1: %%

2: %% Beginning of file 'sample.tex'

3: %%

4: %% Modified 2005 December 5

5: %%

6: %% This is a sample manuscript marked up using the

7: %% AASTeX v5.x LaTeX 2e macros.

8: 

9: \documentclass[12pt,preprint]{aastex}

10: 

11: %% manuscript produces a one-column, double-spaced document:

12: 

13: %%\documentclass[manuscript]{aastex}

14: 

15: %% preprint2 produces a double-column, single-spaced document:

16: 

17: %% \documentclass[preprint2]{aastex}

18: 

19: \newcommand{\vdag}{(v)^\dagger}

20: \newcommand{\myemail}{skywalker@galaxy.far.far.away}

21: \def\kms{km\,s$^{-1}$\,}

22: \def\vs{$v_{s}$\,\,}

23: \def\etal{ et~al.\rm\,}

24: \def\sb{ergs cm$^{-2}$ s$^{-1}$ sr$^{-1}$\,}

25: \def\dem71{DEM L 71\,}

26: \def\chan{{\it Chandra\,}}

27: \def\bn{$I_{B}/I_{N}$\,}

28: \def\tetp{($T_{e}/T_{p}$)$_{0}$\,}

29: \def\fneut{$f_{H^{0}}$\,}

30: \def\memp{$m_{e}/m_{p}$\,}

31: 

32: %% You can insert a short comment on the title page using the command below.

33: %\slugcomment{Not to appear in Nonlearned J., 45.}

34: %\shorttitle{Collapsed Cores in Globular Clusters}

35: %\shortauthors{Djorgovski et al.}

36: 

37: 

38: \begin{document}

39: 

40: \title{The Heating of Thermal Electrons in Fast Collisionless Shocks: The

41: Integral Role of Cosmic Rays}

42: 

43: 

44: %% Use \author, \affil, and the \and command to format

45: %% author and affiliation information.

46: %% Note that \email has replaced the old \authoremail command

47: %% from AASTeX v4.0. You can use \email to mark an email address

48: %% anywhere in the paper, not just in the front matter.

49: %% As in the title, use \\ to force line breaks.

50: 

51: \author{Cara E. Rakowski\altaffilmark{1}}

52: \affil{Space Science Division, Naval Research Laboratory Code 7674R,

53: Washington, D.C. 20375} %\email{crakowski@ssd5.nrl.navy.mil}

54: \author{J. Martin Laming}

55: \affil{Space Science Division, Naval Research Laboratory Code 7674L,

56: Washington, D.C. 20375} %\email{laming@nrl.navy.mil}

57: 

58: \and

59: 

60: \author{Parviz Ghavamian}

61: \affil{Department of Physics and Astronomy, Johns Hopkins University,

62: 3400 N. Charles Street, Baltimore, MD, 21218}

63: \email{parviz@pha.jhu.edu}

64: 

65: \altaffiltext{1}{National Research Council Fellow}

66: 

67: \begin{abstract}

68: Understanding the heating of electrons to quasi-thermal energies at

69: collisionless shocks

70: %%%%

71: has broad implications for plasma astrophysics.

72: It directly impacts the interpretation of X-ray spectra from shocks,

73: is important for understanding how energy is partitioned between the

74: thermal and cosmic ray populations, and provides insight into the

75: structure of the shock itself.

76: %%%%

77: In \citet{GLR07}

78: we presented observational evidence for an inverse square

79: relation between the electron-to-proton temperature ratio and the

80: shock speed at the outer blast waves of supernova remnants

81: in partially neutral interstellar gas. There we outlined how

82: lower hybrid waves generated in the cosmic ray precursor could

83: produce such a relationship by heating the electrons to a common

84: temperature independent of both shock speed and the strength of the

85: ambient magnetic field. Here we explore the mechanism of

86: lower hybrid wave heating of electrons in more detail. Specifically we examine

87: the growth rate of the lower

88: hybrid waves for both the kinetic (resonant) and reactive cases.

89: We find that only the kinetic case

90: exhibits a growing mode.   At low Alfv\'{e}n Mach numbers ($\sim$15)

91: the growth of lower hybrid waves can be faster than the

92: magnetic field

93: amplification by modified Alfv\'{e}n waves.

94: 

95: 

96: \end{abstract}

97: 

98: \keywords{ISM: Cosmic Rays, ISM: Shock Waves, ISM: Supernova Remnants}

99: 

100: \section{Introduction}

101: 

102: The

103: %major acceleration environment

104: main accelerators of

105: cosmic rays (CRs) are widely

106: believed to be high Mach number shocks in collisionless plasma, here

107: loosely defined as plasma where charged particles interact

108: predominantly through plasma waves rather than by Coulomb collisions

109: \citep[for a thorough conceptual and historical review

110: see][]{MalkovDrury2001}. However, a concensus is emerging that

111: CRs are not simply a byproduct of collisionless shocks, but

112: %rather

113: in fact

114: play an integral role in the shock structure, dynamics, and

115: energetics. For example sound waves in a cosmic ray pressure

116: gradient can smooth out the shock jump in cosmic ray modified shocks

117: \citep{Drury86}. More recent analytic and

118: numerical work has shown that modified Alfv\'{e}n waves in the CR precursor may

119: amplify the magnetic field to many times its ambient value by

120: generating perpendicular magnetic field from an initially

121: quasi-parallel geometry %\citep{Bell2004,Bell2005}.

122: \citet{LucekBell2000,BellLucek2001,Bell2004,Bell2005}.

123: %

124: Observational support for dramatic magnetic field amplification ahead of shocks

125: exists in the form of extremely thin X-ray synchrotron rims of

126: supernova remnants (SNRs) such as Cassiopeia A \citep{VinkLaming2003}, SN

127: 1006 \citep{Long2003, Yamazaki2004}, and Tycho's SNR (Warren et al. 2005; Cassam-Chena{\"i} et al. 2007).

128: %

129: As noted by \citet{Cassam2007} there are two possible interpretations for the narrow width,

130: however both require a dramatically amplified magnetic field ahead of the shock. Either the rims are thin

131: because the high magnetic field causes rapid synchrotron

132: cooling of the X-ray emitting electrons or the scale of the rims represents

133: the scale of magnetic field de-amplification behind the shock.

134: %As noted by \citet{Cassam2007}, caution is warranted in the exact

135: %interpretation of the width since it can represent either the synchrotron

136: %cooling length scale behind the shock or the scale for magnetic field de-amplification.

137: %In either case a dramatically amplified magnetic field ahead of the shock is

138: %required.

139: 

140: 

141: In this paper we explore another area where CRs may influence the

142: properties of the shock, namely through the heating of quasi-thermal

143: electrons.

144: For shocks in collisionless plasma the

145: heating of electrons must occur through the damping of waves

146: generated by the other more massive charged particles that dominate

147: the energetics. Given the wide array of possible plasma

148: instabilities at collisionless shocks an observational relationship

149: for electron temperature, $T_{e}$ at the shock front was required to

150: limit theoretical discussions. An inverse relationship between the

151: initial ratio of electron to proton temperatures immediately behind

152: the shock,

153: \tetp,

154: and the shock velocity,

155: \vs,

156:  has been reported in a series of

157: observational papers on supernova remnant

158: shocks \citep{G01,G02,GRHW03,RGH03,R05,GLR07} and has also been noted

159: among the higher Alfv\'en Mach number events in a sample of solar

160: wind shocks \citep{Schwartz88}. In \citet{GLR07}, we focused on

161: shocks propagating into partially neutral gas.

162: Here the collisional excitation of broad and narrow Balmer line emission at the shock

163: front can be used to diagnose

164: \tetp.  In \citet{G01}, \citet{G02}

165: and \citet{GLR07} we described the method of simultaneously

166: constraining \vs\, and \tetp\, via measurement of the

167: width of the broad Balmer line and the ratio of broad to narrow

168: Balmer line flux \citep[see also][]{Heng07a,Heng07b}.

169: Our results are consistent with an inverse square

170: relationship, \tetp $\propto 1/v_{s}^{2}$ for shock speeds above

171: $\sim$400~km~s$^{-1}$ (Ghavamian, Laming \& Rakowski 2007). Given

172: that $T_{p} \propto v_{s}^{2}$ at the shock front by the

173: Rankine-Hugoniot jump conditions, the inverse relationship between

174: equilibration and shock speed implies that the electron temperature

175: itself is nearly {\it constant} $\sim$0.3 keV, independent of shock

176: speed.

177: 

178: The insensitivity of electron temperature to shock velocity suggests a heating mechanism

179: within the extended diffusive cosmic ray precursor ahead of the shock.

180: In this case the

181: electron heating would be more reflective of the generic properties of cosmic ray

182: acceleration and diffusion than tied to the specific attributes of the shock.

183: In contrast, prior work on

184: heating by shock-reflected ions that are confined to within a gyroradius of

185: the (quasi-perpendicular) shock \citep{Cargill88}

186: suggested that \tetp\, would remain constant

187: with shock velocity.

188: In \citet{GLR07} we suggested that lower hybrid waves in the cosmic ray

189: precursor of a perpendicular shock might be a plausible electron heating

190: mechanism.

191: 

192: Lower hybrid waves are electrostatic ion waves directed

193: nearly perpendicular to the magnetic field with a frequency equal

194: to the geometric mean of the electron and ion gyrofrequencies.

195: Electrons that would

196: otherwise screen the ion oscillation are pinned to magnetic field lines.

197: In addition, the group velocity parallel to the magnetic field greatly exceeds the group

198: velocity perpendicular to the field ($\omega /k_{\|} >>\omega /k_{\perp}$).

199: Therefore, the wave can simultaneously resonate with ions moving across the

200: field lines and electrons moving along the field lines, facilitating collisionless

201: energy exchange between them. Based on simple arguments about the width of the

202: cosmic ray precursor and the electron diffusion along the field

203: lines, we showed that electron heating from lower hybrid waves in

204: the cosmic ray precursor would be independent of both the shock

205: speed and the magnetic field. Here we explore this mechanism in

206: more physical detail.

207: 

208: In Section 2 we calculate the growth rate of lower hybrid

209: waves, first examining the kinetic (resonant) then the reactive

210: (non-resonant) case.

211: The treatment here is mathematically similar to the work

212: on modified Alfv\'{e}n waves by \citet{Achterberg83} and

213: \citet{Bell2004,Bell2005} involving the cosmic ray contribution to the plasma dielectric

214: tensor.

215:  The analysis also draws on the work of \citet{L01a}

216: and \citet{L01b} on lower hybrid waves from shock-reflected ions.

217: We compare these growth rates with those for magnetic field amplification,

218: to assess the conditions under which electron heating might occur.

219: %

220: In section 3 we discuss the structure of the cosmic ray shock precursor

221: in more detail. We pay particular attention to the magnetic field geometry,

222: since the excitation of lower  hybrid waves requires a

223: quasi-perpendicular shock. We show schematically how magnetic field

224: amplification and lower-hybrid wave heating might co-exist in the shock

225: precursor for either parallel or perpendicular initial geometries.

226: %

227: We also review some other ideas for electron heating, and make some

228: quantitative predictions from our model for various shock

229: parameters. Included in the appendices are a

230: discussion of cosmic ray diffusion coefficients and a derivation of

231: the resonant growth rate for electromagnetic waves.

232: 

233: 

234: \section{Cosmic Ray Growth Rate of Lower Hybrid Waves}

235: Lower hybrid waves ahead of collisionless shocks have particularly interesting properties.

236: They can have a group velocity away from the shock equal to the

237: shock velocity itself \citep{mcclements97}. {\it This can in principle allow the waves to grow to

238: large amplitudes, even if their intrinsic growth rate is small.}

239: To determine if lower hybrid waves can heat the electrons to the

240: $\sim$0.3 keV temperature observed, we must first calculate the

241: growth rate of this instability to see if it will have sufficient

242: power to overcome the damping effect of the electrons as well as

243: to compete with other instabilities in the precursor.

244: We calculate this growth rate in both kinetic

245: and reactive limits, i.e. either considering the

246: cosmic rays with energies in resonance with the lower hybrid wave frequency or the

247: integrated contribution of the entire distribution, respectively

248: \citep[see e.g][]{melrose86}.

249: Related kinetic and reactive cases were calculated in \citet{L01a}

250: and \citet{L01b},  but only for the case of shock reflected,

251: % added "non-relativistic"...

252: non-relativistic ions

253: gyrating around the magnetic field, represented as a particle beam.

254: Here we begin the discussion with the resonant case.

255: 

256: \subsection{Kinetic Growth Rate}

257: 

258: We model the normalized cosmic ray distribution function diffusing upstream as

259: 

260: \begin{equation}

261: f(p) = \frac{n_{CR}^{\prime}}{4\sqrt{2}(\pi \kappa)^{3/2}p_{t}^{3}}

262: \frac{(2\kappa-3)\Gamma(\kappa)}{\Gamma(\kappa -1/2)}

263: \left[ 1 + \frac{(p_{x}-mv_{s})^{2} + p_{y}^{2} +p_{z}^{2}}

264:                   {2\kappa p_{t}^{2}} \right]^{-\kappa}

265: \end{equation}

266: %

267: where the coordinate system is aligned so that the shock speed $v_s$ lies in the

268: $x$-direction.  Here $p_t$ is defined as the ``thermal'' momentum, which we take

269: to be $\left(3/4\right) mv_s$, and $n_{CR}^{\prime}$ denotes the density of

270: suprathermal particles with distribution function $f\left(p\right)$.

271: The functional form above, known as a

272: ``kappa'' distribution, is often seen for particle distribution

273: functions associated with shocks in the solar wind, and may be derived as equilibrium

274: distributions for a system of particles and waves under certain conditions

275: \citep[see e.g.][]{laming07}, in contrast to

276: a system of particles only which gives a Maxwellian distribution

277: of width $p_t$.

278: The kappa

279: distribution resembles a Maxwellian for $p < p_t$, and in fact for $\kappa\rightarrow

280: \infty$ is exactly a Maxwellian. At higher particle momenta it tends

281: smoothly to a distribution $f\propto p^{-2\kappa }$. Below we shall take $\kappa =2$

282: to model the well known $f(p)\propto p^{-4}$ cosmic ray distribution predicted by

283: diffusive shock acceleration in shocks with a compression ratio of 4.

284: In connecting the cosmic rays to the

285: lower energy particles in this way, we are somewhat blurring the

286: distinction between ``cosmic rays'' and other suprathermal particles reflected from

287: the shock. Hence we denote the combined density of these particles as $n_{CR}^{\prime}$

288: to distinguish it from density of true cosmic rays, $n_{CR}$, which will appear

289: in expressions derived by other authors.

290: Note that all these particles

291: are distinct from the ambient thermal plasma upstream of the shock, which here is

292: considered to be a Maxwellian with much lower temperature than $p_t$ used in

293: Equation (1) for the upstream suprathermals.

294: When discussing the kinetic instability we will be

295: focusing on particles that obey a diffusion equation ahead of the shock, due to

296: their interaction with turbulence, rather than gyrating around field

297: lines.  We will qualify this distinction more carefully

298: below in our discussion of the reactive instability.

299: 

300: The appropriate dispersion relation for Equation (1) can be found from

301: the cold plasma dieletric tensor. For electrostatic waves at frequencies

302: close to the lower hybrid wave frequency we have \citep{L01a,L01b}

303: \begin{equation}

304: K_{L} = 1 + \frac{\omega_{pe}^{2}}{\Omega_{e}^{2}} \sin^{2}\theta

305:  - \frac{\omega_{pi}^{2}}{\omega^{2}}

306:  - \frac{\omega_{pe}^{2}}{\omega^{2}}\cos^{2}\theta

307: + \frac{4\pi q^{2}}{k^{2}} \int \frac{{\bf k} \cdot \partial f/\partial \bf p}

308:     {\omega - \bf k  \cdot \bf v } d^{3}{\bf p}  = 0

309: \end{equation}

310: where $\theta$ denotes the angle between the wavenumber

311: of the perturbation

312: and the preshock magnetic field;

313: $\omega_{px} = (4\pi q^2 n_{x}/m_{x})^{(1/2)}$ is the plasma frequency of a

314: given species $x$

315: (electrons, ions, cosmic rays...) with charge $q$, density $n_{x}$, and mass $m_{x}$;

316: $\Omega_{x} = qB/(m\gamma_L c)$ is the cyclotron frequency of species $x$

317: (with $\gamma_L$ being the Lorentz factor);

318: and unadorned $\omega$ being the lower hybrid wave frequency which is the

319: geometric mean of the electron and proton cyclotron frequencies.

320: %

321: Using the Landau prescription for evaluating the integral at the resonant pole and taking

322: only the imaginary parts of the dielectric tensor equation, we find

323: the growth rate for the lower hybrid waves

324: \begin{equation}

325: %\Rightarrow

326: \gamma = \frac{2\pi^{2}q^{2}}{k^{2}} \frac{\omega^{2}}{\omega_{pi}^{2}+

327: \omega_{pe}^{2}\cos^{2}\theta} \int \delta(\omega - {\bf k} \cdot {\bf v})

328:   {\bf k} \cdot \frac{\partial f}{\partial {\bf p}} d^{3}{\bf p}.

329: \end{equation}

330: 

331: To compute $\gamma$, we take $n_{CR}^{\prime}\propto\exp{-xv_s/D}$ where $D$

332: is the cosmic ray diffusion coefficient and is in principle

333: dependent on the cosmic ray momentum  (making $l=D/v_{s}$ the characteristic

334: diffusive lengthscale).

335: With this substitution we begin the evaluation of the last integral in Equation (3)

336: \begin{equation}

337: \int^{\infty}_{0} 2\pi p_{\bot}fdp_{\bot} =

338: \frac{n_{CR}^{\prime}}{4\sqrt{2}(\pi \kappa)^{3/2}p_{t}^{3}}

339: \frac{(2\kappa-3)\Gamma(\kappa)}{\Gamma(\kappa -1/2)} \frac{2\pi

340: \kappa}{\kappa -1} p_{t}^{2} \left[ 1 +

341: \frac{(p_{x}-mv_{s})^{2}}{2\kappa p_{t}^{2}} \right]^{1-\kappa}

342: e^{-xv_{s}/D},

343: \end{equation}

344: where we have

345: separated

346: out the components of ${\bf p}$ perpendicular to the shock

347: (and ${\bf k}$) from $p_x$.

348: Substituting back into Equation (3) yields a growth rate

349: \begin{eqnarray}

350: & & \gamma = \left( \frac{\pi}{\kappa}\right)^{3/2}\frac{q^{2}}{p_{t}k}

351: \frac{\omega^{3} n_{CR}^{\prime}}{\omega_{pi}^{2}+ \omega_{pe}^{2}\cos^{2}\theta}

352: \frac{(2\kappa-3)\Gamma(\kappa)}{\sqrt{2}\Gamma(\kappa -1/2)}

353: \int \delta(\omega - {\bf k} \cdot {\bf v}) \nonumber \\

354: & & \times \left\{ -\left[1+\frac{(p_{x}-mv_{s})^{2}}{2\kappa

355: p_{t}^{2}}\right]^{-\kappa} \frac{(p_{x}-mv_{s})}{p_{t}^{2}} +

356: \frac{\kappa}{\kappa -1}\frac{xv_{s}}{D^{2}}\frac{\partial

357: D}{\partial p_{x}} \left[ 1 + \frac{(p_{x}-mv_{s})^{2}}{2\kappa

358: p_{t}^{2}} \right]^{1-\kappa} \right\} e^{-xv_{s}/D} dp_{x}.

359: \end{eqnarray}

360: For waves to stay in contact with the shock

361: $w/k\simeq -2v_{s}$ \citep{L01a} in the cold plasma electrostatic limit, i.e. reflected ions returning to the shock

362: excite the waves. This remains generally true when

363: these approximations are

364: relaxed \citep{L01b}, so the $\delta$ function picks out $p_{x} = - 2mv_{s}$

365: (cosmic rays returning to the shock). Hence

366: \begin{eqnarray}

367: & & \gamma = \left( \frac{\pi}{\kappa}\right)^{3/2}

368: \frac{q^{2}\omega^{3} n_{CR}^{\prime} v_{s}^{2}m_{CR}}

369: {p_{t}(\omega_{pi}^{2}+ \omega_{pe}^{2}\cos^{2}\theta)}

370: \frac{(2\kappa-3)\Gamma(\kappa)}{\sqrt{2}\Gamma(\kappa -1/2)}

371: \nonumber \\

372: & & \times\left[1+ \frac{9m^{2}v_{s}^{2}}{2\kappa p_{t}^{2}} \right]^{-\kappa}

373: \left\{ \frac{3mv_{s}}{p_{t}^{2}} + \frac{\kappa xv_{s}}{(\kappa-1) D^{2}}

374: \frac{\partial D}{\partial p_{x}}

375: \left( 1+ \frac{9m^{2}v_{s}^{2}}{2\kappa p_{t}^{2}} \right)

376: \right\} e^{-xv_{s}/D}

377: \end{eqnarray}

378: %

379: Assuming $\partial D/\partial p_{x} = 0$ (see Appendix A

380: on cosmic ray diffusion coefficients) and $\kappa = 2$, only the

381: first term within curly brackets remains. Evaluating the $\Gamma$ functions

382: and substituting $p_{t} = (3/4)mv_{s}$ we arrive at

383: the following expression for the kinetic growth rate of lower hybrid waves

384: %

385: \begin{equation}

386: \gamma =  \frac{8}{225} \frac{\omega_{pi}^{\prime 2} \omega}{\omega_{pi}^{2}

387: + \omega_{pe}^{2}\cos^{2}\theta} e^{-xv_s{}/D}.

388: \end{equation}

389: where $\omega_{pi}^{\prime}$ denotes the plasma frequency for

390: $n_{CR}^{\prime}$. Substituting in the frequency definitions we note

391: that approximately $\gamma \propto (n_{CR}^{\prime}/ n_{i})\Omega_{i}$.

392: %

393: Before proceeding, we pause to compare this growth rate with those for

394: magnetic field amplification. In the case where Alfv\'en waves are resonantly

395: excited, the growth rate is \citep{melrose86,Pelletier2006}

396: \begin{equation}

397: \gamma _{B,res}={3\pi\over 16}{\Omega _i\over v_A}{n_{CR}\over n_i}

398: \left({\cos\theta\over |\cos\theta |}v_s\cos\phi -{4\over 3}v_A-

399: {\pi\over 4}v_s\sin\phi\right)k_{||}r_g

400: \end{equation}

401: where $v_A$ is the Alfv\'en speed, $n_i$ is the density of ions in the background

402: plasma and $r_g$ is the gyroradius of cosmic rays. This expression differs from that

403: in the cited references in the factor $\cos\phi$ and the term in $\sin\phi$,

404: where $\phi$ is the angle between

405: the shock velocity and the magnetic field. At perpendicular shocks,

406: the growth rate of resonant Alfv\'en waves

407: can be neglected, but at parallel shocks may be larger than that for lower hybrid

408: waves, depending on the ratio $n_{CR}^{\prime}/n_{CR}$.  However as we shall argue

409: later, all shocks subject to magnetic field amplification become perpendicular, and this

410: is the geometry where lower hybrid waves are most effectively excited, so we neglect

411: $\gamma _{B,res}$ from here onward. \citet{Bell2004} discovered a nonresonant growth rate

412: for Alfv\'en waves, with approximate growth rate

413: \begin{equation}

414: \gamma _{B,nonres}=\sqrt{{n_{CR}\over n_i}k_{||}v_s\Omega _i-k_{||}^2v_A ^2}

415: \end{equation}

416: which has a maximum value of $M_A\Omega _i n_{CR}/2n_i$.

417: According to \citet{Bell2005}, this instability operates for arbitrary orientations

418: of ${\bf B}$, ${\bf v_s}$ and ${\bf k}$, indicating that it will also amplify

419: magnetic field at perpendicular shocks.

420: Its growth rate is

421: strongest for ${\bf k}||{\bf B}$

422: and zero for ${\bf k}\perp {\bf B}$. Equating the maximum value of $\gamma _{B,nonres}$

423: with the lower hybrid wave growth rate calculated above, we find

424: the critical Alfv\'enic Mach number

425: $M_A\simeq 3n_{CR}^{\prime}/n_{CR}$, such that for higher $M_A$, cosmic rays

426: preferentially amplify magnetic field, and for lower $M_A$ they generate lower hybrid

427: waves. The numerical value depends on the ratio $n_{CR}^{\prime}/n_{CR}$. In the

428: next section we will argue that these two densities should not be the same, and that

429: $n_{CR}^{\prime}>n_{CR}$, following from a consideration of the reactive growth rate

430: for lower hybrid waves.

431: 

432: 

433: \subsection{Reactive Growth Rate}

434: 

435: The reactive case involves the integrated contribution to the growth rate from the entire cosmic ray distribution.

436: Thus we will examine successive orders in an expansion of $f(p)$ to see if they produce any growing modes. Although no

437: instability is found in this process, we do uncover potentially interesting constraints on the properties of $f(p)$.

438: 

439: We consider again the last term in Equation (2), the cosmic ray contribution to the dielectric tensor

440: \citep[e.g.][]{melrose86} which includes a factor that reduced to unity for the resonant case;

441: \begin{equation}

442: K_{L}^{CR} = \frac{4\pi q^{2}}{k^{2}}

443:     \int \frac{{\bf k} \cdot \partial f/\partial \bf p}

444:   {\omega - \bf k  \cdot \bf v } \frac{\bf k  \cdot \bf v}{\omega} d^{3}{\bf p}.

445: \end{equation}

446: For the case of a beam of cosmic ray particles localized around $\bf v_{s}$, one

447: recovers the usual beam reactive instability \citep[equation (A6)]{L01b}. However, as

448: we demonstrate below, for

449: a more physical quasi-isotropic cosmic ray distribution no instability is recovered.

450: We consider the case where cosmic rays drift with velocity ${\bf v_s}$, and

451: waves are generated with ${\bf k}\| {\bf v_s}$.

452: The CR distribution function from appendix A, expanded in terms of the cosine

453: of the angle between ${\bf k}$ and ${\bf v}$ or ${\bf k}$ and ${\bf v_{s}}$, $\cos\alpha = \mu$, is

454: \begin{equation}

455: f = f_{0} + \mu \frac{\partial f}{\partial \mu} + ...

456: = f_{0}(1 + \mu \frac{3v_{s}}{v} + ...),

457: \end{equation}

458: such that $\int \mu v f d \Omega = f_{0}v_{s}$ as before.

459: %

460: Neglecting terms of order $v_{s}^2/v^{2}$, the CR contribution to the dielectric

461: tensor becomes

462: % I changed "=" to "approximately equals", since the expression is no longer exact.

463: % I used this for the next few equations.

464: \begin{equation}

465: K_{L}^{CR} \approx \frac{4\pi q^{2}}{k^{2}} \int_{0}^{\infty} 2\pi

466:     \int_{-1}^{1} \frac{vk^2}{\omega}

467:  \frac{(\mu^{2}+3\mu^{3}v_{s}/v)}{\omega - k v \mu } d\mu

468: \frac{\partial f_{0}}{\partial p}p^{2}dp.

469: \end{equation}

470: %

471: Expanding the numerator into terms divisible by $(\omega/kv - \mu)$ and

472: evaluating the integral over $\mu$ we obtain

473: \begin{equation}

474: K_{L}^{CR} \approx \frac{8\pi^{2} q^{2}}{\omega} \int_{0}^{\infty}

475: \left\{\frac{-2v_{s}}{kv} - \left(\frac{3v_{s}\omega}{kv}+v\right)\frac{2\omega}{k^{2}v^{2}}

476: -\left(\frac{3v_{s}\omega}{kv}+v\right)\frac{\omega^{2}}{k^{3}v^{3}}

477: \ln \left\vert \frac{\omega-kv}{\omega+kv}\right\vert

478: \right\} \frac{\partial f_{0}}{\partial p}p^{2}dp.

479: \end{equation}

480: %

481: We then evaluate this integral in the two limiting cases away from the pole,

482: $\omega\gg kv$ and $\omega\ll kv$.

483: For $\omega\gg kv$;

484: \begin{equation}

485: \ln \left\vert \frac{\omega-kv}{\omega+kv}\right\vert

486: \simeq \frac{-2kv}{\omega} - \frac{2}{3}\left(\frac{kv}{\omega}\right)^{3}- ...

487: \end{equation}

488: leading to

489: \begin{equation}

490: K_{L}^{CR} \approx \frac{8\pi^{2} q^{2}}{\omega} \int_{0}^{\infty}

491: \left\{\frac{-2v_{s}}{kv} - \left(\frac{3v_{s}\omega}{kv}+v\right)\frac{2\omega}{k^{2}v^{2}}

492: +\left(\frac{3v_{s}\omega}{kv}+v\right)

493: \left( \frac{2\omega}{k^{2}v^{2}}+ \frac{2}{3\omega}\right)

494: \right\} \frac{\partial f_{0}}{\partial p}p^{2}dp .

495: \end{equation}

496: %The only term that does not cancel out with another term is

497: All terms in brackets cancel save for one, giving

498: % modified equation below to K_L = ....

499: % Merged equations 16 and 17:

500: \begin{equation}

501: K_{L}^{CR} \approx \frac{8\pi^{2} q^{2}}{\omega} \int_{0}^{\infty}

502: \frac{2v}{3\omega}

503: \frac{\partial f_{0}}{\partial p}p^{2}dp\,\,=\,\,\frac{16\pi^{2} q^{2}}{3 \omega^{2}}

504: \left\{ \left[vp^{2}f_{0}\right]_0^{\infty}

505: -\int_{0}^{\infty} 3f_{0}p^{2}\frac{dp}{\gamma m}

506: \right\}

507: \end{equation}

508: %Integrating by parts,

509: %\begin{equation}

510: %=\frac{16\pi^{2} q^{2}}{3 \omega^{2}}

511: %\left\{ \left[vp^{2}f_{0}\right]_0^{\infty}

512: %-\int_{0}^{\infty} 3f_{0}p^{2}\frac{dp}{\gamma m}

513: %\right\}

514: %\end{equation}

515: %

516: the first term goes to zero

517: so long as $f_{0}(\infty)\rightarrow 0$ faster than $p^{-3}$,

518: and the second term is $\propto n_{CR}/(\gamma m)$ leaving

519: \begin{equation}

520: K_{L}^{CR} \approx - \omega_{pCR}^{2}/\omega^{2}.

521: \end{equation}

522: and

523: \begin{equation}

524: \omega ^2\left(1+{\omega _{pe}^2\over\Omega _e^2}\sin ^2\theta\right)-\omega _{pi}^2

525: -\omega _{pe}^2\cos ^2\theta -\omega _{pCR}^2=0.

526: \end{equation}

527: This simply modifies the $1/\omega^{2}$ term in the dielectric tensor,

528: changing the frequency of the solution but not creating any complex roots,

529: hence no instability is generated.

530: 

531: Likewise, in the case where $\omega \ll kv$;

532: \begin{equation}

533: \ln \left\vert \frac{\omega-kv}{\omega+kv}\right\vert

534: \simeq -\frac{2\omega}{kv} -...

535: \end{equation}

536: leading to

537: \begin{equation}

538: K_{L}^{CR} \approx \frac{8\pi^{2} q^{2}}{\omega} \int_{0}^{\infty}

539: \left\{\frac{-2v_{s}}{kv} - \left(\frac{3v_{s}\omega}{kv}+v\right)\frac{2\omega}{k^{2}v^{2}}

540: +\left(\frac{3v_{s}\omega}{kv}+v\right)

541: \frac{2\omega^{3}}{k^{4}v^{4}}+...

542: \right\} \frac{\partial f_{0}}{\partial p}p^{2}dp .

543: \end{equation}

544: %

545: All but the first term are negligible in this limit, hence

546: \begin{equation}

547: K_{L}^{CR} \approx \frac{16\pi^{2} q^{2}v_{s}}{\omega k} \int_{0}^{\infty}

548: \frac{1}{v} \frac{\partial f_{0}}{\partial p}p^{2}dp .

549: \end{equation}

550: This leads to a full dispersion relation that can be written as

551: \begin{equation}

552: \omega^{2}\left(1+\frac{\omega_{pe}^{2}}{\Omega_e^{2}}\sin^{2}\theta\right) -

553: \omega \frac{4\pi q^{2}v_{s}}{k} 4\pi \int_{0}^{\infty}

554: \frac{1}{v} \frac{\partial f_{0}}{\partial p}p^{2}dp

555: -\left(\omega_{pi}^{2}+ \omega_{pe}^{2}\cos^{2}\theta \right) = 0

556: \end{equation}

557: which also lacks complex roots, regardless of the actual evaluation

558: of the integral.

559: 

560: Higher order terms in the expansion of $f=f_0\left(1+\mu\left(3v_s/v\right)

561: +\mu ^2\left(3v_s/v\right)^2/2+\mu^3\left(3v_s/v\right)^3/6+ ...\right)$ give rise to

562: higher order terms in

563: $\omega$. For $\omega >> kv$, the dispersion relation equation 18 becomes to next

564: highest order

565: \begin{eqnarray}

566: \omega ^3\left(1+{\omega _{pe}^2\over\Omega _e^2}\sin ^2\theta\right)&-\left(\omega _{pi}^2

567: +\omega _{pe}^2\cos ^2\theta +\omega _{pCR}^2-{72\over 5}\pi ^2q^2v_s^2\int{1\over v}

568: {\partial f_{0}\over\partial p}p^2dp\right)\omega\cr &+

569: {72\over 7}\pi ^2q^2kv_s^3\int{1\over v}

570: {\partial f_{0}\over\partial p}p^2dp=0,

571: \end{eqnarray}

572: which is stable since the terms $\propto \int{1\over v}

573: {\partial f_{CR}\over\partial p}p^2dp$ are of order $\sim \omega _{pCR}^2v_s^2/v^2

574: % replaced reference to equation 23 with 22

575: << \omega _{pCR}^2$. When $\omega << kv$ the dispersion relation Equation (22) takes

576: on the next highest order terms

577: \begin{equation}

578: -{24\pi ^2q^2v_s^2\over k^4}\int {1\over v^5}{\partial f_0\over\partial p}p^2dp

579: \left(3\omega ^4 +\omega ^3kv_s\right) -{24\pi ^2q^2v_s^2\over k^2}

580: \int {1\over v^3}{\partial f_0\over\partial p}p^2dp

581: \left(\omega ^2 +{3\over 5}\omega kv_s\right)

582: \end{equation}

583: to become a quartic equation. This has four real solutions so long as

584: $\int {1\over v^5}{\partial f_0\over\partial p}p^2dp > 0$. In fact if

585: $\int {1\over v^5}{\partial f_0\over\partial p}p^2dp < 0$, the addition of higher order

586: terms in the expansion of the cosmic ray distribution function would dramatically

587: alter the character of the solutions, a situation that must be considered unphysical.

588: We require $\int {1\over v^5}{\partial f_0\over\partial p}p^2dp > 0$, which means

589: that at low momenta, $\partial f_0/\partial p > 0$, and the cosmic ray distribution

590: cannot be monatonically decreasing from $v=p=0$. Our use of the kappa distribution in

591: the preceding section may therefore be questioned. However the resonance at $p_x=-2mv_s$

592: places it well into the region of the distribution where the gradient is

593: negative, and so modifications to the low momentum behavior would have very little

594: effect on our result. However this observation does imply that the distribution of

595: particles obeying a diffusion equation ahead of the shock is unlikely to extend down to zero

596: momentum.

597: %, and that

598: Some natural break must exist between the quasi-thermal population

599: gyrating around field lines and the cosmic rays diffusing in

600: turbulence. The forgoing also neglects the cosmic ray induced current in

601: the background plasma. The inclusion of such effects leads to the

602: modification $\omega _{pi}^2 \rightarrow\omega _{pi}^2-\omega

603: _{pCR}^2$, and has no effect on reactive instabilities.

604: 

605: 

606: \section{Discussion}

607: \subsection{Electron Heating or Magnetic Field Amplification?}

608: We have calculated the growth rate for waves that damp by heating electrons, in a

609: cosmic ray shock precursor using similar approximations and techniques to those employed

610: by \citet{Bell2004}.

611: Both the lower hybrid wave heating of electrons and the growth of magnetic field through

612: modified Alfv\'{e}n waves redistribute energy within the cosmic ray precursor. An important

613: question is which of

614: these is more effective, i.e. which grows faster? Above we derived a critical Alfv\'en

615: Mach number, $M_A\simeq 3n_{CR}^{\prime}/n_{CR}$, which divides the regime of magnetic

616: field amplification from that of lower hybrid wave growth. Following from the

617: treatment of the reactive instability above, we estimate $n_{CR}=\int _{p_{inj}}^{\infty}

618: f_{CR}4\pi p^2dp\simeq \left(6/\pi\right) n_{CR}^{\prime}v_s/v_{inj}$

619: where $f_{CR}$ is given by equation 1 and $p_{inj}$ is the injection momentum where

620: particles may begin to participate in a diffusive shock acceleration process. The

621: approximate result $n_{CR}^{\prime}/n_{CR}\simeq v_{inj}/2v_s$, gives

622: $M_A\simeq 1.5v_{inj}/v_s$ as the critical Alfv\'en Mach number.

623: 

624: % added a sentence to beginning of this paragraph:

625: %We can also compute the Alfv\'{e}nic Mach number where lower hybrid wave growth takes

626: %over from the growth of modified Alfv\'{e}n waves.

627: The next step in determining the critical Alfv\'en Mach number is to find an appropriate

628: $v_{inj}$ for the injection of seed particles into the cosmic ray acceleration process.

629: \citet{Zank2006} argue that quasi-perpendicular shocks have similar injection requirements

630: to quasi-parallel shocks, but that highly perpendicular shocks require much higher

631: injection energies. In the case of nonresonant magnetic field generation, we consider

632: the case of a highly perpendicular shock since the generated magnetic field will be

633: perpendicular and much stronger than the initial magnetic field. \citet{Zank2006}

634: give the injection velocity as

635: \begin{equation}

636: v_{inj}=3v_s\left[{1\over\left(r-1\right)^2}+{D_{Bohm}^2\over D_{\perp}^2}\right]^{1/2}

637: \end{equation}

638: where $r$ is the shock compression ratio, and $D_{Bohm}$ and $D_{\perp}$ are

639: the cosmic ray diffusion coefficients in the Bohm limit and in the perpendicular

640: direction respectively. \citet{reville08} give $D_{Bohm}/D_{\perp}\simeq 3$

641: for cosmic rays where $kr_g\sim 1$, and so $v_{inj}\simeq 10 v_s$. Thus the Alfv\'en Mach

642: number at which lower hybrid wave growth takes over from magnetic field amplification

643: should be about 15, unless the magnetic field saturates at a lower value (i.e. higher

644: $M_A$) before this is reached.

645: 

646: The growth of lower hybrid waves is most efficient at a quasi-perpendicular shock, whereas

647: the growth of magnetic field through modified Alfv\'en waves is strongest at a quasi-parallel

648: shock \citep{Bell2005}. This apparent contradiction is actually easily resolved.

649: At an initially quasi-parallel shock, \citet{Bell2005},

650: \citet{reville08} and \citet{zirakashvili08} show

651: that a highly helical magnetic field develops.

652: The distortion of an initially parallel

653: field line is shown schematically in Figure 1, showing the evolution of the shock from

654: quasi-parallel to quasi-perpendicular. A similar schematic in Figure 2 shows the

655: evolution of an initially quasi-perpendicular shock, where magnetic field is amplified

656: orthogonal to the pre-existing magnetic field, but where the shock remains quasi-perpendicular.

657: In both cases a perpendicular field is generated, thus allowing lower hybrid wave growth and electron heating

658: in a region close to the shock as indicated.

659: 

660: Another potential problem is the cavities seen in simulations of the

661: growth of modified Alfv\'{e}n waves \citep[e.g.][]{Bell2005}. The

662: helical field from an initially quasi-parallel geometry naturally

663: creates a filamentary structure, dragging the thermal plasma with

664: it, while cosmic rays tend to accumulate in the low density

665: cavities. This is problematic for our mechanism that requires

666: spatial coincidence between cosmic rays, magnetic field and thermal

667: plasma. A possible solution is that the growth of lower hybrid waves

668: takes over from the growth of modified Alfv\'en waves, so that the

669: cosmic ray driven magnetic field never reaches its final saturated

670: state. \citet{Bell2004}, \citet{reville08} and

671: \citet{zirakashvili08} derive a saturation magnetic field by setting

672: $\gamma _{B,nonres}=0$ in equation (9) to give $\delta B\sim

673: jr_g/4\pi$ or $\delta B^2/8\pi\sim n_{CR}m_iv_sv_{inj}/2$. This

674: gives an Alfv\'en Mach number at saturation of $M_A^2\sim

675: n_i/10n_{CR}$, (assuming $v_{inj}\sim 10 v_s$), which for likely

676: parameters $n_i/n_{CR}\sim 10^3$ gives a value of $M_A$ of similar

677: magnitude but possibly lower than that where the electron heating is

678: expected to take over. Bearing in mind that we took the strongest

679: growth rate for magnetic field amplification to estimate where

680: electron heating takes over, it is quite plausible that the

681: amplified magnetic field never reaches saturation. Also, as the

682: initial shock state becomes more quasi-perpendicular, the growth

683: rate slows down, and the circularly polarized Alfv\'en waves become

684: elliptically polarized, ultimately becoming linearly polarized in

685: the limit of a true perpendicular shock, eliminating the growth of

686: such cavities.

687: 

688: \citet{Pelletier2006} find that the nonresonant instability of

689: \citet{Bell2004} and \citet{Bell2005} dominates over the more familiar resonant instability

690: when the shock velocity $v_s$ is greater than a few times $\epsilon _{CR}c$ where

691: $\epsilon _{CR}$ is the ratio of the cosmic ray energy density to the kinetic energy

692: density of the shock. \citet{niemiec08} simulated the cosmic

693: ray driven amplification of magnetic field in a parallel shock using Particle-In-Cell

694: simulations, which can naturally account for the backreaction of the

695: %magnetic field generation

696: generated magnetic field

697: on the cosmic ray current. In conditions where the nonresonant mode should grow,

698: %they in fact find growth of the resonant mode, and

699: they find magnetic field amplification only to $\delta B\sim B$. The magnetic field

700: again produces filaments, but they do not find

701: %that cosmic rays naturally accumulate

702: cosmic ray accumulation

703: in the filament

704: cavities. They do not find strong growth, and argue that

705: saturation occurs because the incoming flow to the shock is decelerated by the

706: cosmic rays, reducing their relative velocity and hence the cosmic ray current.

707: 

708: For our electron heating model, the precise degree of magnetic field

709: amplification is unimportant so long as the cosmic ray diffusion

710: coefficient remains proportional to $1/B$. It is only necessary that

711: the shock be sufficiently quasi-perpendicular to allow cosmic rays

712: to generate lower hybrid waves. A reduced cosmic ray current does

713: not necessarily produce an appreciable affect on the kinetic growth

714: rate for lower hybrid waves. So long as the current does not vanish,

715: the initial effect of reducing $v_s$ in equation 6 is to bring more

716: cosmic rays into resonance with the lower hybrid waves. Another

717: estimate of the cosmic ray density necessary to heat electrons may

718: come from the long wavelength limit of the magnetic field

719: amplification, when $\gamma _B =\sqrt{{\bf k}\cdot {\bf

720: B}n_{CR}qv_s/n_im_i}$, for both parallel and perpendicular cases

721: \citep{Bell2005}. Electron heating then requires $\gamma\sim\Omega

722: _in_{CR}/n_i >\sqrt{n_{CR}v_s\cos\phi /n_iv_{inj}}\Omega _i$, taking

723: $k=\Omega _i/v_{inj}$ (probably an overestimate), yielding

724: $n_{CR}/n_i > \cos\phi /10$. At $\cos\phi\le\sqrt{m_e/m_i}$,

725: the values typical for lower hybrid wave propagation, the value

726: for $n_{CR}/n_i$ is low enough (0.001 - 0.01) to make electron heating

727: by cosmic rays plausible.

728: 

729: \subsection{Other Electron Heating Mechanisms}

730: Several other researchers have considered the generation of waves in a shock precursor

731: as a means of heating electrons. \citet{ohira07} and \citet{shimada00} have both

732: considered the model of \citet{Cargill88} in more detail, using Particle-In-Cell

733: codes rather than a hybrid approach. Other references

734: \citep{dieckmann00, mcclements01,schmitz02} focus more on the

735: electron injection problem for diffusive shock acceleration, rather than the thermal

736: electron temperature, again invoking various wave modes in a reflected ion precursor.

737: Our principle departure from these works has been to treat similar wave modes upstream

738: of the shock, but excited by cosmic rays undergoing diffusive

739: shock acceleration rather than by quasi-thermal ions reflected from the shock. This

740: allows electron heating to occur over a much more extended upstream region dictated

741: by the cosmic ray diffusion coefficient, $D$, rather than the ion gyroradius.  In addition,

742: expressing the thickness of this region as $l\sim D/v_s$ naturally results in electron

743: heating that is essentially independent of shock speed, as argued from observations of

744: Balmer-dominated shocks in \citet{GLR07}. On the other hand if the thickness of the electron

745: heating region is comparable to the ion gyroradius then $l\propto v_s$

746: and $T_e\,\propto\,v_s^2$.

747: This results in constant $T_e/T_i$ with $v_s$,

748: contrary to what is observed.

749: 

750: A number of other authors have investigated the role of the cross-shock potential

751: in heating the electrons. Inside the (quasi-perpendicular) shock ramp, the magnetic field

752: may ``overshoot'', i.e. increase to a value greatly in excess of its asymptotic downstream

753: strength before decreasing again. The electric field arising from the small charge

754: separation associated with this magnetic field

755: gradient, $E\simeq \partial/\partial x \left(B^2\right)/(8\pi e n_i)$,

756: can decelerate ions and accelerate

757: electrons. Such effects are known to be important at low Mach number shocks where a

758: laminar approximation holds \citep[e.g.][]{scudder86}. At higher Mach numbers, where the

759: shock is

760: %nonstationary,

761: turbulent, the importance of such electric fields is less clear. Electron ${\bf E}\times {\bf B}$ drift along the shock

762: front will result in periods of energy loss as well as energy gain by the cross-shock

763: potential, and hence no net heating. It has been argued \citep{gedalin07} that in certain cases the

764: shock front may be sufficiently thin (length scales of order $c/\omega _{pe}$) that

765: the electrons are effectively demagnetized. One might expect to see electron heating

766: {\em increase} with shock velocity (or $M_A$) once this condition becomes satisfied.

767: Examination of solar wind shocks suggests that such thin shocks are rare at best and certainly not ubiquitous.

768: %Considering the

769: %magnetic field amplification above, the Alfv\'en Mach numbers for SNR shocks are similar to

770: %those for the solar wind, and so we should expect similar uncertainty with regard to

771: %the role of the cross-shock potential in heating electrons.

772: We see no evidence for an increase in electron heating in SNRs up to shock velocities of 6000 km s$^{-1}$

773: \citep[1E 0102.2-7219;][]{hughes00}, and possibly up

774: to 20,000 km s$^{-1}$ \citep[SN 1993J;][]{fransson96}. At higher Mach numbers such as those

775: expected in gamma ray burst afterglows the convective electron gyroradius may easily reduce to

776: less than the electron inertial length, making the cross shock potential a candidate

777: heating electron heating mechanism.

778: 

779: \citet{Schwartz88} have made a survey of a number of solar wind shock crossings observed

780: {\it in situ}. They find $T_e/T_i\propto 1/M_A$ for $M_A$ greater than about 2-3. At lower $M_A$,

781: there is a wide scatter in $T_e/T_i$ about $T_e/T_i\sim 1$. At these slower shocks,

782: $T_e$ correlates very well with the change in ion velocity squared, suggesting that both

783: are due to the same mechanism, presumably the cross shock potential. The

784: % break in behavior

785: switch to $T_e/T_i\propto 1/M_A$

786: at $M_A\sim 2-3$ is possibly due to the onset of

787: %shock nonstationarity

788: turbulent shock structure

789: %for Mach numbers beyond this.

790: at higher Mach numbers.

791: 

792: We can explore the conditions required for the validity of the laminar approximation

793: by adopting the criterion of \citet{tidman71} for the existence of a magnetosonic soliton;

794: \begin{equation}

795: M_S^2/\left(M_S^2-1\right)<M_A^2<4M_S^2\left(M_S^2-4M_S+3+2\ln M_S\right)/

796: \left(M_S^2-1-2\ln M_s\right)^2.

797: \end{equation}

798: The relationship between $M_A$ and $M_S$ predicted by this relation is plotted in

799: Figure 3.  The criterion above indicates that the laminar approximation breaks down

800: at slightly lower Mach numbers in the solar wind shocks than indicated by the behavior of $T_e/T_i$ in

801: \citet{Schwartz88}.  Magnetic field amplification by about an order of magnitude in SNR

802: shocks for the 400 km s$^{-1}$ shocks observed in the Cygnus Loop, where

803: \citet{GLR07} find complete electron-ion equilibration, would bring $M_A$ down to the

804: same range as indicated by \citet{Schwartz88}, possibly suggesting that the cross-shock

805: potential is at work for the lower velocity SNR shocks. At the higher velocity

806: shocks in the \citet{Schwartz88} sample, which are all perpendicular, an empirical relationship

807: $T_e/T_i\propto 1/M_A$ emerges.

808: Such a behavior can be consistent with our model if we make the assumption that

809: the cosmic rays accompanying solar wind shocks are

810: %nonrelativistic.

811: non-relativistic, suprathermal particles.

812: Then the

813: diffusion coefficients take on an extra factor $v_s/c$, assuming that the cosmic

814: ray velocity is proportional to the shock velocity. This extra power of the shock

815: velocity in the diffusion coefficient results in $T_e\propto v_s$ and $T_e/T_i\propto 1/v_s$.

816: 

817: %\subsection{Quantitative Considerations}

818: \subsection{Heating versus Damping of Lower Hybrid Waves and the Width of the Precursor}

819: 

820: We have discussed whether growth of lower hybrid waves may compete with cosmic ray induced

821: magnetic field amplification, but have not yet discussed whether

822: this growth rate is sufficient to balance the damping rate of lower hybrid waves by electrons.

823: To answer this question we compare the electron heating rate from diffusive scattering off the

824: lower hybrid waves with the energy input into the lower hybrid waves from the cosmic ray

825: turbulence.

826: The electron heating rate per unit area of shock is $n_{e}f_R m_e\varkappa _{\|\|}tv_s/2$, where

827: $f_R\simeq \exp\left(-\omega^2/ 2k_{\|}^2 v_{te\|}^2\right)\simeq \exp\left(-2v_s^2m_i/v_{te}^2m_e

828: \right)$ is the fraction of electrons in resonance

829: with the lower hybrid waves (using $\omega /k\simeq 2v_s$),

830: $\varkappa _{\|\|}$ is the parallel electron velocity diffusion coefficient in lower hybrid

831: turbulence, and $t$ is the period of time spent by an electron in the turbulence. This time

832: $t=l/v_s$ where $l$ is the precursor depth. The electron heating is balanced by

833: energy input to the turbulence by cosmic rays with rate $2\gamma E_{turb}l$.

834: Putting $E_{turb}=\left(\delta E^2/8\pi\right)^2\omega _{pe}^2/\Omega _e^2$ and

835: $\varkappa _{\|\|}= q^2\delta E^2k_{\|}^2/4m_e^2k_{\perp}^2\omega $ we deduce a growth rate

836: $\gamma =f_R q^2B^2/16m_im_e\omega c^2 =f_R \omega /16$.

837: The kinetic growth rate derived earlier in Equation (7), 

838: in units of $\Omega_{i}$, is proportional to $n_{CR}/n_i$.  

839: Thus as long as this ratio is comparable to or larger than the 

840: fraction of electrons that are in resonance with the lower hybrid 

841: waves the growth rate outlined above will be sufficient to heat the

842: electrons.

843: 

844: Another constraint on $n_{CR}/n_i$ comes from Equation (9). For

845: magnetic field amplification, we require $n_{CR}/n_i >

846: k_{\|}v_A^2/v_s/\Omega _i$. Since $k_{\|}

847: > 1/r_g$, the gyroradius of cosmic rays at injection, $n_{CR}/n_i >

848: v_A^2/v_{inj}v_s\sim 1/10M_A^2\sim 10^{-3}$ for $M_A\sim 10$ and

849: $v_{inj}\sim 10 v_s$. Taking the maximum growth rate estimated from

850: Equation (9), $\gamma = M_A\Omega_in_{CR}/2n_i\sim\Omega _i/20M_A$,

851: where $l_i=c/\omega _{pi}=v_A/\Omega _i$ is the ion inertial length,

852: we estimate a characteristic length of $v_s/\gamma = 20M_A^2l_i\sim

853: 5\times 10^{10}$~cm.  This requires a cosmic ray diffusion

854: coefficient of order $10^{19}$~cm$^2$~s$^{-1}$. This is considerably

855: smaller than the estimate by \citet{Bell2004}. Taking a

856: characteristic cosmic ray energy of $10^{15}$~eV, \citet{Bell2004}

857: finds a typical growth time for magnetic field of order 100 years.

858: This would yield a characteristic length scale for magnetic field

859: amplification of $\sim 10^{18}$~cm for a 3000 km s$^{-1}$ shock,

860: requiring a cosmic ray diffusion coefficient of $\sim 3\times

861: 10^{26}$~cm$^2$~s$^{-1}$. We suspect that our simple estimate

862: reflects the growth rate while the shock may be considered

863: quasi-parallel, and that magnetic field amplification slows down

864: considerably once it becomes quasi-perpendicular. Therefore in

865: taking a characteristic cosmic ray energy of $10^{15}$~eV,

866: \citet{Bell2004} is taking the lowest energy cosmic rays for which

867: the shock may be considered quasi-parallel, and this result may be

868: considered more realistic.

869: 

870: Further, in \citet{GLR07} we argued that the depth of cosmic ray precursor

871: over which electron heating occurs could not be larger than $\sim 10^8v_s/n_e$ cm, otherwise

872: neutral hydrogen would not survive to encounter the shock front. We suggest here

873: that lower hybrid

874: waves accelerate the small fraction of electrons that happen to be in resonance, and that

875: these accelerated electrons communicate their energy to the rest of the thermal population

876: by Coulomb collisions, with characteristic time scale

877: $10^{10}\left(T/10^8 {\rm K}\right)^{3/2}/n_e$  s.

878: Equating this to $10^8/n_e$ s yields a maximum temperature of

879: $T\sim 10^8\times\left(10^{-2}\right)^{2/3}\simeq 5\times 10^6$ K. This is very close

880: to the temperature found in \citet{GLR07}, 0.3 keV, or $3.5\times 10^6$ K. Put another

881: way, the temperature found in \citet{GLR07} is consistent with electron heating such

882: that neutral hydrogen can survive to encounter the shock front proper. However

883: cosmic ray precursors at the small end of the range considered above ($\sim 10^{11}$~cm)

884: would not allow any significant electron collisional equilibration to occur.

885: Allowing for compressional heating of the electrons as they go through the shock,

886: a precursor electron temperature of order $10^6$~K requires a precursor length of

887: $\sim 10^7v_s/n_e\sim 10^{15}\left(v_s/1000 {\rm ~km~s}^{-1}\right)$~cm, or a

888: minimum cosmic ray diffusion coefficient of $D\sim 10^{23}

889: \left(v_s/1000 {\rm ~km~s}^{-1}\right)^2$~cm$^2$s$^{-1}$. 

890: % The corresponding growth rate requires a maximum 

891: % cosmic ray energy of order $10^{12}$ eV.

892: 

893: The electric field in the lower hybrid waves will be given by the limit derived

894: by \citet{karney78},

895: \begin{equation}

896: \delta E=B\left(\Omega _i\over\omega\right)^{1/3}{\omega\over 4k_{\perp}c}=

897: B\left(\Omega _i\over\omega\right)^{1/3}{v_s\over 2k_{\perp}c}.

898: \end{equation}

899: This is the maximum electric field before ion trapping and heating occurs. \citet{laming07}

900: and references cited therein demonstrate that when $\omega /\sqrt{2}k_{\|}v_{te} <<

901: \omega/\sqrt{2}k_{\perp}v_{ti}$, ions are heated more effectively than electrons above this

902: threshold. In our case, the ions that are heated will be the lower energy part of the

903: suprathermal ion distribution reflected from the shock, i.e. those below the injection

904: threshold for diffusive shock acceleration in Equation (1) or any of its modifications

905: subsequent to the treatment of the reactive lower hybrid wave instability in section 2.2.

906: With the wave electric field given by Equation (27), the electron momentum diffusion

907: coefficient in lower hybrid turbulence varies as $v_s^2$, yielding a constant degree

908: of heating with shock velocity if the time spent in the turbulence varies as $1/v_s^2$,

909: which would be the case if the cosmic rays are obeying a diffusion law.

910: 

911: \section{Summary}

912: We have considered in more detail the speculation of \citet{GLR07} that lower hybrid

913: waves generated in a cosmic ray precursor could be responsible for the electron

914: heating at collisionless shocks in supernova remnants.

915: We find that there do exist growing modes for the resonant or kinetic case, and that

916: the growth rate in this case may be sufficient both to survive the damping by electrons

917: and to compete with magnetic field amplification by modified Alfv\'en waves.

918: Below a certain Alfv\'en Mach number (roughly estimated to be $\sim 15$)

919: the lower hybrid wave growth rate exceeds that of the modified Alfv\'{e}n waves.

920: The modified Alfv\'en wave generation exists for all magnetic field orientations with

921: respect to the shock, but is most effective for quasi-parallel case and always generates new perpendicular field.

922: Lower hybrid waves, on the other hand, require quasi-perpendicular field geometry in order to grow.

923: Thus a schematic picture emerges in which far ahead

924: of the high Mach number shock, modified Alfv\'en waves generate perpendicular field, reducing the effective Mach

925: number closer to the shock front and thus allowing lower hybrid wave growth to occur in a short

926: region before the shock and heat the resonant electrons.

927: A critical Alfv\'en Mach number around 15 suggests magnetic field amplification by about

928: an order of magnitude, similar to what a comparison of the surveys of \citet{GLR07} and

929: \citet{Schwartz88} would suggest, taking in both cases the shock velocity where \tetp $\sim$1

930: starts to break down as that where the laminar shock approximation ceases to hold.

931: 

932: 

933: We have concentrated on the generation of lower hybrid waves, since

934: for these the group velocity can be equal to the shock velocity itself, meaning that

935: the waves can stay in contact with the shock for long time intervals and in principle

936: grow to large amplitudes. However other wave modes that heat electrons are certainly

937: possible, and these, such as the Landau damping of kinetic Alfv\'en waves

938: \citep[e.g.][]{vinas00}, do not

939: require perpendicular shocks as lower hybrid waves do. In fact, \citet{bykov} studied the

940: generic case of heating by turbulent modes in the shock precursor and did identify an area

941: of parameter space for which a near inverse-square relationship between \tetp and shock

942: velocity could be accommodated.

943: %

944: Our model requires that cosmic ray ions be essentially ubiquitous at 

945: SNR shocks, with number densities estimated by various means in section 3.

946: %

947: In a wider context, the idea that cosmic rays are responsible for electron heating

948: at fast shocks reinforces the idea that cosmic rays are an intrinsic component of the

949: collisionless shock phenomenon.

950: 

951: 

952: 

953: \acknowledgments

954: J.M.L. and C.E.R. have been supported by NASA contract NNH06AD66I (LTSA Program)

955: and by basic research funds of the Office of Naval Research. P.G. acknowledges support

956: from NASA contract NAS8-03060. We also appreciate the continuing advice and

957: encouragement of Dr. Jill Dahlburg.

958: 

959: 

960: \appendix

961: 

962: \section{Cosmic Ray Diffusion Coefficients}

963: The parallel spatial cosmic ray diffusion coefficient is most easily obtained from

964: its relation to the pitch angle scattering diffusion coefficient in momentum space.

965: The diffusion coefficient in momentum space is expressed most generally as

966: \citet{melrose86},

967: \begin{equation}

968: D_{\lambda\mu} = \sum_{s=-\infty}^{\infty} \int

969: \frac{8\pi ^2q^{2}}{\hbar}

970: \frac{R_{M}({\bf k})}{\omega_{M}({\bf k})}

971: \vert {\bf e} \cdot {\bf v}({\bf k},{\bf p}, s)\vert ^{2}

972: \delta (\omega_{M} - s\Omega - k_{\|}v_{\|})\triangle\lambda\triangle\mu

973: N_{\mu}({\bf k})\frac{d^{3}k}{(2\pi)^{3}}

974: \end{equation}

975: where $N_{\mu}$ is the number density of wave quanta, $R_M$ is the ratio of electric

976: energy to total energy in the wave, such that $R_M\int N_{\mu}\hbar\omega _Md^3k/\left(2\pi\right)^3

977: =\delta E^2/8\pi $, ${\bf e}$ is the wave polarization vector and ${\bf v}$ is the cosmic ray

978: velocity. For pitch angle scattering by parallel propagating Alfv\'{e}n waves,

979: $\lambda=\alpha$, and so

980: \begin{equation}

981: \triangle\lambda =

982: \hbar\left(\frac{s\Omega}{v_{\bot}}\frac{\partial}{\partial

983: p_{\bot}} + k_{\|}\frac{\partial}{\partial

984: p_{\|}}\right)\lambda =-{\hbar k_{\|}\over

985: p\sin\alpha} .

986: \end{equation}

987: With $\omega_{M} = k_{\|}v_A$, $R_{M} = (v_{\|}^{2})/(2c^{2})$, and

988: ${\bf e}\cdot{\bf v}= v_{\bot}/2$,

989: \begin{eqnarray}

990: \nonumber D_{\alpha\alpha} & = & \int \frac{8 \pi ^2

991: q^{2}}{\hbar\omega}\frac{v_A^{2}}{2c^{2}}

992: \frac{v^{2}\sin^{2}\alpha}{4} \delta (\omega_{M} - s\Omega -

993: k_{\|}v_{\|}) \frac{\hbar^{2}k_{\|}^{2}}{p^{2}\sin^{2}\alpha}

994: \frac{U_{M}({\bf k})}{\hbar \omega}

995: \frac{d^{3}k}{(2\pi)^{3}} \\

996: % & = & \int

997: %\frac{\pi ^2q^{2}k_{\|}^{2} v_{\|}^{2}}{\omega^{2}c^{2}}

998: %\frac{v^{2}}{p^{2}} U_{M}(k_{\|})

999: %\delta (\omega_{M} - s\Omega - k_{\|}v_{\|})

1000: %\frac{dk_{\|}}{2\pi} \\

1001: %& = &

1002: %\frac{\pi ^2q^{2}v_{\|}^{2}}{p^{2}}

1003: %\left(\frac{\Omega}{v_{\|}}\right)^{2}

1004: %\frac{v^{2}}{\omega^{2}c^{2}}

1005: %\frac{U_{M}(k_{\|}=\Omega/v_{\|})}{2\pi v_{\|}} \\

1006: & = & \frac{\pi ^2q^{2}v} {p^{2}c^2 \cos\alpha}

1007: \frac{U_{M}(k_{\|}=\Omega/v_{\|})}{2\pi}

1008: \end{eqnarray}

1009: where we have put, $s=1$ and taken $\omega _M << \Omega$.

1010: 

1011: We now express $D_{\|}$ in terms of $D_{\alpha\alpha}$ by writing

1012: \begin{equation}

1013: f(p,\alpha)=f_{0}(p)+f_{1}(p)\cos\alpha+\onehalf f_{2}(p)\cos^{2}\alpha + ...

1014: \end{equation}

1015: and substituting into the diffusion equation

1016: \begin{equation}

1017: \frac{\partial f}{\partial t} + v_{z}\frac{\partial f}{\partial z} =

1018: \frac{1}{\sin \alpha} \frac{\partial}{\partial \alpha}

1019: \left[\sin\alpha D_{\alpha\alpha} \frac{\partial f}{\partial \alpha} \right].

1020: \end{equation}

1021: %\begin{eqnarray}

1022: %+ \onehalf\cos^{2}\alpha\frac{\partial f_{2}}{\partial t}

1023: %+ v\cos\alpha\frac{\partial f_{0}}{\partial z}

1024: %+ v\cos^{2}\alpha\frac{\partial f_{1}}{\partial z}

1025: %+ \onehalf v\cos^{3}\alpha\frac{\partial f_{2}}{\partial z}

1026: %= & \nonumber \\

1027: %& \frac{\partial}{\partial (\cos \alpha)}

1028: %\left[\sin^{2}\alpha D f_{1} + \sin^{2}\alpha\cos\alpha D f_{2} \right] &

1029: %\end{eqnarray}

1030: %

1031: Upon integrating the result over $\cos\alpha$ we obtain, with $v_z=v\cos\alpha$,

1032: \begin{equation}

1033: \frac{\partial f_{0}}{\partial t} +\frac{1}{3}\frac{\partial f_{2}}{\partial t}

1034: + \frac{v}{3}\frac{\partial f_{1}}{\partial z} = 0.

1035: \end{equation}

1036: Multiplying each side by $\cos \alpha$ and then integrating over $\cos\alpha$ yields

1037: \begin{equation}

1038: \frac{2}{3}\frac{\partial f_{1}}{\partial t} +

1039: \frac{2v}{3}\frac{\partial f_{0}}{\partial z}+

1040: \frac{v}{5}\frac{\partial f_{2}}{\partial z} =- \int_{-1}^{1}\cos\alpha\sin^{2}\alpha Df_{2}d(\cos\alpha).

1041: \end{equation}

1042: With $f_0 >> f_1 >> f_2$,

1043: $f_{2} \simeq -

1044: \frac{2v}{3}\frac{\partial f_{0}}{\partial z}

1045: \left[ \int_{-1}^{1}\cos\alpha\sin^{2}\alpha Dd(\cos\alpha) \right]^{-1}$ which when substituted into

1046: equation A6 allows the identification

1047: \begin{equation}

1048: D_{\|}={2v^2\over 9}\left[\int _{-1}^{1}\cos\alpha\sin ^2\alpha D_{\alpha\alpha}d\left(\cos\alpha\right)

1049: \right]^{-1}.

1050: \end{equation}

1051: With equation A3,

1052: \begin{equation}

1053: D_{\|}={p^2c^2v\over 3\pi q^2 U_M\left(k_{\|}=\Omega

1054: /v_{\perp}\right)}.

1055: \end{equation}

1056: This is a factor of $2\pi$ larger than the equivalent expression

1057: given by \citet{blandford87}, due to a different definition of

1058: $U_M$. Where $U_M\propto k_{\|}^{-\beta }$, $D_{\|}\propto

1059: p^{2-\beta}$, which evaluates to $D_{\|}\propto vp^{1/3}$ or

1060: $D_{\|}\propto vp^{1/2}$ for Kolmogorov or Kraichnan turbulence

1061: respectively. If $v\sim c$, the dependence of $D_{\|}$ on $p$ can

1062: usually be neglected.

1063: 

1064: The perpendicular spatial cosmic ray diffusion coefficient has been

1065: given in terms of $D_{\|}$ by various authors. Based on numerical

1066: experiments, \citet{Marcowith2006} give $D_{\perp}=\eta

1067: ^{2+\epsilon}D_{\|}$ where $\eta =\delta B^2/\left(\delta

1068: B^2+\left<B\right>^2\right)$ and the cosmic ray distribution

1069: function $f\left(p\right)\propto p^{-4-\epsilon}$.

1070: \citet{Shalchi2007} and \citet{Zank2006} give

1071: $D_{\perp}\propto\left(\delta B^2/B_0^2

1072: \right)^{2/3}D_{\|}^{1/3}\left(l_{2D}v\right)^{2/3}$ from analytic

1073: considerations, where $l_{2D}$ is the 2D bendover length scale, the

1074: inverse of the wavenumber where the inertial range onsets, and

1075: consequently has even smaller dependence on the cosmic ray momentum

1076: than the parallel diffusion coefficient for relativistic cosmic rays, and has

1077: the same dependence in the nonrelativistic case.

1078: 

1079: 

1080: \section{Growth Rate for an Electromagnetic Instability}

1081: For completeness, we give here a treatment of the growth rate due to

1082: cosmic rays of electromagnetic waves with frequency in the lower

1083: hybrid range, and show that it is significantly smaller than either

1084: the electrostatic instability, or the growth of modified Alfv\'en

1085: waves. It is relatively easy to show that the reactive instability

1086: of \citet{Bell2004} has higher thresholds and lower growth rates as

1087: the frequency of the electromagnetic wave increases first above the

1088: proton gyrofrequency and then above the electron gyrofrequency. Here

1089: we concentrate on the kinetic instability that might generate

1090: electromagnetic waves in the lower hybrid range, whistlers, adapting

1091: the expressions in \citet{Bell2004} and \citet{Achterberg83};

1092: \begin{eqnarray}

1093: \omega^{2}(K^{T} -1) & =  & \frac{\Omega_{i}c^{2}}{\omega v_{A}^{2}}

1094: \left\{ \tilde{\omega}_{i}^{2}

1095: \mp \frac{k^{2}v_{ti}^{2}}{\Omega_{i}}\tilde{\omega}_{i}

1096: \mp \frac{\Omega_{i}}{n_{e}}k\left( \frac{J_{CR}}{q} - \frac{\omega}{k}N_{CR}

1097: \right) \right\}

1098:  \\

1099: & + & 4\pi q^{2} \int \frac{v_{\bot}/2}{\omega - k_{\|}v_{\|}}

1100: \left\{ (\omega-k_{\|}v_{\|} ) \frac{\partial f}{\partial p_{\bot}}

1101: + k_{\|}v_{\bot} \frac{\partial f}{\partial p_{\|}} \right\} 2\pi

1102: p_{\bot}dp_{\bot} dp_{\|} = c^{2}k^{2}-\omega ^2. \nonumber

1103: \end{eqnarray}

1104: %

1105: Evaluating the two cosmic ray terms,

1106: %

1107: \begin{equation}

1108: \int\int

1109: v_{\bot}\frac{\partial f}{\partial p_{\bot}}2\pi p_{\bot}dp_{\bot}dp_{\|}

1110: =\int_{-\infty}^{\infty}\left\{\left[2\pi p_{\bot}v_{\bot}f \right]_{0}^{\infty}

1111: -\int \frac{4\pi p_{\bot}}{\gamma m}f dp_{\bot}

1112: \right\} dp_{\|} = -\frac{2}{\gamma m}n_{CR}

1113: \end{equation}

1114: and

1115: \begin{eqnarray}

1116: \int\int \frac{v_{\bot}^{2}}{(\omega-k_{\|}v_{\|})}

1117: \frac{\partial f}{\partial p_{\|}}2\pi p_{\bot}dp_{\bot}dp_{\|}

1118: & =  & \int \int

1119: \frac{-p_{\bot}^{2}/(\gamma^{2}m^{2})}{(\omega-k_{\|}v_{\|})}

1120: \frac{(p_{\|}-mv_{s})/p_{t}^{2}}

1121: {\left[1+ ((p_{\|}-mv_{s})^{2}+ p_{\bot}^{2})/(2\kappa p_{t}^{2})\right]^{\kappa+1}}

1122: 2\pi p_{\bot}dp_{\bot}dp_{\|}

1123: \nonumber \\

1124: & = &  \int \int

1125: \frac{p_{\bot}(p_{\|}-mv_{s})/(\gamma^{2}m^{2})}{(\omega-k_{\|}v_{\|})}

1126: \frac{\partial f}{\partial p_{\bot}}2\pi p_{\bot}dp_{\bot}dp_{\|}

1127: \end{eqnarray}

1128: explicitly assuming a $\kappa$ distribution for CR in Equation (B3).

1129: %

1130: After some algebraic manipulation, an integration by parts,

1131: rewriting $f$ as a $\kappa$ distribution and making the

1132: substitution $p_{\bot}^{2} = P, dP = 2p_{\bot}dp_{\bot}$ (B3)

1133: can be written as

1134: %

1135: \begin{equation}

1136: - \int \frac{4\pi(p_{\|}-mv_{s})}{\omega-k_{\|}v_{\|}}

1137: \int \left(\frac{2\kappa p_{t}^{2}}

1138:           {2\kappa p_{t}^{2}+(p_{\|}-mv_{s})^{2}+ P}\right)^{\kappa}

1139: \frac{m^{2}c^{4}+p_{\|}^{2}c^{2}}{(m^{2}c^{2}+p_{\|}^{2}+P)^{2}}

1140: dPdp_{\|}.

1141: \end{equation}

1142: %

1143: The integral over $dP$ can be evaluated using a hypergeometic

1144: function (Gradshteyn \& Ryzhik 1965; 3.197.1) to give

1145: \begin{eqnarray}

1146:  & - \int \frac{4\pi(p_{\|}-mv_{s})}{\omega-k_{\|}v_{\|}}

1147: \frac{(2\kappa p_{t}^{2})^{\kappa}}{(m^{2}+p_{\|}^{2}/c^{2})}

1148: \frac{{\rm B}(1,1+\kappa)}

1149:           {(2\kappa p_{t}^{2}+(p_{\|}-mv_{s})^{2})^{\kappa -1}}

1150: \nonumber \\

1151:  \times &

1152: %{\rm B}(1,1+\kappa)

1153:  _{2}{\rm F}_{1}\left(2,1;2+\kappa;1-\frac{2\kappa p_{t}^{2}+(p_{\|}-mv_{s})^{2}}{m^{2}c^{2}+p_{\|}^{2}}\right)

1154: dp_{\|}

1155: \end{eqnarray}

1156: where ${\rm B}(1,1+\kappa )$ is the beta function. Considering only

1157: the kinetic case, using the Landau prescription for this integral

1158: with the pole at $\omega -k_{\|}v_{\|}$, the imaginary part (i.e.

1159: the portion relevant for the growth rate) is

1160: %\pagebreak

1161: \begin{eqnarray}

1162: {\rm Im}\left(

1163: \int\int \frac{v_{\bot}^{2}}{(\omega-k_{\|}v_{\|})}

1164: \frac{\partial f}{\partial p_{\|}}2\pi p_{\bot}dp_{\bot}dp_{\|} \right)

1165: & = &

1166: 4i \pi^{2} \left(\frac{\gamma m \omega}{k_{\|}} - mv_{s} \right)

1167: \frac{(2\kappa p_{t}^{2})^{\kappa}}

1168:       {m^{2}+ \gamma^{2}m^{2}\omega^{2}/(k_{\|}^{2}c^{2})}

1169: \frac{m}{k_{\|}}

1170: \nonumber \\

1171: & \times &

1172: \frac{{\rm B}(1,1+\kappa)}

1173:       {(2\kappa p_{t}^{2}+(\gamma m \omega/k_{\|}-mv_{s})^{2})^{\kappa -1}}

1174: \nonumber \\

1175: &  \times &

1176:  _{2}{\rm F}_{1}\left(2,1;2+\kappa;

1177: 1-\frac{2\kappa p_{t}^{2}+(\gamma m \omega/k_{\|}-mv_{s})^{2}}

1178: {m^{2}c^{2}+\gamma^{2}m^{2}\omega^{2}/k_{\|}^{2}}\right)

1179: .

1180: \end{eqnarray}

1181: Setting $\omega\rightarrow \omega + i\gamma_{g}$ in the dispersion

1182: relation and taking only the imaginary parts we get

1183: %

1184: \begin{eqnarray}

1185: 0 = \Omega_{i}\gamma_{g}+2\omega\gamma _g \pm

1186: \frac{\Omega_{i}}{\omega^{2}}\gamma_{g}\frac{kJ_{CR}}{n_{e}q}

1187:    (2\pi)^{3} q^{2} k_{\|}\left(\gamma\omega m/k_{\|}- mv_{s}\right)

1188: \frac{(2\kappa p_{t}^{2})^{\kappa}}{(2\kappa p_{t}^{2}+m^{2}v_{s}^{2})^{\kappa -1}}

1189: {\rm B}(1,1+\kappa)

1190: \frac{v_{A}^{2}}{c^{2}} \frac{1}{m^{2}} \frac{m}{k_{\|}}

1191: & & \nonumber \\

1192: \times

1193:  _{2}{\rm F}_{1}\left(2,1;2+\kappa;

1194: 1-\frac{2\kappa p_{t}^{2}+m^{2}v_{s}^{2}}

1195: {m^{2}c^{2}}\right)

1196: \frac{n_{CR}}{4\sqrt{2}(\pi \kappa)^{3/2}p_{t}^{3}}

1197: \frac{(2\kappa-3)\Gamma(\kappa)}{\Gamma(\kappa -1/2)} & &

1198: \end{eqnarray}

1199: %

1200: assuming $\gamma\omega/k_{\|} \ll v_{s}$ and including the normalization of

1201: $f$ and the factor of $k_{\|}/2$ that were omitted during the evaluation of

1202: the integral.

1203: From this we have

1204: \begin{equation}

1205: \gamma_{g} \simeq -\left(\frac{\pi}{2}\right)^{1/2}

1206: \frac{n_{CR}}{n_{i}} {\Omega_{i}^2\over\omega }

1207: \frac{\left(\gamma\omega m/k_{\|}- mv_{s}\right)}{p_{t}}

1208: \frac{2\kappa-3}{\kappa^{1/2}(\kappa +1)}

1209: \frac{\Gamma(\kappa)}{\Gamma(\kappa -1/2)} \times _{2}{\rm

1210: F}_{1}(2,1;2+\kappa;1) .

1211: \end{equation}

1212: Electromagnetic waves in the lower hybrid frequency range are

1213: parallel propagating whistlers, with

1214: \begin{equation}

1215: {k^2c^2\over\omega ^2}\simeq {\omega _{pe}^2\over\omega\left(\Omega

1216: _e-\omega\right)}

1217: \end{equation}

1218: and

1219: \begin{equation}

1220: {\partial\omega\over\partial k}\simeq {2\omega /k\over 1+k^2c^2/\omega _{pe}^2}.

1221: \end{equation}

1222: We assume $\partial\omega /\partial k\propto\partial\omega /\partial k_{\|}\sim v_s$

1223: and hence $\omega /k_{\|}\sim v_s/2$ for $k<< c/\omega _{pe}$. Thus for $\kappa =2$ and

1224: \begin{equation}

1225: \gamma_{g} \simeq \frac{4n_{CR}}{3n_{i}}

1226: {\Omega_{i}^2\over\omega}\left(1-\gamma /2\right).

1227: \end{equation}

1228: This is significantly smaller than the growth of lower hybrid waves, which is

1229: of order $\Omega _in_{CR}/n_i$.  Further, since

1230: whistlers carry energy along magnetic field lines, like Alfv\'en waves, only for

1231: specific shock obliquities will the energy of the waves stay in contact with the shock

1232: and allow large wave intensities to build up. Electromagnetic waves with frequency

1233: above the electron gyrofrequency (O and X modes) have phase velocities greater than

1234: $c$, and so cannot be excited by kinetic instabilities.

1235: 

1236: \begin{thebibliography}{}

1237: \bibitem[Achterberg(1983)]{Achterberg83} Achterberg, A.\ 1983, \aap, 119, 274

1238: 

1239: \bibitem[Bell(2005)]{Bell2005} Bell, A.~R.\ 2005, \mnras, 358, 181

1240: \bibitem[Bell(2004)]{Bell2004} Bell, A.~R.\ 2004, \mnras, 353, 550

1241: \bibitem[Bell \& Lucek(2001)]{BellLucek2001} Bell, A.~R., \& Lucek, S.~G.\ 2001, \mnras, 321, 433

1242: \bibitem[Blandford \& Eichler(1987)]{blandford87}Blandford, R., \& Eichler, D. 1987, Physics Reports,

1243: 154, 1

1244: \bibitem[Bykov \& Uvarov(1999)]{bykov}Bykov, A. M., \& Uvarov, Yu. A. 1999, JETP, 88, 465

1245: \bibitem[Cargill \& Papadopoulos(1988)]{Cargill88} Cargill,

1246: P.~J., \& Papadopoulos, K.\ 1988, \apjl, 329, L29

1247: 

1248: \bibitem[Cassam-Chena{\"i} et al.(2007)]{Cassam2007} Cassam-Chena{\"i}, G., Hughes, J.~P., Ballet, J., \& Decourchelle, A.\ 2007, \apj, 665, 315

1249: \bibitem[Dieckmann et al.(2000)]{dieckmann00}Dieckmann, M. E., McClements, K. G.,

1250: Chapman, S. C., Dendy, R. O., \& Drury, L. O'C. 2000, \aap, 356, 377

1251: \bibitem[Drury \& Falle(1986)]{Drury86} Drury, L.~O., \& Falle,

1252: S.~A.~E.~G.\ 1986, \mnras, 223, 353

1253: \bibitem[Fransson et al.(1996)]{fransson96}Fransson, C., Lundqvist, P., \& Chevalier,

1254: R. A. 1996, \apj, 461, 993

1255: \bibitem[Gedalin et al.(2007)]{gedalin07}Gedalin, M., Balikhin, M. A., \& Eichler, D. 2007,

1256: astro-ph

1257: \bibitem[Ghavamian et al.(2007)]{GLR07} Ghavamian, P.,

1258: Laming, J.~M., \& Rakowski, C.~E.\ 2007, \apjl, 654, L69

1259: \bibitem[Ghavamian et al.(2003)]{GRHW03} Ghavamian, P.,

1260: Rakowski, C.~E., Hughes, J.~P., \& Williams, T.~B.\ 2003, \apj, 590, 833

1261: \bibitem[Ghavamian et al.(2002)]{G02} Ghavamian, P.,

1262: Winkler, P.~F., Raymond, J.~C., \& Long, K.~S.\ 2002, \apj, 572, 888

1263: \bibitem[Ghavamian et al.(2001)]{G01} Ghavamian, P.,

1264: Raymond, J., Smith, R.~C., \& Hartigan, P.\ 2001, \apj, 547, 995

1265: \bibitem[Gradshteyn \& Ryzhik(1965)]{grad1965}Gradshteyn, Yu. V., \&

1266: Ryzhik, I. M. 1965, Table of Integrals, Series, and Products, (New

1267: York: Academic Press)

1268: \bibitem[Heng et al.(2007)]{Heng07b}Heng, K., van Adelsberg, M., McCray, R., \& Raymond, J.~C.\ 2007, \apj, 668, 275

1269: \bibitem[Heng \& McCray(2007)]{Heng07a}Heng, K., \& McCray, R.\ 2007, \apj, 654, 923

1270: \bibitem[Hughes et al.(2000)]{hughes00}Hughes, J. P., Rakowski, C. E., \& Decourchelle, A.

1271: 2000, \apj, 543, L61

1272: \bibitem[Karney(1978)]{karney78}Karney, C. F. F. 1978, Phys. Fluids, 21, 1584

1273: \bibitem[Laming \& Lepri(2007)]{laming07}Laming, J. M., \& Lepri, S. T. 2007, \apj, 660, 1642

1274: \bibitem[Laming(2005)]{Laming2005} Laming, J.~M.\ 2005, \apss, 298, 385

1275: \bibitem[Laming(2001b)]{L01b} Laming, J.~M.\ 2001, \apj, 563, 828

1276: \bibitem[Laming(2001a)]{L01a} Laming, J.~M.\ 2001, \apj, 546, 1149

1277: \bibitem[Long et al.(2003)]{Long2003} Long, K.~S., Reynolds, 

1278: S.~P., Raymond, J.~C., Winkler, P.~F., Dyer, K.~K., 

1279: \& Petre, R.\ 2003, \apj, 586, 1162 

1280: \bibitem[Lucek \& Bell(2000)]{LucekBell2000}

1281: Lucek, S.~G., \& Bell, A.~R.\ 2000, \mnras, 314, 65

1282: 

1283: \bibitem[Malkov \& O'C Drury(2001)]{MalkovDrury2001}

1284: Malkov, M.~A., \& O'C Drury, L.\ 2001, Reports of Progress in Physics, 64, 429

1285: 

1286: \bibitem[Marcowith et al.(2006)]{Marcowith2006} Marcowith, A., Lemoine, M., \& Pelletier, G.\ 2006, \aap, 453, 193

1287: \bibitem[McClements et al.(2001)]{mcclements01}McClements, K. G., Dieckmann, M. E.,

1288: Ynnerman, A., Chapman, S. C., \& Dendy, R. O. 2001, Phys. Rev. Lett. 87, 255002

1289: \bibitem[McClements et al.(1997)]{mcclements97}McClements, K. G., Dendy, R. O., Bingham, R.,

1290: Kirk, J. G., \& Drury, L. O'C. 1997, \mnras, 291, 241

1291: \bibitem[Melrose(1986)]{melrose86}Melrose, D. B. 1986, , Instabilities in Space

1292: and Laboratory Plasmas, (Cambridge: Cambridge University Press)

1293: \bibitem[Niemiec et al.(2008)]{niemiec08}Niemiec, J., Pohl, M., Stroman, T., \&

1294: Nishikawa, K.-I. 2008, astro-ph/0802.2185

1295: \bibitem[Ohira \& Takahara(2007)]{ohira07}Ohira, Y., \& Takahara, F. 2007, \apj,

1296: 661, L171

1297: \bibitem[Parizot et al.(2006)]{Parizot2006} Parizot, E., Marcowith, A., Ballet, J.,

1298: \& Gallant, Y.~A.\ 2006, \aap, 453, 387

1299: \bibitem[Pelletier et al.(2006)]{Pelletier2006} Pelletier, G., Lemoine, M., \&

1300: Marcowith, A.\ 2006, \aap, 453, 181

1301: \bibitem[Rakowski(2005)]{R05} Rakowski, C.~E.\ 2005,

1302: Advances in Space Research, 35, 1017

1303: \bibitem[Rakowski et al.(2003)]{RGH03} Rakowski, C.~E.,

1304: Ghavamian, P., \& Hughes, J.~P.\ 2003, \apj, 590, 846

1305: 

1306: \bibitem[Reville et al.(2007)]{Reville2007} Reville, B., Kirk, J.~G., Duffy, P., \& Sullivan, S.~O 2007, \aap submitted, arXiv:0707.3743

1307: \bibitem[Reville et al.(2008)]{reville08} Reville, B., O'Sullivan, S., Duffy, P., \&

1308: Kirk, J. G. 2008, \mnras in press, astro-ph/0802.0109

1309: \bibitem[Schwartz et al.(1988)]{Schwartz88} Schwartz, S.~J., Thomsen, M.~F., Bame, S.~J.,

1310: \& Stansberry, J.\ 1988, \jgr, 93 12923

1311: \bibitem[Scudder et al.(1986)]{scudder86}Scudder, J. D., Aggson, T. L., Mangeny, A., Lacombe, C.,

1312: \& Harvey, C. C. 1986, \jgr, 91, 11019

1313: \bibitem[Schmitz et al.(2002)]{schmitz02}Schmitz, H., Chapman, S. C., \& Dendy, R. O.

1314: 2002, \apj, 579, 327

1315: \bibitem[Shalchi \& Kourakis(2007)]{Shalchi2007} Shalchi, A., \& Kourakis, I.\ 2007, \aap, 470, 405

1316: \bibitem[Shimada \& Hoshino(2000)]{shimada00}Shimada, N., \& Hoshino, M. 2000, \apj,

1317: 543, L67

1318: \bibitem[Tidman \& Krall(1971)]{tidman71}Tidman, D. A., \& Krall, N. A. 1971, Shock Waves in

1319: Collisionless Plasmas (New York: Wiley-Interscience)

1320: \bibitem[Vi\~nas et a.(2000)]{vinas00}Vi\~nas, A. F., Wong, H. K., \& Klimas, A. J. 2000,

1321: \apj, 528, 509

1322: %\bibitem[Vink et al.(2003)]{Vink2003} Vink, J., Laming, J.~M., Gu, M.~F., Rasmussen, A., \& Kaastra, J.~S.\ 2003, \apjl, 587, L31

1323: 

1324: \bibitem[Vink \& Laming(2003)]{VinkLaming2003} Vink, J., \& Laming, J.~M.\ 2003, \apj, 584, 758

1325: %\bibitem[Vink et al.(2001)]{Vink2001} Vink, J., Laming, J.~M., Kaastra, J.~S., Bleeker, J.~A.~M., Bloemen, H., \& Oberlack, U.\ 2001, \apjl, 560, L79

1326: 

1327: \bibitem[Warren et al.(2005)]{Warren} Warren, J.~S., et al.\

1328: 2005, \apj, 634, 376

1329: \bibitem[Yamazaki et 

1330: al.(2004)]{Yamazaki2004} Yamazaki, R., Yoshida, T., Terasawa, T., Bamba, A., \& Koyama, K.\ 2004, \aap, 416, 595 

1331: \bibitem[Zank et al.(2006)]{Zank2006} Zank, G.~P., Li, G.,

1332: Florinski, V., Hu, Q., Lario, D., \& Smith, C.~W.\ 2006,

1333: Journal of Geophysical Research (Space Physics), 111, 6108

1334: \bibitem[Zirakashvili et al.(2008)]{zirakashvili08}Zirakashvili, V. N., Ptuskin, V. S.,

1335: \& V\"olk, H. J. 2008, \apj in press, astro-ph/0801.4486

1336: 

1337: \end{thebibliography}

1338: 

1339: \clearpage

1340: \begin{figure}

1341: \plotone{f1.eps} \figcaption[f1.eps]{Schematic illustrations of the

1342: amplification of magnetic field by the non-resonant modified Alfv\'{e}n waves in the

1343: shock precursor in the parallel orientation of the

1344: ambient field with respect to the shock normal. The evolution of a single field line

1345: in an exponential purely growing mode is shown.

1346: As the field is amplified, the shock becomes quasi-perpendicular and

1347: the effective $M_{A}$ decreases, eventually to the point where

1348: lower hybrid wave growth takes over, allowing a short region of electron heating.

1349: \label{fig1}}

1350: \end{figure}

1351: 

1352: \clearpage

1353: \begin{figure}

1354: \plotone{f2.eps} \figcaption[f2.eps]{Same as figure 1 but for an initially

1355: perpendicular shock. The evolution of a purely growing mode is illustrated.

1356: The magnetic field amplification is less strong than in the quasi-parallel case, and the shock geometry

1357: remains quasi-perpendicular.

1358: \label{fig2}}

1359: \end{figure}

1360: 

1361: 

1362: \clearpage

1363: \begin{figure}

1364: \plotone{f3.eps} \figcaption[f3.eps]{Allowed range of $M_A$ as a function of

1365: $M_S$ for the existence of a magnetosonic soliton, from Tidman \& Krall (1971). The upper

1366: limit is given by the solid line, the lower limit by the dashed line.

1367: \label{fig3}}

1368: \end{figure}

1369: 

1370: 

1371: \end{document}

1372: 

1373: 

1374: 

1375: dumped text:

1376: 

1377: %The Bell and Lucek mechanism for turbulent amplification of the magnetic field occurs only in the parallel shock orientation because that is where the firehouse instability applies. However here we have been discussing purely the perpindicular case which is needed for the lower hybrid wave generation. How do we reconcile the two?

1378: %One possibility is that the waves generated in the parallel portions of the shock will be converted into other forms when they encounter the perpindicular portions, grossly conserving the energy in turbulence and cosmic rays.

1379: %Another point is that the turbulent field itself will by its nature create many more new regions of perpendicular shock geometry as well as parallel shock geometry.

1380: 

1381: 

1382: %\bibitem[Bell(1987)]{Bell1987MNRAS.225..615B} Bell, A.~R.\ 1987, \mnras, 225, 615

1383: %\bibitem[Bell(1978)]{Bell1978b} Bell, A.~R.\ 1978, \mnras, 182, 443

1384: %\bibitem[Bell(1978)]{Bell1978a} Bell, A.~R.\ 1978, \mnras, 182, 147

1385: %\bibitem[Canizares et al.(1978)]{can78} Canizares, C. R.,

1386: %    Grindlay, J. E., Hiltner, W. A., Liller, W., \&

1387: %    McClintock, J. E.  1978, \apj, 224, 39

1388: 

1389: 

1390: %\bibitem[Shalchi(2007)]{2007A&A...469..839S} Shalchi, A.\ 2007, \aap, 469, 839

1391: %\bibitem[Shalchi(2006)]{2006A&A...453L..43S} Shalchi, A.\ 2006, \aap, 453, L43

1392: %\bibitem[Shalchi et al.(2004)]{2004ApJ...604..675S} Shalchi, A., Bieber, J.~W., \& Matthaeus, W.~H.\ 2004, \apj, 604, 675

1393: %\bibitem[Shalchi \& Schlickeiser(2006)]{2006A&A...454....1S} Shalchi, A., \& Schlickeiser, R.\ 2006, \aap, 454, 1

1394: 

1395: %\bibitem[Schlickeiser et al.(2007)]{2007ApJ...661..185S} Schlickeiser, R., Dohle, U., Tautz, R.~C., \& Shalchi, A.\ 2007, \apj, 661, 185

1396: 

1397: %\bibitem[Zank et al.(2004)]{2004JGRA..10904107Z} Zank, G.~P., Li, G., Florinski, V., Matthaeus, W.~H., Webb, G.~M., \& le Roux, J.~A.\ 2004, Journal of Geophysical Research (Space Physics), 109, 4107

1398: