ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/group/trunk/ripple2/ripple.tex
Revision: 3075
Committed: Thu Nov 9 14:41:45 2006 UTC (17 years, 7 months ago) by gezelter
Content type: application/x-tex
File size: 31668 byte(s)
Log Message:
Import from sources

File Contents

# User Rev Content
1 gezelter 3075 \documentclass[aps,pre,endfloats*,preprint,amssymb,showpacs]{revtex4}
2     \usepackage{epsfig}
3    
4     \begin{document}
5     \renewcommand{\thefootnote}{\fnsymbol{footnote}}
6     \renewcommand{\theequation}{\arabic{section}.\arabic{equation}}
7    
8     %\bibliographystyle{aps}
9    
10     \title{Spontaneous Corrugation of Dipolar Membranes}
11     \author{Xiuquan Sun and J. Daniel Gezelter}
12     \email[]{E-mail: gezelter@nd.edu}
13     \affiliation{Department of Chemistry and Biochemistry,\\
14     University of Notre Dame, \\
15     Notre Dame, Indiana 46556}
16    
17     \date{\today}
18    
19     \begin{abstract}
20     We present a simple model for dipolar membranes that gives
21     lattice-bound point dipoles complete orientational freedom as well as
22     translational freedom along one coordinate (out of the plane of the
23     membrane). There is an additional harmonic surface tension which
24     binds each of the dipoles to the six nearest neighbors on either
25     hexagonal or distorted-hexagonal lattices. The translational freedom
26     of the dipoles allows hexagonal lattices to find states that break out
27     of the normal orientational disorder of frustrated configurations and
28     which are stabilized by long-range antiferroelectric ordering. In
29     order to break out of the frustrated states, the dipolar membranes
30     form corrugated or ``rippled'' phases that make the lattices
31     effectively non-hexagonal. We observe three common features of the
32     corrugated dipolar membranes: 1) the corrugated phases develop easily
33     when hosted on hexagonal lattices, 2) the wave vectors for the surface
34     ripples are always found to be perpendicular to the dipole director
35     axis, and 3) on hexagonal lattices, the dipole director axis is found
36     to be parallel to any of the three equivalent lattice directions.
37     \end{abstract}
38    
39     \pacs{68.03.Hj, 82.20.Wt}
40     \maketitle
41    
42    
43     \section{Introduction}
44     \label{Int}
45     There has been intense recent interest in the phase behavior of
46     dipolar
47     fluids.\cite{Tlusty00,Teixeira00,Tavares02,Duncan04,Holm05,Duncan06}
48     Due to the anisotropic interactions between dipoles, dipolar fluids
49     can present anomalous phase behavior. Examples of condensed-phase
50     dipolar systems include ferrofluids, electro-rheological fluids, and
51     even biological membranes. Computer simulations have provided useful
52     information on the structural features and phase transition of the
53     dipolar fluids. Simulation results indicate that at low densities,
54     these fluids spontaneously organize into head-to-tail dipolar
55     ``chains''.\cite{Teixeira00,Holm05} At low temperatures, these chains
56     and rings prevent the occurrence of a liquid-gas phase transition.
57     However, Tlusty and Safran showed that there is a defect-induced phase
58     separation into a low-density ``chain'' phase and a higher density
59     Y-defect phase.\cite{Tlusty00} Recently, inspired by experimental
60     studies on monolayers of dipolar fluids, theoretical models using
61     two-dimensional dipolar soft spheres have appeared in the literature.
62     Tavares {\it et al.} tested their theory for chain and ring length
63     distributions in two dimensions and carried out Monte Carlo
64     simulations in the low-density phase.\cite{Tavares02} Duncan and Camp
65     performed dynamical simulations on two-dimensional dipolar fluids to
66     study transport and orientational dynamics in these
67     systems.\cite{Duncan04} They have recently revisited two-dimensional
68     systems to study the kinetic conditions for the defect-induced
69     condensation into the Y-defect phase.\cite{Duncan06}
70    
71     Although they are not traditionally classified as 2-dimensional
72     dipolar fluids, hydrated lipids aggregate spontaneously to form
73     bilayers which exhibit a variety of phases depending on their
74     temperatures and compositions. At high temperatures, the fluid
75     ($L_{\alpha}$) phase of Phosphatidylcholine (PC) lipids closely
76     resembles a dipolar fluid. However, at lower temperatures, packing of
77     the molecules becomes important, and the translational freedom of
78     lipid molecules is thought to be substantially restricted. A
79     corrugated or ``rippled'' phase ($P_{\beta'}$) appears as an
80     intermediate phase between the gel ($L_\beta$) and fluid
81     ($L_{\alpha}$) phases for relatively pure phosphatidylcholine (PC)
82     bilayers. The $P_{\beta'}$ phase has attracted substantial
83     experimental interest over the past 30 years. Most structural
84     information of the ripple phase has been obtained by the X-ray
85     diffraction~\cite{Sun96,Katsaras00} and freeze-fracture electron
86     microscopy (FFEM).~\cite{Copeland80,Meyer96} Recently, Kaasgaard {\it
87     et al.} used atomic force microscopy (AFM) to observe ripple phase
88     morphology in bilayers supported on mica.~\cite{Kaasgaard03} The
89     experimental results provide strong support for a 2-dimensional
90     hexagonal packing lattice of the lipid molecules within the ripple
91     phase. This is a notable change from the observed lipid packing
92     within the gel phase.~\cite{Cevc87}
93    
94     Although the results of dipolar fluid simulations can not be directly
95     mapped onto the phases of lipid bilayers, the rich behaviors exhibited
96     by simple dipolar models can give us some insight into the corrugation
97     phenomenon of the $P_{\beta'}$ phase. There have been a number of
98     theoretical approaches (and some heroic simulations) undertaken to try
99     to explain this phase, but to date, none have looked specifically at
100     the contribution of the dipolar character of the lipid head groups
101     towards this corrugation. Before we present our simple model, we will
102     briefly survey the previous theoretical work on this topic.
103    
104     The theoretical models that have been put forward to explain the
105     formation of the $P_{\beta'}$ phase have presented a number of
106     conflicting but intriguing explanations. Marder {\it et al.} used a
107     curvature-dependent Landau-de Gennes free-energy functional to predict
108     a rippled phase.~\cite{Marder84} This model and other related
109     continuum models predict higher fluidity in convex regions and that
110     concave portions of the membrane correspond to more solid-like
111     regions. Carlson and Sethna used a packing-competition model (in
112     which head groups and chains have competing packing energetics) to
113     predict the formation of a ripple-like phase. Their model predicted
114     that the high-curvature portions have lower-chain packing and
115     correspond to more fluid-like regions. Goldstein and Leibler used a
116     mean-field approach with a planar model for {\em inter-lamellar}
117     interactions to predict rippling in multilamellar
118     phases.~\cite{Goldstein88} McCullough and Scott proposed that the {\em
119     anisotropy of the nearest-neighbor interactions} coupled to
120     hydrophobic constraining forces which restrict height differences
121     between nearest neighbors is the origin of the ripple
122     phase.~\cite{McCullough90} Lubensky and MacKintosh introduced a Landau
123     theory for tilt order and curvature of a single membrane and concluded
124     that {\em coupling of molecular tilt to membrane curvature} is
125     responsible for the production of ripples.~\cite{Lubensky93} Misbah,
126     Duplat and Houchmandzadeh proposed that {\em inter-layer dipolar
127     interactions} can lead to ripple instabilities.~\cite{Misbah98}
128     Heimburg presented a {\em coexistence model} for ripple formation in
129     which he postulates that fluid-phase line defects cause sharp
130     curvature between relatively flat gel-phase regions.~\cite{Heimburg00}
131     Kubica has suggested that a lattice model of polar head groups could
132     be valuable in trying to understand bilayer phase
133     formation.~\cite{Kubica02} Bannerjee used Monte Carlo simulations of
134     lamellar stacks of hexagonal lattices to show that large headgroups
135     and molecular tilt with respect to the membrane normal vector can
136     cause bulk rippling.~\cite{Bannerjee02}
137    
138     Large-scale molecular dynamics simulations have also been performed on
139     rippled phases using united atom as well as molecular scale
140     models. De~Vries {\it et al.} studied the structure of lecithin ripple
141     phases via molecular dynamics and their simulations seem to support
142     the coexistence models (i.e. fluid-like chain dynamics was observed in
143     the kink regions).~\cite{deVries05} A similar coarse-grained approach
144     has been used to study the line tension of bilayer
145     edges.\cite{Jiang04,deJoannis06} Ayton and Voth have found significant
146     undulations in zero-surface-tension states of membranes simulated via
147     dissipative particle dynamics, but their results are consistent with
148     purely thermal undulations.~\cite{Ayton02} Brannigan, Tamboli and
149     Brown have used a molecular scale model to elucidate the role of
150     molecular shape on membrane phase behavior and
151     elasticity.~\cite{Brannigan04b} They have also observed a buckled
152     hexatic phase with strong tail and moderate alignment
153     attractions.~\cite{Brannigan04a}
154    
155     The problem with using atomistic and even coarse-grained approaches to
156     study this phenomenon is that only a relatively small number of
157     periods of the corrugation (i.e. one or two) can be realistically
158     simulated given current technology. Also, simulations of lipid
159     bilayers are traditionally carried out with periodic boundary
160     conditions in two or three dimensions and these have the potential to
161     enhance the periodicity of the system at that wavelength. To avoid
162     this pitfall, we are using a model which allows us to have
163     sufficiently large systems so that we are not causing artificial
164     corrugation through the use of periodic boundary conditions.
165    
166     At the other extreme in density from the traditional simulations of
167     dipolar fluids is the behavior of dipoles locked on regular lattices.
168     Ferroelectric states (with long-range dipolar order) can be observed
169     in dipolar systems with non-hexagonal packings. However, {\em
170     hexagonally}-packed 2-D dipolar systems are inherently frustrated and
171     one would expect a dipolar-disordered phase to be the lowest free
172     energy configuration. Therefore, it would seem unlikely that a
173     frustrated lattice in a dipolar-disordered state could exhibit the
174     long-range periodicity in the range of 100-600 \AA (as exhibited in
175     the ripple phases studied by Kaasgard {\it et
176     al.}).~\cite{Kaasgaard03}
177    
178     Is there an intermediate model between the low-density dipolar fluids
179     and the rigid lattice models which has the potential to exhibit the
180     corrugation phenomenon of the $P_{\beta'}$ phase? What we present
181     here is an attempt to find a simple dipolar model which will exhibit
182     this behavior. We are using a modified XYZ lattice model; details of
183     the model can be found in section
184     \ref{sec:model}, results of Monte Carlo simulations using this model
185     are presented in section
186     \ref{sec:results}, and section \ref{sec:discussion} contains our conclusions.
187    
188     \section{2-D Dipolar Membrane}
189     \label{sec:model}
190    
191     The point of developing this model was to arrive at the simplest
192     possible theoretical model which could exhibit spontaneous corrugation
193     of a two-dimensional dipolar medium. Since molecules in the ripple
194     phase have limited translational freedom, we have chosen a lattice to
195     support the dipoles in the x-y plane. The lattice may be either
196     hexagonal (lattice constants $a/b = \sqrt{3}$) or non-hexagonal.
197     However, each dipole has 3 degrees of freedom. They may move freely
198     {\em out} of the x-y plane (along the $z$ axis), and they have
199     complete orientational freedom ($0 <= \theta <= \pi$, $0 <= \phi < 2
200     \pi$). This is essentially a modified X-Y-Z model with translational
201     freedom along the z-axis.
202    
203     The potential energy of the system,
204     \begin{equation}
205     V = \sum_i \left( \sum_{j \in NN_i}^6
206     \frac{k_r}{2}\left( r_{ij}-\sigma \right)^2 + \sum_{j>i} \frac{|\mu|^2}{4\pi \epsilon_0 r_{ij}^3} \left[
207     {\mathbf{\hat u}_i} \cdot {\mathbf{\hat u}_j} -
208     3({\mathbf{\hat u}_i} \cdot {\mathbf{\hat
209     r}_{ij}})({\mathbf{\hat u}_j} \cdot {\mathbf{\hat r}_{ij}})\right]
210     \right)
211     \label{eq:pot}
212     \end{equation}
213    
214     In this potential, $\mathbf{\hat u}_i$ is the unit vector pointing
215     along dipole $i$ and $\mathbf{\hat r}_{ij}$ is the unit vector
216     pointing along the inter-dipole vector $\mathbf{r}_{ij}$. The entire
217     potential is governed by three parameters, the dipolar strength
218     ($\mu$), the harmonic spring constant ($k_r$) and the preferred
219     intermolecular spacing ($\sigma$). In practice, we set the value of
220     $\sigma$ to the average inter-molecular spacing from the planar
221     lattice, yielding a potential model that has only two parameters for a
222     particular choice of lattice constants $a$ (along the $x$-axis) and
223     $b$ (along the $y$-axis). We also define a set of reduced parameters
224     based on the length scale ($\sigma$) and the energy of the harmonic
225     potential at a deformation of 2 $\sigma$ ($\epsilon = k_r \sigma^2 /
226     2$). Using these two constants, we perform our calculations using
227     reduced distances, ($r^{*} = r / \sigma$), temperatures ($T^{*} = 2
228     k_B T / k_r \sigma^2$), densities ($\rho^{*} = N \sigma^2 / L_x L_y$),
229     and dipole moments ($\mu^{*} = \mu / \sqrt{4 \pi \epsilon_0 \sigma^5
230     k_r / 2}$).
231    
232     To investigate the phase behavior of this model, we have performed a
233     series of Metropolis Monte Carlo simulations of moderately-sized (34.3
234     $\sigma$ on a side) patches of membrane hosted on both hexagonal
235     ($\gamma = a/b = \sqrt{3}$) and non-hexagonal ($\gamma \neq \sqrt{3}$)
236     lattices. The linear extent of one edge of the monolayer was $20 a$
237     and the system was kept roughly square. The average distance that
238     coplanar dipoles were positioned from their six nearest neighbors was
239     1 $\sigma$ (on both hexagonal and non-hexagonal lattices). Typical
240     system sizes were 1360 dipoles for the hexagonal lattices and 840-2800
241     dipoles for the non-hexagonal lattices. Periodic boundary conditions
242     were used, and the cutoff for the dipole-dipole interaction was set to
243     4.3 $\sigma$. All parameters ($T^{*}$, $\mu^{*}$, and $\gamma$) were
244     varied systematically to study the effects of these parameters on the
245     formation of ripple-like phases.
246    
247     \section{Results and Analysis}
248     \label{sec:results}
249    
250     \subsection{Dipolar Ordering and Coexistence Temperatures}
251     The principal method for observing the orientational ordering
252     transition in dipolar systems is the $P_2$ order parameter (defined as
253     $1.5 \times \lambda_{max}$, where $\lambda_{max}$ is the largest
254     eigenvalue of the matrix,
255     \begin{equation}
256     {\mathsf{S}} = \frac{1}{N} \sum_i \left(
257     \begin{array}{ccc}
258     u^{x}_i u^{x}_i-\frac{1}{3} & u^{x}_i u^{y}_i & u^{x}_i u^{z}_i \\
259     u^{y}_i u^{x}_i & u^{y}_i u^{y}_i -\frac{1}{3} & u^{y}_i u^{z}_i \\
260     u^{z}_i u^{x}_i & u^{z}_i u^{y}_i & u^{z}_i u^{z}_i -\frac{1}{3}
261     \end{array} \right).
262     \label{eq:opmatrix}
263     \end{equation}
264     Here $u^{\alpha}_i$ is the $\alpha=x,y,z$ component of the unit vector
265     for dipole $i$. $P_2$ will be $1.0$ for a perfectly-ordered system
266     and near $0$ for a randomized system. Note that this order parameter
267     is {\em not} equal to the polarization of the system. For example,
268     the polarization of the perfect antiferroelectric system is $0$, but
269     $P_2$ for an antiferroelectric system is $1$. The eigenvector of
270     $\mathsf{S}$ corresponding to the largest eigenvalue is familiar as
271     the director axis, which can be used to determine a privileged dipolar
272     axis for dipole-ordered systems. The top panel in Fig. \ref{phase}
273     shows the values of $P_2$ as a function of temperature for both
274     hexagonal ($\gamma = 1.732$) and non-hexagonal ($\gamma=1.875$)
275     lattices.
276    
277     \begin{figure}[ht]
278     \centering
279     \caption{Top panel: The $P_2$ dipolar order parameter as a function of
280     temperature for both hexagonal ($\gamma = 1.732$) and non-hexagonal
281     ($\gamma = 1.875$) lattices. Bottom Panel: The phase diagram for the
282     dipolar membrane model. The line denotes the division between the
283     dipolar ordered (antiferroelectric) and disordered phases. An
284     enlarged view near the hexagonal lattice is shown inset.}
285     \includegraphics[width=\linewidth]{phase.pdf}
286     \label{phase}
287     \end{figure}
288    
289     There is a clear order-disorder transition in evidence from this data.
290     Both the hexagonal and non-hexagonal lattices have dipolar-ordered
291     low-temperature phases, and orientationally-disordered high
292     temperature phases. The coexistence temperature for the hexagonal
293     lattice is significantly lower than for the non-hexagonal lattices,
294     and the bulk polarization is approximately $0$ for both dipolar
295     ordered and disordered phases. This gives strong evidence that the
296     dipolar ordered phase is antiferroelectric. We have repeated the
297     Monte Carlo simulations over a wide range of lattice ratios ($\gamma$)
298     to generate a dipolar order/disorder phase diagram. The bottom panel
299     in Fig. \ref{phase} shows that the hexagonal lattice is a
300     low-temperature cusp in the $T^{*}-\gamma$ phase diagram.
301    
302     This phase diagram is remarkable in that it shows an antiferroelectric
303     phase near $\gamma=1.732$ where one would expect lattice frustration
304     to result in disordered phases at all temperatures. Observations of
305     the configurations in this phase show clearly that the system has
306     accomplished dipolar orderering by forming large ripple-like
307     structures. We have observed antiferroelectric ordering in all three
308     of the equivalent directions on the hexagonal lattice, and the dipoles
309     have been observed to organize perpendicular to the membrane normal
310     (in the plane of the membrane). It is particularly interesting to
311     note that the ripple-like structures have also been observed to
312     propagate in the three equivalent directions on the lattice, but the
313     {\em direction of ripple propagation is always perpendicular to the
314     dipole director axis}. A snapshot of a typical antiferroelectric
315     rippled structure is shown in Fig. \ref{fig:snapshot}.
316    
317     \begin{figure}[ht]
318     \centering
319     \caption{Top and Side views of a representative configuration for the
320     dipolar ordered phase supported on the hexagonal lattice. Note the
321     antiferroelectric ordering and the long wavelength buckling of the
322     membrane. Dipolar ordering has been observed in all three equivalent
323     directions on the hexagonal lattice, and the ripple direction is
324     always perpendicular to the director axis for the dipoles.}
325     \includegraphics[width=5.5in]{snapshot.pdf}
326     \label{fig:snapshot}
327     \end{figure}
328    
329     \subsection{Discriminating Ripples from Thermal Undulations}
330    
331     In order to be sure that the structures we have observed are actually
332     a rippled phase and not simply thermal undulations, we have computed
333     the undulation spectrum,
334     \begin{equation}
335     h(\vec{q}) = A^{-1/2} \int d\vec{r}
336     h(\vec{r}) e^{-i \vec{q}\cdot\vec{r}}
337     \end{equation}
338     where $h(\vec{r})$ is the height of the membrane at location $\vec{r}
339     = (x,y)$.~\cite{Safran94} In simple (and more complicated) elastic
340     continuum models, Brannigan {\it et al.} have shown that in the $NVT$
341     ensemble, the absolute value of the undulation spectrum can be
342     written,
343     \begin{equation}
344     \langle | h(q)|^2 \rangle_{NVT} = \frac{k_B T}{k_c |\vec{q}|^4 +
345     \tilde{\gamma}|\vec{q}|^2},
346     \label{eq:fit}
347     \end{equation}
348     where $k_c$ is the bending modulus for the membrane, and
349     $\tilde{\gamma}$ is the mechanical surface
350     tension.~\cite{Brannigan04b}
351    
352     The undulation spectrum is computed by superimposing a rectangular
353     grid on top of the membrane, and by assigning height ($h(\vec{r})$)
354     values to the grid from the average of all dipoles that fall within a
355     given $\vec{r}+d\vec{r}$ grid area. Empty grid pixels are assigned
356     height values by interpolation from the nearest neighbor pixels. A
357     standard 2-d Fourier transform is then used to obtain $\langle |
358     h(q)|^2 \rangle$.
359    
360     The systems studied in this paper have relatively small bending moduli
361     ($k_c$) and relatively large mechanical surface tensions
362     ($\tilde{\gamma}$). In practice, the best fits to our undulation
363     spectra are obtained by approximating the value of $k_c$ to 0. In
364     Fig. \ref{fig:fit} we show typical undulation spectra for two
365     different regions of the phase diagram along with their fits from the
366     Landau free energy approach (Eq. \ref{eq:fit}). In the
367     high-temperature disordered phase, the Landau fits can be nearly
368     perfect, and from these fits we can estimate the bending modulus and
369     the mechanical surface tension.
370    
371     For the dipolar-ordered hexagonal lattice near the coexistence
372     temperature, however, we observe long wavelength undulations that are
373     far outliers to the fits. That is, the Landau free energy fits are
374     well within error bars for all other points, but can be off by {\em
375     orders of magnitude} for a few low frequency components.
376    
377     We interpret these outliers as evidence that these low frequency modes
378     are {\em non-thermal undulations}. We take this as evidence that we
379     are actually seeing a rippled phase developing in this model system.
380    
381     \begin{figure}[ht]
382     \centering
383     \caption{Evidence that the observed ripples are {\em not} thermal
384     undulations is obtained from the 2-d fourier transform $\langle
385     |h^{*}(\vec{q})|^2 \rangle$ of the height profile ($\langle h^{*}(x,y)
386     \rangle$). Rippled samples show low-wavelength peaks that are
387     outliers on the Landau free energy fits. Samples exhibiting only
388     thermal undulations fit Eq. \ref{eq:fit} remarkably well.}
389     \includegraphics[width=5.5in]{fit.pdf}
390     \label{fig:fit}
391     \end{figure}
392    
393     \subsection{Effects of Potential Parameters on Amplitude and Wavelength}
394    
395     We have used two different methods to estimate the amplitude and
396     periodicity of the ripples. The first method requires projection of
397     the ripples onto a one dimensional rippling axis. Since the rippling
398     is always perpendicular to the dipole director axis, we can define a
399     ripple vector as follows. The largest eigenvector, $s_1$, of the
400     $\mathsf{S}$ matrix in Eq. \ref{eq:opmatrix} is projected onto a
401     planar director axis,
402     \begin{equation}
403     \vec{d} = \left(\begin{array}{c}
404     \vec{s}_1 \cdot \hat{i} \\
405     \vec{s}_1 \cdot \hat{j} \\
406     0
407     \end{array} \right).
408     \end{equation}
409     ($\hat{i}$, $\hat{j}$ and $\hat{k}$ are unit vectors along the $x$,
410     $y$, and $z$ axes, respectively.) The rippling axis is in the plane of
411     the membrane and is perpendicular to the planar director axis,
412     \begin{equation}
413     \vec{q}_{\mathrm{rip}} = \vec{d} \times \hat{k}
414     \end{equation}
415     We can then find the height profile of the membrane along the ripple
416     axis by projecting heights of the dipoles to obtain a one-dimensional
417     height profile, $h(q_{\mathrm{rip}})$. Ripple wavelengths can be
418     estimated from the largest non-thermal low-frequency component in the
419     fourier transform of $h(q_{\mathrm{rip}})$. Amplitudes can be
420     estimated by measuring peak-to-trough distances in
421     $h(q_{\mathrm{rip}})$ itself.
422    
423     A second, more accurate, and simpler method for estimating ripple
424     shape is to extract the wavelength and height information directly
425     from the largest non-thermal peak in the undulation spectrum. For
426     large-amplitude ripples, the two methods give similar results. The
427     one-dimensional projection method is more prone to noise (particularly
428     in the amplitude estimates for the non-hexagonal lattices). We report
429     amplitudes and wavelengths taken directly from the undulation spectrum
430     below.
431    
432     In the hexagonal lattice ($\gamma = \sqrt{3}$), the rippling is
433     observed for temperatures ($T^{*}$) from $61-122$. The wavelength of
434     the ripples is remarkably stable at 21.4~$\sigma$ for all but the
435     temperatures closest to the order-disorder transition. At $T^{*} =
436     122$, the wavelength drops to 17.1~$\sigma$.
437    
438     The dependence of the amplitude on temperature is shown in the top
439     panel of Fig. \ref{fig:Amplitude}. The rippled structures shrink
440     smoothly as the temperature rises towards the order-disorder
441     transition. The wavelengths and amplitudes we observe are
442     surprisingly close to the $\Lambda / 2$ phase observed by Kaasgaard
443     {\it et al.} in their work on PC-based lipids.\cite{Kaasgaard03}
444     However, this is coincidental agreement based on a choice of 7~\AA~as
445     the mean spacing between lipids.
446    
447     \begin{figure}[ht]
448     \centering
449     \caption{a) The amplitude $A^{*}$ of the ripples vs. temperature for a
450     hexagonal lattice. b) The amplitude $A^{*}$ of the ripples vs. dipole
451     strength ($\mu^{*}$) for both the hexagonal lattice (circles) and
452     non-hexagonal lattice (squares). The reduced temperatures were kept
453     fixed at $T^{*} = 94$ for the hexagonal lattice and $T^{*} = 106$ for
454     the non-hexagonal lattice (approximately 2/3 of the order-disorder
455     transition temperature for each lattice).}
456     \includegraphics[width=\linewidth]{properties_sq.pdf}
457     \label{fig:Amplitude}
458     \end{figure}
459    
460     The ripples can be made to disappear by increasing the internal
461     surface tension (i.e. by increasing $k_r$ or equivalently, reducing
462     the dipole moment). The amplitude of the ripples depends critically
463     on the strength of the dipole moments ($\mu^{*}$) in Eq. \ref{eq:pot}.
464     If the dipoles are weakened substantially (below $\mu^{*}$ = 20) at a
465     fixed temperature of 94, the membrane loses dipolar ordering
466     and the ripple structures. The ripples reach a peak amplitude of
467     3.7~$\sigma$ at a dipolar strength of 25. We show the dependence
468     of ripple amplitude on the dipolar strength in
469     Fig. \ref{fig:Amplitude}.
470    
471     \subsection{Non-hexagonal lattices}
472    
473     We have also investigated the effect of the lattice geometry by
474     changing the ratio of lattice constants ($\gamma$) while keeping the
475     average nearest-neighbor spacing constant. The antiferroelectric state
476     is accessible for all $\gamma$ values we have used, although the
477     distorted hexagonal lattices prefer a particular director axis due to
478     the anisotropy of the lattice.
479    
480     Our observation of rippling behavior was not limited to the hexagonal
481     lattices. In non-hexagonal lattices the antiferroelectric phase can
482     develop nearly instantaneously in the Monte Carlo simulations, and
483     these dipolar-ordered phases tend to be remarkably flat. Whenever
484     rippling has been observed in these non-hexagonal lattices
485     (e.g. $\gamma = 1.875$), we see relatively short ripple wavelengths
486     (14 $\sigma$) and amplitudes of 2.4~$\sigma$. These ripples are
487     weakly dependent on dipolar strength (see Fig. \ref{fig:Amplitude}),
488     although below a dipolar strength of $\mu^{*} = 20$, the membrane
489     loses dipolar ordering and displays only thermal undulations.
490    
491     The ripple phase does {\em not} appear at all values of $\gamma$. We
492     have only observed non-thermal undulations in the range $1.625 <
493     \gamma < 1.875$. Outside this range, the order-disorder transition in
494     the dipoles remains, but the ordered dipolar phase has only thermal
495     undulations. This is one of our strongest pieces of evidence that
496     rippling is a symmetry-breaking phenomenon for hexagonal and
497     nearly-hexagonal lattices.
498    
499     \subsection{Effects of System Size}
500     To evaluate the effect of finite system size, we have performed a
501     series of simulations on the hexagonal lattice at a reduced
502     temperature of 122, which is just below the order-disorder transition
503     temperature ($T^{*} = 139$). These conditions are in the
504     dipole-ordered and rippled portion of the phase diagram. These are
505     also the conditions that should be most susceptible to system size
506     effects.
507    
508     \begin{figure}[ht]
509     \centering
510     \caption{The ripple wavelength (top) and amplitude (bottom) as a
511     function of system size for a hexagonal lattice ($\gamma=1.732$) at $T^{*} =
512     122$.}
513     \includegraphics[width=\linewidth]{SystemSize.pdf}
514     \label{fig:systemsize}
515     \end{figure}
516    
517     There is substantial dependence on system size for small (less than
518     29~$\sigma$) periodic boxes. Notably, there are resonances apparent
519     in the ripple amplitudes at box lengths of 17.3 and 29.5 $\sigma$.
520     For larger systems, the behavior of the ripples appears to have
521     stabilized and is on a trend to slightly smaller amplitudes (and
522     slightly longer wavelengths) than were observed from the 34.3 $\sigma$
523     box sizes that were used for most of the calculations.
524    
525     It is interesting to note that system sizes which are multiples of the
526     default ripple wavelength can enhance the amplitude of the observed
527     ripples, but appears to have only a minor effect on the observed
528     wavelength. It would, of course, be better to use system sizes that
529     were many multiples of the ripple wavelength to be sure that the
530     periodic box is not driving the phenomenon, but at the largest system
531     size studied (70 $\sigma$ $\times$ 70 $\sigma$), the number of dipoles
532     (5440) made long Monte Carlo simulations prohibitively expensive.
533    
534     \section{Discussion}
535     \label{sec:discussion}
536    
537     We have been able to show that a simple dipolar lattice model which
538     contains only molecular packing (from the lattice), anisotropy (in the
539     form of electrostatic dipoles) and a weak surface tension (in the form
540     of a nearest-neighbor harmonic potential) is capable of exhibiting
541     stable long-wavelength non-thermal surface corrugations. The best
542     explanation for this behavior is that the ability of the dipoles to
543     translate out of the plane of the membrane is enough to break the
544     symmetry of the hexagonal lattice and allow the energetic benefit from
545     the formation of a bulk antiferroelectric phase. Were the weak
546     surface tension absent from our model, it would be possible for the
547     entire lattice to ``tilt'' using $z$-translation. Tilting the lattice
548     in this way would yield an effectively non-hexagonal lattice which
549     would avoid dipolar frustration altogether. With the surface tension
550     in place, bulk tilt causes a large strain, and the simplest way to
551     release this strain is along line defects. Line defects will result
552     in rippled or sawtooth patterns in the membrane, and allow small
553     ``stripes'' of membrane to form antiferroelectric regions that are
554     tilted relative to the averaged membrane normal.
555    
556     Although the dipole-dipole interaction is the major driving force for
557     the long range orientational ordered state, the formation of the
558     stable, smooth ripples is a result of the competition between the
559     surface tension and the dipole-dipole interactions. This statement is
560     supported by the variation in $\mu^{*}$. Substantially weaker dipoles
561     relative to the surface tension can cause the corrugated phase to
562     disappear.
563    
564     The packing of the dipoles into a nearly-hexagonal lattice is clearly
565     an important piece of the puzzle. The dipolar head groups of lipid
566     molecules are sterically (as well as electrostatically) anisotropic,
567     and would not be able to pack hexagonally without the steric
568     interference of adjacent molecular bodies. Since we only see rippled
569     phases in the neighborhood of $\gamma=\sqrt{3}$, this implies that
570     there is a role played by the lipid chains in the organization of the
571     hexagonally ordered phases which support ripples in realistic lipid
572     bilayers.
573    
574     The most important prediction we can make using the results from this
575     simple model is that if dipolar ordering is driving the surface
576     corrugation, the wave vectors for the ripples should always found to
577     be {\it perpendicular} to the dipole director axis. This prediction
578     should suggest experimental designs which test whether this is really
579     true in the phosphatidylcholine $P_{\beta'}$ phases. The dipole
580     director axis should also be easily computable for the all-atom and
581     coarse-grained simulations that have been published in the literature.
582    
583     Our other observation about the ripple and dipolar directionality is
584     that the dipole director axis can be found to be parallel to any of
585     the three equivalent lattice vectors in the hexagonal lattice.
586     Defects in the ordering of the dipoles can cause the dipole director
587     (and consequently the surface corrugation) of small regions to be
588     rotated relative to each other by 120$^{\circ}$. This is a similar
589     behavior to the domain rotation seen in the AFM studies of Kaasgaard
590     {\it et al.}\cite{Kaasgaard03}
591    
592     Although our model is simple, it exhibits some rich and unexpected
593     behaviors. It would clearly be a closer approximation to the reality
594     if we allowed greater translational freedom to the dipoles and
595     replaced the somewhat artificial lattice packing and the harmonic
596     ``surface tension'' with more realistic molecular modeling
597     potentials. What we have done is to present an extremely simple model
598     which exhibits bulk non-thermal corrugation, and our explanation of
599     this rippling phenomenon will help us design more accurate molecular
600     models for corrugated membranes and experiments to test whether
601     rippling is dipole-driven or not.
602     \clearpage
603     \bibliography{ripple}
604     \printfigures
605     \end{document}