1: \documentclass[12pt,preprint]{aastex}
2:
3: \def\kms{km~s$^{-1}$}
4: \def\mv{M_V}
5: \def\msun{$M_{\odot}$}
6: \def\rsun{$R_{\odot}$}
7: \def\lsun{$L_{\odot}$}
8: \def\halpha{H$\alpha$}
9: \def\hbeta{H$\beta$}
10: \def\hgamma{H$\gamma$}
11: \def\hdelta{H$\delta$}
12: \def\Te{T_{\rm eff}}
13: \def\logg{\log g}
14: \def\gta{\lower 0.5ex\hbox{$\buildrel > \over \sim\ $}} %greater than about
15: \def\lta{\lower 0.5ex\hbox{$\buildrel < \over \sim\ $}} %less than about
16:
17: \tighten
18:
19: \begin{document}
20:
21: \title{DISCOVERY OF A COOL, MASSIVE, AND METAL-RICH DAZ WHITE DWARF}
22:
23: \author{A. Gianninas, P. Dufour and P. Bergeron }
24: \affil{D\'epartement de Physique, Universit\'e de Montr\'eal, C.P.~6128,
25: Succ.~Centre-Ville, Montr\'eal, Qu\'ebec, Canada, H3C 3J7.}
26: \email{gianninas, dufourpa, bergeron@astro.umontreal.ca}
27:
28: \begin{abstract}
29:
30: We report the discovery of a new metal-rich DAZ white dwarf, GD
31: 362. High signal-to-noise optical spectroscopy reveals the presence of
32: spectral lines from hydrogen as well as Ca~\textsc{i}, Ca~\textsc{ii},
33: Mg~\textsc{i}, and Fe~\textsc{i}. A detailed model atmosphere
34: analysis of this star yields an effective temperature of
35: $\Te=9740\pm50$~K, a surface gravity of $\logg=9.12 \pm 0.07$, and
36: photospheric abundances of $\log{\rm Ca/H}=-5.2\pm0.1$, $\log{\rm
37: Mg/H}=-4.8\pm0.1$, and $\log{\rm Fe/H}=-4.5\pm0.1$. White dwarf
38: cooling models are used to derive a mass of 1.24
39: \msun\ for GD 362, making it the most massive and metal-rich DAZ
40: uncovered to date. The problems related to the presence of such large
41: metal abundances in a nearby ($d\sim 25$~pc) white dwarf in terms of
42: an accretion scenario are briefly discussed.
43:
44: \end{abstract}
45:
46: \keywords{stars : individual (GD 362) -- stars : abundances -- white dwarfs}
47:
48: \section{INTRODUCTION}
49:
50: The most common type of white dwarfs are those that show hydrogen
51: absorption lines in their optical spectra, the DA stars. A fraction of
52: these, the so-called DAZ stars, also display absorption features from
53: heavier elements, most notably the Ca~\textsc{ii} H and K lines. A
54: recent spectroscopic survey at high dispersion of about 120 cool
55: ($\Te<10,000$~K) DA stars by \citet{zuckerman03} reveals that $\sim
56: 25$\% of the objects in their sample show metallic features. The DAZ
57: phenomenon thus seems much more common than previously believed
58: \citep{billeres97}. The origin of these metals in the atmospheres of
59: DAZ stars is still poorly understood, however. Indeed, as most DAZ
60: stars are comparatively cool and old, it is expected that all elements
61: heavier than hydrogen would have settled rather quickly at the bottom
62: of the atmosphere due to the strong gravitational field present in
63: these stars. Therefore, any metals found in the atmosphere must be
64: provided by an exterior source, such as the interstellar medium or
65: even comets. We refer the reader to \citet{zuckerman03} for a more
66: detailed discussion of the various theories that have been put forth
67: to account for the presence of metals in DAZ stars.
68:
69: We have recently undertaken a spectroscopic survey of white dwarfs
70: drawn from the catalog of \citet{mccook99}. Our aim is to derive the
71: atmospheric parameters for all stars, as well as to confirm the
72: spectroscopic classifications given in the catalog using modern CCD
73: spectroscopy. As part of this survey, we have discovered that GD 362
74: (WD 1729+371, PG) is a unique DAZ white dwarf, with metallic features
75: stronger than in any other member of its class, the Ca~\textsc{ii} H
76: and K lines in particular. GD 362 had initially been classified as a
77: DA white dwarf by \citet{greenstein80} with a note that it could
78: possibly constitute a composite spectrum or a dK subdwarf in a common
79: proper motion binary system. It was later rediscovered in the
80: Palomar-Green survey \citep{green86} where it was classified
81: \textit{sd}, which does not imply a subdwarf type, but rather ``a
82: lower signal-to-noise observation in which two or three Balmer
83: absorption lines of moderate gravity are visible''
84:
85: We present in this {\it Letter} our spectroscopic data for GD 362 in
86: which, besides the usual hydrogen Balmer lines, we have
87: identified spectral lines of Ca~\textsc{i}, Ca~\textsc{ii},
88: Mg~\textsc{i} and Fe~\textsc{i}. By fitting the optical spectrum with
89: a grid of synthetic spectra appropriate for these stars, we determine the
90: effective temperature and surface gravity of GD 362, as well as the
91: abundances of all the elements observed in this star.
92:
93: \section{FITTING PROCEDURE AND SYNTHETIC SPECTRA}
94:
95: \subsection{Observations}
96:
97: The high signal-to-noise (S/N$\sim 110$ per pixel) optical spectrum of
98: GD 362 was obtained over the course of 2 nights using the Steward
99: Observatory 2.3 m telescope equipped with the Boller \& Chivens
100: spectrograph. The 4.5 arcsec slit together with the 600 l/mm grating
101: blazed at $\lambda3568$ in first order provided a spectral coverage
102: from about 3000 to 5250~\AA\ at a resolution of $\sim$ 6~\AA\
103: FWHM. The spectra were reduced using standard IRAF packages. The
104: spectrum is displayed in Figure
105: \ref{fg:f1} where the major spectral features are identified,
106: namely Ca~\textsc{i}, Ca~\textsc{ii}, Mg~\textsc{i}, and Fe~\textsc{i} lines.
107: We also show for comparison the optical spectrum of G74-7,
108: discovered by \citet{lacombe83}, which was the only DAZ star known
109: until the more recent discoveries of G238-44
110: \citep{holberg97}, G29-38 \citep{koester97}, and those reported
111: by \citet{zuckerman03}. The comparison is most striking when we
112: examine in both spectra the relative strengths of the H and K lines of
113: Ca~\textsc{ii} at 3933.66~\AA\ and 3968.67~\AA, respectively. It is
114: worth noting that the region blueward of $\sim3800$~\AA\ is heavily
115: blanketed by Fe \textsc{i} lines.
116:
117: \subsection{Fitting Procedure}
118:
119: The technique for fitting the spectrum is similar to that
120: described at length by \citet{liebert04}. First the continuum
121: used to normalize the spectrum is set by fitting a model spectrum
122: to the data, allowing for a possible wavelength shift, a zero point offset,
123: and higher order terms in $\lambda$ (up to $\lambda^{3}$). The purpose
124: of this procedure is simply to obtain a function that defines best
125: the continuum of the observed spectrum. Once the spectrum has been normalized properly,
126: the atmospheric parameters and chemical abundances are obtained from a
127: grid of model spectra, convolved with a Gaussian profile at 6~\AA\ FWHM,
128: using the nonlinear least-squares method of Levenberg-Marquardt
129: \citep{press86}. This minimization technique also provides formal
130: uncertainties of the fitted parameters through the covariance matrix
131: \citep[see][for details]{BSL}.
132:
133: \subsection{Synthetic Spectra}
134:
135: For the computation of our synthetic spectra, we adopt a grid of LTE
136: model atmospheres with a pure hydrogen composition similar to that
137: described in \citet{liebert04}. We thus make the assumption that the
138: metals have no effect on the atmospheric structure and include them
139: only in the calculation of the emergent fluxes; this assumption will
140: be tested subsequently. Atomic partition functions have been
141: calculated by summing explicitly over the bound states taken directly
142: from TOPBASE. Central wavelengths of the transitions, $gf$ values,
143: energy levels, and damping constants (radiative, Stark, and van der
144: Waals) have been extracted from the GFALL linelist of Kurucz
145: (http://kurucz.harvard.edu/LINELISTS.html).
146:
147: Since our preliminary analysis of GD 362 \citep{gianninas05} revealed
148: that it was a high surface gravity DAZ star, we restrict our model
149: grid calculations to high values of $\logg$. We first begin with the
150: determination of the calcium abundance since calcium produces the most
151: prominent features in the spectrum. Our model grid covers a range of
152: $-5.7 \le \log{\rm Ca/H}\le -4.8$ in steps of 0.3 dex, effective
153: temperatures in the range $9000\le\Te\le10,400$~K in steps of 200 K,
154: and surface gravities in the range $8.75\le\logg\le9.50$ in steps of
155: 0.25 dex. With the calcium abundance fixed, we proceed to determine in
156: a similar way the magnesium and iron abundances in turn, using spectra
157: with magnesium abundances in the range $-5.6 \le \log$ Mg/H $\le -4.4$
158: in steps of 0.4 dex, and iron abundances in the range $-5.0 \le \log$
159: Fe/H $\le -4.4$ in steps of 0.2 dex. At each step, the determination
160: of the effective temperature and surface gravity is refined.
161:
162: \subsection{Atmospheric Parameter and Abundance Determinations}
163:
164: Before undertaking our analysis of GD 362, we wished to test our grid
165: of synthetic spectra, and more importantly our fitting procedure. With
166: this in mind, we fitted the optical spectrum of G74-7 analyzed
167: by \citet{billeres97} who derived $\Te=7260 \pm 40$~K, $\log g=8.03 \pm 0.07$, and
168: $\log {\rm Ca/H}=-8.8 \pm 0.1$. Our own analysis of the same spectrum
169: yields $\Te=7300 \pm 60$~K, $\log g=8.02 \pm 0.11$, and
170: $\log {\rm Ca/H}=-8.9 \pm 0.3$, in excellent agreement with the results of
171: Bill\`eres et al.
172:
173: Our best fit to the spectrum of GD 362 is displayed in Figure
174: \ref{fg:f2}. The atmospheric parameters and corresponding uncertainties are
175: $\Te=9740\pm50$~K, $\logg=9.12\pm 0.07$, $\log {\rm Ca/H}=-5.2\pm0.1$,
176: $\log{\rm Mg/H}=-4.8\pm0.1$, and $\log{\rm Fe/H}=-4.5\pm0.1$ . We are
177: able to obtain a very good fit to the observed spectrum albeit not a
178: perfect one. First, we notice that although the H and K lines of
179: Ca~\textsc{ii} are extremely well reproduced, the Ca~\textsc{i} lines at
180: 4227 and 4585~\AA\ are predicted
181: too strong. We believe that this is caused by an incorrect
182: Ca~\textsc{ii}/Ca~\textsc{i} ratio which could be due either to an
183: erroneous electron density or partition function values obtained from
184: TOPBASE. Second, the Fe~\textsc{i} lines in the bluest portion of the spectrum
185: are not reproduced perfectly. Some of the atomic data for iron is
186: derived from theoretical calculations and is more uncertain than
187: values determined experimentally. We therefore suggest that our
188: derived iron abundance is at worse an upper limit.
189:
190: A posteriori, we computed a blanketed model atmosphere in which we
191: took into account the presence of metals at our derived abundances.
192: The synthetic spectrum calculated from this model
193: is identical to that computed here from our pure
194: hydrogen models, validating our initial assumption that the metals play a
195: negligeable role in the determination of the atmospheric structure.
196:
197: \section{DISCUSSION}
198:
199: Using the evolutionary models of \citet{fontaine01} for DA white
200: dwarfs, we derive a mass of 1.24 \msun. Our derived atmospheric
201: parameters thus make GD 362 the most metal-rich and massive DAZ white
202: dwarf studied thus far, and somewhat hotter than the majority of DAZ
203: stars discovered by \citet{zuckerman03}. In Figure \ref{fg:f3}, we
204: show the calcium abundance as a function of effective temperature for
205: the white dwarfs from \citet{zuckerman03}. We see from its location in
206: Figure \ref{fg:f3} how unique GD 362 truly is. In fact, if we compare
207: with the values listed in Table 4 of
208: \citet{zuckerman03}, the iron and magnesium abundances are nearly
209: solar!
210:
211: A comparison of our calcium abundance determination for GD 362 with
212: those of other DAZ stars from \citet{zuckerman03} with similar
213: effective temperatures reveals that the abundance observed in GD 362
214: is at least a factor of $10^{3}$ higher (see Fig.~\ref{fg:f3}). Since
215: the mass of GD 362 is also higher than average, we need to verify
216: whether this is the reason for the unusual observed photospheric
217: abundances. To do so, we compare the expected steady-state calcium
218: abundance in a massive star at 1.2 \msun\ to that of a normal star
219: with a mass of 0.6 \msun. The steady-state abundance, $X_{\rm ss}$,
220: is given by equation (4) of \citet{dupuis93},
221:
222: \begin{equation}
223: X_{\rm ss}=\frac{\theta \dot{M}}{\Delta M_{\rm cz}}\ \ ,
224: \end{equation}
225:
226: \noindent where $\theta$ is the diffusion time scale of calcium at the
227: bottom of the hydrogen convection zone, $\dot{M}$ is the accretion
228: rate of that element, and $\Delta M_{\rm cz}$ is the mass of the
229: convection zone. If we assume the Bondi-Hoyle accretion rate, then
230: $\dot{M}$ is proportional to $M^2$, where $M$ is the mass of the white
231: dwarf. Then the expected relative abundance of calcium between a 1.2
232: and a 0.6 \msun\ white dwarf can be expressed as
233:
234: \begin{equation}
235: \frac{X_{1.2}}{X_{0.6}} \propto \frac{\theta_{1.2}}{\theta_{0.6}} \ \ \frac{M_{1.2}^{2}}{M_{0.6}^{2}} \ \ \frac{\Delta M_{{\rm cz}, 0.6}}{\Delta M_{{\rm cz}, 1.2}}\ \ ,
236: \end{equation}
237:
238: \noindent where we assume that both stars are moving with the same
239: velocity through an interstellar medium of similar density.
240: Using values of $\theta$ and $\Delta M_{\rm cz}$ generously
241: provided to us by G.~Fontaine (private communication), we obtain that
242:
243: \begin{displaymath}
244: \frac{X_{1.2}}{X_{0.6}} \propto \frac{1}{36} \cdot 4 \cdot 8.5 \approx 1\ \ .
245: \end{displaymath}
246:
247: \noindent We see that even though a more massive star accretes at a higher rate and
248: has a thinner convection zone --- implying a smaller dilution factor,
249: the diffusion time scale at the bottom of the hydrogen convection zone
250: is so much shorter that comparable abundances are predicted.
251:
252: Additionally, the accretion model suffers from the fact there is
253: little interstellar matter within 100 pc from the Sun, as
254: \citet{aanestad93} aptly pointed out. The distance to GD 362
255: can be estimated in two ways. First, the $\Te$ and $\logg$ values
256: inferred for GD 362 together with the photometric calibration of
257: \citet{bwb95} yield an absolute visual magnitude of $\mv=14.31$ (or
258: $M_B=14.48$), which combined with the photographic magnitude of
259: $B_{\rm ph}=16.15$ taken from \citet{green86}, yields a distance
260: estimate of $\sim 22$~pc. Alternatively, we can use the relation
261: between the monochromatic fluxes from our optical spectrum,
262: $f_{\lambda}$, and the theoretical Eddington fluxes, $H_{\lambda}$,
263: given by
264:
265: \begin{equation}
266: f_{\lambda}=4\pi (R/D)^{2}H_{\lambda}\ \ ,
267: \end{equation}
268:
269: \noindent where $R$ and $D$ are the radius of the star and its distance
270: from Earth, respectively, to obtain an independent distance estimate
271: of $\sim 26$~pc, consistent with our previous estimate. In either case, GD
272: 362 seems to be well within the local bubble and far from any high
273: concentrations of interstellar matter. We are therefore at odds to
274: explain the very high metal abundances observed.
275:
276: Another possibility would be that GD 362 is an unresolved degenerate
277: binary composed of a normal DA star and a helium-rich DZ star of
278: comparable luminosities. If this were the case, however, the intrinsic
279: absorption features of each star would be even stronger than observed
280: here, because of the line dilution factor, and the effective
281: temperature inferred from the Balmer lines would thus be grossly {\it
282: underestimated} (Balmer lines get stronger with increasing $\Te$ in
283: this temperature range). In such unresolved systems, the slope of the
284: theoretical spectrum is incompatible with that of the observed spectrum
285: \citep[see, e.g.,][]{bergeron90}. We see on the contrary from Figure
286: \ref{fg:f2} that the slope of the theoretical
287: spectrum is in perfect agreement with the observed slope of GD 362,
288: consistent with the presence of a single star.
289:
290: It is perhaps not too suprising that GD 362 had not been classified as
291: a DAZ white dwarf by \citet{greenstein80} or \citet{green86}. The
292: very strong Ca~\textsc{ii} H and K lines seen on their photographic
293: spectrum, in addition to the narrow Balmer lines, have led these
294: investigators to classify GD 362 as composite or as a lower gravity
295: object. This raises the interesting possibility that many such objects
296: exist and have been overlooked both in the past and in modern surveys
297: such as the Sloan Digital Sky Survey.
298:
299: We thank Gilles Fontaine and Pierre Brassard for useful discussions.
300: We would also like to thank the director and staff of Steward
301: Observatory for the use of their facilities, as well as the referee
302: for useful suggestions. This work was supported in part by the NSERC
303: Canada and by the FQRNT (Qu\'ebec).
304:
305: \clearpage
306:
307: \begin{thebibliography}{}
308:
309: \bibitem[Aanestad et al.(1993)]{aanestad93} Aanestad, P. A., Kenyon, S. J.,
310: Hammond, G. L., \& Sion, E. M. 1993, \aj, 105, 1033
311:
312: \bibitem[Bergeron et al.(1990)]{bergeron90} Bergeron, P., Greenstein, J. L.,
313: \& Liebert, J. 1990, \apj, 361, 190
314:
315: \bibitem[Bergeron et al.(1992)]{BSL} Bergeron, P., Saffer, R. A., \&
316: Liebert, J. 1992, \apj, 394, 228
317:
318: \bibitem[Bergeron et al.(1995)]{bwb95} Bergeron, P., Wesemael, F., \&
319: Beauchamp, A. 1995, \pasp, 107, 1047
320:
321: \bibitem[Bill\`eres et al.(1997)]{billeres97} Bill\`eres, M., Wesemael., F.,
322: Bergeron, P., \& Beauchamp, A. 1997, \apj, 488, 368
323:
324: \bibitem[Dupuis et al.(1993)]{dupuis93} Dupuis, J., Fontaine, G., Pelletier, C.,
325: \& Wesemael F. 1993, \apjs, 84, 73
326:
327: \bibitem[Fontaine et al.(2001)]{fontaine01} Fontaine, G., Brassard, P.,
328: \& Bergeron, P. 2001, \pasp, 113, 409
329:
330: \bibitem[Gianninas et al.(2005)]{gianninas05} Gianninas, A., Bergeron, P., \&
331: Dufour, P. 2005, in 14th European Workshop on White Dwarfs, ASP Conf.~Series,
332: vol.~999, eds. D.~Koester \& S.~Moehler, in press
333:
334: \bibitem[Green et al.(1986)]{green86} Green, R. F., Schmidt, M., \& Liebert,
335: J. 1986, \apjs, 61, 305
336:
337: \bibitem[Greenstein(1980)]{greenstein80} Greenstein, J. L. 1980, \apj, 242, 738
338:
339: \bibitem[Holberg et al.(1997)]{holberg97} Holberg, J. B., Barstow, M. A.,
340: \& Green, E. M. 1997, \apj, 474, L127
341:
342: \bibitem[Koester et al.(1997)]{koester97} Koester, D., Provencal, J.,
343: \& Shipman, H. L. 1997, \aap, 230, L57
344:
345: \bibitem[Lacombe et al.(1983)]{lacombe83} Lacombe, P., Wesemael, F.,
346: Fontaine, G., \& Liebert, J. 1983, \apj, 272, 660
347:
348: \bibitem[Liebert et al.(2004)]{liebert04} Liebert, J., Bergeron, P.,
349: \& Holberg, J. B. 2004, \apjs, in press
350:
351: \bibitem[McCook \& Sion(1999)]{mccook99} McCook, G. P., \& Sion, E. M. 1999,
352: \apjs, 121, 1
353:
354: \bibitem[Press et al.(1986)]{press86} Press, W. H., Flannery, B. P., Teukolsky,
355: S. A., \& Vetterling, W. T. 1986, Numerical Recipes (Cambridge: Cambridge
356: University Press)
357:
358: \bibitem[Zuckerman et al.(2003)]{zuckerman03} Zuckerman, B., Koester, D.,
359: Reid, I. N., \& H\"unsch, M. 2003, \apj, 596, 477
360:
361: \end{thebibliography}
362:
363: \clearpage
364:
365: \figcaption[f1.eps] {Flux calibrated optical spectra of G74-7 ({\it top})
366: and GD 362 ({\it bottom}), normalized to unity
367: at 4500~\AA, and offset from each other by a factor of 0.7. The major
368: metallic features in the spectrum of GD 362 are indicated by tick
369: marks. In both cases, H$\epsilon$ is blended with Ca~\textsc{ii}
370: $\lambda$3969.\label{fg:f1}}
371:
372: \figcaption[f2.eps] {Our best fit ({\it bottom}) to the observed spectrum
373: of GD 362 ({\it top}). The atmospheric parameters obtained from this
374: analysis are indicated along with the mass derived from the
375: evolutionary models of \citet{fontaine01}. \label{fg:f2}}
376:
377: \figcaption[f3.eps] {Calcium abundances, or upper limits, as a function of
378: effective temperature for the white dwarfs taken from Tables 1 and 2
379: of \citet{zuckerman03}. The open circles mark objects where the Ca
380: \textsc{ii} K line was detected, while the crosses denote white dwarfs with
381: only upper limit determinations. The filled circle represents GD
382: 362. \label{fg:f3}}
383:
384: \clearpage
385:
386: \begin{figure}[p]
387: \plotone{f1.eps}
388: \begin{flushright}
389: Figure \ref{fg:f1}
390: \end{flushright}
391: \end{figure}
392:
393: \begin{figure}[p]
394: \plotone{f2.eps}
395: \begin{flushright}
396: Figure \ref{fg:f2}
397: \end{flushright}
398: \end{figure}
399:
400: \begin{figure}[p]
401: \plotone{f3.eps}
402: \begin{flushright}
403: Figure \ref{fg:f3}
404: \end{flushright}
405: \end{figure}
406:
407: \end{document}
408: