1: %\documentclass[preprint,showpacs,aps,prl,superscriptaddress]{revtex4}
2: \documentclass[twocolumn,showpacs,aps,prl,superscriptaddress]{revtex4}
3:
4: \usepackage{graphicx}
5: \usepackage{amsmath}
6: \usepackage{latexsym}
7:
8: \begin{document}
9:
10: \title{Negative Differential Resistance and Astability of the
11: Wigner Solid}
12:
13: \author{G.A. Cs\'athy}
14: \affiliation{Department of
15: Electrical Engineering, Princeton University, Princeton, NJ 08544}
16: %\email{gcsathy@princeton.edu}
17:
18: \author{D.C. Tsui}
19: \affiliation{Department of
20: Electrical Engineering, Princeton University, Princeton, NJ 08544}
21:
22: \author{L.N. Pfeiffer}
23: \affiliation{Bell Labs, Lucent Technologies, Murray Hill, NJ 07974}
24:
25: \author{K.W. West}
26: \affiliation{Bell Labs, Lucent Technologies, Murray Hill, NJ 07974}
27:
28: \date{\today}
29:
30: \begin{abstract}
31: We report an unusual breakdown of the magnetically induced Wigner solid
32: in an exceptional two-dimensional electron gas.
33: The current-voltage characteristic is found to be hysteretic
34: in the voltage biased setup and has a region of negative differential
35: resistance in the current biased setup. When the sample is
36: current biased in the region of negative differential resistance,
37: the voltage on and the current through the sample develop
38: spontaneous narrow band oscillations.
39: %The properties of these oscillations are determined by both the sample
40: %as well as the external circuitry and can be fully explained by a simple model.
41:
42: \end{abstract}
43: \pacs{73.40.-c, 73.20.Qt, 73.63.Hs}
44: \maketitle
45:
46:
47: Two-dimensional electron gases (2DEG) in perpendicular magnetic
48: fields $B$ are fascinating systems in which
49: strong electron-electron interactions lead to the formation of
50: numerous many body ground states. One well known example is the family of
51: over 50 fractional quantum Hall liquids \cite{pan}.
52: Electron solids constitute another class of collective ground states.
53: %Early transport measurements \cite{early,iv,williams}
54: %on the high field insulating phase (HFIP) and
55: %the reentrant insulating phase (RIP) have associated these phases
56: %with the magnetic field induced Wigner solid (WS) \cite{wigner}.
57: %Recent microwave response data \cite{williams,lloyd,lloyd2} provided strong
58: %evidence for this association.
59: The high field insulating phase (HFIP), the reentrant insulating phase (RIP)
60: \cite{early,iv,williams,lloyd,lloyd2},
61: and solids forming close to integer Landau level filling factors
62: $\nu$ \cite{chen} have been identified with magnetic field
63: induced Wigner solids (WS) \cite{wigner}.
64: More exotic solids such as the reentrant integer quantum Hall states (RIQHS)
65: and the electronic liquid crystals of high Landau levels (LL) \cite{lilly}
66: are believed to be realizations of the theoretically predicted bubble and
67: stripe phases \cite{koulakov}, respectively.
68:
69: The RIQHS shares many properties with the solids enlisted
70: above \cite{iv,williams,lloyd,lloyd2,chen}
71: % and with conventional charge density waves (CDW) \cite{gruner}
72: such as non-linear dc current-voltage characteristics ($I$-$V$)
73: with a well defined threshold voltage \cite{jim-iv} and
74: sharp resonances in the microwave conductivity \cite{rupert}.
75: The RIQHS, however, appears to be unique among the
76: $B$-field induced solids since it is the only one
77: exhibiting spontaneous narrow-band oscillations under dc biasing \cite{jim-osc}.
78: The nature of these oscillations remains unexplained to date.
79: Because of the increasing oscillation frequency with an increasing bias
80: current \cite{jim-osc}, the oscillations in the RIQHS were suggested to
81: be akin to the washboard oscillations found in conventional charge
82: density waves (CDW) \cite{fleming}.
83: The frequencies of the RIQHS of the order of 1kHz
84: are, however, orders of magnitude lower than the expected
85: washboard frequencies \cite{jim-osc}.
86:
87: In this Letter we report the observation of spontaneous narrow band oscillations
88: and of negative differential resistance (NDR) in the WS that forms at the highest
89: $B$-fields, i.e. the HFIP and the RIP. We used an exceptional
90: 2DEG confined to a GaAs/AlGaAs quantum well and
91: which has an areal density of $4.8 \times 10^{10}$cm$^{-2}$
92: and a mobility of $13 \times 10^{6}$cm$^2$/Vs.
93: Unlike the RIQHS, the WS conducts through one single channel
94: as all electrons occupy the lowest LL. This property
95: enables us to construct a simple model that explains our observations.
96: We show that the oscillations arise from an astable
97: behavior of the sample with hysteretic $I$-$V$ connected to the external
98: circuitry and that the NDR is a direct consequence of these oscillations.
99:
100: \begin{figure}[!b]
101: \begin{center}
102: \includegraphics[width=2.7in,trim=0.0in 0.0in 0.0in 0.1in]{cFig1.eps}
103: \end{center}
104: \caption{\label{f1}
105: Panel a shows the diagonal resistance $R_{xx}$ as a function of $B$-field
106: at 38mK. Lower panels show the $I$-$V$ and the measurement circuitry
107: in the voltage (panel b) and the current biasing setup (panel c)
108: when the sample $S$ is at 54mK and at 11T. The curve of
109: decreasing current (full symbols) of panel c is offseted by 25nA.
110: Insets show a magnified view of the low current region.
111: Branch $\beta$ of panel c exhibits NDR.
112: }
113: \end{figure}
114:
115: Figure.1a. shows the diagonal resistance $R_{xx}$ of our sample at $T=38$mK
116: measured in a four wire configuration using a small-signal ac excitation.
117: The hashed regions at $\nu<1/5$ and $1/5<\nu<2/9$ mark the HFIP and the RIP.
118: Resonances in the microwave spectrum recently reported in samples cut
119: from the same wafer as ours \cite{lloyd2} are considered the most direct
120: evidence that the RIP and the HFIP is identical to the WS \cite{williams,lloyd,lloyd2}.
121: We perform two-lead dc $I$-$V$ measurements of the RIP and HFIP.
122: As customary in dc measurements, the preamplifiers used are followed
123: by integrators with their time constant set to 0.3s.
124: We thus record $I_{dc}$ and $V_{dc}$, the temporal averages of $I$ and $V$.
125: Two different biasing configurations are utilized on the same ohmic contacts.
126: Unless explicitly specified, the sample is held at 54mK and at 11T.
127:
128: We first employ a voltage biasing configuration shown in Fig.1b.
129: The $I$-$V$ of the $V$-biased sample
130: is non-linear and it exhibits a striking hysteretic behavior.
131: Previously measured $I$-$V$s of the WS
132: were also found non-linear \cite{williams,iv} but,
133: with the exception of one curve from Ref.\cite{williams},
134: none of them showed hysteresis.
135: Perhaps the most intriguing feature of the $I$-$V$ is its simplicity.
136: Indeed, the $I$-$V$ has two linear branches. As the voltage is increased we
137: find that the current stays virtually zero as long as the voltage
138: does not exceed the value at the threshold point $T$.
139: %The resistance of this branch is beyond our measuring capability of tens of G$\Omega$.
140: As the voltage is further increased electrical breakdown occurs when
141: the current suddenly jumps to the upper branch that carries non-zero current.
142: If the $V$ is decreased while the bias point is on the upper branch,
143: the current decreases but it jumps to
144: zero at point $U$ at a voltage lower than that of point $T$.
145: The current at voltages between that of points $U$ and $T$ is not single valued.
146:
147:
148: The second measurement setup, shown in Fig.1c, is the current biased
149: configuration. The $I$-$V$ obtained has a branch that does not support
150: current, one that exhibits an unexpected NDR,
151: and one that is at an angle and it is linear to a good approximation.
152: These branches are labeled in the inset of Fig.1c by
153: $\alpha$, $\beta$, and $\gamma$, respectively.
154: We thus report for the first time NDR in the WS at the highest $B$-fields.
155: % This is the first time NDR has been observed in
156: % how about B=0 ?
157: % insulators of the LL.
158: When superimposed (not shown), the linear branches of the two
159: $I$-$V$s taken with the two different circuits perfectly overlap.
160: The two $I$-$V$s, however, differ in two ways.
161: One difference is the NDR is not present in the $V$-biasing setup.
162: The absence of hysteresis of the $I$-biased configuration,
163: also shown in Fig.1c, is a second major difference.
164: The $I$-$V$ of Fig.1c belongs to the so called
165: S-shaped curves and the NDR is said to be current controlled \cite{sze}.
166:
167: % It is thought that there is no general mechanism that leads to
168: % NDR. The causes of NDR thus have to be examined in each case separately.
169: To understand the origin of the NDR shown in Fig.1c
170: we explore the time dependence of the voltage on and
171: current through the sample. We use the same $I$-biased circuit
172: but we disable the integrators. The bandwidth for $V$ and $I$
173: is limited by the preamplifiers to 700kHz and 8kHz, respectively.
174: In Fig.2a we show the $V$-waveforms corresponding to
175: bias points \#1, \#2, and \#3 of the $I$-$V$ of Fig.2b, an $I$-$V$ which
176: is identical to that of Fig.1c.
177: We find that at bias points \#1 and \#3 the
178: voltage on the sample does not change with time.
179: In fact, if the bias point is anywhere on the linear branches $\alpha$ or
180: $\gamma$ there is a steady state developed.
181: In contrast, when the sample is biased in the NDR region, for example
182: bias point \#2, there are spontaneous $V$-oscillations present on the sample.
183: As seen in Fig.2c, these $V$-oscillations are accompanied by
184: synchronized oscillations of the current through the sample.
185: Since biasing on the branch of the $I$-$V$ with NDR cannot support
186: a steady current flow we conclude that this branch has to be understood as
187: a result of the temporal averaging of the oscillations developed.
188: Oscillations and NDR are observed
189: in our sample for all pairs of contacts but only in the $B$-field induced WS,
190: i.e. the HFIP and RIP shown in Fig.1a,
191: we therefore think that they are due to an intrinsic bulk effect rather
192: than a contact effect.
193: Using the other biasing configuration, i.e. $V$-biasing,
194: we detect no oscillations.
195:
196: \begin{figure}[!t]
197: \begin{center}
198: \includegraphics[width=2.7in,trim=0.0in 0.0in 0.0in 0.1in]{cFig2.eps}
199: \end{center}
200: \caption{\label{f2}
201: Voltage waveforms (panel a) at three different bias points
202: of the $I$-biased $I$-$V$ (panel b). Note the break in the $I$-axis
203: for panel b. At 4.5nA bias (bias point \#2)
204: both $V$ and $I$ oscillate (panel c).
205: Spectra of these oscillations are shown in panel d.
206: }
207: \end{figure}
208:
209: As seen in Fig.2, the $V$ and $I$ waveforms are quite different but
210: they have the same frequency and phase.
211: $V$ has slowly increasing regions that appear to be exponential
212: and that alternate with quick drops. Correlated with this behavior,
213: $I$ is zero then it has a narrow spike.
214: As a consequence, the power spectra of $V$ and $I$
215: shown in Fig.2d consist of discrete lines at the same frequencies but
216: have spectral components of different relative weights.
217: The quality factor is 400 and in various
218: cool-downs we measured values as high as 1000.
219: In addition a comparison of Fig.2a and Fig.2b reveals that there is
220: an intricate connection between the oscillations and the $I$-$V$ as
221: the extremes of the $V$-oscillations coincide with
222: $V_T$ and $V_U$, the voltages of the endpoints $T$ and $U$ of
223: the branches $\alpha$ and $\gamma$ of the $I$-$V$.
224: % bistability due to positive feedback - Schmitt trigger
225:
226: \begin{figure}[!t]
227: \begin{center}
228: \includegraphics[width=2.7in]{cFig3.eps}
229: \end{center}
230: \caption{\label{f3}
231: The dependence of the voltage (panel a) and the current oscillations (panel b)
232: and of the oscillating frequency (panel c) on the biasing conditions.
233: The line in panel c is the prediction of the model described in the
234: text and it has no adjustable parameters.
235: }
236: \end{figure}
237:
238: The bias dependence of the oscillations is summarized in Fig.3.
239: Fig.3a and Fig.3b show that while the amplitude of
240: both the $V$ and $I$-oscillations does not change with bias, the
241: oscillation frequency $f$ is strongly bias dependent.
242: This can better be seen in Fig.3c which displays the
243: increasing $f$ with the average current through the sample $I_{dc}$.
244: We found that $f$ is weakly sublinear and it extrapolates to zero as
245: $I_{dc}$ vanishes.
246: %The time delay between consecutive $I$-spikes is the
247: %period of the oscillations, hence it is a strong function of the bias.
248: The width of the current spikes, shown in Fig.3b,
249: depends only weakly on the bias. The oscillations are still well developed at
250: $I_{dc}=5.7$nA but they disappear for $I_{dc}>5.8$nA. In the narrow
251: current range between the above two values $V$ and $I$ have
252: burst-like behavior (not shown) that is non-periodic and that is strongly
253: influenced by fluctuations in the electrical and thermal environment of
254: the sample.
255:
256: We show now that the $I$-biasing circuit of Fig.1c is incomplete.
257: Indeed, as seen in Fig.2b,
258: there are time periods when the current flowing through the sample in zero while
259: the current through $R_b$ is 4.5nA.
260: Kirchhoff's first law of current conservation thus cannot be satisfied unless
261: an additional circuit element is included. This element is
262: the capacitance $C$ of our cables to ground.
263: The completed circuit consisting of a non-ohmic
264: resistor $S$, the capacitance $C$, and the biasing elements $V_b$, $R_b$
265: is shown in Fig.4. For the model $I$-$V$ of the non-ohmic element
266: we consider two dc stable branches described by $I=0$ for $V<V_T$ and
267: $I=(V-V_0)/r$ for $V>V_U$. From the analysis of this circuit we
268: obtain that when the
269: biasing conditions are such that the load line crosses both branches of the $I$-$V$
270: (not shown), the circuit is bistable, i.e. depending on the biasing history
271: the circuit assumes one of the two stable operating points.
272: However, when the biasing is such that the load line does not
273: cross any of the stable branches, shown in Fig.4, the circuit is
274: astable i.e. it exhibits temporal oscillations between two predetermined states.
275:
276: \begin{figure}[!h]
277: \begin{center}
278: \includegraphics[width=2.7in,trim=0.0in 0.0in 0.0in 0.1in]{cFig4.eps}
279: \end{center}
280: \caption{\label{f5}
281: The complete circuit, the model $I$-$V$ of the sample $S$, and
282: predictions of the model for the time dependences of $V$ and $I$.
283: }
284: \end{figure}
285:
286: We describe now the temporal behavior of the astable circuit.
287: When we connect the battery to the circuit
288: the sample does not conduct. As a result the capacitor starts
289: charging up towards $V_b$ with the time constant $R_bC$.
290: When the voltage $V$ on the sample, which is identical to that on $C$,
291: reaches its threshold value $V_T$
292: the sample becomes a conductor and the capacitor starts to
293: discharge through the sample. During a quick discharge the sample current is
294: bound to take values on the non-zero branch of the $I$-$V$ and
295: the discharge continues until the voltage on the sample drops to $V_U$.
296: At this point the charging process begins anew
297: leading to cyclical variations of $V$ and $I$ that resemble those of Fig.2.
298:
299: The simple $I$-$V$ with linear branches allows an exact description
300: of these processes. After solving the differential equations describing
301: the motion of the charges
302: % and imposing the correct boundary values
303: we obtain the charging and discharging time of the capacitor to be
304: $\tau_1=R_bC \ln (V_b-V_U)/(V_b-V_T)$ and
305: $\tau_2=r C \ln ((V_T-V_0)R_b-(V_b-V_0)r)/((V_U-V_0)R_b-(V_b-V_0)r)$,
306: respectively. However, a simple approximation
307: will be sufficient to explain our data.
308: In the large $R_b$ limit, i.e. for $R_b>>r$ and $V_b>>V_T>V_U$,
309: the time scales above can be expressed as
310: \begin{equation}
311: \tau_1 \simeq C \frac{V_T-V_U}{I_{dc}}, \;\;\;\text{and} \;\;\;
312: \tau_2 \simeq rC \ln \left( \frac{V_T-V_0}{V_U-V_0} \right).
313: \label{eq1}
314: \end{equation}
315: The frequency of the oscillations $f=1/(\tau_1+\tau_2)$
316: is a function of quantities characterizing both the $I$-$V$ of
317: the sample as well as the external biasing circuitry.
318: The parameters $V_T=21.4$mV, $V_U=8.4$mV,
319: $V_0=6.5$mV, $r=144$k$\Omega$ extracted from the $I$-$V$
320: and the cable capacitance of $C \simeq 610$pF yield
321: $\tau_2=0.18$ms which is close to the width of the
322: spikes of Fig.3b.
323: The continuous line on Fig.2c is the prediction of Eq.(\ref{eq1})
324: for the current dependence of $f$ that has no fitting parameters
325: and which is in excellent agreement with our data.
326:
327: \begin{figure}[h]
328: \begin{center}
329: \includegraphics[width=2.7in]{cFig5.eps}
330: \end{center}
331: \caption{\label{f4}
332: The dependence of the frequency of oscillations on the capacitance to ground $C$
333: at $R_b$=50M$\Omega$ (panel a) and on the biasing resistor $R_b$ at $C$=610pF
334: (panel b) at 4.5nA bias.
335: }
336: \end{figure}
337:
338: To further test the model we vary the circuit elements $C$ and $R_b$
339: at a fixed 4.5nA biasing current. As shown in Fig.5a
340: we find that the frequency at fixed $R_b$ is inversely
341: proportional to the capacitance $C$, a behavior that is consistent with
342: the linearly scaling $\tau_1$ and $\tau_2$ with $C$.
343: At a given value of $I_{dc}$ and at
344: fixed value of $C$ in the limit of large $R_b$s
345: the frequency is found to be independent of $R_b$ as seen in Fig.5b.
346: Such a behavior is expected from Eq.\ref{eq1} showing
347: both $\tau_1$ and $\tau_2$ to be $R_b$-independent.
348:
349: According to the described model spontaneous oscillations develop in the
350: WS and in devices with a special non-linear $I$-$V$, i.e.
351: a hysteretic $I$-$V$ in the $V$-biased setup
352: and an $I$-$V$ that exhibits NDR in the $I$-biased setup.
353: These oscillations are known as relaxation oscillations
354: and the circuit is called astable multivibrator or Schmitt trigger \cite{schmitt}.
355: The Shockley diode, thyristor, diac, triac \cite{sze}
356: and the hot electron diode \cite{hed} are examples of
357: semiconductor junction devices with similar $I$-$V$s that have been used
358: in oscillators. The hysteretic $I$-$V$ of the WS, unlike
359: that of the devices above, is a genuine bulk effect that can be explained
360: by the response of the WS to an external bias field in the presence of disorder.
361: This response is not yet fully understood. In one scenario the WS breaks up into
362: domains that are pinned by the disorder and the finite conduction results
363: from motion of the WS domains when the externally applied force overcomes
364: the pinning force \cite{fuku}.
365: Hysteresis results from the gain of kinetic energy from the external field.
366: Such an energy gain is forbidden when the WS is pinned but
367: ensures a flow of charges when the voltage is lowered below its threshold value.
368: In this interpretation the oscillations arise from the
369: switching between the pinned and the sliding solid, a process that is similar
370: to that found in CDW \cite{mihaly}.
371: Lattice defects moving through a pinned WS constitute a second scenario.
372: For example an interstitial electron can move in the periodic potential
373: of the other electrons forming the WS. Such a motion
374: leads to Bloch energy bands and conduction can arise
375: when the external field excites the interstitial electron into the Bloch conduction band.
376:
377: To summarize, the $I$-$V$ of the WS forming at the highest $B$-fields
378: is hysteretic in the $V$-biased setup and
379: exhibits a region of NDR in the $I$-biasing configuration.
380: At special biasing conditions spontaneous voltage and current oscillations
381: are detected that can be accounted for by a simple model.
382: This model predicts similar oscillations for the $B=0$ WS in a Si-MOSFET
383: in which NDR has recently been reported \cite{mos}. Finally we note that
384: due to the different bias circuits a comparison of the results obtained by us
385: and those for the RIQHS \cite{jim-iv,jim-osc}
386: is not straightforward. Several properties of the two phases, however,
387: show resemblance. For both electronic solids the $I$-$V$s are hysteretic,
388: oscillations develop in the audio frequency range,
389: the oscillation frequency increases with increasing bias, and
390: the oscillation amplitude is independent of the bias.
391: This research was funded by the DOE and the NSF.
392:
393: \begin{thebibliography}{29}
394:
395: \bibitem{pan}
396: see in Perspectives in Quantum Hall Effect,
397: edited by S. Das Sarma and A. Pinczuk (Wiley, New York, 1996).
398:
399: \bibitem{early}
400: V.J. Goldman, M. Shayegan, and D.C. Tsui, \prl {\bf 61}, 881 (1988);
401: R.L. Willett {\it et al.}, \prb {\bf 38}, R7881 (1988);
402: H.W. Jiang {\it et al.}, \prl {\bf 65}, 633 (1990).
403:
404: \bibitem{iv}
405: R.L. Willett {\it et al.}, \prb {\bf 40}, 6432 (1989);
406: V.J. Goldman {\it et al.}, \prl {\bf 65}, 2189 (1990);
407: Y.P. Li {\it et al.}, \prl {\bf 67}, 1630 (1991).
408: % H.W. Jiang {\it et al.}, \prb {\bf 44}, 8107 (1991).
409:
410: \bibitem{williams}
411: F.I.B. Williams {\it et al.}, \prl {\bf 66}, 3285 (1991).
412:
413: \bibitem{lloyd}
414: % D.C. Glattli {\it et a.}, Surf. Sc. {\bf 229}, 344 (1990);
415: M.A. Paalanen {\it et al.}, \prb {\bf 45}, 11342 (1992);
416: L.W. Engel {\it et al.}, Solid State Commun. {\bf 104}, 167 (1997);
417: P.D Ye {\it et al.}, \prl {\bf 89}, 176802 (2002).
418:
419: \bibitem{lloyd2}
420: Y.P. Chen {\it et al.}, \prl {\bf 93}, 206805 (2004). % A,B,C
421:
422: \bibitem{chen}
423: Y. Chen {\it et al.}, \prl {\bf 91}, 016801 (2003);
424: R.M. Lewis {\it et al.}, Physica E {\bf 22}, 104 (2004).
425:
426: \bibitem{wigner}
427: E.P. Wigner, Phys. Rev. {\bf 46}, 1002 (1934);
428: Y.E. Lozovik and V.I. Yudson, JETP Letters {\bf 22}, 11 (1975);
429: H. Fukuyama, P.M. Platzman, and P.W. Anderson, \prb {\bf 19}, 5211 (1979).
430:
431: \bibitem{lilly}
432: M.P. Lilly {\it et al.}, \prl {\bf 82}, 394 (1999);
433: R.R. Du {\it et al.}, Solid State Commun. {\bf 109}, 389 (1999).
434:
435: \bibitem{koulakov}
436: A.A. Koulakov, M.M. Fogler, and B.I. Shklovskii, \prl {\bf 76}, 499 (1996).
437:
438: %\bibitem{gruner}
439: %G. Gr\" uner, \rmp {\bf 60}, 1129 (1988).
440:
441: \bibitem{jim-iv}
442: K.B. Cooper {\it et al.}, \prb {\bf 60}, R11285 (1999).
443:
444: \bibitem{rupert}
445: R.M. Lewis {\it et al.}, \prl {\bf 89}, 136804 (2002).
446:
447: \bibitem{jim-osc}
448: K.B. Cooper {\it et al.}, \prl {\bf 90}, 226803 (2003).
449:
450: \bibitem{fleming}
451: R.M. Fleming and C.C. Grimes, \prl {\bf 42}, 1423 (1979).
452:
453: \bibitem{sze}
454: S.M. Sze, Physics of Semiconductor Devices, John Whiley \& Sons, 1981.
455:
456: \bibitem{schmitt}
457: see for example A.S. Sedra and K.C. Smith,
458: Microelectronic Circuits, Oxford University Press, 1998.
459: % p 1002
460:
461: \bibitem{hed}
462: K. Hess {\it et al.}, J. Appl. Phys. {\bf 60}, 3775 (1986);
463: A.M. Belyantsev {\it et al.}, JETP Lett. {\bf 43}, 437 (1986);
464: J. Kolodzey {\it et al.}, IEEE Electron Device Lett. {\bf 9}, 272 (1988).
465:
466: \bibitem{fuku}
467: A.I. Larkin, JETP {\bf 31}, 784 (1970);
468: H. Fukuyama and P.A. Lee, \prb {\bf 17}, 535 (1978).
469: %B.G.A. Normand, P.B. Littlewood, and A.J. Millis, \prb {\bf 46}, 3920 (1992);
470: %H.A. Fertig, \prb {\bf 59}, 2120 (1999);
471: %M.M. Fogler and D.A. Huse, \prb {\bf 62}, 7553 (2000);
472: %R. Chitra, T. Giamarchi, and P. Le Doussal, \prb {\bf 65}, 035312 (2002).
473:
474: \bibitem{mihaly}
475: L. Mih\' aly and G. Gr\" uner, Solid State Commun. {\bf 50}, 807 (1984);
476: R.P. Hall, M. Sherwin, and A. Zettl, \prl {\bf 52}, 2293 (1984).
477:
478: %\bibitem{ridley}
479: %A.L. McWhorter and R.H. Rediker, Proc. IRE {\bf 47}, 1207 (1959),
480: %A. Kittel {\it et al.}, Phys. Lett. A {\bf 147}, 229 (1990);
481: %U. Rau {\it et al.}, \prb {\bf 43}, 2255 (1991).
482:
483: \bibitem{mos}
484: % V.M. Pudalov {\it et al.}, \prl {\bf 70}, 1866 (1993);
485: V.M. Pudalov, J. Phys. (Paris) {\bf 12}, Pr9-331 (2002).
486:
487: \end{thebibliography}
488:
489: \end{document}
490:
491:
492:
493:
494:
495:
496:
497:
498:
499:
500:
501:
502:
503:
504:
505:
506:
507:
508:
509:
510:
511:
512:
513:
514:
515:
516:
517:
518:
519: