1: \documentclass[aps,prb,twocolumn,showpacs,superscriptaddress,floatfix]{revtex4}
2: \usepackage{amsmath,amssymb}
3: \usepackage{graphicx}
4:
5: \newcommand{\bfzero}{{\mathbf{0}}}
6: \newcommand{\bfm}{{\mathbf{m}}}
7: \newcommand{\bfn}{{\mathbf{n}}}
8: \newcommand{\bfq}{{\mathbf{q}}}
9: \newcommand{\bfr}{{\mathbf{r}}}
10: \newcommand{\bfx}{{\mathbf{x}}}
11: \newcommand{\bfy}{{\mathbf{y}}}
12: \newcommand{\bfz}{{\mathbf{z}}}
13: \newcommand{\bfJ}{{\mathbf{J}}}
14: \newcommand{\bfR}{{\mathbf{R}}}
15: \newcommand{\bfS}{{\mathbf{S}}}
16: \newcommand{\bfsigma}{{\boldsymbol{\sigma}}}
17: \newcommand{\bftau}{{\boldsymbol{\tau}}}
18: \newcommand{\bftheta}{{\boldsymbol{\theta}}}
19: \newcommand{\bfvarphi}{{\boldsymbol{\varphi}}}
20: \newcommand{\bfDelta}{{\boldsymbol{\Delta}}}
21:
22: \newcommand{\varA}{{\mathcal{A}}}
23: \newcommand{\varD}{{\mathcal{D}}}
24: \newcommand{\varH}{{\mathcal{H}}}
25: \newcommand{\varK}{{\mathcal{K}}}
26: \newcommand{\varO}{{\mathcal{O}}}
27: \newcommand{\varZ}{{\mathcal{Z}}}
28:
29: \newcommand{\ux}{{\hat\bfx}}
30: \newcommand{\uy}{{\hat\bfy}}
31: \newcommand{\uz}{{\hat\bfz}}
32:
33: \newcommand{\pde}[2]{\frac{\partial{#1}}{\partial{#2}}}
34: \newcommand{\ode}[2]{\frac{d{#1}}{d{#2}}}
35: \newcommand{\Ode}[1]{\frac{d}{d{#1}}}
36:
37: \newcommand{\half}{\frac{1}{2}}
38: \newcommand{\rcp}[1]{\frac{1}{#1}}
39:
40: \newcommand{\avg}[1]{\left\langle#1\right\rangle}
41: \newcommand{\nn}[1]{\left\langle#1\right\rangle}
42: \newcommand{\set}[1]{\left\{#1\right\}}
43:
44: \newcommand{\eqnref}[1]{Eq.~(\ref{#1})}
45: \newcommand{\eqnsref}[1]{Eqs.~(\ref{#1})}
46: \newcommand{\Eqnref}[1]{Equation~(\ref{#1})}
47: \newcommand{\Eqnsref}[1]{Equations~(\ref{#1})}
48: \newcommand{\figref}[1]{Fig.~\ref{#1}}
49: \newcommand{\Figref}[1]{Figure~\ref{#1}}
50: \newcommand{\secref}[1]{Sec.~\ref{#1}}
51: \newcommand{\secsref}[1]{Secs.~\ref{#1}}
52: \newcommand{\Secref}[1]{Section~\ref{#1}}
53: \newcommand{\bibref}[1]{Ref.~\cite{#1}}
54:
55: \newcommand{\XY}{\textit{XY} }
56:
57: \begin{document}
58:
59: \title{Phase Transitions in Models for Coupled Charge-Density Waves}
60:
61: \author{Minchul Lee}
62: \affiliation{Department of Physics, Korea University, Seoul 136-701, Korea}
63:
64: \author{Eun-Ah Kim}
65: \affiliation{Department of Physics, University of Illinois, Champaign, IL 61801, U.S.A.}
66:
67: \author{Jong Soo Lim}
68: \affiliation{Department of Physics, Seoul National University, Seoul 151-747, Korea}
69:
70: \author{M. Y. Choi}
71: \affiliation{Department of Physics, Seoul National University, Seoul 151-747, Korea}
72: \affiliation{Korea Institute for Advanced Study, Seoul 130-722, Korea}
73:
74: \begin{abstract}
75: Various phase transitions in models for coupled charge-density waves are investigated
76: by means of the $\epsilon$-expansion, mean-field theory, and Monte Carlo simulations.
77: At zero temperature the effective action for the system with appropriate commensurability
78: effects is mapped onto the three- or four-dimensional \XY model, depending on
79: spatiotemporal fluctuations, under the corresponding symmetry-breaking fields.
80: It is revealed that the three- and four-dimensional systems display a single transition between
81: the clock order (with broken Z$_M$ symmetry) and disorder.
82: The nature of the phase transition depends crucially on the commensurability factor $M$:
83: For $M \ge 4$, in particular, the transition belongs to the same university class as the \XY model.
84: On the other hand, in the presence of misfit causing frustration in the charge-density wave,
85: the inter-chain coupling is observed to favor either the commensurate state or the
86: incommensurate state depending on the initial configuration; this gives rise to hysteresis around
87: the commensurate-incommensurate transition.
88: Boundaries separating such phases with different symmetries are obtained in the parameter
89: space consisting of the temperature, symmetry-breaking field, fluctuation strength,
90: inter-chain coupling, and misfit.
91: \end{abstract}
92: \bigskip
93: \pacs{64.60.Cn, 71.45.Lr, 64.70.Rh}
94:
95: \maketitle
96:
97: % Introduction
98: %\newpage
99: \section{Introduction}
100:
101: A large number of organic and inorganic solids have crystalline structures in which
102: fundamental structural units form linear chains.\cite{Kagoshima87} In these materials,
103: largely different overlaps of the electronic wave functions in various crystallographic
104: directions lead to strongly anisotropic, so-called quasi-one-dimensional (quasi-1D)
105: electron bands. Among the exotic physical phenomena in quasi-1D materials, charge density waves
106: (CDWs) are of much continual interest. Research into this topic has been stimulated
107: by recent advances in experimental techniques, which now allow direct observation of
108: CDWs and measurement of various static and dynamic properties of CDW systems.\cite{Experiment}
109: In general the material of quasi-1D structure is expected to exist in the form of
110: a bundle of chains rather than of a single chain. In such a bundle of chains,
111: inter-chain tunneling of electrons leads to coupling of the fluctuations on
112: neighboring chains, which may affect the behavior of the system in a crucial way.
113: For example, the system of two coupled incommensurate chains in the weak-coupling limit
114: has been reported to exhibit a complicated commensurate-incommensurate (CI) transition,
115: reminiscent of devil's staircase.\cite{Coutinho81}
116: Obviously, the opening of the gap at the Fermi surface implies that each separate chain is an insulator
117: at low temperatures. However, with the coupling between chains taken into account, expected are
118: various transitions between insulating and metallic phases, the extensive investigation of which is still
119: lacking in spite of the ubiquity of the CDW.
120:
121: In this paper, we investigate nature of the phase transitions in the coupled CDW system.
122: In the absence of misfit the effective action for the (commensurate) system at zero temperature
123: with suitable commensurability effects is mapped onto the \XY model under the corresponding
124: symmetry-breaking field. The effective dimension of the system is given by four if both spatial
125: and temporal fluctuations are significant; otherwise it is three.
126: The phase transition in the resulting three-dimensional (3D) \XY model under the appropriate
127: symmetry-breaking field is examined via the $\epsilon$-expansion and the Monte Carlo method.
128: It is found that there emerges in the system the clock ordered phase (with the Z$_M$ symmetry broken)
129: via a second-order transition, the nature of which depends on the commensurability factor $M$:
130: In particular the critical behavior for $M\ge4$ appears to be the same as that of the 3D \XY model.
131: For the four-dimensional (4D) \XY model, on the other hand, mean-field theory is expected to be accurate
132: and reveals a single transition, which is of the first order for $M=3$ and of the second order otherwise.
133: %
134: We then examine the effects of misfit, which not only change the nature of the order-disorder transition
135: but also brings about a CI transition. In the presence of the misfit, correlations due to the inter-chain
136: coupling are observed to favor, depending on the initial configuration, either the commensurate state or an
137: incommensurate state, which gives rise to hysteresis behavior:
138: While in the cooling process the incommensurate CDWs persist near zero temperature, in
139: the heating process large portions of the system remain in the commensurate state
140: at rather high temperatures.
141:
142: This paper is organized as follows: In \secref{sec:ea}, the effective action at zero temperature is derived
143: for the coupled CDW system and mapped onto the appropriate \XY model according to whether
144: spatial and/or temporal fluctuations are taken into consideration. \Secref{sec:3dxy} is devoted to
145: the investigation of the phase transition in the 3D \XY model under the symmetry-breaking field,
146: which introduces the Z$_M$ symmetry to the system. Here two independent approaches are
147: employed: the $\epsilon$-expansion in \secref{sec:eexp} and the Monte Carlo method in \secref{sec:3dxymc}.
148: \Secref{sec:4dxy} presents the mean-field analysis of the 4D \XY model, demonstrating the first- and
149: second-order nature of the transition under the appropriate symmetry-breaking field.
150: In \secref{sec:cic} the effects of the inter-chain coupling in the presence of misfit are investigated
151: via Monte Carlo simulations, which discloses properties of the CI transition in the system.
152: Combining the results for the CI transition with those for the 3D and 4D \XY models
153: obtained in \secref{sec:3dxy} and \ref{sec:4dxy}, we construct in \secref{sec:pd} schematic
154: phase diagrams for general coupled CDW systems in the 3D space consisting of
155: the temperature, inter-chain coupling, and misfit. Finally, \secref{sec:c} gives a brief summary.
156:
157: %
158: % Effective Action
159: \section{Effective Action\label{sec:ea}}
160:
161: We consider a system of coupled near-commensurate CDW chains along the $z$ direction,
162: each of which is characterized by the commensurability factor $M$ and the position-dependent
163: misfit $\delta_\bfr$ with $\bfr=(x,y,z)$. On the $x$-$y$ plane, the chains are assumed for simplicity
164: to constitute a square array of lattice constant $a\, (\equiv 1)$.
165: Disregarding amplitude fluctuations of the complex CDW order parameter and
166: considering spatial and temporal fluctuations of the phase only,\cite{Gruner94}
167: we write the Hamiltonian in terms of the phase $\phi_\bfr$ of the order parameter at
168: position $\bfr$ and the momentum $p_\bfr = C\partial\phi_\bfr/\partial t$:
169: \begin{equation}
170: \label{eq:chainH}
171: \varH_1 = \int dz \sum_{x,y}
172: \left[ \frac{p_\bfr^2}{2C} + \frac{U_\parallel}{2}\!\left(\pde{\phi_\bfr}{z}{-}\delta_\bfr\right)^2
173: - V_0\cos{M\phi_\bfr} \right],
174: \end{equation}
175: where for the moment the inter-chain coupling has been omitted.
176: The first term and the second term correspond to the change of the total electron kinetic energy
177: due to temporal and spatial fluctuations, respectively,
178: whereas the third term represents the commensurability energy.
179: The dimensionless coupling constants $C$, $V_0$ and $U_\parallel$ depend on such detailed
180: microscopic structure of the system as the density of states at the Fermi level,
181: the effective electron mass renormalized due to the lattice vibration,
182: the electron-phonon coupling strength, and the cutoff energy.\cite{Gruner94}
183:
184: We now consider inter-chain tunneling of electrons between nearest neighboring chains at $(x,y)$ and
185: at $(x',y')$ on the $x$-$y$ plane, i.e., at the same position $z$;
186: this gives rise to the interaction of the form $U_\perp \cos(\phi_\bfr-\phi_{\bfr'})$,
187: where $U_\perp$ is the dimensionless inter-chain coupling constant and higher-order
188: harmonics have been disregarded.
189: With this included, the Hamiltonian $\varH$ for the coupled near-commensurate CDW chains reads
190: \begin{eqnarray}
191: \nonumber
192: \varH & = & \int dz
193: \left\{
194: \sum_{x,y}
195: \left[ \frac{p_\bfr^2}{2C}
196: + \frac{U_\parallel}{2}\!\left(\pde{\phi_\bfr}{z}{-}\delta_\bfr\right)^2
197: - V_0\cos{M\phi_\bfr} \right] \right.\\
198: \label{eq:cdwchains}
199: & & \left.\mbox\qquad\qquad
200: - \sum_{\nn{xy,x'y'}} U_\perp\cos(\phi_\bfr{-}\phi_{\bfr'})
201: \right\},
202: \end{eqnarray}
203: where the \textit{position} $\phi_\bfr$ and the conjugate \textit{momentum} $p_\bfr$ are considered
204: to observe the commutation relation $\left[\phi_\bfr,p_{\bfr'}\right]=i\delta_{\bfr,\bfr'}$,
205: suggesting the position representation $p_\bfr \dot= -i \partial/\partial\phi_\bfr$.
206: In the second summation $\nn{xy,x'y'}$ stands for the nearest neighbor pairs on the $x$-$y$ plane
207: at fixed $z$. Throughout this work, we set $\hbar\equiv 1$, $c\equiv 1$, and
208: the Boltzmann constant $k_B \equiv 1$ .
209:
210: Note that in the presence of the commensurability energy, the misfit cannot be simply gauged away
211: and introduces frustration to the system. For the time being we consider the case of
212: %>
213: strictly commensurate CDW systems without any misfit.
214: %<
215: To investigate the quantum phase transition in this case, driven by quantum fluctuations at zero temperature,
216: we follow the standard procedure\cite{StandardProcedure} to map a $d$-dimensional quantum system
217: to a $(d{+}1)$-dimensional classical system and obtain the corresponding effective action.
218: In \secref{sec:homo}, the system with negligible spatial fluctuations along the chain direction is considered
219: and the corresponding effective action at zero temperature is mapped onto the 3D \XY model,
220: where commensurability effects are described by the appropriate symmetry-breaking field.
221: \Secref{sec:general} discusses the system in the presence of spatial fluctuations.
222: We first pay attention to the classical limit, where temporal fluctuations may be neglected.
223: Here the system is intrinsically 3D, and the effective action is again mapped onto the 3D \XY model.
224: Next, with both spatial and temporal fluctuations considered, the appropriate effective action
225: at zero temperature is identified with the 4D \XY model.
226:
227: \subsection{Homogeneous Case\label{sec:homo}}
228:
229: Although it is in general expected that strong fluctuations are present,
230: we for the moment assume that spatial fluctuations along each chain are negligible,
231: which gives a two-dimensional (2D) system without $z$ dependence.
232: In the absence of the symmetry-breaking field due to commensurability,
233: \eqnref{eq:cdwchains} reduces to the 2D \XY model with kinetic energy
234: \begin{equation}
235: \label{eq:2dxy}
236: \varH = \sum_\bfr \frac{p_\bfr^2}{2C} - \sum_{\nn{\bfr,\bfr'}} U_\perp \cos(\phi_\bfr - \phi_{\bfr'}),
237: \end{equation}
238: where the summation in the second term is to be performed over all nearest-neighboring pairs
239: in the 2D space with $\bfr\equiv(x,y)$. The Hamiltonian in \eqnref{eq:2dxy} has been studied
240: in the context of quantum arrays of Josephson junctions.\cite{JJ}
241: In particular, at zero temperature the 2D quantum system in \eqnref{eq:2dxy} is well known
242: to map onto a 3D classical system via the standard lore.\cite{StandardProcedure}
243: Introducing the imaginary time $\tau$ axis and dividing the interval between $\tau=0$ and $\tau=T^{-1}$
244: into $N$ slices of equal width $\Delta\tau=1/NT$, in the zero temperature limit
245: $(T\rightarrow0)$ we arrive at the partition function of the anisotropic 3D \XY model
246: \begin{equation}
247: \label{eq:an3dxyZ}
248: \varZ =\oint\varD\phi\, \exp\left[ \sum_{\nn{\bfr,\bfr'}} K_{\bfr,\bfr'} \cos(\phi_\bfr{-}\phi_{\bfr'})
249: \right],
250: \end{equation}
251: where $\bfr\equiv(\tau,x,y)$ represents the position in the 3D space, consisting of the (imaginary) time
252: $\tau$ and the 2D space $(x,y)$. The anisotropic coupling is defined on each bond:
253: \begin{equation}
254: \nonumber
255: K_{\bfr,\bfr'} = \left\{
256: \begin{array}{ll}
257: \displaystyle C/\Delta\tau & \mbox{for } \bfr' = \bfr\pm\hat\bftau\Delta\tau,\\ [2mm]
258: U_\perp\Delta\tau & \mbox{for } \bfr' = \bfr\pm \ux \mbox{ or }\bfr\pm \uy.
259: \end{array}
260: \right.
261: \end{equation}
262: Strictly speaking, we should keep $\Delta\tau$ infinitesimal. Without affecting the universality,
263: however, we can rescale the space and time and obtain the partition function of an isotropic 3D \XY model
264: \begin{equation}
265: \varZ = \oint\varD\phi \, e^{-H}
266: \end{equation}
267: with the desired effective action
268: \begin{equation*}
269: -H = K \sum_{\nn{\bfr,\bfr'}} \cos(\phi_\bfr-\phi_{\bfr'}),
270: \end{equation*}
271: where $K^{-1}\equiv (CU_\perp)^{-1/2}$ measures the amount of quantum fluctuations.
272:
273: We then accommodate the commensurability effects at zero temperature, taking advantage of the fact that the
274: symmetry-breaking field term is diagonal in the position basis. Via the same procedure and rescaling as in the
275: absence
276: of the symmetry-breaking field, the effective action for the 3D \XY model obtains
277: \begin{equation}
278: \label{eq:hH}
279: -H = K \sum_{\nn{\bfr,\bfr'}} \cos(\phi_\bfr-\phi_{\bfr'})
280: + h \sum_\bfr \cos M\phi_\bfr,
281: \end{equation}
282: where the symmetry breaking field is given by $h\equiv V_0\Delta\tau = V_0\sqrt{C/U_\perp}$.
283:
284: \subsection{General Case\label{sec:general}}
285:
286: We now turn to the system in which spatial fluctuations in the $z$ direction are present
287: (but still without the misfit), and first consider the case of only spatial fluctuations,
288: with temporal fluctuations negligible. This corresponds to the classical limit in the sense that
289: the momentum and the position decouple and the momentum part, which can be integrated out,
290: does not affect the relevant physics. In this intrinsically 3D case, it is revealing to
291: resort to the discrete formulation and replace the integration $\int dz$ by the summation
292: $\sum_z \Delta z$ with sufficiently small $\Delta z$. Regarding spatial fluctuation-dependent
293: energy as the continuum form of the cosine interaction in the discrete representation,
294: we obtain from \eqnref{eq:cdwchains} the effective Hamiltonian for the 3D \XY model:
295: \begin{equation}
296: \label{eq:sH}
297: -\varH = \sum_{\nn{\bfr,\bfr'}} K_{\bfr,\bfr'} \cos(\phi_\bfr{-}\phi_{\bfr'}) + h\sum_\bfr \cos M\phi_\bfr,
298: \end{equation}
299: where $\bfr\equiv(x,y,z)$ represents lattice sites in the 3D space, and the coupling strength and the
300: symmetry-breaking field are given by
301: \begin{eqnarray*}
302: K_{\bfr,\bfr'} & \equiv &
303: \left\{
304: \begin{array}{ll}
305: U_\perp\Delta z & \mbox{for } \bfr'=\bfr\pm \ux \mbox{ or } \bfr\pm \uy \\ [2mm]
306: U_\parallel/\Delta z & \mbox{for } \bfr'=\bfr\pm \uz\Delta z\\
307: \end{array}\right. \\
308: h & \equiv & V_0\Delta z.
309: \end{eqnarray*}
310: Interestingly enough, \eqnsref{eq:hH} and (\ref{eq:sH}) show that both the (zero-temperature)
311: quantum phase transition in the absence of spatial fluctuations and the (finite-temperature)
312: classical phase transition in the absence of temporal fluctuations are described by the same 3D \XY model
313: under symmetry breaking fields. Quantum fluctuations in the former play the role of thermal fluctuations
314: in the latter, tending to restore symmetry.
315:
316: We finally consider the general case, where both spatial and temporal fluctuations are significant.
317: Adding the kinetic energy term to the 3D effective action in \eqnref{eq:sH} and following the same procedure
318: as that in \secref{sec:homo} at zero temperature, we obtain the 4D \XY model with the effective action
319: \begin{equation}
320: \label{eq:4dxy}
321: -H = \sum_{\nn{i,i'}} \varK_{i,i'} \cos(\phi_i-\phi_{i'}) + h \sum_i \cos M\phi_i,
322: \end{equation}
323: where $i\equiv(\tau,x,y,z)$ denotes 4D space-time lattice sites and the anisotropic coupling
324: and the symmetry-breaking field are given by
325: \begin{eqnarray*}
326: \varK_{i,i'} &\equiv&
327: \left\{
328: \begin{array}{ll}
329: \displaystyle \frac{\Delta\tau}{\Delta z} U_\parallel & \mbox{ for } i' = i\pm\uz\Delta z\\ [2mm]
330: \displaystyle \frac{C}{\Delta z\Delta\tau} & \mbox{ for } i' = i\pm\hat\bftau\Delta\tau\\ [4mm]
331: \Delta z\Delta\tau U_\perp & \mbox{ for other nearest neighbors}
332: \end{array}
333: \right. \\
334: h &\equiv& V_0\Delta z \Delta\tau
335: \end{eqnarray*}
336:
337: In this manner, the coupled near-commensurate CDW systems (without misfit) can be described
338: by the appropriate \XY models under symmetry-breaking fields. Depending on whether spatial
339: and/or temporal fluctuations are present, the effective dimension of the system is determined to be
340: three or four: It is four if both fluctuations are significant at zero temperature and three otherwise.
341: The commensurability effects are described by the symmetry-breaking field,
342: which introduces Z$_M$ symmetry to the system. Such a symmetry-breaking field affects the
343: ground-state symmetry and is expected to be relevant in the sense of the renormalization group (RG)
344: theory. In the limit $M\rightarrow\infty$, however, the Z$_M$ symmetry is hardly distinguishable
345: from the underlying U(1) symmetry in the \XY model. Therefore we expect the phase transition
346: to be crucially dependent upon the commensurability factor $M$, and devote the next two sections
347: to the investigation of the phase transitions in the 3D and 4D \XY models under symmetry-breaking fields.
348:
349: %
350: % 3D \XY Model under Symmetry-Breaking Field
351: \section{3D \XY Model under Symmetry-Breaking Field\label{sec:3dxy}}
352:
353: In this section we investigate the phase transition in the 3D \XY model whose effective action is
354: given by \eqnref{eq:hH}.
355: In the absence of the symmetry-breaking field ($h=0$), the 3D \XY model has been studied both
356: analytically\cite{VortexLoop,Williams88,Shenoy89} and numerically,\cite{3DXYNumericalStudy}
357: revealing that vortex loops do exist and proliferate at the phase transition.
358: Accordingly, the topological scaling idea has been extended to the 3D transition with conventional
359: long-range order, with the scaling procedure of Ref.~\onlinecite{VortexScaling} generalized
360: appropriately for 3D directed loops.\cite{Williams88,Shenoy89}
361: One may then be tempted to extend the study of the 2D model in Ref.~\onlinecite{Jose77}
362: to incorporate the symmetry-breaking term in the 3D \XY model,
363: and combine recursion relations for the vortex fugacity $y_0$ and the field perturbation $y_h$
364: according to the duality relation between them.
365: However, in contrast to the 2D case, the geometric scaling of the coupling constant in three dimensions
366: results in nonzero values of $y_0$ and $y_h$ at the fixed point.\cite{Shenoy89}
367:
368: On the other hand, the $\epsilon$-expansion\cite{Wilson72} is well known to provide a mathematical
369: formalism for calculating critical exponents of the O($n$) model near four spatial dimensions,
370: allowing classification of the universality class of the system.
371: Here we extend the $\epsilon$-expansion approach to incorporate the effects of the symmetry-breaking field.
372: The $\epsilon$-expansion turns out to be useful only for small commensurability factor $M$;
373: we thus supplement its limitations with the Monte Carlo numerical method.
374: In \secref{sec:eexp}, the symmetry-breaking perturbation is treated within the $\epsilon$-expansion
375: and \secref{sec:3dxymc} is devoted to the Monte Carlo simulations of the 3D \XY model
376: under symmetry-breaking fields.
377:
378: %
379: % $\epsilon$-Expansion
380: \subsection{$\epsilon$-Expansion\label{sec:eexp}}
381:
382: We consider the $\epsilon$-expansion for the 3D \XY model, extending the original formulation for the O($n$)
383: model\cite{Wilson72} to incorporate the symmetry-breaking field.
384: To begin with, we employ the two-component continuous local variable or spin
385: $\bfS_\bfr=(S_\bfr^x,S_\bfr^y)$ at each 3D lattice site $\bfr$, the total number of which is denoted by $N$.
386: The constraint that each spin has
387: the unit magnitude is relaxed by the additional weight factor introduced to the partition function:
388: \begin{equation}
389: \varZ = \left(\prod_\bfr \int_{-\infty}^\infty d\bfS_\bfr\right) e^{-[H+W(\bfS_\bfr)]}
390: \end{equation}
391: with the weight factor ($u>0$)
392: \begin{equation*}
393: \exp[-W(\bfS)]=\exp\left[-\frac{1}{2}bS^2-uS^4\right],
394: \end{equation*}
395: which has been expanded up to $\varO(S^4)$. At the physical \XY fixed point the sixth-order term is less
396: relevant in dimension $d=4-\epsilon$ than $d=4$.\cite{Cardy96} With the identification
397: $S_\bfr^x = \cos\phi_\bfr$ and $S_\bfr^y=\sin\phi_\bfr$, the original action in \eqnref{eq:hH} is
398: written in terms of the continuous spin $\bfS_\bfr$:
399: \begin{eqnarray}
400: \nonumber
401: -H & = & \sum_{\bfr,\bfr'} K \bfS_\bfr\cdot\bfS_{\bfr'}
402: {+} \frac{h}{2} \sum_\bfr \left[ (2S^x_\bfr)^M {-} \frac{M}{1!}(2S^x_\bfr)^{M-2} \right.\\
403: \label{eq:eeH}
404: & & \left.\mbox{} \qquad\qquad\quad{+} \frac{M(M{-}3)}{2!}(2S^x_\bfr)^{M-4} {-} \cdots \right].
405: \end{eqnarray}
406: Note that for $M<4$ the order of the action does not exceed $\varO(S^4)$,
407: leading to the Ginzburg-Landau-Wilson (GLW) effective Hamiltonian $\varH=H+W$ to $\varO(S^4)$.
408:
409: We first examine the case $M=2$. In the momentum space representation, keeping only relevant terms to
410: $\varO(q^2)$ and scaling the spin variable according to $\bfS_\bfq = (Ka^{d+2})^{-1/2}\bfsigma_\bfq$,
411: we express the GLW Hamiltonian as
412: \begin{eqnarray}
413: \nonumber
414: -\varH & = & -\half \int_\bfq (r_x{+}q^2)\sigma^x_\bfq\sigma^x_{-\bfq}
415: -\half \int_\bfq (r_y{+}q^2)\sigma^y_\bfq\sigma^y_{-\bfq} \\
416: \label{eq:fa}
417: & & \quad\mbox{} + V_1\int_\bfq \int_{\bfq'} \int_{\bfq''}
418: \sigma^x_\bfq\sigma^x_{\bfq'}\sigma^x_{\bfq''}\sigma^x_{-\bfq-\bfq'-\bfq''} \\
419: \nonumber
420: & & \quad\mbox{} + 2V_2\int_\bfq \int_{\bfq'} \int_{\bfq''}
421: \sigma^x_\bfq\sigma^x_{\bfq'}\sigma^y_{\bfq''}\sigma^y_{-\bfq-\bfq'-\bfq''} \\
422: \nonumber
423: & & \quad\mbox{} +V_3\int_\bfq \int_{\bfq'} \int_{\bfq''}
424: \sigma^y_\bfq\sigma^y_{\bfq'}\sigma^y_{\bfq''}\sigma^y_{-\bfq-\bfq'-\bfq''},
425: \end{eqnarray}
426: where the coefficients are given by
427: \begin{align*}
428: & r_x =\frac{1}{Ka^2}(b-dK-2h) \\
429: & r_y =\frac{1}{Ka^2}(b-dK) \\
430: & V_1 = V_2= V_3= -u < 0
431: \end{align*}
432: and $\int_\bfq \equiv \int d\bfq/(2\pi)^d = (1/Na^d)\sum_\bfq$ with the lattice constant $a$
433: restored for clarity. Here it is shown that the $M=2$
434: symmetry-breaking field gives rise to anisotropy in the quadratic term, making $r_x$ less than $r_y$.
435: Following the standard procedure, we obtain the recursion relations to the leading order in $\epsilon\equiv4-d$:
436: \begin{eqnarray}
437: \nonumber
438: \pde{r_x}{l} &=& 2r_x - \frac{12C}{1+r_x}V_1 - \frac{4C}{1+r_y}V_2 \\
439: \nonumber
440: \pde{r_y}{l} &=& 2r_y - \frac{4C}{1+r_x}V_2 - \frac{12C}{1+r_y}V_3 \\
441: \label{eq:eerr}
442: \pde{V_1}{l} &=& \epsilon V_1 + \frac{36C}{(1+r_x)^2}V_1^2 + \frac{4C}{(1+r_y)^2}V_2^2 \\
443: \nonumber
444: \pde{V_2}{l} &=& \epsilon V_2 + \frac{16C}{(1+r_x)(1+r_y)}V_2^2 \\
445: \nonumber
446: & & \mbox{} + \frac{12C}{(1+r_x)(1+r_y)}V_1V_2 + \frac{12C}{(1+r_x)(1+r_y)}V_2V_3 \\
447: \nonumber
448: \pde{V_3}{l} &=& \epsilon V_3 + \frac{36C}{(1+r_x)^2}V_3^2 + \frac{4C}{(1+r_y)^2}V_2^2
449: \end{eqnarray}
450: with the spatial scale factor $l$ and $C\equiv2^{1-d}\pi^{d/2}/(d/2{-}1)!$.
451:
452: In the absence of the symmetry-breaking field ($h=0$), the parameter space reduces to the 2D space
453: $(r,V)$ since $r_x=r_y\equiv r$ and $V_1=V_2=V_3\equiv V$.
454: In this case the nontrivial fixed point of the recursion relation in \eqnref{eq:eerr} is
455: simply the \XY fixed point, given by
456: \begin{equation}
457: r^* = -\frac{\epsilon}{5},\quad V^* = -\frac{\epsilon}{40C}
458: \end{equation}
459: to $\varO(\epsilon)$. As the symmetry-breaking field is turned on, however, we have $r_x<r_y$
460: at the initial locus and need to examine the RG flow in the full five-dimensional (5D) parameter space
461: $(r_x,r_y,V_1,V_2,V_3)$. To investigate the stability of the \XY fixed point in the 5D parameter space,
462: we linearize \eqnref{eq:eerr} about this point and obtain a stability matrix which yields five eigenvalues
463: together with corresponding scaling exponents:
464: \begin{equation}
465: \label{eq:ev}
466: \begin{array}{c}
467: \displaystyle y_1=2-\frac{\epsilon}{5},\quad y_2=2-\frac{2\epsilon}{5},\quad \\
468: \displaystyle y_3 = -\epsilon,\quad y_4 = -\frac{\epsilon}{5},\quad y_5 = -\frac{4\epsilon}{5}.
469: \end{array}
470: \end{equation}
471: Since the exponents $y_3$, $y_4$, and $y_5$, which correspond to the three eigenvectors spanning
472: the 3D subspace $(V_1,V_2,V_3)$, are all strictly negative, it is concluded that all $V_i$'s (for $i=1,2,3$)
473: are irrelevant at the \XY fixed point. In contrast, both $y_1$ and $y_2$ are positive, indicating that
474: the \XY fixed point is unstable in the 2D subspace $(r_x,r_y)$. The initial locus,
475: which is kept off the \XY fixed point by the symmetry-breaking field, should
476: flow far off the \XY fixed point. Instead it is expected to flow toward the Ising fixed point located at
477: \begin{equation}
478: r_x^* = -\frac{\epsilon}{6},\quad r_y^* = \infty, \quad V_1^* = -\frac{\epsilon}{36C},\quad V_2^*=V_3^*=0.
479: \end{equation}
480: At this point it is more appropriate to take $r^{-1}_y$ as the scaling field, giving the scaling exponents
481: \begin{equation}
482: y_{r_x} = 2-\frac{\epsilon}{3},\qquad y_{r_y^{-1}} = -2.
483: \end{equation}
484: Thus the initial locus with $r_x<r_y$ indeed flows toward the Ising fixed point along the stable
485: $r_y$ direction. In this manner, the symmetry-breaking field for $M=2$ introduces anisotropy in
486: the quadratic term, making the \XY fixed point unstable and generating an Ising or two-state clock fixed point.
487: Revealed accordingly is a single second-order transition between two-state clock order and disorder.
488:
489: We next turn to the case $M=3$, where \eqnref{eq:eeH} shows that terms of linear and third order
490: in the $x$-component of the spin come into play; power counting suggests that these fields are
491: relevant near $d=4$. Owing to the anisotropy associated with the absence of $S^y$ terms, in particular,
492: the cubic term $({S^x})^3$ here may not be removed by mere shift and is not redundant,
493: in contrast to the case of the Ising model. The symmetry-breaking field is thus expected
494: to drive the transition between the disordered phase and the three-state clock ordered phase.
495:
496: Finally the symmetry-breaking field for $M=4$ introduces anisotropy in the quartic term:
497: \begin{equation}
498: \label{eq:M4}
499: \begin{array}{l}
500: V_1=-u + 8a^{d-4}K^{-2}h \\
501: V_2=V_3=-u\ .
502: \end{array}
503: \end{equation}
504: This leads the action in \eqnref{eq:fa} to be unstable for sufficiently large values of the field $h$,
505: making it necessary to consider higher-order terms in the weight function $W(\bfS)$ for stability.
506: Unfortunately, such calculation of higher cumulants is quite a formidable job, and it is
507: very difficult to extend the $\epsilon$-expansion to the case of a high commensurability factor.
508:
509: %
510: % Monte Carlo study
511: \subsection{Monte Carlo Simulations\label{sec:3dxymc}}
512:
513: This section presents the Monte Carlo study of the 3D \XY model in symmetry-breaking fields.
514: To estimate the critical temperatures and determine critical exponents, we have performed Monte Carlo
515: simulations at ``temperatures'' (i.e., quantum or temporal fluctuations) ranging from
516: $K^{-1}=0.5$ to $K^{-1}=5$ on lattices of linear size $L=4$ up to $L=32$
517: for several commensurability factors and field strengths.
518: Measured in simulations are the order parameter $m$ and the susceptibility $\chi$ defined to be
519: \begin{equation}
520: \begin{array}{rcl}
521: m & \equiv & \displaystyle \avg{\left|\rcp{N}\sum_\bfr e^{i\phi_\bfr}\right|} \\ [4mm]
522: \chi & \equiv & \displaystyle \avg{\left|\rcp{N}\sum_\bfr e^{i\phi_\bfr}\right|^2}
523: - \avg{\left|\rcp{N}\sum_\bfr e^{i\phi_\bfr}\right|}^2,
524: \end{array}
525: \end{equation}
526: where $N$ is the number of sites.
527: We have employed the multiple histogram method\cite{MonteCarlo} to interpolate the
528: quantities calculated sparsely in a given range of the temperature to any temperature inside the range.
529: It not only saves a great deal of computing time but also provides the values of quantities
530: at arbitrary temperatures for finite-size scaling, resulting in critical exponents of better accuracy.
531: In fact we have obtained the best collapse of the scaling function such as
532: \begin{equation}
533: \tilde m (L^{1/\nu}t) = L^{\beta/\nu} m(t)
534: % \tilde \chi (L^{1/\nu}t) = L^{-\gamma/\nu}\chi(t)
535: \end{equation}
536: with the reduced temperature $t\equiv T/T_c-1 \equiv K_c/K -1$,
537: by minimizing the measure of error\cite{MonteCarlo}
538: \begin{equation}
539: \sigma_m^2 \equiv \rcp{x_{max}{-}x_{min}} \int_{x_{min}}^{x_{max}} \!\!\!dx
540: \left[\left\langle\!\left\langle\tilde m^2(x)\right\rangle\!\right\rangle
541: - \left\langle\!\left\langle\tilde m(x)\right\rangle\!\right\rangle^2 \right]
542: % \sigma_\chi^2 \equiv \rcp{x_{max}{-}x_{min}} \int_{x_{min}}^{x_{max}}
543: % \!\!\!dx \left[\left\langle\!\left\langle\tilde\chi^2(x)\right\rangle\!\right\rangle
544: % - \left\langle\!\left\langle\tilde\chi(x)\right\rangle\!\right\rangle^2 \right]
545: \end{equation}
546: over the critical temperature and exponents.
547: Here $\left\langle\!\left\langle\cdots\right\rangle\!\right\rangle$ stands for
548: the average over the lattice size and similar relations for the susceptibility have
549: also been considered.
550: %%>
551: We have taken the range $[x_{min}, x_{max}] = [-\Delta x, \Delta x]$ and attempted
552: a series of collapses as $\Delta x$ is diminished. The result has then been extrapolated
553: to the limit $\Delta x\rightarrow 0$.
554:
555: \begin{figure}
556: \centering
557: \includegraphics[width=8cm]{Fig1.eps}
558: \caption{Data collapse of the order parameter $m$ for three different values of the system size $L$,
559: with the best-estimated critical temperature and exponents given in the text.
560: The inset shows the dependence of the error $\sigma_m^2$ on the value of $\nu$,
561: with $T_c$ and $\beta$ fixed at their best estimated values.
562: The data have been taken for $M=2$ and at $h=2.7$; typical error bars are not larger than the symbol sizes.}
563: \label{fig:3dxys}
564: \end{figure}
565: \Figref{fig:3dxys} presents typical collapse of the
566: scaling function with the best-estimated critical temperature $T_c$ and exponents $\beta$ and $\nu$;
567: nice collapse behavior can be observed near the critical temperature $(t=0)$.
568: The inset in \figref{fig:3dxys} discloses how the corresponding error $\sigma_m^2$ depends on
569: the value of the exponent $\nu$. We have thus estimated the error in the obtained critical exponent
570: from the standard deviation of the minimizing values over the sample ensemble.
571: %%<
572: In this way, the phase transition for $M=2$ is found to be of the second order with the critical exponents
573: $\nu=0.63\pm0.01$ and $\beta=0.34\pm0.04$ for $h = 1.2$ and
574: $\nu=0.627\pm0.004$ and $\beta=0.32\pm0.01$ for $h = 2.7$.
575: These results coincide perfectly with the known critical exponents for the 3D Ising model:
576: $\nu=0.630$ and $\beta=0.324$, thus demonstrating the validity of the $\epsilon$-expansion
577: analysis in \secref{sec:eexp}.
578: %%>
579: Here it is of interest that the same universality class was also reported in the quantum Monte Carlo study
580: of a tight-binding model of spinless fermion chains coupled by intra- and inter-chain Coulomb
581: interactions.\cite{Scalapino86} The tight-binding model analysis, though being restricted to the $M=2$ case,
582: could incorporate amplitude fluctuations of the CDW order parameter directly.
583: Indeed the accordance between the two analyses manifests that fluctuations in the phase (rather than in the
584: amplitude) mostly determine the nature of the transition.
585: %%<
586:
587: For $M=3$, on the other hand, we have found no evidence for the first-order transition,
588: and obtained the estimation $\nu=0.596\pm0.007$ and $\beta=0.3\pm0.01$;
589: this is to be compared with the long-standing controversy as to whether the transition
590: %to the three-state clock ordered state
591: in the 3D three-state Potts model is of the first order or continuous.\cite{PottsModel}
592: It is also of interest that the critical behavior for $M\ge4$ is similar to that of the 3D \XY model
593: without any symmetry-breaking field. \Figref{fig:3dxytc} shows the critical temperature
594: (i.e., the critical fluctuation strength) $K_c$ versus the field strength $h$ for $2\le M\le5$.
595: It is observed that for $M\le3$ the critical temperature increases with the field strength
596: while for $M\ge4$ the commensurability energy appears to have no effect on the critical temperature,
597: in the range probed. Such critical behavior for $M\ge4$, similar to that of the ordinary 3D
598: \XY model, reflects that, as noted in \secref{sec:ea}, the Z$_M$ symmetry for large $M$ is
599: indiscernible from the U(1) symmetry underlying in the \XY model.
600: % and the commensurability is important in determining the nature of the transition.
601: %
602: \begin{figure}
603: \centering
604: \includegraphics[width=8cm]{Fig2.eps}
605: \caption{Critical temperature $K_c ^{-1}$ in the 3D \XY model under symmetry-breaking fields
606: versus the symmetry-breaking field strength $h$, for the commensurability factor $M=2, 3, 4$, and 5.
607: Typical error bars are smaller than the symbol size and lines are merely guides to eyes.}
608: \label{fig:3dxytc}
609: \end{figure}
610:
611: %
612: % 4D \XY Model under Symmetry Breaking Field
613: %
614:
615: \section{4D \XY Model under Symmetry-Breaking Field\label{sec:4dxy}}
616:
617: We now study the phase transition in the 4D \XY model, described by the effective action
618: in \eqnref{eq:4dxy}, through the use of the mean-field approximation.
619: Here the mean-field approximation, developed for superconducting arrays in
620: applied magnetic fields,\cite{Shih83} is expected to be accurate since the upper critical dimension
621: of the \XY model is given by $d_u=4$. With the space and time rescaled appropriately,
622: the (mean-field) self-consistent equation reads
623: \begin{equation}
624: \label{eq:ssc}
625: \avg{e^{i\phi}} = \varZ_{MF}^{-1} \int_0^{2\pi} d\phi\, e^{i\phi} e^{-H_{MF}},
626: \end{equation}
627: where $\varZ_{MF}$ is the partition function
628: \begin{equation*}
629: \varZ_{MF} = \int_0^{2\pi} d\phi\, e^{-H_{MF}}
630: \end{equation*}
631: corresponding to the mean-field action
632: \begin{equation}
633: -H_{MF} = 4K(\avg{\cos\phi} \cos\phi + \avg{\sin\phi}) \sin\phi + h \cos M\phi
634: \end{equation}
635: with the rescaled coupling constant $K\equiv \sqrt{CU_\perp}$ and field $h\equiv V_0 \sqrt{C/U_\perp}$.
636:
637: In the absence of the symmetry-breaking field, the self-consistent equation leads to the equation of state
638: for the order parameter $m\equiv\sqrt{\avg{\cos\phi}^2+\avg{\sin\phi}^2}$:
639: \begin{equation}
640: m = \frac{I_1(4Km)}{I_0(4Km)},
641: \end{equation}
642: where $I_n$ is the $n$th modified Bessel function. The system is in the disordered phase for
643: $K<K_c(h{=}0)=1/2$ characterized by $m=0$;
644: beyond $K_c(h{=}0)$ the ordered phase with $m\ne0$ is favored.
645: Similarly, the $M=1$ case can be analyzed easily, where arbitrarily small but positive $h$
646: results in $\avg{\cos\phi}>0$ and $\avg{\sin\phi}=0$. The equation of state is thus
647: \begin{equation}
648: \avg{\cos\phi} = \frac{I_1(4K\avg{\cos\phi}+h)}{I_0(4K\avg{\cos\phi}+h)}.
649: \end{equation}
650:
651: For larger values of $M$, \eqnref{eq:ssc} may be solved numerically with the parameters $K$ and $h$ varied.
652: It is found that there exists a field-dependent transition coupling strength $K_c(h)$ below which
653: \eqnref{eq:ssc} bears only the null solution $\avg{\cos\phi} = \avg{\sin\phi} = 0$.
654: As $K$ is increased beyond $K_c(h)$, nonzero stable solutions emerge.
655: Due to the Z$_M$ symmetry, i.e., the invariance of the action under the shift in the angle by $2\pi n/M$ for
656: any integer $n$, the solution of \eqnref{eq:ssc} can be expressed in terms of the $M$-state clock order
657: parameter $m(K,h)$:
658: \begin{equation}
659: \avg{e^{i\phi}} = m(K,h) e^{2\pi i n/M}
660: \end{equation}
661: with integer $n=0,1,\ldots,M-1$. The critical coupling strength $K_c(h)$ is defined to be the largest value
662: of $K$ satisfying $m(K,h)=0$.
663:
664: \begin{figure}
665: \centering
666: \includegraphics[width=8cm]{Fig3a.eps}\\ (a)\\[1cm]
667: \includegraphics[width=8cm]{Fig3b.eps}\\ (b)\\[1cm]
668: \includegraphics[width=8cm]{Fig3c.eps}\\ (c)
669: \caption{Order parameter $m$ as a function of the coupling $K$ and the field strength $h$ for (a) $M=2$,
670: (b) $M=3$, and (c) $M=4$. Main graphs show $m$ versus $K$ for $h = 0(+), 0.5(\times),
671: 1(\square), 5(\bigcirc)$, and $10(\bigtriangleup)$ from below. Insets display $m(K,h)$ versus $h$ for
672: various values of $K$: From below, (a) $K = 0.275, 0.3, 0.375, 0.5$, and $0.75$;
673: (b) $K = 0.47, 0.475, 0.4875, 0.5$, and $0.75$. The overall behavior of the order parameter for $M>4$
674: is the same as that for $M=4$.}
675: \label{fig:mforder}
676: \end{figure}
677: %
678: \Figref{fig:mforder} shows how the order parameter $m$ depends on the parameters $K$ and $h$
679: for different values of $M$. It is observed that $m$ in general increases rapidly from zero as $K$ exceeds $K_c$
680: and then saturates to unity, with the increase more rapid for larger $h$.
681: While for $M=2$ and $M\ge4$ the transition is found to be of the second order,
682: numerical results for $M=3$ indicate a first-order transition, which is expected from the appearance
683: of the third-order term in the $\epsilon$-expansion. For $M\le3$, the critical coupling strength $K_c(h)$,
684: starting from $K_c(h{=}0)=1/2$, decreases with $h$ almost exponentially to the asymptotic values,
685: 0.250 and 0.462 for $M=2$ and for $M=3$, respectively. For $M\ge4$, on the other hand,
686: the transition point does not depend on the strength of the symmetry-breaking field,
687: giving $K_c=1/2$ regardless of $h$. Again manifested is the crucial role of commensurability
688: in the phase transition.
689:
690: The resulting phase diagram for the 4D \XY model under the symmetry-breaking field is displayed in
691: \figref{fig:mfpd}, where the boundary $K_c^{-1}$ versus $h$ is plotted for different values of $M$.
692: When $K^{-1}<K_c^{-1}$, the Z$_M$ symmetry as well as the U(1) symmetry is broken and identified
693: is the $M$-state clock ordered phase, with one of the minima of $-h\cos M\phi$ favored.
694: When $K^{-1}>K_c^{-1}$, unlike the U(1) symmetry broken explicitly for $h\ne0$,
695: the Z$_M$ symmetry remains unbroken, leading to the disordered phase.
696: %
697: \begin{figure}
698: \centering
699: \includegraphics[width=8cm]{Fig4.eps}
700: \caption{Phase diagram for the 4D \XY model under the symmetry-breaking field, displaying the boundaries
701: between the $M$-state clock order phase (M) and the disordered phase (D) for different values of $M$.
702: Lines for $M = 2$ and $3$ are merely guides to eyes.}
703: \label{fig:mfpd}
704: \end{figure}
705:
706: %
707: % Commensurate-Incommensurate Transition
708: \section{Commensurate-Incommensurate Transition\label{sec:cic}}
709:
710: Up to the present, we have assumed the absence of misfit, so that only the transition between
711: the disordered phase and the $M$-state clock ordered phase, where commensurate CDWs are developed,
712: has been considered in several idealized cases described in \secref{sec:ea}.
713: However, the misfit, being a key ingredient to bring about the commensurate-incommensurate transition
714: in a near-commensurate CDW chain, must be taken into consideration in understanding various transitions
715: in the coupled CDW system.
716:
717: It is well known from the study of the 1D Frenkel-Kontorova model\cite{Choi00} at zero temperature that
718: a single near-commensurate CDW changes from the commensurate state to the incommensurate state
719: when the misfit exceeds a critical value depending on the commensurability energy. At finite temperatures
720: %>
721: the 1D CDW system is always in the incommensurate state.\cite{Sahni81}
722: %<
723: On the other hand, in the coupled CDW system, interactions between the CDW chains may alter the nature
724: of the transition. First of all, the effective dimensions of the system grow to three, giving rise to
725: the persistence of long-range order, as observed in the previous sections. Moreover, the CI transition itself
726: can also be affected by the inter-chain coupling in that the interactions favor either the commensurate state
727: or the incommensurate state, as explained below. To investigate the phase transition of the coupled CDW
728: system, we consider the Hamiltonian in \eqnref{eq:cdwchains} without temporal fluctuations,
729: which makes the problem simpler and allows to focus on the static properties only.
730:
731: At zero temperature the problem is rather simple to solve. The inter-chain coupling term in
732: \eqnref{eq:cdwchains} reaches the minimum when all the phases $\phi_\bfr$ in the $xy$-plane
733: become equal to each other. All the CDW chains, therefore, follow the same phase configurations
734: determined by the 1D Frenkel-Kontorova model, thus undergo the CI transition at the critical misfit
735: given by\cite{Choi00}
736: \begin{equation}
737: \delta_c = \frac{4}{\pi} \sqrt{\frac{V_0}{U_\parallel}},
738: \end{equation}
739: regardless of the inter-chain coupling strength $U_\perp$.
740:
741: The system at finite temperatures is examined by means of the Monte Carlo method. For this we discretize
742: the $z$-axis in \eqnref{eq:cdwchains} as in \secref{sec:general}, and obtain the lattice Hamiltonian
743: \begin{eqnarray}
744: \nonumber
745: \varH & = &
746: \sum_\bfr \frac{U_\parallel}{2} \!\left(\phi_{\bfr+\bfz}{-}\phi_\bfr{-}\delta\right)^2
747: - \sum_\bfr V_0\cos{M\phi_\bfr} \\
748: & & \mbox{}
749: - \sum_z \sum_{\nn{xy,x'y'}} U_\perp\cos(\phi_\bfr{-}\phi_{\bfr'}),
750: \end{eqnarray}
751: where $\Delta z$ has been absorbed into the coupling constants. Since the dimension along the CDW chain
752: is usually longer than the lateral dimensions, the lattice size $L_z$ along the $z$-axis is kept to be two times
753: larger than the other linear sizes in our simulations, and accordingly the inter-chain coupling $U_\perp$ is
754: restricted to be smaller than $U_\parallel$ to avoid inessential finite-size effects.
755: We further set the parameters to $U_\parallel=1$ and $V_0=0.2$ and sweep the inter-chain coupling
756: strength from 0.01 to 0.5 and the misfit $\delta$, which is assumed to be uniform throughout the system,
757: from zero to 0.8 beyond the critical misfit $\delta_c \,(=0.569)$.
758:
759: The order-disorder transition is described conveniently by the (incommensurate) order parameter
760: defined to be
761: \begin{equation}
762: m \equiv \avg{\left|\rcp{N}\sum_\bfr e^{i(\phi_\bfr - \delta z)}\right|}
763: \end{equation}
764: and its susceptibility $\chi$. Note that in the thermodynamic limit this order parameter vanishes
765: not only in the disordered phase but also in the commensurate phase;
766: in the system of finite size it may remain non-zero even in the perfectly commensurate state.
767: We also compute the soliton density $\rho_s$,\cite{Topic90} i.e., the soliton number per site per chain,
768: which characterizes the CI transition.
769: %Its value is bounded to $(1/L_z)(L_z{-}1)\delta/(2\pi/M)$ at the given values of misfit $\delta$ and
770: %commensurability factor $M$.
771: For the detection of any hysteresis, these two physical quantities have been measured in two different ways:
772: in the cooling-down process with randomized initial phase configurations and in the heating-up process
773: starting from the zero-temperature ground state. In each process the system has been equilibrated
774: sufficiently while the temperature is varied gradually.
775:
776: \begin{figure}[!t]
777: \centering
778: \includegraphics[width=8cm]{Fig5.eps}
779: \caption{Behavior of the order-disorder transition temperature $T_D$ with the misfit $\delta$
780: in the $M=2$ coupled CDW system for $U_\parallel=1$ and $V_0=0.2$.
781: Each symbol corresponds to a different value of the inter-chain coupling strength:
782: $U_\perp = 0.01(\square), 0.02(\blacksquare), 0.05(\circ), 0.1(\bullet), 0.2(\triangle),
783: 0.3(\blacktriangle), 0.4(\triangledown)$, and $0.5(\blacktriangledown)$.
784: Typical error bars are smaller than the symbol sizes and lines are merely guides to eyes.}
785: \label{fig:ct}
786: \end{figure}
787: %
788: Shown in \figref{fig:ct} is the dependence of the order-disorder transition temperature $T_D$
789: on the misfit $\delta$ in the $M=2$ CDW system at various inter-chain coupling strengths.
790: It is observed that the misfit tends to reduce the transition temperature:
791: The transition temperature first decreases as the misfit is increased from zero, then saturates
792: when the misfit reaches a value comparable to $\delta_c$.
793: Here stronger inter-chain coupling, which helps to increase the transition temperature,
794: in general weakens the effects of the misfit. Besides the transition temperatures, the critical exponents
795: also change with the misfit. For instance, the critical exponents in the system with $U_\perp=0.1$ are
796: found to be $\nu = 0.79\pm0.02$ and $\gamma=1.33\pm0.01$ at $\delta=0.2$;
797: $\nu = 0.77\pm0.03$ and $\gamma=1.64\pm0.05$ at $\delta=0.4$.
798: This suggests that the introduction of the misfit changes the nature of the ordered phase,
799: making it different from the 3D \XY ordered phase. It should be identified as the
800: incommensurate CDW phase, as demonstrated below.
801:
802: \begin{figure}
803: \centering
804: \includegraphics[width=8cm]{Fig6a.eps}\\ (a)\\[0.5cm]
805: \includegraphics[width=8cm]{Fig6b.eps}\\ (b)
806: \caption{(a) The order parameter $m$ and (b) the soliton density $\rho_s$ versus the temperature $T$
807: in the $M=2$ system of misfit $\delta=0.2$ and size $L_z=32$.
808: Each symbol corresponds to a different value of the inter-chain coupling strength as listed
809: in \figref{fig:ct}. The inset is an enlarged view of the soliton density in the range $0.1\le T\le0.5$.}
810: \label{fig:ccdw_d0.2}
811: \end{figure}
812: %
813: We now draw our attention to the CI transition occurring at lower temperatures.
814: In the absence of misfit ($\delta=0$) the soliton density $\rho_s$ is observed to vanish on average
815: at all temperatures below $T_D$, indicating that the CDWs formed are commensurate even in the
816: presence of thermal fluctuations. Such a commensurate phase is destroyed by the introduction
817: of misfit, even for $\delta<\delta_c$, with the help of thermal fluctuations.
818: \Figref{fig:ccdw_d0.2} exhibits the behaviors of the order parameter and of the soliton density
819: as the temperature is varied in the system with misfit $\delta=0.2$.
820: It is shown that the order parameter reaches its maximum at temperature near $T_D$ and decreases
821: to zero as the temperature is reduced while the soliton density begins to decrease from its maximum
822: value at $T\approx T_D$ and vanishes almost to zero at very low temperatures.
823: Nonzero values of the order parameter and of the soliton density together in the regime
824: $T\lesssim T_D$ correspond to the formation of incommensurate CDWs.
825: Thus the order-disorder transition at $T=T_D$ is identified as the transition between
826: the incommensurate CDW state and the disordered state.
827: In addition, the vanishing soliton density [see the inset in \figref{fig:ccdw_d0.2}] is a sign of
828: the commensurate CDW phase; this indicates the presence of a CI transition driven by thermal fluctuations.
829: The CI transition temperature $T_{IC}$ is defined to be the temperature at which the soliton density
830: %>
831: begins to be non-zero (in the thermodynamic limit).
832: %<
833: Although our computing ability disallow us to determine the precise value of $T_{IC}$,
834: \figref{fig:ccdw_d0.2}(b) manifests that $T_{IC}$ increases with the inter-chain coupling strength.
835: In this sense the inter-chain interactions favor the commensurate state. In contrast,
836: it is also observed that the inter-chain interactions prefer the incommensurate state at the misfit
837: values somewhat larger (but still smaller than $\delta_c$).
838:
839: \begin{figure*}
840: \centering
841: \begin{minipage}{8cm}
842: \centering
843: \includegraphics[width=\textwidth]{Fig7a.eps}
844: (a)
845: \end{minipage}
846: \begin{minipage}{8cm}
847: \centering
848: \includegraphics[width=\textwidth]{Fig7b.eps}
849: (b)
850: \end{minipage}\\[0.5cm]
851: \begin{minipage}{8cm}
852: \centering
853: \includegraphics[width=\textwidth]{Fig7c.eps}
854: (c)
855: \end{minipage}
856: \begin{minipage}{8cm}
857: \centering
858: \includegraphics[width=\textwidth]{Fig7d.eps}
859: (d)
860: \end{minipage}
861: \caption{The order parameter $m$ [(a) and (b)] and the soliton density $\rho_s$ [(c) and (d)] versus the
862: temperature $T$ in the $M=2$ system of misfit $\delta=0.4$ and size $L_z=32$.
863: The left figures [(a) and (c)] and the right ones [(b) and (d)] show the data obtained in the cooling
864: process and in the heating process, respectively. Each symbol corresponds to a different value of the
865: inter-chain coupling strength as listed in \figref{fig:ct}.}
866: \label{fig:ccdw_d0.4}
867: \end{figure*}
868: %
869: \Figref{fig:ccdw_d0.4} compares the order parameter and the soliton density at $\delta=0.4$, measured
870: in two different processes: the cooling process and the heating one.
871: Unlike the case $\delta=0.2$, a hysteresis is evident at $\delta=0.4$ (and also in the undisplayed case of
872: $\delta=0.3$), as shown in \figref{fig:ccdw_d0.4}(a) and (b) for the order parameter and
873: in \figref{fig:ccdw_d0.4}(c) and (d) for the soliton density.
874: In the heating process staring from the commensurate ground state at zero temperature,
875: the commensurate CDWs survive some thermal fluctuations and experience the CI transition at a finite
876: transition temperature. In contrast, as the temperature is lowered in the cooling process,
877: the order parameter keeps increasing and the soliton density saturates to a finite value,
878: unless the inter-chain coupling is sufficiently weak (i.e., for $U_\perp >0.02$).
879: This implies that the system still consists of the incommensurate CDWs at temperatures
880: where commensurate CDWs are supposed to be more stable. For weak coupling ($U_\perp \le 0.02$),
881: on the other hand, no discrepancy between the two processes is observed.
882:
883: It is of interest to compare the hysteresis observed here with the one reported in the specific heat
884: %>
885: around the CI transition in a number of incommensurate systems.\cite{Hysteresis}
886: %<
887: The latter is attributed to the effects of pinning of the incommensurate modulation due to
888: defects or impurities; our observation, in contrast, shows that the hysteresis can appear
889: even in the absence of defects, for which inter-chain interactions are responsible.
890: Namely, the inter-chain coupling operates in different ways depending on the process:
891: In the cooling process, correlations between CDW chains due to the inter-chain interactions
892: hinder each CDW chain from getting into the commensurate state.
893: On the contrary, in the heating process, the inter-chain correlations hold each chain close to
894: the zero-temperature ground state until thermal fluctuations become comparable to the inter-chain
895: interaction energy. This argument is supported by the observation that the order parameter and
896: the soliton density increase abruptly in a narrow region of the temperature, as shown in
897: \figref{fig:ccdw_d0.4}(b) and (d), and such discrepancy becomes manifest as the inter-chain coupling
898: becomes stronger. In addition, in the same narrow region the soliton number and the order parameter
899: fluctuate strongly while the energy fluctuations are relatively weak. This suggests a kind of
900: configurational fluctuations over many metastable states.
901:
902: For large values of the misfit ($\delta\gtrsim\delta_c$), neither commensurate CDW nor hysteresis
903: is observed and the soliton density does not vanish even at zero temperature in any process.
904: Only the transition separating the incommensurate CDW state from the disordered state is thus identified.
905:
906: We have also performed similar simulations for the commensurability factor $M=3$, only to obtain
907: results qualitatively the same as those for $M=2$ presented above. Quantitative differences observed
908: include that for $M=3$ the misfit affects the transition temperature more weakly
909: and the hysteresis takes place even at smaller misfit such as $\delta=0.2$.
910:
911: %
912: % Phase Diagram
913: %
914: \section{Phase Diagram\label{sec:pd}}
915:
916: Combining the results for the 3D and 4D \XY models under symmetry-breaking fields,
917: studied in \secsref{sec:3dxy} and \ref{sec:4dxy},
918: and those for the near-commensurate CDW model, studied in \secref{sec:cic},
919: we are ready to describe the phase transition in the coupled CDW system.
920: First, we focus on the system in the absence of any misfit $(\delta =0)$.
921: For this purpose, it is adequate to consider the 3D space consisting of the three parameters:
922: the temperature $T$, the strength of the symmetry-breaking field $h$,
923: and the amount of quantum or temporal fluctuations $K^{-1}$.
924: On each of the three planes ($T=0$, $K^{-1}=0$, and $h=0$) in the 3D space,
925: appropriate boundaries separating various phases can be plotted through the use of the known results,
926: as shown in \figref{fig:pd_xy}. The phase at a given point in the 3D space can then be speculated
927: by means of finite-size scaling in the region near $T=0$ or the semi-classical methods
928: in the $K^{-1}\approx0$ region.
929: %
930: \begin{figure*}
931: \centering
932: \begin{minipage}{8cm}
933: \centering
934: \includegraphics[width=.75\textwidth]{Fig8a.eps}\\
935: (a)
936: \end{minipage}
937: \begin{minipage}{8cm}
938: \centering
939: \includegraphics[width=.75\textwidth]{Fig8b.eps}\\
940: (b)
941: \end{minipage}\\[0.5cm]
942: \begin{minipage}{8cm}
943: \centering
944: \includegraphics[width=.75\textwidth]{Fig8c.eps}\\
945: (c)
946: \end{minipage}
947: \begin{minipage}{8cm}
948: \centering
949: \includegraphics[width=.75\textwidth]{Fig8d.eps}\\
950: (d)
951: \end{minipage}
952: \caption{Schematic phase diagram for the coupled commensurate CDW system in the $(T, h, K^{-1})$
953: space, depending on spatiotemporal fluctuations and commensurability:
954: the homogeneous case (a) for $M=4$ and (b) for $M\ge5$; the general case (c) for $M=2$ or $3$ and
955: (d) for $M\ge4$. Plotted on each plane are boundaries between various phases, including the disordered
956: phase (D), the $M$-state clock ordered phase (M), the algebraically ordered phase present in the 2D \XY
957: model (2D XY), and the 3D \XY ordered phase (3D XY).}
958: \label{fig:pd_xy}
959: \end{figure*}
960:
961: We first consider the homogeneous case, where spatial fluctuations along each chain are negligible,
962: and show the phase diagram in \figref{fig:pd_xy} for (a) $M=4$ and (b) $M\ge5$.
963: At zero temperature, the system is mapped onto the 3D \XY model as discussed in \secref{sec:ea},
964: and the corresponding phase diagram, obtained in \secref{sec:3dxy}, is sketched on the $h$-$K^{-1}$ plane.
965: On the other hand, in the absence of temporal fluctuations ($K^{-1}=0$), the system maps onto
966: the classical 2D \XY model under symmetry-breaking fields, for which the phase diagram in
967: Ref.~\onlinecite{Jose77} is drawn on the $h$-$T$ plane. Finally, on the $T$-$K^{-1}$ plane with $h=0$,
968: the system is described by the Hamiltonian in \eqnref{eq:2dxy} and expected to display the
969: Berezinskii-Kosterlitz-Thouless transition renormalized by quantum fluctuations.\cite{JJ,Comment}
970: Note that the cases $M=2$ and $3$ are not shown here. In this case presumably Ising/Potts critical lines
971: exist on the $h$-$T$ plane; however, it is not known how these lines connect up to the phase boundary
972: for $h=0$.
973:
974: In the general case with spatial fluctuations present, we obtain the phase diagram shown in
975: \figref{fig:pd_xy}(c) and (d) for $M\le3$ and $M\ge4$, respectively.
976: Owing to the additional dimension along the chain direction, the system maps onto the 3D \XY model
977: in the classical limit, i.e., on the $h$-$T$ plane, whereas 4D \XY model is obtained at zero temperature.
978: Accordingly, on the $h$-$K^{-1}$ plane, the phase diagram of the 4D \XY model obtained in
979: \secref{sec:4dxy} is drawn.
980: It is observed that the 3D \XY ordered phase does not survive the field and is replaced by the $M$-state clock
981: ordered phase for $h\ne0$. For $M\ge4$, however, the commensurability energy affects neither
982: the transition temperature nor the nature of the transition.
983:
984: We next take into consideration the effects of misfit and show in \figref{fig:pd_ccdw}
985: the schematic phase diagram in the 3D space consisting of $T$, $U_\perp^{-1}$, and $\delta$,
986: with $U_\parallel$ and $V_0$ fixed. At $\delta=0$, the system belongs to the same
987: university class as the 3D \XY model under the symmetry-breaking field,
988: for which the phase diagram is drawn on the $T-U_\perp^{-1}$ plane.
989: On the other hand, at zero temperature, each CDW chain undergoes the CI transition with the
990: critical misfit $\delta_c$ which is independent of the inter-chain coupling strength $U_\perp$.
991: The order-disorder transition temperature $T_D$ and the CI transition temperature $T_{IC}$
992: obtained in \secref{sec:cic} produce surfaces depicted by (thick) solid and dashed lines,
993: respectively, in the $(T,U_\perp^{-1},\delta)$ space.
994: The phase diagram shows that for $\delta<\delta_c$ the system undergoes double transitions
995: as the temperature is lowered: first, the order-disorder transition into the incommensurate
996: CDW phase and then the CI transition into the commensurate CDW phase.
997: As revealed in \secref{sec:cic}, a hysteresis takes place around the CI transition point,
998: which manifest itself more clearly as the inter-chain coupling becomes stronger.
999: %
1000: \begin{figure}
1001: \centering
1002: \includegraphics[width=8cm]{Fig9.eps}
1003: \caption{Schematic phase diagram for the coupled CDW system in the $(T, U_\perp^{-1}, \delta)$ space,
1004: with the effects of the misfit taken into account. The surfaces depicted by thick solid lines
1005: and dashed lines separate the incommensurate CDW phase (IC) from the disordered phase (D) and
1006: from the commensurate CDW phase (C), respectively.}
1007: \label{fig:pd_ccdw}
1008: \end{figure}
1009:
1010: %
1011: % Conclusion
1012: %
1013: \section{Conclusion\label{sec:c}}
1014:
1015: To study phase transitions in coupled CDW systems, we have mapped the systems at zero temperature onto
1016: three- or four dimensional \XY models, depending on the spatiotemporal fluctuations,
1017: under symmetry-breaking fields which arise from the commensurability energy.
1018: Such techniques as $\epsilon$-expansion, mean-field theory, and the Monte Carlo method
1019: have then been applied to the obtained \XY models. Revealed is a single second-order transition
1020: between the $M$-state clock order and disorder in both the three-dimensional and four-dimensional
1021: systems, except for the case $M=3$ in the four-dimensional system where the transition is of the first order.
1022: In particular, the commensurability with $M\ge4$ has been observed not to change the properties of
1023: the transition including the critical temperature and exponents.
1024: Combining these zero-temperature ($T=0$) results with the existing results in the absence of quantum
1025: (temporal) fluctuations ($K^{-1}=0$) or of the symmetry-breaking field ($h=0$), we have
1026: constructed boundaries separating various phases on the three planes ($T=0$, $K^{-1}=0$, and $h=0$)
1027: in the three-dimensional $(T, K^{-1}, h)$ space. The boundaries near $T=0$ and near $K^{-1} =0$
1028: can then be speculated through the use of finite-size scaling and semi-classical methods, respectively,
1029: thus giving a schematic phase diagram in the $(T, K^{-1}, h)$ space.
1030:
1031: We have also found via Monte Carlo simulations that the system with nonzero misfit undergoes
1032: a commensurate-incommensurate transition. The inter-chain interactions give rise to the correlations
1033: between neighboring CDWs in such a way that either the commensurate state or the incommensurate
1034: state is favored depending on the initial configuration: In the cooling process the CDWs remain
1035: incommensurate down to almost zero temperature while in the heating process a substantial
1036: amount of the commensurate CDWs survive thermal fluctuations.
1037:
1038: At strong thermal or quantum fluctuations, i.e., at high $T$ or small $K$, the system is in the
1039: disordered phase. No CDW is formed and the system is expected to be metallic. On the contrary,
1040: weak fluctuations (low $T$ and large $K$) favor the $M$-state clock ordered phase,
1041: in which commensurate CDWs are developed as long as $\delta<\delta_c$.
1042: Accordingly, the interaction between the periodicity of the CDW and the underlying lattice periodicity
1043: drives the collective excitation to develop a gap, and the system becomes insulating.
1044: At moderate fluctuations or for large misfit ($\delta>\delta_c$), on the other hand,
1045: incommensurate CDWs emerge. In this case the system may remain conducting through
1046: collective Fr\"ohlich conduction, i.e., via sliding of the CDWs.
1047: Usually, the conductivity via such collective modes is lower than that via uncondensated electrons
1048: in the disordered phase (without CDW). In particular the CDW may be pinned in the presence of impurities,
1049: sharply decreasing the conductivity. As the temperature is lowered, therefore, the system for weak
1050: quantum or thermal fluctuations becomes insulating via three possible routes, depending on
1051: the misfit and the inter-chain interaction:
1052: First, the commensurate CDW phase emerges directly from the high-temperature disordered phase;
1053: second, only the incommensurate CDW phase appears, reducing the conductivity;
1054: third, the incommensurate CDW phase appearing first is followed by the commensurate phase emerging
1055: via the commensurate-incommensurate transition.
1056:
1057: Note that beginning with the Hamiltonian in \eqnref{eq:chainH}, we have taken into account
1058: only phase fluctuations and disregarded amplitude fluctuations. The latter are in general irrelevant
1059: in the RG sense, expected not to affect nature of the phase transition. On the other hand,
1060: there still lacks conclusive understanding of the dynamic properties in various phases.
1061: It is thus desirable to consider the responses to external electromagnetic fields, and
1062: for example, compute the conductivity, which can be obtained from the current or the average
1063: momentum in the presence of appropriate misfit.
1064: Detailed investigation of such dynamic responses is left for further study.
1065:
1066: \section*{Acknowledgments}
1067:
1068: We thank G.S. Jeon for helpful discussions and acknowledge the partial support from the Korea
1069: Science and Engineering Foundation through the SKOREA Program and from the Ministry of Education
1070: of Korea through the BK21 Program.
1071:
1072: %
1073: % References
1074: %
1075: \begin{thebibliography}{99}
1076:
1077: \bibitem{Kagoshima87}
1078: S. Kagoshima, H. Nagasawa, and T. Sambongi, \textit{One-Dimensional Conductors}
1079: (Springer-Verlag, Berlin, 1987).
1080:
1081: \bibitem{Experiment}
1082: See, e.g., B. Gr\'{e}vin, Y. Berthier, G. Collin, and P. Mendels, Phys. Rev. Lett. {\bf
1083: 80}, 2405 (1998);
1084: H.W. Yeom, et al., \textit{ibid}. {\bf 82}, 4898 (1999);
1085: Y. Li, et al., \textit{ibid}. {\bf 83}, 3514 (1999).
1086:
1087: \bibitem{Coutinho81}
1088: S. Coutinho, P. Pitanga, and P. Lederer, Phys. Rev. B {\bf 23}, 4567 (1981).
1089:
1090: \bibitem{Gruner94}
1091: G. Gr\"uner, \textit{Density Waves in Solids} (Addison-Wesley, Reading, 1994).
1092:
1093: \bibitem{StandardProcedure}
1094: M. Wallin, E.S. S{\o}rensen, S.M. Girvin, and A.P. Young, Phys. Rev. B {\bf 49}, 12115
1095: (1994);
1096: S.L. Sondhi, S.M. Girvin, J.P. Carini, and D. Shahar, Rev. Mod. Phys. {\bf 69}, 315
1097: (1997).
1098:
1099: \bibitem{JJ}
1100: S. Doniach, Phys. Rev. B {\bf 24}, 5063 (1981);
1101: E. Simanek, \textit{ibid}. {\bf 32}, 500 (1985);
1102: J.V. Jos\'e, \textit{ibid}. {\bf 29}, 2836 (1984);
1103: S. Kim and M.Y. Choi, \textit{ibid}. {\bf 41}, 111 (1990). See also A. van Otterlo,
1104: K.-H. Wagenblast, R. Fazio, and G. Sch\"on, Phys. Rev. B {\bf 48}, 3316 (1993);
1105: B.J. Kim, J. Kim, S.Y. Park, and M.Y. Choi, \textit{ibid}. {\bf 56}, 395 (1997).
1106:
1107: \bibitem{VortexLoop}
1108: L. Onsager, Nuovo Cimento Suppl. {\bf 6}, 279 (1949);
1109: R.P. Feynman, in \textit{Progress in Low Temperature Physics} Vol. 1, edited by
1110: C. Gorter (North-Holland, Amsterdam, 1955), p. 17.
1111:
1112: \bibitem{Williams88}
1113: G.A. Williams, Phys. Rev. Lett. {\bf 59}, 1926 (1987); {\bf 61}, 1142(E) (1988).
1114:
1115: \bibitem{Shenoy89}
1116: S.R. Shenoy, Phys. Rev. B {\bf 40}, 5056 (1989).
1117:
1118: \bibitem{3DXYNumericalStudy}
1119: G. Kohring, R.E. Shrock, and P. Wills, Phys. Rev. Lett. {\bf 57}, 1358 (1986);
1120: G. Kohring and R.E. Shrock, Nucl. Phys. B {\bf 288}, 397 (1987);
1121: H. Kleinert, \textit{Gauge Fields in Condensed Matter} Vol. I (World Scientific,
1122: Singapore, 1989), pp. 528-530; A.P. Gottlob and M. Hasenbusch, Physica A {\bf 201}, 593 (1993).
1123:
1124: \bibitem{VortexScaling}
1125: J.M. Kosterlitz and D.J. Thouless, J. Phys. C {\bf 6}, 1181 (1973);
1126: J. Kosterlitz, \textit{ibid}. {\bf 7}, 1046 (1974).
1127: % A.P. Young, \textit{ibid}. {\bf 11}, L453 (1978).
1128:
1129: \bibitem{Jose77}
1130: J.V. Jos\'e, L.P. Kadanoff, S. Kirkpatrick, and D.R. Nelson, Phys. Rev. B {\bf 16},
1131: 1217 (1977).
1132:
1133: \bibitem{Wilson72}
1134: K.G. Wilson and M.E. Fisher, Phys. Rev. Lett. {\bf 28}, 240 (1972).
1135:
1136: \bibitem{Cardy96}
1137: J. Cardy, \textit{Scaling and Renormalization in Statistical Physics} (Cambridge
1138: University Press, Cambridge, 1996), pp. 100-101.
1139:
1140: \bibitem{MonteCarlo}
1141: M.E.J Newman and G.T. Barkema, \textit{Monte Carlo Methods in Statistical Physics}
1142: (Oxford University Press, New York, 1999), pp. 219-228.
1143:
1144: \bibitem{Scalapino86}
1145: D.J. Scalapino, R.L. Sugar, and W.D. Toussaint, Phys. Rev. B {\bf 34}, 6367 (1986).
1146:
1147: \bibitem{PottsModel}
1148: J. Rudnick, J. Phys. A {\bf 8}, 1125 (1975);
1149: D. Kim and R.J. Joseph, {\it ibid.} {\bf 8}, 891 (1975);
1150: T.W. Burkhardt, H.J.F. Knops, and M. den Nijs, {\it ibid}. {\bf 9}, L179 (1976);
1151: H.W.J. Bl\"ote and R.H. Swendsen, Phys. Rev. Lett. {\bf 43}, 799 (1979);
1152: B. Nienhuis, E.K. Riedel, and M. Schick, Phys. Rev. B {\bf 23}, 6055 (1981);
1153: J. Lee and J.M. Kosterlitz, {\it ibid}. {\bf 43}, 1268 (1991);
1154: P.D. Scholten and L.J. Irakliotis, {\it ibid}. {\bf 48}, 1291 (1993).
1155:
1156: \bibitem{Shih83}
1157: W.Y. Shih and D. Stroud, Phys. Rev. B {\bf28}, 6575 (1983).
1158:
1159: \bibitem{Choi00}
1160: See, e.g., M.Y. Choi, \textit{Physics of Complex Low-Dimensional Systems} (Hanul, Seoul, 2000), pp. 49-57.
1161:
1162: \bibitem{Sahni81}
1163: P.S. Sahni, Phys. Rev. B {\bf 23}, 1325 (1981).
1164:
1165: \bibitem{Topic90}
1166: B. Topi\u{c}, U. Haeberlen, and R. Blinc, Phys. Rev. B {\bf 42}, 7790 (1990).
1167:
1168: \bibitem{Hysteresis}
1169: R.A. Craven and S.F. Meyer, Phys. Rev. B {\bf 16}, 4583 (1977);
1170: R.M. Fleming, D.E. Moncton, D.B. McWhan, and F.J. DiSalvo, Phys. Rev. Lett. {\bf 45}, 576 (1981);
1171: J.-G. Yoon and S.-I. Kwun, Phys. Rev. B {\bf 35}, 8591 (1987).
1172:
1173: \bibitem{Comment}
1174: Here we disregard the possible existence of re-entrance, which has been a source of
1175: controversy. See Ref.~\onlinecite{JJ}.
1176:
1177: \end{thebibliography}
1178:
1179: \end{document}
1180: