1: \documentclass[12pt,preprint]{aastex}
2:
3: \shorttitle{Images of Vega Dust Ring}
4: \shortauthors{Marsh et al.}
5:
6: \begin{document}
7:
8: \title{Images of Vega Dust Ring at 350 and 450 $\mu$m: New Clues to the
9: Trapping of Multiple-Sized Dust Particles in Planetary Resonances}
10:
11: \author{K. A. Marsh\altaffilmark{1},
12: C. D. Dowell\altaffilmark{1},
13: T. Velusamy\altaffilmark{1},
14: K. Grogan\altaffilmark{1},
15: C. A. Beichman\altaffilmark{2}}
16:
17: \altaffiltext{1}{Jet Propulsion Laboratory, 4800 Oak Grove Drive, Pasadena,
18: CA 91109; Kenneth.A.Marsh@jpl.nasa.gov, cdd@submm.caltech.edu,
19: Thangasamy.Velusamy@jpl.nasa.gov, Keith.Grogan@jpl.nasa.gov}
20: \altaffiltext{2}{California Institute of Technology, IPAC, MS 100-22,
21: Pasadena, CA 91125; chas@ipac.caltech.edu}
22:
23: \pagebreak
24:
25: \begin{abstract}
26: We have used the SHARC II camera at Caltech Submillimeter Observatory
27: to make 350 $\mu$m and 450 $\mu$m images of the Vega dust disk at spatial
28: resolutions (FWHM) of 9\farcs7 and 11\farcs1, respectively. The images show
29: a ring-like morphology (radius $\sim100$ AU) with inhomogeneous structure
30: that is qualitatively different from that previously reported at 850 $\mu$m
31: and longer wavelengths. We attribute the 350/450 $\mu$m emission to
32: a grain population whose characteristic size ($\sim1$ mm) is intermediate
33: between that of the cm-sized grains responsible for emission longward of
34: 850 $\mu$m and the much smaller grains ($\stackrel{<}{_\sim}18$ $\mu$m) in
35: the extensive halo, visible at 70 $\mu$m, discussed by \citet{su05}.
36: We have combined our submillimeter images with
37: {\em Spitzer\/} data at 70 $\mu$m to produce 2-d maps of line-of-sight
38: optical depth (relative column density). These ``tau maps" suggest that the
39: mm-sized grains are located preferentially in three symmetrically-located
40: concentrations. If so, then this structure could be understood in
41: terms of the \cite{wya03} model in which planetesimals are trapped in the
42: mean motion resonances of a Neptune-mass planet at 65 AU, provided
43: allowance is made for the spatial distribution of dust grains to differ from
44: that of the parent planetesimals. The peaks of the tau maps
45: are, in fact, located near the expected positions corresponding to
46: the 4:3 resonance. If this
47: identification is confirmed by future observations, it would resolve an
48: ambiguity with regard to the location of the planet.
49: \end{abstract}
50:
51: \keywords{circumstellar matter --- planetary systems --- stars: individual
52: (\object{Vega})}
53:
54: \section{Introduction}
55:
56: The observable behavior of circumstellar dust particles under the influence of
57: gravity and radiation pressure can provide information on
58: the locations and masses of unseen planets (see, for example,
59: \citet{wya03,wya06,del05}). A particularly suitable object for study
60: is the A0 star Vega at 7.76 pc whose disk is seen nearly face-on, as suggested
61: by the low inclination ($5^\circ$) of the stellar rotation axis \citep{gul94}.
62: Images at 850 $\mu$m and 1.3 mm
63: \citep{hol98,wil02,koe01} indicate a partial dust ring ($r\sim100$ AU)
64: dominated by two unequal clumps
65: interpreted as collisional debris from resonantly-trapped
66: planetesimals. More recent observations with the Multiband Imaging
67: Photometer for {\em Spitzer\/} (MIPS) \citep{su05} have revealed an
68: extensive halo representing small ($\sim2$ $\mu$m) and medium-sized
69: ($\sim18$ $\mu$m) grains blown out from the ring by radiation pressure, the
70: ring itself being dominated by large grains of radii $a>180$ $\mu$m.
71:
72: Dynamical modeling shows the spatial distribution of the ring to be consistent
73: with the 2:1(u) resonance
74: of a Neptune-mass planet which has undergone migration \citep{wya03}.
75: Such a model predicts the existence of other populated resonances (for example
76: 3:2 and 4:3), the detection of which would help confirm the model and
77: provide new constraints. We present new observations
78: at 350 $\mu$m and 450 $\mu$m which bear on this issue.
79:
80: \section{Observations and Data Reduction}
81:
82: Vega was observed at 350 $\mu$m and 450 $\mu$m with the SHARC II
83: camera \citep{dow03} at the Nasmyth focus of the Caltech Submillimeter
84: Observatory (CSO) on UT 2004 Sep 17--19 (350 $\mu$m; $\tau_{\rm 225\,GHz}
85: \simeq 0.041$), 2005 Apr 23 (350 $\mu$m; $\tau_{\rm 225\,GHz}\simeq 0.036$),
86: and 2005 Jun 11--13 (450 $\mu$m; $\tau_{\rm 225\,GHz}\simeq 0.048$), where
87: $\tau_{\rm 225\,GHz}$ represents the zenith value of atmospheric optical
88: depth. The on-source total integration times were 5.0 hr, 2.7 hr and
89: 9.7 hr, respectively. Similarly to our Fomalhaut observations
90: \citep{mar05}, the telescope was scanned in an oscillatory fashion in azimuth
91: and elevation with peak-to-peak amplitudes of $30''-100''$. The Vega images
92: were generated with an iterative code (``sharcsolve") similar to
93: CRUSH\footnote{See http://www.submm.caltech.edu/~sharc/crush/index.htm}.
94: To reduce spatial $1/f$ noise in the center of the image, the sky intensity
95: was forced to be zero beyond $35''$ from the star (analogous to spatial
96: chopping with amplitude $70''$ in all directions), except for the final
97: iteration which allowed non-zero intensities outside $35''$ radius for
98: noise-evaluation purposes. After the application of
99: Gaussian smoothing (with a kernel of 5\farcs5 FWHM),
100: the angular resolutions (FWHM) of the output images were
101: 9\farcs7 and 11\farcs1, at 350 $\mu$m and 450 $\mu$m, respectively.
102: Absolute calibration was accomplished with hourly, interspersed observations
103: of point sources, and is based on assumed Neptune fluxes of $S_{350}=92.2$ and
104: 86.5 Jy in 2004 and 2005, respectively, and
105: $S_{450}=65$ Jy. We estimate 1$\sigma$ calibration accuracies of 30\% and
106: $1''$ in absolute flux and pointing, respectively. The
107: flux error includes the effect of subtraction of the slowly-varying background
108: ($\sim5$\% level over spatial scales of a few tens of arcseconds at both
109: wavelengths).
110:
111: The point source response function (PSF) was obtained from observations of
112: Mars at 350 $\mu$m and Neptune at 450 $\mu$m, rotated and coadded to recreate
113: the rotational smearing due to changing parallactic angle.
114: The nonzero angular diameters of those planets
115: (3\farcs5 and 2\farcs3, respectively) broadened
116: the estimated PSFs by 6\% and 2\%, respectively; the effect
117: on subsequent processing was found to be negligible within prevailing
118: errors.
119:
120: Figure \ref{fig1} shows the observed images at 350 $\mu$m
121: and 450 $\mu$m, before and after subtracting the estimated
122: photospheric contributions of 35 mJy and 21 mJy, respectively.
123: The integrated flux densities of Vega (before photospheric subtraction),
124: within a circular aperture of $30''$
125: radius, at 350 $\mu$m and 450 $\mu$m are $500\pm150$ mJy and $150\pm45$ mJy,
126: respectively.
127:
128: The observed images show clearly the ring morphology of the
129: Vega disk, and suggest inhomogeneous structure. Figure 2
130: shows the 350 $\mu$m intensity variation as a function of azimuth,
131: calculated in an annulus of inner and outer radii 6\farcs9 and 13\farcs9,
132: respectively. Three peaks are apparent, at azimuths of $-90^\circ$, $0^\circ$,
133: and $120^\circ$, with amplitudes of $3\sigma$, $4\sigma$, and $2\sigma$
134: above the local mean level respectively, where $\sigma$ represents the
135: statistical measurement noise.
136:
137: \section{Mapping the Relative Column Density of Dust}
138:
139: We have mapped the line-of-sight optical depth (relative column density of
140: dust) in the Vega disk as a function of 2-d location in the disk plane using
141: the DISKFIT procedure \citep{mar05}, assuming in this case
142: a geometrically-thin face-on disk. The output, referred to as
143: a ``tau map," is estimated using a set of observed images at
144: multiple wavelengths, taking full account of the corresponding PSFs.
145: It is assumed that the
146: local temperature of each dust grain component is determined by the
147: energy balance of individual grains in the stellar radiation field using
148: the results of \citet{bac93}. The current version of the code makes
149: simultaneous estimates of the tau maps corresponding to the different
150: grain components, each of which is characterized by the parameters $\lambda_0$
151: and $\beta$, where $\lambda_0$ represents the wavelength above which the
152: grains radiate inefficiently and has an approximate correspondence with the
153: grain radius, $a$, and $\beta$ is the power-law index of the wavelength
154: dependence of opacity such that $\kappa_\lambda\propto\lambda^{-\beta}$.
155: The spatial resolution of the tau maps exceeds that of
156: the raw images due to implicit deconvolution of the PSFs.
157:
158: Tau maps were made using data at three wavelengths, by combining our
159: 350 \& 450 $\mu$m images with a 70 $\mu$m MIPS image at the fine ($5''$)
160: pixel scale. We employed a two-grain model consisting
161: of the 18 $\mu$m grains which dominate the 70 $\mu$m emission
162: \citep{su05} and a population of larger grains whose size was chosen based
163: primarily on the observed spectral slope between 350 $\mu$m and 70 $\mu$m.
164: Specifically, assuming $\beta=1$ \citep{den00}, the spectral slope
165: implies $T_{\rm dust}\stackrel{<}{_\sim} 50$ K for the larger grains;
166: if we further assume a size-temperature dependence corresponding to the grain
167: composition (silicate-carbon mix) used by \citet{su05} in their Vega modeling,
168: we obtain $a\stackrel{>}{_\sim}100$ $\mu$m. Since this
169: limit was based on the extremes of the flux error bars and of the relative
170: contributions of ring and halo at 70 $\mu$m, we have
171: adopted a larger value (1 mm) as being more likely to be representative of
172: the grains reponsible for 350/450 $\mu$m emission. The results
173: are presented in Figure \ref{fig3}, which shows tau maps
174: for the two grain components (18 $\mu$m and 1 mm)
175: separately and also the total optical depth (lower panel).
176: Superposed on the latter, for comparison, are the positions of
177: previously-reported emission peaks from longer wavelength data at 850 $\mu$m
178: and 1.3 mm \citep{hol98,koe01,wil02}.
179:
180: The tau map for 1 mm grains shows inhomogeneities, the reality of which
181: we have assessed using a $\chi^2$ test based on fits to the 350 $\mu$m
182: and 450 $\mu$m data. Specifically, an azimuthally-uniform ring resulted in
183: a 10\% increase in $\chi^2$ relative to an unconstrained tau map; taking into
184: account the number of independent data points (255), this translates into
185: a relative probability $\sim5\times10^{-6}$ corresponding to a $4.6\sigma$
186: deviation. We therefore conclude that the azimuthal structure we see in the
187: ring is at the 4$\sigma$--5$\sigma$ significance level, consistent with our
188: findings based on the $S/N$ of peaks in the observed 350 $\mu$m image
189: (see Figure \ref{fig2}).
190:
191: The tau map for 18 $\mu$m grains is dominated
192: by a ring of larger scale ($r\simeq140$ AU) than for
193: the 1 mm grains ($r\simeq100$ AU). There is also lower-level diffuse
194: structure masked by surrounding noise spikes, the latter resulting from
195: the radial increase in tau estimation error corresponding to decreasing dust
196: temperature. We can, however,
197: smooth out these fluctuations with azimuthal averaging to
198: show the underlying diffuse structure. Figure \ref{fig4} shows the result
199: in the form of radial profile plots of
200: the 1 mm grain component (filled circles) and the 18 $\mu$m grain
201: component (open circles). The error bars reflect the tau map estimation
202: errors after a $\sqrt{N}$ reduction from azimuthal averaging in
203: annuli of width 17.5 AU. It is apparent that for
204: $r>200$ AU, the radial falloff in column density of 18 $\mu$m grains
205: is consistent with the $1/r$ variation reported by \citet{su05}. Inside
206: 200 AU, however, some of these grains may be concentrated in the
207: the ring structure, but the proportion is subject to model uncertainties
208: related to the larger grain component.
209:
210: \section{Interpretation}
211:
212: The observations show qualitatively different appearance of
213: the Vega dust ring at 350/450 $\mu$m and $\ge850$ $\mu$m. Furthermore,
214: the 850 $\mu$m flux (23 mJy) predicted by our tau map accounts for only a
215: fraction of either of the published values of 850 $\mu$m flux, which
216: are $45.7\pm5.4$ Jy \citep{hol98} and 91 mJy (\citet{hol05}, as quoted
217: by \citet{su05}). The observations can, however, be reconciled using a
218: model in which the ring emission at 350/450 $\mu$m is dominated by
219: 1 mm grains with a $\lambda^{-1}$ opacity law, and the emission longward
220: of 850 $\mu$m is dominated by an additional component of much larger grains
221: with flat spectral opacity ($\beta\sim0$); the latter would require
222: grain sizes of at least a centimeter \citep{pol94}.
223:
224: The morphology of the
225: 850 $\mu$m image (with its single dominant peak at $\sim45^\circ$ in position
226: angle) has been attributed to dynamical effects, the most important of which is
227: the 2:1(u) resonance of a Neptune-mass planet 65 AU from the star \citep{wya03}.
228: Those calculations also show that other resonances are present, principally
229: 3:2 and 4:3, the latter of which shows a 3-fold spatial symmetry
230: consistent with what we see at 350/450 $\mu$m. In this regard we note that
231: for a planet of given mass, the requirement that the dust grains not be
232: dislodged from resonance by radiation pressure sets a lower limit,
233: $a_{\rm crit}$, to the grain size \citep{wya06}. For a Neptune-mass planet
234: orbiting Vega (luminosity $60L_\odot$; mass $2.5M_\odot$),
235: Wyatt's expression leads to $a_{\rm crit}=900$ $\mu$m,
236: consistent with our choice of $a=1$ mm for the tau maps.
237:
238: Based on the 850 $\mu$m data alone, an ambiguity exists for the orbital
239: direction of the planet, shown counterclockwise in Figure 15(b) of
240: \citet{wya03}. But if our identification of the
241: 350/450 $\mu$m structure with the 4:3 resonance is correct, then
242: the ambiguity is resolved; the planet is then located diametrically opposite to
243: the star from the SE clump in our Figure \ref{fig3} and its orbit is
244: clockwise. The predicted locations
245: of the 4:3 resonance peaks are then given by the arrows in Figure \ref{fig2},
246: which are in substantial agreement with observation.
247:
248: It would then remain to explain why the cm-sized grains (seen at 850 $\mu$m) and
249: mm-sized grains (seen at 350/450 $\mu$m) exhibit two different resonance
250: patterns. We note, however, that the analysis by \citet{su05} suggests that
251: the dust has been released during a relatively recent event, most likely
252: the collision of two planetesimals. This being the case, our results could
253: be understood if the two planetesimals involved were located in the 4:3
254: resonance, and that the collision released smaller
255: particles than the ones which were there before, thus affecting the
256: 350/450 $\mu$m appearance much more than that at 850 $\mu$m.
257: The 4:3 resonance is, in fact, a relatively likely place for a collision to
258: occur, since the collision timescale decreases with decreasing orbital
259: semimajor axis, and hence is shorter in the 4:3 resonance than in 2:1.
260:
261: We have simulated this scenario with dynamical modeling based on a
262: Neptune-mass planet at 65 AU, assuming orbital parameters similar to
263: \citet{wya03}, and further assuming that the 4:3 resonance is populated by
264: smaller grains (1 mm) than the other resonances (1 cm). The results are
265: presented in Figure \ref{fig5}. Comparison of the predicted 350 $\mu$m
266: and 850 $\mu$m images with the observations
267: (Figure \ref{fig1} of this paper and \citet{hol98}, respectively)
268: shows that the model successfully accounts for the wavelength dependence
269: of the observed structure.
270: Our analysis therefore suggests that
271: different resonances can be populated by grain populations of different sizes,
272: depending on the collisional history of the planetesimals. This means that
273: the dust distribution does not necessarily mimic the planetesimal
274: distribution as assumed by \citet{wya03}. It will thus be important to confirm
275: the detection of the 4:3 resonance and constrain the grain sizes involved
276: using future observations at multiple wavelengths.
277:
278: \acknowledgments
279:
280: This work was performed by the Jet Propulsion Laboratory,
281: California Institute of Technology, under contract with the
282: National Aeronautics and Space Administration. Research at the Caltech
283: Submillimeter Observatory is supported by NSF grant AST-0229008.
284:
285: \clearpage
286:
287: \begin{thebibliography}{}
288: \bibitem[Backman \& Paresce(1993)]{bac93} Backman, D. E. \& Paresce, F.
289: 1993, in Protostars and Planets III (Tucson: Univ. Arizona Press), 1253
290:
291: \bibitem[Deller \& Maddison(2005)]{del05} Deller, A. T. \& Maddison, S. T.
292: 2005, ApJ, 625, 398
293:
294: \bibitem[Dent et al.(2000)]{den00} Dent, W. R. F., Walker, H. J., Holland,
295: W. S. \& Greaves, J. S. 2000, MNRAS, 314, 702
296:
297: \bibitem[Dowell et al.(2003)]{dow03} Dowell, C. D., et al. 2003, in
298: ``Millimeter and Submillimeter Detectors for Astronomy", ed.
299: T. G. Phillips \& J. Zmuidzinas, Proc. SPIE, 4855, 73
300:
301: \bibitem[Gulliver, Hill \& Adelman(1994)]{gul94} Gulliver, A. F., Hill, G.
302: \& Adelman, S. J. 1994, ApJ, 429, L81
303:
304: \bibitem[Holland et al.(1998)]{hol98} Holland, W. S. et al. 1998,
305: Nature, 392, 788
306:
307: \bibitem[Holland et al.(2005)]{hol05} Holland, W. S. et al. 2005,
308: in preparation
309:
310: \bibitem[Koerner, Sargent \& Ostroff(2001)]{koe01} Koerner, D. W.,
311: Sargent, A. I. \& Ostroff, N. A. 2001, ApJ, 560, L181
312:
313: \bibitem[Marsh et al.(2005)]{mar05} Marsh, K. A., Velusamy, T., Dowell, C. D.,
314: Grogan, K. \& Beichman, C. A. 2005, ApJ, 620, L47
315:
316: \bibitem[Pollack et al.(1994)]{pol94} Pollack, J. B., Hollenbach, D.,
317: Beckwith, S., Simonelli, D. P., Roush, T. \& Fong, W. 1994, ApJ, 421, 615
318:
319: \bibitem[Su et al.(2005)]{su05} Su, K. Y. L. et al. 2005, ApJ, 628, 487
320:
321: \bibitem[Wilner et al.(2002)]{wil02} Wilner, D. J., Holman, M. J.,
322: Kuchner, M. J., \& Ho, P. T. P. 2002, ApJ, 569, L115
323:
324: \bibitem[Wyatt(2003)]{wya03} Wyatt, M. C. 2003, ApJ, 598, 1321
325:
326: \bibitem[Wyatt(2006)]{wya06} Wyatt, M. C. 2006, ApJ, 639, 1153
327:
328: \end{thebibliography}
329: \clearpage
330:
331: \begin{figure}
332: \epsscale{0.95}
333: \plotone{f1.eps}
334: \caption{Observed and photosphere-subtracted images of Vega at 350 $\mu$m
335: and 450 $\mu$m. The intensity scale (shown by the horizontal bar at the top
336: of each image) is in units of mJy arcsec$^{-2}$ and the orientation is
337: such that north is up and east is to the left. The RMS measurement noise on
338: the images is 0.032 and 0.026 mJy arcsec$^{-2}$ at 350 $\mu$m and 450 $\mu$m,
339: respectively.
340: \label{fig1}}
341: \end{figure}
342: \clearpage
343:
344: \begin{figure}
345: \plotone{f2.eps}
346: \caption{Intensity in the Vega disk at 350 $\mu$m wavelength as a function of
347: azimuth around the ring (measured east from north), both with and without
348: subtraction of the estimated photospheric contribution. Error bars represent
349: the standard deviation of measurement noise. The arrows represent the
350: expected locations of the 4:3 resonance peaks for the dynamical model of
351: \cite{wya03} in which dust is trapped in the mean motion resonances of a
352: suspected planet. \label{fig2}}
353: \end{figure}
354: \clearpage
355:
356: \begin{figure}
357: \epsscale{0.44}
358: \plotone{f3.eps}
359: \caption{The estimated line-of-sight optical depth of the Vega disk at a
360: reference wavelength of 350 $\mu$m, based on observed images at 350 $\mu$m,
361: 450 $\mu$m, and 70 $\mu$m, assuming a two-component dust grain model
362: in which the grain radii are 1 mm and 18 $\mu$m. The upper two panels show the
363: optical depth distribution for the two grain components separately, while
364: the lower panel shows the sum of both components. Also shown on the
365: lower plot are the locations of previously reported emission peaks
366: from longer wavelength data as follows: (filled triangle)
367: 850 $\mu$m---\citet{hol98}; ($\times$) 1.3 mm---\citet{wil02}; ($+$)
368: 1.3 mm---\citet{koe01}. \label{fig3}}
369: \end{figure}
370: \clearpage
371:
372: \begin{figure}
373: \epsscale{1.0}
374: \plotone{f4.eps}
375: \caption{Radial profiles of estimated optical depth for the two grain
376: components, characterized by grain radii of 1 mm (filled circles) and
377: 18 $\mu$m (open circles).
378: \label{fig4}}
379: \end{figure}
380:
381: \begin{figure}
382: \epsscale{0.69}
383: \plotone{f5.eps}
384: \caption{Results of dynamical modeling of the Vega dust disk showing the
385: predicted intensity distributions of dust emission at 350 $\mu$m and 850 $\mu$m.
386: Following \cite{wya03} we assumed a Neptune-mass planet at 65 AU, and
387: further assumed that the planet orbits
388: clockwise, is currently located NW of the star and that its 4:3 resonance is
389: populated by smaller grains (1 mm) than the other resonances (1 cm).
390: {\em Upper two panels:\/} The 350 $\mu$m emission resulting from (a) the
391: combined effects of the 2:1(u) and 3:2 resonances, and (b) the
392: 4:3 resonance. The planet is indicated by an encircled cross.
393: {\em Lower two panels:\/} Predicted observational images at (c) 350 $\mu$m and
394: (d) 850 $\mu$m. Images (c) and (d) were produced by adjusting the
395: relative amounts of material in the two grain populations (1 mm and 1 cm)
396: to provide the best match to the observed images; this required
397: equal amounts of material (by cross-sectional area) in each.
398: In this model, the 1 mm grains (4:3 resonance) contribute $80\pm16$\% of
399: the observed flux at 350 $\mu$m and $60\pm12$\% at 850 $\mu$m,
400: while the 1 cm grains (2:1(u) \& 3:2 resonances) contribute
401: 20\% at 350 $\mu$m and 40\% at 850 $\mu$m.
402: All of the above images have been
403: smoothed to the spatial resolution of the observations,
404: corresponding to FWHM of $10''$ at 350 $\mu$m (CSO/SHARC II; present paper) and
405: $14''$ at 850 $\mu$m (JCMT/SCUBA; \citet{hol98}).
406: On the linear pseudocolor intensity scale, white corresponds to peak intensity
407: for each individual image.
408: \label{fig5}}
409: \end{figure}
410: \clearpage
411:
412: \end{document}
413:
414: