1: \documentstyle[twocolumn,prb,aps,epsf]{revtex}

2: 

3: \begin{document}

4: \draft

5: 

6: \wideabs{

7: 

8: \title{Combined effect of Zeeman splitting and spin-orbit interaction on the Josephson current

9: in a S-2DEG-S structure}

10: \author{E. V. Bezuglyi$^{a,b}$, A. S. Rozhavsky$^{a,c}$, I. D. Vagner$^{c}$,

11: and P. Wyder$^{c}$}

12: 

13: \address{$^a$B.Verkin Institute for Low temperature Physics and

14: Engineering, 47 Lenin Avenue, 61164 Kharkov, Ukraine}

15: \address{$^b$Chalmers University of Technology, S-41296,

16: G\"oteborg, Sweden}

17: \address{$^c$Grenoble High Magnetic Field Laboratory, Max-Planck-Institut

18: f\"ur Festk\"orperforschung and Centre National de la Recherche

19: Scientifique, B.P.166, 38042 Grenoble Cedex 9, France}

20: 

21: \maketitle

22: \begin{abstract}

23: We analyze new spin effects in current-carrying state of

24: superconductor-2D electron gas-su\-perconductor (S-2DEG-S) device

25: with spin-polarized nuclei in 2DEG region. The hyperfine

26: interaction of 2D electrons with nuclear spins, described by the

27: effective magnetic field $\bbox{B}$, produces Zeeman splitting of

28: Andreev levels without orbital effects, that leads to the

29: interference pattern of supercurrent oscillations over $B$. The

30: spin-orbit effects in 2DEG cause strongly anisotropic dependence

31: of the Josephson current on the direction of $\bbox{B}$, which may

32: be used as a probe for the spin-orbit interaction intensity. Under

33: certain conditions, the system reveals the properties of

34: $\pi$-junction.

35: \end{abstract}

36: 

37: \pacs{PACS numbers: 74.80.Fp, 31.30.Gs, 71.70.Ej, 73.20.Dx.}

38: }

39: 

40: The spin-orbit (SO) and hyperfine (HF) interactions in GaAs

41: heterojunctions and similar 2D quantum Hall systems attract

42: permanent theoretical and experimental attention. The hyperfine

43: field of the nuclear spin subsystem acting upon the spins of

44: charge carriers may reach $10^4$ G.\cite{Wald} At low

45: temperatures, the nuclear spin relaxation time can be

46: macroscopically long,\cite{Dyak} so the nonequilibrium spin

47: population in heterojunctions, once created, is conserved during

48: hundreds of seconds.\cite{Berg} The Zeeman splitting combined with

49: a strong spin-orbit coupling in GaAs/AlGaAs 2DEG gives rise to a

50: novel class of coherent phenomena, e.g., the spontaneous

51: Aharonov-Bohm effect.\cite{Vagner}

52: 

53: \begin{figure}

54: \epsfxsize=8.5cm\epsffile{figure1.eps} \vspace{3mm} \caption{A

55: model of the superconductor - 2DEG - superconductor device based

56: on GaAs/AlGaAs heterojunction.\cite{Taka} The normal $\bbox{n}$ is

57: directed towards the Al-doped layer.}

58: \end{figure}

59: 

60: In this paper we discuss mesoscopic spin-orbit effects in

61: Josephson current flowing across the S-2DEG-S structure (Fig.\ 1)

62: with polarized nuclei in 2DEG region. The transfer of the

63: Josephson current through the normal conducting layer is provided

64: by the Andreev reflection of electrons and holes at the

65: NS-interfaces, which convert normal electron excitations into

66: Cooper pairs in the superconducting banks. In a pure system with

67: length $d$ smaller than the electron scattering length, the

68: interference between coherent electron states and retro-reflected

69: hole states produces the set of spin-degenerate Andreev energy

70: levels $E_\lambda(\Phi)$, which depend on the quantum numbers

71: $\lambda$ and on the difference $\Phi$ of the order parameter

72: phases in superconducting electrodes.\cite{Kulik} In short

73: structures, $d \ll \xi_0$ ($\xi_0$ is the coherence length in the

74: superconductor), the Josephson current can be presented as the sum

75: of currents transferred by individual Andreev bound states (see,

76: e.g., Ref.\ \onlinecite{Shum} and references therein),

77: %

78: \begin{equation}

79: I(\Phi)=-{2e\over\hbar}\sum\nolimits_\lambda {\partial E_\lambda(\Phi)\over

80: \partial\Phi}\tanh{E_\lambda(\Phi)\over\ 2T},

81: \end{equation}

82: %

83: and must be sensitive to the HF and SO interaction which eliminate

84: spin degeneracy of the Andreev levels.

85: 

86: The contact HF interaction in a semiconductor is described by the

87: Hamiltonian\cite{Schlichter}

88: %

89: \begin{equation}

90: \hat{H}_{hf}=(8\pi/3)\mu_B\gamma_h\sum\nolimits_i\bbox{I}_i

91: \bbox{\sigma}\delta (\bbox{r}-\bbox{R}_i).

92: \end{equation}

93: %

94: Here $\mu_B$ is the Bohr magneton, $\gamma_h$ is the nuclear

95: magneton, and $\bbox{I}_i$, $\bbox{\sigma}$, $\bbox{R}_i$,

96: $\bbox{r}$ are nuclear and charge carrier spins and positions,

97: respectively. It follows from Eq.\ (2) that if the nuclear spins

98: are polarized, $\left\langle\sum\nolimits_i\bbox{I}_i

99: \right\rangle \neq 0$, the charge carrier spins feel the effective

100: HF field $\bbox{B}$ which may cause spin splitting in 2DEG of the

101: order of one tenth of the Fermi energy $E_F$.\cite{Wald,Berg} The

102: influence of the Zeeman splitting solely on a supercurrent was

103: studied first in Ref.\ \onlinecite{BB} for the SFS junction (F

104: denotes ferromagnetic metal). It was shown that the spin splitting

105: suppresses the critical current and produces its oscillations over

106: the intrinsic magnetic field localized within the F-layer.

107: 

108: The SO interaction of a charge carrier with the interface

109: potential in GaAs/AlGaAs heterojunctions is modelled by the

110: Bychkov-Rashba term,\cite{Bychkov,Pfeiffer}

111: %

112: \begin{equation}

113: \hat{H}_{\text{so}}=(\alpha/\hbar)\left[ \bbox{\sigma}

114: \bbox{p}\right] \bbox{n},  \label{Hso}

115: \end{equation}

116: %

117: where $\alpha =0.6\times 10^{-9}$ eV cm for

118: holes,\cite{Bychkov,Stormer} and $\alpha=0.25\times 10^{-9}$ eV cm

119: for electrons,\cite{Bychkov,Stein} $\bbox{\sigma}_i$, $\bbox{p}_i$

120: are the charge carrier spin and momentum, $\bbox{n}$ is the normal

121: to the interface directed towards the Al-doped layer.

122: 

123: The configuration proposed in Fig.\ 1 has the following

124: characteristic features:

125: 

126: i) the nuclear field $\bbox{B}$ is localized outside the

127: superconductors and does not influence the pairing mechanism;

128: 

129: ii) it affects only the electron spins and does not modify the

130: space motion of charge carriers, whereas usual magnetic field

131: causes strong orbital effects and transforms Andreev levels into

132: Landau bands;\cite{GB}

133: 

134: iii) since the SO interaction in Eq.\ (3) is spatially uniform

135: along the 2DEG plane, it does not suppress the supercurrent, in

136: contrast to the SO scattering at nonmagnetic impurities which acts

137: as a depairing factor.\cite{AG} The situation changes in presence

138: of the nuclear field.

139: 

140: To calculate the Josephson current in Eq.\ (1), it is enough to

141: find discrete eigenvalues of the BCS Hamiltonian supplied by the

142: HF and SO interaction terms,

143: %

144: \begin{equation}

145: {\cal H} =\!\!\int\!\! dV\!

146: \left[\psi^+\!\!\left(E(\hat{\bbox{p}})\! +\!

147: \bbox{\sigma}\hat{\bbox{\Lambda}}(\bbox{r})\right) \! \psi +

148: \Delta(\bbox{r})\psi_\uparrow^+\psi_\downarrow^+ \!+ \!h.c.\right]

149: \end{equation}

150: %

151: where $\bbox{\sigma}$ is the vector of Pauli matrices, the spinor

152: $\psi$ is composed from the annihilation operators

153: $\psi_\uparrow$, $\psi_\downarrow$ of the electron with a given

154: spin, and $E(\bbox{p}) = \bbox{p}^2/2m - E_F$ is the energy of a

155: normal electron excitation. For simplicity, we accept a step-wise

156: model for the order parameter $\Delta(\bbox{r})$,

157: %

158: \begin{equation}

159: \Delta(x) = \left\{\begin{array}{ccc}

160: \Delta\exp\left(i\Phi\,\mbox{sign}\,x/2\right), & |x|>d/2,\\ 0, &

161: |x|<d/2.

162: \end{array}\right.

163: \end{equation}

164: %

165: 

166: The vector $\hat{\bbox{\Lambda}}(\bbox{r})$ in Eq.\ (4) describes

167: the HF and SO interactions in accordance with Eqs.\ (2) and (3),

168: %

169: \begin{equation}

170: \hat{\bbox{\Lambda}}(x)\! =\! \left\{\begin{array}{ccc} 0, & |x|>d/2, \\

171: \bbox{h} + \hat{\bbox{w}}, & |x|<d/2,

172: \end{array}\right.\;

173: \bbox{h}\! = \!\mu_B\bbox{B}, \; \hat{\bbox{w}}\! \!=

174: {\alpha\over\hbar} [\hat{\bbox{p}} \bbox{n}].

175: \end{equation}

176: %

177: 

178: In order to consider uniformly the spin and electron-hole states

179: of quasiparticles, it is convenient to express the Hamiltonian in

180: Eq.\ (4) in terms of the 4-spinor field,

181: %

182: \begin{equation}

183: \varphi = \left( \begin{array}{ccc} \psi_\uparrow \\ \psi_\downarrow \\

184: \psi_\downarrow^+ \\ \psi_\uparrow^+

185: \end{array} \right), \qquad

186: \varphi^+ = \left(\psi_\uparrow^+, \; \psi_\downarrow^+, \; \psi_\downarrow,

187: \; \psi_\uparrow\right),

188: \end{equation}

189: %

190: \begin{equation}

191: {\cal H} = {1\over 2}\int dV

192: \varphi^+(\bbox{r})\hat{H}\varphi(\bbox{r}),

193: \end{equation}

194: %

195: \begin{equation}

196: \hat{H} = \tau_z E(\hat{\bbox{p}}) +\! \left\{

197: \begin{array}{lll}

198: \sigma_z[\Delta(x)\tau_+\! +\! \Delta^\ast(x)\tau_-],

199: & |x| > d/2,\\

200: \hat{\bbox{w}} \bbox{\sigma} +h_z\sigma_z +\tau_z

201: \bbox{h}_\parallel\bbox{\sigma}, & |x| < d/2.

202: \end{array}\right.

203: \end{equation}

204: %

205: Here $\bbox{h}_\parallel$ is the component of Zeeman field in 2DEG

206: plane; the Pauli matrices $\sigma$ act upon the spin states, while

207: $\tau$-matrices operate in electron-hole space, e.g.,

208: %

209: \begin{equation}

210: \sigma_z\tau_z = \left(\begin{array}{ccc}

211: \sigma_z & 0 \\ 0 & -\sigma_z

212: \end{array}\right).

213: \end{equation}

214: %

215: 

216: The problem of single-particle spectrum of the Hamiltonian

217: $\cal{H}$ in Eq.\ (8) is equivalent to the solution of the

218: Bogolyubov-de Gennes (BdG) equation for the 4-component wave

219: function $\Psi_\lambda(\bbox{r})$

220: %

221: \begin{equation}

222: \hat{H}\Psi_\lambda(\bbox{r}) = E_\lambda\Psi_\lambda(\bbox{r}).

223: \end{equation}

224: %

225: 

226: Assuming all characteristic energies $\Delta$, $h$, $w$ to be much

227: smaller than $E_F$, we use the quasiclassical representation of

228: $\Psi_\lambda(\bbox{r})$ as the product of rapidly oscillating

229: exponent over the slowly varying envelope $u(x)$:

230: %

231: \begin{equation}

232: \Psi_\lambda(\bbox{r}) = \exp(ispx + ip_y y) u_s(x),\;

233: p=\sqrt{p_F^2-p_y^2}

234: \end{equation}

235: %

236: where $s=\pm 1$ indicates two signs of $x$-component of the

237: electron momentum. The spinor function $u(x)$ obeys Eq.\ (11) with

238: a reduced Hamiltonian, in which the quasiclassical approximation

239: for the kinetic energy and SO operators is used,

240: %

241: \begin{equation}

242: E(\hat{\bbox{p}})\approx sv\hat{p}_x,\; \hat{\bbox{w}} \approx

243: {\alpha\over\hbar} [\bbox{pn}], \; \bbox{p} =(sp,p_y,0), \; v =

244: {p\over m}.

245: \end{equation}

246: %

247: 

248: The matching of the solutions of Eq.\ (11) at the NS interfaces,

249: which are assumed to be completely transparent, yields a

250: dispersion relation,

251: %

252: \begin{equation}

253: {E d\over\hbar v}\! =\! \pi n +\arccos{E\over\Delta}+ s{\Phi\over 2}

254: + \sigma\gamma,

255: \;\; n = 0,\pm 1, \pm 2...,

256: \end{equation}

257: %

258: \begin{equation}

259: \gamma(\bbox{h},\bbox{w})\! =\!

260: \arcsin\!\!\left[\!\sum_{k=\pm1}\!\!{1\!+\!k\bbox{n}_+ \bbox{n}_-

261: \over 2} \sin^2\! {A_+\! +\! kA_-\over 2}\!\right]^{1/2}

262: \end{equation}

263: %

264: \begin{equation}

265: A_\pm = (d/\hbar v)|\bbox{h} \pm \bbox{w}|, \quad \bbox{n}_\pm =

266: (\bbox{h} \pm \bbox{w}) / |\bbox{h} \pm \bbox{w}|

267: \end{equation}

268: %

269: which has the standard structure of the equation for Andreev

270: levels\cite{foot1} in current-carrying SNS junction.\cite{Kulik}

271: An additional term $\sigma\gamma$, where $\sigma = \pm 1$

272: indicates the spin direction, describes Zeeman splitting of the

273: Andreev levels renormalized by the SO interaction. In terms of the

274: BdG wave mechanics, the spin effects change phase relations

275: between the wave functions of the incident and retro-reflected

276: quasiparticles. This produces oscillations of the Andreev levels

277: with the change of the interaction parameters $h$, $w$ which enter

278: the oscillating phase shift $\sigma\gamma$ in Eq.\ (14). As a

279: result, the Josephson current in Eq.\ (1) also reveals

280: oscillations with $h$, $w$ as the manifestation of complicated

281: interference between partial supercurrents carried by individual

282: Andreev bands. In this sense, the effect represents a spin-induced

283: analogue of the Fraunhofer pattern\cite{Bastian} in a planar

284: Josephson junction in external magnetic field. Note that in the

285: absence of Zeeman field ($h=0$), all spin effects disappear:

286: $\gamma(0,\bbox{w}) = 0$.

287: 

288: The most striking manifestation of the SO interaction itself is

289: the anisotropy of Andreev levels with respect to the direction of

290: the Zeeman field $\bbox{h}$; in the absence of the SO interaction

291: ($\bbox{w} = 0$), $E_n(\Phi)$ depends only on the modulus of

292: $\bbox{h}$:

293: %

294: \begin{equation}

295: \gamma(\bbox{h},0) = \arcsin\left[\sin\left( hd/\hbar

296: v\right)\right].

297: \end{equation}

298: %

299: It is helpful to consider this anisotropy for a single electron

300: state with fixed direction of the SO vector $\bbox{w}$. As follows

301: from Eqs.\ (14)-(16), the energy of Andreev level depends only on

302: the angle between $\bbox{h}$ and $\bbox{w}$ and their moduli,

303: being insensitive to the rotation of $\bbox{h}$ around $\bbox{w}$.

304: At $\bbox{h} \parallel \bbox{w}$, the effects of SO interaction

305: vanish, as in Eq.\ (17). These conclusions can be extended to the

306: angle dependence of the Josephson current in Eq.\ (1) in a narrow

307: 2DEG channel which holds a single electron mode ($p = p_F$, $p_y =

308: 0$). In this extreme case, the vectors $\bbox{w}$ of all electrons

309: are directed along the $y$-axis and create a fixed reference frame

310: for the Zeeman vector $\bbox{h}$.

311: 

312: At arbitrary length of the 2DEG region, Eq.\ (14) can be solved

313: only numerically. Below we consider an analytically solvable case

314: of 2DEG channel much shorter than the coherence length $\xi_0$,

315: when the left-hand side of Eq.\ (14) is negligibly small,

316: %

317: \begin{equation}

318: E_\lambda(\Phi) = s\Delta\cos(\Phi/2 + \sigma\gamma), \qquad

319: (s,\sigma)=\pm 1.

320: \end{equation}

321: %

322: 

323: At $\gamma = 0$, Eq.\ (18) describes two spin-degenerated Andreev

324: levels in a superconducting constriction\cite{Furusaki} which

325: traverse across the whole energy gap with the change of $\Phi$ and

326: intersect each other at $\Phi = \pi$.\cite{foot2} The spin effects

327: split each level into two spin-dependent terms and, in addition,

328: expand them into four energy bands which transfer the Josephson

329: current

330: %

331: \begin{eqnarray}

332: \nonumber &\displaystyle I(\Phi) = {e\Delta\over\hbar}\!

333: \int^{p_F}_{-p_F}\!{dp_y\over 2\pi\hbar}\! \sum_{\sigma=\pm 1}

334: \tanh\!\left[{\Delta\over 2T}\cos\!\left({\Phi\over 2}+

335: \sigma\gamma \right)\!\right]

336: %

337: \\  &\displaystyle \times\sin\left(\Phi/ 2+\sigma\gamma\right).

338: \end{eqnarray}

339: %

340: At $h=0$, Eq.\ (19) is reduced to a 2D analogue of the

341: current-phase relationship\cite{KO} for pure constriction.

342: 

343: The set of curves $I(\Phi)$ calculated numerically at $T=0$ for

344: various directions and magnitudes of Zeeman field combined with SO

345: interaction is presented in Fig.\ 2b-d in dimensionless variables

346: %

347: \begin{equation}

348: \bbox{H} = (\hbar v_F/d)\bbox{h}, \qquad  \bbox{W} = (\hbar

349: v_F/d)\bbox{w},

350: \end{equation}

351: %

352: in comparison with those plotted for $W = 0$ in Fig.\ 2a. The

353: common features of these dependencies are drastic variations of

354: the shape of $I(\Phi)$ at $H\sim 1$, and the rapid change of sign

355: of the derivative $dI/d\Phi$ at $\Phi=\pi$ in small field $H$. It

356: is interesting that the curves at $W = 0$ and at $W = 1$,

357: $\bbox{H} \parallel \bbox{y}$ (Fig.\ 2a,b) are similar each to

358: other, as well as the curves for $W = 1$, $\bbox{H} \parallel

359: \bbox{x}$ and $W = 1$, $\bbox{H} \parallel \bbox{z}$ (Fig.\ 2c,d).

360: This reflects the results of our analysis of the anisotropy of

361: Andreev levels in 1D case which appears to be qualitatively

362: applicable for 2D system: the SO effects are relatively small at

363: $\bbox{H} \parallel \bbox{y}$ and approximately isotropic under

364: rotation of Zeeman field around the $y$-axis.

365: 

366: \begin{figure}

367: \epsfxsize=8.5cm\epsffile{figure2.eps} \vspace{3mm}

368: \caption{Current-phase relationships $I(\Phi)$ at $T=0$ normalized

369: on the critical current at $H=0$ for various directions and

370: magnitudes of the Zeeman field and intensities of the SO

371: interaction: $H=0$ (1), 0.4 (2), 0.8 (3), 1.2 (4), 1.6 (5). }

372: \end{figure}

373: 

374: \begin{figure}

375: \epsfxsize=8.5cm\epsffile{figure3.eps} \vspace{3mm} \caption{ The

376: set of dependencies of the critical current

377: $I_c(\bbox{H},T)/I_c(0,0)$ on the dimensionless magnetic field

378: $H$, at $W=1$, $\bbox{H} \parallel \bbox{x}$, $\bbox{y}$,

379: $\bbox{z}$ (solid curves) and $W=0$, $\bbox{H}$ directed

380: arbitrarily (dashed curves). Upper pairs of curves were calculated

381: for $T=0$, lower pairs -- for $T=0.9T_c$.}

382: \end{figure}

383: 

384: In Fig.\ 3 we present oscillations of the critical current

385: $I_c(H)$ depending on the HF field direction and SO interaction

386: intensity, for two values of temperature. At the vicinity of

387: $T_c$, where $I(\Phi)$ has a single harmonic,

388: %

389: \begin{equation}

390: I(\Phi)\approx I_m\sin\Phi, \;\; I_m = {\pi e\Delta^2\over

391: (2\pi\hbar)^2T_c} \int_{-p_F}^{p_F} dp_y \cos 2\gamma,

392: \end{equation}

393: %

394: the critical current $I_c(W,H)=|I_m|$ turns to zero periodically

395: with $H$ like in SFS system.\cite{BB}

396: 

397: The positive sign of $dI/d\Phi$ at $\Phi=\pi$, which occurs at

398: $H\neq 0$ (Fig.\ 2), means that this state can be stable and may

399: produce persistent current in the ground state of a

400: superconducting loop with high enough inductance

401: ($\pi$-junction\cite{pi}). On the other hand, the negative sign of

402: $dI/d\Phi$ at $\Phi = 0$, which occurs within the certain field

403: range at $W = 0$ or at $W \neq 0$, $\bbox{H} \parallel \bbox{y}$,

404: signifies instability of usual ground state with $\Phi = 0$ (note

405: that the SO interaction stabilizes this state at $\bbox{H} \perp

406: \bbox{y}$, at least at $T=0$). The results of a numerical analysis

407: of stability of states $\Phi=0$, $\pi$ within the whole

408: temperature range are shown in Fig.\ 4.

409: 

410: \begin{figure}

411: \epsfxsize=8.5cm\epsffile{figure4.eps}\vspace{2mm} \caption{

412: $H-T/T_c$ diagrams of stability of S-2DEG-S phase states with

413: $\Phi = 0$ and $\Phi=\pi$  for various directions and magnitudes

414: of the Zeeman field and intensities of the SO interaction. Within

415: the gray regions, only usual 0-state is stable: $dI/d\Phi(0) > 0$,

416: $dI/d\Phi(\pi) < 0$; blank regions correspond to the stability of

417: the $\pi$-state only: $dI/d\Phi(0) < 0$, $dI/d\Phi(\pi) > 0$;

418: within the dark regions both derivatives are positive. }

419: \end{figure}

420: 

421: In summary, we have shown that the Josephson current in a

422: mesoscopic S-2DEG-S structure is highly sensitive to the combined

423: action of the Zeeman field and spin-orbit interactions. In

424: particular, the critical current reveals oscillations and

425: anisotropy with respect to the Zeeman field $\bbox{B}$, and the

426: regions of stability at $\Phi =\pi$ (like in $\pi$-junctions)

427: emerge. We assumed hyperfine interaction of electrons with

428: polarized nuclei as the source of electron spin polarization,

429: though similar effects should be observed in external magnetic

430: field lying in the 2DEG-plane (to avoid orbital effects). In order

431: to access the regime of strong interaction ($H\sim 1$, $W \sim 1$)

432: in short 2DEG bridge ($d < \xi_0$) considered here, the

433: interaction energies $h$, $w$ of 2DEG should exceed $\Delta$.

434: Since the HF and SO interaction magnitudes in GaAs/AlGaAs

435: heterojunctions reach at most 1K in temperature scale, the banks

436: of short S-2DEG-S structure should be preferably fabricated from a

437: superconductor with low $T_c\leq 1$K. This restriction can be

438: significantly softened in long ($d \gg \xi_0$) S-2DEG-S junctions

439: where the interaction energies should be comparable with the small

440: distance between Andreev levels: $h,w \sim \hbar v_F/d \ll

441: \Delta$.

442: 

443: \begin{references}

444: 

445: \bibitem{Wald}  K.\ Wald, L.\ P.\ Kouwenhoven, P.\ L.\ McEuen, N.\ C.\ Van der Vaart,

446: and C.\ T.\ Foxon, Phys.\ Rev.\ Lett. {\bf 7}, 1011 (1994).

447: 

448: \bibitem{Dyak}  M.\ I.\ Dyakonov and V.\ I.\ Perel, in {\it Modern Problems in

449: Condensed Matter Sciences}, edited by F.\ Meyer and B.P.\

450: Zakharchenya (North Holland, Amsterdam, 1984) v. 8.

451: 

452: \bibitem{Berg}  A.\ Berg, M.\ Dobers, R.\ R.\ Gerhardts, and K.\ von Klitzing,

453: Phys.\ Rev.\ Lett. {\bf 64}, 2563 (1990).

454: 

455: \bibitem{Vagner}  I.\ D.\ Vagner, A.\ S.\ Rozhavsky, P.\ Wyder, and A.\ Yu.\ Zyuzin,

456: Phys.\ Rev.\ Lett. {\bf 80}, 2417 (1998).

457: 

458: \bibitem{Taka} H.\ Takayanagi,\ T. Akazaki, and J.\ Nitta, Phys.\ Rev.\ Lett. {\bf 75}, 3533

459: (1995).

460: 

461: \bibitem{Kulik} I.\ O. Kulik,\ ZhETF {\bf 57}, 1745 (1969).

462: 

463: \bibitem{Shum} V.\ S.\ Shumeiko, E.\ N.\ Bratus', and G.\ Wendin, Low Temp.\ Phys.

464: {\bf 23}, 181 (1997).

465: 

466: \bibitem{Schlichter} C.\ P.\ Schlichter, in {\it Principles of Magnetic

467: Resonance}, Springer-Verlag, Berlin, 1991, 2-nd edition.

468: 

469: \bibitem{BB} A.\ I.\ Buzdin, L.\ N.\ Bulayevskii, and S.\ V.\ Panyukov, Pis'\-ma ZhETF

470: {\bf 35}, 147 (1982).

471: 

472: 

473: \bibitem{Bychkov}  Yu.\ A.\ Bychkov and E.\ I.\ Rashba, J.\ Phys.\ C {\bf 17}, 609

474: (1984).

475: 

476: \bibitem{Pfeiffer}  P.\ Pfeiffer and W.\ Zawadski, Phys.\ Rev. {\bf B52}, R14332

477: (1995).

478: 

479: \bibitem{Stormer}  H.\ L.\ Stormer, Z.\ Schlesinger, A.\ Chang, D.\ C.\ Tsui,

480: A.\ C.\ Gossard, and W.\ Wiegmann, Phys.\ Rev.\ Lett. {\bf 51},

481: 126 (1983).

482: 

483: \bibitem{Stein}  D.\ Stein, K.\ von Klitzing, and G.\ Wiegmann, Phys. Rev.

484: Lett. {\bf 51}, 130 (1983).

485: 

486: \bibitem{GB} V.\ P.\ Galaiko, and E.\ V.\ Bezuglyi, Sov.\ Phys.\ JETP {\bf 60},

487: 1471 (1971).

488: 

489: \bibitem{AG} A.\ A.\ Abrikosov, and L.\ P.\ Gor'kov, Sov.\ Phys.\ JETP {\bf 15}, 752

490: (1962).

491: 

492: \bibitem{foot1} Since the energy of Andreev ``levels'' in Eq.\ (14) depends on

493: the continuous electron quantum number $v(p_y)$, they actually

494: represent overlapping energy bands.

495: 

496: \bibitem{Bastian} In S-2DEG-S InAs/AlSb structures, the Fraunhofer pattern was

497: observed by G.\ Bastian et al, Phys.\ Rev.\ Lett. {\bf 81}, 1686 (1998).

498: 

499: \bibitem{Furusaki} A.\ Furusaki and M.\ Tsukada, Physica B {\bf 165}, 967

500: (1990).

501: 

502: \bibitem{foot2} Within a more realistic model of semi-transparent NS boundary

503: between different metals, the processes of normal reflection eliminate the

504: level crossing and produce a finite gap between upper and lower level.

505: 

506: \bibitem{KO} I.\ O.\ Kulik and A.\ N.\ Omelyanchouk, Sov.\ J.\ Low Temp.\ Phys.

507: {\bf 3}, 459 (1977).

508: 

509: \bibitem{pi} L.\ N.\ Bulayevskii, V.\ V.\ Kuzii, and A.\ A.\ Sobyanin, Pis'ma ZhETF

510: {\bf 25}, 314 (1977).

511: 

512: \end{references}

513: 

514: \end{document}

515: