1: \documentstyle[preprint,aps,prl,epsf,epsfig]{revtex}
2: \topmargin -0.5cm
3: \begin{document}
4:
5: %\twocolumn[
6: %\hsize\textwidth\columnwidth\hsize\csname @twocolumnfalse\endcsname
7:
8: \title{
9: Weakly nonlinear theory of grain boundary motion in patterns with
10: crystalline symmetry
11: }
12: \author{Denis Boyer}
13: \address{Instituto de F\'\i sica, Universidad
14: Nacional Aut\'onoma de M\'exico, Apartado Postal 20-364,
15: 01000 M\'exico D.F., M\'exico
16: }
17: \author{Jorge Vi\~nals}
18: \address{Laboratory of Computational Genomics, Donald Danforth Plant Science
19: Center, 975 North Warson Rd, St. Louis, Missouri 63132.}
20:
21: \date{\today}
22:
23: \maketitle
24:
25: \begin{abstract}
26: We study the motion of a grain boundary separating two otherwise
27: stationary domains of hexagonal symmetry.
28: Starting from an order parameter equation appropriate for hexagonal
29: patterns, a multiple scale analysis leads to an analytical equation
30: of motion for the boundary that shares many properties with that of
31: a crystalline solid. We find that
32: defect motion is generically opposed by a pinning force that arises
33: from non-adiabatic corrections to the standard amplitude equation. The
34: magnitude of this force depends sharply
35: on the mis-orientation angle between adjacent domains so that
36: the most easily pinned grain boundaries are those with an angle
37: $4^{\circ}\le \theta\le 8^{\circ}$. Although
38: pinning effects may be small, they do not vanish asymptotically
39: near the onset of this subcritical bifurcation, and can be orders of
40: magnitude larger than those present in smectic phases that bifurcate
41: supercritically.
42: \end{abstract}
43:
44: \pacs{05.45.-a, 47.54.+r, 61.43.-j}
45:
46: %\narrowtext
47: %]
48:
49: The microstructure of a condensed phase and the distribution of topological
50: defects largely determine its mechanical and thermodynamical response, as well
51: as the temporal evolution of its non-equilibrium configurations.
52: Continuum or hydrodynamic approaches to phases with broken
53: symmetry are now well understood \cite{re:martin72,re:chaikin95},
54: including the long wavelength description of topological defects
55: \cite{re:kosevich79,re:nelson79}. Nevertheless, a quantitative theory
56: of defect motion (e.g., dislocation glide) remains a difficult
57: problem because it requires short scale phenomena
58: that lie beyond a hydrodynamic theory, and therefore beyond the degree
59: of universality that such a description entails.
60:
61: Bifurcations to states with broken symmetry
62: are also encountered in a variety of physical, chemical and biological systems
63: driven outside of thermodynamic equilibrium.
64: The characteristic length scales involved are much larger than atomic
65: dimensions, therefore making the observation and study of defects substantially
66: easier. Here, amplitude or phase equations that focus on slow modulations of
67: a base periodic pattern play the role of the hydrodynamic description
68: \cite{re:cross93}. It has been noted, for example, that at leading order
69: hexagonal patterns are the dissipative analogues of a two dimensional,
70: isotropic solid \cite{re:walgraef96}.
71: We focus in this paper on grain boundary motion in hexagonal patterns, and
72: find many qualitative and quantitative similarities with grain
73: boundary motion in crystalline solids. In contrast with the latter case,
74: our starting model allows a detailed analysis of the breakdown of the long
75: wavelength description of defect motion through an explicit multiple scale
76: analysis. We are then able to derive analytically several results
77: that are known only qualitatively or empirically in crystalline solids.
78:
79: Hexagonal patterns formed in a spatially extended system (like in Langmuir
80: monolayers \cite{re:sagui95} or block copolymer melts \cite{re:elder01})
81: are often fairly disordered. They consist of randomly
82: oriented grains separated by grain boundaries (arrays of
83: dislocations in the low angle case), and are equivalent by symmetry to a
84: polycrystalline solid.
85: This nonequilibrium microstructure usually
86: evolves on a
87: very slow, fluctuation-dependent time scale due to pinning forces
88: to defect motion. Our focus here is on a coarse grained model of a hexagonal
89: pattern, and, in particular, on the motion of a grain boundary separating
90: two domains with arbitrary mis-orientation. We extend to this case
91: a recent study of defect motion and pinning near a supercritical
92: bifurcation involving stripe patterns (or smectic phases) \cite{re:boyer02}.
93: We show that the coupling between the slow
94: variables of an envelope description of the defect and the
95: underlying (fast) periodicity of the base pattern leads to
96: an effective
97: periodic potential, the close analogue of the Peierls barrier
98: acting on defects in a crystalline solid \cite{re:peierls40}.
99: Furthermore, we find that the magnitude of the potential barrier does not
100: vanish as the subcritical bifurcation point is approached, contrary to
101: the case of smectic phases \cite{re:boyer02}.
102: Hence, not only crystalline phases emerging from a subcritical bifurcation
103: are harder than smectics in terms of the forces acting on topological defects,
104: but also an envelope description that ignores the internal degrees of
105: freedom of defects does not appear to be valid even near the bifurcation
106: point. The self-generated pinning effects discussed here are expected to
107: be a general feature of modulated phases, and can explain, for example,
108: many grain boundary conformations that have been observed experimentally and
109: numerically in block copolymer melts \cite{re:netz97}.
110:
111: We study here the Swift-Hohenberg model of Rayleigh-B\'enard convection
112: with an additional quadratic term to allow the formation of hexagonal
113: patterns \cite{re:cross93,re:walgraef96},
114: \begin{equation}
115: \label{sh}
116: \frac{\partial \psi}{\partial t}=\epsilon \psi-\frac{1}{k_0^4}
117: (k_0^2+\nabla^2)^2\psi+g_2\psi^2-\psi^3.
118: \end{equation}
119: The order parameter $\psi(\vec{x},t)$ is related to the vertical
120: velocity at the mid plane of the convective cell, $\epsilon$ is the
121: reduced Rayleigh number, and $g_{2}$ can be related to deviations
122: from Boussinesq behavior in the working fluid. The uniform solution
123: $\psi = 0$ becomes unstable for $\epsilon > 0$ to a periodic pattern of
124: wavelength $\lambda_{0} = 2\pi/k_{0}$. Hexagonal patterns
125: are stable for $-|\epsilon_m(g_2)|<\epsilon<\epsilon_M(g_2)$, and roll
126: patterns for $\epsilon>\epsilon_M$. In this study, we chose $\epsilon
127: \in [0,\epsilon_M]$ so that only hexagonal patterns are stable.
128: %\begin{figure}
129: %\epsfig{figure=gb1.ps,width=1.5in}
130: %\epsfig{figure=gb2.ps,width=1.5in}
131: %\epsfig{figure=gb3.ps,width=1.5in}
132: %\caption{Order parameter $\psi$ in gray scale. Only a portion of a square grid
133: %of $512^2$ points is shown with spacing $\Delta x=\sqrt{3}/2$
134: %and $\lambda_{0}=8\Delta x$. Three locations are shown in which a planar
135: %grain boundary is stationary.
136: %The two hexagonal domains are defined by the sets
137: %$\{\vec{k}_1,\vec{k}_2,\vec{k}_3\}$ and $\{\vec{k}_4,\vec{k}_5,\vec{k}_6\}$
138: %respectively, with a mis-orientation angle $\theta=\pi/9$.}
139: %\label{figgb}
140: %\end{figure}
141:
142: An approximate stationary solution for a configuration containing a
143: planar grain boundary between two uniform and symmetric domains of
144: hexagons that have a relative mis-orientation angle $\theta$
145: ($0 \le \theta\le\pi/3$, see Fig. \ref{figgb}) can be found by assuming
146: that $\psi_0=\sum_{n=1}^{6} A_{n} e^{i\vec{k}_n\cdot \vec{x}}+c.c.$,
147: where $A_{n}(x)$ are slowly varying envelopes, and $x$ is
148: the coordinate normal to the boundary.
149: As $x \rightarrow -\infty$,
150: $A_{1,2,3} \rightarrow A_0 = (g_2+\sqrt{g_2^2+15\epsilon})/15$
151: exponentially fast outside a boundary layer of thickness $\xi$
152: [$A_{n}(x)=f_{n}(x/\xi)$],
153: whereas $A_{4,5,6} \rightarrow 0$. As $x \rightarrow \infty$,
154: $A_{4,5,6} \rightarrow A_0$ and $A_{1,2,3} \rightarrow 0$. We chose
155: $|\vec{k}_n|=k_{0}$, $\vec{k}_{1,4}
156: =k_0(\cos(\phi)\hat{x}\mp\sin(\phi)\hat{y})$, where
157: $\phi=\pi/6-\theta/2$, and the other vectors $\vec{k}_n$ are obtained
158: from these by rotations of $\pm 2\pi/3$ as indicated in Figure \ref{figgb}.
159: In analogy with crystalline solids, a small angle grain
160: boundary is well described as an array of
161: penta-hepta defects (the dislocations cores of a hexagonal pattern),
162: separated on
163: average by a distance of the order $\lambda_0/\theta$. However, since the
164: projections of $\vec{k}_{1,2,3}$ on the $x$ axis are usually not commensurate,
165: the patterns observed along the boundary are not periodic.
166:
167: We next focus on grain boundary motion and extend our earlier results
168: for stripe patterns \cite{re:boyer02}. The amplitude equation
169: formalism has already been used to study defect dynamics in hexagonal
170: patterns \cite{re:tsimring96}.
171: Amplitude equations are obtained from a standard multiscale analysis of
172: Eq. (\ref{sh}). For example, the equation for $A_1$ follows
173: from the solvability condition
174: \begin{equation}
175: \label{schex}
176: \int_x^{x+l_x}\frac{dx'}{l_x}\lim_{l_y\rightarrow\infty}
177: \int_0^{l_y}\frac{dy'}{l_y}[L(\psi_{0})
178: +g_2\psi_{0}^2-\psi_{0}^3]e^{-i\vec{k}_1\cdot \vec{x}'}=0,
179: \end{equation}
180: where $L$ is a linear operator that follows from Eq. (\ref{sh})
181: \cite{re:manneville90} and $l_x$ a length
182: of $O(\lambda_0)$ to be specified later.
183: When both $\epsilon$ and $g_2 \rightarrow 0$, the length scale of variation
184: of the $A_n$'s ({\it i.e.} the grain boundary thickness $\xi$)
185: is much larger than $\lambda_0$, and only non oscillatory terms
186: contribute to the integral
187: (\ref{schex}). The solvability condition leads to the usual amplitude
188: equations that show that defects are either immobile or move with constant
189: velocity \cite{re:walgraef96}.
190:
191: If, on the other hand, $\epsilon$ and $g_2$ are small but finite,
192: then the amplitudes are not strictly constant within a period $l_x$.
193: Any oscillatory term in (\ref{schex}) of wavevector $\vec{K}$ parallel to
194: the normal $x$-axis does not integrate to zero, and the
195: equation for the slowly varying amplitudes cannot be decoupled from the phase
196: of the defect (yielding \lq\lq non-adiabatic corrections"
197: \cite{re:pomeau86,re:bensimon88b,re:malomed90,re:cross93}).
198: Although these contributions are small
199: (non-analytic in both $\epsilon$ and $g_2$), they may have dramatic effects.
200: In the case of Eq. (\ref{sh}), non-vanishing terms can arise from the cubic
201: nonlinearity, and are proportional to $A_1^2A_4$,
202: $\bar{A}_1\bar{A}_4^2$, $A_1A_3A_5$, $A_1\bar{A}_3\bar{A}_5$, $A_1A_2A_6$ and
203: $A_1\bar{A}_2\bar{A}_6$. An oscillatory term proportional to $\cos(Kx')$ in
204: Eq. (\ref{schex}) contributes to order $\exp(-|K\xi|)$ to the law of
205: grain boundary of motion (see Eqs. (\ref{phex}) and (\ref{pestim}) below).
206: This contribution is simplest
207: when both $\epsilon \ll 1$ and $g_2 \ll 1$ so that
208: $\xi \gg \lambda_0$ and a single mode with lowest $K$ dominates. In
209: this limit, the slowest non-adiabatic corrections are those proportional to
210: $A_1A_2A_6$ and $A_1\bar{A}_2\bar{A}_6$, which have
211: $K=|\vec{k}_2+\vec{k}_6|=2k_0\sin(\theta/2)$ (see Fig. \ref{figgb}).
212: The solvability condition (\ref{schex}) now reads
213: \begin{eqnarray}
214: \frac{\partial A_1}{\partial t} & = & -\frac{\partial F}{\partial\bar{A}_1}
215: \nonumber \\
216: & - &\int_{x_0}^{x_0+l_x}\frac{dx}{l_x}\
217: 6A_1\left(A_2A_6 e^{-2ik_0x\sin(\theta/2)}
218: + {\rm c.c.} \right),
219: \label{A1hex}
220: \end{eqnarray}
221: where $l_x=\lambda_0/[2\sin(\theta/2)]$.
222: The first two terms in Eq. (\ref{A1hex}) represent the standard
223: amplitude equation with the Lyapunov functional
224: \begin{eqnarray}
225: \label{F}
226: F & = & \int d\vec{r}\left[ -\epsilon\sum_{n=1}^6|A_n|^2
227: + \frac{4}{k_0^2} \sum_{n=1}^{6} |D_nA_n|^2 \right.
228: \nonumber \\
229: & - & 2g_{2} ( \bar{A}_1\bar{A}_2\bar{A}_3 +
230: \bar{A}_4\bar{A}_5\bar{A}_6
231: +A_1A_2A_3+A_4A_5A_6 )
232: \nonumber \\
233: & + & \left. \frac{3}{2} \sum_{n=1}^6|A_n|^4
234: + 3\sum_{n< m}|A_n|^2|A_m|^2\right],
235: \end{eqnarray}
236: with $D_n=\partial/\partial {x_n}$ the derivative along $\vec{k}_n$.
237: Equations similar to (\ref{A1hex}) can be derived for the remaining amplitudes
238: $A_{n}$ (not shown).
239: The last term in the r.h.s. of Eq. (\ref{A1hex}) is new and represents
240: the dominant non-adiabatic correction in the limits $\epsilon \ll 1$ and
241: $g_{2} \ll 1$.
242:
243: In order to derive an equation of motion for the grain boundary
244: from the equations for the $A_n$, we first
245: denote by $a_{n}(x)$ ($1\le n\le 6$) the leading order amplitudes of the
246: stationary grain boundary, solutions of the one-dimensional coupled
247: Ginzburg-Landau equations $\partial F/\partial \bar{a}_n=0$.
248: We then look for solutions of the form $A_n(x,t)=a_n(x-x_{gb}(t))$,
249: where $x_{gb}(t)$ is the now time-dependent position of the grain boundary
250: \cite{re:boyer02}. We find,
251: \begin{equation}
252: \label{eqmotion}
253: D\ \dot{x}_{gb}=
254: -p_{\rm hex}\ \sin[2k_0x_{gb}\sin(\theta/2)],
255: \end{equation}
256: where $D=\int_{-\infty}^{\infty} dx\sum_{n=1}^6 (\partial_x a_n)^2$
257: is a friction coefficient, and
258: \begin{eqnarray}
259: \label{phex}
260: p_{\rm hex} & = &{\rm Max}_{\beta} \int_{-\infty}^{\infty}
261: dx \cos \left[ 2k_0\sin(2\theta)x+\beta \right]
262: \left\{ 3 \left[ a_2\partial_x(a_2^2a_6) \right. \right.
263: \nonumber \\
264: & + & \left. a_6\partial_x(a_6^2a_2)] \right]
265: +12 \left[ a_1\partial_x(a_1a_2a_6)+a_3\partial_x(a_3a_2a_6) \right.
266: \nonumber \\
267: & + & \left. \left. a_4\partial_x(a_4a_2a_6)+a_5\partial_x(a_5a_2a_6)
268: \right] \right\}
269: \end{eqnarray}
270: is the (dimensionless) amplitude of a pinning force of wavelength
271: $\lambda_{0}/[2\sin(\theta/2)]$. This periodic force explicitly arises
272: from the coupling between the slowly varying wave amplitudes and the
273: periodicity of the base state in the integral term of Eq. (\ref{A1hex}).
274: Equations (\ref{eqmotion}) and (\ref{phex}) show that
275: a planar grain boundary initially located at an arbitrary position
276: relaxes toward the nearest minimum of the periodic
277: pinning potential.
278: The stationary and stable positions of the boundary are thus discrete,
279: separated from each other by a distance
280: $\Delta x_{gb}=\lambda_0/[2\sin (\theta/2)]>\lambda_0$.
281: The two wavevectors with the smallest projection on the grain
282: boundary normal ($\vec{k}_2$ and $\vec{k}_6$, see Figure \ref{figgb})
283: set the wavevector ($\vec{k}_2+\vec{k}_6$) of the periodic pinning
284: potential.
285: The usual amplitude equation formalism would instead predict that
286: $p_{\rm hex}=0$, and $x_{gb}$ is arbitrary and decoupled from the phase
287: of the pattern. Figure \ref{figgb} shows three successive stable
288: locations of the the grain boundary obtained by numerically solving Eq.
289: (\ref{sh}). The values of $\Delta x_{gb}$ determined numerically
290: agree very well with the analytic result.
291:
292: We expect equations of the form of Eq. (\ref{eqmotion}) to be generic at
293: a bifurcation, and not limited to hexagon-hexagon grain boundaries.
294: In particular, we anticipate that non-adiabatic effects are important in
295: block copolymer melts, and can explain the conformations of
296: planar interfaces observed in these systems \cite{re:netz97}.
297:
298: Remarkably, the present result for a pattern of hexagonal symmetry
299: is analogous to the Peierls force acting
300: on dislocations in crystalline solids \cite{re:peierls40}. Peierls
301: calculated the energy of a single dislocation by summing over the
302: interactions between atoms within two neighboring layers, their displacements
303: given by continuous elasticity as a first approximation.
304: The energy of the dislocation oscillates as a function of its position
305: so that it can glide only if a force of finite amplitude acts on it
306: (the Peierls' force).
307: Here, we find that a similar force acts on assemblies of dislocations
308: organized in arrays. Like in the simpler geometry studied by Peierls,
309: the amplitude of the pinning force decays exponentially with the spatial
310: thickness of the defect $\xi$. From Eq. (\ref{phex}), we find
311: \begin{equation}
312: \label{pestim}
313: p_{\rm hex}\sim c^* A_0^4\ e^{-2k_0\sin(\theta/2)\xi a^*},
314: \end{equation}
315: with $c^*$ and $a^*$ dimensionless
316: constants of order unity that can {\it a priori} depend on the
317: mis-orientation $\theta$. At this point, we note an important distinction
318: between defect motion in hexagonal patterns that emerge at a subcritical
319: bifurcation, and in stripe patterns that bifurcate supercritically. The
320: supercritical case was discussed in ref. \cite{re:boyer02} for
321: a $90^{\circ}$ boundary separating two domains of stripes given
322: by Eq. (\ref{sh}) with $g_2=0$.
323: The corresponding pinning force acting on the boundary, $p_{\rm stripe}$,
324: satisfies a relation similar to Eq. (\ref{pestim}), with
325: $\xi \sim 1/\sqrt{\epsilon}$.
326: Hence $p_{\rm stripe} \sim \exp(-1/\sqrt{\epsilon})\rightarrow 0$,
327: as the control parameter $\epsilon\rightarrow 0$.
328: In the hexagonal phase, however, $\xi(\epsilon=0)
329: \simeq 15\lambda_0/(8\sqrt{6}\pi g_2)$ is finite.
330: Therefore $p_{\rm hex}$, although small, does not vanishes when
331: $\epsilon \rightarrow 0$
332: (nor when $\epsilon\rightarrow-|\epsilon_m(g_2)|$).
333:
334: %\begin{figure}
335: %\vspace{0.7cm}
336: %\epsfig{figure=fig_epsilon.eps,width=5.0in}
337: %\vspace{1.0cm}
338: %\caption{Values of the pinning force as given by Eq. (\ref{phex})
339: %as a function of $\epsilon$
340: %for various values of $g_2$ ($\theta=\pi/9$ in each case). The result
341: %corresponding to a stripe pattern was given in ref.
342: %\protect\cite{re:boyer02}, and is shown as a comparison.}
343: %\label{figphex}
344: %\end{figure}
345:
346: Figure \ref{figphex} shows typical
347: variations of the pinning force as a function of $\epsilon$ for different
348: values of $g_2$ ($\theta=\pi/9$). The curves are given by
349: Eq. (\ref{phex}), where the amplitudes $a_n$ have been obtained
350: by numerically integrating the system of
351: equations $\partial F/\partial\bar{a}_n=0$.
352: For a value of $g_2$ as small as $0.3$, $p_{\rm hex}$
353: is many orders of magnitude larger than $p_{\rm stripe}$.
354: Therefore, non-adiabatic effects in hexagonal, \lq\lq crystalline-like",
355: patterns
356: are difficult to avoid. Defects need to overcome
357: much higher activation barriers, and will be much less easily un-pinned
358: either upon the application of external
359: stresses or random noise (that would be represented by additional terms
360: in the r.h.s. of Eq. (\ref{eqmotion})).
361:
362: Finally we study the dependence of the pinning force with
363: grain boundary mis-orientation $\theta$.
364: Figure \ref{fi:theta} displays $p_{\rm hex}(\theta)$ for different
365: values of the parameters $g_2$ and $\epsilon$.
366: %\begin{figure}
367: %\vspace{0.7cm}
368: %\hspace{-0.3cm}\epsfig{figure=fig_theta.eps,width=5.0in}
369: %\vspace{1.0cm}
370: %\caption{Angular dependence of the pinning force for various
371: %values of $g_2$ and $\epsilon$ obtained analytically with Eq. (\ref{phex}),
372: %and numerically for one of the parameter sets.}
373: %\label{fi:theta}
374: %\end{figure}
375: We find that the least mobile grain boundaries, {\it i.e.} those for which
376: the amplitude
377: of the pinning force is maximal, have a low angle $\theta_M$,
378: typically such that $4^{\circ} \le \theta_M\le 8^{\circ}$. Both $\theta_M$
379: and the overall shape of $p_{\rm hex}(\theta)$
380: depend weakly on $g_2$ and $\epsilon$.
381: Figure \ref{fi:theta} also shows the pinning force vs. $\theta$ obtained
382: from a direct numerical solution of Eq. (\ref{sh}) with $g_2=0.3$ and
383: $\epsilon=0.05$. Here $p_{\rm hex}$ is estimated by fitting grain boundary
384: relaxation trajectories to Eq. (\ref{eqmotion}), with the friction
385: coefficient $D$ assumed
386: to be given by the analytic result. The numerical results compare
387: reasonably well with the theory given the numerical difficulties in
388: tracking the relaxation of the grain boundary in a finite sized system.
389: Although the maximum value is lower and the curve flatter in the numerical
390: case, the right order of magnitude is obtained, as well as a similar
391: range for $\theta_M$. The discrepancy can also be attributed in
392: part to non-adiabatic corrections of higher order to Eq. (\ref{phex}), that
393: may not be negligible at $g_2\sim 0.3$.
394:
395: Our present analysis can qualitatively explain the properties of
396: polycrystalline, partially ordered configurations that are typically observed
397: at long times in many pattern forming systems with this symmetry
398: \cite{re:sagui95,re:elder01}.
399: Although the evolution defined by Eq. (\ref{sh}) is driven by the
400: minimization of a Lyapunov functional,
401: we expect that grain boundaries (and other topological defects) will become
402: pinned at long times as the driving force
403: for microstructure coarsening decreases. Therefore, the system will
404: eventually reach metastable,
405: glassy-like configurations that can only order by slow activated
406: processes (presumably logarithmic in time, as already observed in
407: \cite{re:elder01} with a random forcing
408: term added to Eq. (\ref{sh})). We additionally note that defects observed
409: in cold metals are essentially low angle grain boundaries in the
410: range $5^{\circ}$ and $10^{\circ}$ \cite{re:hull92}, a value that
411: compares very well with the values obtained by the present theory.
412:
413: In summary, we have analyzed the motion of grain boundaries in hexagonal
414: patterns from an order parameter equation and extended the standard
415: Ginzburg-Landau equation for the slowly varying amplitude to
416: incorporate non-analytic corrections.
417: Like in crystalline phases,
418: defect motion is opposed by short range forces with periodicity and
419: amplitude that strongly depend on the mis-orientation angle between domains.
420: Although small, these pinning forces
421: can not be neglected asymptotically at long times in a coarsening system,
422: even near onset, and are orders of magnitude
423: higher than those produced in patterns of smectic symmetry.
424:
425: We are indebted to Fran\c{c}ois Drolet for useful discussions.
426: This research has been supported by the U.S. Department of Energy, contract
427: No. DE-FG05-95ER14566.
428:
429: \bibliographystyle{prsty}
430: %\bibliography{$HOME/mss/references}
431:
432: \begin{thebibliography}{10}
433:
434: \bibitem{re:martin72}
435: P. Martin, O. Parodi, and P. Pershan, Phys. Rev. A {\bf 6}, 2401 (1972).
436:
437: \bibitem{re:chaikin95}
438: P. Chaikin and T. Lubensky, {\em Principles of condensed matter physics}
439: (Cambridge University Press, New York, 1995).
440:
441: \bibitem{re:kosevich79}
442: A. Kosevich, in {\em Dislocations in Solids}, edited by F. Nabarro
443: (North-Holland, New York, 1979), Vol.~1, p.\ 33.
444:
445: \bibitem{re:nelson79}
446: D.R. Nelson and B.I. Halperin, Phys. Rev. B {\bf 19}, 2457 (1979);
447: A.P. Young, Phys. Rev. B {\bf 19}, 1855 (1979).
448:
449: \bibitem{re:cross93}
450: M.C. Cross and P.C. Hohenberg, Rev. Mod. Phys. {\bf 65}, 851 (1993).
451:
452: \bibitem{re:walgraef96}
453: D. Walgraef, {\em Spatio-Temporal Pattern Formation} (Springer Verlag, New
454: York, 1996).
455:
456: \bibitem{re:sagui95}
457: C. Sagui and R.C. Desai, Phys. Rev. E {\bf 49}, 2225 (1994).
458:
459: \bibitem{re:elder01}
460: K.R. Elder, M. Katakowski, M. Haataja, and M. Grant, cond-mat/0107381.
461:
462: \bibitem{re:boyer02}
463: D. Boyer and J. {Vi\~nals}, cond-mat/0110254.
464:
465: \bibitem{re:peierls40}
466: R. Peierls, Proc. Phys. Soc. London {\bf 52}, 34 (1940).
467:
468: \bibitem{re:netz97}
469: R.R. Netz, D. Andelman, and M. Schick, Phys. Rev. Lett. {\bf 79},
470: 1058 (1997), and references therein.
471:
472: \bibitem{re:tsimring96}
473: L. Tsimring, Physica D {\bf 89}, 368 (1996).
474:
475: \bibitem{re:manneville90}
476: P. Manneville, {\em Dissipative Structures and Weak Turbulence} (Academic, New
477: York, 1990).
478:
479: \bibitem{re:pomeau86}
480: Y. Pomeau, Physica D {\bf 23}, 3 (1986).
481:
482: \bibitem{re:bensimon88b}
483: D. Bensimon, B.I. Shraiman, and V. Croquette, Phys. Rev. A {\bf 38}, 5461
484: (1988).
485:
486: \bibitem{re:malomed90}
487: B.A. Malomed, A.A. Nepomnyashchy, and M.I. Tribelsky, Phys. Rev. A {\bf 42}, 7244
488: (1990).
489:
490: \bibitem{re:hull92}
491: D. Hull and D.J. Bacon, {\em Introduction to Dislocations} (Pergamon Press,
492: New York, 1992).
493:
494: \end{thebibliography}
495:
496: \newpage
497: \begin{figure}
498: \epsfig{figure=gb1.ps,width=2.0in}
499: \epsfig{figure=gb2.ps,width=2.0in}
500: \epsfig{figure=gb3.ps,width=2.0in}
501: \vspace{1.0cm}
502: \caption{Order parameter $\psi$ in gray scale. Only a portion of a square grid
503: of $512^2$ points is shown with spacing $\Delta x=\sqrt{3}/2$
504: and $\lambda_{0}=8\Delta x$. Three locations are shown in which a planar
505: grain boundary is stationary.
506: The two hexagonal domains are defined by the sets
507: $\{\vec{k}_1,\vec{k}_2,\vec{k}_3\}$ and $\{\vec{k}_4,\vec{k}_5,\vec{k}_6\}$
508: respectively, with a mis-orientation angle $\theta=\pi/9$.}
509: \label{figgb}
510: \end{figure}
511:
512: \newpage
513: \begin{figure}
514: \vspace{0.7cm}
515: \epsfig{figure=fig_epsilon.eps,width=6.5in}
516: \vspace{1.0cm}
517: \caption{Values of the pinning force as given by Eq. (\ref{phex})
518: as a function of $\epsilon$
519: for various values of $g_2$ ($\theta=\pi/9$ in each case). The result
520: corresponding to a stripe pattern was given in ref.
521: \protect\cite{re:boyer02}, and is shown as a comparison.}
522: \label{figphex}
523: \end{figure}
524:
525: \newpage
526: \begin{figure}
527: \vspace{0.7cm}
528: \hspace{-0.3cm}\epsfig{figure=fig_theta.eps,width=6.5in}
529: \vspace{1.0cm}
530: \caption{Angular dependence of the pinning force for various
531: values of $g_2$ and $\epsilon$ obtained analytically with Eq. (\ref{phex}),
532: and numerically for one of the parameter sets.}
533: \label{fi:theta}
534: \end{figure}
535:
536: \end{document}
537: