0708.3388/ms.tex
1: %%preprint produces a one-column, single-spaced document
2: %%\documentclass[12pt,preprint]{aastex}
3: %% preprint2 produces a double-column, single-spaced document:
4: %% \documentclass[preprint2]{aastex}
5: 
6: 
7: %\documentclass[manuscript]{aastex}
8: \documentclass[12pt,preprint]{aastex}
9: 
10: 
11: \slugcomment{ }
12: 
13: 
14: \shorttitle{Hypercompact HII regions}
15: \shortauthors{Eric Keto}
16: 
17: \begin{document}
18: 
19: 
20: \title{The Early Evolution of Massive Stars: Radio Recombination Line Spectra}
21: 
22: \author{Eric Keto }
23: \affil{Harvard-Smithsonian Center for Astrophysics, 60 Garden St., Cambridge, MA 02138}
24: \and
25: \author{Qizhou Zhang }
26: \affil{Harvard-Smithsonian Center for Astrophysics, 60 Garden St., Cambridge, MA 02138}
27: \and
28: \author{Stanley Kurtz }
29: \affil{CRyA, Universidad Nacional Aut\'onoma de M\'exico, Apdo. Postal 3-72, 
30:        58090 Morelia, Michoac\'an, Mexico}
31: 
32: 
33: \begin{abstract}
34: Velocity shifts and differential broadening 
35: of radio recombination lines
36: are used to estimate the densities and velocities of the ionized gas
37: in several hypercompact and ultracompact 
38: HII regions. 
39: These small HII regions
40: are thought to be at their earliest evolutionary phase and associated with
41: the youngest massive stars.
42: The observations suggest that these HII regions are
43: characterized by high densities, supersonic flows and steep density
44: gradients, consistent with accretion and outflows 
45: that would be associated with the formation of massive stars.
46: 
47: 
48: \end{abstract}
49: 
50: 
51: \keywords{stars: early type --- stars: formation ---  ISM: HII regions}
52: 
53: 
54: \section{Introduction}
55: 
56: Massive star forming regions are marked by small ($\leq 0.1$
57: pc) HII regions \citep{Gaume1995b, Depree1997, Tieftrunk1997, Depree1998, 
58: Wilson2003, Depree2000, Depree2004, Depree2005} 
59: that are characteristically different from many of the
60: few thousand ultracompact HII (UCHII) regions scattered throughout the
61: galaxy \citep{WoodChurchwell1989, Kurtz1994, Walsh1998, Giveon2005a, Giveon2005b}.  
62: First, these very small HII regions have electron
63: densities that are higher ($n_e \geq 10^5$ cm$^{-3}$) than those
64: of UCHII regions ($n_e\sim 10^4$ cm$^{-3}$)
65: \citep{Kurtz2000, KurtzFranco2002, Pratap1992}. 
66: It is unclear how high the densities might be, because 
67: the density cannot be reliably determined from the radio continuum
68: if the gas is optically thick.
69: Second, when
70: observed at centimeter wavelengths, these small HII regions often show
71: very broad radio recombination line (RRL) widths, more than 3 or 4 times the
72: thermal line width \citep{Altenhoff1981, Zijlstra1990, Afflerbach1994,
73: Depree1994, Gaume1995a, Depree1995, Depree1996, Depree1997,
74: Keto1995, Johnson1998, JaffeMP1999, Keto2002a,
75: Sewilo2004, KetoWood2006}.  Not all the small HII regions have this
76: property, but those that do are sometimes referred to as broad
77: recombination line objects.  The breadth of the RRLs could be caused by a 
78: combination of electron impact  (pressure) broadening associated 
79: with high gas density plus broadening caused by spatially unresolved 
80: gas motions. However,
81: the relative contributions
82: of broadening due to gas pressure and gas motions are not known.
83: Third, these HII regions often have
84: continuum spectral energy distributions (SEDs)
85: that are roughly linear with frequency
86: over decades in wavelength from the centimeter to the submillimeter
87: \citep{Hofner1996, Franco2000, Testi2000, Beuther2004, Ignace2004}.
88: The underlying cause of this scaling is not known although spatially
89: unresolved density
90: variations are a possibility.
91: 
92: These small and dense HII regions are sometimes referred to as 
93: hypercompact HII (HCHII) regions.  While the term
94: hypercompact implies a size smaller than ultracompact (UC), the three
95: properties of high electron densities, broad recombination lines, 
96: and linear SEDs, are also seen in some UCHII regions
97: ($\sim 0.1$~pc).  The common characteristic between these unusual
98: UCHII regions and the HCHII regions might be that
99: they are all young and
100: at the earliest evolutionary stage, but the 
101: unusual UCHII regions might
102: contain more or brighter stars than their smaller HCHII counterparts.
103: The HCHII and unusual UCHII regions represent 
104: some of our best sources of information on the processes in the formation of 
105: massive stars, and this motivates our understanding of their special properties.
106: 
107: In this paper we report on multi-frequency RRL observations that
108: address the origin of the three properties of the small HII regions
109: and the processes of massive star formation.  Because 
110: broadening due to pressure is dependent on frequency whereas broadening due
111: to gas motions is independent of frequency, the widths of RRLs
112: observed at more than one frequency can be used to separate the
113: contributions of each broadening mechanism. Since the pressure
114: broadening is proportional to the density, this separation allows a
115: determination of the electron density. Similarly, 
116: since the broadening due to gas motions is proportional to
117: the range of velocities, 
118: we can also measure the magnitude of the gas velocities.
119: 
120: 
121: \section{Observations}\label{observations}
122: 
123: We selected a few well-studied, very bright, very small HII regions 
124: that previous observations suggest have one or more of the 
125: three properties of the small HII regions. For each source
126: we have one observation of the H30$\alpha$ line made with the 
127: Submillimeter Array (SMA) and one or more observations of
128: centimeter wavelength RRLs made with
129: the National Radio Astronomy Observatory
130: Very Large Array (VLA)\footnote{The National Radio Astronomy
131: Observatory is a facility of the National Science Foundation
132: operated under cooperative agreement by Associated Universities,
133: Inc.} 
134: 
135: The SMA observations of the H30$\alpha$ line (231.900959 GHz) in the
136: HII regions G10.6-0.04, NGC7538-IRS1, W51e2, and G28.20-0.04 were made
137: in 2005 September with a spectral resolution of 0.83 km~s$^{-1}$, a
138: bandwidth of 2 GHz, and an angular resolution of 1.0$^{\prime\prime}$.
139: Observations of G45.07+0.14 were made in 2005 October with an angular
140: resolution of 0.4$^{\prime\prime}$.  The noise level obtained in each
141: observation was 60 mJy beam$^{-1}$ channel$^{-1}$.  Analysis of the data
142: indicated that some of the baselines were not well-defined, hence the
143: positions and flux densities of the sources are uncertain.  
144: The standard deviation of the flux of 3C454.4, measured once for each
145: observation,  is
146: 39\% of the average measured value
147: (table \ref{millimetercalibration}).
148: If we exclude the low measurement of 3C454.4 corresponding to 
149: the G45.07+0.14 data,
150: then the standard deviation falls to 14\%
151: of the average.  Our positions from the SMA agree with those of
152: the VLA to better than the angular resolution.  The HII regions 
153: are unresolved by the SMA observations,
154: and hence the widths of their spectral lines are unaffected by baseline
155: errors.  The SMA data were processed in the SMA data reduction package
156: MIR, and in MIRIAD.
157: 
158: Most of the VLA observations were made in 2003 and 2005. 
159: We reprocessed observations of the
160: H76$\alpha$ line in G45.07+0.14 \citep{Garay1986} from the VLA
161: archives.  The observations of the H66$\alpha$ line in G10.6-0.4 were
162: previously reported in \citet{Keto2002a}. The other observations have
163: not previously been published.  Details of the observations are
164: presented in table \ref{VLAobs}. The VLA data were processed in AIPS;
165: calibration data are given in table \ref{centimetercalibration}.
166: The recombination line spectra are shown in figures 1 to 5, and
167: the  line widths and velocities based are given in table
168: \ref{CMrrl}.  
169: 
170: \section{Gas velocities, pressure broadening, and electron densities}\label{derivation}
171: 
172: Radio recombination lines in HII regions are broadened by three contributions:
173: thermal broadening, dynamical broadening due
174: to spatially unresolved motions in the HII region including both ordered flows
175: and turbulence, and pressure broadening
176: due to high electron densities.
177: We seek to estimate the separate contributions of each 
178: of the three broadening components by comparing the widths of RRL 
179: observed at different frequencies. This is possible because the 
180: width due to 
181: thermal and dynamical broadening, is independent of frequency
182: while the width due to pressure broadening 
183: decreases with frequency as $\Delta\nu\sim\nu^{-4}$ 
184: \citep{BrocklehurstSeaton1972, Griem1974}. 
185: Since the dynamical broadening is proportional to the gas velocity while
186: the pressure broadening is proportional to gas density, from a pair of RRL observations
187: we can estimate the velocities and densities in the ionized gas.
188: 
189: We determine the contributions of each of the three broadening
190: mechanisms as follows. We assume that the dynamical broadening, $\Delta\nu_D$, 
191: and the thermal broadening, $\Delta\nu_T$,
192: combine in quadrature,
193: \begin{equation}
194: \Delta\nu_G = \sqrt{\Delta\nu_D^2 + \Delta\nu_T^2}. 
195: \label{eq:equation1}
196: \end{equation}
197: This width due to gas
198: motions, both thermal and spatially unresolved ordered flows and turbulence,
199: combines with the Lorentzian width, $\Delta\nu_L$, due to
200: pressure broadening 
201: to produce a Voigt profile with width, $\Delta\nu_V$.
202: We approximate the width of the Voigt profile
203: by an algebraic expression 
204: \citep[][ equation 2.72]{Olivero1977, GordonS2002},
205: \begin{equation}
206: \Delta\nu_V = 0.5343\Delta\nu_L + [\Delta\nu^2_G + (0.4657\Delta\nu_L)^2]^{1/2}
207: \label{eq:voigtApprox}
208: \end{equation}
209: where $\Delta\nu_V$ is the full-width at half-maximum (FWHM) of the
210: Voigt profile, $\Delta\nu_L$ the FWHM of the Lorentzian, and
211: $\Delta\nu_G$ the FWHM of the Gaussian.  Because of the rapid decrease
212: in pressure broadening with frequency, the width of an RRL at a high
213: enough frequency must be  due solely to thermal broadening 
214: and spatially
215: unresolved gas motions.
216: Therefore we assume that this broadening,
217: $\Delta\nu_G$ in equation \ref{eq:voigtApprox}, is given by the
218: observed width of our highest frequency RRL, H30$\alpha$. 
219: If we compare the width of the H30$\alpha$ line to that of a lower
220: frequency line,
221: any increased width observed in the lower frequency line
222: must be due to pressure broadening 
223: because the width $\Delta\nu_G$ 
224: is independent of frequency 
225: We can then calculate the width due to pressure broadening,  $\Delta\nu_L$,
226: by solving the 
227: quadratic equation \ref{eq:voigtApprox}.  
228: 
229: 
230: The range of gas velocities is estimated from equation (\ref{eq:equation1}).
231: Assuming an ionized gas temperature of 8000 K,
232: the dynamical widths are
233: 15.3, 8.3, 27.1, 18.7 and 53.9 kms$^{-1}$
234: for G10.6-0.4, G28.20-0.04N, G45.07+0.14, W51e2, and
235: NGC 7538-IRS1 respectively.
236: 
237: The electron density may be estimated from the ratio of $\Delta\nu_L$ and $\Delta\nu_T$.
238: For $\alpha$ ($\Delta$N
239: = 1) transitions, the ratio of the Lorentzian width, $\Delta\nu_L$, resulting
240: from pressure broadening, and the thermal width, $\Delta\nu_T$, is given
241: by \citep{Keto1995},
242: \begin{equation}
243: {{\Delta\nu_L}\over{\Delta\nu_T}} = 
244: 1.2 \bigg( {{n_e}\over{10^5}} \bigg)\bigg({{N}\over{92}}  \bigg)^7
245: \label{eq:ratio}
246: \end{equation}
247: where  $n_e$ is the electron
248: density in cm$^{-3}$, and $N$ is the principal quantum number.
249: 
250: As an example, we calculate the electron density in
251: G10.6-0.4 from the H30$\alpha$ and H92$\alpha$ RRL observations as
252: follows.   We use
253: the width of the H30$\alpha$ line, 24.5 kms$^{-1}$ (table \ref{CMrrl}) to estimate
254: $\Delta\nu_G$ at the frequency of the H92$\alpha$ line.  
255: Thus $\Delta\nu_G = 0.678$ MHz.
256: The measured width of the H92$\alpha$ line
257: is $\Delta\nu_V({\rm  H}92\alpha) = 1.03$ MHz. 
258: With these values and equation \ref{eq:voigtApprox}, the estimate of the
259: width due to pressure broadening  is $\Delta\nu_L = 0.578$ MHz
260: or 20.9 kms$^{-1}$. 
261: The thermal width, assuming a temperature of 8000 K, is $\Delta\nu_T =
262: 1.43$ MHz or 19.1 kms$^{-1}$. The width due to 
263: gas motions
264: is $\Delta v_D = \sqrt{24.5^2 - 19.1^2} = 15.3$ kms$^{-1}$. 
265: The electron density from equation \ref{eq:ratio} is 
266: $9.1\times 10^4$ cm$^{-3}$. Similarly, from the observations of
267: the other centimeter RRLs observed in G10.6-0.4,  the electron densities are
268: $7.8\times 10^5$ cm$^{-3}$ from the H66$\alpha$ line and $2.5\times
269: 10^6$ cm$^{-3}$ from the H53$\alpha$ line. Comparing these three
270: RRL, the electron density increases with the RRL
271: frequency.  If there is a density gradient in G10.6-0.4 this would
272: suggest that the higher frequency lines are seeing deeper into the
273: nebula where the electron densities are higher.
274: 
275: \subsection{Uncertainties}
276: 
277: We can check our approximation of negligible pressure broadening
278: in the H30$\alpha$ line from \ref{eq:ratio}.  
279: For the H30$\alpha$ line, this ratio is less than one as long as
280: the density is less than $3\times 10^8$ cm$^{-3}$. Thus our
281: assumption of negligible pressure broadening in the H30$\alpha$
282: line is valid up to this density.
283: 
284: The lines in our sample are not
285: consistently  Gaussians or by Lorentzian. One possibility
286: is that the asymmetric line profiles indicate asymmetries in
287: the gas motions in the HII regions,  for example, ordered flows rather than
288: isotropic turbulence. 
289: We determine the line widths by fitting Gaussians because
290: this provides a reasonable estimate if the
291: observed profiles
292: differ from a Gaussian most strongly in the wings and 
293: are similar in the core where the width is measured. 
294: A comparison of the line widths derived by fitting Gaussians
295: to Voigt profiles provides  an informal
296: estimate of the possible error introduced by this method. If
297: the Voigt profile is composed of a Gaussian and a Lorentzian
298: of equal FWHM, then a Gaussian fit overestimates the 
299: true FWHM of the Voigt by 10\%. 
300: 
301: We find that the procedure of using Gaussian line widths in equation
302: \ref{eq:voigtApprox} to estimate the pressure broadening 
303: is a more reliable method than some alternatives. For
304: example, one could imagine fitting a Voigt profile directly to 
305: an observed low frequency line to determine simultaneously both
306: the Gaussian and Lorentzian components. However, we find
307: this unreliable because the fitting procedure assigns the line width
308: to either the Gaussian or Lorentzian components primarily on the
309: shape of the line wings where the profiles differ most strongly, but
310: the signal-to-noise ratio is the weakest. Alternatively, one might
311: estimate the FWHM directly from the 3 channels with
312: the peak and half-power emission. However, because this estimate 
313: relies on  single channel measurements it is less
314: reliable than a line-fitting procedure that uses measurements at
315: all the channels across the line profile.  
316: 
317: Several lines of evidence, discussed later in the
318: paper suggest that there are density variations within the HII
319: regions.  Because all the HII regions in our sample are unresolved,
320: the measured electron density represents the average within the
321: HII region. 
322: The size of the HII region within the beam
323: (filling factor) does not affect this estimate of the 
324: average density
325: because the electron densities are
326: determined from the line widths rather than the emission
327: intensity. However, the unknown density structure of the HII 
328: region creates some uncertainty. If there are
329: significant variations in optical depth with frequency then
330: it is possible that lines at different frequencies could
331: be generated in different locations within the HII region.
332: 
333: Aside from the systematic effects, the uncertainty in the average 
334: electron density is given
335: by the standard propagation of errors, assuming the errors in the
336: widths are independent. 
337: Uncertainties in the electron densities are listed in table \ref{electrondensities} 
338: and for the line center velocities
339: and widths in table \ref{CMrrl}.
340: 
341: 
342: 
343: \subsection{Electron densities from the radio continuum}
344: 
345: At the high densities in our HII regions, the
346: centimeter continuum emission is optically thick to free-free emission. 
347: This can be
348: determined from the formula 
349: \citep[eqn. A.1b,][]{MezgerHenderson1967}
350: or \citep[eqn. 10,][]{Keto2003},
351: \begin{equation}\label{eq:opticalDepth}
352: \tau_\nu = 8.235\times 10^{-2} \bigg({{T_e}\over{{\rm K}}}   \bigg) ^{-1.35}
353: \bigg({{\nu}\over{{\rm GHz}}}   \bigg) ^{-2.1}
354: \bigg({{\rm n_e^2L}\over{{\rm pc\ cm}{^{-6}}}}   \bigg)
355:  \end{equation}
356: where $T_e$ is the electron temperature and $n_e^2L$ is
357: the electron density squared times the path length.
358: For a characteristic size of 0.01 pc, an HII region has an optical depth
359: of unity at 22 GHz (1.3 cm) if the electron density is $4\times 10^5$ cm$^{-3}$. 
360: This means
361: that the formulae for optically thin emission
362: \citep[e.g.][]{MezgerHenderson1967} are not useful
363: for the HII regions in our sample,
364: and probably not for small, high-density HII regions in general.
365: 
366: If the gas is optically thick to free-free emission, and the
367: electron density is estimated from the radio continuum by incorrectly
368: assuming optically thin emission, then the derived electron density
369: is just that density required to make the optical depth unity.
370: This can be understood by a simple example.
371: The radio free-free
372: emission through a slab of uniform gas density and temperature is,
373: \begin{equation}\label{eq:exactEmission}
374: S_\nu =  2k T_e (\nu^2/c^2)
375: \big[1 - e^{-\tau_\nu}\big] ({\rm Jy/Sr})
376: \end{equation}
377: The optically thin approximation amounts to replacing
378: $[1-\exp(-\tau_\nu)]$ by $\tau_\nu$ so that,
379: \begin{equation}\label{eq:thinEmission}
380: S_\nu \approx  2k T_e (\nu^2/c^2)\tau_\nu
381: \end{equation}
382: whereas, in the optically thick limit,
383: \begin{equation}\label{eq:thickEmission}
384: S_\nu \approx  2k T_e (\nu^2/c^2)
385: \end{equation} 
386: Thus if the HII region is optically thick (equation \ref{eq:thickEmission}) 
387: but $\tau_\nu$ is derived assuming equation \ref{eq:thinEmission}, then the
388: derived value of the optical depth is 1, 
389: and the electron density derived from equation \ref{eq:opticalDepth}
390: is the density required to make $\tau_\nu=1$. 
391: Thus at 22 GHz (1.3 cm),
392: the electron density derived for an HII region with a size of 0.01 pc is never
393: be greater than about $10^5$ cm$^{-3}$ if the optically thin formulae are used --
394: regardless of the true density.  
395: This underestimate also affects
396: quantities derived from the electron density such as 
397: the number of ionizing photons required to
398: balance recombinations which in turn is often used to derive the
399: spectral types of the exciting stars.
400: 
401: Formulas for spherical HII regions that have
402: a density gradient such that the central part of the HII region is optically thick
403: are derived in \citet{Keto2003}. However, these formulae require a 
404: specification of the size of the optically thick region.
405: 
406: 
407: \subsection{The correlation of line widths, velocities, and frequencies}
408: 
409: As shown in table \ref{CMrrl}, the line widths decrease with
410: increasing frequency in all cases.  This suggests the increasing
411: importance of pressure broadening at lower frequencies.  For sources
412: with more than one centimeter RRL observed, the electron densities
413: increase with the increasing frequency of the centimeter line used in
414: the analysis. The one exception is the electron density calculated
415: from the H66$\alpha$ line in NGC7538-IRS1. This line is very much
416: wider and out of character with the other measurements in our sample.
417: Aside from this one exception, the correlation of density with
418: observing frequency indicates that the higher frequency lines are
419: associated with higher density gas.  The line center velocities also
420: increase (redshift) with frequency, except for the lines of
421: NGC7538-IRS1.
422: 
423: The combination of the two correlations, line center
424: velocity with frequency and line width with frequency was first noticed
425: in W3(OH) by \citet{BerulisErshov1983}.  \citet{WelchMarr1987} and
426: \citet{Keto1995} interpreted the dual correlation as evidence for an 
427: accelerating flow within a density gradient \citep{Guilloteau1983}.  
428: A full explanation is given in \citet{Keto1995} (see also 
429: \citet{BrocklehurstSeaton1972})and can be summarized as follows:
430: The peak intensity of RRLs is proportional to
431: the density squared and inversely proportional to the line width. 
432: Thus the peak intensity of optically thin
433: high frequency lines with negligible pressure 
434: broadening is proportional to the density squared.
435: The peak intensity of lower frequency lines is proportional
436: to the first power of the density since the pressure broadened
437: line width is itself proportional to density.
438: Thus in spatially unresolved observations of gas with a density gradient,
439: the higher frequency RRLs are dominated by emission from the higher density
440: gas while the lower frequency RRLs include contributions from both high and
441: low density gas. An observed difference in velocity between the high
442: and low frequency lines thus indicates a difference in velocity between the
443: high and low density gas.
444: 
445: The correlation of velocities and width with frequency therefore indicates
446: that in these HII regions there is both a flow and a density gradient. The two
447: are naturally related by the conservation of mass and the geometry of the
448: flow. In general, if the flow is either outward and diverging or inward and
449: converging then conservation of mass requires that there be a density gradient
450: except in the unique circumstance that
451: the variation of the flow speed exactly cancels the geometric
452: divergence or convergence. 
453: 
454: 
455: 
456: \section{Continuum Spectral Indices}
457: 
458: We determined the spectral energy distrubutions (SEDs) of the HII
459: regions in our sample using our own millimeter and centimeter
460: observations and data collected from the literature. 
461: We selected only those data with
462: angular resolutions close to 1$^{\prime\prime}$ specifically
463: excluding single dish observations. 
464: Most of the data for NGC7538-IRS1 and G45.07+0.14 are from 
465: \citet[][table 1]{Pratap1992} and \citet[][table 3]{Garay1986}, 
466: respectively. The rest of the observations are 
467: listed in table \ref{fluxdata}.
468: Most of these data 
469: do not include estimated uncertainties. However, the calibration
470: of radio data is fairly standard. Many of the reported observations
471: were made at the VLA that typically calibrated flux densities 
472: to better than 20\% at 1.3 cm at the dates of the observations
473: and much better at longer wavelengths.
474: As noted in \S \ref{observations}, we estimated our SMA 
475: observations to be calibrated to better than 39\%. 
476: 
477: The spectral energy distributions are shown in figures 6 through 10.
478: Along with each observed SED, shown as crosses, 
479: the figures show two model SEDs as solid lines.  One model SED
480: is based on a single HII region with a power-law
481: density gradient. The continuum flux density for this model
482: is calculated as in \citet{Keto2003}. 
483: The radio free-free 
484: emission of a spherical HII region seen as an unresolved source is,
485: \begin{equation}
486: S_\nu = 4\pi k T_e \nu^2/c^2  
487: \int^{\theta_0}_0 \theta\big[1 - e^{-\tau_\nu}\big]d\theta
488: \end{equation}
489: Here, $\theta$ is an angular coordinate on the plane of the
490: sky, 
491: $\theta = \sqrt{x^2 + y^2}/D$
492: with coordinate axes $x$ and $y$ are centered on the HII region,
493: $\theta_0 = R_0/D$ where 
494: $R_0$ is
495: the radius of the HII region, and $D$
496: the  distance to the source.
497: An approximation to the free-free optical depth is,
498: \citet[][equation A.1a]{MezgerHenderson1967}, 
499: \begin{equation}
500: \tau_\nu = 8.235\times 10^{-2} \bigg( {{T_e}\over{^\circ K}} \bigg) ^{-1.35}
501:         \bigg( {{\nu}\over{GHz}} \bigg)^{-2.1}
502:         \bigg( {{EM}\over{{\rm pc}\ {\rm cm}^{-6} }} \bigg)
503: \end{equation}
504: where $T_e$ is the electron temperature, and $EM$ is the emission measure, $n_e^2L$.
505: 
506: The model parameters for all sources
507: are listed in table \ref{ModelSED}.  The other model is an SED based
508: on constant density HII regions.  The figures for NGC7538-IRS1 and W51e2 
509: show the SEDs from multi-component models suggested
510: by \citet{Pratap1992} and \citet{Rudolph1990}. These are both based on
511: constant density HII regions plus dust.  The dust emission enters as
512: \citep[][ equation 1]{Pratap1992},
513: \begin{equation}\label{eq:dust}
514: S_\nu({\rm Jy}) = 52.36 {M}_{\odot} 
515: \bigg(  {{0.2}\over{\lambda_{mm}}} \bigg)^{\beta+3}
516: \bigg[ \exp \bigg(  {{14.4}\over{\lambda_{mm}T_d}} \bigg) -1\bigg]^{-1}D^{-2}_{kpc}
517: \end{equation}
518: where $S_\nu$ is the dust emission in Jy, $M$ is the mass of the
519: dusty gas in M$_\odot$, $\beta$ is the spectral index of the dust 
520: emissivity, $T_d$ is the
521: dust temperature in Kelvin, and $D$ is the distance in kpc.
522: 
523: The multi-component model for NGC7538-IRS1 \citep{Pratap1992} nicely
524: illustrates how a density gradient within an HII region can stretch
525: the transition between optically thick and thin frequencies
526: resulting in an SED with an intermediate spectral index. In this
527: model there are two concentric spherical regions each of uniform but different
528: densities. 
529: The smaller, higher density component produces an
530: emission curve that is shifted upward and to the right of the curve
531: for the larger, lower density component (figure 6). The emission from
532: the two components combines to produce an SED of intermediate slope.
533: An HII region with a continuous
534: density gradient can be considered as the logical extension of this
535: two component model to many nested smaller and denser HII
536: regions.  \citet{Lugo2004} also model the NGC 7538-IRS1 HII region with
537: a density gradient that they suggest is due to a wind off a
538: photo-evaporating disk.
539: 
540: The SED modeling combined with the recombination line measurements
541: shows that dust is not a significant contributor to the continuum
542: emission measured at high angular resolution from these sources.  (See
543: also \citet{Kurtz2005}, \citet{Depree1998}, and \citet{Depree2000}.)
544: Consider W51e2 as an example.  As seen in figure 7, the HII region can
545: be modeled with uniform, low-density ionized gas, with the 147 and
546: 230~GHz emission arising from dust \citep{Rudolph1990}.  Figure 7
547: shows that with this approach, the free-free continuum emission at 230
548: GHz would be about a factor of 10 below the observed emission.
549: However, the line-to-continuum ratio should be approximately one at
550: this frequency \citep{GordonS2002}. Hence, if the free-free continuum
551: were this low, the H30$\alpha$ recombination line would also have a
552: much lower flux density --- so low that the line would not be detected
553: in our observations.  Stated another way, the recombination line
554: emission cannot be 10 times greater than the free-free continuum
555: emission at this frequency.  Therefore, most of the continuum emission
556: observed by the interferometer at 231.901 GHz must be free-free, hence
557: the electron density must be higher, and a density gradient is
558: required.
559: 
560: The high frequency emission observed in large beams ($20-30^{\prime\prime}$)
561: around these bright HII regions
562: certainly arises from dust. Again take W51e2 as an example.
563: \citet{Jaffe1984} report a flux density
564: of 1200 Jy~beam$^{-1}$ at 750 GHz and 40$^{\prime\prime}$ angular
565: resolution. We assume, as in the model of \citet{Rudolph1990}, that
566: this emission is from cool dust in the large-scale molecular cloud
567: surrounding the HII region.  
568: Because the dust is cool (50 - 100 K), a large mass of dust is required
569: (4500 M$_\odot$, \citet{Rudolph1990}) to produce the observed 750 GHz flux
570: density.
571: While the emission from this dust is also sufficient to produce the 
572: 90 and 230 GHz flux density 
573: that is observed by interferometers,  the cool dust
574: is not the source of this emission because it is not possible that this
575: large mass of cool dusty gas is within the spatial scale defined by the
576: arc second angular resolution of the interferometers.
577: However, the ionized gas is sufficiently hot
578: ($10^4$ K) that only a small mass (0.01 M$_\odot$), consistent with
579: the mass of the HII regions, is required to produce the 90 - 230 GHz
580: flux density. Similar considerations apply to the multi-component models of
581: other HII regions, for example, the model discussed earlier for
582: NGC7538-IRS1 \citep{Pratap1992}.
583: 
584: On a spatial scale similar to that of HCHII
585: and UCHII regions, continuum emission from dust has been observed 
586: by the SMA in the nearby (1.7 kpc) massive star forming region NGC6334 by
587: \citet{Hunter2006}.  The dusty clumps of arcsecond scale in NGC6334
588: have continuum emission on the order of 1~Jy.  Scaled to the
589: distance of W51, the dust in such clumps would be a minor contributor
590: (60 mJy) to the continuum emission.  Thus, high angular resolution
591: observations of dusty clumps in NGC6334 are consistent with the
592: interpretation of negligible dust contribution to the emission
593: observed at high angular resolution in more distant bright
594: HII regions. 
595: 
596: 
597: \section{Discussion}\label{discussion}
598: 
599: The observations suggest that pressure broadening of the centimeter
600: RRLs is important in small HII regions. This contradicts the
601: interpretation of some previous studies. In particular, on the basis
602: of electron densities calculated by assuming constant gas density and
603: optically thin emission, \citet{Gaume1995a}, \citet{Depree1997}, and
604: \citet{Depree2005} have suggested that the electron densities in
605: NGC7538-IRS1, W49, and SgrB2 are too low to cause significant pressure
606: broadening in the H66$\alpha$ line.  However, the RRL observations
607: suggest that the electron density in NGC7538-IRS1 is two orders of
608: magnitude greater than the $10^5$ cm$^{-3}$ derived by
609: \citet{Gaume1995a} and sufficient to cause significant pressure
610: broadening.  \citet{JaffeMP1999} report observations of millimeter RRL of
611: several sources, including G10.6-0.4 and NGC7538-IRS1.  They concluded
612: that the RRL widths are not due to pressure broadening because the
613: observed widths of different recombination lines do not scale as the
614: fourth power of the frequency.   
615: Their conclusion was based on a comparison of observations with
616: different angular resolutions. It has been demonstrated that low
617: angular resolution observations may suffer confusion from
618: more-extended, lower-density gas \citep{Simpson1973a, Simpson1973b,
619: GS1974, Smirnov1984, GordonS2002}.  We concur with
620: \citet{Gaume1995a, Depree1997, Depree2005, JaffeMP1999} 
621: that broadening due to gas motions is a significant contribution to
622: the observed line widths, but we find that pressure broadening is also significant,
623: particularly for the lower frequency lines.
624: 
625: The intermediate SEDs of small HII regions
626: imply a density gradient as would arise from a divergence or convergence in
627: the flow of ionized gas in an HII region. 
628: Other interpretations are possible. 
629: \citet{Franco2000}, \citet{KurtzFranco2002} and \citet{KimKoo2002} 
630: suggested that the
631: various densities are due to the hierarchical structure of the
632: molecular cloud prior to its ionization by a massive star.
633: The intermediate SEDs can also be created by density variations 
634: other than gradients.
635: \citet{Ignace2004} suggested that
636: an HII region is actually an ensemble of differently-sized clumps,
637: each of uniform but different density. One problem with these models of
638: pre-existing structure or clumps is that their survival time 
639: is approximately their sound crossing time.
640: This timescale seems impossibly short if the high density clumps are
641: substructures within an HCHII  ($t_c \leq 1000$ yrs for $R < 0.01$ pc).
642: 
643: The interpretation of density gradients and resulting high electron
644: densities in small HII regions also contradicts the interpretation of
645: significant dust emission in a number of other studies.  For example,
646: \citet{Rudolph1990, Pratap1992, Testi2000, Schilke1990, Beuther2004,
647:   Beuther2006} model the small HII regions, W51e2, NGC7538-IRS1,
648: G9.62+0.19-F, and Orion-KL-SMA1 respectively, with low and constant
649: density gas and assume that dust makes up the emission at high
650: frequencies.  \citet{Testi2000} also provide an alternative model with
651: a stellar wind rather than dust to supply the high frequency emission.
652: These models with low and constant density have difficulty in
653: explaining the broad RRL line widths. If the density is low, then the
654: widths must be due to dynamical broadening created by a supersonic
655: flow.  However, a constant density is not generally consistent with
656: supersonic flows. In general, a strong density 
657: gradient is required to create the
658: strong pressure gradient needed to drive the flow. 
659: 
660: \section{Conclusions}
661: 
662: 1. The radio recombination line widths increase with decreasing 
663:    frequency in all cases. This indicates the greater importance 
664:    of pressure broadening at lower frequencies.
665: 
666: 2. The electron densities calculated by comparison of 
667:    centimeter recombination lines with the H30$\alpha$ millimeter line
668:    increase with increasing frequency of the centimeter line.
669:    This suggests that the higher frequency lines are generated in 
670:    denser gas and suggest that density gradients exist in small
671:    HII regions.
672: 
673: 3. Supersonic line widths are observed in the H30$\alpha$ RRL.
674:    This line is at a frequency high enough that pressure
675:    broadening is negligible provided that the density
676:    of the emitting gas is less than $10^8$ cm$^{-3}$. The large
677:    line widths suggest the presence of supersonic motions
678:    within the HII regions.
679: 
680: 3. The electron densities in HCHII and UCHII regions are generally
681:    high enough that the gas is optically thick at centimeter
682:    wavelengths. This means that electron densities cannot be calculated
683:    from radio continuum observations 
684:    using formulae that assume optically thin emission.
685:    Instead formulae appropriate for partially
686:    optically thick emission must be used.
687: 
688: 4. At high frequencies ($\geq 100$ GHz) and high angular resolution
689:    ($\sim 1^{\prime\prime}$) the observed continuum emission is mostly
690:    free-free emission and not from dust.  Comparisons based on high
691:    frequency observations with different beam sizes can lead to erroneous
692:    conclusions. High frequency emission at low angular resolution
693:    ($\geq 10^{\prime\prime}$) is probably from dust, but associated with the
694:    large-scale, overlying molecular cloud.
695: 
696: S. Kurtz acknowledges support from UNAM, DGAPA project IN106107.
697: 
698: \clearpage
699: 
700: 
701: \begin{thebibliography}{}
702: 
703: %\setlength\itemsep{0.0pt}%
704: %\setlength\parskip{0.0pt}%
705: 
706: \bibitem[Afflerbach et al.(1994)]{Afflerbach1994}
707: 	Afflerbach, A., Churchwell, E., Hofner, P., \& Kurtz, S. 1994, ApJ, 437, 697
708: 
709: \bibitem[Altenhoff, Stritmatter \& Wendker(1981)]{Altenhoff1981}
710:        Altenhoff, W.J., Stritmatter, P.A., \& Wendker, H.J. 1981, AA, 93, 48
711: 
712: \bibitem[Berulis \& Ershov(1983)]{BerulisErshov1983}
713: 	Berulis, I.I. \& Ershov, A.A., 1983, Soviet Astr. Lett, 9,341
714: 
715: \bibitem[Beuther et al.(2004)]{Beuther2004}
716: 	Beuther, H., Zhang, Q., Greenhill, L. J., Reid, M. J., Wilner, D., Keto, E., 
717: 	Marrone, D., Ho, P. T. P., Moran, J. M., Rao, R., Shinnaga, H., \& Liu, S.-Y. 
718:         2004, ApJ, 616, L31
719: 
720: \bibitem[Beuther et al.(2006)]{Beuther2006}
721: 	Beuther, H., Zhang, Q., Reid, M. J., Hunter, T. R., Gurwell, M., Wilner, D., 
722:         Zhao, J.-H., Shinnaga, H., Keto, E., Ho, P. T. P., Moran, J. M., \& Liu, S.-Y.
723:         2006, ApJ, 636, 323
724: 
725: \bibitem[Brocklehurst \& Seaton(1972)]{BrocklehurstSeaton1972}
726: 	Brocklehurst, M., \& Seaton, M.J. 1972, MNRAS, 157, 179
727: 
728: \bibitem[Brown, Lockman, Knapp(1978)]{Brown1978}
729: 	Brown, R.L., Lockman, F.J., \& Knapp, G.R., 1978, ARAA, 16, 445
730: 	
731: %\bibitem[Carral et al.(1999)]{Carral1999}
732: %	Carral, P., Kurtz, S., Rodriguez, L.F., Marti, J., Lizano, S., Osorio, S., 1999,
733: %	Rev. Mex. AA, 35, 97
734: 
735: %\bibitem[Churchwell(1997)]{Churchwell1997}
736: %	Churchwell, E., ApJ, L59
737: 
738: \bibitem[DePree et al.(1994)]{Depree1994}
739:         DePree, C.G., Goss, W.M., Palmer, P., \& Rubin, R. 1994, ApJ 428, 670
740: 
741: \bibitem[DePree et al.(1995)]{Depree1995}
742:        DePree, C.G., Gaume, R.A., Goss, W.M., \& Claussen, M.J. 1995, ApJ, 451, 284
743: 
744: \bibitem[DePree et al.(1996)]{Depree1996}
745:        DePree, C.G., Gaume, R.A., Goss, W.M., \& Claussen, M.J. 1996, ApJ, 464,788
746: 
747: \bibitem[DePree et al.(1997)]{Depree1997}
748:        DePree C.G., Mehringer, D.M.,\& Goss, W.M. 1997, ApJ, 482, 307
749: 
750: \bibitem[DePree et al.(1998)]{Depree1998}
751:        DePree, C.G., Goss, W.M., \& Gaume, R.A. 1998, ApJ, 500, 847
752: 
753: \bibitem[DePree et al.(2000)]{Depree2000}
754: 	De Pree, C. G.; Wilner, D. J.; Goss, W. M.; Welch, W. J.; McGrath, E.
755: 	2000, ApJ, 540, 308
756: 
757: \bibitem[DePree et al.(2004)]{Depree2004}
758: 	De Pree, C.G., Wilner, D.J., Mercer, A.J., Davis, L.E.,  Goss, W.M., Kurtz, S.
759: 	2004, ApJ, 600, 286
760: 
761: \bibitem[DePree et al.(2005)]{Depree2005}
762:         DePree, C.G., Wilner, D.J., Deblasio, J., Mercer, A.J., \& 
763:         Davis, L.E. 2005, ApJ, 624, L101
764: 
765: \bibitem[Franco et al.(2000)]{Franco2000}
766:         Franco, J., Kurtz, S., Hofner, P., Testi, L., Guillermo, G.-S., \& 
767:         Martos, M. 2000, ApJ, 542, L143
768: 
769: \bibitem[Garay et al.(1985)]{Garay1985}
770: 	Garay, G., Reid, M.J., \& Moran, J.M. 1985, ApJ, 289, 681
771: 	
772: \bibitem[Garay et al.(1986)]{Garay1986}
773: 	Garay, G., Rodriguez, L.F., \& van Gorkom, J.H. 1986, ApJ, 309, 553
774: 	
775: \bibitem[Gaume et al.(1995a)]{Gaume1995a} % NGC7538
776: 	Gaume, R.A., Goss, W.M., Dickel, H.R., Wilson, T.L., \& Johnston, K. 
777:         1995a, ApJ, 438, 776
778: 
779: \bibitem[Gaume et al.(1995b)]{Gaume1995b} % SgrB2
780: 	Gaume, R.A., Claussen, M.J., DePree, C.G., Goss, W.M., \& Mehringer, D.M. 
781:         1995b, ApJ, 449, 663
782: 
783: \bibitem[Giveon et al.(2005a)]{Giveon2005a}
784: 	 Giveon, U.,  Becker, R. H., Helfand, D. J.,  White, R. L., 2005, AJ, 129, 348
785: 
786: \bibitem[Giveon et al.(2005b)]{Giveon2005b}
787: 	 Giveon, U.,  Becker, R. H., Helfand, D. J.,  White, R. L., 2005, AJ, 130, 156
788: 
789: \bibitem[Gordon \& Sorochenko(2002)]{GordonS2002}
790: 	Gordon, M.A. \& Sorochenko, P.N. 2002, {\it Radio Recombination Lines. 
791: 	Their Physics and Astronomical Applications}, (Dordrecht: Kluwer)
792: 
793: \bibitem[Griem(1974)]{Griem1974}
794: 	Griem, H.R. 1974, {\it Spectral Line Broadening by Plasmas}, 
795:         (New York: Academic)
796: 
797: \bibitem[Guilloteau et al.(1983)]{Guilloteau1983}
798: 	Guilloteau, S., Stier, M.T., \& Downes, D., 1983, AA, 126, 10
799: 
800: \bibitem[Gulyaev \& Sorochenko(1974)]{GS1974}
801: 	Gulyaev, S.A., \& Sorochenko, R.L. 1974, Astron. Zh., 51, 1237
802: 
803: \bibitem[Ho \& Haschick(1981)]{HoHaschick1981}
804: 	Ho, P., \& Haschick, A.D. 1981, ApJ, 248, 622
805: 
806: %\bibitem[Ho \& Haschick(1986)]{HoHaschick1986}
807: %	Ho, P., \& Haschick, A.D. 1986, ApJ, 304, 501
808: 
809: %\bibitem[Ho \& Young(1996)]{HoYoung1996}
810: %	Ho, P.T.P., \& Young, L.M. 1996, ApJ, 472, 742
811: 	
812: \bibitem[Hoang-Binh(1972)]{HoangBinh1972}
813: 	Hoang-Binh, D. 1972, M\'{e}m. Roy. Soc. Li\'{e}ge 3, 367
814: 
815: \bibitem[Hofner et al.(1996)]{Hofner1996}
816:         Hofner, P., Kurtz, S., Churchwell, E., Walmsley, C. \& Cesaroni, R. 1996,
817:         ApJ, 460, 359
818: 
819: \bibitem[Hunter et al.(1997)]{Hunter1997}
820: 	Hunter, T., Phillips, T.G., \& Menten, K.M. 1997, ApJ, 478, 283
821: 
822: \bibitem[Hunter et al.(2006)]{Hunter2006}
823: 	Hunter, T.R., Brogan, C.L., Megeath, S.T., Menten, K.M., Beuther, H., 
824:         \& Thorwirth, S. 2006, ApJ, 649, 888
825: 	
826: \bibitem[Ignace \& Churchwell(2004)]{Ignace2004} 
827:         Ignace, R., \& Churchwell, E. 2004, ApJ, 610, 351
828: 
829: \bibitem[Jaffe et al.(1984)]{Jaffe1984}
830: 	Jaffe, D.T., Becklin, E.E., \& Hildebrand, R.H. 1984, ApJ, 279, L51
831: 
832: \bibitem[Jaffe \& Mart\'\i n-Pintado(1999)]{JaffeMP1999}
833:         Jaffe, D.T., \& Mart\'\i n-Pintado, J. 1999, 520, 162
834: 
835: \bibitem[Johnson et al(1998)]{Johnson1998}
836:         Johnson, C.O., DePree, C.G., \& Goss, W.M. 1998, ApJ, 500, 302
837: 
838: %\bibitem[Keto et al.(1987)]{Keto1987}
839: %	Keto, E., Ho, P.T.P., \& Haschick, A.D. 1987, ApJ, 318, 712
840: 
841: %\bibitem[Keto et al.(1988)]{Keto1988}
842: %	Keto, E., Ho, P.T.P., \& Haschick, A.D. 1988, ApJ, 324, 920
843: 
844: %\bibitem[Keto(1990)]{Keto1990}
845: %	Keto, E. 1990, ApJ, 355, 190
846: 
847: %\bibitem[Keto(1991)]{Keto1991}
848: %	Keto, E. 1991, ApJ, 371, 163
849: 
850: \bibitem[Keto et al.(1995)]{Keto1995}
851:         Keto, E., Welch, W., Reid, M., \& Ho, P.T.P. 1995, ApJ, 444, 765
852:         
853: \bibitem[Keto(2002a)]{Keto2002a}
854:         Keto, E. 2002a, ApJ, 568, 754
855: 
856: \bibitem[Keto(2002b)]{Keto2002b}
857:         Keto, E. 2002b, ApJ, 580, 980
858: 
859: \bibitem[Keto(2003)]{Keto2003}
860:         Keto, E. 2003, ApJ, 599, 1196
861: 
862: \bibitem[Keto \& Wood(2006)]{KetoWood2006} 
863: 	Keto, E., \& Wood, K. 2006, \apj, 637, 850
864: 
865: \bibitem[Keto(2006)]{Keto2006}
866:         Keto, E. 2006, ApJ, submitted, astro-ph/0603856
867: 
868: \bibitem[Kim \& Koo(2002)]{KimKoo2002}
869: 	Kim, K.-T., \& Koo, B.-C. 2002, RevMexAA Ser. Conf., 12, 27
870: 
871: \bibitem[Kurtz et al.(1994)]{Kurtz1994}
872: 	Kurtz, S., Churchwell, E., \& Wood, D. 1994, ApJS, 91, 659
873: 
874: \bibitem[Kurtz et al.(1999)]{Kurtz1999}
875: 	Kurtz, S.E., Watson, A.M., Hofner, P., Otte, Birgit, 1999, ApJ, 514, 232
876: 
877: \bibitem[Kurtz(2000)]{Kurtz2000}
878:         Kurtz, S. 2000, RevMexAA Ser. Conf., 9, 169
879: 
880: \bibitem[Kurtz(2005)]{Kurtz2005}
881:         Kurtz, S. 2005, in Proc. 227th Symp. IAU., {\sl Massive Star Birth:
882:         A Crossroads of Astrophysics}, eds: Cesaroni, R., Felli, M., Churchwell, E., \&
883:         Walmsley, M., Cambridge Univ. Press
884: 
885: \bibitem[Kurtz \& Franco(2002)]{KurtzFranco2002}
886: 	Kurtz, S., \& Franco, J., 2002, RevMexAA Ser. Conf., 12, 16
887: 
888: \bibitem[Lugo et al.(2004]{Lugo2004}
889: 	Lugo, J., Lizano, S., \& Garay, G., 2004, ApJ, 614, 807
890: 
891: \bibitem[Mehringer(1994)]{Mehringer1994}
892: 	Mehringer, D. 1994, ApJS, 91, 713
893: 
894: \bibitem[Mezger \& Henderson(1967)]{MezgerHenderson1967}
895: 	Mezger, P.G., \& Henderson, A.P. 1967, ApJ, 147, 471
896: 
897: \bibitem[Olivero \& Longbothum(1977)]{Olivero1977}
898: 	Olivero, J.J. \& Longbothum, R.L., 1977, JQSRT, 17 233
899: 	
900: \bibitem[Parker(1958)]{Parker1958}
901: 	Parker, E.N., 1958, ApJ, 128, 664	
902: 
903: \bibitem[Pratap et al.(1992)]{Pratap1992}
904: 	Pratap, P., Snyder, L.E., \& Batrla, W. 1992, ApJ, 387, 241
905: 
906: \bibitem[Rudolph et al.(1990)]{Rudolph1990}
907: 	Rudolph, A., Welch, W.J., Palmer, P., \& Dubrulle, B. 1990, ApJ, 363, 528
908: 	
909: \bibitem[Sewilo et al.(2004)]{Sewilo2004}
910: 	Sewilo, M., Churchwell, E., Kurtz, S., Goss, W.M., \& Hofner, P. 2004, ApJ, 605, 285
911: 
912: \bibitem[Schilke et al.(1990)]{Schilke1990}
913: 	Schilke, P., Mauersberger, R., Walmsley, C.M., \& Wilson, T.J. 1990, AA, 227, 220
914: 
915: \bibitem[Simpson(1973a)]{Simpson1973a}
916: 	Simpson, J.P. 1973a, ApSS, 20, 187
917: 
918: \bibitem[Simpson(1973b)]{Simpson1973b}
919: 	Simpson, J.P. 1973b, PASP, 85, 479
920: 
921: %\bibitem[Sollins et al.(2004)]{Sollins2004} %W51
922: %	Sollins, P., Zhang, Q., \& Ho, P.T.P. 2004, ApJ, 606, 943
923: 	
924: \bibitem[Sollins \& Ho(2005)]{SollinsHo2005} %bad G10.6-0.04
925: 	Sollins, P., \& Ho, P.T.P. 2005, ApJ, 630, 987
926: 
927: %\bibitem[Sollins et al.(2005a)]{Sollins2005a} %G10.6-0.04 high resolution
928: %	Sollins, P., Zhang, Q., Keto, E., \& Ho, P.T.P. 2005a, ApJ, 624, 49
929: 
930: \bibitem[Sollins et al.(2005b)]{Sollins2005b} %G28.20-0.05
931: 	Sollins, P., Zhang, Q., Keto, E., \&  Ho, P.T.P 2005b, ApJ, 631, 399
932: 
933: \bibitem[Smirnov et al.(1984)]{Smirnov1984}
934: 	Smirnov, G.T., Sorochenko, R.L., \& Pankonin, V. 1984, AA, 135, 116
935: 
936: \bibitem[Su et al.(2004)]{Su2004}
937: 	Su, Y.-N., Liu, S.-Y., Lim, J., Ohashi, N., Beuther, H., Zhang, Q., 
938:         Sollins, P., Hunter, T., Sridharan, T.K., \& Zhao, J.-H. 2004, ApJ, 616, L39
939: 
940: \bibitem[Testi et al.(2000)]{Testi2000}
941:         Testi, L., Hofner, P., Kurtz, S., \& Rupen, M. 2000, AA, 359, L5
942: 
943: \bibitem[Tieftrunk et al.(1997)]{Tieftrunk1997}
944: 	Tieftrunk, A. R., Gaume, R. A., Claussen, M. J., Wilson, T. L., \&
945: 	Johnston, K. J. 1997, AA, 318, 931
946: 
947: \bibitem[Turner \& Matthews(1984)]{TurnerMatthews1984}
948: 	Turner, B.E., \& Matthews, H.E. 1984, ApJ, 277, 164
949: 
950: \bibitem[Ulrich(1976)]{Ulrich1976}
951:         Ulrich, R. 1976, ApJ, 210, 377
952: 
953: \bibitem[Walsh et al.(1998)]{Walsh1998}
954:          Walsh, A.J., Burton, M.G., Hyland, A.R., \& Robinson, G. 1998, MNRAS, 301, 640
955:          
956: \bibitem[Welch \& Marr(1987)]{WelchMarr1987}
957:  	Welch, Wm.J., \& Marr, J., 1987, ApJL, 317, L21
958: 
959: %\bibitem[Wilson et al.(1983)]{Wilson1983}
960: %	Wilson, T.L., Mauersberger, R., Walmsley, C.M., \& Batrla, W., 1983, AA, 127, L19
961: 
962: 
963: \bibitem[Wilson et al.(2003)]{Wilson2003}
964: 	Wilson, T.L., Boboltz, D.A., Gaume, R.A., \& Megeath, S.T.,
965: 	2003, ApJ, 597, 434
966: 
967: \bibitem[Wood \& Churchwell(1989)]{WoodChurchwell1989}
968:         Wood, D.O.S., Wood, \& Churchwell, E. 1989, ApJS, 69, 831
969: 
970: %\bibitem[Young et al.(1998)]{Young1998}
971: %	Young, L.M., Keto, E., \& Ho, P.T.P. 1998, ApJ, 1998, 507
972: 
973: %\bibitem[Zhang \& Ho(1997)]{ZhangHo1997}
974: %	Zhang, Q., \& Ho, P.T.P 1997, ApJ, 488, 241
975: 
976: \bibitem[Zhang et al.(1998)]{ZhangHoOhashi1998}
977: 	Zhang, Q., Ho, P.T.P., \& Ohashi, H. 1998, ApJ, 494, 636
978: 
979: \bibitem[Zijlstra et al.(1990)]{Zijlstra1990} 
980:         Zijlstra, A.A., Pottasch, S.R., Engels, D., Roelfsema, P.R., 
981:         te Lintel Hekkert, P., \& Umana, G.  1990, MNRAS, 246, 217
982: 
983: \end{thebibliography}
984: 
985: 
986: 
987: \clearpage
988: 
989: \begin{deluxetable}{lrrccccclc}
990: \tabletypesize{\scriptsize}
991: \tablecolumns{10}
992: \tablewidth{0pt}
993: \tablecaption{Calibration of Millimeter Observations \label{millimetercalibration}}
994: \tablehead{
995: &   \multicolumn{2}{c}{Flux} &   
996: \colhead{} & \multicolumn{2}{c}{Bandpass}  
997: & \colhead{} & \multicolumn{2}{c}{Gain} \\
998: \cline{3-4} \cline{6-7} \cline{9-10}\\
999: \colhead{Source}  & Band & \colhead{Calibrator}   & \colhead{Flux}  &  & \colhead{Calibrator} & \colhead{Flux} & &
1000:  \colhead{Calibrator}   & \colhead{Flux}     \\
1001:  & & & \colhead{Jy}   & & & \colhead{Jy}  & & & \colhead{Jy} 
1002: } 
1003: \startdata
1004: G10.60 -0.4       & SMA 	& Ceres	& 0.7 	&& 3C454.3	& 16.8	&& PKS 1911-201	& 1.4 \\
1005: G28.20 - 0.04 N  & SMA	& \nodata& \nodata&& 3C454.3	& 14.3	&& PKS 1911-201	& 1.8 \\
1006: G45.07 + 00.14    & SMA	& \nodata& \nodata&& 3C454.3	& 4.6	&& PKS 2025+337	& 0.5 \\
1007: W51e2        	& SMA	& Ceres	& 0.7 	&& 3C454.3	& 16.0	&& PKS 2025+337	& 0.8 \\
1008: NGC7538-IRS1 	& SMA	& Ceres	& 0.5 	&& 3C454.3	& 11.9	&& PKS 2202+422	& 3.0 \\
1009: \enddata
1010: \tablecomments{Flux densities derived assuming a 
1011: flux density for Uranus of 36.6 Jy. }
1012: \end{deluxetable}
1013: 
1014: % VLA OBSERVATIONS
1015: 
1016: \begin{deluxetable}{lrrccccccl}
1017: \tabletypesize{\scriptsize}
1018: %\rotate
1019: \tablecaption{VLA Observations  \label{VLAobs}}
1020: \tablewidth{0pt}
1021: \tablehead{
1022: \colhead{Source} &  \colhead{RA} & \colhead{Dec} & \colhead{Line} & \colhead{Frequency} & \colhead{Spectral} &
1023: \colhead{Band} &
1024: \colhead{Angular} & \colhead{Cont.\tablenotemark{a}} & \colhead{Date} \\
1025: %
1026: &  &  & & & \colhead{Resolution} & \colhead{width} & \colhead{Resolution} & \colhead{rms} & \colhead{Observed} \\
1027: %
1028: & & & &\colhead{GHz} & \colhead{kms$^{-1}$} & Chan. & \colhead{arc second} & \colhead{mJy} 
1029: %
1030: }
1031: \startdata
1032: G10.6 -0.4   & 18 10 28.7  & -19 55 49 & H92$\alpha$ & 8.309383 & 1.76  & 64 & $1.5 \times 1.0$ & 7   & 1989-03-17\\
1033:  & & & H66$\alpha$\tablenotemark{b} & 22.36417 & 2.62  & 64 & $1.7 \times 1.0$ & 9   & 1988-05-23\\
1034:  & & & H53$\alpha$ & 42.95197 & 1.36  & 64 & $1.5 \times 1.3$ & 3.0 & 2003-02-03\\
1035: G28.20 - 0.04 N & 18 42 58.17  & -04 13 57.0 & H53$\alpha$ & 42.95197 & 1.36  & 64 & $0.6 \times 0.4$ & 1.2 & 2005-06-24\\
1036: & & & H53$\alpha$ & 42.95197 & 5.45  & 32 & $0.7 \times 0.5$ & 3.0 & 2005-06-28\\
1037: G45.07+0.14 N& 19 13 22.069 &  10 50 52.5 
1038:                      & H76$\alpha$\tablenotemark{c}   & 14.69000 & 15.95 & 16 & $0.6 \times 0.6$ & 10  & 1984-01-07\\
1039: W51e2    & 19 23 43.913 &  14 30 14.7             & H66$\alpha$ & 22.36417 & 2.62  & 64 & $0.3 \times 0.2$ & 1.6 & 2003-02-01\\
1040: & & & H53$\alpha$ & 42.95197 & 1.36  & 64 & $0.5 \times 0.4$ & 0.9 & 2003-02-01\\
1041: NGC7538-IRS1  & 23 13 45.37  &  61 28 10.4       & H53$\alpha$ & 42.95197 & 5.45  & 31 & $0.2 \times 0.2$ & 0.9 & 2005-04-17\\
1042: \enddata
1043: \tablenotetext{a}{Bandwidth = channels $\times$ spectral resolution}
1044: \tablenotetext{b}{Data from the VLA archives. The original observation is reported in \citet{Keto2002a}.}
1045: \tablenotetext{c}{Data from the VLA archives. The original observation is reported in \citet{Garay1986}.}
1046: \end{deluxetable}
1047: 
1048: \begin{deluxetable}{lccccccccc}
1049: \tabletypesize{\scriptsize}
1050: \tablecolumns{10}
1051: \tablewidth{0pt}
1052: \tablecaption{Calibration of Centimeter Observations \label{centimetercalibration}}
1053: \tablehead{
1054: &   \multicolumn{2}{c}{Flux} &   
1055: \colhead{} & \multicolumn{2}{c}{Bandpass}  
1056: & \colhead{} & \multicolumn{2}{c}{Gain} \\
1057: \cline{3-4} \cline{6-7} \cline{9-10}\\
1058: \colhead{Source}  & Band & \colhead{Calibrator}   & \colhead{Flux}  &  & \colhead{Calibrator} & \colhead{Flux} & &
1059:  \colhead{Calibrator}   & \colhead{Flux}     \\
1060:  & & & \colhead{Jy}   & & & \colhead{Jy}  & & & \colhead{Jy} 
1061: } 
1062: \startdata
1063: G28.20 - 0.04 N 	& Q	& 3C286	& 1.5 	&& 3C273  	& 24.2	&& PKS 1851+005	& 1.9 \\
1064: G45.07 + 00.14 N & U	& 3C286	& 3.5 	&& 3C454.3	& 7.5	&& PKS 1923+210	& 2.1 \\
1065: W51e2        		& Q	& 3C286	& 1.5 	&& 3C273  	& 10.1	&& PKS 1923+210	& 2.6 \\
1066: W51e2        		& K	& 3C286	& 2.5 	&& 3C273  	& 19.5	&& PKS 1923+210	& 3.1 \\
1067: NGC7538-IRS1 		& Q	& 3C48	& 0.5 	&& 3C454.3	& 15.6	&& PKS 2250+558	& 0.8 \\
1068: \enddata
1069: \end{deluxetable}
1070: 
1071: 
1072: 
1073: \begin{deluxetable}{lccccccccc}
1074: \tabletypesize{\scriptsize}
1075: \tablecaption{Recombination Line data \label{CMrrl}}
1076: \tablewidth{0pt}
1077: \tablehead{
1078: \colhead{Source}  & \colhead{Continuum} & 
1079: \colhead{Integrated} & \colhead{Quantum} &
1080: \colhead{Line} & \colhead{$1 \sigma$} & \colhead{Line} & \colhead{$1 \sigma$} & \colhead{Line} & \colhead{$1 \sigma$} \\
1081: &   \colhead{Peak} & 
1082: \colhead{Intensity} & \colhead{Number} &
1083: \colhead{Peak} & & \colhead{Velocity\tablenotemark{a}} & & \colhead{Width\tablenotemark{b}} & \\
1084: &   \colhead{Jy/beam} & \colhead{Jy} & &
1085: \colhead{mJy/beam} & \colhead{mJy/beam} & \colhead{kms$^{-1}$} &  \colhead{kms$^{-1}$} & 
1086: \colhead{kms$^{-1}$} &\colhead{kms$^{-1}$}
1087: }
1088: \startdata 
1089: G10.6-0.4  & 0.017 & 3.14 & 92& 8.1 & 0.2  & 0.6 & 0.5  & 37.5 & 1.2\\
1090: & 0.04 \tablenotemark{c}  & 4.57 & 66& 188.0 & 5.0 & 2.1 & 0.005  & 35.1 & 0.005 \\
1091: &  2.1  & 6.0  & 53& 219.0 & 4.9 & 2.2 & 0.005  & 31.8 & 0.01 \\
1092: &  1.17  & 2.41 & 30 & 1040.0 & 28.5 & 3.4 & 0.3 & 24.5 & 0.8 \\
1093: G28.20-0.04 N& 0.4  & 1.1  & 53& 99.0 & 3.6 & 88.7 & 0.6 & 33.4 & 1.5 \\
1094:  & 0.56  & 0.72 & 30 &930.0 & 22.0  & 92.5 & 0.2 & 20.9 & 0.6 \\
1095: G45.07+0.14 & 0.37 & 0.77 & 76& 21.0 & 2.5 & 45.7 & 3.4 & 56.9 & 8.1\\
1096: & 0.08  & 0.17 & 30 & 92.0 & 10.2 & 49.4 & 1.8 &33.2 & 4.2 \\
1097: W51e2              & 0.15 & 0.30 & 66& 14.2 & 0.4 & 53.8 & 0.7 & 50.9 & 1.6 \\
1098:  & 0.17 & 2.3  & 53& 42.0 & 1.1 & 59.5 & 0.3 & 32.5 & 1.2 \\
1099:  & 1.67  & 3.66 & 30 &870.0 & 51.0 & 59.7 & 0.8 & 26.8 & 1.9\\
1100: NGC7538-IRS1	  & 0.2  & 0.7  & 53& 23.3 & 1.1 & -41.8 & 1.0 & 61.0 & 4.1 \\
1101:  & 2.56  & 2.79 & 30 & 4450.0 & 58.0 & -60.5 & 0.4 & 57.3 & 0.8 \\
1102: \enddata
1103: \tablenotetext{a}{ {Line} center velocity from Gaussian fit.}
1104: \tablenotetext{b}{FWHM from Gaussian fit.}
1105: \tablenotetext{c}{Keto 2002}
1106: 
1107: \end{deluxetable}
1108: 
1109: 
1110: % ELECTRON DENSITIES
1111: 
1112: \begin{deluxetable}{lcccccc}
1113: \tabletypesize{\scriptsize}
1114: %\rotate
1115: \tablecaption{Electron Densities from line widths \label{electrondensities}}
1116: \tablewidth{0pt}
1117: \tablehead{
1118: \colhead{Source} & \colhead{Quantum Number} & \colhead{Line Width} &
1119: \colhead{$1 \sigma$} & \colhead{$n_e$} & \colhead{$1 \sigma$}  \\
1120: &   &    \colhead{kms$^{-1}$} & \colhead{kms$^{-1}$} & \colhead{cm$^{-3}$} & \colhead{cm$^{-3}$} }
1121: \startdata
1122: G10.6-0.4       & 92    & 37.5                  & 1.2   & $9.1\times 10^4$ & $7.3\times 10^3$ \\
1123:                 & 66    & 35.1\tablenotemark{a} & 0.005 & $7.8\times 10^5$ & $3.2\times 10^2$ \\
1124:                 & 53    & 31.8                  & 0.01  & $2.5\times 10^6$ & $3.2\times 10^3$ \\
1125: G28.20-0.04 N   & 92    & 74.4\tablenotemark{b} & 2.6   & $2.9\times 10^5$ & $1.2\times 10^4$   \\
1126:                 & 53    & 33.4                  & 1.5   & $4.1\times 10^6$ & $4.3\times 10^5$   \\
1127: G45.07+0.14     & 76    & 56.9\tablenotemark{d} & 8.1   & $6.1\times 10^5$ & $1.8\times 10^5$   \\
1128:                 & 66    & 42.3\tablenotemark{e} & 2.3   & $6.9\times 10^5$ & $1.6\times 10^5$   \\
1129: W51e2           & 66    & 50.9                  & 1.6   & $1.6\times 10^6$ & $9.0\times 10^4$   \\
1130:                 & 53    & 32.5                  & 1.2   & $2.0\times 10^6$ & $4.0\times 10^5$  \\
1131: NGC7538-IRS1    & 66    & 180\tablenotemark{c}  & ...   & $1.0\times 10^7$ & ...        \\
1132:                 & 53    & 61.0                  & 4.1   & $1.4\times 10^6$ & $1.5\times 10^6$   \\
1133: \enddata
1134: \tablecomments{Electron density, $n_e$,
1135: required to produce the observed increase in the width of the
1136: cm RRL over the mm RRL by pressure broadening. }
1137: \tablenotetext{a}{Keto 2002}
1138: \tablenotetext{b}{Sewilo et al 2004}
1139: \tablenotetext{c}{Gaume et al 1995a}
1140: \tablenotetext{d}{Garay et al 1986. The line width measured in our analysis of this data is larger than reported
1141: in the original reference (48.1 kms$^{-1}$).}
1142: \tablenotetext{e}{Garay et al 1985}
1143: \end{deluxetable}
1144: 
1145: 
1146: \begin{deluxetable}{lcccl}
1147: \tabletypesize{\scriptsize}
1148: %\rotate
1149: \tablecaption{Integrated Intensities from Previous Observations\label{fluxdata}}
1150: \tablewidth{0pt}
1151: \tablehead{
1152: \colhead{Source} & \colhead{Frequency} & \colhead{Flux} & \colhead{Beam} & \colhead{Reference} \\
1153: 		&  \colhead{GHz} 	& \colhead{Jy} 	& \colhead{Arc seconds} }
1154: \startdata
1155: G10.6-0.4	& 23 	& 2.4 		& $0.1 $ 	  & \citet{SollinsHo2005} \\
1156: 	& 22.3	& 4.1 		& $0.2\times 0.1$ & \citet{KetoWood2006} \\
1157: 	& 15 	& 2.8 		& $0.3 $	  & \citet{HoHaschick1981} \\
1158: 	& 15 	& 2.6 		& $0.3\times0.2 $ & \citet{TurnerMatthews1984} \\
1159: 	& 8.3	& 3.1 		& $1.5\times 1.0$ & this paper \\
1160: 	& 5 	& 1.7 		& $1 $ 		  & \citet{HoHaschick1981} \\
1161: G28.20-0.04 N	& 23.7  & 0.98 		& $0.3\times 0.2$ & \citet{Sollins2005b}	\\
1162: 	& 15    & 0.54 		& $0.5$		  & \citet{Kurtz1994}	\\
1163: 	& 8.4   & 0.30 		& $0.9$		  & \citet{Kurtz1994}	\\
1164: 	& 8.3   & 0.30 		& $4.9\times 2.7$ & \citet{Sewilo2004}	\\
1165: G45.07+0.14	& 345   & 2.71 		& $3.0\times 4.0$ & \citet{Su2004}	\\
1166: 	&  98   & 1.04 		& $2.6\times 2.4$ & \citet{Hunter1997}	\\
1167: %W51e2			& 230	& 1.53 & $1.0$ & SMA		\\
1168: W51e2			& 147	& 3.7  		& $2.0$ 	   & \citet{ZhangHoOhashi1998}\\
1169: 		& 88.2  & 1.23 		& $6 \times 5$     & \citet{Rudolph1990}\\
1170: 			& 23.7  & 0.35 		& $5 \times 4$     & \citet{Rudolph1990}\\
1171: 			& 14.7  & 0.26 		& $4.1 \times 3.9$ & \citet{Garay1985}\\
1172: 			& 8.3   & 0.07 		& $2.4 $ 	  & \citet{Mehringer1994}\\
1173: 			& 5.0   & $\leq 0.03$ 	& $4.1 $ 	  & \citet{Mehringer1994}\\
1174: 			& 5.0   & $0.025$ 	& $0.5 $ 	  & \citet{Rudolph1990}\\
1175: \enddata
1176: \end{deluxetable}
1177: 
1178: \begin{deluxetable}{lccccccc}
1179: \tabletypesize{\scriptsize}
1180: \tablecolumns{8}
1181: \tablewidth{0pt}
1182: \tablecaption{Model HII regions for continuum SEDs\label{ModelSED}}
1183: \tablehead{
1184: &  &  \multicolumn{2}{c}{Constant Density} &   \colhead{} &
1185: \multicolumn{3}{c}{Density Gradient} \\
1186: \cline{3-4} \cline{6-8} \\
1187: \colhead{Source} & \colhead{Distance}   & \colhead{radius}    & \colhead{density} & \colhead{} &
1188:  \colhead{radius}   & \colhead{density}    & \colhead{exponent} \\
1189:  & \colhead{kpc}   & \colhead{pc}    & \colhead{cm$^{-3}$} & \colhead{} &
1190:  \colhead{pc}   & \colhead{cm$^{-3}$}   } 
1191: \startdata
1192: G010.6 - 0.4 & 6.0	& 0.05	& $6.0\times 10^4$	&& 0.05	& $3.5\times 10^4$ & -1.4	\\
1193: G28.20 - 0.04 N & 9.1	& 0.03	& $9.0\times 10^4$	&& 0.043	& $1.0\times 10^4$ & -2.5\\
1194: G045.07 + 00.14 & 9.7	& 0.025	& $1.6\times 10^5$	&& 0.05	& $1.5\times 10^4$ & -1.7	\\
1195: W51e2\tablenotemark{a} 	& 7.0	& 0.01	& $3.0\times 10^5$	&& 0.01	& $1.0\times 10^5$ & -2.5	\\
1196: NGC7538-IRS1\tablenotemark{b} 	& 3.5	& 0.02	& $2.7\times 10^6$	&& 0.08	& $2.7\times 10^6$ & -2.0 \\
1197: NGC7538-IRS1\tablenotemark{c} 	& \nodata& 0.08	& $5.0\times 10^4$	&& \nodata	& \nodata & \nodata \\
1198: \enddata
1199: \tablenotetext{a}{SED also includes a contribution from
1200: dust modeled as 4500 M$_\odot$ of cold, dusty gas with a 
1201: temperature of 100 K and a dust emissivity, $\beta=1.5$ from the 
1202: model of \citet{Rudolph1990}.}
1203: \tablenotetext{b}{Constant density model with 3 components, 
1204: 2 HII regions plus dust  from the model of \citet{Pratap1992}.
1205: The first HII component SED also includes a contribution from
1206: dust modeled as 53 M$_\odot$ of cold, dusty gas with a 
1207: temperature of 50 K and a dust emissivity, $\beta=1.0$}
1208: \tablenotetext{c}{Second component of constant density model. }
1209: \end{deluxetable}
1210: 
1211: 
1212: \clearpage
1213: 
1214: \begin{figure}
1215: %\includegraphics[angle=90,width=4.truein]{w51spectra_new}
1216: \includegraphics[angle=90,width=4.truein]{figure1}
1217: \caption{ H66$\alpha$ (dashed blue line), H53$\alpha$ (dotted green line),
1218: and H30$\alpha$ (solid red line) in
1219: W51e2. The flux densities of the H66$\alpha$ line and the H53$\alpha$ lines
1220: have been multiplied by 80 and 27, respectively.}
1221: \label{fig:w51spectra}
1222: \end{figure}
1223: 
1224: \begin{figure}
1225: %\includegraphics[angle=90,width=4.truein]{g28nspectra_new}
1226: \includegraphics[angle=90,width=4.truein]{figure2}
1227: \caption{H53$\alpha$ (dashed blue line) and H30$\alpha$ (solid red line) in
1228: G28.20-0.04N. 
1229: The flux density of the H53$\alpha$ line has been multiplied by 10.}
1230: \label{fig:g28nspectra}
1231: \end{figure}
1232: 
1233: \begin{figure}
1234: %\includegraphics[angle=90,width=4.truein]{n7538spectra_new}
1235: \includegraphics[angle=90,width=4.truein]{figure3}
1236: \caption{ H30$\alpha$ (solid red line) and H66$\alpha$ (dashed blue line)
1237: in NGC7538-IRS1. The flux density of the H66$\alpha$ has been multiplied by
1238: 190.}
1239: \label{fig:n7538spectra}
1240: \end{figure}
1241: 
1242: \begin{figure}
1243: %\includegraphics[angle=90,width=4.truein]{g45spectra_new}
1244: \includegraphics[angle=90,width=4.truein]{f4}
1245: \caption{ H76$\alpha$ (dashed blue line) and H30$\alpha$ (solid red line) in
1246: G45.07+0.13. The flux density of the H76$\alpha$ line has been multiplied by 5.}
1247: \label{fig:g45spectra}
1248: \end{figure}
1249: 
1250: \begin{figure}
1251: %\includegraphics[angle=90,width=4.truein]{g106spectra_new}
1252: \includegraphics[angle=90,width=4.truein]{f5}
1253: \caption{ H92$\alpha$ 
1254: (long-dashed green line), 
1255: H66$\alpha$ 
1256: (short-dashed blue line), 
1257: H53$\alpha$ (dotted black line),
1258: and H30$\alpha$ (solid red line) in G10.6-0.04. The flux densities 
1259: of the H92$\alpha$,
1260: H66$\alpha$, and H53$\alpha$ lines have been multiplied by 
1261: 115, 5.5, and 4.8 respectively.}
1262: \label{fig:g106spectra}
1263: \end{figure}
1264: 
1265: 
1266: \begin{figure}
1267: %\includegraphics[angle=90,width=4.truein]{ngc7538_SED}
1268: \includegraphics[angle=90,width=4.truein]{f6}
1269: \caption{SED NGC 7538 IRS1. Data are from table 1 of \citet{Pratap1992} with
1270: the addition of our 230 GHz SMA observation. The blue lines show a 3 component model
1271: proposed by \citet{Pratap1992} consisting of two HII regions plus dust 
1272: (dotted line). 
1273: The dash-dotted line (of the smaller, higher density component)
1274: is shifted upward and to the right of the dashed line (of the larger, lower
1275: density component. The wavy solid blue line shows the sum of the three
1276: components.
1277: The red solid line that crosses smoothly through the waves of
1278: the three component model shows the SED of a single HII 
1279: region with a power-law density gradient.  See \S \ref{derivation}
1280: and table 7.}
1281: \label{fig:sedn7538}
1282: \end{figure}
1283: 
1284: 
1285: \begin{figure}
1286: %\includegraphics[angle=90,width=4.truein]{w51_SED}
1287: \includegraphics[angle=90,width=4.truein]{f7}
1288: \caption{
1289: Measured flux densities of the W51e2 HII region (crosses). The blue dashed
1290: lines show
1291: a two-component model proposed by \citet{Rudolph1990} consisting of a 
1292: uniform density HII region plus dust (dotted line).
1293: The solid red line is the SED predicted for an HII region with
1294: a density gradient ($n_e\propto r^{-1.5}$) and no significant dust emission.
1295: See \S\ref{derivation} and table 7.
1296: }
1297: \label{fig:sedw51e2}
1298: \end{figure}
1299: 
1300: 
1301: \begin{figure}
1302: %\includegraphics[angle=90,width=4.truein]{g106_new}
1303: \includegraphics[angle=90,width=4.truein]{f8}
1304: \caption{Measured flux densities of the G10.6-0.4 HII region (crosses).
1305: The blue dashed line is a model based on a uniform density HII region;
1306: the red line is a model based on an HII region with a density gradient.}
1307: \label{fig:sedG106}
1308: \end{figure}
1309: 
1310: \begin{figure}
1311: %\includegraphics[angle=90,width=4.truein]{g45_SED}
1312: \includegraphics[angle=90,width=4.truein]{f9}
1313: \caption{ Continuum spectral energy distribution for G45.07+00.14.  
1314: The crosses mark the observations; the blue dashed line
1315:   is a model based on a uniform density HII region; the red line is a
1316:   model based on an HII region with a density gradient.}
1317: \label{fig:sedG45}
1318: \end{figure}
1319: 
1320: \begin{figure}
1321: %\includegraphics[angle=90,width=4.truein]{g28_SED}
1322: \includegraphics[angle=90,width=4.truein]{f10}
1323: \caption{ Continuum spectral energy distribution for G28.20-0.04.
1324:   The crosses mark the observations; the blue dashed line is a model based
1325:   on a uniform density HII region; the red line is a model based on
1326:   an HII region with a density gradient.}
1327: \label{fig:sedG28}
1328: \end{figure}
1329: 
1330: 
1331: \clearpage
1332: 
1333: \end{document}
1334: 
1335: