EXOSAT OBSERVATIONS OF THE ULTRA-SOFT X-RAY BINARY 4U1957+11

8 Pages·0.8 MB·English

by  D. Ricci

A&A manuscript no.
ASTRONOMY
(will be inserted by hand later)
AND
Your thesaurus codes are: ASTROPHYSICS
0.6(24.03.1; 24.01.1; 02.02.1; 19.50.1) 27.1.1995
EXOSAT observations of the ultra{soft X{ray binary
4U1957+11
1;4 2;4 3;4
D. Ricci , G.L. Israel , and L. Stella
1
Dipartimento di Fisica, Universit(cid:18)a di Roma \La Sapienza", P.le Aldo Moro 5,I{00185 Roma, Italy, [email protected]
2
International School for Advanced Studies (SISSA{ISAS), Via Beirut 2{4, I{34014 Trieste, Italy, [email protected]
3
Osservatorio Astronomico di Brera, Via E.Bianchi 46,I{22055 Merate, Italy, [email protected]
4
A(cid:14)liated to I.C.R.A.
Received date; accepted date
Abstract. We present results from the analysis of the two 1987).These(cid:12)ndings clearly indicate that4U1957+11isalow
EXOSATobservationsofthelowmassX-raybinary(LMXRB) mass X-ray binary (LMXRB) with an X-ray luminosity of a
36 (cid:0)1
4U1957+11.The1-20 keVspectrumofthesourceisbest(cid:12)tted few(cid:2)10 ergs s for the estimated distance of (cid:24)6 kpc.
5
by a power law model with exponential cuto(cid:11), that provides X{ray data were also obtained with the Vela 5B and
9
anapproximation ofathermalComptonisation spectrum.The HEAO{1 satellites. No signi(cid:12)cant long-term periodicity be-

n cuto(cid:11) energy ((cid:24) 2 keV), as well as the X-ray colours, are in- tween 20 d and 1yr was found in the Vela 5B data (Pried-
a termediate between thoseofultrasoft sources containing black horsky & Terrel 1984). In the HEAO{1 A-2 X{ray colour{
J
hole candidates (BHCs) and those of soft LMXRBs contain- colour diagram assembled by White and Marshall (1984)

7 ing an old accreting neutron star. We (cid:12)nd no evidence for a 4U1957+11 lies close to the region occupied by ultrasoft
2 black body spectral component. During the 1985 observation sources, most of which are BHCs in their high state, such
thesource(cid:13)uxwas(cid:24)10%higherthaninthe1983observation, as Cyg X{1, LMCX{3, A0620{00 and GX 339{4. A similar

8 andtheMEspectraprovideevidenceforahighenergytail(up conclusion is obtained based on the EXOSAT X-ray colour-
9 to(cid:24)16 keV),whichstrengthensthesimilarity withthespectra colour diagram (Schultz et al. 1989). These results suggest
0 of high state BHCs. However the source luminosity was only that 4U1957+11 might host a BHC. However, in a recent pa-
36 (cid:0)1
1 (cid:24)5(cid:2)10 ergs s . per Singh et al. (1994) report that the EXOSAT spectra of
0 During the 1985 observation the GSPC spectra revealed 4U1957+11 are similar to those of high luminosity LMXRBs
5 thepresenceofa100 eVEWironK-line emission featurewith containing an accreting weakly magnetic neutron star. Based
9 a centroid energy of 7:06(cid:6)0:09 keV; this value is among the on a (cid:24) 33 hr Ginga observation, Yaqoob et al. (1993) con-
h/ highest measured by EXOSAT. The ME light curves showed clude that the source spectrum is well represented by the
p only a low-amplitude variability on timescales of a few hours. sum of a soft component, which they model with a black-
- No evidence for a periodic modulation with periods between body emitting disk, and a hard component (with a power law
o 0.016 and 13000 s wasfound. photon index of (cid:24) 2) detected up to energies of (cid:24) 18 keV.
r
t The derived values of the inner radius and temperature of
s Key words: X-rays: general { X-rays: binaries { Black holes the disk are rin(cosi)1=2 ' 2 km (i is the inclination angle)
a
{ Stars: neutron and kTin '1:5keV, respectively, suggesting a similarity with
LMXRBs containing a neutron star.
InthispaperwepresentadetailedanalysisoftheEXOSAT
spectralandtiming data.Details oftheEXOSATobservations
1. Introduction aresummarisedinSection 2.Section 3and4describeouranal-
ysis of the source spectra and light curves, respectively. Our
4U1957+11 is a relatively poorly studied 20(cid:0)80(cid:22)Jy X{ray results are discussed in Section 5.
source close tothe galactic plane. Discovered with Uhuru(Gi-
acconietal.1974)thesourcewassubsequently associatedwith
abluestar-like optical counterpart(V=18.7,B{V=0.3andU{ 2. The EXOSAT observations of 4U1957+11
B={0.6), thanks to the SAS(cid:0)3 positioning capabilities (Mar-
gon et al. 1978). This is located within 10 arcsec of the po- The European Space Agency's X{ray observatory EX-
sition of 4U1957+11 and has colours similar to those of the OSAT was operational from May 1983 to April 1986. Its low
optical counterpart of Sco X{1 and 4U0614+09. Optical CCD energy imaging telescopes (LE1 and LE2), used in conjunc-
photometry of 4U1957+11 revealed a (cid:24)0.2 mag roughly sinu- ton with the channel multiplier array (CMA), covered the
soidal modulation with a period of 9.3hr, which most likely 0.05{2 keV band and provided broad band (cid:12)lter spectroscopy.
corresponds to the orbital period of the system (Thorstensen TheArgonchambersofthemediumenergy (ME)proportional
counter array operated over the 1{20 keV band and produced
(cid:0)1=2
Sendo(cid:11)printrequeststo:L.Stella spectra(witharesolution of(cid:1)E=E (cid:24)0:2(E=7 keV) )and
2
Table 1. EXOSATobservations and count rates
(cid:0)1
Time Instrument Filter or Energy Range Exposure (s) Counts s
1983 CMA-LE2 3000 (cid:23)ALexan 6203 0.48(cid:6)0:02
August 27 CMA-LE2 Aluminium/Parylene 5814 0.34(cid:6)0:01
CMA-LE2 Boron 6430 0.21(cid:6)0:01
CMA-LE2 Polypropylene 536 0.46 (cid:6)0:04
ME 1{20 keV 20282 66.9(cid:6)0:2
GSPC 3{10 keV 26654 4.69(cid:6)0:02
1985 CMA-LE1 3000 (cid:23)ALexan 15228 0.42(cid:6)0:01
May 28{29 CMA-LE1 Alumimium/Parylene 10201 0.31(cid:6)0:01
CMA-LE1 Boron 11954 0.18(cid:6)0:007
ME 1{20 keV 37265 80.18(cid:6)0:19
GSPC 3{10 keV 36897 5.19(cid:6)0:02
light curves with high throughput. The ME was often oper-
Fig. 1. MEandLEspectrumfromthe1983observation;the\cut-
ated with one half of the detector array o(cid:11)set from the source
o(cid:11)pl"modelhasbeenused.
inordertoprovideasimultaneousmonitoroftheparticleback-
ground; the pointing directions of the two halves were usually
swapped every 3{4 hours in order to minimize the systematic
uncertainties in the background subtraction. The gas scintil-
lation proportional counter (GSPC) provided a factor of (cid:24) 2
improved spectral resolution in the2-20keVband, with afac-
tor of (cid:24)5 lower e(cid:11)ective area than the ME.
EXOSAT observed 4U1957+11 on 1983 August 27 for
about 6 hr and on 1985 May 28{29 for about 11 hr. The data
were obtained from the EXOSAT database available within
HighEnergyAstrophysics Database ServiceattheAstronomi-
cal Observatory ofBrera(Tagliaferri &Stella 1993)except for
the high time resolution light curves presented in Section 4,
which were accumulated by using ESA's on-line interactive
analysis system. Table 1 gives a summary of the source count
rateandexposure timesineachinstrumentduring thetwoob-
servations.Inallcasestheinstrumentalbackgroundwasstable.
3. Energy Spectra
3.1. ME andLE data
The1{20 keVpulse height source spectra fromthe MEArgon
chambers were analysed together with the source count rates
derived from the LE-CMA and each of the (cid:12)lters used. The
data from the ME Xenon chambers were excluded due to sys-
tematic uncertainties in the background subtraction. Several
trial spectral models wereconvolved through theinstrumental
response and (cid:12)tted to the data. Neither a single power law, a
thermal bremsstrahlung, nor ablack body model produced an
acceptable representation of the spectrum of 4U1957+11. We note that the integrated spectrum of an accretion disk
A power-law model with an exponential high-energy cut- locally emitting like a black body provided also a good (cid:12)t to
o(cid:11) (\cuto(cid:11)pl") provided instead a good (cid:12)t to the 1983 data, the 1983 data. However, the standard (cid:11)-viscosity disk model
for a power-law photon index of (cid:0)S (cid:24)0:2 and a cuto(cid:11) energy (\disk" intable 2and3;cf.Shakura &Sunyaev 1973)requires
39 (cid:0)1
of Ecut (cid:24) 2:1 keV (Table 2 and Fig. 1). The column density a disk luminosity of 5:1(cid:2)10 ergs s or (cid:24) 10 times the Ed-
wasconstrained toavalue of(5:2(cid:6)0:8)(cid:2)1020 Hcm(cid:0)2 mainly dington limit for the estimated mass (cid:24)4 M(cid:12) of the accreting
through the LE rates measurements. black hole. Reconciling this luminosity with the observed LX
(cid:14)
requires an unrealistic inclination ofabout 89:9 and thepres-
The thermal Comptonisation model of Sunyaev & ence of X-ray eclipses, which are excluded, e.g., by the 1985
Titarchuck (1980) (\CompST"), of which the power law with EXOSAT light curves. A similarly good (cid:12)t was obtained by
exponential cuto(cid:11) model provides an approximation, gave in- using thesimple black-body disk model adopted byYaqoobet
stead a substantially worse (cid:12)t. In this case an electron tem- al. 1993 (\diskbb" in Table 2 and 3; see also Mitsuda 1984),
perature of kTe (cid:24) 1:4 keV and a Thomson optical depth of in which the disk inner radius and temperature are allowed
1=2
(cid:28)es(cid:24)22:1wereestimated.Forbothmodelsthe1-20 keVlumi- to freely vary. The best (cid:12)t values, rin(cosi) ' 1:9 km and
36 (cid:0)1
nosityofthesourcewasderivedtobeLX '4:6(cid:2)10 ergs s . kTin ' 1:6keV, are similar to those obtained with the Ginga
3
Table 2. Spectral (cid:12)ts ofthe 1983 data. Errors in the spectral parameters are at 68% con(cid:12)dence level.
2
Date Model (cid:31) =dof Best (cid:12)t parameters
1983 cuto(cid:11)pl 21.6/28 (cid:0)S =0:18(cid:6)0:04; Ecut =2:07(cid:6)0:04keV
Aug. 27 compST 65.2/28 kT =1:37(cid:6)0:01keV; (cid:28)es=22:1(cid:6)0:4
disk 30.9/28 M =3:83M(cid:12), M_ =9.9M_Edd, norma= 1.3(cid:2)10(cid:0)3
b
diskbb 36.2/29 Tin =1:63(cid:6)0:01 keV; , norm =8.02
a 2
norm=cos(i)=d10, where i is the inclination of the disk and d10 is the distance in units of10 kpc.
b 2
norm=(Rin=d10) (cid:1)cos(i),where Rin in Km,is the inner disk radius.
Fig. 2. ME andLE spectrumfromthe1985observation.InpanelA thesinglecomponent\cuto(cid:11)pl"modelhasbeenused.InpanelB a
secondpower{lawcomponentwithaslopeof2hasbeenaddedinorderto(cid:12)tthehighenergyexcess.
(cid:14)
spectra. In turn a very high inclination (i > 87:7 ) would be (range between {1.2 and 2.5). Therefore we (cid:12)xed its value to
required ifthederived inner disk radius weretobecompatible 2, compatible with the measurement by Ginga (Yaqoob et al.
2
with the neutron star radius or the marginally stable orbit of 1993).A(cid:31) =dof of48:9=56wasobtained in thiscase.Thesta-
a few solar masses black hole. tistical signi(cid:12)cance oftheadditional powerlawcomponentwas
(cid:0)2
evaluatedthroughanF-testtobe(cid:24)10 (Fig. 2B).Therefore
Thesourcecount rateincreased by(cid:24)20%during the1985
we conclude that the 1985 EXOSAT data con(cid:12)rm the Ginga
observation. The values obtained from the spectral (cid:12)tting of
result of an additional spectral component above (cid:24) 10 keV.
the 1985 data are similar to those discussed above (Table 2),
The 1-20keV source luminosity during the 1985 observation
except for a rather noticeable high energy-excess in the ME
36 (cid:0)1
rates above 10 keV. This excess was not adequately modelled wasLX '5:1(cid:2)10 ergs s ,withthe high energy component
contributing up to (cid:24)10% of the total.
by any of the single component models adopted for the 1983
observation.Thecuto(cid:11)powerlawprovidedthebestsinglecom-
2
ponentmodelwitha(cid:31) of54:8for57degreesoffreedom(dof)
(see Fig. 2 A). To model the high-energy excess an additional
3.2.GSPC observations
spectral component consisting ofa power law wasadded. Ow-
ing to poor statistics, the photon index of this high energy The 3{10 keV GSPC spectra from both observations were (cid:12)t-
power law could not be obtained with an acceptable accuracy tedwith the samecontinuum model that provided thebest (cid:12)t
4
Table 3. Spectral (cid:12)ts ofthe 1985 data.
2
date model (cid:31) =dof Best (cid:12)t parameters
1985 cuto(cid:11)pl 54.8/57 (cid:0)S =0:19(cid:6)0:03; Ecut =2:08(cid:6)0:03 keV
May 28{29 compST 137/57 kT =1:37(cid:6)0:01 keV; (cid:28) =22:13(cid:6)0:26
disk 72.8/56 M =3:8M(cid:12), M_=9.9 M_edd, norm= 1.5(cid:2)10(cid:0)3
diskbb 83.5/58 Tin =1:63(cid:6)0:01 keV; , norm= 9.01
Fig. 3. GSPCspectrumfromthe1985observation.PanelA showsthe(cid:12)tusingthe\cuto(cid:11)pl"modelplusaGaussianline.PanelBshows
theresidualswithrespecttothecontinuum.
totheMEandLEdata(notethatnoevidenceforahighenergy the FWHM. We note that by adopting the alternative X-ray
excess is found in theME databelow 10 keV).Thephotoelec- continuum models discussed in Sect.3.1, the derived line pa-
tricabsorption waskeptconstantatthebest(cid:12)tvalueobtained rameters remain essentially unchanged.
2
from the ME and LE data. The best (cid:12)t model gave a (cid:31) =dof
of 120=100 and 138:8=97 for the 1983 and 1985 observations,
respectively (Fig. 3).
4. Time variability
Adding a Gaussian line pro(cid:12)le with a centroid energy in
the 6{7 keV range did not improve signi(cid:12)cantly the (cid:12)t of the 4.1.Light curves
1983 GSPC spectrum. The 90% con(cid:12)dence upper limit to the
equivalent width of such a feature was (cid:24) 90 eV. On the con- During the observations of 1983 and 1985 the ME provided
2
trary, adding a Gaussian line pro(cid:12)le decreased the (cid:31) =dof of respectively 16and64channelpulse heightanalyser datafrom
the 1985 GSPC spectra to 120:1=95, corresponding to an F- both the source and background detectors. Using these data,
test probability of (cid:24)0:001 (see also Gottwald & White 1990). lightcurves with 0.3125, 0.625 and 30 s time resolution in the
The centroid energy and equivalent width of the line werede- energyranges1{4 keV,4{10 keVand1{10 keVwereextracted
termined to be 7:06(cid:6)0:09 keV and 102(cid:6)23 eV, respectively. from the EXOSAT database. The lightcurves display a small
The width of the line could not be resolved by the GSPC and amplitude aperiodic variability on long timescales which is
an upper limit of 0:9 keV (90% con(cid:12)dence) was obtained for more pronounced in the 1985 observation (see Fig. 4).
5
Fig. 4. 1{10keVMElightcurvesfrombothobservations.Eachbincorrespondstoanintegrationtimeof166s
4.2. Power Spectra con(cid:12)dence upper limits to the modulation semi-amplitude are
given in Table 4 for selected periods.
The lightcurves described above were used to carry out a de-
tailed search for coherent pulsations. Since the 1985 observa-
tion is longer and has a higher average count rate than the Table 4. 3(cid:27) upper limits to the semi-amplitude of pulsations
1983, it allows for a more sensitive search. In order to max- for selected periods
imize the Fourier resolution, and therefore the sensitivity, a
single power spectrum was calculated for the 0.3125, 0.625 Period Upper limits
and 30 s resolved lightcurves of each observation and energy (s) 1983 August 27 1985 May 28{29
range. Missing data points were replaced with the average 1{10 1{4 4{10 1{10 1{4 4{10
count rate. In Fig. 5 the power spectra are marked with \a"
13650 (cid:0) (cid:0) (cid:0) 11:8 6:69 16:8
andnormalised suchthatthewhitenoisearisingfromcounting
10240 (cid:0) (cid:0) (cid:0) 7:71 4:13 11:2
statistics corresponds to a power of2.
8192 9:47 15:3 8:41 2:91 8:41 8:02
Arednoise component is dominant in thefrequency range 4096 4:52 6:32 3:27 0:83 0:85 1:52
(cid:0)5 (cid:0)4
from(cid:24)5(cid:2)10 Hzto(cid:24)3(cid:2)10 Hz.Thiscomponent results 512 0:55 0:69 0:81 0:41 0:52 0:59
from the long{timescale variability visible in the lightcurves. 16a (cid:0) (cid:0) (cid:0) 0:67 (cid:0) (cid:0)
The presence of this continuum power spectrum component 8{0.03a (cid:0) (cid:0) (cid:0) (cid:20)0:85 (cid:0) (cid:0)
a(cid:11)ects the statistical distribution of the power estimates for 0.02a (cid:0) (cid:0) (cid:0) 0:99 (cid:0) (cid:0)
long periods. This means that techniques which assign proba- 0.016a (cid:0) (cid:0) (cid:0) 1:17 (cid:0) (cid:0)
2
bilities to power spectrum peaks assuming the simple (cid:31) dis-
a
From the high time resolution ME data (1{20 keV).
tribution ofpower expected fromcounting statistics noise (see
e.g.Leahyetal.1983)maysigni(cid:12)cantly overestimatetheirsig-
ni(cid:12)cance. Israel & Stella (1995) developed a technique for re- For short periods, up to (cid:24)4000 s, the sensitivity of the
liably searching forpower spectrum peaks, corresponding toa search is mainly limited by counting statistics noise and up-
periodic modulation, in the presence of \coloured" continuum per limits of (cid:20)1%are obtained for any sinusoidal modulation.
power spectrum components. Asearch forcoherent pulsations For longer periods, the source red noise component plays an
is carried out by looking for signi(cid:12)cant peaks above the es- increasingly important role in reducing the sensitivity of the
timated continuum spectrum. If no peaks with a probability search. The corresponding upper limits have values of (cid:24) 3{
of chance occurrence lower than a given threshold are found, 17% for periods between 8000 and 14000 s. The EXOSATob-
thenanupperlimittothesemi-amplitude ofasinusoidal mod- servations were notlong enough to search fora low amplitude
ulation is worked out for each Fourier frequency of the power X{ray modulation at the 9.3 hr orbital period.
spectrum through a generalisation of the method outlined by We analysed also the powerspectra of the0.3125 and 30 s
Leahyetal.(1983).Byusing thistechnique, eachofthepower resolved 1{10,1{4 and 4{10 keVME light curves fromthe(cid:24)6
spectra of4U1957+11wassearched forpeaks withaprobabil- hr 1983 August 27 observation. All the points in the power
ity of (cid:20)0.3% (3(cid:27) con(cid:12)dence level) of exceeding by chance the spectraarewell belowthedetection threshold andthusnosig-
level of the continuum power spectrum components ni(cid:12)cant periodicities are found. Upper limits for selected fre-
For the 11 hr 1985 May 28{29 observation, power spec- quencies aregiveninTable 4andshowninFig. 6(forthe1{10
tra were calculated from the 0.625 and 30 s time resolved keV energy range).
lightcurves in the 1{10, 1{4 and 4{10 keV bands. No signif- ME count rates from the entire energy range of the Ar
icantpeakswerefoundinanyofthepowerspectra.The99.7% chambers (1{20 keV) were also available with a resolution of
6
Fig. 5. Power spectra (a) of the ME lightcurveobtained from the 1{10keV data (panel A) and from the high resolution data in the
entirerangeoftheMEAr chambers(panelB)duringthe1985May28{29observation.The99:7%con(cid:12)dencethresholdforthedetection
of a sinusoidalsignal is shown (b). The bottom curves (c) give the correspondingupper limits to the fractionalsemi-amplitudeof the
modulation.
(cid:24) 8 ms for the 1985 May 28{29 observation. We divided the the characteristic of its Comptonised spectral component is
lightcurve in301intervals, inordertopreventtheDoppler fre- therefore premature.
quencyvariations associatedtotheorbitalmotionfromsmear-
On the other hand the 1985 EXOSAT spectra and, espe-
ingthepowerofasinusoidal signal overmorethanoneFourier
cially, the Ginga spectra show the presence of a high energy
frequency. These intervals were used to calculate 301 power
power-law component above energies of 10 keV. This kind of
spectra of 8192 indipendent Fourier frequencies.
two-component spectra has been observed from several BHCs
The power spectrum resulting from the average of these
intheirhighstate,theluminosity ofwhich,however,isafactor
power spectrawasthen analysed with thesame technique dis-
of >10 higher than that of4U1957+11.
cussed above. No signi(cid:12)cant periodicities were detected. The
upper limits are reported in Table 4 and plotted in Fig. 5. Our decomposition of the 1983 EXOSAT spectrum of
4U1957+11 is similar to that of Singh et al. (1994), who in
their analysis used also the MEXenon chambers data and ex-
5. Discussion
cluded the LE-CMA rates. However, according to these au-
ThemaincomponentoftheEXOSATspectraisconsistentwith thors, the 1985 EXOSAT spectrum of 4U1957+11 consists of
a power law model with exponential cuto(cid:11) characterised by a the sum of a (cid:24)1keV temperature black body, accounting for
spectral indexof(cid:24)0:2andacuto(cid:11)energyof(cid:24)2keVforboth about 40% of the total luminosity, plus a Comptonised spec-
the 1983 and 1985 observations. This spectral form provides trumextendingupto(cid:24)20keV.Inthiscase4U1957+11would
a simple analytical approximation to the thermal Comptoni- posses an X-ray spectrum similar to that of the high luminos-
37 (cid:0)1
sation spectrum of Sunyaev & Titarchuk (1980) and, besides ity LMXRBs (> 10 ergs s ) containing an old neutron star
semplicity, itinvolves noconceptualdi(cid:11)erence. Indeedthermal (the so-called \Z{sources")(White etal. 1988,Hasinger & van
Comptonisation models (and their approximations) are found derKlis 1989).Such aspectral decomposition ofthe1985EX-
to provide very good (cid:12)ts to the main spectral component of OSATdataappearshoweverunlikely,forthefollowingreasons:
a number of high state BHCs and LMXRBs containing old (a)despitethemoderate((cid:24)20%)increaseofsourceluminosity,
weakly magnetic neutron stars (White et al. 1988). theparameters oftheComptonised spectral component would
In high state BHCs this spectral component is somewhat change drastically from the 1983 to the 1985 observation (a
softer then that of 4U1957+11 (Ecut (cid:24) 1:4 keV in LMCX{3 factorof>10and(cid:24)0:1variation inelectrontemperatureand
andLMC X{1;Ricci1995).InLMXRBscontaininganoldneu- Thomson depth, respectively), when it would be virtually in-
tron star the component is instead harder (e.g. Ecut (cid:24)7 keV distinguishable from a power law; (b) neutron star LMXRBs
in 4U1705-44 and Ecut (cid:24) 5 keV in 4U1636-53; Ricci 1995). of luminosity comparable to that of 4U1957+11 do not show
Aclassi(cid:12)cation of thecompact object in 4U1957+11 based on evidence for an additional black body component, with upper
7
Fig. 6. SameasFigure5,butforthe1{10keVMElightcurveofthe1983August27observation.
limits usually in 10{20% range. spectra and light curves were obtained through the High En-
Black body emission disk models can be (cid:12)t to the data in ergyAstrophysicsDatabase ServiceattheBreraAstronomical
place of the cuto(cid:11) power law model (or the Comptonisation Observatory and theon-line interactive analysis system atthe
(cid:14)
model), buttheyrequire anunrealistic inclination ofi>87:7 Space Science Department of ESA, ESTEC. This work was
that cannot be reconciled with the absence of X-ray eclipses. partially supported by ASIgrants.
ThisisalsounlikeotherhighstateBHCsandanumberofhigh
luminosity LMXRBs (White et al. 1988).
References
A Fe K-shell emission feature is detected at an energy of
7:06(cid:6)0:09 keV in the 1985 data. The centroid energy of such EbisawaK.,1991,InstituteofSpaceandAstronauticalScience,
a line is consistent only with K(cid:11) transitions from H-like ions ISAS RN 483
(E0 = 6:96 keV) and is among the highest revealed with the Fabian A.C.,ReesM.J.,Stella L.,White N.E.,1989,MNRAS,
EXOSAT GSPC (Gottwald & White 1991). K(cid:11) transitions 238, 729
from any of the lower ionisation stages of iron require a sub- Giacconi, R.,Murray,S.,Gursky,H.etal.,1974,ApJS,27,37.
stantial blueshift. Similar to the modelling of the Fe K-lines Gottwald M., White N.E., 1991, Lectures Notes in Physics,
from other X-ray binaries and AGNs, such a blueshift could 385, 134
result from (relativistic) Doppler e(cid:11)ects due to bulk plasma Hasinger G.,van der Klis M., 1989, A&A,225, 79
motions in the vicinity of the collapsed object. In particular, Israel G.L.,Stella L.,1995, in preparation.
the observed feature might correspond to the blue horn ofthe Kallman T.,White N.E.,1989, ApJ,341, 955
characteristic pro(cid:12)le producedintheinnermostregion ofarel- Leahy D. A., Darbro W., Elsner R. F. et al., 1983, ApJ, 266,
ativistic accretiondisk(Fabianetal.1989).Thelineequivalent 160.
width ((cid:24) 100 eV) is in the range measured from a number of MargonB.,ThorstensenJ.R.,BowyerS.,1978,ApJ,221,907.
LMXRBs (White et al. 1986). We note that the di(cid:11)erent line Mitsuda Ket al.., 1984, PASJ,36, 74.
centroidenergyderivedbySinghetal.(1994)likelyresultsfrom Priedhorsky W. C.,Terrel J.,1984, ApJ, 280, 661.
their drastically di(cid:11)erent modelling ofthe continuum. Ricci D.,1995, Ph.D. Thesis, in preparation.
TheEXOSATlightcurvesof4U1957+11showonlyamod- Shakura N.I.,Sunyaev R.A.,1973, ApJ, 24, 337.
erate variability ontimescales ofa fewhours, similar toBHCs Schultz N. S.,Hasinger G., J.Tru(cid:127)mper,1989, A&A,225, 48.
in their high state(e.g.LMCX{3;Trevesetal. 1988,Ebisawa Singh K. P., Apparao K. M. V., Kraft R. P., 1994, ApJ, 421,
1991).Asearch for periodicities revealed no coherent modula- 753.
tion. Sunyaev R. A.,Titarchuck L. G.,1980, A&A,86, 121.
TagliaferriG.,StellaL.,1993,Mem.Soc.Astron.Ital.,Vol.XX,
Acknowledgements.WearegratefultoDr.S.Ilovaiskyforhelp- in press.
fulcommentsonanearlierversionofthispaper.TheEXOSAT Thorstensen J.R., 1987, ApJ,312, 739.
8
TrevesA.,Belloni T.,Chiappetti L.etal.,1988,ApJ,325,119
White N.E.,Marshall F. E.,1984, ApJ, 281, 354.
White N.E.,Stella L.,Parmar A.N.,1988, ApJ,324, 363
White N. E., Peacock A., Hasinger G. et al, 1986, MNRAS,
218, 129
Yaqoob T.,Ebisawa K.,Mitsuda K.,1993, MNRAS,264,411.
Noteaddedinproof.Inthe recently distributed preprint \EX-
OSATGSPCironlinecatalog"byGottwaldetal.(1994,A&A,
in press),noiron line detection isreported forthe1985GSPC
spectra of 4U1957+11, despite the fact that the continuum
model is the same as that used in our paper. This is surpising
sincewehaveveri(cid:12)ed thatthedetection condition ofGottwald
2
et al. (a (cid:31) improvement of (cid:21) 13) is met also for the energy
rangeusedintheirpaper.Wecanspeculate thatperhaps their
2
automatic (cid:12)tting procedure failed toreach theminimum (cid:31) in
the case of4U1957+11.
ThisarticlewasprocessedbytheauthorusingSpringer-VerlagLaTEX
A&A style(cid:12)le1990.