1: \documentclass[12pt,preprint]{aastex}
2: %\usepackage{emulateapj5}
3: \newcommand{\be}{\begin{equation}}
4: \newcommand{\ee}{\end{equation}}
5: \newcommand{\bea}{\begin{eqnarray}}
6: \newcommand{\eea}{\end{eqnarray}}
7:
8: \begin{document}
9:
10: \title{Grain Dynamics In Magnetized Interstellar Gas}
11:
12:
13: \author{A. Lazarian \& Huirong Yan}
14:
15:
16: \affil{Department of Astronomy, University of Wisconsin, 475 N. Charter
17: St., Madison, WI53706; lazarian, yan@astro.wisc.edu}
18:
19: \begin{abstract}
20: The interstellar medium is turbulent and this induces relative motions
21: of dust grains. We calculate relative velocities of charged grains
22: in a partially ionized magnetized gas. We account for anisotropy of
23: magnetohydrodynamic (MHD) turbulence, grain coupling with magnetic
24: field, and the turbulence cutoff arising from the ambipolar drag. We obtain
25: grain velocities for turbulence with parameters consistent with those
26: in HI and dark clouds. These velocities are smaller than those in
27: earlier papers, where MHD effects were disregarded. Finally, we consider
28: grain velocities arising from photoelectric emission, radiation pressure
29: and H\( _{2} \) thrust. These are still lower than relative velocities induced by turbulence. We conclude that turbulence should prevent these mechanisms from
30: segregating grains by size.
31: \end{abstract}
32: \keywords{ISM:dust, extinction---kinematics,dynamics---magnetic fields}
33: \section{Introduction}
34:
35: Dust is an important constituent of the interstellar medium (ISM).
36: It interferes with observations in the optical range, but provides
37: an insight to star-formation activity through far-infrared radiation.
38: It also enables molecular hydrogen formation and traces the magnetic
39: field via emission and extinction polarization. The basic properties
40: of dust (optical, alignment etc.) strongly depend on its size distribution.
41: The latter evolves as the result of grain collisions, whose frequency
42: and consequences depend on grain relative velocities.
43:
44: Various processes can affect the velocities of dust grains. Radiation,
45: ambipolar diffusion, and gravitational sedimentation all can bring
46: about a dispersion in grain velocities. It was speculated in de Oliveira-Costa
47: et al. (2000) that starlight radiation can produce the segregation of different
48: sized grains that is necessary to explain a poor correlation of the
49: microwave and \( 100\mu m \) signals of the foreground emission (Mukherjee
50: et al. 2001). If true it has big implications for the CMB foreground
51: studies. However, the efficiency of this segregation depends on grain
52: random velocities, which we study in this paper.
53:
54: Interstellar gas is turbulent (see Arons \& Max 1975). Turbulence
55: was invoked by a number of authors (see Kusaka et al. 1970, Volk et
56: al. 1980, Draine 1985, Ossenkopf 1993, Weidenschilling \& Ruzmaikina
57: 1994) to provide substantial relative motions of dust particles. However, they discussed hydrodynamic turbulence. It is clear
58: that this picture cannot be applicable to the magnetized
59: ISM as the magnetic fields substantially affect fluid dynamics.
60: Moreover dust grains are charged, and their interactions with magnetized
61: turbulence is very different from the hydrodynamic case. This unsatisfactory
62: situation motivates us to revisit the problem and calculate the grain
63: relative motions in magnetized ISM. In what follows, we use the model
64: of MHD turbulence by Goldreich and Sridhar (1995, henceforth GS95),
65: which is supported by recent numerical simulations (Cho \& Vishniac
66: 2000, Maron \& Goldreich 2001, Cho, Lazarian \& Vishniac 2002a, henceforth
67: CLV02). We apply our results to the cold neutral medium (CNM) and a
68: dark cloud to estimate the efficiency of coagulation, shattering and
69: segregation of grains.
70:
71:
72: \section{MHD Turbulence and Grain Motion}
73:
74: Unlike hydrodynamic turbulence, MHD turbulence is anisotropic, with
75: eddies elongated along the magnetic field. This happens because it
76: is easier to mix the magnetic field lines perpendicular to their direction rather than to bend them. The energy of
77: eddies drops with the decrease of eddy size (e.g. \( v_{l}\sim l^{1/3} \)
78: for the Kolmogorov turbulence) and it becomes more difficult for smaller
79: eddies to bend the magnetic field lines. Therefore the eddies get
80: more and more anisotropic as their sizes decrease. As eddies mix the
81: magnetic field lines at the rate \( k_{\bot }v_{k} \), where \( k_{\bot } \)is
82: a wavenumber measured in the direction perpendicular to the local
83: magnetic field and \( v_{k} \) is the mixing velocity at this scale,
84: the magnetic perturbations propagate along the magnetic field lines
85: at the rate \( k_{\parallel }V_{A} \) ,where \( k_{\parallel } \)
86: is the parallel wavenumber and \( V_{A} \) is the Alfven velocity.
87: The corner stone of the GS95 model is a critical balance between those
88: rates, i.e., \( k_{\bot }v_{k} \)\( \sim \) \( k_{\parallel }V_{A} \),
89: which may be also viewed as coupling of eddies and wave-like motions.
90: Mixing motions perpendicular to the magnetic field lines are essentially
91: hydrodynamic (see CLV02) and therefore it is not surprising that the GS95 predicted
92: the Kolmogorov one-dimensional energy spectrum in terms of \( k_{\bot } \),
93: i.e., \( E(k_{\bot })\sim k_{\bot }^{-5/3} \)(see review by Cho, Lazarian \& Yan 2002, henceforth CLY02).
94:
95: The GS95 model describes incompressible MHD turbulence. Recent
96: research suggests that the scaling is approximately true for the dominant Alfvenic modes in a compressible
97: medium with Mach numbers(\( M\equiv V/C_s \)) of the order of unity (Lithwick \& Goldreich
98: 2001, henceforth LG01, CLY02, Cho \& Lazarian 2002, in preparation), which is also consistent
99: with the analysis of observational data (Lazarian \& Pogosyan 2000,
100: Stanimirovic \& Lazarian 2001, CLY02). In what follows we apply the GS95
101: scaling to handle the problem of grain motions.
102:
103: Because of turbulence anisotropy, it is convenient to consider separately
104: grain motions parallel and perpendicular to the magnetic field. The
105: motions perpendicular to the magnetic field are influenced by
106: Alfven modes, while those parallel to the magnetic field are subjected
107: to the magnetosonic modes. The scaling relation for perpendicular motion
108: is \( v_{k}\propto k_{\perp }^{-1/3} \) (GS95). As the eddy turnover
109: time is \( \tau _{k}\propto (k_{\perp }v_{k})^{-1} \), the velocity
110: may be expressed as \( v_{k}\approx v_{max}\left( \tau _{k}/\tau _{max}\right) ^{1/2}, \)
111: where \( \tau _{max}=l_{max}/v_{max} \) is the time-scale for the
112: largest eddies, for which we adopt the fiducial values \( l_{max}=10 \)pc,
113: \( v_{max}=5 \)km/s.
114:
115: Grains are charged and coupled with the magnetic field. If the Larmor
116: time \( \tau _{L}=2\pi m_{gr}c/qB \) is shorter than the gas drag
117: time \( t_{drag} \), grain perpendicular motions are constrained
118: by the magnetic field. In this case, grains have a velocity dispersion
119: determined by the turbulence eddy whose turnover period is \( \sim \tau _{L} \), while grains move with the magnetic field on longer time
120: scales. Since the turbulence velocity grows with the eddy size, the
121: largest velocity difference occurs on the largest scale where grains
122: are still decoupled. Thus, following the approach in Draine (1985),
123: we can estimate the characteristic grain velocity relative to the
124: fluid as the velocity of the eddy with a turnover time equal to \( \tau _{L} \),
125: \begin{equation}
126: \label{vperp}
127: v_{\perp }(a)={v^{3/2}_{max}\over l^{1/2}_{max}}(\rho _{gr})^{1/2}\left( {8\pi ^{2}c\over 3qB}\right) ^{1/2}a^{3/2},
128: \end{equation}
129: and the relative velocity of grains to each other should be approximately equal to the larger one of the grains' velocities, i.e., the the larger grain's velocity,
130: \begin{eqnarray}
131: \delta v_{\perp }(a_{1},a_{2}) & = & {v^{3/2}_{max}\over l^{1/2}_{max}}(\rho _{gr})^{1/2}\left( {8\pi ^{2}c\over 3qB}\right) ^{1/2}[max(a_{1},a_{2})]^{3/2}\nonumber \\
132: & = & 1.4\times 10^{5}cm/s(v_{5}a_{5})^{3/2}/(q_{e}l_{10}B_{\mu })^{1/2},
133: \end{eqnarray}
134: in which \( v_{5}=v_{max}/10^{5} \)cm/s, \( a_{5}=a/10^{-5} \)cm,
135: \( q_{e}=q/1 \)electron, \( l_{10}=l_{max}/10 \)pc, \( B_{\mu }=B/1\mu \)G,
136: and the grain density is assumed to be \( \rho _{gr}=2.6 \)g/cm\( ^{-3} \)
137: .
138:
139: Grain motions parallel to the magnetic field are induced by the compressive
140: component of slow mode with \( v_{\parallel }\propto k_{\parallel }^{-1/2} \)
141: (CLV02, LG01, CLY02). The eddy turnover time is \( \tau _{k}\propto (v_{\parallel }k_{\parallel })^{-1} \),
142: so the parallel velocity can be described as \( v_{\parallel }\approx v_{max}\tau _{k}/\tau _{max} \)\footnote{%
143: We assume that turbulence is driven isotropically at the scale \( l_{max} \).
144: }. For grain motions parallel to the magnetic field the Larmor precession
145: is unimportant and the gas-grain coupling takes place on the translational
146: drag time \( t_{drag} \). The drag time due to collisions with atoms
147: is essentially the time for collision with the mass of gas equal to
148: the mass of grain, \( t^{0}_{drag}=(a\rho _{gr}/n)(\pi /8\mu kT)^{1/2}. \), where \( \mu \) is the mass of gas species.
149: The ion-grain cross-section due to long-range Coulomb force is larger
150: than the atom-grain cross-section (Draine \& Salpeter 1979). Therefore,
151: in the presence of collisions with ions, the effective drag time decreases,
152: \( t_{drag}=\alpha t^{0}_{drag} \), where \( \alpha <1 \) is the
153: function of a particular ISM phase. The characteristic velocity of
154: grain motions along the magnetic field is approximately equal to the
155: parallel turbulent velocity of eddies with turnover time equal to
156: \( t_{drag} \)
157:
158: \begin{equation}
159: \label{vpara}
160: v_{\parallel }(a)=\alpha {v^{2}_{max}\over l_{max}}\left( \frac{\rho _{gr}}{4n}\right) \left( {2\pi \over \mu kT}\right) ^{1/2}a,
161: \end{equation}
162: and the relative velocity of grains for \( T_{100}=T/100 \)K is
163:
164: \begin{eqnarray}
165: \delta v_{\parallel }(a_{1},a_{2}) & = & \alpha {v^{2}_{max}\over l_{max}}\left(\frac{\rho _{gr}}{4n}\right)({2\pi \over \mu kT})^{1/2}[max(a_{1},a_{2})]\nonumber \\
166: & = & (1.0\times 10^{6}cm/s)\alpha v_{5}^{2}a_{5}/(nl_{10}T_{100}^{1/2}),
167: \end{eqnarray}
168:
169:
170: When \( \tau _{L}>t_{drag} \), grains are no longer tied to the magnetic
171: field. Since at a given scale, the largest velocity dispersion is
172: perpendicular to the magnetic field direction, the velocity gradient
173: over the grain mean free path is maximal in the direction perpendicular
174: to the magnetic field direction. The corresponding scaling is analogous
175: to the hydrodynamic case, which was discussed in Draine (1985): \( \delta v(a_{1},a_{2})=v^{3/2}_{max}/ l^{1/2}_{max} t_{drag}^{1/2} \), i.e.,
176: \begin{equation}
177: \delta v(a_{1},a_{2})= \alpha ^{\frac{1}{2}}{v^{3/2}_{max}\over l^{1/2}_{max}}\left( \frac{\rho _{gr}}{4n}\right) ^{\frac{1}{2}}\left( {2\pi \over \mu kT}\right) ^{\frac{1}{4}}[max(a_{1},a_{2})]^{\frac{1}{2}}.\label{HD}
178: \end{equation}
179:
180:
181: Turbulence is damped due to the viscousity when the cascading rate \( v_\perp k_\perp \) equals the damping time \( t_{damp} \)
182: (see Cho, Lazarian \& Vishniac 2002b). If
183: the mean free path for a neutral particle \( l_{n} \), in a partially
184: ionized gas with density \( n_{tot}=n_{n}+n_{i} \), is much less
185: than the size of the eddy in consideration, i.e., \( l_{n}k_{\bot }\ll 1 \),
186: the damping time is \( t_{damp}\sim \nu _{n}^{-1}k_{\perp }^{-2}\sim \left( n_{tot}/n_{n}\right) \left( l_{n}v_{n}\right) ^{-1}k_{\perp }^{-2}, \)
187: where \( \nu _{n} \) is effective viscosity produced by neutrals.
188: In the present paper we consider cold gas with low ionization, therefore
189: the influence of ions on \( l_{n} \) is disregarded. Thus the turbulence
190: cutoff time in neutral medium is
191: \begin{equation}
192: \label{cutoff}
193: \tau _{c}\simeq \left( \frac{l_{n}}{v_{n}}\right) \left( \frac{v_{n}}{v_{max}}\right) ^{\frac{3}{2}}\left( \frac{l_{max}}{l_{n}}\right) ^{\frac{1}{2}}\left( \frac{V_{A}}{v_{max}}\right) ^{\frac{1}{2}}\left( \frac{n_{n}}{n_{tot}}\right) ,
194: \end{equation}
195: where \( v_{n} \) and \( V_{A} \) are, respectively, the velocity
196: of a neutral and Alfven velocity. It is easy to see that for \( \tau _{c} \)
197: longer than either \( t_{drag} \) or \( \tau _{L} \) the grain motions
198: get modified. A grain samples only a part of the eddy before gaining
199: the velocity of the ambient gas. In GS95 picture, the shear rate \( dv/dl \)
200: increases with the decrease of eddy size. Thus for \( \tau _{c}>max\{t_{drag},\tau _{L}\} \),
201: these smallest available eddies are the most important for grain acceleration.
202: Consider first the perpendicular motions. If \( v_{c} \) is the velocity
203: of the critically damped eddy, the distance traveled by the grain
204: is \( \bigtriangleup l\sim v_{c}\times min\lbrace t_{drag},\tau _{L}\rbrace \).
205: Thus the grain experiences the velocity difference \( \bigtriangleup l\times dv/dl\sim v_{c}\times min\lbrace t_{drag},\tau _{L}\rbrace /\tau _{c} \). Due to the critical balance in GS95 model, the shear rate along the magnetic field is \( dv/dl=v_ck_\parallel\sim v_c/(V_A\tau_c) \). Therefore, grain experiences a velocity difference
206: \( V_{A}/v_{c} \) times smaller, i.e., \( \sim v_c^2\times t_{drag}/(V_A\tau _{c}) \).
207:
208:
209: \section{Discussion }
210:
211:
212: \subsection{Shattering and Coagulation }
213:
214: Consider the cold neutral medium (CNM) with temperature \( T=100 \)K, density \( n_{\rm H}=30 \)cm\( ^{-3} \),
215: electron density \( n_{e}=0.045 \)cm\( ^{-3} \), magnetic field
216: \( B\sim 1.3\times 10^{-5} \)G (Weingartner \& Draine 2001a, hereafter WD01a). To account for the Coulomb drag,
217: we use the results by WD01a and get the modified drag time \( t_{drag}=\alpha t^{0}_{drag} \).
218: Using the electric potentials in Weingartner \& Draine (2001b), we
219: get grain charge and \( \tau _{L} \).
220:
221: For the parameters given above, we find that \( t_{drag} \) is larger
222: than \( \tau _{c} \) for grains larger than \( 10^{-6} \)cm, \( \tau _{L} \)
223: is smaller than \( \tau _{c} \) even for grains as large as \( 10^{-5} \)cm.
224: Here, we only consider grains larger than \( 10^{-6} \)cm, which
225: carry most grain mass (\( \sim 80\% \)) in ISM, so we can still use
226: Eq.(\ref{vpara}) to calculate grain parallel velocities and Eq.(\ref{vperp}) to get the perpendicular velocity for grain larger than \( 10^{-5} \)cm. Nevertheless,
227: the perpendicular velocities of grains smaller than \( 10^{-5} \)cm should be estimated as \( v'_{\perp }(a)=v_{c}\times (\tau _{L}/\tau _{c})=v_{max}(\tau _{c}/\tau _{max})^{1/2}(\tau _{L}/\tau _{c})=v_{\perp }(a)(\tau _{L}/\tau _{c})^{1/2}, \)
228: where \( v_{\perp }(a) \) is given by Eq.(\ref{vperp}). The results
229: are shown in Fig.1.
230:
231: The critical sticking velocity were calculated in Chokshi et al. (1993)(see also Dominik \& Tielens 1997).\footnote{There are obvious misprints in the numerical coefficient of Eq.(7) in Chokshi et al.(1993) and the power index of Young's modulus in Eq.(28) of Dominik \& Tielens (1997).} However, experimental work by Blum (2000) shows that the critical velocity is an order of magnitude larger than the theoretical calculation. Thus the collisions can result in coagulation for
232: small silicate grains (\(\leq 3\times 10^{-6} \)cm).
233:
234: With our input parameters, grains do not shatter if the shattering
235: thresholds for silicate is \( 2.7 \)km/s as in Jones et al. (1996).
236: Nevertheless, the grain velocities strongly depend on \( v_{max} \)
237: at the injection scale. For instance, we will get a cutoff \( 6\times 10^{-5} \)cm
238: due to shattering if \( v_{max}=10 \)km/s.
239:
240: For a dark cloud, the situation is different. As the density increases,
241: the drag by gas becomes stronger. Consider a typical dark cloud with
242: temperature \( T=20 \)K, density \( n_{\rm H}=10^{4} \)cm\( ^{-3} \)
243: (Chokshi et al. 1993) and magnetic field \( B\sim 2.3\times 10^{-4} \)G.
244: Assuming that dark clouds are shielded from radiation, grains get charged
245: by collisions with electrons: \( <q>=0.3(r/10^{-5} \)cm) electrons.
246: The ionization in the cloud is \( \chi =n_{e}/n_{tot}\sim 10^{-6} \)
247: and the drag by neutral atoms is dominant. From Eq.(\ref{cutoff}) and
248: the expression for the drag time and the Larmor time, we find \( \tau _{L}<t_{drag} \)
249: for grains of sizes between \( 10^{-6} \)cm and \( 4\times 10^{-6} \)cm,
250: and \( t_{drag}<\tau _{L} \) for grains larger than \( 4\times 10^{-6} \)cm.
251: In both cases, turbulence cutoff \( \tau _{c} \) is smaller than
252: \( t_{drag} \) and \( \tau _{L} \). Thus for the smaller grains,
253: we use Eq.(\ref{vperp}),(\ref{vpara}) to estimate grain velocities.
254: For larger grains, grain velocities
255: are given by Eq.(\ref{HD}).
256:
257: Our results for dark clouds show only a slight difference from the
258: earlier hydrodynamic estimates. Since the drag time \( t_{drag}\propto n^{-1} \),
259: Larmor time \( \tau _{L}\propto B^{-1}\propto n^{-1/2} \), the grain
260: motions get less affected by the magnetic field as the cloud becomes
261: denser. Thus we agree with Chokshi's et al. (1993) conclusion that
262: densities well in excess of \( 10^{4} \)cm\( ^{-3} \) are required
263: for coagulation to occur. Shattering will not happen because the velocities
264: are small, so there are more large grains in dark clouds. This agrees
265: with observations (see Mathis 1990).
266:
267: In the treatment above we disregarded the possibility of direct acceleration
268: of charged grains through their interactions with fluctuating magnetic
269: field. In our next paper we will show that this resonant process is important for a highly ionized
270: medium.
271:
272:
273: \subsection{Grain Segregation and Turbulent Mixing}
274:
275: Our results are also relevant to grain segregation. Grains are the
276: major carrier of heavy elements in the ISM. The issue of grain segregation
277: may have significant influence on the ISM metallicity. Subjected to
278: external forcing, e.g., due to radiation pressure, grains gain size-dependent
279: velocities with respect to gas. WD01a have considered the forces on
280: dust grains exposed to anisotropic interstellar radiation fields.
281: They included photoelectric emission, photodesorption as well as radiation
282: pressure, and calculated the drift velocity for grains of different
283: sizes. The velocities they got for silicate grains in the CNM range
284: from \( 0.1 \)cm/s to \( 10^{3} \)cm/s. Fig.1 shows that the turbulence
285: produces larger velocity dispersions.\footnote{%
286: If reconnection is fast (see Lazarian \& Vishniac 1999), the mixing
287: of grains over large scales is provided by turbulent diffusivity\( \sim v_{max}l_{max} \).
288: On small scales the grain decoupled motions are important.
289: } Thus the grain segregation of very small and large grains speculated
290: in de Oliveira-Costa et al. (2000) is unlikely to happen for typical
291: CNM conditions.
292:
293: A different mechanism of driving grain motions is a residual imbalance
294: in {}``rocket thrust{}'' between the opposite surfaces of a rotating
295: grain (Purcell 1979). This mechanism can provide grain relative motions and preferentially move grains into molecular
296: clouds. It is easy to see that due to averaging caused by grain rotation,
297: the rocket thrust is parallel to the rotation
298: axis. Three causes for the thrust were suggested by Purcell (1979):
299: spatial variation of the accommodation coefficient for impinging atoms,
300: photoelectric emission, and H\( _{2} \) formation. The latter was
301: shown to be the strongest among the three. The uncompensated force
302: in this case arises from the difference of the number of catalytic
303: active sites for H\( _{2} \) formation on the opposite grain surfaces.
304: The nascent H\( _{2} \) molecules leave the active sites with kinetic
305: energy \( E \) and the grain experiences a push in the opposite
306: directions. The number of active sites varies from one grain to another,
307: and we should deal with the expectation value of the force for a given
308: distribution of active sites.
309:
310: Due to internal relaxation of energy (see Lazarian \& Draine
311: 1999a,b, and review by Lazarian 2000) the grain rotational axis tends to be
312: perpendicular to the largest \( b-b \) surface. Adopting the approach in Lazarian \& Draine (1997), we get the mean
313: square root force of H\( _{2} \) thrust on a grain in the shape of
314: a square prism with dimensions \( b\times b\times a \) (\( b>a \))
315:
316: \begin{equation}
317: \label{H}
318: \langle F_{z{\rm H}}\rangle =r^{3/2}(r+1)^{1/2}\gamma (1-y)n_{\rm H}v_{\rm H}a^{2}\left( \frac{2m_{\rm H}E}{\nu }\right) ^{1/2},
319: \end{equation}
320: where \( r=b/2a, \), \( n_{\rm H}\equiv n({\rm H})+2n({\rm H}_2) \), \( y=2n({\rm H}_2)/n_{\rm H} \) is the \( {\rm H_2} \) fraction, \( \gamma \) is the fraction of impinging {\rm H}
321: atoms and \( \nu \) is the number of active sites over the grain
322: surface. The expected grain
323: velocity is \( v=\langle F_{z{\rm H}}\rangle t_{drag}/m \). In the CNM we consider, \( y=0 \), adopting the characteristic values in Lazarian \& Draine (1997), \( r=1, \)
324: \( \gamma =0.2 \), \( E=0.2 \)eV, and the density of active sites \( 10^{11} \)cm\(^{-2}\) so that \( \nu =80(a/10^{-5} \)cm\( )^{2}r(r+1) \),
325: we get the {}``optimistic{}'' velocity shown in Fig 1. For maximal
326: active site density \( 10^{15} \)cm\( ^{-2} \), we get the lower
327: boundary of grain velocity \( v\simeq 3.3(10^{-5}{\rm cm} /a)^{1/2} \)cm/s. The scaling is approximate due to the complexity of coefficient \( \alpha \)(see WD01a Fig.16).
328:
329: Lazarian \& Draine (1999a,b) have shown that subjected to {\rm H}\( _{2} \) torques alone, grains
330: \( \leq 10^{-4} \)cm should experience frequent thermal flipping,
331: which means that the \( F_{z{\rm H}} \) fluctuates. This flipping
332: results from coupling of grain rotational and vibrational degrees
333: of freedom through internal relaxation and would average out \( \langle F_{z{\rm H}}\rangle \).
334: However, the flipping rate depends on the value of the grain angular
335: momentum (Lazarian \& Draine 1999a). If a grain is already spun up to a sufficient velocity,
336: it gets immune to thermal flipping. Radiative torques (Draine \& Weingartner
337: 1996) can provide efficient spin if the grain size is comparable
338: to the wavelength. For a typical interstellar diffuse radiation field,
339: the radiative torques are expected to spin up grains with sizes larger
340: than \( \sim 4\times 10^{-6} \)cm. They will also align grains with
341: rotational axes parallel to the magnetic field. Thus grains should acquire
342: velocities along the magnetic field lines and the corresponding velocities
343: should be compared with those arising from turbulent motions parallel to the magnetic
344: field. It is clear from Fig.1 that for the chosen set of parameters
345: the effect of {\rm H}\( _{2} \) thrust is limited. All in all, we conclude that the radiation effects and H\(_2\) thrust are not efficient for segregating grains in typical ISM conditions.
346: \begin{figure}
347: \centering \leavevmode
348: \resizebox*{0.33\textwidth}{0.25\textheight}{\includegraphics{f1.eps}}
349:
350:
351: \caption{Grain velocities as a function of radii (solid line) in
352: the CNM. Dashdot line represents parallel velocity
353: due to the drag by compressible modes, dotted line refers to perpendicular
354: velocity from the contribution of the drag by Alfven mode, also plotted
355: is the earlier hydrodynamic result (dashed line). The change of the slope
356: is due to the cutoff of turbulence by ambipolar diffusion. The grain
357: velocity driven by {\rm H}\protect\( _{2}\protect \) thrust is plotted
358: to illustrate the issue of grain segregation in the CNM
359: (see text), the part marked by 'o' is nonphysical because thermal
360: flipping is not taken into account. }
361: \end{figure}
362:
363:
364:
365: \section{Summary}
366:
367: We have calculated relative motions of dust grains in a magnetized turbulent
368: fluid taking into account turbulence anisotropy, turbulence damping
369: and grain coupling with the magnetic field. We find that these effects
370: decrease the relative velocities of dust grains compared to the earlier
371: hydrodynamic-based calculations. The difference is substantial in
372: CNM, but less important for dark clouds. For CNM we find that coagulations
373: of silicate grains happen for sizes \(\leq 3\times 10^{-6} \)cm. The
374: force due to {\rm H}\( _{2} \) formation on grain surface might drive small
375: grains (\( <3\times 10^{-6} \)cm) to larger velocities but thermal
376: flipping of grains suppresses the forces for grains less than \( 4\times 10^{-6} \)cm.
377: We conclude that radiation and H\( _2 \) thrust are not capable of segregating grains.
378:
379: We are grateful to John Mathis for reading the manuscript and many
380: important comments. We thank our referee Dr. Stuart Weidenschilling for helpful comments. The research is supported by the NSF grant AST0125544.
381:
382: \begin{thebibliography}{10}
383: \bibitem{}Arons, J. \& Max, C.E. 1975, ApJ, 196, L77
384: \bibitem{}Blum, J. 2000, Space Sci. Rev., 92, 265B
385: \bibitem{}Cho, J. \& Vishniac, E.T. 2000, ApJ, 539, 273
386: \bibitem{}Cho, J., Lazarian, A. \& Yan, H. 2002, ASP, in press, astro-ph/0112366
387: \bibitem{}Cho, J., Lazarian, A. \& Vishniac, E.T. 2002a, ApJ, 564, 000
388: \bibitem{}Cho, J., Lazarian, A. \& Vishniac, E.T. 2002b, ApJL, in press
389: \bibitem{}Chokshi, A., Tielens, A.G.G.A. \& Hollenbach, D. 1993, ApJ, 407, 806
390: \bibitem{}de Oliveira-Costa, A., Tegmark, M., Devies, R.D., Gutierrez, C.M.,
391: Mark J., Haffner, L.M., Jones, A.W., Lasenby, A.N., Rebolo, R., Reynolds,
392: R.J., \& Tufte, S.L., Watson, R.A. 2000, astro-ph/0010527
393: \bibitem{}Dominik, C. \& Tielens, A.G.G.A. 1997, ApJ, 480, 647
394: \bibitem{}Draine, B.T. 1985, in Protostars and Planets II, ed. D.C. Black \&
395: M.S. Matthews (Tucson: Univ. Arizona Press)
396: \bibitem{}Draine, B.T.\& Weingartner, J.C. 1996, ApJ 470, 551
397: \bibitem{}Draine, B.T. \& Salpeter, E.E. 1979, ApJ, 231, 77
398: \bibitem{Goldreich Sridhar 1995}Goldreich, P. \& Sridhar, H. 1995, ApJ, 438, 763
399: \bibitem{}Jones, A.P., Tielens, A.G.G.M. \& Hollenbach, D.J. 1996, ApJ, 469, 740
400: \bibitem{}Kusaka, T., Nakano, T., \& Hayashi, C., 1970, Prog. Theor. Phys.,
401: 44, 1580
402: \bibitem{}Lazarian, A. 2000, in {}``Cosmic Evolution and Galaxy Formation{}'',
403: ASP v.215, eds. Jose Franco, Elena Terlevich, Omar Lopez-Cruz, Itziar
404: Aretxaga, p. 69
405: \bibitem{}Lazarian, A. \& Draine, B.T. 1997, ApJ, 487, 248
406: \bibitem{}Lazarian, A. \& Draine, B.T. 1999a, ApJ, 516, L37
407: \bibitem{}Lazarian, A. \& Draine, B.T. 1999b, ApJ, 520, L67
408: \bibitem{}Lazarian, A. \& Pogosyan, D. 2000, ApJ, 537, 720
409: \bibitem{}Lazarian, A., Vishniac, E.T., 1999, ApJ, 517, 700
410:
411: \bibitem{}Lithwick, Y. \& Goldreich, P. 2001, ApJ, 562, 279
412: \bibitem{}Maron, J. \& Goldreich, P. 2001, ApJ, 554, 1175
413: \bibitem{}Mathis, J.S. 1990, ARA\&A, 28, 37
414: \bibitem{}Mukherjee, P., Jones, A.W., Kneissl, R., Lasenby, A.N. 2001, MNRAS,
415: 320, 224
416: \bibitem{}Ossenkopf, V. 1993, A\&A 280, 617
417: \bibitem{}Purcell, E.M. 1979, ApJ, 231, 404
418: \bibitem{}Stanimirovic, S. \& Lazarian, A. 2001, ApJ, 551, L53
419: \bibitem{}Volk, H.J., Jones, F.C., Morfill, G.E., \& Roser, S. 1980 A\&A, 85,
420: 316
421: \bibitem{}Weidenschilling, S.J. \& Ruzmaikina, T.V. 1994, ApJ, 430, 713
422: \bibitem{}Weingartner, J.C. \& Draine, B.T. 2001a, ApJ, 553, 581
423: \bibitem{}Weingartner, J.C. \& Draine, B.T. 2001b, ApJs, 134, 263
424: \end{thebibliography}
425:
426: \end{document}
427: