SPATIALLY RESOLVED SPITZER IRS SPECTROSCOPY OF THE CENTRAL REGION OF M82

P. Beirão,
1
B. R. Brandl,

1
P. N. Appleton,

2
B. Groves,

1
L. Armus,

3
N. M. Förster Schreiber,

4

J. D. Smith,
5
V. Charmandaris,

6
and J. R. Houck

7

Received 2007 February 7; accepted 2007 November 27

ABSTRACT

We present high spatial resolution (�35 pc) 5Y38 �m spectra of the central region of M82, taken with the Spitzer
Infrared Spectrograph. From these spectra we determined the fluxes and equivalent widths (EWs) of key diagnostic
features, such as the [Ne ii] 12.8 �m, [Ne iii] 15.5 �m, and H2 S(1) 17.03 �m lines, and the broad mid-IR polycyclic
aromatic hydrocarbon (PAH) emission features in six representative regions and analyzed the spatial distribution of
these lines and their ratios across the central region. We find a good correlation of the dust extinction with the CO 1Y0
emission. The PAH emission follows closely the ionization structure along the galactic disk. The observed variations
of the diagnostic PAH ratios across M82 can be explained by extinction effects, within systematic uncertainties. The
16Y18 �m PAH complex is very prominent, and its EWis enhanced outward from the galactic plane. We interpret this
as a consequence of the variation of the UVradiation field. The EWs of the 11.3 �m PAH feature and the H2 S(1) line
correlate closely, and we conclude that shocks in the outflow regions have no measurable influence on the H2 emis-
sion. The [Ne iii]/[Ne ii] ratio is on average low, at �0.18, and shows little variations across the plane, indicating that
the dominant stellar population is evolved (5Y6 Myr) and well distributed. There is a slight increase of the ratio with
distance from the galactic plane of M82, which we attribute to a decrease in gas density. Our observations indicate that
the star formation rate has decreased significantly in the last 5 Myr. The quantities of dust and molecular gas in the
central area of the galaxy argue against starvation and for negative feedback processes, observable through the strong
extraplanar outflows.

Subject headinggs: galaxies: individual (M82) — galaxies: starburst — infrared: galaxies

Online material: color figures

1. INTRODUCTION

M82 (NGC 3034) is an irregular galaxy located at 3.3 Mpc
( Freedman & Madore 1988) in the M81 group. It is the closest
starburst galaxy, seen nearly edge-on, with an inclination angle
of about 80

�
. At infrared wavelengths it is the brightest galaxy on

the sky, with a total infrared luminosity of 3:8 ; 1010 L� (Colbert
et al. 1999). Most of its luminosity originates from the inner
500 pc, hosting intense starburst activity presumably triggered by
a tidal interaction with M81 (e.g., Yun et al. 1993). Evidence for
this interaction comes from the H i streams that connect M81 to
all three (M82, NGC 3077, and NGC 2976) peculiar members
of the inner M81 group (Appleton et al. 1981; Yun et al. 1994).
Recent deep optical images also revealed stars associated with
the H i bridge between M81 and M82 (Sun et al. 2005).

At the distance of M82, 100 corresponds to 15 pc, which allows
spatially resolved studies of the starburst region. Evidence of a
stellar bar �1 kpc long is shown by near-infrared studies (e.g.,
Telesco et al. 1991; Larkin et al. 1994) and mid-infrared [Ne ii]
12.8 �m and millimetric CO emission studies (Lo et al. 1987).
According to Larkin et al. (1994) and Achtermann & Lacy (1995),

there is a rotating ring of ionized gas at a radius of �85 pc and on
the inner side of a ring of molecular gas at �210 pc. Two possible
spiral arms were also identified by Shen & Lo (1995) and Mayya
et al. (2005) at radii of �125 and �400 pc. The starburst of M82
drives a bipolar mass outflow out to several kiloparsecs perpen-
dicular to the plane of the galaxy, which is especially evident in
X-ray and H� (Bregman et al. 1995; Shopbell & Bland-Hawthorn
1998; Lehnert et al. 1999; Cappi et al. 1999; Strickland et al.
2004). Dust has also been detected in the outflow region (Alton
et al. 1999; Heckman et al. 2000; Hoopes et al. 2005; Engelbracht
et al. 2006). The star-forming regions of M82 are predominantly
clustered in the volume enclosed by the molecular gas ring, in-
dicated by the H ii region tracers such as the [Ne ii] 12.8 �m line
and the mid and far-infrared continuum emission (Telesco et al.
1991; Walter et al. 2002; Larkin et al. 1994; Achtermann & Lacy
1995; Lipscy & Plavchan 2004). Near-infrared hydrogen recom-
bination lines also arise in these regions, but they are a more am-
biguous tracer of star formation, as they can also be excited by
shocks, although these are unlikely to dominate.
Near-infrared integral field spectroscopy and ISO SWS mid-

infrared spectroscopy by Förster Schreiber et al. (2001) allowed
a detailed modeling of starburst activity in the central region of
M82 (Förster Schreiber et al. 2003a). These models are consis-
tent with the occurrence of starburst activity in two successive
episodes, about 10and 5 Myr ago,eachlastinga few million years.
However, the spatial studies by Förster Schreiber et al. (2001) and
others covered only near-infrared wavelengths. The large aperture
of the ISO SWS provided a continuous 2.4Y45 �m spectrum but
covered the whole central region of M82. ISOCAM CVF data
(Förster Schreiber et al. 2003b) provided better spatial resolu-
tion, but the spectra only had a spatial resolution of R � 40 and
were shortward of 15 �m.

A

1
Leiden Observatory, Leiden University, P. O. Box 9513, 2300 RA Leiden,

Netherlands.
2
NASA Herschel Science Center, California Institute of Technology, Pasadena,

CA 91125.
3
Spitzer Science Center, California Institute of Technology, Pasadena,

CA 91125.
4
Max-Planck-Institut für extraterrestrische Physik, Garching, Germany.

5
Steward Observatory, University of Arizona, Tucson, AZ 85721.

6
IESL/Foundation for Research andTechnology-Hellas, GR-71110Heraklion,

Greece; and Chercheur Associé, Observatoire de Paris, F-75014, Paris, France.
7
Astronomy Department, Cornell University, 219 Space Sciences Building,

Ithaca, NY 14853.

304

The Astrophysical Journal, 676:304Y316, 2008 March 20
# 2008. The American Astronomical Society. All rights reserved. Printed in U.S.A.



Engelbracht et al. (2006) published Spitzer IRS low-resolution
spectra of a 10 wide strip along the minor axis of M82, inter-
secting the disk at the eastern side. The spectra, taken as part of
the SINGS Legacy project, were combined with Spitzer 8 and
24 �m images, and show that the emission by polycyclic aro-
matic hydrocarbons (PAHs) and H2 molecules extends far out
from the disk (to 6 kpc) in both directions. Engelbracht et al.
(2006) suggest that the filamentary aromatic-dominated emis-
sion either represents dust expelled from the galaxy as a result
of a powerful nuclear superwind or indicates that dust in the halo
is being lit up by the starburst, perhaps coexisting with the ex-
tensive warm H2 molecules. They suggest that this halo dust is
probably a leftover from the interaction with M81.

In this paper we present mid-IR spectral maps at unsurpassed
sensitivity and spatial resolution of the central �0.5 kpc2 of M82,
covering the main contributors to the bolometric luminosity of the
galaxy. Our goal is to provide a spatial and spectrally detailed de-
scription of the physical conditions within the central �500 pc
of M82, to help us to gain an insight on the evolution of the star-
burst activity in this region. This involves the study of the dis-
tribution of the radiation field, gas density, and the physical
properties of PAHs. Of particular interest are the spatial varia-
tions of the fine-structure lines, the excitation of the molecular
hydrogen, and the distribution of the PAH molecules. In x 2 we
describe the observations and data reduction; in x 3 we present
the data; and in x 4 we discuss the scientific results, followed by
our conclusions in x 5.

2. OBSERVATIONS AND DATA REDUCTION

The observations were made with the Infrared Spectrograph
(IRS)

8
spectrometer (Houck et al. 2004) on board the Spitzer

Space Telescope, under the IRS guaranteed time observing pro-
gram. The data were taken on 2005 June 6 using the IRS ‘‘map-
ping mode’’ in all four modules: Short-High (SH; 10Y19 �m)
and Long-High (LH; 14Y38 �m) provide R � 600, while Short-
Low (SL; 5Y14 �m) and Long-Low (LL; 14Y38 �m) give R �
60Y130. Each of the SL and LL modules are further divided into
two subslits, which correspond to diffraction orders: SL1 (7.5Y
14 �m), SL2 (5Y7.5 �m), LL1 (20Y38 �m), and LL2 (14Y20 �m).
The SH map consists of 30 pointings with four cycles each, and
each subsequent pointing is offset by half a slit width parallel to
the slit and about one-third of the slit length along the slit. The
SH map covers an area of 2800 ; 2300. The LH data consists of

12 pointings, with five cycles each, covering an area of 3800 ;
3300. The offsets are equivalent to the SH map. The SL data con-
sists of 120 pointings, with two cycles each, covering an area
over the M82 central region of 55:500 ; 5700. The LL data con-
sists of 22 pointings with two cycles each. Both SL and LL maps
follow the same offsetting scheme as SH. Because of the high
brightness of M82 it was unavoidable that the LL1 data became
saturated near the center of M82, but the LL2 data are still usable.
Figure 1 (left) shows the areas covered by the IRS SL, LH, and
SH maps overlaid on the IRAC 8 �m image from Engelbracht
et al. (2006). The total integration times range from 12 s (for SL
exposures) to 31 s (for LH exposures). Due to issues concerning
the extraction, the LH spectra were not used in this analysis. The
boxes labelled ‘‘A’’ and ‘‘B’’ are the regions where the complete
low-resolution (SL+LL) spectra were extracted.

The data were processed with version 13.2 of the Spitzer re-
duction pipeline (ver. 14 for LL). Observations taken at each po-
sition were combined into spectral cubes using CUBISM (Smith
et al. 2007a), an IDL-based software package designed to com-
bine spectral mapping data sets into three-dimensional spectral
cubes. Bad pixels in the basic calibrated data (BCD) spectra were
manually flagged and then automatically discarded when re-
building the cube. Spectra from off-source positions 1 kpc to the
northeast of the nucleus were used to subtract the background
from the low-resolution spectra. For the high-resolution spectra
we did not subtract a background, since there was no suitable
‘‘sky’’ spectrum available and the high source fluxes strongly
dominate any background emission. The spectral analysis was
done using SMART (Higdon et al. 2004) and PAHFIT (Smith
et al. 2007b).

3. ANALYSIS

On the basis of the SH spectral map coverage we defined six
subregions for which we extracted the spectra from SL and SH
with CUBISM. The location of these regions is shown in Figure 1
(right) overlaid on the IRAC 8 �m image from Engelbracht et al.
(2006). The ISO aperture is also shown to illustrate the increase
in spatial resolution obtained with the Spitzer IRS. Regions 2 and
3 coincide with the peaks of the [Ne ii] emission, regions 1 and
4 are offset to both sides of the galactic disk, the slightly larger
region C covers the nominal center of the galaxy, and the final
region corresponds to the entire area mapped with the SH spec-
trograph. The exact coordinates and sizes are listed in Table 1.

Figure 2 shows the SL (5Y14 �m) and SH (10Y19 �m) spec-
tra extracted for these six regions. The noise is negligible and
the spectra exhibit the classical features of starburst galaxies
(Brandl et al. 2006), such as strong emission features of the

Fig. 1.—Left: Overlay of SL (red) and SH (blue) coverages and selected low-resolution extraction regions on an IRAC 8 �m image. Regions A and B, represented in
yellow, are regions where LL (14Y35 �m) spectra were extracted. Right: Zoom-in of the IRAC 8 �m image with an overlay of the SH map area in blue and the selected
regions (green) from where the SH+SL spectra were extracted. The ISO SWS aperture used in the 12Y27 �m range by Förster Schreiber et al. (2001) is also overlaid in
yellow on the image. Both figures are in logarithmic scaling.

8
The IRS was a collaborative venture between Cornell University and Ball

Aerospace Corporation funded by NASA through the Jet Propulsion Laboratory
and the Ames Research Center.

SPITZER IRS SPECTROSCOPY OF M82 305



PAHs, fine-structure lines, and emission from molecular hydro-
gen, in addition to the broad silicate absorption features around
9.7 and 18 �m, with an underlying continuum of very small grain
emission. The 5Y38 �m wavelength range contains many im-
portant diagnostic lines, such as [Ar ii] 6.99 �m, [Ar iii] 8.99 �m,
[S iv] 10.51 �m, [Ne ii] 12.81 �m, [Ne iii] 15.56 �m, and [S iii]
18.71 �m. All of the detected features are labeled in Figure 2.
The fluxes and ratios of the most relevant fine-structure and H2
emission lines are listed in Table 2. These lines were measured
using Gaussian fits to the line and linear fits to the local contin-
uum. The [Ne ii] line was measured after subtracting the 12.6 �m
PAH feature. [S iv] at 10.5 �m is hard to detect on regions 1 and 4
due to the low signal-to-noise ratio (S/N), and [Ar iii], at 8.99 �m,
is too close to the 8.6 �m PAH feature to be detected in a low-
resolution spectrum. Due to saturation of the LL1 module and
extraction issues of the LH spectra, we could not measure the
[S iii] 33.6 �m line.

Numerous PAH emission features are easily detectable in our
spectra. The fluxes and equivalent widths (EWs) of the strongest
features at 6.2, 7.7, 8.6, 11.3, 12.6, and 14.2 �m, as well as the
16Y18 �m complex are listed in Table 3. Their values were de-
rived using PAHFIT (Smith et al. 2007b), an IDL tool that de-
composes low-resolution spectra of PAH emission sources using
a physically motivated model. This model includes starlight,
thermal dust continuum, resolved dust features and feature blends,
prominent emission lines, and dust extinction. In our case we
merged SL and SH spectra in order to have one spectrum for each
region with the widest wavelength coverage possible. Weaker
PAH features at 5.2, 5.6, 12.0, and 13.55 �m are present in the
spectra in Figure 2 but are not used further in this paper. In Fig-
ure 3 we present an example of a fit to a combined SL+SH spec-
trum of region 2. In the overall spectrum (green) we distinguish
the continuum component (red), PAH component (blue), ionic
lines ( purple), and dust absorption (dashed line).

Silicate absorption affects mainly the wavelength interval from
7Y12 �m. In PAHFIT we have the option to include or exclude
silicate absorption, and the difference between these two cases
can be up to 50% in flux of the 11.3 �m feature. In x 4.3 we dis-
cuss the methods to estimate the magnitude and distribution
of extinction. In addition, there is a significant difference in the
11.3 �m flux between the SH and SL spectra, which can also

reach 50%. This difference can be attributed to a poor fit to the
silicate absorptionfeature in SH, as this module only covers wave-
lengths >10 �m. In this paper we use the SH+SL measurements
for this reason and to include the 17 �m complex.
We also extracted combined SL+LL spectra (5Y38 �m) from

two 1500 ; 2000 regions, A and B (see Fig. 1, left), at a distance of
200 pc above and below the galactic plane of M82, where sat-
uration did not compromise the LL measurements. Their posi-
tions are listed in Table 1. The spectra from regions A and B were
virtually identical in shape, and an average of these spectra is
shown in Figure 4. The spectrum is clearly dominated by strong
PAH emission features and a steeply rising continuum, charac-
teristic of classical starburst galaxies (e.g., Brandl et al. 2006).
The wiggles observed longward of 20 �m are fringes that origi-
nate from interferences in the detector substrate material. At the
low spatial resolution presented by the LL modules, we cannot
see any unusual features in the SL+LL spectra of regions A and
B. Also, as the LL1 slit is saturated at the central region of M82,
we cannot extract any full low-resolution spectra of this region,
connecting it to regions A and B. For these reasons, we do not
make further analysis of LL1 module spectra in this paper.

4. RESULTS AND DISCUSSION

The main gain of our observations over previous work on the
starburst in M82 is the availability of spatially resolved mid-IR
spectroscopy of the central region. Despite their overall simi-
larity, the spectra show distinct variations in the relative strengths
of the spectral features, in particular in the neon fine-structure
lines and PAH features. These variations and their physical causes
are described in the following sections.

4.1. The Morphology of the Starburst Region

The discovery of a series of compact radio supernova rem-
nants along the galactic plane of M82, extending over 600 pc
(Kronberg et al. 1985; Muxlow et al. 1994), is an indication of
very recent and presumably ongoing star formation. The detec-
tion of the ionic high-excitation lines [Ne iii] and [S iv] confirms
the presence of very young massive stars in M82. Ratios using
ionic lines of the same species and different ionization potentials
such as [Ne iii]/[Ne ii], [S iv]/[S iii], and [Ar iii]/[Ar ii] are a useful
measure of the hardness and intensity of the radiation field and

TABLE 1

Characteristics of the Selected Regions

Coordinates Region 1 Region 2 Region 3 Region 4 Center Total Region A Region B

R.A. (J2000.0) ............... 9 55 52.42 9 55 50.63 9 55 52.74 9 55 50.22 9 55 51.19 9 55 51.32 9 55 52.17 9 55 52.68

Decl. (J2000.0)............... +69 40 32.1 +69 40 45.6 +69 40 48.5 +69 40 58.60 +69 40 46.8 +69 40 45.30 +69 41 12.5 +69 40 22.2

Size................................. 6.800 ; 6.800 6.800 ; 6.800 6.800 ; 6.800 6.800 ; 6.800 11.300 ; 11.300 24.900 ; 24.900 14.800 ; 20.300 14.800 ; 20.300

Note.—Units of right ascension are hours, minutes, and seconds, and units of declination are degrees, arcminutes, and arcseconds.

TABLE 2

Main Emission-Line Strengths in Units of 10�19 W cm�2

Region

[Ar ii]

6.99 �m

[S iv]

10.5 �m

[Ne ii]

12.8 �m

[Ne iii]

15.6 �m

[S iii]

18.7 �m

H2 S(1)

17.0 �m

H2 S(2)

12.3 �m S(1)/S(2)

T [S(1)YS(2)]
(K) [Ne iii]/[Ne ii]

1.............. 0.30 � 0.03 0.04 � 0.01 1.04 � 0.03 0.22 � 0.01 0.38 � 0.01 0.12 � 0.01 0.07 � 0.01 1.71 � 0.46 381 � 44 0.21 � 0.02
2.............. 7.82 � 0.23 0.31 � 0.01 25.3 � 0.9 3.42 � 0.09 8.5 � 0.19 0.23 � 0.04 0.25 � 0.03 0.92 � 0.30 586 � 117 0.13 � 0.01
3.............. 5.77 � 0.10 0.33 � 0.01 18.4 � 0.4 3.58 � 0.06 5.92 � 0.18 0.25 � 0.04 0.27 � 0.03 0.93 � 0.28 583 � 109 0.19 � 0.02
4.............. 0.44 � 0.02 0.03 � 0.01 1.73 � 0.04 0.36 � 0.01 0.66 � 0.02 0.08 � 0.003 0.06 � 0.003 1.33 � 0.13 439 � 31 0.21 � 0.02
Center..... 17.1 � 0.47 0.72 � 0.08 55.4 � 0.97 8.53 � 0.15 18.6 � 0.37 0.51 � 0.07 0.67 � 0.07 0.76 � 0.21 700 � 129 0.15 � 0.01
Total ....... 44.8 � 1.49 2.16 � 0.13 141.2 � 3.10 23.6 � 0.38 49.0 � 0.83 2.53 � 0.09 2.47 � 0.11 1.02 � 0.09 536 � 32 0.17 � 0.02

BEIRÃO ET AL.306 Vol. 676



radiation density, and are therefore sensitive to the presence of
young massive stars.

Our measurements of the [Ne iii]/[Ne ii] ratio are shown in
Table 2 for each of the six selected regions. We find 0:13 �
½Ne iii�/½Ne ii� � 0:21, with a median of ½Ne iii�/½Ne ii� ¼ 0:18,
which is consistent with the spatially integrated ratio of
½Ne iii�/½Ne ii� ¼ 0:16 � 0:04 determined by Förster Schreiber
et al. (2001). These values are 30% lower than the average value
of 0.26 for the ISO SWS sample of starburst galaxies (Thornley
et al. 2000) and more than an order of magnitude below the value
of 8.5 foundin thecenter of NGC 5253(Beirão et al. 2006), a low-
metallicity starburst galaxy at about the same distance as M82.

Other ratios, such as [S iv]/[S iii] and [Ar iii]/[Ar ii], could be
used to confirm the results on the [Ne iii]/[Ne ii]. However, as
mentioned in x 3, our measurements of [S iv] in regions 1 and 4
have large errors from noise, and [Ar iii], at 8.99 �m, is too close
to the 8.6 �m PAH feature to be measured accurately from the
low-resolution spectrum, precluding the use of these ratios.

The Spitzer IRS spectral maps allow a study of the spatial var-
iation of the radiation field, based on the [Ne iii]/[Ne ii] ratio.
Due to the presence of several luminous star clusters in the cen-
tral region, one might expect strong variations of [Ne iii]/[Ne ii]
between regions. In Figure 5 we present spectral maps of the two
strongest neon emission lines and a ratio map, overlaid by the
12 �m continuum contours from Achtermann & Lacy (1995).
For both line maps, CUBISM was used to subtract a fitted con-
tinuum map from a total line+continuum map at the same wave-
length on a pixel-by-pixel basis. The [Ne ii] emission in the top
map shows two peaks to either side of the nucleus, a weaker and
a stronger peak, which correspond to the east and west peaks,
respectively, in Achtermann & Lacy (1995). After a close in-
spection, we identify the strong west peak in our maps with the
two continuum emission peaks. The east and west [Ne ii] emis-
sion peaks are identified as a ‘‘ring’’ of ionized gas in Achtermann
& Lacy (1995). The morphology of the [Ne ii] emission follows
the Br� emission in Satyapal et al. (1997) and Förster Schreiber
et al. (2001). The [Ne iii] map also reveals two emission peaks at
the same positions, but with the eastern peak brighter relative to
the western peak.

The bottom map in Figure 5 presents the [Ne iii]/[Ne ii] ratio.
The ratio varies from 0.08Y0.27 throughout the map. The lower
value corresponds to the location of the westernmost clusters,
and there is a significant increase in the ratio farther out from the

galactic plane of M82, from 0.15 to 0.27. Statistical 1 � errors are
�15% and arise mostly from baseline determination errors. The
gray line represents the direction of the X-ray outflow observed
in M82 (Strickland et al. 1997, 2004). It originates from the nu-
cleus ( plus sign) and is perpendicular to the plane of the galaxy.
The peak of X-ray emission is offset from the outflow axis by
30 pc. Although the gradient in the [Ne iii]/[Ne ii] ratio does fol-
low the outflow axis, it appears offset by 500 to the east, associ-
ated with the eastern cluster.

The overall morphology of the ionizing radiation in the region
as revealed by the spectral maps seems to be in agreement with
ground observations of the continuum and [Ne ii] emission by
Achtermann & Lacy (1995). However, the [Ne iii]/[Ne ii] ratio
varies only by a factor of 3 throughout the region, and these var-
iations do not correspond to the position of the emission peaks.
An increase of the [Ne iii]/[Ne ii] ratio is observed farther out
from the galactic plane, but due to the limited spatial coverage of
the map, we cannot determine if this increase is connected to
the outflows. The origins of the observed emission morphology
and the variation of the [Ne iii]/[Ne ii] ratio are discussed in the
following subsection.

4.2. Origins of the Variation of the Radiation Field

The [Ne iii]/[Ne ii] ratio measures the hardness of the radia-
tion field, which is a function of stellar age and metallicity, and is
often parameterized by the effective temperature, TeA, and the
radiation intensity as measured by the ionization parameter U.

9

Given the strength of the [Ne ii] and [Ne iii] lines in M82, and
their proximity in wavelength, which minimizes the effects of
extinction, [Ne iii]/[Ne ii] is the most reliable measure of the
hardness of the radiation field. With the help of starburst models,
it is possible to use the [Ne iii]/[Ne ii] ratio to estimate the ages of
the massive clusters in the region.

Observations with the ISO SWS have been used previously
for this purpose. Förster Schreiber et al. (2001) determined a
spatially integrated ratio of ½Ne iii�/½Ne ii� ¼ 0:16 for the inner
500 pc of M82. Using the photoionization code CLOUDY and

TABLE 3

Main PAH Feature Strengths as Measured with PAHFIT

Region

6.2 �m

Flux
a

EW
b

7.7 �m

Flux
a

EW
b

8.6 �m

Flux
a

EW
b

11.3 �m

Flux
a

EW
b

12.6 �m

Flux
a

EW
b

14.2 �m

Flux
a

EW
b

17 �m complex

Flux
a

EW
b

F(17 �m complex)/

F (6.2 �m)

F(17 �m complex)/

F (11.3 �m)

1................ 8.9 � 0.1 29.9 � 0.1 6.2 � 0.1 10.9 � 0.1 6.0 � 0.1 0.62 � 0.01 4.2 � 0.03 0.47 0.39
0.308 0.254 0.437 2.27 2.06 2.13 1.97

2................ 154.7 � 0.1 528.0 � 0.2 99.3 � 0.1 77.9 � 0.1 64.5 � 0.1 6.1 � 0.1 34.6 � 0.1 0.22 0.44
1.10 0.866 0.848 0.660 0.734 0.717 0.753

3................ 136.3 � 0.1 478.2 � 0.2 83.6 � 0.1 73.7 � 0.1 57.2 � 0.1 5.5 � 0.1 29.1 � 0.1 0.21 0.39
0.827 0.884 0.865 0.784 0.877 0.850 0.720

4................ 13.3 � 0.1 49.6 � 0.1 11.3 � 0.1 15.7 � 0.1 7.7 � 0.1 0.67 � 0.10 7.6 � 0.1 0.57 0.48
0.390 0.320 0.575 1.86 1.52 1.30 2.16

Center....... 375.0 � 0.2 1320.0 � 0.5 242.4.0 � 0.5 202.5 � 0.6 156.3 � 0.3 14.3 � 0.1 99.0 � 0.2 0.26 0.49
1.22 0.991 0.937 0.758 0.791 0.750 0.907

Total ......... 1158.0 � 1.0 4121.7 � 3.4 743.9 � 1.5 736.0 � 1.7 535.8 � 0.7 52.2 � 0.2 327.8 � 0.5 0.28 0.45
1.00 1.00 1.00 1.00 1.00 1.00 1.00

a
Flux in units of 10�19 W cm�2. The quoted errors are only the PAHFIT fit residuals. EWs are normalized to the EWs for the whole SH slit.

b
EW in units of �m.

9
U is defined as U ¼ Q/4�R2nHc, where Q is the production rate of ionizing

photons from the stars, R is the distance between the ionizing cluster and the
illuminated gas cloud, nH is the hydrogen number density of the gas, and c is the
speed of light.

SPITZER IRS SPECTROSCOPY OF M82 307No. 1, 2008



solar metallicity stellar atmosphere models by Pauldrach et al.
(1998), Förster Schreiber et al. (2001) modeled the variations of
line ratios with TeA. Adopting an electron density ne ¼ 300 cm�3
and anionization parameter log U ¼ �2:3, Förster Schreiber et al.
(2003a) found for the ISO value for [Ne iii]/[Ne ii] an effective
temperature of 37; 400 � 400 K and a burst age of 4Y6 Myr.
Other observed ratios were also modeled, such as [Ar iii]/[Ar ii]
and [S iv]/[S iii], giving similar TeA values (within uncertainties).
Independent estimations of cluster ages were done at longer wave-
lengths. Colbert et al. (1999) analyzed far-infrared spectra from
the ISO LWS and fitted line ratios to a combined H ii region and
PDR model. Their best-fit model is an instantaneous starburst of
3Y5 Myr old massive stars, in agreement with Förster Schreiber
et al. (2003a).

If the infrared emission peaks correspond to massive clusters
of stars, we can determine their ages from the measured [Ne iii]/

[Ne ii] ratio and compare them with the above results, using the
photoionization models by Snijders et al. (2007). As input,
Snijders et al. (2007) used massive cluster spectra modeled with
Starburst99, assuming a Salpeter IMF, Mup ¼ 100 M�, Mlow ¼
0:2 M�, and a gas density of 100 cm

�3. This value of gas density
is a factor of 3 lower than log U ¼ �2:3, the Förster Schreiber
et al. (2001) value, but this discrepancy has a small effect on the
[Ne iii]/[Ne ii] ratio. Figure 6 shows the results for a range of
ionization parameters. The selected regions are represented by
horizontal lines. The value of log U ¼ �2:3 derived by Förster
Schreiber et al. (2001) corresponds to a ionization front speed of
q ¼ 1:6 ; 108 cm s�1. For a typical value of q, the ages of the
clusters in each region range from 5 to 6 Myr, in agreement with
the previously determined burst ages. These clusters dominate
the central region of M82 and may be similar to those observed
farther out in M82 in the optical, studied in detail by Smith et al.

Fig. 2.—SL (left) and SH (right) spectra of the six selected regions within the central starburst region of M82. The ‘‘total’’ region corresponds to the total SH map in Fig. 1.
[See the electronic edition of the Journal for a color version of this figure.]

BEIRÃO ET AL.308 Vol. 676



(2006) using the HSTACS, which were found to have an average
age of 6:4 � 0:5 Myr.

As shown in Figure 5, the [Ne iii]/[Ne ii] ratio increases from
0.15 to 0.27 with increasing distance from the galactic plane of
M82. This is counterintuitive, as one might expect a harder ra-
diation field at the location of the most luminous regions along
the plane. Away from the plane, a decrease of gas density relative
to the number of ionizing photons leads to an increase of the ion-
ization parameter, which then causes an increase of the [Ne iii]/
[Ne ii] ratio, as discussed in Thornley et al. (2000). Indeed, Fig-
ure 6 shows that the [Ne iii]/[Ne ii] scales with the ionization
parameter. The variation we observe in the [Ne iii]/[Ne ii] ratio
implies a variation by a factor of 5 in the ionization parameter.
This is equivalent to saying that the gas density decreases 5 times

faster than the radiation field, which decreases as �R�2, with R
being the distance to the ionization source. Shocks could also
contribute to the increase of the [Ne iii]/[Ne ii] ratio in the out-
flow region, but in x 4.5 we show this effect to be minimal. This
hypothesis could be tested using the [S iii] 18.6 �m/[S iii] 33.6 �m
ratio. Unfortunately, due to the problems reported in x 2, we could
not derive an accurate flux for the [S iii] 33.6 �m line in both LL
and LH spectra.

It is important to emphasize that even at higher angular res-
olution, the [Ne iii]/[Ne ii] ratio in M82 remains quite low for an
active starburst. We would have expected a larger variation, with
higher ratios locally corresponding to younger clusters and lower
ratios elsewhere. A comparison with the ISO SWS sample of
starburst galaxies (Thornley et al. 2000) shows that it is actually
lower than most starbursts, despite being closer and better re-
solved. The low [Ne iii]/[Ne ii] ratio could be caused by an aged
stellar population in which starburst activity ceased more than
half a dozen megayears ago—although this possibility seems un-
likely given the large amounts of molecular gas still present at the
center of M82. An edge-on view of the galaxy could also con-
tribute to the low variation of [Ne iii]/[Ne ii] ratio.

As a comparison, we examined the 4:500 ; 4:500 area with the
highest [Ne iii] flux (Fig. 5), comparing the measured [Ne ii] and
[Ne iii] luminosities with the models to determine the enclosed
stellar mass. A single super star cluster of 5 Myr would have a
cluster mass of 106 M�, which is twice the mass of the super star
cluster in NGC 5253 (Turner et al. 2003). A single cluster in a
4:500 ; 4:500 area would correspond to a cluster number density
of �200 kpc�2, which is comparable to the cluster density found
in the fossil starburst region of M82 by de Grijs et al. (2001).

While the average age of the starburst population in M82
appears to be �5 Myr, ongoing star formation (�1 Myr) could
possibly be obscured by recent contributions of older stellar pop-
ulations (>5 Myr) to the neon ratio. Considering this possibility,

Fig. 3.—Decomposition of SL+SH spectrum of region 2. Red solid lines represent the thermal dust continuum components, the thick gray line the total continuum, blue
lines are dust features, while the violet peaks are atomic and molecular spectral lines. The dotted black line indicates the fully mixed extinction which affects all
components, with axis at right. The solid green line is the full fitted model, plotted on the observed flux intensities and uncertainties.

Fig. 4.—Average of the 5Y38 �m low-resolution IRS spectra of the regions A
and B, located approximately 200 pc above and below the galactic plane of M82.

SPITZER IRS SPECTROSCOPY OF M82 309No. 1, 2008



we explore the region with the highest [Ne iii] flux, a strong in-
dicator of the presence of O stars, as an illustrative case to set an
upper limit on the ongoing star formation in M82. This region
has a [Ne iii]/[Ne ii] of �0.10, with a total [Ne iii] luminosity
of 1:41 ; 106 L�. Assuming �1:6 ; 108 cm s�1, a typical young
cluster of 1 Myr has ½Ne iii�/½Ne ii� � 5 and can contribute �50%
to the total [Ne iii] luminosity, or �2.5% per mass relative to the
older 5Y6 Myr old population. This means that the 1 Myr old
population only emits �5% of the total ½Ne ii� þ ½Ne iii� lumi-
nosity of the older 5Y6 Myr old population. Even as an upper
limit, this value indicates that the activity of the starburst has sub-
stantially declined relative to the high star formation rate that
existed 5 Myr ago.

Despite the reduced starburst activity, the presence of CO emis-
sion all over the central region (Fig. 7) shows that the starburst in
M82 still has a largegas reservoir, as pointed out byThornleyet al.
(2000) and Förster Schreiber et al. (2003a). It appears therefore
unlikely that the starburst activity ceased because of lack of fuel.
On theother hand, negativefeedbackmainly throughstrongstellar
winds and supernovae explosions can play an important role in

determining the star formation rate in starbursts. Förster Schreiber
et al. (2003a) calculated a feedback timescale of 1Y10 Myr, which
is in good agreement with the age of the older super star clusters
in the central region of M82.
In summary, we find that [Ne iii]/[Ne ii] ratio is low on average

and increases with distance from the galactic plane of M82. The
increase can be explained by an increase of the ionization pa-
rameter through a drop in gas density. The low [Ne iii]/[Ne ii] ratio
indicates that the dominant population consists of older clusters
(>5 Myr). We cannot rule out the presence of younger clusters
in the central region, but they would be at a much reduced rate
(<5%) of star formation compared to previous epochs. The large
amount of molecular gas still present in the central region argues
against a starvation of starburst activity due to lack of gas. In-
stead, we believe that negative feedback processes are respon-
sible for the observed decline in the star formation rate.

4.3. Extinction

One of the most noticeable features in the spectra in Figure 2
is the absorption feature at 9Y11 �m caused by silicate grains.

Fig. 5.—Spectral maps in the [Ne ii] (top), [Ne iii] (middle), and [Ne iii]/[Ne ii]
(bottom) lines from the IRS SH module, with contour overlays of the 12 �m con-
tinuum emission from Achtermann & Lacy (1995). The regions shown are �2600 ;
2600. At the bottom map, the line is the direction of the of the X-ray outflow, which
is perpendicular to the plane of the galaxy. The cross represents the nucleus.

Fig. 6.—Effect of the cluster age on the [Ne iii]/[Ne ii] ratio. The model curves
are computed for a cluster of 106 M�, assuming Salpeter IMF, Mup ¼ 100 M�,
and Mlow ¼ 0:2 M�. Each curve represents a different ionization parameter, and
the solid curve is the one that approaches the value found by Förster Schreiber
et al. (2001) log U ¼ �2:3. The horizontal lines indicate the [Ne iii]/[Ne ii] for
each selected region, from region 1 to region 4. The first dip in [Ne iii]/[Ne ii]
represents the ageing of the stellar population, after which WR stars are produced,
increasing the [Ne iii]/[Ne ii] ratio. The dip at 6 Myr occurs as most massive stars
die through supernova explosions.

Fig. 7.—Map of �9:8, determined from a simple continuum fit to the SL spectra,
with overlay of CO 1Y0 emission contours from Shen & Lo (1995). The crosses
indicate the [Ne ii] emission peaks. The image was rotated by 55

�
and interpolated.

BEIRÃO ET AL.310 Vol. 676



This feature can be extremely deep especially in ULIRGS (e.g.,
Spoon et al. 2007) and affects the observed fluxes of spectral
lines and features in this region. The strength of this feature is
characterized by the optical depth at 9.8 �m (�9:8), assuming a
simple geometrical dust distribution. For our analysis it is im-
portant to estimate the intensity of this feature and its spatial var-
iation in order to investigate its influence on the PAH strengths,
which we discuss in the following subsections.

To study the distribution of �9:8 in the SL region, we selected
an area of 20 ; 12 pixels in the center of the SL1 map. Within
that area we extracted spectra from 60 spatial resolution elements
(2 ; 2 pixels each).

The apparent optical depth �9:8 is then estimated from the ratio
of the interpolated continuum to the observed flux at 9.8 �m. For
the central region of M82, �9:8 ranges from 0.3 to 3.1, assuming
foreground screen attenuation. We built an extinction map of
10 ; 6 resolution elements from the simple-fit method, which is
shown in Figure 7. The qualitative distribution of �9:8 observed
in this figure is very similar to the qualitative results from the
PAHFIT fitting method. There is a good correlation between the
�9:8 distribution and the CO 1Y0 emission, indicating that dust
and molecular gas coincide in this region. The enhanced �9:8 in
the northwest region of the map indicates an increase of silicate
dust above the galactic plane.

For an independent, and possibly more accurate, estimate of
�9:8, we used PAHFIT on 15 SL+LL1 spectra (5Y20 �m), each
corresponding to four resolution elements fitted only with SL.
With the PAHFIT method, �9:8 ranges from 0 to 2.52, with a
median is 1.34.

The combined SL+LL spectrum is necessary to better constrain
the parameters in PAHFIT. Fitting only SL spectra with PAHFIT
can result in large errors in the calculation of �9:8. Unfortunately,
due to saturation at wavelengths longward of 20 �m, we have
an insufficient spatial coverage of LL data in the central region.
In addition, the LL slits are �1000 wide, providing low spatial
resolution.

For the higher spatial resolution, we used the simple method
described by Spoon et al. (2007) to estimate �9:8. In this method
we approximate the mid-IR continuum at 9.8 �m by a power-law
fit to the flux pivots at 5.5 and 14.5 �m, avoiding the PAH emis-
sion features.

The values derived from both SL+LL and simple-fit methods
agree qualitatively well, but in regions where �9:8 < 1, the differ-
ence between the two methods is greater than a factor of 2. Using
a mixed attenuation law with Av /�9:8 ¼ 16:6 (Rieke & Lebofsky
1985), the range of �9:8 corresponds to 0 < Av < 41:8 for the
SL+LL method. The values of �9:8 from the power-law interpo-
lation method give 5:0 < Av < 51:5. These values are in agree-
ment with Förster Schreiber et al. (2001), who derive 23 < Av <
45 for their selected regions in M82 covering an area closer to the
infrared peaks.

We conclude that both methods, besides their significant un-
certainties in the magnitude of �9:8, have consistently revealed
significant variations in the amount of dust extinction across the
central region. These variations are strong enough to affect the
following interpretation of PAH features.

4.4. Variations of PAH Emission Features

Polycyclic aromatic hydrocarbons (PAHs) are thought to be re-
sponsible for a series of broad emission features that dominate the
mid-infrared spectra of starbursts (e.g., Peeters et al. 2004b). They
are observed in a diverse range of sources, with their strongest
emission originating in photodissociation regions (PDRs), the
interfaces between H ii regions and molecular clouds.

The relative strength of the different PAH bands is expected to
vary with the size and the ionization state of the PAH molecule
(Draine & Li 2001, 2007). Observations of Galactic sources
(e.g., Verstraete et al. 1996; Joblin et al. 1996; Vermeij et al.
2002) have shown that the relative strengths of individual PAH
features dependonthedegree of ionization: CYC stretchingmodes
at 6.2 and 7.7 �m are predominantly emitted by PAH cations,
while the CYH out-of-plane bending mode at 11.3 �m arises
mainly from neutral PAHs (Draine & Li 2001). Thus, the ratios
6.2/11.3 and 7.7/11.3 �m may be used as indicators of the PAH
ionization state. Joblin et al. (1996) have found that the 8.6/11.3 �m
ratio can also be linked to variations in the charge state of the
emitting PAHs, which is supported by laboratory experiments
(Hudgins & Allamandola 1995). Smith et al. (2007b) found band
strength variations of factors of 2Y5 among normal galaxies.

Studies of PAHs in M82 have been done previously using
ISO. Observations with ISOCAM (Förster Schreiber et al. 2003b)
revealed a decrease in the 6.2/7.7 �m ratio and an increase in the
8.6/11.3 �m PAH ratio from the nucleus outward along the ga-
lactic plane. The observations also showed a good spatial cor-
relation of the 8.6/11.3 �m ratio with the CO (1Y0) emission.
These ratio variations are attributed to real differences in the var-
iation of physical characteristics of PAHs across M82, specifi-
cally a higher degree of PAH ionization within the most intense
starburst sites.

We use CUBISM to build maps of ratios of PAH features that
exist in the 6Y14 �m SL spectra. The 7.7 �m feature is split be-
tween the two SL orders, and the 8.6 �m feature is largely in-
fluenced by the 7.7 �m feature, making the local continuum
determination difficult. For these reasons the 6.2 �m feature map
was chosen to be compared to the 11.3 �m feature map. In Fig-
ure 8 we present the 6.2/11.3 �m PAH ratio map, overlaid with
the CO (1Y0) contours from Shen & Lo (1995).

Obviously, the distribution of the 6.2/11.3 �m PAH ratio cor-
relates well with the molecular ring defined by the CO (1Y0) map.
The molecular emission forms two lobes �200 pc from the nu-
cleus, whereas the ionized emission is concentrated around the
nucleus, in the regions where the [Ne ii] and [Ne iii] emissions
peak.

However, the 6.2/11.3 �m PAH ratio may be affected by ex-
tinction. Comparing the CO emission with the extinction map in
Figure 7 and with the PAH ratio map, we see that they correlate

Fig. 8.—IRS spectral map of the PAH ratio 6.2/11.3 �m, with overlay of CO
1Y0 emission contours from Shen & Lo (1995). The map is in logarithmic scaling.
This map was done using CUBISM maps of baseline-subtracted flux at 6.0Y6.5
and 11Y11.7 �m.

SPITZER IRS SPECTROSCOPY OF M82 311No. 1, 2008



well. Hence, we investigate the effects of extinction on the ob-
served PAH ratios in the next subsection.

4.4.1. Influence of Silicate Absorption on PAH Ratios

The fluxes of the PAH bands at 6.2, 7.7, 8.6, and 11.3 �m are
all affected by dust absorption. However, the 11.3 �m is sig-
nificantly more affected by the 9.8 �m silicate feature than the
6.2 �m feature, which affects the measured PAH ratio shown in
Figure 8. To illustrate this influence, we show in Figure 9 the
baseline-subtracted spectra of the selected regions around the
nucleus in M82, normalized to the flux at 7.7 �m to emphasize
the relative flux variations of the 11.3 �m PAH feature. The gray
line is the Chiar & Tielens (2006) extinction law, applied to a
normalized flux of 1.75, assuming �9:8 ¼ 2. For this figure, the
baseline was removed by subtracting a second-order polynomial
fitted to the following wavelengths: 5.5, 6.8, 8.0, 13.2, and
14.5 �m. These wavelengths were chosen to avoid the silicate
feature at 10 �m and the PAH features. The effect of extinction
on the 8.6 and 11.3 �m PAH features is similar, meaning that the
8.6/11.3 �m ratio could possibly be used to study PAH ionization
with minimal concern for extinction, as the difference in flux cor-
rection between 8.6 and 11.3 �m is �15%. However, the 8.6 �m
feature is influenced by the broad PAH feature at 7.7 �m, which
makes local continuum fitting more difficult compared to the
6.2 �m feature.

Figure 9 suggests that the variations of PAH feature ratios in-
volving the 11.3 �m feature will be heavily affected by extinc-
tion, contributing to the distribution of the 6.2/11.3 �m PAH
ratio in Figure 8. The 7.7 �m PAH feature is affected by extinc-
tion to a similar level as the 6.2 �m feature, as shown in Figure 9.
Draine & Li (2001) use the 6.2/7.7 �m ratio for PAH size di-
agnostic and 11.3/7.7 �m for a PAH ionization state diagnostic.
Variations of these ratios reflect real variations of the physical
properties of PAHs in M82 only if their variations are not due to
extinction effects. Figure 10 presents the 6.2/7.7 and 11.3/7.7 �m
PAH ratios calculated from PAHFIT measurements of the same
subregions as in the extinction map. The data in Figure 10 are
corrected for extinction. We took the average of the extinction
methods (continuum fit method and the PAHFIT fit on SL+LL1
spectra; see x 4.3) for correction. The points are dispersed be-
tween the tracks representing totally ionized and neutral PAH
populations, and the error bars show the average of the differ-

ence between the extinctions derived by the two methods. The
triangles are points from the areas where ½Ne iii�/½Ne ii� > 0:24
in the [Ne iii]/[Ne ii] map from Fig. 5. These are regions with
harder radiation field, where a greater number of ionized PAH are
expected to be observed. The squares are points from areas in the
map where ½Ne iii�/½Ne ii� < 0:13. To preserve the clarity of the
plot, only the error bars at these points are represented, as they are
typical for all the points. The arrow represents the effect on the
ratios of a silicate absorption feature with �9:8 ¼ 1:34, which is
the median optical depth for the region as explained in x 4.3.
The 11.3/7.7 �m PAH ratios are dispersed between 0.2 and

0.5. However, the error bar can account for this dispersion, dem-
onstrating how silicate absorption can affect thediagnostic of PAH
ionization state based on this ratio, on scales of the size of the IRS
resolution elements, �35 pc. This implies that the average var-
iations in the ionization state of the grains are relatively small on
scales of 35 pc. These results are to be compared to SINGS re-
sults, in which the 11.3/7.7 �m PAH ratio is found to lie within
the same range as that in the central region of M82 for galaxies
dominated by H ii regions (Smith et al. 2006).
The 6.2/7.7 �m ratio varies between �0.2 and �0.3. This is

well within the range of 0.2Y0.4 reported by Smith et al. (2007b)
for SINGS galaxies dominated by H ii regions. We observe no
significant difference in the 6.2/7.7 �m ratio between regions with
a high [Ne iii]/[Ne ii] ratio and regions with a lower [Ne iii]/[Ne ii]
ratio. Extinction and fitting errors affect the 6.2/7.7 �m ratio by
only �1%, and errors from fitting residuals from PAHFITamount
to less than 2%. However, there are significant uncertainties aris-
ing from the continuum fitting by blackbody curves with effec-
tive temperatures between 35Y300 K. Modifying the number of
blackbody components and their temperatures results in changes
in the PAH ratios in excess of 8%Y12%. The horizontal error bar
represents the average uncertainty of 10%.
The range of 6.2/7.7 �m ratios implies an environment com-

posed of a warm ionized medium and photodissociation regions
(PDRs; Draine & Li 2001). The dispersion in the data could
reflect a real variation of this ratio. The general significance of
the variation of the 6.2/7.7 �m is discussed on Draine & Li (2001).
Assumptions about the stellar radiation intensity, which affect
the 11.3/7.7 �m ratio as well, account for these variations in the
number of C atoms. Following Draine & Li (2001), our observed

Fig. 9.—Baseline subtracted IRS spectra of the selected regions from Fig. 1.
The gray line is the Chiar & Tielens (2006) extinction law, applied to a normalized
flux of 1.75, and assuming �9:8 ¼ 2.

Fig. 10.—The 6.2/7.7 �m PAH ratio vs. the 11.3/7.7 PAH ratio, for the same
subregions as in the absorption map in Fig. 7. The triangles are points from the
areas where ½Ne iii�/½Ne ii� > 0:22, and the squares are points from the areas where
½Ne iii�/½Ne ii� < 0:13. Error bars represent average uncertainties, related to ex-
tinction correction and fitting uncertainties. The arrow represents the effect of a
silicate absorption feature with �9:8 ¼ 1:34. The tracks represent the ratios of a
population of neutral and ionized PAHs.

BEIRÃO ET AL.312 Vol. 676



ratios correspond to a number of carbon atoms in a PAH grain
between 100 and 140 (6:2/7:7 �m � 0:3) and 240 and 320
(6:2/7:7 �m � 0:2).

However, there is no correlation between the 6.2/7.7 �m ratio
with radiation hardness (symbols in Fig. 10) or with any other
resolved spatial structure in the central region. Since the variations
are comparable to the uncertainties in the measurement, the results
are not (yet) significant enough to support PAH size variations at
parsec scales. Similarly, the variations of the 11.3/7.7 �m ratio
are mostly due to extinction and show little support for variations
in PAH ionization throughout the region. However, if PAH sizes
did vary on parsec scales, one would expect strong emission from
the larger grains, which radiate predominantly at larger wave-
lengths, such as the 17 �m complex.

4.4.2. The 17 �m PAH Complex

The16Y18�m wavelength rangecontains the H2 S(1) 17.03 �m
line and the 17 �m PAH complex. This complex is attributed
to a blend of emission features (16.45, 17.03, and 17.37 �m),
which are possibly due to PAH CYCYC bending modes (e.g., Van
Kerckhoven et al. 2000), and emission from PAH clusters, amor-
phous carbon particles, and other PAH-related species (Peeters
et al. 2004a). Following the Draine & Li (2007) models, the 17 �m
complex is mostly emitted by large PAH molecules with 1000Y
2000 carbon atoms, while the 6.2 �m feature is emitted mostly
by smaller PAH molecules with only 200Y300 carbon atoms.
The relatively narrow features at 16.45 and 17.37 �m have been
detected by ISO in Galactic (Moutou et al. 2000) and extraga-
lactic sources (Sturm et al. 2000). Only with the Spitzer IRS
has the whole 17 �m PAH complex been identified and routinely
detected in both normal and starburst galaxies (e.g., Smith et al.
2004; Dale et al. 2006; Brandl et al. 2006), including the outer
regions of M82 (Engelbracht et al. 2006). However, the spatial
variation within galaxies other than the Milky Way has not yet
been studied. Engelbracht et al. (2006) detected this complex in
the disk and in the halo of M82, but their spectral coverage did
not include the central regions. Here we report the characteristics
of this complex in the central kpc of M82.

As seen in Table 3 the ratios between the 17 �m complex and
the 11.3 �m PAH feature, as measured with PAHFIT, range from
0.39 in regions 1 and 3 to 0.49 in the center region, and they do
not show clear spatial correlations. However, the 17/6.2 �m ratio
varies from 0.21 in region 3 to 0.57 in region 4. The ratios be-
tween the regions in the plane and the outward regions vary by a
factor of 2. The latter numbers suggest a significant variation,
possibly due to different PAH sizes. However, correcting for an
optical depth of �9:8 ¼ 2 for the regions around the [Ne ii] peaks
(Fig. 7), the ratio increases by �67%, yielding values of 0.37 for
region 2 and 0.35 for region 3. While substantially increased, the
values are still more than 20% below the values for regions 1
and 4. However, due to uncertainties on the determination of �9:8
we cannot diagnose any variation of PAH sizes based on this
difference.

Figure 11 shows a baseline-subtracted spectrum of the 17 �m
PAH complex normalized to the average flux between 16.2 and
17.6 �m and rebinned to R ¼ 300 to enhance the broad com-
ponents. In this figure we can see the flux variation in these three
main PAH features between the regions. The differences in the
strength of these features are consistent within the uncertainties,
with exceptions being region 4 at 16.45 �m and regions 2 and 3
at 17.37 �m. Moreover, these discrepancies could be explained
by errors in baseline subtraction, which are in the order of 10%.
The 17 �m PAH complex is also seen in the LL spectra as far as
2 kpc from the galactic plane. The model of Draine & Li (2007)

predicts that the relative strength of the individual components
is not sensitive to ionization state or grain size. Our observa-
tions are in good agreement with an invariant shape of this PAH
complex.

4.4.3. The Difference between PAH Emission and VSG Emission

Very small grains (VSGs) are dust particles larger on average
than PAHs, with typical sizes in the range of 1Y150 nm (Desért
et al. 1990). They are excited by stochastic heating and are
thought to be responsible for most of the mid-infrared continuum
emission. The properties of these grains in relation to PAHs have
been studied previously in galactic sources (Verstraete et al. 1996;
Lebouteiller et al. 2007) and in dwarf galaxies (Madden et al.
2006; Wu et al. 2006). These studies focus on the behavior of
VSGs in conditions in which PAHs are destroyed, mainly by
intense stellar radiation in H ii regions or a low-metallicity en-
vironment. Flux differences between VSGs and PAHs are re-
ported in Galactic H ii regions by (e.g., Lebouteiller et al. 2007),
where the PAH/VSG emission ratio increases with the distance
from the cluster.

As shown in Table 3, there are significant differences in the
EWs of the PAH features between the regions in the galactic
plane and the regions outside the galactic plane. However, the
EW shortward of 10 �m behaves differently from the EW long-
ward of 10 �m. Longward of 10 �m, we observe that the EWs of
the PAH features increase outward from the galactic plane. The
11.3 �m EW, for example, increases from 0.660 �m in region 2 to
2.27 �m in region 1. The EWs of the 6.2 and 7.7 �m decrease
outward from the galactic plane. For example, the 6.2 �m EW
decreases from 1.10 �m in region 2 to 0.308 �m in region 1. The
EW of the 8.6 �m feature decreases up to a factor of 2, from
0.865 �m in region 3 to 0.437 �m in region 2. This could be due
to the contribution of the stellar continuum to the local contin-
uum shortward of 10 �m. The stellar continuum contribution
decreases with wavelength and with the distance from the ga-
lactic plane.

The continuum longward of 10 �m is composed of thermal
emission from VSGs. As the PAH flux does not increase outward
from the galactic plane, this can only be due to a decrease of the
local continuum emission relative to the PAH strength. This de-
crease can be explained by several different factors: photodes-
truction, abundance differences between VSGs and PAHs, and

Fig. 11.—Zoom-in on the PAH-dominated emission features in the 16Y18 �m
spectral range of the SH spectra. The spectra were normalized to the average flux
between 16.2 and 17.6 �m and rebinned to a resolution R ¼ 300, to diminish the
noise. The error bars indicate noise at three wavelengths.

SPITZER IRS SPECTROSCOPY OF M82 313No. 1, 2008



differentheatingopacitiesbetween VSGs andPAHs with a varying
radiation field.

Decreasing PAH flux with the hardening of the radiation has
been observed in galactic star-forming regions (Verstraete et al.
1996; Lebouteiller et al. 2007), on small spatial scales near the
luminous clusters (�2 pc), where the PAH destruction largely
surpasses PAH excitation. Whether theconditionsthat lead to PAH
destruction on 2 pc scales can be maintained over much larger
scales, corresponding to the resolution of our maps (�35 pc),
cannot be derived from our data but has been observed in NGC
5253 (Beirão et al. 2006). Using the SINGS sample of galaxies,
Draine et al. (2007) have shown that the fraction of PAH abun-
dance over the total dust abundance decreases with metallicity.
Hence, abundance differences between PAHs and VSGs could
be observed in cases of a strong metallicity gradient. M82 has a
metallicity gradient, but it becomes noticeable only at distances
larger than 1 kpc from the center (Ranalli et al. 2006), which is
greater than the distance from our regions 1 and 4 to the galactic
plane (400 pc).

Our favored explanation is the difference in excitation between
PAHs and VSGs, enhanced by variations of the radiation field.
The mid-infrared PAH features are produced by vibrational-
rotational modes of the PAH molecules, while the VSG contin-
uum emission is mainly produced by thermal radiation. VSGs are
bigger than PAHs, so thermal radiation becomes dominant over
vibrational-rotational transitions. Following Draine & Li (2001)
models, the opacity of the cross-section peaks at FUV wave-
lengths, where hot dust is needed to emit at 10Y20 �m. However,
at the distance from the galactic plane of regions 1 and 4, the
radiation field is still intense enough to excite PAHs but no longer
of a high enough intensity to excite the dust to the same tem-
peratures as in the plane of the galaxy. This provokes the PAH
EW enhancement observed in these regions.

4.5. Excitation of the Warm H2

Molecular hydrogen is the most abundant molecule in the uni-
verse and can be used to probe the properties of the warm mo-
lecular gas in M82. It can be traced in the mid-infrared through
rotational emission lines, which may arise through three different
mechanisms: UV excitation in PDRs surrounding or adjacent to
the H ii regions; shocks that accelerate and modify the gas in a
cloud, collisionally exciting the H2 molecules; or hard X-ray
photons capable of penetrating the molecular clouds and heating
large ionizing columns of gas.

Vibrational-rotational H2 line emission in M82 was studied in
the near-infrared by Pak et al. (2004). By correlating the emis-
sion flux from these lines with [C ii] 157 �m and far-infrared
luminosity, they showed that the H2 emission comes mainly from
the PDRs. ISO observations of M82 detected S(0), S(1), S(2),
S(6), and S(7) rotational lines (Rigopoulou et al. 2002), excited
by UV radiation from massive stars.

The S(1) and S(2) rotational transition lines of H2 are clearly
detected in our SH spectra. The fluxes and temperatures of these
lines are listed in Table 2, along with the S(1)/S(2) ratios. The
temperatures were calculated from the S(1)/S(2) ratios using the
method described by Roussel et al. (2006) assuming an ortho- to
para- ratio of 3. Our derived temperatures are in agreement
with the average temperature value of 450 K, derived with ISO
(Rigopoulou et al. 2002). The H2 S(0) line could not be detected
in any of the LH spectra, as its equivalent width is very low,
supporting our finding that the H2 temperature is indeed rela-
tively high.

If H2 emission is excited mainly by UV radiation in PDRs, the
fluxes of H2 lines and PAH lines should correlate closely, since

PAH emission features arises from the same mechanism in PDRs.
To look for secondary effects, such as shocks, we plot in Fig-
ure 12 the PAH 11.3 �m versus H2 S(1) 17.03 �m fluxes, both
divided by the continuum flux at 14.8Y15.2 �m to reduce the
range covered by the figure. Each data point corresponds to a
resolution element of the CUBISM flux maps taken from SH
spectra. The stars correspond to region 1, squares to region 2,
diamonds to region 3, and triangles to region 4. The error bars
represent the line uncertainties from the measurement. The cor-
relation in Figure 12 shows that the excitation mechanisms for
both species coincide at least on spatial scales of the resolution
of the map, which is 35 pc. While there is a wider dispersion of
data points corresponding to regions 1 and 4 (the regions asso-
ciated with outflows), the systematic measurement errors account
largely for this dispersion. We conclude that shocks in the out-
flow regions have no measurable influence on the H2 emission.
However, we cannot distinguish between UVand local shock ex-
citation, as produced by supernovae or energetic outflows, on
scales smaller than the resolution of our pixels (35 pc).

5. CONCLUSION

We presented spatially resolved mid-infrared spectra of the
central region of M82. The spectra were taken with the Spitzer
Infrared Spectrograph in both the 5Y38 �m low-resolution (R �
65Y130) and the 10Y37 �m high-resolution R � 600 module.
The high signal-to-noise ratio and the continuous spatial and spec-
tral coverage allowed us to study the nucleus of a starburst galaxy
in unsurpassed detail. Our goal was to study the physical con-
ditions of the interstellar medium and their spatial variations
within the central kpc of M82.
Overall, the spectra show the typical features of a starburst:

prominent PAH features, silicate absorption, fine-structure lines,
and a steeply rising continuum. We built a spectral map with
high-resolution spectra, selected six representative regions for
spectral extraction, and studied the variations of the neon ionic
lines and PAH feature emission among the regions.
We attempted to trace the structure of the ionizing radiation,

and therefore the young stellar population of M82, through the
diagnostic emission lines [Ne ii] 12.8 �m and [Ne iii] 15.5 �m.
The overall morphology of the ionizing radiation in the region as
revealed by the spectral maps appears to be in agreement with
previous ground-based observations (Achtermann & Lacy 1995).
There is surprisingly little spatial variation of the [Ne iii]/[Ne ii]

Fig. 12.—PAH 11.3 �m feature over the 15 �m continuum vs. H2 S(1) over
the 15 �m continuum, based on a pixel-by-pixel correlation between SH maps. The
values corresponding to each of the selected regions are represented by different
symbols. The error bars represent line flux measurement errors.

BEIRÃO ET AL.314 Vol. 676



ratio across the disk, and it varies only by a factor of 3 throughout
the central region, with these variations not corresponding to the
position of the emission peaks. We observed an increase of the
[Ne iii]/[Ne ii] ratio by a factor of 2 outward from the galactic
plane, which may be associated with the outflows. We suggest that
the increase of the [Ne iii]/[Ne ii] ratio with distance to the ga-
lactic plane is due to a decrease in gas density rather than a hard-
ening of the field.

The [Ne iii]/[Ne ii] ratio is low on average, which indicates
that the dominant population consists of already evolved clusters
(>5 Myr). We cannot rule out the presence of ongoing star for-
mation in the central region, but it must occur at a much reduced
rate (<5%) compared to previous epochs. This drop is unlikely
to be caused by starvation, as there is still a large amount of mo-
lecular gas present in the central region. It is more likely due to
negative feedback processes causing a decrease in the star for-
mation rate.

There are significant variations in the amount of dust extinc-
tion, which strongly correlate with the CO 1Y0 emission across
the central region. These variations are strong enough to affect
the interpretation of PAH features, but due to limited spectral
coverage the extinction estimates are uncertain.

The flux of the main PAH features correlates spatially with the
flux of the neon ionic lines, and with previous IRAC observa-
tions. Variations in the PAH ratios such as 6.2/11.3 �m were ob-
served across the disk. However, they are strongly affected by the
silicate feature at 10 �m. We studied the variations the 6.2/7.7 �m
and 11.3/7.7 �m PAH ratios, which are diagnostics for the size
and the degree of ionization of PAHs. We found no correlation
between the 6.2/7.7 �m ratio with radiation hardness or with any
other resolved spatial structure in the central region. Since the
variations are comparable to the uncertainties in the measure-
ment, the results are not (yet) significant enough to support PAH
size variations at parsec scales. Similarly, the variations of the
11.3/7.7 �m ratio are mostly due to extinction and show little
support for variations in PAH ionization throughout the region.

The 17 �m PAH complex is very prominent in the center
of M82. We did not find any relative variations within the com-
plex, which is in agreement with predictions. The variations of
the 17/6.2 �m ratio are most likely due to extinction effects. Due
to the uncertainties in the determination of extinction, we did not

consider the remaining variations as a clear indicator of PAH size
variation.

We observed an enhancement of the EWs of the 11.3 �m and
the other PAH features longward of 10 �m outward from the
galactic plane. Several explanations exist for this, but we favor
the variation of the UV radiation field, which excites differently
PAHs and VSGs, given their different sizes.

The S(1) and S(2) rotational transition lines of H2 have been
detected in our spectra throughout the central region. H2 and
PAHs coincide at least on spatial scales of the resolution of the
map, which is 35 pc. We conclude that large-scale shocks in the
outflow regions have no measurable influence on the H2 emis-
sion. However, we cannot distinguish between UVand local shock
excitation as produced by supernovae or energetic outflows, on
scales smaller than the resolution of our pixels (35 pc).

The Spitzer IRS observations of the central region of M82
complements previous studies, not only in mid-infrared, but also
in other wavelengths. Our results demonstrate the importance of
spatially resolved spectroscopy in starburst studies. They helped
to constrain the age of the starburst and confirm results from
other studies and also stressed the importance of a thorough study
of extinction to investigate possible variations of PAH proper-
ties. Further research of the starburst feedback and quenching
processes will elucidate the sharp decrease in star formation in
the last 5 Myr. It would be interesting to see if higher spatial
resolution, as expected from the MIRI on board the James Webb
Space Telescope (JWST ) will discover a wider variation of the
radiation field or signs of variation of PAH sizes an/or ioniza-
tion. In addition, spatially more extended spectral studies of M82
are necessary to study the connection of the ionic line ratios with
the outflows.

We would like to thank L. Snijders for making the Starburst99+
Mappings models available. We would like to thank J. Lacy for
providing the mid-infrared maps from IRTF ground-based ob-
servations. We also thank the anonymous referee, whose many
thoughtful and useful comments greatly improved this paper.
This work is based on observations made with the Spitzer Space
Telescope, which is operated by the Jet Propulsion Laboratory,
California Institute of Technology, under NASA contract 1407.

REFERENCES

Achtermann, J. M., & Lacy, J. H. 1995, ApJ, 439, 163
Alton, P. B., Davies, J. I., & Bianchi, S. 1999, A&A, 343, 51
Appleton, P. N., Davies, R. D., & Stephenson, R. J. 1981, MNRAS, 195,
327

Beirão, P., Brandl, B. R., Devost, D., Smith, J. D., Hao, L., & Houck, J. R.
2006, ApJ, 643, L1

Brandl, B. R., et al. 2006, ApJ, 653, 1129
Bregman, J. N., Schulman, E., & Tomisaka, K. 1995, ApJ, 439, 155
Cappi, M., et al. 1999, A&A, 350, 777
Chiar, J. E., & Tielens, A. G. G. M. 2006, ApJ, 637, 774
Colbert, J. W., et al. 1999, ApJ, 511, 721
Dale, D. A., et al. 2006, ApJ, 646, 161
de Grijs, R., O’Connell, R., & Gallagher, J. S. 2001, AJ, 121, 768
Desért, F.-X., Boulanger, F., & Puget, J. L. 1990, A&A, 237, 215
Draine, B. T., & Li, A. 2001, ApJ, 551, 807
———. 2007, ApJ, 657, 810
Draine, B. T., et al. 2007, ApJ, 663, 866
Engelbracht, C. W., et al. 2006, ApJ, 642, L127
Förster Schreiber, N. M., Genzel, R., Lutz, D., Kunze, D., & Sternberg, A.
2001, ApJ, 552, 544

Förster Schreiber, N. M., Genzel, R., Lutz, D., & Sternberg, A. 2003a, ApJ,
599, 193

Förster Schreiber, N. M., Sauvage, M., Charmandaris, V., Laurent, O., Gallais,
P., Mirabel, I. F., & Vigroux, L. 2003b, A&A, 399, 833

Freedman, W. L., & Madore, B. F. 1988, ApJ, 332, L63

Heckman, T. M., Lehnert, M. D., Strickland, D. K., & Armus, L. 2000, ApJS,
129, 493

Higdon, S. J. U., et al. 2004, PASP, 116, 975
Hoopes, C. G., et al. 2005, ApJ, 619, L99
Houck, J. R., et al. 2004, ApJS, 154, 18
Hudgins, D. M., & Allamandola, L. J. 1995, J. Phys. Chem., 99, 3033
Joblin, C., Tielens, A. G. G. M., Geballe, T. R., & Wooden, T. H. 1996, ApJ,
460, L119

Kronberg, P. P., Biermann, P., & Schwab, F. R. 1985, ApJ, 291, 693
Larkin, J. E., Graham, J. R., Matthews, K., Soifer, B. T., Beckwith, S., Herbst,
T. M., & Quillen, A. C. 1994, ApJ, 420, 159

Lebouteiller, V., Brandl, B., Bernard-Salas, J., Devost, D., & Houck, J. R. 2007,
ApJ, 665, 390

Lehnert, M. D., Heckman, T. M., & Weaver, K. A. 1999, ApJ, 523, 575
Lipscy, S. J., & Plavchan, P. 2004, ApJ, 603, 82
Lo, L. Y., Cheung, K. W., Masson, C. R., Phillips, T. J., Scott, S., & Woody, D. P.
1987, ApJ, 312, 574

Madden, S., Galliano, F., Jones, A. P., & Sauvage, M. 2006, A&A, 446, 877
Mayya, Y. D., Carrasco, L., & Luna, A. 2005, ApJ, 628, L33
Moutou, C., Verstraete, L., Léger, A., Sellgren, K., & Schmidt, W. 2000, A&A,
354, L17

Muxlow, T. W. B., Pedlar, A., Wilkinson, P. N., Axon, D. J., Sanders, E. M., &
de Bruyn, A. G. 1994, MNRAS, 266, 455

Pak, S., Jaffe, D. T., Stacey, G. J., Bradford, C. M., Klumpe, E. W., & Keller, K. D.
2004, ApJ, 609, 692

SPITZER IRS SPECTROSCOPY OF M82 315No. 1, 2008



Pauldrach, A. W. A., Lennon, M., Hoffmann, T. L., Sellmaier, F., Kudritzki, R.-P.,
& Puls, J. 1998, in ASP Conf. Ser. 131, Properties of Hot Luminous Stars, ed.
I. Howarth (San Francisco: ASP), 258

Peeters, E., Mattioda, A. L., Hudgins, D. M., & Allamandola, L. J. 2004a, ApJ,
617, L65

Peeters, E., Spoon, H. W. W., & Tielens, A. G. G. M. 2004b, ApJ, 613, 986
Ranalli, P., Origlia, L., Comastri, A., Maiolino, R., & Makishima, K. 2006, in
The X-Ray Universe, ed. A. Wilson (ESA SP-604; Noordwijk: ESA), 839

Rieke, G. H., & Lebofsky, M. J. 1985, ApJ, 288, 618
Rigopolou, D., Kunze, D., Lutz, D., Genzel, R., & Moorwood, A. F. M. 2002,
A&A, 389, 374

Roussel, H., et al. 2006, ApJ, 646, 841
Satyapal, S., Watson, D. M., Pipher, J. L., Forrest, W. J., Greenhouse, M. A.,
Smith, H. A., Fischer, J., & Woodward, C. E. 1997, ApJ, 483, 148

Shen, J., & Lo L. Y. 1995, ApJ, 445, L99
Shopbell, P. L., & Bland-Hawthorn, J. 1998, ApJ, 493, 129
Smith, J. D., et al. 2004, ApJS, 154, 199
———. 2007a, PASP, 199, 1133
———. 2007b, ApJ, 656, 770
Smith, L. J., Westmoquette, M. S., Gallagher, III, J. S., O’Connell, R. W.,
Rosario, D. J., & de Grijs, R. 2006, MNRAS, 370, 513

Snijders, L., Kewley, L., & van der Werf, P. 2007, ApJ, 669, 269
Spoon, H. W. W., Marshall, J. A., Houck, J. R., Elitzur, M., Hao, L., Armus, L.,
Brandl, B. R., & Charmandaris, V. 2007, ApJ, 654, L49

Strickland, D. K., Heckman, T. M., Colbert, E. J. M. Hoopes, C. G., & Weaver,
K. A. 2004, ApJS, 151, 193

Strickland, D. K., Ponman, T. J., & Stevens, I. R. 1997, A&A, 320, 378
Sturm, E., Lutz, D., Tran, D., Feuchtgruber, H., Genzel, R., Kunze, D.,
Moorwood, A. F. M., & Thornley, M. D. 2000, A&A, 358, 481

Sun, W.-H., et al. 2005, ApJ, 630, L133
Telesco, C. M., Joy, M., Dietz, K. Decher, R., & Campins, H. 1991, ApJ, 369,
135

Thornley, M. D., Förster Schreiber, N. M., Lutz, D., Genzel, R., Spoon, H. W.
W., Kunze, D., & Sternberg, A. 2000, ApJ, 539, 641

Turner, J. L., Beck, S. C., Crosthwaite, L. P., Larkin, J. E., McLean. I. S., &
Meier, D. S. 2003, Nature, 423, 621

Van Kerckhoven, C., et al. 2000, A&A, 357, 1013
Vermeij, R., Peeters, E., Tielens, A. G. G. M., & van der Hust, J. M. 2002,
A&A, 382, 1042

Verstraete, L., Puget, J. L., Falgarone, E., Drapatz, S., Wright, C. M., &
Timmermann, R. 1996, A&A, 315, L337

Walter, F., Weiss, A., & Scoville, N. 2002, ApJ, 580, L21
Wu, Y., Charmandaris, V., Hao, L., Brandl, B. R., Bernard-Salas, J., Spoon, H.
W. W., & Houck, J. R. 2006, ApJ, 639, 157

Yun, M. S., Ho P. T. P., & Lo, K. Y. 1993, ApJ, 411, L17
———. 1994, Nature, 372, 530

BEIRÃO ET AL.316