1: %% LyX 1.3 created this file. For more info, see http://www.lyx.org/.
2: %% Do not edit unless you really know what you are doing.
3: \documentclass[letter,english]{aa}
4: \usepackage{times}
5: \usepackage[T1]{fontenc}
6: \usepackage[latin1]{inputenc}
7: \setcounter{tocdepth}{3}
8: \usepackage{graphicx}
9: \usepackage{amssymb}
10: \usepackage{amsmath}%, amsthm, amssymb}
11: \usepackage[authoryear]{natbib}
12: \makeatletter
13:
14: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% LyX specific LaTeX commands.
15: %% Bold symbol macro for standard LaTeX users
16: \providecommand{\boldsymbol}[1]{\mbox{\boldmath $#1$}}
17:
18:
19: \usepackage{babel}
20: \makeatother
21: \begin{document}
22:
23: \title{Might Carbon-Atmosphere White Dwarfs Harbour\\
24: a New Type of Pulsating Star?
25: }
26:
27: \author{G. Fontaine\inst{1}
28: \and P. Brassard\inst{1}
29: \and P. Dufour\inst{2}
30: }
31:
32: \institute{
33: D\'epartement de Physique, Universit\'e de Montr\'eal, C.P. 6128,
34: Succursale Centre-Ville, Montr\'eal, QC H3C 3J7, Canada
35: \\ \email{fontaine,brassard@astro.umontreal.ca}\\
36: \and Department of Astronomy and Steward Observatory, University of
37: Arizona, 933 North Cherry Avenue, Tucson, AZ 85721, USA
38: \\ \email{dufourpa@as.arizona.edu}
39: }
40:
41: \offprints{G. Fontaine}
42:
43:
44: \date{Received 26 February 2008 / Accepted 14 March 2008}
45:
46: \newpage
47:
48: \abstract{}
49: {In the light of the recent and unexpected discovery of a brand new type
50: of white dwarfs, those with carbon-dominated atmospheres, we examine the
51: asteroseismological potential of such stars. The motivation behind this
52: is based on the observation that past models of carbon-atmosphere white
53: dwarfs have partially ionized outer layers that bear strong resemblance
54: with those responsible for mode excitation in models of pulsating DB
55: (helium-atmosphere) and pulsating DA (hydrogen-atmosphere) white dwarfs.}
56: {We carry out a stability survey of models of carbon-atmosphere white
57: dwarfs following a full nonadiabatic approach. To connect with
58: previously known pulsating white dwarfs, we specifically search for
59: excited $g$-modes with $\ell = 1$ in the period window 80-1200 s. We
60: concentrate on models with $T_{\rm eff}$ $\leq$ 30,000 K, a limit below
61: which the real carbon-atmosphere stars are found, and investigate the
62: effects of changing the surface gravity, the composition of the
63: envelope, and the treatment of the convective efficiency.}
64: {Our exciting main result is that carbon-atmosphere white dwarfs may
65: indeed drive low-order $g$-modes in certain regions of parameter
66: space. For instance, log $g$ = 8.0 models characterized by an envelope
67: composition $X$(C) = $X$(He) = 0.5 and ML2 convection, show a broad
68: instability strip in the range 18,400-12,600 K. This is shifted to
69: 20,800-17,200 K for log $g$ = 9.0 models. In this interval of surface
70: gravity, the excited periods are found between $\sim$ 100 s and $\sim$
71: 700 s, and the shorter periods are excited in the higher-gravity
72: models. Adding carbon in the envelope mixture tends to extinguish
73: pulsational driving.}
74: {Given the right location in parameter space, some carbon-atmosphere white
75: dwarfs are predicted to show pulsational instability against gravity
76: modes. We are eagerly awaiting the results of observational searches for
77: luminosity variations in these stars.}
78:
79: \keywords{stars: oscillations -- stars: white dwarfs}
80: \authorrunning{Fontaine, Brassard, \& Dufour}
81: \titlerunning{Carbon-Atmosphere White Dwarfs as a New Type of Pulsator}
82: \maketitle
83:
84: \section{Introduction}
85:
86: In a recent publication, Dufour et al. (2007) reported on the
87: unexpected discovery of a new type of white dwarf. These are stars with
88: atmospheres dominated by carbon. Until then, white dwarfs cooler than
89: $\sim$ 80,000 K were known to come in only two flavors, those with
90: almost pure hydrogen atmospheres (the so-called DA stars comprising some
91: 75 to 80\% of the known sample), and those with almost pure helium
92: atmospheres (the non-DA's accounting for the rest). Obviously quite rare
93: since none had ever been found before among the several thousands white
94: dwarfs known, the carbon-atmosphere stars uncovered by Dufour et
95: al. (2007) form a small sample of nine faint objects with $g$ $\sim$
96: 18$-$19 culled from the Sloan Digital Sky Survey (SDSS). They are all
97: found in a rather narrow range of effective temperature, between 18,000
98: K and 23,000 K. A typical value of the abundance ratio derived from the
99: less noisy SDSS spectra in the sample is $X$(C)/$X$(He) $\geq$ 99. However,
100: a ratio as small as $X$(C)/$X$(He) $\geq$ 3 cannot be excluded in some
101: of the other spectra according to the more detailed analysis reported by
102: Dufour et al. (2008).
103:
104: Beyond their very existence, which remains to be explained (see, in this
105: context, the qualitative evolutionary scenario proposed by Dufour et
106: al. 2007), carbon-atmosphere white dwarfs have attracted our interest
107: from another point of view. And indeed, it has not escaped our attention
108: that past models of carbon-atmosphere white dwarfs in the range of
109: effective temperature where the real ones are found are characterized by an
110: important outer superficial convection zone, very similar to that found
111: in the pulsating DB (V777 Her) stars (centered around $T_{\rm eff}$
112: $\simeq$ 25,000 K) or in the pulsating DA (ZZ Ceti) stars (found around
113: $T_{\rm eff}$ $\simeq$ 12,000 K). Hence, it follows that
114: carbon-atmosphere white dwarfs could also excite pulsation modes through
115: the same partial ionization/convective driving phenomenon that is at
116: work in these two distinct families of pulsating white dwarfs.
117:
118: In this connection, we draw attention of the reader to Figure 7 of
119: Fontaine \& Van Horn (1976) where some properties of the convection
120: zones found in models of hydrogen-, helium-, and carbon-atmosphere white
121: dwarfs are compared. There are obvious similarities, and this strongly
122: suggests that a stability study for carbon-atmosphere stars would be
123: worth doing. The lead author would never have imagined that the
124: carbon envelope white dwarf models that he examined as part of his
125: Ph.D. thesis some 35 years ago could ever have any relevance with the
126: real world! At the time, this was done only out of theoretical curiosity
127: (see Fontaine 1974) and only the hydrogen- and helium-atmosphere models
128: were then thought to be ``relevant''. And indeed, to our knowledge,
129: the carbon envelope models of Fontaine \& Van Horn (1976) have remained
130: the only ones of the kind to have been published.
131:
132: With this background in mind, we decided to investigate the
133: asteroseismological potential of models of carbon-atmosphere white
134: dwarfs. Further impetus came from the involvement of one of us (P.D.)
135: with other collaborators to search observationally for luminosity
136: variations in these stars. We thought that it would be quite timely to
137: provide the theoretical background in parallel with that search.
138:
139: \section{Equilibrium Models and Stability Survey}
140:
141: In the sort of exploratory work presented in this paper, it is amply
142: sufficient to study relatively simple models of carbon-atmosphere
143: white dwarfs. We thus considered equilibrium structures made of a
144: uniform-composition envelope sitting on top of a C/O core (also with a
145: uniform composition specified by $X$(C) = $X$(O) = 0.5). As a working
146: definition of a ``carbon-dominated'' mixture, we adopted the convention
147: $X$(C) $\geq$ 0.5 for the mass fraction of carbon in the
148: envelope/atmosphere. Hence, we first constructed stellar models with an
149: envelope defined by $X$(C) = $X$(He) = 0.5, representing a limiting case
150: in the composition domain for carbon-atmosphere white dwarfs. In a
151: second step, and to provide a measure of the importance of varying the
152: carbon abundance, we also considered envelope compositions defined by
153: $X$(C) = 0.75 and $X$(He) = 0.25, $X$(C) = 0.9 and $X$(He) = 0.1, as
154: well as $X$(C) =0.99 and $X$(He) = 0.01. We explicitly ignored the
155: possible presence of small traces of hydrogen in our survey.
156:
157: It is worth pointing out here that the white dwarf models that we used are
158: static, but {\sl full} stellar structures nevertheless. To compute them,
159: we took advantage of the scaling law $L(r) \propto M(r)$, as appropriate
160: for a cooling white dwarf (see Brassard \& Fontaine 1994 for more details
161: on this). Given a composition profile, a model is further specified by
162: its effective temperature $T_{\rm eff}$ and surface gravity log $g$.
163: Equivalently, due to the specific mass-radius relationship that
164: applies to white dwarfs, the total mass could be specified instead of
165: the surface gravity, but we prefer the first alternative because it
166: connects better with spectroscopy. In addition, in the range of
167: effective temperature of interest here, convection is present in the
168: envelope of carbon-atmosphere white dwarfs (Fontaine \& Van Horn 1976),
169: and we described it through the so-called ML2 version of the mixing-length
170: theory (see Table 1 of Fontaine, Villeneuve, \& Wilson 1981 for the
171: original definition of ML2). We also briefly explored the effects of
172: changing the modelling of convection by computing some extra models
173: under the assumption of ML3 convection (a version similar to ML2 in
174: terms of the geometry of the eddies, but assuming a ratio of the mixing
175: length to the the local pressure scale height of $\alpha$ = 2 instead of
176: 1 as in ML2).
177:
178: We carried out a stability survey of these white dwarf models with the
179: help of our finite-element pulsation codes (Brassard et al. 1992a;
180: Fontaine et al. 1994). Given what is known about V777 Her and ZZ Ceti
181: pulsators, we specifically searched for unstable dipole ($\ell = 1$)
182: modes in the range of periods from 80 s to 1200 s. These would be
183: $g$-modes. We also concentrated on models cooler than $T_{\rm eff}$ =
184: 30,000 K to connect with the observed carbon-atmosphere white dwarfs of
185: Dufour et al. (2007). We point out that we used a full nonadiabatic
186: approach, but within the so-called frozen convection approximation. This
187: approximation leads to quite reasonable estimates of the location of the
188: blue edge of the ZZ Ceti instability strip in the log $g$--$T_{\rm eff}$
189: plane (see, e.g., the discussion in Brassard \& Fontaine 1999), and the
190: same is true for the V777 Her instability strip (Beauchamp et al.
191: 1999). At the same time, it is well known that the frozen convection
192: approximation leads to red edges of instability strips that are cooler,
193: and sometimes much cooler than they ought to. This caveat should be
194: kept in mind.
195:
196: A last remark about our equilibrium models concerns our choice for the
197: thickness of the outer carbon-dominated envelope, which we picked at a
198: fractional mass depth of log $q_{\rm env}$ $\equiv$
199: log $(1 - M(r)_{\rm env}/M_*)$ = $-$3.0 in all cases. At this depth, the
200: specified uniform envelope composition changes smoothly to the core
201: composition of $X$(C) = $X$(O) = 0.5 in our models. Clearly, the actual
202: pulsation periods must depend on this particular choice due to trapping
203: effects at the composition transition layer (see, e.g., Brassard et
204: al. 1992b). However, as discussed at some length in Quirion, Fontaine ,
205: \& Brassard (2004) in the context of GW Vir white dwarfs, it is not so
206: much the actual periods themselves that matter in the kind of stability
207: study that we carried out, but the {\sl range of unstable periods}. As
208: explicitly demonstrated by Quirion et al. (2004), that interval of
209: unstable periods is very largely independent of the actual choice of
210: log $q_{\rm env}$, as long as the base of the envelope is located much
211: deeper than the driving/damping region. The specific choice of log
212: $q_{\rm env}$ = $-$ 3.0 ensures just that.
213:
214: \begin{figure}[!ht]
215: \begin{center}
216: \begin{tabular}{l}
217: \includegraphics[width=9.2cm,height=9.2cm]{f1.eps}
218: \end{tabular}
219: \end{center}
220: \caption{Period spectra of excited dipole modes in carbon-atmosphere
221: white dwarf models. Three families of models are shown differing in
222: surface gravity (see the color code), but they have otherwise the same
223: envelope composition ($X$(C) = $X$(He) = 0.5) and ML2 convection. Each
224: dot gives the period of a mode, and its size represents a logarithmic
225: measure of the modulus of the imaginary part $\sigma_I$ of the complex
226: eigenfrequency. The bigger the dot, the more unstable the mode.}
227: \end{figure}
228:
229: Figure 1 summarizes some of our results for models with an envelope
230: composition specified by $X$(C) = $X$(He) = 0.5 and ML2 convection. Taking
231: into account the physical background briefly reported on above, we were
232: not really surprised, but still very much thrilled, to find out that
233: pulsation modes can be excited in some carbon-atmosphere white dwarf
234: models. This is because this finding potentially opens the door to the
235: application of the tools of asteroseismology for the further study of
236: this new class of carbon-atmosphere white dwarfs. The black dots in
237: Figure 1 map the expected instability strip for a reference family of
238: models with log $g$ = 8.0. Dipole modes in this family are excited in a
239: range of effective temperature from 18,400 K down to 12,600 K. For the
240: reasons given above, the red edge is probably calculated cooler than it
241: ought to, but the blue edge should be quite secure. The excited modes
242: are all low-order $g$-modes, with a radial order $k$ = 2 for the lowest
243: series of adjacent points (separated by 200 K in $T_{\rm eff}$), to $k$ =
244: 13 for the top three points illustrated. The total interval of excited
245: periods that is mapped for this family of models covers $\sim$ 200 s to
246: $\sim$ 700 s. We point out also that the e-folding times of the excited
247: modes illustrated in the figure are all much smaller than the typical
248: evolutionary timescale of a carbon-atmosphere white dwarf, a few times $10^8$
249: yrs, reaching in some cases (the largest dots in the plot) values as low
250: as 50 to 100 yrs. This means that such modes have plenty of time to
251: develop an observable anplitude.
252:
253: The theoretical blue edge for the log $g$ = 8.0 family of models displayed
254: in Figure 1 may fall a little short if we focus, for a moment, on the
255: current sample of nine carbon-atmosphere white dwarfs discovered by
256: Dufour et al. (2007) since those cluster in a range of $T_{\rm eff}$
257: from 23,000 K to 18,000 K. However, a simple shift in surface gravity
258: easily pushes the blue edge of the instability strip well within that
259: range as shown by the red and blue dots in Figure 1, referring,
260: respectively, to a family of models with log $g$ = 8.5 and log $g$ =
261: 9.0. Note that this displacement of the blue edge toward higher effective
262: temperatures is accompanied by a shift of the range of unstable periods
263: toward shorter values. This is not to be surprising at all because the
264: periods themselves (see, e.g., Brassard et al. 1992b in the case of ZZ
265: Ceti pulsators), i.e., those of given $k$ values, are known to decrease
266: when the surface gravity of a white dwarf increases. A fortiori, this is
267: also true for excited periods. Note further that the fact that the blue
268: edge temperature increases with increasing surface gravity is a
269: phenomenon that is also observed and well understood for the pulsating
270: DB (Beauchamp et al. 1999) and the pulsating DA white dwarfs (see, e.g.,
271: Gianninas, Bergeron, \& Fontaine 2006).
272:
273: \begin{figure}[!ht]
274: \begin{center}
275: \begin{tabular}{l}
276: \includegraphics[width=9.2cm,height=9.2cm]{f2.eps}
277: \end{tabular}
278: \end{center}
279: \caption{Similar to Fig. 1, but the red dots refer, this time, to the
280: reference family of models specified by log $g$ = 8.0, $X$(C) =
281: $X$(He) = 0.5, and ML2 convection. The blue dots were obtained from
282: similar models but using the more efficient ML3 version of the
283: mixing-length theory. The black (green, cyan) points again refer to
284: log $g$ = 8.0, ML2 models, but with an envelope composition defined by
285: $X$(C) = 0.75 and $X$(He) = 0.25 ($X$(C) = 0.9 and $X$(He) = 0.1,
286: $X$(C) = 0.99 and $X$(He) = 0.01).}
287: \end{figure}
288:
289: Figure 2 shows some other results of our exploratory survey, in
290: particular the effects of changing the chemical composition of the
291: envelope. The red dots correspond to our reference family of models with
292: log $g$ = 8.0, ML2 convection, and $X$(C) = $X$(He) = 0.5 in the
293: envelope. All other things being the same, the black dots show a
294: shrinking instability domain when the envelope composition is changed to
295: $X$(C) = 0.75 and $X$(He) = 0.25. There is a decrease of the blue edge
296: temperature from 18,400 K to 16,800 K in that case. The band of
297: predicted unstable dipole modes also reduces to the range $\sim$ 230 s to
298: $\sim$ 580 s. This trend continues to be verified in models with an
299: envelope composition defined by $X$(C) = 0.9 and $X$(He) = 0.1 as
300: indicated by the green dots. The predicted blue edge for this family of
301: log $g$ = 8.0, ML2 models is now at a value $T_{\rm eff}$ = 15,300 K.
302: Although the location of the blue edge does not change much more, the
303: overall instability domain shrinks further as can be seen from the cyan
304: dots associated with models with a very rich carbon composition in the
305: envelope, i.e., $X$(C) = 0.99 and $X$(He) = 0.01. This is a strong
306: indication that the actual amount of carbon in the envelope of real
307: carbon-atmosphere white dwarfs will be a decisive factor in the fate of
308: these stars as pulsators. Finally, the blue dots in Figure 2 illustrate
309: the significant blueward shift (some 2,600 K) of the predicted
310: instability strip and some widening of the instability domain when
311: convection is modelled in terms of the more efficient ML3 version
312: compared to our standard ML2 flavor.
313:
314: \begin{figure*}[!ht]
315: \begin{center}
316: \begin{tabular}{lcr}
317: \includegraphics[scale=0.45,angle=0]{f3a.eps}
318: \includegraphics[scale=0.45,angle=0]{f3b.eps}
319: \end{tabular}
320: \end{center}
321: \caption{\textit{Left panel}: Details of the driving/damping process for
322: a typical $g$-mode excited in a 17,000 K, ML2 model of a hot DQ
323: (carbon-atmosphere) white dwarf with a envelope composition $X$(C) =
324: $X$(He) = 0.5 and a gravity log $g$ = 8.0. The solid curve shows
325: the integrand of the work integral of the mode as a function of
326: fractional mass depth. The dashed curve shows the running work
327: integral, from left to right, toward the surface of the model. The red
328: dotted curve shows the ratio of the convective to total flux. The blue
329: dotted curve gives the run of the Rosseland opacity, to be read on
330: the RHS ordinate axis. The radiative opacity values were taken from the
331: OPAL data base. The maximum in the opacity profile, located at
332: log $q$ $\simeq$ $-$12.41 and corresponding to a temperature $T$
333: $\simeq$ $1.045 \times 10^5$ K, is caused by the partial ionization of
334: He II, C III, and C IV in the envelope mixture. The secondary maximum,
335: located at log $q$ $\simeq$ $-$8.98 and corresponding to a temperature $T$
336: $\simeq$ $1.177 \times 10^6$ K, is caused by the partial ionization of
337: C V, and C VI. The vertical green dotted line on the left
338: (right) gives the location of the base of the atmosphere at optical
339: depth $\tau_R$ = 100 (of the phoptosphere at $\tau_R$ = 2/3).
340: \textit{Right panel}: Similar and for the same $g$-mode, but in
341: a hotter model at 20,000 K for which all modes investigated are
342: stable. It can be seen that the value of the work integral $W$ at the
343: surface (dashed curve) is now negative, contrary to the case in the
344: left panel, and this signifies that the mode is not excited. Note also
345: the two separate convection zones in this model.}
346: \end{figure*}
347:
348: Given these most interesting results, we now turn briefly to the question
349: of the mechanism responsible for exciting pulsation modes in these models.
350: Figure 3 illustrates the details of the driving/damping region in two
351: models culled from the log $g$ = 8.0 family above, one ($T_{\rm eff}$ =
352: 17,000 K) featuring excited $g$-modes, and the other ($T_{\rm eff}$ =
353: 20,000 K) hotter than the blue edge and showing only stable modes.
354: Out of the many $g$-modes found excited in the 17,000 K model (left
355: panel), we have singled out a representative one with indices $k$ = 8
356: and $\ell$ = 1. It has a period of 489.1 s. We also retained the same mode,
357: this time stable and with a period of 451.8 s, in the 20,000 K model
358: (right panel). The figure illustrates a situation which bears strong
359: analogy with the case of the V777 Her and ZZ Ceti pulsators, but is more
360: complicated because of the presence of two maxima in the opacity
361: distribution instead of a single peak in these other pulsators. In the
362: present case, both opacity maxima (caused by two distinct partial
363: ionization zones in the envelope mixture) are ``active'' in the sense
364: that they both contribute to the driving/damping process. What we can
365: observe from the plot is that the regions on the descending side
366: (going in from the surface) of an opacity bump contribute locally to
367: driving, while the deeper adjacent zones, where the opacity plummets to
368: relatively low values, contribute instead to damping. In the 17,000 K
369: model, the two opacity bumps are relatively close to each other and are
370: part of a single convection zone. The damping region in between the two
371: bumps is then relatively narrow (see the left panel) and the overall
372: work integral comes out positive, meaning that the mode is globally
373: excited. In contrast, the two opacity maxima are relatively apart in the
374: 20,000 K model, so much so in fact that they lead to the formation of
375: two distinct, separate convection zones. In that case, there is a much
376: larger region in between the opacity bumps where {\sl radiative} damping
377: becomes so large that the overall work integral comes out negative.
378:
379: \section{Conclusion}
380:
381: Our exploratory survey of the asteroseismological potential of
382: carbon-atmosphere white dwarfs has allowed us to uncover regions of
383: parameter space where that potential can be fulfilled. For example,
384: low-order $g$-modes, typical of other pulsating white dwarfs such as the
385: V777 Her and ZZ Ceti stars, can be excited in a wide range of effective
386: temperature (18,400 K to 12,600 K) for models specified by log $g$ =
387: 8.0, ML2 convection, and an envelope composition given by $X$(C) =
388: $X$(He) = 0.5. Hotter models can become unstable also if their surface
389: gravity is increased or if the assumed convective efficiency used in the
390: model building phase is increased. The actual amount of carbon in the
391: envelope is crucial in determining the location and extent of the
392: instability strip. The extent of that domain tends to diminish when the
393: carbon abundance is increased. If real pulsating stars are found in the
394: current sample of carbon-atmosphere white dwarfs, we would expect them
395: to be massive, to show significant amounts of helium in their
396: atmospheres, and to be possibly undergoing efficient convective
397: transport.
398:
399: We have provided here the basic theoretical framework within which to
400: interpret possible discoveries of pulsations in carbon-atmosphere white
401: dwarfs. We are aware of at least one observational search being
402: currently pursued on that, and we are eagerly awaiting the results. We
403: note that the sample of carbon-atmosphere stars of Dufour et al. (2007)
404: is small, but it is likely, now that we know how to recognize them, that
405: more will be uncovered from future data release from the SDSS. If real
406: pulsators are found among that group, they will constitute the fourth
407: distinct category of pulsating white dwarfs after the GW Vir, V777 Her,
408: and ZZ Ceti stars. More detailed pulsation studies will certainly be
409: warranted in that eventuality.
410:
411: \begin{thebibliography}
412: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
413:
414: \bibitem[]{} Beauchamp, A., Wesemael, F., Bergeron, P., Fontaine, G.,
415: Saffer, R.A., Liebert, J., \& Brassard, P. 1999, ApJ, 516, 887
416:
417: \bibitem[]{} Brassard, P., \& Fontaine, G. 1994, in IAU Colloquium 147,
418: The Equation of State in Astrophysics, ed. G. Chabrier, \& E. Schatzman
419: (Cambridge: Cambridge University Press), 560
420:
421: \bibitem[]{} Brassard, P., \& Fontaine, G. 1999, in ASP Conf. Ser. 173,
422: Theory and Tests of Convection in Stellar Structure, ed. A. Giminez,
423: E.F. Guinan, \& B. Montesinos (San Francisco: ASP), 329
424:
425: \bibitem[]{} Brassard, P., Fontaine, G., Wesemael, F., \& Tassoul,
426: M. 1992b, ApJS, 81, 747
427:
428: \bibitem[]{} Brassard, P., Pelletier, C., Fontaine, G., \& Wesemael,
429: F. 1992a, ApJS, 80, 725
430:
431: \bibitem[]{} Dufour, P., Liebert, J., Fontaine, G., \& Behara, N. 2007,
432: Nature, 450, 522
433:
434: \bibitem[]{} Dufour, P. et al. 2008, in preparation
435:
436: \bibitem[]{} Fontaine, G. 1974, Ph.D. thesis, University of Rochester
437:
438: \bibitem[]{} Fontaine, G., Brassard, P., Wesemael, F., \& Tassoul,
439: M. 1994, ApJ, 428, L61
440:
441: \bibitem[]{} Fontaine, G., \& Van Horn, H.M. 1976, ApJS, 31, 467
442:
443: \bibitem[]{} Fontaine, G., Villeneuve, B., \& Wilson, J. 1981, ApJ, 243,
444: 550
445:
446: \bibitem[]{} Gianninas, A., Bergeron, P., \& Fontaine, G. 2006, AJ, 132,
447: 831
448:
449: \bibitem[]{} Quirion, P.-O., Fontaine, G., \& Brassard, P. 2004, ApJ,
450: 610, 436
451:
452: \end{thebibliography}
453:
454: \end{document}
455: