0810.2540/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: 
3: %% manuscript produces a one-column, double-spaced document:
4: 
5: % \documentclass[manuscript]{aastex}
6: 
7: %% preprint2 produces a double-column, single-spaced document:
8: 
9: % \documentclass[preprint2]{aastex}
10: 
11: 
12: 
13: %% You can insert a short comment on the title page using the command below.
14: 
15: \slugcomment{}
16: 
17: 
18: 
19: \shorttitle{SBS0335-052E Super Star Clusters}
20: \shortauthors{Thompson et al.}
21: 
22: 
23: 
24: \begin{document}
25: 
26: 
27: 
28: \title{Super Star Clusters in SBS0335-052E}
29: 
30: 
31: \author{Rodger I. Thompson}
32: \affil{Steward Observatory, University of Arizona,
33:     Tucson, AZ 85721}
34: \email{rthompson@as.arizona.edu}
35: 
36: \author{Marc Sauvage}
37: \affil{Service d' Astrophysicque, CEA/DAPNIA, Centre d' Etudes de Saclay,}
38: \email{msauvage@cea.fr}
39: 
40: \author{Robert C. Kennicutt}
41: \affil{Institute of Astronomy, Cambridge University, Madingley Road, Cambridge,
42: CB3 0HA United Kingdom}
43: \email{robk@ast.cam.ac.uk}
44: 
45: \author{Charles Engelbracht}
46: \affil{Steward Observatory, University of Arizona,
47:     Tucson, AZ 85721}
48: \email{cengelbracht@as.arizona.edu}
49: 
50: \author{Leonardo Vanzi}
51: \affil{Pontificia Universidad Catolica de Chile, Department of Electrical
52: Engineering, Av. Vicu~na Mackenna 4860,Santiago, Chile}
53: \email{lvanzi@ing.puc.cl}
54: 
55: \author{Glenn Schneider}
56: \affil{Steward Observatory, University of Arizona,
57:     Tucson, AZ 85721}
58: \email{gschneider@as.arizona.edu}
59: 
60: 
61: \begin{abstract}
62: 
63: As one of the lowest metallicity star forming galaxies, with a nucleus of several 
64: super star clusters, SBS0335-052E is the subject of substantial current study. We
65: present new insights on this galaxy based on new and archival high spatial
66: resolution NICMOS and ACS images.  We provide new measurements and limits on the
67: size of several of the SSCs. The images have sufficient resolution to divide
68: the star formation into compact regions and newly discovered extended regions,
69: indicating a bi-modal form of star formation.  The star formation regions are
70: dated via the equivalent width of the Pa$\alpha$ emission and we find that two of 
71: the extended regions of star formation are less than 10 million years old.
72: Our previous finding that stellar winds confine the photo-ionizing flux
73: to small regions around individual stars is consistent with the new observations.
74: This may allow planet formation in what would traditionally be considered
75: a harsh environment and has implications for the number of planets
76: around globular cluster stars.  In addition the images pinpoint the regions 
77: of H$_2$ emission as located in, but not at the center of the two star forming 
78: super star clusters, S1 and S2.
79: 
80: \end{abstract}
81: 
82: \keywords{galaxies:dwarf---galaxies:starburst---galaxies:star clusters---
83: galaxies:individual (SBS0335-052E)}
84: 
85: \section{Introduction} \label{s-i}
86: 
87: SBS0335-052 is a Blue Compact Dwarf galaxy at a redshift of 0.0136,
88: \citep{izo97}, a distance of 54.3 Mpc and with an apparent V magnitude of 16.65 
89: \citep{thu96}. The close group of super star clusters (SSCs) in 
90: the galaxy's center has been and currently is the subject of intense
91: scrutiny. \cite{thm06}, hereinafter TSKEV, provides a recent summary of 
92: previous observations and investigations of the SSCs in SBS0335-052E.
93: Part of the attention is due to the extremely low metallicity of 
94: the galaxy.  The metallicity, as measured by 12 + log(O/H), 
95: in SBS0335-052E ranges between 7.20 and
96: 7.31 \citep{pap06}.  SBS0335-052E has the fourth lowest metallicity of any 
97: star forming galaxy, surpassed only by its western component SBS0335-052W 
98: with the lowest measured metallicity,  J0956+2849, and the well known 
99: I Zw 18 \citep{izo07}. Because of its low metallicity and its now intense star 
100: formation, possibly due to an encounter with the nearby giant
101: spiral galaxy NGC 1376, it represents star formation and feedback 
102: in an environment very much like star formation in the early universe.  
103: SBS0335-052E is obviously in a very early stage of its star formation
104: since the observed stellar mass of $8 \times 10^7$ M$_{\sun}$ is only about
105: 1/25 of the gas mass of $2.1 \times 10^9$ M$_{\sun}$
106: which is again small relative to the dynamical mass of $9 \times 10^9$
107: M$_{\sun}$ \citep{pus01}, most of which must be dark matter.  
108: A main conclusion of TSKEV was
109: that the majority of photo-ionizing radiation from the stars in
110: the actively star forming SSCs is absorbed within a few
111: stellar radii thus delaying the onset of negative feedback due
112: to photo-ionization. This conclusion is validated in the present 
113: study under the assumption that the diffuse Pa$\alpha$ flux around
114: the SSCs S1 and S2 is due to in-situ stars and not from ionizing
115: flux leaking out of the SSCs.
116: 
117: The TSKEV study relied heavily on \emph{Hubble Space Telescope} Near Infrared 
118: Camera and Multi-Object Spectrometer (HST NICMOS) camera 3 observations of 
119: SBS0335-052E which  were part of the SINGS \citep{ken03} large survey
120: of nearby galaxies. The 0.2\arcsec pixel scale of camera 3 is not 
121: sufficient to resolve much of the detail of the feedback mechanism
122: on star formation operating in the clusters To get a higher spatial
123: resolution near infrared picture of the SSCs new observations with
124: the 0.075\arcsec pixel scale NICMOS camera 2 were taken in the
125: Pa$\alpha$ and H$_2$ filters.  The primary goal of this work is to
126: study in detail the feedback effect of the photoionizing flux generated
127: in the super star clusters on past and current star formation.  This study
128: also utilizes archival NICMOS camera 2 images in Pa$\alpha$ and the 
129: broad band filters F160W and F205W from observations taken by K.
130: Johnson (see \S~\ref{s-obs}).  Figure~\ref{fig-map} 
131: shows the location of all of the sources discussed
132: in this paper.  Sources S1-S6 are the sources originally designated 
133: from WFPC2 images in \citet{thu97}. The sources S3-Pa$\alpha$, S7 and
134: S8 are names ascribed by this work.  S3-Pa$\alpha$ is considered a
135: separate source from S3 as it is offset from the continuum source S3
136: and is most likely a completely separate source.  S7 is the
137: source that was designated by a question mark (?) in TSKEV and
138: S8 is diffuse source about $2.5\arcsec$ west of S1 and S2. Unfortunately
139: this source is marked as S7 in \citet{pap06} who did not have the
140: spatial resolution to resolve the source we designate as S7 as a 
141: separate source.  Unlike the NICMOS camera 3
142: images used in TSKEV the new, much higher resolution, NICMOS 
143: camera 2 images clearly resolve S1 and S2 and detect the separation
144: between S3 and S3-Pa$\alpha$.  Sources S3-Pa$\alpha$ and S7 only 
145: appear as slight extensions of the continuum emission in Figure~\ref{fig-map}
146: but are clearly shown in the Pa$\alpha$ contours in Figure~\ref{fig-mcon}.
147: The Advanced Camera for Surveys (ACS) image in Figure~\ref{fig-mcon} is 
148: an archival ACS solar blind channel (SBC) image in the UV F140LP filter 
149: \citep{kun03} which is also discussed in TSKEV.
150: 
151: \section{Observations} \label{s-obs}
152: 
153: The new observations presented in this paper were taken with the
154: HST NICMOS Camera 2 in 4 narrow band 
155: filters (F187N, F190N, F212N and F215N) centered at 1.87 $\micron$, 
156: 1.90 $\micron$, 2.12 \micron\ and 2.15 \micron\ (Proposal ID 10856). 
157: These filters are designed to measure Pa$\alpha$ and the H$_2$ S(1) 
158: lines in the F187N and F212N filters respectively.  The redshift of 
159: SBS0335-052, however, shifts these lines to the F190N and 
160: F215N filters with the shorter wavelength filters acting as the 
161: continuum filters.  The program executed in 8 orbits.  Exposures 
162: were taken in all 4 filters in each orbit without moving HST during
163: an orbit. During each earth occultation the telescope was dithered one step 
164: in a spiral dither pattern with a dither step of $0.5\arcsec$. The
165: F190N filter which images the strong Pa$\alpha$ emission had an 
166: integration time of 320 seconds (Step32, NSamp=17) and the remaining
167: filters all had integration times of 768 seconds (Step64, NSamp=20) in
168: each orbit for a total of 2560 seconds for the F190N filter and 6144
169: seconds in each of the remaining filters.  The field of view of the
170: NICMOS camera 2 is $19.5\arcsec$ by $19.3\arcsec$ which easily encompasses
171: all of the known SSCs.
172: 
173: Additional archival images in the F160W, F205W, F187N and F190N filters 
174: (K. Johnson Proposal ID 10894), \citep{rei08}, were also retrieved. There are
175: 8 images in the F160W and F205W filters.  The F160W images use STEP16
176: with 26 samples and the F205W images use STEP16 with 24 samples for
177: integration times of 304 and 272 seconds respectively.  There are 16
178: on target images in the F187N and F190N filters, again with STEP16 and
179: 26 samples for F187N and 25 samples for F190N for integration times of
180: 304 and 288 seconds. The total integration times of the archival images
181: are: F160W, 2432 seconds; F205W, 2176 seconds; F187N, 4864 seconds; 
182: F190N, 4600 seconds.
183: 
184: Archival ACS HRC and WFC1 images of SBS0335-052 from Proposal 10575
185: (PI Goran \"{O}stlin) in F220W, F330W, F550M, F435M filters and the FR656
186: ramp filter were also retrieved from the STScI MAST archive.  The 
187: combined drizzled images are used with no further processing.  These
188: images along with previous F122M and F140LP images from Proposal 9470 
189: (PI Daniel Kunth) have been used by \citet{ost08} to make H$\alpha$ 
190: and Ly$\alpha$ images. Table~\ref{tbl-obs} gives a summary of the 
191: observational parameters of the images used in this analysis.
192: 
193:  
194: Parallel to the NICMOS observations, images of the spiral galaxy 
195: NGC 1376 in 8 medium band filters were taken  with the ACS WFC. 
196: Two images in one of the 8 filters
197: were taken in each orbit. The ACS observations will be described in a 
198: separate publication. SBS0335-052 is probably engaged in a minor merger 
199: with NGC 1376 and the images were taken to see if there is any evidence
200: of the interaction.  No obvious signs of tidal disruption were seen in
201: the NGC 1376 images.  Given the small mass of SBS0335-052 relative to
202: NGC 1376 the absence of any disruption is not surprising. 
203: 
204: \section{Data Reduction} \label{s-dr}
205: 
206: All of the images were reduced with the NICMOS IDL based pipeline 
207: procedures operating on the raw frames as described in \citet{thm05}.  
208: The new images and the archival images were aligned separately since
209: they were at different orientations and had separate dither patterns.
210: The new images were aligned using offsets determined from the F190N image.  
211: The position of the point-like object
212: S1, described below, was determined by centroiding with the IDL procedure
213: IDP3 \citep{lyt99} for each image.  The same offsets were used for the 
214: other images in each orbit since HST was not moved during an orbit.  The
215: images in each filter were offset in IDP3 to the position of the images
216: in the first orbit and then median combined.  During processing visual 
217: inspection showed that two images, one line image and one continuum
218: image, were affected by a cosmic ray persistence
219: streak.  This occurs when there is a strong cosmic ray event before the 
220: integration begins and the image of the event persists in subsequent images.
221: The pixels involved in the event were identified and then not used in the
222: formation of the median image.  This means that the median is formed from
223: only 7 images rather than 8 for those pixels.
224: The archival images were aligned separately, using the image of S1 in each
225: filter as the fidelity of the pointing between images was not accurately 
226: known.
227: 
228: Subtraction of the associated narrow filter continuum image from the image in 
229: the emission line filter produces an emission line image.  The continuum images 
230: are scaled by the known responses of the line and continuum filters and detector 
231: QE and subtracted from the line image to produce a continuum free image. The 
232: spacing between the narrow line and continuum filters is only 0.03 microns
233: therefore the spectral slope of the SEDs, determined from the ground based 
234: spectra of \citet{van00}, has a negligible effect on the calculated line power. The 
235: Pa$\alpha$ image is the F190N image minus the scaled F187N image and the H$_2$ 
236: image is the F215N image minus the scaled F212N image.  The archival line emission 
237: images were reduced in the same manner and then rotated and 
238: shifted to match the new line images. A straight average was taken of the 
239: two images for the final line emission images.  The F187N image determines the
240: continuum level near the Pa$\alpha$ emission to measure the Pa$\alpha$ 
241: equivalent widths given in Table~\ref{tbl-ne2v}.
242: 
243: The archival wide filter continuum images were also rotated and shifted to match the
244: new images.  A median of all of the dithered F205W images was produced with
245: no shifts to produce a median sky image.  The median sky was then subtracted
246: from the F205W image to remove the thermal sky and instrument background. 
247: Contour plots of the ACS UV F140LP image from the data of \citet{kun04} and
248: contour plots of the new NICMOS continuum and line images are shown in 
249: Figure~\ref{fig-mcon}. Figure~\ref{fig-pa} shows the images of Pa$\alpha$, 
250: H$_2$, and the F160W continuum, along with ACS UV F140LP image. In 
251: these figures north is up in all cases.  Source S8 does not appear in
252: these figures due to the smaller size of the maps and images.
253: 
254: \section{Sizes of the point-like Sources, S1, S5 and S6} \label{s-size}
255: 
256: One aspect of SBS0335-052E, as pointed out earlier by \citet{har06},
257:  is that the majority of the star formation has 
258: occurred in very compact regions rather than being spread diffusely over 
259: the region as occurs in the Orion star forming complex in our own galaxy.  
260: Three of the sources S1, S5 and S6 display Airy rings indicative of a 
261: point-source.  We use NICMOS camera 2 observations of known single sources
262: observed to determine the NICMOS camera 2 PSF to get an estimate of the 
263: true size of these sources.  For the size of the Pa$\alpha$ emission we
264: use the PSF star GD-71 observed in the F190N filter and for the continuum we
265: use the PSF star HD 3670 observed in F160W filter. The PSF image of GD-71
266: is a median combination of 4 images at separate positions on the camera 2
267: detector array.  The images were shifted to a common position using IDP3. 
268: IDP3 uses  bi-cubic interpolation to resample and shift the images. The 
269: HD 3670 PSF 
270: image is the addition of two images again shifted and resampled in IDP3.
271: Each of these images were then resampled in an 8 times finer pixel scale,
272: again by bi-cubic interpolation in IDP3.  The images of S1, S5 and S6 were
273: then also resampled in the same way for comparison with the PSF star images.
274: 
275: The two resampled PSF star images were next convolved with 2D Gaussians with 
276: FWHM values of 3/8, 5/8, 3/4, and 15/16 of the \emph{original} camera 2 
277: 0.075\arcsec pixel size.  The
278: radial profiles of the original PSF, the convolved PSF and the object images
279: at the 8 times finer scale were calculated in IDP3 and then plotted
280: in Figure~\ref{fig-size}. The convolved radial profiles were visually compared 
281: with the source profiles to determine the source size.  The comparisons were
282: only made in the inner $0.15\arcsec$ of the profile to avoid significant 
283: contamination from the diffuse background.  We did not attempt to subtract
284: the background since it is non-uniform with significant structure, particularly
285: for the Pa$\alpha$ background. 
286: 
287: The source S1 is resolved in both the continuum and in Pa$\alpha$ 
288: and is consistent with the radial profile of a Gaussian source with
289: a FWHM of $0.12\arcsec$ to $0.14\arcsec$ convolved with the NICMOS 
290: PSF.  At the 54.3 Mpc distance of SBS0335-052 \citep{thu97} the radius of 
291: S1, taken as 1/2 of the FWHM of the Gaussian, is between 16 and 18 pc. This 
292: is 4 to 5 times 
293: larger than the radius of S1 estimated in TSKEV based on the ACS F140LP
294: image.  That estimate was simply done by eye without a similar 
295: analysis using convolved PSFs and was clearly an underestimate.  This 
296: does not, however, alter any of the conclusions in TSKEV,
297: other than the speculation on the formation of intermediate mass black 
298: holes.  The S1 continuum source may be slightly more compact than
299: the Paschen $\alpha$ but the difference is less than the error in comparing
300: the observed profile with the broadened PSF profiles.  The known ``breathing''
301: of the HST secondary mirror produces temporal changes of the PSF that are
302: approximately plus and minus half the spacing between the PSF profile and the
303: first broadened profile.  The comparison may also be influenced by the 
304: extensive diffuse emission in the area.  The faint source S6 is also resolved 
305: with a size similar to S1 determined in the same manner as for S1.
306: 
307: Source S5 is not resolved in this image.  It is consistent with a
308: point source with an upper limit on the radius of 10 pc. It
309: does not have any Paschen $\alpha$ emission and is therefore probably
310: not undergoing intense star formation.  All of these sources have
311: sizes consistent with current day globular clusters and are most
312: likely the young predecessors of globular clusters.  
313: 
314: \section{The Individual Sources in SBS0335-052E}
315: 
316: The sources S1-S7 lie in a roughly north-south alignment with the
317: northern most source S6 separated from the southern most source
318: S1 by approximately $1.9\arcsec$ which is 490 pc at 54.3 Mpc.  A supernova cavity
319: lies about $4\arcsec$ (~1kpc) north of source S1.  There appears to
320: be a progression of ages of the clusters along the north-south
321: axis (Thuan, Izotov \& Lipovetsky 1997, TSKEV) starting with the supernova
322: cavity in the north to the very young sources S1 and S2 which
323: has lead to the speculation that the SSCs are the result of triggered
324: star formation propagating from the north to the south (TSKEV).
325: In the following discussion we have arbitrarily divided the sources
326: into sources with Pa$\alpha$ emission and those without.  We could
327: have equally well divided the sources into compact (SSCs) and 
328: extended.  In fact the compact sources without Pa$\alpha$ emission
329: are most likely older versions of the compact sources with Pa$\alpha$
330: emission.
331: 
332: \subsection{Sources With Pa$\alpha$ Emission} \label{s-pa}
333: 
334: Five sources, S1, S2, S3-Pa$\alpha$, S7 and S8 along with a diffuse background
335: have detectable Pa$\alpha$ emission and hence measurable star formation
336: rates. Unlike in TSKEV the high resolution NICMOS camera 2 images accurately
337: determine the distribution of Pa$\alpha$ flux between S1 and S2.  In addition 
338: to the Pa$\alpha$ flux associated with the continuum emission of S1 and S2, 
339: the images also show diffuse background Pa$\alpha$ emission in the region of 
340: S1 and S2 which was not evident in the previous lower spatial resolution 
341: observations.  The total diffuse Pa$\alpha$ flux is comparable to the sum of 
342: the fluxes from the sources and is due to either in-situ ionizing stars or to 
343: ionizing radiation that has leaked from the sources S1 and S2. The 
344: former is more likely than the latter (See \S~\ref{ss-ds1ds2}). If the
345: emission is due to ionization by in-situ stars it is not clear whether it 
346: is due to a diffuse distribution of individual stars or a distribution of 
347: smaller clusters that is not resolved in the present observations. 
348: 
349: Two types of photometry were carried out on the sources.  The first is
350: aperture photometry with a 5 pixel diameter (0.378\arcsec) that encompasses
351: all of the flux inside the first Airy ring of a point source.  This is
352: $74\%$ of the total flux of a point source in the F160W filter and $69\%$
353: in the F190N filter.  Larger apertures begin to infringe on adjacent sources.
354: To first order adjacent sources add approximately the same flux to the aperture
355: as falls outside the aperture from the primary sources.  We therefore do 
356: not correct the aperture flux for the percentage of primary object flux in
357: the aperture.  The background was measured in annular rings 3 pixels thick 
358: around the sources.  The inner and outer diameters of the annular rings were 
359: adjusted to avoid flux from adjacent sources. The measured backgrounds, 
360: adjusted to the area of the photometric aperture were subtracted from the 
361: source flux. The photometric error is taken as $10\%$ of the total flux
362: for the bright compact sources and is dominated by the systematics of source
363: overlap and background subtraction. 
364: 
365: The equivalent widths were similarly measured in a 6 pixel (0.454\arcsec)
366: diameter aperture. The continuum measure for the equivalent widths is 
367: provided by the narrow band F187N filter which is adjacent to the F190N
368: band that measures the Pa$\alpha$ flux.  This filter accurately measures 
369: the continuum level at the location of the Pa$\alpha$ line and is free
370: of other contaminating emission lines. The larger aperture was used to 
371: increase the signal to noise of the measurements in the F187N filter which
372: has much lower flux than the Pa$\alpha$ filter or the broad band F160W
373: filter.  The smaller aperture photometry agreed
374: within their signal to noise ratios with the larger aperture measurements.
375: Table~\ref{tbl-flux} gives the line and continuum fluxes for all of the
376: sources along with the equivalent widths of the Pa$\alpha$ line.  It is
377: evident that the equivalent widths of the extended sources S7 and S8 are 
378: significantly higher than for the compact sources S1 and S2.  Again note that S7
379: is the source denoted by a question mark in TSKEV and S8 is the source
380: marked S7 in \citet{pap06}.  Unlike in TSKEV the new observations are deep enough
381: to detect the faint continuum associated with the extended sources and are
382: of high enough resolution to discern the offset between the continuum and
383: line emission in S3.  In the following we discuss the nature of the 
384: sources.
385: 
386: \subsubsection{Sources S1 and S2} \label{ss-s1s2}
387: 
388: The Pa$\alpha$ emission in S1 and S2 is coincident with the continuum
389: emission.  In the new images it is clear that the Pa$\alpha$ emission 
390: in the region of S1 and S2 has at least three components, S1, S2 and diffuse, 
391: therefore, the extinction of A$_V$ = 12.1 derived from the comparison 
392: of Pa$\alpha$ and Br$\alpha$ flux ratios in TSKEV is an average extinction. 
393: The extinction curves of \citet{rie85} were used to compute the extinction.
394: The extinction for any of 
395: the individual components may be different than the average.  If, however,
396: the extinction is primarily foreground then the extinction effects all
397: sources equally and the derivation is valid.  We note that the 1.6 \micron\ 
398: to 2.05 \micron\ color is similar for S1 and S2 indicating that they may have 
399: the same extinction.  The color of the diffuse continuum emission is blue 
400: rather than red, however, which supports the idea of less extinction in the 
401: regions of diffuse emission.
402: 
403: To get an estimate of the spatial distribution of the extinction we 
404: produced a H$\alpha$ image 
405: from the archive ACS HST images of SBS0335-052 from Proposal 10575 
406: (PI G. \"{O}stlin) by subtracting the F550M image from the FR675N ramp
407: filter image.  The F550M image was multiplied by a factor of 0.2 to 
408: account for the differences in throughput. The factor was determined
409: by the requirement that the sources with no Pa$\alpha$ emission, and
410: therefore presumed to have no H$\alpha$ emission, subtract to zero.
411: This does not guarantee an accurate subtraction in the sources with
412: hydrogen line emission since they are generally redder than the 
413: non-emission line sources. The H$\alpha$ image was then resampled to the
414: spatial resolution of the NICMOS Pa$\alpha$ image.  The ratio of the two 
415: images is an indicator of the relative extinction and is 
416: scaled to unity at the location of S1. The differences in wavelength
417: which changes the diameter of the Airy rings limits the accuracy of 
418: this procedure. The extinctions of S1 and S2 are equal within the
419: signal to noise of the image. The extinction of the extended sources 
420: S3-Pa$\alpha$, S7 and S8 are also equal to S1 and S2 within the rather 
421: low signal to noise.  The extinction in the diffuse region is 1/2 of 
422: that of S1 and S2 in terms of absolute flux. We therefore decrease the 
423: extinction in the diffuse flux by a factor of two in computing the properties 
424: of the diffuse emission.  The extinction map shows significant structure
425: due to the differences in spatial resolution and we estimate the 
426: accuracy of the relative extinctions derived from the map to be $20\%$. 
427: 
428: The volume emission measure (${N_e}^2V$) for S1 of $7.9 \times 10^{65}$
429: cm$^{-3}$ is almost exactly the same as the previous measurement in TSKEV of
430: $7.2 \times 10^{65}$, therefore the primary conclusions on the nature
431: of S1 remain valid.  Note that throughout the discussion of the Pa$\alpha$
432: sources the ionized gas temperature is assumed to be 20,000K as discussed
433: in TSKEV.  S1 has a stellar mass of $1.5 \times 10^7$ M$_{\sun}$, see 
434: \S~\ref{s-mass}, with stars of less than 1 M$_{\sun}$ probably not yet on the main
435: sequence.  There are 8650 stars earlier than O9 and the luminosity of
436: the stellar population is $3.4 \times 10^9$ L$_{\sun}$ based on the
437: output of Starburst99 \citep{lei99} if we only account for the Pa$\alpha$ 
438: flux in the point-like object. A further discussion of the Starburst99
439: calculation is given in \S~\ref{s-age}. The star formation rate
440: is 1.2 M$_{\sun}$ per year based on a recombination Case B recombination
441: rate for 20,000K \citep{ost89} and the SFR to Q(H$^0$) ratio given
442: by \citet{ken98}. The specific star formation rate, star formation rate
443: divided by the mass of the S1, of $1.4 \times 10^{-7}$ yr$^{-1}$  is 
444: relatively high. The specific star formation rate for the entire galaxy of 
445: $3.7 \times 10^{-9}$ yr$^{-1}$, however, is quite low.
446: 
447: We now resolve S2 from S1 and obtain a volume emission measure of 
448: $5.9 \times 10^{65}$ cm$^{-3}$ for S2, somewhat less than the arbitrary 
449: half of the total S1 plus S2 emission measure that was assigned to S2 
450: in TSKEV.  The parameters discussed above for S1 simply 
451: scale for S2 by the ratio of the volume emission measures.  The difference
452: between the values of the emission measure for S1 plus S2 in this work
453: versus TSKEV is due to the separation of the sources from the diffuse
454: background, some of which was included in the aperture measurement of
455: Pa$\alpha$ in the previous work. Unlike S1, S2 is clearly not a point 
456: like object. It is irregular in shape and extended in the East-West
457: direction.  Its average spatial extent is $0.05\arcsec$ larger than
458: S1 which puts its radius at 30 pc averaged over the East-West and 
459: North-South directions. 
460: 
461: \subsubsection{Diffuse Emission Around S1 and S2} \label{ss-ds1ds2}
462: 
463: The diffuse emission is the Pa$\alpha$ emission that remains after
464: all of the emission associated with the sources S1-S8 is subtracted.  The 
465: associated diffuse continuum emission is defined in the same way.  The 
466: majority of the diffuse emission, defined in this manner, is in the region 
467: around S1 and S2.  In the commonly used radio emission measure the volume 
468: emission measure of $5.5 \times 10^{65}$ cm$^{-3}$ is equivalent to that of 
469: $1.2 \times 10^4$ O7 stars. The volume emission measure of the diffuse 
470: emission is more than 1/3 of the emission measures of S1 and
471: S2.  This implies that the current star formation is roughly equally 
472: spread between diffuse and compact regions. 
473: We consider two possible interpretations of the diffuse Pa$\alpha$
474: emission.  The first is ionizing radiation from S1 and S2 that escapes the
475: clusters and ionizes the surrounding gas.  The second is an underlying
476: population of young stars.
477: 
478: The volume emission measure of the diffuse gas is $40\%$ of the volume emission
479: measure of S1 and S2. The diffuse emission therefore requires some source
480: of ionizing flux at the level of $40\%$ of that produced by S1 and S2.  This
481: assumes that all of the ionizing photons from the source of ionizing flux
482: for the diffuse emission are absorbed by the diffuse gas and none escape
483: from the region. If the source of ionizing photons is photons that escape
484: S1 and S2 but are trapped by the diffuse gas the mass and luminosity of the 
485: two sources must increase by a similar percentage.  This would place the mass 
486: of S1 and S2 at $2.1 \times 10^7$ M$_{\sun}$ and 
487: $1.6 \times 10^7$ M$_{\sun}$ respectively.  The luminosities for S1 and S2
488: become $4.8 \times 10^9$ L$_{\sun}$ and $3.5 \times 10^9$ L$_{\sun}$. These
489: are still reasonable parameters for the 2 SSCs but they lie at the high
490: side of known clusters, eg. \citep{zha99,dow08}. A difficulty may 
491: be that this makes the percentage of dust re-emitted luminosity of
492: $1.2 \times 10^8$ L$_{\sun}$ \citep{hou04,eng08}, only $20\%$ of the total 
493: luminosity of S1 and S2.  The discussion of the H$_2$ emission in
494: \S~\ref{s-h2}, however, shows that this is a reasonable number.  Another
495: difficulty could arise if the assumption that all of the ionizing flux
496: produces observable Pa$\alpha$ emission is false.  If a large fraction of
497: the ionizing flux escapes the region then the mass and luminosity requirement
498: can become unrealistic.  Also if the diffuse emission is simply ionized gas 
499: with no stellar continuum its Pa$\alpha$ equivalent width should be higher
500: than that of S1 and S2, instead it is lower. The equivalent width calculations
501: described in \S~\ref{s-age} in fact show that the diffuse emission has the
502: lowest equivalent width of any Pa$\alpha$ source. This is very strong evidence
503: against the concept that the diffuse Pa$\alpha$ is due to UV leakage from
504: the clusters.  The discussion in \S~\ref{s-age} also indicates that the 
505: diffuse emission stellar population is most likely older than the other
506: Pa$\alpha$ sources.
507: 
508: The possibility that the diffuse ionization region is produced by an in-situ
509: diffuse distribution of stars is a more plausible explanation.  Accounting for
510: the diffuse emission with in-situ stars requires a stellar mass of $40\%$ 
511: of the combined masses of S1 and S2 if the stars are at the same epoch
512: of evolution as the two SSCs.   The net star formation rate of the diffuse
513: component in this 
514: explanation is 0.85 M$_{\sun}$ yr$^{-1}$.  If this is the correct explanation
515: for the diffuse emission then star formation in SBS0335-052 is occurring
516: in a bi-modal fashion.  The majority of the stellar mass is in the 
517: form of compact objects, S1-S6, but the current star formation is roughly
518: divided 2 to 1 between the compact and diffuse regions whereas current star 
519: formation in the Milky Way appears to be predominantly in diffuse regions. 
520: 
521: \subsubsection{Source S3-Pa$\alpha$} \label{ss-s3pa}
522: 
523: S3-Pa$\alpha$ is a source with a Pa$\alpha$ equivalent width very close to
524: those of S1 and S2.  To delineate the Pa$\alpha$ emitting region from the 
525: continuum source S3 we have designated the emission region S3-Pa$\alpha$. 
526: The separation of S3-Pa$\alpha$ is the reason that \citet{pap06} found
527: an offset between the continuum and the line emission near S3. 
528: As with the diffuse emission the question is whether this is a separate star 
529: forming source or is it gas that is ionized by the double source S3 or 
530: possibly the more distant sources S1 and S2.  S3-Pa$\alpha$ has a volume 
531: emission measure of $2.8 \times 10^{64}$ cm$^{-3}$ which is equal to 
532: that of $2.9 \times 10^3$ O7 stars. If this volume emission measure is due to sources 
533: other than in-situ stars it would have to be due to intercepted ionizing 
534: flux from one of the other sources.
535: 
536: The lack of detectable Pa$\alpha$ flux at the location of S3 argues
537: against it as the source of the ionizing flux.  The only possibility
538: would be if the the region around and in S3 is largely free of gas.  
539: The observed Pa$\alpha$ region subtends 2.2 steradians as seen from
540: the location of S3 if its radial dimension is the same as its extent
541: on the sky and therefore receives approximately $17\%$ of the 
542: emitted flux from S3 if it is emitted isotropically.  Note that as in 
543: the discussion of the diffuse emission we assume that the source of
544: UV flux, the stars, emit radiation isotropically.  Sources can be
545: ``beamed'' by large scale distributions of dust but dust structures
546: simply block ionizing flux, they do not enhance the flux along any
547: particular line of sight.  The volume emission measure for S3 would 
548: then have to be $7.6 \times 10^{65}$ cm$^{-3}$, almost as high as S1.   
549: The requirements for ionization by S1 and S2 are even more formidable.  
550: S3-Pa$\alpha$ subtends $3.4 \times 10^{-2}$ steradians
551: as seen from S1.  The ionizing flux from S1 and S2 would then have to
552: be 370 times larger than listed in Table~\ref{tbl-ne2v}, increasing 
553: their stellar mass to almost $10^{10}$ M$_{\sun}$.  This is clearly
554: not possible. 
555: 
556: The most likely source of ionization of S3-Pa$\alpha$ is in-situ young
557: stars.  This is the explanation that requires the minimum amount of 
558: ionizing flux as is the case for the diffuse flux discussed earlier.
559: The star formation in this source is then in a more
560: extended region than in the SSCs, again showing the bi-modal nature of
561: high pressure, low metallicity star formation.  Another argument in favor of local
562: star formation in S3-Pa$\alpha$ is that the equivalent width of the
563: Pa$\alpha$ flux is almost exactly the same as in S1 and S2.  This 
564: argues that S3-Pa$\alpha$ is the location of early star formation
565: such as in S1 and S2 while the diffuse emission around S1 and S2, 
566: which has a lower equivalent width, is due to an older stage of
567: star formation as is further discussed in \S~\ref{s-age}.
568: 
569: \subsubsection{Sources S7 and S8} \label{ss-s7s8}
570: 
571: Two of the Pa$\alpha$ emitting sources are distinguished by significantly
572: larger equivalent widths than the rest of the sources.  They are
573: S7, the previously known source that was marked with a question mark in
574: TSKEV and S8, $2.59\arcsec$ (664 pc) from S1 as shown in 
575: Figure~\ref{fig-map}.  Note that both S7 and S8 are clearly
576: visible in the images presented by \citet{ost08}, eg. their Figure 3,
577: but were not commented on in the paper.  Both of these sources are extended
578: and far away from any other known ionizing sources, particularly for S8.
579: It appears clear that these must be sites of new star formation and the 
580: ionization is due to in-situ stars.  Their volume emission measures, as
581: shown in Table~\ref{tbl-ne2v}, are about a factor of 10 less than S1 and
582: S2 and the total mass of stars is less.  The star formation rates
583: are 0.14 and 0.043 M$_{\sun}$ yr$^{-1}$ respectively.  The extinction
584: map discussed in \S~\ref{ss-s1s2} gives the same average extinction in
585: S7 and S8 as for S1 and S2, but, the low signal to noise of the continuum
586: make variations by a factor of 2 quite possible.
587: 
588: \subsection{Sources S3, S4, S5 and S6} \label{s-s3456}
589: 
590: The sources S3, S4, S5 and S6 are distinguished by the absence of any associated
591: Pa$\alpha$ emission indicating a lack of ionized gas and current star 
592: formation in these sources.  Their continuum 1.6 to 2.05 \micron\ colors are bluer than
593: S1 and S2 which is most likely a function of the foreground extinction
594: rather than their intrinsic color.  Source S3 deserves special mention 
595: as it may be the most enigmatic of all of the non-Pa$\alpha$ sources. 
596: In the ACS UV F140LP image in Figure~\ref{fig-pa} S3 
597: is clearly double, also seen in Figure 1 right of TSKEV, with a spacing 
598: of $0.08\arcsec$ (23 pc) in the northeast-southwest direction on the sky.  
599: The infrared continuum is centered on the northeast component with no 
600: indication of a southwest source or extension. The continuum fluxes for 
601: S3 are also hard to explain.  At UV wavelengths the two components of S3 
602: are roughly equal in brightness (Figure~\ref{fig-pa} ACS UV F140LP) while 
603: at infrared wavelengths (Figure~\ref{fig-pa} 1.6\micron\ continuum) only 
604: the northeast component is seen but 
605: it is quite bright. A possibility is that the northeast component has an
606: intrinsic brightness much brighter than the southwest but is much more heavily
607: obscured so that at UV wavelengths they appear to be a roughly the
608: same brightness.  This explanation is possible but appears to be a
609: bit contrived. The obscuring dust would have to be a significant distance 
610: in front of the northeast component of S3 otherwise we should see mid and 
611: far infrared emission from that location. On the other hand, it must be
612: small enough to only obscure the northeast source.
613: 
614: The discussion in \S~\ref{s-age} indicates that the non-Pa$\alpha$ sources
615: are older than the other sources which is consistent with triggered star 
616: formation starting in the north and propagating southward.  Only S3 is 
617: inconsistent with this progression since it is very near the young star 
618: forming region S3-Pa$\alpha$.  S5 is somewhat unique in that it is extremely 
619: compact as discussed in \S~\ref{s-size}. S5 could possibly be a young 
620: globular cluster that has undergone core collapse. All of the non-Pa$\alpha$
621: sources have probably had most of the residual gas swept out of them and are 
622: bound young globular clusters.
623: 
624: \section{The ages of the sources} \label{s-age}
625: 
626: Table~\ref{tbl-flux} gives equivalent widths for the Pa$\alpha$ emission
627: sources.  The equivalent width is relative to the continuum measured
628: in the F187N narrow line filter which is adjacent in wavelength to the
629: F190N filter that measures the Pa$\alpha$ flux.  Use of the F160W or 
630: F205W filter images would give higher signal to noise but the uncertainty
631: in the detailed spectrum and the inclusion of other emission lines in the
632: broad filter compromises the accuracy of the equivalent width.  In an
633: attempt to determine the ages of the emission line sources we ran a
634: Starburst99 \citep{lei99} simulation with a stellar mass range of 
635: 0.1-120 M$_{\sun}$, a standard Kroupa IMF, a metallicity of 0.001,
636: time steps of $10^5$ years and an instantaneous burst of star formation.  
637: The luminosity of the Pa$\alpha$ line was
638: calculated from the output Pa$\beta$ luminosity using Case B recombination
639: and a temperature of 20,000K. From this line luminosity and the computed continuum
640: flux at the wavelength of Pa$\alpha$ the equivalent width of Pa$\alpha$ was
641: calculated and plotted versus age in Figure~\ref{fig-eqw}. Note that the 
642: continuum includes both stellar and nebular emission.  The plot is
643: marked with asterisks at the equivalent width appropriate for each source.
644: The position marked for S8 is at the oldest consistent time.  This equivalent width
645: is achieved at two earlier times but the time spent at those values is
646: very brief, $10^5$ and $1.5 \times 10^5$ years, as opposed to $1.3 \times 10^6$
647: years at the marked position.  The dip in equivalent width around 
648: $4 \times 10^6$ years is due to a rise in the infrared continuum rather 
649: than a dip in the Pa$\alpha$ luminosity.  
650: 
651: The parameters of the Starburst99 simulation have little effect
652: on the age sequence and ages of the sources for an instantaneous starburst.
653: It is only the lifetimes of the high mass stars that affect the evolution
654: of the ionizing flux.  Higher metallicities would decrease the ionizing
655: flux per unit mass and different IMFs would change the total mass needed
656: to achieve the observed ionizing fluxes but not the time evolution.  To 
657: check this assumption we produced another Starburst99 simulation with a
658: Salpeter rather than Kroupa IMF.  The resultant equivalent to Figure~\ref{fig-eqw}
659: was indistinguishable from Figure~\ref{fig-eqw}.  These changes only
660:  affect the mass estimate discussed in \S~\ref{s-mass}.  If, however, the 
661: star formation in each of the clusters is not short relative to the present
662: age of the cluster the ages could be quite different from those derived
663: here.  The ages from Figure~\ref{fig-eqw} are only valid if the star formation
664: epoch is short relative to the age of the clusters.
665: 
666: Figure~\ref{fig-eqw} indicates that S7 and S8 are young star forming regions
667: with ages less than 10 million years with S7, the youngest, at about 4 million
668: years.  S1, S2 and S3-Pa$\alpha$ all formed at roughly the same epoch and are now about
669: 10 million years old.  The diffuse emission is from an older group of stars
670: formed a million years before the S1, S2, S3-Pa$\alpha$ group.  The sources
671: without detectable Pa$\alpha$ emission are at least 15 million years old
672: or older.  This shows a general progression in age of the SSCs with oldest
673: sources in the north and the youngest in the south, evidence for triggered
674: star formation running from north to south.  New star formation is now
675: occurring in S7 and the more distant S8.  A 1 km/sec wind would propagate 10
676: pc in 10 million years which makes it doubtful that S8, which is 340 pc from
677: S1 and S2, is causally affected by the SSCs.
678: 
679: \section{Stellar Masses} \label{s-mass}
680: 
681: The Starburst99 output provides an estimate of the stellar mass for those objects
682: with Pa$\alpha$ emission.  The Starburst99 calculation used a stellar mass of
683: $10^8 M_{\sun}$, therefore, the ratio of the observed Pa$\alpha$ power versus
684: the calculated Pa$\alpha$ power times $10^8 M_{\sun}$ at the age given in
685: Figure~\ref{fig-eqw} gives the mass.  The masses are given in the last column of 
686: Table~\ref{tbl-ne2v}.  As discussed in \S~\ref{s-age} the masses are sensitive
687: to the parameters used in the Starburst99 model, particularly on the IMF 
688: and metallicity
689: used.  The metallicity is appropriate for that measured in the clusters but we
690: do not have sufficient information to accurately determine the actual IMF and
691: many of the low mass stars are still probably in their pre-main sequence evolutionary
692: stages. The largest mass components are the two compact sources S1
693: and S2 plus the diffuse component which total to a mass of $4.65 \times 10^7 M_{\sun}$. 
694: If the SSCs with no Pa$\alpha$ are included with an average mass of $10^7 M_{\sun}$ 
695: per SSC then the total mass is on the order of $8 \times 10^7 M_{\sun}$.  Taking the 
696: total gas mass as $2.1 \times 10^9 M_{\sun}$ \citep{pus01} then the ratio of 
697: total stellar mass in the nucleus of SBS0335-052E to the gas mass of 1 to 26 
698: is significantly higher than the average nuclear cluster mass to gas ratio 
699: found by \citet{set08}. The average nuclear cluster mass to gas ratio in 
700: \citet{set08}, however, is dominated by nuclear clusters in large spiral 
701: galaxies rather than in BCD low metallicity galaxies.
702: The objects S3-Pa$\alpha$, S7 and S8, which have less concentrated star formation
703: have significantly less mass, with S7 having only $9.4 \times 10^4 M_{\sun}$.  This
704: may represent star formation in what we would term open clusters and they may not
705: remain bound for very long.
706: 
707: \section{H$_2$ Emission in S1 and S2} \label{s-h2}
708: 
709: A new finding of this investigation is the precise location of the
710: H$_2$ emission in S1 and S2.  Previous ground based observations by
711: \citet{van00} did not have the spatial resolution to resolve the 
712: sources.  The contour map in Figure~\ref{fig-h2} shows separate emission 
713: regions in both S1 and S2.  The emission region in S1 is consistent
714: with being point-like while the emission in S2 is extended as is the 
715: continuum emission.  Both of the regions are offset from the center
716: of the continuum emission.  The H$_2$ emission region in S1 is offset
717: by $0.04\arcsec$ (11.6 pc) to the east from the continuum center and 
718: the H$_2$ region in S2 is offset by the same amount to the south.  
719: These offsets are less than the radii of the clusters so that the 
720: molecular emission regions exist inside the region of Pa$\alpha$
721: emission. There is no chance that the offsets are due to misalignment
722: of the images as the new H$_2$ and Pa$\alpha$ images were taken in
723: the same orbit for each dither position with no change is HST attitude.
724: Taken alone these images clearly show the offset.  The additional 
725: archival Pa$\alpha$ images were then registered to the new Pa$\alpha$
726: images and the same offset was present.  It is remotely possible that
727: the two H$_2$ regions are located along the line of sight to S1 and S2
728: but outside of the clusters.  This might be considered a possibility
729: if only one SSC coincided with a H$_2$ source but it is highly unlikely
730: that two SSCs lie along the line of sight to the only two H$_2$
731: sources. 
732: 
733: The survival of molecular gas inside the sources is another
734: line of evidence that the photo-ionizing flux does not reach throughout
735: the clusters.  The molecular regions are also the most likely sources
736: of the observed mid-infrared emission found by \citet{hou04} and
737: further discussed by \citet{eng08}.  If these are the emission regions
738: and they are small dust regions embedded in the cluster then they are
739: not capable of absorbing all of the luminosity of the cluster and 
740: re-emitting it in the infrared.  This accounts for the low IR luminosity
741: to total luminosity ratio discussed in \S~\ref{ss-ds1ds2}.  Their
742: location inside the cluster also probably accounts for the high
743: temperature of the observed dust emission.
744: 
745: The calculation of the mass of H$_2$ needed to produce the observed
746: emission depends on the temperature of the H$_2$ gas and the ratio
747: of thermal to fluorescent emission, neither of which is accurately
748: available.  \citet{van00} find that the emission is a mixture of
749: fluorescent and thermal emission which makes the line ratios an
750: unreliable monitor of the thermal temperature.  To get a rough 
751: estimate of the amount of H$_2$ needed to produce the observed
752: (1-0) S(1) emission we 
753: calculate the H$_2$ abundance needed for a 2500 K thermal gas
754: using an Einstein A value of $3.471 \times 10^{-7}$ s$^{-1}$
755: \citep{wol98}. This estimate should be good to about a factor
756: of two to three,  modulo the amount of extinction, if the 
757: contributions from thermal and fluorescent emission are about equal. 
758: Table~\ref{tbl-ne2v} gives the mass of H$_2$ needed to produce the
759: observed line strength under the assumption of the same extinction
760: as used in the Pa$\alpha$ calculation.  This is a lower limit on the
761: H$_2$ since the extinction will be significantly higher in the dust
762: protected molecular region.  \citet{hou04} calculate a 9.7 \micron\ 
763: extinction of 0.49 magnitudes which predicts about 2.6 magnitudes 
764: of extinction at 2.12 \micron\ as opposed to the 1.4 magnitudes 
765: derived from the atomic hydrogen lines.  \citet{van00} spectroscopically 
766: observed a line strength that is about 2.2 times larger than that quoted 
767: in the table.  We may have missed some of the extended flux but the value 
768: here is within 1.9$\sigma$ of the spectroscopic value.   
769: 
770: \section{Evolution and Feedback in SBS0335-052E} \label{s-ef}
771: 
772: The primary conclusions on the nature of the SSCs S1 and S2 in TSKEV
773: remain intact.  Although the possibility exists that the diffuse emission
774: around S1 and S2 is due to ionizing radiation from them, the more likely
775: explanation is that the ionization is done by in-situ stars.  The lower
776: equivalent width of the diffuse radiation is strong evidence for this view.
777: If the diffuse emission was due to pure ionized gas with no stars its 
778: equivalent width should be higher rather than lower than the equivalent
779: width of the star rich clusters S1 and S2.  This leaves the conclusion that S1 and 
780: S2 are currently not producing any negative feedback from photo-ionizing
781: flux on star formation in SBS0335-052E intact.  The delay in photo-ionizing
782: feedback may also have implications for planet formation.  The ambient
783: UV radiation field may be significantly less than the fields calculated
784: without considering absorption by stellar winds, eg. \citet{fat08}. This
785: would increase the probability of planet formation even in the environment
786: of a SSC which would normally be considered not conducive to planet formation.
787: It also has implications on whether globular clusters are good sites for
788: future searches for planets.
789: 
790: The deeper and higher spatial resolution images have revealed that there are
791: star formation regions in SBS0335-052E that are not compact SSCs.  Two of 
792: these regions, S7 and S8, appear to be younger than S1 and S2 while the
793: other extended region, S3-Pa$\alpha$ appears to be roughly coeval with 
794: S1 and S2.  The appearance of these extended regions of star formation may 
795: mark the end of the formation of compact stellar clusters and perhaps
796: the beginning of negative photo-ionizing feedback.  Because the volume emission 
797: measures of the extended sources are significantly lower than that of S1 and
798: S2, modulo extinction, they do not contribute much to the net radio emission.
799: If the extended regions have extended HII regions then they should be optically
800: thin.  Future observations with the EVLA should be capable of taking spatially
801: resolved radio images of the different sources to determine whether they
802: are optically thick or thin at radio frequencies.
803: 
804: \citet{elm97} discuss the effect of pressure on the ratio of compact to diffuse
805: star formation.  They claim that high pressure favors the formation of compact
806: clusters as opposed to diffuse star formation. \cite{bek08} has simulated the
807: merger of two dwarf galaxies with stellar masses $4 \times 10^8 M_{\sun}$ and
808: gas masses of $8 \times 10^8 M_{\sun}$. $22\%$ of the new stars in this
809: simulation appear in compact clusters, far more than in simulations at lower
810: pressures.  SBS0335-052 has a gas mass of $2.1 \times 
811: 10^9 M_{\sun}$ and a gas to stellar mass ratio of 25-26 (Pustilnik et al. 2001; this
812: work) therefore the central gas pressure should be significantly higher than 
813: the model considered by \citet{bek08} leading to an increased fraction of star 
814: formation in clusters.
815: 
816: \section{Comparison with Other Recent Work}
817: 
818: One day after the initial submission of this manuscript, new work on
819: SBS0335-052E became publicly available \citep{rei08}, hereinafter RJH, 
820: which reaches some different conclusions than this work.  RJH derive 
821: significantly lower masses and ionizing fluxes than this work, particularly 
822: for the youngest sources, S1 and S2, where the difference is a factor of 15
823: and 10 respectively.  The main diagnostic tool of RJH
824: is SED fitting to the optically observed HST ACS images of \citet{ost08}.
825: This fitting, confined to the first 4 ACS bands leads to significantly
826: lower dust extinction values than are derived in this work and TSKEV.
827: Our A$_V$ value is 12.1 as opposed to 0.5 in RJH.  This is the primary
828: reason for the different conclusions drawn between the two works. Source
829: fluxes redder than the F550M band deviate strongly from the RJH derived
830: SED and must be explained by a combination of hot dust and Extended Red 
831: Emission \citep{wit04}. The extinction model needed to account for the 
832: observed hydrogen emission line ratios is gray from H$\beta$ to Br$\gamma$
833: with 0.9 magnitudes of extinction for each line.  Our model with an A$_V$
834: value of 12.1 and an extinction law given by \citet{rie85} implies significantly
835: more line flux which requires a larger ${N_e}^2V$ and more mass in
836: stars to provide the ionizing flux.  This model does not require the
837: extra emission components to match the observed fluxes.
838: 
839: \section{Summary} \label{s-con}
840: 
841: New high spatial resolution NICMOS camera 2 images of the region of SBS0335-052E 
842: that contains a compact group of SSCs have improved our detailed knowledge of the
843: geometry and physics of the system.  There are 6 point-like objects, 7 if the binary
844: nature of S3 is taken into account, that are either currently SSCs (S1 and S2) or
845: were most likely previously SSCs (S3-S6).  We now know that the continuum source
846: S3 is not at the same position as the Pa$\alpha$ source S3-Pa$\alpha$.  In addition
847: to the compact sources there are 4 diffuse or extended regions of Pa$\alpha$ emission,
848: indicating current star formation in non compact objects.  It should be noted, however,
849: that a diffuse distribution of many compact sources with stellar masses on the order
850: of $10^4$ M$_{\sun}$ would be difficult to differentiate from a diffuse distribution
851: of individual stars.  Two of these regions, S7 and S8 appear to be significantly 
852: younger than the two star forming SSCs.  One of the star forming regions, S8, is 
853: well separated from the compact group of previously known sources.  These regions
854: are the first indication of significant star forming activity in SBS0335-052E in 
855: anything other than very compact clusters. They are significantly less massive than the
856: very compact clusters, modulo extinction, and may be open clusters that will not 
857: remain bound after the gas is expelled. The new images still support the general 
858: picture of delayed photo-ionizing negative feedback from the sources S1 and S2 
859: along with the general picture of propagating star formation from the north to 
860: the south. This delay in feedback may also affect the number of planetary systems
861: that can form in what would normally be considered a very harsh environment.  
862: We have also
863: confirmed that the H$_2$ emission is inside both S1 and S2 but is not at the 
864: exact center of the two objects and probably marks the location of the regions
865: producing the observed mid-infrared emission.
866: 
867: One result that differed from TSKEV is the size of S1 where the possibility
868: was raised that SSCs may be the origin of intermediate mass black holes through
869: the merging of stellar mass black holes created by supernova events in the 
870: cluster.  The new radius of 15 pc is approximately 3 times larger than the 
871: previous  radius and would affect the possibility of forming intermediate 
872: mass black holes. Equation 1 of \citet{zwa06} indicates that the time scale 
873: for forming merged intermediate mass black holes is dependent on radius to 
874: the 3/2 power.  The new radius of S1 would therefore increase the time scale
875: by about a factor of 5.  Even at the new radius,however, SSCs seem to be a 
876: leading candidate for the formation site of greater than stellar mass black 
877: holes. See also the very recent discussion of massive black hole formation
878: in early low metallicity star clusters by \citet{dev08}.
879: 
880: \acknowledgments
881: 
882: We wish to acknowledge very helpful discussions with Kelsey Johnson and to 
883: thank an anonymous referee for comments that have improved this
884: manuscript. This work contains data from the NASA/ESA Hubble Space Telescope 
885: which is
886: operated by the Association of Universities for Research in Astronomy
887: (AURA) Inc. under NASA contract NAS5-26555. 
888: 
889: 
890: \begin{thebibliography}{}
891: 
892: \bibitem[Bekki(2008)]{bek08} Bekki, K. 2008, arXiv:0805.0662v1, MNRAS Letters in press
893: 
894: \bibitem[Devecchi \& Volonteri(2008)]{dev08} Devecchi, B. \& Volonteri, M. 2008,
895: 	arXiv:0810.1057v1
896: 
897: \bibitem[Dowell et al.(2008)]{dow08} Dowell, J. D., Buckalew, B. A. \& Tan, J. C.  2008, 
898: 	\aj, 135, 823
899: 
900: \bibitem[Elmegreen and Efremov(1997)]{elm97} Elmegreen, B. G. \& Efremov, Y. N. 1997,
901: 	\apj, 480, 235
902: 
903: \bibitem[Engelbracht et al.(2008)]{eng08} Engelbracht, C. W., Rieke, G. H.,
904: 	Gordon, K. D., Smith, J.-D. T., Werner, M. W., Moustakas, J., Wilmer, C. N. A.
905: 	\& Vanzi, L. 2008, \apj, 678, 804
906: 
907: \bibitem[Fatuzzo and Adams(2008)]{fat08} Fatuzzo, M. \& Adams, F. 2008, \apj, 
908: 	675, 1361
909: 
910: \bibitem[Harishita \& Hunt(2006)]{har06} Harishita, H. \& Hunt, L. K. 2006, \aap,
911: 	460, 67
912: 
913: \bibitem[Houck et al.(2004)]{hou04} Houck, J. R. et al. 2004, \apjs,
914: 	154, 211
915: 
916: \bibitem[Izotov et al.(1997)]{izo97} Izotov, Y. I., Lipovetsky, V. A.,
917: 	Chaffee, F. H., Foltz, C. B., Guseva, N. G. \& Kniazev, A. Y. 1997,
918: 	\apj, 476, 698 
919: 
920: \bibitem[Izotov \& Thuan(2007)]{izo07} Izotov, Y. I. and Thuan, T. X. 2007
921: 	\apj, 665, 1115 
922: 
923: \bibitem[Kennicutt(1998)]{ken98} Kennicutt, R. C., Jr. 1998, \araa, 36, 189
924: 
925: \bibitem[Kennicutt et al.(2003)]{ken03} Kennicutt, R. C., Jr. et al. 2003,
926: 	\pasp, 115, 928
927: 
928: \bibitem[Kunth et al.(2003)]{kun03} Kunth, D., Leitherer, C., Mas-Hesse, J. M.,
929: 	\"{O}stlin, G. \& Petrosian, A. 2003, \apj, 597, 263
930: 
931: \bibitem[Kunth et al.(2004)]{kun04} Kunth, D., Hayes, M., \"{O}stlin, G.,
932: 	Mas-Hesse, J. M., Leitherer, C. \& Petrosian, A. 2004,
933: 	astro-ph/0407584 v1
934: 
935: \bibitem[Leitherer et al.(1999)]{lei99} Leitherer et al. 1999, \apjs, 123, 3
936: 
937: \bibitem[Lytle et al.(1999)]{lyt99} Lytle, D., Stobie, E., Ferro, A.,
938: 	\& Barg, I. 1999, in ASP Conf. Ser. 172, Astronomical Data
939: 	Analysis Software and Systems VIII, ed. D. Mheringer, R. 
940: 	Plante, \& D. Roberts (San Francisco; ASP), 445
941: 
942: \bibitem[Osterbrock(1989)]{ost89} Osterbrock, D. E. 1989, \emph{Astrophysics
943: 	of Gaseous Nebulae and Active Galactic Nuclei}, (Mill Valley, CA: 
944: 	University Science Books)
945: 
946: \bibitem[\"{O}stlin et al.(2008)]{ost08} \"{O}stlin, G., Hayes, M., Kunth, D.,
947: 	Mas-Hesse, J. M., Leitherer, C., Petrosian, A. \& Atek, H. 2008, 
948: 	astro-ph0803.1174v3
949: 
950: \bibitem[Papaderos et al.(2006)]{pap06} Papaderos, P., Izotov, Y. I., 
951: 	Guseva, N. G., Thuan, T. X. \& Fricke, K. J. 2006, \aap, 454, 119.
952: 
953: \bibitem[Pustilnik et al.(2001)]{pus01} Pustilnik, S. A., Brinks E., Thuan, T. X., 	Lipovetsky, V. A., and Izotov, Y. I. 2001, \aj, 121, 1413
954: 
955: \bibitem[Reines, Johnson \& Hunt(2008)]{rei08} Reines, A. E., Johnson, K E., 
956: 	\& Hunt, L. K. 2008, \aj, in press
957: 
958: \bibitem[Rieke \& Lebofsky(1985)]{rie85} Rieke, G. H., Lebofsky, M. J. 1985, \apj,
959: 	288, 618
960: 
961: \bibitem[Seth et al.(2008)]{set08} Seth, A., Ag\"{u}eros, M., Lee, D. \&
962: 	Basu-Zych 2008, \apj, 678, 116.
963: 
964: \bibitem[Thompson et al.(2005)]{thm05} Thompson, R. I. et al. 2005,
965: 	\aj, 130, 1
966: 
967: \bibitem[Thompson et al.(2006)]{thm06} Thompson, R. I., Sauvage, M., 
968: 	Kennicutt, R. C., Engelbracht, C. W. 2006, \apj, 638, 176 (TSKEV)
969: 
970: \bibitem[Thuan et al.(1996)]{thu96} Thuan, T. X., Izotov, Y. I., \& Lipovetsky, V. A.
971: 	1996, \apj, 463, 120
972: 
973: \bibitem[Thuan, Izotov \& Lipovetsky(1997)]{thu97} Thuan, T. X., Izotov,
974: 	Y. I., \& Lipovetsky, V. A. 1997, \apj, 477, 661
975: 
976: 
977: \bibitem[Vanzi et al.(2000)]{van00} Vanzi, L., Hunt, L. K., Thuan, T. X.
978: 	\& Izotov, Y. I. 2000, \aap, 363, 493
979: 
980: \bibitem[Witt \& Vijh(2004)]{wit04} Witt, A. N., \& Vijh, U. P. 2004,
981: 	in ASP Conf. Ser. 390, Astrophysics of Dust, ed. A. N. Witt, G. C. 
982: 	Clayton,B.T. Draine, (San Francisco: ASP), 115
983: 
984: \bibitem[Wolniewicz et al.(1998)]{wol98} Wolniewicz, L., Simbotin, I., \&
985: 	Dalgarno, A. 1998, \apjs, 115, 293
986: 
987: \bibitem[Zhang \& Fall(1999)]{zha99} Zhang, Q. \& Fall, S.M 1999 \apjl, 527, L81
988: 
989: \bibitem[Zwart et al.(2006)]{zwa06} Zwart, S.F.P., Baumgardt, H., McMillan, S.L.W.,
990: 	Makino, J., Hut, P., \& Ebisuzaki, T. 2006, \apj, 641, 319
991: 
992: \end{thebibliography}
993: 
994: \clearpage
995: 
996: \begin{deluxetable}{ccccccccc}
997: \tabletypesize{\scriptsize}
998: \tablecaption{HST Image Observational Parameters \label{tbl-obs}}
999: \tablewidth{0pt}
1000: \tablehead{
1001: \colhead{Inst.} & \colhead{Prop.} & \colhead{Mode} & \colhead{Filter} &
1002: \colhead{Samp.\tablenotemark{a}} & \colhead{NSamp.\tablenotemark{a}} &
1003: \colhead{Int. time} & \colhead{Ints.} & \colhead{Total} \\
1004: \colhead{} & \colhead{} & \colhead{} & \colhead{} & \colhead{} & 
1005: \colhead{} & \colhead{sec.} & \colhead{\#} & \colhead{sec.}
1006: }
1007: \startdata
1008: NICMOS & 10856 & Cam. 2 & F187N & STEP64 & 20 & 768 & 8 & 6144 \\
1009: NICMOS & 10856 & Cam. 2 & F190N & STEP32 & 17 & 320 & 8 & 2560 \\
1010: NICMOS & 10856 & Cam. 2 & F212N & STEP64 & 20 & 768 & 8 & 6144 \\
1011: NICMOS & 10856 & Cam. 2 & F215N & STEP64 & 20 & 768 & 8 & 6144 \\
1012: NICMOS & 10894 & Cam. 2 & F187N & STEP16 & 26 & 304 & 16 & 4864 \\
1013: NICMOS & 10894 & Cam. 2 & F190N & STEP16 & 25 & 288 & 16 & 4600 \\
1014: NICMOS & 10894 & Cam. 2 & F160W & STEP16 & 26 & 304 & 8 & 2432 \\
1015: NICMOS & 10894 & Cam. 2 & F205W & STEP64 & 24 & 272 & 8 & 2176 \\
1016: ACS & 10575 & WFC1 & F550M & - & - & 430 & 1 & 430 \\
1017: ACS & 10575 & WFC1 & FR656 & - & - & 680 & 1 & 680 \\
1018: ACS & 9470 & SBC & F140LP & - & - & 2700 & 1 & 2700 \\
1019: \enddata
1020: 
1021: \tablenotetext{a}{NICMOS only}
1022: \end{deluxetable}
1023: 
1024: \clearpage
1025: \begin{deluxetable}{ccccccc}
1026: \tabletypesize{\scriptsize}
1027: \tablecaption{Observed Line and Continuum Measurements\tablenotemark{a} \label{tbl-flux}}
1028: \tablewidth{0pt}
1029: \tablehead{
1030: \colhead{Source} & \colhead{Pa$\alpha$} & \colhead{H$_2$} & \colhead{0.14 $\mu$m} & 
1031: \colhead{1.6 $\mu$m} & \colhead{2.05 $\mu$m} & \colhead{Pa$\alpha$ Equivalent Width} \\
1032: \colhead{ } & \colhead{erg s$^{-1}$ cm$^{-2}$} &  \colhead{erg s$^{-1}$ cm$^{-2}$} &
1033:  \colhead{Jy} & \colhead{Jy} & \colhead{Jy} & \colhead{\AA}
1034: }
1035: \startdata
1036: S1 & $9.7 \times 10^{-15}$ & $8.8 \times 10^{-17}$ & $1.7 \times 10^{-5}$ &
1037: $2.6 \times 10^{-5}$ & $ 4.5 \times 10^{-5}$ & 1281 \\
1038: S2 & $7.2 \times 10^{-15}$ & $3.5 \times 10^{-17}$ & $8.6 \times 10^{-6}$ &
1039: $2.3 \times 10^{-5}$ & $ 3.5 \times 10^{-5}$ & 1182 \\
1040: S3-Pa$\alpha$ & $1.5 \times 10^{-15}$ & \nodata & $1.2 \times 10^{-5}$ & 
1041: $1.9 \times 10^{-6}$ & $ 1.1 \times 10^{-5}$ & 1103 \\
1042: S3 & \nodata & \nodata & $1.2 \times 10^{-5}$ & 
1043: $1.1 \times 10^{-5}$ & $ 5.9 \times 10^{-6}$ & \nodata \\
1044: S4 & \nodata & \nodata & $ 4.1 \times 10^{-5}$ & $1.2 \times 10^{-5}$ &
1045: $ 9.2 \times 10^{-6}$ & \nodata \\
1046: S5 & \nodata & \nodata & $ 4.9 \times 10^{-5}$ & $2.4 \times 10^{-5}$ &
1047: $ 1.9 \times 10^{-5}$ & \nodata \\
1048: S6 & \nodata & \nodata & $ 4.1 \times 10^{-6}$ & $4.1 \times 10^{-6}$ &
1049: $ 3.4 \times 10^{-6}$ & \nodata \\
1050: S7  & $1.1 \times 10^{-15}$ & \nodata & $6.0 \times 10^{-7}$ & $3.7 \times 10^{-6}$ &
1051: $ 4.3 \times 10^{-6}$ & 3478\\
1052: S8 & $3.4 \times 10^{-16}$ & \nodata & $ 1.2 \times 10^{-6}$ & $1.2 \times 10^{-6}$ &
1053: $ 1.6 \times 10^{-6}$ & 2079 \\
1054: diffuse & $1.37 \times 10^{-14}$ & \nodata & $ 5.7 \times 10^{-4}$ & $2.6 \times 10^{-5}$ &
1055: $ 2.1 \times 10^{-5}$ & 672 \\
1056: \enddata
1057: 
1058: \tablenotetext{a}{All fluxes are for an aperture diameter of 5 pixels ($0.378\arcsec$) 
1059: and the equivalent widths are for an aperture diameter of 6 pixels ($0.454\arcsec$).
1060: Flux and equivalent width errors are estimated as $10\%$ and are dominated by the 
1061: amount of extended flux.}
1062: \end{deluxetable}
1063: 
1064: \clearpage
1065: 
1066: \begin{deluxetable}{cccccccc}
1067: \tabletypesize{\scriptsize}
1068: \tablecaption{Line Power, ${N_e}^2V$, and Star Formation Rates \label{tbl-ne2v}}
1069: \tablewidth{0pt}
1070: \tablehead{
1071: \colhead{Source} & \colhead{Pa$\alpha$ Line Power\tablenotemark{a}} & 
1072: \colhead{${N_e}^2V$\tablenotemark{b}} & 
1073: \colhead{Ext. Cor. ${N_e}^2V$\tablenotemark{c}} & \colhead{SFR}
1074: & \colhead{Ext. Cor. H$_2$ Pow.} & \colhead{Ext. Cor. H$_2$ Mass} & 
1075: \colhead{Stellar Mass\tablenotemark{d}}\\
1076: \colhead{ } & \colhead{ergs/sec} & \colhead{cm$^{-3}$} & \colhead{cm$^{-3}$} &
1077: \colhead{M$_{\sun}$ yr$^{-1}$} & \colhead{ergs/sec} & \colhead{M$_{\sun}$} & 
1078: \colhead{M$_{\sun}$}
1079: }
1080: \startdata
1081: S1 & $3.4 \times 10^{39}$ & $1.8 \times 10^{65}$ & $7.9 \times 10^{65}$ & 1.2
1082: & $1.1 \times 10^{38}$ & 21. & $1.5 \times 10^7$\\
1083: S2 & $2.5 \times 10^{39}$ & $1.3 \times 10^{65}$ & $5.9 \times 10^{65}$ & 0.9
1084: & $4.5 \times 10^{37}$ & 8.4. & $1.1 \times 10^7$\\
1085: S3-Pa$\alpha$ & $5.3 \times 10^{38}$ & $2.7 \times 10^{64}$ & $1.2 \times 10^{65}$ & 0.18
1086: & \nodata & \nodata & $2.3 \times 10^6$\\
1087: S7 & $3.9 \times 10^{38}$ & $2.0 \times 10^{64}$ & $9.0 \times 10^{64}$ & 0.14
1088: & \nodata & \nodata & $9.4 \times 10^4$\\
1089: S8 & $1.2 \times 10^{38}$ & $6.2 \times 10^{63}$ & $2.8 \times 10^{64}$ & 0.043
1090: & \nodata & \nodata & $2.2 \times 10^5$\\
1091: diffuse & $4.9 \times 10^{39}$ & $2.5 \times 10^{65}$ & $5.5 \times 10^{65}$ & 0.85 
1092: & \nodata & \nodata & $2.05 \times 10^7$\\
1093: \enddata
1094: \tablenotetext{a}{Assumes a distance of 54.3 Mpc and no correction for extinction}
1095: \tablenotetext{b}{No correction for extinction}
1096: \tablenotetext{c}{Assumes Pa$\alpha$ extinction by a factor of 4.5 (A$_V$ = 12.1 
1097: for all sources except for the diffuse emission where it is a factor of 2.25 
1098: (see \S~\ref{ss-s1s2}).}
1099: \tablenotetext{d}{See \S~\ref{s-mass} for the calculation of the stellar mass.}
1100: \end{deluxetable}
1101: 
1102: \clearpage
1103: 
1104: \begin{figure}
1105: \plotone{f1.ps}
1106: \caption{This contour plot of the F160W (1.6 $\micron$) emission from in
1107: SBS0335-052E shows the location of all of the sources described in this
1108: paper.  The sources are displayed with a square root stretch to accommodate
1109: the dynamic range of the source strengths.  The designation of sources S1-S6
1110: are the the sources first named by \citet{thu97}. The sources S3-Pa$\alpha$,
1111: S7 and S8 are named in this paper. S7 was originally marked as ? in
1112: TSKEV. North is up and East is to the left.}
1113: \label{fig-map}
1114: \end{figure}
1115: 
1116: \clearpage
1117: 
1118: \begin{figure}
1119: \epsscale{.8}
1120: \plotone{f2.eps}
1121: \caption{Contour plots of ACS UV, Pa$\alpha$, H$_2$ and Near IR continuum
1122: F160W and F205W images.  The plots have a square root stretch.  Contour lines
1123: do not have the same flux levels between images. The contour plot titled P ALPHA
1124: MINUS S1 is the Pa$\alpha$ image with the point-like source S1 subtracted to show the
1125: underlying emission. North is up and East is to the left.}
1126: \label{fig-mcon}
1127: \epsscale{1.}
1128: \end{figure}
1129: 
1130: \begin{figure}
1131: \epsscale{0.8}
1132: \plotone{f3.eps}
1133: \caption{Linear stretch 40 by 40 pixel NICMOS camera 2 images of SBS0335-052
1134: in the continuum, Pa$\alpha$, H$_2$ and the ACS F140LP UV image.  The images are 
1135: approximately $3\arcsec$ on a side. Note the Airy rings around S1, S5 and S6
1136: (faint) in the 1.6 \micron\ continuum image and around S1 in the Pa$\alpha$
1137: image.  North is up and East is to the left.} 
1138: \label{fig-pa}
1139: \epsscale{1.}
1140: \end{figure}
1141: 
1142: \clearpage
1143: 
1144: \begin{figure}
1145: \plotone{f4.ps}
1146: \caption{This figure plots the radial profile of a) the Paschen $\alpha$ emission
1147: (F190N filter) from
1148: source S1 and the F160W continuum profiles of b) S1, c) S5 and d) S6 compared to the 
1149: radial profiles of a standard PSF star and that star convolved with Gaussian profiles 
1150: with the FWHM values listed in the legend.  The values correspond to 3/8, 5/8, 3/4 and
1151: 15/16 of a pixel respectively.  The profiles do not correspond to the PSF profiles in
1152: the regions beyond $0.15\arcsec$ due to diffuse emission or adjacent sources. See the 
1153: text for a discussion of the source sizes. The names of the sources appear in the 
1154: last row of the in figure legends.
1155: \label{fig-size}}
1156: \epsscale{1.}
1157: \end{figure}
1158: 
1159: \clearpage
1160: \begin{figure}
1161: \plotone{f5.ps}
1162: \caption{Combined contours of the ACS UV emission (blue), Near Infrared continuum (red)
1163: and Pa$\alpha$ (green).  Note the relative offsets of the UV continuum, Near
1164: Infrared continuum and Pa$\alpha$ for the source S3. North is up and East is to
1165: the left.}
1166: \label{fig-comb}
1167: \end{figure}
1168: 
1169: \clearpage
1170: 
1171: \begin{figure}
1172: \plotone{f6.ps}
1173: \caption{The solid line represents the equivalent width of Pa$\alpha$ versus
1174: time calculated from a Starburst99 simulation.  The asterisks are placed on
1175: this line at the location of the equivalent widths measured for each source.
1176: The location of the position of the S8 equivalent width is at the longest time
1177: since the simulated galaxy spends very little time at the same value of the
1178: equivalent width that is achieved at two earlier times. The dip in equivalent
1179: width at about $4 \times 10^6$ years is due to an increase in continuum flux
1180: rather than a decrease in line emission.  It marks the brightest continuum
1181: epoch of the cluster.}
1182: \label{fig-eqw}
1183: \end{figure}
1184: 
1185: \clearpage
1186: 
1187: \begin{figure}
1188: \plotone{f7.ps}
1189: \caption{The near infrared continuum contours (red) and H$_2$ emission (black).
1190: Note the slight offset of the H$_2$ emission to the east of the near
1191: infrared continuum for S1 and to the south for S2. North is up and East is to
1192: the left.}
1193: \label{fig-h2}
1194: \end{figure}
1195: 
1196: \end{document}
1197: