ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/group/trunk/ssdePaper/nptSSD.tex
Revision: 743
Committed: Wed Sep 3 21:36:09 2003 UTC (20 years, 10 months ago) by chrisfen
Content type: application/x-tex
File size: 48323 byte(s)
Log Message:
This is the paper that got scooped

File Contents

# Content
1 \documentclass[prb,aps,times,twocolumn,tabularx]{revtex4}
2 %\documentclass[prb,aps,times,tabularx,preprint]{revtex4}
3 \usepackage{amsmath}
4 \usepackage{graphicx}
5
6 %\usepackage{endfloat}
7 %\usepackage{berkeley}
8 %\usepackage{epsf}
9 %\usepackage[ref]{overcite}
10 %\usepackage{setspace}
11 %\usepackage{tabularx}
12 %\usepackage{graphicx}
13 %\usepackage{curves}
14 %\usepackage{amsmath}
15 %\pagestyle{plain}
16 %\pagenumbering{arabic}
17 %\oddsidemargin 0.0cm \evensidemargin 0.0cm
18 %\topmargin -21pt \headsep 10pt
19 %\textheight 9.0in \textwidth 6.5in
20 %\brokenpenalty=10000
21 %\renewcommand{\baselinestretch}{1.2}
22 %\renewcommand\citemid{\ } % no comma in optional reference note
23
24 \begin{document}
25
26 \title{On the temperature dependent structural and transport properties of the soft sticky dipole (SSD) and related single point water models}
27
28 \author{Christopher J. Fennell and J. Daniel Gezelter{\thefootnote}
29 \footnote[1]{Corresponding author. \ Electronic mail: gezelter@nd.edu}}
30
31 \address{Department of Chemistry and Biochemistry\\ University of Notre Dame\\
32 Notre Dame, Indiana 46556}
33
34 \date{\today}
35
36 \begin{abstract}
37 NVE and NPT molecular dynamics simulations were performed in order to
38 investigate the density maximum and temperature dependent transport
39 for the SSD water model, both with and without the use of reaction
40 field. The constant pressure simulations of the melting of both $I_h$
41 and $I_c$ ice showed a density maximum near 260 K. In most cases, the
42 calculated densities were significantly lower than the densities
43 calculated in simulations of other water models. Analysis of particle
44 diffusion showed SSD to capture the transport properties of
45 experimental very well in both the normal and super-cooled liquid
46 regimes. In order to correct the density behavior, SSD was
47 reparameterized for use both with and without a long-range interaction
48 correction, SSD/RF and SSD/E respectively. In addition to correcting
49 the abnormally low densities, these new versions were show to maintain
50 or improve upon the transport and structural features of the original
51 water model.
52 \end{abstract}
53
54 \maketitle
55
56 %\narrowtext
57
58
59 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
60 % BODY OF TEXT
61 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
62
63 \section{Introduction}
64
65 One of the most important tasks in simulations of biochemical systems
66 is the proper depiction of water and water solvation. In fact, the
67 bulk of the calculations performed in solvated simulations are of
68 interactions with or between solvent molecules. Thus, the outcomes of
69 these types of simulations are highly dependent on the physical
70 properties of water, both as individual molecules and in
71 groups/bulk. Due to the fact that explicit solvent accounts for a
72 massive portion of the calculations, it necessary to simplify the
73 solvent to some extent in order to complete simulations in a
74 reasonable amount of time. In the case of simulating water in
75 bio-molecular studies, the balance between accurate properties and
76 computational efficiency is especially delicate, and it has resulted
77 in a variety of different water
78 models.\cite{Jorgensen83,Berendsen87,Jorgensen00} Many of these models
79 get specific properties correct or better than their predecessors, but
80 this is often at a cost of some other properties or of computer
81 time. As an example, compare TIP3P or TIP4P to TIP5P. TIP5P succeeds
82 in improving the structural and transport properties over its
83 predecessors, yet this comes at a greater than 50\% increase in
84 computational cost.\cite{Jorgensen01,Jorgensen00} One recently
85 developed model that succeeds in both retaining accuracy of system
86 properties and simplifying calculations to increase computational
87 efficiency is the Soft Sticky Dipole water model.\cite{Ichiye96}
88
89 The Soft Sticky Dipole (SSD)\ water model was developed by Ichiye
90 \emph{et al.} as a modified form of the hard-sphere water model
91 proposed by Bratko, Blum, and Luzar.\cite{Bratko85,Bratko95} SSD
92 consists of a single point dipole with a Lennard-Jones core and a
93 sticky potential that directs the particles to assume the proper
94 hydrogen bond orientation in the first solvation shell. Thus, the
95 interaction between two SSD water molecules \emph{i} and \emph{j} is
96 given by the potential
97 \begin{equation}
98 u_{ij} = u_{ij}^{LJ} (r_{ij})\ + u_{ij}^{dp}
99 (\mathbf{r}_{ij},\boldsymbol{\Omega}_i,\boldsymbol{\Omega}_j)\ +
100 u_{ij}^{sp}
101 (\mathbf{r}_{ij},\boldsymbol{\Omega}_i,\boldsymbol{\Omega}_j),
102 \end{equation}
103 where the $\mathbf{r}_{ij}$ is the position vector between molecules
104 \emph{i} and \emph{j} with magnitude equal to the distance $r_ij$, and
105 $\boldsymbol{\Omega}_i$ and $\boldsymbol{\Omega}_j$ represent the
106 orientations of the respective molecules. The Lennard-Jones, dipole,
107 and sticky parts of the potential are giving by the following
108 equations,
109 \begin{equation}
110 u_{ij}^{LJ}(r_{ij}) = 4\epsilon \left[\left(\frac{\sigma}{r_{ij}}\right)^{12}-\left(\frac{\sigma}{r_{ij}}\right)^{6}\right],
111 \end{equation}
112 \begin{equation}
113 u_{ij}^{dp} = \frac{\boldsymbol{\mu}_i\cdot\boldsymbol{\mu}_j}{r_{ij}^3}-\frac{3(\boldsymbol{\mu}_i\cdot\mathbf{r}_{ij})(\boldsymbol{\mu}_j\cdot\mathbf{r}_{ij})}{r_{ij}^5}\ ,
114 \end{equation}
115 \begin{equation}
116 \begin{split}
117 u_{ij}^{sp}
118 (\mathbf{r}_{ij},\boldsymbol{\Omega}_i,\boldsymbol{\Omega}_j)
119 &=
120 \frac{\nu_0}{2}[s(r_{ij})w(\mathbf{r}_{ij},\boldsymbol{\Omega}_i,\boldsymbol{\Omega}_j)\\
121 & \quad \ +
122 s^\prime(r_{ij})w^\prime(\mathbf{r}_{ij},\boldsymbol{\Omega}_i,\boldsymbol{\Omega}_j)]\ ,
123 \end{split}
124 \end{equation}
125 where $\boldsymbol{\mu}_i$ and $\boldsymbol{\mu}_j$ are the dipole
126 unit vectors of particles \emph{i} and \emph{j} with magnitude 2.35 D,
127 $\nu_0$ scales the strength of the overall sticky potential, $s$ and
128 $s^\prime$ are cubic switching functions. The $w$ and $w^\prime$
129 functions take the following forms,
130 \begin{equation}
131 w(\mathbf{r}_{ij},\boldsymbol{\Omega}_i,\boldsymbol{\Omega}_j)=\sin\theta_{ij}\sin2\theta_{ij}\cos2\phi_{ij},
132 \end{equation}
133 \begin{equation}
134 w^\prime(\mathbf{r}_{ij},\boldsymbol{\Omega}_i,\boldsymbol{\Omega}_j) = (\cos\theta_{ij}-0.6)^2(\cos\theta_{ij}+0.8)^2-w^0,
135 \end{equation}
136 where $w^0 = 0.07715$. The $w$ function is the tetrahedral attractive
137 term that promotes hydrogen bonding orientations within the first
138 solvation shell, and $w^\prime$ is a dipolar repulsion term that
139 repels unrealistic dipolar arrangements within the first solvation
140 shell. A more detailed description of the functional parts and
141 variables in this potential can be found in other
142 articles.\cite{Ichiye96,Ichiye99}
143
144 Being that this is a one-site point dipole model, the actual force
145 calculations are simplified significantly. In the original Monte Carlo
146 simulations using this model, Ichiye \emph{et al.} reported a
147 calculation speed up of up to an order of magnitude over other
148 comparable models while maintaining the structural behavior of
149 water.\cite{Ichiye96} In the original molecular dynamics studies of
150 SSD, it was shown that it actually improves upon the prediction of
151 water's dynamical properties 3 and 4-point models.\cite{Ichiye99} This
152 attractive combination of speed and accurate depiction of solvent
153 properties makes SSD a model of interest for the simulation of large
154 scale biological systems, such as membrane phase behavior, a specific
155 interest within our group.
156
157 Up to this point, a detailed look at the model's structure and ion
158 solvation abilities has been performed.\cite{Ichiye96} In addition, a
159 thorough investigation of the dynamic properties of SSD was performed
160 by Chandra and Ichiye focusing on translational and orientational
161 properties at 298 K.\cite{Ichiye99} This study focuses on determining
162 the density maximum for SSD utilizing both microcanonical and
163 isobaric-isothermal ensemble molecular dynamics, while using the
164 reaction field method for handling long-ranged dipolar interactions. A
165 reaction field method has been previously implemented in Monte Carlo
166 simulations by Liu and Ichiye in order to study the static dielectric
167 constant for the model.\cite{Ichiye96b} This paper will expand the
168 scope of these original simulations to look on how the reaction field
169 affects the physical and dynamic properties of SSD systems.
170
171 \section{Methods}
172
173 As stated previously, in this study the long-range dipole-dipole
174 interactions were accounted for using the reaction field method. The
175 magnitude of the reaction field acting on dipole \emph{i} is given by
176 \begin{equation}
177 \mathcal{E}_{i} = \frac{2(\varepsilon_{s} - 1)}{2\varepsilon_{s} + 1}
178 \frac{1}{r_{c}^{3}} \sum_{j\in{\mathcal{R}}} \boldsymbol{\mu}_{j} f(r_{ij})\ ,
179 \label{rfequation}
180 \end{equation}
181 where $\mathcal{R}$ is the cavity defined by the cutoff radius
182 ($r_{c}$), $\varepsilon_{s}$ is the dielectric constant imposed on the
183 system (80 in this case), $\boldsymbol{\mu}_{j}$ is the dipole moment
184 vector of particle \emph{j}, and $f(r_{ij})$ is a cubic switching
185 function.\cite{AllenTildesley} The reaction field contribution to the
186 total energy by particle \emph{i} is given by
187 $-\frac{1}{2}\boldsymbol{\mu}_{i}\cdot\mathcal{E}_{i}$ and the torque
188 on dipole \emph{i} by
189 $\boldsymbol{\mu}_{i}\times\mathcal{E}_{i}$.\cite{AllenTildesley} Use
190 of reaction field is known to alter the orientational dynamic
191 properties, such as the dielectric relaxation time, based on changes
192 in the length of the cutoff radius.\cite{Berendsen98} This variable
193 behavior makes reaction field a less attractive method than other
194 methods, like the Ewald summation; however, for the simulation of
195 large-scale system, the computational cost benefit of reaction field
196 is dramatic. To address some of the dynamical property alterations due
197 to the use of reaction field, simulations were also performed without
198 a surrounding dielectric and suggestions are proposed on how to make
199 SSD more compatible with a reaction field.
200
201 Simulations were performed in both the isobaric-isothermal and
202 microcanonical ensembles. The constant pressure simulations were
203 implemented using an integral thermostat and barostat as outlined by
204 Hoover.\cite{Hoover85,Hoover86} For the constant pressure
205 simulations, the \emph{Q} parameter for the was set to 5.0 amu
206 \(\cdot\)\AA\(^{2}\), and the relaxation time (\(\tau\))\ was set at
207 100 ps.
208
209 Integration of the equations of motion was carried out using the
210 symplectic splitting method proposed by Dullweber \emph{et
211 al.}.\cite{Dullweber1997} The reason for this integrator selection
212 deals with poor energy conservation of rigid body systems using
213 quaternions. While quaternions work well for orientational motion in
214 alternate ensembles, the microcanonical ensemble has a constant energy
215 requirement that is actually quite sensitive to errors in the
216 equations of motion. The original implementation of this code utilized
217 quaternions for rotational motion propagation; however, a detailed
218 investigation showed that they resulted in a steady drift in the total
219 energy, something that has been observed by others.\cite{Laird97}
220
221 The key difference in the integration method proposed by Dullweber
222 \emph{et al.} is that the entire rotation matrix is propagated from
223 one time step to the next. In the past, this would not have been as
224 feasible a option, being that the rotation matrix for a single body is
225 nine elements long as opposed to 3 or 4 elements for Euler angles and
226 quaternions respectively. System memory has become much less of an
227 issue in recent times, and this has resulted in substantial benefits
228 in energy conservation. There is still the issue of an additional 5 or
229 6 additional elements for describing the orientation of each particle,
230 which will increase dump files substantially. Simply translating the
231 rotation matrix into its component Euler angles or quaternions for
232 storage purposes relieves this burden.
233
234 The symplectic splitting method allows for Verlet style integration of
235 both linear and angular motion of rigid bodies. In the integration
236 method, the orientational propagation involves a sequence of matrix
237 evaluations to update the rotation matrix.\cite{Dullweber1997} These
238 matrix rotations end up being more costly computationally than the
239 simpler arithmetic quaternion propagation. On average, a 1000 SSD
240 particle simulation shows a 7\% increase in simulation time using the
241 symplectic step method in place of quaternions. This cost is more than
242 justified when comparing the energy conservation of the two methods as
243 illustrated in figure \ref{timestep}.
244
245 \begin{figure}
246 \includegraphics[width=61mm, angle=-90]{timeStep.epsi}
247 \caption{Energy conservation using quaternion based integration versus
248 the symplectic step method proposed by Dullweber \emph{et al.} with
249 increasing time step. For each time step, the dotted line is total
250 energy using the symplectic step integrator, and the solid line comes
251 from the quaternion integrator. The larger time step plots are shifted
252 up from the true energy baseline for clarity.}
253 \label{timestep}
254 \end{figure}
255
256 In figure \ref{timestep}, the resulting energy drift at various time
257 steps for both the symplectic step and quaternion integration schemes
258 is compared. All of the 1000 SSD particle simulations started with the
259 same configuration, and the only difference was the method for
260 handling rotational motion. At time steps of 0.1 and 0.5 fs, both
261 methods for propagating particle rotation conserve energy fairly well,
262 with the quaternion method showing a slight energy drift over time in
263 the 0.5 fs time step simulation. At time steps of 1 and 2 fs, the
264 energy conservation benefits of the symplectic step method are clearly
265 demonstrated.
266
267 Energy drift in these SSD particle simulations was unnoticeable for
268 time steps up to three femtoseconds. A slight energy drift on the
269 order of 0.012 kcal/mol per nanosecond was observed at a time step of
270 four femtoseconds, and as expected, this drift increases dramatically
271 with increasing time step. To insure accuracy in the constant energy
272 simulations, time steps were set at 2 fs and kept at this value for
273 constant pressure simulations as well.
274
275 Ice crystals in both the $I_h$ and $I_c$ lattices were generated as
276 starting points for all the simulations. The $I_h$ crystals were
277 formed by first arranging the center of masses of the SSD particles
278 into a ``hexagonal'' ice lattice of 1024 particles. Because of the
279 crystal structure of $I_h$ ice, the simulation box assumed a
280 rectangular shape with a edge length ratio of approximately
281 1.00$\times$1.06$\times$1.23. The particles were then allowed to
282 orient freely about fixed positions with angular momenta randomized at
283 400 K for varying times. The rotational temperature was then scaled
284 down in stages to slowly cool the crystals down to 25 K. The particles
285 were then allowed translate with fixed orientations at a constant
286 pressure of 1 atm for 50 ps at 25 K. Finally, all constraints were
287 removed and the ice crystals were allowed to equilibrate for 50 ps at
288 25 K and a constant pressure of 1 atm. This procedure resulted in
289 structurally stable $I_h$ ice crystals that obey the Bernal-Fowler
290 rules\cite{Bernal33,Rahman72}. This method was also utilized in the
291 making of diamond lattice $I_c$ ice crystals, with each cubic
292 simulation box consisting of either 512 or 1000 particles. Only
293 isotropic volume fluctuations were performed under constant pressure,
294 so the ratio of edge lengths remained constant throughout the
295 simulations.
296
297 \section{Results and discussion}
298
299 Melting studies were performed on the randomized ice crystals using
300 constant pressure and temperature dynamics. This involved an initial
301 randomization of velocities about the starting temperature of 25 K for
302 varying amounts of time. The systems were all equilibrated for 100 ps
303 prior to a 200 ps data collection run at each temperature setting,
304 ranging from 25 to 400 K, with a maximum degree increment of 25 K. For
305 regions of interest along this stepwise progression, the temperature
306 increment was decreased from 25 K to 10 and then 5 K. The above
307 equilibration and production times were sufficient in that the system
308 volume fluctuations dampened out in all but the very cold simulations
309 (below 225 K). In order to further improve statistics, five separate
310 simulation progressions were performed, and the averaged results from
311 the $I_h$ melting simulations are shown in figure \ref{dense1}.
312
313 \begin{figure}
314 \includegraphics[width=65mm, angle=-90]{1hdense.epsi}
315 \caption{Average density of SSD water at increasing temperatures
316 starting from ice $I_h$ lattice.}
317 \label{dense1}
318 \end{figure}
319
320 \subsection{Density Behavior}
321 In the initial average density versus temperature plot, the density
322 maximum clearly appears between 255 and 265 K. The calculated
323 densities within this range were nearly indistinguishable, as can be
324 seen in the zoom of this region of interest, shown in figure
325 \ref{dense1}. The greater certainty of the average value at 260 K makes
326 a good argument for the actual density maximum residing at this
327 midpoint value. Figure \ref{dense1} was constructed using ice $I_h$
328 crystals for the initial configuration; and though not pictured, the
329 simulations starting from ice $I_c$ crystal configurations showed
330 similar results, with a liquid-phase density maximum in this same
331 region (between 255 and 260 K). In addition, the $I_c$ crystals are
332 more fragile than the $I_h$ crystals, leading them to deform into a
333 dense glassy state at lower temperatures. This resulted in an overall
334 low temperature density maximum at 200 K, but they still retained a
335 common liquid state density maximum with the $I_h$ simulations.
336
337 \begin{figure}
338 \includegraphics[width=65mm,angle=-90]{dense2.eps}
339 \caption{Density versus temperature for TIP4P\cite{Jorgensen98b},
340 TIP3P\cite{Jorgensen98b}, SPC/E\cite{Clancy94}, SSD without Reaction
341 Field, SSD, and Experiment\cite{CRC80}. }
342 \label{dense2}
343 \end{figure}
344
345 The density maximum for SSD actually compares quite favorably to other
346 simple water models. Figure \ref{dense2} shows a plot of these
347 findings with the density progression of several other models and
348 experiment obtained from other
349 sources.\cite{Jorgensen98b,Clancy94,CRC80} Of the listed simple water
350 models, SSD has results closest to the experimentally observed water
351 density maximum. Of the listed water models, TIP4P has a density
352 maximum behavior most like that seen in SSD. Though not shown, it is
353 useful to note that TIP5P has a water density maximum nearly identical
354 to experiment.
355
356 Possibly of more importance is the density scaling of SSD relative to
357 other common models at any given temperature (Fig. \ref{dense2}). Note
358 that the SSD model assumes a lower density than any of the other
359 listed models at the same pressure, behavior which is especially
360 apparent at temperatures greater than 300 K. Lower than expected
361 densities have been observed for other systems with the use of a
362 reaction field for long-range electrostatic interactions, so the most
363 likely reason for these significantly lower densities in these
364 simulations is the presence of the reaction field.\cite{Berendsen98}
365 In order to test the effect of the reaction field on the density of
366 the systems, the simulations were repeated for the temperature region
367 of interest without a reaction field present. The results of these
368 simulations are also displayed in figure \ref{dense2}. Without
369 reaction field, these densities increase considerably to more
370 experimentally reasonable values, especially around the freezing point
371 of liquid water. The shape of the curve is similar to the curve
372 produced from SSD simulations using reaction field, specifically the
373 rapidly decreasing densities at higher temperatures; however, a slight
374 shift in the density maximum location, down to 245 K, is
375 observed. This is probably a more accurate comparison to the other
376 listed water models in that no long range corrections were applied in
377 those simulations.\cite{Clancy94,Jorgensen98b}
378
379 It has been observed that densities are dependent on the cutoff radius
380 used for a variety of water models in simulations both with and
381 without the use of reaction field.\cite{Berendsen98} In order to
382 address the possible affect of cutoff radius, simulations were
383 performed with a dipolar cutoff radius of 12.0 \AA\ to compliment the
384 previous SSD simulations, all performed with a cutoff of 9.0 \AA. All
385 the resulting densities overlapped within error and showed no
386 significant trend in lower or higher densities as a function of cutoff
387 radius, both for simulations with and without reaction field. These
388 results indicate that there is no major benefit in choosing a longer
389 cutoff radius in simulations using SSD. This is comforting in that the
390 use of a longer cutoff radius results in a near doubling of the time
391 required to compute a single trajectory.
392
393 \subsection{Transport Behavior}
394 Of importance in these types of studies are the transport properties
395 of the particles and how they change when altering the environmental
396 conditions. In order to probe transport, constant energy simulations
397 were performed about the average density uncovered by the constant
398 pressure simulations. Simulations started with randomized velocities
399 and underwent 50 ps of temperature scaling and 50 ps of constant
400 energy equilibration before obtaining a 200 ps trajectory. Diffusion
401 constants were calculated via root-mean square deviation analysis. The
402 averaged results from 5 sets of these NVE simulations is displayed in
403 figure \ref{diffuse}, alongside experimental, SPC/E, and TIP5P
404 results.\cite{Gillen72,Mills73,Clancy94,Jorgensen01}
405
406 \begin{figure}
407 \includegraphics[width=65mm, angle=-90]{betterDiffuse.epsi}
408 \caption{Average diffusion coefficient over increasing temperature for
409 SSD, SPC/E\cite{Clancy94}, TIP5P\cite{Jorgensen01}, and Experimental
410 data from Gillen \emph{et al.}\cite{Gillen72}, and from
411 Mills\cite{Mills73}.}
412 \label{diffuse}
413 \end{figure}
414
415 The observed values for the diffusion constant point out one of the
416 strengths of the SSD model. Of the three experimental models shown,
417 the SSD model has the most accurate depiction of the diffusion trend
418 seen in experiment in both the supercooled and normal regimes. SPC/E
419 does a respectable job by getting similar values as SSD and experiment
420 around 290 K; however, it deviates at both higher and lower
421 temperatures, failing to predict the experimental trend. TIP5P and SSD
422 both start off low at the colder temperatures and tend to diffuse too
423 rapidly at the higher temperatures. This type of trend at the higher
424 temperatures is not surprising in that the densities of both TIP5P and
425 SSD are lower than experimental water at temperatures higher than room
426 temperature. When calculating the diffusion coefficients for SSD at
427 experimental densities, the resulting values fall more in line with
428 experiment at these temperatures, albeit not at standard
429 pressure. Results under these conditions can be found later in this
430 paper.
431
432 \subsection{Structural Changes and Characterization}
433 By starting the simulations from the crystalline state, the melting
434 transition and the ice structure can be studied along with the liquid
435 phase behavior beyond the melting point. To locate the melting
436 transition, the constant pressure heat capacity (C$_\text{p}$) was
437 monitored in each of the simulations. In the melting simulations of
438 the 1024 particle ice $I_h$ simulations, a large spike in C$_\text{p}$
439 occurs at 245 K, indicating a first order phase transition for the
440 melting of these ice crystals. When the reaction field is turned off,
441 the melting transition occurs at 235 K. These melting transitions are
442 considerably lower than the experimental value, but this is not
443 surprising in that SSD is a simple rigid body model with a fixed
444 dipole.
445
446 \begin{figure}
447 \includegraphics[width=85mm]{fullContours.eps}
448 \caption{Contour plots of 2D angular g($r$)'s for 512 SSD systems at
449 100 K (A \& B) and 300 K (C \& D). Contour colors are inverted for
450 clarity: dark areas signify peaks while light areas signify
451 depressions. White areas have g(\emph{r}) values below 0.5 and black
452 areas have values above 1.5.}
453 \label{contour}
454 \end{figure}
455
456 Additional analyses for understanding the melting phase-transition
457 process were performed via two-dimensional structure and dipole angle
458 correlations. Expressions for the correlations are as follows:
459
460 \begin{figure}
461 \includegraphics[width=45mm]{corrDiag.eps}
462 \caption{Two dimensional illustration of the angles involved in the
463 correlations observed in figure \ref{contour}.}
464 \label{corrAngle}
465 \end{figure}
466
467 \begin{multline}
468 g_{\text{AB}}(r,\cos\theta) = \\
469 \frac{V}{N_\text{A}N_\text{B}}\langle\sum_{i\in\text{A}}\sum_{j\in\text{B}}\delta(\cos\theta-\cos\theta_{ij})\delta(r-\left|\mathbf{r}_{ij}\right|)\rangle\ ,
470 \end{multline}
471 \begin{multline}
472 g_{\text{AB}}(r,\cos\omega) = \\
473 \frac{V}{N_\text{A}N_\text{B}}\langle\sum_{i\in\text{A}}\sum_{j\in\text{B}}\delta(\cos\omega-\cos\omega_{ij})\delta(r-\left|\mathbf{r}_{ij}\right|)\rangle\ ,
474 \end{multline}
475 where $\theta$ and $\omega$ refer to the angles shown in the above
476 illustration. By binning over both distance and the cosine of the
477 desired angle between the two dipoles, the g(\emph{r}) can be
478 dissected to determine the common dipole arrangements that constitute
479 the peaks and troughs. Frames A and B of figure \ref{contour} show a
480 relatively crystalline state of an ice $I_c$ simulation. The first
481 peak of the g(\emph{r}) primarily consists of the preferred hydrogen
482 bonding arrangements as dictated by the tetrahedral sticky potential,
483 one peak for the donating and the other for the accepting hydrogen
484 bonds. Due to the high degree of crystallinity of the sample, the
485 second and third solvation shells show a repeated peak arrangement
486 which decays at distances around the fourth solvation shell, near the
487 imposed cutoff for the Lennard-Jones and dipole-dipole interactions.
488 In the higher temperature simulation shown in frames C and D, the
489 repeated peak features are significantly blurred. The first solvation
490 shell still shows the strong effect of the sticky-potential, although
491 it covers a larger area, extending to include a fraction of aligned
492 dipole peaks within the first solvation shell. The latter peaks lose
493 definition as thermal motion and the competing dipole force overcomes
494 the sticky potential's tight tetrahedral structuring of the fluid.
495
496 This complex interplay between dipole and sticky interactions was
497 remarked upon as a possible reason for the split second peak in the
498 oxygen-oxygen g(\emph{r}).\cite{Ichiye96} At low temperatures, the
499 second solvation shell peak appears to have two distinct parts that
500 blend together to form one observable peak. At higher temperatures,
501 this split character alters to show the leading 4 \AA\ peak dominated
502 by equatorial anti-parallel dipole orientations, and there is tightly
503 bunched group of axially arranged dipoles that most likely consist of
504 the smaller fraction aligned dipole pairs. The trailing part of the
505 split peak at 5 \AA\ is dominated by aligned dipoles that range
506 primarily within the axial to the chief hydrogen bond arrangements
507 similar to those seen in the first solvation shell. This evidence
508 indicates that the dipole pair interaction begins to dominate outside
509 of the range of the dipolar repulsion term, with the primary
510 energetically favorable dipole arrangements populating the region
511 immediately outside of it's range (around 4 \AA), and arrangements
512 that seek to ideally satisfy both the sticky and dipole forces locate
513 themselves just beyond this region (around 5 \AA).
514
515 From these findings, the split second peak is primarily the product of
516 the dipolar repulsion term of the sticky potential. In fact, the
517 leading of the two peaks can be pushed out and merged with the outer
518 split peak just by extending the switching function cutoff
519 ($s^\prime(r_{ij})$) from its normal 4.0 \AA\ to values of 4.5 or even
520 5 \AA. This type of correction is not recommended for improving the
521 liquid structure, because the second solvation shell will still be
522 shifted too far out. In addition, this would have an even more
523 detrimental effect on the system densities, leading to a liquid with a
524 more open structure and a density considerably lower than the normal
525 SSD behavior shown previously. A better correction would be to include
526 the quadrupole-quadrupole interactions for the water particles outside
527 of the first solvation shell, but this reduces the simplicity and
528 speed advantage of SSD, so it is not the most desirable path to take.
529
530 \subsection{Adjusted Potentials: SSD/E and SSD/RF}
531 The propensity of SSD to adopt lower than expected densities under
532 varying conditions is troubling, especially at higher temperatures. In
533 order to correct this behavior, it's necessary to adjust the force
534 field parameters for the primary intermolecular interactions. In
535 undergoing a reparameterization, it is important not to focus on just
536 one property and neglect the other important properties. In this case,
537 it would be ideal to correct the densities while maintaining the
538 accurate transport properties.
539
540 The possible parameters for tuning include the $\sigma$ and $\epsilon$
541 Lennard-Jones parameters, the dipole strength ($\mu$), and the sticky
542 attractive and dipole repulsive terms with their respective
543 cutoffs. To alter the attractive and repulsive terms of the sticky
544 potential independently, it is necessary to separate the terms as
545 follows:
546 \begin{equation}
547 \begin{split}
548 u_{ij}^{sp}
549 (\mathbf{r}_{ij},\boldsymbol{\Omega}_i,\boldsymbol{\Omega}_j) &=
550 \frac{\nu_0}{2}[s(r_{ij})w(\mathbf{r}_{ij},\boldsymbol{\Omega}_i,\boldsymbol{\Omega}_j)]\\
551 & \quad \ + \frac{\nu_0^\prime}{2}
552 [s^\prime(r_{ij})w^\prime(\mathbf{r}_{ij},\boldsymbol{\Omega}_i,\boldsymbol{\Omega}_j)],
553 \end{split}
554 \end{equation}
555
556 where $\nu_0$ scales the strength of the tetrahedral attraction and
557 $\nu_0^\prime$ acts in an identical fashion on the dipole repulsion
558 term. For purposes of the reparameterization, the separation was
559 performed, but the final parameters were adjusted so that it is
560 unnecessary to separate the terms when implementing the adjusted water
561 potentials. The results of the reparameterizations are shown in table
562 \ref{params}. Note that both the tetrahedral attractive and dipolar
563 repulsive don't share the same lower cutoff ($r_l$) in the newly
564 parameterized potentials - soft sticky dipole enhanced (SSD/E) and
565 soft sticky dipole reaction field (SSD/RF).
566
567 \begin{table}
568 \caption{Parameters for the original and adjusted models}
569 \begin{tabular}{ l c c c }
570 \hline \\[-3mm]
571 \ Parameters & \ \ \ SSD$^\dagger$\ \ \ \ & \ SSD/E\ \ & \ SSD/RF\ \ \\
572 \hline \\[-3mm]
573 \ \ \ $\sigma$ (\AA) & 3.051 & 3.035 & 3.019\\
574 \ \ \ $\epsilon$ (kcal/mol)\ \ & 0.152 & 0.152 & 0.152\\
575 \ \ \ $\mu$ (D) & 2.35 & 2.418 & 2.480\\
576 \ \ \ $\nu_0$ (kcal/mol)\ \ & 3.7284 & 3.90 & 3.90\\
577 \ \ \ $r_l$ (\AA) & 2.75 & 2.40 & 2.40\\
578 \ \ \ $r_u$ (\AA) & 3.35 & 3.80 & 3.80\\
579 \ \ \ $\nu_0^\prime$ (kcal/mol)\ \ & 3.7284 & 3.90 & 3.90\\
580 \ \ \ $r_l^\prime$ (\AA) & 2.75 & 2.75 & 2.75\\
581 \ \ \ $r_u^\prime$ (\AA) & 4.00 & 3.35 & 3.35\\
582 \\[-2mm]$^\dagger$ ref. \onlinecite{Ichiye96}
583 \end{tabular}
584 \label{params}
585 \end{table}
586
587 \begin{figure}
588 \includegraphics[width=85mm]{gofrCompare.epsi}
589 \caption{Plots comparing experiment\cite{Head-Gordon00_1} with SSD/E
590 and SSD without reaction field (top), as well as SSD/RF and SSD with
591 reaction field turned on (bottom). The insets show the respective
592 first peaks in detail. Solid Line - experiment, dashed line - SSD/E
593 and SSD/RF, and dotted line - SSD (with and without reaction field).}
594 \label{grcompare}
595 \end{figure}
596
597 \begin{figure}
598 \includegraphics[width=85mm]{dualsticky.ps}
599 \caption{Isosurfaces of the sticky potential for SSD (left) and SSD/E \&
600 SSD/RF (right). Light areas correspond to the tetrahedral attractive
601 part, and the darker areas correspond to the dipolar repulsive part.}
602 \label{isosurface}
603 \end{figure}
604
605 In the paper detailing the development of SSD, Liu and Ichiye placed
606 particular emphasis on an accurate description of the first solvation
607 shell. This resulted in a somewhat tall and sharp first peak that
608 integrated to give similar coordination numbers to the experimental
609 data obtained by Soper and Phillips.\cite{Ichiye96,Soper86} New
610 experimental x-ray scattering data from the Head-Gordon lab indicates
611 a slightly lower and shifted first peak in the g$_\mathrm{OO}(r)$, so
612 adjustments to SSD were made while taking into consideration the new
613 experimental findings.\cite{Head-Gordon00_1} Figure \ref{grcompare}
614 shows the relocation of the first peak of the oxygen-oxygen
615 g(\emph{r}) by comparing the original SSD (with and without reaction
616 field), SSD-E, and SSD-RF to the new experimental results. Both the
617 modified water models have shorter peaks that are brought in more
618 closely to the experimental peak (as seen in the insets of figure
619 \ref{grcompare}). This structural alteration was accomplished by a
620 reduction in the Lennard-Jones $\sigma$ variable as well as adjustment
621 of the sticky potential strength and cutoffs. The cutoffs for the
622 tetrahedral attractive and dipolar repulsive terms were nearly swapped
623 with each other. Isosurfaces of the original and modified sticky
624 potentials are shown in figure \cite{isosurface}. In these
625 isosurfaces, it is easy to see how altering the cutoffs changes the
626 repulsive and attractive character of the particles. With a reduced
627 repulsive surface (the darker region), the particles can move closer
628 to one another, increasing the density for the overall system. This
629 change in interaction cutoff also results in a more gradual
630 orientational motion by allowing the particles to maintain preferred
631 dipolar arrangements before they begin to feel the pull of the
632 tetrahedral restructuring. Upon moving closer together, the dipolar
633 repulsion term becomes active and excludes the unphysical
634 arrangements. This compares with the original SSD's excluding dipolar
635 before the particles feel the pull of the ``hydrogen bonds''. Aside
636 from improving the shape of the first peak in the g(\emph{r}), this
637 improves the densities considerably by allowing the persistence of
638 full dipolar character below the previous 4.0 \AA\ cutoff.
639
640 While adjusting the location and shape of the first peak of
641 g(\emph{r}) improves the densities to some degree, these changes alone
642 are insufficient to bring the system densities up to the values
643 observed experimentally. To finish bringing up the densities, the
644 dipole moments were increased in both the adjusted models. Being a
645 dipole based model, the structure and transport are very sensitive to
646 changes in the dipole moment. The original SSD simply used the dipole
647 moment calculated from the TIP3P water model, which at 2.35 D is
648 significantly greater than the experimental gas phase value of 1.84
649 D. The larger dipole moment is a more realistic value and improve the
650 dielectric properties of the fluid. Both theoretical and experimental
651 measurements indicate a liquid phase dipole moment ranging from 2.4 D
652 to values as high as 3.11 D, so there is quite a range available for
653 adjusting the dipole
654 moment.\cite{Sprik91,Kusalik02,Badyal00,Barriol64} The increasing of
655 the dipole moments to 2.418 and 2.48 D for SSD/E and SSD/RF
656 respectively is moderate in the range of the experimental values;
657 however, it leads to significant changes in the density and transport
658 of the water models.
659
660 In order to demonstrate the benefits of this reparameterization, a
661 series of NPT and NVE simulations were performed to probe the density
662 and transport properties of the adapted models and compare the results
663 to the original SSD model. This comparison involved full NPT melting
664 sequences for both SSD/E and SSD/RF, as well as NVE transport
665 calculations at both self-consistent and experimental
666 densities. Again, the results come from five separate simulations of
667 1024 particle systems, and the melting sequences were started from
668 different ice $I_h$ crystals constructed as stated previously. Like
669 before, all of the NPT simulations were equilibrated for 100 ps before
670 a 200 ps data collection run, and they used the previous temperature's
671 final configuration as a starting point. All of the NVE simulations
672 had the same thermalization, equilibration, and data collection times
673 stated earlier in this paper.
674
675 \begin{figure}
676 \includegraphics[width=85mm]{ssdecompare.epsi}
677 \caption{Comparison of densities calculated with SSD/E to SSD without a
678 reaction field, TIP4P\cite{Jorgensen98b}, TIP3P\cite{Jorgensen98b},
679 SPC/E\cite{Clancy94}, and Experiment\cite{CRC80}. The upper plot
680 includes error bars, and the calculated results from the other
681 references were removed for clarity.}
682 \label{ssdedense}
683 \end{figure}
684
685 Figure \ref{ssdedense} shows the density profile for the SSD/E water
686 model in comparison to the original SSD without a reaction field,
687 experiment, and the other common water models considered
688 previously. The calculated densities have increased significantly over
689 the original SSD model and match the experimental value just below 298
690 K. At 298 K, the density of SSD/E is 0.995$\pm$0.001 g/cm$^3$, which
691 compares well with the experimental value of 0.997 g/cm$^3$ and is
692 considerably better than the SSD value of 0.967$\pm$0.003
693 g/cm$^3$. The increased dipole moment in SSD/E has helped to flatten
694 out the curve at higher temperatures, only the improvement is marginal
695 at best. This steep drop in densities is due to the dipolar rather
696 than charge based interactions which decay more rapidly at longer
697 distances.
698
699 By monitoring C$\text{p}$ throughout these simulations, the melting
700 transition for SSD/E was observed at 230 K, about 5 degrees lower than
701 SSD. The resulting density maximum is located at 240 K, again about 5
702 degrees lower than the SSD value of 245 K. Though there is a decrease
703 in both of these values, the corrected densities near room temperature
704 justify the modifications taken.
705
706 \begin{figure}
707 \includegraphics[width=85mm]{ssdrfcompare.epsi}
708 \caption{Comparison of densities calculated with SSD/RF to SSD with a
709 reaction field, TIP4P\cite{Jorgensen98b}, TIP3P\cite{Jorgensen98b},
710 SPC/E\cite{Clancy94}, and Experiment\cite{CRC80}. The upper plot
711 includes error bars, and the calculated results from the other
712 references were removed for clarity.}
713 \label{ssdrfdense}
714 \end{figure}
715
716 Figure \ref{ssdrfdense} shows a density comparison between SSD/RF and
717 SSD with an active reaction field. Like in the simulations of SSD/E,
718 the densities show a dramatic increase over normal SSD. At 298 K,
719 SSD/RF has a density of 0.997$\pm$0.001 g/cm$^3$, right in line with
720 experiment and considerably better than the SSD value of
721 0.941$\pm$0.001 g/cm$^3$. The melting point is observed at 240 K,
722 which is 5 degrees lower than SSD with a reaction field, and the
723 density maximum at 255 K, again 5 degrees lower than SSD. The density
724 at higher temperature still drops off more rapidly than the charge
725 based models but is in better agreement than SSD/E.
726
727 The reparameterization of the SSD water model, both for use with and
728 without an applied long-range correction, brought the densities up to
729 what is expected for simulating liquid water. In addition to improving
730 the densities, it is important that particle transport be maintained
731 or improved. Figure \ref{ssdediffuse} compares the temperature
732 dependence of the diffusion constant of SSD/E to SSD without an active
733 reaction field, both at the densities calculated at 1 atm and at the
734 experimentally calculated densities for super-cooled and liquid
735 water. In the upper plot, the diffusion constant for SSD/E is
736 consistently a little faster than experiment, while SSD starts off
737 slower than experiment and crosses to merge with SSD/E at high
738 temperatures. Both models follow the experimental trend well, but
739 diffuse too rapidly at higher temperatures. This abnormally fast
740 diffusion is caused by the decreased system density. Since the
741 densities of SSD/E don't deviate as much from experiment as those of
742 SSD, it follows the experimental trend more closely. This observation
743 is backed up by looking at the lower plot. The diffusion constants for
744 SSD/E track with the experimental values while SSD deviates on the low
745 side of the trend with increasing temperature. This is again a product
746 of SSD/E having densities closer to experiment, and not deviating to
747 lower densities with increasing temperature as rapidly.
748
749 \begin{figure}
750 \includegraphics[width=85mm]{ssdediffuse.epsi}
751 \caption{Plots of the diffusion constants calculated from SSD/E and SSD,
752 both without a reaction field along with experimental results from
753 Gillen \emph{et al.}\cite{Gillen72} and Mills\cite{Mills73}. The
754 upper plot is at densities calculated from the NPT simulations at a
755 pressure of 1 atm, while the lower plot is at the experimentally
756 calculated densities.}
757 \label{ssdediffuse}
758 \end{figure}
759
760 \begin{figure}
761 \includegraphics[width=85mm]{ssdrfdiffuse.epsi}
762 \caption{Plots of the diffusion constants calculated from SSD/RF and SSD,
763 both with an active reaction field along with experimental results
764 from Gillen \emph{et al.}\cite{Gillen72} and Mills\cite{Mills73}. The
765 upper plot is at densities calculated from the NPT simulations at a
766 pressure of 1 atm, while the lower plot is at the experimentally
767 calculated densities.}
768 \label{ssdrfdiffuse}
769 \end{figure}
770
771 In figure \ref{ssdrfdiffuse}, the diffusion constants for SSD/RF are
772 compared with SSD with an active reaction field. In the upper plot,
773 SSD/RF tracks with the experimental results incredibly well, identical
774 within error throughout the temperature range and only showing a
775 slight increasing trend at higher temperatures. SSD also tracks
776 experiment well, only it tends to diffuse a little more slowly at low
777 temperatures and deviates to diffuse too rapidly at high
778 temperatures. As was stated in the SSD/E comparisons, this deviation
779 away from the ideal trend is due to a rapid decrease in density at
780 higher temperatures. SSD/RF doesn't suffer from this problem as much
781 as SSD, because the calculated densities are more true to
782 experiment. This is again emphasized in the lower plot, where SSD/RF
783 tracks the experimental diffusion exactly while SSD's diffusion
784 constants are slightly too low due to its need for a lower density at
785 the specified temperature.
786
787 \subsection{Additional Observations}
788
789 While performing the melting sequences of SSD/E, some interesting
790 observations were made. After melting at 230 K, two of the systems
791 underwent crystallization events near 245 K. As the heating process
792 continued, the two systems remained crystalline until finally melting
793 between 320 and 330 K. These simulations were excluded from the data
794 set shown in figure \ref{ssdedense} and replaced with two additional
795 melting sequences that did not undergo this anomalous phase
796 transition, while this crystallization event was investigated
797 separately.
798
799 \begin{figure}
800 \includegraphics[width=85mm]{povIce.ps}
801 \caption{Crystal structure of an ice 0 lattice shown from the (001) face.}
802 \label{weirdice}
803 \end{figure}
804
805 The final configurations of these two melting sequences shows an
806 expanded zeolite-like crystal structure that does not correspond to
807 any known form of ice. For convenience and to help distinguish it from
808 the experimentally observed forms of ice, this crystal structure will
809 henceforth be referred to as ice-zero (ice 0). The crystallinity was
810 extensive enough than a near ideal crystal structure could be
811 obtained. Figure \ref{weirdice} shows the repeating crystal structure
812 of a typical crystal at 5 K. The unit cell contains eight molecules,
813 and figure \ref{unitcell} shows a unit cell built from the water
814 particle center of masses that can be used to construct a repeating
815 lattice, similar to figure \ref{weirdice}. Each molecule is hydrogen
816 bonded to four other water molecules; however, the hydrogen bonds are
817 flexed rather than perfectly straight. This results in a skewed
818 tetrahedral geometry about the central molecule. Looking back at
819 figure \ref{isosurface}, it is easy to see how these flexed hydrogen
820 bonds are allowed in that the attractive regions are conical in shape,
821 with the greatest attraction in the central region. Though not ideal,
822 these flexed hydrogen bonds are favorable enough to stabilize an
823 entire crystal generated around them. In fact, the imperfect ice 0
824 crystals were so stable that they melted at greater than room
825 temperature.
826
827 \begin{figure}
828 \includegraphics[width=65mm]{ice0cell.eps}
829 \caption{Simple unit cell for constructing ice 0. In this cell, $c$ is
830 equal to $0.4714\times a$, and a typical value for $a$ is 8.25 \AA.}
831 \label{unitcell}
832 \end{figure}
833
834 The initial simulations indicated that ice 0 is the preferred ice
835 structure for at least SSD/E. To verify this, a comparison was made
836 between near ideal crystals of ice $I_h$, ice $I_c$, and ice 0 at
837 constant pressure with SSD/E, SSD/RF, and SSD. Near ideal versions of
838 the three types of crystals were cooled to ~1 K, and the potential
839 energies of each were compared using all three water models. With
840 every water model, ice 0 turned out to have the lowest potential
841 energy: 5\% lower than $I_h$ with SSD, 6.5\% lower with SSD/E, and
842 7.5\% lower with SSD/RF. In all three of these water models, ice $I_c$
843 was observed to be ~2\% less stable than ice $I_h$. In addition to
844 having the lowest potential energy, ice 0 was the most expanded of the
845 three ice crystals, ~5\% less dense than ice $I_h$ with all of the
846 water models. In all three water models, ice $I_c$ was observed to be
847 ~2\% more dense than ice $I_h$.
848
849 In addition to the low temperature comparisons, melting sequences were
850 performed with ice 0 as the initial configuration using SSD/E, SSD/RF,
851 and SSD both with and without a reaction field. The melting
852 transitions for both SSD/E and SSD without a reaction field occurred
853 at temperature in excess of 375 K. SSD/RF and SSD with a reaction
854 field had more reasonable melting transitions, down near 325 K. These
855 melting point observations emphasize how preferred this crystal
856 structure is over the most common types of ice when using these single
857 point water models.
858
859 Recognizing that the above tests show ice 0 to be both the most stable
860 and lowest density crystal structure for these single point water
861 models, it is interesting to speculate on the favorability of this
862 crystal structure with the different charge based models. As a quick
863 test, these 3 crystal types were converted from SSD type particles to
864 TIP3P waters and read into CHARMM.\cite{Karplus83} Identical energy
865 minimizations were performed on all of these crystals to compare the
866 system energies. Again, ice 0 was observed to have the lowest total
867 system energy. The total energy of ice 0 was ~2\% lower than ice
868 $I_h$, which was in turn ~3\% lower than ice $I_c$. From these initial
869 results, we would not be surprised if results from the other common
870 water models show ice 0 to be the lowest energy crystal structure. A
871 continuation on work studing ice 0 with multipoint water models will
872 be published in a coming article.
873
874 \section{Conclusions}
875 The density maximum and temperature dependent transport for the SSD
876 water model, both with and without the use of reaction field, were
877 studied via a series of NPT and NVE simulations. The constant pressure
878 simulations of the melting of both $I_h$ and $I_c$ ice showed a
879 density maximum near 260 K. In most cases, the calculated densities
880 were significantly lower than the densities calculated in simulations
881 of other water models. Analysis of particle diffusion showed SSD to
882 capture the transport properties of experimental very well in both the
883 normal and super-cooled liquid regimes. In order to correct the
884 density behavior, SSD was reparameterized for use both with and
885 without a long-range interaction correction, SSD/RF and SSD/E
886 respectively. In addition to correcting the abnormally low densities,
887 these new versions were show to maintain or improve upon the transport
888 and structural features of the original water model, all while
889 maintaining the fast performance of the SSD water model. This work
890 shows these simple water models, and in particular SSD/E and SSD/RF,
891 to be excellent choices to represent explicit water in future
892 simulations of biochemical systems.
893
894 \section{Acknowledgments}
895 The authors would like to thank the National Science Foundation for
896 funding under grant CHE-0134881. Computation time was provided by the
897 Notre Dame Bunch-of-Boxes (B.o.B) computer cluster under NSF grant DMR
898 00 79647.
899
900 \bibliographystyle{jcp}
901
902 \bibliography{nptSSD}
903
904 %\pagebreak
905
906 \end{document}