Spatially Resolved PAH Emission Features in Nearby, Low Metallicity, Star-Forming Galaxies
Korey Haynes, John M. Cannon, Evan D. Skillman, Dale C. Jackson, Robert D. Gehrz
aa r X i v : . [ a s t r o - ph . C O ] S e p Astrophysical Journal, in press
Spatially Resolved PAH Emission Features in Nearby, LowMetallicity, Star-Forming Galaxies
Korey Haynes
Department of Physics & Astronomy, Macalester College, 1600 Grand Avenue, Saint Paul,MN 55105Department of Physics & Astronomy, George Mason University, Fairfax, VA 22030 [email protected]
John M. Cannon
Department of Physics & Astronomy, Macalester College, 1600 Grand Avenue, Saint Paul,MN 55105 [email protected]
Evan D. Skillman
Department of Astronomy, University of Minnesota, Minneapolis, MN 55455 [email protected]
Dale C. Jackson
Sandia National Laboratories, Albuquerque, New Mexico [email protected]
Robert D. Gehrz
Department of Astronomy, University of Minnesota, Minneapolis, MN 55455 [email protected]
ABSTRACT
Spitzer /IRS spectral maps are presented forthree nearby, low-metallicity dwarf galaxies (NGC 55, NGC 3109 and IC 5152)for the purpose of examining the spatial distribution and variation of polycyclicaromatic hydrocarbon (PAH) emission. The sample straddles a metallicity of 12+ log(O/H) ≈
8, a transition point below which PAH intensity empirically dropsand the character of the interstellar medium changes. We derive quantitativeradiances of PAH features and atomic lines on both global and spatially-resolvedscales. The
Spitzer spectra, combined with extensive ancillary data from theUV through the mid-infrared, allow us to examine changes in the physical en-vironments and in PAH feature radiances down to a physical scale of ∼
50 pc.We discuss correlations between various PAH emission feature and atomic lineradiances. The (6.2 µ m)/(11.3 µ m), (7.7 µ m)/(11.3 µ m), (8.6 µ m)/(11.3 µ m),(7.7 µ m)/(6.2 µ m), and (8.6 µ m)/(6.2 µ m) PAH radiance ratios are found tobe independent of position across all three galaxies, although the ratios do varyfrom galaxy to galaxy. As seen in other galaxies, we find no variation in the grainsize distribution as a function of local radiation field strength. Absolute PAHfeature intensities as measured by a ratio of PAH/(24 µ m) radiances are seento vary both positionally within a given galaxy, and from one galaxy to anotherwhen integrated over the full observed extent of each system. We examine di-rect comparisons of CC mode PAH ratios (7.7 µ m)/(6.2 µ m) and (8.6 µ m)/(6.2 µ m) to the mixed (CC/CH) mode PAH ratio (7.7 µ m)/(11.3 µ m). We find littlevariation in either mode, and no difference in trends between modes. While thelocal conditions change markedly over the observed regions of these galaxies, theproperties of PAH emission show a remarkable degree of uniformity. Subject headings: galaxies: evolution — galaxies: dwarf — galaxies: irregular —galaxies: individual (NGC 55, NGC 3109, IC 5152)
1. Introduction1.1. On The Origin of PAH Emission Features
The mid-infrared (MIR; ∼ µ m) is a very rich spectral regime. The Rayleigh Jean’stail of the stellar spectral energy distribution (SED) contributes at short wavelengths, andthe rise of the warm dust continuum contributes at long wavelengths. Between them isthe large silicate extinction feature centered around 9.7 µ m. Superposed on this complexcontinuum are numerous broad emission features (notably at 3.3 µ m, 6.2 µ m, 7.7 µ m, 8.6 3 – µ m, 11.3 µ m and 12.6 µ m) and atomic lines (e.g., [S IV] at 10.5 µ m, [Ne II] at 12.8 µ m,[Ne III] at 15.56 µ m, and [S III] at 18.71 µ m, among others). The broad emission features areattributed to vibrational transitions from polycyclic aromatic hydrocarbon (PAH) molecules.PAHs are planar molecules containing tens to thousands of carbon atoms. While theassignment of these features to specific molecules is still an area of very active theoreticaland laboratory investigation, it is generally agreed that this type of molecule (carbon atomsarranged in a system of fused hexagonal rings terminated by carbon-hydrogen single bonds)produces the broad emission features in the MIR spectral range. The four most luminousfeatures in astrophysical sources appear at wavelengths of 6.2 µ m, 7.7 µ m, 8.6 µ m and 11.3 µ m (Leger & Puget 1984). These emission features are produced following the absorptionof energetic (typically far-ultraviolet) photons by PAH molecules; the molecules bend andstretch in fundamental modes that produce infrared photons as this energy is dissipated.There are two principal types of vibrational modes in PAH molecules. The first involvesthe stretching of carbon-carbon (CC) bonds and contributes to the features at 6.2 µ m, 7.7 µ mand 8.6 µ m. The second involves the bending of carbon-hydrogen (CH) bonds; these modescan occur either in the plane (“wagging” modes) or out of the plane of the molecule. Thewagging mode contributes to the 7.7 µ m and 8.6 µ m features, while the out-of-plane bendingmode is less energetic and contributes to the 11.3 µ m feature. For detailed discussions ofthese mode assignments (some of which are still under investigation) we refer the readerto Allamandola et al. (1989), Draine & Li (2007), Tielens (2008), Bauschlicher et al. (2008,2009) and the many references therein.The relative variations in intensity of these features is sensitive to both the physicalsize and to the ionization state (i.e., to the hardness of the radiation field) of the molecules.It is generally accepted that the smallest PAH molecules produce the shortest-wavelengthfeatures, and that increasing the number of carbon atoms leads to increased intensities of thelonger wavelength features. However, this simple interpretation is clouded somewhat by theprediction that the 7.7 µ m band is produced by a mixture of both small and large molecules(Bauschlicher et al. 2009). Evidence is also mounting that the 6.2 µ m feature may arisefrom PAHs with different compositions than the longer-wavelength features, either from theinclusion of N atoms (Bauschlicher et al. 2008, 2009) or from the carriers being among thesimplest C-based ring structures (e.g., protonated naphthalene; Ricks et al. 2009). In thissimple model, the ratios of radiances of long-wavelength PAH features to shorter-wavelengthfeatures provides a diagnostic of the grain size distribution; examining the changes in thegrain size distribution as a function of local conditions is one of the primary goals of thisstudy.The hardness of the radiation field also contributes to the variations in intensity of the 4 –various PAH bands. Laboratory predictions (e.g., Allamandola et al. 1999; Kim & Saykally2002; Bauschlicher 2002) suggest that ionized PAH molecules produce stronger CC modevibrations (e.g., Bakes et al. 2001). Bauschlicher et al. (2008) finds that the intensities ofthe CC modes arising from cationic PAHs are increased by an order of magnitude or morecompared to neutral species. Observationally this manifests itself as a strengthening of theCC modes (i.e., the 6.2 µ m, 7.7 µ m and 8.6 µ m PAH features) relative to the CH modes(primarily the 11.3 µ m feature, although the 8.6 µ m feature is also affected to a lesser extent)in a given source. Quantitative studies have been undertaken to attempt to connect thesevariations of PAH intensities to the physical conditions of the gas in a variety of astrophysicalenvironments (see further discussion below). Galaxies radiate a large percentage of their total luminosity in the infrared, from tens ofpercent in star forming galaxies up to 99% in ultra luminous infrared galaxies (Sanders et al.1988; Sanders & Mirabel 1996). An especially rich segment of the IR region falls between ∼ µ m - 20 µ m, where emission features from PAHs arise in most star-forming galaxies. Theemission from PAHs varies from a few percent up to 20% of the total MIR luminosity innormal galaxies (Helou et al. 2000), but this percentage drops sharply with decreasing metal-licity (Engelbracht et al. 2005; Jackson et al. 2006; Smith et al. 2007). Isolating the processesthat affect PAH emission intensity allows investigation of the physical conditions present instar forming regions (Kennicutt et al. 2003). This in turn leads to enhanced understand-ing of the interplay between PAH emission and star formation, the role of radiation fieldcharacteristics in governing PAH intensities, and the stellar-interstellar medium connection.Many investigations have been conducted on PAH emission in a wide range of galaxytypes, but the observations of low-metallicity galaxies in particular have been challenging,in large part due to the difficulties of obtaining spectra of sufficient sensitivity and spatialresolution from low-mass (and often low surface-brightness) galaxies. Using Infrared Ar-ray Camera (IRAC) imaging, Engelbracht et al. (2005) discovered a threshold metallicityof 12 + log(O/H) ≈
8, below which PAH intensity drops sharply, and these results havebeen confirmed by a number of other studies (e.g., Hogg et al. 2005; O’Halloran et al. 2006;Jackson et al. 2006; Rosenberg et al. 2006; Engelbracht et al. 2008; Wu et al. 2010, and refer-ences therein). This metallicity apparently signifies a fundamental change in the nature ofthe interstellar medium (ISM); the molecular component of more metal-rich galaxies is easilystudied via CO observations. However, in systems more metal-poor than ∼
25% Z ⊙ , CO nolonger traces H accurately and the dominant molecular component by mass remains largely 5 –unconstrained (Taylor et al. 1998; Leroy et al. 2005). In this regard, PAH observations ofmetal-poor galaxies are especially important, as they provide one of our only probes of anotherwise elusive ISM component.Multiple factors have been found to influence the amount of infrared emission from PAHmolecules. The metal content of the host galaxy might be expected to play an importantrole, since the carrier molecules consist in part of heavy elements. Indeed, Rosenberg et al.(2006) find a correlation between diffuse 8 µ m luminosity and metallicity for a large sample ofstar-forming dwarf galaxies. The same investigation also finds a more significant correlationbetween PAH luminosity and current star formation rate (SFR) than with metallicity. TheGordon et al. (2008) study of H II regions in M 101 reaches a similar conclusion, where PAHluminosities are more closely tied to the local ionization index than to the metallicity. Theseresults suggest that multiple factors contribute to the observed PAH luminosity in a givenregion, including process related to both the growth and the stimulation of the molecules.Dwarf galaxies offer canonically “simple” environments in which to explore variations inthe intensities of PAH features and emissions. Their characteristically low masses make themmore susceptible than spiral galaxies to turbulence and feedback (Tenorio-Tagle & Bodenheimer1988), allowing recent disturbances of the galaxy’s dust and gas to be readily seen. Typicaldwarfs display solid body rotation and lack shear, preserving structures in the ISM. Dwarfgalaxies are abundant in the local universe, and were even more so in the past. The lowmetallicities typical of these systems approximate the conditions of high redshift galaxies,revealing the physical properties of some of the most distant galaxies by using some of thenearest as a proxy.Previous studies of PAH emission in dwarf galaxies have sought to explain how thesemolecules form, survive, and are destroyed in low-metallicity systems. Jackson et al. (2006)observed a sample of 15 Local Group dwarfs, including the 3 galaxies in the present study.They find that both metallicity and local radiation field properties influence the intensity ofPAH emission. They also find that diffuse 8 µ m emission (strictly a nonlinear composite ofthe 6.2 µ m, 7.7 µ m and 8.6 µ m PAH features, as well as the underlying warm dust continuum)cannot be predicted by the mass of the galaxy alone. Another concern was whether youngersystems have time to populate their ISM with the requisite metals to grow PAHs in thefirst place, but Jackson et al. (2006) rules out this hypothesis. The galaxies in their sample(and therefore ours) have all populated their asymptotic giant branch (AGB), and thereforehave the necessary metals to grow PAHs (see, for example, the discussion in Tielens 2008),whether or not they display PAH emission. They also dismiss outflow being responsiblefor removing PAHs from low-mass galaxies since the HI distributions for the galaxies intheir sample lack evidence of large-scale blowout. However, the relative importance of PAH 6 –destruction remains an open issue; SNe may be able to destroy PAHs at a rate comparableto the production rate in low-metallicity environments.Environmental differences in the character of PAH emission have already been observedwithin galaxies using Spitzer data (e.g., Smith et al. 2007; Dale et al. 2009). In spiral systemswith large, well-defined structures, this is to be expected. The nuclear regions, spiral arms,inter-arm regions, and extended disks of such galaxies can have vastly different physicalconditions, as well as different chemical compositions and SFRs. In contrast, dwarf galaxiespresent surprisingly uniform metallicity throughout their high surface brightness components(e.g., Kobulnicky & Skillman 1996, 1997). While star formation regions are often scatteredthroughout the disks of dwarfs, these regions appear to have only a minimal effect on themetallicity of the surrounding interstellar gas.The uniform metal content within a given dwarf galaxy is particularly relevant to theissue of stimulating PAH carriers into emission. It has been shown that PAH emissionhighlights the edges of photodissociation regions (PDRs) in spiral galaxies (Giard et al. 1994;Helou et al. 2001, 2004). As a result, it has been proposed that radiation inside PDRs is toointense for the survival of PAH molecules, while at large distances from PDRs the radiationfield would be too weak to stimulate the molecules. There is also evidence that in regionsof hard and/or strong radiation fields, the type of PAH carrier can have a marked effecton the relative PAH intensities (Compi`egne et al. 2007). If radiation field strength were theonly factor affecting the emission from PAH carriers, then the uniform metallicity ISM ofdwarfs would reveal this as a correspondence between the locations of massive stars andthe presence of PAH emission. However, PAH emission has been observed in the quiescentISM of dwarfs via imaging (e.g., Cannon et al. 2006; Jackson et al. 2006), including in thegalaxies observed in this study, so the physical proximity of PAHs to PDRs cannot act alonein governing the amount of emission from PAH molecules in the metal-poor ISM.The spatially resolved properties of PAH emission using spectra have not been investi-gated extensively for low-metallicity systems, though many studies of diffuse emission (arisingfrom multiple PAH features within a bandpass) are now in the literature (see above). Inthis paper, we examine the intensities and spatial variations of several PAH features in threelow-metallicity dwarf galaxies. This approach allows one to separate and study the individ-ual PAH bands contained within the comparatively broad IRAC bandpasses. Our interestis in both how the relative intensities of these features change from one galaxy to the next,and more importantly, within a given galaxy. A comparison of the ratios of PAH bandswithin a galaxy should be instructive as to how environmental factors and PAH intensitiesare related. As one effort to trace these environmental factors, maps of [Ne II] emission arealso extracted from our spectral maps. These images reveal the morphology of ionized gas 7 –and offer insights into how local radiation field properties affect the relative fluxes of thevarious PAH bands.The three galaxies chosen for this sample are NGC 55, NGC 3109, and IC 5152 (seeTable 1). All three are low-metallicity (12% Z ⊙ ≤ Z ≤
49% Z ⊙ ; see below) dwarf galaxieslocated within 2 Mpc. These systems were chosen for their proximity (maximizing spatialresolution) and for their ongoing star formation. Further, the metallicities of the galaxiesin this sample bracket the aforementioned empirical PAH transition metallicity. While oursample size is modest, the spatially resolved data for each system provide new insights intothe variations of PAH intensities in metal-poor environments.This research uses the Infrared Spectrograph (IRS; Houck et al. 2004) Short-Low moduleto create spectral maps of the high IR surface brightness regions of the sample galaxies.IRS is able to cleanly separate the 6.2 µ m, 7.7 µ m, and 8.6 µ m PAH features, as well asto differentiate between the PAH features and the underlying warm dust continuum. Weexamine the radiances of PAH features across the observed regions of the target galaxies.From these measurements, combined with extensive ancillary data from the UV through themid-infrared, it is possible to correlate emission from various gaseous and stellar components.We can then constrain some of the factors suspected of affecting the intensity of PAH emission(e.g., radiation field hardness, PAH molecule size distribution).
2. Observations and Data Reduction
Data were obtained using
Spitzer /IRS for general observer (GO) Program 40457 (PISkillman). Spectra were acquired with the Short-Low (SL) slit (5-15 µ m) for all three targets.Mapping used parallel steps of size 28.35 ′′ , and perpendicular steps of size 1.85 ′′ . NGC 55was observed for a total of 7 hours, using one cycle of ramp duration 14 seconds on 6 paralleland 60 perpendicular steps. 6.2 hours were required to observe IC 5152 with one 60 secondramp on 2 parallel steps and 50 perpendicular steps. NGC 3109 also required 6.2 hours fortwo cycles of 60 second ramps on one parallel step and 70 perpendicular steps. Observationswere completed between November 2007 and January 2008. IRS mapping footprints can beseen in Figure 1, where we overlay the IRS spectral map placement on IRAC 8 µ m images.Note that we targeted the areas of highest IR surface brightness.Basic calibrated data (BCD) from the Spitzer pipeline has been flat-fielded and convertedto specific intensity units (MJy sr − ). We use the CUbe Builder for IRS Spectral Mapping(CUBISM; Smith et al. 2007) for much of the analysis in this investigation. First, the datawere cleaned using a 4-sigma cut, removing hot pixels. Further cleaning was then done by 8 –hand to assure data quality.For IC 5152 and NGC 3109, dedicated sky observations were used for background sub-traction. NGC 55 had a sufficient number of off-target BCD sets that could be used forbackground subtraction. These off-target BCDs were judged to be cleaner than the ded-icated sky data, since more and cleaner astronomical observation requests (AORs) fromthe target data set were available for background subtraction than from the dedicated skyobservations.For stitching together spectral data from the first and second order of the slit, a loworder polynomial was fitted to the SL2 data in the region of overlap. Specific intensity valueswere then averaged in the region of overlap. All data were then smoothed over three pixels.Fitting errors were small, so total errors are presented, except in the region of overlap, wherethe fitting errors were significant. For NGC 55, NGC 3109, and IC 5152, average errors were2%, 17%, and 10% respectively. The only areas of significantly higher error were in theregion of overlap, where errors ranged up to three times higher than the averages outside ofthe overlap region.The program PAHFIT (Smith et al. 2007) was used to fit the spectra and to extractnumerical values describing PAH features and emission lines. PAHFIT was trained on anumber of high S/N ratio galaxies, and uses a simple, physically motivated model thatincludes dust continuum, starlight, emission lines and bands, and extinction. Most com-ponents are fixed, including starlight temperature (5000 K) and central wavelengths andwidths of dust features. The dust continuum is based on eight components represented bymodified blackbodies at a range of temperatures between 35 and 300 K. These componentsare allowed to be zero, though not negative. The extinction is given as a power law withsilicate features peaking at 9.7 µ m and 18 µ m. The line feature wavelengths are allowedto vary by 0.05 µ m, and line widths are allowed to vary by 10% from their default values.Spectral lines are represented by Gaussian profiles, while dust features are represented byDrude profiles, both individual and blended. Because of the highly blended nature of manyof the PAH features, the PAHFIT “Main Power Feature” program was used to combine themost heavily blended features into complexes by adding individual features across a specifiedwavelength interval. PAHFIT is designed to use combined SL and LL Spitzer data in orderto accurately reconstruct the continuum emission; unfortunately, LL data are unavailable forour sample galaxies. Therefore, while the total continuum appears to fit the spectra quitewell, some caution should be used in interpreting the continuum fits themselves (especiallyat wavelengths outside of the 5.5–14.5 µ m range covered by the IRS SL1 and SL2 slits).PAHFIT fits the spectra by minimizing the global χ , and returns statistical uncer-tainties. Global χ values for NGC 55, NGC 3109, and IC 5152 are 59.122, 5.76, and 5.20, 9 –respectively. Average uncertainties in dust features were 2%, 17%, and 7% for the global fitsfor NGC 55, NGC 3109, and IC 5152, respectively.In addition to 1D spectra, 2D spectral maps from various wavelength regions werealso generated using CUBISM’s mapping feature. This procedure allows the user to spec-ify “peak” and “continuum” wavelength intervals. CUBISM then subtracts the specifiedcontinuum from the peak, and creates a map of the resulting emission. For maps coveringspectral regions between 7.4 µ m and 7.6 µ m, the overlap region between SL1 and SL2, amore detailed method was needed. In this case, maps were extracted separately in each slitand averaged together in the area of overlap before being combined.
3. Global Comparisons3.1. Galaxy Properties
As noted above, the galaxies in our sample were selected based on proximity and thepresence of ongoing star formation. These three dwarf irregulars span a modest range inmetallicity ( ∼ λ ± ± ± µ m PAH feature via imaging.Typically the IRAC band 1 or 2 images (at 3.6 µ m and 4.5 µ m, respectively) are assumed toaccurately represent the underlying stellar continuum; these images are scaled and subtractedfrom the IRAC 8 µ image (see detailed discussion in Jackson et al. 2006). The resultingdiffuse 8 µ m emission (which includes contributions from the 6.2 µ m, 7.7 µ m and 8.6 µ mPAH features) in these systems scales approximately with the metallicity. NGC 3109 has thelowest gas-phase abundance ( ∼
12% Z ⊙ ) and the lowest total diffuse 8 µ m flux density (0.06Jy). IC 5152 is slightly more metal-rich ( ∼
18% Z ⊙ ) and harbors a larger total 8 µ m fluxdensity (0.16 Jy). Moving above the threshold metallicity (12 + log(O/H) ≃ ∼
49% Z ⊙ ) is an order of magnitude brighter in the IRAC 8 µ m band(1.4 Jy). Note by examining Figure 1 that our IRS spectral maps cover only the regions ofhighest surface brightness 8 µ m emission; the IRS field of view encompasses 41%, 18%, and57% of the intensity in the IRAC 8 µ m field of view for NGC 55, NGC 3109, and IC 5152,respectively (with no foreground star correction or continuum subtraction applied). 10 –While the scaling of PAH emission (as measured by the diffuse 8 µ m flux density) withmetallicity is expected based on recent empirical results, the comparisons with other globalparameters are more ambiguous. The system with the smallest dynamical mass and currentSFR (IC 5152) is a factor of ∼ µ m band than NGC 3109.While the current SFRs in these two systems differ only at the ∼
10% level, it is perhapsmore surprising that NGC 3109 is ∼
16 times more massive than IC 5152. Moving furtherupward in mass, NGC 55 is the most massive and metal-rich system in the sample. Itscurrent SFR and diffuse 8 µ m flux density are both more than an order of magnitudelarger than in NGC 3109, although the mass and metallicity are only larger by a factor of ∼
3. Taken together, this discussion highlights the complexity of these multiple inter-relatedproperties. While statistical correlations exist between some parameters in large samples(e.g., Rosenberg et al. 2006), it is apparent that individual systems are complex compositesof multiple factors. It is our aim to investigate these variations within three selected systemson a spatially resolved basis in the present investigation. We thus hereafter use the term“global” to identify results or properties obtained by integrating spectra over the IRS fieldsof view as shown in Figure 1. These global results are separated from those attained byindividually integrating over smaller regions within the field of view (see § The global spectra shown in Figures 2, 3, and 4 were created by integrating the IRSspecific intensity over the entire observed area of the galaxy for NGC 55, NGC 3109, andIC 5152, respectively. While these spectra vary significantly in total intensity and signal tonoise ratio, the overall shapes show similar emission patterns. The most prominent featuresin all three galaxies are the 7.7 µ m PAH feature (somewhat blended with the 8.6 µ m PAHfeature), the 11.3 µ m PAH feature, and the blended feature composed of [Ne II] emissionat 12.8 µ m and the 12.6 µ m PAH feature. The [S IV] feature at 10.5 µ m is prominent inNGC 55, weak in IC 5152, and undetectable in NGC 3109.The global spectra are fitted by PAHFIT and decomposed into multiple components,including the thermal dust continuum, stellar continuum, PAH features, and atomic andmolecular emission lines. PAHFIT was set such that the central wavelengths and widths ofthe features were not allowed to vary between galaxies. These decomposed global spectrafor NGC 55, NGC 3109, and IC 5152 are shown in Figures 5, 6, and 7, respectively. Multiplecomponents contribute to each peak visible in the full fit. For example, using the globalspectrum of NGC 55 (Figure 5; highest S/N) as a guide, the 7.7 µ m PAH complex has 11 –contributions from components at 7.6 µ m and 7.8 µ m. Similarly, the minor contributionof the 8.3 µ m PAH feature is evident between the 8.6 µ m and 7.7 µ m complexes, and the11.3 µ m complex separates into two features at 11.2 and 11.3 µ m, respectively. We remindthe reader that the PAHFIT Main Power Feature (see discussion in §
2) explicitly calculatesradiances across these blended complexes.Figures 5, 6, and 7 show that the monochromatic specific intensity at all frequencies ishighest in NGC 55. It is interesting to note that the stellar continuum is weaker comparedto the dust continuum in NGC 55 than in the other two systems; we attribute this to both alarger dust content and a higher star formation rate in NGC 55 than in NGC 3109 or IC 5152.We also note with interest that the fit for NGC 3109 shows contributions from H S(5) and[Ar II] at 6.86 µ m and 6.94 µ m respectively (though the errorbars are appreciable), featuresthat are largely absent from the other two galaxies.The absolute values for the extracted radiances (units of W m − sr − ) of the PAH fea-tures and ionized atomic lines are given in Table 2. As noted above, NGC 55 has the largestPAH luminosity in our sample; it has > ∼ > ∼
10 times higher radiances in each PAHfeature than IC 5152 or NGC 3109, respectively. NGC 55 also has the highest ionized atomicline radiances (compare the [Ne II] 12.8 µ m values). Given these properties, the appearanceof the modest ionization potential [Ar II] line (15.76 eV; Cox 2000) in only NGC 3109 is notsurprising. The other two systems have stronger radiation fields: the [S IV] 10.5 µ m featurein NGC 55 has an ionization potential of 34.79 eV, and the [Ne II] 12.8 µ m line in bothNGC 55 and IC 5152 has an ionization potential of 21.56 eV. The PAH features overwhelmthe [Ar II] line in these systems.We compare the ratios of the radiances of the four prominent PAH features in aglobal sense for each galaxy in Table 3. The global (8.6 µ m)/(11.3 µ m) ratio is the same(0.58 ± µ m)/(11.3 µ m), (7.7 µ m)/(6.2 µ m), and (7.7 µ m)/(11.3 µ m) ratios have more significant scatter (fac-tors as large as 3-4). It is not immediately clear why the (8.6 µ m)/(11.3 µ m) ratio is themost stable across all three galaxies. Considering the small sample size, it is possible that thesmall dispersion is a coincidence. Since the 8.6 µ m and 11.3 µ m PAH features are postulatedto have contributions from CH bending modes (though the CH contribution to the 8.6 µ mfeature is weak when the PAH carriers are ionized; see § µ m)/(6.2 µ m) ratio, which is dominated by the CC stretching modes 12 –when the carriers are ionized) are observed. The constancy of the global (8.6 µ m)/(11.3 µ m)ratio is thus likely due, at least in part, to a similar grain size distribution in this sample. Previous investigations have examined the variations (or lack thereof) in PAH/PAHratios both globally among galaxies (e.g., Smith et al. 2007; Galliano et al. 2008; Hunt et al.2010) and on a spatially resolved basis within galaxies (e.g., Galliano et al. 2008; Gordon et al.2008). From these works, evidence is mounting that the relative intensities of the major PAHfeatures are remarkably uniform, even across a wide range of local and global environments.The Smith et al. (2007) and Galliano et al. (2008) samples include a wide variety of sources(spirals, AGNs, dwarfs, H II regions); the Hunt et al. (2010) sample is composed entirely ofblue compact dwarfs (BCDs); the Gordon et al. (2008) study probes individual H II regionsin M 101. Such a collection of environments harbors a wide range of stellar populations,current SFRs, metal and dust contents, and large-scale dynamical processes. It would thenbe expected that the relative radiances of ionized (6.2 µ m, 7.7 µ m and 8.6 µ m) versus neutral(11.3 µ m), and large versus small (in the same ionization state; e.g., 8.6 µ m versus 6.2 µ m),PAH features would also vary considerably.To examine both theories (grain size and ionization state) across this broad sample ofenvironments, we plot in Figure 8 the three galaxies in this sample, the inner few kiloparsecsof a subset of SINGS galaxies from Smith et al. (2007), a wide range of galaxy types fromGalliano et al. (2008), the sample of BCDs from Hunt et al. (2010), and the star formationregions in M 101 studied by Gordon et al. (2008). This plot is designed to show contribu-tions from larger PAH molecules increasing up the y axis, and contributions from ionized(as compared to neutral) PAH molecules increasing along the x axis. All data representedwere decomposed using PAHFIT, with the exception of the data points from Galliano et al.(2008), who used their own spectral decomposition method (in their study, Galliano et al.2008 use two fitting methods, Spline and Lorentzian; we present only the Lorentzian method,as it more closely approximates the PAHFIT method). Sources were only included on ourplot if the radiances of all four PAH features were available. We stress that the plot containsinformation about both entire galaxies (e.g., the BCDs) and individual regions within galax-ies. Further, the data acquisition techniques (e.g., spectral mapping versus single pointing),analysis techniques (see above), and sensitivities may differ from one data point to another.While we remain mindful of these caveats, this collection of data represents, to our knowl-edge, all currently available information on the radiances of the 6.2 µ m, 7.7 µ m and 8.6 µ mPAH features over a range of ambient metallicities within nearby galaxies. 13 –We find that our three galaxies nicely fit within the overall trend established by thelarger (and broader) samples, which on the whole show a minimal range in size distributionacross an order of magnitude in ionization state (though outliers do exist; see the captionof Figure 8). This implies that a more or less constant grain size distribution exists overthe appreciable range in ionization fraction covered by this sample. Considering the threegalaxies in the present study (shown in red in Figure 8), NGC 55 falls in the center of thelocus of points from the larger compilation; this is not surprising, given its comparativelylarge mass, metallicity, and dust content. IC 5152 also falls within the larger distribution,though at a lower ionization index than NGC 55. It may seem surprising that NGC 3109appears at the highest ionization index (i.e., largest global (7.7 µ m)/(11.3 µ m) ratio) in thisplot; however, this is easily interpreted as its comparatively low S/N spectrum being heavilyweighted toward the region of highest IR surface brightness (and thus most intense radiationfield; we return to this point in § µ m)/(6.2 µ m) and (7.7 µ m)/(11.3 µ m) ratios versus gas-phase oxygen abundances for asubset of the systems in Figure 8. In creating this plot, we implicitly assume no abundancegradients in the dwarfs, and we explicitly exclude global values for spirals, as they have well-documented abundance gradients. We do, however, include the measurements of individualH II regions in M 101 (each with abundance measurements) presented by Gordon et al.(2008). In interpreting this plot, we remain mindful that the dataset is inhomogeneousin the sense that the local conditions within individual star formation regions of a spiraldisk will be quite distinct from those within a dwarf galaxy. We also stress again that thehigh global (7.7 µ m)/(11.3 µ m) ratio in NGC 3109 is heavily weighted toward the region ofhighest IR surface brightness. With these caveats in mind, we find only very weak evidencefor a trend of increasing grain size or ionization index with increasing metal abundance. 14 –This is in agreement with previous studies (e.g., Jackson et al. 2006; Rosenberg et al. 2006;Gordon et al. 2008), where the effect of metallicity has been found to be secondary to otherfactors in controlling the intensities of PAH emission features. While the total amountof metals in a galaxy is clearly an important property in governing the intensity of PAHemission (Engelbracht et al. 2005, 2008), evidence shows that the relative intensities of thePAH features show a remarkable degree of uniformity in the systems considered here.
4. Spatially Resolved Emission
In order to examine the comparative spatial distributions of dust, gas, stellar popu-lations, and radiation field strengths in our sample galaxies, images are shown at variouswavelengths in Figures 10, 11, and 12 for NGC 55, NGC 3109, and IC 5152, respectively.Here we exploit our mapping observational strategy (see § µ m – 9 µ m range). The spatial locationof the regions that contribute are known; a map is then constructed of the emission in agiven wavelength interval (see Smith et al. 2007b for details). The complexity of the spec-trum of each system (see Figures 2, 3, and 4) requires fine-tuning of the spectral extractionregions for each feature individually. We explicitly list the wavelength intervals over whichthe emission maps are created for each system in Table 4.Panels (i), (j), and (k) of Figures 10, 11, and 12 show images created from the IRSdata. The spatial distributions of IRS 11.3 µ m (see panel j of each figure) and broad 8 µ m(integrated from 7.288 µ m – 8.900 µ m in all systems; note that this includes contributionsfrom both the 7.7 µ m and the 8.6 µ m PAH complexes; see panel i of each figure) emissioncorrelate well in all three galaxies. The IRAC and IRS 8 µ m maps also agree in most areas;the small observed differences are most likely due to the larger bandwidth of the IRAC 8 µ mfilter (extending from ∼ µ m – ∼ µ m). As a check on the quality of the absolute fluxcalibration of the IRS spectra, we explicitly compared the total flux density in the IRAC 8 µ m band with the total flux density integrated over the matching IRAC bandpass in the IRSspectrum of each galaxy; while these images are not shown in Figures 10–12, the resultingflux densities agree at the 8%, 13%, and 17% levels for NGC 55, IC 5152, and NGC 3109,respectively.There is very good spatial correlation between emission from PAH features and ionizedgas as indicated by [Ne II] emission (see panel k of each figure) in NGC 55 and IC 5152. This 15 –immediately suggests that the PAH emission is arising from regions with comparatively hardradiation fields (recall that the ionization potential of [Ne II] is ∼ µ m in Figures 10, 11 and 12, we conclude that the multiwavelengthcharacteristics of NGC 55 and NGC 3109 are consistent with variable extinction of up toa few magnitudes in the UV; both systems show good agreement between 11.3 µ m PAHsurface brightness and 4.5 µ m morphology. IC 5152 is more challenging to interpret: itsnear-infrared stellar population bears a strikingly different morphology than the PAH or[Ne II] emission. This can be interpreted as more severe localized extinction than in NGC 55or NGC 3109; such variations have been observed in dwarf galaxies in previous works (e.g.,Cannon et al. 2006). These implied optical and UV extinctions are in agreement with theglobal values derived by Lee et al. (2009), albeit with significant internal variations.In all three galaxies, we examine PAH ratios for individual sources that are spatiallyresolved in our spectral image data cubes. Spectra were extracted from CUBISM maps usingcircular regions of 52.5 pc radius. These extraction regions can be seen in Figures 13 (NGC55), 14 (NGC 3109), and 15 (IC 5152). 52.5 pc is four times the smallest spatial resolutionpossible with the data, or four pixels wide on the IRS map of NGC 55 (D = 2.17 Mpc, themost distant system). At 8.6 µ m, Spitzer ’s diffraction limit is 2.55 ′′ ; 52.5 pc corresponds to5.0 ′′ at the distance of the NGC 55. For NGC 3109 and IC 5152, the extraction regions are8.1 ′′ and 5.5 ′′ , respectively. Extraction regions were chosen to be the same physical size ineach galaxy for optimal comparison of features. 52.5 pc radius regions were judged to best fitphysical variations in structure across the varying angular size of all three galaxies; further,52.5 pc is large enough to analyze on NGC 55, but small enough to allow multiple regions tobe extracted from NGC 3109 and IC 5152. The extracted spectrum of each region was thenfitted with PAHFIT to derive PAH feature and emission line radiances. The results of thesefits are shown in Tables 5, 6, 7, 8, 9, and 10, and discussed in detail in § § § µ m)/(11.3 µ m) PAH radiance ratios span only afactor of ∼ µ m)/(11.3 µ m) ratios is seen for the integrated regions compared to the individual pixelvalues. This is due to the uncertainty cut on the pixels, which removes low S/N pixels fromthe data. This means that the extraction regions, which had no uncertainty cut applied,have the potential to contain pixels of a much lower specific intensity than is possible inthe individual pixel data. Conversely, the individual pixels are therefore weighted towardshigher surface brightness regions of the galaxy in a way the extraction regions are not.This difference is most notable in low S/N regions like Region 2. Region 4, the highestsurface brightness region in NGC 3109, is well within the average range of the rest of thepixels. Examination of Figure 11 shows more diffuse 11.3 µ m emission than other bands inNGC 3109, including the IRS 8 µ m band, which encompasses the 7.7 µ m PAH feature. Fromthis we conclude that integrating over lower surface brightness regions would indeed producea larger contribution from 11.3 µ m compared to 7.7 µ m emission, leading to a smaller ratioof these two values.The small changes (a factor of ∼
5, or just larger than the associated uncertaintiesfor most regions) in the (7.7 µ m)/(11.3 µ m) PAH radiance ratios can be interpreted as arelatively minor variation of the ionization index (i.e., radiation field strength) within a givengalaxy. For NGC 55 and IC 5152, this is to be expected from examination of Figures 10 and12. The agreement in morphology between the [Ne II] emission line and the IRS 8 µ mand 11.3 µ m PAH bands shows that the latter two are primarily arising in regions rich inenergetic photons. The interpretation of the NGC 3109 plot is more ambiguous, primarilybecause there are so many fewer pixels that contribute above the 3 σ level compared to thetwo more massive systems. However, at face value, the variation of the (7.7 µ m)/(11.3 µ m) PAH radiance ratio across NGC 3109 implies that the radiation field strength is highin one specific region (where the individual pixels achieve a S/N ratio above 3) and lowerthroughout most of the system. This is also apparent by examining Figure 11.Two diagnostics of the PAH size distribution are plotted against the (7.7 µ m)/(11.3 µ m) PAH radiance ratio in Figures 16 and 17, the former via the (7.7 µ m)/(6.2 µ m) PAHradiance ratio and the latter via the (8.6 µ m)/(6.2 µ m) PAH radiance ratio. Note that bothof these ratios are dominated by the CC stretching modes in regions where the PAH carriers 17 –are ionized (see the distributions of [Ne II] emission throughout each system in Figures 10,11 and 12). Recall that the 7.7 µ m PAH feature arises from a mixed population of largeand small carriers, so the interpretation of the (7.7 µ m)/(6.2 µ m) ratio as sensitive only tothe PAH size distribution should be treated with more caution than the (8.6 µ m)/(6.2 µ m)ratio. From these plots we see that NGC 55 shows a trend of larger grain sizes (i.e., largerordinate values) in regions of harder radiation field (larger abscissa values), with a moretransparent correlation using the larger difference in grain size (i.e., in the (8.6 µ m)/(6.2 µ m) plot). IC 5152 shows a slight trend in the same sense but the signature is weaker thanin NGC 55. NGC 3109 shows a weak trend in the opposite sense. We again stress that theregion and global radiances in these figures have no minimum specific intensity threshold;hence the reality of the trend in NGC 3109 should be interpreted with caution.In summary, Figures 16 and 17 show variations of radiation field strength and PAHcarrier size distributions on a spatially resolved basis. While very weak trends can be inferredin these data, they are nonetheless consistent with an interpretation of very little variation ingrain size distribution as a function of radiation field strength. This would seem to indicatethat the distribution of PAH grain sizes are long lived, and not significantly affected by localphenomena. Stated differently, the constancy of the PAH/PAH ratios seen amongst galaxies(see detailed discussion in § Data for nineteen circular regions with a radius of 52.5 pc (some of them unavoidablycontaining areas of overlap) were extracted across the observed region of the galaxy (seeFigure 13), and their extracted radiances are shown in Table 5. As expected based onthe results discussed above and shown in Figures 16 and 17, we find very few statisticallysignificant deviations of PAH/PAH radiance ratios derived in these apertures throughoutNGC 55. Table 5 shows the weighted mean values of the (6.2 µ m)/(11.3 µ m), (7.7 µ m)/(11.3 µ m), (8.6 µ m)/(11.3 µ m), (7.7 µ m)/(6.2 µ m) and (8.6 µ m)/(6.2 µ m) radiance ratios andthe standard deviation across this sample. Of the 95 ratios given in the table, we identify11 ratios that deviate at more than the 2 σ level; no variations at the 3 σ level or largerare identified. Only two regions are identified as having variations in more than one band:Region 2 has ∼ σ variations in all ratios except (6.2 µ m)/(11.3 µ m), while Region 16 has 18 – ∼ σ variations in the (7.7 µ m)/(11.3 µ m)m, (8.6 µ m)/(11.3 µ m), and (7.7 µ m)/(6.2 µ m)radiance ratios.We compare the radiances of PAH features with emission lines and broad-band MIPS 24 µ m dust continuum in Table 6. The radiance of the [Ne II] emission line compared to thatof either a neutral (11.3 µ m) or an ionized (8.6 µ m) PAH emission feature show very fewstatistically significant variations within NGC 55. Similarly, the ratio of 24 µ m to PAHradiances is fairly uniform across this galaxy, though the standard deviation is much larger.We draw attention to some interesting complexities in the data in Table 6. First,Region 10 is extremely bright throughout the MIR, and especially so at 24 µ m (compareFigures 10 and 13); it has the largest (24 µ m)/(11.3 µ m) ratio (a > σ deviation from theaverage) and one of the highest ratios of 24 µ m radiance to the 8.6 µ m PAH band. It isluminous in the UV and coincident with a high-surface brightness optical cluster. However,its ratio of [Ne II] to PAH radiance is average. Region 16 is also luminous at 24 µ m, butcontains no UV or optical counterpart and again has average [Ne II] emission line ratioscompared to the PAH features. Finally, Region 19 is luminous at all wavelengths (see Fig-ure 10) and has the largest [Ne II] to PAH ratio in NGC 55. Thus, although the conditionswithin individual regions in NGC 55 are diverse, we find no statistically significant variationsof PAH/PAH radiance ratios in the observed regions of NGC 55. We examine four 52.5 pc radius apertures in NGC 3109. The low S/N ratio throughoutmuch of the observed region limits our exploration to only those areas that are comparativelyIR-bright (see Figure 11). Negligible scatter is seen in the PAH/PAH radiance ratios for theseregions, as seen in Table 7; only one aperture deviates from the weighted mean at the 2 σ significance level (the (6.2 µ m)/(11.3 µ m) ratio for Region 4). This is in marked contrastto the very significant variations seen in the ratio of 24 µ m to PAH radiance in these fourregions of the galaxy. While the weighted mean in Table 8 is affected by the difference inthe errorbars over the small number of apertures, it is clear that Regions 1 and 4 have verydifferent properties.An examination of Figure 11 highlights curious differences between these two regions.Region 1 contains very weak [Ne II] 12.8 µ m emission; it clearly harbors an embedded IRsource, since it is has no associated UV emission, is fairly weak in the IRAC 8 µ m band, butvery luminous at 24 µ m. This suggests a cool thermal dust component at this location. Takentogether, these properties suggest that the radiation field strength is low at the location of 19 –Region 1 and that the PAHs present have a significant neutral component.Regions 3 and 4 are in close physical proximity and present an interesting local envi-ronment with significant variations over small physical scales. Region 4 is very bright in theIRAC 8 µ m band and is coincident with a high surface brightness optical cluster. Region 3is much fainter at 8 µ m, yet it has the largest [Ne II]/(PAH) ratios of any of the regions inNGC 3109.Despite these dramatic changes in environmental properties, the ratios of PAH featurespresented in Table 7 are statistically indistinguishable. Even a cursory examination of thepanels in Figure 11 highlights the complexity of the relationship between PAH emissionand local properties in galaxies (e.g., slope of the local infrared spectral energy distribution(SED), UV intensity, infrared emission line radiance, etc.). The data for NGC 3109, whileonly available for selected regions, provides support for the hypothesis of constant PAH ratioswithin dwarf galaxies. We examine ten 52.5 pc radius apertures within the observed region of IC 5152. Theextracted PAH/PAH radiance ratios are presented in Table 9. Of the 50 ratios shown inthat table, only 1 (the (7.7 µ m)/(6.2 µ m) ratio for Region 5) shows a deviance from theweighted mean at even the 2 σ level. Similarly, as shown in Table 10, there are few variationsof the ratios of [Ne II] to PAH emission, or of the ratios of 24 µ m to PAH emission, that arestatistically significant.The constancy of these ratios again arises from a wide variety of local conditions withinIC 5152. In the most extreme example in our limited sample, the IR, optical, and UV mor-phologies are decidedly different in IC 5152. Most of the UV clusters have associated IRemission, but the converse in not true; a significant IR component exists in this systemthat has no associated UV, optical or near-infrared emission. Similarly, the dominant clus-ter in the UV, optical and near-infrared has very little associated PAH or dust emission.This dichotomy has been seen previously in nearby dwarfs (e.g., Cannon et al. 2006) and issuggestive of a substantial embedded star-forming population in IC 5152.Considering the regions with extracted radiances, Region 10 contains the brightest 24 µ m source in the galaxy; it is coincident with a UV cluster and is bright in the 8 µ m and11.3 µ m PAH bands, as well as luminous in the [Ne II] emission line. While its 24 µ m toPAH ratios are high, the ratios of the PAH emission features themselves are indistinguishablefrom other regions in IC 5152. Region 8 has high surface brightnesses in each of the panels 20 –of Figure 12, including in the [Ne II] emission line. Again, its PAH/PAH ratios appear to beaverage compared to others within the galaxy. At the opposite extreme is Region 6, whichsamples a region of low IR surface brightness. While diffuse UV emission is coincident withthis aperture, its PAH/PAH radiance ratios again appear normal.Taken as a whole, the apertures in IC 5152 lead to the same conclusions as those forNGC 55 and NGC 3109. While substantial local environmental changes occur within thedwarf galaxies of this limited sample, those changes do not produce measurable differencesin the intensities of the major PAH bands. The radiance ratios of these prominent MIRfeatures appear to be insensitive to a wide variety of local environmental factors.
5. Conclusions
Three nearby (D < ∼ ⊙ ≤ Z ≤
49% Z ⊙ ) dwarf galaxies(NGC 55, NGC 3109, and IC 5152) were observed using the Spitzer /IRS Short-Low modulein spectral mapping mode. We examine the resulting spectra on both global (integrated overthe entire field of view, which targets the regions of high IR surface brightness) and spatiallyresolved scales. These data allow us to study the relative intensities of the four prominentPAH bands at 6.2 µ m, 7.7 µ m, 8.6 µ m and 11.3 µ m.Investigation of the global properties of our galaxies reveals the complex interactionsthat influence the intensity of PAH emission in galaxies. While metallicity does correlatewith PAH emission (with a transition metallicity occurring at ∼
25% Z ⊙ ), our examination ofthese three galaxies provides evidence that other global parameters (e.g., mass, metallicity,current SFR) also affect the nature of PAH emission.Over the full area of our spectral maps, the relative ratios of most PAH bands vary by afactor of 3-4 across our sample. The exception is the ratio of (8.6 µ m)/(11.3 µ m) radiances,which is constant on the global level for all three systems. Given that the radiation fieldstrength shows considerable variation across this sample (with the comparatively metal-richISM of NGC 55 showing widespread [Ne II] emission, and the metal-poor ISM of NGC 3109having only a few isolated regions of ionized gas), we interpret the constant value of the (8.6 µ m)/(11.3 µ m) PAH ratio as evidence for a similar PAH size distribution across our sample.It then follows that if the PAH size distribution is similar across our sample, that the grainsize distribution must be fairly long-lived and stable.We compare the global PAH radiance ratios in our sample galaxies with a collectionof similar measurements that probe a wide variety of sources from Smith et al. (2007),Galliano et al. (2008), Gordon et al. (2008), and Hunt et al. (2010). From this comparison, 21 –we see evidence for a similar PAH size distribution across the range of objects explored,which includes spirals, AGN, dwarfs, and H II regions. The ionization indices vary signifi-cantly within these different sources. Taken together, these properties suggest that the localenvironment has a greater impact on the intensity of PAH emission than does the metallicityor the history of PAH formation/destruction in a given galaxy or region.We also examine PAH emission on a spatially resolved basis by extracting maps ofPAH features and emission lines in our sample galaxies. Examining both individual pixels(in regions of high S/N) and apertures of physical radius 52.5 pc, we find that the (7.7 µ m)/(11.3 µ m) ratio (a probe of ionization index or radiation field strength) varies by afactor of ∼ µ m)/(6.2 µ m)ratio as a diagnostic) shows only very minor variations over this range. This agrees with theinterpretations found for other galaxies and suggests that there is very little change in thePAH carrier size distribution as a function of radiation field strength.Previous works have suggested that the relative intensities of the main PAH featuresare essentially constant within a given galaxy (see references in § σ level, and most are at the 2 σ level or lower). In contrast to thediverse environmental variations seen between regions in our sample, the PAH band ratiosare constant within each galaxy.The apparently simple conclusions of this work mask a great deal of complexity in thecanonically “simple” ISM of dwarf galaxies. The uniformity of metal abundance in thesesystems is well-documented but has not yet been explained. The present work suggeststhat this uniformity also extends to the properties of the carriers of the PAH bands. Whenexamining each of these systems, we find a remarkably wide variety of physical conditions:some star formation regions are UV-bright while others are deeply embedded; some regionshave widespread ionizing photons while others are apparently quiescent; some regions aredominated by the red stellar continuum while others have SEDs that rise steeply towardthe far-IR. In response to these diverse local conditions, the intensity of the PAH emissionfeatures do in fact change markedly, in line with expectations based on the strength of thelocal radiation field and the metallicity. However, the relative intensities of these PAH bandsappear to be strikingly uniform over these variable local conditions.This work is based on observations made with the Spitzer Space Telescope , which isoperated by the Jet Propulsion Laboratory, California Institute of Technology, under a con- 22 –tract with NASA. Support for this work was provided by NASA through contract 1321212,issued by JPL/Caltech to J.M.C. at Macalester College. RDG was supported in part byNASA through contracts 1256406 and 1215746 issued by JPL/Caltech to the University ofMinnesota. This research has made use of the NASA/IPAC Extragalactic Database (NED)which is operated by the Jet Propulsion Laboratory, California Institute of Technology, undercontract with the National Aeronautics and Space Administration, and NASA’s AstrophysicsData System. This publication has made use of data products from the Two Micron AllSky Survey, which is a joint project of the University of Massachusetts and the InfraredProcessing and Analysis Center/California Institute of Technology, funded by the NationalAeronautics and Space Administration and the National Science Foundation. We would liketo acknowledge Daniel A. Dale, J.D. Smith, Thomas Varberg, and the
Spitzer
Science Centerfor helpful discussions and support. Finally, we thank the anonymous referee for a carefuland insightful report that improved this manuscript. 23 –
REFERENCES
Allamandola, L. J., Hudgins, D. M., & Sandford, S. A. 1999, ApJ, 511, L115Allamandola, L. J., Tielens, A. G. G. M., & Barker, J. R. 1989, ApJS, 71, 733Bakes, E. L. O., Tielens, A. G. G. M., & Bauschlicher, C. W., Jr. 2001, ApJ, 556, 501Bauschlicher, C. W., Jr. 2002, ApJ, 564, 782Bauschlicher, C. W., Peeters, E., & Allamandola, L. J. 2009, ApJ, 697, 311Bauschlicher, C. W., Jr., Peeters, E., & Allamandola, L. J. 2008, ApJ, 678, 316Cannon, J. M., et al. 2006, ApJ, 647, 293Cannon, J. M., et al. 2006, ApJ, 652, 1170Compi`egne, M., Abergel, A., Verstraete, L., Reach, W. T., Habart, E., Smith, J. D.,Boulanger, F., & Joblin, C. 2007, A&A, 471, 205Cox, A. N. 2000, Allen’s Astrophysical QuantitiesDale, D. A., et al. 2009, ApJ, 693, 1821Draine, B. T., & Li, A. 2007, ApJ, 657, 810Engelbracht, C. W., Gordon, K. D., Rieke, G. H., Werner, M. W., Dale, D. A., & Latter,W. B. 2005, ApJ, 628, L29Engelbracht, C. W., Rieke, G. H., Gordon, K. D., Smith, J.-D. T., Werner, M. W., Mous-takas, J., Willmer, C. N. A., & Vanzi, L. 2008, ApJ, 678, 804Galliano, F., Madden, S. C., Tielens, A. G. G. M., Peeters, E., & Jones, A. P. 2008, ApJ,679, 310Giard, M., Lamarre, J. M., Pajot, F., & Serra, G. 1994, A&A, 286, 203Gordon, K. D., Engelbracht, C. W., Rieke, G. H., Misselt, K. A., Smith, J.-D. T., & Kenni-cutt, R. C., Jr. 2008, ApJ, 682, 336Helou, G., Lu, N. Y., Werner, M. W., Malhotra, S., & Silbermann, N. 2000, ApJ, 532, L21Helou, G., Malhotra, S., Hollenbach, D. J., Dale, D. A., & Contursi, A. 2001, ApJ, 548, L73Helou, G., et al. 2004, ApJS, 154, 253 24 –Hogg, D. W., Tremonti, C. A., Blanton, M. R., Finkbeiner, D. P., Padmanabhan, N., Quin-tero, A. D., Schlegel, D. J., & Wherry, N. 2005, ApJ, 624, 162Houck, J. R., et al. 2004, ApJS, 154, 18Hunt, L. K., Thuan, T. X., Izotov, Y. I., & Sauvage, M. 2010, ApJ, 712, 164Jackson, D. C., Cannon, J. M., Skillman, E. D., Lee, H., Gehrz, R. D., Woodward, C. E., &Polomski, E. 2006, ApJ, 646, 192Jarrett, T. H., Chester, T., Cutri, R., Schneider, S. E., & Huchra, J. P. 2003, AJ, 125, 525Kennicutt, R. C., Jr., et al. 2003, PASP, 115, 928Kim, H.-S., & Saykally, R. J. 2002, ApJS, 143, 455Kobulnicky, H. A., & Skillman, E. D. 1996, ApJ, 471, 211Kobulnicky, H. A., & Skillman, E. D. 1997, ApJ, 489, 636Lee, H., Grebel, E. K., & Hodge, P. W. 2003, A&A, 401, 141Lee, H., McCall, M. L., Kingsburgh, R. L., Ross, R., & Stevenson, C. C. 2003, AJ, 125, 146Lee, J. C., et al. 2009, ApJ, 706, 599Leger, A., & Puget, J. L. 1984, A&A, 137, L5Leroy, A., Bolatto, A. D., Simon, J. D., & Blitz, L. 2005, ApJ, 625, 763Mateo, M. L. 1998, ARA&A, 36, 435O’Halloran, B., Satyapal, S., & Dudik, R. P. 2006, ApJ, 641, 795Ricks, A. M., Douberly, G. E., & Duncan, M. A. 2009, ApJ, 702, 301Rosenberg, J. L., Ashby, M. L. N., Salzer, J. J., & Huang, J.-S. 2006, ApJ, 636, 742Sanders, D. B., & Mirabel, I. F. 1996, ARA&A, 34, 749Sanders, D. B., Soifer, B. T., Elias, J. H., Madore, B. F., Matthews, K., Neugebauer, G., &Scoville, N. Z. 1988, ApJ, 325, 74Skrutskie, M. F., et al. 2006, AJ, 131, 1163Smith, J. D. T., et al. 2007b, PASP, 119, 1133 25 –Smith, J. D. T., et al. 2007a, ApJ, 656, 770Taylor, C. L., Kobulnicky, H. A., & Skillman, E. D. 1998, AJ, 116, 2746Tenorio-Tagle, G., & Bodenheimer, P. 1988, ARA&A, 26, 145Tielens, A. G. G. M. 2008, ARA&A, 46, 289Webster, B. L., & Smith, M. G. 1983, MNRAS, 204, 743Wu, R., Hogg, D. W., & Moustakas, J. 2010, ApJ, submiited (arxiv/0907.1783)
This preprint was prepared with the AAS L A TEX macros v5.2.
Table 1. Galaxy Sample PropertiesTarget Type a M B Distance b c Mass Current Star Diffuse 8 µ mFormation Rate Flux Density d (mag) (Mpc) (dex) (10 M ⊙ ) (M ⊙ yr − ) (Jy)NGC 55 IrrIV -17.5 2.17 8.4 ± × − ± × − ± × − a Type, M B , and mass (calculated from central or rotational velocity, as available) taken from Mateo (1998) b Distance and SFR taken from Lee et al. (2009). SFR based on GALEX FUV integrated photometry andcorrected for extinction based on A
F UV = 7.9E(B – V), except for NGC 3109, where UV attenuation wasunavailable, and H α attenuation (A Hα = 2.5E(B – V), scaled to a factor of 1.8, was used instead c Metallicities taken from Webster & Smith (1983) for NGC 55, Lee et al. (2003a) for IC 5152, and Lee et al.(2003b) for NGC 3109. d Diffuse 8 µ m flux density taken from Jackson et al. (2006) 27 –Table 2. Global Feature Radiances From PAHFIT a Wavelength NGC 55 NGC 3109 IC 51526.2 µ m PAH 76 . ± . . ± . . ± . µ m PAH 226 . ± . . ± . . ± . µ m PAH 43 . ± . . ± . . ± . µ m 7 . ± . . ± . . ± . µ m PAH 73 . ± . . ± . . ± . µ m PAH 29 . ± . . ± . . ± . µ m 8 . ± . . ± . . ± . a Units of 10 − W m − sr − . Table 3. Global PAH Ratios a Target (6.2 µ m)/(11.3 µ m) (7.7 µ m)/(11.3 µ m) (8.6 µ m)/(11.3 µ m) (7.7 µ m)/(6.2 µ m) (8.6 µ m)/(6.2 µ m)NGC 55 1.04 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a Values are ratios of radiances measured in W m − sr − . Table 4. Spectral Extraction Wavelength Regions a Target 6.2 µ m 7.7 µ m 8.6 µ m Broad 8 µ m 11.3 µ m [Ne II] 12.8 µ mNGC 55 5.986-6.513 7.536-7.882 8.403-8.838 7.288-8.900 10.949-11.694 12.625-12.998NGC 3109 6.141-6.358 7.443-7.782 8.217-8.900 7.288-8.900 10.763-11.570 12.688-12.936IC 5152 6.110-6.420 7.474-7.882 8.403-8.838 7.288-8.900 11.073-11.446 12.688-12.998 a Spectral extraction regions are given in units of µ m. Table 5. PAH/PAH Ratios for NGC 55 Extraction Regions a Region (6.2 µ m)/(11.3 µ m) (7.7 µ m)/(11.3 µ m) (8.6 µ m)/(11.3 µ m) (7.7 µ m)/(6.2 µ m) (8.6 µ m)/(6.2 µ m)Region 1 0.93 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a Values are ratios of radiances measured in of W m − sr − . Table 6. Selected Ratios for NGC 55 Extraction Regions a Region [Ne II]/(11.3 µ m) [Ne II]/(8.6 µ m) (MIPS 24 µ m)/(11.3 µ m) (MIPS 24 µ m)/(8.6 µ m)Region 1 0.03 ± b ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a Values are ratios of radiances measured in W m − sr − . b For some regions, errors were below .005. In these cases, the error was rounded up to 0.01.
Table 7. PAH/PAH Ratios for NGC 3109 Extraction Regions a Region (6.2 µ m)/(11.3 µ m) (7.7 µ m)/(11.3 µ m) (8.6 µ m)/(11.3 µ m) (7.7 µ m)/(6.2 µ m) (8.6 µ m)/(6.2 µ m)Region 1 2.45 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a Values are ratios of radiances measured in W m − sr − . Table 8. Selected Ratios for NGC 3109 Extraction Regions a Region [Ne II]/(11.3 µ m) [Ne II]/(8.6 µ m) (MIPS 24 µ m)/(11.3 µ m) (MIPS 24 µ m)/(8.6 µ m)Region 1 0.02 ± ± ± ± ± ± ±
609 808 ± ± ± ± ± ± ± ±
257 533 ± ± ± ± ± a Values are ratios of radiances measured in of W m − sr − . Table 9. PAH/PAH Ratios for IC 5152 Extraction Regions a Region (6.2 µ m)/(11.3 µ m) (7.7 µ m)/(11.3 µ m) (8.6 µ m)/(11.3 µ m) (7.7 µ m)/(6.2 µ m) (8.6 µ m)/(6.2 µ m)Region 1 1.18 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a Values are ratios of radiances measured in W m − sr − . Table 10. Selected Ratios for IC 5152 Extraction Regions a Region [Ne II]/(11.3 µ m) [Ne II]/(8.6 µ m) (MIPS 24 µ m)/(11.3 µ m) (MIPS 24 µ m)/(8.6 µ m)Region 1 0.13 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a Values are ratios of radiances measured in W m − sr − . 38 – D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 03 00 02 55 50 45 40 35 30 25-51 15 003016 003017 003018 003019 0030 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)00 15 15 00 14 45 30-39 09101112131415 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)10 03 30 15 00 02 45 30-26 0506070809101112131415
NGC 55NGC 3109IC 5152
Fig. 1.— IRAC 8 µ m images are shown for each of the target galaxies with the regionobserved by IRS boxed in red. Images have arbitrary scaling to highlight features. Galacticforeground stars are prominent in NGC 3109 and IC 5152 (see in particular the object nearIC 5152, α ≈ δ ≈ -51:16:55). 39 –Fig. 2.— Global IRS spectrum for NGC 55, with prominent emission lines and PAH featureslabeled. Errors are shown in red, and are largest in the region of overlap between the SL1and SL2 slits, between 7.4 µ m and 7.7 µ m. 40 –Fig. 3.— Global IRS spectrum for NGC 3109, with prominent emission lines and PAHfeatures labeled. Errors are shown in red, and are largest in the region of overlap betweenthe SL1 and SL2 slits, between 7.4 µ m and 7.7 µ m. 41 –Fig. 4.— Global IRS spectrum for IC 5152, with prominent emission lines and PAH featureslabeled. Errors are shown in red, and are largest in the region of overlap between the SL1and SL2 slits, between 7.4 µ m and 7.7 µ m. 42 –Fig. 5.— Decomposed global IRS spectrum for NGC 55 produced by PAHFIT. The red linesrepresent thermal dust continuum components at various temperatures, the dashed magentaline the stellar continuum, the gray line the total (dust + stellar) continuum. The blue linerepresents the PAH features, the green line the unresolved atomic and molecular spectrallines, and the black line the full fitted model. 43 –Fig. 6.— Decomposed global IRS spectrum for NGC 3109 produced by PAHFIT. The redlines represent thermal dust continuum components at various temperatures, the dashedmagenta line the stellar continuum, the gray line the total (dust + stellar) continuum. Theblue line represents the PAH features, the green line the unresolved atomic and molecularspectral lines, and the black line the full fitted model. 44 –Fig. 7.— Decomposed global IRS spectrum for IC 5152 produced by PAHFIT. The red linesrepresent thermal dust continuum components at various temperatures, the dashed magentaline the stellar continuum, the gray line the total (dust + stellar) continuum. The blue linerepresents the PAH features, the green line the unresolved atomic and molecular spectrallines, and the black line the full fitted model. 45 –Fig. 8.— Global comparison of PAH/PAH radiance ratios for individual galaxies and se-lected massive star formation regions. The four systems with the lowest (8.6 µ m)/(6.2 µ m)ratios are NGC 4725, NGC 1291, IRAS 23128-5919, and NGC 253, in order of increasing (7.7 µ m)/(11.3 µ m) ratio. This plot is designed to show contributions from larger PAH moleculesincreasing up the y axis, and contributions from ionized PAH moecules increasing along thex axis. The galaxies from our sample fall within the range established by the larger samples. 46 – (b)(a) Fig. 9.— Comparison of PAH/PAH radiance ratios with gas-phase oxygen abundances indwarf galaxies and in individual H II regions in M 101. No uncertainties in metallicity wereavailable for the Hunt et al. (2010) or Galliano et al. (2008) data. Only very weak trendswith metallicity are apparent from this plot. 47 – D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)00 15 04 02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00 (l) D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)00 15 04 02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)00 15 04 02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00
MIPS 24 m[Ne II] 12.8 m IRS 8 m(k) D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)00 15 04 02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)00 15 04 02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00 (i) D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00 (h) D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)00 15 04 02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000) 15 04 02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00 (j) (f)2MASS K S D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00
IRAC 4.5 m (g) µ (e) µ (c)DSS R−Band µ µµ D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000) 15 04 02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00 µ (b)2MASS J GALEX NUV D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)00 15 04 02 00 14 58 56 54 52 50 48-39 10 3011 003012 003013 003014 00 (d)GALEX FUV (a)
Fig. 10.— Multiwavelength images of the region of NGC 55 covered by the IRS spectral map,ordered by increasing wavelength from (a) to (l) as labeled. Contours of 11.3 µ m surfacebrightness are overlaid in each panel at the 1.5, 3, 4.5 and 6 MJy sr − levels. 48 – D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)10 03 06 04 02 00 02 58 56-26 08 15304509 0015304510 00 (l)MIPS 24 m D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)10 03 06 04 02 00 02 58 56-26 08 15304509 0015304510 00 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)10 03 06 04 02 00 02 58 56-26 08 15304509 0015304510 00
IRS 8 m (i)(k) D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)10 03 06 04 02 00 02 58 56-26 08 15304509 0015304510 00 [NeII] 12.8 m D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)04 02 00 02 58 56-26 08 15304509 0015304510 00 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)10 03 06 04 02 00 02 58 56-26 08 15304509 0015304510 00 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)10 03 06 04 02 00 02 58 56-26 08 15304509 0015304510 00 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)04 02 00 02 58 56-26 08 15304509 0015304510 00 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)04 02 00 02 58 56-26 08 15304509 0015304510 00 S (j) (h) (f) µ µ D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)04 02 00 02 58 56-26 08 15304509 0015304510 00 (g) µ µ
IRAC 4.5 m µµ D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)04 02 00 02 58 56-26 08 15304509 0015304510 00
GALEX NUV D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)10 03 06 04 02 00 02 58 56-26 08 15304509 0015304510 00 (b)2MASS J (d)(a)GALEX FUV
Fig. 11.— Multiwavelength images of the region of NGC 3109 covered by the IRS spectralmap, ordered by increasing wavelength from (a) to (l) as labeled. Contours of 11.3 µ msurface brightness are overlaid in each panel at the 0.5, 1.0, 1.5 and 2.0 MJy sr − levels. 49 – D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045
MIPS 24 m D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045 (l) D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045 (k) IRS 8 m[NeII] 12.8 m (i)(h) D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045 S (f)(j) (c)(g) µ D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045 D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045 (e)IRAC 4.5 m 2MASS H DSS R−Band µ µµ µ µ (d) GALEX NUV2MASS J (b) D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 48 46 44 42 40 38-51 16 4517 0015304518 00153045
GALEX FUV (a)
Fig. 12.— Multiwavelength images of the region of IC 5152 covered by the IRS spectral map,ordered by increasing wavelength from (a) to (l) as labeled. Contours of 11.3 µ m surfacebrightness are overlaid in each panel at the 1.0, 1.5, 2.0, 2.5 and 3.0 MJy sr − levels. Theextremely luminous source in the upper right is a Galactic foreground star. 50 – D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)00 15 04 02 00 14 58 56 54 52 50-39 11 15304512 0015304513 001530 8 5611 91317 1214 115 718 1619 10 34 2
NGC 55
Fig. 13.— Full, integrated IRS map of NGC 55 with extraction regions labeled. Regionshave a physical radius of 52.5 pc. 51 – D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)10 03 05 04 03 02 01 00 02 59 58 57-26 08 304509 00153045 4 123
NGC 3109
Fig. 14.— Full, integrated IRS maps of NGC 3109 with extraction regions labeled. Regionshave a physical radius of 52.5 pc. 52 – D E C L I NA T I O N ( J2000 ) RIGHT ASCENSION (J2000)22 02 46 44 42 40 38-51 17 15304518 0015 10 9 8 3 267 5 4 1
IC 5152
Fig. 15.— Full, integrated IRS map of IC 5152 with extraction regions labeled. Regions havea physical radius of 52.5 pc. 53 –Fig. 16.— PAH radiance ratios within each galaxy. Pixel by pixel comparisons are shown inblack, trimmed to show data above the 3 σ level (note that in NGC 3109, which has generallylow S/N, this causes an offset between the pixel cloud and the integrated regions, whichinclude all pixels regardless of S/N level; see discussion in §§
Fig. 15.— Full, integrated IRS map of IC 5152 with extraction regions labeled. Regions havea physical radius of 52.5 pc. 53 –Fig. 16.— PAH radiance ratios within each galaxy. Pixel by pixel comparisons are shown inblack, trimmed to show data above the 3 σ level (note that in NGC 3109, which has generallylow S/N, this causes an offset between the pixel cloud and the integrated regions, whichinclude all pixels regardless of S/N level; see discussion in §§ σ level (note that in NGC 3109, which has generallylow S/N, this causes an offset between the pixel cloud and the integrated regions, whichinclude all pixels regardless of S/N level; see discussion in §§