arXiv:astro-ph/0604007 v1   1 Apr 2006


ar
X

iv
:a

st
ro

-p
h/

06
04

00
7 

v1
   

1 
A

pr
 2

00
6

Mid-Infrared Spectral Diagnostics of Nuclear and Extra-Nuclear

Regions in Nearby Galaxies

D.A. Dale1, J.D.T. Smith2, L. Armus3, B.A. Buckalew3, G. Helou3, R.C. Kennicutt4,2,

J. Moustakas2, H. Roussel5,3, K. Sheth3, G.J. Bendo6,2, D. Calzetti7, B.T. Draine8,

C.W. Engelbracht2, K.D. Gordon2, D.J. Hollenbach9, T.H. Jarrett3, L.J. Kewley10,

C. Leitherer7, A. Li11, S. Malhotra12, E.J. Murphy13, F. Walter5

ABSTRACT

Mid-infrared diagnostics are presented for a large portion of the Spitzer In-

frared Nearby Galaxies Survey (SINGS) sample plus archival data from the In-

frared Space Observatory and the Spitzer Space Telescope. The SINGS dataset

includes low- and high-resolution spectral maps and broadband imaging in the

infrared for over 160 nuclear and extranuclear regions within 75 nearby galaxies

spanning a wide range of morphologies, metallicities, luminosities, and star for-

mation rates. Our main result is that these mid-infrared diagnostics effectively

constrain a target’s dominant power source. The combination of a high ion-

ization line index and PAH strength serves as an efficient discriminant between

AGN and star-forming nuclei, confirming progress made with ISO spectroscopy

on starbursting and ultraluminous infrared galaxies. The sensitivity of Spitzer

1
Department of Physics and Astronomy, University of Wyoming, Laramie, WY 82071; ddale@uwyo.edu

2
Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ 85721

3
California Institute of Technology, MC 314-6, Pasadena, CA 91101

4
Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0HA, United Kingdom

5
Max Planck Institut für Astronomie, Königstuhl 17, 69117 Heidelberg, Germany

6
Astrophysics Group, Imperial College, Blackett Laboratory, Prince Consort Road, London SW7 2AZ United Kingdom

7
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218

8
Princeton University Observatory, Peyton Hall, Princeton, NJ 08544

9
NASA/Ames Research Center, MS 245-6, Moffett Field, CA 94035

10
Institute for Astronomy, 2680 Woodlawn Drive, Honolulu, HI 96822

11
Department of Physics and Astronomy, University of Missouri, Columbia, MO 65211

12
Department of Physics and Astronomy, Arizona State University, Tempe, AZ 85287

13
Department of Astronomy, Yale University, New Haven, CT 06520



– 2 –

allows us to probe fainter nuclear and star-forming regions within galaxy disks.

We find that both star-forming nuclei and extranuclear regions stand apart from

nuclei that are powered by Seyfert or LINER activity. In fact, we identify ar-

eas within four diagnostic diagrams containing >90% Seyfert/LINER nuclei or

>90% H II regions/H II nuclei. We also find that, compared to starbursting

nuclei, extranuclear regions typically separate even further from AGN, especially

for low-metallicity extranuclear environments. In addition, instead of the tradi-

tional mid-infrared approach to differentiating between AGN and star-forming

sources that utilizes relatively weak high-ionization lines, we show that strong

low-ionization cooling lines of X-ray dominated regions like [Si II] 34.82 µm can

alternatively be used as excellent discrimants. Finally, the typical target in this

sample shows relatively modest interstellar electron density (∼ 400 cm−3) and

obscuration (AV ∼ 1.0 mag for a foreground screen), consistent with a lack of

dense clumps of highly obscured gas and dust residing in the emitting regions.

Subject headings: infrared: galaxies — infrared: ISM — galaxies: nuclei —

galaxies: active — H II regions.

1. Introduction

The goal of this study is to explore whether mid-infrared diagnostics developed for lumi-

nous/ultraluminous infrared galaxies and bright Galactic H II regions can be improved upon

and extended to the nuclear and extra-nuclear regions within normal and infrared-faint galax-

ies. A traditional method for characterizing a galaxy’s nuclear power source uses ratios of

optical emission lines such as [O II] 3727Å, Hβ 4861Å, [O III] 5007Å, [O I] 6300Å, Hα 6563Å,

[N II] 6584Å and [S II] 6716,6731Å (e.g., Baldwin, Phillips, & Terlevich 1981; Veilleux & Os-

terbrock 1987; Ho, Filippenko, & Sargent 1997a; Kewley et al. 2001; Kauffmann et al. 2003).

A plot of [O III]/Hβ versus [N II]/Hα, for example, will typically separate Seyferts, LINERs,

and starburst nuclei. Since nuclei are often heavily enshrouded by dust, especially in lumi-

nous and ultraluminous infrared galaxies (LIRGs and ULIRGs), an important limitation to

galaxy optical diagnostics is the effect of extinction. In anticipation of the data stream from

space-based infrared platforms, early theoretical work with photoionization models showed

that infrared ionic fine structure line ratios could profitably enable astronomers to approach

galaxy classification from a new perspective (e.g., Voit 1992; Spinoglio & Malkan 1992). The

advent of sensitive infrared line data from the Infrared Space Observatory was an important

first step to peering more deeply into buried nuclear sources (Genzel et al. 1998; Laurent et

al. 2000; Sturm et al. 2002; Peeters, Spoon, & Tielens 2004). Genzel and collaborators were



– 3 –

the first to show that ionization-sensitive indices based on mid-infrared line ratios correlate

with the strength of polycyclic aromatic hydrocarbon (PAH) emission features. AGN in

particular show weak PAH and large ratios of high-to-low ionization line emission. Inter-

estingly, while Genzel et al. found that 70-80% of their ULIRG sample is mainly powered

by starburst activity, the percentage appears to drop for higher luminosity ULIRGs (e.g.,

Veilleux, Sanders, & Kim 1999, but see also Farrah et al. 2003). Taniguchi et al. (1999)

and Lutz, Veilleux, & Genzel (1999) suggest that optical and infrared classifications agree

if the nuclei of LINER-like ULIRGs are in fact dominated by shocks driven by powerful

supernova winds. Mid-infrared lines observed by ISO were also used to probe the physical

characteristics and evolution of purely starbursting nuclei (Thornley et al. 2000; Verma et

al. 2003) and Galactic H II regions (Vermeij & van der Hulst 2002; Giveon et al. 2002). An

important result stemming from these efforts is that stellar aging effects appear to result in

H II regions generally having higher excitations than starbursting nuclei.

The unprecedented sensitivity and angular resolution afforded by the Spitzer Space

Telescope allow an even more detailed view into the nature of galaxy nuclei (e.g., Armus et

al. 2004; Smith et al. 2004). SINGS takes full advantage of Spitzer’s capabilities by executing

a comprehensive, multi-wavelength survey of 75 nearby galaxies spanning a wide range of

morphologies, metallicities, luminosities, and star formation activity levels (Kennicutt et al.

2003). The sensitivity of Spitzer coupled with the proximity of the SINGS sample allows

dwarf galaxy systems fainter than LFIR ∼ 10
7 L⊙ to be spectroscopically probed in the

infrared for the first time. In addition, prior to Spitzer the only individual extragalactic H II

regions that were detectable with infrared spectroscopy resided in the Local Group (e.g.,

Giveon et al. 2002; Vermeij et al. 2002). In contrast, SINGS provides infrared spectroscopic

data for nearly 100 extragalactic H II regions, residing in systems as near as Local Group

members to galaxies as far as ∼25 Mpc. The SINGS dataset thus samples a wider range

of environments than previously observed with infrared spectroscopy. This diversity in the

SINGS sample provides a huge range for exploring physical parameters with mid-infrared

spectral diagnostics. The high ionization lines historically used in such diagnostics, such as

[O IV] 25.89 µm and [Ne V] 14.32 µm, are relatively weak and can be difficult to detect

in lower luminosity systems. Fortunately, high ionization lines are not the only route to

determining whether a galaxy harbors a strong AGN. Similar to how the [O I] 6300Å and

[O I] 63 µm lines are AGN diagnostics (e.g., Dale et al. 2004a), we show below the utility

of using the comparatively bright [Si II] 34.82 µm mid-infrared line as another effective tool

for deciphering a galaxy’s power source.



– 4 –

2. The Sample

2.1. Galaxy Nuclei

The sample of nuclear targets analyzed in this study derive from the SINGS Third

Data Release. These 50 nuclei come from a wide range of environments and galaxies: low-

metallicity dwarfs; quiescent ellipticals; dusty grand design spirals; Seyferts, LINERs, and

starbursting nuclei of normal galaxies; and systems within the Local and M 81 groups (see

Kennicutt et al. 2003).

2.2. Extra-Nuclear Regions

The 26 extranuclear sources studied in this work also come from the SINGS Third Data

Release. These targets stem from the original set of 39 optically-selected sources listed by

Kennicutt et al. (2003). The optically-selected OB/H II regions cover a large range of

metallicity (0.1-3 Z⊙), extinction-corrected ionizing luminosity (10
49 − 1052 photons s−1),

extinction (AV . 4 mag), radiation field intensity (ionization parameter log U = −2 to −4;

Habing 1968), ionizing stellar temperature (Teff = 35 − 55 kK), and local H2/H I ratio as

inferred from CO (<0.1 to >10). Additional extranuclear targets for the SINGS project have

since been identified, based on their infrared properties, but their observations necessarily

came later than the observations of the optically-selected targets. Those “Second Look”

data will be the focus of a future paper.

3. The Data

The full SINGS observing program and the data processing are described by Kennicutt

et al. (2003), Smith et al. (2004), and Dale et al. (2005). Here we briefly summarize the

spectral observations and data processing relevant to this paper.

3.1. Spitzer Infrared Spectroscopy Observations and Data Processing

High-resolution spectroscopy (R ∼ 600) was obtained in the Short-High (10-19 µm)

and Long-High (19-37 µm) modules, and low-resolution spectroscopy (R ∼ 50 − 100) was

obtained in the Short-Low (5-14 µm) and Long-Low (14-38 µm) modules (Houck et al.

2004a). Figure 1 shows example spectra for a variety of sources (Long-Low data are not



– 5 –

used elsewhere in this work). Nuclei were generally mapped with a 3×5 grid (Short-High and

Long-High) and a 1×18 grid (Short-Low), utilizing half-slit width and half-slit length steps.

Extra-nuclear targets were observed with a similar scheme but with a 1×9 Short-Low grid.

For a subset of nine sources with extended circumnuclear star formation we obtained slightly

larger (6×10) Short-High nuclear maps. Owing to the different angular sizes subtended by

the instruments, the resulting maps are approximately 57′′×31′′ and 57′′×18′′ in Short-Low

(nuclear and extra-nuclear, respectively), 45′′×33′′ in Long-High, and 23′′×15′′ in Short-

High (the 10 expanded Short-High nuclear maps are 40′′×28′′). All integrations are 60 s per

pointing, except the Short-Low nuclear maps are 14 s per pointing. The effective integrations

are longer since each location was covered 2-4 times.

The individual data files for a given spectral map were assembled into spectral cubes

using the software CUBISM (Kennicutt et al. 2003; Smith et al. 2006, in preparation). The

cube input data were pre-processed using version S12.0 of the Spitzer Science Center pipeline.

Various post-processing steps within CUBISM are described by Smith et al. (2004). Short-

Low sky subtraction was enabled via the extended off-source wings of the Short-Low module

(and occasionally using spatially- and temporally-proximate data from the archive when

our off-source wings do not extend to the sky). Several cross-checks on the flux calibration

were made between Short-Low, Short-High, Long-Low, Long-High, MIPS, and IRAC data.

The absolute flux calibration uncertainty for the spectral data is estimated to be 25%; the

uncertainy in line flux ratios is ∼10%.

Though the Short-High, Long-High, and Short-Low cubes all span different solid angles,

the same matched extraction apertures were used for all cubes: one-dimensional spectra were

extracted from the three-dimensional data using ∼23′′×15′′ apertures. Furthermore, the

extraction apertures are centered on the optically-derived coordinates listed by Kennicutt

et al. (2003); the optical coordinates generally coincide with the infrared emission peak.

Emission line and PAH feature fluxes and equivalent widths are derived from continuum-

subtracted Gaussian fits to the lines and first- or second-order polynomial fits to the continua.

3.2. Archival Spectroscopy

ISO-SWS and Spitzer IRS line fluxes are drawn from the literature for a wide variety

of sources including Galactic, Magellanic Cloud, and Local Group H II regions (Vermeij et

al. 2002; Giveon et al. 2002; Peeters et al. 2002), and starburst and active galaxies (Genzel

et al. 1998; Sturm et al. 2002; Verma et al. 2003; Armus et al. 2004; Peeters, Spoon, &

Tielens 2004; Weedman et al. 2005; Haas et al. 2005). Equivalent widths of the 6.2 µm

PAH feature were extracted from archival ISO-PHOT and Spitzer IRS data, when available.



– 6 –

Note that, for a given source, only data from Spitzer or only data from ISO are used; cross-

observatory data are not used in this analysis. The fields-of-view of the ISO-PHOT and

ISO-SWS apertures (respectively 24′′×24′′ and 14′′×20′′ to 20′′×30′′) provide a reasonable

match to the ∼23′′×15′′ extraction apertures described in § 3.1.

3.3. Spitzer Broadband Imaging: 24 µm

The SINGS project did not obtain nearby sky observations in HighRes mode, so if the

underlying hot dust continuum emission is detected, the foreground/background continuum

is not subtracted. SINGS IRS HighRes spectroscopy is designed to measure line fluxes

and thus line equivalent width measures via HighRes are biased by the un-subtracted sky

emission. On the other hand, the SINGS program includes extensive MIPS (Rieke et al.

2004) 24 µm broadband imaging of all galaxies; the 24 µm data can be used to normalize

line fluxes. Sky-subtracted MIPS 24 µm fluxes are extracted from aperture cutouts matched

to the Short-High field-of-view (§ 3.1). See Kennicutt et al. (2003) and Dale et al. (2005)

for further details of the SINGS broadband imaging.

3.4. Optical Spectroscopy Observations and Data Processing

Optical spectrophotometry has been obtained for the SINGS project at the Steward

Observatory Bok 2.3 m and the CTIO 1.5 m telescopes (Moustakas & Kennicutt 2006). A

suite of spectral drift scans, centered on the nuclei, were taken to spatially map the various

regions covered with the IRS spectroscopy program (see Kennicutt et al. 2003 for more

details). The optical spectroscopy has spectral resolution of ∼8Å and covers 3600-7000Å

so that the primary nebular emission lines can be studied (e.g., [O II] 3727Å, Hβ 4861Å,

[O III] 5007Å, Hα 6563Å, [N II] 6584Å). The SINGS optical spectroscopy and emission line

measurements (fluxes, equivalent widths, metallicities, etc.) will be presented by Moustakas

et al. (2006, in preparation).

This work utilizes optical spectral drift scans integrated over the central 20′′×20′′ regions,

approximately matching the ∼23′′×15′′ circumnuclear regions over which the infrared line

and PAH feature fluxes are extracted (see § 3.1).



– 7 –

4. Measured Quantities

4.1. High-Resolution Infrared Spectroscopy: Emission Lines

Several forbidden lines are quite prominent in many of the SINGS nuclear and extra-

nuclear high resolution spectra ([Ne II]12.81 µm, [Ne III]15.56 µm, [S III]18.71 µm, [S III]33.48 µm,

and [Si II]34.82 µm) along with a few other higher ionization lines that are occasionally ob-

served ([S IV]10.51 µm, [Ne V]14.32 µm, and [O IV]25.89 µm; Figure 1). All these lines

except [Si II]34.82 µm are nebular lines from hydrogen gas ionized regions; [Si II]34.82 µm

comes from a wider variety of regions, including both ionized gas and warm atomic gas such

as photodissociation regions and X-ray dominated regions (e.g., Hollenbach & Tielens 1999).

A series of interstellar molecular hydrogen lines are also detected in the SINGS spectra;

these are explored in a separate paper (Roussel et al. 2006, in preparation). The line fluxes

utilized in this work are listed in Tables 1 and 2. Eight galaxies from the SINGS Third Data

Release are not listed in Table 1 because their nuclear regions were too faint to be observed

(or detected if observed).

4.2. Low-Resolution Infrared Spectroscopy: The 6.2 µm PAH Feature

The strength of a PAH feature depends on a complex combination of several parameters

of the interstellar medium, some of which are interlinked: metallicity, dust column density,

the distribution of sizes and ionization states in the dust grain population, and the intensity

and hardness of the interstellar radiation field (e.g., Cesarsky et al. 1996; Thuan et al. 1999;

Sturm et al. 2000; Li & Draine 2001; Draine & Li 2001; Houck et al. 2004b; Galliano

et al. 2005; Engelbracht et al. 2005; Madden et al. 2006; Wu et al. 2006). Due to this

sensitivity to the properties of the interstellar medium, PAH features in the mid-infrared

have been used to characterize the physical state of a system. Previous efforts have utilized

the 6.2 µm feature (Laurent et al. 2000; Peeters, Spoon, & Tielens 2004; Weedman et al.

2005), the 7.7 µm feature (Genzel et al. 1998), or a combination of several PAH features

(e.g., Verstraete et al. 2001; Tran et al. 2001; Peeters et al. 2002; Förster Schreiber et

al. 2004; Armus et al. 2004; Peeters et al. 2004; Madden et al. 2006). For this work we

concentrate on diagnostics that utilize the 6.2 µm feature. The 6.2 µm feature is the only

strong infrared signature of PAHs not blended with an emission line or absorption trough,

not near the wavelength edges of an IRS low resolution module, and for which the blue

and red sides of the “continuum” are easy to define. The equivalent widths of the 6.2 µm

feature are listed in Tables 1 and 2. The equivalent width is used in our diagnostics since the

continuum at 6.2 µm may contain emission from hot dust heated either by star formation or



– 8 –

an AGN, in addition to a fractional contribution from stellar emission; the 6.2 µm equivalent

width is (indirectly) sensitive to the presence of an AGN.

5. Results

5.1. Optical Classifications of Nuclei

Not all of the SINGS galaxies have an optical nuclear classification in the literature

(e.g., Seyfert, LINER, starburst, etc). Moreover, online databases such as NED1 provide a

heterogeneous source for such classifications. Thus we turn to our own optical spectroscopy.

Figure 2 displays a traditional diagnostic diagram (e.g., Baldwin, Phillips, & Terlevich 1981)

using optical spectroscopy and 20′′×20′′ apertures for SINGS galaxy nuclei. Note that such

diagnostics were designed with ratios of lines closely spaced in wavelength to minimize the

effects of extinction. Filled (open) circles in Figure 2 mark galaxies for which the literature

indicates a Seyfert (LINER) nucleus; data points without circles in this diagram represent

galaxies without a LINER or Seyfert classification in the literature. As alluded to in § 1,

the LINER classification can be complicated. Similar to what is observed for Seyferts, the

optical properties of LINERs are consistent with a hard power-law spectrum. But LINER-

type spectra can also be produced via winds, shocks, and cooling flows (Kauffmann et al.

2003). To further complicate the picture, “transition” objects are thought to be LINER

or Seyfert galaxies with substantial contributions from normal star formation (e.g., Ho,

Filippenko, & Sargent 1993; González Delgado et al. 2004).

The dotted lines delineate typical starburst/AGN/LINER boundaries: [O III]5007/Hβ ∼

5 and [N II]6583/Hα ∼ 0.6 (e.g., Veilleux & Osterbrock 1987; Armus, Heckman, & Miley

1989). The long-dashed curve traces the theoretical starburst/AGN boundary of Kewley et

al. (2001), marking the maximum position in this diagram that can be obtained by pure

photoionization. Objects lying above this curve require an additional power source such as

an AGN or shocks; objects lying below this curve may still contain an AGN responsible for

up to ∼30% of the emission line flux ratios. The short-dashed curve traces an empirical

starburst/AGN boundary based on data from tens of thousands of Sloan Digital Sky Survey

galaxies (Kauffmann et al. 2003). The Kauffman et al. curve aims to define a boundary

below which no galaxies contain an AGN. Objects lying in between the Kewley et al. and

the Kauffmann et al. curves are likely to be composite AGN/starburst objects but still

dominated by star formation. It is evident that the classification information available from

1NASA/IPAC Extragalactic Database



– 9 –

the literature is insufficient or incomplete for a handful of SINGS nuclei; there is overall

good agreement between the literature and our classifications, with a few exceptions. The

literature does not indicate that the nuclei of NGC 1291 contains LINER or Seyfert activity,

whereas our spectroscopic data place it squarely in the LINER category. On the other hand,

NED lists NGC 1097, NGC 4321, and NGC 4552 as having LINER or Seyfert nuclei, yet

our optical line ratios suggest they are star-forming galaxies. NGC 3198, NGC 3621, and

NGC4826 appear to lie in a transitional regime between the starburst and LINER/Seyfert

regions.2

Using the Hα/Hβ ratio extracted from our optical spectra and a screen model for the

dust distribution within a galaxy, the SINGS nuclei show modest attenuations, 〈AV 〉 ∼

1.0 mag with a dispersion of 1.0 mag and a maximum of AV ∼ 4.1 mag for NGC 1266. No

sources appear to be heavily buried in the optical so presumably none of the classifications

are skewed by large amounts of dust.3 This result is consistent with the lack of deeply buried

objects in comparisons of SINGS Hα and Spitzer 24 µm data (e.g., Calzetti et al. 2005).

However, as described above, there may be a few transitional objects for which the nuclear

classifications are difficult to interpret in the optical. The main reason behind the analysis

in this section is to provide classifications for sources that do not yet have them from the

literature. Our goal is not to carry out a detailed analysis of the relative merits of classifying

in the optical versus in the infrared, primarily since the SINGS sample is not optimally

suited for such a test. In the next section we turn to exploring new and existing infrared

diagnostics.

5.2. Infrared Spectral Diagnostics of Nuclei and Extra-Nuclear Regions

5.2.1. Emission Line Ratios and PAH Strength

At least 54.9 eV is required to remove an electron from doubly-ionized oxygen. On

the other hand, ionizing neutral neon “only” requires 21.6 eV. Compared to 21.6 eV pho-

tons, 54.9 eV photons are far more likely to stem from accretion-powered disks than star

formation (e.g., OB stars; Smith et al. 2004), and thus the ratio of [O IV]25.89 µm and

[Ne II]12.81 µm depends on the type of source dominating the energetics of the interstel-

2Similar results are found using [S II] 6716,6731Å and [O I] 6300Å data in place of [N II] 6584Å (e.g.,
Kewley et al. 2001).

3The exception is NGC 1377, a deeply obscured system for which the optical data are likely probing only

outer layer (foreground) emission (Roussel et al. 2003, 2006).



– 10 –

lar medium. Furthermore, studies show that PAH features are quite prominent throughout

much of the interstellar medium except for regions characterized by exceptionally hard ra-

diation fields such as those that arise in AGN and the cores of H II regions (Cesarsky et al.

1996; Sturm et al. 2000). A mid-infrared diagnostic diagram first put forth by Genzel et al.

(1998), and later explored by Peeters, Spoon, & Tielens (2004), plots the emission line ratio

[O IV]25.89 µm/[Ne II]12.81 µm versus the strength of a mid-infrared PAH feature. In such

plots AGN sources show enhanced [O IV]25.89 µm emission and comparatively weak PAH

feature strength. However, those two studies focussed on AGN-dominated, ULIRG, and

starburst systems. The upper panel of Figure 3 uses the 6.2 µm PAH feature and the emis-

sion line ratio [O IV]25.89 µm/[Ne II]12.81 µm in a similar mid-infrared diagnostic diagram,

but one that incorporates “normal” (starbursting/star-forming) nuclei and H II regions—

the sensitivity of Spitzer allows us to probe to far fainter levels than heretofore possible.4

As expected, normal nuclei and H II regions extend the previously-observed trend: lower-

luminosity star-forming nuclei and H II regions exhibit comparatively large 6.2 µm equivalent

widths and relatively low ratios of [O IV]25.89 µm/[Ne II]12.81 µm, indicating strong contri-

butions from [Ne II]12.81 µm cooling of H II regions and their PAH-rich photodissociation

region surroundings, and negligible emission from AGN.

High ionization lines like [O IV]25.89 µm are somewhat difficult to detect in many

SINGS sources. Two alternative diagnostic diagrams are also provided in Figure 3. Both

the middle and lower panels involve the more easily detectable [Si II]34.82 µm line, which

has an ionization potential of 8.15 eV. The normalization for the [Si II]34.82 µm line in

the middle and lower panels utilizes strong mid-infrared lines with similar ionization po-

tentials: [Ne II]12.81 µm (21.6 eV) and [S III]33.48 µm (23.3 eV). An advantage to using

[Ne II]12.81 µm as the normalization is that it lies in the much less noisy Short-High mod-

ule of IRS. Conversely, the [S III]33.48 µm line lies in the same Long-High module as the

[Si II]34.82 µm line, which can improve observing efficiency and minimize cross-module un-

certainties involving calibration and aperture matching. In addition, the short wavelength

baseline between the [Si II]34.82 µm [S III]33.48 µm lines minimizes the effects of extinction.

Why are similar trends seen in the three panels? The answer may lie in the physics

of X-ray dominated regions. As pointed out by Maloney, Hollenbach, & Tielens (1996),

the [Si II]34.82 µm line is a strong coolant of X-ray irradiated gas. In X-ray dominated

regions around AGN, the [Si II] emission dominates that from the comparatively small

H II-like regions surrounding the hard-spectrum source. Moreover, X-ray dominated re-

gions can be quite large since hard X-ray photons penetrate large column densities, and

4Additional high ionization line possibilities include [Ne V]14.32 µm and [S IV]10.51 µm, but these are
less frequently detected than [O IV]25.89 µm in SINGS spectra.



– 11 –

the conversion of X-ray energy to infrared continuum and line emission can be very effi-

cient. Maloney, Hollenbach, & Tielens (1996) predict that [Si II]34.82 µm, [O I]63 µm,

[C II]158 µm, and [C I]609 µm are the top four cooling lines within X-ray dominated regions,

with [Si II]34.82 µm having an amplitude 1-10% that of the far-infrared luminosity for an

extremely large range of physical conditions.

An argument based on interstellar density provides another possibility for the high

[Si II]34.82 µm/[Ne II]12.81 µm and SiII 34.82 µm/[S III]33.48 µm ratios in AGN. Kauf-

man, Wolfire, & Hollenbach (2006) show that the ratio [Si II](PDR)/[Si II](H II) increases

with increasing density. In fact, for low density H II regions most of the [Si II] comes from

the H II region and not the surrounding photodissociation region. Moreover, Meijerink &

Spaans (2005) show that the ratio [Si II](XDR)/[Si II](PDR) also increases with increasing

density. AGN may have their emitting gas at higher densities than typically found in star-

bursts and normal galaxies, leading to increased [Si II]34.82 µm/[Ne II]12.81 µm and SiII

34.82 µm/[S III]33.48 µm ratios. In other words, the prominent [Si II]34.82 µm line for AGN

sources may be due to strong [Si II] cooling of X-ray dominated regions or enhanced [Si II]

emission from the surrounding dense photodissociation regions.

A third scenario for enhanced [Si II]34.82 µm in AGN involves the extent to which

silicon is depleted onto dust grains. Heavy elements such as Si, Mg, and Fe may be returned

to the gas phase by dust destruction (e.g., sputtering) in regions subject to strong shocks

caused by stellar winds, starbursts, and AGN activity. So perhaps (gas phase) silicon lines

are stronger in active galaxies due to this effect.

If the strong [Si II]34.82 µm emission is due to the cooling of X-ray dominated regions,

it should be noted that relatively strong low-ionization line emission (e.g., [O I]6300Å and

[O I]63 µm) has previously been observed emanating from the large “partially-ionized re-

gions” surrounding AGN and infrared-bright galaxies (Veilleux & Osterbrock 1987; Armus,

Heckman, & Miley 1989; Veilleux 1991; Spinoglio & Malkan 1992; Osterbrock 1993; Dale et

al. 2004a). Hence, strong low-ionization line emission from AGN is not a new concept. We

take advantage of this concept to present new techniques for distinguishing between AGN

sources and star-forming regions. These techniques rely on an easily-detectable, prominent

cooling line of a low-ionization species associated with X-ray dominated regions, the dense

interstellar material illuminated by power-law radiation fields.

The regions of Figure 3 where Seyferts/LINERs/starbursts mix are quite large, though

the bottom panel perhaps shows a cleaner separation (less mixing) between Seyferts+LINERs

and star-forming regions; only the top left and bottom right extremes allow for a clean

separation between classifications. Short solid lines roughly perpendicular to the dotted

AGN/star-forming curves delineate three regions in the panels of Figure 3. The boundaries



– 12 –

are:

Region[ I − II] : log([O IV]25.89µm/[Ne II]12.81µm) = 10 log(EW[6.2µmPAH]) + 8.0(1)

Region[II − III] : log([O IV]25.89µm/[Ne II]12.81µm) = 1.9 log(EW[6.2µmPAH]) − 0.6(2)

Region[IV − V] : log([Si II]34.82µm/[Ne II]12.81µm) = 5.0 log(EW[6.2µmPAH]) + 4.8(3)

Region[V − VI] : log([Si II]34.82µm/[Ne II]12.81µm) = 1.7 log(EW[6.2µmPAH]) + 0.5(4)

Region[VII − VIII] : log([Si II]34.82µm/[S III]33.48µm) = 10 log(EW[6.2µmPAH]) + 9.7(5)

Region[VIII − IX] : log([Si II]34.82µm/[S III]33.48µm) = 1.1 log(EW[6.2µmPAH]) + 0.3(6)

The population statistics for these regions, provided in Table 4, show that Regions [I+IV+VII]

and [III+VI+IX] are respectively representative (at the >90% level) of Seyferts/LINERs and

star-forming systems. Regions [II+V+VIII], on the other hand, contain a mix of classifica-

tions and thus represent transition regions—either the source classifications in this region

are ambiguous or the region simply contains a more heterogeneous mixture of pure types.

Seyfert nuclei could shift toward the location of star-forming nuclei due to aperture effects—

though the same solid angles are used for extracting the line data, the range of distances

in the sample leads to a range in physical apertures. Conversely, some star-forming nuclei

exhibit relatively large line ratios and small PAH equivalent widths in Figure 3. Perhaps a

fraction of the star-forming regions contain significant numbers of Wolf Rayet stars, leading

to enhanced [O IV] emission (e.g., Schaerer & Stasiǹska 1999). And maybe a decreased PAH

equivalent width results from a relatively low ratio of photodissociation region to H II region

contributions (Laurent et al. 2000).

The dotted lines in Figure 3 represent a variable mix of an AGN nucleus and a “pure”

star-forming region. The anchors for these mixing models are the following:

EW[6.2µmPAH] ≈ 0.01µm, [O IV]25.89µm/[Ne II]12.81µm ≈ 0.4 100% AGN

EW[6.2µmPAH] ≈ 0.7µm, [O IV]25.89µm/[Ne II]12.81µm ≈ 0.01 100% H II

EW[6.2µmPAH] ≈ 0.01µm, [Si II]34.82µm/[Ne II]12.81µm ≈ 2 100% AGN

EW[6.2µmPAH] ≈ 0.7µm, [Si II]34.82µm/[Ne II]12.81µm ≈ 0.4 100% H II

EW[6.2µmPAH] ≈ 0.01µm, [Si II]34.82µm/[S III]33.48µm ≈ 3.5 100% AGN

EW[6.2µmPAH] ≈ 0.7µm, [Si II]34.82µm/[S III]33.48µm ≈ 0.2 100% H II

The dashed line in the upper panel of Figure 3 shows the approximate mixing model of

Genzel et al. (1998; their Figure 5), obtained after empirically deriving a relation between

their 7.7 µm PAH ‘strength’ (line-to-continuum ratio) and the 6.2 µm PAH equivalent width.

Many of the high ionization line data presented by Genzel et al. (1998) were upper limits,

so it is unsurprising that their original curve lies above our curve (though the discrepancy

may also lie in small number statistics).



– 13 –

5.2.2. Line Ratios of Different Ionization States of the Same Element

Figure 4 plots a ratio of doubly- to singly-ionized neon as a function of a ratio of

triply- to doubly-ionized sulfur (see also Verma et al. 2003). Many of the data points in

this plot are for Galactic H II regions (Vermeij et al. 2002; Giveon et al. 2002; Peeters

et al. 2002; see § 3.2). Clearly the neon excitation tracks the sulfur excitation. To first

order, there does not appear to be any sequence in the distribution according to source

classification (Seyfert, starburst, H II, etc). However, the low-metallicity H II regions from

the Magellanic Clouds are preferentially in the high excitation, upper righthand corner of

the diagram. Presumably the diminished line blanketing for low-metallicity sources results

in a harder radiation field and thus higher excitations (see Genzel & Cesarsky 2000; Mart́in-

Hernández et al. 2002; Madden et al. 2006). In addition, AGN sources show somewhat

lower [Ne III]15.6 µm/[Ne II]12.81 µm ratios than exhibited by star-forming sources, and the

locus of the AGN detections lie at slightly higher values of [S IV]10.51 µm/[S III]33.48 µm.

The dotted and solid curves show linear fits to the Seyfert and star-forming sources, and

respectively have slopes of 0.71[±0.12] and 0.75[±0.06].

5.2.3. A Neon, Sulfur, and Silicon Diagnostic

If the neon excitation is plotted as a function of [S III]33.48 µm/[Si II]34.82 µm (Fig-

ure 5), a more obvious separation of the star-forming and AGN-powered data points is

observed. Not only do the low-metallicity Magellanic Cloud regions exhibit a higher neon

excitation, nearly all of the “pure star-forming” nuclei and extra-nuclear regions show rela-

tively elevated ratios in [S III]33.48 µm/[Si II]34.82 µm (see also Figure 3). Note in addition

that many of the filled squares representing starbursting/star-forming nuclei are located be-

tween the H II regions and the AGN. Table 5 quantifies the source type fractions within each

of the four regions delineated by the lines drawn in Figure 5. The boundaries are defined by

curves with the same slope but differing offsets:

log([Ne III]15.56µm/[Ne II]12.81µm) = 8.4 log([S III]33.48µm/[Si II]34.82µm) + γ, (7)

where γ =[+3.3,+1.2,−2.5] for the lines demarcating Regions [I−II,II−III,III−IV]. The num-

bers provided in Table 5 can be used to determine the statistical reliability of a classification

for a galaxy randomly drawn from a mid-infrared line survey. For example, if a galaxy

appears in Region III or Region IV, it should be classified as a star-forming system with a

1σ confidence interval of 84-93% and 91-98%, respectively. Likewise, a galaxy residing in

Region I or Region II should be classified as AGN-powered with a 1σ confidence interval of

83-97% and 73-88%, respectively.



– 14 –

These results can be partially understood in the context of the cooling line physics

introduced above. The [Si II]34.82 µm line is a significant coolant of X-ray ionized regions

or dense photodissociation regions (Hollenbach & Tielens 1999), whereas the [S III]33.48 µm

line is a strong marker of H II regions. In other words, extra-nuclear regions and star-

forming nuclei will show strong signatures of the Strömgren sphere coolant [S III]33.48 µm,

while AGN and their associated X-ray dominated regions or dense photodissociation regions

will exhibit relatively strong [Si II]34.82 µm emission in analogy to the increased strength of

[O I] 6300Å emission in AGN (e.g., Veilleux & Osterbrock 1987). In addition, the fraction

of photodissociation regions falling within each beam will play a role in the line ratios. The

data for Magellanic Cloud and Galactic H II regions stem from smaller physical apertures

and thus are likely to have fractionally higher contributions from Strömgren spheres than

photodissociation regions.

Metallicity may be a factor as well. Since the central regions of galaxies typically

are more abundant in heavy metals (Pagel & Edmunds 1981; McCall 1982; Vila-Costas &

Edmunds 1992; Pilyugin & Ferrini 1998; Henry & Worthey 1999), and as explained above

a lower metallicity can lead to harder radiation fields and thus enhanced high-ionization-to-

low-ionization line ratios, it is possible that this AGN→H II nucleus→H II region sequencing

along the [S III]33.48 µm/[Si II]34.82 µm axis is affected by metallicity. However, the lower

metallicity Magellanic Cloud data are not substantially to the right of the Galactic H II

region data, so the effect is not solely due to metallicity. Alternatively, perhaps some of the

star-forming nuclei have contributions from undetected weak AGN and thus are not “pure”

star-forming nuclei, resulting in a location for star-forming nuclei on this diagram between

AGN and H II regions.

5.2.4. Density Diagnostics

The average line ratio of [S III]18.71 µm-to-[S III]33.48 µm for the SINGS sample is

0.82 with a 1σ dispersion of 0.27. This ratio implies an interstellar electron density of

〈ne〉 ∼ 400
+240
−290 cm

−3 for the ∼ 23′′×15′′ nuclear and extranuclear regions of SINGS galaxies.

The average density is calculated using electron collision strengths from Tayal & Gupta

(1999) and excluding the effects of differential extinction at these mid-infrared wavelengths

(which is shown in Section 5.1 to be relatively small at optical wavelengths). This density on

∼kiloparsec scales is typical of starburst/LINER/Seyfert galaxies (Kewley et al. 2001), but

lower than the ∼ 103 − 104 cm−3 found for high surface brightness H II regions using small

apertures uncontaminated by the surrounding neutral interstellar medium and lower density

H II regions (e.g., Wang et al. 2004). A visual way to explore this doubly-ionized sulfur



– 15 –

line ratio is portrayed in Figure 6 using the aperture-matched 24 µm flux as a normalization

for the line fluxes. The set of dashed lines represent different interstellar electron densities.

The correlation in Figure 6 extends over two orders of magnitude in the line-to-continuum

ratios and encompasses both AGN-dominated and star-formation-dominated sources; a non-

parametric ranking analysis indicates a global correlation at the 7σ level. Linear fits to

the two separate star-forming and Seyfert populations emphasize that, though the trends

for the two populations are similar, the nuclei with Seyfert characteristics (dotted line;

slope 0.91±0.22) differ along the diagonal from the starbursting nuclei and the H II regions

(solid line; slope 0.85±0.05). Star-forming sources show more pronounced [S III]33.48 µm-

to-continuum ratios compared to Seyferts and LINERs, consistent with the notion that

[S III]33.48 µm is an important coolant of H II regions. In addition, we are seeing the effects of

continuum dilution in the line-to-continuum ratio for Seyferts and LINERs, sources for which

hot dust emission is pronounced in the mid-infrared and thus the line-to-continuum ratios

are suppressed (e.g., Laurent et al. 2000; Dale et al. 2001; Xu et al. 2001; Siebenmorgen et

al. 2004). Finally, similar to what is seen in Figures 3 and 5, but perhaps not as prominently,

the data in Figure 6 suggest that the H II regions (open squares) occupy a different portion

of the diagram than star-forming nuclei (filled squares). H II regions have higher line-to-

continuum ratios than star-forming nuclei, which in turn have higher ratios than Seyferts

and LINERs.

6. Summary

We have presented mid-infrared diagnostic diagrams for a large portion of the SINGS

sample supplemented by archival ISO and Spitzer data. A portion of our work solidifies

and extends previous ISO-based mid-infrared work to lower luminosity normal galaxy nuclei

and H II regions using the Spitzer Space Telescope. We also present new diagnostics that

effectively constrain a galaxy’s dominant power source. The power of the diagnostic diagrams

of Genzel et al. (1998; see also Peeters, Spoon, & Tielens 2004) for distinguishing between

AGN and star-forming sources in dusty ULIRGs is that mid-infrared lines and PAH features

are much less affected by extinction than their optical counterparts in a traditional diagnostic

diagram. Unlike diagrams put forth by Genzel et al., which rely on detecting relatively weak

high-ionization lines like [O IV] 25.89 µm and [Ne V] 14.32 µm, we provide a new diagnostic

that utilizes a strong low-ionization line. The advantage of using a line ratio like [Si II]/[Ne II]

is that singly-ionized silicon and neon respectively have ionization potentials of only 8.15 and

12.8 eV, so they can both be observed over a large range of physical conditions. This is similar

in concept to previous efforts that have taken advantage of [O I] lines (e.g., at 6300 Å or

63 µm) that are coolants of the X-ray dominated regions (or dense photodissociation regions)



– 16 –

surrounding AGN. In plots of [O IV]/[Ne II], [Si II]/[Ne II], and [Si II]/[S III] vs. 6.2 µm

PAH equivalent width, we identify regions where >90% of the sources are Seyfert or LINER.

Likewise, additional regions in all three plots show populations comprised of more than 90%

H II regions or star-forming nuclei.

Another useful mid-infrared diagnostic is [Ne III] 15.56 µm/[Ne II] 12.81 µm vs. [S III] 33.48 µm/[Si II] 34.82

This plot tracks the excitation power of the radiation field on one axis, while the other axis

is a relative measure of the cooling of H II regions and X-ray dominated regions (or dense

photodissociation regions). Similar to what is found for the diagnostics mentioned above,

both starbursting nuclei and extranuclear regions stand apart from nuclei that are pow-

ered by accretion-powered disks. Moreover, compared to starbursting nuclei, extranuclear

regions typically separate even further from Seyfert nuclei, especially for low-metallicity en-

vironments. Presumably this extranuclear←→nuclear separation occurs since extranuclear

regions are cleaner representatives of H II regions than starburst nuclei, because their stellar

populations and interstellar medium structure are less complex. Extranuclear regions more

likely contain younger stellar populations since they trace a single burst, as opposed to the

average of multiple star formation episodes for nuclei (e.g., Dale et al. 2004b). Finally, we

note that it is difficult to clearly distinguish between pure Seyfert and pure LINER sources

using these diagnostics.

The line ratio [S III] 18.71 µm/[S III] 33.48 µm yields an average interstellar electron

density of 〈ne〉 ∼ 400
+240
−290 cm

−3 for the ∼ 23′′×15′′ nuclear and extranuclear regions of

SINGS galaxies. This density is much closer (in log space) to the theoretical low density

limit of Tayal & Gupta (1999) than their high density limit, and in fact the data for several

sources are consistent with the low density limiting value. In addition to the interstellar

gas densities being unremarkable, there are no SINGS sources sufficiently obscured by dust

such that optical and infrared diagnostics provide obviously discrepant classifications of the

energy source. This is not surprising, however, since our nuclei exhibit modest extinctions,

〈AV 〉 ∼ 1.0 mag, and normal star-forming galaxy nuclei in general show 0 . AV . 3 mag

(Ho, Filippenko, & Sargent 1997b). This relative transparency means the SINGS sample

is not ideally suited for a detailed comparison of the relative merits of optical and infrared

classifications. However, the diverse SINGS sample of nuclear and extranuclear regions

provides an enormous range of physical parameters, and this has proved critical in developing

the mid-infrared diagnostics in this paper.

Mike Brotherton provided helpful comments. Support for this work, part of the Spitzer

Space Telescope Legacy Science Program, was provided by NASA through Contract Number

1224769 issued by the Jet Propulsion Laboratory, California Institute of Technology under

NASA contract 1407. This research has made use of the NASA/IPAC Extragalactic Database



– 17 –

which is operated by JPL/Caltech, under contract with NASA. This publication makes use

of data products from the Two Micron All Sky Survey, which is a joint project of the

University of Massachusetts and the Infrared Processing and Analysis Center/California

Institute of Technology, funded by the National Aeronautics and Space Administration and

the National Science Foundation.

REFERENCES

Armus, L., Heckman, T.M., & Miley, G.K. 1989, ApJ, 347, 727

Armus, L. et al. 2004, ApJS, 154, L178

Baldwin, J.A., Phillips, M.M., & Terlevich, R. 1981, PASP, 93, 5

Calzetti, D. et al. 2005, ApJ, 633, 871

Cesarsky, D., Lequeux, J., Abergel, A., Perault, M., Palazzi, E., Madden, S., & Tran, D.

1996, A&A, 315, L309

Dale, D.A., Helou, G., Contursi, A., Silbermann, N.A., & Kolhatkar, S. 2001, ApJ, 549, 215

Dale, D.A., Helou, G., Brauher, J.R., Cutri, R.M., Malhotra, S., & Beichman, C.A. 2004a,

ApJ, 604, 565

Dale, D.A. et al. 2004b, ApJ, 601, 813

Dale, D.A. et al. 2005, ApJ, 633, 857

Draine, B.T. & Li, A. 2001, ApJ, 551, 807

Engelbracht, C.W., Gordon, K.D., Rieke, G.H., Werner, M.W., Dale, D.A., & Latter, W.B.

2005, ApJ, 628, L29

Farrah, D., Afonso, J., Efstathiou, A., Rowan-Robinson, M., Fox, M., & Clements, D. 2003,

MNRAS, 343, 585

Förster Schrieber, N.M., Roussel, H., Sauvage, M., & Charmandaris, V. 2004, A&A, 419,

501

Galliano, F., Madden, S.C., Jones, A.P., Wilson, C.D., & Bernard, J.-P. 2005, A&A, 434,

867

Genzel, R. et al. 1998, ApJ, 498, 579



– 18 –

Genzel, R. & Cesarsky, C.J. 2000, ARA&A, 38, 761

Giveon, U., Sternberg, A., Lutz, D., Feuchtgruber, H., & Pauldrach, A.W.A. 2002, ApJ,

566, 880

González Delgado, R.M., Cid Fernandes, R., Pérez, E., Martins, L.P., Storchi-Bergmann,

T., Schmitt, H., Heckman, T., & Leitherer, C. 2004, ApJ, 605, 127

Haas, M., Siebenmorgen, R., Schulz, B., Krügel, E., & Chini, R. 2005, A&A, 442, L39

Habing, H.J. 1968, Bull. Astron. Inst. Netherlands, 19, 421

Henry, R.B.C. & Worthey, G. 1999, PASP, 111, 919

Ho, L., Filippenko, A.V., & Sargent, W.L. 1993, ApJ, 417, 63

Ho, L., Filippenko, A.V., & Sargent, W.L. 1997, ApJS, 112, 315

Ho, L., Filippenko, A.V., & Sargent, W.L. 1997, ApJ, 487, 579

Hollenbach, D. & Tielens, A.G.G.M. 1999, Rev. Mod. Phys., 71, 173

Houck, J.R. et al. 2004a, ApJS, 154, L18

Houck, J.R. et al. 2004b, ApJS, 154, L211

Kauffmann, G. et al. 2003, MNRAS, 346, 1055

Kaufman, M.J., Wolfire, M.G., & Hollenbach, D.J. 2006, ApJ, in press

Kennicutt, R.C. et al. 2003, PASP, 115, 928

Kewley, L.J., Dopita, M.A., Sutherland, R.S., Heisler, C.A., Trevena, J. 2001, ApJ, 556, 121

Laurent, O., Mirabel, I.F., Charmandaris, V., Gallais, P., Madden, S.C., Sauvage, M., Vi-

groux, L., & Cesarsky, C. 2000, A&A, 359, 887

Li, A. & Draine, B.T. 2001, ApJ, 554, 778

Lutz, D., Veilleux, S., & Genzel, R. 1999, ApJ, 517, L13

Madden, S.C., Galliano, F., Jones, A.P., & Sauvage, M. 2006, A&A, 446, 877

Maloney, P.R., Hollenbach, D.J., & Tielens, A.G.G.M. 1996, ApJ, 466, 561



– 19 –

Mart́in-Hernández, N.L., Vermeij, R., Tielens, A.G.G.M., van der Hulst, J.M., & Peeters, E.

2002, A&A, 389, 286

McCall, M.L. 1982, Ph.D. thesis, University of Texas at Austin

Meijerink, R. & Spaans, M. 2005, A&A, 436, 397

Moustakas, J. & Kennicutt, R.C. 2006, ApJ, in press (astro-ph/0511729)

Osterbrock, D.E. 1993, ApJ, 404, 551

Pagel, B.E.J. & Edmunds,M.G. 1981, ARA&A, 19, 77

Peeters, E. et al. 2002, A&A, 381, 571

Peeters, E., Spoon, H.W.W., & Tielens, A.G.G.M. 2004, ApJ, 613, 986

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

Pilyugin, L.S. & Ferrini, F. 1998, A&A, 336, 103

Rieke, G.H. et al. 2004, ApJS, 154, 25

Roussel, H., Helou, G., Beck, R., Condon, J.J., Bosma, A., Matthews, K., & Jarrett, T.H.

2003, ApJ, 593, 733

Schaerer, D. & Stasińska, G. 1999, A&A, 345, L17

Siebenmorgen, R., Freudling, W., Krügel, E., & Haas, M., 2004, A&A, 421, 129

Smith, J.D.T. et al. 2004, ApJS, 154, L199

Spinoglio, L. & Malkan, M.A. 1992, ApJ, 399, 504

Sturm, E., Lutz, D., Tran, D., Feuchtgruber, H., Genzel, R., Kunze, D., Moorwood, A.F.M.,

& Thornley, M.D. 2000, A&A, 358, 481

Sturm, E., Lutz, D., Verma, A., Netzer, H., Sternberg, A., Moorwood, A.F.M., Oliva, E., &

Genzel, R. 2002, A&A, 393, 821

Taniguchi, Y., Yoshino, A., Ohyama, Y., & Nishiura, S. 1999, ApJ, 514, 660

Tayal, S.S., & Gupta, G.P. 1999, ApJS, 526, 544

Thornley, M.D., Förster Schreiber, N.M., Lutz, D., Genzel, R., Spoon, H.W.W., Kunze, D.,

& Sternberg, A. 2000, ApJ, 539, 641



– 20 –

Thuan, T.X., Sauvage, M., & Madden, S. 1999, ApJ, 516, 783

Tran, Q.D. et al. 2001, ApJ, 552, 527

Veilleux, S. & Osterbrock, D.E. 1987, ApJS, 63, 295

Veilleux, S. 1991, ApJ, 369, 331

Veilleux, S., Sanders, D.B., & Kim, D.-C. 1999, ApJ, 522, 139

Verma, A., Lutz, D., Sturm, E., Sternberg, A., Genzel, R., & Vacca, W. 2003, A&A, 403,

829

Vermeij, R. & van der Hulst, J.M. 2002, A&A, 391, 1081

Vermeij, R., Damour, F., van der Hulst, J.M., & Baluteau, J.-P. 2002, A&A, 390, 649

Verstraete, L., Pech, C., Moutou, C., Sellgren, K., Wright, C.M., Giard, M., Léger, A.,

Timmerman, R., & Drapatz, S. 2001, A&A, 372, 981

Vila-Costas, M.B. & Edmunds, M.G. 1992, MNRAS, 259, 121

Voit, G.M. 1992, ApJ, 399, 495

Wang, W., Liu, X.-W., Zhang, Y., & Barlow, M.J. 2004, A&A, 427, 873

Weedman, D.W., Hao, L., Higdon, S.J.U., Devost, D., Wu, Y., Charmandaris, V., Brandl,

B., Bass, E., & Houck, J.R. 2005, ApJ, 633, 706

Wu, Y., Charmandaris, V., Hao, H., Brandl, B.R., Bernard-Salas, J., Spoon, H.W.W., &

Houck, J.R. 2006, ApJ, in press

Xu, C., Lonsdale, C.J., Shupe, D.L., O’Linger, J., & Masci, F. 2001, ApJ, 562, 179

This preprint was prepared with the AAS LATEX macros v5.2.



– 21 –

Table 1. Nuclear Emission Line Fluxes and 6.2 µm PAH Feature Equivalent Widths

Species PAH [SIV] [NeII] [NeIII] [SIII] [OIV] [SIII] [SiII]

Wavelength 6.2µm 10.51µm 12.81µm 15.56µm 18.71µm 25.89µm 33.48µm 34.82µm

ionization (eV) 34.8 21.6 41.0 23.3 54.9 23.3 8.2

NGC 0337 0.56±0.04 · · · 17.4±1.6 8.1±0.5 13.3±0.6 0.5±0.2 17.9±0.5 21.1±0.6

NGC 0584 · · · · · · · · · 2.4±0.7 1.1±0.5 <0.8 · · · · · ·

NGC 0628 0.39±0.08 · · · 6.4±1.3 · · · 2.5±0.3 <0.7 9.1±0.7 5.8±0.8

NGC 1097 0.42±0.01 · · · 328±4 29.0±0.6 89.6±0.5 <5.3 103±3 278±4

NGC 1266 0.29±0.02 · · · 28.0±1.0 10.8±0.6 1.3±0.5 <2.8 2.7±0.6 17.3±3.7

NGC 1377 · · · · · · 3.5±0.5 · · · · · · <3.4 · · · · · ·

NGC 1404 · · · · · · 1.9±0.8 1.3±0.6 · · · <0.7 · · · · · ·

NGC 1566 0.20±0.01 1.1±0.6 16.6±0.8 9.8±0.4 6.8±0.6 6.7±0.5 6.8±0.7 15.2±0.8

NGC 1705 0.42±0.26 3.2±0.8 1.1±0.3 7.0±0.4 3.4±0.4 1.2±0.4 4.9±0.5 3.6±0.5

NGC 2798 0.52±0.02 5.9±1.8 208±2 34.4±0.5 81.6±0.8 9.6±1.8 70.1±2.6 102±4

NGC 2841 0.03±0.03 · · · 5.5±0.5 6.6±0.4 2.7±0.5 0.9±0.1 3.6±0.4 9.7±2.3

NGC 2915 · · · 1.8±0.9 3.1±0.4 11.1±0.4 4.6±0.4 0.4±0.1 6.1±0.8 4.9±0.4

NGC 2976 0.35±0.04 · · · 7.7±0.7 2.7±0.3 6.3±0.4 0.4±0.1 9.2±0.3 8.5±0.5

NGC 3049 0.62±0.04 1.1±0.6 36.4±0.8 6.1±0.4 27.3±0.6 <0.7 24.4±0.4 21.5±0.5

NGC 3190 0.09±0.01 · · · 7.4±0.6 5.6±0.5 1.8±0.6 0.7±0.2 2.0±0.2 6.9±1.7

NGC 3184 0.26±0.04 · · · 17.8±0.5 2.1±0.5 7.8±0.5 <0.7 7.2±0.3 12.0±0.4

NGC 3198 0.31±0.03 · · · 13.9±0.4 0.8±0.3 4.8±0.5 <0.9 5.5±0.5 11.1±2.2

NGC 3265 0.57±0.03 1.5±0.7 28.4±0.3 4.9±0.2 11.7±0.3 <1.2 13.1±0.6 16.9±0.7

Markarian 33 0.51±0.02 17.1±0.8 56.7±1.1 45.3±0.4 52.8±0.7 1.0±0.7 32.8±0.8 25.1±0.9

NGC 3351 0.46±0.01 · · · 165±2 14.0±0.4 60.0±0.5 <2.3 60.5±1.7 128±2

NGC 3521 0.38±0.04 · · · 12.9±0.8 7.9±0.5 3.3±0.7 0.8±0.4 3.6±0.8 16.7±2.3

NGC 3627 0.25±0.01 · · · 21.9±0.9 8.9±0.5 4.7±0.4 1.3±0.7 7.0±1.1 20.2±3.3

NGC 3773 0.65±0.08 3.9±0.7 16.5±0.5 16.1±0.5 14.3±0.6 <0.8 17.0±0.6 13.1±0.7

NGC 3938 0.13±0.03 · · · 5.7±0.5 1.0±0.5 1.1±0.5 0.3±0.1 2.2±0.1 5.9±0.2

NGC 4125 · · · · · · 2.5±0.5 3.4±0.5 · · · <0.9 0.8±0.6 5.4±1.1

NGC 4254 0.54±0.03 · · · 51.7±1.1 6.4±0.4 10.0±0.5 1.3±0.2 15.6±0.6 48.9±1.1

NGC 4321 0.49±0.02 · · · 110±2 12.3±0.4 23.1±0.3 <1.4 25.5±0.8 84.8±0.9

NGC 4536 0.54±0.01 7.5±2.6 307±3 48.9±0.7 130±1 <3.9 160±3 220±3

NGC 4552 · · · · · · 1.8±0.5 4.8±0.8 0.9±0.4 <0.9 0.7±0.3 2.0±0.7

NGC 4569 0.26±0.02 1.0±0.8 32.6±1.2 15.8±0.5 8.8±0.7 2.4±0.6 7.7±0.6 31.5±0.9

NGC 4579 0.09±0.02 · · · 22.8±0.7 12.5±0.5 4.0±0.7 2.2±0.3 3.0±0.4 19.5±0.4

NGC 4725 0.07±0.04 0.8±0.5 1.8±0.5 3.0±0.4 0.3±0.2 1.7±0.2 1.3±0.2 4.6±1.5

NGC 4736 0.16±0.01 · · · 13.7±1.1 14.3±0.5 6.1±0.9 3.8±0.9 9.2±1.0 22.3±1.8

NGC 4826 0.28±0.01 1.8±0.5 99.0±1.6 23.4±0.6 41.8±0.4 2.5±1.1 56.8±1.0 105±1

NGC 5194 0.24±0.02 2.9±0.8 66.2±0.8 36.2±0.6 13.1±0.6 14.9±1.6 18.3±0.4 64.8±0.8

NGC 5408a · · · · · · · · · 1.7±0.4 · · · <0.9 1.5±0.5 1.8±0.5

NGC 5713 0.56±0.02 1.7±0.7 123±2 17.3±0.4 47.5±0.5 2.0±1.0 57.2±1.2 90.3±1.2

NGC 5866 0.07±0.01 · · · 7.5±0.9 5.1±0.3 1.1±0.4 0.7±0.1 4.0±0.3 9.7±0.4

IC 4710 · · · 3.6±0.8 · · · 4.3±1.6 3.6±0.5 <1.3 4.2±0.6 1.4±0.4

NGC 7331 0.09±0.02 0.5±0.2 18.8±0.6 10.3±0.3 6.0±0.4 2.4±0.6 12.2±0.3 29.0±6.3

NGC 7552 0.45±0.01 6.5±1.4 573±31 56.8±1.4 211±1 <15.3 165±5 260±8

NGC 7793 0.36±0.04 · · · 10.3±0.6 2.9±0.5 8.7±0.6 <0.5 9.4±0.3 9.5±0.5

Note. — Fluxes and their (statistical) uncertainties are averaged over ∼23′′×15′′ and listed in units

of 10−9 W m−2 sr−1. Calibration uncertainties are an additional ∼30%. The 6.2 µm PAH feature

equivalent width is given in units of microns.

Note. — Eight galaxies from the SINGS Data Release Three are not listed in this table. No Short-Low,

Short-High, or Long-High data were taken for the optical centers of Holmberg II, IC 2574, DDO 154

and NGC 6822. DDO 053, Holmberg IX, DDO 165, and NGC 5398 (Tololo 89)a were non-detections.

aThe infrared emission peaks outside of the field of view of the spectral maps.



– 22 –

Table 2. Extranuclear Emission Line Fluxes and 6.2 µm PAH Feature Equivalent Widths

Species PAH [SIV] [NeII] [NeIII] [SIII] [OIV] [SIII] [SiII]

Wavelength 6.2µm 10.51µm 12.81µm 15.56µm 18.71µm 25.89µm 33.48µm 34.82µm

ionization (eV) 34.8 21.6 41.0 23.3 54.9 23.3 8.2

NGC 5194 CCM107 0.47±0.02 · · · 32.0±0.6 2.3±0.6 10.6±0.5 1.7±0.3 18.3±0.5 34.1±0.5

NGC 5194 CCM072 0.61±0.02 1.7±0.5 71.9±1.0 3.8±0.4 29.4±0.6 1.4±0.8 36.2±0.5 49.2±0.7

NGC 5194 CCM071 0.66±0.02 · · · 44.3±0.7 5.6±0.5 16.6±0.7 1.3±0.4 25.0±0.6 41.2±0.5

NGC 5194 CCM001 0.69±0.02 · · · 16.4±1.1 3.3±0.5 8.2±0.3 0.6±0.2 13.7±0.5 21.7±0.9

NGC 5194 CCM010 0.69±0.03 · · · 39.2±0.7 6.1±0.4 19.8±0.5 0.5±0.3 29.0±0.5 37.7±0.7

NGC 5194 CCM071A 0.52±0.02 · · · 21.3±0.6 5.7±0.4 11.0±0.5 <0.6 14.4±2.1 12.6±0.4

NGC 3031 HK230 0.44±0.03 · · · 4.1±0.5 0.7±0.5 2.7±0.4 <0.7 3.6±0.5 4.0±0.4

NGC 3031 HK343 0.38±0.03 · · · 7.5±3.1 3.6±0.4 7.0±0.9 <0.8 9.8±0.8 7.0±0.6

NGC 3031 HK453 0.65±0.05 · · · 9.0±0.6 3.4±0.4 8.8±0.9 <0.8 11.7±2.0 9.3±0.6

NGC 3031 HK268 0.55±0.03 1.2±0.7 15.1±0.8 6.0±0.4 13.4±0.7 <0.8 17.3±0.7 13.0±0.6

NGC 3031 HK652 0.68±0.06 · · · 10.3±0.7 2.7±0.5 9.2±1.1 <0.8 10.7±0.8 11.8±0.6

NGC 3031 HK741 0.75±0.05 · · · 8.9±0.8 1.6±0.4 7.3±0.4 <0.9 7.6±0.7 6.5±0.7

NGC 3031 Munch1 2.98±2.29 · · · · · · 1.4±0.2 · · · <1.0 · · · · · ·

NGC 6946 H4 0.65±0.02 3.6±0.7 16.6±0.6 17.5±0.5 14.7±0.4 1.1±0.2 14.2±0.6 14.6±0.5

NGC 6946 HK3 0.63±0.03 8.8±1.3 28.2±0.4 31.9±0.4 30.6±0.4 0.9±0.2 39.0±0.5 25.1±0.5

NGC 6946 H288 0.81±0.04 · · · 18.1±0.4 8.5±0.4 14.0±0.5 <0.6 19.3±0.3 11.8±0.2

NGC 6946 H40 0.84±0.03 1.9±0.8 18.8±0.4 8.5±0.4 15.0±0.4 0.2±0.1 17.3±0.3 15.2±0.5

NGC 6946 H28 1.05±0.45 · · · 9.4±0.5 2.9±0.3 6.8±0.5 <0.5 9.0±0.3 8.2±0.4

NGC 0628 H292 0.55±0.02 2.1±0.9 25.6±0.7 2.7±0.4 20.1±0.6 <1.0 27.4±0.5 13.1±0.7

NGC 0628 H572 0.69±0.03 · · · 12.5±0.7 2.6±0.5 8.1±0.4 <0.9 16.3±0.5 8.2±0.6

NGC 0628 H627 0.60±0.03 3.8±1.3 11.5±0.7 9.5±0.5 10.6±0.4 <0.9 17.8±0.9 9.6±0.5

NGC 0628 H013 0.52±0.07 · · · 4.1±0.7 2.2±0.3 4.9±0.4 <0.7 11.3±1.3 2.9±0.4

HolmbergII HSK45 · · · · · · · · · 3.4±0.5 3.2±0.4 <0.9 2.9±0.7 2.8±0.5

HolmbergII HSK67 · · · · · · · · · 0.7±0.2 · · · <0.7 0.6±0.2 0.9±0.3

HolmbergII HSK70 · · · · · · · · · 1.1±0.5 0.6±0.2 <0.9 0.4±0.2 1.7±0.5

HolmbergII HSK07 · · · · · · · · · 1.1±0.6 0.8±0.5 <1.0 · · · · · ·

Note. — Fluxes and their (statistical) uncertainties are averaged over ∼23′′×15′′ and listed in units of

10−9 W m−2 sr−1. Calibration uncertainties are an additional ∼30%. The 6.2 µm PAH feature equivalent

width is given in units of microns.



– 23 –

Table 3. Archival Sources

Object Type Reference

LMC N160A1 LMC HII Vermeij et al. (2002)

LMC N160A2 LMC HII Vermeij et al. (2002)

LMC N159-5 LMC HII Vermeij et al. (2002)

LMC N157B LMC HII Vermeij et al. (2002)

LMC N4A LMC HII Vermeij et al. (2002)

LMC N11A LMC HII Vermeij et al. (2002)

LMC N83B LMC HII Vermeij et al. (2002)

LMC 30Dor1 LMC HII Vermeij et al. (2002)

LMC 30Dor2 LMC HII Vermeij et al. (2002)

LMC 30Dor3 LMC HII Vermeij et al. (2002)

LMC 30Dor4 LMC HII Vermeij et al. (2002)

SMC N88A SMC HII Vermeij et al. (2002)

SMC N66 SMC HII Vermeij et al. (2002)

SMC N81 SMC HII Vermeij et al. (2002)

NGC 0253 HII nucleus Verma et al. (2003)

IC 342 HII nucleus Verma et al. (2003)

II Zw 40 HII nucleus Verma et al. (2003)

NGC 3034 HII nucleus Verma et al. (2003)

NGC 3256 HII nucleus Verma et al. (2003)

NGC 3690A HII nucleus Verma et al. (2003)

NGC 3690B HII nucleus Verma et al. (2003)

NGC 4038 HII nucleus Verma et al. (2003)

NGC 4945 HII/Seyfert Verma et al. (2003)

NGC 5236 HII nucleus Verma et al. (2003)

NGC 5253 HII nucleus Verma et al. (2003)

NGC 7552 LINER/HII Verma et al. (2003)

WB89 380 A Galactic HII Giveon et al. (2002)

WB89 380 B Galactic HII Giveon et al. (2002)

WB89 399 Galactic HII Giveon et al. (2002)

Note. — The full version would appear as an

electronic table in the online Journal.



– 24 –

Table 4. Classifications by Region in Figure 3

Region Number Seyfert LINER HII extranuclear

of nuclei +HII regions

Sources (%) (%) (%) (%)

I 15 73 20 7 0

II 31 42 23 19 16

III 31 0 10 32 58

IV 12 67 25 8 0

V 39 33 18 21 28

VI 34 0 9 26 65

VII 12 67 25 8 0

VIII 45 31 24 27 18

IX 31 0 0 16 84

Note. — “extranuclear + HII regions” implies SINGS ex-

tranuclear/HII regions in addition to Milky Way and Magel-

lanic Cloud HII regions.

Table 5. Classifications by Region in Figure 5

Region Number Seyfert LINER HII extranuclear

of nuclei +HII regions

Detections (%) (%) (%) (%)

I+II 38 61 32 8 0

III+IV 79 0 3 30 67

I 16 69 31 0 0

II 22 55 32 14 0

III 47 0 4 51 45

IV 32 0 0 0 100

Note. — “extranuclear + HII regions” implies SINGS extranu-

clear/HII regions in addition to Milky Way and Magellanic

Cloud HII regions.



– 25 –

Fig. 1.— Examples of low- and high-resolution spectra for four different types of envi-

ronments found within the SINGS sample. The IRS modules span wavelengths ∼5-7 µm

(Short-Low2), ∼7-14 µm (Short-Low1), ∼14-21 µm (Long-Low2), ∼21-40 µm (Long-Low1),

∼10-19 µm (Short-High), and ∼19-37 µm (Long-High). Each spectral segment has been ex-

tracted from a ∼23′′×15′′ region. The high-resolution data are scaled downwards by 1.0 dex

for clarity. The variable offset between the low- and high-resolution data is largely due to

the lack of sky subtraction for the high-resolution data.



– 26 –

Fig. 2.— A traditional diagnostic diagram is displayed for the SINGS nuclei using optical

data and 20′′×20′′ apertures (uncorrected for reddening). Filled (open) circles mark galaxies

for which the literature indicates a Seyfert (LINER) nucleus. The dotted lines delineate

typical starburst/Seyfert/LINER boundaries: [O III]5007/Hβ ∼ 5 and [N II]6583/Hα ∼ 0.6

(e.g., Armus, Heckman, & Miley 1989). The long-dashed and short-dashed curves respec-

tively trace the starburst/AGN boundaries of Kewley et al. (2001; theoretical) and Kauff-

mann et al. (2003; empirical). Error bars represent 1σ uncertainties.



– 27 –

Fig. 3.— The ratios of mid-infrared forbidden lines as a function of the 6.2 µm PAH fea-

ture equivalent width. SINGS data are displayed in red with 1σ error bars based on the

statistical uncertainties; archival data without error bars are indicated with black symbols

and described in § 3.2. The dotted lines are linear mixing models of a “pure” AGN and a

“pure” star-forming source (see text). The dashed line in the top panel is a similar mixing

model first presented by Genzel et al. (1998). The solid lines and Roman numerals delineate

regions distinguished by Seyferts, LINERs, star formation, etc. (see Table 4 and § 5.2.1).



– 28 –

Fig. 4.— A diagnostic diagram involving ratios of neon and sulfur lines at different ionization

levels is displayed (see, for example, Verma et al. 2003). The data are displayed as described

in Figure 3. The solid line is a linear fit to the detections of star-forming nuclei and H II

regions, while the dotted line is linear fit to the Seyfert detections.



– 29 –

Fig. 5.— A neon, sulfur, and silicon diagnostic diagram involving ratios of lines at different

ionizations is displayed. The data are displayed as described in Figure 3. The lines and

Roman numerals delineate regions distinguished by Seyferts, LINERs, star formation, etc.

(see Table 5 and § 5.2.3).



– 30 –

Fig. 6.— The correlation between two transitions of doubly-ionized sulfur is displayed,

normalized by the flux at 24 µm. This plot includes only SINGS data and 1σ error bars.

The solid line is a linear fit to the detections of star-forming nuclei and H II regions, while

the dotted line is a linear fit to the Seyfert detections. The set of dashed lines represent

different constant interstellar electron densities. Most of the SINGS data are bounded by the

low and high density limiting values, and several are consistent with the low density limiting

value of [S III]18.71 µm/[S III]33.48 µm=0.43 at 0.1 cm−3. The SINGS data typically exhibit

ne ∼ 400 cm
−3.