1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2: %                                                          %

3: %                WL in GaMnAs                        %

4: %                                                          %

5: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

6: 

7: \documentclass[aps,prl,twocolumn,groupedaddress,amsfonts,showpacs]{revtex4}

8: %\documentclass[aps,prl,preprint,groupedaddress,amsfonts,showpacs]{revtex4}

9: \usepackage{graphicx,color}

10: \usepackage{bm}

11: \usepackage[latin1]{inputenc}

12: 

13: \begin{document}

14: 

15: \title{Weak localization in ferromagnetic (Ga,Mn)As nanostructures}

16: 

17: \author{D. Neumaier}

18: \email{daniel.neumaier@physik.uni-regensburg.de}

19: %\homepage{http://www.physik.uni-regensburg.de/forschung/weiss/}

20: \author{K. Wagner}

21: \author{S. Gei{\ss}ler}

22: \author{U. Wurstbauer}

23: \author{J. Sadowski}

24: \author{W. Wegscheider}

25: \author{D. Weiss}

26: 

27: \affiliation{Institut f\"{u}r Experimentelle und Angewandte Physik,

28: Universit\"{a}t Regensburg, 93040 Regensburg, Germany}

29: 

30: \date{\today}

31: 

32: \begin{abstract}

33: 

34: We report on the observation of weak localization in arrays of

35: (Ga,Mn)As nanowires at millikelvin temperatures. The corresponding

36: phase coherence length $L_{\phi}$ is typically between 100 nm and

37: 200 nm at 20 mK. Strong spin-orbit interaction in the material is

38: manifested by a weak anti-localization correction around zero

39: magnetic field.

40: 

41: 

42: \end{abstract}

43: 

44: \pacs{73.43.Jn, 72.25.Dc, 73.43.Qt}%

45: \keywords{}

46: 

47: \maketitle

48: 

49: Quantum corrections to the resistance like weak localization are

50: suppressed by a sufficiently strong perpendicular magnetic field

51: \emph{B} \cite{Bergmann}. Hence the question arises whether such

52: effects can be observed in ferromagnets which have an intrinsic

53: magnetic induction. While few experimental works explored this

54: problem \cite{Aprili,Dumpich}, a definite experimental answer is

55: still lacking. Hence, the advent of the new ferromagnetic

56: semiconductor material (Ga,Mn)As with significantly smaller internal

57: field compared to conventional ferromagnets offers a new opportunity

58: to address such questions. Ferromagnetic semiconductors like

59: (Ga,Mn)As \cite{Ohno} are interesting materials for spintronics as

60: well, as they combine ferromagnetic properties with the versatility

61: of semiconductors \cite{Fabian}. The spin $\frac{5}{2}$-Mn-ions on

62: regular sites of the zinc-blende lattice of the GaAs host  act as

63: acceptors thus providing both holes and magnetic moments. The

64: ferromagnetic order between the Mn-ions is mediated by these holes

65: \cite{Dietl}. By now ferromagnetism in (Ga,Mn)As is well understood,

66: allowing to predict Curie temperatures \cite{Dietl},

67: magnetocrystalline anisotropies \cite{Sawicki} as well as the

68: anisotropic magnetoresistance effect \cite{Baxter}. In this respect

69: (Ga,Mn)As is one of the best understood ferromagnetic materials at

70: all \cite{Jungwirth} and hence suitable as a model system to study

71: quantum corrections to the conductivity.

72: 

73: Interference effects originating from the charge carriers' wave

74: nature are barely explored and understood in ferromagnets in general

75: and in (Ga,Mn)As in particular. To this class of effects belong

76: universal conductance fluctuations (UCF) \cite{Lee}, the

77: Aharonov-Bohm (AB) effect \cite{Webb}, weak localization (WL)

78: \cite{Bergmann}, weak anti-localization (WAL) \cite{Bergmann} and

79: conductivity corrections due to electron-electron interactions (EEI)

80: \cite{Altshuler}. Recently the existence of AB oscillations in

81: ferromagnetic rings was predicted theoretically \cite{Tatara} and

82: subsequently observed in ferromagnetic Fe$_{19}$Ni$_{81}$-

83: \cite{Kasai} and in (Ga,Mn)As-nanorings \cite{Konni}. In (Ga,Mn)As

84: the phase coherence length was extracted from UCFs in nanowires

85: giving typical values  between 90 nm and 300 nm at 20 mK

86: \cite{Konni,Vila}. This raises the question whether WL corrections -

87: or WAL effects - can be observed in ferromagnetic (Ga,Mn)As, a

88: material in which the spin-orbit (SO) interactions for holes in the

89: valence band is quite strong.

90: 

91: Below we report the observation of WL and WAL in ferromagnetic

92: (Ga,Mn)As-wires and films thus demonstrating that WL is not

93: destroyed by the ferromagnets' magnetization. The effect of WL in

94: disordered electronic systems - investigated intensively in the past

95: for non-ferromagnetic materials \cite{Bird} - is due to quantum

96: interference of two  partial waves traveling the same loop type of

97: path in opposite directions. This leads to an enhanced probability

98: of backscattering. As an applied perpendicular $B$-field suppresses

99: the WL the magnetoconductance is positive \cite{Bergmann}. In the

100: presence of SO interaction the spin part of the wave function needs

101: to be taken into account. The two partial waves on time-reversed

102: closed paths experience a spin rotation in opposite direction

103: causing (partially) destructive interference \cite{Bergmann}. So SO

104: interactions leads to reduced backscattering and reverses the sign

105: of the WL, hence called weak anti-localization. A typical signature

106: of WAL is a double dip in the magnetoconductance trace

107: \cite{Bergmann}.

108: 

109: 

110: For the experiments two wafers having a 42 nm and a 20 nm thick

111: (Ga,Mn)As layer were used. Both were grown by low-temperature

112: molecular beam epitaxy deposited on semi-insulating GaAs(001)

113: \cite{Wegscheider}. The nominal Mn concentration of the 42 nm layer

114: was 5.5 \%, of the 20 nm layer 5 \%. The Curie temperature $T_C$ of

115: the as grown layer was 90 K (42 nm) and 55 K (20 nm), respectively.

116: The samples' remanent magnetization was always in-plane. Some of the

117: samples were annealed at 200 °C increasing both carrier density and

118: $T_C$ \cite{Nottingham}. To investigate phase coherent properties

119: Hall-bar mesas, individual nanowires and arrays of wires were

120: fabricated employing optical and electron beam lithography. For

121: nanowire fabrication we used a %Leo Supra 35

122: scanning electron microscope equipped with a nanonic pattern

123: generator and subsequent reactive ion etching. Au contacts to the

124: devices were made by lift-off technique.

125: %after brief in-situ ion beam

126: %etching of the surface to remove native oxide.

127: The characteristic parameters of the samples investigated are listed

128: in Tab. I.

129: 

130: \begin{table}

131: \begin{tabular}[b]{|l|p{0.8cm}|p{0.8cm}|p{0.8cm}|p{0.8cm}|p{0.8cm}|} \hline

132: Sample & 1a & 2 & 2a & 3 & 4 \\\hline

133: \emph{L} ($\mu$m) & 60 & 7.5 & 7.5 & 7.5 & 0.37 \\

134: \emph{w} (nm) & 7200 & 42 & 42 & 35 & 35 \\

135: \emph{t} (nm) & 20 & 42 & 42 & 42 & 42 \\

136: Number of wires \emph{N} & 1 & 25 & 25 & 12 & 1 \\

137: %Dimension & 2D & 1D & 1D & 1D & 1D \\

138: \emph{t}$_{{anneal}}$ at 200 °C (h)& 8.5 & - & 51 & - & - \\

139: \emph{n} ($10^{26}/\textrm{m}^3$) & 1.7 & 3.8 & 9.3 & 3.8 & 3.8 \\

140: %D ($10^{-5}m^2/s$) & 2.5 & 7.6 & 11.8 & 7.6 & 7.6 \\

141: $\rho$ ($10^{-5}\textrm{ }\Omega\textrm{m}$)  & 13 & 3.5& 1.8 & 3.5 & 3.5 \\

142: $T_C$ (K) & 95 & 90 & 150 & 90 & 90 \\

143: \hline

144: \end{tabular}

145: \label{Daten} \caption{Length \emph{L}, width \emph{w} and thickness

146: \emph{t} of the samples. Some of the samples were annealed at 200

147: °C. Resistivity $\rho$ and carrier concentration \emph{n} were taken

148: at \emph{T} = 300 mK.}

149: 

150: \end{table}

151: 

152: Magnetotransport was measured in a top-loading dilution

153: refrigerator. To avoid heating, we used a low frequency (19 Hz) and

154: low current (25 pA to 200 pA) four probe lock-in technique. As we

155: see no effects of saturation for the different experiments (UCF, WL

156: and conductivity decrease) at low $T$, we assume that the effective

157: electron temperature is in equilibrium with lattice and bath

158: temperature even at 20 mK.

159: 

160: To search for WL effects in (Ga,Mn)As wires we measured the

161: resistance of \emph{N} parallel wires to suppress UCFs by ensemble

162: averaging. A corresponding micrograph of sample 2 with 25 wires is

163: shown in Fig. 1a. The sample's conductance as a function of a

164: perpendicular \emph{B} field is shown in Fig. 1b. First we start

165: with a description of the dominant features observed in experiment.

166: The pronounced conductance maxima around \emph{B} $\sim$ 0 are due

167: to the anisotropic magnetoresistance (AMR) effect \cite{Baxter}. For

168: an in-plane magnetization the conductance is higher than for an out

169: of plane orientation of $M$ \cite{Ohno}. The conductance drops with

170: the growth of the magnetization's out-of-plane component and

171: saturates once $M$ is oriented normal to the surface. The positive

172: slope of the conductance for still higher \emph{B} is due to

173: increasing magnetic order \cite{Nagaev}. For temperatures larger

174: than $\sim$ 90 mK the different $G(B)$ traces are shifted but

175: without noticeable change of shape. The decreasing $G$ for

176: decreasing $T$ in Fig. 1b stems from the usual low \emph{T} behavior

177: of the resistance in (Ga,Mn)As which is plotted in Fig. 1c. With

178: decreasing \emph{T} the resistance rises both for wires (Fig. 1c)

179: and extended (Ga,Mn)As films (not shown) and is ascribed to EEI.

180: Similar low \emph{T} behavior has been reported previously for

181: conventional ferromagnets, too \cite{Dumpich,Ono}. According to

182: theory \cite{Lee2} the EEI conductivity correction for 1D systems

183: goes with $-T^{-1/2}$, for 2D systems with  ln(\emph{T}). The

184: corresponding conductance correction

185: $\Delta\sigma=\sigma(T)-\sigma(50mK)$ of our sample 2, taken at

186: \emph{B} = 0 and at \emph{B} = 3 T is plotted in Fig. 1d vs.

187: $T^{-1/2}$. The resulting straight lines for both \emph{B} values

188: demonstrate the expected \emph{T} dependence, prove that the

189: correction is independent of \emph{B} and hence suggest that EEI is

190: indeed accountable for the conductance decrease at low \emph{T}. For

191: the 2D sample 1a, $\Delta\sigma$ was best described by a

192: ln(\emph{T}) dependence (not shown), as expected for EEI in 2D. The

193: novel features which are in the focus of this letter appear at still

194: lower temperatures. At about 50 mK two downward cusps at about

195: $\pm0.4$ T start to become noticeable and have developed to a

196: prominent feature at 20 mK.

197: 

198: \begin{figure}

199: \includegraphics[width=8cm]{Fig1_CM-1}

200: \caption{(a) Electron micrograph of sample 2 with 25 1D-wires in

201: parallel. (b) Conductance of sample 2 for different temperatures

202: measured in a perpendicular magnetic field. To remove the

203: Hall-conductance in this sample, the antisymmetric part of the

204: conductance was subtracted \cite{Hall}. The magnetic field range

205: where the magnetization is rotated from in-plane to out-of-plane is

206: grey-shaded. (c) Increase of resistance with decreasing temperature.

207: This increase stems from EEI as proven by the $-T^{-1/2}$ power law

208: for 1D-systems at \emph{B} = 0 and \emph{B} = 3 T in (d). Here,

209: $\Delta G $ is taken relative to the conductivity at 50 mK.}

210: \end{figure}

211: 

212: To separate the peculiar low \emph{T} conductance features from the

213: "high temperature" background,  $\Delta G=G(\textrm{20 mK})-\alpha

214: G(\textrm{120 mK})$ of four samples was taken and plotted in Fig. 2.

215: The factor $\alpha$ takes the $T$ dependence of $G$ into account and

216: is given by $\alpha=G(\textrm{20 mK})/G(\textrm{120 mK})$. We note,

217: though, that putting $\alpha=1$ does not change $\Delta G$

218: qualitatively as the conductance change is only $\sim10$\%. To

219: compare the different samples, $\Delta G$ was normalized by the

220: number of parallel wires, \emph{N}. All traces in Fig. 2 show a

221: characteristic broad conductance minimum for $|B|<$1 T and a local

222: maximum at $B\sim0$ T. Such $\Delta G(B)$ line shapes are

223: characteristic for WAL in systems with spin orbit interaction.

224: 

225: \begin{figure}

226: \includegraphics[width=7cm]{Fig1_CM-2}

227: \caption{WL contribution for three wire and one 2D sample obtained

228: after subtracting  the 120 mK trace as background conductance. To

229: compare the different samples the total $\Delta G$ was divided by

230: the number of parallel wires. In case of the 2D-sample 1a $\Delta G$

231: was divided by 15 to fit into the graph. Again the grey shaded

232: \emph{B}-range corresponds to the regime where the samples's

233: magnetization follows the external field and changes direction. The

234: red lines are best fits to Eq.(1), discussed in the text. The fit

235: parameters were $L_{\phi}$ = 190 nm and $L_{SO}$ = 85 nm for sample

236: 2a, $L_{\phi}$ = 150 nm and $L_{SO}$ = 70 nm for sample 2 and

237: $L_{\phi}$ = 160 nm and $L_{SO}$ = 70 nm for sample 3. Fitting the

238: 2D sample requires a different formalism which is beyond the scope

239: of the present work.}

240: \end{figure}

241: 

242: To extract the characteristic lengths from the WL correction we

243: compare the data of Fig. 2 with existing theory. In Fig. 3a we

244: particularly compare the WL correction of sample 3, with the

245: standard expression for WL correction in 1D. Since the width

246: \emph{w}  and thickness \emph{t} of our wires are smaller than the

247: phase coherence length $L_\phi$, $w \sim t<L_\phi<<L$ holds and the

248: 1D assumption is justified. The corresponding equation for the

249: conductance correction reads \cite{Altshuler2,Pierre}:

250: \begin{eqnarray}

251: \Delta G=g_s\frac{e^2}{h}\left[\frac{1}{2L}\left(\frac{1}{L_\phi^2}+\frac{1}{3}\frac{w^2}{L_H^2}\right)^{-1/2}\right.\nonumber \\

252: -\left.\frac{3}{2L}\left(\frac{1}{L_\phi^2}+\frac{4}{3L_{SO}^2}+\frac{1}{3}\frac{w^2}{L_H^2}\right)^{-1/2}\right],

253: \end{eqnarray}

254: where $g_s$ is the spin degeneracy. Here, $L_{SO}=\sqrt{D\tau_{SO}}$

255: is the spin-orbit length that characterizes the strength of spin

256: orbit coupling, $L_{\phi}=\sqrt{D\tau_{\phi}}$, and

257: $L_{H}=\sqrt{\hbar/eB}$ is the magnetic length. Eq. (1) is fitted to

258: the WL data in Fig. 3a for sample 3. As the valence band is spin

259: split, the holes are highly spin polarized \cite{Braden}. To account

260: for spin splitting we use either $g_s=1$ (fully spin polarized) or

261: $g_s=2$ (spin degenerate) as adjustable parameter. While the fit for

262: $g_s=1$ nicely matches the conductance minima at $\pm400$ mT as well

263: as the conductance correction $\Delta G$ the fit for $g_s=2$ is less

264: satisfying. The parameters used for the fit were $L_{\phi}=160$ nm,

265: $L_{SO}=70$ nm for $g_s=1$ and $L_{\phi}=90$ nm, $L_{SO}=38$ nm for

266: $g_s=2$ respectively. Also the WL data of the other samples can be

267: nicely modeled by Eq. (1) and $g_s = 1$; the corresponding fits and

268: parameters are given in Fig. 2.

269: 

270: 

271: \begin{figure}

272: \includegraphics[width=8cm]{Fig1_CM-3}

273: \caption{(a) WL correction of sample 3 fitted with the standard 1D

274: WL theory (Eq. 1) for spin degeneracy $g_s=1$ (red) and $g_s=2$

275: (blue). The parameters used for the fit were $L_\phi=160$ nm,

276: $L_{SO}=70$ nm for $g_s=1$ and $L_\phi=90$ nm, $L_{SO}=38$ nm for

277: $g_s=2$, respectively. (b) UCFs measured in an individual 1D-wire

278: made from the same material (sample 4). An electron micrograph of

279: the wire is shown in the lower left inset. The grey shaded regime

280: again corresponds to the magnetic field range where \emph{M} changes

281: direction. The upper inset shows the low field UCFs in a expanded

282: magnetic field scale. The temperature dependence of $\delta G$ ,

283: extracted from the low-field fluctuations, is shown in the lower

284: right inset.}

285: \end{figure}

286: 

287: The size of the weak (anti)-localization contribution in Fig. 2 and

288: Fig. 3a is quite nicely fitted by two parameters, the phase

289: coherence length $L_\phi$ and the spin orbit length $L_{SO}$ which

290: is the characteristic length on which spin flip occurs due to SO

291: interaction. $L_\phi$ can be extracted independently from UCFs

292: measured on individual 1D-wires \cite{Konni,Vila}, so that

293: essentially only one free parameter prevails. To study UCFs we

294: fabricated a single wire, \emph{w} = 35 nm wide and  \emph{L} = 370

295: nm long, from the same material as sample 2 and 3 (sample 4 in Tab.

296: 1). A corresponding electron micrograph  is shown as lower left

297: inset in Fig. 3b. $G(B)$ was measured in a perpendicular

298: \emph{B}-field  from -3 T to 3 T for $T$ between 20 mK and 1 K (for

299: details see \cite{Konni}). Corresponding data taken at  20 mK show

300: pronounced, reproducible UCFs, displayed in Fig. 3b. The root mean

301: square amplitude $\delta G_{rms}=\sqrt{\langle(G-\langle

302: G\rangle)^2\rangle}$ of these fluctuations is connected with

303: $L_\phi$ and the wire length \emph{L} by $\delta

304: G_{rms}\approx(e^2/h)(L_{\phi}/L)^{3/2}f(L_{\phi}/L_{SO})$

305: \cite{Chandrasekar,TDL}. The function $f(L_{\phi}/L_{SO})$ takes

306: spin-orbit interaction into account. For $L_{\phi}/L_{SO} \sim 2.3$

307: we obtain $f(L_{\phi}/L_{SO})\sim 0.53$ \cite{Chandrasekar}.

308: Extracting $L_\phi$ from  $\delta G_{rms}$, taking only  the

309: fluctuations between $\pm400$ mT in Fig. 3b into account, results

310: then in $L_\phi\sim 120$ nm. The temperature dependence of $\delta

311: G_{rms}$, also taken between $\pm400$ mT is displayed in the lower

312: right inset of Fig. 3 and shows the characteristic power law

313: dependence \cite{Konni}. The value of the phase coherence length,

314: extracted independently from UCFs, is thus in surprisingly good

315: agreement with the ones used to fit the WL correction. Hence our

316: analysis suggests that the spin-orbit length $L_{SO}$ ranges between

317: $\sim$70 nm and $\sim$85 nm in our devices.

318: 

319: While WAL was observed  e.g. in non-magnetic p-type (Al,Ga)As/GaAs

320: quantum wells \cite{Pedersen} or in (In,Ga)As quantum wells

321: \cite{Nitta} the observation of WAL-signature in ferromagnetic

322: (Ga,Mn)As comes as a surprise. A recent theory suggests the

323: processes, leading to WAL in nonmagnetic systems, to be totally

324: suppressed in ferromagnets \cite{Dugaev}. In this theory, both the

325: effect of SO scattering from defects as well as the presence of the

326: Bychkov-Rashba term was taken into account. The suppression of WAL

327: in ferromagnets is due to the strong magnetic polarization which

328: excludes contributions from the so-called singlet Cooperon diagrams,

329: responsible for anti-localization. As a consequence, the quantum

330: correction to $G$ is expected to be exclusively negative in

331: ferromagnets, leading to positive magnetoconductance. This clearly

332: contradicts our experimental observation.

333: 

334: While the fits in Figs. 2 and 3a are in good agreement with

335: experiment for $|B|<\textrm{400 mT}$ the concordance at larger $B$

336: is less perfect. The WL/WAL correction is, as a function of

337: increasing \emph{B}, more abruptly suppressed than expected from

338: theory. There is a striking correlation with the magnetic field

339: dependence of the AMR effect. The magnetic field region where the

340: AMR occurs is highlighted by grey shading in Fig 1b, 2, 3a and 3b.

341: Within this \emph{B}-field range the magnetization is rotated from

342: in-plane to out-of-plane. Once the magnetization is out-of-plane the

343: WL correction drops quickly. In the same $B$-field range the

344: fluctuations of an individual wire show a reduced correlation field

345: $B_C$. Corresponding data are displayed in Fig. 3b, magnified in the

346: upper inset. Similar behavior was observed in previous experiments

347: on samples with in-plane easy axis \cite{Konni,Vila} and ad hoc

348: ascribed to the formation of domain walls in \cite{Vila}. Though we

349: can not exclude such a scenario we note that $B_C$ is not a well

350: defined quantity in the regime where the (magnetic) configuration

351: changes.

352: 

353: The observation of WAL, contrary to theoretical expectation, the

354: abrupt suppression of the WL correction once the magnetization is

355: saturated as well as the anomalous $B_C$ in the low \emph{B}-regime

356: suggest that some important ingredients are still missing to

357: describe interference phenomena in (Ga,Mn)As. This is not too

358: surprising as neither the field dependent change of the

359: magnetization direction nor the $\frac{3}{2}$-spin of the involved

360: hole states was taken into account. Especially the latter could add

361: a number of additional interference diagrams not yet treated

362: theoretically.

363: 

364: In summary we have shown that quantum inference effects strongly

365: affect the low temperature conductance of ferromagnetic (Ga,Mn)As.

366: Electron-electron interaction was identified as origin of the

367: decreasing zero-field conductivity. By resolving a clear weak

368: localization signature we demonstrate that interference due to

369: scattering on time reversed paths can exist also in ferromagnetic

370: materials with internal magnetic induction. The corresponding  phase

371: coherence length $L_\phi$ in our material, defining the maximum

372: enclosed area, is between 100 nm and 200 nm at 20 mK and agree with

373: the values extracted from UCFs. The strong spin-orbit interaction in

374: (Ga,Mn)As is manifested by a weak anti-localization contribution at

375: low \emph{B}.

376: 

377: Acknowledgements: We thank K. Richter, I. Adagideli and A. Geim for

378: valuable discussions. Financial support by the Deutsche

379: Forschungsgemeinschaft (DFG) via SFB 689 is gratefully acknowledged.

380: 

381: 

382: %\begin{references}

383: \begin{thebibliography}{00}

384: 

385: \bibitem{Bergmann} G. Bergmann, Phys. Rep. \textbf{107}, 1 (1984).

386: 

387: \bibitem{Aprili} M. Aprili, J. Lesueur, L. Dumoulin, and P. Nédellec, Solid State Commun.

388: \textbf{102}, 41 (1997).

389: 

390: \bibitem{Dumpich} M. Brands, A. Carl, O. Posth, and G. Dumpich, Phys. Rev. B

391: \textbf{39}, 3015 (2005).

392: 

393: \bibitem{Ohno} H. Ohno, Science \textbf{281}, 951 (1998).

394: 

395: \bibitem{Fabian} I. \v{Z}uti\'{c}, J. Fabian, and S. Das Sarma, Rev. Mod. Phys. \textbf{76}, 323 (2004).

396: 

397: %\bibitem{Dietl} T. Dietl, Semicond. Sci. Technol. \textbf{17}, 377

398: %(2002).

399: 

400: \bibitem{Dietl} T. Dietl \textit{et al.}, Science

401: \textbf{287}, 1019 (2000).

402: 

403: \bibitem{Sawicki} M. Sawicki, J. Magn. Magn. Mater. \textbf{300}, 1 (2006) and references therein.

404: 

405: \bibitem{Baxter} D. V. Baxter \textit{et al.}, Phys. Rev. B \textbf{65}, 212407

406: (2002).

407: 

408: \bibitem{Jungwirth} T. Jungwirth \textit{et al.}, Rev. Mod. Phys. \textbf{78}, 809 (2006)

409: 

410: \bibitem{Lee} P. A. Lee, A. D. Stone, and H. Fukuyama, Phys. Rev. B \textbf{35},

411: 1039 (1987).

412: 

413: \bibitem{Webb} S. Washburn and R. Webb, Adv. Phys. \textbf{35}, 375

414: (1986).

415: 

416: \bibitem{Altshuler} B. L. Altshuler, A. G. Aronov, and P. A. Lee, Phys. Rev. Lett. \textbf{44},

417: 1288 (1980).

418: 

419: \bibitem{Tatara} G. Tatara, H. Kohno, E. Bonet, and B. Barbara, Phys. Rev. B \textbf{69},

420: 054420 (2004).

421: 

422: \bibitem{Kasai} S. Kasai, E. Saitoh, and H. Miyajima, Appl. Phys. Lett.

423: \textbf{81}, 316 (2002).

424: 

425: \bibitem{Konni} K. Wagner \textit{et al.}, Phys. Rev. Lett.

426: \textbf{97}, 056803 (2006).

427: 

428: \bibitem{Vila} L. Vila \textit{et al.}, Phys. Rev. Lett.

429: \textbf{98}, 027204 (2007).

430: 

431: \bibitem{Bird} J. J. Lin and J. P. Bird, Journal of Phys.: Cond.

432: Mat. \textbf{14}, 501 (2002).

433: 

434: \bibitem{Wegscheider} M. Reinwald \textit{et al.}, J. Cryst. Growth

435: \textbf{278}, 690 (2005).

436: 

437: \bibitem{Nottingham} K. W. Edmonds \textit{et al.}, Phys. Rev. Lett.

438: \textbf{92}, 037201 (2004)

439: 

440: \bibitem{Dietl2} T. Dietl, H. Ohno, and F. Matsukura, Phys. Rev. B \textbf{63},

441: 195205 (2001).

442: 

443: \bibitem{Nagaev} E. L. Nagaev, Phys. Rev. B \textbf{58}, 816 (1998).

444: 

445: \bibitem{Hall} In sample 2 and 2a  the two voltage leads were

446: fabricated on opposite sites. So, also the Hall conductance was

447: measured. To remove the Hall conductance from the data only the

448: symmetric part of the trace was taken. As this leads to the same

449: result than seen on sample 1a and 3, where the voltage leads are on

450: the same side, physics is not changed by this procedure.

451: 

452: \bibitem{Ono} T. Ono \textit{et al.}, J. Magn. Magn. Mater. \textbf{226}, 1831

453: (2001).

454: 

455: \bibitem{Lee2} P. A. Lee and T. V. Ramakrishnan , Rev. Mod. Phys.

456: \textbf{57}, 287 (1985).

457: 

458: \bibitem{Altshuler2} B. L. Altshuler and A. G. Aronov, JETP Lett \textbf{33},

459: 499 (1981).

460: 

461: \bibitem{Pierre} F. Pierre \textit{et al.}, Phys. Rev. B

462: \textbf{68}, 085413 (2003).

463: 

464: \bibitem{Braden} J. G. Braden, \textit{et al.}, Phys. Rev. Lett.

465: \textbf{91}, 056602 (2003).

466: 

467: 

468: 

469: \bibitem{Chandrasekar} V. Chandrasekar, P. Santhanam, and D. E. Prober, Phys. Rev. B \textbf{42}, 6823

470: (1990).

471: 

472: \bibitem{TDL} At 20 mK $L_\phi<L_T=\sqrt{\hbar D/k_BT}\sim$ 170 nm

473: holds and the thermal lenght $L_T$ needs not be taken into account.

474: 

475: \bibitem{Pedersen} S. Pedersen \textit{et al.}, Phys. Rev. B

476: \textbf{60}, 4880 (1999).

477: 

478: \bibitem{Nitta} T. Koga, J. Nitta, T. Akazaki, and H. Takayanagi,

479: Phys. Rev. Lett. \textbf{89}, 046801 (2002).

480: 

481: \bibitem{Dugaev} V. K. Dugaev, P. Bruno, and J. Barna\'{s}, Phys. Rev. B

482: \textbf{64}, 144423 (2001).

483: 

484: \end{thebibliography}

485: 

486: 

487: % \end{multicols}

488: 

489: \end{document}

490: