0707.0395/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: 
3: \shorttitle{Warm H$_2$ in SINGS galaxies}
4: \shortauthors{Roussel et al.}
5: 
6: \begin{document}
7: 
8: \title{Warm molecular hydrogen in the Spitzer SINGS galaxy sample}
9: 
10: \author{H. Roussel\altaffilmark{1},
11: G. Helou\altaffilmark{2},
12: D.J. Hollenbach\altaffilmark{3},
13: B.T. Draine\altaffilmark{4},
14: J.D. Smith\altaffilmark{5},
15: L. Armus\altaffilmark{2},
16: E. Schinnerer\altaffilmark{1},
17: F. Walter\altaffilmark{1},
18: C.W. Engelbracht\altaffilmark{5},
19: M.D. Thornley\altaffilmark{6},
20: R.C. Kennicutt\altaffilmark{7,5},
21: D. Calzetti\altaffilmark{8},
22: D.A. Dale\altaffilmark{9},
23: E.J. Murphy\altaffilmark{10},
24: C. Bot\altaffilmark{2}
25: }
26: 
27: \email{roussel@mpia-hd.mpg.de}
28: 
29: \altaffiltext{1}{Max-Planck-Institut f\"ur Astronomie, Heidelberg, 69117 Germany}
30: \altaffiltext{2}{California Institute of Technology, Pasadena, CA 91125}
31: \altaffiltext{3}{NASA Ames Research Center, Moffett Field, CA 94035}
32: \altaffiltext{4}{Princeton University, Princeton, NJ 08544}
33: \altaffiltext{5}{Steward Observatory, University of Arizona, Tucson, AZ 85721}
34: \altaffiltext{6}{Bucknell University, Lewisburg, PA 17837}
35: \altaffiltext{7}{IoA, University of Cambridge, UK}
36: \altaffiltext{8}{University of Massachusetts, Amherst, MA 01003}
37: \altaffiltext{9}{University of Wyoming, Laramie, WY 82071}
38: \altaffiltext{10}{Yale University, New Haven, CT 06520}
39: 
40: 
41: \begin{abstract}
42: Results on the properties of warm molecular hydrogen in 57
43: normal galaxies are derived from measurements of H$_2$ rotational transitions
44: in the mid-infrared, obtained as part of the Spitzer Infrared Nearby Galaxies Survey
45: (SINGS). This study extends previous extragalactic surveys of emission lines of H$_2$,
46: the most abundant constituent of the molecular interstellar medium, to fainter
47: and more common systems ($L_{\rm FIR} = 10^7$ to $6 \times 10^{10}$\,L$_{\sun}$)
48: of all morphological and nuclear types.
49: In our sensitive integral-field observations covering kiloparsec-scale areas,
50: the 17\,$\mu$m S(1) transition is securely detected in the nuclear regions of
51: 86\% of SINGS galaxies with stellar masses above $10^{9.5}$\,M$_{\sun}$.
52: The derived column densities of warm H$_2$ (with temperatures above $\sim 100$\,K),
53: even though averaged over large areas, are commensurate with values observed in
54: resolved photodissociation regions; the median of the sample is
55: $3 \times 10^{20}$\,cm$^{-2}$. They amount to a significant fraction
56: of the column densities of total molecular hydrogen, between 1\% and more than 30\%.
57: The power emitted in the sum of the three lowest-energy transitions is on average
58: 30\% of the power emitted in the bright [Si{\small II}] cooling line (34.8\,$\mu$m),
59: and represents about $4 \times 10^{-4}$ of the total infrared power within the same
60: area for star-forming galaxies, which is consistent with excitation in
61: photodissociation regions. The fact that the H$_2$ line intensities scale tightly
62: with the emission in the aromatic bands, even though the average radiation field
63: intensity within the same area varies by a factor ten, can also be understood
64: if both tracers originate predominantly in photodissociation regions, either
65: dense or diffuse. A large fraction of the 25 targets classified as {\small LINER}s
66: or Seyferts, however, strongly depart from the rest of the sample, in having warmer
67: H$_2$ in the excited states, smaller mass fractions of H$_2$ in the warm phase,
68: and an excess of power emitted in the rotational transitions with respect to
69: aromatic bands, the total infrared emission and the [Si{\small II}] line.
70: We propose a threshold in H$_2$ to aromatic band power ratios, allowing the
71: identification of low-luminosity AGNs by an excess H$_2$ excitation.
72: A dominant contribution from shock heating is favored in these objects.
73: Finally, we detect in nearly half the star-forming targets, in particular
74: in low-density central regions of late-type galaxies, non-equilibrium ortho
75: to para ratios, consistent with the effects of pumping by far-ultraviolet
76: photons combined with incomplete ortho-para thermalization by collisions,
77: or possibly non-equilibrium photodissociation fronts advancing into cold gas.
78: \end{abstract}
79:  
80: \keywords{galaxies: ISM -- infrared: galaxies -- infrared: ISM -- ISM: lines and bands --
81: ISM: molecules -- surveys}
82: 
83: 
84: \section{Introduction}
85: 
86: Rotational transitions of molecular hydrogen, lying in the mid-infrared range
87: between 5 and 30\,$\mu$m (Table~\ref{tab_lines}), provide measurements of the mass
88: and temperature distribution of the bulk of the warm molecular gas phase,
89: at temperatures of $\sim 100$ to 1000\,K.
90: By contrast, the transitions from higher vibrational levels in the
91: near-infrared, better studied because more easily observable from the ground,
92: arise from gas at apparent excitation temperatures of
93: more than 1000\,K located in a thin layer of molecular clouds,
94: hosting a negligible fraction (of the order of $10^{-6}$) of the total H$_2$ mass
95: \citep{Black76, Burton92}. The rotational lines are thus more appropriate
96: tracers of molecular gas being exposed to moderate heating, and arise from a much
97: larger volume fraction of the molecular clouds. They constitute one of the
98: most important coolants of warm molecular gas \citep{Neufeld93}.
99: 
100: H$_2$ emission lines have been detected in a wide array of sources, including
101: outflows from young stars \citep{Gautier76, Bally82}, photodissociation regions
102: \citep{Gatley87, Tanaka89}, planetary nebulae \citep{Treffers76, Beckwith78},
103: supernova remnants \citep{Treffers79, Burton89}, large regions at the centers of galaxies
104: \citep{Thompson78, Gatley84} and extranuclear large-scale shocks in galaxy collisions
105: \citep{Herbst90, Sugai97, Appleton06}. The possible excitation mechanisms are
106: accordingly varied. In normal
107: galaxies, the major excitation source is expected to be the far-ultraviolet
108: radiation of massive stars in photodissociation regions, with photon energies between
109: 6 and 13.6\,eV \citep[and references therein]{Hollenbach97}. H$_2$ molecules can be
110: pumped by FUV photons into electronically-excited states, followed by fluorescence
111: and radiative cascade through the vibration-rotation levels of the ground electronic
112: state. Pure fluorescent spectra are produced only if the cascade is not significantly
113: altered by collisions with hydrogen atoms and molecules; if the critical densities
114: for collisional deexcitation are exceeded, a portion of the pump energy is converted
115: to heat by collisions, and the lowest rotational levels are populated by collisions
116: and thermalized. Pure fluorescence is thus much more likely in the vibrational transitions,
117: that have high critical densities, than in the pure rotational transitions considered
118: here, with critical densities below a few $10^3$\,cm$^{-3}$ for S(0) to S(3).
119: Additionally, FUV photons can be absorbed by dust grains, followed by the ejection
120: of photoelectrons that heat the gas. This also results in the thermal excitation
121: of the low-energy levels of H$_2$ by collisions with the warm gas. Besides the radiation
122: of massive stars, a second
123: important source of excitation is shocks, in molecular outflows, supernova remnants
124: or cloud collisions in a disturbed gravitational potential \citep{Shull78, Draine83}.
125: In addition to the above processes, X-rays produced in active nuclei or in supernova
126: remnant shocks can partially ionize and heat the gas over large column densities,
127: leading to H$_2$ excitation by collisions with hydrogen atoms and molecules, and
128: with fast electrons \citep{Lepp83, Draine91, Maloney96}.
129: Finally, H$_2$ molecules can be formed directly into excited states.
130: 
131: Surveys of molecular hydrogen line emission in galaxies have been so far mostly
132: restricted to starbursts, active galactic nuclei and ultraluminous systems,
133: and have been performed mostly in the near-infrared, targetting vibration-rotation
134: lines that arise from
135: upper levels with much higher excitation energies than the mid-infrared lines.
136: It has been speculated that the major source of H$_2$ heating in star-forming galactic
137: nuclei was shocks in supernova remnants, based on comparison of the luminosity of some
138: vibration-rotation H$_2$ lines with a limited number of Galactic templates and with
139: shock models \citep{Moorwood88, Mouri90}. However, scaling individual templates to
140: the integrated emission of galaxies has large inherent uncertainties,
141: and the near-infrared line ratios most often used to discriminate between thermal
142: and non-thermal emission are not always sufficient to distinguish between shocks and
143: fluorescent excitation followed by collisional deexcitation in high-density regions
144: \citep{Sternberg89}.
145: \citet{Puxley88} surveyed starburst galaxies in several vibration-rotation lines,
146: and found that the dominant excitation mechanism was pumping
147: by the far-ultraviolet radiation of massive stars, rather than collisional excitation.
148: \citet{Davies03} reached the same conclusion for a small sample of ultraluminous
149: galaxies, in which the first vibrational level is thermalized by high densities in
150: photodissociation regions.
151: Active nuclei ({\small LINER}s or Seyferts) can show an excess of H$_2$ emission relative
152: to hydrogen recombination lines and aromatic bands \citep[e.g.][]{Moorwood88, Larkin98},
153: but the exact nature of the additional source of excitation, namely X-ray excitation,
154: fluorescence induced by a non-thermal ultraviolet continuum, or shocks induced by
155: dynamical perturbations, is often unclear \citep[e.g.][]{Quillen99}. It is however
156: unlikely that significant H$_2$ emission could arise from interaction between
157: molecular clouds and jets from Seyfert nuclei \citep{Rotaciuc91, Knop01}.
158: 
159: The detection of a rotational line of H$_2$ was first reported by \citet{Beck79}
160: (the S(2) transition at 12.3\,$\mu$m in Orion) from observations at Las Campanas
161: Observatory. It was soon followed by many more ground-based detections, but the
162: majority of data on the rotational spectrum of H$_2$ were produced by the SWS
163: instrument on board ISO \citep[e.g.][]{Lutz00, Rigopoulou02}.
164: Furthermore, with previous infrared spectroscopic capabilities,
165: observations of normal galaxies have proven difficult due to sensitivity
166: limitations, so that our current knowledge is mainly extrapolated from studies of
167: very bright objects, maybe not representative of the general galaxy population.
168: The purpose of this paper is thus to extend previous work to fainter systems than
169: formerly accessible, and to characterize directly the generic properties of the warm
170: molecular hydrogen content of normal galaxies. The SINGS sample
171: \citep[Spitzer Infrared Nearby Galaxies Survey;][]{Kennicutt03}, covering a broad range
172: of infrared luminosities, morphologies and nuclear types, is ideally suited to such
173: a pursuit.
174: 
175: Studies of rotational lines alone, without information on vibrational levels, have
176: very limited diagnostic value concerning the source of excitation, because the
177: low critical densities of the rotational levels make it likely that they will be
178: thermalized most of the time, and thus cannot be used to distinguish between
179: the various heating mechanisms.
180: Because observations of vibration-rotation transitions in the near-infrared are
181: still scarce for normal galaxies, and because they are typically performed in
182: apertures that are not matched to our observations, we did not attempt to include
183: vibrational levels in our analysis. The characterization of excitation mechanisms
184: and physical conditions in the gas would greatly benefit from such
185: information, but would necessitate an additional dedicated survey.
186: 
187: The rotational lines are, however, energetically important and can characterize 
188: the temperature and density conditions of a large mass fraction of the
189: interstellar medium in galaxies, i.e. that consisting of warm molecular gas.
190: From a SWS survey of rotational lines in nearby starburst and Seyfert galaxies,
191: \citet{Valentijn96} and \citet{Rigopoulou02} obtained mass fractions of H$_2$ in
192: the warm phase of several percent. In ultraluminous galaxies observed with Spitzer,
193: \citet{Higdon06} derive much lower mass fractions of warm gas, but the fact that the
194: majority of their sample has only upper limits for the S(0) line makes it possible
195: that the temperatures are overestimated
196: (because computed from the S(1) to S(3) lines only, whenever S(0) is undetected)
197: and thus the masses of warm H$_2$ underestimated.
198: 
199: This paper presents observations of warm molecular hydrogen in nearby galaxies
200: obtained as part of SINGS \citep{Kennicutt03}. From these data, we present
201: quantifications of the temperatures and column densities of warm H$_2$ encountered
202: in kiloparsec-scale areas, mostly nuclear regions,
203: and a comparison of the power emitted in the rotational lines with those produced
204: by [Si{\small II}] at 34.8\,$\mu$m, which is the dominant cooling line
205: of normal galaxies in the mid-infrared range, and by dust.
206: We emphasize the different properties of star-forming regions and nuclei classified
207: as {\small LINER}s or Seyferts, and discuss their H$_2$ excitation mechanisms.
208: The data, analysis methods and observational results are described in Sections
209: \ref{data} to \ref{powers}.\footnote{For easier comparison to future observations
210: and models, ascii flux tables of all the measured quantities are available
211: upon e-mail request.}
212: The interpretation of the main findings is presented in Sections~\ref{excitation}
213: and \ref{syliner}, and a summary of the results and conclusions can be found
214: in Section~\ref{summary}.
215: 
216: 
217: \section{Data and measurements}
218: \label{data}
219: 
220: 
221: \subsection{Targets}
222: 
223: The SINGS sample \citep{Kennicutt03}, comprising 75 galaxies, is intended to be a
224: valuable representative set of local galaxies that
225: are not ultraluminous, and whose moderate distances ensure that the properties of the
226: interstellar medium can be studied at relatively small spatial scales (a few hundreds
227: of parsecs at the shortest wavelengths). Numerous sources with mild starbursts or
228: low-luminosity active nuclei are included. Of this sample, we excluded from the present
229: study the objects that were not observed in spectroscopic mode because of their very
230: low brightness (DDO\,154, Ho\,I, M81\,dwA, M81\,dwB),
231: or containing very little dust and nebular emission within the nuclear
232: area mapped by the high spectral resolution modules
233: (the quiescent ellipticals NGC\,584 and NGC\,1404, the quiescent dwarf galaxies DDO\,53,
234: DDO\,165, Ho\,IX and the asymmetric magellanic galaxies NGC\,4236, NGC\,5398, NGC\,5408
235: and IC\,4710). The dwarf galaxies IC\,2574 and NGC\,5474 were also rejected because they
236: lack observations in some of the spectroscopic modules. Of the two star-forming
237: dwarf galaxies with several extranuclear pointings, Ho\,II and NGC\,6822, we retained
238: only NGC\,6822 here; the regions within Ho\,II are indeed two faint to allow an
239: analysis of the H$_2$ excitation diagram, contrary to some regions within NGC\,6822.
240: Low-mass galaxies with extranuclear pointings will be discussed elsewhere.
241: NGC\,3034 (M\,82) was excluded due to the unavailability of nuclear spectroscopy from
242: SINGS, as well as NGC\,1377, which constitutes a galaxy class of its own very different
243: from the rest of the SINGS sample, and has been discussed separately \citep{Roussel06}.
244: The sample for H$_2$ measurements comprises 66 targets in 57 galaxies
245: (Table~\ref{tab_target}).
246: The pointings are centered either on the nuclear regions (for most targets) or on some
247: bright star-forming complexes (for a few dwarf galaxies and a spiral galaxy).
248: Diffuse regions within galactic disks are not covered by the present study.
249: 
250: The aperture over which we extracted the spectra is the intersection of the various
251: areas covered by all four spectroscopic modules. The central position and solid
252: angle of this aperture, used to measure all the quantities presented in this paper (line
253: and continuum fluxes), is listed for each galaxy in Table~\ref{tab_target}. In practice,
254: the limiting size is that of the maps performed with the high-resolution modules,
255: which were enlarged in a few cases in order to cover the emission from a star-forming
256: circumnuclear ring. At the distances of the targets, the equivalent linear diameters
257: of the apertures range from 60\,pc to 3.8\,kpc (distribution shown in
258: Fig.~\ref{fig:diameters}), and the median is 900\,pc.
259: Although the apertures are in general small fractions of the optical extent of the
260: galaxies, the measurements are still averages over very large and complex areas.
261: It is expected that a large number of disconnected star formation sites, in addition
262: to the nucleus, contribute to the total emission.
263: 
264: 
265: \subsection{Broadband imaging}
266: \label{images}
267: 
268: To estimate flux densities of the dust continuum and of the aromatic bands
269: (also referred to as the emission from PAHs, or polycyclic aromatic hydrocarbons),
270: we used images in the 3.6\,$\mu$m and 7.9\,$\mu$m bands of the IRAC camera \citep{Fazio04},
271: and scan maps in the three bands of the MIPS instrument \citep{Rieke04} at effective
272: wavelengths of 24, 71 and 156\,$\mu$m.
273: Since in early-type galaxies photospheric emission can make an important contribution
274: to 7.9\,$\mu$m fluxes, we subtracted an estimate of this component in order to
275: obtain pure measurements of aromatic band emission. To this effect, we scaled
276: 3.6\,$\mu$m fluxes, assumed to be dominated by stellar emission, as described
277: in \citet{Helou04}. The resulting flux densities are noted $F_{\rm 7.9\,dust}$.
278: 
279: The observing strategy and data reduction are
280: described by \citet{Kennicutt03}. The full width at half maximum of the point spread
281: function (PSF) is close to 2\arcsec\ at 7.9\,$\mu$m, 6\arcsec\ at 24\,$\mu$m,
282: 18\arcsec\ at 71\,$\mu$m and 40\arcsec\ at 156\,$\mu$m. Flux calibration uncertainties
283: are of the order of 10\% in the IRAC bands, and 5\%, 10\% and 15\% in the MIPS 24, 71
284: and 156\,$\mu$m bands, respectively.
285: To correct for the effects of light scattering in IRAC arrays, we applied to flux
286: densities measured from IRAC maps corrective factors that are appropriate for the
287: photometry of extended sources within apertures of arbitrary size
288: \citep[derived by T. Jarrett and published by][]{Dale07}.
289: For our apertures, the correction factor at 7.9\,$\mu$m is of the order of 10\%.
290: 
291: 
292: \subsection{Spectroscopic data}
293: \label{spectra}
294: 
295: The targets were observed in mapping mode with the IRS instrument \citep{Houck04},
296: at low spectral resolution between 5 and 38\,$\mu$m, with the SL and LL slits
297: ($\lambda / \Delta \lambda \approx 60$--130) and at high spectral resolution between
298: 10 and 37\,$\mu$m, with the SH and LH slits ($\lambda / \Delta \lambda \approx 600$).
299: The observing strategy is described by \citet{Kennicutt03} and \citet{Smith04}.
300: The data were pre-processed with the S13 version of the Spitzer Science Center pipeline.
301: Pixels with an abnormal responsivity were masked, and spectral cubes were built
302: with the Cubism software \citep{Smith07a}. The flux calibration
303: was performed as described by \citet{Roussel06}. We checked the accuracy of this
304: procedure by systematically comparing broadband fluxes from imaging observations and
305: from spectra, and line fluxes from high and low spectral resolution spectra, for bright
306: lines that are minimally contaminated by broad aromatic features at low resolution
307: (but note that even if the flux calibrations of the different modules were in perfect
308: agreement, deviations would be expected from slight misalignment between the apertures).
309: We obtain $F_{24}{\rm (MIPS)} / F_{24}{\rm (LL)} = 1.01 \pm 0.04$ (for targets with
310: $F_{24} > 0.025$\,Jy within a diameter of about 50\arcsec),
311: $F_{7.9}{\rm (IRAC)} / F_{7.9}{\rm (SL)} = 0.99 \pm 0.05$ (for targets with
312: $F_{7.9} > 0.025$\,Jy within a diameter of about 30\arcsec\ and accurately determined
313: backgrounds in SL maps),
314: $F_{\rm [SiII]}{\rm (LL)} / F_{\rm [SiII]}{\rm (LH)} = 0.96 \pm 0.14$,
315: $F_{\rm [SIII]~34}{\rm (LL)} / F_{\rm [SIII]~34}{\rm (LH)} = 0.92 \pm 0.20$
316: and $F_{\rm [SIII]~19}{\rm (LL)} / F_{\rm [SIII]~19}{\rm (SH)} = 1.05 \pm 0.21$
317: (for targets with line fluxes above $6 \sigma$).
318: 
319: 
320: \subsection{Measurements}
321: \label{measurements}
322: 
323: The S(0) to S(3) rotational transitions of H$_2$ (Table~\ref{tab_lines}) were measured
324: for all targets. In addition, we measured the S(4) to S(7) transitions in three
325: galaxies in which these lines are bright enough to become detectable
326: at low spectral resolution (see Table~\ref{tab_flux}).
327: 
328: In high spectral resolution data, we defined errors from fluctuations of the
329: pseudo-continuum, which was fitted as an affine function of wavelength
330: ($F_{\nu} = {\rm a} \lambda + {\rm b}$).
331: In SL data, errors at each wavelength were estimated from spatial fluctuations
332: of blank fields within the satellite spectral maps that are automatically obtained
333: when the source lies in the other half of the slit \citep[see][]{Smith04}.
334: Both the fluxes and the errors presented in Table~\ref{tab_flux} were then
335: added linearly for each point of the line profile above the pseudo-continuum.
336: The line profiles were constrained to have a width compatible with the spectral
337: resolution, since the latter is sufficiently low that no line is resolved.
338: Fig.~\ref{fig:lines} shows the line spectra for the representative galaxies
339: NGC\,1097, NGC\,6946, NGC\,7552, NGC\,1266, NGC\,4569 and NGC\,4579.
340: 
341: The S(1) line is usually the brightest.
342: Of the non-dwarf galaxies of the SINGS sample (with stellar masses estimated
343: as by \citet{Lee06} above $10^{9.5}$\,M$_{\sun}$), the nuclear regions of 86\%
344: are securely detected in the S(1) line, with fluxes above three times the
345: measured error. The other 14\% are either ellipticals of the {\small LINER}
346: type, or late-type spirals (Sc-Sd).
347: 
348: There are two galaxies in common between this sample and that of \citet{Rigopoulou02},
349: namely NGC\,7552 and NGC\,6946, the latter from the study of \citet{Valentijn96}.
350: For both, our aperture is larger than the beam of ISO-SWS, which covered an area
351: of 280 to 380 arcsec$^2$. For the lines that were detected with SWS, we obtain
352: fluxes that are higher by factors of 2.3 (S(1) in NGC\,6946), 5.6 (S(0) in NGC\,6946)
353: and 1.1 (S(1) and S(3) in NGC\,7552).
354: The exact placement of the ISO-SWS beam is not known. For NGC\,6946, given this
355: uncertainty, it is conceivable that the H$_2$ emission be twice as bright in our
356: 800 arcsec$^2$ aperture as in the SWS aperture; but the S(0) line flux of
357: \citet{Valentijn96} is inconsistent with our data.
358: 
359: For this study, we estimate total infrared fluxes (TIR) between 3 and 1100\,$\mu$m,
360: defined as a linear combination of 24, 71 and 156\,$\mu$m flux densities.
361: The formula of \citet{Dale02} is used here, and we have checked that replacing
362: it with the more recent prescription by \citet{Draine07a} does not change the
363: following results in any appreciable way.
364: The infrared fluxes are measured within the same area as the other quantities for direct
365: comparison. The PSF width at 156\,$\mu$m is however much larger than the size of our
366: spectroscopic aperture, so that some extrapolation is needed. We first measure MIPS
367: fluxes within the larger aperture used to compare total infrared fluxes with line fluxes
368: measured in the LL module. Then,
369: we scale these fluxes by the ratio of $F_{24}$ measured in the small aperture to
370: $F_{24}$ measured in the larger aperture, which is equivalent to assuming that the
371: spectral energy distribution does not change from an area of $\approx 300$\,arcsec$^2$
372: to an area of $\approx 2000$\,arcsec$^2$. The associated errors are however expected
373: to be small compared with the dynamic range of the quantities discussed in
374: Section~\ref{powers}.
375: Simulations of the overestimation of the far-infrared fluxes caused by
376: the extrapolation, using a simple model of a point-source starburst (with
377: the spectral energy distribution of Mrk\,33) superposed on quasi-uniform
378: emission from low radiation field intensity regions (with the colors
379: of the central regions of NGC\,24 or NGC\,2403), indicate that the effect
380: should be in most cases of the order of 20\% (when the starburst and quiescent
381: components contribute equally at 156\,$\mu$m), and in extreme cases reach a
382: maximum of a factor 2 (when the quiescent component dominates).
383: \citet{Smith07b} reached a similar conclusion (see their Section 3.2).
384: 
385: 
386: \section{Excitation diagrams}
387: 
388: Excitation diagrams provide a convenient visualization of the distribution of level
389: populations and allow first constraints on the excitation mechanisms (thermal or
390: non-thermal) that can produce this distribution. They represent the column density
391: in the upper level of each observed transition $N_{\rm u}$, normalized by its statistical
392: weight $g_{\rm u}$, as a function of the upper level energy $E_{\rm u}$.
393: The flux of a transition can be written as
394: $F = h \nu~ A~ N_{\rm u}~ \Omega / (4 \pi)$, where $A$ is the spontaneous emission
395: probability, $h \nu$ is the transition energy and $\Omega$ is the beam solid angle.
396: In the assumption of local thermodynamic equilibrium, the total column density
397: $N_{\rm tot}$ can be derived from
398: $N_{\rm u} = g_{\rm u}~ N_{\rm tot}~ {\rm exp}(-E_{\rm u}~ /~ (kT))~ /~ Z(T)$, where
399: $g_{\rm u} = (2 I + 1)~ (2 J + 1)$ is the statistical weight (with the spin number
400: $I=0$ for even J or para transitions, and $I=1$ for odd J or ortho transitions),
401: and $Z(T) \sim 0.0247~ T~ /~ (1 - {\rm exp}(-6000\,{\rm K}~ /~ T))$ is the partition
402: function \citep{Herbst96}, valid for $T > 40$\,K.
403: 
404: The apparent excitation temperature can then be derived from each pair of transitions by:
405: \begin{equation}
406: kT = (E_{\rm u2} - E_{\rm u1})~ /~ {\rm ln}(N_{\rm u1} / N_{\rm u2} \times g_{\rm u2} / g_{\rm u1})
407: \end{equation}
408: with $N_{\rm u1} / N_{\rm u2} = F_1 / F_2 \times A_2 / A_1 \times \lambda_1 / \lambda_2$.
409: Since both radiative decay and collisions with H$_2$ change the rotational number $J$
410: by an even number, the ortho and para states are largely 
411: decoupled and should in principle be dealt with independently.
412: 
413: 
414: \subsection{Ortho-para thermalization and departures therefrom}
415: 
416: As emphasized by \citet{Burton92}, the lower rotational levels of H$_2$ will be in
417: collisional equilibrium over a wide range of conditions, because their critical
418: densities are low. Figure~\ref{fig:ncrit} shows the critical densities of all the
419: rotational transitions observable with the IRS instrument, as a function of temperature,
420: computed using the functional form for the collisional de-excitation rate coefficient
421: by H$_2$ given by \citet{Shull82} and the transition probabilities given by
422: \citet{Black76}.
423: The derived critical densities for each line are about an order of magnitude lower
424: than those for collisions with H computed by \citet{Mandy93}, the comparison being
425: made at 600\,K, since \citet{Mandy93} provide results only for high temperatures.
426: 
427: The integrated emission from warm H$_2$ in star-forming galaxies is likely to
428: come predominantly from the densest photodissociation regions (PDRs) within the beam,
429: with densities above $10^3$\,cm$^{-3}$ \citep{Burton92, Kaufman06},
430: in which case the lowest
431: rotational levels will be thermalized. Observations of starburst galaxies with ISO-SWS
432: \citep{Rigopoulou02} as well as ultraluminous galaxies with Spitzer-IRS \citep{Higdon06}
433: are indeed consistent with this expectation. At first sight, the same applies to
434: the galaxies studied here.
435: 
436: However, some of the excitation diagrams show departures from thermalization
437: of ortho levels with para levels,
438: in the sense that the apparent temperatures derived from each pair of transitions
439: of consecutive rotational number are not monotonic as a function of upper level energy.
440: Clear examples are NGC\,1266
441: ($T{\rm (S0-S1)} = (201 \pm 45)$\,K, $T{\rm (S1-S2)} = (465 \pm 34)$\,K
442: and $T{\rm (S2-S3)} = (347 \pm 18)$\,K);
443: NGC\,4254 ($(162 \pm 9)$\,K, $(358 \pm 59)$\,K and $(259 \pm 38)$\,K);
444: and NGC\,4631 ($(127 \pm 8)$\,K, $(342 \pm 39)$\,K and $(268 \pm 25)$\,K).
445: Such deviations from thermalization can be explained by an ortho to para density
446: ratio in the excited states apparently different from the equilibrium value.
447: We have
448: \begin{eqnarray}
449: OPR~ & =~ \frac{OPR_{\rm \,high\,T}}{3}~ \frac{\sum_o (2 I_o + 1)~ (2 J_o + 1)~ {\rm exp}(-E_o~ /~ (kT))}{\sum_p (2 I_p + 1)~ (2 J_p + 1)~ {\rm exp}(-E_p~ /~ (kT))} \nonumber \\
450:      & =~ OPR_{\rm \,high\,T}~ \frac{\sum_o (2 J_o + 1)~ {\rm exp}(-E_o~ /~ (kT))}{\sum_p (2 J_p + 1)~ {\rm exp}(-E_p~ /~ (kT))}
451: \label{eq:opr}
452: \end{eqnarray}
453: where the subscripts $o$ and $p$ designate ortho and para levels respectively
454: ($I_p = 0$ and $I_o = 1$).
455: $OPR_{\rm \,high\,T}$, equal to the actual ortho to para ratio ($OPR$) in the
456: high-temperature limit, expresses deviations from local thermodynamic equilibrium
457: (LTE) if it differs from three.
458: It may be called the effective nuclear spin degeneracy ratio,
459: but will hereafter be called the ortho to para ratio for convenience.
460: In LTE, $OPR \sim 2$ for $T \sim 100$\,K and $OPR \sim 3$ for $T > 200$\,K \citep{Burton92},
461: but $OPR_{\rm \,high\,T} = 3$ at all temperatures.
462: Although $OPR_{\rm \,high\,T} < 3$ may be inferred for the excited states ($J \geq 2$),
463: this does not imply that the ortho to para ratio of the bulk of the gas in the
464: $J = 1$ and $J = 0$ states be out of LTE.
465: In the following, LTE will refer more particularly to the equilibrium
466: between the ortho and para levels, and not of the ortho levels or para
467: levels separately. Extinction effects are discussed in Section~\ref{extinction}
468: and the interpretation of $OPR_{\rm \,high\,T}$ values is postponed to Section~\ref{excitation}.
469: 
470: To derive temperatures and column densities,
471: we first determine whether the excitation diagram is compatible or not with LTE
472: by inserting explicitly the factor $OPR_{\rm \,high\,T} / 3$ in the equations
473: for column densities of the ortho levels, and deriving temperatures from each pair of
474: consecutive transitions as a function of $OPR_{\rm \,high\,T}$, to verify whether these
475: conditions are satisfied:
476: $T{\rm (S0-S1)} \leq T{\rm (S0-S2)} \leq T{\rm (S1-S2)} \leq T{\rm (S1-S3)} \leq T{\rm (S2-S3)}$,
477: since in gas with a distribution of temperatures, ratios of transitions with low-energy
478: upper levels always probe lower excitation temperatures than ratios of transitions
479: with higher-energy upper levels.
480: $T{\rm (S0-S2)}$ and $T{\rm (S1-S3)}$ are independent of $OPR_{\rm \,high\,T}$ and determined
481: directly from the observed fluxes, but $T{\rm (S0-S1)}$, $T{\rm (S1-S2)}$ and $T{\rm (S2-S3)}$
482: depend on $OPR_{\rm \,high\,T}$. For each pair $(p,o)$ = (0,1), (2,1) and (2,3), we have:
483: \begin{equation}
484: k T({\rm S}_p-{\rm S}_o) = (E_{\rm u~o} - E_{\rm u~p})~ /~ \ln(OPR_{\rm \,high\,T} \times R)\,.
485: \end{equation}
486: with $R = F_p / F_o \times A_o / A_p \times \lambda_p / \lambda_o \times (2 J_o + 1)~/~(2 J_p + 1)$.
487: Figure~\ref{fig:diag_temp} shows the corresponding diagram for two galaxies.
488: In case the above condition on the temperatures is satisfied for $OPR_{\rm \,high\,T} = 3$,
489: as illustrated for NGC\,3198, we fix $OPR_{\rm \,high\,T} = 3$~; in the opposite case,
490: illustrated by NGC\,4631, we fit $OPR_{\rm \,high\,T}$ as explained below.
491: The excitation diagrams of all the galaxies, with fits overlaid, are shown in
492: Fig.~\ref{fig:diag_exc}.
493: 
494: 
495: \subsection{Temperatures and column densities}
496: \label{fits}
497: 
498: Since in all cases the excitation diagrams indicate that a single temperature does not fit
499: all the line fluxes, we assume that the H$_2$ emission is the sum of two discrete components
500: of different temperatures, which is enough to reproduce accurately the observed fluxes.
501: In the general case of $OPR_{\rm \,high\,T} = 3$, we perform a least-squares fit of the
502: excitation diagram to determine
503: the parameters of the two discrete components (the lower temperature $T_1$, the upper
504: temperature $T_2$, and their mass fraction) and the normalization by the total
505: column density. The results are listed in Table~\ref{tab_fit}.
506: 
507: When the gas is at a range of temperatures, it is in practice impossible to lift the
508: degeneracy between mass and temperature from the lowest-energy levels.
509: Since the column density has a very steep dependence on $T_1$, we adopt two different
510: procedures to fit the excitation diagrams and ascertain the amplitude of the
511: uncertainties caused by this degeneracy. In the first case, we constrain $T_1$ to
512: exceed the value for which the column density is 20\% higher than the nominal density
513: derived from $T{\rm (S0-S1)}$. In the second case, we leave $T_1$ unconstrained.
514: In the following, both approaches will be retained when discussing results that
515: depend on $T_1$.
516: 
517: For the results not to be biased by systematic sensitivity
518: differences at the wavelengths of the H$_2$ transitions, we also replace the measured
519: errors by a uniform weight. When $OPR_{\rm \,high\,T}$ is allowed to be fitted, we fix $T_2$ at
520: $1.3 \times T{\rm (S1-S3)}$ in the constrained-$T_1$ fits, which was chosen from
521: the median value of $T_2$ in galaxies with $OPR_{\rm \,high\,T} = 3$. In free-$T_1$ fits
522: with $OPR_{\rm \,high\,T} = 3$, the distribution of $T_2 / T{\rm (S1-S3)}$ is large, with a tail
523: of high values; therefore, $T_2$ is first fixed at the median value,
524: $1.14 \times T{\rm (S1-S3)}$, and then at $1.5 \times T{\rm (S1-S3)}$, to probe
525: the full range of most likely values.
526: Finally, when one flux is an upper limit, we fix both $T_1$ at $0.98 \times T{\rm (S0-S1)}$
527: (which increases the total column density by a maximum of $\sim 20$\% with respect to
528: that obtained with $T_1 = T{\rm (S0-S1)}$ but allows a small contribution from hotter
529: gas to the S(0) and S(1) lines), and $T_2$ as above.
530: 
531: For the three galaxies from which more transitions, up to S(7), could be measured,
532: the procedure is the same except that a third component has to be added. The
533: additional parameters are $T_3$ and the mass fraction of the second component,
534: and $T_2$ is fixed at 400\,K.
535: 
536: Several galaxies barely satisfy the criterion on temperatures to have
537: $OPR_{\rm \,high\,T} = 3$, with $T{\rm (S1-S2)} \geq 0.95 \times T{\rm (S1-S3)}$
538: and $T{\rm (S2-S3)} \leq 1.05 \times T{\rm (S1-S3)}$. When $T_1$ is constrained,
539: the quality of their fits can be improved by allowing $OPR_{\rm \,high\,T}$ to vary.
540: For these objects, we provide results with $OPR_{\rm \,high\,T} < 3$. Allowing
541: $OPR_{\rm \,high\,T}$ to be smaller than the equilibrium value has the indirect consequence
542: that the derived column densities are smaller. The amplitude of this effect is
543: indicated in Table~\ref{tab_fit}. Similarly, for NGC\,1705 and NGC\,4552, we provide
544: results with $OPR_{\rm \,high\,T} < 3$ and indicate the change in column density with
545: respect to $OPR_{\rm \,high\,T} = 3$, because although the S(2) transition being an upper
546: limit prevents any reliable determination of $OPR_{\rm \,high\,T}$, the $T_1$ temperatures
547: derived with $OPR_{\rm \,high\,T} = 3$ are the two lowest of the whole sample, raising
548: the suspicion that they might be artifacts of the constraint on $OPR_{\rm \,high\,T}$.
549: We also consider $OPR_{\rm \,high\,T} < 3$ more likely for these galaxies in view of the
550: dependence of $OPR_{\rm \,high\,T}$ on H$_2$ brightness, 
551: discussed later in Section~\ref{excitation}.
552: 
553: The median $T_1$ temperature is 154\,K when the fits are constrained (ranging between 97
554: and 300\,K); when no constraint is applied, the median $T_1$ is 118\,K with
555: $T_2 = 1.14 \times T{\rm (S1-S3)}$, and 161\,K with $T_2 = 1.5 \times T{\rm (S1-S3)}$.
556: The total column densities that we obtained, averaged over kiloparsec-scale regions
557: in galactic centers,
558: range between $10^{19}$ and $2 \times 10^{21}$\,cm$^{-2}$ (for constrained-$T_1$ fits),
559: or $2 \times 10^{22}$\,cm$^{-2}$ (for free-$T_1$ fits), and their medians are
560: respectively $3 \times 10^{20}$\,cm$^{-2}$ and 5--$6 \times 10^{20}$\,cm$^{-2}$
561: (Fig.~\ref{fig:coldens}).
562: This can be compared with typical column densities of resolved photodissociation regions
563: in the Milky Way. In the Orion Bar, column densities of H$_2$ warmer than 400\,K,
564: derived from rotational lines, lie between $10^{20}$ and $10^{21}$\,cm$^{-2}$
565: \citep{Parmar91, Allers05}.
566: Note that because the Orion Bar is observed nearly edge-on, an equivalent PDR
567: seen face-on would have lower column densities.
568: In NGC\,7023, \citet{Fuente99} derived a total column
569: density of $5 \times 10^{20}$\,cm$^{-2}$ for H$_2$ warmer than 300\,K.
570: Thus, if the H$_2$ emission in our targets comes from similar photodissociation
571: regions, they must occupy in general a very large fraction of the observing beam,
572: assuming that they do not overlap on the line of sight.
573: 
574: Figure~\ref{fig:coldens} also shows a clear dependence of the local (nuclear) column
575: density of warm H$_2$ on the total stellar mass of the host galaxy.
576: The stellar mass and the infrared luminosity being correlated for star-forming
577: galaxies, there is a similar dependence on far-infrared luminosities.
578: To first order, the column density of warm H$_2$ shows the same behavior as
579: tracers of molecular gas and star formation rate densities, which suggests
580: that the primary source of H$_2$ heating is the star formation activity in
581: non-AGN galaxies, and the nuclear regions respond to the global mass and luminosity.
582: {\small LINER} and Sy nuclei do not follow the correlation shown
583: by star-forming regions, and tend to have smaller column densities of warm H$_2$.
584: The differences in terms of energy output and excitation mechanisms will be studied
585: in more detail in Sections~\ref{powers}, \ref{excitation} and \ref{syliner}.
586: Since the few extranuclear regions and dwarf galaxies included in the sample
587: do not distinguish themselves from the other star-forming targets in any obvious
588: way, here and in the following, they are not discussed as separate categories.
589: 
590: 
591: \subsection{Optical depth toward H$_2$}
592: \label{extinction}
593: 
594: Consistent with the negligible optical depths inferred from the silicate absorption
595: bands at 10\,$\mu$m and 18\,$\mu$m in most SINGS galaxies \citep{Smith07b},
596: that support the modest values of nebular extinction derived from the Balmer
597: decrement \citep{Dale06}, we assume zero extinction both in the lines and in
598: the dust continuum for all the targets. In eight galactic centers among the
599: SINGS sample (included here),
600: \citet{Smith07b} obtained a better fit in their decomposition of the low
601: spectral resolution spectra by including a finite optical depth in the silicate
602: bands. We expect the warm H$_2$ component to suffer less extinction, on average,
603: than the warm dust continuum, because the two emission sources will not be cospatial
604: in general, and the regions of high optical depth will be confined to compact
605: regions, probably more concentrated than the regions participating in H$_2$ emission
606: \citep[see the striking example of NGC\,1377;][]{Roussel06}.
607: In particular, \citet{Higdon06} did not see any evidence for significant extinction
608: in the rotational H$_2$ lines of ultraluminous galaxies, although these objects
609: are expected to have much higher optical depths than the present sample.
610: In the absence of any quantitative constraint on the differential extinction
611: between the dust and H$_2$, we do not attempt to correct H$_2$ fluxes for extinction.
612: 
613: Using the extinction law of \citet{Moneti01}, valid for the Galactic center, we have
614: $A(9.7\,\mu{\rm m}) / A_{\rm V} = 0.15$, $A(28.2\,\mu{\rm m}) / A(9.7\,\mu{\rm m}) = 0.25$ and
615: $A(17.0\,\mu{\rm m}) / A(9.7\,\mu{\rm m}) = A(12.3\,\mu{\rm m}) / A(9.7\,\mu{\rm m}) = 0.46$\,.
616: Even assuming the same optical depth toward the warm molecular hydrogen
617: as toward the hot dust,
618: the extinction correction would not change significantly the derived column densities.
619: The extinction is modest at 10\,$\mu$m, and therefore negligible at 28\,$\mu$m,
620: the wavelength of the S(0) line which dominates the total column density determination.
621: Extinction effects would however depress the S(1) and S(3) line fluxes with respect
622: to S(0) and S(2), and could thus artificially lower the derived $OPR_{\rm \,high\,T}$.
623: In the following, we put lower limits to $OPR_{\rm \,high\,T}$ values, when less than 3,
624: derived for the eight galaxies with non-zero optical depth at 10\,$\mu$m.
625: 
626: NGC\,3198 is the sample galaxy with the highest optical depth in the silicate
627: feature according to \citet{Smith07b}, but its excitation diagram shows no sign of
628: attenuation of the S(1) and S(3) lines relative to the others, and is consistent with
629: $OPR_{\rm \,high\,T} = 3$ (Fig.~\ref{fig:diag_temp}). The second most obscured galaxy
630: of the present sample is NGC\,1266 (it also has the highest nebular extinction according
631: to \citet{Dale06}, $A_{\rm V} = 4.1$\,mag), for which we derive $OPR_{\rm \,high\,T} < 3$.
632: If this were due to optical depth effects, then the S(3) line at 9.7\,$\mu$m should
633: be more attenuated than the S(1) line at 17.0\,$\mu$m. Since this would be consistent
634: with the excitation diagram, we cannot exclude that the apparently low $OPR_{\rm \,high\,T}$
635: value be an extinction artifact in at least this galaxy.
636: The dissimilar behavior of the two galaxies in terms of differential
637: extinction between H$_2$ and the dust could then arise from different excitation
638: mechanisms and geometries: whereas in the nuclear regions of NGC\,3198, classified
639: as purely H{\small II}, the H$_2$ emission is presumably distributed over a large
640: volume, the H$_2$ emission in the {\small LINER} nucleus of NGC\,1266 may be much
641: more compact, and not produced by star formation processes (see Section~\ref{powers}).
642: For 13 galaxies with negligible silicate extinction in the spectral decomposition
643: performed by \citet{Smith07b}, the excitation diagrams do imply $OPR_{\rm \,high\,T} < 3$,
644: whether a constraint on the lower temperature $T_1$ is applied or not.
645: In addition, of the 6 galaxies found to have non-zero silicate extinction and
646: $OPR_{\rm \,high\,T} < 3$, three would require $\tau{\rm(H_2)} > \tau_{\rm sil}$ in order
647: to obtain $OPR_{\rm \,high\,T} = 3$ after extinction correction (by $\geq 25$\% for
648: NGC\,1266, by a factor $\geq 6$ for NGC\,4631 and by a factor $\geq 3.5$ for NGC\,5866).
649: The three others (NGC\,1482, 4536 and 6946) would require either
650: $\tau{\rm(H_2)} > \tau_{\rm sil}$, or very low $T_1$ temperatures ($\leq 100$\,K).
651: Since it is unlikely that the optical depth toward H$_2$ be higher than toward the
652: dust continuum, we conclude that our finding, discussed in Section~\ref{excitation},
653: is robust against extinction effects.
654: 
655: 
656: \section{Mass fraction in the warm phase}
657: \label{h2co}
658: 
659: In order to estimate the fraction of molecular hydrogen that is heated to temperatures
660: above $\sim 100$\,K, we searched the literature for observed intensities of the 2.6\,mm CO(1-0)
661: line within a beam comparable to the solid angle of our observations. Table~\ref{tab_mass}
662: summarizes the adopted data. The column density of cold H$_2$ as given here is derived
663: from CO velocity-integrated intensities on the main-beam temperature scale, assuming
664: a uniform conversion factor of CO(1-0) intensities to H$_2$ column densities of
665: $2.3 \times 10^{20}$\,cm$^{-2}$/(K\,km\,s$^{-1}$) \citep{Strong88}.
666: We derived aperture corrections to the CO intensities by projecting on a map both
667: the IRS beam and the CO beam. We did not use any deconvolution technique.
668: Whenever possible, a map from the BIMA SONG interferometric survey, including
669: the zero-spacing total intensity \citep{Helfer03}, was used.
670: Otherwise, we used instead the 7.9\,$\mu$m map and assumed the spatial distributions
671: of aromatic bands in emission and CO(1-0) line emission to be similar
672: at the large spatial scales corresponding to our apertures.
673: This can be justified qualitatively by the association of dust with molecular gas
674: and the Schmidt law (for a recent study of the spatially-resolved Schmidt law,
675: see Kennicutt et al. 2007, in preparation).
676: The applied
677: correction factors are listed in Table~\ref{tab_mass}. In some cases, there are
678: several available measurements all giving consistent estimates to within 30\%~;
679: the corresponding unused references are given within parentheses.
680: 
681: There are two major sources of uncertainty in this comparison. The first one is inherent
682: to the difficulty of matching the physical area covered by the IRS integral-field
683: measurements, from single-dish or aperture-synthesis measurements within
684: a different beam. The second dominant source of uncertainty comes
685: from the conversion factor of CO intensities to H$_2$ masses, assumed uniform here.
686: The result of \citet{Strong88} is derived from a comparison of Galactic $\gamma$-ray
687: emission with CO and H{\small I} emission. \citet{Dame01} obtained a consistent
688: conversion factor by extrapolating the gas-to-dust mass ratio measured from H{\small I}
689: and far-infrared emission, in areas devoid of CO emission, to molecular clouds.
690: Both methods provide an estimate of the total H$_2$ column density, including the
691: warm gas as well as the cold gas, for molecular clouds under similar average
692: physical conditions as Galactic clouds.
693: Note however that conversion factors both significantly lower and significantly higher
694: have been derived for normal galaxies. For instance, the recent study of \citet{Draine07b}
695: favors an average value of $4 \times 10^{20}$\,cm$^{-2}$/(K\,km\,s$^{-1}$), based
696: on global gas-to-dust mass ratios in the SINGS sample. In addition,
697: the ratio of H$_2$ column density to CO intensity can vary by at least a factor
698: two, depending on the physical conditions of the regions emitting in CO \citep{Maloney88},
699: even though our observing aperture is large enough to cover a large number of molecular
700: clouds and dilute some of the dispersion in their physical properties.
701: In particular, the conversion factor is expected to be lower for compact and actively
702: star-forming regions than for more diffuse and more quiescent regions.
703: We discount here variations due to metal abundance, since we could find CO measurements
704: for only two low-metallicity targets (NGC\,2915 and NGC\,6822\_A).
705: 
706: Figure~\ref{fig:frac_warm} shows the mass fraction of molecular hydrogen in the warm
707: phase ($T \geq T_1 \approx 100$\,K) as a function of the minimum temperature of the
708: warm component, as determined by the lowest-energy rotational H$_2$ lines.
709: The nuclei classified as star-forming have a relatively narrow range of lower
710: temperatures ($T_1 = 144 \pm 24$\,K for 31 nuclei, with or without CO data,
711: from the constrained fits). However, for nuclear regions classified as
712: {\small LINER}s or Seyferts, the spread in temperatures is higher
713: ($T_1 = 180 \pm 45$\,K for 25 nuclei). No statistically-significant difference
714: exists between the 18 {\small LINER} and 7 Sy nuclei.
715: 
716: A clear anticorrelation exists between the two quantities plotted
717: (partly the result of the degeneracy between temperatures and column densities),
718: which remains intact
719: when restricting the sample to those galaxies for which we could find well-matched
720: CO data (i.e. with correction factors close to unity and with several consistent
721: measurements). The dynamic range in the warm gas mass fraction is much
722: higher than accounted for by the uncertainty on the total H$_2$ mass.
723: The uncertainty on the warm H$_2$ mass for individual objects is however extremely
724: large, owing to the degeneracy between $T_1$, often ill-constrained by the data,
725: and the column density. The example of NGC\,4579 is the most striking
726: (see Table~\ref{tab_fit}).
727: Since its rotational levels up to J=5 are close to thermal equilibrium (at a
728: single temperature of the order of 300-400\,K), such a component at 70\,K
729: as found in the free-$T_1$ fit is unlikely to be real.
730: Because the fits where $T_1$ is unconstrained allow mass fractions in the warm phase
731: that are sometimes unphysical (for example for NGC\,2976 and NGC\,4826),
732: we favor the constrained fits as more plausible, but emphasize that the
733: mass distribution at low temperatures is in general unconstrained.
734: 
735: In the case of constrained-$T_1$ fits,
736: it appears that for a small set of nuclear regions classified as {\small LINER}s
737: or Seyferts, the warm H$_2$ phase consists only of a very small fraction of the total
738: mass, but heated to higher temperatures than in regions classified as purely star-forming.
739: This behavior arises naturally if normal photodissociation region excitation is missing,
740: and if the hotter gas is located in a thin layer of molecular clouds, or has a small
741: filling factor.
742: In the case of free-$T_1$ fits, only NGC\,1316 (Fornax\,A) remains robustly in the part
743: of the diagram with high $T_1$ and mass fraction below 3\%. The average temperature
744: is however still higher for {\small LINER}s and Seyferts than for H{\small II}
745: nuclei, and the average mass fraction in the warm phase likewise lower.
746: The reason for this difference will be further discussed in Section~\ref{syliner},
747: addressing the excitation mechanisms.
748: 
749: 
750: \section{Comparison of the powers emitted by warm H$_2$, [Si{\small II}] and dust in star-forming regions}
751: \label{powers}
752: 
753: In order to empirically quantify the importance of the H$_2$ rotational lines in cooling
754: the interstellar medium of normal galaxies, and to put constraints on the possible
755: excitation mechanisms of H$_2$, discussed in more detail in Section~\ref{excitation},
756: we examine power ratios of H$_2$ to other tracers of the warm interstellar medium
757: extracted from the same observations. The results presented here are independent
758: of any fits to the excitation diagrams. Only the H{\small II} nuclei and complexes
759: are considered, {\small LINER} and Sy nuclei being separately discussed in
760: Section~\ref{syliner}.
761: Since the bulk of warm H$_2$, at the lowest rotational temperatures, emits mostly
762: in the S(0) to S(2) lines, whereas the S(3) line emission has a noticeably higher
763: contribution from hotter H$_2$, probably indicating more mixed excitation sources
764: (anticipating the discussion of excitation mechanisms, see
765: Section~\ref{temp_constraints}), we choose, as the most useful quantification of
766: H$_2$ rotational emission in star-forming targets, the sum of the S(0) to S(2) lines.
767: 
768: 
769: \subsection{Total infrared emission}
770: 
771: In photodissociation regions, almost all the far-ultraviolet power from massive stars
772: that does not escape is absorbed by dust and converted to infrared continuum radiation,
773: or is absorbed by H$_2$. Only a very small fraction of the power absorbed
774: by dust, of the order of 1\%, is converted to photoelectrons that heat
775: the gas, and emerges as infrared lines \citep{Tielens85}.
776: The dominant gas coolants are the [O{\small I}] and [C{\small II}] lines
777: at 63\,$\mu$m and 158\,$\mu$m, but mid-infrared lines, in particular [Si{\small II}]
778: at 34.8\,$\mu$m and the H$_2$ rotational lines, are also energetically significant.
779: Although the transition rate coefficients of H$_2$ are low
780: and the excitation energies relatively high, H$_2$ molecules are dominant in number.
781: 
782: The observed ratios of the power emitted in the sum of the S(0) to S(2) lines
783: to the total dust power emitted in the infrared (TIR; see Section~\ref{measurements})
784: range between $2.5 \times 10^{-4}$ and $7.5 \times 10^{-4}$ for nuclear regions that
785: are not classified as {\small LINER}s or Seyferts (Fig.~\ref{fig:frac_tir}a).
786: These ratios are in agreement with predictions of the photodissociation models of
787: \citet{Kaufman06} for a wide variety of radiation field intensities $G_0$ and hydrogen
788: densities $n$, but a relatively narrow range of $G_0 / n$ ratios, approximately
789: between 0.1 and 1 with $G_0$ in units of $1.6 \times 10^{-3}$\,erg\,s$^{-1}$\,cm$^{-2}$
790: and $n$ in units of cm$^{-3}$. Note that models predict the ratio of the H$_2$ line power
791: to the far-ultraviolet (FUV) power (for photon energies between 6 and 13.6\,eV), rather
792: than the total infrared power. Since the intrinsic FUV flux heating the photodissociation
793: regions is unknown, the comparison between observations and models is here made by assuming
794: an exact conversion of FUV photons to infrared photons.
795: The fraction of dust heating provided by non-FUV photons can however be
796: significant.
797: Allowing for this effect would reduce the derived $G_0 / n$ ratios.
798: The H$_2$ rotational line fluxes predicted by \citet{Kaufman06} are nearly an
799: order of magnitude higher than those from the older models of \citet{Burton92},
800: because of the inclusion of photoelectric heating by PAHs,
801: a better H$_2$ model, and a finer numerical grid near the region of H$_2$ emission.
802: 
803: The inferred $G_0 / n$ ratios are lower than the results of \citet{Malhotra01},
804: who derived the physical conditions of an ensemble of bright star-forming galaxies
805: from the [C{\small II}] and [O{\small I}] lines. They found $G_0 / n$ ratios between
806: about 0.5 and 6, i.e. on average 5 times higher than those indicated here by the
807: rotational H$_2$ lines.
808: A possible explanation is that H$_2$ emission comes from cooler and denser regions
809: than [C{\small II}] and [O{\small I}], 
810: because H$_2$ exists at higher optical depths inside the clouds than C$^+$ and O
811: \citep{Hollenbach97}.
812: The difference in physical conditions could thus merely reflect a different spatial origin.
813: Besides the different locations within PDRs, the two studies also deal with different
814: regions within galaxies: the targets of \citet{Malhotra01} were selected to have
815: most of their line emission encompassed by the ISO-LWS beam of 70\arcsec,
816: whereas our apertures usually cover small fractions of the line and dust emitting
817: areas.
818: Alternatively, the observations of \citet{Malhotra01} could reflect intrinsically
819: different physical conditions because their sample contains galaxies on average
820: brighter and more active than the sample used here.
821: Their far-infrared luminosities \citep[in the definition of][]{Helou88} range from
822: $6 \times 10^7$ to $8 \times 10^{11}$\,L$_{\sun}$, with a median of
823: $1.5 \times 10^{10}$\,L$_{\sun}$, whereas the far-infrared luminosities of the
824: present sample range from $10^7$ to $6 \times 10^{10}$\,L$_{\sun}$, with a median
825: of $3 \times 10^9$\,L$_{\sun}$.
826: The median $F_{60}/F_{100}$ ratio is also higher in the sample of \citet{Malhotra01}
827: (0.57) than in our sample (0.41), indicating higher radiation field intensities
828: on average.
829: The $G_0 / n$ ratios derived by \citet{Malhotra01} however do not display any
830: clear correlation with either infrared luminosity or color.
831: Only NGC\,1482 and NGC\,5713, included in both samples, allow a direct comparison
832: of model results (we discard the {\small LINER} NGC\,1266 because most of its H$_2$
833: emission is not produced by PDRs, as shown in Sect.~\ref{syliner}).
834: For both sources, the H$_2$ line fluxes indicate consistently $G_0 \sim 4000$
835: and $n \sim 1$--$2 \times 10^4$. For NGC\,1482, $G_0$ is in agreement with one of
836: the two models of \citet{Malhotra01}, but $n$ is at least four times higher.
837: For NGC\,5713, $G_0$ is two times higher that that of \citet{Malhotra01}, and
838: $n$ is at least six times higher.
839: In conclusion, we favor differences in spatial origin (both within PDRs and within
840: galaxies) as a likely cause for the different model results.
841: 
842: 
843: \subsection{[Si{\small II}] line emission}
844: 
845: Figure~\ref{fig:frac_tir}b shows the ratio of powers emitted in the H$_2$ rotational
846: lines and in the [Si{\small II}] line. The dispersion in the ratio is very similar
847: to that seen in Fig.~\ref{fig:frac_tir}a, and the [Si{\small II}] line alone emits
848: more power than the sum of the S(0) to S(3) transitions in H{\small II} nuclei.
849: The [Si{\small II}] line has indeed been found to be the brightest mid-infrared
850: cooling line and to scale tightly with the total infrared power both in nuclear
851: and extranuclear regions within the SINGS sample galaxies (Helou et al., in
852: preparation), with only a very slight dependence on the radiation field intensity.
853: We have on average $F{\rm (S0-S2)}/F{\rm ([Si{\small II}])} = 0.3$
854: (ranging between 0.15 and 0.5 for nuclei),
855: and $F{\rm ([Si{\small II}])}/TIR = 2 \times 10^{-3}$.
856: Using the [Si{\small II}] line as a substitute for the total dust emission
857: is advantageous because it is observed at about the same angular resolution
858: as the H$_2$ lines, whereas estimating the total infrared
859: power within these apertures requires a large extrapolation (because of the
860: large width of the point spread function at 70 and 160\,$\mu$m), making the
861: uncertainty on H$_2$/TIR relatively high.
862: The [Si{\small II}] power predicted by the photodissociation region model
863: of \citet{Kaufman06}, with the same physical conditions as above,
864: is however smaller than observed by a factor greater than 3,
865: which implies either that the majority
866: of [Si{\small II}] emission comes from H{\small II} regions in high-metallicity
867: nuclear regions, or that the fraction of silicon incorporated in dust grains
868: is smaller than 90\%.
869: 
870: Only the regions B and C in NGC\,6822 have significantly less [Si{\small II}] emission,
871: with respect to H$_2$ emission, than the nuclear regions of spiral galaxies.
872: Their H$_2$ emission is also slightly overluminous with respect to the
873: aromatic bands (Fig.~\ref{fig:frac_tir}c). This may not be entirely attributable to
874: a metallicity effect, decreasing the abundances
875: of PAHs and silicon, since region A (Hubble\,V) has normal flux ratios,
876: and oxygen abundances are quite uniform in NGC\,6822 \citep{Pagel80}. An alternative
877: explanation is that additional excitation of H$_2$ may be provided in regions B and
878: C, with respect to region A, by shocks in supernova remnants
879: (see the more general discussion in Sect.~\ref{agn_shock}).
880: To our knowledge, no independant evidence exists to test the existence of
881: shocks in these regions. \citet{Chandar00} obtained a normal H{\small II}
882: optical line spectrum at the center of NGC\,6822\_C, but since their beam
883: of 2.5 arcsec$^2$ is only about 1\% of ours, we cannot rule out shock excitation.
884: Finally, given the small distance of NGC\,6822, the regions covered by the
885: IRS aperture are less than 100\,pc in size. Greater fluctuations around
886: the average properties are thus not unexpected.
887: At present, we are unable to decide which scenario is the most likely.
888: 
889: 
890: \subsection{Aromatic bands}
891: \label{pah_power}
892: 
893: Figure~\ref{fig:frac_tir}c shows a remarkable constancy of the power ratio
894: of the H$_2$ rotational lines to the aromatic bands. Among the measured dust
895: and gas observables, PAH emission provides the tightest correlation with H$_2$.
896: Observations of photodissociation regions have shown that the emission from aromatic
897: band carriers and from fluorescently-excited H$_2$ just outside photoionized regions
898: are nearly cospatial, with H$_2$ sometimes seen to extend slightly deeper into molecular
899: clouds \citep{Sellgren90, Graham93, Tielens93, Brooks00, Habart03}.
900: Cospatiality might be expected since both species
901: can be excited by FUV photons. Aromatic band carriers can also be excited
902: by lower-energy photons in the ultraviolet and optical, but with smaller
903: absorption cross-sections \citep[see][]{Li01}, so that FUV photons will dominate
904: the excitation whenever massive stars are present. H$_2$ is however dissociated by
905: FUV photons between 11.3 and 13.6\,eV where it is not self-shielded, whereas
906: PAHs survive the absorption of these photons.
907: Therefore, in the case of relatively dense PDRs (associated with molecular clouds),
908: where collisional heating is expected to be the major origin of the H$_2$ rotational
909: lines, H$_2$ emission should peak at slightly higher optical depth than aromatic bands,
910: in the transition layer between atomic and molecular hydrogen, with $A_{\rm V} > 1$.
911: In addition, PAHs probably cannot be excited as deep into molecular clouds
912: as H$_2$, because at sufficiently high densities they will be coagulated
913: onto grain mantles on short timescales \citep{Boulanger90}.
914: If photodissociation regions dominate the excitation of H$_2$, as consistent with
915: the above results, a tight relation between aromatic band emission and
916: rotational H$_2$ emission can arise
917: only if the physical conditions in PDRs, especially the $G_0 / n$ ratio, are
918: relatively uniform, because H$_2$ fluxes and PAH fluxes depend in very different
919: ways on these two parameters. The condition of relatively constant $G_0 / n$ ratios
920: seems verified in the present sample at least for the average emission within
921: kiloparsec-scale regions (see above). Based on the modelling of [C{\small II}] and
922: [O{\small I}] emission, \citet{Malhotra01} proposed that a regulation of $G_0 / n$
923: might be achieved at the scale of individual PDRs by expanding H{\small II} regions
924: in pressure equilibrium with their surrounding PDRs.
925: 
926: A correlation was previously claimed by \citet{Mouri90}
927: between the 3.3\,$\mu$m band and the v=1-0 S(1) line at 2.12\,$\mu$m for a small
928: sample of starburst and Seyfert galaxies. The dominant source that they propose
929: for H$_2$ excitation, following \citet{Moorwood88}, is however not photodissociation
930: regions, but shocks in supernova remnants. Using the shock models of \citet{Kaufman96}
931: to estimate the sum of the S(0) to S(2) transitions (up to 6\%
932: of the mechanical power, assuming that its totality is dissipated
933: in molecular clouds), and the population synthesis model
934: of \citet{Leitherer99} to estimate both the total mechanical power and the FUV luminosity
935: from continuous star formation with a Salpeter initial mass function, shocks alone
936: are in principle able to produce a significant fraction of the observed H$_2$ emission,
937: but only if the efficiency of conversion of mechanical power into H$_2$ emission
938: is unrealistically high. The rotational line ratios are also inconsistent
939: with shock models, which predict higher temperatures ($T > 1000$\,K) except
940: for very low shock velocities (in which case the power fraction radiated away
941: by rotational H$_2$ lines is lower). If the collective rotational line emission
942: from shocks in supernova remnants is similar to that observed in individual objects
943: such as 3C\,391 and IC\,443 \citep{Reach02}, then this mechanism can provide only
944: a modest fraction of the total H$_2$ emission.
945: In addition, if H$_2$ emission came predominantly from supernova
946: remnants whereas aromatic bands arise mostly in photodissociation regions, the partial
947: deconnection between the two, both temporal and spatial, would manifest itself by a large
948: scatter in the observed relation between H$_2$ and PAH fluxes
949: for galaxies with diverse star formation histories, which is not observed.
950: 
951: More recently, \citet{Rigopoulou02} proposed a relation similar to that
952: presented by \citet{Mouri90}, between the 7.7\,$\mu$m aromatic band and
953: the rotational S(1) line in starburst galaxies.
954: Figure~\ref{fig:frac_tir}c not only confirms this result for lower-luminosity galaxies,
955: but also shows that the dispersion for the whole sample of star-forming nuclei is
956: very small, and much smaller with the aromatic bands than with the
957: 24\,$\mu$m emission (Fig.~\ref{fig:frac_tir}d), which is dominated by the continuum
958: from transiently-heated very small grains, as well as from big grains in intense
959: radiation fields. The quantification of the average H$_2$ to dust power ratios
960: and their dispersions is given in the caption of Fig.~\ref{fig:frac_tir}.
961: The energy coupling between aromatic band carriers and H$_2$ strongly suggests
962: that both are excited predominantly in photodissociation regions,
963: although they may not come from the exact same layers (at the same optical depths
964: within the clouds). We present further analysis in the next section.
965: 
966: A similar correlation, with a similarly small dispersion, was observed between
967: the [C{\small II}] line and aromatic band emission \citep{Helou01}. This relation
968: suggests that aromatic band carriers are the source of a major part of gas
969: heating in photodissociation regions, via the photoelectric effect, at least
970: at modest radiation field intensities, since [C{\small II}] emission is the
971: dominant cooling channel in this case.
972: In the narrow range of physical conditions that seem to apply if the emission from
973: H{\small II} nuclei is interpreted in the framework of photodissociation region models
974: (a dynamic range in $G_0 / n$ of only a factor 10), then the same link between aromatic
975: band carriers and H$_2$ would follow if H$_2$ were heated in relatively dense
976: photodissociation regions by the PAHs. Our results, however, suggest that in
977: nearly half the star-forming targets, the dominant excitation mechanism of the
978: rotational levels may be fluorescence in low-density regions, so that
979: ortho-para thermalization is not achieved by collisions
980: (see Sect.~\ref{excitation}). If the lines are fluorescently excited,
981: the cause underlying the tight relation between H$_2$ and aromatic band emission
982: may be that both are proportional to the incident far-ultraviolet flux which excited
983: them.
984: 
985: 
986: \subsection{Cirrus clouds versus PDRs with high radiation field intensities}
987: 
988: The tight association between H$_2$ emission and aromatic bands (Fig.~\ref{fig:frac_tir}c)
989: may be surprising if one assumes that a significant fraction of aromatic band
990: emission arises from diffuse, mostly atomic regions with low radiation field
991: intensities. The infrared emission of such clouds is often termed
992: cirrus \citep{Low84, Terebey86}. If this were the case, then the
993: scaling of PAH flux with H$_2$ flux could be explained only if a constant fraction of
994: the total FUV flux escaped PDRs and were absorbed in the more diffuse
995: interstellar medium. We stress that we adopt here the definition of photodissociation
996: regions stated by \citet{Hollenbach97}: these are not restricted to the interfaces
997: between bright H{\small II} regions and dense molecular clouds, but
998: apply more generally to all the neutral interstellar medium illuminated by
999: FUV photons (with energies between 6 and 13.6\,eV).
1000: Figure~\ref{fig:frac_pah_frac_highU} demonstrates the great
1001: difficulty of making the idea of an important contribution from the cirrus medium
1002: consistent with the data. It shows the flux ratios of
1003: H$_2$ to PAH emission on the one hand, and 24\,$\mu$m to PAH emission on the other
1004: hand, as a function of $P_{24} \sim \nu_{24} F_{24} / (\nu_{71} F_{71} + \nu_{156} F_{156})$,
1005: estimated within the spectroscopic apertures. The quantity $P_{24}$ is closely related
1006: to $f_{U>100} = f(L_{\rm dust}~; U_{\rm rad} > 100)$, derived from the modelling of
1007: the global spectral energy distributions by \citet{Draine07b}, which is the fraction
1008: of the total dust luminosity emitted by regions with radiation field intensities
1009: $U_{\rm rad}$ higher than 100 times the local average value. The dust luminosity
1010: fraction of cirrus clouds,
1011: $\sim f_{U<10}$, can be evaluated as $1 - c~ f_{U>100}$, if one assumes that
1012: $f_{U>100}$ and $f_{U>10}$ are in constant proportion to each other.
1013: Fig.~\ref{fig:frac_pah_frac_highU} shows that while the $F_{24}$/PAH ratio rises
1014: by about one order of magnitude, the H$_2$/PAH ratio is invariant as a function of
1015: $P_{24}$ or $f_{U>100}$.
1016: The results obtained by replacing $P_{24}$ with $F_{71}/F_{156}$, which has the
1017: same physical significance as $F_{60}/F_{100}$, a more traditional indicator
1018: of the relative importance of H{\small II} regions and cirrus clouds
1019: \citep{Helou86}, are identical.
1020: 
1021: We conclude that aromatic bands are mostly associated with photodissociation
1022: regions (illuminated by FUV photons able to provide H$_2$ excitation).
1023: In addition, since PAHs are excited not only by FUV photons but also by
1024: low-energy photons, the observed constancy of the H$_2$ to PAH ratio imposes
1025: some restrictions on possible variations of the radiation field hardness.
1026: Assuming that cirrus clouds, i.e. PDRs with low radiation field intensities,
1027: receive appreciably softer radiation than PDRs with high radiation field
1028: intensities, it would be difficult to understand how both types of regions
1029: could produce similar H$_2$/PAH ratios. As a corollary, the hypothesis that
1030: cirrus clouds could make a large contribution to H$_2$ and PAH emission in
1031: our targets, although not definitely ruled out, is not favored.
1032: Note that the situation may be different in more quiescent parts of galaxies,
1033: not probed by the present sample, and deserves further investigation.
1034: Measurement of H$_2$ line fluxes in quiescent regions is however challenging,
1035: because they depend steeply on the $G_0$ and $n$ parameters.
1036: 
1037: The above does not preclude a large portion of the H$_2$ and PAH emission
1038: to originate in relatively diffuse molecular gas. Estimates of the optical
1039: depth of the $^{12}$CO(1-0) line, over large areas of the Galaxy, indicate
1040: that the total molecular medium comprises a substantial diffuse component
1041: \citep{Polk88}. We will see in the next section that our data, for a portion
1042: of the targets, do support an important contribution from low-density
1043: PDRs to the total warm H$_2$ emission.
1044: 
1045: 
1046: \section{Excitation mechanisms in star-forming regions}
1047: \label{excitation}
1048: 
1049: In sources with purely stellar activity, H$_2$ emission is expected to arise in
1050: varying proportions from
1051: two main energy sources: the far-ultraviolet radiation of OB stars illuminating PDRs;
1052: and shocks in supernova remnants or other sources, providing collisional heating.
1053: In the first case, the excitation can be thermal, by collisions with gas heated
1054: by photoelectrons, or by inelastic collisions with H$_2$ molecules pumped by FUV
1055: photons. The excitation mechanism can also be fluorescence, followed by radiative
1056: cascade to the ground vibrational state. Heating by supernova remnants is unlikely
1057: to be dominant for two reasons, as we have seen in Sect.~\ref{pah_power}. First,
1058: at the low observed temperatures dominating the warm H$_2$ mass, heating would have
1059: to be provided by slow shocks, which are not efficient enough to compete with PDR
1060: excitation. Second, variations in star formation histories within the sample,
1061: which are shown by Moustakas et al. (2007, in prep.) to be very large (from
1062: population synthesis fitting to optical spectra), would produce more scatter
1063: than observed in the H$_2$/PAH ratio. We conclude that H$_2$ is
1064: heated predominantly by PDRs, and this interpretation is supported both by
1065: energetics arguments and by the close association with aromatic band emission
1066: (Sect.~\ref{powers}). In this section, we focus on
1067: additional constraints on the physical conditions and excitation mechanisms
1068: in PDRs (thermal or fluorescent) from the line ratios and excitation diagrams.
1069: 
1070: 
1071: \subsection{Constraints from the temperature distribution}
1072: \label{temp_constraints}
1073: 
1074: We have seen that the H$_2$ to far-infrared ratios are consistent with values
1075: of $G_0 / n$, the ratio of the average radiation field intensity to the hydrogen
1076: density in PDRs, between about 0.1 and 1. In principle, separate constraints on
1077: $G_0$ and on $n$ can be obtained from the temperature distribution reflected in
1078: the H$_2$ line ratios \citep{Kaufman06}.
1079: Given the complexity of the surveyed regions, however, the interpretation of
1080: the line ratios by comparison with models of a single PDR is severely limited.
1081: First, the emission from many distinct PDRs, presumably showing a large range
1082: of physical conditions, is averaged within the beam. Second, even though the
1083: total emission in the sum of the S(0) to S(3) lines is probably dominated by
1084: PDRs, shocks also related to the star formation activity must be present,
1085: driven in particular by supernova remnants, protostellar outflows, or turbulence
1086: dissipation \citep{Falgarone05}. Since these shocks are characterized by higher
1087: rotational temperatures than PDRs, they would contribute mostly to the S(3) line
1088: \citep{Kaufman96}, and in view of the observed line ratios, negligibly to the
1089: lower-lying transitions.
1090: 
1091: In both cases, the superposition of PDRs of different conditions or of shocks
1092: induces a spread in temperatures, and as a consequence these are not reproduced
1093: by single PDR models. We compared the line ratios of the star-forming targets
1094: with $OPR_{\rm \,high\,T} = 3$ to the predictions of \citet{Kaufman06}. Although
1095: the ranges in $G_0$ and $n$ derived from S(2)/S(0) have a broad overlap with
1096: those derived from S(1)/S(0), the S(3)/S(1) ratios are inconsistent, indicating
1097: more intense radiation fields and higher densities. The least biased tracer of
1098: physical conditions in the bulk of the PDRs is thus the S(1)/S(0) ratio (for
1099: galaxies with $OPR_{\rm \,high\,T} = 3$). It suggests that the average $G_0$
1100: varies between about 100 and 5000 (in units of
1101: $1.6 \times 10^{-3}$\,erg\,s$^{-1}$\,cm$^{-2}$)
1102: and the average $n$ between about 500 and $10^4$\,cm$^{-3}$.
1103: 
1104: 
1105: \subsection{Fluorescent excitation}
1106: 
1107: It is surprising that in several objects,
1108: the S(0) to S(3) transitions are not thermalized, as indicated by deviations
1109: from an apparent $OPR_{\rm \,high\,T}$ of 3. $OPR$ values that
1110: are different from the equilibrium value at the temperature of the H$_2$ gas
1111: arise naturally from fluorescent excitation, if the gas density is lower than
1112: the critical density
1113: for ortho-para equilibration by collisions with H and H$^+$ \citep{Sternberg99}.
1114: This is because the ultraviolet
1115: absorption lines have a greater optical depth in the ortho states than in the
1116: para states, so that the ortho states at a given depth are pumped less.
1117: The apparent $OPR_{\rm \,high\,T}$ values only apply to the states excited by
1118: FUV pumping, and do not imply that the true $OPR_{\rm \,high\,T}$ value
1119: (for the J=1 and J=0 states, where most of the gas resides)
1120: be different from 3. For a total $OPR_{\rm \,high\,T}$ of 3, the $OPR_{\rm \,high\,T}$
1121: of the excited levels is predicted to be close to $\sqrt{3} = 1.7$.
1122: A thorough review of the phenomenon of selective excitation
1123: and its implications for the interpretation of excitation diagrams was provided by
1124: \citet{Sternberg99}. Note that radiative decay in the electronic ground state
1125: and most collisional (de)excitations
1126: always occur at fixed spin number and thus preserve
1127: the ortho or para state. Conversion from one state to the other can be accomplished
1128: by H$_2$ dissociation followed by reformation on dust grains, or by reactions
1129: with protons and hydrogen atoms in the gas phase.
1130: 
1131: Deviations from thermalization had previously remained unseen for rotational
1132: lines of extragalactic sources \citep{Rigopoulou02, Higdon06}.
1133: In the frame of PDR excitation, values $OPR_{\rm \,high\,T} < 3$ may be interpreted
1134: as arising in FUV-pumped gas with sufficiently low densities to prevent ortho-para
1135: equilibration by collisions. However, because this process depends on unknown timescales
1136: for gas heating, cooling, dissociation and reformation, we are unable to quantify
1137: in a simple way the implied density conditions.
1138: Alternatively, it is conceivable that the emission comes from initially cold gas,
1139: that has been heated by slow shocks and has not had time to
1140: reach the equilibrium value of $OPR$ \citep{Timmermann98, Wilgenbus00}.
1141: For H{\small II} nuclei, however, we have seen that photodissociation region
1142: excitation is more likely than shock heating (Sect.~\ref{powers}).
1143: 
1144: Another possibility is that in a fraction of the PDRs within the beam, the photodissociation
1145: front is advancing into cold gas (in LTE at time $t_1$ with $OPR(t_1) \ll 3$ and
1146: $OPR_{\rm \,high\,T}(t_1) = 3$), and the recently-heated gas has not yet had enough time
1147: to reach LTE at time $t_2$ (with $OPR(t_2) \sim OPR(t1)$ and $OPR_{\rm \,high\,T}(t_2) < 3$
1148: according to Equ.~\ref{eq:opr}). This is the interpretation
1149: favored by \citet{Fuente99} and \citet{Habart03} to explain the non-LTE ortho to para ratios
1150: observed in the PDRs of NGC\,7023 and $\rho$ Ophiuchi, respectively. In this scenario, the
1151: observed portion of the interstellar medium would have to contain a much larger fraction
1152: of non-equilibrium PDRs in targets with low $OPR_{\rm \,high\,T}$ than in targets with
1153: $OPR_{\rm \,high\,T} = 3$, and the underlying reason would be unclear. It is also unknown
1154: whether the timescales involved in ortho-para equilibration in our sources are long enough
1155: for this scenario to be viable.
1156: 
1157: Figure~\ref{fig:opr_sb} shows the derived $OPR_{\rm \,high\,T}$ values (fixed to 3 whenever
1158: the temperatures derived from each pair of adjacent transitions were compatible
1159: with this assumption) as a function of the total brightness of the S(0) to S(2)
1160: lines. We have here included {\small LINER}s and Seyferts because they do not
1161: display a different behavior in this diagram. Under the hypothesis of photodissociation
1162: region excitation, and assuming first that sites of star formation occupy
1163: a constant fraction of the observing beam, sources with low H$_2$ brightnesses
1164: should consist of regions with both low densities and low radiation field intensities,
1165: while sources with the highest H$_2$ brightnesses should include a greater fraction
1166: of high-density, high-radiation regions. In this simplified view, low values of
1167: $OPR_{\rm \,high\,T}$, indicating that H$_2$ is not thermalized by collisions, could be
1168: obtained only in the low-brightness sources, as seen generally in Fig.~\ref{fig:opr_sb}.
1169: Variations in the beam filling factor by sites of star formation would then induce
1170: a horizontal scatter, which is indeed very large. Fig.~\ref{fig:opr_sb} also shows
1171: $OPR_{\rm \,high\,T}$ as a function of the average surface brightness in the 24\,$\mu$m band.
1172: The latter quantity incorporates a significant contribution from H{\small II} regions,
1173: and should be dominated by variations in radiation field intensity, rather than variations
1174: in gas density. The fact that the horizontal spread is larger in this diagram than
1175: in the diagram involving the H$_2$ brightness supports our tentative interpretation
1176: in terms of density effects.
1177: In addition, $OPR_{\rm \,high\,T}$ does not show any variation as a function of $F_{71}/F_{156}$,
1178: and only a weak tendency to increase with the quantity $P_{24}$ discussed in the
1179: previous section. Both these quantities are indicators of the average radiation
1180: field intensity;
1181: they are observed to be generally correlated with the gas density, but only weakly.
1182: Although the data do not allow us to truly estimate average densities in photodissociation
1183: regions, they suggest that in a substantial number of the observed nuclear regions,
1184: the emission can be dominated by low density gas,
1185: relative to well-studied Galactic PDRs and starburst galaxies. New modelling is required
1186: to quantify the conditions under which rotational lines indicate $OPR_{\rm \,high\,T} < 3$
1187: in PDRs.
1188: 
1189: The H$_2$ line ratios, compared with the PDR model of \citet{Kaufman06}, do not indicate
1190: lower gas densities, on average, in galaxies with $OPR_{\rm \,high\,T} < 3$ than in
1191: galaxies with $OPR_{\rm \,high\,T} = 3$. However, the densities estimated in this
1192: way are averages within the whole beam, and the densest and warmest regions have a
1193: greater weight, because they are more luminous in H$_2$; if a large spread in densities
1194: exists, with both dense clumps and diffuse PDRs, an increase in the diffuse
1195: fraction may not be easily detectable in the average density, while still leaving
1196: an imprint on $OPR_{\rm \,high\,T}$. We note that the proton density should also
1197: play an important role, since ortho-para conversion is effected by collisions
1198: with H and H$^+$. Whether galaxies with $OPR_{\rm \,high\,T} < 3$ have lower
1199: proton densities in molecular clouds, due to reduced ionization by cosmic rays,
1200: is in principle testable with radio continuum observations
1201: of synchrotron radiation from cosmic ray electrons. The number of sample
1202: galaxies with adequate data, at sufficiently high angular resolution, is however
1203: too small to apply this test.
1204: 
1205: We have seen in Section~\ref{extinction} that extinction effects are unlikely
1206: to modify our results in a statistical sense. The fitted $OPR_{\rm \,high\,T}$ values
1207: are correlated with the brightness of the dust emission,
1208: in the aromatic bands, the 24\,$\mu$m continuum and total infrared emission (not
1209: shown here but similar to Fig.~\ref{fig:opr_sb}), which is a further indirect argument
1210: against extinction effects being responsible for low $OPR_{\rm \,high\,T}$ values.
1211: If extinction played a significant role, then low $OPR_{\rm \,high\,T}$ values
1212: would be seen preferentially in bright and compact regions, which is not the case.
1213: 
1214: The galaxies for which we derive the lowest $OPR_{\rm \,high\,T}$ values (and among the lowest
1215: H$_2$ surface brightnesses, regardless of the constraint on the $T_1$ temperature) are
1216: NGC\,337, NGC\,1705 ($OPR_{\rm \,high\,T}$ is a lower limit), NGC\,2915, NGC\,4552
1217: ($OPR_{\rm \,high\,T}$ is a lower limit) and NGC\,7793. The first three are dwarf galaxies,
1218: NGC\,7793 a very late-type spiral and NGC\,4552 is a small elliptical galaxy classified
1219: as {\small LINER}, with the smallest infrared brightness of the whole sample.
1220: Since smaller gas densities are expected in general in dwarf galaxies than in the
1221: central regions of massive galaxies, except in blue compact dwarfs such as Mrk\,33,
1222: this finding is consistent with our interpretation of small $OPR_{\rm \,high\,T}$ values
1223: in terms of low density. We may also remark that in NGC\,1705, which is a starburst,
1224: no ultracompact H{\small II} region was detected in the radio \citep{Johnson03};
1225: thus no conflict exists with the hypothesis that the photodissociation regions
1226: in NGC\,1705 have low densities.
1227: 
1228: 
1229: \section{Excitation mechanisms in {\small LINER} and Seyfert nuclei}
1230: \label{syliner}
1231: 
1232: A large number of the galaxies classified as {\small LINER}s or Seyferts deviate
1233: significantly from the relations discussed in Section~\ref{powers}, in having a
1234: strong excess of H$_2$ emission with respect to all the other tracers used here,
1235: not only aromatic bands but also [Si{\small II}], the 24\,$\mu$m flux and the total
1236: infrared emission (Fig.~\ref{fig:frac_tir}),
1237: arguing for an alternative excitation mechanism in these
1238: galaxies. The average and dispersion of each power ratio are given separately
1239: for H{\small II} nuclei and for {\small LINER} and Sy nuclei in the figure caption.
1240: Given these results, the quantities by which the two categories are most clearly
1241: separated are the H$_2$ to aromatic band ratio and the H$_2$ to [Si{\small II}] ratio.
1242: In particular, nuclear regions with $F{\rm (S0-S2)}/F_{\rm 7.9\,dust} > 10^{-1.94}$
1243: are likely to be of the {\small LINER} or Sy type at the 99\% confidence level.
1244: 
1245: We thus define an excess of H$_2$ emission, with respect to H{\small II} nuclei,
1246: based on the observed relation with aromatic bands. We choose as the maximal H$_2$
1247: power associated purely with star formation the quantity
1248: $10^{-1.94} \times L_{\rm 7.9\,dust}$.
1249: Aromatic band carriers are thought to be destroyed in intense radiation field environments,
1250: such as H{\small II} region cores \citep{Giard94} and ionized regions around Seyfert nuclei
1251: \citep{Desert88, Voit92}. They would
1252: however survive where H$_2$ is not dissociated, so that an enhancement of the H$_2$/PAH
1253: ratio is more likely caused by a genuine excess of H$_2$ emission, in conditions where
1254: PAHs are not excited, rather than normal H$_2$ emission occurring where PAHs would
1255: have been destroyed.
1256: Furthermore, H$_2$ emission is seen in excess not only with respect to aromatic bands,
1257: but also with respect to [Si{\small II}] and other dust tracers (24\,$\mu$m and
1258: total infrared emission).
1259: [Si{\small II}] emission may be depressed because the ionization state of silicon
1260: becomes higher, but the dust continuum cannot be suppressed like aromatic bands.
1261: Our empirical quantification of the H$_2$ excess is intended to extract the part of
1262: H$_2$ emission that cannot originate in photodissociation regions. The excitation
1263: mechanism that is required to account for this excess, while not exciting PAHs, is
1264: either X-ray irradiation or shock heating.
1265: 
1266: We discuss {\small LINER} and Sy nuclei as a single category, because no detectable
1267: difference exists in their H$_2$ properties. This may be expected for several reasons:
1268: small-number statistics; the fact that the classification of low-luminosity AGNs
1269: can be ambiguous as it depends on the aperture size in particular; and the fact
1270: that the source of H$_2$ excitation might not be directly linked to the nuclear
1271: activity, as the results discussed below suggest.
1272: 
1273: 
1274: \subsection{Heating by X-rays from an active galactic nucleus}
1275: 
1276: The idea that a nuclear X-ray source may modify the chemistry and excitation of the
1277: surrounding molecular clouds through sufficiently large column densities as to
1278: offer a convenient way to identify active nuclei hidden by dust has recently
1279: received much attention. Models predict, in particular, unusual ratios of tracers
1280: of dense molecular gas \citep{Lepp96, Meijerink05}, consistent with observations
1281: of NGC\,1068 \citep{Usero04}. X-ray excitation would also manifest itself in the
1282: properties of H$_2$, the most abundant molecule. To test the hypothesis that the
1283: additional excitation in galaxies showing a significant excess of H$_2$ emission
1284: with respect to aromatic bands (Figure~\ref{fig:frac_tir}c) is predominantly
1285: produced by nuclear X-rays, we have compiled estimated X-ray fluxes
1286: in the 2-10\,keV band obtained from Chandra observations. The data and references
1287: are summarized in Table~\ref{tab_xrays}. The H$_2$ excess is shown as a function
1288: of the X-ray luminosity of the nucleus in Fig.~\ref{fig:xrays}. Here, galaxies
1289: with no H$_2$ excess according to our definition are shown below the dashed line.
1290: 
1291: The spread in Figure~\ref{fig:xrays} is very large.
1292: Since the available X-ray measurements
1293: do not isolate the hard X-ray component, but include soft
1294: emission, a substantial part can be thermal emission from supernova remnants,
1295: as opposed to power-law emission from the activity related to a central supermassive
1296: black hole. Many H{\small II} nuclei with no H$_2$ excess indeed have X-ray luminosities
1297: that are comparable to those of {\small LINER} and Seyfert nuclei. On the other hand,
1298: some galaxies, in particular NGC\,3627, NGC\,4569 and NGC\,5866, are very luminous
1299: in H$_2$ but have only modest X-ray to H$_2$ luminosity ratios compared with
1300: other H$_2$-excess galaxies.
1301: We shall assume a power-law spectrum with a standard photon index of $-1.8$,
1302: as adopted by \citet{Ho01}, to extrapolate the total X-ray luminosity to lower
1303: photon energies. Under this assumption, the luminosity between 0.2 and 10\,keV
1304: is only two times the luminosity between 2 and 10\,keV compiled here.
1305: Up to 10\% of the intrinsic 1-10\,keV luminosity may emerge in the sum of all H$_2$
1306: transitions \citep{Lepp83}, whereas in our comparison we have summed only the three
1307: rotational lines S(0) to S(2). Therefore, pure X-ray excitation of the excess H$_2$
1308: emission should not be energetically possible even in objects such as NGC\,4579 and
1309: NGC\,5195, which have relatively low H$_2$ to X-ray luminosity ratios
1310: in Fig.~\ref{fig:xrays}. Even if the intrinsic emission in soft X-rays
1311: were underestimated by a large factor, X-ray excitation would still be
1312: unlikely to dominate in most cases.
1313: The apparent trend of increasing excess H$_2$ luminosities with increasing
1314: X-ray luminosities in Fig.~\ref{fig:xrays} does not imply a direct excitation of
1315: H$_2$ by X-rays, since it could also be understood in the frame of multi-phase
1316: shocks, produced by supernova remnants or by starburst winds (see below).
1317: From near-infrared line ratios, \citet{Davies05} also reached the conclusion that X-rays
1318: do not contribute significantly to the excitation of H$_2$ in a small sample of
1319: active galactic nuclei, within regions smaller than those sampled here (about 100\,pc).
1320: 
1321: Supernova remnants are another important source of X-rays. More than half their
1322: mechanical energy can be converted to X-rays, mostly with energies below $\sim 500$\,eV
1323: \citep{Draine91}, but the X-ray power is converted to H$_2$ emission in the S(0) to S(2)
1324: lines with low efficiencies, less than $10^{-3}$ \citep{Draine90}. Therefore, in view
1325: of the estimates presented in Fig.~\ref{fig:snshocks} (see the next Section for details),
1326: heating of H$_2$ by X-rays from supernova remnants is completely negligible.
1327: 
1328: 
1329: \subsection{Heating by shocks}
1330: \label{agn_shock}
1331: 
1332: For galaxies such as NGC\,3627 and NGC\,4569, with very high H$_2$ to X-ray luminosity
1333: ratios, an efficient mechanism has to be invoked to account for the H$_2$ brightness.
1334: It has been shown recently that galactic shocks can convert a very large fraction
1335: of the kinetic energy into rotational H$_2$ emission without producing a lot of
1336: X-rays \citep{Appleton06}. We thus propose that large-scale shocks play
1337: a major role. NGC\,3627 is an interacting galaxy in the Leo Triplet, characterized
1338: by severe morphological and kinematical distortions and, in the center, a massive
1339: molecular gas concentration in the form of an asymmetric bar-like structure
1340: \citep{Zhang93}, with a peak in the stellar velocity dispersion that is shifted by
1341: about 3\arcsec\ from the nucleus, and probably strong gas inflow into the nucleus
1342: \citep{Afanasiev05}. NGC\,4569 is a starburst {\small LINER} \citep{Maoz98} having
1343: recently produced a large number of supernova explosions, triggering the expansion
1344: of X-ray, H$\alpha$ \citep{Tschoke01} and synchrotron emission lobes \citep{Chyzy06},
1345: and also has a circumnuclear ring of molecular gas with strong non-circular motions
1346: \citep{Nakanishi05}. The situation could be similar to NGC\,4945, where H$_2$ emission
1347: is seen to follow the innermost part of a starburst outflow, with an extent of
1348: $\sim 200$\,pc \citep{Moorwood96}.
1349: These two examples thus support the idea of a dominant excitation by shocks, triggered
1350: by dynamical perturbations and by a starburst wind, respectively.
1351: 
1352: In order to test the possibility that the excess H$_2$ emission be caused by
1353: supernova remnant shocks, we estimated the required heating efficiencies in the
1354: following way. We first computed the star formation rate required to account for the
1355: excess H$_2$ emission, assuming constant star formation, if all the
1356: mechanical power is converted to the power emitted by the sum of the S(0) to S(2)
1357: lines \citep[using the Starburst99 population synthesis model of][]{Leitherer99}.
1358: We then compared this with the star formation rate estimated from the 24\,$\mu$m
1359: luminosity, using the calibration of \citet{Wu05} for the same initial mass function.
1360: The ratio of the two rates gives an order-of-magnitude estimate of the H$_2$ heating
1361: efficiency $\eta$ that is needed if supernova shocks are invoked as the dominant heating
1362: mechanism, and is shown in Fig.~\ref{fig:snshocks} as a function of the excess H$_2$
1363: luminosity. In this simplified computation, the H$_2$ excess in the star-forming
1364: regions B and C of NGC\,6822 can easily be attributed to supernova remnant shocks,
1365: as they would imply efficiencies of at most 3\%. Besides these regions, supernova
1366: remnants are also a sufficient heating source of the excess H$_2$ for at least six
1367: of the {\small LINER} and Sy nuclei. In NGC\,5195, in particular, the necessary
1368: heating efficiency (including the fraction of the total mechanical power injected
1369: and absorbed in molecular clouds) is very small. Galaxies such as NGC\,4450
1370: and NGC\,4579, on the other hand, unambiguously require much more power than
1371: available in supernova remnants. The excitation source could be shocks triggered by
1372: cloud collisions induced by
1373: gravitational perturbations, maybe in combination with X-rays.
1374: 
1375: NGC\,1316 (Fornax\,A) stands out in Fig.~\ref{fig:frac_warm} as having the highest
1376: $T_1$ temperature ($T_1 > 280$\,K) and the lowest mass fraction in warm phase ($< 1$\%),
1377: and the lowest levels up to J=4 are characterized by a single temperature.
1378: In this particular case,
1379: H$_2$ may be heated by fast shocks caused by the powerful jet that has produced
1380: large-scale radio lobes, and that is observed in the form of knots in the inner
1381: kiloparsec, which may be a signature of interaction with the interstellar medium,
1382: i.e. shocks \citep{Geldzahler84}. Excitation by far-ultraviolet radiation and X-rays
1383: or by slow shocks would produce a large quantity of H$_2$ at temperatures
1384: between 100 and 300\,K \citep{Burton92, Maloney96}, which is ruled out by the data.
1385: NGC\,1316 is the only galaxy of the sample for which the excess H$_2$ may be heated
1386: by an AGN jet.
1387: 
1388: Of the 16 H$_2$-excess galaxies, all have excitation diagrams consistent with
1389: $OPR_{\rm \,high\,T} = 3$ (at least in the free-$T_1$ fits),
1390: except NGC\,1266, NGC\,5866 and NGC\,4125.
1391: NGC\,1266 may be unusual simply by virtue of its significant optical depth,
1392: as discussed in Section~\ref{extinction}. This predominance of $OPR_{\rm \,high\,T} = 3$,
1393: combined with more elevated temperatures than in H{\small II} nuclei
1394: on average (Fig.~\ref{fig:frac_warm}) would be consistent with the excess H$_2$
1395: emission originating in shocks where ortho-para equilibration is fast. It remains
1396: however difficult to test this idea, and to identify the source of these shocks.
1397: 
1398: 
1399: \section{Summary and conclusions}
1400: \label{summary}
1401: 
1402: We present the measurements and results of a survey of the four lowest-energy rotational
1403: transitions of H$_2$, S(0) to S(3), in a local sample of 57 galaxies, from the SINGS program.
1404: For three galaxies in this sample, higher-energy transitions, up to S(7), could be measured.
1405: Characterizing the amount and physical conditions of the warm molecular hydrogen phase
1406: traced by these lines is of prime interest, because molecular hydrogen represents a major
1407: mass fraction of the interstellar medium of normal galaxies, and the warm phase itself
1408: (gas heated to temperatures of $\approx 100$ to 1000\,K) can constitute a substantial
1409: fraction of the total H$_2$.
1410: 
1411: The emission is measured over areas of median size 0.9\,kpc, thus including a large number
1412: of distinct star formation sites and molecular clouds. The sample comprises mostly nuclear
1413: regions (in 47 massive galaxies and 9 dwarf galaxies), of which 45\% are optically
1414: classified as {\small LINER} or Seyfert, as well as 10 extranuclear star-formation
1415: complexes within a dwarf galaxy and a spiral.
1416: With respect to earlier studies of molecular hydrogen emission in galaxies, and
1417: particularly rotational lines, which had focussed on very bright systems (nearby
1418: starbursts and AGNs, as well as ultraluminous galaxies), this paper provides results
1419: on the average properties of warm H$_2$ of relatively faint systems, more representative
1420: of the general population of galaxies.
1421: 
1422: Perhaps the most significant observational results (detailed below) are:
1423: (1) the tight correlation of the powers emitted by the sum of the S(0) to S(2)
1424: lines and by aromatic bands, and the fact that the $F$(S0-S2)/PAH ratio
1425: is insensitive to the marked variations in average radiation field intensities
1426: existing in our sample; (2) the existence of non-equilibrium ortho to para
1427: ratios in the rotational levels, that are weakly correlated with the
1428: surface brightness of the H$_2$ lines. These results call for further modelling
1429: in order to be better understood.
1430: 
1431: {\bf Masses and column densities:}
1432: The total masses of warm H$_2$ within our apertures range between $10^5$\,M$_{\sun}$
1433: and close to $3 \times 10^8$\,M$_{\sun}$ in galaxy nuclear regions. In star formation
1434: complexes of nearby dwarf galaxies (NGC\,2915 and NGC\,6822), we probe warm H$_2$ masses
1435: down to a few $10^3$\,M$_{\sun}$ within equivalent diameters of 60 to 250\,pc.
1436: The mass densities range between 0.2 and 30\,M$_{\sun}$\,pc$^{-2}$.
1437: The column densities that we derive are on average of the same order of magnitude
1438: as the column densities observed in individual Galactic photodissociation regions.
1439: For systems in which H$_2$ is predominantly excited by PDRs,
1440: assuming they have similar characteristics to the Orion Bar,
1441: this implies that they fill most of the observing beam.
1442: 
1443: {\bf H$_2$ mass fraction in the warm phase ($\mathbf T \geq 100$\,K):}
1444: Under a conservative assumption about the distribution of temperatures, we find
1445: that the warm H$_2$ gas makes up between 1\% and more than 30\% of the total H$_2$.
1446: For star-forming galaxies, the median mass fraction in the warm phase is 10\%.
1447: The column density nevertheless has a steep inverse dependence on the temperature,
1448: and we cannot rule out that the unconstrained cool H$_2$
1449: component ($70\,{\rm K} < T < 100$\,K) might in some cases dominate the S(0) and
1450: S(1) emission and account for most of the H$_2$ mass.
1451: 
1452: {\bf PDR excitation in star-forming regions:}
1453: In H{\small II} nuclei, we observe a remarkably narrow range of H$_2$ to aromatic
1454: band flux ratio.
1455: This result argues for photodissociation regions providing most of the power used
1456: for H$_2$ excitation, since aromatic bands are known to arise predominantly from
1457: these regions,
1458: as defined by \citet{Hollenbach97}, that include all the neutral interstellar
1459: medium illuminated by FUV photons.
1460: Two main excitation mechanisms can be at work simultaneously in
1461: PDRs: pumping by FUV photons, followed by fluorescent decay to the ground electronic
1462: state, and collisions with hydrogen atoms and molecules heated by photoelectrons
1463: or FUV pumping.
1464: Comparison with the predictions of PDR models for the ratio of H$_2$ to FUV luminosity
1465: indicates a narrow range of average physical conditions, $G_0 / n$ between 0.1 and 1
1466: with the radiation field intensity $G_0$ in units of
1467: $1.6 \times 10^{-3}$\,erg\,s$^{-1}$\,cm$^{-2}$ and the hydrogen nucleus density
1468: $n$ in units of cm$^{-3}$. The sum of the S(0) to S(2) transitions represents
1469: between $2.5 \times 10^{-4}$ and $7.5 \times 10^{-4}$ of the total infrared power,
1470: and on average 30\% of the [Si{\small II}] line power.
1471: The observed temperatures suggest that the average $G_0$ varies between about
1472: 100 and 5000, and the average $n$ between about 500 and $10^4$\,cm$^{-3}$.
1473: This seems to imply that H$_2$ rotational line emission comes mostly
1474: from molecular clouds illuminated by OB associations.
1475: We have seen however that the estimator of these parameters is biased
1476: (see Section~\ref{temp_constraints}),
1477: and a non-negligible contribution from less dense regions with less intense
1478: radiation fields is not excluded.
1479: 
1480: {\bf Evidence for fluorescence in star-forming regions:}
1481: Previous surveys of rotational lines in galaxies had not revealed any departure
1482: from thermalization of H$_2$, which is a consequence of the relatively low
1483: critical densities of the lower rotational transitions. By contrast, we find that
1484: nearly half the targets in our sample deviate significantly
1485: from local thermodynamic equilibrium in having apparent ortho to para ratios
1486: $OPR_{\rm \,high\,T}$ lower than the equilibrium value of three.
1487: We have seen that this result cannot be an artifact caused by extinction effects.
1488: Low values of $OPR_{\rm \,high\,T}$ may thus be interpreted as evidence of fluorescent
1489: excitation, which naturally leads to low ortho to para ratios in the excited states,
1490: occurring in regions of sufficiently low density that
1491: ortho-para equilibration by collisions is incomplete.
1492: The fraction of relatively diffuse molecular gas in normal
1493: galaxies could thus be far from negligible.
1494: In order to test this idea, it would be desirable to obtain independent estimates
1495: of the average gas density in PDRs. In the absence of any robust estimate of this
1496: quantity, we used the surface brightness of the sum of the S(0) to S(2) transitions
1497: as a tracer of the coupled variations of $n$ and $G_0$ to show that the data are
1498: compatible with an interpretation of low apparent $OPR_{\rm \,high\,T}$ values in terms
1499: of low density PDRs. In particular, the lowest values occur preferentially in very
1500: late type galaxies, that may have a more
1501: diffuse interstellar medium than earlier-type galaxies.
1502: In the present sample, close to half the star-forming targets have low $OPR_{\rm \,high\,T}$
1503: values. We infer that fluorescence can be the predominant excitation mechanism of rotational
1504: H$_2$ lines in normal star-forming galaxies. In more active galaxies, however, collisional
1505: excitation in photodissociation regions is likely to overtake fluorescence, even though
1506: the latter still contributes to gas heating.
1507: Alternatively, low $OPR_{\rm \,high\,T}$ values may be caused by non-equilibrium PDRs,
1508: in which initially cold gas, recently reached by the photodissociation front and heated,
1509: has not had enough time to adjust its ortho to para ratio. A disadvantage of this scenario
1510: is that it does not explain why PDRs would have systematically different properties
1511: in targets with $OPR_{\rm \,high\,T} < 3$ and in those with $OPR_{\rm \,high\,T} = 3$.
1512: We thus emphasize that the cause of non-equilibrium ortho to para ratios in the
1513: rotational levels is not well understood at present.
1514: 
1515: {\bf Differences between H{\small II} and {\small LINER}/Sy nuclei:}
1516: Despite our large observing beam implying that the emission from the immediate vicinity
1517: of the nucleus is diluted in the emission from extended areas decoupled from nuclear
1518: activity, a large fraction of nuclei classified as {\small LINER} or Seyfert distinguish
1519: themselves from purely star-forming nuclei in several ways. In a statistical sense,
1520: the temperatures of the warm H$_2$ phase are slightly higher, and as a corollary the
1521: mass fractions of warm to total H$_2$ are lower
1522: (with a median of 4\% instead of 10\% for H{\small II} nuclei).
1523: The correlation between H$_2$ and
1524: aromatic band emission observed in H{\small II} nuclei also breaks down in
1525: {\small LINER} and Sy nuclei. A large number of them have excess emission in the
1526: H$_2$ lines, with respect to aromatic bands and to [Si{\small II}], which
1527: in general is the brightest cooling line in the mid-infrared range (with possible
1528: contributions from both PDRs and H{\small II} regions as well as X-ray irradiated
1529: gas), and, to a lesser degree, with respect to the total infrared emission.
1530: The fact that less contrast between the different nuclear categories is seen
1531: in the H$_2$/TIR ratio may partly be due to the fact that estimating the 70\,$\mu$m
1532: and 160\,$\mu$m fluxes within our small apertures requires a large extrapolation.
1533: 
1534: {\bf Threshold for nonstellar excitation:}
1535: We propose that nuclear ratios $F{\rm (S0-S2)}/F_{\rm 7.9\,dust} > 10^{-1.94}$,
1536: with observables defined as in our study, are indicative of the {\small LINER}
1537: and Sy categories. It should however be kept in mind that some sources belonging
1538: to these classes are indistinguishable from H{\small II} nuclei. This could be
1539: thought of as mainly a distance and beam dilution effect; but the H$_2$/PAH ratio
1540: behaves contrary to this expectation, since in the present sample it shows a
1541: correlation with distance, rather than an anti-correlation.
1542: None of the quantities derived in this paper shows any dependence on distance,
1543: with this exception of the H$_2$ excess in {\small LINER} and Sy nuclei.
1544: The fact that it tends to increase with distance could be a selection bias,
1545: but it also suggests that the H$_2$ excess is in general spatially extended.
1546: We may also remark that for galaxies like NGC\,1377, whose dust emission is
1547: interpreted as dominated by an extremely young and opaque starburst \citep{Roussel06},
1548: the above criterion to select {\small LINER} and Sy nuclei will not be applicable.
1549: 
1550: {\bf Shock excitation in {\small LINER}/Sy nuclei:}
1551: We interpret the differences in H$_2$/PAH ratios as a genuine excess of H$_2$ emission
1552: (and not a deficit of the other tracers while preserving H$_2$), i.e. as requiring
1553: at least one additional mechanism to excite H$_2$ molecules with respect to PDR heating.
1554: Excitation by nuclear X-rays seems implausible, as models predict much lower heating
1555: efficiencies than what would be necessary to account for the estimated H$_2$ excess,
1556: compared with X-ray luminosities derived from Chandra observations.
1557: We thus favor excess heating by large-scale shocks, caused either by the collective
1558: effect of supernov\ae\ in an aging starburst, or by dynamical perturbations.
1559: An order-of-magnitude estimate suggests that supernova remnant shocks can easily
1560: account for the H$_2$ excess of a fraction of the {\small LINER} and Sy nuclei
1561: (for example NGC\,5195), but do not provide enough mechanical power for the galaxies
1562: with the highest H$_2$/PAH ratios.
1563: For the latter, shocks triggered by dynamical perturbations are the best candidate
1564: to supply the excess H$_2$ heating.
1565: In one case, it is conceivable that the excess H$_2$ emission may be produced
1566: by the interaction of a nuclear jet with the interstellar medium, namely in NGC\,1316
1567: (Fornax\,A). This target has the warmest H$_2$ of the whole sample, with no evidence
1568: for a cool ($T < 300$\,K) component, the lowest mass fraction in the warm phase,
1569: and is remarkable for its large-scale radio lobes.
1570: 
1571: {\bf Ortho-para thermalization in {\small LINER}/Sy nuclei:}
1572: Consistent with the hypothesis that the additional H$_2$ may be caused by shocks
1573: in which ortho-para equilibration is fast, the excitation diagrams of most galaxies
1574: with excess H$_2$ emission are consistent with ortho-para
1575: thermalization ($OPR_{\rm \,high\,T} = 3$), with three exceptions: NGC\,1266,
1576: where we have seen that the apparent $OPR_{\rm \,high\,T} < 3$ may be a result of high
1577: optical depth toward H$_2$, NGC\,5866 and NGC\,4125.
1578: Note that the latter two galaxies are observed edge-on, so that $OPR_{\rm \,high\,T} < 3$
1579: could also be an extinction artefact; this would however imply that the optical depth derived
1580: from the silicate bands be underestimated by large factors (see Sect.~\ref{extinction}).
1581: If excess H$_2$ emission originated from shocks with $OPR_{\rm \,high\,T} < 3$ in the
1582: latter two galaxies, it would imply that the gas was initially cold and had not had
1583: time to reach equilibrium.
1584: 
1585: The rotational H$_2$ lines are most often fainter than the forbidden lines
1586: in the same wavelength range, in particular [Si{\small II}], [Ne{\small II}],
1587: [Ne{\small III}] and [S{\small III}] (with no example of an H$_2$-dominated line
1588: spectrum in the present sample), but are among the dominant coolants of
1589: molecular gas and provide important constraints on the excitation of the warm
1590: molecular
1591: interstellar medium, where a large mass fraction of the gas resides in normal
1592: galaxies. The results presented here are assumed to be representative of
1593: moderate-luminosity galaxies of all types, and can serve as a comparison point
1594: for future studies of distant galaxies.
1595: 
1596: 
1597: 
1598: \acknowledgements
1599: Support for this work, part of the Spitzer Space Telescope Legacy Science
1600: Program, was provided by NASA through an award issued by the Jet Propulsion
1601: Laboratory, California Institute of Technology under NASA contract 1407.
1602: We thank Adam Leroy for a useful discussion.
1603: 
1604: 
1605: 
1606: \begin{thebibliography}{}
1607: \bibitem[Aalto et al.(1995)]{Aalto95}
1608:   Aalto, S., Booth, R.S., Black, J.H., \& Johansson, L.E. 1995, \aap, 300, 369 (A95)
1609: \bibitem[Afanasiev \& Sil'chenko(2005)]{Afanasiev05}
1610:   Afanasiev, V.L., \& Sil'chenko, O.K. 2005, \aap, 429, 825
1611: \bibitem[Albrecht et al.(2004)]{Albrecht04}
1612:   Albrecht, M., Chini, R., Kr\"ugel, E., M\"uller, S.A., \& Lemke, R. 2004, \aap, 414, 141 (A04)
1613: \bibitem[Allers et al.(2005)]{Allers05}
1614:   Allers, K.N., Jaffe, D.T., Lacy, J.H., Draine, B.T., \& Richter, M.J. 2005, \apj, 630, 368
1615: \bibitem[Appleton et al.(2006)]{Appleton06}
1616:   Appleton, P.N., Xu, K.C., Reach, W., et al. 2006, \apj, 639, L51
1617: \bibitem[Bajaja et al.(1991)]{Bajaja91}
1618:   Bajaja, E., Krause, M., Wielebinski, R., \& Dettmar, R.J. 1991, \aap, 241, 411 (B91)
1619: \bibitem[Bally \& Lane(1982)]{Bally82}
1620:   Bally, J., \& Lane, A.P. 1982, \apj, 257, 612
1621: \bibitem[Beck et al.(1979)]{Beck79}
1622:   Beck, S.C., Lacy, J.H., \& Geballe, T.R. 1979, \apj, 234, L213
1623: \bibitem[Beckwith et al.(1978)]{Beckwith78}
1624:   Beckwith, S., Gatley, I., \& Persson, S.E. 1978, \apj, 219, L33
1625: \bibitem[Black \& Dalgarno(1976)]{Black76}
1626:   Black, J.H., \& Dalgarno, A. 1976, \apj, 203, 132
1627: \bibitem[B\"oker et al.(2003)]{Boker03}
1628:   B\"oker, T., Lisenfeld, U., \& Schinnerer, E. 2003, \aap, 406, 87 (B03)
1629: \bibitem[Boulanger et al.(1990)]{Boulanger90}
1630:   Boulanger, F., Falgarone, E., Puget, J.L., \& Helou, G. 1990, \apj, 364, 136
1631: \bibitem[Braine et al.(1993)]{Braine93}
1632:   Braine, J., Combes, F., Casoli, F., et al. 1993, \aaps, 97, 887 (B93)
1633: \bibitem[Brooks et al.(2000)]{Brooks00}
1634:   Brooks, K.J., Burton, M.G., Rathborne, J.M., Ashley, M.C., \& Storey, J.W. 2000, MNRAS, 319, 95
1635: \bibitem[Burton et al.(1992)]{Burton92}
1636:   Burton, M.G., Hollenbach, D.J., \& Tielens, A.G. 1992, \apj, 399, 563
1637: \bibitem[Burton et al.(1989)]{Burton89}
1638:   Burton, M.G., Brand, P.W., Geballe, T.R., \& Webster, A.S. 1989, MNRAS, 236, 409
1639: \bibitem[Chandar et al.(2000)]{Chandar00}
1640:   Chandar, R., Bianchi, L., \& Ford, H.C. 2000, AJ, 120, 3088
1641: \bibitem[Chyzy et al.(2006)]{Chyzy06}
1642:   Chyzy, K.T., Soida, M., Bomans, D.J., et al. 2006, \aap, 447, 465
1643: \bibitem[Claussen \& Sahai(1992)]{Claussen92}
1644:   Claussen, M.J, \& Sahai, R. 1992, \aj, 103, 1134 (C92)
1645: \bibitem[Contini et al.(1997)]{Contini97}
1646:   Contini, T., Wozniak, H., Considere, S., \& Davoust, E. 1997, \aap, 324, 41 (C97)
1647: \bibitem[Curran et al.(2001)]{Curran01}
1648:   Curran, S.J., Polatidis, A.G., Aalto, S., \& Booth, R.S. 2001, \aap, 368, 824 (C01)
1649: \bibitem[Dale et al.(2007)]{Dale07}
1650:   Dale, D.A., Gil de Paz, A., Gordon, K.D., et al. 2007, \apj, 655, 863
1651: \bibitem[Dale et al.(2006)]{Dale06}
1652:   Dale, D.A., Smith, J.D., Armus, L., et al. 2006, \apj, 646, 161
1653: \bibitem[Dale \& Helou(2002)]{Dale02}
1654:   Dale, D.A., \& Helou, G. 2002,  \apj, 576, 159
1655: \bibitem[Dame et al.(2001)]{Dame01}
1656:   Dame, T.M., Hartmann, D., \& Thaddeus, P. 2001, \apj, 547, 792
1657: \bibitem[Davies et al.(2005)]{Davies05}
1658:   Davies, R.I., Sternberg, A., Lehnert, M.D., \& Tacconi-Garman, L.E. 2005, \apj, 633, 105
1659: \bibitem[Davies et al.(2003)]{Davies03}
1660:   Davies, R.I., Sternberg, A., Lehnert, M., \& Tacconi-Garman, L.E. 2003, \apj, 597, 907
1661: \bibitem[D\'esert \& Dennefeld(1988)]{Desert88}
1662:   D\'esert, F.X., \& Dennefeld, M. 1988, \aap, 206, 227
1663: \bibitem[Draine \& Li(2007)]{Draine07a}
1664:   Draine, B.T., \& Li, A. 2007, \apj, 657, 810
1665: \bibitem[Draine et al.(2007)]{Draine07b}
1666:   Draine, B.T., Dale, D.A., Bendo, G., et al. 2007, ApJ, in press (astro-ph/0703213)
1667: \bibitem[Draine \& Woods(1991)]{Draine91}
1668:   Draine, B.T., \& Woods, D.T. 1991, \apj, 383, 621
1669: \bibitem[Draine \& Woods(1990)]{Draine90}
1670:   Draine, B.T., \& Woods, D.T. 1990, \apj, 363, 464
1671: \bibitem[Draine et al.(1983)]{Draine83}
1672:   Draine, B.T., Roberge, W.G., \& Dalgarno, A. 1983, \apj, 264, 485
1673: \bibitem[Elfhag et al.(1996)]{Elfhag96}
1674:   Elfhag, T., Booth, R.S., Hoeglund, B., Johansson, L.E., \& Sandqvist, A. 1996, \aaps, 115, 439 (E96)
1675: \bibitem[Falgarone et al.(2005)]{Falgarone05}
1676:   Falgarone, E., Verstraete, L., Pineau des For\^ets, G., \& Hily-Blant, P. 2005, \aap, 433, 997
1677: \bibitem[Fazio et al.(2004)]{Fazio04}
1678:   Fazio, G., Hora, J.L., Allen, L.E., et al. 2004, \apjs, 154, 10
1679: \bibitem[Fuente et al.(1999)]{Fuente99}
1680:   Fuente, A., Martin-Pintado, J., Rodriguez-Fernandez, N.J., Rodriguez-Franco, A., de Vicente, P., \& Kunze, D. 1999, \apj, 518, L45
1681: \bibitem[Gatley et al.(1987)]{Gatley87}
1682:   Gatley, I., Hasegawa, T., Suzuki, H., et al. 1987, \apj, 318, L73
1683: \bibitem[Gatley et al.(1984)]{Gatley84}
1684:   Gatley, I., Beattie, D.H., Lee, T.J., Jones, T.J., \& Hyland, A.R. 1984, MNRAS, 210, 565
1685: \bibitem[Gautier et al.(1976)]{Gautier76}
1686:   Gautier, T.N., Fink, U., Treffers, R.R., \& Larson, H.P. 1976, \apj, 207, L129
1687: \bibitem[Geldzahler \& Fomalont(1984)]{Geldzahler84}
1688:   Geldzahler, B.J., \& Fomalont, E.B. 1984, \aj, 89, 1650
1689: \bibitem[Giard et al.(1994)]{Giard94}
1690:   Giard, M., Bernard, J.P., Lacombe, F., Normand, P., \& Rouan, D. 1994, \aap, 291, 239
1691: \bibitem[Gordon(1991)]{Gordon91}
1692:   Gordon, M.A. 1991, \apj, 371, 563 (G91)
1693: \bibitem[Graham et al.(1993)]{Graham93}
1694:   Graham, J.R., Serabyn, E., Herbst, T.M., et al. 1993, \aj, 105, 250
1695: \bibitem[Habart et al.(2003)]{Habart03}
1696:   Habart, E., Boulanger, F., Verstraete, L., Pineau des For\^ets, G., Falgarone, E., \& Abergel, A. 2003, \aap, 397, 623
1697: \bibitem[Helfer et al.(2003)]{Helfer03}
1698:   Helfer, T.T., Thornley, M.D., Regan, M.W., et al. 2003, \apjs, 145, 259
1699: \bibitem[Helfer & Blitz(1993)]{Helfer93}
1700:   Helfer, T.T., \& Blitz, L. 1993, \apj, 419, 86 (H93)
1701: \bibitem[Helou et al.(2004)]{Helou04}
1702:   Helou, G., Roussel, H., Appleton P., et al. 2004, \apjs, 154, 253
1703: \bibitem[Helou et al.(2001)]{Helou01}
1704:   Helou, G., Malhotra, S., Hollenbach, D.J., Dale, D.A., \& Contursi, A. 2001, \apj, 548, L73
1705: \bibitem[Helou et al.(1988)]{Helou88}
1706:   Helou, G., Khan, I.R., Malek, L., \& Boehmer, L. 1988, \apjs, 68, 151
1707: \bibitem[Helou(1986)]{Helou86}
1708:   Helou, G. 1986, \apj, 311, L33
1709: \bibitem[Herbst et al.(1996)]{Herbst96}
1710:   Herbst, T.M., Beckwith, S.V., Glindemann, A., Tacconi-Garman, L.E., Kroker, H., \& Krabbe, A. 1996, \aj, 111, 2403
1711: \bibitem[Herbst et al.(1990)]{Herbst90}
1712:   Herbst, T.M., Graham., J.R., Tsutsui, K., Beckwith, S., Matthews, K., \& Soifer, B.T. 1990, \aj, 99, 1773
1713: \bibitem[Higdon et al.(2006)]{Higdon06}
1714:   Higdon, S.J., Armus, L., Higdon, J.L., Soifer, B.T., \& Spoon, H.W. 2006, \apj, 648, 323
1715: \bibitem[Ho et al.(2001)]{Ho01}
1716:   Ho, L.C., Feigelson, E.D., Townsley, L.K., Sambruna, R.M., Garmire, G.P., Brandt, W.N., et al. 2001, \apj, 549, L51 (HF01)
1717: \bibitem[Ho et al.(1997)]{Ho97}
1718:   Ho, L.C., Filippenko, A.V., \& Sargent, W.L. 1997, \apjs, 112, 315
1719: \bibitem[Hollenbach \& Tielens(1997)]{Hollenbach97}
1720:   Hollenbach, D.J., \& Tielens, A.G. 1997, \araa, 35, 179
1721: \bibitem[Horellou et al.(2001)]{Horellou01}
1722:   Horellou, C., Black, J.H., van Gorkom, J.H., Combes, F., van der Hulst, J.M., \& Charmandaris, V. 2001, \aap, 376, 837 (H01)
1723: \bibitem[Houck et al.(2004)]{Houck04}
1724:   Houck, J.R., Roellig, T.L., van Cleve, J., et al. 2004, \apjs, 154, 18
1725: \bibitem[Hubble(1925)]{Hubble25}
1726:   Hubble, E. 1925, \apj, 62, 409
1727: \bibitem[Huber \& Herzberg(1979)]{Huber79}
1728:   Huber, K.P., \& Herzberg, G. 1979, Constants of diatomic molecules, ed. van Nostrand
1729: \bibitem[Israel(2005)]{Israel05}
1730:   Israel, F.P. 2005, \aap, 438, 855 (I05)
1731: \bibitem[Israel et al.(2003)]{Israel03}
1732:   Israel, F.P., Baas, F., Rudy, R.J., Skillman, E.D., \& Woodward, C.E. 2003, \aap, 397, 87
1733: \bibitem[Israel \& Baas(1999)]{Israel99}
1734:   Israel, F.P., \& Baas, F. 1999, \aap, 351, 10 (I99)
1735: \bibitem[Israel et al.(1995)]{Israel95}
1736:   Israel, F.P., Tacconi, L.J., \& Baas, F. 1995, \aap, 295, 599 (I95)
1737: \bibitem[Johnson et al.(2003)]{Johnson03}
1738:   Johnson, K.E., Indebetouw, R., \& Pisano, D.J. 2003, \aj, 126, 101
1739: \bibitem[Kaufman et al.(2006)]{Kaufman06}
1740:   Kaufman, M.J., Wolfire, M.G., \& Hollenbach, D.J. 2006, \apj, 644, 283
1741: \bibitem[Kaufman \& Neufeld(1996)]{Kaufman96}
1742:   Kaufman, M.J., \& Neufeld, D.A. 1996, \apj, 456, 611
1743: \bibitem[Kenney \& Young(1988)]{Kenney88}
1744:   Kenney, J.D., \& Young, J.S. 1988, \apjs, 66, 261 (K88)
1745: \bibitem[Kennicutt et al.(2003)]{Kennicutt03}
1746:   Kennicutt, R.C., et al. 2003, PASP 115, 928
1747: \bibitem[Kilgard et al.(2005)]{Kilgard05}
1748:   Kilgard, R.E., Cowan, J.J., Garcia, M.R., Kaaret, P., Krauss, M.I., McDowell, J.C., et al. 2005, \apjs, 159, 214 (K05)
1749: \bibitem[Kim \& Fabbiano(2003)]{Kim03}
1750:   Kim, D.W., \& Fabbiano, G. 2003, \apj, 586, 826 (KF03)
1751: \bibitem[Kinman et al.(1979)]{Kinman79}
1752:   Kinman, T.D., Green, J.R., \& Mahaffey, C.T. 1979, PASP, 91, 749
1753: \bibitem[Knapen et al.(1996)]{Knapen96}
1754:   Knapen, J.H., Beckman, J.E., Cepa, J., \& Nakai, N. 1996, \aap, 308, 27 (K96)
1755: \bibitem[Knop et al.(2001)]{Knop01}
1756:   Knop, R.A., Armus, L., Matthews, K., Murphy, T.W., \& Soifer, B.T. 2001, \apj, 122, 764
1757: \bibitem[Kohno et al.(2003)]{Kohno03}
1758:   Kohno, K., Ishizuki, S., Matsushita, S., Vila-Vilar\'o, B., \& Kawabe, R. 2003, PASJ, 55, L1 (K03)
1759: \bibitem[Kohno et al.(2002)]{Kohno02}
1760:   Kohno, K., Tosaki, T., Matsushita, S., Vila-Vila\'o, B., Shibatsuka, T., \& Kawabe, R. 2002, PASJ, 54, 541 (K02)
1761: \bibitem[Larkin et al.(1998)]{Larkin98}
1762:   Larkin, J.E., Armus, L., Knop, R.A., Soifer, B.T., \& Matthews, K. 1998, \apjs, 114, 59
1763: \bibitem[Lee et al.(2006)]{Lee06}
1764:   Lee, H., Skillman, E.D., Cannon, J.M., et al. 2006, \apj, 647, 970
1765: \bibitem[Leitherer et al.(1999)]{Leitherer99}
1766:   Leitherer, C., Schaerer, D., Goldader, J.D., et al. 1999, \apjs, 123, 3
1767: \bibitem[Leon et al.(1998)]{Leon98}
1768:   Leon, S., Combes, F., \& Menon, T.K. 1998, \aap, 330, 37 (L98)
1769: \bibitem[Lepp \& Dalgarno(1996)]{Lepp96}
1770:   Lepp, S., \& Dalgarno, A. 1996, \aap, 306, L21
1771: \bibitem[Lepp \& McCray(1983)]{Lepp83}
1772:   Lepp, S., \& McCray, R. 1983, \apj, 269, 560
1773: \bibitem[Leroy et al.(2005)]{Leroy05}
1774:   Leroy, A., Bolatto, A.D., Simon, J.D., \& Blitz, L. 2005, \apj, 625, 763 (L05)
1775: \bibitem[Li \& Draine(2001)]{Li01}
1776:   Li, A., \& Draine, B.T. 2001, \apj, 554, 778
1777: \bibitem[Low et al.(1984)]{Low84}
1778:   Low, F.J., Young, E., Beintema, D.A., et al. 1984, \apj, 278, L19
1779: \bibitem[Lutz et al.(2000)]{Lutz00}
1780:   Lutz, D., Sturm, E., Genzel, R., et al. 2000, \apj, 536, 697
1781: \bibitem[Malhotra et al.(2001)]{Malhotra01}
1782:   Malhotra, S., Kaufman, M.J., Hollenbach, D., et al. 2001, \apj, 561, 766
1783: \bibitem[Maloney et al.(1996)]{Maloney96}
1784:   Maloney, P.R., Hollenbach, D.J., \& Tielens, A.G. 1996, \apj, 466, 561
1785: \bibitem[Maloney \& Black(1988)]{Maloney88}
1786:   Maloney, P., \& Black, J.H. 1988, \apj, 325, 389
1787: \bibitem[Mandy \& Martin(1993)]{Mandy93}
1788:   Mandy, M.E., \& Martin, P.G. 1993, \apjs, 86, 199
1789: \bibitem[Maoz et al.(1998)]{Maoz98}
1790:   Maoz, D., Koratkar, A., Shields, J.C., Ho, L.C., Filippenko, A.V., \& Sternberg, A. 1998, \aj, 116, 55
1791: \bibitem[Mauersberger et al.(1999)]{Mauersberger99}
1792:   Mauersberger, R., Henkel, C., Walsh, W., \& Schulz, A. 1999, \aap, 341, 256 (M99)
1793: \bibitem[Meijerink \& Spaans(2005)]{Meijerink05}
1794:   Meijerink, R., \& Spaans, M. 2005, \aap, 436, 397
1795: \bibitem[Moneti et al.(2001)]{Moneti01}
1796:   Moneti, A., Stolovy, S., Blommaert, J.A., Figer, D.F., \& Najarro, F. 2001, \aap, 366, 106
1797: \bibitem[Moorwood et al.(1996)]{Moorwood96}
1798:   Moorwood, A.F., van der Werf, P.P., Kotilainen, J.K., Marconi, A., \& Oliva, E. 1996, \aap, 308, L1
1799: \bibitem[Moorwood \& Oliva(1988)]{Moorwood88}
1800:   Moorwood, A.F., \& Oliva, E. 1988, \aap, 203, 278
1801: \bibitem[Mouri et al.(1990)]{Mouri90}
1802:   Mouri, H., Kawara, K., Taniguchi, Y., \& Nishida, M. 1990, \apj, 356, L39
1803: \bibitem[Nakanishi et al.(2005)]{Nakanishi05}
1804:   Nakanishi, H., Sofue, Y., \& Koda, J. 2005, PASJ, 57, 905
1805: \bibitem[Nemmen et al.(2006)]{Nemmen06}
1806:   Nemmen, R.S., Storchi-Bergmann, T., Yuan, F., Eracleous, M., Terashima, Y., \& Wilson, A.S. 2006, \apj, 643, 652 (N06)
1807: \bibitem[Neufeld \& Kaufman(1993)]{Neufeld93}
1808:   Neufeld, D.A., \& Kaufman, M.J. 1993, \apj, 418, 263
1809: \bibitem[Nishiyama \& Nakai(2001)]{Nishiyama01}
1810:   Nishiyama, K., \& Nakai, N. 2001, PASJ, 53, 713 (N01)
1811: \bibitem[Pagel et al.(1980)]{Pagel80}
1812:   Pagel, B.E., Edmunds, M.G., \& Smith, G. 1980, MNRAS, 193, 219
1813: \bibitem[Parmar et al.(1991)]{Parmar91}
1814:   Parmar, P.S., Lacy, J.H., \& Achtermann, J.M. 1991, \apj, 372, L25
1815: \bibitem[Planesas et al.(1997)]{Planesas97}
1816:   Planesas, P., Colina, L., \& Perez-Olea, D. 1997, \aap, 325, 81 (P97)
1817: \bibitem[Polk et al.(1988)]{Polk88}
1818:   Polk, K.S., Knapp, G.R., Stark, A.A., \& Wilson, R.W. 1988, \apj, 332, 432
1819: \bibitem[Puxley et al.(1988)]{Puxley88}
1820:   Puxley, P.J., Hawarden, T.G., \& Mountain, C.M. 1988, MNRAS, 234, 29
1821: \bibitem[Quillen et al.(1999)]{Quillen99}
1822:   Quillen, A.C., Alonso-Herrero, A., Rieke, M.J., Rieke, G.H., Ruiz, M., \& Kulkarni, V. 1999, \apj, 527, 696
1823: \bibitem[Reach et al.(2002)]{Reach02}
1824:   Reach, W.T., Rho, J., Jarrett, T.H., \& Lagage, P.O. 2002, \apj, 564, 302
1825: \bibitem[Rieke et al.(2004)]{Rieke04}
1826:   Rieke, G., Young, E.T., Engelbracht, C.W., et al. 2004, \apjs, 154, 25
1827: \bibitem[Rigopoulou et al.(2002)]{Rigopoulou02}
1828:   Rigopoulou, D., Kunze, D., Lutz, D., Genzel, R., \& Moorwood, A.F. 2002, \aap, 389, 374
1829: \bibitem[Rotaciuc et al.(1991)]{Rotaciuc91}
1830:   Rotaciuc, V., Krabbe, A., Cameron, M., et al. 1991, \apj, 370, L23
1831: \bibitem[Roussel et al.(2006)]{Roussel06}
1832:   Roussel, H., Helou, G., Smith, J.D., et al. 2006, \apj, 646, 841
1833: \bibitem[Sage(1993)]{Sage93}
1834:   Sage, L.J. 1993, \aap, 272, 123 (S93)
1835: \bibitem[Sage et al.(1992)]{Sage92}
1836:   Sage, L.J., Salzer, J.J., Loose, H.H., \& Henkel, C. 1992, \aap, 265, 19 (S92)
1837: \bibitem[Sage(1989)]{Sage89}
1838:   Sage, L.J. 1989, \apj, 344, 200 (S89)
1839: \bibitem[Sakamoto et al.(1999)]{Sakamoto99}
1840:   Sakamoto, K., Okumura, S.K., Ishizuki, S., \& Scoville, N.Z. 1999, \apjs, 124, 403 (S99)
1841: \bibitem[Satyapal et al.(2005)]{Satyapal05}
1842:   Satyapal, S., Dudik, R.P., O'Halloran, B., \& Gliozzi, M. 2005, \apj, 633, 86 (SD05)
1843: \bibitem[Satyapal et al.(2004)]{Satyapal04}
1844:   Satyapal, S., Sambruna, R.M., \& Dudik, R.P. 2004, \aap, 414, 825 (S04)
1845: \bibitem[Sellgren et al.(1990)]{Sellgren90}
1846:   Sellgren, K., Tokunaga, A.T., \& Nakada, Y. 1990, \apj, 349, 120
1847: \bibitem[Sheth et al.(2005)]{Sheth05}
1848:   Sheth, K., Vogel, S.N., Regan, M.W., Thornley, M.D., \& Teuben, P.J. 2005, \apj, 632, 217 (S05)
1849: \bibitem[Shioya et al.(1998)]{Shioya98}
1850:   Shioya, Y., Tosaki, T., Ohyama, Y., et al. 1998, PASJ, 50, 317 (S98)
1851: \bibitem[Shull \& Beckwith(1982)]{Shull82}
1852:   Shull, J.M., \& Beckwith, S. 1982, \araa, 20, 163
1853: \bibitem[Shull \& Hollenbach(1978)]{Shull78}
1854:   Shull, J.M., \& Hollenbach, D.J. 1978, \apj, 220, 525
1855: \bibitem[Smith et al.(2007a)]{Smith07a}
1856:   Smith, J.D., Armus, L., Dale, D.A., Roussel, H., Sheth, K., Buckalew, B.A., Helou, G., \& Kennicutt, R.C. 2007a, submitted to PASP
1857: \bibitem[Smith et al.(2007b)]{Smith07b}
1858:   Smith, J.D., Draine, B.T., Dale, D.A., et al. 2007b, \apj, 656, 770
1859: \bibitem[Smith et al.(2004)]{Smith04}
1860:   Smith, J.D., Dale, D.A., Armus, L., et al. 2004, \apjs, 154, 199
1861: \bibitem[Sofue et al.(2003)]{Sofue03}
1862:   Sofue, Y., Koda, J., Nakanishi, H., et al. 2003, PASJ, 55, 17 (S03)
1863: \bibitem[Sternberg \& Neufeld(1999)]{Sternberg99}
1864:   Sternberg, A., \& Neufeld, D.A. 1999, \apj, 516, 371
1865: \bibitem[Sternberg \& Dalgarno(1989)]{Sternberg89}
1866:   Sternberg, A., \& Dalgarno, A. 1989, \apj, 338, 197
1867: \bibitem[Strong et al.(1988)]{Strong88}
1868:   Strong, A.W., Bloemen, J.B., Dame, T.M., et al. 1988, \aap, 207, 1
1869: \bibitem[Sugai et al.(1997)]{Sugai97}
1870:   Sugai, H., Malkan, M.A., Ward, M.J., Davies, R.I., \& McLean, I.S. 1997, \apj, 481, 186
1871: \bibitem[Tacconi et al.(1991)]{Tacconi91}
1872:   Tacconi, L.J., Tacconi-Garman, L.E., Thornley, M., \& van Woerden, H. 1991, \aap, 252, 541 (T91)
1873: \bibitem[Tanaka et al.(1989)]{Tanaka89}
1874:   Tanaka, M., Hasegawa, T., Hayashi, S.S., Brand, P.W., \& Gatley, I. 1989, \apj, 336, 207
1875: \bibitem[Taniguchi et al.(1994)]{Taniguchi94}
1876:   Taniguchi, Y., Murayama, T., Nakai, N., Suzuki, M., \& Kameya, O. 1994, \aj, 108, 468 (T94)
1877: \bibitem[Terebey \& Fich(1986)]{Terebey86}
1878:   Terebey, S., \& Fich, M. 1986, \apj, 309, L73
1879: \bibitem[Thompson et al.(1978)]{Thompson78}
1880:   Thompson, R.I., Lebofsky, M.J., \& Rieke, G.H. 1978, \apj, 222, L49
1881: \bibitem[Tielens et al.(1993)]{Tielens93}
1882:   Tielens, A.G., Meixner, M.M., van der Werf, P.P., et al. 1993, Science, 262, 86
1883: \bibitem[Tielens \& Hollenbach(1985)]{Tielens85}
1884:   Tielens, A.G., \& Hollenbach, D. 1985, \apj, 291, 722
1885: \bibitem[Timmermann(1998)]{Timmermann98}
1886:   Timmermann, R. 1998, \apj, 498, 246
1887: \bibitem[Treffers(1979)]{Treffers79}
1888:   Treffers, R.R. 1979, \apj, 233, L17
1889: \bibitem[Treffers et al.(1976)]{Treffers76}
1890:   Treffers, R.R., Fink, U., Larson, H.P., \& Gautier, T.N. 1976, \apj, 209, 793
1891: \bibitem[Tsch\"oke et al.(2001)]{Tschoke01}
1892:   Tsch\"oke, D., Bomans, D.J., Hensler, G. \& Junkes, N. 2001, \aap, 380, 40
1893: \bibitem[Usero et al.(2004)]{Usero04}
1894:   Usero, A., Garc\'{\i}a-Burillo, S., Fuente, A., Mart\'{\i}n-Pintado, J., \& Rodr\'{\i}guez-Fern\'andez, N.J. 2004, \aap, 419, 897
1895: \bibitem[Valentijn et al.(1996)]{Valentijn96}
1896:   Valentijn E.A., van der Werf, P.P., de Graauw, T., \& de Jong, T. 1996, \aap, 315, L145
1897: \bibitem[Vila-Vilar\'o et al.(1998)]{Vila98}
1898:   Vila-Vilar\'o, B., Taniguchi, Y., \& Nakai, N. 1998, \aj, 116, 1553 (VV98)
1899: \bibitem[Voit(1992)]{Voit92}
1900:   Voit, G.M. 1992, MNRAS, 258, 841
1901: \bibitem[Welch \& Sage(2003)]{Welch03}
1902:   Welch, G.A., \& Sage, L.J. 2003, ApJ, 584, 260 (W03)
1903: \bibitem[Wiklind et al.(1995)]{Wiklind95}
1904:   Wiklind, T., Combes, F., \& Henkel, C. 1995, \aap, 297, 643 (W95)
1905: \bibitem[Wilgenbus et al.(2000)]{Wilgenbus00}
1906:   Wilgenbus, D., Cabrit, S., Pineau des For\^ets, G., \& Flower, D.R. 2000, \aap, 356, 1010
1907: \bibitem[Wu et al.(2005)]{Wu05}
1908:   Wu, H., Cao, C., Hao, C.N., et al. 2005, \apj, 632, L79
1909: \bibitem[Yao et al.(2003)]{Yao03}
1910:   Yao, L., Seaquist, E.R., Kuno, N., \& Dunne, L. 2003, \apj, 588, 771 (Y03)
1911: \bibitem[Young et al.(1995)]{Young95}
1912:   Young, J.S., Xie, S., Tacconi, L., et al. 1995, \apjs, 98, 219 (Y95)
1913: \bibitem[Zhang et al.(1993)]{Zhang93}
1914:   Zhang, X., Wright, M., \& Alexander, P. 1993, \apj, 418, 100
1915: \end{thebibliography}
1916: 
1917: 
1918: \clearpage
1919: 
1920: 
1921: \begin{deluxetable}{llrlrr}
1922: \tablecaption{Observed H$_2$ lines.\tablenotemark{a}
1923: \label{tab_lines}
1924: }
1925: \tablehead{
1926: transition & short    & rest $\lambda$ & spectral & E$_{\rm u}$/k & A \\
1927: v=0        & notation & ($\mu$m)       & order    & (K)           & (10$^{-11}$\,s$^{-1}$)\\
1928: }
1929: \startdata
1930: J=2-0 & S(0) & 28.219 & LH\,14 &  510 & 2.95 \\
1931: J=3-1 & S(1) & 17.035 & SH\,12 & 1015 & 47.6 \\
1932: J=4-2 & S(2) & 12.279 & SH\,17 & 1681 & 275. \\
1933: J=5-3 & S(3) &  9.665 & SL\,1  & 2503 & 980. \\
1934: J=6-4 & S(4) &  8.025 & SL\,1  & 3473 & 2640. \\
1935: J=7-5 & S(5) &  6.910 & SL\,2  & 4585 & 5880. \\
1936: J=8-6 & S(6) &  6.109 & SL\,2  & 5828 & 11400. \\
1937: J=9-7 & S(7) &  5.511 & SL\,2  & 7196 & 20000. \\
1938: \enddata
1939: \tablenotetext{a}{The rotational upper level energies were computed from the molecular
1940: constants given by \citet{Huber79} and the transition probabilities are from \citet{Black76}.}
1941: \end{deluxetable}
1942: 
1943: 
1944: 
1945: \begin{deluxetable}{lrlcl}
1946: \tabletypesize{\footnotesize}
1947: \tablecaption{Targets.\tablenotemark{a}
1948: \label{tab_target}
1949: }
1950: \tablehead{
1951: galaxy & D     & class & center (J2000)   & solid angle \\
1952: ~      & (Mpc) & ~     & RA~~~~~~~~~~~DEC & (arcsec$^2$) \\
1953: }
1954: \startdata
1955: N24   &  8.2 & dwarf & 00 09 56.31 $-$24 57 52.0 & 294. \\
1956: N337  & 24.7 & nuc   & 00 59 49.99 $-$07 34 42.5 & 289. \\
1957: N628  & 11.4 & nuc   & 01 36 41.72 +15 46 59.4 & 300. \\
1958: N855  &  9.6 & dwarf & 02 14 03.64 +27 52 40.6 & 288. \\
1959: N925  & 10.1 & nuc   & 02 27 17.06 +33 34 43.8 & 287. \\
1960: N1097 & 16.9 & nuc   & 02 46 18.77 $-$30 16 30.0 & 809. \\
1961: N1266 & 31.3 & liner & 03 16 00.68 $-$02 25 38.8 & 287. \\
1962: N1291 &  9.7 & liner & 03 17 18.50 $-$41 06 27.7 & 287. \\
1963: N1316 & 26.3 & liner & 03 22 41.61 $-$37 12 28.7 & 287. \\
1964: N1482 & 22.0 & nuc   & 03 54 38.68 $-$20 30 08.5 & 798. \\
1965: N1512 & 10.4 & nuc   & 04 03 53.96 $-$43 20 55.6 & 831. \\
1966: N1566 & 18.0 & sy    & 04 20 00.53 $-$54 56 17.4 & 306. \\
1967: N1705 &  5.8 & dwarf & 04 54 13.26 $-$53 21 39.2 & 802. \\
1968: N2403 &  3.5 & nuc   & 07 36 50.25 +65 36 04.6 & 287. \\
1969: N2798 & 24.7 & nuc   & 09 17 22.99 +42 00 00.8 & 293. \\
1970: N2841 &  9.8 & liner & 09 22 02.75 +50 58 35.5 & 291. \\
1971: N2915 &  2.7 & dwarf & 09 26 11.59 $-$76 37 34.2 & 291. \\
1972: N2976 &  3.5 & dwarf & 09 47 15.63 +67 55 00.3 & 289. \\
1973: N3031 &  3.5 & sy    & 09 55 33.50 +69 03 55.8 & 296. \\
1974: N3049 & 19.6 & nuc   & 09 54 49.66 +09 16 19.2 & 320. \\
1975: N3184 &  8.6 & nuc   & 10 18 17.03 +41 25 28.3 & 287. \\
1976: N3190 & 17.4 & liner & 10 18 05.71 +21 49 56.8 & 304. \\
1977: N3198 &  9.8 & nuc   & 10 19 55.10 +45 32 59.6 & 287. \\
1978: N3265 & 20.0 & nuc   & 10 31 06.85 +28 47 49.1 & 301. \\
1979: Mrk33 & 21.7 & dwarf & 10 32 32.02 +54 24 05.4 & 315. \\
1980: N3351 &  9.3 & nuc   & 10 43 57.83 +11 42 10.9 & 761. \\
1981: N3521 &  9.0 & liner & 11 05 48.51 $-$00 02 10.1 & 285. \\
1982: N3621 &  6.2 & liner & 11 18 16.62 $-$32 48 49.9 & 291. \\
1983: N3627 &  8.9 & sy    & 11 20 15.04 +12 59 31.0 & 301. \\
1984: N3773 & 12.9 & dwarf & 11 38 12.91 +12 06 43.0 & 285. \\
1985: N3938 & 12.2 & nuc   & 11 52 49.45 +44 07 15.0 & 292. \\
1986: N4125 & 21.4 & liner & 12 08 05.73 +65 10 27.9 & 294. \\
1987: N4254 & 20.0 & nuc   & 12 18 49.71 +14 25 01.4 & 299. \\
1988: N4321 & 20.0 & nuc   & 12 22 55.08 +15 49 18.2 & 795. \\
1989: N4450 & 20.0 & liner & 12 28 29.51 +17 05 04.9 & 285. \\
1990: N4536 & 25.0 & nuc   & 12 34 27.22 +02 11 14.3 & 777. \\
1991: N4552 & 20.0 & liner & 12 35 39.71 +12 33 21.9 & 285. \\
1992: N4559 & 11.6 & nuc   & 12 35 57.79 +27 57 36.7 & 295. \\
1993: N4569 & 20.0 & liner & 12 36 49.91 +13 09 46.8 & 287. \\
1994: N4579 & 20.0 & sy    & 12 37 43.70 +11 49 07.1 & 295. \\
1995: N4594 & 13.7 & liner & 12 39 59.58 $-$11 37 22.3 & 289. \\
1996: N4625 &  9.5 & dwarf & 12 41 52.65 +41 16 26.5 & 294. \\
1997: N4631 &  9.0 & nuc   & 12 42 07.84 +32 32 34.8 & 282. \\
1998: N4725 & 17.1 & sy    & 12 50 26.71 +25 30 03.2 & 294. \\
1999: N4736 &  5.3 & liner & 12 50 53.23 +41 07 13.6 & 298. \\
2000: N4826 &  5.6 & liner & 12 56 43.79 +21 41 00.7 & 292. \\
2001: N5033 & 13.3 & sy    & 13 13 27.67 +36 35 38.2 & 311. \\
2002: N5055 &  8.2 & liner & 13 15 49.50 +42 01 46.6 & 313. \\
2003: N5194 &  8.2 & sy    & 13 29 52.94 +47 11 44.2 & 330. \\
2004: N5194\_A\tablenotemark{b}
2005:       &  8.2 & hii   & 13 29 49.54 +47 13 28.6 & 256. \\
2006: N5194\_B\tablenotemark{b}
2007:       &  8.2 & hii   & 13 30 01.69 +47 12 50.9 & 271. \\
2008: N5194\_C\tablenotemark{b}
2009:       &  8.2 & hii   & 13 30 00.02 +47 11 11.9 & 352. \\
2010: N5194\_D\tablenotemark{b}
2011:       &  8.2 & hii   & 13 30 02.76 +47 09 53.0 & 338. \\
2012: N5194\_E\tablenotemark{b}
2013:       &  8.2 & hii   & 13 29 56.84 +47 10 45.9 & 234. \\
2014: N5194\_F\tablenotemark{b}
2015:       &  8.2 & hii   & 13 29 52.93 +47 12 38.7 & 297. \\
2016: N5194\_G\tablenotemark{b}
2017:       &  8.2 & hii   & 13 29 43.99 +47 10 20.7 & 285. \\
2018: N5195 &  8.2 & liner & 13 29 59.80 +47 16 00.1 & 323. \\
2019: N5713 & 26.6 & nuc   & 14 40 11.57 $-$00 17 19.1 & 292. \\
2020: N5866 & 12.5 & liner & 15 06 29.58 +55 45 46.7 & 286. \\
2021: N6822\_A\tablenotemark{c}
2022:       &  0.6 & dwarf & 19 44 52.84 $-$14 43 09.8 & 308. \\
2023: N6822\_B\tablenotemark{c}
2024:       &  0.6 & dwarf & 19 44 50.57 $-$14 52 49.0 & 288. \\
2025: N6822\_C\tablenotemark{c}
2026:       &  0.6 & dwarf & 19 44 48.67 $-$14 52 26.5 & 304. \\
2027: N6946 &  5.5 & nuc   & 20 34 51.79 +60 09 12.6 & 803. \\
2028: N7331 & 15.7 & liner & 22 37 04.11 +34 24 57.5 & 263. \\
2029: N7552 & 22.3 & nuc   & 23 16 10.56 $-$42 35 04.0 & 722. \\
2030: N7793 &  3.2 & dwarf & 23 57 49.79 $-$32 35 28.4 & 296. \\
2031: \enddata
2032: \tablenotetext{a}{
2033: Distances are from \citet{Kennicutt03}.
2034: Nuclear classifications derived from optical spectroscopy, published
2035: by \citet{Smith07b}, were modified for the following targets.
2036: N1097, N1512, N4321 and N7552: Since our aperture includes a bright
2037: star-forming ring, which dominates the dust and line emission, we adopt
2038: the H{\small II} class instead of the {\small LINER} class.
2039: N2841, N4552, N4569, N4594, N4826 and N5195: The Sy class was
2040: changed to {\small LINER} \citep{Ho97}.
2041: N3198 and N3938: The {\small LINER} class was changed to H{\small II} \citep{Ho97}.
2042: The dwarf galaxy class is here arbitrarily defined by a total stellar mass,
2043: estimated as by \citet{Lee06}, below $10^{9.7}$\,M$_{\sun}$.
2044: }
2045: \tablenotetext{b}{A total of 11 extranuclear regions were observed in high-resolution
2046: spectroscopy in NGC\,5194, from which we included 7 in our sample.}
2047: \tablenotetext{c}{Several locations were observed in high-resolution spectroscopy
2048: in this galaxy. We present here results for three among the brightest regions at
2049: 7.9 and 24\,$\mu$m. N6822\_A corresponds to the H{\small II} region Hubble\,V
2050: \citep{Hubble25}, and N6822\_C to the H{\small II} region K$\gamma$ \citep{Kinman79}.}
2051: \end{deluxetable}
2052: 
2053: 
2054: 
2055: \begin{deluxetable}{lrrrrrrrr}
2056: \tabletypesize{\footnotesize}
2057: \rotate
2058: \tablecaption{Fluxes.\tablenotemark{a}
2059: \label{tab_flux}
2060: }
2061: \tablehead{
2062: galaxy & S(0) & S(1) & S(2) & S(3) & [SiII]            & F$_{\rm 7.9\,dust}$ & F$_{24}$ & TIR \\
2063: ~      & \multicolumn{5}{c}{(10$^{-18}$\,W\,m$^{-2}$)} & (Jy)               & (Jy)     & (10$^{-15}$\,W\,m$^{-2}$) \\
2064: }
2065: \startdata
2066: N24   &     $6.6 \pm 2.6$ &    $2.2 \pm 2.4$ &    $3.2 \pm 2.0$ &   $10.3 \pm 6.4$ &     $36.7 \pm 3.6$ & 0.010 & 0.010 &     $18. \pm 3.$ \\
2067: N337  &    $11.6 \pm 3.8$ &   $21.0 \pm 4.9$ &   $15.4 \pm 9.9$ &    $9.7 \pm 5.2$ &    $186.1 \pm 7.0$ & 0.048 & 0.087 &     $89. \pm 9.$ \\
2068: N628  &     $7.9 \pm 2.5$ &   $13.7 \pm 5.3$ &    $0.0 \pm 3.0$ &    $5.7 \pm 8.6$ &     $47.8 \pm 2.8$ & 0.026 & 0.026 &     $39. \pm 3.$ \\
2069: N855  &     $2.9 \pm 0.9$ &   $11.2 \pm 6.5$ &    $0.0 \pm 5.1$ &   $22.4 \pm 7.2$ &     $70.2 \pm 3.8$ & 0.017 & 0.039 &     $53. \pm 7.$ \\
2070: N925  &     $3.8 \pm 1.7$ &    $7.8 \pm 2.9$ &    $3.7 \pm 2.1$ &    $0.0 \pm 8.8$ &     $81.0 \pm 4.5$ & 0.021 & 0.022 &     $37. \pm 4.$ \\
2071: N1097 &  $213.1 \pm 42.6$ & $726.1 \pm 43.3$ & $293.6 \pm 29.4$ & $423.0 \pm 23.4$ & $4921.1 \pm 121.9$ & 1.139 & 3.763 & $2214. \pm 227.$ \\
2072: N1266\tablenotemark{b}
2073:       &    $16.4 \pm 8.5$ &  $148.5 \pm 6.6$ &  $121.8 \pm 7.1$ & $189.8 \pm 11.6$ &   $153.5 \pm 41.5$ & 0.048 & 0.579 &   $438. \pm 30.$ \\
2074: N1291 &     $3.7 \pm 1.8$ &   $29.8 \pm 4.8$ &   $13.9 \pm 8.2$ &  $31.6 \pm 10.0$ &     $75.1 \pm 5.8$ & 0.013 & 0.049 &     $60. \pm 7.$ \\
2075: N1316 &     $1.5 \pm 0.8$ &   $36.0 \pm 6.1$ &   $18.4 \pm 5.9$ &   $84.0 \pm 9.9$ &    $102.4 \pm 6.4$ & 0.025 & 0.094 &   $105. \pm 11.$ \\
2076: N1482 &  $106.8 \pm 35.1$ & $424.0 \pm 53.4$ & $184.0 \pm 22.8$ & $207.5 \pm 26.9$ &  $3954.0 \pm 75.1$ & 1.163 & 2.974 & $1752. \pm 154.$ \\
2077: N1512 &    $25.9 \pm 2.9$ &   $83.6 \pm 8.9$ &  $27.2 \pm 10.4$ &  $48.3 \pm 13.6$ &   $390.7 \pm 13.9$ & 0.091 & 0.150 &   $203. \pm 24.$ \\
2078: N1566 &    $24.0 \pm 2.2$ &  $129.5 \pm 7.1$ &   $55.3 \pm 4.8$ &  $91.6 \pm 17.6$ &    $101.9 \pm 7.8$ & 0.098 & 0.237 &   $235. \pm 19.$ \\
2079: N1705 &     $5.6 \pm 3.9$ &    $2.2 \pm 1.3$ &   $0.0 \pm 10.3$ &   $11.6 \pm 4.0$ &    $86.3 \pm 12.0$ & 0.007 & 0.028 &     $40. \pm 5.$ \\
2080: N2403 &     $5.5 \pm 1.5$ &    $6.6 \pm 3.1$ &    $2.0 \pm 2.0$ &    $4.6 \pm 8.2$ &     $82.0 \pm 5.9$ & 0.029 & 0.022 &     $35. \pm 2.$ \\
2081: N2798 &   $43.2 \pm 25.3$ & $207.8 \pm 22.1$ &   $90.7 \pm 9.2$ & $105.2 \pm 12.1$ &   $840.4 \pm 42.2$ & 0.354 & 1.723 &   $936. \pm 67.$ \\
2082: N2841 &     $3.1 \pm 1.4$ &    $9.6 \pm 2.5$ &    $4.1 \pm 4.5$ &  $28.0 \pm 10.8$ &     $81.1 \pm 5.5$ & 0.010 & 0.027 &     $48. \pm 4.$ \\
2083: N2915 &     $0.8 \pm 0.6$ &    $1.8 \pm 2.1$ &    $3.8 \pm 1.8$ &    $0.0 \pm 8.0$ &     $40.1 \pm 3.7$ & 0.005 & 0.019 &     $25. \pm 3.$ \\
2084: N2976 &     $8.9 \pm 2.4$ &   $15.7 \pm 3.0$ &    $4.9 \pm 3.5$ &    $5.2 \pm 4.2$ &     $66.1 \pm 5.4$ & 0.026 & 0.055 &     $77. \pm 6.$ \\
2085: N3031 &     $6.3 \pm 3.6$ &   $46.1 \pm 6.1$ &   $26.1 \pm 8.1$ &   $65.3 \pm 8.0$ &    $217.9 \pm 9.7$ & 0.081 & 0.337 &   $234. \pm 17.$ \\
2086: N3049 &     $6.4 \pm 2.3$ &   $23.7 \pm 3.8$ &   $11.7 \pm 6.5$ &   $11.1 \pm 5.2$ &   $197.1 \pm 11.9$ & 0.051 & 0.258 &   $141. \pm 13.$ \\
2087: N3184 &     $9.8 \pm 2.3$ &   $22.6 \pm 2.9$ &    $6.3 \pm 3.5$ &   $0.0 \pm 15.9$ &    $105.0 \pm 4.2$ & 0.031 & 0.082 &   $103. \pm 13.$ \\
2088: N3190 &    $18.1 \pm 2.9$ &   $75.3 \pm 8.0$ &   $20.9 \pm 6.4$ &  $71.6 \pm 12.6$ &     $67.1 \pm 7.3$ & 0.051 & 0.068 &   $137. \pm 23.$ \\
2089: N3198 &    $13.2 \pm 4.9$ &   $32.1 \pm 5.4$ &   $10.2 \pm 3.5$ &   $18.7 \pm 8.5$ &     $96.5 \pm 6.6$ & 0.059 & 0.260 &   $200. \pm 20.$ \\
2090: N3265 &     $9.3 \pm 3.2$ &   $30.6 \pm 3.5$ &   $11.4 \pm 7.9$ &   $15.3 \pm 7.4$ &    $159.1 \pm 7.7$ & 0.062 & 0.193 &    $119. \pm 9.$ \\
2091: Mrk33 &    $13.9 \pm 5.6$ &   $29.3 \pm 3.7$ &   $11.5 \pm 5.3$ &   $45.7 \pm 8.6$ &   $245.4 \pm 10.9$ & 0.076 & 0.583 &   $239. \pm 13.$ \\
2092: N3351 &   $62.8 \pm 10.5$ & $218.4 \pm 20.6$ &  $88.3 \pm 10.1$ & $169.8 \pm 19.9$ &  $1858.9 \pm 35.7$ & 0.366 & 1.385 &   $917. \pm 93.$ \\
2093: N3521 &    $17.8 \pm 3.6$ &   $31.2 \pm 6.9$ &   $12.3 \pm 4.0$ &   $20.5 \pm 8.8$ &    $157.5 \pm 4.6$ & 0.101 & 0.106 &   $129. \pm 10.$ \\
2094: N3621 &    $16.7 \pm 2.1$ &   $38.3 \pm 3.2$ &   $17.1 \pm 4.7$ &   $49.0 \pm 8.4$ &    $150.3 \pm 7.5$ & 0.075 & 0.065 &     $88. \pm 6.$ \\
2095: N3627 &    $31.2 \pm 3.9$ & $318.6 \pm 12.2$ &  $141.7 \pm 5.2$ & $209.1 \pm 12.6$ &    $162.1 \pm 9.3$ & 0.190 & 0.406 &   $518. \pm 31.$ \\
2096: N3773 &     $4.8 \pm 3.4$ &    $7.3 \pm 3.2$ &    $3.7 \pm 2.5$ &    $8.3 \pm 8.6$ &    $106.9 \pm 6.1$ & 0.024 & 0.087 &     $67. \pm 7.$ \\
2097: N3938 &     $8.0 \pm 1.2$ &   $13.4 \pm 3.3$ &    $4.2 \pm 3.2$ &    $0.0 \pm 8.0$ &     $57.7 \pm 4.2$ & 0.027 & 0.030 &     $51. \pm 5.$ \\
2098: N4125 &     $2.6 \pm 1.6$ &   $17.6 \pm 4.7$ &   $12.6 \pm 6.4$ &   $18.0 \pm 8.8$ &     $46.8 \pm 5.9$ & 0.005 & 0.015 &     $21. \pm 3.$ \\
2099: N4254 &    $17.6 \pm 1.3$ &   $87.3 \pm 9.5$ &   $46.6 \pm 9.2$ &   $32.5 \pm 8.8$ &   $409.0 \pm 10.6$ & 0.160 & 0.203 &   $231. \pm 23.$ \\
2100: N4321 &   $78.2 \pm 11.7$ & $267.5 \pm 22.8$ & $132.1 \pm 24.6$ & $164.5 \pm 16.2$ &  $1507.3 \pm 27.2$ & 0.392 & 0.730 &   $778. \pm 90.$ \\
2101: N4450 &    $10.3 \pm 1.6$ &   $91.4 \pm 5.2$ &   $33.7 \pm 9.5$ &  $89.0 \pm 11.2$ &     $29.6 \pm 3.1$ & 0.011 & 0.028 &     $55. \pm 9.$ \\
2102: N4536 &  $108.7 \pm 32.3$ & $411.6 \pm 40.5$ & $175.6 \pm 23.3$ & $216.8 \pm 20.7$ &  $3207.4 \pm 70.0$ & 0.813 & 2.401 & $1400. \pm 118.$ \\
2103: N4552 &     $1.8 \pm 1.4$ &    $1.1 \pm 1.5$ &    $0.0 \pm 3.9$ &    $9.1 \pm 4.6$ &     $13.4 \pm 3.9$ & 0.004 & 0.017 &      $6. \pm 1.$ \\
2104: N4559 &    $13.6 \pm 2.1$ &   $17.9 \pm 3.5$ &    $4.0 \pm 1.5$ &  $23.4 \pm 13.7$ &     $81.1 \pm 3.5$ & 0.034 & 0.030 &     $47. \pm 5.$ \\
2105: N4569\tablenotemark{c}
2106:       &    $41.6 \pm 6.9$ &  $316.1 \pm 8.7$ &  $152.4 \pm 5.2$ & $309.4 \pm 18.7$ &   $288.5 \pm 17.6$ & 0.128 & 0.475 &   $365. \pm 40.$ \\
2107: N4579\tablenotemark{d}
2108:       &     $8.9 \pm 2.3$ &  $164.1 \pm 5.2$ & $105.8 \pm 12.4$ & $256.3 \pm 10.0$ &   $175.3 \pm 12.2$ & 0.032 & 0.135 &   $160. \pm 18.$ \\
2109: N4594 &     $1.9 \pm 2.3$ &    $6.4 \pm 3.8$ &    $4.2 \pm 2.7$ &    $4.6 \pm 4.0$ &    $126.1 \pm 7.3$ & 0.017 & 0.068 &     $70. \pm 8.$ \\
2110: N4625 &     $8.0 \pm 1.6$ &    $9.6 \pm 1.9$ &    $5.4 \pm 4.3$ &   $5.0 \pm 11.4$ &     $67.5 \pm 4.8$ & 0.024 & 0.024 &     $33. \pm 5.$ \\
2111: N4631 &    $58.2 \pm 6.4$ & $123.6 \pm 16.1$ &   $60.4 \pm 5.6$ &   $46.8 \pm 9.1$ &  $1350.4 \pm 22.7$ & 0.338 & 0.661 &   $636. \pm 61.$ \\
2112: N4725 &    $11.7 \pm 1.7$ &   $37.1 \pm 5.1$ &   $19.9 \pm 8.1$ &   $38.0 \pm 9.0$ &     $24.9 \pm 3.2$ & 0.011 & 0.035 &     $56. \pm 9.$ \\
2113: N4736 &    $33.2 \pm 6.8$ & $250.8 \pm 16.4$ & $100.1 \pm 10.4$ & $221.7 \pm 13.6$ &   $219.5 \pm 21.4$ & 0.293 & 0.636 &   $678. \pm 50.$ \\
2114: N4826 &    $71.3 \pm 6.5$ & $345.3 \pm 15.8$ & $150.9 \pm 10.9$ & $210.6 \pm 10.3$ &   $927.6 \pm 15.0$ & 0.343 & 0.612 &   $756. \pm 92.$ \\
2115: N5033 &    $36.6 \pm 3.5$ & $182.0 \pm 10.4$ &   $63.5 \pm 3.1$ & $126.9 \pm 19.1$ &   $530.9 \pm 10.0$ & 0.177 & 0.239 &   $311. \pm 45.$ \\
2116: N5055 &    $44.0 \pm 2.4$ &  $158.0 \pm 9.1$ &   $52.0 \pm 6.7$ &  $80.2 \pm 10.6$ &    $295.2 \pm 9.0$ & 0.148 & 0.166 &   $300. \pm 37.$ \\
2117: N5194 &    $17.7 \pm 2.8$ &  $134.4 \pm 9.1$ &   $75.7 \pm 4.7$ & $187.3 \pm 17.7$ &   $634.9 \pm 12.2$ & 0.168 & 0.291 &   $351. \pm 39.$ \\
2118: N5194\_A &  $5.8 \pm 1.3$ &   $16.4 \pm 2.7$ &    $4.3 \pm 2.6$ &    $5.4 \pm 1.8$ &     $27.8 \pm 3.0$ & 0.030 & 0.036 &     $53. \pm 2.$ \\
2119: N5194\_B & $18.1 \pm 2.2$ &   $43.1 \pm 8.2$ &   $12.0 \pm 4.3$ &  $18.1 \pm 10.2$ &    $339.7 \pm 8.5$ & 0.094 & 0.193 &   $201. \pm 13.$ \\
2120: N5194\_C & $23.2 \pm 2.2$ &   $43.4 \pm 6.7$ &    $5.7 \pm 4.5$ &   $14.1 \pm 4.6$ &    $223.6 \pm 6.8$ & 0.079 & 0.115 &    $155. \pm 7.$ \\
2121: N5194\_D & $14.7 \pm 2.8$ &   $30.3 \pm 6.5$ &    $3.9 \pm 3.1$ &    $8.4 \pm 3.0$ &    $222.9 \pm 6.2$ & 0.058 & 0.118 &   $130. \pm 14.$ \\
2122: N5194\_E & $22.7 \pm 1.7$ &   $47.9 \pm 6.0$ &   $19.2 \pm 7.1$ &   $23.3 \pm 3.1$ &    $222.2 \pm 4.8$ & 0.072 & 0.089 &    $116. \pm 8.$ \\
2123: N5194\_F & $19.4 \pm 2.2$ &   $39.2 \pm 5.2$ &   $15.4 \pm 9.1$ &  $21.6 \pm 14.0$ &    $271.7 \pm 5.8$ & 0.075 & 0.110 &    $132. \pm 4.$ \\
2124: N5194\_G & $22.2 \pm 3.0$ &   $49.6 \pm 5.4$ &   $23.6 \pm 2.7$ &  $27.7 \pm 11.2$ &    $425.6 \pm 8.5$ & 0.093 & 0.192 &   $173. \pm 15.$ \\
2125: N5195 &   $54.2 \pm 20.9$ & $310.4 \pm 12.3$ &  $127.7 \pm 6.5$ & $275.0 \pm 17.2$ &   $127.8 \pm 19.6$ & 0.271 & 0.835 &   $600. \pm 55.$ \\
2126: N5713 &    $27.4 \pm 3.6$ & $151.6 \pm 14.3$ &   $50.9 \pm 7.9$ &   $85.6 \pm 7.0$ &   $798.7 \pm 14.8$ & 0.268 & 0.738 &   $500. \pm 54.$ \\
2127: N5866 &    $15.5 \pm 1.5$ &   $90.0 \pm 5.5$ &   $39.1 \pm 7.0$ &   $37.2 \pm 6.5$ &     $78.8 \pm 3.9$ & 0.038 & 0.052 &   $169. \pm 26.$ \\
2128: N6822\_A &  $8.1 \pm 4.3$ &   $21.1 \pm 4.3$ &   $16.0 \pm 4.1$ &   $25.9 \pm 3.8$ &   $187.1 \pm 11.2$ & 0.036 & 0.517 &   $283. \pm 24.$ \\
2129: N6822\_B &  $7.8 \pm 1.3$ &   $11.2 \pm 3.6$ &    $8.8 \pm 2.8$ &   $10.9 \pm 5.1$ &     $19.0 \pm 2.8$ & 0.010 & 0.037 &     $44. \pm 5.$ \\
2130: N6822\_C &  $4.4 \pm 1.9$ &    $8.0 \pm 5.2$ &    $6.2 \pm 5.4$ &    $8.0 \pm 2.9$ &     $16.0 \pm 3.6$ & 0.009 & 0.100 &     $76. \pm 7.$ \\
2131: N6946 &  $227.7 \pm 61.9$ & $636.9 \pm 33.5$ & $271.6 \pm 17.1$ & $286.3 \pm 29.3$ &  $3203.9 \pm 91.9$ & 1.314 & 4.234 & $2354. \pm 193.$ \\
2132: N7331 &    $16.4 \pm 1.6$ &   $49.6 \pm 5.0$ &   $13.8 \pm 1.9$ &   $23.4 \pm 8.3$ &    $213.9 \pm 4.8$ & 0.074 & 0.091 &   $125. \pm 10.$ \\
2133: N7552 & $257.8 \pm 106.9$ & $565.3 \pm 42.6$ & $245.9 \pm 22.5$ & $315.9 \pm 14.8$ & $3619.9 \pm 165.2$ & 1.495 & 7.840 & $3237. \pm 219.$ \\
2134: N7793 &     $7.0 \pm 1.7$ &   $15.0 \pm 3.2$ &   $11.7 \pm 4.6$ &    $5.3 \pm 3.7$ &     $79.7 \pm 4.3$ & 0.041 & 0.046 &     $71. \pm 5.$ \\
2135: \enddata
2136: \tablenotetext{a}{See Sections~\ref{images} and \ref{measurements}.
2137: For NGC\,1266, NGC\,4569 and NGC\,4579, it was possible to measure higher-level
2138: transitions of H$_2$, which are provided in the following notes.}
2139: \tablenotetext{b}{The S(4) to S(7) line fluxes are respectively: $(102.7 \pm 17.4)$, $(241.9 \pm 19.9)$,
2140: $> 18.5$ and $(192.2 \pm 28.6) \times 10^{-18}$\,W\,m$^{-2}$. The S(6) transition was not
2141: accurately measurable because it was observed at low spectral resolution on the blue shoulder
2142: of a bright 6.2\,$\mu$m aromatic band.}
2143: \tablenotetext{c}{The S(4) to S(7) line fluxes are respectively: $(52.6 \pm 27.7)$, $< 252.6$,
2144: $< 61.3$ and $(50.3 \pm 24.2) \times 10^{-18}$\,W\,m$^{-2}$. The S(5) and S(6) transitions are
2145: upper limits because S(5) is contaminated by the [Ar{\small II}] line at 6.985\,$\mu$m and S(6)
2146: is diluted in the blue shoulder of the 6.2\,$\mu$m aromatic band.}
2147: \tablenotetext{d}{The S(4) to S(7) line fluxes are respectively: $(48.7 \pm 10.7)$, $< 324.9$,
2148: $< 147.2$ and $(83.3 \pm 37.5) \times 10^{-18}$\,W\,m$^{-2}$. The same remark as for NGC\,4569
2149: applies.}
2150: \end{deluxetable}
2151: 
2152: 
2153: 
2154: \begin{deluxetable}{lrrrrrrr}
2155: \tabletypesize{\footnotesize}
2156: \tablecaption{Derived temperatures, column densities and $OPR_{\rm \,high\,T}$.\tablenotemark{a}
2157: \label{tab_fit}
2158: }
2159: \tablehead{
2160: galaxy & $T_1$ & $f_1$ & $T_2$ & $f_2$ & $OPR_{\rm \,high\,T}$ & $N_{\rm tot}(T > T_1)$       & $N_{\rm tot}(OPR=3)$ \\
2161: ~      & (K)   & ~     & (K)   & ~     & ~                     & ($10^{20}$\,mol.\,cm$^{-2}$) & ~$/N_{\rm tot}(OPR<3)$ \\
2162: }
2163: \startdata
2164: N24   &    97. & 0.99941 &  769. & 0.00059 & $1.49 \pm 0.91$ &  4.232 & 1.75 \\
2165: ~     &    78. & 0.99980 &  675. & 0.00020 & $1.90 \pm 0.57$ & 15.210 & 1.46 \\
2166: ~     &    90. & 0.99971 &  888. & 0.00029 & $1.45 \pm 0.37$ &  7.314 & 1.73 \\
2167: N337  &   160. & 0.97793 &  402. & 0.02207 & $1.02 \pm 0.15$ &  1.745 & 2.13 \\
2168: ~     &   128. & 0.97803 &  352. & 0.02197 & $1.21 \pm 0.25$ &  3.218 & 0.97 \\
2169: ~     &   165. & 0.98711 &  464. & 0.01289 & $0.86 \pm 0.14$ &  1.824 & 2.19 \\
2170: N628  &   119. & 0.99678 &  393. & 0.00322 & 3               &  2.598 & ~    \\
2171: N855  &   146. & 0.98750 &  578. & 0.01250 & 3               &  0.508 & ~    \\
2172: N925  &   123. & 0.99473 &  494. & 0.00527 & 3               &  1.050 & ~    \\
2173: ~     &    99. & 0.99648 &  421. & 0.00352 & 3               &  2.780 & ~    \\
2174: %N1097 &   140. & 0.98281 &  395. & 0.01719 & 3               & 14.376 & ~    \\
2175: N1097 &   157. & 0.98027 &  422. & 0.01973 & $2.03 \pm 0.36$ & 11.837 & 1.29 \\
2176: ~     &    94. & 0.99033 &  335. & 0.00967 & 3               & 62.925 & ~    \\
2177: N1266\tablenotemark{b}
2178:       &   247. & 0.83691 &  400. & 0.15619 & $1.79 \pm 0.29$ &  1.231 & 1.19 \\
2179: ~     &   204. & 0.82588 &  400. & 0.16935 & $1.80 \pm 0.18$ &  1.594 & 0.84 \\
2180: N1291 &   184. & 0.95420 &  484. & 0.04580 & 3               &  0.415 & ~    \\
2181: ~     &   141. & 0.95010 &  414. & 0.04990 & 3               &  0.762 & ~    \\
2182: N1316 &   300. & 0.96436 & 1456. & 0.03564 & 3               &  0.114 & ~    \\
2183: ~     & $> 278.$ & $> 0.93750$ & $> 900.$ & $< 0.06250$ & 3  &  0.121 & ~    \\
2184: N1482 &   171. & 0.96943 &  407. & 0.03057 & $1.83 \pm 0.25$ &  5.098 & 1.38 \\
2185: ~     &   137. & 0.96699 &  357. & 0.03301 & $2.12 \pm 0.43$ &  8.848 & 1.49 \\
2186: ~     &   175. & 0.98252 &  469. & 0.01748 & $1.58 \pm 0.25$ &  5.378 & 1.40 \\
2187: N1512 &   138. & 0.98887 &  415. & 0.01113 & 3               &  1.883 & ~    \\
2188: ~     &   118. & 0.98955 &  375. & 0.01045 & 3               &  3.338 & ~    \\
2189: %N1566 &   161. & 0.96729 &  419. & 0.03271 & 3               &  3.149 & ~    \\
2190: N1566 &   177. & 0.96436 &  441. & 0.03564 & $2.24 \pm 0.37$ &  2.749 & 1.19 \\
2191: ~     &   115. & 0.97305 &  359. & 0.02695 & 3               &  8.517 & ~    \\
2192: %N1705 &    88. & 0.99980 &  811. & 0.00020 & 3               &  2.208 & ~    \\
2193: N1705 &   127. & 0.99453 &  811. & 0.00547 & $0.49 \pm 0.29$ &  0.494 & 3.78 \\
2194: ~     &    92. & 0.99844 &  711. & 0.00156 & $0.65 \pm 0.14$ &  2.050 & 0.95 \\
2195: ~     &   114. & 0.99727 &  936. & 0.00273 & $0.49 \pm 0.11$ &  0.876 & 2.33 \\
2196: N2403 &   109. & 0.99844 &  449. & 0.00156 & 3               &  2.486 & ~    \\
2197: ~     &    99. & 0.99873 &  419. & 0.00127 & 3               &  4.138 & ~    \\
2198: N2798 &   179. & 0.96182 &  409. & 0.03818 & $1.92 \pm 0.23$ &  5.169 & 1.32 \\
2199: ~     &   144. & 0.95518 &  359. & 0.04482 & $2.17 \pm 0.43$ &  8.460 & 1.16 \\
2200: ~     &   184. & 0.97812 &  472. & 0.02188 & $1.67 \pm 0.27$ &  5.419 & 1.27 \\
2201: N2841 &   137. & 0.99678 & 1030. & 0.00322 & 3               &  0.709 & ~    \\
2202: ~     &   132. & 0.99707 &  970. & 0.00293 & 3               &  0.844 & ~    \\
2203: N2915 &   169. & 0.97090 &  763. & 0.02910 & $1.05 \pm 0.47$ &  0.096 & 1.74 \\
2204: ~     &   116. & 0.98838 &  669. & 0.01162 & $1.30 \pm 0.28$ &  0.305 & 0.55 \\
2205: ~     &   151. & 0.98350 &  880. & 0.01650 & $1.00 \pm 0.20$ &  0.150 & 1.13 \\
2206: %N2976 &   119. & 0.99238 &  340. & 0.00762 & 3               &  2.705 & ~    \\
2207: N2976 &   134. & 0.99199 &  376. & 0.00801 & $1.84 \pm 0.30$ &  2.095 & 1.43 \\
2208: ~     &    86. & 0.99619 &  300. & 0.00381 & 3               & 12.046 & ~    \\
2209: N3031 &   178. & 0.95908 &  523. & 0.04092 & 3               &  0.700 & ~    \\
2210: ~     &   126. & 0.96387 &  433. & 0.03613 & 3               &  1.699 & ~    \\
2211: N3049 &   174. & 0.96611 &  403. & 0.03389 & $1.60 \pm 0.20$ &  0.740 & 1.50 \\
2212: ~     &   141. & 0.96113 &  353. & 0.03887 & $1.83 \pm 0.36$ &  1.238 & 1.17 \\
2213: ~     &   179. & 0.98076 &  465. & 0.01924 & $1.39 \pm 0.22$ &  0.773 & 1.55 \\
2214: N3184 &   127. & 0.99502 &  440. & 0.00498 & 3               &  2.631 & ~    \\
2215: ~     &   118. & 0.99648 &  449. & 0.00352 & 3               &  3.805 & ~    \\
2216: N3190 &   148. & 0.99355 &  574. & 0.00645 & 3               &  3.286 & ~    \\
2217: ~     &   142. & 0.99385 &  554. & 0.00615 & 3               &  3.883 & ~    \\
2218: N3198 &   128. & 0.99316 &  415. & 0.00684 & 3               &  3.438 & ~    \\
2219: ~     &   111. & 0.99385 &  375. & 0.00615 & 3               &  6.283 & ~    \\
2220: %N3265 &   139. & 0.98164 &  375. & 0.01836 & 3               &  1.711 & ~    \\
2221: N3265 &   156. & 0.98105 &  408. & 0.01895 & $2.02 \pm 0.35$ &  1.421 & 1.30 \\
2222: ~     &    95. & 0.98975 &  325. & 0.01025 & 3               &  7.006 & ~    \\
2223: Mrk33 &   124. & 0.99736 &  624. & 0.00264 & 3               &  3.844 & ~    \\
2224: ~     &   115. & 0.99775 &  584. & 0.00225 & 3               &  5.478 & ~    \\
2225: N3351 &   141. & 0.98633 &  437. & 0.01367 & 3               &  4.575 & ~    \\
2226: ~     &   114. & 0.98867 &  387. & 0.01133 & 3               &  9.893 & ~    \\
2227: N3521 &   118. & 0.99531 &  414. & 0.00469 & 3               &  5.715 & ~    \\
2228: ~     &    91. & 0.99687 &  354. & 0.00313 & 3               & 18.837 & ~    \\
2229: N3621 &   126. & 0.99541 &  532. & 0.00459 & 3               &  4.490 & ~    \\
2230: ~     &   110. & 0.99609 &  472. & 0.00391 & 3               &  8.195 & ~    \\
2231: %N3627 &   201. & 0.90293 &  404. & 0.09707 & 3               &  2.833 & ~    \\
2232: N3627 &   220. & 0.91836 &  433. & 0.08164 & $2.42 \pm 0.19$ &  2.672 & 1.12 \\
2233: ~     &   124. & 0.90469 &  344. & 0.09531 & 3               &  7.324 & ~    \\
2234: N3773 &   115. & 0.99678 &  480. & 0.00322 & 3               &  1.671 & ~    \\
2235: ~     &    90. & 0.99805 &  410. & 0.00195 & 3               &  5.446 & ~    \\
2236: N3938 &   118. & 0.99629 &  425. & 0.00371 & 3               &  2.603 & ~    \\
2237: ~     &   102. & 0.99697 &  377. & 0.00303 & 3               &  5.116 & ~    \\
2238: N4125 &   211. & 0.92803 &  481. & 0.07197 & $1.75 \pm 0.24$ &  0.238 & 1.35 \\
2239: ~     &   163. & 0.91826 &  422. & 0.08174 & $1.90 \pm 0.37$ &  0.377 & 1.15 \\
2240: ~     &   212. & 0.95625 &  555. & 0.04375 & $1.56 \pm 0.25$ &  0.262 & 1.35 \\
2241: N4254 &   200. & 0.94111 &  384. & 0.05889 & $1.49 \pm 0.11$ &  1.763 & 1.53 \\
2242: ~     &   167. & 0.90293 &  337. & 0.09707 & $1.59 \pm 0.29$ &  2.368 & 1.14 \\
2243: ~     &   206. & 0.96982 &  444. & 0.03018 & $1.36 \pm 0.22$ &  1.812 & 1.45 \\
2244: N4321 &   164. & 0.97480 &  427. & 0.02520 & $1.76 \pm 0.29$ &  4.001 & 1.41 \\
2245: ~     &   131. & 0.97627 &  374. & 0.02373 & $2.07 \pm 0.43$ &  7.511 & 1.27 \\
2246: ~     &   168. & 0.98477 &  493. & 0.01523 & $1.49 \pm 0.25$ &  4.258 & 1.46 \\
2247: N4450 &   190. & 0.96992 &  536. & 0.03008 & 3               &  1.178 & ~    \\
2248: ~     &   172. & 0.96152 &  476. & 0.03848 & 3               &  1.467 & ~    \\
2249: N4536 &   166. & 0.97344 &  413. & 0.02656 & $1.90 \pm 0.28$ &  5.602 & 1.35 \\
2250: ~     &   133. & 0.97334 &  362. & 0.02666 & $2.23 \pm 0.45$ & 10.170 & 1.35 \\
2251: ~     &   170. & 0.98437 &  476. & 0.01563 & $1.62 \pm 0.27$ &  5.908 & 1.40 \\
2252: %N4552 &    96. & 0.99941 &  986. & 0.00059 & 3               &  1.192 & ~    \\
2253: N4552 &   126. & 0.99463 &  986. & 0.00537 & $0.84 \pm 0.56$ &  0.449 & 2.61 \\
2254: ~     &    89. & 0.99873 &  865. & 0.00127 & $1.09 \pm 0.24$ &  2.156 & 0.54 \\
2255: ~     &   108. & 0.99766 & 1138. & 0.00234 & $0.87 \pm 0.19$ &  0.959 & 1.30 \\
2256: N4559 &   111. & 0.99951 &  775. & 0.00049 & 3               &  6.202 & ~    \\
2257: ~     &   109. & 0.99961 &  825. & 0.00039 & 3               &  6.967 & ~    \\
2258: N4569\tablenotemark{c}
2259:       &   184. & 0.91348 &  400. & 0.08564 & 3               &  4.062 & ~    \\
2260: ~     &   138. & 0.95226 &  400. & 0.04755 & 3               &  9.055 & ~    \\
2261: %N4579\tablenotemark{c}
2262: %      &   275. & 0.56201 &  400. & 0.43198 & 3               &  0.533 & ~    \\
2263: N4579\tablenotemark{d}
2264:       &   286. & 0.55439 &  400. & 0.43999 & $2.80 \pm 0.56$ &  0.534 & 1.00 \\
2265: ~     &    68. & 0.98551 &  400. & 0.01439 & 3               & 21.592 & ~    \\
2266: N4594 &   175. & 0.96621 &  441. & 0.03379 & $1.49 \pm 0.23$ &  0.232 & 1.57 \\
2267: ~     &   138. & 0.96709 &  387. & 0.03291 & $1.71 \pm 0.35$ &  0.421 & 1.06 \\
2268: ~     &   178. & 0.97969 &  509. & 0.02031 & $1.28 \pm 0.21$ &  0.251 & 1.64 \\
2269: N4625 &   133. & 0.99268 &  411. & 0.00732 & $1.27 \pm 0.28$ &  1.856 & 1.85 \\
2270: ~     &   107. & 0.99453 &  361. & 0.00547 & $1.65 \pm 0.35$ &  4.144 & 1.20 \\
2271: ~     &   136. & 0.99551 &  475. & 0.00449 & $1.05 \pm 0.20$ &  1.964 & 2.38 \\
2272: N4631 &   154. & 0.98174 &  386. & 0.01826 & $1.36 \pm 0.19$ &  9.852 & 1.75 \\
2273: ~     &   125. & 0.98174 &  338. & 0.01826 & $1.64 \pm 0.32$ & 18.343 & 0.96 \\
2274: ~     &   159. & 0.98955 &  445. & 0.01045 & $1.14 \pm 0.19$ & 10.160 & 2.01 \\
2275: %N4725 &   137. & 0.98730 &  461. & 0.01270 & 3               &  2.289 & ~    \\
2276: N4725 &   155. & 0.98223 &  481. & 0.01777 & $1.97 \pm 0.50$ &  1.814 & 1.31 \\
2277: ~     &    88. & 0.99482 &  381. & 0.00518 & 3               & 13.562 & ~    \\
2278: N4736 &   180. & 0.96455 &  478. & 0.03545 & 3               &  3.822 & ~    \\
2279: ~     &   151. & 0.95732 &  418. & 0.04268 & 3               &  5.791 & ~    \\
2280: %N4826 &   155. & 0.96357 &  388. & 0.03643 & 3               & 10.172 & ~    \\
2281: N4826 &   176. & 0.96523 &  426. & 0.03477 & $2.05 \pm 0.30$ &  8.766 & 1.27 \\
2282: ~     &    72. & 0.99570 &  328. & 0.00430 & 3               & 223.50 & ~    \\
2283: N5033 &   157. & 0.97900 &  438. & 0.02100 & 3               &  5.255 & ~    \\
2284: ~     &   134. & 0.97920 &  398. & 0.02080 & 3               &  8.435 & ~    \\
2285: N5055 &   142. & 0.98242 &  386. & 0.01758 & 3               &  7.586 & ~    \\
2286: ~     &   114. & 0.98506 &  346. & 0.01494 & 3               & 16.336 & ~    \\
2287: N5194 &   180. & 0.95732 &  521. & 0.04268 & 3               &  1.753 & ~    \\
2288: ~     &   127. & 0.96123 &  431. & 0.03877 & 3               &  4.171 & ~    \\
2289: N5194\_A & 133. & 0.98672 & 349. & 0.01328 & 3               &  1.469 & ~    \\
2290: ~        & 111. & 0.98848 & 319. & 0.01152 & 3               &  2.951 & ~    \\
2291: N5194\_B & 128. & 0.99150 & 373. & 0.00850 & 3               &  4.933 & ~    \\
2292: ~        & 109. & 0.99287 & 343. & 0.00713 & 3               &  9.421 & ~    \\
2293: N5194\_C & 120. & 0.99873 & 458. & 0.00127 & 3               &  6.824 & ~    \\
2294: ~        & 118. & 0.99873 & 448. & 0.00127 & 3               &  7.448 & ~    \\
2295: N5194\_D & 123. & 0.99824 & 429. & 0.00176 & 3               &  4.144 & ~    \\
2296: ~        & 121. & 0.99824 & 419. & 0.00176 & 3               &  4.572 & ~    \\
2297: N5194\_E & 142. & 0.98896 & 406. & 0.01104 & $1.79 \pm 0.33$ &  5.653 & 1.44 \\
2298: ~        &  63. & 0.99922 & 313. & 0.00078 & 3               &  235.5 & ~    \\
2299: N5194\_F & 137. & 0.99102 & 417. & 0.00898 & $1.91 \pm 0.40$ &  4.127 & 1.37 \\
2300: ~        &  85. & 0.99648 & 331. & 0.00352 & 3               & 26.208 & ~    \\
2301: N5194\_G & 148. & 0.98594 & 418. & 0.01406 & $1.64 \pm 0.31$ &  4.041 & 1.53 \\
2302: ~        & 118. & 0.98857 & 367. & 0.01143 & $2.03 \pm 0.43$ &  8.457 & 1.21 \\
2303: ~        & 151. & 0.99160 & 483. & 0.00840 & $1.37 \pm 0.24$ &  4.337 & 1.74 \\
2304: N5195 &   164. & 0.97441 &  468. & 0.02559 & 3               &  6.769 & ~    \\
2305: ~     &   134. & 0.97305 &  408. & 0.02695 & 3               & 11.899 & ~    \\
2306: N5713 &   162. & 0.97266 &  413. & 0.02734 & 3               &  3.901 & ~    \\
2307: ~     &   133. & 0.96875 &  363. & 0.03125 & 3               &  6.684 & ~    \\
2308: N5866 &   194. & 0.94678 &  393. & 0.05322 & $1.88 \pm 0.15$ &  1.676 & 1.33 \\
2309: ~     &   160. & 0.92070 &  344. & 0.07930 & $2.03 \pm 0.37$ &  2.379 & 1.10 \\
2310: ~     &   200. & 0.97158 &  453. & 0.02842 & $1.68 \pm 0.27$ &  1.731 & 1.25 \\
2311: N6822\_A & 158. & 0.98027 & 504. & 0.01973 & $1.53 \pm 0.42$ &  1.094 & 1.56 \\
2312: ~     &   120. & 0.98721 &  442. & 0.01279 & $1.89 \pm 0.41$ &  2.561 & 1.31 \\
2313: ~     &   156. & 0.98799 &  581. & 0.01201 & $1.33 \pm 0.25$ &  1.306 & 1.76 \\
2314: N6822\_B & 140. & 0.99043 & 476. & 0.00957 & $1.29 \pm 0.35$ &  1.609 & 1.82 \\
2315: ~     &   109. & 0.99404 &  417. & 0.00596 & $1.63 \pm 0.36$ &  3.946 & 1.30 \\
2316: ~     &   139. & 0.99414 &  549. & 0.00586 & $1.09 \pm 0.21$ &  1.867 & 2.00 \\
2317: N6822\_C & 148. & 0.98604 & 478. & 0.01396 & $1.33 \pm 0.35$ &  0.716 & 1.77 \\
2318: ~        & 115. & 0.99092 & 419. & 0.00908 & $1.67 \pm 0.34$ &  1.690 & 1.04 \\
2319: ~        & 148. & 0.99150 & 552. & 0.00850 & $1.14 \pm 0.21$ &  0.830 & 2.19 \\
2320: N6946 &   156. & 0.98066 &  399. & 0.01934 & $1.70 \pm 0.26$ & 13.119 & 1.46 \\
2321: ~     &   126. & 0.98105 &  350. & 0.01895 & $2.04 \pm 0.41$ & 24.494 & 1.09 \\
2322: ~     &   161. & 0.98857 &  461. & 0.01143 & $1.43 \pm 0.24$ & 13.603 & 1.62 \\
2323: N7331 &   136. & 0.99043 &  401. & 0.00957 & 3               &  3.976 & ~    \\
2324: ~     &   118. & 0.99023 &  361. & 0.00977 & 3               &  6.658 & ~    \\
2325: N7552 &   144. & 0.98818 &  418. & 0.01182 & $1.77 \pm 0.35$ & 20.019 & 1.43 \\
2326: ~     &   115. & 0.99053 &  367. & 0.00947 & $2.21 \pm 0.46$ & 42.249 & 1.12 \\
2327: ~     &   147. & 0.99277 &  483. & 0.00723 & $1.47 \pm 0.26$ & 21.235 & 1.67 \\
2328: N7793 &   176. & 0.96455 &  381. & 0.03545 & $0.91 \pm 0.09$ &  0.869 & 2.29 \\
2329: ~     &   144. & 0.95400 &  334. & 0.04600 & $1.02 \pm 0.20$ &  1.368 & 1.50 \\
2330: ~     &   181. & 0.98096 &  440. & 0.01904 & $0.79 \pm 0.13$ &  0.900 & 2.62 \\
2331: \enddata
2332: \tablenotetext{a}{
2333: The parameters $f_{\rm i}$ are the mass fractions of the discrete
2334: components at the temperatures $T_{\rm i}$.
2335: For each galaxy, the first line gives the results of the fits where $T_1$
2336: is constrained in order not to overestimate the column density, and the second (third)
2337: line where $T_1$ is unrestricted. For galaxies where the S(2) flux is an upper limit,
2338: only constrained-$T_1$ fits were performed.
2339: When $OPR_{\rm \,high\,T} < 3$, $T_2$ is fixed, and the free-$T_1$ fit results are then
2340: provided for two
2341: different values of $T_2$: $1.14 \times T$(S1-S3) (second line) and $1.5 \times T$(S1-S3)
2342: (third line). See Section~\ref{fits} for explanations.}
2343: \tablenotetext{b}{The third component is characterized by $T_3 = 1415$\,K and $f_3 = 0.00690$
2344: (constrained-$T_1$ fit) or $T_3 = 1455$\,K and $f_3 = 0.00477$ (free-$T_1$ fit).}
2345: \tablenotetext{c}{$T_3 = 1139$\,K and $f_3 = 0.00088$ (constrained-$T_1$ fit),
2346: or $T_3 = 1319$\,K and $f_3 = 0.00019$ (free-$T_1$ fit).}
2347: \tablenotetext{d}{
2348: $T_3 = 1324$\,K and $f_3 = 0.00561$ (constrained-$T_1$ fit), or $T_3 = 1414$\,K and
2349: $f_3 = 0.00010$ (free-$T_1$ fit).}
2350: \end{deluxetable}
2351: 
2352: 
2353: 
2354: \begin{deluxetable}{lrrrrrll}
2355: \tabletypesize{\footnotesize}
2356: \rotate
2357: \tablecaption{Masses in the warm and cold phases.\tablenotemark{a}
2358: \label{tab_mass}
2359: }
2360: \tablehead{
2361: galaxy & $M(T > T_1)$         & $N_{\rm tot}$(cold H$_2$)    & $M(T > T_1)$            & CO beam      & corr. & map & ref CO \\
2362: ~      & ($10^6$\,M$_{\sun}$) & ($10^{20}$\,mol.\,cm$^{-2}$) & ~$/~ M({\rm cold H}_2)$ & (arcsec$^2$) & ~     & ~   & ~ \\
2363: }
2364: \startdata
2365: N24   &   3.13 & ~        & ~        & ~     & ~    & ~    & ~ \\
2366: N337  &  11.61 &   9.4    & 0.18     & 1521. & 1.52 & IRAC & E96 \\
2367: N628  &   3.79 &  10.5    & 0.25     & 627.  & 1.04 & BIMA & S05 (B93, Y95) \\
2368: N855  &   0.50 &   2.3    & 0.22     & 415.  & 1.24 & IRAC & W95 \\
2369: N925  &   1.15 & $< 18.7$ & $> 0.06$ & 386.  & ~    & ~    & S05 \\
2370: N1097 & 103.01 & 183.     & 0.06     & 1590. & 1.74 & IRAC & Y95 (VV98, H93, K03) \\
2371: N1266 &  12.97 &  ~       & ~        & ~     & ~    & ~    & ~ \\
2372: N1291 &   0.42 &  11.7    & 0.04     & 1452. & 2.33 & IRAC & T91 \\
2373: N1316 &   0.85 &  19.9    & 0.006    & 1452. & 2.55 & IRAC & H01 \\
2374: N1482 &  73.91 & 136.     & 0.04     & 1521. & 1.76 & IRAC & E96 \\
2375: N1512 &   6.34 &  ~       & ~        & ~     & ~    & ~    & ~ \\
2376: N1566 &  10.22 &  ~       & ~        & ~     & ~    & ~    & ~ \\
2377: N1705 &   0.50 &  ~       & ~        & ~     & ~    & ~    & ~ \\
2378: N2403 &   0.33 &  23.8    & 0.10     & 855.  & 0.96 & BIMA & E96 \\
2379: N2798 &  34.54 &  46.5    & 0.11     & 855.  & 2.50 & IRAC & E96 \\
2380: N2841 &   0.74 & $< 27.4$ & $> 0.03$ & 370.  & ~    & ~    & S05 \\
2381: N2915 & 0.0077 & $< 6.0$  & $> 0.02$ & 1452. & 2.04 & IRAC & A04 \\
2382: N2976 &   0.28 &  10.6    & 0.20     & 452.  & 1.12 & BIMA & A04 (I05, Y95) \\
2383: N3031 &   0.10 & $< 7.9$  & $> 0.09$ & 3068. & ~    & ~    & S05 \\
2384: N3049 &   3.39 &  31.3    & 0.02     & 573.  & 1.48 & IRAC & C97 \\
2385: N3184 &   2.09 &  40.1    & 0.07     & 441.  & 1.23 & BIMA & S05 (Y95, S93) \\
2386: N3190 &  11.33 &  37.8    & 0.09     & 380.  & 1.14 & IRAC & L98 \\
2387: N3198 &   3.54 &  33.1    & 0.10     & 415.  & 1.31 & IRAC & B93 \\
2388: N3265 &   6.45 &  25.8    & 0.06     & 227.  & 0.76 & IRAC & G91 \\
2389: Mrk33\tablenotemark{b}
2390:       &  21.35 &  16.5    & 0.23     & 380.  & 1.15 & IRAC & S92 (A04) \\
2391: N3351 &  11.30 &  95.1    & 0.05     & 606.  & 0.85 & BIMA & P97 (S05) \\
2392: N3521 &   4.95 &  26.1    & 0.22     & 201.  & 0.68 & BIMA & N01 (Y95) \\
2393: N3621 &   1.88 &  ~       & ~        & ~     & ~    & ~    & ~ \\
2394: N3627 &   2.39 & 196.     & 0.01     & 271.  & 0.95 & BIMA & S05 (Y95) \\
2395: N3773 &   2.98 & $< 13.5$ & $> 0.12$ & 2376. & 5.83 & IRAC & L05 \\
2396: N3938 &   4.23 &  26.4    & 0.10     & 262.  & 0.96 & BIMA & S05 (Y95) \\
2397: N4125 &   1.20 &   6.5    & 0.04     & 415.  & 1.13 & IRAC & W95 \\
2398: N4254 &   7.86 &  81.8    & 0.02     & 201.  & 0.75 & S99  & N01 (K88) \\
2399: N4321 &  47.83 & 133.     & 0.03     & 415.  & 0.74 & BIMA & B93 (S05, K96, H93) \\
2400: N4450 &   5.04 &  16.4    & 0.07     & 131.  & 0.90 & BIMA & S05 \\
2401: N4536 & 102.41 &  50.2    & 0.11     & 855.  & 1.08 & S03  & E96 \\
2402: N4552 &   1.91 &  ~       & ~        & ~     & ~    & ~    & ~ \\
2403: N4559\tablenotemark{c}
2404:       &   9.29 &  13.9    & 0.45     & 355.  & 0.98 & BIMA & S05 \\
2405: N4569 &  17.48 & 200.     & 0.02     & 118.  & 0.73 & BIMA & S05 (N01, K88, H93) \\
2406: N4579 &   2.36 &  36.7    & 0.01     & 118.  & 0.42 & BIMA & S05 (K88) \\
2407: N4594 &   0.47 & $< 6.2$  & $> 0.04$ & 346.  & ~    & ~    & B91 \\
2408: N4625 &   1.85 &   9.2    & 0.20     & 452.  & 1.10 & IRAC & A04 (B03) \\
2409: N4631 &   8.44 &  96.2    & 0.10     & 1590. & 1.76 & IRAC & Y95 \\
2410: N4725 &   5.85 & $< 54.8$ & $> 0.03$ & 210.  & ~    & ~    & S05 \\
2411: N4736 &   1.20 &  81.3    & 0.05     & 767.  & 1.37 & BIMA & S05 (S98, Y95) \\
2412: N4826 &   3.00 & 250.     & 0.04     & 1337. & 1.81 & BIMA & S05 (N01, Y95) \\
2413: N5033 &  10.86 &  96.0    & 0.05     & 855.  & 1.39 & BIMA & C01 (S05, N01, E96, A95) \\
2414: N5055 &   5.98 & 122.     & 0.06     & 645.  & 1.23 & BIMA & S05 (M99, E96, A95, Y95, S93) \\
2415: N5194 &   1.46 &  40.0    & 0.04     & 474.  & 0.98 & BIMA & S05 \\
2416: N5194\_A & 0.95 &   ~     & ~        & ~     & ~    & ~    & ~ \\
2417: N5194\_B & 3.37 &   ~     & ~        & ~     & ~    & ~    & ~ \\
2418: N5194\_C & 6.07 &   ~     & ~        & ~     & ~    & ~    & ~ \\
2419: N5194\_D & 3.53 &   ~     & ~        & ~     & ~    & ~    & ~ \\
2420: N5194\_E & 3.33 &   ~     & ~        & ~     & ~    & ~    & ~ \\
2421: N5194\_F & 3.07 &   ~     & ~        & ~     & ~    & ~    & ~ \\
2422: N5194\_G & 2.91 &   ~     & ~        & ~     & ~    & ~    & ~ \\
2423: N5195 &   5.51 & 110.     & 0.06     & 177.  & 0.58 & IRAC & K02 (S89) \\
2424: N5713 &  30.27 &  50.8    & 0.08     & 1590. & 2.01 & IRAC & Y95 (Y03) \\
2425: N5866 &   2.80 &  50.9    & 0.03     & 346.  & 1.02 & IRAC & W03 (T94) \\
2426: N6822\_A\tablenotemark{b}
2427:       & 0.0047 & 7.8   & 0.14     & 415.  & 1.26 & IRAC & I03 \\
2428: N6822\_B & 0.0062 &  ~       & ~        & ~     & ~    & ~    & ~ \\
2429: N6822\_C & 0.0030 &  ~       & ~        & ~     & ~    & ~    & ~ \\
2430: N6946 &  11.95 & 486.     & 0.03     & 816.  & 0.95 & BIMA & S05 \\
2431: N7331 &   9.69 &  56.0    & 0.07     & 346.  & 0.94 & BIMA & I99 (N01) \\
2432: N7552 & 269.01 & 171.     & 0.12     & 1452. & 1.86 & IRAC & A95 (C92) \\
2433: N7793 &   0.10 &  25.4    & 0.03     & 1452. & 1.87 & IRAC & I95 \\
2434: \enddata
2435: \tablenotetext{a}{We retain here results from
2436: constrained-$T_1$ fits, giving lower masses in the warm phase. Whenever the excitation
2437: diagrams are ambiguous regarding the value of $OPR_{\rm \,high\,T}$, we adopt the mass
2438: of warm H$_2$ derived with $OPR_{\rm \,high\,T} < 3$, which is also smaller than the mass
2439: derived with $OPR_{\rm \,high\,T} = 3$ (see Table~\ref{tab_fit}). References for CO
2440: intensities are given in abbreviated form.
2441: The quantity "corr." designates the factor that was applied to the CO brightness in
2442: order to correct for the different apertures of the CO and H$_2$ observations,
2443: and "map" is the image used to derive this factor
2444: (see Section~\ref{h2co}).}
2445: \tablenotetext{b}{Mrk\,33: The H$_2$ mass fraction in the warm phase may be severely overestimated
2446: in this blue compact galaxy, because of the use of an inappropriate factor to
2447: convert CO flux to total H$_2$ mass. \citet{Israel05} derive a factor about four
2448: times higher than the standard factor used here, which would make the total H$_2$
2449: mass also four times higher.
2450: NGC\,6822\_A: In their detailed study of this region, \citet{Israel03} derive a CO
2451: flux to total H$_2$ mass conversion factor twenty times higher than the standard factor,
2452: so that the same remark as for Mrk\,33 applies.}
2453: \tablenotetext{c}{NGC\,4559 has the highest mass fraction in the warm phase, even when
2454: constraining $T_1$ to be near-maximal. The warm phase is dominated by gas at $\sim 100$\,K.
2455: NGC\,4559 is a quiescent galaxy, as obvious from the low $F_{24}/F_{7.9}$ flux density
2456: ratio within the IRS aperture. The galaxy has little molecular gas, and it is possible
2457: that it is mostly diffuse, and thus more exposed to FUV radiation. The same remarks
2458: may apply, to a lesser degree, to other galaxies of the sample such as NGC\,628.}
2459: \end{deluxetable}
2460: 
2461: 
2462: 
2463: \begin{deluxetable}{lrll}
2464: \tabletypesize{\footnotesize}
2465: \tablecaption{Nuclear X-ray fluxes in the 2-10\,keV band and references.
2466: \label{tab_xrays}
2467: }
2468: \tablehead{
2469: galaxy & $F_{\rm X}$               & ref & note\tablenotemark{a} \\
2470: ~      & (10$^{-17}$ W\,m$^{ -2}$) & ~   & ~ \\
2471: }
2472: \startdata
2473: N628  & 0.59     & K05  & 2 \\
2474: N1097 & 173.     & N06  & 1 \\
2475: N1291 & 7.45     & K05  & 2 \\
2476: N1316 & 10.1     & KF03 & 2 \\
2477: N2841 & 1.06     & HF01 & 1 \\
2478: N3031 & 1020.    & HF01 & 1 \\
2479: N3184 & 1.12     & K05  & 2 \\
2480: N3627 & $< 0.32$ & HF01 & 1 \\
2481: N4125 & 1.07     & S04  & 1 \\
2482: N4321 & $< 1.25$ & HF01 & 1 \\
2483: N4552 & 12.1     & SD05 & 1 \\
2484: N4569 & 7.61     & HF01 & 1 \\
2485: N4579 & 264.     & HF01 & 1 \\
2486: N4594 & 120.     & HF01 & 1 \\
2487: N4725 & 7.15     & HF01 & 1 \\
2488: N4736 & 27.4     & SD05 & 1 \\
2489: N4826 & $< 1.08$ & HF01 & 1 \\
2490: N5033 & 123.     & HF01 & 1 \\
2491: N5194 & 1300.    & S04  & 1 \\
2492: N5195 & 4.45     & K05  & 2 \\
2493: N5866 & 0.18     & SD05 & 1 \\
2494: N7331 & 2.28     & S04  & 1 \\
2495: \enddata
2496: \tablenotetext{a}{1: Quantity given directly in the cited paper
2497: (abbreviated notation as in the list of references). \\
2498: 2: Quantity computed as in \citet{Ho01} (HF01) from counts in the 0.2-8\,keV band.}
2499: \end{deluxetable}
2500: 
2501: 
2502: 
2503: \clearpage
2504: 
2505: 
2506: 
2507: \begin{figure}[!ht]
2508: \vspace*{-1cm}
2509: \resizebox{12cm}{!}{\rotatebox{90}{\plotone{f1.ps}}}
2510: \caption{Histogram of the equivalent diameters of the areas over which the line and continuum
2511: fluxes were integrated. The apertures were defined by the intersection of the spectral maps
2512: in the LH, SH and SL modules of the IRS instrument. The extranuclear regions in NGC\,5194
2513: are shown by the hatched histogram, the regions within dwarfs by the cross-hatched histogram,
2514: the {\small LINER} nuclei by the darker shade and the Sy nuclei by the lighter shade.
2515: }
2516: \label{fig:diameters}
2517: \end{figure}
2518: 
2519: 
2520: 
2521: \begin{figure}[!ht]
2522: \hspace*{-2.5cm}
2523: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2a.ps}}}
2524: \hspace*{-0.55cm}
2525: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2b.ps}}}
2526: \hspace*{-0.55cm}
2527: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2c.ps}}}
2528: \hspace*{-0.55cm}
2529: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2d.ps}}} \\
2530: \hspace*{-2.5cm}
2531: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2e.ps}}}
2532: \hspace*{-0.55cm}
2533: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2f.ps}}}
2534: \hspace*{-0.55cm}
2535: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2g.ps}}}
2536: \hspace*{-0.55cm}
2537: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2h.ps}}} \\
2538: \hspace*{-2.5cm}
2539: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2i.ps}}}
2540: \hspace*{-0.55cm}
2541: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2j.ps}}}
2542: \hspace*{-0.55cm}
2543: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2k.ps}}}
2544: \hspace*{-0.55cm}
2545: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2l.ps}}} \\
2546: \hspace*{-2.5cm}
2547: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2m.ps}}}
2548: \hspace*{-0.55cm}
2549: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2n.ps}}}
2550: \hspace*{-0.55cm}
2551: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2o.ps}}}
2552: \hspace*{-0.55cm}
2553: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2p.ps}}} \\
2554: \hspace*{-2.5cm}
2555: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2q.ps}}}
2556: \hspace*{-0.55cm}
2557: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2r.ps}}}
2558: \hspace*{-0.55cm}
2559: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2s.ps}}}
2560: \hspace*{-0.55cm}
2561: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2t.ps}}}
2562: \caption{Examples of H$_2$ line spectra: the circumnuclear starbursts NGC\,1097,
2563: NGC\,6946 and NGC\,7552, and the three galaxies for which we could estimate the
2564: fluxes of higher transitions than S(3), NGC\,1266, NGC\,4569 and NGC\,4579.
2565: The straight line indicates the fitted pseudo-continuum, and the diamond symbols
2566: show the wavelength range over which the line flux was integrated.}
2567: \label{fig:lines}
2568: \end{figure}
2569: 
2570: \addtocounter{figure}{-1}
2571: \begin{figure}[!ht]
2572: \hspace*{-2.5cm}
2573: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2u.ps}}}
2574: \hspace*{-0.55cm}
2575: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2v.ps}}}
2576: \hspace*{-0.55cm}
2577: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2w.ps}}}
2578: \hspace*{-0.55cm}
2579: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2x.ps}}} \\
2580: \hspace*{-2.5cm}
2581: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2y.ps}}}
2582: \hspace*{-0.55cm}
2583: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2z.ps}}}
2584: \hspace*{-0.55cm}
2585: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2aa.ps}}}
2586: \hspace*{-0.55cm}
2587: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2ab.ps}}} \\
2588: \hspace*{-2.5cm}
2589: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2ac.ps}}}
2590: \hspace*{-0.55cm}
2591: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2ad.ps}}}
2592: \hspace*{-0.55cm}
2593: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2ae.ps}}}
2594: \hspace*{-0.55cm}
2595: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2af.ps}}} \\
2596: \hspace*{-2.5cm}
2597: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2ag.ps}}}
2598: \hspace*{-0.55cm}
2599: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2ah.ps}}}
2600: \hspace*{-0.55cm}
2601: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2ai.ps}}}
2602: \hspace*{-0.55cm}
2603: \resizebox{5cm}{!}{\rotatebox{90}{\plotone{f2aj.ps}}}
2604: \caption{(continued).
2605: }
2606: \end{figure}
2607: 
2608: 
2609: 
2610: \begin{figure}[!ht]
2611: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f3.ps}}}
2612: \caption{Critical densities for collisional deexcitation by H$_2$ for the rotational transitions in the
2613: wavelength range of the IRS instrument.
2614: }
2615: \label{fig:ncrit}
2616: \end{figure}
2617: 
2618: 
2619: 
2620: \begin{figure}[!ht]
2621: \hspace*{-2cm}
2622: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f4a.ps}}}
2623: \hspace*{-0.5cm}
2624: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f4b.ps}}}
2625: \caption{Examples of how the possible range of apparent $OPR_{\rm \,high\,T}$ is determined
2626: for each galaxy.
2627: Thermalization requires the apparent temperatures derived from each pair of transitions
2628: to be monotonic as a function of upper level energy; in particular, the conditions
2629: $T(S1-S2) < T(S1-S3) < T(S2-S3)$ (indicated by the shaded areas) have to be satisfied.
2630: The apparent temperatures derived for NGC\,3198 are compatible with $OPR_{\rm \,high\,T} = 3$,
2631: whereas they require $OPR_{\rm \,high\,T} < 3$ for NGC\,4631 (see text).
2632: }
2633: \label{fig:diag_temp}
2634: \end{figure}
2635: 
2636: 
2637: 
2638: \begin{figure}[!ht]
2639: \vspace*{-1cm}
2640: \hspace*{-2cm}
2641: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5a.ps}}}
2642: \hspace*{-0.5cm}
2643: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5b.ps}}}
2644: \hspace*{-2cm}
2645: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5c.ps}}}
2646: \hspace*{-0.5cm}
2647: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5d.ps}}}
2648: \hspace*{-2cm}
2649: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5e.ps}}}
2650: \hspace*{-0.5cm}
2651: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5f.ps}}}
2652: \vspace*{-1cm}
2653: \caption{Excitation diagrams. The $N_{\rm u} / g_{\rm u}$ ratios are normalized by the S(1)
2654: transition. The dashed line indicates the best fit (see text). Whenever $OPR_{\rm \,high\,T} < 3$
2655: is required by the temperature constraints, a dot-dash line shows for comparison
2656: the fit obtained with $OPR_{\rm \,high\,T} = 3$.
2657: }
2658: \label{fig:diag_exc}
2659: \end{figure}
2660: 
2661: \addtocounter{figure}{-1}
2662: \begin{figure}[!ht]
2663: \hspace*{-2cm}
2664: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5g.ps}}}
2665: \hspace*{-0.5cm}
2666: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5h.ps}}}
2667: \hspace*{-2cm}
2668: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5i.ps}}}
2669: \hspace*{-0.5cm}
2670: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5j.ps}}}
2671: \hspace*{-2cm}
2672: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5k.ps}}}
2673: \hspace*{-0.5cm}
2674: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5l.ps}}}
2675: \caption{(continued).
2676: }
2677: \end{figure}
2678: 
2679: \addtocounter{figure}{-1}
2680: \begin{figure}[!ht]
2681: \hspace*{-2cm}
2682: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5m.ps}}}
2683: \hspace*{-0.5cm}
2684: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5n.ps}}}
2685: \hspace*{-2cm}
2686: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5o.ps}}}
2687: \hspace*{-0.5cm}
2688: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5p.ps}}}
2689: \hspace*{-2cm}
2690: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5q.ps}}}
2691: \hspace*{-0.5cm}
2692: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5r.ps}}}
2693: \caption{(continued).
2694: }
2695: \end{figure}
2696: 
2697: \addtocounter{figure}{-1}
2698: \begin{figure}[!ht]
2699: \hspace*{-2cm}
2700: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5s.ps}}}
2701: \hspace*{-0.5cm}
2702: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5t.ps}}}
2703: \hspace*{-2cm}
2704: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5u.ps}}}
2705: \hspace*{-0.5cm}
2706: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5v.ps}}}
2707: \hspace*{-2cm}
2708: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5w.ps}}}
2709: \hspace*{-0.5cm}
2710: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5x.ps}}}
2711: \caption{(continued).
2712: }
2713: \end{figure}
2714: 
2715: \addtocounter{figure}{-1}
2716: \begin{figure}[!ht]
2717: \hspace*{-2cm}
2718: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5y.ps}}}
2719: \hspace*{-0.5cm}
2720: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5z.ps}}}
2721: \hspace*{-2cm}
2722: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5aa.ps}}}
2723: \hspace*{-0.5cm}
2724: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ab.ps}}}
2725: \hspace*{-2cm}
2726: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ac.ps}}}
2727: \hspace*{-0.5cm}
2728: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ad.ps}}}
2729: \caption{(continued).
2730: }
2731: \end{figure}
2732: 
2733: \addtocounter{figure}{-1}
2734: \begin{figure}[!ht]
2735: \hspace*{-2cm}
2736: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ae.ps}}}
2737: \hspace*{-0.5cm}
2738: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5af.ps}}}
2739: \hspace*{-2cm}
2740: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ag.ps}}}
2741: \hspace*{-0.5cm}
2742: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ah.ps}}}
2743: \hspace*{-2cm}
2744: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ai.ps}}}
2745: \hspace*{-0.5cm}
2746: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5aj.ps}}}
2747: \caption{(continued).
2748: }
2749: \end{figure}
2750: 
2751: \addtocounter{figure}{-1}
2752: \begin{figure}[!ht]
2753: \hspace*{-2cm}
2754: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ak.ps}}}
2755: \hspace*{-0.5cm}
2756: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5al.ps}}}
2757: \hspace*{-2cm}
2758: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5am.ps}}}
2759: \hspace*{-0.5cm}
2760: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5an.ps}}}
2761: \hspace*{-2cm}
2762: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ao.ps}}}
2763: \hspace*{-0.5cm}
2764: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ap.ps}}}
2765: \caption{(continued).
2766: }
2767: \end{figure}
2768: 
2769: \addtocounter{figure}{-1}
2770: \begin{figure}[!ht]
2771: \hspace*{-2cm}
2772: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5aq.ps}}}
2773: \hspace*{-0.5cm}
2774: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ar.ps}}}
2775: \hspace*{-2cm}
2776: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5as.ps}}}
2777: \hspace*{-0.5cm}
2778: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5at.ps}}}
2779: \hspace*{-2cm}
2780: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5au.ps}}}
2781: \hspace*{-0.5cm}
2782: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5av.ps}}}
2783: \caption{(continued).
2784: }
2785: \end{figure}
2786: 
2787: \addtocounter{figure}{-1}
2788: \begin{figure}[!ht]
2789: \hspace*{-2cm}
2790: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5aw.ps}}}
2791: \hspace*{-0.5cm}
2792: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ax.ps}}}
2793: \hspace*{-2cm}
2794: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ay.ps}}}
2795: \hspace*{-0.5cm}
2796: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5az.ps}}}
2797: \hspace*{-2cm}
2798: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5ba.ps}}}
2799: \hspace*{-0.5cm}
2800: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bb.ps}}}
2801: \caption{(continued).
2802: }
2803: \end{figure}
2804: 
2805: \addtocounter{figure}{-1}
2806: \begin{figure}[!ht]
2807: \hspace*{-2cm}
2808: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bc.ps}}}
2809: \hspace*{-0.5cm}
2810: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bd.ps}}}
2811: \hspace*{-2cm}
2812: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5be.ps}}}
2813: \hspace*{-0.5cm}
2814: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bf.ps}}}
2815: \hspace*{-2cm}
2816: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bg.ps}}}
2817: \hspace*{-0.5cm}
2818: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bh.ps}}}
2819: \caption{(continued).
2820: }
2821: \end{figure}
2822: 
2823: \addtocounter{figure}{-1}
2824: \begin{figure}[!ht]
2825: \hspace*{-2cm}
2826: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bi.ps}}}
2827: \hspace*{-0.5cm}
2828: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bj.ps}}}
2829: \hspace*{-2cm}
2830: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bk.ps}}}
2831: \hspace*{-0.5cm}
2832: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bl.ps}}}
2833: \hspace*{-2cm}
2834: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bm.ps}}}
2835: \hspace*{-0.5cm}
2836: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f5bn.ps}}}
2837: \caption{(continued).
2838: }
2839: \end{figure}
2840: 
2841: 
2842: 
2843: \clearpage
2844: 
2845: 
2846: 
2847: \begin{figure}[!ht]
2848: \hspace*{-2cm}
2849: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f6a.ps}}}
2850: \hspace*{-0.5cm}
2851: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f6b.ps}}}
2852: \caption{
2853: {\bf Left:} Histograms of the derived column densities in fits where the lower temperature is
2854: constrained (top) and where it is free to vary (bottom). In the latter case, the solid line
2855: indicates results with low $T_2$ values and the dashed line with high $T_2$ values, in
2856: $OPR_{\rm \,high\,T} < 3$ fits (see text).
2857: {\bf Right:} Column densities as a function of the total stellar mass of the host galaxy,
2858: estimated as by \citet{Lee06}. Star-forming nuclei are represented as diamonds,
2859: the extranuclear regions in NGC\,5194 and NGC\,6822 as squares, {\small LINER} nuclei
2860: as star symbols and Sy nuclei as crosses. By definition, dwarf galaxies have
2861: $M*_{\rm host} < 10^{9.7}$\,M$_{\sun}$.
2862: }
2863: \label{fig:coldens}
2864: \end{figure}
2865: 
2866: 
2867: 
2868: \begin{figure}[!ht]
2869: \resizebox{15cm}{!}{\plotone{new_f7.eps}}
2870: \caption{Fraction of H$_2$ in the warm phase ($T > T_1$) as a function of $T_1$, the lower temperature
2871: of the two components fitted to the rotational lines. Nuclei classified as purely star-forming are
2872: shown as diamonds, and regions in dwarf galaxies as filled diamonds; nuclei classified as
2873: {\small LINER} or Sy are shown as star symbols or triangles, respectively.
2874: The symbol size is increased for targets with the most robust CO flux estimates (see text).
2875: In the free-$T_1$ fits with $OPR_{\rm \,high\,T} < 3$, results with the two values of $T_2$
2876: considered here (see text) are connected by a line segment; the fits with
2877: $T_2 = 1.5 \times T{\rm (S1-S3)}$ produce higher temperatures and lower warm H$_2$ masses
2878: than the fits with $T_2 = 1.14 \times T{\rm (S1-S3)}$.
2879: }
2880: \label{fig:frac_warm}
2881: \end{figure}
2882: 
2883: 
2884: 
2885: \begin{figure}[!ht]
2886: \vspace*{-3.5cm}
2887: \resizebox{15cm}{!}{\plotone{new_f8_ab.eps}}
2888: \vspace*{-0.12cm} \\
2889: \resizebox{15cm}{!}{\plotone{new_f8_cd.eps}}
2890: \vspace*{-0.5cm}
2891: \caption{
2892: Ratio of the power emitted in the sum of the S(0) to S(2) transitions to:
2893: (a) the total infrared power; (b) the [Si{\small II}] line power; (c) the power emitted
2894: in the aromatic bands within the IRAC4 filter (assuming a filter width of 13.9\,THz
2895: and subtracting the stellar emission as mentioned in Sect.~\ref{images});
2896: (d) the power emitted at 24\,$\mu$m within the MIPS1 filter (assuming a filter
2897: width of 3.1\,THz). The symbol coding is as in Fig.~\ref{fig:frac_warm}.
2898: Excluding the regions within NGC\,6822 (the targets with the smallest projected
2899: aperture), as well as NGC\,1705 and NGC\,2915 (the galaxies with the smallest H$_2$
2900: brightness), the average and dispersion of each logarithmic power ratio is:
2901: (a) $-3.36 \pm 0.15$ for H{\small II} nuclei and complexes;
2902: and $-3.12 \pm 0.24$ for {\small LINER} and Sy nuclei;
2903: (b) $-0.53 \pm 0.17$ and $-0.04 \pm 0.42$, respectively;
2904: (c) $-2.19 \pm 0.10$ and $-1.80 \pm 0.34$; (d) $-1.85 \pm 0.28$ and $-1.51 \pm 0.31$.
2905: Note that the H$_2$ flux (abscissa) is almost proportional to the H$_2$ brightness,
2906: because of the quasi-uniform beam.
2907: }
2908: \label{fig:frac_tir}
2909: \end{figure}
2910: 
2911: 
2912: 
2913: \begin{figure}[!ht]
2914: \resizebox{10cm}{!}{\plotone{new_f9.eps}}
2915: \caption{Power ratios of H$_2$ to aromatic bands (shown in Fig.~\ref{fig:frac_tir}c) and 24\,$\mu$m
2916: emission to aromatic bands as a function of
2917: $P_{24} = 1.05 ((\nu_{24} F_{24} - 0.14 \nu_{7.9} F_{\rm 7.9\,dust}) / (\nu_{71} F_{71} + \nu_{156} F_{156}) - 0.035)^{0.75}$,
2918: a quantity closely related to $f(L_{\rm dust}~; U_{\rm rad} > 100)$ defined by
2919: \citet{Draine07b}, which is the fraction of the total dust luminosity coming from
2920: regions with radiation field intensities more than 100 times the average local field.
2921: Only H{\small II} nuclei and complexes are shown. Dwarfs are represented by filled diamonds
2922: and the extranuclear regions within NGC\,5194 by thick squares.
2923: }
2924: \label{fig:frac_pah_frac_highU}
2925: \end{figure}
2926: 
2927: 
2928: 
2929: \begin{figure}[!ht]
2930: \resizebox{15cm}{!}{\plotone{new_f10.eps}}
2931: \caption{Ortho to para ratios as a function of the surface brightness in the sum of the
2932: S(0) to S(2) transitions and as a function of the surface brightness in the 24\,$\mu$m band.
2933: For a number of sources, based on the excitation diagram, we adopted a fixed
2934: $OPR_{\rm \,high\,T} = 3$ (see text). Galaxies with upper limits in the S(2) line, or with
2935: indications of possibly non-negligible optical depth at 10\,$\mu$m, are shown as lower
2936: limits of $OPR_{\rm \,high\,T}$. For free-$T_1$ fits, results obtained from the two values
2937: of $T_2$ considered here are connected by a line segment. The symbol coding is as
2938: in Fig.~\ref{fig:frac_warm}, except that {\small LINER} and Sy nuclei are both shown
2939: as star symbols.}
2940: \label{fig:opr_sb}
2941: \end{figure}
2942: 
2943: 
2944: 
2945: \begin{figure}[!ht]
2946: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f11.ps}}}
2947: \caption{Excess H$_2$ luminosity as a function of 2-10\,keV X-ray luminosity. The excess
2948: H$_2$ emission is defined from the relation between H$_2$ and aromatic band power, shown
2949: in Figure~\ref{fig:frac_tir}c, as the difference between the total H$_2$ emission and the
2950: quantity $10^{-1.94} \times F_{\rm 7.9\,dust}$, which defines the upper envelope of H{\small II}
2951: nuclei. Galaxies with no H$_2$ excess according to this definition were arbitrarily placed
2952: at an ordinate of 2.25 below the dashed line. The size of the symbols is proportional to
2953: the fraction of excess H$_2$ to total H$_2$ emission, coded as in Fig.~\ref{fig:snshocks}.
2954: Sy nuclei are marked by an overlying horizontal bar, and star-forming nuclei by a square.
2955: The large scatter and the high ratios of excess H$_2$ to X-ray luminosities argue against
2956: X-rays playing a dominant role in the additional (nonstellar) excitation of H$_2$ in
2957: {\small LINER}/Sy nuclei (see text).
2958: }
2959: \label{fig:xrays}
2960: \end{figure}
2961: 
2962: 
2963: 
2964: \begin{figure}[!ht]
2965: \resizebox{10cm}{!}{\rotatebox{90}{\plotone{f12.ps}}}
2966: \caption{Heating efficiency required to account for the excess H$_2$ emission by supernova
2967: remnant shocks (ratio of the power in the sum of the S(0)--S(2) lines to the total
2968: mechanical power), as a function of the excess H$_2$ luminosity, defined as in
2969: Fig.~\ref{fig:xrays}. The dashed line indicates the maximum efficiency expected if all
2970: the mechanical power produced by supernova remnants is absorbed in molecular clouds,
2971: for the model parameters of \citet{Kaufman96}.
2972: The size of the symbols is proportional to the fraction of excess H$_2$ to total H$_2$
2973: emission. Sy nuclei are marked by an overlying horizontal bar. The regions B and C within
2974: NGC\,6822 are the only non-AGN targets in this figure.
2975: }
2976: \label{fig:snshocks}
2977: \end{figure}
2978: 
2979: 
2980: \end{document}
2981: