1: %\documentstyle[11pt,epsfig]{article}

2: \documentstyle[11pt, preprint]{aastex}

3: \textwidth=6.5in

4: \textheight=8.25in

5: 

6: \begin{document}

7: 

8: %\voffset=-0.3 true cm

9: %\hoffset=-1.8 true cm

10: 

11: \def\beq{\begin{equation}}

12: \def\eeq{\end{equation}}

13: \def\a{\alpha}

14: \def\t{\theta}

15: \def\r{{\it RHESSI}\,}

16: \def\ie{{\it i.e.}\,}

17: \def\eg{{\it e.g.}\,}

18: \def\ea{et al.\,}

19: \def\3he{$^3$He\,}

20: \def\he4{$^4$He\,}

21: \def\eq{equation\,\,}

22: \def\eqs{equations\,\,}

23: \def\cc{{\rm cm}^{-3}}

24: \def\g{\gamma}

25: \def\loss{{\rm loss}}

26: \def\Coul{{\rm Coul}}

27: \def\sy{{\rm synch}}

28: \def\scat{{\rm scat}}

29: \def\esc{{\rm esc}}

30: \def\eff{{\rm eff}}

31: \def\tr{{\rm cross}}

32: \def\ic{{\rm IC}}

33: \def\br{{\rm brem}}

34: \def\ac{{\rm ac}}

35: \def\min{{\rm min}}

36: \def\max{{\rm max}}

37: \def\tot{{\rm total}}

38: \def\rt{{\rm rad tot}}

39: \def\ln{{\rm ln}}

40: \def\Clog{\ln\Lambda}

41: \def\ltsim{\hbox{\rlap{\raise.5ex\hbox{$<$}}{\lower.7ex\hbox{$\sim$}}}}

42: \def\gtsim{\hbox{\rlap{\raise.5ex\hbox{$>$}}{\lower.7ex\hbox{$\sim$}}}}

43: 

44: \title{Particle Acceleration in Solar Flares and Enrichment of $^3$He and Heavy

45: Ions}

46: 

47: 

48: \author{Vah\'{e} Petrosian\altaffilmark{1,2,3}}

49: 

50: \altaffiltext{1}{Department of Physics, Stanford University, Stanford, CA, 94305

51: email; vahep@stanford.edu}

52: \altaffiltext{2}{Kavli Institute for Particle Astrophysics and Cosmology,

53: Stanford University, Stanford, CA 94305}

54: \altaffiltext{3}{Also Department of Applied Physics}

55: 

56: \begin{abstract}

57: We discuss possible mechanisms of acceleration of particles in solar flares and

58: show that turbulence plays  an 

59: important role in all the mechanism. It is also argued that  stochastic particle

60: acceleration by turbulent plasma waves is the most likely mechanism for

61: production of the high energy electrons and ions responsible for observed

62: radiative signatures of solar flares and for solar energetic particle or SEPs,

63: and that the predictions of this model agrees well with many past and recent

64: high spectral and temporal observations of solar flares. It is shown that, in

65: addition, the model explains many features of   SEPs  that accompany flares. In

66: particular we show that it can  successfully explain the observed extreme

67: enhancement, relative to photospheric values,  of $^3$He ions and the relative

68: spectra of $^3$He and $^4$He. It has also the potential of explaining the

69: relative abundances of most ions including the increasing  enhancements of heavy

70: ions with ion mass or mass-to-charge ratio.

71: 

72: \end{abstract}

73: 

74: \section{Introduction}

75: 

76: The aim of this paper is to investigate to what extent current theoretical

77: models of

78: acceleration of particles can explain the observed characteristics of solar

79: energetic particles (SEPs), in particular the extreme enhancement,

80: relative to photospheric values, of $^3$He ions in many flares, in particular in the so-called

81: impulsive (or He-rich) events. In the next section we  present a general

82: description

83: of the possible acceleration mechanisms for production of particles in solar

84: flare.  We show that plasma waves or turbulence play a major

85: role in all these processes. In \S 3 we describe the model of

86: stochastic

87: acceleration by plasma turbulence in some detail and review its successes in

88: describing observations of solar flares, in particular the radiative signatures

89: produced by accelerated electrons and protons. In \S 4 we address the

90: problem of observed enhancements

91: of \3he and heavy elements.  Here we first

92: present a brief review of some of the earlier models proposed for production of

93: these enhancements and then compare predictions of the stochastic acceleration

94: model with the observations. A brief summary and discussion is presented in \S

95: 5.

96: 

97: 

98: 

99: \section{Particle Acceleration}

100: \label{acceleration}

101: 

102: In this section we first compare various acceleration processes and stress the

103: importance of plasma wave turbulence  ({\bf PWT}) as an agent of

104: acceleration, and then describe the basic scenario and equations for treatment

105: of

106: these processes in the stochastic acceleration ({\bf SA}) model.

107: As described below there is growing evidence that PWT plays an

108: important role in acceleration of particles in general, and in solar flares in

109: particular. The two most commonly used acceleration mechanisms are the

110: following.

111: 

112: {\bf 1. Electric Field Acceleration:}  Electric fields parallel to

113: magnetic fields can accelerate charged particles.

114: For solar flare conditions the Dreicer field ${\bf E}_D = k_BT/(ev\tau_{\rm

115: Coul})\sim 10^{-5}$ V/cm. (Here $k_B, T, v$ and $e$ are the Boltzmann constant,

116: plasma temperature, particle velocity and charge,

117: and $\tau_{\rm Coul}$ is the mean collision time.)

118: Sub-Dreicer fields (${\bf E}\leq{\bf E}_D$) can only accelerate particles up to

119: 10's of keV (for a typical flare

120: length scale

121: $L\sim10^9$~cm)  which is far below the  10's of MeV electrons or $>10$ GeV

122: protons

123: required

124: by observations%

125: \footnote{Dreicer field will be large if the

126: resistivity and density are

127: anomalously high (Tsuneta 1985; Holman 1996a, b).}.  Super-Dreicer fields,

128: which seem to be present in many simulations of reconnection (Drake, 2006, see

129: also

130: Cassak \ea 2006; Hoshino 2006, see also Zenitani \& Hoshino, 2005),

131: accelerate particles at a rate that is faster than $\tau_{\rm Coul}$. This can

132: lead to a runaway and an unstable electron distribution which, as shown

133: theoretically,

134: by laboratory experiments and by the above mentioned simulations,

135: most probably will give rise to PWT (Boris et al.

136: 1970, Holman 1985)%

137: \footnote{Note also that  production of a broad

138: power-law requires a wide range of  potential drops

139: (Litvinenko, 2003).}. {\it In summary the electric fields arising as a result of

140: reconnection cannot be the sole agent of acceleration, but may produce an

141: unstable particle momentum distribution which will produce PWT that can then

142: accelerate particles.}

143: 

144: {\bf 2. Fermi Acceleration:} Nowadays this process has been divided into two

145: kinds.

146: In the original Fermi process particles of velocity $v$ scattering

147: randomly with a (pitch angle cosine $\mu$) diffusion rate $D_{\mu\mu}$ by moving

148:  scattering agents with a velocity $u$

149: gain energy at a rate proportional to $(u/v)^2D_{\mu\mu}$.

150: This, known as a {\it second order Fermi process}, is what we call {\it

151: stochastic acceleration}

152: ({\bf SA}). {\it For solar flares the most likely agent for scattering is PWT.}

153: An alternative process is what is commonly referred to as {\bf Shock}

154: acceleration.

155: Because particles

156: crossing the shock (or any kind of flow convergence) gain energy at a rate which

157: is

158: $\propto u_{sh}/v$, this is known as the {\it first order Fermi} process.

159: Ever since the demonstration by several authors that a very simple version of

160: this process leads to a power law spectrum that agrees approximately with

161: observations of the cosmic rays, shock acceleration is commonly invoked in space

162: and astrophysical

163: plasmas. However, this simple model, though very elegant, has many shortcomings

164: Specially

165: when applied to non-thermal radiation processes.

166: 

167: {\bf Shock} acceleration, as commonly used, requires injection of (sometimes

168: high energy)

169: particles and is unable to accelerate low energy background particles.

170: The simple results also break down when  one considers

171: more realistic models. For example, inclusion of losses

172: (Coulomb at low energies and synchrotron at high) or influence of

173: accelerated particles on the shock structure (see \eg Ellison \ea, 2005: Amato

174: \& Blasi 2005)

175: cause breaks and generally deviations from a simple power law.

176: More importantly, a {\it shock by itself cannot accelerate particles}

177: and requires some

178: scattering agents (most likely PWT) to cause repeated passages of the particles

179: through the shock front

180: (thus the name Fermi).  The rate of

181: energy gain is governed by the scattering rate $D_{\mu\mu}$.

182: This is the case for the so-called quasi-parallel shock, but  shocks

183: are most likely quasi-perpendicular in which case, as pointed out by Jokipii

184: (\eg 1987),

185: the particles may drift along the surface of the shock and get accelerated more

186: efficiently. However, this model also suffers from the above mentioned

187: fundamental limitation and

188: requires injection of particles with velocities $v>u_{sh}\eta$,

189: where $\eta \gg 1$ is the ratio of parallel to perpendicular (to magnetic field)

190: diffusion coefficients. In a more recent simulation of this process Giacolone

191: (2005)

192: shows that a non-thermal tail can be produced from a thermal distribution.

193: However, this model includes

194: significant field fluctuations, which as advocated below can accelerate low

195: energy

196: particles more efficiently

197: than shocks. Thus, it is not obvious what are the

198: relative roles of shock vs SA here.

199: Moreover, for solar flares there is no

200: direct evidence for shocks near the LT

201: during the impulsive phase, and some of the features that make

202: acceleration of cosmic rays by shocks attractive are not present%

203: {\footnote{For example, acceleration by sub Alfv\'enic shocks is

204: questionable (see Kulsrud 2005), and the streaming instability,

205: which is a part of the traditional shock acceleration

206: can be partially suppressed in the presence of PWT (Yan \& Lazarian 2002;

207: Farmer \& Goldreich 2004).}.

208: 

209: {\bf Stochastic Acceleration} is favored because the PWT needed for scattering

210: can also accelerate particles stochastically with the rate $D_{EE}/E^2$,

211: so that shocks may not be always necessary. First, contrary to common belief,

212: Hamilton \& Petrosian 1992 and 

213: Miller \& Reames 1996 have shown that for flare conditions PWT can

214: accelerate the background particles to high energies within the desired

215: time.  More importantly, at low

216: energies or in strongly magnetized plasmas the acceleration rate $D_{EE}/E^2$

217: may exceed the scattering rate $D_{\mu\mu}$ (see Pryadko \& Petrosian, {\bf

218: PP97}).  {\it Thus,

219: for flare conditions low energy particles are accelerated more efficiently by

220: PWT than

221: by shocks}. We note, however, that in gradual flares there may be a

222: need for a second stage acceleration of SEPs by high coronal shocks%

223: \footnote{In practice, \ie mathematically, there

224: is little difference between the two mechanisms (Jones 1994), and the

225: acceleration by turbulence and  shocks can be combined (see below).}.

226: 

227: {\it Irrespective of which process dominates the particle acceleration, it is

228: clear

229: that PWT has a

230: role in all of them. Thus, understanding of the production of PWT and

231: its interaction with particles  is extremely important.}

232: Moreover, {\bf turbulence} is expected to be present in most astrophysical

233: plasmas

234: including solar flares and in and around interplanetary shocks, because the

235: ordinary and magnetic Reynolds numbers for these situations are

236: very large and may be the most efficient channel of energy dissipations in

237: non-equilibrium systems such as solar flares.

238: In recent years there has been a substantial progress

239: in the understanding of

240: MHD turbulence (Goldreich \& Sridhar, 1995, 1997; Lithwick \& Goldreich 2001;

241: Cho \& Lazarian 2002 and 2006). These advances provide new tools for a more

242: quantitative

243: investigation of turbulence and the role it plays

244: in solar flares.

245: 

246: %\input{acceleration.tex}

247: 

248: %\input{SAmodel.tex}

249: 

250: \section{Stochastic Acceleration and Solar Flares}

251: \label{SAandFlares}

252: 

253: 

254: \subsection{Basic Scenario}

255: \label{scenario}

256: 

257: The complete picture of a solar flare involves many phases or steps.

258: After a complex pre-flare buildup, the first

259: phase is the reconnection or the energy release process.  The final consequences

260: of

261: this released energy are the observed radiations (from long wavelength radio to

262: GeV gamma-rays), SEPs and CMEs.  Many processes are involved in

263: conversion of the released energy into radiation. As stressed above

264: we believe that PWT plays an important role in these processes.  We envision the

265: following

266: scenario.

267: Magnetic energy is converted into turbulence by the reconnection process above

268: corona

269: loops which we refer to as the acceleration site or the loop top ({\bf LT})

270: source (see Figure \ref{model}). The turbulence or waves generated on some

271: macroscopic scale (some fraction of reconnection site scale) undergo

272: two kind of interactions. The  first is {\it nonlinear wave-wave interaction}

273: causing

274: them to undergo dissipationless cascade to smaller scales. The second  is

275: {\it wave-particle interaction} which becomes more  important at smaller scales.

276: This

277: damps the turbulence. The lost energy goes into heating the background plasma

278: and/or accelerating particles into a non-thermal

279: tail.

280: The accelerated particles on the open field lines escape the Sun and are

281: observed as SEPs having undergone varied degree of scatterings and acceleration

282: (\eg by interplanetary shocks) during their transport to the Earth. The escaping

283: electrons may also produce type III radio bursts. The particles on the field

284: lines associated with the closed loops spiral down towards the photosphere

285: and produce the observed

286: non-thermal radiations

287: via their interactions with the

288: background particles (and fields) along the loop and primarily at the footpoints

289: ({\bf FPs}).

290: However, most of the energy of the non-thermal particles is lost  by collisions

291: causing heating and evaporation of the colder chromospheric plasma, which is

292: responsible

293: for most of the softer thermal radiation. This process,

294: described by the hydrodynamic equations, has a

295: time-scale comparable to the sound travel time, and

296: is somewhat decoupled from the acceleration-transport process that has

297: a much shorter time-scale. However, the evaporation can modulate the high

298: energy processes by changing the density and temperature in the

299: acceleration site.

300: 

301: \subsection{Formalism}

302: \label{equations}

303: 

304: The mathematical treatment of the processes involved in such a scenario is long

305: and complex. Below we give an outline of the important aspects.

306: 

307: \subsubsection{Kinetic Equations}

308: \label{kinetic}

309: 

310: {\bf The spectrum of PWT} $W({\bf k},t)$ is determined by the wave-wave and

311: wave-particle interactions. The cascade is evaluated  from the rates of

312: wave-wave interactions. For

313: example the three wave interactions can be presented as (see \eg Chandran 2005;

314: Luo \& Melrose 2006)

315: \beq\label{wave-wave}

316: \omega({\bf k}_1) + \omega({\bf k}_2) = \omega({\bf k}_3) \,\,\,\, {\rm and}

317: \,\,\,\, {\bf

318: k}_1+{\bf k}_2={\bf k}_3,

319: \eeq

320: where $\omega$ and ${\bf k}$ are the wave frequency and wavevector. The

321: interaction

322: rates can be represented by the wave diffusion coefficient $D_{ij}$. In

323: general, at large scales the  cascade

324: time $\tau_{\rm cas}\sim k^2/D_{ij}$ is shorter than the damping time

325: $\Gamma(k)^{-1}$ resulting from wave-particle interactions. Thus, the waves

326: undergo dissipationless cascade from the injection scale $k_\min\sim L^{-1}$

327: (with a rate ${\dot Q}^W({\bf k})$) till $k_\max$ where $\tau_{\rm

328: cas}\Gamma=1$, known as the {\it inertial range}. Beyond $k_\max$ the

329: spectrum of waves drops off more rapidly.

330: Adopting the

331: diffusion approximation (see \eg Zhou \& Matthaeus 1990), one

332: can obtain

333: the evolution of the spatially integrated wave spectrum $W({\bf k}, t)$

334: \beq

335: \\{\partial {W} \over \partial t}

336: = {\partial \over\partial k_i}\left[D_{ij}{\partial\over \partial

337: k_j}{W}\right] -

338: \Gamma({\mathbf k}){W} - {{W}\over T^{W}_{\rm esc}({\mathbf k})} + \dot{Q}^{W}.

339: \label{waves}

340: \eeq

341: We will assume that all the wave energy is absorbed so that there is no wave

342: escape, \ie $T^{W}_{\rm esc}\rightarrow\infty$.

343: 

344: The general equation for treatment of the {\bf particle acceleration and

345: transport} is the

346: Fokker-Planck equation

347: for the gyro-phase averaged particle distribution $f(t,s,E,\mu)$ as a function

348: of

349: distance $s$ along the magnetic field lines. This equation is simplified

350: considerably if one can adopt the

351: isotropic

352: approximation (or deal with the pitch-angle averaged distribution) and 

353: impose

354: the homogeneity condition (or deal with distribution integrated over the

355: acceleration region). These are reasonable assumptions for particles at the

356: flare site and  amount to determination of the evolution of the

357: energy spectrum

358: $N(E,t)=\int \int d\mu ds f(t,s,E,\mu)$:

359: \beq

360: {\partial N \over \partial t}

361:  =  {\partial \over \partial E}\left[D_{EE}{\partial N \over \partial E}

362:  - (A-\dot E_L) N\right]

363:  - {N \over T^p_{\rm esc}} +{\dot Q}^p,

364: \label{kineq}

365: \eeq

366: Here  $D_{EE}/E^2, A(E)/E$ describe  the energy

367: diffusion and direct acceleration rates. They are obtained from consideration of

368: the wave-particle interactions, which are often dominated by resonant

369: interactions, specially for low beta (magnetically dominated) plasma, such that

370: \beq\label{resonance}

371: \omega({\bf k})-k\cos\t v\mu = n\Omega/\gamma,

372: \eeq

373: for  waves propagating at an angle $\t$ with respect to the large scale magnetic

374: field,

375: and a particle of velocity $v$, Lorentz factor $\gamma$, pitch angle cosine

376: $\mu$ and gyrofrequency $\Omega$. Here the harmonic number  $n$ (not to be

377: confused with the density)

378: is equal to zero for transit

379: time damping ({\bf TTD}) process. For parallel propagating waves ({\bf PPWs})

380: $n=\pm 1$

381: but for obliquely propagating waves

382: $n= \pm 1, \pm 2, ...$\,.

383: ${\dot E}_L/E$ is the energy loss rate of the particles (due mainly to Coulomb

384: collisions and synchrotron losses) and $\dot{Q}^p$ and the terms with the

385: escape times $T_{\rm esc}$ describe the source and leakage of particles.  (For a

386: more detailed discussion leading to the above equations see Miller et al. 1996

387: and Petrosian \& Liu 2004, {\bf PL04}).

388: 

389: The above two kinetic equations are coupled by the fact that

390: the coefficients of one depend on the spectral distribution of the other.

391: Conservation of energy requires that the energy lost by the waves ${\dot

392: {\cal

393: W}_{\rm tot}}\equiv \int \Gamma({\bf k})W({\bf k})d^3k$ be equal to the energy

394: gained by the

395: particles from the waves;  ${\dot {\cal E}}=\int [A(E)-A_{\rm sh}]N(E)d E$.

396: Representing the

397: energy transfer rate between the waves and particles by

398: $\Sigma ({\bf k}, E)$ this equality implies that

399: \beq

400: \label{coefficients}

401: \Gamma({\bf k})= \int_0^\infty \d E N(E)\Sigma({\bf k}, E),\,\,\,\,

402: A(E)=\int_0^\infty d^3kW({\bf k})\Sigma ({\bf k}, E)+ A_{\rm sh}\ ,

403: \eeq

404: where we have added $A_{\rm sh}$ to represent

405: contributions of other (non-SA) processes affecting the direct acceleration, \eg

406: shocks.

407: 

408: The transport effects and further acceleration of the particles as they travel

409: in the interplanetary space can also be treated by similar equations.

410: 

411: \subsubsection{Dispersion Relations}

412: 

413: At the core of the evaluation of all the coefficients of the kinetic equations

414: described above lies the

415: plasma dispersion relation $\omega({\bf k})$ which determines the

416: characteristics of the waves that can be excited in the plasma, and the rates of

417: wave-wave and wave-particle interactions.

418: For example, in the MHD regime for a cold plasma

419: \beq

420: \label{mhddisp}

421: \omega=v_{\rm A}k\cos\t\,\,\,\,\,  {\rm and}\,\,\,\,\,  \omega=v_{\rm A}k

422: \eeq

423: for the Alfv\'en  and fast magneto-sonic waves, respectively, where $v_{\rm A}$

424: is the Alfv\'en velocity. Beyond the MHD regime a multiplicity of wave modes can

425: be

426: present and the dispersion relation is more complex and is obtained from the

427: following 

428: expressions:

429: \beq\label{dispgeneral}

430: \tan^2\theta = {-P(n_r^2-R)(n_r^2-L)\over (Sn_r^2-REL)(n_r^2-P)}

431: \eeq

432: where $n_r=kc/\omega$ is the refractive index, $S={1\over 2}(R+L)$, and

433: \beq\label{dispterms}

434: P= 1-\sum_i{\omega_{pi}^2\over \omega^2},\ \ \  R=1-\sum_i{\omega_{pi}^2\over

435: \omega^2}\left({\omega\over 

436: \omega+\epsilon_i\Omega_i}\right),\ \ \ {\rm and} \ \ \

437: L= 1-\sum_i{\omega_{pi}^2\over \omega^2}\left({\omega\over 

438: \omega-\epsilon_i\Omega_i}\right).

439: \eeq

440: Here $\omega_{pi}^2={4\pi n_i q_i^2\over m_i}$ and

441: $\Omega_i={|q_i|B\over m_ic}$ are the plasma and gyro frequencies,

442: $\epsilon_i={q_i\over |q_i|}$, and $n_i$, $q_i$, and $m_i$ are the  

443: density, charge, and mass of the background particles (electron, proton and

444: $\alpha$.)

445: Figure \ref{dispersion} shows the more complete picture of the dispersion

446: surfaces  (depicted by the

447: curves) obtained from the well known cold plasma expressions

448: along with the resonant planes

449: in  the ($\omega,\ k_\parallel,\ k_\perp$) space. Intersections

450: between the dispersion

451: surfaces and the resonant planes define the resonant wave-particle interactions.

452: 

453: \begin{figure}[hbtp]

454: \centerline{

455: \includegraphics[height=8.5cm,angle=270]{dispelect.ps}

456: \includegraphics[height=8.0cm,angle=270]{dispprot.ps}

457: }

458: \caption

459: {\scriptsize

460: Dispersion relation (curves) and resonance condition (flat) surfaces

461: for a cold fully ionized H and He (10\% by number) plasma with

462: $\beta_{\rm A}=v_{\rm A}/c=0.012$

463: showing the regions

464: around the electron (left) and proton (right) gyro-frequencies. Only

465: waves with positive $k_\|, k_\perp$ (or $0<\t<\pi/2$) are shown.

466: The mirror image with respect to the ($\omega,\ k_{\perp}$) plane

467: gives the waves propagating in the opposite direction. From high to

468: low frequencies, we have one of the electromagnetic branches (green),

469: upper-hybrid

470: branch (purple), lower-hybrid branch, which also includes the whistler waves

471: (pink),

472: fast-wave branches (yellow), and Alfv\'{e}n

473: branch (black). The effects of a finite temperature modify these curves at

474: frequencies $\omega\sim kv_{\rm th}$, where $v_{\rm th}=\sqrt{2k_{\rm B}T/m}$ is

475: the thermal velocity (see \eg Andr\'{e} 1985). The

476: resonance surfaces are for electrons with $v=0.3c$ and $|\mu|=1.0$ (left: upper

477: $n=1$, lower $n=0$) and

478: $^4$He (right: middle $n=1$) and $^3$He (right: upper $n=1$) ions with

479: $|\mu|=1.0$ and $v=0.01c$. The

480: resonance surfaces for the latter two are the same when $n=0$ (right: lower).}

481: \label{dispersion}

482: \end{figure}

483: 

484: The above dispersion relations are good approximations for low beta plasmas,

485: \beq

486: \label{beta}

487: \beta_p=2(v_s/v_{\rm A})^2=8\pi nk_{\rm B}T/B^2=3.4\times 10^{-2}(n/10^{10}

488: {\rm cm}^{-3})(100 {\rm G}/B)^2(T/10^7{\rm K})\ll 1,

489: \eeq

490: where $v_s=\sqrt{k_{\rm B}T/m_p}$ is the sound speed. For higher beta

491: plasmas,

492: \eg  at higher temperatures, these relations

493: are modified, specially for higher frequencies $\omega\sim kv_{\rm th}$, where

494: $v_{\rm th}=\sqrt{2k_{\rm B}T/m}$. For example, in the MHD regime, in

495: addition

496: to the Alfv\'en mode one gets fast and slow modes with the  dispersion relation

497: (see \eg Sturrock 1994)

498: \beq

499: \label{mhddisp2}

500: (\omega/k)^2={1\over 2}\left[(v_{\rm A}^2+v_s^2)\pm \sqrt{v_{\rm

501: A}^4+v_s^4-2v_{\rm A}^2v_s^2\cos {2\t}}\right],

502: \eeq

503: and the more general dispersion relation is modified in a more complicated way

504: (see \eg Andr\'{e} 1985 or Swanson 1989). The finite temperature imparts an

505: imaginary part to the wave frequency that gives the (Landau) damping rate of the

506: waves (see \eg Swanson 1989 or Pryadko \& Petrosian 1998, 1999 and Cranmer \&

507: van Ballegooijen 2003 for application

508: to solar flare conditions). In general, these rates and the

509: modification of the dispersion relation are known for Maxwellian (sometimes

510: anisotropic) energy distribution of the plasma particles. For non-thermal

511: distributions the damping rates can be evaluated as described in Petrosian, Yan

512: \& Lazarian (2006) using the

513: coupling described in equation  \ref{coefficients}. If the damping due to

514: non-thermal particles is important then the wave and particle kinetic equations

515: (\ref{waves}) and (\ref{kineq}) are coupled. This will be important in large

516: intense flares where a substantial fraction of particles are accelerated into a

517: non-thermal

518: tail. However, most often the damping rate is

519: dominated by the background thermal particles so that the wave and non-thermal

520: particle kinetic

521: equations decouple. This simplifies matters considerably. One can then use the

522: imaginary part of the wave frequency for evaluation of the damping rate using

523: the well-known hot plasma dispersion relation. Moreover, the 'thermal' effects

524: change the real part of the wave frequency only slightly so that

525: often the real part (and the particle diffusion coefficients) can be evaluated

526: using the simpler cold plasma dispersion relation depicted in Figure

527: \ref{dispersion}.

528: 

529: 

530: 

531: \subsubsection{Some Model Results} 

532:  

533: We now demonstrate some of the basic characteristics

534: of our scenario which can be compared with observations. The complete treatment

535: of the problem requires solution of the coupled wave particle kinetic equations

536: for particles at all energies and pitch angles and waves propagating at all

537: angles. In principle, given the plasma density, magnetic field, temperature

538: ($n,\

539: B,\ T$), the geometry of the

540: region

541: (represented by a size $L$ here), the rate and scale of injection  of

542: turbulence $\dot{Q}^W({\bf k}, t)$, one can evaluate the coefficients of

543: equations

544: (\ref{waves}) and (\ref{kineq}) and solve the coupled kinetic equations for

545: determination

546: of the resultant distributions $N(E,t)$ and $W({\bf k},t)$. However,

547: under certain circumstances some simplifications are possible.

548: As mentioned above in most cases the damping by non-thermal particles can be

549: ignored and the equations are decoupled. The wave spectrum then can be

550: approximated by a power law in the inertial range ($W\propto k^{-q}, k_{\rm

551: min}<k<k_{\rm max}$) with a steeper falling spectrum beyond this range. The

552: exponent $q$ will be in the range expected in a Kolmogorov or Kraichnan cascade.

553: Furthermore, as shown in Petrosian \ea (2006), the upper limit $k_{\rm max}$

554: depends strongly on the angle of propagation $\t$ with only quasi-parallel

555: propagating waves {\bf PPWs} reaching small enough scale to be in resonant with

556: low energy

557: particles. Thus, the initial acceleration of thermal particles will be dominated

558: by such waves. Consequently, in what follows here and the next section we assume

559: PPWs with the above spectrum. The following

560: results are based on PL04, where  more details can be found. 

561: 

562: \begin{figure}[htbp]

563: \leavevmode

564: \centerline{

565: %\vspace{-1.0cm}

566: \includegraphics[width=7.5cm]{ppwdispep.eps}

567: \includegraphics[width=7.5cm]{times.eps}

568: }

569: \caption{\scriptsize

570: {\bf Left panel:} The dispersion relation for PPWs (curves)  and the resonant

571: relation (lines, from eq.[\ref{resonance}]) for electrons (top) and protons

572: (bottom, expanded near the origin). Note that the resonant lines start (at

573: $k=0$) from the gyro frequency of the particle (divided by the Lorentz factor

574: $\gamma\sim 1$).

575: {\bf Right panel:} Time scales for a SA acceleration model for solar flare

576: conditions and an

577: assumed

578: spectrum of PWT. Model parameters are indicated in the figure

579: ($\tau_p^{-1}\propto \Omega_e{\cal W}/B^2$ is a

580: characteristic wave-particle interaction rate and $\alpha\propto \sqrt{n}/B$).

581: The acceleration, escape, and

582: loss times are indicated by the dotted, dashed, and solid curves, respectively.

583: The thick

584: (red) curves are for electrons and thin (blue) curves are for protons.}

585: \label{timescales}

586: \vspace{-0.5cm}

587: \end{figure}

588: 

589: 

590: 

591: The left panel of Figure \ref{timescales} shows the

592: dispersion relation for PPW and the resonant conditions (straight lines) for an

593: electron and a proton of the specified energy and pitch angle cosine. The main

594: parameter determining

595: these curves and the resulting resonances (intersections of the lines with

596: the curves) is the ratio of the plasma to gyro frequencies

597: \beq\label{alpha}

598: \alpha=\omega_{pe}/\Omega_e=(m_e/m_p)^{1/2}/\beta_{\rm A}=3.2(n/10^{10}

599: \cc)^{1/2}(100 {\rm G}/B).

600: \eeq

601: The right panel of Figure \ref{timescales} shows the timescales associated with

602: the terms in equation (\ref{kineq}), {\it i.e.} loss

603: time

604: $\tau_{\rm loss}=E/{\dot {E}_L}$, diffusion time $\tau_{\rm diff} \sim

605: E^2/D_{EE}$, acceleration time $\tau_{\rm a}=E/A(E)$, and escape time% 

606: \footnote{This is an approximate

607: relation valid for scattering time $\tau_{\rm scat}\ltsim D_{\mu\mu}^{-1}>$

608: longer and

609: shorter than $T_{\rm cross}\sim L/v$, the time for particles to cross the

610: turbulent region

611: of size $L$ (see PL04).}$T_{\rm esc}= T_{\rm cross}(1+T_{\rm cross}/\tau_{\rm

612: scat})$.

613: 

614: \begin{figure}[htbp]

615: \leavevmode

616: \centerline{

617: \includegraphics[width=7.5cm]{epspectra.eps}

618: \includegraphics[width=7.5cm]{espectra_taup.ps}

619: }

620: \caption{\scriptsize

621: {\bf Left panel:}

622: The spectra of accelerated electrons and protons corresponding to the timescales

623: shown in Figure \ref{timescales} (right). The dotted

624: curves give the spectra at the acceleration site (LT) and the dashed curves give

625: the thick target equivalent

626: spectra of escaping particles. The solid (black) line

627: gives the shape of the background (injected) particle distribution. The hatched

628: areas show the energy ranges for production of  frequently observed ranges of

629: HXRs (for electrons) and gamma-ray lines (for protons). Spectral features occur

630: at

631: the intersections and divergences of these timescales (circle and square signs

632: shown here and in Fig. \ref{timescales}; right).

633: {\bf Right panel:} The dependence of electron spectrum on the  ${\cal

634: W}\propto\tau_p^{-1}$.

635: Higher ${\cal W}$'s produce harder spectra

636: and more acceleration vs heating (see PL04 ).

637: }

638: \label{spectra}

639: \vspace{-0.5cm}

640: \end{figure}

641: 

642: {\it The spectral characteristics} of the particles

643: arise from the interplay among the

644: timescales shown above.

645: Figure \ref{spectra} (left) shows

646: the corresponding spectra  of electrons and protons at the LT

647: ($N_{\rm LT}$, dotted) and the equivalent FP thick target spectra (dashed lines;

648: see \eg Petrosian 1973, Park, Petrosian \& Schwartz  1997)

649: \beq\label{thick}

650: N_{\rm FP}(E)={\dot E}^{-1}_L\int_0^E N_{\rm LT}(E')T_{\rm esc}^{-1}(E')dE'.

651: \eeq

652: Spectral breaks

653: occur at energies (circles and squares) when different terms

654: become important. As evident, in general, SA by PWT  produces a

655: ``quasi-thermal''

656: and a harder non-thermal component; \ie it both ``heats" the plasma and

657: accelerates

658: particles. As also seen in the right panel of Figure \ref{spectra}, the relative

659: strength of the two components depends

660: sensitively on the intensity ${\cal W}\propto \tau_p^{-1}$ of the PWT. For

661: higher densities of turbulence the spectra are harder. In addition,

662: the LT spectra are dominated by the

663: thermal component with a steep non-thermal component, while the FP  spectra

664: are harder with little or non-thermal part.  These agree with the observed

665: features discussed below.

666: For low ${\cal W}$, \eg during the build-up

667: and decay phases of a flare, one expects only plasma heating with essentially no

668: non-thermal component (see below and PL04 ).

669: %differences between the LT and FP sources,

670: %and can  explain the observed pre- and post-impulsive phase behaviors and the

671: %impulsive phase soft-hard-soft evolution

672: 

673: 

674: \subsection{Application to Solar Flares}

675: \label{observe}

676: 

677: Ultimately observations must determine the validity of the models.

678: As described below there have been  extensive comparisons of the SA model with

679: observations. To our knowledge, there have not been similarly detailed

680: comparisons for electric field or shock acceleration.

681: 

682: 

683: \subsubsection{Radiative Signatures of Electrons}

684: \label{electron}

685: 

686: {\bf Spatial Structure and Evolution:}

687: The most direct evidence for the presence of

688: PWT is observations by {\it Yohkoh} of distinct LT impulsive HXR emission

689: (Masuda et al. 1994, Masuda 1994), which  have been

690: shown to be common to almost all {\it

691: Yohkoh} (Petrosian, et al. 2002) and {\it RHESSI} (Liu et al. 2004) flares.

692: Petrosian \&

693: Donaghy (1999) have also shown that these observations require an enhanced pitch

694: angle scattering to confine the particles near the LT. Coulomb collisions cannot

695: be

696: this agent because of high losses they entail.

697: The most likely scattering agent is PWT, which can also accelerate

698: particles. Figure \ref{model} (left) shows a  cartoon of the

699: reconnection and acceleration site

700: for a flaring loop with the red foam representing the PWT produced

701: during reconnection. As reconnection proceeds larger closed loops

702: are formed below this site. This simple picture then predicts  a gradual rise

703: of the LT

704: source accompanied with a continuous increase in separation of the FPs. Most

705: strong flares are

706: complex and involve multiple (or arcade) of loops (\eg July 23, 2002 flare;

707: Krucker \ea 2003), which obscures the simple single loop behavior.

708: In  weaker flares, on the other hand, this behavior is lost in the noise.

709: {\it RHESSI} has detected a strong X-class flare (Nov. 3, 2003)

710: consisting of a single

711: loop which shows exactly this behavior (Liu \ea 2004, see Fig. \ref{model},

712: middle). Similar motion but only for the LT source has been seen in

713: several other {\it RHESSI} flares (see \eg Sui \& Holman 2003).

714: In rare cases, specially when the FP sources are weak or occulted by the sun,

715: one can see a double LT source (Fig. \ref{model} right, from Liu 2007,

716: see also Sui \ea 2005) as envisioned by the cartoon on the left.

717: 

718: 

719: \begin{figure}[hbtp]

720: \leavevmode

721: \centerline{

722: \hspace{0.7cm}

723: \includegraphics[width=5.5cm]{f3a.eps}

724: \hspace{-0.5cm}

725: \includegraphics[height=6.0cm, origin=c, angle=90]{f3b.eps}

726: \hspace{-1.0cm}

727: \includegraphics[width=6.7cm,height=6.0cm]{f3c.eps}

728: %\includegraphics[width=7.0cm,height=6.5cm,angle=90]{f3b.eps}

729: %\includegraphics[width=5.5cm,height=5.0cm]{f3c.eps}

730: }

731: \caption{\scriptsize

732: {\bf Left panel:} A schematic representation of the reconnecting field forming

733: closed

734: loops and open coronal field lines. The red foam represents PWT. {\bf Middle

735: panel:}

736: Temporal evolution of LT and FP HXR sources of Nov. 3, 2003 flare.

737: The symbols indicate the source centroids and the colors show the time with a 20

738: sec

739: interval, starting from black (09:46:20 UT) and ending at red (10:01:00 UT) with

740: contours for the last time. The curves connect schematically

741: the FPs and the LT sources for three different times showing

742: the expected evolution for the model at the left; from Liu \ea (2004). {\bf

743: Right panel:}

744: Image of flare of April 30, 2002, with occulted FPs

745: showing an elongated LT source with two distinct peaks as expected from

746: the model in the left (see Liu et al. 2008). The curves representing the

747: magnetic lines (added by

748: hand)

749: show the presumed occulted FPs below the limb (red line); from Liu (2007).

750: }

751: \label{model}

752: %\vspace {-0.5cm}

753: \end{figure}

754: 

755: 

756: 

757: {\bf Spectral Characteristics:}

758: Observations over a broad energy range (\eg Marschh\"{a}user et al.  1994,

759: Dingus et al. 1994, and

760: Park, Petrosian, \& Schwartz 1997) show several spectral breaks. Petrosian et

761: al. (1994) have shown that these features

762: present in the so-called

763: electron-dominated flares

764: cannot be due to transport or optical depth effects

765:  and are natural

766: consequences of the SA by PWT (Park et al. 1997).

767: Early high resolution observations (Lin et al.  1981)

768: showed spectral steepening below a few tens of keV, which is also consistent

769: with the SA model

770: (Hamilton \& Petrosian 1992). {\it RHESSI}'s imaging spectroscopy has clarified

771: this situation

772: considerably. In general, during the impulsive phase the spectra are dominated

773: by a non-thermal

774: component, with the LT always softer than the FPs, and there is a

775: quasi-thermal component (mostly from the LT)

776: occurring sometimes before the impulsive phase and always

777: during the decay phase, which is in concordance with the features of the

778: accelerated electrons described above.

779: Figure \ref{ionspectra} shows these observed characteristics of the LT and FP

780: sources 

781: for two typical {\it RHESSI} x-ray flares

782: fitted with theoretical spectra from the SA model.

783: 

784: \begin{figure}[hbtp]

785: \leavevmode

786: \centerline{

787: \includegraphics[width=7.5cm, height=5.5cm]{photonspec1.ps}

788: \includegraphics[width=7.5cm, height=5.5cm]{photonspec2.ps}

789: }

790: \caption{\scriptsize

791: {\bf Left panel:} A fit to the total (black), FP (green) and LT (red) spectra of

792: the 

793: first HXR pulse of the Sep.

794: 20th 2002 flare observed by {\it RHESSI}. Model parameters are indicated in the

795: figure. 

796: The dashed and dotted

797: lines are model spectra from equations (\ref{kineq}) and (\ref{thick}) showing

798: the 

799: presence of a

800: quasi-thermal (LT) and a non-thermal (mostly FPs) component. The solid line

801: gives the 

802: sum of the two. The blue

803: dashes indicate the level of the background radiation.

804: {\bf Right panel:} Same as the left panel but for the second HXR peak. Note

805: that 

806: compared with the model parameters

807: for the first HXR pulse, both the gas density and the turbulence energy density 

808: increase during the second pulse,

809: resulting more thermal emission and harder photon spectra.

810: } 

811: \label{ionspectra}

812: \end{figure}

813: 

814: {\bf Spectral Evolution:} It has also been known for

815: sometime that some observations disagree with the model where  \underline{all}

816: the released energy

817: goes directly to non-thermal electrons during {\it the impulsive phase}.

818: Among these are precursor soft X-ray emission referred to as {\it preheating}

819: and

820: a slower than expected temperature decline in the decay phase (see, {\it e.g.}

821: McTiernan et al.  1993)%

822: \footnote{The expected Neupert (1968) relation also does not seem to

823: hold (see Dennis \& Zaro 1993; Veronig et al. 2005).}.

824: In several {\it RHESSI} limb flares Jiang \ea (2006) find that during the decay

825: phase the

826: (resolved) LT

827: source continues to be confined; it does not extend to the FPs as one would

828: expect

829: if the evaporation

830: fills the whole loop with hot plasmas.

831: %Figure \ref{spectra} (right) shows the

832: %temporal evolution of the temperature and emission measure from spectral

833: %fitting,

834: %and the density from the

835: %observed source volume evolution  for the 09/20/2002 flare. The bottom  shows

836: %that the

837: Moreover,  the observed energy decay rate is calculated to be

838: higher than the radiative cooling rate

839: but much lower than the (Spitzer 1961) conduction

840: rate. The latter and the confinement of the LT source require suppression of the

841: conduction and a continuous energy input.

842: PWT can be the agent for both. It can reduce the scattering mean

843: free path and energize the plasma continuously during the decay phase (albeit at

844: a

845: diminishing rate).

846: 

847: The observed spectral  breaks and the temporal evolution also agree with the

848: model results discussed above. During the build-up and decay (\ie pre- and

849: post-impulsive)

850: phases when the level of PWT is low it only manages to ``heat" the plasma with

851: little

852: acceleration into a non-thermal tail. Particle acceleration begins in earnest

853: during

854: the impulsive phase when the rate  of reconnection and production of PWT rises

855: and then falls

856: after reaching high levels. During this rise and fall the spectra undergo

857: soft-hard-soft

858: evolution as observed in most flares.

859: 

860: \begin{figure}[hbtp]

861: \leavevmode

862: \centerline{

863: %\hspace{0.7cm}

864: \includegraphics[width=7.0cm]{epspectra1.eps}

865: %\hspace{-0.5cm}

866: \includegraphics[width=7.0cm]{epalpha.ps}

867: }

868: \caption{\scriptsize

869: {\bf Left panel:}

870: Same as left panel of Figure \ref{spectra} but for a different $\alpha$ showing

871: the dependence of relative electro and proton spectra on $\alpha$.

872: {\bf Right panel:} Variation with $\alpha$ of the ratio of the energies of the

873: accelerated electrons and protons in the spectral energy range shown by the

874: shade region in the left panel.

875: }

876: \label{epratio}

877: \vspace {-0.5cm}

878: \end{figure}

879: 

880: \subsubsection{Radiative Signatures of Protons and Ions}

881: 

882: Observations  of gamma-ray lines in larger solar flares provide

883: evidence for acceleration of protons and ions. The SA model has been

884: the working hypothesis

885: here as well  (see works by Ramaty and collaborators earlier, and Murphy and

886: Share recently). For these

887: flares

888: the ratio of energy content of accelerated electrons to

889: protons

890: (in the hatched range of Fig. \ref{spectra}, left)

891: ranges from 100 to 0.03

892: (Mandzhavidze \& Ramaty 1996);  this ratio is  larger in electron dominated

893: flares.

894: On the other hand, simple theoretical models, for example

895: acceleration by Alfv\'{e}n waves commonly used, favor a much smaller ratio.

896: PL04  show that with a more complete treatment of the

897: wave-particle interaction a barrier appears for proton acceleration,

898: which reduces the number of accelerated protons dramatically,

899: in a way that the above ratio is very sensitive to the plasma parameter

900: $\alpha$ in equation (\ref{alpha}).

901: Figure \ref{epratio} shows this dependence. The left panel is same as the left

902: panel in Figure \ref{spectra} but for a smaller $\alpha$. The right panel shows

903: the variation with $\alpha$ of the ratio of the energy of electron to protons

904: (in the respective shade region).

905: One consequence of this is

906: that the proton acceleration will be more efficient in larger loops, where the

907: $B$ field is

908: weaker, and at late phases, when  due to evaporation $n$ is higher.

909: This can explain the offset of the centroid of the gamma-ray line emissions from

910: that of HXRs seen by {\it RHESSI} (Hurford et al.  2003).

911: It can also account for the observed delay of nuclear line emissions relative

912: to HXR emissions seen in some flares (Chupp et al.  1990).

913: This trend is also true for other ions because the acceleration barrier

914: moves to lower energies with decreasing $\alpha$.

915: %This effect is shown on

916: %Figure \ref{He} (left) for $^4$He and will be true for all ions with the same

917: or

918: %similar charge-to-mass ratios.

919: In general, higher densities, lower magnetic fields favor

920: acceleration of ions

921: versus the electrons and can explain the diversity in their relative observed

922: values both at the flare site and in the interplanetary space.

923: 

924: \subsubsection{SEP Spectra and Abundances}

925: 

926: It is commonly believed that the observed relative abundances of ions in SEPs

927: favor a SA model

928: ({\it e.g.}

929: Mason et al. 1986 and Mazur et al. 1992). More recent observations have

930: confirmed

931: this picture (see

932: Mason et al. 2000, 2002, Reames et al. 1994 and 1997, and Miller 2003).

933: There are similarities and

934: differences among the spectra and abundances of different ions

935: from event to event.

936: One of the most vexing problem of SEPs has been the  enhancement of $^3$He

937: in the so-called {\it impulsive events}, which sometimes can be $3-4$ orders of

938: magnitude above the

939: photospheric value. As stated at the outset these events, which we prefer to

940: call

941: \3he-rich rather than impulsive, is the focus of thees proceedings.  There has 

942: been many attempts to  explain this

943: and the related enhancements of heavy ions observations. This will be the

944: subject of the next sections.

945: 

946: \section{Enhancement of \3he and Heavy Ions}

947: 

948: As stressed in the previous chapters one of the most puzzling aspects of

949: observations of SEPs is the enhancement of \3he and heavy ions  relative to

950: their photospheric

951: values in many of the so-called  impulsive events. Over the years there has been

952: many attempts to account for these observations. In this section we first

953: present a

954: brief

955: review of past attempts at tackling this problem and then describe how well the

956: SA

957: model described above can account for these aspects of solar flares.

958: 

959: 

960: \subsection{Past Works}

961: 

962: 

963: Most of the proposed models, except the  Ramaty and Kozlovsky (1974) model based

964: on  spalation (which has many problems), rely on resonant wave-particle

965: interactions

966: to produce the observed enhancement. The  earliest theoretical

967: models addressing this issue are those of Ibragimov and Kocharov (1977) and Fisk

968: (1978).

969: 

970: \begin{itemize}

971: 

972: \item

973: 

974: Ibragimov and Kocharov claim that under the influence of Langmuir turbulence the

975: plasma ion temperature will increase linearly in time with a rate proportional

976: to the momentum diffusion coefficient. Citing another paper they also claim that

977: the diffusion rate will be proportional to the Coulomb charge times the

978: gyrofrequency which favors heating of \3he relative to protons and \he4. A

979: second stage of acceleration with a velocity or energy threshold for all

980: particles will then accelerate a larger fraction  of \3he ions than \he4 and

981: proton. The parameters of the model can be adjusted to obtain the observed

982: enhancements.

983: \item

984: 

985: Fisk's model relies on the fact that  \3he  can interact more efficiently

986: than

987: \he4 with (electrostatic hydrogen-cyclotron) waves having frequency between

988: proton and \he4 gyrofrequencies. Damping of these waves preferentially heats

989: \3he  ions. There are two major shortcomings in this model. The main

990: difficulty

991: here is that to have sufficient overheating one requires excitation of waves

992: close to \3he  gyrofrequency (=4/3 of \he4) which can be produced if electron

993: to

994: proton temperature ratio and/or the \he4 to hydrogen ratio are large. Again,

995: this model does not answer the full question of production of enhanced

996: population of non-thermal \3he ions. It leaves this to a secondary process which

997: also

998: must satisfy the requirement of having a threshold velocity for acceleration

999: near

1000: the thermal velocity of the heated \3he ions.  Fisk also

1001: tries to explain the enhancement of the heavy elements the same way but relying

1002: on the resonant interactions with the second gyro-harmonics. This aspect also

1003: may not be correct.

1004: 

1005: \item

1006: 

1007: Temerin and Roth (1992) use essentially same idea but with

1008: electromagnetic hydrogen-cyclotron waves. They carry out numerical simulation

1009: based on somewhat artificial model (waves moving down the variable field lines

1010: and \3he  ions moving up that interact when the resonant condition is

1011: satisfied).

1012:  They also produce some heating of \3he  ions but do not discuss the spectrum

1013: nor

1014: the expected ratio of \3he/\he4. Again, like  in the Fisk model first  and

1015: second harmonics

1016: ($n=1$ and 2) are invoked for \3he  and heavy ion heating, respectively.

1017: However, there is no estimate of heavy ion fluxes. Curiously, Fisk is not

1018: acknowledged except in passing mentioning only the shortcoming  mentioned above.

1019: 

1020: \item

1021: 

1022: Miller and Vi\~nas (1993) expand on Temerin and Roth work by

1023: inclusion of more waves and more-realistic plasma conditions. Most of this

1024: paper is devoted to calculating the growth rate  and dispersion relation of

1025: waves excited by a thermal, isotropic ``beam" of electrons superimposed as  a

1026: bump in a somewhat cooler thermal plasma consisting of electrons,

1027: protons and alpha particles. The acceleration of particles is carried out using

1028: orbit calculations for test particles. As in the above works it merely leads to

1029: some heating of iron and \3he  ions. It is then stated that this heating is

1030: more

1031: than what they expect to occur for other ions (\eg CNO) and \he4, respectively.

1032: The required ``beam" densities are much higher than those required for

1033: describing

1034: the Type III radio emission. (These authors provide a good summary and analysis

1035: of earlier works).

1036: 

1037: \item

1038: 

1039: Zhang (1995) uses a more refined version of  the Fisk model and evaluates the

1040: degree of overheating of \3he ions relative to \he4. In addition he includes a

1041: simple calculation of subsequent acceleration by a ``Fermi" process and derives

1042: a general spectrum. However, these spectra are not compared with the observed

1043: ones; they are essentially exponential in form and will not fit well to the

1044: observations. Instead he evaluates the abundance ratios and plots these versus

1045: some important parameters which can be compared with the gross features of the

1046: observations.

1047: 

1048: 

1049: \item

1050: 

1051: Paesold, Kallenbach and Benz (2003)  rely on a somewhat different

1052: aspect of the acceleration process. In their model

1053: waves

1054: are excited by anisotropic distribution of electrons (so-called Fire Hose

1055: Instability) with a growth rate  that is slightly higher at the \3he

1056: gyrofrequency

1057: than that of \he4. It is then claimed that this can lead to a much larger

1058: (exponential)

1059: difference in the wave energy density at the corresponding frequencies. With a

1060: larger wave density at its disposal \3he  acceleration proceed more rapidly. The

1061: acceleration method is used is the same as in Miller and Vi\~nas discussed

1062: above. However, unlike the above works this paper addresses the spectral

1063: differences between \3he  and \he4, but there is no comparison with actual

1064: observed spectra. This model also predicts that there should be i) many more

1065: accelerated protons (which is swept under the rug invoking some energy

1066: limitation) and ii) a \underline{suppressions} of iron and other heavier ions

1067: (whose

1068:  $Q/A$  is less than that of \he4 or CNO) instead of the observed

1069: enhancements.

1070: 

1071: \end{itemize}

1072: 

1073: In summary the common theme among all these models is that they use the

1074: difference between the $Q/A$ of \3he and \he4 and require some non-Maxwellian

1075: electron distribution to excite the waves that then preferentially heat

1076: \3he ions. Some also require very special conditions.

1077: The outcomes of all these works are in general some differential heating with

1078: the actual acceleration

1079: often relegated to a subsequent process. There is little or no comparison with

1080: observed spectra and every model requires some additional factor to explain the

1081: enhancements observed for iron and heavier ions.

1082: 

1083: In contrast to this the model proposed by Liu, Petrosian and Mason (2004, 2006)

1084: based on the SA by PWT (PL04) treats heating and acceleration as a single

1085: process, can

1086: reproduce some of the observed spectral features, and can account for the range

1087: of the observed

1088: enhancement. On the other hand, at this stage of its development this model does

1089: not address the details of the production of the waves or turbulence and relies

1090: on injection, presumably during the reconnection process,  and the subsequent

1091: cascade and damping of the waves. Some of the salient results of this model are

1092: described next.

1093: 

1094: \subsection{Enhancements and Stochastic Acceleration}

1095: 

1096: In this section we describe the role the SA model described in the previous

1097: sections can play in the enrichment of \3he and other elements. The formalism

1098: described above can be applied to the acceleration of all ions in the same way

1099: used to fit the radiative signatures of the accelerated electrons and protons to

1100: observations. All the results presented below are based on acceleration by PPWs

1101: and most of the results are from Liu \ea (2004, 2006).

1102: 

1103: 

1104: 

1105: \begin{figure}[hbtp]

1106: \centerline{

1107: \includegraphics[height=7.5cm,angle=0]{ppwdispHe.eps}

1108: \includegraphics[height=7.5cm,angle=0]{timesHe.eps}

1109: }

1110: \caption

1111: {\scriptsize

1112: {\bf Left panel:} Dispersion relations for parallel propagating waves and

1113: resonance conditions for \3he and \he4 for the specified energies and three

1114: different values of $\alpha$.  Note that the resonance lines for \3he start (at

1115: $k$=0) at its gyrofrequency (divided by $\gamma\sim 1$), which is larger (in

1116: absolute value) than that of \he4. As a result at low energies \3he  has more

1117: resonant intersections (filled circles)  than \he4. {Right panel:} Acceleration,

1118: escape and energy time scales

1119: for \3he an \he4. The main difference here is the shorter acceleration time for

1120: \3he at lower energies because of its stronger interactions with waves.

1121: }

1122: \label{Hetimes}

1123: \vspace{0.5cm}

1124: \end{figure}

1125: 

1126: Figure \ref{Hetimes} (left) shows the dispersion relation near the proton and

1127: \he4 gyrofrequencies and the resonant conditions for two different energies of

1128: \3he and \he4 ions. The presence of fully ionized \he4 (about 8\% of protons) in

1129: the background plasma divides the Alfv\'en wave-proton cyclotron branch into two

1130: distinct modes which we have labeled proton cyclotron (PC) and He cyclotron

1131: (HeC) branches. Because of the gap between these two modes at low values of

1132: the wavevector $k$,  low energy \he4 ions resonate only with the HeC branch

1133: while

1134: \3he ions can resonate with both branches and are

1135: accelerated more efficiently. The right panel of this figure depicts the

1136: relevant

1137: time scales showing the much shorter acceleration time for \3he than \he4 at low

1138: energies. This is the primary cause of the \3he enrichment.

1139: 

1140: As a result of this more efficient acceleration most \3he ions are accelerated

1141: to form a non-thermal distribution. While \he4 ions are heated up forming a low

1142: energy quasi-thermal bump with only a small fraction reaching to high energies.

1143: Figure \ref{Hespectra} shows model spectra fitted to two \3he rich SEP events

1144: displaying the above features.  Another result of

1145: the efficient acceleration is that one expects a high degree of acceleration of

1146: \3he even in smaller events such that the \3he/\he4 abundance ratio is expected

1147: to decrease with increasing fluence of the event. This is what is seen (see Fig.

1148: XXX in Chapter 2) and is demonstrated in the right panel of Figure

1149: \ref{Hespectra} where we show spectra for three different values of the

1150: acceleration rate $\tau_p^{-1}$. For higher rates more of the \he4 ions in the

1151: quasi-thermal tail reach into the non-thermal tail decreasing the \3he

1152: enrichment and the \3he/\he4 ratio.

1153: 

1154: \begin{figure}[hbtp]

1155: \centerline{

1156: \includegraphics[width=5.8cm,angle=0]{he34Aug.eps}

1157: \hspace{-0.5cm}

1158: \includegraphics[width=5.8cm,angle=0]{spectra_taup.ps}

1159: \hspace{-0.5cm}

1160: \includegraphics[width=5.8cm,angle=0]{spectra_alpha.eps}

1161: }

1162: \caption

1163: {\scriptsize Fits to \3he and \he4 spectra of two \3he rich SEP events. Note

1164: that because of the shorter acceleration time of \3he at low energies most of

1165: the \3he is accelerated into super thermal tail while most of \he4 ions are

1166: heated up to a high 'temperature' quasi-thermal distribution with only a smaller

1167: fraction

1168: forming a non-thermal tail. The middle and right panels also show two additional

1169: sets of

1170: model spectra with somewhat higher and lower values of $\tau_p^{-1}$, which is

1171: proportional to the turbulence energy density or the rate of acceleration,  and

1172: $\alpha$, respectively.

1173: }

1174: \label{Hespectra}

1175: \vspace{0.5cm}

1176: \end{figure}

1177: 

1178: The dependence of \3he enrichment on model parameters is further 

1179: demonstrated in Figure \ref{Hetaup}. The most influential parameter is the

1180: acceleration rate $\tau_p^{-1}\propto {\cal W}$ and there is some variation with

1181: $\alpha$ at low

1182: acceleration rates. There is little dependence on the (unknown) spectral

1183: parameters of the turbulence. The trend and the range of the

1184: variations presented here are in agreement with the observations described in

1185: chapter 2. It remains to be seen whether or not the spectral trend presented in

1186: Figure \ref{Hespectra} (middle and right) also agrees with observations. It

1187: should also be possible

1188: to use the observed distributions of \he4 and \3he fluences or peak fluxes to

1189: further test this promising model and constrain its parameters. This will

1190: require a sample of events with known selection criteria and observational

1191: biases.

1192: 

1193: \begin{figure}[hbtp]

1194: \centerline{

1195: \includegraphics[height=7.0cm,angle=0]{ratio_taup.eps}

1196: \includegraphics[height=7.0cm,angle=0]{ratio_taupk.eps}

1197: }

1198: \caption

1199: {\scriptsize Variation with acceleration rate (or level of turbulence) of the

1200: \3he/\he4 fluence ratio at 2 MeV per nucleon for different values of $\alpha$

1201: (left) and the

1202: spectral break wave vector $k_\max$ (right). The black lines are for best fit

1203: values of $\alpha=0.46$ and  $k_\min=0.1 k_\max$ for the 30 Sep. 1999 event

1204: shown

1205: in Figure \ref{Hespectra} (middle and left). These curves  show the possibility

1206: of

1207: obtaining a large range of the ratio of \3he/\he4 and that the main

1208: characteristics of turbulence affecting this ratio is the energy density ${\cal

1209: W}\propto \tau_p^{-1}$ with smaller effect due to $\alpha$ and a relatively

1210: insignificant effect due to unknown details of the turbulence spectrum (see Liu

1211: \ea 2006).

1212: }

1213: \label{Hetaup}

1214: %\vspace{0.5cm}

1215: \end{figure}

1216: 

1217: \section{Summary and Discussion}

1218: 

1219: We have discussed possible mechanism of  accelerations in solar flares and SEPs,

1220: and pointed out that in all

1221: mechanisms plasma  waves and turbulence play a major role. We have then argued

1222: that in solar flares the most likely mechanism is stochastic acceleration  by 

1223: plasma  waves and turbulence and have demonstrated

1224: that the predictions of this model are consistent with most of the observed

1225: flare

1226: radiative signatures. We have also shown that the observed \3he enrichment and

1227: the spectra of \3he and \he4 can  be described by such a model without

1228: requiring any special conditions. The same model that successfully describes the

1229: observed radiation can also account for this long standing puzzle. Unlike

1230: previous models, which addressed only the enhancement (without detailed

1231: comparison with observed spectra) and required  special conditions and/or two

1232: stage acceleration, the  model presented here neither requires special

1233: conditions nor another acceleration mechanism.

1234: 

1235: Another important problem of SEP observations is enhancement of abundances of

1236: heavier ions with the enhancement increasing with ionic mass or mass-to-charge

1237: ratio described in Chapter 2. Here the situation is more complicated because of

1238: the

1239: uncertainty about the charge state of these ions and the observations of

1240: dependence of the charge state on the ion energy discussed in previous chapters.

1241: These observations also favor

1242: acceleration in relatively high densities of the lower corona, where the

1243: stochastic acceleration 

1244: invoked here occurs. Figure  \ref{heavies} (left)  shows abundances of a large

1245: number of ion species obtained from a

1246: composite of many SEP events (red points, Mason, private communication). The

1247: black points and the line show

1248: our preliminary attempt (Liu \& Petrosian in preparation) to fit these

1249: observations. Fits to protons \3he and

1250: \he4 are as above and in plotting the line we have assumed a charge of 16 for

1251: all the elements above iron.

1252: 

1253: \begin{figure}[hbtp]

1254: \centerline{

1255: \includegraphics[height=7.0cm,angle=0]{heavyionflux.ps}

1256: \includegraphics[height=7.0cm,angle=0]{oxygen.ps}

1257: }

1258: \caption

1259: {\scriptsize {\bf Left panel:} A preliminary fit by the SA model to different

1260: ion enhancements obtained from a composite of many events (red points from

1261: Mason, private communication). The open circles and the black curve are the

1262: model

1263: results assuming a charge of 16 for all elements heavier than iron (Liu \&

1264: Petrosian, in preparation). This, of course, is inappropriate for intermediate

1265: ions where the time-independent model described here cannot describe the

1266: observations. {\bf Right panel:} A preliminary result from a time-dependent

1267: treatment showing the evolution of $^{16}$O (dashed lines) spectrum and the time

1268: integrated spectrum (solid green) fitted to  observed spectrum of fluences of

1269: $^{16}$O and \he4 for a typical event.

1270: }

1271: \label{heavies}

1272: %\vspace{0.5cm}

1273: \end{figure}

1274: 

1275: This scenario in the form presented here will encounter difficulty with the

1276: observations of the intermediate ions (\eg CNO). We note that most of the

1277: observed spectra refer to the event fluence. The models described above are

1278: based on the \underline{time-independent} (or the steady state) equations. This

1279: is valid when the acceleration time is shorter than the dynamic time scale of

1280: the

1281: event (\eg event duration), in which case the steady state results are good

1282: approximation. A more accurate analysis requires a \underline{time-dependent}

1283: analysis. Figure \ref{heavies} (right) shows a preliminary result from such a

1284: model (East, Petrosian \& Liu, in preparation) from such a model. The dashed

1285: lines

1286: show the time evolution of the spectrum of $^{16}$O and the solid green line the

1287: time integrated spectrum appropriate for the fluence spectrum shown by the green

1288: squares. This model seems to provide a reasonable fit (except at low energies

1289: where the uncertainties are large) for both  $^{16}$O  and \he4. 

1290: Clearly more data and more detailed model are

1291: required to put this comparison on the same footing as those described above for

1292: electrons, protons and He ions.

1293: 

1294: \section{REFERENCES}

1295: 

1296: %\vspace{0.4cm}

1297: %\setlength{\baselineskip}{8pt}

1298: %{\footnotesize

1299: %\noindent

1300: \def\et{{\it et~al.}}

1301: \def\aa{{\it Astr. \& Astrophys.}}

1302: \def\anyas{{\it Ann. N.Y. Acad. Sci.}}

1303: \def\apj{{\it Ap.~J.},\,\,}

1304: \def\apjl{{\it Ap.~J. (Letters),\,\,}}

1305: \def\ass{{\it Ap. Space Sci.}}

1306: \def\jgr{{\it J.~Geophys. Res.}}

1307: \def\mnras{{\it M.N.R.A.S.}}

1308: \def\nature{{\it Nature}}

1309: \def\prd{{\it Phys. Rev.~D}}

1310: \def\pra{{\it Phys. Rev.~A}}

1311: \def\ssr{{\it Space Sci. Rev.}}

1312: \def\rmp{{\it Rev. Mod. Phys.}}

1313: \def\sphys{{\it Solar Phys.}}

1314: \def\JL{Leach,~J.\ }

1315: \def\VP{Petrosian,~V.\ }

1316: 

1317: 

1318: 

1319: \def\refer{ \par \noindent \hangindent=2pc \hangafter=1}

1320: \baselineskip = 10 true pt

1321: %\bigskip

1322: 

1323: \refer Amato, E., \& Blasi, P. 2005, MNRAS, 364, 76

1324: \refer Andr\'e, Mats 1985, {\it J. Plasma Phys.,} 33, 1

1325: \refer Boris, J. P., Dawson, J. M., Orens, J. H., \& Roberts, K. V. 1970, 

1326: {\it Phys. Rev. Lett.}, 25, 706

1327: \refer Cassak, P. A., Drake, J. F. \& Shay, M. A. 2006, \apj 644, L145

1328: \refer Chandran, B. 2005, {\it Phys. Rev. Lett.}, 95, 265004

1329: \refer Cho, J., \& Lazarian, A. 2002, {\it Phys. Rev. Lett.}, 88, 245001

1330: \refer Cho, J., \& Lazarian, A. 2006, {\it Particle Acceleration by MHD Turbulence}, 

1331: \apj 638, 811

1332: \refer Chupp, E. L. 1990, {\it Science}, 250, 229

1333: \refer Cranmer, S. R., \& van Ballegooijen, A. A. 2003, \apj 594, 573

1334: \refer Dennis, B. R., \& Zarro, D. M. 1993, \sphys, 146, 177

1335: \refer Dingus, B. L. et al. 1994, {\it AIP Conf. Proc.} eds. J. M. Ryan \& W. T. Vestrand, 294, 177

1336: \refer Drake, J. F., Paper presented at {\it Krakow Conference on Relativistic Jets}, June  2006

1337: \refer Ellison, D. C., Decourchelle, A, \& Ballet, J. 2005, \aa,  429, 569

1338: \refer Farmer, A., Goldreich, P. 2004, \apj 604, 671

1339: \refer Fisk, L. A. 1978, \apj 224, 1048

1340: \refer Giacalone, J. 2005, \apjl 628, L37

1341: \refer Goldreich, P., \& Sridhar, S. 1995, \apj 438, 763

1342: \refer Goldreich, P., \& Sridhar, S. 1997, \apj 485, 680

1343: \refer Hamilton, R. J. \& Petrosian, V. 1992, \apj 398, 350

1344: \refer Holman, G. D. 1985, \apj {\bf 293}, 584

1345: \refer Holman, G. D. 1996a, {\it BAAS}, 28, 939

1346: \refer Holman, G. D. 1996b,  {\it AIP Conf. Proc.} eds. R. Ramaty et al., 374, 479

1347: \refer Hoshino, M., Paper presented at {\it COSPAR Scientific Assembly}  Beijing, July 2006

1348: \refer Hurford, G. J. et al. 2003, \apj 595, 77L

1349: \refer Ibragimov, I. A. \& Kocharov, G. E. 1977, Proc. 15th Int, Cosmic Ray Conf. (Plovdiv), 11,  340

1350: \refer Iroshnikov, P., 1963, {\it Astron. Zh.}, 40, 742 (English version: 1964,  Sov. Astron., 7, 566)

1351: \refer Jiang, Y. W. et al. 2006,  \apj 638, 1140 

1352: \refer Jones, F. C. 1994, {\it Ap. J. Suppl.}, 90, 561

1353: \refer Jokipii, R. J. 1987, \apj 313, 842

1354: \refer Kraichnan, R., 1965, Phys. Fluids, 8, 1385

1355: \refer Lin, R. P. et al. 1981, \apjl 251, L109

1356: \refer Lithwick, Y., \& Goldreich, P. 2001, \apj 562, 279 

1357: \refer Litvinenko, Y. E. 2003, {\it Solar Phys.}, 212, 379

1358: \refer Liu, S., Petrosian, V., \& Mason, G. 2004, \apjl 613, L81 

1359: \refer Liu, S., Petrosian, V., \& Mason, G. 2006, \apj 634, 462 

1360: \refer Liu, W. et al. 2004, \apj 611, 53L

1361: \refer Liu, W.  2007, from PhD thesis, Stanford University

1362: \refer Luo, Q. \& Melrose, D. 2006; astro-ph/0602295

1363: \refer Kulsrud, R.M. 2005, {\it Plasma Physics for Astrophysics}, Princeton Univ. Press

1364: \refer Mandzhavidze, N., \& Ramaty, R. 1996, {\it BAAS}, 28, 858

1365: \refer Marschh\"{a}user, H. et al. 1994, {\it AIP Conf. Proc.} eds. J. M. Ryan \& W. T. Vestrand 294, 171

1366: \refer Mason, G. M. et al. 1986, \apj 303, 849

1367: \refer Mason, G. M., Dwyer, J. R., \& Mazur, J. E. 2000, \apj 545, 157L

1368: \refer Mason, G. M. et al. 2002, \apj 574, 1039

1369: %\refer Mason, G. M. et al. 2004, \apj 605, 555

1370: \refer Masuda, S. et al. 1994, {\it Nature}, 371, 495

1371: \refer Masuda, S. 1994, University of Tokyo, Ph.D. Thesis

1372: \refer Mazur, J. E. et al. 1992, \apj 401, 398

1373: \refer McTiernan, J. M. et al. 1993, \apjl 416, L91

1374: \refer Miller, J. A.\ 2003, {\it COSPAR Colloquia Series} Vol. 13, 387

1375: \refer Miller, J. A.,LaRosa, T.N., \& Moore, R.L. 1996, \apj 445, 464

1376: \refer Miller, J. M. \&  Reames, D. V. 1996, {\it AIP Conf. Proc.} eds. R. Ramaty et al., 374, 450

1377: \refer Miller, J.A.\& Vi\~vas, A. F. 1993, ApJ, 412, 386

1378: \refer Neupert, W.M. 1968, \apj 153, L59

1379: \refer  Park, B. T., Petrosian, V., \& Schwartz, R. A. 1997, \apj 489, 358

1380: \refer Paesold, G., Kallenbach, R. \& Benz, A. O. 2003, ApJ 582, 495 

1381: \refer Petrosian, V. 1973, \apj 186, 291

1382: \refer Petrosian, V., \& Donaghy, T. Q. 1999, \apj 527, 945 

1383: \refer Petrosian, V.,  Donaghy, T. Q., \& McTiernan, J. M. 2002, \apj 569, 459 

1384: \refer Petrosian, V., \& Liu, S. 2004, \apj 610, 550 {\bf PL04}

1385: \refer Petrosian, V., McTiernan, J.M., \& Marschh\"{a}user, H. 1994, \apj 434, 744

1386: \refer Petrosian, V,, Yan, H. \& Lazarian, A. 2006, \apj 644, 603 

1387: \refer Pryadko, J., \& Petrosian, V. 1997, \apj 482, 774 

1388: \refer Pryadko, J., \& Petrosian, V. 1998, \apj 495, 377 

1389: \refer Pryadko, J., \& Petrosian, V. 1999, \apj 515, 873 

1390: \refer Ramaty, R. \& Kozlovsky, B. 1974, ApJ 193, 729

1391: \refer Reames, D. V., Meyer, J. P., \& von Rosenvinge, T. T.\ 1994, ApJS, 90, 649

1392: \refer Reames, D. V. et al. 1997, \apj 483, 515

1393: %\refer R\"onnmark, K. 1983, {\it Plasma Phys.}, 25, 669

1394: \refer Sui, L., \& Holman, G. D. 2003, \apjl 596, L251

1395: \refer Sui, L., Holman, G. D., \& Dennis, B. R. 2005, \apj 626, 1102

1396: \refer Swanson, D. G. 1989, {\it Plasma Waves} (New York: Academic Press)

1397: \refer Temerin, M. \&  Roth, I. 1992, ApJ 391, L105

1398: \refer Tsuneta, S. 1985, \apj 290, 353

1399: \refer Veronig, A. M. et al. 2005, \apj 621, 482

1400: \refer Yan, H., \& Lazarian, A. 2002, {\it Phys. Rev. Lett.}, 89, 2881102

1401: \refer Zenitani, S. \& Hoshino, M. 2005, \apj 618, L111

1402: \refer Zhang, T. X. 1995, ApJ, 449, 916

1403: \refer Zhou, Y., \& Matthaeus, W.H. 1990, {\it J. Geophys. Res.} 95, 14881

1404: 

1405: 

1406: \end{document}

1407: 

1408: 

1409: 

1410: