0709.0445/ms.tex
1:  \documentclass[12pt,preprint]{aastex}
2: 
3: \shorttitle{Magnetic Braking \& Protostellar Disks}
4: \shortauthors{Mellon \& Li}
5: 
6: \begin{document}
7: 
8: \title{Magnetic Braking and Protostellar Disk Formation: \\  
9: The Ideal MHD Limit} 
10: 
11: 
12: \author{Richard R. Mellon$\!$\altaffilmark{1} \&  
13: Zhi-Yun Li$\!$\altaffilmark{1} 
14: \altaffiltext{1}{Astronomy Department, University of Virginia, 
15: Charlottesville, VA 22904; rrm8p, zl4h@virginia.edu} }
16: 
17: \begin{abstract}
18: 
19: Magnetic fields are usually considered dynamically important in star formation 
20: when the dimensionless mass-to-flux ratio is close to, or less than, unity
21: ($\lambda~\lesssim~1$). We show that, in disk formation, the requirement is
22: far less stringent. This conclusion is drawn from a set of 2D
23: (axisymmetric) simulations of the collapse of rotating, singular isothermal 
24: cores magnetized to different degrees. We find that a weak field 
25: corresponding to 
26: $\lambda~ \sim 100$ can begin to disrupt the rotationally supported disk 
27: through magnetic braking, by creating regions of rapid, supersonic collapse 
28: in the disk. These regions are separated by one or more centrifugal barriers, 
29: where the rapid infall is temporarily halted. The number of centrifugal 
30: barriers increases with the mass-to-flux ratio $\lambda$. When $\lambda~ 
31: \gtrsim ~100$, they merge together to form a more or less 
32: contiguous, rotationally supported disk. Even though the magnetic
33: field in such a case is extremely weak on the scale of dense cores, 
34: it is amplified by collapse and differential rotation, to the 
35: extent that its pressure dominates the thermal pressure in both the 
36: disk and its surrounding region. For relatively strongly magnetized 
37: cores with $\lambda ~\lesssim ~10$, the disk formation is suppressed 
38: completely, as found previously. A new feature is that the mass 
39: accretion is highly episodic, due to reconnection of the magnetic 
40: field lines accumulated near the center. For rotationally
41: supported disks to appear during the protostellar mass accretion 
42: phase of star formation in dense cores with realistic 
43: field strengths, the powerful magnetic brake must be weakened, perhaps 
44: through nonideal MHD effects. Another possibility is to remove, 
45: through protostellar winds, the material that acts to brake the 
46: disk rotation. We discuss the 
47: possibility of observing a generic product of the magnetic braking, 
48: an extended circumstellar region that is supported by a combination 
49: of toroidal magnetic field and rotation --- a ``magnetogyrosphere'' --- 
50: interferometrically.
51: 
52: \end{abstract}
53: 
54: \keywords{accretion disks --- ISM: molecular clouds and magnetic 
55: fields --- MHD --- stars: formation}
56: 
57: \section{Introduction}
58: 
59: Disks are commonly observed around low-mass young stellar objects. They 
60: play a central role in star formation. It is generally 
61: believed that most of the mass of a Sun-like star is assembled through 
62: a disk \citep*{1987ARA&A..25...23S}. The disks may also be responsible for 
63: regulating the rotation rates of young stars, through disk-stellar 
64: magnetosphere interaction \citep{1991ApJ...370L..39K}, and 
65: for launching powerful 
66: jets and winds through the magnetocentrifugal mechanism 
67: \citep{2000prpl.conf..759K,2000prpl.conf..789S}. 
68: They are also the birthplace of planets.  
69: 
70: Low-mass stars form in dense cores of molecular clouds. The cores are
71: observed to rotate slowly \citep{1993ApJ...406..528G}. It is generally 
72: expected  
73: that the conservation of angular momentum during the core collapse 
74: would automatically lead to the formation of a rotationally supported 
75: disk. The expectation is borne out by both semi-analytic calculations 
76: \citep*[e.g.,][]{1984ApJ...286..529T} and numerical simulations (\citealp[e.g.,][]{1990ApJ...355..651B}; 
77: for reviews of early work, see \citealp{1995ARA&A..33..199B} and
78: \citealp{1998ASPC..148..314B}). 
79: However, the median value of the angular momenta 
80: measured for the cores \citep{1993ApJ...406..528G} is much higher than 
81: that inferred for circumstellar disks \citep{1993prpl.conf..521B}, 
82: by an order of magnitude or more \citep{1995ARA&A..33..199B}. Apparently, a 
83: considerable amount of angular momentum must be lost from the system 
84: in the process of core collapse and disk formation. 
85: 
86: A well-known mechanism for angular momentum removal is magnetic braking. 
87: Magnetic fields are believed to play an 
88: important role in the formation of low-mass stars in relative isolation  
89: \citep{1987ARA&A..25...23S,1993prpl.conf..327M,1999osps.conf..305M}.  
90: Polarization maps of 
91: dust continuum emission have revealed ordered magnetic fields in many
92: dense cores (see \citealp*{2006Sci...313..812G} for a spectacular example). The 
93: field strengths have been determined through Zeeman measurements in a 
94: number of them \citep{2007IAUS..237..141C}. The importance of magnetic 
95: field is usually measured by 
96: the mass-to-flux ratio $\lambda$ in units of the critical value 
97: $(2\pi G^{1/2})^{-1}$. For L1544,  
98: arguably the best studied starless core 
99: \citep{1998ApJ...504..900T,2007A&A...470..221C}, \citet{2000ApJ...537L.139C}
100:  inferred
101: $\lambda \sim 8$ using the line-of-sight components of the field
102: strength and column density. Correcting for projection effects may 
103: reduce $\lambda$ by a factor of 2 or more. In any case, there is 
104: ample observational evidence for both magnetic fields and rotation 
105: on the core scale. How the fields affect disk formation through 
106: magnetic braking of rotation is the focus of our investigation. 
107: 
108: Previous studies of magnetic braking have mostly concentrated on the 
109: core formation phase of cloud evolution prior to the formation of 
110: a central object \citep{1979ApJ...230..204M,1979MNRAS.187..337M,1989MNRAS.241..495N,1994ApJ...432..720B} 
111: or stopped shortly 
112: after a ``seed'' star has formed
113: (\citealp*{1998ApJ...502L.163T,2005A&A...435..385Z,2006ApJ...641..949B, 
114: 2006A&A...457..371F,2007ApJ_MIM_sub};  
115: see, however, recent work by \citealp{2007A&A_HF_sub} and \citealp{2007A&A_HT_sub}).
116: What happens in the later, main protostellar mass accretion phase is 
117: less explored. \citet*[hereafter
118:   ALS03]{2003ApJ...599..363A} 
119: followed numerically the protostellar collapse of 
120: moderately strongly magnetized cores 
121: (with the dimensionless mass-to-flux ratio $\lambda \leq 10$), and 
122: found that rotationally supported disks were not able to form in the 
123: ideal MHD limit. In hindsight, the 
124: suppression of disk formation is not too surprising, given that any 
125: disk that did form would be magnetically linked to a slowly rotating 
126: envelope of much larger moment of inertia and would lose its angular
127: momentum quickly through magnetic braking. The braking is greatly 
128: enhanced by the equatorial pinching of field lines, which lengthens 
129: the magnetic lever arms \citep[ALS03,][]{2006ApJ...647..374G}. 
130: 
131: The catastrophic magnetic braking stood the traditional ``angular 
132: momentum problem'' of star formation on its head; the braking can 
133: be so efficient as to inhibit disk formation for a moderate level 
134: of cloud magnetization in the ideal MHD limit. The question is: 
135: under what condition will the  
136: rotationally supported disk reappear? To address this question, 
137: we go beyond ALS03 in two ways. First, we follow the collapse of 
138: magnetized, rotating dense cores using the spherical (as opposed 
139: to the cylindrical) version of the Zeus2D MHD code \citep{1992ApJS...80..753S,1992ApJS...80..791S}. 
140: The spherical geometry enables us to better resolve the 
141: angular distributions of the matter and magnetic field at small 
142: radii, where most of the magnetic braking is expected to occur. 
143: Second, we explore a wider range in the degree of cloud magnetization, 
144: $4 \leq \lambda \leq 400$. We find that the rotationally supported 
145: disk begins to be disrupted by magnetic braking for a dimensionless  
146: mass-to-flux ratio $\lambda\sim 100$, corresponding to an extremely 
147: weak field strength of order 1~$\mu$G for typical parameters of 
148: low-mass cores. An implication is that $\lambda=1$ is not the only
149: criterion for judging the dynamical importance of a magnetic field,
150: as is often assumed. A relatively weak field of $1 \ll \lambda~ 
151: \lesssim~ 100$ can dominate the angular momentum evolution in the main
152: accretion phase, because the field is amplified both by collapse 
153: and by rotation, and because the disk is braked over many rotation 
154: periods. Indeed, we find that even in the weakest field 
155: case that we have considered ($\lambda=400$), the field is 
156: amplified enough for the magnetic pressure to dominate the thermal 
157: pressure in the disk and its surrounding region, at least in 2D.  
158: The ideal MHD calculations provide a limiting case for 
159: calibrating calculations that include nonideal effects, 
160: particularly ambipolar diffusion, that are currently underway
161: (Mellon \& Li, in preparation).
162: 
163: The rest of the paper is organized as follows. We describe the problem 
164: setup and the code used for computation in \S~\ref{setup}. The numerical 
165: results are presented and discussed in \S~\ref{standard} through
166: \S~\ref{rotation}. In \S~\ref{discussion}, we discuss the generic
167: outcomes of the interplay between magnetic field and rotation during 
168: the protostellar mass accretion phase, including an extended region 
169: supported by a combination of toroidal magnetic field and rotation --- 
170: a ``magnetogyrosphere.'' Plausible 
171: observational evidence for such a structure in the Class 0 source 
172: IRAM 04191 is discussed. The last 
173: section \S~\ref{conclusion} contains the main conclusions. 
174: 
175: \section{Model Formulation}
176: \label{setup}
177: 
178: \subsection{Initial Condition}
179: 
180: The early stage of low-mass star formation can be divided 
181: conceptually into two phases: a prestellar and a protostellar 
182: phase. The two are separated by a pivotal state \citep{1996ApJ...472..211L}, 
183: in which the central density formally goes to infinity. 
184: The pivotal state marks the end of core formation and 
185: the beginning of protostellar mass accretion. It is 
186: characterized by a large (formally infinite) density contrast 
187: between the center and the edge of the core. Idealizing the pivotal (t=0) state 
188: as a static singular isothermal sphere (SIS), \citet{1977ApJ...214..488S} was 
189: able to obtain the well-known self-similar inside-out collapse 
190: solution for the protostellar phase of star formation. The 
191: present investigation is an extension of this work to those 
192: pivotal states that are both magnetized and rotating. 
193: 
194: The self-similarity of the SIS is preserved in the presence of 
195: a dynamically important magnetic field, provided that the 
196: mass-to-flux ratio along each field line is constant. This 
197: condition, termed ``isopedic''  by \citet{1996ApJ...472..211L}, 
198: is roughly satisfied in ambipolar diffusion-initiated 
199: formation of dense cores out of strongly magnetized clouds 
200: \citep{1989ApJ...342..834L,1993ApJ...415..680F}. In more weakly 
201: magnetized clouds, the nearly scale-free prestellar contraction 
202: of the isothermal gas right before the point mass formation 
203: would also drag the magnetic field of the dense core into 
204: a nearly isopedic configuration. ALS03 
205: found a family of self-similar solutions for the isopedically 
206: magnetized, rotating, singular isothermal pivotal state, which 
207: turned out to be toroids. They have carried out an initial 
208: study of the collapse of the pivotal
209: cores described by such solutions. Our investigation is a 
210: refinement and extension of theirs.
211: 
212: The singular isothermal toroids (SITs) are described, in a spherical 
213: polar coordinate system  ($r$, $\theta$, $\varphi$), by 
214: \begin{equation}
215: \rho(r,\theta)={a^2\over 2\pi G r^2}R(\theta),
216: \label{density}
217: \end{equation}
218: \begin{equation}
219: \Phi(r,\theta)={4\pi a^2 r\over G^{1/2}}\phi(\theta), 
220: \label{magflux}
221: \end{equation}
222: \begin{equation}
223: V_\varphi(r,\theta)=a~ v(\theta),
224: \label{rotationspeed}
225: \end{equation}
226: where $a$ is the isothermal sound speed, which we take to be a 
227: constant, and the functions $R(\theta)$, $\phi(\theta)$ and 
228: $v(\theta)$ describe the angular distributions of the density
229: $\rho$, magnetic flux $\Phi$, and rotation speed $V_\varphi$. 
230: We choose a fiducial value for the sound speed $a=0.3$~km/s  
231: (somewhat higher than $0.19$~km/s, the sound speed of molecular 
232: gas at 10~K), to account for the (typically subsonic) nonthermal
233: motions in dense cores \citep{1995mcsf.conf...47M}. The choice of $v(\theta)$ is 
234: discussed in depth in ALS03. Following their example, we set the 
235: rotation speed to a constant, i.e., $v(\theta)=v_0$. The density 
236: and magnetic flux distributions $R(\theta)$ and $\phi(\theta)$ 
237: are determined from a set of coupled ordinary differential 
238: equations (eqs.~[14] through [17] of ALS03) and associated 
239: boundary conditions. 
240: 
241: The toroid solutions are characterized by two free parameters,
242: $v_0$ and $H_0$. The parameter $v_0$ specifies how fast the
243: toroid rotates, while $H_0$ measures the amount of extra mass 
244: over the SIS that is supported by magnetic fields and rotation. 
245: For a given $v_0$, $H_0$ controls the magnetic field strength. 
246: ALS03 considered a number of combinations of $v_0$ and $H_0$,
247: including $v_0=0, 0.125, 0.25$ and $0.5$ for $H_0=0.25$, and 
248: $H_0=0.125, 0.25$ and $0.5$ for $v_0=0.25$. They 
249: found no evidence for a rotationally supported disk in any 
250: of their simulations, which have dimensionless mass-to-flux 
251: ratios ranging from $\lambda=2.77$ to 10. The result indicates 
252: that magnetic 
253: braking is very efficient, even for a magnetic field that is 
254: too weak to provide the bulk of support against self-gravity 
255: as a whole.  
256: On the other hand, a rotationally supported structure (a disk 
257: or a ring) is expected in the limit of zero magnetic 
258: field. As the field strength increases from zero, there must 
259: be a transition from a regime where a rotationally supported
260: structure is formed to a regime where it is completely 
261: suppressed. Quantifying how this transition occurs is one 
262: of the main goals of the current investigation. 
263: 
264: We will first focus on a specific combination of parameters, $v_0=0.5$ 
265: and $H_0=0.4$. For a fiducial isothermal sound speed $a=0.3$~km/s, 
266: the choice $v_0=0.5$ corresponds to an angular speed   
267: 3~km~s$^{-1}$~pc$^{-1}$ on the scale of 
268: $0.05$~pc, the typical radius of a dense core \citep{1995mcsf.conf...47M}. The 
269: angular speed is within, although on the high side of, the range of velocity
270: gradient 
271: inferred observationally by \citet{1993ApJ...406..528G} for a collection 
272: of NH$_3$ cores (see their Fig.~1b). The relatively high rotation 
273: rate is chosen so that any rotationally support structure that 
274: may form is adequately resolved; slower rotations are expected to
275: be braked more easily. For $v_0=0.5$, the 
276: choice $H_0=0.4$ yields a dimensionless mass-to-flux ratio 
277: along each field line $\lambda=4$, consistent with the 
278: value inferred in the well-studied core L1544 \citep{2000ApJ...537L.139C}, 
279: after correcting for projection effects. 
280: It produces a distribution of (vertical) magnetic field on 
281: the equator 
282: \begin{equation}
283: B_{\rm eq}=24.5 \left({a\over 0.3~{\rm km/s}}\right)^2 
284: \left({0.05~{\rm pc} \over r}\right) (\mu G).
285: \label{bequator}
286: \end{equation} 
287: 
288: The rotating, magnetized singular isothermal toroid specified 
289: by $v_0=0.5$ and $H_0=0.4$ is shown in Fig.~\ref{initial}. Compared 
290: with the SIS of the same isothermal sound speed, the SIT has an 
291: enhanced 
292: density on the equator (by a factor of 2.25), due to extra support 
293: by the magnetic field and, to a lesser extent, rotation. The density 
294: is depleted in the polar region as matter settles along 
295: the field lines, producing the toroid appearance. The magnetic 
296: pressure dominates the 
297: thermal pressure in the evacuated polar region, within a half 
298: opening angle of $40^\circ$ of the axis. Outside the region (where 
299: most of matter resides), the thermal pressure dominates, with a 
300: plasma-$\beta=7.68$ on the equator. This magnetic field, although 
301: too weak to prevent the core from collapsing inside out, is strong 
302: enough to suppress the formation of a rotationally supported disk 
303: completely, as we demonstrate below (see also ALS03). Indeed, 
304: even a much weaker magnetic field can modify the process of disk 
305: formation significantly in the ideal MHD limit. 
306: 
307: To quantify the effects of field strength on magnetic braking and 
308: disk formation, we will reduce the field strength of the toroid 
309: shown in Fig.~\ref{initial} by various constant factors everywhere, keeping 
310: the distributions of density and rotation speed fixed. We consider 
311: a wide range in the degree of magnetization, corresponding to 
312: $\lambda=400$, 200, 133, 80, 40, 20, 13.3, 8, and 4 (see Table~1). For 
313: reference, $\lambda=100$ corresponds to a field strength of merely $1~\mu$G 
314: on the core scale of $0.05$~pc (for the fiducial value of sound 
315: speed), 
316: well below the median field strength ($\sim 6~\mu G$) in the 
317: cold neutral structures of HI gas \citep{2005ApJ...624..773H}. The 
318: field strength in dense cores of molecular clouds 
319: is unlikely to be as low as $1~\mu$G, but we include cases
320: with high values of $\lambda$ to illustrate the transition from 
321: disk formation to suppression. For more turbulent (perhaps
322: massive star formation) regions where the effective sound speed 
323: is much higher than 0.3~km/s, even the very large $\lambda$ 
324: cases may become relevant, as we emphasize in the discussion 
325: section. 
326: 
327: A potential drawback of reducing the field strength without changing
328: the distributions of density and rotation speed correspondingly 
329: at $t=0$ is that the initial configuration is out of force balance. 
330: The weakening of magnetic support will induce large-scale infall 
331: motions at times $t >0$. Infall motions on the core scale are 
332: observed, however, in a number of dense cores, including L1544 
333: \citep{1998ApJ...504..900T}, with speeds typically of order half the 
334: sound speed. Such a subsonic motion is present in our standard 
335: model to be discussed in depth in \S~3. The reduction of field 
336: strength by a uniform factor everywhere at $t=0$ has the added 
337: advantage that the core collapse at $t > 0$ remains self-similar, 
338: which provides a powerful check on the numerical solution. The
339: same advantage is preserved when we reduce the fiducial rotation 
340: speed at $t=0$ by a uniform factor everywhere. The effects of 
341: initial rotation speed are discussed separately from those of 
342: initial field strength in \S~\ref{rotation}. 
343: 
344: \begin{deluxetable}{llll}
345: \tablecolumns{4}
346: \tablecaption{Model Parameters\label{tab:model}}
347: \tablewidth{0pt}
348: \tablehead{
349:   \colhead{Model}     & \colhead{$\lambda$} 
350: & \colhead{$v_0$} & \colhead{$B_c^a~(\mu G)$} 
351: }
352: \startdata
353: B0   & $\infty$    & 0.5   & 0.0  \\
354: B1   & 400  & 0.5   & 0.25 \\
355: B2   & 200  & 0.5   & 0.49 \\
356: B3   & 133  & 0.5   & 0.74 \\
357: B4   & 80   & 0.5   & 1.23 \\
358: B5   & 40   & 0.5   & 2.45 \\
359: B6   & 20   & 0.5   & 4.90 \\
360: B7$^b$  & 13.3 & 0.5   & 7.35 \\
361: B8   & 8    & 0.5   & 12.3 \\
362: B9   & 4    & 0.5   & 24.5 \\
363: 
364: R0   & 13.3 & 0.0   & 7.35 \\
365: R1   & 13.3 & 0.125 & 7.35 \\
366: R2   & 13.3 & 0.25  & 7.35 \\
367: 
368: \enddata
369: \tablecomments{(a) Initial equatorial field strength (see 
370: equation~[\ref{bequator}]). (b) Model B7 is the standard model to be
371: discussed in depth in \S~\ref{standard}.}
372: \end{deluxetable}
373: 
374: \subsection{Boundary Conditions}
375: 
376: The protostellar phase of low-mass star formation in magnetized cores
377: is challenging to simulate in the ideal MHD limit. Once a 
378: central point mass is formed, it will carry along with it a 
379: finite amount of magnetic flux. The trapped flux would formally produce 
380: a split magnetic monopole \citep{1993ApJ...417..243G,1997ApJ...475..237L}, 
381: causing the field strength to increase rapidly with decreasing 
382: radius ($\propto r^{-2}$). The rapid 
383: increase of field strength makes the Courant condition associated
384: with the Alfv{\'e}n speed prohibitively small close to the origin, 
385: where the grid is necessarily very fine, especially in the $\theta$
386: direction of the adopted spherical polar coordinate system. 
387: 
388: To alleviate the numerical problem associated with the formation 
389: of a split magnetic monopole, we put the inner boundary of our
390: computation domain at a fixed radius, $r_i=10^{14}$~cm, or 
391: $6.7$~AU. We are unable to follow the evolution of the flow inside 
392: $r_i$. However, for our adopted initial conditions, only 
393: a small fraction of the core mass has a low enough specific 
394: angular momentum to fall through the inner surface (or `` inner
395: hole'') without magnetic braking; for the majority of the 
396: mass, extensive braking
397: must occur in the computation domain to bring it to the inner 
398: surface. On this surface, we impose the standard ``outflow'' 
399: boundary conditions for the hydrodynamic quantities, i.e., the 
400: density, energy and tangential (to the surface) velocity 
401: components are copied from the first active zones into the 
402: ghost zones along the radial direction. The radial velocities 
403: in the ghost zones are set to the lesser of the radial 
404: velocity in the first active zone and zero. These 
405: conditions are a standard feature already implemented in the 
406: Zeus2D code \citep{1992ApJS...80..753S,1992ApJS...80..791S} that is employed for our 
407: simulations.
408: 
409: The inner boundary conditions for the magnetic field require
410: special attention. Since the poloidal components of the 
411: magnetic field are evolved through the method of constrained
412: transport \citep{1988ApJ...332..659E} in Zeus2D, their boundary 
413: conditions are imposed on the 
414: electromotive force (EMF) ${\bf \epsilon}={\bf V}\times 
415: {\bf B}$. We apply the standard Zeus2D ``outflow'' boundary 
416: conditions on the EMF to evolve the poloidal magnetic field. 
417: They produce a poloidal field that varies smoothly across 
418: the inner surface. The toroidal component of the magnetic 
419: field $B_\varphi$ is not evolved 
420: through the  method of constrained transport, and it is 
421: possible to impose boundary conditions on $B_\varphi$ 
422: directly. We pick the simplest possible condition 
423: for the toroidal field in the ghost zones: $B_\varphi =0$. 
424: It is in effect a torque-free boundary condition. The choice 
425: is motivated in part by the realization that, by the time a 
426: piece of core material reaches the (small) inner surface, most 
427: of its angular momentum would have been stripped away already. 
428: The remaining angular momentum would be too little to significantly 
429: twist the strong (split-monopole) field lines that thread the 
430: inner surface \citep[see][for a related discussion]{2006ApJ...647..374G}. 
431: Other prescriptions are possible. For example, \citet{2001MNRAS.322..461S} 
432: enforced the negative stress condition $B_r B_\varphi
433: \leq 0$, whereas \citet{2001ApJ...548..348H} set the transverse 
434: components of the magnetic field to zero in their simulations
435: of black hole accretion in a cylindrical coordinate system. 
436: We have experimented with other prescriptions for $B_\varphi$ 
437: in the ghost zones, including copying values from the first 
438: active zones, which produced a somewhat stronger outflow 
439: along the axis compared with the torque-free case. The 
440: outflow is weak by the standard of the powerful jets and 
441: winds observed in young stars (ALS03), and will be easily 
442: masked by them. In any case, the simple torque-free boundary 
443: condition does not 
444: directly affect the braking of the bulk of the core material 
445: that accretes through the equatorial region (including any 
446: disk that may form), the main focus of our investigation. 
447: 
448: Boundary conditions are also required at the outer boundary. We 
449: set the outer radius of the computation domain to $r_o=2\times 
450: 10^{17}$~cm, slightly 
451: larger than the fiducial core radius of $0.05$~pc. Within this 
452: radius, there is $3.8$~$M_\odot$ of material for the initial 
453: density distribution shown in Fig.~\ref{initial} (for 
454: $a=0.3$~km/s), more than enough to form a typical low-mass 
455: star of 0.5~$M_\odot$. On the outer boundary, we impose the 
456: standard Zeus2D ``outflow'' boundary conditions for all 
457: quantities, including $B_\varphi$. 
458: 
459: We assume that the core collapse remains symmetric with respect 
460: to the polar axis and the equatorial plane at all times. We adopt 
461: the standard Zeus2D boundary conditions for the axis. To enforce 
462: the equatorial boundary conditions, we mirror all quantities
463: (including the three components of the magnetic field) in 
464: the upper quadrant to the lower quadrant at the beginning
465: of each time step, and evolve the governing equations in 
466: both quadrants. This implementation of equatorial boundary
467: conditions, although somewhat more expensive than the one
468: already in Zeus2D, guarantees that the symmetry is enforced
469: exactly. Spontaneous symmetry breaking is observed in MHD 
470: simulations of black hole accretion \citep{2001ApJ...548..348H,2001MNRAS.322..461S}. 
471: We will relax the equatorial symmetry 
472: in future 3D simulations. 
473: 
474: \subsection{Code Modification and Problem Setup} 
475: 
476: For our axisymmetric problem, we use the Zeus2D MHD code of 
477: \citet{1992ApJS...80..753S,1992ApJS...80..791S} 
478: in spherical polar coordinate system. The code is 
479: well suited for our investigation of magnetic braking, since the
480: propagation of (torsional) Alfv\'en waves is treated accurately 
481: with the method of characteristics. The main modification that we 
482: made to the code is to improve the method for obtaining the values
483: of gravitational potential at the inner and outer boundaries; they 
484: are needed for solving the Poisson equation for the self-gravity 
485: of the material inside the computational domain. The method of 
486: multipole expansion in Zeus2D performs poorly when mass is near a 
487: boundary and requires a large number of terms to accurately determine 
488: the gravitational potential.  Instead, we transform Green's function 
489: from the multipole solution to an elliptic integral solution 
490: \citep{1999ApJ...527...86C}. 
491: We can then directly sum up 
492: the contributions of the masses in all active cells to the 
493: gravitational potential at a given location on the boundary, as 
494: in \citet{2004ApJ...616..364F}. We also modified the equation of state 
495: to change smoothly from isothermal to adiabatic 
496: around a critical density of $10^{-13}$~gm~cm$^{-3}$, to mimic 
497: the effects of radiation trapping at high densities. 
498: 
499: To ensure good resolution on both large and small scales, a 
500: logarithmically spaced grid is used in the radial direction. 
501: We divide the computational domain between the inner 
502: ($r_i=10^{14}$~cm) and outer radius ($r_o=2\times 10^{17}$~cm) 
503: into 120 shells, each having a width that is 6\% larger than 
504: that of the shell interior to it. Since 
505: both $\Delta r$ and $r\Delta \theta$ 
506: vary linearly with the radius, the grid does not contain cells 
507: of large aspect ratios often seen in logarithmic grids in 
508: cylindrical coordinate system. The width of the first shell 
509: is $1.10\times 10^{13}$~cm, much smaller than the inner radius
510: $r_i$. In the polar direction, we divide the region between 
511: the upper ($\theta=0$) and lower axis ($\theta=\pi$) into 119 
512: uniform wedges, each with an opening angle of $1.5^\circ$. 
513: As mentioned earlier, half of the grid (between $\theta=\pi/2$ 
514: and $\pi$) is used for enforcing mirror symmetry across the 
515: equator. 
516: 
517: We begin the collapse calculation with a static, dense core. 
518: To initiate the collapse, we add a point 
519: mass of $M_0=3 a^3 r_i/G$ (where $a$ is the isothermal
520: sound speed and $r_i$ the radius of the inner hole) at $t=0$. 
521: This mass is 1.5 times the mass enclosed within $r_i$ for 
522: a SIS of sound speed $a$, and slightly larger than that 
523: enclosed within the same radius for the magnetized SIT shown 
524: in Fig.~\ref{initial} ($2.8 a^3 r_i/G$). As the collapse 
525: proceeds, the density is depleted preferentially in the 
526: polar region, where the magnetic field is strong and matter 
527: slides more or less freely along the field lines to the center 
528: (ALS03). To prevent the Alfv\'en time step from dropping to
529: prohibitively small values, we adopt a density floor of $10^
530: {-19}$~gm~cm$^{-3}$ on the inner half of the (logarithmic) 
531: radial grid (within $5.3\times 10^{15}$~cm). The occasional 
532: use of density floor in the polar region is expected to 
533: have little dynamical effect, since the dynamics there are 
534: controlled by the magnetic field. We have experimented with 
535: different choices of the floor value and confirmed the
536: expectation. 
537: 
538: \section{Standard Model} 
539: \label{standard}
540: 
541: The rotating, magnetized core that we adopt as the initial configuration 
542: for the collapse calculation is specified by two parameters: the 
543: dimensionless mass-to-flux ratio $\lambda$ and the rotation speed 
544: $v_0$ in units of the sound speed. We will explore a wide range of 
545: values for $\lambda$ in \S~\ref{fieldstrength} and several values 
546: for $v_0$ in \S~\ref{rotation}. In this section, we focus on a model 
547: with a particular combination of parameters $\lambda=13.3$ and
548: $v_0=0.5$
549: (Model B7 in Table~1); 
550: it serves as the standard against which others models are compared. 
551: The choice $\lambda=13.3$ corresponds to a moderate field strength of 
552: $7.35$~$\mu$G on the scale of the typical core radius $0.05$~pc 
553: (not much higher than the median field strength of $\sim 6.0$~$\mu$G 
554: inferred by \citet{2005ApJ...624..773H} for the cold neutral 
555: HI structures), and $v_0=0.5$ is chosen to ensure that any rotationally 
556: supported structure that may form is large enough to be adequately 
557: resolved. More importantly, it turns out that the combination 
558: yields a collapse solution that is roughly self-similar in time, 
559: as expected from the self-similar initial configuration adopted. 
560: The self-similarity gives us confidence on the numerically obtained 
561: solution. 
562: 
563: \subsection{Collapse Solution at a Representative Time}
564: 
565: Fig.~\ref{snap} shows a snapshot of the standard model at a 
566: representative time $t=6.4\times 10^{11}$~sec (or $\sim 2\times 
567: 10^4$~yrs after the initiation of collapse), comparable to the
568: typical ages estimated for Class 0 sources. The snapshots at 
569: other times are largely similar (to be quantified below), which 
570: motivates us to introduce a set of dimensionless variables:
571: \begin{equation}
572: \alpha =4\pi G t^2 \rho(r,\theta,t),\ 
573: {\bf b}={G^{1/2} t\over a} {\bf B}(r,\theta,t),\ 
574: {\bf v}={{\bf V}(r,\theta,t)\over a},\ x={r\over a t}.
575: \label{dimensionless}
576: \end{equation}
577: If the collapse is strictly self-similar, the dimensionless density 
578: $\alpha$, magnetic field ${\bf b}$ and velocity ${\bf v}$ would be 
579: functions of the dimensionless self-similar radius $x=r/(at)$ and 
580: polar angle $\theta$ only, and the solution would look identical at 
581: different times except for a scaling factor. Equation (5) can be 
582: used to obtain the dimensional units for any dimensionless quantity
583: to be used below, particularly in figures; the plotted quantities 
584: will be dimensionless unless noted otherwise explicitly. 
585: 
586: Broadly speaking, there are four dynamically distinct regions. At 
587: large (cylindrical) distances, the core material collapses inward, 
588: dragging along magnetic field lines, particularly in the equatorial 
589: region. We term this region the ``collapsing envelope'' or ``Region 
590: I.'' Most of the collapsing material from the envelope is channeled 
591: into the equatorial region, forming a flattened ``pseudodisk'' or 
592: ``Region II.'' The pseudodisk is squeezed from above by an expanding 
593: ``magnetic bubble'' (or ``Region III'') that lies at intermediate 
594: latitudes. The pseudodisk supports the bubble from below. On the 
595: other side of the bubble, closer to the polar axis, lies ``Region 
596: IV,'' where matter slides down well-ordered field lines, forming 
597: a low-density ``polar funnel.'' We consider the outflowing region 
598: near the axis part of ``Region III;'' it is a continuation of the 
599: bubble to large distances. 
600: 
601: The boundaries between the various regions show up most 
602: clearly in the map of the magnetic twist, defined as the ratio of 
603: the toroidal to poloidal field strength $B_\varphi/\vert B_p \vert$ 
604: (see Fig.~\ref{twist}). To be specific, we define the region 
605: dominated by the (negative) toroidal magnetic field (i.e., 
606: $-B_\varphi > \vert B_p \vert$) as the bubble (``Region 
607: III''). It is surrounded by the polar funnel to the left, the 
608: collapsing envelope to the right, and the pseudodisk from 
609: below. The collapsing envelope is separated from the bubble 
610: by a ``magnetic wall,'' produced by the deceleration of the 
611: magnetized collapsing envelope against the bubble; the  
612: deceleration leads to a pileup of the field lines. The 
613: decelerated material slides along the compressed field lines 
614: nearly vertically towards the equator. Before hitting 
615: the equator, the downward moving material is deflected 
616: towards the origin by a strong (thermal) pressure gradient 
617: to join the pseudodisk beneath the magnetic bubble. 
618: 
619: The equatorial part of the collapsing envelope joins onto the 
620: pseudodisk more directly. It slows down near the outer edge of
621: the pseudodisk before being reaccelerated towards the center 
622: by gravity. The deceleration and reacceleration show up more 
623: clearly in Fig.~\ref{vrvphi_eq}, where we plot the radial 
624: component of the velocity on the equator as a function of 
625: radius. The curve of radial speed has a peak around $x_{\rm s} 
626: \approx 0.5$, where $v_r$ is close to zero; indeed, the gas 
627: there is actually slowly expanding rather than collapsing. 
628: We identify the region near the stagnation radius $x_{\rm s}$ 
629: as a ``magnetic barrier'', as opposed to a ``centrifugal 
630: barrier.'' The reason is that, at this location, the rotation
631: speed is well below that needed for the centrifugal force to 
632: balance the gravity (see Fig.~\ref{vrvphi_eq}, where both speeds
633: are plotted).
634: There is tantalizing evidence for such a structure 
635: in the Class 0 source IRAM 04191, as we discuss in 
636: \S~\ref{magnetogyrosphere}. It bounds the pseudodisk from 
637: outside. Note that the infall speeds at large distances are 
638: close to half the sound speed, comparable to that inferred 
639: observationally in the dense core L1544 (as mentioned earlier)
640: and in the envelope of IRAM 04191 \citep{2002A&A...393..927B}, 
641: except near the outer boundary, where the edge effect becomes 
642: significant. The large-scale infall is induced by the reduction 
643: of the full field strength at $t=0$ (from $\lambda=4$ to 
644: 13.3), which weakens the initial magnetic support relative 
645: to self-gravity. 
646: 
647: After passing through the stagnation region at the magnetic 
648: barrier, the pseudodisk material on the equator collapses quickly 
649: inward. The rapid infall leads to a spinup of the collapsing 
650: material, as shown in Fig.~\ref{vrvphi_eq}. 
651: The inward increase in rotation speed 
652: is slowed down around a dimensionless radius $x_{\rm c1}\sim 
653: 0.1$, and levels off near $x_{\rm c2}\sim 0.05$. A glance at
654: Fig.~\ref{twist} shows that this is the region where the magnetic 
655: bubble is initiated. In this region, the radial 
656: infall is temporarily slowed down, as a result of increased 
657: centrifugal support; in other words, a centrifugal barrier 
658: is encountered. The slowdown of infall allows more time for 
659: the magnetic braking to operate. Interior to  $x_{\rm c2}$ is 
660: a strongly magnetized region dominated by the central split 
661: monopole. Here, the rotation speed drops steadily with decreasing 
662: radius, as a result of efficient braking by the strong magnetic 
663: field. The loss of centrifugal support causes the equatorial 
664: material to resume its rapid collapse towards the center. The 
665: rapid infall and slow rotation leave no doubt that a rotationally
666: supported disk is not produced in this standard model. In particular, 
667: the rotation speed at small radii is well below that needed to
668: provide centrifugal support against gravity (compare the top
669: two curves in Fig.~\ref{vrvphi_eq}). Upon arriving at the inner 
670: hole, the infall material has lost most of its original angular 
671: momentum along the way. The questions is: how does the stripping 
672: of angular momentum occur mechanically? 
673: 
674: The key to answering the above question lies in the magnetic 
675: bubble (Region III). It is powered by the pseudodisk (Region 
676: II). The material in the pseudodisk spins up as it collapses, 
677: producing a growing toroidal magnetic field, which in turn 
678: leads to a buildup of magnetic pressure. The 
679: over-pressure in the region is relieved by expansion 
680: along the path of least resistance. Expansion is restricted to 
681: the left of the region or vertically up by the strong monopole 
682: field. It cannot freely expand horizontally outward either; 
683: it is contained by the ram pressure of the infalling material. 
684: These constraints force the over-pressured region to expand 
685: outward at an acute angle with respect to the equatorial 
686: plane, producing a bubble where the bulk of the angular 
687: momentum stripped from the collapsing pseudodisk is stored. 
688: 
689: \subsection{Approximate Self-Similarity in Time Evolution}
690: \label{ss}
691: 
692: Self-similarity not only provides a check on the numerically obtained 
693: solution, but also enables us to discuss the solution in a
694: time-independent way. A good indicator of the self-similarity of 
695: the collapse solution is the evolution of the central point mass
696: $M_0$. In a strictly self-similar solution 
697: (such as Shu's inside-out collapse solution), 
698: the mass should increase linearly with time. The 
699: change of $M_0$ with time for the standard model is shown in 
700: Fig.~\ref{pointmass}. Overall, there is a nearly linear increase 
701: of the point mass with time, except near the origin. The initial 
702: deviation is caused by the finite size of the inner hole (which 
703: breaks the self-similarity of the initial configuration), and 
704: the point mass that was put at the center at $t=0$ to induce 
705: collapse. After a short period of initial adjustment (of 
706: order $10^{11}$~sec), the mass-time curve becomes 
707: remarkably straight, although there are low-amplitude oscillations. 
708: As we discuss in detail in \S~\ref{strongfield}, the oscillation
709: is caused by magnetic flux accumulation near the center, which 
710: produces a split magnetic monopole that reconnects episodically 
711: across the equatorial plane. The reconnection events make it 
712: impossible for the collapse to settle into a strictly steady 
713: state in the self-similar space. Their influence on the collapse
714: solution increases with the strength of the magnetic field. 
715: For the standard model, the field is still weak enough that the 
716: self-similarity is preserved approximately.
717: 
718: To demonstrate the approximate self-similarity more vigorously,  
719: we use a closure relation that any self-similar collapse 
720: solution must satisfy. The relation is derived from the equation 
721: of mass conservation
722: \begin{equation}
723: {\partial M(r,t)\over \partial t}=-2\pi \int^\pi_0 \rho r^2 
724: \sin(\theta)V_r d\theta,
725: \label{masscons}
726: \end{equation}
727: where $M(r,t)$ is the mass enclosed within a radius of $r$ at the instant 
728: of time $t$. It is given by
729: \begin{equation}
730: M(r,t)=M_0(t)+2\pi \int_0^r \int^\pi_0 \rho {\tilde r}^2 
731: \sin(\theta) d\theta d {\tilde r},
732: \label{mass}
733: \end{equation}
734: where $M_0$ is the point mass at the center. In terms of the dimensionless 
735: self-similar variables introduced in equation~(\ref{dimensionless}), the 
736: mass $M(r,t)$ can be written into
737: \begin{equation}
738: M(r,t)={a^3 t\over 2 G} m(x), 
739: \label{mass_scaling}
740: \end{equation}
741: where
742: \begin{equation}
743: m(x)=m_0+\int_0^x \int_0^\pi \alpha({\tilde x},\theta) {\tilde x}^2 
744: \sin(\theta) d\theta d{\tilde x},
745: \label{mass_dimensionless}
746: \end{equation}
747: is the dimensionless mass enclosed within a dimensionless radius $x$,
748: and $m_0$ is the dimensionless point mass at the center. Using 
749: equation~(\ref{mass_scaling}), we can express the left-hand side of 
750: equation~(\ref{masscons}) as 
751: \begin{equation}
752: {\partial M(r,t)\over \partial t} = {a^3\over 2G} \left[m(x)-x 
753: m^\prime(x) \right],
754: \label{chainrule}
755: \end{equation}
756: where $m^\prime=dm/dx= \int_0^\pi \alpha(x,\theta) x^2 \sin(\theta)
757: d\theta$ from equation~(\ref{mass_dimensionless}). Equating the 
758: right-hand side of equation~(\ref{chainrule}) with the right-hand 
759: side of equation~(\ref{masscons}) yields
760: \begin{equation}
761: m(x) = \int_0^\pi \alpha(x,\theta) x^3 \sin(\theta) d\theta+
762: \int_0^\pi \alpha(x,\theta) x^2 \sin(\theta) (-v_r) d\theta
763: =\int_0^\pi \alpha(x,\theta) x^2 \sin(\theta) (x-v_r) d\theta,
764: \label{masscons_dimensionless}
765: \end{equation}
766: which is the equation of mass conservation in the self-similar space.  
767: It has a simple physical meaning. The term on the left-hand side is 
768: the dimensionless mass enclosed within a sphere of radius $r=x a t$ 
769: that expands linearly in time (for a fixed $x$). This mass can be 
770: increased due to either 
771: the expanding volume (which is proportional to the first term on the 
772: right-hand side of the first equality) or advection of material 
773: into the sphere from outside 
774: (proportional to the second term).  Note that the combination $x-v_r$
775: on the right hand side of the second equality is simply the speed with 
776: which the sphere is expanding relative to 
777: the material that moves at a radial speed of $v_r$. 
778: 
779: The dimensionless mass distribution $m(x)$ can be evaluated in another
780: way, using equations~(\ref{mass_dimensionless}), which states that the 
781: mass inside a given radius is simply the sum of all mass (including 
782: the point mass) up to that 
783: radius. In a strictly self-similar solution, the masses determined 
784: independently from equations~(\ref{mass_dimensionless}) and 
785: (\ref{masscons_dimensionless}) should coincide. In our standard 
786: model, we find that the mass distribution $m(x)$ from 
787: equation~(\ref{mass_dimensionless}) remains more or less steady in
788: time everywhere, whereas that from equation~(\ref{masscons_dimensionless})
789: shows large variations from one time to another inside a dimensionless 
790: radius of $\sim 0.3$. The variations are caused by unsteady mass 
791: accretion due to frequent reconnection events. The variations are 
792: much reduced if we average the mass distributions over timescales 
793: that are long compared with the durations of individual reconnection 
794: events but short compared with the total simulation time, which is
795: $\sim 1.6\times 10^{12}$~sec. 
796: The result is illustrated in Fig.~\ref{M_av}, for a representative 
797: period of time between $t=3.0\times 10^{11}$ and $7.5\times 10^{11}$~sec; 
798: the averaging is done over 90 outputs taken at intervals of $5\times 
799: 10^9$~sec. Except for some wiggles within a small radius of $\sim
800: 0.1$, the two 
801: independently determined average mass distributions agree rather 
802: well. The maximum discrepancy (of order $\sim 4\%$) occurs around 
803: $x\sim 0.5$, close to the outer edge of the pseudodisk. Perfect 
804: agreement is not to be expected, given the finite sizes of the inner 
805: and outer boundaries of the computation domain. The good agreement 
806: suggests that the collapse solution fluctuates around a well defined 
807: self-similar solution, with a mass distribution close to those 
808: shown in the figure. 
809: 
810: A similar closure relation can be derived for the angular momentum, 
811: allowing further verification of the approximate self similarity 
812: of this solution. We start from the equation for angular momentum 
813: conservation:
814: \begin{equation}
815: {\partial L(r,t)\over \partial t}=-\int \rho \varpi V_\varphi V_r dS
816: +{1\over 4\pi} \int B_\varphi B_r \varpi dS,
817: \label{angmom_cons}
818: \end{equation}
819: where $L(r,t)$ is the angular momentum enclosed within a sphere
820: of radius $r$ at a time $t$, and the two terms on the right
821: hand side are, respectively, the angular momentum advected into
822: the sphere by fluid motion per unit time and the magnetic torque
823: acting on the surface of the sphere, $S$. The quantity $\varpi$ 
824: is the cylindrical radius. If the collapse is 
825: strictly self-similar, the angular momentum can be written as
826: \begin{equation}
827: L(r,t)={a^5 t^2\over 2 G} l(x),
828: \label{angmom_scaling}
829: \end{equation}
830: where the dimensionless angular momentum $l(x)$ is given by 
831: \begin{equation}
832: l(x)=\int_0^x\int_0^\pi \alpha v_\varphi {\tilde x}^3 
833: \sin^2(\theta) d\theta d{\tilde x}. 
834: \label{angmom_dimensionless}
835: \end{equation}
836: Using equation~(\ref{angmom_scaling}), we can express the left-hand 
837: side of equation~(\ref{angmom_cons}) as 
838: \begin{equation}
839: {\partial L(r,t)\over \partial t} = {a^5 t\over 2G} \left[ 2 l(x)-x 
840: l^\prime(x) \right],
841: \label{chainrule_angmom}
842: \end{equation}
843: where $l^\prime=dl/dx$. Equating the right-hand side of 
844: equation~(\ref{chainrule_angmom}) with the right-hand side 
845: of equation~(\ref{angmom_cons}) yields
846: \begin{equation}
847: l(x) = {1\over 2} \left[ \int_0^\pi \alpha(x,\theta) 
848: v_\varphi 
849: x^4 \sin^2(\theta) d\theta +\int_0^\pi \alpha(x,\theta) v_\varphi 
850: x^3 \sin^2(\theta) (-v_r) d\theta + \int_0^\pi b_r b_\varphi x^3 
851: \sin^2(\theta) d\theta \right] 
852: \label{angmom_cons_dimensionless}
853: \end{equation}
854: which is the equation of angular momentum conservation in the 
855: self-similar space. Compared with the dimensionless equation 
856: for mass conservation (eq.~[\ref{masscons_dimensionless}]), 
857: there is an extra factor of $1/2$ on the right hand side 
858: (resulting from the quadratic rather than linear dependence 
859: of angular momentum on time) and an additional (third) term 
860: inside the square bracket that comes from magnetic braking. 
861: The first two terms have the usual interpretation: they 
862: are, respectively, the increase of angular momentum inside 
863: a sphere of constant dimensionless radius $x$ due to the 
864: expansion of the sphere and matter moving across the surface. 
865: 
866: We plot in Fig.~\ref{L_av} the averaged distributions of angular 
867: momentum computed from equations~(\ref{angmom_dimensionless}) and 
868: (\ref{angmom_cons_dimensionless}), over the same 90 outputs used
869: for Fig.~\ref{M_av}. The two distributions agree remarkably well, 
870: again indicating that the solution is nearly self-similar.
871: 
872: \subsection{Mass and Angular Momentum Distribution in Self Similar Space}
873: 
874: It is instructive to examine the dimensionless mass distribution 
875: $m(x)$ more closely. In Fig.~\ref{M_av}, we plotted the two terms 
876: on the right hand side of equation~(\ref{masscons_dimensionless}) 
877: separately. The contribution to $m(x)$ due to the advection term 
878: (proportional to 
879: the infall speed ${-v_r}$) has a minimum value of $\sim 0.5$ 
880: near the radius $x\sim 0.5$, where the infall is slowed down at the 
881: magnetic barrier. It increases inwards as the temporarily slowed 
882: down material picks up infall speed (see Fig.~\ref{vrvphi_eq}). The 
883: dip is compensated, to a large extent, by a peak in the contribution 
884: due to volume expansion (for a sphere of fixed dimensionless radius 
885: $x$). The peak is a result of the density enhancement in the
886: low-$v_r$, stagnation region.
887: 
888: Note that the dimensionless mass $m(x)$ approaches a finite value as 
889: $x\to 0$. The mass near the origin is simply the central point 
890: mass. It has a dimensionless value of 
891: $m_0 = 2G M_0/(a^3 t) \sim 3.6$. This mass is nearly twice the 
892: classical value for 
893: the inside-out collapse of a singular isothermal sphere (SIS), which 
894: has $m_0=1.95$ in our mass units \citep{1977ApJ...214..488S}. 
895: Part of the increase comes from the fact that the initial configuration 
896: for collapse is, on average, denser than the SIS. Another reason is 
897: that, unlike the SIS, the initial configuration is out of exact force 
898: balance, which induces a subsonic, global 
899: contraction. The mass accretion onto the point mass is hindered, on 
900: the other hand, by rotation, which tends to slow down the collapsing 
901: material through centrifugal force. Indeed, in the absence of a 
902: magnetic field, most of the collapsed material in the standard 
903: model would be rotationally supported rather than falling to the 
904: center. The relatively high rate of {\it central} mass accretion is made 
905: possible by continuous magnetic braking. 
906: 
907: To understand the effects of magnetic braking on the angular momentum 
908: evolution in more detail, we plot in Fig.~\ref{L_av} the individual 
909: terms on the right hand side of 
910: eq.~(\ref{angmom_cons_dimensionless}). 
911: As with the mass distribution $m(x)$, the 
912: contribution to the angular momentum $l(x)$ inside a sphere of 
913: radius $x$ due to advection (second term) shows a dip near 
914: $x\sim 0.5$, as a result of infall deceleration, whereas 
915: that due to volume expansion (first term) shows a peak in the 
916: same region, as a result of angular momentum pileup. A crucial 
917: difference is that, unlike $m(x)$, $l(x)$ does not asymptote to 
918: a finite value near the origin. Rather, it approaches zero at 
919: small radii. The implication is that, {\it although mass falls 
920: onto the central object at a relatively high rate, little 
921: angular momentum is accreted}. The angular momentum removal 
922: can be seen most clearly by comparing the angular momentum 
923: advected into a sphere by mass motion (second term) and that 
924: removed by magnetic braking from the same surface (third term). 
925: >From Fig.~\ref{L_av}, we find that these two terms nearly cancel 
926: each other over an extended region (up to $x\sim 0.4$). The 
927: near cancellation points to a detailed balance of inward and 
928: outward transport of angular momentum, which enables a large 
929: amount of matter to be accreted, but little angular momentum. 
930: The balance is at the heart of the magnetic braking-driven mass 
931: accretion. 
932: 
933: \subsection{Magnetic Bubble} 
934: 
935: The magnetic bubble plays a key role in the angular momentum 
936: redistribution. Angular momentum is transported outward in 
937: this region through both outflowing gas and magnetic torque. In 
938: the standard model, roughly half of the angular momentum advected 
939: across a given radius by the equatorial, infalling, pseudodisk 
940: is carried away by the outflowing material; the other half 
941: by the highly twisted magnetic field. It is the toroidal 
942: field that drives the expansion of the bubble, as in the so-called 
943: ``magnetic tower'' investigated by \citet{2003MNRAS.341.1360L}.
944: The strength of the toroidal field is determined by the ambient pressure 
945: that confines the bubble in the lateral direction. In Fig.~\ref{pressures}, 
946: we plot the total pressure and
947: its various components as a function of angle at a representative
948: radius $x=0.3$ (roughly the middle radius of the bubble). The 
949: total pressure does not vary much at different angles, but it 
950: is dominated by different components in different regions. In the 
951: pseudodisk near the equator ($\theta=\pi/2$), it is dominated by 
952: the thermal pressure, because of high density. The pressure due 
953: to the toroidal magnetic field dominates at intermediate 
954: latitudes inside the bubble. It is more than an order of magnitude 
955: stronger than the pressure due to the poloidal field, and a factor 
956: of $\sim 2.5$ times larger than the thermal pressure. The bubble 
957: is therefore magnetically dominated, with a plasma-$\beta$ less 
958: than unity. The polar region is even more strongly dominated by 
959: the magnetic field, although by the poloidal rather than the 
960: toroidal component. 
961: 
962: The kinematics of the three regions are also quite distinct. Over 
963: most of the equatorial pseudodisk, the infall and rotation speeds 
964: are comparable. Both are supersonic and super Alfv\'enic. The 
965: infall dominates the rotation in the polar funnel. It is supersonic 
966: but sub-Alfv\'enic (because of a high Alfv\'en speed associated 
967: with the strong poloidal field and low density). The bubble, on the 
968: other hand, is a toroidally dominated structure in both magnetic 
969: field (by definition) and velocity field. The mass-weighted average
970: rotation speed inside the bulk of the bubble (within a dimensionless
971: radius $x=1$) is about $2.2~a$, whereas that for the average poloidal 
972: speed is slightly less than the sound speed $a$, which is smaller 
973: than the Alfv\'en speed. The meridian flow inside the bubble is 
974: thus generally sub-Alfv\'enic. It is in contact, through magnetosonic
975: waves, with the base of the bubble, where a toroidal magnetic 
976: flux is constantly generated out of the poloidal flux (dragged in
977: from large distances) by 
978: differential rotation. It is the continuous insertion of new 
979: toroidal flux into the bubble that drives it to expand \citep{2003MNRAS.341.1360L}. 
980: 
981: The sub-Alfv\`enic motions indicate that the bubble can be viewed
982: as being more or less quasi-static in the meridian plane. It is 
983: then natural to ask: what supports the bubble against the gravity 
984: (which comes mostly from the point mass at the radii of the bubble, 
985: with some contribution from the self-gravity)? It turns out that 
986: the support is provided mainly by the rotation and toroidal magnetic 
987: field; the latter acts on the bubble material through a combination 
988: of magnetic pressure gradient and tension force. In some sense, 
989: the slowly expanding bubble is a partly rotationally and partly 
990: magnetically supported structure that forms in place of the  
991: purely rotationally supported structure that would form in the 
992: absence of magnetic braking. It is a generic, potentially 
993: observable structure that we will discuss in some depth in 
994: \S~\ref{magnetogyrosphere}. 
995: 
996: The bubble is a depository of mass and angular momentum. 
997: The dimensionless mass and angular momentum are more or less 
998: constant in time, with small 
999: fluctuations due to reconnection events. The bubble has an 
1000: average mass $m_{\rm b} \sim 2.3$ within a radius of $x=1$, which 
1001: is about 64\% of the central point mass ($m_0\sim 3.6$). It 
1002: carries a total dimensionless angular momentum of $\sim 1.4$. 
1003: The average specific angular momentum is therefore $l_{\rm b}\sim 
1004: 0.61$. It corresponds to a centrifugal radius of 
1005: \begin{equation} 
1006: x_c\sim {2 l_{\rm b}^2 \over m_0} \sim 0.2. 
1007: \label{centrifugalrad}
1008: \end{equation} 
1009: If a fraction of the rotating bubble material were to recollapse 
1010: at a time $t$ in the absence of further magnetic braking, it would 
1011: form a disk of size comparable to the centrifugal radius  
1012: \begin{equation}
1013: r_c \sim 10^3 \left({a\over 0.3 {\rm km/s}}\right) 
1014: \left({t \over 10^5 {\rm yr}}\right)~{\rm AU}
1015: \label{cenrad}
1016: \end{equation}
1017: around the central object. We will discuss the possibility of
1018: late-time disk formation in \S~\ref{magnetogyrosphere}. 
1019:   
1020: Another way to gauge the dynamic state of the bubble is to consider 
1021: its energies. The total (thermal, kinetic and magnetic) energy is 
1022: close to the gravitational binding energy in the potential of the 
1023: central point mass. It is well below the gravitational binding 
1024: energy when the self-gravity of the core is also taken into account. 
1025: The bulk of the bubble material cannot escape to infinity by itself. 
1026: It is expanding slowly outward only because it is being pushed 
1027: continuously from below, as more and more mass and toroidal magnetic 
1028: flux being inserted into the bubble. If for some reason the 
1029: insertion were to stop suddenly, the bubble (at least a 
1030: portion of it) may recollapse to form a disk. The existence 
1031: of a large amount of bound (bubble) material of high specific 
1032: angular momentum not far from the center indicates that, {\it although
1033: magnetic braking can allow mass to fall into the center at
1034: a high rate, the overall angular momentum problem is not 
1035: decisively resolved.}
1036:  We will speculate in the discussion 
1037: section implications of this situation, including the possible 
1038: role of protostellar winds in dispersing away the bubble and 
1039: the angular momentum accumulated in it.
1040: 
1041: \section{Effects of Magnetic Field Strength}
1042: \label{fieldstrength}
1043: 
1044: Having examined in some detail a standard model with a specific 
1045: mass-to-flux ratio of $\lambda=13.3$, we now turn our attention 
1046: to cases with different degrees of magnetization. We will first 
1047: concentrate on the relative weak field cases with $\lambda$ 
1048: between 400 and 13.3 (\S~\ref{weakfield}). The strong field cases
1049: with $\lambda=4$ and $8$ behave quite differently; they will 
1050: be discussed separately in \S~\ref{strongfield}.  
1051: 
1052: \subsection{Relatively Weak Field Cases} 
1053: \label{weakfield}
1054: 
1055: Before going into details of the various collapse solutions, we first
1056: discuss their general behaviors based on the dimensionless central 
1057: mass $m_0$. It is defined as $m_0=2 G M_0/(a^3 t)$, where $M_0$ is
1058: the central mass. The dimensionless mass $m_0(t)$ at any given time is 
1059: simply the averaged mass accretion rate up to that time, normalized by 
1060: $a^3/(2G)$. It is plotted as a function of time in Fig.~\ref{Mstar_LowB} 
1061: for all cases with $\lambda$ between 13.3 and 400. Since mass accretion 
1062: is driven by magnetic braking, which in turn depends on the field 
1063: strength, one expects $m_0$ to decrease as $\lambda$ increases. This 
1064: trend is indeed seen in Fig.~\ref{Mstar_LowB}, except for the three
1065: most weakly magnetized cases ($\lambda=400$, 200 and 133). For these 
1066: extremely weak field cases, the trend is not as clear, especially at 
1067: late times. It turns out that these are the cases where a 
1068: rotationally-supported, equilibrium disk is present.  
1069: They are discussed in \S~\ref{weakextreme}. Even the weakest field 
1070: case has a significant mass accretion, however, at a rate well above 
1071: that in the non-magnetic case ($\lambda=\infty$; the lowest curve in 
1072: the figure) at late times. The initial burst of mass accretion in 
1073: the non-magnetic case comes from the finite size of the inner hole 
1074: and the presence of a small ``seed'' mass that was placed at the
1075: center to induce collapse, rather than any angular momentum 
1076: redistribution. 
1077: 
1078: The initial burst in mass accretion is also present in the magnetized 
1079: cases, although it is relatively inconspicuous in the two strongest
1080: field cases ($\lambda=13.3$ and $20$). In both cases, the (average) 
1081: mass accretion rate quickly approaches a plateau, indicating that 
1082: a nearly self-similar state is reached; the small oscillations are 
1083: related to magnetic flux accumulation in the central hole (see  
1084: \S~\ref{strongfield} below). Indeed, the collapse 
1085: solution in the $\lambda=20$ case is very similar to that of the 
1086: standard $\lambda=13.3$ case discussed in the last section. It will 
1087: not be discussed separately further. 
1088: 
1089: The intermediate cases with $\lambda=40$ and 80 are particularly 
1090: interesting. They represent the transition from the weaker field 
1091: cases where a rotationally-supported, equilibrium disk is 
1092: present 
1093: to those where the disk formation is suppressed. These cases are 
1094: discussed in \S~\ref{transition}. 
1095: 
1096: \subsubsection{Disk Formation in Extremely Weak Field Cases}
1097: \label{weakextreme}
1098: 
1099: The rotating collapse is modified significantly by the magnetic field 
1100: in the case of $\lambda=400$, which corresponds to a field strength 
1101: of merely $0.25~\mu G$ at the typical core radius of 
1102: 0.05~pc. In the absence of any magnetic field, the initial 
1103: configuration collapses into a self-gravitating 
1104: ring (centered at a dimensionless radius $x=0.05$). The ring 
1105: formation is suppressed by the weak field, and a disk is 
1106: produced in its place, as shown in Fig.~\ref{snap_Bpt01}. The figure 
1107: is a snapshot at a representative time $t=2.0\times 10^{12}$~sec, 
1108: plotted in self-similar quantities. The snapshots at other times 
1109: are similar to the one shown. Prominent in the figure is a flatten 
1110: high-density disk, surrounded by a lower density, more puffed up 
1111: ``corona.'' Despite the weakness of the initial field, the 
1112: magnetic pressure due to toroidal field actually dominates the 
1113: thermal pressure, by a factor of $\sim 2.5$ on average. The 
1114: inner part of the disk is more strongly magnetized, with the 
1115: ratio of magnetic to thermal pressures greater than 10 inside 
1116: a radius of $\sim 0.01$, perhaps as a result of a shorter orbital 
1117: period, which allows the field lines to wind up more turns in a 
1118: given time. The poloidal magnetic field inside the disk is, on 
1119: the other hand, very weak, with an energy of order $0.1\%$ of the 
1120: thermal energy on average. The weak poloidal field is susceptible 
1121: to magnetorotational instability \citep[MRI;][]{1998RvMP...70....1B}. 
1122: The role of MRI, as opposed to simple twisting of a pre-existing 
1123: poloidal field, in redistributing angular momentum in the disk 
1124: remains to be quantified (see discussion in \S~\ref{lowbetadisk} 
1125: below). 
1126: 
1127: The disk is rotationally supported. The best evidence for rotational 
1128: support comes from Fig.~\ref{Disk_speeds}, where the rotation speed 
1129: of the material on the equator is compared to the equilibrium 
1130: rotation speed computed from the (total) gravitational potential 
1131: at the representative time shown in Fig.~\ref{snap_Bpt01}. 
1132: Inside the disk (within a radius of about $0.1$), the 
1133: actual rotation speed is close to, but slightly less than, the
1134: equilibrium value. The small deviation is caused by additional
1135: supports from thermal pressure and magnetic forces. Outside
1136: the disk, the rotation speed is substantially less than the
1137: equilibrium value. Also plotted in Fig.~\ref{Disk_speeds} is 
1138: the radial component of the velocity of the material on the 
1139: equator. It shows an outer region that infalls supersonically, 
1140: and an inner region that is nearly static.  
1141: The former is expected of the collapsing envelope, whereas the 
1142: latter is consistent with a rotationally-supported disk 
1143: that is close to an equilibrium in the radial direction. 
1144: The disk does have a small, non-zero, radial velocity, which is 
1145: responsible for transporting material from the outer disk to 
1146: the inner disk. The 
1147: transport, driven by angular momentum redistribution, enables 
1148: mass accretion onto the central object, at 
1149: a dimensionless rate $\sim 0.7$, which is about 1/3 of the 
1150: rate for SIS \citep{1977ApJ...214..488S}.  
1151: 
1152: The rate of mass accretion from the disk to the central object is 
1153: small compared to that from the envelope to the disk. As a result,
1154: the bulk of the material collapsed from the envelope 
1155: is stored in the disk rather 
1156: than the central object. The relative smallness of the central mass 
1157: is also evident from the equilibrium rotation curve shown in 
1158: Fig.~\ref{Disk_speeds}, which deviates significantly from a Keplerian
1159: curve, especially in the outer part of the disk. The deviation, 
1160: particularly the bump near the disk edge around $x\sim 0.1$, is 
1161: an indication that self-gravity is important; indeed, it dominates 
1162: the gravity of the central mass beyond a dimensionless radius 
1163: $\sim 0.04$.  
1164: 
1165: The $\lambda=200$ and 133 cases are qualitatively similar to the
1166: $\lambda=400$ case. They all produce a persistent, 
1167: rotationally supported disk. There are some quantitative 
1168: differences, however. The  
1169: magnetic barrier, barely noticeable in the weakest field 
1170: case, are rather prominent in the stronger field cases. Also, 
1171: as the field strength increases, the disk 
1172: becomes more dynamically active in the radial direction. The 
1173: infall speed can occasionally become supersonic in some (usually 
1174: narrow) regions, indicating that sporadic disk disruption has 
1175: begun already in localized places, especially in the $\lambda=133$ 
1176: case. The disruption becomes more widespread and persistent as the 
1177: field strength increases further. 
1178: 
1179: \subsubsection{Transition from Disk Formation to Suppression}
1180: \label{transition}
1181: 
1182: Sustained disk disruption happens in the case of $\lambda=80$. The 
1183: disruption is illustrated in Fig.~\ref{vr_specmom_Bpt05}, where 
1184: the radial component of the velocity, $v_r$, on the equator 
1185: is plotted at a representative time $t=2\times 10^{12}$~sec. Unlike 
1186: the weaker magnetic field case of $\lambda=400$ shown in 
1187: Fig.~\ref{Disk_speeds}, where the radial velocity nearly vanishes 
1188: everywhere in the disk, $v_r$ is close to zero in four discrete
1189: (narrow) regions, with rapid, transonic or even supersonic, infall 
1190: in between. 
1191: 
1192: The stagnation point at the largest radius is the magnetic barrier,
1193: as in the standard, stronger field case (see Fig.~\ref{vrvphi_eq}).   
1194: In the standard case, there is a second stagnation point interior to 
1195: the magnetic barrier, which we have interpreted as the centrifugal 
1196: barrier, where the infall is slowed down temporarily and the magnetic
1197: braking is locally enhanced. In the 
1198: $\lambda=80$ case, the stagnation points interior to the magnetic
1199: barrier are also centrifugal barriers. The reason for the 
1200: existence of more than one barriers can be seen from the upper curve in 
1201: Fig.~\ref{vr_specmom_Bpt05}, where the specific angular momentum
1202: is plotted as a function of radius. Even though a good fraction (about
1203: half) of the 
1204: angular momentum is removed at the first centrifugal barrier, there
1205: is still a significant amount left interior to it. The material  
1206: recollapses quickly towards the center, spinning up at the same 
1207: time, until a second centrifugal barrier is encountered. The slowdown  
1208: allows more time for the fluid rotation to twist up the field lines. 
1209: The magnetic braking removes enough angular momentum from the material in 
1210: the equatorial region to enable it to recollapse for a third time, 
1211: all the way to near the inner boundary, where the infall is 
1212: rearrested by rotation. In some sense, the rotating material tries 
1213: to settle into a rotationally supported disk, but the attempt is 
1214: repeatedly frustrated by magnetic braking. The result is a collection 
1215: of discrete rings that are temporarily supported by the 
1216: centrifugal force, separated by regions of rapid collapse, where 
1217: the rotational support is insufficient to balance the gravity. 
1218: Note that in 3D, the rings may break up into arcs or clumps, some of which may 
1219: become self-gravitating and form (sub-)stellar objects. 
1220: 
1221: 
1222: The material in the stagnation region near a centrifugal barrier
1223: loses a large fraction of its angular momentum, and collapses 
1224: quickly inwards after exiting the region. Why is it possible to 
1225: remove a large fraction of the angular momentum before the material 
1226: begins collapsing dynamically? One may expect the gas to move 
1227: (inward) out of the stagnation region as soon as its specific
1228: angular momentum 
1229: is reduced slightly below the local equilibrium value, because of an 
1230: imbalance of gravity and centrifugal force. If this were true, the
1231: amount of angular momentum reduction achievable in the region 
1232: will be quite limited. 
1233: However, it takes a finite amount of time for even the full gravity 
1234: to reaccelerate the material to (say) the sound speed from at rest, no 
1235: matter how small a value its angular momentum has been reduced to; 
1236: in many cases, the gravity must first reverse the (slowly) expanding 
1237: motion often found near the centrifugal barrier, which takes
1238: additional time. If 
1239: the magnetic torque is strong enough, it can remove a large fraction, 
1240: if not all, of the angular momentum of the gas within that time. 
1241: In other words, it is the inertia of the stalled gas that enables the 
1242: magnetic braking to reduce the specific angular momentum significantly 
1243: below the local equilibrium value. The efficient, localized, braking 
1244: is responsible for the unique, ``hopping,'' pattern of alternating 
1245: stagnation and rapid infall seen in Fig.~\ref{vr_specmom_Bpt05}.
1246: 
1247: The rotationally supported rings show up clearly in the density map 
1248: (Fig.~\ref{disk_unitvec_Bpt05}). They are overdense regions near 
1249: the equator created by stalled infall, 
1250: which leads to a pileup of matter.
1251: The slowdown also allows more time for magnetic braking to operate. 
1252: The clearest evidence for enhanced braking comes from the outer most 
1253: ring near the dimensionless radius $x\sim 0.1$: there is an outflow 
1254: driven from that location at an acute angle with respect to the 
1255: equatorial plane, as in the magnetic bubble of the standard case. 
1256: Evidence for braking-driven outflows from the inner two rings is 
1257: less clean-cut; they may be masked by the chaotic meridianal flow 
1258: pattern in the puffed-up region outside the equatorial rings. The 
1259: chaotic motions are generally subsonic. They are dominated in 3D by 
1260: supersonic rotation. Indeed, the puffed-up structure is equivalent 
1261: to the magnetic bubble in the standard model: both are supported
1262: by a combination of (mostly toroidal) magnetic fields and rotation.
1263: In some sense, the puffed-up structure is a collection of (relatively 
1264: weak) bubbles, driven from multiple centrifugal barriers whose 
1265: locations vary in time. Such a structure is also present in the 
1266: somewhat stronger field case of $\lambda=40$. We believe it is a 
1267: generic feature of the magnetized rotating collapse when the field
1268: is strong enough to disrupt the rotationally supported disk but 
1269: too weak to channel the outflowing material into a single, coherent
1270: bubble. This feature is discussed further in
1271: \S~\ref{magnetogyrosphere}. 
1272: 
1273: We note that the number of discrete centrifugal barriers depends on 
1274: the field strength. As the field strength decreases, more and more 
1275: centrifugal barriers appear, until they start to overlap and merge 
1276: into a single, more or less contiguous structure --- a rotationally 
1277: supported disk (see the $\lambda=400$ case discussed above). 
1278: Conversely, the disk is replaced by fewer and fewer centrifugal 
1279: barriers as the field strength increases. For the standard, 
1280: $\lambda=13.3$ case, the number of identifiable centrifugal barriers 
1281: is one at most times. This number can drop to zero for more 
1282: strongly magnetized cases, to which we turn our attention next. 
1283: 
1284: \subsection{Relatively Strong Field Cases} 
1285: \label{strongfield}
1286: 
1287: We first focus on the time evolution of the central point mass 
1288: (Fig.~\ref{Mstar_HighB_noscaled}), which provides an overall  
1289: measure of the collapse solutions. The point mass at a give time 
1290: remains similar or even becomes smaller as the field strength 
1291: increases, unlike the moderately weak field ($\lambda=80$ to 13.3) 
1292: cases. The trend is especially clear for the strongest field  
1293: case of the whole set ($\lambda=4$). This change in the dependence 
1294: of point mass on the field strength indicates a change in the 
1295: solution behavior. In the weak field cases, the mass accretion 
1296: is limited by angular 
1297: momentum removal, whose rate generally increases with the field strength. 
1298: Angular momentum removal is no longer the limiting factor in the 
1299: strong field cases: the (poloidal) fields at small radii are 
1300: apparently strong enough, especially after amplification from 
1301: collapse, that a relatively small twist can remove most of the 
1302: angular momentum of the collapsing material and redistribute it 
1303: to large distances. The limiting factor shifts instead to the 
1304: magnetic tension, which replaces the rotation as the main 
1305: impediment to mass accretion. The tension is stronger (relative 
1306: to gravity) for a smaller mass-to-flux ratio $\lambda$. Indeed, 
1307: if $\lambda < 1$ (i.e., a magnetically subcritical core), the 
1308: tension force would prevent core collapse and mass accretion 
1309: altogether. 
1310: 
1311: Unlike angular momentum, the magnetic flux cannot be removed in 
1312: the ideal MHD limit that we are working in. It accumulates near 
1313: the central object --- in the central hole in our simulation. 
1314: The flux accumulation and its sudden release through reconnection 
1315: allow the magnetic tension to suppress mass accretion onto the
1316: central object over an extended period of time even in a 
1317: magnetically supercritical core with an $\lambda$ significantly 
1318: larger than unity. They are responsible for 
1319: the step-like variations of the point mass shown in 
1320: Fig.~\ref{Mstar_HighB_noscaled}. Before discussing magnetic 
1321: reconnection in \S~\ref{reconnection}, we first consider the 
1322: collapse solution at a time representative of the conditions 
1323: before a reconnection event.  
1324: 
1325: \subsubsection{Removal of Angular Momentum in a Magnetic Braking-Driven Wind} 
1326: \label{nodisk}
1327: 
1328: We show in Fig.~\ref{snap_B1pt0} a snapshot of the $\lambda=4$ case 
1329: at a representative time before reconnection. Although one can still
1330: divide the collapse solution into four regions as in the standard 
1331: case (collapsing envelope, equatorial pseudodisk, magnetic bubble 
1332: and polar funnel, see Fig.~\ref{snap}), these regions are less
1333: distinct. In particular, the magnetic bubble (defined as the 
1334: region where the toroidal field is stronger than the poloidal field) 
1335: is much less prominent than in the standard $\lambda=13.3$ case. 
1336: It is located at a larger distance from the center and contains a 
1337: much lower mass and angular momentum. Within a radius of $x=3$, the 
1338: dimensionless mass of the bubble is only 0.18 and angular momentum 
1339: is 0.28. These are to be compared with the bubble mass of $\sim 2.3$ 
1340: and angular momentum of $\sim 1.4$ estimated earlier for the standard 
1341: case. Indeed, the bubble can be more profitably viewed as part of 
1342: a broader magnetic braking-driven wind, where the field lines are
1343: twisted to varying degrees. The (poloidal) field is strong enough 
1344: that even a relatively small twist can exert an appreciable torque 
1345: on the gas: for a fixed twist angle, the magnetic torque is 
1346: proportional to the field strength squared. Compared to the
1347: standard case, the twisted field lines unwind faster because of 
1348: a higher Alfv\'en speed. Another difference is the absence of a 
1349: prominent magnetic wall in which the field lines are severely 
1350: compressed; the stronger fields are more difficult to compress. 
1351: As a result, the collapsing envelope merges more smoothly into 
1352: the equatorial pseudodisk, which is now much thicker. The 
1353: centrifugal force inside the pseudodisk remains substantially 
1354: below the gravity in general. In some sense, parcels of infalling 
1355: material are chasing their centrifugal barriers but unable to 
1356: reach them because of efficient magnetic braking, which continuously 
1357: moves the barriers inward. We conclude that the formation of a 
1358: rotationally supported disk is completely suppressed in the 
1359: relatively strong field regime, in agreement with the earlier 
1360: work of ALS03. 
1361: 
1362: \subsubsection{Magnetic Reconnection and Episodic Mass Accretion}
1363: \label{reconnection}
1364: 
1365: A feature not apparent in the early work is episodic mass accretion. As 
1366: can be seen from Fig.~\ref{Mstar_HighB_noscaled}, most of the central 
1367: mass in the $\lambda=4$ case is accreted in discrete bursts, which 
1368: are caused by magnetic reconnection. To determine how the bursts are
1369: triggered, we focus on the large burst around $t=9.1\times 10^{11}$~sec. 
1370: A snapshot of the inner part of the collapse solution before 
1371: reconnection is shown in the left panel of Fig.~\ref{burst}. Note 
1372: the nearly radial field lines that thread the surface of the inner 
1373: hole. Together with their mirror images (of opposite polarity) 
1374: in the lower quadrant (not shown), they form a split magnetic
1375: monopole. The 
1376: monopole compresses the material in the equatorial region into a 
1377: thin, dense (pseudo)disk, which collapse supersonically. As more 
1378: and more disk material falls into the point mass, more and more 
1379: field lines are added to the central hole. The growing monopole 
1380: pushes on the disk material harder and harder from above and 
1381: below. When field lines of opposite polarities are pressed into 
1382: a single cell on the equator, they reconnect quickly due to 
1383: numerical diffusion. We observe a maximum in the magnetic 
1384: flux that threads the inner hole shortly before the onset of violent 
1385: reconnection. There is a brief period of time when matter falls 
1386: into the central hole but not the field lines. The separation 
1387: indicates that significant magnetic diffusion is present before 
1388: the violent reconnection. The reconnected field lines pull the 
1389: disk material outward (see the right panel of Fig.~\ref{burst}), 
1390: leaving behind an evacuated region that is unable to keep the 
1391: oppositely directing, monopolar field lines apart. As a result, 
1392: the field lines threading the central hole rush onto the disk, 
1393: adding to the reconnected field lines that accelerate the 
1394: equatorial material outward. Except for a tiny amount of mass 
1395: falling into the central hole in the polar region, mass accretion 
1396: is essentially terminated. 
1397: 
1398: As the dense blob (or ring in 3D) shown in the right panel of 
1399: Fig.~\ref{burst} moves outwards, it gradually slows down for several 
1400: reasons. First, it ploughs into infalling material, which cancels 
1401: some of its outward momentum. Second, as the mass of the blob 
1402: increases, through both sweeping up material along its path and 
1403: accretion from above and below along field lines, the gravitational 
1404: pull of the central point mass increases as well. In addition, 
1405: the magnetic tension that pushes the blob outwards from behind 
1406: weakens with time, as the bent field lines straighten up. Sooner 
1407: or later, there will be enough material accumulated for the 
1408: gravity to overwhelm the magnetic forces and pull the matter 
1409: toward the center again. The recollapsing flow in the equatorial 
1410: region typically moves at a speed well below the local free fall 
1411: speed, because it has to drag along with it a rather strong 
1412: magnetic field, which resists the gravitational pull through 
1413: tension force. 
1414: 
1415: Once the bulk of the recollapsing material reaches the inner hole,
1416: it produces a burst in mass accretion. At the same time, the 
1417: accreted material deposits its magnetic flux on the surface of the
1418: inner hole, creating a growing split magnetic monopole that, in 
1419: time, would shut off the mass accretion through reconnection, and 
1420: eject the remaining material to larger distances, starting another
1421: cycle of mass accumulation, collapse, and ejection. In some sense, 
1422: the split monopole acts as an ever tightening ``magnetic gate:'' 
1423: as magnetic flux accumulates, mass 
1424: accretion becomes more and more difficult, until shut off almost 
1425: completely by reconnection; the reconnection allows a large amount of 
1426: magnetic flux (associated with the central point mass) to act on a 
1427: small amount of (disk) matter. 
1428: 
1429: In summary, we have computed the collapse of rotating, strongly 
1430: magnetized cores with much higher angular resolution at small radii 
1431: than the previous calculations that employ a cylindrical coordinate 
1432: system. We confirmed the conclusion of ALS03 that the magnetic 
1433: braking in such cores is too strong for the rotationally supported 
1434: disks to form in the ideal MHD limit. We went one step further, 
1435: showing that the disk suppression holds true even in the presence 
1436: of (numerically triggered) reconnection. The main effect of the 
1437: reconnection events is to generate episodic mass accretion. 
1438: 
1439: \section{Effects of Rotation Speed} 
1440: \label{rotation}
1441: 
1442: The initial rotation speed $v_0=0.5$ (in units of sound speed) 
1443: adopted in the standard 
1444: simulation is on the high side of the observationally inferred 
1445: range. Here we discuss three variants 
1446: of the standard model, with $v_0=0$, $0.125$ and $0.25$, 
1447: respectively. Without rotation, the collapsing envelope is
1448: expected to fall into the central object at a high rate. This 
1449: is indeed the case, as shown in Fig.~\ref{Mstar_diffrot}. The 
1450: dimensionless point mass quickly approaches a nearly constant
1451: value, indicating that the non-rotating collapse is approximately
1452: self-similar. There are small oscillations in the point mass, 
1453: which result from magnetic flux accumulation at the inner hole, 
1454: as in the standard simulation. This episodic behavior is therefore
1455: independent of rotation, in accordance with the interpretation
1456: based on reconnection that we advanced in the last section. The 
1457: value of the dimensionless point mass, $\sim 8$, corresponds to 
1458: a mass accretion rate that
1459: is about 4 times higher than the rate for the collapse of a 
1460: static SIS \citep[1.95;][]{1977ApJ...214..488S}. 
1461: The accretion rate decreases as the 
1462: rotation speed increases, as shown in Fig.~\ref{Mstar_diffrot}. 
1463: For the relatively slowly rotating cases of $v_0=0.125$ and $0.25$, 
1464: the point mass started out as high as that in the non-rotating
1465: case, but gradually approaches a lower, nearly constant value,
1466: indicating that an approximate self-similarity is reached in
1467: these cases as well. 
1468: 
1469: To illustrate the effects of rotation on the structure of the 
1470: magnetized collapsing flow, we show in Fig.~\ref{multi_diffrot} 
1471: a representative snapshot for each of the cases with $v_0=0$,
1472: $0.125$, $0.25$ and $0.5$. In the absence of 
1473: rotation, three of the four dynamically distinct regions that we
1474: have identified in the standard (rotating) model are clearly 
1475: present: the polar funnel flow (Region IV), equatorial pseudodisk 
1476: (region II), and collapsing 
1477: envelope (Region I). They are separated by a magnetic wall, 
1478: which divides the outer collapsing envelope from the inner 
1479: two regions, both of which are 
1480: dominated by the central split magnetic monopole.  
1481: This magnetic structure is remarkably similar to
1482: the one sketched in Fig.~1 of \citet{2006ApJ...647..374G}. Rotation 
1483: changes the appearance of the collapse 
1484: solution by inflating a bubble in between the polar funnel and
1485: the pseudodisk. It is bounded from outside by the envelope. 
1486:  As one may expect intuitively, the size 
1487: (mostly the width) of the bubble increases with the initial 
1488: rotation speed; for the same initial magnetic field, a faster
1489: rotation should wind up the field lines more, generating a 
1490: larger magnetic bubble, as shown in the figure. The difference
1491: in the rotation speed does not appear to change the dynamics of 
1492: the core collapse and magnetic braking fundamentally, however. 
1493: 
1494: \section{Discussion}
1495: \label{discussion}
1496: 
1497: \subsection{Characteristic Mass-to-Flux Ratio for Significant Magnetic
1498:   Braking}
1499: \label{lambdabraking}
1500: 
1501: Our most important result is that the rotationally supported disk
1502: can be disrupted by a weak magnetic field corresponding to a
1503: mass-to-flux ratio $\lambda \gg 1$. This result may seem surprising 
1504: at the first 
1505: sight, in view of the general expectation that the field is 
1506: dynamically important only when $\lambda \lesssim 1$. To illuminate 
1507: the reason behind this seemingly unexpected result, we provide an 
1508: order-of-magnitude estimate of the characteristic mass-to-flux 
1509: ratio for significant magnetic braking from the following equation:
1510: \begin{equation}
1511: M_d l_d \sim {\vert B_{\varphi}\vert B_z \varpi_d \over 4\pi}\ 
1512: (2 \pi \varpi_d^2)\ t,
1513: \label{LeqTaut}
1514: \end{equation}
1515: where $M_d$, $l_d$ and $\varpi_d$ are the mass, specific angular
1516: momentum and radius of the disk in the absence of magnetic 
1517: braking at time $t$, and $B_\varphi$ and 
1518: $B_z$ are the toroidal and vertical components of the magnetic field
1519: on the disk surface. The equation is simply a condition on the rate 
1520: of magnetic braking that would be required to remove all of the disk 
1521: angular momentum within the time $t$. 
1522: 
1523: The disk mass $M_d$ is related to the magnetic flux threading the
1524: disk, $\pi \varpi_d^2 B_z$, through the dimensional mass-to-flux 
1525: ratio $\lambda/(2\pi G^{1/2})$, which is conserved along any field 
1526: line during the core collapse in the ideal MHD limit. Eliminating
1527: $B_z$ in terms of $M_d$ and $\lambda$ from equation (\ref{LeqTaut}),
1528: we have 
1529: \begin{equation}  
1530: \lambda^2 \sim 2 \left( \frac{\vert B_{\varphi}\vert}{B_z}\right) 
1531: \left(\frac{G M_d t}{l_d \varpi_d}\right). 
1532: \label{lambda1}
1533: \end{equation}
1534: To simplify the right hand side of the equation further, we assume 
1535: that the disk is self-gravitating and rotationally supported 
1536: (as in the $\lambda=400$ case shown in Fig.~[\ref{snap_Bpt01}]), with 
1537: a rotation speed $v_d\sim (G M_d/\varpi_d)^{1/2}$ and specific 
1538: angular momentum $l_d\sim v_d \varpi_d$. The above equation then 
1539: becomes
1540: \begin{equation}  
1541: \lambda^2 \sim 4\pi \left( \frac{\vert B_{\varphi}\vert}{B_z}\right)
1542: \left(\frac{t}{p}\right), 
1543: \label{lambda2}
1544: \end{equation}
1545: where $p= 2\pi \varpi_d/v_d$ is the disk rotation period. 
1546: 
1547: Clearly, the characteristic mass-to-flux ratio can greatly exceed 
1548: unity for two reasons: (1) tight wrapping of the disk magnetic 
1549: field, $\vert B_\varphi\vert \gg B_z$, which increases the rate of 
1550: angular momentum removal through magnetic braking ($\propto \vert 
1551: B_\varphi\vert  
1552: B_z$), and (2) long braking time compared to the disk rotation 
1553: period, $t \gg p$. The braking time $t$ can be identified with 
1554: the time for core collapse. 
1555: 
1556: The two factors are not entirely independent of each other. Since 
1557: the toroidal magnetic field is created out of the poloidal field
1558: by rotation, the field line is expected to become increasingly more 
1559: twisted with time. Indeed, it may not be unreasonable to 
1560: approximate the magnetic twist by the number of turns that the 
1561: disk has rotated around the center within time $t$, so that   
1562: \begin{equation}
1563: {{\vert B_\varphi\vert} \over B_z}\sim {t\over p}.
1564: \end{equation}
1565: In this case, the characteristic mass-to-flux ratio reduces to
1566: \begin{equation}  
1567: \lambda \sim 2\sqrt{\pi} \left(\frac{t}{p}\right).  
1568: \label{lambda3}
1569: \end{equation}
1570: 
1571: To obtain concrete numbers, let us consider a disk formed at time
1572: $t$ out of a singular isothermal toroid with an (angle-averaged) 
1573: initial mass distribution $M(r)= 2(1+H_0) a^2 r/G$ and spatially constant 
1574: rotation speed $V_0$. Suppose, as in the case of SIS \citep{1977ApJ...214..488S}, 
1575: half of the mass within a radius $r= a t$ falls into the 
1576: collapsed object, the disk, at any given time $t$. In this case, 
1577: the disk mass $M_d\sim (1+H_0) a^3 t/G$ and specific angular
1578: momentum $l_d \sim V_0 a t/2$, yielding a disk radius
1579: \begin{equation}
1580: \varpi_d \sim {l_d^2 \over G M_d}\sim {V_0^2 t\over 8 (1+H_0) a},
1581: \end{equation}
1582: and rotation period
1583: \begin{equation}
1584: p \sim {\pi \over 4(1+H_0)^2}\left({V_0\over a}\right)^3 t.
1585: \label{period}
1586: \end{equation}
1587: Substituting equation~(\ref{period}) into equation~(\ref{lambda3}),
1588: we finally have 
1589: \begin{equation}
1590: \lambda \sim {8(1+H_0)^2\over \sqrt{\pi}} \left({a\over V_0}\right)^3.   
1591: \label{lambda4}
1592: \end{equation} 
1593:  
1594: For the initial mass distribution with $H_0=0.4$ and rotation speed
1595: $V_0=0.5 a$ adopted in \S~\ref{fieldstrength}, we obtain a 
1596: characteristic mass-to-flux ratio for significant magnetic braking 
1597: of $\lambda \sim 70$. This number is remarkably close to our 
1598: numerically determined value for the disk disruption to begin, 
1599: which is around 100. Given the crudeness of the estimate, the
1600: agreement is encouraging. The analytic estimate brings out the 
1601: key ingredient for efficient disk braking by even weak magnetic
1602: fields: short disk rotation period compared 
1603: to the core collapse time. In addition, it indicates that the disk 
1604: in a more slowly rotating core can be disrupted by a weaker magnetic 
1605: field, which makes physical sense. 
1606: 
1607: Besides the rotation rate, another factor that may also affect 
1608: the value of the mass-to-flux $\lambda$ for significant magnetic 
1609: braking is (non-axisymmetric) gravitational torque. The torque can 
1610: speed up mass accretion through a self-gravitating disk, shortening 
1611: the time available for field line wrapping and magnetic braking. 
1612: This additional means of angular momentum redistribution is expected 
1613: to decrease the efficiency of magnetic braking somewhat. To fully 
1614: quantify this effect, high resolution 3D simulations are required. 
1615: Our 2D (axisymmetric) calculations, while idealized in several ways 
1616: (see \S~\ref{future} below), allow us to concentrate on the role  
1617: of magnetic braking in disk dynamics without extra complications
1618: from gravitational torques and fragmentation. We believe the basic 
1619: conclusion that even weak magnetic fields with $\lambda \gg 1$ are 
1620: dynamically important in disk formation and evolution is robust.
1621: 
1622: \subsection{Magnetogyrosphere}
1623: \label{magnetogyrosphere}
1624: 
1625: Surrounding the strongly braked disk is a vertically extended 
1626: structure where the bulk of the angular momentum of the material 
1627: collapsed from the envelope is parked. The structure is supported 
1628: by a combination of (toroidal) magnetic field and rotation. We term 
1629: it ``magnetogyrosphere,'' to distinguish it from the purely 
1630: magnetically dominated ``magnetosphere'' or the rotationally supported 
1631: disk.
1632: 
1633: An example of the magnetogyrosphere is shown in 
1634: Fig.~\ref{disk_unitvec_Bpt05}, 
1635: where $\lambda=80$. In this particular case, the collapsing envelope material 
1636: is channeled by the magnetic wall surrounding the magnetogyrosphere into 
1637: the equatorial region, where the infalling material crosses multiple 
1638: centrifugal barriers on its way to the center. The braking of the equatorial
1639: material, 
1640: particularly at the multiple centrifugal barriers, is what powers the rather 
1641: chaotic motions inside the slowly expanding 
1642: magnetogyrosphere. The dimensionless size of the magnetogyrosphere is $\sim 
1643: 0.25$, corresponding to a physical size of $\sim 10^3$~AU at a time of 
1644: $2\times 10^{12}$~sec, for a sound speed of $a=0.3$~km/s. By this time, 
1645: the central point mass is about $0.3$~M$_\odot$, whereas the mass of 
1646: the magnetogyrosphere is about $1.1$~M$_\odot$ within $10^3$~AU of the 
1647: origin (we ignore the rotating, outflowing region near the axis at 
1648: larger distances that contains relatively little mass). Most of the 
1649: material collapsed from the envelope is therefore stored in the 
1650: magnetogyrosphere, 
1651: rather than going into the central object. It is prevented from falling 
1652: into the center by the combined effect of magnetic field and
1653: rotation. To be quantitative, 
1654: the mass weighted flow speed in the poloidal plane is only $0.13$~km/s 
1655: inside the magnetogyrosphere, which is less than half of the sound 
1656: speed $a=0.3$~km/s. This is consistent with the fact that the size 
1657: of the magnetogyrosphere is increasing slowly, at a rate of roughly a 
1658: quarter of the sound speed. For comparison, the free fall speed $V_{\rm 
1659: ff}=[2 G M(r)/r]^{1/2}$ at the outer edge of the magnetogyrosphere 
1660: at $10^3$~AU is $1.55$~km/s, an order of magnitude higher than the 
1661: average poloidal speed. The bulk of the magnetogyrospheric material 
1662: is levitated against the gravity by a combination of (mostly 
1663: magnetic) pressure gradient and centrifugal force from rotation; 
1664: the mass averaged rotation speed is $1.23$~km/s, not far from the 
1665: free fall speed. 
1666: 
1667: Qualitatively similar behaviors are found in other moderately weakly
1668: magnetized models of $\lambda=40$, 20, and 13.3. For 
1669: example, in the standard model with $\lambda=13.3$, the magnetic 
1670: braking is strong enough that only a single centrifugal barrier 
1671: exists. After exiting the barrier, the equatorial material falls 
1672: straight to the center whereas the extracted angular momentum is 
1673: stored in a slowly expanding bubble --- the magnetogyrosphere, an 
1674: extended structure again dominated by a combination of magnetic 
1675: field and rotation (see \S~\ref{ss}). Compared with the $\lambda
1676: =80$ case, the magnetogyrosphere contains less material (with a 
1677: mass somewhat less than, rather than well exceeding, the central 
1678: mass), which moves faster on average and in a more ordered fashion 
1679: in the poloidal plane. 
1680: 
1681: To illustrate what the magnetogyrosphere may look like observationally, 
1682: we show in Fig.~\ref{columndensity} the column density distributions 
1683: of the $\lambda=80$ and $13.3$ viewed perpendicular to the axis. In both
1684: cases, there is a highly-flattened, dense equatorial region, sandwiched 
1685: by a more puffed-up structure. The former may potentially be mistaken 
1686: for a rotationally-supported equilibrium disk around the central object;  
1687: in reality, it is a highly dynamic structure that collapses 
1688: supersonically and rotates significantly below the local equilibrium 
1689: rate at most radii, except near the centrifugal barrier(s). The latter 
1690: is the slowly expanding magnetogyrosphere. It is held up mostly by the 
1691: (toroidal) magnetic field in the vertical direction and laterally by 
1692: rotation.
1693: 
1694: \subsubsection{Connection to Observations: IRAM 04191 and HH 211}
1695: 
1696: Magnetogyrospheres should exist in some form. Fundamentally, this 
1697: is because a rotating, collapsing core will necessarily develop a 
1698: differential rotation, which inevitably interacts strongly with a 
1699: poloidal magnetic field, as long as the matter and field are well 
1700: coupled. The interaction is especially strong near the centrifugal 
1701: barrier, where the slowdown of infall motion allows more time for 
1702: the differential rotation to wrap up the field lines. When the field 
1703: is strong enough to remove angular momentum from the collapsing 
1704: material, but too weak to transport most of the removed angular 
1705: momentum out of the system, a magnetogyrosphere develops. In 
1706: principle, the magnetogyrosphere can be distinguished easily 
1707: from the collapsing (inner) envelope, since both the velocity 
1708: and magnetic field are dominated by the toroidal component in 
1709: the former and by the poloidal component in the 
1710: latter. They also have different volume and column densities, which 
1711: may also help separating the two. In practice, the mass distribution 
1712: and especially the kinematics of both the magnetogyrosphere and the 
1713: inner envelope are expected to be modified by the powerful 
1714: protostellar winds, which sweep the ambient medium into bipolar 
1715: molecular outflows. To search for observational evidence for the 
1716: magnetogyrosphere, the best place to start may be those Class 
1717: 0 sources that have relatively narrow molecular outflows. 
1718: 
1719: One of the best studied Class 0 sources is IRAM 04191. The source 
1720: has a low luminosity of $\sim 0.1 ~L_\odot$ \citep*{1999ApJ...513L..57A,2006ApJ...651..945D}, 
1721: which points to a central
1722: stellar object of mass perhaps no more than 0.1~$M_\odot$. The 
1723: object is surrounded by an extended envelope of mass $\sim ~0.5 
1724: M_\odot$ within a radius of 4200~AU, estimated from dust continuum 
1725: emission \citep{1999ApJ...513L..57A}. Molecular line observations have 
1726: established that the envelope is both infalling and 
1727: differentially rotating \citep*{2002A&A...393..927B}. It is well traced 
1728: by $N_2H^+$, except close to the center, where the molecule 
1729: appears to be depleted \citep{2004A&A...419L..35B}. \citet*{2005ApJ...619..948L} 
1730: found that the $N_2H^+$ envelope consists of two kinematically 
1731: distinct parts: a fluffy outer region and a dense inner structure
1732: that resembles a thick, clumpy, tilted ring. The thick ring-like 
1733: structure has an average radius $r_0 \sim 1400~AU$ (or $10^{\prime
1734: \prime}$). The most puzzling, and potentially the most revealing, 
1735: kinematic feature is the inferred infall speed at the radius $r_0$: 
1736: $v_r (r\sim r_0) \sim 0$. The small radial speed implies a pileup 
1737: of infall material, and may have created the dense ring in the 
1738: first place. The stagnation of infall, if confirmed, would have 
1739: strong implications for the dynamics of mass accretion and angular 
1740: momentum evolution. It may mark the transition from the collapsing 
1741: envelope to a circumstellar structure dominated by magnetic fields 
1742: and rotation --- the magnetogyrosphere in our picture.
1743: 
1744: Rotation alone can in principle stop the observed infalling envelope at 
1745: the radius $r_0$, as would be the case near a centrifugal barrier. 
1746: The rotation speed at $r_0$ is estimated to be $V_{\varphi,0}\sim 
1747: 0.16$~km/s, corresponding to a centrifugal force $V_{\varphi,0}^2 / 
1748: r_0 \sim 1.22\times 10^{-8}$~cm/s$^2$ per unit mass. It is to be 
1749: compared with the gravitational acceleration $G [M_{\rm env}(r_0)+M_*]
1750: /r_0^2\sim 7.56 \times 10^{-8}$~cm/s$^2$, where we have assumed 
1751: 0.1~M$_\odot$ for the central stellar mass and 0.15~M$_\odot$ for the 
1752: mass within $r_0=1400$~AU (out of the estimated $\sim 0.5$~M$_\odot$ 
1753: within 4200~AU). The ratio of gravitational to centrifugal forces is 
1754: therefore $\sim 6$, although it may be lower by perhaps up to a factor 
1755: of 2 due to uncertainties in the mass estimates. In any case, the 
1756: inferred rotation does not appear to be faster enough to counter the 
1757: gravity \citep{2005ApJ...619..948L}, which implies that the stagnation 
1758: point $r_0$ is not a centrifugal barrier. This result is consistent with 
1759: the constraint that an optically thick disk, if exists at all, must 
1760: be smaller than 10~AU \citep{2002A&A...393..927B}, well inside the 
1761: stagnation radius $r_0$. 
1762: 
1763: The infalling envelope can also be stopped at the magnetic barrier. 
1764: The barrier is a general feature produced by the poloidal magnetic 
1765: field lines draping over the outer surface of the magnetogyrosphere 
1766: (see Fig.~\ref{snap} and 
1767: Fig.~\ref{disk_unitvec_Bpt05}). The field lines channel the envelope 
1768: material towards the equator, where it is slowed down temporarily, 
1769: before recollapsing towards the center. In the case of IRAM 04191, 
1770: there is evidence that the material interior to the stagnation radius 
1771: $r_0$ is moving inwards at a speed comparable to the sound speed 
1772: \citep{2005ApJ...619..948L}. The rapid infall implies that there is not enough 
1773: angular momentum to hold up the material by rotation. Indeed, there 
1774: is some hint that the region interior to $r_0$ rotates more or less 
1775: as a solid body, which would point to a strong braking, as expected
1776: in a magnetic barrier, where the field is strong and the slowdown 
1777: of infall allows more time for the magnetic braking to operate. 
1778: The braking is apparently so efficient that an 
1779: optically thick disk does not form until within 10~AU of the central 
1780: object, if at all \citep{2002A&A...393..927B}. 
1781: 
1782: If strong magnetic braking is indeed operating inside the stagnation
1783: radius $r_0$ of IRAM 04191, where is the bulk of the extracted angular 
1784: momentum deposited? 
1785: One possibility is the slowly outflowing region (with radial speeds 
1786: of order 0.4~km/s) that shows up in the position-velocity diagram 
1787: along the minor axis of the ring-like $N_2H^+$ structure \citep{2005ApJ...619..948L}. 
1788: This 
1789: region could be part of the magnetogyrosphere. Alternatively, it 
1790: could be material accelerated outwards by the fast protostellar 
1791: wind \citep{2005ApJ...619..948L}. The ambiguity illustrates one of the 
1792: observational challenges in probing the magnetogyrosphere directly;
1793: it may easily be masked by the wind-driven motions. This problem 
1794: may be alleviated using molecules that are little affected by 
1795: wind-interaction. Another complication in direct probing of the 
1796: magnetogyrosphere is the depletion of molecules onto dust grains. 
1797: Even though $N_2H^+$ is among the last molecules to disappear at 
1798: high densities \citep{2002ApJ...569..815T}, it is heavily 
1799: depleted close to the central object in 
1800: IRAM 04191 \citep{2004A&A...419L..35B,2005ApJ...619..948L}. 
1801: 
1802: In the absence of detailed kinematic information, the morphology 
1803: of circumstellar material may provide indirect evidence for the 
1804: magnetogyrosphere. A possible example is the HH 211 system, the 
1805: prototype of the class of highly collimated CO molecular outflows 
1806: \citep{1999A&A...343..571G,1999osps.conf..227B}. It has a pair of 
1807: oppositely directing, fast-moving jets, each enclosed by a narrow, 
1808: slower shell. Recent PdBI observations of the source at a 
1809: resolution of $0.35^{\prime\prime}$ reveals that the outer shell 
1810: is much narrower within about $10^3$~AU of the central object 
1811: (indeed unresolved transversely, Gueth et al., in preparation). 
1812: It appears that, as the outflow propagates away from the central 
1813: region, it is initially confined laterally by a $10^3$~AU-scale 
1814: structure before suddenly expanding into the more widely open 
1815: shell-like structure observed at larger distances. A natural 
1816: possibility for the structure is the magnetogyrosphere, which 
1817: can provide confinement through both thermal pressure and magnetic 
1818: ``hoop'' stresses associated with the toroidal field. Indeed, a 
1819: glance at the density distribution in Fig.~\ref{disk_unitvec_Bpt05} 
1820: (or Fig.~\ref{snap}) reveals a narrow funnel in the polar region 
1821: close to the origin where the outflow can plausibly be confined. 
1822: To strengthen the case for magnetogyrosphere as the confining 
1823: structure at the base of the HH 211 molecular outflow, high resolution 
1824: interferometric observations that probe the kinematics (and ideally 
1825: the magnetic structure) of the confining structure are needed. 
1826: 
1827: To summarize, there is tantalizing evidence for a magnetic braking-driven 
1828: magnetogyrosphere in IRAM 04191 and, to a lesser extent, HH 211. To 
1829: make a stronger case, more high resolution kinematic data are needed, 
1830: perhaps from the recently completed CARMA and upgraded PdBI, and 
1831: especially the upcoming ALMA. A unique model prediction is the 
1832: existence of more than one centrifugal barriers interior to the 
1833: magnetic barrier, depending on the field strength and assuming 
1834: good coupling between the field and matter. Detection of such a 
1835: feature would clinch the case for disk disruption by (repeated) 
1836: magnetic braking. 
1837: 
1838: \subsubsection{Dispersal of Magnetogyrosphere and Late-Time Disk Formation} 
1839: \label{dispersal}
1840: 
1841: If the bulk of the angular momentum of the collapsed material is 
1842: indeed deposited in the magnetogyrosphere, what eventually happens 
1843: to the magnetogyrosphere becomes an important issue in the 
1844: angular momentum evolution of the system. We believe the fate of the 
1845: magnetogyrosphere is tied to that of the dense core material, 
1846: the majority of which is not incorporated into stars. For example, 
1847: in the Taurus molecular clouds, the dense HCO$^+$ cores typically contain 
1848: a few to several solar masses \citep{2002ApJ...575..950O}, whereas 
1849: the typical mass of 
1850: the YSOs is only $\sim 0.5$~M$_\odot$ \citep{1995ApJS..101..117K}. It is
1851: likely that the bulk of the core material is removed by the 
1852: powerful protostellar wind \citep{1987ARA&A..25...23S,2000ApJ...545..364M}. 
1853: Since the specific angular momentum of the core material 
1854: tends to increase with distance, it follows that the protostellar 
1855: wind blows away the majority of not only the mass, but also the 
1856: angular momentum, of the system, independent of magnetic braking 
1857: and magnetogyrosphere formation. 
1858: 
1859: The formation of a magnetogyrosphere in place of a rotationally
1860: supported disk can make the wind-stripping of angular momentum 
1861: potentially more efficient; it should be easier to blow away the 
1862: more spherical, lower density, magnetogyrosphere than a disk. 
1863: The dispersal 
1864: of the magnetogyrospheric material allows the wind to remove the 
1865: angular momentum associated with not only the portion of the 
1866: core material that does not collapse into the central object but 
1867: also the collapsed portion; the latter is parked mostly in the 
1868: magnetogyrosphere. Wind-stripping is a powerful mechanism of 
1869: angular momentum removal that deserves close attention. 
1870: 
1871: The total angular momentum retained in a stellar system may be 
1872: determined 
1873: to a large extent by the amount of the high specific angular 
1874: momentum magnetospheric and/or core material that survives the 
1875: wind-stripping. The left-over material may recollapse towards 
1876: the center, perhaps forming a rotationally supported disk, which 
1877: could persist once the massive, slowly rotating envelope that  
1878: strongly brakes the disk rotation at earlier times has been 
1879: cleared away. It takes only a small fraction of the mass and 
1880: angular momentum of a typical dense core to form a 
1881: minimum solar nebula. Removal of the braking material --- the 
1882: slowly rotating envelope --- is one way for the all-important, 
1883: rotationally supported disk to appear in the collapse of a 
1884: magnetized core. Another possibility, to 
1885: be discussed briefly in \S~\ref{future} below, is through nonideal
1886: MHD effects, such as ambipolar diffusion. 
1887: 
1888: 
1889: \subsection{Low Plasma-$\beta$ Disks Supported by Rotation} 
1890: \label{lowbetadisk}
1891: 
1892: We find persistent, rotationally-supported, equilibrium disks 
1893: only in the three 
1894: weakest field cases, with $\lambda=400$, $200$ and $133$. These 
1895: mass-to-flux ratios correspond to initial values of equatorial 
1896: plasma-$\beta$ of $7.68\times 10^4$, $1.92 \times 10^4$ and $8.53 
1897: \times 10^3$, respectively. One may not expect such weak fields 
1898: to have any significant effect on the dynamics. However, they 
1899: are strongly amplified, through both infalling motion and 
1900: differential rotation. The amplification is so efficient that, 
1901: even in the weakest field case of $\lambda=400$, the magnetic 
1902: pressure dominates the thermal pressure everywhere in the disk 
1903: (except near the equatorial plane, where the 
1904: toroidal field is forced to zero by symmetry), especially in the 
1905: inner part, where the orbital period is shorter. The enormous 
1906: difference in the plasma-$\beta$ between the disk region and the 
1907: collapsing envelope is illustrated in Fig.~\ref{beta}, where we 
1908: plot $\beta$ as a function of radius along a direction $3^\circ$  
1909: away from the midplane. The distribution is similar along other 
1910: angles, including those far from the midplane, where the amplified 
1911: field creates a vertically extended, magnetic tower. The base of 
1912: the tower is a magnetically dominated ``corona'' that replaces 
1913: the classical ``accretion shock'' that bounds the disk in the 
1914: absence of magnetic fields \citep[e.g.,][]{1990ApJ...355..651B}. It 
1915: can be viewed as a weak version of the magnetogyrosphere; the 
1916: bulk of the mass and angular momentum of the collapsed material 
1917: reside in the disk rather than the corona.  
1918: 
1919: The disk magnetic field cannot be amplified indefinitely. 
1920: \citet{2007MNRAS.375.1070B} 
1921: conjectured that the limiting field strength is 
1922: such that the Alfv\'en speed $V_A$ roughly equals the geometric 
1923: mean of the Keplerian speed $V_K$ and the sound speed $a$ of the 
1924: gas. The conjecture is motivated by the 
1925: analysis of \citet{2005ApJ...628..879P}, who showed that the growth 
1926: rate of the magnetorotational instability vanishes when 
1927: $V_A=(2 V_K a)^{1/2}$. It is qualitatively consistent with our 
1928: finding that the plasma-$\beta$ tends to be lower in the inner 
1929: part of the disk where the rotation speed is higher. For a 
1930: quantitative comparison, we plot in Fig.~\ref{beta} the predicted 
1931: plasma $\beta$ distribution based on 
1932: $\beta= 2 a^2/V_A^2=a/V_K$, where $V_K$ is the equilibrium rotation 
1933: speed shown in Fig.~\ref{Disk_speeds}. The rough agreement between 
1934: the predicted and actual distributions lends some support to the 
1935: conjecture, even though the toroidal field in our simulation is 
1936: generated mostly by simple twisting of a pre-existing poloidal 
1937: field, rather than MRI. 
1938: 
1939: The condition for a rotationally supported disk {\it not} to be 
1940: affected significantly by magnetic braking in the 
1941: ideal MHD limit, $\lambda \gtrsim 100$, may be difficult to satisfy 
1942: in the Galaxy at the present time. For a typical low-mass core, 
1943: the field strength must be less than $\sim 1~\mu$G on the 0.05pc
1944: scale. The required strength is 
1945: well below the median value inferred by
1946: \citet{2005ApJ...624..773H} for the cold neutral structures of HI gas. 
1947: A potential exception is the massive cores of molecular clouds, 
1948: where column densities as high as 1~g~cm$^{-2}$ are often inferred 
1949: \citep{1997ApJ...476..730P}. If their field strength is about $15$~$\mu$G 
1950: or less, the mass-to-flux ratio $\lambda$ would be of order 100 
1951: or more. Such a weak field (relative to mass) should not significantly 
1952: disrupt disk formation, particularly around 
1953: massive stars that may form in such cores \citep{2002Natur.416...59M}, 
1954: although it can still be amplified to the extent that its pressure 
1955: dominates the thermal pressure inside the disk, at least in 2D. 
1956: If the cores are more strongly
1957: magnetized, as is the case for the nearest region of ongoing 
1958: massive star formation OMC1 \citep{1999ApJ...520..706C,2004ApJ...616L.111H}, 
1959: the field may suppress the formation of a rotationally supported
1960: disk altogether, as long as it remains well coupled to the 
1961: matter. In this case, most of the mass would be accreted through 
1962: a still rotating but rapidly infalling pseudodisk that is 
1963: strongly braked by magnetic fields. Quantifying this possibility 
1964: for massive star formation fully would require 3D simulations 
1965: that include both radiative transfer \citep*[e.g.,][]{2007ApJ...656..959K} 
1966: and magnetic fields, as well as a treatment of ionization 
1967: and magnetic coupling.
1968: 
1969: \subsection{Future Directions} 
1970: \label{future}
1971: 
1972: We have performed calculations of disk formation in rotating dense 
1973: cores magnetized to different degrees under a number of simplifying
1974: assumptions. These include axisymmetry, ideal MHD, and ignoring 
1975: protostellar winds. The calculations provide a starting point 
1976: for future refinements. 
1977: 
1978: The most straightforward refinement is to increase spatial resolution. 
1979: Higher resolution 2D (axisymmetric) simulations are desirable in the 
1980: weakest field cases, where rotationally-supported, equilibrium 
1981: disks are formed. 
1982: Such disks can evolve through the twisting of a pre-existing, poloidal 
1983: field dragged into the disk from large distances by collapsing
1984: material. The disk evolution can be aided, perhaps even dominated, by 
1985: MRI. To adequately quantify the role of MRI in driving disk evolution, 
1986: one need to resolve the fastest growing MRI mode
1987: \citep{2001ApJ...548..348H,2001MNRAS.322..461S}, whose wavelength 
1988: decreases as the 
1989: field weakens. Another straightforward refinement is to relax the mirror 
1990: symmetry imposed at the equator. We have explored a few cases without 
1991: the equatorial symmetry, and found qualitatively similar results. 
1992: They will be discussed in the future along with 3D simulations. 
1993: 
1994: High resolution, global 3D simulations that include both the core 
1995: and the disk will be a natural, but challenging, extension of our 
1996: weak field calculations. They are needed for determining the 
1997: importance of gravitational torques in redistributing the disk 
1998: angular momentum relative to hydromagnetic stresses \citep[e.g.,][]{2004ApJ...616..364F}, 
1999: including those arising from possible MRI associated 
2000: with the dominant, toroidal component of the disk magnetic field 
2001: that cannot be treated in 2D. 3D calculations are also needed for 
2002: treating potential (sub-)stellar companion formation through 
2003: fragmentation \citep[e.g.,][]{2006ApJ...641..949B,2007ApJ_MIM_sub,2007A&A_HT_sub}.
2004: 
2005: Equally challenging will be detailed treatment of nonideal MHD 
2006: effects. Such effects have been investigated over the years, 
2007: most systematically by Nakano and collaborators 
2008: \citep*[e.g.,][]{2002ApJ...573..199N}, but have yet to be incorporated 
2009: into multi-dimensional {\it protostellar} collapse calculations. 
2010: Particularly vulnerable to non-ideal effects are two prominent
2011: features of our relatively 
2012: strong field cases: the split magnetic monopole near the center 
2013: and the associated episodic mass accretion. Sharp kinks in field 
2014: lines (across the midplane), such as those shown in the left 
2015: panel of  Fig.~\ref{burst}, are expected to be smoothed out 
2016: by nonideal MHD effects \citep[see][for an example]{2006ApJ...647..382S}. 
2017: However, there will still be a 
2018: tendency for the magnetic flux to accumulate at small radii even 
2019: when realistic nonideal effects are present. Indeed, episodic 
2020: mass accretion due to repeated flux 
2021: accumulation and escape can be seen in the nonideal MHD calculations of 
2022: \citet{2005ApJ...618..783T} under the thin-disk 
2023: approximation.
2024: 
2025: Whether the deviation from ideal MHD in a collapsing core is strong 
2026: enough to enable disk formation is uncertain. \citet{2002ApJ...580..987K} 
2027: investigated the effects of ambipolar diffusion on magnetic
2028: braking and disk formation under the self-similar assumption and 
2029: thin-disk approximation. They showed that rotationally supported 
2030: disks can indeed form even in strongly magnetized cores, although
2031: this is by no means guaranteed. A particular complication is the 
2032: hydromagnetic accretion shock of C-type produced by the magnetic 
2033: flux left behind by the mass that has gone into the central object 
2034: \citep{1996ApJ...464..373L,1998ApJ...504..257C}. The increase in field
2035: strength and decrease in infall speed inside the shock and in the
2036: postshock region tend to enhance the magnetic braking 
2037: \citep[in a manner akin to the quasistatic phase of subcritical cloud evolution,][]{1994ApJ...432..720B}, 
2038: whereas the slippage of field lines relative 
2039: to neutral matter in the azimuthal direction tends to decrease 
2040: it. Whether there is enough angular momentum left in the accreting 
2041: material after passing through the C-shock to form an appreciable 
2042: disk is uncertain, as demonstrated by \citet{2002ApJ...580..987K} 
2043: semi-analytically in 1D. Numerical study of this problem in 2D is 
2044: underway (Mellon \& Li, in preparation). 
2045: 
2046: Theory of disk formation will be incomplete without a treatment of 
2047: protostellar wind. It is likely that the wind removes the bulk of 
2048: not only the mass, but also the angular momentum, of a (low-mass) 
2049: star forming core, at least for regions like the Taurus clouds, where
2050: the typical mass of a core is well above that of a YSO. As 
2051: discussed in \ref{dispersal}, there is the possibility of late-time 
2052: disk formation after the wind has stripped away most of the (slowly 
2053: rotating) envelope that impedes disk formation through magnetic 
2054: braking. The role of protostellar wind in the evolution of angular
2055: momentum in star formation in general and disk formation in particular
2056: remains to be quantified. 
2057: The various refinements outlined above are not mutually 
2058: exclusive. However, it will be a daunting, if not impossible, task to 
2059: implement all of them at the same time. Fortunately the calculations 
2060: can be guided by high resolution observations with increasingly 
2061: powerful interferometers, especially ALMA. 
2062: 
2063: \section{Conclusions}
2064: \label{conclusion}
2065: 
2066: We have performed a set of axisymmetric calculations of protostellar
2067: collapse phase of star formation in rotating cores magnetized to
2068: different degrees, expanding on the work of ALS03. The goal is to 
2069: determine how magnetic braking affects disk formation in the ideal 
2070: MHD limit. The main conclusions are: 
2071: 
2072: 1. Protostellar disks can be disrupted by weak magnetic fields 
2073: corresponding to mass-to-flux ratio $\lambda \gg 1$. This is made 
2074: possible by the tight wrapping of the disk field lines, which increases 
2075: the rate of magnetic braking, and the long braking time compared with 
2076: the disk rotation period. In our illustrative calculations, the disk 
2077: disruption begins around $\lambda~ \sim 100$. 
2078: 
2079: 
2080: 2. Magnetic braking disrupts the rotationally supported disk by
2081: creating regions of rapid, supersonic collapse in the disk. These
2082: regions are separated by one or more centrifugal barriers, where 
2083: the rapid infall is halt temporarily and the magnetic braking 
2084: is locally enhanced. The number of centrifugal barriers decreases 
2085: as the core becomes more strongly magnetized. Multiple centrifugal 
2086: barriers, if observed, would clinch the case for disk disruption 
2087: by magnetic braking.
2088: 
2089: 3. Surrounding the strongly braked disk of spatially alternating 
2090: collapse and stagnation is a vertically extended 
2091: structure supported by a combination of (toroidal) magnetic field 
2092: and rotation --- a ``magnetogyrosphere'' --- where the bulk of the 
2093: angular momentum of the collapsed material is stored. The 
2094: magnetogyrosphere may be detectable interferometrically, although 
2095: the detection may be complicated by interaction with protostellar 
2096: winds. We suggest that the wind may play a key role in determining 
2097: not only the mass but also the angular momentum of a stellar system.
2098: 
2099: 4. Even in the extremely weak field cases of $\lambda~ \gtrsim~ 100$,
2100: where a rotationally-supported, equilibrium disk is formed in 
2101: the ideal 
2102: MHD limit, the magnetic field can still be dynamically important. 
2103: It can be amplified to the extent that its pressure dominates the 
2104: thermal pressure both in the disk and its surrounding region, at 
2105: least in 2D. The field may affect the fragmentation properties 
2106: of the disk, and seed the MRI, although high resolution 3D 
2107: simulations are required to fully quantify its dynamical effects. 
2108: 
2109: 5. In relatively strongly magnetized cores of $\lambda~ \lesssim~ 10$, the
2110: disk formation is completely suppressed, and the bulk of the angular
2111: momentum of the collapsed material is ejected out of the system 
2112: via a low-speed, magnetic braking-driven wind, as found previously. 
2113: A new feature is that the mass accretion on to the central object 
2114: is highly episodic, because of magnetic reconnection. How this and 
2115: other behaviors are modified by nonideal MHD effects, particularly
2116: ambipolar diffusion, remains to be quantified. 
2117: 
2118: In summary, our calculations have shown that in the ideal MHD limit 
2119: the magnetic braking is strong enough to prevent the formation 
2120: of a persistent, rotationally supported, protostellar disk in all 
2121: but the unrealistically weak field cases. Since disks are observed 
2122: around many, if not all, young stellar objects, this result presents 
2123: an interesting conundrum that must be resolved. For such a disk to 
2124: form in dense cores magnetized to a realistic level, magnetic 
2125: braking must be weakened one way or another, perhaps through a 
2126: combination of nonideal MHD effects and protostellar winds. 
2127: Quantifying these effects should be a major goal of future research 
2128: in this area. 
2129: 
2130: \acknowledgements
2131: This work was supported in part by NASA (NNG05GJ49G) and NSF (AST-0307368)
2132: grants. It grew out of an earlier collaboration that involved Tony 
2133: Allen and Frank Shu. We thank Tony Allen for technical assistance at 
2134: the beginning of the project, Frank Shu, John Hawley, and Philippe
2135: Andre for helpful 
2136: discussion, and Sebastien Fromang for providing the gravity solver 
2137: used in our work. 
2138: 
2139: \begin{thebibliography}{}
2140: \bibitem[Allen et al.(2003)Allen, Li, \& Shu]{2003ApJ...599..363A} Allen, A., Li, Z.-Y., \& Shu, F.~H.\ 2003, \apj, 599, 363 
2141: \bibitem[Andr{\'e} et al.(1999)Andr{\'e}, Motte, \& Bacmann]{1999ApJ...513L..57A} Andr{\'e}, P., Motte, \ F., \& Bacmann, A.\ 1999, \apjl, 513, L57 
2142: \bibitem[Bachiller \& Tafalla(1999)]{1999osps.conf..227B} Bachiller, R., \& Tafalla, M.\ 1999, in NATO ASIC Proc.~540, The Origin of Stars and Planetary Systems, ed Lada, C.~J. and Kylafis, N.~D. (Norwell, MA: Kluwer Acad. Pub.), 227 
2143: \bibitem[Balbus \& Hawley(1998)]{1998RvMP...70....1B} Balbus, S.~A., \& Hawley, J.~F.\ 1998, Reviews of Modern Physics, 70, 1 
2144: \bibitem[Banerjee \& Pudritz(2006)]{2006ApJ...641..949B} Banerjee, R., \& Pudritz, R.~E.\ 2006, \apj, 641, 949 
2145: \bibitem[Basu \& Mouschovias(1994)]{1994ApJ...432..720B} Basu, S., \& Mouschovias, T.~C.\ 1994, \apj, 432, 720 
2146: \bibitem[Beckwith \& Sargent(1993)]{1993prpl.conf..521B} Beckwith, S.~V.~W., \& Sargent, A.~I.\ 1993, in Protostars and Planets III, ed. E.~H. Levy \& J. Lunine (Tucson:Univ of Arizona Press), 521 
2147: \bibitem[Begelman \& Pringle(2007)]{2007MNRAS.375.1070B} Begelman, M.~C., \& Pringle, J.~E.\ 2007, \mnras, 375, 1070 
2148: \bibitem[Belloche \& Andr{\'e}(2004)]{2004A&A...419L..35B} Belloche, A., \& Andr{\'e}, P.\ 2004, \aap, 419, L35 
2149: \bibitem[Belloche et al.(2002)]{2002A&A...393..927B} Belloche, A., \ Andr{\'e}, P., Despois, D., \& Blinder, S.\ 2002, \aap, 393, 927 
2150: \bibitem[Bodenheimer(1995)]{1995ARA&A..33..199B} Bodenheimer, P.\ 1995, \araa, 33, 199 
2151: \bibitem[Bodenheimer et al.(1990)]{1990ApJ...355..651B} Bodenheimer, P., Yorke, H.~W., Rozyczka, M., \& Tohline, J.~E.\ 1990, \apj, 355, 651 
2152: \bibitem[Boss(1998)]{1998ASPC..148..314B} Boss, A.~P.\ 1998, in ASP Conf. Ser. 148, Origins, ed. C.~E. Woodward, J.~M. Shull, \& H.~A. Thronson, Jr. (San Francisco: ASP), 314 
2153: \bibitem[Ciolek \& K{\"o}nigl(1998)]{1998ApJ...504..257C} Ciolek, G.~E., \& K{\"o}nigl, A.\ 1998, \apj, 504, 257 
2154: \bibitem[Cohl \& Tohline(1999)]{1999ApJ...527...86C} Cohl, H.~S., \& Tohline, J.~E.\ 1999, \apj, 527, 86 
2155: \bibitem[Crapsi et al.(2007)]{2007A&A...470..221C} Crapsi, A., Caselli, P., Walmsley, M.~C., \& Tafalla, M.\ 2007, \aap, 470, 221 
2156: \bibitem[Crutcher(1999)]{1999ApJ...520..706C} Crutcher, R.~M.\ 1999, \apj, 520, 706 
2157: \bibitem[Crutcher \& Troland(2000)]{2000ApJ...537L.139C} Crutcher, R.~M., \& Troland, T.~H.\ 2000, \apjl, 537, L139 
2158: \bibitem[Crutcher \& Troland(2007)]{2007IAUS..237..141C} Crutcher, R.~M., \& Troland, T.~H.\ 2007, in IAU Symp. 237, Triggered Star Formation in a Turbulent ISM, ed. B.~G. Elmegreen \& J. Palous (Cambridge: Cambridge Univ. Press), 141
2159: \bibitem[Dunham et al.(2006)]{2006ApJ...651..945D} Dunham, M.~M., et al.\ 2006, \apj, 651, 945 
2160: \bibitem[Evans \& Hawley(1988)]{1988ApJ...332..659E} Evans, C.~R., \& \ Hawley, J.~F.\ 1988, \apj, 332, 659 
2161: \bibitem[Fiedler \& Mouschovias(1993)]{1993ApJ...415..680F} Fiedler, R.~A., \& Mouschovias, T.~C.\ 1993, \apj, 415, 680 
2162: \bibitem[Fromang et al.(2004)]{2004ApJ...616..364F} Fromang, S., Balbus, S.~A., Terquem, C., \& De Villiers, J.-P.\ 2004, \apj, 616, 364 
2163: \bibitem[Fromang et al.(2006)Fromang, Hennebelle, \& Teyssier]{2006A&A...457..371F} Fromang, S., Hennebelle, P., \& Teyssier, R.\ 2006, \aap, 457, 371 
2164: \bibitem[Galli et al.(2006)]{2006ApJ...647..374G} Galli, D., Lizano, S., Shu, F.~H., \& Allen, A.\ 2006, \apj, 647, 374 
2165: \bibitem[Galli \& Shu(1993)]{1993ApJ...417..243G} Galli, D., \& Shu, F.~H.\ 1993, \apj, 417, 243 
2166: \bibitem[Girart et al.(2006)Girart, Rao, \& Marrone]{2006Sci...313..812G} Girart, J.~M., Rao, R., \& Marrone, D.~P.\ 2006, Science, 313, 812 
2167: \bibitem[Goodman et al.(1993)]{1993ApJ...406..528G} Goodman, A.~A., Benson, P.~J., Fuller, G.~A., \& Myers, P.~C.\ 1993, \apj, 406, 528 
2168: \bibitem[Gueth \& Guilloteau(1999)]{1999A&A...343..571G} Gueth, F., \& Guilloteau, S.\ 1999, \aap, 343, 571 
2169: \bibitem[Hawley \& Krolik(2001)]{2001ApJ...548..348H} Hawley, J.~F., \& Krolik, J.~H.\ 2001, \apj, 548, 348 
2170: \bibitem[Heiles \& Troland(2005)]{2005ApJ...624..773H} Heiles, C., \& Troland, T.~H.\ 2005, \apj, 624, 773 
2171: \bibitem[Hennebell \& Fromang(2007)]{2007A&A_HF_sub} Hennebell, P., \& Fromang, S.\ 2007, \aap, submitted
2172: \bibitem[Hennebell \& Teyssier(2007)]{2007A&A_HT_sub} Hennebell, P., \& Teyssier, R.\ 2007, \aap, submitted
2173: \bibitem[Houde(2004)]{2004ApJ...616L.111H} Houde, M.\ 2004, \apjl, 616, L111 
2174: \bibitem[Kenyon \& Hartmann(1995)]{1995ApJS..101..117K} Kenyon, S.~J., \& Hartmann, L.\ 1995, \apjs, 101, 117 
2175: \bibitem[K{\"o}nigl(1991)]{1991ApJ...370L..39K} K{\"o}nigl, A.\ 1991, \apjl, 370, L39 
2176: \bibitem[K{\"o}nigl \& Pudritz(2000)]{2000prpl.conf..759K} K{\"o}nigl, A., \& Pudritz, R.~E.\ 2000, in Protostars and Planets IV, ed. V. Manning, A. Boss, \& S. Russell (Arizona: Univ of Arizona Press), 759 
2177: \bibitem[Krasnopolsky \& K{\"o}nigl(2002)]{2002ApJ...580..987K} Krasnopolsky, R., K{\"o}nigl, A.\ 2002, \apj, 580, 987 
2178: \bibitem[Krumholz et al.(2007)Krumholz, Klein, \& McKee]{2007ApJ...656..959K} Krumholz, M.~R., Klein, R.~I., \& McKee, C.~F.\ 2007, \apj, 656, 959 
2179: \bibitem[Lee et al.(2005)Lee, Ho, \& White]{2005ApJ...619..948L} Lee, C.-F., Ho, P.~T.~P., \& White, S.~M.\ 2005, \apj, 619, 948 
2180: \bibitem[Li \& McKee(1996)]{1996ApJ...464..373L} Li, Z.-Y., \& McKee, C.~F.\ 1996, \apj, 464, 373 
2181: \bibitem[Li \& Shu(1996)]{1996ApJ...472..211L} Li, Z.-Y., \& Shu, F.~H.\ 1996, \apj, 472, 211 
2182: \bibitem[Li \& Shu(1997)]{1997ApJ...475..237L} Li, Z.-Y., \& Shu, F.~H.\ 1997, \apj, 475, 237 
2183: \bibitem[Lizano \& Shu(1989)]{1989ApJ...342..834L} Lizano, S., \& Shu, F.~H.\ 1989, \apj, 342, 834 
2184: \bibitem[Lynden-Bell(2003)]{2003MNRAS.341.1360L} Lynden-Bell, D.\ 2003, \mnras, 341, 1360 
2185: \bibitem[Machida et al.(2007)Machida, Inutsuka, \& Matsumoto]{2007ApJ_MIM_sub} Machida, M.~N., Inutsuka, S.-i., \& Matsumoto, T.\ 2007, \apj, submitted 
2186: \bibitem[Matzner \& McKee(2000)]{2000ApJ...545..364M} Matzner, C.~D., \& McKee, C.~F.\ 2000, \apj, 545, 364 
2187: \bibitem[McKee \& Tan(2002)]{2002Natur.416...59M} McKee, C.~F., \& Tan, J.~C.\ 2002, \nat, 416, 59 
2188: \bibitem[McKee et al.(1993)]{1993prpl.conf..327M} McKee, C.~F., Zweibel, E.~G., Goodman, A.~A., \& Heiles, C.\ 1993, in Protostars and Planets III, ed. E.~H. Levy \& J. Lunine (Tucson:Univ of Arizona Press), 327 
2189: \bibitem[Mestel \& Paris(1979)]{1979MNRAS.187..337M} Mestel, L., \& Paris, R.~B.\ 1979, \mnras, 187, 337 
2190: \bibitem[Mouschovias \& Ciolek(1999)]{1999osps.conf..305M} Mouschovias, T.~C., \& Ciolek, G.~E.\ 1999,in NATO ASIC Proc.~540, The Origin of Stars and Planetary Systems, ed Lada, C.~J. and Kylafis, N.~D. (Norwell, MA: Kluwer Acad. Pub.), 305 
2191: \bibitem[Mouschovias \& Paleologou(1979)]{1979ApJ...230..204M} Mouschovias, T.~C., \& Paleologou, E.~V.\ 1979, \apj, 230, 204 
2192: \bibitem[Myers(1995)]{1995mcsf.conf...47M} Myers, P. C. 1995, in Molecular Clouds and Star Formation, ed. C. Yuan \& J. You (Singapore: World Scientific), 47
2193: \bibitem[Nakano(1989)]{1989MNRAS.241..495N} Nakano, T.\ 1989, \mnras, 241, 495 
2194: \bibitem[Nakano et al.(2002)Nakano, Nishi, \& Umebayashi]{2002ApJ...573..199N} Nakano, T., Nishi, R., \& Umebayashi, T.\ 2002, \apj, 573, 199 
2195: \bibitem[Onishi et al.(2002)]{2002ApJ...575..950O} Onishi, T., Mizuno, A., Kawamura, A., Tachihara, K., \& Fukui, Y.\ 2002, \apj, 575, 950 
2196: \bibitem[Pessah \& Psaltis(2005)]{2005ApJ...628..879P} Pessah, M.~E., \& Psaltis, D.\ 2005, \apj, 628, 879 
2197: \bibitem[Plume et al.(1997)]{1997ApJ...476..730P} Plume, R., Jaffe, D.~T., Evans, N.~J., II, Martin-Pintado, J., \& Gomez-Gonzalez, J.\ 1997, \apj, 476, 730 
2198: \bibitem[Shu(1977)]{1977ApJ...214..488S} Shu, F.~H.\ 1977, \apj, 214, 488 
2199: \bibitem[Shu et al.(2006)]{2006ApJ...647..382S} Shu, F.~H., Galli, D., Lizano, S., \& Cai, M.\ 2006, \apj, 647, 382 
2200: \bibitem[Shu et al.(1987)Shu, Adams, \& Lizano]{1987ARA&A..25...23S} Shu, F.~H., Adams, F.~C., \& Lizano, S.\ 1987, \araa, 25, 23 
2201: \bibitem[Shu et al.(2000)]{2000prpl.conf..789S} Shu, F.~H., Najita, J.~R., Shang, H., \& Li, Z.-Y.\ 2000, in Protostars and Planets IV, ed. V. Manning, A. Boss, \& S. Russell (Arizona: Univ of Arizona Press), 789 
2202: \bibitem[Stone \& Norman(1992a)]{1992ApJS...80..753S} Stone, J.~M., \& Norman, M.~L.\ 1992, \apjs, 80, 753 
2203: \bibitem[Stone \& Norman(1992b)]{1992ApJS...80..791S} Stone, J.~M., \& Norman, M.~L.\ 1992, \apjs, 80, 791 
2204: \bibitem[Stone \& Pringle(2001)]{2001MNRAS.322..461S} Stone, J.~M., \& Pringle, J.~E.\ 2001, \mnras, 322, 461 
2205: \bibitem[Tafalla et al.(1998)]{1998ApJ...504..900T} Tafalla, M., Mardones, D., Myers, P.~C., Caselli, P., Bachiller, R., \& Benson, P.~J.\ 1998, \apj, 504, 900 
2206: \bibitem[Tafalla et al.(2002)]{2002ApJ...569..815T} Tafalla, M., Myers, P.~C., Caselli, P., Walmsley, C.~M., \& Comito, C.\ 2002, \apj, 569, 815 
2207: \bibitem[Tassis \& Mouschovias(2005)]{2005ApJ...618..783T} Tassis, K., \& Mouschovias, T.~C.\ 2005, \apj, 618, 783 
2208: \bibitem[Terebey et al.(1984)Terebey, Shu, \& Cassen]{1984ApJ...286..529T} Terebey, S., Shu, F.~H., \& Cassen, P.\ 1984, \apj, 286, 529 
2209: \bibitem[Tomisaka(1998)]{1998ApJ...502L.163T} Tomisaka, K.\ 1998, \apjl, 502, L163 
2210: \bibitem[Ziegler(2005)]{2005A&A...435..385Z} Ziegler, U.\ 2005, \aap, 435, 385 
2211: \end{thebibliography}
2212: 
2213: 
2214: 
2215: \begin{figure}
2216: \epsscale{0.75}
2217: \plotone{f1.eps}
2218: \caption{Rotating, magnetized singular isothermal toroid (SIT) as the 
2219: initial configuration for collapse calculation. Plotted are the 
2220: isodensity contours (dashed lines, with two contours per decade 
2221: and $10^{-19}$~g~cm$^{-3}$ labeled) and the lines 
2222: of magnetic field (solid). 
2223: } 
2224: \label{initial}
2225: \end{figure}
2226: 
2227: 
2228: \begin{figure}
2229: \epsscale{1.0}
2230: \plotone{f2.eps}
2231: \caption{Snapshot of the standard collapse solution at a representative
2232: time. Shown in color is the distribution of dimensionless density in
2233: the meridian plane, with the lines of 
2234: magnetic field and velocity vectors superposed. 
2235: The four dynamically distinct regions are labeled, corresponding to 
2236: the collapsing envelope (Region I), equatorial pseudodisk (II), magnetic 
2237: bubble (III) and polar funnel (IV). See the text for discussion.} 
2238: \label{snap}
2239: \end{figure}
2240: 
2241: \begin{figure}
2242: \epsscale{1.0}
2243: \plotone{f3.eps}
2244: \caption{Map of the magnetic twist. Superposed on the map are the 
2245: lines of magnetic field (black) and 
2246: velocity vectors (arrows). The toroidally dominated magnetic
2247: bubble (enclosed within the blue contour labeled with ``-1'') 
2248: is surrounded by the polar funnel flow to the left, collapsing
2249: envelope to the right, and equatorial pseudodisk from below. } 
2250: \label{twist}
2251: \end{figure}
2252: 
2253: \begin{figure}
2254: \epsscale{0.75}
2255: \plotone{f4.eps}
2256: \caption{Velocity distribution of the standard model on the equator. 
2257: Plotted are the radial (dashed), rotational (solid) speeds of the 
2258: fluid, and the rotation speed needed for support against gravity 
2259: (dot-dashed).} 
2260: \label{vrvphi_eq}
2261: \end{figure}
2262: 
2263: \begin{figure}
2264: \epsscale{0.75}
2265: \plotone{f5.eps}
2266: \caption{Evolution of the central point mass for the standard model. 
2267: The nearly linear increase of the mass with time indicates that the 
2268: collapse solution is approximately self-similar. } 
2269: \label{pointmass}
2270: \end{figure}
2271: 
2272: 
2273: \begin{figure}
2274: \epsscale{0.75}
2275: \plotone{f6.eps}
2276: \caption{Time averaged distributions of the dimensionless mass $m(x)$
2277:   enclosed within an expanding sphere of fixed dimensionless radius 
2278: $x$, computed in two different ways, using  
2279: eqs~(\ref{mass_dimensionless}) (solid line) and
2280: (\ref{masscons_dimensionless}) (dotted). The averaging is 
2281: done between $3.0\times 10^{11}$ and $7.5\times 10^{11}$~sec. The 
2282: agreement between the two curves is an indication that the collapse 
2283: solution is approximately self-similar. Also 
2284: plotted are the two terms on the right hand side of
2285: eq.~(\ref{masscons_dimensionless}), showing the mass change   
2286: due to volume expansion (dashed line) and mass crossing the 
2287: sphere (dot-dashed).
2288: }
2289: \label{M_av}
2290: \end{figure}
2291: 
2292: \begin{figure}
2293: \epsscale{0.75}
2294: \plotone{f7.eps}
2295: \caption{Time averaged distributions of the dimensionless angular 
2296: momentum $l(x)$ enclosed within an expanding sphere of fixed dimensionless 
2297: radius $x$, computed in two different ways, using 
2298: eqs.~(\ref{angmom_dimensionless}) (solid line) and
2299: (\ref{angmom_cons_dimensionless}) (dotted). The averaging is done 
2300: between $3.0\times 10^{11}$ and $7.5\times 10^{11}$~sec. Also 
2301: plotted are the three terms on the right hand side of 
2302: eq.~(\ref{angmom_cons_dimensionless}), showing the angular 
2303: momentum change due to volume expansion (dashed), mass crossing the
2304: sphere (dot-dashed), and magnetic braking (dot-dot-dashed). The 
2305: sum of the last two terms is plotted as the long dashed line, 
2306: showing that the angular momentum advected into a sphere of
2307: radius $x~\lesssim ~0.4$ by fluid motion is almost completely removed 
2308: by magnetic braking. 
2309: }
2310: \label{L_av}
2311: \end{figure}
2312: 
2313: \begin{figure}
2314: \epsscale{0.75}
2315: \plotone{f8.eps}
2316: \caption{Time averaged angular distributions of the total pressure and its
2317: components at a representative radius $x=0.3$ for the standard
2318: case. The total pressure (solid line) is nearly 
2319: constant in $\theta$, but is dominated by different components in different 
2320: regions. The thermal pressure (dot-dashed) dominates in the
2321: pseudodisk, the toroidal 
2322: magnetic pressure (dashed) dominates in the bubble, and the poloidal magnetic 
2323: pressure (dotted) dominates in the polar region.  }
2324: \label{pressures}
2325: \end{figure}
2326: 
2327: \begin{figure}
2328: \epsscale{0.75}
2329: \plotone{f9.eps}
2330: \caption{Time evolution of dimensionless point mass for relatively weak 
2331: magnetic field cases. The mass-to-flux ratio increases from the standard 
2332: model ($\lambda=13.3$) on the top to the non-magnetized case ($\lambda=\infty$) 
2333: on the bottom.}
2334: \label{Mstar_LowB}
2335: \end{figure}
2336: 
2337: \begin{figure}
2338: \epsscale{1.0}
2339: \plotone{f10.eps}
2340: \caption{A snapshot of the disk formed in the weakest field case 
2341: of $\lambda=400$. The density is plotted in color contours, with 
2342: the magnetic fields lines in black 
2343: and the velocity field as white arrows. The dimensionless 
2344: density of $3\times10^{3}$ (white contour) provides a fiducial  
2345: boundary for the disk.}
2346: \label{snap_Bpt01}
2347: \end{figure}
2348: 
2349: 
2350: \clearpage
2351: 
2352: \begin{figure}
2353: \epsscale{0.75}
2354: \plotone{f11.eps}
2355: \caption{Infall (dashed line) and rotation (solid)  speeds on the equator at the representative 
2356: time for the weakest field case of $\lambda=400$. The rotational support for the disk is 
2357: seen by comparing the rotation speed to the equilibrium value (dot-dashed), and by the 
2358: corresponding drop in the infall speed. }
2359: \label{Disk_speeds}
2360: \end{figure}
2361: 
2362: \begin{figure}
2363: \epsscale{0.75}
2364: \plotone{f12.eps}
2365: \caption{Radial velocity (dashed curve) and specific angular momentum 
2366: (solid) on the equator at a representative time $t=2\times
2367:   10^{12}$~sec for the relatively weak field case of $\lambda=80$. 
2368: They are in units of, respectively, the sound speed $a$ and $a^2 t$.  
2369: Note the abrupt drops in angular momentum at the stagnation points of the
2370: collapsing flow.}
2371: \label{vr_specmom_Bpt05}
2372: \end{figure}
2373: 
2374: \begin{figure} 
2375: \epsscale{1.0}
2376: \plotone{f13.eps}
2377: \caption{Map of density distribution in a transition case ($\lambda=80$), 
2378: at the representative time. Transient high density rings form instead of 
2379: a disk as infalling material piles up near centrifugal radii, then
2380: recollapses after losing a large fraction of its angular momentum 
2381: through magnetic braking. White contours are magnetic field lines, 
2382: and arrows are unit velocity vectors.}
2383: \label{disk_unitvec_Bpt05}
2384: \end{figure}
2385: 
2386: \begin{figure}
2387: \epsscale{0.75}
2388: \plotone{f14.eps}
2389: \caption{Time evolution of dimensionless point mass for relatively strong
2390: magnetic field cases. The mass-to-flux ratio decreases from top to
2391: bottom, the opposite of the trend seen in the low field cases.  Note 
2392: the step-like behavior, especially in the highest field case,
2393: indicative of episodic mass accretion.}
2394: \label{Mstar_HighB_noscaled}
2395: \end{figure}
2396: 
2397: \begin{figure}
2398: \epsscale{1.0}
2399: \plotone{f15.eps}
2400: \caption{Snapshot of the strongly magnetized ($\lambda=4$) core 
2401: representative of the state before a ``reconnection'' event. It is to 
2402: be compared with the standard case shown in Fig.~\ref{snap}. 
2403: Enclosed within the red contour is the region where the magnetic
2404: field is dominated by the toroidal component. The four dynamically 
2405: distinct regions are labelled here as in Fig.~\ref{snap}.}
2406: \label{snap_B1pt0}
2407: \end{figure}
2408: 
2409: \begin{figure}
2410: \epsscale{1.0}
2411: \plotone{f16.eps}
2412: \caption{Density distribution (color map), velocity field (white arrows) and
2413:   magnetic field lines (black lines) before (left panel) and after
2414:   (right) a reconnection event. 
2415: }
2416: \label{burst}
2417: \end{figure}
2418: 
2419: \begin{figure}
2420: \epsscale{0.75}
2421: \plotone{f17.eps}
2422: \caption{Dimensionless central point masses for collapsing cores of the 
2423: same initial magnetic field ($\lambda=13.3$) but different rotation 
2424: speeds. The accretion rate decreases as the rotation rate increases.  
2425: Also plotted for comparison is the point mass for SIS (dashed line).}
2426: \label{Mstar_diffrot}
2427: \end{figure}
2428: 
2429: \begin{figure} 
2430: \epsscale{1.0}
2431: \plotone{f18.eps}
2432: \caption{Snapshots of collapsing cores of the same initial magnetic field but
2433: different rotation speeds of $v_o=0$ (upper left panel), 0.125 (upper
2434: right), 0.25 (lower left), 0.5 (lower right). The magnetic bubble,
2435: enclosed  within the red contour, becomes thinner with less rotation, 
2436: disappearing entirely in the non-rotating case.}
2437: \label{multi_diffrot}
2438: \end{figure}
2439: 
2440: \begin{figure}
2441: \epsscale{1.0}
2442: \plotone{f19.eps}
2443: \caption{Column density distributions for the $\lambda=80$ case 
2444: at $t=2\times 10^{12}$~sec (left panel) and the $\lambda=13.3$ 
2445: case at $t=8.3\times 10^{11}$~sec (right) case viewed perpendicular 
2446: to the axis. In each case, a dense equatorial pseudodisk is 
2447: surrounded by an extended structure supported by a combination 
2448: of magnetic fields and rotation --- a magnetogyrosphere. 
2449: The length and column density are in units of $cm$ and 
2450: $g~cm^{-2}$ respectively.  
2451: } 
2452: \label{columndensity}
2453: \end{figure}
2454: 
2455: \begin{figure}
2456: \epsscale{0.75}
2457: \plotone{f20.eps}
2458: \caption{Plasma-$\beta$ parameter along a direction $3^\circ$ away from 
2459: the midplane at a representative time $t=2\times 10^{12}$~sec (solid 
2460: line). Plotted for comparison is a plausible limiting value estimated 
2461: analytically (dashed line). See text for discussion.} 
2462: \label{beta}
2463: \end{figure}
2464: 
2465: \end{document}
2466: 
2467: 
2468: