1: % \documentclass[prl,aps,twocolumn,showpacs]{revtex4}
2: % \documentclass[prl,aps]{revtex4}
3: %\usepackage[dvips]{graphicx}
4: %\usepackage{dcolumn}
5: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
6: %
7: \documentclass[prl,aps,twocolumn,showpacs]{revtex4}
8: % \documentclass[prl,aps]{revtex4}
9: \usepackage[dvips]{graphicx}
10: \usepackage{dcolumn}
11: \newcommand{\be}{\begin{equation}}
12: \newcommand{\ee}{\end{equation}}
13: \newcommand{\bea}{\begin{eqnarray}}
14: \newcommand{\eea}{\end{eqnarray}}
15: \newcommand{\nn}{ \nonumber}
16: \newcommand{\ds}{\displaystyle}
17: \begin{document}
18: \topmargin=-10mm
19: % \Large
20:
21:
22:
23:
24:
25: \title{Theory of the dc Magnetotransport in Laterally Modulated Quantum Hall Systems Near Filling $\ds \ {\cal V}=\frac{1}{2}$}
26:
27: \author{Natalya Zimbovskaya$^{1,2,3}$, Godfrey Gumbs$^2$ and Joseph L. Birman$^3$ }
28:
29:
30: \affiliation{
31: $^1$Department of Physics and Astronomy, St. Cloud State
32: University, 720 Fourth Avenue South, St. Cloud MN, 56301;\\
33: $^2$Department of Physics and Astronomy
34: Hunter College of the City University of New York,
35: 695 Park Avenue, New York, NY 10021; \\ $^3$Department of Physics City College of the City University of New York,
36: Convent Avenue and 138th Street, New York, NY 10031 }
37:
38: \date{\today}
39:
40:
41: \begin{abstract}
42: A quasiclassical theory for dc magnetotransport in a modulated quantum
43: Hall system near filling factor $\nu=\frac{1}{2}$ is presented. A weak
44: one-dimensional electrostatic potential acts on the two-dimensional
45: electron gas. Closed form analytic expressions are obtained for the
46: resistivity $\rho_\perp$ corresponding to a current at right angles to the
47: direction of the modulation lines as well as a smaller component
48: $\rho_{||}$ for a current along the direction of the modulation lines. It
49: is shown that both resistivity components are affected by the presence of
50: the modulation. Numerical results are presented for $\rho_\perp$ and
51: $\rho_{||}$ and show reasonable agreement with the results of recent
52: experiments.
53: \end{abstract}
54:
55:
56: \pacs{73.43.Cd, 73.63.-b,05.60.-k}
57:
58: \maketitle
59:
60:
61: \section{\small 1. Introduction and Background}
62: \label{sec1}
63:
64: It is well known that the transport properties of a two-dimensional
65: electron gas (2DEG) in high magnetic fields providing a half filling of
66: the lowest Landau level ($\nu= 1/2$) are well described by the
67: theory of Halperin, Lee and Read (HLR) \cite{one,two} which corresponds to
68: a physical picture in which electrons are decorated by attached quantum
69: flux tubes. These are the relevant quasiparticles of the system -- so-called composite fermions (CFs). The CFs are spinless fermionic
70: quasiparticles with charge $-e$ which move in a reduced effective magnetic
71: field $B_{eff}= B-4\pi\hbar nc/e$, where $n$ is the electron sheet
72: density. At $\nu=1/2$, the CFs form a Fermi sea and exhibit a
73: Fermi surface (FS). For the homogeneous 2DEG, the FS of the CFs is a
74: circle of radius $p_F=\sqrt{4\pi n\hbar^2}$ in quasimomentum space.
75:
76: According to the theory of HLR, the resistivity tensor of the 2DEG is
77: given by $\tensor{\rho}=\tensor{\rho}_{cf}+\tensor{\rho}_{cs}$, where
78: $\tensor{\rho}_{cf}$ is the contribution from CFs whereas
79: $\tensor{\rho}_{cs}$ arises from the fictitious effective magnetic field
80: in the Chern-Simons formulation of the theory. Here, $\tensor{\rho}_{cs}$
81: has only off-diagonal elements. Therefore, the diagonal elements of the
82: resistivity tensor of the 2DEG are the same as those for the CF
83: resistivity tensor $\tensor{\rho}_{cf} =\tensor{\sigma}_{cf}^{-1}$, where
84: $\tensor{\sigma}_{cf}$ is the CF conductivity tensor.
85:
86: Some interesting features of the magnetoresistivity near $\nu=1/2$ have
87: been reported in the transport experiments of a 2DEG whose density is
88: modulated by a weak one-dimensional electrostatic potential. This includes
89: a well-defined minimum at $\nu=1/2$ in the resistivity
90: $\rho_\perp$ corresponding to a current at right angles to the direction
91: of the modulation lines \cite{three,four} as well as a smaller component
92: $\rho_{||}$ with a maximum at $\nu=1/2$ for a current along the
93: lines of the modulation \cite{four}. In addition, oscillations in these
94: two components of the resistivity were observed close to filling factor
95: $\nu=1/2$. Previous theories for dc magnetotransport in modulated quantum
96: Hall systems near one half filling based on the Boltzmann transport
97: equation for the CF distribution function \cite{five,six,seven} are
98: succesfull in accounting for the minimum in the magnetoresistivity $
99: \rho_\perp.$ However, these theories, as well as semiclassical studies of
100: dc magnetotransport properties of modulated 2DEG at low magnetic field
101: \cite{eight,nine,ten,eleven}, fail to explain the magnetic field
102: dependence of $ \rho_{||}.$ Up to present the effect of modulations on $
103: \rho_{||} $ was treated as a pure quantum effect which cannot be described
104: within semiclassical calculations, although it remains exhibited within
105: that range of magnetic fields where the semiclassical approach could be
106: applied \cite{twelve,thirteen,fourteen,fifteen}.
107:
108: We think that the existing theory is mistaken at this point, and the
109: magnetic field dependence of both resistivity components near $\nu = 1/2 $
110: is dominated with similar mechanism and could be analyzed within a
111: semiclassical approximation. This can be shown if we adopt a correct
112: procedure of averaging of CF response functions over the period of
113: modulations \cite{sixteen,seventeen} which gives semiclassical analogs for
114: the response functions obtained as a result of quantum mechanical
115: calculations. The proposed procedure differs from that used in the
116: existing semiclassical theories \cite{eight,nine,ten,eleven} and provides
117: different results for transport coefficients of a modulated 2DEG. In
118: contrast with the corresponding conclusions of the earlier works, our
119: results for the both components of magnetoresistivity tensor are
120: consistent with those obtained as a semiclassical limit of quantum
121: mechanical calculations (see e.g. \cite{thirteen}). They also demonstrate
122: a better agreement with experiments on a dc magnetotransport in a
123: modulated 2DEG in low magnetic fields.
124:
125: In the main body of the paper we use a simplified semiquantitative
126: approach, resembling that used in earlier works of Beenakker \cite{eight}
127: and Gerhardts \cite{nine}. In the experiments of Ref.4 the mean free path
128: $ l $ of the CFs is larger than their cyclotron radius $ R $ but smaller
129: or of the same order as the period of modulations $ \lambda.$ In this
130: "local" regime $ R, l< \lambda$ this method of calculations gives
131: reasonably good approximation for the transport coefficients we seek.
132:
133:
134:
135: In the Appendix, we present calculations of the magnetoresistivity
136: components based on the transport Boltzmann equation within a relaxation
137: time approximation. The results of this analysis corroborate reasonably
138: with the semiquantitative approach developed in the main body of the
139: paper.
140:
141:
142: \section{\small 2. General Formulation of the Problem}
143: \label{sec2}
144:
145: We now consider a sinusoidal density modulation with a single harmonic of
146: period $\lambda=2\pi/g$ along the $y$ direction and given by $\Delta
147: n({\bf r}) =\Delta n\sin(gy)$. This density modulation influences the CF
148: system through the appearance of an additional inhomogeneous magnetic
149: field $\Delta B({\bf r}) =-4\pi \hbar c\Delta n(y)/e$, which is
150: proportional to the density modulation $\Delta n(y)$ as well as through
151: the external modulating electric field. Following Ref.\ \cite{seven}, we
152: parameterize the electric potential screened by the 2DEG as
153: $$
154: \displaystyle{U(y)=\frac{\Delta n}{n}E_F\frac{\sin(gy)}{e}},
155: $$
156: where $E_F$ is the Fermi energy of
157: the unmodulated CF system.
158:
159: Our starting point is the Lorentz force equations describing the
160: CF motion along the cyclotron orbits
161: % f1
162: \bea
163: \frac{dp_x}{dt}& = &-\frac{e}{c}B(y)v_y ; \nn \\ \nn \\
164: \frac{dp_y}{dt} & = &
165: \frac{e}{c}B(y)v_{x}+e\frac{dU(y)}{dy}\ ,
166: \label{e1}
167: \eea%%%%%%%%nd{equation}
168: where $p_x,p_y$ and $v_x,v_y$ are the components of the quasimomentum and
169: velocity of the CF and $B(y)=B_{eff}-4\pi\hbar\Delta n(y)c/e$.
170:
171:
172: For weak modulations, $\Delta n/n\ll 1$, we may assume that inhomogeneous
173: terms in Eq.\ (\ref{e1}) are small for all values of $B_{eff}$ except when
174: the filling factor is close to $\nu=1/2$, where $B_{eff}\to 0$. In this
175: limit, we can write the CF velocity as ${\bf v} ={\bf v}_0+\delta{\bf v}$,
176: where ${\bf v}_0$ is the uniform field velocity and the correction
177: $\delta{\bf v}$ is due to the density modulation. For a circular CF--FS,
178: we
179: have $v_{x0}=v_F\cos\Omega t,\ \ v_{y0}=v_F\sin\Omega t,\ \ \Delta
180: n(y)\approx\Delta n \sin(gY\\-gR\cos\Omega t)$, where $\Omega$ is the
181: cyclotron frequency, $v_F$ is the Fermi velocity for CFs, $R=v_F/\Omega$
182: is the cyclotron radius, and $Y$ is the $y$ coordinate of the guiding
183: center. Substituting these results for ${\bf v}$ and $\Delta n(y)$ into
184: Eq.\ (\ref{e1}) and keeping only the terms to first order in the
185: perturbation, we obtain
186: % f2
187: %%\begin{mathletters}
188: %%\label{e2}
189: \be%%\begin{eqnarray}
190: \frac{d(\delta v_x)}{dt} = -\Omega\delta v_y-\frac{\Delta B}{B_{eff}}\Omega v_F\sin(\Omega t)
191: \sin(gY-gR\cos(\Omega t))\ ,
192: \ee
193: %%\nonumber\\ \nonumber\\ &\times&
194: %%\label{e2a}
195: %% \ee %%\end{eqnarray}
196: \bea %% \begin{eqnarray}
197: \frac{d(\delta v_y)}{dt} & = & \Omega\delta v_x+\frac{\Delta B}{B_{eff}}\Omega v_F \cos(\Omega t)
198: \sin(gY- gR \cos(\Omega t))
199: \nonumber\\ \nonumber\\
200: & +& \frac{1}{2} \frac{\Delta n}{n} v_F^2 g \cos(gY-gR\cos(\Omega t))\ .
201: \eea %%\label{e2b} \end{eqnarray}
202: %%\end{mathletters}
203: Apart from its effect for a modulation potential in a low magnetic field,
204: it has been shown to order $\Delta n/n$ that the modulating potential
205: $U(y)$ gives rise to spatially inhomogeneous corrections to the chemical
206: potential and Fermi velocity of the quasiparticles as well as their
207: scattering rates \cite{ten}. For the CF system, $\Delta n/n$ may be
208: treated as a much smaller parameter than $\Delta B/B_{eff}$. Therefore, we
209: neglect the corrections to the CF Fermi velocity in Eqs. (12) and (13) as
210: well as the corrections to the relaxation time.
211:
212: We would like to point out that when some effects of electric modulation
213: are neglected, one misses the possibility of describing an asymmetry in
214: the shape near the minimum of $\rho_\perp$ near $B_{eff}=0$ which was
215: observed in the experiments \cite{three,four}. It was shown by Zwerschke
216: and Gerhardts \cite{seven} that the small asymmetry of $\rho_\perp$ as a
217: function of $B_{eff}$ near $B_{eff}=0$ originates from the effect of
218: interference of the applied electric modulations and induced magnetic
219: modulations. We believe that the observed asymmetry \cite{four} in the
220: shape near the maximum in $\rho_{||}$ is also due to the same reason.
221: However, we do not analyze this effect here to avoid extra complications
222: arising from additional calculations.
223:
224: To lowest order in the modulating field, the corrections $\delta v_x$ and
225: $\delta v_y$ are periodic over the unperturbed cyclotron orbit, as was
226: used in Ref.\ \cite{nine}. With this assumption, we calculate the
227: averages of Eqs. (2) and (3) over the cyclotron orbit. After a
228: straightforward calculation, we obtain the following results for the
229: components $V_x$ and $V_y$ of the velocity of the guiding center
230: % f4-5
231: \bea
232: V_x(Y)&=&\frac{1}{2\pi}\int_0^{2\pi} d\psi\ \delta v_x(Y,\psi) = v_F\cos(gY)
233: \nn \\ \nn \\ & \times &
234: \left[\frac{\Delta B}{B_{eff}}J_1(gR)-\frac{1}{2}gR\frac{\Delta n}
235: {n}J_0(gR) \right] , \\ \nn \\
236: %%%\ee \begin{eqnarray}
237: V_y(Y)&=&\frac{1}{2\pi}\int_0^{2\pi} d\psi\ \delta v_y(Y,\psi)=0 \ ,
238: \label{e3b}
239: \end{eqnarray}
240: where $\psi=\Omega t$, and $J_{0}(gR) $, $J_{1}(gR)$ are Bessel functions
241: of the first kind. We now calculate the CF conductivity by assuming, as
242: before \cite{seventeen}, that $v_x(Y)=v_{x0}+V_x(Y)$ and that the
243: cyclotron frequency $\Omega$ can be replaced by
244: $\Omega(Y)=\Omega+\Delta\Omega(Y)$, where $\Delta\Omega(Y)$ is the
245: correction to the cyclotron frequency due to the inhomogeneous effective
246: magnetic field, averaged over the cyclotron orbit,
247: % f6
248: \bea
249: \Omega(Y)&=&\Omega\left\{1+\frac{\Delta B}{B_{eff}}\frac{1}{2\pi}\int_0^{2\pi}d\psi\
250: \sin(gY-gR\cos\psi) \right\}
251: \nn \\ \nn \\
252: &=& \Omega\left\{1+\frac{\Delta B}{B_{eff}}\sin(gY)J_0(gR)\right\}\ .
253: \label{e4}
254: \eea
255: %+1
256: Within the local limit $ R << \lambda $ the guiding center velocity (3)
257: becomes neiligible, and the conductivity tensor $ \tensor{\sigma}_{cf} (Y)
258: $ has the Drude form with the cyclotron frequency $ \Omega $ replaced by $
259: \Omega (Y).$
260:
261:
262:
263:
264: We now introduce the current density of CFs, averaged over the period of
265: the modulation,
266: % f6
267: \begin{equation}
268: {\bf j}=<{\bf j}(Y)>=\frac{g}{2\pi}\int_{-\pi/g}^{\pi/g}dY\ {\bf j}(Y)
269: \ .
270: \label{e6}
271: \end{equation}
272: Within the local limit ${\bf j}(Y)$ related to a driving electric field by the usual linear relation:
273: % f7
274: \begin{equation}
275: {\bf j}(Y)=\tensor{\sigma}_{cf}(Y)\cdot \left({\bf E}+\Delta{\bf E}(Y)\right)
276: \label{e7}
277: \end{equation}
278: with ${\bf E}$
279: denoting the external electric field and $\Delta{\bf E}(Y)$ the
280: inhomogeneous contribution to the total electric field due to the density
281: modulation, averaged over the cyclotron orbit. To proceed we define the
282: effective CF
283: conductivity by the relation
284: % f8
285: \begin{equation}
286: {\bf j}=\tensor{\sigma}_{cf}\cdot
287: <{\bf E}+\Delta {\bf E}(Y)>
288: . \label{e8}
289: \end{equation}
290:
291: When an electrostatic modulation is applied along the $y$ direction, it
292: affects only the current along the direction of the modulation lines
293: \cite{ten}, so within the geometry chosen here, $j_y$ does not depend on
294: the $y$ coordinate. This result follows from the continuity equation.
295: Consequently, we assume that $j_y$ is independent of $Y$. Also, since
296: $E_x$ does not depend on $Y$, we can obtain closed form analytic
297: expressions for the effective CF conductivity tensor from Eq.\ (\ref{e8}).
298: To get these expressions we first assume that $E_x=0$ and solve for
299: the conductivity to
300: obtain
301: % f9
302: \begin{equation}
303: \sigma_{cf}^{yy}=\left< \frac{1}{\sigma_{cf}^{yy}(Y)} \right>^{-1}\ .
304: \label{e9}
305: \end{equation}
306: Solving again for $E_y$ and assuming that $<E_y(Y)>=0$, we obtain
307: % f10
308: \begin{equation}
309: \sigma_{cf}^{yx}=-\sigma_{cf}^{xy}=
310: \left< \frac{\sigma_{cf}^{yx}(Y)}{\sigma_{cf}^{yy}(Y)} \right>
311: \left< \frac{1}{\sigma_{cf}^{yy}(Y)} \right>^{-1}\ .
312: \label{e10}
313: \end{equation}
314: Finally, substituting the result for $E_y(Y)$ into the expression for the
315: $x$ component of Eq.\ (\ref{e8}), and making use of
316: \[
317: <j_x(Y)>=\left< \frac{\sigma_{cf}^{xy}(Y)}{\sigma_{cf}^{yy}(Y)} \right>j_y\ ,
318: \]
319: we obtain
320: % f11
321: \begin{equation}
322: \sigma_{cf}^{xx}=\left<\sigma_{cf}^{xx}+ \frac{\sigma_{cf}^{xy}(Y)}{\sigma_{cf}^{yy}(Y)}
323: \left(\sigma_{cf}^{yx} -\sigma_{cf}^{yx}(Y) \right)\right> \ .
324: \label{e11}
325: \end{equation}
326: Now we define the effective CF magnetoresistivity
327: $\tensor{\rho}_{cf}$ as $\tensor{\sigma}_{cf}^{-1}$. The above results
328: for the effective conductivity components are valid in the local regime,
329: regardless of the dc magnetotransport experiments. We use this to analyze
330: the experimental data. Assuming that the current flows along the
331: modulation lines ($j_{y} = 0$), we obtain the following result for the
332: magnetoresistivity
333: \begin{equation}
334: \rho_{||}=\rho_{xx}=\left\{<\sigma_{cf}^{xx}(Y)>+
335: \left<\frac{\sigma_{cf}^{xy}(Y)^2}{\sigma_{cf}^{yy}(Y)} \right>
336: \right\}^{-1} ;
337: \label{newe12}
338: \end{equation}
339:
340: Similarly, when current is driven across the modulation lines, and
341: $<j_{x}>=0$ we get the expression for $\rho_{\perp}$ in the form:
342: % f13
343: \begin{equation}
344: \rho_{\perp}=\rho_{yy}=
345: \left<\frac{1}{\sigma_{cf}^{yy}(Y)}\right> -
346: \frac{\displaystyle{
347: \left<\frac{\sigma_{cf}^{xy}(Y)}{\sigma_{cf}^{yy}(Y)}\right>^2}}
348: {\displaystyle{<\sigma_{cf}^{xx}(Y)> +
349: \left<\frac{\sigma_{cf}^{xy}(Y)^2}{\sigma_{cf}^{yy}(Y)} \right>}}\ .
350: \label{newe13}
351: \end{equation}
352: %+2
353: It follows from these results that in the local regime the transverse
354: resistivity $ \rho_\perp $ shows a positive magnetoresistance
355: proportional to $ \big (\Omega \tau \Delta B / B_{eff}\big )^2,$ whereas
356: the longitudinal resistivity remains unchanged at the presence of
357: modulations and takes on its Drude value which agrees with the current
358: theory \cite{ten}.
359:
360: To analyze the influence of modulations on the magnetotransport
361: characteristics within a broader range of magnetic fields $R
362: \stackrel{<}{\sim}\lambda$ we take into account corrections to the usual
363: Drude conductivity originating from the guiding center drift. For this
364: purpose we now introduce the conductivity tensor $ \tensor{\sigma}_{cf}
365: (Y)$ in the form:
366: % f15
367: \begin{equation}
368: \sigma_{cf}^{\alpha\beta}(Y)=\frac{e^2m_c\tau}{2\pi\hbar^2}\sum_{k=-\infty}^{\infty}
369: \frac{v_{k\alpha}(Y)v_{-k\beta}(Y)}{1+ik\Omega(Y)\tau}\ \ ,
370: \label{e5}
371: \end{equation}
372: where $m_c$ is the cyclotron mass of the CF, $\tau$ is the relaxation
373: time, $v_{k\beta}(Y)$ denotes the Fourier series coefficients of the
374: velocity components given by \\\\
375: $ \ds v_{kx}=\frac{v_F}{2}(\delta_{k,1}+\delta_{k,-1})+V_x(Y)\delta_{k,0} \qquad $ and \\
376: $\ds v_{ky}=\frac{iv_F}{2}(\delta_{k,1}-\delta_{k,-1}) $. \\\\
377: %+3
378: Similar expression for the "nonlocal" conductivity was used by Gerhardts
379: \cite{nine} (see Eqs. (2.10), (2.11) of the above paper), and earlier by
380: Beenakker \cite{eight}.
381:
382:
383: Starting from the definition (\ref{e5}), keeping only those terms up to
384: order $(\Delta B/B_{eff})^2$ and using
385: \[
386: \frac{\Delta B}{B_{eff}}\Omega \tau=\frac{\Delta n}{n}\frac{p_F}{\hbar}\ell=
387: \frac{\Delta n}{n}k_F\ell\ ,
388: \]
389: we obtain the following results for the effective CF conductivity components (8)--(10):
390: % f16
391: \bea
392: \sigma_{cf}^{xx}&=&\frac{\sigma_0}{1+(\Omega\tau)^2}\left\{1+\frac{1}{2}
393: \left(\frac{\Delta n}{n}k_F\ell\right)^2\frac{(\Omega\tau)^2}{1+(\Omega\tau)^2} J_0^2(gR)\right\}
394: \nn \\ \nn \\
395: &+& \sigma_0\left( \frac{\Delta B}{B_{eff}}\right)^2
396: \left( J_1(gR)-\frac{g}{2k_F} J_0(gR) \right)^2 ,
397: \\ \nn \\ %%f17
398: \sigma_{cf}^{yx}&=& -\sigma_{cf}^{xy}
399: \nn \\ &=&
400: -\frac{\sigma_0\Omega\tau}{1+(\Omega\tau)^2}
401: \left\{1-\frac{1}{2}
402: \left(\frac{\Delta n}{n}k_F\ell\right)^2\frac{J_0^2(gR)}{1+(\Omega\tau)^2}
403: \right\} , \\ \nn \\
404: %%% \ee % f18 \begin{equation}
405: \sigma_{cf}^{yy}&=&\frac{\sigma_0}{1+(\Omega\tau)^2}\left\{1-\frac{1}{2}
406: \left(\frac{\Delta n}{n}k_F\ell\right)^2\frac{J_0^2(gR)}{1+(\Omega\tau)^2}
407: \right\} ,
408: \label{e16}
409: \eea %%%%quation}
410: where, in this notation, $\sigma_0=ne^2\ell/p_F$ is the Drude
411: conductivity for CFs. The last term in Eq.(16) for
412: $\sigma_{cf}^{xx}$ shows the drift of the guiding center and represents
413: the diffusion of CFs in the $x$ direction. This can be verified by
414: calculating the corresponding $x$ component of the diffusion coefficient $\delta {\cal D}$. Following Ref.\ \cite{eight}, we have
415: \bea % f19
416: \delta {\cal D}&=&\tau<V_x^2(Y)> \nn \\ \nn \\
417: &=&\frac{v_F^2\tau}{2}
418: \left(\frac{\Delta B}{B_{eff}}\right)^2
419: \left(J_1(gR)-\frac{g}{2k_F}
420: J_0(gR)\right)^2 . \nn \\
421: \label{e17}
422: \eea
423: Substituting Eq.\ (\ref{e16}) into the Einstein relation
424: $\sigma_{cf}^{\alpha\beta} = Ne^2{\cal D}_{\alpha\beta}$, where $N$ is the
425: density of states for CFs, we obtain the last term in Eq.\
426: (19).
427:
428: Using our expressions (16)--(18) we
429: may calculate the diagonal components of the electronic resistivity tensor
430: from the CF conductivity tensor $\tensor{\sigma}_{cf}$ given by Eqs. (11) and (12).
431: When the density modulation is very weak, i.e., $\displaystyle
432: {\frac{\Delta n}{n}k_F\ell\ll 1}$, the corrections to the
433: magnetoresistivity due to the density modulation are small and may be
434: neglected. In this case, the nonuniform part of the effective magnetic
435: field does not significantly alter the dc transport. For stronger
436: modulation, i.e., $\displaystyle{\frac{\Delta n}{n}k_F\ell\sim 1}$, the
437: resistivity components are changed considerably. Keeping only the largest
438: contributions when $\Delta B/B_{eff}$ is treated as a small parameter, we
439: obtain the following approximate results for the components of the
440: electron resistivity $\rho_{\perp}=\rho_{yy}$ and $\rho_{||}=\rho_{xx}$
441: corresponding to current flowing perpendicular and parallel, respectively,
442: to the modulation lines.
443: % f20
444: \bea
445: \rho_{\perp}&=&\frac{1}{\sigma_0}\left\{1 + \left(\frac{\Delta
446: n}{n}k_F\ell\right)^2 \left[J_1^2(gR) + \frac{1}{2}J_0^2(gR) \right]\right\} , \nn \\ \\
447: \label{e18}
448: % f21
449: \rho_{||}&=& \frac{1}{\sigma_0}\left\{ 1-\left(\frac{\Delta
450: n}{n}\frac{k_F}{g}\right)^2 (gR)^2 \right.
451: \nn \\ \nn \\
452: &\times & \left.
453: \left[J_1(gR) - \frac{1}{2}\frac{g}{k_F}J_0(gR) \right]^2 \right \} .
454: \label{e19}
455: \eea
456: %+4
457: We arrived at the expressions (20), (21) for magnetoresistivities as a
458: result of a semiquantitative analysis described above. Our analysis is a
459: merely semiquantitative one for it employs the relation (7) in "nonlocal"
460: $(R \stackrel{<}{\sim}\lambda)$ calculations whereas this relation is
461: completely adequate only within the extremely local regime $ (R <<
462: \lambda).$ However, similar considerations were carried out before
463: \cite{eight,nine}, and it was shown \cite{eight,nine,ten} that they yield
464: results which basically agree with those obtained with proper calculations
465: based on the Boltzmann transport equation. Here, we also use the latter to
466: justify our results for resistivity components (see Appendix).
467:
468: It follows from Eqs. (20) and (21) that both resistivities are influenced by the
469: one-dimensional modulation, ruling out the local limit $ R << \lambda. $
470: This disagrees with the results obtained in Refs.\
471: \cite{five,six,seven,eight,nine,ten,eleven}, where it was stated that only
472: one component of the resistivity, i.e., $\rho_\perp$, changes as a result
473: of the modulation. Comparison of our theory with the existing
474: semiclassical theories shows that the difference in the results arises
475: from the difference in definitions of the effective magnetoresistivity.
476: Here, we define $\tensor{\rho}_{cf}$ as $\tensor{\sigma}_{cf}^{-1}$ where
477: $\tensor{\sigma}_{cf}$ is introduced by Eq.\ (\ref{e8}), whereas Menne and
478: Gerhardts \cite{ten} use a different definition for the effective
479: magnetoresistivity, namely:
480: $$
481: \tensor{\rho} <{\bf j}> = {\bf E}.$$
482: In other words, we first introduce the averaged effective conductivity and then we
483: calculate the magnetoresistivity tensor, as an inverse of the effective
484: conductivity. This is in contrast with Refs.\
485: \cite{five,six,seven,eight,nine,ten,eleven} where the average was carried
486: out last \cite{eighteen}.
487:
488: Our point to justify the averaging procedure based on the definition (7)
489: is that the expressions for transport coefficients obtained either with
490: quantum mechanical or with classical calculations have to be consistent at
491: low magnetic fields where the cyclotron quantum $ \hbar \Omega $ is minute
492: compared to the Fermi energy of the system. Quantum mechanical
493: calculations of the magnetoconductivity
494: \cite{twelve,thirteen,fourteen,fifteen} insert summation over quantum
495: numbers labeling eigenstates of the unperturbed (homogeneous) 2DEG in an
496: external magnetic field, including the guiding center coordinate $ Y.$ In
497: the semiclassical limit the resultant conductivity passes into a classical
498: conductivity tensor averaged over the period of modulations, therefore the
499: latter is an accurate semiclassical analog of the conductivity calculated
500: within the proper quantum mechanical approach. Our definition of the
501: effective conductivity (9) agrees with that, so it is correct. A similar
502: definition was already used to analyze the response of a modulated 2DEG to
503: the surface acoustic wave \cite{sixteen}. On the contrary, the approach of
504: \cite{five,six,seven,eight,nine,ten,eleven} is mistaken. Direct
505: calculations of the averaged magnetoresistivity components proposed by
506: Beenakker \cite{eight} and further developed by Menne and Gerhardts
507: \cite{ten} give results which disagree with the corresponding expressions
508: obtained within the semiclassical (low field) limit of quantum mechanical
509: calculations \cite{thirteen}. This difference is insignificant for the
510: component $ \rho_\perp$ but it is crucial for $ \rho_{||}.$
511:
512:
513:
514: Our results for $\rho_{\perp}$ and $\rho_{||}$ are different from those
515: obtained in \cite{seventeen}. The reason is that here we averaged the CF
516: conductivity based on the definition in Eq.\ (\ref{e8}) consistent with
517: the continuity equation, whereas in Ref.\ \cite{seventeen} the
518: conductivity $\sigma_{cf}$ was obtained as a simple spatial average of
519: $\sigma_{cf}(Y)$. This procedure led to invalid results for the
520: resistivity component $\rho_{\perp}$ which did not compare well with the
521: experimental results.
522:
523:
524: In comparing the results of our paper in Eqs.(20) and
525: (21) with the experimental results in Ref.\cite{four}, we note
526: that these expressions cannot be applied for filling factors close to
527: $\nu=1/2$ because they are only valid when $\Delta B/B_{eff}\ll
528: 1$. However, we may use them when the filling is not close to
529: $\nu=1/2$ in order to analyze the dependence of the resistivity on
530: magnetic field. When $ gR \sim 1$ and $\displaystyle{\frac{\Delta
531: n}{n}k_F\ell\sim 1}$, the correction to the resistivity $\rho_\perp$ is of
532: the order of unity and increases with the effective magnetic field
533: $B_{eff}$ with a minimum at $B_{eff}=0$ corresponding to
534: $\nu=1/2$, as has been observed experimentally \cite{three,four}.
535: For the range of magnetic field when the effective magnetic field is not
536: small, our theory gives good qualitative agreement with that in Refs.\
537: \cite{five,six,seven}. Also, our Eq. (20) gives results for
538: $\rho_{||}$ in qualitative agreement with experiment over a range of
539: effective magnetic fields corresponding to $2<gR<4$, unlike Refs.\
540: \cite{five,six,seven}. The magnitude of $\rho_{||}$ is smaller than
541: $\rho_{\perp}$.
542:
543:
544: \section{III. Numerical Results}
545: \label{sec3}
546:
547: We compare our numerical results obtained from Eqs.\ (\ref{e18}) and
548: (\ref{e19}) with the experimental data by choosing the following values in
549: our numerical calculations, $n=1.1\times 10^{15}m^{-2},\ \ p_F=1.2\times
550: 10^{-26}\ kg.m/s, \ \lambda=0.5\ \mu m,\ \ell=1.0\ \mu m$ and $\Delta
551: n/n=0.025$. Plots of $\rho_{xx}$ and $\rho_{yy}$ as functions of $B_{eff}$
552: are presented in Fig. 1.
553: \begin{figure}[t]
554: \begin{center}
555: \includegraphics[width=9.0cm,height=10cm]{cf.eps}
556: \caption{
557: Plots of $\rho_{\perp}=\rho_{yy}$ and $\rho_{||}=\rho_{xx}$ in
558: units of the Drude resistivity $\sigma_0^{-1}$ in zero magnetic field, as
559: functions of the effective magnetic field $B_{eff}$ for $\Delta
560: n/n=0.025$. Here, $n=1.1\times 10^{15}m^{-2},\ \ p_F=1.2\times 10^{-26}\
561: kg.m/s$ $\lambda=0.5\ \mu m$ and $\ell=1.0\ \mu m$.}
562: \label{rateI}
563: \end{center}
564: \end{figure}
565: The oscillations in both $\rho_{xx}$
566: and $\rho_{yy}$ are due to the density modulation and depend on the
567: wavelength $\lambda$. The ratio $\rho_{xx}/\rho_{yy}$ for the range of
568: magnetic fields shown in Fig. 1 is in good agreement with
569: experiment \cite{four}. In the immediate vicinity of $\nu= 1/2 $,
570: where the condition $\frac{\Delta B}{B_{eff}}<1$ is not satisfied, our
571: theory breaks down because the magnetic field dependence is strongly
572: influenced by the channeled orbits of CFs which cannot be included in this
573: formalism.
574:
575:
576:
577:
578:
579: \section{IV. Summary and Concluding Remarks}
580: \label{sec4}
581:
582: In this paper, we use a quasiclassical theory based on the Beenakker
583: approximation for dc magnetotransport in a modulated quantum Hall system
584: near filling factor $\nu=1/2$. Assuming that a weak
585: one-dimensional electrostatic potential acts on the two-dimensional
586: electron gas, we obtain closed form analytic expressions for the
587: resistivity $\rho_\perp$ corresponding to a current at right angles to the
588: direction of the modulation lines as well as a smaller component
589: $\rho_{||}$ for a current along the direction of the modulation lines.
590: Numerical results are presented for $\rho_\perp$ and $\rho_{||}$ and show
591: some of the features observed experimentally. Our analytic results are not
592: valid at filling factors too close to $\nu=1/2$ because our
593: approximation scheme assumes that $\Delta B/B_{eff}$ is a small parameter
594: which can be applied as a perturbation parameter. A completely different
595: approach must be used for filling factors nearer $\nu=1/2$ and
596: will be considered elsewhere.
597:
598: Finally, we point out that expressions similar to our results for the
599: magnetoresistivity in Eqs. (20) and (21) can also be used
600: for a semiquantitative analysis of dc madnetotransport in a modulated 2DEG
601: in a weak nonquantizing magnetic field when the modulation is magnetic in
602: nature. This enables us to explain qualitatively the so-called
603: ``antiphase" oscillations of the magnetoresistivity $\rho_{||}.$ It was
604: observed in experiments \cite{twelve} and confirmed with the calculations
605: based on quantum mechanical approach \cite{thirteen} that both
606: magnetoresistivity components of a 2DEG in a one-dimensional lateral
607: superlattice show oscillations of the same period at low magnetic fields
608: where quantum oscillations of the electron density of states at the
609: Fermi surface (DOS) are negligible. Oscillations of $ \rho_\perp$ are
610: identified as a commensurability effect (Weiss oscillations) which is also
611: described within a semiclassical approach. As for antiphase oscillations
612: of $ \rho_{||}$, the current semiclassical theory is enable to explain
613: them, therefore it is proposed \cite{twelve,thirteen,fourteen,fifteen}
614: that these oscillations are dominated by quantum mechanism, namely, by
615: modulation-induced amplitude oscillations of DOS. The above explanation is
616: hardly correct. It is true that at the presence of modulations quantum
617: oscillations of DOS are superimposed with the commensurability
618: oscillations. The same effect can be observed in conventional metals, as
619: it was shown before \cite{eighteen}. However, at low magnetic fields the
620: quantum correction to DOS goes to zero, modulated or not. Only effects
621: originating from classical mechanisms survive within this semiclassical
622: limit. Besides, it is well known that periods of quantum oscillations of
623: DOS and semiclassical commensurability oscillations differ, and their
624: ratio is of the order of $\hbar g/p_F$ where $ p_F $ is the Fermi
625: momentum of the considered system. Based on coincidence of the periods of
626: low field oscillations of both magnetoresistivity components we conclude
627: that they have the same nature and origin. So, we treat the low-field
628: oscillations of $ \rho_{||} $ as semiclassical commensurability
629: oscillations which are misinterpreted as a quantum effect by the existing
630: theory. Our results (19) and (39) confirm this conclusion. We do not
631: believe that this contradicts the results of \cite{thirteen} based on
632: quantum calculations. As in the case of Weiss oscillations, consistent
633: results can be obtained within a classical approach as well as a
634: semiclassical limit of quantum calculations, and we believe that our
635: approach present a correct qualitative semiclassical description of the
636: antiphase oscillations of $\rho_{||}, $ restoring the self-consistence
637: of the of the theory of magnetotransport in modulated 2D electron systems.
638:
639:
640: {\it \bf Acknowledgments:}
641: NAZ thanks G.M. Zimbovsky for help with the manuscript.
642: GG acknowledges the support in part from a NATO
643: Grant \# CRG-972117 (U.S. - U.K. Collaborative Grant), the City University
644: of New York PSC-CUNY grants \#664279 and \#669456 as well as grant \#
645: 4137308-02 from the NIH.
646: JLB acknowledges support in part from an "in service" grant
647: from the PSC-CUNY research program.
648:
649:
650:
651: \section{V. appendix}
652: \label{sec5}
653:
654: In this Appendix, we present a derivation of the expressions for the
655: components of the magnetoresistivity based on the Boltzmann transport
656: equation. Our goal is to show that both $\rho_{||}$ and $\rho_\perp$ can
657: be affected by weak density modulations along the $y$ direction with
658: $\Delta n(y)=\Delta n\ \sin(gy)$, where $g$ is a constant. For simplicity,
659: we neglect a small direct effect due to the screened electric modulating
660: potential $U(y)$ on the response functions and we concentrate on the
661: effect of modulations of the effective magnetic field.
662:
663: The CF current density can be written as
664: \begin{equation} %%f22
665: {\bf j}(y)=Ne^2\int_0^{2\pi}\frac{d\psi}{2\pi}\ {\bf v}(\psi)
666: \Phi(y,\psi)\ ,
667: \label{ae1}
668: \end{equation}
669: where $N$ is the CF density of states on their Fermi surface and the
670: distribution function $\Phi(y,\psi)$ satisfies the linearized Boltzmann equation
671: \begin{equation} %f23
672: v_y\frac{\partial \Phi}{\partial y}+(\Omega+\Delta\Omega(y))
673: \frac{\partial \Phi}{\partial \psi}+C[\Phi]={\bf E}\cdot{\bf v}\ .
674: \label{ae2}
675: \end{equation}
676: Here, ${\bf E}$ is an external electric field. The collision integral
677: $C[\Phi]$ in our calculations below is taken in a relaxation time
678: approximation, with the relaxation towards the local equilibrium
679: distribution function, i.e.,
680: \begin{equation} %f24
681: C[\Phi]=\frac{1}{\tau}\left[\Phi(y,\psi)-\int_0^{2\pi}
682: \frac{d\psi}{2\pi}\ \Phi(y,\psi) \right] \ .
683: \label{ae3}
684: \end{equation}
685: The average of this $C[\Phi]$ over $\psi$ vanishes. As a result, the
686: current density in Eq.\ (\ref{ae1}) has to satisfy the continuity
687: equation.
688:
689: To proceed, we separate from the distribution function the homogeneous
690: term $\Phi(\psi)$ which describes the linear response of the CF system in
691: the absence of modulations. For this term, we use the expression given in
692: \cite{ten}, i.e.,
693: \begin{equation} %f25
694: \Phi(\psi)=\rho_0\tau{\bf v}\cdot{\bf j}_0\ ,
695: \label{ae4}
696: \end{equation}
697: where $\rho_0$ is the Drude resistivity ($\rho_0=\frac{1}{\sigma_0}$) and
698: ${\bf j}_0$ is the CF current density for the unmodulated system.
699:
700: Expressing $\Phi(y,\psi)$ as
701: \begin{equation} %f26
702: \Phi(y,\psi)=\Phi(\psi)+\rho_0\tau \chi(y,\psi)\ ,
703: \label{ae5}
704: \end{equation}
705: we obtain the following transport equation
706: \begin{equation} %f27
707: v_y\frac{\partial \chi}{\partial y}+(\Omega+\Delta\Omega(y))
708: \frac{\partial \chi}{\partial \psi}+C[\chi] =
709: \Delta\Omega(y)
710: \left(v_yj_x^0-v_xj_y^0 \right) \ .
711: \label{ae6}
712: \end{equation}
713:
714: We now expand $\chi(y,\psi)$ as a Fourier series in its spatial variable, i.e.,
715: \begin{equation} %28
716: \chi(y,\psi)=\chi_0(\psi)+\sum_{n=1}^\infty \left(
717: \chi_n e^{igny}+\chi_n^\ast e^{-igny} \right)\ .
718: \label{ae7}
719: \end{equation}
720: This leads to a system of equations for the Fourier components $\chi_n(\psi)$ given by
721: \bea %f29
722: \frac{\partial \chi_0}{\partial \psi}&+&\frac{1}{\Omega}C[\chi_0]= i\frac{\Delta\Omega}{2\Omega}\left(\frac{\partial\chi_1^\ast}
723: {\partial \psi}-\frac{\partial\chi_1}{\partial\psi} \right)
724: , \\ \nn \\ %% \label{ae8} \end{equation}
725: %f30
726: igR\sin(\psi)\chi_1 &+& \frac{\partial\chi_1}{\partial\psi}+\frac{1}{\Omega}
727: C[\chi_1]=\frac{i}{2}\frac{\Delta\Omega}{\Omega}\left(v_xj_y^0-
728: v_yj_x^0 \right) \nn \\ \nn \\ &+&
729: \frac{i}{2}\frac{\Delta\Omega}{\Omega}
730: \left(\frac{\partial\chi_0}{\partial\psi}
731: - \frac{\partial\chi_2}{\partial\psi} \right) .
732: \label{ae9}
733: \eea
734: The equations determining the Fourier components $\chi_n$ for $n\geq 2$ are given by
735: \begin{equation} %f31
736: ingR\
737: \chi_n+\frac{\partial\chi_n}{\partial\psi}+\frac{1}{\Omega}C[\chi_n]
738: =i\frac{\Delta\Omega}{2\Omega}\left(
739: \frac{\partial\chi_{n-1}}{\partial\psi}-\frac{\partial\chi_{n+1}}{\partial\psi} \right)\ .
740: \label{ae10}
741: \end{equation}
742:
743: It follows from Eqs. (29)-(\ref{ae10}) that $\chi_1$ and
744: $\chi_1^\ast$ are of the order of $\Delta B/B_{eff}$, whereas
745: $\chi_0,\chi_2$ and $\chi_2^\ast$ are of order $(\Delta B/B_{eff})^2$ and
746: all Fourier components with $n>2$ are of order $(\Delta B/B_{eff})^3$ or
747: smaller. Therefore, retaining in the expansion of $\chi(y,\psi)$ those
748: terms no less than the square of $\Delta B/B_{eff}$, we can omit all terms
749: with $n>2$ in the Fourier series (\ref{ae7}). Then, looking for the
750: solutions of (\ref{ae8})-(\ref{ae10}) in the limit when impurity
751: scattering is weak, i.e., $\Omega\tau\gg 1$, we obtain the main
752: approximations for the desired Fourier components, namely,
753: \bea %f32
754: \chi_0(\psi)&=& \left(\frac{\Delta\Omega}{\Omega}\right)^2\sin(gR\cos\psi)\ QE_x \ ,
755: \\ \nn \\
756: \chi_1(\psi)&=& \frac{\Delta\Omega}{\Omega}\exp(igR\cos\psi)\
757: QE_x \ ,
758: \\ \nn \\
759: \chi_2(\psi)&=&-\frac{i}{2}\left(\frac{\Delta\Omega}{\Omega}\right)^2\exp(igR\cos\psi)\ QE_x n
760: \nn \\ \nn \\ &+&
761: \frac{i}{2}\left(\frac{\Delta\Omega}{\Omega}\right)^2\exp(2igR\cos\psi)\ SE_x \ ,
762: \\ \nn \\
763: Q & =&\frac{v_F}{2}\sigma_0\Omega\tau\frac{J_1(gR)}{1-J_0^2(gR)}
764: \ , \\ \nn \\
765: S&=&\frac{v_F}{2}\sigma_0\Omega\tau\frac{J_0(gR)J_1(gR)}
766: {(1-J_0^2(gR))(1+J_0(2gR))} \ .
767: \eea
768: Substituting the derived results in Eqs. (32)-(34) into
769: the Fourier expansion (\ref{ae7}), we obtain the distribution function
770: $\chi(y,\psi)$ which we use to calculate the CF current density for a
771: modulated system.
772:
773:
774: We see that only the $j_x$ component has an extra term due to the
775: modulation, whereas $j_y$ remains unchanged and does not depend on the $y$
776: coordinate. This agrees with the continuity equation and is a motivation
777: for us to calculate the averaged CF current density in a simple way:
778: \begin{equation} %%f37
779: <{\bf j}(y)>=<\tensor{\sigma}_{cf}(y)>\cdot {\bf E}\ .
780: \label{ae15}
781: \end{equation}
782: We obtain
783: \begin{equation} %%f38
784: \sigma_{cf}^{xx}= <\sigma_{cf}^{xx}(y)>=\frac{\sigma_0}{1+(\Omega\tau)^2} +
785: \sigma_0\left(\frac{\Delta B}{B_{eff}} \right)^2\frac{J_1^2(gR)} {1-J_0^2(gR)}\ .
786: \label{ae16}
787: \end{equation}
788: The second term in (\ref{ae16}) represents the diffusion of the CFs along
789: the $x$ direction due to the drift of the guiding center. Other components
790: of the CF conductivity tensor remain unchanged due to the modulation and
791: we obtain the following expressions for the electron magnetoresistivity:
792: \bea %f39
793: \rho_{yy}=\rho_\perp\approx\frac{1}{\sigma_0}\left\{1+\left(
794: \frac{\Delta n}{n}k_Fl\right)^2\frac{J_1^2(gR)}{1-J_0^2(gR)} \right\} , &&
795: \label{ae17}
796: \\ \nn \\
797: \rho_{xx}=\rho_{||}\approx\frac{1}{\sigma_0}\left\{1-\left(
798: \frac{\Delta n}{n} k_FR\right)^2\frac{J_1^2(gR)}{1-J_0^2(gR)} \right\} .&&
799: \label{ae18}
800: \eea
801: %+5
802: These results (38), (39) are obtained using the main approximation
803: in the expansion of the distribution function
804: $\chi (y, \psi)$ in the expansion of the distribution function $ \chi
805: (y,\psi) $ in the inverse powers of the parameter $ \Omega \tau >> 1. $
806: Taking into account higher order terms in this expansion does not change
807: the main result, namely, that both resistivities are influenced by the
808: modulations. Corresponding calculations are lengthy, and we do not present
809: them here.
810:
811: So, our analysis based on the Boltzmann equation gives results
812: which do not contradict our semiquantitative approach which we employed in
813: Section 2. We have shown that both magnetoresistivity components
814: are affected by the modulation, although the effect on $\rho_\perp$ is
815: larger than on $\rho_{||}$. Apart from the denominator $(1-J_0^2(gR))$,
816: the first terms in the expressions for the corrections due to modulation
817: in our results (20) and (\ref{e19}) coincide with the corrections
818: in the expressions (\ref{ae17}) and (\ref{ae18}).
819:
820: We now return to the discrepancy between our results and those in Refs.
821: \cite{five,six,seven,eight,nine,ten,eleven}. We use of the definitions in
822: \cite{ten} for the effective magnetoresistivity tensor
823: $\tensor{\rho}_{eff} \cdot<{\bf j}>={\bf E}$ and their result
824: $\tensor{\rho}_{{\cal D}} {\bf j}={\bf E}+{\bf \Delta}$, where
825: $\tensor{\rho}_{{\cal D}}$ is the Drude resistivity tensor and the vector
826: ${\bf \Delta}$ is (see \cite{ten})
827: \begin{equation}
828: \Delta_{x,y}=\mp\frac{2\rho_0\tau}{v_F^2}\left<\int_0^{2\pi}
829: \frac{d\psi}{2\pi} \ \Delta\Omega(y)v_{y,x}\chi(y,\psi)\right>\ .
830: \label{ae19}
831: \end{equation}
832: We conclude that for weak modulation, when $<{\bf j}>\approx{\bf j}_0$, we have
833: \begin{equation}
834: \left(\tensor{\rho}_{{\cal D}}-\tensor{\rho}_{eff} \right)\cdot
835: {\bf j}_0={\bf \Delta}\ .
836: \label{ae20}
837: \end{equation}
838: This means that the components of the vector ${\bf \Delta}$ are linear
839: combinations of the components of the corrections of the linear
840: resistivity tensor. As in the local limit, the effective resistivity
841: tensor is directly calculated in \cite{ten}, avoiding calculations of the
842: effective conductivity first.
843:
844: Substituting our expression for $\chi(y,\psi)$ into (\ref{ae19}), we can
845: obtain ${\bf \Delta}$, and then the components of the electron
846: magnetoresistivity. The results corroborate the corresponding results of
847: \cite{ten} and we arrive at their conclusion that only $\rho_\perp$ is
848: affected by the modulation. This confirms our conclusion that the root of
849: the disagreement between our results and those of
850: \cite{five,six,seven,eight,nine,ten,eleven} is in the different ways of
851: averaging the response functions over the period of the modulation.
852:
853:
854:
855:
856: %\vskip 0.2in
857:
858:
859: \begin{references}
860:
861:
862: \bibitem{one} B.I. Halperin, P.A. Lee, and N.A. Read, Phys.
863: Rev. B {\bf 47}, 7312 (1993).
864:
865: \bibitem{two} B.I. Halperin in ``{\em Low-Dimensional Semiconductor
866: Structures},'' edited by . Das Sarma and A. Pinczuk (Wiley, NY, 1996).
867:
868: \bibitem{three} J.H S, K. von Klitzing, D. Weiss, and W. Wegschneider,
869: Phys. Rev. Lett. {\bf 80}, 4538 (1998).
870:
871: \bibitem{four} R.W. Willett, K.W. West, and L.N. Pfeiffer,
872: Phys. Rev. Lett. {\bf 83}, 2624 (1999); see also the review article by
873: R.W. Willett, Advances in Physics, {\bf 46}, 447 (1997).
874:
875: \bibitem{five} F. von Oppen, A. Stern, and B.I. Halperin, Phys. Rev.
876: B {\bf 80}, 4434 (1998).
877:
878: \bibitem{six} A.D. Merlin, P. W\"olfle, Y. Levinson, and O. Entin Wohlman,
879: Phys. Rev. Lett. {\bf 81}, 1070 (1998).
880:
881: \bibitem{seven} S.D.U. Zwerscke and R.R. Gerhardts, Phys. Rev. Lett.
882: {\bf 83}, 2616 (1999).
883:
884: \bibitem{eight} C.W.J. Beenakker, Phys. Rev. Lett. {\bf 62}, 2020,
885: (1989).
886:
887: \bibitem{nine} R.R. Gerhardts, Phys. Rev. B {\bf 53}, 11064 (1996).
888:
889: \bibitem{ten} R. Menne and R.R. Gerhardts, Phys. Rev. B {\bf 57}, 1707
890: (1998).
891:
892: \bibitem{eleven} A.D. Mirlin and P. Wolfle, Phys. Rev. B {\bf 58}, 12986
893: (1998).
894:
895: \bibitem{twelve} D.Weiss, K.v. Klitzing, K. Ploog and G. Weimann,
896: Europhys. Lett {\bf 8}, 179 (1989).
897:
898: \bibitem{thirteen} C. Zhang and R.R. Gerhardts, Phys. Rev. B {\bf 41},
899: 12850, (1990).
900:
901: \bibitem{fourteen} A. Manolescu, R.R. Gerhardts, M. Suhrke and U. Rossler,
902: Phys. Rev. B {\bf 63}, 115322 (2001).
903:
904: \bibitem{fifteen} J. Grob and R.R. Gerhardts, Phys. Rev. B {\bf 66},
905: 155321 (2002).
906:
907: \bibitem{sixteen} Y. Levinson, O. Entin-Wohlman, A.D. Mirlin and P.
908: Wolfle, Phys. Rev. B {\bf 58}, 7113 (1998).
909:
910: \bibitem{seventeen} N.A. Zimbovskaya and J.L. Birman, Phys. Rev. B {\bf
911: 60}, 12174 (1999), Solid State Comm. {\bf 116}, 21 (2000).
912:
913: \bibitem{eighteen} There is one exception, namely the paper of
914: A.D. Mirlin and P. Wolfle \cite{eleven}, where the authors claim that they
915: first introduce the averaged effective conductivity, which seems to be the
916: same approach as adopted in our work. However, the comparison of
917: \cite{eleven} with the work of Menne and Gerhardts \cite{ten} shows that
918: the quantity which is claimed there to be the effective conductivity $
919: \tensor \sigma, $ actually is the different quantity, namely, an inverse
920: of the effective magnetoresistivity tensor introduced in \cite{ten}.
921:
922: \bibitem{nineteen}
923: See e.g. N. Zimbovskaya, Local Geometry of the Fermi
924: Surface and High-Frequency Phenomena in Metals, Ch. 7. Springer-Verlag,
925: New York, 2001.
926:
927:
928:
929: \end{references}
930:
931:
932:
933: \vspace{2mm}
934:
935:
936:
937:
938:
939:
940:
941:
942: %\end{multicols}
943:
944:
945: \end{document}
946:
947:
948:
949:
950:
951:
952:
953:
954:
955:
956:
957:
958:
959:
960: