1: %  This is a LaTeX file for a paper by A.A.Ovchinnikov

2: % Figures are accessible as ps-files.

3: \documentstyle[aps,twocolumn]{revtex}

4: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

5: %TCIDATA{OutputFilter=LATEX.DLL}

6: %TCIDATA{LastRevised=Wed Aug 07 10:44:39 2002}

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

8: %TCIDATA{CSTFile=revtex.cst}

9: 

10: \input{epsf}

11: \renewcommand{\baselinestretch} {1.35}

12: \topmargin -27pt

13: \textwidth 6.3in

14: \textheight 8.8in

15: \def\be{\begin{eqnarray}}

16: \def\ee{\end{eqnarray}}

17: \def\b{\beta}

18: \def\d{\delta}

19: \def\g{\gamma}

20: \def\i{\infty}

21: \def\k{\kappa}

22: \def\l{\lambda}

23: \def\o{\omega}

24: \def\ol{\overline}

25: \def\p{\partial}

26: \def\s{\sigma}

27: \def\s{\sigma}

28: \def\e{\epsilon}

29: \def\f{\varphi}

30: \def\t{\theta}

31: \def\r{\rho}

32: \def\p{\partial}

33: 

34: \begin{document}

35: \title{Nonlinear Hall effect in a time-periodic electric

36: \ field and related phenomena.}

37: \author{A.A.~Ovchinnikov}

38: \address{{\it Max Planck Institute for Physics of Complex Systems, Dresden;\\

39: Joint Institute of Chemical Physics, RAS, Moscow.}}

40: 

41: \wideabs{

42: \maketitle

43: 

44: 

45: \begin{abstract}

46: 

47: A motion of neutral and charged particles located nearby a metal surface under

48: joint action of time-periodic electric field $E(t)$ directed normally to

49: the surface and permanent magnetic field $H$ directed along the surface has

50: been considered. It has been shown that due to a nonlinearity of the system

51: there exists a directed transport of the particles perpendicularly to both

52: magnetic and electric fields with velocity which is proportional to $E^2H$.

53: The velocity has been evaluated for charged (ions, electrons etc.) and

54: neutral (atoms, molecules, nano-size clusters etc.) particles adsorbed on a

55: metal surface. Corresponding surface electric current has been found as a

56: function of a frequency of the electric field and material parameters. It

57: has been noted that the similar phenomena appear in a bulk of electrically

58: inhomogeneous media.

59: \end{abstract}

60: }

61: 

62: PACS: 05.45.-a, 05.60.-k

63: 

64: The directed transport of particles under action of a time-periodic force is

65: now well established phenomenon in various non-linear systems. The details

66: of this transport has been studied for one-dimensional motion of particle in

67: the space-periodic potential $U(x)$ and a time periodic force having general

68: form $F(t)=F_{0}\cos {\omega t}+F_{1}\cos {(2\omega t+\theta )}$. (See

69: review \cite{1} and recent works \cite{2,3,4,5}.) Main physical reason for

70: such directed motion of the particles is due to the so called latent

71: asymmetry of driving force $F(t)$. Though the time average of $F(t)$ is zero

72: the time average $\overline{F^{3}(t)}$ doesn't vanish and depends on $\theta

73: $. Nonlinearity of the system leads to mixing of harmonics such a way that

74: average velocity is not zero and directed along $\overline{F^{3}(t)}$.

75: 

76: The aim of this paper is to show that the similar time- asymmetry arises in

77: a permanent magnetic field even in the case of purely harmonic external

78: force $F(t)$. The phenomenon is closely related with a conventional Hall

79: effect in conducting media. In difference with the latter the DC electric

80: current is arising on interface under a joint action of the harmonic

81: electric field directed normally to the surface and a permanent magnetic

82: field directed along the surface. The effect is of the second order with

83: respect to the amplitude of the electric field. For this reason we'll call

84: it the nonlinear Hall (NHL) effect.

85: 

86: Firstly we calculate the transverse velocity of a single charged particle

87: located on a metallic surface in the AC electric field and longitudinal

88: permanent magnetic field. Then, in Sections 2,3 we consider the similar

89: phenomenon in a surface electron gas of metals and semiconductors. In

90: Section 4 we evaluate the transport velocity of neutral particles in above

91: conditions. Finally we give a phenomenological description of the NLH effect

92: in inhomogeneous conducting media (Section 5).

93: 

94: \section{Directed motion of single charged particle.}

95: 

96: In order to demonstrate the physical mechanism of DC current let us consider

97: a motion of a single ion with a mass $M$ on metallic surface. A complete

98: description of an interaction between the ion and metallic surface is a

99: quite complicated problem. In framework of a density functional approach it

100: has been discussed in a recent monograph by Liebsch \cite{6}. There are a

101: few solidly established facts we are going to use in our paper. The

102: interaction between the ion and the metallic surface consists mainly of two

103: parts. The first part is an attractive Coulomb force between the ion and an

104: image charge having a potential $V_{im}(z)$

105: \begin{equation}

106: V_{im}(z)=-Q^{2}/(2z\epsilon )  \label{1}

107: \end{equation}

108: where $z$ is a distance between the ion and an edge of the metal surface and

109: $\epsilon $ is a static dielectric constant. We gave here the simplest

110: expression for such force at large $z$ neglecting the dynamics of the metal

111: electrons. More rigorous consideration includes an integration over the

112: complex surface polarisability, which modifies the $z$- dependence of $%

113: V_{im}(z)$. For the sake of simplicity we will use (1) with some

114: dimensionless parameter $c_{1}$

115: \begin{equation}

116: V_{im}(z)=-c_{1}Q^{2}/z  \label{2}

117: \end{equation}

118: The second part of the interaction is connected with a short-range repulsive

119: exchange potential $V_{ex}(z)$ which we choose in exponential form

120: \begin{equation}

121: V_{ex}(z)=c_{2}a_{1}\exp \{-(z-z_{e})/a_{1}\}  \label{3}

122: \end{equation}

123: were $z_{e}$ is an equilibrium distance between the phy\-sisorbed ion and

124: metal surface, $a_{1}$ is a radius of exchange forces on surface. The $c_{2}$

125: is chosen from a equilibrium condition

126: \begin{equation}

127: V(z)=V_{im}(z)+V_{ex}(z)  \label{4}

128: \end{equation}

129: \begin{equation}

130: \partial V/\partial z|_{z=z_{e}}=0  \label{5}

131: \end{equation}

132: \begin{equation}

133: c_{2}=c_{1}Q^{2}/z_{e}^{2}  \label{6}

134: \end{equation}

135: The motion of the physisorbed ion with a velocity $\nu $ leads to electronic

136: excitations in the metal and therefore to a friction force for such motion.

137: 

138: The friction force may be written as \cite{6}

139: \begin{equation}

140: {\bf f}=-M(\eta _{\parallel }\nu _{\parallel }+\eta _{\perp }\nu _{\perp })

141: \label{7}

142: \end{equation}

143: Here ${\bf \nu }$ is the particle velocity and $\eta _{\parallel }$, $\eta

144: _{\perp }$ are the friction coefficients for the motion parallel and normal

145: to the surface. If the charge is located outside the electronic density of

146: metal, it can be shown that $\eta _{\parallel }=\eta _{\perp }/2$ \cite{6}.

147: At large ion surface distances $z$ the friction parameter $\nu _{\perp }$

148: depends on $z$ as

149: \begin{equation}

150: \nu _{\perp }={\frac{{Q^{2}}}{M}}{\frac{{\omega _{F}}}{{k_{F}\omega _{p}^{2}l%

151: }}}{\frac{1}{{z^{3}}}}  \label{8}

152: \end{equation}

153: where $\hbar \omega _{F},k_{F},l$ and $\omega _{p}$ are a Fermi energy, a

154: Fermi momentum, a mean free path of electrons and a plasmon frequency

155: respectively.

156: 

157: For rough estimation it is convenient to use more simple formula

158: \begin{equation}

159: \nu _{\perp }={\frac{{Q^{2}}}{{M\sigma _{0}z^{3}}}}  \label{9}

160: \end{equation}

161: Here $\sigma _{0}$ is a bulk conductivity of a metal. The latter gives a

162: correct order of magnitude for $\nu _{\parallel }$ and $\nu _{\perp }$.

163: 

164: Having in mind all these facts one can readily write the equations of motion

165: of the ion under a joint action of the AC electric field $E_{0}\cos {\omega t%

166: }$ directed along $z$ (normally to the surface) and a permanent magnetic

167: field directed parallel to the surface along y-axis, $H_{y}=H$.

168: 

169: \begin{equation}

170: M{\ddot{z}}=-M\nu (z){\dot{z}}+{\frac{1}{c}}Q{\dot{x}}H-{\frac{\partial V}{%

171: \partial z}}+\xi _{z}(t)+QE_{0}(t)  \label{10}

172: \end{equation}

173: \begin{equation}

174: M{\ddot{x}}=-Mv_{\parallel }(z){\dot{x}}+{\frac{Q{\dot{z}}H}{c}}+\xi _{x}(t)

175: \label{11}

176: \end{equation}

177: Here $c$ is a light velocity. We included also into these equations the

178: random forces $\xi _{x(z)}(t)$ which reproduce the thermal fluctuations of

179: the ion velocity. As usual, they have the white noise correlations of the

180: following type

181: 

182: \begin{eqnarray}

183: &<&\xi _{x}(t_{1})\xi _{x}(t_{2})>=2\nu _{\parallel }(z)k_{B}T\delta

184: (t_{1}-t_{2})  \nonumber \\

185: &<&\xi _{z}(t_{1})\xi _{z}(t_{2})>=2\nu _{\perp }(z)k_{B}T\delta

186: (t_{1}-t_{2})  \label{12}

187: \end{eqnarray}

188: where $k_{B}$ is a Boltzmann constant and $T$ is a temperature of the system.

189: 

190: In spite of the comparative complexity of these equations they can be

191: analyzed analytically in the limit of the very small electric and magnetic

192: fields. At small amplitudes $E_{0}$, $H$ of external fields a displacement

193: of the ion $\delta z=z-z_{e}$ from its equilibrium position is quite small

194: and the first equation (10) could be chosen in a linear form

195: \begin{equation}

196: M{\ddot{z}}=-2M{\dot{z}}\nu _{0}-k_{e}\delta z+QE_{0}(t)  \label{13}

197: \end{equation}

198: where $k_{e}$ is a force constant of a vibration of ion along the z-axis

199: \begin{equation}

200: k_{e}={\frac{\partial ^{2}V}{\partial z^{2}}}\Bigr|_{z=z_{e}}  \label{14}

201: \end{equation}

202: and

203: \begin{equation}

204: \nu _{0}=2\nu _{\perp }(z_{e})  \label{15}

205: \end{equation}

206: We neglected in (10) the thermal fluctuations ($T=0$) and the Lorentz force

207: term. This is correct if we are interested in a linear contribution with

208: respect to magnetic field into the velocity of the transport. However we

209: have to keep this term in (11).

210: 

211: The nonlinearity of Eq.(11) is connected with damping term, which in

212: accepted approximation could be written as

213: \begin{equation}

214: -M\nu _{0}(1-\gamma \delta z){\dot{x}}  \label{16}

215: \end{equation}

216: Here $\gamma $ is a derivative of the damping function $\nu $ taken at

217: equilibrium distance $z_{e}$. Then, the Eq.(11) took the following form

218: \begin{equation}

219: M{\ddot{x}}=-M\nu _{0}(1-\gamma \delta z){\dot{x}}+{\frac{QH}{c}}{\dot{z}}

220: \label{17}

221: \end{equation}

222: The periodic solution of (13) is

223: \begin{equation}

224: \delta z(t)=Re{\Bigl\{{\frac{E_{0}(Q/M)e^{i\omega t}}{{\omega

225: _{e}^{2}-\omega ^{2}+i\nu _{0}2\omega }}}\Bigr\}}  \label{18}

226: \end{equation}

227: Here $\omega _{e}$ is an eigen-frequency of the physisorbed ion, $\omega

228: _{e}^{2}=k_{e}/M$. Then the periodic solution of the second linear equation

229: (17) could be written as the following

230: 

231: \begin{equation}

232: \begin{array}{ll}

233: {\dot{x}}=&\int_{0}^{\infty }d\xi \lbrack QH/cM]\exp [-\xi \nu _{0}]\delta {%

234: \dot{z}}(t-\xi )\times \\

235: &~~~\exp {\{-\gamma \nu _{0}\int_{t-\xi }^{t}\delta z(\tau

236: )d\tau \}}\\

237: \end{array}

238: \label{19}

239: \end{equation}

240: We are interested in the transverse velocity ${\dot{x}}$ averaged over the

241: period of the vibration in the external electric field $E_{0}(t)$. Making

242: use the necessary averaging we have a final expression

243: \begin{equation}

244: {\bar{v}}(\omega )={\frac{Q^{3}E_{0}^{2}H\gamma }{2cM^{3}}}{\frac{\omega ^{2}%

245: }{{\omega ^{2}+{\bf \nu }_{0}^{2}}}}{\frac{1}{{(\omega _{e}^{2}-\omega

246: ^{2})^{2}+4\omega ^{2}{\bf \nu }_{0}^{2}}}}  \label{20}

247: \end{equation}

248: The function ${\overline{v}}(\omega )$ has a resonance character and reaches

249: its maximal value at $\omega =\omega _{e}$. Since $\nu _{0}\sim M^{-1}$ and

250: $\omega _{e}^{2}\sim M^{-1}$ the ${\overline{v}}(\omega _{e})$ depends quite

251: weakly on the mass of ion. As a rule, $\nu _{0}\ll \omega _{e}$ (damping is

252: comparatively small) and

253: \begin{equation}

254: \overline{v}(\omega _{e})={\frac{Q^{3}E_{0}^{2}H\gamma }{2cM^{3}4\omega

255: _{e}^{2}\nu _{0}^{2}}}

256: \label{21}

257: \end{equation}

258: The behavior of $\overline{v}(\omega )$ as function of $\omega $ is given in

259: Fig.1 for three different values of mass. On the same figure we gave the

260: results of numerical solutions of Eqs.(10,11).

261: 

262: Let us estimate the possible velocities $\overline{v}$ at reasonable

263: experimental parameters

264: $$

265: \begin{array}{ll}

266: M\sim 10^{-24}g,~~&\gamma \sim 10^{8}cm^{-1},\\

267: \omega_{e}\sim 10^{12}sec^{-1},~~ &\nu_{0}\sim 0.01\omega_{e}\\

268: \end{array}

269: $$

270: and $Q$ is an electron charge. Then

271: \[

272: \overline{v}(\omega _{e})\sim 10^{-3}E_{0}^{2}H,

273: \]

274: where $E_{0}$ and $H$ have to be taken in emu system. For example, if $H\sim

275: 10^{3}Gauss$ and $E_{0}\sim 100V/cm$, $\overline{v}(\omega _{e})\sim 1$

276: cm/sec.

277: 

278: In order to clarify the physical source of the directed transport of the

279: particle we give a qualitative analysis of the equations(13) and (17). First

280: of all we neglect the dependence of the damping coefficient $\nu (z)$ on $z$%

281: , $\gamma =0$. Then, system of equations (13) and (17) became a linear one.

282: The solution of this system can be written as following

283: 

284: \begin{eqnarray}

285: \delta z &=&a_{0}(\omega )\cos {(\omega t+\varphi _{z}(\omega ))}  \nonumber

286: \\

287: x(t) &=&a_{0}(\omega )b_{0}(\omega )\cos {(\omega t+\varphi _{z}(\omega

288: )+\varphi _{x}(\omega ))}  \label{22}

289: \end{eqnarray}

290: where $a_{0}(\omega )$ and $\varphi _{z}(\omega )$ are modulus and argument

291: of a complex function

292: \[

293: {\frac{E_{0}Q}{M}}{\frac{1}{{\omega _{e}^{2}-\omega ^{2}+i2\nu _{0}\omega }}%

294: }

295: \]

296: and $b_{0}(\omega )$ and $\varphi _{x}(\omega )$ are modulus and argument of

297: a complex function ${%

298: %TCIMACRO{\dfrac{\omega _{L}}{{\nu _{0}+i\omega }}}%

299: %BeginExpansion

300: {\displaystyle{\omega _{L} \over {\nu _{0}+i\omega }}}%

301: %EndExpansion

302: }$.

303: 

304: A closed trajectory of the particle in the plane $(\delta z,x)$ is an

305: ellipse obeyed the following equation (Fig.2)

306: \begin{equation}

307: {\frac{(x-\delta zb_{0}\cos {\varphi _{x}})^{2}}{{a_{0}^{2}b_{0}^{2}\sin ^{2}%

308: {\varphi _{x}}}}}+{\frac{\delta z^{2}}{a_{0}^{2}}}=1  \label{23}

309: \end{equation}

310: If either $\omega _{L}/\nu _{0}$ or $\omega /\nu _{0}$ trends to zero the

311: ellipse turns into a line. An angle $\psi $ between axes $z$ and a largest

312: axis of the ellipse is proportional to $b_{0}(\omega )$. Points $A$ and $%

313: A^{\prime }$ on the Fig.2 are the points of return of the trajectory in the $%

314: x$-direction. At points $C$ and $C^{\prime }$ the trajectiry cross the line $%

315: \delta z=0$, and dashed lines $AB$ and $A^{\prime }B^{\prime }$ are parallel

316: to this line. In fact, the ellipse (23) is an attractor of the system

317: (13),(17).

318: 

319: Consider a small deformation of the trajectory taking into account the

320: dependence of $\eta (z)$ on $z$. According to (16)the damping coefficient $%

321: \eta (z)$ is higher at $\delta z<0$ ($\gamma >0$) and lower at $\delta z>0$.

322: After switching on this dependence the trajectory of the particle between

323: points $C$ and $B$ will be shifted a little bit to the right side. But this

324: shift will be compensated completely by the shift to the left side on $%

325: AC^{\prime }$ part of the trajectory. The same is true for $AC$ and $%

326: C^{\prime }B^{\prime }$ parts of the trajectory. Non-zero shift originates

327: from $AB$ and $A^{\prime }B^{\prime }$ parts. Indeed the velocity on the $AB$

328: part is positive and the trajectory will be shifted to the right side if we

329: diminish the damping coefficient of the particle. The velocity on $A^{\prime

330: }B^{\prime }$ is negative and if we increase the damping the trajectory will

331: be shifted also to the right side. All in all it leads to the shift of whole

332: trajectory to the right which is equivalent to the transport of particle in

333: this direction. The qualitative consideration given above does not depends

334: on details of the attractor and holds for any non-linear dynamics of the

335: system at.

336: 

337: Averaging the Eq.(17) over period of time we arrive to the following

338: equality for the shift in $x$-direction

339: \begin{equation}

340: {\overline{\delta x}}=\gamma \int \delta z{\dot{x}}dt  \label{24}

341: \end{equation}

342: The integral in (24) is nothing but an area of the ellipse $S$ and we have a

343: very simple formula

344: \begin{equation}

345: {\overline{\delta x}}=\gamma S  \label{25}

346: \end{equation}

347: Eq.(25) could be used directly for the calculation of ${\overline{v}}(\omega

348: )$ and of course it leads to the same result as (20).

349: 

350: Two additional remarks to the main expression (20). If the external AC

351: electric field is a sum of two electric fields with two different amplitudes

352: and frequencies $\omega _{1}$ and $\omega _{2}$ the effect is additive

353: \[

354: {\overline{v}}={\overline{v}}_{1}(\omega _{1})+{\overline{v}}_{2}(\omega

355: _{2})

356: \]

357: where $v_{1(2)}(\omega _{1(2)})$ are average velocities (20) with $%

358: E_{0}=E_{1(2)}$ and $\omega =\omega _{1(2)}$, correspondingly. Moreover, any

359: non-thermal electric noise in system leads to a directed transport of

360: charged particles. Corresponding average velocity could be obtained by

361: integration of (20) over a frequency distribution of the electric noise.

362: 

363: Now, if we have two weakly coupled particles with opposite charges and

364: different masses $M_{1}$ and $M_{2}$ the average velocity of their mass

365: centre can be defined as

366: \[

367: {\overline{v}}(\omega )={\frac{{M_{1}\overline{{v_{1}}}(\omega )-M_{2}%

368: \overline{{v_{2}}}(\omega )}}{{M_{1}+M_{2}}}}

369: \]

370: leading to some neutral current in system. More realistic situation with

371: strongly coupled charged particles will be considered in Section 4.

372: 

373: \section{The NLH effect on metal surface.}

374: 

375: The similar phenomenon takes play inside a metal on the boundary separating

376: two media having different conductivities. In this section we consider the

377: metal-insulator or metal-vacuum boundary as simplest example of such

378: interface. Consider dynamics of electronic gas in surface layer of metal

379: under an influence of external AC electric field $E_0(t)=E_0\cos{\omega t}$

380: directed perpendicularly to surface of the metal (along $z$ axis) and in the

381: magnetic field $H$ directed along the surface, axis $y$. At this geometry a

382: density of electronic gas, $\rho (z,t)$, is a function of normal coordinate $%

383: z$ and a time $t$. There are two velocity components of the electronic gas:

384: the normal component $v(z,t)$ and the tangential component $u(z,t)$ both of

385: which are functions of only $z$ and $t$. The electronic gas vibrates

386: perpendicularly to the surface. Due to the nonlinearity of this vibrations a

387: normal velocity $v(z,t)$ and density of electron gas $\rho (z,t)$ are

388: periodic functions of time with frequency $\omega$ having all overtones:

389: \[

390: v=v_1(z)\cos{(\omega t+\varphi_1)}+v_2(z)\cos{(2\omega t+\varphi_2)}\ldots

391: \]

392: \[

393: \rho=\rho_0+\rho_1(z)\cos{(\omega t+\varphi_3)}+\rho_2(z)\cos{(2\omega

394: t+\varphi_4)}\ldots

395: \]

396: The above amplitudes and phases of harmonics as functions of $z$ are a

397: subject of calculations. The Lorentz forces induced by magnetic field lead

398: to tangential vibration of electronic gas with amplitude which is

399: proportional to permanent magnetic field. Corresponding tangential velocity $%

400: u(z,t)$ has the same form as $v(z,t)$ with its own amplitudes and phases.

401: The density of tangential current is $j(z,t)=u(z,t)\rho(z,t)$. The time

402: averaging of $j(z,t)$ over period gives the DC component of the current.

403: 

404: Since the width of a surface layer and all other characteristic lengths

405: greater than atomic distances one can readily use hydrodynamics approach

406: (see \cite{6,7}) for description of system. In planar geometry all variables

407: are the functions of only one coordinate $z$ and a periodic functions of $t$

408: with frequency $\omega $. The main equations look as following \cite{6,7}

409: \begin{eqnarray}

410: &&\rho \Biggl({\frac{{\partial v}}{{\partial t}}}+v{\frac{{\partial v}}{{%

411: \partial z}}}\Biggr)=-{\frac{{\partial p}}{{\partial \rho }}}{\frac{{%

412: \partial \rho }}{{\partial z}}}+{\frac{{2\pi e}}{{m}}}Q(z)  \nonumber \\

413: &-&{\frac{{e}}{{m\mu }}}\rho v-{\frac{e}{m}}E_{0}(t)\rho -{\frac{{He}}{{mc}}}%

414: \rho u+\eta {\frac{{\partial ^{2}v}}{{\partial z^{2}}}}  \label{26}

415: \end{eqnarray}

416: \begin{equation}

417: \rho \Biggl({\frac{{\partial u}}{{\partial t}}}+v{\frac{{\partial u}}{{%

418: \partial z}}}\Biggr)={\frac{{He}}{{mc}}}\rho v-{\frac{{e}}{{m\mu }}}\rho

419: u+\eta {\frac{{\partial ^{2}u}}{{\partial z^{2}}}}  \label{27}

420: \end{equation}

421: \begin{equation}

422: {\frac{{\partial \rho }}{{\partial t}}}=-{\frac{{\partial }}{{\partial z}}}%

423: (v\rho )  \label{28}

424: \end{equation}

425: \begin{equation}

426: Q(z,t)=2q(z)-q(\infty )  \label{29}

427: \end{equation}

428: Here $p$ is a pressure of electronic gas in metal, $\mu $ is a mobility of

429: electron in metal, c is the light velocity, m is effective mass of

430: electrons, $\eta $ is a viscosity coefficient of an electron fluid. A

431: quantity

432: \begin{equation}

433: q(z,t)=e\int_{0}^{z}(\rho (z^{\prime },t)-\rho _{0})dz^{\prime }  \label{30}

434: \end{equation}

435: is a surface density of charge (together with positive charge of a lattice)

436: accumulated in a layer between $z=0$ and $z$. The boundary conditions for

437: this system are quite simple. Both velocities $v,u$ are equal to zero at the

438: edge of metal ($z=0$).

439: 

440: We are looking for a periodic solution of the equations (22-25). The surface

441: electric current directed along the surface and averaged over the period of

442: time is

443: \begin{equation}

444: I_{H}=\int_{0}^{\infty }e{\overline{\rho (z,t)u(z,t)}}dz  \label{31}

445: \end{equation}

446: The bar in Eq.(31) means the time averaging.

447: 

448: The equation (28) can be rewritten via the $q(z,t)$ as

449: \begin{equation}

450: {\frac{dq}{dt}}=-ev(z,t)\rho (z,t)  \label{32}

451: \end{equation}

452: From this equation we see immediately that the time averaged normal

453: component of the current

454: \[

455: {\overline{j}_{z}}(z)=0

456: \]

457: Then, using the Eq.(27) we can get the following formula for time averaged

458: tangential component of the current

459: \begin{equation}

460: \omega _{\mu }{\overline{j}_{x}(z)}=-e{\frac{\partial }{\partial z}}[{%

461: \overline{\rho (z,t)u(z,t)v(z)}}]+e{\overline{\eta {\frac{\partial ^{2}u}{%

462: \partial z^{2}}}}}  \label{33}

463: \end{equation}

464: The full surface electric current $I_{H}$ can be obtained by integration

465: over $z$ and looks as following

466: \begin{equation}

467: I_{H}=-{\frac{e\rho _{0}}{\omega _{\mu }}}{\overline{u_{\infty }v_{\infty }}}%

468: -{\frac{e}{m\omega _{\mu }}}{\frac{\partial \eta }{\partial \rho }}%

469: \int_{0}^{\infty }{\overline{{\frac{\partial \rho }{{\partial z}}}{\frac{%

470: \partial u}{{\partial z}}}}}dz  \label{34}

471: \end{equation}

472: where $v_{\infty }(t)$ and $u_{\infty }(t)$ are the normal and tangential

473: velocities for large $z$. The constant

474: \[

475: \kappa =m/{%

476: %TCIMACRO{\dfrac{\partial \eta }{\partial \rho }}%

477: %BeginExpansion

478: {\displaystyle{\partial \eta  \over \partial \rho }}%

479: %EndExpansion

480: }

481: \]

482: has a dimension of a kinematic viscosity $cm^{2}/sec$. At small electric and

483: magnetic fields first non-vanishing contribution to the $I_{H}$ can be

484: obtained if we know $q(z,t)$, $u(z,t)$ and $v(z,t)$ in a linear

485: approximation.

486: 

487: Taking the limit $z\rightarrow \infty $ we come to three equations for

488: asymptotic values $q_{\infty }=q(\infty ,t)$, $u_{\infty }=u(\infty ,t)$, $%

489: v_{\infty }=v(\infty ,t)$

490: \begin{equation}

491: {\frac{{d}}{dt}}q_{\infty }=-\rho _{0}v_{\imath }  \label{35}

492: \end{equation}

493: \begin{equation}

494: {\frac{{d}}{dt}}v_{\infty }=-v_{\imath }\omega _{\mu }+{\frac{e}{m}}%

495: E_{0}(t)-2\beta q_{\imath }(t)  \label{36}

496: \end{equation}

497: \begin{equation}

498: {\frac{{d}}{dt}}u_{\infty }=v_{\imath }\omega _{L}-u_{\imath }\omega _{\mu }

499: \label{37}

500: \end{equation}

501: Here $\omega _{\mu }=e/m\mu $, $\beta ={2\pi e^{2}/m}$ and $\omega

502: _{L}=eH/mc $ is a Larmor frequency. Periodic solutions of these linear

503: equations give the boundary conditions at $z\rightarrow \infty $. And $%

504: q(0,t)=0$ by definition. Since one needs only the periodic solution of

505: (26-29) the initial conditions are arbitrary. The system of equations

506: (26-29) has been analyzed both by perturbation theory with respect to

507: external electric field and numerically using standard NAG routines.

508: 

509: The periodic solution of linear equations (35-37) can be written in

510: following form

511: 

512: \begin{eqnarray}

513: v_{\infty }(t) &=&c_{e}Re[a(\omega )e^{i\omega t}]  \nonumber \\

514: q_{\infty }(t) &=&[\epsilon _{0}c_{e}/\omega _{\mu }]\rho _{0}Re[b(\omega

515: )e^{i\omega t}]  \label{38} \\

516: \eta _{\infty }(t) &=&\epsilon _{0}hc_{e}Re[c(\omega )e^{i\omega t}]

517: \nonumber

518: \end{eqnarray}

519: Here we introduced a dimensionless functions

520: \begin{eqnarray}

521: a(\omega ) &=&{\frac{{\omega _{\mu }\omega }\epsilon _{0}}{{(\omega

522: _{s}^{2}-\omega ^{2})+i\omega \omega _{{_{\mu }}}}}}  \nonumber \\

523: b(\omega ) &=&{\frac{{i\epsilon _{0}\omega _{\mu }^{2}}}{{(\omega

524: _{s}^{2}-\omega ^{2})+i\omega \omega _{{_{\mu }}}}}}  \label{39} \\

525: c(\omega ) &=&a(\omega ){\frac{{\omega }_{{L}}}{{\omega _{\mu }+i\omega }}}%

526: \text{{,}}  \nonumber

527: \end{eqnarray}

528: where $\epsilon _{0}$ and $h$ are the dimensionless electric and magnetic

529: fields

530: \[

531: \epsilon _{0}={\frac{eE_{0}}{{\omega _{\mu }c_{e}m}}},~~~h={\frac{eH}{{%

532: mc\omega _{\mu }}}},~~~\omega _{s}^{2}={\frac{2\pi e^{2}\rho _{0}}{m}}

533: \]

534: Using (26-29) as boundary conditions at an infinite $z$ we arrive to the

535: following solutions of the Eqs.(26-29)

536: 

537: \begin{eqnarray}

538: v(z,t) &=&c_{e}Re[a(\omega )e^{i\omega t}(1-e^{-\xi k})]  \nonumber \\

539: q(z,t) &=&\rho _{0}Re[b(\omega )e^{i\omega t}(1-e^{-\xi k})]  \label{40} \\

540: u(z,t) &=&c_{e}Re[c(\omega )e^{i\omega t}(1-e^{-\xi k})]  \nonumber

541: \end{eqnarray}

542: 

543: Here $\xi $ is a dimensionless coordinate, $z=c_{e}\xi /\omega _{\mu }$ ,

544: and complex $k=k_{1}+ik_{2}$ is defined from the equation

545: \begin{equation}

546: (k_{1}+ik_{2})^{2}=[2\omega _{s}^{2}-\omega ^{2}+i\omega \omega _{\mu

547: }]/\omega _{\mu }^{2},~~k_{2}>0,  \label{41}

548: \end{equation}

549: We are choosing a root which has $k_{2}>0$. Omitting intermediate

550: calculations we give a final expression for $I_{H}$ in following form

551: \begin{equation}

552: I_{H}=-{\frac{{ec_{e}^{2}\rho _{0}}}{{2\omega _{\mu }}}}h\epsilon {_{0}}%

553: ^{2}\gamma _{1}(\omega )+{\frac{\kappa \rho _{0}e}{2}}\gamma _{2}(\omega );

554: \label{42}

555: \end{equation}

556: $\gamma _{1}$ and $\gamma _{2}$ are dimensionless functions of a frequency $%

557: \omega $ of the external field

558: \[

559: \begin{array}{ll}

560: \gamma _{1}(\omega )= & a^{\ast }(\omega )c(\omega )+c^{\ast }(\omega

561: )a(\omega ); \\

562: \gamma _{2}(\omega )= & [k^{2}k^{\ast }c^{\ast }(\omega )b(\omega )+h.c.][{%

563: k+k^{\ast }}]^{-1}

564: \end{array}

565: \]

566: Both of them have resonance character and reach their peaks at $\omega

567: =\omega _{s}$ as it shown on Fig.3. Maximal value of them are $\gamma

568: _{1}(\omega _{s})=2\omega _{\mu }^{2}/\omega _{s}^{2}$ and $\gamma

569: _{2}(\omega _{s})=1$. For metals at low temperature as a rule $\omega _{\mu

570: }^{2}/\omega _{s}^{2}\ll 1$. At very high frequency $\omega \gg \omega _{s}$

571: both $\gamma _{1}(\omega )$ and $\gamma _{2}(\omega )$ decrease as $1/\omega

572: ^{4}$. At low frequency $\omega \ll \omega _{\mu}$ they behave as $\omega

573: ^{2}$: $\gamma _{1}(\omega )\sim \omega _{\mu }^{2}\omega ^{2}/\omega

574: _{s}^{4}$ and $\gamma _{2}(\omega )=\omega ^{2}/\omega _{s}^{2}$.

575: 

576: Thus, at low temperature main contribution to the surface current $I_{H}$

577: comes from the second term of (38), depending on the viscosity of electronic

578: gas. According to \cite{7a} the viscosity of electronic Fermi liquid in

579: metal at low temperature could be estimated as

580: \begin{equation}

581: \eta \cong m\rho _{0}c_{e}d(\epsilon _{F}/T)^{2}  \label{43}

582: \end{equation}

583: where $d$ is a lattice constant and $\epsilon _{F}$ is a Fermi energy of a

584: metal.

585: 

586: Thus, at low temperature the surface current $I_{H}$ can be expressed as

587: \[

588: I_{H}=-{\frac{\omega ^{2}e\rho _{0}dc_{e}}{4\omega _{s}^{2}}}\Biggl({\frac{%

589: \epsilon _{F}}{k_{B}T}}\Biggr)^{2}\Biggl({\frac{\mu E}{c_{e}}}\Biggr)^{2}%

590: \Biggl({\frac{\mu M}{c}}\Biggr)

591: \]

592: It is easy to see that the temperature dependence of the effect is

593: determined by the factor $\sigma ^{3}(T)/T^{2}$ ($\sigma (T)$ is a

594: conductivity of the metal).

595: 

596: To demonstrate the value of the effect let us take material parameters of

597: pure copper at low temperature, say, T=30~K:

598: \[

599: d=10^{-8}~cm,~~~~c_{e}=10^{8}~{\frac{cm}{sec}},~~~~\rho

600: _{0}=10^{22}~cm^{-3}.

601: \]

602: The conductivity behave as $T^{-5}$:

603: \[

604: \sigma (T)=\sigma _{R}(T_{R}/T)^{5}

605: \]

606: where $\sigma _{R}=10^{6}~om^{-1}$ is a room temperature conductivity and $%

607: T_{R}=300~K$.

608: 

609: Finally, for the pure cooper at $T=300~K$ we have

610: \[

611: I_{H}=-0.1\nu ^{2}E_{0}^{2}H

612: \]

613: where $\nu $ is expressed in MHz, $E_{0}$ in a V/cm, $H$ in a $Tesla$, $I_{H}$

614: is in a $\mu A/cm$. At $E_{0}=10~V/cm$, $H=0.01~T$, $\nu =10~MHz$ one has $%

615: I_{H}=\sim 1.0~\mu A/cm$.

616: 

617: \section{NLH current in an injection layer of a semiconductor.}

618: 

619: In the previous Section we have considered the dynamics of\ electron gas

620: inside a metal \ on the boundary metal-semiconductor. If a work function for

621: transition of electrons from metal side to the semiconductor is not too

622: high electrons of the metal can be injected into the semiconductor creating

623: an injection layer of electrons inside of the semiconductor. The injection

624: layer can give a contribution to the NLH current on the interface

625: metal-semiconductor. An estimation of the NLH current in this injection

626: layer is the subject of present section. First of all, let us consider the

627: equilibrium electron density in the semiconductor side. Free energy of

628: electron gas per unit area of the surface has the following form

629: \cite[Liebsch]{6}

630: \begin{eqnarray}

631: &F =\int\limits_{0}^{\infty }f(\rho ,T)dz+2\pi e^{2}\int\limits_{0}^{\infty

632: }\rho (z)dz\int\limits_{0}^{\infty }z\rho (z)dz-  \nonumber \\

633: &e^{2}\pi \int\limits_{0}^{\infty }\int\limits_{0}^{\infty

634: }\rho (z)|z-z^{\prime }|\rho (z^{\prime })dzdz^{\prime }+\psi

635: _{0}\int\limits_{0}^{\infty }\rho (z)dz  \label{44}

636: \end{eqnarray}

637: where $f=(\rho (z),T)$ is a density of the free energy of injected electron

638: gas, which is a function of electron density $\rho (z)\ $and temperature $T$%

639: . At small density $\rho (z)\ $the $f=(\rho ,T)$ is a free energy of ideal

640: classical gas.

641: 

642: \[

643: f(\rho ,T)=T\rho \ln \rho /{\rho }_{T}

644: \]

645: 

646: We use the units where Boltzmann constant $k_{B}=1$. $\Psi _{0}$ is.a work

647: function for transfer electron from metal to the semiconductor. Minimization

648: of $F$\ with respect to $\rho (z)$ leads to the well known equation for

649: equilibrium density $\rho _{e}(z)$

650: 

651: \begin{eqnarray}

652: {\displaystyle{{\partial f} \over {\partial \rho }}}%

653: &+{2}\pi e^{2}z\int\limits_{0}^{\infty }\rho _{e}(z)dz

654: -e^{2}2\pi

655: \int\limits_{0}^{\infty }\rho _{e}(z^{\prime })|z-z^{\prime }|dz^{\prime }

656: \nonumber \\

657: &+2\pi e^{2}\int\limits_{0}^{\infty }z\rho _{e}(z)dz+\Psi _{0} =0\nonumber\\

658: \label{45}

659: \end{eqnarray}

660: 

661: For Boltzmann electron gas the solution of \ (45) looks as following

662: 

663: \begin{equation}

664: \rho _{e}(z)={\frac{z_{d}q_{\infty }^{2}}{(zz_{d}q_{\infty }+1)^{2}}}

665: \label{46}

666: \end{equation}

667: Here \ $q_{\infty }=\int_{0}^{\infty }\rho (z)dz\ $and

668: 

669: \[

670: z_{d}={%

671: %TCIMACRO{\dfrac{2\pi e^{2}}{T}}%

672: %BeginExpansion

673: {\displaystyle{2\pi e^{2} \over T}}%

674: %EndExpansion

675: },

676: \]

677: is full injected charge per unit of area of surface. An equality of chemical

678: potential of the metal and the semiconductor on the surface of the metal

679: 

680: \begin{equation}

681: \frac{{\partial f}}{{\partial \rho }}\Biggl|_{\rho =\rho _{e}(0)}=-\Psi _{0}

682: \label{47}

683: \end{equation}

684: gives

685: \begin{equation}

686: \rho _{e}(0)=\rho _{T}e^{-\Psi /T}  \label{48}

687: \end{equation}

688: and

689: \begin{equation}

690: q_{\infty }=(\rho _{T}/z_{d})^{1/2}e^{-\Psi /2T}  \label{49}

691: \end{equation}

692: In fact, the equations (45-49) summarise briefly the simplest theory of

693: Schottky barrier on the metal-semiconductor interface \cite{8}. Under

694: action of time-periodic external electric field $E_{0}\sin {\omega t}$ the

695: non-linear oscillations of the density $\rho (z,t)$ takes place. A system of

696: equations describing these oscillations looks very much like that of the

697: metal-vacuum interface equations (26-29), but the background positive charge

698: $\rho _{0}$ in Eq.(29) has to be taken zero.

699: 

700: Another important difference between the metal-insulator and

701: metal-semiconductor cases is connected with boundary conditions for

702: velocities $v(z,t)$ and $u(z,t)$ at $z=0$. There are two limit cases. In the

703: first one, the exchange of charges over the surface is quite slow which

704: means that total charge in the semiconductor is conserving under action of

705: the AC electric field and $v(0,t)=0$, $u(0,t)=0$ at the boundary. If we

706: neglect the viscosity a total NLH current vanishes.

707: 

708: Let us outline briefly the solution of the hydrodynamics equations (26-29)

709: in this case. At small electric $E_{0}$ and magnetic $H$\ fields and small

710: viscosity of electron liquid the first equation (26) can be rewritten as

711: \begin{equation}

712: \rho \omega _{\mu }=-c_{e}^{2}{\frac{{\partial \rho }}{{\partial z}}}+{\rho

713: 4\pi (\rho (z)-\rho }_{{\infty }}{)+eE(t)\rho }  \label{50}

714: \end{equation}

715: where

716: \begin{equation}

717: {\rho (z,t)=}\int_{0}^{\infty }\rho (z)dz  \label{51}

718: \end{equation}

719: and

720: \begin{equation}

721: c_{e}^{2}={\rho \frac{{\partial }^{{2}}{f}}{{\partial \rho }^{{2}}}}

722: \label{52}

723: \end{equation}

724: For Boltzmann gas $c_{e}^{2}=\sqrt{T/m}$\ . The equation of continuity can

725: be rewritten as

726: 

727: \begin{equation}

728: {\frac{{\partial q}}{{\partial t}}=-v\rho }  \label{53}

729: \end{equation}

730: In the equation for $u(z,t)$\ we have to keep small terms connected with the

731: magnetic field (Lorentz force) and viscosity. Then, this equation can be

732: rewritten in the following form

733: 

734: \begin{equation}

735: {\frac{{\partial u\rho }}{{\partial t}}+\frac{{\partial }}{{\partial z}}%

736: vu\rho ={\omega }_{{L}}v\rho -{\omega }\mu u\rho +\frac{{\eta }}{{m}}\frac{{%

737: \partial }^{{2}}{u}}{{\partial z}^{{2}}}}  \label{54}

738: \end{equation}

739: 

740: Having in mind that we are interested in the full surface current averaged

741: over period of oscillation of external electric field

742: 

743: \begin{equation}

744: I_{H}{=e}\int_{0}^{\infty }\overline{\rho u}dz  \label{55}

745: \end{equation}

746: we arrive to the expression

747: 

748: \begin{equation}

749: I_{H}{=}\frac{{e}}{{\omega _{\mu }}}\int_{0}^{\infty }\frac{{\eta }}{{m}}{%

750: \frac{{\partial }^{{2}}{u}}{{\partial z}^{{2}}}}dz  \label{56}

751: \end{equation}

752: Here again the bar means the averaging over period of oscillation. We took

753: into account also that

754: 

755: \begin{equation}

756: \overline{{j}_{{z}}}=\overline{v\rho }{=0}  \label{57}

757: \end{equation}

758: due to the second equation.

759: 

760: As a rule, the viscosity of electronic gas $\eta $ depends on the density of

761: electron and, thus, depend on $z$\ . It is more convenient to rewrite (56) as

762: 

763: \begin{equation}

764: I_{H}{=-}\frac{{e}}{{\omega _{\mu }}\ m}\int_{0}^{\infty }({\frac{{\partial

765: \eta }}{{\partial \rho }}})\frac{{\partial \rho }}{{\partial z}}{\frac{{%

766: \partial u}}{{\partial z}}}dz  \label{58}

767: \end{equation}

768: Further we shall show that

769: 

770: \[

771: \frac{{\partial u}}{{\partial z}}\Biggl|_{z=0}=0

772: \]

773: At small viscosity and small ${\omega /\omega _{\mu }}$\ (low frequency of

774: external field)

775: 

776: \[

777: {u=\frac{{{\omega }_{{L}}}}{{\omega _{\mu }}}v}

778: \]

779: and we can express the $I_{H}$\ via normal velocity $v(z,t)$\ and density $%

780: \rho (z,t)$\ found from equation (50)

781: 

782: \begin{equation}

783: I_{H}{=-}\frac{{e{\omega }_{{L}}}}{m{\omega _{\mu }}^{{2}}}\int_{0}^{\infty

784: }({\frac{{\partial \eta }}{{\partial \rho }}})\frac{{\partial \rho }}{{%

785: \partial z}}{\frac{{\partial u}}{{\partial z}}}dz  \label{59}

786: \end{equation}

787: The viscosity of electron gas is unknown function of density of the gas.

788: Father, for estimation, we assume that is $%

789: %TCIMACRO{\dfrac{{1}}{{m}}}%

790: %BeginExpansion

791: {\displaystyle{{1} \over {m}}}%

792: %EndExpansion

793: %TCIMACRO{\dfrac{{\partial \eta }}{{\partial \rho }}}%

794: %BeginExpansion

795: {\displaystyle{{\partial \eta } \over {\partial \rho }}}%

796: %EndExpansion

797: $ is approximately equal to kinematic viscosity of electronic gas,

798: 

799: \[

800: \kappa {=}%

801: %TCIMACRO{\dfrac{{1}}{{m}}}%

802: %BeginExpansion

803: {\displaystyle{{1} \over {m}}}%

804: %EndExpansion

805: %TCIMACRO{\dfrac{{\partial \eta }}{{\partial \rho }}}%

806: %BeginExpansion

807: {\displaystyle{{\partial \eta } \over {\partial \rho }}}%

808: %EndExpansion

809: ,

810: \]

811: taken at equilibrium surface density of electron gas, ${\rho }_{eq}(0).$Thus

812: (60) turns into the following expression

813: 

814: \begin{equation}

815: I_{H}{=-}\frac{{e{\omega }_{{L}}}}{{\omega _{\mu }}^{{2}}}\kappa

816: \int_{0}^{\infty }dz\frac{{\partial \rho }}{{\partial z}}{\frac{{\partial v}%

817: }{{\partial z}}}  \label{60}

818: \end{equation}

819: which we will use for estimation of nonlinear Hall surface current.

820: 

821: The solution of the eqs. (50) will be performed by theory of perturbation

822: with respect to a small amplitude of electric field $E_{0}$ . After

823: introduction of dimensionless time $\tau =\omega t$\ , coordinate $z={%

824: %TCIMACRO{\dfrac{{c_{{e}}\xi }}{\sqrt{{\omega \omega _{\mu }}}}}%

825: %BeginExpansion

826: {\displaystyle{{c_{{e}}\xi } \over \sqrt{{\omega \omega _{\mu }}}}}%

827: %EndExpansion

828: }$, electric field ${\epsilon }_{{0}}={%

829: %TCIMACRO{\dfrac{eE_{0}}{m{c_{{e}}}\sqrt{{\omega \omega _{\mu }}}}}%

830: %BeginExpansion

831: {\displaystyle{eE_{0} \over m{c_{{e}}}\sqrt{{\omega \omega _{\mu }}}}}%

832: %EndExpansion

833: }$\ and new function $\lambda (z,t)$.

834: 

835: \begin{equation}

836: q_{(z,t)}=q_{\infty }(1-\alpha \lambda (z,t))  \label{61}

837: \end{equation}

838: the (50) turns into the following equation

839: 

840: \begin{equation}

841: {\frac{{\partial }\lambda }{{\partial \tau }}=\frac{{\partial }^{{2}}\lambda

842: }{{\partial \xi }^{{2}}}+2\eta \frac{{\partial }\lambda }{{\partial \xi }}%

843: +\epsilon \frac{{\partial }\lambda }{{\partial \xi }}\cos \tau }  \label{62}

844: \end{equation}

845: which has to be solved with boundary conditions

846: 

847: \[

848: \lambda (\xi )\rightarrow 0\ \ \text{at}\ \ \xi \rightarrow \infty

849: \]

850: 

851: \[

852: \lambda (0)=\frac{{1}}{{\alpha }}

853: \]

854: Here we use the following notation

855: 

856: \[

857: {\alpha }={%

858: %TCIMACRO{

859: %\dfrac{\sqrt{{\omega \omega _{\mu }}}{c_{{e}}}m_{0}}{2\pi q_{\infty }e^{2}}}%

860: %BeginExpansion

861: {\displaystyle{\sqrt{{\omega \omega _{\mu }}}{c_{{e}}}m_{0} \over 2\pi q_{\infty }e^{2}}}%

862: %EndExpansion

863: ,}\beta ={%

864: %TCIMACRO{\dfrac{{c_{{e}}}}{\sqrt{{\omega \omega _{\mu }}}}}%

865: %BeginExpansion

866: {\displaystyle{{c_{{e}}} \over \sqrt{{\omega \omega _{\mu }}}}}%

867: %EndExpansion

868: }

869: \]

870: 

871: \[

872: {v}=-\beta \omega (\frac{{\partial }\lambda }{{\partial \tau }})/{(\frac{{%

873: \partial }\lambda }{{\partial \xi }})}

874: \]

875: Note that we are looking for the periodic solution of (63). If we make a

876: substitution

877: 

878: \begin{equation}

879: \lambda =-({\frac{{\partial \varphi }}{{\partial \xi }})/\varphi }

880: \label{63}

881: \end{equation}

882: the nonlinear eq. (63) turns into linear one

883: 

884: \begin{equation}

885: {\frac{{\partial \varphi }}{{\partial \tau }}=\frac{{\partial }^{{2}}{%

886: \varphi }}{{\partial \xi }^{{2}}}+\epsilon \cos \tau \frac{{\partial \varphi

887: }}{{\partial \xi }}}  \label{64}

888: \end{equation}

889: with the boundary conditions

890: 

891: \[

892: {\varphi }(\xi )\rightarrow \infty \ \ \text{at}\ \ \xi \rightarrow \infty

893: \]

894: 

895: \[

896: \frac{{\partial \varphi }}{{\partial \xi }}\Biggl|_{{\xi }=0}=\frac{{1}}{{%

897: \alpha }}{\varphi (0)}

898: \]

899: 

900: The perturbative analysis of linear Eq. (65) is rather trivial and we give

901: only the final result

902: 

903: \begin{equation}

904: \lambda (\xi ,t)=\lambda _{0}(\xi ){+\epsilon }\lambda _{1}(\xi ,t)

905: \label{65}

906: \end{equation}

907: 

908: \begin{equation}

909: \lambda _{0}=\frac{{1}}{\xi +{\alpha }}  \label{66}

910: \end{equation}

911: \begin{equation}

912: \lambda _{1}={\epsilon }\left[ \frac{{1}}{\xi +{\alpha }}\frac{{\partial

913: a(\xi )}}{{\partial \xi }}-\frac{{1}}{(\xi +{\alpha )}^{{2}}}{a(\xi )}\right]

914: e^{i\tau }  \label{67}

915: \end{equation}

916: where ${a(\xi )}$\ has the form

917: 

918: \begin{equation}

919: {a(\xi )}=i\left[ \frac{{1}}{1+k{\alpha }}e^{-k\xi }-1\right] ,k=\frac{{i+1}%

920: }{\sqrt{2}}  \label{68}

921: \end{equation}

922: Omitting intermediate calculations we give the expression for $I_{H}$\ in

923: the following form

924: 

925: \begin{equation}

926: I_{H}{=-e}\frac{b{{\omega }_{{L}}}}{{\omega _{\mu }}^{{2}}}\frac{q_{\infty

927: }\alpha }{\beta }{\epsilon }^{{2}}\int_{0}^{\infty }\frac{{1}}{(\xi +{\alpha

928: )}^{{4}}}v({\xi })d{\xi }  \label{69}

929: \end{equation}

930: where

931: 

932: \[

933: v({\xi })=\overline{\left( \frac{{\partial \eta }_{1}}{{\partial \tau }}{%

934: \frac{{\partial \eta }_{1}}{{\partial \xi }}}\right) }/\left( {\frac{{%

935: \partial \eta }_{0}}{{\partial \xi }}}\right) ^{2}

936: \]

937: since is proportional to $\sqrt{{\omega }}$\ ,the function $R(\alpha )$ ,

938: 

939: \begin{equation}

940: R(\alpha )=\int_{0}^{\infty }\frac{{1}}{{(\xi +\alpha )}^{{4}}}v({\xi })d{%

941: \xi }  \label{70}

942: \end{equation}

943: is actually a function of frequency of external electric field $\omega $\.

944:  Numerical calculation shows that $R(0)=-0.21$. Collecting all multipliers

945: we get the following expression for the $I_{H}$

946: 

947: \begin{equation}

948: I_{H}{=-\frac{3{H}}{2{\pi c}}R(0)\omega \kappa \Biggl(\frac{{\mu }E_{0}}{{%

949: c}_{{e}}}\Biggr)}^{{2}}  \label{71}

950: \end{equation}

951: As an example let us choose reasonable parameters

952: 

953: \[

954: {\rho }_{{\infty }}=10^{12}%

955: {\displaystyle{{1} \over {cm}^{{3}}}}%

956: ,~~~~{\kappa }=10^{3}~{\frac{cm^{2}}{sec}},

957: \]

958: \[\mu =10^{5}~\frac{cm^{2}}{%

959: V\sec },~~~~{v}=10^{7}~{\frac{cm}{sec}}.

960: \]

961: Then, $I_{H}$\ can be written as

962: 

963: \[

964: I_{H}{=10}^{{-5}}{H\omega }E_{0}^{2}

965: \]

966: where $I_{H}$ is expressed in a $%

967: %TCIMACRO{\dfrac{\mu A}{em}}%

968: %BeginExpansion

969: {\displaystyle{\mu A \over cm}}%

970: %EndExpansion

971: $ , $H$ in a $Oe$, $E_{0}$ in  a $V/cm$ and ${\omega }$\ in a $MHz$.

972: 

973: \section{Magnetophoresis of neutral particles on the surface.}

974: 

975: Consider now the neutral particle (atom, molecule or any nano-size particle)

976: located on the metal surface under joint action of normal AC-electric and

977: tangent permanent magnetic fields. As it has been shown in Section 1 the

978: driving force which causes the directed transport of particle on the surface

979: depends on a mass, charge and details of an interaction between particles

980: and the surface of metal. If negative and positive charges coupled in a

981: neutral pair as it takes place in atoms and molecules then the driving

982: forces acting on positive and negative charges are directed on opposite

983: sides, but their values, generally speaking, can be different. Thus, the

984: resulting force acting on the particle is not necessary zero leading to the

985: directed transport of the particles. As it is well-known the Lorentz forces

986: acting on the neutral particle in a magnetic field $H$ can be written as

987: \begin{equation}

988: {\bf F}_{l}=\frac{1}{c}\left[ \frac{{\partial }{\bf D}}{{\partial }t}\times

989: {\bf H}\right]  \label{72}

990: \end{equation}

991: where ${\bf D}$ is a dipole moment of the system of the charged particles.

992: On the other hand, the dipole moment ${\bf D}$ of the neutral particle in an

993: external electric field ${\bf E}_{0}\cos (\omega t)$ is an algebraic sum of

994: a permanent dipole moment of the particle ${\bf D}_{0}$ and the dipole

995: moment ${\bf D}_{1}(t)$ induced by an external time-periodic electric field $%

996: {\bf E}_{0}\cos (\omega t)$. Thus, we have

997: \begin{equation}

998: {\bf D}(t)={\bf D}_{0}+\alpha (\omega ){\bf E}_{0}\cos (\omega t)  \label{73}

999: \end{equation}

1000: where $\alpha (\omega )$ is a frequency depending polarisibility of the

1001: particle. As a rule, for neutral particle absorbed on a metal surface vector

1002: ${\bf D}_{0}$ is directed normally to the surface. The potential of

1003: interaction between the dipole particle and metallic surface $V(z)$,

1004: consists of two terms \cite{6}.

1005: 

1006: The Coulomb attraction potential

1007: \begin{equation}

1008: V_{id}(z)=-\frac{{\bf D}^{2}}{4z^{3}\epsilon }  \label{74}

1009: \end{equation}

1010: and a repulsive exchange potential $V_{ed}(z)$, which we again (see Section

1011: 1) choose in an exponential form

1012: \begin{equation}

1013: V_{ed}(z)=c_{d}a_{d}\exp \left( -\frac{z-z_{d}}{a_{d}}\right)  \label{75}

1014: \end{equation}

1015: 

1016: Here $a_{d}$ is a radius of exchange forces on the metallic surface and $%

1017: z_{d}$ is an equilibrium distance between the particle and the surface, $%

1018: c_{d}$ is determined from an equilibrium condition

1019: \begin{equation}

1020: V(z)=V_{id}(z)+V_{ed}(z),\qquad \left. \frac{\partial V}{\partial z}\right|

1021: _{z=z_{d}}=0  \label{76}

1022: \end{equation}

1023: 

1024: As in the case of ion-metal system the friction coefficient, $\nu $, for

1025: the dipole particle located on a metal surface depends on the distance $z$

1026: (see \cite{6}). At large distances $z$, the $\nu (z)$ is proportional to $%

1027: 1/z^{3}$.

1028: 

1029: Similar to Section 1 we choose this dependence in a form

1030: \begin{equation}

1031: \nu (z)=\nu _{0}[1-\gamma _{d}(z-z_{d})]  \label{77}

1032: \end{equation}

1033: where $\nu _{0}=\nu (z_{d})$ and $\gamma _{d}=-\left.

1034: %TCIMACRO{\dfrac{\partial \nu (z)}{\partial z}}%

1035: %BeginExpansion

1036: {\displaystyle{\partial \nu (z) \over \partial z}}%

1037: %EndExpansion

1038: \right| _{z=z_{d}}$.

1039: 

1040: At small deviation from the equilibrium position of the particle the

1041: equations of motion for its gravity center can be written as following

1042: 

1043: \begin{eqnarray}

1044: M\ddot{z} &=&-M\nu (z)\dot{z}-\omega _{d}^{2}M(z-z_{d})-Q_{d}E_{0}(t)

1045: \nonumber \\

1046: M\ddot{x} &=&-M\nu (z)\dot{x}+\frac{H}{c}\frac{\partial D}{\partial t}

1047: \label{78}

1048: \end{eqnarray}

1049: 

1050: Here $\omega _{d}$ is an eigenfrequency of particle vibration on the surface

1051: and

1052: \[

1053: Q_{d}=\frac{3D_{0}\alpha (\omega )}{z_{d}^{4}}

1054: \]

1055: 

1056: Using (73) we get

1057: \begin{equation}

1058: \frac{\partial D}{\partial t}=-\omega \alpha (\omega )E_{0}\sin (\omega t)

1059: \label{79}

1060: \end{equation}

1061: 

1062: The system of equations (78) is complete analog of that in the Section

1063: 1. Omitting the details of the calculation we give the expression for

1064: tangent velocity $\frac{dx}{dt}$ averaged over a period of oscillation of

1065: the external effective electric field

1066: \begin{equation}

1067: \overline{\dot{x}}=v(\omega )=\frac{\gamma }{4}\left[ g_{0}(\omega

1068: )g_{1}^{\ast }(\omega )+g_{1}(\omega )g_{0}^{\ast }(\omega )\right]

1069: \label{80}

1070: \end{equation}

1071: where complex functions $g_{0}(\omega )$ and $g_{1}(\omega )$ have form

1072: \begin{eqnarray}

1073: g_{0}(\omega ) &=&\frac{Q_{d}E_{0}}{M(\omega _{d}^{2}-\omega ^{2}+i\nu

1074: _{0}\omega )}  \label{81} \\

1075: g_{1}(\omega ) &=&\frac{E_{0}\alpha (\omega )Hi\omega }{Mc(i\omega +\nu

1076: _{0})}  \label{82}

1077: \end{eqnarray}

1078: Final expression of the $v(\omega )$ has a form

1079: \begin{equation}

1080: v(\omega )=\gamma \frac{3D_{0}}{z_{d}^{4}M^{2}c}\frac{\alpha ^{2}(\omega

1081: )E_{0}^{2}H\omega ^{2}}{\omega ^{2}+\nu _{0}^{2}}\frac{\omega _{d}^{2}-\nu

1082: _{0}^{2}-\omega ^{2}}{\left( \omega _{d}^{2}-\omega ^{2}\right) ^{2}+\nu

1083: _{0}^{2}\omega ^{2}}  \label{83}

1084: \end{equation}

1085: 

1086: At small $\omega $ the velocity $v(\omega )$ is decreasing as $\omega ^{2}$

1087: in complete analogy with corresponding velocity of charged particle.

1088: However, in difference with the Section 1 $v(\omega )$ changes its sign at $%

1089: \omega =\sqrt{\omega _{d}^{2}+\nu _{0}^{2}}$. The general form of $v(\omega

1090: ) $ as a function of frequency is presented on Fig.4.

1091: 

1092: If the dipole moment of the particle $D_{0}$ is proportional to its volume

1093: (as it takes place for ferroelectric medium) then $v(\omega )$ does not

1094: change much with the increase of the radius of the particle and its mass.

1095: 

1096: If the external electric field $E_{0}(t)$ is a sum of two harmonics with

1097: different frequencies $\omega _{1}$ and $\omega _{2}$ the resulting

1098: velocity is additive,

1099: \begin{equation}

1100: v(\omega )=v_{1}(\omega _{1})+v_{1}(\omega _{2})  \label{84}

1101: \end{equation}

1102: 

1103: For atomic size particles (atom or molecule) the maximal velocity could be

1104: estimated as

1105: \begin{equation}

1106: v=10^{-12}E_{0}^{2}H  \label{85}

1107: \end{equation}

1108: where $v$ is given in cm/sec, $E_{0}$ in V/cm and $H$ in Oe.

1109: 

1110: \section{NLH effect phenomenology.}

1111: 

1112: In previous Sections we have shown that the joint action of the permanent

1113: magnetic and AC-electric fields leads to the direct transport of particles

1114: located nearly surface. An important role in consideration plays gradient of

1115: the damping constant the system metal-particle. Generalizing all examples of

1116: the previous Section we can assume that the NHL effect takes place in a

1117: bulk of the material if there exists a non-zero gradient of conductivity

1118: tensor

1119: {$%\sigma _{ij}$\}. Then a covariant expression of the NLH current density $%

1120: j_{i}$ could be chosen as

1121: 

1122: \begin{equation}

1123: j_{i}=\lambda (\omega ){\frac{{\partial \sigma _{ij}}}{{\partial z_{k}}}}{%

1124: \dot{E}}_{k}{\dot{E}}_{l}\epsilon _{jlm}H_{m}  \label{86}

1125: \end{equation}

1126: where ${\dot{E}}_{k}$ is a time derivative of an external electric field and

1127: $H$ is magnetic field. In a case of isotropic medium $\sigma _{ij}=\delta

1128: _{ij}\sigma $ the density of current is

1129: \begin{equation}

1130: {\bf j}=\lambda ({\bf \nabla }\sigma {\dot{{\bf E}}})[{\dot{{\bf E}}}\times

1131: {\bf H}]  \label{87}

1132: \end{equation}

1133: Here $\lambda $ is a phenomenological coefficient depending on frequency and

1134:  material parameters.

1135: 

1136: To demonstrate the physical consequences of (88) consider the cylindrical

1137: system shown on Fig.5. The time periodic voltage is applied between inner

1138: and external surface of the cylindrical sample. The permanent magnetic

1139: fields $H$\ is directed along main axes of the cylinder. We assume also that

1140: the conductivity of the metal decreases along the radius of the sample, ${%

1141: %TCIMACRO{\dfrac{{\partial \sigma }}{{\partial r}}}%

1142: %BeginExpansion

1143: {\displaystyle{{\partial \sigma } \over {\partial r}}}%

1144: %EndExpansion

1145: <0}$.

1146: 

1147: In the chosen cylindrical geometry there exist only angular component of

1148: the current density $j_{\varphi }(r)$\ which according to (88) can be

1149: expressed as

1150: 

1151: \begin{equation}

1152: j_{\varphi }(r)=\lambda (\omega ){\frac{{\partial \sigma }}{{\partial r}}%

1153: \dot{{\bf E}}}^{{2}}{\bf H}  \label{88}

1154: \end{equation}

1155: 

1156: Now, let us assume also, that the conductivity ${\sigma }$ , $%

1157: \omega $\ and thickness of the sample are chosen such a way that the

1158: electric field $E(t)$\ penetrates over the sample.Then the additional

1159: magnetic field $H_{ad}(r)$\ induced by this current can be written as

1160: 

1161: \begin{equation}

1162: H_{i}(r)=-{\frac{4\pi }{{c}}}\int_{r_{1}}^{r}\overline{j_{\varphi }}(r

1163: ){dr

1164: = \frac{4\pi }{{c}}}(r-r_{1})\overline{j_{\varphi }}

1165: \label{89}

1166: \end{equation}

1167: where $r_{1}$\ is an radius of the sample. There exist also diamagnetic

1168: contribution to the induced magnetic field which is proportional to

1169: diamagnetic susceptibility, $\chi $\ ,of the material

1170: 

1171: \begin{equation}

1172: H_{d}=\chi H  \label{90}

1173: \end{equation}

1174: 

1175: Summing up both terms we come to the conclusion that the full induced

1176: magnetic field is positive if

1177: 

1178: \begin{equation}

1179: -\lambda {\frac{{\partial \sigma }}{{\partial r}}}\overline{{\dot{{\bf E}}}^{%

1180: {2}}}{\frac{4\pi }{{c}}}l_{d}>|\chi |  \label{91}

1181: \end{equation}

1182: 

1183: Actually, the condition (91) is a condition of stability of the system to

1184: the generation of magnetic field in the sample under action of time periodic

1185: electric field $E(t)$.The details of phenomenon will be considered in the

1186: future.

1187: 

1188: The main obstacle for observation of the NLH effect in AC electric field

1189: seems to be connected with an absorption of ultrahigh frequency

1190: electromagnetic field into material. The time-periodic components of Hall

1191: current which is linear with respect to external electric field is greater

1192: than a permanent component of the NLH current and may lead to heating up of

1193: the sample. On the other hand, at low (or ultralow) temperature the

1194: absorption and warming up could be made small enough for such observation.

1195: 

1196: %\newpage

1197: 

1198: \begin{references}

1199: \bibitem{1}  F.J\"{u}licher, A.Ajdari and J.Prost, Rev.Mod.Phys. {\bf 69}%

1200: ,1369 (2000).

1201: 

1202: \bibitem{2}  K.N.Alekseev at al. Phys.Rev.Lett. {\bf 80},2669 (1998).

1203: 

1204: \bibitem{3}  M.Grifoui and P.H\"{a}nggi, Physics Reports {\bf 304}, 229

1205: (1998).

1206: 

1207: \bibitem{4}  O.Yevtushenko, S.Flach, Y.Zolotoryuk and A.A.Ovchinnikov,

1208: Europhys. Lett. {\bf 54},141 (2001). Flach S., Yevtushenko O. and Zolotoruk

1209: Y., Phys. Rev. Lett. {\bf 84},2358 (2000).

1210: 

1211: \bibitem{5}  S.Scheidl, V.M.Vinocur, cond-mat/0201008

1212: 

1213: \bibitem{6}  A.Liebsch, Elictronic Excitations at Metal Surfaces, Plenum

1214: Press, NY and London, 1997.

1215: 

1216: \bibitem{7}  M.Corvi, W.Schaich, Phys.Rev. {\bf B33}, 3688 (1986).

1217: 

1218: \bibitem{7a}  E.M.Lifshitz, L.P.Pitaevsky, Physical Kinetics,

1219: Nauka, Moscow, 1979.

1220: 

1221: \bibitem{8}  J.Bardeen, Phys.\-Rev., {\bf 71}, 717 (1947); W.Schott\-ky,

1222: Z.Physik, {\bf 118}, 570 (1940).

1223: 

1224: \bibitem{9}  W.M\"{o}nch, Semoconductor Surfaces and Interfaces, 2nd

1225: Edition, Springer-Verlag, Berlin-Heidelberg, NY, 1995.

1226: \end{references}

1227: 

1228: \bigskip

1229: 

1230: \newpage

1231: {\bf Captions to Figures} \vspace{0.15in}

1232: 

1233: Fig.~1. Frequency dependence of the velocity (20) of a single charged

1234: particle at $\nu_0=0.1\o_e$ for three different masses $m=M/M_0=0.25$,

1235: $M/M_0=1.0$ and $M/M_0=4.0$.

1236: 

1237: Fig.~2.  An attractor of the system (13-17). The A and A' are points of

1238: return in x direction. Arrows on the trajectory show the direction of

1239: the particle rotation, $\psi$ is an angle of inclination of the ellipse.

1240: 

1241: Fig.~3. Frequency dependence of $\g_1(\o )$ and $\g_2(\o )$  at

1242: $\o_{\mu}=0.1\o_s$.

1243: 

1244: Fig.~4. Frequency dependence of the velocity (83) of neutral particle

1245: at $\nu_0=0.1\o_d$.

1246: 

1247: Fig.~5. The principle set up for observation of NHL effect. Electric field

1248: $E_0(t)$ is directed along the radius of the cylinder sample.

1249: 

1250: 

1251: \end{document}

1252: