WARM MOLECULAR GAS IN M51: MAPPING THE EXCITATION TEMPERATURE
AND MASS OF H2 WITH THE SPITZER INFRARED SPECTROGRAPH

Gregory Brunner
1

Department of Physics and Astronomy, Rice University, Houston, TX 77005; gbrunner@rice.edu

Kartik Sheth and Lee Armus

Spitzer Science Center, Caltech, Pasadena, CA 91125

Mark Wolfire and Stuart Vogel

Department of Astronomy, University of Maryland, College Park, MD 20741

Eva Schinnerer

Max-Planck-Institut für Astronomie, Königstuhl 17, 69117 Heidelberg, Germany

George Helou

Spitzer Science Center, Caltech, Pasadena, CA 91125

Reginald Dufour

Department of Physics and Astronomy, Rice University, Houston, TX 77005

John-David Smith

University of Arizona, Steward Observatory, Tucson, AZ 85721

and

Daniel A. Dale

Department of Physics and Astronomy, University of Wyoming, Laramie, WY 82071

Received 2007 July 13; accepted 2007 October 10

ABSTRACT

We have mapped the warm molecular gas traced by the H2 S(0)–H2 S(5) pure rotational mid-infrared emission lines
over a radial strip across the nucleus and disk of M51 (NGC 5194) using the Infrared Spectrograph (IRS) on the Spitzer
Space Telescope. The six H2 lines have markedly different emission distributions. We obtained the H2 temperature and
surface density distributions by assuming a two-temperature model: a warm (T ¼ 100 300 K) phase traced by the low
J [S(0)–S(2)] lines and a hot phase (T ¼ 400 1000 K) traced by the high J [S(2)–S(5)] lines. The lowest molecular gas
temperatures are found within the spiral arms (T �155 K), while the highest temperatures are found in the inter-arm
regions (T > 700 K). The warm gas surface density reaches a maximum of 11 M� pc

�2 in the northwest spiral arm,
whereas the hot gas surface density peaks at 0.24 M� pc

�2 at the nucleus. The spatial offset between the peaks in the
different phases suggests that the warm phase is more efficiently heated by star formation activity and the hot phase is
more efficiently heated by nuclear activity. The warm H2 is found in the dust lanes of M51 and is generally spatially co-
incident with the cold molecular gas traced by CO emission, consistent with excitation of the warm phase in dense pho-
todissociation regions. The hot H2 is most prominent in the nuclear region. Here, the hot H2 coincides with [O iv]
(25.89 �m) and X-ray emission indicating that shocks and/or X-rays are responsible for exciting this phase.

Subject headinggs: galaxies: individual (M51) — galaxies: ISM — ISM: molecules

1. INTRODUCTION

Star formation and galactic evolution are connected via the
molecular gas in a galaxy. In the Milky Way, star formation oc-
curs in molecular clouds, although not all clouds are actively
forming stars. On a global, galactic scale, star formation may be
triggered whenever the molecular gas surface density is enhanced,
for example, by a spiral density wave (Vogel et al. 1988), by in-
creased pressure orgasdensity in galactic nuclei (Young & Scoville
1991; Sakamoto et al. 1999; Sheth et al. 2005), by hydrodynamic
shocks along the leading edge of bars (Sheth et al. 2000, 2002),
and in the transition region at the ends of bars (Kenny & Lord
1991; Sheth et al. 2002). How does this star formation affect the
surrounding molecular gas? How is it heated and what is the dis-
tribution of the gas temperatures? How does the mass of the
warm and hot gas vary from region to region? We address these

questions using spectral line maps from a radial strip across the
grand-design spiral galaxy, M51.
M51 (the Whirlpool galaxy, NGC 5194) is a nearby, face-on

spiral galaxy that is rich in molecular gas. Its proximity (assumed
to be 8.2 Mpc; Tully 1988), face-on orientation, and grand-
design spiral morphology make it the ideal target for studies of
theinterstellar medium (ISM) across distinct dynamical, chemical,
and physical environments in a galaxy. Studies of the molecular
gaswithinM51haverevealedgiantmolecularassociations(GMAs)
along the spiral arms (Vogel et al. 1988; Rand & Kulkarni 1990;
Aalto et al. 1999), a reservoir of molecular gas in the nuclear
region that is massive enough to fuel the active galactic nucleus
(AGN; Scoville et al. 1998), and spiral density wave triggered
star formation in molecular clouds (Vogel et al. 1988). In addition
to being well studied at millimeter and radio wavelengths, M51
has also been studied at X-ray, UV, optical, near-infrared, infrared,
and submillimeter wavelengths (Palumbo et al. 1985; Terashima
et al. 1998; Scoville et al. 2001; Calzetti et al. 2005; Matsushita
et al. 2004; Meijerink et al. 2005).

1
Visiting Graduate Student Fellow, Spitzer Science Center, Caltech, Pasadena,

CA 91125.

316

The Astrophysical Journal, 675:316–329, 2008 March 1

# 2008. The American Astronomical Society. All rights reserved. Printed in U.S.A.



In this paper we present maps of the H2 S(0)–H2 S(5) pure ro-
tational mid-infrared lines over a strip across M51 created from
Spitzer Space Telescope (Spitzer) Infrared Spectrograph (IRS)
spectral mapping mode observations. The mid-infrared H2 lines
trace the warm (T ¼ 100 1000 K) phase of H2 and we use these
lines to model the H2 excitation temperature, mass (Rigopoulou
et al. 2002; Higdon et al. 2006), and ortho-to-para ratio (Neufeld
et al. 1998, 2006) across the M51 strips.

2
We use the inferred dis-

tributions to place constraints on the energy injection mechanisms
(i.e., radiative heating, shocks, turbulence) that heat the warm
molecular gas phase of the ISM.

2. OBSERVATIONS AND DATA REDUCTION

2.1. Spectral Data

We mapped a radial strip across M51 using the short-low (SL;
5–14.5 �m) and long-low (LL; 14-38 �m) modules of the Spitzer
IRS in spectral mapping mode (Houck et al. 2004). The radial
strips were 324 00 ; 57 00 and 295 00 ; 51 00 in the SL and LL, re-

spectively. Each slit position was mapped twice with half-slit spac-
ings. In total, 1412 spectra were taken in the SL and 100 were
taken in the LL. Integration times for individual spectra were
14.6 s in both the SL and LL. Dedicated off-source background
observations were taken for the SL observations. Backgrounds
for the LL observations were taken from outrigger data collected
while the spacecraft was mapping in the adjacent module. Fig-
ure 1 presents the astronomical observation requests (AORs) over-
laid on the Spitzer Infrared Array Camera (IRAC) 8 �m image of
M51.

The spectra were assembled from the basic calibration data
(BCD) into spectral data cubes for each module using CUBISM
(Kennicutt et al. 2003; Smith et al. 2004, 2007a). Background
subtraction and bad pixel removal were done within CUBISM.
The individual BCDs were processed using the S14.0 version of
the Spitzer Science Center (SSC) pipeline. In CUBISM, the SL
and LL data cubes have 1.8500 and 5.0800 pixels, respectively. The
pixel size is half the full width at half-maximum (FWHM) of the
point-spread function (PSF) at the red end of a given module. In
principle the PSF should vary with wavelength, but since the PSF
is undersampled at the blue end of the module it is approximately
constant across a given module. So the approximate resolution of

2
While we are always exploring the warm phase of H2, in this paper we refer

to a warm and a hot phase corresponding to temperatures of T ¼ 100 300 K and
T ¼ 400 1000 K, respectively.

Fig. 1.—Overlay of the IRS footprints on the Spitzer IRAC 8 �m image of M51. The three parallel strips mark the SL (5–14.5 �m) observations with the white strip
denoting the SL2 (5–7.5 �m) coverage and the gray strip denoting the SL1(7.5–14.5 �m) coverage. The central strip is where the SL1 and SL2 overlap along with the LL (14–
38 �m) radial strip, giving us complete coverage of the mid-infrared spectrum across the nucleus and disk of M51. The small off-galaxy strip is the SL background observation.

MOLECULAR HYDROGEN IN NGC 5194 317



the SL and LL modules and maps of spectral features observed in
the SL and LL modules is 3.700 and 10.100, respectively.

We created continuum-subtracted line flux maps of the H2
S(0)–H2 S(5) lines using a combination of PAHFIT (Smith et al.
2007b) and our own code. PAHFIT is a spectral fitting routine
that decomposes IRS low-resolution spectra into broad PAH
features, unresolved line emission, and grain continuum with the
main advantage being that it allows one to recover the full line
flux of any blended features. Several H2 lines are blended with
PAH features and atomic lines in IRS low-resolution spectra: H2
S(1) with the 17.0 �m PAH complex, H2 S(2) with the 12.0 and
12.6 �m PAH complexes, H2 S(4) with the 7.8 and 8.6 �m PAH
complexes, and H2 S(5) with the [Ar ii] (6.9 �m) line. PAHFIT
also solves for the foreground dust emission and dereddens the
emitted line intensities. Our code concatenates SL1 and SL2, and
LL1 and LL2 data cubes into two cubes, one for SL and one for
LL, and smoothes each map in the cubes by a 3 pixel ; 3 pixel
box, conserving the flux, to increase the signal-to-noise ratio of
the spectra. Then, for each pixel, the spectrum is extracted and
PAHFIT is run to decompose it. Our code saves the location of
the pixel on the sky along with the PAHFIToutput (i.e., integrated
line flux, line FWHM, line equivalent width, the uncertainty in the
line flux, the fit to the entire spectrum and the fit to the continuum)
for each spectrum and uses this information to construct line flux
maps for all of the mid-infrared features. In addition to creating
line maps, our code creates maps of line FWHM, line equivalent
width, and uncertainty in the flux and data cubes of the fit to the

entire spectrum, the fit to the continuum, a continuum-subtracted
data cube, and a residual data cube. These parameters (the uncer-
tainty in the line flux, line FWHM, and line equivalent width) are
all determined within PAHFIT. For a more thorough discussion
of how PAHFIT determines quantities such as the uncertainty in
the integrated line flux, line FWHM, and line equivalent width,
see Smith et al. (2007b).
Figure 2 presents sample Spitzer IRS low-resolutionspectra and

the PAHFITspectral decomposition for three different single-pixel
regions across the M51 strip. The SL and LL spectra are plotted
separately because we decomposed the SL and LL data cubes
separately. The locations of the regions are marked on the maps
of the H2 surface density in Figure 5. While we have decomposed
the entire spectrum and created maps of all of PAH features and
spectral lines, in this paper we focus primarily onthe H2 linemaps.

2.2. Ancillary Data: CO (J = 1–0), Optical V Band, H�,
and X-Ray Observations

In this section we briefly discuss the ancillary data that we have
used in order to understand H2 excitation in M51. We take note of
the image resolutions for comparison to the Spitzer IRS beam. The
Berkeley-Illinois-Maryland Association (BIMA) CO (J ¼ 1 0)
map was acquired as part of the BIMA Survey of Nearby Gal-
axies (SONG; Regan et al. 2001; Helfer et al. 2003). At the dis-
tance of M51, the SONG beam (5:8 00 ; 5:1 00) subtends 220 pc ;
190 pc. The optical V band and continuum-subtracted H� þ ½N ii�
images of M51 were obtained from the Spitzer Infrared Nearby

Fig. 2.—Shown are sample Spitzer IRS low-resolution spectra and the PAHFITspectral decomposition for three different single-pixel regions across the M51 strip. The SL
and LL spectra are plotted separately because we decomposed the SL and LL data cubes separately. The locations of the regions are marked on the maps of the H2 surface
density in Fig. 5.

BRUNNER ET AL.318 Vol. 675



Galaxies Survey (SINGS) archive.
3
The native pixel scale for

both images is 0.300 and the angular resolutions are �100. X-ray
emission from M51 was observed by the Advanced CCD Im-
aging Spectrometer (ACIS) on the Chandra X-Ray Observatory
on 2000 June 20. The resolution of the image is �100. The X-ray
image that we use has been presented and discussed in Terashima
& Wilson (2001).

3. RESULTS

3.1. The Distribution of H2 Emission

We have detected and mapped H2 emission from the six lowest
pure rotational H2 lines (Fig. 3). The maps reveal remarkable
differences in the distribution of H2 emission in M51. H2 S(0)
emission is strongest in the northwest inner spiral arm peaking
at a flux of 3:7 ; 10�18 W m�2 and decreases by a factor of 2 in
the nuclear region. In contrast, the H2 S(1) emission peaks in
the nucleus of the galaxy at 1:0 ; 10�17 W m�2 and has an ex-
tension of equal brightness toward the northwest inner spiral
arm. In the spiral arm itself, the H2 S(0) peak is offset from the
H2 S(1) emission by 10

00 (�380 pc). We detect emission from
the H2 S(0) and H2 S(1) lines to the outer limit of our radial strip,
�6 kpc from the nucleus of the galaxy. In the outer spiral arms,
the H2 S(0) flux is a factor of 2 times lower than in the inner north-
west spiral arm and the H2 S(1) flux is a factor of 5 times lower
than in the nucleus.

The H2 S(2)–H2 S(5) maps show different molecular gas dis-
tributions within M51 through each H2 line. The brightest H2

S(2) emission is from the nucleus at 2:2 ; 10�18 W m�2. We also
see bright H2 S(2) emission from the inner northwest spiral
arm at half the flux of the nuclear peak. The H2 S(3) peak at the
nucleus is 1:4 ; 10�17 W m�2, a factor of �6 greater than the H2
S(2) nuclear peak. There is also a linear barlike structure in H2
S(3) emission across the nucleus of the galaxy at a P:A: � �100
and length of �1.5 kpc. The emission peaks in the H2 S(2) and
H2 S(3) maps are not spatially coincident. For instance, in the north-
west inner spiral arm the brightest H2 S(3) emission is further down
the spiral arm than the brightest H2 S(2) emission. Offsets like
these suggest variations in the excitation temperature from re-
gion to region within a galaxy, and even within a spiral arm.

The H2 S(4) and H2 S(5) lines are brightest at the nucleus with
fluxes of 3.1 and 8:0 ; 10�18 W m�2, respectively. The H2 S(4)
line shows emission in the nucleus and in the spiral arm to the
west. In the spiral arm to the west of the nucleus, the H2 S(4) flux
is 2:1 ; 10�18 W m�2. This is notable because the spiral arm to
the southwest of the nucleus is very bright in CO and studies
have revealed very high molecular gas column densities in the
southwest inner spiral arm (Lord & Young 1990; Aalto et al. 1999).
H2 S(5) emission is asymmetric in the nucleus and mimics the
morphology of the H2 S(3) line with extended emission to the north
of the nucleus. The trail of intensity peaks along the long northeast
edge of the H2 S(4) and H2 S(5) maps is not real. It is likely due to
difficulty smoothing the maps near the edges. These regions do not
affect the determination of the H2 temperature and surface density
due to the offset of the SL strip relative to the LL strip.

Based on the uncertainty maps generated while creating the
H2 line maps, errors in the H2 S(0) line flux that range from 15%
to 75% with the largest uncertainties being found in the interarm
regions and the outer northwest and southeast spiral arms. Errors3 See http://data.spitzer.caltech.edu/popular/sings.

Fig. 3.—Maps of the H2 S(0) (top left), H2 S(1) (top middle), H2 S(2) (top right), H2 S(3) (bottom left), H2 S(4) (bottom middle), and H2 S(5) (bottom right) line fluxes across
the SL and LL strips that we mapped with the Spitzer IRS. The H2 S(0) and H2 S(1) maps are created from the LL data cube. The H2 S(2), H2 S(3), H2 S(4), and H2 S(5) maps are
created from the SL data cube. The gray scale is in units of W m�2. Contour levels are at 3:7 ; 10�19, 7:3 ; 10�19, 1:1 ; 10�18, 1:5 ; 10�18, 1:8 ; 10�18, 2:2 ; 10�18, and
2:9 ; 10�18 W m�2 for H2 S(0); 1:1 ; 10

�18, 2:1 ; 10�18, 3:2 ; 10�18, 4:3 ; 10�18, 5:4 ; 10�18, 6:4 ; 10�18, 7:5 ; 10�18, 8:6 ; 10�18, and 9:6 ; 10�18 W m�2 for H2 S(1);
2:2 ; 10�19, 4:4 ; 10�19, 6:7 ; 10�19, 8:9 ; 10�19, and 1:1 ; 10�18 W m�2 for H2 S(2); 1:3 ; 10

�18, 4:0 ; 10�18, 6:7 ; 10�18, 9:4 ; 10�18, and 1:2 ; 10�17 W m�2 for H2 S(3);
1:0 ; 10�18 and 2:0 ; 10�18 W m�2 for H2 S(4); and 8:0 ; 10

�19, 4:0 ; 10�18, and 7:3 ; 10�18 W m�2 for H2 S(5). The vertical axis is the right ascension and the horizontal axis is
the declination. In all of the maps, north is up, east is to the left, and the plus sign denotes the nucleus of the galaxy. The different spiral arm regions are labeled on the H2 S(0)
and H2 S(1) maps in order to aid in discussing the molecular gas morphologies. The bar in the bottom right corner of each map represents 1 kpc. The box around each map
represents the SL or LL strip that we mapped.

MOLECULAR HYDROGEN IN NGC 5194 319No. 1, 2008



in the H2 S(1) line fluxes range from 10% to 70% with the largest
uncertainties being found at the distant northwest and southeast
portions of the M51 strip. Errors in the H2 S(2) and H2 S(3) line
fluxes range from 10% to 85% with the largest uncertainties
being found toward the interarm regions. Errors in the H2 S(4)
and H2 S(5) line fluxes are �30% and 20% within a 1 kpc radius
of the nucleus and become higher moving radially away.

3.2. Mapping H2 Excitation Temperature and Surface Density
across M51

3.2.1. Modeling H2 Excitation Temperature and Surface Density

The pure rotational lines of molecular hydrogen provide a
powerful probe of the conditions of the ISM by placing constraints
on the energy injection that excites H2. For example, Neufeld et al.
(2006) discuss shock excitation of H2 and Kaufman et al. (2006)
discuss H2 excitation in photodissociation regions (PDRs). Us-
ing the maps of H2 emission, we modeled the H2 temperature

and mass distribution over the radial strip in M51 following the
methods described in Rigopoulou et al. (2002) and Higdon et al.
(2006).
First, we smoothed the H2 S(1)–H2 S(5) maps to the resolution

of the H2 S(0) map, 10.1
00. The maps were then interpolated to the

same spatial grid. Excitation diagrams across the strip were de-
rived from the Boltzmann equation

Ni=N ¼ g(i)=Z Texð Þ½ � exp �Ti=Texð Þ; ð1Þ

where g(i) is the statistical weight of state i, Z(Tex) is the parti-
tion function, Ti is the energy level of a given state, and Tex is the
excitation temperature. N and Ni are the total column density and
the column density of a given state i and Ni is determined directly
from the measured extinction-corrected flux by

Ni ¼ 4� ; Cux(i)= �A(i)h�(i)½ �; ð2Þ

TABLE 1

H
2
Parameters

Transition

Wavelength

(�m)

Rotational State

(J )

Energy

(E k�1)
A

(s�1)
Statistical Weight

(g)

H2(0–0) S(0) .................... 28.22 2 510 2.94 ; 10
�11 5

H2(0–0) S(1) .................... 17.04 3 1015 4.76 ; 10
�10 21

H2(0–0) S(2) .................... 12.28 4 1682 2.76 ; 10
�9 9

H2(0–0) S(3) .................... 9.66 5 2504 9.84 ; 10
�9 33

H2(0–0) S(4) .................... 8.03 6 3474 2.64 ; 10
�8 13

H2(0–0) S(5) .................... 6.91 7 4586 5.88 ; 10
�8 45

Note.—The statistical weight (g) is (2J þ 1)(2I þ 1) where I equals 1 for odd J transitions (ortho transitions) and I equals 0 for
even J transitions (para transitions).

Fig. 4.—Excitation diagrams and the fits to the warm and hot H2 phases taken from three different single-pixel regions along the M51 strip. The different regions are
marked on the H2 surface density maps in Fig. 5.

BRUNNER ET AL.320 Vol. 675



where A(i) is the Einstein A-coefficient, �(i) is the frequency
of state i, � is the solid angle of the beam, and h is Planck’s con-
stant. Table 1 lists the values for the wavelength, rotational state,
Einstein A-coefficient, energy, and statistical weight of the pure
rotational levels of H2.

In order to derive temperature and surface density distributions
we assume a two-temperature model for the H2. To determine the
hot (T ¼ 400 1000 K) phase temperature, we do a least-squares
fit to the H2 S(2)–H2 S(5) column densities in the excitation dia-
gram at every pixel in our maps. We then subtract the contribution
of the hot phase from the lower J lines and do a least-squares fit to
the column densities of H2 S(0)–H2 S(2) lines at every pixel in our
maps to determine the temperature distribution of the warm (T ¼
100 300 K) phase. The warm and hot phase surface density dis-
tributions are derived from the column densities of the H2 S(0) and
H2 S(3) lines, respectively. The column densities are easily con-
verted to mass surface density by determining the mass of the
warm and hot H2 within every pixel.

Figure 4 shows excitation diagrams and the fits to the warm
and hot H2 phases for three different regions across the M51
strip. The three regions are marked on the H2 surface density maps
in Figure 5. In the nuclear region (Region 2) the ortho and para
levels appear to lie along the same curve, indicating an ortho-to-
para ratio (OPR) of 3; the excitation diagrams do not exhibit the
‘‘zigzag’’ characteristic of a nonequilibrium H2 OPR (Neufeld
et al. 1998; Fuente et al. 1999). The slope of the curve appears to
decrease as the rotational state increases, indicating that we are
sampling a continuous range of H2 temperatures with the higher
rotational states [S(2)–S(5)] probing a hotter phase of H2 than
the lower rotational states [S(0)–S(2)]. In the southeast and
northwest spiral arms (Regions 1 and 3, respectively) the lower J
[H2 S(0)–H2 S(3)] levels exhibit an OPR of �3. The H2 S(4)
measurement shows significant scatter in the excitation diagrams
outside of the nuclear region of M51. This would indicate that
the OPR is less than 3, however, due to the low signal-to-noise
ratio of the H2 S(4) map, we do not believe that the OPR deter-
mined from the H2 S(4) flux reflects the OPR of the warm H2. We
assume an OPR of 3 when deriving the H2 temperature and mass
distributions, which is consistent with Roussel et al. (2007) who
found an OPR of 3 for the nuclear region and four of seven H ii
regions studied in M51.

3.2.2. Warm and Hot H2 Excitation Temperature and Surface
Density Distributions

Figure 5 presents the warm (left) and hot (right) H2 surface
density distributions across the M51 strip.

4
The highest gas sur-

face density for the warm H2 phase is in the inner northwest spi-
ral arm at 11 M� pc

�2. The gas surface density in the outer
northwest and southeast spiral arms is maximum at the center
of the spiral arms at 3.5 and 1.0 M� pc

�2, respectively. The hot
phase surface density is highest in the nucleus and interior to the

Fig. 5.—Warm (T ¼ 100 300 K) H2 (left) and hot (T ¼ 400 1000 K) H2 (right) surface density distributions. The surface density distributions are in units of M� pc�2.
Contours are overplotted for clarity. The warm H2 surface density contour levels are at 1.10, 2.21, 3.32, 4.43, 5.55, and 8.85 M� pc

�2. The hot H2 contour levels are at intervals
of 10% of 0.25 M� pc

�2. The hot H2 surface density distribution is derived from the fit to the H2 S(2)–H2 S(5) lines and the warm H2 surface density distribution is derived from
the fit to the H2 S(0)–H2 S(2) lines, corrected for the contribution of the hot H2 phase. The three circles denote the regions of the spectra and excitation diagrams in Figs. 2
and 4, respectively.

Fig. 6.—Warm (T ¼ 100 300 K) H2 surface density distribution (in contours)
compared to the warm H2 temperature distribution (in gray scale, in units Kelvin).
The warm H2 temperature and surface density distributions are derived from the fit
to the excitation diagrams across the strip for the H2 S(0)–H2 S(2) lines, corrected
for the contribution of the hot (T ¼ 400 1000 K) H2 phase. Surface density con-
tour levels are at 1.10, 2.21, 3.32, 4.43, 5.55, and 8.85 M� pc

�2 (same as in Fig. 5).
The nonrectangular shape of the map is due to the slight offset of the Spitzer IRS SL
strip relative to the LL strip. In both the northwest and southeast corners of the
maps, the temperature and surface density distributions could not be determined
due to no detection of the H2 S(0) line.

4
Note that the nonrectangular shape of the strip is due to the offset between

the SL and LL strips.

MOLECULAR HYDROGEN IN NGC 5194 321No. 1, 2008



inner spiral arm at 0.24 M� pc
�2. The gas surface density of

the hot phase in the spiral arms is 3–5 times lower than that of
the nuclear region.

Figures 6 and 7 show the distribution of the warm and hot
H2 temperatures (in gray scale) with the contours of the warm
and hot H2 surface densities overlaid, respectively. In both cases
we see that the temperature and surface density are inversely
correlated with the hottest temperatures corresponding to re-
gions of lowest surface density. For both the warm and hot phases,
we see that the temperature is higher in the interarm regions than
in the spiral arms. This is real and does not result from lower
signal-to-noise ratio in the interarm regions. We tested this by
extracting spectra over 0.76 kpc2 interarm regions between the
northwest inner arm and northwest arm and between the southeast
inner arm and southeast arm and found warm H2 temperatures of
177 and 175 K and hot H2 temperatures of 691 and 690 K, re-
spectively. These temperatures are higher than the 150–165 K
warm H2 temperatures and the 550–650 K hot H2 temperatures
found in the spiral arms.

Figure 8 compares the warm (in gray scale) and hot (in con-
tours) H2 surface density distributions. The warm H2 mass dis-
tribution peaks in the northwest inner spiral arm and the hot H2
mass distribution peaks in the nucleus and in the region interior
to the northwest spiral arm. The warm-to-hot H2 mass ratio is not
constant across the galaxy but is lowest (�12) in the nucleus of
the galaxy and increases to 170 and 136 in the southeast and
northwest spiral arms, respectively.

4. DISCUSSION

4.1. Effects of Beam Averaging on Studies of Extragalactic H2

Previous studies have used aperture averages over entire ga-
lactic nuclei to derive the physical conditions of the molecular
gas (Rigopoulou et al. 2002; Higdon et al. 2006; Roussel et al.
2007). In M51, Roussel et al. (2007) find that within the central
330 arcsec2 (4:61 ; 105 pc2), the warm H2 phase has a temper-
ature of 180 K and a surface density of 3.2 M� pc

�2 (a total mass

of Mwarm ¼ 1:5 ; 106 M�). Roussel et al. (2007) have also mea-
sured the temperature of the hot phase (although they do not mea-
sure the mass in the hot phase) and find a hot H2 temperature of
521 K.
Having spatially resolved spectra over a strip across M51, we

can investigate the behavior of the warm and hot H2 phases on
smaller scales. In the nuclear region of M51 we see that the warm-
phase temperature peaks at 192 K and decreases radially toward
the inner spiral arms. The warm H2 surface density at the nucleus
is 4.4 M� pc

�2 and decreases over a 0.5 kpc radius surrounding
the nucleus. To check the consistency of our results against those
of Roussel et al. (2007), we averaged the warm-phase temper-
ature over a similar 412 arcsec2 (5:76 ; 102 pc2) aperture and
found that the warm phase temperature and surface density are
186 K and 2.8 M� pc

�2, respectively. Our results are consistent
with previous studies of warm H2 done at lower resolution and
suggest that previous studies of the warm H2 temperature and mass
have yielded average values rather than maximum (or minimum)
values.

4.2. Distinguishing the H2 Excitation Mechanisms

The warm-to-hot H2 surface density ratio varies across M51,
suggesting that the H2 excitation mechanisms have different ef-
fects on the warm and hot phases. The largest warm H2 surface
densities are found in the spiral arms, suggesting that the warm
phase is associated with star formation activity. The largest hot
H2 surface densities are found in the nuclear region of M51. This
suggests that where nuclear activity is the dominant excitation
mechanism, a larger fraction of the H2 can be maintained in the
hot phase and that H2 is generally more efficiently heated by
nuclear activity. This is further supported by the fact that the hot
H2 surface density is �660 K in the nuclear region and decreases
to �530 and 550 K in the inner northwest and southeast spiral
arms, respectively. Roussel et al. (2007) find that H2 is generally
heated by massive stars in PDRs; however, Seyferts and LINERs
show evidence for the dominance of other excitation mechanisms
such as X-rays or shocks. By comparing the spatial distribution
of the warm and hot H2 to optical V-band imagery, H�, and CO
(J ¼ 1 0) emission, we can investigate the relationship between
star formation and H2. In addition, by comparing the warm
and hot H2 distributions to X-ray observations and the spatial

Fig. 7.—Hot (T ¼ 400 1000 K) H2 surface density distribution (in contours)
compared to the hot H2 temperature distribution (in gray scale, in units kelvin). The
hot H2 temperature and surface density distributions are derived from the fit to the
excitation diagrams across the strip for the H2 S(2)–H2 S(5) lines. Surface density
contour levels are at intervals of 10% of 0.25 M� pc

�2 (same as in Fig. 6). The high
temperature and surface densityof thehot phasetowardthe northwest and southeast
ends of the strips is due to the lack of detection of extended emission from the H2
S(4) and H2 S(5) lines. In these regions (beyond the northwest and southeast spiral
arms) we are unable to accurately determine the hot phase temperature and surface
density.

Fig. 8.—Warm (T ¼ 100 300 K) H2 surface density (in gray scale) compared
to the hot (T ¼ 400 1000 K) H2 surface density (in contours). Contour levels for
the hot H2 surface density distribution are at intervals of 10% of the maximum
surface density (0.25 M� pc

�2). The gray scale is in units of M� pc
�2.

BRUNNER ET AL.322 Vol. 675



distribution of [O iv] (25.89 �m) emission, we can investigate
X-ray and shock heating of H2 within M51.

Kaufman et al. (2006) show that within galaxies, where the tele-
scope beam size is generally kiloparsecs across, H2 emission could
serve to probe the average physical conditions in the surfaces of
molecular clouds. In Figure 9, we compare the warm (left) and
hot H2 (right) mass distributions to the cold molecular gas traced
by CO (J ¼ 1 0) emission. The warm and hot H2 phases appear

to trace the bright CO emission in the northwest and southeast
spiral arms. Comparison of CO to the individual H2 S(0)–H2 S(3)
line intensity maps in Figure 10 also shows that the H2 in the spiral
arms traces the bright CO emission. In the spiral arms, both the
warm and hot H2 phases are found in PDRs and are associated
with star formation activity. The most striking result is that in the
inner spiral arms, we see that the CO is offset toward the nucleus
from the warm H2 mass (in Fig. 9). The offset between the peaks in

Fig. 9.—Left: Comparison of CO intensity (in gray scale) to the warm (T ¼ 100 300 K) H2 surface density (in contours). The CO intensity is in units of Jy km s�1.
The warm H2 surface density contours are the same as in Figs. 5 and 6. Right: Comparison of CO intensity (in gray scale) to the hot (T ¼ 400 1000 K) H2 surface
density (in contours). The CO intensity is in units of Jy km s�1. The hot H2 surface density contours are the same as in Figs. 5 and 7.

Fig. 10.—Comparison of the CO emission to the H2 S(0) (top left), H2 S(1) (top right), H2 S(2) (bottom left), and H2 S(3) (bottom right) emission. The CO emission
maps are in units of Jy km s�1. Contour levels for H2 S(0), H2 S(1), H2 S(2), and H2 S(3) are the same as in Fig. 3.

MOLECULAR HYDROGEN IN NGC 5194 323No. 1, 2008



CO and warm H2 mass is 7
00 in the northwest inner spiral arm and

500 in the southeast inner spiral arm. We believe that these offsets
are real, with one possible explanation being that the H2 is tracing
the regions of active star formation within the giant molecular
associations.

In Figures 11 and 12, we compare the warm (left) and hot
(right) H2 mass distributions to H� emission and an optical
V-band image, respectively. The H� image reveals the brightest
H ii regions and the V-band image shows the dust lanes in M51.
In general, the warm and hot H2 concentrations are not cospatial
with the brightest H� emission regions in the spiral arms, with

the one exception being that the warm H2 mass in the northwest
and southeast inner spiral arms appears to trace the H� emission.
The warm H2 mass contours show that local peaks in H2 mass are
found within the dust lanes, with the H� emission spatially offset
toward the leading edge of the spiral arms. An example of this is
in the northwest spiral arms where we see the H2 mass offset from
the H� spiral arms with local peaks being found in the dust lanes.
In Figures 13 and 14, we compare the H2 S(0)–H2 S(3) line-
intensity maps to H� emission and V-band imagery, respectively.
Comparison of the H2 S(0) map to H� reveals that the strongest
H2 emission in the northwest and southeast inner spiral arms is

Fig. 11.—Left: Comparison of H� (in gray scale) to the warm (T ¼ 100 300 K) H2 surface density (in contours). The H� image is in units of counts s�1. The warm
H2 surface density contours are the same as in Figs. 5 and 6. Right: Comparison of H� (in gray scale) to the hot (T ¼ 400 1000 K) H2 surface density (in contours). The
H� image is in units of counts s�1. The hot H2 surface density contours are the same as in Figs. 5 and 7.

Fig. 12.—Left: Comparison of the V-band image (in gray scale) to the warm (T ¼ 100 300 K) H2 surface density (in contours). The V band shows the dust lanes within
M51. The V-band image is in units of counts s�1. The warm H2 surface density contours are the same as in Figs. 5 and 6. Right: Comparison of V-band image (in gray scale) to
the hot (T ¼ 400 1000 K) H2 surface density (in contours). The V-band image is in units of counts s�1. The hot H2 surface density contours are the same as in Figs. 5 and 7.

BRUNNER ET AL.324 Vol. 675



coincident with H� emission; however, the other H2 S(0) spiral
arms show the strongest emission in the dust lanes, offset from the
H� spiral arms. The largest offsets are seen in the southeast spiral
arm where the H2 S(0) emission is offset from the H� spiral arm by
�1500 (560 pc). H2 S(1) emission appears to follow the dust lanes
and the H2 S(1) intensity subsides into the H� spiral arms. H2 S(2)
and H2 S(3) emission is also found in the dust lanes; however, there
areinstances (such asin the southeast spiral arm) where the H2 emis-
sion appears to be found straddling the dust lane and H� spiral arm.

The [O iv] (25.89 �m) line can be excited in fast shocks (vs >
100 km s�1; Lutz et al. 1998), by photoionization in Wolf-Rayet
stars (Schaerer & Stasinska 1999), or by an AGN (Smith et al.
2004). We have mapped the blended [O iv] (25.89 �m) + [Fe ii]
(25.99 �m) line across M51. While the map contains the
contribution of both lines, we have examined the SINGS high-
resolution spectra and find that averaged over a 365 arcsec2 re-
gion around the nucleus of M51, the [O iv] line flux is 2.7 times

higher than the [Fe ii] flux.
5
In the high-resolution data cubes

of the nuclear region, the [O iv] line flux dominates the [O iv] +
[Fe ii] blend and the [O iv] emission distribution derived from
the low resolution spectra accurately reflects the distribution seen
in the high resolution spectra. In Figure 15, we compare the [O
iv] surface brightness to the warm (left) and hot (right) H2 distri-
butions. The [O iv] emission is brightest in the nuclear region at
1:2 ; 10�17 W m�2, and within a 0.5 kpc radius of the nucleus
the peak is coincident with the peak in the mass of the hot H2.
[O iv] surface brightness decreases from the nucleus to the inner
spiral arm by 50%. We resolve weaker [O iv] emission within the
warm and hot H2 spiral arms. The [O iv] surface brightness is
a factor of �6 lower in the spiral arms than the peak inten-
sity found in the nucleus. The [O iv] emission within the nuclear

Fig. 13.—Comparison of H� emission to the H2 S(0) (top left), H2 S(1) (top right), H2 S(2) (bottom left), and H2 S(3) (bottom right) emission. The H� image is in units
of counts s�1. Contour levels for H2 S(0), H2 S(1), H2 S(2), and H2 S(3) are the same as in Fig. 3.

5
SINGS data cubes and spectra for M51 can be found at http://irsa.ipac

.caltech.edu/data/SPITZER/SINGS/galaxies/ngc5194.html.

MOLECULAR HYDROGEN IN NGC 5194 325No. 1, 2008



Fig. 14.—Comparison of the V-band image to the H2 S(0) (top left), H2 S(1) (top right), H2 S(2) (bottom left), and H2 S(3) (bottom right) emission. The V-band image
is in units of counts s�1. Contour levels for H2 S(0), H2 S(1), H2 S(2), and H2 S(3) are the same as in Fig. 3.

Fig. 15.—Left: Comparison of the [O iv] (25.89 �m) emission (in gray scale) to the warm (T ¼ 100–300 K) H2 surface density distribution (in contours). Warm H2
surface density contours are the same as in Figs. 5 and 6. Right: Comparison of the [O iv] (25.89 �m) emission (in gray scale) to the hot (T ¼ 400–1000 K) H2 surface
density distribution (in contours). Hot H2 surface density contours are the same as in Figs. 5 and 7. The [O iv] (25.89 �m) emission is in units of W m

�2.



region of M51 is likely due to the weak Seyfert 2 nucleus (Ford
et al. 1985) and is possibly associated with shocked gas from the
outflows of the AGN. The peak of the [O iv] emission coincides
with the nuclear peak in hot H2 mass, indicating that the hot H2
phase in the nuclear region of the galaxy is AGN or shock heated.
In the nuclear region we observe a factor of 12 times greater warm
H2 mass than the hot H2 mass. The warm H2 mass is much greater

within the spiral arms than within the nucleus and the warm-to-
hot mass ratio is lowest in the nuclear region where the [O iv] flux
is greatest. In general, nuclear activity and shocks heat the ISM
more efficiently, causing the higher surface density of hot H2.

X-ray studies of M51 have revealed bright X-ray emission
from the nucleus, the extranuclear cloud (XNC, to the south of
the nucleus), and the northern loop (Terashima & Wilson 2001;

Fig. 16.—Left: Comparison of the smoothed 0.5–10 keV X-ray emission band (in gray scale) to the warm (T ¼ 100 300 K) H2 surface density distribution (in contours).
The X-ray image has been smoothed to the same resolution as the warm H2 surface density map. X-ray emission is in units of counts. The warm H2 surface density contours are
the same as in Figs. 5 and 6. Right: Comparison of the smoothed 0.5–10 keV X-ray emission band (in gray scale) to the hot (T ¼ 400 1000 K) H2 surface density distribution
(in contours). The hot H2 surface density distribution contours are the same as in Figs. 5 and 7.

Fig. 17.—Comparison of the 0.5–10 keV X-ray emission band (in gray scale) to the H2 S(2) (top left), H2 S(3) (top right), H2 S(4) (bottom left), and H2 S(5) (bottom
right) emission in the nuclear region of M51. X-ray emission is in units of counts. The H2 S(2) and H2 S(3) emission contours are at 10% of their peak values
(2:20 ; 10�18 and 1:35 ; 10�17 W m�2, respectively). The H2 S(4) contours are at 1:0 ; 10

�18 and 2:0 ; 10�18 W m�2 and the H2 S(5) contours are at 8:0 ; 10
�19, 4:0 ;

10�18, and 7:3 ; 10�18 W m�2.

MOLECULAR HYDROGEN IN NGC 5194 327



Maddox et al. 2007). A radio jet that is believed to be shock
heating the ISM has been observed emanating from the south of
the nucleus toward the XNC in 6 cm imagery (Crane & van der
Hulst 1992). In Figure 16, we compare the smoothed 0.5–10 keV
band X-ray image to the warm (left) and hot (right) H2 mass dis-
tributions. The 0.5–10 keV band has been smoothed to the resolu-
tion of the warm and hot H2 mass distributions, and the nucleus,
XNC, and northern loop are indistinguishable in the smoothed
image. X-ray emission is brightest in the nucleus and decreases
into the northwest spiral arm that contains the greatest H2 mass.
There appears to be very little connection between the 0.5–10 keV
X-ray band and the warm H2 mass distribution. The peak in X-ray
emission is coincident with the hot H2 mass peak. The brightest
0.5–10 keV X-ray emission originates from the nucleus and is
oriented north to south, similar to the [O iv] (25.89 �m) emission.
The peak in X-ray emission is located within the peak in hot H2
mass, suggesting that X-rays play an important role in produc-
ing the hot H2 phase. While there is a correlation between X-ray
emission and the hot H2 phase, H2 excitation by X-rays cannot be
distinguished from H2 excitation by shocks. In Figure 17 we com-
pare the X-ray surface brightness to the H2 S(2)–H2 S(5) maps
and find that the nuclear H2 emission appears to be correlated with
the X-ray source.

5. CONCLUSIONS

We have spectrally mapped a strip across M51 using the Spitzer
IRS low-resolution modules. We used the spatially resolved spec-
tra to map H2 S(0)–H2 S(5) lines across the strip. We find:

1. The distribution of H2 emission in M51 varies with H2
rotational level. H2 S(0) emission is brightest in the spiral arms
of the galaxy while the higher J transitions show the strongest
emission toward the nucleus. The H2 S(1) line is brightest in the
nuclear region and is offset from the peak in H2 S(0) intensity in
the inner northwest spiral arm by 1000. The H2 S(2) and H2 S(3)
maps show H2 emission in the nucleus, spiral arms, and interarm
regions of M51 and bar structure aligned north to south is apparent
in H2 S(3) emission. H2 S(4) and H2 S(5) emission is resolved in
the nuclear region of M51.

2. The different distributions of H2 emission in M51 indicate
significant spatial variations in H2 temperature and surface den-
sity. Using the low J [S(0)–S(2)] lines to trace the warm (T ¼ 100
300 K) H2, we find that the warm H2 temperature is highest in the
nuclear region at 192 K and the warm H2 surface density peaks in
the northwest inner spiral arm at 11 M� pc

�2. Using the higher J
[S(2)–S(5)] lines to trace the hot (T ¼ 400 1000 K) H2, we find
that the hot H2 temperature is lowest in the inner spiral arms (500–

550 K) and increases to �600 K in the nucleus where the largest
hot H2 surface density is found to be 0.24 M� pc

�2.
3. The warm and the hot H2 surface density distributions

are not cospatial and the warm-to-hot surface density ratio varies
across M51. The warm H2 surface density distribution peaks in
the northwest spiral arm and is offset from the hot mass peak by
1100. The hot H2 surface density distribution shows two peaks,
one in the nucleus of M51 and one located interior to the north-
west inner spiral arm of M51. Variations in the warm-to-hot H2
ratio and differences in the distributions of the H2 line emission
across M51 suggest that the warm H2 is mostly produced by UV
photons in star-forming regions while the hot H2 is mostly pro-
duced by shocks or X-rays associated with nuclear activity.

4. The warm H2 follows the cold molecular gas traced by
CO in the spiral arms of M51, indicating that the warm phase is
associated with the surface layers of dense molecular clouds.
The H2 S(0)–H2 S(3) contours trace the CO; however, within the
spiral arms, the peaks in H2 can be offset from the peaks in CO
intensity.

5. Comparing the distributions of H2 to V-band imagery
and H� emission reveals that the warm and hot H2 in the spiral
arms is found in the dust lanes rather than spatially coincident with
the H� emission.

6. [O iv] (25.89 �m) emission and X-ray intensity peak in
the nuclear region of M51 and their peaks are spatially coin-
cident with the peak in hot H2 surface density. This implies that
the hot H2 is more efficiently heated by the AGN, shocks (pos-
sibly associated with the AGN), or X-rays associated with the
AGN. The spatial distributions of the [O iv] emission and X-ray
surface brightness are very similar, preventingthe characterization
of the primary excitation mechanism (shocks or X-rays) of the hot
H2 phase in the nuclear region.

The author graciously acknowledges the Spitzer Science Cen-
ter Visiting Graduate Student Fellowship program and commit-
tee for providing support for this research. The author would
like to specifically acknowledge the program coordinators, Phil
Appleton and Alberto Noriega-Crepso. The authors would also
like to thank the anonymous referee who helped to clarify our
results and improve our discussion. Partial support for the com-
pletion and preparation for publication of this study by the author
was provided by AURA grant GO10822.1 to Rice University.

Facilities: Spitzer Science Center (SSC), Spitzer Space Telescope
(SST), Berkeley-Illinois-Maryland Association (BIMA).

REFERENCES

Aalto, S., Hüttemeister, S., Scoville, N. Z., & Thaddeus, P. 1999, ApJ, 522,
165

Calzetti, D., et al. 2005, ApJ, 633, 871
Crane, P. C., & van der Hulst, J. M. 1992, AJ, 103, 1146
Ford, H. C., Crane, P. C., Jacoby, G. H., Lawrie, D. G., & van der Hulst, J. M.
1985, ApJ, 293, 132

Fuente, A., Martı́n-Pintado, J., Rodrı́guez-Fernández, N. J., Rodrı́guez-Franco,
A., de Vicente, P., & Kunze, D. 1999, ApJ, 518, L45

Helfer, T. T., Thornley, M. D., Regan, M. W., Wong, T., Sheth, K., Vogel, S. N.,
Blitz, L., & Bock, D. C. J. 2003, ApJS, 145, 259

Higdon, S. J. U., Armus, L., Higdon, J. L., Soifer, B. T., & Spoon, H. W. W.
2006, ApJ, 648, 323

Houck, J. R., et al. 2004, ApJS, 154, 18
Kaufman, M. J., Wolfire, M. G., & Hollenbach, D. J. 2006, ApJ, 644, 283
Kennicutt, R. C., et al. 2003, PASP, 115, 928
Kenny, J. D. P., & Lord, S. D. 1991, ApJ, 381, 118
Lord, S. D., & Young, J. S. 1990, ApJ, 356, 135

Lutz, D., Kunze, D., Spoon, H. W. W., & Thornley, M. D. 1998, A&A, 333,
L75

Maddox, L. A., Cowan, J. J., Kilgard, R. E., Schinnerer, E., & Stockdale, C. J.
2007, AJ, 133, 2559

Matsushita, S., et al. 2004, ApJ, 616, L55
Meijerink, R., Tilanus, R. P. I., Dullemond, C. P., Israel, F. P., & van der Werf,
P. P. 2005, A&A, 430, 427

Neufeld, D. A., Melnick, G. J., & Harwit, M. 1998, ApJ, 506, L75
Neufeld, D. A., et al. 2006, ApJ, 649, 816
Palumbo, G. G. C., Fabbiano, G., Fransson, C., & Trinchieri, G. 1985, ApJ,
298, 259

Rand, R. J., & Kulkarni, S. R. 1990, ApJ, 349, L43
Regan, M. W., Thornley, M. D., Helfer, T. T., Sheth, K., Wong, T., Vogel, S. N.,
Blitz, L., & Bock, D. C. J. 2001, ApJ, 561, 218

Rigopoulou, D., Kunze, D., Lutz, D., Genzel, R., & Moorwood, A. F. M. 2002,
A&A, 389, 374

Roussel, H., et al. 2007, ApJ, 669, 959

BRUNNER ET AL.328 Vol. 675



Sakamoto, K., Okumura, S. K., Ishizuki, S., & Scoville, N. Z. 1999, ApJ, 525,
691

Schaerer, D., & Stasinska, G. 1999, A&A, 345, L17
Scoville, N. Z., Polletta, M., Ewald, S., Stolovy, S. R., Thompson, R., & Rieke,
M. 2001, AJ, 122, 3017

Scoville, N. Z., Yun, M. S., Armus, L., & Ford, H. 1998, ApJ, 493, L63
Sheth, K., Regan, M. W., Vogel, S. N., & Teuben, P. J. 2000, ApJ, 532,
221

Sheth, K., Vogel, S. N., Regan, M. W., Thornley, M. D., & Teuben, P. J. 2005,
ApJ, 632, 217

Sheth, K., Vogel, S. N., Regan, M. W., Tueben, P. J., Harris, A. I., &
Thornley, M. D. 2002, AJ, 124, 2581

Smith, J. D. T., et al. 2004, ApJS, 154, 199
———. 2007a, PASP, 119, 1133
———. 2007b, ApJ, 656, 770
Terashima, Y., Ptak, A., Fujimoto, M. I., Kunieda, H., Makishima, K., &
Sherlemitsos, P. J. 1998, ApJ, 496, 210

Terashima, Y., & Wilson, A. S. 2001, ApJ, 560, 139
Tully, R. B. 1988, Nearby Galaxies Catalog (Cambridge: Cambridge Univ. Press)
Vogel, S. N., Kulkarni, S. R., & Scoville, S. Z. 1988, Nature, 334, 402
Young, J. S., & Scoville, N. Z. 1991, ARA&A, 29, 581

MOLECULAR HYDROGEN IN NGC 5194 329No. 1, 2008