1: %%
2: %% This is the second Orion paper by Huff & Stahler
3: %%
4: %%
5:
6:
7: \documentclass[12pt,preprint]{aastex}
8:
9:
10: \newcommand{\vdag}{(v)^\dagger}
11: \newcommand{\Msun}{M_\odot}
12:
13: \shorttitle{Cluster Formation}
14: \shortauthors{Huff \& Stahler}
15:
16: \begin{document}
17:
18: \title{Cluster Formation in Contracting Molecular Clouds}
19:
20: \author{E. M. Huff and Steven W. Stahler}
21: \affil{Astronomy Department, University of California,
22: Berkeley, CA 94720}
23:
24: \email{ehuff@astro.berkeley.edu \,\, Sstahler@astro.berkeley.edu}
25:
26: \begin{abstract}
27: We explore, through a simplified, semi-analytic model, the formation of
28: dense clusters containing massive stars. The parent cloud spawning the cluster
29: is represented as an isothermal sphere. This sphere is in near force balance
30: between self-gravity and turbulent pressure. Self-gravity, mediated by
31: turbulent dissipation, drives slow contraction of the cloud, eventually leading
32: to a sharp central spike in density and the onset of dynamical instability. We
33: suggest that, in a real cloud, this transition marks the late and rapid
34: production of massive stars.
35:
36: We also offer an empirical prescription, akin to the Schmidt law, for
37: low-mass star formation in our contracting cloud. Applying this prescription to
38: the Orion Nebula Cluster, we are able to reproduce the accelerating star
39: formation previously inferred from the distribution of member stars in the HR
40: diagram. The cloud turns about 10~percent of its mass into low-mass stars before
41: becoming dynamically unstable. Over a cloud free-fall time, this figure
42: drops to 1~percent, consistent with the overall star formation efficiency of
43: molecular clouds in the Galaxy.
44:
45: \end{abstract}
46:
47: \keywords{open clusters and associations: individual (Orion Nebula Cluster) ---
48: stars: formation --- stars: pre-main-sequence --- ISM: clouds}
49:
50: \section{Introduction}
51:
52: There is growing evidence that the formation of stellar groups is a relatively
53: slow process. More specifically, a star cluster appears within its parent
54: molecular cloud over a period long compared to the cloud's free-fall time,
55: as gauged by the mean gas density. \citet{tan06} have summarized several lines
56: of argument leading to this conclusion. The gas clumps believed to form massive
57: clusters look round, indicating that they are in force balance, and not a
58: state of collapse. Massive clusters themselves have smooth density profiles,
59: again in contrast to a dynamical formation scenario. The observed flux in
60: protostellar outflows indicates a slow accretion rate, and therefore a long
61: star formation time scale. Finally, placement of young clusters in the HR
62: diagram yields age spreads in excess of typical free-fall times (see also
63: Palla \& Stahler 2000).
64:
65: Many researchers have performed direct numerical simulations of molecular
66: clouds; their results also bear on the issue of the star formation time scale.
67: In a typical simulation, the computational volume is filled with a magnetized,
68: self-gravitating gas that has a turbulent velocity field. If the turbulence is
69: only impressed initially, it dies away in a crossing time, and most of the
70: gas condenses into unresolvably small structures \citep[e.g.,][]{k98}. Since
71: the crossing and free-fall times are similar in a molecular cloud, some authors
72: have concluded that all clouds produce stars rapidly, while in a state of
73: collapse \citep{ha01}. Others have used empirical arguments to make the same
74: point \citep{el00}. This view is at odds with the observations concerning
75: cluster-forming clouds cited above. Moreover, the simulations show that, if
76: turbulence is driven throughout the calculation, the rate of star formation
77: can be reduced to a more modest level \citep{mk04}.\footnote{The actual rate of
78: condensation depends on the magnitude of the {\it sonic length}, i.e., the
79: size scale of turbulent eddies whose velocity matches the local sound speed
80: \citep{v03,km05}.} It is plausible that the turbulence is indeed driven by
81: the cloud's self-gravity, a point we shall amplify later.
82:
83: The emerging picture, then, is that molecular clouds both evolve and create
84: internal clusters in a quasi-static fashion. That is, the structure as a
85: whole is nearly in force balance, until it is eventually destroyed by the
86: ionizing radiation and winds from the very stars it spawns. The inferred masses
87: of all clouds larger than dense cores greatly exceed the corresponding Jeans
88: value, evaluated using the gas kinetic temperature. Thus, self-gravity must be
89: opposed by some force beyond the relatively weak thermal pressure gradient. The
90: extra support is generally attributed to MHD waves generated by internal,
91: turbulent motion \citep{es04}. This motion, which is modeled in the numerical
92: simulations just described, imprints itself on molecular line transitions,
93: giving them their observed, superthermal width \citep{am75,f92}.
94:
95: In this paper, we follow the quasi-static contraction of a spherical,
96: cluster-forming cloud. Contraction is facilitated by the turbulent dissipation
97: of energy. This investigation continues and extends an earlier one that was
98: part of our study of the Orion Nebula Cluster (ONC) \citep[Paper I]{hs06}.
99: Here we track in more detail the changing structure of a generic cloud, taken
100: to be in near balance between self-gravity and turbulent pressure. We find that
101: contraction eventually causes the density profile to develop a sharp, central
102: spike. Such a region of growing density is a plausible environment for the
103: birth of massive stars \citep[see][]{sp00}.
104:
105: We also track, using a simple, empirical prescription, the formation of low-mass
106: stars in our contracting cloud. For reasonable parameter values, stellar births
107: occur throughout the cloud over a period of order $10^7$~yr. The global rate of
108: star formation rises with time monotonically, i.e., the formation
109: accelerates. Extended, accelerating production of stars is also found
110: empirically when one analyzes clusters in the HR diagram \citep{ps00}. Indeed,
111: it is not difficult to match specifically the global acceleration documented
112: in the ONC. Here, the star formation rate depends on cloud density in the same
113: manner as the classic Schmidt law.
114:
115: In Section~2 below, we present the basic physical assumptions underlying our
116: model. We also give a convenient, nondimensional scheme. In Section~3, we
117: introduce our treatment of turbulent dissipation, and calculate the interior
118: evolution of the cloud as it contracts toward the high-density state. Section~4
119: offers our prescription for low-mass star formation, and compares the resulting
120: birthrate with the ONC data. Finally, Section~5 discusses the broader
121: implications of our findings, as well as their utility for future work.
122:
123: \section{Formulation of the Problem}
124: \subsection{Physical Assumptions} \label{assumptions}
125:
126: We focus on molecular cloud clumps that are destined to produce the
127: highest-density clusters, i.e., those containing massive stars near their
128: centers. \citet{sh03} used CS observations to study a sample of 63 clumps
129: already containing massive stars, as evidenced by water maser emission. These
130: clouds are nearly round, with median projected axis ratios of 1.2. It is
131: thus a reasonable approximation, and certainly a computationally advantageous
132: one, to take our model cloud to be spherically symmetric.
133:
134: The clumps observed by \citet{sh03} have a median radius of 0.32~pc and mass
135: of $920~\Msun$. A cloud of this size and mass has internal, turbulent motion
136: well in excess of the sound speed, where the latter is based on the typical
137: gas kinetic temperature of 10~K \citep{l81}. This bulk motion excites a
138: spectrum of MHD waves, i.e., perturbations to the interstellar magnetic field
139: threading the cloud \citep{fp86}. Such waves exert an effective pressure that
140: can, at least in principle, support the cloud against global collapse
141: \citep{p90}.
142:
143: \citet{fa93} studied the mechanical forcing due to MHD waves propagating in a
144: one-dimensional, self-gravitating slab. They considered two cases: a slab with
145: an embedded magnetic field oriented parallel to the slab plane, and one
146: with an internal field in the normal direction. In the first case, Fatuzzo
147: \& Adams showed that magnetosonic waves provide a normal force. In the
148: second, it is Alfv\'en waves that exert the force, also in the normal direction.
149: Thus, Fattuzzo \& Adams verified explicitly that the waves counteract gravity,
150: even in the absence of wave damping.
151:
152: \cite{mz95} extended this result. Using the pioneering analysis of \citet{d70},
153: they showed that Alfv\'en waves generated by a turbulent wave field exert an
154: isotropic pressure, regardless of the background geometry. McKee \& Zweibel
155: derived a simple dependence of the wave pressure $P$ on the local density:
156: \begin{equation}
157: P\,\propto\,\rho^{1/2}\,\,.
158: \end{equation}
159: If we ignore the relatively small thermal pressure, then the cloud can be
160: described as an \hbox{$n = -2$} polytrope.
161:
162: Are the structures of real clouds consistent with this polytropic wave pressure?
163: One, indirect, argument indicates they are not. McKee \& Zweibel also
164: demonstrated that $P$ is proportional to $\rho$ times the square of the
165: (randomly oriented) velocity fluctuation $\delta v$. It follows that
166: \begin{equation}
167: \delta v \,\propto\, \rho^{-1/4}
168: \end{equation}
169: in this model. Now the density in an \hbox{$n\,=\,-2$} polytrope tends to
170: approach a power law outside the central plateau, such that $\rho$ is
171: proportional to $r^{-4/3}$. From equation (2), it follows that $\delta v$ is
172: proportional to $r^{1/3}$.
173:
174: Consider the nearly spherical cloud, now gone, that produced the ONC. This cloud
175: was recently driven off by the Trapezium stars, which themselves have ages of
176: about $10^5$~yr \citep{ps01}. The disruption itself occurred well within the
177: cluster crossing time of about $10^6$~yr. Hence, the present-day velocity
178: dispersion of the {\it stars} should reflect the prior $\delta v$ of the
179: {\it gas}. But the dispersion of ONC proper motions has negligible variation
180: from the center to the outskirts of the cluster \citep{jw88}. These
181: measurements span at least a decade in radius, over which $\delta v$ should
182: vary by a factor of 2.2, according to the polytropic relation.
183:
184: Our conclusion, based on this admittedly limited evidence, is that a more
185: realistic model of the internal turbulence has a spatially constant velocity
186: dispersion.\footnote{Inside giant cloud complexes, the observed velocity
187: dispersion increases with the size of the substructure \citep{om02}.
188: Again, we are focusing on a single clump, where such considerations do not
189: apply.} If we further appropriate the relationship between $P$ and
190: $\delta v$ derived by McKee \& Zweibel, we are then positing an isothermal
191: equation of state:
192: \begin{equation}
193: P\,=\,\rho\,a_T^2
194: \end{equation}
195:
196: Here $a_T$, the effective isothermal sound speed, is taken to be a fixed
197: constant at a given instant of time. This same quantity varies temporally;
198: indeed, this latter variation essentially drives the cloud's evolution. We
199: emphasize that $a_T$ does not, as in ordinary gas dynamics, give the
200: magnitude of random, microsopic velocities. Instead, this quantity
201: represents, however crudely, the bulk motion of turbulent eddies; these
202: eddies create the pressure $P$ via MHD waves.
203:
204: Since we are modeling the cloud as an isothermal sphere, we face the familiar
205: difficulty that its mass is infinite unless the configuration is bounded
206: externally. We therefore picture the cloud as being surrounded by a low-density,
207: high-temperature medium with an associated pressure $P_\circ$. This latter
208: quantity is also the pressure at the boundary of our spherical cloud. The cloud
209: density at the boundary, $\rho_\circ$, is found from equation (3), given
210: knowledge of $a_T^2$.
211:
212: \subsection{Nondimensional Scheme} \label{nondim}
213:
214: The mathematical description of a self-gravitating, isothermal cloud in
215: hydrostatic balance is well known \citep[see][Chap.~9]{sp04}. All
216: structural properties follow from the isothermal Lane-Emden equation:
217: \begin{equation}
218: {1\over{\xi^2}}\,{{d{\phantom \xi}}\over{d\xi}}\!
219: \left(\xi^2\,{{d\psi}\over{d\xi}}\right)\,=\,{\rm exp}\,(-\psi)\,\,,
220: \end{equation}
221: with boundary conditions \hbox{$\psi(0)\,=\,\psi^\prime(0)\,=\,0$}. Here,
222: $\psi$ is the dimensionless form of the gravitational potential $\phi_g$:
223: \begin{equation}
224: \psi \,\equiv\, \phi_g/ a_T^2 \,\,.
225: \end{equation}
226: The nondimensional radius $\xi$ is obtained from the dimensional $r$
227: using $G$, $a_T^2$, and the central density $\rho_c$:
228: \begin{equation}
229: \xi\,\equiv\,\left({{4\,\pi\,G\,\rho_c}\over{a_T^2}}\right)^{1/2} r\,\,.
230: \end{equation}
231:
232: Equation (4) was derived using both Poisson's equation and the condition
233: of hydrostatic equilibrium. The latter may be recast as a relation between the
234: density at any radius, $\rho$, its central value, $\rho_c$, and the potential:
235: \begin{equation}
236: \rho \, =\, \rho_c\,{\rm exp}\left(-\psi\right)\,\,.
237: \end{equation}
238: The full, dimensional mass $M_\circ$ follows by integration of $\rho$ over mass
239: shells, Using equation (4) to evaluate the integral, one finds
240: \begin{equation}
241: M_\circ \,=\, {a_T^3\over{\sqrt{4\,\pi\,\rho_c\,G^3}}}
242: \left(\xi^2\,{d\psi\over{d\xi}}\right)_\circ \,\,.
243: \end{equation}
244: Here, the subscript denotes the cloud boundary. Similarly, we shall
245: use $R_\circ$ for the radius of that point, where the internal cloud
246: pressure has fallen to $P_\circ$.
247:
248: In this standard formulation, all nondimensional variables are
249: defined through the basic quantities $a_T^2$, $\rho_c$, and $G$.
250: Although the standard variables will remain useful, the scheme itself
251: is not well suited to describing cloud evolution at fixed mass. For
252: this purpose, we shall also utilize a second nondimensional scheme,
253: based on $M_\circ$, $P_\circ$, and $G$.
254:
255: Let $\lambda$ be the new nondimensional radius, and $\alpha$ the
256: nondimensional effective sound speed. These are defined through
257: \begin{equation}
258: \lambda \,\equiv\,
259: {{P_\circ^{1/4}}\over{G^{1/4}\,M_\circ^{1/2}}}\ r \,\,\,,
260: \end{equation}
261: and
262: \begin{equation}
263: \alpha^2 \,\equiv\,{a_T^2\over{G^{3/4}\,M_\circ^{1/2}
264: \,P_\circ^{1/4}}} \,\,.
265: \end{equation}
266: We further denote as $\delta$ the nondimensional density:
267: \begin{equation}
268: \delta \,\equiv\,{{G^{3/4}\,M_\circ^{1/2}}
269: \over P_\circ^{3/4}}\ \rho \,\,\,.
270: \end{equation}
271: Since we will be discussing temporal evolution, we define a
272: nondimensional time through
273: \begin{equation}
274: \tau \,\equiv\,{{G^{1/8}\,P_\circ^{3/8}}
275: \over M_\circ^{1/4}}\ t \,\,\,.
276: \end{equation}
277:
278: It will be useful to relate new nondimensional quantities to
279: old ones. Thus, equation (6) tells us $\lambda$ as a function
280: of $\xi$:
281: \begin{equation}
282: \lambda \,=\, \sqrt{\alpha^2\over{4\,\pi\,\delta_c}}\ \xi \,\,.
283: \end{equation}
284: For the central density appearing here, $\delta_c$, we use
285: equation (7), evaluated at the cloud boundary:
286: \begin{equation}
287: \delta_c \,=\, {
288: {{\rm exp}\left(\psi_\circ\right)}\over\alpha^2} \,\,.
289: \end{equation}
290: Finally, $\alpha$ itself may be written in terms of standard
291: variables by using equation (8):
292: \begin{equation}
293: \alpha^4 \,=\, \sqrt{4\,\pi}
294: \left(\xi^2\,{{d\psi}\over{d\xi}}\right)_\circ^{-1}
295: {\rm exp}\left(\psi_\circ/2\right) \,\,.
296: \end{equation}
297:
298: \section{Cloud Evolution}
299:
300: \subsection{Internal Structure \label{internal}}
301:
302: We now consider a sequence of isothermal spheres of fixed mass, all embedded in
303: the same external pressure. We may describe each structure using the new,
304: nondimensional variables. The sequence is characterized by a single parameter,
305: the center-to-edge density contrast; we shall denote this ratio as $\beta$.
306: From equation (7), $\beta$ can also be written as
307: \begin{equation}
308: \beta \,=\, {\rm exp}\left(\psi_\circ\right)
309: \end{equation}
310: Since $\psi (\xi)$ is a known function, there is a one-to-one correspondence
311: between our fundamental parameter $\beta$ and $\xi_\circ$, the old,
312: nondimensional radius. The potential $\psi$ increases monotonically with $\xi$,
313: so $\beta$ likewise increases with $\xi_\circ$. The lowest value of $\beta$ is
314: unity, corresponding to \hbox{$\xi_\circ \,=\, 0$}.
315:
316: The internal velocity dispersion $\alpha$ varies along our sequence. We may
317: track this change through equation (15). Thus, for each selected $\beta$, we
318: first find $\psi_\circ$ from equation (16). From knowledge of the function
319: $\psi (\xi)$, we find the corresponding $\xi_\circ$, as well as
320: $\left(d\psi/d\xi\right)_\circ$. Equation (15) then yields $\alpha$.
321:
322: It is equally straightforward to obtain the internal density profile,
323: $\delta (\lambda)$, of any model. Knowing $\psi_\circ$ and $\alpha$,
324: equation (14) gives the central density, $\delta_c$. Proceeding outward,
325: equation (13) gives the value of $\xi$ corresponding to each $\lambda$. Again
326: using $\psi (\xi)$, equation (7) yields the density ratio,
327: \hbox{$\delta/\delta_c \,=\, {\rm exp}(\psi)$}. When we get to the edge,
328: \hbox{$\lambda \,=\, \lambda_\circ$}, we find that
329: \hbox{$\delta_\circ/\delta_c \,\equiv\, \beta^{-1} \,=\,
330: {\rm exp}(-\psi_\circ)$}, in agreement with equation (16).
331:
332: Figure~1 displays graphically the change of the cloud's structure as a function
333: of $\beta$. Here we have plotted the radius, $\lambda (\beta)$, of selected mass
334: shells. As expected, a shell in the deep interior monotonically shrinks. Other
335: shells, however, turn around. With rising $\beta$, an increasing fraction of
336: the cloud mass starts to expand. Such expansion costs energy. Thus,
337: configurations of very high $\beta$ are not physically accessible, as we shall
338: see.
339:
340: \subsection{Enthalpy and Dynamical Stability}
341:
342: The lower dashed, horizontal line in Figure~1 corresponds to a $\beta$-value of
343: 14.1. This is the Bonnor-Ebert density contrast. In the standard analysis,
344: clouds of higher contrast are dynamically unstable \citep{e55,b56}. We recall,
345: however, that this instability arises from perturbations of a cloud {\it held
346: at fixed temperature}. In contrast, our sequence has varying effective sound
347: speed. The Bonner-Ebert contrast no longer marks a stability transition.
348: Nevertheless, this value, which we denote as $\beta_{\rm min}$, is still of
349: interest. It signifies, at least in an approximate way, the point where
350: self-gravity starts to overwhelm external pressure as the main compressive
351: force. Our description of cloud evolution will henceforth focus on such
352: gravity-dominated configurations, i.e., those for which
353: \hbox{$\beta\,>\,\beta_{\rm min}$}.
354:
355: To analyze stability in the present sequence, we first need to invoke
356: thermodynamics. We showed in Paper~I that energy dissipation in an isothermal
357: cloud results in a decrease of the total enthalpy. Returning to dimensional
358: notation, equation (A10) stated
359: \begin{equation}
360: {{dH}\over{dt}} \,=\, -L \,\,,
361: \end{equation}
362: where $L$ is the luminosity. The enthalpy $H$ is the generalization, to a
363: self-gravitating gas, of the classic definition:
364: \begin{equation}
365: H \,\equiv\, E_{\rm therm} \,+\, E_{\rm grav} \,+\, P_\circ\,V \,\,,
366: \end{equation}
367: where $E_{\rm therm}$ and $E_{\rm grav}$ are the thermal and gravitational
368: potential energies, respectively, and $V$ is the cloud volume.
369:
370: To evaluate $E_{\rm therm}$, we employ the general expression for a
371: nonrelativistic gas, $(3/2)\int\!P\,dV$. Using equation (3) for $P$, this
372: integral becomes $(3/2) M_\circ\,a_T^2$. Instead of evaluating $E_{\rm grav}$
373: directly, we invoke the virial theorem, in the form
374: \begin{equation}
375: E_{\rm grav} \,=\, -2\,E_{\rm therm} \,+\,3\,P_\circ\,V \,\,.
376: \end{equation}
377: After expressing the cloud volume in terms of the radius, the enthalpy is
378: \begin{equation}
379: H \,=\, -{3\over 2}\,M_\circ\,a_T^2 \,+\,
380: {{16\,\pi}\over 3}\,P_\circ\,R_\circ^3 \,\,.
381: \end{equation}
382: If we define a nondimensional enthalpy $h$ through
383: \begin{equation}
384: h \,\equiv\, {H \over{G^{3/4}\,M_\circ^{3/2}\,P_\circ^{1/4}}} \,\,,
385: \end{equation}\
386: then equation (20) becomes
387: \begin{equation}
388: h \,=\, -{3\over 2}\,\alpha^2 \,+\,
389: {{16\,\pi}\over 3}\,\lambda_\circ^3 \,\,.
390: \end{equation}
391:
392: Figure~2 shows $h$ along our sequence of clouds. Again, we restrict ourselves
393: to gravity-dominated configurations, for which
394: \hbox{$\beta \,>\, \beta_{\rm min}$}. We also recall that $\beta$ increases
395: monotonically along the sequence. Plotted here against $\alpha$, the enthalpy
396: dips to a minimum, then spirals inward toward a point. The latter corresponds
397: to the singular isothermal sphere. For this special configuration, it may be
398: shown that \hbox{$\alpha^2\,=\,(\pi/2)^{1/4}$} and
399: \hbox{$\lambda_\circ \,=\, ({1/{8\pi}})^{1/4}$}. Thus, the limiting
400: value of $h$ is $-0.187$.
401:
402: However, this limiting value is never reached in the course of evolution. As
403: long as the cloud releases energy into space, so that \hbox{$L\,>\,0$},
404: equation~(17) tells us that the enthalpy declines. Thus, the last accessible
405: configuration coincides with the minimum-enthalpy point in Figure~2.
406: Numerically, we find that \hbox{$h\,=\,-0.50$} for this cloud. The corresponding
407: density contrast $\beta$ is 370.
408:
409: Consider now two configurations with identical values of $h$, very slightly
410: above the minimum. These clouds, like all those in the sequence, have the same
411: mass. We may view them as extremal states attained by the minimum-enthalpy
412: cloud in the course of a normal mode of oscillation. Here, we are assuming that
413: the cloud radiates negligible energy during an oscillation period, so that $h$
414: remains constant. The two endstates are in precise force balance; intermediate
415: states depart only slightly from this condition. In the small-amplitude limit,
416: the oscillation has zero frequency, and the unperturbed, minimum-enthalpy,
417: state represents a stability transition.
418:
419: In summary, an isothermal cloud becomes dynamically unstable at a density
420: contrast $\beta$ of 370, provided the global enthalpy is held fixed
421: during any oscillatory perturbation. This important fact was first discovered by
422: \citet{c03} in the course of a general analysis of isothermal configurations.
423: \footnote{Chavanis finds a slightly higher critical $\beta$ of 390. His
424: minimum enthalpy value, in our units, is \hbox{$h\,=\,-0.493$}.} Note again the
425: marked contrast with the traditional, Bonnor-Ebert result. The much lower
426: critical density contrast in that case (\hbox{$\beta \,=\, 14.1$}) arises
427: because the cloud releases $-$ and draws in $-$ as much energy as necessary to
428: remain isothermal, even during a single oscillation period. This assumption
429: would be inconsistent with our picture that the cloud is quasi-statically
430: contracting due to relatively slow, turbulent dissipation.
431:
432: The minimum-enthalpy state thus marks the natural endpoint of the cloud's
433: evolution. We denote as $\beta_{\rm max}$ the corresponding density contrast,
434: and display this limit as the upper dashed, horizontal line in Figure~1. Clouds
435: with higher density contrast, including the singular isothermal sphere, are
436: inaccessible.
437:
438: \subsection{Turbulent Dissipation} \label{turb}
439:
440: Although we have drawn a number of conclusions regarding the changing structure
441: of our model cloud, we have yet to discuss its temporal evolution. The
442: quasi-static contraction envisioned here is facilitated by the release of
443: energy. This emission must arise at the shock interface between colliding,
444: turbulent eddies. Typical fluid speeds are the virial value, i.e., less than
445: 10~km~s$^{-1}$ for the clouds of interest. Hence, the shocks radiate through
446: far-infrared and submillimeter photons from low-lying transitions of molecules.
447: The cloud is optically thin to such photons. The luminosity $L$ in
448: equation~(17) is thus generated from the full interior.
449:
450: Consider, then, a representative volume of the cloud. The numerical simulations
451: mentioned previously have modeled the dynamics of a magnetized gas subject to
452: an impressed, turbulent velocity field. Even if the fluid disturbances begin as
453: incompressible Alfv\'en waves, efficient mode conversion produces compressible
454: MHD waves that steepen and shock \citep[see, e.g.,][]{g78}. Signficant energy is
455: dissipated during the characteristic crossing time of the largest eddies. For
456: example, \citet{m99} found that
457: \begin{equation}
458: {\dot\epsilon} \,=\, -\eta\,{V^3_{\rm turb}\over\lambda} \,\,.
459: \end{equation}
460: Here $\dot\epsilon$ is the energy loss rate per unit mass of gas, $V_{\rm turb}$
461: the average (rms) eddy speed, and $\lambda$ the dominate wavelength of the
462: impressed turbulence. The empirical constant $\eta$ was measured by Mac~Low to
463: be about 0.4.
464:
465: In the simulations, turbulence is impressed on an arbitrary scale. Indeed, the
466: question of what drives the turbulence remains controversial \citep{m04}. Here,
467: we recall the key fact that the mean, interior velocities match the virial value
468: over a large range of cloud sizes and masses \citep{l81}. It is likely,
469: therefore, that self-gravity constitutes the basic driving mechanism, although
470: a quantitative model is still lacking. If this basic idea is correct, then the
471: ``dominant wavelength" in equation~(23) should be comparable to the cloud size.
472: We therefore adopt, as our expression for the cloud luminosity $L$, a
473: mass-integrated version of this relation:
474: \begin{equation}
475: L \,=\, \eta\,{{M_\circ\,a_T^3}\over {4\,R_\circ}} \,\,,
476: \end{equation}
477: where the prefactor $\eta$ does not necessarily have the value found by
478: Mac~Low. The factor of 4 in the denominator reflects the fact that the
479: largest mode corresponds to overall expansion or contraction of the cloud
480: \citep{ma02}.
481:
482: \subsection{Quasi-Static Contraction}
483:
484: To follow the cloud evolution in time, we use our prescribed luminosity to
485: alter the global enthalpy. Combining equations (17), (20), and (24), we
486: recast the result into nondimensional terms:
487: \begin{equation}
488: -\eta\,{\alpha^3\over\lambda_\circ}\,=\,
489: -{6}\,{{d\alpha^2}\over{d\tau}} \,+\,
490: 64\,\pi\lambda_\circ^2\,{{d\lambda_\circ}\over{d\tau}} \,\,.
491: \end{equation}
492:
493: The dependent variables $\alpha$ and $\lambda_\circ$ are already known
494: implicitly in terms of $\xi_\circ$. (Recall equations (13) $-$ (15).) We may
495: thus regard equation~(25) as giving the dependence of $\tau$ on this same
496: quantity. Since $\eta$ is still unknown, we use instead the combination
497: $\eta\tau$:
498: \begin{equation}
499: {{d(\eta\tau)}\over{d\xi_\circ}} \,=\,
500: {{6\,\lambda_\circ}\over{\alpha^3}}\,{{d\alpha^2}\over{d\xi_\circ}} \,-\,
501: {{64\,\pi\,\lambda_\circ^3}\over{\alpha^3}}\,
502: {{d\lambda_\circ}\over{d\xi_\circ}}\,\,.
503: \end{equation}
504: We integrate this equation numerically, setting \hbox{$\eta\tau\,=\,0$} at
505: \hbox{$\xi_\circ\,=\,6.5$}, the value at the Bonnor-Ebert density contrast,
506: and ending at \hbox{$\xi_\circ\,=\,25$}, the minimum-enthalpy state. Over this
507: evolutionary span, $\eta\tau$ increases by 0.96.
508:
509: Figure~3 shows that neither the effective sound speed nor the cloud radius
510: vary greatly during this period. The former increases monotonically, with a
511: fractional change of 10~percent by the end. The radius gently decreases most of
512: the time. Just before the unstable state is reached, the cloud surface begins to
513: swell, in agreement with Figure~1.
514:
515: The temporal change of the central density is much more dramatic. As seen in
516: Figure~4, $\delta_c$ increases slowly at first, and then accelerates strongly
517: at the end. Given the behavior of the cloud radius, this rapid compression
518: evidently involves a small fraction of the total cloud volume. The left panel
519: of Figure~5 displays the evolution of the full density profile. From bottom to
520: top, the associated values of $\eta\tau$ are 0, 0.15, 0.23, and the final
521: value, \hbox{$\eta\tau \,=\, 0.24$}.
522:
523: Because the inner portion of the cloud undergoes such rapid compression, one
524: may question the basic assumption of quasi-static behavior. Are all mass shells
525: really moving at subsonic velocity? At the very center, the velocity must fall
526: to zero at each time. But what of shells just outside the center?
527:
528: Figure~6 shows the velocity profiles for the final three times depicted in the
529: left panel of Figure~5. Since the temporal variable in equation~(26) is
530: $\eta\tau$, we plot the ratio $v/\eta\alpha$. For \hbox{$\eta\tau \,=\,0.60$}
531: and 0.92, the velocity is subsonic throughout, even if $\eta$ were unity. In
532: fact, plausible $\eta$-values are less than unity, as indicated earlier; we
533: shall see in Section~4 below that \hbox{$\eta\,\approx\,0.3$} best matches the
534: ONC data. In the final profile, corresponding to the minimum-enthalpy state,
535: the peak velocities occur in the deep interior, and are mildly supersonic
536: (\hbox{$v\,\gtrsim\,\alpha$}).
537:
538: \section{Modeling the ONC}
539:
540: \subsection{Prescription for Star Formation}
541:
542: At present, we have scant knowledge of how the birthrate of stars scales with
543: the properties of the cloud medium spawning these objects. Within the solar
544: neighborhood, \citet{s59} found the star formation rate to be proportional to
545: the square of the local density; here, the rate is measure per unit volume.
546: Schmidt's law thus states that the formation rate per gas mass scales
547: linearly with the cloud density.\footnote{In Schmidt's original formulation,
548: the gas in question was HI. We now know, of course, that the relevant clouds
549: are molecular.}
550:
551: On the scale of galactic disks, it is established that the formation rate per
552: unit disk area rises as $\Sigma^n$, where $\Sigma$ is the total gas surface
553: density, and the exponent $n$ is about 1.4 \citep{ke98}. However, it is not
554: straighforward to relate this important finding to the present study. Each
555: areal patch in the galactic observations comprises numerous molecular
556: complexes, any one of which is far larger than the clumps of direct interest
557: here.
558:
559: Returning to our model, if the clump indeed undergoes slow contraction, then
560: its overall star formation rate must increase with time. It is plausible that
561: the local rate within each mass shell rises with that shell's density $\rho$.
562: Following Schmidt, we posit a power-law dependence:
563: \begin{equation}
564: {\dot m}_\ast \,=\, {\epsilon\over t_1}\,\left({\rho\over\rho_1}\right)^n
565: \,\, .
566: \end{equation}
567: Here, ${\dot m}_\ast$ is the mass in stars forming per unit time, per unit
568: cloud mass. The fiducial density and time, $\rho_1$ and $t_1$, are those from
569: equations (11) and (12), respectively:
570: \begin{mathletters}
571: \begin{eqnarray}
572: \rho_1 \, & \equiv & \, {P_\circ^{3/4}\over{G^{3/4}\,M_\circ^{1/2}}} \\
573: t_1 \, & \equiv & \, {M_\circ^{1/4}\over{G^{1/8}\,P_\circ^{3/8}}} \,\,.
574: \end{eqnarray}
575: \end{mathletters}
576: Finally, the exponent $n$ is to be set by matching to observations. We stress
577: that the prescription in equation (27) applies to low-mass stars. The formation
578: of massive objects is a separate phenomenon. In our model, this occurs only in
579: the high-density, central region of the final, minimum-enthalpy state.
580:
581: Equation~(27) contains a nondimensional efficiency factor $\epsilon$. Since
582: only a fraction of the cloud mass turns into stars, we expect $\epsilon$ to be
583: well under unity. If, for simplicity, we assume this parameter to be the same
584: in all mass shells, then integration of equation~(27) yields the total mass per
585: unit time in new stars:
586: \begin{equation}
587: {\dot M_\ast} \,=\, {{4\,\pi\,\epsilon}\over{t_1\,\rho_1^n}}
588: \int_0^{R_\circ} \! \rho^{n+1}\,r^2\,dr \,\,.
589: \end{equation}
590: We may conveniently recast this fornula in terms of the effective sound speed:
591: \begin{equation}
592: {\dot M_\ast} \,=\, \epsilon\,{a_T^3\over G}\ {\cal I} \,\,.
593: \end{equation}
594: Here, the nondimensional quantity ${\cal I}$ is expressed using the
595: traditional, polytropic variables:
596: \begin{equation}
597: {\cal I} \,\equiv\, {{\alpha^{1-2n}}\over{\sqrt{4\pi}}}
598: {\rm e}^{(n-1/2)\psi_\circ}
599: \int_0^{\xi_\circ}\!{\rm e}^{-(n+1)\psi}\,\xi^2\,d\xi \,\,.
600: \end{equation}
601:
602: Our prescription gives a finite star formation rate for clouds of arbitrarily
603: low density. This is clearly an oversimplification. There is no evidence, for
604: example, that HI clouds form stars at all. Even within the molecular domain,
605: it may be that stars form only above some threshold density. In their study of
606: the Rosette cloud complex, \citet{w95} found that only clumps which are
607: strongly self-gravitating (as assessed by comparison of velocity dispersions,
608: masses, and sizes) have internal stars. While a more complete model should
609: account for this threshold effect, we shall not include it explicitly, but
610: simply limit our discussion to self-gravitating clumps
611: (\hbox{$\beta\,>\,\beta_{\rm min}$}) that are capable of forming stars.
612:
613: \subsection{Comparison with Observations}
614:
615: In Paper I, we empirically determined the star formation history of the ONC. The
616: database of \citet{h97}, together with theoretical pre-main-sequence tracks
617: \citep{ps99}, allowed us to assign masses and contraction ages.\footnote
618: {Recently, \citet{j07} has redetermined the ONC distance as 390~pc, rather
619: than the 470~pc used by \citet{h97}. If correct, this distance shift will
620: systematically lower stellar luminosities, and therefore increase their ages.}
621: In the detailed analysis, we restricted our attention to the 244 members with
622: \hbox{$M_\ast > 0.4\ \Msun$}; the Trapezium stars themselves have
623: \hbox{$M_\ast \geq 7\ \Msun$}, and are thus already on the main sequence.
624: This sample is statistically complete, in the sense that the oldest stars do
625: not fall below the observational sensitivity limit.
626:
627: Based on our results from Paper~I, Figure~7 shows ${\dot M}_\ast(t)$, the mass
628: production per unit time, as a function of stellar age. Here, we have binned
629: the data in age intervals of $10^6$~yr. We have also extrapolated from our
630: subsample to all stellar masses. We did so by multiplying the accumulated mass
631: at each epoch by a factor of 1.3. This factor accounts for the missing stars
632: with \hbox{$M_\ast\,<\,0.4\ \Msun$}, according to the field-star initial mass
633: function of \citet{s98}.
634:
635: In order to use equation (30) to describe the ONC, we need the effective sound
636: speed $a_T$ as a function of time. Our numerical model gives the nondimensional
637: functional relation $\alpha (\eta\tau)$ (recall Fig. 3). Similarly, the
638: quantity ${\cal I}$ contains $\psi_\circ$ and $\xi_\circ$, which we also know
639: as functions of $\eta\tau$. Converting these relations to dimensional form
640: requires that we set the cloud mass $M_\circ$ and background pressure $P_\circ$.
641:
642: We now make the critical assumption that the parent cloud of the ONC was, just
643: prior to its dispersal, in the minimum-enthalpy state. Then equations (9) and
644: (10) may be combined to yield
645: \begin{equation}
646: M_\circ \,=\, f_1\, {{R_\circ\,a_T^2}\over G}
647: \end{equation}
648: and
649: \begin{equation}
650: P_\circ \,=\, f_2\, {{G\,M_\circ^2}\over R_\circ^4} \,\,,
651: \end{equation}
652: where $R_\circ$ and $a_T$ refer to the final cloud state. The results of our
653: numerical integration give \hbox{$f_1 \,=\,2.0$} and \hbox{$f_2 \,=\,0.028$}. We
654: take $R_\circ$ to equal the radius of the stellar cluster (2.5~pc; see
655: Hillenbrand 1997), and identify $a_T$ with the observed {\it stellar} velocity
656: dispersion (2.4~km~s$^{-1}$; see Jones \& Walker 1988). We then find that
657: \hbox{$M_\circ \,=\,6900\ \Msun$} and
658: \hbox{$P_\circ\,=\,1.1\times 10^{-10}\ {\rm dyne~cm}^{-2}$}. The latter is
659: about 300 times the canonical value in the diffuse interstellar medium, i.e.,
660: that bounding HI clouds \citep{wo95}.
661:
662: It remains only to adjust $\epsilon$, $\eta$, and the exponent $n$, until the
663: theoretical star formation rate ${\dot M}_\ast$, as given by equation~(30),
664: matches the empirical one. We use a standard implementation of the
665: Levenberg-Marquardt fitting algorithm \citep[ \S~15.5]{num88}. The likelihood
666: function that we maximize incorporates a uniform error in the star formation
667: rate at each epoch of $\sigma\,=\,3\ \Msun\ {\rm Myr}^{-1}$. Here we have
668: used \hbox{$\sigma \,\approx\,\sqrt{N}\langle M\rangle$}, where $N$ is the
669: median number of stars produced per $10^6$~yr, and \hbox{$\langle M\rangle$}
670: is the average stellar mass at the appropriate epoch. This formula assumes that
671: the number of stars in each bin is Poisson distributed about the mean predicted
672: by the model, i.e., we neglect the observational contribution to the error.
673:
674: The dashed curve in Figure~7 shows our theoretical rate as a function of
675: stellar age, along with the optimal values for the three parameters. As
676: predicted, $\epsilon$ is small ($2\times 10^{-4}$), signifying a low efficiency
677: for stellar production. Specifically, the cloud converts 8~percent of its mass
678: ($550~\Msun$) into low-mass stars. \footnote{This figure is higher if many of
679: the embedded, near-infrared sources seen behind the ONC are members of the
680: original cluster \citep{ad95,hc00}.} The parameter $\eta$ is also small (0.3),
681: indicating that the cloud contracts over an interval long compared with the
682: free-fall time, $t_{\rm ff}$. The latter is 1.6~Myr for our initial cloud
683: state. Finally, the best-fit value of $n$ (1.4) lies close to unity. Thus, the
684: original star formation law of Schmidt may hold quite generally. The 1-sigma
685: errors in $\epsilon$, $\eta$, and $n$, are, respectively, $1\times 10^{-4}$,
686: $0.08$, and $0.09$. These figures would have been larger had we included
687: observational sources of error.
688:
689: We note also that the basic physical characteristics of our model clouds are
690: consistent with the clump properties inferred from observations. Our fiducial
691: density $\rho_\circ$ is equivalent to a molecular hydrogen number density of
692: 520~cm$^{-3}$. As can be seen in Figure~5, the average interior density is
693: higher by an order of magnitude. The mean visual extinction of a cloud, using
694: $M_\circ/\pi\,R_\circ^2$ as the typical column density, is 11~mag. These
695: figures are in general accord with the findings of \citet{w95} for
696: self-gravitating clumps in the Rosette complex. Moreover, $A_v$, measured
697: inward from the edge, quickly exceeds unity in all our models. Thus, the
698: hydrogen is indeed in molecular form throughout the bulk of the interior, as
699: is appropriate for a star-forming cloud.
700:
701: \section{Discussion}
702:
703: In this paper we have adopted a conceptually simple model of cloud contraction
704: and stellar group formation. The cloud is a self-gravitating sphere, supported
705: against collapse by the motion of turbulent eddies. We assigned an effective
706: pressure to this motion; our formulation implicitly assumes the eddy speed to
707: be spatially constant, though varying in time. Cloud evolution is mediated by
708: the slow leakage of energy, presumed to occur through internal, shock
709: dissipation.
710:
711: This model, supplemented by a Schmidt-type prescription for the star formation,
712: can account not only for the empirically known history of the ONC, but also
713: for more general characteristics of stellar birth. Consider, for example, the
714: issue of formation efficiency. In the spirit of \citet{km05}, we may define a
715: nondimensional star formation rate per free-fall time:
716: \begin{equation}
717: \epsilon_{\rm ff} \,\equiv\,
718: {t_{\rm ff}\over t_{\rm ev}} \,
719: {{\Delta M_\ast}\over M_\circ} \,\,.
720: \end{equation}
721: Here, $t_{\rm ev}$ is the time over which the cloud produces stars, while
722: $\Delta M_\ast$ is the total mass in these objects. Using our ONC model, and
723: setting \hbox{$t_{\rm ev}\,=\,1\times 10^7~{\rm yr}$}, we find
724: \hbox{$\epsilon_{\rm ff} \,=\, 0.014$}. \citet{ze74} long ago pointed out that
725: only about 1~percent of the Galaxy's molecular gas can become stars in a cloud
726: free-fall time, to reproduce the observed, global star formation rate. The
727: agreement here suggests that giant complexes create stars principally through
728: their slowly contracting, internal clumps, as we have modeled.
729:
730: We have not described, in any quantitative way, the physics underlying the
731: turbulent dissipation. Our best-fit value of $\eta$ for the ONC is similar to
732: that found in numerical simulations of turbulent clouds \citep{m99}. However,
733: our physical picture is quite different. All simulations to date, which focus on
734: an isolated, interior volume, find the turbulent energy dying away. In our
735: model, the mean turbulent speed increases with time (see Figure~3). Future,
736: global simlations of self-gravitating clouds supprted by turbulent pressure
737: should show this effect.
738:
739: Our best-fit value of $n$ agrees, perhaps fortuitously, with that originally
740: proposed by \citet{s59}. For an $n$-value of unity, the star formation rate per
741: cloud mass scales with the gas density. Did this proportionality really hold
742: in the ONC? The righthand panel of Figure~5 suggests that it did, at least
743: roughly. Here, the solid curve is the density profile of our final,
744: minimum-enthalpy, cloud model. The dot-dashed curve is the current number
745: density of ONC stars, as reconstructed from the observed, projected number
746: density (see Fig.~3 of Paper~I). The similarity of the two curves indeed
747: suggests that stars trace the mass distribution of the parent cloud. The same
748: point is evident when comparing the projected stellar density with CO
749: contours of the remnant gas \citep[see][Fig.~12.27]{sp04}.
750:
751: Of course, all stars travel some distance from their birth sites. They do not
752: move ballistically, but are subject to the gravitational potential of the
753: parent cloud. Because of the star formation law expressed in equation~(27),
754: stellar births are indeed concentrated toward the cloud center, but there will
755: inevitably be some outward diffusion. In a future paper, we hope to track this
756: process through a direct, numerical simulation.
757:
758: The minimum-enthalpy cloud that terminates our dynamical sequence is dynamically
759: unstable. How do we interpret this instability in a more realistic setting? The
760: essential fact is that the self-gravity of the gas becomes so strong that it
761: leads to rapid, internal contraction, perhaps even true collapse of the central
762: region. It is tempting to link this event with the formation of high-mass
763: stars. While the physics of massive star formation is far from clear, the
764: collapse or coalescence of dense, gaseous structures appears to play a key
765: role \citep{sp00,mt03}. In the specific case of the ONC, the Trapezium stars
766: are, of course, centrally located, and appear to be of relatively recent
767: origin \citep{ps01}.
768:
769: In comparing our model with data from the ONC, we have accepted at face value
770: the stellar ages inferred from the placement of each object in the HR diagram.
771: The age spread within clusters remains a contentious issue. For the ONC,
772: \citet{p05} have found that four nominally older stars are depleted in lithium,
773: as would be expected. Such findings are inconsistent with the assertion by
774: \citet{h01} that the ostensible age spread primarily reflects observational
775: uncertainties.
776:
777: \citet{pk07} accept the higher ages, but hypothesize that all such stars were
778: gravitationally captured from somewhat older, neighboring clusters. As the
779: authors themselves note, the existence of such neighboring systems is unclear.
780: The other subassociations within the Orion complex are too young. One
781: possibility is that there were a large number of nearby small groups producing
782: low-mass stars and then dispersing. Pending more direct evidence for such
783: groups, we continue to believe that the nominal age spreads in both the ONC
784: and other systems are real.
785:
786: In the present model, we have taken the cloud to be of fixed mass. This
787: assumption may be acceptable for the ONC progenitor cloud, at least until
788: the point when the Trapezium stars ionized and drove off the gas. In clouds
789: producing low-mass T associations, the latter process does not occur. Yet these
790: clouds are still dispersed, presumably through stellar winds. In our next
791: paper, we will generalize our model of cloud contraction to include the effect
792: of continuous mass loss. We will thus achieve a fuller picture of stellar
793: group formation, a process of importance not only locally, but on galactic
794: scales.
795:
796:
797: \acknowledgments
798:
799: We are grateful to Steve Shore for pointing out the work of P. Chavanis on
800: the thermodynamics of self-gravitating spheres. This project was supported by
801: NSF grant AST-0639743.
802:
803:
804: \begin{thebibliography}{}
805: \bibitem[Ali \& Depoy(1995)]{ad95} Ali, B. \& Depoy, D. L. 1995, \aj, 109, 709
806: \bibitem[Arons \& Max(1975)]{am75} Arons, J. \& Max, C. E. 1975, \apjl, 196,
807: L77
808: \bibitem[Bonnor(1956)]{b56} Bonnor, W. B. 1956, \mnras, 116, 351
809: \bibitem[Chavanis(2003)]{c03} Chavanis, P. H. 2003, \aap, 401, 15
810: \bibitem[Dewar(1970)]{d70} Dewar, R. L. 1970, Phys.Fluids, 13, 2710
811: \bibitem[Ebert(1955)]{e55} Ebert, R. 1955, \zap, 37, 322
812: \bibitem[Elmegreen(2000)]{el00} Elmegreen, B. G. 2000, \apj, 530, 277
813: \bibitem[Elmegreen \& Scalo(2004)]{es04} Elmegreen, B. G. \& Scalo, J. 2004,
814: \araa, 42, 211
815: \bibitem[Falgarone \& Puget(1986)]{fp86} Falgarone, E. \& Puget, J. L. 1986,
816: \aap, 162, 235
817: \bibitem[Falgarone et al.(1992)]{f92} Falgarone, E., Puget, J.-L., \&
818: P\'erault, M. 1992, \aap, 257, 730
819: \bibitem[Fatuzzo \& Adams(1993)]{fa93} Fatuzzo, M. \& Adams, F. C. 1993,
820: \apj, 412, 146
821: \bibitem[Goldstein(1978)]{g78} Goldstein, M. L. 1978, \apj, 219, 700
822: \bibitem[Hartmann(2001)]{h01} Hartmann, L. 2001, AJ, 121, 1030
823: \bibitem[Hartmann et al.(2001)]{ha01} Hartmann, L., Ballesteros-Paredes, J., \&
824: Bergin, E. A. 2001, \apj, 562, 852
825: \bibitem[Hillenbrand(1997)]{h97} Hillenbrand, L. A. 1997, \aj, 113, 1733
826: \bibitem[Hillenbrand \& Carpenter(2000)]{hc00} Hillenbrand, L. A. \&
827: Carpenter, J. M. 2000, \apj, 540, 236
828: \bibitem[Huff \& Stahler(2006)]{hs06} Huff, E. M. \& Stahler, S. W. 2006,
829: \apj, 644, 355
830: \bibitem[Jeffries(2007)]{j07} Jeffries, R. D. 2007, astro-ph/0701186
831: \bibitem[Jones \& Walker(1988)]{jw88} Jones, B. F. \& Walker, M. F. 1988, \aj,
832: 95, 1755
833: \bibitem[Kennicutt(1998)]{ke98} Kennicutt, R. C. 1998, \apj, 498, 541
834: \bibitem[Klessen et al.(1998)]{k98} Klessen, R. S., Burkert, A., \&
835: Bate, M. R. 1998, \apjl, 501, L205
836: \bibitem[Krumholz \& McKee(2005)]{km05} Krumholz, M. R. \& McKee, C. F. 2005,
837: \apj, 630, 250
838: \bibitem[Larson(1981)]{l81} Larson, R. B. 1981, \mnras, 194, 809
839: \bibitem[Mac~Low(1999)]{m99} Mac~Low, M.-M. 1999, \apj, 524, 169
840: \bibitem[Mac~Low(2004)]{m04} Mac~Low, M.-M. 2004, \apss, 289, 323
841: \bibitem[Mac~Low \& Klessen(2004)]{mk04} Mac~Low, M.-M. \& Klessen, R. S.
842: 2004, Rev. Mod. Phys., 76, 125
843: \bibitem[Matzner(2002)]{ma02} Matzner, C. D. 2002, \apj, 566, 302
844: \bibitem[McKee \& Tan(2003)]{mt03} McKee, C. F. \& Tan, J. C. 2003, \apj,
845: 585, 859
846: \bibitem[McKee \& Zweibel(1995)]{mz95} McKee, C. F. \& Zweibel, E. G. 1995,
847: \apj, 440, 686
848: \bibitem[Ossenkopf \& Mac~Low(2002)]{om02} Ossenkopf, V. \& Mac~Low, M.-M.,
849: 2002, \aap, 390, 307
850: \bibitem[Palla \& Stahler(1999)]{ps99} Palla, F. \& Stahler, S. W. 1999,
851: \apj, 525, 772
852: \bibitem[Palla \& Stahler(2000)]{ps00} Palla, F. \& Stahler, S. W. 2000,
853: \apj, 540, 255
854: \bibitem[Palla \& Stahler(2001)]{ps01} Palla, F. \& Stahler, S. W. 2001,
855: \apj, 553, 299
856: \bibitem[Palla et al.(2005)]{p05} Palla, F., Randich, S., Flaccomio, E., \&
857: Pallavicini, R. 2005, \apjl, 626, L49
858: \bibitem[Pflamm-Altenburg \& Kroupa(2007)]{pk07} Pflamm-Altenburg, J. \&
859: Kroupa, P. 2007, \mnras, 375, 855
860: \bibitem[Press et al.(1988)]{num88} Press, W. H., Teukolsky, S. A., Vetterling,
861: W. T., \& Flannery, B. P. 1988, Numerical Recipes, Cambridge U. Press
862: \bibitem[Pudritz(1990)]{p90} Pudritz, R. E. 1990, \apj, 350, 195
863: \bibitem[Scalo(1998)]{s98} Scalo, J. 1998, in The Stellar Initial Mass Function,
864: ed. G. Gilmore, I. Parry, \& S. Ryan, Cambridge: Cambridge U. Press,
865: 32
866: \bibitem[Schmidt(1959)]{s59} Schmidt, M. 1959, \apj, 129, 243
867: \bibitem[Shirley et al.(2003)]{sh03} Shirley, Y. L., Evans, N. J., Young,
868: K. E., Knez, C., \& Jaffe, D. T. 2003, \apjs, 149, 375
869: \bibitem[Stahler \& Palla(2004)]{sp04} Stahler, S. W. \& Palla, F.
870: 2004, The Formation of Stars, Wiley-VCH
871: \bibitem[Stahler et al.(2000)]{sp00} Stahler, S. W., Palla, F., \& Ho, P. T. P.
872: 2000, in Protostars and Planets IV, ed. V. Mannings, A. P. Boss, \&
873: S. S. Russell, Tucson: U. of Arizona Press, 327
874: \bibitem[Tan et al.(2006)]{tan06} Tan, J. C., Krumholz, M. R., \&
875: McKee, C. F. 2006, \apj, 641, L121
876: \bibitem[V\'azquez-Semadeni et al.(2003)]{v03} V\'azquez-Semadeni, E.,
877: Ballesteros-Paredes, J., \& Klessen, R. S. 2003, \apjl, 585, L131
878: \bibitem[Williams et al.(1995)]{w95} Williams, J. P., Blitz, L., \& Stark,
879: A. A. 1995, \apj, 451, 252
880: \bibitem[Wolfire et al.(1995)]{wo95} Wolfire, M. G., Hollenbach, D., McKee,
881: C. F., Tielens, A.G.G.M., \& Bakes, E.L.O. 1995, \apj, 443, 152
882: \bibitem[Zuckerman \& Evans(1974)]{ze74} Zuckerman, B. \& Evans, N.J. 1974,
883: \apjl, 192, L149
884: \end{thebibliography}
885:
886: \clearpage
887:
888:
889: \begin{figure}
890: \epsscale{.80}
891: \plotone{f1.eps}
892: \caption{Evolution of the cloud's internal structure. Shown are the radii of
893: selected, Lagrangian mass shells as a function of the density contrast $\beta$.
894: From left to right, the five shells enclose 0.2, 0.4, 0.6, 0.8, and 1.0 times
895: the total cloud mass. Also indicated are the minimum $\beta$-value for
896: self-gravitating clouds and the maximum value for dynamical stability.}
897: \end{figure}
898: \clearpage
899:
900: \begin{figure}
901: \plotone{f2.eps}
902: \caption{The run of specific enthalpy $h$ along the cloud sequence. This
903: quantity is plotted as a function of the effective sound speed $\alpha$. The
904: dotted portion of the curve pertains to clouds which have too low a density
905: contrast to be self-gravitating.}
906: \end{figure}
907: \clearpage
908:
909: \begin{figure}
910: \plotone{f3.eps}
911: \caption{Evolution of the cloud radius ({\it solid curve}) and the effective
912: sound speed ({\it dashed curve}). Notice that the time coordinate $\eta\tau$
913: starts at the first self-gravitating configuration, i.e., that for which
914: \hbox{$\beta \,=\, \beta_{\rm min}$}.}
915: \end{figure}
916: \clearpage
917:
918: \begin{figure}
919: \plotone{f4.eps}
920: \caption{Evolution of the nondimensional central density. As in Figure~3, the
921: time $\eta\tau$ is measured from the first self-gravitating cloud.}
922: \end{figure}
923: \clearpage
924:
925: \begin{figure}
926: \plotone{f5.eps}
927: \caption{{\it Left panel:} Evolution of the density as a function of radius.
928: From bottom to top, the corresponding values of $\eta\tau$ are 0, 0.60, 0.92,
929: and 0.96. The latter value corresponds to the minimum-enthalpy state.
930: {\it Right panel:} The density profile of the minimum-enthalpy state
931: ({\it solid curve}) compared to the reconstructed stellar number density in the
932: ONC ({\it dot-dashed curve}).}
933: \end{figure}
934: \clearpage
935:
936: \begin{figure}
937: \plotone{f6.eps}
938: \caption{Evolution of the velocity profile. In order of
939: deepening minima, the curves correspond to $\eta\tau$-values of 0.60, 0.92,
940: and 0.96. Note that the velocity is normalized to $\eta\alpha$.}
941: \end{figure}
942: \clearpage
943:
944: \begin{figure}
945: \plotone{f7.eps}
946: \caption{Total star formation rate in the ONC as a function of time. The
947: latter is actually shown as the stellar age. The solid histogram uses the
948: empirical ages from Paper~I, binned in 1~Myr intervals. The dashed curve is the
949: theoretical prediction. Also shown are the best-fit values for the model's
950: three free parameters: $\eta$, $\epsilon$, and $n$.}
951: \end{figure}
952: \clearpage
953:
954:
955: \end{document}
956:
957:
958: