1: 

2: \documentclass{elsart}

3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%

4: 

5: \begin{document}

6: 

7: \runauthor{}

8: 

9: \begin{frontmatter}

10: \title{Giant magnetoresistance in quantum magnetic contacts}

11: 

12: \author[KSU]{L.R. Tagirov\thanksref{Someone}},

13: \author[KSU,KFTI]{B.P. Vodopyanov},

14: \author[KSU]{B.M. Garipov}

15: 

16: \address[KSU]{Kazan State University, Kazan 420008,

17: Russian Federation}

18: \address[KFTI]{Kazan Physico-Technical Institute of RAS, Kazan 420029,

19: Russian Federation}

20: \thanks[Someone]{Corresponding author, E-mail address: Lenar.Tagirov@ksu.ru \\

21: Postal address: Theoretical Physics Department, Kazan State University, 

22: 420008 Kazan, Russia ; FAX: 007-8432-387201}

23: 

24: \begin{abstract}

25: We present calculations of quantized conductance and magnetoresistance 

26: in nanosize point contacts between two ferromagnetic metals. 

27: When conductance is open for only one conduction electrons spin-projection,

28: the magnitude of magnetoresistance is limited by the rate of

29: conduction electron spin-reversal processes. For the case when both spin-channels 

30: contribute to the conductance we analyze the influence of the point contact 

31: cross-section  asymmetry on the giant megnetoresistance. Recent experiments 

32: on magnetoresistance of magnetic point contacts are discussed in the 

33: framework of the developed theory.

34: 

35: PACS numbers: 74.80.Dm; 74.50.+r; 74.62.-c

36: \end{abstract}

37: \begin{keyword}

38: Magnetic point contacts; giant magnetoresistance; conductance quantization

39: \end{keyword}

40: \end{frontmatter}

41: 

42: \newpage

43: 

44: \section{Introduction}

45: 

46: Since experiments with two-dimensional electron gas in a semiconductor \cite

47: {Wees,Wharam} it is demonstrated that electric conduction is quantized, and

48: elementary conductance quantum is equal to $2e^2/h$. When measured on tiny

49: contacts of nonmagnetic semiconductors and metals the conductance

50: quantization is limited to low temperatures by thermal fluctuations, and the

51: factor 2 is attributed to the two-fold spin degeneracy of conduction

52: electron states. Recently, sharp conductance quantization steps have been

53: observed in nanosize point contacts of ferromagnetic metals at room

54: temperature \cite{Kramer,Oshima,Ott,Ono}. It was possible, because phonon

55: and magnon assisted relaxation processes are quenched due to a large, $\sim

56: 1eV$, exchange splitting of the conduction band. In addition, Oshima and

57: Miyano \cite{Oshima} found an indication of the odd-valued number $N$ of

58: open conductance channels ($\sigma =N(e^2/h)$) in nickel point contacts from

59: room temperature up to 770K. Ono \textit{et al.} \cite{Ono} presented an

60: evidence of switching from $2e^2/h$ conductance quantum to $e^2/h$ quantum

61: at room temperature in the nickel nanocontacts of another morphology.

62: Obviously, the change of conductance quantum from $2e^2/h$ to $e^2/h$ is a

63: result of lifting-off the spin degeneracy of the conduction band. Recent

64: calculations \cite{Imamura,Zvezdin} confirmed the $e^2/h$ conductance

65: quantization in ferromagnetic metals, which is due to non-synchronous

66: opening of ''up'' and ''down'' spin-channels in the point contact conduction.

67: 

68: New pulse to studies of electric transport in ferromagnets has been given by

69: observation of giant magnetoresistance (GMR) in nanosize magnetic contacts

70: by Garc\'ia \textit{et al.} \cite{Garcia1,Garcia2,Garcia3}.

71: Magnetoresistance magnitudes of 280\% for Ni-Ni \cite{Garcia1} and 200\% for

72: Co-Co \cite{Garcia2} nanosize contacts were obtained at room temperature.

73: Somewhat smaller ($\sim 30\%$), but also very large for a single interface,

74: magnetoresistance was observed in Fe-Fe point contacts \cite{Garcia3}. In

75: these experiments there is a huge spread in the measured values of

76: magnetoresistance, drawn as a function of conductance at ferromagnetic

77: alignment of magnetizations in contacting ferromagnetic domains

78: (F-conductance). The spread of MR points for Ni-Ni and Co-Co contacts is

79: extremely large at F-conductance lying in the range of 2-8 elementary

80: conductances $e^2/h$. The above mentioned observations of conductance

81: quantization steps in point junctions of ferromagnetic metals at room

82: temperature give anticipation that conduction quantization is responsible

83: for the giant magnitude and the giant fluctuations of magnetoresistance in

84: tiny magnetic contacts.

85: 

86: In this article we develop a simple model of conductance and

87: magnetoresistance for nanosize magnetic contacts in the regime of

88: conductance quantization (quantum magnetic contacts), proposed in the

89: previous work \cite{Tagirov2}. In \cite{Tagirov2} we argued, that if only

90: one conduction electron spin-channel is open at F-conductance, then the

91: magnitude of GMR is limited only by spin-flip processes of conduction

92: electrons when passing through the point contact \cite{Referee}. Then the

93: magnetic nanocontact serves as quantum spin-valve. When both spin-channels

94: of F-conduction are open we established, that GMR is a multivalued function

95: of conductance at ferromagnetic alignment of magnetizations (at least at low

96: temperatures and absence of disorder). This means that if the conductance is

97: quantized, different samples, having the same F-conductance, reveal

98: different magnetoresistance. Distribution of magnetoresistance values is not

99: normal or flat in the statistical sense. Rather, at fixed F-conductance

100: values, smaller magnitudes are much more probable than the maximal ones. The

101: width of distribution is extremely large for the first few open

102: F-conductance channels. Thus, we concluded that the giant raw-data

103: fluctuations observed in the experiments by Garc\'ia \textit{et al.} \cite

104: {Garcia1,Garcia2,Garcia3} might be the consequence of conduction

105: quantization. In the present study we focus out attention on the influence

106: of the point contact cross-section asymmetry on GMR. We find that number of

107: open conduction channels, at which conductance shows up for the

108: antiferromagnetically aligned magnetizations, depends not only on the

109: conduction band spin-polarization \cite{Tagirov2}, but also on the aspect

110: ratio of the contact cross-section. We discuss the above mentioned as well

111: as very recent experiments on GMR in magnetic nanocontacts.

112: 

113: \section{Calculation of conductance and magnetoresistance}

114: 

115: We consider a model of two ferromagnetic, single-domain half-spaces

116: contacting via a narrow and short neck with typical length from one to

117: several nanometers. For the F-alignment of domains the magnetization is

118: homogeneous along the constriction, therefore current carriers move in a

119: constant potential created by the magnetization. At antiferromagnetically

120: (AF) aligned domains a domain wall (DW) is created inside the neck. Then,

121: the carriers move in a potential landscape created by the domain wall.

122: According to the general quantum-mechanical prescription, any inhomogeneity

123: in the potential energy landscape results in a reflection of quasiparticle

124: wave function, which evokes an additional electric resistance \cite{Falicov}. 

125: For the free DW between unconstrained domains this domain-wall resistance

126: is very small because the profile of DW is smooth, and free domain wall

127: width $\delta _0$ is large, typically in the range 15-150 nm for the strong

128: elemental ferromagnets like Co, Fe and Ni \cite{Hubert}. However, if DW is

129: created in the constriction, then the wall width $\delta $ is approximately

130: equal to the length of the neck $l$, which is at least an order of magnitude

131: shorter than $\delta _0$ \cite{Bruno}. The sharpening of DW leads to huge

132: enhancement of quasiparticle reflection from DW \cite{Tagirov1}, as well as

133: some increase of impurity scattering \cite{Levy}. When the external DC

134: magnetic field aligns the domains magnetizations parallel (F-alignment), it

135: eliminates DW and domain-wall reflection, which results in essential

136: decrease of resistance, \textit{i.e.} leads to GMR \cite{Tagirov1}.

137: 

138: Now we consider the regime of quantized conductance through the nanosize

139: neck. The cross-section size of the neck is assumed very small, typically

140: about 1 nm, so that the transverse motion of electron in the neck is

141: quantized. In our previous work \cite{Tagirov2} we considered the neck of

142: cylindrical cross-section, in this paper we give solution for the neck of

143: rectangular shape. The length of the neck, $l$, is considered shorter than

144: the electron mean free path, that is why the electron transport through the

145: neck is ballistic. Actually, the neck is a conducting bridge which plays the

146: role of a quantum filter. It selects from the continuous domain of

147: quasiparticle incidence angles only those, which meet the allowed (and

148: quantized) transverse momentum in the channel, and satisfy the energy and

149: momentum conservation laws. For the particular calculations of conductance

150: we may use the ballistic-limit versions of the formulas (14), (18) and (19)

151: of our work \cite{Tagirov1}: 

152: \begin{equation}

153: \sigma ^F=\sigma _{\uparrow \uparrow }+\sigma _{\downarrow \downarrow }=

154: \frac{e^2}h\widetilde{\sum_{m,n}}\left\{ D_{\uparrow \uparrow

155: }(x_{mn})+D_{\downarrow \downarrow }(x_{mn})\right\} ,  \label{eq1}

156: \end{equation}

157: \begin{equation}

158: \sigma ^{AF}=\frac{2e^2}h\widetilde{\sum_{m,n}}D_{\uparrow \downarrow

159: }(x_{mn}).  \label{eq2}

160: \end{equation}

161: The same formulas may be also obtained within the Landauer-B\"uttiker

162: scattering formalism \cite{Landauer,Buttik}. In the above expressions 

163: $\sigma ^F(\sigma ^{AF})$ is the conductance at ferromagnetic

164: (antiferromagnetic) alignment of domains, $\sigma _{\alpha \alpha }$ is the

165: conductance for the $\alpha $-th spin-channel, and $x_{mn}=\cos \theta $ is

166: the cosine of the quasiparticle incidence angle $\theta $, measured from the

167: longitudinal symmetry axis of the neck, indices $m$ and $n$ refer to quantum

168: numbers of transverse motion in the neck. $D_{\alpha \beta }(x)$ is the

169: quantum-mechanical transmission coefficient for the connecting channel (see

170: below). The magnetoresistance is defined as follows: 

171: \begin{equation}

172: MR=\frac{R^{AF}-R^F}{R^F}=\frac{\sigma ^F-\sigma ^{AF}}{\sigma ^{AF}}.

173: \label{eq3}

174: \end{equation}

175: 

176: Quantization of transverse motion in the channel obliges the parallel to the

177: interface projection of the incident quasiparticle momentum to satisfy the

178: requirement: 

179: \begin{equation}

180: p_{\parallel }=p_{F\alpha }\sin \theta =p_{mn}\equiv \hbar \lambda _{mn},

181: \label{eq4}

182: \end{equation}

183: where $p_{F\alpha }$ is the Fermi momentum for the $\alpha $-th

184: spin-channel, $\lambda _{mn}$ is the quantized wave number (see definition

185: below). This is the first basic selection rule, which comes from

186: quantization. Tilde in (\ref{eq1}) and (\ref{eq2}) means that the summations

187: should be done over the open conduction channels satisfying the condition: 

188: \begin{equation}

189: x_{mn}\equiv \cos \theta =\sqrt{1-(\hbar \lambda _{mn}/p_{F\alpha })^2}\leq

190: 1.  \label{eq5}

191: \end{equation}

192: When the magnetizations alignment is ferromagnetic, the Fermi momenta on

193: both sides of the contact are equal in each, spin-up and spin-down, channel,

194: respectively. The energy and momentum conservations are satisfied, and the

195: transmission coefficients are equal to unity. At the antiferromagnetic

196: alignment the conservation of the parallel to the interface momentum 

197: ($p_{\Vert }\equiv p_{F1\alpha }\sin \theta _1=p_{F2\alpha }\sin \theta _2$ ,

198: where the subscript 1 or 2 labels left- or right-hand side of the contact,

199: respectively) introduces the additional selection rule into Eq. (\ref{eq4}): 

200: \begin{equation}

201: p_{F\alpha }=\min (p_{Fj\uparrow },p_{Fj\downarrow }).  \label{eq6}

202: \end{equation}

203: This selection rule is strictly valid, if the electron spin conserves upon

204: transmission through the DW. We believe, that conservation is realized in

205: the atomic-size point contacts, when the length of the connecting channel is

206: comparable with the Fermi wave-length of the current carriers. It was argued

207: that the above scenario is realized, if the DW width $\delta <\delta _s$,

208: where $\delta _s=\min (v_F/\omega _Z,v_FT_1)$, $T_1$ is the longitudinal

209: relaxation rate time of the carriers magnetization \cite{Tagirov1}, and 

210: $\omega _z$ is the Larmore precession frequency \cite{Gregg}. Imamura 

211: \textit{et al.} \cite{Imamura} justified the above hypothesis for a quantum 

212: DW by numerical calculations for the linear chain of spins.

213: 

214: We perform concrete calculations for the neck of rectangular cross-section,

215: which models a contact with asymmetric cross-section. The solution to the

216: Schr\"odinger equation for the electron moving in the neck is sought in the

217: form 

218: \begin{equation}

219: \Psi (x,y,z)=\Phi (z)\sin \frac{\pi nx}a\sin \frac{\pi my}b.  \label{eq8}

220: \end{equation}

221: The function $\Phi (z)$ describes motion along the channel, it obeys the

222: equation 

223: \begin{equation}

224: \hbar ^2\frac{\partial ^2\Phi }{\partial z^2}+\left( p_{F0}^2-\lambda

225: _{mn}^2\hbar ^2+2MU(z)\right) \Phi =0,  \label{eq9}

226: \end{equation}

227: where $U(z)=zE_{ex}/l$ is the potential landscape created in the neck by the

228: constrained domain wall, $E_{ex}$ is the conduction band exchange energy

229: splitting, $M$ is the conduction electron mass, $a$ and $b$ are the width

230: and height of the neck and $p_{F0}$ is the Fermi momentum in absence of

231: conduction band splitting. In Eqs. (\ref{eq8}) and (\ref{eq9}) $m$ and $n$

232: are positive integer quantum numbers. The discrete function $\lambda _{mn}$

233: is given by 

234: \begin{equation}

235: \lambda _{mn}=\pi \sqrt{\left( \frac na\right) ^2+\left( \frac mb\right) ^2.}

236: \label{eq10}

237: \end{equation}

238: The choice of potential energy $U(z)$ in the form of linear function of $z$

239: is based on the calculations by Bruno (Ref. \cite{Bruno}, Fig. 2). Eq. (\ref

240: {eq9}) has an exact solution in terms of Airy functions, the explicit

241: expression for the transmission coefficient $D_{\alpha \beta }$ is given in

242: Ref. \cite{Tagirov2}.

243: 

244: A numerical routine consists of the summation over the consecutive values of

245: the roots $\lambda _{mn}$ satisfying the constraints, Eqs. (\ref{eq4}) and 

246: (\ref{eq5}). At the antiferromagnetic alignment the minority Fermi momentum

247: of the either spin projection should be used instead of $p_{F\alpha }$ in

248: Eqs. (\ref{eq4}) and (\ref{eq5}) to calculate the conductance $\sigma ^{AF}$, 

249: Eq. (\ref{eq2}). The results are displayed on Figures 1 and 2, important

250: for the discussion conduction band spin-polarization parameter $\gamma $ is

251: defined as: $\gamma =p_{F\downarrow }/p_{F\uparrow }\leq 1$. Calculations

252: revealed that the results depend on the absolute value of $p_{F\uparrow }$ ,

253: we have chosen $\hbar ^{-1}p_{F\uparrow }=1$\AA $^{-1}$ for the presentation.

254: 

255: \section{Results of calculations}

256: 

257: Fig.1 displays the results of calculations for the neck of the square

258: cross-section ($b=a$) and $\gamma =0.68$. Panel (a) shows the dependence 

259: of F- and AF-conductances on the channel radius. $l$ and $\lambda

260: =lp_{F\uparrow }\hbar ^{-1}$ are the length in \AA\ and dimensionless length

261: of the connecting channel, respectively. The chosen value, $\lambda =10.0$,

262: corresponds to the connecting channel length 10\AA\ (1 nm). Panel (b) shows

263: the dependence of magnetoresistance on the channel size $a$. The panels (c)

264: and (d) display the magnetoresistance against F-conductance for the sloping

265: (c) and the step-like (d) potential landscapes in the channel, the latter

266: one is the limiting case of very sharp DW. Physically, Fig.1 demonstrates

267: the case, when the AF-alignment conduction opens in the interior part of the

268: first F-conductance plateau. It allows us to make the following conclusions:

269: 1) the F-alignment conductance is spin-dependent, the conductions of

270: spin-channels open asynchronously (panel (a)), thus resulting in $e^2/h$

271: quantization of conductance \cite{Imamura,Zvezdin}; 2) if some number of

272: conduction channels are open for the F-alignment ($\sigma ^F$ is finite),

273: but there is no conduction for the AF-alignment ($\sigma ^{AF}=0$), then,

274: according to definition Eq. (\ref{eq3}), MR diverges. Magnetoresistance is

275: infinite in the idealized model with no reversal of the carriers spin upon

276: transmitting the neck. In a more realistic treatment the magnitude of MR of

277: this quantum spin-valve is restricted by the spin-flip process, which gives

278: rise to a finite AF-conductance at any number of open F-conductance

279: channels. It is the quantum spin-valve regime; 3) the magnitude of MR beyond

280: the quantum spin-valve regime is well above 200\% for very moderate

281: polarization of the conduction band ($\gamma =0.68$, see discussion below);

282: 4) the magnetoresistance has very sharp and high peak, when the first

283: channel for AF conductance step appears (panel (b) in correlation with

284: panel(a)); 5) sudden jumps in magnetoresistance, followed by practically

285: flat plateaus, appear at the moments when new F-alignment spin-up

286: conductance channel opens. They persist until the spin-down projection opens

287: new channel (panel (b) in correlation with panel (a)); 6) panels (c) and (d)

288: show that the magnetoresistance drawn as a function of quantized F-alignment

289: conductance is a multivalued function of F-conductance, $\sigma ^F$ \cite

290: {Tagirov2}. The issue 4) leads to weakly disperse, or even non-disperse

291: behavior of magnetoresistance at certain numbers of open F-alignment

292: channels: $N^F=4,5,7,11,13,17...$ (see panels (c) and (d)). Non-disperse

293: behavior of MR comes if the AF-conductance is practically independent on the

294: contact size when a new F-conductance channel opens (see panel (a)). The

295: issue 6) means, that if the temperature and disorder effects can be

296: neglected, several values of magnetoresistance correspond to the same number

297: of open conductance channels for the F-alignment of magnetizations (abscissa

298: in the panels (c) and (d)). The overall width of distributions of MR points,

299: which belong to the same value of the quantized F-conductance, may be

300: comparable with maximal value of MR, \textit{i.e.} magnetoresistance

301: acquires giant fluctuations because of conductance quantization.

302: 

303: Next, we change the aspect ratio $\varepsilon $ of the sides of the

304: rectangular cross-section, $\varepsilon =b/a.$ Fig. 2 is drawn with 

305: $\varepsilon =1.5$. Main changes compared to Fig. 1 can be summarized as

306: follows: 1) the AF-conductance opens now at three open channels of

307: F-conductance ($\sigma ^F=3e^2/h$, panels (c) and (d)); 2) the range of the

308: neck sizes with zero AF-conductance (quantum spin-valve) becomes broader; 3)

309: the overall magnitudes of MR increase (panels (b)-(d)); 4) magnetoresistance

310: points appear at almost every number of open F-conductance channels (compare

311: panels (c) and (d) of Figs. 1 and 2). We emphasize the issue which has

312: important implication to point contact GMR experiments: number of open

313: F-conductance channels, at which the AF conductance opens ($\sigma ^F=

314: 2e^2/h$ in Fig.1 and $\sigma ^F=3e^2/h$ in Fig.2) depends not only on the

315: polarization of the conduction band \cite{Tagirov2}, but also on the

316: asymmetry of the point contact cross-section.

317: 

318: \section{Discussion of the results}

319: 

320: There are techniques, which provide the information about the

321: spin-polarization of the ferromagnet's conduction band at the Fermi energy.

322: These are the ferromagnet-insulator-superconductor tunneling spectroscopy

323: (see Ref. \cite{TedMes} and references therein) and the Andreev-reflection

324: spectroscopy \cite{Soulen98,Soulen99,Buhrman,Nadgorny,Strijk1,Strijk2}. The

325: tunneling spectroscopy suggests the following estimates for the mean values

326: of conduction band polarization parameter $\gamma $: 0.6 for permalloy 

327: (Ni$_{80}$Fe$_{20}$); 0.63 for pure Ni; 0.48 for Co and 0.43 for Fe. From the

328: Andreev-reflection spectroscopy we obtain the ranges for the values of 

329: $\gamma $: $\sim 0.68$ for permalloy; $\sim 0.62-0.72$ for Ni; $\sim 0.6-0.68$

330: for Co; $\sim 0.62-0.64$ for Fe. Observing the tunneling and

331: Andreev-reflection data on $\gamma $ and our figures we may confirm our

332: conclusion made from calculations for the cylindrical neck in Ref. \cite

333: {Tagirov2}: using realistic values of $\gamma $ we may reproduce maximal

334: values as well as giant fluctuations of MR data obtained by Garc\'ia 

335: \textit{\ et al.} \cite{Garcia1,Garcia2,Garcia3}. However, the agreement 

336: between the theory and the experiment on Fig. 3 in \cite{Tagirov2} could be 

337: even better, if some MR points would appear at neighboring number of open 

338: F-conductance channels. Comparison of Figs. 1 and 2 of the present work with 

339: Fig. 3 from Ref. \cite{Tagirov2} shows, that varying the aspect ratio in the range 

340: $\sim 1.0-2.0$ one may get a desired re-assignment of some MR points to number 

341: of conduction channels, and to improve agreement between the theory and the

342: experiment.

343: 

344: Independent on the actual shape of the neck, when its length is comparable

345: or longer than the cross-section size, the dipole-dipole anisotropy energy

346: may cause fluctuations between Bloch, N\'eel or more complicated types of

347: domain walls. Coey \textit{et al.} concluded \cite{Coey1,Coey2} that giant

348: MR of a nanocontact may be reduced somewhat by these fluctuations, but not

349: eliminated. In recent calculations Zhuravlev \textit{et al.} \cite{Zhurav1}

350: also predicted giant values and fluctuations of MR in segmented nanowires,

351: when conductance of the wire is quantizes.

352: 

353: When the cross-section of the point contact is very small, so that

354: F-conduction is open for only one spin-channel, the magnitude of GMR is

355: limited from above by the spin-reversal rate of conduction electrons upon

356: passing through the neck \cite{Tagirov2}. Our calculations show (see panels

357: (a) of Figs. 1 and 2), that higher the polarization of the conduction band

358: and larger the asymmetry of the cross-section, then wider the range of neck

359: sizes, at which the regime of quantum spin-valve can be realized. Magnetic

360: half-metal contacts with $\sim $100\% polarization of conduction band would

361: be almost always quantum spin-valves at nanometer range of size. In very

362: recent experiments \cite{Chopra,Garcia4} the ballistic magnetoresistance

363: (BMR) in the range 3000-4000\% has been observed in Ni point contact. These

364: really giant MR values can be easily reproduced in the ballistic regime of

365: quasiclassical conductance, Eq. (23) of Ref. \cite{Tagirov1}, for a moderate

366: polarization of the conduction band: with $\gamma =0.2$ ($P$(DOS$)=100\cdot

367: (1-\gamma )/(1+\gamma )=67\%$) we get $MR=3090\%$, and with $\gamma 

368: =0.18$ ($P=70\%$) we get $MR=4140\%$. However, Garc\'\i a \textit{et al.} 

369: \cite{Garcia4} reported also in the footnote, Ref. 9, that few times GMR up to

370: 100000\% was observed in magnetic nanocontacts. Concerning this information,

371: we may guess that this huge magnetoresistance could be actually the result

372: of the quantum spin-valve realization. In contrast to the explanation

373: proposed in \cite{Garcia4}, the quantum spin-valve hypothesis does not need

374: in almost completely (100\%) polarized conduction band to predict 100000\%

375: effect. Theoretically, these numbers may appear even at experimentally

376: approved polarizations of Ni conduction band in the range 35-45\% \cite

377: {TedMes,Soulen98,Soulen99,Buhrman,Nadgorny,Strijk1}. It seems, that quantum

378: spin-valve concept brings us to the upper physical limit of

379: magnetoresistance for a non-superconducting spin-valve-type device. The true

380: infinite (but positive) magnetoresistance can be reached in the

381: proximity-effect superconducting spin-valve (PRESUS-valve) proposed in \cite

382: {Tagirov3,Buzdin}.

383: 

384: In conclusion, we have investigated theoretically the giant

385: magnetoresistance of a nanosize magnetic point contact in the regime of

386: conductance quantization. Concrete calculations have been made for the neck

387: of rectangular cross-section, and dependence of GMR on the asymmetry of

388: cross-section has been studied. Results of calculations show that taking

389: into consideration possible asymmetry of the point contact cross-section one

390: may improve agreement between the theory and the experiment. We argued, that

391: if conductance is open for only one spin-channel, the MR magnitude of this

392: quantum spin-valve is limited by the spin-reversal rate of conduction

393: electrons. For larger areas of the nanocontact the magnetoresistance becomes

394: a multivalued function of the conductance $\sigma ^F$ at ferromagnetic

395: alignment of contacting magnetic domains. This multivalued behavior of MR

396: (which may be treated as giant reproducible fluctuations of MR) is the

397: intrinsic property of quantum magnetic nanocontacts. This property survives

398: for every shape of the nanocontact and disorder, provided that: 1)

399: conductance at the ferromagnetic alignment is quantized (steps are not

400: destroyed); 2) the domain wall in the constriction is effectively sharp.

401: When observed experimentally, such MR distributions should not be

402: interpreted as being due to poor reliability or reproducibility of

403: experimental data.

404: 

405: \section{Acknowledgments}

406: 

407: The authors have benefited from collaboration with Prof. K.B. Efetov. The

408: work is supported by URFI grant N$^0$ 01.01.061 and by BRHE grant REC-007.

409: 

410: \begin{thebibliography}{99}

411: \bibitem{Wees}  B.J. van Wees, H. van Houten, C.W.J. Beenakker, J.G.

412: Williamson, L.P. Kouwenhoven, D. van der Marel, C.T. Foxton, Phys. Rev.

413: Lett. \textbf{60} (1988) 848.

414: 

415: \bibitem{Wharam}  D.A. Wharam, T.J. Thornton, R. Newbury, M. Pepper, H.

416: Ahmed, J.E.F. Frost, D.G. Hasko, D.C. Peacock, D.A. Ritchie, G.A.C. Jones,

417: J. Phys. C \textbf{21} (1988) L209.

418: 

419: \bibitem{Kramer}  J.L. Costa-Kr\"amer, Phys. Rev. B \textbf{55} (1997) 4875.

420: 

421: \bibitem{Oshima}  H. Oshima, K. Miyano, Appl. Phys. Lett. \textbf{73 }(1998)

422: 2203.

423: 

424: \bibitem{Ott}  F. Ott, S. Barberan, J.G. Lunney, J.M.D. Coey, P. Berthet,

425: A.M. de Leon-Guevara, A. Revcolevschi, Phys. Rev. B \textbf{58} (1998) 4656.

426: 

427: \bibitem{Ono}  T. Ono, Y. Ooka, H. Miyajima, Appl. Phys. Lett. \textbf{75}

428: (1999) 1622 .

429: 

430: \bibitem{Imamura}  H. Imamura, N. Kobayashi, S. Takahashi, S. Maekawa, 

431: Phys. Rev. Lett. \textbf{84} (2000) 1003.

432: 

433: \bibitem{Zvezdin}  A.K. Zvezdin, A.F. Popkov, JETP Lett. \textbf{71} (2000)

434: 209.

435: 

436: \bibitem{Garcia1}  N. Garc\'ia, M. Mu\~noz, Y.-W. Zhao, Phys. Rev. Lett. 

437: \textbf{82} (1999) 2923.

438: 

439: \bibitem{Garcia2}  G. Tatara, Y.-W. Zhao, M. Mu\~noz, N. Garc\'ia, Phys.

440: Rev. Lett. \textbf{83} (1999) 2030.

441: 

442: \bibitem{Garcia3}  N. Garc\'ia, M. Mu\~noz, Y.-W. Zhao, Appl. Phys. Lett. 

443: \textbf{76} (2000) 2586.

444: 

445: \bibitem{Tagirov2}  L.R. Tagirov, B.P. Vodopyanov, K.B. Efetov, Phys. Rev. B 

446: \textbf{65} (2002) 214419.

447: 

448: \bibitem{Referee}  We acknowledge the referee of the work \cite{Tagirov2}

449: who stimulated the analysis of the one-spin-projection conductance regime.

450: 

451: \bibitem{Falicov}  G.G. Cabrera, L.M. Falicov, Phys. Stat. Solidi B 

452: \textbf{61} (1974) 539.

453: 

454: \bibitem{Hubert}  A. Hubert, R. Sch\"afer, \emph{Magnetic Domains}

455: (Springer, Berlin, 1988).

456: 

457: \bibitem{Bruno}  P. Bruno, Phys. Rev. Lett. \textbf{83} (1999) 2425.

458: 

459: \bibitem{Tagirov1}  L.R. Tagirov, B.P. Vodopyanov, K.B. Efetov, Phys. Rev. B 

460: \textbf{63} (2001) 104428.

461: 

462: \bibitem{Levy}  P.M. Levy, Sh. Zhang, Phys. Rev. Lett. \textbf{79} (1997)

463: 5110.

464: 

465: \bibitem{Landauer}  R. Landauer, IBM J. Res. Dev. \textbf{32} (1988) 306.

466: 

467: \bibitem{Buttik}  M. B\"uttiker, IBM J. Res. Dev. \textbf{32} (1988) 317.

468: 

469: \bibitem{Gregg}  J.F. Gregg, W. Allen, K. Ounadjela, M. Viret, M. Hehn, S.M.

470: Thompson, J.M.D. Coey, Phys. Rev. Lett. \textbf{77} (1996) 1580.

471: 

472: \bibitem{TedMes}  P.M. Tedrow and R. Meservey, Phys. Rep. \textbf{238}

473: (1994) 173.

474: 

475: \bibitem{Soulen98}  R.J. Soulen Jr., J.M. Byers, M.S. Osofsky, B. Nadgorny,

476: T. Ambrose, S.F. Cheng, P.R. Broussard, C.T. Tanaka, J. Nowack, J.S.

477: Moodera, A. Barry, J.M.D. Coey, Science \textbf{282} (1998) 85.

478: 

479: \bibitem{Soulen99}  R.J. Soulen Jr., M.S. Osofsky, B. Nadgorny, T. Ambrose,

480: P. Broussard, S.F. Cheng, J. Byers, C.T. Tanaka, J. Nowack, J.S. Moodera, G.

481: Laprade, A. Barry, J.M.D. Coey, J. Appl. Phys. \textbf{85} (1999) 4589.

482: 

483: \bibitem{Buhrman}  S.K. Upadhyay, A. Palanisami, R.N. Louie, R.A. Buhrman,

484: Phys. Rev. Lett. \textbf{81} (1998) 3247.

485: 

486: \bibitem{Nadgorny}  B. Nadgorny, R.J. Soulen, M.S. Osofsky, I.I. Mazin, G.

487: Laprade, R.J.M. van de Veerdonk, A.A. Smits, S.F. Chang, E.F. Skelton, S.B.

488: Qadri, Phys. Rev. B \textbf{61} (2000) 3788.

489: 

490: \bibitem{Strijk1}  G.J. Strijkers, Y. Ji, F.Y. Yang, C.L. Chien, J.M. Byers,

491: Phys. Rev. B \textbf{63} (2001) 104510.

492: 

493: \bibitem{Strijk2}  Y. Ji, G.J. Strijkers, F.Y. Yang, C.L. Chien, J.M. Byers,

494: A. Anguelouch, G. Xiao, A. Gupta, Phys. Rev. Lett. \textbf{86} (2001) 5585.

495: 

496: \bibitem{Coey1}  J.M.D. Coey, L. Berger, Y. Labaye, Phys.Rev. B \textbf{64}

497: (2001) 020407.

498: 

499: \bibitem{Coey2}  Y. Labaye, L. Berger, J.M.D. Coey, J. Appl. Phys. 

500: \textbf{91} (2002) 5341.

501: 

502: \bibitem{Zhurav1}  M.Ye. Zhuravlev, H.O. Lutz, A.V. Vedyayev, Phys. Rev. B 

503: \textbf{63} (2001) 174409.

504: 

505: \bibitem{Chopra}  H.D Chopra, S.Z. Hua, Phys. Rev. B \textbf{66} (2002)

506: 020403(R).

507: 

508: \bibitem{Garcia4}  H. Wang, H. Cheng, N. Garc\'\i a, cond-mat/0207516 (22

509: July 2002).

510: 

511: \bibitem{Tagirov3}  L.R. Tagirov, Phys. Rev. Lett. \textbf{83} (1999) 2058.

512: 

513: \bibitem{Buzdin}  A.I. Buzdin, A.V. Vedyayev, N.V. Ryzhanova, Europhys.

514: Lett. \textbf{48} (1999) 686.

515: \end{thebibliography}

516: 

517: %\newpage

518: 

519: \begin{center}

520: \textbf{Figure captions}

521: \end{center}

522: 

523: Fig. 1. The dependence of conductance (a), and MR (b) on the cross-sectional

524: size of the neck $a$, $\varepsilon =1.0$. Panels (c) and (d) show

525: dependencies of MR on the number of the open conductance channels at the

526: F-alignment of the magnetizations. The maximal MR=563\% for the step-like

527: potential at $\sigma ^F=2e^2/h$ is not shown.

528: 

529: Fig. 2. The same as in Fig.1, but for $\varepsilon =1.5$. The maximal

530: MR=758\% for the step-like potential and MR=322\% for the sloping potential

531: at $\sigma ^F=3e^2/h$ are not shown.

532: 

533: \end{document}

534: