1: % Title: Non-equilibrium supercurrent through
2: % mesoscopic ferromagnetic weak links
3: %
4: % Authors: Tero T. Heikkil\"a, Frank K. Wilhelm, and
5: % Gerd Sch\"on
6: % File name: sfs.tex
7: % Manuscript #:
8:
9: \documentclass[debug,preprint]{epl}
10:
11: \usepackage{cite}
12: \usepackage{amsmath}
13: \usepackage{amssymb}
14:
15: \def\baselinestretch{1,0}
16:
17: \shorttitle{Supercurrent in ferromagnetic weak links}
18: \shortauthor{T. T. Heikkil\"a \etal}
19:
20: \title{Non-equilibrium supercurrent through mesoscopic ferromagnetic
21: weak links}
22: \author{T. T. Heikkil\"a\inst{1,2}\thanks{E-mail:
23: \email{Tero.T.Heikkila@hut.fi}} \and F. K. Wilhelm\inst{1,3} \and
24: G. Sch\"on\inst{1}}
25:
26: \hyphenation{Karls-ruhe}
27:
28: \institute{
29: \inst{1}Institut f\"ur Theoretische Festk\"orperphysik --
30: Universit\"at Karlsruhe, D-76128 Karlsruhe, Germany\\
31: \inst{2}Materials Physics Laboratory -- Helsinki University of
32: Technology, FIN-02015 HUT, Finland\\
33: \inst{3}Quantum Transport Group -- TU Delft, 2600 GA Delft, The
34: Netherlands}
35:
36: \pacs{74.50.+r}{Proximity effects, weak links, tunnelling phenomena,
37: and Josephson effects}
38: \pacs{71.70.Ej}{Spin-orbit coupling, Zeeman and Stark splitting,
39: Jahn-Teller effect}
40: \pacs{75.30.Et}{Exchange and superexchange interactions}
41:
42: \begin{document}
43:
44: \date{\today}
45:
46: \maketitle
47: \begin{abstract}
48: We consider a mesoscopic normal metal, where the spin
49: degeneracy is lifted by a ferromagnetic exchange field or
50: Zeeman splitting, coupled to two superconducting reservoirs.
51: As a function of the exchange field or the distance between the
52: reservoirs, the supercurrent through this device
53: oscillates with an exponentially decreasing envelope.
54: This phenomenon is similar to the tuning of a
55: supercurrent by a non-equilibrium quasiparticle
56: distribution between two voltage-biased reservoirs.
57: We propose a device combining the exchange field and
58: non-equilibrium effects, which allows us to observe a range of novel
59: phenomena. For instance, part of the field-suppressed supercurrent can
60: be recovered by a voltage between the additional probes.
61: \end{abstract}
62:
63: Externally controlled weak links in mesoscopic superconducting circuits
64: have been at the focus of interest in recent years \cite{baselmans}.
65: The possibility to control the quasiparticle distribution
66: by external voltage probes allows tuning the supercurrent through the
67: device (mesoscopic SNS transistors).
68: It has been predicted that devices with tunnel junctions
69: \cite{volkov} and systems with good metallic contacts
70: \cite{wilhelm} can enter a peculiar
71: mesoscopic non-equilibrium state at low temperatures, which
72: even allows reversing the supercurrent, turning the system into a
73: $\pi$-junction. This phenomenon has been verified experimentally
74: \cite{baselmans}.
75:
76: Another phenomenon of high interest in superconducting mesoscopics is
77: the combination of ferromagnetic (F) elements with superconductors (S)
78: \cite{jiang,demler,lazar,fazio,seviour}. A strong exchange
79: interaction $h$ in the ferromagnet is expected to suppress the
80: superconducting proximity effect, and hence also the supercurrent.
81: (Several recent experiments \cite{petrashov,lawrence,giroud} do not
82: confirm this expectation, a fact which, at this stage, is not understood.)
83: For weak fields, the supercurrent through a SFS weak link and the
84: transition temperature of a SF multi-layer are predicted to oscillate
85: \cite{buzdin,radovic,khusainov} as a
86: function of the field, or of the width $d$ of the ferromagnet.
87: The latter defines a characteristic energy scale,
88: the Thouless energy, which in the diffusive limit is $E_{\rm T}=\hbar D/d^2$,
89: proportional to the diffusion constant $D$.
90:
91: In this article, we show that the non-equilibrium-controlled
92: supercurrent in mesoscopic SNS transistors\cite{wilhelm}
93: and the supercurrent in
94: SFS weak links are formally equivalent, although one is
95: tuned by varying the distribution function, while the other is
96: controlled by modifications of equilibrium spectral functions. Combining
97: the two phenomena, we can recover by an applied voltage part of the
98: supercurrent which is suppressed by the exchange field. Thereby, one
99: can measure the exchange field in the weak link.
100:
101: For definiteness we consider a quasi-one-dimensional system depicted
102: schematically in fig.~\ref{fig:twoprobe} assuming a three-dimensional
103: system with structural changes only in one direction. The
104: magnetism in the weak link, or the Zeeman splitting, is accounted for
105: by the energy $\sigma h$ of a spatially homogeneous exchange field
106: coupling to the electron spin $\sigma = \pm 1$. In the diffusive
107: limit, the system can be described by the Usadel equation for
108: quasiclassical Green's functions \cite{demler,fazio,BWBSZ}. While the
109: equilibrium results of the present work can also be obtained in the
110: imaginary-time Matsubara formalism, we have chosen to use the
111: real-time Keldysh technique in order to include also non-equilibrium
112: processes. Then we have
113: \begin{align}
114: D\partial_x^2\theta=&-2i(E-\sigma h)\sinh \theta + 2\Delta \cosh
115: \theta+ \frac{D}{2} (\partial_x \chi)^2 \sinh 2
116: \theta\label{eq:usadel1}\\
117: j_E(E,h)&=\sinh^2 \theta \partial_x \chi\; , \qquad \partial_x
118: j_E=0 \, ,
119: \label{eq:usadel2}
120: \end{align}
121: where $\theta(E,h,x)$ and $\chi(E,h,x)$ are complex variables
122: parametrising the quasiclassical diagonal and off-diagonal Green's
123: function $G(E,h,x) = \cosh(\theta) $ and $F(E,h,x) = \sinh(\theta)
124: \exp(i\chi)$. For a system of length $d$, eq.~(\ref{eq:usadel1})
125: introduces a natural energy scale $E_T=D/d^2$. Hence, one way to
126: tune the relevant energies is by varying the length $d$.
127: Deep in the superconducting electrodes the exchange field or Zeeman
128: splitting vanishes. For simplicity, we
129: assume a bulk BCS solution up to the interfaces,
130: $\theta_{\rm S}=\arctan(\Delta/E)$, $\chi_{\rm S}=\pm
131: \phi/2$ in the superconducting electrodes with amplitudes $\Delta$ and
132: phase difference $\phi$ of the order parameters of the two
133: superconductors. Furthermore, we assume clean interfaces, and neglect the
134: reduction of Andreev reflection expected in spin-polarised systems
135: \cite{dejong,falko,jedema}. We expect the error due to these
136: approximations to be only quantitative (for the latter point, see the
137: discussions below).
138:
139: The imaginary part of the conserved `spectral supercurrent', $j_E$, in
140: eq.~(\ref{eq:usadel2}) enters into the observable supercurrent as
141: \begin{equation}
142: j_{\rm S}(\phi)=\frac{d}{4}\sum_{\sigma=\pm 1} g_{{\rm N}\sigma}
143: \int_{-\infty}^\infty \upd E (1-2f(E)) \text{Im}\{j_E(E,\sigma h)\}.
144: \label{eq:observablesc}
145: \end{equation}
146: Here $f(E)$ is the distribution function of quasiparticles in the
147: weak link, which in the absence of applied voltages reduces to
148: the equilibrium Fermi distribution
149: $f^{\text{eq}}$. Furthermore, $g_{{\rm N}\sigma}=2e^2 N_{0\sigma}
150: D_\sigma$ is the normal-state conductivity for spin $\sigma$, and
151: $N_{0\sigma}$ is the corresponding normal-state density of states
152: at the Fermi level. Our approach (eqs.~(\ref{eq:usadel1}),
153: (\ref{eq:usadel2})) assumes spin-independent densities of states and
154: diffusion constants. It is valid at low fields $h$,
155: when the variation in the densities of states is small,
156: $N_{0\uparrow}-N_{0\downarrow} \ll
157: (N_{0\uparrow}+N_{0\downarrow})/2$. In this case we may put
158: $g_{{\rm N}\uparrow}=g_{{\rm N}\downarrow} \equiv g_{\rm N}$. The
159: distribution function $f$ in general is obtained from kinetic
160: equations \cite{BWBSZ}, but for the moment, we assume thermal equilibrium.
161:
162: \begin{figure}
163: \includegraphics{twoprobe.eps}
164: \includegraphics[clip]{ssc.eps}
165: \caption{Schematic picture of the studied S-F-S structure.}
166: \label{fig:twoprobe}
167: \caption{Spectral supercurrent for different exchange fields $h$ at
168: $\phi=\pi/2$ as a function of energy. The exchange fields are
169: expressed in the units of the Thouless energy $E_{\rm T}$. The
170: variation in the peak heights is due to a finite magnitude of the
171: order parameter $\Delta=50 E_{\rm T}$.}
172: \label{fig:spectralsc}
173: \end{figure}
174:
175: It is instructive to see how the spectral supercurrent Im$\{j_E\}$
176: depends on energy $E$ and exchange fields $h$. It is
177: plotted in fig.~\ref{fig:spectralsc} for a phase difference
178: $\phi=\pi/2$ between the superconducting electrodes. For $h=0$, the
179: function Im$\{j_E\}$ is antisymmetric around the Fermi
180: surface. At low energies $E \lesssim E_{\rm T}$, it vanishes until some
181: phase-dependent $E_c(\phi)$. At larger energies it
182: increases sharply, and then decreases exponentially, oscillating
183: between positive and negative values. The exchange field
184: shifts the position of the symmetry point from $E=0$ to $E=\sigma
185: h$ and for a superconducting gap $\Delta$ of the order of $h$,
186: distorts the symmetry. Since $\Delta$ serves as an upper cutoff,
187: which is not shifted, the overall magnitude of the spectral
188: supercurrent decreases when $h$ becomes comparable to $\Delta$.
189:
190: \begin{figure}[t]
191: \twofigures[clip,width=.4375\linewidth]{scvshwithT.eps}{scvshwithD.eps}
192: \caption{Supercurrent $j_{\rm S}(\phi=\pi/2)$ as a function of exchange
193: field $h/E_{\rm T}$ through the structure depicted in
194: fig.~\ref{fig:twoprobe} for different temperatures $T/E_T$. The
195: superconducting order parameter $\Delta=1000 E_T$.}
196: \label{fig:IcvshwithT}
197: \caption{SFS supercurrent $j_{\rm S}(\phi=\pi/2)$ as a function of
198: exchange field $h/E_{\rm T}$ for different values $\Delta$ of the
199: superconducting order parameter at $T=0$.}
200: \label{fig:IcvshwithDelta}
201: \end{figure}
202:
203: In equilibrium we have $1-2f(E)=\tanh(E/2T)$. This term and the sum of
204: the spectral supercurrent $\sum_\sigma j_E(E,\sigma h)$ are antisymmetric around
205: $E=0$. Hence, for the discussion of the total supercurrent $j_{\rm S}$ we
206: can concentrate on the part $E > 0$. At low $T
207: \lesssim E_T$ the supercurrent $j_{\rm S}$ is given by an alternating
208: sum over the decreasing areas under the oscillating function
209: Im$\{j_E\}$ measured from the $E$-axis (see
210: fig.~\ref{fig:spectralsc}). At $h=0$, the
211: positive first term dominates the sum and yields a large supercurrent
212: $j_{\rm S}$. Increasing $h$ shifts the negative peak from $E<0$ to $E>0$,
213: hence decreasing $j_{\rm S}$, and even reversing its sign. At finite
214: temperature, the low-energy part, $E \lesssim T$, is effectively
215: cut off, hence $j_{\rm S}$ decreases in amplitude. This result is
216: illustrated in
217: Figs.~\ref{fig:IcvshwithT} and \ref{fig:IcvshwithDelta}, where
218: $j_{\rm S}(\phi=\pi/2)$ is plotted as a function of different exchange
219: fields at different temperatures and for different bulk order
220: parameters $\Delta$. Analogous results can be obtained for a constant
221: exchange field by varying the distance $d$ of the superconducting
222: reservoirs and through it the Thouless energy $E_T$.
223:
224: In the regime where $j_{\rm S}(\phi=\pi/2)$ is negative, the
225: junction forms a so-called `$\pi$-junction' \cite{buzdin}, since
226: the ground state of the system, with no supercurrent
227: flowing between the two superconductors, is reached for a phase
228: difference equal to $\pi$. The supercurrent--phase relation for
229: different exchange
230: fields is plotted in fig.~\ref{fig:scvsphase}, showing the crossover
231: from an ordinary behaviour to a $\pi$-state. A closer analysis of
232: fig.~\ref{fig:IcvshwithT} shows that the precise value of $h/E_{\rm
233: T}$ where the crossover occurs depends weakly on temperature,
234: since higher values of $T$ smoothen the oscillations
235: between positive and negative contributions to $j_{\rm S}$. At $h=0$,
236: $j_{\rm S}(\phi=\pi/2)$ is positive for any $T$. It remains positive
237: as long as the thermal energy dominates over the exchange, $T\gg h$.
238: With increasing $T$ the cross-over to a $\pi-$junction is shifted
239: towards higher fields. This dependence was probably observed in
240: ref.~\cite{veretennikov}. It is an alternative way to verify
241: the current reversal to what has been discussed in previous
242: proposals, where typically one requires many different samples with
243: varying widths~\cite{jiang} but otherwise equal parameters.
244: The crossover is illustrated in fig.~\ref{fig:crossovervsT} for a few
245: values of $h/E_T$.
246:
247:
248: \begin{figure}
249: \begin{center}
250: \twofigures[clip,width=.4375\linewidth]{scvsphase.eps}{scThcrit.eps}
251: \caption{Supercurrent $j_{\rm S}(\phi)$ as a function of phase for
252: different exchange fields $h/E_{\rm T}$ in the regime where the
253: crossover from the ordinary 0-state to the $\pi$-state occurs for the
254: first time. Here, $T=0$ and $\Delta=1000E_T$.}
255: \label{fig:scvsphase}
256: \caption{Crossover from the $\pi$-state ($j_{\rm S}(\phi=\pi/2)<0$) to
257: the 0-state ($j_{\rm S}(\phi=\pi/2)>0$) as a function of temperature $T$ for
258: a few values of the exchange field $h/E_{\rm T}$.}
259: \label{fig:crossovervsT}
260: \end{center}
261: \end{figure}
262:
263: By shifting the variable of integration $E$ in
264: eq.~(\ref{eq:observablesc}) by $\sigma h$ one finds
265: \begin{equation}
266: j_{\rm S}(\phi)=\frac{dg_{\rm N}}{2}\sum_\sigma \int_{-\infty}^\infty
267: \upd E (1-f^{\text{eq}}(E+\sigma h)) {\text{Im}}\{j_E(E)\}.
268: \label{eq:equivalence}
269: \end{equation}
270: The shifted distribution function $f=1/2\sum_\sigma f^{\rm eq}(E+\sigma h)$
271: has the same form as the two-step distribution function measured in
272: ref.~\cite{pothier}. There it appeared as the solution of a kinetic
273: equation in the centre of a diffusive metal between two normal probes with
274: voltage $eV=\pm 2h$ in the limit where the inelastic scattering length
275: is longer than the distance between the two normal reservoirs.
276: The spectral supercurrent in general still depends on the exchange field
277: via the boundary conditions. However, if the superconducting gap
278: $\Delta$ is much larger than the exchange field, $\Delta \gg h$,
279: this dependence can be neglected. In this limit, the supercurrent in
280: the presence of an exchange field is the same as for
281: a non-equilibrium distribution four-probe structure described in
282: ref.~\cite{wilhelm} (see fig.~\ref{fig:fourprobe} for a schematic picture).
283:
284: It is interesting to note that this behaviour of the diffusive-limit
285: supercurrent as functions of the exchange field and the external
286: potential is very similar to the supercurrent through a multi-probe
287: structure in the clean limit. This limit has been described by
288: Dobrosavljevi$\acute{\text{c}}$-Gruji$\acute{\text{c}}$ \etal
289: \cite{dobro} for the ferromagnetic two-probe case and by van Wees
290: \etal \cite{vanwees} including a voltage in a non-magnetic three-probe
291: setup. In this case, the supercurrent is carried by the Andreev
292: levels, whose energies are controlled by the exchange field \cite{dobro}, and
293: whose occupation can be tuned by the voltage \cite{vanwees}. With both
294: parameters, for example, the system can be driven into a $\pi$-state.
295:
296: \begin{figure}[ht]
297: \twofigures[clip,width=.4375\linewidth]{fourprobe.eps}{scvsVandhtwo.eps}
298: \caption{Four-probe setup for studying the non-equilibrium effects on
299: the supercurrent. It is assumed that $L \gg d_{\rm S}$ and that the
300: superconductors lie in the middle of the normal wire ($y=0$) so that the
301: distribution function has the two-step form between the
302: superconductors. Furthermore, we expect that the four-probe setup does
303: not notably alter the spectral supercurrent obtained from a
304: quasi-one-dimensional calculation (for a more detailed discussion, see
305: refs. \cite{wilhelm,wilhelmpro}).}
306: \label{fig:fourprobe}
307: \caption{Supercurrent $j_{\rm S}(\phi=\pi/2)$ of the four-probe structure at
308: different fields as a function of the voltage $V$ between the normal
309: probes. In the calculations for the main picture, the magnitude of the order
310: parameter was set to $\Delta=100E_{\rm T}$, and at the inset,
311: $\Delta=10E_{\rm T}$, thereby showing that even when $\Delta$ is of
312: the order of $h$ and $eV$, a local maximum is obtained at $eV=2h$.}
313: \label{fig:voltageandfield}
314: \end{figure}
315:
316: We can combine the effects of exchange field and non-equilibrium
317: distribution~\cite{wilhelm,wilhelmpro} by considering the structure in
318: fig.~\ref{fig:fourprobe}, where the magnetic material is placed between
319: superconductors and normal voltage leads.
320: Here, the distribution function is
321: \begin{equation}
322: f(E,y)=(\frac{1}{2}-\frac{y}{L})f^{\text{eq}}(E+eV/2) + (\frac{1}{2} +
323: \frac{y}{L})f^{\text{eq}}(E-eV/2),
324: \end{equation}
325: exhibiting the two-step form observed by Pothier et
326: al. \cite{pothier}. In this case, if the
327: superconducting reservoirs are located around $y=0$ and provided
328: $\Delta \gg h,eV$, the observable supercurrent can be written as a sum
329: of four terms,
330: \begin{equation}
331: \begin{split}
332: j_{\rm S}(\phi)=\frac{dg_{\rm N}}{8} \int_{-\infty}^\infty \upd E
333: (1-2f^{\text{eq}}(E)) {\text{Im}}\{&j_E(E-h-eV/2)+
334: j_E(E-h+eV/2)+\\&j_E(E+h-eV/2)+j_E(E+h+eV/2)\}.
335: \end{split}
336: \end{equation}
337: For example, if the potential is exactly twice the exchange field,
338: $eV=2h$, due to the antisymmetry of Im$\{j_E\}$, we have
339: \begin{equation}
340: j_{\rm S}(\phi)=\frac{1}{2}\left(j_{\rm S}^{\text{SFS}}(\phi,0)+
341: j_{\rm S}^{\text{SFS}}(\phi,2h)\right) \approx
342: \frac{1}{2}j_{\rm S}^{\text{SFS}}(\phi,0).
343: \label{eq:screcover}
344: \end{equation}
345: The latter approximate equality holds if $h \gg E_{\rm T}$. Here,
346: $j_{\rm S}^{\text{SFS}}(\phi,h)$ is the supercurrent through the SFS
347: structure with the exchange field $h$ in the weak link. Hence, one
348: can use the external potential to recover half of the zero-field
349: supercurrent. This is illustrated in fig.~\ref{fig:voltageandfield},
350: where the supercurrent of the four-probe structure is plotted as a
351: function of voltage with a few magnitudes of fields.
352:
353: The results summarised by eq.~(\ref{eq:screcover}) provide a way
354: to measure the exchange field and at the same time to
355: explore the applicability of the simplified model
356: for the ferromagnet used here and previously
357: \cite{buzdin,fazio,seviour}. When the voltage-dependent supercurrent
358: $j_{\rm S}(V)$ reaches a
359: maximum, $eV$ should equal $2h$. Deviations could occur as, for
360: instance, this model neglects the band-structure effects \cite{mazin}
361: important in the ferromagnets. Also, to be able to measure the actual
362: supercurrent through a typical ferromagnet with Curie temperature
363: $T_{\text{Cu}} \gg \Delta$, the ratio $h/E_T$ has to be made small by
364: fabricating very thin weak links. Moreover, our assumption of the
365: diffusive regime requires $d \gg l_{\text{el}}$, and a quantitative
366: agreement cannot be expected for thin structures. Finally, due to
367: the strong
368: electron-electron interactions in ferromagnets, producing a short
369: inelastic relaxation length, the normal probes should be fabricated rather
370: close to each other to obtain the two-step form for the distribution
371: function. For conventional ferromagnets the exchange field is large,
372: which would correspond to enormous voltages.
373: However, we expect that our model is approximately valid for
374: setups constructed from ferromagnetic alloys with $T_{\text{Cu}}$ of
375: the order of the superconducting critical temperature
376: \cite{veretennikov}, or in situations where $h$ can be related to the
377: Zeeman splitting in magnetic fields much weaker than the
378: superconducting critical field.
379:
380: In summary, we have calculated the supercurrent through a
381: ferromagnetic weak link as a function of the exchange field in the
382: ferromagnet. In the calculations, the Keldysh technique was used to
383: provide a description of non-equilibrium effects. We found that when
384: $\Delta \gg h$, the problem is formally equivalent to the four-probe
385: measurement of the supercurrent through a normal-metal weak
386: link. Furthermore, we showed that applying a non-equilibrium potential
387: in the transverse direction, one can recover half of the supercurrent
388: of a ferromagnet with an exchange field $h \gg E_{\rm T}$, as compared
389: to the supercurrent in the absence of $h$.
390:
391: \acknowledgments
392:
393: We thank B. Pannetier, M. Giroud and M. Fogelstr\"om for
394: discussions. This work was supported by the Helsinki University of
395: Technology, the DFG through SFB 195 and the EU through the EU-TMR
396: network ``Superconducting nano-circuits''. While writing this paper,
397: we became aware of a related work by Yip \cite{yip}, who introduced
398: the exchange-field term as the Zeeman splitting of the
399: current-carrying density of states.
400:
401: \begin{thebibliography}{25}
402:
403: \bibitem{baselmans} \Name{Morpurgo A., van Wees B. J. \and Klapwijk T. M.}
404: \REVIEW{Appl. Phys. Lett.}{72}{1998}{966}; \Name{Baselmans J. J. A.,
405: Morpurgo A., van Wees B. J., \and Klapwijk
406: T. M.} \REVIEW{Nature}{397}{1999}{43}.
407: \bibitem{volkov} \Name{Volkov A. F.}
408: \REVIEW{Phys. Rev. Lett.}{74}{1995}{4730}.
409: \bibitem{wilhelm} \Name{Wilhelm F. K., Sch\"on G., \and Zaikin A. D.}
410: \REVIEW{Phys. Rev. Lett.}{81}{1998}{1682}.
411: \bibitem{jiang} \Name{Jiang J. S., Davidovi$\acute{\text{c}}$ D., Reich
412: D. H. \and Chien C. L.} \REVIEW{Phys. Rev. Lett.}{74}{1995}{314}
413: \bibitem{demler} \Name{Demler E. A., Arnold G. B. \and Beasley M. R.}
414: \REVIEW{Phys. Rev. B.}{55}{1997}{15174}.
415: \bibitem{lazar} \Name{Lazar L., Westerholt K., Zabel H., Tagirov
416: L. R., Goryunov Yu. V., Garif'yanov N. N. \and Garifullin I. A.}
417: \REVIEW{Phys. Rev. B}{61}{2000}{3711}.
418: \bibitem{fazio}\Name{Fazio R. \and Lucheroni C.}
419: \REVIEW{Europhys. Lett.}{45}{1999}{707}.
420: \bibitem{seviour} \Name{Seviour R., Lambert C. J. \and Volkov A. F.}
421: \REVIEW{Phys. Rev. B}{59}{1999}{6031}.
422: \bibitem{petrashov} \Name{Petrashov V. T., Antonov V. N., Maksimov
423: S. V., Shaikhaidarov R. Sh.} \REVIEW{JETP Lett.}{59}{1994}{551};
424: \Name{Petrashov V. T., Sosnin I. A., Cox I., Parsons A. \and Troadec
425: C.} \REVIEW{Phys. Rev. Lett.}{83}{1999}{3281}.
426: \bibitem{lawrence} \Name{Lawrence M. D. \and Giordano N}
427: \REVIEW{J. Phys.: Condens. Matter}{8}{1996}{L563}.
428: \bibitem{giroud} \Name{Giroud M., Courtois H., Hasselbach K., Mailly
429: D. \and Pannetier B.} \REVIEW{Phys. Rev. B}{58}{1998}{R11872}.
430: \bibitem{buzdin} \Name{Buzdin A. I., Bulaevskii L. N. \and Panyukov
431: S. V.} \REVIEW{JETP Lett.}{35}{1982}{178}; \Name{Buzdin A. I. \and
432: Kupriyanov M. Yu.} \REVIEW{JETP
433: Lett.}{52}{1990}{487}; \SAME{53}{1991}{321}; \Name{Buzdin A. I.,
434: Bujicic, B. \and Kupriyanov M. Yu}
435: \REVIEW{Sov. Phys. JETP}{74}{1992}{124}.
436: \bibitem{radovic} \Name{Radovi$\acute{\text{c}}$ Z.,
437: Dobrosavljevi$\acute{\text{c}}$-Gruji$\acute{\text{c}}$ L., Buzdin
438: A. I. \and Clem J. R.} \REVIEW{Phys. Rev. B}{38}{1988}{2388};
439: \Name{Radovi$\acute{\text{c}}$ Z., Ledvij M.,
440: Dobrosavljevi$\acute{\text{c}}$-Gruji$\acute{\text{c}}$ L., Buzdin
441: A. I. \and Clem J. R.} \REVIEW{Phys. Rev. B}{44}{1991}{759}.
442: \bibitem{khusainov} \Name{Khusainov M. G. \and Proshin Yu. N.}
443: \REVIEW{Phys. Rev. B}{56}{1997}{R14 283}.
444: \bibitem{BWBSZ} \Name{Belzig W., Wilhelm F. K., Bruder C., Sch\"on
445: G. \and Zaikin A. D.} \REVIEW{Superlatt. and
446: Microstruct.}{25}{1999}{1251}.
447: \bibitem{dejong}\Name{de Jong, M. J. M. \and Beenakker, C. W. J.}
448: \REVIEW{Phys. Rev. Lett.}{74}{1995}{1657}.
449: \bibitem{falko}\Name{Fal'ko, V. I., Lambert, C. J. \and Volkov,
450: A. F.} \REVIEW{JETP Lett.}{69}{1999}{532};
451: \REVIEW{Phys. Rev. B}{60}{1999}{15394}.
452: \bibitem{jedema}\Name{Jedema, F. J., van Wees B. J., Hoving B. H.,
453: Filip A. T. \and Klapwijk T. M.} [cond-mat/9901323].
454: \bibitem{veretennikov} \Name{Veretennikov A. V., Ryazanov V. V.,
455: Oboznov V. A., Rusanov, A. Yu., Larkin, V. A. \and Aarts J.}
456: \REVIEW{Physica B}{284-288}{2000}{495}; \Name{Aarts J.} private communication.
457: \bibitem{pothier} \Name{Pothier H., Gu$\acute{\text e}$ron S., Birge
458: N. O., Esteve D. \and Devoret M. H.}
459: \REVIEW{Phys. Rev. Lett.}{79}{1997}{3490}.
460: \bibitem{dobro}
461: \Name{Dobrosavljevi$\acute{\text{c}}$-Gruji$\acute{\text{c}}$ L.,
462: Ziki$\acute{\text{c}}$ R. \and Radovi$\acute{\text{c}}$ Z.}
463: [cond-mat/9911339].
464: \bibitem{vanwees} \Name{van Wees B. J., Lenssen K.-M. H. \and Harmans
465: C. J. P. M.} \REVIEW{Phys. Rev. B}{44}{1991}{470}.
466: \bibitem{wilhelmpro} \Name{Wilhelm F. K., Sch\"on G. \and Zaikin
467: A. D.} \REVIEW{Physica B}{280}{2000}{418}.
468: \bibitem{mazin} \Name{Mazin I. I.}
469: \REVIEW{Phys. Rev. Lett.}{83}{1999}{1427}.
470: \bibitem{yip} \Name{Yip S. K.} [cond-mat/0002395].
471: \end{thebibliography}
472:
473: \end{document}
474:
475:
476: