1: \documentclass[preprint2, longabstract]{emulateapj}
2: %\usepackage{graphics,amsmath, amsthm, amssymb,upquote}
3: \def\hii{\relax \ifmmode {\rm H\,{\sc ii}}\else H\,{\sc ii}\fi}
4:
5: \shorttitle{Tracing the [FeII]/[NeII] ratio within star forming dwarf galaxies: a \emph{Spitzer} IRS archival survey}
6:
7: \shortauthors{O'Halloran et al.}
8:
9:
10: \begin{document}
11:
12: \title{Tracing the [FeII]/[NeII] ratio and its relationship with other ISM indicators within star forming dwarf galaxies: a \emph{Spitzer} IRS archival study.}
13:
14:
15: \author{B. O'Halloran\altaffilmark{1}, S. C. Madden\altaffilmark{2}, N. P. Abel\altaffilmark{3}}
16:
17: \email{boh@physics.gmu.edu}
18:
19: \altaffiltext{1}{Dept. of Physics \& Astronomy, George Mason University, Fairfax, VA 22030, USA}
20:
21: \altaffiltext{2}{Service d'Astrophysique, CEA, Saclay, Orme des Merisiers 91191, Gif-sur-Yvette, France}
22:
23: \altaffiltext{3}{Department of Physics, University of Cincinnati, Cincinnati, OH 45221, USA}
24:
25:
26:
27:
28:
29: \begin{abstract}
30:
31: Archival \emph{Spitzer} observations of 41 starburst galaxies that span a wide range in metallicity reveal for the first time a correlation between
32: the [FeII]/[NeII] 26.0/12.8 $\mu$m ratio and the electron gas density as traced by the 18.7/33.4 $\mu$m [SIII] ratio, with the [FeII]/[NeII] ratio
33: decreasing with increasing gas density. The correlations of the [FeII]/[NeII] ratio, the PAH peak to continuum strength \& metallicity found in an
34: earlier paper were confirmed for a larger sample. We also find a strong correlation between the gas density and the PAH peak to continuum strength.
35: Using shock and photoionization models, we see the driver of the observed [FeII]/[NeII] ratios is metallicity. The majority of [FeII] emission in low
36: metallicity galaxies may be shock-derived, whilst at high metallicity, the [FeII] emission may be instead dominated by contributions from \hii\ and
37: in particular from dense PDR regions. However, the observed [FeII]/[NeII] ratios may instead be following a metallicity-abundance relationship, with
38: iron being less depleted onto grains in low metallicity galaxies - a result that would have profound implications for the use of iron emission lines
39: as unambiguous tracers of shocks.
40:
41:
42: \end{abstract}
43:
44:
45: \keywords{galaxies: starburst - galaxies: stellar content - ISM: lines and bands - infrared: galaxies }
46:
47:
48: \section{Introduction}
49:
50:
51:
52: The presence of massive stars within starbursts undoubtedly plays a huge role in determining the physical conditions within the local interstellar
53: medium (ISM). High-mass (M$_{init}$ $\geq$ 8M$_{\odot}$) stars formed within typical starbursts drastically affect the dynamics of the surrounding
54: ISM, through not only the release of ionizing photons which destroy molecular material, but also via supernovae (SNe) which provide thermal and
55: kinetic energy input into the ISM. The effects of photoionization by high-mass stars on the observed dearth of PAHs have recently been investigated
56: \citep{mad06,wu06}, while in \cite{oha06} (hereafter Paper I), we examined a sample of 18 galaxies of varying metallicity (from 1/50th to super
57: solar) with high star formation rates in order to determine whether supernova-driven shocks do indeed play a role in the PAH deficit in low
58: metallicity environments. If we consider the ratio of the 26 $\mu$m [FeII] line and the 12.8 $\mu$m [NeII] line as a tracer of the strength of
59: supernova shocks, we found a strong anti-correlation suggesting that strong supernova-driven shocks are indeed present within low metallicity
60: galaxies. Furthermore, the PAH deficit within these objects may indeed be linked to the presence and strength of these shocks. However, it has not as
61: yet been conclusively proved that shocks are the dominant process behind the PAH deficit. As already noted, photoionization and perhaps delayed
62: injection of dust into the ISM \citep{galliano08} each make their own contributions to the PAH deficit, but as yet, it remains unclear as to which of
63: these three processes is the most dominant. Additionally, is the relationship between the [FeII]/[NeII] ratio and the PAH emission, as indicated in
64: Paper I, really probing a causal effect? It could very well be that this relationship may be coincidental, or that physical or selection effects may
65: be the primary drivers for the observed relationship. It would therefore be advisable to further explore the nature of the ISM within such
66: star-forming environments using additional mid-IR probes in order to expand upon our understanding of the [FeII]/[NeII] ratio, and by extension, its
67: relationship with key ISM indicators such as PAH strength and metallicity.
68:
69:
70:
71:
72:
73: \section{Observations and data analysis}
74:
75: To accomplish this, we have expanded our sample from the 18 objects presented in Paper I by including archival IRS observations of 23 additional
76: objects, bringing the sample total to 41. The full list of targets is given in Table 1, and listed by increasing metallicity. These galaxies range in
77: metallicity from extremely low (such as I Zw 18, with $ \it Z/Z_{\odot}$= 1/50) to super-solar metallicity ($\geq 1~\it Z_{\odot}$) galaxies such as
78: NGC 7714. To differentiate between low and high-metallicity galaxies, we use a metallicity (12 + log [O/H]) cut-off value of 8.85 - just less than
79: solar, as per Paper I. The galaxies range widely in morphology from blue compact dwarfs such as I Zw 18 to spirals such as NGC 7714 and IC 342. None
80: of the galaxies in our sample are known to harbour AGNs. This is important, as PAH destruction can occur close to an AGN due to the hard ionization
81: environment (e.g. Sturm et al. 2000). In addition [FeII] emission can be elevated in galaxies harbouring AGN. One possible exception is NGC 7714,
82: which is optically classified as a LINER \citep{tho02}. However, the lack of [NeV] emission at 14 and 24 $\mu$m and the absence of any evidence
83: for an obscured AGN by recent Chandra imaging \citep{bran04,smi05} strongly suggests that NGC 7714 is a pure starburst and it is therefore included
84: in our sample.
85:
86:
87:
88: We extracted low and high resolution archival spectral data from the Short-Low (SL) (5.2 - 14.5 $\mu$m), Short-High (SH) (9.9 - 19.6 $\mu$m)
89: and Long-High (LH) (18.7 - 37.2 $\mu$m) modules of the {\it Spitzer} Infrared Spectrograph (IRS). The datasets were derived from a number of
90: {\it Spitzer} Legacy, GO and GTO programs released to the {\it Spitzer} Data Archive, and consisted of either spectral mapping or staring
91: observations. We obtained fluxes for the nuclear positions only from the mapping observations. All the staring observations were centered on the
92: galaxy's nucleus. The data were preprocessed by the {\it Spitzer} Science Center (SSC) data reduction pipeline version 15.3\footnote[4]{{\it
93: Spitzer} Observers Manual, URL: http://ssc.spitzer.caltech.edu/documents/som/} before being downloaded. Further processing was done within the IRS
94: BCD-level data reduction package {\it SPICE}, v 2.0.1\footnote[5]{URL: http://ssc.spitzer.caltech.edu/postbcd/spice.html}. Both the low and high
95: resolution spectra were extracted by {\it SPICE} using the full-aperture extraction method. As the [SIII] emission is extended beyond the extent of
96: the high resolution slits in particular, we applied the ALCF and SLCF extended source corrections to the spectra as part of the \emph{SPICE}
97: extraction\footnote[6]{URL: http://ssc.spitzer.caltech.edu/postbcd/doc/spice.pdf}. The slit for the SH and LH modules is too small for background
98: subtraction to take place and separate SH or LH background observations do not exist for any of the galaxies in this sample. For the SL module,
99: background subtraction was done using either a designated background file when available or by subtracting the data from one nod position from the
100: other along the slit. In some cases, the slit was enveloped in the host galaxy and background subtraction could not take place. For both high and low
101: resolution spectra, the ends of each order where the noise increases significantly were manually clipped, as were hot pixels.
102:
103: For the high resolution observations, we required matched extractions in terms of angular extent from both the SH \& LH slits, in order to accurately
104: derive densities based upon the 18 and 33 $\mu$m [SIII] lines. The ionization potential of [SIII] is $\sim$35 eV, with the [SIII] emission arising
105: from gas ionized by young stars for our sample. With a relatively low ionization potential, [SIII] emission may be quite easily extended in spatial
106: extent beyond the nuclear single-slit pointings for SH and LH. Indeed, comparisons of ISO and IRS data \citep{dudik07} show [SIII] emission within
107: star-forming AGN that extends beyond the size of both the SH and LH slits. Given that [SIII] emission is extended and we would expect to see a higher
108: flux in the larger LH slit, any derived [SIII] line ratio will be artificially depressed. To avoid such aperture effects between the SH and LH slits,
109: we scaled the LH flux by multiplying it by the ratio of the SH/LH slit angular sizes.
110:
111:
112:
113: As a number of the sample starbursts at high metallicity are known to be dusty systems with large obscuration, the extracted line fluxes were then
114: corrected for extinction for the sample as a whole. The fluxes and statistical errors, plus the derived line ratios, are presented in Tables 1 and 2
115: respectively, obtained using the IDL-based analysis package \emph{SMART}, v.6.2.6 \citep{hig04}. In all cases detections were defined when the line
116: flux was at least 3$\sigma$. For the SL, SH and LH modules the spectral resolution was $\lambda$/$\delta\lambda$ 60-127 and $\sim$600 respectively,
117: with FWHM in the order of 1.9 x 10$^{-1}$ $\mu$m for the PAH feature in the low-res spectrum, and 2-4 x 10$^{-2}$ $\mu$m for the fine structure lines
118: in the high resolution spectra. For the absolute photometric flux uncertainty we conservatively adopt 15\%, based on the assessed values given by the
119: Spitzer Science Center (SSC) over the lifetime of the mission. This error is calculated from multiple observations of various standard stars
120: throughout the Spitzer mission by the SSC. The dominant component of the total error arises from the uncertainty at mid-IR wavelengths in the stellar
121: models used in calibration and is systematic rather than Gaussian in nature. The 6.2 $\mu$m PAH strength was determined according to the prescription
122: adopted in a number of previous studies \citep{rig99,fors04}. While the 7.7 and 8.6 $\mu$m PAH lines are stronger, the 8.6 $\mu$m line can suffer
123: extinction effects due to a potential silicate absorption feature at 9 $\mu$m, and both of these pose difficulties in determining the continuum
124: level. For this work, the continuum was determined at the center of the 6.2 $\mu$m feature, using a first order linear fit to the 5.5 and 11.5 $\mu$m
125: bandpass. The PAH strength was then calculated as the ratio of the 6.2 $\mu$m feature peak intensity to the underlying continuum. We assume a
126: Gaussian fit while fitting the PAH and fine structure lines with \emph{SMART}.
127:
128:
129: \section{Results}
130:
131: In Fig 1, we plot [FeII]/[NeII] vs the 6.2 $\mu$m PAH peak to continuum ratio to check if the relationship between the [FeII]/[NeII] ratio and the
132: PAH peak to continuum strength first noted in Paper I, using the 6.2 $\mu$m PAH feature instead of the 7.7 $\mu$m, as one can perform continuum
133: fitting with a higher degree of confidence with the 6.2 $\mu$m feature, still holds for the expanded sample, as noted above. Employing a Spearman
134: rank correlation analysis \citep{ken76} to assess the statistical significance of this trend yields a correlation coefficient of r$_{s}$ of -0.705
135: and P$_{s}$ of 2.69 x 10$^{-6}$, confirming the significant anti-correlation between PAH strength and the [FeII]/[NeII] ratio as seen in Paper I. The
136: Spearman rank correlation technique has the advantage of being non-parametric, robust to outliers and does not presuppose a linear relation. We plot
137: the [FeII]/[NeII] ratio versus the metallicity in Fig.2, and we again see a strong anti-correlation (r$_{s}$ of -0.777 and P$_{s}$ of 9.36 x
138: 10$^{-7}$) between [FeII]/[NeII] and metallicity as seen in paper I.
139:
140: \begin{figure}
141: \resizebox{\columnwidth}{!} {\includegraphics{f1.eps}}
142: \caption{Plot of the [FeII]/[NeII] ratio versus the PAH 6.2 micron peak to continuum ratio for the extended sample. The observed trend confirms
143: the relationships between the [FeII]/[NeII] ratio and the PAH strength seen in Paper I.}
144: \end{figure}
145:
146: \begin{figure}
147: \resizebox{\columnwidth}{!} {\includegraphics{f2.eps}}
148:
149: \caption{Plot of the [FeII]/[NeII] ratio versus the metallicity for the extended sample. The observed trend confirms
150: the strong relationship between the [FeII]/[NeII] ratio and metallicity seen in Paper I.}
151: \end{figure}
152:
153:
154: Based upon what we have already seen from Paper I and from Figs.~1 and 2, if the [FeII]/[NeII] ratio is truly indicative of the strength of
155: supernova-driven shocks within the extent of the high resolution slits, one would expect the passage of such shocks to affect conditions within the
156: local ISM in quite a substantive manner. Shocks, in addition to removing dust and PAH from the ISM \citep{reach00,reach06}, should also be adept in
157: removing gas - one would therefore expect the propagation of intense SNe-driven shocks into the ISM of star forming regions to greatly impact on the
158: density of the gas. In order to determine how the gas density within these nuclear star forming regions corresponds with the strength of the
159: supernova-driven shocks, we require a reliable mid-IR tracer to probe the gas density. The [SIII] 18.7/33.4 $\mu$m line ratio provides such a
160: reliable mid-IR tracer of the gas density, as it is ideal for probing gas (with high critical densities) in the regions surrounding starbursts,
161: especially as the [SIII] lines are starburst dominated \citep{ver04}. This ratio is sensitive to changes in the density for the 50 $\leq$ n$_{e}$
162: $\leq$ 10$^{4}$ cm$^{-3}$, but is insensitive to changes in temperature \citep{rigo96}. In Fig. 3, we plot the logarithm of the ratio of the [SIII]
163: fluxes versus the [FeII]/[NeII] flux ratio. There is a strikingly strong trend between the two line ratios, with high [FeII]/[NeII] values
164: corresponding to lower [SIII] ratios. Using the Spearman rank correlation analysis, we get (r$_{s}$) of -0.891 between the [SIII] ratio and the
165: [FeII]/[NeII] with a probability of chance correlation (P$_{s}$) of 7.43 x 10$^{-8}$, indicating a significant anti-correlation. Interestingly, we
166: see a general decrease in the metallicity of the object with lower [SIII] ratios, corresponding to higher [FeII]/[NeII].
167:
168: \begin{figure}
169: \resizebox{\columnwidth}{!} {\includegraphics*{f3.eps}}
170:
171: \caption{Plot of the 18.7/33.4 $\mu$m [SIII] ratio as a function of the [FeII]/[NeII] ratio for the extended sample. The [SIII] ratio correlates
172: strongly with the [FeII]/[NeII] ratio, with high values of the [FeII]/[NeII] ratio corresponding to low [SIII] ratios and metallicity.}
173:
174:
175: \end{figure}
176:
177: \begin{figure}
178: \resizebox{\columnwidth}{!} {\includegraphics*{f4.eps}}
179:
180: \caption{Plot of the 18.7/33.4 $\mu$m [SIII] ratio as a function of the 6.2 $\mu$m PAH peak to continuum strength for the extended sample. The
181: [SIII] ratio correlates quite well with the PAH strength, indicating that the gas density scales with the increased strength of PAH emission.}
182:
183:
184: \end{figure}
185:
186: We again plot the logarithm of the ratio of the [SIII] fluxes in Fig. 4, but this time against the 6.2 micron PAH peak/continuum ratio. As with Fig.
187: 3, we see a strong trend, but this time with the PAH strength increasing with log [SIII] - the PAH strength is increasing with higher gas density.
188: Again running a Spearman rank correlation analysis, we get r$_{s}$ of 0.750 and P$_{s}$ of 9.64 x 10$^{-7}$, confirming a significant correlation
189: between log [SIII] and PAH strength. We again see that a general trend exists with PAH strength increasing with metallicity and increasing values of
190: log [SIII].
191:
192:
193: \section{Discussion}
194:
195: \subsection{Can we explain the observed relationships?}
196:
197: \subsubsection{Are there other factors which may be influencing the observed relationships?}
198:
199: As we have seen from Figs. 1-4, a series of relationships would seem to exist between the [FeII]/[NeII] ratio, the metallicity, the PAH strength and
200: the [SIII] ratio, with low values of the [SIII] ratio corresponding with high [FeII]/[NeII] ratios and weak/no PAH emission. However, before we press
201: ahead with more detailed investigations of the nature of these relationships, it is prudent to first check whether other factors, such as extinction
202: and aperture effects might be responsible.
203:
204: \subsubsection{Extinction}
205:
206: As we have noted previously in Section 2, we corrected the IRS fluxes for extinction, with A$_{V}$ ranging from $\sim$ 2 to 30 mag. To determine the
207: correction, we used the ratios of $\rm H\beta/H\alpha$ and $\rm Br\beta/Br\alpha$ where available from the literature. Emissivity coefficients from
208: \cite{storey95} for case B recombination with an electron temperature of 10$^{4}$ K and density of $100~\rm cm^{-3}$ were used to calculate the
209: intrinsic line ratios for the H recombination lines. We note that while the density used in these calculations is lower than that seen in some of our
210: sample, we noted little change in the intrinsic line ratios for denser environments. We adopted a \cite{draine89} extinction law above 8.5 $\mu$m,
211: and the \cite{lutz99} galactic extinction law below this point. As the observed relationships remain post correction, we conclude that extinction can
212: be ruled out as a driver for the observed correlations. For the optical [SII] lines used lines used as comparisons to the \emph{MAPPINGS} output, we
213: obtained and used from the literature extinction corrected fluxes where available.
214:
215:
216: \subsubsection{Aperture effects}
217:
218: As noted earlier, the ionization potential of [SIII] is $\sim$35 eV, with the [SIII] emission arising from gas ionized by young stars. With a
219: relatively low ionization potential, [SIII] emission may be quite easily extended in spatial extent beyond the nuclear single-slit pointings for SH
220: and LH. To compensate, we applied the SLCF and ALCF extended source corrections to the spectra. Additionally, as we are looking to derive a physical
221: quantity, namely the gas density, over a similarly sized physical area given the extent of the [SIII] emission, it is therefore crucial that we use
222: [SIII] fluxes derived from slits of equal angular size. To do this, we obtained a fraction of the LH full slit flux by multiplying it by the SH/LH
223: angular size ratio, as noted in Section 2. The final, corrected fluxes are given in Table 1.
224:
225: \subsection{Modeling the observed relationships}
226:
227: Since we can rule out extinction and aperture effects as drivers for the observed correlations, we can be confident that the correlations seen in
228: Figs. 1-4 are indeed true physical relationships between the strength of the [FeII]/[NeII] ratio and a number of ISM indicators, with the primary
229: culprit for the observed relationships being the passage of SNe-driven shocks. However, while we have focussed up to now with a supernova-derived
230: shock origin for the behaviour of the [FeII]/[NeII] ratio, it may not be the only explanation. For example, \cite{izotov06} note sign of strong
231: depletion of iron onto dust grains, and gradual destruction of those grains on a time scale of a few Myr, based on a survey of metal-poor galaxies
232: from the 3rd release of the SDSS. Such a process could undoubtedly drastically affect the nature of Fe emission, and by extension the behaviour of
233: the [FeII]/[NeII] ratio, within our sample - we may instead be probing an abundance-driven relationship and the other observed relationships
234: presented here would be purely coincidental to the [FeII]/[NeII] ratio. In order to explore if this indeed is the true cause or if shocks alone can
235: explain the observed relationships, we used shock and standard \hii\ - PDR models in an effort to model the observed [FeII]/[NeII] relationship for a
236: wide variety of environments. We used both the \emph{MAPPINGS III} photoionization/shock code \citep{sutherland93} and the Cloudy photoionization
237: code \citep{ferland98} in order to determine the relative proportions to the [FeII] emission from shocks and \hii/PDRs, and by extension, the driving
238: process behind the behaviour of the observed [FeII]/[NeII] line ratio. In addition to the mid-IR output, we also wished to explore if a similar
239: relationship is present at optical wavelengths. To this end, we use the ratio of extinction corrected [SII] fluxes at 6716 and 6731~\AA~taken from
240: the literature and compared the observed ratios with the\emph{ MAPPINGS }output.
241:
242:
243: \subsubsection{MAPPINGS models}
244:
245:
246:
247:
248: In an effort to quantify the true nature of the [FeII]/[NeII] ratio, we initially constructed a number of \emph{MAPPINGS} models where density,
249: metallicity and luminosity were allowed to vary. The theoretical \hii\ region models were generated by the \emph{MAPPINGS III} code, which uses as
250: input the EUV fields predicted by the stellar population synthesis models \emph{STARBURST99} \citep{leitherer99}. The photoionization modeling and
251: shock models carried out with \emph{MAPPINGS III} for this analysis are described in \cite{dopita96} and \cite{kewley01}, and are described briefly
252: here. To model the \hii\ region spectrum, \emph{MAPPINGS} assumes the metallicity and the shape of the EUV spectrum are defined (through the
253: \emph{STARBURST99} input), and characterizes the local ionization state by a local ionization parameter:
254:
255: \begin{equation}
256: q = \frac{S_{H^{\emph{0}}}}{n}
257: \end{equation} where \emph{S} is the ionizing photon flux through a unit area, \emph{n} is the local number density of hydrogen atoms and \emph{q} is the maximum
258: velocity of an ionization front that can be driven by the local radiation field \citep{dopita00, kewley02}. We used a 5 Myr continuous
259: star-formation SED from \emph{STARBURST99} as input to the \emph{MAPPINGS III} code. The parameters of the SED consisted of a Saltpeter IMF with a
260: power law of 2.35 and a star-formation rate of 1 $\it M_{\odot}$ yr$^{-1}$.
261:
262: We then constructed a series of plane parallel, isobaric models with \emph{P/k} = 10$^{5}$ cm$^{-3}$ K, where either the gas density, metallicity or
263: the luminosity for the \emph{STARBURST99} input were allowed to vary, with the other two variable remaining constant. For the \emph{STARBURST99}
264: input models, the density, metallicity and luminosity were varied as follows:
265:
266: $\bullet$ \textbf{\emph{Density:}} 10-1000 cm$^{-3}$; metallicity (0.4\emph{Z$_{\odot}$}) and luminosity (10$^{43}$ ergs) remain constant;
267:
268: $\bullet$ \textbf{\emph{Metallicity:}} 0.05 - 2\emph{Z$_{\odot}$}; density (100 cm$^{-3}$) and luminosity (10$^{43}$ ergs) remain constant;
269:
270: $\bullet$ \textbf{\emph{Luminosity:}} 10$^{41}$ - 10$^{45}$ ergs; density (100 cm$^{-3}$) and metallicity (0.4\emph{Z$_{\odot}$}) remain constant.
271:
272: Elemental abundances for 0.5-2.0\emph{Z$_{\odot}$} were adopted for the variable metallicity models, while for the variable density and luminosity
273: models, 0.4\emph{Z$_{\odot}$} metallicity abundances were used.
274:
275: Dust physics is treated explicitly through absorption, grain charging and photoelectric heating and the use of a standard MRN grain size
276: distribution for both carbonaceous and silicaceous grain types \citep{mat77}. The \emph{MAPPINGS} code took the undepleted solar abundances to be
277: those of \cite{anders89}. For non-solar metallicities, \emph{MAPPINGS} assumes that both the dust model and the depletion factors are unchanged. All
278: elements except nitrogen and helium are taken to be primary nucleosynthesis elements. As for the shock models, we use the pre-run shock models from
279: the \emph{MAPPINGS} website, which are based upon a grid of \emph{n} = 1.0 models with \emph{v} = 100-1000 km/s in steps of 25 km/s. The magnetic
280: parameter \emph{B = p/$\sqrt{n}$} is fixed to 3 in all calculations. This value corresponds to equipartition between thermal and magnetic pressures
281: \citep{dopita96}.
282:
283:
284: \begin{figure}
285: \resizebox{\columnwidth}{!}{\includegraphics*{f5.eps}}
286:
287: \caption{Plots of the [SIII] (top) and [SII] (bottom) ratios vs [FeII]/[NeII] for the extended galaxy sample, overlaid with model data from \emph{MAPPINGS}. The dot-dash line
288: indicates the low density limit.}
289:
290: \end{figure}
291:
292:
293: From the \emph{MAPPINGS} output from the variable metallicity model, we saw that whilst the [SIII] and [SII] values remain relatively constant (and
294: by extension, the density), the [FeII]/[NeII] ratio varies quite considerably. With the variable density model, interestingly, we saw that
295: [FeII]/[NeII] does not drop with increasing density contrary to what we observe (Fig. 3), but rather slightly increases, contrary to what we observe.
296: Finally, for the variable luminosity model, both the [FeII]/[NeII] and unsurprisingly the [SIII] and [SII] values, remained constant. From the
297: \emph{MAPPINGS} model data, it would seem that the change in metallicity is primarily responsible for the observed drop in the [FeII]/[NeII] ratio.
298: In order to directly compare the photoionization and shock model output with the IRS and optical data, we then set about defining models that would
299: replicate the sort of density, metallicity and luminosity parameters typically seen in the sample. Since the low metallicity objects tend to have
300: lower densities and luminosities, we generated \emph{MAPPINGS} models that gradually increased the metallicity, density and luminosity, over the
301: ranges outlined above, moving from low to high metallicity (0.05, 0.2, 0.4 and solar metallicity), low to high density (20-1000 cm$^{-3})$ and low to
302: high luminosity (10$^{42}$ to 10$^{44}$ ergs). As we increased the metallicity, we concurrently moved to high densities and higher luminosities. As
303: with the earlier \emph{MAPPINGS} models, we used a 5 Myr continuous star-formation SED from \emph{STARBURST99} as input to the \emph{MAPPINGS III}
304: code \citep{leitherer99}. The parameters of the SED consisted of a Saltpeter IMF with a power law of 2.35 and a star-formation rate of 1 $\it
305: M_{\odot}$ yr$^{-1}$. Elemental abundances for 1/20\emph{Z$_{\odot}$}, 2/5\emph{Z$_{\odot}$} and 1\emph{Z$_{\odot}$} were again adopted.
306:
307:
308:
309: Using the observed [SII] 6716 \& 6731~\AA, [SIII] 18 \& 33 $\mu$m plus the [FeII] 26 $\mu$m and [NeII] 12.8 $\mu$m fluxes, we obtained the respective
310: ratios and performed best fit line calculations to the [SIII] vs [FeII]/[NeII] and [SII] vs [FeII]/[NeII] model data, where the line fluxes are the
311: product of both the photoionization and shock model outputs. Using the best fit models, we plotted the model ratio data in conjunction with the
312: observed IRS and optical data ratios (Fig. 5), with the model data derived from the second $\it MAPPINGS$ run describing very well the behaviour of
313: both sets of observed line ratios.
314:
315: \subsubsection{Cloudy models}
316:
317: The slight increase in the [FeII] emission with density seen from the \emph{MAPPINGS} output poses the following question - where does this increase
318: originate from? Is this increase due to an increased contribution to the [FeII] emission by dense \hii\ regions and PDRs? If so, what is the level of
319: this contribution? To resolve this, we used the spectral synthesis code Cloudy to try and determine how much of a contribution \hii\ regions and PDRs
320: make to the overall [FeII] emission, how this contribution is affected by the gas metallicity (as suggested by \emph{MAPPINGS}), and the effect of
321: the \hii/PDR contribution has on the overall [FeII]/[NeII] ratio. We used the developmental version of the spectral synthesis code Cloudy, last
322: described by \cite{ferland98}. \cite{abel05} and \cite{shaw05} describe recent advances in its treatment of PDR chemistry, while \cite{vanhoof04}
323: describes the grain physics. As with the \emph{MAPPINGS} models, we used a 5 Myr continuous star-formation SED from \emph{STARBURST99} as input to
324: Cloudy. The parameters of the SED consisted of a Saltpeter IMF with a power law of 2.35 and a star-formation rate of 1 $\it M_{\odot}$ yr$^{-1}$.
325:
326: Our model geometry consists of a plane-parallel slab illuminated on one side by a source of UV radiation. Our calculations start at the hot,
327: illuminated face of the slab, where all the hydrogen is ionized. We end our calculations at two locations, when the hydrogen ionization fraction
328: falls below 1\% (for the pure \hii\ region calculations, see below) and when the fraction of hydrogen in the form of \emph{H$_{2}$}(\emph{(2 $\times$
329: n(H$_{2}$))/{n(H)$_{total}$})} exceeds 90\%. Beyond this depth, we expect there to be little contribution to either [FeII] or [NeII] emission.
330:
331: We parameterize the ionizing continuum with the `ionization parameter', which is the dimensionless ratio of hydrogen ionizing flux to density, \[U =
332: \frac{Q_H}{4 \pi R^2 N_H c}\] where \emph{Q$_{H}$} is the total number of hydrogen-ionizing photons emitted by the central object per second. For all
333: calculations, we chose Log[\emph{U}] = -2.5, a value typical in \hii\ regions \citep{veilleux87}. The ionization parameter in \emph{MAPPINGS} is
334: related to the one used in Cloudy by the equation:
335:
336: \begin{equation}
337: U = \frac{q}{c}
338: \end{equation}
339:
340: We also include cosmic rays in our calculations. Primary and secondary cosmic ray ionization processes are treated as described in Appendix C of
341: \cite{abel05}, with an assumed cosmic ray ionization rate of 5 $\times$ $10^{-17}$ s$^{-1}$. Cosmic rays are a significant heating source and also
342: drive the ion-molecule chemistry deep in the molecular cloud. However, since our calculations stop short of the molecular cloud, for our models
343: cosmic rays are only of secondary importance. Our calculations include the major ionization processes that can affect the ionization structure. This
344: includes all stages of ionization for the lightest 30 elements.
345:
346: Dust is known to play an important role in both \hii\ regions and PDRs \citep{draine03}. We self-consistently determine the grain temperature and
347: charge as a function of grain size and material, for the local physical conditions and radiation field. This determines the grain photoelectric
348: heating of the gas, an important gas heating process, as well as collisional energy exchange between the gas and dust. We also treat stochastic
349: heating of grains as outlined in \cite{guhathakurta89}, which can affect the dust continuum shape. We include grain charge transfer as a general
350: ionization - recombination process, as described in Appendix B of \cite{abel05}. The rates at which H$_{2}$ forms on grain surfaces is derived using
351: the temperature and material-dependent rates given in \cite{cazaux02}. The assumed grain size distribution is representative of the star-forming
352: regions. The ratio of total to selective extinction, \emph{R$_{V}$} = \emph{A$_{V}$}/(\emph{A$_{B}$}-\emph{A$_{V}$}), is a good indicator of the
353: size distribution of grains \citep{cardelli89}. \cite{calzetti00} derived an average value for \emph{R$_{V}$} $\sim$ 4.3. We therefore use the
354: \emph{R$_{V}$} = 4 grain size distribution given in \cite{weingartner01}. We also include size-resolved PAHs in our calculations, with the same size
355: distribution used by \cite{bakes94}. The abundance of carbon atoms in PAHs that we use, \emph{n$_{C}$}(PAH)/\emph{n$_{H}$}, is 3 $\times$ 10$^{-6}$.
356: PAHs are thought to be destroyed by hydrogen ionizing radiation and coagulate in molecular environments (see, for instance, \cite{omont86}). We
357: model this effect by scaling the PAH abundance by the ratio of H$^{0}$/H$_{tot}$(n$_{C}$(PAH)/n$_{H}$ = 3 $\times$ 10$^{-6}$
358: [\emph{n}(H$^{0}$)/\emph{n}(H$_{tot}$)]).
359:
360: The assumed gas-phase and dust abundances are extremely important in our calculations. We also assume gas-phase abundances based on an average of the
361: abundances in the Orion Nebula derived by \cite{baldwin91}, \cite{rubin91}, and \cite{osterbrock92}. The abundances by number are He/H = 0.095, C/H=
362: 3 $\times$ 10$^{-4}$; O/H= 4 $\times$ 10$^{-4}$, N/H= 7 $\times$ 10$^{-5}$, and Ar/H= 3 $\times$ 10$^{-6}$. We also assume default grain abundance
363: of A$_{V}$/\emph{N}(H$_{tot}$) = 5 x 10$^{-22}$ mag cm$^{2}$. We then scale the metal and dust abundance such that the final value of \emph{Z} =
364: 12+log[\emph{O/H}] ranges from 7 to 10, in increments of 1 dex.
365:
366: \begin{figure}
367: \resizebox{\columnwidth}{!} {\includegraphics*{f6.eps}}
368:
369: \caption{ Cloudy photoionization model output of the [FeII]/[NeII] ratio as a function of density and \emph{Z}, with contributions to the[FeII] emission by \hii\ regions only.}
370:
371:
372: \end{figure}
373:
374: \begin{figure}
375: \resizebox{\columnwidth}{!} {\includegraphics[angle=0]{f7.eps}}
376:
377: \caption{ Cloudy photoionization model output of the [FeII]/[NeII] ratio as a function of density and \emph{Z}
378: with contributions to the [FeII] emission from both \hii\ and PDR regions.}
379:
380: \end{figure}
381:
382:
383:
384: There is one major exception to our assumed abundances, and that is the \emph{Fe/H} ratio. \cite{rodriguez05} discuss the variation in \emph{Fe/H}
385: with \emph{Z}, and as noted earlier, \cite{izotov06} also determined the variation in elemental abundances with \emph{Z} from the Sloan Digital Sky
386: Survey, including Ne and Fe. For these two elements, \cite{izotov06} provide the following equations to derive their respective abundances (for this
387: work, we use the notation \emph{Z} instead of \emph{X}, as used in \cite{izotov06}:
388:
389: \begin{equation}
390: Log(Ne/O) = 0.097\emph{Z}- 1.542
391: \end{equation}
392:
393: and
394:
395: \begin{equation}
396: Log(Fe/O) = -0.606\emph{Z} + 2.994
397: \end{equation}
398:
399: where \emph{Z} = 12 + Log(\emph{O/H}), we find that:
400:
401: \begin{equation}
402: Log \left[\frac{Ne}{Fe}\right] = -4.536 + 0.703Z.
403: \end{equation}
404:
405: In all calculations, we scale the Fe abundance so that equation 3 is always obeyed.
406:
407: The relationship between density, temperature, and pressure is determined by the assumed equation of state. We perform two sets of calculations, one
408: where we stop the calculation at the hydrogen ionization front and one where we include the PDR. For the pure \hii\ region models we assume constant
409: density, with the density ranging from Log[\emph{n}$_{H}$] 0.5 to Log[\emph{n}$_{H}$] = 4, in increments of 0.5 dex. For the \hii\ + PDR
410: calculations, we assume constant pressure, with thermal and radiation pressure being the dominant contributors to the total pressure. We do not
411: include turbulent pressure in the equation of state, although we assume a small turbulent broadening of 1 km s$^{-1}$ for its effects on line optical
412: depths. Such an equation of state neglects shocks or other time-dependent effects. In the constant pressure scenario, we specify \emph{n}$_{H}$ at
413: the illuminated face, then solve for density based how the pressure contributions change with depth. Our densities range over Log[\emph{n}$_{H}$] =
414: 2 to 4, in increments of 0.5 dex. Overall, our combination of \emph{n}$_{H}$ and \emph{Z} represent 52 calculations.
415:
416:
417:
418:
419:
420: In Fig. 6, we plot the [FeII]/[NeII] line ratio as a function of $\it Z$ and density for the \hii\ region only. In Fig. 7, we again plot the
421: [FeII]/[NeII] line ratio as a function of $\it Z$ and density, but this time including the PDR in addition to the \hii\ region. As one can see, the
422: relative \hii\ contribution dominates over the PDR contribution at low densities (and metallicities), with the PDR contribution becoming more
423: dominant as one moves to higher densities ($\sim$ 10$^{3}$ cm$^{-3}$) and metallicities. Previous IR studies of starburst \hii/PDR regions
424: \citep{carral94,lord96} support this scenario. For very dense \hii\ regions where the PDRs are irradiated by intense FUV, and thus PDR-derived [FeII]
425: emission is dominant. Correspondingly at lower densities ($\sim$10$^{2}$ cm$^{-3}$), the \hii\ regions are larger, the PDRs lie further from the
426: stars, and the resultant lowered FUV fluxes and densities do not excite [FeII] in the PDR, and thus the \hii\ region dominates the \hii/PDR
427: contribution to [FeII] production for such environments.
428:
429:
430:
431: \subsubsection{What is driving the Fe/Ne relationship?}
432:
433: From the $\it MAPPINGS$ and $\it Cloudy$ simulations and the comparisons with the IRS and optical data (Fig. 5), it would seem that metallicity is
434: the primary driver for the behaviour of the [FeII]/[NeII] ratio, as metallicity determines the origin (shocks vs \hii\ and/or PDRs) and strength of
435: the [FeII] emission. We can reproduce the observed [FeII]/[NeII] dependence on \emph{Z} if we assume the Izotov variation of \emph{Fe/Ne} with
436: \emph{Z}. Cloudy reproduces the [FeII]/[NeII] ratio decrease with increases in density (for the pure H II region only) because of the following: the
437: critical density for electron excitation of [Fe II] is 1.2 x 10$^{4}$ cm$^{-3}$, while for [Ne II] it is 5 x 10$^{5}$ cm$^{-3}$. The ratio of [Fe
438: II]/[NeII] thus depends on the ratio:
439:
440: \[\frac{(1 + n_{critical}(neon)/n(\emph{H}))} {(1 + n_{critical}(iron)/n(\emph{H}))}\]
441:
442:
443: In the low density limit, this ratio is $\sim$42. However, for a density of 1 x 10$^{4}$ cm$^{-3}$, which is the high end of our parameter space,
444: this ratio is only $\sim$20 because [Fe II] emission is suppressed by the higher density. The reason the ratio increases with density in Fig. 7 (\hii
445: + PDR) is because [Fe II] is excited by atomic hydrogen in the PDR in addition to electrons in the \hii\ region. The critical density of [Fe II]
446: with H atoms is $\sim$1 x 10$^{6}$ cm$^{-3}$. Therefore [Fe II] emits readily in the PDR, while [Ne II] does not emit because all the neon is
447: neutral. However, we do note that the \hii\ and PDR contribution to the [FeII] emission is surpassed by the contribution from shocks at the lowest
448: metallicities and densities, based on comparison with the \emph{MAPPINGS} output and the observed data - perhaps only $\sim$10\% of the [FeII]
449: emission comes from \hii\ and PDRs with that contribution rising to $\sim$80-90\% as one moves to higher metallicity. As we move to roughly solar
450: metallicity, it would seem likely that the dominant mechanism for production of [FeII] (and thus the behaviour of the [FeII]/[NeII] ratio at high
451: metallicities) is from \hii\ and PDRs in combination rather than from SNe shocks.
452:
453:
454: \subsubsection{Can we tie in the behaviour of the gas density with the [FeII]/[NeII] ratio, metallicity and PAH strength?}
455:
456: While abundances may play a role in the overall behavior of the [FeII]/[NeII] ratio, the shock scenario would seem to fit the observed variation of
457: the gas density and the PAH ratio with metallicity. Assuming that shocks are indeed behind the observed correlations, the following scenarios would
458: seem to account for the relationship between the [FeII]/[NeII] and [SIII] ratios and the PAH strength. In very low metallicity dwarf starburst
459: environments, such as I Zw 18, the observed gas density is phenomenally low, in the order of $\sim$20-100 cm$^{-3}$. From Fig. 3 and the models, it
460: would seem that low density correlates strongly with the presence of strong supernovae shocks, a harsh radiation environment which is lacking in
461: populations of dust and PAH, as noted by the lack of PAH emission (Figs. 1 and 4) and generally flat slopes of the dust continua. It would seem
462: likely therefore that we are probing the tenuous gas within a superbubble cavity that has been carved by the cumulative effects of numerous
463: supernova-driven shocks. Previous surveys of low metallicity dwarf galaxies have found evidence for such superbubbles
464: \citep{strickland00,calzetti04,ott05}, with ionized gas expanding at a rate of several tens of km/s driven by mechanical energy from starbursts
465: deposited into the ISM \citep{calzetti04}.
466:
467: For high metallicity environments, the picture becomes somewhat more complicated. Such objects are of course still the sites of high levels of
468: massive star formation - the high metallicity starburst M82 is after all the template for starburst activity - but we see a much more benign
469: environment for the survivability of dust and PAH. A number of new factors play a role in affecting star formation and its effects on the local
470: environment. First of all, from the Cloudy and \emph{MAPPINGS} models plus the observed data, we note that the dominant contribution to the [FeII]
471: flux no longer comes from shocks, but rather from \hii\ regions and from PDRs in particular - the lack of shocks propagating into the ISM provides a
472: more benign, denser environment for the build-up of dust and PAH. Superbubbles are present as in low metallicity galaxies, however supernova-driven
473: shocks are no longer the dominant mechanism driving the expansion but rather stellar winds from massive stars. As a consequence, less mechanical
474: energy is available to drive breakout and outflow and the superbubble can remain enclosed, especially where the gravitational potential of the galaxy
475: is steep \citep{skill95,burkert04,calzetti04}. Given the inability of weaker shocks to drive out material and clear the local ISM, it is not
476: surprising that we should see a buildup of metals, and the survivability of dust and PAH formed locally.
477:
478:
479: \subsubsection{Are abundances driving the Fe/Ne relationship?}
480:
481: Whilst the SNe-driven shock hypothesis is convincing in terms of the enhanced [FeII] emission and corresponding low PAH emission, it is the not the
482: only plausible explanation for the observed correlation. As we have already noted, from \cite{izotov06} it has been suggested that the observed
483: levels of [FeII] emission may instead be abundance driven. Using equations 1 \& 2 to derive how the \emph{Fe/Ne} ratio varies with metallicity, we
484: see that the \emph{Fe/Ne} ratio decreases with increasing \emph{Z} in a fashion very similar to the [FeII]/[NeII] vs.~\emph{Z} correlation,
485: suggesting that the variation in the [FeII]/[NeII] emission ratio with \emph{Z} may be a consequence of varying \emph{Fe/Ne} abundance with \emph{Z}.
486: Indeed, the slope of \emph{Fe/Ne} vs. \emph{Z} from \cite{izotov06} is very similar to the slope of the best fitting line from Fig 2 (m = -0.68) -
487: our observational dataset points to this being a possibility, and the \emph{MAPPINGS} and Cloudy models further lend support to this scenario.
488:
489:
490: However given the similarity in terms of behaviour of our observational dataset with the \cite{izotov06} results, it begs the question - what may be
491: causing the change in \emph{Fe/Ne} with \emph{Z}? Since neon is not depleted \citep{juett06,yao06}, the observed relationship would then be due to
492: variations in \emph{Fe}/\emph{H} rather than \emph{Ne}/\emph{H}. So what is causing the iron depletion? Since iron is known to be depleted onto
493: grains \citep{juett06,yao06}, the increased variation in \emph{Fe}/\emph{H} would suggest that there is either more grain destruction, or less grain
494: formation, as one moves to lower metallicities.
495:
496: If this is correct, the reason for the observed [FeII]/[NeII] with \emph{Z} trend is that Fe is less depleted in low metallicity galaxies. This would
497: have a profound impact on ISM studies, as the usefulness of iron emission as a pure indicator of shocks may be compromised. However, this is not
498: conclusive, and we need more observations to deduce for certain the physical process which is controlling the [FeII] emission. In order to
499: definitively figure out if shocks are truly responsible for the [FeII] emission at low metallicity, we require high resolution IR spectroscopy to
500: unambiguously determine the presence of shocks within the velocity profiles. The upcoming availability of such high resolution spectroscopic data
501: from {\it Herschel} and {\it SOFIA} will be crucial to determine the true nature of iron emission at low metallicity and we look forward to
502: furthering this line of investigation in a future paper.
503:
504: \section{Conclusions}
505:
506: Using archival \emph{Spitzer} observations of 41 starburst galaxies that span a wide range in metallicity, we found a correlation between the ratio
507: of emission line fluxes of [FeII] at 26 $\mu$m and [NeII] at 12.8 $\mu$m and the electron gas density as traced by the 18.7/33.4 $\mu$m [SIII] ratio,
508: with the [FeII]/[NeII] flux ratio decreasing with increasing gas density. We also find a strong correlation between the gas density and the PAH peak
509: to continuum strength. The correlation of the [FeII]/[NeII] ratio and the PAH peak to continuum strength found in \cite{oha06} was confirmed for a
510: larger sample. Using shock and photoionization models, we see that metallicity is the primary driver for the observed behaviour of the [FeII]/[NeII]
511: ratio. It may very well be that the majority of [FeII] emission at low metallicity may be shock-derived, whilst at high metallicity, the [FeII]
512: emission is dominated by contributions from \hii\ and in particular, from PDR regions, and that at higher metallicity shocks may not play as
513: significant a role in removing gas, PAH and dust from the ISM, unlike in low metallicity systems. However, the observed [FeII]/[NeII] emission may
514: instead be following a metallicity-abundance relationship, with the iron being less depleted in low metallicity galaxies, a result that would have
515: profound implications for the use of Fe emission lines as unambiguous tracers of shocks. Follow up high resolution spectroscopic observations will be
516: required to determine if this is indeed the case, or if the detected iron emission is truly tracing emission from the passage of shocks.
517:
518: \acknowledgments
519:
520: We are grateful to S. Satyapal for her comments and guidance. BO'H gratefully acknowledges financial support from NASA grant NAG5-11432. NPA
521: acknowledges National Science Foundation support under Grant No. 0094050 , 0607497 to The University of Cincinnati. This work is based on archival
522: data obtained with the \emph{Spitzer} Space Telescope, which is operated by the Jet Propulsion Laboratory, California Institute of Technology under a
523: contract with NASA. The IRS was a collaborative venture between Cornell University and Ball Aerospace Corporation funded by NASA through the Jet
524: Propulsion Laboratory and Ames Research Center. \emph{SMART} was developed at Cornell University and is available through the \emph{Spitzer} Science
525: Center at Caltech. The \emph{SPICE} reduction package has been developed by and released through the \emph{Spitzer} Science Center.
526:
527:
528: \begin{thebibliography}{}
529:
530: \bibitem[Abel et al.(2005)]{abel05}
531: Abel N.P., Ferland G.J., Shaw G. \& van Hoof P.A.M., 2005, \apjs, 161, 65
532:
533: \bibitem[Anders \& Grevesse(1989)]{anders89}
534: Anders E. \& Grevesse N., 1989, Geochimica et Cosmochimica Acta, 53, 197
535:
536: \bibitem[Bakes \& Tielens(1994)]{bakes94}
537: Bakes E.L.O. \& Tielens A.G.G.M., 1994, \apj, 427, 822
538:
539: \bibitem[Baldwin et al.(1991)]{baldwin91}
540: Baldwin J.A., Ferland G.J., Martin P.G. et al., 1991 \apj, 374, 580
541:
542: \bibitem[Brandl et al.(2004)]{bran04}
543: Brandl B.R., Devost D., Higdon S.J.U. et al., 2004, \apjs, 154, 188
544:
545: \bibitem[Burkert(2004)]{burkert04}
546: Burkert A., 2004, The Formation and Evolution of Massive Young Star Clusters, ASP Conference Series, vol. 322., p.489
547:
548: \bibitem[Calzetti et al.(2000)]{calzetti00}
549: Calzetti D., Armus L., Bohlin R.C. et al., 2000, \apj, 533, 682
550:
551: \bibitem[Calzetti et al.(2004)]{calzetti04}
552: Calzetti D., Harris J., Gallagher J. et al., 2004, \aj, 127, 1405
553:
554: \bibitem[Cardelli et al.(1989)]{cardelli89}
555: Cardelli J.A., Clayton G.C. \& Mathis J.S., 1989, \apj, 345, 245
556:
557: \bibitem[Carral et al.(1994)]{carral94}
558: Carral P., Hollenbach D.J., Lord S.D. et al., 1994, \apj, 423, 223
559:
560: \bibitem[Cazaux \& Tielens(2002)]{cazaux02}
561: Cazaux S. \& Tielens A.G.G.M., 2002, \apj, 575, 29
562:
563:
564: \bibitem[Dopita \& Sutherland(1996)]{dopita96}
565: Dopita M.A. \& Sutherland R.S., 1996, \apjs, 102, 61
566:
567: \bibitem[Dopita(2000)]{dopita00}
568: Dopita M.A., 2000, Ap\&SS, 272, 79
569:
570:
571: \bibitem[Draine(1989)]{draine89}
572: Draine B. T., 1989, Interstellar Extinction in the Infrared, European Space Agency, p.93.
573:
574: \bibitem[Draine(2003)]{draine03}
575: Draine B.T., 2003, \araa, 41, 241
576:
577: \bibitem[Dudik(2007)]{dudik07}
578: Dudik R.P., Weingartner J.C., Satyapal S. et al., 2007, \apj, 684, 71.
579:
580: \bibitem[Ferland et al.(1998)]{ferland98}
581: Ferland G.J., Korista K.T., Verner D.A. et al., 1998, \pasp, 110, 761
582:
583: \bibitem[F\"{o}rster Schreiber et al.(2004)]{fors04}
584: F\"{o}rster Schreiber, N. M., Roussel, H., Sauvage, M., Charmandaris, V., 2004, \aap, 419, 501
585:
586: \bibitem[Galliano(2008)]{galliano08}
587: Galliano F., Dwek E. \& Chanial P., 2008, \apj, 672, 214
588:
589: \bibitem[Guhathakurta \& Draine(1989)]{guhathakurta89}
590: Guhathakurta P. \& Draine B.T, 1989, \apj, 345, 230
591:
592: \bibitem[Higdon et al.(2004)]{hig04}
593: Higdon S.J.U., Devost D., Higdon J.L. et al., 2004, \pasp, 116, 975
594:
595: \bibitem[Hogg et al.(2005)]{hogg05}
596: Hogg D.W., Tremonti C.A., Blanton M.R. et al., 2005, \aap, 388, 439
597:
598: \bibitem[Hoyos \& Diaz(2006)]{hoyos06}
599: Hoyos C. \& Diaz A.I., 2006, \mnras, 365, 454
600:
601: \bibitem[Ho et al.(1997)]{ho97}
602: Ho L.C., Filippenko A.V. \& Sargent W.L.W., 1997 \apjs, 112, 315
603:
604: \bibitem[Izotov \& Thuan(2002)]{izo02}
605: Izotov Y.I. \& Thuan T.X., 2002, \apj, 567, 875
606:
607: \bibitem[Izotov et al.(2004)]{izo04}
608: Izotov Y.I., Papaderos P., Guseva N.G. et al., 2004, \aap, 421, 539
609:
610: \bibitem[Izotov et al.(2006)]{izotov06}
611: Izotov Y.I., Stasinska G., Meynet G. et al., 2006, \aap, 448, 955
612:
613: \bibitem[Juett et al.(2006)]{juett06}
614: Juett A.M., Schulz N.S., Chakrabarty D., Gorczyca T.W., 2006, \apj, 648, 1066
615:
616: \bibitem[Kendall \& Stuart(1976)]{ken76}
617: Kendall M. \& Stuart A. 1976, In: The Advanced Theory of Statistics, Vol. 2, Macmillian
618:
619: \bibitem[Kewley et al.(2001)]{kewley01}
620: Kewley L.J., Dopita M.A., Sutherland R.S., Heisler C.A. \& Trevena J., 2001, \apj, 556, 121
621:
622: \bibitem[Kewley \& Dopita(2002)]{kewley02}
623: Kewley L.J. \& Dopita M.A., 2002, \apjs, 142, 35
624:
625: \bibitem[Kong \& Cheng(1999)]{kong99}
626: Kong X. \& Cheng F.Z., 1999, \aap, 351, 477
627:
628: \bibitem[Leitherer et al.(1999)]{leitherer99}
629: Leitherer C., Schaerer D., Goldader J.D. et al., 1999, \apjs, 123, 3
630:
631: \bibitem[Lord et al.(1996)]{lord96}
632: Lord S.D., Hollenbach D.J., Haas M.R. et al., 1996, \apj, 465, 703
633:
634: \bibitem[Lutz, Veilleux \& Genzel(1999)]{lutz99}
635: Lutz D., Veilleux S., \& Genzel R., 1999, \apj 517, 13
636:
637: \bibitem[Madden et al.(2006)]{mad06}
638: Madden S.C., Galliano F., Jones A.P. \& Sauvage, M., 2006, \aap, 446, 877
639:
640: \bibitem[Mathis, Rumpl \& Nordsieck(1977)]{mat77}
641: Mathis J.S., Rumpl W. \& Nordsieck, K. H., 1977, \apj, 217, 425
642:
643: \bibitem[O'Halloran, Satyapal \& Dudik(2006)]{oha06}
644: O'Halloran B., Satyapal S. \& Dudik R.P., 2006, \apj, 641, 795
645:
646: \bibitem[Omont(1986)]{omont86}
647: Omont A., 1986, \aap, 164, 159
648:
649: \bibitem[Osterbrock et al.(1992)]{osterbrock92}
650: Osterbrock D.E., Tran H.D. \& Veilleux S., 1992, \apj, 389, 305
651:
652: \bibitem[Ott et al.(2005)]{ott05}
653: Ott J., Walter F. \& Brinks E., 2005, \mnras, 358, 1423
654:
655: \bibitem[Reach \& Rho(2000)]{reach00}
656: Reach W.T. \& Rho J., 2000, \apj, 544, 843
657:
658: \bibitem[Reach et al.(2006)]{reach06}
659: Reach W.T., Rho J., Tappe A. et al., 2006, \aj, 131, 479
660:
661: \bibitem[Rigopoulou et al.(1996)]{rigo96}
662: Rigopoulou D., Lutz D., Genzel R. et al., 1996, \aap, 315, 125
663:
664: \bibitem[Rigoupolou et al.(1999)]{rig99}
665: Rigopoulou, D., Spoon, H. W. W., Genzel, R. et al., 1999, \aj, 118, 2625
666:
667: \bibitem[Rodriguez \& Rubin(2005)]{rodriguez05}
668: Rodriguez M. \& Rubin R.H., 2005, \apj, 626, 900
669:
670: \bibitem[Rubin et al.(1991)]{rubin91}
671: Rubin R.H., Simpson J.P., Haas M.R. \& Erickson E.F., 1991, \apj, 374, 564
672:
673: \bibitem[Satyapal et al.(2005)]{sat05}
674: Satyapal S., Dudik R.P., O'Halloran B. \& Gliozzi M., 2005, \apj, 633, 86
675:
676: \bibitem[Scoville et al.(2000)]{sco00}
677: Scoville N.Z., Evans A.S., Thompson R. et al., 2000, \aj, 119, 991
678:
679: \bibitem[Shaw et al.(2005)]{shaw05}
680: Shaw G., Ferland G.J., Abel N.P. et al., 2005, \apj, 624, 794
681:
682: \bibitem[Skillman \& Bender(1995)]{skill95}
683: Skillman E.D. \& Bender R., 1995, RMxAC, 3, 255
684:
685: \bibitem[Smith et al.(2005)]{smi05}
686: Smith B.J., Struck C. \& Nowak M.A., 2005, \aj, 129, 1350
687:
688: \bibitem[Storey \& Hummer(1995)]{storey95}
689: Storey, P.J. \& Hummer D.G., 1995, \mnras, 272, 41
690:
691: \bibitem[Strickland \& Stevens(2000)]{strickland00}
692: Strickland D.K. \& Stevens I.R., 2000, \mnras, 304, 511
693:
694: \bibitem[Sturm et al.(2000)]{stu00}
695: Sturm E., Lutz D., Tran D. et al., 2000, \aap 358, 481
696:
697: \bibitem[Surace \& Sanders(1999)]{sur99}
698: Surace J.A. \& Sanders D. B., 1999, \apj, 512, 162
699:
700: \bibitem[Sutherland \& Dopita(1993)]{sutherland93}
701: Sutherland R.S. \& Dopita M.A., 1993, \apjs, 88, 253
702:
703: \bibitem[Thomas et al.(2002)]{tho02}
704: Thomas H.C., Dunne L., Clemens, M.C. et al., 2002, \mnras, 329, 747
705:
706: \bibitem[Vacca \& Conti(1992)]{vacca92}
707: Vacca W. \& Conti P., 1992, \apj, 401, 543
708:
709: \bibitem[van Hoof et al.(2004)]{vanhoof04}
710: van Hoof P.A.M., Weingartner J.C., Martin P.G. et al., 2004, \mnras, 350, 1330
711:
712: \bibitem[Verma et al.(2004)]{ver04}
713: Verma A., Lutz D., Sturm E. et al., 2004, \aap, 403, 829
714:
715: \bibitem[Veilleux \& Osterbrock(1987)]{veilleux87}
716: Veilleux S. \& Osterbrock D.E., 1987, \apjs, 63, 295
717:
718: \bibitem[Weingartner \& Draine(2001)]{weingartner01}
719: Weingartner J.C. \& Draine B.T., 2001, \apj, 548, 296
720:
721: \bibitem[Wu et al.(2006)]{wu06}
722: Wu Y., Charmandaris V., Hao L. et al., 2006, \apj, 639, 157
723:
724: \bibitem[Yao et al.(2006)]{yao06}
725: Yao Y., Schulz N., Wang Q.D. \& Nowak M., 2006, \apj, 653, 121
726: \end{thebibliography}
727:
728:
729: \begin{deluxetable*}{lcrrrrr}
730: \tabletypesize{\scriptsize} \tablecaption{IRS flux data.} \tablewidth{0pt} \tablehead{\colhead{Target} &
731: \colhead{Metallicity} &
732: \colhead{F([NeII])} &
733: \colhead{F([SIII]) } &
734: \colhead{F([FeII]) } &
735: \colhead{F([SIII]) } \\
736: \colhead{} &
737: \colhead{12+log(O/H)} &
738: \colhead{@ 12.8 $\mu$m} &
739: \colhead{@ 18.7 $\mu$m} &
740: \colhead{@ 26.0 $\mu$m} &
741: \colhead{@ 33.4 $\mu$m} \\
742: \colhead{} &
743: \colhead{} &
744: \colhead{\it (x 10$^{-20}$ W cm$^{-2}$)} &
745: \colhead{\it (x 10$^{-20}$ W cm$^{-2}$)} &
746: \colhead{\it (x 10$^{-20}$ W cm$^{-2}$)} &
747: \colhead{\it (x 10$^{-20}$ W cm$^{-2}$)} \\
748: \colhead{(1)} &
749: \colhead{(2)} &
750: \colhead{(3)} &
751: \colhead{(4)} &
752: \colhead{(5)} &
753: \colhead{(6)} }
754:
755:
756: \startdata
757: I Zw 18 & 7.2 & 0.030$\pm$0.003 & 0.006$\pm$0.001 & 0.051$\pm$0.015 & 0.011$\pm$0.006\\
758: SBS 0335-052 & 7.3 & 0.014 $\pm$ 0.003 & 0.013$\pm$0.003 & 0.018$\pm$0.006 & 0.021$\pm$0.008 \\
759: HS 2236+1344 & 7.5 & 0.032 $\pm$ 0.002 & 0.063$\pm$0.008 & 0.056$\pm$0.016 & 0.112$\pm$0.011 \\
760: SDSS J0519+0007 & 7.5 & 0.207$\pm$0.014 & 0.048$\pm$0.014 & 0.155$\pm$0.056 & \\
761: SBS 1415+437 & 7.6 & 0.068$\pm$0.013 & 0.026$\pm$0.007 & 0.076$\pm$0.024 & 0.045$\pm$0.021 \\
762: HS 0837+4717 & 7.6 & 0.121$\pm$0.010 & 0.020$\pm$0.005 & 0.177$\pm$0.036 & 0.033$\pm$0.011 \\
763: Tol 65 & 7.6 & 0.024$\pm$0.013 & 0.011$\pm$0.002 & 0.040$\pm$0.005 & 0.022$\pm$0.008 \\
764: HS 1442+4250 & 7.6 & 0.226$\pm$0.010 & 0.024$\pm$0.004 & 0.127$\pm$0.041 & 0.038$\pm$0.004 \\
765: Mrk 25 & 7.8 & 1.074$\pm$0.156 & 0.847$\pm$0.104 & 0.647$\pm$0.128 & 0.870$\pm$0.026 \\
766: SBS 1030+583 & 7.8 & 0.075$\pm$0.014 & 0.167$\pm$0.011 & 0.028$\pm$0.007 & 0.149$\pm$0.043 \\
767: Mrk 209 & 7.8 & 0.068$\pm$0.017 & 0.070$\pm$0.024 & 0.077$\pm$0.020 & 0.124$\pm$0.024 \\
768: Mrk 36 & 7.9 & 0.281$\pm$0.071 & 0.051$\pm$0.010 & 0.130$\pm$0.024 & 0.085$\pm$0.006 \\
769: SBS 0917+527 & 7.9 & 0.228$\pm$0.058 & 0.115$\pm$0.024 & 0.074$\pm$0.007 & 0.084$\pm$0.008 \\
770: SBS 1152+579 & 7.9 & 0.928$\pm$0.300 & 0.071$\pm$0.037 & 0.305$\pm$0.018 & 0.076$\pm$0.024 \\
771: Mrk 170 & 7.9 & 0.053 $\pm$ 0.015 & 0.033$\pm$0.002 & 0.048$\pm$0.006 & 0.058$\pm$0.009 \\
772: UM 448 & 8.0 & 1.430$\pm$0.018 & 0.397$\pm$0.031 & 0.442$\pm$0.084 & 0.563$\pm$0.099 \\
773: SBS 0946+558 & 8.0 & 0.053$\pm$0.015 & 0.034$\pm$0.022 & 0.026$\pm$0.003 & 0.034$\pm$0.006 \\
774: II Zw 40 & 8.1 & 0.452$\pm$0.061 & 1.375$\pm$0.026 & 0.115$\pm$0.031 & 1.274$\pm$0.304 \\
775: Mrk 930 & 8.1 & 0.219$\pm$0.074 & 0.170$\pm$0.019 & 0.112$\pm$0.026 & 0.189$\pm$0.056 \\
776: Mrk 996 & 8.1 & 0.297$\pm$0.022 & 0.271$\pm$0.044 & 0.073$\pm$0.011 & 0.184$\pm$0.046 \\
777: Tol 1924-416 & 8.1 & 0.464$\pm$0.039 & 0.247$\pm$0.050 & 0.058$\pm$0.001 & 0.271$\pm$0.073 \\
778: II Zw 70 & 8.1 & 0.143$\pm$0.012 & 0.178$\pm$0.041 & 0.182$\pm$0.053 & 0.268$\pm$0.075 \\
779: NGC 4670 & 8.2 & 0.482$\pm$0.173 & 0.240$\pm$0.039 & 0.309$\pm$0.042 & 0.321$\pm$0.029 \\
780: Mrk 450 & 8.2 & 0.125$\pm$0.009 & 0.083$\pm$0.029 & 0.050$\pm$0.015 & 0.096$\pm$0.059 \\
781: Mrk 5 & 8.2 & 0.106$\pm$0.013 & 0.076$\pm$0.020 & 0.075$\pm$0.019 & 0.103$\pm$0.027 \\
782: Mrk 1329 & 8.2 & 0.305$\pm$0.015 & 0.257$\pm$0.054 & 0.060$\pm$0.002 & 0.196$\pm$0.036 \\
783: NGC 5253 & 8.2 & 0.572$\pm$0.075 & 1.172$\pm$0.039 & 0.434$\pm$0.100 & 1.749$\pm$0.249 \\
784: IC 342 & 8.3 & 34.239$\pm$5.285 & 9.902$\pm$0.156 & 5.793$\pm$0.025 & 9.872$\pm$1.771 \\
785: Haro 3 & 8.3 & 2.016$\pm$0.084 & 1.431$\pm$0.064 & 0.306$\pm$0.042 & 1.243$\pm$0.090 \\
786: UM 311 & 8.3 & 0.126$\pm$0.010 & 0.0873$\pm$0.021 & 0.056$\pm$0.009 & 0.101$\pm$0.042 \\
787: Mrk 1499 & 8.4 & 0.150$\pm$0.014 & 0.319$\pm$0.008 & 0.069$\pm$0.017 & 0.393$\pm$0.018 \\
788: UGC 4274 & 8.5 & 0.710$\pm$0.030 & 0.402$\pm$0.036 & 0.308$\pm$0.079 & 0.393$\pm$0.093 \\
789: NGC 7793 & 8.7 & 0.318$\pm$0.077 & 0.295$\pm$0.010 & 0.088$\pm$0.025 & 0.178$\pm$0.069 \\
790: NGC 4194 & 8.8 & 8.464$\pm$0.169 & 2.151$\pm$0.042 & 1.491$\pm$0.336 & 2.151$\pm$0.466 \\
791: NGC 253 & 8.9 & 199.753$\pm$1.235 & 42.928$\pm$0.780 & 28.433$\pm$1.154 & 29.012$\pm$0.416 \\
792: He 2-10 & 8.9 & 24.671$\pm$1.677 & 9.034$\pm$0.156 & 2.300$\pm$0.091 & 6.072$\pm$0.990 \\
793: NGC 7714 & 8.9 & 8.546$\pm$0.169 & 2.665$\pm$0.016 & 1.012$\pm$0.121 & 1.200$\pm$0.194 \\
794: M82 & 9.0 & 948.875$\pm$25.765 & 44.638$\pm$0.047 & 66.546$\pm$2.118 & 18.416$\pm$0.411 \\
795: NGC 3049 & 9.0 & 2.800$\pm$0.353 & 1.008$\pm$0.180 & 0.834$\pm$0.016 & 0.780$\pm$0.104 \\
796: NGC 1482 & 9.1 & 100.483$\pm$4.179 & 13.346$\pm$0.104 & 1.759$\pm$0.073 & 4.905$\pm$0.119 \\
797: NGC 2903 & 9.3 & 84.257$\pm$1.871 & 4.456$\pm$0.026 & 7.563$\pm$0.057 & 1.979$\pm$0.267 \\
798:
799: \enddata
800:
801:
802:
803: \tablecomments{Columns: (1) Common source names; (2) Metallicity of the galaxy; (3) Extinction corrected flux of the 12.8 $\mu$m [NeII] fine
804: structure line in Watts per centimeter squared; (4) Extinction corrected flux of the 18.7 $\mu$m [SIII] fine structure line in Watts per centimeter
805: squared; (5) Extinction corrected flux of the 26.0 $\mu$m [FeII] fine structure line in Watts per centimeter squared; (6) Extinction corrected flux
806: of the 33.4 $\mu$m [SIII] fine structure line in Watts per centimeter squared.}
807: \end{deluxetable*}
808:
809:
810: \begin{deluxetable*}{lcccc}
811: \tabletypesize{\scriptsize} \tablecaption{IRS and optical flux ratios.} \tablewidth{0pt} \tablehead{\colhead{Target} &
812: \colhead{PAH 6.2 micron P/C} &
813: \colhead{[FeII]/[NeII]} &
814: \colhead{log ([SIII] 18.7/33.4 $\mu$m)} &
815: \colhead{log ([SII] 6716/6731~\AA)} \\
816: \colhead{(1)} &
817: \colhead{(2)} &
818: \colhead{(3)} &
819: \colhead{(4)} &
820: \colhead{(5)}}
821:
822:
823: \startdata
824:
825: I Zw 18 & 1.2 & 1.74 & -0.29 & 1.32\tablenotemark{a} \\
826: SBS 0335-052 & 1.23 & 1.29 & -0.22 & 1.2\tablenotemark{b} \\
827: HS 2236+1344 & 1.88 & 1.75 & -0.25 & 1.15\tablenotemark{c} \\
828: SDSS J0519+0007 & 1.89 & 0.75 & 1.17\tablenotemark{c} \\
829: SBS 1415+437 & 1.57 & 1.11 & -0.24 & 1.15\tablenotemark{b} \\
830: HS 0837+4717 & 1.66 & 1.47 & -0.22 & 1.1\tablenotemark{c} \\
831: Tol 65 & 1.41 & 1.68 & -0.29 & 1.21\tablenotemark{c} \\
832: HS 1442+4250 & 1.81 & 0.56 & -0.20 & - \\
833: Mrk 25 & 2.56 & 0.60 & -0.11 & - \\
834: SBS 1030+583 & 1.67 & 0.37 & 0.05 & 1.25\tablenotemark{b} \\
835: Mrk 209 & 2.27 & 1.12 & -0.25 & 1.28\tablenotemark{a} \\
836: Mrk 36 & 1.82 & 0.46 & -0.23 & 1.07\tablenotemark{b} \\
837: SBS 0917+527 & 2.1 & 0.33 & 0.04 & 1.14\tablenotemark{b} \\
838: SBS 1152+579 & 2.41 & 0.33 & -0.14 & 1.17\tablenotemark{b} \\
839: Mrk 170 & 2.2 & 0.90 & -0.25 & \\
840: UM 448 & 1.63 & 0.31 & -0.15 & 1.07\tablenotemark{b} \\
841: SBS 0946+558 & 2.5 & 0.50 & -0.01 & 1.11\tablenotemark{b} \\
842: II Zw 40 & 2.6 & 0.26 & 0.03 & 1.15\tablenotemark{b} \\
843: Mrk 930 & 2.5 & 0.51 & -0.04 & 1.15\tablenotemark{b} \\
844: Mrk 996 & 3.3 & 0.24 & 0.17 & - \\
845: Tol 1924-416 & 2.8 & 0.13 & -0.04 & 1.13\tablenotemark{b} \\
846: II Zw 70 & 2.1 & 1.27 & -0.18 & 1.2\tablenotemark{b} \\
847: NGC 4670 & 3.75 & 0.64 & -0.13 & - \\
848: Mrk 450 & 2.64 & 0.40 & -0.07 & 1.17\tablenotemark{c} \\
849: Mrk 5 & 2.91 & 0.71 & -0.13 & 1.11\tablenotemark{b} \\
850: Mrk 1329 & 3.88 & 0.40 & 0.07 & 1.06\tablenotemark{c} \\
851: NGC 5253 & 3.21 & 0.71 & -0.13 & 0.91\tablenotemark{b} \\
852: IC 342 & 3.42 & 0.17 & 0.00 & 0.83\tablenotemark{e} \\
853: Haro 3 & 4.18 & 0.15 & 0.06 & 1.23\tablenotemark{a} \\
854: UM 311 & 3.27 & 0.44 & -0.06 & 0.92\tablenotemark{b} \\
855: Mrk 1499 & 4.27 & 0.46 & -0.09 & - \\
856: UGC 4274 & 3.92 & 0.43 & 0.01 & 0.9\tablenotemark{e} \\
857: NGC 7793 & 4.78 & 0.28 & 0.22 & - \\
858: NGC 4194 & 3.66 & 0.18 & 0.00 & - \\
859: NGC 253 & 3.05 & 0.14 & 0.17 & - \\
860: He 2-10 & 4.1 & 0.08 & 0.17 & 1.04\tablenotemark{f} \\
861: NGC 7714 & 4.22 & 0.12 & 0.35 & 0.88\tablenotemark{b} \\
862: M82 & 5.07 & 0.07 & 0.38 & 0.8\tablenotemark{e} \\
863: NGC 3049 & 5.23 & 0.30 & 0.11 & 0.87\tablenotemark{b} \\
864: NGC 1482 & 3.7 & 0.02 & 0.43 & - \\
865: NGC 2903 & 5.26 & 0.09 & 0.35 & 1.11\tablenotemark{e} \\
866:
867: \enddata
868:
869:
870:
871:
872:
873:
874: \tablecomments{Columns:(1) Common source
875: names; (2) PAH 6.2 micron peak to continuum ratio; (3) [FeII]/[NeII] line ratio; (4) Logarithm of the [SIII] 18.7 to 33.4 micron line ratio; (5)
876: Logarithm of the [SII] 6716 to 6731~\AA~line ratio. Sources for [SII] ratios (a): \cite{kong99}; (b) \cite{hoyos06}; (c) \cite{izo04}; (d)
877: \cite{izo02}; (e) \cite{ho97}; (e) \cite{ho97}; (e) \cite{vacca92}.}
878: \end{deluxetable*}
879:
880:
881:
882:
883:
884:
885:
886:
887:
888:
889:
890:
891: \end{document}
892: