1: \documentclass[11pt,preprint2]{emulateapj}
2:
3: %\documentclass[12pt,preprint]{aastex}
4:
5: % preprint preprint2 or manuscript
6: %\documentclass[article]{aa}
7: %\usepackage{epsfig}
8: %\usepackage{epsfig,natbib}
9: \shorttitle{The molecular envelope of VY~CMa}
10: \shortauthors{Muller et al. 2006}
11:
12: \begin{document}
13:
14: \title{The molecular envelope around the red supergiant VY~CMa}
15:
16: \author{Muller S., % \altaffilmark{1},
17: Dinh-V-Trung, % \altaffilmark{1},
18: Lim J., % \altaffilmark{1},
19: Hirano N.} % \altaffilmark{1}
20: \affil{Academia Sinica Institute of Astronomy and Astrophysics (ASIAA), P.O. Box 23-141, Taipei, 106 Taiwan}
21: \email{muller, trung, jlim, hirano@asiaa.sinica.edu.tw}
22: \author{Muthu C.} % \altaffilmark{2}}
23: \affil{Aryabhatta Research Institute of Observational Sciences (ARIES), Manora Peak, Nainital 263129, India}
24: \email{muthu@aries.ernet.in}
25: \and
26: \author{Kwok S.} % \altaffilmark{3}
27: \affil{Department of Physics, Faculty of Science, University of Hong Kong}
28: \email{sunkwok@hkucc.hku.hk}
29:
30: \begin{abstract}
31: We present millimeter interferometric observations of the molecular envelope around the red
32: supergiant \object{VY~CMa} with the SubMillimeter Array (SMA)
33: \footnote{The Submillimeter Array is a joint project between the Smithsonian Astrophysical Observatory
34: and the Academia Sinica Institute of Astronomy and Astrophysics and is funded by the Smithsonian
35: Institution and the Academia Sinica.}. The high angular resolution
36: ($<$ 2$\arcsec$) allows us to derive the structure of the envelope as observed in the 1.3 mm
37: continuum, $^{12}$CO(2-1), $^{13}$CO(2-1) and SO(6$_5$-5$_4$) lines emission. The circumstellar
38: envelope is resolved into three components: a dense, compact and dusty central component,
39: embedded in a more diffuse and extended envelope plus a high velocity component. We construct
40: a simple model, consisting of a spherically symmetric slowly expanding envelope and bipolar
41: outflows with a wide opening angle ($\sim$ 120$\degr$) viewed close to the line of sight
42: (i = 15$\degr$). Our model can explain the main features of the SMA data and previous
43: single-dish CO multi-line observations. An episode of enhanced mass loss along the bipolar
44: direction is inferred from our modelling. The SMA data provide a better understanding of the
45: complicated morphology seen in the optical/IR high resolution observations.
46:
47: \end{abstract}
48:
49: \keywords{circumstellar matter -- supergiants -- stars: individual (VY Canis Majoris) -- radio lines: stars}
50:
51: \section{Introduction}
52:
53: The terminal stage in the evolution of massive stars (M $>$ 10 M$_\odot$) is characterized by
54: very short timescales ($\sim$ 10$^4$ yr) and drastic changes in their immediate
55: circumstellar environment. In the red supergiant phase, periods of large mass loss release
56: a thick circumstellar envelope of molecular gas and dust. Understanding the mass loss process
57: from red supergiants is extremely important in that it will eventually govern their evolution
58: as well as the structure of their surrounding envelope. In a galactic context, supergiants
59: contribute to the enrichment of the interstellar medium in dust and heavy elements and participate
60: to the global galactic chemical evolution.
61:
62: VY Canis Majoris (VY~CMa) is one of the brightest known red supergiants. Its distance has long
63: been debated \citep{her70, lad78}, but there is now a consensus on a value of 1.5 kpc,
64: consistent with the velocity and proper motions of H$_2$O masers \citep{ric98}. The stellar
65: luminosity is estimated to $\sim$ 2 10$^5$ L$_\odot$ \citep{mon99}. The stellar
66: mass is thought to be $\ge$ 15 M$_\odot$. The basic properties of VY~CMa are summarized in Table 1.
67:
68: VY~CMa is characterized by an apparent very high mass loss rate, of order 2 -- 4 10$^{-4}$
69: M$_\odot$/yr (see {\em e.g.} \citealt{bow83, dan94, sta95, har01}), which produced a very
70: thick circumstellar envelope of dust and molecules, which almost completely obscurs the
71: central star. The optical/infrared morpholgy of the nebulosity around VY~CMa is very complex,
72: showing arcs, filaments, bright knots in a very complicated arrangement \citep{smi01, mon99}.
73: Long slit spectroscopy \citep{smi04, hum05} reveals a very complex kinematics with evidence
74: for localized and episodic mass loss events. \citet{hum05} showed that the prominent arcs
75: are expanding relative to the star and represent discrete ejection events. The ejection dates
76: are ranging from 200 to 1000 years ago and state that VY~CMa has had a period of major activity
77: during the past 1000 years. These works however rely on scattered lights which may reflect only
78: the illuminated part of the nebula and the origin of the features seen in the optical and near
79: infrared images is very uncertain.
80:
81: We note that in a recent paper, \citet{dec06} propose a theoretical model, based on a spherically
82: symmetric envelope, that requires time varying mass-loss rates to reproduce previous single-dish CO
83: multi-line spectra. The high angular resolution view offered by the SubMillimeter Array (SMA) now
84: provides new constraints on the structure and kinematics of the circumstellar molecular envelope
85: of this peculiar object. In this paper, we present our SMA observations and discuss about the
86: morphology and kinematics of the molecular envelope around VY~CMa. We propose a simple model to
87: explain our data and compare with observations at other wavelengths.
88:
89: \section{Observations}
90:
91: We obtained 1.3 mm interferometric observations of VY~CMa with the SubMillimeter Array (SMA,
92: \citealt{ho04}), which consists of eight 6 m antennas, located on top of the Mauna Kea, Hawaii.
93: The phase reference of the array was set to
94: R.A. = 07$^{\rm h}$22$^{\rm m}$58$\fs$328 and Dec. = $-$25$\degr$46$\arcmin$03$\farcs$19.
95:
96: VY~CMa was observed at 2 different epochs and with 2 different configurations. The first
97: observation was carried out on 2004 July 31$^{\rm st}$, with 7 antennas in a compact configuration,
98: including projected baselines from 8 to 99 m. The system temperatures at 1 mm were between 100
99: and 200 K for the different antennas. In order to improve the angular resolution, a second
100: track was observed on 2005 January 19$^{\rm th}$, with 7 antennas in an extended configuration.
101: The projected baselines ranged from 18 to 192 m. The system temperatures were between 200 and
102: 400 K. The bandpass was calibrated using Uranus and/or Jupiter, while Callisto set the flux
103: calibration. The nearby quasars 0607$-$157 and 0730$-$116 were regularly observed for the gain
104: calibration. After calibration, the visibilities from both tracks have been combined together.
105: The data reduction and calibration were done under MIR/IDL. Imaging and data analysis were
106: entirely processed under GILDAS\footnote{http://www.iram.fr/IRAMFR/GILDAS/}.
107:
108: The heterodyne receivers were tuned to allow multi-line observations including the
109: $^{12}$CO(2-1) line in the upper side band (USB) and the $^{13}$CO(2-1) and SO(6$_5$-5$_4$)
110: lines in the lower side band (LSB). The SMA correlator bandwidth is 2 GHz wide for each
111: sidebands, which are separated by 10 GHz. The instrumental spectral resolution was 0.8125
112: MHz, but we smoothed it to 3.25 MHz ({\em i.e.} $\sim$ 4 km~s$^{-1}$) for the $^{12}$CO(2-1)
113: and SO(6$_5$-5$_4$) lines and to 8.125 MHz ({\em i.e.} $\sim$ 11 km~s$^{-1}$) in the case of
114: $^{13}$CO(2-1).
115:
116: We obtained the map of the 1.3 mm (225 GHz) continuum emission by averaging the line-free
117: channels from both LSB and USB, resulting in a total bandwidth of 3 GHz. To maximize the
118: angular resolution, we applied uniform weighting to the continuum visibilities, leading
119: to a synthesized beam size of 1.38$\arcsec$ x 1.23$\arcsec$ (PA = 169$\degr$). We estimate
120: a rms noise level of 8 mJy/beam in this continuum map.
121:
122: The channel maps of the $^{12}$CO(2-1) and SO(6$_5$-5$_4$) line emission were obtained
123: using natural weighting and gaussian tapering with a full width at half maximum (FWHM) of
124: 130 m, making a compromise between brigthness sensitivity and angular resolution. The
125: $^{13}$CO(2-1) map was derived using a taper distance of 80 m to increase the signal-to-noise
126: ratio. The resulting synthesized beam sizes and 1$\sigma$ noise levels are reported in Table 2.
127: The continuum visibilities were subtracted before imaging. All the maps were deconvolved
128: with a standard CLEAN/CLARK algorithm.
129:
130: At 230 GHz, the FWHM of the 6 m SMA antennas primary beam is 54$\arcsec$. The shortest
131: projected baseline being 8 m, we may loose informations on structures larger than 40$\arcsec$.
132: Nevertheless, the resulting coverage of the uv-plane is rather scarce until 18 m. We may
133: therefore also miss part of the emission associated with structures larger than 18$\arcsec$,
134: although we show in \S\ref{molecularspectra} from the comparison of the $^{12}$CO(2-1)
135: SMA and single dish data, that the missing flux is limited to less than 20 \%.
136:
137: \section{Results}
138:
139: \subsection{1.3 mm continuum emission} \label{cont}
140:
141: \citet{lip05} observed VY~CMa at centimeter wavelengths using the VLA. They do not resolve
142: the source at an angular resolution of $\sim$ 1$\arcsec$ at 1.3 cm. They also present Keck
143: mid-infrared imaging and show that the dust emission is elongated in the south-west direction
144: (PA = 227\degr) with FWHM $\sim$ 0.4$\arcsec$ and 0.3$\arcsec$ at 11.7 and 17.9 $\micron$
145: respectively. From diffraction-limited speckle interferometry, \citet{wit98} measure a dust
146: shell size of 138 x 205 mas (2.17 $\micron$), 80 x 116 mas (1.28 $\micron$) and 67 x 83 mas
147: (0.8 $\micron$) with roughly a north-south elongation. Thus, at near-IR wavelengths, the size
148: of the dust-emitting region decreases towards shorter wavelengths, progressively tracing warmer
149: dust close to the star.
150:
151: As seen from Fig.\ref{figcont}, the continuum emission is not resolved by our synthesized beam
152: ($\sim$ 1.4$\arcsec$). The continuum visibilities are well fitted by a point source of 288 $\pm$
153: 25 mJy (with 3$\sigma$ rms), located at a position +0.15$\arcsec$ $\pm$ 0.01$\arcsec$ to the east
154: and $-$0.05$\arcsec$ $\pm$ 0.01$\arcsec$ to the south, relatively to the phase center. Our
155: measurement of the flux density is consistent with the value of 270 $\pm$ 40 mJy from previous
156: SMA observations at 215 GHz \citep{shi04} but differs from the previous bolometric values of
157: 630 $\pm$ 65 mJy obtained with the JCMT \citep{kna93} and 378 $\pm$ 20 mJy with the SEST
158: \citep{wal91}. Part of the difference could be explained by contamination from line emission
159: in the bolometric observations. We note, however, that even the $^{12}$CO(2-1) line, presumably
160: the strongest in this frequency band, contributes only for about 2\% and 7\% of the total flux
161: density measurements with the JCMT and SEST, respectively. On the other hand, we cannot exclude
162: the presence of an extended low surface brigthness emission that could have been missed in our
163: observations. Independant checks of the continuum flux density in the upper and lower side bands
164: give consistent values within 10\%, implying that our measurements are not severly contaminated
165: by weak line emission.
166:
167: Because the continuum emission is of order of $\sim$ 1 mJy in the centimeter wavelenghts
168: \citep{lip05} and given that the contribution of the central star (black body emission)
169: amounts for S$_*$ $\sim$ 10 mJy, {\em i.e.} is negligible with respect to the total 1.3 mm
170: continuum emission S$_\nu$, this latter should arise almost completely from dust thermal emission.
171: % S(Jy) = pi theta**2*BB(nu,T) where theta = 15 AU / 1.5 kpc.
172: Under the Rayleigh-Jeans approximation and if the dust emission is optically thin, we can
173: estimate the dust mass M$_{d}$ within our 1.4$\arcsec$ beam, according to:
174: \begin{equation}
175: M_{d} = \frac{2 c^2 D^2 a_g \rho_g (S_\nu - S_*)}{3 Q_\nu k T_d \nu^2}
176: \end{equation}
177: \noindent where D is the distance to the source, $\rho_g$ and a$_g$ are the mass density and
178: radius of the dust grains, Q$_\nu$ is the grain emissivity and $T_d$ the average temperature
179: of the dust envelope. We assume typical values for oxygen rich stars, a$_g$ = 0.2 $\micron$,
180: $\rho_g$ = 3.5 g.cm$^{-3}$, Q$_\nu$ = 5.65 10$^{-4}$($\nu$ / 274.6 GHz), following
181: \citet{kna93}. The ratio of IR fluxes indicates a temperature T$_d$ $\sim$ 250 K in the dust
182: shell \citep{lip05}. Accordingly, we derive a dust mass of 1.5 10$^{-3}$ M$_\odot$, close
183: to the value M$_d$ = 1.3 10$^{-3}$ M$_\odot$ given by \citet{lip05}.
184: Assuming that the dust is uniformly distributed within a radius comparable to
185: our synthesized beam (upper limit of 1.4$\arcsec$), the optical depth towards
186: the central star would be $\tau_{1.3mm}$ $\sim$ 0.01.
187:
188: Furthermore, assuming an expanding velocity V$_{\rm exp}$ = 15 km~s$^{-1}$ (similar to the
189: escape velocity of $\sim$ 20 km~s$^{-1}$ at the dust formation radius, see also \S\ref{model})
190: leads to an age of order of 300 years. The average mass loss rate over this timescale is
191: 4.7 10$^{-4}$ M$_\odot$/yr, assuming an isotropic mass loss and a dust-to-gas ratio of 100.
192: This value is slightly higher than previous measurements for VY~CMa (see {\em e.g.}
193: \citealt{dan94,har01}) and comparable to mass loss rates estimated for other supergiants
194: \citep{net87}. \citet{smi01} and \citet{hum05} showed that localized and short-term
195: ejection events in VY~CMa could in fact reach 10$^{-3}$ M$_\odot$/yr.
196:
197: On the other hand, as pointed out by the referee, there is a possibility that the continuum
198: dust emission originates from a circumstellar disk, as advocated earlier by \citet{bow83} and
199: \citet{ric98} from observations of OH and H$_2$O maser emissions. However, the angular resolution
200: of our current data is not enough to constrain this possibility.
201:
202:
203: \subsection{Molecular line emission}
204:
205: \subsubsection{Description of the molecular line emission} \label{molecularline}
206:
207: The $^{12}$CO(2-1) emission (Fig.\ref{fig12co}) extends to $\sim$ 7$\arcsec$ around the central
208: star and is detected over a (LSR) velocity range $-$25 to 70 km~s$^{-1}$. There are three
209: distinct emission peaks in the channel maps: a first one at a velocity of about $-$5 km~s$^{-1}$
210: with an offset of 1$\arcsec$ to the east, a second one close to the systemic velocity of VY~CMa
211: (V$_{SYS}$ $\sim$ 22 km~s$^{-1}$), centered on the stellar position and a third one around +45
212: km~s$^{-1}$ with an offset of 1$\arcsec$ to the west. The line emission is stronger in the
213: western redshifted part of the nebula, where it reaches $\sim$ 5 Jy/beam. The velocity channels
214: around the systemic velocity of VY~CMa (12 $<$ V$_{LSR}$ $<$ 33 km~s$^{-1}$) show a north-south
215: elongated structure of roughly 5$\arcsec$ x 3$\arcsec$, embedded in a more diffuse envelope.
216: A second weak peak is seen 3$\arcsec$ to the west.
217:
218: Our $^{13}$CO(2-1) channel maps (Fig.\ref{fig13co}), although of low signal-to-noise, present
219: globally the same features as seen in $^{12}$CO(2-1). The two peaks are well detected at
220: velocities $\sim$ $-$5 and +50 km~s$^{-1}$, and separated by $\sim$ 2$\arcsec$. The emission
221: around the systemic velocity is weakly detected. The channel maps of the SO(6$_5$-5$_4$) emission
222: (Fig.\ref{figso}) also show the same features as that of the $^{12}$CO(2-1). The integrated
223: intensity maps for the three species are presented in Fig.\ref{figint}.
224:
225: The morphology and kinematics of the $^{12}$CO(2-1) emission around VY~CMa do not reproduce
226: those of a typical spherical expanding shell. We interpret the east and west offset bright
227: peaks, with high velocity relatively to the systemic velocity, as the signature of a east-west
228: bipolar outflow. \citet{shi03} already suggested the presence of a bipolar outflow around VY~CMa
229: from the butterfly like shape of the SiO(1-0) emission. The idea was further supported by SiO
230: masers linear polarization measurements \citep{shi04}, which result in a mean polarization
231: position angle of 72$\degr$, in the axis of the SiO bipolar structure. Previous observations of
232: OH \citep{bow83,bow93} and H$_2$O \citep{ric98} masers also show some elongation in a
233: similar direction with position angles of 50$\degr$ and 60$\degr$, respectively. However, the
234: maser emission is clumpy and might not trace the complete geometry of the outflow.
235:
236: In addition, the 1.3 mm continuum and CO emission around the systemic velocity indicate the
237: presence of a central dense component. The position-velocity diagrams (Fig.\ref{figpv}), along
238: the east-west and north-south directions, {\em i.e.} along and perpendicular to the outflow axis,
239: support our interpretation. Along the east-west direction, the high velocity bipolar outflow and
240: the slowly expanding shell appear clearly as two distinct kinematic components. The bipolar
241: outflow emission shows a point symmetry relative to the stellar position. It extends to roughly
242: 5$\arcsec$ from the star and has a velocity up to 45 km~s$^{-1}$ with respect to the systemic
243: velocity. This corresponds to a timescale of about 800 yr, comparable to the past period of strong
244: mass loss activity \citep{hum05,dec06}. The slowly expanding envelope can be seen out to a
245: radius of roughly 5$\arcsec$. Given an expansion velocity of 15 km~s$^{-1}$ (see \S\ref{model}),
246: the corresponding timescale is 2400 yr. The overall features are similar in the $^{12}$CO(2-1) and
247: SO(6$_5$-5$_4$) position-velocity diagrams, indicating that they arise from the same regions.
248:
249:
250: \subsubsection{Comparison with single dish spectra} \label{molecularspectra}
251:
252: We present in Fig.\ref{figspec} the spectra of the $^{12}$CO(2-1), $^{13}$CO(2-1) and
253: SO(6$_5$-5$_4$) lines, taken at the central position and convolved to a beam size of FWHM
254: 19$\farcs$7, equal to the JCMT beam at 230 GHz. The $^{12}$CO(2-1) line profile obtained from our
255: interferometric data can then be directly compared to the single-dish $^{12}$CO(2-1) spectrum
256: \citep{kem03} at the same angular resolution. The $^{12}$CO(2-1) SMA spectrum shows three
257: peaks at V$_{LSR}$ $\sim$ $-$5, 20 and 45 km~s$^{-1}$, similar with the JCMT spectrum. The high
258: velocity channels (V$_{LSR}$ $<$ 5 and $>$ 35 km~s$^{-1}$) reach the same brightness temperature
259: as the single-dish observations, indicating that the interferometer recovered all of the emission
260: for these velocities. Close to the systemic velocity of VY~CMa, {\em i.e.} for 5 $<$ V$_{LSR}$ $<$ 30
261: km~s$^{-1}$, there is a small difference between the two profiles, although less than $\sim$ 20\%
262: of the emission is lost by the interferometer. We estimate the integrated intensity of the
263: $^{12}$CO(2-1) line to 65 K~km~s$^{-1}$, comparable to the value of 66 $\pm$ 7 K~km~s$^{-1}$
264: reported by \citet{kem03}. The integrated intensities of the $^{13}$CO(2-1) and SO(6$_5$-5$_4$)
265: lines are estimated to 4 K~km~s$^{-1}$ and 28 K~km~s$^{-1}$, respectively. The 3$\sigma$ rms
266: uncertainty is $\sim$ 1 K~km~s$^{-1}$.
267:
268:
269: \subsubsection{SO emission}
270:
271: While carbon chemistry is mostly absent in the circumstellar envelope (CSE) of O-rich stars,
272: most of the carbon being locked up in CO, sulfur chemistry is expected to be relatively active.
273: S-bearing species are indeed commonly observed in the massive CSE around O-rich stars
274: \citep{sah92, omo93}.
275:
276: Under chemical thermodynamical equilibrium, \citet{tsu73} showed that the sulfur in CSEs is mostly
277: in the form of H$_2$S. Later non-equilibrium chemical models \citep{sca80, nej88, wil97}
278: predict that a large proportion of sulfur should go into SO and SO$_2$, as a result of the reactions:
279: S + OH $\longrightarrow$ SO + H, HS + O $\longrightarrow$ SO + H and SO + OH $\longrightarrow$
280: SO$_2$ + H. OH is a natural product of the photodissociation of H$_2$O, and VY~CMa exhibits intense
281: maser emission for these two species \citep{bow83,ric98}. On the other hand, the production of
282: HS requires highly endothermic reactions. From our SMA observations, the SO and CO morphology and
283: kinematics are similar, implying that SO is present both close to the star and in the bipolar
284: outflow. Observations of \citet{san00} towards the bipolar nebula OH~231.8+4.2 show a similar trend.
285: These authors argue for the presence of SO both in the central component and in the accelerated lobes,
286: with no significant difference in abundance between the two components.
287:
288: Assuming optically thin emission and an excitation temperature of T$_{ex}$ = 40 K \citep{sah92},
289: we estimate a SO column density $\sim$ 10$^{16}$ cm$^{-2}$ towards the center. The
290: corresponding SO abundance relative to H$_2$ is $\sim$ 10$^{-6}$.
291:
292: High angular resolution observations of other S-bearing species such as H$_2$S and SO$_2$ would
293: provide a better understanding of the sulfur chemistry and put stronger constraints for chemical
294: models.
295:
296:
297: \section{A simple model of the envelope} \label{model}
298:
299: We have constructed a simple model in order to better understand the geometry and physical
300: conditions within the circumstellar envelope around VY~CMa. The combination of our high angular
301: resolution CO(2-1) data together with previous single dish observations of multiple CO rotational
302: transitions \citep{kem03} provides very useful constraints for our model and leads us to a
303: coherent picture of the envelope.
304:
305: The geometry of our model is illustrated in Fig.\ref{figsketch}. The envelope is assumed to be
306: axisymmetric and radially expanding. The bipolar outflow direction defines the symmetry axis
307: of the envelope. We project the envelope directly into a regular 3-D grid. At each grid point,
308: the gas density, kinetic temperature and expansion velocity can be specified. The emerging CO
309: intensity along each ray through the envelope is calculated by integrating directly the standard
310: radiation transfer equation. We take into account 15 rotational levels of the CO molecule in its
311: vibrational ground state. The populations on the different rotational levels, which are necessary
312: for the calculation of the line opacity and source function at each grid point, are determined by
313: solving the statistical equilibrium equations within the framework of the large velocity gradient
314: formalism. The collision rates between CO and molecular hydrogren are taken from \citet{flo85} and
315: calculated for different temperatures using the prescription of \citet{jon75}. Because the envelope
316: around VY~CMa is oxygen rich, we adopt an abundance f = [CO]/[H$_2$] = 3 10$^{-4}$ as used in the
317: analysis of \citet{kem03}. The local line profile, which is needed to integrate the radiative
318: transfer equation, is determined through the turbulence velocity of the molecular gas. We assume
319: a turbulence velocity of 1 km~s$^{-1}$ in the slowly expanding envelope and 3 km~s$^{-1}$ in the
320: bipolar lobes, respectively. This assumption is reasonable as a higher outflow velocity in the
321: lobe would result in more turbulent environment.
322:
323: As discussed in \S\ref{molecularline}, two distinct kinematic components are clearly present in the
324: nebula around VY~CMa: (1) a slowly expanding envelope slightly elongated in the north-south direction
325: surrounded by a diffuse halo with some extension toward the north-west direction and (2) a high
326: velocity bipolar outflow oriented in the east-west direction. However, to keep our model as simple
327: as possible, we assume that the slowly expanding component is spherically symmetric. We adopt an
328: expansion velocity of 15 km~s$^{-1}$ for this component, which allows us to satisfactorily fit the
329: central peak seen in single-dish CO observations. The presence of diffuse emission suggests
330: that the gas density in the slowly expanding envelope peaks more sharply toward the center than for
331: the case of constant mass loss.
332: We therefore adopt a simple density distribution n($r$) = 4.5 10$^5$ (10$^{16}$ cm/$r$)$^{2.5}$, which
333: deviates slightly from the usual density distribution generated by constant mass loss. This adopted
334: distribution corresponds to a mass loss rate $\dot{M}$ = 4.5 10$^{-5}$ (10$^{16}$ cm/$r$)$^{0.5}$
335: M$_\odot$/yr, or a decrease in mass loss rate by a factor of 4 between the inner radius (R$_{\rm in}$
336: and the outer radius (R$_{\rm out}$). The kinetic temperature of the molecular gas in the oxygen-rich
337: circumstellar envelopes with constant mass loss has been known to roughly follow a power law T($r$)
338: $\sim$ $r^{-\alpha}$ from the modelling works of \citet{gol76} and \citet{jus94}. Although the mass
339: loss rate in the slowly expanding envelope of VY~CMa is not strictly constant, resulting probably in
340: some slight change in the temperature profile in comparison to the simple constant mass loss case,
341: we expect that qualitatively, the temperature profile still follows the power law. Therefore, we
342: assume in our calculations, for the sake of simplicity, that the kinetic temperature varies as
343: T($r$)= 500 K (2.25 10$^{15}$ cm/$r$)$^{0.7}$, which is similar to that used by \citet{kem03}
344: to model the oxygen-rich circumstellar envelopes.
345:
346: In order to reproduce the velocity gradient seen in the position-velocity
347: diagrams (Fig.\ref{figpv}), the outflow velocity in the bipolar lobes is allowed to increase linearly
348: with radial distance from the central star. In addition, the outflow velocity in the bipolar lobes
349: also varies as a function of the polar angle $\theta$ measured from the bipolar axis. The strong
350: emission peaks at large velocity shifts suggest the presence of an enhanced density shell within the
351: bipolar lobes. However, this density enhancement is more pronounced in the receeding lobe as clearly
352: seen in the position-velocity diagram. With the density enhancement included in the model, we find
353: that the bipolar lobes and the slowly expanding shell can account for the highly unusual triple peak
354: profile of single-dish CO lines observed by \citet{kem03} (see Fig.\ref{figspecco}).
355:
356: The intensities of the CO lines calculated along individual lines of sight are used to form the
357: model channel maps (Fig.\ref{figchanmodel}). We convolve these model channel maps with a gaussian
358: beam with specified FWHM to produce the model CO line profiles, which are indicated by the dashed
359: lines and compared with the JCMT observations of \citet{kem03} in Fig.\ref{figspecco}.
360:
361: We infer from our modelling that the velocity of the gas in the lobes increases linearly from
362: 15 km~s$^{-1}$, which is similar to the expansion velocity of the spherical envelope, to
363: 45 km~s$^{-1}$ at the outer radius. The bipolar outflow is found to be very open, with an opening
364: angle of 120$\degr$, and inclines at 15$\degr$ from the line of sight. All the model parameters,
365: which provide a reasonable fit to both the SMA and single dish data, are shown in Table 3.
366:
367: Although relatively simple, our model provides a qualitative understanding of the single dish data
368: obtained by \citet{kem03} and also of the structure of the envelope (Figs.\ref{figspecco},
369: \ref{figchanmodel}, \ref{figpvmodel}). The high angular resolution SMA data show that the CO(2-1)
370: emission originating from the high velocity bipolar lobes produces a double horn profile at roughly
371: $\pm$20 km~s$^{-1}$ from the systemic velocity. The central broad peak of the CO(2-1) line comes
372: from the slow component. Higher transitions of CO probe progressively warmer and denser gas located
373: in the inner part of the envelope. However, both the density and temperature of the gas in the inner
374: region of the bipolar outflow, as suggested by our model, are lower than that in the slowly
375: expanding envelope. As a result, the central peak becomes more prominent in higher transitions while
376: the strength of the emission from the bipolar lobes decreases noticeably. In addition, the lower
377: outflow velocity in the inner part of the bipolar lobes also leads to a narrower full width at zero
378: velocity for higher transitions of CO. This behavior can be clearly seen in Fig.\ref{figspecco}.
379:
380: By comparing the model prediction for the strength of the $^{13}$CO(2-1) line with our SMA
381: observations, we estimate the isotopic ratio $^{12}$C/$^{13}$C = 60.
382: This ratio is significantly larger than the values $^{12}$C/$^{13}$C = 6 and 12
383: found in the red supergiants $\alpha$ Ori and $\alpha$ Sco, respectively \citep{har84, hin76}.
384: It is likely that, prior to the current mass loss episode, VY~CMa experienced no significant
385: mixing of H burning products to the surface.
386: Our model predicts that most of the $^{13}$CO(2-1) line emission should arise from the bipolar
387: outflow (Fig.\ref{fig13comodel}), which is consistent with our SMA imaging data (Fig.\ref{fig13co}).
388:
389: We emphasize here that our model is constructed to explain the general features of the high
390: resolution SMA data and single-dish CO observations. Other features such as the slight elongation
391: of the slowly expanding shell and the more complicated structure within the bipolar lobes are not
392: considered. However, our model could serve as a useful starting point when higher angular
393: resolution data of CO(2-1) and higher transitions become available in the future.
394:
395:
396: \section{Discussion}
397:
398: \subsection{Connection with optical imaging/spectroscopy data}
399:
400: Our high angular resolution CO(2-1) data bring a clearer overall picture of the molecular
401: envelope around the enigmatic red supergiant VY~CMa. The envelope consists of two kinematic
402: components: a slowly expanding shell slightly elongated in the north-south direction and a high
403: velocity bipolar outflow oriented in the east-west direction with small inclination angle to the
404: line of sight. The presence of these two kinematic components have already been hinted in previous
405: observations of H$_2$O and OH masers \citep{bow83, bow93, ric98}. Given the small inclination
406: angle $\sim$ 15$\degr$ of the bipolar outflows to the line of sight, as derived from our model and
407: assuming uniform illumination in the nebula, one would expect that the envelope appears only
408: slightly asymmetric with a brighter east lobe (nearer side). However, optical and near IR images
409: obtained by \citet{mon99} and \citet{smi01} all show a highly irregular and asymmetric nebula around
410: VY~CMa. The nebula is seen to the west and southwest of the bright point-like source identified as
411: the obscured central star VY~CMa and is likely the result of a non-isotropic illumination by the
412: central star. Comparing the optical images with our SMA data, we note that the optical nebulosity
413: corresponds to the location of the west (far side) lobe and just a small part of the east (near side)
414: lobe. It is therefore likely that the light from VY~CMa escapes preferentially in the direction of
415: the west bipolar lobe and scattering on dust brings a fraction of the light back toward the observer.
416:
417: The strong 1.3 mm continuum emission detected in our data (see \S\ref{cont}) confirms the presence
418: of a thick inner dust shell surrounding VY~CMa. The average optical depth would be so high, if the
419: distribution of material in the dust shell is smooth, that all the stellar light would be completely
420: extincted. The complex nebula seen in the optical and near IR then suggests that stellar light must
421: be able to escape through some opening or holes in the dust shell, creating search light beams that
422: illuminate the nebula in an irregular pattern. We note that a similar suggestion was made by
423: \citet{wal78} to explain the subtle but peculiar change in the relative position between the star
424: and the brightest knot in the nebula.
425:
426: The thick and compact inner dust shell detected in 1.3 mm continuum emission might be the region
427: where TiO and ScO bands \citep{her74, wal86} seen in emission originate. The calculated
428: excitation temperatures of these molecular bands range from 600 to 800 K, suggesting that they
429: come from a hot and dense region located close to the star.
430:
431: High resolution optical and near IR images of \citet{mon99} and \citet{smi01} show a very
432: prominent long curved tail or arc to the north-west of VY~CMa together with several fainter
433: arcs and condensations in the envelope. The location of the long northwest arc roughly
434: coincides with the extended and somewhat diffuse CO(2-1) emission in the western part of
435: the envelope (Fig.\ref{fig12co}). Together with strong central emission core, this extended
436: emission is conspicuous in all channel maps redshifted with respect to the central star, most
437: prominent in the velocity range 21 -- 33 km~s$^{-1}$ and 46 -- 54 km~s$^{-1}$. The orientation
438: of this feature changes from west to north-west with increasing velocity upon closer inspection.
439: We can identify a similar feature in the channel maps of the SO emission at the same velocity,
440: although there are some slight differences at higher redshifted velocities, which are probably
441: related to local excitation of the SO molecule. It is therefore possible that the scattering of
442: stellar light by dust in this structure produces the prominent northwest arc seen in optical and
443: near IR images. Future observations of molecular lines at higher angular resolution and detailed
444: modelling of the light scattering by dust inside the envelope will help to clarify the relation
445: between the material distribution and the optical and near IR appearance of the nebula.
446:
447: Recently, \citet{hum05} have obtained high resolution long slit spectroscopic data of VY~CMa.
448: By measuring the velocity of the reflected stellar absorption lines and K I emission lines they
449: suggest that the prominent northwest arc is expanding at a velocity of $\sim$ 50 km~s$^{-1}$ on
450: the plane of the sky. More surprising is the existence of diffuse gas and dust more or less
451: stationary with respect to the central star, mostly along the slit positioned to the southeast
452: of VY~CMa (slit V in their Fig.1). Thus, components with very different kinematics seem to
453: co-exist in the nebula. To reconcile the optical spectroscopic data with our SMA results, we
454: note that for the simple case of single scattering, the light scattered by moving reflective
455: material (dust particles) within the envelope is Doppler shifted by
456: V$_{\rm exp}$(1 $-$ cos$\omega$), where V$_{\rm exp}$ is the expansion velocity and $\omega$ is
457: the angle measured from the line of sight to the radial vector of the scattering point (see
458: Fig.\ref{figsketch}). The velocity shift can be seen to be largest in the west lobe, which is
459: receeding along the line of sight, because $\omega$ $>$ 90$\degr$. If the association of the
460: extended feature in the west lobe as traced by CO emission and the northwest arc is correct,
461: we would expect the velocity shift induced by scattering of moving dust to vary from
462: 15 km~s$^{-1}$ for material in the arc moving close to the plane of the sky, which is part of
463: the slowly expanding shell, to $\sim$ 50 km~s$^{-1}$ or higher for material within the bipolar
464: lobe. This simple estimate of the velocity shift is consistent with the measurement of
465: \citet{hum05}. For absorption lines seen in scattering through the east lobe, the velocity
466: shift is expected to be close to zero because the angle $\omega$ is small. That might explain
467: the presence of dust and gas almost stationary with respect to the star as suggested by
468: \citet{hum05}. However, similar behavior seen in resonant K I lines are still difficult to
469: understand.
470:
471:
472: \subsection{Mass loss history}
473:
474: There are strong evidences that the mass loss rate in VY~CMa is time variable, such as the
475: prominent arc-like features seen in optical \citep{smi01} and the relative strength of the CO
476: rotational line profiles \citep{dec06}.
477:
478: Our modelling results suggest that the mass loss is varying as $\sim$ 4.5 10$^{-5}$
479: (10$^{16}$ cm/$r$)$^{0.5}$ M$_\odot$/yr in the slowly expanding envelope (V$_{\rm exp}$ =
480: 15 km~s$^{-1}$), {\em i.e.} that the mass loss from VY~CMa is increasing with time, reaching
481: $\sim$ 6.4 10$^{-5}$ M$_\odot$/yr in the inner part of the envelope (R$_{\rm in}$) or about
482: 100 years ago. In the bipolar lobes, the density enhancement between 5 -- 7 10$^{16}$ cm
483: represents an episode of strong mass loss about 350 to 500 years ago, for the expansion
484: velocity of 45 km~s$^{-1}$. The nominal mass loss rate at 6 10$^{16}$ cm is 10$^{-3}$
485: M$_\odot$/yr. A similar episode of intense mass loss has also been inferred from modelling
486: results of \citet{dec06}, although their assumption of spherical symmetry is inadequate to
487: describe the geometry and kinematics of the nebula around VY~CMa.
488:
489: The current mass loss from VY~CMa is better traced by dust continuum emission in the mid
490: infrared or millimeter and sub-millimeter wavelengths. We derived a mass loss rate of
491: 4.7 10$^{-4}$ M$_\odot$/yr from 1.3 mm dust continuum emission. This value is very close
492: to that obtained by \citet{har01} from fitting infrared spectroscopic data. As already noted
493: by \citet{dec06}, the difference between mass loss rate derived from CO line profiles and
494: dust continuum suggests that the dust continuum emission originates from the inner dense and
495: hot region (a few tens of stellar radii from the central star), which is not well traced by
496: CO low rotational lines.
497:
498:
499: \subsection{Origin of the complex envelope around VY~CMa}
500:
501: The origin of the bipolar outflow and the north-south elongated structure in VY~CMa is still
502: unclear at present. However, we note the similarity in morphology (bipolar outflows and/or
503: disk-like structures) between the envelope of VY~CMa and that around several AGB and post-AGB
504: stars such as $\pi^1$ Gru \citep{chi06}, V Hya \citep{hir04}, M~2-56 \citep{cas01}
505: and X Her \citep{nak05}. The current popular model proposed by \citet{mor87} to explain the
506: striking bipolar morphology and the presence of collimated fast outflow in AGB and post-AGB
507: envelopes involves a low mass binary companion. The companion can have strong effect on the
508: structure of the envelope. Its gravitational pull can focus the AGB wind toward the orbital
509: plane to create an equatorial density enhancement. In addition, the accretion of wind material
510: into the companion leads to the formation of an accretion disk, which might drive the collimated
511: high velocity outflow. Although the overall morphology of the envelope is similar to that around
512: other AGB and post-AGB stars, the high velocity wind from VY~CMa, as suggested by our data and
513: modelling results, is only weakly collimated with very wide opening angle. It is conceivable
514: that the interaction between an unseen companion and VY~CMa could produce equatorial density
515: enhancement, possibly identified as the elongated structure and induce the collimated fast wind
516: to precess. Subsequent interaction with the nebula would result in wide bipolar lobe. The
517: prominent curved nebulous tail seen in scattered light and the extended and curved structure
518: seen in our CO(2-1) data might actually trace the precessing outflow. To explain the prominent
519: nebulous tail, \citet{mon99} also suggest the existence of a rotating jet and estimate the
520: rotation period of a few thousands years. That time scale might be related to the orbital period
521: of an unseen companion around VY~CMa. However, high resolution images from \citet{smi01} suggest
522: that this structure is actually much more complicated than expected from the model of \citet{mon99}
523: and probably more consistent with a localized ejection event on the stellar surface. Our currrent
524: SMA data cannot distinguish between these possibilities. Future high angular resolution observations
525: and monitoring are necessary to determine the distribution of material and kinematics in order to
526: test the presence of a companion.
527:
528: VY~CMa might have magnetic activity, resembling solar flares or eruptions, as suggested by
529: \citet{smi01} in order to explain the presence of numerous arcs and bright knots seen in
530: scattered light in the nebula. We speculate further that the bipolar flow with wide opening
531: angle and the equatorial density enhancement could be similar to what is observed in the
532: solar wind. The fast solar wind originates from coronal holes located near the poles where
533: magnetic field lines are open and the slower but denser wind comes from the equatorial region.
534: The geometry of the solar wind is closely coupled to the dipolar field of the Sun. In the case
535: of VY~CMa, the presence and strength of the magnetic field can only be inferred indirectly
536: from observations of SiO, H$_2$O and OH masers. Observations of Zeeman effect in SiO maser
537: lines suggest a strength of 0 -- 2 G \citep{her06} for the magnetic field around
538: VY~CMa. Using simple estimates, \cite{sok98} and \cite{sok02} suggest that a magnetic field
539: of about 1 G can significantly influence the outflow.
540: %The B field is expected to dominate
541: %the dynamics of the outflow only when the magnetic pressure is larger than the ram pressure
542: %of the wind, {\em i.e.} B of the order of $>$10 G.
543: The observed strength of the magnetic field
544: in VY~CMa thus suggests that it might affect the dynamics of the wind and if the field is organised
545: on large scale, {\em i.e.} dipolar configuration, it could explain the structure of the envelope
546: as seen in our SMA data.
547:
548:
549: \section{Conclusion}
550:
551: In this paper, we present high angular resolution observations of the circumstellar molecular
552: envelope of the red supergiant VY~CMa obtained with the SubMillimeter Array. Thanks to the
553: wide frequency coverage of the SMA, we were able to trace the envelope in the $^{12}$CO(2-1),
554: $^{13}$CO(2-1) and SO(6$_5$-5$_4$) lines and dust continuum emission. We detected a strong and
555: unresolved continuum emission at the center of the nebula. The derived dust mass is 1.5 10$^{-3}$
556: M$_\odot$, implying a lower limit of the recent mass loss of 4.7 10$^{-4}$ M$_\odot$/yr over
557: the last 300 years. The CO(2-1) emission is resolved into two distinct kinematic components: a
558: slowly (15 km~s$^{-1}$) expanding shell elongated in the north-south direction and a high
559: velocity (45 km~s$^{-1}$) component with a clear velocity gradient along the east-west direction.
560: Similar structures are also seen in the $^{13}$CO(2-1) and SO(6$_5$-5$_4$) emission. We interpret
561: the high velocity component as a bipolar outflow with a wide opening angle ($\sim$ 120$\degr$),
562: for which the symmetry axis is oriented close to the line of sight (i = 15$\degr$). Using a
563: simple model, we could explain the main features of the CO(2-1) emission in our SMA data as well
564: as the previous single-dish CO multi-line observations. Density enhancement in the outflows,
565: representing a previous episode of high mass loss, is inferred from our model. Our data
566: complement previous high resolution optical spectroscopy and optical/IR images and provide a
567: better understanding of the complicated morphology of the nebula around VY~CMa. Future high
568: resolution and high sensitivity observations of CO lines, including high-J transitions, will
569: allow us to study the structure of the envelope and its physical properties in more detail.
570:
571:
572:
573: \acknowledgements
574:
575: We are very grateful to F. Markwick-Kemper for providing us her JCMT CO spectra on VY~CMa.
576: This work is supported in part by Academia Sinica and by grants from the National Science Council
577: of Taiwan (NSC) to D.-V-T. (NSC 94-2112-M-001-008), J.L., and S.K. This research made use of NASA's
578: Astrophysics Data System, as well as the SIMBAD database operated at CDS, Strasbourg, France.
579:
580:
581:
582:
583:
584: \begin{thebibliography}{}
585:
586: \bibitem[Bowers et al.(1983)]{bow83}{Bowers, P. F., Johnston, K. J. \& Spencer, J. H., 1983, \apj, 274, 733}
587: \bibitem[Bowers et al.(1993)]{bow93}{Bowers, P. F., Claussen, M. J. \& Johnston, K. J., 1993, \aj, 105, 284}
588: \bibitem[Castro-Carrizo et al.(2001)]{cas01}{Castro-Carrizo, A., Lucas, R., Bujarrabal, V., Colomer, F. \& Alcolea, J., 2001, \aap, 368, L34}
589: \bibitem[Chiu et al.(2006)]{chi06}{Chiu, P.-J., Hoang, C.-T., Dinh-V-Trung, et al., 2006, \apj, 645, 605}
590: \bibitem[Danchi et al.(1994)]{dan94}{Danchi, W. C., Bester, M., Degiacomi, C. G., Greenhill, L. J. \& Townes, C. H., 1994, \aj, 107, 1469}
591: \bibitem[Decin et al.(2006)]{dec06}{Decin, L., Hony, S., de Koter, A., et al., 2006, \aap, 456, 549}
592: \bibitem[de Jong, Chu \& Dalgarno(1975)]{jon75}{de Jong, T., Chu S.-I. \& Dalgarno, A., 1975, \apj, 199, 69}
593: \bibitem[Flower \& Launay(1985)]{flo85}{Flower, D. R. \& Launay, J. M., 1985, \mnras, 214, 271}
594: \bibitem[Goldreich \& Scoville(1976)]{gol76}{Goldreich, P. \& Scoville, N., 1976, \apj, 205, 144}
595: \bibitem[Harris \& Lambert(1984)]{har84}{Harris, M. J. \& Lambert, D. L., 1984, \apj, 281, 739}
596: \bibitem[Harwit et al.(2001)]{har01}{Harwit, M., Malfait, K., Decin, L., et al., 2001, \apj, 557, 844}
597: \bibitem[Herbig(1970)]{her70}{Herbig, G. H., 1970, \apj, 162, 557}
598: \bibitem[Herbig(1974)]{her74}{Herbig, G. H., 1974, \apj, 188, 533}
599: \bibitem[Herpin et al.(2006)]{her06}{Herpin, F., Baudry, A., Thum, C., Morris, D. \& Wiesemeyer, H., 2006, \aap, 450, 667}
600: \bibitem[Hinkle et al.(1976)]{hin76}{Hinkle, K. H., Lambert, D. L. \& Snell, R. L., 1976, \apj, 210, 684}
601: \bibitem[Hirano et al.(2004)]{hir04}{Hirano, N., Shinnaga, H.; Dinh-V-Trung, et al., 2004, \apjl, 616, L43}
602: \bibitem[Ho et al.(2004)]{ho04}{Ho, P. T. P., Moran, J. M. \& Lo, K. Y., 2004, \apj, 616, L1}
603: \bibitem[Humphreys et al.(2005)]{hum05}{Humphreys, R. M., Davidson, K., Ruch, G. \& Wallerstein, G., 2005, \aj, 129, 492}
604: \bibitem[Justtanont et al.(1994)]{jus94}{Justtanont, K., Skinner, C. J. \& Tielens, A. G. G. M., 1994, \apj, 435, 852}
605: \bibitem[Kemper et al.(2003)]{kem03}{Kemper, F., Stark, R., Justtanont, K., et al., 2003, \aap, 407, 609}
606: \bibitem[Knapp et al.(1993)]{kna93}{Knapp, G. R., Sandell, G. \& Robson, E. I., 1993, \apjs, 88, 173}
607: \bibitem[Lada \& Reid(1978)]{lad78}{Lada, C. J. \& Reid, M. J., 1978, \apj, 219, 95}
608: \bibitem[Lipscy et al.(2005)]{lip05}{Lipscy, S. J., Jura, M. \& Reid, M. J., 2005, \apj, 626, 439}
609: \bibitem[Monnier et al.(1999)]{mon99}{Monnier, J. D., Tuthill, P. G., Lopez, B., et al., 1999, \apj, 512, 351}
610: \bibitem[Morris(1987)]{mor87}{Morris, M., 1987, \pasp, 99, 1115}
611: \bibitem[Nakashima(2005)]{nak05}{Nakashima, J., 2005, \apj, 620, 943}
612: \bibitem[Nejad \& Millar(1988)]{nej88}{Nejad, L. A. M. \& Millar, T. J., 1988, \mnras, 230, 76}
613: \bibitem[Netzer \& Knapp(1987)]{net87}{Netzer, N. \& Knapp, G. R., 1987, \apj, 323, 734}
614: \bibitem[Omont et al.(1993)]{omo93}{Omont, A., Lucas, R., Morris, M. \& Guilloteau, S., 1993, \aap, 267, 490}
615: \bibitem[Perryman et al.(1997)]{per97}{Perryman, M. A. C., Lindergren L., Kovalesky J., et al., 1997, \aap, 323, L49}
616: \bibitem[Richards et al.(1998)]{ric98}{Richards, A. M. S., Yates, J. A. \& Cohen, R. J., 1998, \mnras, 299, 319}
617: \bibitem[Sahai \& Wannier(1992)]{sah92}{Sahai, R. \& Wannier, P. G., 1992, \apj, 394, 320}
618: \bibitem[S\'anchez-Contreras et al.(2000)]{san00}{S\'anchez-Contreras, C., Bujarrabal, V., Neri, R. \& Alcolea, J., 2000, \aap, 357, 651}
619: \bibitem[Scalo \& Slavsky(1980)]{sca80}{Scalo, J. M. \& Slavsky, D. B., 1980, \apj, 239, L73}
620: \bibitem[Shinnaga et al.(2003)]{shi03}{Shinnaga, H., Claussen, M. J., Lim, J., Dinh-van-Trung \& Tsuboi, M., 2003, in ASSL Vol. 283, Mass-Losing Pulsating Stars and their Circumstellar Matter, eds. Nakada, Honma, Y. M. \& Seki M. (Dordrecht: Kluwer), 393}
621: \bibitem[Shinnaga et al.(2004)]{shi04}{Shinnaga, H., Moran, J. M., Young, K. H. \& Ho, P. T. P., 2004, \apj, 616, L47}
622: \bibitem[Smith(2004)]{smi04}{Smith, N., 2004, \mnras, 349, L31}
623: \bibitem[Smith et al.(2001)]{smi01}{Smith, N., Humphreys, R. M., Davidson, K., et al., 2001, \aj, 121, 1111}
624: \bibitem[Soker(1998)]{sok98}{Soker, N., 1998, \mnras, 299, 1242}
625: \bibitem[Soker \& Zoabi(2002)]{sok02}{Soker, N. \& Zoabi, E., 2002, \mnras, 329, 204}
626: \bibitem[Stanek et al.(1995)]{sta95}{Stanek, K. Z., Knapp, G. R., Young, K. \& Phillips, T. G., 1995, \apjs, 100, 169}
627: \bibitem[Tsuji(1973)]{tsu73}{Tsuji, T., 1973, \aap, 23, 411}
628: \bibitem[Wallerstein(1978)]{wal78}{Wallerstein, G., 1978, Observatory, 98, 224}
629: \bibitem[Wallerstein(1986)]{wal86}{Wallerstein, G., 1986, \aap 164, 101}
630: \bibitem[Walmsley(1991)]{wal91}{Walmsley, C. M., Chini, R., Kreysa, E., et al., 1991, \aap, 248, 555}
631: \bibitem[Willacy \& Millar(1997)]{wil97}{Willacy, K. \& Millar, T. J., 1997, \aap, 324, 237}
632: \bibitem[Wittkowski et al.(1998)]{wit98}{Wittkowski, M., Langer, N. \& Weigelt, G., 1998, \aap, 340, L39}
633: \bibitem[Wright et al.(1990)]{wri90}{Wright, M. C.H., Carlstrom, J. E., Plambeck R. L. \& Welch W. J., 1990, \aj, 99,1299}
634:
635: \end{thebibliography}
636:
637: %\clearpage
638:
639: \begin{table}
640: \begin{tabular}{lll}
641: \multicolumn{3}{l}{Table 1: Basic data of VY~CMa} \\
642: \hline
643: \hline
644: R.A. (J2000) & 07$^{\rm h}$22$^{\rm m}$58$\fs$3315 & (1) \\ %07 22 58.3315 -25 46 03.174
645: Dec. (J2000) & $-$25$\degr$46$\arcmin$03$\farcs$174 & (1) \\
646: Distance & 1.5 kpc & (2) \\
647: Angular scale & 1$\arcsec$ $\sim$ 1500 AU $\sim$ 2.25 10$^{16}$ cm & \\
648: Luminosity L$_*$ & $\sim$ 2 10$^5$ L$_\odot$ & (3) \\
649: Effective temperature T$_{eff,*}$ & 2800 K & (3) \\
650: Stellar radius R$_*$ & 1.2 10$^{14}$ cm & (3) \\ % 600 Ro = 4.2e11 m = 2.8 AU % Mon99 R*=5.5 mas
651: Stellar mass M$_*$ & $\ge$ 15 M$_\odot$ & \\
652: Mass loss rate $\dot{M}$ & $\sim$ 2 -- 4 10$^{-4}$ M$_\odot$/yr & (4) \\
653: \hline
654: \end{tabular}
655: \tablerefs{ (1) from the Hipparcos catalogue, \citet{per97}; (2) \citet{lad78}; (3) \citet{mon99}; (4) \citet{bow83}, \citet{dan94}, \citet{sta95}, \cite{har01}.}
656: \end{table}
657:
658: %\clearpage
659:
660: \begin{table}
661: \begin{tabular}{lccccc}
662: \multicolumn{3}{l}{Table 2: SMA line observations towards VY~CMa.} \\
663: \hline
664: \hline
665: \multicolumn{1}{c}{Line} & Rest freq. & Synthesized beam & P.A. & Vel. res. & 1$\sigma$ rms \\
666: & (GHz) & (arcsec) & (deg.) & (km~s$^{-1}$) & (mJy/beam) \\
667: \hline
668: $^{12}$CO(2-1) & 230.538 & 1.94 x 1.67 & 0 & 4.2 & 140 \\
669: $^{13}$CO(2-1) & 220.399 & 2.65 x 2.30 & 147 & 11 & 65 \\
670: SO(6$_5$-5$_4$) & 219.949 & 1.99 x 1.80 & 155 & 4.4 & 110 \\
671: \hline
672: \end{tabular}
673: \end{table}
674:
675: %\clearpage
676:
677: \begin{table}
678: {\small
679: \begin{tabular}{lll}
680: \multicolumn{3}{l}{Table 3: Parameters of our model} \\
681: \hline
682: \hline
683: Envelope inner radius (R$_{\rm in}$) & 5 10$^{15}$ cm & \\
684: Envelope outer radius (R$_{\rm out}$) & 8 10$^{16}$ cm & \\
685: Bipolar lobe inner radius (R$_{\rm lobe}$) & 1.5 10$^{16}$ cm & \\
686: Opening angle of the bipolar lobes & 120$^o$ & \\
687: Envelope expansion velocity & 15 km~s$^{-1}$ & \\
688: Expansion velocity in the bipolar lobes & \multicolumn{2}{l}{[30 - 10 tan$\theta$/tan(60$^o$)]
689: (r - R$_{\rm lobe}$)/(R$_{\rm out}$ - R$_{\rm lobe}$) + 15 km~s$^{-1}$ $^\dagger$} \\
690: Gas density in the envelope & 4.5 10$^5$(10$^{16}$ cm/r)$^{2.5}$ cm$^{-3}$ & \\
691: Gas temperature in the envelope & 500 (2.25 10$^{15}$ cm/r)$^{0.7}$ K & \\
692: Gas density in the bipolar lobes & 6 10$^4$ cm$^{-3}$ & for R$_{\rm lobe}$ $<$ r $<$ 5 10$^{16}$ cm \\
693: ... & 10$^5$ cm$^{-3}$ & for 5 10$^{16}$ cm $<$ r $<$ 7 10$^{16}$ cm \\
694: ... & 10$^5$ (7 10$^{16}$ cm/r)$^2$ (cm$^{-3}$) & for r $>$ 7 10$^{16}$ cm \\
695: Gas temperature in the bipolar lobes & 150 (1.5 10$^{16}$ cm/r)$^{0.7}$ K & for R$_{\rm lobe}$ $<$ r $<$ 5 10$^{16}$ cm \\
696: ... & 35 K & for 5 10$^{16}$ cm $<$ r $<$ 7 10$^{16}$ cm \\
697: ... & 35 (7 10$^{16}$ cm/r) K & for r $>$ 7 10$^{16}$ cm \\
698: Inclination angle & 15$^o$ & \\
699: Position angle of the nebula axis & 90$^o$ & \\
700: f=[CO]/[H$_2$] & 3 10$^{-4}$ & \\
701: $^{12}$C/$^{13}$C & 60 & \\
702: Total molecular gas mass in the envelope & 0.15 M$_\odot$ & \\
703: \hline
704: \end{tabular}
705: }
706: \end{table}
707:
708: %\clearpage
709:
710: \begin{figure}
711: \includegraphics[width=10cm]{f1bw.eps}
712: \caption{1.3 mm continuum emission map of VY~CMa. The synthesized beam is 1.38$\arcsec$ x
713: 1.23$\arcsec$ (PA = 169$\degr$) as indicated by the ellipse in the bottom left corner.
714: Contour levels are every 40 mJy/beam (5$\sigma$).
715: The positions of the SiO \citep{wri90}, H$_2$O \citep{bow93} and OH \citep{bow83}
716: maser centroids are indicated.
717: The 1.3 mm continuum peak is located at R.A. = 07$^{\rm h}$22$^{\rm m}$58$\fs$339 and
718: Dec. = $-$25$\degr$46$\arcmin$03$\farcs$24 (J2000).}
719: \label{figcont}
720: \end{figure}
721:
722: %\clearpage
723:
724: \begin{figure}
725: \includegraphics[height=11cm]{f2bw.eps}
726: \caption{Channel maps of the $^{12}$CO(2-1) line emission. The synthesized beam sizes
727: 1.94$\arcsec$ x 1.67$\arcsec$ (PA = 0$\degr$) and is shown in the bottom left corner of
728: the figure. Contour levels are drawn every 0.55 Jy/beam or 3.9 K ($\sim$ 4$\sigma$).
729: The velocity resolution is 4.2 km~s$^{-1}$. The cross indicates the position of
730: the phase center.} \label{fig12co}
731: \end{figure}
732:
733: %\clearpage
734:
735: \begin{figure}
736: \includegraphics[height=8cm]{f3bw.eps}
737: \caption{Channel maps of the $^{13}$CO(2-1) line emission. The synthesized beam sizes
738: 2.65$\arcsec$ x 2.30$\arcsec$ (PA = 147$\degr$) and is shown in the bottom left corner
739: of the figure. Contour levels are drawn every 0.2 Jy/beam or 0.82 K ($\sim$ 3$\sigma$).
740: The velocity resolution is 11 km~s$^{-1}$. The cross indicates the position of the phase
741: center.} \label{fig13co}
742: \end{figure}
743:
744: %\clearpage
745:
746: \begin{figure}
747: \includegraphics[height=11cm]{f4bw.eps}
748: \caption{Channel maps of the SO(6$_5$-5$_4$) line emission. The synthesized beam sizes
749: 1.99$\arcsec$ x 1.80$\arcsec$ (PA = 155$\degr$) and is shown in the bottom left corner
750: of the figure. Contour levels are drawn every 0.44 Jy/beam or 3.1 K ($\sim$ 4$\sigma$).
751: The velocity resolution is 4.4 km~s$^{-1}$. The cross indicates the position of the phase
752: center.} \label{figso}
753: \end{figure}
754:
755: %\clearpage
756:
757: \begin{figure}
758: \includegraphics[width=15cm]{f5bw.eps}
759: \caption{Integrated maps of the $^{12}$CO(2-1), $^{13}$CO(2-1), SO(6$_5$-5$_4$) emission
760: lines. The synthesized beam sizes are indicated in the bottom left corner of each frame.
761: Contour levels are every 140 (7$\sigma$), 30 (3$\sigma$), and 105 (7$\sigma$) K~km~s$^{-1}$
762: respectively. Offsets refer to the position of the phase center.} \label{figint}
763: \end{figure}
764:
765: %\clearpage
766:
767: \begin{figure}
768: \includegraphics[height=7cm]{f6bw.eps}
769: \caption{Position - velocity diagrams along the east-west and north-south directions for
770: the $^{12}$CO(2-1) and SO(6$_5$-5$_4$) line emissions. Countour levels every 0.60 and 0.48
771: Jy/beam (4$\sigma$), respectively.}
772: \label{figpv}
773: \end{figure}
774:
775: %\clearpage
776:
777: \begin{figure}
778: \includegraphics[width=10cm]{f7.eps}
779: \caption{Spectra of the $^{12}$CO(2-1), $^{13}$CO(2-1) and SO(6$_5$-5$_4$) lines smoothed to a
780: 19$\farcs$7 FWHM beam. The $^{12}$CO(2-1) spectrum observed with the JCMT by \citet{kem03} is
781: shown as a dashed line in the upper panel. The velocity resolution is 4.2, 11 and 4.4 km~s$^{-1}$
782: from top to bottom.}
783: \label{figspec}
784: \end{figure}
785:
786: %\clearpage
787:
788: \begin{figure}
789: \includegraphics[width=13cm]{f8bw.eps}
790: \caption{Sketch of the geometry of our model. The central star VY~CMa (S) is embedded in a spherically
791: symmetric slowly expanding envelope (light grey). Part of the nebula is occupied by a bipolar outflow
792: with a wide opening angle (dotted pattern). Regions of density enhancement within the bipolar lobes
793: are also included (filled bands). The line of sight OS intercepts the bipolar outflow axis with an
794: inclination angle of 15$\degr$. A stellar photon is scattered at point P inside the envelope to the
795: observer. The angle $\omega$ is defined as $\hat{OSP}$.}
796: \label{figsketch}
797: \end{figure}
798:
799: %\clearpage
800:
801: \begin{figure}
802: \includegraphics[width=10cm]{f9.eps}
803: \caption{Spectra of the CO(2-1), CO(3-2), CO(6-5) and CO(7-6) lines observed by \citet{kem03}
804: using JCMT are shown (full lines) together with the predicted CO line profiles from our model
805: (dotted lines).}
806: \label{figspecco}
807: \end{figure}
808:
809: %\clearpage
810:
811: \begin{figure}
812: \includegraphics[height=7cm]{f10.eps}
813: \caption{SMA spectrum of the $^{13}$CO(2-1) line convolved with a 19$\farcs$7 FWHM beam (full
814: line) together with the output spectrum from our model (dotted line). The model spectrum is
815: smoothed to the same velocity resolution of the SMA data (11 km~s$^{-1}$). A ratio
816: $^{12}$C/$^{13}$C = 60 was inferred from our modelling.}
817: \label{fig13comodel}
818: \end{figure}
819:
820: %\clearpage
821:
822: \begin{figure}
823: \includegraphics[width=10cm]{f11bw.eps}
824: \caption{Predicted channel maps of the CO(2-1) emission from the envelope around VY~CMa.
825: Contour levels are the same as in Fig.\ref{fig12co}.}
826: \label{figchanmodel}
827: \end{figure}
828:
829: %\clearpage
830:
831: \begin{figure}
832: \includegraphics[height=7cm]{f12bw.eps}
833: \caption{Predicted position-velocity diagram of the CO(2-1) emission along the east-west (left frame)
834: and north-south (right frame) direction.}
835: \label{figpvmodel}
836: \end{figure}
837:
838: \end{document}
839: