ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/group/trunk/multipoleSFPaper/multipoleSFPaper.tex
Revision: 3066
Committed: Tue Oct 24 16:11:42 2006 UTC (17 years, 8 months ago) by gezelter
Content type: application/x-tex
File size: 54396 byte(s)
Log Message:
working on the paper

File Contents

# User Rev Content
1 chrisfen 3064 %\documentclass[prb,aps,twocolumn,tabularx]{revtex4}
2 gezelter 3066 \documentclass[11pt]{article}
3 chrisfen 3064 \usepackage{tabls}
4     \usepackage{endfloat}
5     \usepackage[tbtags]{amsmath}
6     \usepackage{amsmath,bm}
7     \usepackage{amssymb}
8     \usepackage{mathrsfs}
9     \usepackage{setspace}
10     \usepackage{tabularx}
11     \usepackage{graphicx}
12     \usepackage{booktabs}
13     \usepackage{colortbl}
14     \usepackage[ref]{overcite}
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 gezelter 3066 \AtBeginDelayedFloats{\renewcommand{\baselinestretch}{2}} %doublespace captions
24     \let\Caption\caption
25     \renewcommand\caption[1]{%
26     \Caption[#1]{}%
27     }
28     \setlength{\abovecaptionskip}{1.2in}
29     \setlength{\belowcaptionskip}{1.2in}
30 chrisfen 3064
31     \begin{document}
32    
33 gezelter 3066 \title{Pairwise Alternatives to the Ewald Sum: Applications
34     and Extension to Point Multipoles}
35 chrisfen 3064
36     \author{Christopher J. Fennell and J. Daniel Gezelter \\
37     Department of Chemistry and Biochemistry\\ University of Notre Dame\\
38     Notre Dame, Indiana 46556}
39    
40     \date{\today}
41    
42     \maketitle
43     %\doublespacing
44    
45     \begin{abstract}
46 gezelter 3066 The damped, shifted-force electrostatic potential has been shown to
47     give nearly quantitative agreement with smooth particle mesh Ewald for
48     energy differences between configurations as well as for atomic force
49     and molecular torque vectors. In this paper, we extend this technique
50     to handle interactions between electrostatic multipoles. We also
51     investigate the effects of damped and shifted electrostatics on the
52     static, thermodynamic, and dynamic properties of liquid water and
53     various polymorphs of ice. We provide a way of choosing the optimal
54     damping strength for a given cutoff radius that reproduces the static
55     dielectric constant in a variety of water models.
56 chrisfen 3064 \end{abstract}
57    
58     %\narrowtext
59    
60     %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
61     % BODY OF TEXT
62     %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
63    
64     \section{Introduction}
65    
66 gezelter 3066 Over the past several years, there has been increasing interest in
67     pairwise methods for correcting electrostatic interactions in computer
68 chrisfen 3064 simulations of condensed molecular
69 gezelter 3066 systems.\cite{Wolf99,Zahn02,Kast03,Beck05,Ma05,Fennell06} These
70     techniques were developed from the observations and efforts of Wolf
71     {\it et al.} and their work towards an $\mathcal{O}(N)$ Coulombic
72     sum.\cite{Wolf99} Wolf's method of cutoff neutralization is able to
73     obtain excellent agreement with Madelung energies in ionic
74     crystals.\cite{Wolf99}
75 chrisfen 3064
76 gezelter 3066 In a recent paper, we showed that simple modifications to Wolf's
77     method could give nearly quantitative agreement with smooth particle
78     mesh Ewald (SPME) for quantities of interest in Monte Carlo
79     (i.e. configurational energy differences) and Molecular Dynamics
80     (i.e. atomic force and molecular torque vectors).\cite{Fennell06} We
81     described the undamped and damped shifted potential (SP) and shifted
82     force (SF) techniques. The potential for damped form of the SF method
83     is given by
84 chrisfen 3064 \begin{equation}
85     \begin{split}
86     V_\mathrm{DSF}(r) = q_iq_j\Biggr{[}&
87     \frac{\mathrm{erfc}\left(\alpha r\right)}{r}
88     - \frac{\mathrm{erfc}\left(\alpha R_\mathrm{c}\right)}{R_\mathrm{c}} \\
89     &+ \left(\frac{\mathrm{erfc}\left(\alpha R_\mathrm{c}\right)}{R_\mathrm{c}^2}
90     + \frac{2\alpha}{\pi^{1/2}}
91     \frac{\exp\left(-\alpha^2R_\mathrm{c}^2\right)}{R_\mathrm{c}}
92     \right)\left(r-R_\mathrm{c}\right)\ \Biggr{]}
93     \quad r\leqslant R_\textrm{c}.
94     \label{eq:DSFPot}
95     \end{split}
96     \end{equation}
97 gezelter 3066 (In this potential and in all electrostatic quantities that follow,
98     the standard $4 \pi \epsilon_{0}$ has been omitted for clarity.)
99 chrisfen 3064
100 gezelter 3066 The damped SF method is an improvement over the SP method because
101     there is no discontinuity in the forces as particles move out of the
102     cutoff radius ($R_\textrm{c}$). This is accomplished by shifting the
103     forces (and potential) to zero at $R_\textrm{c}$. This is analogous to
104     neutralizing the charge as well as the force effect of the charges
105     within $R_\textrm{c}$.
106    
107     To complete the charge neutralization procedure, a self-neutralization
108     term needs to be included in the potential. This term is constant (as
109     long as the charges and cutoff radius do not change), and exists
110     outside the normal pair-loop. It can be thought of as a contribution
111     from a charge opposite in sign and equal in magnitude to the central
112     charge, but which has been spread out over the surface of the cutoff
113     sphere. This term is calculated via a single loop over all charges in
114     the system. For the undamped case, the self term can be written as
115 chrisfen 3064 \begin{equation}
116 gezelter 3066 V_\textrm{self} = \frac{1}{2 R_\textrm{c}} \sum_{i=1}^N q_i^{2},
117 chrisfen 3064 \label{eq:selfTerm}
118     \end{equation}
119     while for the damped case it can be written as
120     \begin{equation}
121 gezelter 3066 V_\textrm{self} = \left(\frac{\alpha}{\sqrt{\pi}}
122     + \frac{\textrm{erfc}(\alpha
123     R_\textrm{c})}{2R_\textrm{c}}\right) \sum_{i=1}^N q_i^{2}.
124 chrisfen 3064 \label{eq:dampSelfTerm}
125     \end{equation}
126     The first term within the parentheses of equation
127 gezelter 3066 (\ref{eq:dampSelfTerm}) is identical to the self term in the Ewald
128 chrisfen 3064 summation, and comes from the utilization of the complimentary error
129     function for electrostatic damping.\cite{deLeeuw80,Wolf99}
130    
131 gezelter 3066 The SF, SP, and Wolf methods operate by neutralizing the total charge
132     contained within the cutoff sphere surrounding each particle. This is
133     accomplished by creating image charges on the surface of the cutoff
134     sphere for each pair interaction computed within the sphere. The
135     damping function applied to the potential is also an important method
136     for accelerating convergence. In the case of systems involving
137     electrostatic distributions of higher order than point charges, the
138     question remains: How will the shifting and damping need to be
139     modified in order to accommodate point multipoles?
140 chrisfen 3064
141 gezelter 3066 \section{Electrostatic Damping for Point
142     Multipoles}\label{sec:dampingMultipoles}
143 chrisfen 3064
144 gezelter 3066 To apply the SF method for systems involving point multipoles, we
145     consider separately the two techniques (shifting and damping) which
146     contribute to the effectiveness of the DSF potential.
147 chrisfen 3064
148 gezelter 3066 As noted above, shifting the potential and forces is employed to
149     neutralize the total charge contained within each cutoff sphere;
150     however, in a system composed purely of point multipoles, each cutoff
151     sphere is already neutral, so shifting becomes unnecessary.
152    
153     In a mixed system of charges and multipoles, the undamped SF potential
154     needs only to shift the force terms between charges and smoothly
155     truncate the multipolar interactions with a switching function. The
156     switching function is required for energy conservation, because a
157     discontinuity will exist in both the potential and forces at
158     $R_\textrm{c}$ in the absence of shifting terms.
159    
160     To damp the SF potential for point multipoles, we need to incorporate
161     the complimentary error function term into the standard forms of the
162     multipolar potentials. We can determine the necessary damping
163     functions by replacing $1/r$ with $\mathrm{erfc}(\alpha r)/r$ in the
164     multipole expansion. This procedure quickly becomes quite complex
165     with ``two-center'' systems, like the dipole-dipole potential, and is
166     typically approached using spherical harmonics.\cite{Hirschfelder67} A
167     simpler method for determining damped multipolar interaction
168     potentials arises when we adopt the tensor formalism described by
169     Stone.\cite{Stone02}
170    
171     The tensor formalism for electrostatic interactions involves obtaining
172     the multipole interactions from successive gradients of the monopole
173     potential. Thus, tensors of rank zero through two are
174 chrisfen 3064 \begin{equation}
175 gezelter 3066 T = \frac{1}{r_{ij}},
176     \label{eq:tensorRank1}
177 chrisfen 3064 \end{equation}
178     \begin{equation}
179 gezelter 3066 T_\alpha = \nabla_\alpha \frac{1}{r_{ij}},
180     \label{eq:tensorRank2}
181 chrisfen 3064 \end{equation}
182 gezelter 3066 \begin{equation}
183     T_{\alpha\beta} = \nabla_\alpha\nabla_\beta \frac{1}{r_{ij}},
184     \label{eq:tensorRank3}
185     \end{equation}
186     where the form of the first tensor is the charge-charge potential, the
187     second gives the charge-dipole potential, and the third gives the
188     charge-quadrupole and dipole-dipole potentials.\cite{Stone02} Since
189     the force is $-\nabla V$, the forces for each potential come from the
190     next higher tensor.
191 chrisfen 3064
192 gezelter 3066 As one would do with the multipolar expansion, we can replace $r^{-1}$
193     with $\mathrm{erfc}(\alpha r)\cdot r^{-1}$ to obtain damped forms of the
194     electrostatic potentials. Equation \ref{eq:tensorRank2} generates a
195     damped charge-dipole potential,
196 chrisfen 3064 \begin{equation}
197     V_\textrm{Dcd} = -q_i\mu_j\frac{\cos\theta}{r^2_{ij}} c_1(r_{ij}),
198     \label{eq:dChargeDipole}
199     \end{equation}
200     where $c_1(r_{ij})$ is
201     \begin{equation}
202     c_1(r_{ij}) = \frac{2\alpha r_{ij}e^{-\alpha^2r^2_{ij}}}{\sqrt{\pi}}
203     + \textrm{erfc}(\alpha r_{ij}).
204     \label{eq:c1Func}
205     \end{equation}
206    
207 gezelter 3066 Equation \ref{eq:tensorRank3} generates a damped dipole-dipole potential,
208 chrisfen 3064 \begin{equation}
209 gezelter 3066 V_\textrm{Ddd} = 3\frac{(\boldsymbol{\mu}_i\cdot\mathbf{r}_{ij})
210     (\boldsymbol{\mu}_j\cdot\mathbf{r}_{ij})}{r^5_{ij}}
211     c_2(r_{ij}) -
212     \frac{\boldsymbol{\mu}_i\cdot\boldsymbol{\mu}_j}{r^3_{ij}}
213     c_1(r_{ij}),
214     \label{eq:dampDipoleDipole}
215 chrisfen 3064 \end{equation}
216 gezelter 3066 where
217 chrisfen 3064 \begin{equation}
218     c_2(r_{ij}) = \frac{4\alpha^3r^3_{ij}e^{-\alpha^2r^2_{ij}}}{3\sqrt{\pi}}
219     + \frac{2\alpha r_{ij}e^{-\alpha^2r^2_{ij}}}{\sqrt{\pi}}
220     + \textrm{erfc}(\alpha r_{ij}).
221     \end{equation}
222     Note that $c_2(r_{ij})$ is equal to $c_1(r_{ij})$ plus an additional
223     term. Continuing with higher rank tensors, we can obtain the damping
224     functions for higher multipole potentials and forces. Each subsequent
225     damping function includes one additional term, and we can simplify the
226     procedure for obtaining these terms by writing out the following
227     generating function,
228     \begin{equation}
229     c_n(r_{ij}) = \frac{2^n(\alpha r_{ij})^{2n-1}e^{-\alpha^2r^2_{ij}}}
230     {(2n-1)!!\sqrt{\pi}} + c_{n-1}(r_{ij}),
231     \label{eq:dampingGeneratingFunc}
232     \end{equation}
233     where,
234     \begin{equation}
235     m!! \equiv \left\{ \begin{array}{l@{\quad\quad}l}
236     m\cdot(m-2)\ldots 5\cdot3\cdot1 & m > 0\textrm{ odd} \\
237     m\cdot(m-2)\ldots 6\cdot4\cdot2 & m > 0\textrm{ even} \\
238     1 & m = -1\textrm{ or }0,
239     \end{array}\right.
240     \label{eq:doubleFactorial}
241     \end{equation}
242     and $c_0(r_{ij})$ is erfc$(\alpha r_{ij})$. This generating function
243 gezelter 3066 is similar in form to those obtained by Smith and Aguado and Madden
244     for the application of the Ewald sum to
245 chrisfen 3064 multipoles.\cite{Smith82,Smith98,Aguado03}
246    
247     Returning to the dipole-dipole example, the potential consists of a
248     portion dependent upon $r^{-5}$ and another dependent upon
249 gezelter 3066 $r^{-3}$.
250 chrisfen 3064 $c_2(r_{ij})$ and $c_1(r_{ij})$ dampen these two parts
251 gezelter 3066 respectively. The forces for the damped dipole-dipole interaction, are
252     obtained from the next higher tensor, $T_{\alpha \beta \gamma}$,
253 chrisfen 3064 \begin{equation}
254     \begin{split}
255     F_\textrm{Ddd} = &15\frac{(\boldsymbol{\mu}_i\cdot\mathbf{r}_{ij})
256     (\boldsymbol{\mu}_j\cdot\mathbf{r}_{ij})\mathbf{r}_{ij}}{r^7_{ij}}
257     c_3(r_{ij})\\ &-
258     3\frac{\boldsymbol{\mu}_i\cdot\mathbf{r}_{ij}\cdot\hat{\boldsymbol{\mu}}_j +
259     \boldsymbol{\mu}_j\cdot\mathbf{r}_{ij}\cdot\hat{\boldsymbol{\mu}}_i +
260     \boldsymbol{\mu}_i\cdot\boldsymbol{\mu}_j\cdot\mathbf{r}_{ij}}
261     {r^5_{ij}} c_2(r_{ij}),
262     \end{split}
263     \label{eq:dampDipoleDipoleForces}
264     \end{equation}
265 gezelter 3066 Using the tensor formalism, we can dampen higher order multipolar
266     interactions using the same effective damping function that we use for
267     charge-charge interactions. This allows us to include multipoles in
268     simulations involving damped electrostatic interactions.
269 chrisfen 3064
270 gezelter 3066 \section{Applications of DSF Electrostatics}
271 chrisfen 3064
272 gezelter 3066 Our earlier work on the SF method showed that it can give nearly
273     quantitive agreement with SPME-derived configurational energy
274     differences. The force and torque vectors in identical configurations
275     are also nearly equivalent under the damped SF potential and
276     SPME.\cite{Fennell06} Although these measures bode well for the
277     performance of the SF method in both Monte Carlo and Molecular
278     Dynamics simulations, it would be helpful to have direct comparisons
279     of structural, thermodynamic, and dynamic quantities. To address
280     this, we performed a detailed analysis of a group of simulations
281     involving water models (both point charge and multipolar) under a
282     number of different simulation conditions.
283 chrisfen 3064
284 gezelter 3066 To provide the most difficult test for the damped SF method, we have
285     chosen a model that has been optimized for use with Ewald sum, and
286     have compared the simulated properties to those computed via Ewald.
287     It is well known that water models parametrized for use with the Ewald
288     sum give calculated properties that are in better agreement with
289     experiment.\cite{vanderSpoel98,Horn04,Rick04} For these reasons, we
290     chose the TIP5P-E water model for our comparisons involving point
291     charges.\cite{Rick04}
292 chrisfen 3064
293 gezelter 3066 The soft sticky dipole (SSD) family of water models is the perfect
294     test case for the point-multipolar extension of damped electrostatics.
295     SSD water models are single point molecules that consist of a ``soft''
296     Lennard-Jones sphere, a point-dipole, and a tetrahedral function for
297     capturing the hydrogen bonding nature of water - a spherical harmonic
298     term for water-water tetrahedral interactions and a point-quadrupole
299     for interactions with surrounding charges. Detailed descriptions of
300     these models can be found in other
301     studies.\cite{Liu96b,Chandra99,Tan03,Fennell04}
302    
303     In deciding which version of the SSD model to use, we need only
304     consider that the SF technique was presented as a pairwise replacement
305     for the Ewald summation. It has been suggested that models
306     parametrized for the Ewald summation (like TIP5P-E) would be
307     appropriate for use with a reaction field and vice versa.\cite{Rick04}
308     Therefore, we decided to test the SSD/RF water model, which was
309     parametrized for use with a reaction field, with the damped
310     electrostatic technique to see how the calculated properties change.
311    
312 chrisfen 3064 The TIP5P-E water model is a variant of Mahoney and Jorgensen's
313     five-point transferable intermolecular potential (TIP5P) model for
314     water.\cite{Mahoney00} TIP5P was developed to reproduce the density
315 gezelter 3066 maximum in liquid water near 4$^\circ$C. As with many previous point
316     charge water models (such as ST2, TIP3P, TIP4P, SPC, and SPC/E), TIP5P
317     was parametrized using a simple cutoff with no long-range
318     electrostatic
319 chrisfen 3064 correction.\cite{Stillinger74,Jorgensen83,Berendsen81,Berendsen87}
320     Without this correction, the pressure term on the central particle
321     from the surroundings is missing. When this correction is included,
322     systems of these particles expand to compensate for this added
323     pressure term and under-predict the density of water under standard
324 gezelter 3066 conditions. In developing TIP5P-E, Rick preserved the geometry and
325     point charge magnitudes in TIP5P and focused on altering the
326     Lennard-Jones parameters to correct the density at 298~K. With the
327 chrisfen 3064 density corrected, he compared common water properties for TIP5P-E
328     using the Ewald sum with TIP5P using a 9~\AA\ cutoff.
329    
330 gezelter 3066 In the following sections, we compare these same properties calculated
331     from TIP5P-E using the Ewald sum with TIP5P-E using the damped SF
332     technique. Our comparisons include the SF technique with a 12~\AA\
333     cutoff and an $\alpha$ = 0.0, 0.1, and 0.2~\AA$^{-1}$, as well as a
334     9~\AA\ cutoff with an $\alpha$ = 0.2~\AA$^{-1}$.
335 chrisfen 3064
336 gezelter 3066 \subsection{The Density Maximum of TIP5P-E}\label{sec:t5peDensity}
337 chrisfen 3064
338     To compare densities, $NPT$ simulations were performed with the same
339     temperatures as those selected by Rick in his Ewald summation
340     simulations.\cite{Rick04} In order to improve statistics around the
341     density maximum, 3~ns trajectories were accumulated at 0, 12.5, and
342     25$^\circ$C, while 2~ns trajectories were obtained at all other
343     temperatures. The average densities were calculated from the later
344     three-fourths of each trajectory. Similar to Mahoney and Jorgensen's
345     method for accumulating statistics, these sequences were spliced into
346     200 segments, each providing an average density. These 200 density
347     values were used to calculate the average and standard deviation of
348     the density at each temperature.\cite{Mahoney00}
349    
350     \begin{figure}
351     \includegraphics[width=\linewidth]{./figures/tip5peDensities.pdf}
352     \caption{Density versus temperature for the TIP5P-E water model when
353 gezelter 3066 using the Ewald summation (Ref. \citen{Rick04}) and the SF method with
354     varying cutoff radii and damping coefficients. The pressure term from
355     the image-charge shell is larger than that provided by the
356     reciprocal-space portion of the Ewald summation, leading to slightly
357     lower densities. This effect is more visible with the 9~\AA\ cutoff,
358     where the image charges exert a greater force on the central
359     particle. The error bars for the SF methods show the average one-sigma
360     uncertainty of the density measurement, and this uncertainty is the
361     same for all the SF curves.}
362 chrisfen 3064 \label{fig:t5peDensities}
363     \end{figure}
364     Figure \ref{fig:t5peDensities} shows the densities calculated for
365 gezelter 3066 TIP5P-E using differing electrostatic corrections overlaid with the
366     experimental values.\cite{CRC80} The densities when using the SF
367     technique are close to, but typically lower than, those calculated
368 chrisfen 3064 using the Ewald summation. These slightly reduced densities indicate
369     that the pressure component from the image charges at R$_\textrm{c}$
370     is larger than that exerted by the reciprocal-space portion of the
371 gezelter 3066 Ewald summation. Bringing the image charges closer to the central
372 chrisfen 3064 particle by choosing a 9~\AA\ R$_\textrm{c}$ (rather than the
373     preferred 12~\AA\ R$_\textrm{c}$) increases the strength of the image
374     charge interactions on the central particle and results in a further
375     reduction of the densities.
376    
377     Because the strength of the image charge interactions has a noticeable
378     effect on the density, we would expect the use of electrostatic
379     damping to also play a role in these calculations. Larger values of
380     $\alpha$ weaken the pair-interactions; and since electrostatic damping
381     is distance-dependent, force components from the image charges will be
382     reduced more than those from particles close the the central
383     charge. This effect is visible in figure \ref{fig:t5peDensities} with
384 gezelter 3066 the damped SF sums showing slightly higher densities; however, it is
385     clear that the choice of cutoff radius plays a much more important
386     role in the resulting densities.
387 chrisfen 3064
388 gezelter 3066 All of the above density calculations were performed with systems of
389     512 water molecules. Rick observed a system size dependence of the
390     computed densities when using the Ewald summation, most likely due to
391     his tying of the convergence parameter to the box
392 chrisfen 3064 dimensions.\cite{Rick04} For systems of 256 water molecules, the
393     calculated densities were on average 0.002 to 0.003~g/cm$^3$ lower. A
394     system size of 256 molecules would force the use of a shorter
395 gezelter 3066 R$_\textrm{c}$ when using the SF technique, and this would also lower
396     the densities. Moving to larger systems, as long as the R$_\textrm{c}$
397     remains at a fixed value, we would expect the densities to remain
398     constant.
399 chrisfen 3064
400 gezelter 3066 \subsection{Liquid Structure of TIP5P-E}\label{sec:t5peLiqStructure}
401 chrisfen 3064
402     The experimentally determined $g_\textrm{OO}(r)$ for liquid water has
403     been compared in great detail with the various common water models,
404     and TIP5P was found to be in better agreement than other rigid,
405     non-polarizable models.\cite{Sorenson00} This excellent agreement with
406     experiment was maintained when Rick developed TIP5P-E.\cite{Rick04} To
407 gezelter 3066 check whether the choice of using the Ewald summation or the SF
408 chrisfen 3064 technique alters the liquid structure, the $g_\textrm{OO}(r)$s at
409     298~K and 1~atm were determined for the systems compared in the
410     previous section.
411    
412     \begin{figure}
413     \includegraphics[width=\linewidth]{./figures/tip5peGofR.pdf}
414     \caption{The $g_\textrm{OO}(r)$s calculated for TIP5P-E at 298~K and
415     1atm while using the Ewald summation (Ref. \cite{Rick04}) and the {\sc
416     sf} technique with varying parameters. Even with the reduced densities
417 gezelter 3066 using the SF technique, the $g_\textrm{OO}(r)$s are essentially
418 chrisfen 3064 identical.}
419     \label{fig:t5peGofRs}
420     \end{figure}
421     The $g_\textrm{OO}(r)$s calculated for TIP5P-E while using the {\sc
422     sf} technique with a various parameters are overlaid on the
423     $g_\textrm{OO}(r)$ while using the Ewald summation in
424     figure~\ref{fig:t5peGofRs}. The differences in density do not appear
425     to have any effect on the liquid structure as the $g_\textrm{OO}(r)$s
426 gezelter 3066 are indistinguishable. These results do indicate that
427 chrisfen 3064 $g_\textrm{OO}(r)$ is insensitive to the choice of electrostatic
428     correction.
429    
430 gezelter 3066 \subsection{Thermodynamic Properties of TIP5P-E}\label{sec:t5peThermo}
431 chrisfen 3064
432 gezelter 3066 In addition to the density and structual features of the liquid, there
433     are a variety of thermodynamic quantities that can be calculated for
434     water and compared directly to experimental values. Some of these
435     additional quantities include the latent heat of vaporization ($\Delta
436     H_\textrm{vap}$), the constant pressure heat capacity ($C_p$), the
437     isothermal compressibility ($\kappa_T$), the thermal expansivity
438     ($\alpha_p$), and the static dielectric constant ($\epsilon$). All of
439     these properties were calculated for TIP5P-E with the Ewald summation,
440     so they provide a good set for comparisons involving the SF technique.
441 chrisfen 3064
442     The $\Delta H_\textrm{vap}$ is the enthalpy change required to
443     transform one mole of substance from the liquid phase to the gas
444     phase.\cite{Berry00} In molecular simulations, this quantity can be
445     determined via
446     \begin{equation}
447     \begin{split}
448 gezelter 3066 \Delta H_\textrm{vap} &= H_\textrm{gas} - H_\textrm{liq} \\
449     &= E_\textrm{gas} - E_\textrm{liq}
450     + P(V_\textrm{gas} - V_\textrm{liq}) \\
451     &\approx -\frac{\langle U_\textrm{liq}\rangle}{N}+ RT,
452 chrisfen 3064 \end{split}
453     \label{eq:DeltaHVap}
454     \end{equation}
455 gezelter 3066 where $E$ is the total energy, $U$ is the potential energy, $P$ is the
456 chrisfen 3064 pressure, $V$ is the volume, $N$ is the number of molecules, $R$ is
457     the gas constant, and $T$ is the temperature.\cite{Jorgensen98b} As
458     seen in the last line of equation (\ref{eq:DeltaHVap}), we can
459     approximate $\Delta H_\textrm{vap}$ by using the ideal gas for the gas
460     state. This allows us to cancel the kinetic energy terms, leaving only
461 gezelter 3066 the $U_\textrm{liq}$ term. Additionally, the $pV$ term for the gas is
462 chrisfen 3064 several orders of magnitude larger than that of the liquid, so we can
463     neglect the liquid $pV$ term.
464    
465     The remaining thermodynamic properties can all be calculated from
466     fluctuations of the enthalpy, volume, and system dipole
467     moment.\cite{Allen87} $C_p$ can be calculated from fluctuations in the
468     enthalpy in constant pressure simulations via
469     \begin{equation}
470     \begin{split}
471 gezelter 3066 C_p = \left(\frac{\partial H}{\partial T}\right)_{N,P}
472 chrisfen 3064 = \frac{(\langle H^2\rangle - \langle H\rangle^2)}{Nk_BT^2},
473     \end{split}
474     \label{eq:Cp}
475     \end{equation}
476     where $k_B$ is Boltzmann's constant. $\kappa_T$ can be calculated via
477     \begin{equation}
478     \begin{split}
479     \kappa_T = \frac{1}{V}\left(\frac{\partial V}{\partial p}\right)_{N,T}
480 gezelter 3066 = \frac{(\langle V^2\rangle_{NPT} - \langle V\rangle^2_{NPT})}
481     {k_BT\langle V\rangle_{NPT}},
482 chrisfen 3064 \end{split}
483     \label{eq:kappa}
484     \end{equation}
485     and $\alpha_p$ can be calculated via
486     \begin{equation}
487     \begin{split}
488     \alpha_p = \frac{1}{V}\left(\frac{\partial V}{\partial T}\right)_{N,P}
489 gezelter 3066 = \frac{(\langle VH\rangle_{NPT}
490     - \langle V\rangle_{NPT}\langle H\rangle_{NPT})}
491     {k_BT^2\langle V\rangle_{NPT}}.
492 chrisfen 3064 \end{split}
493     \label{eq:alpha}
494     \end{equation}
495     Using the Ewald sum under tin-foil boundary conditions, $\epsilon$ can
496     be calculated for systems of non-polarizable substances via
497     \begin{equation}
498     \epsilon = 1 + \frac{\langle M^2\rangle}{3\epsilon_0k_\textrm{B}TV},
499     \label{eq:staticDielectric}
500     \end{equation}
501     where $\epsilon_0$ is the permittivity of free space and $\langle
502     M^2\rangle$ is the fluctuation of the system dipole
503     moment.\cite{Allen87} The numerator in the fractional term in equation
504     (\ref{eq:staticDielectric}) is the fluctuation of the simulation-box
505     dipole moment, identical to the quantity calculated in the
506     finite-system Kirkwood $g$ factor ($G_k$):
507     \begin{equation}
508     G_k = \frac{\langle M^2\rangle}{N\mu^2},
509     \label{eq:KirkwoodFactor}
510     \end{equation}
511     where $\mu$ is the dipole moment of a single molecule of the
512     homogeneous system.\cite{Neumann83,Neumann84,Neumann85} The
513     fluctuation term in both equation (\ref{eq:staticDielectric}) and
514     \ref{eq:KirkwoodFactor} is calculated as follows,
515     \begin{equation}
516     \begin{split}
517     \langle M^2\rangle &= \langle\bm{M}\cdot\bm{M}\rangle
518     - \langle\bm{M}\rangle\cdot\langle\bm{M}\rangle \\
519     &= \langle M_x^2+M_y^2+M_z^2\rangle
520     - (\langle M_x\rangle^2 + \langle M_x\rangle^2
521     + \langle M_x\rangle^2).
522     \end{split}
523     \label{eq:fluctBoxDipole}
524     \end{equation}
525     This fluctuation term can be accumulated during the simulation;
526     however, it converges rather slowly, thus requiring multi-nanosecond
527     simulation times.\cite{Horn04} In the case of tin-foil boundary
528     conditions, the dielectric/surface term of the Ewald summation is
529 gezelter 3066 equal to zero. Since the SF method also lacks this
530 chrisfen 3064 dielectric/surface term, equation (\ref{eq:staticDielectric}) is still
531     valid for determining static dielectric constants.
532    
533     All of the above properties were calculated from the same trajectories
534     used to determine the densities in section \ref{sec:t5peDensity}
535     except for the static dielectric constants. The $\epsilon$ values were
536     accumulated from 2~ns $NVE$ ensemble trajectories with system densities
537     fixed at the average values from the $NPT$ simulations at each of the
538     temperatures. The resulting values are displayed in figure
539     \ref{fig:t5peThermo}.
540     \begin{figure}
541     \centering
542     \includegraphics[width=4.5in]{./figures/t5peThermo.pdf}
543     \caption{Thermodynamic properties for TIP5P-E using the Ewald summation
544 gezelter 3066 and the SF techniques along with the experimental values. Units
545 chrisfen 3064 for the properties are kcal mol$^{-1}$ for $\Delta H_\textrm{vap}$,
546     cal mol$^{-1}$ K$^{-1}$ for $C_p$, 10$^6$ atm$^{-1}$ for $\kappa_T$,
547     and 10$^5$ K$^{-1}$ for $\alpha_p$. Ewald summation results are from
548     reference \cite{Rick04}. Experimental values for $\Delta
549     H_\textrm{vap}$, $\kappa_T$, and $\alpha_p$ are from reference
550     \cite{Kell75}. Experimental values for $C_p$ are from reference
551     \cite{Wagner02}. Experimental values for $\epsilon$ are from reference
552     \cite{Malmberg56}.}
553     \label{fig:t5peThermo}
554     \end{figure}
555    
556 gezelter 3066 For all of the properties computed, the trends with temperature
557     obtained when using the Ewald summation are reproduced with the SF
558     technique. One noticeable difference between the properties calculated
559     using the two methods are the lower $\Delta H_\textrm{vap}$ values
560     when using SF. This is to be expected due to the direct weakening of
561     the electrostatic interaction through forced neutralization. This
562     results in an increase of the intermolecular potential producing lower
563     values from equation (\ref{eq:DeltaHVap}). The slopes of these values
564     with temperature are similar to that seen using the Ewald summation;
565     however, they are both steeper than the experimental trend, indirectly
566     resulting in the inflated $C_p$ values at all temperatures.
567 chrisfen 3064
568     Above the supercooled regime, $C_p$, $\kappa_T$, and $\alpha_p$ values
569     all overlap within error. As indicated for the $\Delta H_\textrm{vap}$
570 gezelter 3066 and $C_p$ results, the deviations between experiment and simulation in
571     this region are not the fault of the electrostatic summation methods
572     but are due to the geometry and parameters of the TIP5P class of water
573     models. Like most rigid, non-polarizable, point-charge water models,
574     the density decreases with temperature at a much faster rate than
575     experiment (see figure \ref{fig:t5peDensities}). This reduced density
576     leads to the inflated compressibility and expansivity values at higher
577     temperatures seen here in figure \ref{fig:t5peThermo}. Incorporation
578     of polarizability and many-body effects are required in order for
579     water models to overcome differences between simulation-based and
580     experimentally determined densities at these higher
581 chrisfen 3064 temperatures.\cite{Laasonen93,Donchev06}
582    
583     At temperatures below the freezing point for experimental water, the
584 gezelter 3066 differences between SF and the Ewald summation results are more
585 chrisfen 3064 apparent. The larger $C_p$ and lower $\alpha_p$ values in this region
586     indicate a more pronounced transition in the supercooled regime,
587 gezelter 3066 particularly in the case of SF without damping.
588 chrisfen 3064
589 gezelter 3066 This points to the onset of a more frustrated or glassy behavior for
590     the undamped and weakly-damped SF simulations of TIP5P-E at
591     temperatures below 250~K than is seen from the Ewald sum. Undamped SF
592     electrostatics has a stronger contribution from nearby charges.
593     Damping these local interactions or using a reciprocal-space method
594     makes the water less sensitive to ordering on a short length scale.
595     We can recover nearly quantitative agreement with the Ewald results by
596     increasing the damping constant.
597    
598 chrisfen 3064 The final thermodynamic property displayed in figure
599     \ref{fig:t5peThermo}, $\epsilon$, shows the greatest discrepancy
600 gezelter 3066 between the Ewald and SF methods (and with experiment). It is known
601     that the dielectric constant is dependent upon and is quite sensitive
602     to the imposed boundary conditions.\cite{Neumann80,Neumann83} This is
603     readily apparent in the converged $\epsilon$ values accumulated for
604     the SF simulations. Lack of a damping function results in dielectric
605 chrisfen 3064 constants significantly smaller than those obtained using the Ewald
606     sum. Increasing the damping coefficient to 0.2~\AA$^{-1}$ improves the
607     agreement considerably. It should be noted that the choice of the
608 gezelter 3066 ``Ewald coefficient'' ($\kappa$) and real-space cutoff values also
609     have a significant effect on the calculated static dielectric constant
610     when using the Ewald summation. In the simulations of TIP5P-E with the
611     Ewald sum, this screening parameter was tethered to the simulation box
612     size (as was the $R_\textrm{c}$).\cite{Rick04} In general, systems
613     with larger screening parameters reported larger dielectric constant
614     values, the same behavior we see here with {\sc sf}; however, the
615     choice of cutoff radius also plays an important role.
616 chrisfen 3064
617 gezelter 3066 \subsubsection{Dielectric Constants for TIP5P-E and SSD/RF}\label{sec:t5peDielectric}
618 chrisfen 3064
619 gezelter 3066 In the previous section, we observed that the choice of damping
620     coefficient plays a major role in the calculated dielectric constant
621     for the SF method. Similar damping parameter behavior was observed in
622     the long-time correlated motions of the NaCl crystal.\cite{Fennell06}
623     The static dielectric constant is calculated from the long-time
624     fluctuations of the system's accumulated dipole moment
625     (Eq. (\ref{eq:staticDielectric})), so it is quite sensitive to the
626     choice of damping parameter. Since $\alpha$ is an adjustable
627     parameter, it would be best to choose optimal damping constants such
628     that any arbitrary choice of cutoff radius will properly capture the
629     dielectric behavior of the liquid.
630 chrisfen 3064
631     In order to find these optimal values, we mapped out the static
632     dielectric constant as a function of both the damping parameter and
633 gezelter 3066 cutoff radius for TIP5P-E and for a point-dipolar water model
634     (SSD/RF). To calculate the static dielectric constant, we performed
635     5~ns $NPT$ calculations on systems of 512 water molecules and averaged
636     over the converged region (typically the later 2.5~ns) of equation
637     (\ref{eq:staticDielectric}). The selected cutoff radii ranged from 9,
638     10, 11, and 12~\AA , and the damping parameter values ranged from 0.1
639     to 0.35~\AA$^{-1}$.
640 chrisfen 3064
641     \begin{table}
642     \centering
643 gezelter 3066 \caption{Static dielectric constants for the TIP5P-E and SSD/RF water models at 298~K and 1~atm as a function of damping coefficient $\alpha$ and
644     cutoff radius $R_\textrm{c}$. The color scale ranges from blue (10) to red (100).}
645 chrisfen 3064 \vspace{6pt}
646 gezelter 3066 \begin{tabular}{ lccccccccc }
647 chrisfen 3064 \toprule
648     \toprule
649 gezelter 3066 & \multicolumn{4}{c}{TIP5P-E} & & \multicolumn{4}{c}{SSD/RF} \\
650     & \multicolumn{4}{c}{$R_\textrm{c}$ (\AA )} & & \multicolumn{4}{c}{$R_\textrm{c}$ (\AA )} \\
651     \cmidrule(lr){2-5} \cmidrule(lr){7-10}
652     $\alpha$ (\AA$^{-1}$) & 9 & 10 & 11 & 12 & & 9 & 10 & 11 & 12 \\
653 chrisfen 3064 \midrule
654 gezelter 3066 0.35 & \cellcolor[rgb]{1, 0.788888888888889, 0.5} 87.0 & \cellcolor[rgb]{1, 0.695555555555555, 0.5} 91.2 & \cellcolor[rgb]{1, 0.717777777777778, 0.5} 90.2 & \cellcolor[rgb]{1, 0.686666666666667, 0.5} 91.6 & & \cellcolor[rgb]{1, 0.5, 0.5} 116 & \cellcolor[rgb]{1, 0.5, 0.5} 119.2 & \cellcolor[rgb]{1, 0.5, 0.5} 131.4 & \cellcolor[rgb]{1, 0.5, 0.5} 130 \\
655     & \cellcolor[rgb]{1, 0.892222222222222, 0.5} & \cellcolor[rgb]{1, 0.704444444444444, 0.5} & \cellcolor[rgb]{1, 0.72, 0.5} & \cellcolor[rgb]{1, 0.6666666666667, 0.5} & & \cellcolor[rgb]{1, 0.5, 0.5} & \cellcolor[rgb]{1, 0.5, 0.5} & \cellcolor[rgb]{1, 0.5, 0.5} & \cellcolor[rgb]{1, 0.5, 0.5} \\
656     0.3 & \cellcolor[rgb]{1, 0.995555555555556, 0.5} 77.7 & \cellcolor[rgb]{1, 0.713333333333333, 0.5} 90.4 & \cellcolor[rgb]{1, 0.713333333333333, 0.5} 90.4 & \cellcolor[rgb]{1, 0.646666666666667, 0.5} 93.4 & & \cellcolor[rgb]{1, 0.5, 0.5} 100 & \cellcolor[rgb]{1, 0.5, 0.5} 118.8 & \cellcolor[rgb]{1, 0.5, 0.5} 116 & \cellcolor[rgb]{1, 0.5, 0.5} 122 \\
657     0.275 & \cellcolor[rgb]{0.653333333333333, 1, 0.5} 61.9 & \cellcolor[rgb]{1, 0.933333333333333, 0.5} 80.5 & \cellcolor[rgb]{1, 0.811111111111111, 0.5} 86.0 & \cellcolor[rgb]{1, 0.766666666666667, 0.5} 88 & & \cellcolor[rgb]{1, 1, 0.5} 77.5 & \cellcolor[rgb]{1, 0.5, 0.5} 105 & \cellcolor[rgb]{1, 0.5, 0.5} 118 & \cellcolor[rgb]{1, 0.5, 0.5} 125.2 \\
658     0.25 & \cellcolor[rgb]{0.537777777777778, 1, 0.5} 56.7 & \cellcolor[rgb]{0.795555555555555, 1, 0.5} 68.3 & \cellcolor[rgb]{1, 0.966666666666667, 0.5} 79 & \cellcolor[rgb]{1, 0.704444444444445, 0.5} 90.8 & & \cellcolor[rgb]{0.5, 1, 0.582222222222222} 51.3 & \cellcolor[rgb]{1, 0.993333333333333, 0.5} 77.8 & \cellcolor[rgb]{1, 0.522222222222222, 0.5} 99 & \cellcolor[rgb]{1, 0.5, 0.5} 113 \\
659     0.225 & \cellcolor[rgb]{0.5, 1, 0.768888888888889} 42.9 & \cellcolor[rgb]{0.566666666666667, 1, 0.5} 58.0 & \cellcolor[rgb]{0.693333333333333, 1, 0.5} 63.7 & \cellcolor[rgb]{1, 0.937777777777778, 0.5} 80.3 & & \cellcolor[rgb]{0.5, 0.984444444444444, 1} 31.8 & \cellcolor[rgb]{0.5, 1, 0.586666666666667} 51.1 & \cellcolor[rgb]{1, 0.995555555555556, 0.5} 77.7 & \cellcolor[rgb]{1, 0.5, 0.5} 108.1 \\
660     0.2 & \cellcolor[rgb]{0.5, 0.973333333333333, 1} 31.3 & \cellcolor[rgb]{0.5, 1, 0.842222222222222} 39.6 & \cellcolor[rgb]{0.54, 1, 0.5} 56.8 & \cellcolor[rgb]{0.735555555555555, 1, 0.5} 65.6 & & \cellcolor[rgb]{0.5, 0.698666666666667, 1} 18.94 & \cellcolor[rgb]{0.5, 0.946666666666667, 1} 30.1 & \cellcolor[rgb]{0.5, 1, 0.704444444444445} 45.8 & \cellcolor[rgb]{0.893333333333333, 1, 0.5} 72.7 \\
661     & \cellcolor[rgb]{0.5, 0.848888888888889, 1} & \cellcolor[rgb]{0.5, 0.973333333333333, 1} & \cellcolor[rgb]{0.5, 1, 0.793333333333333} & \cellcolor[rgb]{0.5, 1, 0.624444444444445} & & \cellcolor[rgb]{0.5, 0.599333333333333, 1} & \cellcolor[rgb]{0.5, 0.732666666666667, 1} & \cellcolor[rgb]{0.5, 0.942111111111111, 1} & \cellcolor[rgb]{0.5, 1, 0.695555555555556} \\
662     0.15 & \cellcolor[rgb]{0.5, 0.724444444444444, 1} 20.1 & \cellcolor[rgb]{0.5, 0.788888888888889, 1} 23.0 & \cellcolor[rgb]{0.5, 0.873333333333333, 1} 26.8 & \cellcolor[rgb]{0.5, 1, 0.984444444444444} 33.2 & & \cellcolor[rgb]{0.5, 0.5, 1} 8.29 & \cellcolor[rgb]{0.5, 0.518666666666667, 1} 10.84 & \cellcolor[rgb]{0.5, 0.588666666666667, 1} 13.99 & \cellcolor[rgb]{0.5, 0.715555555555556, 1} 19.7 \\
663     & \cellcolor[rgb]{0.5, 0.696111111111111, 1} & \cellcolor[rgb]{0.5, 0.736333333333333, 1} & \cellcolor[rgb]{0.5, 0.775222222222222, 1} & \cellcolor[rgb]{0.5, 0.860666666666667, 1} & & \cellcolor[rgb]{0.5, 0.5, 1} & \cellcolor[rgb]{0.5, 0.509333333333333, 1} & \cellcolor[rgb]{0.5, 0.544333333333333, 1} & \cellcolor[rgb]{0.5, 0.607777777777778, 1} \\
664     0.1 & \cellcolor[rgb]{0.5, 0.667777777777778, 1} 17.55 & \cellcolor[rgb]{0.5, 0.683777777777778, 1} 18.27 & \cellcolor[rgb]{0.5, 0.677111111111111, 1} 17.97 & \cellcolor[rgb]{0.5, 0.705777777777778, 1} 19.26 & & \cellcolor[rgb]{0.5, 0.5, 1} 4.96 & \cellcolor[rgb]{0.5, 0.5, 1} 5.46 & \cellcolor[rgb]{0.5, 0.5, 1} 6.04 & \cellcolor[rgb]{0.5,0.5, 1} 6.82 \\
665 chrisfen 3064 \bottomrule
666     \end{tabular}
667 gezelter 3066 \label{tab:DielectricMap}
668 chrisfen 3064 \end{table}
669 gezelter 3066
670 chrisfen 3064 The results of these calculations are displayed in table
671 gezelter 3066 \ref{tab:DielectricMap}. The dielectric constants for both models
672     decrease linearly with increasing cutoff radii ($R_\textrm{c}$) and
673     increase linearly with increasing damping ($\alpha$). Another point
674     to note is that choosing $\alpha$ and $R_\textrm{c}$ identical to
675     those used with the Ewald summation results in the same calculated
676     dielectric constant. As an example, in the paper outlining the
677     development of TIP5P-E, the real-space cutoff and Ewald coefficient
678     were tethered to the system size, and for a 512 molecule system are
679     approximately 12~\AA\ and 0.25~\AA$^{-1}$ respectively.\cite{Rick04}
680     These parameters resulted in a dielectric constant of 92$\pm$14, while
681     with SF these parameters give a dielectric constant of
682 chrisfen 3064 90.8$\pm$0.9. Another example comes from the TIP4P-Ew paper where
683     $\alpha$ and $R_\textrm{c}$ were chosen to be 9.5~\AA\ and
684     0.35~\AA$^{-1}$, and these parameters resulted in a dielectric
685 gezelter 3066 constant equal to 63$\pm$1.\cite{Horn04} Calculations using SF with
686     these parameters and this water model give a dielectric constant of
687     61$\pm$1. Since the dielectric constant is dependent on $\alpha$ and
688     $R_\textrm{c}$ with the SF technique, it might be interesting to
689     investigate the dielectric dependence of the real-space Ewald
690     parameters.
691 chrisfen 3064
692 gezelter 3066 We have also mapped out the static dielectric constant of SSD/RF as a
693     function of $R_\textrm{c}$ and $\alpha$. It is apparent from this
694     table that electrostatic damping has a more pronounced effect on the
695     dipolar interactions of SSD/RF than the monopolar interactions of
696     TIP5P-E. The dielectric constant covers a much wider range and has a
697     steeper slope with increasing damping parameter.
698    
699     Although it is tempting to choose damping parameters equivalent to the
700     Ewald examples, the results of our previous work indicate that values
701     this high are destructive to both the energetics and
702 chrisfen 3064 dynamics.\cite{Fennell06} Ideally, $\alpha$ should not exceed
703     0.3~\AA$^{-1}$ for any of the cutoff values in this range. If the
704     optimal damping parameter is chosen to be midway between 0.275 and
705     0.3~\AA$^{-1}$ (0.2875~\AA$^{-1}$) at the 9~\AA\ cutoff, then the
706     linear trend with $R_\textrm{c}$ will always keep $\alpha$ below
707     0.3~\AA$^{-1}$ for the studied cutoff radii. This linear progression
708     would give values of 0.2875, 0.2625, 0.2375, and 0.2125~\AA$^{-1}$ for
709     cutoff radii of 9, 10, 11, and 12~\AA. Setting this to be the default
710 gezelter 3066 behavior for the damped SF technique will result in consistent
711 chrisfen 3064 dielectric behavior for these and other condensed molecular systems,
712     regardless of the chosen cutoff radius. The static dielectric
713     constants for TIP5P-E simulations with 9 and 12\AA\ $R_\textrm{c}$
714     values using their respective damping parameters are 76$\pm$1 and
715     75$\pm$2. These values are lower than the values reported for TIP5P-E
716     with the Ewald sum; however, they are more in line with the values
717     reported by Mahoney {\it et al.} for TIP5P while using a reaction
718     field (RF) with an infinite RF dielectric constant
719 gezelter 3066 (81.5$\pm$1.6).\cite{Mahoney00}
720    
721     We can use the same trend of $\alpha$ with $R_\textrm{c}$ for SSD/RF
722     and for a 12~\AA\ $R_\textrm{c}$, the resulting dielectric constant is
723     82.6$\pm$0.6. This value compares very favorably with the experimental
724     value of 78.3.\cite{Malmberg56} This is not surprising given that
725     early studies of the SSD model indicate a static dielectric constant
726     around 81.\cite{Liu96} The static dielectric constants for SSD/RF
727     simulations with 9 and 12\AA\ $R_\textrm{c}$ values using their
728     respective damping parameters are 82.6$\pm$0.6 and 75$\pm$2.
729    
730     As a final note, aside from a slight
731 chrisfen 3064 lowering of $\Delta H_\textrm{vap}$, using these $\alpha$ values does
732     not alter the other other thermodynamic properties.
733    
734     \subsubsection{Dynamic Properties}\label{sec:t5peDynamics}
735    
736 gezelter 3066 To look at the dynamic properties of TIP5P-E when using the SF
737 chrisfen 3064 method, 200~ps $NVE$ simulations were performed for each temperature
738     at the average density reported by the $NPT$ simulations using 9 and
739     12~\AA\ $R_\textrm{c}$ values using the ideal $\alpha$ values
740     determined above (0.2875 and 0.2125~\AA$^{-1}$). The self-diffusion
741     constants ($D$) were calculated using the mean square displacement
742     (MSD) form of the Einstein relation,
743     \begin{equation}
744     D = \lim_{t\rightarrow\infty}
745     \frac{\langle\left|\mathbf{r}_i(t)-\mathbf{r}_i(0)\right|^2\rangle}{6t},
746     \label{eq:MSD}
747     \end{equation}
748     where $t$ is time, and $\mathbf{r}_i$ is the position of particle
749     $i$.\cite{Allen87}
750    
751     \begin{figure}
752     \centering
753     \includegraphics[width=3.5in]{./figures/waterFrame.pdf}
754     \caption{Body-fixed coordinate frame for a water molecule. The
755     respective molecular principle axes point in the direction of the
756     labeled frame axes.}
757     \label{fig:waterFrame}
758     \end{figure}
759     In addition to translational diffusion, orientational relaxation times
760     were calculated for comparisons with the Ewald simulations and with
761     experiments. These values were determined from the same 200~ps $NVE$
762     trajectories used for translational diffusion by calculating the
763     orientational time correlation function,
764     \begin{equation}
765     C_l^\alpha(t) = \left\langle P_l\left[\hat{\mathbf{u}}_i^\alpha(t)
766     \cdot\hat{\mathbf{u}}_i^\alpha(0)\right]\right\rangle,
767     \label{eq:OrientCorr}
768     \end{equation}
769     where $P_l$ is the Legendre polynomial of order $l$ and
770     $\hat{\mathbf{u}}_i^\alpha$ is the unit vector of molecule $i$ along
771     principle axis $\alpha$. The principle axis frame for these water
772     molecules is shown in figure \ref{fig:waterFrame}. As an example,
773     $C_l^y$ is calculated from the time evolution of the unit vector
774     connecting the two hydrogen atoms.
775    
776     \begin{figure}
777     \centering
778     \includegraphics[width=3.5in]{./figures/exampleOrientCorr.pdf}
779     \caption{Example plots of the orientational autocorrelation functions
780     for the first and second Legendre polynomials. These curves show the
781     time decay of the unit vector along the $y$ principle axis.}
782     \label{fig:OrientCorr}
783     \end{figure}
784     From the orientation autocorrelation functions, we can obtain time
785     constants for rotational relaxation. Figure \ref{fig:OrientCorr} shows
786     some example plots of orientational autocorrelation functions for the
787     first and second Legendre polynomials. The relatively short time
788     portions (between 1 and 3~ps for water) of these curves can be fit to
789     an exponential decay to obtain these constants, and they are directly
790     comparable to water orientational relaxation times from nuclear
791     magnetic resonance (NMR). The relaxation constant obtained from
792     $C_2^y(t)$ is of particular interest because it describes the
793     relaxation of the principle axis connecting the hydrogen atoms. Thus,
794     $C_2^y(t)$ can be compared to the intermolecular portion of the
795     dipole-dipole relaxation from a proton NMR signal and should provide
796     the best estimate of the NMR relaxation time constant.\cite{Impey82}
797    
798     \begin{figure}
799     \centering
800     \includegraphics[width=3.5in]{./figures/t5peDynamics.pdf}
801     \caption{Diffusion constants ({\it upper}) and reorientational time
802 gezelter 3066 constants ({\it lower}) for TIP5P-E using the Ewald sum and SF
803 chrisfen 3064 technique compared with experiment. Data at temperatures less than
804     0$^\circ$C were not included in the $\tau_2^y$ plot to allow for
805     easier comparisons in the more relevant temperature regime.}
806     \label{fig:t5peDynamics}
807     \end{figure}
808     Results for the diffusion constants and orientational relaxation times
809     are shown in figure \ref{fig:t5peDynamics}. From this figure, it is
810     apparent that the trends for both $D$ and $\tau_2^y$ of TIP5P-E using
811 gezelter 3066 the Ewald sum are reproduced with the SF technique. The enhanced
812 chrisfen 3064 diffusion at high temperatures are again the product of the lower
813     densities in comparison with experiment and do not provide any special
814     insight into differences between the electrostatic summation
815 gezelter 3066 techniques. With the undamped SF technique, TIP5P-E tends to
816 chrisfen 3064 diffuse a little faster than with the Ewald sum; however, use of light
817     to moderate damping results in indistinguishable $D$ values. Though
818 gezelter 3066 not apparent in this figure, SF values at the lowest temperature
819 chrisfen 3064 are approximately twice as slow as $D$ values obtained using the Ewald
820     sum. These values support the observation from section
821     \ref{sec:t5peThermo} that there appeared to be a change to a more
822 gezelter 3066 glassy-like phase with the SF technique at these lower
823 chrisfen 3064 temperatures, though this change seems to be more prominent with the
824 gezelter 3066 {\it undamped} SF method, which has stronger local pairwise
825 chrisfen 3064 electrostatic interactions.
826    
827     The $\tau_2^y$ results in the lower frame of figure
828     \ref{fig:t5peDynamics} show a much greater difference between the {\sc
829     sf} results and the Ewald results. At all temperatures shown, TIP5P-E
830     relaxes faster than experiment with the Ewald sum while tracking
831 gezelter 3066 experiment fairly well when using the SF technique, independent
832 chrisfen 3064 of the choice of damping constant. Their are several possible reasons
833     for this deviation between techniques. The Ewald results were
834 gezelter 3066 calculated using shorter (10ps) trajectories than the SF results
835     (25ps). A quick calculation from a 10~ps trajectory with SF with
836 chrisfen 3064 an $\alpha$ of 0.2~\AA$^{-1}$ at 25$^\circ$C showed a 0.4~ps drop in
837     $\tau_2^y$, placing the result more in line with that obtained using
838     the Ewald sum. This example supports this explanation; however,
839     recomputing the results to meet a poorer statistical standard is
840     counter-productive. Assuming the Ewald results are not entirely the
841     product of poor statistics, differences in techniques to integrate the
842     orientational motion could also play a role. {\sc shake} is the most
843     commonly used technique for approximating rigid-body orientational
844     motion,\cite{Ryckaert77} whereas in {\sc oopse}, we maintain and
845     integrate the entire rotation matrix using the {\sc dlm}
846     method.\cite{Meineke05} Since {\sc shake} is an iterative constraint
847     technique, if the convergence tolerances are raised for increased
848     performance, error will accumulate in the orientational
849     motion. Finally, the Ewald results were calculated using the $NVT$
850 gezelter 3066 ensemble, while the $NVE$ ensemble was used for SF
851 chrisfen 3064 calculations. The additional mode of motion due to the thermostat will
852     alter the dynamics, resulting in differences between $NVT$ and $NVE$
853     results. These differences are increasingly noticeable as the
854     thermostat time constant decreases.
855    
856    
857     \subsection{SSD/RF}
858    
859     In section \ref{sec:dampingMultipoles}, we described a method for
860 gezelter 3066 applying the damped SF technique to systems containing point
861 chrisfen 3064 multipoles. The soft sticky dipole (SSD) family of water models is the
862     perfect test case because of the dipole-dipole (and
863     charge-dipole/quadrupole) interactions that are present. As alluded to
864     in the name, soft sticky dipole water models are single point
865     molecules that consist of a ``soft'' Lennard-Jones sphere, a
866     point-dipole, and a tetrahedral function for capturing the hydrogen
867     bonding nature of water - a spherical harmonic term for water-water
868     tetrahedral interactions and a point-quadrupole for interactions with
869     surrounding charges. Detailed descriptions of these models can be
870     found in other studies.\cite{Liu96b,Chandra99,Tan03,Fennell04}
871    
872     In deciding which version of the SSD model to use, we need only
873 gezelter 3066 consider that the SF technique was presented as a pairwise
874 chrisfen 3064 replacement for the Ewald summation. It has been suggested that models
875     parametrized for the Ewald summation (like TIP5P-E) would be
876     appropriate for use with a reaction field and vice versa.\cite{Rick04}
877     Therefore, we decided to test the SSD/RF water model, which was
878     parametrized for use with a reaction field, with this damped
879     electrostatic technique to see how the calculated properties change.
880    
881     \subsubsection{Dipolar Damping}
882    
883     \begin{table}
884     \caption{Properties of SSD/RF when using different electrostatic
885     correction methods.}
886     \footnotesize
887     \centering
888     \begin{tabular}{ llccc }
889     \toprule
890     \toprule
891     & & Reaction Field [Ref. \citen{Fennell04}] & Damped Electrostatics & Experiment [Ref.] \\
892     & & $\epsilon = 80$ & $R_\textrm{c} = 12$\AA ; $\alpha = 0.2125$~\AA$^{-1}$ & \\
893     \midrule
894     $\rho$ & (g cm$^{-3}$) & 0.997(1) & 1.004(4) & 0.997 [\citen{CRC80}]\\
895     $C_p$ & (cal mol$^{-1}$ K$^{-1}$) & 23.8(2) & 27(1) & 18.005 [\citen{Wagner02}] \\
896     $D$ & ($10^{-5}$ cm$^2$ s$^{-1}$) & 2.32(6) & 2.33(2) & 2.299 [\citen{Mills73}]\\
897     $n_C$ & & 4.4 & 4.2 & 4.7 [\citen{Hura00}]\\
898     $n_H$ & & 3.7 & 3.7 & 3.5 [\citen{Soper86}]\\
899     $\tau_1$ & (ps) & 7.2(4) & 5.86(8) & 5.7 [\citen{Eisenberg69}]\\
900     $\tau_2$ & (ps) & 3.2(2) & 2.45(7) & 2.3 [\citen{Krynicki66}]\\
901     \bottomrule
902     \end{tabular}
903     \label{tab:dampedSSDRF}
904     \end{table}
905     The properties shown in table \ref{tab:dampedSSDRF} compare
906     quite well. The average density shows a modest increase when
907     using damped electrostatics in place of the reaction field. This comes
908     about because we neglect the pressure effect due to the surroundings
909     outside of the cutoff, instead relying on screening effects to
910     neutralize electrostatic interactions at long distances. The $C_p$
911     also shows a slight increase, indicating greater fluctuation in the
912     enthalpy at constant pressure. The only other differences between the
913     damped and reaction field results are the dipole reorientational time
914     constants, $\tau_1$ and $\tau_2$. When using damped electrostatics,
915     the water molecules relax more quickly and exhibit behavior very
916     similar to the experimental values. These results indicate that not
917     only is it reasonable to use damped electrostatics with SSD/RF, it is
918     recommended if capturing realistic dynamics is of primary
919     importance. This is an encouraging result because the damping methods
920     are more generally applicable than reaction field. Using damping,
921     SSD/RF can be used effectively with mixed systems, such as dissolved
922     ions, dissolved organic molecules, or even proteins.
923    
924     \section{Application of Pairwise Electrostatic Corrections: Imaginary Ice}
925    
926     In an earlier work, we performed a series of free energy calculations
927     on several ice polymorphs which are stable or metastable at low
928     pressures, one of which (Ice-$i$) we observed in spontaneous
929     crystallizations of an SSD type single point water
930     model.\cite{Fennell05} In this study, a distinct dependence of the
931     free energies on the interaction cutoff and correction technique was
932 gezelter 3066 observed. Being that the SF technique can be used as a simple
933 chrisfen 3064 and efficient replacement for the Ewald summation, it would be
934     interesting to apply it in addressing the question of the free
935     energies of these ice polymorphs with varying water models. To this
936     end, we set up thermodynamic integrations of all of the previously
937 gezelter 3066 discussed ice polymorphs using the SF technique with a cutoff
938 chrisfen 3064 radius of 12~\AA\ and an $\alpha$ of 0.2125~\AA . These calculations
939     were performed on TIP5P-E and TIP4P-Ew (variants of the root models
940     optimized for the Ewald summation) as well as SPC/E, and SSD/RF.
941    
942     \begin{table}
943     \centering
944     \caption{Helmholtz free energies of ice polymorphs at 1~atm and 200~K
945 gezelter 3066 using the damped SF electrostatic correction method with a
946 chrisfen 3064 variety of water models.}
947     \begin{tabular}{ lccccc }
948     \toprule
949     \toprule
950     Model & I$_\textrm{h}$ & I$_\textrm{c}$ & B & Ice-$i$ & Ice-$i^\prime$ \\
951     \cmidrule(lr){2-6}
952     & \multicolumn{5}{c}{(kcal mol$^{-1}$)} \\
953     \midrule
954     TIP5P-E & -11.98(4) & -11.96(4) & -11.87(3) & - & -11.95(3) \\
955     TIP4P-Ew & -13.11(3) & -13.09(3) & -12.97(3) & - & -12.98(3) \\
956     SPC/E & -12.99(3) & -13.00(3) & -13.03(3) & - & -12.99(3) \\
957     SSD/RF & -11.83(3) & -11.66(4) & -12.32(3) & -12.39(3) & - \\
958     \bottomrule
959     \end{tabular}
960     \label{tab:dampedFreeEnergy}
961     \end{table}
962     The results of these calculations in table \ref{tab:dampedFreeEnergy}
963     show similar behavior to the Ewald results in the previous study, at
964     least for SSD/RF and SPC/E which are present in both.\cite{Fennell05}
965     The Helmholtz free energies of the ice polymorphs with SSD/RF order in
966     the same fashion, with Ice-$i$ having the lowest free energy; however,
967     the Ice-$i$ and ice B free energies are quite a bit closer (nearly
968     isoenergetic). The SPC/E results show the near isoenergetic behavior
969     when using the Ewald summation.\cite{Fennell05} Ice B has the lowest
970     Helmholtz free energy; however, all the polymorph results overlap
971     within error.
972    
973     The most interesting results from these calculations come from the
974     more expensive TIP4P-Ew and TIP5P-E results. Both of these models were
975     optimized for use with an electrostatic correction and are
976     geometrically arranged to mimic water following two different
977     ideas. In TIP5P-E, the primary location for the negative charge in the
978     molecule is assigned to the lone-pairs of the oxygen, while TIP4P-Ew
979     places the negative charge near the center-of-mass along the H-O-H
980     bisector. There is some debate as to which is the proper choice for
981     the negative charge location, and this has in part led to a six-site
982     water model that balances both of these options.\cite{Vega05,Nada03}
983     The limited results in table \ref{tab:dampedFreeEnergy} support the
984     results of Vega {\it et al.}, which indicate the TIP4P charge location
985     geometry is more physically valid.\cite{Vega05} With the TIP4P-Ew
986     water model, the experimentally observed polymorph (ice
987     I$_\textrm{h}$) is the preferred form with ice I$_\textrm{c}$ slightly
988     higher in energy, though overlapping within error. Additionally, the
989     less realistic ice B and Ice-$i^\prime$ structures are destabilized
990     relative to these polymorphs. TIP5P-E shows similar behavior to SPC/E,
991     where there is no real free energy distinction between the various
992     polymorphs, because many overlap within error. While ice B is close in
993     free energy to the other polymorphs, these results fail to support the
994     findings of other researchers indicating the preferred form of TIP5P
995     at 1~atm is a structure similar to ice
996     B.\cite{Yamada02,Vega05,Abascal05} It should be noted that we are
997     looking at TIP5P-E rather than TIP5P, and the differences in the
998     Lennard-Jones parameters could be a reason for this dissimilarity.
999     Overall, these results indicate that TIP4P-Ew is a better mimic of
1000     real water than these other models when studying crystallization and
1001     solid forms of water.
1002    
1003     \section{Conclusions}
1004    
1005     \section{Acknowledgments}
1006     Support for this project was provided by the National Science
1007     Foundation under grant CHE-0134881. Computation time was provided by
1008     the Notre Dame Center for Research Computing.
1009    
1010     \newpage
1011    
1012     \bibliographystyle{achemso}
1013     \bibliography{multipoleSFPaper}
1014    
1015    
1016     \end{document}