White Dwarfs in Cataclysmic Variables : An Update

In this review, we summarize what is currently known about the surface temperatures of accreting white dwarfs in nonmagnetic and magnetic cataclysmic variables (CVs) based upon synthetic spectral analyses of far ultraviolet data. We focus only on white dwarf surface temperatures, since in the area of chemical abundances, rotation rates, WD masses and accretion rates, relatively little has changed since our last review, pending the results of a large HST GO program involving 48 CVs of different CV types. The surface temperature of the white dwarf in SS Cygni is re-examined in the light of its revised distance. We also discuss new HST spectra of the recurrent nova T Pyxidis as it transitioned into quiescence following its April 2011 nova outburst.


Introductory Overview
The white dwarfs in cataclysmic variables (hereafter CVWD) are the central engines of the observed outbursts, either as potential wells for the release of gravitational energy during accretion (dwarf nova -DN), or as the sites of explosive thermonuclear runaway (TNR) shell burning (classical novae), steady shell burning (supersoft X-ray binaries) or instantaneous collapse and total thermonuclear detonation if the WD reaches the Chandrasekhar limit (Type Ia supernova?SN Ia).Therefore, the accreting WDs serve as probes of explosive evolution and accretion physics and diffusion, as they bear the thermal, chemical and rotational imprint of their long term accretion and thermonuclear history (Sion 1991(Sion & 1995 Deeper physical insights however require a larger number of chemical abundances, rotation rates, surface temperatures, mass accretion rates, and masses for each spectroscopic subclass of CVs.Only then can any definitive conclusions be drawn.In order to adequately sample the parameter space (M wd , i, Ṁ , T ef f , P orb ) of the DNs, Nova-like variables, and magnetic CVs, a large GO program was approved in Cycle 20 (B.Gänsicke, Principal Investigator) to secure high quality COS spectra for CV classes underrepresented in the current overall CV sample.The data is in hand and undergoing analysis at the time of this writing.Ultimately, we hope to obtain data for > 30 CVs per class.Therefore, in this review our focus is restricted to white dwarf surface temperatures, since in the area of chemical abun-dances, rotation rates, WD masses and accretion rates, relatively little has changed since our last review in the 2011 Palermo meeting proceedings.In section 2, our FUV analysis techniques are briefly summarized, section 3 we address how the revised (shorter) distance to SS Cygni affects the results of our analysis of the FUSE + HST STIS spectra of SS Cygni by Sion et al. (2010), in section 4 we tabulate and display the current distribution of CVWD surface temperatures versus orbital period and in the final section, we include some remarks on new HST spectra of the recurrent nova T Pyxidis.

Synthetic Spectral Analysis of FUV Spectra of CVWDs
We have modeled the FUV spectra of WDs in CVs during DN quiescence and Nova-like low states from IUE, FUSE and HST (FOS, GHRS, STIS, COS) archival data, and through our ongoing collaboration with past HST surveys led by B.Gänsicke, and P. Szkody.
The IUE archival spectra and high quality FUSE and HST FUV data are fit with the latest versions of the TLUSTY/SYNSPEC model photosphere code and model accretion disk codes (Hubeny 1988;Hubeny & Lanz 1995).We are taking into account the BL explicitly in our modeling of the FUV spectra of disk accreting systems, by replacing the very inner rings of the standard accretion disk model with high temperature rings in agreement with the temperature and density in the boundary layer.This improves the model fitting at the shorter wavelengths.As an example of our CVWD photosphere fitting, in Fig. 1, we display a WD solar composition fit to the HST STIS spectrum of V442 Cen with E(B-V) = 0.10.The WD model has T=47,000K ±2000K and Log(g) = 8.3 ± 0.2, V sin(i) = 300km/s ±50 km/s (Sion et al. 2008).In Table 1, we have tabulated what we regard as the most reliably secured WD surface temperatures.The first column gives the name of the CV, the second column the CV subtype, the third column, the orbital period in minutes, the fifth column the WD T ef f and the sixth column, the reference for the temperature.These temperatures are derived from a variety FUV spectra and all are known with a precision of at least 3000K and in the majority of cases, better that +/-2000K.With the paucity of FUV-derived temperatures per CV subclass, one should not ignore usable FUV data from IUE, FUSE and HUT because the "quality" of the data is deemed inferior to HST STIS and COS.
The current distribution of WD T ef f against the orbital period P orb is displayed in Fig. 2. (see also Fig. 4 in Townsley & Gänsicke 2009).Note the continued relatively sparse coverage in temperature of the WDs in CVs above the period gap (P orb > 3hr), compared with the coverage below the period gap.
It is possible that the apparent trend toward higher temperatures with increasing P orb (i.e. higher long term average < Ṁ >) could be due to observational selection since the WD T ef f 's were derived primarily in the FUV where the Planckian peak occurs for hotter accreting WDs.The interstellar reddening E(B-V) of the recurrent nova T Pyxidis is a critical parameter in the determination of the best-fitting model parameters.
The UV spectra exhibit a minimum near 2175 A which is due to the interstellar extinction.Since the GALEX spectrum is the most reliable (i.e.highest S/N ratio) in that wavelength region, we used the GALEX spectrum to determine E(B-V).The value of E(B-V) for which the 2175 A feature disappears from the dereddened spectrum, E(B-V)=0.35, is taken as the E(B-V) value towards T Pyx (see Fig. 3).We use this value to deredden the IUE, GALEX and HST spectra, and we also consider the effects of different reddening values on our results.
The HST STIS and COS spectra obtained in December 2012 and July 2013 are identical.Thus, we co-added them to improve the signal-to-noise (S/N).We have found that the pre-outburst IUE and GALEX spectra together with the HST post-outburst spectrum.We note that the UV flux has remained constant not only before the outburst, but it has now come back precisely to the same value.This is an indication that The accreted envelope is larger than the ejected envelope, except for a value of E(B-V)=0.25 combined with a WD mass ∼ 0.9M and larger.However, the best results were obtained for a larger value of the reddening E(B − V ) > 0.30.
Using the value of the reddening we derived, together with the new distance estimate of 4.8 kpc (Sokoloski et al. 2013), we fit the observed (back-toquiescence) HST spectra with disk models for different WD masses.We then computed the accreted mass over a period of 44yrs, which we then compared to the estimates of the ejected envelope mass (during the 2011 outburst).We recapitulate our results in Fig. 4, where we also consider different reddening values for the sake of completness.Our main finding is that the mass of the accreted material between the last two outbursts is larger than the mass of the ejected envelope in the last outbursts, unless the reddening is E(B-V)=0.25 (or smaller) AND the mass of the WD is ∼ 0.9M or larger.However, such a small value of the reddening leads to a very bad fit and has therefore to be rejected.

Figure 1 :
Figure 1: The best-fitting single temperature WD fit to the HST/STIS spectrum of V442 Cen.The model consists of a 47,000K WD with Log(g)=8.3and scaled to a distance of 328pc.

3
The White Dwarf in SS Cygni:The VLBI and Corrected Hubble FGS Distance Schreiber & Lasota (2007), and references therein, pointed out that the previously published Hubble FGS parallax of SS Cygni, with a distance of 166 +/-12 pc, posed serious problems for the disc instability model (DIM) of dwarf nova outbursts because it would fail to explain the absolute magnitude during outburst.With the new VLBI-derived distance (Miller-Jones et al. 2013) to SS Cygni (114 pc), and the corrected HST distance (Nelan & Bond 2013) of 120 pc instead of 166 pc, the concern for the validity of DIM is alleviated.An obvious question is: How does the new, shorter distance affect the analysis of Sion et al. (2010) regarding their detection of the WD during quiescence and their derivation of the WD's surface temperature from the modeling of FUSE + HST STIS spectra?To answer this question, we have carried out model fits using the new distance in the range of 114 pc to 120 pc.When we used single, steady state disk models, we obtained the same results as inSion et al. (2010).The disk does not fit the flux in the shorter wavelengths unless the accretion rate is very large, which is inconsistent with dwarf nova quiescence, and the model-derived distance is far too large.When we combined a model disk with a model white dwarf photosphere (using the Bitner et al.WD mass as in Sion et al. (2010), the correct distance to SS Cygni was obtained for a low accretion rate and a lower temperature white dwarf.However, the fits with the WD + disk are inferior to the fits where the WD dominates the FUV flux.In order to match the best fit solution to the FUSE + HST STIS wavelength range in Sion et al. (2010), the white dwarf must have T ef f = 45000K -50000K for a distance of 112 pc to 120 pc and a white dwarf mass M wd = 0.95M .This value of the WD mass is within the error range of the Bitner et al. mass.Our conclusion about the T ef f of the WD in SS Cygni during quiescence is essentially unchanged from Sion et al. (2010).The effective temperatures T ef f of the WDs (obtained during dwarf nova quiescence and nova-like low states when the WD is exposed), are the critical key to revealing the thermal response of the WD to mass accretion and the long term accretion rate < Ṁ > which, through compressional heating by the accreted material, is linked to the WD surface temperature (Sion 1995; Townsley & Bildsten 2003; Townsley & Gänsicke 2009).
For example, Copperwheat et al. (2010) derive a T ef f = 10 − 15, 000K in the optical for the WD in the eclipsing DN IP Peg which has P orb = 3.8 h while the eclipsing system SDSS1006 with P orb = 4.46 h, has T ef f = 16, 500 ± 2000K (Southworth et al. 2009) in the optical.Both of these objects should contain much hotter WDs (higher < Ṁ >) for their orbital periods.

Figure 2 :
Figure 2: Effective White Dwarf Temperature as a function of the orbital period.The references for the indivdual temperatures can be found in Sion et al. 2008, Townsley & Gänsicke 2009, Araujo-Betancor 2005a & b and references therein).The traditional magnetic braking above the period gap is shown between the parallel diagonal solid lines.On the right hand side are the time-averaged accretion rates correpsonding to the temperature scale on the left hand side of the diagram.Shown for comparison between the dotted lines is the long term evolutionary path of a 0.8 solar mass white dwarf (with an initial core temperature of 30 million degrees K) which has undergone 1000 nova outburst cycles accreting at the long term rate of 10 −8 M /yr

Figure 3 :Figure 4 :
Figure 3: The merged Galex-IUE spectrum of T Pyx has been dereddened for different values of E(B-V) as indicated on the right.The 2175 Å feature associated with the reddening is clearly seen in absoprtion for low balues of E(B-V), and it appears as extra flux for large values of E(B-V).We deduce that the reddening towards T Pyx must be E(B-V)=0.35,the value for which the 2175 Å feature vanishes.

Table 1 :
The temperature of CV white dwarfs