1: \documentclass[useAMS,usenatbib]{mn2e}
2:
3: \title[The Eccentric Accretion Disc of A0620-00]
4: {The Eccentric Accretion Disc of the Black Hole A0620-00\thanks{This
5: paper includes data gathered with the 6.5-m Magellan Telescopes
6: located at Las Campanas Observatory, Chile.}}
7: \author[Neilsen, Steeghs, \& Vrtilek]
8: {J.~Neilsen,$^{1}$\thanks{E-mail: jneilsen@cfa.harvard.edu (JN);
9: dsteeghs@cfa.harvard.edu (DS); svrtilek@cfa.harvard.edu (SDV)}
10: D.~Steeghs,$^{2,3}$ S.D.~Vrtilek$^{2}$\\
11: $^{1}$Harvard University Department of Astronomy, 60 Garden Street,
12: MS-10, Cambridge, MA 02138\\
13: $^{2}$Smithsonian Astrophysical Observatory, 60 Garden Street,
14: Cambridge, MA 02138\\
15: $^{3}$Department of Physics, University of Warwick, Coventry CV4 7AL, UK}
16: \date{Released 2007 Xxxxx XX}
17:
18: \pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{2007}
19:
20: \usepackage{latexsym}
21: \usepackage{color}
22: \usepackage{graphicx}
23: \usepackage{fancyheadings}
24: \usepackage{lscape}
25:
26: \begin{document}
27:
28: \label{firstpage}
29:
30: \maketitle
31:
32: \begin{abstract}
33: We present spectroscopic observations of the quiescent black hole
34: binary A0620-00 with the the 6.5-m Magellan Clay telescope at Las Campanas
35: Observatory. We measure absorption-line radial velocities of the
36: secondary and make the most precise determination to date
37: ($K_{2}=435.4\pm0.5$ km s$^{-1}$). By fitting the rotational broadening of
38: the secondary, we
39: refine the mass ratio to $q=0.060\pm0.004$; these results, combined with
40: the orbital period, imply a minimum mass for
41: the compact object of $3.10\pm0.04$ M$_{\sun}.$ Although
42: quiescence implies little accretion activity, we find that the disc
43: contributes $56\pm7$ per cent of the light in B and V, and is subject to
44: significant flickering. Doppler maps of the Balmer lines reveal bright
45: emission from the gas stream-disc impact point and unusual
46: crescent-shaped features. We also find that the disc centre of symmetry does
47: not coincide with the predicted black hole velocity. By comparison
48: with SPH simulations, we identify this source with an eccentric
49: disc. With high S/N, we pursue modulation tomography of
50: H$\alpha$ and find that the aforementioned bright regions are strongly
51: modulated at the orbital period. We interpret this modulation in the
52: context of disc precession, and discuss cases for the accretion disc
53: evolution.
54: \end{abstract}
55:
56: \begin{keywords}
57: accretion, accretion discs --- stars: individual: A0620-00 --- binaries: close
58: \end{keywords}
59:
60: \section{Introduction}
61: A0620-00 (V616 Mon) is the prototype Soft X-ray Transient, a class of
62: low-mass binary stars which exhibit infrequent but intense X-ray
63: bursts (Gelino, Harrison, and Orosz, 2001). In 1975 it became the brightest
64: X-ray nova ever detected, at approximately 50 Crab \citep{Elvis75},
65: and it was the first nova to be identified with a black
66: hole primary (McClintock \& Remillard, 1986; hereafter MR86). MR86
67: measured an orbital period of 7.75 hr and a radial velocity
68: semiamplitude for the K-type secondary of 457 km s$^{-1}$, leading to a mass
69: function $f(M)=3.18$ M$_{\sun};$ estimates of $K_{2}$ and $f(M)$ have
70: decreased slightly since then (i.e. 433 km s$^{-1}$ and 3.09 M$_{\sun}$)
71: (Marsh, Robinson, \& Wood 1994, hereafter MRW94). Given this minimum
72: mass, it is likely that A0620-00 is a black hole.
73:
74: A substantial amount of work has gone into the analysis of A0620-00 in
75: the last twenty years, with particular emphasis on ellipsoidal
76: variations in the light curve and the contamination of the K-star flux by
77: light from the accretion disc. As yet, no real consensus has been
78: reached, mostly due to the complexity of the lightcurves. While
79: ellipsoidal variations are obvious, they are highly asymmetric
80: (Leibowitz, Hemar, \& Orio 1998); the origin of the asymmetry is
81: undetermined. Modelling this lightcurve, \citet{Gelino01} determined in
82: inclination of 41$\pm 3\degr$, invoking starspots to explain the
83: asymmetries. Shahbaz, Naylor, and Charles (1994)
84: found a 90 per cent confidence interval of $i=$30--45$\degr$ given the
85: mass ratio of A0620-00, modelling their asymmetries with the bright
86: spot where the accretion stream hits the disc.
87:
88: Lightcurve modelling is also complicated by the variability of
89: the disc itself. To quantify ellipsoidal variations, most
90: authors assume the disc to be constant, and justify the claim by
91: noting that A0620-00 is quiescent. They do not mention that estimates
92: of the disc contamination range from $<$3 per cent \citep{Gelino01}
93: to $\la50$ per cent (MR86). The contribution from this disc is
94: not only unclear, but apparently not constant. More than half of A0620-00's
95: 58-year burst cycle has passed, and it is important to note that
96: quiescent does not mean inactive. We will argue that the variability
97: of the accretion disc cannot be neglected. In order to determine
98: definitively the mass of the compact object, it is very
99: important to understand the structure and variation of the
100: accretion disc. MRW94 made enormous progress towards this goal.
101: In 2004, Shahbaz et al. (hereafter S04) noticed signatures of an
102: eccentric disc not seen in previous Doppler maps, but lacked the phase
103: coverage to verify their hypothesis.
104:
105: Therefore, as follow-up to the work of MRW94 and S04, and as part of a
106: Doppler imaging survey of black hole and neutron star binaries, we
107: undertook phase-resolved optical spectroscopy of
108: A0620-00. In $\S 2$ we describe our observational methods and data
109: reduction. In $\S 3$ we measure the radial velocity of the secondary
110: star, the system mass ratio, and attempt to quantify flickering. In
111: $\S 4$ we present Doppler images of the accretion disc at several
112: wavelengths, investigate evidence for an eccentric disc, and report
113: results of modulation tomography of the
114: H$\alpha$ line. We discuss conclusions from the variability of the
115: disc and our Doppler maps in $\S 5.$
116:
117: \section{OBSERVATIONS}
118:
119: We observed A0620-00 with the Low-Dispersion Survey Spectrograph
120: (LDSS3) at the f/4 focus of the 6.5-m Clay
121: telescope at Las Campanas observatory on 2006 December
122: 14--16. We acquired 48 spectra using the VPH Blue grism and a
123: long 0.75 arcsec slit. By shifting the slit 4$\degr$ redward, we were
124: able to observe H$\alpha$ with the superior resolution of the Blue grism
125: (2.3 \AA~$\equiv$ 130 km s$^{-1}$), covering 4250--7035 \AA. To minimize the
126: effects of atmospheric dispersion, we observed at parallactic
127: angle. Our exposure times ranged from 420 s to 1200 s, with an average
128: of 660 s (a total of 8.81 hours on the source).
129:
130: Each night we observed the flux standard HILT 600
131: with the same instrumental setup as A0620-00. On all nights the seeing
132: was generally comparable to our slit width, but occasionally spiked
133: as high as 1.8 arcsec due to wind, and slit losses prevented precise
134: flux calibration. On 2006 Dec 15 we also
135: observed two K3/K4 stars, HD 18298 and HD 7142, as velocity standards,
136: again with the same optical setup. As the secondary is constrained to
137: be later than K3V \cite{FR07}, we expect more accurate
138: results for HD 7142, which is listed as K3/K4III in the SIMBAD
139: database (HD 18298 is listed as K3IIICN). For each pointing, we obtained
140: comparison HeNeAr arc lamp spectra after every 3--7 spectra, depending
141: on the current exposure time. We list the observations in Table 1.
142:
143: \setcounter{table}{0}
144: \begin{table}
145: \caption{Observation log}
146: \label{tab1}
147: \begin{tabular}{@{}cccc}
148: \hline
149: Source & Date & \# of Spectra & T$_{\rm exp}$ (s)\\
150: \hline
151: A0620-00 & 2006 Dec 14 & 11 & 873 \\
152: A0620-00 & 2006 Dec 15 & 23 & 698 \\
153: A0620-00 & 2006 Dec 16 & 14 & 433 \\
154: HILT 600 & 2006 Dec 14 & 1 & 20 \\
155: HILT 600 & 2006 Dec 15 & 3 & 20 \\
156: HILT 600 & 2006 Dec 16 & 3 & 20 \\
157: HD 18298 & 2006 Dec 15 & 3 & 5 \\
158: HD 7142 & 2006 Dec 15 & 5 & 5 \\
159: \hline
160: \end{tabular}
161:
162: \medskip T$_{\rm exp}$ is the average exposure time for the source.
163: \end{table}
164: We used standard IRAF routines for basic data reduction
165: (zero-subtraction, flat-fielding, and spectral extraction). We
166: extracted our spectra in multispec format, attempting \textit{apall}'s
167: optimal extraction using nominal LDSS3
168: gain, readout noise, and full-well values. The routine also performs
169: standard extraction and generates error bars. A CCD defect running
170: across our spectra prevented reliable fits to the spatial profile for
171: optimal extraction, so we used the normally extracted spectra
172: instead. This choice did not significantly degrade our S/N, and should
173: have a negligible effect on our presented results. After wavelength
174: calibration, we passed the spectra and their errors to the software
175: package \small{MOLLY}\normalsize~for cleaning and analysis.
176:
177: \section{ANALYSIS}
178: \subsection{The spectrum of A0620-00}
179:
180: \setcounter{figure}{0}
181: \begin{figure}
182: \includegraphics[width=84 mm]{f1.eps}
183: \caption{The average normalized spectrum of A0620-00. The strongest
184: features are, in order of increasing wavelength, H$\gamma,$
185: H$\beta,$ He\,{\sc i} 5875, H$\alpha,$ He\,{\sc ii} 6678, and
186: telluric absorption lines. We also show the scaled spectrum of HD
187: 7142 for reference.}
188: \label{fig1}
189: \end{figure}
190:
191: In Figure 1 we present the average normalized spectrum of A0620-00. The
192: spectrum shows a number of strong features (originating from the disc)
193: and many relatively weak K-star absorption lines. We overplot the K-dwarf
194: template HD 7142 for reference. We have
195: extremely high S/N H$\alpha$ and H$\beta$ lines, both observed by
196: MRW94, as well as lower S/N H$\gamma$ and He\,{\sc i} 6678\AA. The
197: non-detections of He\,{\sc ii} lines indicates the scarcity of
198: ionizing radiation. Our H$\alpha$ line is stronger relative to the
199: continuum than in 1994 by approximately 50 per cent. This relative
200: brightening is not surprising given the decreased fraction of light
201: contributed by the secondary. We show a close-up of H$\alpha$ in panel
202: a of Figure 2. The profile has two strong symmetric
203: peaks. It is also interesting that our
204: H$\beta$ line also shows two strong peaks (as opposed to a
205: single-peaked line seen by MRW94).
206:
207: The other feature of note is located on top of the interstellar sodium
208: doublet near 5890 \AA. We show a close-up of this line in
209: panel b of Figure 2. The line suffers significant extinction by
210: interstellar sodium, so it is not possible to identify its peak
211: unequivocally, but it seems to be He\,{\sc i} 5875. It
212: appears to have some structure, and trailed spectra
213: suggest a double-peaked profile, but our attempts to correct for
214: interstellar absorption and create a Doppler map ($\S$ 4) were
215: unsuccessful.
216:
217: \setcounter{figure}{1}
218: \begin{figure}
219: \includegraphics[width=84 mm]{f2.eps}
220: \caption{Close-ups of a) the H$\alpha$ line and b) the feature near
221: 5890\AA. Intensities are as in Figure 1.}
222: \label{fig2}
223: \end{figure}
224:
225: \subsection{The radial velocity of the secondary star}
226:
227: Following MRW94, we first measure the radial velocity of V616 Mon in
228: order to estimate the mass function. Masking out emission lines,
229: telluric absorption, and
230: interstellar sodium lines in our velocity standards, and normalizing,
231: we cross-correlated our spectra against these K-dwarf templates. The
232: results of the following sections are presented in Table 2.
233:
234: \setcounter{figure}{2}
235: \begin{figure*}
236: \includegraphics[width=150 mm]{f3.eps}
237: \caption{Radial velocity of V616 Mon. Two cycles are shown
238: for clarity. Filled circles correspond to the template HD 18298, and
239: open circles to HD 7142. Note from Table 2 that the main difference
240: is in $\gamma.$}
241: \label{fig3}
242: \end{figure*}
243:
244: As discussed in MRW94, this process requires several adjustments for
245: A0620-00: rotational broadening and orbital smearing. The following
246: analysis was completed for each template. We performed a
247: preliminary broadening of the template with MRW94's value of $v\sin
248: i=83$ km s$^{-1}$. Then, after cross-correlating, we shifted our spectra into
249: the rest frame of the template, which is essentially the rest frame of
250: the secondary. We made 48 copies of the template and smeared each
251: according to the formula $s=2\pi V_{i} T_{i}/P,$ where $P$ is the orbital
252: period and $V_{i}$ and $T_{i}$ are the radial velocity and exposure
253: time for the $i$th A0620-00 spectrum. With exposure times of up to 1200
254: s, it is necessary to correct for smearing because $s$ is comparable
255: to $v\sin i.$ We averaged the smeared templates, rotationally broadened the
256: result with a value between 30 and 135 km s$^{-1}$, and optimally
257: subtracted the final template from the average object spectrum. We
258: assume a limb-darkening coefficient $\varepsilon$ of 0.65 (Wade \&
259: Rucinski 1985, and references therein), but perform these calculations
260: for $\varepsilon=0.45-0.85$ to evaluate our systematics. The
261: optimal subtraction routine returns the fraction $f$ of light
262: contributed by the secondary and a $\chi^{2}$ value. We fit
263: fourth-order polynomials to $\chi^{2}$ to identify the appropriate $f$
264: and $v\sin i.$ Results are included in Table 2.
265:
266: The measured rotational broadening of V616 Mon ($80\pm2$ for HD 18298
267: and $83\pm2$ for HD 7142) is consistent with
268: MRW94, but $f$ is not, even though $f$ and $v\sin i$ were strongly
269: correlated (i.e. we were able to identify them with the same
270: $\chi^{2}$ minimum). MRW94 found that the secondary contributes $\sim
271: 94$ per cent of the light near H$\alpha$ and $\sim 85$ per cent of the
272: light near H$\beta.$ We observed both wavelengths simultaneously, and
273: find that the secondary contributes only 44 per cent of the light in
274: $B$ and $V$ (near 5500 \AA). This conclusion is effectively
275: independent of the limb-darkening coefficient. We will discuss
276: systematic uncertainties on $f$ in $\S3.4$.
277:
278: Finally, we performed another cross-correlation of our spectra against
279: the template, this time with the appropriate value of $v\sin i.$ The
280: results are shown in Figure 3. Although we took arc exposures
281: frequently to minimize the effects of flexure, we found arc scales to drift
282: by $\sim$0.5 \AA~over the course of a few hours. As our dispersion
283: (0.69 \AA~pixel$^{-1}$ ) corresponds to 37.9 km s$^{-1}$ pixel$^{-1}$, it was
284: necessary to compensate for arc drift. We did this by
285: cross-correlating the telluric lines near 6900 \AA~and shifting out
286: the resulting velocities (generally around 10 km s$^{-1}$). In this way we
287: guaranteed a common (heliocentric) rest frame for all 48 spectra.
288:
289: We fit a function of the form $V=K_{2}\sin(2\pi
290: (t-t_{0})/P)+\gamma$ for each template. $\gamma$ represents the
291: systemic velocity in the template frame. Although we have very high
292: S/N, the baseline ($\sim$7 orbits) was insufficient
293: to make a reliable independent determination of the orbital period, so
294: we fixed $P$ according to the ephemeris of MR86. To get an
295: accurate measure of our statistical uncertainties, we performed 10,000
296: Monte Carlo simulations for each fit, assuming each radial velocity to be
297: distributed normally around the measured value, increasing the errors
298: from cross-correlation by a factor of three. For our final values,
299: we averaged the results from both templates and took the standard
300: deviation of the combined distribution as the uncertainty. The
301: resulting uncertainty is not purely statistical, as the use
302: of multiple templates includes some systematic errors
303: (e.g. template mismatch). These results can be found in Table 2.
304:
305: \setcounter{table}{1}
306: \begin{table*}
307: \begin{minipage}{150mm}
308: \caption{Fits to rotational broadening and radial velocities.}
309: \label{tab2}
310: \begin{tabular}{@{}cccccccccc}
311: \hline
312: Template & $\varepsilon$ & $v\sin i$ & $f$ & $K_{2}$ (km s$^{-1}$)
313: & t$_{0}-2454084$ & $\gamma$ (km s$^{-1}$) & $q$ & $f(M_{1})$
314: (M$_{\sun}$) & $\chi^{2}$\\
315: & & (km s$^{-1}$) $\pm2$ & $\pm0.001$ & $\pm0.4$ & $\pm4$E-5 &
316: $\pm8$ & $\pm0.004$ & $\pm0.03$ &\\
317: \hline
318: HD 18298 & 0.45 & 77 & 0.441 & 435.8 &
319: 0.69487 & 11 & 0.051& 3.06 & 1.06\\
320: ~& 0.55 & 78 & 0.441 & 435.8 &
321: 0.69487 & 11 & 0.052 & 3.07 & 1.06\\
322: ~& 0.65 & 80 & 0.440 & 435.8 &
323: 0.69487 & 11 & 0.057 & 3.09 & 1.05\\
324: ~ & 0.75 & 80 & 0.441 & 435.8 &
325: 0.69487 & 11 & 0.056 & 3.09 & 1.06\\
326: ~ & 0.85 & 81 & 0.441 & 435.8 &
327: 0.69487 & 11 & 0.058 & 3.10 & 1.06\\
328: HD 7142 & 0.45 & 80 & 0.433 & 435.0 &
329: 0.69484 & -9 & 0.056 & 3.07 &1.03 \\
330: ~ & 0.55 & 80 & 0.433 & 435.0 &
331: 0.69484 & -9 & 0.057 & 3.08 & 1.03 \\
332: ~ & 0.65 & 83 & 0.449 & 434.9 &
333: 0.69484 & -9 & 0.063 & 3.11 & 1.02 \\
334: ~ & 0.75 & 83 & 0.433 & 435.0 &
335: 0.69484 & -9 & 0.061 & 3.10 & 1.03 \\
336: ~ & 0.85 & 84 & 0.433 & 435.0 &
337: 0.69484 & -9 & 0.064 & 3.12 & 1.03\\
338: \hline
339: \end{tabular}
340: \end{minipage}
341: \end{table*}
342:
343: All fits are comparable in quality and have
344: excellent $\chi^{2}$. We find consistent values for $K_{2}$ and
345: excellent agreement in $t_{0}$ -- we
346: observed inferior conjunction on the second night -- and our
347: measurements are independent of limb-darkening ($\varepsilon$=0.65 is
348: marginally preferred). We will consider the systemic velocities only
349: briefly. The uncertainties in $\gamma$ are dominated by the systematic
350: uncertainty (one fifth of a pixel, or $\sim$7.6 km s$^{-1}$). We were unable
351: to find a cataloged radial velocity for HD 18298, but from
352: \citet{Malaroda01}, the radial velocity of HD 7142 is 32.8 km
353: s$^{-1}$. Given that MRW94 report a systemic velocity of 22 km
354: s$^{-1}$, we consider these fits to be accurate; the choice of
355: template does not significantly affect $K_{2},~t_{0},$ or $v\sin i.$
356: In summary, we adopt $v\sin i=82\pm2$ and $K_{2}=435.4\pm0.5$ km
357: s$^{-1}.$ In addition, we find that inferior conjunction of the
358: mass donor star, which defines the zero point $t_{0}$ of our
359: ephemeris, occurs at HJD (UTC) 2454084.69485$\pm$0.00005. This
360: latest $t_{0}$ is consistent with the ephemeris of MR86 (within 0.5
361: $\sigma).$
362:
363: \subsection{The mass ratio}
364:
365: To measure the mass ratio, we use
366: Paczynski's (1971) formula relating the rotational broadening and
367: $K_{2}$ to $q$ for Roche lobe-filling stars,
368: \begin{equation}
369: \frac{v\sin i}{K_{2}}=0.462[(1+q)^{2}q]^{1/3}.
370: \end{equation}
371: This gives
372: $q=0.060\pm 0.004,$ where the uncertainty includes variation between
373: templates. This value is consistent with MRW94's measurement
374: (by the same method) of $0.064 \pm 0.01,$ and with their value of
375: $q=0.067\pm 0.01,$ obtained by calculating models on a grid over the
376: Roche lobe, including gravity darkening and quadratic limb
377: darkening. Although it might be possible with our improved resolution
378: to distinguish between the Paczynski approximation and the grid
379: models, since the grid model correction was far from
380: significant at their $1\sigma$ level, we opt to take our results as
381: accurate. This choice is validated by the weak (at best) dependence of
382: our measurements on limb darkening, and we report uncertainties large
383: enough to account for any systematics in $\varepsilon.$
384:
385: Given the orbital period, $K_{2},$ and $q$, the minimum masses for
386: both objects are:
387: \begin{eqnarray*}
388: f(M_{1})&=& 3.10\pm0.04~\rm{M}_{\sun}\\
389: f(M_{2})&=& 0.19\pm0.02~\rm{M}_{\sun},
390: \end{eqnarray*}
391: where the true mass goes like M$_{\rm min}/\sin^{3}i.$ As noted in many
392: papers (MRW94; Shahbaz et al. 1994), the mass of a maximally
393: rotating neutron star with the stiffest equation of state is 3.2
394: M$_{\sun};$ if causality is the only constraint, the absolute upper
395: limit is 3.76 M$_{\sun}$ \citep{FI87}.
396: Using the constraints $39\degr\leq i \leq 75\degr$ \citep{Gelino01}, we find
397: 3.4 M$_{\sun}\leq$ M$_{1}\leq 12.6$ M$_{\sun},$ with the most likely value
398: at 11.1 M$_{\sun}.$ We thus improve the precision of the mass function,
399: but it remains a possibility that A0620-00 is a stiff, massive neutron star.
400:
401: \subsection{Ellipsoidal variations and the variability of the disc}
402:
403: Part of the difficulty in measuring the parameters of A0620-00, and a
404: possible source of systematic error, is the tidal distortion of the
405: secondary. We assume, for example, in our radial velocity
406: measurements, that the secondary's centre of light coincides with its
407: centre of mass. For a star filling its Roche lobe with some
408: gravitational and limb darkening, this assumption is not likely to be
409: valid. Following \citet{Sterne41}, we checked for this distortion in
410: our data by fitting an extra $-K_{\rm ell}\sin(4\pi\phi)$ term to our
411: radial velocity curve. Orbital elements from \citet{Gelino01} lead
412: to a predicted $K_{\rm ell}=3.0$ km s$^{-1}$, but we found $K_{\rm
413: ell}\sim 0.7$ km s$^{-1}$ (less than the error bars on each radial
414: velocity), and an insignificant improvement to $\chi^{2}.$ It is
415: possible that our data have insufficient S/N to make this
416: measurement. This particular effect is small, but we must be careful
417: not to dismiss the systematic uncertainties from ellipsoidal effects.
418:
419: %READ STERN 1941
420: %based on values from gelino, e.g. a=4.47 and r2=0.8 i use sterne to
421: %calculate dv ~ 3sin2L, where v ~ k2 cosL. transform v to k2 sin x
422: %gives v ~ k2 sin (L+90). then L+pi/2 = 2pi phi. 2L = 4pi phi - pi. so
423: %2L= 4 pi phi - pi. sin 2L = sin(4 pi phi - pi) = -sin(4 pi phi)
424:
425: Ellipsoidal variations,
426: particularly in the lightcurve, receive a great deal of attention
427: because of the constraints they place on the orbital parameters,
428: especially the inclination. In practice, one usually attributes all variation
429: to the secondary (assuming the disc to be constant). However, the fact
430: that the fraction of light contributed by the disc changes by a
431: factor of nine over ten years casts significant doubt on that
432: particular assumption. In turn, this evolution must be accounted for
433: when interpreting lightcurves spanning a long period of time.
434:
435: Given that we have good spectral resolution and relatively good phase
436: coverage, we have attempted to characterize the ellipsoidal
437: variations of the source during our observations. While it is most
438: common to address variations in the lightcurve, effects of tidal distortion
439: should also be apparent in the secondary light fraction
440: and the line equivalent widths. In the rest of this section we
441: quantify these variations.
442:
443: \setcounter{figure}{3}
444: \begin{figure}
445: \includegraphics[width=84 mm]{f4.eps}
446: \caption{a) Equivalent width of the H$\alpha$ line. b)
447: Secondary light fraction as measured from 5000--5800 \AA~and
448: 6000--6500 \AA~for the
449: template HD 7142. Differences from HD 18298 are less than 1$\sigma.$
450: These plots indicate a flickering source of light. The solid,
451: dotted, and dashed lines are the predicted light fraction for a
452: constant disc and ellipsoidal modulations, for inclinations of
453: 40$\degr$, 60$\degr$, and 80$\degr$, respectively. See Equation 2.}
454: \label{fig4}
455: \end{figure}
456: If the disc is a constant diluting source of light, then continuum
457: variations, and therefore variations in the equivalent widths of
458: lines, maybe be attributed to the tidal distortion of the
459: secondary. In Figure 4a we show the equivalent width of the H$\alpha$
460: line as a function of orbital phase. The asymmetric ellipsoidal
461: variations observed by \citet{Gelino01} are obvious, along with
462: noticeable scatter. Given a S/N of $\sim 95$ at H$\alpha$, it seems
463: reasonable to interpret this as flickering, albeit undersampled,
464: rather than noise. We will discuss the equivalent widths in more
465: detail shortly.
466:
467: Figure 4b shows the secondary light fraction
468: $f$ as a function of orbital phase. We can evaluate the variability of
469: the disc with a back-of-the-envelope calculation of $f(\phi).$ We
470: assume the secondary continues to exhibit ellipsoidal variations of
471: 0.077 mag (MRW94 using $i=40\degr$), or flux variations of 7 per cent, and
472: that the disc is constant in time. Since the disc contributes
473: 56 per cent of the light, it should be roughly 22 per cent brighter than the
474: secondary. Then in units of the mean secondary flux,
475: $F_{2}=1-0.07\cos(4\pi\phi)$ and $F_{1}=1.22,$ and
476: \begin{equation}
477: f=\frac{1-0.07\cos(4\pi\phi)}{2.22-0.07\cos(4\pi\phi)}.
478: \end{equation} The inclination-dependent amplitude of
479: ellipsoidal variations is tabulated in MRW94. Equation 2 is plotted
480: along with $f$ for inclinations of 40, 60, and 80$\degr.$ We have
481: matched the mean relatively well, and to some extent the functional
482: form, but it is obvious that our assumptions do not hold. First, the
483: disc appears anomalously bright between $\phi=$0.7--0.9. As
484: this phase interval was observed on the second night only, we cannot
485: speculate if the dip in $f$ indicates a flare or a long-lived bright
486: region of the disc, like a warp, visible only at this phase.
487:
488: The second and more troubling discrepancy is the large amplitude of
489: variation of $f.$ As is apparent from Figure 4b, excluding
490: $\phi=$0.7--0.9, higher inclinations are preferred, even those which
491: are ruled out by the lack of eclipses in this source. Accepting
492: momentarily the inclination determined by
493: \citet{Gelino01}, this figure illustrates very clearly the significance of the
494: assumption that the disc is a constant source of light; by requiring a
495: constant disc, we could overestimate the inclination by several tens
496: of degrees. Now it is reasonable to assume that the secondary has not evolved
497: substantially in the last twenty years, so we must conclude that the
498: discrepancy between the predicted and observed light fractions is
499: related to activity in the disc. For example, a component of disc
500: light modulating at the orbital period could reproduce the effect easily.
501:
502: Unfortunately, there is no a priori way to determine in advance the
503: viability of an inclination measurement. SMARTS data from the last ten
504: years show that A0620-00 goes through periods of quiescence, where ellipsoidal
505: modulations are observed, and periods of erratic variability which
506: gradually swamps the smooth component (Cantrell \& Bailyn 2007,
507: private communication). In these active periods, the source is
508: extremely variable on timescales from minutes (our observations) to
509: years (SMARTS), and the variability is highly
510: phase-dependent, so that ellipsoidal modulations cannot be reliably
511: measured; the current active state has persisted since December of 2003.
512: Associating this variability with the disc, we warn
513: against measurements of $i$ during such periods. We can expect,
514: furthermore, a strong correlation between flickering and the light
515: fraction. There is no reason $f$ cannot be small and constant
516: simultaneously, but an increase in light from the disc can only mean an
517: increase in accretion activity, for which flickering is highly
518: probable.
519:
520: This is admittedly a grim assessment of the situation, but lightcurve
521: estimates of the inclination made during periods of variability
522: or non-negligible contamination by disc light are unreliable. The
523: variation in $f$ itself is a final source of systematic
524: uncertainty. We have a very precise measurement of the dilution fraction
525: for the average spectrum, but that precision is only meaningful if $f$ is not
526: variable. Instead, we take the standard deviation in $f(\phi)$ as our
527: uncertainty: $f=44\pm7$ per cent, and summarize the discussion above
528: by reminding the reader that quiescence is not inactivity, and ought
529: not be used to justify invalid assumptions.
530:
531: Consider now the equivalent width. Since the continuum varies
532: slowly over our lines, it is trivial to show that
533: the line equivalent width is given by \begin{equation}
534: EW~(\textrm{\AA})\simeq\frac{EW_{\rm disc}}{1+f}, \end{equation}
535: where $EW_{\rm disc}$ is the ratio of flux integrated
536: over the line to the disc
537: continuum, and $f$ is the secondary light fraction. If the disc varies
538: uniformly or not at all, $EW_{\rm disc}$ should be a constant for any
539: given line. We can then attribute variations in the equivalent
540: width to variations in $f,$ which would be ellipsoidal in
541: nature. However, measurements of the equivalent width with the
542: secondary subtracted should reveal the reliability of this shaky assumption.
543:
544: The equivalent width of our unsubtracted H$\alpha$ line (open circles)
545: is shown in Figure 5; it
546: should be compared to the same plot from MRW94. To facilitate this
547: comparison, we have overlaid their best-fitting line. We find good
548: agreement with their phasing and amplitude, but an increase in average
549: equivalent width of approximately 13 \AA. As noted earlier, the
550: H-alpha line is brighter relative to the continuum than it was in 1994, so
551: this increase is reasonable. Also shown in Figure 5 (filled
552: circles) is the equivalent width of the same line after subtraction of
553: the secondary. To achieve this result, which indicates the magnitude
554: of fluctuations in the disc, we normalized, broadened, and
555: smeared the template, subtracted $f(\phi)$ times the template from
556: each spectrum, and set the continuum to one. The dip between phases
557: 0.7 and 0.9 corresponds to the possible flare seen in the light fraction.
558:
559: We also note that the subtracted modulations are in phase with
560: the unsubtracted equivalent widths. As the secondary is expected to
561: contribute more light at longer wavelengths, the phasing may be an
562: artefact of improper secondary subtraction, because the strongest
563: contributions to $f$ come from 5000--5800 \AA. However, this effect should be
564: small because absorption lines up to 6500 \AA~were used in the
565: measurement. Even if this is the case, the scatter in $EW_{\rm disc}$
566: cannot be explained by the combined noise in $f$ and $EW.$ We suggest
567: that the most probable extra source of scatter is physical variability
568: of the disc, i.e. flickering. In this case, the fact that the
569: subtracted equivalent width is not constant suggests that the disc
570: fluctuations cannot be spatially uniform.
571:
572: \setcounter{figure}{4}
573: \begin{figure}
574: \includegraphics[width=84 mm]{f5.eps}
575: \caption{Measured equivalent widths for the H$\alpha$ line from
576: A0620-00. Open circles correspond to unsubtracted spectra; filled
577: circles were calculated after subtracting the template HD
578: 7142. Again, differences between templates were insignificant, and
579: the errors are smaller than the circles because of the extreme
580: signal-to-noise. We have plotted MRW94's best fit for comparison.}
581: \label{fig5}
582: \end{figure}
583: Our measurements of flickering are confirmed by nearly
584: simultaneous observations with all four IRAC bands on Spitzer and the
585: 1.2m FLWO telescope (McClintock 2007, private communication). These
586: observations, taken in the last week of November 2006, show strong erratic
587: variability which is well-correlated between telescopes. Neither
588: ellipsoidal variations nor the orbital period is obvious through
589: these fluctuations. The conclusion is clear: the
590: assumption that the disc is constant is not valid, though it may be a
591: good approximation if the disc contribution to the continuum is
592: negligible (for example, when \citet{Gelino01} found $f\ga 97$ per cent
593: and $i=41\pm 3\degr$). To investigate further the spatial
594: and temporal intensity of the disc, we present the results of Doppler
595: tomography and modulation tomography in $\S$ 4.
596:
597: \section{Doppler Imaging}
598: In this section we discuss the results of Doppler tomography of the
599: emission lines from A0620-00. To create each image, we subtracted the
600: scaled HD 7142 spectrum, performed a linear
601: fit to the continuum around each line, normalized, and set the
602: surrounding continuum to zero (required by
603: \small{DOPPLER}\normalsize). MRW94 express some uncertainty as to the
604: propriety of interpolating over the H$\alpha$ line in the template
605: spectrum. Doing so removes an image of the donor star in the map,
606: and it is not clear which is the appropriate choice. However, we
607: found that Doppler maps including the light from the secondary do not
608: show donor emission, so we chose to interpolate over the line. Then we binned
609: the spectra to a uniform velocity scale and passed them to the
610: \small{DOPPLER}\normalsize~routine, which computes the maximum entropy
611: image that reproduces the observed line profiles during the course of
612: an orbital period. For an excellent summary of the method, see
613: \citet{Marsh01}.
614:
615: \subsection{The maps}
616: In Figures 6--9 we present Doppler tomograms of the strong emission
617: lines in A0620-00. In each figure, the left column is, from top to
618: bottom: the observed trailed spectrum, the Doppler map, and
619: the fitted data. The right column is the observed data minus a
620: simulated symmetric part, the asymmetric part of the Doppler map, and
621: its fitted data. In the maps, we plot the Roche lobe of the secondary,
622: the ballistic trajectory of the gas stream (lower line), and the
623: Keplerian velocity of the disc along the stream (upper line).
624:
625: In Figure 6 we show the maps of H$\alpha$. The most obvious feature is
626: the bright spot, which corresponds to the gas stream impact point. As
627: found by MRW94 and S04, in our maps the gas
628: stream trajectory and the Keplerian disc velocity along t he stream
629: (the two lines plotted in the map panels) bracket the bright spot. We
630: can interpret this as post-shock emission, originating
631: somewhat inside the outer edge of the disc \citep{Marsh90}. Using the
632: ballistic trajectory, we locate the spot at
633: $r=0.6\pm0.05R_{\rm L1}.$ If the disc velocities are Keplerian and the
634: spot moves with the disc, we find the outer edge near
635: $r=0.45\pm0.05R_{\rm L1}.$ This is some indication of our systematic
636: uncertainties, but we shall suggest shortly that the larger disc is
637: more likely.
638:
639: The other features of note are the two
640: crescents at $\sim 7$ o'clock and $\sim 2$ o'clock. We shall address
641: their origin shortly. The fitted data are rather
642: messy, mainly due to the presence of those features and the strong
643: flickering. Because the crescents and the bright
644: spot so thoroughly dominate the image, the symmetric part is too
645: bright, and we oversubtract to make the asymmetric part. Still, we do
646: detect some emission from the bright spot. The trail itself is not
647: particularly sinusoidal: it is more of a zigzag than an S-wave. The
648: discrepancy is best seen in the simulated asymmetric trail (bottom
649: right, Figure 6), but is apparent in the top panels as well. In the
650: data panel (top left), the trail is diagonal between $\phi=-3$ and
651: $\phi=-2.5$, but nearly horizontal at $\phi=-2$. It is difficult to
652: interpret this as the S-wave of a circular orbit. If the orbit is not
653: circular, it may be a combination of the ballistic trajectory of the
654: stream and the motion of the disc, or it may be an elliptical orbit.
655:
656: In Figure 7 we show maps of
657: H$\beta$. Again, we see the crescent features; their shape is more
658: apparent despite the lower S/N at H$\beta.$ For the same reason, we
659: have less trouble computing the symmetric part, and our asymmetric
660: part shows some emission along the stream. While the corresponding
661: trail does look nicer than its H$\alpha$ counterpart, the two seem to
662: be consistent. In Figure 8 we show maps of H$\gamma$, and in Figure 9 we
663: show maps of He\,{\sc i} 6678. These two maps, at much
664: lower S/N, reveal bright spot emission with some contribution from the
665: stream itself.
666:
667: \setcounter{figure}{5}
668: \begin{figure*}
669: \includegraphics[width=150 mm]{f6.eps}
670: \caption{H$\alpha$ Doppler maps. Notice in the middle left panel that
671: the bright spot is located between the gas stream trajectory and the
672: Keplerian velocity of the disc along the stream, and the brighter
673: crescent-shaped features.}
674: \label{fig6}
675: \end{figure*}
676: \setcounter{figure}{6}
677: \begin{figure*}
678: \includegraphics[width=150 mm]{f7.eps}
679: \caption{H$\beta$ Doppler maps. The bright spot is again between the
680: gas stream and the Keplerian disc velocity, and the crescents are
681: obvious. The gas stream itself is clearly visible in the asymmetric
682: part (middle right).}
683: \label{fig7}
684: \end{figure*}
685: \setcounter{figure}{7}
686: \begin{figure*}
687: \includegraphics[width=150 mm]{f8.eps}
688: \caption{H$\gamma$ Doppler maps. Only the bright spot is prevalent.}
689: \label{fig8}
690: \end{figure*}
691: \setcounter{figure}{8}
692: \begin{figure*}
693: \includegraphics[width=150 mm]{f9.eps}
694: \caption{He\,{\sc I} 6678 Doppler maps. Again, only the bright spot is
695: apparent.}
696: \label{fig9}
697: \end{figure*}
698:
699: \subsection{The eccentric disc of A0620-00}
700: In the previous sections, we have presented evidence for a bright
701: flickering disc which contributes more than half of the light from the
702: source in the optical. It is already clear that the ``no intrinsic
703: variability'' assumption of Doppler tomography is not satisfied by
704: A0620-00. We should therefore exercise caution in interpreting our
705: Doppler maps. Tomography is a robust technique, but the physical
706: relevance of its results depends on the extent to which its
707: assumptions are violated. For example, we assume that all velocity
708: vectors corotate with the binary. However, if the disc is large
709: enough, it may reach an orbital resonance with the secondary and
710: become eccentric by tidal distortions. It will then precess, even in
711: the corotating frame. For the mass ratio of A0620-00, the dominant
712: resonance is 3:1 \citep{WK91}. Recent SPH simulations by
713: \citet{Foulkes04} show that the manifestations of eccentric discs
714: are bright emission between the gas stream and the Keplerian disc
715: velocity along the stream, non-sinusoidal S-waves, crescent-shaped
716: features in Doppler maps, and the shifting of the map centre of
717: symmetry away from (0,-$K_{1}$).
718:
719: S04 observed the first three features and invoked an eccentric disc,
720: but they also point out one possible objection: MRW94 found a clean circular
721: disc. However, we have exposure times comparable to MRW94 with a
722: larger telescope; our higher signal to noise may have enabled us to
723: detect this phenomenon. The other explanation is that
724: the crescent was absent in 1994, and the source has changed. It is
725: thought that A0620-00 has an outburst recurrence time of 58 years, and more
726: than half of that time has passed since 1975, so the system should be
727: gearing up for a new outburst. In the disc instability model of
728: outbursts, the disc steadily recharges between outbursts, growing in
729: size and density until it becomes unstable. In the context of this
730: model, it is quite possible that since 1994, we
731: have actually watched the disc expand towards the 3:1 resonance
732: (r=0.66$R_{\rm L1}$) and become distorted. If this resonance is inside
733: the disc, then the distortion time-scale is $q^{-2}P_{\rm orb},$ or
734: about ninety days \citep{FKR02}. If not, the process is slower,
735: but a change over twelve years for A0620-00 seems reasonable.
736:
737: Now if the disc is actually circular, the Doppler map should be
738: radially symmetric about the point (0,-$K_{1}$), and by locating the
739: centre of the disc, we can identify the radial velocity of the black
740: hole \citep{Steeghs02}. Otherwise, the eccentricity of the
741: disc should be evident in a discrepancy between the observed and
742: predicted locations of this point. We therefore
743: implemented a search for the disc centre of symmetry, starting at the
744: predicted point and extending $\pm$200 km s$^{-1}$ in $V_{\rm x}$ and
745: $V_{\rm y},$ subtracting the symmetric part, squaring, and computing
746: the mean and standard deviation. The point with the lowest mean
747: corresponds to the centre of symmetry. We iterated our search,
748: updating the centre and improving the resolution until we found the
749: minimum.
750:
751: Since our disc is very structured, it was difficult to find a region
752: of the image unaffected by bright spots, so we performed our search on
753: a smoothed version of the map. We then ran 10,000 Monte Carlo
754: simulations to find the centre of symmetry, using the standard
755: deviation of the residuals as the uncertainty to be sampled. Combining
756: the results for our H$\alpha$ and H$\beta$ maps, we find the center of
757: symmetry at (80,-220)$\pm$(40,20) km s$^{-1}$, well outside the
758: uncertainty in $K_{1}.$ Since we cannot explain a factor of nine of implied
759: increase in $q,$ we take this result, coupled with the bright spot
760: locationa and the crescent features in the maps, as evidence for the
761: non-zero eccentricity of the disc.
762:
763: Finally, we recognize that as the disc precesses, its apparent centre
764: of symmetry should move, as this point roughly corresponds to the mean
765: radial velocity of the disc. We made Doppler images of the first night
766: and the two halves of the second night, and performed the above search
767: for the centre of symmetry for each. However, the low phase coverage
768: of these maps results in large uncertainties, and we are unable to
769: determine a trend. The feasibility of such a measurement also depends on
770: the portion of the disc participating in eccentricity and precession,
771: and the exact precession period, which is generally estimated at $\sim
772: 30$ orbits. With improved spectral resolution, a sequence of 3--4 full
773: nights (for sufficient phase coverage per night) sampling the
774: precession period should allow the motion of the center of symmetry of
775: the disc to be resolved.
776:
777: \subsection{Modulation tomography}
778: It is also possible to relax the assumption of Doppler tomography that
779: the source flux is constant throughout the orbit \citep{Steeghs03}. The
780: new technique of modulation tomography allows not only the imaging of
781: average line emission from an accretion disc, but also maps harmonic
782: variations on the orbital period. The technique is robust and
783: flexible, but requires somewhat better S/N than standard tomographic
784: imaging. Therefore we only consider our H$\alpha$ profile here. The
785: process of creating modulation maps is quite similar to standard
786: tomography, and is described in \citet{Steeghs03}. The map is presented
787: in Figure 10.
788:
789: \setcounter{figure}{9}
790: \begin{figure*}
791: %\includegraphics[angle=270,width=150 mm]{f10.eps}
792: \includegraphics[width=150 mm]{f10.eps}
793: \caption{H$\alpha$ modulation map. The observed data (top left) are
794: well-reproduced by the fitted data (top right). The middle left
795: panel shows the constant part of the disc, and the middle right
796: panel shows the amplitude of modulation. The bottom row, left and
797: right, are the cosine and sine components of variation,
798: respectively. The crescent regions are obviously modulated at the
799: orbital period.}
800: \label{fig10}
801: \end{figure*}
802: It is clear from the fitted data (top right), that modulation
803: tomography does a much better job reproducing the trail than standard
804: Doppler mapping. Whereas standard mapping could only reach
805: $\chi^{2}=35,$ we were able to attain $\chi^{2}=9.75$ with modulation
806: tomography. The poor $\chi^{2}$ is due to flickering. The constant
807: portion of the map (middle row, left
808: panel) looks quite similar to the maps presented in previous sections,
809: though the bright spot is significantly smaller. We still place it off
810: the gas stream trajectory, but it is much less diffuse, and is
811: located, as before, near $r=0.6R_{\rm L1}.$ It is clear from this image
812: that the inner edge of the disc in velocity space, and thus the outer
813: edge of the disc in physical space, extends as far as 0.7$R_{\rm L1}.$ The
814: crescents are not nearly as prevalent in the constant part, but are
815: apparent in the modulation maps. The crescent placed at 7 o'clock in
816: the standard map appears in the $\cos\phi$ map (bottom left) and the
817: crescent placed at 2 o'clock appears in the $\sin\phi$ map (bottom
818: right). The full modulating amplitude is shown in the middle row,
819: right panel.
820:
821: For an erratic source like A0620-00, these maps must be interpreted
822: carefully. Modulation tomography, as mentioned, maps only harmonic
823: variations, and the demonstrated flickering is hardly
824: harmonic. However, we have shown the eccentric disc to be a viable
825: explanation for the observed phenomena, and the regions responsible
826: for the crescent emission are not constant by any means. It is clear
827: from the constant image
828: that the inner edge of the disc in velocity space, and thus the outer
829: edge of the disc in physical space, extends as far as 0.7$R_{\rm L1},$
830: lending credence to the 3:1 resonance argument. For typical
831: disc precession, superhumps are observed with periods $3-5$ per cent off the
832: orbital period \citep{WK91}. If indeed this modulating
833: emission is caused by viscous dissipation at the outer edge of an
834: eccentric precessing disc, there should be some power at $P_{\rm orb},$
835: though a map of variation at some superhump period $P_{\rm sh}$ would
836: reveal larger amplitudes. As yet, such a map is not
837: possible. None the less, modulation tomography has provided yet another
838: piece of evidence for our eccentric precessing disc model, and the
839: fact that there is a component of disc light modulating at the orbital
840: period could explain the observed light fraction from the secondary.
841:
842: \section{Discussion}
843:
844: We have presented spectroscopic analysis of the black hole binary
845: A0620-00. We measure an absorption-line radial velocity
846: $K_{2}=435.4\pm0.5$ km s$^{-1}$. With two measurements of
847: the rotational broadening of the secondary, we find a mass ratio of
848: $q=0.060\pm0.004$ and a minimum mass of
849: $3.10\pm0.04$ M$_{\sun}$ for the primary object. With the most
850: likely inclination of 41$\degr$ from \citet{Gelino01}, measured in
851: $J,~H,$ and $K$, the black
852: hole has a mass of 11.1 M$_{\sun}.$ The strong infrared flickering
853: discussed earlier, in conjunction with unexplained smooth variability
854: in the lightcurve and uncertainty in the disc spectrum itself, makes
855: it difficult to estimate the true uncertainty in the inclination. The
856: range of reported inclinations, 31$\degr$ to 70.5$\degr$(Gelino et
857: al. 2001 and references therein), results in a black hole mass between
858: 3.7 M$_{\sun}$ and 22.7 M$_{\sun}$. Until the nature and variability
859: of the light from the disc is revealed in full, this conservative
860: error estimate must be sufficient.
861:
862: We also find that the secondary contributes $44\pm7$ per cent of the light
863: near 5500 \AA. As this means that the disc contributes a significant
864: fraction of the light, especially in emission line regions, it becomes
865: important to assess the variability of the disc, particularly if the
866: inclination is to be determined by lightcurve modelling. As noted, it
867: is common to assume that the disc is a constant source of light. While
868: it may be valid when $f$ is large, three observational points
869: cast doubt on this assumption:
870: \begin{enumerate}
871: \item S04 performed a detailed study of flares from
872: A0620-00, which in their observations have amplitudes nearing 20 per
873: cent of the source flux.
874: \item Measurements of the fraction $f$ of light contributed by the
875: secondary have not been consistent. MR86 found $40\pm10$ per cent at
876: 5100 \AA, MRW94 found $94\pm3$ per cent at H$\alpha$, and
877: \citet{Gelino01} found $f\ga97$ per cent in $J,~H,$ and $K$,
878: assuming a \textit{constant} diluting source of light.
879: \item Observations in all four Spitzer bands, taken approximately two
880: weeks before our observations on the Clay telescope, show strong
881: flickering which is highly correlated with simultaneous R-band
882: lightcurves from the 1.2m telescope on Mt. Hopkins.
883: \end{enumerate}
884: While McClintock, Horne, and Remillard (1995) rightly point out that the
885: absorption line strength of the template will affect the observed
886: dilution fraction, unless it can be shown that the measured fraction is
887: strongly correlated with template line strength, a physical origin for
888: this variation cannot be ruled out. Future studies could assess the
889: true dependence of $f$ on template star, as well as the long-term
890: variability of $f$, by observing both BS 753 (MWR94's template) and HD
891: 7142. Our measurements of the H$\alpha$ equivalent
892: widths, when compared to those of MRW94, suggest a physically real
893: origin for the variation, because equivalent widths are independent of
894: the template star and the instrument. So there appears to be evidence
895: for a disc which is brighter relative to the secondary than it used to be.
896:
897: Indeed, the modulations of the equivalent width even point to a
898: physically non-uniform flickering, because the disc emission lines
899: vary relative to the non-stellar continuum. We can tentatively
900: identify the flickering with the crescents, neither of which was
901: present in 1994, so our conclusion seems viable. The line emission
902: and the continuum may have different radial emissivity
903: dependencies, which could result in slower modulations. We also detect
904: several bright regions in the disc: one near the gas stream impact
905: point, and two crescent-shaped regions on opposite sides of the
906: disc. The reality of these features, as well as the non-uniform
907: flickering, are confirmed by modulation tomography of the H$\alpha$
908: disc line, which reveals variation near the bright
909: crescents.
910:
911: First noticed by S04, the crescent-like features may
912: indicate an eccentric disc, which is predicted for systems like A0620-00
913: with small mass ratios and large discs. It seems that we have observed
914: an eccentric disc, but let us consider the evidence. From our
915: observations, the following are clear:
916: \begin{enumerate}
917: \item The disc is bright and variable. The brightness is evident in the
918: increased dilution of the secondary spectrum, and the variability
919: is clear from a number of phenomena. First, the trailed H$\alpha$
920: line shows clear evidence of flickering events. Second, the
921: subtracted equivalent width of the same line is variable beyond
922: explanation by noise alone. Third, the phase-resolved light
923: fraction cannot be reproduced by ellipsoidal variability on
924: top of a constant source of light.
925:
926: \item The disc extends to the 3:1 tidal resonance. This is a simple
927: point, clear from the Doppler and modulation maps.
928:
929: \item The disc is not centered on the radial velocity of the black
930: hole.
931:
932: \item The disc is not radially symmetric, but characterized by bright
933: crescent-shaped regions.
934:
935: \item The crescent regions are modulated at the orbital period.
936: \end{enumerate}
937:
938: Each of these points alone would be insufficient evidence to
939: conclude that the accretion disc is eccentric and
940: precessing. But with the exception of a direct image of the elliptical
941: disc, we can present a complete and coherent argument that this is
942: the case. The disc has grown to tidal resonance, where the enhanced
943: disc viscosity results in bright and variable rims of extra dissipation.
944: We then observe crescents of extra dissipation at relatively low
945: velocities, as expected. Given the viscous effects, it is predicted
946: that the disc will receive a gravitational torque from the secondary,
947: and begin to precess. The asymmetries introduced here shift the
948: velocity center of disc emission away from the black hole, and we find
949: that the disc is not centered on the black hole in velocity
950: space. Furthermore, given the beat period between the precession and
951: orbital motion, the regions of viscous dissipation should be modulated
952: at roughly the orbital period. Modulation tomography reveals this to
953: be the case.
954:
955: In retrospect, knowing that portions of the disc are modulated on the
956: orbital period, we look closer at the fraction of light contributed by
957: the secondary star, and see that it is not well fit by ellipsoidal
958: modulations for a system at the inclination of A0620. But if another
959: component of the system was variable on the orbital period, as we have
960: observed the disc to be, then there is no need for concern. The
961: physical picture, a precessing elliptical disc torqued by the
962: secondary star, predicts and produces all the phenomena we have
963: discussed in our data, which are of high quality.
964:
965: To put it another way, the eccentric disc hypothesis is nicely
966: self-consistent. It explains why and how the accretion disc has
967: changed, allows the disc to be large enough for the growth of
968: eccentric modes, and predicts the phenomena that we observe. It is
969: unfortunately not possible at this point to make an estimate of the
970: disc eccentricity. \citet{Smith07} have shed a great deal of light on
971: the evolution of disc eccentricity and energy dissipation with 3D SPH
972: simulations. They find that systems with $q$ between 0.08 and 0.24
973: develop low-mass eccentric discs withsuperhumps; for $q=0.0526,$ the
974: disc exhibits a short-lived superhump and decaying eccentricity. All
975: mass ratios show enhanced dissipation in the disc from the
976: thermal-tidal instability, even without the eccentric modes.
977:
978: Since we have not observed a superhump, we cannot place A0620-00 in either
979: category. If it falls in the more extreme group, the disc eccentricity
980: is likely zero (reached after about 300 orbital periods)
981: \citep{Smith07}. In that case, the steady state is a massive disc. If the
982: steady state is very long-lived, and the disc continues to grow, this
983: could explain the enormous intensity of novae like A0620-00. If
984: it fits among
985: the less extreme mass ratios, the disc eccentricity is around 0.1--0.2, and a
986: superhump should be observable with better photometry and a longer
987: baseline \citep{Smith07}. A0620-00 may also be at a transition
988: between those cases, and its evolution might be somewhat more
989: erratic, as suggested by the SMARTS data discussed earlier. For
990: example, it may toggle between states of quiescence,
991: superhumps, and variability (like what we have observed here). It
992: might, then, be erroneous to interpret this recent increase in
993: brightness as the build towards outburst.
994:
995: While we have strong evidence that the accretion disc around the
996: black hole has grown out to the tidal distortion radius, evolved into
997: an eccentric disc, and started to precess, further study is required
998: to verify our conclusion. Data from SMARTS, FLWO, and Spitzer will
999: further quantify flickering, and may reveal a superhump, or some new
1000: period consistent with our results, and future programs of tomography
1001: will track the evolution of the accretion disc. In anticipation of the
1002: impending outburst, and in light of progress in simulations, we
1003: suggest that this well-studied system not be
1004: disregarded or ignored, for it affords us the opportunity to watch the
1005: evolution of an accretion disc from quiescence to outburst, and the
1006: chance to test models for disc instabilities in X-ray novae.
1007:
1008: \section*{Acknowledgements}This research was supported by the NSF
1009: grant AST-0507637, the Harvard
1010: University Graduate School of Arts and Sciences (JN), and a SAO Clay
1011: Fellowship (DS). We wish to thank Cara Rakowski for help with the
1012: observations, Tom Marsh for use of his software packages, Jack
1013: Steiner for many useful discussions, and the reviewer for a number of
1014: constructive comments which improved the quality of the paper.
1015:
1016: \begin{thebibliography}{}
1017:
1018: %Nova Mon
1019: \bibitem[\protect\citeauthoryear{Elvis et al.}{1975}]{Elvis75}
1020: Elvis M., Griffiths C.G., Turner M.J.L., Page C.,~1975, IAU
1021: Circ, 2184
1022:
1023: %Accretion Power in Astrophysics
1024: \bibitem[\protect\citeauthoryear{Frank et al.}{2002}]{FKR02}
1025: Frank J., King A., Raine D.,~2002, Accretion Power in
1026: Astrophysics. Cambridge University Press, Cambridge, UK
1027:
1028: %Simulations of Spectral Lines from an Eccentric Precessing Accretion
1029: %Disc
1030: \bibitem[\protect\citeauthoryear{Foulkes et al.}{2004}]{Foulkes04}
1031: Foulkes S.B., Haswell C.A., Murray J.R., Rolfe D.J.,~2004,
1032: MNRAS, 349, 1179
1033:
1034: %On the Maximum Mass of a Maximally Rotating Neutron Star
1035: \bibitem[\protect\citeauthoryear{Friedman \& Ipser}{1987}]{FI87}
1036: Friedman J.L., Ipser J.R.,~1987, ApJ, 314, 594
1037:
1038: %Near-Infrared Light Curves of the Black Hole Binary A0620-00
1039: \bibitem[\protect\citeauthoryear{Froning \& Robinson}{2001}]{FR01}
1040: Froning C.S., Robinson E.L.,~2001, AJ, 121, 2212
1041:
1042: %Near Infrared Spectra of the Black Hole Binary A0620-00
1043: \bibitem[\protect\citeauthoryear{Froning et al.}{2007}]{FR07}
1044: Froning C.S., Robinson E.L., Bitner M.A.,~2007, ApJ, in press
1045:
1046: %A Multiwavelength, Multiepoch Study of the Soft X-ray Transient
1047: %Prototype, V616 Monocerotis (A0620-00)
1048: \bibitem[\protect\citeauthoryear{Gelino et al.}{2001}]{Gelino01}
1049: Gelino D.M., Harrison T.E., Orosz J.A.,~2001, AJ, 122, 2668
1050:
1051: %The Black Hole A0620-00 and its Accretion Disc
1052: \bibitem[\protect\citeauthoryear{Johnston et al.}{1989}]{Johnston89}
1053: Johnston H.M., Kulkarni S.R., Oke J.B.,~1989, ApJ, 345, 492
1054:
1055: %Regularities in the Long-Term Optical Lightcurve of the Black Hole
1056: %Candidate Binary A0620-00 (V616 Mon)
1057: \bibitem[\protect\citeauthoryear{Leibowitz et al.}{1998}]{Leibowitz98}
1058: Leibowitz E.M., Hemar S., Orio M.,~1998, MNRAS, 300, 463L
1059:
1060: %Stellar Radial Velocities 1991-1998
1061: \bibitem[\protect\citeauthoryear{Malaroda et al.}{2001}]{Malaroda01}
1062: Malaroda S., Levato H., Galliani S.,~2001, VizieR On-line Data Catalog:
1063: III/216.
1064:
1065: %Doppler Tomography
1066: \bibitem[\protect\citeauthoryear{Marsh}{2001}]{Marsh01}
1067: Marsh T.R.,~2001, in Boffin H.M.J., Steeghs D., Cuypers J., eds, LNP
1068: Vol. 573, Astrotomography: Indirect Imaging Methods in Observational
1069: Astronomy. Springer-Verlag, Berlin, p. 1
1070:
1071: %Doppler Imaging of the Dwarf Nova U Geminorum
1072: \bibitem[\protect\citeauthoryear{Marsh et al.}{1990}]{Marsh90}
1073: Marsh T.R., Horne K., Schlegel E.M., Honeycutt R.K., Kaitchuck
1074: R.H.~1990, ApJ, 364, 637
1075:
1076: %Spectroscopy of A0620-00: the Mass of the Black Hole and an Image of
1077: %its Accretion Disc.
1078: \bibitem[\protect\citeauthoryear{Marsh et al.}{1994}]{MRW94}
1079: Marsh T.R., Robinson E.L., Wood J.H.,~1994, MNRAS, 266, 137 (MRW94)
1080:
1081: %The Black Hole Binary A0620-00
1082: \bibitem[\protect\citeauthoryear{McClintock \& Remillard}{1986}]{MR86}
1083: McClintock J.E., Remillard R.A.,~1986, ApJ, 308, 110 (MR86)
1084:
1085: %The Dim Inner Accretion Disc of the Quiescent Black Hole A0620-00
1086: \bibitem[\protect\citeauthoryear{McClintock et al.}{1995}]{MHR95}
1087: McClintock J.E., Horne K., Remillard R.A.,~1995, ApJ, 442, 358
1088:
1089: %Evolutionary Processes in Close Binary Systems
1090: \bibitem[\protect\citeauthoryear{Paczynski}{1971}]{Paczynski71}
1091: Paczynski B.,~1971, ARA\&A, 9, 183
1092:
1093: %The Mass of the Black Hole in A0620-00
1094: \bibitem[\protect\citeauthoryear{Shahbaz et al.}{1994}]{Shahbaz94}
1095: Shahbaz T., Naylor T., Charles P.A.,~1994, MNRAS, 268, 756
1096:
1097: %Infrared Spectroscopy of V616 Monaceras (=A0620-00): the Accretion
1098: %Disc Contamination
1099: \bibitem[(Shahbaz et al. 1999)]{Shahbaz99}
1100: Shahbaz T., Bandyopadhyay R.M., Charles P.A.,~1999, A\&A, 346, 82
1101:
1102: %Optical Spectroscopy of Flares from the Black Hole X-ray Transient
1103: %A0620-00 in Quiescence
1104: \bibitem[\protect\citeauthoryear{Shahbaz et al.}{2004}]{Shahbaz04}
1105: Shahbaz T., Hynes R.I., Charles P.A., Zurita C., Casares J.,
1106: Haswell C.A., Araujo-Betancor S., Powell C.,~2004, MNRAS, 354, 31 (S04)
1107:
1108: %Comprehensive Simulations of Superhumps
1109: \bibitem[\protect\citeauthoryear{Smith et al.}{2007}]{Smith07}
1110: Smith A.J., Haswell C.A., Murray J.R., Truss M.R., Foulkes
1111: S.B.,~2007, MNRAS, 378, 785
1112:
1113: %Extending Emission-Line Doppler Tomography: Mapping Modulated Line
1114: %Flux
1115: \bibitem[\protect\citeauthoryear{Steeghs}{2003}]{Steeghs03}
1116: Steeghs D.,~2003, MNRAS, 344, 448
1117:
1118: %The Mass Donor of Scorpius X-1 Revealed
1119: \bibitem[\protect\citeauthoryear{Steeghs \& Casares}{2002}]{Steeghs02}
1120: Steeghs D., Casares J.,~2002, ApJ, 568, 273
1121:
1122: %Notes on Binary Stars. IV. A Source of Spurious Eccentricity in
1123: %Spectroscopic Binaries
1124: \bibitem[\protect\citeauthoryear{Sterne}{1941}]{Sterne41}
1125: Sterne T.E.,~1941, PNAS, 27, 168
1126:
1127: %Linear and Quadratic Limb-Darkening Coefficients for a Large Grid of
1128: %LTE Atmosphere Models
1129: \bibitem[\protect\citeauthoryear{Wade \& Rucinski}{1985}]{Wade85}
1130: Wade R.A., Rucinski S.M.,~1985, A\&SS, 60, 471
1131:
1132: %Superhumps, Resonances, and Accretion Discs
1133: \bibitem[\protect\citeauthoryear{Whitehurst \& King}{1991}]{WK91}
1134: Whitehurst R., King A.,~1991, MNRAS, 249, 25
1135:
1136: \end{thebibliography}{}
1137:
1138: \label{lastpage}
1139:
1140: \end{document}
1141: