1:
2:
3: \documentclass{emulateapj}
4: \usepackage{lscape}
5: %%\documentclass[12pt,preprint]{aastex}
6: %%\usepackage{emulateapj5}
7:
8: %macros for commonly used symbols
9: \newcommand{\mm}{millimeter}
10: \newcommand\submm{submillimeter}
11: \newcommand\fir{far-infrared}
12: \newcommand\mir{mid-infrared}
13: \newcommand\nir{near-infrared}
14: \newcommand{\lsun}{\mbox{L$_\odot$}}% Lsun
15: \newcommand{\msun}{\mbox{M$_\odot$}}% Msun
16: \newcommand{\td}{\mbox{$T_D$}}
17: \newcommand{\mean}[1]{\mbox{$\langle#1\rangle$}} %generic mean for defined qu.
18: \newcommand{\opacity}{\mbox{$\kappa(\nu)$}} % opacity as func. of freq.
19: \newcommand{\av}{\mbox{$A_V$}} % Visual Extinction
20:
21: \shorttitle{1.1~mm Surveys of Serpens, Perseus, and Ophiuchus}
22: \shortauthors{Enoch et al.}
23:
24: \begin{document}
25:
26: \title{Comparing Star Formation on Large Scales in the c2d Legacy Clouds:
27: Bolocam 1.1~mm Dust Continuum Surveys of Serpens, Perseus, and Ophiuchus}
28:
29: \author{Melissa L. Enoch \altaffilmark{1}, Jason Glenn \altaffilmark{2}, Neal J. Evans II \altaffilmark{3},
30: Anneila I. Sargent \altaffilmark{1}, Kaisa E. Young \altaffilmark{3,4}, and Tracy L. Huard \altaffilmark{5}}
31:
32: \email{MLE: menoch@astro.caltech.edu}
33: \altaffiltext{1}{Division of Physics, Mathematics \& Astronomy, California
34: Institute of Technology, Pasadena, CA 91125}
35: \altaffiltext{2}{Center for Astrophysics and Space
36: Astronomy, 389-UCB, University of Colorado, Boulder, CO 80309}
37: \altaffiltext{3}{The University of Texas at Austin, Astronomy Department,
38: 1 University Station C1400, Austin, TX, 78712-0259}
39: \altaffiltext{4}{Department of Physical Sciences, Nicholls State University,
40: Thibodaux, Louisiana 70301}
41: \altaffiltext{5}{Harvard-Smithsonian Center for Astrophysics, 60 Garden St.,
42: Cambridge, MA 02138}
43:
44: \begin{abstract}
45: We have undertaken an unprecedentedly large 1.1~millimeter
46: continuum survey of three nearby star forming clouds using
47: Bolocam at the Caltech Submillimeter Observatory.
48: We mapped the largest areas in each cloud at millimeter or submillimeter
49: wavelengths to date: 7.5 deg$^2$ in Perseus (Paper I), 10.8
50: deg$^2$ in Ophiuchus (Paper II), and 1.5 deg$^2$ in Serpens with a
51: resolution of 31\arcsec, detecting 122, 44, and 35 cores,
52: respectively. Here we report on results of the Serpens survey and
53: compare the three clouds. Average measured angular core sizes and
54: their dependence on resolution suggest that many of the observed
55: sources are consistent with power-law density profiles. Tests of
56: the effects of cloud distance reveal that linear resolution strongly
57: affects measured source sizes and densities, but not the shape of the
58: mass distribution. Core mass distribution slopes in Perseus and
59: Ophiuchus ($\alpha=2.1\pm0.1$ and $\alpha=2.1\pm0.3$) are consistent
60: with recent measurements of the stellar IMF, whereas the Serpens
61: distribution is flatter ($\alpha=1.6\pm0.2$). We also compare the
62: relative mass distribution shapes to predictions from turbulent
63: fragmentation simulations. Dense cores constitute less than 10\% of
64: the total cloud mass in all three clouds, consistent with other
65: measurements of low star-formation efficiencies. Furthermore, most
66: cores are found at high column densities; more than 75\% of 1.1~mm
67: cores are associated with $\av \gtrsim 8$~mag in Perseus, 15~mag in
68: Serpens, and $20-23$~mag in Ophiuchus.
69: \end{abstract}
70:
71:
72: \keywords{stars: formation --- ISM: clouds --- ISM: individual
73: (Serpens, Perseus, Ophiuchus) --- submillimeter}
74:
75: \section{Introduction}
76:
77: Large-scale physical conditions in molecular clouds influence the
78: outcome of local star formation, including the stellar initial mass
79: function, star-formation efficiency, and the spatial distribution of
80: stars within clouds \citep[e.g.][]{evans99}. The physical processes
81: that provide support of molecular clouds and control the
82: fragmentation of cloud material into star-forming cores remain
83: a matter of debate. In the classical picture magnetic fields
84: provide support and collapse occurs via ambipolar diffusion
85: \citep[e.g.][]{shu78}, but many simulations now suggest that
86: turbulence dominates both support and fragmentation \citep[for a
87: review see][]{mlk04}.
88:
89: Dense prestellar and protostellar condensations, or cores \citep[for
90: definitions and an overview see][]{dif05}, provide a crucial link
91: between the global processes that control star formation on large
92: scales and the properties of newly-formed stars. The mass and spatial
93: distributions of such cores retain imprints of the fragmentation
94: process, prior to significant influence from later protostellar
95: stages such as mass ejection in outflows, core dissipation, and
96: dynamical interactions. These cold (10~K), dense ($n
97: >10^4$~cm$^{-3}$) cores are most easily observed at millimeter and
98: submillimeter wavelengths where continuum emission from cold dust
99: becomes optically thin and traces the total mass. Thus, complete
100: maps of molecular clouds at millimeter wavelengths are important for
101: addressing some of the outstanding questions in star formation.
102:
103:
104: Recent advances in millimeter and submillimeter wavelength continuum
105: detectors have enabled a number of large-scale surveys of nearby
106: molecular clouds \citep[e.g.][]{john04, kirk05, hatch05, enoch06,
107: stanke06, young06}. In addition to tracing the current and future
108: star-forming activity of the clouds on large scales, millimeter and
109: submillimeter observations are essential to understanding the
110: properties of starless cores and the envelopes of the most deeply
111: embedded protostars \citep[for more on the utility of millimeter
112: observations, see][]{enoch06}.
113:
114:
115: We have recently completed 1.1~mm surveys of Perseus
116: \citep[][hereafter Paper~I]{enoch06} and Ophiuchus \citep[][hereafter
117: Paper~II]{young06} with Bolocam at the Caltech Submillimeter
118: Observatory (CSO). In this work we present a similar 1.1~mm survey of
119: Serpens, completing our three-cloud study of nearby northern
120: star-forming molecular clouds. Unlike previous work, our surveys not
121: only cover the largest area in each cloud to date, but the uniform
122: instrumental properties allow a comprehensive comparison of the cloud
123: environments in these three regions. A comparison of the results for
124: all three clouds provides insights into global cloud conditions and
125: highlights the influence that cloud environment has on properties of
126: star forming cores.
127:
128:
129: Background facts on the Perseus and Ophiuchus molecular clouds are
130: discussed in Papers I and II. The Serpens molecular cloud is an
131: active star formation region at a distance of $d=260\pm10$~pc
132: \citep{stra96}. Although the cloud extends more then 10~deg$^2$ as
133: mapped by optical extinction \citep{cam99}, most observations of the
134: region have been focused near the main Serpens cluster at a Right
135: Ascension (R.A.) of $18^{h}30^{m}$ and declination (decl.) of
136: $1\degr15\arcmin$ (J2000).
137:
138: The Serpens cluster is a highly extincted region with a high density
139: of young stellar objects (YSOs), including a number of Class~0
140: protostars. It has been studied extensively at near-infrared,
141: far-infrared, submillimeter, and millimeter wavelengths
142: \citep[e.g.][]{ec92, hb96, lars00, davis99, ced93, ts98}. Some
143: recent work has also drawn attention to a less well known cluster to
144: the south, sometimes referred to as Serpens/G3-G6
145: \citep{djup06,harv06}. Beyond these two clusters no continuum
146: millimeter or submillimeter continuum surveys have been done that
147: could shed light on large-scale star formation processes.
148:
149: Following Papers I and II, we utilize the wide-field mapping
150: capabilities of Bolocam, a large format bolometer array at the CSO,
151: to complete millimeter continuum observations of 1.5~deg$^2$ of the
152: Serpens cloud. These observations are coordinated to cover the area
153: mapped with \textit{Spitzer} Space Telescope IRAC and MIPS
154: observations of Serpens from the ``Cores to Disks''
155: \citep[c2d][]{eva03} legacy project. While millimeter and
156: submillimeter observations are essential to understanding the
157: properties of dense prestellar cores and protostellar envelopes,
158: infrared observations are necessary to characterize the protostars
159: embedded within those envelopes. In a future paper (M. Enoch et al. 2007,
160: in preparation) we will take advantage of the overlap between our
161: 1.1~mm maps and the c2d \textit{Spitzer} Legacy maps to characterize
162: the deeply embedded and prestellar populations in Serpens.
163:
164:
165: Here the Bolocam observations and data reduction (\S~\ref{obssect}),
166: general cloud morphology and source properties (\S~\ref{ressect}),
167: and summary (\S~\ref{serpsum}) for Serpens are briefly presented. In
168: \S~\ref{compsect} we compare the millimeter survey results for
169: Serpens, Perseus, and Ophiuchus. First, we outline our operational
170: definition of a millimeter core including instrumental effects in
171: \S~\ref{coresect}, and discuss the observational biases introduced by
172: different cloud distances in \S~\ref{convsect}. Physical implications
173: of source sizes and shapes, and differences between the clouds, are
174: discussed in \S~\ref{sizecompsect}. We examine source densities and
175: compare the cloud mass versus size distributions in
176: \S~\ref{dencompsect}. We analyze the core mass distributions and
177: their relation to cloud turbulence in \S~\ref{masscompsect}, and the
178: spatial distributions of the core samples and clustering in
179: \S~\ref{clustsect}. The relationship of dense cores to the surrounding
180: cloud column density and the mass fraction in dense cores are discussed in
181: \S~\ref{threshsect} and \S~\ref{efficsect}.
182: We end with a summary in \S~\ref{sumsect}.
183:
184:
185: \section{Serpens Observations and Data Reduction}\label{obssect}
186:
187: Observations and data reduction for Serpens follow the same
188: methodology as for Perseus (Paper I) and Ophiuchus (Paper II). The
189: data reduction techniques we have developed for the molecular cloud
190: data are described in detail in Paper~I and Paper~II, including
191: removal of sky noise, construction of pointing and calibration
192: models, application of iterative mapping, and source extraction.
193: Given these previous descriptions, only details specific to Serpens
194: will be presented here. Further information about the Bolocam
195: instrument and reduction pipeline is also available in
196: \citet{laurent05}.
197:
198: \subsection{Observations}
199:
200: As for Perseus and Ophiuchus, millimeter continuum observations of
201: Serpens were made with
202: Bolocam\footnote{\url{http://www.cso.caltech.edu/bolocam}} at the CSO
203: on Mauna Kea, Hawaii. Bolocam is a 144-element bolometer array that
204: operates at millimeter wavelengths and is designed to map large fields
205: \citep{glenn03}. Observations of Serpens were carried out at a
206: wavelength of $\lambda=1.1$~mm, where the field of view is $7\farcm5$
207: and the beam is well approximated by a Gaussian with full-width at
208: half-maximum (FWHM) size of 31$\arcsec$. The instrument has a
209: bandwidth of 45 GHz at $\lambda=1.1$mm, excluding the CO (J=2-1) line
210: to approximately 99\%.
211:
212: The 1.1mm observations were designed to cover a region with $\av \ge
213: 6$~magnitudes ($\av \ge 6^m$) in the visual extinction map of
214: \citet{cam99}, shown in Figure~\ref{c2dfig}. As demonstrated in
215: Figure~\ref{c2dfig}, this ensures that the Bolocam observations
216: overlap as closely as possible with the region of Serpens observed
217: with \textit{Spitzer} IRAC and MIPS as part of the c2d Legacy project.
218: In practice, the Bolocam survey covers a region slightly larger than
219: the IRAC map, and slightly smaller than the MIPS map.
220:
221: Serpens was observed during two separate runs in 2003 May 21-June 9
222: and 2005 June 26-30. During the 2003 run, 94 of the 144 channels
223: were operational, compared to 109 during the 2005 run. Scans of
224: Serpens were made at a rate of $60\arcsec$~sec$^{-1}$ with no
225: chopping of the secondary. The final map consists of 13 scans from
226: 2003 in good weather, and 17 scans from 2005 in somewhat poorer
227: conditions. Each scan covered the entire 1.5 deg$^2$ area and took
228: $35-40$ minutes to complete depending on the scan direction. Scans
229: were made in two orthogonal directions, approximately half in R.A.
230: and half in decl. This strategy allows for good cross-linking in
231: the final map, sub-Nyquist sampling, and minimal striping from $1/f$
232: noise. Scans were observed in sets of three offset by $-43\arcsec$,
233: $0\arcsec$, $+43\arcsec$ to optimize coverage.
234:
235: In addition, small maps of pointing sources were observed
236: approximately every 2 hours, and at least one primary flux
237: calibration source, including Neptune, Uranus, and Mars, was observed
238: each night. Several larger beam maps of planets were also made
239: during each run, to characterize the Bolocam beam at 1.1~mm.
240:
241: \subsection{Pointing and Flux Calibration}\label{pointsect}
242:
243: A pointing model for Serpens was generated using two nearby pointing
244: sources, G~34.3 and the quasar 1749+096. After application of the
245: pointing model, a comparison to the literature SCUBA $850~\micron$
246: positions of four bright known sources \citep{davis99} in the main
247: Serpens cluster indicated a constant positional offset of
248: ($\delta$R.A., $\delta$decl.) = (5$\arcsec$, $-10\arcsec$). We
249: corrected for the positional offset, but estimate an uncertainty in
250: the absolute pointing of $10 \arcsec$, still small compared to the
251: beam size of 31\arcsec. The relative pointing errors, which cause
252: blurring of sources and an increase in the effective beam size, should
253: be much smaller, approximately $5 \arcsec$. Relative pointing errors
254: are characterized by the rms pointing uncertainty, derived from the
255: deviations of G~34.3 from the pointing model.
256:
257: The Bolocam flux calibration method makes real-time corrections for
258: the atmospheric attenuation and bolometer operating point using the
259: bolometer optical loading, by calculating the calibration factor as a
260: function of the bolometer DC resistence. Calibrator maps of Neptune,
261: Uranus, and G~34.3, observed at least once per night, were used to
262: construct a calibration curve for each run. A systematic uncertainty
263: of approximately 10\% is associated with the absolute flux
264: calibration, but relative fluxes should be much more accurate.
265:
266: \subsection{Cleaning and Iterative mapping}\label{cleansect}
267:
268: Aggressive sky subtraction techniques are required for Bolocam data to
269: remove sky noise, which dominates over the astronomical signal before
270: cleaning. As in Papers I and II, we remove sky noise from the Serpens
271: scans using Principal Component Analysis (PCA) cleaning
272: (\citet{laurent05} and references therein), by subtracting 3 PCA
273: components.
274:
275: As described in Paper~I, PCA cleaning removes some source flux from
276: the map as well as sky noise, necessitating the use of an iterative
277: mapping procedure to recover lost astronomical flux. In this
278: procedure, sky subtraction is refined by iteratively removing a source
279: model (derived from the cleaned map) from the raw data, thereby
280: reducing contamination of the sky template by bright sources.
281: Simulations show that at least 98\% of source flux density is
282: recovered after 5 iterations, with the exception of very large
283: ($\gtrsim 100\arcsec$ FWHM) faint sources, for which we only recover
284: 90\% or less of the true flux density (See Paper~I). We
285: conservatively estimate a 10\% residual photometric uncertainty from
286: sky subtraction after iterative mapping, in addition to the 10\%
287: uncertainty in the absolute flux calibration.
288:
289: Data from the 2003 and 2005 observing runs were iteratively mapped
290: separately because they required different pointing and flux
291: calibration models. After iterative mapping the two epochs were
292: averaged, weighted by the square root of the observational coverage.
293:
294:
295: \subsection{Source Identification}\label{idsect}
296:
297: Millimeter cores are identified in an optimally filtered map using the
298: source extraction method described in Paper~I. Because the signal
299: from a point source lies in a limited frequency band, we can use an
300: optimal (Wiener) filter in Fourier space to attenuate $1/f$ noise at
301: low frequencies, as well as high frequency noise above the signal
302: band. The optimal filter preserves the resolution of the map and the
303: peak brightness of point sources but reduces the rms noise per pixel
304: by approximately $\sqrt{3}$, thereby optimizing the source
305: signal-to-noise (S/N). Extended sources will have slightly enhanced
306: peak values in the optimally filtered map. After optimal filtering,
307: the map is trimmed to remove areas of low coverage. Note that the
308: optimally filtered map is used for source detection only; all
309: photometry is measured in the unfiltered map, and all maps displayed
310: here are unfiltered.
311:
312: Observational coverage, which depends on the scan strategy, number of
313: scans, and number of bolometer channels, was very uniform for Serpens;
314: trimming regions where the coverage was less than 30\% of the peak
315: coverage was equivalent to cutting off the noisy outer edges of the
316: map. The average coverage of the Serpens map is 1600 hits per pixel,
317: where a hit means a bolometer passed over this position, with
318: differences in hits per pixel across the map of 18\%. The average
319: coverage corresponds to an integration time of 13 minutes per pixel,
320: although individual pixels are not independent because the map is
321: over-sampled.
322:
323: For each pixel, the local rms noise is calculated in small (45~
324: arcmin$^2$) boxes using a noise map from which sources have been
325: removed. This noise map is derived as part of the iterative mapping
326: process (see Paper~I). A simple peak-finding routine identifies all
327: pixels in the optimally filtered map more than $5\sigma$ above the
328: local rms noise level, and source positions are determined using an
329: IDL centroiding routine. Each new source must be separated by at
330: least a beam size from any previously identified source centroid. The
331: centroid is a weighted average position based on the surface
332: brightness within a specified aperture, and is computed as the
333: position at which the derivatives of the partial sums of the input
334: image over (y,x) with respect to (x,y) equal zero. A given centroid
335: is considered ``well defined'' as long as the computed derivatives are
336: decreasing. All sources were additionally inspected by eye to remove
337: spurious peaks near the noisier edges of the map.
338:
339:
340:
341: \section{Serpens Results}\label{ressect}
342:
343: The final $10\arcsec$ pixel$^{-1}$ Serpens map is shown in
344: Figure~\ref{mapfig}, with the well known northern Serpens cluster,
345: Cluster~A \citep{harv06}, indicated, as well as the southern cluster,
346: Cluster B (Serpens/G3-G6). Covering a total area of 1.5 deg$^2$, or
347: 30.9~pc$^2$ at a distance of 260~pc, the map has a linear resolution
348: of $7.3\times10^3$~AU.
349:
350: We identify 35 sources above the $5\sigma$ detection limit in the
351: Serpens map. The $5\sigma$ limit is based on the local rms noise, and
352: is typically 50 mJy beam$^{-1}$. The noise map for Serpens is shown
353: in Figure~\ref{noisefig}, with the positions of identified sources
354: overlaid. The mean rms noise is 9.5 mJy~beam$^{-1}$, but is higher
355: near bright sources. Most of the 18\% variations in the local rms
356: noise occur in the main cluster region, where calculation of the noise
357: is confused by residual artifacts from bright sources. Such artifacts
358: must contribute significantly to noise fluctuations; coverage
359: variations alone would predict rms noise variations of only $\sqrt{18}
360: = 4\%$. This means that faint sources near bright regions have a
361: slightly lower chance of being detected than those in isolation.
362:
363: Source positions are listed in Table~\ref{srctab}, and identified by
364: red circles in Figure~\ref{sourcefig}. Figure~\ref{sourcefig} also
365: shows magnifications of the more densely populated source regions,
366: including Cluster A and Cluster B. We do not see any circularly
367: symmetric extended emission on scales $\gtrsim 3\arcmin$ in the map.
368: It should, in principle, be possible to recover symmetric structures
369: up to the array size of $7\farcm5$, but our simulations show that
370: sources $\gtrsim 4\arcmin$ in size are severely affected by cleaning
371: and therefore difficult to fully recover with iterative mapping. The
372: map does contain larger filamentary structures up to $8\arcmin$ long.
373: In particular, the long filament between Cluster A and Cluster B is
374: reminiscent of the elongated ridge near B1 in Perseus (Paper~I). The
375: Serpens filament does not contain the bright compact sources at either
376: end that are seen in the B1 ridge, however.
377:
378:
379: Previous millimeter-wavelength maps of Cluster A, such as the 1.1~mm
380: UKT14 map of \citet{ced93} and the $850~\micron$ SCUBA map of
381: \citet{davis99}, generally agree with our results in terms of
382: morphology and source structure. Not all of the individual
383: $850~\micron$ sources are detected by our peak finding routine,
384: presumably due to the poorer resolution of Bolocam ($30\arcsec$)
385: compared to SCUBA ($14\arcsec$), but most can, in fact, be identified
386: by eye in the Bolocam map. An IRAM 1.3~mm continuum map of Cluster B
387: with 11\arcsec\ resolution \citep{djup06} is visually quite similar to
388: our Bolocam map of the region. We detect each of the four 1.3~mm
389: sources identified (MMS1-4), although the 1.3~mm triplet MMS1 is seen
390: as a single extended source in our map.
391:
392: Most of the brightest cores, in particular those in Cluster A, are
393: associated with known YSOs including a number of Class~0 objects
394: \citep{hb96,harv06}. All bright 1.1~mm sources are aligned with
395: bright $160~\micron$ emission in the \textit{Spitzer} MIPS map of
396: Serpens observed by the c2d legacy project \citep{harv07}. Fainter
397: millimeter sources are usually associated with extended $160~\micron$
398: filaments, but do not necessarily correspond to point sources in the
399: MIPS map. Conversely, one bright extended region of $160~\micron$
400: emission just south of Cluster A contains no 1.1~mm sources. This
401: area also exhibits extended emission at 70 and 24~$\micron$, which may
402: be indicative of warmer, more diffuse material than that of the dense
403: cores detected at 1.1~mm.
404:
405: Despite the low rms noise level achieved in Serpens, very few sources
406: are seen outside the main clusters; most of the area that we mapped
407: appears devoid of 1.1~mm emission, despite being in a region of high
408: extinction. Figure~\ref{avfig} shows a comparison between the
409: Bolocam millimeter map (gray-scale) and visual extinction (contours)
410: derived from c2d near- and mid-infrared Spitzer data.
411:
412: The majority of sources (90--95\%) detected by IRAC and MIPS in the
413: c2d clouds have spectral energy distributions characteristic of
414: reddened stars. Thus we have measures of the visual extinction for
415: many lines of sight through the molecular clouds imaged by c2d.
416: Line-of-sight extinction values are derived by fitting the R$_V$=5.5
417: dust model of \citet{wd01} to the near-infrared through mid-infrared
418: SED \citep{deliv}. For each of the three clouds, the derived
419: line-of-sight extinctions were convolved with uniformly spaced
420: 90\arcsec\ Gaussian beams to construct an extinction map.
421:
422: Extinction maps used throughout this paper for Serpens, Perseus, and
423: Ophiuchus are derived from c2d data by this method. The c2d
424: extinction maps accurately trace column densities up to $\av \sim
425: 40^m$, but are relatively insensitive to small regions of high
426: volume density, because they rely on the detection of background
427: stars. Thus the extinction maps are complementary to the 1.1~mm
428: observations, which trace high volume density structures (see
429: \S~\ref{coresect}). In Figure~\ref{avfig}, the \av\ map for Serpens
430: is smoothed to an effective resolution of $2\farcm5$.
431:
432: As can be seen in Figure~\ref{avfig}, nearly all Serpens millimeter
433: sources lie within regions of high visual extinction ($\av \ge 10^m$)
434: and, in particular, all bright Bolocam sources are associated with
435: areas of $\av \ge 15^m$. Nevertheless, there are a number of high
436: extinction areas ($\av \ge 12^m$) with no detectable 1.1~mm sources.
437: A similar general trend was noted in both Perseus (Paper~I) and
438: Ophiuchus (Paper~II), with relatively few sources found outside the
439: major groups and clusters associated with the highest extinction.
440: \S~\ref{threshsect} examines the relationship between $A_V$ and 1.1~mm
441: sources in more detail.
442:
443: \subsection{Source Properties}\label{statsect}
444:
445: \subsubsection{Positions and Photometry}
446:
447: Positions, peak flux densities, and S/N for the 35 1.1~mm sources
448: identified in the Bolocam map of Serpens are listed in
449: Table~\ref{srctab}. The S/N ratio is measured in the optimally
450: filtered map, whereas photometry and all other source properties are
451: measured in the unfiltered, surface brightness normalized map. The
452: peak flux density per beam ($I_{\nu}$) is given in mJy beam$^{-1}$
453: (1~mJy beam$^{-1} = 0.04$~MJy~sr$^{-1}$). Uncertainties in
454: photometry are calculated from the local rms beam$^{-1}$, calculated
455: as in \S~\ref{idsect}. An additional systematic error of 15\% is
456: associated with all flux densities, from the absolute calibration
457: uncertainty and the systematic bias remaining after iterative mapping.
458: Table~\ref{srctab} also lists the most commonly used name from the
459: literature for known sources, and indicates if the 1.1~mm source is
460: coincident (within $60\arcsec$) with a MIPS $24~\micron$ source from
461: the c2d database \citep{harv07}.
462:
463: Table~\ref{phottab} lists photometry in fixed apertures of diameter
464: $40\arcsec$, $80\arcsec$, and $120\arcsec$, and the total integrated
465: flux density ($S_{\nu}$). Integrated flux densities are measured
466: assuming a sky value of zero, and include a correction for the
467: Gaussian beam so that a point source has the same integrated flux
468: density in all apertures. No integrated flux density is given if
469: the distance to the nearest neighboring source is smaller than the
470: aperture diameter. The total flux density is integrated in the
471: largest aperture ($30\arcsec-120\arcsec$ diameters in steps of
472: $10\arcsec$) that is smaller than the distance to the nearest
473: neighboring source. Uncertainties are $\sigma_{ap} = \sigma_{mb}
474: (\theta_{ap}/\theta_{mb})$, where $\sigma_{mb}$ is the local rms
475: beam$^{-1}$ and ($\theta_{ap}$, $\theta_{mb}$) are the aperture and
476: beam FWHM respectively.
477:
478: Peak and total flux density distributions for the 35 1.1~mm sources
479: in Serpens are shown in Figure~\ref{fluxfig}, with the $5\sigma$
480: detection limit indicated. In general, source total flux densities
481: are larger than peak flux densities because most sources in the map
482: are extended, with sizes larger than the beam. Both distributions
483: look bimodal; all of the sources in the brighter peak are in either
484: Cluster A or Cluster B. The mean peak flux density of the sample is
485: 0.5~Jy beam$^{-1}$, and the mean total flux density is 1.0~Jy, both
486: with large standard deviations of order the mean value. Peak
487: \av\ values of the cores, calculated from the peak flux density as in
488: \S~\ref{masssec}, are indicated on the upper axis.
489:
490:
491: \subsubsection{Sizes and Shapes}
492:
493: Source FWHM sizes and position angles (PA, measured east of north)
494: are measured by fitting an elliptical Gaussian after masking out
495: nearby sources using a mask radius equal to half the distance to the
496: nearest neighbor. The best fit major and minor axis sizes and PAs are
497: listed in Table~\ref{phottab}. Errors given are the formal fitting
498: errors and do not include uncertainties due to residual cleaning
499: effects, which are of order $10-15\%$ for the FWHM and $5\degr$ for
500: the PA (see Paper~I).
501:
502: As can be seen in Figure~\ref{sizefig}, the minor axis FWHM values are
503: fairly narrowly distributed around the sample mean of $49\arcsec$,
504: with a standard deviation of $12\arcsec$. The major axis FWHM have a
505: similarly narrow distribution with a mean of $63\arcsec$ and a scatter
506: of $12\arcsec$. On average sources are slightly elongated, with a
507: mean axis ratio at the half-max contour (major axis FWHM / minor axis
508: FWHM) of 1.3.
509:
510: A morphology keyword for each source is also given in
511: Table~\ref{phottab}, to describe the general source shape and
512: environment. Keywords indicate if the source is multiple (within
513: $3\arcmin$ of another source), extended (major axis full-width at
514: $2\sigma > 1\arcmin$), elongated (axis ratio at $4\sigma > 1.2$),
515: round (axis ratio at $4\sigma < 1.2$), or weak (peak flux densities
516: less than 5 times the rms per pixel in the unfiltered map). The
517: majority (28/35) of sources are multiple by this definition, and
518: nearly all (32/35) are extended at the $2\sigma$ contour.
519:
520:
521: \subsubsection{Masses, Densities, and Extinctions}\label{masssec}
522:
523: The total mass $M$ of gas and dust in a core is proportional to the
524: total flux density $S_{\nu}$, assuming the dust emission at 1.1~mm is
525: optically thin and both the dust temperature and opacity are
526: independent of position within a core:
527: \begin{equation}
528: M = \frac{d^2 S_{\nu}}{B_{\nu}(T_D) \kappa_{\nu}}, \label{masseq}
529: \end{equation}
530: where $\kappa_{1.1mm}=0.0114$~cm$^2$~g$^{-1}$ is the dust opacity per
531: gram of gas, $d=260$~pc is the distance, and $T_D=10$~K is the dust
532: temperature. A gas to dust mass ratio of 100 is included in
533: $\kappa_{1.1mm}$. The dust opacity is interpolated from
534: \citet[][Table~1 column 5]{oh94}, for dust grains with thin ice
535: mantles, coagulated for $10^5$ years at a gas density of
536: $10^6$~cm$^{-3}$. This opacity has been found to be the best fit in a
537: number of radiative transfer models \citep{evans01, shir02, young03}.
538:
539: Masses calculated in this manner, assuming a dust temperature of
540: $T_D=10$~K for all sources, are listed in Table~\ref{phottab}.
541: Uncertainties given are from the uncertainty in the total flux density
542: only; additional uncertainties from $\kappa$, $T_D$, and $d$ together
543: introduce a total uncertainty in the mass of up to a factor of 4 or
544: more (also see Paper~I). The dust opacity is uncertain by up to a
545: factor of two or more \citep{oh94}, owing to large uncertainties in
546: the assumed dust properties, and possible positional variations of
547: $\kappa_{\nu}$ within a core. Variations in both $\kappa_{\nu}$ and
548: the cloud distance $d$ have smaller effects than dust temperature
549: uncertainties for the range of plausible values
550: ($\kappa_{1.1mm}=0.005-0.02$~cm$^2$~g$^{-1}$; $d=200-300$~pc;
551: $T_D=5-30~K$).
552:
553: Assuming $T_D =10$~K is a reasonable compromise to cover both
554: prestellar and protostellar sources, based on the results of more
555: detailed radiative transfer models \citep[][see also discussion in
556: Paper I]{evans01,shir02}. A temperature of 10~K will result in
557: overestimates of the masses of protostellar cores, which will be
558: warmer on the inside, by up to a factor of three (for a 20~K source).
559: Temperature errors are not as problematic as they might seem, however,
560: as most of the envelope mass is located at large radii and low
561: temperatures (see also Paper I).
562:
563: The total mass of the 35 1.1~mm cores in Serpens is $92\msun$, which
564: is only 2.7\% of the total cloud mass (3470 \msun). The cloud mass
565: is estimated from the c2d visual extinction map using
566: $N($H$_2)/\av\ = 0.94 \times 10^{21}$~mag~cm$^{-2}$ \citep{bohlin78}
567: and
568: \begin{equation}
569: M\mathrm{(cloud)} = d^2 m_H \mu_{H_2} \Omega
570: \sum{N(\mathrm{H}_2)}\label{avmasseq}
571: \end{equation}
572: where $d$ is the distance, $\Omega$ is the solid angle, $\sum$
573: indicates summation over all $\av>2^m$ pixels in the extinction map,
574: and $ \mu_{H_2}=2.8$ is the mean molecular weight per H$_2$ molecule.
575:
576: Figure~\ref{massfn} shows the differential mass function of all 1.1~mm
577: sources in Serpens. The point source detection limit of $0.13
578: M_{\sun}$ is indicated, as well as the 50\% completeness limit for
579: sources of FWHM 55$\arcsec$ ($0.35 M_{\sun}$), which is the average
580: size of the sample. Completeness is determined from Monte Carlo
581: simulations of simulated sources inserted into the raw data and run
582: through the reduction pipeline, as described in Paper~I. The best fit
583: power law slope to $dN/dM \propto M^{-\alpha}$ is shown ($\alpha =
584: 1.6$), as well as the best fit lognormal slope.
585:
586: Our mass distribution has a flatter slope than that found by
587: \citet{ts98} from higher resolution ($5\arcsec$) OVRO observations
588: ($\alpha = -2.1$ for $M>0.35\msun$). This may be due in part to the
589: fact that most of our detections, at least 25/35, lie outside the
590: $5\farcm 5 \times 5\farcm 5$ area observed by \citet{ts98}.
591: The resolution differences of the observations may also contribute
592: significantly; for example, a number of our bright sources break down
593: into multiple objects in the $5 \arcsec$ resolution map.
594:
595: From the peak flux density $I_{\nu}$ we calculate the central H$_2$
596: column density for each source:
597: \begin{equation}
598: N(\mathrm{H}_2) = \frac{I_{\nu}}{\Omega_{mb} \mu_{H_2} m_H
599: \kappa_{\nu} B_{\nu}(T_D)}. \label{aveq}
600: \end{equation}
601: Here $\Omega_{mb}$ is the beam solid angle, $m_H$ is the mass of
602: hydrogen, $\kappa_{1.1mm} = 0.0114$~cm$^2$g$^{-1}$ is the dust opacity
603: per gram of gas, $B_{\nu}$ is the Planck function, $T_D$=10~K is the
604: dust temperature, and $\mu_{H_2}=2.8$ as above. From column density
605: we convert to extinction: $\av = N($H$_2)/ 0.94 \times
606: 10^{21}$~mag~cm$^{-2}$ \citep{bohlin78}, adopting $R_V = 3.1$. We
607: note, however, that this relation was determined for the diffuse
608: interstellar medium and may not be ideal for the highly extincted
609: lines of sight probed here.
610:
611: The resulting peak \av\ values are listed in Table~\ref{phottab}.
612: The average peak \av\ of the sample is 41$^m$ with a large standard
613: deviation of 55$^m$ and a maximum \av\ of 256$^m$. Extinctions
614: calculated from the millimeter emission are generally higher than
615: those from the c2d visual extinction map by approximately a factor of
616: 7, likely a combination of both the higher resolution of the Bolocam
617: map ($30''$ compared with $90''$) and the fact that the extinction
618: map cannot trace the highest volume densities because it relies on
619: the detection of background sources. Grain growth in dense cores
620: beyond that included in the dust opacity from \citet{oh94} could also
621: lead to an overestimate of the \av\ from out 1.1~mm data.
622:
623: Also listed in Table~\ref{phottab} is the mean particle density:
624: \begin{equation}
625: \mean{n} = \frac{M}{\frac{4}{3} \pi R^3 m_H \mu_p}, \label{deneq}
626: \end{equation}
627: where $M$ is the total mass, $R$ is the linear deconvolved half-width
628: at half-max size, and $\mu_p=2.33$ is the mean molecular weight per
629: particle. The median of the source mean densities is
630: $2.3\times10^5$~cm$^{-3}$, with values ranging from $3.1\times10^4$
631: to $6.1\times10^6$~cm$^{-3}$.
632:
633: \section{Serpens Summary}\label{serpsum}
634:
635: We have completed a 1.1~mm dust continuum survey of Serpens, covering
636: $1.5$~deg$^2$, with Bolocam at the CSO. We identify 35 1.1~mm
637: sources in Serpens above a $5\sigma$ detection limit which is
638: 50~mJy~beam$^{-1}$, or $0.13\msun$, on average. The sample has an
639: average mass of $2.6\msun$, and an average source FWHM size of
640: $55\arcsec$. On average, sources are slightly elongated with a mean
641: axis ratio at half-max of 1.3. The differential mass distribution of
642: all 35 cores is consistent with a power law of slope $\alpha = 1.6\pm
643: 0.2$ above $0.35 \msun$. The total mass in dense 1.1~mm cores in
644: Serpens is $92\msun$, accounting for 2.7\% of the total cloud mass,
645: as estimated from our c2d visual extinction map.
646:
647:
648: \section{Three-Cloud Comparison of Perseus, Ophiuchus, and Serpens}\label{compsect}
649:
650: The survey of Serpens completes a three-cloud study investigating the
651: properties of millimeter emission in nearby star-forming molecular
652: clouds: Perseus \citep{enoch06}, Ophiuchus \citep{young06}, and
653: Serpens. Having presented the results for the Serpens cloud, we now
654: compare the three clouds.
655:
656: Our large-scale millimeter surveys of Perseus, Ophiuchus, and Serpens,
657: completed with the same instrument and reduction techniques, provide
658: us with a unique basis for comparing the properties of 1.1~mm emission
659: in a variety of star-forming environments. In the following sections
660: we examine similarities and differences in the samples of
661: star-forming cores, and discuss implications for physical properties
662: of cores as well as global cloud conditions.
663:
664:
665: \subsection{What is a core?}\label{coresect}
666:
667: Before comparing results from the three clouds, we first describe our
668: operational definition of a millimeter core. The response of Bolocam
669: to extended emission, together with observed sensitivity limits
670: determine the type of structure that is detectable in our 1.1~mm maps
671: The Bolocam 1.1~mm observations presented here are sensitive to
672: sub-structures in molecular clouds with volume density $n\gtrsim 2
673: \times 10^4$~cm$^{-3}$. One way to see this is to calculate the mean
674: density along the curve defined by the detection as a function of
675: size in each cloud.
676:
677: Figure~\ref{mvscomp} demonstrates how the completeness in each cloud
678: varies as a function of source size. Plotted symbols give the total
679: mass versus linear deconvolved size for all sources detected in each
680: cloud, and lines indicate the empirically derived 50\% completeness
681: limits. For Ophiuchus the average completeness curve is plotted; as
682: the rms noise varies considerably in Ophiuchus, some regions have
683: higher or lower completeness limits than the curve shown here.
684: Completeness is determined from Monte Carlo simulations by adding
685: simulated sources to the raw data, processing them in the same way as
686: the real data, and attempting to detect them using our peak-finding
687: algorithm (see Paper~I). We are biased against detecting large
688: diffuse sources because we detect sources based on their peak flux
689: density, whereas the mass is calculated from the total flux, which
690: scales approximately as the size squared.
691:
692: Calculating mean densities along the 50\% completeness curve in each
693: cloud yields $n_{lim}\sim 3-7\times 10^4$~cm$^{-3}$ in Serpens,
694: $n_{lim}\sim 2-4\times 10^4$~cm$^{-3}$ in Perseus, and $n_{lim}\sim
695: 10-30\times 10^4$~cm$^{-3}$ in Ophiuchus. By comparison, the mean
696: cloud density as probed by the extinction map is approximately
697: 1000~cm$^{-3}$ in Serpens, 220~cm$^{-3}$ in Perseus, and 390~cm$^{-3}$
698: in Ophiuchus. To be identified as a core, therefore, individual
699: structures must have a mean density $n\gtrsim 2\times 10^4$~cm$^{-3}$,
700: and a contrast compared to the average background density of at least
701: 30-100. The mean cloud density is estimated from the total cloud mass
702: (\S~\ref{masssec}) and assumes a cloud volume of
703: $\mathrm{V}=\mathrm{A}^{1.5}$, where A is the area of the extinction
704: map within the $\av=2$ contour.
705:
706: Although we are primarily sensitive to cores with high density
707: contrast compared to the background, it is clear that there is
708: structure in the 1.1~mm map at lower contrasts as well, and that many
709: cores are embedded within lower density filaments. The total mass in
710: each of the 1.1~mm maps, calculated from the sum of all pixels
711: $>5\sigma$, is approximately twice the mass in dense cores:
712: 176\msun\ versus 92\msun\ in Serpens, 376\msun\ versus 278\msun\ in
713: Perseus, and 83\msun\ versus 44\msun\ in Ophiuchus, for ratios of
714: total 1.1~mm mass to total core mass of 1.9, 1.4, and 1.9
715: respectively. Thus about half the mass detectable at 1.1~mm is not
716: contained in dense cores, but is rather in the ``foothills'' between
717: high density cores and the lower density cloud medium.
718:
719: Structures that meet the above sensitivity criteria and are identified
720: by our peak-finding routine are considered cores. Our peak-finding
721: method will cause extended filaments to be broken up into several
722: separate ``cores'' if there are local maxima in the filament separated
723: by more than one beam size, and if each has a well-defined centroid
724: (see \S~\ref{idsect}). There is some question as to whether these
725: objects should be considered separate sources or a single extended
726: structure, but we believe that our method is more reliable for these
727: data than alternative methods such as Clumpfind \citep{will94}.
728: In Paper~I we found that faint extended sources in our maps, which one
729: would consider single if examining by eye, are often partitioned into
730: multiple sources by Clumpfind.
731: Using our method, one filamentary
732: structure in Serpens is broken up into several sources, as are two
733: filaments in Perseus.
734:
735: Monte Carlo tests were done to quantify biases and systematic errors
736: introduced by the cleaning and iterative mapping process, which affect
737: the kind of structure we can detect. Measured FWHM, axis ratios, and
738: position angles are not significantly affected by either cleaning or
739: iterative mapping for sources with FWHM$\lesssim 120''$; likewise, any
740: loss of flux for such sources has an amplitude less than that of the
741: rms noise. Sources with FWHM$\gtrsim 200''$ are detected, but with
742: reduced flux density (by up to 50\%), and large errors in the measured
743: FWHM sizes of up to a factor of two.
744:
745: The limitation on measurable core sizes of approximately $120''$
746: corresponds to $3\times10^4$~AU in Perseus and Serpens, and
747: $1.5\times10^4$~AU in Ophiuchus. Note that these sizes are of
748: order the median core separation in each cloud (see
749: \S~\ref{clustsect}), meaning that we are just as likely to be limited
750: by the crowding of cores as by our sensitivity limits in the
751: measurement of large cores. The dependence of measurable core size on
752: cloud distance can be seen in Figure~\ref{mvscomp}, where the
753: completeness rises steeply at smaller linear deconvolved sizes for
754: Ophiuchus than for Perseus or Serpens. Thus we are biased against
755: measuring large cores in Ophiuchus compared to the other two clouds.
756: Although there are cores in our sample with sizes up to
757: $3\times10^4$~AU (Figure~\ref{mvscomp}), most cores have sizes
758: substantially smaller than the largest measurable value, again
759: indicating that we are not limited by systematics.
760:
761: To summarize, the Bolocam 1.1~mm observations presented here naturally
762: pick out sub-structures in molecular clouds with high volume density
763: ($n\gtrsim 2\times 10^4$~cm$^{-3}$). These millimeter cores have a
764: contrast of at least $30-100$ compared to the average cloud density as
765: measured by the visual extinction map ($2-10\times10^3$~cm$^{-3}$).
766: Many cores are embedded in lower density extended structures, which
767: contribute approximately half the mass measurable in the Bolocam
768: 1.1~mm maps. Finally, Monte Carlo tests indicate that we can detect
769: cores with intrinsic sizes up to approximately $120''$.
770:
771: \subsection{Distance effects}\label{convsect}
772:
773: To test the effects of instrumental resolution and its dependence on
774: distance, we convolve the Ophiuchus map with a larger beam to simulate
775: putting it at approximately the same distance as Perseus and Serpens.
776: After convolving the unfiltered Ophiuchus map to $62''$ resolution, we
777: apply the optimal filter and re-compute the local rms noise, as
778: described for Serpens (\S~\ref{idsect}). The pixel scale in the
779: convolved map is still 10\arcsec\ pixel$^{-1}$, but the resolution is
780: now $62''$ and the rms is lower than in the original map, with a
781: median value of 17~mJy beam$^{-1}$ in the main L~1688 region. Source
782: detection and photometry is carried out in the same way as for the
783: original map, with the exception that a 62\arcsec\ beam is assumed.
784:
785: We detect 26 sources in the degraded-resolution map, or 40\% fewer
786: than the 44 sources in the original map. Therefore a number of
787: sources do become confused at lower resolution. The basic source
788: properties for the original and degraded-resolution samples, including
789: angular deconvolved sizes, axis ratios, mass distribution, and mean
790: densities, are compared in Figure~\ref{convplots}. Here the angular
791: deconvolved size is defined as $\theta_{dec} = \sqrt{
792: \theta_{meas}^2 - B^2}$, where $\theta_{meas}$ is the geometric mean
793: of the measured minor and major axis FWHM sizes and $B$ is the
794: pointing-smeared beam FWHM ($32.5\arcsec$ and $62\arcsec$ for the
795: original and degraded resolution maps, respectively).
796:
797: We note that the average source size in the degraded-resolution
798: map is nearly twice that in the original Ophiuchus map
799: (average angular deconvolved size of $61''$ versus $98''$). Such a large
800: size difference cannot be fully accounted for by blending of sources,
801: as even isolated sources show the same effect. This behavior provides
802: clues to the intrinsic intensity profile of the sources. For example,
803: a gaussian intensity profile or a solid disk of constant intensity
804: will both have measured deconvolved sizes that are similar in maps
805: with $31''$ and $62''$ beams. Conversely, a power law intensity
806: profile will have a larger measured size in the $62''$ resolution map.
807:
808: The ratio of angular deconvolved size to beam size
809: ($\theta_{dec}/\theta_{mb}$; Figure~\ref{sizetconv}) are similar for
810: the degraded-resolution (median $\theta_{dec}/\theta_{mb}=1.7$) and
811: original (median $\theta_{dec}/\theta_{mb}=1.5$) samples, further
812: evidence for power law intensity profiles. An intrinsic gaussian or
813: solid disk intensity profile will result in $\theta_{dec}/\theta_{mb}$
814: values in the degraded-resolution map that are approximately half
815: those in the original map, while a $1/r^2$ intensity profile results
816: in similar values in the degraded-resolution and original map (0.9
817: versus 1.3). We discuss source profiles further in
818: Section~\ref{sizecompsect} below.
819:
820: Sources in the degraded-resolution map appear slightly more elongated,
821: with an average axis ratio at the half-maximum contour of $1.3\pm0.2$
822: compared to $1.2\pm0.2$ for the original map. Larger axis ratios
823: are expected for blended sources in a lower-resolution map. The slope
824: of the mass distribution is not significantly changed: $2.0\pm0.4$ for
825: the degraded-resolution sample compared to $2.1\pm0.3$ for the
826: original sample. The factor of two increase in deconvolved sizes do
827: create lower mean densities in the degraded-resolution sample, however
828: (median density of $1.6\times10^5$~cm$^{-3}$ compared to
829: $5.8\times10^5$~cm$^{-3}$ for the original sample). The effect of
830: resolution on mean densities will be discussed further in
831: \S~\ref{dencompsect}.
832:
833: Given the small number of sources in the degraded-resolution map, we carry out
834: the three cloud comparison below using the original Ophiuchus map to
835: mitigate uncertainties from small number statistics. Any notable
836: differences between the original and degraded-resolution Ophiuchus results will
837: be discussed where appropriate.
838:
839: \subsection{Physical Implications of Source Sizes and Shapes}\label{sizecompsect}
840:
841: To compare sources in clouds at different distances, we first look at
842: the linear deconvolved size: $D_{dec} = d \sqrt{ \theta_{meas}^2 -
843: B^2}$, where $d$ is the cloud distance, $\theta_{meas}$ is measured
844: angular size, and $B$ is the beam size. The distributions of
845: linear deconvolved source sizes for the three clouds are shown in
846: Figure~\ref{dsizecomp}, left. Sources in Ophiuchus have smaller
847: deconvolved sizes than those in Perseus or Serpens by almost a
848: factor of two, with mean vales of $7.5\times10^3$~AU in Ophiuchus
849: compared to $1.2\times10^4$~AU in Serpens and $1.5\times10^4$~AU in
850: Perseus. There is a systematic uncertainty in the deconvolved size
851: associated with the uncertainty in the effective beam size, which
852: becomes larger with larger pointing errors. The effective beam in any
853: of the three clouds may be as large as $35\arcsec$, which would
854: decrease deconvolved sizes negligibly, by up to $10^3$~AU depending on the
855: distance and measured size.
856:
857:
858: While there are possible physical explanations for intrinsic size
859: differences, for instance a denser medium with a shorter Jeans length
860: should produce smaller cores on average, we are more likely seeing a
861: consequence of the higher linear resolution in Ophiuchus, as
862: discussed in \S~\ref{convsect}. Thus cores in Serpens and Perseus
863: would likely appear smaller if observed at higher resolution, and measured
864: linear deconvolved sizes should be regarded as upper limits.
865: To reduce the effects of distance, we examine the ratio of angular
866: deconvolved size to beam size ($\theta_{dec}/\theta_{mb}$;
867: Figure~\ref{dsizecomp}, right). We found in \S~\ref{convsect} that
868: $\theta_{dec}/\theta_{mb}$ does not depend strongly on the linear
869: resolution, but does depend on the intrinsic source intensity
870: profile.
871:
872: If the millimeter sources follow power law density distributions,
873: which do not have a well defined size, then \citet[][hereafter
874: Y03]{young03} show that $\theta_{dec}/\theta_{mb}$ depends on the
875: index of the power-law, and not on the distance of the source. So,
876: for example, if sources in Perseus and Ophiuchus have the same
877: intrinsic power law profile, the mean $\theta_{dec}/\theta_{mb}$
878: should be similar in the two clouds, and the mean linear deconvolved
879: size should be twice as small in Ophiuchus because it lies at half the
880: distance. This is precisely the behavior we observe, suggesting that
881: many of the detected 1.1~mm sources have power law density profiles.
882:
883: Considering that a number of the 1.1~mm sources have internal
884: luminosity sources (M. Enoch et al. 2007, in preparation), and that
885: protostellar envelopes are often well described by power law profiles
886: \citep[][Y03]{shir02}, this is certainly a plausible scenario.
887: According to the correlation between $\theta_{dec}$ and density power
888: law exponent $p$ found by Y03, median $\theta_{dec}/\theta_{mb}$
889: values of 1.7 in Perseus, 1.5 in Ophiuchus and 1.3 in Serpens would
890: imply average indices of $p=1.4$, 1.5, and 1.6 respectively. These
891: numbers are consistent with mean $p$ values found from radiative
892: transfer modeling of Class 0 and Class I envelopes \citep[$p\sim
893: 1.6$,][Y03]{shir02}, although the median for those samples is
894: somewhat higher ($p\sim 1.8$). Note that source profiles could
895: deviate from a power-law on scales much smaller than the beam size, or
896: on scales larger than our size sensitivity ($200''$) without
897: affecting our conclusions.
898:
899: Perseus displays the widest dispersion of angular sizes, ranging
900: continuously from $1-4 \theta_{mb}$. By contrast, more than half the
901: sources in Serpens and Ophiuchus are within $0.5 \theta_{mb}$ of
902: their respective mean values. Although there is a group of Ophiuchus
903: sources at large sizes in Figure~\ref{dsizecomp}, note that the
904: degraded-resolution Ophiuchus sample displays a very narrow range of
905: sizes (Figure~\ref{sizetconv}), similar to Serpens. The observed size
906: range in Perseus would correspond to a wide range of power law
907: indices, from very shallow ($p\sim1$) to that of a singular isothermal
908: sphere ($p=2$). A more likely possibility, however, is that sources
909: with large $\theta_{dec}/\theta_{mb}$ do not follow power law density
910: profiles.
911:
912: The axis ratio at the half-maximum contour is a simple measure of
913: source shape. Figure~\ref{axiscomp} shows the distribution of 1.1~mm
914: source axis ratios in the three clouds. Our simulations suggest that
915: axis ratios of up to 1.2 can be introduced by the data reduction
916: (Paper~I), so we consider sources with an axis ratio $<1.2$ to be
917: round, and those with a ratio $>1.2$ to be elongated. Sources in
918: Ophiuchus tend to be round, with a mean axis ratio of 1.2, but note
919: that the mean axis ratio in the degraded-resolution Ophiuchus sample
920: is 1.3. The average axis ratio in Serpens is 1.3, and Perseus
921: sources exhibit the largest axis ratios with a mean of 1.4 and a tail
922: out to 2.7.
923:
924: We found in Paper~II that Ophiuchus sources were more elongated at
925: the $4\sigma$ contour than at the half-max contour, as would be the
926: case for round cores embedded in more elongated filaments. A similar
927: situation is seen in Serpens; the average axis ratio at the $4\sigma$
928: contour is 1.4 in all three clouds. Thus cores in Perseus are
929: somewhat elongated on average, while objects in Serpens and Ophiuchus
930: appear more round at the half-max contour but elongated at the
931: $4\sigma$ contour, suggesting round cores embedded in filamentary
932: structures.
933:
934: In addition to angular sizes, Y03 also note a relationship between
935: axis ratio and density power law exponent, finding that aspherical
936: sources are best modeled with shallower density profiles. The inverse
937: proportionality between $p$ and axis ratio demonstrated in Figure~25
938: of Y03 suggests power law indices in all three clouds between 1.5 and
939: 1.7. These values are consistent with those inferred from the
940: average angular deconvolved source sizes, and the wider variation of
941: axis ratios in Perseus again points to a larger range in $p$ for that
942: cloud.
943:
944: \subsection{Densities and the Mass versus Size Distribution}\label{dencompsect}
945:
946: Mean densities calculated using linear deconvolved FWHM sizes appear
947: to be significantly higher in Ophiuchus, where the median of the mean
948: densities of the sample is $5.4\times10^5$~cm$^{-3}$, than in Serpens
949: (median density $2.2\times10^5$~cm$^{-3}$) or Perseus (median density
950: $1.6\times10^5$~cm$^{-3}$), as seen in Figure~\ref{dencomp}, left.
951: There is a large scatter with standard deviation of order twice the
952: mean value in all three clouds. Sources in Ophiuchus tend to be less
953: massive than in the other two clouds, so the larger mean densities can
954: be entirely attributed to smaller deconvolved sizes in the Ophiuchus
955: sample, which are sensitive to the shape of the intrinsic density
956: distribution (Y03). As noted in \S~\ref{convsect}, linear resolution
957: has a strong systematic effect on deconvolved sizes, and consequently
958: on mean densities. The median density of the degraded-resolution Ophiuchus
959: sample is $1.6\times10^5$~cm$^{-3}$, similar to both Perseus and
960: Serpens.
961:
962: We additionally calculate mean densities using the full-width at
963: $4\sigma$ size rather than the FWHM size (Figure~\ref{dencomp},
964: right) to test the hypothesis that mean density differences are
965: largely an effect of how source sizes are measured. Using this
966: definition, differences between the clouds are less pronounced, with
967: median densities of $1.2 \times 10^5$~cm$^{-3}$ in Perseus, $1.3
968: \times 10^5$~cm$^{-3}$ in Serpens, and $2.0 \times 10^5$~cm$^{-3}$ in
969: Ophiuchus. These numbers suggest that source mean densities are less
970: dependent on cloud distance when measured at the radius where the
971: source merges into the background, rather than at the half-max.
972:
973: Figure~\ref{mvscomp2} again displays the source total mass versus size
974: distribution, using the angular deconvolved size ($\theta_{dec}$, in
975: units of the beam size) rather than the linear deconvolved size.
976: With the exception of two low mass compact sources, 1.1~mm sources in
977: Serpens exhibit a wide range of masses and a narrow range of sizes.
978: Sources in Perseus, in contrast, demonstrate a wide range in both mass
979: and size. This difference likely reflects a wider variety of physical
980: conditions in the Perseus cloud, including the existence of more
981: sources outside the main cluster regions. In such lower density
982: regions, sources may be more extended than in densely populated
983: groups. This idea is supported by our observations: Perseus sources
984: within the NGC~1333 region are smaller on average (mean size of
985: $59\pm13''$) than those in the rest of the cloud (mean size of
986: $71\pm21''$). Sources in Ophiuchus have a more bimodal distribution,
987: with the majority occupying a narrow range of sizes, and a smaller
988: group with $\theta_{dec}>3\theta_{mb}$.
989:
990: \subsection{Fragmentation and the Core Mass Distribution}\label{masscompsect}
991:
992: Differential ($dN/dM$) core mass distributions (CMDs) of 1.1~mm
993: sources in Serpens, Perseus, and Ophiuchus are shown in
994: Figure~\ref{masscomp}, with those of Perseus and Ophiuchus scaled
995: down for clarity. Error bars reflect $1\sigma$ $\sqrt{N}$ statistical
996: uncertainties. Dashed lines indicate empirical 50\% completeness
997: limits for average sized sources in each cloud, which are determined
998: as described in \S~\ref{coresect} and in Paper~I. Mass distributions
999: include all 1.1~mm cores in each cloud, including those that may be
1000: associated with embedded protostellar sources.
1001:
1002: The shape of the Ophiuchus and Perseus mass distributions are quite
1003: similar above their respective completeness limits ($M \gtrsim 0.5
1004: \msun$ in Ophiuchus and $M \gtrsim 0.8 \msun$ in Perseus). Fitting a
1005: power law ($dN/dM \propto M^{-\alpha}$) to the CMDs, we find that they
1006: both have a best fit slope of $\alpha=2.1$, although the error is
1007: larger on the slope for Ophiuchus ($\sigma_{\alpha} = 0.3$) than for
1008: Perseus ($\sigma_{\alpha} = 0.1$). The slope of the Serpens CMD is
1009: marginally different ($\alpha = 1.6 \pm 0.2$), being flatter than in
1010: the other two clouds by approximately $2 \sigma$.
1011:
1012: The two-sided Kolmogorov-Smirnov test indicates a high probability
1013: (46\%) that the Perseus and Ophiuchus mass distributions are
1014: representative of the same parent population. Conversely, the
1015: probabilities that the Serpens core masses are sampled from the same
1016: population as the Perseus (prob$=12$\%) or Ophiuchus (prob$=5$\%)
1017: masses are much lower. Although it is possible that the lower linear
1018: resolution in Perseus and Serpens has lead to larger masses via
1019: blending, the test we conducted to increase the effective beam size in
1020: the Ophiuchus map did not appreciably change the shape of the
1021: Ophiuchus mass distribution (see Figure~\ref{convplots}).
1022:
1023: If the shape of the CMD is a result of the fragmentation process, then
1024: the slope of the CMD can be compared to models, e.g. of turbulent
1025: fragmentation. \citet{padoan02} argue that turbulent fragmentation
1026: naturally produces a power law with $\alpha = 2.3$, consistent with
1027: the slopes we measure in Perseus and Ophiuchus ($\alpha=2.1$), but not
1028: with Serpens ($\alpha=1.6$). Recently \citet[][hereafter
1029: BP06]{paredes06} have questioned that result, finding that the
1030: shape of the CMD depends strongly on the Mach number of the turbulence
1031: in their simulations. The BP06 smoothed particle hydrodynamics (SPH)
1032: simulations show that higher Mach
1033: numbers result in a larger number of sources with lower mass and a
1034: steep slope at the high mass end (their Figure~5). Conversely, lower
1035: Mach numbers favor sources with higher mass, resulting in a smaller
1036: number of low mass sources, more high mass cores, and a shallower
1037: slope at the high mass end.
1038:
1039: Using an analytic argument, \citet{padoan02} also note a relationship
1040: between core masses and Mach numbers, predicting that the mass of the
1041: largest core formed by turbulent fragmentation should be inversely
1042: proportional to the square of the Alfv\'{e}nic Mach number on the
1043: largest turbulent scale $\mathcal{M}_{A}^2$. Given that our ability
1044: to accurately measure the maximum core mass is limited by resolution,
1045: small number statistics, and cloud distance differences, we focus
1046: here on the overall CMD shapes.
1047:
1048: To compare our observational results to the simulations of BP06, we
1049: estimate the sonic Mach number $\mathcal{M} = \sigma_v/c_s$ in each
1050: cloud. Here $\sigma_v$ is the observed rms velocity dispersion, $c_s
1051: = \sqrt{kT/\mu m_H}$ is the isothermal sound speed, and $\mu = 2.33$
1052: is the mean molecular weight per particle. Large $^{13}$CO maps of
1053: Perseus and Ophiuchus observed with FCRAO at a resolution of
1054: 44\arcsec\ are publicly available as part of ``The COMPLETE Survey of
1055: Star Forming Regions''\footnote{http://cfa-www.harvard.edu/COMPLETE/}
1056: \citep[COMPLETE;][]{good04,ridge06}. Average observed rms velocity
1057: dispersions kindly provided by the COMPLETE team are $\sigma_v =
1058: 0.68$~km~s$^{-1}$ in Perseus, $\sigma_v = 0.44$~km~s$^{-1}$ in
1059: Ophiuchus, and $\sigma_v = 0.92$~km~s$^{-1}$ in Serpens (J. Pineda,
1060: personal communication). These were measured by masking out all
1061: positions in the map that have peak temperatures with a S/N less than
1062: 10, fitting a Gaussian profile to each, and taking an average of the
1063: standard deviations.
1064:
1065: We note that the value of $\sigma_v= 0.68$~km~s$^{-1}$ for Perseus is
1066: smaller than a previous measurement of $\sigma_v= 2.0$~km~s$^{-1}$
1067: based on AT\&T Bell Laboratory 7~m observations of a similar area of
1068: the cloud \citep{pad03}. The smaller value derived by the COMPLETE
1069: team is most likely a consequence of the method used: a linewidth is
1070: calculated at every position and then an average of these values is
1071: taken. In contrast to calculating the width of the averaged spectrum
1072: this method removes the effects of velocity gradients across the
1073: cloud. The different resolutions of the surveys (44\arcsec\ and 0.07
1074: km~s$^{-1}$ for the COMPLETE observations, 100\arcsec\ and 0.27
1075: km~s$^{-1}$ for the Padoan et al. observations) may also play a role.
1076: Sources of uncertainty in the linewidth measurement are not
1077: insignificant, and include the possibility that $^{13}$CO is optically
1078: thick to an unknown degree, and the fact that line profiles are not
1079: necessarily well fit by a gaussian, especially in Perseus where the
1080: lines sometimes appear double-peaked.
1081:
1082: Assuming that the sound speed is similar in all three clouds, the
1083: relative velocity dispersions suggest that turbulence is more
1084: important in Serpens than in Perseus or Ophiuchus, by factors of
1085: approximately 1.5 and 2, respectively. Mach numbers calculated using
1086: the observed $\sigma_v$ and assuming a gas kinetic temperature of 10~K
1087: are $\mathcal{M}=4.9$ (Serpens), $\mathcal{M}=3.6$ (Perseus), and
1088: $\mathcal{M}=2.3$ (Ophiuchus). We focus on Serpens and Ophiuchus, as
1089: they are the most different. Serpens is observed to have a higher
1090: Mach number than Ophiuchus, but the CMD indicates a larger number of
1091: high mass cores, a shallower slope at the high mass end, and a dearth
1092: of low mass cores compared to Ophiuchus. This result is contrary to
1093: the trends found by BP06, which would predict a steeper slope and more
1094: low mass cores in Serpens.\footnote{It should be noted that the shape of the CMDs presented by BP06 depend on the type of numerical code used, and not on the cloud Mach number alone. Here we have compared to the BP06 SPH simulations (their Figure~5); for the BP06 total variation diminishing (TVD) method (their Figure~4), a larger number of low mass cores are again seen for higher Mach numbers, but the CMD slopes at the high mass end are independent of Mach number.}
1095:
1096: One core of relatively high mass is measured in the degraded-resolution
1097: map of Ophiuchus, as is expected for blending at lower
1098: resolution (Figure~\ref{convplots}), making the difference between the
1099: Serpens and Ophiuchus CMDs less dramatic. The degraded-resolution CMD
1100: for Ophiuchus still has a steeper slope ($\alpha=2.0\pm0.4$) than the
1101: Serpens CMD, however, and a larger fraction of low mass sources: 81\%
1102: of sources in the degraded-resolution Ophiuchus sample have $M<2M_{\sun}$
1103: compared to 49\% in the Serpens sample.
1104:
1105: More accurate measurements of the Mach number and higher resolution
1106: studies of the relative CMD shapes will be necessary to fully test
1107: the BP06 prediction, given that uncertainties are currently too
1108: large to draw firm conclusions. As both numerical simulations and
1109: observations improve, however, the observed CMD and measurements of
1110: the Mach number will provide a powerful constraint on turbulent star
1111: formation simulations.
1112:
1113: Comparing the shape of the CMD to the stellar initial mass function
1114: (IMF) may give insight into what determines final stellar masses: the
1115: initial fragmentation into cores, competitive accretion, or feedback
1116: processes. The shape of the local IMF is still uncertain
1117: \citep{scalo05}, but recent work has found evidence for a slope of
1118: $\alpha=2.5-2.8$ for stellar masses $M \gtrsim 1 \msun$, somewhat
1119: steeper than the slopes we observe in all three clouds. For example,
1120: \citet{reid02} find $\alpha = 2.5$ above 0.6 \msun, and $\alpha = 2.8$
1121: above 1 \msun. \citet{chab03} suggests $\alpha = 2.7$
1122: ($M>1~M_{\sun}$), while \citet{schroder03} finds $\alpha = 2.7$ for
1123: $1.1 < M < 1.6$ \msun\ and $\alpha = 3.1$ for $1.6 < M < 4$ \msun.
1124: For reference, the Salpeter IMF has a slope of $\alpha=2.35$
1125: \citep{sal55}.
1126:
1127: For comparison to the IMF, we would ideally like to construct a CMD
1128: that includes starless cores only, so that it is a measure of the
1129: mass initially available to form a star. Although we cannot separate
1130: prestellar cores from more evolved objects with millimeter data
1131: alone, we are currently comparing the Bolocam maps with c2d
1132: \textit{Spitzer} Legacy maps of the same regions, which will allow us
1133: to distinguish prestellar cores from those with internal luminosity
1134: sources (M. Enoch et al. 2007, in preparation).
1135:
1136:
1137: \subsection{Clustering}\label{clustsect}
1138:
1139: We use the two-point correlation function,
1140: \begin{equation}
1141: w(r) = \frac{H_s(r)}{H_r(r)} - 1,
1142: \end{equation}
1143: as a quantitative measure of the clustering of cores. Here $H_s(r)$
1144: is the number of core pairs with separation between $log(r)$ and
1145: $log(r)+dlog(r)$, and $H_r(r)$ is similar but for a random
1146: distribution (see Paper~I). Thus $w(r)$ is a measure of excess
1147: clustering over a random distribution, as a function of separation.
1148:
1149: Figure~\ref{corrncomp} plots the cloud two-point correlation
1150: functions, with the best fit power laws, $w(r) \propto r^n$, shown.
1151: The average linear source FWHM size is indicated, as well as the
1152: minimum possible separation (the beam size), and the linear map size.
1153: The best fit slopes to $w(r)$ for Serpens ($n=-1.5 \pm 0.2$), Perseus
1154: ($n=-1.18 \pm 0.06$), and Ophiuchus ($n=-1.5 \pm 0.1$) are consistent
1155: within $2\sigma$, but there is an indication both from the slopes and
1156: a visual examination of the plot that $w(r)$ falls off more steeply
1157: in Serpens and Ophiuchus than in Perseus. A shallower slope suggests
1158: that clustering remains strong out to larger scales in Perseus than in
1159: the other two clouds.
1160:
1161: We do find that a broken power law with slopes of $n=-0.75$
1162: ($1.3\times 10^4 < r < 6.3\times 10^4$AU) and $n=-3.3$ ($r >
1163: 6.3\times 10^4$AU) is a better fit to the Serpens correlation
1164: function than a single power law, indicating that clustering remains
1165: strong to intermediate scales, as in Perseus, but then drops off
1166: quickly. Also shown in Figure~\ref{corrncomp} is $w(r)$ for the
1167: degraded-resolution Ophiuchus sample (gray hatched curve). The
1168: degraded-resolution sample is fit by a shallower slope
1169: ($n=-1.2\pm0.3$) because there are fewer sources at small
1170: separations, but it is broadly consistent with the original Ophiuchus
1171: sample.
1172:
1173: Our general conclusion that clustering of 1.1~mm sources remains
1174: strong over larger scales in Perseus than in Serpens and Ophiuchus can
1175: also be reached by visually examining each map. Perseus has highly
1176: clustered regions spread over a larger area as well as a number of
1177: more distributed sources, whereas Serpens and Ophiuchus have a single
1178: main cluster with fewer small groups spread throughout the cloud.
1179:
1180: A few caveats should be noted here. First, the fact that clustering
1181: seems to extend over larger scales in Perseus could simply be due to
1182: the fact that Perseus has more widely separated regions of star
1183: formation, and the physical association of all these regions has not
1184: been firmly established. Second, the map of Serpens covers a much
1185: smaller linear area than that of Perseus (30~pc$^2$ versus
1186: 140~pc$^2$); we need to be sure that the steepening of the slope in
1187: Serpens is not caused by the map size. This was confirmed by taking
1188: a piece of the Perseus map equal in size to the Serpens map and
1189: recalculating $w(r)$. Although the value of the slope changed
1190: slightly, the smaller map size did not cause the slope to steepen at
1191: large separations. A more serious issue is that the overall
1192: amplitude, but not the slope, of $w(r)$ depends on how large an area
1193: the random distribution $H_r(r)$ covers. We choose each random
1194: distribution such that the largest pair separation is similar to the
1195: largest pair separation in the real data. For Serpens and Ophiuchus
1196: this means that the random distribution does not cover the entire
1197: observed area.
1198:
1199: Finally, we look at two other measures of clustering: the peak number
1200: of cores per square parsec, and the median separation of cores.
1201: Although the median separation of cores is much smaller in Ophiuchus
1202: ($6.2\times10^3$~AU) than in Perseus ($2.6\times10^4$~AU) or Serpens
1203: ($2.3\times10^4$~AU), the median separation in the degraded-resolution
1204: Ophiuchus map ($2.6\times10^4$~AU) is consistent with the other two
1205: clouds. Imposing a uniform flux limit of
1206: $75 \mathrm{mJy} \times(\frac{250\mathrm{pc}}{d})^2$ across all three clouds,
1207: equivalent to the $5\sigma$ limit in the shallowest map (Perseus),
1208: does not significantly change these results.
1209:
1210: We calculate the number of cores within one square parsec at each
1211: point in the map, and take the peak value to be the peak number of
1212: cores per square parsec. The peak values are $12 ~\mathrm{pc}^{-2}$
1213: in Serpens, $22 ~\mathrm{pc}^{-2}$ in Perseus, and $24
1214: ~\mathrm{pc}^{-2}$ in the original Ophiuchus map. Differences in
1215: linear resolution and completeness both have an effect in this case;
1216: the peak number in Ophiuchus falls to 20~pc$^{-2}$ when using a
1217: uniform flux limit, and to 12~pc$^{-2}$ in the degraded-resolution
1218: map. The peak number of cores per parsec provides further evidence
1219: that clustering is stronger in Perseus than in the other two clouds.
1220:
1221:
1222: %\subsection{Relationship to Cloud Column Density and Efficiency of Forming Cores}\label{threshsect}
1223: \subsection{Relationship to Cloud Column Density}\label{threshsect}
1224:
1225: In contrast to the extinction map, which is a measure of the general
1226: cloud (line-of-sight averaged) column density, the 1.1~mm map is
1227: sensitive only to regions of high volume density (see
1228: \S~\ref{coresect}). A comparison of the two tells us, therefore,
1229: about the relationship between dense star-forming structures and the
1230: column density of the larger-scale cloud. A visual comparison of the
1231: 1.1~mm maps of each cloud with visual extinction maps derived from
1232: the reddening of background stars (e.g. Figure~\ref{avfig}) suggests
1233: that 1.1~mm cores are generally found in regions of the cloud with
1234: high \av.
1235:
1236: Figure~\ref{avcomp} quantifies the relationship between dense cores
1237: and the surrounding cloud column density by plotting the cumulative
1238: fraction of 1.1~mm cores in each cloud as a function of cloud \av. In
1239: all three clouds the majority of cores are found at high cloud column
1240: density ($\av>7^m$). The \av\ levels above which 75\% of cores are
1241: found in each cloud are indicated by thin lines in Figure~\ref{avcomp}: 75\% of
1242: 1.1~mm cores in Perseus, Serpens, and Ophiuchus are found at visual
1243: extinctions of $\av \gtrsim 8^m$, $\av \gtrsim 15^m$, and $\av \gtrsim
1244: 23^m$, respectively. Although there is not, in general, a strict extinction
1245: threshold for finding cores, below these \av\ levels the likelihood of finding a
1246: 1.1~mm core is very low.
1247: Only in Ophiuchus does there appear to be a true
1248: \av\ threshold; only two cores are found at $\av < 17^m$ in this cloud.
1249:
1250: The cumulative distribution for the degraded resolution Ophiuchus
1251: sample is nearly identical to the original Ophiuchus sample, but
1252: shifted to lower \av\ by 1--2~mag. For the degraded resolution map,
1253: 75\% of cores are found at $\av \gtrsim 20^m$. Based on this result and the
1254: value for the original Ophiuchus sample, we adopt $\av\sim20-23^m$ for the 75\%
1255: level in Ophiuchus. Requiring that all cores in the Perseus and
1256: Serpens sample meet the detection threshold for Ophiuchus
1257: (approximately 110~mJy) changes the distributions negligably.
1258: Only for Serpens is the 75\% level increased slightly, from $\av \sim15^m$ to $16^m$.
1259:
1260:
1261: Cloud to cloud differences could indicate variations in the core
1262: formation process with environment, differing degrees of
1263: sub-structure in the clouds, or varying amounts of foreground
1264: extinction. Note that in Papers I and II we found extinction
1265: ``thresholds'' for Perseus and Ophiuchus, using a different analysis
1266: that looked at the probability of finding a 1.1~mm core as a function
1267: of \av, of $\av \sim 5^m$ and $\av \sim 9^m$, respectively. Those
1268: values were derived using 2MASS-only \av\ maps, rather than the c2d
1269: extinction maps used here.
1270:
1271:
1272: \citet{john04} have suggested an extinction threshold for forming
1273: dense cores in Ophiuchus at $\av \sim 15^m$, based on the lowest \av\
1274: at which SCUBA cores were observed to have sizes and fluxes
1275: consistent with those of stable Bonnor-Ebert spheres. We find that
1276: most Bolocam cores in Ophiuchus are found at even higher extinctions
1277: (75\% at $\av>20-23^m$); the discrepancy likely arises from
1278: differences in the extinction maps used. \citet{john04} used a
1279: 2MASS-derived extinction map, while the c2d extinction map used here
1280: includes IRAC data as well and probes somewhat higher \av\ values.
1281: \citet{hatch05} find no evidence for an extinction threshold in
1282: Perseus using $850~\micron$ SCUBA data, but \citet{kirk06}, also using
1283: SCUBA data, find a threshold of $\av \sim5^m$. Despite differences in
1284: detail, the relative 75\% \av\ levels found here (highest in Ophiuchus
1285: and lowest in Perseus) are consistent with the relative values of
1286: previous extinction threshold measurements in Perseus and Ophiuchus.
1287:
1288:
1289: An extinction threshold has been predicted by \citet{mckee89} for
1290: photoionization-regulated star formation in magnetically supported
1291: clouds. In this model star formation is governed by ambipolar
1292: diffusion, which depends on ionization levels in the cloud. Core
1293: collapse and star formation will occur only in shielded regions of a
1294: molecular cloud where $\av \gtrsim 4-8^m$. \citet{john04} note
1295: that it is not clear how turbulent models of star formation would produce
1296: an extinction threshold for star-forming objects, and
1297: interpret such a threshold as evidence for magnetic support.
1298:
1299:
1300: \subsection{Efficiency of Forming Cores}\label{efficsect}
1301:
1302: Another interesting measure of global cloud conditions is the fraction
1303: of cloud mass contained in dense cores. Table~\ref{avtab} lists the
1304: cloud area and cloud mass within increasing \av\ contours, calculated
1305: from the c2d visual extinction maps, as well as the total mass in
1306: cores within the same extinction contour. Cloud masses are
1307: calculated from the c2d \av\ maps (\S~\ref{masssec},
1308: Eq.~\ref{avmasseq}). The total cloud mass within $\av>2^m$ is
1309: 3470~\msun for Serpens, 7340~\msun for Perseus, and 3570~\msun for
1310: Ophiuchus.
1311:
1312: Within a given \av\ contour, the mass ratio is defined as the ratio of
1313: total core mass to total cloud mass, and is a measure of the
1314: efficiency of core formation at that \av\ level. For example, the
1315: mass ratio at the $\av=2^m$ contour is equivalent to the fraction of
1316: total cloud mass that is contained in dense cores. In each cloud
1317: 1.1~mm cores account for less than 5\% of the total cloud mass. The
1318: mass ratios at $\av=2^m$ are similar in all three clouds: 3.8\% in
1319: Perseus, 2.7\% in Serpens, and 1.2\% in Ophiuchus. If we restrict
1320: ourselves to $\av>6^m$ in each cloud, which is reasonable given that
1321: the Serpens map was only designed to cover $\av>6^m$, the mass ratio
1322: is still low in all three clouds ($<10$\%), and remains higher in
1323: Perseus (7\%) than in Serpens (4\%) or Ophiuchus (2\%). Low mass
1324: ratios are consistent with measurements of the overall star-formation
1325: efficiency of $1-2\%$, which suggests that molecular cloud material
1326: is relatively sterile \citep[e.g.][]{evans99,leis89}.
1327:
1328: \citet{john04} find a mass ratio of 2.5\% in a survey of approximately
1329: $4$~deg$^2$ of Ophiuchus, quite similar to our result. Note, however,
1330: that the \citet{john04} core masses (and mass ratio) should be
1331: multiplied by a factor of 1.5 to compare to our values, due to
1332: differences in assumed values of $T_D$, $\kappa_{\nu}$, and $d$. In a
1333: similar analysis of 3.5 deg$^2$ in Perseus, also using SCUBA
1334: $850~\micron$ data, \citet{kirk06} found a mass ratio for $\av>0$ of
1335: only 1\%. The difference between this result and our value arises
1336: primarily from the smaller core masses of those authors, which should
1337: be multiplied by 2.5 to compare to ours.
1338:
1339: In all three clouds the mass ratio rises with increasing \av\ contour,
1340: indicating that in high extinction regions a greater percentage of
1341: cloud mass has been assembled into cores, consistent with the idea
1342: that star formation is more efficient in dense regions. Although
1343: this is an intuitively obvious result, it is not a necessary one. If,
1344: for example, a constant percentage of cloud mass were contained in
1345: dense cores at all column densities, there would be a large number of
1346: dense cores lying in a low \av\ background. On the other hand, a
1347: molecular cloud might consist of large regions of uniformly high
1348: extinction in which we would find no 1.1~mm cores because there is no
1349: sub-structure, and millimeter cores require high density contrast (see
1350: \S~\ref{coresect}).
1351:
1352: At higher \av\ levels, mass ratios vary considerably from cloud to
1353: cloud. The mass ratio remains fairly low in Ophiuchus, with a
1354: maximum value of 9\% at $\av=30^m$. In contrast, the mass ratio
1355: rises rapidly in Serpens to 65\% at $\av=30^m$, which may suggest
1356: that Serpens has formed cores more efficiently than Ophiuchus at high
1357: \av.
1358:
1359: \section{Summary}\label{sumsect}
1360:
1361: This work completes a three-cloud study of the millimeter continuum
1362: emission in Perseus, Ophiuchus, and Serpens. We examine similarities
1363: and differences in the current star formation activity within the
1364: clouds using large-scale 1.1~mm continuum maps completed with Bolocam
1365: at the CSO. In total, our surveys cover nearly 20 deg$^2$ with a
1366: resolution of 31\arcsec\ (7.5 deg$^2$ in Perseus, 10.8 deg$^2$ in
1367: Ophiuchus, and 1.5 deg$^2$ in Serpens), and we have assembled a
1368: sample of 200 cores (122 in Perseus, 44 in Ophiuchus, and 35 in
1369: Serpens). Point mass detection limits vary from approximately 0.1 to
1370: 0.2 \msun\ depending on the cloud. The results presented here
1371: provide an unprecedented global picture of star formation in three
1372: clouds spanning a range of diverse environments.
1373:
1374: These Bolocam 1.1~mm observations naturally select dense cores with
1375: $n\gtrsim 2\times 10^4$~cm$^{-3}$ and density contrast compared to the
1376: background cloud of at least $30-100$. We test instrumental biases
1377: and the effects of cloud distance by degrading the resolution of the
1378: Ophiuchus map to match the distance of Perseus and Serpens. We find
1379: that linear resolution strongly biases measured linear deconvolved
1380: source sizes and mean densities, but not the mass distribution slope.
1381: Angular deconvolved sizes are less strongly affected by cloud
1382: distance.
1383:
1384: Rather than a true physical difference, the small mean linear
1385: deconvolved sizes in Ophiuchus ($0.8 \times 10^4$~AU) compared to
1386: Perseus ($1.5 \times 10^4$~AU) and Serpens ($1.2 \times 10^4$~AU) are
1387: likely a result of observing sources with power law density profiles,
1388: which do not have a well defined size, at a distance of 125~pc in
1389: Ophiuchus versus 250~pc in Perseus and 260~pc in Serpens. The
1390: observed mean angular deconvolved sizes and axis ratios in each cloud
1391: suggest average power law indices ranging from $p=1.4$ to 1.7 (Y03).
1392:
1393: Sources in Perseus exhibit the largest range in sizes, axis ratios,
1394: and densities, whereas sources in both Serpens and Ophiuchus display a
1395: fairly narrow range of sizes for a large range of masses. We suggest
1396: that this is indicative of a greater variety of physical conditions in
1397: Perseus, supported by the fact that Perseus contains both dense
1398: clusters of millimeter sources and more isolated distributed objects.
1399: A wide range in angular deconvolved sizes may also imply a range in
1400: the power law index of source profiles in Perseus (Y03).
1401:
1402: The slope of the clump mass distribution for both Perseus and
1403: Ophiuchus is $\alpha=2.1$, marginally different than the Serpens slope
1404: of $\alpha=1.6$. Only Perseus and Ophiuchus are consistent within the
1405: substantial errors with the stellar initial mass function ($\alpha
1406: \sim 2.5-2.8$) and with the slope predicted for turbulent
1407: fragmentation ($\alpha=2.3$) by \citet{padoan02}.
1408:
1409:
1410: Turbulent fragmentation simulations by BP06 predict that higher cloud
1411: Mach numbers should result in a large number of low mass cores, and
1412: low Mach numbers in a smaller number of higher mass cores. Given the
1413: measured Mach numbers of $\mathcal{M}=4.9$ in Serpens, 3.6 in Perseus
1414: and 2.3 in Ophiuchus, our observed core mass distribution (CMD) shapes
1415: are inconsistent with the turbulent fragmentation prediction from
1416: BP06. We cannot rule out a turbulent fragmentation scenario, however,
1417: due to uncertainties in the observations and in our assumptions.
1418:
1419: We argue that clustering of 1.1~mm sources remains stronger out to
1420: larger scales in Perseus, based on the slope of the two-point
1421: correlation function (-1.5 in Serpens and Ophiuchus, and -1.2 in
1422: Perseus). This result is supported by the fact that the peak number
1423: of cores per square parsec is larger in Perseus
1424: ($22~\mathrm{pc}^{-2}$) than in Serpens ($12 ~\mathrm{pc}^{-2}$) or
1425: the degraded-resolution Ophiuchus map ($12 ~\mathrm{pc}^{-2}$).
1426:
1427: Finally, we examine relationship between dense cores and the local
1428: cloud column density, as measured by visual extinction (\av).
1429: Extinction thresholds for star formation have been suggested based on
1430: both theory and observation \citep{mckee89, john04}. Although in
1431: general we do not observe a strict \av\ threshold, dense 1.1~mm cores
1432: do tend to be found at high \av: 75\% of cores in Perseus are found
1433: at $\av\gtrsim 8^m$, in Serpens at $\av\gtrsim 15^m$, and in
1434: Ophiuchus at $\av\gtrsim 20-23^m$. Our results confirm that forming
1435: dense cores in molecular clouds is a very inefficient process, with
1436: 1.1~mm cores accounting for less than 10\% of the total cloud mass in
1437: each cloud. This result is consistent with measurements of low star
1438: formation efficiencies of a few percent from studies of the stellar
1439: content of molecular clouds \citep[e.g.][]{evans99}.
1440:
1441: While millimeter-wavelength observations can provide a wealth of
1442: information about the detailed properties of star forming cores as
1443: well as insight into the large scale physical properties of molecular
1444: clouds, they do not tell a complete story. Detecting and
1445: understanding the youngest embedded protostars currently forming
1446: within those cores requires information at mid- to far-infrared
1447: wavelengths. The Bolocam maps for all three clouds presented here
1448: are coordinated to cover the same regions as the c2d \textit{Spitzer}
1449: Legacy IRAC and MIPS maps of Serpens, Perseus, and Ophiuchus.
1450: Combining millimeter and \textit{Spitzer} data for these clouds will
1451: allow us to separate starless cores from cores with embedded
1452: luminosity sources and to better understand the evolution of cores
1453: through the early Class 0 and Class I protostellar phases. Synthesis
1454: and analysis of the combined data set is currently underway (M. Enoch
1455: et al. 2007, in preparation).
1456:
1457: \acknowledgments
1458:
1459:
1460: The authors are grateful to Paul Harvey, Yancy Shirley, and the
1461: anonymous referee for comments that significantly improved this work.
1462: We would like to thank members of the Bolocam team for instrumental
1463: and software support, including James Aguirre, Jack Sayers, Glenn Laurent,
1464: and Sunil Golwala. We also thank the Lorentz Center in Leiden for
1465: hosting several meetings that contributed to this paper. Support for
1466: this work, part of the Spitzer Legacy Science Program, was provided by
1467: NASA through contracts 1224608 and 1230782 issued by the Jet
1468: Propulsion Laboratory, California Institute of Technology, under NASA
1469: contract 1407. Bolocam was built and commissioned under grants
1470: NSF/AST-9618798 and NSF/AST-0098737. KEY was supported by
1471: NASA under Grant NGT5-50401, issued
1472: through the Office of Space Science. Additional support came from
1473: NASA Origins grant NNG04GG24G to NJE and NSF grant AST 02-06158 to JG.
1474: MLE acknowledges support of a Moore Fellowship and an NSF Graduate
1475: Research Fellowship.
1476:
1477:
1478: %%%%%%%%%%%%%% References %%%%%%%%%%%%%%%%%%%%%%
1479:
1480: \begin{thebibliography}{}
1481: \bibitem[Ballesteros-Paredes et al.(2006)]{paredes06}
1482: Ballesteros-Paredes, J., Gazol, A., Kim, J., Klessen, R. S.,
1483: Jappsen, A-K., \& Tejero, E., 2006, \apj, 637, 384
1484: \bibitem[Bohlin et al.(1978)]{bohlin78} Bohlin, R.~C., Savage, B.~D.,
1485: \& Drake, J.~F.\ 1978, \apj, 224, 132
1486: \bibitem[Cambr\'{e}sy(1999)]{cam99}Cambr\'{e}sy, L. 1999, A\&A, 345,
1487: 965
1488: \bibitem[Casali, Eiroa, \& Duncan(1993)]{ced93} Casali, M. M., Eiroa,
1489: C., Duncan, W. D. 1993, A\&A, 275, 195
1490: \bibitem[Chabrier(2003)]{chab03} Chabrier, G. 2003, \pasp, 115, 763
1491: \bibitem[Davis et al.(1999)]{davis99} Davis, C. J., Matthews, H. E.,
1492: Ray, T. P., Dent, W. R. F., \& Richer, J. S. 1999, \mnras, 309,
1493: 141
1494: \bibitem[Di Francesco et al.(2005)]{dif05} Di Francesco, J., Evans, N.J. II,
1495: Caselli, P., Myers, P.C., Shirley, Y., Aikawa, Y., \& Tafalla, M.,
1496: "An Observational Perspective of Low-Mass Dense Cores I: Internal Physical and
1497: Chemical Properties," Protostars and Planets V, 2005
1498: \bibitem[Djupvik et al.(2006)]{djup06} Djupvik, A. A., Andr\'{e}, Ph.,
1499: Bontemps, S., Motte, F., Olofsson, G., G{\aa}lfalk, M., \&
1500: Flor\'{e}n, H.-G. 2006, A\&A, 485, 789
1501: \bibitem[Eiroa \& Casali(1992)]{ec92} Eiroa, C. \& Casali,
1502: M. M.. 1992, A\&A, 262, 468
1503: \bibitem[Enoch et al.(2006)]{enoch06}Enoch, M. L., et al. 2006, \apj,
1504: 638, 293
1505: \bibitem[Evans(1999)]{evans99} Evans, N. J. 1999, \araa, 37, 311
1506: \bibitem[Evans et al.(2001)]{evans01} Evans, N. J., II, Rawlings,
1507: J. M. C., Shirley, Y. L., \& Mundy, L. G. 2001, \apj, 557, 193
1508: \bibitem[Evans et al.(2003)]{eva03}Evans, N. J., II, et al. 2003,
1509: \pasp, 115, 965
1510: \bibitem[Evans et al.(2006)]{deliv} Evans, N. J., II, et al. 2006, Final
1511: Delivery of Data from the c2d Legacy Project: IRAC and MIPS (Pasadena: Caltech),
1512: http://data.spitzer.caltech.edu
1513: \bibitem[Glenn et al.(2003)]{glenn03} Glenn, J., et al. 2003,
1514: Proc. SPIE, 4855, 30
1515: \bibitem[Goodman et al.(2004)]{good04} Goodman, A. A., \& the COMPLETE
1516: Team, 2004, in Star Formation in the Interstellar Medium (San
1517: Francisco: ASP)
1518: \bibitem[Harvey et al.(2006)]{harv06} Harvey, P. M., et al. 2006,
1519: \apj, 644, 307
1520: \bibitem[Harvey et al.(2007)]{harv07} Harvey, P. M., et al. 2007,
1521: \apj, in press
1522: \bibitem[Hatchell et al.(2005)]{hatch05} Hatchell, J., Richer, J. S.,
1523: Fuller, G. A., Qualtrough, C. J., Ladd, E. F., \& Chandler, C. J.
1524: 2005, \aap, 440, 151
1525: \bibitem[Huard et al.(2006)]{huard06} Huard, T. L., et al. 2006, \apj, 640,
1526: 391
1527: \bibitem[Hurt \& Barsony(1996)]{hb96} Hurt, R. L., Barsony, M. 1996,
1528: \apj, 460, L45
1529: \bibitem[Johnstone et al.(2000)]{john00} Johnstone, D., Wilson, C. D.,
1530: Moriarty-Schieven, G., Joncas, G., Smith, G., Gregersen, E., \& Fich,
1531: M. 2000, \apj, 545, 327
1532: \bibitem[Johnstone et al.(2004)]{john04} Johnstone, D., DiFrancesco,
1533: J., \& Kirk, H. 2004, \apjl, 45, 611L
1534: \bibitem[J{\o}rgensen et al.(2006)]{jorg06} J{\o}rgensen, J., et
1535: al. 2006, \apj, submitted
1536: \bibitem[Kirk et al.(2005)]{kirk05} Kirk, J. M., Ward-Thompson, D., \&
1537: Andr\'{e}, P. 2005, \mnras, 360, 1506
1538: \bibitem[Kirk, Johnstone \& Di Francesco(2006)]{kirk06} Kirk, H.,
1539: Johnstone, D., \& Di Francesco, J. 2006, \apj, 646, 1009
1540: \bibitem[Larson(1981)]{larson81} Larson, R. B., 1981, \mnras, 194, 809
1541: \bibitem[Lada et al.(1999)]{lada99} Lada, C. J., Alves, J., \& Lada,
1542: E. A. 1999, in The Physics and Chemistry of the Interstellar Medium,
1543: eds. V. Ossenkopf, J. Stutzki, G. Winnewisser, 161
1544: \bibitem[Laurent et al.(2005)]{laurent05} Laurent, G., et al. 2005,
1545: \apj, 623, 742
1546: \bibitem[Larsson et al.(2000)]{lars00} Larsson, B., et al. 2000 \aap, 363, 253
1547: \bibitem[Leisawitz et al.(1989)]{leis89} Leisawitz, D., Bash, F., \& Thaddeus, P.
1548: 1989, \apjs, 70, 731
1549: \bibitem[Mac Low \& Klessen(2004)]{mlk04} Mac Low, M.-M. \& Klessen, R. S. 2004,
1550: Reviews of Modern Physics, 76, 125
1551: \bibitem[McKee(1989)]{mckee89} McKee, C. F. 1989, \apj, 345, 782
1552: \bibitem[Myers(1998)]{myers98} Myers, P. C. 1998, \apj, 496, L109
1553: \bibitem[Ossenkopf \& Henning(1994)]{oh94} Ossenkopf, V., \& Henning, Th.
1554: 1994, \aap, 291, 943
1555: \bibitem[Padoan \& Nordlund(2002)]{padoan02} Padoan, P., \& Nordlund,
1556: {\AA}.\ 2002, \apj, 576, 870
1557: \bibitem[Padoan et al.(2003)]{pad03} Padoan, P., Goodman, A. A., \& Juvela, M.
1558: 2003, \apj, 588 881
1559: \bibitem[Reid et al.(2002)]{reid02} Reid, I.~N., Gizis, J.~E., \&
1560: Hawley, S.~L.\ 2002, \aj, 124, 2721
1561: \bibitem[Ridge et al.(2006)]{ridge06} Ridge, N. A., et al. 2006, \aj, 131, 2921
1562: \bibitem[Salpeter(1955)]{sal55}Salpeter, E. E. 1955, \apj, 121, 161
1563: \bibitem[Scalo (2005)]{scalo05} Scalo, J., 2005 in The Stellar Initial Mass
1564: Function Fifty Years Later, Kluwer Academic Publishers, ed. E. Corbelli, F. Palla,
1565: and H. Zinnecker, p. 23
1566: \bibitem[Schr{\"o}der \& Pagel(2003)]{schroder03} Schr{\"o}der, K.-P.,
1567: \& Pagel, B.~E.~J. 2003, \mnras, 343, 1231
1568: \bibitem[Shirley et al.(2002)]{shir02} Shirley, Y. L., Evans, N. J., II, \&
1569: Rawlings, J. M. C. 2002, \apj, 575, 337
1570: \bibitem[Shu et al.(1978)]{shu78} Shu, F. H., Adams, F. C., \& Lizano, S.
1571: 1978, \araa, 25, 23
1572: \bibitem[Stanke et al.(2006)]{stanke06} Stanke, T., Smith, M. D.,
1573: Gredel, R., Khanzadyan, T. 2006, A\&A, 447, 609
1574: \bibitem[Straizys et al.(1996)]{stra96} Straizys, V., Cernis, K., \&
1575: Bartasiute, S. 1996, Baltic Astronomy, 5, 125
1576: \bibitem[Testi \& Sargent(1998)]{ts98} Testi, L. \& Sargent,
1577: A. I. 1998, \apj, 508, L91
1578: \bibitem[Weingartner \& Draine(2001)]{wd01} Weingartner, J. C. \& Draine, B. T.
1579: 2001, \apj, 548, 296
1580: \bibitem[Williams et al.(1994)]{will94} Williams, J. P., deGeus, E. J.,
1581: \& Blitz, L. 1994, ApJ, 428, 693
1582: \bibitem[Young et al.(2003)]{young03} Young, C. H., Shirley, Y. L.,
1583: Evans, N. J., II, \& Rawlings, J. M. C. 2003, \apjs, 145, 111
1584: \bibitem[Young et al.(2006)]{young06}Young, K. E., et al. 2006, \apj,
1585: 644, 326
1586: \end{thebibliography}
1587:
1588:
1589: %%%%%%%%%%%%%%%%%% Tables %%%%%%%%%%%%%%%%%%%%%
1590:
1591: \clearpage
1592: \input{tab1}
1593:
1594: \clearpage
1595: % \LongTables % optionally
1596: \begin{landscape}
1597: \input{tab2}
1598: \clearpage
1599: \end{landscape}
1600: %\clearpage
1601: %\thispagestyle{empty}
1602: %\input{tab2}
1603:
1604: \clearpage
1605: \input{tab3}
1606:
1607: \clearpage
1608:
1609: %%%%%%%%%%%%%%%%%%% Figures %%%%%%%%%%%%%%%%%%
1610:
1611: \epsscale{0.65}
1612: \begin{figure}
1613: \plotone{f1.eps}
1614: \caption{Bolocam 1.1~mm (thick line) and Spitzer c2d IRAC (thin line)
1615: and MIPS (gray line) coverage of Serpens overlaid on the \citet{cam99}
1616: visual extinction map. The area observed with IRAC was chosen to cover
1617: $A_V \ge 6^m$ in this portion of the cloud. Our Bolocam survey covers
1618: the same area as the IRAC and a slightly smaller area than the MIPS
1619: observations.\label{c2dfig}}
1620: \end{figure}
1621:
1622: \epsscale{0.9}
1623: \begin{figure}
1624: \notetoeditor{Figure 2 should be full-page.}
1625: \plotone{f2.eps}
1626: \caption{Bolocam 1.1~mm map of 1.5~deg$^2$ (31~pc$^2$ at $d=260$~pc)
1627: in the Serpens molecular cloud. Bolocam has a resolution of
1628: $31\arcsec$, and the map is binned to 10$\arcsec$~pixel$^{-1}$.
1629: The average $1\sigma$ rms noise is 9.5~mJy beam$^{-1}$.
1630: \label{mapfig}}
1631: \end{figure}
1632:
1633: \epsscale{0.9}
1634: \begin{figure}
1635: \plotone{f3.eps}
1636: \caption{Map of the local 1$\sigma$ rms noise per beam in the Serpens
1637: map. Positions of the 35 sources are indicated by white plus
1638: symbols. The average rms noise is 9.5 mJy~beam$^{-1}$, varying by
1639: 18\% across the map. The noise is higher near bright sources due
1640: to sky subtraction residuals. \label{noisefig}}
1641: \end{figure}
1642:
1643: \epsscale{0.95}
1644: \notetoeditor{Figure 4 should be full-page, and color in the printed version.}
1645: \begin{figure}
1646: \plotone{f4.eps}
1647: \caption{Bolocam 1.1~mm map of Serpens with the positions of the 35
1648: sources detected above $5\sigma$ indicated by red circles. Inset
1649: maps magnify the most densely populated source regions, including
1650: the well known northern Cluster A, Cluster B to the south, and an
1651: elongated filament reminiscent of the B1 ridge in Perseus (Paper~I).
1652: Despite the low rms noise level reached (9.5 mJy~beam$^{-1}$), few
1653: sources are seen outside the cluster regions. Many bright 1.1~mm
1654: sources are associated with YSOs \citep{harv06}, and all are
1655: coincident with 160~$\micron$ emission.
1656: \label{sourcefig}}
1657: \end{figure}
1658:
1659: \epsscale{0.65}
1660: \begin{figure}
1661: \notetoeditor{Figure 5 should be full-page if possible.}
1662: \plotone{f5.eps}
1663: \caption{Visual extinction (\av) contours calculated from c2d Spitzer
1664: maps, as described in \S~\ref{masssec}, overlaid on the gray-scale
1665: 1.1~mm map. Contours are $\av= 6,9,12,15,20,25^m$ and are smoothed
1666: to an effective resolution of $2\farcm5$. All bright 1.1~mm cores
1667: are found in regions of high ($>15^m$) extinction, but not all high
1668: \av\ areas are associated with strong millimeter sources.
1669: \label{avfig}}
1670: \end{figure}
1671:
1672: \epsscale{1.}
1673: \begin{figure}
1674: \plotone{f6.eps}
1675: \caption{Distribution of peak (dashed line) and total (solid line)
1676: flux densities of the 35 1.1~mm sources in Serpens. Peak \av\ values
1677: derived from the 1.1~mm peak flux densities using Eq.~\ref{aveq} are
1678: shown on the upper axis. The mean peak flux density of the sample
1679: is $0.5$~Jy~beam$^{-1}$, the mean peak \av\ is 40$^m$, and the
1680: mean total flux density is 1.0~Jy. The $5\sigma$ detection limit
1681: of 0.05~Jy (dotted line) is relatively uniform across the
1682: cloud. \label{fluxfig}}
1683: \end{figure}
1684:
1685: \begin{figure}
1686: \plotone{f7.eps}
1687: \caption{Distribution of source FWHM minor axis (dashed line) and
1688: major axis (solid line) sizes, determined by an elliptical gaussian
1689: fit. The mean FWHM sizes are $49\arcsec$ (minor axis) and
1690: $63\arcsec$ (major axis), and the mean axis ratio (major/minor) is
1691: 1.3. \label{sizefig}}
1692: \end{figure}
1693:
1694:
1695: \begin{figure}
1696: \plotone{f8.eps}
1697: \caption{ Differential mass distribution of the 35 detected 1.1~mm
1698: sources in Serpens for masses calculated with $T_D = 10$~K.
1699: Dotted lines indicate the point source detection limit and the
1700: empirically derived 50\% completeness limit for sources with the
1701: average FWHM size of $55\arcsec$. The best fit power law ($\alpha
1702: = 1.6\pm0.2$) is shown, as well as the best fit lognormal
1703: function. \label{massfn}}
1704: \end{figure}
1705:
1706: \begin{figure}
1707: \plotone{f9.eps}
1708: \caption{Completeness as a function of linear deconvolved source size
1709: in Serpens, Perseus, and Ophiuchus. Symbols show the distribution
1710: of source mass versus deconvolved size in each of the three clouds,
1711: where the size is the linear deconvolved average FWHM. Curves are
1712: empirical 50\% completeness limits determined from Monte Carlo
1713: simulations, and demonstrate the dependence of completeness on
1714: source size and cloud distance.
1715: The beam FWHM of $31''$ corresponds to approximately $8\times 10^3$~AU
1716: in Serpens and Perseus, and $4\times10^3$~AU in Ophiuchus.
1717: Error bars for average sized
1718: sources near the detection limit in each cloud are also shown, as
1719: estimated from the results of Monte Carlo simulations and pointing
1720: uncertainties of approximately $10\arcsec$.
1721: \label{mvscomp}}
1722: \end{figure}
1723:
1724:
1725: \epsscale{0.75}
1726: \begin{figure}
1727: \plotone{f10.eps}
1728: \caption{Comparison of the basic source properties for the original
1729: (black curve) and degraded-resolution, or convolved, (gray hatched
1730: curve) Ophiuchus source samples, which contain 44 and 26 cores
1731: respectively. (\textit{a}) Measured angular deconvolved sizes are
1732: larger in the degraded-resolution map than in the original map by
1733: approximately a factor of two (respective mean values of $98''$ and
1734: $61''$). (\textit{b}) Sources tend to be slightly more elongated in
1735: the degraded-resolution map, with an average axis ratio of
1736: $1.3\pm0.2$ compared to $1.2\pm0.2$ in the original
1737: map. (\textit{c}) The slope of the CMD is not significantly changed
1738: for the degraded-resolution sample, but a number of low mass cores
1739: are blended into a few higher mass sources. (\textit{d}) Larger
1740: deconvolved sizes lead to lower mean densities for the
1741: degraded-resolution sample (median values of
1742: $1.6\times10^5$~cm$^{-3}$ and
1743: $5.8\times10^5$~cm$^{-3}$).}\label{convplots}
1744: \end{figure}
1745:
1746: \epsscale{1.}
1747: \begin{figure}
1748: \plotone{f11.eps}
1749: \caption{Ratio of angular deconvolved size to beam size
1750: ($\theta_{dec}/\theta_{mb}$) for the original and degraded-resolution,
1751: or convolved, Ophiuchus maps. Note that $\theta_{mb}$ is $31''$ for
1752: the original map and $62''$ for the degraded-resolution map.
1753: Measured $\theta_{dec}/\theta_{mb}$ values are similar for the
1754: degraded-resolution (median $\theta_{dec}/\theta_{mb}=1.7$) and
1755: original (median $\theta_{dec}/\theta_{mb}=1.5$) samples, providing
1756: evidence for power law intensity profiles. An intrinsic gaussian or
1757: solid disk intensity profile will result in
1758: $\theta_{dec}/\theta_{mb}$ values in the degraded-resolution map
1759: that are nearly half those in the original map, while a $1/r^2$
1760: intensity profile results in similar values in the
1761: degraded-resolution and original map (0.9 vs
1762: 1.3).}\label{sizetconv}
1763: \end{figure}
1764:
1765:
1766: \begin{figure}
1767: \plottwo{f12a.eps}{f12b.eps}
1768: \caption{\textit{Left}: Comparison of the distribution of deconvolved
1769: linear source sizes in Serpens (solid line), Perseus (shaded), and
1770: Ophiuchus (dotted line). Histograms are plotted as the fraction of
1771: total sources in the cloud as a function of deconvolved size in AU.
1772: Estimates of the minimum resolvable source size (dashed lines) are
1773: based on smearing of the beam by pointing errors. Although measured
1774: angular source sizes in Ophiuchus are similar to those in Perseus and
1775: Serpens, the deconvolved sizes are much smaller due to the different
1776: cloud distances. \textit{Right}: Similar, but sizes are measured as
1777: the angular deconvolved size in units of the beam FWHM. For isolated
1778: sources with power law density profiles, $\theta_{dec}/\theta_{mb}$ is
1779: inversely proportional to the power law index and is independent of
1780: cloud distance \citep{young03}.
1781: \label{dsizecomp}}
1782: \end{figure}
1783:
1784: \begin{figure}
1785: \plotone{f13.eps}
1786: \caption{Comparison of the distribution of axis ratios, where the
1787: ratio is calculated at the half-max contour. Axis ratios $<1.2$
1788: are considered round, and $>1/2$ elongated, based on Monte Carlo
1789: simulations. Sources are primarily round in Ophiuchus and Serpens,
1790: with mean axis ratio of 1.2 and 1.3, respectively. Perseus
1791: exhibits the most elongated sources, with a mean axis ratio of 1.4
1792: and a distribution tail extending up to 2.7. \label{axiscomp}}
1793: \end{figure}
1794:
1795: \epsscale{1.}
1796:
1797: \begin{figure}
1798: \plottwo{f14a.eps}{f14b.eps}
1799: \caption{Comparison of the distributions of core mean densities.
1800: \textit{Left:} The mean density is calculated from the total mass
1801: (Table~\ref{phottab}) and the linear deconvolved source size,
1802: assuming spherical cores. Despite the lower masses of cores in
1803: Ophiuchus compared to Serpens and Perseus, mean densities in
1804: Ophiuchus are higher on average due to small linear deconvolved
1805: sizes (see Figure~\ref{dsizecomp}). The mean densities in the
1806: degraded-resolution Ophiuchus sample are more consistent with Perseus and
1807: Serpens, suggesting cloud distance has a strong effect.
1808: \textit{Right:} Similar, except that the mean density is calculated
1809: using the full-width at $4\sigma$ size rather than of FWHM size.
1810: Differences between the clouds are diminished; thus the mean density
1811: is less dependent on cloud distance when the source size is measured
1812: at the radius where the source merges into the background, rather
1813: than at the half-max.
1814: \label{dencomp}}
1815: \end{figure}
1816:
1817: \begin{figure}
1818: \plotone{f15.eps}
1819: \caption{ Source total mass versus angular deconvolved size (in units
1820: of the beam FWHM) for Serpens, Perseus, and Ophiuchus. The angular
1821: size is less dependent on cloud distance than the linear deconvolved
1822: source size (\S~\ref{convsect}).
1823: Sources in Perseus fill more of the parameter space than those in
1824: Serpens or Ophiuchus, suggesting a wider range of physical conditions
1825: in that cloud.
1826: Empirically derived 50\% completeness limits are also shown (lines).
1827: \label{mvscomp2}}
1828: \end{figure}
1829:
1830: \epsscale{0.9}
1831: \begin{figure}
1832: \plotone{f16.eps}
1833: \caption{ Comparison of the differential core mass distributions
1834: (CMDs) of sources in Serpens, Perseus, and Ophiuchus. The
1835: Ophiuchus curve has been scaled by 1/10, and the Perseus curve by
1836: 1/15 for clarity. Uncertainties reflect $\sqrt{N}$ counting
1837: statistics only. Vertical dotted lines show the 50\% completeness
1838: limits for average sized sources in each cloud (55\arcsec\ FWHM
1839: in Serpens, 68\arcsec\ in Perseus and 69\arcsec\ in Ophiuchus). The
1840: shape of the Perseus and Ophiuchus distributions are quite similar
1841: ($\alpha = 2.1$) in the region where both CMDs are complete. The
1842: shape of the Serpens CMD is marginally different
1843: ($\alpha=1.6$) from the other two clouds. \label{masscomp} }
1844: \end{figure}
1845:
1846: \epsscale{0.8}
1847: \begin{figure}
1848: \plotone{f17.eps}
1849: \caption{ Comparison of the correlation function $w(r)$ for the three
1850: clouds, with power law fits. For each cloud the average source FWHM
1851: size, resolution limit, and linear map size in AU are indicated.
1852: The best fit slope to $w(r) \propto r^n$ is $n=-1.5\pm 0.2$ in
1853: Serpens, $n=-1.18\pm 0.06$ in Perseus, and $n=-1.5\pm 0.1$ in
1854: Ophiuchus. The slope found by \citet{john00} for a smaller map of
1855: Ophiuchus is shallower ($-0.75$). The slope of $w(r)$ derived for
1856: the degraded-resolution Ophiuchus sample (hatched gray curve) is
1857: also shallower ($-1.2\pm0.3$), but the curve appears consistent
1858: with the original sample. Differences in the slope of the
1859: correlation function likely trace differences in clustering
1860: properties of millimeter cores in the clouds. \label{corrncomp}}
1861: \end{figure}
1862:
1863: \epsscale{1.0}
1864: \begin{figure}
1865: \plotone{f18.eps}
1866: \caption{ Cumulative fraction of 1.1~mm cores as a function of cloud
1867: \av\, for sources in Perseus, Serpens, and Ophiuchus. The cloud
1868: \av\ is derived from the reddening of background stars using 2MASS
1869: and IRAC data as described in \S~\ref{ressect}. Thin solid lines
1870: denote the \av\ levels above which 75\% of dense cores are found in each cloud.
1871: More than 75\% of 1.1~mm cores lie above $\av\sim8^m$ in Perseus,
1872: $\av\sim15^m$ in Serpens, and $\av\sim23^m$ in Ophiuchus. Results
1873: for the degraded resolution Ophiuchus map are shown as well (thin
1874: dotted line), but do not vary substantially from the original
1875: Ophiuchus sample, with 75\% of cores above $\av\sim20^m$.
1876: \label{avcomp} }
1877: \end{figure}
1878:
1879: \end{document}
1880: