ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/group/trunk/multipoleSFPaper/multipoleSFPaper.tex
Revision: 3065
Committed: Tue Oct 24 06:27:31 2006 UTC (18 years, 9 months ago) by chrisfen
Content type: application/x-tex
File size: 56789 byte(s)
Log Message:
fixed the bib file and some references

File Contents

# Content
1 %\documentclass[prb,aps,twocolumn,tabularx]{revtex4}
2 \documentclass[12pt]{article}
3 \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
24 \begin{document}
25
26 \title{Pairwise electrostatic corrections for monopolar and multipolar systems}
27
28 \author{Christopher J. Fennell and J. Daniel Gezelter \\
29 Department of Chemistry and Biochemistry\\ University of Notre Dame\\
30 Notre Dame, Indiana 46556}
31
32 \date{\today}
33
34 \maketitle
35 %\doublespacing
36
37 \begin{abstract}
38 \end{abstract}
39
40 %\narrowtext
41
42 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
43 % BODY OF TEXT
44 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
45
46 \section{Introduction}
47
48 Over the past several years, there has been increased interest in
49 pairwise methods for correcting electrostatic interaction in computer
50 simulations of condensed molecular
51 systems.\cite{Wolf99,Zahn02,Kast03,Fennell06}
52
53 \section{Shifted Force Pairwise Electrostatic Correction}
54
55 In our previous look at pairwise electrostatic correction methods, we
56 described the undamped and damped shifted potential {\sc sp} and
57 shifted force {\sc sf} techniques.\cite{Fennell06} These techniques
58 were developed from the observations and efforts of Wolf {\it et al.}
59 and their work towards an $\mathcal{O}(N)$ Coulombic sum.\cite{Wolf99}
60 The potential for the damped form of the {\sc sp} method is given by
61 \begin{equation}
62 V_{\textrm{DSP}}(r) = q_iq_j\left(\frac{\textrm{erfc}\left(\alpha r\right)}{r}
63 - \frac{\textrm{erfc}\left(\alpha R_\textrm{c}\right)}{R_\textrm{c}}\right)
64 \quad r\leqslant R_\textrm{c},
65 \label{eq:DSPPot}
66 \end{equation}
67 while the damped form of the {\sc sf} method is given by
68 \begin{equation}
69 \begin{split}
70 V_\mathrm{DSF}(r) = q_iq_j\Biggr{[}&
71 \frac{\mathrm{erfc}\left(\alpha r\right)}{r}
72 - \frac{\mathrm{erfc}\left(\alpha R_\mathrm{c}\right)}{R_\mathrm{c}} \\
73 &+ \left(\frac{\mathrm{erfc}\left(\alpha R_\mathrm{c}\right)}{R_\mathrm{c}^2}
74 + \frac{2\alpha}{\pi^{1/2}}
75 \frac{\exp\left(-\alpha^2R_\mathrm{c}^2\right)}{R_\mathrm{c}}
76 \right)\left(r-R_\mathrm{c}\right)\ \Biggr{]}
77 \quad r\leqslant R_\textrm{c}.
78 \label{eq:DSFPot}
79 \end{split}
80 \end{equation}
81 Overall, the damped {\sc sf} method is considered an improvement over
82 the {\sc sp} method because there is no discontinuity in the forces as
83 particles move out of the cutoff radius ($R_\textrm{c}$). This is
84 accomplished by shifting the forces (and potential) to zero at
85 $R_\textrm{c}$. This is analogous to neutralizing the charge and the
86 force effect of the charges within $R_\textrm{c}$.
87
88 As we mentioned in the previous study, to complete the charge
89 neutralization procedure, a self-neutralization term needs to be
90 included in the potential. This term exists outside the pair-loop, and
91 can be thought of as a charge opposite in sign and equal in magnitude
92 to the central particle, spread out over the entire cutoff
93 sphere. This term is calculated via a single loop over all charges in
94 the system. For the undamped case, the self-term can be written as
95 \begin{equation}
96 V_\textrm{self} = \sum_{i=1}^N\frac{q_iq_i}{2R_\textrm{c}},
97 \label{eq:selfTerm}
98 \end{equation}
99 while for the damped case it can be written as
100 \begin{equation}
101 V_\textrm{Dself} = \sum_{i=1}^Nq_iq_i\left(\frac{\alpha}{\sqrt{\pi}}
102 + \frac{\textrm{erfc}(\alpha R_\textrm{c})}{2R_\textrm{c}}\right).
103 \label{eq:dampSelfTerm}
104 \end{equation}
105 The first term within the parentheses of equation
106 (\ref{eq:dampSelfTerm}) is identical to the self-term in the Ewald
107 summation, and comes from the utilization of the complimentary error
108 function for electrostatic damping.\cite{deLeeuw80,Wolf99}
109
110 \section{Multipolar Electrostatic Damping}\label{sec:dampingMultipoles}
111
112 As stated above, the {\sc sf} method operates by neutralizing the
113 charge pair force interactions (and potential) within the cutoff
114 sphere with shifting and by damping the electrostatic
115 interactions. Now we would like to consider an extension of these
116 techniques to include point multipole interactions. How will the
117 shifting and damping need to be modified in order to accommodate point
118 multipoles?
119
120 Of the two component techniques, the easiest to adapt is
121 shifting. Shifting is employed to neutralize the cutoff sphere;
122 however, in a system composed purely of point multipoles, the cutoff
123 sphere is already neutralized. This means that shifting is not
124 necessary between point multipoles. In a mixed system of monopoles and
125 multipoles, the undamped {\sc sf} potential needs only to shift the
126 force terms of the monopole and smoothly truncate the multipole
127 interactions with a switching function. The switching function is
128 required in order to conserve energy, because a discontinuity will
129 exist in both the potential and forces at $R_\textrm{c}$ in the
130 absence of shifting terms.
131
132 If we want to dampen the {\sc sf} potential, then we need to
133 incorporate the complimentary error function term into the multipole
134 potentials. The most direct way to do this is by replacing $r^{-1}$
135 with erfc$(\alpha r)\cdot r^{-1}$ in the multipole expansion. In the
136 multipole expansion, rather than considering only the interactions
137 between single point charges, the electrostatic interaction is
138 reformulated such that it describes the interaction between charge
139 distributions about central sites of the respective sets of
140 charges.\cite{Hirschfelder67} This procedure is what leads to the
141 familiar charge-dipole,
142 \begin{equation}
143 V_\textrm{cd} = -q_i\frac{\boldsymbol{\mu}_j\cdot\mathbf{r}_{ij}}{r^3_{ij}}
144 = -q_i\mu_j\frac{\cos\theta}{r^2_{ij}},
145 \label{eq:chargeDipole}
146 \end{equation}
147 and dipole-dipole,
148 \begin{equation}
149 V_\textrm{dd} = 3\frac{(\boldsymbol{\mu}_i\cdot\mathbf{r}_{ij})
150 (\boldsymbol{\mu}_j\cdot\mathbf{r}_{ij})}{r^5_{ij}} -
151 \frac{\boldsymbol{\mu}_i\cdot\boldsymbol{\mu}_j}{r^3_{ij}},
152 \label{eq:dipoleDipole}
153 \end{equation}
154 interaction potentials.
155
156 Using the charge-dipole interaction as an example, if we insert
157 erfc$(\alpha r)\cdot r^{-1}$ in place of $r^{-1}$, a damped
158 charge-dipole results,
159 \begin{equation}
160 V_\textrm{Dcd} = -q_i\mu_j\frac{\cos\theta}{r^2_{ij}} c_1(r_{ij}),
161 \label{eq:dChargeDipole}
162 \end{equation}
163 where $c_1(r_{ij})$ is
164 \begin{equation}
165 c_1(r_{ij}) = \frac{2\alpha r_{ij}e^{-\alpha^2r^2_{ij}}}{\sqrt{\pi}}
166 + \textrm{erfc}(\alpha r_{ij}).
167 \label{eq:c1Func}
168 \end{equation}
169 Thus, $c_1(r_{ij})$ is the resulting damping term that modifies the
170 standard charge-dipole potential (Eq. (\ref{eq:chargeDipole})). Note
171 that this damping term is dependent upon distance and not upon
172 orientation, and that it is acting on what was originally an
173 $r^{-3}$ function. By writing the damped form in this manner, we
174 can collect the damping into one function and apply it to the original
175 potential when damping is desired. This works well for potentials that
176 have only one $r^{-n}$ term (where $n$ is an odd positive integer);
177 but in the case of the dipole-dipole potential, there is one part
178 dependent on $r^{-3}$ and another dependent on $r^{-5}$. In order to
179 properly damping this potential, each of these parts is dampened with
180 separate damping functions. We can determine the necessary damping
181 functions by continuing with the multipole expansion; however, it
182 quickly becomes more complex with ``two-center'' systems, like the
183 dipole-dipole potential, and is typically approached with a spherical
184 harmonic formalism.\cite{Hirschfelder67} A simpler method for
185 determining these functions arises from adopting the tensor formalism
186 for expressing the electrostatic interactions.\cite{Stone02}
187
188 The tensor formalism for electrostatic interactions involves obtaining
189 the multipole interactions from successive gradients of the monopole
190 potential. Thus, tensors of rank one through three are
191 \begin{equation}
192 T = \frac{1}{4\pi\epsilon_0r_{ij}},
193 \label{eq:tensorRank1}
194 \end{equation}
195 \begin{equation}
196 T_\alpha = \frac{1}{4\pi\epsilon_0}\nabla_\alpha \frac{1}{r_{ij}},
197 \label{eq:tensorRank2}
198 \end{equation}
199 \begin{equation}
200 T_{\alpha\beta} = \frac{1}{4\pi\epsilon_0}
201 \nabla_\alpha\nabla_\beta \frac{1}{r_{ij}},
202 \label{eq:tensorRank3}
203 \end{equation}
204 where the form of the first tensor gives the monopole-monopole
205 potential, the second gives the monopole-dipole potential, and the
206 third gives the monopole-quadrupole and dipole-dipole
207 potentials.\cite{Stone02} Since the force is $-\nabla V$, the forces
208 for each potential come from the next higher tensor.
209
210 To obtain the damped electrostatic forms, we replace $r^{-1}$ with
211 erfc$(\alpha r)\cdot r^{-1}$. Equation \ref{eq:tensorRank2} generates
212 $c_1(r_{ij})$, just like the multipole expansion, while equation
213 \ref{eq:tensorRank3} generates $c_2(r_{ij})$, where
214 \begin{equation}
215 c_2(r_{ij}) = \frac{4\alpha^3r^3_{ij}e^{-\alpha^2r^2_{ij}}}{3\sqrt{\pi}}
216 + \frac{2\alpha r_{ij}e^{-\alpha^2r^2_{ij}}}{\sqrt{\pi}}
217 + \textrm{erfc}(\alpha r_{ij}).
218 \end{equation}
219 Note that $c_2(r_{ij})$ is equal to $c_1(r_{ij})$ plus an additional
220 term. Continuing with higher rank tensors, we can obtain the damping
221 functions for higher multipole potentials and forces. Each subsequent
222 damping function includes one additional term, and we can simplify the
223 procedure for obtaining these terms by writing out the following
224 generating function,
225 \begin{equation}
226 c_n(r_{ij}) = \frac{2^n(\alpha r_{ij})^{2n-1}e^{-\alpha^2r^2_{ij}}}
227 {(2n-1)!!\sqrt{\pi}} + c_{n-1}(r_{ij}),
228 \label{eq:dampingGeneratingFunc}
229 \end{equation}
230 where,
231 \begin{equation}
232 m!! \equiv \left\{ \begin{array}{l@{\quad\quad}l}
233 m\cdot(m-2)\ldots 5\cdot3\cdot1 & m > 0\textrm{ odd} \\
234 m\cdot(m-2)\ldots 6\cdot4\cdot2 & m > 0\textrm{ even} \\
235 1 & m = -1\textrm{ or }0,
236 \end{array}\right.
237 \label{eq:doubleFactorial}
238 \end{equation}
239 and $c_0(r_{ij})$ is erfc$(\alpha r_{ij})$. This generating function
240 is similar in form to those obtained by researchers for the
241 application of the Ewald sum to
242 multipoles.\cite{Smith82,Smith98,Aguado03}
243
244 Returning to the dipole-dipole example, the potential consists of a
245 portion dependent upon $r^{-5}$ and another dependent upon
246 $r^{-3}$. In the damped dipole-dipole potential,
247 \begin{equation}
248 V_\textrm{Ddd} = 3\frac{(\boldsymbol{\mu}_i\cdot\mathbf{r}_{ij})
249 (\boldsymbol{\mu}_j\cdot\mathbf{r}_{ij})}{r^5_{ij}}
250 c_2(r_{ij}) -
251 \frac{\boldsymbol{\mu}_i\cdot\boldsymbol{\mu}_j}{r^3_{ij}}
252 c_1(r_{ij}),
253 \label{eq:dampDipoleDipole}
254 \end{equation}
255 $c_2(r_{ij})$ and $c_1(r_{ij})$ dampen these two parts
256 respectively. The forces for the damped dipole-dipole interaction,
257 \begin{equation}
258 \begin{split}
259 F_\textrm{Ddd} = &15\frac{(\boldsymbol{\mu}_i\cdot\mathbf{r}_{ij})
260 (\boldsymbol{\mu}_j\cdot\mathbf{r}_{ij})\mathbf{r}_{ij}}{r^7_{ij}}
261 c_3(r_{ij})\\ &-
262 3\frac{\boldsymbol{\mu}_i\cdot\mathbf{r}_{ij}\cdot\hat{\boldsymbol{\mu}}_j +
263 \boldsymbol{\mu}_j\cdot\mathbf{r}_{ij}\cdot\hat{\boldsymbol{\mu}}_i +
264 \boldsymbol{\mu}_i\cdot\boldsymbol{\mu}_j\cdot\mathbf{r}_{ij}}
265 {r^5_{ij}} c_2(r_{ij}),
266 \end{split}
267 \label{eq:dampDipoleDipoleForces}
268 \end{equation}
269 rely on higher order damping functions because we perform another
270 gradient operation. In this manner, we can dampen higher order
271 multipolar interactions along with the monopole interactions, allowing
272 us to include multipoles in simulations involving damped electrostatic
273 interactions.
274
275 \section{Applications of Pairwise Electrostatic Corrections: Water}
276
277 In our earlier work, we performed a statistical comparison of both
278 energetics and dynamics of the {\sc sf} and {\sc sp} methods, showing
279 that these pairwise correction techniques are nearly equivalent to the
280 Ewald summation in the simulation of general condensed
281 systems.\cite{Fennell06} It would be beneficial to have some specific
282 examples to better illustrate the similarities and differences of the
283 pairwise (specifically {\sc sf}) and Ewald techniques. To address
284 this, we decided to perform a detailed analysis involving water
285 simulations. Water is a good test in that there is a great deal of
286 detailed experimental information as well as an extensive library of
287 results from a variety of theoretical models. In choosing a model for
288 comparing these correction techniques, it is important to consider a
289 model that has been optimized for use with corrected electrostatics so
290 that the calculated properties are in better agreement with
291 experiment.\cite{vanderSpoel98,Horn04,Rick04} For this reason, we
292 chose the TIP5P-E water model.\cite{Rick04}
293
294 \subsection{TIP5P-E}
295
296 The TIP5P-E water model is a variant of Mahoney and Jorgensen's
297 five-point transferable intermolecular potential (TIP5P) model for
298 water.\cite{Mahoney00} TIP5P was developed to reproduce the density
299 maximum anomaly present in liquid water near 4$^\circ$C. As with many
300 previous point charge water models (such as ST2, TIP3P, TIP4P, SPC,
301 and SPC/E), TIP5P was parametrized using a simple cutoff with no
302 long-range electrostatic
303 correction.\cite{Stillinger74,Jorgensen83,Berendsen81,Berendsen87}
304 Without this correction, the pressure term on the central particle
305 from the surroundings is missing. When this correction is included,
306 systems of these particles expand to compensate for this added
307 pressure term and under-predict the density of water under standard
308 conditions. When using any form of long-range electrostatic
309 correction, it has become common practice to develop or utilize a
310 reparametrized water model that corrects for this
311 effect.\cite{vanderSpoel98,Fennell04,Horn04} The TIP5P-E model follows
312 this practice and was optimized for use with the Ewald
313 summation.\cite{Rick04} In his publication, Rick preserved the
314 geometry and point charge magnitudes in TIP5P and focused on altering
315 the Lennard-Jones parameters to correct the density at 298~K. With the
316 density corrected, he compared common water properties for TIP5P-E
317 using the Ewald sum with TIP5P using a 9~\AA\ cutoff.
318
319 In the following sections, we compared these same water properties
320 calculated from TIP5P-E using the Ewald sum with TIP5P-E using the
321 {\sc sf} technique. In the above evaluation of the pairwise
322 techniques, we observed some flexibility in the choice of parameters.
323 Because of this, the following comparisons include the {\sc sf}
324 technique with a 12~\AA\ cutoff and an $\alpha$ = 0.0, 0.1, and
325 0.2~\AA$^{-1}$, as well as a 9~\AA\ cutoff with an $\alpha$ =
326 0.2~\AA$^{-1}$.
327
328 \subsubsection{Density}\label{sec:t5peDensity}
329
330 As stated previously, the property that prompted the development of
331 TIP5P-E was the density at 1 atm. The density depends upon the
332 internal pressure of the system in the $NPT$ ensemble, and the
333 calculation of the pressure includes a components from both the
334 kinetic energy and the virial. More specifically, the instantaneous
335 molecular pressure ($p(t)$) is given by
336 \begin{equation}
337 p(t) = \frac{1}{\textrm{d}V}\sum_\mu
338 \left[\frac{\mathbf{P}_{\mu}^2}{M_{\mu}}
339 + \mathbf{R}_{\mu}\cdot\sum_i\mathbf{F}_{\mu i}\right],
340 \label{eq:MolecularPressure}
341 \end{equation}
342 where d is the dimensionality of the system, $V$ is the volume,
343 $\mathbf{P}_{\mu}$ is the momentum of molecule $\mu$, $\mathbf{R}_\mu$
344 is the position of the center of mass ($M_\mu$) of molecule $\mu$, and
345 $\mathbf{F}_{\mu i}$ is the force on atom $i$ of molecule
346 $\mu$.\cite{Melchionna93} The virial term (the right term in the
347 brackets of equation \ref{eq:MolecularPressure}) is directly dependent
348 on the interatomic forces. Since the {\sc sp} method does not modify
349 the forces, the pressure using {\sc sp} will be identical to that
350 obtained without an electrostatic correction. The {\sc sf} method
351 does alter the virial component and, by way of the modified pressures,
352 should provide densities more in line with those obtained using the
353 Ewald summation.
354
355 To compare densities, $NPT$ simulations were performed with the same
356 temperatures as those selected by Rick in his Ewald summation
357 simulations.\cite{Rick04} In order to improve statistics around the
358 density maximum, 3~ns trajectories were accumulated at 0, 12.5, and
359 25$^\circ$C, while 2~ns trajectories were obtained at all other
360 temperatures. The average densities were calculated from the later
361 three-fourths of each trajectory. Similar to Mahoney and Jorgensen's
362 method for accumulating statistics, these sequences were spliced into
363 200 segments, each providing an average density. These 200 density
364 values were used to calculate the average and standard deviation of
365 the density at each temperature.\cite{Mahoney00}
366
367 \begin{figure}
368 \includegraphics[width=\linewidth]{./figures/tip5peDensities.pdf}
369 \caption{Density versus temperature for the TIP5P-E water model when
370 using the Ewald summation (Ref. \citen{Rick04}) and the {\sc sf} method
371 with various parameters. The pressure term from the image-charge shell
372 is larger than that provided by the reciprocal-space portion of the
373 Ewald summation, leading to slightly lower densities. This effect is
374 more visible with the 9~\AA\ cutoff, where the image charges exert a
375 greater force on the central particle. The error bars for the {\sc sf}
376 methods show the average one-sigma uncertainty of the density
377 measurement, and this uncertainty is the same for all the {\sc sf}
378 curves.}
379 \label{fig:t5peDensities}
380 \end{figure}
381 Figure \ref{fig:t5peDensities} shows the densities calculated for
382 TIP5P-E using differing electrostatic corrections overlaid on the
383 experimental values.\cite{CRC80} The densities when using the {\sc sf}
384 technique are close to, though typically lower than, those calculated
385 using the Ewald summation. These slightly reduced densities indicate
386 that the pressure component from the image charges at R$_\textrm{c}$
387 is larger than that exerted by the reciprocal-space portion of the
388 Ewald summation. Bringing the image charges closer to the central
389 particle by choosing a 9~\AA\ R$_\textrm{c}$ (rather than the
390 preferred 12~\AA\ R$_\textrm{c}$) increases the strength of the image
391 charge interactions on the central particle and results in a further
392 reduction of the densities.
393
394 Because the strength of the image charge interactions has a noticeable
395 effect on the density, we would expect the use of electrostatic
396 damping to also play a role in these calculations. Larger values of
397 $\alpha$ weaken the pair-interactions; and since electrostatic damping
398 is distance-dependent, force components from the image charges will be
399 reduced more than those from particles close the the central
400 charge. This effect is visible in figure \ref{fig:t5peDensities} with
401 the damped {\sc sf} sums showing slightly higher densities; however,
402 it is apparent that the choice of cutoff radius plays a much more
403 important role in the resulting densities.
404
405 As a final note, all of the above density calculations were performed
406 with systems of 512 water molecules. Rick observed a system size
407 dependence of the computed densities when using the Ewald summation,
408 most likely due to his tying of the convergence parameter to the box
409 dimensions.\cite{Rick04} For systems of 256 water molecules, the
410 calculated densities were on average 0.002 to 0.003~g/cm$^3$ lower. A
411 system size of 256 molecules would force the use of a shorter
412 R$_\textrm{c}$ when using the {\sc sf} technique, and this would also
413 lower the densities. Moving to larger systems, as long as the
414 R$_\textrm{c}$ remains at a fixed value, we would expect the densities
415 to remain constant.
416
417 \subsubsection{Liquid Structure}\label{sec:t5peLiqStructure}
418
419 A common function considered when developing and comparing water
420 models is the oxygen-oxygen radial distribution function
421 ($g_\textrm{OO}(r)$). The $g_\textrm{OO}(r)$ is the probability of
422 finding a pair of oxygen atoms some distance ($r$) apart relative to a
423 random distribution at the same density.\cite{Allen87} It is
424 calculated via
425 \begin{equation}
426 g_\textrm{OO}(r) = \frac{V}{N^2}\left\langle\sum_i\sum_{j\ne i}
427 \delta(\mathbf{r}-\mathbf{r}_{ij})\right\rangle,
428 \label{eq:GOOofR}
429 \end{equation}
430 where the double sum is over all $i$ and $j$ pairs of $N$ oxygen
431 atoms. The $g_\textrm{OO}(r)$ can be directly compared to X-ray or
432 neutron scattering experiments through the oxygen-oxygen structure
433 factor ($S_\textrm{OO}(k)$) by the following relationship:
434 \begin{equation}
435 S_\textrm{OO}(k) = 1 + 4\pi\rho
436 \int_0^\infty r^2\frac{\sin kr}{kr}g_\textrm{OO}(r)\textrm{d}r.
437 \label{eq:SOOofK}
438 \end{equation}
439 Thus, $S_\textrm{OO}(k)$ is related to the Fourier-Laplace transform
440 of $g_\textrm{OO}(r)$.
441
442 The experimentally determined $g_\textrm{OO}(r)$ for liquid water has
443 been compared in great detail with the various common water models,
444 and TIP5P was found to be in better agreement than other rigid,
445 non-polarizable models.\cite{Sorenson00} This excellent agreement with
446 experiment was maintained when Rick developed TIP5P-E.\cite{Rick04} To
447 check whether the choice of using the Ewald summation or the {\sc sf}
448 technique alters the liquid structure, the $g_\textrm{OO}(r)$s at
449 298~K and 1~atm were determined for the systems compared in the
450 previous section.
451
452 \begin{figure}
453 \includegraphics[width=\linewidth]{./figures/tip5peGofR.pdf}
454 \caption{The $g_\textrm{OO}(r)$s calculated for TIP5P-E at 298~K and
455 1atm while using the Ewald summation (Ref. \cite{Rick04}) and the {\sc
456 sf} technique with varying parameters. Even with the reduced densities
457 using the {\sc sf} technique, the $g_\textrm{OO}(r)$s are essentially
458 identical.}
459 \label{fig:t5peGofRs}
460 \end{figure}
461 The $g_\textrm{OO}(r)$s calculated for TIP5P-E while using the {\sc
462 sf} technique with a various parameters are overlaid on the
463 $g_\textrm{OO}(r)$ while using the Ewald summation in
464 figure~\ref{fig:t5peGofRs}. The differences in density do not appear
465 to have any effect on the liquid structure as the $g_\textrm{OO}(r)$s
466 are indistinguishable. These results indicate that the
467 $g_\textrm{OO}(r)$ is insensitive to the choice of electrostatic
468 correction.
469
470 \subsubsection{Thermodynamic Properties}\label{sec:t5peThermo}
471
472 In addition to the density, there are a variety of thermodynamic
473 quantities that can be calculated for water and compared directly to
474 experimental values. Some of these additional quantities include the
475 latent heat of vaporization ($\Delta H_\textrm{vap}$), the constant
476 pressure heat capacity ($C_p$), the isothermal compressibility
477 ($\kappa_T$), the thermal expansivity ($\alpha_p$), and the static
478 dielectric constant ($\epsilon$). All of these properties were
479 calculated for TIP5P-E with the Ewald summation, so they provide a
480 good set for comparisons involving the {\sc sf} technique.
481
482 The $\Delta H_\textrm{vap}$ is the enthalpy change required to
483 transform one mole of substance from the liquid phase to the gas
484 phase.\cite{Berry00} In molecular simulations, this quantity can be
485 determined via
486 \begin{equation}
487 \begin{split}
488 \Delta H_\textrm{vap} &= H_\textrm{gas} - H_\textrm{liq.} \\
489 &= E_\textrm{gas} - E_\textrm{liq.}
490 + p(V_\textrm{gas} - V_\textrm{liq.}) \\
491 &\approx -\frac{\langle U_\textrm{liq.}\rangle}{N}+ RT,
492 \end{split}
493 \label{eq:DeltaHVap}
494 \end{equation}
495 where $E$ is the total energy, $U$ is the potential energy, $p$ is the
496 pressure, $V$ is the volume, $N$ is the number of molecules, $R$ is
497 the gas constant, and $T$ is the temperature.\cite{Jorgensen98b} As
498 seen in the last line of equation (\ref{eq:DeltaHVap}), we can
499 approximate $\Delta H_\textrm{vap}$ by using the ideal gas for the gas
500 state. This allows us to cancel the kinetic energy terms, leaving only
501 the $U_\textrm{liq.}$ term. Additionally, the $pV$ term for the gas is
502 several orders of magnitude larger than that of the liquid, so we can
503 neglect the liquid $pV$ term.
504
505 The remaining thermodynamic properties can all be calculated from
506 fluctuations of the enthalpy, volume, and system dipole
507 moment.\cite{Allen87} $C_p$ can be calculated from fluctuations in the
508 enthalpy in constant pressure simulations via
509 \begin{equation}
510 \begin{split}
511 C_p = \left(\frac{\partial H}{\partial T}\right)_{N,p}
512 = \frac{(\langle H^2\rangle - \langle H\rangle^2)}{Nk_BT^2},
513 \end{split}
514 \label{eq:Cp}
515 \end{equation}
516 where $k_B$ is Boltzmann's constant. $\kappa_T$ can be calculated via
517 \begin{equation}
518 \begin{split}
519 \kappa_T = \frac{1}{V}\left(\frac{\partial V}{\partial p}\right)_{N,T}
520 = \frac{(\langle V^2\rangle_{N,P,T} - \langle V\rangle^2_{N,P,T})}
521 {k_BT\langle V\rangle_{N,P,T}},
522 \end{split}
523 \label{eq:kappa}
524 \end{equation}
525 and $\alpha_p$ can be calculated via
526 \begin{equation}
527 \begin{split}
528 \alpha_p = \frac{1}{V}\left(\frac{\partial V}{\partial T}\right)_{N,P}
529 = \frac{(\langle VH\rangle_{N,P,T}
530 - \langle V\rangle_{N,P,T}\langle H\rangle_{N,P,T})}
531 {k_BT^2\langle V\rangle_{N,P,T}}.
532 \end{split}
533 \label{eq:alpha}
534 \end{equation}
535 Using the Ewald sum under tin-foil boundary conditions, $\epsilon$ can
536 be calculated for systems of non-polarizable substances via
537 \begin{equation}
538 \epsilon = 1 + \frac{\langle M^2\rangle}{3\epsilon_0k_\textrm{B}TV},
539 \label{eq:staticDielectric}
540 \end{equation}
541 where $\epsilon_0$ is the permittivity of free space and $\langle
542 M^2\rangle$ is the fluctuation of the system dipole
543 moment.\cite{Allen87} The numerator in the fractional term in equation
544 (\ref{eq:staticDielectric}) is the fluctuation of the simulation-box
545 dipole moment, identical to the quantity calculated in the
546 finite-system Kirkwood $g$ factor ($G_k$):
547 \begin{equation}
548 G_k = \frac{\langle M^2\rangle}{N\mu^2},
549 \label{eq:KirkwoodFactor}
550 \end{equation}
551 where $\mu$ is the dipole moment of a single molecule of the
552 homogeneous system.\cite{Neumann83,Neumann84,Neumann85} The
553 fluctuation term in both equation (\ref{eq:staticDielectric}) and
554 \ref{eq:KirkwoodFactor} is calculated as follows,
555 \begin{equation}
556 \begin{split}
557 \langle M^2\rangle &= \langle\bm{M}\cdot\bm{M}\rangle
558 - \langle\bm{M}\rangle\cdot\langle\bm{M}\rangle \\
559 &= \langle M_x^2+M_y^2+M_z^2\rangle
560 - (\langle M_x\rangle^2 + \langle M_x\rangle^2
561 + \langle M_x\rangle^2).
562 \end{split}
563 \label{eq:fluctBoxDipole}
564 \end{equation}
565 This fluctuation term can be accumulated during the simulation;
566 however, it converges rather slowly, thus requiring multi-nanosecond
567 simulation times.\cite{Horn04} In the case of tin-foil boundary
568 conditions, the dielectric/surface term of the Ewald summation is
569 equal to zero. Since the {\sc sf} method also lacks this
570 dielectric/surface term, equation (\ref{eq:staticDielectric}) is still
571 valid for determining static dielectric constants.
572
573 All of the above properties were calculated from the same trajectories
574 used to determine the densities in section \ref{sec:t5peDensity}
575 except for the static dielectric constants. The $\epsilon$ values were
576 accumulated from 2~ns $NVE$ ensemble trajectories with system densities
577 fixed at the average values from the $NPT$ simulations at each of the
578 temperatures. The resulting values are displayed in figure
579 \ref{fig:t5peThermo}.
580 \begin{figure}
581 \centering
582 \includegraphics[width=4.5in]{./figures/t5peThermo.pdf}
583 \caption{Thermodynamic properties for TIP5P-E using the Ewald summation
584 and the {\sc sf} techniques along with the experimental values. Units
585 for the properties are kcal mol$^{-1}$ for $\Delta H_\textrm{vap}$,
586 cal mol$^{-1}$ K$^{-1}$ for $C_p$, 10$^6$ atm$^{-1}$ for $\kappa_T$,
587 and 10$^5$ K$^{-1}$ for $\alpha_p$. Ewald summation results are from
588 reference \cite{Rick04}. Experimental values for $\Delta
589 H_\textrm{vap}$, $\kappa_T$, and $\alpha_p$ are from reference
590 \cite{Kell75}. Experimental values for $C_p$ are from reference
591 \cite{Wagner02}. Experimental values for $\epsilon$ are from reference
592 \cite{Malmberg56}.}
593 \label{fig:t5peThermo}
594 \end{figure}
595
596 As observed for the density in section \ref{sec:t5peDensity}, the
597 property trends with temperature seen when using the Ewald summation
598 are reproduced with the {\sc sf} technique. One noticeable difference
599 between the properties calculated using the two methods are the lower
600 $\Delta H_\textrm{vap}$ values when using {\sc sf}. This is to be
601 expected due to the direct weakening of the electrostatic interaction
602 through forced neutralization. This results in an increase of the
603 intermolecular potential producing lower values from equation
604 (\ref{eq:DeltaHVap}). The slopes of these values with temperature are
605 similar to that seen using the Ewald summation; however, they are both
606 steeper than the experimental trend, indirectly resulting in the
607 inflated $C_p$ values at all temperatures.
608
609 Above the supercooled regime, $C_p$, $\kappa_T$, and $\alpha_p$ values
610 all overlap within error. As indicated for the $\Delta H_\textrm{vap}$
611 and $C_p$ results discussed in the previous paragraph, the deviations
612 between experiment and simulation in this region are not the fault of
613 the electrostatic summation methods but are due to the geometry and
614 parameters of the TIP5P class of water models. Like most rigid,
615 non-polarizable, point-charge water models, the density decreases with
616 temperature at a much faster rate than experiment (see figure
617 \ref{fig:t5peDensities}). This reduced density leads to the inflated
618 compressibility and expansivity values at higher temperatures seen
619 here in figure \ref{fig:t5peThermo}. Incorporation of polarizability
620 and many-body effects are required in order for water models to
621 overcome differences between simulation-based and experimentally
622 determined densities at these higher
623 temperatures.\cite{Laasonen93,Donchev06}
624
625 At temperatures below the freezing point for experimental water, the
626 differences between {\sc sf} and the Ewald summation results are more
627 apparent. The larger $C_p$ and lower $\alpha_p$ values in this region
628 indicate a more pronounced transition in the supercooled regime,
629 particularly in the case of {\sc sf} without damping. This points to
630 the onset of a more frustrated or glassy behavior for TIP5P-E at
631 temperatures below 250~K in the {\sc sf} simulations, indicating that
632 disorder in the reciprocal-space term of the Ewald summation might act
633 to loosen up the local structure more than the image-charges in {\sc
634 sf}. The damped {\sc sf} actually makes a better comparison with
635 experiment in this region, particularly for the $\alpha_p$ values. The
636 local interactions in the undamped {\sc sf} technique appear to be too
637 strong since the property change is much more dramatic than the damped
638 forms, while the Ewald summation appears to weight the
639 reciprocal-space interactions at the expense the local interactions,
640 disagreeing with the experimental results. This observation is
641 explored further in section \ref{sec:t5peDynamics}.
642
643 The final thermodynamic property displayed in figure
644 \ref{fig:t5peThermo}, $\epsilon$, shows the greatest discrepancy
645 between the Ewald summation and the {\sc sf} technique (and experiment
646 for that matter). It is known that the dielectric constant is
647 dependent upon and quite sensitive to the imposed boundary
648 conditions.\cite{Neumann80,Neumann83} This is readily apparent in the
649 converged $\epsilon$ values accumulated for the {\sc sf}
650 simulations. Lack of a damping function results in dielectric
651 constants significantly smaller than those obtained using the Ewald
652 sum. Increasing the damping coefficient to 0.2~\AA$^{-1}$ improves the
653 agreement considerably. It should be noted that the choice of the
654 ``Ewald coefficient'' value also has a significant effect on the
655 calculated value when using the Ewald summation. In the simulations of
656 TIP5P-E with the Ewald sum, this screening parameter was tethered to
657 the simulation box size (as was the $R_\textrm{c}$).\cite{Rick04} In
658 general, systems with larger screening parameters reported larger
659 dielectric constant values, the same behavior we see here with {\sc
660 sf}; however, the choice of cutoff radius also plays an important
661 role. This connection is further explored below as optimal damping
662 coefficients for different choices of $R_\textrm{c}$ are determined
663 for {\sc sf} in order to best capture the dielectric behavior.
664
665 \subsubsection{Dielectric Constant}\label{sec:t5peDielectric}
666
667 In the previous, we observed that the choice of damping coefficient
668 plays a major role in the calculated dielectric constant. This is not
669 too surprising given the results for damping parameter influence on
670 the long-time correlated motions of the NaCl crystal in our previous
671 study.\cite{Fennell06} The static dielectric constant is calculated
672 from the long-time fluctuations of the system's accumulated dipole
673 moment (Eq. (\ref{eq:staticDielectric})), so it is going to be quite
674 sensitive to the choice of damping parameter. We would like to choose
675 optimal damping constants such that any arbitrary choice of cutoff
676 radius will properly capture the dielectric behavior of the liquid.
677
678 In order to find these optimal values, we mapped out the static
679 dielectric constant as a function of both the damping parameter and
680 cutoff radius. To calculate the static dielectric constant, we
681 performed 5~ns $NPT$ calculations on systems of 512 TIP5P-E water
682 molecules and averaged over the converged region (typically the later
683 2.5~ns) of equation (\ref{eq:staticDielectric}). The selected cutoff
684 radii ranged from 9, 10, 11, and 12~\AA , and the damping parameter
685 values ranged from 0.1 to 0.35~\AA$^{-1}$.
686
687 \begin{table}
688 \centering
689 \caption{TIP5P-E dielectric constant as a function of $\alpha$ and
690 $R_\textrm{c}$. The color scale ranges from blue (10) to red (100).}
691 \vspace{6pt}
692 \begin{tabular}{ lcccc }
693 \toprule
694 \toprule
695 & \multicolumn{4}{c}{$R_\textrm{c}$ (\AA )} \\
696 \cmidrule(lr){2-5}
697 $\alpha$ (\AA$^{-1}$) & 9 & 10 & 11 & 12 \\
698 \midrule
699 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 \\
700 & \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} \\
701 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 \\
702 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 \\
703 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 \\
704 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 \\
705 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 \\
706 & \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} \\
707 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 \\
708 & \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} \\
709 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 \\
710 \bottomrule
711 \end{tabular}
712 \label{tab:tip5peDielectricMap}
713 \end{table}
714 The results of these calculations are displayed in table
715 \ref{tab:tip5peDielectricMap}. Coloring of the table is based on the
716 numerical values within the cells and was added to better facilitate
717 interpretation of the displayed data and highlight trends in the
718 dielectric constant behavior. This makes it easy to see that the
719 dielectric constant is linear with respect to $\alpha$ and
720 $R_\textrm{c}$ in the low to moderate damping regions, and the slope
721 is 0.025~\AA$^{-1}\cdot R_\textrm{c}^{-1}$. Another point to note is
722 that choosing $\alpha$ and $R_\textrm{c}$ identical to those used with
723 the Ewald summation results in the same calculated dielectric
724 constant. As an example, in the paper outlining the development of
725 TIP5P-E, the real-space cutoff and Ewald coefficient were tethered to
726 the system size, and for a 512 molecule system are approximately
727 12~\AA\ and 0.25~\AA$^{-1}$ respectively.\cite{Rick04} These
728 parameters resulted in a dielectric constant of 92$\pm$14, while with
729 {\sc sf} these parameters give a dielectric constant of
730 90.8$\pm$0.9. Another example comes from the TIP4P-Ew paper where
731 $\alpha$ and $R_\textrm{c}$ were chosen to be 9.5~\AA\ and
732 0.35~\AA$^{-1}$, and these parameters resulted in a dielectric
733 constant equal to 63$\pm$1.\cite{Horn04} Calculations using {\sc sf}
734 with these parameters and this water model give a dielectric constant
735 of 61$\pm$1. Since the dielectric constant is dependent on $\alpha$
736 and $R_\textrm{c}$ with the {\sc sf} technique, it might be
737 interesting to investigate the dielectric dependence of the real-space
738 Ewald parameters.
739
740 Although it is tempting to choose damping parameters equivalent to
741 these Ewald examples, the results of our previous work indicate that
742 values this high are destructive to both the energetics and
743 dynamics.\cite{Fennell06} Ideally, $\alpha$ should not exceed
744 0.3~\AA$^{-1}$ for any of the cutoff values in this range. If the
745 optimal damping parameter is chosen to be midway between 0.275 and
746 0.3~\AA$^{-1}$ (0.2875~\AA$^{-1}$) at the 9~\AA\ cutoff, then the
747 linear trend with $R_\textrm{c}$ will always keep $\alpha$ below
748 0.3~\AA$^{-1}$ for the studied cutoff radii. This linear progression
749 would give values of 0.2875, 0.2625, 0.2375, and 0.2125~\AA$^{-1}$ for
750 cutoff radii of 9, 10, 11, and 12~\AA. Setting this to be the default
751 behavior for the damped {\sc sf} technique will result in consistent
752 dielectric behavior for these and other condensed molecular systems,
753 regardless of the chosen cutoff radius. The static dielectric
754 constants for TIP5P-E simulations with 9 and 12\AA\ $R_\textrm{c}$
755 values using their respective damping parameters are 76$\pm$1 and
756 75$\pm$2. These values are lower than the values reported for TIP5P-E
757 with the Ewald sum; however, they are more in line with the values
758 reported by Mahoney {\it et al.} for TIP5P while using a reaction
759 field (RF) with an infinite RF dielectric constant
760 (81.5$\pm$1.6).\cite{Mahoney00} As a final note, aside from a slight
761 lowering of $\Delta H_\textrm{vap}$, using these $\alpha$ values does
762 not alter the other other thermodynamic properties.
763
764 \subsubsection{Dynamic Properties}\label{sec:t5peDynamics}
765
766 To look at the dynamic properties of TIP5P-E when using the {\sc sf}
767 method, 200~ps $NVE$ simulations were performed for each temperature
768 at the average density reported by the $NPT$ simulations using 9 and
769 12~\AA\ $R_\textrm{c}$ values using the ideal $\alpha$ values
770 determined above (0.2875 and 0.2125~\AA$^{-1}$). The self-diffusion
771 constants ($D$) were calculated using the mean square displacement
772 (MSD) form of the Einstein relation,
773 \begin{equation}
774 D = \lim_{t\rightarrow\infty}
775 \frac{\langle\left|\mathbf{r}_i(t)-\mathbf{r}_i(0)\right|^2\rangle}{6t},
776 \label{eq:MSD}
777 \end{equation}
778 where $t$ is time, and $\mathbf{r}_i$ is the position of particle
779 $i$.\cite{Allen87}
780
781 \begin{figure}
782 \centering
783 \includegraphics[width=3.5in]{./figures/waterFrame.pdf}
784 \caption{Body-fixed coordinate frame for a water molecule. The
785 respective molecular principle axes point in the direction of the
786 labeled frame axes.}
787 \label{fig:waterFrame}
788 \end{figure}
789 In addition to translational diffusion, orientational relaxation times
790 were calculated for comparisons with the Ewald simulations and with
791 experiments. These values were determined from the same 200~ps $NVE$
792 trajectories used for translational diffusion by calculating the
793 orientational time correlation function,
794 \begin{equation}
795 C_l^\alpha(t) = \left\langle P_l\left[\hat{\mathbf{u}}_i^\alpha(t)
796 \cdot\hat{\mathbf{u}}_i^\alpha(0)\right]\right\rangle,
797 \label{eq:OrientCorr}
798 \end{equation}
799 where $P_l$ is the Legendre polynomial of order $l$ and
800 $\hat{\mathbf{u}}_i^\alpha$ is the unit vector of molecule $i$ along
801 principle axis $\alpha$. The principle axis frame for these water
802 molecules is shown in figure \ref{fig:waterFrame}. As an example,
803 $C_l^y$ is calculated from the time evolution of the unit vector
804 connecting the two hydrogen atoms.
805
806 \begin{figure}
807 \centering
808 \includegraphics[width=3.5in]{./figures/exampleOrientCorr.pdf}
809 \caption{Example plots of the orientational autocorrelation functions
810 for the first and second Legendre polynomials. These curves show the
811 time decay of the unit vector along the $y$ principle axis.}
812 \label{fig:OrientCorr}
813 \end{figure}
814 From the orientation autocorrelation functions, we can obtain time
815 constants for rotational relaxation. Figure \ref{fig:OrientCorr} shows
816 some example plots of orientational autocorrelation functions for the
817 first and second Legendre polynomials. The relatively short time
818 portions (between 1 and 3~ps for water) of these curves can be fit to
819 an exponential decay to obtain these constants, and they are directly
820 comparable to water orientational relaxation times from nuclear
821 magnetic resonance (NMR). The relaxation constant obtained from
822 $C_2^y(t)$ is of particular interest because it describes the
823 relaxation of the principle axis connecting the hydrogen atoms. Thus,
824 $C_2^y(t)$ can be compared to the intermolecular portion of the
825 dipole-dipole relaxation from a proton NMR signal and should provide
826 the best estimate of the NMR relaxation time constant.\cite{Impey82}
827
828 \begin{figure}
829 \centering
830 \includegraphics[width=3.5in]{./figures/t5peDynamics.pdf}
831 \caption{Diffusion constants ({\it upper}) and reorientational time
832 constants ({\it lower}) for TIP5P-E using the Ewald sum and {\sc sf}
833 technique compared with experiment. Data at temperatures less than
834 0$^\circ$C were not included in the $\tau_2^y$ plot to allow for
835 easier comparisons in the more relevant temperature regime.}
836 \label{fig:t5peDynamics}
837 \end{figure}
838 Results for the diffusion constants and orientational relaxation times
839 are shown in figure \ref{fig:t5peDynamics}. From this figure, it is
840 apparent that the trends for both $D$ and $\tau_2^y$ of TIP5P-E using
841 the Ewald sum are reproduced with the {\sc sf} technique. The enhanced
842 diffusion at high temperatures are again the product of the lower
843 densities in comparison with experiment and do not provide any special
844 insight into differences between the electrostatic summation
845 techniques. With the undamped {\sc sf} technique, TIP5P-E tends to
846 diffuse a little faster than with the Ewald sum; however, use of light
847 to moderate damping results in indistinguishable $D$ values. Though
848 not apparent in this figure, {\sc sf} values at the lowest temperature
849 are approximately twice as slow as $D$ values obtained using the Ewald
850 sum. These values support the observation from section
851 \ref{sec:t5peThermo} that there appeared to be a change to a more
852 glassy-like phase with the {\sc sf} technique at these lower
853 temperatures, though this change seems to be more prominent with the
854 {\it undamped} {\sc sf} method, which has stronger local pairwise
855 electrostatic interactions.
856
857 The $\tau_2^y$ results in the lower frame of figure
858 \ref{fig:t5peDynamics} show a much greater difference between the {\sc
859 sf} results and the Ewald results. At all temperatures shown, TIP5P-E
860 relaxes faster than experiment with the Ewald sum while tracking
861 experiment fairly well when using the {\sc sf} technique, independent
862 of the choice of damping constant. Their are several possible reasons
863 for this deviation between techniques. The Ewald results were
864 calculated using shorter (10ps) trajectories than the {\sc sf} results
865 (25ps). A quick calculation from a 10~ps trajectory with {\sc sf} with
866 an $\alpha$ of 0.2~\AA$^{-1}$ at 25$^\circ$C showed a 0.4~ps drop in
867 $\tau_2^y$, placing the result more in line with that obtained using
868 the Ewald sum. This example supports this explanation; however,
869 recomputing the results to meet a poorer statistical standard is
870 counter-productive. Assuming the Ewald results are not entirely the
871 product of poor statistics, differences in techniques to integrate the
872 orientational motion could also play a role. {\sc shake} is the most
873 commonly used technique for approximating rigid-body orientational
874 motion,\cite{Ryckaert77} whereas in {\sc oopse}, we maintain and
875 integrate the entire rotation matrix using the {\sc dlm}
876 method.\cite{Meineke05} Since {\sc shake} is an iterative constraint
877 technique, if the convergence tolerances are raised for increased
878 performance, error will accumulate in the orientational
879 motion. Finally, the Ewald results were calculated using the $NVT$
880 ensemble, while the $NVE$ ensemble was used for {\sc sf}
881 calculations. The additional mode of motion due to the thermostat will
882 alter the dynamics, resulting in differences between $NVT$ and $NVE$
883 results. These differences are increasingly noticeable as the
884 thermostat time constant decreases.
885
886
887 \subsection{SSD/RF}
888
889 In section \ref{sec:dampingMultipoles}, we described a method for
890 applying the damped {\sc sf} technique to systems containing point
891 multipoles. The soft sticky dipole (SSD) family of water models is the
892 perfect test case because of the dipole-dipole (and
893 charge-dipole/quadrupole) interactions that are present. As alluded to
894 in the name, soft sticky dipole water models are single point
895 molecules that consist of a ``soft'' Lennard-Jones sphere, a
896 point-dipole, and a tetrahedral function for capturing the hydrogen
897 bonding nature of water - a spherical harmonic term for water-water
898 tetrahedral interactions and a point-quadrupole for interactions with
899 surrounding charges. Detailed descriptions of these models can be
900 found in other studies.\cite{Liu96b,Chandra99,Tan03,Fennell04}
901
902 In deciding which version of the SSD model to use, we need only
903 consider that the {\sc sf} technique was presented as a pairwise
904 replacement for the Ewald summation. It has been suggested that models
905 parametrized for the Ewald summation (like TIP5P-E) would be
906 appropriate for use with a reaction field and vice versa.\cite{Rick04}
907 Therefore, we decided to test the SSD/RF water model, which was
908 parametrized for use with a reaction field, with this damped
909 electrostatic technique to see how the calculated properties change.
910
911 \subsubsection{Dipolar Damping}
912
913 \begin{table}
914 \caption{Properties of SSD/RF when using different electrostatic
915 correction methods.}
916 \footnotesize
917 \centering
918 \begin{tabular}{ llccc }
919 \toprule
920 \toprule
921 & & Reaction Field [Ref. \citen{Fennell04}] & Damped Electrostatics & Experiment [Ref.] \\
922 & & $\epsilon = 80$ & $R_\textrm{c} = 12$\AA ; $\alpha = 0.2125$~\AA$^{-1}$ & \\
923 \midrule
924 $\rho$ & (g cm$^{-3}$) & 0.997(1) & 1.004(4) & 0.997 [\citen{CRC80}]\\
925 $C_p$ & (cal mol$^{-1}$ K$^{-1}$) & 23.8(2) & 27(1) & 18.005 [\citen{Wagner02}] \\
926 $D$ & ($10^{-5}$ cm$^2$ s$^{-1}$) & 2.32(6) & 2.33(2) & 2.299 [\citen{Mills73}]\\
927 $n_C$ & & 4.4 & 4.2 & 4.7 [\citen{Hura00}]\\
928 $n_H$ & & 3.7 & 3.7 & 3.5 [\citen{Soper86}]\\
929 $\tau_1$ & (ps) & 7.2(4) & 5.86(8) & 5.7 [\citen{Eisenberg69}]\\
930 $\tau_2$ & (ps) & 3.2(2) & 2.45(7) & 2.3 [\citen{Krynicki66}]\\
931 \bottomrule
932 \end{tabular}
933 \label{tab:dampedSSDRF}
934 \end{table}
935 The properties shown in table \ref{tab:dampedSSDRF} compare
936 quite well. The average density shows a modest increase when
937 using damped electrostatics in place of the reaction field. This comes
938 about because we neglect the pressure effect due to the surroundings
939 outside of the cutoff, instead relying on screening effects to
940 neutralize electrostatic interactions at long distances. The $C_p$
941 also shows a slight increase, indicating greater fluctuation in the
942 enthalpy at constant pressure. The only other differences between the
943 damped and reaction field results are the dipole reorientational time
944 constants, $\tau_1$ and $\tau_2$. When using damped electrostatics,
945 the water molecules relax more quickly and exhibit behavior very
946 similar to the experimental values. These results indicate that not
947 only is it reasonable to use damped electrostatics with SSD/RF, it is
948 recommended if capturing realistic dynamics is of primary
949 importance. This is an encouraging result because the damping methods
950 are more generally applicable than reaction field. Using damping,
951 SSD/RF can be used effectively with mixed systems, such as dissolved
952 ions, dissolved organic molecules, or even proteins.
953
954 \subsubsection{Dielectric Constant}
955
956 \begin{table}
957 \centering
958 \caption{SSD/RF dielectric constant as a function of $\alpha$ and $R_\textrm{c}$. The color scale ranges from blue (10) to red (100).}
959 \vspace{6pt}
960 \begin{tabular}{ lcccc }
961 \toprule
962 \toprule
963 & \multicolumn{4}{c}{$R_\textrm{c}$ (\AA )} \\
964 \cmidrule(lr){2-5}
965 $\alpha$ (\AA$^{-1}$) & 9 & 10 & 11 & 12 \\
966 \midrule
967 0.35 & \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 \\
968 & \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} \\
969 0.3 & \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 \\
970 0.275 & \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 \\
971 0.25 & \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 \\
972 0.225 & \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 \\
973 0.2 & \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 \\
974 & \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} \\
975 0.15 & \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 \\
976 & \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} \\
977 0.1 & \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 \\
978 \bottomrule
979 \end{tabular}
980 \label{tab:ssdrfDielectricMap}
981 \end{table}
982 In addition to the properties tabulated in table
983 \ref{tab:dampedSSDRF}, we mapped out the static dielectric constant of
984 SSD/RF as a function of $R_\textrm{c}$ and $\alpha$ in the same
985 fashion as in section \ref{sec:t5peDielectric} with TIP5P-E. It is
986 apparent from this table that electrostatic damping has a more
987 pronounced effect on the dipolar interactions of SSD/RF than the
988 monopolar interactions of TIP5P-E. The dielectric constant covers a
989 much wider range and has a steeper slope with increasing damping
990 parameter. We can use the same trend of $\alpha$ with $R_\textrm{c}$
991 used with TIP5P-E, and for a 12~\AA\ $R_\textrm{c}$, the resulting
992 dielectric constant is 82.6$\pm$0.6. This value compares very
993 favorably with the experimental value of 78.3.\cite{Malmberg56} This
994 is not surprising given that early studies of the SSD model indicate a
995 static dielectric constant around 81.\cite{Liu96}
996
997 \section{Application of Pairwise Electrostatic Corrections: Imaginary Ice}
998
999 In an earlier work, we performed a series of free energy calculations
1000 on several ice polymorphs which are stable or metastable at low
1001 pressures, one of which (Ice-$i$) we observed in spontaneous
1002 crystallizations of an SSD type single point water
1003 model.\cite{Fennell05} In this study, a distinct dependence of the
1004 free energies on the interaction cutoff and correction technique was
1005 observed. Being that the {\sc sf} technique can be used as a simple
1006 and efficient replacement for the Ewald summation, it would be
1007 interesting to apply it in addressing the question of the free
1008 energies of these ice polymorphs with varying water models. To this
1009 end, we set up thermodynamic integrations of all of the previously
1010 discussed ice polymorphs using the {\sc sf} technique with a cutoff
1011 radius of 12~\AA\ and an $\alpha$ of 0.2125~\AA . These calculations
1012 were performed on TIP5P-E and TIP4P-Ew (variants of the root models
1013 optimized for the Ewald summation) as well as SPC/E, and SSD/RF.
1014
1015 \begin{table}
1016 \centering
1017 \caption{Helmholtz free energies of ice polymorphs at 1~atm and 200~K
1018 using the damped {\sc sf} electrostatic correction method with a
1019 variety of water models.}
1020 \begin{tabular}{ lccccc }
1021 \toprule
1022 \toprule
1023 Model & I$_\textrm{h}$ & I$_\textrm{c}$ & B & Ice-$i$ & Ice-$i^\prime$ \\
1024 \cmidrule(lr){2-6}
1025 & \multicolumn{5}{c}{(kcal mol$^{-1}$)} \\
1026 \midrule
1027 TIP5P-E & -11.98(4) & -11.96(4) & -11.87(3) & - & -11.95(3) \\
1028 TIP4P-Ew & -13.11(3) & -13.09(3) & -12.97(3) & - & -12.98(3) \\
1029 SPC/E & -12.99(3) & -13.00(3) & -13.03(3) & - & -12.99(3) \\
1030 SSD/RF & -11.83(3) & -11.66(4) & -12.32(3) & -12.39(3) & - \\
1031 \bottomrule
1032 \end{tabular}
1033 \label{tab:dampedFreeEnergy}
1034 \end{table}
1035 The results of these calculations in table \ref{tab:dampedFreeEnergy}
1036 show similar behavior to the Ewald results in the previous study, at
1037 least for SSD/RF and SPC/E which are present in both.\cite{Fennell05}
1038 The Helmholtz free energies of the ice polymorphs with SSD/RF order in
1039 the same fashion, with Ice-$i$ having the lowest free energy; however,
1040 the Ice-$i$ and ice B free energies are quite a bit closer (nearly
1041 isoenergetic). The SPC/E results show the near isoenergetic behavior
1042 when using the Ewald summation.\cite{Fennell05} Ice B has the lowest
1043 Helmholtz free energy; however, all the polymorph results overlap
1044 within error.
1045
1046 The most interesting results from these calculations come from the
1047 more expensive TIP4P-Ew and TIP5P-E results. Both of these models were
1048 optimized for use with an electrostatic correction and are
1049 geometrically arranged to mimic water following two different
1050 ideas. In TIP5P-E, the primary location for the negative charge in the
1051 molecule is assigned to the lone-pairs of the oxygen, while TIP4P-Ew
1052 places the negative charge near the center-of-mass along the H-O-H
1053 bisector. There is some debate as to which is the proper choice for
1054 the negative charge location, and this has in part led to a six-site
1055 water model that balances both of these options.\cite{Vega05,Nada03}
1056 The limited results in table \ref{tab:dampedFreeEnergy} support the
1057 results of Vega {\it et al.}, which indicate the TIP4P charge location
1058 geometry is more physically valid.\cite{Vega05} With the TIP4P-Ew
1059 water model, the experimentally observed polymorph (ice
1060 I$_\textrm{h}$) is the preferred form with ice I$_\textrm{c}$ slightly
1061 higher in energy, though overlapping within error. Additionally, the
1062 less realistic ice B and Ice-$i^\prime$ structures are destabilized
1063 relative to these polymorphs. TIP5P-E shows similar behavior to SPC/E,
1064 where there is no real free energy distinction between the various
1065 polymorphs, because many overlap within error. While ice B is close in
1066 free energy to the other polymorphs, these results fail to support the
1067 findings of other researchers indicating the preferred form of TIP5P
1068 at 1~atm is a structure similar to ice
1069 B.\cite{Yamada02,Vega05,Abascal05} It should be noted that we are
1070 looking at TIP5P-E rather than TIP5P, and the differences in the
1071 Lennard-Jones parameters could be a reason for this dissimilarity.
1072 Overall, these results indicate that TIP4P-Ew is a better mimic of
1073 real water than these other models when studying crystallization and
1074 solid forms of water.
1075
1076 \section{Conclusions}
1077
1078 \section{Acknowledgments}
1079 Support for this project was provided by the National Science
1080 Foundation under grant CHE-0134881. Computation time was provided by
1081 the Notre Dame Center for Research Computing.
1082
1083 \newpage
1084
1085 \bibliographystyle{achemso}
1086 \bibliography{multipoleSFPaper}
1087
1088
1089 \end{document}