1: %\documentclass[aps,prb,preprint,groupedaddress]{revtex4}
2: %\documentclass[aps,prl,preprint,superscriptaddress]{revtex4}
3: \documentclass[aps,prl,twocolumn,amssymb,groupedaddress,showpacs,floatfix]{revtex4}
4: \usepackage{graphicx}
5: % You should use BibTeX and apsrev.bst for references
6: % Choosing a journal automatically selects the correct APS
7: % BibTeX style file (bst file), so only uncomment the line% below if necessary.
8: \bibliographystyle{apsrev}
9: \def\be{\begin{equation}}
10: \def\ee{\end{equation}}
11: \def\ba{\begin{eqnarray}}
12: \def\ea{\end{eqnarray}}
13: \def\bc{\begin{center}}
14: \def\ec{\end{center}}
15: \def\sgn{{\rm sgn}}
16:
17: \begin{document}
18: \title{Magnetotransport in 2D electron systems with a Rashba spin-orbit interaction}
19:
20: \author{M. V. Cheremisin, A. S. Furman}
21: \affiliation{A.F.Ioffe Physical-Technical Institute,
22: St.Petersburg, Russia}
23: \date{\today}
24:
25: \begin{abstract}
26: The beating pattern of Shubnikov-de Haas oscillations in 2D
27: electron system in the presence of a Rashba zero-field spin
28: splitting is reproduced. It is shown, taking into account the
29: Zeeman splitting, that the explicit formulae for the node position
30: well describes the experimental data. The spin-orbit interaction
31: strength obtained is found to be magnetic field independent in an
32: agreement with the basic assumptions of the Rashba model.
33: \end{abstract}
34:
35: \pacs{73.20.At,71.43.Qt,72.25.Dc, 73.61.-r}
36:
37: \maketitle
38:
39: There has been growing interest in the zero-magnetic-field spin
40: splitting\cite{Lommer85,Luo88,Das90} of the 2D electron gas
41: (2DEG), associated with the spin-orbit interaction (SOI) caused by
42: the structural inversion asymmetry in
43: heterostructures\cite{Rashba}. Application of a gate voltage
44: \cite{Nitta97,Engels97} is known to be the most effective method
45: to control the SOI strength. These 2D systems have been suggested
46: for application in future spintronics devices, such as spin-based
47: field-effect transistors\cite{Datta90}, spin-interference
48: devices\cite{Quan94,Nitta99}, and nonmagnetic spin filters based
49: on a resonant tunneling structure\cite{Koga02a}. Usually, the
50: beating-pattern analysis of Shubnikov-de Haas oscillations ( SdHO
51: ) \cite{Das89} and the weak antilocalization method \cite{Koga02b}
52: are used to determine the SOI strength in 2D systems. However, the
53: former approach is known to lead to a certain controversy in
54: determining the zero-field spin splitting $\Delta$. Namely, the
55: spin splitting deduced from the SdHO beating node position at
56: finite fields \cite{Das89} is different from that $\Delta$
57: expected for $B=0$. In the present paper, this discrepancy is
58: attributed to the contribution of the nonzero Zeeman spin
59: splitting at finite fields. We support our idea by a rigorous
60: analysis of the SdHO beating pattern caused by SOI spin splitting.
61: The beating node positions reported in \cite{Das89} agree well
62: with those predicted by the theory. Then, we demonstrate that the
63: SOI strength is independent of the magnetic field.
64:
65: Let us consider a 2DEG in the x-y plane, subjected to a magnetic
66: field. In the Landau gauge, the one-electron Hamiltonian including
67: the Rashba spin-orbit term \cite{Rashba} is given by
68: \begin{equation}
69: H=\frac{({\bf p}+e{\bf A})^{2}}{2m}+\frac{\alpha}{\hbar}[{\bf
70: \sigma}({\bf p}+e{\bf A})]{\bf n}+\frac{g\mu_{B}}{2}({\bf \sigma
71: B}) \label{hamiltonian}
72: \end{equation}
73: where $\bf{p}$ is the 2D momentum; $m$, the effective mass; $g$,
74: the Zeeman factor; $\mu_{B}$, the Bohr magneton; and, ${\bf n}$,
75: the unit vector in the z-direction. Then, $\bf{ \sigma}$ is the
76: Pauli spin matrix; $\bf{B}$, the total magnetic field; and,
77: $\alpha$, the SOI strength.
78:
79: It has been shown \cite{Rashba} that the solution to
80: Eq.(\ref{hamiltonian}) has an explicit form in the case of a
81: perpendicular magnetic field ${\bf B}=B_{z}=B$. The spectrum for
82: dimensionless energy $\varepsilon=E/\mu$ ($\mu$ is the Fermi
83: energy) is given by \cite{Rashba}
84: \begin{eqnarray}
85: \varepsilon_{0}=\eta\beta,
86: \label{spectrum} \\
87: \varepsilon_{n}^{\pm}=\eta(n \pm \sqrt{\gamma^{2}n+\beta^{2}}), n
88: \ge 1 \nonumber
89: \end{eqnarray}
90: where $\eta=\hbar\omega_{c}/\mu $ is the dimensionless magnetic
91: field; $\omega_{c}= \frac{eB}{mc}$, the cyclotron frequency;
92: $\beta=\frac{1}{2}(1-\chi)$, the term containing the Zeeman spin
93: splitting; $\chi=\frac{g m}{2m_{0}}$, the spin susceptibility; and
94: $n$, an integer similar to that in the conventional description of
95: the Landau levels(LL). Then, according to Ref.\cite{Rashba}
96: $\gamma=\sqrt{\frac{\delta}{\eta}}=\frac{\alpha k_{F}}{\mu
97: \sqrt{\eta}}$, where $\delta$ is the dimensionless SOI strength
98: parameter; $\Delta=2 \alpha k_{F}$, the zero-field spin-orbit
99: splitting at the Fermi energy; and $\hbar k_{F}$, the Fermi
100: momentum. Usually, the typical Fermi energy $\mu \sim 80 $meV
101: exceeds the SOI- induced splitting $\Delta \sim 1 $meV ( see
102: \cite{Das89} ), and, therefore $\delta \ll 1 $. It is noteworthy
103: that the conventional spin-up(down) energy states associated with
104: $n$-th LL number correspond to $\varepsilon_{n}^{+}$ and
105: $\varepsilon_{n+1}^{-}$ states respectively. In the absence of
106: SOI, Eq.(\ref{spectrum}) reproduces well-known LL energy spectrum.
107:
108: In contrast to the conventional formalism extensively used to find
109: the low-B magnetoresistivity, we use the alternative approach
110: \cite{Kirby73,Cheremisin01a,Cheremisin01b} which allows to resolve
111: magnetotransport problem in both the SdHO and Integer Quantum Hall
112: Effect (IQHE) modes. Moreover, this method was successfully used
113: in a recent paper \cite{Cheremisin05} to reproduce the SdHO
114: beating structure in the presence of the zero-field valley
115: splitting (Si-MOSFET 2D system), and in both the crossed- and
116: tilted- field configurations. Following the argumentation put
117: forward in Ref.\cite{Cheremisin01b}, well above the classically
118: strong magnetic field range $\omega_{c}\tau \gg 1$, where $\tau$
119: is the momentum relaxation time, 2DEG can be assumed
120: dissipationless in strong quantum limit when the cyclotron energy
121: $\hbar \omega_{c}$ exceeds both the thermal energy $kT$ and the
122: energy related to LL-width $\hbar/\tau_{q}$. Here, $\tau_{q}$ is
123: the quantum relaxation time. Under the above assumptions
124: $\sigma_{xx}, \rho_{xx} \simeq 0$. Nevertheless, routine dc
125: measurements yield \cite{Cheremisin01b} the finite
126: magnetoresistivity associated with a combination of the Peltier
127: and Seebeck thermoelectric effects. Within the scenario suggested
128: \cite{Cheremisin01b}, we obtain the above magnetoresistivity in
129: the form
130: \begin{equation}
131: \rho= \rho_{yx}\frac{\alpha_{2D}^2}{L} \label{magnetoresistivity}
132: \end{equation}
133: where $\alpha_{2D}$ is the 2DEG thermoelectric power; $\rho
134: _{yx}^{-1}=Nec/B$, the Hall resistivity; $N=- {\partial \Omega
135: \overwithdelims()
136: \partial \mu }_{T}$, the 2D density, $\Omega=-kT \Gamma \sum \limits_{n}\ln
137: \left(1+\exp \left(\frac{\mu -\varepsilon_{n}}{kT}\right)\right)$,
138: the thermodynamic potential; $\Gamma=\frac{eB}{hc}$, the
139: zero-width LL density of states;
140: $L=\frac{\pi^{2}k_{B}^{2}}{3e^{2}}$, the Lorentz number; $k_{B}$,
141: the Boltzmann constant. In fact, the 2D thermoelectric power in
142: strong magnetic fields is a universal quantity \cite{Girvin82},
143: proportional to the entropy per electron: $\alpha_{2D}
144: =-{\frac{S}{eN}}$, where $S=-{\partial \Omega \overwithdelims()
145: \partial T }_{\mu}$ is the entropy. Both $S,N$, and, therefore, $\alpha_{2D},\rho$
146: are universal functions of the dimensionless temperature
147: $\xi=\frac{kT}{\mu}$ and the magnetic field $\eta=2/\nu$, where
148: $\nu=N_{0}/\Gamma$ is the conventional filling factor, and
149: $N_{0}=\frac{m}{\pi \hbar ^{2}}\mu$ is the zero-field density of
150: the strongly degenerate 2DEG in the absence of a SOI-induced
151: splitting.
152:
153: Using the Lifshitz-Kosevich formalism and, then, neglecting finite
154: LL-width( $\hbar/\tau_{q} \rightarrow 0$ ), we derive in Appendix
155: asymptotic formulae for $\Omega$, and, hence, for
156: $N,S,\rho_{yx},\rho$, which are valid at low temperatures and weak
157: magnetic fields $\xi, \eta \ll 1$:
158: \begin{eqnarray}
159: N=N_{0}\xi F_{0}(1/\xi )+2\pi \xi N_{0} \sum
160: \limits_{k=1}^{\infty} \frac{\sin (2\pi k/\eta)}{\sinh
161: (r_{k})}R(\eta),
162: \label{Lifshitz} \\
163: S=S_{0}-2\pi ^{2}\xi k_{B}N_{0} \sum \limits_{k=1}^{\infty }\Phi
164: (r_{k})\cos(2\pi k/\eta)R(\eta), \nonumber
165: \end{eqnarray}
166: where $S_{0}= k_{B}N_{0}(2\xi F_{1}(1/\xi)-F_{0}(1/\xi))$ is the
167: entropy at $B=0$; $F _{n}(z)$, the Fermi integral; and $\Phi
168: (z)=\frac{1-z\coth (z)}{z\cdot \sinh (z)}$. At $B=0$ both the
169: thermopower and 2D density are constants, i.e.
170: $\alpha_{2D}=\frac{\pi^{2}\xi^{2}}{3}\frac{k_{B}}{e}, N=N_{0}$,
171: hence the magnetoresistivity is given by zero-field asymptote
172: $\rho= \frac{h}{e^{2}} \frac{\pi^{2}\xi^{2}\eta}{6}$. According to
173: Eq.(\ref{Lifshitz}), for actual first-harmonic case( $k=1$ ) the
174: magnetoresistivity can be viewed as the zero-field background, on
175: which the rapid SdHO modulated by long-period beatings( see
176: Fig.\ref{f2} ) are superimposed. It's worthwhile to mention that
177: at the beat nodes( i.e. when the form-factor at $k=1$ vanishes )
178: the magnetoresistivity is given by zero-field asymptote. This is
179: not, however, the case of low temperatures and(or) high magnetic
180: fields when the high-order terms($k>1$) in Eq.(\ref{Lifshitz}) may
181: determine the amplitude of magnetoresistivity at the beat nodes.
182: It turns out that the data reported in \cite{Das89} point to the
183: above feature.
184:
185: We now analyze in detail the form-factor $R(\eta)$( see Appendix )
186: which determines the beating pattern of $S,N$ and, hence, $\rho$.
187: For the actual first-harmonic case (i.e., $k=1$), the beating
188: nodes can be observed when $R(\eta)=0$ or
189: \begin{equation}
190: \sqrt{\beta^{2}+\frac{\delta}{\eta^{2}}}=\frac{j}{4}, \label{node}
191: \end{equation}
192: where we neglect the small quadratic term $\delta^{2}/4 \eta^{2}
193: \ll \delta/\eta^{2}$ evaluating Eq.(\ref{N+/-}). Then, $j=1,3..$
194: is the beating node index. We emphasize that the first node cannot
195: be observed in experiments, performed, for example, in Ref.
196: \cite{Das89}. Indeed, for real 2D
197: In$_{x}$Ga$_{1-x}$As/In$_{0.52}$Al$_{0.48}$As system (
198: $m=0.049m_{0}$, $g \simeq 4$ ) we find $\beta=0.45$, and,
199: therefore Eq.(\ref{node}) cannot be satisfied for $j=1$. With the
200: help of Eq.(\ref{node}), we analyze the nodes, reported in
201: \cite{Das89} for three different samples, and then plot the
202: dependence of the zero-field SOI splitting at the Fermi energy
203: $\Delta$ against the node index( see Fig.\ref{f1} ), starting from
204: $j=3$. For these samples $\Delta$ is nearly constant within the
205: actual range of the magnetic fields, therefore we obtain the
206: respective mean values $\Delta_{0}$ denoted in Table
207: \ref{tab:table1}. Note that the minor deviation of $\Delta$ with
208: respect to its mean value in high-field limit( low-index nodes )
209: can be associated with possible magnetic field dependence of the
210: g-factor. In contrast, the non-parabolicity effects \cite{Hu99}
211: seem to be irrelevant \cite{Bychkov90} for the actual low-field
212: case $B<1$T.
213:
214: \begin{figure} \vspace*{0.5cm}
215: \includegraphics[scale=0.75]{Fig1} \caption[]{\label{f1}
216: Zero-field SOI splitting at the Fermi energy vs node index $j$,
217: deduced from the experimental data \cite{Das89}, with the help of
218: the node condition specified by Eq.(\ref{node}). Dotted lines(
219: from top to bottom ) represent the mean values $\Delta_{0}$ for
220: samples B,A,C respectively.} \vspace*{-0.5cm}
221: \end{figure}
222:
223: We emphasize that the node condition similar to Eq.(\ref{node})
224: was previously discussed in literature. Following the analysis
225: done in Ref.\cite{Bychkov90}, the nodes occur when the
226: spin-orbit-split subbands are shifted one with respect another by
227: half a period at the Fermi energy. Namely, $1 \simeq
228: \varepsilon_{n}^{+}=(\varepsilon_{n+s}^{-}+\varepsilon_{n+s+1}^{-})/2$,
229: where $s=0,1,2...$ corresponds to the node index as $j=1+2s$. For
230: actual high LL-number case $n \gg 1$ this condition reproduces
231: Eq.(\ref{node}).
232:
233: \begin{table}
234: \caption{\label{tab:table1} Transport data( at 4.2K ) and the
235: Zero-field spin-orbit splitting at Fermi energy for
236: In$_{x}$Ga$_{1-x}$As/In$_{0.52}$Al$_{0.48}$As 2D system reported
237: in Ref.\cite{Das89}}
238: \begin{ruledtabular}
239:
240:
241: \begin{tabular}{cccccccc}
242: Sample($x$) &$\mu_{0} \times 10^{4}$,cm$^{2}$/Vs &$n \times 10^{12}$cm$^{-2}$ &$\mu$,meV &$\Delta_{0}$,meV \\
243: \hline A(0.65)&13.4 &1.75 &78 &2.34 \\
244: B(0.60)&9.5 &1.65 &74 &2.57 \\
245: C(0.53)&6.8 &1.46 &65 &1.63 \\
246:
247: \end{tabular}
248: \end{ruledtabular}
249: \end{table}
250:
251: Let us discuss the conventional method \cite{Das89} often used to
252: extract the zero-field SOI splitting at the Fermi energy.
253: According to phenomenological arguments put forward by Das et al
254: \cite{Das89,Das90}, the nodes may occur when $\cos \left(
255: \pi\frac{\Delta_{tot}}{\hbar\omega_{c}}\right )=0$ or
256: $\Delta_{tot}=\pm \frac{j}{2} \hbar\omega_{c}$, where the total
257: spin splitting at the Fermi energy between spin-down
258: $\varepsilon_{n+1}^{-}$ and spin-up $\varepsilon_{n}^{+}$ states
259: yields $\Delta_{tot}=\hbar \omega_{c}-\sqrt{(2\beta\hbar
260: \omega_{c})^{2}+\Delta^{2}}$. As expected, the total spin
261: splitting $\Delta_{tot}$ coincides with the zero-field $-\Delta$
262: and the Zeeman $\chi \hbar \omega_{c}$ spin splitting in low( high
263: ) magnetic field limit respectively. With the help of the
264: dimensionless units the node condition suggested by Das et al
265: reads $ \sqrt{\beta^{2}+\frac{\delta}{\eta^{2}}}=\frac{1 \pm
266: j/2}{2}$, hence, reproduces our result if one selects "$+$" set at
267: $j \ge 1$. We argue that straightforward procedure ( see
268: Fig.\ref{f1} ) used to extract $\Delta_{0}$ is, however,
269: preferable compare to zero-field extrapolation method suggested in
270: Ref.\cite{Das89,Das90}. Indeed, for low-density samples and(or)
271: under the temperature enhanced conditions the SdHO amplitude is
272: suppressed, hence, the low-field nodes become hidden. In this case
273: the zero-field extrapolation method \cite{Das89} may lead to a
274: subsequent errors.
275:
276: \begin{figure} \vspace*{0.5cm}
277: \includegraphics[scale=0.75]{Fig2}
278: \caption[]{\label{f2} SdHO beating pattern calculated with the
279: help of Eq.(\ref{Lifshitz}) at $k=1$ for sample A($x=0.65$)
280: \cite{Das89}: $N_{0}=1.75*10^{12}$ cm$^{-2}$, $m=0.049m_{0}$,
281: $g=4$, $\Delta_{0}=2.34$meV, $T=1.6$K. Arrows show the beating
282: nodes at $j=3,5..$. Zero-field asymptote is represented by dotted
283: line} \vspace*{-0.5cm}
284: \end{figure}
285:
286: Let us now reproduce( see Fig.\ref{f2} ) the SdHO beating pattern
287: with the nodes occurred in a typical sample( sample A($x=0.65$) in
288: \cite{Das89}) at $B=0.873;0.46;0.291;0.227;0.183;0.153$T using
289: Eq.(\ref{Lifshitz}), and previously extracted value of zero-field
290: SOI splitting $\Delta_{0}=2.34$meV. It's worthwhile to mention
291: that our results differ with respect to those, which can be
292: obtained within the conventional formalism in the following: (i)
293: the low-field quantum interference, classical magnetoresistivity
294: and 3D substrate parallel resistivity\cite{Das89} background are
295: excluded within our approach; (ii) in contrast to conventional
296: SdHO analysis, our method determines the absolute value of
297: magnetoresistivity, and, moreover, can lead to a gradual
298: transition \cite{Cheremisin01b} from the SdHO to the IQHE mode.
299:
300: We argue that the noticeable increase in SdHO amplitude was
301: observed \cite{Das89} at $B \simeq 0.3$T. This value satisfies the
302: criterion of the classically strong magnetic field since
303: $\omega_{c}\tau=4$ while the corresponding cyclotron energy $\hbar
304: \omega_{c}=8.2$K correlates with that $\sim 9.8$K expected from
305: T-dependent SdHO-damping factor, i.e. when $2\pi^{2}\xi/\eta \sim
306: 1$. We conclude that the energy associated with LL width $\sim
307: \hbar/\tau_{q}$ is less or at least equal to the thermal energy.
308: The above estimates point to validity of zero-width LL model in
309: this particular case. Nevertheless, since both the temperature and
310: finite LL width known to suppress the SdHO amplitude in a rather
311: similar manner, we esteem reasonable to reproduce in Fig.\ref{f2}
312: the SdHO beating pattern using somewhat higher temperature
313: $T=1.6$K than that $T=0.5$K reported in \cite{Das89}.
314:
315: Note that our approach provides a correct number of oscillations
316: between the adjacent nodes. For example, the number of
317: oscillations confined between $j=3,5$ nodes (37) correlates with
318: that (35) observed in \cite{Das89}. A minor point is that our
319: approach predicts a somewhat lower amplitude of SdHO, compared
320: with that in the experiment \cite{Das89}. For example, for $j=3$
321: node( $B=0.873$T in Ref.\cite{Das89} ) we obtain $\rho=0.0035$Ohm.
322: Actually, one would expect the same order of magnitude for SdHO
323: amplitude between the proximate nodes ( see $j=3,5$ in
324: Fig.\ref{f2} ). Our estimation is, however, less than both the
325: absolute magnetoresistivity $40$Ohm at $B=0.873$T and SdHO
326: amplitude $\sim 5$Ohm reported in Ref.\cite{Das89}.
327:
328: In conclusion, we demonstrated the relevance of the
329: approach\cite{Cheremisin01b} regarding the beating pattern of SdHO
330: caused by Rashba spin-orbit interactions. Taking into account the
331: Zeeman splitting, the rigorous analysis of experimental data
332: \cite{Das89} suggests a B-independent strength of the Rashba SOI.
333: The above finding is consistent with the general theoretical
334: assumptions \cite{Rashba}. Our approach can be helpful for
335: estimation of the SOI strength.
336:
337: The authors wish to thank Prof. N.Averkiev and Dr. S.Tarasenko for
338: helpful comments. This study was supported by the Russian
339: Foundation for Basic Research (grant 03-02-17588) and LSF (
340: Weizmann Institute).
341:
342: \section{Appendix}
343: \label{Lifshitz-Kosevich formalism}
344:
345: Using the conventional Poisson formulae
346: \begin{equation}
347: \sum\limits_{m_{0}}^{\infty }\varphi (n)=\int\limits_{a}^{\infty
348: }\varphi (n)dn+2\text{Re}\sum\limits_{k=1}^{\infty
349: }\int\limits_{a}^{\infty }\varphi (n)e^{2\pi ikn}dn,
350: \label{Poisson}
351: \end{equation}
352: where $m_{0}-1<a<m_{0}$, $m_{0}$ the initial value of the
353: summation, the thermodynamic potential can be represented as the
354: sum $ \Omega=\Omega_{0}+\Omega_{\sim}$ of the zero-field and
355: oscillating parts as follows
356: \begin{eqnarray}
357: \Omega_{0}=-N_{0} \mu \xi^{2} F_{1}(1/\xi),
358: \label{LL_Omega1} \\
359: \Omega_{\sim}=-N_{0} \mu \eta \xi {\text Re}
360: \sum\limits_{k=1}^{\infty }\int\limits_{0}^{\infty }e^{2\pi ikn}
361: \ln \left( 1+e^{\frac{1-\varepsilon_{n}^{\pm}}{\xi}} \right )d n,
362: \nonumber
363: \end{eqnarray}
364: where $F_{n}(z)$ is the Fermi integral. For simplicity, we omit
365: the SOI-induced splitting in zero-field term $\Omega_{0}$ because
366: $\delta \ll 1 $. The special interest of the present paper is in
367: the oscillating term $\Omega_{\sim}$ of thermodynamic potential,
368: which can be strongly affected by spin-orbit-split
369: subbands($\pm$). After a simple integration by parts, the
370: oscillating term yields
371: \begin{equation}
372: \Omega_{\sim}=N_{0} \mu {\text Re} \sum\limits_{k=1}^{\infty
373: }\frac{i \eta }{2\pi k } \int \limits_{0}^{\infty } \frac{e^{2\pi
374: ikn^{\pm}}}{1+e^{\frac{\varepsilon-1}{\xi}}}d\varepsilon
375: \label{LL_Omega2}
376: \end{equation}
377: Using Eq.(\ref{spectrum}), for a certain energy we calculate the
378: actual high-order LL-like numbers, associated with both the
379: spin-orbit-split subbands as
380: \begin{equation}
381: n^{\pm}(\varepsilon)=\frac{\varepsilon}{\eta}+\frac{\gamma^{2}}{2}
382: \pm \sqrt{\beta^{2}+ \frac{\gamma^{2}\varepsilon}{\eta}+
383: \frac{\gamma^{4}}{4}}. \label{N+/-}
384: \end{equation}
385: It should be noted that the integrand equation in
386: Eq.(\ref{LL_Omega2}) is a rapidly oscillating function, which is,
387: in addition, strongly damped when $\varepsilon
388: > 1$. The major part of the magnitude of the integral results from the
389: energy range close to the Fermi energy, when $\varepsilon \sim 1$.
390: Therefore, $n^{\pm}(\varepsilon)$ can be regarded as smooth
391: functions of energy, and, hence, can be re-written as $n^{\pm}=
392: n^{\pm}_{1}+{\partial n^{\pm} \overwithdelims()\partial
393: \varepsilon}_{1}(\varepsilon-1)$, where we use the designation
394: $n^{\pm}_{1}=n^{\pm}(1)$ . Under the above assumption, we can
395: change the lower limit of integration to $-\infty$ and then use
396: the textbook expression $\int\limits_{-\infty}^{\infty}\frac{e^{i
397: k y}}{1+e^{y}}dy=\frac{-i\pi}{\sinh(\pi k)}$ for the integral of
398: the above type. Finally, the thermodynamic potential yields
399: \begin{equation}
400: \Omega=\Omega_{0} + N_{0} \mu 2 \pi^{2} \xi^{2}
401: \sum\limits_{k=1}^{\infty }\frac{\cos(\pi k
402: (n^{+}_{1}+n^{-}_{1}))R(\eta)}{r_{k}\sinh{r_{k}}}
403: \label{LL_Omega3}
404: \end{equation}
405: where we assume that ${\partial n^{\pm} \overwithdelims()\partial
406: \varepsilon}_{1}\sim \frac{1}{\eta}$ is valid for the actual case
407: of high- order Landau levels $n^{\pm} \gg 1$, and $r_{k}=2\pi
408: ^{2}\xi k/\eta$ is a dimensionless parameter related to T-damping
409: of SdH amplitude. Then, $R(\eta)=\cos(\pi k(n^{+}_{1}-n^{-}_{1}))$
410: is the form-factor. The oscillatory part of the thermodynamic
411: potential consists of rapid oscillations $\cos(\pi
412: k(n^{+}_{1}+n^{-}_{1})) \simeq \cos(2\pi k/\eta)$, on which
413: long-period beatings governed by the form-factor are superimposed.
414: As expected, the form-factor is reduced in absence of SOI to a
415: field-independent constant $R(\eta)=\cos(2\pi k \beta)$, and,
416: therefore, the beating structure is absent. Using the conventional
417: thermodynamic definition, we can easily obtain both the entropy
418: and the density of 2D electrons, specified by
419: Eq.({\ref{Lifshitz}}).
420:
421: \smallskip
422:
423: \begin{thebibliography}{99}
424: \bibitem{Lommer85} G. Lommer, F. Malcher, and U. Rössler, Phys. Rev. B, {\bf 32}, 6965 (1985).
425: \bibitem{Luo88} J. Luo, H. Munekata, F.F. Fang, and P.J. Stiles, Phys. Rev. B {\bf 38}, 10142 (1988).
426: \bibitem{Das89} B. Das, D.C. Miller, S. Datta, R. Reifenberger, W.P. Hong, P.K. Bhattacharaya, J. Singh, and M. Jaffe, Phys. Rev. B {\bf 39}, 1411 (1989).
427: \bibitem{Das90} B. Das, S. Datta, R. Reifenberger, Phys. Rev. B {\bf 41}, 8278 (1990).
428: \bibitem{Rashba} E.I. Rashba, Fiz. Tverd. Tela (Leningrad) {\bf 2}, 1224 (1960)
429: [Sov. Phys. Solid State {\bf 2}, 1109 (1960)]; Y.A. Bychkov and
430: E.I. Rashba, J. Phys. C {\bf 17}, 6039 (1984).
431: \bibitem{Bychkov90} Y.A. Bychkov, V.I. Mel'nikov, and E.I. Rashba, Zh. Eksp. Teor. Fiz. {\bf 98},
432: 717 (1990), [Sov. Phys. JETP {\bf 71}, 401 (1990)].
433: \bibitem{Nitta97} J. Nitta, T. Akazaki, H. Takayanagi, and T. Enoki, Phys. Rev. Lett. {\bf 78}, 1335 (1997).
434: \bibitem{Engels97} G. Engels, J. Lange, Th. Schäpers and H. Lüth, Phys. Rev. B {\bf 55}, R1958 (1997).
435: \bibitem{Datta90} S. Datta and B. Das, Appl. Phys. Lett. {\bf 56}, 665 (1990).
436: \bibitem{Quan94} Tie-Zheng Qian and Zhao-Bin Su, Phys. Rev. Lett. {\bf 72}, 2311 (1994).
437: \bibitem{Nitta99} J. Nitta, F.E. Meijer, and H. Takayanagi, Appl. Phys. Lett. {\bf 75}, 695 (1999).
438: \bibitem{Koga02a} T. Koga, J. Nitta, H. Takayanagi, and S. Datta, Phys. Rev. Lett. {\bf 88}, 126601 (2002).
439: \bibitem{Koga02b} T. Koga, J. Nitta, T. Akazaki, and H. Takayanagi, Phys. Rev. Lett. {\bf 89}, 046801 (2002).
440: \bibitem{Kirby73} C.G.M. Kirby and M.J. Laubitz, Metrologia {\bf 9}, 103 (1973).
441: \bibitem{Cheremisin01a} M.V. Cheremisin, Zh. Eksp. Teor. Fiz. {\bf 119}, 409 (2001), [Sov. Phys. JETP, {\bf 92}, 357, 2001].
442: \bibitem{Cheremisin01b} M.V. Cheremisin, Physica E, {\bf 28}, 393 (2005).
443: \bibitem{Cheremisin05} M.V. Cheremisin, Physica E {\bf 27}, 151 (2005).
444: \bibitem{Girvin82} S.M. Girvin and M. Jonson, J.Phys.C {\bf 15}, L1147 (1982).
445: \bibitem{Hu99} Can-Ming Hu, J.Nitta, T.Akazaki et al, Phys. Rev. B {\bf 60}, 7736 (1999).
446: \bibitem{Dobers88} M. Dobers, K. von Klitzing, G. Weimann, Phys. Rev. B {\bf 38}, 5453 (1988).
447: \bibitem{Dobers89} M. Dobers, J.P.Viereit, Y. Guldner et al, Phys. Rev. B {\bf 40}, 8075 (1989).
448:
449: \end{thebibliography}
450:
451:
452: \end{document}
453: