0710.0003/ms.tex
1: %% Referee manuscript produces a one-column, double-spaced document:
2: %\documentclass[12pt,manuscript]{aastex}
3: 
4: %% preprint produces a one-column, single-spaced document:
5: %\documentclass[12pt,preprint]{aastex}
6: % preprint2 produces a double-column, single-spaced document:
7: %\documentclass[12pt,preprint]{aastex}
8: 
9: %% Emulate ApJ document - for a nice 2 column compact form
10: \documentclass{emulateapj}
11: 
12: 
13: \usepackage{amssymb}
14: \usepackage{graphicx}
15: 
16: 
17: \slugcomment{Accepted by ApJ, September 28th, 2007}
18: \shorttitle{Elemental Abundances of BCDs}
19: \shortauthors{Wu et al.}
20: 
21: 
22: \begin{document}
23: 
24: 
25: \title{Elemental Abundances of Blue Compact Dwarfs from mid-IR Spectroscopy with Spitzer}
26: 
27: \author{Yanling Wu\altaffilmark{1}, J. Bernard-Salas\altaffilmark{1},
28:   V. Charmandaris\altaffilmark{2,3}, V.  Lebouteiller\altaffilmark{1},
29:   Lei Hao\altaffilmark{1}, B.~R. Brandl\altaffilmark{4}, J.~R. Houck\altaffilmark{1}}
30: 
31: \altaffiltext{1}{Astronomy Department, Cornell University, Ithaca, NY 14853}
32: 
33: \altaffiltext{2}{University of Crete, Department of Physics, GR-71003,
34: Heraklion, Greece}
35: 
36: \altaffiltext{3}{IESL/Foundation for Research and Technology - Hellas,
37:   GR-71110, Heraklion, Greece and Chercheur Associ\'e, Observatoire de
38:   Paris, F-75014, Paris, France}
39: 
40: \altaffiltext{4}{Leiden Observatory, Leiden University, P.O. Box 9513,
41: 2300 RA Leiden, The Netherlands}
42: 
43: 
44: 
45: \email{wyl@astro.cornell.edu, jbs@isc.astro.cornell.edu,
46:   vassilis@physics.uoc.gr, vianney@isc.astro.cornell.edu, 
47:   haol@astro.cornell.edu, brandl@strw.leidenuniv.nl,
48:   jrh13@cornell.edu}
49: 
50: 
51: 
52: \begin{abstract}
53:   We present a study of elemental abundances in a sample of thirteen
54:   Blue Compact Dwarf (BCD) galaxies, using the $\sim$10--37$\mu$m high
55:   resolution spectra obtained with Spitzer/IRS. We derive the
56:   abundances of neon and sulfur for our sample using the infrared
57:   fine-structure lines probing regions which may be obscured by dust
58:   in the optical and compare our results with similar infrared studies
59:   of starburst galaxies from ISO.  We find a good correlation between
60:   the neon and sulfur abundances, though sulfur is under-abundant
61:   relative to neon with respect to the solar value. A comparison of
62:   the elemental abundances (neon, sulfur) measured from the infrared
63:   data with those derived from the optical (neon, sulfur, oxygen)
64:   studies reveals a good overall agreement for sulfur, while the
65:   infrared derived neon abundances are slightly higher than the
66:   optical values.  This indicates that either the metallicities of
67:   dust enshrouded regions in BCDs are similar to the optically
68:   accessible regions, or that if they are different they do not
69:   contribute substantially to the total infrared emission of the host
70:   galaxy.
71: 
72: \end{abstract}
73: 
74: \keywords{dust, extinction ---
75:   galaxies:starburst ---
76:   abundances}
77: 
78: 
79: \section{Introduction}
80: 
81: Blue Compact Dwarf Galaxies (BCDs) are dwarf galaxies with blue
82: optical colors resulting from one or more intense bursts of
83: star-formation, low luminosities (M$_B>-$18) and small sizes. The
84: first BCD discovered was I\,Zw\,18 by \citet{Zwicky66}, which had the
85: lowest oxygen abundance observed in a galaxy \citep{Searle72}, until
86: the recent study of the western component of SBS0335-052
87: \citep{Izotov05}.  Although BCDs are defined mostly by their
88: morphological parameters, they are globally found to have low heavy
89: element abundances as measured from their HII regions
90: (1/30\,Z$_\odot$$\sim$1/2\,Z$_\odot$). The low metallicity of BCDs is
91: suggestive of a young age since their interstellar medium is
92: chemically unevolved. However, some BCDs do display an older stellar
93: population and have formed a large fraction of their stars more than
94: 1Gyr ago \citep[see][]{Loose85,Aloisi07}. The plausible scenario that
95: BCDs are young is intriguing within the context of Cold Dark Matter
96: models which predict that low-mass dwarf galaxies, originating from
97: density perturbations much less massive than those producing the
98: larger structures, can still be forming at the current epoch. However,
99: despite the great success in detecting galaxies at high redshift over
100: the past few years, bona fide young galaxies still remain extremely
101: difficult to find in the local universe
102: \citep{Kunth86,Kunth00,Madden06}. This is likely due to the
103: observational bias of sampling mostly luminous more evolved galaxies
104: at high redshifts. If some BCDs are truly young galaxies, they would
105: provide an ideal local laboratory to understand the galaxy formation
106: processes in the early universe.
107: 
108: Over the past two decades, BCDs have been studied extensively in many
109: wavelengths using ground-based and space-born instruments \citep [for
110: a review see][]{Kunth00}.  In the FUV, the Far Ultraviolet
111: Spectroscopic Explorer (FUSE) has been used to study the chemical
112: abundances in the neutral gas in several BCDs \citep{Thuan02,
113:   Aloisi03, Lebouteiller04}. Optical spectra have been obtained for a
114: large number of BCDs and display strong narrow emission lines
115: resulting from the intensive star-formation processes that take place
116: in these systems \citep{Izotov97, Izotov99b, Pustilnik05, Salzer05}.
117: The Infrared Space Observatory (ISO) revealed unexpectedly that
118: despite their low metallicities, BCDs, such as SBS\,0335-052E, could
119: still have copious emission from dust grains \citep{Thuan99, Madden00,
120:   Madden06, Plante02}. More recently, the Spitzer Space Telescope
121: \citep{Werner04} has been used to observe these metal-poor dwarf
122: systems in order to study their dust continuum properties and the
123: polycyclic aromatic hydrocarbon (PAH) features \citep{Houck04b,
124:   Hogg05, Engelbracht05, Rosenberg06, Wu06, OHalloran06, Hunt06,
125:   Wu07}. Finally, radio observations have also been performed for
126: several BCDs to study their HI kinematics and distribution
127: \citep{Thuan04} and thermal/non-thermal continuum emission properties
128: \citep{Hunt05}.
129: 
130: \begin{deluxetable*}{lrllllcc}
131:   \tabletypesize{\scriptsize}
132:   \setlength{\tabcolsep}{0.02in}
133:   \tablecaption{Observing Parameters of the Sample\label{tab1}}
134:   \tablewidth{0pc}
135:   \tablehead{
136:     \colhead{Object Name} & \colhead{RA} & \colhead{Dec} & \colhead{AORKEY} &
137:     \colhead{Observation} & \colhead{Redshift} & \multicolumn{2}{c}{On-source Time (Sen)}\\
138:     \colhead{} & \colhead{(J2000)} & \colhead{(J2000)} & \colhead{} & \colhead{Date} &  \colhead{} & 
139:     \colhead{SH} & \colhead{LR} \\
140:   }
141:   \startdata
142:   
143: Haro11              & 00h36m52.5s  & -33d33m19s &  9007104 & 2004-07-17  &  0.0206 & 480     & 240      \\
144: NGC1140             & 02h54m33.6s  & -10d01m40s &  4830976 & 2004-01-07  &  0.0050 & 480     & 240      \\
145: SBS0335-052E        & 03h37m44.0s  & -05d02m40s & 11769856 & 2004-09-01  &  0.0135 & 1440    & 960      \\
146: NGC1569             & 04h30m47.0s  & +64d50m59s &  9001984 & 2004-03-01  & $\sim$ 0& 480     & 240      \\
147: IIZw40              & 05h55m42.6s  & +03d23m32s &  9007616 & 2004-03-01  &  0.0026 & 480     & 240      \\
148: UGC4274             & 08h13m13.0s  & +45d59m39s & 12076032 & 2004-10-23  &  0.0015 & 120     &  56      \\ 
149:                     &              &            & 12626688 & 2004-11-11  &         & 120     &  56      \\ 
150: IZw18               & 09h34m02.0s  & +55d14m28s &  9008640 & 2004-03-27  &  0.0025 & 480     & 240      \\
151:                     &              &            & 16205568 & 2005-12-16  &         & 2880    & 1440     \\
152: VIIZw403            & 11h27m59.9s  & +78d59m39s &  9005824 & 2004-12-09  & $\sim$ 0& 480     & 240      \\
153: Mrk1450             & 11h38m35.6s  & +57d52m27s &  9011712 & 2004-12-12  &  0.0032 & 480     & 240      \\
154: UM461               & 11h51m33.3s  & -02d22m22s &  9006336 & 2005-01-03  &  0.0035 & 480     & 240      \\
155:                     &              &            & 16204032 & 2006-01-14  &         & 1440    & \nodata  \\
156: SBS1210+537A        & 12h12m55.9s  & +53d27m38s &  8989952 & 2004-06-06  & \nodata & 480     & 240      \\
157: Tol1214-277         & 12h17m17.1s  & -28d02m33s &  9008128 & 2004-06-28  &  0.0260 & 480     & 240      \\
158: Tol65               & 12h25m46.9s  & -36d14m01s &  4829696 & 2004-01-07  &  0.0090 & 480     & 240      \\
159: UGCA292             & 12h38m40.0s  & +32d46m01s &  4831232 & 2004-01-07  &  0.0010 & 480     & 240      \\
160: Tol1304-353         & 13h07m37.5s  & -35d38m19s &  9006848 & 2004-06-25  &  0.0140 & 480     & 240      \\
161: Pox186              & 13h25m48.6s  & -11d37m38s &  9007360 & 2004-07-14  &  0.0039 & 480     & 240      \\
162: CG0563              & 14h52m05.7s  & +38d10m59s &  8992512 & 2005-05-30  &  0.0324 & 240     & 120      \\
163: CG0598              & 14h59m20.6s  & +42d16m10s &  8992256 & 2005-03-19  &  0.0575 & 480     & 240      \\
164: CG0752              & 15h31m21.3s  & +47d01m24s &  8991744 & 2005-03-19  &  0.0211 & 480     & 240      \\
165: Mrk1499             & 16h35m21.1s  & +52d12m53s &  9011456 & 2004-06-05  &  0.0090 & 480     & 240      \\
166: {\rm [RC2]}A2228-00 & 22h30m33.9s  & -00d07m35s &  9006080 & 2004-06-24  &  0.0052 & 480     & 240      \\
167: 
168: \enddata
169: 
170: \tablecomments{The coordinates and redshifts of the objects are cited from the NASA/IPAC Extragalactic Database (NED).
171:   In this paper, we only include the analysis of thirteen out of twenty-two sources which have SNRs sufficient for our 
172:   abundance study. CG0563, CG0598 and CG0752 are included in the original sample as BCD candidates, however, they appear 
173:   to be more starburst like \citep[see][]{Hao07}. Thus even though they have high SNR, we do not include these three sources 
174:   in this study.}
175: \end{deluxetable*}
176: 
177: 
178: 
179: Metallicity is a key parameter that influences the formation and
180: evolution of both stars and galaxies. Detailed studies of the
181: elemental abundances of BCDs have already been carried out by several
182: groups \citep{Izotov99b, Kniazev03, Shi05} and the well known
183: metallicity-luminosity relation has also been studied in detail in the
184: environment of dwarf galaxies \citep{Skillman89, Hunter99,
185:   Melbourne02}. However, because these studies were performed in the
186: optical, they were limited by the fact that the properties of some of
187: the deeply obscured regions in the star-forming galaxies may remain
188: inaccessible due to dust extinction at these wavelengths. In fact,
189: \citet{Thuan99}, using ISO, have shown that the eastern component of
190: SBS\,0335-052 does have an embedded super star cluster (SSC) that is
191: invisible in the optical while contributing $\sim$75\% to the
192: bolometric luminosity \citep[see also][]{Plante02,Houck04b}, even
193: though it has very low metallicity (12+log(O/H)=7.33), which would in
194: principle imply a low dust content. In addition to probing the dust
195: enshrouded regions, emission in the infrared also has the advantage
196: that the lines accessible at these wavelengths are less sensitive to
197: the electron temperature fluctuations than the corresponding optical
198: lines of the same ion. In the infrared, more ionization stages of an
199: element become available as well. The improved sensitivity of the
200: Infrared Spectrograph (IRS\footnote{The IRS was a collaborative
201:   venture between Cornell University and Ball Aerospace Corporation
202:   funded by NASA through the Jet Propulsion Laboratory and the Ames
203:   Research Center.})  \citep{Houck04a} on Spitzer has enabled us to
204: obtain for the first time infrared spectra for a much larger sample of
205: BCDs than was previously possible \citep{Thuan99, Madden00, Verma03,
206:   Martin06}, thus motivating this study to probe the heavy element
207: abundances in BCDs.
208: 
209: In this paper, we analyze Spitze/IRS spectra of thirteen BCDs and
210: present elemental abundances of neon and sulfur, which are both
211: primary elements produced by the same massive stars in the nuclear
212: synthesis processes. In section 2, we describe the sample selection,
213: observations and data reduction. We present our results on the
214: chemical abundances in section 3, followed by a comparison of the
215: optical and infrared derived abundances in section 4. We show the
216: interplay between the abundances and PAH emission in section 5.
217: Finally, we summarize our conclusions in section 6.
218: 
219: 
220: \section{Observations and Data Reduction}
221: 
222: As part of the {\em IRS} Guaranteed Time Observation (GTO) program, we
223: have compiled a large sample of BCD candidates selected from the
224: Second Byurakan Survey (SBS), Bootes void galaxies
225: \citep{Kirshner81,Popescu00}, and other commonly studied BCDs. Details
226: on the low-resolution spectra of the sample have been published by Wu
227: et al. (2006).
228: 
229: We acquired the targets using the red (22$\mu$m) IRS peak-up camera in
230: high accuracy mode to locate the mid-IR centroid of the source and
231: then offset to the appropriate slit using the standard IRS staring
232: observing mode. Subsequently, we obtained the 10$-$37\,$\mu$m spectra
233: for our sources using the {\em IRS} Short-High (SH, 9.9-19.6\,$\mu$m)
234: and Long-high (LH, 18.7-37.2\,$\mu$m) modules. No spectra of the
235: background were obtained. The AORkey and on-source integration time of
236: the twenty-two sources for which we obtain high-resolution
237: spectroscopy are given in Table \ref{tab1}.  In this paper, we focus
238: on the description and analysis of thirteen BCDs from our program
239: which have a signal-to-noise ratio (SNR) high enough for our elemental
240: abundance study.
241: 
242: The data were processed by the {\em Spitzer} Science Center (SSC)
243: pipeline version 13.2. The three-dimensional data cubes were converted
244: to two-dimensional slope images after linearization correction,
245: subtraction of darks, and cosmic-ray removal. The reduction of the
246: spectral data started from intermediate pipeline products ``droop''
247: files, which only lacked stray light and flat-field correction.
248: Individual pointings to each nod position of the slit were co-added.
249: The data from SH and LH were extracted using the full slit extraction
250: method of a script version of {\em IRS} data analysis package SMART
251: \citep{Higdon04} from the median of the combined images. The
252: 1-dimensional spectra were flux calibrated by multiplying by a
253: relative spectral response function (RSRF), which was created from the
254: {\em IRS} standard star, $\xi$ Dra, for which accurate templates were
255: available \citep{Cohen03,Sloan07}. As a final cosmetic step the ends
256: of each order where the noise increases significantly were manually
257: clipped. No scaling was needed between the adjacent orders within the
258: same module. For faint sources, the spectra taken at nod position 1
259: are severely affected by fringing problems, thus we only use the data
260: at nod position 2 for these sources in our study.
261: 
262: 
263: The fine-structure lines, [SIV]\,$\lambda$10.51$\mu$m,
264: [NeII]\,$\lambda$12.81$\mu$m, [NeIII]\,$\lambda$15.55$\mu$m and
265: [SIII]\,$\lambda\lambda$18.71$\mu$m, 33.48$\mu$m are clearly present
266: in the majority of the BCDs in our sample\footnote {See Fig. 3 of
267:   \citet{Wu06}, or \citet{Hao07} for the reduced spectra.}. We measure
268: the line fluxes by fitting them with a Gaussian profile. Even though
269: the [SIII] line is visible in both SH and LH we only use the
270: 18.71\,$\mu$m line for deriving the ionic abundance. This is because
271: by using lines within the same module (SH), we remove uncertainties
272: due to the different sizes of the SH and LH slits. Furthermore, the
273: [SIII]\,$\lambda$33.48$\mu$m line is near the cut off of the LH module
274: where the sensitivity drops dramatically, making its measurement more
275: uncertain in some cases.
276: 
277: 
278: 
279: \section{Elemental Abundances of BCDs}                                        
280: 
281: In this section, we derive the neon and sulfur abundances using the
282: new infrared data of our BCD sample. The elemental abundances of the
283: compact HII regions in our Galaxy \citep{Martin02,Simpson04}, in the
284: Magellanic Clouds \citep{Vermeij02}, as well as in starburst galaxies
285: \citep{Verma03, Martin06} have already been studied extensively in the
286: past using ISO, and more recently with {\em Spitzer}
287: \citep{Rubin07}.  There are several methods to derive the chemical
288: abundances \citep [for a review see][]{Stasinska07}. Here we use an
289: empirical method, which derives ionic abundances directly from the
290: observed lines of the relavent ions.  To do so, we need to have the
291: flux of at least one hydrogen recombination line, the dust extinction,
292: as well as the electron temperature and density of the interstellar
293: medium (ISM).
294: 
295: 
296: \subsection{Electron density and temperature}
297: 
298: The electron density (N$_e$) could in principle be determined by
299: comparing the measured ratio of [SIII]18.71$\mu$m/33.48$\mu$m to the
300: expected theoretical value using the corresponding $S$-curve
301: \citep[see][]{Houck84}. However, the values for the ratio are in the
302: horizontal part of the $S$-curve and as a result we cannot use these
303: two lines to accurately constrain the electron density of these
304: systems.  Furthermore, the 33.48\,$\mu$m [SIII] line is located at the edge
305: of the LH slit, where the sensitivity drops dramatically, and
306: considering that these emission lines are weak, the measured line flux
307: for the 33.48\,$\mu$m [SIII] has a large uncertainty. Since the infrared
308: determination of elemental abundances does not depend strongly on the
309: density, we adopted the optically derived electron densities from the
310: literature, which range from 10 to 3000\,cm$^{-3}$ (see Table \ref
311: {tab2}). For sources where such information is not available, we adopt
312: a typical electron density of 100\,cm$^{-3}$ in this paper.
313: 
314: \begin{deluxetable*}{lccccccrc}
315:   \tabletypesize{\scriptsize}
316:   \setlength{\tabcolsep}{0.02in}
317:   \tablecaption{Optical Properties of the Sources\label{tab2}} 
318:   \tablewidth{0pc}
319:   \tablehead{
320:     \colhead{Object} & \multicolumn{4}{c}{F(H$\beta$) ($\times$10$^{-14}$ergs~s$^{-1}$cm$^{-2}$)} &
321:     \colhead{E$_{B-V}$} & \colhead{T$_e$} & \colhead{N$_e$} & \colhead{ref} \\
322:     \cline{2-5}
323:     \colhead{} &  \colhead{Hu$\alpha$-derived} & \colhead{H$\alpha$-derived} & \colhead{radio-derived} &
324:     \colhead{optical} & \colhead{mag} & \colhead{K} & \colhead{cm$^{-3}$} & \colhead{} \\
325:   }
326:   \startdata
327:   
328:   Haro11        & 252$\pm$40  & \nodata    & \nodata    & \nodata    & 0.41  & 13700 & 10  & (1)\\
329:   NGC1140       &  67$\pm$9   & \nodata    & \nodata    & \nodata    & 0.10  & 10000 & 100 & (2)\tablenotemark{a} \\
330:   SBS0335-052E  & $<$22.4     & \nodata    &6.5$\pm$0.06& \nodata    & 2.76  & 20000 & 200 & (3),(4),(5)\\
331:   NGC1569       & 393$\pm$19  & \nodata    & \nodata    & \nodata    & 0.65  & 12000 & 100 & (6)\\ 
332:   IIZw40        & 363$\pm$10  & 282$\pm$22 & \nodata    & \nodata    & 0.79  & 13000 & 190 & (2),(7) \\
333:   UGC4274       & $<$59.6     &11.5$\pm$1.0& \nodata    & \nodata    & 0     & 10000 & 62  & (7),(8)\tablenotemark{a}\\
334:   IZw18         & $<$12.6     & 8.0$\pm$0.6& 6.1$\pm$0.6& \nodata    & 0.08  & 19000 & 100 & (7),(9),(10)\\
335:   VIIZw403      & $<$24.6     &11.8$\pm$1.2& \nodata    & \nodata    & 0     & 14800 & 100 & (7),(11)\\
336:   Mrk1450       & $<$14.8     &15.8$\pm$1.4& \nodata    & \nodata    & 0.10  & 12500 & 100 & (7),(12)\\
337:   UM461         & $<$21.0     & \nodata    & \nodata    &13.6$\pm$0.6& 0.08  & 16100 & 200 & (13),(14)\\
338:   Tol1214-277   & 23.3$\pm$2.8& \nodata    & \nodata    & \nodata    & 0.03  & 19790 & 400 & (15)\\  
339:   Tol65         & $<$18.2     & 9.3$\pm$0.4& \nodata    & \nodata    & 0.08  & 17320 & 50  & (7),(15)\\ 
340:   Mrk1499       & 14.6$\pm$1.9&15.1$\pm$1.3& \nodata    & \nodata    & 0.17  & 12600 & 3267& (7),(16)\\ 
341:   \enddata
342: 
343:   \tablecomments{When no data are available in the literature, the 
344:     electron temperature and densities are assumed to be 10\,000\,K (
345:     for NGC1140 and UGC4274) and 100~cm$^{-3}$ (for NGC1140 and NGC1569) 
346:     respectively. When the Hu\,$\alpha$ line is detected, the H$\beta$
347:     flux is preferentially derived from this line. The upper limits 
348:     are listed for the non-detections. For the remaining sources,
349:     we derived the H$\beta$ flux either from the thermal component of the
350:     radio continuum, or from the extinction corrected H$\alpha$ flux
351:     inside the SH slit. For UM461, the H$\beta$ flux is for the whole 
352:     galaxy from the integrated optical spectra.}  
353:   \tablerefs{(1) \citet{Bergvall02}, (2) \citet{Guseva00},
354:     (3) \citet{Izotov06}, (4) \citet{Hunt04}, (5) \citet{Houck04b},
355:     (6) \citet{Kobulnicky97}, (7) \citet{Gildepaz03}, (8)
356:     \citet{Ho97}, (9) \citet{Cannon05}, (10) \citet{Izotov99a}, (11)
357:     \citet{Izotov97}, (12) \citet{Izotov94}, (13) \citet{Moustakas06},
358:     (14) \citet{Izotov98}, (15) \citet{Izotov01}, (16) \citet{Kong02}
359:     }
360:    
361: \end{deluxetable*}
362: 
363: 
364: Since BCDs in general have low metallicities, it is expected that BCDs
365: would have relatively high electron temperatures (T$_e$). Fortunately,
366: the infrared lines are much less sensitive to the uncertainties in the
367: electron temperature compared to the optical \citep{Bernard-Salas01}.
368: As a result we adopt electron temperatures derived from the optical
369: studies found in the literature. For sources which do not have direct
370: measurements (which happen to be high metallicity sources in our
371: sample, e.g. NGC1140, UGC4274), we use a representative temperature of
372: T$_e$\,=\,10,000\,K (see Table \ref {tab2}).
373: 
374: 
375: \subsection{Ionized hydrogen flux estimates}
376: 
377: In order to derive the ionic abundances, we also need to obtain
378: information on at least one hydrogen recombination line (usually
379: H$\beta$). For sources in which the Humphreys\,$\alpha$ (Hu$\alpha$,
380: 12.37\,$\mu$m) line is detected, we use it to convert to H$\beta$.
381: From the extinction corrected Hu$\alpha$ flux, we estimate the
382: H$\beta$ flux using the values given in the tables of \citet{Hummer87}
383: for Case B recombination. This has the advantage that we do not need
384: to correct for aperture effects because the neon and sulfur lines we use
385: to derive the abundances also reside in the same module (SH). The
386: 12.37\,$\mu$m line is generally very weak and it is detected in six of
387: our thirteen sources. For SBS0335-052E and IZw18, which have radio
388: continuum data, we calculate the H$\beta$ flux from the thermal
389: free-free emission using the fraction included in the aperture of our
390: SH slit at the time of the observation, and no correction for
391: extinction is needed. For the remaining four sources which have
392: H$\alpha$ images available from \citet{Gildepaz03}, we overlay the SH
393: slit on each H$\alpha$ image to calculate what fraction of the
394: H$\alpha$ emission is included in our slit. Then we derive the
395: corresponding H$\beta$ flux inside the SH slit from the
396: extinction-corrected H$\alpha$ flux. Finally, for UM461, we use the
397: H$\beta$ flux of the whole galaxy from the integrated spectra
398: \citep{Moustakas06}. UM461 is a very compact source, thus fully
399: encompassed within the SH slit.
400: 
401: \subsection{Extinction correction}
402: 
403: All our measurements, both the fluxes from the H$\alpha$ images, as
404: well as the infrared lines, have been corrected for extinction. From
405: the low-resolution spectra of our BCDs \citep{Wu06}, we find that most
406: of the sources do not show a strong 9.7\,$\mu$m silicate feature, thus
407: indicating an intrinsically relatively low dust extinction. Due to the
408: faintness of our targets at $\lambda<$9.7\,$\mu$m, the determination
409: of the continuum on the blue side of the 9.7\,$\mu$m feature is often
410: poor and the uncertainties in estimating the mid-IR extinction from
411: this are rather large \citep[see discussion in][]{Spoon07}. Thus for
412: sources where optical spectra are available, we adopt the E$_{\rm
413:   B-V}$ values calculated from the hydrogen recombination lines in the
414: literature. Only for SBS\,0335-052E, which has a high quality mid-IR
415: spectrum and evidence for an embedded SSC, the optical E$_{B-V}$ is
416: not a good estimate of its extinction, thus we use the extinction
417: estimated from the depth of the silicate feature \citep{Houck04b}.
418: Throughout this paper, we adopt the \citet{Fluks94} extinction law,
419: though using the \citet{Draine03} law would produce very similar
420: results. The E$_{\rm B-V}$ magnitudes are provided in Table
421: \ref{tab2}.
422: 
423: \subsection{Neon and sulfur  abundances determination}
424: 
425: We use the electron density and temperature given in Table \ref
426: {tab2}, together with the fine structure lines detected in the {\em
427:   IRS} high-resolution spectra, to calculate the abundances of neon
428: and sulfur. The equation used to determine the ionic abundance is
429: described in \citet{Bernard-Salas01} and it is:
430: 
431: \begin{equation}
432:   \frac{N_{\mathrm{ion}}}{N_{\mathrm{p}}}= \frac{I_{\mathrm{ion}}}{I_
433:     {\mathrm{H_{\beta}}}} N_{\mathrm{e}}
434:   \frac{\lambda_{\mathrm{ul}}}{\lambda_{\mathrm{H_{\beta}}}} \frac
435:   {\alpha_{\mathrm{H_{\beta}}}}{A_{\mathrm{ul}}}
436:   \left( \frac{N_{\mathrm{u}}}{N_{\mathrm{ion}}} \right)^{-1}
437:   \label{eq_abun_ch2}
438: \end{equation}
439: 
440: where I$_{\mathrm{ion}}$/I$_{\mathrm{H\beta}}$ is the measured flux of
441: the fine-structure line normalized to H$\beta$; N$_{\mathrm{P}}$ is
442: the density of the ionized hydrogen; $\lambda_{\mathrm{ul}}$ and
443: $\lambda_{\mathrm{H\beta}}$ are the wavelengths of the line and
444: H$\beta$; $\alpha_{\mathrm{H\beta}}$ is the effective recombination
445: coefficient for H$\beta$; A$_{\mathrm{ul}}$ is the Einstein
446: spontaneous transition rate for the line and N$_\mathrm
447: u$/N$_{\mathrm{ion}}$ is the ratio of the population of the level from
448: which the line originates to the total population of the ion. This
449: ratio is determined by solving the statistical equilibrium equation
450: for a five level system and normalizing the total number of ions to be
451: unity \citep{Osterbrock89}. The effective collisional strengths used
452: to derive the population of levels were obtained from the appropriate
453: reference of the IRON project
454: \citep{Hummer93}\footnote{http://www.usm.uni-muenchen.de/people/ip/iron-project.html}.
455: 
456: \begin{deluxetable*}{lrrrrrrrrr}
457: \tabletypesize{\scriptsize}
458:   \setlength{\tabcolsep}{0.02in}
459:   \tablecaption{Fine-structure Line Fluxes and Ionic Abundances\label{tab3}}
460:   \tablewidth{0pc}
461:   \tablehead{
462:     \colhead{} & \multicolumn{4}{c}{Line Fluxes ( $\times10^{-14}$ergs~cm$^{-2}$~s$^{-1}$)} & \colhead{} &\multicolumn{4}{c}{Ionic Abundances ($\times10^{-6}$)}\\
463:     \cline{2-5}\cline{7-10}
464:    \colhead{Object} & \colhead{[SIV](10.51$\mu$m)} & \colhead{[NeII](12.81$\mu$m)} & \colhead{[NeIII](15.55$\mu$m)} 
465:     & \colhead{[SIII](18.71$\mu$m)} & \colhead{} & \colhead{N$_{\rm SIV}$/N$_p$} & \colhead{N$_{\rm NeII}$/N$_p$} & \colhead{N$_{\rm NeIII}$/N$_p$}& 
466:     \colhead{N$_{\rm SIII}$/N$_p$} \\
467:   }
468:   \startdata  
469:   Haro11        & 43.95$\pm$0.14  &   31.23$\pm$0.23  & 98.14$\pm$0.65  &   45.24$\pm$0.25  & & 0.353$\pm$0.056 &    15.4$\pm$2.4   &  23.2$\pm$3.7   &    1.89$\pm$0.30  \\
470:   NGC1140       & 12.14$\pm$0.13  &   11.41$\pm$0.20  & 37.79$\pm$0.32  &   20.04$\pm$0.11  & & 0.375$\pm$0.051 &    22.5$\pm$3.0   &  35.5$\pm$4.8   &    3.56$\pm$0.48 \\
471:   SBS0335-052E  &  1.56$\pm$0.03  &   $<$0.14         &  1.43$\pm$0.02  &    0.43$\pm$0.03  & & 0.527$\pm$0.011 &   $<$2.27         &  11.2$\pm$1.6   &   0.535$\pm$0.038 \\
472:   NGC1569       &147.26$\pm$0.55  &   15.70$\pm$0.17  &175.85$\pm$0.59  &   72.70$\pm$0.23  & & 0.817$\pm$0.040 &    5.28$\pm$0.26  &  28.2$\pm$1.4   &    2.12$\pm$0.10  \\
473:   IIZw40        &185.72$\pm$1.73  &    6.24$\pm$0.11  &112.65$\pm$0.83  &   45.80$\pm$0.21  & &  1.12$\pm$0.03  &    2.23$\pm$0.07  &  19.3$\pm$0.6   &    1.37$\pm$0.04  \\
474:   UGC4274       &  4.18$\pm$0.12  &    8.87$\pm$0.11  & 12.84$\pm$0.12  &   12.63$\pm$0.11  & & 0.743$\pm$0.068 &     102$\pm$9     &  70.3$\pm$6.1   &    13.2$\pm$1.2 \\  
475:   IZw18         &  0.48$\pm$0.03  &    0.09$\pm$0.01  &  0.46$\pm$0.02  &    0.23$\pm$0.02  & & 0.125$\pm$0.015 &    1.41$\pm$0.21  &  3.56$\pm$0.38  &   0.276$\pm$0.036 \\
476:   VIIZw403      &  0.99$\pm$0.05  &   $<$0.142        &  0.88$\pm$0.02  &    0.87$\pm$0.02  & & 0.311$\pm$0.035 &   $<$2.79         &  8.41$\pm$0.88  &    1.39$\pm$0.14  \\
477:   Mrk1450       &  7.34$\pm$0.04  &    1.34$\pm$0.03  &  9.61$\pm$0.04  &    4.37$\pm$0.03  & & 0.928$\pm$0.223 &   10.70$\pm$2.57  &  36.9$\pm$8.9   &    2.98$\pm$0.71  \\
478:   UM461         &  4.58$\pm$0.04  &    0.16$\pm$0.02  &  2.83$\pm$0.03  &   0.812$\pm$0.03  & & 0.636$\pm$0.029 &    1.38$\pm$0.18  &  11.8$\pm$0.5   &   0.563$\pm$0.032  \\
479:   Tol1214-277   &  0.88$\pm$0.02  &   $<$0.10         &  0.61$\pm$0.02  &   $<$0.19         & & 0.061$\pm$0.007 &   $<$0.413        &  1.23$\pm$0.15  &   $<$0.059         \\
480:   Tol65         &  0.69$\pm$0.02  &   $<$0.14         &  0.90$\pm$0.02  &    0.32$\pm$0.02  & & 0.129$\pm$0.007 &   $<$1.62         &  5.07$\pm$0.25  &   0.293$\pm$0.022 \\
481:   Mrk1499       &  1.74$\pm$0.02  &    1.88$\pm$0.03  &  5.03$\pm$0.03  &    2.91$\pm$0.02  & & 0.286$\pm$0.037 &    16.5$\pm$2.2   &  21.6$\pm$2.8   &    2.28$\pm$0.30  \\
482:   \enddata
483:   \tablecomments{The observed line fluxes are measured from the IRS high resolution spectra of the sources. Background emission has not been subtracted, but 
484:     this does not affect the flux of the fine-structure lines.}
485: \end{deluxetable*}
486: 
487: The most important ionization stages of neon and sulfur are available
488: in the infrared. We detected [SIV]\,$\lambda$10.51\,$\mu$m and
489: [NeIII]\,$\lambda$15.56\,$\mu$m in all of our BCDs while the
490: [NeII]\,$\lambda$12.81\,$\mu$m line is detected in nine objects and
491: [SIII]\,$\lambda$18.71\,$\mu$m in twelve sources (see Table \ref
492: {tab3}). For sulfur, SII has its strongest emission lines in the
493: optical.  \citet{Vermeij02} have shown that SII is typically less than
494: 10\% of SIII in the Magellanic cloud HII regions they studied. Because
495: BCDs are typically high-excitation objects, we do not expect to have a
496: significant contribution from SII. The optical studies reveal that the
497: ionic abundance of SII is typically 10$-$20\% of that of SIII
498: \citep{Izotov94,Izotov97}, thus we keep in mind that this might result
499: in a $\sim$10\% underestimate in our infrared derived sulfur
500: abundance.  This is addressed further in the following subsection where
501: we discuss the properties of the individual objects. The presence of
502: the [OIV]\,$\lambda$25.89\,$\mu$m line in most of our sources (except for
503: VIIZw403, UM461, Tol65 and Mrk1499) indicates that some [NeIV] might
504: be present, even though the [OIV] line is typically weak. For example,
505: in Tol1214-277, where [OIV] is detected, the flux of
506: [NeIV]$\lambda$4725 from optical spectra is less than 1\% of that of
507: [NeIII]$\lambda$3868 \citep{Izotov04}, indicating that the
508: contribution of [NeIV] is not significant. \citet{Tsamis05} also have
509: shown in their photoionization modelling study of 30 Doradus that the
510: contribution from NeIV to the total neon abundance is much less than
511: 1\%.  As a result, rather than applying ionization correction factors
512: (ICFs) which could introduce an unknown systematic uncertainty in our
513: study, we sum the corresponding ionic abundances for the most
514: important ionization stages to determine the total element abundance
515: and the results are shown in Table \ref{tab3}.
516: 
517: Two successive stages of ionization of a given element (X) can be used
518: to measure the state of the ionization of the ISM, which, on first
519: order, depends on the ionization parameter, U, and the hardness of the
520: ionizing radiation \citep{Vilchez88}. For a given U, the ratio
521: X$^{+i+1}$/X$^{+i}$ is an indicator of the hardness of the stellar
522: radiation field. Here we use the ionic pairs of Ne$^{++}$/Ne$^+$ and
523: S$^{+3}$/S$^{++}$ as indicators of the hardness of the ionization field
524: and plot them in Fig. 1. It is clear from the figure that these two
525: ratios increase proportionally.  This indicates that the hardening of
526: the radiation field affects similarly the various elements in the full
527: range of the ionizing continua.
528: 
529: \begin{figure}
530:   \epsscale{1.2}
531:   \plotone{f1.eps}
532:   \caption{A plot of Ne$^{++}$/Ne$^+$ vs S$^{3+}$/S$^{++}$, which
533:     traces the hardness of the radiation field. The solid line is a
534:     fit to the data.}
535:   \label{fig:fig1}
536: \end{figure}
537: 
538: 
539: The major uncertainty in our study of the elemental abundances
540: originates from the H$\beta$ fluxes used as references. For sources
541: where the 12.37\,$\mu$m Hu$\alpha$ line is detected, we have no
542: uncertainties from matching the slit apertures, and the extinction
543: correction is very small. For sources where we estimate the H$\beta$
544: flux from the thermal radio continuum emission, we are free of
545: extinction corrections, even though we still need to properly account
546: for the radio emission which falls within the IRS slit. For BCDs with
547: no Hu\,$\alpha$ emission or radio data available in the literature, we
548: estimate the H$\beta$ flux using archival H$\alpha$ images of the objects.
549: This is clearly more challenging because in addition to the aperture
550: corrections, the extinction correction is higher which could introduce
551: a systematic uncertainty as large as 25\%. In the following analysis, we
552: indicate the sources with or without the detection of Hu\,$\alpha$
553: line using different symbols on the plots (see Fig. 2, 3 and 4). The
554: measurement uncertainties in the infrared lines are typically only a few
555: percent, but could grow to $\sim$10\% in the case of fainter sources.
556: The uncertainties we list in Table \ref{tab3} only account for measurement
557: uncertainties.
558: 
559: \begin{deluxetable*}{lrrrrrrr}
560: \tabletypesize{\scriptsize}
561:   \setlength{\tabcolsep}{0.02in}
562:   \tablecaption{Elemental Abundances\label{tab4}}
563:   \tablewidth{0pc}
564:   \tablehead{
565:     \colhead{Object} & \colhead{Ne/H$ (\times$10$^{-6}$)} &
566:   \colhead{S/H ($\times$10$^{-6}$)} & \colhead{O/H ($\times$10$^{-5}$)} & 
567:     \colhead{Ne (Z$_\odot$)} & \colhead{S (Z$_\odot$)} & \colhead{O (Z$_\odot$)} & \colhead{ref (O/H)} \\ 
568:   }
569:   \startdata  
570:   Haro11        &  38.6$\pm$4.4  & 2.24$\pm$0.31  & 7.94  &  0.33$\pm$0.04 & 0.16$\pm$0.02  & 0.16  & (1) \\
571:   NGC1140       &  58.0$\pm$5.7  & 3.94$\pm$0.48  & 28.8  &  0.48$\pm$0.05 & 0.28$\pm$0.03  & 0.59  & (2) \\
572:   SBS0335-052E  &  [11.2, 13.5]\tablenotemark{a}  & 1.06$\pm$0.04 & 1.95$\pm$0.09 &  [0.09, 0.11]\tablenotemark{a} & 0.08$\pm$0.003 & 0.03$\pm$0.001 & (3) \\
573:   NGC1569       &  33.5$\pm$1.4  & 2.94$\pm0.11$  & 15.8$\pm$0.5  &  0.28$\pm$0.01 & 0.21$\pm$0.01  & 0.32$\pm$0.01 & (4) \\
574:   IIZw40        &  21.5$\pm$0.6  & 2.49$\pm$0.05  & 12.2$\pm$0.4  &  0.17$\pm$0.005& 0.18$\pm$0.004 & 0.25$\pm$0.01 & (2) \\
575:   UGC4274       &  172.3$\pm$10.9&13.94$\pm$1.20  & 33.3  &  1.4$\pm$0.1   & 1.0$\pm$0.1    & 0.68  & (5) \\  
576:   IZw18         &  5.0$\pm$0.4   & 0.40$\pm$0.04  & 1.47$\pm$0.09 &  0.04$\pm$0.003& 0.03$\pm$0.003 & 0.03$\pm$0.002 & (6) \\
577:   VIIZw403      &  [8.4, 11.2]\tablenotemark{a}   & 1.70$\pm$0.14 & 4.90$\pm$0.11  &  [0.07, 0.09]\tablenotemark{a}  & 0.12$\pm$0.01  & 0.10$\pm$0.002 & (7) \\
578:   Mrk1450       &  47.6$\pm$9.2  & 3.98$\pm$0.74  & 9.55$\pm$0.11 &  0.40$\pm$0.08 & 0.28$\pm$0.06  & 0.19$\pm$0.002 & (8) \\
579:   UM461         &  13.2$\pm$0.5  & 1.20$\pm$0.04  & 6.10$\pm$0.40 &  0.11$\pm$0.01 & 0.09$\pm$0.003 & 0.12$\pm$0.01 & (9) \\
580:   Tol1214-277   &  [1.2, 1.6]\tablenotemark{a}    & [0.06,0.12]\tablenotemark{b} & 3.45$\pm$0.10 & [0.010, 0.013]\tablenotemark{a} & [0.004, 0.008]\tablenotemark{b} & 0.07$\pm$0.002 & (10) \\
581:   Tol65         &  [5.1, 6.7]\tablenotemark{a}    & 0.42$\pm$0.07 & 3.48$\pm$0.10 &  [0.04,0.06]\tablenotemark{a}   & 0.03$\pm$0.002 & 0.07$\pm$0.02 & (10) \\
582:   Mrk1499       &  38.1$\pm$3.6  & 2.57$\pm$0.30  & 13.2 &  0.32$\pm$0.03 & 0.18$\pm$0.02  & 0.27 & (11) \\
583:   \enddata
584:   \tablecomments{We adopt the following values for the solar abundances: (Ne/H)$_\odot$=1.2$\times$10$^{-4}$, (S/H)$_\odot$=1.4$\times$10$^{-5}$ and 
585:     (O/H)$_\odot$=4.6$\times$10$^{-4}$ (See section 4.1).}
586:   \tablenotetext{a}{We provide a range for the neon abundance by taking as the lower value the ionic abundance of NeIII, 
587:     while to obtain the upper value we add the upper limit for NeII.}
588:   \tablenotetext{b}{We provide a range for the sulfur by taking as the lower value the ionic abundance of SIV,
589:     while to obtain the upper value we add the upper limit of SIII.}
590:   \tablerefs{(1) \citet{Bergvall02}, (2) \citet{Guseva00},
591:     (3) \citet{Izotov06}, (4) \citet{Kobulnicky97}, (5) \citet{Ho97}, 
592:     (6) \citet{Izotov99a}, (7) \citet{Izotov97}, (8) \citet{Izotov94}, 
593:     (9) \citet{Izotov98}, (10) \citet{Izotov01}, (11) \citet{Kong02}
594:     }
595:    
596:    
597: \end{deluxetable*}
598: 
599: 
600: 
601: \subsection{Individual Objects}
602: 
603: Some details on the individual objects and how we measured the
604: necessary parameters are provided in the following paragraphs.
605: 
606: 
607: {\bf Haro11:} The 12.37\,$\mu$m Hu$\alpha$ line is clearly detected in
608: the SH spectrum.  The source has multiple nuclei, designated as A, B
609: and C \citep{Bergvall02}.  The optical spectra from which the oxygen
610: abundance was derived corresponds to the central region. Because all
611: the lines we are using are from the SH spectrum, we are relatively
612: free from extinction problems and there is no need for aperture
613: correction. The contribution of SII may add $\sim$10\% to the total
614: sulfur abundance.
615: 
616: {\bf NGC1140:} As with Haro11, the 12.37\,$\mu$m Hu$\alpha$ line is
617: also clearly detected in the SH spectrum of NGC1140.  No measurement
618: was found in the literature for the electron temperature or density
619: for this source, thus we assume 10000\,K and 100\,cm$^{-3}$
620: respectively, which are typical for BCDs. However, because its oxygen
621: abundance is more than half solar, the electron temperature could also
622: be lower. If we use a T$_e$ of 5000\,K and re-derive the abundances,
623: we find that the neon abundance would increase from 5.8$\times$
624: 10$^{-5}$ to 8.5$\times$10$^{-5}$ and the sulfur abundance from
625: 3.9$\times$10$^{-6}$ to 5.5$\times$10$^{-6}$. The contribution of SII
626: may add $\sim$10\% to the total sulfur abundance.
627: 
628: {\bf SBS0335-052E:} We do not detect the 12.37\,$\mu$m Hu$\alpha$ line
629: in the SH spectrum.  We use the thermal component of the 5\,GHz radio
630: continuum from \citet{Hunt04} to convert to the H$\beta$ flux, and
631: thus no extinction correction is needed. [NeII] is not detected and we
632: provide the upper limit in Table \ref{tab3}. The ionic abundance of
633: SII is $\sim$25\% of the SIII determined from the optical study of
634: \citet{Izotov06} and may add 12\% to the total sulfur abundance.  We
635: also find higher neon and sulfur abundances compared with the oxygen
636: and possible implications are discussed in \citet{Houck04b}.
637: 
638: {\bf NGC1569:} The 12.37\,$\mu$m Hu$\alpha$ line is clearly detected
639: in the SH spectrum, thus providing a direct estimate of the H$\beta$
640: flux inside the SH slit. This galaxy is extended in the mid-IR
641: \citep[see Fig. 4 in][]{Wu06} and optical spectra have been taken for
642: several of the bright knots by \citet{Kobulnicky97} in order to study
643: the chemical gradient and inhomogeneities of the source. These authors
644: found very little variation in the metallicity. Our infrared-derived
645: abundances are in rough agreement with the optical results. The ionic
646: abundance of SII is $\sim$13\% that of the SIII estimated from the
647: optical study \citep{Kobulnicky97} and may add 9\% to the total sulfur
648: abundance.
649: 
650: {\bf IIZw40:} We use the 12.37\,$\mu$m Hu$\alpha$ line to convert to
651: H$\beta$ flux. The derived neon and sulfur abundances agree with each
652: other but appear to be lower than oxygen with respect to the solar
653: values. If we estimate the H$\beta$ flux from the H$\alpha$ image of
654: \citet{Gildepaz03}, we find a value $\sim$22\% lower than the first,
655: which would increase the neon and sulfur abundances by 22\%
656: accordingly. This suggests, that when using the H$\beta$ flux derived
657: from H$\alpha$ image, the systematic error in our measurement is
658: probably no better than 25\%.  The ionic abundance of SII is
659: $\sim$13\% that of SIII from the optical study of \citet{Guseva00} and
660: may add 7\% to the total sulfur abundance.
661: 
662: {\bf UGC4274:} We use an H$\beta$ flux derived from the H$\alpha$
663: image of \citet{Gildepaz03}. Similarly to NGC1569, the galaxy is
664: extended and the peak of the infrared centroid is displaced from the
665: optical peak position. The derived neon and sulfur abundances are both
666: super solar. The oxygen abundance is not directly available from the
667: literature, thus we use the [NII]/H$\alpha$ method \citep{Ho97,
668:   Denicolo02} to derive O/H, which bears a large uncertainty. The
669: contribution of SII may add $\sim$10\% to the total sulfur abundance.
670: 
671: {\bf IZw18:} The H$\beta$ flux is derived from the thermal component
672: of the radio continuum inside the SH slit.  The ionic abundance of SII
673: is $\sim$21\% that of SIII from the optical study of \citet{Izotov99a}
674: and may add 15\% to the total sulfur abundance. A more detailed
675: discussion on this object can be found in \citet{Wu07}.
676: 
677: {\bf VIIZw403:} The H$\beta$ flux is derived by using the H$\alpha$
678: image from \citet{Gildepaz03}.  The optical and infrared centroids do
679: not overlap. The SH spectrum for this source is noisy and
680: [NeII]\,$\lambda$12.81\,$\mu$m is not detected, thus we have only a lower limit
681: on the neon abundance.  The upper limit of the [NeII] line indicates
682: that it could add less than 25\% to the total elemental abundance of
683: neon which is presented in Table \ref{tab4}. The ionic abundance of
684: SII is $\sim$24\% that of SIII from the optical work of
685: \citet{Izotov97} and may add 20\% to the total sulfur abundance.
686: 
687: {\bf Mrk1450:} As with VIIZw403, we use the hydrogen flux as given by
688: the H$\alpha$ flux included in the SH slit.  Neon and sulfur
689: abundances appear to be higher than the oxygen abundance with respect
690: to the solar values. The ionic abundance of SII is $\sim$18\% that of
691: SIII from the optical study of \citet{Izotov94} and may add 12\% to
692: the total sulfur abundance.
693: 
694: {\bf UM461:} The H$\beta$ flux is estimated from the integrated
695: spectra of \citet{Moustakas06}, and the source is small enough to be
696: fully included by the SH slit. The derived neon and sulfur abundances
697: are in good agreement with oxygen metallicity from the optical. The
698: ionic abundance of SII is $\sim$17\% that of SIII from the optical
699: \citep{Izotov98} and may add 8\% to the total sulfur abundance.
700: 
701: {\bf Tol1214-277:} The 12.37\,$\mu$m Hu$\alpha$ line is marginally
702: detected in the SH spectrum, and we use this line to derive the
703: H$\beta$ flux. The spectrum is very noisy and no
704: [NeII]\,$\lambda$12.81\,$\mu$m or [SIII]\,$\lambda$18.71\,$\mu$m lines
705: are visible. We use the ionic abundances of [NeIII] and [SIV] as the
706: lower limits of the neon and sulfur abundances.  The upper limit of
707: [NeII] amounts to 33\% of the ionic abundance of [NeIII] while the
708: upper limit of [SIII] is nearly as much as the abundance of [SIV].  We
709: give the range of their elemental abundances in Table \ref{tab4}. The
710: ionic abundance of SII is $\sim$18\% that of SIII from the optical
711: study of \citet{Izotov01} and may add $\sim$9\% to the total sulfur
712: abundance.
713: 
714: {\bf Tol65:} We use the H$\alpha$ image \citep{Gildepaz03} to
715: calculate the H$\beta$ flux. [NeII]\,$\lambda$12.81\,$\mu$m is not
716: detected in the SH spectrum, thus our derived neon abundance is a
717: lower limit. The upper limit of NeII is $\sim$32\% of the ionic
718: abundance of NeIII, and the range of neon abundance is provided in
719: Table \ref{tab4}.  The metallicity of sulfur relative to solar is
720: lower than oxygen. The ionic abundance of SII is $\sim$26\% that of
721: SIII from the optical \citep{Izotov01} and may add 19\% to the total
722: sulfur abundance.
723: 
724: {\bf Mrk1499:} The 12.37\,$\mu$m Hu$\alpha$ line is detected in the SH
725: spectrum. There was a bad pixel at the peak of this line and we
726: interpolated its value using adjacent pixel values. The infrared
727: centroid appears to be slightly shifted from the optical one. The
728: H$\beta$ flux inside the SH slit derived from the H$\alpha$ image is
729: very similar to that derived from the Hu$\alpha$ line (within
730: $\sim$10\%). The contribution of SII may add $\sim$10\% to the total
731: sulfur abundance.
732: 
733: \section{Discussion}
734: 
735: In this section, we compare the abundances derived from the IR spectra
736: with the results from the optical studies.
737: 
738: \subsection{Solar abundances}
739: 
740: The solar photospheric abundances have changed drastically over the
741: past few years \citep[see discussion in][]{Pottasch06}. In some cases,
742: the solar abundance estimates for the same element from different
743: authors could differ by a factor of two. In this paper, when we quote
744: the elemental abundance in solar units (e.g. Table \ref{tab4}), we
745: adopt the following values: (Ne/H)$_\odot$=1.2$\times$10$^{-4}$ from
746: \citet{Feldman03}, (S/H)$_\odot$=1.4$\times$10$^{-5}$ from
747: \citet{Asplund05}, and (O/H)$_\odot$=4.9$\times$10$^{-4}$ from
748: \citet{Prieto01}. We should note though that \citet{Asplund05} have
749: also reported that (Ne/H)$_\odot$=6.9$\times$10$^{-5}$ and
750: \citet{Grevesse98} have given (S/H)$_\odot$=2.1$\times$10$^{-5}$,
751: while \citet{Anders89} have reported that
752: (O/H)$_\odot$=8.5$\times$10$^{-4}$. These values represent the
753: extremes for the solar neon, sulfur and oxygen abundance
754: determinations.  \citet{Pottasch06} have shown that the higher neon
755: solar value is favored in their sample of planetary nebulae (PNe). In
756: the present paper we indicate the range of solar abundances in our
757: plots and we also discuss the impact of different solar values on our
758: results.
759: 
760: 
761: \subsection{Neon and sulfur abundances}
762: 
763: As discussed in \citet{Thuan95}, it is assumed --- and has been shown
764: in the optical --- that the $\alpha$-element-to-oxygen abundance
765: ratios do not vary with oxygen abundance. This is due to the fact that
766: those elements are produced by the same massive stars
767: (M$>$10M$_{\odot}$) responsible for oxygen production. We test this
768: assumption in the infrared by studying the abundances of neon and
769: sulfur as derived from our IRS data.
770: 
771: \begin{figure}
772:   \epsscale{1.2}
773:   \plotone{f2.eps}
774:   \caption{A plot of the Ne/H vs S/H abundance. Our BCDs are indicated by the
775:     filled circles if Hu\,$\alpha$ is detected and by diamonds if it is not detected. The sources 
776:     that are shown by only the error bars are those that have non-detection of [NeII] 
777:     or [SIII] (See note of Table \ref {tab4}). The starburst galaxies from \citet{Verma03} 
778:     are marked with the stars. We use the grey band to indicate the locus on the plot where the 
779:     ratio of Ne/S would be consistent with the different solar values for neon and sulfur abundances.}
780:   \label{fig:fig2}
781: \end{figure}
782: 
783: 
784: In Fig. 2, we plot the abundances of neon and sulfur as derived from
785: our high-resolution data. \citet{Verma03} have found a positive
786: correlation between the neon and argon abundances for their sample of
787: starburst galaxies while their data show no correlation between the
788: sulfur and neon and/or argon abundances (indicated as stars on Fig.
789: 2). However, our sources show that the neon and sulfur abundances
790: scale with each other. In the same figure, we also plot the
791: proportionality line of the ratio for (Ne/S)$_\odot$. The maximum and
792: minimum values of the solar neon and sulfur abundances are indicated
793: by the width of the grey band, and we find that most of our BCDs have
794: ratios above those values.
795: 
796: 
797: \begin{figure}
798:   \epsscale{1.2}
799:   \plotone{f3.eps}
800:   \caption{The abundance ratio of Ne/S as a function of Ne/H. The
801:     symbols are the same as in Fig. 1. The grey area indicates the
802:     range for the values of the Ne/S ratio encountered in the solar
803:     neighborhood.}
804:   \label{fig:fig3}
805: \end{figure}
806: 
807: 
808: In Fig. 3, we examine the ratio of Ne/S as a function of Ne/H derived
809: from the IRS data. We see no correlation between the ratio of Ne/S
810: with respect to Ne/H, in agreement with the results of
811: \citet{Thuan95}.  The average Ne/S ratio we found for these BCDs is
812: $\sim$11.4. This is similar with the Ne/S ratio of 14.3 in the Orion
813: Nebula \citep{Simpson04}.  The Ne/S ratios found in the HII regions of
814: M83 are higher ($\sim$24$-$42) and decrease with increasing
815: deprojected galactocentric radii \citep{Rubin07}.  These authors have
816: also found that the Ne/S ratios in M33 are $\sim$12$-$21
817: \citep{Rubin06}. We notice that all the BCDs, as well as the HII
818: regions in M33, M83 and the Orion Nebula, have larger Ne/S ratios than
819: those found in the solar neighborhood, while their Ne/S ratios from
820: optical measurements are more consistent with the solar ratios.  The
821: higher Ne/S ratios from infrared studies have already been found in
822: other type of objects, including PNe, starburst galaxies, and other
823: HII regions \citep{Marigo03,Verma03, Henry04,Pottasch06}. Some of
824: these works point out that the solar sulfur abundance could have been
825: underestimated and they also appear to favor the higher values for the
826: solar neon abundance.  It could also be due to a differential
827: depletion of sulfur onto dust grains compared to neon as suggested by
828: several studies \citep{Simpson90,Verma03,Pottasch06, Bernard-Salas07}.
829: 
830: \begin{deluxetable*}{lcccccccc}
831:   \tabletypesize{\scriptsize}
832:   \setlength{\tabcolsep}{0.02in}
833:   \tablecaption{PAH measurements of the sample\label{tab5}}
834:   \tablewidth{0pc}
835:   \tablehead{
836:     \colhead{Object Name} & \multicolumn{4}{c}{PAH EW ($\mu$m)} & \multicolumn{4}{c}{Integrated Flux ($\times10^{-14}$ergs~cm$^{-2}$~s$^{-1}$})\\
837:     \colhead{} & \colhead{6.2\,$\mu$m} & \colhead{7.7\,$\mu$m} & \colhead{8.6\,$\mu$m} & 
838:     \colhead{11.2\,$\mu$m} &  \colhead{6.2\,$\mu$m} & \colhead{7.7\,$\mu$m} & \colhead{8.6\,$\mu$m} & 
839:     \colhead{11.2\,$\mu$m} \\
840:   }
841:   \startdata
842:  Haro11      &   0.120$\pm0.003$  &  0.227$\pm0.005$  &  0.042$\pm0.003$  &   0.099$\pm0.002$  &  78.3$\pm1.8$     &    180$\pm4$       & 25.4$\pm1.6$      &   67.2$\pm1.3$ \\ 
843:  NGC1140     &   0.479$\pm0.033$  &  0.568$\pm0.024$  &  0.091$\pm0.006$  &   0.525$\pm0.032$  &  42.1$\pm1.4$     &   78.3$\pm1.8$     & 11.1$\pm0.7$      &   50.2$\pm2.5$ \\
844:  SBS0335-052E&$<$0.035            & \nodata           & \nodata           &$<$0.015            &  $<$1.7           &  \nodata           & \nodata           &   $<$1.0 \\
845:  NGC1569     &   0.232$\pm0.013$  &  0.380$\pm0.019$  &  0.007$\pm0.005$  &   0.129$\pm0.008$  &  50.4$\pm2.4$     &    118$\pm6$       &  2.7$\pm1.6$      &   68.1$\pm4.3$ \\
846:  IIZw40      &   0.044$\pm0.007$  &  0.044$\pm0.006$  &  0.006$\pm0.005$  &   0.033$\pm0.006$  &  13.2$\pm2.1$     &   19.0$\pm2.8$     &  2.8$\pm1.2$      &   21.1$\pm3.7$ \\
847:  UGC4274     &   0.423$\pm0.032$  &  0.497$\pm0.049$  &  0.105$\pm0.012$  &   0.495$\pm0.021$  &  23.2$\pm1.2$     &   43.3$\pm2.3$     &  7.7$\pm0.9$      &   29.1$\pm0.9$ \\ 
848:  IZw18       &$<$0.233            & \nodata           & \nodata           &$<$0.116            &  $<$0.4           &  \nodata           & \nodata           &   $<$0.1 \\  
849:  Mrk1450     &   0.207$\pm0.045$  &  0.337$\pm0.143$  &  0.084$\pm0.020$  &   0.112$\pm0.017$  &   1.8$\pm0.4$     &    2.8$\pm0.9$     &  0.7$\pm0.2$      &    1.6$\pm0.2$ \\ 
850:  UM461       &$<$0.663            & \nodata           & \nodata           & $<$0.134           &  $<$1.5           &  \nodata           & \nodata           &   $<$1.6 \\
851:  Mrk1499     &   0.408$\pm0.040$  &  0.526$\pm0.016$  &  0.156$\pm0.011$  &   0.721$\pm0.035$  &    5.5$\pm0.3$    &   10.9$\pm0.3$     &  2.3$\pm0.2$      &    7.3$\pm0.2$ \\
852: 
853:  \enddata \tablecomments{Contrary to \citet{Wu06}, the wavelengths
854:    which we have chosen to determine the underlying continuum are
855:    fixed in the above measurements, which explains the minor
856:    discrepancies between these two studies. The symbol `` \nodata''
857:    indicates that no PAH EW measurement was possible and it is mostly
858:    due to the low SNR of the corresponding spectrum. In this case, the
859:    determination of the continuum is highly uncertain and could
860:    significantly affect the value of the PAH EW. For SBS0335-052E, the
861:    SNR is high enough, but no PAH features can be identified in its
862:    mid-IR spectrum. We do not derive upper limits for the PAH EWs at
863:    7.7\,$\mu$m and 8.6\,$\mu$m, because no reliable templates for
864:    those features are available.}
865:   
866: \end{deluxetable*}
867: 
868: 
869: 
870: \subsection{Comparison with optically derived abundances}
871: 
872: As has been discussed earlier in this paper, one of the major
873: advantages of using IR lines for estimating abundances is that one can
874: probe emission from dust enshrouded regions without the uncertainties
875: introduced from extinction corrections.  In this subsection, we
876: compare the newly derived neon and sulfur abundances of our BCD sample
877: using the infrared data, with the abundances of the same elements as
878: well as those for oxygen estimated from optical studies available in
879: the literature.
880: 
881: In order to determine elemental abundances, the optical studies
882: usually adopt a two-zone photoionized HII region model: a
883: high-ionization zone with temperature T$_e$(OIII) and a low-ionization
884: zone with temperature T$_e$(OII). Because the infrared lines are much
885: less sensitive to variations in T$_{e}$ (which affect the calculation
886: on the population of the various atomic levels), in the subsequent
887: calculations, we assume a constant electron temperature for our
888: analysis.
889: 
890: 
891: \begin{figure}
892:   \epsscale{2.2}
893:   \plottwo{f4a.eps}{f4b.eps}
894:   \caption{a)The Ne/H vs O/H abundances. The symbols are the same as in Fig. 1. The grey band 
895:     represents the locus on the plot where sources of different metallicities would have a Ne/O
896:     intrinsic ratio similar to the values found in the solar neighborhood. 
897:     b) Same as a), but for S/H vs O/H abundances.}
898:   \label{fig:fig4}
899: \end{figure}
900: 
901: 
902: In Fig. 4 we compare the infrared derived neon and sulfur abundances
903: of the BCD sample with their optical metallicities, using the oxygen
904: abundances from the literature (see also Table \ref{tab4}).  The
905: infrared neon abundances with respect to solar are slightly higher
906: than the oxygen (see Fig. 4a), but overall, there is a good agreement
907: between the infrared measurements and the optical results. This
908: indicates that there is little dust enshrouded gas, or that these
909: regions have similar metallicities, or that they do not contribute much
910: to the total integrated emission of the galaxies. In SBS0335-052E, as
911: well as Mrk1450, UGC4274 and Haro11, the infrared derived metallicity
912: of neon is more than twice that of oxygen compared to the solar
913: values.  This could be due to the different regions we are probing
914: using the infrared, which could be obscured in the optical.
915: SBS0335-052E is known to have an embedded super star cluster that is
916: invisible in the optical \citep[see][]{Houck04b}. Both Haro11 and
917: UGC4274 are somewhat extended, and detailed spectroscopy over the
918: whole galaxy would be needed to directly compare the infrared results
919: with the optical ones. For Mrk1450, which is a point source for
920: Spitzer/IRS, the abundance we derived using the infrared data is
921: higher than the optical value and it is likely due to the fact that
922: the optical estimates are more susceptible to the uncertainties in the
923: electron temperature. If we lower T$_e$ from 12\,500\,K to 10\,000\,K,
924: the oxygen abundance calculated from the optical data would double,
925: while the neon abundance measured from the infrared would only
926: increase by 5\%. In Fig. 4b we show that the infrared derived sulfur
927: abundances agree well with the oxygen abundance relative to solar.
928: 
929: For nine out of the thirteen objects we studied, optically derived
930: neon and sulfur abundances are also available in the literature. The
931: optical data have matched apertures, in most cases an accurate T$_e$
932: measurement, a direct measure of the H$\beta$ flux, and reddening
933: derived from the same data.  Our infrared results are less sensitive
934: to the electron temperature, but affected by the uncertainty in the
935: H$\beta$ flux if the Hu\,$\alpha$ line is not detected and the
936: uncertainty could be as large as 25\%.  In Fig. 5a, we plot the
937: optically derived Ne/H against the infrared Ne/H. The solid line is
938: the 1:1 proportion line for the infrared and optical derived
939: abundances and we find that most of the sources are located above this
940: line. A possible explanation for the higher infrared derived neon
941: abundances could be the presence of dust enshrouded regions which
942: might have higher heavy element abundances. This could explain the
943: higher infrared metallicity found in SBS0335-052E, but if this were
944: true for the whole sample, it should also apply to the sulfur
945: abundances.  However, when we plot the S/H (Optical) against the S/H
946: (IR) in Fig. 5b, we find that the sources are located on both sides of
947: the 1:1 proportion line. Alternatively, this could be due to the
948: difference in the determination of element abundance using the
949: infrared and optical methods. Because only NeIII is detected in the
950: optical regime, the total elemental abundance of neon from the optical
951: study is heavily dependent on the adopted ICF. In the infrared the
952: [NeII] line is detected in most of our sources and we do not use any
953: ICF for our study. Thus the higher neon abundances derived from the
954: infrared data compared to the optical results could also be due to the
955: large uncertainty in the ICF used in the optical studies. Another
956: possibility is that the temperature of the NeIII ion is lower than
957: that of the OIII ion as found in some PNe \citep{Bernard-Salas02}.
958: Because the optical studies are based on T$_e$(OIII) when calculating
959: the ionic abundance of NeIII, that could also result in an
960: underestimate of its abundance.
961: 
962: \begin{figure}
963:   \epsscale{2.2}
964:   \plottwo{f5a.eps}{f5b.eps}
965:   \caption{a)The infrared derived Ne/H as a function of the optical
966:     Ne/H abundances. The sources shown with only the error bars are
967:     those where [NeII] or [SIIII] is not detected. The solid line 
968:     represents the 1:1 line for the optical and IR derived abundances.  
969:     b) Same as a), but for S/H (Optical) vs S/H (IR) abundances.}
970:   \label{fig:fig5}
971: \end{figure}
972: 
973: 
974: 
975: Finally, we explore the variation in the ratio of Ne/S as derived from
976: the infrared and the optical data, and plot them as a function of O/H
977: in Fig. 6. Interestingly, we find that with the exception of VIIZw403,
978: in all BCDs, the infrared derived Ne/S ratios are higher than the
979: corresponding optical values. \citet{Izotov99b} indicate an average
980: Ne/S ratio of 6.9 for the 54 super giant HII regions they study in 50
981: BCDs using optical spectroscopy. The average of the optically derived
982: Ne/S ratio for the nine sources in our sample is 6.5$\pm$1.8 while the
983: corresponding ratio from the infrared data is 11.4$\pm$2.9,
984: statistically higher than the optical results. As we mentioned
985: earlier, no SII abundances could be determined with the Spitzer/IRS
986: data.  However, in high-excitation objects such as BCDs, SII does not
987: contribute much ($\sim$10\%) to the total elemental abundance, thus it
988: could not account for the factor of two difference in the ratio of
989: Ne/S.  Furthermore, such a discrepancy is not a result of metallicity
990: because we see no correlation in the dispersion between the infrared and
991: optically derived ratios of Ne/S with respect to O/H. If we divide the
992: sources into a high-excitation group with sources that have a
993: detectable [OIV]$\lambda$25.89\,$\mu$m line and a low-excitation group
994: with sources without any detection of [OIV], we still see no clear
995: trend. Thus the ionization field cannot explain the difference in the
996: Ne/S ratios either. Moreover, no correlation is found between the Ne/S
997: ratio and the extinction to the source. Even though the largest
998: uncertainty in the infrared derived abundances comes from the
999: uncertainty in the H$\beta$ flux in cases where no Hu\,$\alpha$ is
1000: detected, when calculating the ratio of Ne/S, the H$\beta$ flux
1001: cancels out, thus our infrared determined values for Ne/S should be
1002: fairly reliable. One possibility for the observed discrepancy could be
1003: that the difference in the infrared and optical results is again due
1004: to the ICF adopted in the optical studies.
1005: 
1006: 
1007: \begin{figure}
1008:   \epsscale{1.2}
1009:   \plotone{f6.eps}
1010:   \caption{The abundance ratio of Ne/S as a function of O/H. The infrared derived Ne/S ratios are 
1011:     indicated by the filled circles while the optical derived Ne/S ratios are shown by the open 
1012:     circles. The dashed line is the average Ne/S ratio from the infrared derived elemental abundances 
1013:     while the dotted line is average Ne/S ratio from the optical derived elemental abundances. 
1014:     The dash-dotted line represents the average Ne/S ratio from \citet{Izotov99b}. }
1015:   \label{fig:fig6}
1016: \end{figure}
1017: 
1018: 
1019: 
1020: 
1021: \section{Metallicities and PAHs}
1022: 
1023: The observed suppression of PAH emission in low metallicity
1024: environment is still an open topic \citep{Madden00,Madden06,
1025:   Engelbracht05, Wu06, OHalloran06}.  One explanation on the absence
1026: of PAHs is that there is not enough carbon to form into PAH molecules
1027: in such metal-poor environments while an alternative possibility could
1028: be that the strong ionization field in low metallicity environments
1029: may destroy the fragile PAH rings.
1030: 
1031: Previous studies relating the PAHs and metallicity were all based on
1032: the oxygen abundance measured from the optical, which could be
1033: associated with regions that are different from the ones where PAH
1034: emission originates. Because we have obtained the infrared measured neon
1035: and sulfur abundances, it is particularly interesting to compare the
1036: PAH strength with our metallicity estimates. Several BCDs in our
1037: sample do show PAH emission features and an analysis of the PAH
1038: properties of our sample based on IRS low resolution spectra was
1039: presented in \citet{Wu06}. We have obtained deeper spectra for some of
1040: the sources, so for consistency, we have re-measured all features in
1041: our sample and the results are given in Table \ref{tab5}.  None of the
1042: sources with metallicity lower than $\sim$0.12\,Z$_\odot$ show any
1043: detectable PAHs. For the sources where PAHs were detected, we
1044: calculate the equivalent widths (EWs) of the 6.2,7.7,8.6 and
1045: 11.2\,$\mu$m features and find that the correlation between the PAH
1046: EWs and the metallicity is weak. This is not unexpected because as
1047: discussed in \citet{Wu06}, the strength of PAHs in low-metallicity
1048: environments is a combination of creation and destruction effects. As
1049: a result the dependency of PAH strength on the metallicity alone could
1050: have a substantial scatter.
1051: 
1052: 
1053: 
1054: 
1055: 
1056: \section{Conclusions}
1057: 
1058: We have studied the neon and sulfur abundances of thirteen BCDs using
1059: Spitzer/IRS high-resolution spectroscopy.  Our analysis was based
1060: on the fine-structure lines and the hydrogen recombination line
1061: detected in the SH spectra of the {\em IRS}, combined with the radio
1062: continuum, H$\alpha$ images and integrated optical spectral data in
1063: some cases. We find a positive correlation between the neon and sulfur
1064: abundances, though sulfur appears to be more under-abundant than neon
1065: (with respect to solar). The ratio of Ne/S for our sources is on
1066: average 11.4$\pm$2.9, which is consistent with what has been found in
1067: other HII regions using infrared data. However, this average ratio
1068: appears to be higher than the corresponding optical value of
1069: 6.5$\pm$1.8 (in BCDs), which could be due to the adopted ICFs in the
1070: optical studies. When comparing the newly derived neon and sulfur
1071: abundances with the oxygen abundances measured from the optical lines,
1072: we find a good overall agreement.  This indicates that there are few
1073: completely dust enshrouded HII regions in our BCDs, or if such HII
1074: regions are present, they have similar metallicities to the ones
1075: probed in the optical. Finally, the infrared derived neon and sulfur
1076: abundances also correlate, with some scatter, with the corresponding
1077: elemental abundances derived from the optical data.
1078: 
1079: 
1080: \acknowledgments We thank Robert Kennicutt whose detailed comments
1081: help to improve this manuscript. We also thank Shannon Gutenkunst and
1082: Henrik Spoon for helpful discussions as well as an anonymous referee
1083: whose insightful suggestions helped to improve this manuscript. This
1084: work is based in part on observations made with the Spitzer Space
1085: Telescope, which is operated by the Jet Propulsion Laboratory,
1086: California Institute of Technology, under NASA contract 1407. Support
1087: for this work was provided by NASA through Contract Number 1257184
1088: issued by JPL/Caltech.
1089: 
1090: 
1091: \acknowledgments 
1092: 
1093: 
1094: 
1095: \begin{thebibliography}{}
1096: 
1097: \bibitem[Anders \& Grevesse(1989)]{Anders89} 
1098:   Anders, E., \& Grevesse, N.\ 1989, \gca, 53, 197 
1099: 
1100: \bibitem[Allende Prieto et al.(2001)]{Prieto01} 
1101:   Allende Prieto, C., Lambert, D.~L., \& Asplund, M.\ 2001, \apjl, 556, L63 
1102: 
1103: \bibitem[Aloisi et al.(2007)]{Aloisi07}
1104:   Aloisi, A. et al.\ 2007, astro-ph/0702216, Proceedings of the IAU Symp. 241 
1105:   ``Stellar Populations as Building Blocks of Galaxies''
1106: 
1107: \bibitem[Aloisi et al.(2003)]{Aloisi03} 
1108:   Aloisi, A., Savaglio, S., Heckman, T.M., Hoopes, C.~G., Leitherer, C., \& Sembach, K.~R.\ 2003, \apj, 595, 760 
1109: 
1110: \bibitem[Asplund et al.(2005)]{Asplund05} 
1111:   Asplund, M., Grevesse, N., \& Sauval, A.~J.\ 2005, Cosmic Abundances as Records of Stellar Evolution and Nucleosynthesis, 336, 25 
1112: 
1113: \bibitem[Bergvall{\ \& Ouml}stlin(2002)]{Bergvall02} 
1114:    Bergvall, N., {\ \& Ouml}stlin, G.\ 2002, \aap, 390, 891  
1115: 
1116: \bibitem[Becker et al.(1995)]{Becker95} 
1117:   Becker, R.~H., White, R.~L., \& Helfand, D.~J.\ 1995, \apj, 450, 559 
1118: 
1119: \bibitem[Bernard Salas et al.(2007)]{Bernard-Salas07}
1120:   Bernard Salas, J., et al.\ 2007,  \apj, in press, astro-ph/0709.3292
1121: 
1122: \bibitem[Bernard Salas et al.(2002)]{Bernard-Salas02} 
1123:   Bernard Salas, J., Pottasch, S.~R., Feibelman, W.~A., \& Wesselius, P.~R.\ 2002, \aap, 387, 301 
1124: 
1125: \bibitem[Bernard Salas et al.(2001)]{Bernard-Salas01} 
1126:   Bernard Salas, J., Pottasch, S.~R., Beintema, D.~A., \& Wesselius, P.~R.\ 2001, \aap, 367, 949 
1127: 
1128: \bibitem[Cannon et al.(2005)]{Cannon05} 
1129:   Cannon, J.~M., Walter, F., Skillman, E.~D., \& van Zee, L.\ 2005, \apjl, 621, L21 
1130: 
1131: \bibitem[Cohen et al.(2003)]{Cohen03} 
1132:   Cohen, M., Megeath, S.~T., Hammersley, P.~L., Mart{\'{\i}}n-Luis, F., \& Stauffer, J.\ 2003, \aj, 125, 2645 
1133: 
1134: \bibitem[Denicol{\'o} et al.(2002)]{Denicolo02} 
1135:   Denicol{\'o}, G., Terlevich, R., \& Terlevich, E.\ 2002, \mnras, 330, 69 
1136: 
1137: \bibitem[Draine(2003)]{Draine03} 
1138:   Draine, B.~T.\ 2003, \araa, 41, 241 
1139: 
1140: \bibitem[Engelbracht et al.(2005)]{Engelbracht05} 
1141:   Engelbracht, C.~W., Gordon, K.~D., Rieke, G.~H., Werner, M.~W.,  Dale, D.~A., \& Latter, W.~B.\ 2005, \apjl, 628, L29
1142: 
1143: \bibitem[Feldman \& Widing(2003)]{Feldman03} 
1144:   Feldman, U., \& Widing, K.~G.\ 2003, Space Science Reviews, 107, 665 
1145: 
1146: \bibitem[Fluks et al.(1994)]{Fluks94} 
1147:   Fluks, M.~A., Plez, B., The, P.~S., de Winter, D., Westerlund, B.~E., \& Steenman, H.~C.\ 1994, \aaps, 105, 311
1148: 
1149: \bibitem[Gil de Paz et al.(2003)]{Gildepaz03} 
1150:   Gil de Paz, A., Madore, B.~F., \& Pevunova, O.\ 2003, \apjs, 147, 29 
1151: 
1152: \bibitem[Grevesse \& Sauval(1998)]{Grevesse98} 
1153:   Grevesse, N., \& Sauval, A.~J.\ 1998, Space Science Reviews, 85, 161 
1154: 
1155: \bibitem[Guseva et al.(2000)]{Guseva00} 
1156:   Guseva, N.~G., Izotov,    Y.~I., \& Thuan, T.~X.\ 2000, \apj, 531, 776 
1157: 
1158: \bibitem[Hao et al.(2007)]{Hao07}
1159:   Hao, L., et al., 2007, in preparation
1160: 
1161: 
1162: \bibitem[Henry et al.(2004)]{Henry04} 
1163:   Henry, R.~B.~C., Kwitter, K.~B., \& Balick, B.\ 2004, \aj, 127, 2284 
1164: 
1165: \bibitem[Higdon et al.(2004)]{Higdon04} 
1166:   Higdon, S.~J.~U., et al.\ 2004, \pasp, 116, 975 
1167: 
1168: \bibitem[Ho et al.(1997)]{Ho97} 
1169:   Ho, L.~C., Filippenko, A.~V., \& Sargent, W.~L.~W.\ 1997, \apjs, 112, 315 
1170: 
1171: \bibitem[Hogg et al.(2005)]{Hogg05} 
1172:   Hogg, D.~W., Tremonti, C.~A., Blanton, M.~R., Finkbeiner, D.~P., Padmanabhan, N., Quintero, A.~D., 
1173:   Schlegel, D.~J., \& Wherry, N.\ 2005, \apj, 624, 162 
1174: 
1175: \bibitem[Houck et al.(2004a)]{Houck04a} 
1176:   Houck, J.~R., et al.\ 2004a, \apjs, 154, 18 
1177: 
1178: \bibitem[Houck et al.(2004b)]{Houck04b} 
1179:    Houck, J.~R., et al.\ 2004b, \apjs, 154, 211 
1180: 
1181: \bibitem[Houck et al.(1984)]{Houck84} 
1182:   Houck, J.~R., Shure, M.~A., Gull, G.~E., \& Herter, T.\ 1984, \apjl, 287, L11 
1183: 
1184: \bibitem[Hummer et al.(1993)]{Hummer93} 
1185:   Hummer, D.~G., Berrington, K.~A., Eissner, W., Pradhan, A.~K., Saraph, H.~E., \& Tully, J.~A.\ 1993, \aap, 279, 298 
1186: 
1187: \bibitem[Hummer \& Storey(1987)]{Hummer87} 
1188:   Hummer, D.~G., \& Storey, P.~J.\ 1987, \mnras, 224, 801 
1189: 
1190: \bibitem[Hunt et al.(2006)]{Hunt06} 
1191:   Hunt, L.~K., Thuan, T.~X., Sauvage, M., \& Izotov, Y.~I.\ 2006, \apj, 653, 222 
1192: 
1193: \bibitem[Hunt et al.(2005)]{Hunt05} 
1194:   Hunt, L., Bianchi, S., \& Maiolino, R.\ 2005a, \aap, 434, 849 
1195: 
1196: \bibitem[Hunt et al.(2004)]{Hunt04} 
1197:    Hunt, L.~K., Dyer, K.~K., Thuan, T.~X., \& Ulvestad, J.~S.\ 2004, \apj, 606, 853 
1198: 
1199: \bibitem[Hunter \& Hoffman(1999)]{Hunter99} 
1200:   Hunter, D.~A., \& Hoffman, L.\ 1999, \aj, 117, 2789 
1201: 
1202: \bibitem[Izotov et al.(1994)]{Izotov94} 
1203:   Izotov, Y.~I., Thuan, T.~X., \& Lipovetsky, V.~A.\ 1994, \apj, 435, 647 
1204: 
1205: \bibitem[Izotov et al.(1997)]{Izotov97} 
1206:   Izotov, Y.~I., Thuan, T.~X., \& Lipovetsky, V.~A.\ 1997, \apjs, 108, 1 
1207: 
1208: \bibitem[Izotov \& Thuan(1998)]{Izotov98} 
1209:   Izotov, Y.~I., \& Thuan, T.~X.\ 1998, \apj, 500, 188 
1210: 
1211: \bibitem[Izotov et al.(1999a)]{Izotov99a} 
1212:   Izotov, Y.~I., Chaffee, F.~H., Foltz, C.~B., Green, R.~F., Guseva, N.~G., \& Thuan, T.~X.\ 1999, \apj, 527, 757 
1213: 
1214: \bibitem[Izotov \& Thuan(1999b)]{Izotov99b} 
1215:   Izotov, Y.~I., \& Thuan, T.~X.\ 1999, \apj, 511, 639 
1216: 
1217: \bibitem[Izotov et al.(2001)]{Izotov01} 
1218:   Izotov, Y.~I., Chaffee, F.~H., \& Green, R.~F.\ 2001, \apj, 562, 727 
1219: 
1220: \bibitem[Izotov et al.(2004)]{Izotov04} 
1221:   Izotov, Y.~I., Papaderos, P., Guseva, N.~G., Fricke, K.~J., \& Thuan, T.~X.\ 2004, \aap, 421, 539 
1222: 
1223: \bibitem[Izotov et al.(2005)]{Izotov05} 
1224:   Izotov, Y.~I., Thuan, T.~X., \& Guseva, N.~G.\ 2005, \apj, 632, 210 
1225: 
1226: 
1227: \bibitem[Izotov et al.(2006)]{Izotov06} 
1228:   Izotov, Y.~I., Schaerer, D., Blecha, A., Royer, F., Guseva, N.~G., \& North, P.\ 2006, \aap, 459, 71 
1229: 
1230: \bibitem[Kirshner et al.(1981)]{Kirshner81} 
1231:   Kirshner, R.~P., Oemler, A., Jr., Schechter, P.~L., \& Shectman, S.~A.\ 1981, \apjl, 248, L57 
1232: 
1233: \bibitem[Kniazev et al.(2003)]{Kniazev03} 
1234:   Kniazev, A.~Y., Grebel, E.~K., Hao, L., Strauss, M.~A., Brinkmann, J., \& Fukugita, M.\ 2003, \apjl, 593, L73 
1235: 
1236: \bibitem[Kobulnicky \& Skillman(1997)]{Kobulnicky97} 
1237:   Kobulnicky, H.~A., \& Skillman, E.~D.\ 1997, \apj, 489, 636 
1238: 
1239: \bibitem[Kong et al.(2002)]{Kong02} 
1240:   Kong, X., Cheng, F.~Z., Weiss, A., \& Charlot, S.\ 2002, \aap, 396, 503 
1241: 
1242: \bibitem[Kunth \& \"{O}stlin(2000)]{Kunth00} 
1243:   Kunth, D.~ \& \"{O}stlin, G.\ 2000, \aapr, 10, 1 
1244: 
1245: \bibitem[Kunth \& Sargent(1986)]{Kunth86} 
1246:   Kunth, D., \& Sargent, W.~L.~W.\ 1986, \apj, 300, 496 
1247: 
1248: \bibitem[Lebouteiller et al.(2004)]{Lebouteiller04} 
1249:   Lebouteiller, V., Kunth, D., Lequeux, J., Lecavelier des Etangs, A., D{\'e}sert, J.-M., 
1250:   H{\'e}brard, G., \& Vidal-Madjar, A.\ 2004, \aap, 415, 55 
1251: 
1252: \bibitem[Loose \& Thuan(1985)]{Loose85} 
1253:   Loose, H.-H., \& Thuan, T.~X.\ 1985, Star-Forming Dwarf Galaxies and Related Objects, 73 
1254: 
1255: \bibitem[Madden et al.(2006)]{Madden06} 
1256:   Madden, S.~C., Galliano, F., Jones, A.~P., \& Sauvage, M.\ 2006, \aap, 446, 877 
1257: 
1258: \bibitem[Madden(2000)]{Madden00}
1259:   Madden, S.~C.\ 2000, New Astronomy Review, 44, 249
1260: 
1261: \bibitem[Marigo et al.(2003)]{Marigo03} 
1262:   Marigo, P., Bernard-Salas, J., Pottasch, S.~R., Tielens, A.~G.~G.~M., \& Wesselius, P.~R.\ 2003, \aap, 409, 619 
1263: 
1264: \bibitem[Mart{\'{\i}}n-Hern{\'a}ndez et al.(2002)]{Martin02} 
1265:   Mart{\'{\i}}n-Hern{\'a}ndez, N.~L., et al.\ 2002, \aap, 381, 606 
1266: 
1267: \bibitem[Mart{\'{\i}}n-Hern{\'a}ndez et al.(2006)]{Martin06} 
1268:   Mart{\'{\i}}n-Hern{\'a}ndez, N.~L., Schaerer, D., Peeters, E., Tielens, A.~G.~G.~M., \& Sauvage, M.\ 2006, \aap, 455, 853 
1269: 
1270: \bibitem[Melbourne \& Salzer(2002)]{Melbourne02} 
1271:   Melbourne, J., \& Salzer, J.~J.\ 2002, \aj, 123, 2302 
1272: 
1273: \bibitem[Moustakas \& Kennicutt(2006)]{Moustakas06} 
1274:   Moustakas, J., \& Kennicutt, R.~C., Jr.\ 2006, \apjs, 164, 81 
1275: 
1276: \bibitem[O'Halloran et al.(2006)]{OHalloran06} 
1277:   O'Halloran, B., Satyapal, S., \& Dudik, R.~P.\ 2006, \apj, 641, 795 
1278: 
1279: \bibitem[Osterbrock(1989)]{Osterbrock89}
1280:   Osterbrock, D.~E.\ 1989, Astrophysics of gaseous nebulae and active galactic nuclei. Research supported by 
1281:   the University of California, John Simon Guggenheim Memorial Foundation, University of Minnesota, et al. 
1282:   Mill Valley, CA, University Science Books, 1989, 422p
1283: 
1284: \bibitem[Peimbert \& Torres-Peimbert(1992)]{Peimbert92} 
1285:   Peimbert, M., \& Torres-Peimbert, S.\ 1992, \aap, 253, 349 
1286: 
1287: \bibitem[Plante \& Sauvage(2002)]{Plante02} 
1288:   Plante, S., \& Sauvage, M.\ 2002, \aj, 124, 1995 
1289: 
1290: \bibitem[Popescu \& Hopp(2000)]{Popescu00} 
1291:   Popescu, C.~C., \& Hopp, U.\ 2000, \aaps, 142, 247
1292:  
1293: \bibitem[Pottasch \& Bernard-Salas(2006)]{Pottasch06} 
1294:   Pottasch, S.~R., \& Bernard-Salas, J.\ 2006, \aap, 457, 189 
1295: 
1296: \bibitem[Pustilnik et al.(2005)]{Pustilnik05} 
1297:   Pustilnik, S.~A., et al.\ 2005, \aap, 442, 109 
1298: 
1299: \bibitem[Rosenberg et al.(2006)]{Rosenberg06} 
1300:   Rosenberg, J.~L., Ashby, M.~L.~N., Salzer, J.~J., \& Huang, J.-S.\ 2006, \apj, 636, 742 
1301: 
1302: \bibitem[Rubin et al.(2007)]{Rubin07} 
1303:   Rubin, R.~H., et al.\ 2007, \mnras, 377, 1407 
1304: 
1305: \bibitem[Rubin et al.(2006)]{Rubin06}
1306:   Rubin, R.~H. et al.\ 2006, in {\em Galaxy Evolution Across the Hubble Time}, IAU Symposium 235, 
1307:   Eds.F. Combes \& J. Palous, (Cambridge U. Press) (in press)
1308: 
1309: \bibitem[Salzer et al.(2005)]{Salzer05} 
1310:   Salzer, J.~J., Jangren, A., Gronwall, C., Werk, J.~K., Chomiuk, L.~B., Caperton, K.~A., Melbourne, 
1311:   J., \& McKinstry, K.\ 2005, \aj, 130, 2584 
1312: 
1313: \bibitem[Searle \& Sargent(1972)]{Searle72} 
1314:   Searle, L., \& Sargent, W.~L.~W.\ 1972, \apj, 173, 25 
1315: 
1316: \bibitem[Shi et al.(2005)]{Shi05} 
1317:   Shi, F., Kong, X., Li, C., \& Cheng, F.~Z.\ 2005, \aap, 437, 849 
1318: 
1319: \bibitem[Simpson \& Rubin(1990)]{Simpson90} 
1320:   Simpson, J.~P., \& Rubin, R.~H.\ 1990, \apj, 354, 165 
1321: 
1322: \bibitem[Simpson et al.(2004)]{Simpson04} 
1323:   Simpson, J.~P., Rubin, R.~H., Colgan, S.~W.~J., Erickson, E.~F., \& Haas, M.~R.\ 2004, \apj, 611, 338 
1324: 
1325: \bibitem[Skillman et al.(1989)]{Skillman89} 
1326:   Skillman, E.~D., Kennicutt, R.~C., \& Hodge, P.~W.\ 1989, \apj, 347, 875 
1327: 
1328: \bibitem[Sloan et al.(2007)]{Sloan07}
1329:   Sloan,G., et al., 2007, in prep
1330: 
1331: \bibitem[Smith et al.(2007)]{Smith07} 
1332:   Smith, J.~D.~T., et al.\ 2007, \apj, 656, 770 
1333: 
1334: \bibitem[Spoon et al.(2007)]{Spoon07} 
1335:   Spoon, H.~W.~W., Marshall, J.~A., Houck, J.~R., Elitzur, M., Hao, L., Armus, L., Brandl, B.~R., 
1336:   \& Charmandaris, V.\ 2007, \apjl, 654, L49 
1337: 
1338: \bibitem[Stasi{\'n}ska (2007)]{Stasinska07}
1339:   Stasi{\'n'}ska, G.\ 2007, astro-ph/0704.0348, Lectures given at the XVIII Canary Island Winterschool 
1340:   ``The emission line Universe'', to be published by Cambridge University Press
1341: 
1342: \bibitem[Thuan et al.(2004)]{Thuan04} 
1343:   Thuan, T.~X., Hibbard, J.~E., \& L{\'e}vrier, F.\ 2004, \aj, 128, 617 
1344: 
1345: \bibitem[Thuan et al.(2002)]{Thuan02} 
1346:   Thuan, T.~X., Lecavelier des Etangs, A., \& Izotov, Y.~I.\ 2002, \apj, 565, 941 
1347: 
1348: \bibitem[Thuan et al.(1999)]{Thuan99} 
1349:   Thuan, T.~X., Sauvage, M., \& Madden, S.\ 1999, \apj, 516, 783 
1350: 
1351: \bibitem[Thuan et al.(1995)]{Thuan95}
1352:   Thuan, T.~X., Izotov, Y.~I., \& Lipovetsky, V.~A.\ 1995, \apj, 445, 108
1353: 
1354: \bibitem[Tsamis \& P{\'e}quignot(2005)]{Tsamis05} 
1355:   Tsamis, Y.~G., \& P{\'e}quignot, D.\ 2005, \mnras, 364, 687 
1356: 
1357: \bibitem[Verma et al.(2003)]{Verma03} 
1358:   Verma, A., Lutz, D., Sturm, E., Sternberg, A., Genzel, R., \& Vacca, W.\ 2003, \aap, 403, 829 
1359: 
1360: \bibitem[Vermeij et al.(2002)]{Vermeij02} 
1361:   Vermeij, R., Peeters, E., Tielens, A.~G.~G.~M., \& van der Hulst, J.~M.\ 2002, \aap, 382, 1042 
1362: 
1363: \bibitem[Vilchez \& Pagel(1988)]{Vilchez88} 
1364:   Vilchez, J.~M., \& Pagel, B.~E.~J.\ 1988, \mnras, 231, 257 
1365: 
1366: \bibitem[Werner et al.(2004)]{Werner04}
1367:   Werner, M., et al.\ 2004, \apjs, 154,1
1368: 
1369: \bibitem[Wu et al.(2006)]{Wu06} 
1370:   Wu, Y., Charmandaris, V., Hao, L., Brandl, B.~R., Bernard-Salas, J., Spoon, H.~W.~W., \& Houck, J.~R.\ 2006, \apj, 639, 157 
1371: 
1372: \bibitem[Wu et al.(2007)]{Wu07}
1373:   Wu, Y., et al.\ 2007,  \apj, 662, 952 
1374: 
1375: \bibitem[Zwicky(1966)]{Zwicky66} 
1376:   Zwicky, F.\ 1966, \apj, 143, 192 
1377: 
1378: 
1379: \end{thebibliography}
1380: 
1381: \end{document}
1382: 
1383: 
1384: 
1385: