1: %Tom has modified this version
2: %\documentclass[preprint2]{/home1/anish/tex/sty/apj/aastex52/aastex}
3: \documentclass[preprint]{aastex}
4:
5: \newcommand\OUCHII{UCH\,{\sc ii}}
6: \newcommand\UCHII{UCH}
7: \newcommand\HII{H\,{\sc ii}}
8: \newcommand\CII{C\,{\sc ii}}
9: \newcommand\HI{H\,{\sc i}}
10: \newcommand\Hi{H110$\alpha$}
11: \newcommand\He{He110$\alpha$}
12: \newcommand\Ca{C110$\alpha$}
13: \newcommand\kms{km~s$^{-1}$}
14: \newcommand\cmt{cm$^{-2}$}
15: \newcommand\CeO{C$^{18}$O}
16: \newcommand\tCO{$^{12}$CO}
17: \newcommand\thCO{$^{13}$CO}
18: \newcommand\NHthree{NH$_3$}
19: \newcommand\muG{$\mu$G}
20: \newcommand\Msun{M$_{\odot}~$}
21: \newcommand\Lsun{L$_{\odot}~$}
22: \newcommand\cmthree{cm$^{-3}~$}
23: \newcommand\cmsix{cm$^{-6}~$}
24: \newcommand\dycm{dyne cm$^{-2}~$}
25: \newcommand\etal{et al.}
26: \newcommand\be{\begin{equation}}
27: \newcommand\ee{\end{equation}}
28: \newcommand\bea{\begin{eqnarray}}
29: \newcommand\eea{\end{eqnarray}}
30: \newcommand\La{Ly${\alpha}$~}
31: \newcommand\ddeg{$^{o}$}
32:
33: %
34: % Define \lsim and \gsim per J. S.'s fmt:
35: %
36: \catcode`\@=11
37: \newcommand{\gsim}{${\mathrel{\mathpalette\@versim>}}$}
38: \newcommand{\lsim}{${\mathrel{\mathpalette\@versim<}}$}
39: \newcommand{\@versim}[2]{\lower 2.9truept \vbox{\baselineskip 0pt \lineskip
40: 0.5truept \ialign{$\m@th#1\hfil##\hfil$\crcr#2\crcr\sim\crcr}}}
41: \catcode`\@=12
42:
43: \newcommand\bbeta{\boldmath$\beta$}
44:
45:
46: %\slugcomment{Draft 4}
47: \shorttitle{Carbon recombination lines toward \OUCHII\ regions}
48: \shortauthors{D. Anish Roshi et al.}
49:
50:
51: \begin{document}
52:
53: \title{An 8.5 GHz Arecibo survey of Carbon Recombination Lines
54: toward Ultra-compact \HII\ regions: Physical properties of dense molecular material}
55:
56: \author{D. Anish Roshi}
57: \affil{Raman Research Institute, Bangalore, India 560 080 and \\
58: National Radio Astronomy Observatory\altaffilmark{1}, Green Bank, WV 24944, USA}
59: \email{anish@rri.res.in}
60:
61: \author{Dana. S. Balser}
62: \affil{National Radio Astronomy Observatory, Green Bank, WV 24944, USA }
63: \email{dbalser@nrao.edu}
64:
65: \author{T. M. Bania}
66: \affil{Institute for Astrophysical Research, 725 Commonwealth Avenue, \\
67: Boston University, Boston, MA 02215, USA.}
68: \email{bania@bu.edu}
69:
70: \author{W. M. Goss}
71: \affil{National Radio Astronomy Observatory, Socorro, NM 87801, USA }
72: \email{mgoss@nrao.edu}
73:
74: \and
75:
76: \author{C. G. De Pree}
77: \affil{Department of Physics and Astronomy, Agnes Scott College,
78: 141 East College Avenue, Decatur, GA 30030. }
79: \email{cdepree@agnesscott.edu}
80:
81:
82: \altaffiltext{1}{The National Radio Astronomy Observatory is a
83: facility of the National Science Foundation, operated
84: under a cooperative agreement by Associated
85: Universities, Inc.}
86:
87:
88: \begin{abstract}
89: We report here on a survey of carbon recombination lines (RLs) near 8.5 GHz toward
90: 17 ultra-compact \HII\ regions (\UCHII s). Carbon RLs are detected in 11 directions,
91: indicating the presence of dense photodissociation regions (PDRs) associated
92: with the \UCHII s. In this paper, we show that the carbon RLs provide
93: important, complementary information on the kinematics and physical properties
94: of the ambient medium near \UCHII s. Non-LTE models for
95: the carbon line forming region are developed, assuming that the PDRs
96: surround the \UCHII s, and we constrained the model parameters
97: by multi-frequency RL data. Modeling shows that
98: carbon RL emission near 8.5 GHz is dominated by stimulated emission and
99: hence we preferentially observe the PDR material that is in front of the \UCHII\
100: continuum. We find that the relative motion between ionized gas and the
101: associated PDR is about half that estimated earlier, and has
102: an RMS velocity difference of 3.3 \kms.
103: %+/- 2.9 km/sec.
104: Our models also give estimates for the PDR density and pressure.
105: We found that the neutral density of PDRs is typically
106: $>$ 5 $\times$ 10$^5$ \cmthree
107: and \UCHII s can be embedded in regions with high ambient pressure.
108: Our results are consistent with a pressure confined \HII\ region model
109: where the stars are moving relative to the cloud core. Other models
110: cannot be ruled out, however. Interestingly,
111: in most cases, the PDR pressure is an order of magnitude larger
112: than the pressure of the ionized gas. Further investigation is
113: needed to understand this large pressure difference.
114: \end{abstract}
115:
116: \keywords{
117: ISM:\HII\ regions --- ISM: general --- line:formation --- radio lines:ISM ---
118: stars:formation --- surveys
119: }
120:
121: \section{Introduction}
122:
123: Observations of molecular lines toward \UCHII s
124: have shown that the natal clouds harboring the \HII\ regions
125: have densities $>$ 10$^5$ \cmthree\ and temperatures
126: between 100 -- 200 K (see \nocite{c02}Churchwell 2002).
127: The presence of such dense clouds has other
128: observable effects. In particular, far-ultraviolet (FUV) photons
129: (6.0 -- 13.6 eV) from OB stars should produce photo-dissociation regions
130: (PDRs) in the neutral material close to the \UCHII\
131: (see review by \nocite{ht97}Hollenbach \& Tielens 1997).
132: Gas phase carbon in these regions is ionized by photons
133: in the energy range 11.3 -- 13.6 eV.
134: The physical conditions in these PDRs are ideally suited for
135: producing observable radio recombination lines (RLs)
136: of carbon (\nocite{nwt94}Natta, Walmsley \& Tielens 1994;
137: \nocite{jetal05}Jeyakumar \etal\ 2005).
138:
139: To date, most objects that have been searched for carbon
140: RL emission are extended \HII\ regions. These \HII\ regions are
141: often density bound and the molecular densities near them are low.
142: Thus, not surprisingly, these searches have detected carbon lines
143: only toward a few extended \HII\ regions. \UCHII s form in or near
144: dense molecular material and hence they appear to be ideal targets
145: for detecting carbon RLs. A handful of recent VLA
146: observations was successful in detecting
147: carbon RLs toward \UCHII s. For example, observations
148: toward W48A and W49 have detected, respectively, the 76$\alpha$ (14697.314 MHz)
149: and 92$\alpha$ (8313.528 MHz) transitions of carbon
150: (\nocite{rgaj05}Roshi \etal\ 2005,
151: \nocite{detal05}De Pree \etal\ 2005).
152: The angular resolutions of these observations
153: (5\arcsec ~and 2\arcsec ~respectively) were sufficient to infer that
154: the line emission is associated with the \UCHII s.
155: Moreover, the detected RLs have velocities similar to that of
156: the molecular clouds harboring the \UCHII s, implying that
157: the line forming region is embedded in the dense material
158: near the \HII\ regions.
159:
160: Multi-frequency carbon RL observations from PDRs associated with
161: \UCHII s can be used: (1) to estimate the physical
162: properties of the dense molecular material
163: (\nocite{jetal05}Jeyakumar \etal\ 2005) and
164: (2) to test whether \UCHII s are pressure
165: confined (\nocite{rgaj05}Roshi \etal\ 2005).
166: With these aims, we made a survey of carbon RLs near 8.5 GHz with the Arecibo
167: telescope\footnote{Arecibo Observatory is part of the National Astronomy
168: and Ionosphere Center (NAIC), which is operated by Cornell University
169: under a cooperative agreement with the National Science Foundation.}
170: toward 17 \UCHII s. The source selection was made such that at least one source
171: from each of the six morphological types suggested by
172: \nocite{wc89b}Wood \& Churchwell (1989b) and
173: \nocite{kcw94}Kurtz, Churchwell \& Wood (1994) for \UCHII s
174: is observed in the survey
175: (see Table~\ref{tab1}). Another criterion used for selection was that
176: the continuum flux densities at 2 cm of the \UCHII s are $>$ 150 mJy
177: (\nocite{wc89b}Wood \& Churchwell 1989b; \nocite{kcw94}Kurtz \etal\ 1994).
178: %The details of the observations,
179: %data analysis and modeling of the line forming region
180: %are presented in this paper.
181: In \S\ref{sec:obs} we discuss the observations and data analysis procedure.
182: The results of the survey are given in \S\ref{sec:result}.
183: In \S\ref{sec:model} we present the details of modeling of the carbon RL
184: emission toward a selected subset of \UCHII s from our sample.
185: Our conclusions are given in \S\ref{sec:con}.
186:
187: \section{Observations and Data Analysis}
188: \label{sec:obs}
189:
190: The observations were made at frequencies near 8.5 GHz with the Arecibo Telescope
191: on May 30 through June 4 2003. At the observed frequency, the telescope
192: has a beam-width (FWHM) of 28\arcsec. The wide bandwidth of the 8 to 10 GHz
193: receiver system of the telescope was used to simultaneously observe
194: four RL transitions (89$\alpha$, 90$\alpha$, 91$\alpha$ and 92$\alpha$)
195: of carbon, hydrogen and helium from two polarizations. In addition to
196: these lines, the H113$\beta$ line (8878.730 MHz) was present in the
197: frequency band of the spectrometer used for observing the 90$\alpha$ RL
198: transitions. The auto-correlation spectrometer was configured in the four sub-band mode
199: to observe the four RL transitions. The transitions were selected such that the
200: frequency range was free of radio frequency interference (RFI).
201: %Moreover, these transitions are located in the frequency range
202: %where the telescope gain is maximum.
203: % Arecibo's maximum gain is at 430 MHz.....
204: We used 25 MHz bandwidth for each sub-band, which corresponds to a velocity
205: range of $\sim$ 840 \kms. This velocity range was adequate
206: to observe RLs of carbon, helium and hydrogen in each sub-band. We selected 9 bit
207: sampling in the correlator to minimize effects due to RFI and to maximize the
208: sensitivity of the observations. In the chosen configuration, the correlator provides
209: 1024 spectral channels for each sub-band and for each polarization.
210: Before observing the target sources, the setup was tested by observing
211: the `extended' \HII\ region G43.17+0.0 (\nocite{l89}Lockman 1989).
212:
213: Our observations were one of the first scientific
214: uses of the 8 -- 10 GHz system and hence there were several factors
215: affecting telescope performance. Some of these factors are: (1)
216: an inaccurate telescope pointing model and (2) irregularities of the
217: primary surface panel settings. We, moreover,
218: had to use a `gain curve' which is poorly sampled in azimuth and
219: elevation. The gain curve essentially provides the telescope gain
220: (i.e. antenna temperature in K per Jy) for different azimuths
221: and zenith angles. This curve is needed to calibrate the
222: observed spectra in Jy, since the gain of the Arecibo telescope
223: is a function of azimuth and zenith angle. The gain curve is
224: derived from observations toward a set of calibrator sources.
225: We had to use data toward calibrator sources observed during
226: the period 01 March to 19 July, 2003 to derive the gain curve.
227: The number of calibrators observed in this
228: period was not large enough to provide good sampling of the
229: telescope gain over azimuth and zenith angle.
230: (Unfortunately the receiver system was modified a few weeks after our
231: observations so we could not improve upon this gain curve.)
232: During the 15 months subsequent to our observations the
233: performance of the 8--10 GHz system, the data quality
234: and calibration procedure have all been significantly improved.
235: Due to the performance limitations during our observations, however,
236: we were forced to devise several analysis procedures in order
237: to minimize the calibration errors. In addition to the software provided
238: by Arecibo observatory, we developed programs in IDL to implement
239: these procedures.
240:
241: Since the telescope pointing model was not accurate we observed
242: calibrator sources B1843+098, B1857+129 and B2018+295 at intervals
243: of about 45 minutes. We made azimuth--zenith angle continuum cross
244: scans in order to measure pointing offsets. These offsets were used
245: to correct the telescope pointing before observing the target sources.
246: We found that the pointing offset was as large as 16\arcsec\
247: ($\sim$ 0.6 $\times$ FWHM beam width) in some cases. Because the position could,
248: of course, not be corrected during observations pointing errors have
249: resulted in an additional uncertainty in the amplitude calibration of our spectra.
250: We estimate below that the RL amplitudes reported here have an error of 20\%.
251:
252: Careful bandpass calibration is important for
253: our observations since the RL flux density is
254: less than a percent of the system equivalent flux density.
255: To do this we observed a
256: target source for 5 minutes, then switched
257: to an off-source position and acquired data for the
258: same amount of time. The off-source position was selected
259: such that it followed the same azimuth--zenith angle track
260: as that of the target source during the integration.
261: After each on- and off-source measurement pair,
262: spectra were taken with a noise calibration signal, turned on
263: and off for 30 sec, respectively. These data were used
264: to derive the system temperature.
265:
266: We configured the spectrometer to have an integration time of 1 second.
267: This short averaging time was selected in case the spectra
268: were affected by RFI. Examination of the data showed that the four
269: bands were not affected by any RFI. The spectrometer provides the
270: Fourier transforms of the measured auto-correlation function after
271: applying uniform weighting. The spectral resolution achieved is about
272: 1 \kms\ which is adequate to resolve the carbon RLs.
273:
274: The noise calibration scans and the gain curve were used to calibrate the
275: spectrum in units of Jy. We noticed that the total power in the 1 second
276: integrations was not stable for some of the sub-bands. Therefore we visually
277: examined the calibrated spectra for relatively high RMS and discrepant spectral
278: baselines and excised these data. Less than 1\%\ of the data were excised.
279:
280: The velocity resolution of the sub-bands were not the same because
281: they are centered at different sky frequencies. For the carbon RLs
282: observed here the velocity resolution can change by as much as 10\%.
283: Moreover the spectra of the four sub-bands are not sampled at the
284: same velocities. Since the spectra were obtained by Fourier transforming
285: the measured autocorrelation functions with uniform weighting, the filter
286: response of the spectrum was a sinc function. We therefore resampled the
287: spectra using sinc interpolation in order to have an identical velocity
288: sampling for all sub-bands. We referenced all sub-bands to the velocity
289: scale and resolution of the spectrum corresponding to the 89$\alpha$
290: transition. A demonstration of our resampling process is
291: given in Fig.~\ref{fig0}, where the average spectrum obtained toward G61.48+0.09
292: and the resampled average spectra of the four sub-bands are shown.
293: The velocity range is restricted to show the carbon RL, which has the smallest
294: line width (4.7 \kms) and hence is affected most by any error in the resampling
295: processes. The good agreement between the carbon RL profiles proves the
296: efficacy of the resampling process. The four resampled spectra of a scan
297: were then averaged to produce a single calibrated spectrum
298: for each polarization.
299:
300: We examined the amplitude of the hydrogen RLs in the individual on/off
301: pairs for each source. For some
302: sources the hydrogen RL amplitude changed during the integration.
303: Since the RL intensity is not expected to change, this variation in line
304: amplitude is due to pointing errors and/or errors in the estimated antenna
305: gain. To minimize the error in the final spectrum and to maximize the
306: signal-to-noise ratio, we weighted the spectrum of each scan by the ratio of
307: the hydrogen RL amplitude to the spectral variance before averaging.
308: The variance was estimated using spectral values in the velocity range
309: where RLs were not present and after removing a residual baseline.
310: During the observations we choose to apply a Doppler correction for each sub-band
311: separately and hence the spectra corresponding to different scans could be
312: averaged without any further velocity shift. The frequency offset used for each
313: sub-band was based on the rest frequency of the carbon RL observed in that sub-band.
314:
315: The spectral baselines of the final calibrated spectra were fit with
316: a third-order (or less) polynomial model using only the line-free
317: channels. The continuum flux density was determined to be the mean
318: value of this baseline model. The spectral line profiles were modeled
319: using a minimum number of Gaussian components that were able to fit
320: the hydrogen, helium, carbon, and H133$\beta$ lines without any
321: constraints on the model parameters. In one object there was evidence
322: for a weak heavy ion (heavier than C$^+$ ion) RL in the residuals.
323: The Gaussian parameters of
324: this profile were obtained by keeping the model parameters of the
325: other four RLs constant.
326:
327: The error in the line flux density in the final average spectrum can be
328: estimated by comparing the hydrogen RL amplitude in the final spectrum
329: with that obtained from individual scans. The comparison showed that
330: the line flux density was within 15--18\% of the maximum RL amplitude
331: observed for a source but clearly biased toward lower
332: values. However, the spectrum with maximum amplitude itself may not
333: have the correct flux density in all cases, since sampling of the
334: gain curve is non-optimal. So we further compared the continuum
335: flux densities of two sources (G32.80+0.19 and G70.29+1.60) with those
336: obtained from existing 3.6 cm interferometric images (see Table~\ref{tab3}).
337: The continuum flux densities obtained from our observations (see Table~\ref{tab2})
338: were measured with respect to
339: the off-source position used for bandpass calibration. The two sources
340: were selected based on negligible sky contribution to the continuum emission
341: at the off-source position, as inferred from existing continuum surveys
342: (\nocite{retal90}Reich \etal\ 1990; \nocite{aetal79}Altenhoff \etal\ 1979).
343: Neglecting the small wavelength difference, the flux densities obtained
344: from our observations are within 10\% of those estimated from interferometric
345: data. Based on this comparison and the lower line flux densities in the
346: final spectra compared to the observed maximum values, we estimate a
347: $1\sigma$ error of 20\% for all RL flux densities obtained from our survey data.
348:
349: \section{Results}
350: \label{sec:result}
351:
352: Table~\ref{tab1} gives the source names, morphological type and J2000 coordinates
353: of the 17 \UCHII s observed in the survey.
354: The effective integration time given in Table~\ref{tab1} is eight times the
355: actual on-source observing time, since the final spectra are obtained by averaging
356: all the four RL transitions from two polarizations. The calibrated spectra
357: toward the observed sources are shown in Fig.~\ref{fig1}a and \ref{fig1}b.
358: Table~\ref{tab2} gives the line parameters obtained from Gaussian modeling
359: of the final spectra together with the $1\sigma$ error on these parameters
360: provided by the non-linear least square fitting routine. The RMS is obtained
361: from the final spectrum using values in the velocity range free of any RL
362: emission. The continuum flux densities of the observed sources, which are
363: measured with respect to the off-source position used for band-shape measurement,
364: are also listed. This flux density is the mean value of the polynomial baseline
365: model removed from the final calibrated spectra (see \S\ref{sec:obs}). Note that
366: these values, in all cases, are not the flux densities of the observed sources
367: since continuum emission may be present in the off source position.
368:
369: As mentioned above, we observed the \HII\ region G43.17+0.0 to test the
370: observing setup. We have detected hydrogen, helium, carbon and H113$\beta$
371: lines toward G43.17+0.0 indicating that our observing procedures were correct.
372: The coordinates of the \HII\ region and line parameters are given in
373: Table~\ref{tab1}. The final spectrum obtained toward G43.17+0.0 is
374: shown in Fig.~\ref{fig1}b. Since G43.17+0.0 is an `extended' \HII\
375: region we will not discuss this source further.
376:
377: \subsection{Carbon recombination lines}
378: \label{sec:crl}
379:
380: Carbon RLs were detected toward 11 \UCHII s (65\% detection rate) out of the
381: 17 observed sources. The detection of carbon RLs toward a large number of
382: \UCHII s indicates that a majority of these \HII\ regions have associated
383: dense PDRs. The sources where carbon RLs were not detected probably also
384: have associated dense PDRs; we believe that the non-detections are due to
385: sensitivity limitations. The ratio of carbon to hydrogen RLs obtained
386: from the 11 detections has a mean value of 0.11 (1$\sigma$ error 0.09).
387: The expected carbon line flux density estimated using this ratio and
388: the detected hydrogen line flux density is less than the RMS noise in
389: the spectrum of the sources where carbon RLs were not detected.
390: Thus we conclude that our data are consistent with the hypothesis that all
391: \UCHII s in our survey have associated dense PDR material.
392:
393: Based on the continuum morphology of \UCHII s,
394: \nocite{wc89b}Wood \& Churchwell (1989b) and \nocite{kcw94}Kurtz \etal\ (1994)
395: have classified these regions into 7 types:
396: unresolved/spherical, cometary, core-halo, Gaussian,
397: shell, multiple peak and
398: irregular. \UCHII s observed in this survey are selected from
399: these different morphological types (see Table~\ref{tab1}). Carbon
400: lines have been detected toward \UCHII s belonging to all
401: types except for the irregular morphologies. Observations toward a
402: larger sample of \UCHII s are, however, needed to establish a
403: definite relationship between the formation of PDRs and morphology of
404: \UCHII s.
405:
406: The observed width of the carbon RLs ranges between 3 and 14 \kms,
407: with a median value of 6.3 \kms. The expected thermal line width
408: from these regions is $<$ 1.9 \kms ~(corresponding to a gas temperature
409: of 1000 K; \nocite{nwt94}Natta \etal\ 1994), indicating that
410: non-thermal motions in the PDR dominate.
411: Earlier observations of the NH$_3(1,1)$ transition toward \UCHII s showed
412: that their line widths are typically between 1.5 -- 4 \kms, with a median
413: value of $\sim$ 3 \kms\ (\nocite{cwc90}Churchwell,
414: Walmsley \& Cesaroni 1990). Thus the
415: median width of carbon RLs is a factor of two larger than that of
416: NH$_3(1,1)$. The line widths in both cases are dominated by non-thermal
417: motion. Higher angular resolution observations are needed to understand the
418: line width difference and the relationship between the molecular and carbon
419: line forming regions. The different line widths may indicate that the two
420: lines originate from different regions. Carbon lines may
421: originate from regions closer to the boundary of the \UCHII , as expected
422: in a molecular cloud--\HII\ region interface.
423:
424: Fig.~\ref{fig2} shows the relationship between the continuum
425: and carbon RL flux densities observed toward \UCHII s. Carbon RL intensity
426: clearly correlates well with the continuum emission. At 8.5 GHz,
427: continuum emission originates from the ionized gas associated with the \UCHII\
428: while carbon RLs originate from PDR gas outside the \HII\ region.
429: Thus the correlation between the two observed quantities indicates stimulated
430: emission of carbon lines by the background continuum radiation. This correlation
431: confirms the result by \nocite{nwt94}Natta \etal\ (1994) that,
432: near 8.5 GHz, carbon RL emission is dominated by stimulated
433: emission (see also \S\ref{sec:model}).
434:
435: The dominance of stimulated emission for carbon lines implies that the
436: RLs near 8.5 GHz are detected from the PDR on the near side
437: of the \UCHII . Further, modeling of RL emission shows that
438: the line intensity from the PDR at the
439: far side would typically be a factor of five weaker than that
440: originating from the front and hence is not detectable in the present
441: observations (see \S\ref{sec:model}). We used this information
442: to study the relative motion between the PDRs and \UCHII s
443: (\nocite{rgaj05}Roshi \etal\ 2005).
444: The observed central velocity
445: of the hydrogen line represents the mean velocity of the ionized gas
446: with respect to the observer. Therefore the difference between the
447: central velocities of the carbon and hydrogen lines provides a measure
448: of the relative motion of the PDR with respect to the ionized
449: region. The velocity difference ranges between $-$5.2 and +4.3 \kms,
450: with an RMS value of 3.3 \kms. Note that the
451: mean error of the relative velocity is only 0.4 \kms. Also, the RMS
452: velocity difference between the hydrogen and carbon RLs
453: is 3 times larger than that of hydrogen and helium RLs
454: (see \S\ref{sec:hy}). Earlier
455: studies of the relative motion between the ionized gas in compact
456: \HII\ regions and the molecular gas associated with them have shown
457: velocity differences ranging between $-$12 and +8 \kms\
458: (\nocite{fetal90}Forster \etal\ 1990). Our observations show a
459: velocity difference that is about half of that inferred from
460: molecular line data.
461:
462:
463: \subsection{Hydrogen, Helium, H113\bbeta\ and Heavy ion recombination lines}
464: \label{sec:hy}
465:
466: Hydrogen RLs were detected toward all the observed sources. The line profiles
467: can in most cases be modeled by a single component Gaussian. For 4 sources
468: (G34.26+0.15, G37.87$-$0.40, G45.12+0.13 and G70.29+1.60), the observed
469: line profile is better approximated by a two component Gaussian (see Table~\ref{tab2}).
470: Line width ranges between 23.5 and 69.7 \kms. The broad ($>$ 50 \kms)
471: hydrogen lines show non-Gaussian profiles. Such broad lines have been observed earlier
472: toward \UCHII s (\nocite{jm99}Jaffe \& Martin-Pintado 1999,
473: \nocite{detal04}De Pree \etal\ 2004). Helium and H113$\beta$
474: RLs were detected toward sources where the RMS noise in the spectrum
475: is less than about 10\% of the peak hydrogen line flux density.
476: G60.88$-$0.13 is an exception -- the H113$\beta$ and carbon line were
477: detected toward this source but no helium line was detected. A heavy ion
478: RL was detected toward the source G61.48+0.09.
479:
480: Since helium and hydrogen RLs originate from the same ionized gas,
481: no difference in their central velocities is expected.
482: The relative velocity ranges between $-$1.8 and +2.3 \kms\
483: with a RMS velocity difference of 1.1 \kms. Since the mean error of the
484: relative velocities is 0.6 \kms\ we conclude that this RMS value
485: is not significant. (Note that when multiple hydrogen RL components are present
486: we used the mean velocity of hydrogen RL components
487: to compute the relative velocity.) The velocity differences
488: between the hydrogen and H113$\beta$ RLs show a mean value of
489: 3.6 \kms. The origin of this velocity difference is probably due to the fact
490: that the H113$\beta$ line is blended with the H129$\gamma$ line. We have
491: modeled noiseless Gaussian profiles assuming that the H113$\beta$
492: transition is 2.5 times more intense than the H129$\gamma$ line, a shift
493: in LSR velocity of 15 \kms\ between the two line centers (based on the
494: known frequencies), and that both lines have a FWHM of 25 \kms. A
495: single Gaussian component model fit to the blended profile produces a
496: shift in velocity of 3.67 \kms, consistent with our results.
497:
498:
499: %The origin of this velocity difference
500: %could not be definitively traced, but could result from a combination
501: %of residual baseline structure and blending of the H113$\beta$ line
502: %with that of the H129$\gamma$ transition.
503:
504: \section{Physical properties of the PDRs}
505: \label{sec:model}
506:
507: The physical properties of the PDR can be determined by modeling
508: multi-frequency carbon line data (\nocite{nwt94}Natta \etal\ 1994;
509: \nocite{jetal05}Jeyakumar \etal\ 2005). The observed carbon line intensity
510: is a function of the background radiation intensity since the carbon
511: line emission is dominated by stimulated emission (\nocite{nwt94}Natta \etal\ 1994).
512: For the present case, the background radiation is due to the thermal
513: continuum emission from the \UCHII s. Therefore, for modeling,
514: we carefully selected a subset of \UCHII s from our sample
515: whose properties could be estimated with data available in the
516: literature. Table~\ref{tab3} summarizes the parameters of the
517: \UCHII s selected for modeling.
518: Interferometric images of the selected sources near 3.6 cm wavelength
519: with visibilities measured at angular scale comparable to the
520: Arecibo beam are available (\nocite{aetal96}Afflerbach \etal\ 1996;
521: \nocite{kcw94}Kurtz \etal\ 1994). Continuum flux
522: density and angular size of the \UCHII s are obtained from these images.
523: Note that for sources where both compact and `extended' emission are present
524: the angular size is roughly obtained from the half-power contour
525: of the extended emission. Thus these source sizes are different from the
526: values given by \nocite{aetal96}Afflerbach \etal\ (1996) in their paper.
527: \nocite{aetal96}Afflerbach \etal\ (1996) have estimated
528: the electron temperature, $T_e$, of the ionized gas in most of the selected \UCHII s.
529: We assumed a temperature of 10$^4$ K for the source G43.24$-$0.05 since no
530: temperature estimate is available. Using these temperatures and angular sizes,
531: we derive the emission measures, $EM$s, of the \UCHII s that are consistent
532: with the observed continuum flux density at 3.6 cm. We assume that the ionized
533: region has cylindrical geometry with length equal to the diameter
534: to derive the $EM$. (The correction factor
535: given by \nocite{mh67}Mezger \& Henderson (1967) that needs to
536: be used while obtaining the $EM$ for such geometry
537: is not applied since for most sources the flux density distribution cannot be
538: approximated as a simple Gaussian.)
539: These $EM$s along with angular sizes and distances to the sources
540: are used to estimate the electron density, $n_e$.
541:
542:
543: The PDR parameters that determine the carbon line intensity are
544: gas temperature, $T_{PDR}$, electron density, $n_e^{PDR}$, and PDR thickness along
545: the line of sight, $l$. We considered homogeneous slabs of carbon line forming region
546: in front and back of the \UCHII s to estimate the RL emission. The line
547: brightness temperature, $T_{LB}$, due to the slab in the front side is given by
548: (\nocite{s75}Shaver 1975)
549: \begin{eqnarray}
550: T_{LB} & = & T_{0bg,\nu} e^{-\tau_{C\nu}}(e^{-\tau_{L\nu}} - 1)\nonumber \\
551: & & \mbox{} + T_{PDR}\left( \frac{b_m \tau_{L\nu}^* + \tau_{C\nu}}{\tau_{L\nu} + \tau_{C\nu}}
552: (1 - e^{-(\tau_{L\nu} + \tau_{C\nu})}) - (1 - e^{-\tau_{C\nu}})\right)
553: \label{eq:line}
554: \end{eqnarray}
555: where $T_{0bg,\nu}$ is the background radiation temperature and
556: $\tau_{C\nu}$ is the continuum optical depth of the PDR. The non-LTE line
557: optical depth of the spectral transition from energy state $m$ to $n$,
558: $\tau_{L\nu}$, is $\tau_{L\nu} = b_n \beta_n \tau_{L\nu}^*$
559: where $\tau_{L\nu}^*$ is the LTE line optical depth,
560: $b_n$ and $\beta_n$ are the departure coefficients of state $n$. For the
561: PDR, $\tau_{L\nu}^*\; \propto\; n_e^{PDR} n_{C^+} l$, where $n_{C^+}$ is the carbon
562: ion volume number density in the PDR. For the present calculations we assumed
563: that $n_e^{PDR} = n_{C^+} = n_e$ so $\tau_{L\nu}^*\; \propto\; n_e^2 l$.
564: Thus, for a homogeneous PDR, the electron density provided by the modeling will
565: be the true electron density. However, for an inhomogeneous PDR,
566: the actual electron density depends on just how clumpy the PDR is and
567: the model value will be the square root of the second moment of the
568: spatial distribution of electron density.
569: The background temperature is given by
570: \begin{equation}
571: T_{0bg,\nu} = T_e (1 - e^{-\tau_{C\nu}^{UCH}}) \; ,
572: \end{equation}
573: where $\tau_{C\nu}^{UCH}$ is the continuum optical depth of the \UCHII . The line
574: temperature from the PDR on the far (back) side is obtained from Eq.~\ref{eq:line}
575: by setting $T_{0bg,\nu} = 0$. The line brightness temperature is
576: finally converted to flux density using the source angular size given
577: in Table~\ref{tab3}.
578:
579: Modeling starts with the computation of
580: the non-LTE departure coefficients for a given
581: electron density, gas temperature and background radiation
582: field. The coefficients are obtained using the program developed originally by
583: \nocite{bs77} Brocklehurst \& Salem (1977) and later modified by
584: \nocite{pae94} Payne, Anantharamaiah \& Erickson (1994).
585: These coefficients are then used to estimate the carbon line flux density
586: as a function of frequency. We combined our data at 8.5 GHz with the
587: existing RL data at 4.8 GHz to constrain the electron density and
588: PDR thickness for a given gas temperature. The 4.8 GHz data
589: (\nocite{aetal02}Araya \etal\ 2002; \nocite{wetal03}Watson \etal\ 2003)
590: are listed in Table~\ref{tab3}. We found that, for a given temperature,
591: the electron density and PDR thickness can be well constrained using the
592: data at these two frequencies. Particularly, the upper limit near 4.8 GHz provides
593: a stringent lower limit on the electron density in the PDR (see also \S\ref{sec:rob}).
594:
595: Typically the PDR gas temperature is in the range 300 -- 1000 K
596: (\nocite{nwt94}Natta \etal\ 1994). Table~\ref{tab4} lists the PDR electron
597: density and thickness for temperatures 300 K and 500 K. Note that for higher
598: gas temperature the estimated electron density will be larger. The carbon
599: RL flux densities predicted by the Table~\ref{tab4} models are shown in
600: Fig.~\ref{fig3} which plots for each source the model carbon flux density as a
601: function of frequency.
602: The model electron densities are converted to neutral densities using the cosmic
603: abundance of carbon (3.9 $\times$ 10$^{-4}$; \nocite{m74}Morton 1974) and a
604: depletion factor of 25\% (\nocite{nwt94}Natta \etal\ 1994). The neutral density,
605: $n_H$ (number density
606: of hydrogen atom), is in most cases greater than 5 $\times$ 10$^5$ \cmthree
607: (see Table~\ref{tab4}).
608:
609: An important inference from modeling is the dominance of stimulated emission
610: for carbon lines at frequencies \lsim\ 15 GHz. Line emission
611: from the PDR slab in front of the \UCHII\ is amplified
612: by the background continuum radiation from the ionized gas.
613: The line flux density is typically 5 times larger than that
614: from the PDR slab behind the \UCHII . Thus
615: the sensitivity of our observations
616: is sufficient to detect only carbon line emission from
617: the front of the \UCHII . As discussed in \S\ref{sec:crl},
618: this fact can be used to study the relative motion of
619: the PDR with respect to the \UCHII .
620:
621: \subsection{Pressure}
622:
623: We use the derived physical properties of the PDRs and the \UCHII s
624: to estimate the pressure in these regions.
625: The total gas pressure inside the \UCHII\ is the sum of thermal
626: pressure and turbulent pressure. Because there are no direct
627: measurements of the magnetic field strength in \UCHII s or PDRs
628: we do not include any contribution to the pressure from magnetic fields.
629: %Magnetic fields inside \HII\ regions are typically $\sim$ 10 -- 20 $\mu$G
630: %(e.g. \nocite{hc80}Heiles \& Chu 1980) and hence do not contribute
631: %significantly to the total pressure.
632: The total gas pressure is given by
633: \begin{equation}
634: P_{UCH} = 2 k n_e T_e + n_e \mu m_H v_{turb}^2 \mbox{~~dyne cm$^{-2}$},
635: \label{eq:pres}
636: \end{equation}
637: where $k$ is the Boltzmann constant, $\mu = 1.4$ is the effective mass
638: in amu of a pure H + He gas with He fraction taken as 10\% by number,
639: $m_H$ is the hydrogen mass in gm and $v_{turb}$ is
640: the turbulent velocity inside the \UCHII\ in units of cm s$^{-1}$.
641: The line profile due to turbulence is considered to be
642: Gaussian with an FWHM of $v_{turb}$. We estimated $v_{turb}$ from the
643: observed width of helium lines after removing the contribution due to
644: thermal motion. (Hydrogen lines were not used since
645: some of the \UCHII s have multiple line components.) The values for
646: electron density and temperature are taken from Table~\ref{tab3}. The
647: estimated pressures of the ionized gas in \UCHII s are typically $<$
648: 2 $\times$ 10$^{-7}$ \dycm (see Table~\ref{tab3}). Our calculations
649: show that the turbulent pressure in \UCHII s is about an order of magnitude
650: larger than the thermal pressure.
651:
652: The total gas pressure in the PDR is given by
653: \be
654: P_{PDR} = k n_H T_{PDR} + n_H \mu m_H v_{C-turb}^2 \mbox{~~\dycm},
655: \ee
656: where $v_{C-turb}$ is the turbulent velocity
657: in cm s$^{-1}$ estimated from the FWHM of the observed carbon RLs after removing
658: the thermal line width. To estimate
659: PDR pressure we consider that in the region where carbon is
660: ionized, the neutral material is predominantly in atomic hydrogen form
661: (see \nocite{ht97}Hollenbach \& Teilens 1997).
662: %The value of magnetic field strength in the PDR is not known and hence
663: %we do not include its contribution in the calculation of PDR pressure.
664: The estimated total gas pressure in the PDR is typically 10$^{-6}$ \dycm
665: (see Table~\ref{tab4}). As in the case of the \UCHII , the turbulent
666: pressure in the PDR is about one order of magnitude larger than the thermal pressure.
667: This importance of turbulent pressure was noted earlier by
668: \nocite{xetal96}Xie \etal\ (1996).
669:
670: \subsection{Robustness of the PDR parameters and Pressure estimates}
671: \label{sec:rob}
672:
673: For modeling the PDR properties we have used the carbon line parameters
674: obtained from our survey. However, as discussed in \S\ref{sec:obs},
675: the line flux densities are uncertain by up to 20\%. To study
676: the effect of this uncertainty, we
677: estimated the physical properties of the PDR by changing the
678: line flux density near 8.5 GHz by $\pm$ 20\% and
679: compared them with the values given in
680: Table~\ref{tab4}. We found that the estimated neutral densities
681: and PDR thickness change by about a factor of two and three
682: respectively. Fig.~\ref{fig4} shows the result of such a study
683: toward G32.80+0.19. The PDR parameters and pressures obtained
684: in this study are given in Table~\ref{tab5}.
685: %line flux density was increased by 20\% for $T_{PDR} = 300$ K
686: %are $n_e^{PDR} = 1000$ \cmthree, $l = $ 6.2 $\times$ 10$^{-4}$ pc
687: %and those for $T_{PDR} = 500$ K are $n_e^{PDR} = 2000$ \cmthree,
688: %$l = $ 4.5 $\times$ 10$^{-4}$ pc. When the line flux density is
689: %reduced by 20\% the parameters obtained are $n_e^{PDR} = 300$ \cmthree,
690: %$l = $ 2.6 $\times$ 10$^{-3}$ pc and $n_e^{PDR} = 400$ \cmthree,
691: %$l = $ 3.6 $\times$ 10$^{-3}$ pc for gas temperatures of 300 K
692: %and 500 K respectively.
693: We conclude that the
694: lower limits on neutral densities and pressure of the PDR given
695: in Table~\ref{tab4} will reduce only by about a factor of two
696: due to the measurement uncertainty of the carbon RL flux
697: density at 8.5 GHz.
698:
699: \section{Discussion}
700: \label{sec:con}
701:
702: A survey of recombination lines near 8.5 GHz toward 17 \UCHII s
703: using the Arecibo telescope detected carbon lines from 11 sources (65\%
704: detection rate). In this paper, we have shown that the carbon RLs provide
705: important, complementary information on the kinematics and properties
706: of the ambient medium near \UCHII s. The RLs were used
707: to study the relative motion between the carbon line forming region
708: and the \UCHII s. We obtained a RMS velocity difference of 3.3 \kms.
709: Earlier, \nocite{fetal90}Forster \etal\ (1990) compared hydrogen RL and
710: molecular line velocities toward nine compact \HII\ regions to
711: study their relative motion. They concluded that the RL velocities
712: are offset with respect to molecular lines by typically $\pm$ 6 \kms.
713: The relative velocity difference obtained from our observations is about
714: a factor of two smaller than that measured by \nocite{fetal90}Forster \etal\ (1990).
715: These authors argue that the larger velocity difference they have measured is
716: consistent with the expansion of a 10$^4$ K \HII\ region at
717: the sound speed of $\sim$ 15 \kms. The lower velocity difference
718: seen in our observations may
719: indicate that \UCHII s are not expanding at sound speeds in
720: ionized gas.
721:
722: The dynamical lifetime for \UCHII s of a few times $10^3$ years
723: was estimated assuming that the \HII\ region is
724: expanding at the speed of sound in the ionized gas.
725: However, the lifetime deduced from the observed number of \UCHII s
726: is a few times $10^5$ years.
727: This discrepancy between the two lifetimes is referred to
728: as the ``lifetime problem'' of \UCHII s
729: (\nocite{wc89a}Wood \& Churchwell 1989a). Several hypothesis
730: have been proposed to explain the long lifetime of \UCHII s
731: (see \nocite{c99}Churchwell 1999). For example,
732: \nocite{drg95}De Pree, Rodr\'iguez \& Goss (1995)
733: suggested that if high density ($\sim$ 10$^7$ \cmthree),
734: warm ($\sim$ 100 K) molecular material is present
735: in the vicinity of \UCHII s, it may be able to pressure confine \UCHII s
736: that form there and thus extend their lifetime.
737: This hypothesis is supported by
738: recent molecular line observations (eg. \nocite{ketal00}Kurtz \etal\ 2000),
739: which show that the temperature and density of the
740: molecular cloud harboring \UCHII s are $>$ 50 K and $\sim$ 10$^7$ \cmthree
741: respectively. Modeling of carbon RL data in our observations also indicates that
742: \UCHII s are embedded in neutral regions with high ambient pressure.
743:
744: In the pressure-confined nebula scenario, if the massive
745: star is stationary with respect to the molecular core, then
746: no relative motion between the ionized gas and the PDR is expected.
747: However, our carbon line observations indicate a non-zero relative motion
748: between the PDR and the ionized gas (see \S\ref{sec:crl}).
749: If the star is moving with respect to the cloud core, such relative
750: motion is expected even though the ionized gas is pressure confined.
751: The star motion can also produce the different observed
752: morphologies of \UCHII s (\nocite{gf04}Garcia-Segura \& Franco 2004).
753: However, relative motion between the PDR and ionized gas is expected
754: in other models proposed to solve
755: the long lifetime of \UCHII s and their morphology. For example, in
756: the model proposed by \nocite{kk01}Kim \& Koo (2001), the ionizing
757: star is at the edge of a dense, hot core and the `champagne flow' through the
758: hierarchical structure of the molecular cloud produces the extended
759: emission that has been observed around \UCHII s. The PDR, in this model,
760: resides in the dense, hot core. The presence of the `champagne flow' can shift
761: the centroid of the RL emission from the ionized gas relative to the
762: line emission from the PDR. This shift results in a velocity difference
763: between the hydrogen and carbon RLs from the \UCHII s in our observations.
764: Higher angular resolution observations of carbon RLs may provide
765: enough information to distinguish between these models.
766:
767: Interestingly, the PDR pressure we derive, in most cases,
768: is an order of magnitude larger than the total pressure in \UCHII s.
769: In the classical treatment of the pressure driven evolution
770: of \UCHII s, a dense shell of neutral material is formed
771: surrounding the \HII\ region
772: (\nocite{gf96}Garcia-Segura \& Franco 1996). If
773: the PDR responsible for the carbon line emission
774: resides in this neutral shell then its pressure
775: can be high. However, if ionization front is trapped within the neutral
776: material for $\sim$ 10$^5$ years, then the sound crossing
777: length scale ($\sim$ 0.2 pc; sound speed $\sim$ 2 \kms)
778: is larger than the typical molecular cloud core sizes ($\le$ 0.1 pc;
779: Churchwell 2002). Thus the neutral shell would have diffused out
780: in a fraction of the expected lifetime of the \UCHII s and hence
781: the ambient pressure outside the \UCHII\ is most likely that of the natal
782: molecular cloud. The large pressure difference, therefore, may
783: indicate that the dominant pressure inside the \UCHII\
784: is due to other physical processes such as stellar wind. Further investigation
785: with high angular resolution, multi-frequency RL and continuum observations
786: together with modeling of the evolution
787: of \HII\ regions surrounded by a dense PDR is needed to understand this
788: large pressure difference.
789:
790: \acknowledgments
791:
792: We are grateful to Phil Perillat (NAIC) for his valuable assistance
793: during observations and for his analysis software.
794: We thank Chris Salter (NAIC) for
795: his generous help in processing the calibration data
796: and the telescope operators for their support during the observations.
797:
798:
799: \begin{thebibliography}{}
800:
801: \bibitem{aetal96}
802: Afflerbach, A., \etal\ 1996, ApJS, 106, 423
803:
804: \bibitem{aetal79}
805: Altenhoff, W. J., \etal\ 1979, A\&AS, 35, 23
806:
807: \bibitem{aetal02}
808: Araya, E., Hofner, P., Churchwell, E., Kurtz, S., 2002, ApJS, 138, 63
809:
810: \bibitem{bs77}
811: Brocklehurst, M., Salem, M. 1977, Computer Phys. Commun., 13, 39
812:
813: %\bibitem{cetal94}
814: %Cesaroni, R. \etal\ 1994, A\&A, 288, 903
815:
816: \bibitem{c02}
817: Churchwell, E., 2002, ARAA, 40, 27
818:
819: \bibitem{c99}
820: Churchwell, E., 1999, The Origin of Stars and Planetary Systems,
821: Eds C. J. Lada and N. D. Kylafis, Kluwer Academic Publishers, p.515
822:
823: \bibitem{cwc90}
824: Churchwell, E., Walmsley, C. M., Cesaroni, R., 1990, A\&AS, 83, 119
825:
826: \bibitem{drg95}
827: De Pree, C. G., Rodr\'iguez, L. F., Goss, W. M., 1995, Rev. Mex. Astron. Astrofis, 31, 39
828:
829: \bibitem{detal04}
830: De Pree, C. G., \etal\ 2004, ApJ, 600, 286
831:
832: \bibitem{detal05}
833: De Pree, C. G., \etal\ 2005, in preparation.
834:
835: \bibitem{fetal03}
836: Fish, V. L., \etal\ 2003, ApJ, 587, 701
837:
838: \bibitem{fetal90}
839: Forster, J. R., \etal\ 1990, A\&A, 231, 473
840:
841: \bibitem{gf96}
842: Garcia-Segura, G., Franco, J., 1996, ApJ, 469, 171
843:
844: \bibitem{gf04}
845: Garcia-Segura, G., Franco, J., 2004, Gravitational Collapse:
846: From Massive Stars to Planets, Eds. G. García-Segura,
847: G. Tenorio-Tagle, J. Franco, H. W. Yorke, Revista Mexicana
848: de Astronomia y Astrofisica (Serie de Conferencias), Vol. 22, p.131
849:
850: %\bibitem{hc80}
851: %Heiles, C., Chu, Y. H., 1980, ApJ, 235L, 105
852:
853: %\bibitem{hetal94}Hollenbach, D., Johsstone, D., Lizano, S.,
854: %Shu, F., 1994, ApJ, 428, 654
855:
856: \bibitem{ht97}
857: Hollenbach, D. J., Tielens, A. G. G. M., 1997, ARA\&A, 35, 179
858:
859: \bibitem{jm99}
860: Jaffe, D. T., Martin-Pintado, J., 1999, ApJ, 520, 162
861:
862: \bibitem{jetal05}
863: Jeyakumar, S., \etal\ 2005, in preparation
864:
865: \bibitem{kk01}
866: Kim, K-T., Koo, B-C., 2001, ApJ, 549, 979
867:
868: \bibitem{kcw94}
869: Kurtz, S., Churchwell, E., Wood, D. O. S., 1994, ApJS, 91, 659
870:
871: \bibitem{ketal99}
872: Kurtz, S., \etal\ 1999, ApJ, 514, 232
873:
874: \bibitem{ketal00}
875: Kurtz, S., \etal\ 2000, Protostars and Planets IV, University of Arizona Press,
876: Tucson, Eds Mannings, V., Boss, A.P., Russell, S. S., p. 299
877:
878: \bibitem{l89}
879: Lockman, F. J., 1989, ApJS, 71, 469
880:
881: \bibitem{mh67}
882: Mezger, P. G., Henderson, A. P., 1967, ApJ, 147, 471
883:
884: \bibitem{m74}
885: Morton, D. C, 1974, ApJL 193, L35
886:
887: \bibitem{nwt94}
888: Natta, A., Walmsley, C. M., Tielens, A. G. G. M., 1994, ApJ, 428, 209
889:
890: \bibitem{pae94}
891: Payne H. E., Anantharamaiah K. R., Erickson W. C., 1994, ApJ, 430, 690
892:
893: \bibitem{retal90}
894: Reich, W., \etal\ 1990, 990, A\&AS, 85, 633
895:
896: \bibitem{rgaj05}
897: Roshi, D. A., Goss, W. M, Anantharamaiah, K. R., Jeyakumar, S., 2005, submitted to ApJ
898:
899: \bibitem{s75}
900: Shaver P. A., 1975, Pramana, 5, 1
901:
902: \bibitem{wetal03}
903: Watson, C., \etal\ ApJ, 2003, 587, 714
904:
905: \bibitem{wc89a}
906: Wood, D. O. S., Churchwell, E., 1989a, ApJ, 340, 265
907:
908: \bibitem{wc89b}
909: Wood, D. O. S., Churchwell, E., 1989b, ApJS, 69, 831
910:
911: \bibitem{xetal96}
912: Xie, T., Mundy, L. G., Vogel, S. N., Hofner, P., 1996, ApJ, 473, L131
913:
914: \end{thebibliography}
915:
916: \include{tab1}
917:
918: \include{tab2}
919:
920: \include{tab3}
921:
922: \include{tab4}
923:
924: \include{tab5}
925:
926: \begin{figure}
927: \plotone{f1.ps}
928: \caption{A demonstration of the spectral resampling procedure.
929: The final average spectrum obtained toward G61.48+0.09
930: is shown on the top. The four spectra plotted in the bottom panel correspond
931: to the RL transitions 89$\alpha$, 90$\alpha$, 91$\alpha$ and 92$\alpha$. They
932: are obtained by averaging the data for each sub-band separately and by resampling
933: at the same velocity resolution as that of the spectrum shown on top.
934: The velocity range is restricted to show the carbon RL (narrow line feature),
935: which has the smallest line width (4.7 \kms) and hence is affected most
936: by any error in the resampling process. The broad line feature is the
937: helium RL. The good agreement between the RL profiles shows the efficacy of the
938: resampling process.
939: \label{fig0} }
940: \end{figure}
941:
942: \begin{figure}
943: \plotone{f2a.ps}
944: \caption{Recombination line spectra toward 11 \UCHII s for which
945: carbon RLs have been detected in
946: the Arecibo survey. The spectra are obtained by averaging four
947: RL transitions (89$\alpha$, 90$\alpha$,
948: 91$\alpha$ and 92$\alpha$) near 8.5 GHz. The strongest line in the figure is the
949: hydrogen RL. The carbon, helium, H113$\beta$ and heavy ion RLs are marked
950: in the spectra where these lines are detected. The spectrum obtained
951: toward the `extended' \HII\ region G43.17+0.00 is also shown in the figure.
952: \label{fig1} }
953: \end{figure}
954:
955: \begin{figure}
956: \plotone{f2b.ps}
957: \caption{(Figure 2 continued)}
958: \end{figure}
959:
960: \begin{figure}
961: \plotone{f3.ps}
962: \caption{
963: Flux density of carbon RLs observed toward \UCHII s plotted against
964: the continuum flux density. The line
965: intensity correlates well with the continuum emission indicating the dominance
966: of stimulated emission for carbon RLs. The correlation
967: coefficient is 66\%. The error bar on carbon RL flux density is
968: the $\pm 1\sigma$ error obtained from the Gaussian modeling of line profiles.
969: For continuum flux density the error bar is $\pm$ 20\%.
970: \label{fig2} }
971: \end{figure}
972:
973: \begin{figure}
974: \plotone{f4a.ps}
975: \caption{Carbon line flux density as a function of frequency from models toward
976: a subset of \UCHII s in the Arecibo survey. The electron temperatures used for
977: modeling are 300 K (solid line) and 500 K (dashed line). The electron
978: density and PDR thickness of the models are given in Table~\ref{tab4}. The curves
979: are obtained for model parameters that produce the observed carbon line
980: flux density near 8.5 GHz
981: (marked as circle with $\pm$ 1$\sigma$ error obtained from Gaussian modeling
982: of line profiles) and are consistent wit the 3$\sigma$ upper limit of carbon
983: line flux density observed near 4.8 GHz.
984: \label{fig3} }
985: \end{figure}
986:
987: \begin{figure}
988: \plotone{f4b.ps}
989: \caption{(Figure 4 continued)}
990: \end{figure}
991:
992: \begin{figure}
993: \plotone{f5.ps}
994: \caption{
995: %The plot demonstrates the robustness of the estimated PDR parameters toward G32.80+0.19.
996: Carbon line flux density as a function of frequency
997: for PDR models toward G32.80+0.19. Model line flux density with gas temperatures
998: 300 K (left) and 500 K (right) are plotted. The solid line
999: corresponds to the electron density and PDR thickness given in Table~\ref{tab4} for
1000: G32.80+0.19. Line flux density for model parameters (see Table~\ref{tab5}) that
1001: are consistent with the upper limit at 4.8 GHz and RL flux densities of 19.4
1002: and 13.0 mJy at 8.5 GHz are also shown by the `dash' and `dash-dot' lines respectively.
1003: These RL flux densities are $\pm$ 20 \% of the value given in Table~\ref{tab4}.
1004: The carbon line flux density near 8.5 GHz is
1005: marked as circle with $\pm$ 20\% error bar.
1006: The 3$\sigma$ upper limit of carbon line flux density observed near 4.8 GHz
1007: is also shown in the plots. The lower limit on electron density reduces
1008: only by about a factor of two from that given in Table~\ref{tab4} when the carbon line
1009: flux density is reduced by 20\% (see \S\ref{sec:rob}).
1010: \label{fig4} }
1011: \end{figure}
1012:
1013: \end{document}
1014:
1015: