Mon.Not.R.Astron.Soc.000,000–000 (0000) Printed1February2008 (MNLATEXstylefilev1.4) Irradiation of the secondary star in X-ray Nova Scorpii 1994 (=GRO J1655–40) T. Shahbaz1, P. Groot2, S.N. Phillips1, J. Casares3, P.A. Charles1, J. van Paradijs2,4 1Universityof Oxford, Department of Physics, Nuclear Physics Laboratory, Keble Road, Oxford, OX13RH, UK 2Astronomical Institute “Anton Pannekoek”, University of Amsterdam and Centerfor HighEnergy Astrophysics, Kruislaan, 403, 1098 SJ Amsterdam, The Netherlands 3Institutode Astrof´ısica de Canarias 38200 La Laguna, Tenerife,Spain 0 4PhysicsDepartment, UAH, Hunstville, Alabama 35899, USA 0 0 2 1February2008 n a J ABSTRACT 3 1 We have obtained intermediate resolution optical spectra of the black-hole candidate Nova Sco 1994 in June 1996, when the source was in an X-ray/optical active state 1 (R∼15.05). We measure the radial velocity curve of the secondary star and obtain a v semi-amplitude of 279±10kms−1; a value which is 30 per cent larger than the value 7 obtained when the source is in quiescence. Our large value for K2 is consistent with 2 60+9 per cent of the secondary star’s surface being heated; compared to 35 per cent, −7 2 which is what one would expect if only the inner face of the secondary star were 1 irradiated. Effects such as irradiation-induced flows on the secondary star may be 0 0 important in explaining the observed large value for K2. 0 Key words: accretion, accretion discs – binaries: close – stars: individual: X-ray / h Nova Sco 1994 (GRO J1655–40)– X-rays: stars. p - o r st 1 INTRODUCTION of 228.2±2.2 kms−1. Shahbaz et al. (1999) using only a quiescent data, determined the true radial velocity semi- v: The soft X-raytransients, asubclass of thelow-mass X-ray amplitudeK2=215.5±2.4kms−1whichgivesarevisedvalue i binariesdistinguishedbytheirX-rayoutbursts,haveproved for the mass function of f(M) = 2.73±0.09 M⊙. They also X to be an ideal hunting ground for stellar-mass black hole measured the rotational broadening of the secondary star ar candidates (Tanaka & Shibazaki 1996). The system Nova which then gives the binary mass ratio q∼0.39 (=M2/M1, Sco1994(=GROJ1655–40)isparticularlyinteresting,since whereM1 andM2 arethemassesofthecompactobjectand as well as being asource of superluminal jets (Zhanget al., secondary star respectively). 1994;Harmonetal.,1995),itsopticalbrightnessandpartial The effect of heating of the secondary is to shift the eclipse features mean that it is one of the few systems that ‘effectivecentre’ of thesecondary, weighted bythestrength hasyielded areliable estimatefor themass of thecollapsed of the absorption lines, from the centre of mass of the star. star. One expects that this results in a significant distortion of theradial velocity curveand rendersa sinusoidal fit clearly Nova Sco 1994 was discovered on July 27 1994 with inadequate,leadingtoaspuriouslyhighradialvelocitysemi- BATSE on board the Compton Gamma Ray Observatory amplitude. In order to quantify this effect we have deter- (Zhang et al., 1994). It has been studied extensively dur- mined the radial velocity variations of the secondary star ing the past few years in X-rays and at optical and ra- in Nova Sco 1994, when it was in outburst and compared dio wavelengths (Bailyn et al., 1995a and b, Zhang et al., ourresults with others obtained using datataken when the 1995, van der Hooft et al. 1998). Strong evidence that source was in different X-ray states. the compact object in Nova Sco 1994 is a black hole was presented by Bailyn et al. (1995b) who initially estab- lished a spectroscopic period of 2.601 ± 0.027 days, clas- sified the secondary as an F2–F6iv type star and sug- 2 OBSERVATIONS AND DATA REDUCTION gested a mass function f(M)=3.16±0.15 M⊙. An im- 2.1 Spectroscopy proved value of f(M)=3.24±0.09 M⊙ was presented by Orosz & Bailyn (1997) using both quiescent and out- Intermediate resolution optical spectra of Nova Sco 1994 burst data, derived from a radial velocity semi-amplitude were obtained on 1996 June 20–24 with the 1.54-m Dan- (cid:13)c 0000RAS 2 T. Shahbaz, P. Groot, S.N. Phillips, J. Casares, P.A. Charles, J. van Paradijs Table 1.Logofobservations Table 2.Hαequivalentwidth Night# #ofspectra seeing(′′) Orbitalphase Night# Orbitalphase EW(˚A) covered Night1 11 2.0 0.722–0.834 Night1 0.78 9.8±0.5 Night2 5 2.6 0.100–0.225 Night2 0.16 11.9±0.3 Night3 11 2.3 0.501–0.592 Night3 0.54 10.6±0.4 Night4 10 1.6 0.883–0.964 Night4 0.93 3.8±010 Night5 10 3.0 0.270–0.353 Night5 0.32 10.4±0.8 ishTelescopeattheEuropeanSouthernObservatory(ESO) Hαemissionline(meanequivalentwidthof7.5±0.06˚A)and in Chile using the Danish Faint Object Spectrograph and a much weaker Hei 6678˚A (equivalent width of 0.4±0.03˚A) Camera (DFOSC). We used grating #8 which gave a dis- can be seen. In Table 2 we list the Hα equivalent width for persion of 1.26 ˚A perpixeland a wavelength coverage from thenightlyaverages.TheFeiabsorptionblendat6485,6496, 5865–8336˚A. The Loral 2048×2048 CCD was used, binned 6499,6502˚Aisalsovisible.Thesefeaturesareusedtodeter- by a factor two in the spatial direction in order to reduce mine the radial velocity of the secondary star (see section the readout noise, but not binned in the dispersion direc- 4). The 6613˚A diffuseinterstellar band is also present. tion.Theseeingduringtheobservationswaspoorandvari- TheemissionlinesinNovaSco1994aredouble-peaked, able (see section 2.2) so we used a slit width of 2′.′5 on the which is presumably a consequence of the system being at first night and then 2′.′0 for the other nights. This resulted highinclination.Wecancomparetheobservedpeak-to-peak in spectral resolutions of 7.6 ˚A and 5.5 ˚A for the first and half separation of the Hα emission line (which arises from other nights respectively. Wavelength calibration was per- the accretion disc) with the projected velocity of the outer formed using a Cu-Ararc. A total of 47 spectra were taken disc edge. In a binary system with a mass ratio > 0.25 it eachhavingexposuretimesof1800s(seeTable1fordetails). is generally assumed that the accretion disc cannot grow The data reduction and analysis was performed us- ing the Starlink figaro package, the pamela routines of larger than thetidal truncation radius, rd (Paczynski1977; Whitehurst 1988; Osaki, Hirose & Ichikawa 1993), which is K.HorneandthemollypackageofT.R.Marsh.Removalof approximatelygivenbyrd =0.60a/(1+q) for0.03<q<1, theindividualbiassignal was achievedthrough subtraction where a is the binary separation (Warner 1995). Given the of a median bias frame. Small scale pixel-to-pixel sensitiv- systemparameters(Porb=2.62168days;q∼0.39;i∼69de- ityvariationswereremovedwith aflat-fieldframe prepared from observations of a tungsten lamp. One-dimensional grees; M1 ∼6.7 M⊙ see Shahbaz et al., 1999) the minimum value for the projected velocity of the accretion disc rim spectra were extracted using the optimal-extraction algo- is∼394 kms−1.Theobservedpeak-to-peakhalfseparation rithm of Horne (1986), and calibration of the wavelength of the Hα emission line (see Figure 1) in late June 1996 is scale was achieved using 5th order polynomial fits which 385±8 kms−1(measuredbyfittingtheprofilewithadouble gaveanrmsscatterof0.03˚A.Thestabilityofthefinalcali- Gaussian), which implies that the accretion disc is close to brationwasverifiedwiththeOHskylineat 6300.3˚A whose its maximum possible size. Soria et al. (1998) estimate the position was accurate towithin 0.1 ˚A. Hα half peak-to-peak separation to be <350/2 kms−1and <550/2 kms−1for their August/September 1994 and June 2.2 Photometry 1996 Hα observations respectively, velocities much lower than expected, suggesting that the Hα emission line arises Using the same setup as for the spectroscopy, we also ob- from non-Keplerian regions/flows in theaccretion disc. tainedlimitedBessellr-bandimagesofNovaSco1994every night. The data were debiased using a median bias frame, but not flat-fielded,as nonewere taken. These images were used to estimate the seeing each night (see Table 1). We 4 THE RADIAL VELOCITY OF THE applied aperture photometry to Nova Sco 1994 and several SECONDARY STAR nearby comparison stars within the field of view. Johnson V- and R-band magnitudes of these comparison stars were The radial velocities of the F-type secondary star in Nova made available to us by J. Orosz. We determined the rela- Sco1994 were measured from thespectrabythemethodof tivemagnitudeofNovaSco1994withrespecttothreestars cross-correlation(Tonry&Davis1979)withatemplatestar. having a range of colours [(V −R)=0.49, 0.77 and 1.23]. Priortocross-correlation thespectrawereinterpolatedonto Assuming that the colour correction between the two filter a logarithmic wavelength scale (pixel size 55 kms−1) using systems is small (<0.05 mags; similar to the accuracy of asin x/xinterpolation schemetominimizedatasmoothing our photometry) and that Nova Sco 1994 has a colour in (Stoveretal.1980),andthennormalised.Thetemplatestar the same range as the comparison stars used, we estimate spectrum(HR2906;F6v)wasthenartificiallybroadenedby R∼15.05 for NovaSco 1994. 90 kms−1(Shahbazetal.,1999)toaccountfortherotational velocityofthesecondarystar.Notethattheorbitalsmearing oftheNovaSco1994 spectrathroughthe1800s exposureis atmostonly10 kms−1,muchlessthantheresolutionofthe 3 THE SPECTRA OF NOVA SCO 1994 data. Only regions of thespectrum devoid of emission lines InFigure1weshowthevariance-weightedaverageandalso (6400-6520˚A) were used in thecross-correlation. Theradial thenightlyaveragesoftheNovaSco1994spectra.Astrong velocity of the template star (derived using the position of (cid:13)c 0000RAS,MNRAS000,000–000 Irradiation of the secondary star in X-ray Nova Scorpii 1994 (=GRO J1655–40) 3 theHαabsorptionlinetobe−7 kms−1)wasthenaddedto we expect the total radiated power for X-ray transients in theradial velocities of NovaSco1994. bothoutburstandquiescencetolie(Chen,Shrader&Livio Using the orbital ephemeris given by van der Hooft et 1997). Note that there is a correlation between X-ray lu- al(1998) wephase-foldedandbinnedtheheliocentricradial minosity and the observed radial velocity semi-amplitude; velocities (see Figure 2). From figure 2, it can be seen that the higher the X-ray luminosity the larger the observed ra- theradialvelocitymeasurementatphase0.2doesnotfitthe dial velocity semi-amplitude, exactly as expected. We can general pattern of the sinusoidal modulation present in the useourmodel toestimate theX-rayluminosity at thetime data. This data point was the total of three radial velocity whenOrosz&Bailyn(1997) tooktheirradialvelocitymea- measurements taken on the second night (21st June 1996). surements.WefindLx ∼5×1035 ergs−1whichisconsistent Althoughtheseeingandqualityofthespectratakenduring with theBATSEupperlimit. thisnightwerenotasgoodastheothers,noobviousreason could be found as to why these spectra gave much lower radial velocities than expected. A sine wave fit to all the datapointsdoesnotgiveanadequatefit(χ2ν=6.9).However, 6 IRRADIATION OF THE SECONDARY STAR removing the discrepant data point and then performing a It has been known for some time, especially in studies of sine wave fit yields a χ2ν of 1.5, a semi-amplitude K2 = dwarf novae and polars, that substantial heating of the 279±10 kms−1, systemic velocity γ = −155±7 and a secondary star shifts the effective centre of the secondary, phase shift of −0.043±0.005φ (1-σ errors are given). We weighted by the strength of the absorption lines, from the also fitted the radial velocity curve with an eccentric orbit, centre of mass of the star. This results in a significant dis- but found thefit tobe less than 50 percent significant. tortion of the radial velocity curve leading to a spuriously high semi-amplitude and a radial velocity curve that may be eccentric. Davey & Smith (1992) describe a procedure for detectingtheeffectsof irradiation on theradial velocity 5 THE EFFECT OF IRRADIATION ON THE curve of the secondary star, whereby one tests the signifi- SECONDARY STAR’S RADIAL VELOCITY cance of an eccentricity in the orbital solution. However, it Three absorption line radial velocity curves have been ob- shouldbenotedthat,althoughourdatadoesnotallowthis tained for Nova Sco 1994, using the same absorption fea- eccentricitytest,duetothepoororbitalphasecoverage,we tures of the F6iv secondary star and the standard method can use the spuriously high radial velocity semi-amplitude of cross-correlation. However, in each case the system was to show that X-ray heating is present. observed to be in a different X-ray state. A sinusoidal fit In order to investigate the effects of X-ray heating on to the outburst data taken in April/May 1995 of Orosz thesecondarystar’sradialvelocitycurveweusedthemodel & Bailyn (1998) gives a radial velocity semi-amplitude of describedbyPhillips,Shahbaz&Podsiadlowski(1999).The Kobs=230±2 kms−1. During this period BATSE did not model uses a crude treatment for X-ray heating, since no detect the source, so we can only put an upper limit of satisfactory robust model exists for the effects of external 2.4×1036 ergs−1(<0.03photonsergcm−2s−1intheBATSE heatinginstars.However,itservestoillustratetheextreme 20-350 keV energy range) to the X-ray luminosity of the effectsofX-rayheating.Itshouldbenotedthatthefirstor- source. This upper limit alone does not allow us to state dermodelofX-rayheatingbyBrett&Smith(1993),which unequivocally that the source was not active at X-ray en- does not include energy transport effects, does show that ergies, but optical observations suggest that thesource was the whole temperature structure of the outer layers of the not in quiescence (V=16.5; Orosz & Bailyn 1998). In sec- secondary is upset by external heating. Figure 3 shows the tion4wedeterminedK2=279±10kms−1fromdatatakenin effects of different amounts of X-ray luminosity on the sec- June1996whenRXTEASM(2-12keV)observationsgivean ondary star’s radial velocity amplitude. Kobs is computed X-ray luminosity of Lx=6.8×1037 ergs−1, and the R-band by fitting the predicted curve with an eccentric orbit. The brightness was ∼ 1 mag brighter than its quiescent value. regions on the secondary star that are heated do not con- The BATSE (20–350 keV) count rate was at least a factor tribute to the absorption line flux. The maximum possible of 4 higher than in April/May 1995. Shahbaz et al. (1999) change that irradiation can have on Kobs, based purely on determined the true radial velocity of the secondary star geometry, is 15 per cent. However, from our data presented (K2=215.5±2.4 kms−1)in1998May/June,whenthesource inthispaper,we observe∆K2/K2=0.30±0.05, which when was finally in optical quiescence. The only X-ray quiescent comparedwithmaximumpossiblevaluebasedongeometry, observationswereobtainedduringMarch1996usingASCA is significant at the3-σ level. (1-10 keV; Robinson et al., 1997) which gave Lx=2×1032 In Figure 4 we show how much of the secondary star’s ergs−1. surfaceneedstobeheatedinordertoproducetheobserved In Figure 3 we show the observed radial velocity am- radial velocity amplitude. Wefind that based purely on ge- plitudes relative to the quiescent value as a function of the ometry 35 per cent of the secondary star’s surface is di- observedX-rayluminosityatthetimeofthemeasurements. rectly heated by X-rays produced at the compact object. We have converted the X-ray luminosities, which were ob- (This fraction only depends on the shape and size of the served with different instruments, into a common energy secondary star, which in turn is determind by the q. Us- range (0.4–10 keV) using a hydrogen column density of ing the extreme values for q (Shahhazet al., 1999), we find Nh =0.89×1022 cm−2 andaphotonpower-lawmodelwith that this fraction changes by less than 1 per cent.) How- indices 2.8 and 1.5 for the X-ray high and quiescent states ever, in order to produce the observed large radial velocity respectively (see Table 3; Zhang et al., 1997; Robinson et semi-amplitude, 60+9 per cent of the secondary star needs −7 al.,1997;Hameuryetal.,1997).Thisenergyrangeiswhere to be heated. The 1-σ uncertanties quoted here were esti- (cid:13)c 0000RAS,MNRAS000,000–000 4 T. Shahbaz, P. Groot, S.N. Phillips, J. Casares, P.A. Charles, J. van Paradijs Table 3.RadialvelocityandX-rayluminositymeasurements Kobs Lx power-lawindex Corrected Lx (kms−1) (ergs−1) Nh=0.89×1022 cm−1 (ergs−1) 215.5±2.4 2.0×1032 ASCA(1-10keV) 1.5 2.1×1032 232±2 <2.4×1036 BATSE(20-350kev) 2.8 <1.6×1037 279±10 6.8×1037 RXTE(2–12 keV) 2.8 8.5×1037 nated using the 1-σ uncertanties in ∆K2/K2. This result accreting,steadilyburningwhitedwarfirradiatestheaccre- may seem surprising at first, since one expects only the re- tion disk and the secondary star, as suggested by van den gions of thesecondary star facing thecompact object to be Heuvel et al. (1992). A simple description of energy trans- irradiatedandyetourresultimpliesthatsomeoftheregions portonthesecondarysurfacewasusedandthenintegrated notdirectlyseenbythecompactobject arealsoaffectedby overthewholesurface,whileconservingthetotalluminosity. irradiation.However,oneshouldnotethateffectssuchasX- They found that significant energy transport of the irradi- ray scattering and irradiation-induced flows on the surface ated fluxtonon-illuminated parts on thesecondary surface of the secondary star (Phillips & Podsiadlowski 1999) can is requiredto simulate theobserved lightcurve, particularly increase thefraction of thesecondary star that responds to aroundtheprimaryeclipse,whentheshadowedhemisphere the X-ray source. Note that the regions on the secondary of thesecondary is in view. star that are shadowed by the accretion disc will be indi- Recent models for irradiation-induced flows in binary rectly heated by such mechanisms. Therefore Kobs can be stars have been computed by Martin & Davey (1995). largerthanthatexpectedfromheatingtheinnerfaceofthe Theyconsideredcirculationingently-heatedsecondarystars secondary star alone. (wheretheincidentfluxislessthantheintrinsicflux).Their 2-dimensional calculations included the effects of the Cori- olis force and showed upwelling of hot material being car- ried preferentially towards the direction of rotation of the 7 DISCUSSION star. They also concluded that all secondary stars should Theexistenceofcirculationinrotatingstarswasfirstproved show asymmetric heating, because of the presence of Cori- in 1924 by von Zeipel (von Zeipel 1924). He demonstrated olis forces. Phillips (1999) has recently extended the study that for a rotating homogeneous star, the radiative trans- of circulation to 3-dimensions. As well as including the ef- portequationandequationofconservationofenergycannot fects of X-ray irradiation i.e. the anisotropic heating of the be fulfilled simultaneously. This results in the formation of irradiatedsurface,andtheeffectsofsurfaceradiationstress, meridional motions. Inorder tomaintain astationary state healsoconsidersthelarge-scaleeffectsoftherotationofthe as assumed, one has to demand that these meridional mo- systemandincludesanapproximatetreatmentoftheCorio- tions contribute to the energy transport. In the case of an lisforce.HisresultssuggestarealisticanalysisoftheCoriolis irradiated rotating star, the situation is still more compli- force is essential for a full description of stellar circulation. cated, since the radiation will induce additional circulation In order to study the extent of irradiation of the sec- currents. ondary star one requires good quality spectrophotometric Evidence for the existence of significant irradiation- studiesthroughoutanX-rayoutburst,duringwhichthelevel driven circulation is provided byseveral sources. For exam- of X-ray irradiation and induced heating changes. This will ple,theanalysisoftheopticallightcurveofHZHerculishas allowthesurfaceintensitydistributionacrossthesecondary shown this to be heated by its accompanying X-ray source startobemapped(seeRutten&Dhillon1994andDavey& HER X-1. Although the main features of the optical light Smith1996), from which effectssuchas irradiation-induced variationarewellunderstood(HZHerisbrightwhentheX- circulation or star-spots can beinvestigated. raysourceisinfrontofit,itsbrightnessisreducedduringthe occultation oftheX-raysourcebythesecondary),themin- imum at phase 0.0 is sharper than expected and indicates some additional source of optical radiation at this phase. ACKNOWLEDGEMENTS Strittmatter et al. (1973) tried to explain this via the illu- We would like to thank the referee, Prof. Robert Smith for minationofthediskbyHZHer.Otherattemptsweremade useful comments. by Pringle (1973) and Bahcall, Joss & Avni (1974). How- ever, the most successful explanation was due to Kippen- hahn & Thomas (1979). They estimated the energy trans- portedfromtheX-rayilluminatedpartofthestellarsurface REFERENCES totheshadowed side, and demonstrated that theminimum BahcallJ.N.,JossB.C.,AvniY.,1974,ApJ,191,211 at phase 0.0 could be reasonably well accounted for (X-ray BailynC.D.etal.,1995a,Nat,374,701 heating without horizontal transport leads to a flat mini- BailynC.D.,OroszJ.A.,McClintockJ.E.,RemillardR.A.,1995b, mum at phase 0.0). Nat,378,157 In addition, Schandl et al. (1997) found circulation to BrettJ.M.,SmithR.C.,1993,MNRAS,264,641 be necessary in order to accurately model the optical light ChenW.,ShraderC.R.,LivioM.,1997,ApJ,491,312 curveof CAL 87, an eclipsing supersoft X-raysource. They DaveyS.C.,SmithR.C.,1992,MNRAS,257,476 calculated the light curve based on the assumption that an DaveyS.C.,SmithR.C.,1996,MNRAS,280,481 (cid:13)c 0000RAS,MNRAS000,000–000 Irradiation of the secondary star in X-ray Nova Scorpii 1994 (=GRO J1655–40) 5 HameuryJ.-M.,LasotaJ.-P.,McClintockJ.E.,NarayanR.,1997, 489,234 HarmonB.A.,etal.,1995, IAUCirc.6205 HorneK.,1986,PASP,98,609 KippenhahnR.,ThomasH.C.,1979, A&A,75,281 MartinT.J.,DaveyS.C.,1995,MNRAS,275,31 OroszJ.A.,BailynC.D.,1997, ApJ,477,876 Osaki Y., Hirose M., Ichikawa S., 1993, in Accretion Disks in CompactStellarSystems,edJ.C.,Wheeler(Singapore,World Scientific, 272 Paczynski B.,1977, ApJ,216,822 PhillipsS.N.,ShahbazT.,PodsiadlowskiPh.,1999,MNRAS,304, 839 PhillipsS.N.,PodsiadlowskiPh.,1999,MNRAS,inprep. PhillipsS.N.,1999,D.Philthesis,UniversityofOxford. PringleJ.E.,1973,Nature,243,90 RobinsonC.,etal.,1997,ApJ,inprep. RuttenR.,DhillonV.S.,1994,A&A,288,773 SchandlS.,Meyer-HofmeisterE.,MeyerF.,1997,A&A,318,73, Shahbaz T., van der Hooft F., Casares J., Charles P.A., van ParadijsJ.,1999,MNRAS,306,89 Soria R., Wickramasinghe D.T., Hunstead R.W., Wu K., 1998, ApJ,495,L95 Stover, R.L., Robinson, E.L., Nather, R.E., Montemayer, T.J., 1980,ApJ,240,597 Strittmatter P.A., Scott J., Whelan J., Wickramasinghe D.T., WoolfN.J.,1973, A&A,25,275 TanakaY.,ShibazakiN.,1996,ARA&A,23,607 Tonry,J.,Davis,M.,1979,AJ,84,1511 vandenHeuvelE.P.J.,BhattacharyaD.,NomotoK.,Rappaport S.,1992,A&A,262,97 vanderHooftF.etal.,1998,A&A,329,538 vonZeipelH.,1924,MNRAS,84,665 WarnerB.,1995,inCataclysmicVariablesStars,CambridgeUni- versityPress,Cambridge,57 WhitehurstR.,1988, MNRAS,232,35 WilsonC.A.,HarmonB.A.,ZhangS.N.,PaciesasW.S.,Fishman G.J.,1995,IAUCirc.6152. Zhang S.N., Wilson C.A., Harmon B.A., Fishman G.J., Wilson P.B., Paciesas W.S., Scott M., Rubin B.C., 1994, IAU Circ. 6046 Zhang S.N., Harmon B.A., Paciesas W.S., Fishman G.J., 1995, IAUCirc.6209 ZhangS.N.etal.,1997, ApJ,479,381 (cid:13)c 0000RAS,MNRAS000,000–000 6 T. Shahbaz, P. Groot, S.N. Phillips, J. Casares, P.A. Charles, J. van Paradijs Figure 1.Top:Doppleraveragespectrum ofNovaSco1994intherestframeofthesecondarystarandanF6star(HR2906). Bottom: Nightly averaged spectra of Nova Sco 1994 (first night at the top, last night at the bottom). The spectra have been normalized and shiftedverticallyforclarity.ISindicatestheinterstellar6613˚A line. (cid:13)c 0000RAS,MNRAS000,000–000 Irradiation of the secondary star in X-ray Nova Scorpii 1994 (=GRO J1655–40) 7 Figure 2. The radial velocity curve of the secondary star in Nova Sco 1994. The dashed curve is a sinusoidal fit to the data points markedwithastar.ThedatahavebeenfoldedontheorbitalephemerisgivenbyvanderHooftetal.,(1998) and1.5orbitalcyclesare shown. (cid:13)c 0000RAS,MNRAS000,000–000 8 T. Shahbaz, P. Groot, S.N. Phillips, J. Casares, P.A. Charles, J. van Paradijs Figure3. TheeffectsofdifferentamountsofX-rayheatingonthesecondarystar’sradialvelocitysemi-amplitude.Weshowthefractional change inKobs as afunctionof Lx.Kobs iscomputed byfitting thepredicted curvewithaneccentric orbit. Thethreeobserved radial velocitymeasurements arealsoshown.Thetwolowercurvesshowtheeffects ofX-rayheatingbasedpurelyongeometryi.e.onlythose elements of area on the secondary star that aredirectly seen by the X-raysource areirradiated. The top curve (dashed line) show the effects ofindirectX-rayheating, calculated byextending theradiationhorizonasseenbytheX-raysourcebyafurther24degrees(see Figure4).TheeffectofthisadditionalheatingistoproduceavalueforKobs whichis30percentlargerthanexpectedpurelybasedon geometry. (cid:13)c 0000RAS,MNRAS000,000–000 Irradiation of the secondary star in X-ray Nova Scorpii 1994 (=GRO J1655–40) 9 Figure4.Theirradiatedsecondarystar’sRochelobeinthe(x−z)plane.Thecompactobjectisatcoordinates(1,0).Wehaveassumed Lx =8.5×1037 ergs−1 and the extreme geometrical case witha massratio of q=0.44, an inclinationangle of i=71◦ and adiscangle of2◦.Thedenseshadedregionshowsthearea(35percent)thatisirradiateddirectlybyX-raysproducedatthecompactobject;these regions do not contribute to the observed absorption line flux. The less dense region shows the area (60 per cent) which must also be heatedindirectlyinordertoproducethelargeobservedradialvelocitysemi-amplitude. (cid:13)c 0000RAS,MNRAS000,000–000