1: \documentclass[aps,amsmath,preprint]{revtex4}
2: %\documentclass[aps,amsmath]{revtex4}
3: \usepackage{graphics,epsf,epsfig}
4: \usepackage{bm}
5:
6: \begin{document}
7:
8: \title{Shear induced grain boundary motion for lamellar phases
9: in the weakly nonlinear regime}
10: \author{Zhi-Feng Huang and Jorge Vi\~nals}
11: \affiliation{
12: School of Computational Science and Information Technology,
13: Florida State University, Tallahassee, Florida 32306-4120
14: }
15: \date{\today}
16:
17: \begin{abstract}
18:
19: We study the effect of an externally imposed oscillatory shear on
20: the motion of a grain boundary that separates differently oriented
21: domains of the lamellar phase of a diblock copolymer. A direct numerical
22: solution of the Swift-Hohenberg equation in shear flow is used for the
23: case of a transverse/parallel grain boundary in the limits of weak
24: nonlinearity and low shear frequency. We focus on the region of
25: parameters in which both transverse and parallel lamellae are linearly
26: stable. Shearing leads to excess free energy in the transverse region
27: relative to the parallel region, which is in turn dissipated by net motion of
28: the boundary toward the transverse region. The observed boundary motion is a
29: combination of rigid advection by the flow and order parameter
30: diffusion. The latter includes break up and reconnection of lamellae, as well
31: as a weak Eckhaus instability in the boundary region for sufficiently
32: large strain amplitude that leads to slow wavenumber readjustment.
33: The net average velocity is seen to increase with frequency and strain
34: amplitude, and can be obtained by a multiple scale expansion of the
35: governing equations.
36:
37: \end{abstract}
38:
39: \maketitle
40:
41: \section{Introduction}
42: \label{sec_intro}
43:
44: Below an order-disorder temperature, nanoscale phases of various
45: symmetries can be found in block copolymer melts
46: \cite{fredrickson96,larson99}. For example, for diblock
47: copolymers that consist of two chemically distinct but covalently bonded
48: monomers, six distinct phases that result from microphase separation have been
49: documented \cite{fredrickson96}. They differ by the symmetry of their
50: composition modulation. One of the most actively investigated
51: microphases is the lamellar phase, which can be observed around 50-50
52: composition (symmetric mixture). However, when melts are processed by thermal
53: quenching or solution-casting from a disordered phase, a macroscopic
54: size sample usually exhibits polycrystalline configurations comprised
55: of many locally ordered but randomly oriented domains (grains) and
56: large amounts of topological defects, such as grain boundaries,
57: dislocations, and disclinations. Such state of
58: partial ordering is undesirable for many applications, with significant
59: affects on e.g., the optical and mechanical performance of the material.
60: Therefore, the realization and control of long-range orientational
61: order of domains has been of great interest in both experimental
62: and theoretical studies.
63:
64: Several methods have been used experimentally to achieve
65: microstructural alignment in block copolymer melts since the
66: discovery of flow-induced alignment by Keller \textit{et
67: al.} \cite{keller70}. One of the most
68: common ways to induce global order in bulk samples is to
69: impose an oscillatory or steady shear between two parallel plates,
70: which not only has the advantage of easily characterizing the shear
71: stress and monitoring the aligning progress, but also can
72: bring some interesting new physics \cite{fredrickson96,larson99,%
73: hadziioannou79,koppi92,winey93,wiesner97,chen98,hamley01}. For the
74: case of lamellar morphology that we are interested in here, the two
75: most often observed alignments relative to the imposed shear are parallel
76: (with the lamellar layers parallel to the shearing plane), and
77: perpendicular (with the lamellar layers normal to the vorticity
78: of the shear flow). Both alignments have been
79: observed in nearly symmetric diblock copolymer
80: systems such as poly(ethylenepropylene)-poly(ethylethylene) (PEP-PEE)
81: \cite{koppi92} and polystyrene-polyisoprene (PS-PI) \cite{winey93,%
82: wiesner97}. Which of these two possible
83: alignments is selected in a given experiment depends on
84: temperature, and shear strain amplitude and frequency
85: \cite{larson99,chen98,hamley01}.
86: A third alignment direction, the so called transverse
87: orientation in which the lamellar normal is parallel to the shear
88: direction, has been found to coexist with parallel orientation for
89: entangled poly(styrene-b-ethylenepropylene) (S-EP) diblock copolymer
90: at high frequency and in the strong segregation limit \cite{pinheiro96}.
91: Other studies have focused on the kinetics of global alignment and
92: have shown that the ordering rate increases with shear frequency,
93: strain amplitude, and temperature \cite{chen98,gupta96}. Experiments
94: on PS-PI diblocks further show
95: nonlinear effects of the strain amplitude on the alignment rate, and
96: that the time scale for the development of alignment exhibits a power law
97: dependence on the strain amplitude, with an exponent equal to $-3$
98: or $-5$ depending on the stage of alignment and in different
99: frequency regimes \cite{gupta96}. Furthermore, many experiments have
100: indicated that the motion of topological defects plays an important
101: role in the global alignment of microdomains under shear, including
102: the evolution of kink band defects and tilt boundaries \cite{winey98},
103: as well as the migration and annihilation of partial focal conic
104: defects, boundaries, and tilt walls \cite{chen97,chen98}.
105:
106: A coarse-grained description of a diblock copolymer melt has been
107: used, both in the weak segregation \cite{leibler80,%
108: ohta86,fredrickson87} and strong segregation \cite{ohta86} regimes,
109: to theoretically understand the effect of shear flow and the
110: mechanisms of morphology evolution
111: \cite{cates89,fredrickson94,drolet99,chen02,doi96,ren01}. The analytic
112: studies of Cates and Milner \cite{cates89}, and Fredrickson
113: \cite{fredrickson94} focused on the order-disorder transition from
114: the isotropic state to the lamellar phase and the related alignment
115: dynamics for systems subjected to steady shear flow. Below the transition
116: temperature, the stability of uniform
117: lamellar structures under an oscillatory shear flow has been addressed
118: in both two-dimensional (2D) \cite{drolet99} and three-dimensional
119: (3D) \cite{chen02} cases. Bifurcation diagrams, including
120: secondary instabilities (Eckhaus and zigzag) were given. These analyses
121: found that although all three types of lamellar orientations could be
122: linearly stable under specific conditions, the stability range of
123: the perpendicular orientation is larger than that of the parallel
124: one, which in turn is larger than the region of stability of the
125: transverse orientation. The perpendicular alignment was also shown to
126: be the preferred orientation following the decay of unstable parallel or transverse
127: lamellae. In addition to
128: these analytic work, computer simulations have been performed to
129: investigate the dynamics of lamellar alignment in bulk samples under
130: steady \cite{doi96} or oscillatory \cite{ren01} shear flow. The
131: effects of strain amplitude and shear frequency on the degree of sample
132: alignment have been examined, as well as annihilation
133: processes of defects (such as dislocations and disclinations)
134: \cite{ren01}. Also, domain
135: coarsening in 2D diblock copolymer melts has been addressed although in
136: the absence of shear \cite{boyer01R,boyer02,shiwa02}.
137: Little is known about the effect of shear on coarsening
138: of block copolymers.
139:
140: Few of the previous theoretical studies discussed above focused
141: on the detailed dynamics and quantitative properties of topological
142: defects motion under shear, which are crucial for the understanding
143: of alignment and coarsening. In this paper we
144: study the detailed mechanisms of grain boundary
145: motion under an oscillatory shear flow building upon stability results
146: of uniform lamellar patterns \cite{drolet99,chen02}.
147: We use a simplified 2D configuration which involves only two lamellar
148: domains of parallel and transverse orientations respectively, separated
149: by a grain boundary, and focus on shears of small amplitude (less than
150: $50\%$) and low angular frequency. Compared to previous
151: numerical simulations \cite{doi96,ren01,boyer01R,boyer02,shiwa02},
152: the aspect ratio (defined as the ratio between
153: the lateral extent of the system and the lamellar wavelength) is
154: larger, so that important dynamic features associated with grain
155: boundary motion, such as diffusive relaxation of lamellae, phase
156: shift of boundary velocity, and wavenumber adjustment in the transverse
157: region, can be quantitatively analyzed.
158: These features were absent in earlier work on lamellar (roll)
159: systems without shear \cite{cross93,manneville83,tesauro87,boyer01},
160: and we argue below that they are important when the system is under
161: oscillatory shear.
162:
163: This paper is organized as follows: In Sec. \ref{sec_model}
164: we introduce a dimensionless model equation based on the Swift-Hohenberg
165: equation to describe the dynamic evolution of symmetric diblock
166: copolymer melts under shear, and describe the grain boundary
167: configuration used. The numerical results including grain
168: boundary velocity and
169: lamellar wavenumber are presented in Sec. \ref{sec_num}.
170: We derive amplitude equations governing the system evolution in
171: Sec. \ref{sec_ampl}, and compare the results with the direct solutions
172: of Sec. \ref{sec_num}. Finally, in Sec. \ref{sec_concl}, we summarize
173: our results and discuss the physical origin of the phenomena observed.
174:
175: \section{Model equation and grain boundary configuration}
176: \label{sec_model}
177:
178: The system under consideration is a symmetric diblock copolymer
179: melt below the order-disorder transition temperature $T_{\rm ODT}$.
180: For length scales larger than the microscopic monomer
181: scale (i.e., at a mesoscopic level) and time scales long enough compared
182: to the molecular relaxation of the polymer chains, a coarse-grained
183: description of the block copolymer melt can be used, with an order
184: parameter field $\psi({\bf r},t)$ representing the local density
185: difference of the two constituent monomers. In the weak segregation
186: limit, i.e., close to $T_{\rm ODT}$, a coarse-grained free energy
187: functional has been derived \cite{leibler80,ohta86,fredrickson87}:
188: \begin{equation}
189: F[\psi] = \int d{\bf r} \left \{ -\frac{\tau}{2} \psi^2 + \frac{u}{4}
190: \psi^4 + \frac{\xi}{2} \left [ \left ( \nabla^2 + {q_0^*}^2 \right )
191: \psi \right ] ^2 \right \},
192: \label{eq_F}
193: \end{equation}
194: where $\tau$ denotes a reduced temperature variable which is a
195: measure of the distance from the order-disorder transition and is
196: positive below $T_{\rm ODT}$, and $q_0^*=2\pi/\lambda_0^*$ is the
197: wavenumber of the periodic lamellar structure. Under
198: the assumption that changes in the local composition field
199: $\psi$ are driven by the free energy minimization and advection by the flow,
200: the order parameter $\psi({\mathbf r},t)$ obeys a
201: time-dependent Ginzburg-Landau equation
202: \begin{equation}
203: \frac{\partial \psi}{\partial t} + {\bf v} \cdot {\bf \nabla} \psi
204: = -\Lambda \frac{\delta F}{\delta \psi}.
205: \label{eq_G-L}
206: \end{equation}
207: Here ${\bf v}$ is the local velocity field, $\Lambda$ is an Onsager
208: kinetic coefficient and can be written as $\Lambda=M {q_0^*}^2$
209: (with $M$ a mobility).
210:
211: We introduce a length scale $1/q_0^*$, a time
212: scale $1/\Lambda \xi {q_0^*}^4$ (which is the characteristic
213: polymer relaxation time and $\simeq 1/D{q_0^*}^2$, with $D$ the
214: chain diffusivity of the copolymer), and an order parameter scale
215: $(\xi {q_0^*}^4 /u)^{1/2}$. Given PEP-PEE-2 as an example, we have
216: $\lambda_0^* \sim 30$ nm, $D \sim 10^{-11}$ cm$^2$/s for temperatures close
217: to $T_{\rm ODT}$, and hence the length scale here is about 5 nm and
218: the time scale is about 0.03 s. Consequently, the rescaled composition
219: field $\psi$ obeys a dimensionless Swift-Hohenberg equation
220: \cite{swift77} with an advection term
221: \begin{equation}
222: \frac{\partial \psi }{\partial t} + {\bf v} \cdot {\bf{\nabla}}
223: \psi= \epsilon \psi - (\nabla^2 + q_0^2)^2 \psi - \psi^3,
224: \label{eq_s-h}
225: \end{equation}
226: where $\epsilon = \tau/\xi {q_0^*}^4$ and we have $0 < \epsilon \ll 1$
227: in the weak segregation regime considered here. Also, $q_0=1$
228: although the symbol $q_0$ is retained in what follows for clarity of presentation.
229:
230: As shown in Fig. \ref{fig_conf}, we
231: consider a 2D reference state below the order-disorder transition
232: containing a planar grain boundary that separates two semi-infinite
233: ordered domains A and B. Initially both domains are in the
234: lamellar state with the same wavenumber $q_0$ but
235: oriented along different directions. We are interested here in the
236: case of a
237: $90^{\circ}$ grain boundary with two mutually perpendicular lamellar
238: sets A and B, a configuration that is known to be stable
239: against small perturbations in the absence of shear
240: \cite{manneville83,tesauro87,cross93}. The two domains are under an
241: imposed shear flow
242: \begin{equation}
243: {\mathbf v}_0 = \frac{da}{dt} z {\bf \hat x}
244: =\gamma \omega \cos (\omega t) ~ z ~{\bf \hat x},
245: \label{eq_v_0}
246: \end{equation}
247: where $da/dt$ represents the shear rate with strain
248: $a(t)=\gamma \sin(\omega t)$, $\omega$ is the angular frequency,
249: $\gamma$ the strain amplitude, and all quantities are assumed
250: dimensionless. Lamellae of domain A are transverse (with the
251: wavevector components $q_x=q_0$ and
252: $q_z=0$) at $t=0$, and those in region B parallel ($q_x=0$ and
253: $q_z=q_0$). Parallel lamellae B are marginal
254: to the shear and not distorted, while transverse
255: lamellae A are compressed, with both orientation and wavelength
256: changing following the imposed shear as shown schematically in
257: Fig. \ref{fig_conf}. Thus, we anticipate that the
258: grain boundary will not remain stationary even though both A
259: and B lamellae are linearly stable under shear. Net motion results
260: from the free energy difference between region A (compressed) and
261: region B (unchanged), as well as diffusive
262: relaxation of the order parameter as shown below.
263:
264: The stability of a uniform configuration of either parallel or
265: transverse lamellae under shear flow has been given in
266: Refs. \onlinecite{drolet99} and \onlinecite{chen02}. There exists a critical
267: strain amplitude $\gamma_c$ above which the lamellar structure of a
268: given orientation melts, with $\gamma_c \rightarrow \infty$ for parallel
269: lamellae of wavenumber $q=q_0$, and small $\gamma_c$ for transverse
270: orientation. The stability diagrams presenting secondary instability
271: boundaries (for zigzag and Eckhaus modes) for 2D system have also
272: been given in Ref. \onlinecite{drolet99}. We focus below solely on shears
273: for which both uniform parallel and transverse lamellae are linearly
274: stable. In addition, we consider the case in which shear effects are
275: of the same order of magnitude as diffusive relaxation of the order
276: parameter. Otherwise, at one extreme lamellae are passively advected
277: by the flow, whereas at the other, diffusion dominates.
278: If the velocity ${\bf v}$ in Eq. (\ref{eq_s-h}) can be approximated as ${\bf v}_0$
279: (which is the case under certain conditions \cite{fredrickson94},
280: including neglecting back flows due to osmotic stresses, and any
281: viscosity contrast between the microphases), then
282: the interesting range of $\gamma$ is such that the advection contribution
283: due to the imposed shear (${\bf v}\cdot {\bf \nabla} = (da/dt)z\partial_x$
284: in Eq. (\ref{eq_s-h})) is ${\cal O} (\epsilon)$. As will
285: be further discussed in Sec. \ref{sec_ampl} in connection with our
286: multiple scale analysis, this requires $\gamma \sim {\cal O}
287: (\epsilon^{1/4})$, an assumption that will be used in what follows.
288:
289: \section{Numerical results}
290: \label{sec_num}
291:
292: We first introduce a time dependent sheared frame of reference
293: \cite{drolet99,chen02} in which the imposed shear flow vanishes. It is
294: defined by the
295: non-orthogonal basis set $\{ {\bf e}_{x'}={\bf \hat x}$,
296: ${\bf e}_{z'}=a(t){\bf \hat x}+{\bf \hat z} \}$ shown in Fig.
297: \ref{fig_conf}. In this sheared frame, we have the coordinates
298: $x'=x-a(t) z$ and $z'=z$. Also, the corresponding reciprocal basis
299: set is defined as $\{ {\bf g}_{x'}={\bf \hat x}-a(t){\bf \hat z}$,
300: ${\bf g}_{z'}={\bf \hat z} \}$, with wavevector components expressed as
301: $q_{x'}=q_x$ and $q_{z'}=a(t)q_x + q_z$. After coordinate
302: transformation the Swift-Hohenberg equation (\ref{eq_s-h})
303: becomes
304: \begin{equation}
305: \frac{\partial \psi }{\partial t} = \epsilon \psi
306: - ({\nabla'}^2 + q_0^2)^2 \psi - \psi^3,
307: \label{eq_s-h'}
308: \end{equation}
309: where the modified Laplacian operator takes the form
310: $$
311: {\nabla'}^2 = \left[ 1+a^2(t) \right] \partial^2_{x'} - 2 a(t) \partial_{x'}
312: \partial_{z'} + \partial^2_{z'}.
313: $$
314: The critical value $\gamma_{c}$ for neutral
315: stability is independent of the shear frequency $\omega$ and given by
316: \begin{equation}
317: \gamma_c = \left ( \frac{-b + \sqrt{b^2 - 4dc}}{2d} \right )^{1/2},
318: \label{eq_gamma_c}
319: \end{equation}
320: with $b=q_{x'}^2 (2{\beta}^2+4q_{z'}^2-2q_0^2)/2$, $c={\beta}^4-2q_0{\beta}^2
321: +q_0^4 -\epsilon$, and $d=3q_{x'}^4/8$ (here ${\beta}^2=q_{x'}^2 +
322: q_{z'}^2$). The results for secondary instabilities were already
323: presented in Ref. \onlinecite{drolet99}.
324:
325: We numerically solve Eq. (\ref{eq_s-h'}) with periodic
326: boundary conditions (in the sheared frame) by using the pseudo-spectral
327: algorithm described in Ref. \onlinecite{cross94}, and also detailed
328: in the appendix of Ref. \onlinecite{drolet99}. The equation has been
329: discretized on a $1024 \times 1024$ square grid in the
330: sheared frame. In most of our calculations, we choose a mesh
331: spacing $\Delta x'=\Delta z'=\lambda_0/8$, corresponding to 8 grid
332: points per wavelength $\lambda_0$ ($=2\pi/q_0$) and so $\Delta q
333: =1/128$ for the wavenumber spacing. A second order, semi-implicit time stepping
334: algorithm is used, with a time step $\Delta t=0.1$. In order to
335: remain within the weak segregation regime we set $\epsilon=0.04$ and
336: the angular frequency $\omega$ chosen is ${\cal O} (\epsilon)$. Initial
337: conditions are obtained by numerical solution of
338: Eq. (\ref{eq_s-h'}) without shear, i.e., $a(t)=0$, from a
339: starting configuration consisting of two symmetric grain boundaries
340: located at $x'=L_{x'}/4$ and $3 L_{x'}/4$ (with $L_{x'}=1024 \Delta x'$
341: the extent of the system in the $x'$ direction), separating a parallel domain B
342: from two surrounding regions of transverse lamellae. This
343: configuration is allowed to evolve without shear until a stationary
344: solution is reached. This stationary solution is used as the initial
345: condition for the integration with $a(t) \neq 0$.
346:
347: \subsection{Grain boundary motion due to shear and diffusive
348: relaxation of the order parameter}
349: \label{subsec_motion}
350:
351: If $\gamma > \gamma_c= (8\epsilon/3)^{1/4}$, the stability limit of
352: transverse lamellae (Eq. (\ref{eq_gamma_c})), A lamellae melt to
353: a disordered state ($\psi=0$), and B lamellae
354: invade region A. If $\gamma < \gamma_{c}$,
355: but still large enough to result in a secondary instability
356: (Eckhaus, cross-roll, or zigzag) of transverse lamellae at a given
357: frequency (e.g., $\gamma=0.5$ for $\epsilon=0.04$ and small $\omega$),
358: our calculations show that small domains of parallel lamellae
359: form within the bulk transverse region A as a result of the instability.
360: These domains evolve, and connect with each other and with
361: the approaching grain boundary to increase the extent of the parallel
362: region B. In both cases, the final configuration of the system is a uniform
363: lamellar structure of parallel orientation.
364: More interesting phenomena are observed in the
365: range of $\gamma$ and $\omega$ in which both parallel and transverse
366: bulk regions are linearly stable, and when the contributions from
367: shear flow and order parameter diffusion are of the same order, as described
368: at the end of Sec. \ref{sec_model}. A typical transient configuration in this
369: parameter range is shown in Fig. \ref{fig_shear}, with $\gamma=0.4$,
370: $\omega=0.04$, and grid spacing $\Delta x=\lambda_0/32$. The
371: configuration shown corresponds to $t=2.25 T_0$, where $T_0=2\pi / \omega$
372: is the shear period, and is presented in the laboratory frame basis set
373: $\{ {\bf \hat x}, {\bf \hat z} \}$.
374:
375: Our analysis of the transient evolution of the configuration under
376: shear is based on the average location of the grain boundary $x'_{\rm
377: gb}$. To determine this quantity we use a relation similar to that of Ref.
378: \onlinecite{boyer01}: $B(x'_{\rm gb})=\sqrt{3} \sum_{i=1}^{n}
379: [\psi(x'_{\rm gb}, i\lambda_0) - \psi(x'_{\rm gb}, (i-1/2)\lambda_0)]
380: /4n = \delta$, with $n$ the number of pairs of lamellae in the $z'$ direction,
381: and $\delta$ a quantity ${\cal O}(\epsilon)$. The value used here is
382: $\delta = \epsilon/4$. The grain boundary velocity $v'_{\rm gb}$ is defined as
383: the time rate of change of $x'_{\rm gb}$. Representative results
384: (in the sheared reference frame) for $x'_{\rm gb}$ and $v'_{\rm gb}$
385: as a function of time are shown in Figs. \ref{fig_xstar}
386: and \ref{fig_vgb}. Two distinct features can be clearly distinguished
387: as illustrated in Fig. \ref{fig_shear}:
388: First, transverse lamellae in the bulk rigidly follow the
389: oscillatory shear flow, leading to a periodic change
390: in their orientation. Second, during part of the cycle the region of
391: parallel lamellae B grows as transverse lamellae in the boundary
392: region break up and reconnect as parallel lamellae (cross roll
393: instability). This process is partially reversed during the
394: rest of the cycle. The grain boundary exhibits oscillatory motion with
395: a nonzero net average as shown in Figs. \ref{fig_xstar} (location) and
396: \ref{fig_vgb} (velocity). Those portions of the shear cycle in which
397: broken transverse lamellae recombine correspond to the segments in
398: Fig. \ref{fig_xstar} with decreasing $x'_{\rm gb}$, or negative velocity
399: $v'_{\rm gb}$ in Fig. \ref{fig_vgb}.
400:
401: Whereas rigid distortion of transverse lamellae is the dominant response
402: to shear in the bulk, order parameter diffusion is important in the
403: boundary region, and is the mechanism that enables dissipation of
404: the excess free energy stored in the bulk transverse region due to shearing.
405: On the one hand, transverse lamellae are elastically compressed by the shear,
406: resulting in a net free energy increase in region A relative to B. As
407: a consequence, the elastic contribution to the system's energy is expected
408: to drive grain boundary motion toward the transverse domain
409: at all times during the shear cycle. On the other hand, backward
410: motion is also observed in Figs. \ref{fig_xstar} and \ref{fig_vgb} in
411: those portions of the shear cycle in which the magnitude
412: of the shear strain $a(t)=\gamma \sin (\omega t)$ is small, indicating
413: diffusive relaxation of the order parameter field near the
414: grain boundary. The competition between the two determines the net
415: rate of advance of the boundary. This competition is illustrated in
416: Fig. \ref{fig_vgb}(b). For $\gamma=0.3$ and
417: $0.4$ (with the same frequency $\omega=0.01$), the figure shows that
418: within one period $T_0$, a smaller strain amplitude $\gamma$
419: corresponds to larger temporal range of negative velocity, as well as
420: to a larger phase lag between the boundary velocity and the imposed
421: shear. The same effect can be seen in Fig. \ref{fig_vgb}(a), as lowering
422: the frequency enhances the effect of diffusion over elasticity.
423:
424: The net velocity of the average location of the boundary is positive
425: as shown in Figs. \ref{fig_v_w} and \ref{fig_v_gamma}. The figures
426: plot the temporal average of velocity over a period $\langle
427: v'_{\rm gb} \rangle = \int_0^{T_0} dt
428: v'_{\rm gb}(t) /T_0$ as a function of $\omega$ and $\gamma$.
429: The velocity increases sharply at very small $\omega$ and saturates at large
430: $\omega$. Diffusive relaxation is more pronounced at lower $\omega$,
431: consistent with the calculations shown in
432: Fig. \ref{fig_vgb}(a). $\langle v'_{\rm gb}\rangle$ also depends on strain amplitude
433: $\gamma$, as seen in Figs. \ref{fig_v_w} and \ref{fig_v_gamma}. The
434: average boundary velocity increases approximately as $\langle v'_{\rm gb}\rangle
435: \sim \gamma^{\alpha}$ with $\alpha \sim 4$. The range of strain
436: amplitude accessible to our calculations is limited by the restriction that
437: $\gamma \sim {\cal O}(\epsilon^{1/4})$. As discussed above,
438: the transverse lamellar region is unstable for larger $\gamma$, whereas
439: for smaller $\gamma$, diffusion of order parameter dominates.
440: Note that the power law dependence is consistent with experiments
441: in PS-PI copolymers \cite{chen98,gupta96}, in which the
442: rate of global alignment of a bulk sample is a power law of the strain
443: amplitude, with an exponent in the range $3-5$.
444:
445: \subsection{Wavenumber adjustment in the transverse region}
446: \label{subsec_wavenumber}
447:
448: Bulk transverse lamellae elastically compressed by the shear would have
449: a wavenumber $q_0 \sqrt{1+a^2(t)}$ in the laboratory frame, or
450: $q_{x'}=q_0$, constant, in the sheared frame. Order parameter
451: diffusion in the boundary region, however, is seen to lead to a
452: wavenumber modification of transverse lamellae relative to what would
453: be expected from rigid deformation. We show in Fig. \ref{fig_qx}
454: the structure factor $|\psi_{q_{x'}}|$ along the $x'$ direction
455: defined as
456: the Fourier transform of the order parameter $\psi$ at fixed $z'$
457: (e.g., at $z'=L_{z'}/2$, with $L_{z'}=1024 \Delta z'$ the system size
458: along $z'$ direction). For large enough $\gamma$ ($\gamma > 0.1$), the peak
459: in $|\psi_{q_{x'}}|$ shifts away from $q_0=1$ with time,
460: asymptotically reaching a constant $q^m_{x'} < q_0$
461: (Fig. \ref{fig_qx}(a)). We have studied this wavenumber compression
462: for different frequencies $\omega$ and strain amplitudes $\gamma$. We
463: find that the
464: value of $q^m_{x'}$ is independent of $\omega$, but that it increases
465: with decreasing $\gamma$ (Fig. \ref{fig_qx}(b)):
466: $\delta q^m_{x'}=q_0-q^m_{x'} =5 \Delta q$ (with $\Delta
467: q=1/128$) for $\gamma=0.4$ (solid line), $ \delta q^m_{x'} = 3 \Delta q$ for
468: $\gamma=0.3$ (dotted line), $ \delta q^m_{x'} = \Delta q$ for
469: $\gamma=0.2$ (dashed line), and
470: $\delta q^m_{x'} = 0$ for $\gamma=0.1$ (thin solid line). These values
471: correspond to the disappearance of $5$, $3$, $1$, and $0$
472: lamellae in region A respectively. We further discuss this finding
473: in Sec. \ref{sec_ampl}, and in the discussion section
474: \ref{sec_concl}. In addition to the motion in the position of the peak
475: of the structure factor, Fig. \ref{fig_qx}(a) also shows a decreasing
476: amplitude and broadening of the peak. This is due to the finite size
477: of the configuration; as the grain boundary moves, the region occupied
478: by transverse lamellae decreases.
479:
480: The power spectrum of region B $|\psi_{q_{z'}}|$ as a function of
481: $q_{z'}$ (the Fourier
482: transform of $\psi$ at fixed $x'$ position, results not shown here)
483: is unaffected by the shear flow. Its maximum is located at
484: $q_z=q_{z'}=q_0$, without any visible shift in time.
485:
486: \section{Multiple scale analysis and amplitude equations}
487: \label{sec_ampl}
488:
489: A multiple scale analysis of the type frequently used to derive amplitude
490: equations close to instability thresholds
491: \cite{newell69,tesauro87,cross93} can be introduced here to derive an
492: equation of motion for the grain boundary in the limit of weak
493: segregation $\epsilon \ll 1$. As shown in Fig.
494: \ref{fig_conf}, we first define a time dependent, orthogonal
495: basis set $\{ {\bf e}_{x_A} = ({\bf \hat x}-a(t){\bf \hat z})/
496: (1+a^2(t))$, ${\bf e}_{z_A} = (a(t){\bf \hat x}+{\bf \hat z})/
497: (1+a^2(t)) \}$ (different from the non orthogonal sheared frame
498: $\{ {\bf e}_{x'}$, ${\bf e}_{z'} \}$ used in Sec.
499: \ref{sec_num}). This frame is the orthogonal
500: frame of reference attached to the transverse region. The corresponding
501: coordinates are $x_A=x-a(t)z$ and $z_A=a(t)x+z$. We then introduce an
502: anisotropic coordinate scaling to define slowly varying amplitudes of
503: $\exp(iq_0 x_A)$ as $X_A=\epsilon^{1/2} x_A$ and $Z_A=\epsilon^{1/4}
504: z_A$. We retain the laboratory frame coordinates in region B
505: $\{ {\bf \hat x}$, ${\bf \hat z} \}$ with a base mode given by
506: $\exp(iq_0 z)$. Its slowly varying amplitude is a function of the
507: rescaled variables $X_B=\epsilon^{1/4} x$ and $Z_B=\epsilon^{1/2} z$.
508: We then expand the order parameter field $\psi$ as
509: \begin{equation}
510: \psi=\frac{1}{\sqrt{3}} \left ( A e^{iq_0 x_A}
511: + B e^{iq_0 z} +{\rm c.c.} \right ),
512: \label{eq_expan}
513: \end{equation}
514: where both complex amplitudes $A$ and $B$ are functions of the slow
515: spatial scales $X_A$, $Z_A$, $X_B$, and
516: $Z_B$, and of a slow time scale $T=\epsilon t$. The requirement of
517: slow amplitude change restricts our analysis to low frequencies $\omega
518: \sim {\cal O}(\epsilon)$, and we further focus on sufficiently small
519: shear amplitudes so that advection and local diffusion balance. This requires
520: ${\bf v} \cdot \nabla=(da/dt)z\partial_x \sim
521: (\nabla^2+q_0^2)^2 \sim \epsilon$ according to Eq.
522: (\ref{eq_s-h}). Considering that in the transverse region
523: $\partial_x=\partial_{x_A} + a \partial_{z_A}$, $\partial_z=
524: -a \partial_{x_A} + \partial_{z_A}$, $z=(-a x_A + z_A)/(1+a^2)$,
525: as well as the spatial $(x_A, z_A)$ and temporal scalings, we
526: require that $\gamma \sim {\cal O}(\epsilon^{1/4})$.
527: The same relationship follows from the scalings appropriate for the
528: parallel region.
529:
530: Following standard multiple scale procedure \cite{tesauro87,cross93},
531: we introduce the expansions $\partial_x \rightarrow \partial_{x_A}
532: + \epsilon^{1/4} \partial_{X_B} + \epsilon^{1/2} (\partial_{X_A}
533: + {\bar a} \partial_{Z_A})$, $\partial_z \rightarrow \partial_z
534: + \epsilon^{1/4} (-{\bar a}\partial_{x_A} + \partial_{Z_A})
535: +\epsilon^{1/2} \partial_{Z_B} - \epsilon^{3/4} {\bar a}
536: \partial_{X_A}$, and $\partial_t+(da/dt)z\partial_x \rightarrow
537: \epsilon [\partial_T + {\dot \gamma}_{\Omega} (X_A \partial_{Z_A}
538: +Z_B \partial_{X_B})] + {\cal O}(\epsilon^{5/4})$ (here ${\bar a}$
539: and ${\dot \gamma}_{\Omega}$ are defined by $a=\epsilon^{1/4}{\bar a}$
540: and $da/dt=\epsilon^{5/4}{\dot \gamma}_{\Omega}$), and we derive
541: the following amplitude equations at ${\cal O}(\epsilon^{3/2})$ from
542: Eq. (\ref{eq_s-h}),
543: \begin{eqnarray}
544: \partial_t A &=& \left [ \epsilon - \frac{da}{dt}
545: x_A \partial_{z_A} - \left ( 2iq_0 \partial_{x_A} + \partial_{z_A}^2
546: - q_0^2 a^2 \right )^2 \right ]A - |A|^2 A -2|B|^2 A,
547: \label{eq_A_local}\\
548: \partial_t B &=& \left [ \epsilon - \frac{da}{dt}
549: z \partial_x - \left (\partial_x^2 + 2iq_0\partial_z \right )^2
550: \right ] B -|B|^2 B - 2|A|^2 B.
551: \label{eq_B_local}
552: \end{eqnarray}
553: For $\gamma=0$, these equations reduce to those of Refs.
554: \onlinecite{manneville83} and \onlinecite{tesauro87}.
555:
556: Equations (\ref{eq_A_local}) and (\ref{eq_B_local}) are expressed
557: in two different coordinates systems. We next transform them
558: to a common sheared frame $\{ {\bf e}_{x'}$, ${\bf e}_{z'} \}$,
559: by using the relations $x_A=x'$ and $z_A=ax'+(1+a^2)z'$. The base
560: state of the order parameter is still given by Eq. (\ref{eq_expan})
561: (with $x_A$ replaced by $x'$), and to ${\cal O}(\epsilon^{3/2})$, the resulting 2D
562: amplitude equations were already given in Ref. \onlinecite{huang03}.
563: Here we further assume a planar grain boundary in the sheared frame,
564: for which the dependence of the amplitudes on the coordinate ($z'$) parallel
565: to the grain boundary can be ignored. The complex amplitudes $A$
566: and $B$ satisfy the one-dimensional (1D) equations
567: \begin{equation}
568: \partial_t A = \left [\epsilon- ( 2 i q_0 \partial_{x'}
569: -q_0^2 a^2 )^2 \right ] A -|A|^2 A -2 |B|^2 A,
570: \label{eq_A}
571: \end{equation}
572: and
573: \begin{equation}
574: \partial_t B = \left [\epsilon- ( -2 i q_0 a \partial_{x'}
575: + \partial^2_{x'} )^2 \right ] B -|B|^2 B -2 |A|^2 B.
576: \label{eq_B}
577: \end{equation}
578: At ${\cal O}(\epsilon^{3/2})$, two contributions from the shear flow
579: remain in Eqs. (\ref{eq_A}) and (\ref{eq_B}). The first one
580: involves the term $-2 i q_0 a \partial_{x'}$ in Eq. (\ref{eq_B})
581: and the term $(q_0^2 a^2)(2 i q_0 \partial_{x'})$ obtained from
582: the expansion of Eq. (\ref{eq_A}), which is non
583: negligible only in the grain boundary region and leads to diffusive
584: relaxation of the order parameter. The second is
585: $q_0^4 a^4 A$ in Eq. (\ref{eq_A}). This term is uniform in the
586: entire region A, and reflects the contribution from advection of
587: transverse lamellae by the flow.
588:
589: We had analytically calculated the velocity
590: of the grain boundary from these amplitude equations
591: in Ref. \onlinecite{huang03} by assuming
592: that the amplitudes can be approximated by
593: $A(x',t) \simeq A(x'-x'_{\rm gb}(t))$ and
594: $B(x',t) \simeq B(x'-x'_{\rm gb}(t))$. We found that the velocity is
595: proportional to the
596: free energy difference between the transverse and parallel phases, in
597: agreement with previous studies in the absence of flow
598: \cite{manneville83,boyer01,boyer02}. Also, the results gave the
599: correct order of magnitude of the
600: average velocity, but we noted quantitative discrepancies
601: \cite{huang03}. We argued that the adiabatic approximation for
602: $A(x',t)$ and $B(x',t)$ given cannot incorporate diffusive relaxation
603: of the order parameter in the
604: boundary region so that the calculation only yields an upper
605: bound to the net boundary velocity. Since we have argued in
606: Sec. \ref{sec_num} that this
607: relaxation is important, we turn here to a numerical determination of
608: the boundary velocity.
609:
610: We present next the results of the numerical solution of
611: Eqs. (\ref{eq_A}) and (\ref{eq_B}). We initially consider
612: a region of parallel lamellae B surrounded by two regions
613: of transverse lamellae A, and use periodic boundary conditions in the
614: integration. Both amplitudes $A$ and $B$ are complex variables.
615: A pseudo-spectral method is applied, with a Crank-Nicholson scheme
616: used for the linear terms, and a second-order Adams-Bashford scheme
617: used for the nonlinear terms. The
618: instantaneous location of the grain boundary $x'_{\rm gb}(t)$ is
619: defined by the condition $|B(x'_{\rm gb})|=\epsilon/4$, and its
620: velocity $v'_{\rm gb}$ is found by taking the time derivative of
621: $x'_{\rm gb}(t)$. In order to compare with the 2D
622: results of the original model shown in Sec. \ref{sec_num}, we set the
623: system size $L=1024$, the time step $\Delta t=0.1$, and the grid
624: spacing $\Delta x'=\lambda_0/8$. As was the case there, the initial
625: condition for $A$ and $B$ is provided by the steady solution of the
626: amplitude equations in the absence of shear.
627: Our results for the boundary velocity for $\gamma=0.4$, $\omega=0.01$,
628: and $\omega = 0.04$ are shown as dotted curves in
629: Fig. \ref{fig_vgb}(a), both in good
630: agreement with the direct solution of original model equation (\ref{eq_s-h'})
631: (symbols in the figure). The time averaged velocity $\langle v'_{\rm
632: gb} \rangle$ is shown by the dashed lines in Figs. \ref{fig_v_w} and
633: \ref{fig_v_gamma}.
634:
635: In order to further analyze the wavenumber readjustment process discussed in
636: Sec. \ref{subsec_wavenumber}, we show our results in terms of the phase
637: $\phi_A$ of the complex amplitude $A$. In the sheared frame we define
638: $A=|A| \exp(i\phi_A)$ and plot $\phi_A$ as a function of
639: grid index $i_{x'} = x'/\Delta x'$ in Fig. \ref{fig_phaseA}.
640: Near the boundary, the phase becomes linear in space: $\phi_A
641: \propto -\delta q \cdot x'$ (the linear behavior is clearer
642: for a larger system size, as seen by comparing Fig. \ref{fig_phaseA}(a)
643: ($L=1024$) with \ref{fig_phaseA}(b) ($L=4096$)), indicating a local
644: wavenumber change $q_{x'} \rightarrow q_0-\delta q$. This is also in
645: agreement with the direct solution of the original model as shown in
646: Fig. \ref{fig_qx}. Note also that the region of linearity (right side of
647: dot-dashed line in Fig. \ref{fig_phaseA}) increases with time, indicating that
648: the re-adjustment of the local wavelength of the transverse lamellae
649: first occurs at the boundary, and then progressively propagates into
650: the bulk. At late times (e.g., $t=40 T_0$ in Fig. \ref{fig_phaseA}(a)),
651: the wavenumber change can be observed in the whole transverse domain,
652: with $q_{x'}$ corresponding to the stationary peak position
653: of the structure factor $|\psi_{q_{x'}}|$ presented in
654: Fig. \ref{fig_qx}(a) ($t \geq 40 T_0$ there). We find that
655: $\delta q = \delta q^m_{x'}$, the wavenumber shift discussed in
656: Sec. \ref{sec_num}, by determining $\delta q$ from the slope of the
657: dotted line in Figs. \ref{fig_phaseA}(a)and (b).
658:
659: \section{Discussion and conclusions}
660: \label{sec_concl}
661:
662: A coarse grained order parameter model has been used to study
663: the motion of a grain boundary separating two regions of uniform
664: parallel and transverse lamellae under an imposed shear flow. The
665: motion of the boundary is oscillatory, and
666: the driving force for motion is the excess energy stored in the elastically
667: strained transverse phase that can only be relieved through diffusive
668: relaxation of the order parameter in the boundary region. Diffusive
669: relaxation, however, is complex as the response of the order parameter
670: field is out of phase with the shear, and lamellae break up and reconnect
671: during each of the cycles. As expected, the effects of diffusive relaxation
672: are more pronounced for small shear strain and low angular frequency,
673: as seen in both the time dependent behavior of boundary velocity (Fig.
674: \ref{fig_vgb}) or the time averaged velocity (Figs. \ref{fig_v_w} and
675: \ref{fig_v_gamma}). Although under the conditions of the study both
676: transverse and parallel orientations are linearly stable, we observe net
677: motion of the boundary toward the region occupied by transverse lamellae.
678:
679: As the boundary moves over time, we have observed that the wavenumber
680: of the transverse lamellae does not remain constant and equal to
681: $q_0$. Instead, it is slowly readjusted by wavenumber diffusion, as
682: shown by both direct solution of the coarse grained model
683: of Sec. \ref{sec_num} and the corresponding complex amplitude
684: equations of Sec. \ref{sec_ampl}. The wavenumber shift
685: $\delta q^m_{x'}$ is approximately independent of shear frequency $\omega$,
686: but strongly dependent on the strain amplitude $\gamma$. In order to understand
687: the physical origin of wavenumber compression (or
688: lamellae expansion) that occurs in the transverse region A, we
689: focus on the lamellae near the grain boundary, since the calculation
690: in Sec. \ref{sec_ampl} shows that
691: wavenumber change is initiated at the boundary, and then propagates into
692: the bulk. Since the amplitude of the transverse lamellae go to zero
693: at the boundary region, it is possible to create or eliminate lamellar
694: planes there in a way that it is not possible in the bulk for the parameters of
695: our study \cite{drolet99}. First consider the stability diagram of the
696: Swift-Hohenberg model (\ref{eq_s-h}) at zero shear
697: \cite{greenside84}. For fixed $\epsilon$ and $q_x > q_0$, the closest
698: instability boundary in the $q_x$-$\epsilon$ diagram is the Eckhaus
699: instability given by
700: \begin{equation}
701: \epsilon = 12 (q_x - q_0)^2.
702: \label{eq_Eck}
703: \end{equation}
704: For $\epsilon=0.04$ and $q_0=1$, we have the Eckhaus boundary at
705: $q_x^{\rm E} = q_0+(\epsilon /12)^{1/2}
706: =1.0577$. In our case $q_x$ is the wavenumber of the transverse
707: lamellae in the lab frame, and equal to $q_0 \sqrt{1+a^2(t)}$ if we
708: assume that the lamellae are rigidly distorted by the shear.
709: The maximum value of $q_x$ is then $q_x^{\rm
710: max}=q_0\sqrt{1+\gamma^2}$ so that
711: for $\gamma=0.4$ we have $q_x^{\rm max} > q_x^{\rm E}$. Although this
712: is not sufficient to destabilize the bulk transverse region (according
713: to the stability diagrams in Ref. \onlinecite{drolet99} obtained by
714: Floquet analysis over the entire period of the oscillation), it
715: appears to be sufficient to induce lamellar elimination at the grain
716: boundary region, as seen in Figs. \ref{fig_qx} and
717: \ref{fig_phaseA}. The figures show that $\delta q^m_{x'}=5 \Delta q$,
718: corresponding to the elimination of $5$ transverse lamellae. As $\gamma$
719: decreases $q_x^{\rm max}$ also decreases, becoming smaller
720: than $q_x^{\rm E}$, and eventually lamellae elimination is expected to cease.
721: This is consistent with our results shown in Fig. \ref{fig_qx}(b). With
722: decreasing value of $\gamma$ from $0.4$ to $0.1$, the number of
723: lost transverse lamellae decreases from $5$ to $0$.
724:
725: In summary, for small shear strains and low frequencies such that
726: diffusion of order parameter is of the same order as advection by the
727: flow, the excess free energy in the transverse region relative to the
728: marginal parallel region is dissipated through order parameter
729: diffusion in the grain boundary region. The latter includes break up
730: and reconnection of transverse lamellae, and a weak Eckhaus instability
731: developing at the grain boundary that diffuses into the bulk
732: transverse lamellae leading to dynamical wavenumber readjustment. A
733: weakly nonlinear analysis, as well as the amplitude equations derived,
734: capture quantitatively all aspects of grain
735: boundary motion, including the boundary velocity and the wavenumber
736: readjustment. The order parameter distribution in the boundary region
737: can be represented crudely by introducing an
738: adiabatic approximation into the amplitude equations, which gives a
739: reasonable approximation to the net boundary velocity toward the
740: transverse region, and be well reproduced by the direct solution
741: of the amplitude equations.
742: Although our study is confined to the case of a transverse/parallel
743: grain boundary in two dimensions, we expect that our results will
744: qualitatively hold for both three dimensional transverse/parallel
745: and transverse/perpendicular cases. In three dimensions, however,
746: there is a completely different type of tilt boundary, that between
747: parallel and perpendicular lamellae. Both orientations are marginal with
748: respect to the shear. This configuration is currently under
749: investigation.
750:
751:
752: \begin{acknowledgments}
753: This work was supported by the National Science Foundation
754: under grant DMR-0100903.
755: \end{acknowledgments}
756:
757:
758: \begin{thebibliography}{}
759:
760: \bibitem{fredrickson96}
761: G. H. Fredrickson and F. S. Bates, Annu. Rev. Mater. Sci.
762: {\bf 26}, 501 (1996).
763:
764: \bibitem{larson99}
765: R. G. Larson, \textit{The structure and Rheology of Complex Fluids}
766: (Oxford University Press, New York, 1999).
767:
768: \bibitem{keller70}
769: A. Keller, E. Pedemonte, and F. M. Willmouth, Kolloid-Z. Z. Polym.
770: \textbf{238}, 385 (1970); Nature \textbf{225}, 538 (1970).
771:
772: \bibitem{hadziioannou79}
773: G. Hadziioannou, A. Mathis, A. Skoulios, Colloid Polym. Sci.
774: \textbf{257}, 136 (1979).
775:
776: \bibitem{koppi92}
777: K. A. Koppi, M. Tirrell, F. S. Bates, K. Almdal, and R. H. Colby,
778: J. Phys. II (France) \textbf{2}, 1941 (1992).
779:
780: \bibitem{winey93}
781: K. I. Winey, S. S. Patel, R. G. Larson, and H. Watanabe,
782: Macromolecules \textbf{26}, 2542 (1993).
783:
784: \bibitem{wiesner97}
785: U. Wiesner, Macromol. Chem. Phys. \textbf{198}, 3319 (1997);
786: D. Maring and U. Wiesner, Macromolecules \textbf{30}, 660 (1997).
787:
788: \bibitem{chen98}
789: Z.-R. Chen and J. A. Kornfield, Polymer \textbf{39}, 4679 (1998).
790:
791: \bibitem{hamley01}
792: I. W. Hamley, J. Phys.: Condens. Matter \textbf{13}, R643 (2001).
793:
794: \bibitem{pinheiro96}
795: B. S. Pinheiro, D. A. Hajduk, S. M. Gruner, and K. I. Winey,
796: Macromolecules \textbf{29}, 1482 (1996).
797:
798: \bibitem{gupta96}
799: V. K. Gupta, R. Krishnamoorti, Z.-R. Chen, J. A. Kornfield,
800: S. D. Smith, M. M. Satkowski, and J. T. Grothaus, Macromolecules
801: \textbf{29}, 875 (1996).
802:
803: \bibitem{winey98}
804: D. L. Polis and K. I. Winey, Macromolecules \textbf{31}, 3617 (1998);
805: L. Qiao and K. I. Winey, \textit{ibid.} \textbf{33}, 851 (2000).
806:
807: \bibitem{chen97}
808: Z.-R. Chen, A. M. Issaian, J. A. Kornfield, S. D. Smith, J. T. Grothaus,
809: and M. M. Satkowski, Macromolecules \textbf{30}, 7096 (1997).
810:
811: \bibitem{leibler80}
812: L. Leibler, Macromolecules \textbf{13}, 1602 (1980).
813:
814: \bibitem{ohta86}
815: T. Ohta and K. Kawasaki, Macromolecules \textbf{19}, 2621 (1986).
816:
817: \bibitem{fredrickson87}
818: G. H. Fredrickson and E. Helfand, J. Chem. Phys. \textbf{87}, 697
819: (1987).
820:
821: \bibitem{cates89}
822: M. E. Cates and S. T. Milner, Phys. Rev. Lett. \textbf{62}, 1856
823: (1989).
824:
825: \bibitem{fredrickson94}
826: G. H. Fredrickson, J. Rheol. \textbf{38}, 1045 (1994).
827:
828: \bibitem{drolet99}
829: F. Drolet, P. Chen, and J. {Vi\~nals}, Macromolecules \textbf{32},
830: 8603 (1999).
831:
832: \bibitem{chen02}
833: P. Chen and J. {Vi\~nals}, Macromolecules \textbf{35}, 4183 (2002).
834:
835: \bibitem{doi96}
836: H. Kodama and M. Doi, Macromolecules \textbf{29}, 2652 (1996).
837:
838: \bibitem{ren01}
839: S. R. Ren, I. W. Hamley, P. I. C. Teixeira, and P. D. Olmsted,
840: Phys. Rev. E \textbf{63}, 041503 (2001).
841:
842: \bibitem{boyer01R}
843: D. Boyer and J. Vi\~nals, Phys. Rev. E \textbf{64}, 050101(R) (2001).
844:
845: \bibitem{boyer02}
846: D. Boyer and J. Vi\~nals, Phys. Rev. E \textbf{65}, 046119 (2002).
847:
848: \bibitem{shiwa02}
849: Y. Yokojima and Y. Shiwa, Phys. Rev. E \textbf{65}, 056308 (2002).
850:
851: \bibitem{cross93}
852: M. C. Cross and P. C. Hohenberg, Rev. Mod. Phys. \textbf{65}, 851
853: (1993).
854:
855: \bibitem{manneville83}
856: P. Manneville and Y. Pomeau, Phil. Mag. A \textbf{48}, 607 (1983).
857:
858: \bibitem{tesauro87}
859: G. Tesauro, and M. C. Cross, Phil. Mag. A \textbf{56}, 703 (1987).
860:
861: \bibitem{boyer01}
862: D. Boyer and J. Vi\~nals, Phys. Rev. E \textbf{63}, 061704 (2001).
863:
864: \bibitem{swift77}
865: J. Swift and P. C. Hohenberg, Phys. Rev. A \textbf{15}, 319 (1977).
866:
867: \bibitem{cross94}
868: M. C. Cross, D. Meiron, and Y. Tu, Chaos \textbf{4}, 607 (1994).
869:
870: \bibitem{newell69} A. C. Newell and J. A. Whitehead, J. Fluid Mech.
871: \textbf{38}, 279 (1969); L. A. Segel, {\it ibid.} \textbf{38}, 203
872: (1969).
873:
874: \bibitem{huang03} Z. F. Huang, F. Drolet, and J. Vi\~nals,
875: Macromolecules, in press.
876:
877: \bibitem{greenside84}
878: H. S. Greenside and W. M. Coughran, Jr., Phys. Rev. A \textbf{30},
879: 398 (1984).
880:
881: \end{thebibliography}
882:
883: \newpage
884:
885: \begin{figure}
886: \centerline{\epsfig{figure=fig1.eps,width=5in}}
887: \caption{Schematic representation of the two dimensional grain
888: boundary configuration under oscillatory shear flow studied in this
889: paper. Both the non orthogonal sheared frame $\{{\bf e}_{x'}, {\bf e}_{z'}\}$,
890: and the auxiliary frame $\{{\bf e}_{x_{\rm A}},
891: {\bf e}_{z_{\rm A}}\}$ are indicated.
892: }
893: \label{fig_conf}
894: \end{figure}
895:
896: \begin{figure}
897: \centerline{\epsfig{figure=fig2.eps,width=3in}}
898: %\centerline{\epsfig{figure=fig2.pdf,width=3in}}
899: %\centerline{\includegraphics[width=3in]{fig2.jpg}}
900: \caption{Grain boundary configuration (in gray scale) at time $t=2.25 T_0$
901: obtained by numerically solving the Swift-Hohenberg equation
902: (\ref{eq_s-h'}) with $\epsilon=0.04$,
903: $\gamma=0.4$, $\omega=0.04$, and $\Delta x = \lambda_0/32$. The
904: configuration shown here has been transformed back to the laboratory frame
905: $\{ {\bf \hat x}, {\bf \hat z} \}$. As time evolves, the transverse
906: domain is invaded by parallel lamellae, until a uniform parallel
907: configuration occupies the whole system.
908: }
909: \label{fig_shear}
910: \end{figure}
911:
912: \begin{figure}
913: \centerline{\epsfig{figure=fig3.eps,width=4.5in}}
914: \caption{Relative grain boundary displacement in the sheared frame
915: $(x'_{\rm gb}-x'_{\rm gb}(t=0))/\lambda_{0}$ as a function of time $t/T_{0}$ for
916: $\epsilon=0.04$, $\gamma=0.4$, $\omega=0.04$, and $\Delta x=
917: \lambda_0/8$.
918: }
919: \label{fig_xstar}
920: \end{figure}
921:
922: \begin{figure}
923: \centerline{\epsfig{figure=fig4a.eps,width=4.in}}
924: \centerline{\epsfig{figure=fig4b.eps,width=4.in}}
925: \caption{Boundary velocity $v'_{\rm gb}$ as
926: a function of time obtained from direct numerical integration of
927: Eq. (\ref{eq_s-h'}) with $\epsilon=0.04$ and different values of
928: $\omega$ and $\gamma$. (a) Fixed strain amplitude $\gamma=0.4$,
929: $\omega=0.01$ (filled triangles), and $\omega = 0.04$ (open
930: circles). The dotted lines are the corresponding results obtained from
931: numerical integration of the amplitude equations (\ref{eq_A}) and (\ref{eq_B}).
932: (b) Results at constant frequency $\omega=0.01$, but different strain
933: amplitudes $\gamma=0.4$ (solid line), and $0.3$ (dashed line).
934: }
935: \label{fig_vgb}
936: \end{figure}
937:
938: \begin{figure}
939: \centerline{\epsfig{figure=fig5.eps,width=4.5in}}
940: \caption{Temporal average of the grain boundary velocity $\langle
941: v'_{\rm gb}\rangle$ as a function of frequency $\omega$. The symbols
942: correspond to the solution of the order parameter model Eq. (\ref{eq_s-h'}) for
943: $\gamma=0.4$ (circles), and $0.3$ (squares), while the corresponding
944: dashed lines are obtained from the amplitude equations (\ref{eq_A})
945: and (\ref{eq_B}).
946: }
947: \label{fig_v_w}
948: \end{figure}
949:
950: \begin{figure}
951: \centerline{\epsfig{figure=fig6.eps,width=4.5in}}
952: \caption{Log-log plot of average velocity $\langle v'_{\rm gb}
953: \rangle$ versus strain amplitude $\gamma$ for $\omega=0.01$
954: (squares), $0.02$ (circles), and $0.1$ (triangles) from
955: direct solution of Eq. (\ref{eq_s-h'}). Also shown
956: (dashed lines) are the corresponding results given by the amplitude
957: equations (\ref{eq_A}) and (\ref{eq_B}).
958: }
959: \label{fig_v_gamma}
960: \end{figure}
961:
962: \begin{figure}
963: \centerline{\epsfig{figure=fig7a.eps,width=4.in}}
964: \centerline{\epsfig{figure=fig7b.eps,width=4.in}}
965: \caption{One dimensional structure factor for transverse lamellae
966: $|\psi_{q_{x'}}|$ along the
967: $x'$ direction at $z'=L_{z'}/2$ as a function of wavenumber $q_{x'}$.
968: (a), different times from
969: top to bottom: $t=30 T_0$, $40 T_0$, $50 T_0$ and $60 T_0$,
970: with $\gamma=0.4$ and $\omega=0.04$.
971: (b), different strain amplitudes: $\gamma=0.4$ and $t=100 T_0$ (solid
972: curve), $\gamma = 0.3$ and $t=300 T_0$ (dotted curve),
973: $\gamma = 0.2$ and $t=1200 T_0$ (dashed curve), and $\gamma = 0.1$ and
974: $t=12000 T_0$ (thin solid curve). Here $\omega=0.1$, and
975: $\Delta t=0.1$ except for $\gamma=0.1$ in which case we have used
976: $\Delta t=0.2$.
977: In both cases (a) and (b), the vertical dot-dashed line indicates the location of
978: the wavenumber $q_{x'}=q_0=1$.
979: }
980: \label{fig_qx}
981: \end{figure}
982:
983: \begin{figure}
984: \centerline{\epsfig{figure=fig8a.eps,width=4.in}}
985: \centerline{\epsfig{figure=fig8b.eps,width=4.in}}
986: \caption{Phase $\phi_A$ of the complex amplitude $A$ as a function of
987: position in the sheared frame (in terms of the grid index $i_{x'}$)
988: with $\Delta x'= \lambda_0/8$, $\gamma=0.4$, and $\omega=0.04$.
989: Two system sizes are shown: (a) $L=1024$ at times
990: (from left to right): $t=2 T_0$, $5 T_0$, $10 T_0$, $20 T_0$, and
991: $40 T_0$ (to be compared with Fig. \ref{fig_qx}(a)); and (b) $L=4096$
992: at times (from left to right) $t=10 T_0$, $50 T_0$, $100 T_0$, and
993: $200 T_0$. The vertical dot-dashed lines indicate the instantaneous grain boundary
994: positions $x'_{\rm gb}$, and a dotted line with slope $-\delta
995: q^m_{x'} \cdot \Delta x'=-5\Delta x'/128$ is also shown for reference.
996: }
997: \label{fig_phaseA}
998: \end{figure}
999:
1000: \end{document}
1001: