1: %% --------------------------------------------------------------

2: %% epsilon Eri disk Spitzer Fab4 GTO paper

3: %% Dana Backman,  SOFIA / SETI Institute

4: %% submitted 25 May 2008

5: %% resubmitted 5 September 2008

6: % --------------------------------------------------------------

7: 

8: %% ----------------------- Header -------------------------------

9: %%

10: %\documentclass[manuscript]{aastex}		% double spaced

11: \documentclass[12pt,preprint]{aastex} 		% single spaced

12: %\documentclass[preprint2]{aastex}		% two columns

13: 

14: %% ---------------------- Packages ------------------------------

15: %%

16: \usepackage{natbib}

17: %% \usepackage{apjfonts}   % nicer fonts which will ps2pdf w/o bitmapping

18: 

19: %% ---------------------- Personal Commands ---------------------

20: %%

21: %% % definitions

22: 

23: \newcommand{\kms}{km s$^{-1}$}

24: \newcommand{\solarmass}{\ensuremath{ \mathnormal{M}_{\Sun} }}

25: \newcommand{\mass}{\ensuremath{ \mathnormal{M} }}

26: \def\asec{$^{\prime\prime}$}

27: \def\amin{$^{\prime}$}

28: \def\Msun{~M$_{\odot}$}

29: \def\Lsun{~L$_{\odot}$}

30: \def\deg{$^{\circ}$}

31: \def\um{~$\mu$m}

32: \def\Teff{T$_{eff}$}

33: 

34: %% ---------------------- Begin Document ------------------------

35: %%

36: 

37: \begin{document}

38: 

39: \title{Epsilon Eridani's Planetary Debris Disk: \\

40: Structure and Dynamics based on \\

41: Spitzer and CSO Observations}

42: 

43: %\slugcomment{revised and resubmitted September 5, 2008} 

44: \shortauthors{Backman et al. } 

45: \shorttitle{Epsilon Eridani Debris Disk}

46: 

47: \author{D. Backman}

48: \affil{SOFIA \& SETI Institute}

49: \author{M. Marengo}

50: \affil{Harvard-Smithsonian Center for Astrophysics}

51: \author{K. Stapelfeldt}

52: \affil{Jet Propulsion Laboratory, California Institute of Technology}

53: \author{K. Su}

54: \affil{Steward Observatory, University of Arizona}

55: \author{D. Wilner}

56: \affil{Harvard-Smithsonian Center for Astrophysics}

57: \author{C. D. Dowell}

58: \affil{Jet Propulsion Laboratory, California Institute of Technology}

59: \author{D. Watson}

60: \affil{University of Rochester}

61: \author{J. Stansberry, G. Rieke}

62: \affil{Steward Observatory, University of Arizona}

63: \author{T. Megeath\altaffilmark{1}, G. Fazio}

64: \affil{Harvard-Smithsonian Center for Astrophysics}

65: \author{M. Werner}

66: \affil{Jet Propulsion Laboratory, California Institute of Technology}

67: 

68: \altaffiltext{1}{Now at the University of Toledo (Ohio)}

69: 

70: \begin{abstract}

71: 

72: {\it Spitzer} and Caltech Submillimeter Observatory (CSO) images

73: and spectrophotometry of $\epsilon$ Eridani at wavelengths from

74: 3.5 to 350 $\mu$m reveal new details of its bright debris disk.

75: The 350 $\mu$m map confirms the presence of a ring at

76: $r =$ 11--28\asec~(35--90 AU) observed previously at

77: longer sub-mm wavelengths.  The {\it Spitzer}

78: mid- and far-IR images do not show the ring, but rather

79: a featureless disk extending from within a few arcsec of the

80: star across the ring to $r \sim$ 34\asec~(110 AU). 

81: 

82: The spectral energy distribution (SED) of the debris system

83: implies a conplex structure. A model constrained by the surface

84: brightness profiles and the SED indicates that the sub-mm ring

85: emission is primarily from large ($a \sim$ 135 $\mu$m) grains,

86: with smaller ($a \sim$ 15 $\mu$m) grains also present in and

87: beyond the ring.  The {\it Spitzer} IRS and MIPS SED-mode

88: spectrophotometry data clearly show the presence of spatially

89: compact excess emission at $\lambda \ga 15~\mu$m that requires

90: the presence of two additional narrow belts of dust

91: within the sub-mm ring's central void. The innermost belt at

92: $r \sim$ 3 AU is composed of silicate dust.

93: 

94: A simple dynamical model suggests that dust produced collisionally

95: by a population of about 11 M$_\oplus$ of planetesimals in the

96: sub-mm ring could be the source of the emission from both in

97: and beyond the sub-mm ring.  Maintaining the inner belts and the

98: inner edge to the sub-mm ring may require the presence of three

99: planets in this system including the candidate radial velocity

100: object.

101: 

102: \end{abstract}

103: 

104: \keywords{

105: infrared: stars -- circumstellar matter -- planetary systems -- 

106: stars: individual (epsilon Eridani)}

107: 

108: \section{Introduction}

109: \label{sec:intro}

110: 

111: The very nearby (3.22 pc; HIP 16537) K2 V star $\epsilon$ Eridani

112: (HR 1084, HD 22049, GJ 144) was found by IRAS to have a 

113: bright semi-resolved far-infrared excess due to circumstellar dust

114: (Gillett 1984; Aumann 1985).  The system shows $\geq 1$ Jy

115: of excess at 60--200 $\mu$m (IRAS 03305-0937; Walker \& Heinrichsen

116: 2000 and references therein), and a moderate-significance excess at

117: 25 $\mu$m.

118: 

119: VLTI angular diameter measurements plus evolutionary models

120: imply an age for $\epsilon$ Eri of 850 Myr (Di Folco et al.\ 2004),

121: consistent with previous determinations based on Ca II activity

122: (Henry et al.\ 1996), Li abundances (Song et al.\ 2000),

123: and gyrochronology (Barnes 2007).

124: For $\epsilon$ Eri, as well

125: as the other IRAS-discovered prototype debris disk systems

126: $\alpha$ Lyr (Vega), $\alpha$ PsA (Fomalhaut), and $\beta$ Pic,

127: (Gillett 1984; 1986) later dubbed the ''Fab Four'', estimated

128: time scales for dust destruction by Poynting-Robertson (P-R) drag and

129: mutual collisions are significantly less than the respective system ages.

130: Thus, the observed dust is not primordial but must

131: be ''second-generation'' material released relatively

132: recently, by collisions or other activity, from larger parent bodies.

133: These debris disks are therefore understood to

134: represent late stages, or remnants, of the planet formation

135: process.  Many more debris disks have been discovered since the

136: prototypes (see reviews by Backman \& Paresce 1993 (Protostars

137: and Planets III); Lagrange et al.\ 2000 (Protostars and Planets

138: IV); Meyer et al.\ 2007 (Protostars and Planets V)).

139: $\epsilon$ Eri is especially interesting for its

140: proximity, its age of nearly 1 Gyr, and its K main sequence spectral

141: type for which relatively few debris disks are presently known

142: (Trilling et al.\ 2008).

143: 

144: The $\epsilon$ Eri system was mapped at 850 and 450 $\mu$m

145: by Greaves et al.\ (1998; 2005), revealing a nearly circular ring

146: extending more than 25\asec~(80 AU) from the star, with peak

147: surface brightness at about 18\asec~($\sim$ 60 AU) and a central

148: cavity with radius about 10\asec~($\sim$ 30 AU).

149: Sch\"{u}tz et al. (2004) detected the ring at 1.2 mm and confirmed

150: its spatial extent.

151: Marengo et al.\ (2006)

152: made a deep search at wavelengths of 3.5 to 7.9 $\mu$m

153: for planetary-mass companions between 14\asec~(45 AU) and

154: 2.9\amin~(560 AU) from the star with {\it Spitzer}'s IRAC camera,

155: and also set upper limits to extended emission.

156: Proffitt et al.\ (2004) used HST/STIS observations to determine an

157: upper limit to optical scattered light from the region of the

158: sub-mm ring.

159: Di Folco et al.\ (2007) set an upper limit

160: on near-IR (K' band) scattered light brightness

161: within 1\asec~(3 AU) of $\epsilon$ Eri

162: via CHARA interferometric measurements.

163: 

164: Dent et al.\ (2000), Li, Lunine \& Bendo (2003),

165: and Sheret, Dent \& Wyatt (2004) calculated disk models

166: constrained by the surface brightness distribution in

167: the sub-mm images and pre-{\it Spitzer} SED data,

168: examining alternatives of porous or solid

169: grain structures and silicate or silicate-plus-ice compositions.

170: Bright spots in the sub-mm ring have been interpreted as

171: dust density enhancements caused by a planet orbiting in or

172: near the ring (Liou \& Zook 1999; Ozernoy et al.\ 2000;

173: Quillen \& Thorndike 2002; Deller \& Maddison 2004;

174: Greaves et al.\ 2005).

175: 

176: Tentative detection of a Jovian-mass planet

177: orbiting $\epsilon$ Eri with

178: a semi-major axis of 3.4 AU, well inside the sub-mm ring,

179: has been made via radial velocity measurements

180: (Campbell et al.\ 1988; Hatzes et al.\ 2000), possibly

181: confirmed astrometrically (Benedict et al.\ 2006).

182: Das \& Backman (1992) suggested that

183: such a planet might influence the location of the warm

184: dust detected by IRAS.

185: Detailed models of the gravitational effect of a planet 

186: close to the star on possible nearby dust distributions

187: were calculated by Moran, Kuchner \& Holman (2004).

188: 

189: The present paper is part of the {\it Spitzer} all-instrument GTO

190: ''Fab Four'' program making comprehensive studies of the prototype

191: planetary debris disks, to put our solar system into evolutionary

192: context and to search for signs of planets

193: (Fomalhaut:  Stapelfeldt et al.\ 2004;

194: Vega: Su et al.\ 2005;

195: $\beta$ Pic:  Chen et al.\ 2007).

196: {\it Spitzer} obtained the first far-IR images

197: of $\epsilon$ Eri, which yield broadband

198: photometry and disk surface brightness

199: profiles as well as high-resolution comparisons with prior

200: sub-mm maps.  Those {\it Spitzer} images are combined here

201: with medium-resolution IR spectroscopy plus a 350 $\mu$m map

202: from the CSO SHARC II camera.

203: Together, these data provide a unique opportunity

204: to study the nearest known exo-Kuiper belt and exo-zodiacal

205: dust system in detail.

206: 

207: Section 2 of this paper describes the {\it Spitzer} and CSO

208: observations of $\epsilon$ Eri and the data reduction.

209: Section 3 addresses detailed analyses of the data including

210: first-order corrections of image and spectrophotometric data

211: plus determination of the photosphere SED and IR excess.

212: Section 4 describes derivation of

213: a model of grain spatial and size distributions

214: fitted to the disk's multi-$\lambda$ surface brightness

215: distributions and over-all SED.

216: Section 5 discusses implications of the model results,

217: compares them with other observations and SED-based

218: models of the $\epsilon$ Eri disk,

219: and presents results of a simple dynamical model

220: calculation connecting $\epsilon$ Eri's disk properties

221: with the history of our solar system's Kuiper belt.

222: 

223: \section{Observations and Data Reduction} 

224: \label{sec:obs+redux}

225: 

226: {\it Spitzer} observations of $\epsilon$

227: Eri were taken in Guaranteed Time Observing Program 90 (P.I.

228: Michael Werner). 

229: Data obtained with various instruments and facilities were used as follows:

230: (1) {\it Spitzer} IRAC image photometry measures the photospheric SED

231: from 3.6 to 7.9 $\mu$m, wavelengths that show no detectable contribution

232: from the debris disk.

233: (2) {\it Spitzer} MIPS data at 24, 70, and 160 $\mu$m

234: provide mid- and far-IR images of the disk.

235: (3) {\it Spitzer} MIPS SED-mode data give source flux

236: densities from 55 to 90 $\mu$m with low spectral resolution,

237: plus source size in one dimension versus wavelength.

238: (4) {\it Spitzer} IRS spectra link the mid-IR

239: broadband photometric points,

240: determine the shortest wavelength of excess (highest

241: abundant dust temperature, minimum dust orbit radius),

242: and allow a search for mineralogical features.

243: (5) The CSO 350 $\mu$m map can be compared

244: with the {\it Spitzer} far-IR images as well as

245: with prior 450 and 850 $\mu$m maps.

246: Surface brightness profiles from the 24, 70, 160 and 350 $\mu$m

247: data together strongly constrain the disk model.

248: (6) The {\it Spitzer} MIPS images and CSO map are also integrated

249: to yield broadband photometric points for the entire disk,

250: supplemented by IRAS catalog photometry using re-calculated color

251: corrections, plus previously published ISO photometry.

252: 

253: \subsection{Spitzer IRAC data}

254: \label{sec:irac_data}

255: 

256: Images at 3.6 to 7.9 $\mu$m of $\epsilon$ Eri and its surroundings

257: were made with {\it Spitzer}'s InfraRed Array Camera (IRAC;

258: Fazio et al.\ 2004) in 2004 January and February.

259: Those observations were designed to look 

260: for wide-separation low-mass companions, but also provided accurate

261: photometry of the star via Point Spread Function (PSF) fitting

262: of non-saturated portions of $\epsilon$ Eri and standard star images.

263: Details of the observations and data reduction procedures

264: are described by Marengo et al.\ (2006).

265: The IRAC data were reprocessed for the present paper to take

266: advantage of an updated instrument calibration plus

267: availability of additional reference star observations

268: to generate a new PSF.

269: 

270: Each of the two IRAC observations consisted of a 36-position dither

271: of the source on the array, for a total of 2 x 3380 seconds integration time.

272: The new PSF reference stars were $\epsilon$ Indi (K4.5 V), Vega, Fomalhaut

273: and Sirius.  (The debris disks around Vega and Fomalhaut

274: are not detected at IRAC wavelengths.)

275: The total field of view had a width of 5.78\amin~due to the dither,

276: slightly larger than the IRAC frame width of 5.21\amin.  Data 

277: reduction and calibration were done using the {\it Spitzer}

278: Science Center (SSC) 

279: pipeline version S15 combined with the post-BCD IRACproc software

280: (Schuster, Marengo \& Patten 2006).

281: 

282: \subsection{Spitzer MIPS data}

283: \subsubsection{MIPS 24, 70 and 160 $\mu$m Images}

284: 

285: Observations of $\epsilon$ Eri were made using all three

286: imaging channels of the Multiband Imaging 

287: Photometer for {\it Spitzer} (MIPS; Rieke et al.\ 2004).

288: Images at 24 $\mu$m were made on 2004 January 29 using two

289: different observing strategies.  (1) The standard 

290: 24 $\mu$m MIPS photometry dither pattern was executed at four cluster target 

291: positions in a square pattern 8.29\asec~(3.25 pixels) wide.

292: This procedure

293: yielded an exposure time of 4$\times$(16$\times$3) seconds

294: for the region within  

295: 2.5\amin~of the star, with a high degree of sub-pixel sampling to 

296: facilitate PSF subtraction.  Observations of

297: the K4 III star HD 217382 made on 2003 November 19 with the identical

298: dither/cluster 

299: target pattern were used as a PSF reference.

300: (2) To probe a wider field of view for possible companions, another

301: $\epsilon$ Eri 24 $\mu$m dataset was taken using a $3\times3$ raster map.  

302: The resulting mosaic covers a 0.25\deg$\times$0.25\deg~region 

303: centered on the star to an exposure depth of 2$\times$(16$\times$3) seconds.

304: 

305: Two types of MIPS 70 $\mu$m image observations were also taken

306: on 2004 January 29.  (1) With the default pixel scale

307: ($9.96$\asec/pixel), eight cycles of the large-field photometry

308: dither pattern were used, providing a total exposure 

309: time of 151 seconds per pixel.

310: (2) With the fine pixel scale ($5.2$\asec/pixel), one cycle of the small 

311: field photometry dither pattern was made at each of 12 cluster target 

312: positions, arranged in a rectangular grid with 16.22\asec~(3.25-pixel)

313: spacing, for an effective exposure time of 12$\times$(10$\times$3)

314: seconds per pixel.

315: On 2005 April 4 a deeper fine scale image was obtained with an

316: effective exposure time of 12$\times$(10$\times$10) seconds per pixel.

317: 

318: Two MIPS 160 $\mu$m image datasets were obtained.

319: (1) The first set was taken on 2004 January 29, using 2 cycles of

320: small-field photometry at each of 9 cluster target positions spaced

321: on a 36\asec~rectangular grid, and 3-second exposures, for an exposure

322: time of 9$\times$27 seconds per pixel.  (2) Another set

323: was obtained on 2006 Feb 19 using the same observing

324: strategy except that 10-second exposures were used,

325: for an exposure time of 9$\times$90 seconds.

326: 

327: The MIPS 160 $\mu$m camera suffers from a spectral 

328: leak by which stray light in the wavelength range 1-1.6 $\mu$m

329: produces a false image partially overlapping the actual

330: 160 $\mu$m source image.

331: This ghost image is several times brighter than the real disk

332: emission in the case of $\epsilon$ Eri, and must be subtracted

333: accurately to enable study of the circumstellar emission.

334: The spectral leak is field-dependent, and therefore can only be 

335: properly subtracted using a reference star that was observed with the

336: same dither  

337: parameters as the science target.  Pre-launch observation planning did

338: not anticipate this requirement, so no appropriate reference star

339: observation was available initially to leak-subtract the $\epsilon$ Eri

340: dataset.

341: 

342: Good leak subtraction was achieved using 160 $\mu$m

343: images taken on 2004 August 6

344: of the debris disk star $\tau$ Ceti (type G8)

345: with the same dither pattern as for $\epsilon$ Eri.

346: Only the spectral leak, and not the photosphere or disk excess,

347: is detected at 160 $\mu$m from $\tau$ Ceti

348: (Stapelfeldt et al., in preparation).

349: The subtraction procedure involved empirically determining the maximum

350: normalization factor for the leak reference source such that its subtraction

351: from the science target did not produce noticeable residuals below the

352: background level.  The subtractions were performed using mosaics

353: in detector coordinates of all the dithered observations.

354: Observations of Achernar ($\alpha$ Eri) were

355: also obtained for $\epsilon$ Eri leak subtraction reference,

356: on 2005 August 30.  The Achernar data did not support a good

357: leak subtraction for $\epsilon$ Eri,

358: probably because of the large color difference between

359: the two stars (type B3 versus K2).  

360: 

361: The MIPS instrument team's Data Analysis Tool (DAT; Gordon 

362: et al.\ 2005) was used for basic reduction (dark subtraction, flat

363: fielding/illumination correction) of the SSC S11 data products.

364: The 24 $\mu$m images were processed 

365: to remove a vertical ''jailbar'' pattern along the detector columns

366: that appears during observations of bright sources.  The known transient 

367: behaviors associated with the MIPS 70 $\mu$m array were removed

368: from the default-scale data by median column 

369: subtraction and time-filtering the images with an excluded region of 15

370: pixels ($\sim$ 150\asec) centered on the source

371: (for details see Gordon et al.\ 2007).  For the 70 $\mu$m

372: fine-scale data, off-source exposures were subtracted from

373: preceding on-source exposures to remove the time-dependent

374: background column offsets.

375: Flux calibration factors of

376: 0.0454 MJy sr$^{-1}$ (24 $\mu$m),

377: 702 MJy sr$^{-1}$ (70 $\mu$m default-scale),

378: 2894 MJy sr$^{-1}$ (70 $\mu$m fine-scale), and

379: 44.7 MJy sr$^{-1}$ (160 $\mu$m)

380: per MIPS raw data unit

381: were applied to the MIPS image data

382: (Engelbracht et al.\ 2007; Gordon et al.\ 2007;

383: Stansberry et al.\ 2007).

384: The processed data were then combined using

385: World Coordinate System (WCS) information to produce final mosaics

386: with pixels half the size of the physical pixel scale.

387: 

388: \subsubsection{MIPS SED-mode Data}

389: 

390: MIPS Spectral Energy Distribution (SED-mode) data

391: were obtained on 2005 September 4 using 10 cycles of

392: 10 second exposures with a +3\amin~chopper throw. The total

393: integration time was 629 seconds.  Basic reduction of the data to produce

394: calibrated versions of the individual exposures was done using version

395: 3.06 of the MIPS DAT code.  MIPS SED-mode observations provide

396: chopped pairs of images.  The off-source images were pair-wise

397: subtracted from the on-source images, and the results mosaicked

398: to produce an image of the 55--90 $\mu$m spectrum of $\epsilon$ Eri.

399: The mosaic was generated using 4.9\asec~pixels, yielding an image

400: 65 pixels across in the spectral dimension by 44 pixels in the spatial

401: dimension.  This image was then boxcar-smoothed over 5 adjacent rows,

402: resulting in spectral resolution $R =$ 15--20.

403: 

404: \subsection{Spitzer IRS data}

405: 

406: Mid-infrared spectra of $\epsilon$ Eri were obtained in

407: 2004 February, 2004 August, and 2005 January with the

408: {\it Spitzer} Infrared Spectrograph (IRS; Houck et al.\ 2004).

409: Using the low-spectral resolution

410: ($R \sim$ 60--120)

411: spectrographs SL1 (7.5--14 $\mu$m), LL2 (14--21 $\mu$m)

412: and LL1 (20--38 $\mu$m), observations were made with

413: five slit positions separated

414: by half a slit width (1.8\asec) in SL1 and

415: ten slit positions separated by half a slit width

416: (4.8\asec) in LL2 and LL1.

417: The star was centered in the scan pattern in each case.

418: Staring-mode observations were also obtained using the

419: low-resolution SL2 (5.3--7.5 $\mu$m) and high resolution

420: ($R \sim$ 600) SH (10--20 $\mu$m) and

421: LH (19--37 $\mu$m) modules.  Those observations were at two nod

422: positions, with the star displaced from the center of the slit

423: by a third of the slit length.

424: Total integration times per slit position or nod were 84 seconds

425: in the SL1 spectral map, 70 seconds each in the LL2 and LL1 spectral

426: maps, and 54 seconds, 72 seconds, and 144 seconds respectively

427: for the SL2, SH and LH observations.

428: Each set of observations was preceded by a high-accuracy

429: pointing peak-up observation of the neighboring star SAO 130582

430: using the IRS Blue (15 $\mu$m) camera.

431: 

432: Data reduction began with products of the {\it Spitzer}

433: Science Center's S11 IRS data pipeline.  Further processing

434: involved removal, by interpolation in the spectral direction, of

435: permanently bad and ''rogue'' pixels identified in the IRS

436: dark-current data for all campaigns up to and including the

437: one in which each $\epsilon$ Eri spectrum was taken.

438: The IRS instrument team's SMART software was then used to extract

439: 1-D spectra from the 2-D long-slit (SL, LL) or echelle (SH, LH) images.

440: Sky emission in the SL and LL observations was measured by a linear fit

441: to the signals at points along the slit farther than 40\asec~from the star,

442: and subtracted as part of the extraction.  The

443: $\epsilon$ Eri system appeared to be spatially unresolved

444: in the SL observations, so an extraction window was used that was

445: fitted to the instrumental point-spread function at the longest

446: wavelengths of each low-resolution module, 4 pixels wide

447: in each case.  A full-slit extraction was performed for

448: the high-spectral-resolution (SH and LH) observations.

449: The sky emission measured in the

450: low-spectral-resolution observations was small enough compared

451: with the flux received from $\epsilon$ Eri and its disk

452: that sky subtraction in the SH and LH observations

453: could be neglected.

454: 

455: Reference SL and LL observations of $\alpha$ Lac (A1 V)

456: plus SH and LH observations of $\xi$ Dra (K1 III)

457: were obtained in the same manner as the $\epsilon$ Eri observations.

458: The spectra of $\epsilon$ Eri were calibrated via division by

459: spectra of the respective reference star taken at the corresponding

460: nod position, multiplication by a synthetic template spectrum of the

461: reference star (Cohen 2004, private communication), and tben nod-averaging.

462: Reduced spectra from each spectrograph module were then

463: combined to produce complete 5.3--38 $\mu$m spectra.

464: Small spectrograph module-to-module differences

465: at wavelengths of overlap were assumed to be due to minor

466: telescope pointing-induced throughput losses, and removed by

467: scaling up the spectra from smaller-slit modules to

468: match those from the larger-slit modules in the overlap range.

469: 

470: Linearity and saturation in the IRS detector arrays is corrected

471: in the data pipeline, but $\epsilon$ Eri and its disk are bright

472: enough that linearity and saturation effects in the

473: low-spectral-resolution modules are

474: large and the corrections uncertain.  The main impact

475: of these effects is that the LL spectra

476: had significantly lower effective signal-to-noise ratio and

477: spectrophotometric accuracy than the SH and LH observations

478: due to residual inaccuracy of the LL saturation correction.

479: The final spectrum used in this paper was assembled from

480: SL, SH and LH data; LL data will not be discussed further.

481: The uncertainty in flux density is 5\% below

482: about 10 Jy, increasing to about 10\% for flux

483: densities as high as 20 Jy.

484: 

485: \subsection{CSO Data}

486: \label{sec:cso_data}

487: 

488: Observations of $\epsilon$ Eri at 350 $\mu$m

489: were made during parts of three nights in

490: 2003 January and two nights in 2005 December using the

491: SHARC II 12$\times$32 bolometer array (Dowell et al.\ 2003)

492: at the Caltech Submillimeter Observatory (CSO).

493: The weather conditions were excellent, with 225 GHz (1.3 mm)

494: zenith opacity ranging from 0.028 to 0.048.

495: The total integration time on the source was 11 hours.

496: Data were obtained by rapid scanning (without chopping)

497: in Lissajous patterns with periods of 14--20 seconds

498: and amplitudes of $\pm$15\arcsec~to $\pm$40\asec~in each

499: of two axes.

500: 

501: Each SHARC II detector spans 4.6\asec, and the

502: diffraction-limited beam size is 9\asec~(FWHM), with

503: a PSF determined by coadding images from

504: a variety of calibration sources

505: interspersed with the $\epsilon$ Eri observations.

506: Data reduction used the {\it sharcsolve} software package that

507: iteratively decomposes fixed source emission, time variable

508: atmospheric emission, and detector offsets.  Best results for

509: subtraction of the temporally and spatially variable telluric

510: emission were achieved by assuming that the average emission

511: outside a 35\asec~radius from the star was zero, at the cost

512: of filtering out possible more widely extended emission.

513: Calibration was based

514: on the integrated map signals within a 35\asec~radius, following

515: correction for atmospheric absorption, from sub-mm

516: standards OH231.8 (18 Jy), CRL618 (19 Jy), o Ceti (2.3 Jy),

517: Ceres (13 Jy in 2003 January), and Juno (8.0 Jy in 2005 December).

518: The uncertainly in the absolute flux scale is estimated to

519: be 25\%, and the RMS noise level in the reduced 350 $\mu$m

520: image is $\sim 3$ mJy per smoothed beam.

521: 

522: \section{Results}

523: \label{sec:results}

524: 

525: \subsection{Images and Spatial Structure}

526: 

527: \subsubsection{MIPS Far-IR Images}

528: 

529: The top row of Figure 1 displays the MIPS 24, 70 and 160 $\mu$m

530: images after standard {\it Spitzer} pipeline processing.

531: The bottom row of Figure 1 shows the output from a further level of

532: processing.  Specifically: (1) Figure 1d is the result of subtracting

533: a scaled PSF from the 24 $\mu$m image in 1a, (2) Figure 1e shows a

534: 70 $\mu$m fine pixel-scale image for comparison with the default

535: pixel-scale image in 1b, and (3) Figure 1f is the result of compensating

536: for the 160 $\mu$m filter's short-wavelength leak apparent in 1c.

537: Radial profiles of the processed images are presented and discussed

538: in Section 4.2.1.

539: 

540: The 24 $\mu$m PSF-subtracted data show an image core

541: size of 5.6\asec$\times$5.4\asec, consistent with

542: the values of 5.52\asec$\times$5.48\asec~for an unresolved

543: blue (photospheric) source determined by Su et al.\ (2008).

544: The stripes in the processed data are PSF subtraction

545: residuals, mainly noise arising from different sampling

546: of the source and PSF reference by the pixel grid.

547: Examination of the 24 $\mu$m PSF wings shows evidence

548: for slight excess surface brightness

549: at the position of the first Airy dark ring ($r \sim$ 7\asec)

550: but with only $\sim1\sigma$ significance.

551: 

552: The far-IR source is resolved at 70 and 160 $\mu$m.

553: At 70 $\mu$m (fine pixel-scale image)

554: the observed source has a compact core

555: with FWHM of 24.2\asec$\times$22.3\asec~elongated

556: approximately north-south, superimposed on a relatively

557: flat pedestal of emission roughly 1\amin~in diameter.

558: The core 70 $\mu$m emission lies mostly {\it interior} to

559: the sub-mm ring, and is clearly extended relative to

560: the size of unresolved point sources, which is

561: 15.3\asec$\times$14.4\asec~(Su et al.\ 2008).

562: The leak-subtracted 160 $\mu$m source can be approximated

563: by Gaussian profiles with FWHM 62\asec$\times$52\asec,

564: also elongated north-south.

565: This distribution is also extended

566: relative to a point source (39\asec~diffraction FWHM).

567: The stellar point source contributes about

568: 10\% of the total system flux density at 70 $\mu$m

569: and 5\% at 160 $\mu$m.

570: 

571: The 70 and 160 $\mu$m images contain a substantial surprise.

572: Although the far-IR source FWHM is about the size of the sub-mm ring,

573: the far-IR images do not show the ring-with-deep-hole

574: aspect observed at longer wavelengths.

575: Instead, the far-IR surface brightness distribution is continuous

576: across the central region and the sub-mm ring,

577: extending well beyond the ring,

578: with no significant enhancement at the position of the ring.

579: The far-IR source is close to circularly symmetric,

580: with possible slight extension north-south as in the

581: 350 and 450 $\mu$m source maps, but unlike the 850 $\mu$m maps

582: that are extended east-west or northeast-southwest

583: (Greaves et al.\ 2005).

584: 

585: To test for far-IR detection

586: of surface brightness features seen in the sub-mm ring,

587: an azimuthal average of the 70 $\mu$m fine-scale image

588: was subtracted from the image itself.

589: Figure 2a shows the result of this analysis.

590: The subtracted image is smooth to $\la$ 20\% of

591: the surface brightness in the ring, smaller than the

592: amplitude of features in prior sub-mm maps.

593: The 70 $\mu$m image therefore does not show

594: the azimuthal surface brightness variations

595: observed at longer wavelengths. A similar result, albeit with

596: lower signal-to-noise ratio, is obtained with the

597: 160 $\mu$m image (Figure 2b).

598: 

599: \subsubsection{MIPS SED-mode Source Size}

600: 

601: Figure 3a shows the MIPS SED-mode source FWHM

602: (along the slit) versus wavelength from Gaussian fits

603: to the spatial profiles of the signal at each row of the mosaic image.

604: The FWHM are 21, 23.5, and 38\asec~at 55, 70 and 90 $\mu$m,

605: respectively, consistent with the 70 and 160 $\mu$m image source sizes.

606: Analysis of spectrophotometric data from the

607: MIPS-SED instrument, presented in Figure 3b,

608: is described in Sections 3.3.1 and 4.3.2.

609: The small-scale variations in FWHM and in

610: flux density versus wavelength relative to simple

611: monotonic trends, as well as the steep increase in FWHM

612: longward of 87 $\mu$m, are not real but represent

613: systematic uncertainties in the reduction procedures,

614: which dominate photometric uncertainties.

615: 

616: \subsubsection{CSO Sub-mm Map}

617: 

618: Figure 4 displays the CSO SHARC II 350 $\mu$m map.

619: The sub-mm ring that has been observed previously at

620: 450 and 850 $\mu$m is apparent here, with approximately the same spatial scale.

621: The 350 $\mu$m image has flux consistent with zero excess in

622: the central beam (FWHM $=$ 29 AU)

623: after subtraction of a model 8 mJy point source corresponding

624: to the stellar photosphere emission, representing

625: 2\% of the total flux in the map (Table 1).

626: The 350 $\mu$m ring may show north-south extension, as in

627: the {\it Spitzer} 70 and 160 $\mu$m images and the JCMT

628: 450 $\mu$m map, but differing from the 850 $\mu$m morphology.

629: 

630: The average value and dispersion of the surface brightness around

631: the ring at 350 $\mu$m is S$_{\nu} = 0.15 \pm 0.04$ mJy per square arcsec.

632: The brightest 350 $\mu$m spot, in the north to north-east sector of

633: the ring, has S$_{\nu}$ $\sim$ 0.24 mJy per square arcsec,

634: about 2$\sigma$ above the average.

635: The location and shape of the 350 $\mu$m ring are confirmed by

636: the 2003 and 2005 data sets analyzed independently, but the

637: azimuthal substructure is not.

638: 

639: \subsection{Broadband Photometry}

640: 

641: \subsubsection{Stellar Photosphere}

642: 

643: Precise determination of $\epsilon$ Eri's photospheric SED was

644: necessary to measure the excess emission from the disk, especially

645: in the mid--IR where the system SED is only slightly above the stellar

646: continuum.

647: The star has m$_v$ = +3.74,  B-V = +0.88 (Mendoza et al.\ 1978)

648: and V-K of +2.11 (Koornneef 1983a).

649: Average main sequence B-V colors are +0.82 and

650: +0.92 for K0 and K2 temperature classes, respectively (Johnson 1966).

651: Average V-K colors are +2.00 and +2.25 

652: for K1 and K2, respectively (Koornneef 1983b).  These data

653: imply that $\epsilon$ Eri is a little bluer than the average K2 V

654: star, with interpolated temperature class between K1 and K2.

655: Flower (1996) gives T$_{eff}$ = 5090 K for B-V = 0.88.

656: 

657: Flux densities from $\epsilon$ Eri at 3.6, 4.5,

658: 5.7 and 7.9 $\mu$m determined by PSF-fitting of {\it Spitzer}

659: IRAC data (Section 2.1) have ratios nicely consistent with expectations

660: for a photosphere of this star's temperature and luminosity.

661: The photosphere model SED used in this paper is a Kurucz

662: model for T = 5000 K (nearest match to 5090 K) and log g = 4.5,

663: normalized to the four IRAC flux densities

664: via least-squares minimization.  The RMS of the fit to those

665: points is 1.5\%.  The result is consistent with ground-based

666: photometry at 2.2 $\mu$m and shorter wavelengths

667: to better than 2\% precision.  Figure 5 displays the IRAC

668: data points and fitted photosphere in relation to other

669: broadband measurements discussed below.  Table 1

670: presents extrapolated photospheric flux densities which

671: can be regarded as having 2\% uncertainty.

672: 

673: \subsubsection{Photometry from Image Data}

674: 

675: Since the disk is not resolved at 24 $\mu$m, the total emission

676: from the disk at that wavelength was

677: obtained using both aperture photometry and PSF-fitting photometry

678: optimized for point sources; for details see Su et al.\ (2006).

679: A total of $2.04 \pm 0.04$ Jy was estimated for the star and disk

680: combined, which is 18\% above the expected photosphere.

681: No color correction was applied at 24 $\mu$m because the observed

682: SED slope in that wavelength region implies a MIPS filter correction

683: factor of $\sim$ 1.

684: 

685: The total flux density of the system at 70 and 160 $\mu$m was

686: integrated over a circular aperture with radius

687: 64\asec~centered at the star.

688: The total disk emission in the 70 $\mu$m and 160 $\mu$m bands

689: is respectively 1.50 and 0.93 Jy after application of

690: color correction factors of 0.893 and 0.971

691: appropriate for a 55 K blackbody.

692: 

693: IRAS photometry of $\epsilon$ Eri spans an important range of

694: wavelengths, from the red end of the photospheric SED to beyond

695: the peak of the excess. 

696: The {\it Spitzer} IRS spectrum, MIPS images and MIPS SED-mode

697: data allow a refined

698: estimate of the continuum slope across the IRAS 25 and 60 $\mu$m bands

699: and hence more accurate color corrections, which were applied here.

700: 

701: Figure 5 and columns 1 \& 2 of Table 1 present the merged

702: system (star+disk) broadband SED

703: including {\it Spitzer} MIPS photometry from

704: image data, integrated flux density from the CSO 350 $\mu$m map,

705: newly color-corrected IRAS catalog flux densities,

706: ISO 200 $\mu$m flux density,

707: and previously published integrated flux densities

708: at 450 and 850 $\mu$m from Greaves et al.\ (2005).

709: The peak of the SED is between $\lambda$ $\sim$ 70 and 100 $\mu$m.

710: 

711: ISO photometry of $\epsilon$ Eri (Walker \& Heinrichsen 2000)

712: yielded a flux density of 1880 mJy at 200 $\mu$m (cf.\ predicted

713: photosphere value of 25 mJy), far above the 160 $\mu$m point.

714: The {\it Spitzer} 160 $\mu$m image, as well as the 450 and 850 $\mu$m

715: sub-mm maps, show significant background emission extended

716: beyond 40\asec~from the star that would have been included

717: in the large long-wavelength effective beam sizes of both ISO

718: (200 $\mu$m photometric aperture diameter 90\asec) and

719: IRAS (100 $\mu$m detector footprint $\sim$ 180\asec$\times$300\asec).

720: Numerous objects with near-IR colors of galaxies that would be

721: strong sources at $\lambda \ga$ 100 $\mu$m are detected in

722: {\it Spitzer} IRAC images of the $\epsilon$ Eri field

723: (Marengo et al.\ 2006).

724: The presence of this complex background emission probably

725: accounts for the high flux density measured by ISO;

726: the present observations indicate that much of the ISO

727: flux must come from beyond the sub-mm ring.

728: The IRAS 100 $\mu$m and ISO 200 $\mu$m photometry

729: (Figure 5) therefore provide upper limits

730: to the system flux densities at those wavelengths.

731: A less confused view of the source at 100--200 $\mu$m

732: should be possible eventually with deep Herschel PACS and SOFIA HAWC

733: observations.

734: 

735: \subsection{Spectrophotometric Data}

736: \label{sec:spectrophotometry}

737: 

738: \subsubsection{MIPS SED-mode}

739: 

740: With $\epsilon$ Eri expected to appear as an extended source along

741: the MIPS SED slit, simple aperture photometry measurements cannot be

742: made without admitting significant background noise.  To address

743: that problem, a Gaussian was fit to the spatial profile of the signal

744: along each row of the SED mosaic.  The integrated product of the Gaussian

745: width and height gives the source response at each wavelength.

746: The resulting $\epsilon$ Eri spectrum in instrumental units was converted

747: to physical units by reducing and extracting an archival spectrum of the

748: calibration star Canopus ($\alpha$ Car, G8 III) in the same way.

749: The Canopus spectrum

750: was normalized to give a flux density of 2.15 Jy at 70 $\mu$m

751: (the star's 3.11 Jy source flux density multiplied by a 69\% slit

752: throughput factor appropriate for a point source).

753: The result was then corrected for

754: the spectral response function by assuming Canopus has a Rayleigh-Jeans

755: spectral slope in the far-IR.  The flux normalization and spectral

756: response function derived from Canopus were then applied to the

757: $\epsilon$ Eri spectrophotometry.

758: 

759: Figure 3b presents the resulting MIPS SED-mode spectrum 

760: from 55 to 90 $\mu$m, giving the detected flux density transmitted

761: through the MIPS SED slit independent of assumptions about the true

762: spatial extent of the source.

763: The slope of the spectrum, declining to longer wavelengths,

764: is due to the intrinsic shape of the star+disk SED

765: combined with wavelength-dependent slit losses as the size

766: of the diffracted and extended emission region varies relative

767: to the 20\asec-wide MIPS SED entrance aperture.

768: Linking the MIPS SED-mode spectrophotometric data

769: to a disk model is described in Section 4.3.2.

770: 

771: \subsubsection{IRS}

772: 

773: {\it Spitzer} IRS spectrophotometric data for the disk plus stellar

774: photosphere are included in Figure 5.

775: The photosphere model described in Section 3.2.1

776: is overplotted for comparison.

777: A constant offset of 173 mJy was subtracted from the IRS

778: data, determined by making the average excess above the

779: photosphere model from 5 to 12 $\mu$m equal zero.

780: This is consistent with the observed lack of significant excess

781: emission in bradband photometry at 5.7, 7.9 and 12 $\mu$m (Table 1).

782: That offset represents only 1--3\% of the total source

783: flux in that wavelength range, well within the IRS calibration

784: uncertainty.

785: The resulting rescaled IRS mid-IR SED is in close

786: agreement with the MIPS 24 $\mu$m image photometry

787: as well as with the lower-precision IRAS 25 $\mu$m broadband point.

788: 

789: The Figure 5 panel inset shows the

790: spectrum from 10 to 30 $\mu$m after subtraction of the photosphere.

791: Significant departure of the

792: $\epsilon$ Eri SED from the photosphere begins at about 15 $\mu$m.

793: The mid-IR SED is remarkable in having a very steep rise from

794: 15 to 20 $\mu$m, a ''plateau'' of almost constant flux density

795: from about 20 to 30 $\mu$m, and then the beginning of another steep

796: rise beyond 30 $\mu$m.

797: 

798: \section{Modeling}

799: \label{sec:modeling}

800: 

801: \subsection{General Considerations}

802: \label{sec:general}

803: 

804: The data presented above provide new information on the

805: $\epsilon$ Eri circumstellar dust spatial distribution, new

806: spectral information regarding warm dust components, and

807: separation of disk emission from confusing background sources.

808: A model of particle spatial distributions and physical properties

809: can now be produced that will allow improved understanding of dust

810: lifetimes and dynamics, which in turn allows inferences about

811: locations of planetary companions, and about the evolution of this

812: system for comparison with our solar system.

813: 

814: The modeling process involved: (1) derivation of

815: intrinsic 70, 160 and 350 $\mu$m 1-D radial surface brightness

816: profiles that, when convolved with the respective PSFs, match the

817: observed radial profiles; (2) calculation of a physical model

818: for the outer disk with specific grain spatial distributions,

819: size distributions and compositions that yields the intrinsic

820: 70, 160, and 350 $\mu$m surface brightnesses inferred from the

821: first step; (3) derivation of another physical model for warm

822: material detected by IRAS and {\it Spitzer} IRS

823: but not resolved in the imaging data, then testing that the

824: predicted long-wavelength emission from the inner

825: material would be consistent with the 70, 160 and 350 $\mu$m

826: radial profiles; and finally, 4) checking that the over-all

827: physical model, including both resolved and unresolved material,

828: correctly predicts the observed spectral energy distribution of IR excess

829: from 15 to 850 $\mu$m and also the MIPS SED-mode narrow-slit

830: flux densities plus 1-D source sizes or limits.

831: As mentioned in Section 3.3.1, the MIPS SED fluxes need to

832: be corrected for slit losses, and that can only be done

833: by estimating the corrections iteratively using synthetic

834: images derived from the model (Section 4.3.2).

835: 

836: The 350 $\mu$m map shows a deficit of sub-mm emission

837: at $r \la 35$ AU relative to 35--90 AU, similar to previous sub-mm maps.

838: In contrast, the 24, 70 and 160 $\mu$m image profiles and photometry

839: plus mid-IR spectrophotometry and broadband photometry,

840: as well as the original IRAS detection, together

841: indicate the presence of significant amounts of warm material lying

842: at $r \la 35$ AU, within the central void in the sub-mm ring.

843: 

844: The surface brightness in the high-S/N 70 $\mu$m images

845: is continuous and monotonic out to

846: r $\ga$ 45 arcsec (140 AU), far outside the sub-mm ring.

847: However, trial PSF convolutions

848: of model intrinsic brightness distributions

849: revealed that the outer boundary of the 70 $\mu$m

850: emission is actually located at $r \sim$ 34\asec~(110 AU),

851: with apparently further extended emission representing the PSF wings.

852: In comparison, nearly all of the observed sub-mm surface

853: brightness is contained in the ring between about 35 and 90 AU

854: (11--28\asec) (Greaves et al.\ 2005; also, see

855: Sch\"{u}tz et al.\ 2004 regarding the source diameter at 1.2 mm).

856: The region outside the sub-mm ring, beyond $r = 28$\asec~(90 AU),

857: is here designated the ''halo''.

858: 

859: An outer limit to the 160 $\mu$m emission is more difficult to

860: determine due to background source confusion discussed above,

861: but the data and preliminary modeling indicate that the over-all

862: spatial distribution of material observed at 160 $\mu$m much more

863: closely resembles the 70 $\mu$m than the 350 $\mu$m emission.

864: Significant emission at 350 $\mu$m

865: does not extend beyond 28\asec~(90 AU).

866: 

867: \subsection{Outer Fully Resolved Disk}

868: 

869: \subsubsection{Radial Surface Brightness Profiles}

870: 

871: The 70 $\mu$m images have small ellipticity

872: and also are azimuthally smooth (Figure 2),

873: justifying uses of 1-D average radial profiles centered

874: on the star position for detailed comparison of model

875: surface brightness distributions to observations.

876: Based on inspection of the radial profiles and preliminary

877: modeling, fiducial radii of $r =$ 11, 28, and

878: 34\asec~(35, 90, and 110 AU), representing transition

879: points in some of the surface brightness distributions,

880: were chosen as primary points for matching models to observations.

881: 

882: Intrinsic surface brightness profiles at 70, 160 and

883: 350 $\mu$m were constructed by choosing surface brightness

884: values at the three fiducial radii and interpolating between those

885: points with simple power laws.

886: Production of model PSF-convolved radial profiles for

887: trial-and-error comparison with the

888: observations involved expansion of model 1-D radial profiles

889: into 2-D pseudo-images, convolution of each pseudo-image with

890: the respective wavelength's 2-D PSF derived for a point source

891: with $\epsilon$ Eri's temperature using STinyTim (Krist 2006),

892: and then collapsing the convolved model 2-D map back into a 1-D

893: profile.  After each trial, differences between predicted

894: and observed surface brightness profiles were used to estimate

895: new values at the fiducial radii for the next trial.

896: 

897: The results of modeling the 24, 70, 160 and 350 $\mu$m

898: surface brightness profiles are displayed in Figure 6.

899: The actual locations of changes in profile slopes near the

900: fiducial radii are uncertain to probably 10\% in radius, i.e.\ the

901: modeling process had approximately that much spatial resolution.

902: Note that these intrinsic surface brightness profile

903: determinations were aimed merely to match the 

904: observed profiles, without reference at this step

905: in the modeling process to physically motivated grain

906: properties, which are treated in the next section.

907: 

908: \subsubsection{Physical Grain Models}

909: 

910: The Heidelberg DDS code (Wolf \& Hillenbrand 2005)

911: was used to estimate grain properties

912: and surface densities that would yield the surface brightness values

913: determined in the preceding modeling step.

914: Grains were assumed to be solid (non-porous) and spherical.

915: DDS settings represented

916: $\epsilon$ Eri as a 5090 K blackbody with 0.33\Lsun.

917: Choice of grain composition, grain minimum and maximum sizes,

918: size distribution power law exponent, and mass surface densities

919: at the three fiducial radii allowed calculation of emitted surface

920: brightness for comparison

921: with the intrinsic 70, 160 and 350 $\mu$m surface brightness

922: values determined at those locations.

923: Integrated disk flux densities and grain masses

924: could then be calculated using

925: surface densities and grain properties interpolated

926: between the values at the fiducial radii.

927: 

928: The large difference in appearance of the sub-mm maps versus

929: the {\it Spitzer} 70 and 160 $\mu$m images led to a search

930: for model disk components such that:

931: (1) material having high sub-mm and low far-IR

932: emissivity would dominate in the sub-mm ring region,

933: coexisting with (2) material having low sub-mm and high far-IR

934: emissivity extending across the ring into the halo region.

935: Many combinations of grain properties, size distributions

936: and spatial distributions can match those constraints.

937: A simple model can be built of

938: components that separately fit the 70 \& 160 $\mu$m images

939: and the 350 $\mu$m map.

940: 

941: The final model for the resolved disk combines:

942: (1) large (radii $a =$ 100--200 $\mu$m) amorphous H$_2$O ice

943: grains located in the sub-mm ring, plus (2) small

944: ($a =$ 6--23 $\mu$m) ''astronomical'' silicate grains having

945: a broader spatial distribution that includes the ring.

946: Experiments varying the value of the grain size

947: distribution exponent indicated that the conventional

948: collision equilibrium value of $x = -3.5$ could be used

949: for both populations although other values are not excluded.

950: Net emissivities integrated over the size distributions

951: are approximately equivalent to single-size particles with

952: radii of 135 $\mu$m and 15 $\mu$m for the large icy and

953: small silicate grains, respectively.

954: 

955: \subsection{Two Inner Warm Belts}

956: 

957: \subsubsection{Dust Properties and Locations}

958: 

959: {\it Spitzer} IRS, MIPS, and MIPS-SED data

960: reveal the presence of material within the

961: sub-mm ring's central void.

962: PSF analysis of 70 and 160 $\mu$m image surface brightness

963: profiles shows emission closer to the star than

964: 11\asec~(35 AU) but also reveals that the emission

965: does not extend all the way to $r = 0$.

966: The local maximum and plateau in the IR excess

967: around 20--30 $\mu$m indicate emission at $T \sim 100-150$ K,

968: confirming the location of material at $r < 35$ AU, because at 35 AU,

969: $T_{dust}$ would be only $\sim$ 40 K for blackbody grains

970: or $\sim$ 60 K for 1-$\mu$m silicate grains.

971: Quadrature subtraction of a point source

972: size from the MIPS-SED 1-D source extent at 55 $\mu$m

973: (Figure 3a) yields an emission radius of

974: very roughly 8\asec~(25 AU).

975: 

976: Modeling the properties of the inner material involved

977: first noting that the marked change in SED slope at about 20 $\mu$m

978: (Figure 5) is easily explained as a silicate mineralogical

979: band (e.g.\, Chen et al.\ 2006).  Trial-and-error experiments with

980: the Heidelberg DDS code showed that the detectability of the band sets

981: a rough upper limit on grain radius $ a \leq 20/2\pi$ $\sim$ 3 $\mu$m.

982: In contrast, "generic" grains of that size without mineralogical attributes

983: produce an emission SED that is significantly broader than observed,

984: as do ice grains.

985: 

986: The 15-30 $\mu$m segment of the $\epsilon$ Eri SED can be matched

987: approximately by emission from a belt of warm ''astronomical''

988: silicate grains with characteristic size $\sim$ 3 $\mu$m,

989: located about 3 AU ($\sim 1$\asec) from the star.

990: This innermost warm belt is responsible for $\ga$ 95\%

991: of the system's excess emission at 25 $\mu$m first

992: detected by IRAS.  Note that dust located 3 AU from $\epsilon$ Eri

993: would be entirely contained within the IRS slits

994: that have a minimum width of 4.7\asec~(15 AU).

995: 

996: The precise position of the ''3 AU'' belt is constrained

997: by relations between grain size, radiative efficiency,

998: and temperature. For example, if the model grain size $a$ is

999: set at 2.2 $\mu$m instead of 3 $\mu$m, the belt must move to

1000: $r \sim 3.8$ AU to have the correct color temperature.

1001: However, grain efficiency declining steeply to longer

1002: wavelengths for such small grains would

1003: unacceptably ''erode'' the 20--30 $\mu$m SED plateau.

1004: Thus, the grain size and belt location are

1005: determined to perhaps 25\% precision.

1006: There is no separate information regarding

1007: an upper limit to the belt's width,

1008: but that can be taken to be equivalent to the uncertainty

1009: in belt location, i.e. $\sim$ 25\%.

1010: 

1011: The rising IRS spectrum near 35 $\mu$m (Figure 5)

1012: cannot be fit by emission from the material inferred at $\sim$ 3 AU,

1013: which has an SED decreasing beyond 30 $\mu$m.  The steep upward slope

1014: from 35 to 60 $\mu$m, characterized by a color temperature of about 55 K,

1015: plus the relatively narrow SED peak at 70--100 $\mu$m, locate the

1016: next lower-temperature population of grains also

1017: in a belt with narrow ranges of temperature and particle size.

1018: The precise properties of this material are

1019: especially constrained by the MIPS SED-mode 55--90 $\mu$m

1020: flux densities, but because of slit losses, those data

1021: must be renormalized using models of the source

1022: geometry, as described in the next section.

1023: 

1024: A belt of $a \sim 8 \mu$m ''astronomical'' silicate grains

1025: located at r $\sim 20$ AU ($\sim 6$\asec) satisfies

1026: the SED and source size constraints.

1027: Models placing the belt at 20 AU

1028: can be distinguished from alternate locations at 10 or 30 AU.

1029: Material at 20 AU could also explain the slight filling-in

1030: of the first Airy dark ring at 24 $\mu$m, $r~\sim$ 23 AU

1031: (Figure 6a).

1032: Silicate grains of this size and temperature would not

1033: show mineralogical features in the {\it Spitzer} data,

1034: so the composition of this belt is consistent with,

1035: but not determined to be, silicate.

1036: 

1037: Models with continuous distributions of material extending

1038: across the sub-mm void produced too much total mid- and far-IR

1039: flux and/or did not match the surface brightness profiles

1040: and source size measurements. The warm material is clearly

1041: restricted to annular zones with relatively empty gaps.

1042: 

1043: \subsubsection{Modeling Details, Including MIPS SED Normalization}

1044: 

1045: Convergence on final models for the two inner warm belts

1046: involved iterative comparison of model SEDs from the

1047: Heidelberg DDS software with the IRS 15-35 $\mu$m

1048: and 55--90 $\mu$m (MIPS SED-mode) flux densities not already

1049: accounted for by the model of the resolved disk.

1050: Trial models yielded intrinsic surface brightness profiles

1051: at the observed wavelengths that were:

1052: (1) convolved with the 24 $\mu$m PSF to compare with the observed

1053: unresolved source,

1054: (2) convolved with a derived {\it Spitzer} 55 $\mu$m PSF

1055: and MIPS SED-mode slit mask to compare with the observed

1056: source 55 $\mu$m 1-D FWHM, and

1057: (3) added to the already-established 70, 160 and 350 $\mu$m

1058: intrinsic surface brightness profiles of the disk and then

1059: PSF-convolved to check that the profiles were not significantly

1060: perturbed by the extra warm material.

1061: 

1062: The model properties of these two belts within the sub-mm void

1063: interact, in that they both produce significant flux density at

1064: $\lambda$ $=$ 30--35 $\mu$m.  For example, changing the second

1065: warm belt's properties toward smaller spatial radius and/or

1066: particle radii produces greater 30--35 $\mu$m model emission,

1067: requiring the innermost belt to have greater spatial radius

1068: and/or particle radii so as to preserve the total mid-IR SED shape.

1069: 

1070: The slit losses and

1071: flux normalization of the MIPS SED data were derived

1072: by generating 2-D synthetic images of the source using model radial profiles

1073: of the disk computed at the MIPS SED wavelengths. This analysis was restricted

1074: to the two extremes of the MIPS SED wavelength range, 55 and 90

1075: $\mu$m, using the midpoint at 70 $\mu$m as

1076: the flux normalization reference.

1077: 

1078: Synthetic 2-D

1079: images at 55, 70 and 90 $\mu$m were derived by convolving azimuthal

1080: projections of the radial disk model profiles with the STinyTIM PSF, and then

1081: adding a point source rescaled to the stellar flux at each wavelength.

1082: The resulting 70 $\mu$m model ''image'' closely matches the

1083: 70 $\mu$m observations (Figure 6b).

1084: A software mask with the size of the MIPS SED slit was then

1085: centered on the synthetic

1086: images to measured the fraction $q_\nu$ of the total flux recovered

1087: through the SED-mode slit:  58\%, 47\% and 38\% respectively at

1088: 55, 70 and 90 $\mu$m.

1089: The overall flux normalization factor $N^{flux}_{70}$ at 70

1090: $\mu$m is derived by rescaling the SED in Figure 3b to have

1091: a 70 $\mu$m flux density identical to the MIPS 70 $\mu$m image total

1092: flux density (1.688 Jy, Table~1)

1093: which yields $N^{flux}_{70} \simeq 1.78$. The

1094: correction for the other two wavelengths can be derived

1095: by factoring out the slit loss at 70 $\mu$m and multiplying

1096: by the slit losses at 55 or 90 $\mu$m:

1097: 

1098: % equation 1

1099: \begin{equation}

1100: N^{flux}_\nu = N^{flux}_{70} \times \frac{q_{\nu}}{q_{70}}

1101: \end{equation}

1102: 

1103: \noindent

1104: This procedure produced corrected MIPS SED flux densities

1105: of $1.62 \pm 0.25$ Jy at 55 $\mu$m and $1.64 \pm 0.25$ Jy at 90 $\mu$m.

1106: Uncertainties were estimated by quadrature addition of the

1107: 70 $\mu$m uncertainty to the uncertainties in the uncorrected

1108: flux densities plotted in Figure 3b.

1109: Once the photospheric contribution is subtracted, an excess from the

1110: disk of 1.30 $\pm$ 0.25 Jy at 55 $\mu$m and 1.52 $\pm$ 0.25 Jy

1111: at 90 $\mu$m is obtained.

1112: 

1113: \subsection{Model Summary}

1114: 

1115: As described in the preceding section,

1116: the over-all model of $\epsilon$ Eri's circumstellar material

1117: includes:  (1) particles with low far-IR emissivity

1118: and high sub-mm emissivity, consistent with the properties of

1119: radius $a =$ 100--200 $\mu$m (effective $a \sim$ 135 $\mu$m)

1120: amorphous H$_2$O ice grains in the sub-mm ring at $r =$ 35--90 AU,

1121: (2) particles with high far-IR emissivity and low

1122: sub-mm emissivity, consistent with the properties of

1123: $a =$ 6--23 $\mu$m (effective $a \sim$ 15 $\mu$m)

1124: ''astronomical'' silicate grains at $r =$ 35--110 AU,

1125: corresponding to the sub-mm ring plus an exterior halo,

1126: (3) a narrow belt at $\sim$ 3 AU (T $\sim$ 120 K) of small

1127: ($a \sim$ 3 $\mu$m) silicate grains, and

1128: (4) a narrow belt at $\sim$ 20 AU (T $\sim$ 55 K) of small grains

1129: ($a \sim$ 8 $\mu$m) of undetermined, but possibly silicate,

1130: composition.

1131: Specific properties of the model components are presented

1132: in Table 2.

1133: 

1134: Figure 7 displays the SED of the complete model and the

1135: separate disk components compared with the photometric and

1136: spectrophotometric data (model photosphere SED subtracted).

1137: {\it Spitzer} mid- and far-IR data especially

1138: reveal a complicated SED shape that strongly constrains

1139: the temperatures, locations, and grain sizes of warm unresolved

1140: material.  The model is not unique but was built from the fewest

1141: components with the simplest assumptions that produced

1142: a good match to all the available data.

1143: Many alternate models were tested, resulting in confidence that:

1144: (a) 70 and 160 $\mu$m emission does not extend beyond 110 AU;

1145: (b) 350 $\mu$m emission does not extend beyond 90 AU;

1146: (c) 350 $\mu$m emission does not extend inside 35 AU;

1147: (d) the dominant emission in the sub-mm ring is not

1148: from silicate particles;

1149: (e) 70 and 160 $\mu$m emission within 35 AU

1150: must include non-emitting gaps, with emission

1151: restricted to annular zones; and,

1152: (f) grains in the innermost warm belt

1153: at $r \sim 3$ AU have silicate composition.

1154: 

1155: The model rises less steeply than the IRS

1156: observations at 10--18 $\mu$m.  The excess flux in

1157: the model is less than 3\% of the emission from

1158: the system in that range.  This may be related to uncertainty

1159: in the photospheric model subtraction at short wavelengths

1160: where the IR excess is barely significant, and may also

1161: indicate the presence of crystalline silicates, with a

1162: sharp 20 $\mu$m spectral feature, combined with

1163: the amorphous silicates assumed in the model.

1164: 

1165: The top panel of Figure 8 displays the model radial profile

1166: of perpendicular optical depth $\tau_\perp$, i.e.\ fractional

1167: (unitless) surface density ($m^2$ of grain cross-sectional

1168: area per $m^2$ of disk surface area; e.g.\ Backman 2004)

1169: and in the bottom panel the corresponding mass surface density

1170: inferred from the optical depth and model grain properties.

1171: The slopes of the two functions across the sub-mm ring and halo

1172: are not equal because the

1173: model includes smooth variation in the particle size distribution

1174: between the fiducial radii, so the ratio of particle cross section

1175: (and thus optical depth) to mass surface density varies slightly

1176: with radius.  The optical depths and mass densities of the two

1177: inner rings in Figure 8 are calculated by assuming

1178: $\Delta$$r$ values of 1 AU at 3 AU and 2 AU at 20 AU;

1179: they would scale inversely with $\Delta$$r$.

1180: 

1181: The total inferred mass in {\it detected} grains,

1182: i.e.\ with sizes

1183: from a few microns to a few hundred microns, carrying most

1184: of the radiating surface area, is roughly $5 \times 10^{-3}$

1185: M$_\oplus$ $\sim$ 0.4 M$_{\rm Luna}$,

1186: 90\% of which is in the large (''icy'') grain

1187: component of the sub-mm ring.

1188: The total bolometric fractional luminosity ($L_d/L_*$)

1189: of all the disk components is 1.1$\times$10$^{-4}$.

1190: 

1191: Figure 9 is a 2-D representation of the $\epsilon$ Eri

1192: debris disk system in linear spatial scale, showing the

1193: sub-mm ring, halo, and inner warm belts.

1194: The position of the innermost warm belt is compared

1195: in the inset figure with one orbit solution for the candidate

1196: radial velocity planet (Section 5.4).

1197: 

1198: \section{Discussion}

1199: \label{sec:discussion}

1200: 

1201: \subsection{Some Implications of the Disk Model}

1202: 

1203: The derived surface density profile for the

1204: small grain component (Figure 8) has a nearly constant value

1205: from 35 to 110 AU, consistent with dynamics controlled

1206: by Poynting-Robertson radiation drag or its corpuscular

1207: wind analog (Section 5.3).  In contrast, the large grain

1208: component has surface density rising from 35 to 90 AU.

1209: The larger grains are less subject to drag forces, so

1210: their radial distribution may reflect

1211: that of the underlying parent body population.

1212: 

1213: The void interior to the sub-mm ring is inferred to be partly

1214: filled by two additional belts of particles with mean sizes

1215: in both cases smaller than in the sub-mm ring.

1216: The large grain population in the ring appears to be absent

1217: in the void, even though 100 $\mu$m-size ice grains could

1218: persist against sublimation for as long as $\epsilon$ Eri's age,

1219: 850 Myr, at distances as small as 3 AU from the star

1220: (see equations 16 and 17 in Backman \& Paresce 1993

1221: for expressions of ice sublimation time scales).

1222: Moreover, the mid-IR SED, plus the 70 $\mu$m surface brightness

1223: profile in particular, indicate that there are gaps in the distribution

1224: of material from just beyond 3 AU to almost 20 AU,

1225: and from just beyond 20 AU to 35 AU.

1226: A mechanism is needed to prevent significant amounts

1227: of both small and large grains from drifting via P-R drag from

1228: the sub-mm ring into the inner void (Section 5.4).

1229: 

1230: The total mass of detected grains in $\epsilon$ Eri's innermost

1231: belt is about 0.1$\times$ the mass estimated for the

1232: warm belt detected around the 230 Myr-old A star

1233: $\zeta$ Lep (Chen \& Jura 2001).

1234: The size and structure of the $\epsilon$ Eri system, with

1235: two warm belts in a central void surrounded by a cold disk,

1236: and silicate dust identified in the innermost ring, bear

1237: striking resemblance to the circumstellar material

1238: in the much younger HD 113766 system (10-16 Myr, F3/F5 binary)

1239: (Lisse et al.\ 2008).

1240: 

1241: An upper limit to the ratio of ice versus silicate grain mass

1242: for the sub-mm ring region in which they co-exist is about 16.

1243: This ratio is intermediate between values for

1244: solar composition ($\sim$ 50) and for

1245: comets ($\sim$ 2-5, e.g.\ Altenhoff et al.\ 2002).

1246: 

1247: \subsection{Comparison with Prior Studies of $\epsilon$ Eri}

1248: 

1249: \subsubsection{Sub-mm Ring Features}

1250: 

1251: As noted above, the sub-mm ring at 350 $\mu$m (Figure 4)

1252: has the same inner and outer diameters observed in previously

1253: published 850 and 450 $\mu$m maps (Greaves et al.\ 1998; 2005),

1254: but 350 $\mu$m ring surface brightness features do not in

1255: general align with 450 or 850 $\mu$m features.

1256: The (JCMT / SCUBA) 450 and 850 $\mu$m data 

1257: were obtained via dithered (''jiggle''-)

1258: mapping, whereas the (CSO / SHARC II) 350 $\mu$m map was made

1259: via continuous spiral scanning.  Real features should

1260: be robust against such differences in data acquisition and

1261: reduction procedures.

1262: 

1263: The strongest 450 and 850 $\mu$m feature, about 1.5$\times$

1264: brighter than the ring average, is in the south-east sector.

1265: A south-east feature appears in some form in all the maps

1266: and was suggested to be a real part of the $\epsilon$ Eri

1267: system by Greaves et al.\ (2005) based on its motion

1268: over the 1997--2002 time span. Poulton, Greaves \& Cameron

1269: (2006) made a careful statistical analysis of this issue.

1270: Only a weak feature is seen at that location in the 350 $\mu$m

1271: image.  Other fairly prominent 850 $\mu$m ring features,

1272: especially in the south-west sector,

1273: were considered by Greaves et al.\ to be background sources

1274: although they have no clear counterparts at 450 $\mu$m.

1275: 

1276: The strongest 350 $\mu$m feature is

1277: located in the north-north-east section of the ring, with

1278: a surface brightness of about 0.24 mJy per square arcsec.

1279: Its location corresponds approximately to a weak

1280: 450 $\mu$m feature and a moderately strong 850 $\mu$m

1281: feature that could also be part of the $\epsilon$ Eri

1282: system, considering proper motion plus orbital motion.

1283: Recall (Section 3.1.3) that the 350 $\mu$m data,

1284: when split into equal parts

1285: and composed into separate maps, have about the same level

1286: of disagreement regarding azimuthal features as

1287: between maps at different wavelengths, and between

1288: the 850 $\mu$m maps for 1997--2002 versus 2000--2002 

1289: (Greaves et al.\ 2005).

1290: As discussed in Section 3.1.3 and displayed in Figure 2,

1291: no surface brightness features are seen in the ring at

1292: 70 and 160 $\mu$m.

1293: 

1294: Real density features need not have the same surface brightness

1295: ratios versus wavelength as the main ring.  For example,

1296: smaller grains trapped in planetary

1297: resonances can have increased libration widths, smoother

1298: distributions, and show less conspicuous clumping than

1299: large grains, so that some features prominent at long

1300: wavelengths may be less apparent at short wavelengths

1301: (Wyatt 2006).  Azimuthal features with surface brightness

1302: that does not vary progressively with wavelength, however,

1303: are difficult to reconcile with physical models.

1304: 

1305: The disagreement between details of the different sub-mm maps

1306: illustrates the difficulty of working at wavelengths with

1307: large variable sky opacity and  emission.

1308: These comparisons imply that the {\it Spitzer} far-IR images

1309: and CSO 350 $\mu$m maps cannot confirm the ring

1310: surface brightness features reported at 450 or 850 $\mu$m.

1311: 

1312: \subsubsection{Optical \& Near-IR Upper Limits}

1313: 

1314: The brightness of scattered light from circumstellar material can be

1315: estimated from:

1316: 

1317: % equation 2

1318: \begin{equation}

1319: R_{\lambda, r} = (~\tau_\perp~/~4\pi~)~(~d^2_{\rm pc}~/~r^2_{\rm

1320:  AU}~)~(~P~\phi(\gamma)~/~cos(i)~) 

1321: \end{equation}

1322: %

1323: 

1324: \noindent

1325: in which R is the ratio, at a given wavelength and position,

1326: of circumstellar surface brightness per square arcsec

1327: to the star's brightness; $\tau_\perp$ is the

1328: perpendicular optical depth of the circumstellar material;

1329: $d_{\rm pc}$ is the system distance from the observer;

1330: $r_{\rm AU}$ is the distance of the material

1331: from the star; $P$ is the geometric albedo; $\phi(\gamma)$

1332: is the phase function value for the star-dust-observer angle $\gamma$;

1333: and $cos(i)$ is the inclination of the disk to the plane of the sky,

1334: $i = 0$ representing face-on presentation.

1335: A value of $cos(i)$ $\sim$ 1 can be used for the nearly face-on

1336: $\epsilon$ Eri disk.

1337: A value $\phi(\gamma)~\sim~0.5$ can be assumed

1338: for particles much larger than the

1339: wavelength of interest in an approximately face-on disk.

1340: 

1341: Proffitt et al.\ (2004) used the STIS spectrographic camera on HST

1342: to search for scattered light from the sub-mm

1343: ring region.  They set an upper limit of $m_{\rm STIS} \ga +25$ for

1344: circumstellar light at $r \sim 55$ AU (17\asec) in a wide bandpass with

1345: $\lambda_{eff} \sim 0.6$ $\mu$m.

1346: Assuming an albedo of $P = 0.1$ for the model large grains,

1347: which have the

1348: highest optical depth in the sub-mm ring region (Figure 6a),

1349: yields a  predicted scattered light surface

1350: brightness at 55 AU almost exactly equal to the HST/STIS

1351: limit.  Assuming instead an albedo of 0.9, possible

1352: for fresh pure ice surfaces, would yield an optical surface

1353: brightness almost 2.5 mags brighter than the limit.

1354: In comparison, the model of the 2nd warm belt

1355: ($r \sim 20$ AU $\sim 6$\asec) predicts, an optical surface

1356: brightness of $m_V \sim +21.7$ per square arcsec assuming

1357: $P = 0.1$ and radial width $\Delta$$r = 2$ AU, brightness

1358: scaling inversely with $\Delta$$r$.

1359: 

1360: Di Folco et al.\ (2007) report near--IR (K' band, 2.1 $\mu$m)

1361: interferometric observations of $\epsilon$ Eri and $\tau$ Ceti 

1362: with the CHARA array.  They resolved the stellar diameters in

1363: both cases and detected circumstellar emission near $\tau$

1364: Ceti but set an upper limit of 0.6\% of the photosphere for

1365: emission within 1\asec~(3 AU) of $\epsilon$ Eri.  The present

1366: model innermost belt predicts a total scattered light brightness

1367: of approximately 3\% of the Di Folco et al.\ upper limit

1368: if the grain near-IR albedo is assumed to be 0.1.  The direct

1369: thermal emission at K' band from that belt, in fact from the

1370: entire disk, would be many orders of magnitude fainter than

1371: the scattered light brightness.

1372: 

1373: \subsubsection{SED Models}

1374: 

1375: Dent et al.\ (2000),

1376: Li, Lunine \& Bendo (2003) (hereafter LLB03),

1377: and Sheret, Dent \& Wyatt (2004) (hereafter SDW04)

1378: combined pre-{\it Spitzer} broadband photometry and sub-mm map data to

1379: produce models of the $\epsilon$ Eri disk with specific grain

1380: spatial distributions, size distributions and compositions.

1381: Given the limited data available to constrain them, those models are

1382: consistent with the model presented in Section 4 of this paper

1383: based on higher-precision observations over a much broader

1384: wavelength range.

1385: 

1386: Dent et al.\ (2000) assumed non-porous

1387: grains, with a simplified emissivity law determined

1388: only by particle size without mineralogical features.

1389: Their model requires material as close as

1390: 10 AU from the star with $\sim10$\% of the surface

1391: density in the main ring.

1392: LLB03 and SDW04 calculated models

1393: assuming real grain mineralogies, including a range of porosities

1394: (fraction of silicate grain volume occupied by vacuum or ice)

1395: ranging from $p =$ 0 to 0.9.  SDW04 included their own

1396: sub-mm photometric data regarding $\epsilon$ Eri

1397: to supplement published maps.

1398: LLB03's best-fit model involved highly porous

1399: ($p = 0.9$) grains, whereas SDW04's suite of

1400: models favored non-porous (solid) grains.

1401: The present model assumes non-porous grains

1402: and agrees qualitatively with the SDW04 result

1403: that the grain size distribution cannot extend down to

1404: 1 $\mu$m without producing too much short-$\lambda$ emission.

1405: SDW04 remarked that the lack of $\mu$m-scale grains

1406: is a puzzle given the inability of $\epsilon$ Eri's

1407: radiation field to eject grains of that size.

1408: Radiation pressure and corpuscular wind effects

1409: are discussed further in Section 5.3.

1410: 

1411: The total circumstellar grain mass derived by LLB03 was

1412: $\sim7 \times 10^{-3}$ M$_\oplus$ $\sim$ 0.6 M$_{Luna}$,

1413: compared with 0.1 M$_{Luna}$ in SDW04's model and

1414: 0.4 M$_{Luna}$ in the current model.  The differences

1415: between these mass estimates result from differing assumptions

1416: about the particle properties and size ranges,

1417: since each group is matching approximately the same

1418: system fluxes, requiring similar total particle

1419: radiating areas.

1420: 

1421: \subsection{Particle Dynamics}

1422: 

1423: \subsubsection{Grain Ejection via Wind versus Radiation Pressure}

1424: 

1425: Small grains are subject to radiation pressure (e.g.\ Burns et al.\

1426: 1979) and corpuscular winds (e.g.\ Plavchan et al.\ 2005)

1427: that tend to remove them from their source

1428: region, either rapidly outward by direct ejection for particles smaller

1429: than the ''blowout'' size, or, for larger particles,  slowly inward

1430: toward encounters with planets or sublimation near the star.

1431: Grains also collide destructively.

1432: Competition between rates of destructive processes versus rate of grain

1433: resupply from larger parent bodies determines a

1434: debris disk's equilbrium density and also whether there is net mass

1435: flow outward or inward from the grain source (e.g.\ Wyatt 2005; Meyer

1436: et al.\ 2007).  These processes will be considered in this and the

1437: following subsection in view of the $\epsilon$ Eri system's

1438: characteristics.  Porous and non-spherical grains would be affected

1439: more strongly by radiation and wind than the spherical and non-porous

1440: grains assumed here, so destruction and drift

1441: time scales calculated below should be considered upper limits.

1442: 

1443: The direct (radial) radiation pressure (force per area) opposing

1444: gravitational attraction between star and grain is:

1445: 

1446: % equation 3

1447: \begin{equation}

1448: P_r(radiation) = L_* / 4 \pi r^2 c

1449: \end{equation}

1450: %

1451: 

1452: \noindent

1453: whereas the direct wind pressure is:

1454: 

1455: % equation 4

1456: \begin{equation}

1457: P_r(wind) = v_{wind} (dM_* / dt) / 4 \pi r^2 

1458: \end{equation}

1459: %

1460: 

1461: \noindent

1462: The wind mass loss rate for $\epsilon$ Eri has been determined to be 

1463: $\sim 3-4 \times 10^{10}$ kg s$^{-1}$

1464: (Stevens 2005; Wood et al.\ 2005),

1465: about 20--30$\times$ the solar value.

1466: Assuming conservatively that young, chromospherically active

1467: $\epsilon$ Eri has the same wind terminal speed of $\sim$ 400 km/sec

1468: as the sun's, radial wind pressure is negligible

1469: in the $\epsilon$ Eri system, less than 5\% as strong as

1470: radiation pressure.  The star's mass/luminosity ratio is low

1471: enough that solid silicate-density particles should be stable

1472: against direct ejection down to sizes of a few $\times$ 0.1 $\mu$m.

1473: 

1474: This analysis indicates that:  (1) the model grain size lower

1475: limits of a few microns in the sub-mm ring and inner unresolved

1476: belts are not defined by radiation pressure or

1477: neutral wind ''blowout'', and (2) radiation and wind

1478: pressures in this system are too weak to move substantial

1479: amounts of material outward, for example from the sub-mm ring

1480: into the halo.

1481: 

1482: \subsubsection{Poynting-Robertson and Wind Azimuthal Drag versus Collisions}

1483: 

1484: The following two equations compare the azimuthal pressure

1485: $P_{\theta}$ on orbiting grains respectively due to

1486: radiation (P-R drag) and a neutral wind.

1487: 

1488: % equation 5

1489: \begin{equation}

1490: P_{\theta}(radiation) = v_{orb} L_* / 4 \pi r^2 c^2 

1491: \end{equation}

1492: %

1493: 

1494: % equation 6

1495: \begin{equation}

1496: P_{\theta}(wind) = v_{orb} (dM_* / dt) / 4 \pi r^2 

1497: \end{equation}

1498: %

1499: 

1500: \noindent

1501: Using the values given above for wind mass and speed,

1502: the radial drift rate for particles around $\epsilon$ Eri

1503: is enhanced by a factor of about 20 relative to the purely

1504: radiation-driven P-R drag rate (numerically similar,

1505: by coincidence, to the $\epsilon$ Eri wind mass loss

1506: rate relative to solar).

1507: 

1508: The timescale for grain-grain mutual collisions is:

1509: 

1510: % equation 7

1511: \begin{equation}

1512: t_{coll} \sim P_{orb} / 8 \tau_\perp

1513: \end{equation}

1514: %

1515: 

1516: \noindent

1517: in which $P_{orb}$ is the orbit period and $\tau_\perp$ is

1518: the perpendicular optical depth.

1519: Cross-velocities as low as a few hundred meters per second

1520: are sufficient to destroy small silicate and ice grains.

1521: The solar system's asteroid and Kuiper belts both

1522: have inclination and eccentricity ranges more than

1523: large enough to result in routinely shattering

1524: collisions, which may also be

1525: assumed to pertain to the $\epsilon$ Eri disk.

1526: 

1527: At the inner edge of the sub-mm ring, $r = 35$ AU,

1528: where P-R drag is most competitive with collisions,

1529: the collision time scale for the large ($a \sim 135$ $\mu$m) grains

1530: is $t_{coll} \sim 2 \times 10^{5}$ yrs, whereas the wind-drag time scale

1531: for inward drift by $\Delta r = 1$ AU is $\sim 8 \times 10^5$ yrs.

1532: Thus, even when enhanced by wind, virtually all the large icy grains

1533: should be destroyed by collisions {\it in situ} and not be able

1534: to drift inward to fill the central void.

1535: For the small grains ($a \sim 15$ $\mu$m), on the other hand,

1536: the collision destruction time scale

1537: (including cross-sectional area of both small and large

1538: grains as targets) is $\sim 1.5 \times 10^{5}$ yrs,

1539: approximately the same as the wind-drag time scale

1540: for these particles to drift 1 AU.

1541: Given that the collision process is statistical, e.g.\

1542: $\sim$ 1\% of grains can be expected to survive 5 $t_{coll}$,

1543: some fraction of the small grain population {\it could}

1544: travel inward from the sub-mm ring into the central void before

1545: colliding with other grains.

1546: 

1547: At the inner and outer edges of the halo,

1548: $r = 90$ and 110 AU respectively,

1549: the collision time scale for the small grains is also

1550: comparable to the $\Delta r = 1$ AU wind-drag drift time.

1551: Thus, if grains are produced in the halo region they would

1552: tend to stay in place or flow inward toward the sub-mm ring,

1553: rather than outward.

1554: 

1555: \subsubsection{Collisional Scattering}

1556: 

1557: Is there a mechanism other than radiation or wind pressure that could

1558: move material from the sub-mm ring to the outer halo, or does the

1559: presence of significant material in the halo require a particle source

1560: (presumably collisions of planetesimals) at $r \ga$ 90 AU?

1561: In fact, scattering of material into new orbits after collisions

1562: could suffice.

1563: 

1564: The {\it vis viva} equation expresses the

1565: orbit properties resulting from velocity other than

1566: circular orbit velocity at a given location,

1567: 

1568: % equation 8

1569: \begin{equation}

1570: v^2 = G M_* ( 2 / r  -  1 / a_{orb})

1571: \end{equation}

1572: %

1573: 

1574: \noindent

1575: in which $v$ is the speed at radius $r$, and $a_{orb}$ is

1576: the orbit semi-major axis.

1577: 

1578: Under what conditions would grain collisions in

1579: the main, heavily populated ring be able to supply

1580: material to the outer halo?

1581: If grains or parent bodies collide at $r$

1582: with average cross-speeds of $\xi \times v_{circ}$,

1583: corresponding to relative orbit inclinations of order

1584: $\delta$$i \sim \pm \xi$ rad and/or eccentricity ranges of

1585: order $\delta$$e \sim \xi/2$, significant numbers of

1586: fragments will leave the collision with

1587: $\delta v \geq \xi$$v$ relative to the

1588: motion of the colliding bodies' center-of-mass.

1589: Collisions at $r \sim$ 90 AU with $\xi = 0.05$

1590: would send some fragments into new orbits reaching

1591: to 110 AU, the outer edge of the halo.

1592: Therefore the extended halo observed at 70 and 160 $\mu$m

1593: could be the result of collisions among planetesimals

1594: and fragments moving with modest departure from

1595: circular planar motion in the sub-mm ring.

1596: If collsions are in fact the mechanism spreading material

1597: from the ring into the halo, the lack of material

1598: beyond 110 AU could indicate an upper limit

1599: on inclinations and eccentricities in the ring.

1600: 

1601: \subsubsection{Particle Dynamics Summary}

1602: 

1603: Radiation plus wind pressure should not be able to eject

1604: solid silicate-density particles larger than sub-micron size

1605: from the $\epsilon$ Eri system, but the disk model has a

1606: deficit of particles smaller than $a \sim 6~\mu$m in the

1607: sub-mm ring.  Consideration of particle destruction time

1608: scales suggests that a time span of order 10$^5$ years may have

1609: elapsed since the last significant injection of small particles.

1610: The small grain population has a spatial distribution indicating

1611: overall control by azimuthal drag. Some small grains

1612: near the inner edge of the sub-mm ring should be able

1613: to drift into the central void faster than their collision

1614: destruction time scale.  The fact that

1615: this does not appear to be happening may indicate the presence

1616: of a barrier planet preventing inward drift (Section 5.4).

1617: 

1618: The large grains in the sub-mm ring have collision time

1619: scales much shorter than drift time scales, so they

1620: are seen near their locations of creation, and their

1621: radial mass density profile may track that of the

1622: planetesimal parent bodies.

1623: 

1624: \subsection{Planets Around $\epsilon$ Eri}

1625: \label{sec:infer_planets}

1626: 

1627: Radial velocity measurements of $\epsilon$ Eri

1628: have suggested the presence of a planetary companion,

1629: hereafter designated Planet A, with M~$\sin{i}=$ 0.86 M$_{Jup}$

1630: and an orbit period of about 6.9 years (Hatzes et al.\ 2000

1631: and references therein).  The corresponding orbital

1632: semi-major axis is 3.4 AU, near the location inferred

1633: for the innermost warm dust belt.

1634: 

1635: The eccentricity of 0.70 $\pm$ 0.04 reported for

1636: the 3.4 AU planet by Benedict et al. (2006) corresponds

1637: to periastron and apoastron distances of 1.0 and 5.8 AU, respectively.

1638: A giant planet with this orbit would quickly clear that region

1639: not only of dust particles but also the parent planetesimal belt

1640: needed to resupply them, inconsistent with the

1641: dust distribution implied by the {\it Spitzer} observations.

1642: The disk's 70, 160 and 350 $\mu$m aspect ratios and position angles

1643: reported here do not correspond to values of those quantities determined

1644: from the 850 $\mu$m map, calling into question the

1645: system inclination value used to combine the

1646: astrometric and radial velocity data.

1647: Butler et al.\ (2006) report a different orbital eccentricity

1648: of 0.25 $\pm$ 0.23, which could

1649: allow the planet to orbit entirely outside the innermost dust belt.

1650: 

1651: As discussed in Section 5.3, some of the small grains in the

1652: sub-mm ring should be able to drift into the central

1653: void and begin to fill it on timescales of $10^6$ to $10^7$ years.

1654: A mechanism is needed to prevent grains from moving inward past

1655: 35 AU.  Liou \& Zook (1999) and Moro-Martin et al.\ (2005) modeled

1656: the ability of massive planets to define the inner edges of

1657: Kuiper belt-like dust rings. A hypothetical planet orbiting

1658: at the edge of the main $\epsilon$ Eri sub-mm ring is here designated

1659: Planet C.  The gravitational influence

1660: of a planet near or in the ring has also

1661: been proposed (Liou \& Zook 1999; Quillen \& Thorndike 2002;

1662: Greaves et al.\ 2005) as the cause of ring density enhancements

1663: corresponding to surface brightness features reported in

1664: sub-mm imaging that may or may not be real (Section 5.2.1).

1665: 

1666: Confinement of a narrow debris belt at $r \sim 20$ AU

1667: could indicate the presence of a third planet, here

1668: designated Planet B.

1669: Deller \& Maddison (2005) calculated dynamical models

1670: of the $\epsilon$ Eri system including effects of planetary

1671: masses, radiation pressure, and stellar wind, and

1672: found that a planet orbiting at the inner edge of the sub-mm

1673: ring would not by itself be able to define the ring's edge.

1674: Rather, the effect of an additional Jovian-mass planet

1675: with semi-major axis $a \simeq 20$ AU is needed.

1676: Greaves et al.\ (2005) proposed that the mass

1677: controlling some of the possible sub-mm ring features

1678: may be located well inside the ring, at $r \sim$ 24 AU,

1679: based on the orbital period suggested

1680: by apparent motion of those features over 7 years.

1681: 

1682: Summarizing all the observations and models bearing

1683: on the presence of planets around $\epsilon$ Eri, there may

1684: be indications of three different planets in the system.

1685: Planet A: the long-suspected but still unconfirmed

1686: Jovian-mass planet with $a_{orb} = 3.4$ AU that may be associated

1687: with the innermost warm debris belt detected by {\it Spitzer}.

1688: Planet B: also perhaps Jovian-mass,

1689: associated with the second warm debris belt at $r \sim 20$ AU

1690: inferred in this paper, also possibly

1691: helping define the sub-mm ring inner edge via 2:1 resonance,

1692: keeping the zone between 20 and 35 AU clear.

1693: (Note that a 1 $M_{Jup}$ planet with

1694: age 0.85 Gyr located 6\asec~from $\epsilon$ Eri would not

1695: have been seen by {\it Spitzer} but could be detected by JWST.)

1696: Planet C: located at $r \sim 35$ AU, with mass less

1697: than a few $\times$ 0.1 $M_{Jup}$ according to dynamical

1698: models, preventing small grains from drifting inward

1699: past the inner edge of the sub-mm ring.

1700: 

1701: \subsection{Simple Collisional-Evolutionary Model}

1702: 

1703: A dynamical model of the $\epsilon$ Eri sub-mm ring

1704: was calculated assuming equilibrium between collisional

1705: production of grains and removal by P-R plus corpuscular wind drag,

1706: based on previous simple dynamical models of the solar system's Kuiper

1707: belt (Backman, Dasgupta \& Stencel 1995) and solar-type circumstellar

1708: debris disks observed by {\it Spitzer} (Meyer et al.\ 2007).

1709: This model indicates that the

1710: sub-mm ring can be maintained by collisions

1711: of 11 M$_\oplus$ of parent bodies with assumed diameters of 10 km

1712: and material densities of 1000 kg m$^{-3}$.  This agrees

1713: with estimates made by Sheret, Dent \& Wyatt (2004) and

1714: Greaves et al.\ (2005) using similar assumptions.  Thus, the

1715: $\epsilon$ Eri system as observed can be in quasi-equilibrium.

1716: 

1717: {\it Spitzer} observations of Vega

1718: revealed a far-infrared disk extending across the position of

1719: a sub-mm ring into an outer halo (Su et al. 2005),

1720: similar to the $\epsilon$ Eri system.  However, the Vega

1721: disk's properties require recent injection of circumstellar material

1722: to explain the abundance of short-lived particles.

1723: The main difference between the two systems is that $\epsilon$ Eri's

1724: luminosity (including wind)

1725: is low relative to Vega's, so there is no

1726: direct ejection of grains, P-R plus wind azimuthal drag rates are

1727: lower, and equilibrium resupply by continuous minor collisions

1728: can be effective.

1729: 

1730: A simple evolutionary model that estimates loss of belt

1731: mass via destruction of planetesimal parent bodies, with subsequent

1732: removal of fragments by the dynamical processes listed above,

1733: was described and applied to the solar system's Kuiper belt,

1734: asteroid belt, and extrasolar analogs by Meyer et al.\ (2007).

1735: The same model, tuned to the properties of the $\epsilon$

1736: Eri system, predicts that for an initial mass of planetesimal

1737: parent bodies in the range 15--30 M$_\oplus$, similar to the

1738: inferred original mass of the Kuiper belt (e.g.\ Stern \& Colwell

1739: 1997), about 11 M$_\oplus$ would remain at a system age of 0.85 Gyr.

1740: In other words, $\epsilon$ Eri's current debris disk is similar

1741: in total mass to what the solar system's Kuiper belt

1742: mass would have been at the age of $\epsilon$ Eri had there

1743: not been a prior Late Heavy Bombardment (LHB) clearing event

1744: (e.g.\ Levison et al.\ 2007).

1745: 

1746: Collisional evolution is dissipative, eventually converging

1747: on a planetesimal population that has a time scale for further

1748: collisions proportional to the system age, nearly independent of the

1749: initial conditions, if there are no external effects.

1750: However, to transform the young solar system's

1751: Kuiper belt or $\epsilon$ Eri's

1752: debris disk into a low-mass system like the

1753: present-day Kuiper belt, with at most a few

1754: $\times$ 0.1 M$_\oplus$ at age 4.5 Gyr (Luu \& Jewitt 2002),

1755: requires additional mass reduction processes

1756: beyond quasi-steady state collisional grinding and debris removal

1757: (Meyer et al.\ 2007 and references therein).

1758: For example, in our solar system, the LHB event,

1759: possibly driven by a late episode of planet migration, may have

1760: destroyed most of the Kuiper belt's mass relatively sudddenly,

1761: and similar events may happen in other systems with ages

1762: of a Gyr or more (Levison et al.\ 2007 and references therein).

1763: Alternatively, a process that is continually at work in our solar

1764: system, planetary pertubations pulling planetesimals out of the

1765: Kuiper belt, could, with the help of collisional grinding,

1766: transform $\epsilon$ Eri's 11 M$_\oplus$ disk at $t =$ 0.85 Gyr

1767: into the solar system's 0.1 M$_\oplus$ Kuiper belt at 4.5 Gyr

1768: if the perturbative removal process

1769: supplementing collisions were to operate with an

1770: exponential time scale of about 1.5 Gyr.

1771: In contrast, $\tau$ Ceti's debris disk has an inferred

1772: mass of 1.2 M$_\oplus$ at an age of 8 Gyr, so it may

1773: not have had a major disruptive episode (Greaves et al.\ 2004).

1774: 

1775: \section{Summary}

1776: 

1777: The planetary debris disk around $\epsilon$ Eridani

1778: was observed using {\it Spitzer}'s IRAC and MIPS cameras,

1779: {\it Spitzer}'s IRS and MIPS SED-mode spectrophotometers,

1780: and the SHARC II bolometer array on the CSO sub-millimeter telescope.

1781: 

1782: The slight 25 $\mu$m excess detected by IRAS is confirmed

1783: photometrically by the {\it Spitzer} IRS and MIPS data

1784: but not resolved in the 24 $\mu$m image.

1785: The CSO 350 $\mu$m map confirms the sub-mm ring

1786: at $r \sim$ 11--28\asec~(35--90 AU),

1787: previously observed at 450 and 850 $\mu$m,

1788: about 2$\times$ the width of our solar system's Kuiper belt.

1789: Analyses of the 70 and 160 $\mu$m images reveal emission

1790: extending from within a few arcsec of the star outward

1791: to $r \sim$ 34\asec~(110 AU),

1792: both inside and outside the sub-mm ring.

1793: The {\it Spitzer} 70 and 160 $\mu$m images do not have enhanced

1794: surface brightness at the position of the ring.

1795: The 350 $\mu$m surface brightness varies

1796: noticeably with azimuth around the ring but the

1797: locations of the brightest features do not correspond

1798: between 350 $\mu$m data sets reduced separately,

1799: or with with the brightest features in the longer-wavelength maps,

1800: calling into question the reality of the sub-mm ring azimuthal

1801: asymmetries.

1802: 

1803: The over-all system spectral energy distribution is complex,

1804: with significant excess relative to the

1805: photosphere beginning at $\lambda \sim 15~\mu$m.

1806: The excess rises rapidly

1807: to 20 $\mu$m, flattens from 20 to 30 $\mu$m, rises rapidly again

1808: from 30 $\mu$m to a peak between 70 and 100 $\mu$m, and then

1809: decreases toward longer wavelengths.

1810: 

1811: A model of the disk,

1812: constrained by all the available data,

1813: combines two populations of material in the resolved disk:

1814: (1) large (radii $ a \sim$ 100--200 $\mu$m) solid grains

1815: in the sub-mm ring, which have high emissivity at sub-mm wavelengths

1816: but low emissivity at 70 and 160 $\mu$m, consistent with H$_2$O ice

1817: composition, plus (2) smaller ($ a \sim$ 6--23 $\mu$m) solid grains in

1818: both the sub-mm ring and its exterior ''halo''

1819: (r = 25--34\asec, 90--110 AU) which have high emissivity

1820: in the far-IR and low emissivity at sub-mm wavelengths,

1821: consistent with silicate composition.

1822: The mid- and far-IR spectrophotometry and radial source profiles

1823: imply two additional narrow belts of material,

1824: one at r $\sim$ 3 AU (1\asec) with a weak

1825: 20 $\mu$m silicate emission feature,

1826: and the other at r $\sim$ 20 AU (6\asec)

1827: of undetermined, but possibly silicate, composition.

1828: 

1829: The grains in the

1830: halo may be spread there from the ring region via

1831: the effect of collisions.  Definition

1832: of the inner edge of the sub-mm ring, and confinement of the

1833: two rings of material closer to the star, probably require

1834: gravitational influences of three planets.

1835: A dynamical model of the disk indicates that the observed

1836: populations of grains in the sub-mm ring and outer halo can

1837: be supplied in equilibrium by

1838: collisions of about 11 M$_\oplus$ of km-scale parent

1839: bodies located in the sub-mm ring.

1840: 

1841: \acknowledgements

1842: 

1843: This paper is based in part on observations made

1844: with the {\it Spitzer Space Telescope} that is operated by the Jet 

1845: Propulsion Laboratory, California Institute of Technology, under

1846: NASA contract 1407, and in part on observations from the

1847: Caltech Submillimeter Observatory (CSO).

1848: Support for this work was provided by NASA through awards

1849: issued by JPL/Caltech, including subcontracts 1278243

1850: and 1255094 respectively to the SETI Institute and the

1851: University of Arizona, and also by NASA grant NNG05GI81G.

1852: Analyses in this paper made use of the SIMBAD

1853: database operated at CDS, Strasbourg France.

1854: 

1855: We are grateful to Jane Morrison and Myra Blaylock, University of Arizona,

1856: for help with MIPS image data processing, and Paul Smith,

1857: University of Arizona, for guidance in MIPS SED-mode data processing.

1858: We thank Peter Plavchan and Eric Becklin for helpful discussions.

1859: 

1860: \begin{thebibliography}

1861: 

1862: \bibitem[Altenhoff et al.(2002)]{altenhoff2002} Altenhoff, W. J.,

1863:  Bertoldi, F., Menten, K. M., Sievers, A., Thum, C., \& Kreysa, E. 2002,

1864:  \aap{}, 391, 353

1865: 

1866: \bibitem[Aumann(1985)]{aumann1985} Aumann, H. H. 1985, \pasp{}, 97,

1867:  885 

1868: 

1869: \bibitem[Backman(2004)]{backman2004} Backman, D. 2004, in ''Debris

1870:  Disks and the Formation of Planets:  A Symposium in Memory of Fred

1871:  Gillett'', eds. L. Caroff, L.J. Moon, D. Backman, \& E. Praton,

1872:  ASP Conference Series 324, 9

1873: 

1874: \bibitem[Backman et al.(1995)]{backman1995} Backman, D. E.,

1875:  Dasgupta, A., \& Stencel, R.E. 1995, \apjl{}, 450, L35

1876: 

1877: \bibitem[Backman \& Paresce(1993)]{backman1993} Backman, D., \&

1878:  Paresce, F.  1993, in Protostars and Planets III, eds.\ Levy E. H.

1879:  \& Lunine, J. I., University of Arizona Pess, Tucson, p. 1253

1880: 

1881: \bibitem[Barnes(2007)]{barnes2007} Barnes, S. A. 2007, \apj{}, 669,

1882:  1167 

1883: 

1884: \bibitem[Benedict et al.(2006)]{benedict2006} Benedict, G. F. et

1885:  al. 2006, \aj{}, 132, 2206 

1886: 

1887: \bibitem[Burns et al.(1979)]{burns1979} Burns, J. A., Lamy, P. L. \&

1888:  Soter, S. 1979, \icarus{}, 40, 1

1889: 

1890: \bibitem[Butler et al.(2006)]{butler2006} Butler, R.P. et al. 2006,

1891:  \apj{}, 646, 505 

1892: 

1893: \bibitem[Campbell et al.(1988)]{campbell1988} Campbell, B. W., Walker,

1894:  G. A. H. \& Yang, S. 1988, \apj{} 331, 902

1895: 

1896: \bibitem[Chen \& Jura(2001)]{chen2001} Chen, C. H., \& Jura, M., 2001,

1897: \apj{}, 560, L171

1898: 

1899: \bibitem[Chen et al.(2006)]{chen2006} Chen, C. H. et al. 2006, \apjs{},

1900:  166, 351

1901: 

1902: \bibitem[Chen et al.(2007)]{chen2007} Chen, C. H. et al. 2007, \apj{},

1903:  666, 466

1904: 

1905: \bibitem[Cohen(2003)]{cohen2003} Cohen, M. Megeath, S. T., Hammerseley,

1906:  P. L., Mart\'in-Luis, F. \& Stauffer, J. 2003, \aj{}, 125, 2645

1907: 

1908: \bibitem[Das \& Backman(1992)]{das1992} Das, S. R. \& Backman,

1909:  D. E. 1992, \baas{}, 24, 1228

1910: 

1911: \bibitem[Deller \& Maddison(2005)]{deller2005} Deller, A.T. \&

1912:  Maddison, S.T. 2005, \apj{}, 625, 398 

1913: 

1914: \bibitem[Dent et al.(2000)]{dent2000} Dent, W.R.F., Walker, H.J.,

1915:  Holland, W.S., Greaves, J.S. 2000, \mnras{}, 314, 702

1916: 

1917: \bibitem[Di Folco et al.(2004)]{difolco2004} Di Folco, E., Th\'evenin,

1918:  F., Kervella, P. Domiciano de Souza, A., Coud\'e du Foresto, V.,

1919:  S\'ngransan, D. \& Morel, P. 2004, \aap{}, 426, 601

1920: 

1921: \bibitem[Di Folco et al.(2007)]{difolco2007} Di Folco, E. et al. 2007,

1922:  \aap{}, 475, 234

1923: 

1924: \bibitem[Dowell et al.(2003)]{dowell2003} Dowell, C. D. et al. 2003, in

1925: ``Millimeter and Submillimeter Detectors for Astronomy'',

1926: eds. T. G. Phillips \& J. Zmuidzinas, Proc. SPIE, 4855, p. 73

1927: 

1928: \bibitem[Engelbracht et al.(2007)]{engelbracht2007} Engelbracht, C. W.

1929:  et al. 2007, \pasp{}, 119, 994

1930: 

1931: \bibitem[Fazio et al.(2004)]{fazio2004} Fazio, G. G. et al. 2004,

1932:  \apjs{}, 154, 10

1933: 

1934: \bibitem[Flower(1996)]{flower1996} Flower, P. J. 1996, \apj{}, 469,

1935:  355 

1936: 

1937: \bibitem[Gillett(1984)]{gillett1984} Gillett, F.C. 1984, invited

1938: oral presentation at Protostars and Planets II

1939: 

1940: \bibitem[Gillet(1986)]{gillet1986} Gillett, F.C. 1986, in ``Light on

1941:  dark matter''; Proceedings of the First Infra-Red Astronomical

1942:  Satellite Conference,  Reidel Publishing, Dordrecht, p. 61

1943: 

1944: \bibitem[Gordon et al.(2005)]{gordon2005} Gordon, K. D. et al. 2005,

1945:  \pasp{}, 117, 503 

1946: 

1947: \bibitem[Gordon et al.(2007)]{gordon2007} Gordon, K. D. et al. 2007,

1948:  \pasp{}, 119, 1019 

1949: 

1950: \bibitem[Greaves et al.(1998)]{greaves1998} Greaves, J. et al. 1998,

1951:  \apj{}, 506, 133 

1952: 

1953: \bibitem[Greaves et al.(2004)]{greaves2004} Greaves, J.S.,

1954:  Wyatt, M.C., Holland, W.S., \& Dent, W.R.F., 2004, M.N.R.A.S.,

1955:  351, L54

1956: 

1957: \bibitem[Greaves et al.(2006)]{greaves2005} Greaves, J. et al. 2005,

1958:  \apj{}, 619, 187 

1959: 

1960: \bibitem[Hatzes et al.(2000)]{hatzes2000} Hatzes, A. et al. 2000,

1961:  \apjl{}, 544, L145 

1962: 

1963: \bibitem[Henry et al.(1996)]{henry1996} Henry, T.J., Soderblom, D.R.,

1964:  Donahue, R.A., Baliunas, S.L. 1996, \aj{}, 111, 439 

1965: 

1966: \bibitem[Houck et al.(2004)]{houck2004} Houck, J.R. et al. 2004,

1967:  \apjs{}, 154, 18 

1968: 

1969: \bibitem[Johnson(1996)]{johnson1996} Johnson, H. L. 1986, \araa{}, 4,

1970:  193 

1971: 

1972: \bibitem[Koornneef(1983a)]{koornnees1983a} Koornneef, J. 1983a,

1973:  \aaps{}, 51, 489

1974: 

1975: \bibitem[Koornneef(1983b)]{koornneef1983b} Koornneef, J. 1983b,

1976:    \aap{}, 128, 84

1977: 

1978: \bibitem[Krist(2006)]{krist2006} Krist, J.E. 2006 Spitzer Tiny TIM User's Guide Version 2.0

1979: 

1980: \bibitem[Lagrange et al.(2000)]{lagrange2000} Lagrange, A.-M.,

1981:  Backman D. E., Artymowicz, P. 2000, in Protostars and Planets IV,

1982:  Mannings, V., Boss,  A.P., and Russell, S.S. eds, Univ. Arizona

1983:  Pess, Tucson, p. 639 

1984: 

1985: \bibitem[Levison et al.(2007)]{levison2007} Levison, H.F.,

1986:  Morbidelli, A., Gomes, R., \& Backman, D., 2007,

1987:  in Reipurth, B., Jewitt, D. and

1988:  Kiel, K., eds., Protostars and Planets V, Univ. Arizona Pess,

1989:  Tucson, p. 669

1990: 

1991: \bibitem[Li et al.(2003)]{li2003} Li, A., Lunine, J. I., and Bendo,

1992:  G. J. 2003, \apjl{}, 598, L51

1993: 

1994: \bibitem[Lisse et al.(2008)]{lisse2008} Lisse, C. M., Chen, C. H., Wyatt, M. C., and Morlok, A. 2008, \apj{}, 673, 1106

1995: 

1996: \bibitem[Liou \& Zook(1999)]{liou1999} Liou, J.-C. and Zook,

1997:  H. A. 1999, \aj{}, 118, 580 

1998: 

1999: \bibitem[Luu \& Jewitt(2002)]{luu2002}  Luu, J.X., \& Jewitt, D.C.,

2000:  2002, A.R.A.A., 40, 63

2001: 

2002: \bibitem[Marengo et al.(2006)]{marengo2006} Marengo, M., Megeath,

2003:  S.T., Fazio, G.G.,  Stapelfeldt, K.R., Werner, M.W. \& Backman,

2004:  D.E. 2006, \apj{}, 647, 1437 

2005: 

2006: \bibitem[Mendoza et al.(1978)]{mendoza1978} Mendoza, E. E., Gomez,

2007:  V. T. \& Gonzalez, S. 1978, \aj{}, 83, 606

2008: 

2009: \bibitem[Meyer et al.(2007)]{meyer2007} Meyer, M.R., Backman, D.E.,

2010:  Weinberger, A.J.,  Wyatt, M.C. 2007, in Reipurth, B., Jewitt, D. and

2011:  Kiel, K., eds., Protostars and Planets V, Univ. Arizona Pess,

2012:  Tucson, p. 573

2013: 

2014: \bibitem[Moran et al.(2004)]{moran2004} Moran, S. M., Kuchner,

2015:  M. J. \& Holman, M. J. 2004, \apj{}, 612, 1163

2016: 

2017: \bibitem[Moro-Martin et al.(2005)]{moro-martin2005} Moro-Martin, A.,

2018: Wolf, S., \& Malhotra, R., 2005, \apj{}, 621, 1079

2019: \bibitem[Ozernoy et al.(2000)]{ozernoy2000} Ozernoy, L. M., Gorkavyi,

2020:  N. N., Mather, J. C., Taidakova, T. A. 2000, \apj{}, 537, 147 

2021: 

2022: \bibitem[Plavchan et al.(2005)]{plavchan2005} Plavchan, P., Jura,

2023:  M. \& Lipscy, S. J. 2005, \apj{}, 631, 1161

2024: 

2025: \bibitem[Poulton et al.(2006)]{poulton2006} Poulton, C. J., Greaves,

2026:  J. S. \& Cameron, A. C. 2006, \mnras{}, 372, 53

2027: 

2028: \bibitem[Proffitt et al.(2004)]{proffitt2004} Proffitt, C.R. et

2029:  al. 2004, \apj{}, 612, 481 

2030: 

2031: \bibitem[Quillen \& Thorndike(2002]{quillen2002} Quillen, A.C.,

2032:  Thorndike, S. 2002, \apjl{}, 578, L149 

2033: 

2034: \bibitem[Riecke et al.(2004)]{rieke2004} Rieke, G. et al. 2004,

2035:  \apjs{}, 154, 25 

2036: 

2037: \bibitem[Schuster et al.(2006)]{schuster2006} Schuster, M., Marengo,

2038:  M. \& Patten B. 2006, SPIE Meeting, Orlando, FL. \#6720-65

2039: 

2040: \bibitem[Sch\"utz et al.(2004)]{schutz2004} Sch\"utz, O., Nielbock,

2041:  M., Wolf, S., Henning, Th. \& Els, S. 2004, \aap{}, 414, L9

2042: 

2043: \bibitem[Sheret et al.(2004)]{sheret2004} Sheret, I., Dent, W.R.F. \& 

2044:  Wyatt, M.C. 2004, \mnras{}, 348, 1282

2045: 

2046: \bibitem[Stern \& Colwell(1997)]{stern1997} Stern, S. A. \& Colwell,

2047:  J. E. 1997, \apj{}, 490, 879

2048: 

2049: \bibitem[Stevens(2005)]{stevens2005} Stevens, I.R. 2005, \mnras{},

2050:  356, 1053 

2051: 

2052: \bibitem[Song et al.(2000)]{song2000} Song, I., Caillault,

2053:  J.-P. Barrado y Navascu\'es, D., Stauffer, J.R. \& Randich, S. 2000,

2054:  \apjl{}, 533, L41 

2055: 

2056: \bibitem[Stansberry et al.(2007)]{stansberry2007} Stansberry, J. A. et

2057:  al. 2007, \pasp{}, 119, 1038

2058: 

2059: \bibitem[Stapelfeldt et al.(2004)]{stapelfeldt2004} Stapelfeldt,

2060:  K.R. et al. 2004, \apjs{}, 154, 458 

2061: 

2062: \bibitem[Su et al.(2008)]{su2008} Su, K. Y., Rieke, G. H.,

2063:  Stapelfeldt, K. R., Smith, P. S., Bryden, G., Chen, C. H. \&

2064:  Trilling, D. E. 2008, in press, arXiv:0804.2924

2065: 

2066: \bibitem[Su et al.(2005)]{su2005} Su, K.Y. et al. 2005, \apj{}, 628,

2067:  487 

2068: 

2069: \bibitem[Su et al.(2006)]{su2006} Su, K. Y. et al. 2006, \apj{}, 653,

2070:  675 

2071: 

2072: \bibitem[Trilling et al.(2008)]{trilling2008} Trilling, D. E. et

2073:  al. 2008, \apj{}, 674, 1086 

2074: 

2075: \bibitem[Walker \& Heinrichsen(2000)]{walker2000} Walker, H.J. \&

2076:  Heinrichsen, I. 2000, \icarus{}, 143, 147 

2077: 

2078: \bibitem[Wolf \& Hillenbrandt(2005)]{wolf2005} Wolf, S. \&

2079:  Hillenbrand, L. A. 2005, Comp.\ Phy.\ Comm., 171, 208

2080: 

2081: \bibitem[Wood et al.(2005)]{wood2005} Wood, B. E., M\"uller, H.-R.,

2082: Zank, G. P., Linsky, J. L., and Redfield, S. 2005, \apjl{}, 628, L143

2083: 

2084: \bibitem[Wyatt(2005)]{wyatt2005} Wyatt, M. C. 2005, \aap{}, 433, 1007

2085: 

2086: \bibitem[Wyatt(2006)]{wyatt2006} Wyatt, M. C. 2006, \apj{}, 639, 1153

2087: 

2088: \end{thebibliography}

2089: 

2090: %%

2091: %% Tables

2092: %%

2093: %

2094: \clearpage 

2095: 

2096: % Table 1

2097: \begin{deluxetable}{rrrrl}

2098: \tablecaption{$\epsilon$~Eri Broadband Photometry}

2099: \tablehead{

2100: \colhead{$\lambda$ [$\mu$m]} &

2101: \colhead{Total $F_\nu$ [Jy]} &

2102: \colhead{Star $F_\nu$ [Jy]} &

2103: \colhead{Excess $F_\nu$ [Jy]} &

2104: \colhead{Notes}}

2105: \startdata

2106:    3.550 & 64.320 $\pm$ 1.140 & 64.320 &   ---  & IRAC\\

2107:    4.493 & 39.800 $\pm$ 0.700 & 39.800 &   ---  & IRAC\\

2108:    5.731 & 26.110 $\pm$ 0.470 & 26.110 &   ---  & IRAC\\

2109:    7.872 & 14.910 $\pm$ 0.260 & 14.910 &   ---  & IRAC\\

2110:   12.000 &  6.690 $\pm$ 0.540 &  6.570 &  0.130 & Color-corrected IRAS\\

2111:   23.680 &  2.042 $\pm$ 0.040 &  1.726 &  0.315 & MIPS\\

2112:   25.000 &  1.900 $\pm$ 0.280 &  1.550 &  0.350 & Color-corrected IRAS\\

2113:   60.000 &  1.750 $\pm$ 0.250 &  0.270 &  1.480 & Color-corrected IRAS\\

2114:   71.420 &  1.688 $\pm$ 0.207 &  0.189 &  1.499 & MIPS\\

2115:  100.000 &  1.990 $\pm$ 0.300 &  0.100 &  1.890 & Color-corrected IRAS\\

2116:  156.000 &  0.970 $\pm$ 0.186 &  0.039 &  0.931 & MIPS\\

2117:  200.000 &  1.880 $\pm$ 0.280 &  0.020 &  1.860 & ISO$^1$\\

2118:  350.000 &  0.366 $\pm$ 0.050 &  0.008 &  0.358 & CSO / SHARC II\\

2119:  450.000 &  0.250 $\pm$ 0.020 &  0.005 &  0.245 & JCMT / SCUBA$^2$\\

2120:  850.000 &  0.037 $\pm$ 0.003 &  0.001 &  0.036 & JCMT / SCUBA$^2$\\

2121: \enddata

2122: \tablerefs{(1) Walker \& Heinrichsen (2000); (2) Greaves et al. (2005)}

2123: \end{deluxetable}

2124: 

2125: \clearpage

2126: 

2127: % Table 2

2128: \begin{deluxetable}{lcccccc}

2129: \tablecaption{Model Components}

2130: \tablehead{

2131: \colhead{component} &

2132: \colhead{$r$ [AU]} &

2133: \colhead{$M_{\rm T}$ [$M_\oplus$]} &

2134: \colhead{$\alpha$} &

2135: \colhead{$a$ [$\mu$m]} &

2136: \colhead{$x$} &

2137: \colhead{$f$}}

2138: \startdata

2139: W1 & 3 & $1.8 \times 10^{-7}$ & -- & 3.0 & -- & $3.3 \times 10^{-5}$ \\

2140: W2 & 20 & $2.0 \times 10^{-5}$ & -- & 8.0 & -- & $3.4 \times 10^{-5}$ \\

2141: RS & 35--90 & $2.0 \times 10^{-4}$ & +0.01 & 6.0--23  & -3.5 & $3.0 \times 10^{-5}$ \\

2142: RL & 35--90 & $4.2 \times 10^{-3}$ & +1.05 & 100--200  & -3.5 & $4.4 \times 10^{-6}$ \\

2143: HS & 90--110 & $2.5 \times 10^{-4}$ & +0.15 & 15-23 & -3.5 & $4.8 \times 10^{-6}$ \\

2144: \enddata

2145: \tablecomments{Columns:  (1) model component: 

2146: W1 = warm belt 1, W2 = warm belt 2, 

2147: RS = sub-mm ring, small grains; RL = sub-mm ring, large grains;

2148: HS = halo, small grains;  (2) location; (3) total mass;

2149: (4) mass surface density exponent, assumed to be zero for the W1 and

2150: W2 components, fitted to data for the other components;  (5) grain radius;

2151: (6) assumed grain size distribution exponent;

2152: (7) fractional luminosity, $L_d/L_*$}

2153: \end{deluxetable}

2154: 

2155: %%

2156: %% Figures and Captions.

2157: %%

2158: 

2159: %

2160: % 6-PANEL MIPS IMAGES

2161: \clearpage

2162: 

2163: %\markright{Figure 1}

2164: \onecolumn

2165: \begin{figure}[t]

2166: \begin{center}

2167: \includegraphics[width=0.85\textwidth]{f1.eps}

2168: \end{center}

2169: \figcaption{Images of $\epsilon$ Eri in the three {\it Spitzer} MIPS channels.

2170: N is up and E to the left in all the panels; 10\asec~$=$ 32 AU at

2171: $\epsilon$ Eri's distance.

2172: A cross marks the stellar position at the epoch of each observation,

2173: taking into account $\epsilon$ Eri's proper motion of $\sim$ 1\asec~per year.

2174: Dark is positive.  Circles indicate FWHM beam sizes.

2175: {\bf The left column} shows 24 $\mu$m images 

2176: in log stretch.  These have been reconstructed from multiple dithered 

2177: exposures and are oversampled by a factor of four.  Panel $(a)$ is the 

2178: direct image, dominated by the stellar PSF.  Panel $(d)$ shows this image 

2179: after subtracting a PSF reference star scaled to the photospheric

2180: flux density.  The remaining excess (18\% of the total 24 $\mu$m

2181: emission) appears largely as an unresolved point source.

2182: {\bf The middle column} shows 70 $\mu$m images in linear stretch.

2183: Panel $(b)$

2184: is the default pixel-scale image used for photometric measurements,

2185: and panel $(e)$ shows the fine pixel-scale image.

2186: {\bf The right column} shows 160 $\mu$m images in linear stretch.

2187: Panel $(c)$

2188: is the direct image, and panel $(f)$

2189: shows this image after subtraction of the

2190: ghost image arising from the spectral leak.

2191: }

2192: \label{mips-data}

2193: \end{figure}

2194: %

2195: 

2196: %

2197: % 70 AND 160 MICRON DATA MINUS AZIMUTHALLLY SMOOTHED

2198: \clearpage

2199: 

2200: %%\markright{Figure 2}

2201: \onecolumn

2202: \begin{figure}[t]

2203: \begin{center}

2204: \includegraphics[width=0.45\textwidth]{f2.eps}

2205: \end{center}

2206: \figcaption{

2207: (a) difference between the {\it Spitzer}

2208: MIPS fine-scale 70 $\mu$m image and

2209: its azimuthal average.

2210: (b) the same for the MIPS 160~$\mu$m image.

2211: The vertical bars are due to

2212: incomplete correction of column artifacts.

2213: The remaining residuals are consistent with noise

2214: from the subtraction, indicating absence of surface brightness

2215: features in the sub-mm ring within the sensitivity limits.

2216: }

2217: \label{div_bumps_70f}

2218: \end{figure}

2219: %

2220: 

2221: %

2222: % MIPS SED-MODE FLUX AND FWHM

2223: \clearpage

2224: 

2225: %\markright{Figure 3}

2226: \begin{figure}[t]

2227: \begin{center}

2228: \includegraphics[width=0.65\textwidth]{f3a.eps}

2229: \includegraphics[width=0.65\textwidth]{f3b.eps}

2230: \end{center}

2231: \figcaption{

2232: MIPS SED-mode measurements of $\epsilon$ Eri.

2233: (a) (solid line) FWHM of the

2234: $\epsilon$ Eri disk emission versus wavelength, extracted by Gaussian

2235: fitting along the slit at PA 85\deg, versus (dashed line) the same for

2236: Canopus, an unresolved

2237: calibration star with no IR excess.  $\epsilon$ Eri is

2238: clearly resolved across the wavelength range of the MIPS SED

2239: measurements.

2240: (b) Star+disk spectrum windowed by the 20\asec-wide slit,

2241: without aperture correction.

2242: The dotted lines indicated the range of photometric

2243: uncertainty based on uncertainty in Gaussian fits

2244: across the slit.  Subsequent model-dependent flux normalization

2245: of these data is discussed in Section 4.4.2.

2246: }

2247: \label{mips-sed}

2248: \end{figure}

2249: %

2250: 

2251: %

2252: % SHARCII CSO 350 MICRON IMAGE

2253: \clearpage

2254: 

2255: %\markright{Figure 4}

2256: \onecolumn

2257: \begin{figure}[t]

2258: \begin{center}

2259: \includegraphics[width=0.95\textwidth,angle=-90]{f4.eps}

2260: \end{center}

2261: \figcaption{

2262: Caltech Submillimeter Observatory (CSO) SHARC II 350 $\mu$m map.

2263: Dark is positive.

2264: The circle's diameter indicates 12.7\asec~resolution to which the

2265: map was smoothed, equivalent

2266: to 41 AU at the distance of $\epsilon$ Eri.

2267: A scaled PSF representing the relatively small 350 $\mu$m

2268: brightness of the central star has been subtracted.

2269: }

2270: \label{350um-CSO}

2271: \end{figure}

2272: %

2273: 

2274: %

2275: % MERGED BROADBAND SED (EXCLUDING MIPS SED)

2276: \clearpage

2277: 

2278: %\markright{Figure 5}

2279: \onecolumn

2280: \begin{figure}[t]

2281: \begin{center}

2282: \includegraphics[width=0.65\textwidth,angle=-90]{f5.eps}

2283: \end{center}

2284: \figcaption{

2285: Merged broadband SED of the $\epsilon$ Eri system.  Numerical values are

2286: presented in Table 1.  The thin solid curve is a Kurucz photosphere for

2287: T = 5000 K, log g = 4.5, fit to the four {\it Spitzer} IRAC photometric

2288: points at 3.6 to 7.9 $\mu$m.

2289: The thick curve is the combined {\it Spitzer} IRS SL + SH + LH

2290: spectrum rescaled to have zero average excess at 5--12~$\mu$m. The

2291: IRAS 100~$\mu$m point and the ISO 200~$\mu$m point are

2292: considered upper limits due to background contamination. 

2293: The inset panel shows the mid-infrared segment of the SED from

2294: {\it Spitzer} IRS (thin solid line), after subtraction of the

2295: photosphere model displayed in the main figure. The thick line

2296: is the same spectrum smoothed with a

2297: 111-point ($\sim 3.3$ $\mu$m) boxcar.

2298: }

2299: \label{merged-SED}

2300: \end{figure}

2301: %

2302: 

2303: %

2304: % RADIAL PROFILES

2305: \clearpage

2306: 

2307: %\markright{Figure 6}

2308: \onecolumn

2309: \begin{figure}[t]

2310: \begin{center}

2311: \includegraphics[width=0.65\textwidth,angle=-90]{f6.eps}

2312: \end{center}

2313: \figcaption{

2314: Radial profiles of images at 24, 70, 160 $\mu$m

2315: ({\it Spitzer} MIPS) and the 350 $\mu$m (CSO) map.

2316: In each wavelength panel the dashed line with error bars represents

2317: the azimuthal average of the image data after subtraction

2318: of the stellar photosphere PSF. Observational error bars are

2319: estimated as the rms of the azimuthal average.

2320: The thick solid line is the disk model after convolution with

2321: the relevant instrumental PSF, fitted to the observations.

2322: The discontinuous thin solid line is the model

2323: intrinsic surface brightness before PSF convolution (Section 4.2).

2324: }

2325: \label{1d_profiles}

2326: \end{figure}

2327: %

2328: 

2329: % 

2330: % OVER-ALL SYSTEM SED AND MODEL

2331: \clearpage

2332: 

2333: %\markright{Figure 7}

2334: \onecolumn

2335: \begin{figure}[t]

2336: \begin{center}

2337: \includegraphics[width=0.65\textwidth,angle=-90]{f7.eps}

2338: \end{center}

2339: \figcaption{

2340: Observed SED of the $\epsilon$ Eri disk, after subtraction of

2341: the stellar photosphere, compared with the model SED.

2342: Individual photometric points include MIPS SED

2343: flux densities at 55 and 90~$\mu$m, iteratively

2344: aperture-corrected using model images, and

2345: rescaled according to the measured MIPS 70 $\mu$m image total

2346: flux. The thick solid line is the IRS SL + SH + LH combined spectrum,

2347: rescaled to have zero average excess for $\lambda = 5-12 \mu$m.

2348: The thin solid line is the total model flux (sum of all four dust

2349: components). The two dashed lines are the contributions of the two

2350: unresolved inner belts. The

2351: dot-dashed line is the contribution of the

2352: small ($a \sim 15~\mu$m) silicate grains in the sub-mm ring and halo.

2353: The dotted line is the contribution of the

2354: large ($a \sim 135~\mu$m) ice grains in the sub-mm ring.

2355: }

2356: \label{irs+phot}

2357: \end{figure}

2358: %

2359: 

2360: %

2361: % NORMAL OPTICAL DEPTH & MASS SURFACE DENSITY

2362: \clearpage

2363: 

2364: %\markright{Figure 8}

2365: \onecolumn

2366: \begin{figure}[t]

2367: \begin{center}

2368: \includegraphics[width=0.65\textwidth]{f8.eps}

2369: \end{center}

2370: \figcaption{

2371: (a) Perpendicular optical depth $\tau_\perp$ versus

2372: position for the five model disk components.

2373: The two unresolved inner belts (dashed lines),

2374: outer ''icy'' ring (dotted line), and

2375: ring+halo ''silicate'' dust populations

2376: (dot-dashed line) are shown.

2377: (b) Mass surface density versus radius corresponding

2378: to the optical depth profiles in panel (a), assuming

2379: grain compositions

2380: and material densities described in the text.

2381: The difference in slopes between the optical depth

2382: and mass surface density profiles is due to

2383: radial variation in model particle properties.}

2384: \label{tauprofile}

2385: \end{figure}

2386: %

2387: 

2388: %

2389: % SYSTEM CARTOON

2390: \clearpage

2391: 

2392: %\markright{Figure 9}

2393: \onecolumn

2394: \begin{figure}[t]

2395: \begin{center}

2396: \includegraphics[width=0.85\textwidth,angle=-90]{f9.eps}

2397: \end{center}

2398: \figcaption{

2399: Representation of the $\epsilon$ Eri debris disk model components.

2400: The small-scale dotted ellipse is one solution for the orbit of the

2401: suggested radial velocity planet that appears to be inconsistent with

2402: the innermost warm debris belt's position.

2403: }

2404: \label{cartoon}

2405: \end{figure}

2406: %

2407: 

2408: \end{document}

2409: