Optical and Supersoft X-ray Light Curve Models of Classical Nova V2491 Cygni: A New Clue to the Secondary Maximum
aa r X i v : . [ a s t r o - ph . S R ] F e b TO APPEAR IN THE A STROPHYSICAL J OURNAL , L
ETTERS
Preprint typeset using L A TEX style emulateapj v. 08/22/09
OPTICAL AND SUPERSOFT X-RAY LIGHT CURVE MODELS OF CLASSICAL NOVA V2491 CYGNI: A NEW CLUETO THE SECONDARY MAXIMUM I ZUMI H ACHISU
Department of Earth Science and Astronomy, College of Arts and Sciences, University of Tokyo, Komaba, Meguro-ku, Tokyo 153-8902, Japan
AND M ARIKO K ATO
Department of Astronomy, Keio University, Hiyoshi, Kouhoku-ku, Yokohama 223-8521, Japan to appear in the Astrophysical Journal, Letters
ABSTRACTV2491 Cygni (Nova Cygni 2008 No.2) was detected as a transient supersoft X-ray source with the
Swift
XRTas early as 40 days after the outburst, suggesting a very massive white dwarf (WD) close to the Chandrasekharlimit. We present a unified model of near infrared, optical, and X-ray light curves for V2491 Cyg, and haveestimated, from our best-fit model, the WD mass to be 1 . ± . M ⊙ with an assumed chemical compositionof the envelope, X = 0 . Y = 0 . X CNO = 0 . X Ne = 0 .
10, and Z = 0 .
02 by mass weight. We stronglyrecommend detailed composition analysis of the ejecta because some enrichment of the WD matter suggeststhat the WD mass does not increase like in RS Oph, which is a candidate of Type Ia supernova progenitors.V2491 Cyg shows a peculiar secondary maximum in the optical light curve as well as V1493 Aql and V2362Cyg. Introducing magnetic activity as an adding energy source to nuclear burning, we propose a physicalmechanism of the secondary maxima.
Subject headings: novae, cataclysmic variables — stars: individual (V1493 Aql, V2362 Cyg, V2491 Cyg) —stars: mass loss — X-rays: binaries INTRODUCTION
Classical novae show a wide variety of timescales andshapes in the optical light curves (e.g., Payne-Gaposchkin1957). Among various shapes of nova light curves, V1493Aql (Nova Aquilae 1999 No.1) shows an impressive sec-ondary maximum about 50 days after the outburst (e.g.,Bonifacio et al. 2000; Venturini et al. 2004), although thephysical mechanism of the secondary maximum is not under-stood yet. Recent two novae, V2362 Cyg (Nova Cygni 2006)and V2491 Cyg (Nova Cygni 2008 No.2), also show a similartype of single secondary maximum, at about 250 and 15 daysafter the outburst, respectively. These three novae form a widevariety set of timescales, i.e., about 15, 50, and 250 days at thesecondary maximum and of secondary peak heights, i.e., 1.1,2.8, and 3.6 mag, respectively, (see, e.g., Kimeswenger et al.2008), which provide us a new clue to the mechanism of thesecondary maxima.In this Letter, we propose a strong magnetic activity as themechanism of the secondary maxima observed in V2491 Cyg,V1493 Aql, and V2362 Cyg, using the white dwarf (WD) pa-rameters obtained from light curve fittings based on an op-tically thick wind model of nova outbursts (Kato & Hachisu1994). In §2, we briefly describe our numerical method andlight curve fitting of V2491 Cyg. In §3, we propose an idea ofstrong magnetic activity in the WD envelope and estimate thetimescales of the secondary maxima. V1493 Aql and V2362Cyg show very different timescales of the secondary maxi-mum, both of which are also explained by the same mecha-nism in §4. Conclusions follow in §5. MODELING OF NOVA OUTBURSTS
Optically thick wind model
Electronic address: [email protected] address: [email protected]
After a thermonuclear runaway sets in on a mass-accretingWD, its photospheric radius expands greatly to R ph & R ⊙ and the WD envelope settles in a steady-state. We have fol-lowed evolutions of novae by connecting steady state solu-tions along the decreasing envelope mass sequence. We solvea set of equations, that is, the continuity, equation of motion,radiative diffusion, and conservation of energy, from the bot-tom of the hydrogen-rich envelope through the photosphereassuming spherical symmetry. Winds are accelerated deep in-side the photosphere so that they are called “optically thickwinds.” As one of the boundary conditions for our numeralcode, we assume that photons are emitted at the photosphereas a blackbody with the photospheric temperature of T ph . X-ray flux is estimated directly from the blackbody emission, butinfrared and optical fluxes are calculated from free-free emis-sion by using the physical values of our wind solutions. Weneglect the effect of ash helium layer, which may be piled upbeneath the hydrogen burning zone, for all nova calculationsexcept the RS Oph case (Hachisu et al. 2007). Our methodand various physical properties of these wind solutions havealready been published (e.g., Hachisu & Kato 2001a,b, 2004,2006, 2007; Hachisu et al. 1996, 1999a,b, 2000, 2003, 2007,2008; Kato 1983, 1997, 1999; Kato & Hachisu 1994).The light curves of our optically thick wind model are pa-rameterized by the WD mass ( M WD ), chemical compositionof the envelope ( X , Y , X CNO , X Ne , and Z ), and the envelopemass ( ∆ M env , ) at the outburst (day 0). Details of our lightcurve fittings are described in Hachisu & Kato (2006, 2007)and Hachisu et al. (2007, 2008). Light curve fitting (V2491 Cyg)
V2491 Cyg was discovered by Nishiyama and Kabashimaat mag 7.7 on 2008 April 10.728 UT (Nakano et al. 2008).The nova was not detected on April 8.831 UT (limiting mag14). The exact outburst day is unknown, so we assume here Hachisu & Kato F IG . 1.— Model light curves of V2491 Cyg for three white dwarf masses,1 . M ⊙ (green line), 1 . M ⊙ (thick blue line), and 1 . M ⊙ (thin blackline), together with the observations. See Hachisu & Kato (2006) for moredetails of light curve fitting. Two arrows indicate epochs when the windstops (day 40) and when the hydrogen shell-burning ends (day 50) in the1 . M ⊙ WD model.
Dashed lines : Supersoft X-rays are probably not de-tected during the wind phase because of self-absorption by wind itself (see,e.g. Hachisu & Kato 2003b). Large open circles at the right end of each op-tical light curve denote the epoch when the optically thick wind stops. Weobtain the best-fit model for the envelope chemical composition of X = 0 . Y = 0 . X CNO = 0 . X Ne = 0 .
10, and Z = 0 . Large open triangles : Ob-servational X-ray (0 . - . Swift (Page et al.2009). Optical and near IR observational data of I ( open diamonds ), R ( cir-cles with a plus ), V ( filled squares ), y ( filled circles ), and visual ( small opencircles ) are taken from AAVSO (American Association of Variable Star Ob-servers) and VSOLJ (Variable Star Observers League in Japan). The F λ ∝ t - law ( magenta ) is added for the nebular phase, i.e., after the wind stops, where t is the time after the outburst. that t OB = 2454566 . P orb = 0 . . - .
05 mag.The rise or decay time of supersoft X-ray flux is an im-portant indicator of the WD mass (e.g., Hachisu & Kato2006, 2007; Hachisu et al. 2008). The best fit model withthe
Swift observation (Page et al. 2009) is the WD mass of1 . ± . M ⊙ as shown in Figure 1. Here we adopt thechemical composition of X = 0 . Y = 0 . X CNO = 0 . X Ne = 0 .
10, and Z = 0 .
02 to reproduce a short (10 days)supersoft X-ray duration (see Hachisu & Kato 2006, 2007;Hachisu et al. 2007, 2008, for dependence on chemical com-position).Figure 1 also shows optical and near IR light curves of the I , R , V , y , and visual magnitudes. Our theoretical light curvesof free-free emission reasonably fit with the observation untilday ∼
50 except the secondary peak. After that, the observa-tional data deviate probably due to the contribution of nebularemission lines.Thus we conclude that the WD of V2491 Cyg is as mas-sive as 1 . M ⊙ from our light curve fittings with the supersoftX-ray, optical, and near IR data. This WD mass of V2491Cyg is a bit less massive but comparable to that of the re-current nova RS Oph (Hachisu et al. 2007), the supersoft X-ray duration of which lasted 60 days starting from day 30,much longer than the 10 days in V2491 Cyg. Hachisu et al.(2007) explained this long duration of supersoft X-ray phaseof RS Oph by incorporating heat flux from a hot helium layerunderneath the hydrogen burning zone, which develops inmass-increasing WDs. In V2491 Cyg, however, the supersoftX-ray phase lasts only 10 days, indicating that no thick he-lium layer develops beneath the hydrogen burning zone and,therefore, the WD mass is not increasing (see discussion inHachisu et al. 2007). Tomov et al. (2008) suggested a pos- F IG . 2.— Five energy densities of the WD envelope are plotted against theradius, i.e., thermal energy ( ε th : red ), thermal gas energy ( ε th , gas : magenta ),rotational kinetic energy ( ε rot : blue ), magnetic energy ( ε mag : black ), andwind kinetic energy ( ε wind : green ). Five sequential stages are plotted for ε rot but only two stages [the first ( thick solid ) and the last ( thin solid )] for the otherenergy densities. Open circles indicate the photosphere while open squarescorrespond to the critical point of each wind solution. Various parameters aresummarized in the figure. sibility that V2491 Cyg is a recurrent nova, for which we ex-pect that the WD mass is increasing (Hachisu & Kato 2001b).One of the strongest clues to this question is the chemicalcomposition of the ejecta. We strongly recommend compo-sition analysis of V2491 Cyg to clarify whether or not theejecta are contaminated by WD matter. Suzaku
X-ray spec-tra analyzed by Takei et al. (2009) suggested oxygen/neon-rich ejecta. If so, the WD mass is decreasing and, as a result,V2491 Cyg is not a candidate of Type Ia supernova progeni-tors. MAGNETIC ACTIVITY (V2491 CYG)
V2491 Cyg shows a single secondary maximum in the op-tical and near IR light curves, which we cannot reproduce byour evolution model. Here we propose magnetic activities asa new energy source of the secondary maximum.The prenova X-ray was detected in V2491 Cyg with
Swift (Ibarra & Kuulkers 2008). Ibarra et al. (2008) suggestedthat the prenova X-ray spectra are more like those seenfrom magnetic rather than non-magnetic cataclysmic vari-ables. Takei et al. (2009) reported the first detection of super-hard X-rays (15 -
60 keV) with the
Suzaku
HXD at day 10, thespectrum of which has a power law distribution, suggesting anon-thermal origin, and unlikely to arise from thermal emis-sion from internal shocks. This superhard component was notseen in their second
Suzaku observation at day 30. These twoobservations point toward strong magnetic activities on theWD surface.Therefore, our idea on the secondary maximum is based onadditional energy release associated with rotating magneticfield. We assume that V2491 Cyg is a polar system with mag-netic field as strong as B ∼ G on the WD surface (e.g.,Warner 1995). Before the nova outburst, the WD magneticfield rotates synchronously with the WD spin as well as thebinary orbital motion. After the outburst, the nova envelopeexpands largely and rotates differentially due to local angularmomentum conservation. Then, the magnetic field no morerotates synchronously with the WD spin because the magnetictension is not strong enough to keep the whole envelope rotat-ing with the WD spin. We expect that the differential rotationamplifies magnetic field which drives strong magnetic activi-ties. As the nova outburst proceeds, the density of the WD en-velope is gradually decreasing due mainly to wind mass lossight Curve Models of Novae 3
TABLE 1E
POCH OF S ECONDARY MAXIMUM IN OUR NOVA MODELS object max a P orb M a M WD ε rot ≈ ε mag (day) (day) ( M ⊙ ) ( R ⊙ ) ( M ⊙ ) (day)V2491 Cyg 15 0.0958 0.18 1.0 1.32 131.3 151.27 18V1493 Aql 50 0.156 0.34 1.4 1.2 401.15 491.1 62V2362 Cyg 250 0.207 0.48 1.6 0.75 2000.7 2400.65 330 a epoch of secondary maximum b calculated from equation (5) and then the magnetic field eventually recovers synchronousrotation. The strong magnetic activities end at this stage, cor-responding to the end of a secondary maximum. Since theadditional energy source disappears, the light curve goes backto a “normal” one as shown in Figure 1.To confirm this idea we estimate the total thermal energyof gas plus radiation ( ε th = ε th , gas + ε th , rad ), rotational kineticenergy ε rot , wind kinetic energy ε wind , and magnetic energy ε mag , which are calculated from ε th = ε th , gas + ε th , rad = 32 kT µ m H ρ + aT , (1) ε rot = 12 ρ (cid:0) r Ω spin (cid:1) , (2) ε wind = 12 ρ v , (3) ε mag = B π (cid:18) rR WD (cid:19) - . (4)The temperature T , density ρ , and wind velocity v wind aretaken from our wind solutions of the best-fit model of V2491Cyg (1 . M ⊙ WD). Here we assume the dipole magnetic fieldof B = 3 × G at the WD surface (e.g., Warner 1995, forV1500 Cyg) and that the WD spin period is the same as theorbital period, 2 π/ Ω spin = P spin = P orb . Figure 2 shows theenergy densities for five sequential stages during the novaoutburst. The first model ( thick solid ) has the largest pho-tospheric radius and corresponds to a stage at/near the op-tical maximum. We see that at/near the optical maximum, ε mag . ε rot at r & × cm. This indicates that the magneticfield is differentially rotating outside of r ∼ × cm. Then,the rotation energy density decreases with time and eventually ε mag becomes comparable to ε rot at the stage of the smallestphotospheric radius. After that, the magnetic field probablygains synchronous rotation with the WD spin. We expect thatthe magnetic activities have a peak at ε mag ≈ ε rot . This con-dition is satisfied at day 15 for our best-fit model, being veryconsistent with the time of the secondary maximum.Next we estimate the epoch when the companion emergesfrom the nova envelope. If the mass of the donor star is esti-mated from Warner’s (1995) empirical formula, i.e., M M ⊙ ≈ . (cid:18) P orb hours (cid:19) / , for 1 . < P orb hours < M = 0 . M ⊙ , which corresponds to the separa-tion of a = 1 . R ⊙ , and the effective Roche lobe radius of F IG . 3.— Same as Fig. 1, but for V1493 Aql (Nova Aql 1999 No.1). Ourbest-fit model is M WD = 1 . M ⊙ for the envelope chemical composition of X = 0 . Y = 0 . X CNO = 0 . X Ne = 0 .
03, and Z = 0 .
02. Observationaldata are taken from AAVSO and VSOLJ. We further add the data of IAUCirc. 7223, 7225, 7228, 7232, 7258, 7273, 7313.F IG . 4.— Same as Fig. 1, but for V2362 Cyg (Nova Cyg 2006). Ourbest-fit model is M WD = 0 . M ⊙ for the envelope chemical composition of X = 0 . Y = 0 . X CNO = 0 .
35, and Z = 0 .
02. Observational data are takenfrom AAVSO and VSOLJ. the primary component (WD) of R ∗ = 0 . R ⊙ . When thephotospheric radius of the nova envelope shrinks to near theorbit (the separation a ), the condition of ε rot < ε mag is sat-isfied as shown in Figure 2. This indicates that the epochof the secondary maximum is shortly after the companionemerges from the nova envelope. We here implicitly assumethat strong magnetic field connects the WD and the compan-ion (like in polar systems). The mechanism of activity we sup-pose is magnetic reconnection. Strong magnetic reconnectionoccurs between the WD and the companion. When the WDphotosphere is larger than the companion’s orbit, magnetic re-connection may occur deep inside the photosphere but the gaspressure (or gas thermal energy) at the reconnection region ismuch larger than the magnetic pressure (magnetic energy), sothe gas is not easily accelerated by magnetic force. On theother hand, when the gas pressure becomes smaller than themagnetic pressure, i.e., when the photosphere shrinks to theorbit (see Fig.2), then the gas is easily accelerated by mag-netic force and the envelope gas is massively ejected. Thusthis process increases the mass-loss rate around/near the sec-ondary maximum. V1493 AQL AND V2362 CYG
Hachisu & KatoV1493 Aql was discovered by Tago at mag 8.8 on July13.558 UT. The nova was not detected on his films of July5 and 9 (limiting mag 11 and 10.5, respectively). There-fore, we assume here that the outburst day is t OB = 2451372 . ∼ . ± . M ⊙ for the chemical composition of X = 0 . Y = 0 . X CNO = 0 . X Ne = 0 .
03, and Z = 0 .
02. The V and visual magnitudes deviate from our model light curveabout 100 days after the outburst. This is due probably to thecontribution of emission lines such as [O III ] in the nebularphase. The orbital period of P orb = 0 .
156 days was obtainedby Dobrotka et al. (2006) from the modulations with a verysmall amplitude of 0.015 mag. The mass of the donor star isestimated to be M = 0 . M ⊙ from equation (5). The cor-responding separation is a = 1 . R ⊙ and the effective radiusof the WD Roche lobe is R ∗ = 0 . R ⊙ . We have obtainedthe epoch of ε mag ≈ ε rot to be day 49 for our 1 . M ⊙ WDmodel as listed in Table 1. This timescale is consistent withthe peak of the secondary maximum and also the emergenceof the companion star from the WD envelope.V2362 Cyg was discovered by Nishimura at mag 10.5 on2006 April 2.807 UT. The nova was not detected on his filmstaken on March 28 (limiting mag 12) or on earlier patrol filmsback to 2001. Therefore, we assume here the outburst dayof t OB = 2453827 . I observation until day ∼
500 except the secondary maximum.The best-fit model is the WD mass of 0 . ± . M ⊙ for thechemical composition of X = 0 . Y = 0 . X CNO = 0 . Z = 0 .
02 (see, e.g., Munari et al. 2008, for observed val-ues). The R and y magnitudes slightly but the V and visualmagnitudes largely deviate from our model light curve afterthe secondary maximum. The orbital period of P orb = 0 .
207 days was obtained by Goranskij et al. (2008) from the mod-ulations with an amplitude of 0.11 mag. If we take the donormass of M = 0 . M ⊙ from equation (5), the separation is a = 1 . R ⊙ , and the WD Roche lobe is R ∗ = 0 . R ⊙ . Wehave examined the epoch of ε mag ≈ ε rot to be day 240 for ourmodel of 0 . M ⊙ WD as listed in Table 1. This timescaleis again reasonably consistent with the peak of the secondarymaximum and also the emergence of the companion star fromthe WD envelope. CONCLUSIONS
We have estimated the WD mass of the classical novaeV2491 Cyg by comparing our free-free light curves with theoptical and near infrared observations as well as by compar-ing our blackbody X-ray light curves with the
Swift
XRT data.The best-fit model is the 1 . ± . M ⊙ WD for the chemicalcomposition of X = 0 . Y = 0 . X CNO = 0 . X Ne = 0 . Z = 0 .
02. We strongly recommend composition analysisof the ejecta because the enrichment of WD matter providesinformation whether the WD mass increases or not. We havealso estimated the WD mass of V1493 Aql and V2362 Cygto be M WD = 1 . ± .
05 and 0 . ± . M ⊙ , respectively.For these three novae, the epoch of secondary maximum isconsistently explained by our magnetic activity model if themagnetic activity reaches maximum at ε mag ≈ ε rot in the WDenvelope. We strongly recommend search for magnetic activ-ities for these three novae even in quiescence.We thank D. Takei for providing us with their machine read-able X-ray data of V2491 Cyg and also AAVSO and VSOLJfor the optical and near infrared data for V2491 Cyg, V1493Aql, and V2362 Cyg. We are grateful to the anonymous ref-eree for useful comments and to D. Takei, M. Tsujimoto, andS. Kitamoto for stimulating discussion on V2491 Cyg. Thisresearch has been supported in part by the Grant-in-Aid forScientific Research (20540227) of the Japan Society for thePromotion of Science.in the WDenvelope. We strongly recommend search for magnetic activ-ities for these three novae even in quiescence.We thank D. Takei for providing us with their machine read-able X-ray data of V2491 Cyg and also AAVSO and VSOLJfor the optical and near infrared data for V2491 Cyg, V1493Aql, and V2362 Cyg. We are grateful to the anonymous ref-eree for useful comments and to D. Takei, M. Tsujimoto, andS. Kitamoto for stimulating discussion on V2491 Cyg. Thisresearch has been supported in part by the Grant-in-Aid forScientific Research (20540227) of the Japan Society for thePromotion of Science.