Gianninas 2011

Accepted for publication in the Astrophysical Journal
Preprint typeset using LATEX style emulateapj v. 5/2/11
A SPECTROSCOPIC SURVEY AND ANALYSIS OF BRIGHT, HYDROGEN-RICH WHITE DWARFS

A. Gianninas1,3 , P. Bergeron1 , and M. T. Ruiz2,3
1 Département de Physique, Université de Montréal, C.P. 6128, Succ. Centre-Ville, Montréal, Québec H3C 3J7, Canada
2 Departamento de Astronomı́a, Universidad de Chile, Casilla 36-D, Santiago, Chile and
3 Visiting astronomers at Las Campanas Observatory operated by Carnegie Institution of Washington
Accepted for publication in the Astrophysical Journal
ABSTRACT
arXiv:1109.3171v1 [astro-ph.SR] 14 Sep 2011
We have conducted a spectroscopic survey of over 1300 bright (V ≤ 17.5), hydrogen-rich white
dwarfs based largely on the last published version of the McCook & Sion catalog. The complete results
from our survey, including the spectroscopic analysis of over 1100 DA white dwarfs, are presented.
High signal-to-noise ratio optical spectra were obtained for each star and were subsequently analyzed
using our standard spectroscopic technique where the observed Balmer line profiles are compared
to synthetic spectra computed from the latest generation of model atmospheres appropriate for these
stars. First, we present the spectroscopic content of our sample, which includes many misclassifications
as well as several DAB, DAZ, and magnetic white dwarfs. Next, we look at how the new Stark
broadening profiles affect the determination of the atmospheric parameters. When necessary, specific
models and analysis techniques are used to derive the most accurate atmospheric parameters possible.
In particular, we employ M dwarf templates to obtain better estimates of the atmospheric parameters
for those white dwarfs that are in DA+dM binary systems. Certain unique white dwarfs and double-
degenerate binary systems are also analyzed in greater detail. We then examine the global properties
of our sample including the mass distribution and their distribution as a function of temperature. We
then proceed to test the accuracy and robustness of our method by comparing our results to those
of other surveys such as SPY and SDSS. Finally, we revisit the ZZ Ceti instability strip and examine
how the determination of its empirical boundaries are affected by the latest line profile calculations.
Subject headings: binaries: spectroscopic – mass function – stars: fundamental parameters – stars:
luminosity function – stars: oscillations – surveys – techniques: spectroscopic –
white dwarfs
1. INTRODUCTION particularly well studied despite their relative bright-

For many years, the catalog of McCook & Sion (1999, ness, which makes it easier to obtain high S/N spectro-
hereafter MS99) was the most important reference in the scopic observations. Indeed, many of these white dwarfs
white dwarf community including a total of 2249 spectro- have only a spectral classification, based on photographic
scopically identified white dwarf stars. Similarly, the now spectra, without any further analysis ever having been
regularly updated online incarnation1 is used by white performed, let alone with modern CCD spectroscopy.
dwarf researchers on a daily basis and has grown to in- Therefore, we believe it is both important and worth-
clude nearly 13,000 entries as this publication goes to while to study these stars, the more so that it is much
press. This is due in large part to the increasingly large more feasible to conduct follow-up observations (i.e. high
number of white dwarfs discovered in the Sloan Digital resolution spectroscopy, high-speed photometry, etc.) for
Sky Survey (SDSS). The SDSS’s huge database of newly objects of interest, if they are bright.
discovered white dwarfs has permitted the discovery of Indeed, some of the early results from our survey have
previously unknown types of white dwarfs like the carbon already allowed us to make a few discoveries of our own.
atmosphere “Hot DQ” stars (Dufour et al. 2008), white The most notable of these is likely the discovery of the
dwarfs with gaseous disks (Gänsicke et al. 2008), and unique nature of GD 362, originally reported as a highly
even white dwarfs with oxygen dominated atmospheres metallic and massive DAZ white dwarf (Gianninas et al.
(Gänsicke et al. 2010). However, the fact remains that 2004), it was later discovered that the main atmospheric
most of the SDSS data lack the necessary signal-to-noise constituent is helium (Zuckerman et al. 2007). Even
ratio (S/N) to measure accurately the atmospheric pa- more important was the discovery of an infrared ex-
rameters for each star (Gianninas et al. 2005). This is cess due to the presence of a circumstellar debris disk
largely due to the rather faint white dwarfs observed by (Becklin et al. 2005; Kilic et al. 2005) around this star.
the SDSS and the practice of having a set exposure time This lone discovery essentially opened the door to a
for every field. Consequently, the fainter the star, the whole new area of white dwarf research with many groups
lower the S/N of the data. On the other hand, white launching observational campaigns with the specific goal
dwarfs from MS99 are comparatively bright, especially of discovering debris disks around other cool, metal-
if we restrict ourselves to stars with V ≤ 17.5. Histori- enriched white dwarfs. We have also successfully used
cally, many of the white dwarfs from MS99 have not been our survey to refine the empirical determination of the
boundaries of the ZZ Ceti instability strip and con-
firm its purity (Gianninas et al. 2005). We were thus
gianninas@astro.umontreal.ca, bergeron@astro.umontreal.ca, mtruiz@das.uchile.cl
1 http://www.astronomy.villanova.edu/WDCatalog/index.html able to predict the variability of several new ZZ Ceti
2 GIANNINAS, BERGERON & RUIZ
pulsators, while also confirming the non-variability of cluded in our analysis. Second, we have DA+dM binary
other DA white dwarfs near the ZZ Ceti instability strip systems whose Balmer-line profiles are contaminated, to
(Gianninas et al. 2006). varying degrees, by the spectrum of the dM secondary.
Ours is not the only survey to have recently looked As a whole, we wish to present a detailed analysis of
at a substantial number of bright white dwarfs. The all the hydrogen-line white dwarfs from MS99 for which
ESO SN Ia Progenitor Survey (SPY, Koester et al. 2001) we have obtained optical spectra. In Section 2 we will
also observed about 800 bright white dwarfs, of which present the spectroscopic content of our survey. Sec-
many were selected from MS99. It is not surprising tion 3 will deal with the model atmospheres used for
then that there is a rather large overlap between the the analysis of our data, and we will present in Section 4
SPY sample and our own. However, in the case of the spectroscopic analysis of our entire sample, including
SPY, the aim of the survey was to obtain multi-epoch, particular analyses for certain subsets of white dwarfs.
high resolution spectra in order to detect the radial ve- Subsequently, in Section 5 we will examine the global
locity variations of double-degenerate binary systems. properties of our sample including the mass distribution
These could then lead to the identification of possible and the mass distribution as a function of effective tem-
SN Ia progenitors, which were ultimately what they perature, and how these compare to other large surveys.
were looking to find. Results of the analysis of the DA We will then revisit the PG luminosity function, and the
white dwarfs from SPY are summarized in Koester et al. empirical determination of the ZZ Ceti instability strip.
(2009). Finley et al. (1997) also presented an analysis Finally, we end with some concluding remarks in Section
of 174 hot (Teff > 25,000 K) DA white dwarfs selected 6.
from MS99 as well as the ROSAT (Pounds et al. 1993)
and EUVE (Bowyer et al. 1994) surveys. Similarly, 2. SPECTROSCOPIC CONTENT
Marsh et al. (1997a,b) and Vennes et al. (1996, 1997, The goal of our survey was to observe all the white
1998) also presented analyses of DA white dwarfs selected dwarfs listed as DA in MS99, including all subtypes
from ROSAT and EUVE while Vennes et al. (2002) an- (DAB, DAH/P, DAO, DAZ, DA+dM), with a visual
alyzed over 200 DA white dwarfs discovered in the 2dF magnitude V ≤ 17.5. Along the way, several DA stars
QSO Redshift Survey (Smith et al. 2005). Other impor- were added to the sample through various other projects
tant studies of large samples of DA white dwarfs include and collaborations the Montreal group was involved in.
the works of Liebert et al. (2005, hereafter LBH05) and In the final reckoning, we had defined a sample of 1503
Limoges & Bergeron (2010) who looked at the complete white dwarfs for which we wanted to obtain high S/N
samples from the Palomar-Green (Green et al. 1986, PG) optical spectra. After nearly 8 years of observing, and
and Kiso (Noguchi et al. 1980; Kondo et al. 1984, KUV) a total of 19 observing runs in both the northern and
surveys, respectively. These studies defined statistically southern hemispheres, we succeeded in securing spectra
significant samples of white dwarfs in an effort to derive for 1360 of the 1503 white dwarfs we had selected. This
the luminosity function of the disk white dwarf popula- translates to a 90% completion rate. Indeed, the major-
tion. Finally, an analysis of over 7000 DA white dwarfs ity of the missing spectra are those of the least bright
from the SDSS Data Release 4 (DR4) was presented in stars in our sample and objects observable only from the
Kepler et al. (2007) with particular attention to the mass southern hemisphere where we were unable to secure ad-
distribution of the sample. Tremblay et al. (2011a) per- ditional telescope time.
formed a new analysis of the aforementioned sample of In all, 726 spectra were obtained specifically for the
DA white dwarfs from the SDSS and showed that for the purposes of this survey since it began in 2003. Of those
analysis of ensemble properties, such as the mass distri- 726 spectra, 568 were obtained between 2003 September
bution, the SDSS data are reliable down to S/N ∼ 15. and 2011 April at Steward Observatory’s 2.3 m telescope
However, the precise determination of the atmospheric equipped with the Boller & Chivens spectrograph. The
parameters for individual stars remains problematic with 4.′′ 5 slit together with the 600 line mm−1 grating blazed
such low S/N observations, as in the case of the ZZ Ceti at 3568 Å in first order provide a spectral coverage from
instability strip (Gianninas et al. 2005).
In an effort to derive the most accurate atmospheric about 3000 to 5250 Å at a resolution of ∼ 6 Å (FWHM).
parameters possible, we also want to incorporate all An additional 18 spectra were obtained in 2006 April
available improvements to the model atmospheres that and 2007 January at the 1.6 m telescope of the Ob-
we use to study these white dwarfs. With that in mind, servatoire du Mont-Mégantic where the 600 line mm−1
we will be using models that include the improved theo- grating provided a spectral coverage from about 3100
retical calculations of the Balmer-line profiles, including to 7500 Å at a similar resolution. Furthermore, a total
non-ideal effects, by Tremblay & Bergeron (2009, here- of 140 spectra from the southern hemisphere were se-
after TB09). Seeing as our spectroscopic technique relies cured during the course of two observing runs in 2007
on a detailed comparison between observed and theoret- March and 2007 October, respectively. First, 84 spec-
ical Balmer lines, it is crucial that we employ the most tra were obtained at ESO’s 3.6 m telescope at La Silla,
accurate and up to date models at our disposal. In the Chile with the ESO Faint Object Spectrograph and Cam-
same vein, two sub-categories of white dwarfs contained era (v.2) (EFOSC2). The #7 grism and a 1.′′ 0 slit pro-
within our sample required particular attention in or- vided a spectral coverage from about 3300 to 5200 Å
der to obtain accurate atmospheric parameters. First, with a resolution of ∼ 6 Å (FWHM). The remaining 56
there are the hot DA and DAO stars exhibiting the so- southern spectra were obtained at Carnegie Observato-
called Balmer-line problem. A comprehensive and de- ries’ 2.5 m Irénée du Pont Telescope at Las Campanas,
tailed analysis of these stars has already been presented Chile with the Boller & Chivens spectrograph. The 1.′′ 5
by Gianninas et al. (2010) and those results will be in- slit with the 600 line mm−1 grating blazed at 5000 Å pro-
A SPECTROSCOPIC SURVEY & ANALYSIS OF BRIGHT, HYDROGEN-RICH WHITE DWARFS 3
Figure 1. Distribution of visual magnitudes V for our sample Figure 2. Distribution of S/N for all the stars in our spectroscopic
selected from MS99 (bold histogram) and for the white dwarfs survey. The distribution peaks at the 50 − 60 bin and more than
for which we have obtained optical spectra (hatched histogram). 50% of the spectra have S/N ≥ 70.
In comparison, the distribution for the SDSS sample as of DR4
(Eisenstein et al. 2006) is also shown (dashed histogram). Note displayed in Figure 2. The S/N2 is determined by an esti-
that the scale is different for the McCook & Sion sample (left) and mate of the rms noise per pixel in the continuum between
the SDSS sample (right) and that we use the SDSS g magnitudes
as equivalent to V . ∼ 4500 and ∼ 4700 Å. Over 52% of our sample have S/N
> 70 and almost 80% above 50. Only the faintest objects
in our sample have S/N that is significantly lower. As
vided a spectral coverage from about 3500 to 6600 Å at such, we have achieved our goal of obtaining high quality
data for the majority of the white dwarfs in our survey.
a slightly better resolution of ∼ 3 Å (FWHM). There
As previously stated, many of these stars have never
are also 169 spectra that were kindly provided by C.
been studied using modern CCD spectroscopy. Conse-
Moran (1999, private communication). Of these, eight quently, it is not surprising to discover that many of the
were obtained at the South African Astronomical Ob-
spectral classifications are erroneous; many of the ob-
servatory’s (SAAO) 1.9 m Radcliffe Telescope using the
jects listed in MS99 are not even white dwarfs. We list
Grating Spectrograph. The remaining 161 spectra were in Table 1 68 objects that are incorrectly classified as
obtained at the Isaac Newton Telescope (INT), on La
white dwarfs in MS99. We provide the spectral classi-
Palma, with the Intermediate Dispersion Spectrograph
fication from MS99 as well as our revised spectroscopic
(IDS). In both cases, the spectra have a resolution of classification for each object along with the appropri-
∼ 3 Å (FWHM). Finally, the sources of the spectra for ate references. Some of these have been known to be
0950+139, and 1136+667 (two DAO white dwarfs) are misclassified for some time but are still listed as white
described in Gianninas et al. (2010). For the remain- dwarfs in the online version of MS99. Of the 68 stars
ing 464 stars, we have used archival spectra that have listed in Table 1, 20 were recently reported as being mis-
been obtained throughout the years by the Montréal classified objects in Limoges & Bergeron (2010) and we
group during the course of various other projects (e.g. refer the reader to Figure 1 of Limoges & Bergeron for
Bergeron et al. 1992, LBH05). These were acquired at the spectra of those stars. In this paper, we report the
Steward Observatory’s 2.3 m telescope with the same in- erroneous classification of an additional 27 objects for
strument and setup described above. the first time. The spectra of these 27 white dwarfs are
We show in Figure 1 the final distribution of visual displayed in Figure 3. Many of these are sdB stars that
magnitudes V for our sample of white dwarfs (see Fig- show strong Balmer lines that are not as broad as those of
ure 1 of Gianninas et al. 2009, as a comparison). We note their white dwarf cousins due to their lower surface grav-
that 10 white dwarfs have no magnitude listed in the lit- ity. However, it is not too hard to see how these objects
erature. The bold histogram shows the distribution for could easily have been mistaken as white dwarfs, espe-
the entire sample while the hatched histogram represents cially using photographic spectra. On the other hand,
the stars for which we have successfully obtained spectra. it is harder to explain how several main sequence stars
We see that the distribution peaks near V = 16.5 and were just as easily misclassified. Indeed, even a quasar
that the majority of our stars have V ≤ 16.5. We con- (0713+745) has found its way into MS99. Furthermore,
trast this result with the sample of DA white dwarfs from MS99 still lists stars like BD+28 4211 (2148+286) and
the SDSS DR4 (Eisenstein et al. 2006). Although the Feige 34 (1036+433), popular spectrophotometric stan-
SDSS sample contains over six times more white dwarfs,
the vast majority of these have 20.5 ≥ g ≥ 18.5, which 2 For the majority of our spectra, obtained at Steward Observa-
makes for a much fainter sample overall. The distribu- tory’s 2.3 m telescope, there are ∼3.2 pixels per resolution element
tion of S/N for our entire sample of 1360 optical spectra is but this value varies for the other instrument setups employed.
dards that are known hot subdwarfs. In addition to the Lacombe et al. 1983) as well as GD 362 (1729+371,
misclassified objects, several stars had two separate en- Gianninas et al. 2004). Although, as previously men-
tries in MS99 under two different names. Although virtu- tioned, GD 362 has since been shown to have a helium-
ally all of these have since been corrected, we report here dominated atmosphere (Zuckerman et al. 2007), a dis-
that at least two still remain. The entries for 0525+271 covery made possible by the detection of a very faint
and 0526+271 refer to the same star, which also hap- helium line at 5877 Å. However, the dominant absorp-
pens to be one of the misclassified objects in Table 1, tion lines in the spectrum, which define the spectral class
and 1100+604 and 1104+602 are also one in the same. of an object, remain the hydrogen and metal lines, as
We also note that 1959+059 is GD 226, not GD 266 as such the spectral type should at most be updated to
listed in MS99 (Giclas et al. 1965). DAZB. In addition, we report for the first time that
In Figure 4 we show a series of representative DA 0543+436 and 1054−226 are also members of the DAZ
spectra in order of decreasing effective temperature, all class. In particular, 1054−226 was initially discovered
of which have 75 ≥ S/N ≥ 65. We have chosen this by Subasavage et al. (2007) but their discovery spectrum
range of S/N values to emphasize the mean quality of lacked the necessary S/N to detect the H and K lines
our data, which is quite high overall. We refer the reader of calcium. In addition, there are eight white dwarfs
to Gianninas et al. (2010, see Figure 1) for a detailed whose spectra are completely devoid of absorption lines
presentation of the DAO white dwarfs contained within in the observed wavelength range. However, with the ex-
our sample. Spectra for the 25 magnetic white dwarfs ception of 0648+368 and 1055−072, the other six cool
in our sample are displayed in Figure 5 as a function DA stars have detections of Hα (Bergeron et al. 2001;
of their continuum slope. In many of these we see the Silvestri et al. 2001), hence their DA spectral type. We
classical Zeeman splitting of the hydrogen lines for some- also note that 0727+482 is actually a double-degenerate
what lower magnetic field strengths on the order of ∼1 to system (Strand et al. 1976). Finally, we also include the
∼10 MG. On the other hand, stars with much stronger peculiar spectrum obtained for 1015+076. It seems that
fields, on the order of 10’s to 100’s of MG, are rendered 1015+076 is a DA white dwarf whose spectrum is com-
almost unrecognizable as DA stars. Of note in Figure 5 pletely dominated by an unresolved, background G dwarf
are possibly four new magnetic white dwarfs. With re- (Farihi et al. 2005). Indeed, the SPY spectrum analyzed
gards to 1018−103 and 1610+330, there is little doubt in Voss (2006) shows a perfectly normal DA white dwarf
as to their magnetic nature. On the other hand, the with Teff ∼ 25,600 K and log g ∼ 7.70.
cases of 0321−026 and 0350+098 are more problematic. Our spectroscopic sample also contains a total of 47
In the case of 0321−026, the Balmer lines predicted by DA+dM binary systems, all of which have been previ-
our best fit model are too deep and we obtain a some- ously identified. In order to obtain accurate atmospheric
what high log g value (see Section 4.6). This is often the parameters for the DA white dwarfs in these binary sys-
case with magnetic white dwarfs as the Zeeman splitting tems, we need to remove the contamination from the
broadens the observed line profiles leading to a higher dM secondary. With that in mind, we have obtained
log g value. We suggest that 0321−026 might harbor follow-up spectra for all the DA+dM systems in our
a weak magnetic field and a spectrum with higher S/N sample, with the exception of five (0208−153, 0309−275,
could reveal weak Zeeman splitting such as that seen in 0419−487, 1541−381, 1717−345), covering a larger spec-
0637+477. Second, 0350+098 was detected by ROSAT tral range from 3500 to 7000 Å. These spectra were ob-
(Fleming et al. 1996) and we therefore conclude that it tained at Steward Observatory’s 2.3 m telescope using
must be a relatively hot star, and the slope would seem the same instrument and setup as described above except
to corroborate that conclusion. Unfortunately, nothing that here we used the 400 line mm−1 grating blazed at
else is known about this object. The absence of any spec- 4800 Å to achieve the desired wavelength coverage. The
tral features could be an indication of a strong magnetic one drawback was that the use of this particular grating
field that has completed diluted any absorption lines that degrades our spectral resolution to ∼ 9 Å (FWHM) from
would otherwise be present. Spectropolarimetric obser-
vations would be needed to confirm the magnetic nature our usual ∼ 6 Å. However, this slightly reduced spectral
of this star. We also note that three of the magnetic resolution is still sufficient for our purposes. We present
white dwarfs are members of double-degenerate binary in Figure 7 what we will refer to as our red spectra for 42
systems. Namely, 0945+245 (LB 11146, Liebert et al. of the 47 DA+dM binaries in our sample. These will be
1993), 1440+753 (RE J1440+750, Vennes et al. 1999), coupled with our usual optical spectra (3000 – 5250 Å),
and 1506+229 (CBS 229, Gianninas et al. 2009). which we will call our blue spectra, for the analysis of the
Finally, in Figure 6 we display spectra for white DA+dM systems. Several of the tell-tale features found
dwarfs of other miscellaneous spectral types included in the spectra of DA+dM binaries can be seen in Fig-
in our sample. This includes five DAB stars, six DAZ ure 7. We see in these spectra how the contribution from
white dwarfs, and eight cool DA. Among the DAB the M dwarf becomes more significant at longer wave-
white dwarfs we find HS 0209+0832 (0209+085) and lengths. More specifically, we see the TiO absorption
GD 323 (1302+597), which have been shown to be spec- band near ∼ 4950 Å that produces the “kink” in the red
troscopically variable (Heber et al. 1997; Pereira et al. wing of Hβ. We also note the presence of the strong Na
2005, respectively), while the other three DAB stars D line near 5895 Å as well as emission, from the M dwarf,
(0339+002, 0453−295, and 1115+166) are actually in the core of several Balmer lines, especially at Hα.
DA+DB binary systems (Limoges & Bergeron 2010; As a summary, we present in Table 2 a breakdown of
Wesemael et al. 1994; Bergeron & Liebert 2002, respec- the white dwarfs in our sample listing the number of stars
tively). Regarding the DAZ white dwarfs, our sample in- in each spectral class.
cludes the first DAZ ever discovered, G74-7 (0208+396,
Table 1
Misclassified Objects
WD Name ST (MS99) ST (this work) Notes

0021−234 Ton S 155 DAWK sdB 1
0031−274 Ton S 163 DA1 sdB 1
0107−342 GD 687 DA3 sdB 1
0109−264 Ton S 201 DA1 sdB 1
0113−243 GD 693 DA6 sdB 2
0154−071 PB 8949 DA1 sdB 1
0200−171 G272-B5B DA sdB 3
0222+314 KUV 02222+3124 DA sdB 2
0240+341 KUV 02409+3407 DA MS(dF) 2
0258+184 PG 0258+184 DA sdB 1
0340−243 Ton S 372 DAWK MS(dF) 3
0341−248 Ton S 374 DA sdB 3
0459+280 CTI 045934.1+280335 DA MS(dF) 3
0509+168 KUV 05097+1649 DA MS(dF) 4
0510+163 KUV 05101+1619 DA MS(dF) 2
0526+271 KUV 05260+2711 DA sdB 2
0529+261 KUV 05296+2610 DA sdB 2
0627+299 KUV 06274+2958 DA sdB 2
0628+314 KUV 06289+3126 DA sdB 2
0713+584 GD 294 DA4 sdB 3
0713+745 Mrk 380 DA QSO 3
0720+304 SA 51-822A DA2 sdB 5
0752+412 KUV 07528+4113 DA? sdB 2
0852+602 PG 0852+602 DA2 sdB 3
0920+366 CBS 98 DA1 sdB 3
0927+388 KUV 09272+3854 DA sdB 2
0930+376 KUV 09306+3740 DA sdB 2
0932+396 KUV 09327+3937 DA sdB 2
0937+395 KUV 09372+3933 DA sdB 2
0939+071 PG 0939+072 DA7 MS(dF) 3
0943+371 KUV 09436+3709 DA sdB 2
0944−090 GD 104 DA1 sdO 6
0946+381 KUV 09467+3809 DA sdB 2
1000+220 Ton 1145 DA6 MS(dF) 3
1008−179 EC 10081−1757 DA1 sdO 7
1036+433 Feige 34 DA sdO 8
1057+307 CSO 64 DA5 MS(dF) 3
1121+145 PG 1121+145 DA1 sdB 9
1133−528 BPM 21065 DA5 MS(dK) 3
1137+311 CSO 105 DA5 MS(dF) 3
1207−032 PG 1207−033 DA4 sdB 1
1214+032 LP 554-63 DA dM 3
1256+286 KUV 12562+2839 DA? sdB 2
1304+313 PB 3322 DA sdB 2
1357+280 CTI 135700.6+280448 DA MS(dF) 3
1412+612 HS 1412+6115 DAH sdO 3
1412−049 PG 1412−049 DA DA+dM 10
1433−270 BPS CS 22874-57 DA sdB 3
1501+300 ... DA MS(dF) 3
1509−028.1 LP 622-13 DA: MS(dF) 3
1514+033 PG 1514+034 DAWK sdB 3
1524+438 CBS 246 DABZ4 sdO 11
1525+433 GD 344 DA sdB 3
1532+239 Ton 241 DA sdB 3
1544+009 LB 898 DA1 sdO 12
1603+175 KUV 16032+1735 DAH? sdO+dK 2
1611+390 KUV 16118+3906 DA sdB 2
1616−390 Ton 264 DA4 MS(dF) 3
1659+442 PG 1659+442 DA sdB 3
2122+157 PG 2122+157 DA sdB 1
2148+286 BD+28 4211 DA sdO 13
2204+071 PG 2204+071 DA sdOB 10
2309+258 KUV 23099+2548 DA sdB 2
2323−241 G275-B16A DA MS(dF) 3
2329−291 Ton S 102 DAWK sdB 1
2333−002 PB 5462 DA2 sdO 14
2343+043 PB 5529 DA sdB 3
2352−255 G275-B17B DA sdB 3
References. — (1) Lisker et al. 2005; (2) Limoges & Bergeron 2010; (3)
this work; (4) Kawka et al. 2004; (5) Finley et al. 1997; (6) Stroeer et al.
2007; (7) Christlieb et al. 2001; (8) Farihi et al. 2005; (9) Green et al. 1986;
(10) LBH05; (11) Eisenstein et al. 2006; (12) Catalán et al. 2008; (13)
Kilkenny et al. 1988; (14) Hügelemeyer et al. 2006.
Figure 3. Optical spectra of stars misclassified as DA white dwarfs in MS99 and reported for the first time in this work. The objects are
approximately ordered as a function of their slope from upper left to bottom right. The spectra have been normalized to a continuum set
to unity and vertically shifted for clarity. 0713+745 (left panel, first object at the top) is a quasar.
Figure 4. Optical spectra for a subsample of the DA white dwarfs from our survey of MS99. The spectra are normalized at 4500 Å and
shifted vertically for clarity. The effective temperature decreases from upper left to bottom right. All the spectra have 75 ≥ S/N ≥ 65.
Figure 5. Optical spectra for 25 magnetic (or suspected magnetic) DA white dwarfs from our survey of MS99. The objects are approxi-
mately ordered as a function of their slope from upper left to bottom right. The spectra are normalized at 4500 Å and shifted vertically
for clarity.
Figure 6. Optical spectra for 20 white dwarfs from our survey of miscellaneous spectral types. The objects are approximately ordered as
a function of their slope from upper left to bottom right. The spectra are normalized at 4600 Å and shifted vertically for clarity.
3. MODEL ATMOSPHERES
Table 2
3.1. Pure hydrogen atmosphere models Breakdown of Spectral Types
For the analysis of the vast majority of white dwarfs
in our sample, we have employed model atmospheres DA DAO DAZ DAB DAmag DA+dM cool DA Total
and synthetic spectra appropriate for DA white dwarfs. 1171 30 6 5 25 47 8 1292
These are described at length in LBH05 and refer-
ences therein. Briefly, these are pure hydrogen, plane- erly fit the spectra of these stars, we will need to com-
parallel model atmospheres where non-local thermody- bine our pure hydrogen models with pure helium mod-
namic equilibrium (NLTE) effects are explicitly taken els. The helium models we have employed are described
into account above Teff = 20,000 K, and energy trans- in Bergeron et al. (2011), which are based on the up-
port by convection is included in cooler models follow- dated grid of white dwarf models described in TB09 in
ing the ML2/α = 0.8 prescription of the mixing-length which the improved Stark profiles of neutral helium of
theory (see Tremblay et al. 2010). The one major up- Beauchamp et al. (1997) have been incorporated. These
grade from the models described in LBH05 is that we models are comparable to those described by Beauchamp
now use the new and improved Stark broadening profiles (1995) and Beauchamp et al. (1996), with the exception
of TB09 that account for nonideal effects directly in the that at low temperatures (Teff < 10,800 K), the free-free
line profile calculation. However, since we will wish to absorption coefficient of the negative helium ion of John
compare our results with those of other large surveys, (1994) is now used. The models are in local thermody-
such as SPY, we have also fit our sample using models namic equilibrium (LTE) and include convective energy
computed with the older broadening profiles of Lemke transport within the mixing length theory. As in the
(1997), and consequently with the ML2/α = 0.6 version analysis of Beauchamp et al. (1999), the parametrization
of the mixing-length theory (Bergeron et al. 1995), since described as ML2/α = 1.25 is implemented.
they were still in use at the time that the SPY analysis 4. SPECTROSCOPIC ANALYSIS
was performed. We can thus compare the independent
results on a level playing field. Beyond that, perform- 4.1. Fitting technique
ing our analysis with both sets of hydrogen line profiles The method used for fitting the observations re-
will also allow us to explore the implications of these lies on the so-called spectroscopic technique developed
improved models on the analysis of DA white dwarfs. by Bergeron et al. (1992), which has been refined by
Bergeron et al. (1995) and more recently by LBH05, and
3.2. Hydrogen atmosphere models with CNO includes the Balmer lines from Hβ to H8. The first step
Although the pure hydrogen models described above is to normalize the flux from an individual line, in both
are sufficient in most cases, some white dwarfs present observed and model spectra, to a continuum set to unity
unique challenges that require somewhat more elaborate at a fixed distance from the line center. The compar-
models. This is the case with the hot DAO white dwarfs ison with model spectra, which are convolved with the
as well as the hot DA stars that exhibit the so-called appropriate Gaussian instrumental profile (3, 6, 9, and
Balmer-line problem (DA+BP), which manifests itself as 12 Å), is then carried out in terms of these line shapes
an inability to fit all the Balmer lines simultaneously with only. The most sensitive aspect of this fitting technique
consistent atmospheric parameters. Though the models is to define the continuum of the observed spectra. The
used to properly fit the spectra of these stars are pre- approach is slightly different depending on the tempera-
sented and described at length in Gianninas et al. (2010), ture range in question. For stars in the interval 16,000
for the benefit of the reader, we briefly explain them here. & Teff & 9000 K, where the Balmer lines are at their
The above mentioned models have all been computed strongest, the normalization procedure involves the use
using the TLUSTY atmosphere code and the accompa- of several pseudo-Gaussian profiles (Bergeron et al. 1995,
nying spectrum synthesis code SYNSPEC, both devel- and references therein). This procedure is quite reliable
oped by Hubeny & Lanz (1995). The models are com- as the sum of the pseudo-Gaussian profiles represents a
puted in NLTE and include carbon, nitrogen, and oxy- good approximation to the observed Balmer lines. Out-
gen (CNO) at solar abundances (Asplund et al. 2005). side of this temperature range, the method becomes more
As Werner (1996) demonstrated, the inclusion of CNO unstable as the continuum between the Balmer lines be-
reduces the temperature in the upper layers of the at- comes essentially linear. Consequently, for stars with Teff
mosphere, which in turn produces deeper cores for the > 16,000 K and Teff < 9000 K, we rely on our synthetic
hydrogen Balmer lines, thus overcoming the Balmer-line spectra to reproduce the observed spectrum, including
problem. As clearly stated by Gianninas et al. (2010), a wavelength shift, as well as several order terms in λ
CNO acts as a proxy for all metals that may be present (up to λ6 ) using the nonlinear least-squares method of
in the atmosphere. As such, the inclusion of CNO at Levenberg-Marquardt (Press et al. 1986). The normal
solar abundances should in no way be considered as a points are then fixed at the points defined by this smooth
determination of the abundance of CNO in these stars. model fit. Note that the values of Teff and log g at this
The model atmospheres used for the DAO and DA+BP stage are meaningless since too many fitting parameters
stars are identical, save for the inclusion of helium in the are used, and the model just serves as a smooth fitting
DAO models in order to reproduce the He ii λ4686 line. function to define the continuum of the observed spec-
trum. Once the lines are normalized to a continuum set
3.3. Helium atmosphere models to unity, we use our grid of model spectra to determine
In Figure 6, we presented three DA+DB binary sys- Teff and log g in terms of these normalized profiles only.
tems that are included in our sample. In order to prop- Our minimization technique again relies on the nonlinear
Figure 7. Optical spectra for 42 DA+dM binary systems covering the range 3500–7000 Å. The spectra are normalized to unity at 6500
Å and shifted vertically for clarity.
least-squares method of Levenberg-Marquardt, which is cooling sequence.
based on a steepest descent method. Sample fits of 20 In Figure 9, we present the comparison between the at-
DA white dwarfs, covering almost the entire tempera- mospheric parameters measured using the older Lemke
ture range of our sample are displayed in Figure 8. We (1997) and the new TB09 Stark profiles for the 1151 DA
see that the combination of high S/N spectra with the stars in our sample. In the lower panel, we see that
procedures described above allow us to achieve a proper the correlation between the old and new values of Teff is
normalization in each case. Hence, we are able to obtain very tight although the overall trend is to higher temper-
an excellent agreement between the observed and pre- atures. This effect is somewhat more evident at higher
dicted line profiles for nearly every star. The spectrum Teff . In the upper panel of Figure 9, we compare the old
of one star, 0342+374, features emission lines from its and new values of surface gravity. Unlike with the values
associated planetary nebula. These emission lines can of Teff we note here a much more obvious shift toward
interfere with our normalization procedure but also with higher measurements of log g when using the new TB09
our fitting technique and so we simply exclude the af- Stark broadening profiles. These results are consistent
fected spectral ranges from both the normalization and with those presented in TB09.
fitting procedure. In Figure 10 we examine how these differences in the
measured atmospheric parameters translate to the deter-
4.2. Effects of new Stark profile calculations mination of the mass distribution as a function of Teff .
As we have mentioned, in this study we will be passing Using the old profiles, we see in the top panel of Fig-
from the use of the old Lemke (1997) profiles to the new ure 10 that besides the high-log g problem at lower tem-
TB09 Stark broadening profiles that include nonideal ef- peratures (Tremblay et al. 2010, and Section 5.5 below),
fects directly in the line profile calculation. In doing the mass distribution dips near 13,000 K. On the other
so, it is important to understand what effects these new hand, the new profiles seem to even out the mass distri-
profiles will have on our atmospheric parameter deter- bution with the bulk of the stars showing masses between
minations. TB09 explored some of these differences (see 0.55 M⊙ and 0.7 M⊙ before the high-log g problem takes
their Figure 12) using the PG sample of LBH05, but this hold at lower Teff . As such, the new profiles permit an im-
limited them to a particular range of Teff whereas we can proved determination of the mass distribution as a func-
examine the effects along virtually the entire white dwarf tion of Teff over the older models that employ Lemke’s
Figure 8. Model fits (red) to the individual Balmer line profiles (black) of a sample of DA white dwarfs taken from Figure 4 in order of
decreasing effective temperature. The lines range from Hβ (bottom) to H8 (top), each offset by a factor of 0.2. The best-fit values of Teff
and log g are indicated at the bottom of each panel.
Figure 11. Model fits (red) to the optical spectra (black) for the
three DA+DB binary systems in our sample. The atmospheric
parameters for each solution are given in the figure. Both the
observed and theoretical spectra are normalized to a continuum
set to unity and the spectra are vertically shifted for clarity.
about 13,000 K where the high-log g problem begins to
manifest itself.
In the following sections we will focus on the spectro-
scopic analysis of all the objects and systems that require
particular attention. These include white dwarfs of spec-
Figure 9. Comparison of log g (top) and Teff (bottom) values tral types DAB, DAZ, DA+dM binary systems, magnetic
derived from fits using the improved models of TB09 and the VCS white dwarfs, and a selection of other unique objects that
profiles of Lemke (1997) for 1151 DA white dwarfs. The dotted demanded a more detailed analysis. The motivation here
lines represent the 1:1 correlation.
is to ensure that we determine the most accurate atmo-
spheric parameters possible for each star. With that in
mind, we also note that from this point forward, we will
be using the new Stark profiles of TB09 exclusively.
4.3. DAB white dwarfs

In Figure 6 we displayed five white dwarfs classified
as DAB stars. In other words, besides the hydrogen
Balmer lines, the spectra of these stars also feature
lines due to neutral helium. For three of these ob-
jects (0339+002, 0453−295, 1115+166), previous anal-
yses have shown that fits assuming a single star do not
produce satisfactory results. In contrast, fits performed
under the assumption that the spectrum is that of a
DA+DB double-degenerate binary system have proven
successful (Limoges & Bergeron 2010; Wesemael et al.
1994; Bergeron & Liebert 2002, respectively). With that
in mind, we proceed to fit the spectra of these systems
by combining our usual pure hydrogen models with the
pure helium models described in Section 3.3. When fit-
ting DA+DB model spectra, the total flux of the sys-
Figure 10. Mass vs. Teff distribution for 1151 DA white dwarfs tem is obtained from the sum of the monochromatic Ed-
in our sample using the Stark broadening profiles of Lemke (1997)
(top) and the improved calculations of TB09 (bottom). Lines of dington fluxes of the individual components, weighted
constant mass at 0.55 M⊙ and 0.7 M⊙ are shown as a reference. by their respective radius. The stellar radii are ob-
tained from evolutionary models similar to those de-
Stark broadening profiles. Specifically, the TB09 profiles scribed in Fontaine et al. (2001) but with C/O cores,
produce a relatively constant mass distribution down to q(He) ≡ log MHe /M⋆ = 10−2 and q(H) = 10−4 , which
of Teff and log g than Bergeron & Liebert (2002) for the
DA component but identical parameters for the DB.
The case of HS 0209+0832 is different from the above
three systems as it is a true DAB white dwarf, as first
reported by Jordan et al. (1993). In other words, it is
a single star with both hydrogen and neutral helium
lines present in its optical spectrum. Furthermore, like
GD 323, this star presents neutral helium lines that are
variable, in particular the He i λ4471 line (Heber et al.
1997). However, contrary to GD 323, Wolff et al. (2000)
were able to explain the variability as resulting from the
accretion of pockets of interstellar matter of varying den-
sities leading to different accretion rates over the course
of time. We show in Figure 12 the best fit to our optical
spectrum of HS 0209+0832. Besides the usual Balmer
lines, we also fit the He i λ4471 line in order to mea-
sure the helium abundance. To perform the fit, we have
used the grid of homogeneously mixed H/He models de-
scribed by Gianninas et al. (2010) for DAO white dwarfs,
with some modifications. Namely, we have extended the
grid down to Teff = 30,000 K. However, we do not in-
clude CNO in the models as there is no evidence of the
Balmer-line problem in HS 0209+0832.
Our new determinations (Teff = 36,550, log g = 8.00)
are quite comparable to the previous measurements
of Heber et al. (1997, Teff = 36,100, log g = 7.91) and
Wolff et al. (2000, Teff = 35,500, log g = 7.90) with our
Figure 12. Model fits (red) to the hydrogen Balmer lines and higher values likely due to the use of the new TB09 Stark
the He i λ4471 line in the observed optical spectrum (black) of broadening profiles. Conversely, we determine a much
the DAB white dwarf HS 0209+0832. The lines range from Hβ lower helium abundance of log He/H = −3.04 as com-
(bottom) to H8 in addition to He i λ4471 (top), each offset by a
factor of 0.2. The best-fit values of Teff , log g, and log He/H are pared to the values compiled in Table 2 of Wolff et al.
indicated at the bottom of the figure. (2000) that hover around log He/H ∼ −2.0 with the low-
est value at log He/H = −2.32 (Heber et al. 1997). It
are representative of hydrogen atmosphere white dwarfs, is important to note that our spectrum was obtained in
and q(He) = 10−2 and q(H) = 10−10 , which are repre- 2002 December, more than three years after the last ob-
sentative of helium atmosphere white dwarfs. servation listed in Table 2 of Wolff et al. (2000). Further-
The fitting technique for these DA+DB systems first more, Wolff et al. state that “A lower rate would reduce
requires that we normalize both the observed and syn- the abundance within a few months due to the short dif-
thetic spectra to a continuum set to unity. The calcula- fusion timescale”. Hence, the most logical explanation
tion of χ2 is then carried out in terms of these normalized for our considerably lower helium abundance determi-
line profiles only. The atmospheric parameters, Teff and nation is that in the intervening time, HS 0209+0832
log g, for the DA white dwarf and Teff for the DB white traversed a region of space where the helium density was
dwarf are considered free parameters. We set log g = lower, leading to a reduced accretion rate.
8.0 for the DB components in 0339+002 and 0453−295 Finally, we do not attempt to analyze GD 323. In-
since the helium lines are rather weak and are not espe- deed, to date there has not been a model that has been
cially sensitive to the surface gravity. The helium lines able to achieve a satisfactory fit to the optical spectrum
are comparatively stronger in 1115+166 so for the fit of of GD 323 despite many attempts throughout the years.
this star, we allow log g to vary. We refer the reader to Pereira et al. (2005) for a compre-
Our best fits to the spectra of 0339+002, 0453−295, hensive summary of the various models that have been
and 1115+166 are shown in Figure 11. We obtain excel- elaborated over the years in order to explain both the
lent fits for all three systems with the fit for 1115+166 be- optical spectrum of GD 323, and the observed spectral
ing exceptionally good. For 0339+002, we obtain higher variability.
values of Teff and log g for the DA component as com-
pared to the results of Limoges & Bergeron (2010) with 4.4. DAZ white dwarfs
a comparable value of Teff for the DB component. This Our sample contains the six DAZ white dwarfs shown
is not an altogether surprising result for the DA star in Figure 6. With the exception of GD 362, we fit the
since we are using the new profiles of TB09 in our analy- remaining five DAZ white dwarfs to obtain Teff , log g, as
sis. With regards to 0453−295, we obtain a substantially well as the calcium abundance. In order to do so, we
higher Teff and log g for the DA component as compared computed a small grid of synthetic spectra based on our
to Wesemael et al. (1994) and also a higher Teff for the grid of pure hydrogen atmospheres and added calcium
DB white dwarf. However, the models used in their anal- in the calculation of the synthetic spectrum only, as de-
ysis are considerably outdated and they also performed scribed in Billères et al. (1997). The inclusion of calcium
a four parameter fit allowing the surface gravity of the in the calculation of the synthetic spectrum alone, and
DB star to vary. Finally, we get somewhat higher values not in the model atmosphere itself, is justified since the
Figure 13. Model fits (red) to the hydrogen Balmer lines and the Ca H & K lines in the observed optical spectra (black) of five DAZ
white dwarfs. The lines range from Hβ (bottom) to H8 (top), each offset by a factor of 0.2. We note that the Ca H line is blended with
Hǫ. The best-fit values of Teff , log g, and log Ca/H are indicated at the bottom of each panel.
Figure 14. Same as Figure 8 for five DA+dM binary systems for which we do not have red spectra. In these cases, Hβ (green) is omitted
from the fitting procedure as the only contaminated spectral line.
Figure 15. Same as Figure 8 for four additional DA+dM binary systems where little to no contamination is detected in their blue spectra.
trace amounts of calcium will have no effect whatsoever

on the thermodynamic structure of the atmosphere (see
Gianninas et al. 2004). Our grid covers Teff from 6000 to
9000 K in steps of 500 K, log g from 7.0 to 9.5 in steps
of 0.5 dex and log Ca/H from −7.0 to −9.5 in steps of
0.5 dex. Although our standard spectroscopic technique
works well even without the inclusion of calcium in the
synthetic spectra, the fact is that the calcium H line (at
3968 Å) is blended with Hǫ (at 3970 Å). Since the upper
Balmer lines are especially sensitive to the surface grav-
ity, any failure to properly model the line profile would
lead to a less precise measurement of log g. Therefore, we
are compelled to include the calcium lines in our spectra
to avoid this potential source of uncertainty.
Our best fits of the five DAZ stars using the above
grid are displayed in Figure 13. We have employed es-
sentially the same fitting procedure used for our regular
DA white dwarfs but here we have three free parameters:
Teff , log g, and the calcium abundance. Consequently, we
have extended the fitting region around Hǫ blueward in
order to properly include the Ca K line in the fit. We
can see that we are able to reproduce the calcium lines Figure 16. Best fits of the observed red spectra (black) for four
of our DA+dM binary systems using our combination of synthetic
rather well in all cases. Furthermore, we can compare the DA spectra and dM spectral templates (red). The spectra are
atmospheric parameters of the three known DAZ stars, normalized to unity at 5100 Å and shifted vertically for clarity.
G74-7 (0208+396), LTT 13742 (1257+278), and G180-
43 (1633+433), to previous determinations. Specifically, and we obtain excellent fits to their observed line pro-
Billères et al. (1997) obtained Teff = 7260 K, log g = 8.03, files without the need to omit any lines. First, the red
and log Ca/H = −8.80 for G74-7, and Zuckerman et al. spectrum of 0116−231 shows virtually no contamination
(2003) obtained Teff = 8481 K, log g = 7.90, and log from a companion, as we can see in the left panel of Fig-
Ca/H = −8.07 for LTT 13742, and Teff = 6569 K, log g ure 7. Similarly, the red spectrum of 0308+096 is also
= 8.08, and log Ca/H = −8.63 for G180-43. These devoid of contamination except for emission at Hα. In
values are quite comparable to the values we measure contrast, the blue spectrum 0232+035 seems devoid of
here. Indeed, some variation is expected considering any contamination while its red spectrum has emission
our models contain several improvements over the ones lines in the cores of all the Balmer lines. Finally, the blue
used in the previous analyses. With respect to the two spectrum of 1247−176 shows weak emission for some of
newly identified DAZ stars, G96-53 (0543+436) and LP the Balmer lines while we see strong emission in the core
849-31 (1054−226), their atmospheric parameters seem of all the Balmer lines in its red spectrum (see the middle
to suggest that they are rather typical of DAZ white panel of Figure 7). In the first two cases, there is insuffi-
dwarfs within the same temperature range (see Figure 5 cient contamination in the blue spectra to warrant using
of Zuckerman et al. 2003). As new detections, these last the procedure described below to subtract the contami-
two DAZ stars would be prime candidates for harbor- nation due to the secondary. In the latter two cases, the
ing circumstellar dust disks as the source of the metals red spectrum is considerably more contaminated than
present in the atmosphere. Infrared observations would the blue spectrum. As such, our procedure would likely
be required in order to ascertain, through the detection only make matters worse. Furthermore, these seemingly
of an infrared excess, whether a circumstellar debris disk contradictory detections and non-detections of emission
is indeed present around these white dwarfs (Kilic et al. lines from the M dwarf secondary in the latter two cases
2006). are likely due to the fact the red and blue spectra were
obtained at different phases of the binary orbit. Finally,
as described above for 0342+374, we exclude the cen-
4.5. DA+dM binary systems tre of each Balmer line from the fitting procedure for
As stated in Section 2, we have a total of 47 DA+dM 0308+096, 1247−176, and 1717−345, as they are con-
binary systems in our sample. For nine of those systems, taminated by emission lines from their M dwarf compan-
we employ our standard fitting technique, as described ions.
above. The fits for these nine systems are displayed in For the remaining 38 systems, we will take advan-
Figures 14 and 15. Of these nine, five are the DA+dM tage of the red spectra we obtained (see Figure 7) in
systems, listed earlier, for which we were unable to se- order to remove the contamination due to the presence
cure red spectra (see Section 2). In these cases, we simply of the M dwarf. There have been several studies in
omit Hβ, the only contaminated Balmer line, from our the last few years that have analyzed DA+dM systems
fitting procedure and fit only the Balmer lines from Hγ from the SDSS (Silvestri et al. 2006; Heller et al. 2009;
to H8. We see in Figure 14 that we are able to achieve a Rebassa-Mansergas et al. 2010). One important distinc-
more than satisfactory fit to the remaining four Balmer tion between these studies and our own is that in the
lines. On the other hand, the remaining four DA+dM above cases the authors were interested in recovering
systems, shown in Figure 15, display little to no con- both the spectrum of the white dwarf and the spectrum
tamination from their companion in their blue spectra of the M dwarf in an effort to study the system as a
whole and obtain atmospheric parameters for both stars.

In contrast, we are only interested in obtaining a more Table 3
Parameters for Magnetic White Dwarfs
accurate measurement of the parameters for the DA star.
Our fitting procedure for these systems combines the WD Name Bp (MG) Teff (K) log g Notes
synthetic spectra computed from our pure hydrogen
model atmospheres with the M dwarf spectral templates 0253+508 KPD 0253+5052 17 15,000 8.00 1
0329+005 KUV 03292+0035 12.1 26,500 ... 2
compiled and presented by Bochanski et al. (2007). We 0553+053 LTT 17891 20 5500 8.00 3
attempt to fit a function of the form 0816+376 GD 90 9 14,000 8.00 4,5
0945+245 LB 11146a 670 14,500 8.50 6,7,8
1017+366 Ton 1206 65 16,000 ... 9
Fobs = [FDA (Teff , log g) + a1 · FdM (Sp. type)] 1031+234 Ton 527 1000 15,000 8.50 10
1440+753 RE J1440+750 15 42,000 8.80 6,11
× (a2 + a3 λ + a4 λ2 + a5 λ3 ) 1506+399 CBS 229 20 17,000 ... 6,12,13
1533−057 PG 1533−057 31 20,000 8.50 1,14
to the observed spectrum. FDA is the flux due to the DA 1639+537 GD 356 13 7510 8.14 15,16
1900+705 LFT 1446 320 16,000 8.58 15
white dwarf and depends on Teff and log g while FdM , the 2316+123 KUV 23162+1220 45 11,000 8.00 3
flux from the dM component, is a function solely of the
M dwarf spectral type, from M0 to M9. The remaining References. — (1) Achilleos & Wickramasinghe 1989; (2)
Gänsicke et al. 2002; (3) Putney & Jordan 1995; (4) Putney 1997;
free parameters are a1 , which is the relative contribution (5) Martin & Wickramasinghe 1984; (6) Double-degenerate; (7)
of the synthetic DA spectrum and the M dwarf spectral Liebert et al. 1993; (8) Glenn et al. 1994; (9) Saffer et al. 1989; (10)
template, a2 , a scaling factor between the composite syn- Schmidt et al. 1986; (11) Vennes et al. 1999; (12) Gianninas et al. 2009;
(13) Vanlandingham et al. 2005; (14) LBH05; (15) Bergeron et al. 2001;
thetic spectrum and the observed spectrum, and the co- (16) Ferrario et al. 1997.
efficients a3−5 of a third order polynomial that is meant
to account for errors in the flux calibration. In total,
there are eight free parameters when we fit the compos- Hβ and Hγ, we are now able to fit them quite well with
ite spectra. This means that the values of Teff , log g, the contamination removed. The log g value is virtually
and spectral type that we obtain are meaningless since identical whereas we determine a Teff that is over 2000 K
there are too many parameters used in the fitting pro- lower. Finally, we note the greatly improved agreement
cess. However, as we mentioned above, our only interest between the slope of the observed spectrum and the spec-
is to obtain the necessary function to subtract from the troscopic solution.
composite spectrum in order to remove the dM contam-
ination. We show in Figure 16 four examples of the fits 4.6. Magnetic and cool white dwarfs
we obtain to our composite DA+dM spectra. We see In Figure 5 we presented the spectra of 25 magnetic, or
that we are able to achieve some very good fits to the suspected to be magnetic, white dwarfs that are found in
observed spectra. It is interesting to note the varying our sample. For most of these white dwarfs, our standard
degrees of contamination from one system to the next. spectroscopic technique is inadequate since the Balmer
For example, 1443+336 shows only light contamination line profiles are severely distorted and/or completely de-
in contrast to 1132−298 where the white dwarf is nearly stroyed by a strong magnetic field. However, some of
drowned out by the flux from its M dwarf companion. the more weakly magnetic stars can be fit with our spec-
Indeed, in the case of 1132−298, and several other sys- troscopic method. We note that we do not exclude any
tems, when letting all the free parameters listed above part of the observed line profiles from our fitting pro-
vary, the fitting procedure produced rather poor results. cedure. Consequently, the values of Teff and log g de-
In cases such as those, we forced Teff and log g to the rived from these fits can only be considered, at best, as
values obtained from fitting our blue spectra of the same rough estimates of the atmospheric parameters. In Fig-
objects, similar to the fits presented in Figure 14, and this ure 18 we present the fits for nine weakly magnetic white
allowed us to achieve satisfactory results as evidenced by dwarfs. For 1658+440, 1220+234, 1018−103, 0637+477,
1132−298. 2329+267, and 1503−070, the Zeeman splitting of the
Once the spectroscopic fit to the composite spectrum Balmer lines is fairly obvious. In the cases of 1350−090
is obtained, we subtract from the spectroscopic solution and 2359−434, the spectroscopic solution predicts the
the contribution of the DA component [FDA (Teff , log g)]. Balmer lines to be deeper than observed, this is also
In other words, we subtract a synthetic DA spectrum evidence of a weak magnetic field, which is indepen-
interpolated to the values of Teff , and log g, as determined dently confirmed through spectropolarimetric measure-
by the fit to the composite spectrum, from the composite ments (Schmidt & Smith 1994; Aznar Cuadrado et al.
spectroscopic solution. The residual spectrum generated 2004, respectively). Finally, the reason why we clas-
by this procedure represents the contribution of the M sified 0321−026 as magnetic becomes evident here as
dwarf only. This residual spectrum is then interpolated the Balmer lines are again predicted to be too strong.
onto the wavelength grid of the blue spectrum of the same We note that all nine stars have log g values which
object. Finally, we subtract the residual spectrum from are substantially higher than the canonical value of
the blue spectrum removing the contamination from the log g = 8.0, this is not in the least surprising since
companion. it is known that magnetic white dwarfs have masses
The next step is to fit the corrected blue spectrum with which are higher than the average for normal DA stars
our standard fitting technique. We show in Figure 17 an (Wickramasinghe & Ferrario 2000). When trying to fit
example of the fits for 1643+143 before and after the these magnetic stars with non-magnetic models, this phe-
correction has been applied. We see quite a dramatic nomenon is exacerbated by the Zeeman splitting which
change here. Indeed, where we had to omit the fits to broadens the lines even further causing our spectroscopic
Figure 17. Comparison of the spectroscopic solutions for 1643+143 before (top) and after (bottom) subtraction of the dM component.
The left panels show the fits (red) to the observed Balmer lines (black) from Hβ to H8 where fits in green indicate spectral lines that are
not taken into account in the fitting procedure. The right panels compare the observed spectrum (black) with a synthetic spectrum (red)
interpolated at the values of Teff and log g obtained from the spectroscopic fit, both are normalized to unity at 4600 Å. Note the scales for
the relative flux are different in the left and right panels.
solutions to require even higher log g values. There are
no better examples of this effect than 1658+440 and Table 4
Parameters for Cool White Dwarfs
1503−070 whose log g values have actually been extrap-
olated outside of our model grid. WD Name Teff (K) log g M /M⊙ MV Notes
The 16 remaining magnetic white dwarfs are simply
beyond the scope of this work. The same can be said of 0121+401 G133-8 5340 7.90 0.52 14.65 1
0213+427 G134-22 5600 8.12 0.66 14.72 1
the cool white dwarfs shown in the right panel of Fig- 0551+468 LP 159-32 5380 8.01 0.59 14.78 1
ure 6. For this reason, these white dwarfs will not be 0648+368 GD 78 5700 ... ... ... 2
further analyzed here. However, in the interest of com- 0727+482A G107-70A 5020 7.92 0.53 15.03 1,3
pleteness, we have searched throughout the literature in 0727+482B G107-70B 5000 8.12 0.66 15.33 1,3
1055−072 LFT 753 7420 8.42 0.85 13.91 1
order to compile atmospheric parameter determinations, 1750+098 G140-B2B 9527 ... 1.17 ... 4
and other derived quantities, for these white dwarfs. 1820+609 G227-28 4780 7.83 0.48 15.16 1
First, Table 3 lists the properties for 13 of the 16 remain-
DAZ
ing magnetic white dwarfs, including the magnetic field
strength (assuming a magnetic dipole), Teff , and log g, 1729+371 GD 362 10,540 8.24 ... ... 5
where available, along with the references to the original References. — (1) Bergeron et al. 2001; (2) Angel et al. 1981; (3)
analyses. No information could be found for 0350+098, Double-degenerate; (4) Silvestri et al. 2001; (5) Zuckerman et al. 2007.
and 1610+330 is identified as a magnetic white dwarf for
the first time in this work. Finally, 0239+109 is ana-
lyzed at length in the next section. Similarly, Table 4 necessary references. We also include in Table 4 the pa-
lists the atmospheric parameters for all the cool white rameters for GD 362 from Zuckerman et al. (2007) as we
dwarfs in our sample. Specifically, we indicate Teff , log g, have not computed appropriate helium models, which
mass, and the absolute visual magnitude along with the include the necessary metals, for a proper re-analysis of
this particular object.
Figure 19. Model fits (red) to the observed Balmer lines (black)
for the three peculiar white dwarfs in our sample using the hot
seed (top) and the cold seed (bottom) for the determination of
the spectroscopic solution. The atmospheric parameters for each
solution are given in the figure. Both the observed and theoretical
spectra are normalized to a continuum set to unity and the lines
range from Hβ (bottom) to H8 (top), each offset by a factor of 0.2.
Figure 18. Same as Figure 8 for nine weakly magnetic DA white
dwarfs.
We note that from here on, we shall not take into con-
sideration the values compiled in the two tables described
above in our analysis of the global properties of our spec-
troscopic sample. We wish to include only white dwarfs
whose atmospheric parameters were determined directly
by us, as we endeavor to maintain the homogeneity of
our analysis.
4.7. Peculiar Objects

In the following section, we will take a closer look
at three white dwarfs (0239+109, 0737−384, and
0927−173) whose analysis demanded particular atten-
tion in order to be certain we are obtaining the correct
values for their atmospheric parameters. In all three
cases, although the hot solution seemed to be the cor-
rect choice, based on our Balmer-line fits, the slopes of
the corresponding spectroscopic solutions proved incom-
patible with those of the observed spectra. The fits for
the three stars in question are presented in Figure 19.
We see that for all three stars, the hot solutions (top
panels) seem to match the observed spectra rather well,
whereas the cold solutions are completely incompatible
with the data. We now take a closer look at each object
in turn.
4.7.1. 0239+109 (G4-34)
The star 0239+109 is actually one of the 25 magnetic Figure 20. Fits to the energy distribution of 0239+109 and
white dwarfs we presented in Figure 5, however we will 0927−173 with pure hydrogen models. The BV I (BV only for
see that this case is not quite so simple. This star was 0927−173) and JHKS photometric observations are represented
analyzed in Bergeron et al. (1990a) and they had sim- by error bars, while the model fluxes are shown as filled circles.
The atmospheric parameters corresponding to the photometric so-
ilar difficulties in properly fitting the Balmer-line pro- lution are given in the figure.
files (see their Figure 1). To be more exact, the slope of
their observed spectrum clearly suggested a cooler spec-

troscopic solution despite the fact that the hot solution
produced the better fit to the Balmer-line profiles. As a
result, their conclusion was that 0239+109 must in fact
be an unresolved DA+DC binary system. More recently,
this star was observed as part of the SPY survey and
Koester et al. (2009) classified it instead as a magnetic
white dwarf. Indeed, the SPY spectrum shown in Fig-
ure 7.3 of Voss (2006) shows what seems to be Zeeman
splitting of both Hα and Hβ. Our spectrum of 0239+109
is one of the many provided to us by C. Moran, and as
a result, we have no information regarding the observa-
tions (airmass, etc.). However, as we mentioned earlier,
when trying to fit the Balmer-line profiles of this star, we
encounter the same incoherence noted by Bergeron et al.
(1990a), i.e. the hot solution produces the better fit to
the observed Balmer lines, but the slope of the spectro-
scopic solution is incompatible with that of our observed Figure 21. Model fit (red) to the optical spectrum (black) for
spectrum, which suggests a much cooler temperature. the DA+DC binary 0239+109. The atmospheric parameters cor-
Furthermore, we have in our archives the older spectrum responding to the spectroscopic solution are given in the figure.
used in the Bergeron et al. (1990a) analysis and obtained Both the observed and theoretical spectra are normalized to a con-
tinuum set to unity.
by Greenstein (1986). The slopes of the two independent
observations are in perfect agreement so we are confident
that there is nothing wrong with the data.
In order to get an independent estimate of Teff , we
exploit the available photometry and fit the overall
optical-near infrared spectral energy distribution com-
bining the BRI photometry from the USNO-B Catalog
(Monet et al. 2003) with the available Two Micron All
Sky Survey (2MASS, Cutri et al. 2003) JHKs magni-
tudes. Synthetic colors are obtained using the proce-
dure outlined in Holberg & Bergeron (2006) based on
the Vega fluxes taken from Bohlin & Gilliland (2004).
The method used to fit the photometric data is de-
scribed in Bergeron et al. (2001). Since we have no par-
allax measurement, we assume log g = 8.0. The best
fit to the photometry is presented in the top panel of
Figure 20. Although the photometric fit is not per-
fect, overall we see that this must be a cool object with
Teff ∼ 8600 K. To reconcile the discrepancy between the
slopes of the hot solution and that of our optical spec-
trum, taking into account the Teff value from the photo- Figure 22. Relative energy distributions for our best composite
metric fit, the only possibility is that there are two stars DA+DC fit displayed in Figure 21. The dashed line represents the
instead of one. Hence, we believe that in the end, both contribution of the DA component while the dotted line represents
the contribution of the DC component, both properly weighted
Bergeron et al. (1990a) and Koester et al. (2009) were by their respective radius. The red line corresponds to the total
right. In other words, the system is an unresolved double- monochromatic flux of the composite system superimposed on our
degenerate binary composed of a magnetic DA star and spectrum of 0239+109 (black line), which has been scaled to the
a cooler DC white dwarf, a system analogous to G62-46 flux of the composite model at 4600 Å.
(Bergeron et al. 1993). In order to get a more accurate
measure of the atmospheric parameters for 0239+109, 4.7.2. 0737−384 (NGC 2451-6)
we use a procedure identical to the one employed for the This white dwarf is a supposed member of the open
DA+DB binary systems but here we assume a DA and cluster NGC 2451 as first reported by Koester & Reimers
a DC component. Figure 21 shows our best fit to the (1985). However, this claim was later challenged by
composite spectrum and our results are very similar to Platais et al. (2001) based on the proper motion of the
those obtained by Bergeron et al. (1990a). As a final star as compared to the proper motion of the cluster.
check, we plot in Figure 22 the model flux for the com- The analysis of Koester & Reimers (1985) yielded Teff =
bined DA+DC spectroscopic solution and compare the 31,000 ± 3000 K, which would seem to concur with our
slope with that of our optical spectrum of 0239+109 and determination assuming the hot solution is indeed the
see that the agreement is nearly perfect. We note, how- correct one. Unfortunately, the slope of the hot solu-
ever, that despite the excellent agreement between our tion does not match the slope of our observed spectrum,
solution and the data, the magnetic nature of the DA which suggests a much cooler temperature for this star.
component means that our measured values of Teff and How can this clear discrepancy between the slopes of our
log g for this star should be considered approximate. observed spectrum and that of the hot solution be ex-
plained? First, we verified our observing logs to see if
Figure 24. Model fit (red) to the optical spectrum (black) for
Figure 23. Comparison of the observed spectrum (black) with the DA+DC binary 0927−173. The atmospheric parameters cor-
the synthetic spectrum (red) corresponding to the hot solution pre- responding to the spectroscopic solution are given in the figure.
sented in Figure 19 for 0737−384. The observed spectra have been Both the observed and theoretical spectra are normalized to a con-
corrected for interstellar reddening with values of E(B − V ) as in- tinuum set to unity.
dicated in the figure. Both the observed and synthetic spectra are
normalized at 4600 Å and offset vertically for clarity.
our observations were obtained at high airmass. This

could lead to a significant loss of flux in the blue por-
tion of the spectrum, due to atmospheric extinction, if
the instrument is not properly rotated to match the par-
allactic angle. As it turns out, 0737−384 was observed
at an airmass of z = 1.08 so any such effects would be
minimal and we discount that possibility. Next, we at-
tempted to fit 0737−384 as a DA+DC system, but the
slope of the combined model does not match the slope
of the observed spectrum, as was the case in the previ-
ous section with 0239+109, thus a different solution is
required to explain the discrepancy. In the analysis of
Koester & Reimers (1985), a value of E(B − V ) = 0.05
is adopted for the reddening toward NGC 2451 based on
several earlier studies (see references therein). Hence, we
speculated that our spectrum might actually be suffering
from interstellar reddening. We used the above value of
interstellar reddening to correct our observed spectrum, Figure 25. Relative energy distributions for our best composite
following the prescription of Seaton (1979), but the ef- DA+DC fit displayed in Figure 24. The dashed line represents the
fect is negligible as we see in the top panel of Figure 23. contribution of the DA component while the dotted line represents
the contribution of the DC component, both properly weighted
However, when querying the NASA/IPAC Infrared Sci- by their respective radius. The red line corresponds to the total
ence Archive’s Dust Extinction maps3 in the direction monochromatic flux of the composite system surperimposed on our
of NGC 2451, we get a value of E(B − V ) = 0.684. spectrum of 0927−173 (black line), which has been scaled to the
This is significantly larger than the previously deter- flux of the composite model at 4600 Å.
mined values. Alternatively, if we enter the coordinates
for 0737−384, we obtain E(B − V ) = 0.465. This would 4.7.3. 0927−173 (LP 787-49)
seem to suggest that the extinction in the direction fo As in the previous case, our first instinct was to ver-
the cluster is quite patchy. For our purposes, we adopt ify if there might be a problem with our observations of
the value of E(B − V ) = 0.465. When we apply this new 0927−173. As it turns out, 0927−173 was observed at
value of reddening to our data we see, in the bottom of an airmass of z = 1.54, which puts it at much greater
Figure 23, that the slopes are now in much better agree- risk of suffering from atmospheric extinction. However,
ment. We therefore conclude that the hot solution for other objects observed on the same night and at higher
0737−384 is indeed the correct one and the discrepancy airmass (z ≥ 1.80) do not manifest the same problem and
between the slopes of the observed spectrum and the the slopes of their optical spectra are in perfect agree-
spectroscopic solution are due to interstellar reddening. ment with those of their spectroscopic solutions. Once
The spectroscopic fit to the corrected spectrum yields again, we are confident that there is nothing wrong with
Teff = 30,270 K and log g = 7.83. the data itself. Furthermore, previous determinations of
Teff for this object suggest that it is a rather cool DA
3 http://irsa.ipac.caltech.edu/applications/DUST/, derived us-
star. Indeed, Kilkenny et al. (1997) find Teff = 7000 K
ing the data and techniques from Schlegel et al. (1998) while Zuckerman et al. (2003) find Teff = 8000 K, in both
Table 5
Atmospheric Parameters of DA White Dwarfs from MS99
WD Name ST Teff (K) log g M /M⊙ MV D (pc) Notes

0000+171 PG 0000+172 DA2.4 21,130 (325) 8.00 (0.05) 0.63 (0.03) 10.68 108
0000−186 GD 575 DA3.3 15,350 (265) 7.97 (0.05) 0.60 (0.03) 11.19 105
0001+433 RE J0003+433 DA1.1 46,850 (1244) 9.05 (0.10) 1.23 (0.04) 11.24 118
0004+061 PHL 670 DA2.1 24,400 (386) 8.51 (0.05) 0.94 (0.03) 11.23 100
0004+330 GD 2 DA1.0 49,980 (898) 7.77 (0.06) 0.59 (0.02) 8.79 103
0005−163 G158-132 DA3.4 14,920 (252) 7.93 (0.05) 0.57 (0.03) 11.18 104 1
0008+424 LP 192-41 DA7.0 7200 (107) 8.12 (0.07) 0.66 (0.05) 13.64 21
0009+501 G217-37 DA7.6 6620 (103) 8.40 (0.09) 0.85 (0.06) 14.42 10 1
0009−058 G158-39 DA4.8 10,560 (156) 8.22 (0.06) 0.74 (0.04) 12.34 54
0010+280 PG 0010+281 DA1.9 27,220 (407) 7.87 (0.05) 0.57 (0.02) 9.99 152
0011+000 Wolf 1 DA5.2 9710 (140) 8.24 (0.05) 0.75 (0.04) 12.67 34 1
0012−173 PHL 752 DA.9 55,810 (1321) 7.88 (0.08) 0.65 (0.04) 8.85 379
0013−241 Ton S 147 DA2.6 19,340 (301) 8.04 (0.05) 0.65 (0.03) 10.89 78
0014+097 PG 0014+098 DA3.5 14,410 (273) 7.80 (0.05) 0.50 (0.03) 11.06 91 2
0016−220 Ton S 142 DA3.7 13,610 (346) 7.88 (0.05) 0.54 (0.03) 11.29 64
0016−258 MCT 0016-2553 DA4.4 11,370 (166) 8.20 (0.05) 0.72 (0.03) 12.09 63 3
0017+061 PHL 790 DA1.7 29,120 (463) 7.82 (0.06) 0.55 (0.03) 9.75 100
0018−339 BPM 46232 DA2.2 22,420 (339) 7.87 (0.04) 0.56 (0.02) 10.38 73
0019+150 PHL 802 DA1.6 31,710 (484) 8.14 (0.06) 0.73 (0.03) 10.07 292
0023+388 G171-B10A DA4.6 10,980 (164) 8.21 (0.05) 0.73 (0.03) 12.21 56
...
Note. — Table 5 is available in its entirety in the electronic edition of the Astrophysical Journal. A
portion is shown here for guidance regarding its form and content. (1) Photometrically constant; (2) DA+dM
binary; (3) ZZ Ceti; (4) Balmer-line problem; (5) based on non-magnetic models; (6) DA+DC binary; (7)
DA+DB binary; (8) Suffers from interstellar reddening.
cases the temperature is determined solely from a mea- includes all the normal DA white dwarfs in our sample
surement of (B − V ). Combining the photometry from as well as all the stars we have analyzed in detail in the
Kilkenny et al. (1997) with the available measurements preceding sections including the DAB and DAZ stars,
of JHKs from 2MASS, we performed a photometric fit and the DA+DB and DA+dM binaries. We also include
using the same procedure described in Section 4.7.1. The the nine weakly magnetic white dwarfs from Section 4.6
results of this fit are displayed in the bottom panel of and the three peculiar objects from Section 4.7. In all, a
Figure 20. Assuming log g = 8.0, our fit to the available total of 1265 white dwarfs and their atmospheric param-
photometry yields Teff = 7530 K, in excellent agreement eters are listed in Table 5. The entries in Table 5 are or-
with the previous determinations, and we conclude that dered by their WD numbers, and we list the values of Teff
the cold solution must be the correct one. But how do and log g as well as masses derived from the evolutionary
we reconcile this temperature with the poor fit to the models of Wood (1995) with thick hydrogen layers. Al-
Balmer lines with the cold solution? The most likely ex- though the Montreal group has computed evolutionary
planation is that 0927−173 is a DA+DC binary system models of their own (Fontaine et al. 2001), these are bet-
where the continuum flux of the DC component dilutes ter suited for the cooler end of the white dwarf cooling se-
the observed Balmer lines leading to the results seen in quence. At higher temperatures, the Wood and Fontaine
Figure 19. In order to fit the composite spectrum, we em- et al. models are quite comparable and, in fact, the Wood
ploy exactly the same method as we did with the DA+DB models have the benefit of using post-AGB evolutionary
systems analyzed in Section 4.3. Once again, Teff and sequences as a starting point (M. A. Wood, private com-
log g for the DA component and Teff for the DC compo- munication). Furthermore, low-mass white dwarfs, be-
nent are the free parameters as we fix log g = 8.0 for the low 0.46 M⊙ and Teff < 50,000 K, are likely helium core
DC star. The results of our fit are shown in Figure 24 white dwarfs, and we rely instead on the evolutionary
where we see that the fit to the composite spectrum is models from Althaus et al. (2001) for those stars. For
nearly perfect. We obtain Teff = 10,000 K and a rather masses higher than 1.3 M⊙ , we use the zero temperature
high surface gravity of log g = 8.91 for the DA compo- calculations of Hamada & Salpeter (1961). The values in
nent, paired to a significantly cooler DC component with parentheses represent the uncertainties of each parame-
Teff = 5600 K. More importantly, in Figure 25 we com- ter, calculated by combining the internal error, which is
pare the model flux for the combined DA+DC system the dominant source of uncertainty, obtained from the
to our observed spectrum and we see that the slopes are covariance matrix of the fitting algorithm with the ex-
in excellent agreement. Consequently, we conclude that ternal error, obtained from multiple observations of the
the observed slope of 0927−173 can be explained by the same object, estimated for DA stars at 1.2% in Teff and
additional presence of an unresolved companion, a cool 0.038 dex in log g (see LBH05 for details). The effect on
DC white dwarf. the uncertainties caused by different values of S/N are in-
cluded in the external error but are minimized by the fact
5. GLOBAL PROPERTIES AND DISCUSSION that the majority of our data have rather high S/N (see
5.1. Adopted atmospheric parameters Figure 12 of Gianninas et al. 2005). In addition, we list
absolute visual magnitudes determined using the pho-
The atmospheric parameters we have adopted for all tometric calibrations from Holberg & Bergeron (2006).
the stars in our sample are summarized in Table 5. This
We also provide an updated spectroscopic classification 5.2. Comparison with SPY

for each star, with temperature indices4 based on our In establishing the reliability of the results of large
determinations of Teff . surveys such as the one presented here, we believe it is
By coupling the absolute magnitudes we have deter- both interesting and important to compare our results
mined with the apparent magnitudes available for each with other studies that include many of the same ob-
star in the literature, we computed spectroscopic dis- jects. As mentioned earlier, recent analyses of large num-
tances that we also list in Table 5. However, these de- bers of white dwarfs were presented both by LBH05 and
terminations come with the caveat that many of these Limoges & Bergeron (2010), and although virtually all of
apparent magnitudes are photographic or, when V was the white dwarfs in those studies are included here, we
not available, we have substituted a B magnitude, or are also using the exact same observations and analysis
a g magnitude from the SDSS, or even a Strömgren y techniques making any meaningful comparison a rather
magnitude. Consequently, our distance estimates should futile endeavor. Instead, what we require is an indepen-
be viewed with caution. Additionally, we applied an ex- dent analysis, which is exactly what the SPY sample ana-
tinction correction of AV =1.4685 when determining the lyzed by Koester et al. (2009) represents. The DA white
distance to 0737−384. We note that our distance of 142 dwarfs in the SPY sample had initially been analyzed
pc for this star also places it outside NGC 2451 (∼180 in Voss (2006), however Koester et al. (2009) have re-
pc, Platais et al. 2001). analyzed the same stars with a new generation of models
Since there is a growing interest in identifying white presented in Koester (2010). We must stress that these
dwarfs in the local neighborhood, in particular those models did not yet include the improved Stark profiles of
within 20 pc (Holberg et al. 2008), we note that there TB09. Consequently, as previously discussed, we will be
is no strong evidence for new white dwarfs in our sam- using the results obtained with the Lemke (1997) broad-
ple within that distance, with the exception perhaps ening profiles in order to compare with the SPY sample.
of 0213+396 (16 pc). The distance to 1503−070 of 7 More importantly, the white dwarfs from SPY have been
pc is obviously underestimated since we fitted this star observed and analyzed with models and techniques com-
with non-magnetic models (see Figure 18); the measured pletely independent from our own and thus represent a
trigonometric parallax actually suggests a distance of ∼ truly distinct data set.
25 pc. We also have several objects with distances within The sample presented in Koester et al. (2009) con-
20 pc that are too cool (Teff < 6500 K) to derive meaning- sists of a total of 615 DA white dwarfs of which 362
ful atmospheric parameters with the spectroscopic tech- are in common with our sample. The reason there is
nique, in particular log g. One must also remember that not a greater overlap is two-fold. First, many of the
the spectroscopic distances of cool DA stars located in stars selected for the SPY project were taken from the
the temperature range where the high-log g problem is Hamburg-ESO (Christlieb et al. 2001) and Hamburg-
encountered are clearly underestimated since their lu- Schmidt (Homeier et al. 1998) surveys and these stars
minosity is also underestimated (smaller radii); this in- were not yet listed in MS99. Second, as we mentioned
cludes 0213+396, mentioned above, at Teff = 9370 K. earlier, many of the white dwarfs for which we were un-
Finally, special notes have been added in Table 5 to able to obtain spectra were situated in the southern hemi-
designate objects that are known ZZ Ceti pulsators as sphere, and this also contributed to the smaller overlap.
well as stars that are photometrically constant. We note In order to be sure we are comparing all stars on an
that all the stars analyzed in Gianninas et al. (2010) are equal footing, we choose to compare only single DA white
also listed in Table 5. This includes both the DAO and dwarfs; any known double-degenerate systems or mag-
DA+BP stars, with a note for all stars that exhibit the netic white dwarfs are excluded. Furthermore, we also
Balmer-line problem. For the DA white dwarfs that have omit stars with Teff > 50,000 K and Teff < 8000 K for the
been determined to belong to DA+DB, DA+DC, and same reasons outlined in Koester et al. (2009). Namely,
DA+dM systems, we list only the parameters of the DA the models used by Koester et al. (2009) do not include
component. Finally, we include a note for the weakly effects due to departures from local thermodynamic equi-
magnetic stars that the listed parameters were deter- librium at higher temperatures, and the spectra become
mined with non-magnetic models. less sensitive to changes in log g at lower temperatures.
These last two criteria reduce the number of viable white
5.1.1. Erratum: 0102−185 dwarfs to 330.
The results of our comparison are presented in Fig-
One of the reported DA+BP stars from ure 26. We see that with the exception of a couple of
Gianninas et al. (2010), 0102−185, is actually a outliers near Teff ∼ 14,000 K, the agreement between
much cooler DA white dwarf with Teff = 23,410 K and our Teff values and those from SPY is excellent. The two
log g = 7.78, in contrast to the parameters previously outliers (0318−021 and 0937−103) are stars where both
reported. An error during data reduction associated the cold and hot seeds lead to the same spectroscopic
the wrong spectrum to this star. The spectrum that solution (see LBH05 for a discussion about the hot and
was analyzed as 0102−185 is actually a spectrum of cold solutions). Clearly, this is not the case in the SPY
2211−495 from the same observing run. This reduces analysis of the same objects as they derive larger values
the number of DA+BP stars from 18 to 17. for Teff in both cases. In contrast, the log g determina-
tions show a clear trend where our measurements tend
4 See MS99 for a description of how the temperature index is
toward higher values than those from SPY. Koester et al.
calculated. (2009) had noted this problem (see their Figure 3) and
5 Taken from http://irsa.ipac.caltech.edu/applications/DUST/,
see Section 4.7.2
discussed several possibilities in an attempt to under-
Figure 26. Comparison of log g (top) and Teff (bottom) values Figure 27. Comparison of log g (top) and Teff (bottom) values
between this work using the Lemke (1997) Stark profiles and the between this work using the Lemke (1997) Stark profiles and the
results of Koester et al. (2009) for 330 DA white dwarfs common results of our fits to the SPY spectra using the same models. The
to both samples. The dotted lines represent the 1:1 correlation. dotted lines represent the 1:1 correlation.
stand this discrepancy. In particular, they point out that

differences between the “Bergeron” and “Koester” mod-
the log g values (top panel). There is more scatter in the
els could be the cause. Indeed, LBH05 showed a com-
log g plot but the overall agreement is markedly better
parison between their results and those of several other
than was seen in Figure 26. Since we obtain a satis-
studies. Three of those analyses used Koester models
factory agreement between the atmospheric parameters
and all three seemed to exhibit the same trend we see
measured from both sets of spectra, we must conclude
here. Another possibility is that the apparent shift in
that there is no problem, a priori, with the SPY spectra
log g could be caused by differences in the fitting proce-
themselves. Differences between our results and those
dures. Finally, Koester et al. (2009) speculate that the
of Koester et al. (2009) must necessarily arise from dif-
very nature of the SPY data could be the origin of the
ferences in the models or the fitting techniques used.
problem. The SPY spectra were obtained with the UV-
Nonetheless, the fact that the Teff determinations agree
Visual Echelle Spectrograph (UVES) at the Very Large
so well in both of the previous comparisons is encour-
Telescope (VLT), and the stitching together of the var-
aging. Furthermore, when using an independent data
ious orders might have introduced artifacts in the spec-
set and relying on our own theoretical framework, we
tra. For the complete discussion, we refer the reader to
succeed in getting both Teff and log g values that match
Koester et al. (2009).
rather well.
In an effort to better understand this shift in log g val-
ues, we decided to independently fit the SPY spectra,
which were kindly provided to us by D. Koester, using 5.3. Comparison with SDSS
our models and fitting procedures. In Figure 27 we show Another independent source of spectra is the SDSS.
the comparison between these new values derived from Despite the thousands of new white dwarfs discovered
the SPY spectra and those from our spectra, all ana- in the SDSS, very few are in common with our sample.
lyzed with the Lemke (1997) Stark profiles. First, we As we showed in Figure 1, the much fainter sample of
note that the agreement between the Teff determinations white dwarfs observed by the SDSS is the reason the two
(bottom panel) remains very good. Second, and more samples are so mutually exclusive. Nonetheless, we have
importantly, we do not observe any systematic shift in retained 83 DA white dwarfs in common with both sam-
Figure 29. Comparison of absolute magnitudes derived from par-

allax measurements with the absolute magnitudes derived from our
spectroscopic determinations of Teff and log g for 101 white dwarfs.
The open circles represent known or suspected double-degenerate
binary systems while the triangles correspond to central stars of
planetary nebulae. The dotted lines represent the 1:1 correlation.
with those obtained from parallax measurements. The

added incentive here is that the absolute magnitudes ob-
tained from parallaxes are completely model indepen-
dent. We have 92 DA white dwarfs in our sample with
available parallax measurements from either the Yale
Figure 28. Comparison of log g (top) and Teff (bottom) values Parallax Catalogue (van Altena et al. 1995), or Hippar-
derived from our observations and from fits to SDSS spectra of
83 DA white dwarfs common to both samples. The dotted lines cos (Perryman & ESA 1997) and 9 central stars of plan-
represent the 1:1 correlation. etary nebulae with parallaxes taken from Harris et al.
(2007). We present in Figure 29 the comparison between
ples6 . We obtained the SDSS spectra for these stars from the MV values from parallax measurements and those
the SDSS SkyServer7 and proceeded to fit them with the obtained through our spectroscopic analysis. We can see
new models that include the TB09 Stark broadening pro- that the overall agreement is quite good and this rein-
files. The comparison between our Teff and log g deter- forces our confidence that our results are accurate.
minations and those obtained from the SDSS spectra is There are a number of obvious outliers in this figure
presented in Figure 28. As in the case of the compari- that can be easily explained as unresolved double de-
son with SPY, we note a very good agreement between generates (open circles), the most important of which
both sets of Teff values (bottom panel). However, we is L870-2 (0135−052), discovered by Saffer et al. (1988).
also note a second similarity with our SPY comparison as But this is not always the case. For instance, we find
the log g values also seem to be systematically shifted to- MV (spec) = 11.68 for Ross 548 (0133−016, ZZ Ceti it-
wards lower values from those determined from the SDSS self), for a corresponding distance of 31 pc, while the
spectra. This concurs with the results of Tremblay et al. distance obtained from the trigonometric parallax mea-
(2011a) who explained that any differences must be due surement from the Yale Parallax Catalogue is more than
to the different spectra employed as they are being an- twice as large at 67 pc. We cannot find an easy ex-
alyzed within the same theoretical framework and the planation for this discrepancy unless Ross 548 is also
same fitting method. Tremblay et al. (2011a) conclude an unresolved degenerate binary. As well, the abso-
that issues with the reduction of the SDSS data still re- lute magnitudes of cool DA stars found in the temper-
main and are the reason for the observed shift. ature range where the high-log g problem is found are
likely to be overestimated; this accounts for the clump
5.4. Comparison with parallax measurements of objects away from the 1:1 correlation in the range
Yet another way of testing the reliability of our results 12.5 < MV < 13.5.
is to compare our derived values of absolute magnitudes We must finally mention that the observed agreement
is very good within the parallax uncertainties for both
6 We omit DA+dM binaries, DAO stars, and magnetic white model grids. As TB09 pointed out, despite the fact that
dwarfs from this comparison. the new models yield higher values of Teff and log g (i.e.,
7 http://cas.sdss.org/dr7/en/
smaller radii), the two effects nearly cancel each other out
and the predicted luminosities (or MV ) remain largely The overall shape of the distribution is in good agreement
unchanged. with the latest determination of the SDSS DA mass dis-
tribution presented by Tremblay et al. (2011a), the only
5.5. Mass distribution other large scale analysis of DA white dwarfs employing
We display in Figure 30, the mass distribution as a the new TB09 Stark profiles. In both cases, the mass
function of Teff for all the stars listed in Table 5. As distribution is strongly peaked near 0.6 M⊙ . The other
previously noted, the bulk of the distribution is centered notable features are the high-mass tail and the low-mass
around 0.6 M⊙ for Teff > 13,000 K. Below this temper- component that peaks near 0.45 M⊙ . The mean mass
ature, the well known “high-log g problem” manifests it- of the entire sample is hM i = 0.661 M⊙ . However, if
self. Several solutions to this problem have been explored we consider only stars with Teff ≥ 13,000 K, we get a
over the years. First, Bergeron et al. (1990b) showed lower mean mass of hM i = 0.638 M⊙ . This is higher
how helium, spectroscopically invisible below 13,000 K, than the value of 0.613 M⊙ obtained by Tremblay et al.
might explain this problem. If analyzed with pure hydro- (2011a) for the SDSS DR4 sample. However, as stated
gen models, DA white dwarfs containing helium would in Tremblay et al., the mean mass is sensitive to outliers
appear to have higher log g values because the presence and as we will discuss below, the MS99 sample contains
of helium increases the pressure in the atmosphere, which a number of high-mass white dwarfs that are systemati-
is analogous to increasing the surface gravity. Recently, cally missed by many surveys. This fact is also evidenced
Tremblay et al. (2010) attempted to detect the presence in Figure 18 of Tremblay et al. (2011a) where we note a
of the He i λ5877 line in six DA white dwarfs using high dearth of stars for M > 1.0 M⊙ and Teff ≥ 20,000 K.
resolution spectroscopy. They found no evidence of the We note in Figure 30 a number of rather high mass
helium line and concluded that helium was not the solu- white dwarfs. In particular, if we restrict ourselves to
tion. The reason helium was thought to be the key in- stars with Teff > 13,000 K, thus avoiding the troublesome
gredient in explaining the high-log g problem is because lower temperature regime where the high-log g problem
it was expected that convective mixing would bring it to reigns, we find there are 20 white dwarfs with a mass
the surface. In addition, properly calibrating the mixing M ≥ 1.1 M⊙ . Of these, 11 were detected by ROSAT
length theory used to describe convection was deemed (Fleming et al. 1996). This trend is not at all surprising.
crucial. Bergeron et al. (1995) addressed this issue but By definition, white dwarfs with a higher mass have a
even with the mixing length properly calibrated, the smaller radius, which, in turn, means they are intrinsi-
high-log g problem remained. The advent of the TB09 cally fainter at a given Teff . As such, magnitude-limited
profiles was also thought to represent a possible element UV-excess surveys (PG, KUV), and many of the other
of the solution. However, as we have seen, the new pro- surveys looking for bright blue objects (MCT, EC) would
files succeed in “straightening out” the mass distribution, have skipped these stars because they are fainter. On the
but do not seem to contribute in fixing the masses at other hand, X-ray surveys like the one conducted with
lower Teff . One of the few remaining avenues of research ROSAT, catalog all sources, regardless of the bright-
is the computation of full 3D hydrodynamic models of ness of the object. Consequently, since ours is not a
convection as a replacement to the mixing length theory magnitude-limited sample, these white dwarfs make their
that has always been used, but ultimately constitutes way into our sample by virtue of being included in MS99.
an approximation. The preliminary results from such At the other end of the mass spectrum, the presence
computations, as presented by Tremblay et al. (2011b), of stars below the dotted lines in Figure 30 is significant.
show that these models produce corrections to the values Those isochrones include the main sequence lifetime of
of log g that are of the correct magnitude and direction the white dwarf’s progenitor. This represents a cutoff
when compared to shifts deduced from the analysis of of sorts in that the possible progenitor stars would not
large samples of DA white dwarfs. Hence, it seems that have had sufficient time to become white dwarfs within
the solution to the high-log g problem has finally been the lifetime of the Galaxy, nor to have cooled down to
nailed down. those temperatures. However, it is clear that there exists
At the hotter end of the white dwarf sequence, we a non-negligible low-mass component in the mass distri-
note in Figure 30 that the distribution remains essen- bution. The presence of these stars can be explained if
tially continuous around 0.6 M⊙ unlike the previous de- we consider binary star evolution under different guises.
termination shown in Figure 3 of Gianninas et al. (2009) First, several of these stars are known to be unresolved
where there is a noticeable “dip” toward lower masses at double-degenerate binary systems that have been dis-
higher Teff . Our new result is a consequence of the new covered through searches for radial velocity variations
determinations of Teff and log g for the DAO stars from (Maxted & Marsh 1999; Maxted et al. 2000). In such
Gianninas et al. (2010). As it was demonstrated, the use cases, the measured atmospheric parameters are mislead-
of the CNO models to solve the Balmer-line problem also ing if we assume only a single star is present. A second
resulted in higher masses for the DAO stars, consistent possibility involves the merger of two white dwarfs. In-
with the remainder of the hydrogen-rich white dwarf pop- deed, the recent results presented by Kilic et al. (2011,
ulation. and references therein) seem to suggest that there are
We show in Figure 31 the histogram that represents several very low mass (∼ 0.2 M⊙ ) white dwarfs that are
the mass distribution of our sample regardless of effective in close binary systems. If these white dwarfs merge, they
temperature. Besides the full distribution, we also plot could potentially form a white dwarf with a mass near
the mass distributions for stars above and below Teff = 0.4 M⊙ . Finally, there is the possibility of a helium core
13,000 K (in blue and red, respectively). The reason for white dwarf (Althaus et al. 2001). Helium-core white
this split is that we want to analyze the mass distribution dwarfs are formed when stars lose a significant amount of
for stars that are not affected by the high-log g problem. their envelope’s mass to a companion before the progen-
Figure 30. Mass distribution as a function of Teff for all the stars listed in Table 5. The solid red lines represent isochrones that take into
account only the white dwarf cooling time whereas the dotted red lines include also the main sequence lifetime. Each isochrone is labeled
by its age in Gyr. Isochrones for τ < 1 Gyr are taken from Wood (1995) whereas those for τ ≥ 1 Gyr are taken from Fontaine et al. (2001).
clear: these low-mass white dwarfs represent a separate

evolutionary channel distinct from the usual post-AGB
models that explain the evolution of the vast majority of
white dwarfs.
5.6. Revisiting the PG luminosity function
Besides revising the atmospheric parameters of all the
normal DA stars from the complete sample with the use
of the TB09 profiles, there is also a number of white
dwarfs whose atmospheric parameters are further refined
from the LBH05 analysis as they are members of cer-
tain distinct classes of objects. In particular, there are
16 DA+dM binaries, 10 DAO stars8 , 5 DA+BP stars,
7 magnetic white dwarfs and finally the DA component
from the 1115+166 DA+DB binary system. For the mag-
netic white dwarfs not included in Table 5, we adopt the
parameters listed in Table 3. With these up to date at-
mospheric parameters in hand, we proceed to recompute
the PG luminosity function in exactly the same manner
described in LBH05. We compare our new determina-
tion of the PG luminosity function with that of LBH05
in Figure 32. We see some variations in the individ-
ual magnitude bins with respect to the LBH05 result,
Figure 31. Mass distribution for all the stars listed in Table 5 but this is to be expected considering the revised at-
(solid line histogram). Also shown are the mass distributions for mospheric parameters we have employed in computing
stars with Teff > 13,000 K (hatched blue histogram) and for stars the luminosity function. We can also compare the space
with Teff < 13,000 K (hatched red histogram). The mean mass
and dispersion for each distribution are indicated in the figure in densities, in other words, the total number of DA white
units of solar masses. dwarfs per pc−3 . This is determined by integrating the
8 PG 1305−017 is included in this sample but, as out-
itor reaches the tip of the red giant branch. Regardless lined in Gianninas et al. (2010), the parameters obtained by
which of these scenarios is the most likely to account for Bergeron et al. (1994) using their stratified models are adopted,
the population of low-mass white dwarfs, one thing is as such, the parameters for this star remain unchanged.
Figure 33. Teff - log g distribution for DA white dwarfs with

high-speed photometric measurements. The cyan circles represent
the 56 ZZ Ceti stars in our sample and the red circles are the
145 photometrically constant DA stars from Table 5. The dotted
lines represent the empirical boundaries of the instability strip as
established in Gianninas et al. (2006) whereas the dashed lines cor-
respond to our new determinations. The error bars represent the
average uncertainties of the spectroscopic method in the region of
the ZZ Ceti instability strip.
Figure 32. Luminosity function of all DA stars in the complete

PG sample using the atmospheric parameters derived in this work
(solid line) presented in half-magnitude bins, assuming a scale
height for the Galaxy of z0 = 250 pc. The dashed line represents
the results of LBH05 shown here for comparison.
area under the luminosity function. LBH05 had deter-
mined a value9 of 3.07 × 10−4 pc−3 for MV < 12.75. Our
new determination of the PG luminosity function yields
a slightly lower value of 2.92 × 10−4 pc−3 for the same
magnitude range. This represents only a 5.1% difference
with the value from LBH05. At first glance, one might
think that the higher temperatures produced by the new
TB09 Stark profiles means that stars are intrinsically
brighter than we once thought, meaning they can poten-
tially be further away thus reducing the local density of Figure 34. Differences in Teff determinations between the im-
white dwarfs. However, our earlier comparison between proved Stark profiles of TB09 and the VCS profiles of Lemke (1997)
as a function of Teff for stars near the ZZ Ceti instability strip. The
the absolute magnitudes obtained from spectroscopy and dotted line represents the 1:1 correlation.
those determined from parallax measurements showed
that the intrinsic brightness of these white dwarfs has
not changed in an appreciable way since the stellar radii
are also smaller (higher masses). Under these circum-
stances, the luminosity function should remain almost
unaffected. However, closer inspection of the individual
values of Teff , log g, and MV reveals that small differ-
ences do cause stars to jump from one magnitude bin to
the next, which can ultimately affect the determination
of the space density of white dwarfs.
5.7. ZZ Ceti instability strip
In Table 5 we list a number of stars that have been
observed in high-speed photometry in order to ascertain
if they are photometrically variable. In total, we identify
145 white dwarfs that are photometrically constant and
56 stars that are known ZZ Ceti pulsators. Using this
photometric sample of 201 white dwarfs, we wish to re- Figure 35. Same as Figure 33 but for our entire sample of white
visit the empirical determination of the boundaries of the dwarfs. The black circles represent the stars for which we do not
have any high-speed photometric data. The dashed lines corre-
9 LBH05 actually give a value of 5.0×10−4 pc−3 but this number spond to our new determination of the empirical boundaries of the
is inaccurate for reasons unknown to one of the co-authors of both ZZ Ceti instability strip.
studies (P.B.) who obtained the correct number provided here.
ZZ Ceti instability strip. In particular, we wish to see vate communication). Since the discovery of its variabil-
how the use of the TB09 Stark broadening profiles has ity is more recent, it is likely that the Moran spectrum
affected the location of the instability strip as compared was not averaged over several pulsation cycles as is nec-
to our last result in Gianninas et al. (2007). We plot in essary in order to obtain a proper measure of the atmo-
Figure 33 the location of these 201 white dwarfs in the spheric parameters for ZZ Ceti stars. Indeed, Voss et al.
Teff -log g plane. For the purposes of comparison, we plot (2007) report a period of 1350 s and an amplitude of
as dotted lines the empirical boundaries of the instability 5.69 mma, consistent with cooler ZZ Ceti stars that have
strip as determined in Gianninas et al. (2007). Our new longer periods and larger amplitudes as compared to
determination of these boundaries is shown as dashed their hotter counterparts. These large amplitudes sig-
lines. First, we notice that the entire strip has shifted by nify that Teff can easily swing back and forth by several
a few hundred kelvins towards higher Teff . Second, we hundred kelvins resulting in the star moving in and out of
see that the strip has also widened by several hundred the instability strip (Fontaine & Brassard 2008). How-
kelvins. These changes are a direct result of the use of ever, we finally secured a new spectrum for 1959+059,
the new TB09 broadening profiles. We plot in Figure 34 averaged over several pulsation cycles, during our last
differences in Teff obtained by using the old Lemke (1997) observing run in 2011 April. The new values do not sub-
profiles versus the calculations of TB09, focusing on the stantially change the position of the star in the Teff -log g
range of Teff where the ZZ Ceti instability strip lies. We plane and it remains outside the empirical boundaries of
see that the new values of Teff are systematically higher the instability strip. This is not altogether inconsistent
than the old values, this explains why the whole instabil- with the notion of a pure instability strip in the context
ity strip is shifted toward higher effective temperatures. of the difficulty of precisely defining the red edge of the
However, we also notice that ∆Teff increases as a function strip, as discussed above.
of Teff from ∼ 200 K near Teff = 11,000 K, to ∼ 600 K at The second offending star is HS 1612+5528, the non-
Teff = 13,000 K. This is the reason the instability strip variable white dwarf lying in the heart of the instability
is now wider than previously established. strip. This star was reported as “not observed to vary”
One aspect of the instability strip that remains un- by Voss et al. (2006). Gianninas et al. (2009) indepen-
changed is its purity. With the exception of the two dently observed HS 1612+5528 and also detected no vari-
“offending” stars discussed below (one pulsator and one ations down to a limit of 0.2%. However, there are ZZ
non-variable), only pulsating white dwarfs are to be Ceti stars with amplitudes as low as 0.05%. It is possible
found within the instability strip and only photometri- that in both cases the star was observed during a period
cally constant white dwarfs without, within the uncer- of destructive interference. Alternatively, HS 1612+5528
tainties. We stress that the significance of a pure insta- could represent the first ZZ Ceti star whose pulsations
bility strip is twofold. First, if the strip is pure, one can are hidden from us due to geometric considerations. In
predict the variability of white dwarfs whose atmospheric other words, since we cannot resolve the surface of the
parameters places them within the confines of the insta- star, the excited pulsation mode, or modes, might pro-
bility strip. Second, if the strip is pure, then it represents duce no detectable variations if the combined effect of the
an evolutionary phase through which all hydrogen-rich alternating hot and cold areas is nil. This would likely
white dwarfs must pass. As a result, asteroseismological require a chance alignment where, for example, the rota-
analyses of these stars can yield important information tion axis of the star is perpendicular to our line of sight.
about their internal structure, in particular the thickness Whatever the case may be, we are planning to secure
of the hydrogen layer. This in turn can be applied to the new high-speed photometry for this star during the next
the entire population of hydrogen-rich white dwarf stars. trimester to settle, once and for all, the issue of its pho-
We also note that the blue edge seems to be quite tometric status.
clearly defined whereas there is a certain level of am- Finally, in Figure 35 we add all the remaining stars
biguity associated with the location of the red edge. As from our survey whose photometric status is unknown.
far as the blue edge is concerned, the sharp transition We can identify a total of 14 ZZ Ceti candidates that lie
between non-variable to variable stars is probably a con- within the empirical boundaries of the instability strip.
sequence of very specific conditions being required for the However, as Gianninas et al. (2006) demonstrated, it is
driving mechanism to ignite pulsations at the surface of important and worthwhile to observe all white dwarfs
the white dwarf. On the other hand, the dying off of that are near either edge of the instability strip as well.
pulsations at the red edge is not so well understood and It could be argued that adding a few more ZZ Ceti stars
it is speculated that the interaction between convection to the mix is not especially important since hundreds
and the pulsations is what ultimately signals the demise have now been identified through the SDSS and follow-
of variability. It is conceivable that this process is not up work. However, the SDSS variables are, for the most
as clear cut and thus the exact temperature at which a part, a few magnitudes fainter than the white dwarfs
star will cease pulsating is more ambiguous, producing a shown here. Besides the discovery lightcurves, it is hard
red edge whose precise location is more difficult to nail to conceive of meaningful asteroseismological studies of
down. the ZZ Ceti stars from SDSS. In contrast, these brighter
As mentioned above, there are two notable exceptions candidates could be studied on small to medium-sized
to the “pure” instability strip. The first of the two offend- telescopes equipped with proper instrumentation.
ing white dwarfs is 1959+059. This is a ZZ Ceti star that The most intriguing of all the candidates is certainly
lies outside the instability strip near ∼ 11,000 K. This 1659+662 (GD 518). With log g = 9.08, it would easily
star was first reported as being variable by Voss et al. become the most massive ZZ Ceti star ever discovered,
(2007). However, our initial spectrum for this object was surpassing even 1236−495 (BPM 37093, LTT 4816) the
one of the many provided to us by C. Moran (1999, pri- current heavyweight of the group. High-speed photo-
metric measurements of this object are being planned as hundred white dwarfs were in common with both samples
well. but were observed and analyzed in a completely indepen-
dent fashion. We showed that our Teff determinations are
6. CONCLUSION in excellent agreement with those of Koester et al. (2009)
We have presented the results of our spectroscopic sur- despite a trend towards higher log g values in the Koester
vey of hydrogen-rich white dwarfs from MS99. First, we et al. analysis.
discovered that many stars listed in the catalog are still Our mass distribution yields a mean mass of 0.638 M⊙
erroneously classified as white dwarfs. In particular, we when considering only white dwarfs with Teff ≥ 13,000 K.
list a total of 68 stars that are misclassified, 27 of which The mean mass is somewhat higher than other determi-
need to be reclassified as the result of our observations. nations, but it is understood to be caused by high-mass
We then examined the spectroscopic content of our sur- white dwarfs that are missed in many surveys but in-
vey. The majority of stars in our sample are simple DA cluded here. We also note the presence of a low-mass
white dwarfs but a number of them show more than just component, consistent with the results of most other
the characteristic Balmer lines of hydrogen. Indeed, our large scale studies of DA white dwarfs. The low-mass
sample includes both DAB and DAZ white dwarfs as well stars are necessarily the product of binary evolution ei-
as DA+dM binaries and magnetic white dwarfs. In addi- ther through mergers or a common envelope phase where
tion, there are several white dwarfs that are cool enough a significant fraction of their mass may have been lost.
that no Balmer-lines are present, at least not within the We also recomputed the PG luminosity function and
spectral range covered by our observations. With the found a space density which is only slightly lower than
exclusion of the magnetic stars and the cool, featureless that determined by LBH05. It would be easy to point the
white dwarfs, we have analyzed the spectra of all the finger at the new TB09 Stark broadening profiles since
white dwarfs in our sample. In order to do so, and with they yield higher temperatures, but they also yield higher
the goal of obtaining the most accurate atmospheric pa- log g measurements that translate to smaller radii. These
rameters possible, we have endeavored to use the most facts coupled together produce two very similar values of
modern, up-to-date, and appropriate model atmospheres MV . However, even small differences can cause stars to
in our analyses of these white dwarfs. switch magnitude bins, which explains our marginally
First, for the analysis of all the hydrogen-rich white lower value.
dwarfs we have used models that include the new Finally, we re-examined the ZZ Ceti instability strip
Stark broadening profiles from TB09. The analysis and found that the TB09 Stark broadening profiles pro-
of the hot DAO and DA+BP stars was presented in duce a wider instability strip that is also systematically
Gianninas et al. (2010) and featured models to which shifted toward higher Teff . Furthermore, we have identi-
CNO, at solar abundances, was added in order to over- fied over a dozen new ZZ Ceti candidates among which
come the Balmer-line problem. The results of that might be the most massive ZZ Ceti star ever discovered.
analysis were incorporated here. Furthermore, we used High-speed photometric observations are planned for all
updated helium atmosphere models to analyze several candidates in the very near future.
DA+DB double-degenerate binary systems and we de- We thank the director and staff of Steward Obser-
veloped a technique that uses M dwarf templates to cor- vatory and of the Carnegie Observatories for the use
rect for the contamination in the composite spectra of of their facilities. We are also grateful to D. Koester
DA+dM binary systems. We also analyzed several DAZ for providing us with all of the SPY spectra, and M.-
stars by including calcium in the calculation of the syn- M. Limoges for obtaining several of the optical spectra
thetic spectrum only. Two of these are newly identified analyzed in this paper. This work was supported in
DAZ stars and they should be checked for any infrared part by the NSERC Canada and by the Fund FQRNT
excess. This would betray the presence of a circumstel- (Québec). M. T. R. acknowledges support from FON-
lar disk that would be the logical source of the metal DAP (15010003) and Proyecto BASAL PB06 (CATA).
pollutants. P. B. is a Cottrell Scholar of Research Corporation for
In switching from the older Stark broadening profiles of Science Advancement.
Lemke (1997) to those of TB09, we also explored how our
atmospheric parameters have changed as a result. We REFERENCES
observed that the values of Teff are shifted toward higher
temperatures while the values of log g show a more obvi- Achilleos, N., & Wickramasinghe, D. T. 1989, ApJ, 346, 444
ous and significant trend to towards higher values. With Althaus, L. G., Serenelli, A. M., & Benvenuto, O. G. 2001,
MNRAS, 323, 471
that said, we also saw how the new profiles improved Angel, J. R. P., Borra, E. F., & Landstreet, J. D. 1981, ApJS, 45,
the overall shape of the mass distribution as a function 457
of Teff . The distribution now remains largely constant Asplund, M., Grevesse, N., & Sauval, A. J. 2005, Cosmic
around 0.60 M⊙ until ∼ 13,000 K where the well known Abundances as Records of Stellar Evolution and
high log g problem takes over. The new 3D hydrody- Nucleosynthesis, 336, 25
Aznar Cuadrado, R., Jordan, S., Napiwotzki, R., Schmid, H. M.,
namic models of convection presented in Tremblay et al. Solanki, S. K., & Mathys, G. 2004, A&A, 423, 1081
(2011b) have finally, and convincingly, solved this prob- Beauchamp, A. 1995, Ph.D. Thesis, Univ. de Montréal
lem by replacing the more approximate mixing length Beauchamp, A., Wesemael, F., & Bergeron, P. 1997, ApJS, 108,
theory used to this day. 559
We also wanted to test the reliability of our results by Beauchamp, A., Wesemael, F., Bergeron, P., Fontaine, G., Saffer,
R. A., Liebert, J., & Brassard, P. 1999, ApJ, 516, 887
comparing them with those of an independent study. The Beauchamp, A., Wesemael, F., Bergeron, P., Liebert, J., & Saffer,
Koester et al. (2009) analysis of DA white dwarfs from R. A. 1996, in ASP Conf. Ser. 96, Hydrogen-Deficient Stars, ed.
the SPY sample was ideal for this purpose since several S. Jeffery & U. Heber (San Francisco, CA: ASP), 295
Becklin, E. E., Farihi, J., Jura, M., Song, I., Weinberger, A. J., & Jordan, S., Heber, U., Engels, D., & Koester, D. 1993, A&A, 273,
Zuckerman, B. 2005, ApJ, 632, L119 L27
Bergeron, P., et al. 2011, arXiv:1105.5433 Kawka, A., Vennes, S., & Thorstensen, J. R. 2004, AJ, 127, 1702
Bergeron, P., Leggett, S. K., & Ruiz, M. T. 2001, ApJS, 133, 413 Kepler, S. O., Kleinman, S. J., Nitta, A., Koester, D.,
Bergeron, P., & Liebert, J. 2002, ApJ, 566, 1091 Castanheira, B. G., Giovannini, O., Costa, A. F. M., &
Bergeron, P., Greenstein, J. L., & Liebert, J. 1990a, ApJ, 361, 190 Althaus, L. 2007, MNRAS, 375, 1315
Bergeron, P., Ruiz, M.-T., & Leggett, S. K. 1993, ApJ, 407, 733 Kilic, M., Brown, W. R., Allende Prieto, C., Agüeros, M. A.,
Bergeron, P., Saffer, R. A., & Liebert, J. 1992, ApJ, 394, 228 Heinke, C., & Kenyon, S. J. 2011, ApJ, 727, 3
Bergeron, P., Wesemael, F., Beauchamp, A., Wood, M. A., Kilic, M., von Hippel, T., Leggett, S. K., & Winget, D. E. 2006,
Lamontagne, R., Fontaine, G., & Liebert, J. 1994, ApJ, 432, ApJ, 646, 474
305 Kilic, M., von Hippel, T., Leggett, S. K., & Winget, D. E. 2005,
Bergeron, P., Wesemael, F., Fontaine, G., & Liebert, J. 1990b, ApJ, 632, L115
ApJ, 351, L21 Kilkenny, D., Heber, U., & Drilling, J. S. 1988, South African
Bergeron, P., Wesemael, F., Lamontagne, R., Fontaine, G., Saffer, Astronomical Observatory Circular, 12, 1
R. A., & Allard, N. F. 1995, ApJ, 449, 258 Kilkenny, D., O’Donoghue, D., Koen, C., Stobie, R. S., & Chen,
Billères, M., Wesemael, F., Bergeron, P., & Beauchamp, A. 1997, A. 1997, MNRAS, 287, 867
ApJ, 488, 368 Koester, D. 2010, Mem. Soc. Astron. Italiana, 81, 921
Bochanski, J. J., West, A. A., Hawley, S. L., & Covey, K. R. 2007, Koester, D., et al. 2001, A&A, 378, 556
AJ, 133, 531 Koester, D., & Reimers, D. 1985, A&A, 153, 260
Bohlin, R. C., & Gilliland, R. L. 2004, AJ, 127, 3508 Koester, D., Voss, B., Napiwotzki, R., Christlieb, N., Homeier,
Bowyer, S., Lieu, R., Lampton, M., Lewis, J., Wu, X., Drake, D., Lisker, T., Reimers, D., & Heber, U. 2009, A&A, 505, 441
J. J., & Malina, R. F. 1994, ApJS, 93, 569 Kondo, M., Noguchi, T., & Maehara, H. 1984, Ann. Tokyo
Catalán, S., Isern, J., Garcı́a-Berro, E., Ribas, I., Allende Prieto, Astron. Obs., 20, 130
C., & Bonanos, A. Z. 2008, A&A, 477, 213 Lacombe, P., Wesemael, F., Fontaine, G., & Liebert, J. 1983,
Christlieb, N., Wisotzki, L., Reimers, D., Homeier, D., Koester, ApJ, 272, 660
D., & Heber, U. 2001, A&A, 366, 898 Lemke, M. 1997, A&AS, 122, 285
Cutri, R. M., et al. 2003, The IRSA 2MASS All-Sky Point Source Liebert, J., Bergeron, P., & Holberg, J. B. 2005, ApJS, 156, 47
Catalog, NASA/IPAC Infrared Science Archive (LBH05)
Dufour, P., et al. 2010, ApJ, 718, 647 Liebert, J., Bergeron, P., & Saffer, R.A. 1991, in 7th European
Dufour, P., Fontaine, G., Liebert, J., Schmidt, G. D., & Behara, Workshop on White Dwarfs, NATO ASI Series, ed. G. Vauclair
N. 2008, ApJ, 683, 978 & E. M. Sion (Dordrecht: Kluwer Academic Publishers), 409
Eisenstein, D. J., et al. 2006, ApJS, 167, 40 Liebert, J., Bergeron, P., Schmidt, G. D., & Saffer, R. A. 1993,
Farihi, J., Becklin, E. E., & Zuckerman, B. 2005, ApJS, 161, 394 ApJ, 418, 426
Farihi, J., Hoard, D. W., & Wachter, S. 2006, ApJ, 646, 480 Limoges, M.-M., & Bergeron, P. 2010, ApJ, 714, 1037
Ferrario, L., Wickramasinghe, D. T., Liebert, J., Schmidt, G. D., Lisker, T., Heber, U., Napiwotzki, R., Christlieb, N., Han, Z.,
& Bieging, J. H. 1997, MNRAS, 289, 105 Homeier, D., & Reimers, D. 2005, A&A, 430, 223
Finley, D. S., Koester, D., & Basri, G. 1997, ApJ, 488, 375 Marsh, M. C., et al. 1997a, MNRAS, 286, 369
Fleming, T. A., Snowden, S. L., Pfeffermann, E., Briel, U., & Marsh, M. C., et al. 1997b, MNRAS, 287, 705
Greiner, J. 1996, A&A, 316, 147 Martin, B., & Wickramasinghe, D. T. 1984, MNRAS, 206, 407
Fontaine, G., & Brassard, P. 2008, PASP, 120, 1043 Maxted, P. F. L., & Marsh, T. R. 1999, MNRAS, 307, 122
Fontaine, G., Brassard, P., & Bergeron, P. 2001, PASP, 113, 409 Maxted, P. F. L., Marsh, T. R., & Moran, C. K. J. 2000,
Gänsicke, B. T., Euchner, F., & Jordan, S. 2002, A&A, 394, 957 MNRAS, 319, 305
Gänsicke, B. T., Koester, D., Girven, J., Marsh, T. R., & Steeghs, McCook, G. P., & Sion, E. M. 1999, ApJS, 121, 1 (MS99)
D. 2010, Science, 327, 188 Monet, D. G., et al. 2003, AJ, 125, 984
Gänsicke, B. T., Koester, D., Marsh, T. R., Rebassa-Mansergas, Noguchi, T., Maehara, H., & Kondo, M. 1980, Ann. Tokyo
A., & Southworth, J. 2008, MNRAS, 391, L103 Astron. Obs., 18, 55
Gianninas, A., Bergeron, P., Dupuis, J., & Ruiz, M. T. 2010, Pereira, C., Bergeron, P., & Wesemael, F. 2005, ApJ, 623, 1076
ApJ, 720, 581 Perryman, M. A. C., & ESA 1997, ESA Special Publication, 1200
Gianninas, A., Bergeron, P., & Fontaine, G. 2007, in ASP Conf. Platais, I., Kozhurina-Platais, V., Barnes, S., Girard, T. M.,
Ser. 372, 15th European Workshop on White Dwarfs, ed. R. Demarque, P., van Altena, W. F., Deliyannis, C. P., Horch, E.
Napiwotzki & M. R. Burleigh, (San Francisco, CA:ASP), 577 2001, AJ, 122, 1486
Gianninas, A., Bergeron, P., & Fontaine, G. 2006, AJ, 132, 831 Pounds, K. A., et al. 1993, MNRAS, 260, 77
Gianninas, A., Bergeron, P., & Fontaine, G. 2005, ApJ, 631, 1100 Press, W. H., Flannery, B. P., Teukolsky, S. A., & Vetterling, W.
Gianninas, A., Bergeron, P., & Ruiz, M. T. 2009, J. Phys. Conf. T. 1986, Numerical Recipes (Cambridge: Cambridge Univ.
Ser., 172, 012021 Press)
Gianninas, A., Dufour, P. & Bergeron, P. 2004, ApJ, 617, L57 Putney, A. 1997, ApJS, 112, 527
Giclas, H. L., Burnham, R., & Thomas, N. G. 1965, Lowell Putney, A., & Jordan, S. 1995, ApJ, 449, 863
Observatory Bulletin, 6, 155 Rebassa-Mansergas, A., Gänsicke, B. T., Schreiber, M. R.,
Glenn, J., Liebert, J., & Schmidt, G. D. 1994, PASP, 106, 722 Koester, D., & Rodrı́guez-Gil, P. 2010, MNRAS, 402, 620
Green, R. F., Schmidt, M., & Liebert, J. 1986, ApJS, 61, 305 Saffer, R. A., Liebert, J., & Olszewski, E. M. 1988, ApJ, 334, 947
Greenstein, J. L. 1986, ApJ, 304, 334 Saffer, R. A., Liebert, J., Wagner, R. M., Sion, E. M., &
Hamada, T., & Salpeter, E. E. 1961, ApJ, 134, 683 Starrfield, S. G. 1989, AJ, 98, 668
Harris, H. C., et al. 2007, AJ, 133, 631 Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500,
Heber, U., Napiwotzki, R., Lemke, M., & Edelmann, H. 1997, 525
A&A, 324, L53 Schmidt, G. D., & Smith, P. S. 1994, ApJ, 423, L63
Heller, R., Homeier, D., Dreizler, S., & Østensen, R. 2009, A&A, Schmidt, G. D., West, S. C., Liebert, J., Green, R. F., &
496, 191 Stockman, H. S. 1986, ApJ, 309, 218
Holberg, J. B., & Bergeron, P. 2006, AJ, 132, 1221 Seaton, M. J. 1979, MNRAS, 187, 73P
Holberg, J. B., Sion, E. M., Oswalt, T., McCook, G. P., Foran, S., Silvestri, N. M., et al. 2006, AJ, 131, 1674
& Subasavage, J. P. 2008, AJ, 135, 1225 Silvestri, N. M., Oswalt, T. D., Wood, M. A., Smith, J. A., Reid,
Homeier, D., Koester, D., Hagen, H.-J., Jordan, S., Heber, U., I. N., & Sion, E. M. 2001, AJ, 121, 503
Engels, D., Reimers, D., & Dreizler, S. 1998, A&A, 338, 563 Smith, R. J., Croom, S. M., Boyle, B. J., Shanks, T., Miller, L., &
Hubeny, I., & Lanz, T. 1995, ApJ, 439, 875 Loaring, N. S. 2005, MNRAS, 359, 57
Hügelmeyer, S. D., Dreizler, S., Homeier, D., Krzesiǹski, J., Strand, K. A., Dahn, C. C., & Liebert, J. W. 1976, BAAS, 8, 506
Werner, K., Nitta, A., & Kleinman, S. J. 2006, A&A, 454, 617 Stroeer, A., Heber, U., Lisker, T., Napiwotzki, R., Dreizler, S.,
John, T. L. 1994, MNRAS, 269, 871 Christlieb, N., & Reimers, D. 2007, A&A, 462, 269
Subasavage, J. P., Henry, T. J., Bergeron, P., Dufour, P., Hambly, Voss, B. 2006, Ph.D. Thesis, Univ. Kiel
N. C., & Beaulieu, T. D. 2007, AJ, 134, 252 Voss, B., Koester, D., Østensen, R., Kepler, S. O., Napiwotzki,
Tremblay, P.-E., & Bergeron, P. 2009, ApJ, 696, 1755 (TB09) R., Homeier, D., & Reimers, D. 2006, A&A, 450, 1061
Tremblay, P.-E., Bergeron, P., & Gianninas, A. 2011a, ApJ, 730, Voss, B., Koester, D., Østensen, R., Napiwotzki, R., Homeier, D.,
128 & Reimers, D. 2007, in ASP Conf. Ser. 372, 15th European
Tremblay, P.-E., Bergeron, P., Kalirai, J. S., & Gianninas, A. Workshop on White Dwarfs, ed. R. Napiwotzki & M. R.
2010, ApJ, 712, 1345 Burleigh, (San Francisco, CA:ASP) 583
Tremblay, P.-E., Ludwig, H.-G., Steffen, M., Bergeron, P., & Werner, K. 1996, ApJ, 457, L39
Freytag, B. 2011b, A&A, 531, L19 Wesemael, F., et al. 1994, ApJ, 429, 369
van Altena, W. F., Lee, J. T., & Hoffleit, E. D. 1994, The General Wickramasinghe, D. T., & Ferrario, L. 2000, PASP, 112, 873
Catalogue of Trigonometric Parallaxes (New Haven: Yale Univ. Wolff, B., Jordan, S., Koester, D., & Reimers, D. 2000, A&A,
Obs.) 361, 629
Vanlandingham, K. M., et al. 2005, AJ, 130, 734 Wood, M. A. 1995, in Proceedings of the 9th European Workshop
Vennes, S., Christian, D. J., & Thorstensen, J. R. 1998, ApJ, 502, on White Dwarfs, eds. D. Koester & K. Werner (Berlin:
763 Springer), 41
Vennes, S., Ferrario, L., & Wickramasinghe, D. T. 1999, MNRAS, Zuckerman, B., Koester, D., Melis, C., Hansen, B. M., & Jura, M.
302, L49 2007, ApJ, 671, 872
Vennes, S., Smith, R. J., Boyle, B. J., Croom, S. M., Kawka, A., Zuckerman, B., Koester, D., Reid, I. N., Hünsch, M. 2003, ApJ,
Shanks, T., Miller, L., & Loaring, N. 2002, MNRAS, 335, 673 596, 477
Vennes, S., Thejll, P. A., Galvan, R. G., & Dupuis, J. 1997, ApJ,
480, 714
Vennes, S., Thejll, P. A., Wickramasinghe, D. T., & Bessell, M. S.
1996, ApJ, 467, 782

Gianninas 2011

Uploaded by

Copyright:

Available Formats

Gianninas 2011

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Gianninas 2011

Uploaded by

Copyright:

Available Formats

Accepted for publication in the Astrophysical Journal

Preprint typeset using LATEX style emulateapj v. 5/2/11

A SPECTROSCOPIC SURVEY AND ANALYSIS OF BRIGHT, HYDROGEN-RICH WHITE DWARFS

1. INTRODUCTION particularly well studied despite their relative bright-

WD Name ST (MS99) ST (this work) Notes

4.3. DAB white dwarfs

trace amounts of calcium will have no effect whatsoever

whole and obtain atmospheric parameters for both stars.

4.7. Peculiar Objects

their observed spectrum clearly suggested a cooler spec-

our observations were obtained at high airmass. This

WD Name ST Teff (K) log g M /M⊙ MV D (pc) Notes

We also provide an updated spectroscopic classification 5.2. Comparison with SPY

stand this discrepancy. In particular, they point out that

Figure 29. Comparison of absolute magnitudes derived from par-

with those obtained from parallax measurements. The

clear: these low-mass white dwarfs represent a separate

Figure 33. Teff - log g distribution for DA white dwarfs with

Figure 32. Luminosity function of all DA stars in the complete

You might also like