1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass{emulateapj}
3: \usepackage{amsmath}
4: \usepackage{amssymb}
5: \usepackage{graphicx}
6:
7: \newcommand\bmath[1] {\mbox{\boldmath$\rm #1$}}
8:
9: %\bibliographystyle{mn2e}
10: \usepackage{natbib}
11: \bibliographystyle{aa}
12:
13:
14: %%% Abbreviations
15: % Common journal abbreviations
16: %\newcommand\aap{A\&A} %.......................Astronomy and Astrophysics
17: %\newcommand\aaps{A\&AS} %.....................Astronomy and Astrophysics Supplement Series
18: %\newcommand\aj{AJ} %..........................Astronomical Journal
19: %\newcommand\apj{ApJ} %........................Astrophysical Journal
20: %\newcommand\apjl{ApJL} %......................Astrophysical Journal Letters
21: %\newcommand\apjs{ApJS} %......................Astrophysical Journal Supplement Series
22: %\newcommand\apss{Ap\&SS} %....................Astrophysics and Space Science
23: %\newcommand\araa{ARAA} %......................Annual Reviews of Astronomy and Astrophysics
24: %\newcommand\josa{J.~Opt.~Soc.~Am.} %..........Journal of the Optical Society of America
25: %\newcommand\jpp{J.~Plasma Phys.} %............Journal of Plasma Physics
26: %\newcommand\mnras{MNRAS} %....................Monthly Notices of the Royal Astronomical Society
27: %\newcommand\nat{Nature} %.....................Nature
28: %\newcommand\ptrs{Phil.~Trans.~Roy.~Soc.} %....Philosophical Transactions of the Royal Society
29: %\newcommand\physrep{Phys.~Rep.} %.............Physics Reports
30: %\newcommand\plb{Phys. Lett. B} %..............Physics Letters B
31: %\newcommand\pp{Phys.~Plasma} %................Physical Plasmas
32: %\newcommand\prl{Phys.~Rev.~Lett.} %...........Physical Review Letters
33: %\newcommand\pr{Phys.~Rev.} %..................Physical Review
34: %\newcommand\prd{Phys.~Rev.~D} %...............Physical Review D
35: %\newcommand\pre{Phys.~Rev.~E} %...............Physical Review E
36: %\newcommand\pasj{Publ.~Astron.~Soc.~Japan} %..Publications of the Astronomical Society of Japan
37: %\newcommand\pasp{PASP} %......................Publications of the Astronomical Society of the Pacific
38: %\newcommand\prs{Proc.~Roy.~Soc.} %............Proceedings of the Royal Society
39:
40: % Standard abbreviations
41: \newcommand\cf{{cf.}} %.......confer
42: \newcommand\eg{{e.g.}} %......exempli gratia
43: \newcommand\ie{{i.e.}} %......id est
44: \newcommand\etal{{et al.}} %..and others
45: \newcommand\etc{{etc}} %......et cetera
46:
47: % Unit abbreviations
48: %%% Temperature
49: \newcommand\K{{\rm K}} %...........Kelvin
50: %%% Time
51: \newcommand\s{{\rm s}} %...........seconds
52: \newcommand\ns{{\rm n}\s} %........nanoseconds
53: \newcommand\mus{\mu\s} %...........microseconds
54: \newcommand\ms{{\rm m}\s} %........milliseconds
55: \newcommand\ks{{\rm k}\s} %........kiloseconds
56: \newcommand\yr{{\rm yr}} %.........years
57: \newcommand\Myr{{\rm M}\yr} %......megayears
58: \newcommand\Gyr{{\rm G}\yr} %......gigayears
59: %%% Frequency
60: \newcommand\Hz{{\rm Hz}} %.........Hertz
61: \newcommand\muHz{\mu\Hz} %.........microhertz
62: \newcommand\mHz{{\rm mHz}} %.......millihertz
63: \newcommand\MHz{{\rm MHz}} %.......Megahertz
64: \newcommand\GHz{{\rm GHz}} %.......Gigahertz
65: \newcommand\THz{{\rm THz}} %.......Terahertz
66: %%% Length
67: \newcommand\m{{\rm m}} %...........meters
68: %\newcommand\fm{{\rm f}\m} %........femtometers
69: \newcommand\nm{{\rm n}\m} %........nanometers
70: \newcommand\mum{\mu\m} %...........micrometers
71: \newcommand\mm{{\rm m}\m} %........millimeters
72: \newcommand\cm{{\rm c}\m} %........centimeters
73: \newcommand\km{{\rm k}\m} %........kilometers
74: % Angnstroms are already \AA
75: \newcommand\pc{{\rm pc}} %.........parsecs
76: \newcommand\kpc{{\rm k}\pc} %......kiloparsecs
77: \newcommand\Gpc{{\rm G}\pc} %......gigaparsecs
78: \newcommand\au{{\rm au}} %.........astronomical units
79: %%% Mass
80: \newcommand\g{{\rm g}} %...........grams
81: \newcommand\mug{\mu\g} %...........micrograms
82: \newcommand\mg{{\rm m}\g} %........milligrams
83: \newcommand\kg{{\rm k}\g} %........kilograms
84: %\newcommand\Ms{M_\odot} %...........solar masses
85: %%% Energy
86: \newcommand\eV{{\rm eV}} %.........electron volts
87: \newcommand\keV{{\rm k}\eV} %......kiloelectron volts
88: \newcommand\MeV{{\rm M}\eV} %......megaelectron volts
89: \newcommand\GeV{{\rm G}\eV} %......gigaelectron volts
90: \newcommand\TeV{{\rm T}\eV} %......teraelectron volts
91: \newcommand\PeV{{\rm P}\eV} %......petaelectron volts
92: \newcommand\XeV{{\rm X}\eV} %......exaelectron volts
93: \newcommand\ZeV{{\rm Z}\eV} %......zetaelectron volts
94: \newcommand\erg{{\rm erg}} %.......ergs
95: \newcommand\J{{\rm J}} %...........joules
96: %%% Magnetic Fields
97: \newcommand\G{{\rm G}} %...........gauss
98: \newcommand\nG{{\rm n}\G} %........nanogauss
99: \newcommand\muG{\mu\G} %...........microgauss
100: \newcommand\mG{{\rm m}\G} %........milligauss
101: \newcommand\kG{{\rm k}\G} %........kilogauss
102: \newcommand\T{{\rm T}} %...........tesla
103: \newcommand\muT{\mu\T} %...........microtesla
104: \newcommand\mT{{\rm m}\T} %........millitesla
105: %%% Angular Resolution
106: % ^\circ is already degrees
107: % ' is alrady minutes
108: % `` is already seconds
109: \newcommand\mas{{\rm mas}} %.......milli-arcseconds
110: \newcommand\muas{\mu{\rm as}} %....micro-arcseconds
111: \newcommand\nas{{\rm nas}} %.......nano-arcseconds
112:
113: \newcommand\NH{{${\rm N}_2 {\rm H}^+$}}
114:
115: \renewcommand\d{{\rm d}}
116: \newcommand\e{{\rm e}}
117:
118: \begin{document}
119:
120: \title{Oscillating Starless Cores: The Nonlinear Regime}
121:
122: \author{Avery E. Broderick\altaffilmark{1}, Eric Keto\altaffilmark{2},
123: Charles J. Lada\altaffilmark{3}, Ramesh Narayan\altaffilmark{4}}
124: \affil{Harvard-Smithsonian Center for Astrophysics, 60 Garden
125: Street, Cambridge, MA 02138, USA}
126: \altaffiltext{1}{abroderick@cfa.harvard.edu}
127: \altaffiltext{2}{keto@cfa.harvard.edu}
128: \altaffiltext{3}{clada@cfa.harvard.edu}
129: \altaffiltext{4}{rnarayan@cfa.harvard.edu}
130:
131: \shorttitle{OSCILLATING STARLESS CORES: THE NONLINEAR REGIME}
132: \shortauthors{BRODERICK ET AL.}
133:
134:
135: \begin{abstract}
136: In a previous paper, we modeled the oscillations of a
137: thermally-supported (Bonnor-Ebert) sphere as non-radial, linear
138: perturbations following a standard analysis developed for stellar
139: pulsations. The predicted column density variations and molecular
140: spectral line profiles are similar to those observed in the Bok
141: globule B68 suggesting that the motions in some starless cores may
142: be oscillating perturbations on a thermally supported equilibrium structure.
143: However, the linear analysis is unable to address several questions, among
144: them the stability, and lifetime of the perturbations.
145: In this paper we simulate the oscillations using a three-dimensional numerical
146: hydrodynamic code. We find that the oscillations are damped
147: predominantly by non-linear mode-coupling, and the damping time scale is
148: typically many oscillation periods, corresponding to a few million
149: years, and persisting over the inferred lifetime of gobules.
150: \end{abstract}
151:
152: \keywords{stars: formation, hydrodynamics, radiation mechanisms:
153: non-thermal, line: profiles, ISM: globules, ISM: clouds}
154:
155: \maketitle
156:
157:
158: \section{Introduction}
159:
160: Recent observations suggest that the internal structure of some of
161: the small isolated molecular clouds known as starless cores or Bok
162: globules \citep{Bok1948} is well approximated by a balance of thermal and
163: gravitational forces on which is superposed a pattern of non-radial
164: oscillations
165: \citep{Lada2003}. Previously,
166: the internal oscillations of these starless cores have been noted
167: in simulations of collapse \citep{Hennebelle2003}, magnetized clouds
168: \citep{Galli2005}, modeled as one-dimensional radial oscillations by
169: means of a numerical simulation that includes radiative energy losses
170: \citep{KetoField2005} and as three-dimensional non-radial
171: perturbations assuming isothermal conditions
172: \citep{Keto2006}. Analysis of these models and comparison of
173: simulated molecular spectral line profiles with those observed in the
174: dark clouds L1544 \citep{Caselli2002a} and B68 \citep{Lada2003}
175: indicate that the motions in some dark clouds are consistent with
176: large-scale hydrodynamic oscillations, i.e., sound waves.
177: Nevertheless, a number of questions remain as yet unanswered.
178:
179: Is the underlying assumption of the oscillations as perturbations valid for
180: the amplitudes required to produce the observed velocities?
181: Based on the observations of B68, the perturbation amplitudes in
182: our previous three-dimensional model were $\sim25$\%.
183: Does the linear model correctly
184: predict the velocity field for such large amplitudes
185: or would significant differences appear in the
186: non-linear regime?
187:
188:
189: What is the lifetime of the perturbations and thus the velocities
190: within the clouds? For hydrodynamic oscillations to remain a viable
191: explanation of the velocity field observed in B68 they must have
192: lifetimes in excess of a crossing time. Supersonic oscillations are strongly
193: damped via shocks. However, in the subsonic regime the dominant
194: damping mechanism is less clear. In the one-dimensional model \citep{KetoField2005}, the
195: dissipation of the oscillations occurs via radiative losses with an
196: estimated timescale of $10\,\Myr$. In the
197: non-radial perturbative model \citep{Keto2006} damping is explicitly ignored.
198: For mode amplitudes larger than linear perturbations,
199: non-linear mode-mode coupling provides an additional
200: potentially significant damping mechanism.
201:
202: Are the oscillations stable or growing? Are there preferred modes?
203: In general we expect the oscillations to damp over time, and we expect
204: that the longest-lived modes of oscillation would be the long wavelength
205: modes. Neither the one-dimensional non-linear analysis nor the three-dimensional
206: linear analysis is able to adequately address this question.
207:
208: In this paper we address these questions by modeling the oscillations of dark clouds
209: with a three-dimensional, non-linear,
210: numerical hydrodynamic simulation.
211: We find that the dominant damping mechanism is
212: indeed non-linear mode coupling, the longest wavelength mode lifetimes
213: are on the order of a few $\Myr$ and produce velocity and density
214: fields qualitatively similar to those observed. Section \ref{CM}
215: summarizes the numerical methods employed, sections \ref{QP} and
216: \ref{DP} discuss the viability of quadrupole and dipole oscillation
217: models, and the conclusions are contained in section \ref{C}.
218:
219:
220:
221: \section{Computational Modeling} \label{CM}
222:
223: \begin{figure}[tb]
224: \begin{center}
225: \includegraphics[width=\columnwidth]{f1.eps}
226: \caption{The density (solid) and pressure (dashed) profiles of the
227: unperturbed isothermal cloud (Bonnor-Ebert sphere). Near the
228: surface of the cloud ($r/R=1$) the density rapidly decreases.
229: However, the transition region in the equation of state softens this
230: drop over many grid zones to avoid difficulties in resolving the
231: pressure gradients.} \label{fig:profiles}
232: \end{center}
233: \end{figure}
234:
235:
236: \begin{figure}[tb]
237: \begin{center}
238: \includegraphics[width=\columnwidth]{f2.eps}
239: \caption{The dimensionless amplitude of the first quadrupole p-mode,
240: the (1,2,2) mode.
241: The dashed line is a quadratic fit to the logarithm of the
242: amplitude, with an e-folding time of roughly $2\,\Myr$ at $t=0$.
243: The squares indicate the times for which the frames in Figures
244: \ref{fig:q} were computed.
245: The dotted and long-dashed lines show the only other two significant
246: long-wavelength oscillations, listed by their $nlm$.} \label{fig:quad_amp}
247: \end{center}
248: \end{figure}
249:
250: \begin{figure*}[t!]
251: \begin{center}
252: \begin{tabular}{cc}
253: \includegraphics[width=0.48\textwidth]{f3a.eps}
254: &
255: \includegraphics[width=0.48\textwidth]{f3b.eps}
256: %\\
257: %(a) & (b)
258: \end{tabular}
259: \end{center}
260: \caption{The equatorial gas density, velocity, and CS(2-1) line
261: profiles of the quadrupolar mode after the first half-period, (left),
262: and the second full period, (right). These correspond to the times
263: marked by black squares in Figure \ref{fig:quad_amp}. Despite having
264: oscillated for nearly
265: $5\times10^{5}\,\yr$, the quadrupolar structure is still clearly
266: evident in the density \& velocity fields as well as in the spectral
267: line profiles. Spectral line profiles colored blue (red) indicate
268: approaching (retreating) gas on the near side relative to the
269: observer.} \label{fig:q}
270: \end{figure*}
271:
272:
273: \begin{figure}[tbh!]
274: \begin{center}
275: \includegraphics*[width=\columnwidth]{f4.eps}
276: \end{center}
277: \caption{A map of the CS(2-1) line profile overlaid upon the gas column
278: density for the quadrupolar oscillation after nearly
279: $5\times10^5\,\yr$. This is viewed from $\lambda=30^\circ$, $\theta=30^\circ$, for
280: which the pattern of spectral line profiles is similar to the Bok
281: globule B68 (\cf~Figure 7 of \citealt{Keto2006}). The
282: gas density is shown as a filled contour map, with contours equally
283: spaced in $10\%$ increments of the maximum value and the $50\%$
284: contour is marked by the thick black line. The CS(2-1) line profiles
285: are associated with lines of sight centered in each subplot. Blue (red)
286: lines correspond to approaching (retreating) gas on the nearside of
287: the cloud relative to the observer. Note the prominent double reversal of the
288: line-of-sight motions and the substantial distortion of
289: the cloud shape, as seen in B68 (and in the same relative
290: orientation).}\label{fig:qcds}
291: \end{figure}
292:
293:
294:
295: \begin{figure}[tb]
296: \begin{center}
297: \includegraphics[width=\columnwidth]{f5.eps}
298: \caption{The dimensionless amplitude of the first dipole p-mode, the
299: (1,1,1) mode.
300: The dashed line is a quadratic fit to the logarithm of the
301: amplitude, with an e-folding time of roughly $2\,\Myr$ at $t=0$.
302: The squares indicate the times for which the frames in Figure
303: \ref{fig:d} were computed.
304: The long-dashed line shows the $(0,2,2)$ mode, the only other significant
305: oscillation.} \label{fig:dip_amp}
306: \end{center}
307: \end{figure}
308:
309: \begin{figure*}[tbh]
310: \begin{center}
311: \begin{tabular}{cc}
312: \includegraphics[width=0.48\textwidth]{f6a.eps}
313: &
314: \includegraphics[width=0.48\textwidth]{f6b.eps}
315: \\
316: (a) & (b)
317: \end{tabular}
318: \end{center}
319: \caption{The equatorial gas density, velocity, and CS(2-1) line
320: profiles of the dipole mode after the first half-period, (a),
321: and the second full period, (b). These correspond to the times
322: marked by black squares in Figure \ref{fig:dip_amp}. Despite having
323: oscillated for more than
324: $5\times10^{5}\,\yr$, the dipolar structure is still clearly
325: evident in the density \& velocity fields as well as in the spectral
326: line profiles. Spectral line profiles colored blue (red) indicate
327: approaching (retreating) gas on the near side relative to the
328: observer. Note that the spectra along the bottom would seem to
329: imply expansion while those along the top would imply collapse.
330: Furthermore, spectra from the side suggests rotation.
331: Thus inferring the global state of the cloud from line profiles may
332: be complicated by the presence of large-scale
333: oscillations.}\label{fig:d}
334: \end{figure*}
335:
336:
337: \begin{figure}[tbh!]
338: \begin{center}
339: \includegraphics*[width=\columnwidth]{f7.eps}
340: \end{center}
341: \caption{A map of the CS(2-1) line profile overlaid upon the gas column
342: density for the dipolar oscillation after more than
343: $5\times10^5\,\yr$. This is viewed from the same angle as shown in
344: Figure \ref{fig:qcds}, ($\lambda=30^\circ$, $\theta=30^\circ$). The
345: gas density is shown as a filled contour map, with contours equally
346: spaced in $10\%$ increments of the maximum value and the $50\%$
347: contour is marked by the thick black line. The CS(2-1) line profiles
348: are associated with lines of sight centered in each subplot.
349: Blue (red)
350: lines correspond to approaching (retreating) gas on the nearside of
351: the cloud relative to the observer.
352: Note the prominent single reversal, similar to that
353: associated with cloud rotation. For the dipole mode the distortion
354: in the column density map is substantially less than that observed
355: for the quadrupolar mode.}\label{fig:dcds}
356: \end{figure}
357:
358:
359: \subsection{Hydrodynamic Evolution}
360: Because the thermal gas heating and cooling time, via collisional coupling
361: to dust at high densities \citep{Burke1983} and molecular line
362: radiation at low densities \citep{Goldsmith2001}, is short
363: in comparison to the typical oscillation period
364: (on the order of $10^5\,\yr$), the oscillations of dark-starless cores are well modeled
365: with an isothermal equation of state.
366: In particular, we use a
367: barotropic equation of state with an adiabatic index of unity, in
368: which case the isothermal evolution is also adiabatic. Since the
369: unbounded isothermal gas sphere is unstable, we truncate the solution
370: at a given density contrast ($\rho(R)/\rho_c$), producing the
371: standard Bonnor-Ebert sphere solution. This is done by inducing a phase
372: change in the equation of state, \ie,
373: \begin{equation}
374: P =
375: \left\{
376: \begin{aligned}
377: \rho \frac{kT}{\mu m_p} &&& \text{if } \rho\ge\rho(R)\,,\\
378: \left[ \frac{\rho}{\rho(R)} \right]^\epsilon \rho(R) \frac{k T}{\mu m_p} &&& \text{if } \rho(R) > \rho \ge \delta\rho(R)\frac{T}{T_{\rm ext}}\,,\\
379: \rho \frac{kT_{\rm ext}}{\mu m_p} &&& \text{otherwise}\,,
380: \end{aligned}
381: \right.
382: \end{equation}
383: where the intermediate state, with $\delta=0.8$ and $\epsilon =
384: \log(\delta)/\log(\delta T/T_{\rm ext}) \ll 1$ chosen to make $P(\rho)$
385: continuous, is introduced to avoid numerical artifacts at the
386: surface, and $T_{\rm ext}\gg T$. The resulting radial density and
387: pressure profiles are shown in Figure \ref{fig:profiles}. Note that
388: there is a transition region, associated with the intermediate state in
389: the equation of state, which distributes the sudden drop in density
390: over many grid zones. Despite this, the pressure is continuous
391: (necessarily) and nearly constant outside the surface of the
392: Bonnor-Ebert sphere. We find that generally our results are
393: insensitive to the particular equation of state of the exterior hot
394: gas, given that it is sufficiently dilute. This is appropriate for
395: globules like B68 which are known to be surrounded by hot, low density
396: gas.
397:
398: We employ a three-dimensional,
399: self-gravitating hydrodynamics code to follow the long term mode
400: evolution. Details regarding the numerical algorithm and specific
401: validation for oscillating gas spheres (in that case a white dwarf)
402: can be found in \citet{BroderickRathore2006}, and thus will only be briefly
403: summarized here.
404:
405: The code is a second order accurate (in space and time) Eulerian
406: finite-difference code, and has been demonstrated to have a low
407: diffusivity. Because the equation of state is barotropic, we use the
408: gradient of the enthalpy instead of the pressure in the Euler equation
409: since this provides better stability in the unperturbed configuration
410: \citep{BroderickRathore2006}. The Poisson equation is solved via
411: spectral methods, with boundary conditions set by a multipole
412: expansion of the matter on the computational domain.
413:
414: In order to assess the suitability of the code for application to
415: isothermal gas spheres we tested the stability to monopolar
416: perturbations as a function of central-to-surface density
417: contrast, finding instability for values above $13.5$. This is in approximate
418: agreement with the prediction of instability for density constrasts
419: above 14.3 \citep{FosterChevalier1993}.
420: The deviation from $14.3$ is likely
421: due to the effective atmosphere produced by the exterior hot matter,
422: and in particular the material in the transition region of the
423: equation of state.
424:
425: Henceforth, the specific physical parameters of the unperturbed model that we
426: use, chosen to be roughly applicable to B68, are
427: $\rho_c = 1.185\times10^{-18}\,\g \cm^{-3}$, $\rho(R)/\rho_c = 12.5$, and
428: $T = 10\,\K$. In all cases the simulations were performed at resolutions of
429: $64^3$, $128^3$ and $256^3$. Comparison of the latter two indicated
430: that the numerical solution had converged by these resolutions. The
431: results presented in the following sections are from the simulations
432: at $256^3$.
433:
434: The initial perturbed states are computed in the linear approximation
435: using the standard formalism of adiabatic \citep[which in this case is also
436: isothermal, ][]{Keto2006} stellar oscillations \citep{Cox1980}.
437: The initial conditions of the perturbed gas sphere are then given by
438: the sum of the equilibrium state and the linear perturbation. We
439: employ the same definition for the dimensionless mode amplitude as
440: given in the Appendix of \citet{Keto2006}.
441:
442: Throughout the evolution, the amplitude of a given oscillation mode,
443: denoted by its radial ($n$) and angular ($l$ \& $m$) quantum numbers,
444: may be estimated by the explicit integral:
445: \begin{equation}
446: A_{nlm} = \omega_{nlm}^{-1}\int \d^3\!x \,\rho\, \bmath{v}\cdot\bmath{\xi}_{nlm}^\dagger\,,
447: \label{eq:amp}
448: \end{equation}
449: where $\bmath{v}$ is the gas velocity, $\bmath{\xi}_{nlm}$ is the mode
450: displacement eigenfunction, $\omega_{nlm}$ is the mode frequency and
451: the dagger denotes hermitian conjugation.\footnote{For stellar
452: oscillations the mode amplitudes may be determined from the density
453: perturbation alone using the velocity potential. However, when
454: there is a non-vanishing surface density, as is the case for the
455: Bonnor-Ebert sphere, this is not possible and the velocity field
456: must be used.}
457:
458: \subsection{Molecular Line Simulations}
459:
460:
461: Assuming that the gas contains the usual interstellar molecules, we simulate molecular line
462: observations of the hydrodynamic model by calculating the non-LTE (non-thermodynamic
463: equilibrium) emission of the
464: CS(2-1) transition using our three-dimensional
465: ALI (accelerated $\Lambda$ iteration) code \citep{Keto2004}.
466: The densities, velocities, and constant temperature are taken from the hydrodynamic simulation.
467: The line broadening is assumed purely thermal.
468:
469:
470: %As in \citet{Keto2006}, we follow \cite{Bergin2001}, \cite{Tafalla2002}, and \cite{Tafalla2004} and approximate
471: %the dependence of the abundance
472: %on the density, $n$, as an exponential.
473: %\begin{equation}
474: %X(n) = X_0 \exp (-n/n_d)
475: %\end{equation}
476: %For CS we use an undepleted abundance of $X_0 = 6 \times 10^{-9}$ and critical
477: %density of $n_d = 10^{4}$ cm$^{-3}$. \citet{Tafalla2002} cite a
478: %range in values derived from
479: %their observations of $X_0 = 3 - 9 \times 10^{-9}$ and
480: %$n_d = 1.5 - 4.0 \times 10^4$.
481:
482:
483:
484: The abundance of CS is based on a simplified model of CO chemistry from \citet{KetoCaselli2008}.
485: We briefly describe the model here. The abundance
486: of CO in the gas is decreased by two effects, by photodissociation at the edge of the cloud and
487: by freezing onto dust grains in the high density center. We calculate the reduction in the CO
488: abundance due to these two affects and assume that the same reduction affects the CS molecule.
489: We assume, as in \citet{Keto2006}, an undepleted abundance of CS of
490: $6\times 10^{-9}$. This is consistent with \citet{Tafalla2002}, who
491: cite a range in values derived from their observations of $3 - 9 \times 10^{-9}$.
492:
493: \citet{TielensHollenbach1985} suggest that at the edge of the cloud where photodissociation is
494: important, the dominant cycle for the formation and destruction of CO is
495: \begin{equation}
496: {\rm C}^+ \rightarrow {\rm CH}_2 \rightarrow {\rm CO} \rightarrow {\rm C} \rightarrow {\rm C}^+\,.
497: \end{equation}
498: The time scale for the formation of CH$_2$ by radiative association is longer than for the
499: formation of CO from CH$_2$ and O. Therefore we assume that the rate of formation of CO is
500: given by that of CH$_2$. The CO cycle may then be described by three rate equations
501: for the creation of C$^+$, C, and CO plus the conservation condition that the total amount of
502: carbon is contained in these three species. The rate equations for photodissociation
503: depend on the mean visual extinction at each point, which we calculate using the
504: machinery of the three-dimensional radiative transfer code.
505:
506: The distribution of CO between the gas and grain phases is controlled by the rate of depletion from the
507: gas phase by freezing onto dust grains and the rate of
508: desorption off dust grains by cosmic rays. The steady state abundance of CO in the gas
509: phase is then given by the ratio of the depletion time to the sum of the depletion and desorption
510: times. Because the depletion time scales inversely with the collision rate while the desorption time
511: scales inversely with the cosmic ray ionization rate, the depletion of CO is dependent only on
512: the gas density because the cosmic ray ionization rate is the same throughout the cloud.
513:
514:
515:
516: \section{Quadrupole Perturbations} \label{QP}
517: The primary mode discussed in \citet{Keto2006} was the first
518: quadrupole p-mode, \ie, the $n=1$, $l=m=2$ oscillation mode. In
519: order to reproduce the observed velocities it was necessary to
520: consider a dimensionless amplitude of $0.25$, which produced
521: near-vanishing densities in some portions of the cloud, and is thus
522: clearly well into the non-linear regime.
523: %\subsection{Lifetime}
524:
525: The evolution of the dimensionless amplitude of the $(1,2,2)$ mode, as
526: defined by equation (\ref{eq:amp}), is shown as a function
527: of time in Figure \ref{fig:quad_amp}. The decay is well approximated
528: by an exponential. However, the strongly non-linear nature of the
529: decay is evident in the changing decay constant (resulting in a noticeable
530: bowing of the fitted line in Figure \ref{fig:quad_amp}), which yields an initial
531: e-folding time of approximately $2\,\Myr$, corresponding to roughly
532: $8$ oscillation periods ($16$ pattern periods).
533: Since the intrinsic dissipation in the code has previously been shown
534: to be small \citep{BroderickRathore2006}, the decay in the amplitude is
535: dominated by non-linear mode interactions, resulting in the transfer
536: of energy from large-scale, low-order modes to smaller-scale,
537: higher-order modes
538: \footnote{Generally the amplitude evolution may be expected to depend upon the
539: initial mode spectrum. We investigated the significance of this
540: dependence by introducing higher order modes with randomly chosen
541: amplitudes and phases. We found that as long as the
542: quadrupole mode remained dominant (in terms of $A_{nlm}$) the results
543: were unchanged.}. However, mass loss driven by large velocity
544: perturbations and mode coupling in
545: the artificial atmosphere (see Figure \ref{fig:profiles}) also
546: contribute to the decay. It is not clear if these processes are
547: artifacts of our numerical simulation and the manner in which we
548: initially prepare the cloud or if they may be found in nature as well.
549: Thus our estimate of the lifetime should be considered a
550: {\em lower limit}.
551:
552:
553: The only two oscillations of comparable wavelength (less than
554: approximately $R/5$) that reach significant amplitudes are the $(0,2,2)$
555: and $(0,4,4)$ modes, whose couplings are likely enhanced by the Cartesian
556: geometry of the underlying computational grid. In particular, there
557: is no evidence for an inverse-cascade, implying that the gas cloud
558: will not fragment. Indeed, this can be explicitly seen in the density
559: and velocity structures in the equatorial plane (see Figure \ref{fig:q})
560: in which the mode energy is primarily dissipated into small scale
561: motions. Thus even vigorous oscillations (at least up to amplitudes of
562: 25\%) do not appear to be sufficient to split the cloud into a
563: binary.
564:
565:
566: Despite the decay in amplitude, the quadrupole mode structure is
567: still apparent in the density distribution of the cloud after a number
568: of oscillation periods. This is also apparent in the substantially
569: distorted H$_2$ column densities, shown in Figure \ref{fig:qcds} for a
570: viewing angle that reproduces the generic features of B68
571: ($\theta$ and $\lambda$ corresponding to the lattitude and longitude,
572: respectively, of the line of sight and oriented such that
573: $\lambda=0^\circ$/$90^\circ$ corresponding to a view from the
574: bottom/left of the page in Figure \ref{fig:q}). Even after
575: many oscillations the column density maps are generally qualitatively
576: similar to observations of dark cores which often show clouds with an
577: aspect ratio of approximately 2:1
578: \citep{Ward-ThompsonMotteAndre1999, Bacmann2000}.
579:
580: Similarly, the complex velocity patterns, as inferred from the shape
581: of the CS(2-1) line profiles, are present after multiple oscillations.
582: To the left and below the plots of the density and velocity
583: structure in Figure \ref{fig:q}, the CS(2-1) line profiles are shown
584: for a number of lines of sight. Comparison of the early time and late
585: time line profiles suggests that the small scale motions, the result
586: of damping of the initial oscillation, do not significantly alter the
587: line shapes or, as a direct consequence, the inferred line-of-sight
588: velocity distribution throughout the cloud. Due to the intrinsically
589: complex nature of the oscillation velocity field, the inferred
590: velocity map can mimic expansion, contraction, rotation, or a combination of
591: these. For example, the velocity patterns inferred by the line shapes
592: shown in Figure \ref{fig:q} appear to correspond to rotation, while if
593: these were rotated by $45^\circ$ (diagonal lines of sight) they would
594: appear to correspond to expansion or contraction.
595:
596: The structure of the inferred velocity map can become substantially
597: more complex as we approach the pole (the polar angle $\theta$
598: approaches $0^\circ$). As seen in Figure \ref{fig:qcds}, for
599: $\theta\simeq30^\circ$ and azimuthal angle $\lambda\simeq30^\circ$
600: a quadrupolar oscillation can reproduce the general features of B68.
601: This includes a double reversal of the line-of-sight velocity across
602: the cloud (represented by the change from red to blue to red in the
603: line profiles) and the significantly distorted gas column density. In
604: fact, for the view shown, the relative orientation of the column
605: density distortion and the velocity double reversal is similar to that
606: observed in B68! This continues to be the case even after $5\times10^5\,\yr$.
607:
608: As anticipated in \citet{Keto2006}, the qualitative
609: features of the linear column density and velocity maps are still
610: present in the non-linear computation. In particular, both the
611: anisotropic nature of the column densities and the complex velocity
612: structure persist for a number of oscillation periods.
613: Nevertheless, the non-linear decay of the amplitude, and subsequent
614: small scale motions, does become significant on timescales of a $\Myr$
615: as evidenced by the decay of the mode amplitude in Figure
616: \ref{fig:quad_amp}. As a consequence, non-linear mode damping
617: appears to be the dominant constraint upon whether such an oscillation
618: can be clearly observed in velocity maps.
619:
620:
621:
622: \section{Dipole Perturbations} \label{DP}
623: Although not explicitly considered in \citet{Keto2006}, dipole
624: oscillations are a possible alternative interpretation to rotation,
625: and perhaps also capable of explaining the velocity structure of B68.
626: For this reason we considered the first dipole p-mode, \ie, the
627: $n=l=m=1$ oscillation mode, as well.
628:
629: As a consequence of its longer wavelength, the amplitude of the
630: dipole mode must be larger than that of the quadrupole oscillation
631: mode to produce similar velocities. Hence, we chose a dimensionless
632: amplitude of $0.5$. Similar to the quadrupole case, this is also
633: likely to be strongly non-linear.
634:
635:
636: As in the quadrupole case, the dipole oscillation decays
637: primarily into small scale motions. In both the quadrupole and dipole
638: case the cloud sheds a few percent of its mass from the outer layers,
639: contributing to the decay of the oscillation. In the dipole case this
640: also produces a notable shift in the center of mass of the cloud as a
641: result of the asymmetric mass loss through the boundary of the
642: computational domain.
643:
644:
645: As with the quadrupolar oscillations the gas column densities are
646: initially strongly distorted. However, in contrast to the quadrupole
647: mode, after the first oscillation period (approximately
648: $1.5\times10^5\,\yr$) they have become roughly circular (see, \eg, Figure
649: \ref{fig:dcds}). This suggests that large amplitude dipole modes
650: alone are insufficient to generate the observed anisotropy in the
651: shapes of dark cores.
652:
653: Figure \ref{fig:dip_amp} shows the evolution of the dimensionless
654: amplitude for the first dipole p-mode. Again the evolving decay
655: constant implies that this is a strongly non-linear process. However,
656: in this case no oscillations with similar amplitudes develop within
657: the duration of the simulation (the largest being the $(0,2,2)$ mode,
658: which again is likely due to the Cartesian nature of the grid). The
659: mode lifetime is on the order of $2\,\Myr$, considerably longer than
660: the mode oscillation period. As before, there is no indication of
661: cloud fragmentation.
662:
663:
664: Nevertheless, the velocity maps inferred from the CS (2-1)
665: line shapes, shown in Figures \ref{fig:d}, retain their dipolar
666: structure, even after many oscillations. Note that
667: this implies that inferring the global dynamical state of the cloud
668: from line profiles alone is significantly complicated by the possible
669: existence of large-scale oscillations. As shown in Figures
670: \ref{fig:d} \& \ref{fig:dcds}, for a variety of viewing angles, the
671: motion appears to mimic rotation, expansion or contraction, all while
672: in the same dynamical state. This is still true after the column
673: density maps become nearly spherical.
674:
675:
676: \section{Conclusions} \label{C}
677: Despite their strongly non-linear nature, large-amplitude oscillations
678: of pressure-supported molecular clouds reproduce many of the
679: qualitative features of observations of starless cores. These
680: include the complex velocity structures and highly distorted column
681: densities found in B68 and L1544, as well as mimicking rotation,
682: expansion, and contraction depending upon the mode structure and
683: viewing angle. This may complicate efforts to discern the global
684: dynamical state of these objects from observations of molecular
685: line shapes alone.
686:
687: However, it may be possible to distinguish oscillations from collapse
688: by comparing the spectral line profiles and velocities of volatile
689: molecules such as N2H+ and NH3 against those of more refractory
690: species such as CO, HCO+, or CS \citep{KetoField2005}. In a simple
691: breathing mode oscillation, the gas velocities are highest towards the
692: outer regions of the cloud where the abundance of the carbon species
693: is not depleted by freeze-out. In contrast, in the gravitational
694: collapse of unstable BE spheres, the gas velocities are highest in the
695: center of the cloud where the emission from the non-depleting nitrogen
696: molecules is highest.
697:
698: We note that we have restricted our attention to hydrodynamic
699: oscillations of pressure-bounded, thermally-supported
700: spheres. This description seems appropriate for globules such as
701: B68 that are surrounded by hotter gas and have
702: density profiles that closely match those of Bonnor-Ebert
703: spheres. Whether the Bonnor-Ebert description is appropriate for
704: embedded cores remains to be determined. For example, if magnetic
705: stresses contribute significantly to cloud support, then additional
706: classes of oscillations exist, associated with the magnetic field
707: \citep[see,\eg,][]{Hennebelle2003,Galli2005}.
708:
709: The lifetimes of hydrodynamic oscillations of starless cores like
710: B68 are sufficiently long (few $10^6$ yr) to be a significant
711: fraction of the expected lifetime of a molecular cloud ($10^7$
712: yr). The oscillation lifetimes are dominantly limited by non-linear
713: mode-mode coupling. Nevertheless, large amplitude oscillations
714: persist for many periods (and thus many sound crossing times), and
715: are therefore not expected to be particularly rare. There are no
716: signs of cloud fragmentation, and thus amplitudes comparable to those
717: discussed here are insufficient to produce binaries without the
718: inclusion of additional physical processes.
719:
720:
721: \acknowledgments
722: A.E.B. gratefully acknowledges the support of an ITC Fellowship from
723: Harvard College Observatory.
724:
725: \begin{thebibliography}
726:
727: \bibitem[Alves, Lada, \& Lada(2001)]{AlvesLadaLada2001}
728: Alves, J., Lada, C., \& Lada, E. 2001, \nat, 409, 159
729:
730: \bibitem[Bacmann et al.~(2000)]{Bacmann2000}
731: Bacmann, A., Andre, A.P., Puget, J.-L., Abergel, A., Bontemps, S. \& Ward-Thompson, D. 2000, \aap, 361, 555
732:
733:
734: \bibitem[Bergin et al.~(2001)]{Bergin2001}
735: Bergin, E., Ciardi, D., Lada, C., Alves, J., Lada, E., 2001, \apj, 557, 209
736:
737: \bibitem[Bok (1948)]{Bok1948}
738: Bok, B., 1948, {\it Centennial Symposia}, Harvard Observatory Monographs No. 7, Harvard-College Observatory, Cambridge
739: \bibitem[Broderick \& Rathore (2006)]{BroderickRathore2006}
740: Broderick, A.~E. and {Rathore}, Y., 2006, \mnras, 372, 923
741:
742: \bibitem[Burke \& Hollenbach (1983)]{Burke1983}
743: Burke, J.~R. \& Hollenbach, D.~J., 1983, \apj, 265, 223
744:
745: \bibitem[Caselli et al.(2002)]{Caselli2002a}
746: Caselli, P., Walmsley, C., Zucconi, A., Tafalla, M., Dore, L. \& Myers, P. 2002a, \apj, 565, 331
747:
748: \bibitem[Cox(1980)]{Cox1980}
749: Cox, J.P., 1980, Theory of Stellar Pulsations, Princeton University Press: Princeton
750:
751: \bibitem[Foster \& Chevalier (1993)]{FosterChevalier1993}
752: Foster, P. \& Chevalier, R., 1993, \apj, 416, 303
753:
754: \bibitem[Galli (2005)]{Galli2005}
755: Galli, D., 2005, \mnras, 359, 1083
756:
757: \bibitem[Goldsmith (2001)]{Goldsmith2001}
758: Goldsmith, P., 2001, \apj, 557, 736
759:
760: \bibitem[Hennebelle (2003)]{Hennebelle2003}
761: Hennebelle, P., 2003, \aap, 411, 9
762:
763: \bibitem[Keto et al.(2004)]{Keto2004}
764: Keto, E., Rybicki, G., Bergin, E., Plume, R., 2004, \apj, 613, 355
765:
766: \bibitem[Keto \& Field(2005)]{KetoField2005}
767: Keto, E. \& Field, G., 2005, \apj, 635, 1151
768:
769: \bibitem[Keto et al.(2006)]{Keto2006}
770: Keto, E., Broderick, A., Lada, C.J. \& Narayan, R., 2006, \apj, 652, 1366
771:
772: \bibitem[Keto \& Caselli(2008)]{KetoCaselli2008}
773: Keto, E. \& Caselli, P., 2008, {\em in preparation}
774:
775: \bibitem[Lada et al.(2003)]{Lada2003}
776: Lada, C., Bergin, E., Alves, J., Huard, T., 2003, \apj, 586, 286
777:
778: \bibitem[Tafalla et al.~(2002)]{Tafalla2002}
779: Tafalla, M., Myers, P., Caselli, P., Walmsley, C. \& Comito, C., 2002, \apj, 569, 815
780:
781: \bibitem[Tielens \& Hollenbach(1985)]{TielensHollenbach1985}
782: Tielens, A.G.G.M. \& Hollenbach, D., 1985, \apj, 291, 722
783:
784: \bibitem[Ward-Thompson, Motte, \& Andre(1999)]{Ward-ThompsonMotteAndre1999}
785: Ward-Thompson, D., Motte, F. \& Andre, P. 1999, \mnras, 305, 143
786:
787:
788: \end{thebibliography}
789:
790:
791:
792:
793:
794: \end{document}
795:
796:
797:
798:
799: The lifetimes of hydrodynamic oscillations of starless cores like
800: B68 are sufficiently long (few $10^6$ yr) to be a significant
801: fraction of the expected lifetime of a molecular cloud ($10^7$
802: yr). The oscillation lifetimes are dominantly limited by non-linear
803: mode-mode coupling. Nevertheless, large amplitude oscillations
804: persist for many periods (and thus many sound crossing times), and
805: are therefore not expected to be particularly rare, though it remains
806: to be determined if this is true for cores not embedded in hot, low density
807: envelopes. There are no signs of cloud fragmentation, and thus
808: amplitudes comparable to those discussed here are insufficient to
809: produce binaries without the inclusion of additional physical
810: processes.
811:
812: Finally, we note that we have restricted our attention to hydrodynamic
813: oscillations. For clouds in which magnetic stresses contribute
814: significantly to the cloud support additional classes of oscillations
815: exist, associated with the magnetic field
816: \citep[see, \eg,][]{Hennebelle2003,Galli2005}. However,the fact that
817: for a number of globules, including B68, the fact that the
818: measured density profile so closely fits the Bonnor-Ebert profile
819: implies that for at least a substantial sub-class of these objects our
820: hydrodynamic analysis is likely sufficient.
821: