1: %\documentclass[preprint2]{aastex}
2: %\documentclass[12pt,preprint]{aastex}
3: \documentclass{emulateapj}
4:
5: \newcommand{\lO} {\lambda_{\Omega}}
6: \newcommand{\led} {\lambda_{e}}
7: \newcommand{\cAi} {c_{Ai}}
8: \newcommand{\cAn} {c_{An}}
9: \newcommand{\RM} {R_{M}}
10: \newcommand{\mbfB}{\mathbf{B}}
11: \newcommand{\mbfu}{\mathbf{u}}
12: \newcommand{\mbfui}{\mathbf{u}_i}
13: \newcommand{\mbfun}{\mathbf{u}_n}
14: \newcommand{\mbfz}{\mathbf{z}}
15: \newcommand{\mbfn}{\mathbf{n}}
16: \newcommand{\mbfk}{\mathbf{k}}
17: \newcommand{\mbfx}{\mathbf{x}}
18: \newcommand{\mbfv}{\mathbf{v}}
19: \newcommand{\mbfnabla}{\mathbf{\nabla}}
20: \newcommand{\prt}{\partial}
21: \newcommand{\f} {\frac}
22: \newcommand{\ddx}{\partial_x}
23: \newcommand{\ddy}{\partial_y}
24: \newcommand{\ddt}{\partial_t}
25: \newcommand{\lomega}{\lambda_\Omega}
26:
27: \begin{document}
28:
29: \title{Fragmentation of Shocked Flows:
30: Gravity, Turbulence and Cooling}
31:
32: \author{Fabian Heitsch\altaffilmark{1}}
33: \author{Lee W. Hartmann\altaffilmark{1}}
34: \author{Andreas Burkert\altaffilmark{2}}
35: \altaffiltext{1}{Dept. of Astronomy, University of Michigan, 500 Church St.,
36: Ann Arbor, MI 48109-1042, U.S.A}
37: \altaffiltext{2}{Universit\"ats-Sternwarte M\"unchen, Scheinerstr. 1, 81679 M\"unchen, Germany}
38: \lefthead{Heitsch et al.}
39: \righthead{Fragmentation of Shocked Flows}
40:
41: \begin{abstract}
42: The observed rapid onset of star formation
43: in molecular clouds requires rapid formation of dense fragments which
44: can collapse individually before being overtaken by global gravitationally-driven
45: flows. Many previous investigations have suggested that supersonic turbulence
46: produces the necessary fragmentation,
47: without addressing however the source of this
48: turbulence. Motivated by our previous (numerical) work on the flow-driven
49: formation of molecular clouds, we investigate the expected timescales
50: of the dynamical and thermal instabilities leading to the rapid fragmentation of gas swept
51: up by large-scale flows, and compare them with global gravitational
52: collapse timescales. We identify parameter regimes in gas density, temperature and
53: spatial scale within which a given instability will dominate.
54: We find that dynamical instabilities disrupt large-scale coherent flows via generation of
55: turbulence, while strong thermal fragmentation
56: amplifies the resulting low-amplitude density perturbations, thus leading to small-scale,
57: high-density fragments as seeds for {\em local} gravity to act upon.
58: Global gravity dominates only on the largest scales; large-scale gravitationally-driven
59: flows promote the formation of groups and clusters of stars formed by
60: turbulence, thermal fragmentation, and rapid cooling.
61: \end{abstract}
62: \keywords{gravitation --- instabilities --- turbulence --- methods:analytical
63: --- stars:formation --- ISM:clouds}
64:
65: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
66: %
67: %\section{Motivation}
68: %
69: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
70: \section{Motivation}\label{s:motivation}
71:
72: There is increasing evidence that star formation (in the solar neighborhood) follows rapidly
73: on molecular cloud formation (\citealp{2001ApJ...562..852H}; \citealp{2007RMxAA..43..123B};
74: \citealp{2007ApJ...668.1064E}).
75: This evidence suggests that the density
76: enhancements in which stars form are produced {\em during the cloud formation phase} --
77: and not after the cloud has formed, as implicitly assumed by the initial conditions
78: of many numerical models.
79: Global gravitational modes can sweep up material at the edges of clouds
80: \citep{2004ApJ...616..288B}, leading to the formation of filaments
81: \citep{2007ApJ...657..870V,2008ApJ...674..316H}. Yet for local gravitational collapse to win,
82: the cloud needs to be seeded with local perturbations early on \citep{2004ApJ...616..288B,2007ApJ...654..988H}.
83: Not only need these density enhancements to arise early, but they also need to be
84: non-linear, i.e. the cloud must acquire strong small-scale density perturbations
85: during its formation. Thus, a rapid fragmentation mechanism other -- and faster! -- than gravity is needed.
86:
87: The scenario of flow-driven cloud formation, in which
88: large-scale shocked flows of atomic hydrogen in the warm neutral medium
89: (WNM) assemble molecular clouds
90: (\citealp{1999ApJ...515..286B}; \citealp{2001ApJ...562..852H}), offers an elegant
91: mechanism for the rapid fragmentation of the proto-cloud and the generation of
92: non-linear density seeds. The key to the rapid fragmentation is a combination
93: of strong dynamical and thermal instabilities in shocked flows
94: (\citealp{1999A&A...351..309H} and \citealp{2000ApJ...532..980K} for one-dimensional
95: models; \citealp{2002ApJ...564L..97K} for 2D models including H$_2$ formation;
96: \citealp{2005A&A...433....1A}, \citealp{2005ApJ...633L.113H,2006ApJ...648.1052H},
97: \citealp{2007A&A...465..431H} and \citealp{2007A&A...465..445H}
98: emphasizing the formation of cold neutral medium clouds in 2D; \citealp{2006ApJ...643..245V},
99: \citealp{2007ApJ...657..870V} and \citealp{2008ApJ...674..316H} for three-dimensional models,
100: the latter two including self-gravity).
101: Rather than generating turbulence by imposing e.g. a velocity field chosen ad hoc in
102: Fourier space, density and velocity structures arise naturally during the formation
103: of the cloud in this scenario.
104:
105: In this study, we attempt to throw some light on the relative importance of
106: various classes of instabilities at play, via a consideration of their timescales.
107: We aim to assess whether fragmentation processes other than gravity are acting
108: rapidly enough during the formation of molecular clouds to allow the rapid onset
109: of local star formation.
110: Thermal effects -- strong cooling and thermal instability
111: (TI, \citealp{1965ApJ...142..531F}) play a dominant role,
112: weakening the effective equation of state in a
113: shocked flow and thus allowing rapid fragmentation. The TI derives its
114: strength from a combination with turbulence which generates density perturbations
115: on small scales. The resulting rapid growth of small-scale perturbations allows the rapid
116: onset of {\em local} gravitational collapse, before global collapse modes overwhelm
117: any pre-existing low-amplitude density variations. Hence, in the context of flow-driven
118: cloud formation, thermal fragmentation is the key to the rapid onset of star formation.
119:
120: The expressions for the timescales and the instability conditions are given in
121: \S\ref{s:timecond}. We discuss the parameter regimes for the instabilities
122: and their consequences in \S\ref{s:discussion} and summarize in \S\ref{s:summary}.
123:
124: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
125: %
126: %\section{Timescales and Instability Conditions\label{s:timescales}}
127: %
128: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
129: \section{Timescales and Instability Conditions}\label{s:timecond}
130: The instabilities enabling the rapid fragmentation of shocked flows are
131: associated with characteristic timescales. These
132: will serve as a vehicle to estimate the relative importance
133: of the instabilities for fragmentation and turbulence generation.
134:
135: Since we are mainly interested in the competition between global gravity
136: and local fragmentation processes,
137: we restrict ourselves to the discussion of four instabilities which can
138: be split into two groups of physical processes, namely condensation
139: processes (driven by cooling [\S\ref{ss:ti}] and gravity [\S\ref{ss:gi}])
140: and generic fluid instabilities (ram pressure imbalance [\S\ref{ss:NTSI}]
141: and shear flows [\S\ref{ss:KHI}]).
142: This list of instabilities is by no means exhaustive (\S\ref{ss:oi}).
143: Our choice is to some extent guided by the results of our numerical simulations
144: \citep{2008ApJ...674..316H}, and by the notion of cloud formation in large-scale, organized flows
145: such as spiral arms of galaxies
146: \citep{2003ApJ...599.1157K,2006ApJ...646..213K,2007ApJ...668.1064E,2007MNRAS.376.1747D,2008MNRAS.tmp..285D},
147: or expanding supernova shells (e.g. \citealp{1977ApJ...214..725E}; see also discussion in
148: \citealp{2007RMxAA..43..123B}).
149:
150: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
151: %\subsection{Thermal Instability}
152: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
153: \subsection{Thermal Instability (TI) and Cooling Timescales}\label{ss:ti}
154: The TI rests on the balance between heating and cooling processes.
155: The astrophysically most relevant mode is the isobaric
156: condensation mode (e.g. \citealp{1995ASPC...80..328B}).
157: %new begin
158: The condensation mode's linear growth
159: rate is independent of the wave number as long as the perturbation can
160: adjust to an isobaric state and under neglection of heat conduction,
161: however, perturbations on
162: smaller scales will grow first \citep{2000ApJ...537..270B}.
163: Thus, in the linear regime, the condensation mode is limited to scales smaller
164: than the sound crossing length
165: %new end
166: (see also \citealp{2003LNP...614..213V} and \citealp{2007A&A...465..431H}),
167: \begin{equation}
168: \lambda_c = \tau_c\,c_s,\label{e:cond_soundcross}
169: \end{equation}
170: where
171: \begin{equation}
172: \tau_c=\frac{k_BT}{n\Lambda}\label{e:ts_cool}
173: \end{equation}
174: is the cooling timescale with temperature $T$ and particle density $n$.
175: Thermal energy losses due to optically thin, collisionally excited atomic lines
176: are given by the cooling function $\Lambda(T)$ in erg~cm$^3$~s$^{-1}$. The sound
177: speed is
178: \begin{equation}
179: c_s\equiv\sqrt\frac{\gamma k_B T}{\mu},\label{e:csound}
180: \end{equation}
181: with the mean molecular mass $\mu$.
182: Since the TI spans a range of 2 orders of magnitude in temperature,
183: condition~(\ref{e:cond_soundcross}) can become quite restrictive, but
184: it varies strongly with temperature and the strength of the cooling
185: (see Fig.~\ref{f:coolcurve}).
186: The scale $\lambda_c$ can drop to a few tenths of a parsec for parameters typical of the WNM.
187:
188: %new begin
189: In the non-linear regime, and
190: on scales $\lambda>\lambda_c$, the gas can still cool isochorically such that the resulting
191: pressure drop will generate waves (e.g. \citealp{1995ASPC...80..328B}), leading to additional
192: fragmentation.
193: In that sense, equation~(\ref{e:cond_soundcross}) is not a strict upper limit for the TI, since
194: it only refers to the (linear) evolution of the condensation mode.
195: Nevertheless, we will use it as a proxy for estimating the importance of the TI since it gives
196: the largest scale out of which a single coherent cold fragment can form and thus is emphasizing
197: the local nature of the TI.
198: %new end
199: On small scales, the condensation mode is limited by the Field (\citeyear{1965ApJ...142..531F})
200: length, below which heat conduction will become important
201: (see e.g. \citealp{2004ApJ...602L..25K}).
202: The condensation mode will grow if
203: \begin{equation}
204: \left(\f{\prt{\cal L}}{\prt T}\right)_n
205: - \f{n_0}{T_0}\left(\f{\prt{\cal L}}{\prt n}\right)_T<0\label{e:cond_ti},
206: \end{equation}
207: where ${\cal L} \equiv n\Lambda-\Gamma$ is the loss-heat function in
208: erg~s$^{-1}$ \citep{1965ApJ...142..531F}.
209:
210: The growth timescale of the condensation mode under the conditions
211: (\ref{e:cond_soundcross}) and (\ref{e:cond_ti}) is given by
212: \begin{equation}
213: \tau_{TI}\equiv\f{\gamma\,k_B\,T}{(\gamma-1)(n\,\prt{\cal L}_n-T\,\prt{\cal L}_T)}
214: \label{e:ts_TI}
215: \end{equation}
216: with the partial derivatives $\prt{\cal L}_n\equiv \prt{\cal L}/\prt n$ and
217: $\prt{\cal L}_T\equiv \prt{\cal L}/\prt T$.
218: Equation~(\ref{e:ts_TI}) includes the condition for the condensation mode,
219: i.e. for $\tau_{TI}<0$, the condensation mode does not grow.
220:
221: Thermal effects still may be important in the absence of the TI, in which case the
222: thermal timescale is given by the more general cooling time (eq.~[\ref{e:ts_cool}]).
223: A short $\tau_c$ does not necessarily entail TI.
224: Rather, a short $\tau_c$ in the presence of a more or less constant heating term
225: means that the gas cools rapidly to a minimum temperature and then stays at that
226: temperature effectively isothermal, i.e. the effective adiabatic exponent
227: $\gamma_{eff}=1$, as in $P\propto n^{\gamma_{eff}}$. A $\gamma_{eff}<0$ indicates
228: thermal instability: with increasing density, the pressure drops. For
229: $0<\gamma_{eff}<1$, fragmentation can still be enhanced in the
230: presence of an external (gravitational or ram) pressure.
231:
232: We derive the cooling function from a combination of the rates
233: quoted by \citet{1972ARA&A..10..375D} and \citet{1995ApJ...443..152W}
234: for $T<10^4$~K, while for $T>10^4$~K, we use the tabulated curves
235: of \citet{1993ApJS...88..253S}.
236: Figure~\ref{f:coolcurve} summarizes the cooling curves for a range of
237: ionization degrees $x_i$ and metallicities, corresponding to solar
238: abundances \citep{1972ARA&A..10..375D} and abundances typical for
239: the LMC and SMC \citep{1999IAUS..190..266G}.
240:
241: \begin{figure}
242: \includegraphics[width=\columnwidth]{f1.eps}
243: % \includegraphics[width=\columnwidth]{../figures/allcoolf_all.eps}
244: \caption{\label{f:coolcurve}Composite cooling curve (see text)
245: for various ionization degrees and for metallicities
246: corresponding to the solar neighborhood, the LMC and the SMC.}
247: \end{figure}
248:
249: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
250: %\subsection{Gravity}
251: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
252: \subsection{Gravity, Jeans Instability (GI)\label{ss:gi}}
253: A (one-dimensional) region becomes gravitationally unstable if
254: \begin{equation}
255: \rho>\f{(c_s\,k)^2}{4\pi G},\label{e:cond_jeans}
256: \end{equation}
257: with the wave number $k$ and mass density $\rho\equiv\mu n$. The corresponding time scale is
258: \begin{equation}
259: \tau_{GI}=(4\pi G\rho-c_s^2k^2)^{-1/2},\label{e:ts_GI}
260: \end{equation}
261: again including the condition for instability. Gravitational collapse occurs for
262: $\tau_{GI}\in \mathbb{R}$.
263:
264: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
265: %\subsection{Non-Linear Thin Shell Instability (NTSI)}
266: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
267: \subsection{Non-Linear Thin Shell Instability (NTSI)\label{ss:NTSI}}
268: The NTSI \citep{1994ApJ...428..186V}
269: arises in a shock-bounded slab, when ripples in a two-dimensional slab
270: focus incoming shocked material and produce density fluctuations.
271: The growth rate of the NTSI is mostly controlled by $k\eta$, the product
272: of the wave number of the slab perturbation $k=2\pi/L$, and the amplitude of the
273: slab's initial displacement $\eta$ (equivalently, the amplitude of the
274: collision interface's geometrical perturbation).
275: The instability is driven by lateral transport of longitudinal momentum,
276: i.e. if the inflow is parallel to the $x$ direction, and the slab is in the
277: $y$-$z$-plane, $x$-momentum is transported laterally in $y$ (and $z$),
278: collecting at the focal points of the perturbed slab. The efficiency of lateral
279: momentum transport is key to the development of the instability, since
280: it is the imbalance of ram pressure at the focal points that eventually propels
281: matter forward, driving the growth of the slab's perturbation.
282: \citet{1994ApJ...428..186V} derived a growth timescale of
283: \begin{equation}
284: \tau_{NTSI} \approx [c_sk (k\eta)^{1/2}]^{-1}.\label{e:vishniac}
285: \end{equation}
286: \citet{1996NewA....1..235B} found that at constant $\eta$ and for small $k$,
287: equation~(\ref{e:vishniac}) yields only a lower limit, while for large $k$, the
288: analytical growth rates agree well with the numerical results.
289: The reason for this seems to lie in the efficiency of deflecting the incoming
290: flow: for small $k$, a small fraction of the incoming flow's momentum
291: is converted to lateral motions, while a large part compresses the
292: slab (depending on the equation of state, this could lead to an increase in energy
293: losses).
294: For the parameter study (\S\ref{s:discussion}), we set $\eta\equiv L/4$.
295: Keeping $k$ constant and varying $\eta$ affects the NTSI in a similar way as
296: keeping $\eta$ constant and changing $k$, with a weaker dependence on $\eta$.
297: Magnetic fields can suppress the NTSI if the ram pressure of the incoming
298: flow drops below the magnetic pressure \citep{2007ApJ...665..445H}.
299: The NTSI is an efficient dynamical focusing mechanism for large-scale gas streams,
300: resulting in the rapid build-up of massive cores in the focal points
301: \citep{2003NewA....8..295H} as possible sites for massive star formation \citep{2008ApJ...674..316H}.
302:
303: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
304: %\subsection{Kelvin-Helmholtz-Instability (KHI)}
305: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
306: \subsection{Kelvin-Helmholtz-Instability (KHI)}\label{ss:KHI}
307: The NTSI converts the highly compressible modes of the inflows
308: into shear flows at the flanks of the perturbed slab. These
309: shear flows give rise to the KHI, which thus is secondary to
310: the NTSI, but which also is the main turbulence generation
311: mechanism. The turbulence in turn leads to the saturation of the NTSI
312: \citep{1994ApJ...428..186V}. In the simplest -- incompressible -- scenario
313: of the KHI \citep{1961hhs..book.....C}, the shear layer has constant density, and the
314: velocity profile is a step function. In this case, the growth timescale is
315: given by the velocity difference $\Delta U$ and the wave number as
316: \begin{equation}
317: \tau_{KHI} = (k\Delta U)^{-1},\label{e:ts_KHI}
318: \end{equation}
319: i.e. the system is unconditionally unstable. This is definitely the most extreme case:
320: magnetic field components parallel to the shear flows can stabilize against the KHI, and
321: for compressible flows, the system will be stable for all those wave numbers whose effective
322: Mach number is larger than a critical value \citep{1968RvMP...40..652G}. For a
323: detailed discussion of the KHI, see \citet{2008arXiv0802.2497P}.
324:
325: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
326: %\subsection{Other Instabilities}
327: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
328: \subsection{Other Instabilities}\label{ss:oi}
329: The list of instabilities and their effects considered here is by no means exhaustive.
330: For instance, \citet{2006ApJ...652.1331I} discuss a corrugation instability of evaporation fronts
331: appearing at the interface between the WNM and the CNM on timescales of approximately $\sim 1$~Myr
332: in the WNM, and down to $0.01$~Myr in the CNM. This -- so they argue -- is on the
333: order of the cooling timescales in each of the media, so that the instability
334: could contribute to the generation of turbulence in the ISM. Hence the corrugation instability
335: will play an important role for the dynamics of the WNM/CNM. However, as equation~(47) and
336: Figure~9 of \citet{2006ApJ...652.1331I} demonstrate, the growth timescale of the corrugation
337: instability is in the (for our application) interesting range of $\lesssim 1Myr$ only for
338: sub-parsec scales, which are substantially smaller than the large-scale flows sweeping up
339: whole clouds as envisaged in the current study.
340:
341: In a similar vein, we do not discuss turbulence generation by the TI alone. This possibility
342: has been investigated extensively (e.g. \citealp{2000ApJ...537..270B}; \citealp{2002ApJ...569L.127K};
343: \citealp{2003LNP...614..213V}). \citet{2005ARA&A..43..337C} points out that because of the limited
344: energy reservoir, thermal pressure variations in a bistable ISM are expected to have negligible dynamical
345: effects, requiring additional sources or triggers for turbulence generation (e.g. \citealp{2002ApJ...580L..51K}).
346:
347: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
348: %\subsection{Criteria for Instability Dominance}
349: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
350: \subsection{Criteria for Instability Dominance}\label{ss:criteria}
351:
352: The characteristic timescales
353: (eqs.~[\ref{e:ts_TI},\ref{e:ts_GI},\ref{e:vishniac},\ref{e:ts_KHI}])
354: can be used to derive criteria for the dominance of a given
355: instability in terms of physical quantities. We chose (particle) density and
356: temperature, to facilitate a straight-forward comparison to ISM regimes. Attributes
357: in square brackets in each of the following subsection titles stand for the line-styles
358: used in Figure~\ref{f:ntplane}. Not all combinations are physically relevant. For
359: instance, $\tau_c\ll\tau_{GI}$ generally for gravitationally unstable regimes
360: (\S\ref{ss:gravvstherm}). The ratio of specific heat capacities is set to
361: $\gamma\equiv 5/3$, and the mean molecular weight $\equiv 2.36$. Obviously, these
362: choices are not valid over the full parameter range discussed, but they will considerably
363: simplify the discussion.
364:
365: \subsubsection{Gravity threshold [green dashed]}
366: A region will be gravitationally unstable if
367: \begin{eqnarray}
368: n &>& \f{\gamma k_B}{4\pi G\mu^2}\,k^2\,T\label{e:gi:rho}\\
369: &\approx& 7.4\times 10^1 \frac{T}{[\mbox{K}]}\,\left(\frac{L}{[\mbox{pc}]}\right)^2\,\mbox{cm}^{-3}..
370: \nonumber
371: \end{eqnarray}
372: This is just equation~(\ref{e:cond_jeans}) expressed in terms of temperature.
373:
374: \subsubsection{GI dominates NTSI and KHI [red dashed]}
375: If condition~(\ref{e:gi:rho}) is fulfilled, gravitation will dominate the NTSI if
376: \begin{eqnarray}
377: n &>& \f{\gamma\,k_B}{4\pi G\mu^2}\,k^3\,\eta\,T\label{e:gi_ntsi:rho}\\
378: &\approx& 4.7\times10^2 \frac{\eta}{[\mbox{pc}]}\,\left(\frac{L}{[\mbox{pc}]}\right)^{-3}\,
379: \frac{T}{[\mbox{K}]}\,\mbox{cm}^{-3},\nonumber
380: \end{eqnarray}
381: where $\eta$ is the displacement of the slab. Note the strong dependence on
382: the wave number $k$, which hints at the power of dynamical fragmentation to
383: prevent global gravitational collapse.
384:
385: The condition for the KHI is similar to conditions~(\ref{e:gi:rho}) and (\ref{e:gi_ntsi:rho}).
386: The GI dominates over the KHI for
387: \begin{eqnarray}
388: n &>& \f{\gamma\,k_B}{4\pi G\mu^2}\,{\cal M}\,k^2\,T\label{e:gi_khi:rho}\\
389: &\approx& 7.4\times 10^1 {\cal M}\,\frac{T}{[\mbox{K}]}\,\left(\frac{L}{[\mbox{pc}]}\right)^{-2}\,\mbox{cm}^{-3},
390: \nonumber
391: \end{eqnarray}
392: where ${\cal M}$ is the Mach number of the shear flow. This is
393: the most beneficial case for the KHI: for supersonic flows,
394: the KHI is not unconditionally unstable anymore.
395:
396: Since conditions~(\ref{e:gi_ntsi:rho}) and (\ref{e:gi_khi:rho}) each depend
397: linearly on the temperature, they will behave similarly in the $(n,T)$, plane.
398: Without loss of generality, we can combine them in \S\ref{s:discussion}.
399:
400: \subsubsection{TI dominates NTSI and KHI [red solid]}
401: The TI dominates the NTSI for
402: \begin{eqnarray}
403: n &>&\left(\f{\gamma}{\mu}\right)^{1/2}\,k^{3/2}\,\eta^{1/2}\,\f{(k_BT)^{3/2}}{\Lambda(T)}\label{e:ti_ntsi:rho}\\
404: &\approx& 5.4\times 10^{-4}\left(\frac{\eta}{[\mbox{pc}]}\right)^{1/2}\left(\frac{L}{[\mbox{pc}]}\right)^{-3/2}
405: \left(\frac{T}{[\mbox{K}]}\right)^{3/2}\nonumber\\
406: &\times&\left(\frac{10^{26}\Lambda(T)}{[\mbox{erg cm$^3$ s$^{-1}$}]}\right)^{-1}
407: \,\mbox{cm}^{-3}.
408: \nonumber
409: \end{eqnarray}
410: This just mirrors the fact that with increasing density the cooling becomes stronger.
411: A similar condition can be derived for the KHI by replacing $(k\eta)^{1/2}$ by the
412: velocity difference multiplied by the wave number.
413:
414: \subsubsection{Sound crossing time limit for the TI [blue solid]}
415: %begin new
416: As discussed in \S\ref{ss:ti}, we use the sound crossing scale (eq.~\ref{e:cond_soundcross})
417: for the TI's condensation mode as a proxy to estimate the importance of the TI. It leads
418: %end new
419: to a density threshold above which
420: the condensation mode of the TI cannot be excited:
421: \begin{eqnarray}
422: n &<& \left(\f{\gamma}{\mu}\right)^{1/2}\,k\,\f{(k_BT)^{3/2}}{\Lambda(T)}\label{e:ti:rho}\\
423: &\approx& 2.2\times 10^{-4}\left(\frac{L}{[\mbox{pc}]}\right)^{-1}
424: \left(\frac{T}{[\mbox{K}]}\right)^{3/2}\left(\frac{10^{26}\Lambda(T)}{[\mbox{erg cm$^3$ s$^{-1}$}]}\right)^{-1}.
425: \nonumber
426: \end{eqnarray}
427: The smaller the scales, the less restrictive is the limit on the density for the
428: condensation mode. The temperature dependence is given by the detailed
429: form of the cooling curve.
430:
431: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
432: %
433: %\section{Discussion}
434: %
435: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
436: \section{Discussion}\label{s:discussion}
437:
438: The conditions summarized in \S\ref{ss:criteria} allow us to
439: identify the dominant instabilities in the two-dimensional $(n,T)$ parameter plane.
440: Most of the conditions (\ref{e:gi:rho})-(\ref{e:ti:rho}) depend on the
441: spatial scale. Figure~\ref{f:ntplane} summarizes the regimes in four diagrams,
442: at characteristic (fixed) scales of $0.1,1.0,10$ and $10^2$~pc.
443: Table~\ref{t:key} provides a key to the line styles and colors used in
444: Figure~\ref{f:ntplane}. We will discuss the dominant instabilities
445: (\S\ref{ss:TIdom}--\ref{ss:tivsdi}) and an evolutionary sequence
446: of a fluid parcel in the ISM (\S\ref{ss:evol}).
447:
448: \begin{figure*}
449: \begin{center}
450: \includegraphics[width=\textwidth]{f2.eps}
451: % \includegraphics[width=\textwidth]{../figures/rhotempplane_four.eps}
452: \end{center}
453: \caption{\label{f:ntplane}Summary of the instability regimes in
454: $(n,T)$-space, for four
455: characteristic spatial scales as indicated in the top left corners.
456: The diagonal black long-dashed lines denote lines
457: of constant pressure as indicated. The thick black line between $1<\log T<4$
458: represents the thermal equilibrium curve due to a balance between heating and
459: cooling. The horizontal thick black lines
460: at the bottom of each panel denote the ranges of the two thermally unstable regimes.
461: Fragmentation by cooling dominates in all regions with right-slanted hashing (red, blue
462: and black), while gravity dominates fragmentation in the left-slanted hashed region (green).
463: Red stands for dominance of dynamical instabilities over cooling/TI. Blue/black indicates
464: where fragmentation by cooling dominates. The blue ribbon and the black horizontal lines
465: together indicate ranges of strict thermal instability.
466: See Table~\ref{t:key} for a summary of the line styles and colors.}
467: \end{figure*}
468:
469: \begin{deluxetable*}{llc}
470: \tablewidth{0pt}
471: \tablecaption{Key to Figure~\ref{f:ntplane}\label{t:key}}
472: \tablehead{\colhead{line style}&\colhead{meaning}&\colhead{equation}}
473: \startdata
474: hashed, slanted right & fragmentation by cooling or TI & \\
475: hashed, slanted left & fragmentation by gravity & \\
476: red hashed & dynamics dominate cooling & \\
477: blue hashed & cooling/TI dominates dynamics & \\
478: black hashed & cooling (but not TI) dominates dynamics & \\
479: \hline
480: green dash & Jeans criterion & (\ref{e:gi:rho})\\
481: red dash & GI dominates NTSI& (\ref{e:gi_ntsi:rho})\\
482: red solid & TI dominates NTSI& (\ref{e:ti_ntsi:rho})\\
483: blue solid & TI crossing time & (\ref{e:ti:rho})
484: \enddata
485: % \tablecomments{}
486: \end{deluxetable*}
487:
488: \subsection{Thermally Dominated Regime\label{ss:TIdom}}
489: The most prominent feature in Figure~\ref{f:ntplane} is the large extent of
490: the thermally dominated parameter regime (right-slanted hashing; blue and black).
491: This regime is limited by the growing importance of dynamical instabilities (red) towards
492: low densities, and by the Jeans condition (eq.~\ref{e:gi:rho}) towards high densities.
493: Towards low densities (red), the thermal timescales get longer than the dynamical
494: timescales, so that the gas will tend to behave adiabatically.
495: Within the blue hashed region, cooling dominates dynamical instabilities, leading
496: to the TI within the temperature ranges indicated by the thick black
497: horizontal bars. Outside the strictly thermally unstable regions, cooling still
498: dominates and can lead to fragmentation when an external (ram or gravitational) pressure is applied.
499: Thus, while the TI dominates the dynamics only within certain temperature ranges,
500: thermal effects (strong cooling) continue to dominate all through the black
501: hashed region at higher densities. The upper edge of the blue ribbon is determined
502: by the sound crossing time condition~(\ref{e:ti:rho}), which is -- strictly -- only
503: applicable to the TI. There are
504: two thermally unstable regimes (according to eq.~[\ref{e:cond_ti}]):
505: The lower one between $10^2<T<10^4$~K connects the warm and cold neutral medium (WNM and CNM),
506: while the upper (of lesser strength) in the range of $T\approx 10^5$ connects the hot
507: and warm ionized medium (HIM and WIM).
508:
509: The curved thick black line between $10<T<10^4$~K denotes the thermal equilibrium curve,
510: where heating terms balance cooling terms. At $T\approx 10^4$~K and $n<1$~cm$^{-3}$, the
511: ISM behaves quasi-isothermal. The signature of the thermal instability is a pressure loss
512: with increasing density (compare to dashed lines of constant pressure). Moving to
513: high densities and lower temperatures, the effective equation of state tends back to
514: isothermality but stays sub-isothermal, i.e. $\gamma_{eff}<1$. The thermal equilibrium curve
515: is describing an approximate evolutionary sequence of a fluid element from the WNM to the CNM
516: (see \S\ref{ss:evol}).
517:
518: Before we compare the relative strengths of the instabilities,
519: Figure~\ref{f:tidetail} offers a more detailed view of the TI in the WNM. As orientation and
520: for comparison with Figure~\ref{f:ntplane}, the lines of constant pressure and
521: the sound crossing time limit (eq.~[\ref{e:ti:rho}]) for the TI have been indicated in
522: the same style as in Figure~\ref{f:ntplane}.
523: The color-shaded region denotes the growth timescale (eq.~[\ref{e:ts_TI}]) of the TI
524: (see color bar to the right of the diagrams) in all locations where the TI can be
525: excited. There are two points to notice here:
526:
527: \begin{figure*}
528: \begin{center}
529: \includegraphics[width=\textwidth]{f3.eps}
530: % \includegraphics[width=\textwidth]{../figures/rhotempplane_detail.eps}
531: \end{center}
532: \caption{\label{f:tidetail}A more detailed look at the TI in
533: $(n,T)$-space, for four characteristic spatial scales as indicated in the top
534: left corners. The diagonal black long-dashed lines denote lines
535: of constant pressure as indicated. The thick black line between $1<\log T<4$
536: represents the thermal equilibrium curve due to a balance between heating and
537: cooling. The horizontal thick black lines
538: at the bottom of each panel denote the ranges of the two thermally unstable regimes.
539: The blue line denotes the
540: maximum density given by the sound crossing time limit (eq.~[\ref{e:ti:rho}]).
541: The color-shaded region shows the growth timescale of the TI (eq.~[\ref{e:ts_TI}]).
542: Light green corresponds to a timescale of $\approx 1$~Myr.}
543: \end{figure*}
544:
545: (1) Looking at the upper left panel ($L=0.1$~pc),
546: the timescale $\tau_{TI}$ {\em can} become extremely short; however, around the
547: thermally unstable region, it is on the order of $1$~Myr (light green, traced
548: by the thermal equilibrium curve), comparable to dynamical
549: timescales. This leads to a noticeable amount of thermally unstable gas (e.g.
550: \citealp{2003ApJ...586.1067H} for observations of neutral hydrogen, and
551: \citealp{2001ApJ...557L.121G}, \citealp{2002ApJ...577..768S},
552: \citealp{2005A&A...433....1A}, \citealp{2006ApJ...648.1052H}
553: for evidence from numerical models).
554: For the isothermal branch at $T\approx 10^4$~K, the TI is essentially
555: absent. Thus, the lower the density on the (warm) isothermal branch, the higher the
556: compression due to e.g. shocks needs to be to trigger the TI. Since the post-shock
557: density in an isothermal gas scales with the square of the Mach number, this condition
558: is not overly restrictive.
559:
560: (2) With increasing spatial scale (solid blue line in the remaining three panels of
561: Fig.~\ref{f:tidetail}), the upper density limit for the TI drops due to the sound
562: crossing time condition~(\ref{e:ti:rho}). Strictly, the TI will only be excited if
563: there is a color-shaded region below the blue line, so to speak. This leaves us with
564: the somewhat puzzling result that for scales $L>10$~pc, the TI seemingly cannot be excited.
565: The solution is two-fold: (a) At large scales (and lower densities/higher temperatures),
566: the upper TI at $10^5$~K kicks in, and (b) the notion of fixed scales is misleading.
567: The TI will be triggered by compressions and/or turbulent mixing equivalent to a
568: reduction of spatial scale. Moreover, since it is a condensation mode,
569: we cannot regard the evolution of the TI at a fixed scale, but need to follow a
570: fluid parcel (see \S\ref{ss:evol}).
571:
572: \subsection{Gravitation against Thermal Instability\label{ss:gravvstherm}}
573: The Jeans criterion (eq.~[\ref{e:gi:rho}]) is fulfilled for all pairs $(n,T)$ above
574: the green dashed line in Figure~\ref{f:ntplane} -- e.g., for densities
575: $n>10^5$~cm$^{-3}$ and temperatures $T\approx 10$~K at $0.1$~pc. With increasing
576: scale, the $(n,T)$-regime dominated by
577: gravitation extends more and more to lower densities and higher temperatures.
578: At large scales ($\approx 10$~pc), gravity starts to dominate the parameter space,
579: {\em indicating the importance of global gravitational modes}.
580:
581: \begin{figure*}
582: \begin{center}
583: \includegraphics[width=\textwidth]{f4.eps}
584: % \includegraphics[width=\textwidth]{../figures/rhotempplane_pair.eps}
585: \end{center}
586: \caption{\label{f:tipair}Comparison of characteristic times for self-gravity and cooling in
587: $(n,T)$-space, for four
588: characteristic spatial scales as indicated in the top left corners.
589: The diagonal black long-dashed lines denote lines
590: of constant pressure as indicated. The thick black line between $1<\log T<4$
591: represents the thermal equilibrium curve due to a balance between heating and
592: cooling. The horizontal thick black lines
593: at the bottom of each panel denote the ranges of the two thermally unstable regimes.
594: The blue line denotes the
595: maximum density given by the sound crossing time limit (eq.~[\ref{e:ti:rho}]).
596: The color-shaded triangles show the ratio $\tau_c/\tau_{GI}$
597: (eqs.~[\ref{e:ts_cool}], [\ref{e:ts_GI}]) for the pairs of $(n,T)$
598: subject to the GI. Yellow would correspond to $\tau_c = \tau_{GI}$ -- obviously,
599: $\tau_c < \tau_{GI}$ for all gravitationally dominated pairs of $(n,T)$.}
600: \end{figure*}
601:
602: Figure~\ref{f:tipair} compares the relative strengths of gravitational over thermal
603: condensation\footnote{We are using ``thermal condensation'' for condensation
604: due to strong cooling, not -- as it occasionally happens in the literature -- for
605: gravitational collapse.}. Shown are the same plots as in the previous figures, but
606: now overlaid with the ratio $\tau_c/\tau_{GI}$ for the gravitationally dominated
607: $(n,T)$-regime (eq.~\ref{e:ts_GI}). Yellow (see color bar of $\tau_c/\tau_{GI}$)
608: denotes $\tau_c = \tau_{GI}$. The cooling timescales are substantially
609: shorter than any gravitational timescales over the whole parameter range accessible
610: to the GI. If the TI can be excited (e.g between $80<T<5000$~K), this means that
611: {\em thermal condensations will be the dominant fragmentation process}. For parameters
612: $(n,T)$ not subjected to the TI, the short cooling timescales mean only that any
613: excess energy due to compressions (by gravity!) will be efficiently radiated away.
614:
615: \subsection{Gravitational against dynamical effects\label{ss:givsdi}}
616: The relative importance of gravity and dynamics can be read off
617: equations~(\ref{e:gi:rho})--(\ref{e:gi_khi:rho}). The discerning line
618: for GI against NTSI is shown in Figure~\ref{f:ntplane} as a red dashed line
619: -- obviously, the NTSI ceases to be important before densities high enough
620: for gravitational collapse can be reached. Equations~(\ref{e:gi:rho}) and
621: (\ref{e:gi_khi:rho}) on the other hand show that for Mach numbers ${\cal M}>1$,
622: the KHI can still dominate in gravitationally unstable gas. This is the regime
623: of turbulent fragmentation in a Jeans-unstable medium discussed in detail
624: by \citet{2004RvMP...76..125M}.
625:
626: \subsection{Thermal against dynamical effects\label{ss:tivsdi}}
627: Dynamical instabilities dominate below the red (NTSI) and the black (KHI) lines
628: (Fig.~\ref{f:ntplane}), generally however in the red hashed region.
629: They impose a lower density threshold on the importance of
630: cooling (and of the TI within the applicable temperature range), i.e. the gas
631: will tend to behave more and more adiabatically for those regions in $(n,T)$-space.
632:
633: Figure~\ref{f:tidi} shows the ratio of the cooling time scale (eq.~[\ref{e:ts_cool}])
634: over $\tau_{NTSI}$ (eq.~[\ref{e:ti_ntsi:rho}]), where the latter is used as a proxy
635: of dynamical instabilities. Since the cooling timescale does not depend on the spatial
636: scale, dynamical instabilities can become dominant for small scales ($L=0.1\cdots 1.0$~pc)
637: in the WNM ($T\approx 10^4$~K). With increasing density, cooling dominates dynamics, but the
638: TI is still limited by condition~(\ref{e:ti:rho}).
639:
640: Dynamical instabilities entail the generation of turbulence --
641: especially at the high Reynolds numbers in the ISM. The turbulent cascade
642: leads to a spectrum of density perturbations, populating a wide range
643: of spatial scales. In other words: instead of proceeding at a fixed
644: scale, the TI will act on all density perturbations within its parameter range
645: (see Fig.~\ref{f:tidetail}) at scales smaller than the sound crossing scale simultaneously.
646:
647: \begin{figure*}
648: \begin{center}
649: \includegraphics[width=\textwidth]{f5.eps}
650: % \includegraphics[width=\textwidth]{../figures/rhotempplane_tidi.eps}
651: \end{center}
652: \caption{\label{f:tidi}Comparison of characteristic times for dynamical instabilities
653: and cooling in $(n,T)$-space, for four
654: characteristic spatial scales as indicated in the top left corners.
655: The diagonal black long-dashed lines denote lines
656: of constant pressure as indicated. The thick black line between $1<\log T<4$
657: represents the thermal equilibrium curve due to a balance between heating and
658: cooling. The horizontal thick black lines
659: at the bottom of each panel denote the ranges of the two thermally unstable regimes.
660: The blue line denotes the
661: maximum density given by the sound crossing time limit (eq.~[\ref{e:ti:rho}]).
662: The color-shaded regions show the ratio $\tau_c/\tau_{DI}$ (eqs.~[\ref{e:ts_cool}],
663: [\ref{e:ti_ntsi:rho}]). Note that the ratio decreases monotonically with increasing
664: density.}
665: \end{figure*}
666:
667: \subsection{An Evolutionary Sequence\label{ss:evol}}
668:
669: The strength and importance of the TI is based on the one hand on the fact that
670: it grows on the smallest scales first \citep{2000ApJ...537..270B} and thus
671: can feed on ubiquitous small-scale structures generated by turbulence. On the other
672: hand, the condensation mode of the TI does not proceed at a fixed scale:
673: a fluid parcel which gets thermally unstable will contract to smaller scales and thus
674: ``delay'' running into the sound crossing scale condition~(\ref{e:ti:rho}).
675:
676: To see this, consider the thermal equilibrium
677: curve (solid black curve in e.g. Figs.~\ref{f:ntplane} and \ref{f:tidetail}).
678: Numerical simulations of the fragmentation of shocked flows in a thermally unstable
679: medium demonstrate that the gas generally follows this thermal equilibrium curve
680: (see Fig.~3 of \citealp{2005A&A...433....1A}, and Fig.~14 of \citealp{2006ApJ...648.1052H}),
681: so that it can be interpreted as an evolutionary sequence from the WNM to the CNM.
682: While the equilibrium curve
683: is well defined at the high-density end due to extremely short cooling time scales
684: (see Fig.~\ref{f:tidetail}), the scatter especially in the thermally unstable regime can be
685: substantial and depends largely on the ratio between the dynamical and the cooling
686: timescales (Figs.~14 and 15 of \citealp{2006ApJ...648.1052H}).
687:
688: A fluid parcel\footnote{Strictly, the notion of a fluid {\em parcel} is slightly inappropriate.
689: As the numerical models consistently show, the density structures tend to be filamentary rather than
690: round. In that sense, the scale refers to the shortest axis of the filament/sheet.}
691: in the WNM/WIM starts out at a scale of e.g. $\sim 1$~pc and a density of a little
692: under $1$~cm$^{-3}$. The parcel may be compressed due to a shock wave, or
693: turbulence may lead to density variations. In any case, the density increases,
694: propelling the parcel upwards along the isothermal branch of the equilibrium curve,
695: reducing the cooling timescales (Fig.~\ref{f:tidetail}) and thus eventually lowering
696: the temperature. The fluid parcel will follow (approximately) the equilibrium
697: curve along its sharp left-turn into the thermally unstable regime. Looking at the
698: 1~pc panel, by then, the fluid parcel will have left the region of thermal instability
699: (Fig.~\ref{f:ntplane}). However, the compression above not only increased the density,
700: but also reduced the scale, so that effectively the thermally unstable regime (blue hashed)
701: shifts upwards to higher densities, moving us from the $1$~pc panel to the $0.1$~pc panel in
702: Figures~\ref{f:ntplane} or \ref{f:tidetail}.
703:
704: Once the fluid parcel has entered the thermally unstable regime, it evolves (more or less) along lines of
705: constant pressure towards higher densities and lower temperatures. In the
706: condensation mode, this is equivalent to a further shrinking of the fluid parcel's volume,
707: so that the blue ribbon of thermal instability essentially ``moves'' to smaller scales with the fluid parcel
708: on its way along the equilibrium curve.
709: This ``co-moving'' or Lagrangian behavior is crucial for the strength and the
710: importance of the condensation mode of the TI in the ISM.
711:
712: Yet the above statement calls for a caveat. A closer inspection of equation~(\ref{e:ti:rho}) reveals the
713: following. Assuming that the fluid parcel evolves under constant pressure $P_0$
714: -- actually, for dynamical times substantially longer than thermal timescales, the fluid parcel will
715: follow approximately the thermal equilibrium curve (\citealp{2002ApJ...577..768S};
716: and Fig.~\ref{f:ntplane}), which is sub-isobaric
717: as indicated by $\gamma_{eff}<0$ in the thermally unstable regime --
718: and with a constant cooling strength $\Lambda_0$, the sound crossing scale condition~(\ref{e:ti:rho}) translates into an
719: expression for the outer length scale $L_T$ of thermal instability,
720: \begin{equation}
721: L_T < a_0\,n^{-5/2}\label{e:sc_t},
722: \end{equation}
723: equivalent to equation~(\ref{e:cond_soundcross}), with the constant
724: \begin{equation}
725: a_0\equiv 2\pi\,\left(\frac{\gamma\,k_B}{\mu}\right)^{1/2}\,\frac{P_0^{3/2}}{\Lambda_0}.
726: \end{equation}
727: A spherical blob of constant mass $m_0$ will contract as
728: \begin{equation}
729: L_V = b_0\,n^{-1/3},\label{e:sc_v}
730: \end{equation}
731: with the constant
732: \begin{equation}
733: b_0\equiv\left(\frac{6\,m_0}{\pi\,\mu}\right)^{1/3}.
734: \end{equation}
735: Thus,
736: \begin{equation}
737: L_T \propto L_V\,n^{-13/6},
738: \end{equation}
739: meaning that the sound crossing scale (eq.~[\ref{e:cond_soundcross}]) shrinks more
740: rapidly by a factor of $\approx n^2$ than the size of the fluid parcel. Even
741: a one-dimensional compression $L_V\propto n^{-1}$ would still result in
742: $L_T\propto L_V\,n^{-3/2}$. This would move the fluid parcel out of the thermally
743: unstable regime. If the fluid parcel contracts ``faster''
744: than the sound crossing time criterion allows, the condensation
745: mode will be suppressed, but the isochoric mode of the TI still will be excited.
746: This is specifically the case for the black hashed regions (Fig.~\ref{f:ntplane})
747: where thermal timescales are substantially shorter than the dynamical timescales.
748: The thermal pressure of the fluid parcel can then drop below that of the surrounding medium,
749: and the resulting waves will lead to further fragmentation
750: (smaller scales) and compression, triggering again the condensation mode. In that
751: sense, while the co-moving picture is a simplification, the more realistic
752: evolution of a fluid parcel involving isobaric and isochoric modes will lead
753: to even more fragmentation.
754:
755: Eventually, our fluid parcel will reach the sub-isothermal branch of the thermal
756: equilibrium curve below $\approx 80$~K, entering the gravitationally dominated
757: regime (as discussed earlier, this regime is characterized also by extremely short
758: cooling timescales). In the presence of an ``external'' pressure (due to e.g.
759: gravity), the sub-isothermal equation of state can lead to further fragmentation.
760:
761: However, gravity will not only play a local role. At large scales, global gravity
762: can be dominant (Fig.~\ref{f:ntplane}, e.g. 10~pc). Global gravitational modes
763: in combination with a finite extent of the cloud can easily lead to the sweep-up
764: of material {\em at the edges} of the cloud (\citealp{2004ApJ...616..288B}; see
765: \citealp{2007ApJ...657..870V} and \citealp{2008ApJ...674..316H} for numerical
766: evidence). Due to the rapid evolution of the {\em local} (thermal and dynamical) instabilities,
767: the resulting filaments will have substructure, leading to further fragmentation into cores.
768:
769: The thermal instability thus offers a short-cut to reach the gravitationally dominated
770: regime from the warm, isothermal branch (at $T=10^4$~K). Instead of having to compress
771: the gas isothermally by a factor of $\approx 10^4$ -- corresponding to a Mach number
772: of $10^2$ -- only a modest compression is needed in the warm isothermal branch
773: to move the fluid parcel up the thermally stable equilibrium line and ``push it over the ledge''.
774: This allows star formation in environments without high pressurization and/or deep gravitational
775: potentials.
776:
777: To summarize, the generation of turbulence due to dynamical instabilities (red hashed regime)
778: introduces density variations at smaller scales, triggering the thermal instability, which
779: in turn amplifies the density perturbations. The growth of the thermal instability is eventually
780: limited by the mass reservoir available, and thus by
781: the external pressure \citep{2002ApJ...569L.127K,2002ApJ...580L..51K}. This could take the form
782: of global gravity, cloud collisions, expanding shells or galaxy mergers. Gravity acts locally
783: on the high-density seeds, allowing rapid collapse before edge effects of global gravity
784: \citep{2004ApJ...616..288B} overrun the local perturbations.
785:
786: \subsection{Limitations\label{ss:limitations}}
787:
788: Obvious limitations of the study include neglecting the effect of structure in the inflows,
789: a generic cooling curve, the missing time-dependence and neglecting magnetic fields.
790: Each of these is discussed in turn below.
791:
792: \subsubsection{The Structure-less Inflows}
793: Our analysis approaches the problem of structured cloud formation
794: from the angle of a ``worst case'' scenario, using the most
795: unfavorable initial conditions to generate turbulence and substructure, by considering
796: (as in our numerical work) the large-scale flows to be uniform. Obviously -- as pointed out earlier
797: \citep{2006ApJ...648.1052H} -- this assumption neglects the possibility that the
798: flows themselves already might contain substructure. This however is not problematic,
799: as our goal is to demonstrate that substructure can arise from uniform flows with
800: only small, long-wavelength initial perturbations.
801: The introduction of structure in the inflows -- specifically a mixture of WNM/CNM instead
802: of just WNM -- would make it even easier to form substructure. Moreover, introducing
803: substructure in the inflows raises the question of just what form should be assumed.
804: The most common initial conditions for numerical simulations of cloud formation
805: comprise density perturbations (e.g. \citealp{2005AIPC..784..318I}), velocity
806: ``noise'' (e.g. \citealp{2007ApJ...657..870V}), or a turbulent velocity distribution
807: (e.g. \citealp{2007A&A...465..445H}), but the theoretical justification of these assumptions
808: is unclear. Moreover, it is far from obvious that
809: driving mechanisms such as that of a spherical stellar wind, H II region expansion, or
810: supernova bubble expansion are highly turbulent and structured in the absence of
811: interaction with the surrounding medium. By ignoring substructure in the inflows we
812: can consider the most general (and unfavorable) cases for generating turbulent substructure.
813:
814: \subsubsection{Cooling Curve\label{sss:limitcool}}
815: Our cooling curve at low temperatures is mainly limited by the fact that it does
816: not include molecular line cooling or the formation of molecules. Molecular cooling
817: would lower the cooling timescales further, driving the cooling curve more to
818: isothermality. The formation of H$_2$ lowers the sound speed by a factor of $\sqrt{2}$
819: and the pressure by $2$, thus introducing an additional pressure loss at high densities.
820: While the thermal equilibrium curve will describe the evolution of a fluid parcel
821: approximately correctly from the WNM to the CNM, the details of the evolution at
822: high densities and low temperature are less realistic.
823:
824: The heating and cooling functions $\Gamma$ and $\Lambda$ were defined for solar
825: abundances and a Galactic UV background. Reducing the metallicity by e.g. a
826: factor of $10$ to values typical for the SMC reduces the cooling strength
827: below $T\approx 10^4$~K by approximately the same factor (see Fig.~\ref{f:coolcurve}).
828: Thus, the blue ribbon in Figure~\ref{f:ntplane} would move up to higher densities.
829:
830: \subsubsection{Time dependence}
831: Obviously, the analysis does not include any time evolution (although
832: a rough idea of the dynamical evolution can be gleaned from following
833: the thermal equilibrium curve connecting the WNM and CNM, \S\ref{ss:evol}).
834: Dynamical effects (turbulence, shocks) are expected to render thermal effects
835: (cooling, TI) even stronger, specifically by providing small-scale seeds, and by
836: allowing the TI to act on many scales simultaneously. This can clearly be seen in the high-resolution
837: two-dimensional simulations by \citet{2007A&A...465..445H} and \citet{2007A&A...465..431H}.
838: For a detailed discussion of (fully
839: developed) turbulence in combination with the TI, see e.g. \citet{2003LNP...614..213V}
840: and \citet{2005A&A...433....1A}. The latter authors point out that increasing the level of
841: turbulence will lower the CNM fraction, an effect depending on the ratio of dynamical over
842: thermal timescales (see also \citealp{2006ApJ...648.1052H}).
843:
844: Certain simplifications had to be made, e.g. the Mach number
845: for the KHI (eq.~\ref{e:ts_KHI}) was set to $1$.
846: While this is a good approximation for many regimes in the ISM that exhibit trans-sonic
847: flows, it will fail in cases of extreme shear flow conditions such as HVCs or
848: molecular clouds (however, see Fig.~9 of \citealp{2006ApJ...648.1052H}).
849:
850: \subsubsection{Magnetic Fields}
851: Magnetic fields are not included.
852: In relevance to the flow-driven cloud formation scenario,
853: they would introduce a threshold criterion
854: for the KHI \citep{1961hhs..book.....C} and the NTSI \citep{2007ApJ...665..445H},
855: and they may affect the evolution of the TI \citep{1965ApJ...142..531F}.
856: Critical field strengths and angles for the suppression of the condensation mode
857: in one-dimensional converging flows have been identified by \citep{2000A&A...359.1124H}.
858: Two-dimensional models of flow-driven cloud formation with the magnetic field
859: perpendicular to the inflows \citep{2008arXiv0801.0486I} demonstrate that while
860: the fields can suppress the formation of high-density clouds, thermal fragmentation
861: still occurs, leading to highly filamentary cold HI clouds. These results lighten
862: somewhat the restrictive low limits on the perpendicular field strengths derived from
863: one-dimensional arguments \citep{2004ApJ...612..921B}.
864: On a larger (galactic spiral arm) scale, numerical models by
865: \citet{2003ApJ...599.1157K} and \citet{2006ApJ...646..213K} demonstrate the importance of magnetic
866: fields for channeling gas streams.
867:
868: %\newpage
869:
870: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
871: %
872: %\section{Summary}
873: %
874: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
875: \section{Summary}\label{s:summary}
876:
877: To form stars, molecular clouds need to fragment
878: on scales vastly smaller than their overall dimensions. Self-gravity in combination
879: with supersonic turbulence allows the rapid fragmentation of Jeans-unstable
880: regions \citep{1981MNRAS.194..809L,1993ApJ...419L..29E,2004RvMP...76..125M}.
881: Yet unless the density perturbations in molecular
882: clouds are non-linear {\em before} the onset of global gravitational collapse,
883: global gravity tends to win and ``overrun'' any local perturbation \citep{2004ApJ...616..288B}.
884: Thus, a strong fragmentation mechanism is needed {\em during the formation of the
885: cloud} to provide {\em simultaneously} the observed turbulence as well as
886: the initial density seeds for the rapid {\em local} gravitational
887: collapse mandated by the observed rapid onset of star formation
888: \citep{2001ApJ...562..852H,2002ApJ...578..914H}.
889:
890: The flow-driven formation of molecular clouds in large-scale flows
891: of atomic hydrogen provides a natural fragmentation mechanism due to
892: a combination of dynamical and strong thermal instabilities (see references in
893: \S\ref{s:motivation}). As
894: demonstrated by numerical models, this fragmentation
895: mechanism does not need to recur on imposed turbulent velocity fields, but arises
896: from the physical conditions in the WNM/CNM. The rapid fragmentation of the
897: shocked flows will form high-density seeds for local gravity to take over,
898: while gravitational forces on the cloud scale can sweep up material
899: into filaments \citep{2008ApJ...674..316H}.
900:
901: The above outline of the connection between cloud formation and the initial
902: conditions for star formation mainly rests on evidence from numerical simulations. In
903: the present study we discussed the timescales of the instabilities
904: dominating the evolution of shocked flows. Guided
905: by our earlier work, we identify four instabilities, namely two condensation
906: mechanisms (gravity and thermal instability) and two fluid instabilities
907: (NTSI and KHI). We determine the parameter regime in density, temperature and
908: scale in which each of the instabilities is expected to dominate.
909: These are our findings and their implications:
910:
911: \begin{enumerate}
912: \item The (dynamical and thermal) instabilities leading to the rapid fragmentation
913: of shocked flows dominate on small scales for reasonable parameters of the
914: WNM and CNM (Fig.~\ref{f:ntplane}).
915: \item Cooling timescales (of $\approx 1$~Myr
916: in the thermally unstable regime) are generally substantially shorter than
917: gravitational timescales for that same density and temperature regime
918: (Fig.~\ref{f:tipair}). Thus, thermal fragmentation dominates gravitational
919: fragmentation during cloud formation.
920: \item The strength of the condensation mode of the TI derives from two effects,
921: namely that it grows on the smallest scales
922: first, and that it is essentially a co-moving
923: (or Lagrangian) mode, i.e. the unstable region shrinks, thus keeping
924: ``longer'' below the sound crossing scale $\lambda_c$ (eq.~[\ref{e:ti:rho}]). Feeding on turbulent density
925: perturbations below $\lambda_c$ will allow the TI to grow on many scales simultaneously.
926: \item The preference of small scales by the TI entails the preference to form
927: sheets and filaments: the shorter axes are more susceptible to become unstable.
928: This is an obvious parallel to the GI, with the -- equally obvious but crucial -- difference
929: that the GI will act globally {\em and} locally.
930: \item The fragmentation due to thermal effects is crucial for the formation of
931: small-scale, high-density perturbations {\em during the formation of the
932: molecular cloud}, to provide the seeds for rapid {\em local} gravitational
933: collapse before global gravity dominates the dynamics completely.
934: \item The all-important role of the TI due to the intermittent nature of the turbulence
935: driven by dynamical instabilities allows an early fragmentation of the
936: converging flows,
937: before (local) gravity can take over. This supports numerical findings that
938: the (molecular) core mass spectrum might be set to some extent
939: \citep{2008arXiv0801.2257D} early on during cloud formation
940: by thermal fragmentation \citep{2007A&A...465..445H,2008ApJ...674..316H}.
941: \end{enumerate}
942:
943: \acknowledgements
944: We thank the referee for a critical and timely report.
945: The arguments presented lean heavily on
946: the evidence from numerical simulations performed at the National Center for
947: Supercomputing Applications (AST 060031). FH is supported by the University
948: of Michigan and NASA grant NNG06GJ32G.
949: This work has made use of NASA's Astrophysics Data System.
950:
951: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
952: %
953: %\references
954: %
955: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
956:
957: %\bibliographystyle{apj}
958: %\bibliography{./references}
959:
960: \begin{thebibliography}{54}
961: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
962:
963: \bibitem[{{Audit} \& {Hennebelle}(2005)}]{2005A&A...433....1A}
964: {Audit}, E. \& {Hennebelle}, P. 2005, \aap, 433, 1
965:
966: \bibitem[{{Balbus}(1995)}]{1995ASPC...80..328B}
967: {Balbus}, S.~A. 1995, in Astronomical Society of the Pacific Conference Series,
968: Vol.~80, The Physics of the Interstellar Medium and Intergalactic Medium, ed.
969: A.~{Ferrara}, C.~F. {McKee}, C.~{Heiles}, \& P.~R. {Shapiro}, 328
970:
971: \bibitem[{{Ballesteros-Paredes} \& {Hartmann}(2007)}]{2007RMxAA..43..123B}
972: {Ballesteros-Paredes}, J. \& {Hartmann}, L. 2007, Revista Mexicana de
973: Astronomia y Astrofisica, 43, 123
974:
975: \bibitem[{{Ballesteros-Paredes} {et~al.}(1999){Ballesteros-Paredes},
976: {V{\'a}zquez-Semadeni}, \& {Scalo}}]{1999ApJ...515..286B}
977: {Ballesteros-Paredes}, J., {V{\'a}zquez-Semadeni}, E., \& {Scalo}, J. 1999,
978: \apj, 515, 286
979:
980: \bibitem[{{Bergin} {et~al.}(2004){Bergin}, {Hartmann}, {Raymond}, \&
981: {Ballesteros-Paredes}}]{2004ApJ...612..921B}
982: {Bergin}, E.~A., {Hartmann}, L.~W., {Raymond}, J.~C., \& {Ballesteros-Paredes},
983: J. 2004, \apj, 612, 921
984:
985: \bibitem[{{Blondin} \& {Marks}(1996)}]{1996NewA....1..235B}
986: {Blondin}, J.~M. \& {Marks}, B.~S. 1996, New Astronomy, 1, 235
987:
988: \bibitem[{{Burkert} \& {Hartmann}(2004)}]{2004ApJ...616..288B}
989: {Burkert}, A. \& {Hartmann}, L. 2004, \apj, 616, 288
990:
991: \bibitem[{{Burkert} \& {Lin}(2000)}]{2000ApJ...537..270B}
992: {Burkert}, A. \& {Lin}, D.~N.~C. 2000, \apj, 537, 270
993:
994: \bibitem[{{Chandrasekhar}(1961)}]{1961hhs..book.....C}
995: {Chandrasekhar}, S. 1961, {Hydrodynamic and hydromagnetic stability}
996: (International Series of Monographs on Physics, Oxford: Clarendon, 1961)
997:
998: \bibitem[{{Cox}(2005)}]{2005ARA&A..43..337C}
999: {Cox}, D.~P. 2005, \araa, 43, 337
1000:
1001: \bibitem[{{Dalgarno} \& {McCray}(1972)}]{1972ARA&A..10..375D}
1002: {Dalgarno}, A. \& {McCray}, R.~A. 1972, \araa, 10, 375
1003:
1004: \bibitem[{{Dib} {et~al.}(2008){Dib}, {Brandenburg}, {Kim}, {Gopinathan}, \&
1005: {Andre}}]{2008arXiv0801.2257D}
1006: {Dib}, S., {Brandenburg}, A., {Kim}, J., {Gopinathan}, M., \& {Andre}, P. 2008,
1007: ArXiv e-prints, 801
1008:
1009: \bibitem[{{Dobbs} \& {Bonnell}(2007)}]{2007MNRAS.376.1747D}
1010: {Dobbs}, C.~L. \& {Bonnell}, I.~A. 2007, \mnras, 376, 1747
1011:
1012: \bibitem[{{Dobbs} \& {Bonnell}(2008)}]{2008MNRAS.tmp..285D}
1013: ---. 2008, \mnras, 285
1014:
1015: \bibitem[{{Elmegreen}(1993)}]{1993ApJ...419L..29E}
1016: {Elmegreen}, B.~G. 1993, \apjl, 419, L29+
1017:
1018: \bibitem[{{Elmegreen}(2007)}]{2007ApJ...668.1064E}
1019: ---. 2007, \apj, 668, 1064
1020:
1021: \bibitem[{{Elmegreen} \& {Lada}(1977)}]{1977ApJ...214..725E}
1022: {Elmegreen}, B.~G. \& {Lada}, C.~J. 1977, \apj, 214, 725
1023:
1024: \bibitem[{{Field}(1965)}]{1965ApJ...142..531F}
1025: {Field}, G.~B. 1965, \apj, 142, 531
1026:
1027: \bibitem[{{Garnett}(1999)}]{1999IAUS..190..266G}
1028: {Garnett}, D.~R. 1999, in IAU Symposium, Vol. 190, New Views of the Magellanic
1029: Clouds, ed. Y.-H. {Chu}, N.~{Suntzeff}, J.~{Hesser}, \& D.~{Bohlender},
1030: 266--272
1031:
1032: \bibitem[{{Gazol} {et~al.}(2001){Gazol}, {V{\'a}zquez-Semadeni},
1033: {S{\'a}nchez-Salcedo}, \& {Scalo}}]{2001ApJ...557L.121G}
1034: {Gazol}, A., {V{\'a}zquez-Semadeni}, E., {S{\'a}nchez-Salcedo}, F.~J., \&
1035: {Scalo}, J. 2001, \apjl, 557, L121
1036:
1037: \bibitem[{{Gerwin}(1968)}]{1968RvMP...40..652G}
1038: {Gerwin}, R.~A. 1968, Reviews of Modern Physics, 40, 652
1039:
1040: \bibitem[{{Hartmann}(2002)}]{2002ApJ...578..914H}
1041: {Hartmann}, L. 2002, \apj, 578, 914
1042:
1043: \bibitem[{{Hartmann} {et~al.}(2001){Hartmann}, {Ballesteros-Paredes}, \&
1044: {Bergin}}]{2001ApJ...562..852H}
1045: {Hartmann}, L., {Ballesteros-Paredes}, J., \& {Bergin}, E.~A. 2001, \apj, 562,
1046: 852
1047:
1048: \bibitem[{{Hartmann} \& {Burkert}(2007)}]{2007ApJ...654..988H}
1049: {Hartmann}, L. \& {Burkert}, A. 2007, \apj, 654, 988
1050:
1051: \bibitem[{{Heiles} \& {Troland}(2003)}]{2003ApJ...586.1067H}
1052: {Heiles}, C. \& {Troland}, T.~H. 2003, \apj, 586, 1067
1053:
1054: \bibitem[{{Heitsch} {et~al.}(2005){Heitsch}, {Burkert}, {Hartmann}, {Slyz}, \&
1055: {Devriendt}}]{2005ApJ...633L.113H}
1056: {Heitsch}, F., {Burkert}, A., {Hartmann}, L.~W., {Slyz}, A.~D., \& {Devriendt},
1057: J.~E.~G. 2005, \apjl, 633, L113
1058:
1059: \bibitem[{{Heitsch} {et~al.}(2008){Heitsch}, {Hartmann}, {Slyz}, {Devriendt},
1060: \& {Burkert}}]{2008ApJ...674..316H}
1061: {Heitsch}, F., {Hartmann}, L.~W., {Slyz}, A.~D., {Devriendt}, J.~E.~G., \&
1062: {Burkert}, A. 2008, \apj, 674, 316
1063:
1064: \bibitem[{{Heitsch} {et~al.}(2006){Heitsch}, {Slyz}, {Devriendt}, {Hartmann},
1065: \& {Burkert}}]{2006ApJ...648.1052H}
1066: {Heitsch}, F., {Slyz}, A.~D., {Devriendt}, J.~E.~G., {Hartmann}, L.~W., \&
1067: {Burkert}, A. 2006, \apj, 648, 1052
1068:
1069: \bibitem[{{Heitsch} {et~al.}(2007){Heitsch}, {Slyz}, {Devriendt}, {Hartmann},
1070: \& {Burkert}}]{2007ApJ...665..445H}
1071: ---. 2007, \apj, 665, 445
1072:
1073: \bibitem[{{Hennebelle} \& {Audit}(2007)}]{2007A&A...465..431H}
1074: {Hennebelle}, P. \& {Audit}, E. 2007, \aap, 465, 431
1075:
1076: \bibitem[{{Hennebelle} {et~al.}(2007){Hennebelle}, {Audit}, \&
1077: {Miville-Desch{\^e}nes}}]{2007A&A...465..445H}
1078: {Hennebelle}, P., {Audit}, E., \& {Miville-Desch{\^e}nes}, M.-A. 2007, \aap,
1079: 465, 445
1080:
1081: \bibitem[{{Hennebelle} \& {P{\'e}rault}(1999)}]{1999A&A...351..309H}
1082: {Hennebelle}, P. \& {P{\'e}rault}, M. 1999, \aap, 351, 309
1083:
1084: \bibitem[{{Hennebelle} \& {P{\'e}rault}(2000)}]{2000A&A...359.1124H}
1085: ---. 2000, \aap, 359, 1124
1086:
1087: \bibitem[{{Hueckstaedt}(2003)}]{2003NewA....8..295H}
1088: {Hueckstaedt}, R.~M. 2003, New Astronomy, 8, 295
1089:
1090: \bibitem[{{Inoue} \& {Inutsuka}(2008)}]{2008arXiv0801.0486I}
1091: {Inoue}, T. \& {Inutsuka}, S.-i. 2008, ArXiv e-prints, 801
1092:
1093: \bibitem[{{Inoue} {et~al.}(2006){Inoue}, {Inutsuka}, \&
1094: {Koyama}}]{2006ApJ...652.1331I}
1095: {Inoue}, T., {Inutsuka}, S.-i., \& {Koyama}, H. 2006, \apj, 652, 1331
1096:
1097: \bibitem[{{Inutsuka} {et~al.}(2005){Inutsuka}, {Koyama}, \&
1098: {Inoue}}]{2005AIPC..784..318I}
1099: {Inutsuka}, S.-I., {Koyama}, H., \& {Inoue}, T. 2005, in American Institute of
1100: Physics Conference Series, Vol. 784, Magnetic Fields in the Universe: From
1101: Laboratory and Stars to Primordial Structures., ed. E.~M. {de Gouveia dal
1102: Pino}, G.~{Lugones}, \& A.~{Lazarian}, 318--328
1103:
1104: \bibitem[{{Kim} \& {Ostriker}(2006)}]{2006ApJ...646..213K}
1105: {Kim}, W.-T. \& {Ostriker}, E.~C. 2006, \apj, 646, 213
1106:
1107: \bibitem[{{Kim} {et~al.}(2003){Kim}, {Ostriker}, \&
1108: {Stone}}]{2003ApJ...599.1157K}
1109: {Kim}, W.-T., {Ostriker}, E.~C., \& {Stone}, J.~M. 2003, \apj, 599, 1157
1110:
1111: \bibitem[{{Koyama} \& {Inutsuka}(2000)}]{2000ApJ...532..980K}
1112: {Koyama}, H. \& {Inutsuka}, S.-I. 2000, \apj, 532, 980
1113:
1114: \bibitem[{{Koyama} \& {Inutsuka}(2002)}]{2002ApJ...564L..97K}
1115: {Koyama}, H. \& {Inutsuka}, S.-i. 2002, \apjl, 564, L97
1116:
1117: \bibitem[{{Koyama} \& {Inutsuka}(2004)}]{2004ApJ...602L..25K}
1118: ---. 2004, \apjl, 602, L25
1119:
1120: \bibitem[{{Kritsuk} \& {Norman}(2002{\natexlab{a}})}]{2002ApJ...580L..51K}
1121: {Kritsuk}, A.~G. \& {Norman}, M.~L. 2002{\natexlab{a}}, \apjl, 580, L51
1122:
1123: \bibitem[{{Kritsuk} \& {Norman}(2002{\natexlab{b}})}]{2002ApJ...569L.127K}
1124: ---. 2002{\natexlab{b}}, \apjl, 569, L127
1125:
1126: \bibitem[{{Larson}(1981)}]{1981MNRAS.194..809L}
1127: {Larson}, R.~B. 1981, \mnras, 194, 809
1128:
1129: \bibitem[{{Mac Low} \& {Klessen}(2004)}]{2004RvMP...76..125M}
1130: {Mac Low}, M.-M. \& {Klessen}, R.~S. 2004, Reviews of Modern Physics, 76, 125
1131:
1132: \bibitem[{{Palotti} {et~al.}(2008){Palotti}, {Heitsch}, {Zweibel}, \&
1133: {Huang}}]{2008arXiv0802.2497P}
1134: {Palotti}, M.~L., {Heitsch}, F., {Zweibel}, E.~G., \& {Huang}, Y.~. 2008, ArXiv
1135: e-prints, 802
1136:
1137: \bibitem[{{S{\'a}nchez-Salcedo} {et~al.}(2002){S{\'a}nchez-Salcedo},
1138: {V{\'a}zquez-Semadeni}, \& {Gazol}}]{2002ApJ...577..768S}
1139: {S{\'a}nchez-Salcedo}, F.~J., {V{\'a}zquez-Semadeni}, E., \& {Gazol}, A. 2002,
1140: \apj, 577, 768
1141:
1142: \bibitem[{{Sutherland} \& {Dopita}(1993)}]{1993ApJS...88..253S}
1143: {Sutherland}, R.~S. \& {Dopita}, M.~A. 1993, \apjs, 88, 253
1144:
1145: \bibitem[{{V{\'a}zquez-Semadeni} {et~al.}(2003){V{\'a}zquez-Semadeni}, {Gazol},
1146: {Passot}, \& {et al.}}]{2003LNP...614..213V}
1147: {V{\'a}zquez-Semadeni}, E., {Gazol}, A., {Passot}, T., \& {et al.} 2003, in
1148: Lecture Notes in Physics, Berlin Springer Verlag, Vol. 614, Turbulence and
1149: Magnetic Fields in Astrophysics, ed. E.~{Falgarone} \& T.~{Passot}, 213--251
1150:
1151: \bibitem[{{V{\'a}zquez-Semadeni} {et~al.}(2007){V{\'a}zquez-Semadeni},
1152: {G{\'o}mez}, {Jappsen}, {Ballesteros-Paredes}, {Gonz{\'a}lez}, \&
1153: {Klessen}}]{2007ApJ...657..870V}
1154: {V{\'a}zquez-Semadeni}, E., {G{\'o}mez}, G.~C., {Jappsen}, A.~K.,
1155: {Ballesteros-Paredes}, J., {Gonz{\'a}lez}, R.~F., \& {Klessen}, R.~S. 2007,
1156: \apj, 657, 870
1157:
1158: \bibitem[{{V{\'a}zquez-Semadeni} {et~al.}(2006){V{\'a}zquez-Semadeni}, {Ryu},
1159: {Passot}, {Gonz{\'a}lez}, \& {Gazol}}]{2006ApJ...643..245V}
1160: {V{\'a}zquez-Semadeni}, E., {Ryu}, D., {Passot}, T., {Gonz{\'a}lez}, R.~F., \&
1161: {Gazol}, A. 2006, \apj, 643, 245
1162:
1163: \bibitem[{{Vishniac}(1994)}]{1994ApJ...428..186V}
1164: {Vishniac}, E.~T. 1994, \apj, 428, 186
1165:
1166: \bibitem[{{Wolfire} {et~al.}(1995){Wolfire}, {Hollenbach}, {McKee}, {Tielens},
1167: \& {Bakes}}]{1995ApJ...443..152W}
1168: {Wolfire}, M.~G., {Hollenbach}, D., {McKee}, C.~F., {Tielens}, A.~G.~G.~M., \&
1169: {Bakes}, E.~L.~O. 1995, \apj, 443, 152
1170:
1171: \end{thebibliography}
1172:
1173:
1174: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1175: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1176: \end{document}
1177: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1178: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1179:
1180:
1181: