1: \documentclass{emulateapj}

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

3: \usepackage{graphicx}

4: \usepackage{apjfonts}

5: 

6: 

7: 

8: \newcommand{\intensunits}{\mbox{erg s$^{-1}$ cm$^{-2}$ Hz$^{-1}$}}

9: \newcommand{\kmsend}{\mbox{km s$^{-1}$}}

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

11: \newcommand{\hgyr}{\mbox{h$^{-1}$Gyr }}

12: \newcommand{\hmyr}{\mbox{h$^{-1}$Myr }}

13: \newcommand{\hkpc}{\mbox{h$^{-1}$kpc }}

14: \newcommand{\hpc}{\mbox{h$^{-1}$pc }}

15: \newcommand{\hmsun}{\mbox{h$^{-1}$M$_{\sun}$ }}

16: \newcommand{\hmsunend}{\mbox{h$^{-1}$M$_{\sun}$}}

17: \newcommand{\msun}{\mbox{M$_{\sun}$ }}

18: \newcommand{\msunend}{\mbox{M$_{\sun}$}}

19: \newcommand{\lsun}{\mbox{L$_{\sun}$ }}

20: \newcommand{\lsunend}{\mbox{L$_{\sun}$}}

21: \newcommand{\lir}{\mbox{L$_{\rm IR}$}}

22: \newcommand{\lhcn}{\mbox{L$_{\rm HCN}$}}

23: %\newcommand{\lhcnjone}{\mbox{L$_{\rm HCN (J=1-0)}$}}

24: \newcommand{\lhcnjone}{\mbox{HCN (J=1-0)}}

25: \newcommand{\lco}{\mbox{L$_{\rm CO}$}}

26: \newcommand{\lmolecule}{\mbox{L$_{\rm molecule}$}}

27: \newcommand{\lmol}{\mbox{L$_{\rm mol}$}}

28: \newcommand{\ltwentyfive}{\mbox{L$_{\rm 25}$}}

29: \newcommand{\lcojone}{\mbox{CO (J=1-0)}}

30: \newcommand{\lcojthree}{\mbox{CO (J=3-2)}}

31: \newcommand{\ncrit}{\mbox{$n_{\rm crit}$}}

32: 

33: 

34: %\newcommand{\lcojone}{\mbox{L$_{\rm CO (J=1-0)}$}}

35: %\newcommand{\lcojthree}{\mbox{L$_{\rm CO (J=3-2)}$}}

36: \newcommand{\cmthree}{\mbox{cm$^{-3}$}}

37: \newcommand{\cmtwo}{\mbox{cm$^{-2}$}}

38: \newcommand{\msunyr}{\mbox{M$_{\sun}$yr$^{-1}$ }}

39: \newcommand{\msunyrend}{\mbox{M$_{\sun}$yr$^{-1}$}}

40: \newcommand{\htwo}{\mbox{H$_2$}}

41: \newcommand{\mhtwo}{\mbox{$M_{H2}$}}

42: \newcommand{\z}{\mbox{$z$}}

43: \newcommand{\zapprox}{\mbox{$z \approx$}}

44: \newcommand{\zga}{\mbox{$z \ga$}}

45: \newcommand{\zla}{\mbox{$z \la$}}

46: \newcommand{\zsim}{\mbox{$z\sim$ }}

47: \newcommand{\msigma}{\mbox{$M_{\rm BH}$-$\sigma_{\rm v}$ }}

48: \newcommand{\magorrian}{\mbox{$M_{\rm BH}$-$M_{\rm bulge}$ }}

49: \newcommand{\Lbol}{L_{\rm {bol}}}

50: \newcommand{\MBH}{M_{\rm{BH}}}

51: \newcommand{\Msun}{M_{\odot}}

52: \newcommand{\Lsun}{L_{\odot}}

53: \newcommand{\tQ}{\tau_{QSO}}

54: \newcommand{\myr}{\rm {Myr}}

55: \newcommand{\Mvir}{M_{\rm{vir}}}

56: \newcommand{\Vvir}{V_{\rm{vir}}}

57: 

58: \shorttitle{Dense Molecular Gas in Galaxy Nuclei}

59: 

60: \shortauthors{Narayanan et al.}

61: 

62: \slugcomment{ApJL - accepted}

63: 

64: 

65: 

66: \begin{document}

67: \title{The Star Formation Rate - Dense Gas Relation in the Nuclei of

68:   Nearby Galaxies} \author{Desika Narayanan\altaffilmark{1,2}, Thomas

69:   J. Cox\altaffilmark{1,3}, and Lars Hernquist\altaffilmark{1}}

70: \altaffiltext{1}{Harvard-Smithsonian Center for Astrophysics, 60 Garden Street,

71:   Cambridge, MA 02138, USA}

72: \altaffiltext{2}{dnarayanan@cfa.harvard.edu}

73: \altaffiltext{3}{W.M. Keck Postdoctoral Fellow}

74: 

75: \begin{abstract}

76: 

77: We investigate the relationship between the star formation rate (SFR)

78: and dense molecular gas mass in the nuclei of galaxies. To do this, we

79: utilize the observed 850 $\mu$m luminosity as a proxy for the infrared

80: luminosity (\lir) and SFR, and correlate this with the observed CO

81: (J=3-2) luminosity. We find tentative evidence that the \lir-CO

82: (J=3-2) index is similar to the Kennicutt-Schmidt (KS) index ($N

83: \approx$ 1.5) in the central $\sim$1.7 kpc of galaxies, and flattens

84: to a roughly linear index when including emission from the entire

85: galaxy. This result may imply that the volumetric Schmidt relation is

86: the underlying driver behind the observed SFR-dense gas correlations,

87: and provides tentative confirmation for recent numerical models. While

88: the data exclude the possibility of a constant \lir-CO (J=3-2) index

89: for both galaxy nuclei and global measurements at the $\sim$80\%

90: confidence level, the considerable error bars cannot preclude

91: alternative interpretations.

92: 

93: 

94: 

95: %\begin{itemize}

96: 

97: %\item re-run methods using normal linfit and make sure answers are the same

98: 

99: %\item add in the critical density of CO (J=3-2) somehwere in the

100: %sample selection. also note that the \lir-co J3=2 relation was found

101: %to be linear by n05 and postdicted as such by n08b.

102: %\end{itemize}

103: 

104: 

105: 

106: \end{abstract}

107: \keywords{stars: formation -- galaxies: ISM, evolution, starburst --

108: radio lines: ISM, galaxies}

109: 

110: \section{Introduction}

111: The rate at which stars form in galaxies has long been parameterized

112: in terms of a power-law involving the gas density \citep{sch59}.  HI

113: and CO observations of galaxies have shown that the star formation

114: rate (SFR) is related to the surface gas density via $\Sigma_{\rm SFR}

115: \propto \Sigma_{\rm gas}^{1.4-1.5}$ \citep{ken98a}. Theoretically, a

116: power-law index of $\sim$1.5 controlling the SFR is appealing. If a

117: constant fraction of gas is converted into stars over a free fall

118: time, the relation SFR $\propto \rho_{\rm gas}^{1.5}$ results.

119: 

120: Observations of molecular gas in galaxies have provided a more complex

121: view of SFR relations. Local surveys have shown that the SFR (as

122: traced by the \lir) is proportional to the CO (J=1-0) luminosity

123: \citep[$L'$;][]{gao04a} to the $\sim$1.5 power, consistent with

124: observed surface density SFR relations \citep{ken98a}. However, recent

125: observations have revealed a tight, linear relation between HCN

126: (J=1-0) and the \lir \ in galaxies \citep{gao04a,gao04b}. Because the

127: J=1-0 rotational transition of HCN has a relatively high critical

128: density ($n_{\rm crit} \sim$10$^{5}$ \cmthree), this has been

129: interpreted as a more fundamental SFR relation such that the

130: volumetric SFR is linearly related to the dense, star-forming

131: molecular gas. A linear relation between \lir \ and the luminosity

132: from a high critical density tracer CO (J=3-2) in a similar sample of

133: galaxies has provided confirming evidence for this relation and

134: advocated a similar interpretation \citep{nar05}.

135: 

136: An alternative interpretation to the linear relation between \lir

137: \ and dense gas tracers has been put forth from the theoretical

138: side. First, \citet{kru07} utilized models of individual giant

139: molecular clouds (GMCs) with a lognormal density distribution function

140: coupled with escape probability radiative transfer calculations. These

141: authors found that the central issue driving the observed molecular

142: SFR relations was the relationship between the fraction of the cloud's

143: gas above the critical density of the molecular transition. When, on

144: average, the gas density is higher than the line's critical density

145: (<$n_{\rm cloud}$> >> \ncrit), the bulk of the gas is thermalized in

146: the line, and the line luminosity increases linearly with increasing

147: mean gas density (<$n$>). In this scenario, a relationship between SFR

148: and \lmol \ is expected to have index similar to the underlying

149: (volumetric) Schmidt index. This is reminiscent of the observed

150: relation between \lir \ and CO (J=1-0) in galaxies. Alternatively,

151: when only a small fraction of the gas is thermalized (<$n_{\rm

152:   cloud}$> << \ncrit), the line luminosity will increase superlinearly

153: with <$n$>, and the SFR-\lmol \ relation will have index

154: less than that of the underlying Schmidt index (e.g. the observed

155: \lir-HCN J=1-0 relation).

156: 

157: \citet{nar08b} furthered these models of GMCs by applying 3D non-LTE

158: radiative transfer calculations to hydrodynamic simulations of

159: isolated galaxies and equal mass binary galaxy mergers.  A key finding

160: in these numerical models was that the observed relations are only

161: found when global measurements are made. Higher spatial resolution

162: observations of e.g. the nuclei of galaxies would probe gas in which a

163: larger fraction of the gas is thermalized than in unresolved

164: observations of the entire galaxy. In this case, <$n$>

165: could potentially become comparable to the critical density of the

166: line, and an SFR-\lmol \ index of $\sim$1.5 would be expected even for

167: high critical density molecular lines. In both sets of models, the

168: fundamental relation is the Schmidt relation with index 1.5. The

169: observed SFR-\lmol \ relations were simply manifestations of the

170: underlying Schmidt law.

171: 

172:  A key difference exists between the numerical models of \citet{kru07}

173:  and \citet{nar08b}, and the interpretations of

174:  \citet[][]{gao04a,gao04b}, \citet{nar05} and \citet{wu05}. The models

175:  find the driving relation is the volumetric Schmidt relation whereas

176:  the latter set of observations cite a more fundamental relation as

177:  that between SFR and dense molecular gas. {\it Tests that distinguish

178:    between these interpretations are crucial to understanding global

179:    SFR relations in galaxies}.  \citet{nar08b} offered a direct

180:  prediction from their models that observations of molecular lines

181:  with critical density higher than that of HCN (J=1-0) should break

182:  the linear trend seen between SFR and HCN (J=1-0). Specifically, for

183:  extremely high critical density lines, \ncrit will be even larger

184:  than <$n_{\rm galaxy}$> than in the case of HCN (J=1-0); In these

185:  cases, the relation between SFR and \lmol \ is directly predicted to

186:  be sublinear. Indeed, confirming evidence for this trend has been

187:  found by Bussmann et al. (2008; submitted), who found a sublinear

188:  \lir-HCN (J=3-2) index in remarkable agreement with the predictions

189:  of \citet{nar08b}.

190: 

191: An alternative generic feature of both the \citet{kru07} and

192: \citet{nar08b} simulations which may serve as a test of the models is

193: a superlinear SFR-\lmol \ relation for high critical density tracers

194: when <$n_{\rm galaxy}$> is sufficiently high. One

195: potential manifestation of this is a break in the linear SFR-HCN

196: (J=1-0) relation in systems with extremely high infrared luminosity

197: (e.g. hyper-luminous infrared galaxies; \lir >

198: 10$^{13}$\lsunend). Here, the models predict a steepening in the

199: SFR-\lmol \ index toward $N$=1.5 \citep{kru07}. Tentative evidence for

200: this may have been found by \citet{gao07} in high redshift systems,

201: though the potential contribution of active galactic nuclei (AGN) to

202: the \lir \ may drive a similar signature.

203: 

204: In order to avoid the muddying effects of central AGN, a potential

205: alternative to this test is to observe the SFR-\lmol \ relation in

206: galactic nuclei, where the local mean gas density may be higher than

207: the globally averaged mean gas density. In this {\it Letter}, we

208: utilize literature data in order to examine the relationship between

209: SFR (traced by the 850$\mu$m flux) and dense molecular gas (traced by

210: \lcojthree). The aim is to help distinguish between the competing

211: interpretations of the linear SFR-dense gas relations in galaxies.

212: 

213: \section{Literature Data}

214: \label{sec:obs}

215: 

216: %\begin{deluxetable}{ccc}

217: %\tabletypesize{\scriptsize}

218: %\tablecaption{Sample Galaxies\label{table:galaxies}}

219: %\tablewidth{0pt}

220: %\tablehead{

221: %\colhead{Galaxy} & \colhead{$F_{\rm 25 \mu \rm m}$} & 

222: %\colhead{log$_{\rm 10}$ ($L_{\rm CO J=3-2}$}) \\

223: %& Jy & K-km s$^{-1}$\\

224: %} %close tablehead

225: %\startdata

226: %UGC 5376	 	&0.62		&4.06	\\   

227: %NGC 3110		&1.13		&5.23	\\   

228: %IRAS 1017+08     	&0.55		&5.39	\\   

229: %NGC 3221		&0.93		&4.69	\\   

230: %NGC 3367		&1.98		&4.28	\\   

231: %IRAS 1056+61	        &1.27		&5.81	\\  

232: %Arp 148		        &0.71		&4.98	\\   

233: %NGC 3583		&0.77		&4.17	\\   

234: %NGC 3994/5      	&0.90		&4.24	\\   

235: %NGC 4045		&1.02		&4.01	\\   

236: %IRAS 1211+03	        &0.66		&5.89	\\   

237: %NGC 4273		&1.65		&4.33	\\   

238: %IRAS 1222-06    	&2.28		&5.07	\\   

239: %NGC 4418		&9.67		&4.59	\\  

240: %NGC 4433		&1.52		&4.80	\\   

241: %NGC 4793		&1.51		&4.45	\\   

242: %NGC 5020		&0.72		&4.60	\\   

243: %UGC 8387		&1.42		&5.72	\\  

244: %NGC 5104		&0.74		&5.25	\\   

245: %NGC 5256		&1.07		&5.16	\\   

246: %NGC 5257/8      	&1.34		&5.15	\\   

247: %UGC 8739		&0.42		&5.32	\\   

248: %NGC 5371		&0.97		&4.25	\\   

249: %NGC 5394	        &1.40		&4.87	\\   

250: %NGC 5433		&0.85		&5.12	\\   

251: %NGC 5600		&0.72		&3.83	\\   

252: %NGC 5653		&1.37		&4.92	\\   

253: %NGC 5665		&0.82		&4.16	\\   

254: %NGC 5676		&1.70		&4.06	\\   

255: %NGC 5713		&2.84		&4.24	\\   

256: %NGC 5792		&1.00		&4.31	\\   

257: %NGC 5900		&0.70		&4.57	\\   

258: %I Zw 107		&1.42		&5.53	\\   

259: %NGC 5930		&1.60		&4.16	\\   

260: %IRAS 1525+36     	&1.31		&5.64	\\   

261: %NGC 5936		&1.47		&5.02	\\   

262: %NGC 5937		&1.47		&4.64	\\   

263: %NGC 5953		&1.58		&4.42	\\   

264: %NGC 5962		&1.04		&4.26	\\   

265: %NGC 5990		&1.60		&4.85	\\   

266: %\enddata

267: %\end{deluxetable}

268: 

269: 

270: 

271: In order to investigate the relation between SFR and dense molecular

272: gas over a variety of physical spatial extents in galaxies, we require

273: a beam matched set of observations in both high critical density

274: molecular line and SFR tracer with a sufficient number of galaxies.

275: While matching the highest resolution and sensitivity infrared data of

276: nearby galaxies \citep[e.g SINGS; ][]{ken03} with high spatial

277: resolution observations of a high critical density tracer

278: \citep[e.g. ][]{kri07} would be ideal, few galaxies exist at the

279: intersection of such surveys. Most other traditional infrared surveys

280: are insufficient as they typically report global measurements, rather

281: than higher spatial resolution data. Further complications arise on

282: the dense gas tracer side. Because of the prodigious observing time

283: necessary to map large numbers of galaxies with an interferometer,

284: most dense gas surveys are done with a single dish and report on lines

285: at $\sim$mm wavelengths \citep[e.g.][]{baa08}. In order to obtain the

286: highest spatial resolution possible from single dish surveys,

287: higher-frequency observations (e.g. $\lambda$<1 mm) must be employed.

288: 

289: The largest beam-matched sample of a dense gas tracer and \lir

290: \ tracer is the JCMT CO (J=3-2) survey of \citet{yao03} and the SCUBA

291: Local Universe Galaxy Survey \citep{dun00}. \citet{yao03} used the

292: 850$\mu$m images of these galaxies to scale the total \lir \ to match

293: the $\sim$15$\arcsec$ beam of the CO (J=3-2) observations.  As such,

294: while the number statistics are relatively small, to our knowledge

295: this sample comprises the largest available for this type of analysis.

296: Moreover, using CO (J=3-2) as a tracer of dense molecular gas has the

297: attractive quality that both global measurements \citep{nar05} and

298: simulated unresolved observations from simulations \citep{nar08b} have

299: shown a linear relation between \lir \ and CO (J=3-2) luminosity in

300: galaxies.

301: 

302: In order to avoid single galaxies at the extrema of the \lir \ range

303: artificially biasing the fits, we impose nominal luminosity cuts in

304: the sample, considering galaxies in the range \lir=[10$^9$,

305:   2$\times$10$^{11}$ \lsunend]. This excludes galaxies which may be

306: interacting, thus allowing us to use galaxy distance as a proxy for

307: mean density (\S~\ref{sec:results}). This results in a total sample

308: size of 40 galaxies, excluding only 4 galaxies from the parent sample.

309: 

310: 

311: 

312: \section{Results}

313: \label{sec:results}

314: 

315: %\begin{figure}

316: %\includegraphics[scale=0.375,angle=90]{f1.ps}

317: %\caption{Histogram of number of galaxies within given distances. The

318: %  number of galaxies rises sharply near distances of $\sim$50 Mpc. We

319: %  therefore begin our analysis at this distance bin to maximize the

320: %  statistical significance of the investigated

321: %  sample.\label{figure:disthist}}

322: %\end{figure}

323: 

324: \begin{figure}

325: \includegraphics[scale=0.5]{f1.ps}

326: \caption{The \lir-\lcojthree \ relation for galaxies in each of our

327:   distance bins. The slope is seen to flatten as observations probe

328:   more global measurements.\label{figure:lirvlco32}}

329: \end{figure}

330: 

331: 

332: 

333: \begin{figure}

334: \includegraphics[scale=0.375,angle=90]{f2.ps}

335: \caption{The \lir-CO (J=3-2) index as a function of limiting galaxy

336:   distance. The physical scale covered by the beam is on the top

337:   axis. Observations of the nearby galaxies probe the nuclear region

338:   of the galaxies. In this high <$n$> regime, the \lir-CO (J=3-2)

339:   index tentatively lies near the volumetric Schmidt index, also

340:   consistent with recent modeling efforts by \citet{kru07} and

341:   \citet{nar08b}. As observations become unresolved and more low

342:   density gas is folded into the beam, the index tends toward

343:   unity. \label{figure:lirvlco32_index}}

344: \end{figure}

345: 

346: 

347: 

348: We have binned our sample of galaxies into bins in distance. Because

349: we are considering observations of a fixed beamsize, this is

350: equivalent to binning the galaxies in terms of physical radial extent

351: from the nucleus probed by the observations. 

352: %In Figure~\ref{figure:disthist}, we show the distribution of galaxies

353: %within given distance bins. 

354: The number of galaxies rises sharply to $\sim$25 when considering

355: galaxies with a maximum distance of $\sim$50 Mpc. We thus utilize 50

356: Mpc as the first distance bin in order to maximize the statistical

357: significance of the result.

358: 

359: In Figure~\ref{figure:lirvlco32}, we show the \lir-\lcojthree

360: \ relation for each distance bin. In

361: Figure~\ref{figure:lirvlco32_index}, we plot the best fit log(\lir)

362: -log(\lcojthree) \citep[L';][]{gao04a} slope as a function of limiting

363: distance.  On the top axis we label the physical extent of the

364: beamsize.%\footnote{We are considering the

365: %  molecular line 'luminosity' which is the typical method for

366: %  reporting the velocity integrated molecular line

367: %  intensity:\\ $L'\approx \pi/$(4 ln 2)$\theta_{\rm mb}^2 I_{\rm mol}

368: %  d_L^2$(1+\z)$^{-3}$ where $\theta_{\rm mb}$ is the main beam angular

369: %  size, $I_{\rm mol}$ is the molecular line intensity, $d_L$ is the

370: %  luminosity distance, and \z \ is the redshift \citep{gao04a}.}

371: The fits were done utilizing the publically available Monte Carlo

372: Markov Chain fitting packages of \citet{kel07}. The Bayesian routines

373: assume the intrinsic distribution can be well approximated via a

374: mixture of Gaussians, and returns the posterior probability

375: distribution function (pdf) of potential slopes to the fit. Unlike

376: standard linear regression techniques, the fit is not done with the

377: assumption that the abscissa values are known exactly while

378: measurement errors exist only for the ordinate. Rather, errors are

379: allowed in both axes. This is important as errors are only reported

380: for the literature CO measurements \citep{yao03}, and to conform with

381: standard literature fits, we are required to fit \lir \ as a function

382: of the CO luminosity. The points in

383: Figure~\ref{figure:lirvlco32_index} are the median of the returned

384: distribution of slopes, and the error bars denote the standard

385: deviation in the pdf.

386: 

387: The slopes in the \lir-\lcojthree \ relation tentatively show a

388: similar trend to what would be expected from the models of

389: \citet{kru07} and \citet{nar08b}. Specifically, in the regions of high

390: <$n$> where the bulk of the gas may be thermalized, the SFR-\lmol

391: \ relation for high critical density tracers tends toward 1.5

392: (e.g. the points at D$\approx$50 Mpc in

393: Figure~\ref{figure:lirvlco32_index}; this corresponds to the central

394: $\sim$1.7 kpc for most of the galaxies in this bin). When considering

395: more unresolved, global observations of galaxies which fold in

396: significant amounts of diffuse gas (thus lowering the effective <$n$>

397: probed by the observations), the mean density of the galaxy drops

398: below the critical density, and the observed SFR-\lmol \ index is less

399: than that of the underlying volumetric Schmidt index. Here, we see

400: that the more unresolved observations approach a slope of unity,

401: consistent with the linear slope found between \lir \ and CO (J=3-2)

402: luminosity in a nearly identical sample of galaxies by

403: \citet{nar05}. We note that the slope is still moderately superlinear

404: in these bins, though consistent with the range of results expected by

405: the \citet{nar08b} models. This occurs because the maximum distance

406: bins still contain the relatively nearby (D<50 Mpc) galaxies, which

407: increases the \lir-\lcojthree \ index. Ideally one would like to

408: exclude the nearby galaxies from the most distant bins, but low number

409: statistics prevent this experiment.

410: 

411: 

412: \section{Discussion}

413: \label{sec:discussion}

414: The trends seen in Figure~\ref{figure:lirvlco32_index} may be

415: consistent with a picture in which the observed SFR-\lmol \ relations

416: are driven globally by the relationship between <$n_{\rm gas}$> and

417: \ncrit. In contrast, if the linear relation between \lir \ and HCN

418: (J=1-0) or CO (J=3-2) were indicative of a more fundamental SFR

419: relation in galaxies in terms of dense molecular gas, one would expect

420: the relation to remain linear even in the nuclei of galaxies. With a

421: potentially model distinguishing result such as this one, obvious

422: questions regarding its robustness arise.

423: 

424: First, we caution that the error bars presented in the lowest distance

425: bins in Figure~\ref{figure:lirvlco32_index} are rather large. This

426: owes to the small sample sizes of these bins ($\sim$25 galaxies for

427: the lowest distance bins). It is possible to investigate what the

428: probability is that the same \lir-\lcojthree \ index properly

429: characterizes both the lowest and highest distance bins. To do this,

430: we remind the reader that each point (and associated error bars) in

431: Figure~\ref{figure:lirvlco32_index} is a pdf for potential

432: \lir-\lcojthree \ indices at each distance bin. In

433: Figure~\ref{figure:cumdist}, we plot the (normalized) difference in

434: the pdf's from the highest and lowest distance bins in

435: Figure~\ref{figure:lirvlco32_index}. We additionally show the

436: cumulative distribution function in the same plot. A value of 0 in the

437: pdf difference is expected at points when the same index characterizes

438: both the high and low distance bins. As can be seen, the nuclear

439: \lir-\lcojthree \ index is systematically weighted toward larger

440: numbers than the global value. That said, the error bars in the

441: nuclear distance bin are not insignificant. The probability that the

442: \lir-\lcojthree \ indices from the nuclear and global distance bins

443: are the same $\pm$0.25(0.5) is $\sim$17(33)\%. Thus, while the

444: tentative trends seen in Figure~\ref{figure:lirvlco32_index} are

445: indeed probable, they are by no means robust. In this sense, surveys

446: to increase the number statistics will be required to confirm/refute

447: this potential result.

448: 

449: 

450:  

451: %While the data for the lowest distance bins in

452: %Figure~\ref{figure:lirvlco32_index} have considerable uncertainty, it

453: %is possible to rule out a scenario in which each distance bins can be

454: %characterized with the same \lir-\lcojthree \ slope. If we consider

455: %each point in Figure~\ref{figure:lirvlco32_index} as a pdf, then the

456: %product of the individual distributions will give the probability that

457: %all distance bins have a given \lir-\lcojthree \ index. In

458: %Figure~\ref{figure:probdist}, we show this product of probabilities as

459: %a function of \lir-\lcojthree \ index. For reference, the most

460: %probable index for each distance bin in

461: %Figure~\ref{figure:lirvlco32_index} (the solid points) typically had a

462: %probability of $\sim$10\%. As Figure~\ref{figure:probdist} shows, the

463: %probability for any single \lir-\lcojthree \ index satisfying the data

464: %points for each distance bin is negligibly small. This is a relatively

465: %powerful statement as the standard interpretation of the linear

466: %SFR-dense gas relation in galaxies would suggest that all distance

467: %bins in Figure~\ref{figure:lirvlco32_index} would have the same

468: %(linear) index. That said, the considerable error bars do not preclude

469: %the alternative that the distance bins have different indices than

470: %those in Figure~\ref{figure:lirvlco32_index}, without all having the

471: %same index.

472: 

473: Second, we can inquire as to the validity of \lir \ as an SFR

474: indicator. In particular, the \lir \ from galaxies has the potential

475: to be contaminated by central AGN. That said, the galaxies in the

476: sample presented here are relatively low luminosity (global \lir \ <

477: 10$^{12}$ \lsunend). Both observational studies \citep{tra01} and

478: theoretical works \citep{cha07a} suggest that galaxies with the \lir \

479: range investigated here have their IR luminosity largely powered by

480: star formation.

481: 

482: %If the galaxies did indeed contain AGN, it is feasible that targeted

483: %observations toward the nuclear regions would have a larger fraction

484: %of the \lir \ powered by a central AGN than global observations of the

485: %same galaxies. However, the lowest distance bin included in

486: %Figure~\ref{figure:lirlco_index} is nearly 2 kpc in radius, and thus

487: %likely contains 

488: 

489: Third, we can question whether or not the trend in the \lir-CO (J=3-2)

490: index seen with distance is equivalent to a trend in <$n$>. To do

491: this, in Figure~\ref{figure:meandens}, we plot the normalized <$n$> of

492: the galaxy versus the galaxy distance.  For the density determination,

493: the volume is derived by assuming the gas resides in a disk with

494: radius equivalent to the physical extent probed by the beam at that

495: distance, and a constant scale height in each galaxy. The molecular

496: gas mass is taken from the CO (J=1-0) luminosity, utilizing a CO-\htwo

497: \ conversion factor appropriate for this sample of galaxies

498: \citep{yao03}.  The CO (J=1-0) observations are beam matched to the CO

499: (J=3-2) observations presented here, so the molecular gas mass is

500: indeed the gas mass within the radial extent of the CO (J=3-2)

501: beam. The mean density is normalized to account for a variety of

502: uncertain parameters (e.g. disk scale height, molecular gas volume

503: filling factor, density profile of GMCs). Because we are only

504: concerned with the trend in <$n_{\rm galaxy}$> with distance,

505: normalizing the density has no

506: consequence. Figure~\ref{figure:meandens} shows that the mean density

507: probed by the molecular line observations drops as we consider

508: galaxies successively farther away. Alternatively said, unresolved

509: observations of galaxies probe lower <$n$> than observations toward

510: galactic nuclei. This implies that the tentative trend in \lir-CO

511: (J=3-2) index seen in Figure~\ref{figure:lirvlco32_index} owes to the

512: relationship between the mean density of the gas in the beam and

513: \ncrit \ of the emission line.

514: 

515: If larger samples verify the results of

516: Figure~\ref{figure:lirvlco32_index}, then these observed trends may

517: have several implications. The models of \citet{kru07} and

518: \citet{nar08b} which predict an SFR-\lmol \ index of 1.5 when <$n$> is

519: high are predicated on an underlying relation controlling the SFR: SFR

520: $\propto$ $n^{1.5}$. The results presented here tentatively support a

521: scenario in which the SFR can be described by such a power-law (rather

522: than a linear relation), though we reiterate that the error bars in

523: this study are substantial.

524: 

525: \begin{figure}

526: \includegraphics[angle=90,scale=0.375]{f3.ps}

527: \caption{Solid Line: Normalized difference in PDFs of \lir-\lcojthree

528:   \ indices for lowest distance bin and largest distance bin. Dotted

529:   line: Cumulative distribution function for shown PDF. This can be

530:   interpreted as the PDF of the \lir-\lcojthree \ indices for the

531:   nuclear and global observations as being the

532:   same.  \label{figure:cumdist}}

533: \end{figure}

534: 

535: 

536: \begin{figure}

537: \includegraphics[angle=90,scale=0.375]{f4.ps}

538: \caption{The normalized <$n$> within the CO (J=3-2) beam as a function

539:   of galaxy distance. The <$n$> within the beam drops as a function of

540:   distance, showing that the more unresolved observations are probing

541:   lower <$n$> gas. This implies that the relationship seen in

542:   Figure~\ref{figure:lirvlco32_index} owes to a dropping <$n$> in the

543:   farthest objects. \label{figure:meandens}}

544: \end{figure}

545: 

546: %The results here additionally provide further support for the

547: %interpretation of Bussmann et al. (2008; submitted) who found a

548: %sublinear \lir-HCN (J=3-2) relation with a slope of $\sim$0.75 in

549: %unresolved observations of nearby galaxies. This result was predicted

550: %by \citet{nar08b}, and interpreted as a direct result of a scenario in

551: %which <$n_{\rm galaxy}$> << \ncrit, and the SFR was controlled by the

552: %volumetric Schmidt relation. In contrast, the standard interpretation

553: %of the linear SFR-HCN (J=1-0) relation would suggest that even higher

554: %critical density tracers than HCN (J=1-0) (e.g. HCN J=3-2) would be

555: %similarly linear with \lir. The results of Bussmann et al. (2008)

556: %along with the relations presented here are suggestive of a scenario

557: %in which the linear \lir-HCN (J=1-0) and CO (J=3-2) relations in

558: %unresolved observations of galaxies are simply a consequence of a

559: %particular average <$n_{\rm galaxy}$>-\ncrit \ combination.

560: 

561: 

562: In principle, we could perform the same experiment shown in

563: Figure~\ref{figure:lirvlco32_index} with the CO (J=1-0)

564: transition. Indeed, Nobeyama 45m observations of CO (J=1-0) in many of

565: the same galaxies exist, thus providing the opportunity for a

566: beam-matched comparison to this dataset \citep{yao03}. However, the

567: error bars on these observations are sufficiently large that the

568: slopes on the \lir-\lcojone \ indices for the lowest distance bins (D

569: < 100 Mpc) are consistent with numerous interpretations (including no

570: correlation at all).

571: 

572: The lack of clean data in low \ncrit \ transitions does not preclude

573: us from conjecturing what results one might expect in a similar

574: experiment with CO (J=1-0). Naively, one might imagine the \lir-CO

575: (J=1-0) index to be $\sim$1.5 for all distance bins owing to the

576: relatively low critical density of this ground state

577: transition. However, the more likely scenario are results similar to

578: the trend seen in Figure~\ref{figure:lirvlco32_index} - a slope of

579: $\sim$1.5 for the lowest distance bins, and a flattening to a slope of

580: $\sim$1. The reason for this is global observations of galaxies in

581: this relatively low \lir \ range have a linear \lir-CO (J=1-0)

582: relation \citep{gao04a,gao04b}. It is only when considering galaxies

583: with \lir $\ga$10$^{11}$ \lsun in the sample that unresolved

584: observations result in an \lir-\lcojone \ index of 1.5. In the

585: interpretation of \citet{kru07} and \citet{nar08b}, this would result

586: from lower luminosity galaxies having most of their gas subthermal in

587: this transition.

588: 

589: Finally, we emphasize that though the results presented here advocate

590: an interpretation of the linear \lir-dense gas relations consistent

591: with \citet{kru07} and \citet{nar08b}, this is not to say that some

592: high critical density tracers such as HCN (J=1-0) are not applicable

593: as SFR tracers. By virtue of their linearity with \lir, insomuch that

594: the \lir \ can be used as an adequate proxy for the SFR, so can global

595: measurements of HCN (J=1-0) and CO (J=3-2).

596: 

597: \section{Conclusions and Ways Forward}

598: \label{sec:conclusions}

599: 

600: We have studied the relationship between the SFR and dense molecular

601: content in the nuclei of galaxies by utilizing the observed \lir \ as

602: a proxy for the SFR, and CO (J=3-2) as a proxy for dense molecular

603: gas. We find tentative evidence that the \lir-CO (J=3-2) index is

604: superlinear in galactic nuclei with value $N \sim$ 1.5, while

605: flattening as the observations measure more global conditions. 

606: 

607: The error bars in this potential trend are considerable however, and

608: that the results here serve as motivation for future observational

609: experiments investigating the \lir-dense gas relation in a variety of

610: environments. One potential way to increase the sample size would be

611: to obtain additional beam-matched sub-mm and CO (J=3-2) observations

612: of the nuclei of nearby galaxies with 15 m sub-mm telescopes (e.g. the

613: JCMT or ASTE). Alternatively, high spatial resolution observations of

614: nearby galaxies with ancillary IR maps (e.g. the SINGS sample), would

615: provide a rich dataset with \lir \ and dense gas measurements for

616: numerous regions with varying physical conditions in individual

617: galaxies. Additional data to constrain the gas density (e.g. CO J=1-0)

618: would allow the \lir-\lcojthree\ index to be studied truly as a

619: function of gas density, as opposed to galaxy distance. These

620: experiments will help to confirm/refute the potential results of

621: Figure~\ref{figure:lirvlco32_index}.

622: 

623: 

624: \acknowledgements We thank Lihong Yao and Brandon Kelly for helpful

625: conversations. Support for this work was provided in part by the Keck

626: Foundation.

627: 

628: 

629: 

630: 

631: 

632: \begin{thebibliography}

633: 

634: 

635: \bibitem[{{Baan} {et~al.}(2008){Baan}, {Henkel}, {Loenen}, {Baudry}, \&

636:   {Wiklind}}]{baa08}

637: {Baan}, W.~A., {Henkel}, C., {Loenen}, A.~F., {Baudry}, A., \& {Wiklind}, T.

638:   2008, \aap, 477, 747

639: 

640: \bibitem[{{Chakrabarti} {et~al.}(2007){Chakrabarti}, {Cox}, {Hernquist},

641:   {Hopkins}, {Robertson}, \& {Di Matteo}}]{cha07a}

642: {Chakrabarti}, S., {Cox}, T.~J., {Hernquist}, L., {Hopkins}, P.~F.,

643:   {Robertson}, B., \& {Di Matteo}, T. 2007, \apj, 658, 840

644: 

645: \bibitem[{{Dunne} {et~al.}(2000){Dunne}, {Eales}, {Edmunds}, {Ivison},

646:   {Alexander}, \& {Clements}}]{dun00}

647: {Dunne}, L., {Eales}, S., {Edmunds}, M., {Ivison}, R., {Alexander}, P., \&

648:   {Clements}, D.~L. 2000, \mnras, 315, 115

649: 

650: \bibitem[{{Gao} {et~al.}(2007){Gao}, {Carilli}, {Solomon}, \& {Vanden

651:   Bout}}]{gao07}

652: {Gao}, Y., {Carilli}, C.~L., {Solomon}, P.~M., \& {Vanden Bout}, P.~A. 2007,

653:   \apjl, 660, L93

654: 

655: \bibitem[{{Gao} \& {Solomon}(2004{\natexlab{a}})}]{gao04a}

656: {Gao}, Y. \& {Solomon}, P.~M. 2004{\natexlab{a}}, \apjs, 152, 63

657: 

658: \bibitem[{{Gao} \& {Solomon}(2004{\natexlab{b}})}]{gao04b}

659: ---. 2004{\natexlab{b}}, \apj, 606, 271

660: 

661: \bibitem[{{Kelly}(2007)}]{kel07}

662: {Kelly}, B.~C. 2007, \apj, 665, 1489

663: 

664: \bibitem[{{Kennicutt}(1998)}]{ken98a}

665: {Kennicutt}, Jr., R.~C. 1998, \araa, 36, 189

666: 

667: \bibitem[{{Kennicutt} \& {et}(2003)}]{ken03}

668: {Kennicutt}, Jr., R.~C. \& {et}, a. 2003, \pasp, 115, 928

669: 

670: \bibitem[{{Krips} {et~al.}(2007){Krips}, {Neri}, {Garcia-Burillo}, {Martin},

671:   {Combes}, {Gracia-Carpio}, \& {Eckart}}]{kri07}

672: {Krips}, M., {Neri}, R., {Garcia-Burillo}, S., {Martin}, S., {Combes}, F.,

673:   {Gracia-Carpio}, J., \& {Eckart}, A. 2007, ArXiv e-prints, 712

674: 

675: \bibitem[{{Krumholz} \& {Thompson}(2007)}]{kru07}

676: {Krumholz}, M.~R. \& {Thompson}, T.~A. 2007, \apj, 669, 289

677: 

678: \bibitem[{{Narayanan} {et~al.}(2007){Narayanan}, {Cox}, {Shirley}, {Dave},

679:   {Hernquist}, \& {Walker}}]{nar08b}

680: {Narayanan}, D., {Cox}, T.~J., {Shirley}, Y., {Dave}, R., {Hernquist}, L., \&

681:   {Walker}, C.~K. 2007, ArXiv e-prints, 711

682: 

683: \bibitem[{{Narayanan} {et~al.}(2005){Narayanan}, {Groppi}, {Kulesa}, \&

684:   {Walker}}]{nar05}

685: {Narayanan}, D., {Groppi}, C.~E., {Kulesa}, C.~A., \& {Walker}, C.~K. 2005,

686:   \apj, 630, 269

687: 

688: \bibitem[{{Schmidt}(1959)}]{sch59}

689: {Schmidt}, M. 1959, \apj, 129, 243

690: 

691: \bibitem[{{Tran} \& {et}(2001)}]{tra01}

692: {Tran}, Q.~D. \& {et}, a. 2001, \apj, 552, 527

693: 

694: \bibitem[{{Wu} {et~al.}(2005){Wu}, {Evans}, {Gao}, {Solomon}, {Shirley}, \&

695:   {Vanden Bout}}]{wu05}

696: {Wu}, J., {Evans}, II, N.~J., {Gao}, Y., {Solomon}, P.~M., {Shirley}, Y.~L., \&

697:   {Vanden Bout}, P.~A. 2005, \apjl, 635, L173

698: 

699: \bibitem[{{Yao} {et~al.}(2003){Yao}, {Seaquist}, {Kuno}, \& {Dunne}}]{yao03}

700: {Yao}, L., {Seaquist}, E.~R., {Kuno}, N., \& {Dunne}, L. 2003, \apj, 588, 771

701: 

702: \end{thebibliography}

703: 

704: 

705: \end{document}

706: