cond-mat0108497/st.tex
1: 
2: \documentclass{epl}
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: %TCIDATA{Created=Tue Aug 07 09:48:28 2001}
5: %TCIDATA{LastRevised=Mon Aug 13 18:16:16 2001}
6: 
7: \institute{
8: \inst{1} Physics Department and Center for Stochastic
9: Processes in Science and Engineering, 
10: Virginia Polytechnic Institute and State University,
11: Blacksburg, VA. 24061-0435, USA \\
12: \inst{2} Lehrstuhl Werkstoffkunde und Technologie der Metalle (WTM), 
13: Universit\"{a}t Erlangen-N\"{u}rnberg, 
14: Martensstr.~5, 91058 Erlangen, Germany.
15: }
16: 
17: \pacs{05.70.Ln}{Nonequilibrium and irreversible thermodynamics}
18: \pacs{64.75.+g}{Solubility, segregation, and mixing; phase separation}
19: \pacs{66.30.-h}{Diffusion in solids}
20: 
21: %\input{tcilatex}
22: 
23: \begin{document}
24: 
25: \title{Vacancy-mediated domain growth in a driven lattice gas. }
26: 
27: \author{B. Schmittmann\inst{1} \and M. Thies\inst{2}}
28: \maketitle
29: 
30: \vspace{-.5cm}
31: 
32: \begin{abstract}
33: Using Monte Carlo simulations and a mean-field theory, we study domain growth 
34: in a driven lattice gas. Mediated by a single vacancy, two
35: species of particles (``charges'') unmix, so that a disordered initial
36: configuration develops charge-segregated domains which grow logarithmically
37: slowly. An order parameter is defined and shown to satisfy dynamic scaling.
38: Its scaling form is computed analytically, in excellent agreement with
39: simulations.
40: \end{abstract}
41: 
42: \vspace{-0.8cm} 
43: 
44: \emph{Introduction. }Driven diffusive lattice gases, first
45: introduced by Katz, et.al.~\cite{KLS}, display generic non-equilibrium
46: features in their whole phase space~\cite{DL17}. Even when interactions are
47: due to excluded volume constraints alone, complex phase diagrams can be
48: induced by local dynamic rules~\cite{SHZ,ABC-ss,ABC-t} which violate
49: detailed balance~\cite{DB}. While disordered phases and universal properties
50: near continuous transitions are quite well understood by now, ordered phases
51: have proven far more complex~\cite{DL17}. In particular, studies of phase
52: ordering and coarsening in \emph{driven} systems have revealed surprising
53: behaviors, quite distinct from those~\cite{GB} exhibited by systems evolving
54: towards an \emph{equilibrium} final state. For example, when driven, the
55: two-dimensional (2D) Ising lattice gas develops non-universal domain
56: morphologies which grow in a highly anisotropic fashion and do not satisfy
57: dynamic scaling~\cite{CAL}. In 1D, ordered domains grow with time as 
58: $t^{1/2}$~\cite{CB}, in contrast to the equilibrium $t^{1/3}$ law~\cite{CKS}.
59: Analytic results are sparse and focus primarily on establishing growth 
60: laws~\cite{CB,ABC-t,2D-c,2D-nc}. Clearly, further studies of domain growth far
61: from equilibrium are needed before a more general framework can be
62: established.
63: 
64: In this letter, we report a detailed dynamic scaling analysis for
65: defect-mediated domain growth in a driven three-state lattice gas~\cite{SHZ}%
66: . Two species of particles, labelled ``positive'' and
67: ``negative'', diffuse on a periodic lattice by hopping onto  
68: a single empty site (the defect). Only nearest-neighbor 
69: jumps are allowed, biased by an ``electric'' field $E$ aligned with a 
70: lattice axis. At \emph{finite} vacancy concentration, this system
71: exhibits a phase transition, controlled by drive and particle density, from
72: a disordered, ``free-flowing'' phase to an ordered, ``jammed'' phase~\cite
73: {SHZ,BSZ,LZ}. Ordered configurations are spatially inhomogeneous: positive
74: and negative particles form adjacent strips transverse to $E$, 
75: impeding each other. A mean-field theory predicts the 
76: structure and stability limits of both phases~\cite{SHZ,LZ,VZS}.
77: Models of this type have been invoked to describe water-in-oil
78: microemulsions~\cite{waterdroplets}, gel electrophoresis~\cite{gelelectro}
79: and traffic flow~\cite{trafficflow}.
80: 
81: Remarkably, even the presence of a \emph{single} vacancy suffices to order
82: an initially random system. This problem, in both its static and dynamic
83: aspects, is an example of a much-wider ranging class of interacting random
84: walk and defect-mediated domain growth problems. The vacancy plays the role
85: of a highly mobile defect~\cite{VMD}, whose motion restructures its
86: environment (i.e., the particle configuration), while the latter provides a
87: feedback through the local jump rates. In the resulting steady state,
88: investigated in detail in~\cite{TS}, particles form a charge-segregated
89: strip, with the vacancy \emph{localized} at one of the interfaces. The
90: density profiles obey characteristic scaling forms, controlled by drive and
91: system size.
92: 
93: Here, we focus on the \emph{time evolution} of this system in 2D,
94: seeking to understand how ordered domains form,
95: grow and finally saturate. We present Monte Carlo (MC) data, 
96: supported by an approximate solution of the
97: time-dependent mean-field equations. The ordering process exhibits three
98: regimes: the initial formation of the strip, an extended
99: logarithmic growth regime and the crossover to steady state. We
100: demonstrate that the system satisfies dynamic scaling in the second
101: regime and compute the scaling function analytically, in excellent
102: agreement with the data. We conclude with some suggestions for further
103: study. More details can be found in~\cite{long}.
104: 
105: \emph{The microscopic model. }Our model is defined on a 2D square lattice of 
106: $L_{x}\times L_{y}$ sites with fully periodic boundary conditions (PBC). The
107: occupation of each site is represented by a spin variable $s_{xy}$ taking
108: the values $+1,-1,0$, if a positive or negative charge or the vacancy is
109: present at site ($x,y$). For simplicity, we restrict our study to equal
110: densities of positive and negative charges. 
111: At each MC step (MCS), the vacancy exchanges with a
112: randomly chosen nearest neighbor $s_{xy}$, with Metropolis rate~%
113: \cite{metropolis}, $\min \left\{ 1,\exp (s_{xy}E\delta y)\right\} $. Here, $%
114: \delta y=0,\pm 1$ is the change of the particle's $y$--coordinate due to the
115: jump. $E$ denotes the bias, uniform in space and time and directed along
116: positive $y$.
117: 
118: 
119: %=============================Fig. 1================
120: \begin{figure}[tbp]
121: \input epsf \hfill \hfill 
122: \begin{minipage}{0.16\textwidth}
123:   \epsfxsize = \textwidth \epsfysize = 1.5\textwidth \hfill
124:   \epsfbox{./6lat2.ps} \hfill
125:     \begin{center} (a) \end{center}
126:   \end{minipage}
127: \hfill \hfill 
128: \begin{minipage}{0.16\textwidth}
129:   \epsfxsize = \textwidth \epsfysize = 1.5\textwidth \hfill
130:   \epsfbox{./6lat3.ps} \hfill
131:     \begin{center} (b) \end{center}
132:   \end{minipage}
133: \hfill \hfill \vspace{0.02\textwidth} %\par
134: %\hspace{1.7cm}
135: \begin{minipage}{0.16\textwidth}
136:   \epsfxsize = \textwidth \epsfysize = 1.5\textwidth \hfill 
137:   \epsfbox{./6lat4.ps}
138:     \begin{center} (c) \end{center}
139:   \end{minipage}
140: \hfill \hfill 
141: \begin{minipage}{0.16\textwidth}
142:   \epsfxsize = \textwidth \epsfysize = 1.5\textwidth \hfill
143:   \epsfbox{./6lat5.ps}
144:     \begin{center} (d) \end{center}
145:   \end{minipage}
146: \hfill \hfill 
147: \begin{minipage}{0.16\textwidth}
148:   \epsfxsize = \textwidth \epsfysize = 1.5\textwidth \hfill
149:   \epsfbox{./6lat6.ps}
150:     \begin{center} (e) \end{center}
151:   \end{minipage}
152: \hfill \hfill \vspace{-.4cm} 
153: \caption{Snapshots of a $16\times 24$ lattice at $E=0.8$, taken after 
154: (a) $10^{3}$, (b) $10^{4}$, (c) $10^{5}$, (d) $10^{6}$, 
155: (e) $10^{7}$ MC steps. 
156: Negative (positive) particles are colored black (white), and the
157: hole is gray. $E$ points upwards. The initial configuration was
158: random.}
159: 
160: \vspace{-.4cm}
161: \end{figure}
162: %=======================================================
163: 
164: 
165: This deceptively simple dynamics induces a \emph{%
166: charge segregation} process on the lattice, as illustrated by Fig.~1. 
167: Starting from a random
168: configuration (not shown), the system remains disordered for early times
169: (Fig.~1a). Eventually, the hole begins to segregate the two species: In 
170: Fig.~1b, an interface between regions of opposite charge begins to develop. Due
171: to PBC, a second interface must also form; this occurs at a time set by $%
172: L_{y}$. After $10^{5}$ MCS (Fig.~1c), the segregation of charges is already
173: quite apparent. Clearly, the field drives the hole \emph{towards }the lower
174: (the ``downstream''), and \emph{away from} the upper (the ``upstream''),
175: interface. Typical configurations are homogeneous in the transverse direction.
176: For our choice of parameters, the interfaces are well separated 
177: \cite{TS}. Since any increase of order requires a series of \emph{%
178: field-suppressed} exchanges, the approach to steady state is very slow
179: (Figs.~1c-e).
180: 
181: 
182: In the following, we provide a brief quantitative analysis of this process,
183: beginning with our MC results. 
184: The control parameters of our study are $E$, ranging from $0.2$ to $2.0$,
185: and the system size, $L_{x}=16$ or $40$, $16\leq L_{y}\leq 72$. 
186: The initial configuration is random. Time is measured in MCS. All of our data are averaged 
187: (denoted by $\left\langle \cdot \right\rangle $) over 
188: $100$ independent runs (samples), resulting in good statistics 
189: with errors well below $5\%$.
190: 
191: To probe the growth of the ordered domain, we measure averaged \emph{hole
192: and charge density profiles}, defined via $\phi (y,t){}\,\equiv
193: \,\left\langle \frac{1}{L_{x}}\sum_{x}(1-s_{xy}^{2})\right\rangle $ and $%
194: \psi (y,t)\,\equiv \,\left\langle \frac{1}{L_{x}}\sum_{x}s_{xy}\right\rangle 
195: $. Due to translational invariance, strips can be centered at any $y$; thus,
196: care must be taken when averaging: In each sample, we determine the maximum
197: of the hole density over a sufficient time interval. This maximum marks the
198: downstream interface. The charge density profiles from different samples are
199: now shifted so that these maxima coincide, and averages can be taken.
200: Clearly, this procedure is not very reliable for early times, since no or
201: only faint strips have developed yet; however, our focus here concerns later
202: stages of the growth process. Then, fluctuations of the downstream interface
203: position are rather small and very slow, so that its location can be
204: determined very accurately.
205: 
206: The evolution of the charge density profiles proceeds in three stages, as
207: illustrated in Fig.~2 for systems with different $L_{y}$. At early times,
208: the downstream interface forms and equilibrates quite rapidly, reaching its
209: steady state form at about $10^{5}$ MCS, independent of $L_{y}$. Near this
210: interface, the charges are already perfectly segregated, with the charge
211: profile saturating at $\pm 1$. Two fronts, one on each side, separate the
212: ordered from the remaining disordered region. The larger systems now enter a
213: second regime, during which both fronts travel slowly outwards, increasing
214: the width of the ordered region. This process is also independent of $L_{y}$%
215: : the centers of the fronts move out as $\ln t$, while their shape remains
216: unchanged. Eventually, in the third regime, the two fronts merge, due to
217: PBC, and the upstream interface equilibrates. The system has now reached
218: steady state. For smaller systems (including the one shown in Fig.~1), the
219: crossover to steady state sets in before well-developed fronts can form.
220: 
221: %====================Fig. 2====================
222: 
223: \begin{figure}[tbp]
224: \input{epsf} \vspace{-.5cm} \hspace{2.5cm} 
225: \begin{minipage}{.4\textwidth}
226:     \epsfxsize = 1.3\textwidth \epsfysize = \textwidth 
227:   \epsfbox{42.ps} 
228:   \end{minipage}
229: \vspace{-.4cm} 
230: \caption{Charge density profiles for $L_{x}=16$, $E=0.8$, and a range of 
231: $L_{y}$, at different MC times (arrows). Some data
232: points, away from the central region, are not shown for clarity.}
233: \label{cd_Ly24324872_E08}
234: \vspace{-.5cm}
235: \end{figure}
236: 
237: %=============================================
238: 
239: As a quantitative probe,  
240: we define an order parameter, $\bar{Q}\equiv
241: \sum_{y}\psi ^{2}(y,t)$~\cite{TS} which measures the area, and hence the
242: degree of order, under the (averaged, squared) charge profile. A central
243: result of our work, Fig.~3 demonstrates that $\bar{Q}$ exhibits dynamic
244: scaling in the logarithmic growth regime: Excellent data collapse is
245: observed, for a range of $E$, $L_{x}$ and $L_{y}$ if $E\bar{Q}$ is plotted
246: vs the scaling variable $\tilde{t}\equiv tE^{3}/L_{x}$. This confirms our
247: picture of the ordering process: the motion and shape of the fronts is
248: independent of $L_{y}$, and $\bar{Q}(\tilde{t})\propto \ln \tilde{t}$, since
249: each escape of the vacancy is an activated process (exponentially suppressed
250: by $E$).
251: 
252: Two comments are in order. First, we should emphasize that $\bar{Q}$ differs
253: from an order parameter used previously, namely $Q\equiv
254: \left\langle \sum_{y}\left( \frac{1}{L_{x}}\sum_{x}s_{xy}\right)
255: ^{2}\right\rangle $~\cite{SHZ}. The latter can be measured directly from the
256: configurations, without the shifting procedure. Essentially, $Q$
257: counts the \emph{total} number of ordered rows transverse to the external
258: field. It is therefore sensitive to the presence of \emph{multiple strips}
259: which can emerge during the ordering process, especially in larger systems.
260: As a result, the growth of $Q$ exhibits a weak $L_{y}$-dependence. In
261: contrast, the shifting procedure averages out multiple strips, in favor of
262: the largest (dominant) strip. Here, we focus on $\bar{Q}$ but note that the
263: study of multiple strips remains an outstanding problem. Second, our
264: scenario of the ordering process relies on having a well-developed
265: downstream interface sandwiched between two fully charge-segregated regions.
266: Our steady-state studies~\cite{TS} show that this is the case
267: provided $EL_{y}\geq 18$. Thus, the system with $L_{y}=24$, $E=0.8$ sets a
268: lower limit.
269: 
270: \emph{Analytic approach. }Finally, we turn to an analytic description, 
271: with the aim of deriving the appropriate scaling variables and the 
272: slope of the $\ln \tilde{t}$ law. 
273: Our starting point is a set of mean-field equations for the
274: (coarse-grained) local hole and charge densities. Due to particle number
275: conservation, the continuum version of our model takes the form of two
276: continuity equations which can be derived from a microscopic
277: master equation. Their form has been well established~\cite{SHZ,VZS,TS}.
278: Proceeding directly to the equations for the profiles, we obtain 
279: \begin{eqnarray}
280: \partial _{t}\phi (y,t) &=&\partial \left\{ \partial \phi +E\phi \psi
281: \right\}    \\
282: \partial _{t}\psi (y,t) &=&\partial \left\{ \phi \partial \psi -\psi
283: \partial \,\phi -E\phi \right\}   \label{prof}
284: \end{eqnarray}
285: Here, $\partial $ denotes a spatial derivative in the drive direction, and
286: terms of $O(\phi ^{2})$ have been neglected, since they reflect the presence
287: of multiple vacancies. The continuum limit has been taken by letting the
288: lattice constant vanish at finite $E$. The equations have to be supplemented
289: with appropriate boundary conditions and the constraints on total mass and
290: charge.
291: 
292: %===================Fig. 3===============================
293: 
294: \begin{figure}[tbp]
295: \input{epsf} \vspace{-.5cm} \hspace{2.5cm} 
296: \begin{minipage}{.4\textwidth}
297:     \epsfxsize = 1.3\textwidth \epsfysize = \textwidth 
298:   \epsfbox{416.ps} 
299:   \end{minipage}
300: \vspace{-.2cm}
301: \caption{Scaling plot of order parameter. The straight line, with slope
302: $2.1\pm0.1$, is
303: a fit to the data.}
304: \label{logfit}
305: \vspace{-.3cm} 
306: \end{figure}
307: 
308: %===========================================================
309: 
310: Solving these coupled, nonlinear PDE's is of course highly nontrivial.
311: However, the data suggest a few approximations which
312: allow for considerable analytic progress. Since we wish to capture 
313: the logarithmic growth regime, we
314: center the downstream interface at the origin ($y=0$) and focus on the front
315: moving in the positive $y$-direction (the other front being a mirror image,
316: of course).
317: By the onset of the second regime, this front is located 
318: well away from $y=0$, since 
319: the charge density at the downstream
320: interface has already equilibrated (cf. Fig.~2).
321: Moreover, MC data
322: show that the hole density profile is strongly localized at the origin and
323: does not change significantly as time progresses. Therefore, we replace it in Eq.~(\ref{prof})
324: by its \emph{steady state} form $\phi _{o}(y)$~\cite{TS} 
325: which decays, to excellent approximation, as 
326: $c^{-1}\exp (-E|y|)$ on both sides. Here, $c\propto L_{x}E^{-1}$ is a
327: normalization, such that $\phi _{o}(y)$ is the \emph{probability} 
328: of finding the vacancy at site ($x,y$) in the system. 
329: Introducing the scaling variables, 
330: $z\equiv Ey$ and $%
331: \tau \equiv c^{-1}E^{2}t$, we arrive~\cite{long} at a parameter-free equation,
332: valid for $y>0$: 
333: \begin{equation}
334: \partial _{\tau }\psi (z,\tau )=\exp (-z)\left[ \partial _{z}^{2}\psi -\psi
335: +1\right] .  \label{psi}
336: \end{equation}
337: Clearly, initial and boundary conditions are needed. To study 
338: the motion of the front well before saturation
339: occurs, we may allow $0<z<\infty $~\cite{c}. The disordered phase far ahead
340: of the front is modelled by $\lim_{z\rightarrow \infty }\psi (z,\tau )=0$
341: for all $\tau <\infty $. Further, the fully ordered phase well behind the
342: front should be independent of $t$, since it is unaffected by further
343: growth; thus, we demand that $\psi (z,\tau )\rightarrow 1$ for $\tau
344: \rightarrow \infty $ and all $0<z<\infty $.
345: 
346: Some comments are in order. First, since $c\propto L_{x}/E$, the
347: characteristic scaling of MC time with $E^{3}/L_{x}$ already emerges.
348: Second, since $\bar{Q}\equiv 2\int_{0}^{\infty }\psi
349: ^{2}(y,t)dy=2E^{-1}\int_{0}^{\infty }\psi ^{2}(z,\tau )dz$, our mean field
350: theory predicts data collapse if $\bar{Q}E$ is plotted vs $tE^{3}/L_{x}$, as
351: borne out by Fig.~3.
352: 
353: Since the moving front retains its shape, we may seek a
354: solution in the form $\psi (z,\tau )\equiv f(w)$, with $%
355: w\equiv z-z_{o}(\tau )$. Clearly, $z_{o}(\tau )$ describes the front
356: position. Eq.~(\ref{psi}) becomes
357: 
358: \begin{equation}
359: -\dot{z}_{o}e^{z_{o}+w}f^{\prime }=f^{\prime \prime }-f+1,  \label{front}
360: \end{equation}
361: where $\dot{z}_{o}\equiv $ $\partial _{\tau }z_{o}$ and $f^{\prime }\equiv
362: \partial _{w}f$, with $\lim_{w\rightarrow +\infty }f(w)=0$ and $%
363: \lim_{w\rightarrow -\infty }f(w)=1$. Also, $f^{\prime }$ differs notably
364: from zero only in a (finite) neighborhood of $w=0$. Multiplying both
365: sides of (\ref{front}) by $f^{\prime }$, and integrating from $-\infty $
366: to $+\infty $, we arrive at $\dot{z}_{o}e^{z_{o}}\int_{-\infty }^{+\infty
367: }dwe^{w}f^{\prime 2}=\frac{1}{2}.$ Obviously, $\int_{-\infty }^{+\infty
368: }dwe^{w}f^{\prime 2}\equiv \kappa $ is a positive, nonvanishing \emph{%
369: numerical} constant. The time dependence of the front position (with simple
370: initial condition $z_{o}(0)=0$) now follows as $z_{o}(\tau )=\ln \left(
371: 1+\tau /2\kappa \right) $. It is logarithmic, as expected.
372: 
373: Returning to Eq.~(\ref{front}), and using $\dot{z}_{o}e^{z_{o}}=1/2\kappa $,
374: the shape of the front, subject to the specified boundary conditions, is given by $f(w)=2\kappa e^{-w}\left[ 1-\exp (-e^{w}/2\kappa)\right]$. 
375: The associated order parameter is now easily computed: $\bar{Q}\,E=2\ln \tau
376: -0.927...+O(1/\tau )$, for large times $\tau >>\kappa $. Gratifyingly, the
377: key features of the MC data, namely, the correct scaling variables, the
378: logarithmic growth law and even its amplitude (i.e., the factor $2$) are
379: reproduced by our solution. Of course, a single additive constant is
380: required to match the time scale of the simulations.
381: 
382: \emph{Conclusions. }In this letter, we presented MC and analytic results for
383: domain growth in a simple driven lattice gas. A single vacancy rearranges
384: positive and negative particles into charge-segregated strips 
385: transverse to the drive. Focusing on the largest strip, 
386: we find that it grows logarithmically
387: with time, via field-suppressed excursions of the vacancy away from the
388: center of the strip. The ordered domain is separated from the remaining
389: disordered region by two well-defined moving fronts which retain their shape
390: during the growth process. Eventually, the two fronts merge due to our BC's,
391: and the system crosses over into steady state. During the logarithmic growth
392: regime, we observe dynamic scaling of a suitably defined order parameter, 
393: $\bar{Q}$, if $\bar{Q}E$ is plotted vs $tE^{3}/L_{x}$. These findings
394: are supported by 
395: an (approximate) solution of a set of mean-field equations for
396: the charge and hole density profiles. To leading order in $t$, 
397: the scaling function 
398: $\bar{Q}E$ grows as $2\ln \left( tE^{3}/L_{x}\right) $, 
399: in good agreement with the data.
400: 
401: Several questions remain open. First, multiple-strip configurations form
402: easily, especially in larger systems, and our focus on the dominant strip
403: washes out any secondary ones, due to the shifting procedure. 
404: Unfortunately, a full
405: dynamic scaling study of observables sensitive to multiple strips demands
406: much greater computational effort. Whether similar coarsening behavior
407: emerges remains to be explored. A second natural extension of our study
408: would allow for multiple vacancies, i.e., a finite hole density. Under these
409: conditions, particles first form ``clouds'' which coarsen and eventually
410: merge into multiple strips transverse to the field. These strips then
411: continue to coarsen until a single charge-segregated strip survives. A
412: numerical solution of the 1D mean-field equations~\cite{kertesz} for the
413: multiple-strip regime indicates that logarithmic growth persists, consistent 
414: with the arguments of Kafri, et.al.~\cite{2D-c}.
415: However, a detailed analysis, which identifies the relevant dynamic scaling
416: variables and functions, and incorporates the crossover from clouds to
417: strips, is still outstanding.
418: 
419: \acknowledgments
420: 
421: We thank E.~Sch\"{o}ll, R.K.P.~Zia, G.~Korniss, and Z.~Toroczkai for
422: valuable discussions. This research is supported in part by the National
423: Science Foundation through the Division of Materials Research.
424: 
425: 
426: \begin{thebibliography}{99}
427: \vspace{-.3cm}
428: 
429: \bibitem{KLS}  
430: 	\Name{Katz S., Lebowitz J.L. \and Spohn H.} 
431: 	\REVIEW{Phys.~Rev.~B}{28}{1983}{1655} \and
432: 	\REVIEW{J.~Stat.~Phys.}{34}{1984}{497}.
433: 
434: \bibitem{DL17}  
435: 	\Name{Schmittmann B. \and Zia R.K.P.} 
436: 	\Book{Phase Transition and Critical Phenomena}
437: 	\Vol{17}
438: 	\Editor{Domb C. \and Lebowitz J.L.}
439: 	\Publ{Academic, New York}
440: 	\Year{1995}.
441: 
442: \bibitem{SHZ}  
443: 	\Name{Schmittmann B., Hwang K., \and Zia R.K.P.}
444: 	\REVIEW{Europhys.~Lett.}{19}{1992}{19}. 
445: 
446: \bibitem{ABC-ss}  
447: 	\Name{Evans M.R., Kafri Y., Kodulevy H.M., \and Mukamel D.}
448: 	\REVIEW{Phys.~Rev.~Lett.}{80}{1998}{425}
449: 
450: \bibitem{ABC-t}  
451: 	\Name{Evans M.R., Kafri Y., Kodulevy H.M., \and Mukamel D.}  
452: 	\REVIEW{Phys.~Rev.~E}{58}{1998}{2764}. 
453: 	See [12] for further references to similar 1D models. 
454: 
455: \bibitem{DB}  
456: 	\Name{Mukamel D.} in 
457: 	\Book{Soft and Fragile Matter: Nonequilibrium
458: 	Dynamics, Metastability and Flow} 
459: 	\Editor{Cates M.E. \and Evans M.R.}
460: 	\Publ{Institute of Physics Publishing, Bristol}
461: 	\Year{2000}. 
462: 
463: \bibitem{GB}  
464: 	\Name{Gunton J.D., San Miguel M., \and Sahni P.S.} in
465: 	\Book{Phase Transitions and Critical Phenomena}
466: 	\Vol{8}
467:         \Editor{Domb C. \and Lebowitz J.L.}
468: 	\Publ{Academic, New York}
469:         \Year{1983}; 
470: 	\Name{Bray A.J.}
471:         \REVIEW{Adv.~Phys.}{43}{1994}{357}.
472: 
473: \bibitem{CAL} 
474: 	\Name{Yeung C., Rogers T., Hernandez-Machado A., \and Jasnow D.}
475: 	\REVIEW{J.~Stat.~Phys.}{66}{1992}{1071}; 
476: 	\Name{Alexander F.J., Laberge C.A., Lebowitz J.L., \and Zia R.K.P.}
477: 	\REVIEW{J.~Stat.~Phys.}{82}{1996}{1133}; 
478:        	\Name{Rutenberg A.D. \and Yeung C.}
479: 	\REVIEW{Phys.~Rev.~E}{60}{1999}{2710}.
480: 
481: \bibitem{CB}  
482: 	\Name{Cornell S.J. \and Bray A.J.}
483: 	\REVIEW{Phys.~Rev.~E}{54}{1996}{1153}; 
484: 	\Name{Spirin V., Krapivsky P.L., \and Redner S.}
485: 	\REVIEW{Phys.~Rev.~E}{60}{1999}{2670}.
486: 
487: \bibitem{CKS}  
488: 	\Name{Cornell S.J., Kaski K., \and Stinchcombe R.B.}
489: 	\REVIEW{Phys.~Rev.~B}{44}{1991}{12263}
490: 	\Name{Majumdar S.N., Huse D.A., \and Lubachevsky B.D.}
491: 	\REVIEW{Phys.~Rev.~Lett.}{73}{1994}{182}. 
492: 
493: \bibitem{2D-c}  
494: 	\Name{Kafri Y., Biron D., Evans M.R., \and Mukamel D.}
495: 	\REVIEW{Eur.~Phys.~J.~B}{16}{2000}{669} .
496: 
497: \bibitem{2D-nc}  
498: 	\Name{Evans M.R., Kafri Y., Levine E., \and Mukamel D.}
499: 	\REVIEW{Phys.~Rev.~E}{62}{2000}{7619}. 
500: 
501: \bibitem{BSZ}  
502: 	\Name{Bassler K.E., Schmittmann B. \and Zia R.K.P.}
503: 	\REVIEW{Europhys.~Lett.} {24}{1993}{115}.
504:  
505: \bibitem{LZ}  
506: 	\Name{Leung K.-t. \and Zia R.K.P.}
507: 	\REVIEW{Phys.~Rev.~E}{56}{1997}{308}.
508: 
509: \bibitem{VZS}  
510: 	\Name{Vilfan I., Zia R.K.P., \and Schmittmann B.}
511: 	\REVIEW{Phys.~Rev.~Lett.}{73}{1994}{2071}.
512: 
513: \bibitem{waterdroplets}  
514: 	\Name{Aertsens M. \and Naudts J.}
515: 	\REVIEW{J.~Stat.~Phys.}{62}{1990}{609}.
516: 
517: \bibitem{gelelectro}  
518: 	\Name{Rubinstein M.}
519: 	\REVIEW{Phys.~Rev.~Lett.}{59}{1987}{1946};
520: 	\Name{Duke T.A.J.}
521: 	\REVIEW{Phys.~Rev.~Lett.}{62}{1989}{2877}); 
522: 	\Name{Shnidman Y.} in
523: 	\Book{Mathematical and Industrial Problems IV}
524: 	\Editor{Friedman A.}
525: 	\Publ{Springer, Berlin}
526: 	\Year{1991}; 
527: 	\Name{Widom B., Viovy J.L., \and Desfontaines A.D.}
528: 	\REVIEW{J.~Phys.~I (France)}{1}{1991}{1759}.
529: 
530: \bibitem{trafficflow}  
531: 	\Name{Biham O., Middleton A.A., \and Levine D.}
532: 	\REVIEW{Phys.~Rev.~A}{46}{1992}{R6124};
533: 	\Name{Leung K.-t.}
534: 	\REVIEW{Phys.~Rev.~Lett.}{73}{1994}{2386};
535: 	\Name{Molera J.M., Martinez F.C. \and Cuesta J.A.}
536: 	\REVIEW{Phys.~Rev.~E}{51}{1995}{175}; 
537: 	\Name{Chowdhury D., Santen L., \and Schadschneider A.}
538: 	\REVIEW{Phys.~Rep.}{329}{2000}{199}.
539: 
540: \bibitem{VMD}  
541: 	\Name{Toroczkai Z., Korniss G., Schmittmann B., \and Zia R.K.P.}
542: 	\REVIEW{Europhys. Lett.}{40}{1997}{281};
543: 	\Name{Triampo W., Aspelmeier T., \and Schmittmann B.}
544: 	\REVIEW{Phys.~Rev.~E}{61}{2000}{2386}; 
545: 	\Name{Aspelmeier T., Schmittmann B., \and Zia R.K.P.}
546: 	\REVIEW{Phys.~Rev.~Lett.}{87}{2001}{65701}.
547: 
548: \bibitem{TS}   
549: 	\Name{Thies M. \and Schmittmann B.}
550: 	\REVIEW{Phys.~Rev.~E}{61}{2000}{184}.
551: 
552: \bibitem{long}  
553: 	\Name{Schmittmann B. \and Thies M.} to be published.
554: 
555: \bibitem{metropolis}  
556: 	\Name{Metropolis N., Rosenbluth A.W., Rosenbluth M.N.,
557: 	Teller A.H. \and Teller E.}
558: 	\REVIEW{J.~Chem.~Phys.}{21}{1953}{1087}.
559: 
560: \bibitem{c}  These boundary conditions are already implicit in the
561: normalization of $\phi _{o}(y)$.
562: 
563: \bibitem{kertesz}  
564: 	\Name{Kert\'{e}sz J. \and Ramaswamy R.}
565: 	\REVIEW{Europhys.Lett.}{28}{1994}{617}.
566: \end{thebibliography}
567: 
568: \end{document}
569: 
570: 
571: 
572: 
573: 
574: 
575: 
576: 
577: