1: \documentclass{emulateapj}

2: %\documentclass[12pt, preprint]{aastex}

3: 

4: %\usepackage{epsf}

5: %\usepackage{psfig}

6: % \epsfig{usepackage}

7: %\documentclass{aastex}

8: %\documentstyle[emulateapj,epsf]{article}

9: 

10: \submitted{Accepted for publication in ApJ}

11: 

12: \def\deg{\ifmmode^\circ\else$^\circ$\fi}

13: \def\solar{\ifmmode_{\mathord\odot}\else$_{\mathord\odot}$\fi}

14: \def\er{\relax} \def\sr{\relax}

15: \def\ls{{_<\atop^{\sim}}}

16: \def\gs{{_>\atop^{\sim}}}

17: \def\mic{~$\mu$m}

18: \def\etc{{\it etc.}}

19: \def\cf{{\it cf.}}

20: \def\ie{{\it i.e.}}

21: \def\eg{{\it e.g.}}

22: \def\iras{{\it IRAS}}

23: \def\iue{{\it IUE}}

24: \def\rosat{{\it Rosat}}

25: \def\ginga{{\it Ginga}}

26: \def\et{{et al.~}}

27: 

28: \def\lsun{L$_{\odot}$}

29: \def\msun{m$_{\odot}$}

30: 

31: \def\mic{$\mu${\rm m}}

32: \def\lir{{\rm L}$_{IR}$}

33: 

34: \def\hf{\mbox{$H_{1.6}$}}

35: \def\arcs{\ifmmode {''}\else $''$\fi}

36: \def\arcm{\ifmmode {'}\else $'$\fi}

37: \def\parcs{\sa=.07em \sb=.03em

38:      \ifmmode $\rlap{.}$^{\scriptscriptstyle\prime\kern -\sb\prime}$\kern -\sa$

39:      \else \rlap{.}$^{\scriptscriptstyle\prime\kern -\sb\prime}$\kern -\sa\fi}

40: \def\parcm{\sa=.08em \sb=.03em

41:      \ifmmode $\rlap{.}\kern\sa$^{\scriptscriptstyle\prime}$\kern-\sb$

42:      \else \rlap{.}\kern\sa$^{\scriptscriptstyle\prime}$\kern-\sb\fi}

43: 

44: \def\Msun{M$_{\odot}$}

45: \def\Myr{\Msun/yr}

46: \def\kp{{\rm K}$^{\prime}$}

47: \def\lya{{\rm Ly}$\alpha$}

48: \def\Ls{{ L$^{*}$}}

49: \def\han {\mbox{{\rm H}$\alpha$}}

50: \def\ha{\han~}

51: \def\xray{\hbox{X-ray}}

52: \def\brg {{\rm Br}$\gamma$}

53: 

54: \def\spose#1{\hbox to 0pt{#1\hss}}

55: \def\simlt{\mathrel{\spose{\lower 3pt\hbox{$\mathchar"218$}}

56:      \raise 2.0pt\hbox{$\mathchar"13C$}}}

57: \def\simgt{\mathrel{\spose{\lower 3pt\hbox{$\mathchar"218$}}

58:      \raise 2.0pt\hbox{$\mathchar"13E$}}}

59: \def\lsim{\rlap{$<$}{\lower 1.0ex\hbox{$\sim$}}}

60: \def\gsim{\rlap{$>$}{\lower 1.0ex\hbox{$\sim$}}}

61: 

62: \usepackage[usenames]{color}

63: 

64: \newcommand{\tmp}[1]{{\color{Bittersweet}\ignorespaces{\bf [}{\sf #1}\unskip{\bf ]}~}}

65: 

66: \begin{document}

67: 

68: \title{PAH Emission from Ultraluminous Infrared Galaxies\altaffilmark{1}}

69: 

70: \author{V. Desai\altaffilmark{2},

71: L. Armus\altaffilmark{3},

72: H.W.W. Spoon\altaffilmark{4},

73: V. Charmandaris\altaffilmark{5,6},

74: J. Bernard-Salas\altaffilmark{4},

75: B.R. Brandl\altaffilmark{7},

76: D. Farrah\altaffilmark{4},

77: B.T. Soifer\altaffilmark{2,3},

78: H.I. Teplitz\altaffilmark{3},

79: P.M. Ogle\altaffilmark{3},

80: D. Devost\altaffilmark{4},

81: S.J.U. Higdon\altaffilmark{8},

82: J.A. Marshall\altaffilmark{4},

83: J.R. Houck\altaffilmark{4}

84: }

85: 

86: \altaffiltext{1}{Based on observations obtained with the \textit{Spitzer Space

87: Telescope}, which is operated by the Jet Propulsion Laboratory,

88: California Institute of Technology, under NASA contract 1407}

89: 

90: \altaffiltext{2}{Division of Physics, Math \& Astronomy, California

91: Institute of Technology, Pasadena, CA 91125}

92: 

93: \altaffiltext{3}{Spitzer Science Center, MS 220-6, Caltech, Pasadena, CA 91125}

94: 

95: \altaffiltext{4}{Cornell University, Ithaca, NY 14853}

96: 

97: \altaffiltext{5}{University of Crete, Department of Physics, P.O. Box

98: 2208 GR-71003, Heraklion, Greece}

99: 

100: \altaffiltext{6}{IESL/Foundation for Research and Technology - Hellas,

101: GR-71110, Heraklion, Greece and Chercheur Associ\'e, Observatoire de

102: Paris, F-75014, Paris, France}

103: 

104: \altaffiltext{7}{Leiden University, P.O. Box 9513, 2300 RA Leiden, The

105: Netherlands}

106: 

107: \altaffiltext{8}{Georgia Southern University, Statesboro, GA 30460}

108: 

109: \altaffiltext{9}{The IRS was a collaborative venture between Cornell

110: University and Ball Aerospace Corporation funded by NASA through the

111: Jet Propulsion Laboratory and the Ames Research Center}

112: 

113: \begin{abstract}

114: 

115: We explore the relationships between the Polycyclic Aromatic

116: Hydrocarbon (PAH) feature strengths, mid-infrared continuum

117: luminosities, far-infrared spectral slopes, optical spectroscopic

118: classifications, and silicate optical depths within a sample of 107

119: ULIRGs observed with the Infrared Spectrograph on the {\it Spitzer

120: Space Telescope}.  The detected 6.2 $\micron$ PAH equivalent widths

121: (EQWs) in the sample span more than two orders of magnitude

122: ($\sim$0.006--0.8 $\micron$), and ULIRGs with HII-like optical spectra

123: or steep far-infrared spectral slopes ($S_{25} / S_{60} < 0.2$)

124: typically have 6.2 $\micron$ PAH EQWs that are half that of

125: lower-luminosity starbursts.  A significant fraction ($\sim$40--60\%)

126: of HII-like, LINER-like, and cold ULIRGs have very weak PAH EQWs.

127: Many of these ULIRGs also have large ($\tau_{9.7} > 2.3$) silicate

128: optical depths.  The far-infrared spectral slope is strongly

129: correlated with PAH EQW, but not with silicate optical depth.  In

130: addition, the PAH EQW decreases with increasing rest-frame 24

131: $\micron$ luminosity.  We argue that this trend results primarily from

132: dilution of the PAH EQW by continuum emission from dust heated by a

133: compact central source, probably an AGN.  High luminosity,

134: high-redshift sources studied with {\it Spitzer} appear to have a much

135: larger range in PAH EQW than seen in local ULIRGs, which is consistent

136: with extremely luminous starburst systems being absent at low

137: redshift, but present at early epochs.

138: 

139: \end{abstract}

140: 

141: \keywords{infrared:galaxies -- galaxies:active -- galaxies:starburst}

142: 

143: \section{Introduction}

144: \label{sec:Introduction}

145: 

146: Ultraluminous Infrared Galaxies (ULIRGs) have bolometric luminosities

147: comparable to quasars (${\rm L}_{\rm bol} \gs 10^{12}$ \lsun), but

148: emit nearly all of this energy at mid- and far-infrared wavelengths.

149: Although ULIRGs are rare in the local Universe and comprise only a few

150: percent of infrared-bright galaxies \citep{Soifer87,Sanders03}, they

151: account for a rapidly increasing fraction of all star-formation

152: activity at high-redshift, and they may dominate the far-infrared

153: background at $z > 2$

154: \citep{Franceschini01,LeFloch05,PerezGonzalez05}.  It has been

155: suggested that they play a role in the formation of both quasars and

156: elliptical galaxies \citep{Kormendy92,SandersMirabel96,Scott02}.

157: 

158: To understand the roles that ULIRGs play in the formation of massive

159: galaxies, the global star formation history of the universe, and the

160: creation of the cosmic infrared background, it is essential to

161: determine the extent to which Active Galactic Nuclei (AGN) and star

162: formation contribute to their enormous luminosities.  Several studies

163: have used optical

164: \citep{deGrijp85,Osterbrock85,Armus87,Armus89,Veilleux95,Kim98,Veilleux99}

165: and near-infrared

166: \citep{Goldader95,Veilleux97,Veilleux99b,Murphy99,Murphy01,Burston01,Davies03,Dannerbauer05}

167: spectroscopy to determine the main source of ionization of the

168: line-emitting gas in local ULIRGs.  These studies show that while the

169: majority of ULIRGs have optical and near-infrared spectra similar to

170: those of starburst galaxies, the fraction of ULIRGs displaying the

171: spectroscopic signatures of an AGN, either broad permitted lines or

172: high-ionization narrow-lines, increases with infrared luminosity.  In

173: addition, evidence for AGN activity is more commonly found among

174: ULIRGs with flatter far-infrared spectral slopes (those classified as

175: ``warm" with $ S_{25} / S_{60} \ge 0.2$ as measured with IRAS), than

176: among ULIRGs with steeper far-infrared spectral slopes (those

177: classified as ``cold" with $S_{25} / S_{60} < 0.2$).

178: 

179: A principle uncertainty of these optical and near-infrared studies is

180: the extent to which the physical conditions of the visible gas are

181: linked to those in the deeply embedded nucleus, where the bulk of the

182: energy is generated \citep{Soifer00}.  Emission at longer wavelengths

183: can penetrate more heavily obscured regions close to the nuclear power

184: source.  Therefore, diagnostics derived from mid-infrared spectroscopy

185: are potentially a better probe of the dominant energy source in the

186: nuclear regions. The mid-infrared spectra of ULIRGs are comprised of

187: continuum emission from heated dust grains; broad emission features

188: associated with Polycyclic Aromatic Hydrocarbons (PAHs), the most

189: prominent of which appear at 6.2, 7.7, 8.6, 11.3, and 12.7 $\micron$;

190: absorption by amorphous silicates centered at 9.7 and 18 $\micron$;

191: and atomic fine-structure lines of Ne, O, Si, and S covering a large

192: range in ionization potential.  Diagnostic diagrams using combinations

193: of fine-structure line ratios, mid-infrared spectral slope, and PAH

194: feature strengths have been used to classify bright ULIRGs, based on

195: the expectations that ULIRGs with central AGN produce high ionization

196: lines, have flatter spectral slopes, and display lower PAH equivalent

197: widths than those without

198: \citep[e.g.][]{Genzel98,Lutz98,Rigopoulou99,Laurent00,Tran01,Sturm02,Armus04,Armus06,Armus07,Farrah07}.

199: For sources at high-redshift ($z \ge 1$), however, it is generally not

200: possible to place useful limits on the high-ionization fine-structure

201: lines.  The PAH emission, together with the infrared spectral slope,

202: are often the only mid-infrared diagnostics available for studying

203: high-redshift ULIRGs.

204: 

205: The first attempts to classify the energy generation mechanism in

206: ULIRGs using mid-infrared PAH features made use of data from the

207: \textit{Infrared Space Observatory}

208: \citep{Lutz98,Lutz99,Rigopoulou99,Tran01}.  For these studies, ULIRGs

209: were classified based on the strengths of their 7.7 $\micron$ PAH

210: line-to-continuum ratios (L/C, defined as the ratio of the peak height

211: of the 7.7 $\micron$ PAH feature to the level of the underlying 7.7

212: $\micron$ continuum).  The PAH-based classifications showed that most

213: ($\sim$80\%) ULIRGs have strong PAH emission indicative of star

214: formation as the dominant power source.  However, the highest

215: luminosity sources tend to have weak (or undetected) PAH emission,

216: interpreted as a sign of increased AGN activity.  Recently,

217: \citet{Imanishi07} analyzed the mid-infrared spectra of 48 nearby

218: ULIRGs observed with the Infrared Spectrograph \citep[IRS;][]{Houck04}

219: on board the \textit{Spitzer Space Telescope}.  These authors suggest

220: that 30-50\% of these ULIRGs, all of which have HII-like or LINER-like

221: optical classifications, harbor buried AGN.

222: 

223: In this paper we present the mid-infrared spectroscopic properties of

224: a large sample of 107 ULIRGs that comprise the IRS GTO team ULIRG

225: survey.  The sample includes a large fraction ($\sim$40\%) of warm

226: sources, which tend to have higher infrared luminosities than cold

227: ULIRGs.  The warm ULIRGs are a critical population to study, since it

228: has been suggested that they mark the transition between

229: staburst-dominated ULIRGs and QSOs \cite[e.g.~][]{Sanders88b}.  Because

230: of their high luminosities, they provide a much-needed baseline over

231: which to extrapolate to the extremely high luminosities (L$_{\rm IR} >

232: 10^{13}$ L$_{\odot}$) seen at higher redshift.

233: 

234: The sensitivity of the IRS offers two distinct advantages over

235: previous studies with ISO.  First, we can measure the PAH strengths

236: with respect to the dust continuum (the equivalent width, or hereafter

237: EQW) over nearly two orders of magnitude, allowing us to quantify the

238: PAH emission in many ULIRGs that previously only had upper

239: limits. This is critical for fitting trends with luminosity and

240: spectral slope, as many ULIRGs have much weaker PAH emission features

241: than are typically found in low-luminosity starburst galaxies (see

242: \S{\ref{sec:medianspectra}}).  Second, the Short-Low and Long-Low IRS

243: modules (5--38 $\micron$) offer greatly increased wavelength coverage

244: compared to ISOPHOT-S, allowing us to accurately fit the silicate

245: absorption (at 9.7 and 18 $\micron$) and multiple PAH features in all

246: the ULIRGs.  Here, we have chosen to fit the 9.7 $\micron$ silicate

247: absorption feature since it is the strongest and the 6.2 and 11.3

248: $\micron$ PAH features because they are both well-isolated and easily

249: measured.  Previously the 7.7 $\micron$ feature was often used, which

250: is not ideal because absorption at both shorter and longer wavelengths

251: can mimic a peak at 7.7 $\micron$ \citep{Spoon02} and its integrated

252: flux can be difficult to measure due to blending with the adjacent 8.6

253: $\micron$ feature.  By correlating the PAH emission with the silicate

254: absorption, it is possible to gain insight into the distribution of

255: the heating source(s) and the grains.  In the following analysis, we

256: use ${\rm H_0}=70$ km s$^{-1}$ Mpc$^{-1}$, $\Omega_{\rm m}=0.3$, and

257: $\Lambda=0.70$.

258: 

259: \section{Observations \& Data Reduction}

260: \label{sec:Observations}

261:  

262: The 107 ULIRGs presented in this paper were chosen primarily from the

263: IRAS 1-Jy survey \citep{Kim98}, the IRAS 2-Jy survey

264: \citep{Strauss92}, and the FIRST/IRAS radio-far-infrared sample of

265: \citet{Stanford00}.  The ULIRGs span a redshift range of $0.018 < z <

266: 0.93$, have IRAS 60 $\micron$ flux densities between $0.14 < {\rm

267: S}_{60} / {\rm Jy} < 103$, and have integrated 40--500 $\micron$

268: luminosities in the range $11.7 < \log({\rm L}_{\rm FIR} / {\rm

269: L}_{\odot}) < 13.13$.  The sample consists of 41 warm ULIRGs and 66

270: cold ULIRGs, based upon their rest-frame far-infrared flux densities

271: at 25 and 60 $\micron$.  Many of our highest luminosity sources were

272: taken from the \citet{Stanford00} FIRST/IRAS survey.  While

273: radio-bright AGN may preferentially populate the highest luminosities

274: of the Stanford et al.~survey, the selected ULIRGs have radio to

275: far-infrared flux density ratios consistent with infrared-selected

276: starburst galaxies of lower luminosity.  The sample is specifically

277: designed to include a larger fraction of warm and infrared luminous

278: sources than would be found in a pure flux- or volume-limited sample,

279: for the express purpose of allowing a careful study of the

280: mid-infrared spectral properties as functions of infrared color and

281: luminosity.

282: 

283: All ULIRGs were observed in Staring Mode with both sub-slits (orders)

284: of each of the Short-Low (SL) and Long-Low (LL) modules of the IRS.

285: Each target was acquired by performing a high accuracy IRS peak-up on

286: the target itself, or by peaking up on a nearby 2MASS star and

287: offsetting to the target.  Each galaxy was observed at two nod

288: positions within each of the IRS sub-slits (SL1, SL2, LL1, LL2). The

289: resulting spectra have a spectral resolution of R $\sim 80$ over the

290: 5--38 $\micron$ wavelength range.

291: 

292: All spectra were reduced using the S14 IRS pipeline at the Spitzer

293: Science Center.  This reduction includes ramp fitting, dark sky

294: subtraction, droop correction, linearity correction, and wavelength

295: and flux calibration.  The flux calibration sources were HD173511 for

296: the SL and LL2 modules and KsiDra for the LL1 module. For a given

297: module, order, and nod position, the background in the two-dimensional

298: spectrum was removed by subtracting the combined two-dimensional data

299: taken with the same module, but adjacent order.  One-dimensional

300: spectra were extracted from the background-subtracted two-dimensional

301: spectra using the SMART data reduction package \citep{Higdon04}.  The

302: adopted extraction apertures are four pixels at the blue end of each

303: order and expand linearly with wavelength.

304: 

305: The 6.2 and 11.3 $\micron$ PAH features were measured by integrating

306: the flux above a spline-interpolated continuum.  For those sources

307: with strong 6 $\micron$ water ice absorption, the apparent 6.2

308: $\micron$ EQW has been corrected by using an inferred 6.2 $\micron$

309: continuum defined by a spline interpolation between 5--26 $\micron$,

310: using pivot points at 5.2, 5.6, 7.8 (in the case of very weak PAH

311: emission), 14, and 26 $\micron$.  This correction results in a lower

312: 6.2 $\micron$ PAH EQW, and is correct under the assumption that the

313: PAH emission is unaffected by the bulk of the ice absorption.  A full

314: description of the ice-fitting procedure is given by \cite{Spoon07},

315: who first present the measured 6.2 $\micron$ EQWs for many of these

316: ULIRGs (see \S 3.2 and Figure 2 for a description of the correction

317: and Figure 1 for an indication of the size of the correction).  The

318: PAH emission at 11.3 $\micron$ is affected by the broad 9.7 $\micron$

319: silicate absorption feature, which can be deep in ULIRGs.  However,

320: lacking a strong understanding of how the obscuring dust is

321: distributed, we have not attempted to correct for it.  Of the 107

322: ULIRGs analyzed in this paper, the 6.2 $\micron$ PAH feature is

323: detected for all but 11 sources, and the 11.3 $\micron$ PAH feature is

324: detected for all but 12 sources.

325: 

326: In the following, we investigate the variation of the PAH features

327: with rest-frame continuum fluxes at 5.5, 24, 25, and 60 $\micron$.

328: The 5.5 $\micron$ flux is taken to be the average continuum betweeen

329: 5.3 and 5.8 $\micron$.  The 24 $\micron$ flux was calculated by

330: convolving the IRS spectrum with the MIPS 24 $\micron$ bandpass.  In

331: the 10 cases where the IRS spectrum did not extend over the full

332: wavelength range of the wide IRS bandpass due to trimming of pipeline

333: artifacts (e.g.~fringes), it was extended by linearly interpolating in

334: log-log space between the IRS spectrum and the 60 $\micron$ IRAS flux

335: density \citep{Moshir90}.  No color-correction was made to place these

336: filter fluxes on the MIPS absolute scale, but these corrections are

337: small ($<$5\%) based on the spectral shape of the ULIRGs within the 24

338: $\micron$ bandpass (see the MIPS Data Handbook).  The 25 $\micron$

339: rest-frame flux was estimated by taking the median of the trimmed and

340: (when needed) interpolated IRS spectrum between 24.7 and 25.3

341: $\micron$.  The rest-frame 60 $\micron$ luminosity was estimated by

342: linearly interpolating in log-log space between the observed IRAS 60

343: and 100 $\micron$ flux densities.  For 12 ULIRGs, this interpolation

344: likely underestimates the 60 $\micron$ luminosity, either because the

345: object is at a redshift high enough ($z > 0.65$) that the observed 100

346: $\micron$ IRAS observation corresponds to a rest-frame wavelength less

347: than 60 $\micron$, or because the 100 $\micron$ observation resulted

348: in only a limit, rather than a detection.

349: 

350: Classifications based on optical spectra are available from the

351: literature for 64 of the 107 ULIRGs in our sample.  Of these, 15 have

352: HII-like optical spectra, 22 are LINER-like, and 27 are Seyfert-like.

353: 

354: \section{Results}

355: 

356: \subsection{PAH strength and silicate absorption versus spectral slope and optical classification}

357: \label{sec:medianspectra}

358: 

359: The median IRS spectra for warm and cold ULIRGs are shown in the

360: left-hand panel of Figure \ref{fig:medianspectra_warmcold}.  Cold

361: ULIRGs have a median 6.2 $\micron$ PAH EQW that is a factor of six

362: higher than the median measured for warm ULIRGs (0.24 versus 0.04

363: $\micron$; see Table \ref{table:pahew_v_class}). The right-hand panel

364: of Figure \ref{fig:medianspectra_warmcold} shows the average starburst

365: spectrum from \citet{Brandl06} plotted over the median cold ULIRG

366: spectrum.  Cold ULIRGs have a median 6.2 $\micron$ PAH EQW which is

367: about half the median for lower-luminosity starbursts (see also

368: \citet{Imanishi07}).  The difference in the PAH EQWs in the median

369: warm and cold ULIRG spectra reflects the strong trend between PAH EQW

370: and mid-infrared spectral type shown in Figure \ref{fig:pah_v_slope}.

371: There is a huge range in measured PAH EQW among the sample ULIRGs.

372: The full range in PAH EQW for those ULIRGs with detected emission is

373: 0.006--0.864 $\micron$ for the 6.2 $\micron$ feature, and 0.006--1.169

374: $\micron$ for the 11.3 $\micron$ feature.  The smallest upper limits

375: are 0.005 and 0.004 $\micron$ for the 6.2 and 11.3 $\micron$ features,

376: respectively.  While the median 6.2 $\micron$ PAH EQW among cold/HII

377: ULIRGs is about half as large as for local starburst galaxies, the

378: median 11.3 $\micron$ PAH EQW in cold/HII ULIRGs is 80\% of that

379: measured in starburst galaxies.  The cold ULIRGs and starburst

380: galaxies track each other over the limited range of $ 0.06 <S_{25} /

381: S_{60} < 0.2$ (see Figure \ref{fig:pah_v_slope}).

382: 

383: Figure \ref{fig:medianspectra_warmcold} also demonstrates that the

384: median cold source shows stronger silicate absorption at both 9.7 and

385: 18 $\micron$ than the median warm source (1.36 versus 0.65 for

386: $\tau_{9.7}$, the optical depth at 9.7 $\micron$), and the median cold

387: ULIRG has stronger absorption than the average starburst (1.36 versus

388: 0.22).  In Figure \ref{fig:tau_v_slope}, we plot the apparent silicate

389: optical depth as a function of far-infrared spectral slope.  While the

390: ULIRGs with $\tau_{9.7} > 1$ have steeper far-infrared spectra than

391: the sources with $\tau_{9.7} < 1$, we find no real correlation (see

392: also \citet{Imanishi07,Hao07}).  In fact, many sources with very high

393: silicate optical depths ($\tau_{9.7} > 2$) would also be classified as

394: warm ULIRGs.  All of the ULIRGs with Seyfert-1 optical spectra, and

395: many, but not all, of the ULIRGs with Seyfert-2 optical spectra have

396: $\tau_{9.7} < 1$ and $ S_{25} / S_{60} > 0.2$.

397: 

398: The median HII, LINER, and Seyfert ULIRG IRS spectra are shown in the

399: left-hand panel of Figure \ref{fig:medianspectra_optical}. The median

400: and mean values of the 6.2 and 11.3 $\micron$ PAH EQW for each of

401: these spectral classes is given in Table \ref{table:pahew_v_class}.

402: As expected, the HII-like ULIRGs have the largest median 6.2 $\micron$

403: PAH EQW (0.28 $\micron$) while the ULIRGs classified as Seyferts have

404: the lowest (0.04 $\micron$).  The median PAH EQW for infrared cold

405: ULIRGs is comparable to the median for those classified as HII-like,

406: while the median PAH EQW for infrared warm sources is similar to the

407: median for Seyfert-like ULIRGs.  There are a number of ULIRGs with

408: HII-like or LINER-like spectra that have unusually low 6.2 $\micron$

409: PAH EQW ($\lsim$0.15 $\micron$, or about 1/4 that found for pure

410: starburst galaxies).  Of the 22 LINER-like (15 HII-like) ULIRGs in the

411: sample, 12 (six) have 6.2 $\micron$ PAH EQW $\le 0.15$ $\micron$.

412: Similarly, of the 66 cold ULIRGs, 23 have a 6.2 $\micron$ PAH EQW $\le

413: 0.15$ $\micron$.

414: 

415: The right-hand panel of Figure \ref{fig:medianspectra_optical} shows

416: the median HII and LINER ULIRG spectra overplotted with the average

417: starburst spectrum from \citet{Brandl06} and the average

418: infrared-bright LINER spectrum from \citet{Sturm06}.  Although sources

419: with HII-like optical spectra have the largest PAH EQWs among ULIRGs,

420: the median values are only about $50\%$ of those measured for pure

421: starburst galaxies, the same result as found for the cold ULIRGs.

422: HII-like ULIRGs also have stronger silicate absorption at 9.7 and 18

423: $\micron$ than seen in lower-luminosity starbursts (1.52 versus 0.22

424: for $\tau_{9.7}$).  Compared to infrared-bright LINERs, LINER-like

425: ULIRGs have $\sim 25$\% smaller PAH EQWs, and stronger silicate

426: absorption ($\tau_{9.7} = 2.11$ for the LINER-like ULIRGs while it is

427: only 0.84 for the IR-bright LINERs).

428: 

429: 

430: \subsection{PAH equivalent width versus continuum luminosity}

431: \label{sec:pahew_v_lum}

432: 

433: In Figure \ref{fig:pahew_v_lum24}, we plot the 6.2 and 11.3 $\micron$

434: PAH EQW versus 24 $\micron$ rest-frame luminosity.  This continuum

435: wavelength was chosen because it samples the peak emission of warm

436: ($\sim$120 K) thermally-emitting dust, it can be measured much more

437: accurately than LIR in many cases, it is free from strong emission or

438: absorption features, and it allows direct comparison to a variety of

439: low-redshift samples observed with the MIPS 24 $\micron$ filter and to

440: $z \sim 2$ samples observed with the MIPS 70 $\micron$ filter.  For

441: reference, we estimate that $\log_{10}(\nu L_{\nu}(24 \micron) / {\rm

442: LIR}) = -1.06 \pm 0.47$ for cold ULIRGs and $-0.79 \pm 0.38$ for warm

443: ULIRGs.  These estimates were made by applying the

444: \citet{SandersMirabel96} prescription for converting IRAS photometry

445: into LIR to the nearest ($z < 0.2$) ULIRGs in our sample.  Figure

446: \ref{fig:pahew_v_lum24} shows a trend of decreasing 6.2 and 11.3

447: $\micron$ PAH EQW with increasing rest-frame 24 $\micron$ luminosity.

448: There is significant scatter in this relation, and ULIRGs with low PAH

449: EQW span a wide range in luminosity.  Nevertheless, the presence of a

450: correlation is verified by a highly significant Spearman correlation

451: coefficient, whether or not upper limits are included in the same way

452: as detections in the calculation.

453: 

454: We have divided the ULIRG sample into six luminosity bins chosen to

455: span the full range in 24 $\micron$ luminosity ($(0.03-8)$ $\times$

456: 10$^{12}$ L$_{\odot}$), and to contain approximately equal numbers

457: (17-18) of ULIRGs.  The median 6.2 and 11.3 $\micron$ PAH EQWs drop by

458: an order of magnitude between the first and last luminosity bins (see

459: Table \ref{table:pahew_v_lum}).  The median ULIRG spectrum in each 24

460: $\micron$ luminosity class is shown in Figure \ref{fig:medianspectra}.

461: The median spectra were constructed using all sources in a given

462: luminosity bin, even those for which only an upper limit on the PAH

463: EQW could be measured.  Moving from lowest to highest 24 $\micron$

464: luminosity (top to bottom in Figure \ref{fig:medianspectra}), the PAH

465: features diminish and the spectrum flattens.  While the median

466: spectrum in the highest luminosity bin has a noticeably smaller

467: silicate absorption than the lower-luminosity median spectra, there is

468: a large range in the apparent silicate optical depth among the most

469: luminous sources.  This is consistent with the results of

470: \citet{Spoon07}, who find a dichotomy in silicate optical depths among

471: sources with small 6.2 $\micron$ PAH EQW (see also

472: \S{\ref{sec:tau_v_pahew}).

473: 

474: 

475: \subsection{Silicate absorption versus PAH equivalent width}

476: \label{sec:tau_v_pahew}

477: 

478: Figure \ref{fig:fork} shows the silicate optical depth versus 6.2

479: $\micron$ PAH EQW for the ULIRGs in our sample.  Such a plot was first

480: presented by \citet{Spoon07}, who discussed the presence of two

481: discrete tracks, or branches.  Figure \ref{fig:fork} shows that the

482: horizontal branch is populated almost exclusively by infrared warm

483: sources, while the diagonal branch is populated by about half of the

484: warm sources and virtually all of the cold sources.  Examining the

485: ULIRGs for which we have optical spectroscopic classifications, the

486: diagonal branch harbors about half of the Seyfert-like ULIRGs, and

487: nearly all of the HII-like and LINER-like ULIRGs.  The horizontal

488: branch is populated mainly by ULIRGs showing Seyfert-like spectra,

489: including all of the Type I sources.  ULIRGs with Seyfert Type 2

490: optical spectra are found on both branches, spanning (nearly) the full

491: range in silicate optical depth.  Most of the ULIRGs without optical

492: spectroscopic classifications lie on the diagonal branch.  As noted in

493: \S{\ref{sec:medianspectra}}, a significant fraction of the HII and

494: LINER-like ULIRGs have very low 6.2 $\micron$ PAH EQWs ($\le$0.15

495: $\micron$).  Of these, almost all of the LINERs (11/12) and half (3/6)

496: of the HII-like ULIRGs have $\tau_{9.7} \ge 2.5$, implying A$_{V} \ge

497: 45$ mag.

498: 

499: 

500: \section{Discussion}

501: \label{sec:Discussion}

502: 

503: The median 6.2 $\micron$ PAH EQWs of cold (or HII-like) ULIRGs is

504: about 50\% of that seen in lower-luminosity starbursts.  Warm ULIRGs

505: have 6.2 $\micron$ PAH EQWs that are up to two orders of magnitude

506: smaller than those found in starbursts.  There are several possible

507: explanations for the extremely large range in PAH EQW and the overall

508: low PAH EQW in ULIRGs compared to starbursts.  Some ULIRGs may host a

509: central AGN, and the associated soft X-ray and ultraviolet radiation

510: may destroy PAH molecules \citep[e.g.][]{Aitken85,Voit92}, or

511: thermalize the emission.  The AGN may also heat dust to significant

512: temperatures ($>$400 K), resulting in excess continuum emission which

513: would lower the measured 6.2 $\micron$ PAH EQW, in efffect diluting

514: the starburst-driven PAH emission with respect to the AGN-powered, hot

515: dust continuum.  For either of these explanations, $55\pm13$\% of the

516: LINER-like ULIRGs and $40\pm15$\% of the HII-like ULIRGs in our sample

517: would harbor buried AGN.  In comparison, \citet{Lutz99} find the

518: percentage of LINER-like ULIRGs classified by ISO as AGN to be

519: $17\pm15$\% and the percentage of HII-like ULIRGS classified by ISO as

520: AGN to be $20\pm12$\%.  We are thus finding a much higher fraction of

521: LINERs and HII-like ULIRGs with weak PAH emission, although, given the

522: small sample sizes and the fact that the percentage is comparable to

523: the statistical errors in the ISO sample, our results are formally

524: consistent with those of \citet{Lutz99} at the 2$\sigma$ level.  Our

525: results are more comparable to those of \citet{Imanishi07}, who find

526: that 30--50\% of HII- and LINER-like ULIRGs contain dominant buried

527: AGN. Of the 12 LINER-like and 6 HII-like ULIRGs with low PAH EQWs, 9

528: were observed with the high-resolution modules of the IRS

529: \citep{Farrah07}.  Of these, none have detectable [NeV]14.3$\micron$,

530: [NeV]24.3$\micron$, or [OIV]24.9$\micron$ line emission.  Upper limits

531: on the [NeV]14.3$\micron$ to [NeII]12.8$\micron$ line flux ratio range from

532: 0.02 to 0.3, corresponding to starburst-dominated ULIRGs with less

533: than $5-25$\% AGN contribution \citep{Armus07}.

534: However, it is important to note that many of these sources are deeply

535: buried, as evidenced by their large silicate optical depths (see

536: \S{\ref{sec:tau_v_pahew}).  As a result, the

537: [NeV]14.3$\micron$ emission may be highly extinguished.  In addition,

538: it is plausible that a completely buried starburst exists in these

539: sources, as in NGC 1377 \citep {Roussel06}.

540: 

541: The 6.2 $\micron$ PAH EQW in ULIRGs could also be (intrinsically)

542: smaller than that found in pure starburst galaxies if star formation

543: in ULIRGs is accompanied by larger quantities of hot dust emission

544: than is typically found in lower-luminosity starbursts.  The dust

545: would then have to ``see" more of the stellar radiation and compete

546: more effectively for UV photons, thus raising its average temperature,

547: something that is not unreasonable given the compact nature of the

548: infrared emission in ULIRGs. However, if this were the case, then both

549: the 6.2 and the 11.3 $\micron$ PAH EQW would be similarly depressed.

550: However, among cold ULIRGs, the median 11.3 $\micron$ PAH EQW is 80\%

551: of that found in starbursts, while the median 6.2 $\micron$ PAH EQW is

552: only 50\%.  A combination of dilution and extinction offers a more

553: likely explanation.  In this scenario, the 6.2 $\micron$ PAH EQW is

554: diluted by hot dust continuum produced by a central ionizing source

555: interior to the star-forming regions.  The dilution of the 11.3

556: $\micron$ PAH EQW is less significant because of its proximity to the

557: broad silicate absorption feature at 9.7 $\micron$.  The hot dust

558: continuum responsible for diluting the 6.2 $\micron$ PAH EQW is

559: heavily extincted at 11.3 $\micron$, and therefore does not readily

560: dilute the 11.3 $\micron$ PAH EQW.  This is seen in Figure

561: \ref{fig:fork}.  The left-hand panels show that the 6.2 $\micron$ PAH

562: EQWs of ULIRGs on the upper branch deviate strongly from the

563: starbursts.  In contrast, the right-hand panels indicate that ULIRGs

564: have nearly starburst-like 11.3 $\micron$ PAH EQWs because the strong

565: extinction at this wavelength overwhelms all but the strongest

566: contribution from hot dust.

567: 

568: Finally, extinction is not likely to play a large role in the reduced

569: 6.2 $\micron$ PAH EQWs found in ULIRGs compared to starbursts.

570: Although ULIRGs on average display stronger silicate absorption

571: features than lower-luminosity starbursts, larger column densities of

572: dust would lead to depressed PAH EQWs only if the PAH emission were

573: more extinguished than the continuum.  Such a situation is unlikely,

574: since the continuum flux is emitted by hot dust that must lie close to

575: the ionization source.

576: 

577: In a further attempt to disentangle reduced PAH emission from

578: increased hot dust emission as the driving force behind the low 6.2

579: $\micron$ PAH EQWs in ULIRGs, we plot the 6.2 $\micron$ PAH luminosity

580: against the rest-frame 5.5 $\micron$ luminosity in Figure

581: \ref{fig:pahlum_v_lum5p5}.  Low-luminosity starbursts from

582: \citet{Brandl06} follow a correlation between these two quantities,

583: characterized by $\log_{10}$[L(6.2 $\micron$ PAH)] = -0.49 + 0.96

584: $\times$ $\log_{10}[\nu L_{\nu}(5.5 \micron)]$.  ULIRGs do not lie on

585: the extrapolation of this relation to higher luminosities. Instead,

586: ULIRGs appear offset to higher 5.5 $\micron$ luminosities for their

587: measured 6.2 $\micron$ PAH luminosities.  Not surprisingly, sources

588: with flatter far-infrared spectra and those with Seyfert-like optical

589: spectra are, as a group, the most displaced towards higher 5.5

590: $\micron$ luminosity.  However, the range of 6.2 $\micron$ PAH

591: luminosity within the different classes (both optical and infrared) is

592: the same.  It has been suggested that an evolutionary link exists

593: between ULIRGs and QSOs, such that cold ULIRGs evolve into warm ULIRGs

594: on their way to becoming QSOs \citep[e.g.][]{Sanders88}.  Figure

595: \ref{fig:pahlum_v_lum5p5} implies that the transition between cold and

596: warm ULIRGs involves primarily an increase in hot dust, rather than

597: the suppression of PAH emission. \citet{Lutz98} also suggested that

598: dilution of the EQW by hot dust was the dominant effect among ULIRGs.

599: 

600: Also plotted in the lower left-hand panel of Figure

601: \ref{fig:pahlum_v_lum5p5} are $\sim$80 PG QSOs at $z < 0.5$ (Ogle \&

602: Antonucci 2007, in preparation; Shi et al. 2007, submitted).  Most

603: have only upper limits for the 6.2 micron PAH EQW, but 15 have

604: detected values.  The 5.5 $\micron$ continuum

605: luminosities of ULIRGs and PG QSOs span the same range.  

606: %Particularly

607: %interesting is the fact that many PG QSOs have 5.5 $\micron$

608: %luminosities more similar to cold ULIRGs than to warm ULIRGs.  Thus,

609: %if all PG QSOs evolve from warm ULIRGs, the transition must involve

610: %both a reduction in PAH luminosity for most sources and a decrease in

611: %hot dust emission for at least some sources.  This is plausible if AGN

612: %feedback is effective in halting star formation and clearing out some

613: %of the hot dust.  

614: The 15 PG QSOs with detected 6.2 $\micron$ PAH

615: features show no evidence for a correlation in Figure

616: \ref{fig:pahlum_v_lum5p5}.  This is in contrast with the results of

617: \citet{Schweitzer06}, who find a correlation between the 7.7 $\micron$

618: PAH luminosity and the 6 $\micron$ continuum luminosity among a sample

619: of 25 PG QSOs, including 15 with upper limits on the PAH luminosity

620: and 11 with detections.  Based on the assumptions that the 6 $\micron$

621: luminosity is due to hot dust heated by an AGN, while the PAH

622: luminosity is emitted by star-forming regions, they argue that their

623: observed correlation is evidence for a starburst-AGN connection among

624: PG QSOs.  However, we find no such correlation when considering the 6.2

625: $\micron$ PAH feature.

626: 

627: While the flatter spectra and lower 6.2 $\micron$ PAH EQWs of many of

628: the ULIRGs can readily be explained by an excess of hot dust emission,

629: the silicate optical depth, at least in some cases, seems inconsistent

630: with a simple model for the emission.  The 9.7 $\micron$ silicate

631: optical depth in ULIRGs varies from $-0.2 < \tau_{9.7} < 4$, and is on

632: average deeper than in starburst galaxies.  The radiative transfer

633: models of \citet{Levenson07} suggest that the geometry of the

634: obscuring material in ULIRGs determines both the depth of the silicate

635: feature and the infrared spectral slope.  According to these models,

636: the large apparent silicate optical depth ($\tau_{9.7} >$ 1--2)

637: observed in some ULIRGs requires an optically and geometrically thick,

638: smooth dust component, since a clumpy dust distribution can only

639: result in a shallow ($\tau_{9.7} < 1$) absorption feature.  All of the

640: Seyfert-like ULIRGs would then be viewed through this clumpy dust

641: distribution, exhibiting flat infrared spectra, weak PAH EQWs, and

642: shallow (or nonexistent) silicate absorption features.  If this simple

643: model is correct, a correlation between silicate optical depth and

644: spectral slope might be expected.  However, we find no such

645: correlation (see Figure \ref{fig:tau_v_slope}).  In fact, a

646: significant number of the sources with the most extreme absorption are

647: warm (see Figure \ref{fig:fork}).  These ULIRGs also tend to have low

648: 6.2 $\micron$ PAH EQWs ($<0.15 \micron$).  A majority of the sources

649: with large silicate optical depth are classified optically as LINERs.

650: 

651: Among local ULIRGs, the sources with the highest rest-frame 24

652: $\micron$ luminosities have the lowest 6.2 $\micron$ PAH EQWs (see

653: Figure \ref{fig:pahew_v_lum24}).  The fit to this correlation is shown

654: as the black line in Figure \ref{fig:pahew_v_lum24_medians}, and is

655: given by $\log_{10}(6.2 \ \micron \ {\rm PAH \ EQW} [\micron]) = (7.71

656: \pm 0.07) + (-0.723 \pm 0.006) \times \log_{10}(\nu L_{\nu}(24

657: \micron) [L_{\odot}])$.  According to this fit, ULIRGs with rest-frame

658: 24 $\micron$ luminosities above about $10^{12}$ L$_{\odot}$ have 6.2

659: $\micron$ PAH EQWs below 0.1--0.2 $\micron$.  However, very luminous

660: infrared sources with strong PAH emission (6.2 $\micron$ PAH EQW $>$

661: 0.3-0.5 $\micron$) have recently been found in high-redshift samples

662: that were selected in the submillimeter, which is dominated by cool

663: dust emission \citep{Lutz05,MenendezDelmestre07,Valiante07}.  For

664: example, of the five IRS spectra of submillimeter sources presented by

665: \citet{MenendezDelmestre07}, two have the wavelength coverage to

666: measure the 6.2 $\micron$ PAH feature.  These two

667: submillimeter-selected ULIRGs have rest-frame EQWs of approximately

668: 0.6 and 0.8 $\micron$, comparable to the strongest PAH emission

669: measured for local starburst galaxies.  Their positions are marked in

670: Figure \ref{fig:pahew_v_lum24_medians}, and they are clearly well

671: above the fit to the local ULIRGs.  Additional examples have been

672: found in samples selected at 160 $\micron$

673: \citep[MIPSJ142824;][]{Desai06} or a combination of 24 $\micron$ flux

674: and mid-infrared color \citep{Yan05,Sajina07}.  These are also plotted

675: in Figure \ref{fig:pahew_v_lum24_medians}.  However, sources with high

676: rest-frame 24 $\micron$ luminosities and large 6.2 $\micron$ PAH EQW

677: are rare in bright (f$_{24} > 0.7$ mJy) flux-limited samples selected

678: at 24 $\micron$ \citep[e.g.][]{Houck05,Weedman06}; these samples

679: appear dominated by AGN.  ULIRGs with high rest-frame 24 $\micron$

680: luminosities and strong PAH emission are absent from local samples,

681: yet they obviously exist at high redshift, and they can be found by

682: selecting on color, cold dust emission, or by probing to fainter flux

683: levels in the mid-infrared (observed $f_{24} < 0.3$ mJy).  At $z < 1$,

684: the generation of large amounts of warm dust seems to require the

685: presence of a dominant AGN.  The absence of extremely luminous

686: starburst-dominated ULIRGs at low redshift represents a real effect,

687: and not simply an observational bias.

688: 

689: 

690: \section{Summary and Conclusions}

691: 

692: Using a large sample of 107 ULIRGs observed with the IRS on board the

693: \textit{Spitzer Space Telescope}, we investigate the relationships

694: between the 6.2 and 11.3 $\micron$ PAH EQWs, continuum luminosity,

695: infrared spectral slope, optical spectroscopic classification, and

696: silicate optical depth.  Our results can be summarized as follows:

697: 

698: \begin{enumerate}

699: 

700: \item There is an extremely large range in detected 6.2 $\micron$ PAH

701: EQWs among ULIRGs, spanning more than two orders of magnitude from

702: $\sim$0.006--0.9 $\micron$.  ULIRGs classified as starburst-dominated

703: based on their far-infrared colors or optical spectra have median 6.2

704: $\micron$ PAH EQWs which are 50\% of that measured in low-luminosity

705: starburst galaxies.  An excess of hot dust, which is most prominent in

706: the far-infrared warm or Seyfert-like ULIRGs, appears to be the cause

707: of the decreased PAH EQW, as opposed to extinction or grain

708: destruction.  Approximately 55\% of LINER-like ULIRGs, 40\% of

709: HII-like ULIRGs, and 35\% of cold ULIRGs have very low 6.2 $\micron$

710: PAH EQWs, indicative of buried AGN or starbursts with supressed PAH

711: emission.  Most of these ULIRGs also have high apparent silicate

712: optical depths at 9.7 $\micron$.

713: 

714: %\item The 11.3 $\micron$ PAH EQWs of starburst-dominated ULIRGs

715: %deviate less strongly from those of lower-luminosity starburst

716: %galaxies.  This suggests dilution of the PAH EQW by excess hot dust

717: %emission from a central deeply embedded source.  In AGN-dominated

718: %ULIRGs, no systematic differences between the 6.2 and 11.3 $\micron$

719: %PAH EQWs are observed, indicating dilution of the PAH EQW by excess

720: %hot dust emission from a central source that is only lightly

721: %enshrouded.

722: 

723: %\item The 6.2 $\micron$ PAH EQW is strongly anticorrelated with

724: %mid-infrared spectral slope ($f_{\nu}(25)/f_{\nu}(60)$).  This is

725: %likely due to the presence of hot dust leaking out of the core,

726: %and diluting the 6.2 $\micron$ PAH EQW of surrounding star

727: %formation.

728: 

729: \item The apparent silicate optical depth is not correlated with

730: far-infrared spectral slope.  The ULIRGs with the strongest silicate

731: absorption features have small 6.2 $\micron$ PAH EQWs and flat

732: spectral slopes in the far-infrared.  Their strong silicate absorption

733: implies large columns of cold dust, but their small PAH EQWs and

734: far-infrared colors imply substantial amounts of hot dust emission.

735: For these sources, a model in which a warm source is viewed through

736: either a smooth optically thick or clumpy dust shell seems

737: inconsistent with the data.

738: 

739: %\item The handful of Type 1 Seyfert-like ULIRGs all have low apparent

740: %silicate optical depths, however, suggesting that in these ULIRGs with

741: %low apparent silicate optical depth have clumpy dust distributions,

742: %while ULIRGs with high apparent silicate optical depth have smooth

743: %dust distributions which are optically and geometrically thick

744: %\citep{Nenkova02,Levenson07,Spoon07}.  

745: 

746: %\item Approximately 55\% of LINER-like ULIRGs, 40\% of HII-like

747: %ULIRGs, and 35\% of cold ULIRGs have very low 6.2 $\micron$ PAH EQWs

748: %indicative of buried nuclei.  Most of these also have high apparent

749: %silicate optical depths at 9.7 $\micron$.

750: 

751: \item ULIRGs with the largest rest-frame 24 $\micron$ luminosities

752: have the smallest 6.2 $\micron$ PAH EQWs.  A fit to our sample

753: suggests $\log_{10}(6.2 \ \micron \ {\rm PAH \ EQW} [\micron]) = (7.71

754: \pm 0.07) + (-0.723 \pm 0.006) \times \log_{10}(\nu L_{\nu}(24

755: \micron) [L_{\odot}])$.  The range in PAH EQWs among luminous,

756: high-redshift sources observed with Spitzer is much larger than

757: implied by this correlation due to the presence of luminous sources

758: with very large 6.2 $\micron$ EQWs.  Extremely luminous sources of all

759: types are rare at low-redshift, but those that do exist seem to be AGN

760: dominated.

761: 

762: \end{enumerate}

763: 

764: \acknowledgements

765: 

766: We would like to thank R. Chary, M. Lacy, J. Surace, and A. Sajina for

767: insightful discussions.  Support for this work was provided by NASA

768: through an award issued by JPL/Caltech.

769: 

770: \bibliographystyle{apj}

771: %\bibliography{/scr5/desai/latex/papers/references}

772: \bibliography{references}

773: 

774: \clearpage

775: 

776: \begin{deluxetable}{cccccccccc}

777:   \tabletypesize{\scriptsize}

778:   \tablecaption{6.2 and 11.3 $\micron$ PAH EQWs versus spectral class}

779:   \tablecolumns{10}

780:   \tablewidth{0pc}

781:   \tablehead{

782:     \colhead{} & 

783:     \colhead{} & 

784:     \multicolumn{2}{c}{uncorrected $6.2\mu {\rm m}$ PAH EQW} &

785:     \colhead{} &

786:     \multicolumn{2}{c}{corrected $6.2\mu {\rm m}$ PAH EQW} &  

787:     \colhead{} & 

788:     \multicolumn{2}{c}{11.3 $\micron$ PAH EQW}\\

789:     \cline{3-4} \cline{6-7} \cline{9-10} \\

790:     \colhead{spectral class} &

791:     \colhead{number}   &

792:     \colhead{median} &

793:     \colhead{average} &

794:     \colhead{} &

795:     \colhead{median} &

796:     \colhead{average} & 

797:     \colhead{} &

798:     \colhead{median} &

799:     \colhead{average} \\

800:     \colhead{} &

801:     \colhead{} &

802:     \colhead{($\mu$m)} &

803:     \colhead{($\mu$m)} &

804:     \colhead{} &

805:     \colhead{($\micron$)} &

806:     \colhead{($\micron$)} &

807:     \colhead{} &

808:     \colhead{($\micron$)} &

809:     \colhead{($\micron$)}

810: }

811: 

812: \startdata

813: HII   & 15 & 0.32 & 0.30 $\pm$ 0.16 & & 0.28 & 0.27 $\pm$ 0.18 & & 0.39 & 0.41 $\pm$ 0.20\\

814: LINER & 22 & 0.19 & 0.27 $\pm$ 0.20 & & 0.11 & 0.17 $\pm$ 0.14 & & 0.50 & 0.46 $\pm$ 0.18\\

815: S1    &  6 & 0.02 & 0.04 $\pm$ 0.05 & & 0.02 & 0.04 $\pm$ 0.05 & & 0.03 & 0.03 $\pm$ 0.03\\

816: S2    & 21 & 0.05 & 0.10 $\pm$ 0.11 & & 0.05 & 0.08 $\pm$ 0.07 & & 0.09 & 0.16 $\pm$ 0.15\\

817: S1+S2 & 27 & 0.05 & 0.09 $\pm$ 0.10 & & 0.04 & 0.07 $\pm$ 0.07 & & 0.07 & 0.13 $\pm$ 0.14\\

818:       &    & & & &      &                 & &      &                \\

819: Cold  & 66 & 0.30 & 0.32 $\pm$ 0.18 & & 0.24 & 0.27 $\pm$ 0.18 & & 0.50 & 0.52 $\pm$ 0.24\\

820: Warm  & 41 & 0.04 & 0.08 $\pm$ 0.09 & & 0.04 & 0.06 $\pm$ 0.08 & & 0.07 & 0.13 $\pm$ 0.13\\

821: \enddata

822: 

823: \tablecomments{Statistics on the 6.2 and 11.3 $\micron$ PAH EQWs of

824: the ULIRGs in our sample, broken down by spectral class (Column 1).

825: Column 2 gives the number of ULIRGs in each class.  Columns 3 and 4

826: show the median and average 6.2 $\micron$ PAH EQWs, respectively,

827: uncorrected for ice absorption.  Columns 5 and 6 give the same

828: statistics for the ice-corrected 6.2 $\micron$ PAH EQW (see

829: \citet{Spoon07} for details on this correction).  Column 7 shows the

830: median and average 11.3 $\micron$ PAH EQWs, uncorrected for the

831: underlying silicate absorption.  All medians and averages were

832: computed by including sources with upper limits as though they were

833: detections.  Optical classifications are from

834: \citet{Duc97,Veilleux97,Veilleux95,Cutri94,Allen91,Armus89,Kleinmann88,Sanders88b}.

835: The far-infrared classifications of cold ($f_{\nu}(25

836: \micron)/f_{\nu}(60 \micron) < 0.2$) and warm ($f_{\nu}(25

837: \micron)/f_{\nu}(60 \micron) \ge 0.2$) are based on the rest-frame 25

838: and 60 $\micron$ values calculated as described in

839: \S{\ref{sec:Observations}}.}

840: 

841: \label{table:pahew_v_class}

842: \end{deluxetable}

843: 

844: \begin{deluxetable}{ccccccccccc}

845:   \tabletypesize{\scriptsize}

846:   \tablecaption{6.2 and 11.3 $\micron$ PAH EQW versus 24 $\micron$ luminosity}

847:   \tablecolumns{11}

848:   \tablewidth{0pc}

849:   \tablehead{

850:     \multicolumn{2}{c}{24 $\micron$}& \colhead{} & \multicolumn{2}{c}{uncorrected 6.2 $\micron$ PAH EQW} & \colhead{} & \multicolumn{2}{c}{corrected 6.2 $\micron$ PAH EQW} & \colhead{} & \multicolumn{2}{c}{11.3 $\micron$ PAH EQW}\\

851:     \cline{1-2} \cline{4-5} \cline{7-8} \cline{10-11}\\

852:     \colhead{Class} &

853:     \colhead{Luminosity}   &

854:     \colhead{} &

855:     \colhead{median} &

856:     \colhead{average} &

857:     \colhead{} &

858:     \colhead{median} &

859:     \colhead{average} & 

860:     \colhead{} & 

861:     \colhead{median} & 

862:     \colhead{average}\\

863:     \colhead{} &

864:     \colhead{($10^{12}$L$_\odot$)} &

865:     \colhead{} &

866:     \colhead{($\mu$m)} &

867:     \colhead{($\mu$m)} &

868:     \colhead{} &

869:     \colhead{($\mu$m)} &

870:     \colhead{($\mu$m)} & 

871:     \colhead{} &

872:     \colhead{($\mu$m)} &

873:     \colhead{($\mu$m)}

874: }

875: \startdata

876: 1 & $0.03-0.20$ & & 0.43 & 0.47 $\pm$ 0.16 & & 0.41 & 0.41 $\pm$ 0.18 & & 0.58 & 0.61 $\pm$ 0.20\\

877: 2 & $0.20-0.30$ & & 0.33 & 0.34 $\pm$ 0.18 & & 0.29 & 0.30 $\pm$ 0.17 & & 0.56 & 0.52 $\pm$ 0.24\\

878: 3 & $0.30-0.44$ & & 0.16 & 0.23 $\pm$ 0.17 & & 0.10 & 0.17 $\pm$ 0.13 & & 0.44 & 0.39 $\pm$ 0.18\\

879: 4 & $0.44-0.63$ & & 0.14 & 0.17 $\pm$ 0.11 & & 0.12 & 0.15 $\pm$ 0.11 & & 0.31 & 0.30 $\pm$ 0.15\\

880: 5 & $0.63-1.15$ & & 0.10 & 0.12 $\pm$ 0.10 & & 0.06 & 0.07 $\pm$ 0.05 & & 0.19 & 0.31 $\pm$ 0.40\\

881: 6 & $1.15-8.00$ & & 0.03 & 0.04 $\pm$ 0.05 & & 0.02 & 0.04 $\pm$ 0.05 & & 0.05 & 0.08 $\pm$ 0.08\\

882: \enddata

883: 

884: \tablecomments{Same as Table \ref{table:pahew_v_class}, except the

885: ULIRG sample is broken down into luminosity bins, rather than spectral

886: class.  Each bin contains 17-18 ULIRGs.}

887: \label{table:pahew_v_lum}

888: \end{deluxetable}

889: 

890: %%FIGURE 1

891: 

892: \begin{figure}

893: \centerline{

894: \plottwo{f1a.ps}{f1b.ps}

895: }

896: 

897: \caption{\textit{Left:} The median IRS spectra of ULIRGs classified as

898: cold (red) and warm (blue), arbitrarily normalized for display

899: purposes. \textit{Right:} The median spectrum of cold ULIRGs (red)

900: compared to the average starburst spectrum from \citet{Brandl06}

901: (black), both normalized at 24 $\micron$.}

902: 

903: \label{fig:medianspectra_warmcold}

904: \end{figure}

905: 

906: %%FIGURE 2

907: \begin{figure}

908: \centerline{

909: \includegraphics[scale=1.0]{f2.eps}

910: }

911: \caption{The 6.2 $\micron$ (top) and 11.3 $\micron$ (bottom) PAH EQW

912:   versus spectral slope.  The ULIRGs are color-coded by optical

913:   spectroscopic classification: ULIRGs with HII-like optical spectra

914:   are plotted as red stars, LINER-like ULIRGs are shown as green

915:   squares, ULIRGs with Seyfert 1 optical spectra are shown as light

916:   blue triangles, ULIRGs with Seyfert 2 optical spectra are shown as

917:   dark blue upside-down triangles, and sources with unknown optical

918:   spectroscopic classifications are shown as white circles.  The black

919:   dots represent the starburst galaxies from \citet{Brandl06}.  The

920:   dotted vertical line represents the division between warm and cold

921:   sources adopted for this paper ($f_{\nu}(25) / f_{\nu}(60) = 0.2$),

922:   where cold sources lie to the left of the line and warm sources lie

923:   to the right.}

924: 

925: \label{fig:pah_v_slope}

926: \end{figure}

927: 

928: %%FIGURE 3

929: \begin{figure}

930: \centerline{

931: \includegraphics[scale=0.6]{f3.eps}

932: }

933: 

934: \caption{The 9.7 $\micron$ silicate optical depth versus the

935:   mid-infrared spectral slope.  The ULIRGs are color-coded by optical

936:   spectroscopic classification, as in Figure \ref{fig:pah_v_slope}.

937:   The vertical dotted line is $f_{\nu}(25) / f_{\nu}(60) = 0.2$, the

938:   division between warm and cold sources adopted for this paper.  The

939:   spectral slope for the ULIRGs was computed from the IRS spectra as

940:   described in \S{\ref{sec:Observations}} and the spectral slope for

941:   the starbursts was computed from IRAS fluxes taken from

942:   \citet{Brandl06}.}

943: 

944: \label{fig:tau_v_slope}

945: \end{figure}

946: 

947: %%FIGURE 4

948: 

949: \begin{figure}

950: \centerline{

951: \plottwo{f4a.ps}{f4b.ps}

952: }

953: 

954: \caption{\textit{Left:} The colored lines show the median IRS spectra

955: of ULIRGs optically classified as HII-like (red), LINER-like (green),

956: and Seyfert-like (blue).  The black line shows the median spectrum of

957: ULIRGs lacking optical classifications.  The normalizations are

958: arbitrary.  \textit{Right:} The median spectrum of HII-like ULIRGs

959: compared to the average starburst spectrum from \citet{Brandl06} (top

960: red and black lines respectively) and the median spectrum of

961: LINER-like ULIRGs compared to the average infrared-bright LINER

962: spectrum from \citet{Sturm06} (bottom green and black lines

963: respectively).  For wavelengths longer than 10 $\micron$, the

964: high-resolution \citet{Sturm06} spectrum was convolved with a Gaussian

965: profile with a FWHM of 0.2 $\micron$ to approximately match the

966: resolution of the low-resolution ULIRG spectrum.  The comparison

967: spectra were normalized to match the HII- and LINER-like ULIRGs at 24

968: $\micron$.}

969: 

970: \label{fig:medianspectra_optical}

971: \end{figure}

972: 

973: %%FIGURE 5

974: 

975: \begin{figure}

976: \includegraphics[scale=1.0]{f5.eps}

977: 

978: \caption{The 6.2 $\micron$ (top row) and 11.3 $\micron$ (bottom row)

979:   PAH EQWs versus 24 $\micron$ rest-frame luminosity.  The ULIRG

980:   sample is color-coded by warm ($f_{\nu}(25 \micron)/f_{\nu}(60

981:   \micron) \ge 0.2$) and cold ($f_{\nu}(25 \micron)/f_{\nu}(60

982:   \micron) < 0.2$) mid-infrared spectral slope (left) and by optical

983:   spectroscopic classification as in Figure \ref{fig:tau_v_slope}

984:   (right).  The average PAH EQWs of starburst galaxies from

985:   \citet{Brandl06} are indicated by solid horizontal lines and the

986:   $\pm$1 $\sigma$ dispersions in these averages are shown by dotted

987:   horizontal lines.  Upper limits ($3\sigma$) to the PAH EQWs are

988:   indicated by downward pointing arrows. The six luminosity classes

989:   (1-6; see text and Table \ref{table:pahew_v_lum}) are labeled on

990:   the bottom of each panel and are alternately shaded for clarity.}

991: 

992: \label{fig:pahew_v_lum24}

993: \end{figure}

994: 

995: %%FIGURE 6

996: 

997: \begin{figure}

998: \centerline{ \includegraphics[scale=0.8]{f6.ps}}

999: 

1000: \caption{Median IRS ULIRG spectra for the six 24 $\micron$ rest-frame

1001: luminosity bins shown in Figure \ref{fig:pahew_v_lum24}, arranged such

1002: that the luminosity increases from top to bottom. The grey bands

1003: highlight the positions of the 6.2 and 11.3 $\micron$ PAH features.}

1004: 

1005: \label{fig:medianspectra}

1006: \end{figure}

1007: 

1008: %%FIGURE 7

1009: 

1010: \begin{figure}

1011: \centerline{

1012: \includegraphics[scale=1.0]{f7.eps}}

1013: 

1014: \caption{The strength of the 9.7 $\micron$ silicate absorption feature

1015: versus the 6.2 (left) and 11.3 (right) $\micron$ PAH EQWs.  The ULIRGs

1016: are color-coded according to spectral slope and optical classification

1017: in the left and right panels, respectively.  The colored symbols are

1018: identical to those in Figure \ref{fig:pahew_v_lum24}.  The small black

1019: dots represent the starbursts from \citet{Brandl06}.}

1020: 

1021: \label{fig:fork}

1022: 

1023: \end{figure}

1024: 

1025: %%FIGURE 8

1026: 

1027: \begin{figure}

1028: \centerline{

1029: \includegraphics[scale=1.0]{f8a.eps}}

1030: \bigskip

1031: \bigskip

1032: \centerline{

1033: \includegraphics[scale=1.0]{f8b.eps}}

1034: 

1035: \caption{Luminosity in the 6.2 $\micron$ PAH feature as a function of

1036:   5.5 $\micron$ luminosity.  In the top panels, the ULIRGs are

1037:   color-coded according to spectral slope (left) and optical

1038:   classification (right) using the same symbols as in Figure

1039:   \ref{fig:pahew_v_lum24}.  The solid black line is a fit to the

1040:   starbursts analyzed by \citet{Brandl06}.  The bottom panel features

1041:   expanded x and y axes to show the locations of the \citet{Brandl06}

1042:   starbursts (red diamonds) and $\sim$80 $z < 0.5$ PG QSOs, displayed

1043:   as cyan squares (detections) and arrows (upper limits).  In this

1044:   bottom panel, the ULIRGs are shown as black circles.}

1045: 

1046: \label{fig:pahlum_v_lum5p5}

1047: \end{figure}

1048: 

1049: 

1050: %%FIGURE 9

1051: 

1052: \begin{figure}

1053: \centerline{

1054: \includegraphics[scale=1.0]{f9.eps}}

1055: 

1056: \caption{Median (open circles) and average (filled circles) 6.2

1057: $\micron$ PAH EQW versus 24 $\micron$ luminosity.  Both the medians

1058: and the means were computed by including sources with PAH EQW upper

1059: limits in the same way as detections.  The error bars were computed as

1060: the dispersion in the PAH EQWs in each luminosity bin, and are

1061: centered on the average values in this plot.  The x-position of each

1062: point is the median of the 24 $\micron$ luminosities of the sources in

1063: each luminosity bin.  The thick red, blue, and black lines are fits to

1064: the cold ULIRGs, the warm ULIRGs, and all of the ULIRGs, respectively.

1065: Sources with upper limits on the PAH EQW were included in the fits as

1066: detections with 10\% error bars.  The horizontal thin and dotted black

1067: lines have the same meaning as in Figure \ref{fig:pahew_v_lum24}, as

1068: do the luminosity bin labels across the bottom.  The green points are

1069: the high-redshift ($z \sim 2$) ULIRGs from \citet{Sajina07} with

1070: measured 6.2 $\micron$ PAH EQWs, and rest-frame 24 $\micron$

1071: luminosities estimated from the 70 $\micron$ MIPS photometry.  The

1072: large black star is MIPS J142824.0+352619, a bright 160 $\micron$

1073: source at $z=1.3$.  The rest-frame 24 $\micron$ luminosity for this

1074: object was computed by via log-log interpolation of the photometry

1075: presented in \citet{Borys06}, and the 6.2 $\micron$ PAH EQW was taken

1076: from \citet{Desai06}.}

1077: 

1078: \label{fig:pahew_v_lum24_medians}

1079: \end{figure}

1080: 

1081: \end{document}

1082: 

1083: 

1084: 

1085: 

1086: 

1087: 

1088: 

1089: