1: \documentclass[showpacs,preprintnumbers,amsmath,amssymb,superscriptaddress]{revtex4}
2: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb,superscriptaddress]{revtex4}
3: %\documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb,superscriptaddress]{revtex4}
4:
5: \usepackage{graphicx}
6: \usepackage{dcolumn}
7: \usepackage{bm}
8:
9: \begin{document}
10:
11: \preprint{J. Phys.: Condens. Matter \textbf{20}, 245102 (2008)}
12:
13: \title{Coexistence of hexatic and isotropic phases in two-dimensional Yukawa systems}
14:
15: \author{Wei-Kai Qi}
16: \affiliation{Institute of Theoretical Physics, Lanzhou University,
17: Lanzhou $730000$, China}
18:
19: \author{Shao-Meng Qin}
20: \affiliation{Institute of Theoretical Physics, Lanzhou University,
21: Lanzhou $730000$, China}
22:
23: \author{Xiao-Ying Zhao}
24: \affiliation{Institute of Theoretical Physics, Lanzhou University,
25: Lanzhou $730000$, China}
26:
27: \author{Yong Chen}
28: \altaffiliation{Author to whom correspondence should be addressed.
29: Email: ychen@lzu.edu.cn}
30: \affiliation{Institute of Theoretical Physics, Lanzhou University, Lanzhou $730000$, China}
31: \affiliation{Key Laboratory for Magnetism and Magnetic materials of the Ministry of Education, Lanzhou University, Lanzhou $730000$, China}
32:
33: \date{\today}
34:
35: \begin{abstract}
36: We performed Brownian dynamics simulations on the melting of
37: two-dimensional colloidal crystals in which particles interact via a
38: Yukawa potential. A stable hexatic phase was found in the Yukawa
39: systems, but we also found that the melting of Yukawa systems is a
40: two-stage melting, which is inconsistent with the
41: Kosterlitz-Thouless-Halperin-Nelson-Young (KTHNY) theory. A two-phase
42: coexistence region between the stable hexatic phase and the isotropic
43: liquid phase was found. The behavior of point defects in the
44: coexistence region is very complicated. The emergence of some
45: unstable free disclinations and grain boundaries was a characteristic
46: representative of the isotropic liquid phase, and a large number of
47: free dislocations indicated the existence of a hexatic phase. This
48: indicates the existence of a phase of hexatic-isotropic liquid phase
49: coexistence. The big picture in the melting of a two-dimensional
50: Yukawa system is that first the system undergoes a transition
51: induced by the formation of free dislocations, then it goes through a
52: phase coexistence, and finally, it comes into an isotropic fluid
53: phase. This melting process is consistent with experiments and
54: simulations.
55: \end{abstract}
56:
57: \pacs{64.70.D-, 82.70.Dd, 61.72.Lk}
58: % 64.70.D-, Solid¨Cliquid transitions
59: % 82.70.Dd, colloids
60: % 61.72.Lk, Linear defects: dislocations, disclinations
61:
62: \maketitle
63:
64: \section{\label{sec:level1}INTRODUCTION}
65:
66: In contrast to the case of melting in three-dimensional systems, it
67: has by now been well established that in two-dimensional (2D)
68: crystals, long-range positional order does not exist due to
69: long-wavelength fluctuations~\cite{mer}. Despite this, in a 2D
70: crystal, there exists a special kind of long-range bond orientational
71: order. A microscopic scenario of 2D melting has been posited in the
72: form of the Kosterlitz-Thouless-Halperin-Nelson-Young (KTHNY)
73: theory~\cite{two,kt,nh,Yp}. The KTHNY theory predicts a new phase,
74: the so-called hexatic phase, that exists between the solid and
75: liquid phases in 2D melting~\cite{Dn}.
76:
77: According to the KTHNY theory, the melting of a two-dimensional
78: system is a two-stage transition. In the first stage we start with
79: the two-dimensional system in a solid phase, which has both
80: quasi-long-range positional order and long-range bond orientation
81: order; the system then undergoes a continuous transition and becomes
82: to a hexatic phase with short-range positional order and
83: quasi-long-range orientational order. In the second stage, another
84: continuous transition drives the hexatic phase to an isotropic liquid
85: phase in which both positional and bond orientational order have
86: short ranges.
87:
88: The KTHNY theory predicts the unbinding of topological defects to
89: break the symmetry in the two-stage transitions. The physical driving
90: force behind the two-stage transitions is the dissociation of bound
91: defect pairs, specifically pairs of dislocation (solid$\to$hexatic)
92: and pairs of disclinations (hexatic$\to$liquid). Two-dimensional
93: systems are characterized by two different order parameters, namely,
94: the orientational and translational order, corresponding to the two
95: types of topological defects. Dissociation of the dislocation pairs
96: causes the translational symmetry to be broken, and dissociation of
97: free dislocations melts causing the orientational symmetry to be
98: broken.
99:
100: In recent years, a large number of experiments and computer
101: simulations have indicated that there is, indeed, a two-stage melting
102: scenario in two-dimensional systems as prescribed by the KTHNY
103: theory. The hexatic phase has been observed in colloidal
104: crystals~\cite{Hexatic1, Hexatic3}, magnetic bubble
105: arrays~\cite{Hexatic4}, and the freestanding liquid-crystal
106: films~\cite{Hexatic5}. The hexatic phase not only appears in 2D
107: systems, but also in three-dimensional systems, such as in layered
108: smectic liquid crystals~\cite{Hexatic6}, dense solutions of
109: DNA~\cite{Hexatic7,Hexatic8} and high temperature
110: superconductors~\cite{Hexatic9}.
111:
112: Although the KTHNY theory is currently preferred, a different
113: theoretical approach, evoking grain-boundary-induced melting, was a
114: first-order transition suggested by Chui~\cite{chui}. Using a
115: low-density approximation, Chui found that the grain boundaries might
116: be generated before the dislocations unbind when the core energy of
117: dislocations is sufficiently small ($E_{c}\leq 2.84\textit{k}_{B}T$);
118: Thus, he predicted a first-order transition. One may note that the
119: condensation of geometrical defects is also a first-order
120: transition~\cite{Gd1,Gd2}.
121:
122: Several computer simulations on 2D melting have favored a first-order
123: phase transition or, at most, a weak first-order transition. In these
124: simulations, the hexatic phase was not observed. It was argued that
125: this transition might depend on the specific properties of systems
126: being studied such as their inter-particle potential. The transition
127: seems to be first-order in hard-core systems~\cite{ja}, but to be
128: second-order behavior in dipole-dipole interactions~\cite{bo}. Some
129: simulations for Lennard-Jones systems have discovered that the
130: hexatic phase is metastable~\cite{Kn}. In the case of Yukawa systems,
131: Naidoo and Schnitker found that the defect topology was very
132: complicated, and the predictions of the KTHNY theory were
133: violated~\cite{Ns}.
134:
135: The hexatic phase has now been indeed observed in 2D colloidal
136: crystals~\cite{zm1,My,Ta}. Two-dimensional colloidal particles are a
137: good experimental system for studying the 2D melting. The advantage
138: of such systems is that the charged colloidal crystals in aqueous
139: suspensions are observable under the microscope where, due to the
140: particle size, the individual colloid motions can be directly
141: observed. At the same time, the particles are still small enough to
142: perform thermally driven motion and can be considered as a
143: statistical ensemble in thermal equilibrium. In 1987, Murray \textit{et al.} experimented on these colloidal suspensions of simple
144: polystyrene spheres, and the results indicated that the two-step
145: melting was the same as prescribed by the KTHNY theory. However, the
146: experimental evidence for two-dimensional colloidal suspensions
147: remains a hotly debated subject of controversy. Tang et al. also
148: observed the two-step melting process, but they indicated that the
149: melting was first-order, which is consistent with the picture
150: developed by Chui.
151:
152: One commonly encounters the following essential question: does the
153: melting of two-dimensional screened coulomb colloidal systems (or
154: Yukawa systems) follow the scenario of KTHNY theory? In
155: Ref.~\cite{Ns}, the answer is that they found an intermediate phase,
156: but not a true hexatic phase, in their simulations. In the other
157: hand, Murray et al. observerd the hexatic phase in the screened
158: Coulomb system. A similar result has been obtained by Tang et al.,
159: but it worth noticing that although the hexatic phase was observed,
160: the possibility of the coexistence of the hexatic phase with the
161: isotropic liquid phase could not be ruled out in their
162: experiments~\cite{soft2}.
163:
164: In this paper, we study the melting of two-dimensional charged
165: colloidal crystals, where we present a Brownian-dynamics simulation
166: on a two-dimensional Yukawa system. We focus on the existence of the
167: stable hexatic phase and the coexistance of an isotropic-hexatic
168: phase. Furthermore, we study the defect structure and compare it with
169: former results from experiments and simulations. In Murray and Van
170: Winkle's work, the two-step KTHNY melting was found by a correlation
171: length analysis, however, the paired dislocations in the solid phase
172: and the free dislocations in the hexatic phase were not found. Tang
173: \emph{et al.} have also obtained a similarly contradictory result. By
174: Voronoi constructions analysis, we find that the paired dislocations
175: exist in the solid phase and only a few unstable free dislocations
176: are found in the hexatic phase.
177:
178: The organization of this paper is as follows. In Sec. II, we describe
179: the Brownian-dynamics simulation methods. The results are presented
180: and discussed in section III. Here, we calculate the orientational
181: and pair correlation functions, and then examine the two-phase
182: coexistence regions and defect topology. In the last section, we
183: summarize our results.
184:
185: \section{\label{sec:level1}The Model}
186: The well-known Dejaguin-Landu-Verwey-Overbeek (DLVO) theory gives us
187: a good description for the effective pair interaction of the
188: one-component model in the colloidal systems. The DLVO potential
189: consists of an electrostatic repulsion and a van der Waals
190: attraction. Normally this is managed in a Yukawa or screened Coulomb
191: form, which only retains the electrostatic part~\cite{Yw0}. For a
192: dilute charged stabilized colloidal system in which many-body
193: interactions can be ignored~\cite{Yw1,Yw2,Yw3}, a pairwise Yukawa
194: potential is defined as
195: \begin{equation}
196: V(r) = U_{0} \frac{\sigma}{r} \exp \left( - \lambda
197: \frac{r-\sigma}{\sigma} \right ), \label{eq01}
198: \end{equation}
199: where $U_{0}$ is the energy and $\sigma$ is the scale length. The
200: screening parameter $\lambda$ describes the 'softening' of the
201: particles: when $\lambda$ increases from zero to infinity, the cores
202: of interacting particles change from very soft to extremely
203: hard~\cite{2d,BD1,BD2}. In our simulations, we assumed that the
204: Yukawa potential between the particles is very soft (the screening
205: parameter $\lambda = 8$).
206:
207: Now we briefly describe the standard Brownian-dynamics simulation,
208: which is based on a finite difference integration of the irreversible
209: Langevin equations. The equation of motion for an individual colloid
210: $i$ is
211: \begin{equation}
212: \xi \dot{\textbf{r}}_{i} (t) = \textbf{F}_{i} (t) + \textbf{R}(t),
213: \label{eq02}
214: \end{equation}
215: where $i=1,\ldots,N$ labels the $N$ particles, $\xi$ is the friction
216: coeffcient, and $\xi=1$ in simulation units. $\textbf{R}(t)$ is the
217: Langevin random force of the solvent, and $\textbf{F}_{i}(t)$ is the
218: total inter-particle force on particle $i$. Here the hydrodynamic
219: interaction is ignored. The finite difference integration is
220: \begin{equation}
221: \textbf{r}_{i}(t+\Delta t) = \textbf{r}_{i}(t) + \textbf{F}_{i}(t)
222: \Delta(t) + (\Delta\textbf{r})_{R} + \textit{O}(\Delta t)^2,
223: \label{eq03}
224: \end{equation}
225: where $(\Delta \textbf{r})_{R}$ is a random displacement sampled from
226: a Gaussian distribution of zero mean and variance
227: \begin{equation}
228: \overline{(\Delta\textbf{r})^2_{R}} = 4 D_{0} \Delta t. \label{eq04}
229: \end{equation}
230: Here $D_{0}=\textit{k}_{B}T/\eta$ is the short-time diffusion
231: coefficient, $\textit{k}_{B}$ is the Boltzmann Constant and $T$ is
232: the temperature. The coefficient in Eq.~(\ref{eq04}) is $2$ in
233: one-dimensional systems. We used reduced units such that $U_{0}=1$,
234: $\sigma=1$, and $\rho=N/V=1$. In all simulations, we tuned the
235: reduced temperature $T^{*}=k_BT/U_0$ and the other parameters
236: $\sigma$, $U_{0}$, $\rho$, and $\lambda$ were fixed. We used a
237: periodically repeated rectangular simulation box with $N=2500$
238: particles and started from a triple lattice. The cutoff $r_{c}$ was
239: set as $4.1$. We considered only a triangular lattice since it is the
240: most densely packed lattice in two-dimensions and is thus favored by
241: nature. The scale of the simulation box is in the ratio $2:\sqrt{3}$
242: with the length of the $x$-axis of our simulation box $55.836$ in
243: order to minimize the finite-size effects. One can find more details
244: about this simulation in Ref.~\cite{2d}.
245:
246: To characterize the translational order, we calculate the pair
247: correlation function. It is defined by
248: \begin{equation}
249: g(r) = \rho^{-2} \left \langle \sum_{i,j \neq i}
250: \delta(\textbf{r}_{i}) \delta(\textbf{r}_{j} - \textbf{r}) \right
251: \rangle, \label{eq05}
252: \end{equation}
253: where $\rho$ is the $2D$ particle density. The bond-orientational
254: function is
255: \begin{equation}
256: g_{6}(r) = \left\langle \psi_{6}^{*} (\textbf{r}') \psi_{6}
257: (\textbf{r}' - \textbf{r}) \right\rangle, \label{eq06}
258: \end{equation}
259: where $\psi_{6}(\textbf{r})$ is the local bond orientational order
260: parameter
261: \begin{equation}
262: \psi_{6}(\textbf{r}_{m}) = \frac{1}{N_{b}} \sum_{n=1}^{N_{b}} e^{6i
263: \theta_{mn}}. \label{eq07}
264: \end{equation}
265: Here $N_{b}$ denotes the number of the nearest neighbor of the $m$th
266: particle, and $\theta_{mn}$ is the angle between the particles $i$
267: and $j$ with an arbitrary, but fixed, reference axis. According to
268: the KTHNY theory, the bond-orientational function $g_{6}(r)$ will
269: have an algebraic decay in a hexatic phase, and an exponential decay
270: in the liquid phase. Before the dislocation unbinding transition
271: occurs, the hexatic phase is anisotropic and the bond orientational
272: correlation function is
273: \begin{equation}
274: g_{6}(r)\propto r^{-\eta_{6}(T)}, \label{eq08}
275: \end{equation}
276: where $ \eta_{6} = \frac{18\emph{k}_{B}T}{\pi K_{A}}$. $K_{A}$ is
277: called the Frank Constant which describes the coupling constant
278: related to distortions of the bond-angle field. The KTHNY theory
279: predicts that the disclination unbinding transition is also
280: continuous, and that it occurs when the value of the Frank Constant
281: falls below $72 k_{B} T/\pi$.
282:
283: \section{\label{sec:level1}Results and discussions}
284:
285: In order to study the phase behaviors and identify the existence of
286: the hexatic phase, we computed the pair distribution functions and
287: bond orientational correlation functions. As there are large
288: fluctuations near the critical point, it is difficult to obtain the
289: phase boundaries precisely, however, what we are interested in is the
290: process of melting transition and phase behavior in the middle of the
291: intermediate region. In our simulations, it takes a sufficient amount
292: of time to reach equilibrium when a two-stage continuous melting
293: transition occurs. Normally, the simulation reaches equilibrium after
294: bout $50 \tau_{B}$, and the simulation results are gathered within
295: the range of $10 \tau_{B}$ ($\tau_{B} = \sigma^{2} \xi / U_{0}$).
296:
297: \subsection{\label{sec:level2}The translational and orientational order}
298:
299: \begin{figure*}
300: \begin{center}
301: \includegraphics[width=1\textwidth]{fig01.eps}
302: \caption{The pair correlation function $g(r)$ and its FFT analysis in
303: Yukawa systems with $N=2500$ and $\rho=1.0$ for different
304: temperatures, (a) $T^{*}=0.200$, (b) $T^{*}=0.500$, (c)
305: $T^{*}=0.605$, (d) $T^{*}=0.630$.} \label{Fig:1}
306: \end{center}
307: \end{figure*}
308:
309: Fig.~\ref{Fig:1}(e-h) shows the pair correlation functions $g(r)$ for
310: different temperatures. In this setting (the 2D crystal), there is
311: only an orientational symmetry; here there is no true translational
312: order as it can never really be "long-range". The systems are in the
313: solid phase for $T^{*}=0.200$ and $0.500$ and the oscillations of the
314: pair distribution function persist over the entire range. It was
315: found that the translational correlation function decays
316: algebraically. The behavior was different however for $T^{*}=0.605$
317: and $0.630$, and the oscillations died out quickly, which indicating
318: that short-range translational order does exist in the system.
319:
320: The fast Fourier transforms (FFT) of the pair distribution functions
321: are also presented in Fig.~\ref{Fig:1}(a-d). At low temperature
322: (solid phase), the first peak that indicates the periodic structure
323: of the system is very sharp. The second and third peaks also are
324: obvious (see Fig.~\ref{Fig:1}(a, b)). As we increase the temperature,
325: the third peak becomes unnoticeable. For $T^{*}=0.605$, the second
326: peak is not obvious, but the first peak was lower than it was for
327: the systems in the solid phase. At the temperature $T^{*}=0.630$, the
328: system is in the liquid phase. Here, none of the peaks were obvious
329: and the first peak was much lower than it was for the system in the
330: solid phase (see Fig.~\ref{Fig:1}(c, d)). With the increasing of
331: temperature, it show that the first peak moves considerably in
332: Fig.~\ref{Fig:1}(a-d). Due to crystal lattice is periodic, the first peak is
333: very sharp in solid phase. When the system melting into liquid phase,
334: the crystal lattice was broken. So it indicates that the
335: quis-long-range position order in solid phase became into short range
336: in liquid phase.
337: \begin{figure}
338: \includegraphics[width=0.5\textwidth]{fig02.eps}
339: \caption{(Color online) Orientational correlation function $g_{6}(r)$
340: as a function of the reduced temperature $T^{*}$ in a log-log plot.
341: The curves decays algebraically, which implies the existence of the
342: hexatic phase. The straight line, with slope $1/4$, is a guide for
343: the eyes.} \label{Fig:2}
344: \end{figure}
345:
346: In Fig.~\ref{Fig:2}, we plotted the simulation results of the bond
347: orientational functions $g_{6}(r)$ for different reduced
348: temperatures. When $g_{6}(r)$ does not decay, it means that the
349: system is in the solid phase with a long-range bond-orientational
350: order. When the reduced temperature rises to near the disclination
351: unbinding regions, $g_{6}(r)$ decays algebraically with an exponent
352: near $1/4$, which implies the existence of the hexatic phase as
353: predicted by the KTHNY theory. As in our simulations, it is easy to
354: see the hexatic phase where the bond-orientational functions decay
355: algebraically with $\eta(T^{*})$ near to $1/4$ (see $T^* = 0.605$ in
356: Fig.~\ref{Fig:2}). With further increases in temperature, the system
357: becomes a disordered liquid and $g_{6}(r)$ decays exponentially. It
358: should be noted that $g_{6}(r)$ decays algebraically with an exponent
359: $0.54$ at $T=0.61$, which is faster than $1/4$ from the KTHNY theory.
360:
361: \begin{figure}
362: \includegraphics[width=0.5\textwidth]{fig03.eps}
363: \caption{The Lindemann parameter $\gamma_{m}$. The lines drawn are
364: only guides for the eyes.} \label{Fig:3}
365: \end{figure}
366:
367: To locate the solid-to-hexatic phase transition temperature, we use
368: the 2D Lindemann melting criterion introduced by Bedanov and
369: Gadiyak\cite{lindemann}. As the mean square displacement $\langle
370: u^{2}\rangle$ diverges in a two-dimensional crystal, they suggested
371: the Lindemann parameter
372: \begin{equation}
373: \gamma_{m}=\langle(u_{j}-u_{j+1})^{2}\rangle/a^{2}
374: \end{equation}
375: where the indices $j$ and $j+1$ refer to neighboring particles. At
376: the melting point, these authors found a critical value
377: $\gamma_{m}^{c}=0.033$. Here we measure the Lindemann parameter at
378: different temperatures. At the melting point $T_{m}^{*}$, the
379: Lindemann parameter grows sharply, indicating a vanishing of the
380: positional symmetry. In our simulations, a sharp growth of
381: $\gamma_{m}$ is observed at $T_{m}^{*}=0.520$ (see Fig.~\ref{Fig:3}).
382:
383: \subsection{\label{sec:level2}The coexistence of hexatic and isotropic phases}
384:
385: Indeed, there exists an algebraic decay of the bond-orientational
386: correlation function $g_{6}(r)$. However, we observed that $\eta_{6}$
387: is larger than $1/4$ (see Fig.~\ref{Fig:2}), which is not consistent
388: with the prediction of KTHNY theory. In light of this, it has been
389: conjectured that a coexistence of hexatic and isotropic phases may
390: appear; this was pointed out by H. H. von Gr\"unberg et
391: al~\cite{soft2}. The coexistence region is important evidence of a
392: first order transition~\cite{2d}, however, in order to identify this
393: coexistence we need to work carefully.
394:
395: \begin{figure}
396: \includegraphics[width=0.5\textwidth]{fig04.eps}
397: \caption{Particle trajectory at $T^{*}=0.609$. This shows an apparent
398: two-phase equilibrium in two-dimension Yukawa systems. This
399: trajectory is plotted during $6 \tau_{B}$.} \label{Figtr}
400: \end{figure}
401:
402: Fig.~\ref{Figtr} plots the trajectory of the particles at
403: $T^{*}=0.609$ for a section of the simulation box during $6\tau_{B}$.
404: The existence of solid-like and liquid-like patches in
405: Fig.~\ref{Figtr} has been interpreted as evidence for phase
406: coexistence, however, a similar path also appears in the purely
407: hexatic phase. Given this, we have to present a more reliable and
408: measurable method to identify whether the systems is the pure hexatic
409: phase or in a two-phase coexistence.
410:
411: In order to confirm the phase coexistence, we used the recipe
412: suggested by Strandburg \textit{et al}~\cite{ch6}. The angular
413: susceptibility $\chi_{6}$ for different length scales is defined by
414: \begin{equation}
415: \chi_6=\bigg<\bigg|\frac{1}{N}\sum_{l}\frac{1}{N_l}\sum_{n}e^{6i\theta_{mn}}\bigg|^2\bigg>
416: \end{equation}
417: where the sum on $l$ is over all particles, the sum on $n$ is over
418: the nearest neighbors, $n_l$ is the number of nearest neighbors of
419: particle $l$, and $N$ is the number of particles in the system.In the
420: solid phase, the angular susceptibility $\chi_{6}$ is large due to
421: the long-range order. On the other hand, $\chi_{6}$ is small in the
422: fluid phase. If the system exhibits a two-phase coexistence, one
423: might expect that the distributions of $\chi_{6}$ for sufficiently
424: small length scales could be modeled by a combination of solid and
425: fluid distributions. However, in the case of a homogeneous hexatic
426: phase, varying the size of the subsystems should not lead to any
427: qualitative changes in the distribution of $\chi_{6}$ (as shown in
428: Ref.~\cite{bo} for systems with dipole-dipole interactions).
429:
430: We calculate $\chi_{6}$ in our simulations for many different length
431: scales by dividing the system of $2500$ particles into subsystems
432: containing an average of $128$, $64$, $16$, and $4$ particles. These
433: subsystems are also periodically repeated as rectangular simulation
434: boxes. $\chi_{6}$ is calculated every $100$ passes. If the system is
435: homogeneous, varying the size of the subsystems should not lead to
436: any qualitative changes in the distribution of $\chi_{6}$.
437: Contrarily, for a two-phase region, the probability distribution of
438: $\chi_{6}$ at a sufficiently small length scale could be modeled by a
439: curve with two peaks, reflecting a combination of two-phase
440: distributions.
441:
442: \begin{figure*}
443: \begin{center}
444: \includegraphics[width=0.9\textwidth]{fig05.eps}
445: \caption{The probability distributions of $\chi_{6}$ for different
446: temperatures: (a) $T^{*}=0.500$, (b) $T^{*}=0.600$, (c)
447: $T^{*}=0.602$, (d) $T^{*}=0.605$, (e) $T^{*}=0.609$, and (f)
448: $T^{*}=0.630$. The symbols in the plots indicate the number of
449: particles in different subsystems.} \label{Figk}
450: \end{center}
451: \end{figure*}
452:
453: Fig.~\ref{Figk} shows the distributions of $\chi_{6}$ for subsystems
454: of $128$, $64$, $16$, and $4$ particles with different temperatures.
455: In the region of low temperature or crystal phase, there is no
456: qualitative change (see Fig.~{\ref{Figk}}(a)). The probability
457: distributions of $\chi_{6}$ always remain singly peaked as the size
458: of the subsystems is varied. This indicates that the system is in the
459: homogeneous solid phase for low temperatures. In Fig.~\ref{Figk}(b)
460: and (c), we presented the probability distribution of $\chi_{6}$ at
461: $T^{*}=0.600$ and $T^{*}=0.602$ where a power law decayed of
462: $g_{6}(r)$ was found. By varying the size of the subsystems, the
463: probability distribution of $\chi_{6}$ has a single peak, which means
464: that there is only a homogeneous phase and the two-phase coexistence
465: can be ruled out. One can conclude that there is a stable hexatic
466: phase at $T^{*}=0.600$ and $0.602$.
467:
468: In Fig.~\ref{Figk}(d) and (e), the probability distributions of
469: $\chi_{6}$ at $T^{*}=0.605$ and $0.609$ could be modeled by a curve
470: with two peaks for $n=64$ and $128$. This clearly demonstrates the
471: two-phase coexistence since the peaks reflect a combination of solid
472: and fluid distributions. It should be noted that the $g_{6}(r)$
473: decays algebraically at both temperatures.
474:
475: Fig.~\ref{Figk}(f) depicts the probability distribution of $\chi_{6}$
476: at $T^{*}=0.630$. In the fluid phase, the distributions are peaked
477: near zero. Varying the size of the subsystems, the distribution of
478: $\chi_{6}$ cannot lead to any qualitative changes so we can conclude
479: that the 2D Yukawa systems melt to a pure liquid phase.
480:
481: \subsection{\label{sec:level2}Topological defects}
482:
483: \begin{figure}
484: \includegraphics[width=0.4\textwidth]{fig6_01.eps}\\
485: \includegraphics[width=0.4\textwidth]{fig6_02.eps}\\
486: \includegraphics[width=0.4\textwidth]{fig6_03.eps}\\
487: \caption{Distribution of topological defects at $T^{*}=0.50$ (top),
488: $0.600$ (middle), and $0.630$ (down). The $\vartriangle$ denotes a
489: disclination of unit positive strength (five nearest neighbors) and
490: the $\square$ denotes a disclination of unit negative strength (seven
491: nearest neighbors).} \label{Figtr2}
492: \end{figure}
493:
494: In the KTHNY theory, the two-dimensional melting is caused by the
495: unbinding of topological defect pairs~\cite{kt,nh,Yp}. First,
496: dislocation pairs unbind and then, the disclination pairs unbind.
497: Since dislocations are associated with an additional half row of
498: atoms, they can be quite effective at breaking up translational
499: order. However, dislocations are less disruptive of orientational
500: correlations. And the disclination breaks the orientational order.
501: Thus it is important to determine whether these defects are indeed
502: important in the melting process. Murray et al. observed that islands
503: of sixfold coordinated particle are surrounded by a network of grain
504: boundaries of fourfold and sevenfold coordinated particles in the
505: fluid phase~\cite{Hexatic9}. In the hexatic phase, they failed to find
506: free dislocations, but these grain boundaries did not completely
507: disappear. The neighboring grains began to orient with respect to
508: each other. In solid phase, they also failed to find paired
509: dislocations.
510:
511: Normally, the defect structure can be pictured by using Voronoi
512: polygons where we identify disclinations as particles with five or
513: seven nearest neighbors. A Voronoi polygon is defined as the boundary
514: of a region enclosing a particle, which is closer to every point of
515: the region than to any others~\cite{V1,V2}. A disclination is located
516: at a particle with five or seven vertices in its Voronoi polygon. In
517: a perfect triangular colloidal crystal, all particles are sixfold.
518: Disclinations are described as particles having five and seven
519: neighbors. A disclination with positive unit strength is located at a
520: particle with five near neighbors and that with negative unit
521: strength is located near seven neighbors. A dislocation may be viewed
522: as a tightly bound pair of disclinations. In the solid phase,
523: dislocations are bound very strongly to a potential that increases as
524: the square of the separation.
525:
526: We studied the topological configurations by using Voronoi cell
527: pictures to identify the position of the defect. We did this by
528: showing the positions of disclinations for a single configuration
529: after a long run. For a single configuration in a solid phase at
530: $T^{*}=0.500$ (see Fig.~\ref{Figtr2}(a)), nearly all of the
531: dislocations occur in pairs. Fig.~\ref{Figtr2}(b) shows the defect
532: structure in an intermediate region between the isotropic liquid and
533: the solid at $T^{*}=0.600$. It was found that there exist free
534: dislocations (see arrow $1$), which supports the KTHNY theory that
535: melting is the unbinding of dislocation pairs.
536:
537: In Fig.~\ref{Figtr2}(c), the system is clearly in an isotropic liquid
538: phase. There are a large number of defects and free disclinations,
539: and the defect structure is very complicated. Clustering of
540: dislocations is observed~\cite{Ns}: there exists a pair of fivefold
541: coordination (5-coordination) bindings with an 8-coordination (shown
542: in Fig.~\ref{Figtr2}(c) by arrow $1$) or a pair of 7-coordination
543: bindings with a 4-coordination (in Fig.~\ref{Figtr2}(c) shown by
544: arrow $2$). These topological defects are unstable, however, and
545: vanish very quickly.
546:
547: \begin{figure}
548: \includegraphics[width=0.4\textwidth]{fig07.eps}
549: \caption{Distribution of topological defects at $T^{*}=0.605$, where
550: the system is in a two-phase coexistence. A small amount of the free
551: dislocations appeared. $\vartriangle$ denotes a disclination of unit
552: positive strength (five nearest neighbors), and $\square$ denotes a
553: disclination of unit negative strength (seven nearest neighbors).}
554: \label{Figtr3}
555: \end{figure}
556:
557: The distributions of topological defects in the two-phase coexistence
558: region are shown in Fig.~\ref{Figtr3}. There is a large number of
559: free dislocations, and only a few free disclinations. Notice that the
560: translational order is broken by the emergence of free dislocations.
561: Since the number of disclinations is not enough to break up the bond
562: orientational order, the bond orientational function still decays
563: algebraically at the two-phase coexistence region. We observe that
564: defects are likely a characteristic of grain boundaries, that is, the
565: clusters consisting of dislocations and dislocations pair up, or
566: small dislocations form a loop, such defects were also shown in
567: Tang's work. The appearance of grain boundaries leads to the
568: first-order transition suggested by Chui~\cite{chui}.
569:
570: \begin{figure}[t]
571: \includegraphics[width=0.5\textwidth]{fig08.eps}
572: \caption{The fractions of $6$-coordinated, $5$-coordinated, and
573: $7$-coordinated particles. At low temperature, the number of defects
574: is very small. When the temperature above $0.500$, $N_{6}/N$
575: decreases rapidly.} \label{Figd2}
576: \end{figure}
577:
578: Fig.~\ref{Figd2} plots the fractions of $6$-coordinated,
579: $5$-coordinated, and $7$-coordinated particles. At low temperature,
580: all particles are nearly 6-coordinated, and the number of defects is
581: very small. When the temperature reaches $0.500$, $N_{6}/N$ shows
582: rapidly decreasing behavior. At $T^{*}=6.05$, almost $20\%$ of
583: particles are attached to the defects, which is consist with the
584: results in Ref.~\cite{Ta}. As the defect fraction rises above $30\%$,
585: the system melts into a liquid phase. At the liquid phase, the number
586: of $5$-coordinated particles is much more than the number of
587: $7$-coordinated particles due to the emergence of the $8$-coordinated
588: particles and the effect of boundaries.
589:
590: We have observed the mechanism of defects in the two-dimensional
591: Yukawa system~\cite{mv}. At low temperature, the paired dislocation
592: is formed or annihilated. The formation of a binding dislocation pair
593: can be viewed as the simultaneous formation of two sevenfold
594: coordinated particles and two fivefold coordinated particles from
595: four sixfold coordinated particles. When the temperature rises to the
596: hexatic phase, the dislocation pair dissociates. In the solid phase,
597: the unbinding of dislocations is unstable, and these unbound
598: dislocations will quickly bind. This is in contrast to the stable
599: free dislocation is found in the hexatic phase. Grain boundaries and
600: unstable disclinations appears throughout the region of two-phase
601: coexistence. Given this, one may conjecture that the isotropic liquid
602: phase can be characterized by the existence of stable defect
603: clustering.
604:
605: \section{\label{sec:level1}Conclusions}
606:
607: In this paper, we performed Brownian dynamics simulations to study
608: the melting of 2D colloidal crystals with Yukawa interactions, and
609: two-stage melting is found. The hexatic phase in melting of 2D
610: charged colloidal crystals was indeed observed~\cite{My}. Moreover,
611: the hexatic-liquid phase coexistence was observed as well. Such
612: coexistence was also confirmed by Tang \textit{et al.}~\cite{Ta}.
613:
614: We calculated the pair correlation function and bond-orientational
615: correlation functions. At low temperature, due to the
616: quasi-long-range positional order in 2D systems, the oscillations of
617: the pair correlation function persist over the entire range. On the
618: other hand, the oscillations of the pair correlation function died
619: quickly at higher temperature, and it was shown that the positional
620: order becomes shot-range. By using the 2D Lindemann melting
621: criterion, we found that the melting temperature is
622: $0.530$($\pm0.01$). An algebraic decay with $\eta$ near $1/4$ of the
623: bond orientational correlation function was observed at the
624: temperature $0.605$($\pm0.01$). By ruling out the coexistence, we
625: verified that this is a pure phase at the temperatures between
626: $0.530$ and $0.605$. As the bond orientational correlation function
627: decays algebraically, we concluded that the pure phase is a stable
628: hexatic phase with the quasi-long-range bond-orientational order.
629:
630: We found that the quasi-long-range bond-orientational order still
631: exists in the coexistence region in finding the algebraic decays of
632: the bond orientational functions. The emergence of unstable free
633: disclinations and grain boundaries is a characteristic representative
634: of an isotropic liquid phase, and a large number of free dislocations
635: is a characteristic representative of a hexatic phase. This indicates that
636: there was indeed a coexistence of hexatic-isotropic liquid phases. In a word,
637: the melting of two-dimensional Yukawa systems is a two-stage melting.
638: Firstly, the system first undergoes a transition induced by the
639: formation of free dislocations, the system then goes through a phase
640: coexistence, and finally moves into an isotropic fluid phase.
641:
642: \begin{acknowledgments}
643: X.Y.Z. acknowledges financial support of the National Talent Training
644: Fund in Basic Research. Y.C. was supported by the SRF for ROCS, SEM,
645: and by the Interdisciplinary Innovation Research Fund for Young
646: Scholars, Lanzhou University.
647: \end{acknowledgments}
648:
649: \begin{thebibliography}{99}
650:
651: \bibitem{mer}N. D. Mermin, Phy. Rev. \textbf{158}, 383 (1967).
652: %Absence of ferromagnetism or antiferromagnetism if one- or two- dimensional isotrpoic Heisenberg models.
653:
654: \bibitem{two}K. J. Strandburg, Rev. Mod. Phys. \textbf{60}, 161 (1988).
655: %Two-dimensional melting.
656:
657: \bibitem{kt}J. M. Kosterlitz and D. J. Thouless, J. Phys. C. \textbf{6}, 1181 (1973).
658: %Ordering, metastability and phase transitions in two-dimensional systems.
659:
660: \bibitem{nh}D. R. Nelson and B. I. Halperin, Phys. Rev. B. \textbf{19}, 2457 (1979).
661: %Dislocation-mediated melting in two dimensions.
662:
663: \bibitem{Yp}A. P. Young, Phys. Rev. B. \textbf{19}, 1855 (1979).
664: %Melting and the vector Coulomb gas in two dimensions.
665:
666: \bibitem{Dn}D. R. Nelson, \textit{Defects and Geometry in Condensed Matter Physics}(Cambridge University Press, Cambridge, 2002).
667:
668: \bibitem{Hexatic1}C. M. Murray, \textit{Bond-Orientational Order in Condensed Matter Systems}, edited by K. Strandurg (Springer, Berlin, 1992).
669:
670: \bibitem{Hexatic3}K. Zahn and G. Maret, Phys. Rev. Lett. \textbf{85}, 3656 (2000).
671: %Dynamic criteria for melting in two dimensions.
672:
673: \bibitem{Hexatic4}R. Seshadri and R. M. Westervelt, Phys. Rev. Lett. \textbf{66}, 2774 (1991).
674: %Hexatic-to-Liqiud transition in two-dimensional magnetic-bubble lattices.
675:
676: \bibitem{Hexatic5}C. F. Chou, A. J. Jin, S. W. Hui, C. C. Huang, and J. T. Ho, Science \textbf{280}, 1424 (1998).
677: %Multiple-Step Melting in Two-Dimensional Hexatic Liquid-Crystal Films.
678:
679: \bibitem{Hexatic6}C. C. Huang, Adv. Phys. \textbf{42}, 343 (1993).
680: %Thermal-properties of stacked hexatic phase in liquid-crystals.
681:
682: \bibitem{Hexatic7}D. H. Van Winkle, A. Chatterjee, R. Link, and R. L. Rill, Phys. Rev. E. \textbf{55}, 4354 (1997).
683: %Magnetic-field alignment of cholesteric liquid-crystalline DNA
684:
685: \bibitem{Hexatic8}R. L. Rill, T. E. Strzelecka, M. W. Davidson, and D. H. Van Winkle, Physica A \textbf{176}, 87 (1991).
686: %Ordered phases in concentrated DNA solutions
687:
688: \bibitem{Hexatic9}C. A. Murray, P. L. Gammel, D. J. Bishop, D. B. Mitzi, and A. Kapitulnik, Phys. Rev. lett. \textbf{64}, 2312 (1990).
689: %Observation of a hexatic vortex glass in flux lattices of the high-Tc superconductor Bi_{2.1}Sr_{1.9}Ca_{0.9}Cu_{2}O_{8+¦Ä}
690:
691: \bibitem{di1}C. Eisenmann, U. Gasser, P. Keim, G. Maret, and H. H. von G\"unberg, Phys. Rev. Lett. \textbf{95}, 185502 (2005).
692: %Pair Interaction of Dislocations in Two-Dimensional Crystals.
693:
694: \bibitem{gk}H. H. von G\"unberg, P. Keim, K. Zahn, and G. Maret, Phys. Rev. Lett. \textbf{93}, 255703 (2004).
695: %Elastic Behavior of a Two-Dimensional Crystal Near Melting.
696:
697: \bibitem{chui}S. T. Chui, Phys. Rev. B. \textbf{28}, 178 (1983).
698: %Grain-boundary theory of melting in two dimensions.
699:
700: \bibitem{Gd1}M. A. Glaser and N. A. Clark, Adv. Chem. Phys. \textbf{83}, 543 (1993).
701: %Melting and liquid structure in two dimensions.
702:
703: \bibitem{Gd2}Y. Lansac, M. A. Glaser, and N. A. Clark, Phys. Rev. E \textbf{73}, 041501 (2006).
704: %Discrete elastic model for two-dimensional melting.
705:
706: \bibitem{ja}A. Jaster, Phys. Rev. E \textbf{59}, 2594 (1999).
707: %Computer simulations of the two-dimensional melting transition using hard disks.
708:
709: \bibitem{bo}S. Z. Lin, B. Zheng, and S. Trimper, Phys. Rev. E \textbf{73}, 066106 (2006).
710: %Computer simulations of two-dimensional melting with dipole-dipole interactions
711:
712: \bibitem{Kn}K. Chen, T. Kaplan, and N. A. Clark, Phys. Rev. Lett. \textbf{74}, 4019 (1995).
713: %Melting in Two-Dimensional Lennard-Jones Systems: Observation of a Metastable Hexatic Phase.
714:
715: \bibitem{Ns}K. J. Naidoo and J. Schnitker, J. Chem. Phys. \textbf{100}, 3114 (1994).
716: %Melting of two-dimensional colloidal crystals: A simulation study of the Yukawa system.
717:
718: \bibitem{zm1}K. Zahn, R. Lenke, and G. Maret, Phys. Rev. Lett. \textbf{82}, 2721 (1999).
719: %Two-Stage Melting of Paramagnetic Colloidal Crystals in Two Dimensions.
720:
721: \bibitem{My}C. A. Murry and D. H Van Winkle, Phys. Rev. Lett. \textbf{58}, 1200 (1987).
722: %Experimental observation of two-stage melting in a classical two-dimensional screened Coulomb system.
723:
724: \bibitem{Ta}Y. Tang, A. J. Armstrong, R. C. Mockler, and W. J. O'Sullivan, Phys. Rev. Lett. \textbf{62}, 2401 (1989).
725: %Free-expansion melting of a colloidal monolayer.
726:
727: \bibitem{soft2}H. H. von Gr\"unberg, P. Keim, and G. Maret, Soft Matter (Vol.3): Colloidal Order from Entropic and Surface Forces, Edited by G. Gompper and M. Schick, Wiley-VCH (2007).
728:
729: \bibitem{Yw0}E. J. W. Verwey and J. T. G. Overbeek, \textit{Theory of the stability of Lyophobic Collids} (Amsterdam: Elsevier, 1948).
730: %
731: \bibitem{Yw1}H. L\"owen, P. A. Madden, and J. P. Hansen, Phys. Rev. Lett. \textbf{68}, 1081 (1992).
732: %Ab initio description of counterion screening in colloidal suspensions.
733:
734: \bibitem{Yw2}Y. Chen, Chinese Phys. Lett. \textbf{20}, 1626 (2003).
735: %Effect of Size Polydispersity on Melting of Charged Colloidal Systems.
736:
737: \bibitem{Yw3}J. Dobnikar, Y. Chen, R. Rzehak, and H. H. von Gr\"unberg, J. Chem. Phys. \textbf{119}, 04971 (2003); J. Dobnikar, Y. Chen, R. Rzehak, and H. H. von Gr\"unberg, J. Phys.: Condens. Matter \textbf{15}, S263 (2003).
738:
739: %Many-body interactions and the melting of colloidal crystals.
740: %Many-body interactions in colloidal suspensions
741:
742: \bibitem{2d}X. Qi, Y. Chen, Y. Jin, and Y. H. Yang, J. Kor. Phys. Soc. \textbf{49}, 1682 (2006).
743: %Structure and Brownian dynamics of the two-dimensional Yukawa fluid.
744:
745: \bibitem{BD1}H. L\"owen, J. Phy.: Condens. Matter \textbf{4}, 10105 (1993).
746: %Bond-Orientational Order in Melting of Colloidal Crystals.
747:
748: \bibitem{BD2}H. L\"owen and G. Szamel, J. Phys.: Condens. Matter \textbf{5}, 2295 (1993).
749: %Long-time self-diffusion coefficient in colloidal suspensions: theory versus simulation.
750:
751: \bibitem{lindemann}V. M. Bedanov and G. V. Gadiyak, Phys. Lett. A \textbf{109}, 289 (1985).
752:
753: \bibitem{ch6}K. J. Strandburg, J. A. Zollweg, and G. V. Chester, Phys. Rev. B \textbf{30}, 2755 (1984).
754: %Bond-angular order in two-dimensional Lennard-Jones and hard-disk systems.
755:
756: \bibitem{V1}G. Voronoi, J. Reine Ang. Math. \textbf{134}, 198 (1908).
757:
758: \bibitem{V2}B. Boots, A. Okabe, and K. Sugihara, \textit{Spatial Tessellations: Concepts and Applications of Voronoi Diagrams} (Wiley, New York, 1992).
759:
760: \bibitem{mv}A movie is archived at the website: $http://player.youku.com/player.php/sid/XMjE5OTk0MDg=/v.swf$
761:
762: \end{thebibliography}
763:
764: \end{document}
765: