1: \documentclass[12pt,preprint]{aastex}
2:
3: \newcommand{\mv} {M_V}
4: \newcommand{\halpha} {H$\alpha$}
5: \newcommand{\hbeta} {H$\beta$}
6: \newcommand{\hgamma} {H$\gamma$}
7: \newcommand{\hdelta} {H$\delta$}
8: \newcommand{\hepsilon} {H$\epsilon$}
9: \newcommand{\JHKs} {JHK_S}
10: \newcommand{\msun} {$M_{\sun}$}
11: \newcommand{\rsun} {$R_{\sun}$}
12: \newcommand{\lsun} {$L_{\sun}$}
13: \newcommand{\Te} {T_{\rm eff}}
14: \newcommand{\logg} {\log g}
15: \begin{document}
16:
17: \title{The Ratio of Helium- to Hydrogen-Atmosphere White Dwarfs:
18: Direct Evidence for Convective Mixing}
19:
20: \author{P.-E. Tremblay and P. Bergeron}
21: \affil{D\'epartement de Physique, Universit\'e de Montr\'eal, C.P.~6128,
22: Succ.~Centre-Ville, Montr\'eal, Qu\'ebec, Canada, H3C 3J7.}
23: \email{tremblay@astro.umontreal.ca, bergeron@astro.umontreal.ca}
24:
25: \begin{abstract}
26:
27: We determine the ratio of helium- to hydrogen-atmosphere white dwarf
28: stars as a function of effective temperature from a model atmosphere
29: analysis of the infrared photometric data from the Two Micron All Sky
30: Survey combined with available visual magnitudes. Our study
31: surpasses any previous analysis of this kind both in terms of the
32: accuracy of the $\Te$ determinations as well as the size of the
33: sample. We observe that the ratio of helium- to hydrogen-atmosphere
34: white dwarfs increases gradually from a constant value of $\sim0.25$
35: between $\Te=15,000$~K and 10,000~K to a value twice as large in the
36: range $10,000 > \Te > 8000$ K, suggesting that convective mixing,
37: which occurs when the bottom of the hydrogen convection zone reaches
38: the underlying convective helium envelope, is responsible for this
39: gradual transition. The comparison of our results with an approximate
40: model used to describe the outcome of this convective mixing process
41: implies hydrogen mass layers in the range $M_{\rm H}/M_{\rm
42: tot}=10^{-10}$ to $10^{-8}$ for about 15\% of the DA stars that
43: survived the DA to DB transition near $\Te\sim 30,000$~K, the remainder
44: having presumably more massive layers above $M_{\rm H}/M_{\rm
45: tot}\sim10^{-6}$.
46:
47: \end{abstract}
48:
49: \keywords{Convection --- stars: atmospheres --- stars: interiors --- white dwarfs}
50:
51: \section{INTRODUCTION}
52:
53: The Two Micron All Sky Survey (2MASS) represents one the largest set
54: of homogeneous infrared photometric data for all areas of the sky
55: applicable to white dwarf stars of various spectral types. Indeed, the
56: 2MASS survey contains $JHK_S$ magnitudes for almost all point sources
57: of the sky up to the magnitude thresholds of $J \sim 16.8$, $H \sim
58: 16.5$, and $K_S\sim 16.2$ for detections with any formal
59: uncertainty \citep[see for instance Fig.~1 of][]{tremblay07}. This yields a nearly
60: complete magnitude limited survey of white dwarfs that allows us to
61: trace a portrait of the local population of white dwarfs, assuming we
62: can properly identify the white dwarfs in this survey. To that effect,
63: the most complete compilation is that of the Villanova White Dwarf
64: Catalog\footnote{http://www.astronomy.villanova.edu/WDCatalog/index.html}
65: (WDC), which contains more than 5500 spectroscopically identified
66: white dwarfs reported in the literature. The interest of such a
67: sizeable sample of homogeneous infrared photometric observations is to
68: reach an unprecedented level in the statistical analysis of white
69: dwarf stars. In addition, \citet{tremblay07} have demonstrated the
70: overall reliability of the 2MASS Point-Source Catalog (PSC) photometry
71: for the purpose of detailed comparisons with the predictions from
72: white dwarf model atmospheres.
73:
74: Studies of the spectral evolution of white dwarfs attempt to
75: understand how the surface composition and the corresponding spectral
76: signature evolve along the white dwarf cooling sequence (see
77: \citealt{fontaine87,fontaine91} for a review). Because of the intense
78: gravitational field present in these stars, hydrogen will float to the
79: surface in the absence of competing mechanisms, while heavier elements
80: will sink below the photospheric regions. Indeed, the majority of
81: white dwarfs have optical spectra that are completely dominated by
82: strong hydrogen Balmer lines --- the DA stars --- with
83: hydrogen-dominated atmospheres. We also find that almost 25\% of white
84: dwarfs have helium-dominated atmospheres, mostly DO, DB, DQ, and DZ
85: stars, and some of the DC stars as well. The fact that not all white
86: dwarfs have hydrogen-rich atmospheres imply that either some stars are
87: born with hydrogen-deficient atmospheres and remain that way
88: throughout their evolution, or alternatively, that several physical
89: mechanisms such as diffusion, accretion, convection, radiation
90: pressure, and stellar winds are competing with gravitational settling
91: in determining their surface composition as they cool off. This
92: particular issue can be resolved by analyzing the distribution of
93: spectral types as a function of effective temperature (or any
94: temperature index) for a large sample of white dwarf stars, in a way
95: similar to the analysis of \citet{sion84} nearly twenty five years
96: ago.
97:
98: Before discussing the physical processes that will affect the spectral
99: evolution of hydrogen- and helium-atmosphere white dwarfs, it is
100: important first to consider the constraints brought about by the
101: studies of the hottest white dwarfs. The canonical mass fractions of
102: light elements in white dwarfs considering the mass threshold for
103: residual nuclear burning are $M_{\rm He}/M_{\rm tot} \sim 10^{-2}$ and
104: $M_{\rm H}/M_{\rm tot} \sim 10^{-4}$. However, multiple excursions on
105: the AGB nuclear burning phase, during which very late helium flashes
106: could remove essentially all of the hydrogen \citep{werner06}, could also produce
107: white dwarfs with total hydrogen masses much lower than the canonical
108: value of $10^{-4}$. The first class would correspond to the
109: progenitors of the hottest DA stars while the second class would
110: represent the progenitors of hot white dwarf stars with
111: hydrogen-deficient atmospheres --- the PG 1159 stars at very high
112: effective temperatures, and the DO stars at slightly lower
113: temperatures, whose spectra are dominated by He\textsc{ii} lines. It
114: has traditionally been believed that by the time DO stars cool off to
115: $\Te\sim45,000$~K, residual amounts of hydrogen ($M_{\rm H}/M_{\rm
116: tot} \sim 10^{-16}$) thoroughly mixed in their helium-rich envelope
117: would gradually accumulate to the surface, turning all
118: helium-atmosphere white dwarfs into DA stars
119: \citep{fontaine87,fontaine91}. This scenario would account
120: for the existence of the so-called DB gap, a range of temperature
121: between $\Te=45,000$~K and 30,000~K where all white dwarfs have a DA spectral
122: type. However, the recent discovery in the Sloan Digital Sky Survey
123: of several DB stars within the gap \citep{eisenstein06a} implies that the total mass
124: of hydrogen left in the envelope of DO stars can be even smaller than
125: previously believed. It is then generally admitted that the
126: significant increase in the number of helium-atmosphere DB white
127: dwarfs below $\Te\sim30,000$~K can be explained in terms of the convective
128: dilution of the superficial hydrogen atmosphere by the underlying
129: helium convective envelope, provided that the hydrogen layer is
130: sufficiently thin
131: \citep[$M_{\rm H}/M_{\rm tot} \sim 10^{-15}$,][]{macdonald91}.
132:
133: At even lower effective temperatures ($\Te\lesssim12,000$~K),
134: hydrogen-atmosphere white dwarfs get a second opportunity to evolve
135: into stars with helium-dominated atmospheres when the superficial
136: hydrogen layer becomes convective over a significant fraction of its
137: depth. This process is illustrated in Figure \ref{fg:f1} where the
138: extent of the hydrogen convection zone is displayed as a function of
139: decreasing effective temperature for a 0.6 \msun\ DA white dwarf,
140: based on evolutionary models with thick hydrogen layers similar to
141: those described by \citet{fon01} and kindly provided to us by
142: G.~Fontaine and P.~Brassard. These calculations show that if the
143: hydrogen envelope is thin enough, the bottom of the hydrogen
144: convection zone may eventually reach the underlying and more massive
145: convective helium layer, resulting in a mixing of the hydrogen and
146: helium layers
147: \citep{strittmatter71,shipman72,baglin73,koester76,vauclair77}. Figure \ref{fg:f1}
148: also indicates that the effective temperature at which this mixing
149: occurs will depend on the thickness of the hydrogen envelope. The
150: thicker the envelope, the lower the mixing temperature; if the
151: hydrogen layer is more massive than $M_{\rm H}/M_{\rm tot} \sim
152: 10^{-6}$, mixing will never occur.
153:
154: The simplest physical model that can be used to describe this
155: convective mixing process is to assume that hydrogen and helium are
156: homogeneously mixed. Since the helium convection zone is much more
157: massive ($M_{\rm He-conv}/M_{\rm tot}\sim10^{-6}$) than the hydrogen
158: layer when mixing occurs, it is generally assumed that a DA star would
159: be transformed into a helium-atmosphere white dwarf of type DB, DQ,
160: DZ, or DC with only a trace abundance of hydrogen. More detailed
161: calculations discussed by \cite{fontaine91} confirm these
162: predictions. If such a process takes place in cool white dwarfs, we
163: then expect the ratio of helium- to hydrogen-atmosphere white dwarfs
164: to increase at low effective temperatures. A comparison between the
165: observed ratio as a function of $\Te$ and the theoretical expectations
166: would make it possible to estimate the thickness of the hydrogen
167: layers in DA white dwarfs. In turn, these determinations could be
168: compared to independent measurements of the hydrogen layer mass
169: inferred from ZZ Ceti asteroseismology.
170:
171: The best approach to study the occurrence of this convective mixing
172: process begins with the statistical analysis of large samples of cool
173: white dwarfs for which we can determine the main atmospheric
174: constituent and the effective temperature, or any analogue temperature
175: index such as the absolute $V$ magnitude. \citet{sion84} was the first
176: to study this problem by estimating the non-DA to DA ratio at low
177: effective temperatures using a proper motion sample of 695 white
178: dwarfs drawn from the catalog of spectroscopically identified white
179: dwarfs available at that time. The results from Figure 1 of
180: \cite{sion84} are reproduced here in Figure \ref{fg:f2} in terms of
181: the non-DA to DA ratio (rather than the absolute number of DA and
182: non-DA stars) as a function of the absolute visual magnitude $M_V$.
183: Looking at the results, we notice a first increase in this ratio at
184: $M_V>11.25$, which corresponds to $\Te\sim15,000$ K for a 0.6 \msun\
185: white dwarf, a temperature much in excess of the theoretically
186: predicted value for the convective mixing process. A second increase
187: occurs at $M_V\sim 12.5$, or $\Te \sim 10000$ K, but the observed
188: ratio drops suddenly afterwards and this increase is probably not
189: significant. Then a third increase occurs at $M_V> 13$, or
190: $\Te<8000$~K. Clearly, the evidence based on these results are not
191: exactly convincing, and our current picture of the situation below
192: 12,000 K is at best sketchy. The global portrait did not change
193: significantly since the analysis of
194: \citet{sion84}, and this remains the only available proof quoted in the literature
195: of the transformation of some DA stars into non-DA stars
196: (see also \citealt{greenstein86}).
197:
198: Our goal is to improve upon the analysis of \citet{sion84} by using
199: the 2MASS photometric sample combined with the WDC database to
200: identify white dwarfs and determine their atmospheric composition, and
201: most importantly to obtain more accurate temperature determinations
202: using a full model atmosphere analysis of the $VJHK_S$ photometry. We
203: first describe in \S~2 the 2MASS photometric white dwarf sample used
204: in our analysis. The determination of the atmospheric parameters is
205: then presented in \S~3 and the uncertainties related to our approach
206: are discussed at length in \S~4. The evolution of the ratio of helium-
207: to hydrogen-atmosphere white dwarfs as a function of effective
208: temperature is then determined in \S~5. Our conclusions follow in
209: \S~6.
210:
211: \section{DEFINITION OF THE 2MASS WHITE DWARF SAMPLE}
212:
213: Our starting point is the Villanova White Dwarf Catalog (WDC) of
214: spectroscopically identified white dwarfs. We took the most recent
215: electronic version of the catalog (August 2006), which includes a
216: total of 5557 white dwarfs. We made a first selection by taking only
217: the objects with at least one published visual magnitude\footnote{We
218: have neglected magnitudes prior to 1972, photographic magnitudes, $B$
219: magnitudes, as well as those flagged as uncertain.} and with at least
220: one measurement brighter than $V=18.5$. Figure
221: \ref{fg:f3} shows the predicted $V$ magnitude for which $H_{\rm 2MASS}$ = 16.5 --- a
222: representative value for the 2MASS detection limit in this band --- as
223: a function of $\Te$ for our pure hydrogen and pure helium model
224: atmospheres at $\log g=8$. Only white dwarfs located below these
225: curves at a given effective temperature can be detected by 2MASS in
226: the $J$ and $H$ bandpasses. For white dwarfs with $\Te >$ 5000 K, the
227: $V < 18.5$ criterion is sufficient to find a maximum number of white
228: dwarfs. Finally, since our study is aimed at tracing the spectral
229: evolution of cool white dwarfs, we neglected all objects with early
230: spectral classifications (e.g., PG 1159, DO, etc.) or for which the
231: spectral classification was unavailable altogether. Thus, a total of
232: 1473 remaining white dwarf stars define our initial WDC sample.
233:
234: We then searched the 2MASS PSC for $JHK_S$ photometric observations of
235: all white dwarfs from this initial WDC sample. We used the GATOR batch
236: file tool and a $20''$ search window centered on a set of improved
237: coordinates measured by J.~B.~Holberg (2005, private
238: communication). In most instances, multiple sources were found within
239: the search window and we unambiguously identified each object by
240: comparing the 2MASS atlas with the finding charts available from the
241: online version of the Villanova WDC. For several objects close to the
242: detection limits, only one or two magnitudes are detected with formal
243: uncertainties. We neglect all objects with only one detected magnitude
244: since in such cases, only one degree of freedom would be available in
245: our minimization procedure described in \S~3. We recovered 940
246: objects in the 2MASS PSC from the 1473 objects contained in our
247: initial WDC sample. The remaining 533 objects are eliminated for the
248: following reasons: 508 are too faint to be detected by 2MASS in at
249: least two bands, 15 are bright enough in the 2MASS atlas but are not
250: part of the 2MASS PSC, and 10 could not be unambiguously identified
251: from the comparison of the 2MASS atlas and the published finding
252: charts.
253:
254: A spectral energy distribution is then built for each object by
255: combining the 2MASS $JHK_S$ photometry with the $V$ magnitude (or
256: Str\"omgren $y$) taken from the WDC. In addition, about 10\% of our
257: $V$ magnitudes come from unpublished USNO photometry
258: \citep{dahn07}. When more than one visual magnitude is available, we
259: take an unweighted average, unless we had reasons to believe that a
260: given measurement was less accurate than others. We also assume a
261: typical uncertainty of 0.05 mag for the visual magnitudes.
262:
263: \section{ATMOSPHERIC PARAMETER DETERMINATIONS}
264:
265: We classify the white dwarfs in our sample into two broad categories,
266: those with hydrogen atmospheres (DA stars, including DAB and DAZ
267: stars) and those with helium atmospheres (DB, DZ, and DQ stars as well
268: as helium-atmosphere objects with weak hydrogen features such as
269: Ross 640 and L745-46A; the case of DC stars is discussed further
270: below). When the spectral classification is ambiguous, we check in the
271: literature to confirm the presence of hydrogen lines compatible with
272: the estimated effective temperature. We then fit the $VJHK_S$ (or
273: $VJH$) photometry with our grids of pure hydrogen or pure helium model
274: atmospheres described in
275: \citet{tremblay07} and references therein. Our fitting technique is
276: similar to that outlined in \citet{bergeron97,bergeron01}. Briefly,
277: the magnitudes are converted into monochromatic fluxes using the zero
278: points defined in \citet{holberg06} for photon counting devices, but
279: using the transmission function of \citet{bessell90} for $V$ and the
280: 2MASS transmission functions of \citet{cohen03} for $JHK_S$. The
281: resulting fluxes are then compared with those predicted from the model
282: atmospheres, properly averaged over the same bandpasses. The
283: effective temperature and the solid angle $\pi(R/D)^2$ (with $R$ the
284: radius of the star and $D$ the distance to Earth) are considered free
285: parameters in the $\chi ^2$ minimization procedure. We assume a value
286: of $\log g=8.0$ throughout. The surface gravity has very little
287: effect on the predicted $(V-J)$ color indices of white dwarfs. The
288: results shown in Figure \ref{fg:f4} indicate that a difference of 1
289: dex in $\log g$ corresponds to a color difference smaller than 0.05
290: mag at all temperatures. This slight difference, which is of the order
291: of our color uncertainties, allows us to fix the surface gravity at
292: $\log g=8.0$ for all objects. Since the mass distributions of
293: hydrogen- and helium-atmosphere white dwarfs have to a good
294: approximation the same mean value
295: \citep{bergeron92,beauchamp95,liebert05,voss07}, this simplifying
296: assumption should not affect the results of our analysis
297: significantly.
298:
299: For the featureless DC stars, we consider both hydrogen- and helium-atmosphere
300: solutions. If the hydrogen solution is above $\Te=5000$~K, we adopt
301: the helium solution since in principle, H$\alpha$ should have been
302: detected spectroscopically according to the photometric analyses of
303: \citet{bergeron97,bergeron01}. However, if the hydrogen solution is
304: below this temperature, it is impossible to determine the main
305: atmospheric constituent from the restricted available photometry, and
306: therefore we set the limit of our analysis to $\Te>5000$~K.
307:
308: There is also a significant fraction of binary white dwarfs in our
309: sample. If the companion is a cool main-sequence star, the observed
310: $JHK_S$ fluxes are substantially larger than those predicted by the
311: models, and the $(J-H)$ color index takes a large positive value
312: \citep{tremblay07}, which typically yields a bad fit to the energy
313: distribution. We eliminated 90 binary candidates satisfying at least
314: two of the following criteria\footnote{16 of these did not exactly
315: meet two criteria but they have poor fits and atmospheric parameters
316: that are incompatible with the published spectral type.}: (1) a
317: photometric temperature below 6000 K incompatible with the spectral
318: type, (2) an observed flux at $H$ larger than that at $J$, and (3) a
319: flux at $J$ larger than ten times the flux at $V$. We have also
320: eliminated 9 binary systems that are partially resolved in the 2MASS
321: atlas, but whose colors in the the 2MASS PSC are likely to be
322: contaminated. The large majority of our binary candidates have been
323: previously identified as such in the literature.
324:
325: Sample fits for eight bright ($11 < V < 13.5$) and well studied white
326: dwarfs in our sample are displayed in Figure \ref{fg:f5}. We notice in
327: each case that the best fit is in excellent agreement with the
328: $VJHK_S$ photometry, for both hydrogen- or helium-atmosphere white
329: dwarfs. We also compare in Table 1 the photometric temperatures for
330: these objects with independent spectroscopic or photometric effective
331: temperature determinations reported in the literature. This comparison
332: indicates that our technique applied to cool white dwarfs of various
333: spectral types yields $\Te$ values with a precision of the order of
334: 5\% compared to other methods, provided we have accurate photometric
335: data.
336:
337: \section{ATMOSPHERIC PARAMETER UNCERTAINTIES}
338:
339: We attempt in this section to better quantify the uncertainties of our
340: effective temperature determinations in order to determine the optimal
341: bin widths for the histograms used in our study of convective mixing,
342: presented in section \S~5. These uncertainties can be first quantified
343: by looking at the rate of change of the $(V-J)$ color index as a
344: function of $\Te$. We show in Figure \ref{fg:f6} the effective
345: temperature variation, $\Delta \Te$, for a corresponding change of the
346: $(V-J)$ color index of $\Delta (V-J)=0.08$, a value that defines the
347: mean uncertainty of this color index in our sample. We notice that the
348: $\Te$ uncertainties increase rapidly towards high effective
349: temperatures, and for this reason, we restrict our analysis to
350: $\Te<15,000$~K.
351:
352: We compare in Figure \ref{fg:f7} our photometric $\Te$ determinations
353: for 133 cool white dwarfs in common with the analyses of
354: \citet{bergeron97,bergeron01}, which are based on independent
355: $BVRIJHK$ photometric data and in several cases, trigonometric
356: parallax measurements from which surface gravities can be
357: determined. To be internally consistent, we have revised their
358: atmospheric parameters using our own fitting procedure based on
359: the slightly different synthetic photometric calibration of \citet{holberg06}.
360: As well, we consider only the stars that have been observed in at
361: least the $J$ and $H$ bands. We can see that the agreement is
362: excellent in this temperature range and that the $VJHK_S$ photometry
363: is sufficient to constrain the $\Te$ values with the precision sought
364: in our analysis.
365:
366: We present in Figure \ref{fg:f8} a similar comparison of our
367: photometric temperatures with independent values obtained from line
368: profile fitting of DA and DB stars. The spectroscopic
369: temperatures for the 198 DA stars in common with our sample are taken
370: from the ongoing spectroscopic survey of A.~Gianninas (2007, private
371: communication). We notice in this case that both samples are generally
372: in good agreement, but several discrepancies are also observed where
373: the spectroscopic temperatures are significantly larger than the
374: photometric values ($\Delta \Te>0$), in particular at high
375: temperatures. A first explanation for this discrepancy is that above
376: $\Te > 11,000$~ K, the $(V-J)$ color index is a slowly varying
377: function of effective temperature (see Fig. \ref{fg:f6}), and thus
378: white dwarfs with large temperature uncertainties contaminate this
379: particular region. This is confirmed by the fact that most white
380: dwarfs showing a large discrepancy are near the 2MASS detection limit
381: (objects not detected at $K_S$ in 2MASS are shown by open circles in
382: Fig. \ref{fg:f8}). Therefore, we add the constraint that only white
383: dwarfs detected in all three $JHK_S$ bands are allowed in our analysis
384: for $\Te>11,000$~K. The second group of objects showing a positive
385: $\Delta \Te$ are unresolved double degenerate systems, three of which
386: are identified in Figure \ref{fg:f8} by filled squares. In all cases
387: the photometric temperature is substantially lower than the
388: spectroscopic value, and both spectroscopic and photometric fits are
389: excellent. It is therefore nearly impossible to eliminate double white
390: dwarf systems from our analysis. There remain 10 other objects with a
391: $\Delta \Te$ larger than 20\% for which the photometric fit is good
392: and there is no clear explanation for a discrepancy. These could
393: also correspond to unresolved degenerate binaries, although a
394: closer inspection rather suggests that the $V$ magnitudes, which are
395: old measurements, are probably in error.
396:
397: In the bottom panel of Figure \ref{fg:f8} we compare our photometric
398: temperatures for 18 DB stars with those determined spectroscopically
399: by \citet{beauchamp95,beauchamp96}. We see that the agreement is excellent
400: although the spectroscopic temperatures are systematically larger than
401: the photometric values by about 5\%. This small shift could be due to
402: the neglect of small traces of hydrogen or metals in both analyses, or
403: to inaccuracies in the treatment of the broadening theory for neutral
404: helium lines at low effective temperatures, which will mainly affect
405: the spectroscopic temperatures \citep[see, e.g.,][]{kepler07}. The
406: effective temperature dispersions observed in the figures above
407: indicate that histograms with bin widths of 1000 K below
408: $\Te=11,000$~K and of 2000~K above this temperature are
409: acceptable. These values are reproduced in Figures \ref{fg:f6} to
410: \ref{fg:f8}.
411:
412: \section{RESULTS ON CONVECTIVE MIXING}
413:
414: We present in Table 2 the number of hydrogen- and helium-atmosphere
415: white dwarfs in different effective temperature bins for our final
416: 2MASS sample of 447 objects (after the temperature cutoff at
417: $\Te=15,000$ and $5000$~K and the removal of 2MASS data with low
418: accuracy above $\Te=11,000$~K have been applied). This sample is nearly
419: twice as large as that used in the analysis of
420: \citet{sion84} in the same range of effective temperature.
421: We also display in Figure \ref{fg:f9} the corresponding histogram of
422: the number of hydrogen- and helium-atmosphere white dwarfs as a
423: function of $\Te$. The total number of stars in
424: each bin below $\Te=11,000$~K is comparable, a result that can be
425: explained by the competition between the increasing luminosity
426: function and the decreasing detection probability of cooler, fainter
427: objects. We further observe that hydrogen-atmosphere DA white dwarfs
428: are always the dominant type. Consequently, convective mixing --- if
429: this process occurs at all --- should affect only a small fraction of
430: cool DA stars.
431:
432: For a magnitude limited sample --- here the $H$ magnitude ---, we
433: probe a limited volume of space. This volume is defined in terms of
434: the distance at which a given white dwarf can be detected at the
435: limiting magnitude of the survey. However, this distance is different
436: for hydrogen- and helium-atmosphere white dwarfs because they have
437: different continuum opacity sources and corresponding monochromatic
438: fluxes. Figure \ref{fg:f10} presents the ratio of the volume probed by
439: hydrogen and helium white dwarfs as a function of effective
440: temperature for a magnitude limit in the $H$ bandpass. While the
441: assumption of a fixed value of $\log g$ was appropriate for
442: determining the effective temperatures, this may not be the case for
443: the volume effect discussed here. Indeed, white dwarfs stars with
444: large radii (or low masses) can be detected at larger distances. The
445: differences in the mass distributions of DA and DB stars could
446: therefore affect the ratio \citep[see also][]{eisenstein06a}. While
447: the results of \cite{beauchamp96} suggest that the mass distribution
448: of DB stars is narrower than that of DA stars, the recent results of
449: \citet{voss07} reveal instead that both distributions are comparable.
450: Hence the small residual differences in the low and high mass tails
451: are not expected to affect our results significantly.
452:
453: To take into account the volume effect, we henceforth correct the
454: ratio of the number of helium- to hydrogen-atmosphere white dwarfs,
455: $N_{\rm He-atm}/N_{\rm H-atm}$, by the corresponding factor given in
456: Figure 11. This ratio corrected for the volume effect is reported in
457: Table 2 and displayed in Figure \ref{fg:f11} as a function of
458: effective temperature. The error bars are statistical and they are
459: mainly governed by the limited number of helium-atmosphere objects in
460: each bin. We also show in the same figure the ratio obtained by
461: shifting the temperature bins in Figure \ref{fg:f9} by 500 K.
462:
463: We can immediately notice one important trend in the observed ratio of
464: helium- to hydrogen-atmosphere white dwarfs displayed in Figure
465: \ref{fg:f11}. At high effective temperatures, this ratio takes a
466: constant value of $\sim0.25$ and it shows a gradual increase from
467: $\Te\sim10,000$~K to a fairly constant value twice as large
468: ($\sim0.5$) at 8000~K. The only physical mechanism that could account
469: for this increase is the convective mixing of thin hydrogen layer with
470: the more massive underlying helium envelope. These results further
471: imply that about 15\% of all DA stars\footnote{The transformation
472: implies the following change of ratio $N_{\rm He-atm}/N_{\rm
473: H-atm}$~$\rightarrow$~$(N_{\rm He-atm}+N_{\rm H-transformed})/(N_{\rm
474: H-atm}-N_{\rm H-transformed})$.} between $\Te=15,000$~K and 10,000~K
475: have turned into helium-atmosphere white dwarfs --- either DQ, DZ (or
476: DZA), or DC --- by the time they reach 8000~K. Note that the
477: results of our analysis become too uncertain above the temperature
478: range displayed here (see Fig.~\ref{fg:f6}).
479:
480: We can gain a better understanding of our results by looking at the
481: simple theoretical model of convective mixing discussed in the
482: Introduction. As the white dwarf cools off, the bottom of the hydrogen
483: convection zone will eventually reach the helium convection zone,
484: provided the hydrogen layer is thin enough (see Fig.~\ref{fg:f1}),
485: with the mixing temperature depending on the thickness of the hydrogen
486: layer. White dwarfs with thicker hydrogen layers will mix at lower
487: effective temperatures. A comparison of the results shown in Figures
488: \ref{fg:f1} and \ref{fg:f11} implies that there are very few DA stars
489: between $\Te\sim15,000$~K and 10,000~K --- the region where ZZ Ceti
490: stars are found --- with hydrogen layer masses smaller than $M_{\rm
491: H}/M_{\rm tot}\sim10^{-10}$. The gradual increase of the observed
492: ratio from 10,000~K down to 8000~K further implies that only $15\%$ of
493: DA white dwarfs in the 10$-$15,000~K temperature range have
494: superficial hydrogen layers with a mass in the range $M_{\rm H}/M_{\rm
495: tot}\sim10^{-10}$ to $10^{-8}$. Note that this range of layer mass
496: depends sensitively on the assumed convective efficiency
497: (ML2/$\alpha=0.6$ in Fig.~\ref{fg:f1}). Envelope models with a more
498: efficient parameterization of the convective energy transport would
499: imply even thicker hydrogen layers \citep[see Fig.~9
500: of][]{tassoul90}. The lack of evidence for convective mixing at lower
501: effective temperatures between $\Te=8000$~K and 5000~K implies a
502: discontinuity in the hydrogen layer mass distribution, with the large
503: majority ($\sim85$\%) of all DA stars between $\Te=15,000$~K and
504: 10,000~K having hydrogen layer masses above $M_{\rm H}/M_{\rm
505: tot}\sim10^{-6}$ or so. The constant ratio of $\sim 0.25$ in this
506: temperature range also implies that $\sim 20$\% of all DA stars in the
507: DB gap (or DB ``trough'') between $\Te=45,000$~K and 30,000~K have
508: turned into DB stars by the time they reach 15,000~K; we cannot tell
509: from our results at which effective temperature the DA to DB
510: transition occurs precisely.
511:
512: The numbers given above could change slightly if one takes into
513: account the presence of traces of metals (including carbon) and
514: hydrogen in the model atmosphere calculations of DQ and DZ
515: stars. Indeed, the atmospheric structure of cool helium-atmosphere
516: white dwarfs is affected significantly by small traces of metals or
517: hydrogen, which then provide most of the free electrons
518: \citep{provencal02, dufour05, dufour07}. These additional electrons
519: increase the contribution of the He$^-$ free-free opacity, which in
520: turn reduces the atmospheric pressure. The results of Dufour et
521: al.~(2005, 2007) reveal that photometric temperatures of DQ and DZ
522: stars can be systematically reduced by 500~K on average when metals
523: are included in the model calculations. Therefore, this could affect
524: the precise temperature dependence of the change in the ratio of
525: helium- to hydrogen-atmosphere white dwarfs observed in Figure
526: \ref{fg:f11}.
527:
528: We finally note that our results are compatible with the
529: asteroseismological analyses of ZZ Ceti stars, which suggest that most
530: ZZ Ceti stars have fairly thick hydrogen layers of the order of
531: $M_{\rm H}/M_{\rm tot}\sim 10^{-6}$ and higher
532: \citep[see, e.g.,][]{fontaine92,bergeron93,bradley98,brassard06,pech06}.
533: Only a small fraction (about 15\%) of all ZZ Ceti stars should have
534: thin layers according to our results. Furthermore, our results are
535: also in qualitative agreement with the analysis of DZ stars by
536: \citet{dufour07} who suggested that the presence of hydrogen in cool
537: DZA white dwarfs could be explained better as the outcome of
538: convectively mixed DA stars rather than due to accretion from the
539: interstellar medium. This suggestion is based on estimates of the
540: total mass of hydrogen present in the mixed hydrogen and helium
541: envelope of DZ and DZA stars. Measurements of the hydrogen abundances
542: in these stars suggest hydrogen layer masses in about the same range
543: as that inferred here
544: \citep[see Fig.~13 of][]{dufour07}.
545:
546: \section{CONCLUSION}
547:
548: We have presented a detailed study of the spectral evolution of cool
549: white dwarfs following the pioneering effort of \citet{sion84}. In
550: particular, we have determined the ratio of helium- to
551: hydrogen-atmosphere white dwarfs below $\Te=15,000$~K as a function of
552: effective temperature by increasing the size of the sample by a factor
553: of 2, and by obtaining more precise measurements of the effective
554: temperature of these stars through model atmosphere fits to Johnson
555: $V$ (or Str\"omgren $y$) and 2MASS $JHK_S$ photometric
556: observations. We have confirmed that in this temperature range, the
557: evolution of hydrogen- and helium-atmosphere white dwarfs is not
558: independent, and we have revealed in greater detail the role of the
559: convective mixing process responsible for the coupling between these
560: two white dwarf types. We have found that about 15$\%$ of cool
561: hydrogen-atmosphere DA white dwarfs between $\Te=15,000$~K and
562: 10,000~K are transformed into helium-atmosphere non-DA white dwarfs at
563: lower temperatures. Using a basic model for convective mixing, our
564: results imply in turn that 15\% of the DA stars that have survived the
565: DA to DB transition have hydrogen layer masses in the range $M_{\rm
566: H}/M_{\rm tot}\sim10^{-10}$ to $10^{-8}$, although the exact values
567: depend on the assumed convective efficiency in the evolutionary model
568: calculations. The remaining DA stars would presumably have more
569: massive hydrogen layers, probably in excess of $M_{\rm H}/M_{\rm
570: tot}\sim10^{-6}$.
571:
572: Further analysis of the convective mixing problem would require a
573: significantly larger photometric sample, such as the {\it ugriz}
574: photometric sample of the Sloan Digital Sky Survey (SDSS). This sample
575: covers a few areas of the sky with a completeness of around 95\% up to
576: $g=20$. However, most spectroscopic follow-ups have relied on
577: color-color diagrams to identify white dwarfs
578: \citep{kl04,eisenstein06b}, and this method is strongly biased towards
579: hot white dwarfs ($\Te > 12000$~K). Therefore, the SDSS white dwarfs
580: identified so far cannot be used to study the convective mixing
581: process at low effective temperatures. One way to proceed to identify
582: cooler white dwarfs in the SDSS is through proper motion-color
583: diagrams. \citet{harris06} have used this method to identify nearly
584: 6000 white dwarfs but a complete spectroscopic analysis similar to
585: that of \citet{kilic06} has yet to be performed to take full advantage
586: of this sample.
587:
588: \acknowledgements
589: We thank A.~Gianninas for a careful reading of our manuscript, and
590: G.~Fontaine and P.~Brassard for providing with us with the details of
591: their evolutionary models. We are also grateful to H.~C.~Harris for
592: allowing us to use some of the USNO $V$ magnitudes prior to
593: publication. This work was supported in part by the NSERC Canada and
594: by the Fund FQRNT (Qu\'ebec). P. Bergeron is a Cottrell Scholar of
595: Research Corporation.
596:
597: \clearpage
598:
599: \begin{thebibliography}{}
600:
601: \bibitem[Baglin \& Vauclair(1973)]{baglin73} Baglin, A., \& Vauclair, G. 1973, \aap, 27, 307
602:
603: \bibitem[Beauchamp(1995)]{beauchamp95} Beauchamp, A.\ 1995, Ph. D. Thesis, Universit\'e de Montr\'eal
604:
605: \bibitem[Beauchamp et al.(1996)]{beauchamp96} Beauchamp, A., Wesemael, F., Bergeron, P., Liebert, J., \& Saffer, R. A. 1996, in ASP Conf. Ser. Vol. 96, Hydrogen-Deficient Stars, ed. S. Jeffery \& U. Heber (San Francisco: ASP), 295
606:
607: \bibitem[Bergeron et al.(1993)]{bergeron93} Bergeron, P., et al. 1993, AJ, 106, 1987
608:
609: \bibitem[Bergeron et al.(2001)]{bergeron01} Bergeron, P., Leggett, S.K., \& Ruiz, M.T. 2001, \apj, 133, 413
610:
611: \bibitem[Bergeron et al.(1997)]{bergeron97} Bergeron, P., Ruiz, M.T., \& Leggett, S.K. 1997, \apj, 133, 413
612:
613: \bibitem[Bergeron et al.(1992)]{bergeron92} Bergeron, P., Saffer, R.A., \& Liebert, J. 1992, \apj, 394, 228
614:
615: \bibitem[Bessell(1990)]{bessell90} Bessell, M.S.\ 1990, \pasp, 102, 1181
616:
617: \bibitem[Bradley(1998)]{bradley98} Bradley, P.A., 1998, \apj, 552, 326
618:
619: \bibitem[Brassard \& Fontaine(2006)]{brassard06} Brassard, P., \& Fontaine, G.\ 2006, Mem.~S.~A.~It., 77, 439
620:
621: \bibitem[Cohen et al.(2003)]{cohen03} Cohen, M., Wheaton, W.A., \& Megeath, S.T.\ 2003, \aj, 126, 1090
622:
623: \bibitem[Dahn et al.(2007)]{dahn07} Dahn, C.C., et al. 2007, in preparation
624:
625: \bibitem[Dufour et al.(2005)]{dufour05} Dufour, P., Bergeron, P., \& Fontaine, G.\ 2005, \apj, 627, 404
626:
627: \bibitem[Dufour et al.(2007)]{dufour07} Dufour, P., Bergeron, P., Liebert, J., Harris, H.C., Knapp, G.R., Anderson, S.F., Hall, P.B., Strauss, M.A., Collinge, M.J., \& Edwards, M.C. 2007, \apj, 663, 1291
628:
629: \bibitem[Eisenstein et al.(2006a)]{eisenstein06a} Eisenstein, D.J., et al. 2006a, \aj, 132, 676
630:
631: \bibitem[Eisenstein et al.(2006b)]{eisenstein06b} Eisenstein, D.J., et al. 2006b, \apjs, 167, 40
632:
633: \bibitem[Fontaine et al.(2001)]{fon01} Fontaine, G., Brassard, P., \& Bergeron, P. 2001, \pasp, 113, 409
634:
635: \bibitem[Fontaine et al.(1992)]{fontaine92} Fontaine, G., Brassard, P., Bergeron, P., \& Wesemael, F.\ 1992, \apjl, 399, L91
636:
637: \bibitem[Fontaine \& Wesemael(1987)]{fontaine87} Fontaine, G., \& Wesemael, F.\ 1987, in IAU Colloquium 95, The Second Conference on Faint Blue Stars, eds. A. G. Davis Philip, D. S. Hayes, \& J. Liebert (Schenectady: L. Davis), 319
638:
639: \bibitem[Fontaine \& Wesemael(1991)]{fontaine91} Fontaine, G., \& Wesemael, F. 1991, in IAU Symp.~145, Evolution of Stars: The Stellar Abundance Connection, ed.~G.~Michaud \& A.~Tutukov (Dordrecht: Reidel), 421
640:
641: \bibitem[Greenstein(1986)]{greenstein86} Greenstein, J.L. 1986, \apj, 304, 334
642:
643: \bibitem[Harris et al.(2006)]{harris06} Harris, H.C., et al. 2006, \aj, 131, 571
644:
645: \bibitem[Holberg \& Bergeron(2006)]{holberg06} Holberg, J.B., \& Bergeron, P.\ 2006, \aj, 132, 1221
646:
647: \bibitem[Kepler et al.(2007)]{kepler07} Kepler, S.O., Kleinman, S.J., Nitta, A., Koester, D., Castanheira, B.G., Giovannini, O., Costa, A.F.M., \& Althaus, L. 2007, \mnras, 375, 1315
648:
649: \bibitem[Kilic et al.(2006)]{kilic06} Kilic, M., Munn, J.A., Harris, H.C., Liebert, J., von Hippel, T., Williams, K.A., Metcalfe, T.S., Winget, D.E., \& Levine, S.E. 2006, \aj, 131, 582
650:
651: \bibitem[Kleinman et al.(2004)]{kl04} Kleinman, S.J., et al. 2004, \apj, 607, 426
652:
653: \bibitem[Koester(1976)]{koester76} Koester, D.\ 1976, \aap, 52, 415
654:
655: \bibitem[Liebert et al.(2005)]{liebert05} Liebert, J., Bergeron, P., \& Holberg, J.B. 2005, \apjs, 156, 47
656:
657: \bibitem[MacDonald \& Vennes(1991)]{macdonald91} MacDonald, J., \& Vennes, S.\ 1991, \apj, 371, 719
658:
659: \bibitem[Oswalt et al.(1991)]{oswalt91} Oswalt, T.D., Sion, E.M., Hammond, G., Vauclair, G., Liebert, J.W., Wegner, G., Koester, D., \& Marcum, P.M.\ 1991, \aj, 101, 583
660:
661: \bibitem[Pech \& Vauclair(2006)]{pech06} Pech, D., \& Vauclair, G. 2006, \aap, 453, 219
662:
663: \bibitem[Provencal et al.(2002)]{provencal02} Provencal, J.L., Shipman, H.L., Koester, D., Wesemael, F., \& Bergeron, P.\ 2002, \apj, 568, 324
664:
665: \bibitem[Shipman(1972)]{shipman72} Shipman, H.L. 1972, \apj, 177, 723
666:
667: \bibitem[Sion(1984)]{sion84} Sion, E.M.\ 1984, \apj, 282, 612
668:
669: \bibitem[Strittmatter \& Wickramasinghe(1971)]{strittmatter71} Strittmatter, P.A., \& Wickramasinghe, D.T. 1971, \mnras, 152, 47
670:
671: \bibitem[Tassoul et al.(1990)]{tassoul90} Tassoul, M., Fontaine, G., \& Winget,
672: D. E. 1990, \apjs, 72, 335
673:
674: \bibitem[Tremblay \& Bergeron(2007)]{tremblay07} Tremblay, P.-E., \& Bergeron, P. 2007, \apj, 657, 1013
675:
676: \bibitem[Vauclair \& Reisse(1977)]{vauclair77} Vauclair, G., \& Reisse, C. 1977, \aap, 61, 415
677:
678: \bibitem[Voss et al.(2007)]{voss07} Voss, B., Koester, D., Napiwotzki, R., Christlieb, N., \& Reimers, D. 2007, \aap, 470, 1079
679:
680: \bibitem[Werner \& Herwig(2006)]{werner06} Werner, K., \& Herwig, F. 2006, \pasp, 118, 183
681:
682: \end{thebibliography}
683:
684: \clearpage
685: \include{tab1}
686:
687: \clearpage
688: \include{tab2}
689:
690: \clearpage
691:
692: \figcaption[f1.eps]
693: {Location of the hydrogen convection zone ({\it hatched region}) as a
694: function of effective temperature in the pure hydrogen envelope of a 0.6
695: \msun\ DA white dwarf calculated with the ML2/$\alpha=0.6$ version of the
696: mixing-length theory (from G.~Fontaine \& P.~Brassard 2006, private
697: communication). The depth is expressed as the fractional mass above
698: the point of interest with respect to the total mass of the star. The
699: thick solid line corresponds to the photosphere ($\tau_R\sim1$).
700: \label{fg:f1}}
701:
702: \figcaption[f2.eps]
703: {The ratio of non-DA to DA white dwarfs as a function of absolute
704: visual magnitude for the sample of \citet{sion84}. The error bars
705: correspond to statistical uncertainties. The temperature scale for
706: pure hydrogen model atmospheres with $M=0.6$ \msun\ is shown for
707: comparison at the top of the figure.
708: \label{fg:f2}}
709:
710: \figcaption[f3.eps]
711: {Predicted $V$ magnitude as a function of $\Te$ for pure hydrogen
712: ({\it solid line}) and pure helium ({\it dashed line}) model
713: atmospheres at $\log g=8$ and for which $H_{\rm 2MASS}=16.5$. The
714: turnover of the hydrogen-atmosphere curve below 5000 K is due to the
715: onset of the collision-induced absorptions by molecular hydrogen.
716: \label{fg:f3}}
717:
718: \figcaption[f4.eps]
719: {Difference in $(V-J)$ color indices as a function of $\Te$ between
720: $\log g=8.5$ and 7.5 model atmospheres with pure hydrogen ({\it solid
721: line}) and pure helium ({\it dashed line}) compositions. The general
722: behavior of these curves can be explained in terms of the various
723: opacity sources that peak at slightly different effective temperatures
724: for these two $\log g$ values.\label{fg:f4}}
725:
726: \figcaption[f5.eps]
727: {Sample fits to the observed $VJHK_S$ photometric observations ({\it
728: error bars}) of eight bright white dwarfs with hydrogen and helium
729: atmospheres. The model fluxes are shown by open circles and the
730: atmospheric parameters are given in each panel.\label{fg:f5}}
731:
732: \figcaption[f6.eps]
733: {Effective temperature variation, $\Delta \Te$, for a corresponding
734: change of the $(V-J)$ color index of $\Delta (V-J)=0.08$ as a function
735: of $\Te$ for our pure hydrogen ({\it solid line}) and pure helium
736: ({\it dashed line}) model atmospheres. The dotted horizontal lines
737: represent the various bin widths of the histograms used in
738: \S~5.\label{fg:f6}}
739:
740: \figcaption[f7.eps]
741: {Differences in effective temperature determinations based on our
742: 2MASS sample ($T_{\rm phot}$) and the photometric $BVRIJHK$ data
743: ($T_{\rm BRL}$) of \citet{bergeron97,bergeron01} for the 133 cool
744: white dwarfs in common. The solid line represents a perfect match
745: between both data sets, while the dashed lines correspond to
746: temperature uncertainties equal to the bin widths of the histograms
747: used in \S~5.\label{fg:f7}}
748:
749: \figcaption[f8.eps]
750: {\textit{Top panel:} Differences between our photometric $\Te$ values
751: and those determined spectroscopically for a common sample of 198 DA
752: white dwarfs. The small open circles represent stars for which the
753: 2MASS fluxes are close to the detection limit with formal
754: uncertainties in only two bands. The filled squares correspond to
755: known double degenerate systems. The horizontal and dashed lines have
756: the same meaning as in Figure \ref{fg:f7}. \textit{Bottom panel:} Same
757: as top panel but for 18 DB white dwarfs.
758: \label{fg:f8}}
759:
760: \figcaption[f9.eps]
761: {Histogram of the number of hydrogen-atmosphere ({\it single hatched})
762: and helium-atmosphere ({\it double hatched}) white dwarfs from the
763: 2MASS sample as a function of $\Te$. The number of stars in each bin
764: is per unit of $10^3$~K in $\Te$.
765: \label{fg:f9}}
766:
767: \figcaption[f10.eps]
768: {The ratio of the observed space volume for helium-atmosphere and
769: hydrogen-atmosphere white dwarfs in a $H$ magnitude limited survey as a
770: function of $\Te$.
771: \label{fg:f10}}
772:
773: \figcaption[f11.eps]
774: {The ratio of the number of helium- to hydrogen-atmosphere white dwarfs
775: as a function of effective temperature, corrected for the volume
776: effect displayed in Figure \ref{fg:f10} and discussed in the
777: text. The filled circles use the binning defined in Figure
778: \ref{fg:f9} while the open circles present the results
779: for the temperature bins shifted by 500~K. The error bars are
780: statistical.
781: \label{fg:f11}}
782:
783: \begin{figure}[p]
784: \plotone{f1.eps}
785: \begin{flushright}
786: Figure \ref{fg:f1}
787: \end{flushright}
788: \end{figure}
789:
790: \begin{figure}[p]
791: \plotone{f2.eps}
792: \begin{flushright}
793: Figure \ref{fg:f2}
794: \end{flushright}
795: \end{figure}
796:
797: \begin{figure}[p]
798: \plotone{f3.eps}
799: \begin{flushright}
800: Figure \ref{fg:f3}
801: \end{flushright}
802: \end{figure}
803:
804: \begin{figure}[p]
805: \plotone{f4.eps}
806: \begin{flushright}
807: Figure \ref{fg:f4}
808: \end{flushright}
809: \end{figure}
810:
811: \begin{figure}[p]
812: \plotone{f5.eps}
813: \begin{flushright}
814: Figure \ref{fg:f5}
815: \end{flushright}
816: \end{figure}
817:
818: \begin{figure}[p]
819: \plotone{f6.eps}
820: \begin{flushright}
821: Figure \ref{fg:f6}
822: \end{flushright}
823: \end{figure}
824:
825: \begin{figure}[p]
826: \plotone{f7.eps}
827: \begin{flushright}
828: Figure \ref{fg:f7}
829: \end{flushright}
830: \end{figure}
831:
832: \begin{figure}[p]
833: \plotone{f8.eps}
834: \begin{flushright}
835: Figure \ref{fg:f8}
836: \end{flushright}
837: \end{figure}
838:
839: \begin{figure}[p]
840: \plotone{f9.eps}
841: \begin{flushright}
842: Figure \ref{fg:f9}
843: \end{flushright}
844: \end{figure}
845:
846: \begin{figure}[p]
847: \plotone{f10.eps}
848: \begin{flushright}
849: Figure \ref{fg:f10}
850: \end{flushright}
851: \end{figure}
852:
853: \begin{figure}[p]
854: \plotone{f11.eps}
855: \begin{flushright}
856: Figure \ref{fg:f11}
857: \end{flushright}
858: \end{figure}
859:
860: \end{document}
861: