Using GALEX-SDSS-PanSTARRS-HST-Gaia to understand post-AGB evolution

  • Luciana Bianchi
  • Graziela R. Keller
  • Ralph Bohlin
  • Martin Barstow
  • Sarah Casewell
Original Article
Part of the following topical collections:
  1. UV Surveys, the Needs and the Means


Hot WDs in binary systems with a less evolved star are particularly invaluable astrophysical probes, the unevolved companion enabling better derivation of distance and age than is usually possible for post-AGB objects, and therefore also of their radius and luminosity. But hot white dwarfs (WD) are elusive at all wavelengths except the UV (Bianchi et al. 2011a). From our GALEX UV source catalogs (Bianchi et al. 2011a,b, 2014, 2017) matched to SDSS, we identified thousands of candidate hot WDs including WDs in binary systems consisting of a hot WD and a companion of spectral type from A to M. The identification and preliminary characterization of the stellar parameters is based on the analysis of the photometric SED from far-UV to z-band.

We have observed subsamples of the UV-selected WDs with the Hubble Space Telescope (HST) to better characterize their stellar parameters. We obtained (1) UV spectroscopy with STIS and analyzed the UV spectra together with optical SDSS spectra, and (2) multi-band imaging with WFC3 (\(0.04^{\prime\prime}/\mbox{pixel}\)) to measure angular separation and individual SEDs of the pair’s components in binary systems. In our HST/WFC3 sample of 59 hot-WD binaries with late-type companions, we found that at least a dozen have possibly evolved without exchanging mass. The UV STIS spectroscopy led to the revision of previous results based on optical spectra only, because of the often undetectable or unquantifiable contribution of the hot component to the optical fluxes.


Astronomical data bases: surveys Stars: white dwarfs Stars: AGB and post-AGB Stars: binaries: general (Stars:) variables Galaxy: stellar content 

1 Introduction. Searching for the elusive hot white dwarf population in the Milky Way

Stars with initial mass up to \(\sim8 M_{\odot}\) end their evolution as white dwarfs (WD) with a narrow mass range, mostly below \(0.8M_{\odot}\) (e.g. Kepler et al. 2007; Holberg et al. 2008; Herwig 2005). They reach the WD cooling track after evolving through the asymptotic giant branch (AGB) and planetary nebula (PN) phases, in which they shed much of their mass. The ejected material enriches the interstellar medium (ISM) with newly synthesized nuclear products (e.g. Herwig 2005; Marigo 2001, 2012; Karakas 2010; Karakas et al. 2002; Weiss and Ferguson 2009). Because the IMF is skewed towards lower masses, the WD progenitors, i.e. the intermediate-mass stars, are the main providers of carbon, nitrogen, and other important elements.

In spite of their importance for the chemical evolution of galaxies, the late evolutionary phases are still poorly understood, in particular concerning AGB mass-loss processes and the efficiency of the so-called “third dredge-up”, which brings chemically enriched material from inner layers to the surface. The initial-final mass relation (IFMR), which maps the final WD mass to the progenitor’s initial mass, is critical for understanding the yield of chemical elements. In spite of much recent work (e.g., Cummings et al. 2018) it sorely lacks observational constraints, especially for the most massive WD progenitors which are the most relevant for the chemical enrichment. Significant differences in CNO predictions result from different assumptions (e.g., di Criscienzo et al. 2016; Karakas and Lugaro 2017; Karakas et al. 2018; Ventura et al. 2018). The scarcity of observed post-AGB objects in the hottest \(T_{\mathrm{eff}}\) regime (Sion 2011), where they are difficult to identify from optical surveys (because of their low optical luminosity, and high \(T_{\mathrm{eff}}\) to which optical colors are saturated), and their uncertain placement on the post-AGB evolutionary tracks due to the difficulty of establishing their distances (except for the nearest objects), are largely responsible for the persisting uncertainties.

Hot WDs in binaries in particular offer invaluable clues to stellar evolution. Interacting binaries such as cataclysmic variables and supersoft X-ray sources offer insight on accretion processes and close binary evolution: these types of systems are rather well studied because they are easy to detect; WD mergers or single-denegerate systems have been suggested as possible Ia supernovae progenitors (e.g., Yungelson and Livio 2000; Maoz et al. 2014). Instead, binaries with a hot WD and a main-sequence (MS) or giant companion, separated enough to have not exchanged mass, are extremely hard to identify unless UV data are available, the presence of the very-hot, low-optical-luminosity star being often inconspicuous or indistinguishable from optical data (Fig. 5). For example, the current census of Sirius-like systems (binary or multiple star systems with a WD and a companion of spectral type K or earlier) is largely incomplete beyond 20 pc from the Sun (Holberg et al. 2013). These binaries are important for studying the initial-final mass relation, the degenerate mass-radius relation (Holberg et al. 2012), and binary evolution. However, because detecting a hot WD (small radius) close to a cooler star requires UV data, these binaries remained elusive, until our GALEX + SDSS catalogs yielded several thousand new candidates.

From GALEX UV surveys matched to optical photometry and spectroscopy, we extracted unprecedented samples of hot-WD binaries with unevolved companions (Bianchi et al. 2011a, 2018a). Over half of the sample are unknown stars, and most of the known ones were not previously identified as including a WD. Out of thousands of Sirius-like system candidates which we identified from GALEX FUV, NUV data and SDSS optical photometry and spectroscopy, we selected two samples for follow-up observations with the Hubble Space Telescope (HST). The pairs are not resolved in GALEX and SDSS images (Figs. 1, 4, 5), however the 7-band photometry \(\mathit{FUV}\), \(\mathit{NUV}\), \(u\), \(g\), \(r\), \(i\), \(z\), analyzed with synthetic stellar model grids, enabled a successful selection of candidates, and an initial characterization of a sample about two orders of magnitudes larger than previous ones (Bianchi et al. 2018a). Gaia data release 2 (DR2) provides direct measurements of distance for many of these systems when the cool star is detected by Gaia (the hot WD companion is almost always too faint) but our sample includes objects at large distances, up to beyond 1 kpc, and parallax errors are very large for most of the sample.
Fig. 1

Top: Cut-out images of one of the STIS-UV-spectroscopy targets; from left: SDSS color-composite and separate \(u\)\(g\), \(r\) images (resolution \(\sim1.4^{\prime\prime}\), \(\mbox{image size} =40^{\prime\prime}\times40^{\prime\prime}\)), and STIS acquisition image (resolution \(0.05^{\prime\prime}/\mbox{pxl}\), image size about \(2^{\prime\prime}\), optical broad band). This target is semiresolved in the HST acquisition image. Second row: (left) SED fit of the GALEX FUV, NUV + SDSS \(u\), \(g\), \(r\), \(i\), \(z\) photometry, from which binarity is inferred, and initial estimates of stellar parameters are derived, used for target selection of the HST follow-up project. Right: a \(\chi^{2}\) representation (color bar indicates \(\chi^{2}\) values) of the model fit to the stellar spectrum, from which three parameters \(T_{\mathrm {eff}}\), \(\log g\) and \(E_{B\mbox{--}V}\) are concurrently derived. Bottom plot: the STIS spectrum (black) plotted with the range of acceptable models (orange); thin black lines are flux error (on individual points). Adapted from Keller et al. (2018)

We have chosen to focus on these systems because they offer the ideal test-bench to constrain post-AGB evolution: (i) the mass of the less-evolved star is a lower limit to that of the WD progenitor (more evolved, hence initially more massive), (ii) the distance can be determined for a MS or giant star much better than it is possible for a WD, therefore WD luminosity and radius can also be derived, (iii) the age of the less evolved star yields an evolutionary time for the system hence, (iv) the mass of the WD and of its progenitor (or lower limit) can be constrained (IFMR).

This paper introduces two different ongoing projects aimed at validating the identification and characterization of hot WD in our larger GALEXxSDSS sample, as presented at the NUVA workshop 2017. In Sect. 2 we report on our ongoing study from HST/STIS UV spectroscopy of fifteen GALEX-selected hot WDs, some single stars and some in binary systems with a cooler companion (Bianchi’s HST-GO-13397). In Sect. 3 we report on another project (Bianchi’s HST 14119), in which we imaged selected binaries with HST’s WFC3 camera, at 218, 275, 336, 475, and 606 nm, to measure the separation (or upper limit) of the stellar pair down to the limit corresponding to the WFC3/UVIS resolution, \(0.04^{\prime\prime}/\mbox{pxl}\), and the individual SEDs for resolved pairs.

2 Characterizing binaries with a hot white dwarf from UV HST/STIS and optical SDSS spectroscopy

For fifteen sources selected from our GALEXxSDSS samples, part presumed single hot WDs and part hot WDs with a cooler stellar companion, we obtained HST/STIS spectra with the G140L and G230L gratings, covering a total range of 1150–3200 Å at resolution \(R \sim1000\).

In Figs. 1, 2 and 3 we show the optical ground-based (SDSS) and STIS acquisition (filter: 50CCD) images, and best-fit models of the composite UV+optical spectra of one target, yielding for the cool star \(T_{\mathrm{eff}}=3300~\mbox{K}\), \(\log g=5.0\) (fitted with MARCS models, Gustafsson et al. 2008). Gaia DR2 gives a parallax of 1.53 mas with an error of about 10%, i.e. a distance \(D=654~\mbox{pc}\). The best-fit model for the hot component has \(T_{\mathrm{eff}}=29{,}000~\mbox{K}\) and \(\log g = 7.0\) (using a grid of models of Bianchi et al. 2011a computed with the TLUSTY code, described by Hubeny and Lanz 1995). At the distance of 654 pc, the best-fit model scaled to the observed flux implies a radius of \(R_{\mathrm{WD}} \sim0.02 R_{\odot}\) and \(L_{\mathrm{bol}}\sim0.25 L_{\odot}\). Uncertainties are large especially in \(\log g\), which propagate to the derived parameters. Figure 3 illustrates the sensitivity of the Lyα line in STIS UV spectra to the stellar parameters and to even tiny amounts of extinction.
Fig. 2

Fit of the STIS UV plus SDSS optical spectra of the target shown in the previous figure. The lower panel is an enlarged view of the UV spectrum. In the fit, an interstellar (ISM) absorption of Lyα, with column density consistent with the \(E_{B\mbox{--}V}\) value used to deredden the spectrum, is added to the stellar model, to further constrain the derived parameters (see next figure). Adapted from Keller et al. (2018)

Fig. 3

The sensitivity of the STIS G140L spectra (Lyα line) to temperature and gravity of the hot WD shown in the previous figures, in this \(T_{\mathrm{eff}}\) regime. The red/green/blue models have reddening applied with \(E_{B\mbox{--}V}=0.005,0.01,0.02\) respectively. TLUSTY models from the grid of Bianchi et al. (2011a) are convolved to the STIS/G140L line-spread function

Figure 4 shows the best-fit model results for another target which again has a hot and a cool component. We derive \(T _{\mathrm{eff}}=40{,}000~\mbox{K}\), \(\log g=8.0\) for the hot star and \(T _{\mathrm{eff}}=3{,}500~\mbox{K}\), \(\log g=5.0\) for the cool star; Rebassa-Mansergas et al. (2010) had derived \(T_{\mathrm{eff}}=8{,}187\mbox{--}8{,}282~\mbox{K}\), \(\mbox{type}=\mbox{M}+\mbox{DA}\), \(\mbox{distance} = 668\mbox{--}823~\mbox{pc}\) from analysis of the SDSS spectrum, and Heller et al. (2009) a distance of 648.3–439.6 pc. Our fit including UV spectra, which accounts for the contribution of the hot star to the optical fluxes, implies a larger distance; assuming a distance of \(\sim1~\mbox{kpc}\) the WD radius (derived from the scaled model fit) would be \(\sim0.03R_{\odot}\), and the separation of \(0.223^{\prime\prime}\) measured in the ACQ image would translate to 0.0011 pc, implying that the two stars have not exchanged mass during their previous evolutionary phases. This object has a photometric detection in Gaia DR2 but not a parallax measurement, implying again a large distance. Analysis is in progress with new grids of models, to refine the parameters.
Fig. 4

Top: Cut-out images for another UV-spectroscopy target from Bianchi’s program 13397; this binary is well resolved in the STIS acquisition image. As in Fig. 1, SDSS u, g, r, and STIS acquisition images are arranged from left to right. Lower panels: The UV + optical spectra of the system, fitted with stellar models of a hot WD (TLUSTY model) and a cool star (MARCS model). Previous analysis based on SDSS spectra only (Rebassa-Mansergas et al. 2010) had obtained a different \(T_{\mathrm{eff}}\) for the cool star, and by consequence also a different distance was inferred

3 Characterizing binaries with a hot white dwarf from high-resolution HST/WFC3 imaging

We have observed fifty-nine Sirius-like systems identified and selected from our large GALEXxSDSS matched samples with the WFC3 camera onboard HST. Each target was imaged in five filters, at 218, 275, 336, 475, and 606 nm. Two or three dithered sub-exposures were taken in each filter, to improve the resulting photometry. The WFC3 scale is \(0.04^{\prime \prime}/\mbox{pxl}\), allowing us to resolve and measure the separation of stellar pairs, or derive an upper limit much more stringent than from the GALEX and SDSS data (resolution \(\sim5^{\prime\prime}\) and \(1.4^{\prime\prime}\) respectively) where none of these systems is resolved. For the resolved pairs, we can measure the individual SEDs which allow us to derive concurrently stellar parameters and extinction. There is no far-UV capability in the WFC3 filter complement, however the far-UV flux has essentially no contamination from the cool star companion (Fig. 5), and GALEX FUV can be used to extend the SED of the hot WD.
Fig. 5

Examples of WFC3 imaging from Bianchi’s HST program 14119, fully resolving some of the binary targets. For each target, we show on the left the GALEX (\(\sim4.2/5.^{\prime\prime}\) resolution FUV/NUV) and SDSS images (\(\sim1.4^{\prime\prime}\) resolution) (top), and the five-filter WFC3 images (lower panels), from near-UV to 606 nm; Right: the two-component SED-fitting of the unresolved GALEX + SDSS photometry (radii are given for a reference distance of 1 kpc)

Figure 5 shows cut-outs from GALEX and SDSS images, and from the HST WFC3 5-filter images, for some binaries resolved with HST and an unresolved one. Only with HST data we can measure the separation, or a stringent upper limit. Combined with Gaia direct distance measurements, available only for a subsample in DR2, we will be able to derive also the current stellar parameters for both components, and, comparing these with stellar evolutionary models, to also infer the age of the system, and the initial mass for the WD progenitor. About a dozen targets are fully resolved so that the two separate components can be reliably measured, other targets appear elongated: we are currently performing tests with artificial star photometry to extricate stellar parameters for these (Bianchi et al. 2018b).

4 Summary and future work

From GALEX UV surveys matched to optical photometry and spectroscopy, we extracted unprecedented samples of hot-WD binaries with unevolved companions. HST follow-up was performed for two sub-samples: several pairs, unresolved in GALEX-SDSS imaging, can be fully resolved with HST. Other cases appear elongated, and work is in progress to establish upper limits to the separation. Based on simple calculations, considering a range of possible mass ratios and that total mass and momentum may be conserved or not, we estimate that for separations \(\geq4000 R _{\odot}\), a rather conservative limit, no mass exchange has occurred even in the AGB phase of the WD progenitor, thus the stars have likely evolved as two single stars. Pairs with smaller separation may have exchanged mass, and must be interpreted with evolutionary models for interacting binaries. Gaia parallax measurements in DR2 are only available for a subsample, and very few of our targets have parallaxes with small errors. Therefore, our sample, stretching to distances up to a few kpc in low-extinction sightlines, significantly extends in volume and complements the Gaia WD sample, which is limited to \(\sim120~\mbox{pc}\) from the Sun (Babusiaux et al. 2018). The fraction of non-interacting binaries from our HST samples can be compared to the figures recently derived for unevolved massive and intermediate-mass stars. Our preliminary results indicate that a smaller fraction of resolved binaries is found in late evolutionary stages with respect to the incidence of binaries in earlier evolutionary phases, for which recent works reported \(\gtrsim50\%\), up to 70% in O-type stars, possibly lower but less well constrained for ∼B types, many of the O-star binaries likely having interacted (e.g., Sana 2017; Moe and di Stefano 2016; Caballero-Nieves et al. 2014). A lower percentage of detached binaries at late evolutionary stages, if confirmed, would imply that a good fraction of the initial binaries interact or merge.

For stars in resolved pairs, where measurements of separation and individual stellar parameters are possible, by combining our results with Gaia distances (when available) we will also be able to constrain initial and current WD mass, age and luminosity. Results from both HST samples confirm that a fraction of the targets can be spatially resolved with HST; the UV spectra proved to be decisive in revising spectral parameters over analyses based on optical spectra alone.

More information on our UV catalogs and related projects can be found at the author’s \(\mathit{UVSKY}\) web site



The work made use of GALEX, SDSS, and Gaia data downloaded from the MAST archive, and of data from HST programs HST-GO-13397 and HST-GO-14119 (L. Bianchi P.I.). Support for Program numbers HST-GO-13397 and HST-GO-14119 was provided by NASA through a grant from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Incorporated, under NASA contract NAS5-26555.


  1. Babusiaux, C., van Leeuwen, F., Barstow, M.A., et al. (Gaia collaboration): Astron. Astrophys. (2018, in press) Google Scholar
  2. Bianchi, L., Shiao, B., et al.: (2018a, in preparation) Google Scholar
  3. Bianchi, L., Keller, G., et al.: (2018b, in preparation) Google Scholar
  4. Bianchi, L., et al.: Mon. Not. R. Astron. Soc. 411, 2770 (2011a). ADSCrossRefGoogle Scholar
  5. Bianchi, L., Herald, J., Efremova, B., et al.: Astrophys. Space Sci. 335, 161 (2011b). ADSCrossRefGoogle Scholar
  6. Bianchi, L., Conti, A., Shiao, B.: Adv. Space Res. 53, 900–912 (2014) ADSCrossRefGoogle Scholar
  7. Bianchi, L., Shiao, B., Thilker, D.: Astrophys. J. Suppl. Ser. 230, 24 (2017) ADSCrossRefGoogle Scholar
  8. Caballero-Nieves, S., et al.: Astrophys. J. 147, 40 (2014) Google Scholar
  9. Cummings, et al.: Astrophys. J. (2018, in preparation) Google Scholar
  10. di Criscienzo, M., et al.: Mon. Not. R. Astron. Soc. 462(1), 395 (2016) ADSCrossRefGoogle Scholar
  11. Gustafsson, B., Edvardsson, B., Eriksson, K., Jørgensen, U.G., Nordlund, Å., Plez, B.: Astron. Astrophys. 486, 951 (2008) ADSCrossRefGoogle Scholar
  12. Heller, R., Homeier, D., Dreizler, S., Ostensen, R.: Astron. Astrophys. 496, 191 (2009) ADSCrossRefGoogle Scholar
  13. Herwig, F.: Annu. Rev. Astron. Astrophys. 43, 435 (2005) ADSCrossRefGoogle Scholar
  14. Holberg, J.B., Sion, E.M., Oswalt, T., McCook, G.P., Foran, S., Subasavage, J.P.: Astron. J. 135, 1225 (2008) ADSCrossRefGoogle Scholar
  15. Holberg, J., et al.: Astron. J. 143, 68 (2012) ADSCrossRefGoogle Scholar
  16. Holberg, J., et al.: Mon. Not. R. Astron. Soc. 435, 2077 (2013) ADSCrossRefGoogle Scholar
  17. Hubeny, I., Lanz, T.: Astrophys. J. 439, 875 (1995) ADSCrossRefGoogle Scholar
  18. Karakas, A.I.: Mon. Not. R. Astron. Soc. 403, 1413 (2010) ADSCrossRefGoogle Scholar
  19. Karakas, A., Lugaro, M.: JPS Conf. Proc. 14, 010403 (2017). Google Scholar
  20. Karakas, A.I., Lattanzio, J.C., Pols, O.R.: Publ. Astron. Soc. Aust. 19(4), 515 (2002) ADSCrossRefGoogle Scholar
  21. Karakas, A., et al.: Mon. Not. R. Astron. Soc. 477, 421 (2018) ADSCrossRefGoogle Scholar
  22. Keller, G.R., Bianchi, L., et al.: (2018, in preparation) Google Scholar
  23. Kepler, S.O., Kleinman, S.J., Nitta, A., Koester, D., Castanheira, B.G., Giovannini, O., Costa, A.F.M., Althaus, L.: Mon. Not. R. Astron. Soc. 375, 1315 (2007) ADSCrossRefGoogle Scholar
  24. Maoz, D., et al.: Annu. Rev. Astron. Astrophys. 52, 187 (2014) CrossRefGoogle Scholar
  25. Marigo, P.: Astron. Astrophys. 370, 194 (2001) ADSCrossRefGoogle Scholar
  26. Marigo, P.: In: IAUS 283: Planetary Nebulae: an Eye to the Future, p. 87 (2012) Google Scholar
  27. Moe, M., di Stefano, R.: arXiv:1606.05347 (2016)
  28. Rebassa-Mansergas, A., et al.: Mon. Not. R. Astron. Soc. 402, 620 (2010) ADSCrossRefGoogle Scholar
  29. Rebassa-Mansergas, A., et al.: Mon. Not. R. Astron. Soc. 419, 806 (2012) ADSCrossRefGoogle Scholar
  30. Sana, H.: In: IAUS 329: The Lives and Death-Throes of Massive Stars, p. 110 (2017) Google Scholar
  31. Sion, E.: S1, arXiv:1111.6652 (2011)
  32. Ventura, P., et al.: Mon. Not. R. Astron. Soc. 477, 438 (2018) ADSCrossRefGoogle Scholar
  33. Weiss, A., Ferguson, J.W.: Astron. Astrophys. 508, 1343 (2009) ADSCrossRefGoogle Scholar
  34. Yungelson, R., Livio, M.: Astrophys. J. 528, 108 (2000) ADSCrossRefGoogle Scholar

Copyright information

© Springer Nature B.V. 2018

Authors and Affiliations

  1. 1.Department of Physics & AstronomyThe Johns Hopkins UniversityBaltimoreUSA
  2. 2.Science Systems and Applications, Inc.LanhamUSA
  3. 3.Space Telescope Science InstituteBaltimoreUSA
  4. 4.Department of Physics & AstronomyUniversity of LeicesterLeicesterUK

Personalised recommendations