1: \documentclass[10pt,preprint]{aastex}
2:
3: \newcommand{\myemail}{keivan.stassun@vanderbilt.edu}
4: \newcommand{\bd}{2M0535--05}
5: \newcommand{\model}{{\sc wd}}
6:
7: \shorttitle{Brown-Dwarf Eclipsing Binary}
8: \shortauthors{Stassun et al.}
9:
10:
11: \begin{document}
12:
13: \title{A Surprising Reversal of Temperatures in the Brown-Dwarf
14: Eclipsing Binary 2MASS J05352184$-$0546085}
15:
16: \author{Keivan G.\ Stassun\altaffilmark{1},
17: Robert D.\ Mathieu\altaffilmark{2}, and
18: Jeff A.\ Valenti\altaffilmark{3} }
19:
20: \altaffiltext{1}{Department of Physics \& Astronomy,
21: Vanderbilt University, Nashville, TN 37235; \myemail}
22: \altaffiltext{2}{Department of Astronomy, University of
23: Wisconsin--Madison, Madison, WI, 53706}
24: \altaffiltext{3}{Space Telescope Science Institute, Baltimore,
25: MD 21218}
26:
27: \begin{abstract}
28: The newly discovered brown-dwarf eclipsing binary
29: \object{2MASS J05352184$-$0546085} provides a
30: unique laboratory for testing the predictions of theoretical models of
31: brown-dwarf formation and evolution. The finding that the lower-mass
32: brown dwarf in this system is hotter than its higher-mass companion
33: represents a challenge to brown-dwarf evolutionary models, none of
34: which predict this behavior. Here we present updated determinations
35: of the basic physical properties of \bd, bolstering the surprising
36: reversal of temperatures with mass in this system. We compare these
37: measurements with widely used brown-dwarf evolutionary tracks, and
38: find that the temperature reversal can be explained by some models
39: if the components of \bd\ are mildly non-coeval, possibly consistent
40: with dynamical simulations of brown-dwarf formation. Alternatively,
41: a strong magnetic field on the higher-mass brown dwarf might explain
42: its anomalously low surface temperature, consistent with emerging
43: evidence that convection is suppressed in magnetically active, low-mass
44: stars. Finally, we discuss future observational and theoretical work
45: needed to further characterize and understand this benchmark system.
46: \end{abstract}
47:
48:
49: \keywords{stars: low mass, brown dwarfs---binaries: eclipsing---stars:
50: fundamental parameters---stars: formation---stars:
51: individual (2MASS J05352184$-$0546085)}
52:
53:
54: \section{Introduction\label{intro}}
55: Born with masses between the least massive stars
56: \citep[$\approx 0.072\; {\rm M}_\odot$;][]{chabrier00}
57: and the most massive planets
58: \citep[$\approx 0.013\; {\rm M}_\odot$;][]{burrows01},
59: brown dwarfs at once extend our
60: understanding of the formation and evolution of both stars and planets.
61: In the decade since the discovery of the first brown dwarfs
62: \citep{nakajima95,rebolo95},
63: the study of star- and planet-formation has increasingly trained its
64: attention on these objects that, while being neither star
65: nor planet, may provide key insights to understanding the origins
66: of both \citep{basri00}.
67:
68: Such an understanding must be founded on accurate measurements
69: of fundamental physical properties---masses, radii, and
70: luminosities. Unfortunately, the number of objects for which these
71: properties can be measured directly is extremely small. Though
72: rare, eclipsing binaries have long been employed as ideal
73: laboratories for directly measuring fundamental stellar parameters
74: \citep[e.g.][]{andersen91}. The power of eclipsing binaries lies
75: in their provision of masses and radii with only the most basic of
76: theoretical assumptions. With the addition of atmosphere models the
77: ratio of effective temperatures can also be derived, and luminosities
78: can then be determined directly through the Stefan-Boltzmann law
79: without knowledge of distance. Finally, considering the two objects'
80: properties together permits study of the binary as twins at birth
81: whose evolutionary histories differ because of their different masses.
82:
83: Thus the recent discovery of a brown-dwarf eclipsing binary system
84: in the Orion Nebula Cluster, \object{2MASS J05352184$-$0546085}
85: (hereafter \bd), offers a unique laboratory with which to
86: directly and accurately test the predictions of theoretical
87: models of brown-dwarf formation and evolution \citep[][hereafter
88: Paper~I]{stass06}. In Paper~I, we presented a preliminary analysis of
89: the orbit and of the mutual eclipses in \bd\ to directly measure the
90: masses and radii of both components, as well as the ratio of their
91: effective temperatures. Our mass measurements reveal both objects in
92: this young binary system to be substellar, with masses of $M_1 = 55$
93: M$_{\rm Jup}$ and $M_2 = 35$ M$_{\rm Jup}$, accurate to $\sim 10\%$. In
94: addition, from the observed eclipse durations and orbital velocities
95: we directly measured the radii of the brown dwarfs to be $R_1 = 0.67$
96: R$_\odot$ and $R_2 = 0.51$ R$_\odot$, accurate to $\sim 5\%$, and
97: representing the first direct measurements of brown-dwarf radii. Such
98: large radii are generally consistent with theoretical predictions of
99: young brown dwarfs in the earliest stages of gravitational contraction.
100:
101: Surprisingly, however, we reported in Paper~I that the lower-mass
102: brown dwarf has an effective temperature that is slightly---but
103: significantly---warmer than its higher-mass companion. Such a reversal
104: of temperatures with mass is not predicted by any theoretical model
105: for coeval brown dwarfs. This finding has potentially important
106: ramifications for theoretical brown-dwarf evolutionary tracks, and
107: thus for our understanding of brown-dwarf formation more generally.
108:
109: In this paper, we present supplementary light-curve and radial-velocity
110: measurements (\S\ref{data}) which we use to refine our determination
111: of the basic physical properties of \bd\ (\S\ref{results}). We verify
112: the finding of a temperature reversal with mass in this system, and
113: quantitatively compare the empirically determined physical parameters
114: with those predicted by theoretical evolutionary tracks. We then
115: briefly explore the implications of the temperature reversal in
116: \bd\ for theoretical models of brown-dwarf formation and evolution
117: (\S\ref{discussion}). Specifically, we consider whether the temperature
118: reversal can be explained by dynamical brown-dwarf formation scenarios
119: and/or by the physical effects of strong surface fields on young
120: brown dwarfs. We summarize our conclusions in \S\ref{conclusions}.
121:
122: \section{Data and Methods\label{data}}
123:
124: \subsection{Spectroscopic observations: Radial velocities
125: and spectral types\label{spectra}}
126: We observed \bd\ with the {\it Phoenix} spectrograph on Gemini South
127: on eight separate nights from December 2002 to January 2003 (first
128: reported in Paper~I) and again on 03 March 2005 (newly reported
129: here). All nine observations were obtained with the same instrument
130: setup. Following \citet{mazeh02}, we observed the wavelength range
131: 1.5515--1.5585 $\mu$m (central wavelength 1.555 $\mu$m; $H$ band)
132: at a resolving power of $R \approx 30000$ (slit size of $0\farcs35$).
133: Exposure times were
134: varied between 1.0 and 3.3 hr by the queue observers based on sky
135: conditions. The typical exposure time was 2.6 hr (i.e.\ $\sim 0.01$
136: orbital phase; see \S\ref{lightcurve}), divided into 6 telescope nods
137: in an {\tt abbaab} pattern for the purposes of sky subtraction and
138: cosmic-ray rejection. An observation of the {\sc coravel-elodie}
139: high-precision radial-velocity standard \objectname{HD~50778}
140: \citep[K4 III;][]{udry99} was obtained immediately before or after
141: each observation of \bd\ to monitor the instrument wavelength zero
142: point. In addition, observations were obtained of a grid of late-type
143: spectral standards, with spectral types of M0 to M9, selected from
144: \citet{mazeh02} and \citet{mohanty03}. The standard-star observations
145: are summarized in Table~\ref{standards}.
146:
147: The observations were processed using Interactive Data Language
148: procedures developed specifically for optimal extraction of {\it
149: Phoenix} spectra. The background subtraction step includes logic
150: to compensate for variations in the night sky lines due to grating
151: motion or brightness variations between nods. A bootstrap procedure
152: is used to determine a wavelength solution from ThNeAr calibration
153: spectra obtained immediately before and after the science exposures.
154: First an {\it a priori} wavelength dispersion is determined by matching a
155: {\it Phoenix} spectrum of a K giant with the KPNO FTS atlas of Arcturus
156: \citep{hinkle95}. Next, a small wavelength shift (less than 1 km s$^{-1}$)
157: is applied to match the mean wavelengths of the three strongest
158: ThNeAr lines as a function of position along the slit. Finally, a
159: multi-Gaussian fit is used to empirically determine the wavelengths
160: of the remaining ThNeAr lines. The individual exposures from the
161: different slit positions are combined and interpolated onto a uniform
162: dispersion and then continuum normalized using sixth-order polynomial
163: fits to the high points in the spectra. Typical signal-to-noise ratios
164: of the extracted spectra were $\sim 15$ per resolution element for
165: the nine observations of \bd\ and $\sim 50$ per resolution element
166: for the standard stars.
167:
168: Absolute heliocentric radial velocities of the late-type standards
169: were determined relative to the {\sc coravel-elodie} standard star
170: HD~50778 (see above). To minimize the possibility of introducing
171: systematic velocity offsets due to large differences in spectral
172: type between this K4 standard and the late-M standards, we applied
173: velocity corrections in a step-wise fashion, determining the velocity
174: shift of each sequentially later type standard by cross-correlating
175: it with the standard from the previous step (e.g., M0 relative to K4,
176: M1 relative to M0, and so on). The radial velocities reported below
177: are thus formally on the {\sc coravel-elodie} system \citep{udry99}.
178:
179: Radial velocities of the two components of \bd\ were determined
180: via the technique of Broadening Functions (BFs) described by
181: \citet{rucinski99}. As discussed in that study, BFs are less prone
182: than simple cross-correlation techniques to ``peak pulling," in which
183: closely spaced correlation features in double-lined binaries can
184: alter the positions of the correlation peak centroids, particularly
185: when the velocity separation of the components is on the order of
186: the spectral resolution. The BF analysis requires a radial-velocity
187: template spectrum that is well matched in spectral type to the target;
188: we determined that the M6.5 star LHS~292 from our grid of standards
189: was the best template for this purpose (see below).
190:
191: In seven of the nine {\it Phoenix} spectra of \bd\ the BFs
192: show two clear peaks, an example of which is shown in Fig.\
193: \ref{bf-fig}. For these seven observations, radial velocities and their
194: uncertainties were determined using two-Gaussian fits to the BFs
195: \citep[e.g.][]{rucinski99}. We designate the component producing
196: the stronger peak in the BFs as the `primary' and the component
197: producing the weaker peak in the BFs the `secondary' (as we show
198: below, the primary component so defined is also the more massive
199: of the pair). One of the remaining two observations was by chance
200: taken during eclipse of the primary (orbital phase 0.75; see Fig.\
201: \ref{Ilc-fig}) and thus appears as a single-lined spectrum; the BF
202: peak was fit with a single Gaussian and its velocity assigned to the
203: secondary. The other observation occurs shortly after eclipse of the
204: secondary (orbital phase 0.12) and thus appears as a blended spectrum
205: producing a single, broadened BF peak; in this case two Gaussians
206: were fit, their initial centroids estimated from the preliminary orbit
207: solution of Paper~I and their widths fixed based on the widths of the
208: well-separated BF peaks from the other spectra. Radial velocities and
209: their uncertainties so measured are summarized in Table~\ref{rv-data}
210: and displayed in Fig.\ \ref{rv-fig}.
211:
212: We found that the BF peaks of both components appeared strongest when
213: we used LHS~292, an old M6.5 dwarf (see Table~\ref{standards}),
214: as the radial-velocity template in the BF analysis.
215: The BF peaks were $\sim 20\%$ weaker when
216: we used the M6 or M7 dwarfs as templates, and weaker still when we
217: used earlier or later templates.
218: This suggests that the spectral types of both
219: components of \bd\ are also $\sim$M6.5.
220: Indeed, we found in Paper~I
221: (cf.\ Fig.\ 3 in that paper) that the isolated spectrum of the
222: primary component of \bd\ very closely matches that of LHS~292,
223: and that the $JHK$ colors
224: of \bd\ imply a spectral type in combined light of M6.5$\pm 0.5$
225: with negligible reddening.
226: Taken together, the available evidence suggests that the components of
227: \bd\ have very similar spectral types of $\sim$M6.5$\pm 0.5$,
228: corresponding to $T_{\rm eff} \approx 2700$~K
229: \citep{slesnick04,goli04}.
230:
231: While our light-curve analysis below (\S\ref{analysis}) yields a
232: $T_{\rm eff}$ ratio of $\sim 1$, reinforcing the conclusion that
233: the components of \bd\ are very similar in spectral type,
234: we caution that our assignment of absolute spectral types is
235: preliminary and subject to systematic uncertainty.
236: The differences in spectral features seen in late-M standard stars
237: are extremely subtle for dwarfs with spectral types M4--M8
238: \citep[e.g.][]{bender05,prato02},
239: and the low S/N of our \bd\ spectra prohibits a detailed
240: analysis of spectral features.
241: Moreover, the low surface gravities and strong magnetic fields of
242: young, low-mass objects
243: can bias classification of their spectra, in the sense that the true
244: spectral types may be 1--2 subtypes later than inferred from comparison
245: to old, high-gravity dwarfs \citep{mohanty03}.
246: Thus, for our purposes here, we emphasize
247: that the spectral match between \bd\ and the adopted radial-velocity
248: template is sufficiently good to permit precise radial velocity
249: measurements, and that the spectral types of the components of \bd\
250: are evidently very similar to one another.
251:
252: Because the components of \bd\ have such similar spectral
253: types, the relative BF peak areas reflect the two components'
254: relative contributions to the total light of the system
255: \citep[e.g.][]{bayless06}. In particular, the primary component
256: apparently contributes approximately 50\% more flux than the
257: secondary at 1.555 $\mu$m (see Fig.\ \ref{bf-fig}). This flux
258: ratio provides an important additional constraint for removing
259: degeneracies in the determination of physical parameters, as we
260: discuss in \S\ref{analysis}.
261:
262: The component masses that we determine in \S\ref{analysis}
263: follow directly from an orbit solution fit to the observed radial
264: velocities, and thus the uncertainties in the derived masses (and
265: in other properties that in turn depend on those masses) depend
266: sensitively on uncertainties in the radial velocity measurements. The
267: mean precisions of the radial-velocity measurements from the
268: BF analysis are formally 1.7 and 1.9 km~s$^{-1}$ for the \bd\
269: primary and secondary, respectively (see $\sigma_{\rm RV}$ column of
270: Table~\ref{rv-data}). However, the accuracy of these measurements
271: can potentially be degraded by various systematic effects. At
272: the instrument level, the stability of the wavelength zero-point
273: is of particular concern. To assess this, we cross-correlated the
274: first observation of the {\sc coravel-elodie} standard star HD~50778
275: against each subsequent observation of this star, providing a measure
276: of the instrument stability over the course of our {\it Phoenix}
277: observations. The resulting radial velocities of HD~50778 exhibit
278: an r.m.s.\ scatter of 0.45 km~s$^{-1}$ on the nights that \bd\
279: was observed, which is much smaller than the random errors in the
280: individual measurements of \bd. Another possible source of systematic
281: error is spectral-type mismatch between \bd\ and the radial-velocity
282: template. However, as discussed above, the radial-velocity template
283: that we used in the BF analysis appears to match the spectral types
284: of the \bd\ components very well. Moreover, the components of \bd\
285: have very similar spectral types to one another, so that any systematic
286: effects due to spectral mismatch with the template should affect both
287: components similarly, in which case only the center-of-mass velocity
288: will be affected.
289:
290: Finally, the observed radial velocities can be affected by spots
291: on the surfaces of the brown dwarfs that cause phase-dependent
292: distortions in the shapes of the spectral lines, and in fact we
293: show in \S\ref{spots} that spots are clearly present on one or
294: both of the components in \bd. While the data suggest that this
295: effect is small, we have not yet incorporated a physical spot
296: model into our analysis and thus any spot signatures remain as
297: potential systematics in the radial velocities. In V1174 Ori, for
298: example, a young eclipsing binary with photometric spot amplitudes
299: similar to those observed in \bd, we found that the resulting
300: radial-velocity distortions were as large as $\sim 1$ km~s$^{-1}$ at
301: certain orbital phases \citep{stass04}. Analyses of other spotted,
302: low-mass eclipsing systems find similar effects; for example, GU Boo
303: shows distortions of $\sim 0.5$ km~s$^{-1}$ \citep{lopez05} and YY
304: Gem shows distortions of $\sim 1$ km~s$^{-1}$ \citep{torres02}. Any
305: spot-induced radial-velocity distortions in \bd\ would need to be
306: 2--3 times larger than in these systems to be comparable to the
307: random measurement errors of $\sim 1.5$--2 km~s$^{-1}$ \citep[for
308: an example of such a system, see][]{neuhauser98}. Encouragingly,
309: the residuals of the radial-velocity measurements with respect to
310: the final orbit solution of \S\ref{analysis} ($\chi_\nu^2 = 0.7$
311: and 1.2 for the primary and secondary velocities, respectively; see
312: Table~\ref{rv-data}) do not indicate that systematic effects dominate
313: the radial-velocity measurement errors.
314:
315: \subsection{Photometric observations: Light curves\label{lightcurve}}
316: We have continued to photometrically monitor \bd\ intensively with
317: the 1.0-m and 1.3-m telescopes at CTIO and, as of this writing,
318: possess a total of 2404 $I_C$-band measurements spanning the time
319: period 1994 December to 2006 April. The observing campaign to date is
320: summarized in Table~\ref{obs-table}, and the individual measurements
321: are provided in Table~\ref{lc-data}. We have excluded a small number
322: of observations with photometric errors larger than 0.1 mag. Thus,
323: photometric errors on the individual measurements have a range of
324: 0.01--0.1 mag, with a mean error of 0.024 mag and a median error of
325: 0.020 mag.
326:
327: A period search based on the phase dispersion minimization (PDM)
328: technique of \citet{stellingwerf78} reveals an unambiguous period of
329: $P = 9.7795557 \pm 0.000019$ d. The PDM technique is well suited to
330: periodic variability that is highly non-sinusoidal in nature, as is
331: the case for most detached eclipsing binaries. This updated period is
332: slightly shorter than, but not inconsistent with, the period reported
333: in Paper~I (see Table~\ref{properties-table}).
334:
335: In Fig.~\ref{Ilc-fig} we show the $I_C$-band light curve of \bd\
336: folded on this period. Two distinct eclipses are clearly evident
337: and cleanly separated in phase, as is typical of fully detached
338: eclipsing binaries. One eclipse---the `primary eclipse' at orbital
339: phase $\sim 0.075$---is notably deeper than the other; this marks
340: the time in the orbit when the hotter component is eclipsed which,
341: as discussed in Paper~I and below in \S\ref{reversal}, in this
342: system is the lower-mass component.
343:
344: In addition, the time from primary to secondary eclipse is longer
345: than that from secondary to primary eclipse, indicating an eccentric
346: orbit. More quantitatively, the two parameters that determine the
347: shape and orientation of the orbit---the eccentricity, $e$, and the
348: argument of periastron, $\omega$---can be estimated from geometrical
349: considerations relating the orbital period, the durations of the
350: eclipses (6.8 hr and 10.1 hr, respectively), and their 6.5-day
351: separation in time \citep[e.g.][]{kallrath99},
352: from which we estimate $e \approx 0.35$ and $\omega \approx 216^\circ$.
353: These initial estimates of $e$ and $\omega$ agree well with the more
354: precise values that we obtain from detailed light-curve modeling and
355: analysis below (\S\ref{analysis}; Table \ref{properties-table}).
356:
357: In addition to the extensive $I_C$-band light curve measurements,
358: we have also now obtained a set of near-infrared ($JHK$) light curves.
359: These are presented and analyzed in a subsequent paper \citep{gomez06},
360: but we note here that the same features observed
361: in the $I_C$-band light curve are also seen at these other wavelengths.
362:
363: \subsection{Analysis\label{analysis}}
364: As in Paper~I, we have performed a simultaneous analysis of the
365: {\it Phoenix} radial velocities and the $I_C$-band light curve
366: using the eclipsing-binary algorithms of \citet[][updated 2005;
367: hereafter \model]{wd05} as implemented in the {\sc phoebe} code
368: of \citet{prsa05}. \model\ has become standard in the modeling
369: and analysis of all manner of eclipsing binary systems, having
370: grown in sophistication over the past 35 years to include, e.g.,
371: phase-dependent projection effects in eccentric orbits, radial-velocity
372: perturbations arising from proximity and eclipse effects, apsidal
373: motion, and reflection and limb-darkening effects. In its most advanced
374: implementation, the code can also treat the effects of surface spots,
375: model atmospheres, and asynchronous rotation, including any attendant
376: gravity brightening and radial-velocity perturbations arising from
377: non-sphericity and from phase-dependent variations in the shapes of
378: the components.
379:
380: For the present study, we add information from the out-of-eclipse
381: variations in the light curve to model asynchronous rotation
382: of the components and as a first step toward accounting for the
383: presence of surface spots (\S\ref{spots}). As part of our ongoing
384: study of \bd, we plan to incorporate progressively more advanced
385: treatments, including a full investigation
386: of surface spots and inclusion of state-of-the-art brown-dwarf model
387: atmospheres. Here, as in Paper~I, we use simple blackbody spectra in
388: our light curve models. While brown-dwarf spectra are known to exhibit
389: strong departures from simple blackbodies, we show below that this
390: effect is unlikely to significantly alter our principal findings;
391: in particular, the reversal of temperatures with mass that we find
392: in \bd\ is probably not the result of improperly modeled spectra. On
393: the other hand, understanding the temperature reversal may ultimately
394: require a proper treatment of surface spots and their influence on
395: the time- and aspect-dependent emergent flux from the system.
396:
397: \subsubsection{Spot-related variations in the light curve\label{spots}}
398: To begin, we ran an initial \model\ fit to the light-curve and
399: radial-velocity data (Tables \ref{rv-data} and \ref{lc-data}) using
400: as initial values the system's orbital and physical parameters from
401: Paper~I, together with the updated orbital period from this study
402: (\S\ref{lightcurve}). The best-fit parameters and their formal
403: uncertainties are summarized in Table~\ref{properties-table}. The
404: reduced $\chi^2$ of this fit is large ($\chi_\nu^2=3.10$), and
405: the r.m.s.\ of the residuals relative to this fit (0.033 mag) is
406: accordingly larger than expected given the typical error on the
407: photometric measurements (\S\ref{lightcurve}). This suggests an
408: additional source of variability in the light curve---spots, for
409: example---not included in the model.
410:
411: Surface spots are most commonly manifested as low-amplitude, periodic,
412: roughly sinusoidal variations superposed on the eclipse light curve
413: \citep[e.g.][]{stass04}. We searched for such low-amplitude, periodic
414: variations in the light curve of \bd\ by applying a Lomb-Scargle
415: periodogram \citep{scargle82} analysis to those portions of the
416: light-curve data obtained outside of eclipse. We performed this
417: periodogram analysis separately on each epoch of light-curve data (see
418: Table~\ref{obs-table}), as spot-related variations often show evolution
419: in amplitude and phase with time \citep{stass99,torres02,stass04}.
420:
421: We find statistically significant periodic signals in all but the
422: first two epochs, these earlier epochs possessing relatively few
423: measurements. While the amplitudes of these periodic signals vary
424: slightly from epoch to epoch, they are always small ($\Delta m \lesssim
425: 0.06$ mag, peak-to-peak). Importantly, the periods of these signals are
426: consistent across epochs, with a value of $P_{\rm spot} \approx 3.3$~d.
427: Spot-related variability observed in other eclipsing binaries is
428: similarly characterized by photometric amplitudes of a few percent and
429: periods that are constant over many epochs \citep[e.g.][]{ribas03},
430: most likely reflecting the rotation period of one or both binary
431: components. The implication here is that at least one of the brown
432: dwarfs in \bd\ is nearly always spotted, and rotates with $P_{\rm rot}
433: \approx 3.3$~d.
434:
435: With a rotation period that is shorter than the orbital period by
436: a factor of (almost exactly) 3, the rotational and orbital angular
437: velocities of \bd\ have evidently not yet become pseudo-synchronized
438: with one another \citep[where the rotational angular velocity is
439: synchronized to the orbital angular velocity at periastron;][]{hut81},
440: which for the eccentricity of \bd\ would give $P_{\rm rot} / P_{\rm
441: orb} \approx 1/2$. Instead, the rotation of at least one component
442: in \bd\ is super-synchronous. Such super-synchronicity is probably
443: not surprising given the extreme youth of \bd\ (see below) and that
444: young brown dwarfs tend to be rapid rotators \citep{mohanty03}; tidal
445: effects have likely not yet had sufficient time to synchronize the
446: system \citep[e.g.][]{meibom05}.
447:
448: A more detailed discussion of rotation in \bd\ is beyond the scope of
449: this paper. For our present purposes, we proceed to adopt the observed
450: $P_{\rm rot} = 3.3$~d in our final \model\ fit as a constraint on
451: the oblateness of both \bd\ components. In addition, we rectify the
452: $I_C$-band light curve by subtracting a sinusoid of $P = 3.3$~d, with
453: the amplitude and phase determined separately for each epoch of data,
454: as described above. The aim of this rectification procedure is to
455: ``remove" the spot signal from the light curve and to thereby allow the
456: \model\ model to achieve goodness of fit without introducing additional
457: spot-fitting parameters that are poorly constrained at present. We
458: caution, however, that this procedure is {\it not} equivalent to
459: modeling the spots physically. The inclusion of additional light
460: curves at multiple wavelengths into our analyses \citep{gomez06}
461: will ultimately be required to constrain the physical properties of
462: the spots and to permit full modeling of their effects on the observed
463: light curves and radial velocities. Bearing these caveats in mind, we
464: proceed with our analysis below using the rectified $I_C$-band light
465: curve, and revisit the possible influence of spots in \S\ref{reversal}.
466:
467: \subsubsection{Determination of physical parameters\label{finalfit}}
468: The final \model\ solution was determined by simultaneously fitting
469: the {\it Phoenix} radial velocities and the $I_C$-band light curve,
470: rectified as described above. The resulting orbital and physical
471: parameters of \bd\ are summarized in Table~\ref{properties-table},
472: the fit to the radial velocities shown in Fig.\ \ref{rv-fig},
473: the fit to the light curve displayed in Fig.\ \ref{Ilc-fig},
474: and the system geometry illustrated in Fig.\ \ref{orbit-fig}.
475: With the light curve rectified, the
476: \model\ fit is now excellent; the r.m.s.\ of the residuals is 0.021
477: mag, comparable to the typical error on the photometric measurements
478: (\S\ref{lightcurve}), and the goodness-of-fit is correspondingly very
479: good, $\chi_\nu^2 = 1.035$.
480:
481: A comparison of the orbital and physical properties of
482: \bd\ derived from the rectified and unrectified light curves
483: (Table~\ref{properties-table}) reveals mostly minor, statistically
484: insignificant, differences in the fit parameters. However, a few
485: parameters do show differences that are larger than their formal
486: uncertainties. For example, both $e$ and $\omega$ differ by $\sim
487: 2\sigma$ (although these parameters are strongly correlated with one
488: another, so their uncertainties are not independent), suggesting
489: that the rectification procedure is not entirely free of small,
490: systematic couplings to some parameters. Thus, while a physical
491: treatment of spots in the \model\ model will ultimately be required
492: to achieve full accuracy in the determination of some parmaters, the
493: effects of spots on the most fundamental physical parameters that we
494: derive for \bd---the orbital parameters, in particular, and thus the
495: component masses and radii---are probably insignificant. Moreover,
496: if other parameters remain susceptible to spot-modeling effects,
497: these effects are evidently very subtle.
498:
499: We note that there exists an alternative
500: \model\ solution to the one presented in Table~\ref{properties-table},
501: with nearly identical best-fit parameters except that the
502: component radii are reversed. Formally, the goodness-of-fit of this
503: alternative solution is inferior, with $\chi_\nu^2 = 1.062$, but
504: it is nonetheless acceptable. This particular type of degeneracy
505: is a general feature of eclipsing-binary solutions \citep[see,
506: e.g.,][]{stass04,lopez05}. Since only the component radii are
507: significantly different in the two solutions, the degeneracy can be
508: broken by adding the additional constraint of the luminosity ratio,
509: which is generally very different in the two solutions. In this
510: case, the preferred solution from Table~\ref{properties-table}
511: gives an $H$-band luminosity ratio of $L_1/L_2 = 1.63 \pm 0.13$,
512: whereas the alternative solution gives $L_1/L_2 = 0.69 \pm 0.06$. The
513: luminosity ratio predicted by the alternative solution---in which
514: the secondary is both warmer and larger, and thus more luminous,
515: than the primary---is clearly inconsistent with the ratio inferred
516: from the observed spectra, which imply $L_1/L_2 \sim 1.5$ in the $H$
517: band (see Fig.\ \ref{bf-fig}).
518: %consistent in both sense and magnitude with the
519: %solution in Table~\ref{properties-table}.
520:
521: \section{Results\label{results}}
522: \subsection{Masses and Radii}
523: With masses of $M_1 = 0.0569 \pm 0.0046\; {\rm M}_\odot$ and $M_2
524: = 0.0358 \pm 0.0028\; {\rm M}_\odot$, the components of \bd\ are
525: proven to be {\it bona fide} brown dwarfs. Moreover, with radii
526: of $R_1 = 0.674 \pm 0.023 \; {\rm R}_\odot$ and $R_2 = 0.485 \pm
527: 0.018 \; {\rm R}_\odot$, these brown dwarfs are more comparable
528: in their physical dimensions to young, low-mass stars than to
529: old brown dwarfs, consistent with their being very young (see
530: below). Importantly, these mass and radius measurements are distance
531: independent. Moreover, systematic errors do not appear to dominate
532: over random measurement uncertainties (\S\ref{spectra}). Thus the
533: masses and radii of \bd\ reported here are accurate to 8\% and 3\%,
534: respectively. In \S\ref{tracks} we make use of these measurements for
535: an initial examination of the predictions of theoretical brown-dwarf
536: evolutionary models.
537:
538: \subsection{Reversal of temperatures\label{reversal}}
539: The physical properties that we have determined for the brown dwarfs in
540: \bd\ are surprising in at least one important respect: The less-massive
541: secondary is hotter than the primary. Specifically, we find $T_{\rm
542: eff,2}/T_{\rm eff,1} = 1.064 \pm 0.004$ (Table~\ref{properties-table}),
543: which follows from the relative depths of the eclipses and bolometric
544: corrections from the model atmospheres (blackbodies in this case),
545: with small corrections for differences in occulted areas that occur due
546: to the mildly eccentric orbit (Fig.\ \ref{orbit-fig}). This finding
547: is highly statistically significant. Such a reversal of temperatures
548: with mass is not predicted by any theoretical model for coeval brown
549: dwarfs---in which temperature increases monotonically with mass---and
550: thus merits closer scrutiny. Here we consider the possible effects
551: of non-blackbody atmospheres and surface spots.
552:
553: Brown dwarf spectra deviate significantly from an ideal blackbody,
554: due primarily to the strong wavelength dependence of molecular
555: opacity. The corresponding redistribution of flux into or out of any
556: specific wavelength depends on temperature, potentially affecting the
557: ratio of eclipse depths. However, the primary and secondary of \bd\
558: have very similar effective temperatures (\S\ref{spectra}), so both
559: components should have similar deviations from a blackbody.
560:
561: To check this quantitatively, we have examined up-to-date synthetic
562: models of brown-dwarf spectra for temperatures appropriate to the
563: components of \bd. In particular, we use the solar-metallicity {\sc
564: cond}\footnote{At the temperatures of the components of \bd, the
565: {\sc dusty} atmosphere models of these authors are very similar and
566: would work equally well for our purposes. See also \citet{mohanty03}.}
567: atmosphere models of \citet{allard01}. As the masses and radii
568: of \bd\ imply $\log g \approx 3.5$ for both components, we use the
569: {\sc cond} models with this $\log g$ value also. The {\sc cond}
570: models incorporate recent opacities, including $\sim 500$ million
571: molecular lines, as well as dust formation and condensation at very
572: low temperatures. (For a more detailed discussion of these models, and
573: their applicability to young brown dwarfs in particular, we point the
574: reader to the extensive discussion in \citet{mohanty03}). The results
575: of our analysis are summarized in Fig.\ \ref{speccomp-fig}. We find
576: that, while dramatic departures from simple blackbodies are clearly
577: present in the model spectra, the ratio of surface brightnesses in the
578: $I$ band is consistent with that predicted from simple blackbodies
579: to within 2\%, which implies a $T_{\rm eff}$ ratio that is consistent
580: with the blackbody value to within 0.5\%.
581:
582: As an additional check, we calculated the $T_{\rm eff}$'s of the
583: components of \bd\ independently at 1.25, 1.65, and 2.2 $\mu$m.
584: We use the ratio of surface fluxes at 0.8 $\mu$m (measured directly
585: from the ratio of eclipse depths in the $I_C$-band light curve;
586: \S\ref{analysis}) and the radii (measured directly from the eclipse
587: durations and the orbital velocities; Table~\ref{properties-table}),
588: together with an assumed distance to the Orion Nebula Cluster of 450
589: pc, empirical bolometric corrections from the literature
590: \citep{goli04}, and the observed $JHK$ magnitudes at these wavelengths
591: from \citet{carpenter01}:
592: $K=13.58 \pm 0.02$, $J-H = 0.640 \pm 0.015$, and $H-K = 0.385 \pm 0.015$.
593: We derive temperatures of $T_1 = 2730$ K and
594: $T_2 = 2875$ K (1.25 $\mu$m), $T_1 = 2675$ K and $T_2 = 2860$ K (1.65
595: $\mu$m), and $T_1 = 2680$ K and $T_2 = 2795$ K (2.2 $\mu$m). In all
596: cases, the derived $T_{\rm eff}$'s are consistent with a spectral type
597: of $\sim$M6.5, and in all cases the reversal of temperatures persists.
598:
599: Another potential contributor to non-blackbody spectra is the
600: presence of hot or cold spots (akin to solar plage or sunspots) on
601: the brown dwarfs, leading to spectra that derive from a combination
602: of temperatures. Cool spots have now been found to be present on many
603: brown dwarfs \citep[e.g.][]{scholz05} and may thus be expected on
604: one or both brown dwarfs in \bd\ as well. Indeed, we find a clear
605: periodic signal in the out-of-eclipse portions of the $I_C$-band
606: light curve, with a period of $P_{\rm spot} \approx 3.3$ d and a
607: semi-amplitude of $\sim 0.03$ mag (\S\ref{spots}); almost certainly
608: this is a rotationally modulated spot signal.
609:
610: Such spot signals are commonly observed in the light curves of
611: close eclipsing binaries, though modeling these effects is often
612: a largely cosmetic exercise with little effect on the resulting
613: stellar properties. For example, in our analysis of the young
614: eclipsing binary \object{V1174 Ori} \citep{stass04}, we successfully
615: reproduced the observed out-of-eclipse light-curve variations by
616: including spots in the light-curve model, but the derived system
617: parameters were altered only negligibly as a result \citep[see
618: also][]{torres02,ribas03,lopez05}.
619:
620: On the other hand, accurate modeling of spot effects can be crucially
621: important for the proper interpretation of certain systems. W~UMa
622: stars of the so-called ``W type" are a particularly good case
623: in point. A defining characteristic of these systems is that the
624: deeper eclipse corresponds to the occultation of the physically
625: smaller and lower-mass secondary, similar to what is observed in
626: \bd. One interpretation advanced early on by \citet{rucinski74}
627: is that the secondary is hotter than the primary (typically by
628: $\sim 5\%$), perhaps due to thermal effects arising from mass- and
629: energy-transfer in these contact systems. However, an alternate
630: interpretation \citep[e.g.][]{eaton80} is that cool spots on the
631: primary, if preferentially located along the the eclipsed latitude and
632: more-or-less uniformly distributed in longitude, can have the effect
633: of lowering the surface brightness of the eclipsed regions on the
634: primary during transit by the secondary, thus making the eclipse of the
635: primary shallower. Such a model is generally very
636: delicate in the details, and requires careful arrangement of the
637: putative spots in order that they produce only mild ($\lesssim 0.05$
638: mag) out-of-eclipse variations in the light curve; the inferred spot
639: configurations can resemble a ``leopard print" pattern in some cases
640: \citep[e.g.][]{linnell91}.
641:
642: We are now experimenting with more sophisticated light-curve models
643: that include the effects of spots.
644: In the meantime, we emphasize here that the ratio of eclipsed surface
645: brightnesses implied by our current light-curve model is robust;
646: the higher-mass primary really does radiate less per unit area
647: than the secondary, at least at those surface elements that are
648: eclipsed by the secondary. Thus, in what follows, we continue to
649: explore the implications of a temperature reversal in \bd.
650:
651: \subsection{Physical association with the Orion
652: Nebula star-forming region: Evidence for youth}
653: Strong evidence for the physical association of \bd\ with the
654: Orion Nebula Cluster, and hence of its youth, is provided by its
655: center-of-mass velocity. The observed value of $\gamma = 24.1
656: \pm 0.4$ km s$^{-1}$ (Table~\ref{properties-table}) is within
657: 1 km s$^{-1}$ of the systemic radial velocity ($25 \pm 1.5$ km
658: s$^{-1}$) of kinematic members of this active star-forming region
659: \citep{stass99,sicilia-aguilar05}.
660:
661: In addition, we can derive a distance to \bd\ by comparing
662: the total system luminosity to its observed flux. To
663: calculate the luminosity, we use the directly measured radii
664: (Table~\ref{properties-table}) together with the effective
665: temperatures and apply the Stefan-Boltzmann relation, $L=4\pi R^2
666: \sigma_B T_{\rm eff}^4$. The adopted spectral type of M6.5$\pm$0.5
667: for the \bd\ primary (\S\ref{spectra}) implies $T_{\rm eff,1} =
668: 2715 \pm 100$~K based on recent calibrations of the $T_{\rm eff}$
669: scale for brown dwarfs \citep{mohanty03,slesnick04,goli04}, though
670: as mentioned earlier, systematic uncertainties in the $T_{\rm
671: eff}$ scale may be as large as $\sim 200$~K. From the measured
672: temperature ratio of $T_{\rm eff,2}/T_{\rm eff,1} = 1.064 \pm
673: 0.004$ (Table~\ref{properties-table}), we obtain $T_{\rm eff,2} =
674: 2820 \pm 105$~K. This then gives component luminosities of $L_1 =
675: 0.0223 \pm 0.0034\; {\rm L}_\odot$ and $L_2 = 0.0148 \pm 0.0023\;
676: {\rm L}_\odot$, and a total system luminosity of $L = 0.0372 \pm
677: 0.0060\; {\rm L}_\odot$.
678: Adopting an apparent magnitude of $m_K=13.58\pm 0.02$ \citep{carpenter01}
679: and bolometric corrections appropriate for the observed $T_{\rm eff}$'s
680: \citep{goli04} yields a derived distance to \bd\ of $456 \pm 34$
681: pc, assuming no extinction.
682: This distance determination is consistent with
683: the distance to the Orion Nebula of $480 \pm 80$ pc \citep{genzel89}.
684:
685: The Orion Nebula Cluster is extremely young, with an age that has been
686: estimated to be just $1^{+2}_{-1}$ Myr \citep{palla99,hill97}. With
687: a center-of-mass velocity and distance that are both consistent with
688: membership in this cluster, \bd\ is probably also very young, likely
689: having formed within the past few Myr. In addition, as discussed
690: in Paper~I, the $JHK$ colors of \bd\ place an upper limit on the
691: extinction of $A_V< 0.75$, and thus limit the amount of remnant
692: material available to the brown dwarfs for ongoing accretion.
693: Given the eclipsing nature of the system, any disk material would
694: necessarily be seen edge-on and would thus produce a large amount of
695: extinction and reddening. Thus, while the colors of \bd\ are formally
696: consistent with a small amount of interstellar extinction and/or a
697: small amount of remnant disk material, the currently observed masses
698: are unlikely to change significantly over time. These brown dwarfs
699: will likely forever remain brown dwarfs.
700:
701: \subsection{Comparison to theoretical brown-dwarf evolutionary tracks
702: \label{tracks}}
703: While we expect that our ongoing analysis of \bd\ will further improve
704: the accuracy of the measured system parameters, the present accuracy is
705: sufficient to permit an initial examination of theoretical brown-dwarf
706: evolutionary models. Here we consider two of the more widely used
707: sets of tracks: those of \citet{baraffe98}\footnote{We use the tracks
708: with convection mixing-length parameter $\alpha=1.0$, as recent
709: analyses of young, low-mass stars and brown dwarfs suggest that such
710: low-efficiency convection best matches the observed physical properties
711: of these objects \citep[e.g.][]{stass04,mathieu06,torres06}.} and
712: those of \citet[][updated 1998]{dantona98}. These models differ in
713: their treatment of brown-dwarf atmospheres and of energy transport
714: (convection) in brown-dwarf interiors, as well as in the choice of
715: initial conditions \citep[for a more in-depth discussion of differences
716: in these models, see][]{siess00,baraffe02}.
717:
718: Fig.\ \ref{models-fig} compares the observed radii, effective
719: temperatures, and luminosities of \bd\ with the values theoretically
720: predicted by these models for brown dwarfs with masses of $M_1 =
721: 0.0569 \pm 0.0046\; {\rm M}_\odot$ and $M_2 = 0.0358 \pm 0.0028\;
722: {\rm M}_\odot$ (Table~\ref{properties-table}).
723:
724: We note first that, generally speaking, the agreement between the
725: observed and theoretically predicted properties of \bd\ is quite
726: good. The models predict that these very young brown dwarfs should,
727: at an age of $\sim 1$ Myr, be significantly larger, warmer, and more
728: luminous than their older counterparts---and that is in fact what
729: we see. Indeed, between 1 and 100 Myr, a brown dwarf with a mass of
730: 0.057 ${\rm M}_\odot$ (the mass of the \bd\ primary) is predicted
731: to shrink by 500\%, cool by several hundred K (or, equivalently,
732: go from an M spectral class to an L spectral class), and dim by 1.5
733: orders of magnitude (Fig.\ \ref{models-fig}). The recently measured
734: radius ($R = 0.12\; {\rm R}_\odot$) of the old and low-mass ($M =
735: 0.09\; {\rm M}_\odot$) star in the eclipsing binary system OGLE-TR-122
736: \citep{pont05} confirms that stars with near-brown-dwarf masses have
737: very small radii ($\sim 1$ R$_{\rm Jup}$) when they are old. Thus,
738: the fact that \bd\ comprises young brown dwarfs that are both large
739: and luminous---and even simply that they are of M spectral class,
740: as predicted---is a testament to the generally good predictive power
741: of current theoretical models of brown dwarfs.
742:
743: At a more detailed level, the models of \citet{dantona98} predict radii
744: and luminosities that are more consistent with the observed values. The
745: difference is most pronounced in the radii; the \citet{baraffe98}
746: models under-predict both radii by $\sim 10\%$. This finding of
747: under-predicted model radii is similar in sense and magnitude to that
748: found from recent efforts to derive the physical properties of low-mass
749: stars and brown dwarfs through detailed modeling of their spectra
750: \citep{mohanty04a,mohanty04b}. In addition, while both sets of models
751: perform reasonably well with respect to the observed luminosities,
752: the agreement is somewhat better for the \citet{dantona98} models;
753: it appears possible that more accurate measurements will reveal an
754: under-prediction of luminosity by the \citet{baraffe98} models for
755: the lower-mass secondary.
756:
757: But perhaps more importantly, the reversal of temperatures in \bd\
758: remains puzzling; the relationship between effective temperature and
759: mass is predicted by both sets of models to be monotonic for brown
760: dwarfs of the same age. However, this expectation disappears for two
761: brown dwarfs of differing age. The observed effective temperatures can
762: be seen as consistent with the \citet{dantona98} models if the primary
763: is taken as being modestly younger than the secondary ($\Delta\tau
764: \approx 0.5$ Myr). Indeed, the temperatures, radii, and luminosities of
765: \bd\ all remain in marginally good agreement with these models for such
766: an age difference, at a mean age of 1 Myr (Fig.\ \ref{models2-fig}),
767: if we also adjust the observed $T_{\rm eff}$ scale cooler by 70 K.
768: Such a shift is well within the current systematic uncertainty of the
769: brown-dwarf $T_{\rm eff}$ scale (\S\ref{spectra}). Thus, a question
770: raised by our findings is whether current brown-dwarf formation theory
771: can accommodate a scenario in which two brown dwarfs---that are part of
772: the same, very young, binary system---can have sufficiently different
773: ages to allow for reversed temperatures as we have observed in \bd.
774:
775:
776: \section{Discussion\label{discussion}}
777:
778: \subsection{\bd\ as a case study in dynamical brown-dwarf formation?}
779: In truth, we have very little understanding about how binaries
780: with components as close as those in \bd\ are formed \citep[see,
781: e.g.,][]{bonnell01,tohline02}. (\bd\ is, after PPl~15 in the Pleiades
782: \citep{basri99}, the shortest-period brown-dwarf binary system
783: yet discovered). With a few exceptions \citep[e.g.][]{tohline01},
784: fission seems to be ruled out. {\it In situ} mechanisms involving
785: dynamic cloud fragmentation and disk fragmentation have not yet proven
786: successful in creating such close binaries, but remain to be fully
787: developed. In particular, the role of orbital migration in the presence
788: of a massive circumbinary disk needs to be considered. Certainly,
789: if the two brown dwarfs formed together as a close binary, we have
790: essentially no {\it a priori} expectations for whether they should
791: be observed to be coeval to within 0.5 Myr.
792:
793: Alternatively, recent theoretical work \citep{reipurth01}, as well
794: as detailed numerical simulations \citep{bate02a,bate02b,bate05},
795: suggest that dynamical interactions may be integral to the
796: formation of brown dwarfs. The argument is essentially that strong
797: gravitational interactions in multiple-body encounters provide
798: a feasible mechanism for disrupting the accretion process, and
799: thereby preventing the accumulation of mass by objects that would
800: otherwise have become stars. This hypothesis remains under debate
801: \citep[e.g.][]{maxted05}. Still, it is tempting to speculate that
802: the components of \bd\ did not form together as a binary, but rather
803: formed separately---with the primary forming later---and then were
804: later married through a dynamical interaction. In such a scenario, it
805: is possible that the resulting binary system---comprising two objects
806: that were not originally formed together---may exhibit seemingly
807: peculiar characteristics, such as the observed reversal of effective
808: temperatures, that in fact reveal the non-coeval nature of the system.
809:
810: There are at least two specific scenarios involving multiple-body
811: interactions that might pertain to the origin of \bd. One involves
812: a low-mass binary pair that interacts with a more massive third body
813: from elsewhere in the cluster. Simulations show that, if the binary
814: is not simply broken apart by the encounter, the lower-mass member
815: of the binary is ejected and replaced by the massive incoming object
816: \citep{bate02a}. The resulting binary would then likely consist of
817: non-coeval members. A serious concern with this scenario in the context
818: of \bd\ is that it is unlikely in three-body encounters within clusters
819: that the most massive object would have a mass of only 0.06 M$_\odot$.
820:
821: Thus, a second scenario is that several objects formed in the
822: fragmentation of a small molecular core, forming an unstable multiple
823: system. The consequent rapid dynamical evolution of the system may have
824: both terminated accretion and formed a hard binary \citep{bate02b}. In
825: such a fragmentation scenario the relative core-collapse times and
826: evolutionary zero points might easily vary by 0.5 Myr, of order the
827: age difference needed to explain the temperature reversal of \bd\
828: in the context of some evolutionary models (Fig.\ \ref{models2-fig}).
829:
830: The position of \bd\ might be taken as further evidence that
831: dynamics have been important in its history. With a projected
832: separation of 2.8 pc from the center of the Orion Nebula, \bd\
833: is situated more than 10 core radii from the center of the
834: Orion Nebula Cluster \citep{hillenbrand98}; perhaps the pair was
835: ejected from the cluster center during a dynamical encounter. The
836: one-dimensional velocity dispersion in the cluster is $\sim 2$ km
837: s$^{-1}$ \citep{vanaltena88,hillenbrand98,stass99,sicilia-aguilar05},
838: so to reach its current position in 1 Myr, if ejected from the cluster
839: core, \bd\ would need a somewhat higher-than-usual velocity of $\sim 3$
840: km s$^{-1}$ in the plane of the sky. However,
841: the measured center-of-mass radial velocity of \bd\
842: does not deviate significantly from the cluster velocity.
843: An alternative
844: interpretation for the position of \bd\ at the outskirts of the
845: cluster may be that it is not a member of the Cluster proper, but
846: rather a member of the more widely distributed population of young
847: stars in the region surrounding the Orion Nebula \citep{warren78}. The
848: binary's proper motion is the critical measurement needed to establish
849: whether \bd\ was ejected from the cluster center.
850:
851: \subsection{Missing physics in theoretical brown-dwarf evolutionary
852: tracks?}
853: An alternative explanation is that the evolutionary tracks are
854: deficient with respect to some critical physical ingredient(s). For
855: example, the presence of strong magnetic fields on one or both brown
856: dwarfs in \bd\ could be affecting energy transport and thereby altering
857: their physical structure.
858:
859: Indeed, recent analyses of several young, low-mass eclipsing binaries
860: \citep[e.g.][]{stass04,covino04,torres06} indicate systematic
861: discrepancies in the models. In particular, the observed effective
862: temperatures are cooler than expected, and the observed radii larger
863: than expected. These discrepancies are especially pronounced among
864: the most magnetically active stars (as traced by X-ray emission and
865: other proxies). One possible interpretation is that strong magnetic
866: fields inhibit energy transport in these otherwise fully convective
867: stars, resulting in a decrease of the surface temperature and a
868: corresponding increase in radius so as to radiate the same total flux
869: \citep{montalban06}. Such an interpretation would be consistent
870: with, and would help explain, the emerging observational evidence
871: for suppressed convection in young, low-mass stars \citep[e.g.][and
872: references therein]{stass04,mathieu06,torres06}.
873:
874: In this context, the temperature reversal in \bd\ might be taken as
875: evidence for a strong magnetic field on the higher-mass primary that
876: is causing a sufficient decrease in its surface temperature to make
877: it effectively cooler than the lower-mass secondary. By inference,
878: the secondary would be interpreted as being less magnetically
879: active. The observational evidence is strong for magnetic activity
880: in young, low-mass stars and brown dwarfs of early- and mid-M type
881: \citep[e.g.][]{mohanty02,stass04b,preibisch05}. Moreover, the evidence
882: in fact shows a marked decline in magnetic activity at very late M
883: types \citep[e.g.][]{gizis00,mohanty02}, suggesting that brown dwarfs
884: with roughly the mass of the \bd\ secondary and below are not capable
885: of generating strong fields. The idea of a magnetically active primary
886: would also be consistent with the primary being heavily spotted (see
887: \S\ref{reversal}).
888:
889: If we are to explain the anomalously low effective temperature of
890: the \bd\ primary in this way, we should then also expect its radius
891: to be larger than theoretically predicted, as the one effect goes
892: hand in hand with the other \citep{montalban06}. As discussed above
893: (\S\ref{tracks}), whether the \bd\ radii agree with theoretical
894: predictions depends on one's choice of model. The \citet{baraffe98}
895: tracks do indeed suggest over-sized radii in \bd\ (for {\it both}
896: components), though this may simply reflect the truncation of
897: those tracks at 1 Myr. On the other hand, the \citet{dantona98} tracks
898: agree with the observed radii very well (Fig.\ \ref{models-fig}),
899: with no need to invoke missing physics.
900:
901: \section{Summary and Conclusions\label{conclusions}}
902: \bd\ is the first known eclipsing binary system comprising two
903: brown dwarfs. Satisfying both kinematic and distance requirements
904: for physical association with the young ($\sim 1$ Myr) Orion Nebula
905: Cluster, \bd\ provides the only direct, accurate measurements of
906: the fundamental physical properties of newly formed sub-stellar
907: objects. The masses that we measure are accurate to $\sim 10\%$,
908: the radii accurate to $\sim 5\%$, the ratio of effective temperatures
909: accurate to $\sim 1\%$, and all are distance independent. As such,
910: \bd\ represents an important benchmark for theoretical models of
911: brown-dwarf formation and evolution.
912:
913: Encouragingly, we find that current brown-dwarf evolutionary tracks
914: are, broadly speaking, successful in predicting the fundamental
915: physical properties of these young brown dwarfs. More quantitatively,
916: of the two sets of theoretical models considered here, we find that
917: the models of \citet{dantona98} yield mass-radius and mass-luminosity
918: relationships that best agree with the empirically determined ones. The
919: models of \citet{baraffe98} predict radii and luminosities that are
920: 1.5--2$\sigma$ smaller than the observed values.
921:
922: However, the reversal of component effective temperatures with mass
923: in \bd\ is unexpected and unexplained. We have considered here two
924: possible interpretations of this intriguing result. The first is that
925: the components of \bd\ are mildly non-coeval, with the higher-mass
926: primary being $\sim 0.5$ Myr younger than the secondary. The models of
927: \citet{dantona98} are in fact consistent with the observed temperature
928: reversal for such an age difference. A second hypothesis is that
929: strong magnetic activity on the primary is inhibiting convection,
930: and thereby lowering its surface temperature.
931:
932: Neither of these interpretations is wholly satisfying, and neither
933: is obviously discreditable. Binary formation theory is largely
934: silent on the subject of coevality at the level of $\sim 0.5$ Myr,
935: and theorists have long warned the star-formation community about
936: the limited applicability of evolutionary track chronometry at such
937: early ages (the model zero-points being arbitrary in most cases). In
938: addition, while the observational evidence is strong that young brown
939: dwarfs can be magnetically active, the effects of magnetic fields
940: on brown-dwarf structure and evolution have yet to be consistently
941: modeled or fully understood.
942:
943: \acknowledgments
944: This research is supported by NSF grant AST-0349075, and by a Cottrell
945: Scholar award from the Research Corporation, to K.G.S., and by NSF
946: grant AST-9731302 to R.D.M.
947: This work is based in part on observations obtained through queue
948: program GS-2002B-Q-11 at the Gemini Observatory, which is operated
949: by the Association of Universities for Research in Astronomy, Inc.,
950: under a cooperative agreement with the NSF on behalf of the Gemini
951: partnership: the National Science Foundation (United States), the
952: Particle Physics and Astronomy Research Council (United Kingdom),
953: the National Research Council (Canada), CONICYT (Chile), the
954: Australian Research Council (Australia), CNPq (Brazil) and CONICET
955: (Argentina).
956: The WIYN Observatory is a joint facility of the University of
957: Wisconsin--Madison, Indiana University, Yale University, and the
958: National Optical Astronomy Observatories.
959:
960:
961: \begin{thebibliography}{}
962:
963: \bibitem[Allard et al.(2001)]{allard01} Allard, F., Hauschildt, P.~H.,
964: Alexander, D.~R., Tamanai, A., \& Schweitzer, A.\ 2001, \apj, 556, 357
965:
966: \bibitem[Andersen(1991)]{andersen91} Andersen, J.\ 1991, \aapr, 3, 91
967:
968: \bibitem[Baraffe et al.(1998)]{baraffe98} Baraffe, I., Chabrier, G.,
969: Allard, F., \& Hauschildt, P.~H.\ 1998, \aap, 337, 403
970:
971: \bibitem[Baraffe et al.(2002)]{baraffe02} Baraffe, I., Chabrier, G.,
972: Allard, F., \& Hauschildt, P.~H.\ 2002, \aap, 382, 563
973:
974: \bibitem[Basri(2000)]{basri00} Basri, G.\ 2000, \araa, 38, 485
975:
976: \bibitem[Basri \& Mart{\'{\i}}n(1999)]{basri99} Basri, G., \&
977: Mart{\'{\i}}n, E.~L.\ 1999, \aj, 118, 2460
978:
979: \bibitem[Bate \& Bonnell(2005)]{bate05} Bate, M.~R., \& Bonnell,
980: I.~A.\ 2005, \mnras, 356, 1201
981:
982: \bibitem[Bate et al.(2002a)]{bate02a} Bate, M.~R., Bonnell, I.~A.,
983: \& Bromm, V.\ 2002a, \mnras, 332, L65
984:
985: \bibitem[Bate et al.(2002b)]{bate02b} Bate, M.~R., Bonnell, I.~A.,
986: \& Bromm, V.\ 2002b, \mnras, 336, 705
987:
988: \bibitem[Bayless \& Orosz(2006)]{bayless06} Bayless, A.~J., \&
989: Orosz, J.~A.\ 2006, \apj, 651, 1155
990:
991: \bibitem[Bender et al.(2005)]{bender05} Bender, C., Simon, M.,
992: Prato, L., Mazeh, T., \& Zucker, S.\ 2005, \aj, 129, 402
993:
994: \bibitem[Bonnell(2001)]{bonnell01} Bonnell, I.~A.\ 2001, IAU Symposium,
995: 200, 23
996:
997: \bibitem[Burrows et al.(2001)]{burrows01} Burrows, A., Hubbard, W.~B.,
998: Lunine, J.~I., \& Liebert, J.\ 2001, Reviews of Modern Physics, 73, 719
999:
1000: \bibitem[Carpenter et al.(2001)]{carpenter01} Carpenter, J.~M.,
1001: Hillenbrand, L.~A., \& Skrutskie, M.~F.\ 2001, \aj, 121, 3160
1002:
1003: \bibitem[Carpenter(2001)]{2mass} Carpenter, J.~M.\ 2001, \aj, 121, 2851
1004:
1005: \bibitem[Chabrier \& Baraffe(2000)]{chabrier00} Chabrier, G., \&
1006: Baraffe, I.\ 2000, \araa, 38, 337
1007:
1008: \bibitem[Covino et al.(2004)]{covino04} Covino, E., Frasca, A.,
1009: Alcal{\'a}, J.~M., Paladino, R., \& Sterzik, M.~F.\ 2004, \aap,
1010: 427, 637
1011:
1012: \bibitem[D'Antona \& Mazzitelli(1997)]{dantona98} D'Antona, F., \&
1013: Mazzitelli, I.\ 1997, Memorie della Societa Astronomica Italiana,
1014: 68, 807
1015:
1016: \bibitem[Eaton et al.(1980)]{eaton80} Eaton, J.~A., Wu, C.-C., \&
1017: Rucinski, S.~M.\ 1980, \apj, 239, 919
1018:
1019: \bibitem[Genzel \& Stutzki(1989)]{genzel89} Genzel, R., \& Stutzki,
1020: J.\ 1989, \araa, 27, 41
1021:
1022: \bibitem[Gizis et al.(2000)]{gizis00} Gizis, J.~E., Monet, D.~G.,
1023: Reid, I.~N., Kirkpatrick, J.~D., Liebert, J., \& Williams, R.~J.\
1024: 2000, \aj, 120, 1085
1025:
1026: \bibitem[Golimowski et al.(2004)]{goli04} Golimowski, D.~A., et al.\
1027: 2004, \aj, 127, 3516
1028:
1029: \bibitem[G\'{o}mez Maqueo Chew et al.(2007)]{gomez06}
1030: G\'{o}mez Maqueo Chew, Y., Stassun, K.~G., Vaz, L.~P.~R., \& Mathieu,
1031: R.~D.\ 2007, in preparation
1032:
1033: \bibitem[Hillenbrand \& Hartmann(1998)]{hillenbrand98} Hillenbrand,
1034: L.~A., \& Hartmann, L.~W.\ 1998, \apj, 492, 540
1035:
1036: \bibitem[Hillenbrand(1997)]{hill97} Hillenbrand, L.~A.\ 1997, \aj,
1037: 113, 1733
1038:
1039: \bibitem[Hinkle et al.(1995)]{hinkle95} Hinkle, K., Wallace, L., \&
1040: Livingston, W.\ 1995, \pasp, 107, 1042
1041:
1042: \bibitem[Hut(1981)]{hut81} Hut, P.\ 1981, \aap, 99, 126
1043:
1044: \bibitem[Kallrath \& Milone(1999)]{kallrath99} Kallrath, J., \&
1045: Milone, E.~F.\ 1999, Eclipsing binary stars : modeling and analysis
1046: / Joseph Kallrath, Eugene F.~Milone.~ New York : Springer, 1999.~
1047: (Astronomy and astrophysics library)
1048:
1049: \bibitem[Linnell(1991)]{linnell91} Linnell, A.~P.\ 1991, \apj, 383, 330
1050:
1051: \bibitem[Linnell(1987)]{linnell87} Linnell, A.~P.\ 1987, \apj, 316, 389
1052:
1053: \bibitem[L{\'o}pez-Morales \& Ribas(2005)]{lopez05} L{\'o}pez-Morales,
1054: M., \& Ribas, I.\ 2005, \apj, 631, 1120
1055:
1056: \bibitem[Mathieu et al.(2006)]{mathieu06} Mathieu, R.~D., Baraffe, I.,
1057: Simon, M., Stassun, K.G., \& White, R.\ 2006, Protostars and Planets V,
1058: in press
1059:
1060: \bibitem[Maxted \& Jeffries(2005)]{maxted05} Maxted, P.~F.~L., \&
1061: Jeffries, R.~D.\ 2005, \mnras, 362, L45
1062:
1063: \bibitem[Mazeh et al.(2002)]{mazeh02} Mazeh, T., Prato, L., Simon,
1064: M., Goldberg, E., Norman, D., \& Zucker, S.\ 2002, \apj, 564, 1007
1065:
1066: \bibitem[Meibom \& Mathieu(2005)]{meibom05} Meibom, S., \& Mathieu,
1067: R.~D.\ 2005, \apj, 620, 970
1068:
1069: \bibitem[Mohanty et al.(2004b)]{mohanty04b} Mohanty, S., Jayawardhana,
1070: R., \& Basri, G.\ 2004, \apj, 609, 885
1071:
1072: \bibitem[Mohanty et al.(2004a)]{mohanty04a} Mohanty, S., Basri, G.,
1073: Jayawardhana, R., Allard, F., Hauschildt, P., \& Ardila, D.\ 2004,
1074: \apj, 609, 854
1075:
1076: \bibitem[Mohanty \& Basri(2003)]{mohanty03} Mohanty, S., \& Basri,
1077: G.\ 2003, \apj, 583, 451
1078:
1079: \bibitem[Mohanty et al.(2002)]{mohanty02} Mohanty, S., Basri, G.,
1080: Shu, F., Allard, F., \& Chabrier, G.\ 2002, \apj, 571, 469
1081:
1082: \bibitem[Montalb{\'a}n \& D'Antona(2006)]{montalban06} Montalb{\'a}n,
1083: J., \& D'Antona, F.\ 2006, \mnras, 713
1084:
1085: \bibitem[Nakajima et al.(1995)]{nakajima95} Nakajima, T., Oppenheimer,
1086: B.~R., Kulkarni, S.~R., Golimowski, D.~A., Matthews, K., \& Durrance,
1087: S.~T.\ 1995, \nat, 378, 463
1088:
1089: \bibitem[Neuhaeuser et al.(1998)]{neuhauser98} Neuhaeuser, R., et al.\
1090: 1998, \aap, 334, 873
1091:
1092: \bibitem[Palla \& Stahler(1999)]{palla99} Palla, F., \& Stahler,
1093: S.~W.\ 1999, \apj, 525, 772
1094:
1095: \bibitem[Pont et al.(2005)]{pont05} Pont, F., Melo, C.~H.~F., Bouchy,
1096: F., Udry, S., Queloz, D., Mayor, M., \& Santos, N.~C.\ 2005, \aap,
1097: 433, L21
1098:
1099: \bibitem[Prato et al.(2002)]{prato02} Prato, L., Simon, M.,
1100: Mazeh, T., McLean, I.~S., Norman, D., \& Zucker, S.\ 2002, \apj, 569, 863
1101:
1102: \bibitem[Preibisch et al.(2005)]{preibisch05} Preibisch, T., et al.\
1103: 2005, \apjs, 160, 401
1104:
1105: \bibitem[Pr{\v s}a \& Zwitter(2005)]{prsa05} Pr{\v s}a, A., \& Zwitter,
1106: T.\ 2005, \apj, 628, 426
1107:
1108: \bibitem[Rebolo et al.(1995)]{rebolo95} Rebolo, R., Zapatero-Osorio,
1109: M.~R., \& Martin, E.~L.\ 1995, \nat, 377, 129
1110:
1111: \bibitem[Reipurth \& Clarke(2001)]{reipurth01} Reipurth, B., \&
1112: Clarke, C.\ 2001, \aj, 122, 432
1113:
1114: \bibitem[Ribas(2003)]{ribas03} Ribas, I.\ 2003, \aap, 398, 239
1115:
1116: \bibitem[Rucinski(1999)]{rucinski99} Rucinski, S.\ 1999, ASP
1117: Conf.~Ser.~185: IAU Colloq.~170: Precise Stellar Radial Velocities,
1118: 185, 82
1119:
1120: \bibitem[Rucinski(1974)]{rucinski74} Rucinski, S.~M.\ 1974, Acta
1121: Astronomica, 24, 119
1122:
1123: \bibitem[Scargle(1982)]{scargle82} Scargle, J.~D.\ 1982, \apj, 263, 835
1124:
1125: \bibitem[Scholz \& Eisl{\"o}ffel(2005)]{scholz05} Scholz, A., \&
1126: Eisl{\"o}ffel, J.\ 2005, \aap, 429, 1007
1127:
1128: \bibitem[Sicilia-Aguilar et al.(2005)]{sicilia-aguilar05}
1129: Sicilia-Aguilar, A., et al.\ 2005, \aj, 129, 363
1130:
1131: \bibitem[Siess et al.(2000)]{siess00} Siess, L., Dufour, E., \&
1132: Forestini, M.\ 2000, \aap, 358, 593
1133:
1134: \bibitem[Slesnick et al.(2004)]{slesnick04} Slesnick, C.~L.,
1135: Hillenbrand, L.~A., \& Carpenter, J.~M.\ 2004, \apj, 610, 1045
1136:
1137: \bibitem[Stassun, Mathieu, \& Valenti(2006)]{stass06} Stassun, K.~G.,
1138: Mathieu, R.~D., \& Valenti, J.~A.\ 2006, \nat, 440, 311 (Paper~I)
1139:
1140: \bibitem[Stassun et al.(2004a)]{stass04} Stassun, K.~G., Mathieu,
1141: R.~D., Vaz, L.~P.~R., Stroud, N., \& Vrba, F.~J.\ 2004a, \apjs,
1142: 151, 357
1143:
1144: \bibitem[Stassun et al.(2004)]{stass04b} Stassun, K.~G., Ardila,
1145: D.~R., Barsony, M., Basri, G., \& Mathieu, R.~D.\ 2004b, \aj, 127, 3537
1146:
1147: \bibitem[Stassun et al.(1999)]{stass99} Stassun, K.~G., Mathieu,
1148: R.~D., Mazeh, T., \& Vrba, F.~J.\ 1999, \aj, 117, 2941
1149:
1150: \bibitem[Stellingwerf(1978)]{stellingwerf78} Stellingwerf, R.~F.\
1151: 1978, \apj, 224, 953
1152:
1153: \bibitem[Tohline(2002)]{tohline02} Tohline, J.~E.\ 2002, \araa, 40, 349
1154:
1155: \bibitem[Tohline \& Durisen(2001)]{tohline01} Tohline, J.~E., \&
1156: Durisen, R.~H.\ 2001, IAU Symposium, 200, 40
1157:
1158: \bibitem[Torres et al.(2006)]{torres06} Torres, G., Lacy, C.~H.,
1159: Marschall, L.~A., Sheets, H.~A., \& Mader, J.~A.\ 2006, \apj, 640, 1018
1160:
1161: \bibitem[Torres \& Ribas(2002)]{torres02} Torres, G., \& Ribas, I.\
1162: 2002, \apj, 567, 1140
1163:
1164: \bibitem[Udry et al.(1999)]{udry99} Udry, S., Mayor, M., \& Queloz,
1165: D.\ 1999, ASP Conf.~Ser.~185: IAU Colloq.~170: Precise Stellar Radial
1166: Velocities, 185, 367
1167:
1168: \bibitem[van Altena et al.(1988)]{vanaltena88} van Altena, W.~F., Lee,
1169: J.~T., Lee, J.-F., Lu, P.~K., \& Upgren, A.~R.\ 1988, \aj, 95, 1744
1170:
1171: \bibitem[Warren \& Hesser(1978)]{warren78} Warren, W.~H., Jr., \&
1172: Hesser, J.~E.\ 1978, \apjs, 36, 497
1173:
1174: \bibitem[Wilson \& Devinney(1971)]{wd05} Wilson, R.~E., \& Devinney,
1175: E.~J.\ 1971, \apj, 166, 605
1176:
1177: \end{thebibliography}
1178:
1179: \clearpage
1180:
1181: \begin{deluxetable}{lcrc}
1182: \tablecolumns{4}
1183: \tablewidth{0pt}
1184: \tablecaption{Standard stars\label{standards}}
1185: \tablehead{
1186: \colhead{Name} & \colhead{SpTy} & \colhead{Exp.\ time (s)} & \colhead{Ref.}
1187: }
1188: \startdata
1189: GJ~328 & M0 & 180 & \citet{mazeh02} \\
1190: GJ~382 & M1.5 & 180 & \citet{mazeh02} \\
1191: GJ~447 & M4 & 120 & \citet{mohanty03} \\
1192: GJ~406 & M6 & 120 & \citet{mohanty03} \\
1193: LHS~292 & M6.5 & 3600 & \citet{mazeh02} \\
1194: LHS~3003 & M7 & 900 & \citet{mohanty03} \\
1195: GJ~3655 & M8 & 3600 & \citet{mohanty03} \\
1196: BRIB 1507$-$0229 & M9 & 3600 & \citet{mohanty03} \\
1197: \enddata
1198: \end{deluxetable}
1199:
1200: \clearpage
1201:
1202: \begin{deluxetable}{rrrrr}
1203: \tablecolumns{5}
1204: \tablewidth{0pt}
1205: \tablecaption{Radial velocity measurements of \bd\label{rv-data}}
1206: \tablehead{
1207: \colhead{HJD\tablenotemark{a}} & \colhead{Phase\tablenotemark{b}} &
1208: \colhead{R.V.\tablenotemark{c}} & \colhead{$\sigma_{\rm RV}$} &
1209: \colhead{$(O-C)$\tablenotemark{d}} \\
1210: \colhead{} & \colhead{} & \colhead{km s$^{-1}$} & \colhead{km s$^{-1}$} &
1211: \colhead{km s$^{-1}$}
1212: }
1213: \startdata
1214: \cutinhead{Primary}
1215: 2452623.74805 & 0.75 & \nodata & \nodata & \nodata \\
1216: 2452624.75701 & 0.85 & 9.59 & 1.41 & 1.34 \\
1217: 2452625.72555 & 0.95 & 2.00 & 2.22 & 1.32 \\
1218: 2452626.65154 & 0.04 & 9.84 & 1.75 & $-$2.79 \\
1219: 2452649.72938 & 0.40 & 38.02 & 1.41 & 1.30 \\
1220: 2452650.67281 & 0.50 & 34.21 & 1.14 & 0.31 \\
1221: 2452655.74512 & 0.02 & 6.61 & 2.56 & $-$0.98 \\
1222: 2452656.70824 & 0.12 & 25.58 & 1.28 & $-$0.84 \\
1223: 2453432.57920 & 0.45 & 37.09 & 2.11 & 1.62 \\
1224: & & & \multicolumn{2}{c}{$\chi_\nu^2 = 0.7$} \\
1225: \cutinhead{Secondary}
1226: 2452623.74805 & 0.75 & 29.69 & 1.17 & $-$2.50 \\
1227: 2452624.75701 & 0.85 & 48.53 & 1.73 & $-$0.41 \\
1228: 2452625.72555 & 0.95 & 61.35 & 2.07 & 0.40 \\
1229: 2452626.65154 & 0.04 & 42.43 & 2.97 & 0.43 \\
1230: 2452649.72938 & 0.40 & 4.18 & 1.24 & 0.37 \\
1231: 2452650.67281 & 0.50 & 5.07 & 2.00 & $-$3.21 \\
1232: 2452655.74512 & 0.02 & 50.04 & 2.35 & 0.05 \\
1233: 2452656.70824 & 0.12 & 22.58 & 1.31 & 2.45 \\
1234: 2453432.57920 & 0.45 & 6.64 & 2.64 & 0.85 \\
1235: & & & \multicolumn{2}{c}{$\chi_\nu^2 = 1.2$} \\
1236: \enddata
1237: \tablenotetext{a}{Heliocentric Julian Date.}
1238: \tablenotetext{b}{Orbital phase, relative to ephemeris of
1239: Table~\ref{properties-table}.}
1240: \tablenotetext{c}{Heliocentric radial velocity.}
1241: \tablenotetext{d}{Residual relative to \model\ solution (see text).}
1242: \end{deluxetable}
1243:
1244: \clearpage
1245:
1246: \begin{deluxetable}{rccr}
1247: \tablecolumns{4}
1248: \tablewidth{0pt}
1249: \tablecaption{Summary of $I_C$-band time-series photometry\label{obs-table}}
1250: \tablehead{
1251: \colhead{Obs.} & \colhead{Telescope} &
1252: \colhead{HJD range\tablenotemark{a}} &
1253: \colhead{$N_{\rm obs}$\tablenotemark{b}}
1254: }
1255: \startdata
1256: 1 & KPNO 0.9-m & 49699.70 -- 49703.93 & 40 \\
1257: 2 & WISE 1.0-m & 49698.35 -- 49714.42 & 47 \\
1258: 3 & WIYN 0.9-m & 52227.78 -- 52238.01 & 102 \\
1259: 4 & WIYN 0.9-m & 52595.75 -- 52624.95 & 144 \\
1260: 5 & SMARTS 0.9-m & 52622.57 -- 52631.51 & 122 \\
1261: 6 & SMARTS 1.3-m & 52922.73 -- 53081.57 & 347 \\
1262: 7 & SMARTS 0.9-m & 53011.57 -- 53024.77 & 205 \\
1263: 8 & SMARTS 1.3-m & 53280.74 -- 53340.73 & 230 \\
1264: 9 & SMARTS 1.0-m & 53373.56 -- 53386.79 & 184 \\
1265: 10 & SMARTS 1.3-m & 53403.53 -- 53445.60 & 338 \\
1266: 11 & SMARTS 1.3-m & 53646.82 -- 53727.69 & 204 \\
1267: 12 & SMARTS 1.0-m & 53719.59 -- 53727.83 & 190 \\
1268: 13 & SMARTS 1.3-m & 53745.64 -- 53846.48 & 251 \\
1269: \enddata
1270: \tablenotetext{a}{Range of Heliocentric Julian Dates ($2400000+$).}
1271: \tablenotetext{b}{Number of observations.}
1272: \end{deluxetable}
1273:
1274: \clearpage
1275:
1276: \begin{deluxetable}{rrrr}
1277: \tablecolumns{4}
1278: \tablewidth{0pt}
1279: \tablecaption{Differential $I_C$-band light curve of \bd\label{lc-data}}
1280: \tablehead{
1281: \colhead{HJD\tablenotemark{a}} & \colhead{$\Delta m$\tablenotemark{b}} &
1282: \colhead{$\sigma_m$} & \colhead{Obs.\tablenotemark{c}}
1283: }
1284: \startdata
1285: 2453686.70525 & 0.006 & 0.020 & 11 \\
1286: 2453686.71293 & 0.016 & 0.054 & 11 \\
1287: 2453687.76841 & $-$0.004 & 0.052 & 11 \\
1288: 2453687.77617 & $-$0.006 & 0.052 & 11 \\
1289: 2453687.78389 & 0.020 & 0.029 & 11 \\
1290: 2453687.79158 & $-$0.002 & 0.029 & 11 \\
1291: 2453688.73523 & $-$0.001 & 0.020 & 11 \\
1292: 2453688.74295 & $-$0.008 & 0.061 & 11 \\
1293: 2453688.75066 & 0.022 & 0.020 & 11 \\
1294: 2453688.75835 & 0.030 & 0.020 & 11 \\
1295: 2453689.77018 & 0.262 & 0.025 & 11 \\
1296: 2453689.77783 & 0.220 & 0.020 & 11 \\
1297: 2453689.79309 & 0.237 & 0.062 & 11 \\
1298: 2453690.71340 & 0.008 & 0.020 & 11 \\
1299: \enddata
1300: \tablenotetext{a}{Heliocentric Julian Date}
1301: \tablenotetext{b}{Differential $I_C$ magnitude (arbitrary zero-point).}
1302: \tablenotetext{c}{Source of measurement (see Table~\ref{obs-table}).}
1303: \tablecomments{The full table is available in the electronic version of
1304: the Journal. A portion is shown here for guidance regarding its form
1305: and content.}
1306: \end{deluxetable}
1307:
1308: \clearpage
1309:
1310: \begin{deluxetable}{lrrr}
1311: \rotate
1312: \tablecolumns{4}
1313: \tablewidth{0pt}
1314: \tablecaption{Orbital and physical parameters of
1315: \bd\label{properties-table}}
1316: \tablehead{
1317: \colhead{} & \colhead{Paper~I} & \multicolumn{2}{c}{This study} \\
1318: \colhead{} & \colhead{} & \colhead{Unrectified\tablenotemark{a}} & \colhead{Rectified\tablenotemark{a}}
1319: }
1320: \startdata
1321: Orbital period, $P$ [d] & $9.779621 \pm 0.000042$ & \multicolumn{2}{c}{$9.779556 \pm 0.000019$\tablenotemark{b}} \\
1322: Time of periastron (Bessellian year), $T_0$ & $2001.863650 \pm 0.000095$ & $2001.863903 \pm 0.000160$ & $2001.863765 \pm 0.000071$ \\
1323: Eccentricity, $e$ & $0.3225 \pm 0.0060$ & $0.3354 \pm 0.0049$ & $0.3276 \pm 0.0033$ \\
1324: Orientation of periastron, $\omega$ [$^\circ$] & $215.4 \pm 1.1$ & $219.2 \pm 1.4$ & $217.0 \pm 0.9$ \\
1325: Semi-major axis, $a \sin i$ [{\sc a.u.}] & $0.0398 \pm 0.0010$ & $0.0406 \pm 0.0016$ & $0.0406 \pm 0.0010$ \\
1326: Center-of-mass velocity, $\gamma$ [km s$^{-1}$] & $24.1 \pm 0.4$ & $24.1 \pm 0.4$ & $24.1 \pm 0.4$ \\
1327: Mass ratio, $q \equiv M_2/M_1$ & $0.625 \pm 0.018$ & $0.622 \pm 0.022$ & $0.631 \pm 0.015$ \\
1328: Total mass, $(M_1 + M_2) \sin^3 i$ [M$_\odot$] & $0.0880 \pm 0.0076$ & $0.0932 \pm 0.0111$ & $0.0932 \pm 0.0073$ \\
1329: Inclination, $i$ [$^\circ$] & $88.8 \pm 0.2$ & $89.4 \pm 0.3$ & $89.2 \pm 0.2$ \\
1330: Primary semi-amplitude, $K_1$ [km s$^{-1}$] & \nodata & $18.37 \pm 1.01$ & $18.49 \pm 0.67$ \\
1331: Secondary semi-amplitude, $K_2$ [km s$^{-1}$] & \nodata & $29.55 \pm 1.24$ & $29.30 \pm 0.81$ \\
1332: Primary mass, $M_1$ [M$_\odot$] & $0.0541 \pm 0.0046$ & $0.0575 \pm 0.0069$ & $0.0572 \pm 0.0045$ \\
1333: Secondary mass, $M_2$ [M$_\odot$] & $0.0340 \pm 0.0027$ & $0.0358 \pm 0.0043$ & $0.0360 \pm 0.0028$ \\
1334: Primary radius, $R_1$ [R$_\odot$] & $0.669 \pm 0.034$ & $0.673 \pm 0.037$ & $0.675 \pm 0.023$ \\
1335: Secondary radius, $R_2$ [R$_\odot$] & $0.511 \pm 0.026$ & $0.485 \pm 0.029$ & $0.486 \pm 0.018$ \\
1336: Primary gravity, $\log g_1$ & \nodata & $3.62 \pm 0.14$ & $3.62 \pm 0.10$ \\
1337: Secondary gravity, $\log g_2$ & \nodata & $3.54 \pm 0.14$ & $3.54 \pm 0.09$ \\
1338: Effective temperature ratio, $T_2/T_1$ & $1.054 \pm 0.006$ & $1.062 \pm 0.006$ & $1.064 \pm 0.004$ \\
1339: \enddata
1340: \tablenotetext{a}{\model\ solutions based on fits to the unrectified
1341: and rectified $I_C$-band light curve (see \S\ref{analysis}).}
1342: \tablenotetext{b}{Uncertainty in the period is from a phase dispersion
1343: minimization \citep{stellingwerf78} analysis on the $I_C$-band light curve
1344: (see \S\ref{lightcurve}); the period is held fixed in the \model\ fit
1345: and its uncertainty propagated into the uncertainties of derived
1346: quantities.}
1347: \end{deluxetable}
1348:
1349: \clearpage
1350:
1351: \figcaption[f1.ps]{\label{bf-fig}
1352: Determination of the radial velocities of the primary and secondary
1353: components of \bd\ using the Broadening Function (BF) analysis of
1354: \citet{rucinski99}. The example BF shown here (solid line) is from the
1355: spectrum of HJD 2452649 (see Table~\ref{rv-data}), where the primary
1356: and secondary are at near maximum velocity separation. An M6.5 dwarf
1357: was used as the radial-velocity template (see \S\ref{spectra}). The
1358: dashed line shows a two-gaussian fit to the BF, from which velocity
1359: centroids and their formal uncertainties are determined. The ratio
1360: of the peak areas from this fit is 1.6, indicating that the primary
1361: dominates the light of the system, contributing $\sim 60\%$ more
1362: flux than the secondary at the wavelength of these observations
1363: (1.555 $\mu$m). }
1364:
1365: %\clearpage
1366:
1367: \figcaption[f2.ps]{
1368: \label{rv-fig}
1369: Radial velocity measurements of \bd. The individual radial
1370: velocity measurements from Table~\ref{rv-data} are plotted
1371: (primary measurements in
1372: green, secondary measurements in red), folded on the orbital period and
1373: phased to the time of periastron (see Table~\ref{properties-table}).
1374: The solid lines are \model\ models based on a simultaneous fit to the
1375: radial velocity measurements and the rectified $I_C$-band light curve
1376: (see Table~\ref{lc-data} and Fig.\ \ref{Ilc-fig}). Distortions in the
1377: model curves near phases 0.075 and 0.75 are due to the brief occultations
1378: of the approaching and receding limbs of each component when it is
1379: eclipsed. Residuals are shown at bottom. Note: This figure appears in
1380: color in the electronic version of the journal.
1381: }
1382:
1383: %\clearpage
1384:
1385: \figcaption[f3.ps]{
1386: \label{Ilc-fig}
1387: $I_C$-band light curve of \bd. The individual photometric
1388: measurements from Table~\ref{lc-data} are plotted, rectified using
1389: sinusoidal fits to the out-of-eclipse portions of the light curve (see
1390: \S\ref{analysis}), folded on the orbital period and phased to the
1391: time of periastron (see Table~\ref{properties-table}). The solid line
1392: is a \model\ model based on a simultaneous fit to the rectified
1393: $I_C$-band light curve and the {\it Phoenix} radial velocities (see
1394: Table~\ref{rv-data}). Residuals are shown at bottom. Insets show
1395: detail around primary and secondary eclipses, which in this system
1396: correspond to the eclipses of the secondary and primary components,
1397: respectively. The r.m.s.\ residual is 0.02 mag, comparable to the mean
1398: photometric error. The reduced $\chi^2$ of the fit is 1.035. Note:
1399: this figure appears in color in the electronic version of the journal.
1400: }
1401:
1402: %\clearpage
1403:
1404: \figcaption[f4.ps]{
1405: \label{orbit-fig}
1406: The geometry of the orbital plane of \bd\ in the rest frame of the
1407: primary (more massive) brown dwarf is illustrated, to scale, at primary
1408: eclipse (orbital phase 0.075; see Fig.\ \ref{Ilc-fig}). The brown
1409: dwarfs are represented by filled circles, their radii also to scale
1410: (the adopted rotation period of $P_{\rm rot} = 3.3$ d constrains the
1411: oblateness of both components to be less than 0.5\%). The position
1412: of periastron (orbital phase 0.0) is indicated by a star symbol, and
1413: small arrows indicate the direction of the secondary's orbit about
1414: the primary. The observer is to the left, as indicated by the arrow.
1415: Note that, at this orbital phase corresponding to the deeper eclipse,
1416: the secondary (less massive) brown dwarf is the one eclipsed.
1417: }
1418:
1419: %\clearpage
1420:
1421: \figcaption[f5.ps]{
1422: \label{speccomp-fig}
1423: We use the {\sc cond} atmosphere models of \citet{allard01} to check
1424: whether the non-blackbody spectra of brown dwarfs may explain the
1425: unexpected reversal of effective temperatures with mass in \bd. {\it
1426: Upper panel}: {\sc cond} model spectra at the effective temperatures
1427: of the primary (green) and secondary (red) components of \bd\ are
1428: compared with simple blackbodies of the same temperatures (dashed
1429: lines). Dotted vertical lines demarcate the approximate bandpass of the
1430: $I_C$ filter used for our light-curve observations. {\it Bottom left}:
1431: The flux ratio of the {\sc cond} models in the upper panel is shown
1432: relative to the blackbody flux ratio as a function of wavelength in
1433: the $I_C$ band. Also
1434: shown is the $I_C$-band filter bandpass profile (dotted curve) and
1435: the bandpass-weighted mean {\sc cond}-to-blackbody flux ratio (dashed
1436: line), which is 1.02 in this case; that is, the {\sc cond} flux ratio
1437: is within 2\% of the blackbody value. {\it Bottom right}: Same as
1438: {\it bottom left}, except showing the $T_{\rm eff}$ ratio implied
1439: by the {\sc cond} flux ratio. The bandpass-weighted value is 1.069
1440: (dashed line), within 0.5\% of the value of $1.064 \pm 0.004$ found in
1441: our \model\ analysis (\S\ref{analysis}) using simple blackbodies. Note:
1442: This figure appears in color in the electronic version of the journal.
1443: }
1444:
1445: %\clearpage
1446:
1447: \figcaption[f6.ps]{
1448: \label{models-fig}
1449: Comparisons of observations with theoretical models of young brown
1450: dwarfs. The measured radii, effective temperatures, and luminosities of
1451: \bd\ (symbols with error bars) are compared to the values predicted by
1452: two sets of theoretical models \citep{baraffe98,dantona98} of young
1453: brown dwarfs. Solid curves show the predicted evolution from 0.1 to
1454: 100 Myr for brown dwarfs with masses equal to those measured for
1455: the primary (green) and secondary (red) brown dwarfs in \bd. (The
1456: theoretical calculations of \citet{baraffe98} do not extend to
1457: ages less than 1 Myr.) Dashed curves bracketing the solid curves
1458: represent the $1\sigma$ measurement uncertainties in those masses. Observed
1459: and predicted values are generally in good agreement, particularly with the
1460: \citet{dantona98} models. Note that the effective temperature of the
1461: primary is predicted by both sets of models to be warmer than that of
1462: the secondary at any particular age, but that the \citet{dantona98}
1463: models allow the primary to be cooler than the secondary if it is
1464: sufficiently younger (see also Fig.\ \ref{models2-fig}). Note: This
1465: figure appears in color in the electronic version of the journal.
1466: }
1467:
1468: %\clearpage
1469:
1470: \figcaption[f7.ps]{
1471: \label{models2-fig}
1472: Same as Fig.\ \ref{models-fig}, except with the \bd\ components
1473: arbitrarily separated in age by 0.5 Myr ($\tau_1 = 0.75$ Myr, $\tau_2 =
1474: 1.25$ Myr). This demonstrates that the temperature reversal observed in
1475: \bd\ may in fact be marginally consistent with the \citet{dantona98}
1476: models if the \bd\ primary is taken to be slightly younger than
1477: the secondary. Note: This figure appears in color in the electronic
1478: version of the journal.
1479: }
1480:
1481: \clearpage
1482: \begin{figure}
1483: \plotone{f1.ps}
1484: \centerline{f1.ps}
1485: \end{figure}
1486:
1487: \clearpage
1488:
1489: \begin{figure}
1490: \rotate
1491: \plotone{f2.ps}
1492: \centerline{f2.ps}
1493: \end{figure}
1494: \clearpage
1495:
1496: \begin{figure}
1497: \rotate
1498: \plotone{f3.ps}
1499: \centerline{f3.ps}
1500: \end{figure}
1501: \clearpage
1502:
1503: \begin{figure}
1504: \plotone{f4.ps}
1505: \centerline{f4.ps}
1506: \end{figure}
1507: \clearpage
1508:
1509: \begin{figure}
1510: \plotone{f5.ps}
1511: \centerline{f5.ps}
1512: \end{figure}
1513: \clearpage
1514:
1515: \begin{figure}
1516: \rotate
1517: \plotone{f6.ps}
1518: \centerline{f6.ps}
1519: \end{figure}
1520: \clearpage
1521:
1522: \begin{figure}
1523: \plotone{f7.ps}
1524: \centerline{f7.ps}
1525: \end{figure}
1526:
1527: \end{document}
1528:
1529:
1530: \end{document}
1531:
1532: