1: %\documentclass[prb,preprint,showpacs]{revtex4}

2: 

3: \documentclass[prb,tighten]{revtex4}

4: 

5: %\documentclass[pre,twocolumn,showpacs]{revtex4}

6: 

7: %\documentclass[pre,preprint,draft,showpacs,tighten]{revtex4}

8: 

9: %\usepackage[spanish]{babel}

10: 

11: %\usepackage[cp850]{inputenc}

12: 

13: %\usepackage[latin1]{inputenc}

14: 

15: %\usepackage{showkeys}% Muestra etiquetas de formulas, figuras, referencias,...

16: 

17: % Include figure files

18: 

19: %\usepackage{dcolumn}% Align table columns on decimal point

20: 

21: %\usepackage{bm}% bold math

22: 

23: %\psdraft

24: 

25: 

26: 

27: 

28: 

29: 

30: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

31: 

32: \usepackage{epsfig}

33: 

34: \usepackage{graphicx}

35: 

36: 

37: 

38: %TCIDATA{OutputFilter=Latex.dll}

39: 

40: %TCIDATA{LastRevised=Friday, December 23, 2005 14:49:30}

41: 

42: %TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}

43: 

44: 

45: 

46: 

47: 

48: \begin{document}

49: 

50: 

51: 

52: \title{Mass and heat fluxes for a binary granular mixture at low-density}

53: \author{Vicente Garz\'{o}}

54: \email[E-mail: ]{vicenteg@unex.es}

55: \address{Departamento de F\'{\i}sica, Universidad de Extremadura, E-06071

56: Badajoz, Spain}

57: \author{Jos\'e Mar\'{\i}a Montanero}

58: \email[E-mail: ]{jmm@unex.es}

59: \address{Departamento de Electr\'onica e Ingenier\'{\i}a Electromec\'anica,

60: Universidad de Extremadura, E-06071 Badajoz, Spain}

61: \author{James W. Dufty}

62: \email[E-mail: ]{dufty@phys.ufl.edu}

63: \address{Department of Physics, University of Florida, Gainesville, Florida

64: 32611}

65: 

66: 

67: 

68: \begin{abstract}

69: 

70: The Navier--Stokes order hydrodynamic equations for a low density

71: granular mixture obtained previously from the Chapman--Enskog

72: solution to the Boltzmann equation are considered further. The six

73: transport coefficients associated with mass and heat flux in a

74: binary mixture are given as functions of the mass ratio, size

75: ratio, composition, and coefficients of restitution. Their

76: quantitative variation across this parameter set is demonstrated

77: using low order Sonine polynomial approximations to solve the

78: exact integral equations. The results are also used to quantify

79: the violation of the Onsager reciprocal relations for a granular

80: mixture. Finally, the stability of the homogeneous cooling state

81: is discussed.

82: 

83: \end{abstract}

84: 

85: 

86: 

87: \pacs{ 05.20.Dd, 45.70.Mg, 51.10.+y, 47.50.+d}

88: \date{\today}

89: \maketitle

90: 

91: \draft

92: 

93: 

94: 

95: \section{Introduction}

96: \label{sec1}

97: 

98: The relevance and context of a hydrodynamic description for

99: granular gases remains a controversial topic. At sufficiently low

100: density the origin of hydrodynamics can be studied from Boltzmann

101: kinetic theory. \cite{GS95,BP04} To obtain hydrodynamics from the

102: Boltzmann kinetic equation the essential assumption is the

103: existence of a ``normal'' solution, defined to be one for which

104: all space and time dependence occurs through the macroscopic

105: hydrodynamic fields. \cite{CC70} This solution, together with the

106: macroscopic balance equations, leads to a closed set of

107: hydrodynamic equations for these fields. The Chapman--Enskog

108: method provides a constructive means to obtain an approximation to

109: such a solution for states whose spatial gradients are not too

110: large. In this general context, the study of hydrodynamics for

111: granular gases is the same as that for normal gases.

112: 

113: 

114: 

115: The details of the Chapman--Enskog method have been carried out

116: for a one granular component gas to obtain the Navier--Stokes

117: order hydrodynamic equations, together with exact integral

118: equations determining the transport coefficients occurring in

119: these equations. \cite{BDKS98} These integral equations have been

120: solved approximately using Sonine polynomial expansions and

121: compared to both Direct Monte Carlo simulation of the Boltzmann

122: equation and molecular dynamics simulation of the gas.

123: \cite{BRMC99} Good agreement is obtained even for relatively

124: strong degrees of dissipation. The results support the formal

125: theoretical analysis and the claim that hydrodynamics is not

126: limited to the quasi-elastic limit.

127: 

128: 

129: The analysis for multicomponent granular gases is much more

130: complicated than for a one component gas. Most of the previous

131: attempts \cite{JM89} were made for {\em nearly} elastic spheres

132: where the equipartition of energy can be considered as an

133: acceptable assumption. In addition, according to this level of

134: approximation, the inelasticity is only accounted for by the

135: presence of a sink term in the energy balance equation, so that

136: the expressions for the transport coefficients are the same as

137: those obtained for normal fluids. \cite{MCK83} However, the

138: failure of energy equipartition in multicomponent granular gases

139: \cite{GD99} has been also confirmed by computer simulations

140: \cite{computer} and even observed in real experiments. \cite{exp}

141: Although the possibility of nonequipartition was already pointed

142: out many years ago, \cite{JM87} it has not been until recently

143: that a systematic study of the effect of nonequipartition on

144: transport has been carefuly analyzed. In this context, Garz\'o and

145: Dufty \cite{GD02} have carried out a derivation of the

146: Navier--Stokes hydrodynamic equations for a binary mixture at

147: low-density that accounts for nonequipartition of granular energy.

148: These equations and associated transport coefficients provide a

149: somewhat more stringent test of the analysis since the parameter

150: space is much larger. There are now many more transport

151: coefficients, given as functions of the three independent

152: coefficients of restitution, size ratio, mass ratio, and

153: composition. As in the one component case, explicit expressions

154: for the transport coefficients requires also to consider Sonine

155: polynomial expansions. The accuracy of this approach has been

156: confirmed by comparison with Monte Carlo simulations of the

157: Boltzmann equation in the cases of the shear viscosity \cite{MG03}

158: and the tracer diffusion \cite{GM04} coefficients. Exceptions to

159: this agreement are extreme mass or size ratios and strong

160: dissipation, although these discrepancies between theory and

161: simulation diminish as one considers more terms in the Sonine

162: polynomial approximation. \cite{GM04}

163: 

164: 

165: 

166: 

167: Since the dependence of the shear viscosity coefficient on the

168: parameters of the mixture (masses, sizes, concentration,

169: coefficients of restitution) has been widely studied in a previous

170: work, \cite{MG03}  a primary objective here is to demonstrate the

171: variation of the six transport coefficients associated with the

172: mass and heat flux in this parameter space, using the same Sonine

173: polynomial approximation as was found applicable for the one

174: component gas. To set the context for these quantitative results

175: two qualitative and potentially confusing issues are briefly noted

176: at the outset.

177: 

178: 

179: 

180: There is some ambiguity regarding the hydrodynamic temperature in

181: a mixture since temperatures for each species can be defined in

182: addition to the global temperature. What should be the

183: hydrodynamic fields? For normal gases the answer is clear. These

184: are set by the slow variables (at large space and time scales)

185: associated with conserved quantities. In addition to species

186: number and momentum, only the total kinetic energy is conserved,

187: so only one global temperature occurs as a hydrodynamic field. The

188: energy is no longer conserved for granular gases, but it remains a

189: slow variable if the cooling rate is not too large. Thus, in this

190: case as well, only the global temperature should appear among the

191: hydrodynamic fields.

192: 

193: 

194: Nevertheless, the species temperatures play a new and interesting

195: secondary role. For a normal gas, there is a rapid velocity

196: relaxation in each fluid cell to a local equilibrium state on the

197: time scale of a few collisions. Subsequently, the equilibration

198: among cells occurs via the hydrodynamic equations. In each cell

199: the species velocity distributions are characterized by the

200: species temperatures. These are approximately the same due to

201: equipartition, and the hydrodynamic relaxation occurs for the

202: single common temperature. \cite{CC70} A similar rapid velocity

203: relaxation occurs for granular gases in each small cell, but to a

204: universal state different from local equilibrium and one for which

205: equipartition no longer occurs. Hence, the species temperatures

206: $T_{i}$ are different from each other and from the overall

207: temperature $T$ of the cell. \cite{GD99} Nevertheless, the time

208: dependence of all temperatures is the same in this and subsequent

209: states, $T_{i}(t)=\gamma _{i}T(t)$. This implies that the species

210: temperatures do not provide any new dynamical degree of freedom.

211: They still characterize the shape of the partial velocity

212: distributions and affect the quantitative averages calculated with

213: these distributions. The transport coefficients for granular

214: mixtures therefore have new quantitative effects arising from the

215: time independent temperature ratio $\gamma _{i}$ for each species.

216: \cite{GD02} This dependence is illustrated here as well.

217: 

218: 

219: In some earlier works, \cite{JM87,GDH05} additional equations for

220: each species temperature have been included among the hydrodynamic

221: set. This is possible since the overall temperature is determined

222: from these by $T(t)=\sum_{i}x_{i}T_{i}(t)$, where $ x_{i}$ is the

223: concentration of species $i$. However, this is an unnecessary

224: complication, describing additional kinetics beyond hydrodynamics

225: that is relevant only on the time scale of a few collisions. As

226: described above (and supported by molecular dynamics simulations

227: \cite{DHGD02}) the dynamics of all temperatures quickly reduces to

228: that of $T(t)$. The remaining time independent determination of

229: the $\gamma _{i}$ follows directly from the condition that the

230: cooling rates of all temperatures be the same.

231: 

232: 

233: 

234: A second confusing issue is the context of the Navier--Stokes

235: equations considered here. The derivation of the Navier--Stokes

236: order transport coefficients does not limit their application to

237: weak inelasticity. For this reason the results reported below

238: include a domain of both weak and strong inelasticity, $0.5\leq

239: \alpha \leq 1$. The Navier--Stokes hydrodynamic equations

240: themselves may or may not be limited with respect to inelasticity,

241: depending on the particular states considered. The derivation of

242: these equations by the Chapman--Enskog method assumes that

243: relative changes in the hydrodynamic fields over distances of the

244: order of a mean free path are small. For normal gases this can be

245: controlled by the initial or boundary conditions. It is more

246: complicated for granular gases. In some cases (e.g., steady states

247: such as the simple shear flow problem \cite{SGD04}) the boundary

248: conditions imply a relationship between the coefficient of

249: restitution and some hydrodynamic gradient -- the two cannot be

250: chosen independently. Consequently, there are examples for which

251: the Navier--Stokes approximation is never valid or is restricted

252: to the quasi-elastic limit. \cite{SGD04} However, the transport

253: coefficients characterizing the Navier--Stokes order hydrodynamic

254: equations are well-defined functions of $\alpha$, regardless of

255: the applicability of those equations (e.g., the coefficients of a

256: Taylor series for some function $f(\alpha,x)$ in powers of $x$ may

257: be defined for all $\alpha$ at each order, even when the series

258: cannot be truncated at that order).

259: 

260: 

261: 

262: The plan of the paper is as follows. First, in Sec.\ \ref{sec2}

263: the hydrodynamic equations and associated fluxes to Navier--Stokes

264: order are recalled, and the expressions for the transport

265: coefficients for heat and mass transport are given to leading

266: Sonine polynomial approximation. The elastic limit is discussed to

267: aid in the interpretation of these expressions. Next, in Sec.\

268: \ref{sec3} the results for these six coefficients are illustrated

269: for a common coefficient of restitution  and same size ratio as

270: functions of $\alpha $ at a composition $x_{1}=0.2$ for several

271: values of the mass ratio. With the exception of thermal

272: conductivity, the deviations from normal gas values are largest at

273: small $\alpha $ and large mass ratio. The usual Onsager relations

274: among these coefficients for normal gases is then noted and tested

275: for the granular gas in Sec.\ \ref{sec4}. Since the underlying

276: basis for these relations (time reversal symmetry) no longer holds

277: for granular systems, the expected violation is demonstrated as a

278: function of $\alpha $ for the same conditions. The stability of a

279: special homogeneous solution to the mixture hydrodynamic equations

280: is then studied, and some comments on the implications of the

281: instability found are offered. Finally, the results are summarized

282: and discussed in the last section.

283: 

284: 

285: 

286: \section{Boltzmann kinetic theory for the mass and heat fluxes}

287: \label{sec2}

288: 

289: 

290: 

291: We consider a binary mixture of \emph{inelastic}, smooth, hard

292: spheres of masses $m_{1}$ and $m_{2}$, and diameters $\sigma _{1}$

293: and $\sigma _{2}$. The inelasticity of collisions among all pairs

294: is characterized by three independent constant coefficients of

295: normal restitution $\alpha _{11}$, $ \alpha _{22}$, and $\alpha

296: _{12}=\alpha _{21}$, where $\alpha _{ij}$ is the coefficient of

297: restitution for collisions between particles of species $i$ and

298: $j$. \ It is assumed that the density of each species is

299: sufficiently low that their velocity distribution functions are

300: accurately described by the coupled set of {\em inelastic}

301: Boltzmann kinetic equations. The precise form of these equations

302: is given in Ref.\ \onlinecite{GD02} and will not be required here.

303: These equations imply the exact macroscopic balance equations for

304: the particle number density of each species, $n_{i}\left(

305: \mathbf{r},t\right) $, flow velocity $\mathbf{u}\left(

306: \mathbf{r},t\right) $, and temperature, $T\left(

307: \mathbf{r},t\right) $, \cite{GD02}

308: \begin{equation}

309: D_{t}n_{i}+n_{i}\nabla \cdot \mathbf{u}+\frac{\nabla \cdot

310: \mathbf{j}_{i}}{ m_{i}}=0\;,\hspace{0.3in}i=1,2  \label{2.01}

311: \end{equation}

312: \begin{equation}

313: D_{t}\mathbf{u}+\rho ^{-1}\nabla \mathsf{P}=0\;,  \label{2.02}

314: \end{equation}

315: \begin{equation}

316: D_{t}T-\frac{T}{n}\sum_{i}\frac{\nabla \cdot

317: \mathbf{j}_{i}}{m_{i}}+\frac{2}{ 3n}\left( \nabla \cdot

318: \mathbf{q}+\mathsf{P}:\nabla \mathbf{u}\right) =-\zeta T\;.

319: \label{2.03}

320: \end{equation}

321: In the above equations, $D_{t}=\partial _{t}+\mathbf{u}\cdot

322: \nabla $ is the material derivative, $\rho =m_{1}n_{1}+m_{2}n_{2}$

323: is the total mass density, $\mathbf{j}_{i}$ is the particle number

324: flux for species $i$, $\mathbf{q}$ is the heat flux, $\mathsf{P}$

325: is the pressure tensor, and $\zeta$ is the cooling rate. For the

326: two component mixture considered here there are six independent

327: fields, $n_{1},$ $n_{2},$ $T,$ $\mathbf{u}$. To obtain a closed

328: set of hydrodynamic equations, expressions for ${\bf j}_{i}$,

329: $\mathbf{q}$, $ \mathsf{P}$, and $\zeta $ must be given in terms

330: of these fields. Such expressions are called ``constitutive

331: equations''. It is convenient to give these constitutive equations

332: in terms of a different set of experimentally more accessible

333: fields, $x_{1},$ $p,$ $T,$ $\mathbf{u}$, where $

334: x_{1}=n_{1}/\left( n_{1}+n_{2}\right) $ is the composition of

335: species $1$, and $p=\left( n_{1}+n_{2}\right) T$ is the

336: hydrostatic pressure. This is simply a change of variables, so

337: that Eqs.\ (\ref{2.01})--(\ref{2.03}) become

338: \begin{equation}

339: D_{t}x_{1}+\frac{\rho }{n^{2}m_{1}m_{2}}\nabla \cdot

340: \mathbf{j}_{1}=0\;, \label{2.04}

341: \end{equation}

342: \begin{equation}

343: D_{t}p+p\nabla \cdot \mathbf{u}+\frac{2}{3}\left( \nabla \cdot

344: \mathbf{q}+ \mathsf{P}:\nabla \mathbf{u}\right) =-\zeta p,

345: \label{2.05}

346: \end{equation}

347: \begin{equation}

348: D_{t}\mathbf{u}+\rho ^{-1}\nabla \mathsf{P}=0\;,  \label{2.06}

349: \end{equation}

350: \begin{equation}

351: D_{t}T-\frac{T}{n}\sum_{i}\frac{\nabla \cdot

352: \mathbf{j}_{i}}{m_{i}}+\frac{2}{3n}\left( \nabla \cdot

353: \mathbf{q}+\mathsf{P}:\nabla \mathbf{u}\right) =-\zeta T\;.

354: \label{2.07}

355: \end{equation}

356: 

357: 

358: 

359: The constitutive equations up to the Navier--Stokes order have

360: been obtained from the Boltzmann equation in Ref.\

361: \onlinecite{GD02} with the results

362: \begin{equation}

363: \mathbf{j}_{1}=-\left( \frac{m_{1}m_{2}n}{\rho }\right) D\nabla

364: x_{1}-\frac{ \rho }{p}D_{p}\nabla p-\frac{\rho }{T}D^{\prime

365: }\nabla T,\quad \mathbf{j}_{2}=-\mathbf{j}_{1},  \label{2.1}

366: \end{equation}

367: \begin{equation}

368: \mathbf{q}=-T^{2}D^{\prime \prime }\nabla x_{1}-L\nabla p-\lambda

369: \nabla T, \label{2.2}

370: \end{equation}

371: \begin{equation}

372: P_{k\ell }=p\delta _{k\ell }-\eta \left( \nabla _{\ell

373: }u_{k}+\nabla _{k}u_{\ell }-\frac{2}{3}\delta _{k\ell }\nabla

374: \cdot \mathbf{u}\right) , \label{2.3}

375: \end{equation}

376: \begin{equation}

377: \label{2.2.1} \zeta =\zeta _{0}+\mathcal{O}(\nabla^2)

378: \end{equation}

379: The transport coefficients $\{D,D_{p},D^{\prime},D^{\prime \prime

380: },L,\lambda ,\eta \}$ verify a set of coupled linear integral

381: equations which can be solved approximately by using the leading

382: terms in a Sonine polynomial expansion. This solution provides

383: explicit expressions for the transport coefficients in terms of

384: the coefficients of restitution and the parameters of the mixture

385: (masses, sizes, and composition). The above expressions for mass

386: and heat fluxes can be defined in a variety of equivalent ways

387: depending on the choice of driving forces used. For systems with

388: elastic collisions, the specific set of gradients contributing to

389: each flux is restricted by fluid symmetry, Onsager's relations

390: (time reversal invariance), and the form of entropy production.

391: \cite{GM84} In this case, one usual representation leads to the

392: mass and heat fluxes proportional to $(\nabla \mu_{i})_T$ and

393: $\nabla T$, where $\mu_i$ (defined below) is the chemical

394: potential per unit mass. However, for \emph{inelastic} systems

395: only fluid symmetry holds and so there is more flexibility in

396: representing the fluxes and identifying the corresponding

397: transport coefficients. In particular, a third contribution

398: proportional to $\nabla p$ appears in both fluxes. Some care is

399: required in comparing transport coefficients in different

400: representations using different independent gradients for the

401: driving forces.

402: 

403: 

404: 

405: The cooling rate to lowest order in the gradients is $\zeta =\zeta

406: _{0}(x_{1},p,T)$. There are no contributions to first order in the

407: gradient for the low density Boltzmann equation. The general form

408: including second order gradient contributions is displayed in

409: Appendix A. These second order terms have been calculated for a

410: one component gas \cite{BDKS98} and found to be very small. Here,

411: these second order contributions to the cooling rate will be

412: neglected.

413: 

414: 

415: 

416: Substitution of the Navier--Stokes constitutive equations, (\ref

417: {2.1})-(\ref{2.3}), into the exact balance equations,

418: (\ref{2.04})-(\ref{2.07} ), gives the Navier--Stokes hydrodynamic

419: equations for a binary mixture

420: \begin{equation}

421: D_{t}x_{1}=\frac{\rho }{n^{2}m_{1}m_{2}}\nabla \cdot \left(

422: \frac{m_{1}m_{2}n }{\rho }D\nabla x_{1}+\frac{\rho }{p}D_{p}\nabla

423: p+\frac{\rho }{T}D^{\prime }\nabla T\right) \;,  \label{2.4}

424: \end{equation}

425: \begin{eqnarray}

426: \left( D_{t}+\zeta \right) p+\frac{5}{3}p\nabla \cdot \mathbf{u}

427: &=&\frac{2}{ 3}\nabla \cdot \left( T^{2}D^{\prime \prime }\nabla

428: x_{1}+L\nabla

429: p+\lambda\nabla T\right)   \nonumber \\

430: &&+\frac{2}{3}\eta \left( \nabla _{\ell }u_{k}+\nabla _{k}u_{\ell

431: }-\frac{2}{ 3}\delta _{k\ell }\nabla \cdot \mathbf{u}\right)

432: \nabla _{\ell }u_{k}, \label{2.5}

433: \end{eqnarray}

434: \begin{eqnarray}

435: \left( D_{t}+\zeta \right) T+\frac{2}{3}p\nabla \cdot \mathbf{u}

436: &=&-\frac{T }{n}\frac{m_{2}-m_{1}}{m_{1}m_{2}}\nabla \cdot \left(

437: \frac{m_{1}m_{2}n}{ \rho }D\nabla x_{1}+\frac{\rho }{p}D_{p}\nabla

438: p+\frac{\rho

439: }{T}D^{\prime}\nabla T\right)   \nonumber \\

440: &&+\frac{2}{3n}\nabla \cdot \left( T^{2}D^{\prime \prime }\nabla

441: x_{1}+L\nabla p+\lambda \nabla T\right)   \nonumber \\

442: &&+\frac{2}{3n}\eta \left( \nabla _{\ell }u_{k}+\nabla _{k}u_{\ell

443: }-\frac{2 }{3}\delta _{k\ell }\nabla \cdot \mathbf{u}\right)

444: \nabla _{\ell }u_{k}, \label{2.6}

445: \end{eqnarray}

446: \begin{equation}

447: D_{t}u_{\ell }+\rho ^{-1}\nabla _{\ell }p=\rho ^{-1}\nabla

448: _{k}\eta \left( \nabla _{\ell }u_{k}+\nabla _{k}u_{\ell

449: }-\frac{2}{3}\delta _{k\ell }\nabla \cdot \mathbf{u}\right) \;.

450: \label{2.7}

451: \end{equation}

452: For the chosen set of fields $n=p/T$ and $\rho =p\left[ \left(

453: m_{1}-m_{2}\right) x_{1}+m_{2}\right] /T$. These equations are

454: exact to second order in the spatial gradients for a low density

455: Boltzmann gas.

456: 

457: 

458: 

459: \subsection{Mass flux}

460: \label{subsec2.1}

461: 

462: 

463: The mass flux contains three transport coefficients, the diffusion

464: coefficient $D$, the pressure diffusion coefficient $D_{p}$, and

465: the thermal diffusion coefficient $D^{\prime }$. Explicit

466: expressions for these were obtained in Ref.\ \onlinecite{GD02}

467: using a first Sonine approximation. Dimensionless forms are

468: defined by

469: 

470: \begin{equation}

471: D=\frac{\rho T}{m_{1}m_{2}\nu _{0}}D^{\ast },\quad

472: D_{p}=\frac{nT}{\rho \nu _{0}}D_{p}^{\ast },\quad D^{\prime

473: }=\frac{nT}{\rho \nu _{0}}D^{\prime}{}^{\ast }.  \label{2.8}

474: \end{equation}

475: Here, $\nu _{0}=\sqrt{\pi }n\sigma _{12}^{2}v_{0}$,

476: $\sigma_{12}=(\sigma_1+\sigma_2)/2$, and  $v_{0}=\sqrt{

477: 2T(m_{1}+m_{2})/m_{1}m_{2}}$ is a thermal velocity defined in

478: terms of the temperature $T$ of the mixture. The explicit forms

479: are then

480: \begin{equation}

481: D^{\ast }=\left[ \left( \frac{\partial }{\partial

482: x_{1}}x_{1}\gamma _{1}\right) _{p,T}+\left( \frac{\partial \zeta

483: ^{\ast }}{\partial x_{1}} \right) _{p,T}\left( 1-\frac{\zeta

484: ^{\ast }}{2\nu ^{\ast }}\right) D_{p}^{\ast }\right] \left( \nu

485: ^{\ast }-\frac{1}{2}\zeta ^{\ast }\right)^{-1},  \label{2.9}

486: \end{equation}

487: \begin{equation}

488: D_{p}^{\ast }=x_{1}\left[ \gamma _{1}-\frac{\mu (1+\delta )}{1+\mu

489: \delta } \right] \left( \nu ^{\ast }-\frac{3}{2}\zeta ^{\ast

490: }+\frac{\zeta ^{\ast 2}}{ 2\nu ^{\ast }}\right) ^{-1},

491: \label{2.10}

492: \end{equation}

493: \begin{equation}

494: D^{\prime }{}^{\ast }=-\frac{\zeta ^{\ast }}{2\nu ^{\ast

495: }}D_{p}^{\ast }. \label{2.11}

496: \end{equation}

497: In these equations,

498: \begin{equation}

499: \gamma _{1}=\frac{T_{1}}{T}=\frac{\gamma }{1+x_{1}(\gamma

500: -1)},\quad \gamma _{2}=\frac{T_{2}}{T}=\frac{1}{1+x_{1}(\gamma

501: -1)},  \label{2.11a}

502: \end{equation}

503: where $\mu = m_{1}/m_{2}$ is the mass ratio, $\mu

504: _{ij}=m_{i}/(m_{i}+m_{j}),$ $\delta = x_{1}/x_{2}$, and $\gamma

505: =T_{1}/T_{2}$. The detailed forms for the temperature ratio

506: $\gamma ,$ dimensionless collision rate $\nu ^{\ast }$, and

507: dimensionless cooling rate $\zeta ^{\ast }$ are given in Appendix

508: A. Since $\mathbf{j}_{1}=-\mathbf{j}_{2}$ and $\nabla

509: x_{1}=-\nabla x_{2}$, it is expected that $D^{\ast }$ should be

510: symmetric with respect to interchange of particles $1$ and $2$

511: while $D_{p}^{\ast }$ and $D^{\prime }{}^{\ast}$ should be

512: antisymmetric. This can be easily verified by noting that $

513: x_{1}\gamma _{1}+x_{2}\gamma _{2}=1$.

514: 

515: 

516: 

517: In the case of elastic collisions $\alpha _{ij}=1$, $\zeta ^{\ast

518: }=0$, $ \gamma =1$, and Eqs.\ (\ref{2.9})--(\ref{2.11}) become

519: \begin{equation}

520: D^{\ast }=\frac{3}{8}\frac{1+\delta }{1-\mu _{12}(1-\delta

521: )},\quad D_{p}^{\ast }=x_{1}\frac{(1-\mu )}{1+\mu \delta }D^{\ast

522: }, \quad D^{\prime }{}^{\ast }=0. \label{2.11e}

523: \end{equation}

524: These coincide with known results obtained for elastic collisions

525: in the first Sonine approximation. \cite{CC70} Recently, it has

526: been shown that the estimate given by the first Sonine

527: approximation for the diffusion coefficient $D$ (in the very

528: dilute concentration limit $x_{1}\rightarrow 0$) compares quite

529: well with Monte Carlo simulations, \cite{GM04} except for the

530: cases in which the gas particles are much heavier and/or much

531: larger than impurities. For these extreme cases, the second Sonine

532: approximation to $D$ improves the accuracy of the kinetic theory

533: results. \cite{GM04}

534: 

535: 

536: 

537: 

538: \subsection{Heat flux}

539: \label{subsec2.2}

540: 

541: 

542: The heat flux requires going up to the second Sonine

543: approximation. The transport coefficients $D^{\prime \prime }$,

544: $L$, and $\lambda $ appearing in the heat flux (\ref{2.2}) are

545: given by \cite{GD02}

546: 

547: \begin{equation}

548: D^{\prime \prime }=-\frac{5}{2}\frac{n}{(m_{1}+m_{2})\nu

549: _{0}}\left[ \frac{ x_{1}\gamma _{1}^{3}}{\mu _{12}}d_{1}^{\prime

550: \prime }+\frac{x_{2}\gamma _{2}^{3}}{\mu _{21}}d_{2}^{\prime

551: \prime }-\left( \frac{\gamma _{1}}{\mu _{12}}-\frac{\gamma

552: _{2}}{\mu _{21}}\right) D^{\ast }\right] ,  \label{2.12}

553: \end{equation}

554: \begin{equation}

555: L=-\frac{5}{2}\frac{T}{(m_{1}+m_{2})\nu _{0}}\left[

556: \frac{x_{1}\gamma _{1}^{3}}{\mu _{12}}\ell _{1}+\frac{x_{2}\gamma

557: _{2}^{3}}{\mu _{21}}\ell _{2}-\left( \frac{\gamma _{1}}{\mu

558: _{12}}-\frac{\gamma _{2}}{\mu _{21}} \right) D_{p}^{\ast }\right]

559: ,  \label{2.13}

560: \end{equation}

561: \begin{equation}

562: \lambda =-\frac{5}{2}\frac{nT}{(m_{1}+m_{2})\nu _{0}}\left[ \frac{

563: x_{1}\gamma _{1}^{3}}{\mu _{12}}\lambda _{1}+\frac{x_{2}\gamma

564: _{2}^{3}}{\mu_{21}}\lambda _{2}-\left( \frac{\gamma _{1}}{\mu

565: _{12}}-\frac{\gamma _{2}}{ \mu _{21}}\right) D^{\prime }{}^{\ast

566: }\right] ,  \label{2.14}

567: \end{equation}

568: where the expressions for the (dimensionless) Sonine coefficients

569: $\{d_{i}^{\prime \prime },\ell _{i},\lambda _{i}\}$ are displayed

570: in Appendix \ref{appA}. In Eqs.\ (\ref{2.12})--(\ref{2.14}) it is

571: understood that the coefficients $D^{\ast }$, $D_{p}^{\ast }$, and

572:  $D^{\prime \ast }$ are given by

573:  Eqs.\ (\ref{2.9})--(\ref{2.11}), respectively (first

574: Sonine approximation). As expected, our results show that

575: $D^{\prime \prime }$ is antisymmetric with respect to the change

576: $1\leftrightarrow 2$ while $L$ and $ \lambda $ are symmetric.

577: Consequently, in the case of mechanically equivalent particles

578: ($m_{1}=m_{2}\equiv m$, $\sigma _{1}=\sigma _{2}\equiv \sigma$,

579: $\alpha _{ij}\equiv \alpha $), the coefficient $D^{\prime \prime}$

580: vanishes.

581: 

582: 

583: 

584: An equivalent representation is given in terms of the heat flow

585: $\mathbf{J}_{q}$ defined as

586: \begin{equation}

587: \mathbf{J}_{q}\equiv

588: \mathbf{q}-\frac{5}{2}T\sum_{i}\frac{\mathbf{j}_{i}}{

589: m_{i}}=\mathbf{q}-\frac{5}{2}T\frac{m_{2}-m_{1}}{m_{1}m_{2}}\mathbf{j}_{1},

590: \label{2.15}

591: \end{equation}

592: where in the second equality use has been made of the requirement

593: $\mathbf{j} _{1}=-\mathbf{j}_{2}$. The difference between

594: $\mathbf{q}$ and $\mathbf{J} _{q}$ is a heat flow due to

595: diffusion. In addition, for elastic collisions, $ \mathbf{J}_{q}$

596: is the flux conjugate to the temperature gradient in the form of

597: the entropy production where the contribution coming from the mass

598: flux couples only to the gradient of the chemical potentials. The

599: thermal conductivity in a mixture is generally measured in the

600: absence of diffusion, i.e., when $\mathbf{j}_{1}=\mathbf{0}$. To

601: identify this coefficient, we have to express $\mathbf{J}_{q}$ in

602: terms of $\mathbf{j}_{1}$, $\nabla T$, and $\nabla p$. The

603: corresponding coefficient of $\nabla T$ defines the thermal

604: conductivity. \cite{GM84} According to Eq.\ (\ref{2.1}), the

605: gradient of mole fraction $\nabla x_{1}$ is

606: \begin{equation}

607: \nabla x_{1}=-\frac{\rho }{m_{1}m_{2}nD}\mathbf{j}_{1}-\frac{\rho

608: ^{2}}{ m_{1}m_{2}p}\frac{D^{\prime }}{D}\nabla T-\frac{\rho

609: ^{2}}{m_{1}m_{2}np} \frac{D_{p}}{D}\nabla p.  \label{2.16}

610: \end{equation}

611: The expression of $\mathbf{J}_{q}$ is obtained by substituting

612: Eq.\ (\ref {2.2}) into Eq.\ (\ref{2.15}) and eliminating $\nabla

613: x_{1}$ by using the identity (\ref{2.16}). Thus, the heat flow is

614: given by

615: \begin{equation}

616: \mathbf{J}_{q}=-\kappa \nabla T+\frac{\rho

617: p}{m_{1}m_{2}n^{2}}\kappa_{\text{T}}\mathbf{j}_{1}-L_{p}\nabla p,

618: \label{2.17}

619: \end{equation}

620: where

621: \begin{equation}

622: \kappa =\lambda -\frac{\rho ^{2}T^{2}}{m_{1}m_{2}p}\frac{D^{\prime \prime

623: }D^{\prime }}{D},  \label{2.18}

624: \end{equation}

625: \begin{equation}

626: \kappa _{\text{T}}=T\frac{D^{\prime \prime

627: }}{D}-\frac{5}{2}\frac{n}{\rho } (m_{2}-m_{1}),  \label{2.19}

628: \end{equation}

629: \begin{equation}

630: L_{p}=L-\frac{\rho ^{2}T}{n^{2}m_{1}m_{2}}\frac{D^{\prime \prime }D_{p}}{D}.

631: \label{2.20}

632: \end{equation}

633: As in the elastic case, the coefficient $\kappa $ is the thermal

634: conductivity while $\kappa _{\text{T}}$ is called the thermal-diffusion

635: factor or Dufour coefficient. There is a \emph{new} contribution

636: proportional to $\nabla p$ not present in the elastic case that defines the

637: transport coefficient $L_{p}$.

638: 

639: 

640: For elastic collisions, $L_{p}=0$ \cite{comment} and the

641: expressions derived here for $\kappa $ and $\kappa _{\text{T}}$

642: coincide with those obtained for a gas-mixture of elastic hard

643: spheres. \cite{CC70} Furthermore, in the case of mechanically

644: equivalent particles, the Dufour coefficient $\kappa _{\text{T}}$

645: vanishes as expected and the heat flow (\ref {2.17}) can be

646: written as

647: \begin{equation}

648: \mathbf{J}_{q}=-\overline{\kappa }\nabla T-\overline{\mu }\nabla n,

649: \label{2.21}

650: \end{equation}

651: where

652: \begin{equation}

653: \overline{\kappa }=\kappa +nL_{p}=\frac{25}{32}\left( \frac{mT}{\pi }\right)

654: ^{1/2}\sigma ^{-2}\frac{1+c}{\nu _{\kappa }-2\zeta ^{\ast }},  \label{2.22}

655: \end{equation}

656: \begin{equation}

657: \overline{\mu }=TL_{p}=\frac{75}{32}\frac{T}{n}\left(

658: \frac{mT}{\pi }\right) ^{1/2}\sigma ^{-2}\zeta ^{\ast }\left(

659: \frac{2}{3}\frac{1+c}{\nu _{\kappa }-2\zeta ^{\ast

660: }}+\frac{1}{3}\frac{c}{\zeta ^{\ast }}\right) \left(2\nu _{\kappa

661: }-3\zeta ^{\ast }\right) ^{-1},  \label{2.23}

662: \end{equation}

663: \begin{equation}

664: \nu _{\kappa }=\frac{1}{3}(1+\alpha )\left[

665: 1+\frac{33}{16}(1-\alpha )+\frac{ 19-3\alpha }{1024}c\right] ,

666: \label{2.24}

667: \end{equation}

668: \begin{equation}

669: c=\frac{32(1-\alpha )(1-2\alpha ^{2})}{81-17\alpha +30\alpha

670: ^{2}(1-\alpha )} ,\quad \zeta ^{\ast }=\frac{5}{12}(1-\alpha

671: ^{2})\left( 1+\frac{3}{32} c\right) .  \label{2.25}

672: \end{equation}

673: Note that in writing Eq.\ (\ref{2.21}) use has been made of the

674: relation $ \nabla p=n\nabla T+T\nabla n$. Equations

675: (\ref{2.21})--(\ref{2.25}) are the same as those obtained for the

676: one component granular gas. \cite{BDKS98} This confirms the

677: relevant known limiting cases for the granular mixture results

678: described here.

679: 

680: 

681: \section{Transport coefficients}

682: \label{sec3}

683: 

684: 

685: 

686: The transport coefficients depend on many parameters: $\left\{

687: x_{1},T,m_{1}/m_{2},\sigma _{1}/\sigma _{2}, \alpha _{11}, \alpha

688: _{22}, \alpha _{12}\right\} $. This complexity exists in the

689: elastic limit as well, so the primary new feature is the

690: dependence on the coefficients of restitution  $ \alpha _{ij}$

691: being different from unity. To illustrate the differences between

692: granular and normal gases the transport coefficients are

693: normalized to their values in the elastic limit. Then, the

694: dependence on the overall temperature scales out. Also, only the

695: simplest case of a common coefficient of restitution ($\alpha

696: _{11}=\alpha _{22}=\alpha _{12}\equiv \alpha $) and common size

697: $\omega \equiv \sigma _{1}/\sigma _{2}=1$ is considered. This

698: reduces the parameter set to three quantities:

699: $\{m_{1}/m_{2},x_{1},\alpha \}$.

700: 

701: 

702: In Figs.\ \ref{fig1}--\ref{fig6}, we plot the above transport

703: coefficients as functions of the coefficient of restitution

704: $\alpha $ for $x_{1}=0.2$, and several values of the mass ratio

705: $\mu$. It is understood that all coefficients have been reduced

706: with respect to their elastic values, except in the cases of

707: $D^{\prime }$ and $L_{p}$ since both coefficients vanish for

708: elastic collisions. In these latter two cases, we have considered

709: the reduced coefficients $D^{\prime }{}^{\ast }$ defined by Eq.\

710: (\ref{2.9}) and $ L_{p}^{\ast }=-\left[ (5/2)T\nu

711: _{0}/(m_{1}+m_{2})\right]^{-1}L_{p}$. Figure \ref{fig1} shows the

712: mutual diffusion coefficient as a function of $ \alpha $ for three

713: mass ratios $\mu=0.5,1,$ and $4$. There is a monotonic increase of

714: the coefficient with decreasing $\alpha $ in all cases. Moreover,

715: that effect increases as the mass of the dilute species increases.

716: This is consistent with an observed singular behavior in the

717: extreme case of tracer diffusion for a massive particle.

718: \cite{SD01} The velocity distributions in a granular gas are no

719: longer Maxwellian, \cite{GD99} and the difference is measured by

720: the coefficients $c_{i}$ that appear in the expressions for the

721: transport coefficients. \cite{GD02} The dashed curves in Fig.\

722: \ref{fig1} correspond to $c_{1}=c_{2}\rightarrow 0$, the Maxwell

723: limit. In this case it is seen that the distortion of the

724: Maxwellian is not very important for this transport coefficient

725: (see however, discussion of Fig.\ \ref{fig4} below). Figure

726: \ref{fig2} shows that the pressure diffusion coefficient has a

727: very similar behavior. The thermal diffusion coefficient vanishes

728: in the elastic limit and remains small and slightly negative when

729: the dilute species has small mass ratio, as illustrated in Fig.\

730: \ref{fig3}. However as the mass ratio becomes large it becomes

731: large and positive for strong dissipation. The effect of different

732: species temperatures is also shown on this graph. The dashed

733: curves correspond to setting $\gamma =T_{1}/T_{2}\rightarrow 1$.

734: This is seen to yield large errors, particularly in the case of

735: large mass ratio (temperature differences are greater for

736: mechanically different particles), indicating the real

737: quantitative effect of two different species temperatures in

738: granular gases.

739: \begin{figure}[tbp]

740: \begin{center}

741: \resizebox{7.5cm}{!}{\includegraphics{fig1new.eps}}

742: \end{center}

743: \caption{Plot of the reduced mutual diffusion coefficient

744: $D(\protect\alpha )/D(1)$ as a function of the coefficient of

745: restitution $\protect\alpha$ for $x_1=0.2$, $\protect\omega=1$ and

746: $\protect\mu=0.5$ (a), $\protect\mu=1$ (b), and $\protect\mu=4$

747: (c). The dashed lines correspond to the approximation

748: $c_1=c_2=0$.} \label{fig1}

749: \end{figure}

750: \begin{figure}[tbp]

751: \begin{center}

752: \resizebox{7.5cm}{!}{\includegraphics{fig2new.eps}}

753: \end{center}

754: \caption{Plot of the reduced pressure diffusion coefficient

755: $D_p(\protect \alpha)/D_{p}(1)$ as a function of the coefficient

756: of restitution $\protect \alpha$ for $x_1=0.2$, $\protect\omega=1$

757: and $\protect\mu=0.5$ (a), $ \protect\mu=2$ (b), and

758: $\protect\mu=4$ (c).} \label{fig2}

759: \end{figure}

760: \begin{figure}[tbp]

761: \begin{center}

762: \resizebox{7.5cm}{!}{\includegraphics{fig3.eps}}

763: \end{center}

764: \caption{Plot of the reduced thermal diffusion coefficient

765: $D^{^{\prime }\ast }(\protect\alpha )$ as a function of the

766: coefficient of restitution $ \protect\alpha $ for $x_{1}=0.2$,

767: $\protect\omega =1$ and $\protect\mu =0.5$ (a), $\protect\mu =2$

768: (b), and $\protect\mu =4$ (c). The dashed lines correspond to

769: $\protect\gamma =1$.} \label{fig3}

770: \end{figure}

771: 

772: 

773: The thermal conductivity is shown in Fig.\  \ref{fig4}, with the

774: same monotonic increase with increasing dissipation. The dashed

775: lines ($c_i=0$) indicate a significant effect of the distortion of

776: the reference distribution function from its Maxwellian form.

777: Presumably, this is due to the fact that the thermal conductivity

778: $\kappa$ depends on a higher velocity moment than the mutual

779: diffusion coefficient $D$ and is more sensitive to the larger

780: distortions at higher velocities. We also observe that there is

781: little mass dependence when the dilute species is lighter.

782: However, when the dilute species is more massive there is a

783: significant decrease in the thermal conductivity, opposite to the

784: case of diffusion. In contrast, the Dufour coefficient does have a

785: dependence on the mass ratio more like diffusion (Fig.\

786: \ref{fig5}). Finally, the coefficient $L_{p}^{\ast }$ is

787: illustrated in Figure \ref{fig6}. Again there is weak dependence

788: on the mass ratio until it becomes larger for the dilute species.

789: 

790: 

791: In summary, the mass and heat flux transport coefficients for a

792: granular mixture differ significantly from those for a normal gas

793: mixture even at moderate dissipation. In most cases (thermal

794: conductivity is an exception) the differences increase with

795: decreasing $\alpha $, depend weakly on the mass ratio when the

796: dilute species ($x_1/x_2<1$) is lighter than the excess species

797: ($m_{1}/m_{2}\leq 1$) but increase significantly in the opposite

798: case ($m_{1}/m_{2}>1$).

799: 

800: \begin{figure}[tbp]

801: \begin{center}

802: \resizebox{7.5cm}{!}{\includegraphics{fig4new.eps}}

803: \end{center}

804: \caption{Plot of the reduced thermal conductivity coefficient

805: $\protect\kappa (\protect\alpha)/\protect\kappa(1)$ as a function

806: of the coefficient of restitution $\protect\alpha$ for $x_1=0.2$,

807: $\protect\omega=1$ and $\protect \mu=0.5$ (a), $\protect\mu=1$

808: (b), and $\protect\mu=4$ (c). The dashed lines correspond to the

809: approximation $c_1=c_2=0$.} \label{fig4}

810: \end{figure}

811: \begin{figure}[tbp]

812: \begin{center}

813: \resizebox{7.5cm}{!}{\includegraphics{fig5new.eps}}

814: \end{center}

815: \caption{Plot of the reduced Dufour coefficient

816: $\protect\kappa_T(\protect \alpha)/\protect\kappa_{T}(1)$ as a

817: function of the coefficient of restitution $\protect\alpha$ for

818: $x_1=0.2$, $\protect\omega=1$ and $\protect \mu=0.5$ (a),

819: $\protect\mu=2$ (b), and $\protect\mu=4$ (c).} \label{fig5}

820: \end{figure}

821: \begin{figure}[tbp]

822: \begin{center}

823: \resizebox{7.5cm}{!}{\includegraphics{fig6.eps}}

824: \end{center}

825: \caption{Plot of the reduced coefficient $L_p^*(\protect\alpha)$

826: as a function of the coefficient of restitution $\protect\alpha$

827: for $x_1=0.2$, $ \protect\omega=1$ and $\protect\mu=0.5$ (a),

828: $\protect\mu=1$ (b), and $ \protect\mu=4$ (c).} \label{fig6}

829: \end{figure}

830: 

831: 

832: \section{Onsager's reciprocal relations}

833: \label{sec4}

834: 

835: 

836: 

837: In the usual language of the linear irreversible thermodynamics

838: for ordinary fluids, \cite{GM84} the constitutive equations for

839: the mass flux (\ref{2.1}) and heat flow (\ref{2.17}) is written

840: \begin{equation}

841: \mathbf{j}_{i}=-\sum_{i}L_{ij}\left( \frac{\nabla \mu

842: _{j}}{T}\right) _{T}-L_{iq}\frac{\nabla T}{T^{2}}-C_{p}\nabla p,

843: \label{4.1}

844: \end{equation}

845: \begin{equation}

846: \mathbf{J}_{q}=-L_{qq}\nabla T-\sum_{i}L_{qi}\left( \frac{\nabla

847: \mu _{i}}{T} \right) _{T}-C_{p}^{\prime }\nabla p,  \label{4.2}

848: \end{equation}

849: where

850: \begin{equation}

851: \left( \frac{\nabla \mu _{i}}{T}\right) _{T}=\frac{1}{m_{i}}\nabla

852: \ln (x_{i}p),  \label{4.3}

853: \end{equation}

854: $\mu _{i}$ being the chemical potential per unit mass. Here, the

855: coefficients $L_{ij}$ are the so-called Onsager phenomenological

856: coefficients. For \emph{normal} fluids, Onsager showed \cite{GM84}

857: that time reversal invariance of the underlying microscopic

858: equations of motion implies important restrictions on the above

859: set of transport coefficients

860: \begin{equation}

861: L_{ij}=L_{ji},\quad L_{iq}=L_{qi},\quad C_{p}=C_{p}^{\prime }=0.

862: \label{4.4}

863: \end{equation}

864: The first two symmetries are called reciprocal relations as they

865: relate transport coefficients for different processes. The last

866: two are statements that the pressure gradient does not appear in

867: any of the fluxes even though it is admitted by symmetry. Even for

868: a one component fluid, Onsager's theorem is significant as it

869: leads to Fourier's law for the heat flow rather than (\ref{2.21}),

870: i.e. $\overline{\mu }=0$. Since there is no time reversal symmetry

871: for granular fluids, Eqs. (\ref{4.4}) cannot be expected to apply.

872: However, since explicit expressions for all transport coefficients

873: are at hand, the quantitative extent of the violation can be

874: explored.

875: 

876: 

877: To make connection with the previous sections it is first

878: necessary to transform Eqs.\ (\ref{4.1})--(\ref{4.3}) to the

879: variables $x_{1},$ $p,$ $T.$ Since $\nabla x_{1}=-\nabla x_{2}$,

880: Eq.\ (\ref{4.3}) implies

881: \begin{equation}

882: \frac{(\nabla \mu _{1})_{T}-(\nabla \mu _{2})_{T}}{T}=\frac{n\rho

883: }{\rho _{1}\rho _{2}}\left[ \nabla x_{1}+\frac{n_{1}n_{2}}{n\rho }

884: (m_{2}-m_{1})\nabla \ln p\right] .  \label{4.5}

885: \end{equation}

886: The coefficients $L_{ij}$ then can be easily obtained in terms of

887: those of the previous sections. The result is

888: \begin{equation}

889: L_{11}=-L_{12}=-L_{21}=\frac{m_{1}m_{2}\rho _{1}\rho _{2}}{\rho

890: ^{2}}D,\quad L_{1q}=\rho TD^{\prime },  \label{4.6}

891: \end{equation}

892: \begin{equation}

893: L_{q1}=-L_{q2}=\frac{T^{2}\rho _{1}\rho _{2}}{n\rho }D^{\prime

894: \prime }- \frac{5}{2}\frac{T\rho _{1}\rho _{2}}{\rho

895: ^{2}}(m_{2}-m_{1})D,\quad L_{qq}=\lambda -\frac{5}{2}\rho

896: \frac{m_{2}-m_{1}}{m_{1}m_{2}}D^{\prime }, \label{4.7}

897: \end{equation}

898: \begin{equation}

899: C_{p}\equiv \frac{\rho}{p} D_{p}-\frac{\rho _{1}\rho _{2}}{p\rho

900: ^{2}}(m_{2}-m_{1})D, \label{4.8}

901: \end{equation}

902: \begin{equation}

903: C_{p}^{\prime }\equiv

904: L-\frac{5}{2}\frac{T}{p}\frac{m_{2}-m_{1}}{m_{1}m_{2}}C_{p}-

905: \frac{n_{1}n_{2}}{np\rho }T^{2}(m_2-m_1)D^{\prime \prime}.

906: \label{4.9}

907: \end{equation}

908: The Onsager's relation $L_{12}=L_{21}$ is already evident since

909: the diffusion coefficient $D$ is symmetric under the change

910: $1\leftrightarrow 2$ , as discussed following Eq.\ (\ref{2.11a})

911: 

912: 

913: 

914: Imposing Onsager's relation $L_{1q}=L_{q1}$ yields

915: \begin{equation}

916: D^{\prime \prime }=\frac{5}{2}\frac{n}{T\rho

917: }(m_{2}-m_{1})D+\frac{n\rho ^{2} }{T\rho _{1}\rho _{2}}D^{\prime

918: },  \label{4.10}

919: \end{equation}

920: while the condition $C_{p}=C_{p}^{\prime }=0$ leads to the

921: following additional requirements

922: \begin{equation}

923: D_{p}=\frac{\rho _{1}\rho _{2}}{\rho ^{3}}(m_{2}-m_{1})D,

924: \label{4.11}

925: \end{equation}

926: \begin{equation}

927: \frac{5}{2}\frac{T}{p}(m_{1}-m_{2})\left[ \frac{n_{1}n_{2}}{\rho

928: ^{2}}(m_{2}-m_{1})D- \frac{\rho }{m_{1}m_{2}}D_{p}\right]

929: =pL-\frac{n_{1}n_{2}}{n\rho } T^{2}(m_{2}-m_{1})D^{\prime \prime

930: }.  \label{4.12}

931: \end{equation}

932: Since the relations (\ref{4.10})--(\ref{4.12}) involve transport

933: coefficients that have been determined in the first Sonine

934: approximation, we restrict here our discussion to this level of

935: approximation. In this case, $ d_{i}^{\prime \prime }=\ell

936: _{i}=\lambda _{i}=0$ so that Onsager's theorem, Eqs.\

937: (\ref{4.10})--(\ref{4.12}), gives the conditions

938: \begin{equation}

939: P(\alpha _{ij})\equiv \left[ \gamma _{1}-1+\mu (1-\gamma

940: _{2})\right] \frac{ D^{\ast }}{\mu }+\frac{1}{5}\frac{(1+\delta

941: )(1+\mu \delta )}{\mu \delta } \frac{\zeta ^{\ast }}{\nu ^{\ast

942: }}D_{p}^{\ast }=0,  \label{4.13}

943: \end{equation}

944: \begin{equation}

945: Q(\alpha _{ij})\equiv D_{p}^{\ast }-x_{1}\frac{(1-\mu )}{(1+\mu

946: \delta )} D^{\ast }=0,  \label{4.14}

947: \end{equation}

948: \begin{equation}

949: R(\alpha _{ij})\equiv \frac{1+\mu }{\mu }\left( \gamma _{1}-\mu

950: \gamma _{2}\right) Q(\alpha _{ij})=0.  \label{4.15}

951: \end{equation}

952: 

953: 

954: In the elastic limit, the reduced coefficients $D_{p}^{\ast }$ and

955: $D^{\ast } $ are given by Eq.\ (\ref{2.11}) and these conditions

956: are verified. Also, for mechanically equivalent particles with

957: arbitrary $\alpha $, $\gamma _{i}=1$ and $D_{p}^{\ast }=0$ so that

958: $P(\alpha )=Q(\alpha )=R(\alpha )=0$. Nevertheless, beyond these

959: limit cases, Onsager's relations do not apply (as expected). At

960: this macroscopic level the origin of this failure is due to the

961: cooling of the reference state as well as the occurrence of

962: different kinetic temperatures for both species. Figures

963: \ref{fig7}, \ref{fig8}, and \ref{fig9} show the dependence of the

964: quantities $P$, $Q$, and $R$, respectively, on the (common)

965: coefficient of restitution $\alpha_{ij}\equiv \alpha$ for mass

966: ratios $\mu =0.5,2,$ and $4$.

967: \begin{figure}[tbp]

968: \begin{center}

969: \resizebox{7.5cm}{!}{\includegraphics{fig7.eps}}

970: \end{center}

971: \caption{Plot of $P(\protect\alpha)$ as a function of

972: $\protect\alpha$ for $ x_1=0.2$, $\protect\omega=1$ and

973: $\protect\mu=0.5$ (a), $\protect\mu=2$ (b), and $\protect\mu=4$

974: (c).} \label{fig7}

975: \end{figure}

976: \begin{figure}[tbp]

977: \begin{center}

978: \resizebox{7.5cm}{!}{\includegraphics{fig8.eps}}

979: \end{center}

980: \caption{Plot of $Q(\protect\alpha)$ as a function of

981: $\protect\alpha$ for $ x_1=0.2$, $\protect\omega=1$ and

982: $\protect\mu=0.5$ (a), $\protect\mu=2$ (b), and $\protect\mu=4$

983: (c).} \label{fig8}

984: \end{figure}

985: \begin{figure}[tbp]

986: \begin{center}

987: \resizebox{7.5cm}{!}{\includegraphics{fig9.eps}}

988: \end{center}

989: \caption{Plot of $R(\protect\alpha)$ as a function of

990: $\protect\alpha$ for $ x_1=0.2$, $\protect\omega=1$ and

991: $\protect\mu=0.5$ (a), $\protect\mu=2$ (b), and $\protect\mu=4$

992: (c).} \label{fig9}

993: \end{figure}

994: 

995: 

996: 

997: \section{Linearized hydrodynamic equations and stability}

998: \label{sec5}

999: 

1000: 

1001: 

1002: In contrast to normal fluids, the Navier--Stokes hydrodynamic

1003: equations\ (\ref {2.4})--(\ref{2.7}) have non-trivial solutions

1004: even for spatially homogeneous states,

1005: \begin{equation}

1006: \label{5.0}

1007: \partial _{t}x_{1H}=0=\partial _{t}u_{H\ell },

1008: \end{equation}

1009: \begin{equation}

1010: \left[ \partial _{t}+\zeta \left( x_{1H},T_{H},p_{H}\right)

1011: \right] T_{H}=0, \hspace{0.3in}\left[ \partial _{t}+\zeta \left(

1012: x_{1H},T_{H},p_{H}\right) \right] p_{H}=0.  \label{5.1}

1013: \end{equation}

1014: where the subscript $H$ denotes the homogeneous state. Since the

1015: dependence of the cooling rate $\zeta \left(

1016: x_{1H},T_{H},p_{H}\right)$ on $x_{1H},T_{H},p_{H}$ is known (see

1017: Appendix A), these first order nonlinear equations can be solved

1018: for the time dependence of the homogeneous state. The result is

1019: the familiar Haff's law cooling law for $T(t)$ at constant

1020: density. \cite{BP04}. As discussed above, each species temperature

1021: also has the same time dependence but each with a different value

1022: \cite{GD99} In this section, the hydrodynamics for small initial

1023: spatial perturbations of this homogeneous cooling state (HCS) is

1024: discussed. For normal fluids such perturbations decay in time

1025: according to the hydrodynamic modes of diffusion (shear, thermal,

1026: mass) and damped sound propagation.  The analysis is for fixed

1027: coefficients of restitution different from unity in the long

1028: wavelength limit. It will be seen here that the corresponding

1029: modes for a granular fluid are then quite different from those for

1030: a normal fluid. An alternative analysis with fixed long wavelength

1031: and coefficient of restitution  approaching unity leads to usual

1032: normal fluid modes. Thus, the nature of hydrodynamic modes is non

1033: uniform with respect to the inelasticity and the wavelength of the

1034: perturbation.

1035: 

1036: 

1037: Let $\delta y_{\alpha }(\mathbf{r},t)=y_{\alpha

1038: }(\mathbf{r},t)-y_{H\alpha }(t)$ denote the deviation of

1039: $\{x_{1},\mathbf{u},T,p\}$ from their values in the HCS. If the

1040: initial spatial perturbation is sufficiently small, then for some

1041: initial time interval these deviations will remain small and the

1042: hydrodynamic equations (\ref{2.4})--(\ref{2.7}) can be linearized

1043: with respect to $\delta y_{\alpha }(\mathbf{r},t)$. This leads to

1044: partial differential equations with coefficients that are

1045: independent of space but which depend on time since the HCS is

1046: cooling. As in the one component case, \cite{BDKS98,G05} this time

1047: dependence can be eliminated through a change in the time and

1048: space variables, and a scaling of the hydrodynamic fields.  We

1049: introduce the following dimensionless space and time variables:

1050: \begin{equation}

1051: \tau =\int_{0}^{t}dt^{\prime }\nu_{H}(t^{\prime }),\quad

1052: \mathbf{r}^{\prime }= \mathbf{r/}\ell _{H},  \label{5.2}

1053: \end{equation}

1054: where $v_{0H}=\sqrt{2T_{H}(m_{1}+m_{2})/m_{1}m_{2}}$ is the

1055: thermal velocity introduced above, $\ell_{H}=1/\sqrt{\pi }

1056: n_{H}\sigma _{12}^{2}$ is an effective mean free path, and

1057: $\nu_{H}(t)=v_{0H}(t)/\ell _{H}$ is the effective collision

1058: frequency. The dimensionless time scale is therefore an average

1059: number of collisions up to the time $t$. A set of Fourier

1060: transformed dimensionless variables are then defined by

1061: \begin{equation}

1062: \delta y_{\mathbf{k}\alpha }(\tau )=\int

1063: d\mathbf{r}^{\prime}\;e^{-i\mathbf{k} \cdot

1064: \mathbf{r}^{\prime}}\delta y_{\alpha }(\mathbf{r}^{\prime },

1065: \tau), \label{5.3}

1066: \end{equation}

1067: \begin{equation}

1068: \rho_{\mathbf{k}}(\tau )=\frac{\delta x_{1\mathbf{k}}(\tau

1069: )}{x_{1H}},\quad \mathbf{w}_{\mathbf{k}}(\tau )=\frac{\delta

1070: \mathbf{u}_{\mathbf{k}}(\tau )}{ v_{0H}(\tau )},  \label{5.4}

1071: \end{equation}

1072: \begin{equation}

1073: \theta _{\mathbf{k}}(\tau )=\frac{\delta T_{\mathbf{k}}(\tau

1074: )}{T_{H}(\tau )} ,\quad \Pi _{\mathbf{k}}(\tau )=\frac{\delta

1075: p_{\mathbf{k}}(\tau )}{ p_{H}(\tau )}.  \label{5.5}

1076: \end{equation}

1077: In terms of these variables the linearized hydrodynamic equations

1078: for the set $\{\rho _{\mathbf{k}},\mathbf{w}_{\mathbf{k}},\theta

1079: _{\mathbf{k}},\Pi _{ \mathbf{k}}\}$ separate into a degenerate

1080: pair of equations for the two transverse velocity components

1081: $w_{\mathbf{k}\perp }$ (orthogonal to $ \mathbf{k}$)

1082: \begin{equation}

1083: \left( \frac{\partial }{\partial \tau }-\frac{\zeta ^{\ast

1084: }}{2}+\eta ^{\ast }k^{2}\right) w_{\mathbf{k}\perp }=0,

1085: \label{5.6}

1086: \end{equation}

1087: and a coupled set of equations for $\rho _{\mathbf{k}},\theta

1088: _{\mathbf{k} },\Pi _{\mathbf{k}}$, and the longitudinal velocity

1089: component $w_{\mathbf{k} ||}$ (parallel to $\mathbf{k}$)

1090: \begin{equation}

1091: \frac{\partial \delta z_{\mathbf{k}\alpha }(\tau )}{\partial \tau

1092: }=\left( M_{\alpha \beta }^{(0)}+ikM_{\alpha \beta

1093: }^{(1)}+k^{2}M_{\alpha \beta }^{(2)}\right) \delta

1094: z_{\mathbf{k}\beta }(\tau ),  \label{5.7}

1095: \end{equation}

1096: where now $\delta z_{\mathbf{k}\alpha }(\tau )$ denotes the four

1097: variables $ \left( \rho _{\mathbf{k}},\theta _{\mathbf{k}},\Pi

1098: _{\mathbf{k}},w_{\mathbf{k }||}\right) $. The matrices in this

1099: equation are

1100: \begin{equation}

1101: M^{(0)}=\left(

1102: \begin{array}{cccc}

1103: 0 & 0 & 0 & 0 \\

1104: -x_{1}\left( \frac{\partial \zeta ^{\ast }}{\partial x_{1}}\right)

1105: _{T,p} &

1106: \frac{\zeta ^{\ast }}{2} & -\zeta ^{\ast } & 0 \\

1107: -x_{1}\left( \frac{\partial \zeta ^{\ast }}{\partial x_{1}}\right)

1108: _{T,p} &

1109: \frac{\zeta ^{\ast }}{2} & -\zeta ^{\ast } & 0 \\

1110: 0 & 0 & 0 & \frac{\zeta ^{\ast }}{2}

1111: \end{array}

1112: \right) ,  \label{5.8}

1113: \end{equation}

1114: \begin{equation}

1115: M^{(1)}=\left(

1116: \begin{array}{cccc}

1117: 0 & 0 & 0 & 0 \\

1118: 0 & 0 & 0 & -\frac{2}{3} \\

1119: 0 & 0 & 0 & -\frac{5}{3} \\

1120: 0 & 0 & -\frac{1}{2}\frac{\mu _{12}}{x_{1}\mu +x_{2}} & 0

1121: \end{array}

1122: \right) ,  \label{5.9}

1123: \end{equation}

1124: \begin{equation}

1125: M^{(2)}=\left(

1126: \begin{array}{cccc}

1127: -\frac{1}{2}\frac{\mu x_{1}+x_{2}}{1+\mu }D^{\ast } &

1128: -\frac{1}{2x_{1}}\frac{ \mu x_{1}+x_{2}}{1+\mu }D^{\prime

1129: }{}^{\ast } & -\frac{1}{2x_{1}}\frac{\mu

1130: x_{1}+x_{2}}{1+\mu }D_{p}^{\ast } & 0 \\

1131: -x_{1}\left( \frac{2}{3}D^{\prime \prime }{}^{\ast }-\frac{1-\mu

1132: }{2(1+\mu )} D^{\ast }\right)  & \frac{1-\mu }{2(1+\mu )}D^{\prime

1133: }{}^{\ast }-\frac{2}{3} \lambda ^{\ast } & -\frac{2}{3}L^{\ast

1134: }+\frac{1-\mu }{2(1+\mu )}D_{p}^{\ast

1135: } & 0 \\

1136: -\frac{2}{3}x_{1}D^{\prime \prime }{}^{\ast} & -\frac{2}{3}

1137: \lambda ^{\ast } & -\frac{2}{3}L^{\ast } & 0 \\

1138: 0 & 0 & 0 & -\frac{4}{3}\eta ^{\ast }

1139: \end{array}

1140: \right) .  \label{5..10}

1141: \end{equation}

1142: In these equations, $x_{i}=n_{iH}/n_{H}$, $\zeta ^{\ast }=\zeta

1143: _{H}/\nu _{H} $ and the reduced transport coefficients $D^{\ast

1144: }$, $ D_{p}^{\ast }$, and $D^{\prime }{}^{\ast }$ are given by

1145: Eqs.\ (\ref{2.9})--(\ref{2.11} ), respectively. Moreover, the

1146: reduced Navier--Stokes transport coefficients are

1147: \begin{equation}

1148: \eta ^{\ast }=\frac{\nu _{H}\eta }{\rho _{H}v_{0H}^{2}},

1149: \label{5.11}

1150: \end{equation}

1151: \begin{equation}

1152: D^{\prime \prime }{}^{\ast }=\frac{\nu _{H}T_{H}D^{\prime \prime

1153: }}{ n_{H}v_{0H}^{2}},\quad L^{\ast }=\frac{\nu

1154: _{H}L}{v_{0H}^{2}},\quad \lambda ^{\ast }=\frac{\nu _{H}\lambda

1155: }{n_{H}v_{0H}^{2}},  \label{5.12}

1156: \end{equation}

1157: where $\rho_H=m_1n_{1H}+m_2n_{2H}$.

1158: 

1159: It is instructive to consider first the solutions to these

1160: equations in the extreme long wavelength limit, $k=0$. In this

1161: case, the eigenvalues or hydrodynamic modes are given by

1162: \begin{equation}

1163: s_{\perp }=\frac{1}{2}\zeta ^{\ast },\hspace{0.3in}s_{n}=\left(

1164: 0,0,-\frac{1}{2} \zeta ^{\ast },\frac{1}{2}\zeta ^{\ast }\right),

1165: \label{5.14}

1166: \end{equation}

1167: where $s_n$ refers to the longitudinal modes. Two of the

1168: eigenvalues are positive, corresponding to growth of the initial

1169: perturbation in time. Thus, some of the solutions are unstable.

1170: The two zero eigenvalues represent marginal stability solutions,

1171: while the negative eigenvalue gives stable solutions. For general

1172: initial perturbations all modes are excited. These modes

1173: correspond to evolution of the fluid due to uniform perturbations

1174: of the HCS, i.e. a global change in the HCS parameters. The

1175: unstable modes are seen to arise from the initial perturbations

1176: $w_{\mathbf{k}\perp }(0)$ or $w_{\mathbf{k}||}(0)$. The marginal

1177: modes correspond to changes in the composition at fixed pressure,

1178: density, and velocity, and to changes in $\Pi_{{\bf

1179: k}}-\theta_{{\bf k}}$ at constant composition and velocity. The

1180: decaying mode corresponds to changes in the temperature or

1181: pressure for $\Pi_{{\bf k}}=\theta_{{\bf k}}$. The unstable modes

1182: may appear trivial since they are due entirely to the

1183: normalization of the fluid velocity by the time dependent thermal

1184: velocity. However, this normalization is required by the scaling

1185: of the entire set of equations to obtain time independent

1186: coefficients.

1187: 

1188: At finite wave vectors, these instabilities give rise to real

1189: growth of spatial perturbations. The linear growth of the

1190: transverse modes is simply given by

1191: \begin{equation}

1192: w_{\mathbf{k}\perp }(\tau )=w_{\mathbf{k}\perp }(0)\exp

1193: (\frac{1}{2}\zeta ^{\ast }-\eta ^{\ast }k^{2})\tau .  \label{5.23}

1194: \end{equation}

1195: The instability for the two shear modes is removed at sufficiently

1196: large $ k>k_{\perp }^{\text{c}}$, where

1197: \begin{equation}

1198: k_{\perp }^{\text{c}}=\left( \frac{\zeta ^{\ast }}{2\eta ^{\ast

1199: }}\right) ^{1/2}.  \label{5.25}

1200: \end{equation}

1201: The wave vector dependence of the remaining four modes is more

1202: complex. This is illustrated in Fig.\ \ref{fig10} showing the real

1203: parts of the modes $ s\left( k\right) $ for $\alpha _{ij}=0.9,$

1204: $\sigma _{1}/\sigma _{2}=1$, $ x_{1}=0.2$, and $m_{1}/m_{2}=4$.

1205: The $k=0$ values are those of (\ref{5.14}), corresponding to six

1206: hydrodynamic modes with two different degeneracies. The shear mode

1207: degeneracy remains at finite $k$ but the other is removed at any

1208: finite $k$. At sufficiently large $k$ a pair of real modes become

1209: equal and become a complex conjugate pair at all larger wave

1210: vectors, like a sound mode. The smaller of the unstable modes is

1211: that associated with the longitudinal velocity, which couples to

1212: the scalar hydrodynamic fields. It becomes negative at a wave

1213: vector smaller than that of Eq.\ (\ref{5.25}) and gives the

1214: threshold for development of spatial instabilities.

1215: 

1216: 

1217: 

1218: The results obtained here for the mixture show no new surprises

1219: relative to the earlier work for a one component gas,

1220: \cite{BDKS98,G05} with only the addition of the stable mass

1221: diffusion mode. Of course, the quantitative features can be quite

1222: different since there are additional degrees of freedom with the

1223: parameter set $\left\{ x_{1},T,m_{1}/m_{2},\sigma _{1}/\sigma

1224: _{2},\alpha _{ij}\right\} .$ Also, the manner in which these

1225: linear instabilities are enhanced by the nonlinearities may be

1226: different from that for the one component case.

1227: \begin{figure}[tbp]

1228: \begin{center}

1229: \resizebox{7.5cm}{!}{\includegraphics{fig10.eps}}

1230: \end{center}

1231: \caption{Dispersion relations for $\protect\alpha =0.9$,

1232: $x_{1}=0.2$, $ \protect\omega =1$ and $\protect\mu =4$.}

1233: \label{fig10}

1234: \end{figure}

1235: 

1236: 

1237: 

1238: \section{Summary and Discussion}

1239: \label{sec6}

1240: 

1241: 

1242: The Navier--Stokes order hydrodynamic equations have been

1243: discussed for a low density granular binary mixture. The form of

1244: the momentum flux is the same as for a one component gas, with

1245: only the value of the viscosity changed (see Appendix \ref{appA}).

1246: Since the dependence of the viscosity on the parameters of the

1247: mixture has been widely explored in a previous paper, \cite{MG03}

1248: attention has been focused here on the mass and heat fluxes and

1249: their associated transport coefficients. There is no phenomenology

1250: involved as the equations and the transport coefficients have been

1251: derived systematically from the inelastic Boltzmann equation by

1252: the Chapman-Enskog procedure. Consequently, there is no \emph{a

1253: priori} limitation on the degree of inelasticity, size and mass

1254: ratios, or composition. For practical purposes, the integral

1255: equations determining the transport coefficients have been solved

1256: by truncated expansions in Sonine polynomials. This is expected to

1257: fail at extreme values of size or mass ratio, \cite{MG03,GM04} but

1258: the results are quite accurate otherwise.

1259: 

1260: The hydrodynamic equations are the same as for a normal gas,

1261: except for a sink in the energy equation due to granular cooling,

1262: and additional transport coefficients in the mass and heat flux

1263: constitutive equations. The latter arise because the usual

1264: restrictions of irreversible thermodynamics no longer apply. These

1265: restrictions include Onsager reciprocal relations among various

1266: transport coefficients, and the extent to which these are violated

1267: has been demonstrated in Section 4. It has been verified that the

1268: results described here reduce to those for a normal mixture in the

1269: elastic limit, \cite{CC70} and to those for a one component

1270: granular gas \cite{BDKS98} when the species are mechanically

1271: identical.

1272: 

1273: As is the case for a normal gas, the hydrodynamic fields include

1274: only the global temperature even though two species temperatures

1275: can be defined. For a normal gas the species temperatures rapidly

1276: approach the global temperature due to equipartition. For the

1277: granular gas the species temperatures approach different values,

1278: but with the same time dependence as the global temperature. The

1279: transport coefficients have an additional dependence on the

1280: composition due to this time independent ratio of species

1281: temperatures. This has been illustrated in Fig.\ \ref{fig3} for

1282: the thermal diffusion coefficient, where the effect is seen to be

1283: large for large mass ratio (implying very different species

1284: temperatures. Generally, it was seen that the deviation due to

1285: inelasticity is enhanced for greater mechanical differences

1286: between the species.

1287: 

1288: The linear equations for small perturbations of the special

1289: homogeneous cooling solution to the hydrodynamic equations were

1290: obtained and discussed. In order to characterize the solutions in

1291: terms of modes, it is necessary to introduce dimensionless fields

1292: so that the time dependence of the reference HCS is eliminated.

1293: The resulting equations exhibit a long wavelength instability for

1294: three of the modes. This is quite similar to the case of a one

1295: component granular gas, \cite{BDKS98,G05} and in fact the same

1296: modes are unstable here. The additional diffusion mode for two

1297: species behaves as for a normal fluid. The consequences of this

1298: instability for a binary mixture were not studied here. This

1299: entails an analysis of the dominant nonlinearities which has not

1300: been performed as yet. Since there are additional degrees of

1301: freedom would be interesting to see if the density clustering that

1302: occurs for a one component system is more complex here (e.g.,

1303: species segregation).

1304: 

1305: \acknowledgments

1306: 

1307: Partial support of the Ministerio de Ciencia y Tecnolog\'{\i}a

1308: (Spain) through Grant No.  FIS2004-01399 (partially financed by

1309: FEDER funds) in the case of V.G. and ESP2003-02859 (partially

1310: financed by FEDER funds) in the case of J.M.M. is acknowledged. V.

1311: G. also acknowledges support from the European Community's Human

1312: Potential Programme HPRN-CT-2002-00307 (DYGLAGEMEM).

1313: 

1314: 

1315: \appendix

1316: \section{Some explicit expressions}

1317: \label{appA}

1318: 

1319: 

1320: 

1321: Navier--Stokes hydrodynamics retains terms up through second order

1322: in the gradients. As a scalar, the cooling rate has the most

1323: general form at this order given by

1324: \begin{eqnarray}

1325: \zeta  &=&\zeta _{0}+\zeta _{u}\nabla \cdot \mathbf{u}+\zeta

1326: _{x}\nabla ^{2}x_{1}+\zeta _{T}\nabla ^{2}T+\zeta _{p}\nabla

1327: ^{2}p+\zeta _{TT}\left( \nabla T\right) ^{2}+\zeta _{xx}\left(

1328: \nabla x_{1}\right) ^{2}\nonumber\\

1329: & &  +\zeta _{pp}\left( \nabla p\right) ^{2}+\zeta _{Tx}\left(

1330: \mathbf{\nabla }T\right) \cdot \left( \mathbf{\nabla }x_{1}\right)

1331: +\zeta _{Tp}\left( \mathbf{\nabla }T\right) \cdot \left(

1332: \mathbf{\nabla } p\right) +\zeta _{px}\left( \mathbf{\nabla

1333: }p\right) \cdot \left( \mathbf{ \nabla }x_{1}\right) \nonumber\\

1334: & & +\zeta _{uu}\left( \nabla _{i}u_{j}\right) \left( \nabla

1335: _{i}u_{j}\right) .  \label{a01}

1336: \end{eqnarray}

1337: For a low density gas the first order gradient term vanishes,

1338: \cite{BDKS98} $\zeta _{u}=0$. However, for higher densities $\zeta

1339: _{u}$ is different from zero. \cite{GD99a} The second order terms

1340: have been left implicit in equations (\ref{2.5}) and (\ref{2.6})

1341: for the temperature and the pressure. As noted in the text, these

1342: second order terms have been calculated for a one component fluid

1343: \cite{BDKS98} and found to be very small relative to corresponding

1344: terms from the fluxes. Consequently, they have been neglected in

1345: the linearized hydrodynamic equations (\ref{5.7}).

1346: 

1347: 

1348: \subsection{Mass flux parameters}

1349: 

1350: 

1351: 

1352: The transport coefficients are expressed in terms of a number of

1353: dimensionless parameters. For completeness, they are listed here.

1354: In general they depend on the reference distribution functions in

1355: the Chapman--Enskog expansion, which are not Maxwellians.

1356: \cite{GD99} The deviation of the reference distributions from

1357: their Maxwellian forms is measured by the cumulants $c_i$.

1358: \cite{GD99} As shown in Sec.\ \ref{sec3}, while the influence of

1359: these coefficients on the transport coefficients is not quite

1360: important in the case of the coefficients associated with the mass

1361: flux (see Fig.\ \ref{fig1}, for example), not happens the same in

1362: the case of the heat flux (see Fig.\ \ref{fig4}, for example)

1363: where the influence of $c_i$ is not negligible for strong

1364: dissipation. However, in order to offer a simplified theory the

1365: parameters given in this Appendix have neglected the corrections

1366: due to the cumulants $c_i$. The full expressions for the transport

1367: coefficients can be found in Ref.\ \onlinecite{GD02}.

1368: 

1369: 

1370: The temperature ratio $\gamma =T_{1}/T_{2}$ is determined from the

1371: condition

1372: \begin{equation}

1373: \label{n1} \zeta_1^*=\zeta_2^*=\zeta^*,

1374: \end{equation}

1375: where the dimensionless cooling rate, $\zeta_i ^{\ast}=\zeta_i

1376: /\nu _{0}$ ($\nu _{0}$ is the average frequency defined below

1377: (\ref{2.8})) is

1378: \begin{eqnarray}

1379: \zeta_1^{\ast } &=&\frac{2}{3}\sqrt{2\pi }\left( \frac{\sigma

1380: _{1}}{\sigma _{12}}\right) ^{2}x_{1}\theta _{1}^{-1/2}\left(

1381: 1-\alpha _{11}^{2}\right)

1382: \nonumber \\

1383: &&+\frac{4}{3}\sqrt{\pi }x_{2}\mu _{21}\left( \frac{1+\theta

1384: }{\theta }\right) ^{1/2}\left( 1+\alpha _{12}\right) \theta

1385: _{2}^{-1/2}\left[ 2-\mu _{21}\left( 1+\alpha _{12}\right) \left(

1386: 1+\theta \right) \right]. \label{a03}

1387: \end{eqnarray}

1388: The expression for $\zeta_2^*$ can be easily obtained by

1389: interchanging $1\leftrightarrow 2$. Here, $\theta _{1}=1/(\mu

1390: _{21}\gamma _{1})$, $\theta _{2}=1/(\mu _{12}\gamma _{2})$, $\mu

1391: _{ij}=m_{i}/(m_{i}+m_{j})$, $\delta =x_{1}/x_{2}$, and $\theta

1392: =\theta _{1}/\theta _{2}=\mu /\gamma $. The temperature ratios

1393: $\gamma_i$ are related to the temperature ratio $\gamma$ through

1394: Eqs.\ (\ref{2.11a}). In the quasielastic limit ($\alpha_{ij}\ll

1395: 1$), the temperature ratio $\gamma$ has the simple form

1396: \begin{eqnarray}

1397: \gamma  &\rightarrow&1+\frac{1}{2\mu _{12}\mu _{21}}\left\{ \left(

1398: \mu _{12}x_{1}-\mu

1399: _{21}x_{2}\right) \left( 1-\alpha _{12}\right) \right.   \nonumber \\

1400: &&\left. +\frac{1}{\sqrt{2}}\left[ \left( \frac{\sigma

1401: _{22}}{\sigma _{12}} \right) ^{2}x_{2}\sqrt{\mu _{12}}\left(

1402: 1-\alpha _{22}\right) -\left( \frac{ \sigma _{11}}{\sigma

1403: _{12}}\right) ^{2}x_{1}\sqrt{\mu _{21}}\left( 1-\alpha

1404: _{11}\right) \right] \right\} .  \label{a02}

1405: \end{eqnarray}

1406: In this limit, the temperature ratio is a linear function of the

1407: hydrodynamic field $x_{1}$. The dimensionless frequency $\nu

1408: ^{\ast}$ appearing in the expressions of the transport

1409: coefficients associated with the mass flux is

1410: \begin{equation}

1411: \nu ^{\ast }=\frac{4}{3}\mu _{21}\frac{1+\mu \delta }{1+\delta

1412: }\left( \frac{ 1+\theta }{\theta }\right) ^{1/2}\theta

1413: _{2}^{-1/2}(1+\alpha _{12}). \label{a04}

1414: \end{equation}

1415: The cooling rate $\zeta ^{\ast}$ and frequency $\nu ^{\ast }$ are

1416: also functions of the hydrodynamic field $x_{1}$. However, all the

1417: parameters above are independent of the temperature and density.

1418: 

1419: 

1420: 

1421: \subsection{Heat and momentum flux parameters}

1422: 

1423: 

1424: 

1425: As in the last section effects due to the distortion of the

1426: reference Maxwellian are neglected. In the case of the heat flux

1427: $\mathbf{q}$, Eq.\ ( \ref{2.2}), the transport coefficients

1428: $D^{\prime \prime }$, $L$, and $ \lambda $ are given by Eqs.\

1429: (\ref{2.12})--(\ref{2.14}), respectively. By using matrix

1430: notation, the (dimensionless) Sonine coefficients $ d_{i}^{\prime

1431: \prime }$, $\ell _{i}$, and $\lambda _{i}$ verify the coupled set

1432: of six equations \cite{GD02}

1433: \begin{equation}

1434: \Lambda _{\sigma \sigma ^{\prime }}X_{\sigma ^{\prime }}=Y_{\sigma

1435: }. \label{a1}

1436: \end{equation}

1437: where $X_{\sigma ^{\prime }}$ is the column matrix

1438: \begin{equation}

1439: \mathbf{X}=\left(

1440: \begin{array}{c}

1441: d_{1}^{\prime \prime } \\

1442: d_{2}^{\prime \prime } \\

1443: \ell _{1} \\

1444: \ell _{2} \\

1445: \lambda _{1} \\

1446: \lambda _{2}

1447: \end{array}

1448: \right) ,  \label{a2}

1449: \end{equation}

1450: and $\Lambda _{\sigma \sigma ^{\prime }}$ is the matrix

1451: \begin{equation}

1452: \Lambda =\left(

1453: \begin{array}{cccccc}

1454: \nu _{11}-\frac{3}{2}\zeta ^{\ast } & \nu _{12} & -\left(

1455: \frac{\partial \zeta ^{\ast }}{\partial x_{1}}\right) _{p,T} & 0 &

1456: -\left( \frac{\partial

1457: \zeta ^{\ast }}{\partial x_{1}}\right) _{p,T} & 0 \\

1458: \nu _{21} & \nu _{22}-\frac{3}{2}\zeta ^{\ast } & 0 & -\left(

1459: \frac{\partial \zeta ^{\ast }}{\partial x_{1}}\right) _{p,T} & 0 &

1460: -\left( \frac{\partial

1461: \zeta ^{\ast }}{\partial x_{1}}\right) _{p,T} \\

1462: 0 & 0 & \nu _{11}-\frac{5}{2}\zeta ^{\ast } & \nu _{12} & -\zeta

1463: ^{\ast } & 0

1464: \\

1465: 0 & 0 & \nu _{21} & \nu _{22}-\frac{5}{2}\zeta ^{\ast } & 0 &

1466: -\zeta ^{\ast }

1467: \\

1468: 0 & 0 & \zeta ^{\ast }/2 & 0 & \nu _{11}-\zeta ^{\ast } & \nu _{12} \\

1469: 0 & 0 & 0 & \zeta ^{\ast }/2 & \nu _{21} & \nu _{22}-\zeta ^{\ast

1470: }

1471: \end{array}

1472: \right) .  \label{a3}

1473: \end{equation}

1474: The column matrix $Y_{\sigma }$ has the elements

1475: \begin{equation}

1476: Y_{1}=D^{\ast }\left( \tau _{12}-\frac{\zeta ^{\ast }}{x_{1}\gamma

1477: _{1}^{2}} \right) -\frac{1}{\gamma _{1}^{2}}\left( \frac{\partial

1478: \gamma _{1}}{

1479: \partial x_{1}}\right) _{p,T},\hspace{0.3in}Y_{2}=-D^{\ast }\left( \tau

1480: _{21}-\frac{\zeta ^{\ast }}{x_{2}\gamma _{2}^{2}}\right)

1481: -\frac{1}{\gamma _{2}^{2}}\left( \frac{\partial \gamma

1482: _{2}}{\partial x_{1}}\right) _{p,T}, \label{a5}

1483: \end{equation}

1484: \begin{equation}

1485: Y_{3}=D_{p}^{\ast }\left( \tau _{12}-\frac{\zeta ^{\ast

1486: }}{x_{1}\gamma _{1}^{2}}\right) ,\hspace{0.3in}Y_{4}=-D_{p}^{\ast

1487: }\left( \tau _{21}-\frac{ \zeta ^{\ast }}{x_{2}\gamma

1488: _{2}^{2}}\right) ,  \label{a7}

1489: \end{equation}

1490: \begin{equation}

1491: Y_{5}=-\frac{1}{\gamma _{1}}+D^{\prime }{}^{\ast }\left( \tau

1492: _{12}-\frac{ \zeta ^{\ast }}{x_{1}\gamma _{1}^{2}}\right)

1493: ,\hspace{0.3in}Y_{6}=-\frac{1}{ \gamma _{2}}-D^{\prime }{}^{\ast

1494: }\left( \tau _{21}-\frac{\zeta ^{\ast }}{ x_{2}\gamma

1495: _{2}^{2}}\right) .  \label{a9}

1496: \end{equation}

1497: The dimensionless collision integrals $\tau _{12}$, $\nu _{11}$,

1498: and $\nu _{12}$ are given, respectively, by

1499: \begin{eqnarray}

1500: \tau _{12} &=&\frac{4}{3}\sqrt{\frac{\mu _{21}}{2}}\left(

1501: \frac{\sigma _{1}}{ \sigma _{12}}\right) ^{2}\gamma

1502: _{1}^{-3/2}(1-\alpha _{11}^{2}) \nonumber

1503: \label{a11} \\

1504: &&+\frac{4}{15}\mu _{21}^{-1}\gamma _{1}^{-4}(1+\alpha

1505: _{12})\left( \theta _{1}+\theta _{2}\right) ^{-1/2}\left( \theta

1506: _{1}\theta _{2}\right) ^{-3/2}\left( \frac{x_{2}}{x_{1}}A-\gamma

1507: B\right) ,

1508: \end{eqnarray}

1509: \begin{eqnarray}

1510: \label{a12}

1511:  \nu _{11} &=&\frac{16}{15}x_{1}\left( \frac{\sigma

1512: _{1}}{\sigma _{12}} \right) ^{2}(2\theta _{1})^{-1/2}(1+\alpha

1513: _{11})\left[ 1+\frac{33}{16}

1514: (1-\alpha _{11})\right]   \nonumber \\

1515: &&+\frac{2}{15}x_{2}\mu _{21}(1+\alpha _{12})\left( \frac{\theta

1516: _{1}}{ \theta _{2}(\theta _{1}+\theta _{2})}\right) ^{3/2}\left(

1517: E-5\frac{\theta _{1}+\theta _{2}}{\theta _{1}}A\right) ,

1518: \end{eqnarray}

1519: \begin{equation}

1520: \label{a13}

1521:  \nu _{12}=-\frac{2}{15}x_{2}\frac{\mu _{21}^{2}}{\mu

1522: _{12}}(1+\alpha _{12})\left( \frac{\theta _{1}}{\theta _{2}(\theta

1523: _{1}+\theta _{2})}\right) ^{3/2}\left( F+5\frac{\theta _{1}+\theta

1524: _{2}}{\theta _{2}}B\right) .

1525: \end{equation}

1526: In the above equations we have introduced the quantities

1527: \cite{misprints}

1528: \begin{eqnarray}

1529: \label{a14} A &=&5(2\beta _{12}+\theta _{2})+\mu _{21}(\theta

1530: _{1}+\theta _{2})\left[ 5(1-\alpha _{12})-2(7\alpha _{12}-11)\beta

1531: _{12}\theta _{1}^{-1}\right]

1532: \nonumber \\

1533: &&+18\beta _{12}^{2}\theta _{1}^{-1}+2\mu _{21}^{2}\left( 2\alpha

1534: _{12}^{2}-3\alpha _{12}+4\right) \theta _{1}^{-1}(\theta

1535: _{1}+\theta _{2})^{2}-5\theta _{2}\theta _{1}^{-1}(\theta

1536: _{1}+\theta _{2})

1537: \end{eqnarray}

1538: \begin{eqnarray}

1539: \label{a15} B &=&5(2\beta _{12}-\theta _{1})+\mu _{21}(\theta

1540: _{1}+\theta _{2})\left[ 5(1-\alpha _{12})+2(7\alpha _{12}-11)\beta

1541: _{12}\theta _{2}^{-1}\right]

1542: \nonumber \\

1543: &&-18\beta _{12}^{2}\theta _{2}^{-1}-2\mu _{21}^{2}\left( 2\alpha

1544: _{12}^{2}-3\alpha _{12}+4\right) \theta _{2}^{-1}(\theta

1545: _{1}+\theta _{2})^{2}+5(\theta _{1}+\theta _{2})

1546: \end{eqnarray}

1547: \begin{eqnarray}

1548: \label{a16} E &=&2\mu _{21}^{2}\theta _{1}^{-2}(\theta _{1}+\theta

1549: _{2})^{2}\left( 2\alpha _{12}^{2}-3\alpha _{12}+4\right) (5\theta

1550: _{1}+8\theta _{2})

1551: \nonumber \\

1552: &&-\mu _{21}(\theta _{1}+\theta _{2})\left[ 2\beta _{12}\theta

1553: _{1}^{-2}(5\theta _{1}+8\theta _{2})(7\alpha _{12}-11)+2\theta

1554: _{2}\theta

1555: _{1}^{-1}(29\alpha _{12}-37)-25(1-\alpha _{12})\right]   \nonumber \\

1556: &&+18\beta _{12}^{2}\theta _{1}^{-2}(5\theta _{1}+8\theta

1557: _{2})+2\beta

1558: _{12}\theta _{1}^{-1}(25\theta _{1}+66\theta _{2})  \nonumber \\

1559: &&+5\theta _{2}\theta _{1}^{-1}(11\theta _{1}+6\theta

1560: _{2})-5(\theta _{1}+\theta _{2})\theta _{1}^{-2}\theta

1561: _{2}(5\theta _{1}+6\theta _{2})

1562: \end{eqnarray}

1563: \begin{eqnarray}

1564: \label{a17} F &=&2\mu _{21}^{2}\theta _{2}^{-2}(\theta _{1}+\theta

1565: _{2})^{2}\left( 2\alpha _{12}^{2}-3\alpha _{12}+4\right) (8\theta

1566: _{1}+5\theta _{2})

1567: \nonumber \\

1568: &&-\mu _{21}(\theta _{1}+\theta _{2})\left[ 2\beta _{12}\theta

1569: _{2}^{-2}(8\theta _{1}+5\theta _{2})(7\alpha _{12}-11)-2\theta

1570: _{1}\theta

1571: _{2}^{-1}(29\alpha _{12}-37)+25(1-\alpha _{12})\right]   \nonumber \\

1572: &&+18\beta _{12}^{2}\theta _{2}^{-2}(8\theta _{1}+5\theta

1573: _{2})-2\beta

1574: _{12}\theta _{2}^{-1}(66\theta _{1}+25\theta _{2})  \nonumber \\

1575: &&+5\theta _{1}\theta _{2}^{-1}(6\theta _{1}+11\theta

1576: _{2})-5(\theta _{1}+\theta _{2})\theta _{2}^{-1}(6\theta

1577: _{1}+5\theta _{2})

1578: \end{eqnarray}

1579: Here, $\beta _{12}=\mu _{12}\theta _{2}-\mu _{21}\theta _{1}$. The

1580: corresponding expressions for $\tau _{21}$, $\nu _{22}$, and $\nu

1581: _{21}$ can be inferred from Eqs.\ (\ref{a11})--(\ref{a17}) by

1582: interchanging $ 1\leftrightarrow 2$. For elastic collisions, the

1583: expressions (\ref{a11})--( \ref{a17}) reduce to those obtained for

1584: hard sphere mixtures. \cite{CC70bis}

1585: 

1586: 

1587: The solution to Eq.\ (\ref{a1}) is

1588: \begin{equation}

1589: X_{\sigma }=\left( \Lambda ^{-1}\right) _{\sigma \sigma ^{\prime

1590: }}Y_{\sigma ^{\prime }}.  \label{a18.0}

1591: \end{equation}

1592: This relation provides an explicit expression for the coefficients

1593: $ d_{i}^{\prime \prime }$, $\ell _{i}$, and $\lambda _{i}$ in

1594: terms of the coefficients of restitution  and the parameters of

1595: the mixture. Their explicit forms are

1596: \begin{eqnarray}

1597: \label{a18} d_1^{\prime \prime}&=&\frac{1}{\Delta}\left\{2\left[2

1598: \nu_{12}Y_2-Y_1(2\nu_{22}-3\zeta^*)\right]\left[\nu_{12}\nu_{21}-\nu_{11}\nu_{22}

1599: +2(\nu_{11}+\nu_{22})\zeta^*-4\zeta^{*2}\right]\right.\nonumber\\

1600: && +2\left( \frac{\partial \zeta ^{\ast }}{\partial x_{1}}\right)

1601: _{p,T}(Y_3+Y_5)\left[2\nu_{12}\nu_{21}+2\nu_{22}^2-\zeta^*(

1602: 7\nu_{22}-6\zeta^{*})\right]\nonumber\\

1603: & & \left.-2\nu_{12}\left( \frac{\partial \zeta ^{\ast }}{\partial

1604: x_{1}}\right)_{p,T}(Y_4+Y_6)\left(2\nu_{11}+2\nu_{22}-7\zeta^*\right)\right\},

1605: \end{eqnarray}

1606: \begin{eqnarray}

1607: \label{a19} \ell_1&=&\frac{1}{\Delta}\left\{-2Y_3\left[2

1608: (\nu_{12}\nu_{21}-\nu_{11}\nu_{22})\nu_{22}+\zeta^*(7\nu_{11}\nu_{22}-5\nu_{12}\nu_{21}+2\nu_{22}^2

1609: -6\nu_{11}\zeta^*-7\nu_{22}\zeta^*+6\zeta^{*2})\right]\right.\nonumber\\

1610: &&

1611: +2Y_4\nu_{12}\left[2\nu_{12}\nu_{21}-2\nu_{11}\nu_{22}+2\zeta^*(\nu_{11}+\nu_{22})

1612: -\zeta^{*2}\right]\nonumber\\

1613: & &

1614: \left.+2Y_5\zeta^*\left[2\nu_{12}\nu_{21}+\nu_{22}(2\nu_{22}-7\zeta^*)+6\zeta^{*2}\right]

1615: -2\nu_{12}\zeta^*Y_6\left[2(\nu_{11}+\nu_{22})-7\zeta^*\right]

1616: \right\},

1617: \end{eqnarray}

1618: \begin{eqnarray}

1619: \label{a20} \lambda_1&=&\frac{1}{\Delta}\left\{-Y_3\zeta^*\left[2

1620: \nu_{12}\nu_{21}+\nu_{22}(2\nu_{22}-7\zeta^*)+6\zeta^{*2}\right]+\nu_{12}\zeta^*

1621: Y_4\left[2(\nu_{11}+\nu_{22})-7\zeta^*\right] \right. \nonumber\\

1622: &

1623: &-Y_5\left[4\nu_{12}\nu_{21}(\nu_{22}-\zeta^*)+2\nu_{22}^2(5\zeta^*-2\nu_{11})+2\nu_{11}

1624: (7\nu_{22}\zeta^*-6\zeta^{*2})+5\zeta^{*2}(6\zeta^*-7\nu_{22})\right]\nonumber\\

1625: & & \left.

1626: +\nu_{12}Y_6\left[4\nu_{12}\nu_{21}+2\nu_{11}(5\zeta^*-2\nu_{22})+\zeta^*(10\nu_{22}-

1627: 23\zeta^*)\right]\right\},

1628: \end{eqnarray}

1629: where

1630: \begin{equation}

1631: \label{a21}

1632: \Delta=\left[4(\nu_{12}\nu_{21}-\nu_{11}\nu_{22})+6\zeta^*(\nu_{11}+\nu_{22})-9\zeta^{*2}\right]

1633: \left[\nu_{12}\nu_{21}-\nu_{11}\nu_{22}+2\zeta^*(\nu_{11}+\nu_{22})-4\zeta^{*2}\right].

1634: \end{equation}

1635: The expressions for $d_2^{\prime \prime}$, $\ell_2$, and

1636: $\lambda_2$ can be obtained from Eqs.\ (\ref{a18})--(\ref{a20}) by

1637: setting $1\leftrightarrow 2$. From the above expressions one can

1638: easily get the transport coefficients $D^{\prime }$, $L$ and

1639: $\lambda $ from Eqs.\ (\ref{2.12})--(\ref{2.14}), respectively.

1640: They are functions of $x_{1}$ but independent of temperature and

1641: pressure.

1642: 

1643: 

1644: 

1645: The pressure tensor $P_{k,\ell }$ is given by

1646: \begin{equation}

1647: P_{k\ell }=p\delta _{k\ell }-\eta \left( \nabla _{\ell

1648: }u_{k}+\nabla _{k}u_{\ell }-\frac{2}{3}\delta _{k\ell }\nabla

1649: \cdot \mathbf{u}\right) , \label{a22}

1650: \end{equation}

1651: where $\eta $ is the shear viscosity coefficient. Its expression

1652: can be written as \cite{GD02}

1653: \begin{equation}

1654: \eta =\frac{nT}{\nu _{0}}\left( x_{1}\gamma _{1}^{2}\eta

1655: _{1}+x_{2}\gamma _{2}^{2}\eta _{2}\right) ,  \label{a22.1}

1656: \end{equation}

1657: with

1658: \begin{equation}

1659: \eta _{1}=\frac{2\gamma _{2}(2\lambda _{22}-\zeta ^{\ast

1660: })-4\gamma _{1}\lambda _{12}}{\gamma _{1}\gamma _{2}\left[ \zeta

1661: ^{\ast 2}-2\zeta ^{\ast }(\lambda _{11}+\lambda _{22})+4(\lambda

1662: _{11}\lambda _{22}-\lambda _{12}\lambda _{21})\right] },

1663: \label{a23}

1664: \end{equation}

1665: \begin{equation}

1666: \eta _{2}=\frac{2\gamma _{1}(2\lambda _{11}-\zeta ^{\ast

1667: })-4\gamma _{2}\lambda _{21}}{\gamma _{1}\gamma _{2}\left[ \zeta

1668: ^{\ast 2}-2\zeta ^{\ast}(\lambda _{11}+\lambda _{22})+4(\lambda

1669: _{11}\lambda _{22}-\lambda _{12}\lambda _{21})\right] }.

1670: \label{a24}

1671: \end{equation}

1672: The dimensionless quantities $\lambda _{ij}$ are given by

1673: \cite{GD02,MG03}

1674: \begin{eqnarray}

1675: \lambda _{11} &=&\frac{16}{5\sqrt{2}}x_{1}\left( \frac{\sigma

1676: _{1}}{\sigma _{12}}\right) ^{2}\theta _{1}^{-1/2}\left[

1677: 1-\frac{1}{4}(1-\alpha _{11})^{2}

1678: \right]   \nonumber  \label{a25} \\

1679: &&+\frac{8}{15}x_{2}\mu _{21}(1+\alpha _{12})\theta

1680: _{1}^{3/2}\theta _{2}^{-1/2}\left[ 6\theta _{1}^{-2}(\mu

1681: _{12}\theta _{2}-\mu _{21}\theta

1682: _{1})(\theta _{1}+\theta _{2})^{-1/2}\right.   \nonumber \\

1683: &&\left.+\frac{3}{2}\mu _{21}\theta _{1}^{-2}(\theta _{1}+\theta

1684: _{2})^{1/2}(3-\alpha _{12})+5\theta _{1}^{-1}(\theta _{1}+\theta

1685: _{2})^{-1/2}\right],

1686: \end{eqnarray}

1687: \begin{eqnarray}

1688: \lambda _{12} &=&\frac{8}{15}x_{2}\frac{\mu _{21}^{2}}{\mu

1689: _{12}}(1+\alpha _{12})\theta _{1}^{3/2}\theta _{2}^{-1/2}\left[

1690: 6\theta _{2}^{-2}(\mu _{12}\theta _{2}-\mu _{21}\theta

1691: _{1})(\theta _{1}+\theta

1692: _{2})^{-1/2}\right.   \nonumber  \label{a26} \\

1693: &&\left.+\frac{3}{2}\mu _{21}\theta _{2}^{-2}(\theta _{1}+\theta

1694: _{2})^{1/2}(3-\alpha _{12})-5\theta _{2}^{-1}(\theta _{1}+\theta

1695: _{2})^{-1/2}\right].

1696: \end{eqnarray}

1697: The corresponding expressions for $\lambda _{22}$ and $\lambda

1698: _{21}$ can be inferred from Eqs.\ (\ref{a23}) and (\ref{a24}) by

1699: interchanging $ 1\leftrightarrow 2$.

1700: 

1701: The program for calculating the cooling rates, the temperature

1702: ratio and the transport coefficients of the binary mixture can be

1703: obtained on request from the authors.

1704: 

1705: 

1706: \begin{thebibliography}{99}

1707: 

1708: \bibitem{GS95}A. Goldshtein and M. Shapiro, ``Mechanics of collisional motion of granular materials.

1709: Part 1. General hydrodynamic equations,'' J. Fluid Mech. {\bf

1710: 282}, 41 (1995); J. J. Brey, J. W. Dufty, and A. Santos,

1711: ``Dissipative dynamics for hard spheres,'' J. Stat. Phys. {\bf

1712: 87}, 1051 (1997).

1713: 

1714: \bibitem{BP04}N. V. Brilliantov and T. P\"oschel, {\em Kinetic Theory

1715: of Granular Gases} (Oxford University Press, Oxford, 2004).

1716: 

1717: 

1718: 

1719: \bibitem{CC70}S. Chapman and T. G. Cowling, {\em The Mathematical Theory of Nonuniform Gases}

1720: (Cambridge University Press, Cambridge, 1970).

1721: 

1722: 

1723: \bibitem{BDKS98}J. J. Brey, J. W. Dufty, C. S. Kim, and A. Santos,

1724: ``Hydrodynamics for granular flow at low density,'' Phys. Rev. E

1725: {\bf 58}, 4638 (1998); J. J. Brey and D. Cubero, ``Hydrodynamic

1726: transport coefficients of granular gases,''  in {\em Granular

1727: Gases} in {\em Lectures Notes in Physics}, edited by T. P\"oschel

1728: and S. Luding (Springer, Berlin, 2001), p. 59.

1729: 

1730: 

1731: \bibitem{BRMC99} See for instance, J. J. Brey, M. J. Ruiz-Montero,

1732: and D. Cubero, ``On the validity of linear hydrodynamics for

1733: low-density granular flows described by the Boltzmann equation,''

1734: Europhys. Lett. {\bf 48}, 359 (1999); J. J. Brey, M. J.

1735: Ruiz-Montero, D. Cubero, and R. Garc\'{\i}a-Rojo, ``Self-diffusion

1736: in freely evolving granular gases,'' Phys. Fluids {\bf 12}, 876

1737: (2000); V. Garz\'o and J. M. Montanero, ``Transport coefficients

1738: of a heated granular gas,'' Physica A {\bf 313}, 336 (2002); J. M.

1739: Montanero, A. Santos, and V. Garz\'o, ``DSMC evaluation of the

1740: Navier-Stokes shear viscosity of a granular fluid,'' in {\em

1741: Rarefied Gas Dynamics 24}, edited by M. Capitelli (AIP Conference

1742: Proceedings, 72, 2005), p. 803.

1743: 

1744: 

1745: 

1746: \bibitem{JM89}J. T. Jenkins and F. Mancini,

1747: ``Kinetic theory for binary mixtures of smooth, nearly elastic

1748: spheres,'' Phys. Fluids A {\bf 1}, 2050 (1989); P. Zamankhan,

1749: ``Kinetic theory for multicomponent dense mixtures of slightly

1750: inelastic spherical particles,''  Phys. Rev. E {\bf 52}, 4877

1751: (1995); B. Arnarson and J. T. Willits, ``Thermal diffusion in

1752: binary mixtures of smooth, nearly elastic spheres with and without

1753: gravity,'' Phys. Fluids {\bf 10}, 1324 (1998); J. T. Willits and

1754: B. Arnarson, ``Kinetic theory of a binary mixture of nearly

1755: elastic disks,'' Phys. Fluids {\bf 11}, 3116 (1999); M. Alam, J.

1756: T. Willits, B. Arnarson, and S. Luding, ``Kinetic theory of a

1757: binary mixture of nearly elastic disks with size and mass

1758: disparity ,'' Phys. Fluids {\bf 14}, 4085 (2002); B. Arnarson and

1759: J. T. Jenkins, ``Binary mixtures of inelastic hard spheres:

1760: Simplified constitutive theory,'' Phys. Fluids {\bf 16}, 4543

1761: (2004.)

1762: 

1763: \bibitem{MCK83}M. L\'opez de Haro, E. G. D. Cohen, and J. M.

1764: Kincaid, ``The Enskog theory for multicomponent mixtures. I.

1765: Linear response theory,'' J. Chem. Phys. {\bf 78}, 2746 (1983).

1766: 

1767: \bibitem{GD99}V. Garz\'o and J. W. Dufty, ``Homogeneous cooling

1768: state for a granular mixture,'' Phys. Rev. E  {\bf 60}, 5706

1769: (1999).

1770: 

1771: 

1772: \bibitem{computer}See for instance, J. M. Montanero and V. Garz\'o,

1773: ``Monte Carlo simulation of the homogeneous cooling state for a

1774: granular mixture,'' Granular Matter {\bf 4}, 17 (2002); A. Barrat

1775: and E. Trizac, ``Lack of energy equipartition in homogeneous

1776: heated binary granular mixtures,'' Granular Matter {\bf 4}, 57

1777: (2002); R. Clelland and C. M. Hrenya, ``Simulations of a

1778: binary-sized mixture of inelastic grains in rapid shear flow,''

1779: Phys. Rev. E {\bf 65}, 031301 (2002); S. R. Dahl, C. M. Hrenya, V.

1780: Garz\'o, and J. W. Dufty, ``Kinetic temperatures for a granular

1781: mixture,'' Phys. Rev. E {\bf 66}, 041301 (2002); R. Pagnani, U. M.

1782: B. Marconi, and A. Puglisi, ``Driven low density granular

1783: mixtures,'' Phys. Rev. E {\bf 66}, 051304 (2002); D. Paolotti, C.

1784: Cattuto, U. M. B. Marconi, and A. Puglisi,``Dynamical properties

1785: of vibrofluidized granular mixtures,''  Granular Matter {\bf 5},

1786: 75 (2003); P. Krouskop and J. Talbot, ``Mass and size effects in

1787: three-dimensional vibrofluidized granular mixtures,'' Phys. Rev. E

1788: {\bf 68}, 021304 (2003); H. Wang, G. Jin, and Y. Ma, ``Simulation

1789: study on kinetic temperatures of vibrated binary granular

1790: mixtures,'' Phys. Rev. E {\bf 68}, 031301 (2003); J. J. Brey, M.

1791: J. Ruiz-Montero, and F. Moreno, ``Energy partition and segregation

1792: for an intruder in a vibrated granular system under gravity,''

1793: Phys. Rev. Lett. {\bf 95}, 098001 (2005).

1794: 

1795: 

1796: \bibitem{exp}R. D. Wildman and D. J. Parker, `` Coexistence of two granular

1797: temperatures in binary vibrofluidized beds,'' Phys. Rev. Lett.

1798: {\bf 88}, 064301 (2002); K. Feitosa and N. Menon, ``Breakdown of

1799: energy equipartition in a 2D binary vibrated granular gas,'' Phys.

1800: Rev. Lett. {\bf 88}, 198301 (2002).

1801: 

1802: 

1803: \bibitem{JM87}J. Jenkins and F. Mancini, ``Balance laws and constitutive relations

1804: for plane flows of a dense bianry mixture of smooth, nearly

1805: elastic, circular disks,'' J. Appl. Mech. {\bf 54}, 27 (1987).

1806: 

1807: \bibitem{GD02}V. Garz\'o and J. W. Dufty, ``Hydrodynamics for a granular binary

1808: mixture at low density,'' Phys. Fluids {\bf 14}, 1476 (2002).

1809: 

1810: \bibitem {MG03}J. M. Montanero and V. Garz\'o, `Shear viscosity for a heated granular binary mixture

1811: at low-density,'' Phys. Rev. E {\bf 67}, 021308 (2003).

1812: 

1813: 

1814: \bibitem{GM04}V. Garz\'o and J. M. Montanero, ``Diffusion of impurities in a

1815: granular gas,'' Phys. Rev. E {\bf 69}, 021301 (2004).

1816: 

1817: 

1818: \bibitem{GDH05}L. Huilin, D. Gidaspow, and E. Manger, ``Kinetic theory of

1819: fluidized binary granular mixtures,'' Phys. Rev. E {\bf64}, 061301

1820: (2001); M.F. Ramahan, J. Naser, and P. J. Witt, ``An unequal

1821: temperature kinetic theory: description of granular flow with

1822: multiple particle classes,'' Powder Tech. {\bf 138}, 82 (2003); J.

1823: E. Galvin, S. R. Dahl, and C. M. Hrenya, ``On the influence of the

1824: equipartition-of-energy assumption in kinetic theories for

1825: rapidly-flowing, granular mixtures,'' J. Fluid Mech. {\bf 528},

1826: 207 (2005).

1827: 

1828: 

1829: \bibitem{DHGD02}S. R. Dahl, C. M. Hrenya, V. Garz\'o, and J. W.

1830: Dufty, ``Kinetic temperatures for a granular mixture,'' Phys. Rev.

1831: E {\bf 66}, 041301 (2002).

1832: 

1833: 

1834: 

1835: \bibitem{SGD04}A. Santos, V. Garz\'o, and J. Dufty, ``Inherent rheology

1836: of a granular fluid in uniform shear flow,'' Phys. Rev. E {\bf

1837: 69}, 061303 (2004).

1838: 

1839: 

1840: \bibitem{GM84}S. R. de Groot and P. Mazur, {\em Nonequilibrium Thermodynamics} (Dover, New York, 1984).

1841: 

1842: 

1843: 

1844: \bibitem{comment} Note that this coefficient is not strictly zero in

1845: the approximation used here due to the fact that while the set

1846: $\{d_{i}'', \ell_{i}, \lambda_{i}\}$ has been evaluated in the

1847: second Sonine approximation, the coefficients $D^*$, $D_p^*$, and

1848: $D'^*$ have been estimated in the first Sonine approximation.

1849: However, the magnitude of $L_p$ is negligible.

1850: 

1851: \bibitem{SD01}A. Santos and J. W. Dufty, ``Critical behavior of a heavy

1852: particle in a granular fluid,'' Phys. Rev. Lett. {\bf 86}, 4823

1853: (2001); ``Nonequilibrium phase transition for a heavy particle in

1854: a granular fluid,'' Phys. Rev. E {\bf 64}, 051305 (2001).

1855: 

1856: 

1857: \bibitem{G05}V. Garz\'o, ``Instabilities in a free granular fluid described

1858: by the Enskog equation,'' Phys. Rev. E {\bf 72}, 021106 (2005).

1859: 

1860: \bibitem{GD99a}V. Garz\'o and J. W. Dufty, ``Dense fluid transport for inelastic hard spheres,''

1861: Phys. Rev. E {\bf 59}, 5895 (1999).

1862: 

1863: \bibitem{misprints}Some misprints occur in the expressions given

1864: in Ref.\ \onlinecite{GD02} for these quantities. The expressions

1865: given here correct such results.

1866: 

1867: \bibitem{CC70bis}See Ref.\ \onlinecite{CC70}, Chap. 9, Eqs.\ (9.6,1)--(9.6,10).

1868: 

1869: 

1870: 

1871: 

1872: 

1873: \end{thebibliography}

1874: 

1875: 

1876: 

1877: 

1878: \end{document}

1879: