1: %%%%%%%%%%%%%% aggregation in self-assembled colloidal alloys %%%%%%%%%%%%%
2: % =========================================================================
3: % Dr. K. W. Yu
4: % THE CHINESE UNIVERSITY OF HONG KONG
5: % Department of Physics, Shatin, New Territories, Hong Kong
6: % email: kwyu@phy.cuhk.edu.hk phone: (852) 2609 6100 fax: (852) 2603 5204
7: % =========================================================================
8: %%%%%%%%%%%%%% aggregation in self-assembled colloidal alloys %%%%%%%%%%%%%
9: % md.tex (last updated March 16, 2002)
10: %
11: \documentstyle[preprint,aps,epsfig]{revtex}
12:
13: \def\be{\begin{equation}}
14: \def\ee{\end{equation}}
15:
16: \begin{document}
17: \draft
18: \title{Heterogeneous aggregation in binary colloidal alloys}
19: \author{Andrew C. T. Wong and K. W. Yu}
20: \address{Department of Physics, The Chinese University of Hong Kong, \\
21: Shatin, New Territories, Hong Kong, China }
22:
23: \maketitle
24:
25: \begin{abstract}
26: Molecular dynamics (MD) simulation has been employed to study the
27: nonequilibrium structure formation of two types of particles
28: in a colloidal suspension, driven by type-dependent forces.
29: We examined the time evolution of structure formation as well as on
30: the structural properties of the
31: resulting aggregation by studying the radial distribution function (RDF).
32: The resulting aggregation is well described by a binary colloidal gelation.
33: We compared the structural properties to those for one type of particles.
34: From the MD results, it is evident that there are significant differences
35: between the RDF's of the two cases.
36: Moreover, we found that the average coordination number is generally larger
37: in the monodisperse case for all area fractions considered. Thus, by means
38: of heterogeneous aggregation, it is possible to obtain a wide variety of
39: structures while more close-packed structures are formed for monodisperse
40: colloidal aggregation.
41: \end{abstract}
42: \vskip 5mm
43: \pacs{PACS Numbers: 83.80.Gv}
44:
45: \section{Introduction}
46:
47: Soft condensed matter are materials which can easily be deformed by
48: external stresses, electric and magnetic fields,
49: or even by thermal fluctuations \cite{Chaikin}.
50: These materials typically possess structures which are much larger than
51: atomic or molecular scales; the structure and dynamics at the mesoscopic
52: scales determine the macroscopic physical properties.
53: The goal of our research is to model and understand this relationship.
54: These materials can be synthesized by means of colloidal self-assembly.
55: In what follows, we will describe a model colloidal system in which the
56: particles interact through type-dependent forces, so as to study
57: the relationship between the interaction at the mesoscopic scales and
58: the macroscopic properties.
59:
60: There is a long-standing yet fundamental question in physics:
61: Given all the interactions between the particles, what will be the
62: resulting structure that the particles will form?
63: For monodisperse case, usually close-packed structures will be formed,
64: e.g., hexagonal structure in two dimensions and face-centered cubic or
65: hexagonal close-packed structures in three dimensions.
66: By allowing two different types of particles, driven by two types of forces,
67: we are prepared to show that a rich variety of possible structures will
68: be formed, as we will study in more detail below.
69:
70: The plan of the paper is as follows.
71: In the next section, we will discuss the molecular dynamics simulation
72: method. In section III, we will present the results.
73: Discussion and conclusion on our results will be given.
74:
75: \section{Molecular Dynamics Simulation}
76:
77: The colloidal system studied in this work is a two-dimensional one
78: consisting of circular particles of the same diameter $d$,
79: suspended in a viscous fluid. There is a pairwise interparticle force
80: ${\bf F}_{ij}=-\nabla U({\bf r}_{ij})$ between particles $i$ and $j$,
81: where $U$ is a potential and ${\bf r}_{ij}={\bf r}_i - {\bf r}_j$.
82: The equation of motion of a particle is given by
83: \be
84: m \frac{d^2{\bf r}_{i}}{dt^2}= \sum_{j\neq i}{\bf F}_{ij}
85: - w\frac{d {\bf r}_{i} }{dt} ,
86: \ee%1
87: where the first term on the right-hand side describes the total force
88: acting on the particle $i$,
89: while the second term denotes the viscous drag (with coefficient $w$),
90: exerted by the fluid.
91: We have neglected the Brownian motion, which is a valid assumption for
92: mesoscale objects.
93: The variables can be rescaled as
94: $t=t_0 t^*$, ${\bf r}=d{\bf r^*}$, and ${\bf F}=F_0{\bf F^*}$,
95: with $t_0=m/w$ and $F_0$ being the typical magnitude of the
96: interparticle force. The rescaled equation of motion can be written as
97: \be
98: {\bf \ddot{r}}_i^*=A \sum_{j\neq i}{\bf F}_{ij}^* - {\bf \dot{r}}^*_i ,
99: \label{eom}
100: \ee%2
101: where $A=F_0 t_0/w$.
102: In a highly viscous medium, $A$ is very small and the particles are
103: in overdamped motion, i.e., ${\bf \ddot{r}}_i^* \approx 0$.
104: In this case, Eq.~(\ref{eom}) reduces to
105: \be
106: {\bf \dot{r}}^*_i=A \sum_{j\neq i}{\bf F}_{ij}^* ,
107: \ee%3
108: We further define $\tilde{t}^*=A t^*$ and obtain the reduced equation
109: of motion for our simulation:
110: \be
111: \frac{d {\bf r}_i^* }{d \tilde{t}^*}=\sum_{j\neq i}{\bf F}_{ij}^* .
112: \ee%4
113: Thus we have a set of simultaneous first-order differential equations
114: to solve. In the overdamped situation, the final structure is independent
115: of $A$, but the time for the structure formation is inversely proportional
116: to $A$.
117: The initial configurations consist of $N$ particles randomly dispersed in
118: a square simulation cell, with periodic boundary conditions being imposed
119: in two directions.
120: Thus, we include the forces acting on a particle due to all other particles
121: in the simulation cell as well as their periodic images, calculated within
122: a square centered at that particle which is of the same size of the
123: simulation cell. The simulations are run for a sufficiently long time so
124: that the particles have no further movements as time goes on.
125:
126: The simulation cell is set to the scales $25d \times 25d$ and we ran
127: simulations of area fraction of particles
128: $\Phi=0.13$, $0.19$, $0.25$, $0.31$, $0.37$ and $0.44$
129: corresponding to the number of particles
130: $N=100$, 150, 200, 250, 300, 350 respectively.
131: The simple Euler algorithm has been used throughout the simulation with
132: a time step $\delta \tilde{t}^* = 5\times 10^{-4}$.
133: Larger time steps may result in unphysically large velocities because
134: of serious overlapping between the particles.
135:
136: We consider two cases. The first case consists of one type of particles,
137: interact through attractive force, while the second case consists of
138: equal number of different types of particles A and B.
139: The initial positions and velocities of the particles are imposed by a
140: random distribution inside the simulation cell without particle overlapping.
141: It is difficult to generate for $N$ larger than 350 because of serious
142: particle overlapping.
143:
144: \subsection{One type of particles with attractive force}
145:
146: In this case all particles attract each other at short distance and
147: repel each other at long distance. The potential between the partilces is
148: \be
149: U(r)= u_0 \exp(-(r-d)/\xi_0) - u_1 \exp(-(r-d)/\xi_1).
150: \ee%5
151: The first term denotes a repulsion, with a range $\xi_0$ and magnitude
152: $u_0$. The second term denotes an attraction, with a range $\xi_1$ and
153: magnitude $u_1$. We set $\xi_0>\xi_1$ and $u_1>u_0$ so that the particles
154: attract each other at short distance.
155: In order to avoid too much overlap when the particles approach and finally
156: touch, we add a steep repulsive potential of the form $b/r^{11}$
157: to the interparticle potential.
158: Thus, the force is calculated by the negative gradient of the potential
159: \be
160: F(r)=\frac{11b}{r^{12}}+\frac{u'}{\xi_0}\exp(-(r-d)/\xi_0)
161: -\frac{u_1}{\xi_1}\exp(-(r-d)/\xi_1)
162: \ee%6
163: where
164: $$
165: u'=u_0-{b\over d^{11}},\ {\rm and}\ \
166: b=\frac{u_0/\xi_0-u_1/\xi_1}{1/\xi_0 d^{11}-11/d^{12}}.
167: $$
168: In this way, the two particles attract each other when they approach,
169: while the force vanish when $r=d$, and repel each other when they
170: overlap. If we rescale the force by $b=F_0 d^{12} b^*$, $u'=F_0 d u'^*$,
171: $u_0=F_0 d u_0^*$ and $u_1=F_0 d u_1'^*$, we obtain the dimensionless force:
172: \be
173: F^*(r)=\frac{11b^*}{ {r^*}^{12} }
174: +\frac{u'^*}{\xi_0^*}\exp(-(r^*-1)/\xi_0^*)
175: -\frac{u_1^*}{\xi_1^*}\exp(-(r^*-1)/\xi_1^*) ,
176: \ee%7
177: where
178: $$
179: b^*=\frac{u_0^*/\xi_0^*-u_1^*/\xi_1^*}{1/\xi_0^* -11}\ {\rm and}\ \
180: u'^*=u_0^*-b^* .
181: $$
182: Here $\xi_0^*$, $\xi_1^*$, $u_0^*$ and $u_1^*$ are adjustable parameters.
183: In this simulation, the parameters used are $\xi_0^*=0.5$, $\xi_1^*=0.3$,
184: $u_0^*=5$, and $u_1^*=6$.
185:
186: \subsection{Two type of particles}
187:
188: The system consist equal number of A and B particles of total area
189: fraction $\Phi$. For unlike particles, there are short range attractive
190: and repulsive forces between the particles, same as for one type of
191: particles above. The potential between two particles is
192: \be
193: U_-(r)=u_0 \exp(-(r-d)/\xi_0) - u_1 \exp(-(r-d)/\xi_1),
194: \ee%8
195: while for the same type of particles, there is always a repulsive
196: potential between them
197: \be
198: U_+(r)=u_0 \exp(-(r-d)/\xi_0).
199: \ee%9
200: Again, we impose a steep repulsive potential of the form $b/r^{11}$
201: to the interparticle potential.
202: Likewise, the rescaled dimensionless force are:
203: \be
204: F_-^*(r)=\frac{11b^*}{ {r^*}^{12} }
205: +\frac{u'^*}{\xi_0^*}\exp(-(r^*-1)/\xi_0^*)
206: -\frac{u_1^*}{\xi_1^*}\exp(-(r^*-1)/\xi_1^*)
207: \ee%10
208: \be
209: F_+^*(r)=\frac{11b^*}{ {r^*}^{12} }
210: +\frac{u'^*}{\xi_0^*}\exp(-(r^*-1)/\xi_0^*)
211: \ee%11
212: respectively. The same parameters are used as in the monodisperse case.
213:
214: \section{Results}
215:
216: We calculate the radial distribution function to study the stucture
217: of the aggregations. The normalized two dimensional
218: radial distribution function is defined as \cite{Allen}
219: \be
220: g_0(r^*)=\frac{1}{8 \Phi N \Delta }\sum_i\sum_{j\neq i}
221: \frac{\delta(r^*-r_{ij}^*)}{r_{ij}^*}
222: \ee
223: where $\Delta$ is the width of histogram of $g_0$. $g_0$ is normalised
224: such that it tends to 1 at long distances.
225:
226: For the analysis of simulations of two types of particles, we introduce
227: partial radial distribution functions, namely,
228: $g_{AA}$ is the contribution of $g_0$ by radial distances between
229: particles of type A, $g_{BB}$ the
230: contributions of type B, and $g_{AB}$ the contributions of radial
231: distances between different types A and B.
232: It is clear that $g_0=g_{AA}+g_{BB}+g_{AB}$. The distances are counted
233: over the particles in the cells and
234: their periodic images, in the same way as calculating the forces.
235: The calculations are averaged over 20
236: ensembles for better statistics.
237:
238: We also calculated the mean coordination number, counted over the
239: particles and their periodic images and averaged over 20 ensembles.
240:
241: 1. Nonequilibrium structure formation for two types of particles, driven
242: by type-dependent forces: We will focus on the time evolution of
243: structure formation as well as on the structural properties of the
244: resulting aggregation by studying the radial distribution function (RDF).
245: We plot the time series of $g_0$ of aggregation in Fig.1 with four
246: panels (a)--(d), from initial to final time, both for one type and two
247: types of particles (shown on the same figure). The same figure will be
248: used in describing results in item 2 below.
249:
250: We plot the partial RDF's ($g_{AA}$, $g_{BB}$, $g_{AB}$) as well as the total
251: RDF ($g_0$) in Fig.2 with six panels (a)--(f), corresponding to the six area
252: fractions studied, for the binary colloidal case.
253: The resulting aggregation is well described by a binary alloy gelation.
254:
255: 2. We will compare the structural properties to those for one type of
256: particles (monodisperse case).
257: We plot the RDF ($g_0$) in Fig.3 with six panels (a)--(f),
258: corresponding to the six area fractions studied, for the monodisperse case.
259: From the results, it is evident that there are significant differences
260: between the RDF's of the two cases. In this regard, we will also study
261: the statistical geometry by computing the average coordination number of
262: the resulting aggregation.
263:
264: We plot the average coordination number $\langle Z\rangle$ versus area
265: fraction in Fig.4 both for one type and two types of particles.
266: We find that the average coordination number is generally larger in the
267: monodisperse case for all area fractions considered. It turns out that more
268: close-packed structures (in fact hexagonal) are formed for monodisperse
269: colloidal aggregation.
270:
271: 3. We will also study the partial RDF's in more detail. By symmetry,
272: the resulting binary colloidal alloy is statistically bipartite,
273: i.e., $g_{AA}=g_{BB}$. That means that the resulting lattice can be
274: decomposed into two statistically identical lattice.
275: According to our results in Fig.2, the resulting aggregation can be
276: decomposed into local square order (tetravalent sites) and local
277: hexagonal order (trivalent sites). In the limit of a large attractive
278: force between the different types of particles, however,
279: we may enhance tetravalent sites at the expense of trivalent sites.
280: In this regard, we will do additional simulations for the largest area
281: fraction (corresponding to $N=350$), with a larger $u_1=12$,
282: other parameters being fixed, so as to suppress the trivalent sites.
283: In Fig.5, we plot the partial RDF's for the $u_1=12$ case.
284: The $u_1=6$ case is plotted on the same figure for comparison.
285:
286: \section{Discussion and conclusion}
287:
288: Here a few comments on our results are in order.
289: The self-assembly of two types of particles can be realized by a recent
290: experimental demonstration of DNA-assisted self-assembly of nanoparticles
291: \cite{DNA}. While the present investigation has been on isotropic
292: interparticle forces, it is instructive to extend the present work to
293: polydisperse electrorheological (ER) fluids, in which the suspended
294: particles can have different dielectric permittivities \cite{Yu}.
295: In ER fluids, the polarized particles aggregate under the influence of
296: anisotropic dipolar forces. By tuning the strength of the applied field,
297: it is possible to realize a structure transformation from the body-centered
298: tetragonal to the face-centered cubic structures \cite{Lo}. Thus, by
299: considering two different types of particles driven both by isotropic and
300: anisotropic interparticle forces, we may obtain a diversity of structures
301: with potential applications in photonic band-gap materials.
302:
303: In conclusion, we have performed a detailed molecular dynamics simulation
304: to study the nonequilibrium structure formation for two types of particles
305: in a colloidal suspension, driven by type-dependent forces.
306: We examined the time evolution of structure formation as well as on
307: the structural properties of the resulting aggregation by studying the
308: radial distribution function (RDF).
309: The resulting aggregation is well described by a binary colloidal gelation.
310: We find that the average coordination number is generally larger in the
311: monodisperse case for all area fractions considered. Thus, by means
312: of heterogeneous aggregation, it is possible to obtain a wide variety of
313: structures while more close-packed structures are formed for monodisperse
314: colloidal aggregation.
315:
316: \section*{Acknowledgments}
317: This work was supported by the Research Grants Council of the Hong Kong
318: SAR Government under grant CUHK 4245/01P.
319: K.W.Y. acknowledges the hospitality received during his participation in
320: the Adriatico Research Conference on Interaction and Assembly of
321: Biomolecules, hosted by the International Center for Theoretical Physics
322: at Italy, where the present work was initiated.
323: We acknowledge useful discussion with Dr. Jones T. K. Wan.
324:
325: \begin{references}
326: \bibitem{Chaikin} P. M. Chaikin and T. C. Lubensky,
327: {\em Principles of Condensed Matter Physics},
328: (Cambridge University Press, UK, 2000).
329:
330: \bibitem{Allen} M. P. Allen, in {\em Methods in Molecular Simulation},
331: (UMIST, Manchester, UK, 2001); see also
332: M. P. Allen and D. J. Tildesley, {\em Computer Simulation
333: of Liquids}, (Clarendon Press, Oxford, 1989).
334:
335: \bibitem{DNA} C. A. Mirkin, R. L. Letsinger, R. C. Mucic and J. J. Storhoff,
336: Nature {\bf 382}, 607 (1996).
337:
338: \bibitem{Yu} K. W. Yu and Jones T. K. Wan, Comput. Phys. Commun.
339: {\bf 129}, 177 (2000).
340:
341: \bibitem{Lo} C. K. Lo and K. W. Yu, Phys. Rev. E {\bf 64}, 031501 (2001).
342:
343: \end{references}
344:
345: \begin{figure}[h]
346: \caption{Plot the time series of $g_0$ of aggregation with four panels (a)--(d),
347: from initial to final time, both for one type and two types of particles
348: (shown on the same figure).}
349: \end{figure}
350:
351: \begin{figure}[h]
352: \caption{Plot the partial RDF's ($g_{AA}$, $g_{BB}$, $g_{AB}$) as well as the total
353: RDF ($g_0$) with six panels (a)--(f), corresponding to the six area
354: fractions studied, for the binary colloidal case.}
355: \end{figure}
356:
357: \begin{figure}[h]
358: \caption{Plot the RDF ($g_0$)with six panels (a)--(f),
359: corresponding to the six area fractions studied,
360: for the monodisperse case.}
361: \end{figure}
362:
363: \begin{figure}[h]
364: \caption{Plot the average coordination number $\langle Z\rangle$ versus area
365: fraction both for one type and two types of particles.}
366: \end{figure}
367:
368: \begin{figure}[h]
369: \caption{Plot the partial RDF's for the $u_1=20$ case.
370: The $u_1=6$ case is plotted on the same figure for comparison.}
371: \end{figure}
372:
373: \newpage
374: \centerline{\epsfig{file=fig1.eps,width=\linewidth}}
375: \centerline{Fig.1/Wong and Yu}
376:
377: \newpage
378: \centerline{\epsfig{file=fig2.eps,width=\linewidth}}
379: \centerline{Fig.2/Wong and Yu}
380:
381: \newpage
382: \centerline{\epsfig{file=fig3.eps,width=\linewidth}}
383: \centerline{Fig.3/Wong and Yu}
384:
385: \newpage
386: \centerline{\epsfig{file=fig4.eps,width=\linewidth}}
387: \centerline{Fig.4/Wong and Yu}
388:
389: \newpage
390: \centerline{\epsfig{file=fig5.eps,width=\linewidth}}
391: \centerline{Fig.5/Wong and Yu}
392: \end{document}
393: