LONG-ORBITAL-PERIOD PREPOLARS CONTAINING EARLY K-TYPE DONOR STARS. BOTTLENECK ACCRETION MECHANISM IN ACTION 6 1 G. Tovmassian1, D. Gonz´alez–Buitrago1, S. Zharikov1 0 2 Instituto de Astronom´ıa, Universidad Nacional Aut´onoma de M´exico, Apartado Postal 877, Ensenada, Baja California, 22800 M´exico b e gag, dgonzalez, [email protected] F 4 1 D. E. Reichart6, J. B. Haislip6, K. M. Ivarsen6, A. P. LaCluyze6, J. P. Moore6 ] Department of Physics and Astronomy, University of North Carolina at Chapel Hill, Campus Box 3255, R Chapel Hill, NC 27599, USA S and . h p - o A. S. Miroshnichenko7 r t Department of Physics and Astronomy, University of North Carolina at Greensboro, Greensboro, NC s 27402-6170, USA a [ 2 v 0 ABSTRACT 4 2 6 We studied two objects identified as a cataclysmic variables (CVs) with periods exceeding 0 thenaturalboundaryforRoche-lobe-fillingzero-agemainsequence(ZAMS)secondarystars. We . present observational results for V1082Sgr with a 20.82 hr orbital period, an object that shows 1 0 a low luminosity state when its flux is totally dominated by a chromospherically active K star 6 with no signs of ongoing accretion. Frequent accretion shutoffs, together with characteristics of 1 emission lines in a high state, indicate that this binary system is probably detached, and the : v accretionofmatteronthemagneticwhite dwarftakesplacethroughstellarwindfromtheactive i donor star via coupled magnetic fields. Its observational characteristics are surprisingly similar X to V479And, a 14.5 hr binary system. They both have early K-type stars as a donor stars. r a We argue that, similar to the shorter-period prepolars containing M dwarfs, these are detached binarieswithstrongmagneticcomponents. Theirmagneticfieldsarecoupled,allowingenhanced stellarwindfromtheKstartobecapturedandchanneledthroughthebottleneckconnectingthe two stars onto the white dwarf’s magnetic pole, mimicking a magnetic CV. Hence, they become interactive binaries before they reach contact. This will help to explain an unexpected lack of systems possessing white dwarfs with strong magnetic fields among detached white+red dwarf systems. Subject headings: (stars:) binaries, cataclysmic variables – stars: individual (V1082Sgr; V479And) 1. Introduction corresponding Roche lobe and losing matter to a white dwarfcompanion (Warner 1995). CVs with Cataclysmic variables (CVs) are semidetached periodsover10hrarerare,whichisanaturalcon- binary systems consisting of a red star filling its 1 sequence of the requirement for a (nearly) main 2. Observations sequence donor star to match its Roche lobe size 2.1. Spectroscopy. (Ritter 2012). CVs with longer periods should contain late-type stars that have departed the Time-resolved spectroscopy of V1082Sgr was zero-age main sequence (ZAMS) in order to com- performedwiththe2.1mtelescopeoftheObserva- ply with the latter condition. torioAstron´omicoNacional1 atSanPedroMa´rtir, Relevant to this study is a small group of Baja California, M´exico (OAN SPM) in 2012 and objects that were thought to be low-accretion- 2013 with the Boller & Chivens spectrograph,us- rate polars, that is, magnetic CVs with an ac- ing a 600 and 1200 grooves mm−1 grating with a cretion rate of ∼ 10−13M⊙yr−1 (Schwope et al. 15µm2048×2048pixelMarconi2CCD,withspec- 2002). However, recently it has been argued tral resolutions of 4.1˚A and 1.8˚A, respectively. that these are in fact a detached pair of white The standard long-slit reduction of the data with and red dwarfstars (Webbink & Wickramasinghe a variance weighting extraction was made using 2005;Schmidt et al.2005). Amodelwasproposed IRAF2 proceduresafterapplyingbiassubtraction. according to which the magnetic white dwarf ac- Only cleaning cosmic rays, which are abundant cretes matter captured from the wind of a mag- on 1200s exposures,were made with the external neticallyactiveMdwarf. Hencetheyweredubbed task lacos (van Dokkum 2001). prepolars containing M-dwarf companions, which The wavelength calibration was made with the failed to achieve Roche lobe contactin their post- help of an arc lamp taken every 10th exposure. common-envelope evolution. The theoretical ba- The Boller & Chivens spectrograph, made origi- sis for a model in which the accretion is fueled nally for photographic plates, was later adopted by a stellar wind from the M dwarf and collected for the CCD camera, a much heavier detector. throughinterlockedmagneticfieldsofbinarycom- Thatintroducesstrongflexesandsignificantshifts ponents was proposed earlier by Li et al. (1994, in the wavelength’s zero point, which we usually 1995). correct by using a strong sky line in each spec- Here we present a new study of V1082Sgr, trum. At moderate zenith heights, those shifts an object with an extremely long period and can reach1−1.5˚A, and the transitionfrom expo- identified as a CV. It shows a number of out- sure to exposure is smooth, so a reasonable cor- standing observational features in a wide range rection is attainable. But because V1082Sgr is a of wavelengths. V1082Sgr was discovered by southernobject,itremainsquitelowinSPM(alti- Cieslinski et al. (1998). Thorstensen et al. (2010) tude <35◦), evenwhenthe objectpassesthrough foundabsorptionlinesfromaK-typestarinitsop- the meridian. As a result, we had severe prob- ticalspectrum andusingthem determined the or- lems with wavelength calibration, and a fraction bital period of 0.868 days. They tentatively iden- of our spectra were worthless for radial velocity tified V1082Sgr as nova-like. Bernardini et al. (RV) studies. (2013)conductedanextensiveX-rayobservational The spectra of the object were flux calibrated study, revealing that V1082Sgr is a highly vari- using spectrophotometric standardstars observed able source, with variations on a wide range of during the same night. The instrument cannot timescales, from hours to months. They found automatically rotate the slit to the corresponding that the X-ray spectrum is similar to a mag- parallactic angle, and for simplicity we routinely neticCV.Theobjectsharesmanypropertieswith usedanE–Wslitorientation. Inamajorityofob- V479And (Gonz´alez-Buitrago et al. 2013). Tak- servationsthe slit width waskeptnarrow(180µm ing into account that there are only a few objects = 2 arcsec) for better resolution. These two fac- identified as CVs in that orbital period range, we tors make a correct flux calibration difficult. But dwell on their similarities to understand the un- derlying reasons. We discuss both objects in this 1http: www.astrossp.unam.mx paper and develop a qualitative model to explain 2IRAF is distributed by the National Optical Astronomy them. Observatory, whichisoperated bytheAssociationof Uni- versities for Research in Astronomy (AURA) under coop- erativeagreement withtheNationalScienceFoundation. 2 in 2014’s observational run we acquired a couple of low and another few of higher (300 and 1200 grooves mm−1) resolution spectra, FWHM=8.5 and 2.2 ˚A, with a wide 350µm slit to circumvent this flux calibration problem. A number of K- typespectralstandardsfromCenarro et al.(2007) wereobservedalongwiththeobjectwiththesame instrumental settings. Two selected K2IV stars, BSNS104 and HD197964, are presented here as spectral identification templates. We used the xcsao procedure in IRAF to cross-correlate the observed spectra with a standard K star in the λ5150−5850˚ArangeinordertomeasuretheRVs of the absorption lines. A variety of standards from K2 to K4 were used, with no significant dif- ferences in the obtained velocities. The emission line parameters reported in this paperweremeasuredbyfittingasingleLorentzian becausetheline profilesarebestdescribedbythis function. We tried single- and double-Gaussian methods (Schneider & Young 1980) to determine theRVs. Thelatterisdesignedtoestimatetheor- Fig. 1.— The V-band light curve of V1082Sgr bitalmotionoftheaccretingstarbymeasuringthe over half a year. The individual measurements velocity of the inner parts of the disk. Regardless are plotted as tiny dots, and filled pentagons cor- ofthemethod,theresultingRVvaluesaresimilar. respond to a nightly average magnitude. The variability may have a periodic component at We make extensive use of spectra kindly made around 29 days, which is overplotted as a sine available to us by John Thorstensen (referred to curve. here as JT spectra; Thorstensen et al. 2010). 2.2. Photometry eight days (2014 February-March, see also Ta- Time-resolved V-band photometry was ob- ble 1). We used two of the three instruments tained using the 0.41 m Ritchey–Chr´etien tele- on board of the Swift gamma-ray burst explorer (see Gehrels et al. 2004): the X-ray Telescope scopeatthe PanchromaticRoboticOpticalMoni- toring andPolarimetryTelescopes(PROMPT)at (XRT; e.g. Burrows et al. 2005) and the Ultravio- CerroTololoInter-AmericanObservatory(CTIO) let/Optical Telescope (UVOT; e.g. Roming et al. 2005). inChile,withtheapogeecamerathatmakeuseof E2V CCDs. The log of photometric observations The bservations with the XRT were made pre- is also given in Table 1. The data were reduced dominantly in photon-counting mode (PC). A withIRAF,andtheimageswerecorrectedforbias light curve and spectrum were extracted within a and flat fields before aperture photometry. Flux 25-pixel-radiuscirclecenteredonthemaximumof calibration was performed using secondary stan- theemissionfromV1082Sgr,whichincludesmore dard stars from the same field. than 95% of the source flux. The background was taken from an annulus of 30-pixel inner ra- 2.3. UV and X-ray observations dius and 60-pixel outer radius, also centered on the source position. This wasdone using the soft- V1082Sgr was observed with the Swift tele- ware XSELECT version 2.4. Observations with scope as a target of opportunity (ToO tar- the UVOT were made using the UVW1, UVM2, get ID: 31252) with a total on-source exposure and U filters, centered at 2600,2246,and 3465 ˚A, time of 26.37 ks, divided into four observations respectively. Thelightcurvewasextractedfroma of approximately 6500 s, each performed for 10-pixel-radiusregionusingtheUVOTMAGHIST 3 Fig. 2.— The spectra of V1082Sgr in a high, intermediate, and low states. In the high state it appears as a typical CV. In the low state, emission lines vanish altogether, but there is weak H emission. α tool version 1.12. Observations of the object performed by Swift Table 1: Log of photometric observations. in2008and2012andreportedbyBernardini et al. (2013) are also included in the analysis for com- Date Exp. Filter Telescope Total pleteness. s Instrument (days) 3. High and Low states 19/08/08 327 UVW2 SWIFT 0.3 20/08/08 825 UVM2 SWIFT 3.7 V1082Sgr exhibits strong variability on differ- 28/08/08 514 UVM2 SWIFT 0.5 ent timescales. We observed the object regularly 22/02/14 286 UVM2 SWIFT 0.7 for five months. The photometric variability ap- 01/03/14 1160 UVW2 SWIFT 1.7 pears to be irregular and does not show any pe- 08/03/14 1214 UUUU SWIFT 1.7 riodic signal at the frequency corresponding to 15/03/14 744 UVW1 SWIFT 1.8 the orbital period reported by Thorstensen et al. 22/02/14 1636 UVM2 SWIFT 0.3 (2010). However, on a much longer timescale, 01/03/14 1636 UVW2 SWIFT 0.3 there might be some cyclical activity. In Figure1 08/03/14 1636 U SWIFT 0.3 we present the observed light curve of the object 15/03/14 1636 UVW1 SWIFT 0.3 in the V band. Individual measurementsareplot- 06/13 180 V APOGEE 19 ted as tiny dots, and the nightly average magni- 07/13 180 V APOGEE 18 tudesaredenotedbyfilledpentagons. Thesystem 08/13 180 V APOGEE 15 varieswithamplitudeasmuchas0.4magthrough 09/13 180 V APOGEE 14 the course of each night. The full amplitude of 10/13 180 V APOGEE 08 variability reaches 1.5mag, changing from 13.6 at 11/11/13 180 V APOGEE 01 the maximum to 14.8 mag at the minimum. The maximumtominimumcyclesoccurapproximately 4 every 29 days. We tentatively modeled the light UVandwasbarelyvisibleintheX-ray. Thecount curvewithasinusoid. Onlysixcycleswerecovered rate was ≤0.002ctss−1. during our observations. Since we know that in the low state the donor The spectral properties of V1082Sgr change star dominates the flux in the opticaldomain and strongly with the luminosity. Two distinctive there is little evidence of ongoing accretion, we states can be singled out. In the high state it might expect that the X-ray flux is formed by has intense emission lines, most notably of the the K-type star. Bernardini et al. (2013) esti- Balmerseriesaccompaniedbyneutralandionized mated that V1082Sgr in the low state has a flux helium. The Balmer decrement is rather steep of6.1×10−14ergcm−2s−1 correspondingtolumi- (Hα/Hβ/Hγ/Hδ/Hǫ=1.3/1.0/0.85/0.65/0.54). nosity Lx = 7.3×1030ergs−1 for a distance of 1 Heiiiseverpresentinthespectrawhenitisinthe kpc. Thisis consistentwiththeupper limitofthe high state, and its intensity is comparable to that soft (0.1 - 2.0 keV) X-ray luminosities of RS CVn of H . In addition, a blend of fluorescent lines systems,whicharegenerallyfoundintherangeof β of N iii and C iii around λ4645˚A is prominently 1029−1032ergs−1 (Drake et al. 1992). present. Also clearly visible are absorption lines The average X-ray spectrum of the object in belonging to the secondary star. In Figure2 a broad 0.3–100keV range is 100 times brighter, example spectra of the object in the high, the reflecting the high state. The spectrum is con- intermediate, and the low states are presented. sistent with a small X-ray-emitting region hav- As the object becomes fainter, the lines also ingplasmatemperaturestypicalofamagnetically weaken. However, the low-state spectrum is confinedaccretionflow,likeinpolars. TheMwd = highly unusual for a CV because it contains 0.64±0.04M massofthewhitedwarfwasfetched ⊙ pure radiation from the secondary star. Spectra fromthefittothecompositeXMM-NewtonEPIC resembling an isolated K star with practically no andSwiftBATspectrum(Bernardini et al.2013). emission lines have been observed in this system on many occasions in different epochs. In total 3.1. Revisiting the X-ray light curve we have 10 occurrences of a low state in the Bernardini et al. (2013) also reported a promi- JT spectra and one of our own (as a criteria nent brightening , dubbed as a ”flare”. That 5–6 being considered, the equivalent width (EW) hourflare-likebrighteningintheX-raylightcurve of H being less than 4.0˚A). We witnessed the α emergence of emission lines of H and Heii from canbe interpretedas partofanirregularvariabil- β ity,butitalsocanbetheresultofamagneticpole one night to another, with no significant change transitingthelineofsight. Theentirespanoftime in the continuum levels. during whichSuzakutook exposuresofV1082Sgr Bernardini et al. (2013) report that V1082Sgr wasslightly longerthan30hr, longerthan the or- is alsoa highlyvariablesourcealsointhe UVand bitalperiodoftheobject,butnotenoughtocover X-rays. The UV measurements are presented in two cycles. If we assume it was not a flare but or- Figure3. The data were taken in four different bitalmodulation,thenwecanfolditwiththePorb filters,unfortunately. Thatdoes nothelp ininter- and compare it to V479And, a very similar ob- pretingtheUVlightcurveorindeterminingtimes ject contemplated in this paper. The X-ray light whentheobjectwasinalowstate,i.e. whenthere curves of both objects are presented in Figure4, was no active accretion going on. However, there each folded with their corresponding orbital pe- are recurring observations obtained in the same riod and ephemeris. The interesting thing about UVM2 and UVW2 Swift UVOT filters. A com- this plot is that V479And was observed for two parison among them and an analysis of the spec- continuous orbital cycles, and we know for sure tral energy distribution (SED) (presented below that the brightening there is due to the orbital in Section3.4) show that the object probably was modulation. In both cases,the X-ray-emittingre- in a truly low state only during the 2012 obser- gionislocatedonthesideofthewhitedwarffacing vationscorrespondingtoHJD2456099-6100. Dur- the donor star and hence comes into the sight of ing that period of time, two exposures (3400 and view right after the phase zero. The duration of 6000s long) were obtained over two consecutive brightening is similar, and so is the relative flux, days, and the source was persistently faint in the althoughV479Andismuchfainter. Belowwedis- 5 W2 M2 W2M 2 M2 W2 U W1 Fig. 3.— The Swift UVOT magnitudes of the V1082Sgr. The object shows strong variability. The UVOT filters used in the observations were set arbitrarily according to the configuration of the instrument at the moment of observation. The filters are marked on top of the panels and the data are plotted in different shades of blue to help tell them apart. cuss in more detail the similarities of these two certaintyofthe estimate. Therearemany spectra objects and their common identity. ofZAMSKstarsavailableincatalogs,andonecan find a satisfactory continuum flux fit among K2– 3.2. Absorption lines K4V stars, but they fail to match all absorption features. Particularly,thetroughcorrespondingto V1082Sgr experiences low states and, as was the MgH band blueward from the Mgb triplet is mentioned earlier,the spectrum of the object fea- poorlyfit,andtheMgHfeaturearoundλ4770˚Ais tures only the late-type companion of the binary notcompatiblewiththemain-sequenceluminosity during these episodes. We only hadone chanceto class3. We think the K2IV spectrum represents a observeV1082Sgrspectroscopicallyina lowstate better match to the observed spectra. (JD=2456893). We used a low-resolution setting Two selected K2IV stars (BSNS104 and and a wide slit, hence we covered a wide wave- lengthrangewithreliablefluxcalibration. Onthe HD197964; Cenarro et al. 2007) scaled to the SPM spectrum are overplotted. They were ob- next night, the object already featured emission lines, so it probably was on the path of brighten- served on the same night with the same instru- ing. The low-state spectrum is presented in Fig- mental settings. BSNS104 represents a perfect match, while HD197964 deviates at longer wave- ures2and5. JThasobtained14low-statespectra taken at different epochs. It is not excluded that lengths. However, HD197964 tallies better with the objectthanBSNS104inthe higher-resolution the flux calibration of their spectra is not very spectra regarding the depth of absorption fea- precise (it is certainly off at the blue end of the spectrum) and the flux is underestimated. Our tures. The higher-resolution spectra obtained on the next nightarepresentedinthe rightside pan- spectrumandthe averageofJT spectra,scaledto thelevelofSPM,aredisplayedintheleftpanelsof els. Emission lines have already reappeared in Figure5, where large portions of spectra are plot- V1082Sgr,butthecontinuumandabsorptionlines ted inthree verticalpanels coveringthe entire ob- served range. Thorstensen et al. (2010) suggested 3AgoodfitisachievedwithK5starHD283916=SAO76803 (Jacoby etal. 1984), but its spectral classification is dis- aK4spectraltypeclassificationforthedonorstar putedbyMalyutoetal.(1997),wholistsK2III.Hencewe of V1082Sgr, but they also mentioned a large un- prefernottorelyonthisstandard. 6 | | Fig. 5.— Details of absorption features of V1082Sgr as compared to comparison stars of known spectral type and luminosity class. The black and gray lines are the spectra of the object from SPM and JT; the latter is scaled in flux to overlap with the former. The spectrum of BSNS104 is represented by an orange line. Thedark-bluelineisthespectrumofHD197964,anotherK2IVstarclassifiedassuchbyCenarro et al. (2007). They both were observed from SPM at the same night and the same instrumental settings as the object. The depth of lines in the midlle part of the optical range is better matched by HD197964, but it deviates at the red end of the spectrum (this might be a flux calibration problem). were not affected yet. The spectra in these pan- ≈7−9˚A;FWZI≈40−50˚AofH )(seeFigure2). β els are normalized. There is a close resemblance In the high state, the EWs of H are in a −30 to α of the spectra of the object obtained at SPM, by −40 ˚A range. In Figure6 the profiles of H and α JT and HD197964. Of course, spectral classifica- Heii are presented. Also included in the plot is a tionofthedonorstarshouldbedonewithcaution, prominentblendofNiiiandCiiiaroundλ4645˚A. since it may have large spots and, depending on It is formed by a continuum fluorescence, as ar- which part of the star surface was observed, the gued by Williams & Ferguson (1983) and is evi- spectral class can vary by two digits. denceofastrongUVcontinuum. Thequickresur- TheorbitalperiodofV1082Sgrwasdetermined gence of intense emission lines with little change byJTusing acomplexofabsorptionfeatures. We of continuum and the Balmer decrement asserts repeated the analysis by adding some reliable RV that the emission lines are formed in optically measurements obtained by us (adding a longer thin, sparse gas. The deficiency of the G-band timebase)andfoundnodeviationfromtheperiod andNiii/Ciiiprofileisprobablyevidenceofape- determinedbyJT.Wepresenttheabsorptionlines culiar chemical composition of the donor star. RV curve in the bottom panel of Figure7 only for The emission line profiles are single peaked comparison with the corresponding Heii and H and symmetric. It would be almost impossible β RV curves. to determine the orbital period by them. How- ever,whenfoldedwithknownorbitalperiods,they 3.3. Emission lines start to make sense. A periodic pattern is clearly present,anditisinastrictcounter-phasefromthe The emission lines of V1082Sgr in a high state absorptionlines,asdemonstratedinFigure7. The are very intense. Their intensity is strongly vari- measurements of the RVs of H and Heii are no- able, but they remain relatively narrow (FWHM β 7 Fig. 4.— X-ray light curves of V1082Sgr and Fig. 6.— Example of emission line profiles in a V479And folded with their corresponding or- high and low states. In high state the lines are bital periods and ephemeris fetched from the mostly symmetric, single peaked, relatively nar- absorption-line RVs. The coverage of V1082Sgr row with wings somewhat wider than a Gaussian is not sufficient to prove that the brightness peak profile. TheyarebestdescribedbyasingleLoren- is repeating from orbit to orbit, but of V479And zian. In the low state H is narrow and is not α it is. accompanied by other Balmer or helium lines. tably dispersedrelativeto the best-fitsine curves. line. It has FWHM ≈ 6˚A in a 2˚Apixel−1 reso- The reason is not the accuracy of the measure- lution spectra. The EW in the low state reaches ments, but the intrinsic, chaotic velocities of the −2.0˚A, a value very common for chromospheri- gas superposed on orbital motion. Nevertheless, callyactiveKstars(Houdebine2012). TheRVsof bothsinecurvesshowastronganticorrelationwith H line with EW>−4.0˚A tend to follow the mo- α theabsorptionlinesformedatthedonorstar,with tion corresponding to the secondary. This is not thesemiamplitude ofHβ being30%lessthanthat surprising, assuming that it has a chromospheric of Heii. Meanwhile, Hα does not show any pe- origin. The RVs of the Hα line in low-state spec- riodic variability, just an erratic spread of values. tra are marked by filled pentagons in the bottom TheparametersofthesinefittoHβ,Heiiandthe panel of Figure7. In Figure8 the residual spectra absorption lines are presented in Table2. after subtraction of a K2IV from the low-state Inthe low state we only getto measurethe H spectrum is presented. The red dashed line cor- α responds to a zero flux. Caii H&K lines become visible in the residual spectrum along with a nar- Table 2: Radial velocity fit parameters. rowHα andareclearlyofachromosphericnature. Line γ RV Phase shift rms The presence of a faint Hβ line is evidence of en- ID (km s−1) (km s−1) HJD0 (km s−1) hanced activity. Abs. 41.8±3.5 54±4.2 0.0† 19.4 H 24.8±3.5 21.7±5.5 0.50 15.5 3.4. Spectral energy distribution β Heii 32.2±4.5 33.2±7.5 0.52 24.9 The Roche lobe size of a K2 star at a 20hr or- † Fixed; Porb given by the JT ephemeris bital period is 1.6 <R/R < 1.8 for a range of ⊙ 8 Fig. 8.— The residual spectrum between the object and K2,IV star BSNS104 scaled to the continuum at 5500˚Aisshown. Visible spectralfeaturesonthe residualspectrumareH ,atinyH andCaiiandprobably α β the G band. With a grey line a similar residual spectrum is plotted of the JT spectrum, which has higher spectral resolution but suffers from flux calibration on the blue end of the spectrum. We included it in the plot to demonstrate the narrowness of H , leaving no doubts about its chromospheric nature. α white dwarf masses from 0.5 to 1.2M . Mean- i.e. a period of time when the contribution from ⊙ while, a K2V star has a radius 0.83R , and if theaccretion-fueledprocessesisnegligible. During ⊙ the donor star is still on ZAMS, it would hardly such periods, we observe a pure spectrum of the fill even half of its corresponding Roche lobe. It late star in the optical domain. It would be natu- will take 20-40 Gyr for a K star to evolve to the ral to assume that the UV flux at such moments size of the Roche lobe. The shortest distance to is dominated by the white dwarf. The tempera- V1082Sgrwouldbe550pcifthedonorisaZAMS ture of the white dwarf could not be constrained star and 1400pc if it is close to filling its Roche from the available data. However, we can make a lobe, assuming that the average visual magnitude rough estimate from the energy balance. A white (mV = 14.7) at the low state emanates entirely dwarf should have a radius of Rwd ≈ 0.01Rcool from the donor. if the cool star is close to the main sequence, In Figure9 the SED of the object (black lines and two times smaller if it is about Roche lobe and symbols) is presented from the IR to UV. size. Accordingly, it must have a temperature The spectra, as well as photometric measure- ranging from Teff ∼ 17000 to 30000K to pro- ments, are correctedfor the interstellar reddening videsufficientluminosityandbeobservableatthe by assuming E(B − V) = 0.15 and a standard low state. Presented in Figure9 are blackbod- R = 3.1 (Schlegel et al. 1998) and are recalcu- ies of 17 and 30 kK effective temperature (long- V lated for other wavelengths using the Fitzpatrick dashed lines), describing a possible contribution (1999) parametrization. Worth noting is that the from the white dwarf. The red short-dashed line hydrogencolumndensityNH =9×1020cm−2used is a black-body of 4800K representing the cool in the X-ray spectral analysis corresponds to the star. A realistic combination of values for the same extinction. whitedwarfradiusandthetemperaturestemming from L ∼ T4 R2 (D/1kpc)2 relation, where D is It is not clear in which luminosity state of the eff wd distance to the object, confirmscorrectnessofour objectsomephotometricdatawereobtained. The assumptionthatinthelowstateweobservepurely photometric UV data are apparently available in both states. Particularly, Swift UVOT UVM2 stellar components. band measurements are stretched over a wide en- The light-green, short-dashed line roughly cor- ergyrange;partofthemaroundHJD2456717(see responds to the flux difference between the high Figure3) are certainly takenduring the low state, 9 Fig. 7.— Radialvelocitymeasurementsandsinu- Fig. 9.— Spectral energy distribution of soidalfitsofdifferentspectrallines. Inthebottom V1082Sgr in low and high states presented by panel, the RVs of a complex of absorption lines dark lines and symbols. All available photomet- are plotted with open squares. Filled pentagons ric measurements in IR (2MASS) and UV (Swift) correspond to the RVs of the H chromospheric regardlessoftheluminositystateoftheobjectare α emission line measured in the low-state spectra. plotted with the star symbols. The yellow-red In the upper panel, the measurements of emission solid line is a spectrum of K3IV (Pickles 1998). lines in higher states are presented. The RVs cor- Thedashedlinesarecalculatedblackbodiesscaled responding to H are plotted with open symbols to pass through observed points. The red short- β and a dashed line, and Heii by filled symbols and dashed line corresponds to a 4800K, the light- a continuous line. The data are folded with the blue long-dashed line to 17000K, and the violet orbital period and repeated twice for illustrative to 30000K temperature bb. The gray line is a purposes. sum of K3IV and 17000K bb. The light-green short-dashed line is a rough differential flux be- tween high and low states. and low states. Dash-dotted lines are power-law4 F ∼ λ−α with α = 1 and 2.33. Most CVs ra- λ diate in the UV in the form of a power law with tialevidence,but itis acomplementaryargument indexes found within that range (e.g. de Martino to the ones made in Section3.3 that an accretion 1999). It is possible to find a family of observed diskorastream,commoninCVs,isnotthesource points laying on a single power law, except for of the excess radiation. the U-band data, which might be elevated by the An alternative can be the cyclotron radiation Balmer jump. Nevertheless, some measurements since in the case of low-rate magnetic accretion it are clearly off from being part of a straight line. provides the dominant cooling mechanism. Such The additional emission appearing during high radiation is clearly observed in prepolars contain- states in the UV does not fit easily into a black ing Mdwarfsas adonorstar. Ofcourse,inpolars body or a power-law model. This is not substan- and prepolars with shorter periods (P .8hr), orb thecyclotronradiationisalsoobservedintheopti- 4None of the blackbody or power curves are fitted to the cal domain. However,we could not find any cred- observeddata(thequantityandqualitydonotallowoneto ible signs of cyclotron lines (wide humps in the performanymeaningfulfit)andarepresentedtoillustrate thepossibleconfigurationofthebinary. continuum) in the optical spectra of V1082Sgr in 10