1: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb,prb]{revtex4}
2: \usepackage{graphicx}% Include figure files
3: \usepackage{dcolumn}% Align table columns on decimal point
4: \usepackage{bm}% bold math
5:
6: %\documentclass[prb,amssymb,twocolumn,showpacs,superscriptaddress,floatfix]{revtex4}
7: %\usepackage{graphicx}
8: %\usepackage{bm}
9: %\usepackage{mathptmx}
10:
11:
12: \begin{document}
13:
14: \preprint{APS/123-QED}
15:
16: \title{Critical behavior of repulsive dimers on square lattices at $2/3$ monolayer coverage}
17:
18: \author{F. Rom\'a}
19: \affiliation{Centro At{\'{o}}mico Bariloche, R8402AGP San Carlos
20: de Bariloche, R\'{\i}o Negro, Argentina}
21: \affiliation{Departamento
22: de F\'{\i}sica, Universidad Nacional de San Luis, Chacabuco 917,
23: D5700BWS San Luis, Argentina}
24: \author{J. L. Riccardo}
25: \affiliation{Departamento de F\'{\i}sica, Universidad Nacional de
26: San Luis, Chacabuco 917, D5700BWS San Luis, Argentina}
27: \author{A.J. Ramirez-Pastor}
28: \affiliation{Departamento de F\'{\i}sica, Universidad Nacional de
29: San Luis, Chacabuco 917, D5700BWS San Luis, Argentina}
30:
31:
32: \date{\today}
33:
34: \begin{abstract}
35: Monte Carlo simulations and finite-size scaling theory have been
36: used to study the critical behavior of repulsive dimers on square
37: lattices at $2/3$ monolayer coverage. A ``zig-zag" (ZZ) ordered
38: phase, characterized by domains of parallel ZZ strips oriented at
39: $\pm 45^o$ from the lattice symmetry axes, was found. This ordered
40: phase is separated from the disordered state by a order-disorder
41: phase transition occurring at a finite critical temperature. Based
42: on the strong axial anisotropy of the ZZ phase, an orientational
43: order parameter has been introduced. All the critical quantities
44: have been obtained. The set of critical exponents suggests that
45: the system belongs to a new universality class.
46: \end{abstract}
47:
48: \pacs{68.35.Rh, 64.60.Cn, 68.43.De, 05.10.Ln}
49:
50: \maketitle
51:
52: %................................................................
53: \section{Introduction}
54:
55: The study of critical phenomena and phase transitions is a major
56: and long standing topic in statistical physics.
57: ~\cite{Stanley,Fisher,Kawasaki,Baxter,Yeomans,Goldenfeld,Domb}
58: Particularly, the two-dimensional lattice-gas model~\cite{Hill}
59: with repulsive interactions between the adparticles has received
60: considerable theoretical and experimental interest because it
61: provides the theoretical framework to study structural
62: order-disorder transitions occurring in many adsorbed monolayer
63: films.~\cite{Dash,Taub,Somorjai,Schick1,Binder0,Binder1,Landau1,Binder2,Landau2,Landau3,Schick2,Schick3,Schick4,Patrykiejew}
64: Most studies have been devoted to adsorption of particles with
65: single occupancy. The problem becomes considerably difficult when
66: particles occupy two adjacent lattice sites (dimers).
67: Consequently, there have been a few studies devoted to
68: order-disorder transitions associated to dimer adsorption with
69: repulsive lateral interactions. Among them, the structural
70: ordering of interacting dimers has been analyzed by A. J. Phares
71: et al.~\cite{Phares} The authors calculated the entropy of dimer
72: on semi-infinite $M \times N$ square lattice ($N \rightarrow
73: \infty$) by means of transfer matrix techniques. They concluded
74: that there are a finite number of ordered structures. As it arose
75: from simulation analysis,~\cite{SURFSCI3} only two of the
76: predicted structures survive at thermodynamic limit. In fact, in
77: Ref.~\onlinecite{SURFSCI3}, the analysis of the phase diagram for
78: repulsive nearest-neighbor interactions on a square lattice
79: confirmed the presence of two well-defined structures: a $c(4
80: \times 2)$ ordered phase at $\theta=1/2$ and a ``zig-zag" (ZZ)
81: order at $\theta = 2/3$, being $\theta$ the surface coverage.
82:
83: The thermodynamic implication of such a structural ordering was
84: demonstrated through the analysis of adsorption
85: isotherms,~\cite{LANG5} the collective diffusion
86: coefficient~\cite{SURFSCI2} and the configurational
87: entropy~\cite{LANG6} of dimers with nearest-neighbor repulsion.
88: Later, Monte Carlo (MC) simulations and finite-size scaling (FSS)
89: techniques have been used to study the critical behavior of
90: repulsive linear $k$-mers in the low-coverage ordered structure
91: (at $\theta=1/2$).~\cite{PRB4,PRB5} A ($2k \times 2$) ordered
92: phase, characterized by alternating lines, each one being a
93: sequence of adsorbed $k$-mers separated by $k$ adjacent empty
94: sites, was found. The critical temperature and critical exponents
95: were calculated. The results revealed that the system does not
96: belong to the universality class of the two-dimensional Ising
97: model. The study was extended to triangular lattices.~\cite{PRB6}
98: In this case, the exponents obtained for $k>1$ and
99: $\theta=k/(2k+1)$ are very close to those characterizing the
100: critical behavior of $k$-mers ($k>1$) on square lattices at
101: $\theta=1/2$.
102:
103: Recently, by using MC simulations and finite-scaling techniques,
104: R\.zysko and Bor\'owko have studied a wide variety of systems in
105: presence of
106: multisite-occupancy.~\cite{BORO1,BORO2,BORO3,BORO5,BORO4} Among
107: them, attracting dimers in the presence of energetic
108: heterogeneity;~\cite{BORO1} heteronuclear dimers consisting of
109: different segments A and B adsorbed on square
110: lattices;~\cite{BORO2,BORO3,BORO5} and trimers with different
111: structures adsorbed on square lattices.~\cite{BORO4} In these
112: leading papers, a rich variety of phase transitions was reported
113: along with a detailed discussion about critical exponents and
114: universality class.
115:
116: Summarizing, although there have been various studies for
117: monolayers at half coverage, to the author's knowledge, there are
118: no conclusive studies on the characteristics of the transition
119: phase of repulsive dimers on a square lattice at $2/3$ coverage.
120: In the present contribution we attempt to remedy this situation.
121: For this purpose, extensive MC simulations in the canonical
122: ensemble complemented by analysis using FSS techniques have been
123: applied. The FSS study has been divided in two parts. Namely, $1)$
124: a conventional FSS in terms of the normalized scaling variable
125: $\epsilon \equiv T/T_c -
126: 1$,~\cite{Fisher,Binder0,Privman,Privman1} where $T_c$ is the
127: critical temperature; and $2)$ an extended FSS,
128: ~\cite{GARTENHAUS,CAMPBELL} where $\sigma \equiv 1-T_c/T$, instead
129: of $\epsilon$, is used. Our results led the determination of the
130: critical temperature separating the transition between a
131: disordered state and the ZZ ordered phase occurring at $2/3$
132: coverage and the critical exponents characterizing the phase
133: transition.
134:
135: The outline of the paper is as follows: In Sec. II we describe the
136: dimer lattice-gas model. The order parameter and the simulation
137: scheme are introduced in Secs. III and IV, respectively. Finally,
138: the results and general conclusions are presented in Sec. V.
139:
140: %................................................................
141: \section{The model}
142:
143: In this section, the lattice-gas model for dimer adsorption is
144: described. The surface is represented as a simple square lattice
145: in two-dimensions consisting of $M=L \times L$ adsorptive sites,
146: where $L$ is the size of the system along each axis. The
147: homonuclear dimer is modelled as $2$ monomers at a fixed
148: separation, which equals the lattice constant $a$. In the
149: adsorption process, it is assumed that each monomer occupies a
150: single adsorption site and the admolecules adsorb or desorb as one
151: unit, neglecting any possible dissociation. The high-frequency
152: stretching motion along the molecular bond has not been considered
153: here.
154:
155: \begin{figure}[t]
156: \includegraphics[width=4cm,clip=true,angle=90]{figure1.eps}
157: \caption{Snapshot of the ordered phase for dimers at $\theta
158: =2/3$.} \label{figure1}
159: \end{figure}
160:
161: In order to describe the system of $N$ dimers adsorbed on $M$
162: sites at a given temperature $T$, let us introduce the occupation
163: variable $c_i$ which can take the following values: $c_i=0$ if the
164: corresponding site is empty and $c_i = 1$ if the site is occupied.
165: Under this consideration, the Hamiltonian of the system is given
166: by,
167: \begin{equation}
168: H = w \sum_{\langle i,j \rangle} c_i c_j - N w+ \epsilon_o
169: \sum_{i} c_i \label{h}
170: \end{equation}
171: where $w$ is the nearest-neighbor (NN) interaction constant which
172: is assumed to be repulsive (positive),$\langle i,j \rangle$
173: represents pairs of NN sites and $\epsilon_o$ is the energy of
174: adsorption of one given surface site. The term $Nw$ is subtracted
175: in Eq.~(\ref{h}) since the summation over all the pairs of NN
176: sites overestimates the total energy by including $N$ bonds
177: belonging to the $N$ adsorbed dimers.
178:
179: %................................................................
180: \section{Order parameter}
181:
182: Given the inherent anisotropy of the adparticles, it is convenient
183: to define a related order parameter. In this section, we will
184: briefly refer to a recently reported order parameter
185: $\delta$,~\cite{PRB5} which measures the orientation of the
186: admolecules in the ordered structure.
187:
188: Fig.~\ref{figure1} shows one of the possible configurations of the
189: ordered ZZ structure appearing for dimers at $2/3$ monolayer.
190: Though the degeneracy of this phase is high, the entropy per
191: lattice site tends to zero in the thermodynamic
192: limit~\cite{LANG6}. The figure suggests a simple way to build an
193: order parameter. In fact, any realization of the ZZ structure
194: implies the orientation of the particles along one of the lattice
195: axis.~\cite{foot1} Then, all the available configurations can be
196: grouped in two sets, according to this orientation. Taking
197: advantage of this property, we define the order parameter as:
198: \begin{equation}
199: \delta = \left | \frac{N_v-N_h}{N} \right | \label{fi2}
200: \end{equation}
201: where $N_v$ ($N_h$) represents the number of dimers aligned along
202: the vertical (horizontal) axis and $N= N_v + N_h$.
203:
204: When the system is disordered $(T>T_c)$, the two orientations
205: (vertical or horizontal) are equivalent and $\delta$ is zero. As
206: the temperature is decreased below $T_c$, the dimers align along
207: one direction and $\delta$ is different from zero. Thus, $\delta$
208: appears as a proper order parameter to elucidate the phase
209: transition.
210:
211: %................................................................
212: \section{Monte Carlo method}
213:
214: The lattices were generated fulfilling the following conditions:
215:
216: \begin{itemize}
217: \item[1)] The sites were arranged in a square lattice of side $L$
218: ($M = L \times L$), with conventional periodic boundary
219: conditions. \item[2)] Because the surface was assumed to be
220: homogeneous, the interaction energy between the adsorbed dimer and
221: the atoms of the substrate $\epsilon_o$ was neglected for sake of
222: simplicity. \item[3)] In order to maintain the lattice at $2/3$
223: coverage, $\theta=2N/M=2/3$, the number of dimers on the lattice
224: was fixed as $N = M/3$. \item[4)] Appropriate values of $L$ ($=60,
225: 72, 84, 96, 108$) were used in such a way that the ZZ adlayer
226: structure is not altered by boundary conditions.
227: \end{itemize}
228:
229: In order to study the critical behavior of the system, we have
230: used an exchange MC method.~\cite{Hukushima,Earl} As in Ref.
231: ~\onlinecite{Hukushima}, we build a compound system that consists
232: of $m$ noninteracting replicas of the system concerned. The $i$-th
233: replica is associated with a heat bath at temperature $T_i$ [or
234: $\beta_i=1/(k_B T_i)$, $k_B$ being the Boltzmann constant]. To
235: determine the set of temperatures, $\{T_i \}$, we set the highest
236: temperature, $T_1$, in the high-temperature phase where relaxation
237: (correlation) time is expected to be very short and there exists
238: only one minimum in the free energy space. On the other hand, the
239: lowest temperature, $T_m$, is set in the low-temperature phase
240: whose properties we are interested in. Finally, the difference
241: between two consecutive temperatures, $T_i$ and $T_{i+1}$ with
242: $T_i > T_{i+1}$, is set as $\Delta T = \left(T_{1} - T_m
243: \right)/(m-1)$ (equally spaced temperatures).
244:
245: Under these conditions, the algorithm to carry out the simulation
246: process is built on the basis of two major subroutines: {\it
247: replica-update} and {\it exchange}.
248:
249: \noindent {\it Replica-update}: Interchange vacancy-particle and
250: diffusional relaxation. The procedure is as follows: (a) One out
251: of the $m$ replicas is randomly selected (for example the $i$-th
252: replica). (b) A dimer and a pair of nearest-neighbor empty sites,
253: both belonging to the replica chosen in (a), are randomly selected
254: and their coordinates are established. Then, an attempt is made to
255: interchange its occupancy state with probability given by the
256: Metropolis rule, ~\cite{Metropolis}:
257: \begin{equation}
258: P = \min \left\{1,\exp\left( - \beta_i \Delta H \right) \right\}
259: \end{equation}
260: where $\Delta H$ is the difference between the Hamiltonians of the
261: final and initial states. (c) A dimer is randomly selected. Then,
262: a displacement is attempted (following the Metropolis scheme), by
263: either jumps along the dimer axis or reptation through a $90^o$
264: rotation of the dimer axis, where one of the dimer centers remains
265: in its position (interested readers are referred to Fig.~ 1 in
266: Ref. ~\onlinecite{SURFSCI2} for a more complete description of the
267: reptation mechanism). This procedure (diffusional relaxation) must
268: be allowed in order to reach equilibrium in a reasonable time.
269:
270: \noindent {\it Exchange}: Exchange of two configurations $X_i$ and
271: $X_{i'}$, corresponding to the $i$-th and $i'$-th replicas,
272: respectively, is tried and accepted with probability
273: $W\left(X_i,\beta_i| X_{i'},\beta_{i'}\right)$. In general, the
274: probability of exchanging configurations of the $i$-th and $i'$-th
275: replicas is given by, ~\cite{Hukushima}
276: \begin{equation}
277: W\left(X_i,\beta_i| X_{i'},\beta_{i'}\right)=\left\{
278: \begin{array}{cc}
279: 1 & {\rm for}\ \ {\Delta \leq 0} \\
280: \exp(-\Delta) & {\rm for}\ \ {\Delta>0}
281: \end{array}
282: \right.
283: \end{equation}
284: where $\Delta=\left( \beta_i - \beta_{i'} \right)\left[ H(X_{i'})
285: - H(X_{i}) \right]$. As in Ref. ~\onlinecite{Hukushima}, we
286: restrict the replica-exchange to the case $i'= i+1$.
287:
288: The complete simulation procedure is the following:
289:
290: \begin{itemize}
291:
292: \item[1)] Initialization.
293:
294: \item[2)] Replica-update.
295:
296: \item[3)] Exchange.
297:
298: \item[4)] Repeat from step 2) $m \times M $ times. This is the
299: elementary step in the simulation process or Monte Carlo step
300: (MCS).
301:
302: \end{itemize}
303:
304: The initialization of the compound system of $m$ replicas, step
305: 1), is as follows. By starting with a random initial condition,
306: the configuration of the replica $1$ is obtained after $n_1$
307: MCS$'$ at $T_1$ (MCS$'$ consists of $M$ realizations of the
308: replica-update subroutine). Second, for $i=\{ 2,....,m \}$, the
309: configuration of the $i$-th replica is obtained after $n_1$
310: MCS$'$ at $T_i$, taking as initial condition the configuration of
311: the replica to $T_{i-1}$. This method results more efficient than
312: a random initialization of each replica.
313:
314: The procedure 1)-4) is repeated for all lattice sizes. For each
315: lattice, the equilibrium state can be well reproduced after
316: discarding the first $n_2$ MCS. Then, averages are taken over
317: $n_{MCS}$ successive MCS. As it was mentioned above, a set of
318: equally spaced temperatures is chosen in order to accurately
319: calculate the physical observables in the close vicinity of $T_c$.
320:
321: The thermal average $\langle ... \rangle$ of a physical quantity
322: $A$ is obtained through simple averages:
323: \begin{equation}
324: {\langle A\rangle}=\frac{1}{n_{MCS}} \sum_{t=1}^{n_{MCS}} A
325: \left[X_i(t)\right].
326: \end{equation}
327: In the last equation, $X_i$ stands for the state of the $i$-th
328: replica (at temperature $T$). Thus, the specific heat $C$ (in
329: $k_B$ units) is sampled from energy fluctuations:
330: \begin{equation}
331: \frac{C}{k_B}= \frac{1}{(L k_B T)^2} [\langle H^2 \rangle -
332: \langle H \rangle ^2].
333: \end{equation}
334: The quantities related with the order parameter, such as the
335: susceptibility $\chi$, and the reduced fourth-order cumulant $U$
336: introduced by Binder, ~\cite{Binder2} are calculated as:
337: \begin{equation}
338: \chi = \frac{L^2}{k_BT} [ \langle \delta^2 \rangle - \langle
339: \delta \rangle^2]
340: \end{equation}
341: and
342: \begin{equation}
343: U = 1 -\frac{\langle \delta^4\rangle} {3\langle
344: \delta^2\rangle^2}. \label{cum}
345: \end{equation}
346:
347: Finally, in order to discuss the nature of the phase transition,
348: the fourth-order energy cumulant, $U_E$, was obtained as:
349: \begin{equation}
350: U_E=1 -\frac{\langle H^4\rangle} {3\langle H^2 \rangle^2}
351: \label{cume}
352: \end{equation}
353:
354: \begin{table}
355: \caption{Parameters used in the MC runs. For all lattice sizes we
356: have chosen $T_1=0.25$ and $T_m=0.15$ (the temperatures are in
357: units of $w/k_B$).}
358: \begin{ruledtabular}
359: \begin{tabular}{ccccc}
360: $L$ & $m$ & $n_1$ & $n_2$ & $n_{MCS}$ \\
361: \hline
362: $60$ & $101$ & $10^5$ & $10^5$ & $3$ x $10^5$ \\
363: $72$ & $121$ & $10^5$ & $10^5$ & $3$ x $10^5$ \\
364: $84$ & $141$ & $2$ x $10^5$ & $2$ x $10^5$ & $5$ x $10^5$ \\
365: $96$ & $161$ & $4$ x $10^5$ & $4$ x $10^5$ & $8$ x $10^5$ \\
366: $108$ & $141$ & $5$ x $10^5$ & $5$ x $10^5$ & $10^6$ \\
367: \end{tabular}
368: \end{ruledtabular}
369: \end{table}
370:
371: %................................................................
372: \section{Results and conclusions}
373:
374: The critical behavior of the present model has been investigated
375: by means of the computational scheme described in the previous
376: section and FSS techniques. The values of the parameters used in
377: each MC run are shown in Table I. In addition, all the simulation
378: calculations were obtained by averaging over $50$ MC runs.
379:
380: Because the replica temperatures were chosen equally spaced, the
381: acceptance probability of the replica-exchange decreases in the
382: critical temperature region, reaching a minimum whose value is
383: always greater than $50 \%$. The equilibration has been tested by
384: studying how the results vary when the simulation times $n_2$ and
385: $n_{MCS}$ are successively increased by factors of $2$. We
386: require that the last three results for all observables agree
387: within error bars. This simple method is shown to be useful to
388: test equilibration [see, for instance, Ref.~\onlinecite{KATZ}].
389: All calculations were carried out using the parallel cluster BACO
390: of Universidad Nacional de San Luis, Argentina. This facility
391: consists of 60 PCs each with a 3.0 GHz Pentium-4 processor.
392:
393: \begin{figure}
394: \includegraphics[width=6cm,clip=true]{figure2.eps}
395: \caption{(Color online) Size dependence of the order parameter as
396: a function of temperature. Inset: dependence of $\langle \delta
397: \rangle$ on $L^{-1}$ for different regimes of $T$ as indicated.
398: The error in each measurement is smaller that the size of the
399: symbols.} \label{figure2}
400: \end{figure}
401:
402: \begin{figure}
403: \includegraphics[width=6cm,clip=true]{figure3.eps}
404: \caption{(Color online) Size dependence of the susceptibility as a
405: function of temperature. The error in each measurement is smaller
406: that the size of the symbols.} \label{figure3}
407: \end{figure}
408:
409: \begin{figure}
410: \includegraphics[width=6cm,clip=true]{figure4.eps}
411: \caption{(Color online) $U$ versus $k_BT/w$, for different sizes.
412: From their intersections one obtained $k_BT_c/w$. In the inset,
413: the data are plotted over a wider range of temperatures.}
414: \label{figure4}
415: \end{figure}
416:
417: \begin{figure}
418: \includegraphics[width=6cm,clip=true]{figure5.eps}
419: \caption{(Color online) Temperature variation of $U_E$ for various
420: lattice sizes.} \label{figure5}
421: \end{figure}
422:
423: The conventional FSS implies the following behavior of $C$,
424: $\langle \delta \rangle$, $\chi$ and $U$ at criticality,
425: \begin{equation}
426: C=L^{\alpha/\nu} \tilde C(L^{1/\nu} \epsilon) \label{cal}
427: \end{equation}
428: \begin{equation}
429: \langle \delta \rangle = L^{-\beta/\nu} \tilde \delta(L^{1/\nu}
430: \epsilon) \label{phi}
431: \end{equation}
432: \begin{equation}
433: \chi= L^{\gamma/\nu}\tilde \chi(L^{1/\nu} \epsilon) \label{chi}
434: \end{equation}
435: \begin{equation}
436: U=\tilde U(L^{1/\nu} \epsilon) \label{cum}
437: \end{equation}
438: \noindent for $L \rightarrow \infty$, $\epsilon \rightarrow 0$
439: such that $L^{1/\nu} \epsilon $= finite, where ($ \epsilon \equiv
440: T/T_c - 1$). Here $\alpha$, $\beta$, $\gamma$ and $\nu$ are the
441: standard critical exponents of the specific heat ( $C \sim
442: |\epsilon|^{-\alpha}$ for $\epsilon \rightarrow 0$, $L \rightarrow
443: \infty $), order parameter ($\langle \delta \rangle \sim
444: -\epsilon^{\beta} $ for $\epsilon\rightarrow 0^-$, $L\rightarrow
445: \infty$),
446: susceptibility($\chi \sim |\epsilon|^\gamma
447: $ for $\epsilon \rightarrow 0$, $L\rightarrow \infty$) and
448: correlation length $\xi$ ($\xi \sim |\epsilon|^{-\nu}$ for
449: $\epsilon \rightarrow 0, L \rightarrow \infty$), respectively.
450: $\tilde C, \tilde \delta, \tilde \chi $ and $\tilde U$ are scaling
451: functions for the respective quantities. In the case of extended
452: FSS, ~\cite{GARTENHAUS,CAMPBELL} $\epsilon$ and $L$ in
453: eqs.~(\ref{cal}-\ref{cum}) are replaced by $\sigma \equiv 1-T_c/T$
454: and $L\sqrt{T}$, respectively.
455:
456: \begin{figure}[t]
457: \includegraphics[width=5.5cm,clip=true]{figure6a.eps}
458: \includegraphics[width=5.5cm,clip=true]{figure6b.eps}
459: \includegraphics[width=5.5cm,clip=true]{figure6c.eps}
460: \caption{(Color online) (a) Log-log plot of the size dependence of
461: the maximum values of derivatives of various thermodynamic
462: quantities used to determine $\nu$. (b) Log-log plot of the size
463: dependence of the maximum value of the susceptibility, the point
464: of inflection of the order parameter and the maximum value of the
465: derivative of the order parameter used to determine $\gamma$ and
466: $\beta$, respectively. (c) $K_c(L)$ vs. $L^{-1/\nu}$ from several
467: quantities as indicated. From extrapolation one obtains the
468: estimation of the critical temperature. In all cases, dotted lines
469: correspond to linear fits of the data and $L=60, 72,84,96,108$. }
470: \label{figure6}
471: \end{figure}
472:
473: \begin{figure}
474: \includegraphics[width=5.8cm,clip=true]{figure7a.eps}
475: \includegraphics[width=5.8cm,clip=true]{figure7b.eps}
476: \includegraphics[width=5.8cm,clip=true]{figure7c.eps}
477: \caption{(Color online) Conventional data collapsing for: (a) the
478: curves in Fig.~\ref{figure2}; (b) the curves in
479: Fig.~\ref{figure3}; and (c) the curves in Fig.~\ref{figure4}.}
480: \label{figure7}
481: \end{figure}
482:
483: \begin{figure}
484: \includegraphics[width=5.8cm,clip=true]{figure8a.eps}
485: \includegraphics[width=5.8cm,clip=true]{figure8b.eps}
486: \includegraphics[width=5.8cm,clip=true]{figure8c.eps}
487: \caption{(Color online) Same as Fig.~\ref{figure6} for the
488: extended scaling scheme.} \label{figure8}
489: \end{figure}
490:
491: \begin{figure}
492: \includegraphics[width=5.8cm,clip=true]{figure9a.eps}
493: \includegraphics[width=5.8cm,clip=true]{figure9b.eps}
494: \includegraphics[width=5.8cm,clip=true]{figure9c.eps}
495: \caption{(Color online) Same as Fig.~\ref{figure7} for the
496: extended scaling scheme.} \label{figure9}
497: \end{figure}
498:
499: We start with the calculation of the order parameter
500: (Fig.~\ref{figure2}), susceptibility (Fig.~\ref{figure3}) and
501: cumulant (Fig.~\ref{figure4}) plotted versus $k_BT/w$ for several
502: lattice sizes. Due to computational limitations,~\cite{foot2} the
503: curves in Fig.~\ref{figure2} do not clearly show the existence, at
504: thermodynamic limit, of a finite temperature below which the order
505: parameter is different from zero. In order to clarify this point,
506: the inset in Fig.~\ref{figure2} shows the dependence of $\langle
507: \delta \rangle$ on $L^{-1}$ for constant $T$. As it can be
508: observed, $\langle \delta \rangle$ tends to a finite value (zero)
509: for $T<T_c$ ($T>T_c$).
510:
511: From the intersections of the curves in Fig.~\ref{figure4} one
512: gets the estimation of the critical temperature. In this case,
513: $k_BT_c/w=0.182(1)$, which is in good agreement with the value
514: previously reported in the literature~\cite{SURFSCI3}. In
515: Ref.~\onlinecite{SURFSCI3}, the critical temperature was obtained
516: from the peaks of the curves of the specific heat versus
517: temperature (at fixed coverage) and coverage (at fixed
518: temperature). A more rigorous study was not possible due to the
519: lack of an adequate order parameter. In the inset, the data are
520: plotted over a wider range of temperatures, exhibiting the typical
521: behavior of the cumulants in presence of a continuous phase
522: transition. With respect to the value of the cumulant at the
523: transition temperature, $U^*$, this quantity was calculated by
524: plotting $U^*(L)$ vs $L^{-1/\nu}$ ~\cite{Ferren}, where the value
525: of $U^*(L)$ was obtained by fixing $K$($\equiv w/k_BT$) at our
526: estimate for $K_c$($ \equiv w/k_BT_c$) and looking at the cumulant
527: there (this is not shown here for brevity). In the thermodynamic
528: limit we obtained $U^*=0.57(1)$. This value is consistent with
529: the cumulant crossings shown in Fig.~\ref{figure4}.
530:
531: In order to discard the possibility that the phase transition is a
532: first-order one, the energy cumulants [Eq.~(\ref{cume})] have been
533: measured. As it is well-known, the finite-size analysis of $U_E$
534: is a simple and direct way to determine the order of a phase
535: transition~\cite{Binder2,Challa,Vilmayr}. Fig.~\ref{figure5}
536: illustrates the energy cumulants plotted versus $k_BT/w$ for
537: different lattice sizes ranging between $L=24$ and $L=60$. The
538: values of the parameters used in the MC runs were $m=21$,
539: $n_1=10^5$, $n_2=10^5$, $n_{MCS}=3 \times 10^5$, $T_1=0.25$ and
540: $T_m=0.15$. As it is observed, $U_E$ has the characteristic
541: behavior of a continuous phase transition: the minima in the
542: cumulants tend to $2/3$ as the lattice size is increased. This
543: indicates that the latent heat is zero in the thermodynamic limit,
544: which reinforces the arguments given in the paragraphs above.
545:
546: Next, the critical exponents will be calculated. As stated in
547: Refs.~\onlinecite{Ferren,Janke,Bin}, the critical exponent $\nu$
548: can be obtained by considering the scaling behavior of certain
549: thermodynamic derivatives with respect to the inverse temperature
550: $K$, for example, the derivative of the cumulant and the
551: logarithmic derivatives of $\langle \delta \rangle$ and $\langle
552: \delta^2 \rangle$. In Fig.~\ref{figure6}(a) we plot the maximum
553: value of these derivatives as a function of system size on a
554: log-log scale~\cite{foot3}. The results for $1/\nu$ from these
555: fits are given in the figure. Combining these three estimates we
556: obtain $\nu=1.16(3)$ (see Table II). Once we know $\nu$, the
557: critical exponent $\gamma$ can be determined by scaling the
558: maximum value of the susceptibility~\cite{Ferren,Janke}. Our data
559: for $\chi|_{\rm max}$ are shown in Fig.~\ref{figure6}(b). The
560: value obtained for $\gamma$ is indicated in the figure and listed
561: in Table II.
562:
563: On the other hand, the standard way to extract the exponent ratio
564: $\beta/\nu$ is to study the scaling behavior of $\langle \delta
565: \rangle$ at the point of inflection, i. e., at the point where $d
566: \langle \delta \rangle / d K $ is maximal. Since these points
567: should scale as usual, $\left(K^{\langle \delta \rangle}_{\rm
568: inf}- K_c \right)L^{1/\nu} \equiv \epsilon L^{1/\nu} ={\rm
569: const}$, we expect~\cite{Janke}
570: \begin{equation}
571: {\langle \delta \rangle |}_{\rm inf}=L^{-\beta/\nu} \tilde
572: \delta(\epsilon L^{1/\nu}) \propto L^{-\beta/\nu}, \label{betanu}
573: \end{equation}
574: where ${\langle \delta \rangle |}_{\rm inf}$ is the value of
575: $\langle \delta \rangle$ at the point of inflection. In addition,
576: since the derivative with respect to $K$ picks up a factor
577: $L^{1/\nu}$ from the argument of the scaling function $\tilde
578: \delta$,
579: \begin{equation}
580: \left. \frac{d \langle \delta \rangle}{d K} \right \vert _{\rm
581: max}=L^{\left(-\beta/\nu + 1/\nu \right)} \tilde \delta'(\epsilon
582: L^{1/\nu}) \propto L^{\left(1-\beta \right)/\nu}. \label{betabe}
583: \end{equation}
584:
585: The scaling of ${\langle \delta \rangle |}_{\rm inf}$ is shown in
586: Fig.~\ref{figure6}(b). The linear fit through all data points
587: gives $\beta^{\left({\langle \delta \rangle |}_{\rm
588: inf}\right)}=0.24(6)$. In the case of $d \langle \delta \rangle/d
589: K |_{\rm max}$ [see Fig.~\ref{figure6}(b)], the value obtained
590: from the fit is $\beta^{\left(d \langle \delta \rangle/d K |_{\rm
591: max}\right)}=0.20(8)$. Combining the two estimates, we obtain the
592: final value $\beta=0.22(6)$, which is indicated in the figure and
593: listed in Table II.
594:
595:
596: The finite-size scaling
597: theory~\cite{Binder0,Privman,Privman1,Ferren} allows for various
598: efficient routes to estimate $T_c$ from MC data. One of this
599: method, which was used in Fig.~\ref{figure4}, is from the
600: temperature dependence of $U(T)$ for different lattice sizes. An
601: independent procedure to determine $T_c$ will be used in the
602: following analysis. The method relies on the extrapolation of the
603: positions $K_c(L)$ of the maxima of various thermodynamic
604: quantities, which scale with system size
605: like~\cite{Binder0,Privman,Privman1,Ferren}
606: \begin{equation}
607: K_c(L)=K_c(\infty)+{\rm const}.L^{-1/\nu}. \label{nunu}
608: \end{equation}
609: Fig.~\ref{figure6}(c) shows a plot of $K_c(L)$ vs. $L^{-1/\nu}$
610: for the maxima of the slopes of $\langle \delta \rangle$, $\left(
611: d \langle \delta \rangle / d K \right)_{\rm max} $; $U$, $\left( d
612: U / d K \right)_{\rm max} $; $\ln \langle \delta \rangle$, $\left(
613: d \ln \langle \delta \rangle / d K \right)_{\rm max} $; $\ln
614: \langle \delta^2 \rangle$, $\left( d \ln \langle \delta^2 \rangle
615: / d K \right)_{\rm max} $, as well as of the susceptibility,
616: $\chi_{\rm max}$. The lines are fits of the data to
617: Eq.~(\ref{nunu}) with $\nu=1.16$. From extrapolation one obtains
618: $K^{(A)}_c(\infty)$ [or $k_BT^{(A)}_c(\infty)/w$] for the
619: different observables $A$. In this case,
620: $k_BT^{(U)}_c(\infty)/w=0.1824(12)$; $k_BT^{(\ln \langle \delta
621: \rangle)}_c(\infty)/w=0.1825(14)$; $k_BT^{(\ln \langle \delta^2
622: \rangle)}_c(\infty)/w=0.1825(16)$; $k_BT^{(\langle \delta
623: \rangle)}_c(\infty)/w=0.1820(20)$; and
624: $k_BT^{(\chi)}_c(\infty)/w=0.1823(5)$. Combining these estimates
625: we find a final value $k_BT_c/w=0.1823(7)$, which coincides,
626: within numerical errors, with the value calculated from the
627: crossing of the cumulants.
628:
629: Strong corrections to scaling were observed to be present at small
630: lattices and we excluded $L<60$ data from the calculations. On the
631: other hand, the $L \geq 60$ data are not good enough to include
632: corrections to scaling in the estimation of the critical exponents
633: and critical temperature. Thus, the quoted errors in our results
634: do not include systematic errors due to corrections to scaling
635: that possibly could affect our data.
636:
637: The scaling behavior can be further tested by plotting $\langle
638: \delta \rangle L^{\beta/\nu}$ vs $|\epsilon|L^{1/\nu}$, $\chi
639: L^{-\gamma/\nu}$ vs $\epsilon L^{1/\nu}$ and $U$ vs $|\epsilon
640: |L^{1/\nu}$ and looking for data collapsing. Using our best
641: estimates $k_BT_c/w=0.182$, $\nu=1.16$, $\beta=0.22$ and
642: $\gamma=1.98$, we obtain very satisfactory scaling as it is shown
643: in Fig.~\ref{figure7}. This study leads to independent controls
644: and consistency checks of the values of all the critical
645: exponents. The collapses in Fig.~\ref{figure7} were calculated by
646: following a conventional FSS scheme.
647:
648: In the next, the analysis of Figs.~\ref{figure6} and \ref{figure7}
649: is repeated, this time applying the extended FSS scheme mentioned
650: before. As it is shown in Fig.~\ref{figure8}, the values obtained
651: for the critical temperature $k_BT_c/w=0.1822(7)$ and the critical
652: exponents $\nu=0.18(3)$, $\beta=0.23(6)$ and $\gamma=2.04(5)$ (see
653: Table II) are in excellent agreement with those calculated using
654: the conventional FSS. In addition, we apply the extended FSS
655: scheme to collapse the data by plotting $U$, $\langle \delta
656: \rangle$ and $\chi$ in terms of the variables $\sigma \equiv
657: 1-T_c/T$ and $L\sqrt{T}$. The results are shown in
658: Fig.~\ref{figure9}. The behavior of the critical quantities for
659: the extended FSS scheme is consistent with the behavior observed
660: by following the conventional FSS scheme, which reinforces the
661: robustness of the scaling analysis introduced here.
662:
663: The critical exponents obtained by conventional and extended FSS,
664: along with the fixed point value of the cumulants, $U^*=0.57(1)$,
665: obtained in Fig.~\ref{figure4}, suggest that the phase transition
666: occurring for repulsive dimers on square lattices at $2/3$
667: monolayer coverage belongs to a new universality class. We say
668: ``suggest" because the lattice sizes studied here do not allow us
669: to exclude a more complex critical behavior, for example, the
670: presence of a tricritical point~\cite{BORO5,WILDING1,WILDING2}. In
671: order to elucidate this point, future studies on the whole phase
672: diagram, including variables as coverage and an external field
673: (coupled to the order parameter) breaking the orientational
674: geometry of the phase, will be carried out.
675:
676:
677: \begin{table}[t]
678: \caption{Critical exponents obtained from the fits shown in
679: Figs.~\ref{figure6} and \ref{figure8}.}
680: \begin{ruledtabular}
681: \begin{tabular}{ccc}
682: %exponent & conventional FSS & extended FSS & 2D Ising model \\
683: exponent & conventional & extended \\
684: & FSS & FSS \\
685: \hline
686: $\nu$ & $1.16(3)$ & $1.18(3)$ \\
687: $\beta$ & $0.22(6)$ & $0.23(6)$ \\
688: $\gamma$& $1.98(7)$ & $2.04(5)$ \\
689: $\alpha$& $\cdots$ & $\cdots$ \\
690: \end{tabular}
691: \end{ruledtabular}
692: \end{table}
693:
694: Finally, we will briefly refer to the specific heat exponent
695: $\alpha$. The ``roughness" of the curves of $C$ prevents a direct
696: determination of $\alpha$. However, the usual hyperscaling
697: relations inequalities of Rushbrooke, $\alpha+2\beta+\gamma \geq
698: 2$, and Josephson, $d \nu + \alpha \geq 2$ (being $d$ the
699: dimension of the space), predict a negative specific heat exponent
700: $\alpha \approx -0.3$. This finding is consistent with the
701: preliminary results obtained in the present study. Namely, even
702: though the fluctuations in the energy are of the order of the
703: statistical errors in the simulation results (reason for which the
704: data are not shown here), it is possible to observe that the
705: maxima of the curves of the specific heat remain practically
706: constant as $L$ is increased. In other words, the specific heat
707: seems not to diverge on approaching the transition. However, the
708: present data do not allow us to be conclusive on this point and
709: more work is needed to clearly elucidate the specific heats
710: behavior of the model.
711:
712: In summary, we have addressed the critical properties of repulsive
713: dimers on two-dimensional square lattices at $2/3$ coverage. The
714: results were obtained by using exchange MC simulations and FSS
715: theory. The choice of an adequate order parameter [as defined in
716: Eq.~(\ref{fi2})] along with the exhaustive study of FSS presented
717: here allow us $1)$ to confirm previous results in the
718: literature,~\cite{Phares,SURFSCI3} namely, the existence of a
719: continuous phase transition at $2/3$ coverage; $2)$ to calculate
720: the critical temperature characterizing this transition; and $3)$
721: to obtain the complete set of static critical exponents for the
722: reported transition. Though it is not possible to exclude the
723: existence of a more complex critical behavior, the results suggest
724: that the phase transition belongs to a new universality class.
725:
726: \acknowledgments This work was supported in part by CONICET,
727: Argentina, under Grant No. PIP 6294 and the Universidad Nacional
728: de San Luis, Argentina, under the Grants Nos. 328501 and 322000.
729:
730: %.....................................................................
731: \begin{thebibliography}{99}
732:
733: \bibitem{Stanley} H. E. Stanley, \textit{Introduction to Phase Transitions and Critical
734: Phenomena} (Oxford University Press, New York, 1971).
735:
736: \bibitem{Fisher} M. E. Fisher, \textit{Critical Phenomena} (Academic Press, London, 1971).
737:
738: \bibitem{Kawasaki} K. Kawasaki, \textit{Phase Transitions and Critical
739: Phenomena}, edited by C. Domb and M. S. Green (Academic Press,
740: London, 1972), Vol. 2, pag. 443.
741:
742: \bibitem{Baxter} R. J. Baxter, \textit{Exactly solved models in statistical mechanic} (Academic Press, London, 1982).
743:
744: \bibitem{Yeomans} J. M. Yeomans, \textit{Statistical Mechanics of Phase Transitions} (Clarendon Press, Oxford, 1992).
745:
746: \bibitem{Goldenfeld} N. Goldenfeld, \textit{Lectures on Phase Transitions and the Renormalization Group}
747: (Addison-Wesley, Reading, MA, 1992).
748:
749: \bibitem{Domb} \textit{Phase Transitions and Critical Phenomena}, edited by C. Domb and J. L. Lebowitz
750: (Academic Press, London, 2001), Vol. 19.
751:
752: \bibitem{Hill} T. L. Hill, J. Chem. Phys. {\bf 17}, 520 (1949).
753:
754: \bibitem{Dash} \textit{Phase Transitions in Surface Films}, edited by J. G. Dash and J. Ruvalds
755: (Plenum, New York, 1980).
756:
757: \bibitem{Taub} \textit{Phase Transitions in Surface Films 2}, edited by H. Taub, G. Torso, H. J. Lauter,
758: and S. C. Fain, Jr. (Plenum, New York, 1991).
759:
760: \bibitem{Somorjai} G. A. Somorjai and M. A. Van Hove, \textit{Adsorbed Monolayers on Solid Surfaces}
761: (Springer-Verlag, New York, 1979).
762:
763: \bibitem{Schick1} M. Schick, J. S. Walker and M. Wortis, Phys. Rev. B {\bf 16}, 2205 (1977).
764:
765: \bibitem{Binder0} K. Binder, {\it Applications of the Monte Carlo Method in Statistical
766: Physics. Topics in Current Physics} \rm (Springer, Berlin, 1984),
767: Vol. 36.
768:
769: \bibitem{Binder1} K. Binder and D. P. Landau, Phys. Rev. B {\bf 21}, 1941 (1980).
770:
771: \bibitem{Landau1} D. P. Landau, Phys. Rev. B {\bf 27}, 5604 (1983).
772:
773: \bibitem{Binder2} K. Binder and D. P. Landau, Phys. Rev. B {\bf 30}, 1477 (1984).
774:
775: \bibitem{Landau2} D. P. Landau and K. Binder, Phys. Rev. B {\bf 31}, 5946 (1985).
776:
777: \bibitem{Landau3} D. P. Landau and K. Binder, Phys. Rev. B {\bf 41}, 4633 (1990).
778:
779: \bibitem{Schick2} E. Domany, M. Schick, J. S. Walker, and R. B. Griffiths, Phys. Rev. B {\bf 18}, 2209 (1978).
780:
781: \bibitem{Schick3} E. Domany and M. Schick, Phys. Rev. B {\bf 20}, 3828 (1979).
782:
783: \bibitem{Schick4} M. Schick, Prog. Surf. Sci. {\bf 11}, 245 (1981).
784:
785: \bibitem{Patrykiejew} A. Patrykiejew, S. Sokolowski, and K. Binder,Surf. Sci. Rep. {\bf 37}, 207 (2000).
786:
787: \bibitem{Phares} A. J. Phares, F. J. Wunderlich, J. D Curley, and D. W.
788: Grumbine, Jr., J. Phys. A: Math. Gen. {\bf 26}, 6847 (1993).
789:
790: \bibitem{SURFSCI3} A. J. Ramirez-Pastor, J. L. Riccardo, and V. D.
791: Pereyra, Surf. Sci. {\bf 411}, 294 (1998).
792:
793: \bibitem{LANG5} A. J. Ramirez-Pastor, J. L. Riccardo, and V. D.
794: Pereyra, Langmuir {\bf 16}, 10167 (2000).
795:
796: \bibitem{SURFSCI2} M. S. Nazzarro, A. J. Ramirez-Pastor, J. L. Riccardo, and V. D.
797: Pereyra, Surf. Sci. {\bf 391}, 267 (1997).
798:
799: \bibitem{LANG6} F. Rom\'a, A. J. Ramirez-Pastor, and J. L.
800: Riccardo, Langmuir {\bf 16}, 9406 (2000).
801:
802: \bibitem{PRB4} F. Rom\'a, A. J. Ramirez-Pastor, and J. L.
803: Riccardo, Phys. Rev. B {\bf 68}, 205407 (2003).
804:
805: \bibitem{PRB5} F. Rom\'a, A. J. Ramirez-Pastor, and J. L.
806: Riccardo, Phys. Rev. B {\bf 72}, 035444 (2005).
807:
808: \bibitem{PRB6} P. M. Pasinetti, F. Rom\'a, J. L. Riccardo and
809: A. J. Ramirez-Pastor, Phys. Rev. B {\bf 74}, 155418 (2006).
810:
811: \bibitem{BORO1} M. Bor\'owko and W. R\.zysko, J. Colloid Interface Sci. {\bf 244}, 1
812: (2001).
813:
814: \bibitem{BORO2} W. R\.zysko and M. Bor\'owko, J. Chem. Phys. {\bf 117}, 4526 (2002).
815:
816: \bibitem{BORO3} W. R\.zysko and M. Bor\'owko, Surf. Sci. {\bf 520}, 151 (2002).
817:
818: \bibitem{BORO5} W. R\.zysko and M. Bor\'owko, Surf. Sci. {\bf 600}, 890 (2006).
819:
820: \bibitem{BORO4} W. R\.zysko and M. Bor\'owko, Phys. Rev. B {\bf 67}, 045403 (2003).
821:
822: \bibitem{Privman} V. Privman, \textit{Finite Size Scaling and Numerical Simulation of Statistical Systems}
823: (World Scientific, Singapore, 1990).
824:
825: \bibitem{Privman1} V. Privman, P. C. Hohenberg, and A. Aharony, \textit{Phase Transitions and Critical
826: Phenomena}, edited by C. Domb and J. L. Lebowitz (Academic Press,
827: New York, 1991), Vol. 14.
828:
829: \bibitem{GARTENHAUS} S. Gartenhaus and W. S. McCullough, Phys. Rev. B {\bf 38}, 11688 (1988).
830:
831: \bibitem{CAMPBELL} I. A. Campbell, K. Hukushima, and H. Takayama, Phys. Rev. Lett. {\bf 97}, 117202 (2006).
832:
833: \bibitem{foot1}
834: As in the case of the $(2k \times 2)$ ordered phase occurring for
835: linear $k$-mers on square lattices at $1/2$ monolayer coverage,
836: ~\cite{PRB5} the phase transition at $2/3$ monolayer coverage is
837: accomplished by a breaking of the translational and orientational
838: symmetries.
839:
840: \bibitem{Hukushima} K. Hukushima and K. Nemoto, J. Phys. Soc.
841: Jpn. {\bf 65}, 1604 (1996).
842:
843: \bibitem{Earl} D. J. Earl and M. W. Deem, Phys. Chem. Chem. Phys. {\bf 7}, 3910 (2005).
844:
845: \bibitem{Metropolis} N. Metropolis, A.W. Rosenbluth, M.N. Rosenbluth, A.H. Teller
846: and E. Teller, J. Chem. Phys. {\bf 21}, 1087 (1953).
847:
848:
849: \bibitem{KATZ} H. G. Katzgraber, M. K\"{o}rner, and A. P. Young, Phys. Rev. B {\bf
850: 73}, 224432 (2006).
851:
852: \bibitem{foot2}
853: As it is well-known, MC simulations of $k$-mers at equilibrium are
854: very time consuming. Consequently, the finite-size scaling study
855: was carried out for lattice sizes up to $L=108$, with an effort
856: reaching almost the limits of our computational capabilities.
857:
858: \bibitem{Challa} M. S. S. Challa, D. P. Landau and K. Binder, Phys. Rev. B {\bf 34}, 1841 (1986).
859:
860: \bibitem{Vilmayr} K. Vollmayr, J. D. Reger, M. Scheucher and K. Binder, Z. Phys. B: Condens. Matter {\bf 91}, 113 (1993).
861:
862: \bibitem{Ferren} A. M. Ferrenberg and D. P. Landau, Phys. Rev. B {\bf 44}, 5081 (1991).
863:
864: \bibitem{Janke} W. Janke, M. Katoot and R. Villanova, Phys. Rev. B {\bf 49}, 9644 (1994).
865:
866: \bibitem{Bin} K. Binder and E. Luijten, Physics Reports {\bf 344}, 179 (2001).
867:
868: \bibitem{foot3}
869:
870: In this paper, the temperature derivatives were taken by averaging
871: the slopes of two adjacent data points as follows: $\frac{1}{2}
872: \left[\frac{y_{i+1}-y_{i}}{x_{i+1}-x_{i}} +
873: \frac{y_{i}-y_{i-1}}{x_{i}-x_{i-1}} \right]$. Error bars were
874: estimated by propagation of errors from the last equation. On the
875: other hand, the coordinates of each maximum were calculated by
876: fittting a three-order polynomial to a set of between 15 and 20
877: data points around this critical value. Polynomials of order 2 and
878: 4 were also used as fitting functions and the results did not
879: change significantly.
880:
881:
882: \bibitem{WILDING1} N. B. Wilding, and P. Nielaba, Phys. Rev. E {\bf 53}, 926 (1996).
883:
884: \bibitem{WILDING2} N. B. Wilding, F. Schmid, and P. Nielaba, Phys. Rev. E {\bf 58},
885: 2201 (1998).
886:
887:
888: \end{thebibliography}
889: %.....................................................................
890: \end{document}
891:
892: By following a standard algorithm [W. H. Press, S. A. Teukolsky,
893: W. T. Vetterling, and B. P. Flannery, Numerical Recipes in C: The
894: art of scientific computing (Cambridge University Press, New York,
895: 1992)],
896:
897:
898: Finite-size corrections to scaling were observed to be present at
899: small lattices and we excluded $L<60$ data from the calculations.
900: On the other hand, the $L \geq 60$ The quoted error bars in our
901: estimates do not include systematic errors due to corrections to
902: scaling that possibly could affect our data.
903:
904: Here only statistical errors are given. It is difficult to
905: quantify systematic errors due to corrections to scaling.
906:
907: Unfortunately, our data for fully frustrated XY model for L > 14
908: turn out to be rather noisy.
909: