1: %\documentclass[manuscript]{aastex}
2: %\documentclass[12pt,preprint]{aastex}
3: \documentclass[12pt,preprint]{aastex}
4:
5: %\documentclass{emulateapj}
6: %\usepackage{apjfonts}
7: \usepackage{graphicx}
8:
9:
10: \shorttitle{FU Ori model} \shortauthors{Zhu et al.}
11:
12: \begin{document}
13:
14: \title{THE HOT INNER DISK OF FU ORI}
15:
16: \author{Zhaohuan Zhu\altaffilmark{1}, Lee Hartmann\altaffilmark{1}, Nuria Calvet\altaffilmark{1},Jesus Hernandez\altaffilmark{1,3}, James
17: Muzerolle\altaffilmark{2}, Ajay-Kumar Tannirkulam\altaffilmark{1}}
18:
19: \altaffiltext{1}{Dept. of Astronomy, University of Michigan, 500
20: Church Street, Ann Arbor, MI
21: 48109;zhuzh@umich.edu,lhartm@umich.edu,ncalvet@umich.edu,hernandj@umich.edu,hernandj@umich.edu}
22: \altaffiltext{2}{Steward Observatory, 933 North Cherry Avenue,
23: University of Arizona,Tucson, AZ 85721;jamesm@as.arizona.edu}
24: \altaffiltext{3}{Centro de Investigaciones de Astronomia, Apartado
25: Postal 264, Merida 5101-A, Venezuela} \email{}
26:
27: \newcommand\msun{\rm M_{\odot}}
28: \newcommand\lsun{\rm L_{\odot}}
29: \newcommand\rsun{\rm R_{\odot}}
30: \newcommand\msunyr{\rm M_{\odot}\,yr^{-1}}
31: \newcommand\be{\begin{equation}}
32: \newcommand\en{\end{equation}}
33: \newcommand\cm{\rm cm}
34: \newcommand\kms{\rm{\, km \, s^{-1}}}
35: \newcommand\K{\rm K}
36: \newcommand\etal{{\rm et al}.\ }
37: \newcommand\sd{\partial}
38:
39: \begin{abstract}
40: We have constructed a detailed radiative transfer disk model which
41: reproduces the main features of the spectrum of the outbursting
42: young stellar object FU Orionis from $\sim$ 4000 \AA\, to $\sim 8
43: \mu$m. Using an estimated visual extinction $A_{V} \sim 1.5$, a
44: steady disk model with a central star mass $\sim 0.3 M_{\odot}$ and
45: a mass accretion rate $\sim 2 \times 10^{-4} M_{\odot}$yr$^{-1}$, we can
46: reproduce the spectral energy distribution of FU Ori quite well.
47: Higher values of extinction used in previous analysis (A$_{V}$$\sim$ 2.1) result
48: in spectral energy distributions which are less well-fitted by a
49: steady disk model, but might be explained by extra energy
50: dissipation of the boundary layer in the inner disk. With the
51: mid-infrared spectrum obtained by the Infrared Spectrograph (IRS) on
52: board the {\em Spitzer Space Telescope}, we estimate that the outer
53: radius of the hot, rapidly accreting inner disk is $\sim$ 1 AU using disk
54: models truncated at this outer radius. Inclusion of radiation from
55: a cooler irradiated outer disk might reduce the outer limit of
56: the hot inner disk to $\sim$ 0.5 AU.
57: In either case, the radius is inconsistent with a pure thermal instability model for the
58: outburst. Our radiative transfer model implies that the central disk
59: temperature $T_{c} \ge 1000$~K out to $\sim$ 0.5 $-$ 1 AU, suggesting that
60: the magnetorotational instability can be supported out to that
61: distance. Assuming that the $\sim 100$~yr decay timescale in
62: brightness of FU Ori represents the viscous timescale of the hot
63: inner disk, we estimate the viscosity parameter to be $\alpha \sim
64: 0.2 - 0.02$ in the outburst state, consistent with numerical
65: simulations of the magnetorotational instability in disks. The
66: radial extent of the high $\dot{M}$ region is inconsistent with the
67: model of Bell \& Lin, but may be consistent with theories
68: incorporating both gravitational and magnetorotational
69: instabilities.
70: \end{abstract}
71:
72: \keywords{accretion disks, circumstellar matter, stars: formation, stars: pre-main sequence}
73:
74: \section{Introduction}
75: The FU Orionis systems are a small but remarkable class of variable young objects
76: which undergo outbursts in optical light of 5 magnitudes or more \citep{herbig77}, with a F-G
77: supergiant optical spectra and K-M supergiant
78: near-infrared spectra dominated by deep CO overtone absorption.
79: While the rise times for outbursts are usually very short ($\sim$ 1-10 yr), the
80: decay timescales range from decades to centuries. The FU Ori
81: objects also show distinctive reflection nebulae, large infrared
82: excesses of radiation, wavelength dependent spectral types, and ``double-peaked''
83: absorption line profiles \citep{Lee96}. The frequency of these
84: outbursts is uncertain; in recent years an increasing number of heavily extincted
85: potential FU Ori objects have been identified on the basis of their spectroscopic
86: characteristics at near-infrared wavelengths \citep{kenyon93,reipurth97,sandell98,aspin03}.
87:
88: The accretion disk model for FU Ori objects proposed by Hartmann
89: $\&$ Kenyon (1985, 1987a, 1987b) and \cite{kenyon88} can explain the
90: peculiarities enumerated above in a straightforward manner.
91: Outbursts are known in other accreting disk systems and may be the
92: result of a common mechanism (e.g., Bell $\&$ Lin 1994). The high
93: temperature of the inner disk produces the observed F-G supergiant
94: optical spectrum, while the cooler outer disk produces an infrared
95: spectrum having the spectral type of a K-M supergiant. The Keplerian
96: rotation of the disk can produce double-peaked line profiles as
97: often observed, with peak separation decreasing with increasing
98: wavelength of observations, since the inner hotter disk which
99: produces the optical spectrum rotates faster than the outer cooler
100: disk which produces the infrared spectrum \citep{Lee96}.
101:
102: While the accretion disk model has been successful so far, it is
103: important to continue to test it and to derive further insights into
104: the origins of the accretion outbursts as new observations become
105: available. In particular, the mid-infrared spectrum obtained with
106: the Infrared Spectrograph (IRS) on the {\em Spitzer Space Telescope}
107: provides important constraints on the outer edge of the hot inner
108: disk \citep{jgreen06}, which in turn can test theories of outbursts.
109: For example, the thermal instability model of \cite{bell94} predicts
110: an outer radius for the outburst region of only $\sim 20 \rsun$,
111: which is smaller than that estimated by \cite{jgreen06} on the basis
112: of simple blackbody disk models. \cite{armitage01} suggested that a
113: combination of gravitational instability (GI) with triggering of the
114: magnetorotational instability (MRI) might also explain FU Ori
115: outbursts; in this model the high accretion rate region in outburst
116: would be much larger, of order 0.5 AU. Vorobyov \& Basu (2005,2006)
117: suggested that FU Ori outbursts could also be produced by the
118: accretion of clumps formed in a gravitationally unstable disk. This
119: model predicts high accretion rates at its inner radius of 10 AU.
120: These differing predictions for the size of the outburst region
121: should be testable with detailed models of the spectral energy
122: distribution (SED).
123: Furthermore, if we can
124: constrain the size of the outburst region, the timescales of decay
125: may provide quantitative estimates of the viscosity, vital for
126: understanding the evolution of young protoplanetary disks.
127:
128:
129:
130: In this paper we present new steady, optically-thick accretion disk spectrum calculations,
131: and compare the results with observations from the ultraviolet
132: to the mid-infrared of FU Ori$-$the prototype of these systems.
133: In \S 2 we describe the observational material used to constrain the models, while
134: in \S 3 we describe the methods used to calculate the disk models.
135: We show in \S 4 that the steady disk model reproduces the variation of the observed
136: spectral features over this very large wavelength range quite well.
137: In \S 5 we consider some implications of our results for disk viscosities, outburst
138: mechanisms, and disk masses, and summarize our conclusions in \S 6.
139:
140: \section{Optical and infrared data}
141:
142: An essential part of the spectral energy distribution (SED) of FU Ori for our model is the
143: IRS spectrum from {\em Spitzer} observed on March 4, 2004 (see Green et al. 2006 for details).
144: As FU Ori has been fading slowly over the last 60 years, we need to obtain data at other
145: wavelengths near the same time as the IRS spectrum to construct a complete spectral
146: energy distribution (SED).
147:
148: Optical photometry
149: (UBVR) was obtained in 2004 at the Maidanak Observatory
150: \citep{ibrahimov99,jgreen06}. As FU Ori exhibits small, irregular variability
151: on timescales much less than a year, we averaged the Maidanak data
152: to form the mean values (with RMS uncertainty in each band $\sim 0.02$ magnitude)
153: used in this paper. In addition, we include an optical spectrum of
154: FU Ori obtained on March 27th, 2004, near
155: the time of the IRS observations, using the 1.5m telescope of the
156: Whipple Observatory with the FAST Spectrograph \citep{fabricant98}, equipped
157: with the Loral 512 $\times$ 2688 CCD. This instrument provides
158: 3400 {\AA} of spectral coverage centered
159: at 5500 {\AA}, with a resolution of 6 \AA. The spectrum was wavelength calibrated
160: and combined using standard IRAF
161: routines \footnote{IRAF is distributed by the National Optical Astronomy Observatories,
162: which are operated by the Association of Universities for Research in Astronomy, Inc.,
163: under cooperative agreement with the National Science Foundation.}. The spectrum was
164: corrected for the relative system response using the IRAF {\it
165: sensfunc} task. The optical spectrum was placed on an absolute flux scale
166: using the averaged 2004 UBVR photometry.
167:
168: Near-infrared fluxes were determined from several sources. 2MASS JHK
169: photometry is available from October 7th, 1999, five years before
170: IRS observations. We also include data from \cite{reipurth2004} on
171: December 15, 2003, using the NAOJ 8 m Subaru Telescope on Mauna Kea
172: and the Infrared Camera and Spectrograph (IRCS). Finally, we
173: obtained JHK magnitudes on December 17th 2005, using the Near-IR
174: Imager/Spectrometer TIFKAM on the 2.4 m MDM telescope at Kitt Peak.
175: Although the 2.4m telescope had to be defocused
176: to avoid saturation because FU Ori is so bright,
177: our JHK magnitudes of 6.57, 5.91 and 5.33, respectively,
178: are in reasonable agreement with the JHK$_s$ 2MASS magnitudes of
179: 6.519, 5.699, 5.159, respectively, indicating that the FU Ori
180: fluxes have not changed dramatically over that six year period.
181: The \cite{reipurth2004} observations are on a different system,
182: but when converted to fluxes yield reasonably similar results.
183: We therefore adopted the 2MASS magnitudes as our standard
184: in this wavelength region.
185:
186: We then incorporated the KSPEC spectrum (resolution R$\sim$500,
187: 1.15-2.42$\mu$m ) observed on Dec 15th, 1994 by \cite{greene96} and
188: the observation with SpeX (2.1-4.8$\mu$m) on the Infrared Telescope
189: Facility (IRTF; Rayner et al. 1998) obtained during the nights of January
190: 5-6th, 2001 (see Muzerolle et al. 2003 for details). We placed these
191: spectra on an absolute scale by convolving the spectra with the
192: relative spectral response curve (RSRs) of the 2MASS
193: system \citep{cohen03} and tying the overall result to the 2MASS
194: K$_s$ magnitude.
195:
196: The collected observations are shown in Figure \ref{fig:1}, where we assume an A$_{V}$
197: of 1.5 instead of the usually assumed value of $\sim$ 2.2 (Kenyon \etal 1988; KHH88).
198: %I change \S 4 to KHH88 to make it clear
199: The previously discussed variation of spectral type with wavelength is immediately
200: apparent in Figure \ref{fig:1};
201: the optical spectrum corresponds to a G type spectrum as estimated
202: from the atomic metal lines while
203: % I get rid of (\S 4) to reduce the references to later chapter
204: the near-IR and IRS wavelength ranges are dominated by molecular bands indicative
205: of a much later spectral type. While the optical and infrared spectra are
206: dominated by absorption features, dust emission dominates longward of
207: 8 $\mu$m, as shown by the silicate emission features at 10 and 20 $\mu$m.
208:
209: \section{Model}
210:
211: In this paper we only model the absorption spectrum ($\lambda$$<$8 $\mu$m)
212: which is produced by the centrally-heated accretion disk. The silicate emission
213: features are produced by externally-heated regions of the outer disk or possibly
214: a circumstellar dusty envelope (Kenyon \& Hartmann 1991; Green \etal 2006);
215: we defer modeling of this region to a subsequent paper.
216:
217: We follow the method of Calvet \etal (1991a,b) to calculate the disk
218: spectrum. In summary, we calculate the emission from the atmosphere of
219: a viscous, geometrically thin,
220: optically thick accretion disk with constant mass accretion rate
221: $\dot{M}$ around a star with mass M and radius R. The disk height
222: $H$ is assumed to vary with the distance from the rotational axis of
223: the star $R$ as $H=H_{0}(R/R_{i})^{9/8}$, where we take $H_{0}=0.1
224: R_{i}$ and $R_{i}$ is the radius of the central star. This
225: approximation is not very accurate but it only affects the local
226: surface gravity of the disk atmosphere, which has only a small
227: effect on the emergent spectrum. Viscous dissipation in the
228: atmosphere is neglected, which is a good approximation for cases of
229: interest \citep{HK91}. Thus, radiative equilibrium holds in the disk
230: atmosphere, and the surface flux is determined by the viscous energy
231: generation in the deeper disk layers. This constant radiative flux
232: through the disk atmosphere can be characterized by the effective
233: temperature distribution of the steady optically-thick disk,
234: \begin{equation}
235: T_{eff}^{4}= \frac{3GM\dot{M}}{8\pi\sigma R^{3}} \left [ 1-\left (
236: \frac{R_{i}}{R} \right ) ^{1/2} \right ] \,. \label{eq:teff}
237: \end{equation}
238: This equation predicts that the maximum disk temperature $T_{max}$
239: occurs at $1.36 R_{i}$ and then decreases to zero at $R = R_i$. The
240: latter is unphysical, and so we modify equation (\ref{eq:teff}) so
241: that when the radius is smaller than 1.36 R$_{i}$, we assume that
242: the temperature is constant and equal to $T = T_{max}$. (e.g.,
243: KHH88) The vertical temperature structure at each
244: radius is calculated using the gray-atmosphere approximation in the
245: Eddington limit, adopting the Rosseland mean optical depth $\tau$.
246:
247: The opacity of atomic and molecular lines has been calculated using
248: the Opacity Distribution Function (ODF) method \citep{kurucz04,
249: kurucz042,castelli05}. Briefly, the ODF method is a statistical
250: approach to handling line blanketing when millions of lines are
251: present in a short wavelength range \citep{kurucz74}. For a
252: given temperature and pressure, the absorption coefficient for each
253: line is exactly computed, then the profiles of all the lines in each
254: small interval $\Delta \nu$ are rearranged monotonically as a
255: function of $\nu$. The opacity increases as the frequency increases
256: in this $\Delta \nu$. A step function with 12 subintervals in
257: frequency is used to represent this monotonic function. The height
258: of every step is the averaged opacity of the monotonic function in
259: this subinterval. The width of the 12 subintervals are $\Delta
260: \nu$/10 for 9 intervals and $\Delta \nu$/20, $\Delta \nu$/30,
261: $\Delta \nu$/60 for the last 3 intervals where the monotonic
262: function increases steeply and reaches the maximum. Thus, in every
263: interval $\Delta \nu$ we obtain 12 representative averaged opacities
264: at a given temperature and pressure. We then construct an opacity
265: table as a function of $\lambda$ for each temperature and pressure.
266: The wavelength grid is the same as used in the code DFSYNTHE
267: \citep{castelli05} from 8.976 nm to 10000 nm with 328 BIG intervals
268: (resolution ~30-100) and 1212 LITTLE intervals (resolution ~60-500).
269: However, we extend the temperature grid and pressure grid to lower
270: temperatures and pressures ($\log T$ ranging from 1.5 to 5.3, $\log
271: P$ from -24 to 8 in cgs units) than DFSYNTHE, so that it can be used in our disk
272: % add in cgs units
273: models for which the temperatures and pressures reach lower values
274: than in typical stellar atmosphere models. Because the line opacity
275: usually varies by several orders of magnitude within a line width,
276: the ODF method is substantially more accurate than either straight
277: or harmonic means.
278:
279: The line list is taken from Kurucz's CD-ROMs Nos. 1, 15, 24, and 26
280: \citep{kurucz05}. Not only atomic lines but also many molecular
281: lines (C$_{2}$, CN, CO, H$_{2}$, CH, NH, OH, MgH, SiH, SiO, TiO,
282: H$_{2}$O) are included. The opacities of TiO and H$_{2}$O, the most
283: important molecules in the infrared, are from \cite{PS97} and
284: \cite{S98}. We do not include a detailed calculation of the dust
285: condensation process; instead, we add the dust opacity when T$<$1500
286: K (the condensation temperature of silicates at typical disk
287: densities). We use the ISM dust model of \cite{draine84,draine87} to
288: represent the dust in the disk. This assumption is unlikely to be
289: correct but mostly affects the continuum at the very longest
290: wavelengths and in particular the silicate emission features, which
291: we do not model in this paper.
292:
293: At low temperatures complex chemical processes occur which are not included in the Kurucz data.
294: We have not calculated the low temperature
295: molecular opacity in detail;
296: instead, we have assumed that the abundance ratio between different types
297: of molecules below 700 K is the same as the ratio at 700 K. This is unimportant
298: for our purposes because dust opacity dominates at such low temperatures.
299:
300: In Figure \ref{fig:ross} we show the Rosseland mean opacity
301: calculated for solar abundances and a turbulent velocity $v = 2
302: \kms$ for different values of $\rho /T_6^3$ to be compared with
303: \cite{alexander94}, where $\rho$ is the density in gcm$^{-3}$ and
304: $T_6$ is the temperature in millions of degrees. The results closely
305: match the more detailed calculations by \cite{alexander94} and
306: \cite{ferguson05} except near 1500 K, where our dust opacity rises
307: more rapidly due to our neglect of dust condensation processes.
308:
309: \cite{calvet91b} showed that the near-infrared spectrum of FU Ori
310: could be well-modeled with the then-current water vapor opacities
311: and a similar calculational method to the one we are using here.
312: We now have a much better set of opacities, and can treat the optical spectrum.
313: To test the code at optical wavelengths, we
314: compare the model spectrum from the annulus with T$_{eff}$=5300 K
315: with the observed spectrum of SAO 21446 which is a G1 supergiant
316: \citep{jacoby84}. The spectrum of SAO 21446 has been obtained using
317: the Intensified Reticon Scanner(IRS) on the No.1 90 cm telescope at
318: Kitt Peak by \cite{jacoby84}. Here we have convolved this spectrum
319: to a resolution of 300 to agree with our model spectrum.
320: %LH where does the observed spectrum come from????
321: As shown in Figure \ref{fig:f3},
322: our model reproduces the continuum spectrum fairly well
323: except for a few strong features. These discrepancies are probably
324: due to the limitation of the grey atmospheric assumption.
325:
326: Figure \ref{fig:f4} shows the emergent intensities with 55$^{o}$ emergent angle
327: from different annuli of
328: our final disk model (parameters can be found in \S 4), which has
329: 45 annuli. The 55$^{o}$ inclination angle is estimated from the near-infrared interferometric
330: observations by \cite{malbet05} and mid-infrared interferometric observations by Quanz \etal (2006).
331: The radii of these annuli are chosen to
332: increase exponentially from $r = R_i$ to $1000 R_i$.
333: At the first annulus ($r =1 R_i$), we can see a significant Balmer jump at 0.36 $\mu$m, similar to
334: the spectra of early type stars. As the radius increases, the temperatures of the annuli
335: become lower and lower. Infrared molecular features appear at larger annuli with
336: effective temperatures less than 5000K ($r > 3 R_i$). For the outermost annuli with effective
337: temperature less than 1500 K ($r > 15 R_i$), the molecular features have almost disappeared
338: because dust opacity dominates.
339: The final spectrum is the result of the addition of the fluxes from each of these annuli
340: weighted by the appropriate annular surface area.
341:
342: In addition to these calculations aimed at comparing with low-resolution spectra,
343: we also calculated high-resolution spectra in restricted wavelength regions. For these calculations,
344: the vertical structure of every annulus is calculated as above,
345: and then this structure is imported into the program
346: SYNTHE \citep{kurucz81,kurucz93}. SYNTHE is a suite of programs which solve the
347: radiative transfer equations in LTE with a very high spectrum resolution ($\sim$500,000).
348: Each annular spectrum is broadened with the rotational profile appropriate to a Keplerian
349: disk and then the summed spectrum is calculated (e.g., KHH88).
350: These calculations are very time-consuming, and were used only
351: to produce a high resolution spectrum of a small wavelength range around 6170 \AA\
352: for the purpose of estimating the central mass, and for examining the CO first-overtone lines.
353: % get rid of \S 4 to reduce the reference to later section
354:
355: \section{Comparison with observations}
356:
357: The SED of a steady, optically-thick disk model is determined by
358: two parameters: the product of the mass accretion
359: rate and the mass of the central star, $\dot{M} M$, and the inner radius $R_i$.
360: One observational constraint comes from the observed spectral lines and/or the peak
361: of the SED (after reddening correction), which determines the maximum temperature
362: of the steady disk model,
363: \begin{equation}
364: T_{max}=0.488 \left ( \frac{3GM\dot{M}}{8\pi R_{i}^{3}\sigma} \right
365: )^{1/4}\,.\label{eq:tmax}
366: \end{equation}
367: The other observational constraint is given by the true luminosity
368: of a flat disk L$_{d}$,
369: \begin{equation}
370: L_{d} = 2 \pi d^{2} {F \over \cos i} = { G M \dot{M} \over 2
371: R_i}\,,\label{eq:Ld}
372: \end{equation}
373: %correct the equation
374: where $d$ is the distance of FU Ori,
375: $i$ is the inclination angle of the disk to the line of sight, and
376: $F$ is the observed total flux corrected for extinction.
377: If the distance and inclination are known, we can solve for the inner disk radius
378: and thus for the product $M \dot{M}$. If we then further use the observed
379: rotational velocities, it is then possible to derive independent values of the
380: central mass $M$ and the accretion rate $\dot{M}$, in the manner outlined by
381: KHH88.
382:
383: The best constraints on $T_{max}$ come from the optical spectrum, as the
384: uncertain extinction makes it difficult to constrain by overall SED
385: fitting. From the lines of our optical FAST spectrum
386: we derive a spectral type of $\sim$ G2 for FU Ori using the methods of \cite{jesus04}, which
387: agrees with previous results \citep{herbig77}.
388: However, it is not straightforward to derive a value of extinction and thus $T_{max}$
389: from this spectral type determination; the disk spectrum is not that of a single (standard)
390: star with a well-defined effective temperature, but instead it is a combination of hotter
391: and cooler regions, with an overall spectrum that clearly varies with the wavelength
392: of observation in a complicated way.
393: To address this problem, we constructed several disk models with a modest range of
394: $T_{max}$ and compared them with the observed optical spectrum dereddened by
395: differing amounts of visual extinction $A_V$, as shown in Figure \ref{fig:f5}.
396:
397: The uppermost model in Figure \ref{fig:f5} has $T_{max} = 7240$~K,
398: comparable to that of the FU Ori disk model of KHH88. The model
399: optical spectrum is too flat longward of $\sim 3900$~\AA\ in
400: comparison with the observations dereddened by the $A_V = 2.2$,
401: suggested by KHH88, or for $A_V=1.9$. At the other extreme, we find
402: that the bottom model, with $T_{max} = 5840$~K, matches the observed
403: spectrum dereddened by only $A_V = 1.3$ quite well. However, we
404: suspect that our gray-atmosphere approximation somewhat
405: underestimates the amount of line blanketing in the blue optical
406: region, especially shortward of the Ca II resonance lines at 3933
407: and 3968 \AA\, leading us to suspect that this agreement is somewhat
408: misleading. We therefore provide our best estimate of the extinction
409: as $A_V = 1.5 \pm 0.2$, and adopt the model which best matches the
410: mean extinction with $T_{max} = 6420$~K as our standard model.
411:
412: In this connection it is worth noting that KHH88 did not actually
413: find a good fit for the optical spectrum of FU Ori using a GOI
414: standard star and extinctions $A_V > 2$. They speculated that the
415: discrepancy might be alleviated by an improved treatment of limb
416: darkening in the disk. However, Hartmann \& Kenyon (1985) had found
417: that a G2I standard star and $A_V = 1.55$ provided a good match to
418: the optical spectrum between 3900 \AA\ and 7400 \AA\,, consistent
419: with our results here.
420:
421: %LH should we make that figure in \AA\ rather than in log lambda?
422:
423: The accretion disk luminosity $L$ depends upon the distance,
424: inclination, and dereddened flux. We adopt a distance of $\sim$ 500
425: pc, consistent with membership in the general Orion region (Herbig
426: 1977). The inclination 55$^{o}$ is adopted
427: (Malbet et al.2005, Quanz et al. 2006). Using these
428: parameters, observed total flux and $A_V = 1.5$, we obtain the true
429: luminosity $L_{d}=8.66\times10^{35} \rm ergs \,
430: s^{-1}$$\sim$226$L_{\odot}$ according to equation \ref{eq:Ld}. As
431: mentioned before, fitting the observed spectrum yields $T_{max} =
432: 6420$~K. We derive $M\dot{M} = 7.2 \times 10^{-5}
433: M_{\odot}^{2}$yr$^{-1}$ and $R_i = 5 R_{\odot}$ by solving equation \ref{eq:tmax} and equation
434: \ref{eq:Ld} simultaneously.
435: %get rid of apparent luminosity calculation and add solving equation 2 and 3 to make it clear.
436:
437: Figure \ref{fig:f6} shows the predicted spectrum for the adopted
438: parameters and several values of the outer radius. Our computed
439: spectrum matches the observations at wavelengths
440: shortward of about $4 \mu$m, indicating that the innermost
441: region of the inner disk is reasonably well reproduced by the adopted
442: $\dot{M} M$ and $R_i$ parameters.
443:
444: Because the optical and near-infrared SED is well-matched by the model, we may
445: then proceed to estimate the central stellar mass and thus the accretion rate.
446: We used the observed optical rotation from KHH88 to constrain the central star mass.
447: We computed the line profiles around 6170 $\dot{A}$ using SYNTHE, cross
448: correlated it with a non-rotating disk spectrum, and compared it with the
449: observed cross-correlation line profile given in KHH88. The uncertainty in
450: fitting the line profile widths is about $\pm$~20\%. However, systematic
451: errors in the inclination angle are probably more important for the mass
452: estimate, and for that reason we give an uncertainty
453: $0.3 M_{\odot}\pm0.1 M_{\odot}$ (
454: consistent with the earlier estimate of KHH88 assuming $i = 50^{\circ}$).
455: For a mass of 0.3 $M_{\odot}$, $\dot{M} \sim 2.4 \times 10^{-4} M_{\odot}$yr$^{-1}$.
456: Parameters of the model are given in Table 1.
457:
458: As shown in Figure \ref{fig:f6},
459: we find that regardless of extinction and disk parameters, steady accretion models
460: which fit the optical to near-infrared region predict too much
461: emission in the IRS range for very large outer radii. This can only be
462: remedied by truncating the hot inner disk. The estimate of the
463: outburst models of \cite{bell94} suggested an outer radius of the
464: high state of $\sim 20 \rsun$ (with somewhat different parameters
465: for the accretion rate and inner disk radius); as shown in Figure
466: \ref{fig:f6} this truncation radius fails to explain the SED. A
467: truncation radius of $R_{out} \sim 210 R_{\odot} \sim 1$~AU provides
468: a better fit to the flux at $\sim 5 - 8 \mu$m, although is somewhat
469: low in the $3 \mu$m band.
470: Our results are reasonably consistent with
471: \cite{jgreen06}, considering the limitations of their blackbody modeling only out to 5 microns.
472:
473: As discussed above,
474: models with outer radii larger than $\sim$200$R_{\odot}$ overpredict the flux at $\sim 5 - 8 \mu$m.
475: At even longer wavelengths, absorption features are no longer present; instead,
476: silicate dust emission features are seen in IRS spectra at $\sim 10$ and $20 \mu$m.
477: As discussed by \cite{jgreen06}, these silicate features are signatures of heating
478: from above, rather than from internal viscous dissipation, and may be produced in
479: upper atmosphere of the outer disk which absorb light from the central disk.
480: An outer disk is expected to be present, since
481: a disk radius of $\sim 0.5 - 1$~AU is extremely small by standards of typical T Tauri
482: disks, for which $R_{out} \sim 100$~AU or more are common (e.g., Simon, Dutrey,
483: \& Guilloteau 2000). Moreover,
484: such a small radius would imply a low-mass optically-thick disk,
485: making it difficult
486: to explain the observed submm flux from FU Ori
487: (Sandell \& Weintraub 2001). So, it is very likely that an outer disk is present with a lower
488: accretion rate, which could contribute some flux in the $5 - 8 \mu$m wavelength
489: range and thus reduce the radius of the hot, high-accretion rate region.
490: An outer disk accretion rate of about $10^{-5} \msunyr$ is comparable to some of
491: the highest infall rates estimated for (not heavily embedded) protostars (e.g., Kenyon,
492: Calvet \& Hartmann 1993); higher accretion rates would imply implausibly short
493: evolutionary timescales $M/\dot{M} \sim 3 \times 10^4$~yr. In any event, if the
494: outer disk accretion rate is not significantly smaller than the inner disk accretion
495: rate, the problem of a large outer radius yielding too much flux remains.
496: More likely, the emission from the outer disk is dominated by irradiation from the
497: inner disk independent of the (lower) accretion rate there \citep{turner97}, as indicated
498: by the presence of emission rather than absorption features.
499:
500: In the \cite{bell94} thermal instability models for FU Ori outbursts,
501: the outer disk accretion rate is of order $10^{-1}$ of the inner disk accretion
502: rate in outburst. We have therefore investigated the effect of an outer disk
503: accretion rate of $2.4 \times 10^{-5} \msunyr$ on the SED. The dotted curves
504: in Figure \ref{fig:f6} show the results of adding in such outer disk emission
505: from disk internal viscous dissipation.
506: The effect is relatively small so that our estimate of the outer radius of
507: the hot disk with the high accretion rate is relatively robust. Of course
508: we cannot rule out other kinds of temperature distributions, such as a smooth
509: decline of accretion rate from, say, 100 to 200 $\rsun$ and
510: the irradiation effect from the inner disk to the outer disk.
511: %These uncertainties
512: %in interpreting the SED do not affect our conclusions in any major way (\S 5).
513: We are presently carrying out more detailed calculations of the outer
514: disk, including irradiation by the inner disk. Preliminary results indicate
515: that the addition of the irradiated outer disk emission
516: will decrease our hot disk outer radius estimate by no more than a factor of two (Zhu et al., in preparation).
517:
518:
519: %; we do not model this
520: %dust emission in this paper. If we include the irradiated outer disk to explain the silicate emission,
521: % the emission from the outer disk will increase, which will contribute more in
522: %the mid-infrared. In this case our fitted outer radius of the inner disk has to decrease to
523: %have the total flux of both the inner and outer disk still fit the IRS spectrum.
524: %Our preliminary results suggest the addition of the irradiated outer disk emission
525: % will decrease our outer radius of
526: %the hot disk by no more than a factor of two.
527:
528: %We find that regardless of extinction and disk parameters, models
529: %which fit the optical to near-infrared region predict too much
530: %emission in the IRS range for large outer radii. This can only be
531: %remedied by truncating the hot inner disk. The estimate of the
532: %outburst models of \cite{bell94} suggested an outer radius of the
533: %high state of $\sim 20 \rsun$ (with somewhat different parameters
534: %for the accretion rate and inner disk radius); as shown in Figure
535: %\ref{fig:f6} this truncation radius fails to explain the SED. A
536: %truncation radius of $R_{out} \sim 210 R_{\odot} \sim 1$~AU provides
537: %a better fit to the flux at $\sim 5 - 8 \mu$m, although is somewhat
538: %low in the $3 \mu$m band. As mentioned above, including the outer irradiated disk may reduce this
539: %radius, but no more than a factor of two.
540: %Our results are reasonably consistent with
541: % \cite{jgreen06}, considering the limitations of their blackbody modelling only out to 5 microns.
542: %
543:
544: %An outer disk radius of $\sim 0.5 - 1$~AU is extremely small by standards of typical T Tauri
545: %disks, for which $R_{out} \sim 100$~AU or more are common (e.g., Simon, Dutrey,
546: %\& Guilloteau 2000). In addition, such a small radius would imply a low-mass optically-thick disk,
547: %making it difficult
548: %to explain the observed submm flux from FU Ori
549: %(Sandell \& Weintraub 2001). It is very likely that an outer disk is present with a lower
550: %accretion rate, which could contribute some flux in the $5 - 8 \mu$m wavelength
551: %range and thus reduce the radius of the hot, high-accretion rate region.
552: %An outer disk accretion rate of about $10^{-5} \msunyr$ is comparable to some of
553: %the highest infall rates estimated for (not heavily embedded) protostars (e.g., Kenyon,
554: %%Calvet \& Hartmann 1993); higher accretion rates would imply implausibly short
555: %evolutionary timescales $M/\dot{M} \sim 3 \times 10^4$~yr. In any event, if the
556: %outer disk accretion rate is not significantly smaller than the inner disk accretion
557: %rate, the problem of a large outer radius yielding too much flux remains.
558: %More likely, the emission from the outer disk is dominated by irradiation from the
559: %inner disk independent of the (lower) accretion rate there \citep{turner97}, as indicated
560: %by the presence of emission rather than absorption features.
561: %
562:
563: %In the \cite{bell94} thermal instability models for FU Ori outbursts,
564: %the outer disk accretion rate is of order $10^{-1}$ of the inner disk accretion
565: %rate in outburst. We have therefore investigated the effect of an outer disk
566: %accretion rate of $2.4 \times 10^{-5} \msunyr$ on the SED. The dotted curves
567: %in Figure \ref{fig:f6} show the results of adding in such outer disk emission.
568: %The effect is relatively small so that our estimate of the outer radius of
569: %the hot disk with the high accretion rate is relatively robust. Of course
570: %we cannot rule out other kinds of temperature distributions, such as a smooth
571: %decline of accretion rate from, say, 100 to 200 $\rsun$ and
572: %the irradiation effect from the inner disk to the outer disk.
573: %These uncertainties
574: %in interpreting the SED do not affect our conclusions in any major way (\S 5).
575:
576: We explored the effects of a larger visual extinction
577: by scaling the observations to A$_{V}$=2.2. As shown in
578: Figure \ref{fig:7}, this makes only a very slight difference to the
579: long-wavelength spectrum and does not change our estimate of the outer
580: radius.
581:
582: To examine our consistency with previous results,
583: we calculated the CO first overtone high resolution spectrum
584: with $R_{out}=210 R_{\odot}$ using SYNTHE program and compared it with the
585: high resolution spectrum from \cite{lee04}. The comparison is shown in Figure \ref{fig:CO}.
586: We found that we needed to adopt a turbulent velocity of 4 ${\rm km \, s^{-1}}$ to
587: obtain lines that are deep enough in comparison with the observations.
588: This is the sound speed of gas with temperature around 4000 K, slightly
589: supersonic for the annulus with radius larger than 4 stellar radius;
590: %LH Zhaohuan! by how much! and also refer to Hartmann, Hinkle and Calvet!!!
591: Hartmann, Hinkle, \& Calvet (2004) similarly found that slightly supersonic
592: turbulence was needed to explain the first-overtone CO lines, and pointed out
593: that this would not be surprising in the context of turbulence driven
594: by the magnetorotational instability.
595:
596:
597: In Figure \ref{fig:f9} we show an expanded view of the wavelength
598: region between $\sim 3$ and $8 \mu$m. The model accounts for the
599: overall shape of the SED reasonably well, and yields a reasonable
600: strength for the 6.8$\mu$m water vapor absorption feature. However,
601: the 5.8$\mu$m water vapor feature of the model is not as strong as
602: it is in the observations (\citep{jgreen06}). In this wavelength
603: %add one reference
604: region the contributions from the dust-dominated regions of the disk
605: are significant, and the contribution of this dust (relatively
606: featureless) continuum emission could reduce the strength of the
607: absorption feature. To investigate this possibility, we reduced the
608: dust opacity to only 1$\%$ that of
609: \cite{draine84,draine87} and recalculated the model (with a slightly
610: larger outer radius). As can be seen in Figure \ref{fig:f9}, this
611: reduction in dust opacity strengthens the absorption features,
612: although the predicted 5.8$\mu$m feature is still not strong enough,
613: perhaps suggesting some difficulty with the opacities. It is not
614: clear whether we require dust depletion, because we have not
615: considered the temperature dependence of dust condensation in our
616: model. We also cannot rule out uncertainties due to our
617: % get rid of \S 3 to make it more smooth to read.
618: simple treatment of the vertical temperature structure of the disk.
619:
620: \section{Discussion}
621:
622: \subsection{The steady accretion disk model and the inner boundary condition}
623:
624: We have shown that the steady accretion disk model can
625: quantitatively account for the observed SED of FU Ori over a factor
626: of nearly 20 in wavelength, from 4000 \AA\ to about $8 \mu$m. This
627: is accomplished with only three adjustable parameters: $R_i$,
628: $\dot{M} M$, and $R_{out}$. No alternative model proposed for FU
629: Ori has explained the SED over such a large wavelength range. The
630: accretion disk explanation is bolstered by this work.
631:
632: Within the
633: context of the steady disk model, however, there remains the
634: question of the appropriate inner boundary condition. The steady
635: disk model considered here only accounts for half of the accretion
636: luminosity ($L = G \dot{M} M/(2 R_i)$) potentially available from
637: accretion onto a slowly rotating star. As \cite{kenyon89} showed
638: from ultraviolet spectra, there is no evidence in FU Ori for
639: boundary layer emission which would account for this missing
640: luminosity, up to $\sim$ 10-20\% of the total luminosity; and if our
641: lower estimate of extinction is correct, the potential boundary
642: layer emission would be much less. This leaves as possibilities that
643: the missing accretion energy is going into spinning up the central
644: star, expanding the central star, being radiated over a larger
645: radial distance of the disk, or some combination of all three
646: effects.
647:
648: Hartmann \& Kenyon (1985; also Kenyon \etal 1989)
649: pointed out that the 5 R$_{\odot}$ inner radius implied by steady disk models is
650: considerably larger than that of the typical T Tauri star ($R \sim 2 \rsun$), and suggested that this might be due to the
651: disk dumping large amounts of thermal energy into the star, expanding its
652: outer layers.
653:
654: Popham \etal (1993, 1996) suggested that the the missing boundary
655: layer energy could be distributed over a significant range of radii
656: in the disk at high accretion rates. They found that as $\dot{M}$
657: increases to $\sim$ 10$^{-4}$ M$_{\odot}$yr$^{-1}$ the dynamical
658: boundary layer (the radial extent of the region where $\Omega$ drops
659: from Keplerian angular velocity to 0) grows to 10-20 $\%$ of a
660: stellar radius and the thermal boundary layer (the radial width over
661: which the boundary layer luminosity is radiated) grows to the point
662: that it becomes impossible to distinguish the boundary layer from
663: the inner parts of the disk. Popham \etal (1996) showed that this
664: model would imply somewhat slower than Keplerian rotation in the
665: innermost disk. In
666: this paper we find no evidence for extra radiation if A$_{V}$=1.5.
667: We cannot rule out a slightly larger A$_{V}$, which would allow some
668: extra dissipation of kinetic energy in the inner disk. However, the
669: shape of the spectrum in the blue-optical region (Figure \ref{fig:f5}) suggests
670: %change \S 4 to figure 5 to be more specific
671: that the higher extinction value is not preferred, limiting the
672: amount of excess radiation to a modest fraction of the total
673: luminosity. We speculate that the missing energy is mostly going
674: into expanding the outer stellar layers, with perhaps a small amount
675: of excess radiation that is difficult to discern given the
676: uncertainties in the extinction.
677:
678: \subsection{$R_{out}$ and $\alpha$}
679:
680: Our estimate of the outer radius of the high state of FU Ori is around 200 R$_{\odot}$, although the
681: irradiation effect from the hot inner disk to the outer disk may decrease this value to no less than 100 $\rsun$.
682: \cite{bell94} assume low viscosity parameters ($\alpha$$\sim$10$^{-3}$)
683: and derive a small outer radius ( R$_{out}$$\sim$20R$_{\odot}$) for the high $\dot{M}$,
684: hot inner disk during outbursts. Our model of FU Ori is inconsistent with such a small hot region; instead, we require the high $\dot{M}$ region to be an order of magnitude larger in radius.
685:
686: FU Ori has been fading slowly in brightness over the last $\sim$~70 years \citep{ibrahimov99}.
687: If we attribute this fading to the emptying out of mass from the inner hot disk
688: onto the central star, this timescale gives us a way to constrain the viscosity in
689: this region. We make the simple assumption that the decay timescale is the viscous timescale
690: \begin{equation}
691: t_v \sim R^{2}/\nu\,,
692: \end{equation}
693: where the viscosity is
694: $\nu = \alpha c_{s}^{2}/\Omega$, $\alpha$ is the viscosity parameter (Shakura \&
695: Sunyaev 1973), $c_s$ is the sound speed in the disk midplane, and $\Omega$ is the
696: Keplerian angular frequency at $R$. We use the isothermal sound speed with a mean molecular
697: weight of 2.3, appropriate for molecular hydrogen. We then need an estimate of the central
698: temperature to evaluate $t_v$. Our radiative transfer
699: model shows that the effective temperature is $T_{eff} \sim 800$~K at the outer radius
700: of $R \sim 200 \rsun$. This is a lower limit for the central temperature as the absorption features in
701: the spectrum show that the central temperature is higher than the surface temperature.
702: The disk is likely to be very optically thick in this region,
703: such that the central temperature is considerably larger than the effective temperature
704: due to radiative trapping. We choose 1500 K as the central temperature
705: above which MRI is active and thus high $\alpha$ can sustain high mass accretion rate. We therefore find
706: \begin{equation}
707: t_v \sim 141\, {\rm yrs} \times \left (\frac{M}{0.3M_{\odot}}
708: \right)^{1/2} \left (\frac{R}{210R_{\odot}} \right )^{1/2} \left
709: (\frac{T}{1500 K} \right )^{-1} \left (\frac{\alpha}{10^{-1}}
710: \right )^{-1}\,. \label{eq:tvisc}
711: \end{equation}
712: where $M$ is the mass of the central star, $R$ is the outer radius
713: of the high mass accretion disk, $T$ is the central disk temperature
714: of the high mass accretion disk, and $\alpha$ is the viscosity
715: parameter.
716: %change Tcri to T, and add an explanation
717: If we estimate a typical decay timescale of FU Ori as $\sim 100$~yr, then we
718: find $\alpha \sim 0.14$. For the smaller hot disk radius with an
719: irradiated outer disk, our preliminary results yield (Zhu et al., in preparation)
720: R $\ge$0.5 AU and thus $\alpha \sim 0.1$.
721:
722: It is important to recognize that equation (\ref{eq:tvisc}) is only an order of magnitude
723: estimate. The decay timescale of accretion can be affected by whatever mechanism makes
724: the transition from high to low state; for example, in the thermal instability model,
725: the decay time may not be the timescale for emptying out half the material in the
726: affected region of the disk, but only that (smaller) amount necessary to shut off the
727: thermal instability. Nevertheless, it is suggestive that we find a larger value of
728: $\alpha$ in the high state than the $10^{-3}$ of \cite{bell94}.
729:
730: A value
731: of $\alpha \sim 0.2 - 0.02$ is roughly consistent with estimates from
732: compact systems with accretion disks \citep{king2007} as well as
733: with numerical simulations of the magnetorotational instability (MRI) in ionized
734: disks (Balbus \& Hawley 1988). In this connection, we note that the
735: requirement for enough thermal ionization to initiate a robust MRI is
736: a central temperature $T \gtrsim 1000$~K (Gammie 1996), comparable to
737: our observationally-based estimates ($T_{eff}$$\sim$800 K at $R$ $\sim$ 200 $\rsun$).
738: %add T$_{eff}$$\sim$800 K at $R \sim 200 \rsun$
739:
740: \subsection{Thermal and other instabilities}
741:
742: The thermal instability model was originally suggested to account for outbursts
743: in dwarf nova systems (Faulkner, Lin, \& Papaloizou 1983) and has attractive
744: properties for explaining FU Ori outbursts \citep{bell94}.
745: However, the observed large outer radius, 0.5 $-$ 1 AU,
746: of the high accretion rate portion of the FU Ori disk
747: poses difficulties for the pure thermal instability model.
748: Even in the situation explored by \cite{2004MNRAS.353..841L},
749: in which the thermal instability is triggered
750: by a massive planet in the inner disk, the outer radius
751: of the high state is still the same as in \cite{bell94}, $\sim$ 20 $\rsun$.
752: The relatively high temperatures ($2-4 \times 10^3$~K) required to trigger
753: the thermal instability (due to the ionization of hydrogen)
754: are difficult to achieve at large radii; they require large surface densities which in turn
755: produce large optical depths, trapping the radiation internally and making
756: the central temperature much higher than the surface (effective) temperature.
757:
758: To illustrate the problem, we calculated thermal equilibrium ``S''
759: curves (e.g., Faulkner \etal 1983) for our disk parameters at $R =
760: 210 \rsun$ but various values of $\alpha$. The equilibrium curve
761: represents energy balance between viscous energy generation and
762: radiative losses,
763: \begin{equation} F_{vis}=F_{rad}
764: \end{equation}
765: Thus,
766: \begin{equation}
767: \frac{9}{4}\alpha\Omega\Sigma c_{s}^{2}=2\sigma T_{eff}^{4}\,,
768: \end{equation}
769: where $\Sigma$ is the surface density and $c_{s}$ is the sound speed
770: at the local central disk. For an optically-thick disk, we have
771: T$_{c}^{4}$=$\frac{3}{8}$$\tau_{R}$T$_{eff}^{4}$, where
772: $\tau_{R}$=$\kappa_{R}$$\Sigma$ and $\kappa$ is the Rosseland mean
773: opacity \citep{hubeny90}.
774: %LH actually the standard estimate is T_c^4 = (3/8) \tau T_e^4; Hubeny 1990
775: Then,
776: \begin{equation}
777: T_{c}^{3}=\frac{27}{64\mu\sigma}\alpha\Omega\Sigma^{2} {\cal{R}}_{c} \kappa
778: \,,
779: \end{equation}
780: where $\cal{R}$$_{c}$ is the gas constant, and $\mu$ is the mean
781: molecular weight . As long as we know T$_{c}$, we can derive
782: T$_{eff}$ and the corresponding $\dot{M}$ at this radius. Because
783: the Rosseland mean opacity is also dependent on temperature and
784: pressure, iteration of the disk structure calculation is needed to
785: derive the equilibrium curve.
786:
787: We show in Figure \ref{fig:f10} the equilibrium curve at R=210 R$_{\odot}$ for
788: five values of $\alpha$: 10$^{-1}$, 10$^{-2}$, 10$^{-3}$, 10$^{-4}$, 10$^{-5}$.
789: In discussing our results, we define $\Sigma_{A}$ to be the
790: highest stable low state for a given $\alpha$.
791: If at any radius the surface density increases above
792: $\Sigma_{A}$, the disk can no longer stay stable on the
793: lower branch and strong local heating begins.
794: If the $\alpha$ of the low state is the same as the $\alpha$ of the high state,
795: then $\alpha$$\ge$10$^{-2}$; in this case, to trigger the thermal instability at
796: $\Sigma_{A}$, the mass accretion rate in the low state would need to be $\sim$10$^{-3}$
797: (see Figure \ref{fig:f10}), which is higher than the $\dot{M}$ of the high state
798: that we have determined. To have a mass accretion in the low state at least one order
799: of magnitude lower than that of the high state, we would need $\alpha$$\le$10$^{-4}$ for
800: %change 10^-5 to 10^-4
801: the low state (see Figure \ref{fig:f10}).
802: This is comparable to the low state $\alpha$ used by \cite{bell94},
803: but very much lower than our high state estimate. Changing the outer radius to R$\sim$100 $\rsun$
804: does not change these results by more than a factor of 10.
805:
806: With $\alpha$=10$^{-4}$, the critical surface density at 210 R$_{\odot}$
807: %change 10^-5 to 10^-4
808: is $\Sigma_A \sim 10^{6}$ g cm$^{-2}$. Then we can estimate the Toomre
809: $Q$ parameter,
810: \begin{equation}
811: Q = {c_s \Omega \over \pi G \Sigma_A} \sim 0.1 \,,
812: \end{equation}
813: where $Q < 1$ implies gravitational instability. Thus, a thermal instability model
814: for FU Ori implies a massive, probably gravitationally-unstable disk, which
815: raises the question: could the outburst be driven by gravity instead?
816:
817: \cite{armitage01} proposed an alternative model of disk outbursts
818: in which material piles up at radii of order 1-2 AU, consistent with layered
819: disk models \citep{gammie96}, achieving repeated outbursts with high accretion
820: on size scales much closer to our inferred $R_{out}$. In this model, gravitational
821: torques lead to increased accretion, which in turn heats up the inner disk
822: until temperatures are high enough (800 K in their model) to trigger the MRI,
823: which results in very much higher viscosities and rapid accretion. The \cite{armitage01}
824: model does not give exactly FU Ori-type outbursts; instead, the accretion events
825: last for $\sim 10^{4}$ yr and achieve peak accretion rates of only $\sim 10^{-5} \msunyr$.
826: However, \cite{armitage01} suggested that thermal instabilities might be triggered
827: by their gravitationally-driven outbursts. Simple models with somewhat different
828: parameters demonstrate that it is indeed possible to trigger thermal instabilities
829: in the context of this general model (Gammie, Book, \& Hartmann 2007, in preparation),
830: resulting in outbursts more similar to that of FU Ori.
831:
832: Vorobyov \& Basu (2005,2006) suggest FU Ori outburst could be
833: explained by the accretion of clumps formed in a gravitationally
834: unstable disk. The disk becomes gravitationally unstable because of
835: the infall of the matter from the envelope. The spiral arms and
836: clumps form and grow in the unstable disk. The redistribution of
837: mass and angular momentum leads to centrifugal disbalance in the
838: disk and further triggers the outburst when the dense clumps are
839: driven into the central star. The authors derive a mass accretion
840: rate at 10 AU of $\sim$10$^{-4}$M$_{\odot}$/yr, which could result in
841: too much emission from the outer disk; this issue requires further
842: exploration.
843:
844: The rapid rise time of the outburst of FU Ori in the B band, $\sim$ 1 year,
845: provides the best evidence for a thermal instability.
846: We can test this idea very crudely by using our steady disk model to
847: estimate the range of radii in the disk which can contribute to the B band
848: magnitude in the light curve, finding that only the inner
849: $R \lesssim 20 \rsun$ disk are involved.
850: If we then assumed that the rise in B light is due to the inward
851: propagation of a thermal instability front across this region at a
852: (maximum) speed $\sim \alpha c_s$ \citep{lin85,bell95}, and
853: assume a sound speed characteristic of $2-4 \times 10^3$~K
854: (the minimum temperature for instability), then we
855: derive $\alpha \sim 0.1$.
856: While this calculation is very crude, it does suggest consistency with the
857: idea of a thermal instability with
858: a viscosity parameter comparable to that which we inferred from the
859: decaying light curve (\S 5.2).
860:
861: Finally, we estimate the amount of mass in the inner disk.
862: An upper mass limit for the inner disk can be estimated by assuming Q
863: $\sim 1$ (the approximate limit for strong gravitational instability)
864: at $R$ $\sim$ 200 $\rsun$. Adopting a central temperature $T \sim 1000 $~K,
865: the gravitational instability limit leads to an approximate surface density
866: limit $\Sigma \lesssim 10^{5} g cm^{-2}$ and thus a maximum inner disk mass
867: $\sim \pi\Sigma R^{2} \sim 0.05 \msun$. This leads us to an inner disk mass
868: estimate $0.05 \msun \gtrsim M_{in} \gtrsim 0.01 \msun$, where the lower limit
869: comes from requiring the outburst mass accretion rate of $\sim 10^{-4} \msunyr$
870: to be sustained for the observed decay timescale of $\sim 100$~yr.
871: %add this paragraph
872:
873:
874: \section{Conclusions}
875:
876: Using the latest opacities and the ODF method, we have constructed a
877: new detailed radiative transfer disk model which reproduces the main
878: features of the FU Ori spectrum from $\sim$4000$\dot{A}$ to about 8
879: $\mu$m. The SED of FU Ori can be fitted with a steady disk model and
880: no boundary layer emission if A$_{V}$$\sim$1.5, a somewhat
881: lower valued extinction than
882: previously estimated. A larger A$_{V}$ would imply extra heating at
883: small radii which would result from dissipation of kinetic energy as
884: material accretes onto the central star. With A$_{V}$=1.5, the
885: inclination estimated from near-IR interferometry and the
886: observed rotation of the inner disk, we estimate that the central
887: star has a mass $\sim 0.3 M_{\odot}$ (typical of low-mass TTS) and a disk mass
888: accretion rate of $\sim 2.4 \times 10 ^{-4}\msunyr$.
889:
890: Relying on the Infrared Spectrograph Spectrum of FU Ori from
891: the Spitzer Space Telescope presented by \cite{jgreen06}, we estimate that
892: the outer radius of the hot, rapidly accreting region of the inner disk
893: is $\sim$1 AU, although the irradiation effect may decrease this value by
894: no more than a factor of two(Zhu et al., in preparation).
895: Either way, this is inconsistent with the pure thermal instability
896: models of \cite{bell94} who adopted a low $\alpha$ viscosity parameter
897: $\sim$10$^{-3}$-10$^{-4}$.
898: If we assume that the observed decay timescale of
899: FU Ori ($\sim$100 years) is the viscous transport time from the outer edge of the hot region,
900: then we derive $\alpha \sim 0.2 - 0.02$, , comparable to the predictions of simulations
901: of the magnetohydrodynamic (MHD) turbulence. The effective temperature
902: at the outer edge of the hot region in our model is $\sim$800 K
903: ; the presence of absorption features shows that the central disk temperature must be
904: higher. Thus, our model suggests that the central disk temperature is greater than
905: the T$_{c}$$\sim$1000 K required to maintain full MHD turbulence by thermal
906: ionization out to $\sim$ 1 AU.
907:
908: We show that pure thermal instability models
909: have difficulty in explaining the outburst of FU Ori, and suggest that models
910: including gravitational torques and MRI activation,
911: such as those of \cite{armitage01}, are more promising.
912:
913: \acknowledgments We thank Dr. R. Kurucz and Dr. F. Castelli for making
914: their line list and ATLAS codes available. They also provide us very
915: helpful suggestions to help us run their codes smoothly. We also
916: thank Mansur Ibrahimov for communicating recent photometry of FU Ori
917: and Thomas P. Greene for providing the near-infrared spectrum of FU
918: Ori. This work was supported in part by NASA grant NNG06GJ32G and the
919: University of Michigan.
920:
921: \begin{thebibliography}
922:
923: \bibitem[Alexander \& Ferguson(1994)]{alexander94} Alexander,
924: D.~R., \& Ferguson, J.~W.\ 1994, \apj, 437, 879
925:
926: \bibitem[Armitage et al.(2001)]{armitage01} Armitage, P.~J., Livio, M., \& Pringle, J.~E.\ 2001, \mnras, 324, 705
927:
928: \bibitem[Aspin \& Reipurth(2003)]{aspin03} Aspin, C., \&
929: Reipurth, B.\ 2003, \aj, 126, 2936
930:
931: \bibitem[Balbus \& Hawley(1998)]{1998RvMP...70....1B} Balbus, S.~A., \&
932: Hawley, J.~F.\ 1998, Reviews of Modern Physics, 70, 1
933:
934: \bibitem[Bell \& Lin(1994)]{bell94} Bell, K.~R., \& Lin,
935: D.~N.~C.\ 1994, \apj, 427, 987
936:
937: \bibitem[Bell et al.(1995)]{bell95} Bell, K.~R., Lin,
938: D.~N.~C., Hartmann, L.~W., \& Kenyon, S.~J.\ 1995, \apj, 444, 376
939:
940: \bibitem[Calvet et al.(1991b)]{calvet91b} Calvet, N., Hartmann,
941: L., \& Kenyon, S.~J.\ 1991, \apj, 383, 752
942:
943:
944: \bibitem[Calvet et al.(1991a)]{calvet91} Calvet, N., Patino, A.,
945: Magris, G.~C., \& D'Alessio, P.\ 1991, \apj, 380, 617
946:
947: \bibitem[Castelli(2005)]{castelli05} Castelli, F.\ 2005, Memorie
948: della Societa Astronomica Italiana Supplement, 8, 34
949:
950: \bibitem[Castelli \& Kurucz(2004)]{kurucz04} Castelli, F., \&
951: Kurucz, R.~L.\ 2004, ArXiv Astrophysics e-prints, arXiv:astro-ph/0405087
952:
953: \bibitem[Clarke et al.(2005)]{clarke05} Clarke, C., Lodato, G.,
954: Melnikov, S.~Y., \& Ibrahimov, M.~A.\ 2005, \mnras, 361, 942
955:
956: \bibitem[Cohen et al.(2003)]{cohen03} Cohen, M., Wheaton,
957: W.~A., \& Megeath, S.~T.\ 2003, \aj, 126, 1090
958:
959: \bibitem[Cohen \& Woolf(1971)]{cohen71} Cohen, M., \& Woolf,
960: N.~J.\ 1971, \apj, 169, 543
961:
962: \bibitem[Draine \& Lee(1984)]{draine84} Draine, B.~T., \& Lee,
963: H.~M.\ 1984, \apj, 285, 89
964:
965: \bibitem[Draine \& Lee(1987)]{draine87} Draine, B.~T., \& Lee,
966: H.~M.\ 1987, \apj, 318, 485
967:
968: \bibitem[Fabricant et al.(1998)]{fabricant98} Fabricant, D.,
969: Cheimets, P., Caldwell, N., \& Geary, J.\ 1998, \pasp, 110, 79
970:
971: \bibitem[Faulkner et al.(1983)]{1983MNRAS.205..359F} Faulkner, J., Lin,
972: D.~N.~C., \& Papaloizou, J.\ 1983, \mnras, 205, 359
973:
974: \bibitem[Ferguson et al.(2005)]{ferguson05} Ferguson, J.~W.,
975: Alexander, D.~R., Allard, F., Barman, T., Bodnarik, J.~G., Hauschildt,
976: P.~H., Heffner-Wong, A., \& Tamanai, A.\ 2005, \apj, 623, 585
977:
978: \bibitem[Gammie(1996)]{gammie96} Gammie, C.~F.\ 1996, \apj, 457,
979: 355
980:
981: \bibitem[Gammie(1999)]{gammie99} Gammie, C.~F.\ 1999, ASP
982: Conf.~Ser.~160: Astrophysical Discs - an EC Summer School, 160, 122
983:
984: \bibitem[Green et al.(2006)]{jgreen06} Green, J.~D., Hartmann,
985: L., Calvet, N., Watson, D.~M., Ibrahimov, M., Furlan, E., Sargent, B., \&
986: Forrest, W.~J.\ 2006, \apj, 648, 1099
987:
988: \bibitem[Greene \& Lada(1996)]{greene96} Greene, T.~P., \& Lada,
989: C.~J.\ 1996, \aj, 112, 2184
990:
991: \bibitem[Goodrich(1987)]{goodrich87} Goodrich, R.~W.\ 1987, \pasp,
992: 99, 116
993:
994: \bibitem[Hartmann et al.(2004)]{lee04} Hartmann, L., Hinkle,
995: K., \& Calvet, N.\ 2004, \apj, 609, 906
996:
997:
998: \bibitem[h (1985)]{hk85} Hartmann, L., \& Kenyon, S.~J.\ 1985, \apj, 299, 462
999:
1000: \bibitem[h (1987)a]{hk87a} Hartmann, L., \& Kenyon, S.~J.\ 1987a, \apj, 312, 243
1001:
1002: \bibitem[h (1987)b]{hk87b} Hartmann, L., \&
1003: Kenyon, S.~J.\ 1987b, \apj, 312, 243
1004:
1005: \bibitem[Hartmann \& Kenyon(1991)]{HK91} Hartmann, L., \&
1006: Kenyon, S.~J.\ 1991, IAU Colloq.~129: The 6th Institute d'Astrophysique de
1007: Paris (IAP) Meeting: Structure and Emission Properties of Accretion Disks,
1008: 203
1009:
1010: \bibitem[Hartmann \& Kenyon(1996)]{Lee96} Hartmann, L., \&
1011: Kenyon, S.~J.\ 1996, \araa, 34, 207
1012:
1013: \bibitem[Herbig(1977)]{herbig77} Herbig, G.~H.\ 1977, \apj, 217,
1014: 693
1015:
1016: \bibitem[Hernandez(2005)]{jesusphd} Hern\'{a}ndez, J.\ 2005, Ph.D. Thesis, Postgrado de F\'{\i}sica Fundamental, Universidad
1017: de Los Andes, Venezuela.
1018:
1019: \bibitem[Hern{\'a}ndez et al.(2004)]{jesus04} Hern{\'a}ndez,
1020: J., Calvet, N., Brice{\~n}o, C., Hartmann, L., \& Berlind, P.\ 2004, \aj,
1021: 127, 1682
1022:
1023: \bibitem[Hubeny(1990)]{hubeny90} Hubeny, I.\ 1990, \apj, 351,
1024: 632
1025:
1026: \bibitem[Ibrahimov(1999)]{ibrahimov99} Ibrahimov, M.~A.\ 1999,
1027: Informational Bulletin on Variable Stars, 4691, 1
1028:
1029: \bibitem[Jacoby et al.(1984)]{jacoby84} Jacoby, G.~H., Hunter,
1030: D.~A., \& Christian, C.~A.\ 1984, \apjs, 56, 257
1031:
1032: \bibitem[Kenyon \& Hartmann(1991)]{kenyon91} Kenyon, S.~J., \&
1033: Hartmann, L.~W.\ 1991, \apj, 383, 664
1034:
1035: \bibitem[Kenyon et al.(1993)]{kenyon93} Kenyon, S.~J., Hartmann,
1036: L., Gomez, M., Carr, J.~S., \& Tokunaga, A.\ 1993, \aj, 105, 1505
1037:
1038: \bibitem[Kenyon et al.(1988)]{kenyon88} Kenyon, S.~J., Hartmann,
1039: L., \& Hewett, R.\ 1988, \apj, 325, 231 (KHH88)
1040:
1041: \bibitem[Kenyon et al.(1989)]{kenyon89} Kenyon, S.~J., Hartmann,
1042: L., Imhoff, C.~L., \& Cassatella, A.\ 1989, \apj, 344, 925
1043:
1044: \bibitem[King et al.(2007)]{king2007} King, A.~R., Pringle,
1045: J.~E., \& Livio, M.\ 2007, \mnras, 376, 1740
1046:
1047: \bibitem[Kurucz(1993)]{kurucz93} Kurucz, R.~L.\ 1993, Kurucz
1048: CD-ROM, Cambridge, MA: Smithsonian Astrophysical Observatory, |c1993,
1049: December 4, 1993,
1050:
1051: \bibitem[Kurucz(2005)]{kurucz05} Kurucz, R.~L.\ 2005, Memorie
1052: della Societa Astronomica Italiana Supplement, 8, 86
1053:
1054: \bibitem[Kurucz \& Avrett(1981)]{kurucz81} Kurucz, R.~L., \&
1055: Avrett, E.~H.\ 1981, SAO Special Report, 391,
1056:
1057: \bibitem[Kurucz et al.(1974)]{kurucz74} Kurucz, R.~L.,
1058: Peytremann, E., \& Avrett, E.~H.\ 1974, Washington : Smithsonian
1059: Institution : for sale by the Supt.~of Docs., U.S.~Govt.~Print.~Off.,
1060: 1974., 37
1061:
1062: \bibitem[Lin et al.(1985)]{lin85} Lin, D.~N.~C., Faulkner,
1063: J., \& Papaloizou, J.\ 1985, \mnras, 212, 105
1064:
1065: \bibitem[Lodato \& Clarke(2004)]{2004MNRAS.353..841L} Lodato, G., \&
1066: Clarke, C.~J.\ 2004, \mnras, 353, 841
1067:
1068: \bibitem[Malbet et al.(2005)]{malbet05} Malbet, F., et al.\
1069: 2005, \aap, 437, 627
1070:
1071: \bibitem[Mould et al.(1978)]{mould78} Mould, J.~R., Hall,
1072: D.~N.~B., Ridgway, S.~T., Hintzen, P., \& Aaronson, M.\ 1978, \apjl, 222,
1073: L123
1074:
1075: \bibitem[Muzerolle et al.(2003)]{muzerolle03} Muzerolle, J.,
1076: Calvet, N., Hartmann, L., \& D'Alessio, P.\ 2003, \apjl, 597, L149
1077:
1078: \bibitem[Papaloizou et al.(1983)]{papaloizou83} Papaloizou, J.,
1079: Faulkner, J., \& Lin, D.~N.~C.\ 1983, \mnras, 205, 487
1080:
1081: \bibitem[Partridge \& Schwenke(1997)]{PS97} Partridge, H.,
1082: \& Schwenke, D.~W.\ 1997, \jcp, 106, 4618
1083:
1084: \bibitem[Popham et al.(1993)]{popham93} Popham, R., Narayan, R.,
1085: Hartmann, L., \& Kenyon, S.\ 1993, \apjl, 415,
1086:
1087: \bibitem[Popham et al.(1996)]{1996ApJ...473..422P} Popham, R., Kenyon, S.,
1088: Hartmann, L., \& Narayan, R.\ 1996, \apj, 473, 422
1089:
1090: \bibitem[Quanz et al.(2006)]{quanz2006} Quanz, S.~P., Henning,
1091: T., Bouwman, J., Ratzka, T., \& Leinert, C.\ 2006, \apj, 648, 472L127
1092:
1093: \bibitem[Rayner et al.(1998)]{rayner98} Rayner, J.~T., Toomey,
1094: D.~W., Onaka, P.~M., Denault, A.~J., Stahlberger, W.~E., Watanabe, D.~Y.,
1095: \& Wang, S.-I.\ 1998, \procspie, 3354, 468
1096:
1097: \bibitem[Reipurth \& Aspin(1997)]{reipurth97} Reipurth, B., \&
1098: Aspin, C.\ 1997, \aj, 114, 2700
1099:
1100: \bibitem[Reipurth \& Aspin(2004)]{reipurth2004} Reipurth, B., \&
1101: Aspin, C.\ 2004, \apjl, 608, L65
1102:
1103: \bibitem[Sandell \& Aspin(1998)]{sandell98} Sandell, G., \&
1104: Aspin, C.\ 1998, \aap, 333, 1016
1105:
1106: \bibitem[Sandell \& Weintraub(2001)]{2001ApJS..134..115S} Sandell, G., \&
1107: Weintraub, D.~A.\ 2001, \apjs, 134, 115
1108:
1109: \bibitem[Sbordone et al.(2004)]{kurucz042} Sbordone, L.,
1110: Bonifacio, P., Castelli, F., \& Kurucz, R.~L.\ 2004, Memorie della Societa
1111: Astronomica Italiana Supplement, 5, 93
1112:
1113: \bibitem[Schwenke(1998)]{S98} Schwenke, D.~W.\ 1998,
1114: Chemistry and Physics of Molecules and Grains in Space.~ Faraday
1115: Discussions No.~109, 321
1116:
1117: \bibitem[Shakura \& Sunyaev(1973)]{SS1973} Shakura, N.~I., \&
1118: Sunyaev, R.~A.\ 1973, \aap, 24, 337
1119:
1120: \bibitem[Simon et al.(2000)]{2000ApJ...545.1034S} Simon, M., Dutrey, A., \&
1121: Guilloteau, S.\ 2000, \apj, 545, 1034
1122:
1123: \bibitem[Turner et al.(1997)]{turner97} Turner, N.~J.~J.,
1124: Bodenheimer, P., \& Bell, K.~R.\ 1997, \apj, 480, 754
1125:
1126: \bibitem[Vorobyov \& Basu(2005)]{2005ApJ...633L.137V} Vorobyov, E.~I., \&
1127: Basu, S.\ 2005, \apjl, 633, L137
1128:
1129: \bibitem[Vorobyov \& Basu(2006)]{2006ApJ...650..956V} Vorobyov, E.~I., \&
1130: Basu, S.\ 2006, \apj, 650, 956
1131:
1132:
1133: \end{thebibliography}
1134:
1135:
1136: \clearpage
1137: \begin{table}
1138: \begin{center}
1139: \caption{Best fit model \label{tab1}}
1140: \begin{tabular}{clllll}
1141:
1142: \tableline\tableline
1143: M$_{star}$/M$_{\odot}$ & $\dot{M}$ & R$_{i}$/R$_{\odot}$ & R$_{out}$/R$_{\odot}$ & $\theta$\tablenotemark{a} & Distance/pc \\
1144: \tableline
1145: 0.3 &2.4$\times$10$^{-4}$M$_{\odot}$/yr &5 &210 &55$^{o}$ &500 \\
1146: \tableline
1147: \end{tabular}
1148: \tablenotetext{a}{Inclination angle of the disk from \cite{malbet05}}
1149: \end{center}
1150: \end{table}
1151: \clearpage
1152:
1153: \begin{table}
1154: \begin{center}
1155: \caption{Optical lines used to determine the spectral type \label{tab2}}
1156: \begin{tabular}{clcl}
1157: \tableline\tableline
1158: Lines \tablenotemark{a}& Metals & Spectral type \\
1159: \tableline
1160: 4047.00 & Fe I+Sc I & F8 \\
1161: 4226.00 & Ca I & G1 \\
1162: 4271.00 & Fe I & G4 \\
1163: 4305.00 & CH(Gband) & G1 \\
1164: 4458.00 & Mn I+Fe I & G1 \\
1165: 5329.00 & Fe I & G2 \\
1166: 5404.00 & Fe I & G3 \\
1167: 6162.00 & Ca I+TiO & F9 \\
1168: \tableline
1169: \end{tabular}
1170: \tablenotetext{a}{From \cite{jesusphd}}
1171: \end{center}
1172: \end{table}
1173:
1174: \clearpage
1175:
1176: \begin{figure}
1177: \epsscale{.80} \plotone{f1.eps} \caption{The spectral energy distribution of FU Ori.
1178: Collected photometric and spectroscopic data of FU Ori with A$_{v}$=1.5 (see \S 4).
1179: The solid circles with error bars are the UBVR photometry from the Maidanak Observatory,
1180: the squares are 2MASS JHK photometry, and the open circles are JHK'L' photometry from \cite{reipurth2004}.
1181: The dashed curve is the FAST spectrum from 3650 \AA \ to 7500 \AA,
1182: the dotted curve is the KSPEC spectrum from 1.15 $\mu$m to 2.42 $\mu$m \citep{greene96},
1183: and the solid curve is the SpeX spectrum from 2.1 $\mu$m to 4.8 $\mu$m
1184: \citep{muzerolle03}, both scaled to the 2MASS photometry.
1185: Finally, the dot-dash curve is the IRS spectrum from 5 $\mu$m to 35 $\mu$m
1186: from \cite{jgreen06}. The spectra were put on as absolute flux scale as discussed in
1187: \S 2. }
1188: \label{fig:1}
1189: \end{figure}
1190:
1191: \clearpage
1192: \begin{figure}
1193: \epsscale{.80} \plotone{f2.eps} \caption{Our calculated Rosseland mean opacity
1194: as a function of temperature for solar composition. Each curve is
1195: labeled with the value of log($\rho /T_6^3$) to be compared
1196: with \cite{alexander94}, where $T_6$ is the temperature in millions of degrees. }
1197: \label{fig:ross}
1198: \end{figure}
1199:
1200: \begin{figure}
1201: \epsscale{.80} \plotone{f3.eps} \caption{The spectrum predicted by
1202: the annulus with T$_{eff}$=5300 K (solid line) and the spectrum of
1203: SAO 21446 which is a G1 I star \citep{jacoby84} ( dotted line).
1204: The spectra are scaled to the same flux at 7000 \AA. }
1205: \label{fig:f3}
1206: \end{figure}
1207:
1208: \begin{figure}
1209: \epsscale{.80}
1210: \plotone{f4.eps}
1211: \caption{Emergent intensities from selected annuli of the disk model calculated for a 55 degree emergent angle.
1212: %add 55 degree emergent angle
1213: Effective temperatures of the annuli at 1, 3, 5, 6, 8, 10, 15, 20, 25, and 30 R$_{i}$
1214: are 6420, 4660, 3400, 2480, 2130, 1630, 1310, 1110, and 977 K, respectively.
1215: Molecular absorption features in the near-IR arise from cool regions of the disk,
1216: which do not contribute to the optical spectrum.}
1217: \label{fig:f4}
1218: \end{figure}
1219:
1220:
1221: \begin{figure}
1222: \epsscale{.80}
1223: \plotone{f5.eps}
1224: \caption{Observed UBVR photometry and optical spectra
1225: with different A$_{V}$: 2.2,1.9,1.5,1.3 (dotted lines, from top to bottom)
1226: and model spectra with T$_{max}$: 7240, 6770, 6420, 5840
1227: (solid lines, from top to bottom). The total flux spectra are produced by
1228: weighting the intensities (Figure 4) by the appropriate annular surface areas.
1229: %add the last sentence
1230: } \label{fig:f5}
1231: \end{figure}
1232:
1233: \begin{figure}
1234: \epsscale{.80} \plotone{f6.eps} \caption{Comparison between
1235: observed data and disk models for different values of the outer
1236: radii. All the black lines and dots are observed spectra and
1237: photometry data as shown in figure \ref{fig:1}. All the color lines
1238: are model spectra. The parameters of the disk models are
1239: inner disk radius 5 R$_{\odot}$, and $M \dot{M}$$\sim$
1240: 7.2$\times$10$^{-5}$M$_{\odot}$/yr. The corresponding T$_{max}$ is
1241: 6420 K.
1242: The four model
1243: spectra from top to bottom are for the disk models truncated at 5000
1244: R$_{\odot}$, 210 R$_{\odot}$, 50 R$_{\odot}$ and 20 R$_{\odot}$. The
1245: solid lines are model spectra for a high mass accretion rate region
1246: with outer radii as shown. The dotted lines are spectra for models
1247: where an outer region with mass accretion one order of magnitude
1248: smaller has been added. For the disk model truncated at 5000
1249: R$_{\odot}$, the dotted line is coincident with the solid line.
1250: The extinction parameter is $A_{V}$=1.5.
1251: %change the caption
1252: } \label{fig:f6}
1253: \end{figure}
1254:
1255: \begin{figure}
1256: \epsscale{.80} \plotone{f7.eps} \caption{Two truncated models
1257: fitting the observations corrected by A$_{V}$=2.2 and A$_{V}$=1.5.
1258: Only optical photometries and IRS spectra (5-30 microns) are shown.
1259: The square points are the photometries corrected by A$_{V}$=2.2,
1260: while the round points are the photometries corrected by
1261: A$_{V}$=1.5. The IRS spectra corrected by these two A$_{V}$ are
1262: coincident. The solid line is the model fitting to observations with
1263: A$_{V}$=2.2, while the dotted line is the model fitting to
1264: observations with A$_{V}$=1.5. Both of the two models have the same
1265: outer radius of the high state R$_{out}$=210 R$_{\odot}$.}
1266: \label{fig:7}
1267: % change the caption
1268: \end{figure}
1269:
1270: \begin{figure}
1271: \epsscale{.80}
1272: \plotone{f8.eps}
1273: %LH set v = v(turb) and i = 55 in the figure labeling
1274: \caption{CO first overtone absorption band around 2.3 $\mu$m.
1275: The solid line is the observed CO first overtone absorption
1276: feature from \cite{lee04} and the dotted line is the model spectrum after being
1277: convolved to the same spectrum resolution. The turbulent velocity is 4 ${\rm km \, s^{-1}}$ and the inclination angle is 55 degrees.} \label{fig:CO}
1278: \end{figure}
1279:
1280:
1281:
1282: \begin{figure}
1283: \epsscale{.80} \plotone{f9.eps} \caption{The mid-infrared part of
1284: the observed spectra in Figure \ref{fig:1} compared with the model
1285: spectra. The solid line is the model with R$_{out}$=210 R$_{\odot}$
1286: and the dotted line is the model with 1 \% dust opacity (dust
1287: opacity
1288: reduced by a factor of 100) and R$_{out}$=225
1289: R$_{\odot}$.
1290: %LH
1291: We do not have the CO fundamental opacities in our model and thus
1292: cannot reproduce these strong spectral features at 4.6 $\mu$m.}
1293: \label{fig:f9}
1294: \end{figure}
1295:
1296: \begin{figure}
1297: \epsscale{1.00} \plotone{f10.eps} \caption{ The equilibrium curves
1298: for thermal-instability at R=210 R$_{\odot}$ with five values of
1299: $\alpha$: 10$^{-1}$,10$^{-2}$,10$^{-3}$,10$^{-4}$,10$^{-5}$(from top
1300: to bottom).
1301: %LH Zhaohuan, read the following and check/fix
1302: The dotted line is $\dot{M}=2.4\times10^{-5}M_{\odot}/yr$, set to be
1303: one order of magnitude lower accretion rate than the accretion rate
1304: of the inner disk. If the outer disk $\dot{M}$ were one order of
1305: magnitude higher than this (so that the outer disk and inner disk
1306: accretion rates are the same) the mid-infrared flux would
1307: %change the caption to make our explanation more clear
1308: be much larger than observed (see discussion in \S 4). The open
1309: circles denote the critical surface densities $\Sigma_{A}$ for
1310: triggering the thermal instability at each value of $\alpha$ (see \S
1311: 5.3) } \label{fig:f10}
1312: \end{figure}
1313:
1314:
1315: \end{document}
1316: