthatother
J non-pulsatingULXsmayhostNSs. Wefoundalsothatabove2keVthespectrumofknownpulsatingULXs

3 is harder than that of the majority of the other ULXs of the sample, with only IC 342 X-1 and Ho IX X-1
1 displaying spectra of comparable hardness. We thus suggest that these two ULXs may host an accreting NS
andencouragesearchesforperiodicpulsationsintheflux.

]
Subjectheadings:stars: neutron–(stars:) pulsars: general–X-rays: individual(NGC7793P13, NGC5907
E
X-1,IC342X-1,HoIXX-1,HoIIX-1,NGC5408X-1,NGC1313X-1,NGC1313X-2,
H
NGC5204X-1,NGC55ULX1,NGC5643ULX1,NGC6946X-1)
.
h
p
1. INTRODUCTION optically-thick corona close to the BH (e.g. Middleton et al.
-
o Ultraluminous X-ray sources (ULXs) are a class of extra- 2015). However,thequalityofULXspectraisgenerallylow
r galactic, point-likesourceswithluminosityfrom∼ 1039 erg andtheinterpretationoftheirspectralpropertiesisstillmatter
t
s s−1 up to 1042 erg s−1, in excess of the Eddington limit ofdebate.
a By exploiting the recent availability of broadband X-ray
for a 10M black hole (BH) (e.g. Fabbiano 2006). Such
[ (cid:12) data obtained with the XMM-Newton and NuSTAR satellites,
a high X-ray luminosity suggests that they host either inter-
we carried out a comprehensive spectral analysis of a large
1 mediate mass BHs accreting at sub-Eddington rate (Colbert
sample of the brightest ULXs. Several previous investiga-
v & Mushotzky 1999; Miller et al. 2004), or super-Eddington
tions focussed on the phenomenological characterization of
5 accreting BHs of stellar mass or of stellar origin (e.g. Glad-
9 stoneetal.2009;Feng&Soria2011). Inparticular,themost the spectra of ULXs (e.g Gonc¸alves & Soria 2006; Stobbart
5 luminous ULXs (> 5 × 1040 erg s−1) are usually consid- etal.2006;Gladstoneetal.2009;Pintore&Zampieri2012;
3 ered as the best intermediate mass BH candidates (e.g. Sut- Suttonetal.2013;Pintoreetal.2014). Motivatedbythedis-
0 tonetal.2012),whilelessluminoussourcescouldbesuper- covery of X-ray pulsations from M82 X-2 and subsequently
. fromNGC7793P13andNGC5907X-1,weadoptedasim-
1 EddingtonaccretingBHswithmassesintherange5–80M(cid:12) ple spectral model that was extensively used to fit the X-ray
0 (e.g.Zampieri&Roberts2009).Therecentdiscoveryofthree
continuum of accreting magnetic NSs, especially X-ray pul-
7 pulsatingneutronstars(NSs)intheULXM82X-2(Bachetti
satorsinGalactichighmassX-raybinaries(formoredetails
1 etal.2014),NGC7793P13andNGC5907X-1(Israeletal.
seeWhiteetal.1983,1995;Coburnetal.2001). Themodel
: 2016a,Israeletal.2016b)provedthatULXscanbepowered
v consists of a power-law with a high-energy exponential cut-
byobjectsdifferentfromBHs.
i off plus a blackbody; we find that in most cases the ULX
X TheULXsspectraaregenerallydescribedintermsofacold
spectraarewellfitbysuchamodel, indicatingthatdifferent
(kT ∼ 0.1−0.6 keV) soft thermal component coupled to a
r interpretationsoftheULXspectraarepossible.
a high-energycomponentwithacut-offbelow10keV(e.g.Sut-
Wedescribethedatareductioninsection2,wepresentthe
tonetal.2013;Bachettietal.2013;Pintoreetal.2014;Mid-
results of our spectral analysis in section 3 and we discuss
dletonetal.2015). Thesoftcomponentispossiblyassociated
theminsection4.
tothephotosphereofstrongoutflowsejectedfromtheaccre-
tiondiscatsuper-Eddingtonrates(e.g.Poutanenetal.2007;
2. OBSERVATIONSANDDATAANALYSIS
Ohsuga & Mineshige 2011), while the high-energy compo-
nentmightcomefromthebareinneraccretiondiscorfroman We analyzed a sample of 12 ULXs, including two ULX
pulsars,locatedatdistanceslowerthan15Mpc,whichshows
a broad luminosity range spanning from ∼ 1039 erg s−1 to
1INAF-IASFMilano,viaE.Bassini15,I-20133Milano,Italy ∼ 1041 erg s−1. The sources have available XMM-Newton
2NAF-OsservatorioAstronomicodiPadova, Vicolodell’Osservatorio
and, in some cases, simultaneous (or nearly simultaneous)
5,I-35122Padova,Italy
NuSTAR observations of good quality which allow us to put
3INAF-OsservatorioastronomicodiRoma,ViaFrascati44,I-00078,
MonteporzioCatone,Italy robustconstraintsonthehighenergycurvatureoftheirX-ray
4INAF,OsservatorioAstronomicodiBrera, viaBrera28, 20121Mi- spectrum(i.e. withatleast10000countsintheXMM-Newton
lano,Italy EPIC instruments, as shown in Gladstone et al. 2009). Our
2 Pintoreetal.
TABLE1
LOGOFTHEOBSERVATIONSOFTHEULXSANALYZEDINTHISWORKANDORDEREDWITHINCREASINGRA.
Source. Obs.ID Date Exp. Obs.ID Date Exp Epoch
XMM-Newton (ks) NuSTAR (ks)
0724810401 2013-11-12 33.5 30002039005 2013-11-12 113 1
NGC5907X-1 0729561301 2014-07-09 42.0 80001042002 2014-07-09 57.1 2
- - - 80001042004 2014-07-12 56.3 2
0693760401 2013-11-25 46.0 - - - 1
NGC7793P13
0748390901 2014-12-10 47.0 - - - 2
0028740101 2001-11-15 28.3 - - - 1
NGC55ULX-1
0655050101 2010-05-24 124.0 - - - 2
0150280301 2003-12-21 10.3 - - - 1
NGC1313X-1
0693851201 2012-12-22 125.2 30002035004 2012-12-16 127.0 2
0405090101 2012-12-16 123.3 - - - 1
NGC1313X-2
0693851201 2012-12-22 125.2 30002035004 2012-12-16 127.0 2
0693850601 2012-08-11 59.9 30002032003 2012-08-10 98.6 1
IC342X-1
0693851301 2012-08-17 60.0 30002032005 2012-08-16 127.4 2
0200470101 2004-04-15 93.5 - - - 1
HoIIX-1 0724810101 2013-09-09 12.3 30001031002 2013-09-09 31.4 2
- - - 30001031003 2013-09-09 79.4 2
0724810301 2013-09-17 12.0 30001031005 2013-09-17 111.1 2
0200980101 2004-09-26 117 - - - 1
- - - 30002033005 2012-11-11 40.7 2
HoIXX-1 0693851701 2012-11-12 9.92 30002033006 2012-11-11 35.2 2
0693851801 2012-11-14 13.8 30002033008 2012-11-14 14.5 2
0693851101 2012-11-16 13.3 30002034010 2012-11-15 49.0 2
0405690201 2006-11-19 43.5 - - - 1
NGC5204X-1
0693851401 2013-04-21 16.9 30002037002 2013-04-19 96.0 2
NGC5408X-1 0653380501 2011-01-28 124.1 - - - 1
NGC5643ULX-1 0744050101 2014-08-27 114.0 - - - 1
NGC6946X-1 0691570101 2012-10-21 114.3 - - - 1
sampleissimilartothoseanalyzedinseveralpreviousstudies ting was carried out using XSPEC v.12.8.2 (Arnaud 1996).
(e.g.Stobbartetal.2006;Gladstoneetal.2009;Pintoreetal. ThespectraofthepnandthetwoMOScameras, and(when
2014;Middletonetal.2015)andoffersanoverallpictureof available)oftheNuSTARdetectorswerefittogether,foreach
the spectral properties displayed by ULXs. In order to ex- epoch. A multiplicative factor was included to account for
tract individual spectra with better statistics, observations of possible inter-calibration uncertainties, which were in most
thesamesourceobtainedwithintimeintervalsofafewdays cases smaller than 10%. We considered the 0.3–10 keV and
weremergedifnosignificantspectralvariationswerepresent. 3–70 keV energy range for EPIC and NuSTARdata, respec-
Data taken at very different times and/or showing different tively.
spectralstateswereinsteadconsideredasindividualobserva-
tions. Additionallytothisselection,wealsoconsidered,for 3. RESULTS
eachsource,theobservationswiththelargestsourcespectral 3.1. Broadbandspectralanalysis
variability. WelistinTable1theobservationsthatareusedin
We fit the spectra with an absorbed power-law with an
thiswork.
exponential cut-off characterized by two parameters, the
The data of the XMM-Newton EPIC instrument (Stru¨der
cut-off energy E and the folding energy E : F(E) =
c f
et al. 2001; Turner et al. 2001) were reduced using SAS
v.14.0.0.WeextractedspectrafromeventswithPATTERN≤4 k · E−Γe−(E−Ec)/Ef (HIGHECUT×POWERLAW in XSPEC),
withtheexponentialpartactingforE > E . Inmostcases,
for EPIC-pn (single- and double-pixel events) and PATTERN c
an additional soft component, that we model with a black-
≤ 12forEPIC-MOS(single-andmultiple-pixelevents). We
body for simplicity, is needed. This model was extensively
set‘FLAG=0’inordertoexcludebadpixelsandeventscom-
applied in the past to fit the X-ray continuum of accreting
ingfromtheCCDedgesandremovedtimeintervalswithhigh
magnetic NSs, notably X-ray pulsators in high mass X-ray
background. Thespectrawereextractedfromcircularregions
binaries. In those systems the hard component is believed
with radius of 30(cid:48)(cid:48) and 65(cid:48)(cid:48) for source and background, re-
to originate in a magnetic accretion-column close to the NS
spectively.
surface, whereas the soft blackbody-like component is usu-
TheNuSTARdatawerereducedwiththestandardpipeline,
ally ascribed to emission from the inner edge of the disk at
basedontheNuSTARDataAnalysisSoftwarev1.3.0(NUS-
themagnetosphericboundary(e.g.Whiteetal.1995;Coburn
TARDAS) in the HEASOFT FTOOLS v6.16. We obtained
etal.2001).
cleaned event files by applying the standard corrections.
NGC5907X-1wasobservedsixtimeswithXMM-Newton
Spectrawereextractedfromcircularregionsofradius40(cid:48)(cid:48)and
(e.g. Israeletal.2016b,Fuerstetal.2016)andgoodquality
60(cid:48)(cid:48)forsourceandbackground,respectively.
spectra are available for the observations of February 2003,
AllXMM-NewtonandNuSTARspectrawererebinnedsoas
February 2012, November 2013 and July 2014. Simultane-
to have at least 25 and 100 counts per bin, respectively, so
ous and quasi-simultaneous NuSTAR data are available for
thatminimumχ2 fittingtechniquescouldbeused;modelfit-
the 2013 and 2014 observations. The fit of the 2003, 2012
Thesearchfornewultraluminouspulsars 3
for spectral variability, as already reported in the literature
(e.g. Sutton et al. 2013; Pintore et al. 2014; Middleton et al.
NGC 7793 P13 2015). To illustrate the range of possible states in which the
sourcescanbefound,weplotinFig.1thetworepresentative
10-2 spectra of each ULX, choosing for each source the observa-
NGC 5907 X-1
tions with the hardest and softest spectrum (see section 2).
These are the observations reported in Table 1, with corre-
spondingbestfitparameterslistedinTable2.
Generally,the“pulsator-like”modelgivesgoodfits,except
IC 342 X-1
-4 for a few cases in which the soft thermal component is not
10
strictly required (i.e. IC 342 X-1 in August 2012, and NGC
Ho IX X-1
1313 X-2 in December 2012). In all other cases, blackbody
NGC 5643 ULX-1 components with temperatures in the range ∼0.15–0.3 keV
] arerequired. Avarietyofpower-lawspectralslopesandcut-
y units 10-6 NGC 1313 XN-2GC 1313 X-1 offTphaerraemlaettievreslyislaforguenχd2(Tvaablulees2o).btainedinthefitofHoIXX-
r
ra 1arecausedbysomeexcessemissionathighenergy,possibly
bit indicating the presence of a further hard tail component, as
r
[a NGC 5204 X-1 discussedinWaltonetal.(2014a). Wenotethathardtailsare
)
E observed also in the spectra of some Galactic accreting NSs
2f( 10-8 (e.g.McClintock&Remillard2006).
E
Finally,thespectraofNGC55ULX1,NGC5408X-1,and
Ho II X-1 NGC6946showsomeresidualsaround1keV,asalreadyre-
NGC 6946 X-1
ported in literature (e.g. Middleton et al. 2014). These fea-
tureshavebeenassociatedlatelytoablendingofunresolved
10-10 NGC 5408 X-1 emissionandabsorptionlinesproducedbythephotosphereof
a possible outflowing wind in a scenario of super-Eddington
accretiononastellarmasscompactobject(Pintoetal.2016).
NGC 55 ULX-1
For the three sources, we find that good fits are obtained
with the addition of one or two broad Gaussian absorption
-12 lines centered at ∼0.7 keV and ∼1.2 keV and with widths
10
σ ∼0.07−0.13keV.
3.2. Characterizationofthehardcomponent
1 10 Wefoundthatthe“pulsator-like”modelcandescribewell
Energy (keV) the spectra of virtually all the ULXs of our sample. Differ-
ent broad band spectral shapes and best fit parameters char-
FIG.1.—Unfolded(E2F(E))spectraofalltheULXsofoursample. For
acterize different sources (see Fig. 1 and Table 2). This is
eachsource,weshowthetwomostdifferentspectralstatesobtainedbyfitting
themwiththemodelforaccretingNSsdescribedinthetext.Fordisplaypur- particularly evident at low energy, where, besides the effect
poses,thespectrahavebeenrebinnedonlyandthefluxisarbitrarilyshifted of different absorption column densities (from ∼ 5 × 1020
ontheyaxis. cm−2 to8×1022 cm−2), therelativeimportanceofthesoft
component can differ significantly (even if similar tempera-
and 2014 spectra give results consistent within the errors turesaroundkT∼0.2keVarefound). Spectraldifferencesare
with absence of spectral variability, as also found by Fuerst apparentalsoathigherenergies,wherethespectrumisdomi-
et al. (2016). We therefore report here only the results from natedbythecut-offpower-lawcomponent. Inordertostudy
the 2014 observations which are of higher quality and can ingreaterdetaildifferencesintheharderspectralcomponent
be complemented with NuSTAR data. They can be fit by (where the effects due to absorption and the additional soft
a HIGHECUT×POWERLAW alone (no soft component) with componentarenegligible),searchforcorrelationswithother
photon index Γ∼1.5, E = 5.5 keV and E = 8.3 keV. The sourceproperties, andpossiblygatherinformationonthein-
c f
2013spectrum, obtainedwhenthesourcewasaboutafactor nermostregionsoftheaccretionflowclosetothecompactob-
of2-3fainterthanin2014, isslightlysofter(Γ ∼1.9), butit ject,wecharacterizethehigh-energyspectrathroughacolor-
canstillbefitbyacut-offpower-lawwithouttheadditionof colordiagram. Thiswasdonebyusingthefluxesinthe2–4
asoftcomponent. keV, 4–6 keV and 6–30 keV energy bands derived from the
ForNGC7793P13,weusetwoobservationstakenin2013 best-fitofTable2. Wepresentthecolor-colordiagraminfig-
and2014(noNuSTARdataareavailable). Bothobservations ure2,wherewedefinedthehardnessasthe(6–30keV)/(4–6
givesimilarresults,requiringathermalcomponentwithtem- keV)ratioandthesoftnessas(2–4keV)/(4–6keV)ratio. We
peraturekT∼0.2keVinadditiontoacut-offpower-lawwith notethatthefluxesabove10keVareestimateddirectlyfrom
Γ∼1.1. AnotherXMM-Newtonobservation,obtainedinMay thespectraonlyinthecaseswhereNuSTARobservationsare
2012whenthesourcewasmuchfainter,providesalesscon- available. Intheothercases,thefluxesarecalculatedfroman
strainingspectrum,butconsistentinspectralshapewiththose extrapolationofthe0.3–10keVbest-fitmodel. Asexpected,
of2013and2014. sources displaying spectral variations move across different
For all other ULXs in our sample, several XMM-Newton regions in this diagram (note again that we pre-selected the
and NuSTAR observations are available. We analyze all the mostdifferentspectra ofeachsource)Severalof thesources
availablegoodqualityspectra,findinginmostcasesevidence with parameters around hardness∼1.5 and softness∼1.5 be-
4 Pintoreetal.
TABLE2
SPECTRALRESULTSOBTAINEDBYADOPTINGTHECONTINUUMMODELTBABS×(BBODY+HIGHECUT×POWERLAW).
Src. Epoch TBABS BBODY HIGHECUT POWERLAW GABS
NH kTin Norm Ec Ef Γ Norm Energy σ Strenght LXa χ2/dof
1022cm−2 keV 10−6 keV keV 10−3 keV keV 1040ergs−1
NGC5907X-1 1 0.76+−00..0076 - - 6.0+−01..95 8.0+−21..58 1.85+−00..0089 0.19+−00..0022 - - - 4.2 149.44/173
2 0.77+−00..0044 - - 5.5+−00..55 8.3+−00..87 1.48+−00..0045 0.34+−00..0022 - - - 10.9 442.90/442
NGC7793P13 1 0.07+−00..0022 0.22+−00..0022 1.5+−00..32 5.3+−01..42 4.3+−10..78 1.06+−00..0089 0.093+−00..00019 - - - 0.21 357.49/325
2 0.10+−00..0022 0.19+−00..0043 1.1+−00..54 6.6+−01..32 4.5+−31..41 1.11+−00..0047 0.24+−00..0011 - - - 0.53 629.78/563
NGC55ULX-1 1 0.291+−00..005414 0.173+−00..00222 29.3+−55..82 1.424+−00..1029 1.48+−00..4342 0.95+−00..5614 8.1+−32.8 - - - 0.26 700.77/688
2 0.41+−00..0065 0.15+−00..0011 4.0+−00..67 2.00+−00..109 1.5+−00..43 2.3+−00..45 5.9+−11..98 0.75+−00..0011 0.07+−00..0021 0.038+−00..00109 0.23 270.40/243
NGC1313X-1 1 0.3+−00..11 0.24+−00..0068 19+−1124 2.9+−02..79 4.6+−72..25 1.7+−01..52 1.1+−00..88 - - - 1.67 250.54/229
2 0.283+−00..00108 0.205+−00..000088 9.4+−00..66 6.1+−00..59 10.4+−11..41 1.74+−00..0036 5.8+−00..34 - - - 1.07 1703.40/1718
NGC1313X-2 1 0.32+−00..0011 - - 3.4+−00..43 7.7+−10.9 1.60+−00..0044 0.68+−00..0022 - - - 1.05 820.64/860
2 0.19+−00..0044 0.23+−00..0034 1.7+−11..11 1.02+−00..0066 1.8+−00..43 0.9+−00..44 0.28+−00..0046 - - - 0.28 1020.57/987
IC342X-1 1 1.045+−00..008775 0.219+−00..002284 8.7+−42..15 6.66+−00..4556 10.6+−11..42 1.62+−00..005665 0.51+−00..0045 - - - 0.43 937.17/981
2 0.99+−00..002266 - - 6.38+−00..567 15+−21..49 1.915+−00..00331 0.89+−00..0033 - - - 0.51 1061.76/1063
HoIIX-1 1 0.12+−00..0011 0.211+−00..000067 2.0+−00..23 2.0+−00..23 8.7+−21..46 2.1+−00..11 1.6+−00..11 - - - 1.2 1317.83/1317
2 0.10+−00..0022 0.17+−00..0011 13.1+−22..20 5.3+−00..35 8.6+−00..76 1.93+−00..0068 1.00+−00..0078 - - - 0.88 800.29/785
HoIXX-1 1 0.160+−00..000088 0.209+−00..000088 8.7+−00..55 6.0+−00..45 10.6+−22..61 1.38+−00..0033 0.70+−00..0022 - - - 1.4 2191.03/2043
2 0.239+−00..000088 0.21+−00..0065 2.1+−00..99 3.6+−00..22 5.7+−00..22 1.43+−00..0044 2.8+−00..11 - - - 3.8 1546.80/1433
NGC5204X-1 1 0.08+−00..0023 0.240+−00..00029 6.9+−11..32 2.8+−02..58 12.4+−152.4 2.0+−00..13 0.45+−00..0038 - - - 0.81 837.88/821
2 0.05+−00..0033 0.18+−00..0022 3.6+−10..08 4.0+−01..71 7.4+−11..57 1.7+−00..13 0.23+−00..0035 - - - 0.50 414.06/414
NGC5408X-1 1 0.09+0.01 0.162+0.005 14.8+1.5 2.0+0.3 5.7+2.3 2.0+0.2 0.32+0.04 0.75+−00..0012 0.07+−00..0033 0.03+−00..0021 0.87 1035.47/990
−0.01 −0.005 −1.6 −0.2 −1.4 −0.2 −0.05 1.17+0.03 0.08+0.04 0.023+0.02
−0.03 −0.03 −0.009
NGC5643ULX-1 1 0.20+−00..0011 - - 3.9+−00..33 6.5+−10..19 1.46+−00..0044 0.160+−00..000055 - - - 2.98 865.07/934
NGC6946X-1 1 0.40+−00..0076 0.13+−00..0012 25+−188 1.8+−00..34 7.0+−42..73 1.9+−00..23 0.25+−00..0054 0.69+−00..0034 0.13+−00..0043 0.15+−00..1035 0.9 613.68/637
aUnabsorbedluminosityinthe0.3–20keVenergyband;theadopteddistancesare:17.14Mpc,3.58Mpc,2.73Mpc,3.77Mpc,3.27Mpc,5.32Mpc,4.25Mpc,1.99Mpcand4.76Mpc(Tullyetal.2013)for
NGC5907,NGC7793,IC342,HoIX,HoII,NGC5408,NGC1313,NGC55andNGC5204,respectively;5.5Mpc(Smithetal.2007)forNGC6946and13.9Mpc(Sandersetal.2003)forNGC5643.
comeeithersofter(i.e. movetowardtheupper-leftcornerof pulsarsareIC342X-1andHoIXX-1.
the figure) or harder (i.e. move toward the lower-right cor-
4. DISCUSSION
ner),butnoneofthemshowsbothbehaviors. Thisleadstoa
distinctionbetweentwodifferentclassesofsofterandharder Inpreviousstudies,ULXspectracharacterizedbyacurva-
ULXs. Asimilarconclusionwasreachedine.g. Suttonetal. tureabove3–5keVandasoftexcessbelow∼1keVwerede-
(2013)andPintoreetal.(2014)basedonadifferentspectral scribed in terms of physically-motivated spectral models for
analysis. accreting BHs of different masses (e.g. Stobbart et al. 2006;
WehighlightthatthetwoknownULXpulsarsinthesam- Gladstoneetal.2009;Pintore&Zampieri2012;Suttonetal.
ple lie in the region of the color-color diagram that corre- 2013;Bachettietal.2013;Waltonetal.2013).Inspiredbythe
spondstothehardestspectra. Theirhardnessvaluesareclose recent discovery of pulsations in three ULXs (Bachetti et al.
tothemaximumbutstillwithinthebroaddistributionofthe 2014;Israeletal.2016a,b),weappliedhereamodelthathas
ULXs in our sample. We applied a Kolmogorov-Smirnov beencommonlyusedtofittheX-raycontinuumofaccreting
test (KS) to the distribution of the softness of pulsating and magneticNSs(e.g.Whiteetal.1995)togoodqualityXMM-
non-pulsating ULXs finding a KS-statistics of ∼ 88% and a Newton and NuSTAR spectra of a sample of bright ULXs.
p-value of ∼ 0.013 that they are drawn from the same sam- Suchmodel, HIGHECUT×POWERLAWplusasoftblackbody,
ple. Althoughwiththecurrentlimitedsampleitisdifficultto was introduced empirically (see White et al. 1995, and ref-
obtain firm conclusions, it is interesting to note that also the erences therein) and later found to be similar to the spectra
ULXpulsarM82X-2hasaratherhardspectrum5(Brightman calculatedfrommodelsofbulkandthermalComptonization
etal.2016a,b). in magnetic accretion models (Becker & Wolff 2007) which
We note that NGC 7793 P13 attained the lowest values of havealsobeenemployedtofitthespectraofafewX-raypul-
the softness ratio of the whole sample. During the epoch sars in HMXBs (Farinelli et al. 2016). The shape of its ex-
1 observation, when no pulsations were found in the timing ponentialcutoffisregulatedbytwoparameters(thee-folding
data (Israel et al. 2016b, a), the spectrum of NGC 5907 X-1 energyEf andthecutoffenergyEc),throughwhichthespec-
wasclosetothecenterofthedistributioninFig.2. In2014, trawithvastlydifferentcurvatureobservedinaccretingX-ray
whenpulsationsweredetected,thepositionofNGC5907X- pulsars at high energies (usually > 7−10 keV) can be fit-
1inthecolor-colordiagramwasclosertothatofNGC7793 ted.Inthisworkwefoundthatthis“pulsator-like”continuum
P13. Hence,whenpulsationswereseen,thetwoULXpulsars model, together with a soft component, can be successfully
weresignificantlyhard. Theonlysourcesthat,atleastonone usedtodescribealsotheoverallshapeofthespectraofbright
epoch, had spectral parameters similar to those of the ULX ULXs6. This suggests also that accreting BH-based models
usually adopted for ULXs (e.g. disc blackbody plus Comp-
5Unfortunately,goodqualityXMM-Newtonspectraldataarenotavailable
fortheULXpulsarM82X-2,owingtocontaminationfromthebrightnearby 6AswithothercontinuummodelsforULXs,inafewcasestheadditionof
ULXM82X-1. broadabsorption-likefeatures(e.g.Pintoetal.2016)and/orahighenergytail
Thesearchfornewultraluminouspulsars 5
10
NGC 7793 P13
NGC 5907 X-1
IC 342 X-1
Ho IX X-1
NGC 5643 ULX-1
NGC 1313 X-1
NGC 1313 X-2
NGC 5204 X-1
Ho II X-1
NGC 6946 X-1
NGC 5408 X-1
NGC 55 ULX-1
s
s
e
n
t
f
o
S
1
0.5 1 1.5 2 2.5 3
Hardness
FIG.2.—Color-colordiagramobtainedfromtheratiosofthefluxestimatedintheenergyranges2–4keV,4–6keVand6–30keVandestimatedfromthe
best-fitsoftable2.
tonization) do not necessarily have the interpretative power whether BH-models or the pulsator-like model we adopted
thathasbeenattributedtothem. provide a statistically better fit to the X-ray continuum of
Inourspectralfits,thesoftblackbodycomponenthasatem- bright ULXs. The two types of models resemble each other
peratureof∼0.1-0.3keVinallsources,butitsrelativeimpor- inthattheybothcompriseasoftcomponent(ablackbodyor
tancediffersacrossdifferentspectraandsources. Theequiva- adisk-blackbody,whichisverydifficulttodistinguishatthe
lentradiusofthethermallyemittingregion,inferredfromthe softX-rayenergieswhereitisseen)andahardercomponent
normalizationoftheblackbodycomponent,istypicallylarger comprisingapowerlaw-likespectrumwithexponentialcutoff,
than100km.Thisvaluerulesoutanyassociationwiththepu- whichlikelyoriginatesfromComptonizationinanaccretion
tative NS surface, but may be consistent with the size of the diskcoronainthecaseofBHsandfromthepost-shockregion
emittingregionofthediscclosetotheco-rotationradius,asin of a magnetically-funneled accretion column in the case of
thecaseofsomeX-raypulsatorsandULXpulsars(Israeletal. NSs. Theshapeofthecut-offcanbesmootherorsteeperthan
2016a,b). Alternatively, since pulsating ULXs have super- a simple exponential in the pulsator-like model. In order to
Eddingtonaccretionrates, thesoftcomponentmayoriginate ascertainwhatobservationswouldberequiredtodistinguish
in a powerful outflow that is ejected from the accretion disc between BH and NS-like spectral models we simulated the
(e.g. Poutanen et al. 2007; Ohsuga & Mineshige 2011; Mid- spectraofIC342X-1andNGC1313X-1byusingthebest-fit
dletonetal.2015). DISKBB+COMPTTmodel(seee.g. Middletonetal.2015)and
The best fit values derived with the HIGHECUT a1MsexposuretimewithXMM-NewtonandNuSTAR.Byfit-
×POWERLAW for the hard component span a fairly large tingthesespectrawithaHIGHECUT×POWERLAWmodel,we
range,bothinspectralslopeandcut-offandfoldingenergies, foundstructuredspectralresidualsathighenergy(>10keV),
as observed also in galactic accreting X-ray pulsars. This notobservedwithaDISKBB+COMPTTmodel. Thesamesim-
component is believed to arise in the post-shock region of ulationwascarriedoutbasedonthespecificationsofthefu-
the accretion column above the surface of the neutron star. ture X-ray mission Athena+7 where a discrimination of the
Recentanalyticalcalculations,aswellas2Dradiation-hydro- twomodelsrequiredanexposuretimeofatleast∼100ksin
simulationsofaccretionontomagnetizedNSs, showthatfor theWFIinstrument. However, wehighlightthatthesimple
magneticfieldsof∼ 1014 Gorlarger,veryhighluminosities DISKBB+COMPTTmodel,whenusingNuSTARdata,hasbeen
(up to 2-3 orders of magnitude above the Eddington limit) challengedinthesourcesHoIXX-1(Waltonetal.2014b),al-
can be produced (Mushtukov et al. 2015; Kawashima et al. readyindicatingthatULXspectradonotcompletelysuitwith
2016). aBHdescriptiononly.
Basedoncurrentlyavailabledata,itisnotpossibletoassess The relatively large number of parameters in the BB +
HIGHECUT×POWERLAW model hampers a straightforward
(e.g.Waltonetal.2014b;Mukherjeeetal.2015)provednecessaryinorder
tomakethefitacceptable 7http://www.the-athena-x-ray-observatory.eu/
6 Pintoreetal.
comparisonbetweenthesources,butaclearpictureemerges bealsonotedthatX-rayluminosityswingoftheULXpulsars
from a color-color diagram in which only the range above 2 is much larger ((cid:38) a factor of 100) than that of IC 342 X-1
keVisconsidered,whereabsorptioneffectsarenegligibleand and Ho IX X-1 (a factor of ∼ 5 only). Following Israel et
thereisno(ormarginal)contaminationfromthesoftcompo- al. (2016a,b)andassumingthatthesourcesareallviewedat
nent. This clearly shows that the ULX pulsars, when pulsa- comparableinclinationanglesandhavecomparablebeaming,
tionsaredetected,haveaharderspectrumthanthatofthema- weexpectthatthehighertheluminositythehigherthemag-
jorityoftheotherULXsofthesample(notethatthespectrum neticfield.Israeletal.(2016a,b)estimatedadipolarmagnetic
ofNGC5907X-1wassomewhatsofterin2013,whennopul- fieldof∼ 3×1013 GforNGC5907X-1and∼ 5×1012 G
sationsfromthissourceweredetected). Therearenoinstru- for NGC 7793 P13. Since IC 342 X-1 and Ho IX X-1 have
mentaleffectsthatfavorthedetectionofpulsations,ifpresent, luminositiesof∼1040ergs−1,wecanroughlyguessadipo-
in harder sources, but we cannot exclude that for some pro- lar magnetic field of ∼1013 G for both sources if their spin
cesses intrinsic to the sources the flux modulation increases periodwillturnouttobeintherangeofsecondsaswell.
when their spectra become harder. It is possible to specu-
late that the harder sources are those observed more directly
intheinnerregionsclosetothecompactobjects(orpossibly 5. CONCLUSION
fromanaccretioncolumn, Fuerstetal.2016)andthismight We adopted a commonly used spectral model for accret-
facilitate the detection of pulsations in the case of accreting ingX-raypulsatingNSintheGalaxytodescribethespectra
NS; while for softer sources the inner regions might be pos- of a sample of bright ULXs. We found that this model well
siblyoccluded(becauseofhigheranglesofthelineofsight) describesthespectraofmostULXs, suggestingthatspectral
by cold, optically thick plasmas which reduce the pulsation models based on accreting BHs may not have the interpre-
coherence. tative power that has been ascribed to them up to now. We
None of the other sources showed high-energy spectra as conclude that other ULXs may be found to host NSs accret-
hardasthehardestspectraoftheULXpulsars,exceptforIC ingatsuper-Eddingtonrates. Weshowedalsothattheknown
342X-1andHoIXX-1.Thespectralparametersofthesetwo pulsatingULXsareamongstthehardestsourcesofoursam-
sourcesoverlap,atleastononeepoch,withthoseoftheULX ple,withIC342X-1andHoIXX-1havingcomparablyhard
pulsars (see Fig. 2). Hence we suggest that IC 342 X-1 and spectra.Basedontheirspectralpropertiesalone,wethussug-
Ho IX X-1 may be good candidates NS ULXs. On the ba- gestthesetwosourcesarelikelycandidateNSULXs.
sisoftheirspectrumalone,NGC5408X-1,HoIIX-1,NGC
1313X-1andX-2,NGC5643ULX1andNGC6946X-1are
instead less likely to contain NSs. The failure to detect pul- ACKNOWLEDGEMENTS
sations so far in IC 342 X-1 and Ho IX X-1 may be caused We thank the anonymous referee for his/her helpful com-
byintermittentactivityofperiodicfluxmodulationand/orby ments. We acknowledge financial contribution from the
the presence of a large period derivative. However it should agreementASI-INAFI/037/12/0andPRININAF2014.
REFERENCES
Arnaud,K.A.1996,inASPConf.Ser.,SanFranciscoCA,Vol.101,Jacoby, Miller,J.M.,Fabian,A.C.,&Miller,M.C.2004,ApJ,614,L117
G.H.andBarnes,J.,eds.,AstronomicalDataAnalysisSoftwareand Mukherjee,E.S.,Walton,D.J.,Bachetti,M.,&Harrison,F.A.e.a.2015,
SystemsV.,17 ApJ,808,64
Bachetti,M.,Harrison,F.A.,Walton,D.J.,etal.2014,Nature,514,202 Mushtukov,A.A.,Suleimanov,V.F.,Tsygankov,S.S.,&Poutanen,J.
Bachetti,M.,Rana,V.,Walton,D.J.,etal.2013,ApJ,778,163 2015,MNRAS,454,2539
Becker,P.A.,&Wolff,M.T.2007,ApJ,654,435 Ohsuga,K.,&Mineshige,S.2011,ApJ,736,2
Brightman,M.,Harrison,F.,Walton,D.J.,etal.2016a,ApJ,816,60 Pinto,C.,Middleton,M.J.,&Fabian,A.C.2016,Nature,533,64
Brightman,M.,Harrison,F.A.,Barret,D.,etal.2016b,ApJ,829,28 Pintore,F.,&Zampieri,L.2012,MNRAS,420,1107
Coburn,W.,Heindl,W.A.,Gruber,D.E.,etal.2001,ApJ,552,738 Pintore,F.,Zampieri,L.,Wolter,A.,&Belloni,T.2014,MNRAS,439,3461
Colbert,E.J.M.,&Mushotzky,R.F.1999,ApJ,519,89 Poutanen,J.,Lipunova,G.,Fabrika,S.,Butkevich,A.G.,&Abolmasov,P.
Fabbiano,G.2006,ARA&A,44,323 2007,MNRAS,377,1187
Farinelli,R.,Ferrigno,C.,Bozzo,E.,&Becker,P.A.2016,A&A,591,A29 Sanders,D.B.,Mazzarella,J.M.,Kim,D.-C.,Surace,J.A.,&Soifer,B.T.
Feng,H.,&Soria,R.2011,NewAstronomyReviews,55,166 2003,AJ,126,1607
Fuerst,F.,Walton,D.J.,Stern,D.,etal.2016,ArXive-prints, Smith,B.J.,Struck,C.,Hancock,M.,etal.2007,AJ,133,791
arXiv:1610.00258 Stobbart,A.-M.,Roberts,T.P.,&Wilms,J.2006,MNRAS,368,397
Gladstone,J.C.,Roberts,T.P.,&Done,C.2009,MNRAS,397,1836 Stru¨der,L.,etal.2001,A&A,365,L18
Gonc¸alves,A.C.,&Soria,R.2006,MNRAS,371,673 Sutton,A.D.,Roberts,T.P.,&Middleton,M.J.2013,MNRAS,435,1758
Israel,G.L.,Papitto,A.,Esposito,P.,Stella,L.,&Zampieri,e.a.2016a, Sutton,A.D.,Roberts,T.P.,Walton,D.J.,Gladstone,J.C.,&Scott,A.E.
ArXive-prints,arXiv:1609.06538 2012,MNRAS,423,1154
Israel,G.L.,Belfiore,A.,Stella,L.,etal.2016b,ArXive-prints, Tully,R.B.,Courtois,H.M.,Dolphin,A.E.,Fisher,J.R.,&etal.2013,AJ,
arXiv:1609.07375 146,86
Kawashima,T.,Mineshige,S.,Ohsuga,K.,&Ogawa,T.2016,PASJ,68,83 Turner,M.J.L.,etal.2001,A&A,365,L27
McClintock,J.E.,&Remillard,R.A.2006,inCompactstellarX-ray Walton,D.J.,Harrison,F.A.,Grefenstette,B.W.,Miller,J.M.,&etal.
sources,ed.W.H.G.LevinandM.vanderKlis(Cambridge:Cambridge 2014a,ArXive-prints,arXiv:1402.2992
UniversityPress),157 —.2014b,ApJ,793,21
Middleton,M.J.,Heil,L.,Pintore,F.,Walton,D.J.,&Roberts,T.P.2015, Walton,D.J.,Miller,J.M.,Harrison,F.A.,etal.2013,ApJ,773,L9
MNRAS,447,3243 White,N.E.,Nagase,F.,&Parmar,A.N.1995,X-rayBinaries,1
Middleton,M.J.,Walton,D.J.,Roberts,T.P.,&Heil,L.2014,MNRAS, White,N.E.,Swank,J.H.,&Holt,S.S.1983,ApJ,270,711
438,L51 Zampieri,L.,&Roberts,T.P.2009,MNRAS,400,677