The Astrophysical Journal, 646:161–173, 2006 July 20 A # 2006. The American Astronomical Society. All rights reserved. Printed in U.S.A. MID-INFRARED SPECTRAL DIAGNOSTICS OF NUCLEAR AND EXTRANUCLEAR REGIONS IN NEARBY GALAXIES D. A. Dale,1 J. D. T. Smith,2 L. Armus,3 B. A. Buckalew,3 G. Helou,3 R. C. Kennicutt Jr.,2,4 J. Moustakas,2 H. Roussel,3,5 K. Sheth,3 G. J. Bendo,2,6 D. Calzetti,7 B. T. Draine,8 C. W. Engelbracht,2 K. D. Gordon,2 D. J. Hollenbach,9 T. H. Jarrett,3 L. J. Kewley,10 C. Leitherer,7 A. Li,11 S. Malhotra,12 E. J. Murphy,13 and F. Walter5 Received 2006 February 24; accepted 2006 March 31 ABSTRACT Mid-infrared diagnostics are presented for a large portion of the Spitzer Infrared Nearby Galaxies Survey (SINGS) sample plus archival data from ISO and Spitzer. The SINGS data set 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 formation rates. Our main result is that these mid-infrared diagnostics effectively constrain a target’s dominant power source. The combination of a high- ionization line index and PAH strength serves as an efficient discriminant between AGNs and star-forming nuclei, confirming progress made with ISO spectroscopy on starbursting and ultraluminous infrared galaxies. The sensitivity of Spitzer 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 areas 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 AGNs, especially for low-metallicity extranuclear environments. In addition, instead of the traditional mid- infrared approach to differentiating between AGNs 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 m can alternatively be used as excellent discriminants. Finally, the typical target in this sample shows relatively modest À3 interstellar electron density (400 cm ) 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: galaxies: active — galaxies: nuclei — H ii regions — infrared: galaxies — infrared: ISM Online material: color figures, machine-readable table 1. INTRODUCTION [O iii] k5007, [O i] k6300, H k6563, [N ii] k6584, and [S ii] kk6716, 6731 (e.g., Baldwin et al. 1981; Veilleux & Osterbrock The goal of this study is to explore whether mid-infrared di- 1987; Ho et al. 1997b; Kewley et al. 2001; Kauffmann et al. agnostics developed for luminous/ultraluminous infrared gal- 2003). A plot of [O iii]/H versus [N ii]/H , for example, will axies (LIRGs/ULIRGs) and bright Galactic H ii regions can be typically separate Seyfert galaxies, LINERs, and starburst nu- improved on and extended to the nuclear and extranuclear re- clei. Since nuclei are often heavily enshrouded by dust, espe- gions within normal and infrared-faint galaxies. A traditional cially in LIRGs and ULIRGs, an important limitation to galaxy method for characterizing a galaxy’s nuclear power source uses optical diagnostics is the effect of extinction. In anticipation of ratios of optical emission lines such as [O ii] k3727, H k4861, the data stream from space-based infrared platforms, early theo- retical work with photoionization models showed that infrared 1 Department of Physics and Astronomy, University of Wyoming, Laramie, ionic fine-structure line ratios could profitably enable astrono- WY 82071; [email protected]. 2 mers to approach galaxy classification from a new perspective Steward Observatory, University of Arizona, 933 North Cherry Avenue, (e.g., Voit 1992; Spinoglio & Malkan 1992). The advent of sen- Tucson, AZ 85721. 3 California Institute of Technology, MC 314-6, Pasadena, CA 91101. sitive infrared line data from the Infrared Space Observatory 4 Institute of Astronomy, University of Cambridge, Madingley Road, Cam- (ISO) was an important first step to peering more deeply into bridge CB3 0HA, UK. buried nuclear sources (Genzel et al. 1998; Laurent et al. 2000; 5 Max-Planck-Institut fu¨rAstronomie,Ko¨nigstuhl 17, 69117 Heidelberg, Sturm et al. 2002; Peeters et al. 2004b). Genzel and collaborators Germany. 6 Astrophysics Group, Imperial College, Blackett Laboratory, Prince Consort were the first to show that ionization-sensitive indices based on Road, London SW7 2AZ, UK. mid-infrared line ratios correlate with the strength of polycyclic 7 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, aromatic hydrocarbon (PAH) emission features. Active galactic MD 21218. 8 nuclei (AGNs) in particular show weak PAH and large ratios of Princeton University Observatory, Peyton Hall, Princeton, NJ 08544. high- to low-ionization line emission. Interestingly, while Genzel 9 NASA Ames Research Center, MS 245-6, Moffett Field, CA 94035. 10 Institute for Astronomy, University of Hawaii, 2680 Woodlawn Drive, et al. (1998) found that 70%–80% of their ULIRG sample is Honolulu, HI 96822. mainly powered by starburst activity, the percentage appears to 11 Department of Physics and Astronomy, University of Missouri, Columbia, drop for higher luminosity ULIRGs (e.g., Veilleux et al. 1999; MO 65211. but see also Farrah et al. 2003). Taniguchi et al. (1999) and Lutz 12 Department of Physics and Astronomy, Arizona State University, Box 871504, Tempe, AZ 85287. et al. (1999) suggest that optical and infrared classifications 13 Department of Astronomy, Yale University, P.O. Box 208101, New agree if the nuclei of LINER-like ULIRGs are in fact dominated Haven, CT 06520. by shocks driven by powerful supernova winds. Mid-infrared 161 162 DALE ET AL. Vol. 646 lines observed by ISO were also used to probe the physical char- 3. THE DATA acteristics and evolution of purely starbursting nuclei (Thornley The full SINGS observing program and the data processing et al. 2000; Verma et al. 2003) and Galactic H ii regions (Vermeij are described by Kennicutt et al. (2003), Smith et al. (2004), and & van der Hulst 2002; Giveon et al. 2002). An important result Dale et al. (2005). Here we briefly summarize the spectral ob- stemming from these efforts is that stellar aging effects appear to servations and data processing relevant to this paper. result in H ii regions generally having higher excitations than starbursting nuclei. 3.1. Spitzer Infrared Spectroscopy Observations The unprecedented sensitivity and angular resolution afforded and Data Processing by the Spitzer Space Telescope allow an even more detailed view into the nature of galactic nuclei (e.g., Armus et al. 2004; Smith High-resolution spectroscopy (R 600) was obtained in the et al. 2004). The Spitzer Infrared Nearby Galaxies Survey (SINGS) Short-High (10–19 m) and Long-High (19–37 m) modules, takes full advantage of Spitzer’s capabilities by executing a com- and low-resolution spectroscopy (R 50 À 100) was obtained in prehensive, multiwavelength survey of 75 nearby galaxies span- the Short-Low (5–14 m) and Long-Low (14–38 m) modules ning a wide range of morphologies, metallicities, luminosities, (Houck et al. 2004a). Figure 1 shows example spectra for a va- and star formation activity levels (Kennicutt et al. 2003). The riety of sources (Long-Low data are not used elsewhere in this work). Nuclei were generally mapped with a 3 ; 5 grid (Short- sensitivity of Spitzer, coupled with the proximity of the SINGS ; sample, allows dwarf galaxy systems fainter than L 107 L High and Long-High) and a 1 18 grid (Short-Low), utilizing FIR half-slit width and half-slit length steps. Extranuclear targets to be spectroscopically probed in the infrared for the first time. In ; addition, prior to Spitzer the only individual extragalactic H ii were observed with a similar scheme but with a 1 9 Short-Low grid. For a subset of nine sources with extended circumnuclear regions that were detectable with infrared spectroscopy resided ; in the Local Group (e.g., Giveon et al. 2002; Vermeij et al. 2002). star formation we obtained slightly larger (6 10) Short-High In contrast, SINGS provides infrared spectroscopic data for nearly nuclear maps. Owing to the different angular sizes subtended by 00 ; 00 100 extragalactic H ii regions, residing in systems as near as Local the instruments, the resulting maps are approximately 57 31 00 ; 00 Group members to galaxies as far as 25 Mpc. The SINGS data and 57 18 in Short-Low (nuclear and extranuclear, respec- 00 ; 00 00 ; 00 set thus samples a wider range of environments than previously tively), 45 33 in Long-High, and 23 15 in Short-High 00 ; 00 observed with infrared spectroscopy. This diversity in the SINGS (the 10 expanded Short-High nuclear maps are 40 28 ). All sample provides a huge range for exploring physical parameters integrations are 60 s per pointing, except the Short-Low nuclear with mid-infrared spectral diagnostics. The high-ionization lines maps, which are 14 s per pointing. The effective integrations are historically used in such diagnostics, such as [O iv]25.89m and longer since each location was covered 2–4 times. [Nev] 14.32 m, are relatively weak and can be difficult to detect The individual data files for a given spectral map were assem- in lower luminosity systems. Fortunately, high-ionization lines bled into spectral cubes using the software CUBISM (Kennicutt are not the only route to determining whether a galaxy harbors a et al.