1: %% ****** Start of file template.aps ****** %
2: %%
3: %%
4: %% This file is part of the APS files in the REVTeX 4 distribution.
5: %% Version 4.0 beta 4 of REVTeX, May 24, 2000.
6: %%
7: %%
8: %% Copyright (c) 2000 The American Physical Society.
9: %%
10: %% See the REVTeX 4 README file for restrictions and more information.
11: %%
12: %
13: % This is a template for producing files for use with REVTEX 4.0 beta
14: % Copy this file to another name and then work on that file.
15: % That way, you always have this original template file to use.
16: %
17: % Group addresses by affiliation; use superscriptaddress for long
18: % author lists, or if there are many overlapping affiliations.
19: % For Phys. Rev. appearance, change preprint to twocolumn.
20: % Add 'draft' option to mark overfull boxes with black boxes
21: % Add 'showpacs' option to make PACS codes appear
22: \documentclass[aps,twocolumn,groupedaddress]{revtex4}
23: %\setlength{\topmargin}{0.3in}
24:
25: %\documentclass[aps,preprint,superscriptaddress]{revtex4}
26: %\documentclass[aps,twocolumn,groupedaddress]{revtex4}
27:
28: \usepackage[oztex]{graphicx}
29:
30:
31: \begin{document}
32: % You should use BibTeX and apsrev.bst for references
33: \bibliographystyle{apsrev}
34:
35: % Use the \preprint command to place your local institutional report
36: % number on the title page in preprint mode.
37: % Multiple \preprint commands are allowed.
38: %\preprint{}
39: \preprint{LA-UR-00-2319}
40:
41: %Title of paper
42: \title[Na$^+$(aq)]{The hydration number of Na$^+$ in liquid water}
43: % Optional argument for running titles on pages
44: %\title[]{}
45:
46: % repeat the \author .. \affiliation etc. as needed
47: % \email, \thanks, \homepage, \altaffiliation all apply to the current
48: % author. Explanatory text should go in the []'s, actual e-mail
49: % address or url should go in the {}'s for \email and \homepage.
50: % Please use the appropriate macro for the type of information
51:
52: % \affiliation command applies to all authors since the last
53: % \affiliation command. The \affiliation command should follow the
54: % other information
55:
56: \author{Susan B. Rempe}
57: %\email[]{Your e-mail address}
58: %\homepage[]{Your web page}
59: %\thanks{}
60: %\altaffiliation{}
61: \affiliation{Theoretical Division, Los Alamos National Laboratory, Los
62: Alamos, New Mexico 87545 USA}
63:
64: \author{Lawrence R. Pratt}
65: %\email[]{Your e-mail address}
66: %\homepage[]{Your web page}
67: %\thanks{}
68: %\altaffiliation{}
69: \affiliation{Theoretical Division, Los Alamos National Laboratory, Los
70: Alamos, New Mexico 87545 USA}
71:
72: %Collaboration name if desired (requires use of superscriptaddress
73: %option in \documentclass). \noaffiliation is required (may also be
74: %used with the \author command).
75: %\collaboration{}
76: %\noaffiliation
77:
78: \date{\today}
79:
80: \begin{abstract}
81: % insert abstract here
82: An `ab initio' molecular dynamics simulation of a Na$^+$ ion in
83: aqueous solution is presented and discussed. The calculation treats a
84: Na$^+$ ion and 32 water molecules with periodic boundary conditions on
85: a cubic volume determined by an estimate of zero partial molar volume
86: for this solute in water at normal density and at a temperature of
87: 344$\pm$24~K. Analysis of the last half of the 12 ps trajectory shows
88: 4.6 water molecules occupying the inner hydration shell of the Na$^+$
89: ion on average, with 5 being the most probable occupancy. The
90: self-diffusion coefficient observed for the Na$^+$ is
91: 1.0$\times$10$^{-5}$~cm$^2$/s. The quasi-chemical theory of solutions
92: provides the framework for two more calculations. First a
93: complementary calculation, based on electronic structure results for
94: ion-water clusters and a dielectric continuum model of outer sphere
95: hydration contributions, predicts an average hydration shell occupancy
96: of 4.0. This underestimate is attributed to the harmonic
97: approximation for the clusters in conjunction with the approximate
98: dielectric continuum model treatment of outer sphere contributions.
99: Finally, a maximum entropy fitting of inner sphere occupancies that
100: leads to insensitive composite free energy approximations suggests a
101: value in the neighborhood of -68~kcal/mol for the hydration free
102: energy of Na$^+$(aq) under these conditions with no contribution supplied for
103: packing or van der Waals interactions.
104:
105: {\it keywords:} ab initio molecular dynamics,
106: dielectric continuum, electronic structure, hydration, information
107: theory, quasi-chemical theory, sodium ion
108: \end{abstract}
109: % insert suggested PACS numbers in braces on next line
110: \pacs{}
111:
112: %\maketitle must follow title, authors, abstract and \pacs
113: \maketitle
114:
115: % body of paper here - Use proper section commands
116: % References should be done using the \cite, \ref, and \label commands
117: %\section{}
118: %\label{}
119: %\subsection{}
120: %\subsubsection{}
121:
122: \section{Introduction}
123: Solvation of simple ions in aqueous solution is not yet fully
124: understood despite its fundamental importance to chemical and
125: biological processes. For example, disagreement persists regarding the
126: hydration number of the Na$^+$ ion in liquid water. A pertinent
127: problem of current interest centers on the selectivity of biological
128: ion channels; it seems clear that the selective transport of K$^+$
129: relative to Na$^+$ ions in potassium
130: channels\cite{kchannel:98,guidoni:99,laio:99} depends on details of
131: the ion hydration that might differ for K$^+$ relative to Na$^+$.
132:
133: Experimental efforts to define the hydration structure of Na$^+$(aq)
134: using diffraction\cite{caminiti:80,skipper:89} and
135: spectroscopic\cite{Michaellian:78} methods produce a hydration number
136: ranging between four and six\cite{Ohtaki:93}. Simulations have
137: obtained a range of values, but most predict six water molecules in
138: the inner hydration sphere of the Na$^+$ ion%
139: \cite{Heinzinger:79,Mezei:81,Impey:83,Chandrasekhar:84,%
140: bounds:85,wilson:85,Zhu:91,Heinzinger:93,lee:94,toth:96,obst:96,%
141: Koneshan:98b}. An `ab initio' molecular dynamics simulation produced
142: five inner shell water molecules neighboring
143: Na$^+$(aq)\cite{schwegler:00}.
144:
145: An important limitation of theoretical studies of ion hydration
146: concerns the sufficiency of model force fields used in classical
147: statistical mechanical calculations. In the most customary
148: approaches, interatomic force fields used in theories or simulations
149: derive from empirical fits of a parameterized model to a variety of
150: experimental data. `Ab initio' molecular dynamics avoids this
151: intermediate modeling step by approximate solution of the electronic
152: Schroedinger equation for each configuration of the
153: nuclei\cite{marx:99,alfe:2000}. This technique thus goes significantly
154: beyond conventional simulations regarding the accuracy of the force
155: fields. It also augments theories built more directly on electronic
156: structure studies of ion-water complexes by adopting approximate
157: descriptions of the solution environment of those
158: complexes\cite{rempe:99,pratt:98,feature,martin:97,pratt:99,G98}.
159:
160: Relative to conventional simulations, `ab initio' molecular dynamics
161: simulations also have some important limitations due to the high
162: computational demand. Applications of the method have been restricted
163: to small systems simulated for short times. For example, an `ab
164: initio' molecular dynamics study\cite{schwegler:00} of the Na$^+$(aq)
165: ion comparable to the present work obtained a thermal trajectory
166: lasting 3 ps after minimal thermal aging. The present work, though
167: still limited to relatively small systems, pushes such calculations to
168: longer times that might permit more precise determination for
169: Na$^+$(aq) of primitive hydration properties. The analysis here
170: utilizes the last half of a 12~ps thermal trajectory. The
171: quasi-chemical
172: theory\cite{rempe:99,pratt:98,feature,martin:97,pratt:99} and separate
173: electronic structure calculations on Na(H$_2$O)$_n{}^+$ complexes
174: assist in this analysis.
175:
176:
177: \begin{figure}
178: \begin{center}
179: \leavevmode
180: \includegraphics{fig1.eps}
181: \end{center}
182: \caption{Structures from `ab initio' molecular dynamics calculations.
183: In the top panel, the Na$^+$ ion has five (5) inner shell water
184: molecule neighbors. The bottom panel shows the four-coordinate
185: structure produced 70~fs later. The bonds identify water oxygen atoms
186: within 3.1~\AA\ of the Na$^+$ ion. The hydrogen, sodium, and oxygen
187: atoms are shown as open, black, and gray circles, respectively.}
188: \label{sim:fig}
189: \end{figure}
190:
191:
192: \section{Methods}
193:
194: The system consisted of one Na$^+$ ion and 32 water molecules in a
195: cubic box with edge 9.86518~\AA\ and periodic boundary conditions. The
196: dimensions of the box correspond to a water density of 1 g/cm$^3$ and
197: zero partial molar volume for the solute. Initial conditions were
198: obtained as in an earlier `ab initio' molecular dynamics simulation
199: on Li$^+$(aq)\cite{rempe:99}. In that earlier work, an optimized
200: structure for the inner sphere Li(H$_2$O)$_6{}^+$ complex was
201: equilibrated with 26 water molecules under conventional simulation
202: conditions for liquid water, utilizing a current model force field and
203: assuming a partial molar volume of zero. In the present calculation,
204: the same pre-equilibrated system was used as an initial configuration
205: for the `ab initio' molecular dynamics except that an optimized
206: structure for the inner sphere Na(H$_2$O)$_6{}^+$ complex replaced the
207: hexa-coordinated Li$^+$ structure. Constant pressure or constant water
208: activity simulations, defined by intensive rather than extensive
209: variables, probably would produce a more useful characterization of
210: the solvent thermodynamic state for these small systems, but those
211: alternatives are currently impractical.
212:
213:
214: Molecular dynamics calculations based upon a gradient-corrected
215: electron density functional description of the electronic structure
216: and interatomic forces were carried out on this Na$^+$(aq) system
217: utilizing the VASP program\cite{vasp1}. The ions were represented by
218: ultrasoft pseudopotentials\cite{vasp2} and a kinetic energy cutoff of
219: 31.5 Rydberg limited the plane wave basis expansions of the valence
220: electronic wave functions. The equations of motion were integrated in
221: time steps of 1~fs, which is small enough to sample the lowest
222: vibrational frequency of water. A thermostat constrained the system
223: temperature to 300~K during the first 4.3 ps of simulation time. After
224: removing the thermostat, the temperature rose slightly and then
225: leveled off by 6 ps to an average of 344 $\pm$ 24~K. During the
226: simulation, the initial $n$=6 hydration structure relaxed into $n$=4
227: and $n$=5 alternatives, such as those shown in Fig.~\ref{sim:fig}. All
228: analyses reported here rely on the trajectory generated subsequent to
229: the 6 ps of aging with the system at a temperature elevated from room
230: temperature.
231:
232:
233: \begin{figure}
234: \begin{center}
235: \leavevmode
236: \includegraphics[scale=0.6]{g_NaO.eps}
237: \end{center}
238: \caption{Radial distribution function g$_{\mathrm{NaO}}$(r) and number
239: n(r) of oxygen atoms neighboring the Na$^+$ ion. Error estimates of
240: $\pm$ 2$\sigma$ are also plotted for the radial distribution function.
241: $\sigma$ was estimated by dividing the observed trajectory into four
242: blocks of approximate duration 1.5~ps; those blocks were assumed to
243: provide independent observations. The first minimum in the g(r)
244: function is at r=3.12~\AA\ where g(r) falls to 0.2. Here an average of
245: 4.6 oxygen atoms surround the Na$^+$ ion.}
246: \label{gor:fig}
247: \end{figure}
248:
249:
250: \section{Results}
251: The ion-oxygen radial distribution function is shown in
252: Fig.~\ref{gor:fig}. The first maximum occurs at a radius of 2.35~\AA\
253: from the Na$^+$ ion and the minimum at radius 3.12~\AA\ demarcates the
254: boundary of the first and innermost hydration shell. An average of
255: $\langle n\rangle$=4.6 water molecules occupy the inner hydration
256: shell. Fig.~\ref{neigh:fig1} tracks the instantaneous number of water
257: oxygen atoms found within the first hydration shell of the Na$^+$,
258: defined by radius r$\leq$3.12~\AA\ for the upper panel. The fractions
259: $x_4$ and $x_5$ of four-coordinate and five-coordinate hydration
260: structures, respectively, constitute $x_4$=40\% and $x_5$=56\% of the
261: last 6 ps of the simulation. Structures in which the Na$^+$ ion
262: acquires six innershell water molecules occur with a 4\% frequency,
263: while structures with three and seven innershell water molecules occur
264: less than 1\% of the time. Analysis of the mean-square displacement
265: of the Na$^+$ ion (Fig.~\ref{diff:fig}) produces a self-diffusion
266: constant of 1.0$\times$10$^{-5}$ cm$^2$/s, which agrees reasonably
267: well with an experimental result of 1.33x10$^{-5}$cm$^2$/s\cite{hertz:73}.
268:
269:
270:
271: These results correspond coarsely with an `ab initio' molecular
272: dynamics calculation on this system carried-out
273: independently\cite{schwegler:00}. The most probable inner shell
274: occupancy found there was also five, but the probabilities of n=4 and
275: n=6 were reversed from what we find here. This difference may be
276: associated with the lower temperature used in Ref~\cite{schwegler:00}.
277:
278:
279:
280: One motivation for this study arises from the quasi-chemical theory of
281: solutions. According to this formulation, $x_0$ contributes a `chemical'
282: contribution to $\mu_{Na^+}^{ex}$, the excess chemical potential or
283: absolute hydration free energy of the ion in liquid
284: water\cite{pratt:99},
285: \begin{eqnarray}
286: \beta\mu_{\mathrm{Na}^+}{}^{ex} = \ln x_0 - \ln\left[ \left\langle {e^{-\beta\Delta
287: U}}\prod\limits_j {(1-b_{\mathrm{Na}^+ j})} \right\rangle_0 \right].
288: \label{gqca}
289: \end{eqnarray}
290: Here the inner shell is defined by specifying a function
291: $b_{{\mathrm{Na}^+} j}$ that is equal to one (1) when solvent molecule
292: j is inside the defined inner shell and zero (0) otherwise; $\Delta
293: U$ is the interaction energy of the solvent with the solute Na$^+$
294: that is treated as a test particle, $\beta^{-1}$=k$_B$T, and the
295: subscript zero associated with $ \left\langle \ldots \right\rangle_0$
296: indicates a test particle average \cite{pratt:99}. The second term on
297: the right-hand side of Eq.~(\ref{gqca}) is the excess chemical
298: potential of the solute lacking inner shell solvent molecules whereas
299: the first term is the free energy of allowing solvent molecules to
300: occupy the inner shell. The validity of Eq.~(\ref{gqca}) has been
301: established elsewhere\cite{pratt:99}. The second term on the right of
302: Eq.~\ref{gqca} is the outer sphere contribution to the excess chemical
303: potential in contrast to the first or chemical term.
304:
305:
306: \begin{figure}[b!]
307: \begin{center}
308: \leavevmode
309: \includegraphics[scale=0.65]{bonds_bothr.eps}
310: \end{center}
311: \caption{The solid line in the upper plot depicts the number of oxygen
312: atoms within a radius of 3.12 {\AA} from the Na$^+$ at each
313: configuration in the molecular dynamics simulation. A radius of 2.68
314: {\AA} defines the nearest oxygen neighbors in the lower plot. The
315: dashed lines show the kinetic energy per atom during the simulation,
316: plotted after removal of the 300~K thermostat at 4.3~ps. The axis on
317: the right refers to the kinetic energy values. In the upper plot, an
318: average of 4.6 water molecules surround the Na$^+$ ion, while an
319: average of 4.0 water molecules surround the ion in the lower plot.}
320: \label{neigh:fig1}
321: \end{figure}
322:
323:
324:
325:
326: The utility of this quasi-chemical formulation is the
327: suggestion\cite{hummer:cp:2000} of more detailed study of the $x_n$,
328: the fractions of n-coordinate hydration structures found in solution,
329: on the basis of the equilibria forming inner shell complexes of
330: different aggregation number:
331: \begin{eqnarray} \mathrm{Na(H_2O)_{m=0}{}^+}\ +\ \mathrm{nH_2O}
332: \rightleftharpoons \mathrm{Na(H_2O)_n{}^+}~.
333: \label{reaction} \end{eqnarray}
334: Utilizing the chemical equilibrium ratios
335: \begin{eqnarray}
336: K_n={\rho_{\mathrm{Na(H_2O)_n}{}^+} \over
337: \rho_{\mathrm{H_2O}}{}^n\rho_{\mathrm{Na(H_2O)_{m=0}{}^+}}
338: }~,
339: \label{Kn-ob}
340: \end{eqnarray}
341: the normalized $x_n$ can be expressed as
342: \begin{eqnarray}
343: x_n={{K_{n}\rho _{\mathrm{H_2O}}{}^n} \over {\sum\limits_{m\ge 0}
344: {K_{m}\rho _{\mathrm{H_2O}}{}^m}}}~.
345: \label{cluster-var}
346: \end{eqnarray}
347: The $\rho_\sigma$ are the number densities and, in particular,
348: $\rho_{\mathrm{H_2O}}$ is the molecule number density of liquid water.
349: If the medium external to the clusters is neglected, the equilibrium
350: ratios, denoted as $K_n{}^{(0)}$, can be obtained from electronic
351: structure calculations on the complexes, assuming for the thermal
352: motion of the atoms the harmonic approximation evaluated at the
353: calculated minimum energy configuration. Finally utilization of a
354: dielectric continuum approximation for the outer sphere contributions
355: to the chemical potential gives a natural, though approximate,
356: quasi-chemical
357: model\cite{rempe:99,pratt:98,feature,martin:97,pratt:99,G98}.
358:
359: \begin{figure}[t!]
360: \begin{center}
361: \leavevmode
362: \includegraphics[scale=0.65]{diff_Na_12283.eps}
363: \end{center}
364: \caption{Mean-square displacement of the Na$^+$ ion plotted with respect
365: to the time interval analyzed. Analysis of the slope from 200-400 ps gives
366: a diffusion constant of 1.0$\times$10$^{-5}$ cm$^2$/s.}
367: \label{diff:fig}
368: \end{figure}
369:
370:
371: For the present problem, the quasi-chemical approximation was implemented
372: following precisely the procedures of the earlier study of
373: Li$^+$(aq)\cite{rempe:99}, except that the sodium ion cavity radius
374: for the dielectric model calculation was assigned as
375: R$_{Na^+}$=3.1~\AA, the distance of the first minimum of the radial
376: distribution function of Fig.~\ref{gor:fig}. The temperature and
377: density used were 344~K and 1.0~g/cm$^3$ and the value of the bulk
378: dielectric constant was 65.3\cite{uematsu:80}.
379:
380: Results of the calculations are summarized in Fig.~\ref{qc:fig}. The
381: electronic structure results are consonant with those found previously
382: for the Li$^+$ ion. The n=4 inner sphere gas-phase complex has the
383: lowest free energy. Although outer sphere placements are obtained for
384: additional water molecules in the minimum energy structures of larger
385: clusters, attention is, nevertheless, here restricted to inner sphere
386: structures. The mean occupation number predicted by this
387: quasi-chemical model is $\langle n \rangle$ = 4.0; the computed
388: absolute hydration free energy of the Na$^+$ ion under these
389: conditions is -103~kcal/mol, not including any repulsive force
390: (packing) contributions. An experimental value for Na$^+$ ion in
391: liquid water at room temperature is -87~kcal/mol\cite{marcus:94}.
392:
393: \begin{figure}[b!]
394: \vspace{0.2in}
395: \begin{center}
396: \leavevmode
397: \includegraphics[scale=0.45]{na_qc_new.eps}
398: \end{center}
399: \caption{Free energies for Na$^+$ ion hydration in liquid water as a
400: function of the number of inner shell water neighbors at T=344~K and
401: $\rho_{\mathrm{H_2O}}$=1~g/cm$^3$. The lowest results (open
402: diamonds) show quasi-chemical approximate values for the liquid,
403: labelled according to the quasi-chemical interpretation. This graph
404: indicates that the n=4 inner sphere structure is most probable under
405: these conditions. The radius used for the Na$^+$ ion here is 3.1~\AA,
406: though a substantial reduction of this value produced only a minor
407: change in the inferred absolute hydration free energy; otherwise the
408: procedure is the same as in previous
409: reports\protect\cite{rempe:99,pratt:99}. The absolute hydration free
410: energy predicted here is -103~kcal/mol. The results marked
411: $\Delta$G$^{(0)}$ (filled circles) are the free energies predicted for
412: the reaction Na$^+$ + n~H$_2$O in an ideal gas at p = 1~atm
413: $\equiv{\bar p}$ and T=344~K. The minimum value is at n=4. The middle
414: graph (crosses) add to the ideal gas results the `replacement'
415: contribution reflecting the formal density of the
416: water molecules $\mathrm{-n RT}\, \ln \left[ \mathrm{RT}
417: \rho_{\mathrm{H_2O}} / {\bar p} \right] = \mathrm{-n*5.03}$
418: kcal/mol with T=344~K, and $\rho_{\mathrm{H_2O}}$ =1~g/cm$^3$.}
419: \label{qc:fig}
420: \end{figure}
421:
422: Because of the significance of $x_0$ [Eq.~\ref{gqca}], we fitted
423: several model distributions \{$x_n$\} based on different ideas to the
424: `ab initio' molecular dynamics results. The varying success of those
425: models in inferring $x_0$ was enlightening. An instructive selection of those models is
426: shown in Fig.~\ref{inf:fig} and we describe those results here.
427:
428: First, we note that though the preceeding quasi-chemical approximation
429: does not agree closely with the 'ab initio' molecular dynamics simulation, the populations
430: obtained from the quasi-chemical approximation, $\hat x_n$, can serve
431: as a default model for a maximum entropy inference of
432: $x_n$\cite{hummer:cp:2000}. In this approach we model
433: \begin{equation}
434: \ln x_j=\ln \hat x_j -\lambda _0-j\lambda _1-j(j-1)\lambda _2/2 -
435: \ldots~,
436: \label{maxent}
437: \end{equation}
438: with Lagrange multipliers $\lambda_k$ adjusted to conform to
439: available moment information
440: \begin{equation}
441: \left\langle {n \choose j} \right\rangle =\sum\limits_k {x_k}{k
442: \choose j}
443: \label{moments}
444: \end{equation}
445: for j = 0, 1, 2, \ldots. In view of the limited data available, use
446: of more than two moments produced operationally ill-posed fitting
447: problems.
448:
449: One difficulty with this specific approach is that the `ab initio'
450: molecular dynamics produced $x_7\!\!>$0 in contrast to the electronic
451: structure methods that found no minimum energy
452: hepta-coordinated inner-sphere clusters. Since
453: the observed $x_7$ is likely to be relatively less accurate and is
454: furthest away from the desired n=0 element, we excluded n=7
455: configurations of the `ab initio' molecular dynamics, renormalized
456: the probabilities $x_n$, and recalculated the moments.
457: As the upper panel in Fig.~\ref{inf:fig} shows, this simple maximum
458: entropy model is qualitatively satisfactory although not
459: quantitatively convincing. The fitted model significantly disagrees
460: with the observed $x_3$. The chemical contribution suggested by
461: Fig.~\ref{inf:fig} is approximately -70~kcal/mol. Using the Born
462: formula, $-q^2(1 - 1/\epsilon)/2R$ with R=3.12~\AA, to estimate the
463: outer sphere contributions represented by the last term in
464: Eq.~\ref{gqca}, then the net absolute hydration free energy falls in
465: the neighborhood of -115~kcal/mol. Since experimental values for the
466: absolute hydration free energy at room temperature center around
467: -90~kcal/mol, this comparison shows that the present free energy
468: results are not to be interpreted quantitatively, but rather as
469: indicative of the present state of the theory.
470:
471: \begin{figure}[t!]
472: \begin{center}
473: \leavevmode
474: \includegraphics[scale=0.7]{fit_all.eps}
475: \end{center}
476: \caption{Results for the inference of $x_0$ from `ab initio' molecular
477: dynamics information. The solid points represent the information
478: extracted from the molecular dynamics simulation, the dotted lines are
479: the default models, and the solid lines show the fit achieved by the
480: information theory approach. In the top panel, the hepta-occupancy was
481: excluded, the probabilities were renormalized on this condition, and
482: the quasi-chemical default model was used together with the moments
483: $\langle n \rangle$=4.633 and $\langle n(n-1)/2 \rangle\rangle$=8.577.
484: In the middle panel, all $x_n$ observed in the simulation were
485: included, $\langle n \rangle$=4.642 and $\langle n(n-1)/2
486: \rangle\rangle$=8.624, and the ideal gas (or Gibbs default) model was
487: used along with the same moments as above. The bottom panel shows the
488: results using the inner sphere radius R=2.68~\AA\ and the moments
489: $\langle n \rangle$=4.046, and $\langle n(n-1)/2 \rangle\rangle$=6.393
490: with the Gibbs default model. }
491: \label{inf:fig}
492: \end{figure}
493:
494: A second approach focused on testing a default model that supplies a
495: nonzero ${\hat x}_7$; we used the Gibbs default model $\hat{x}_n
496: \propto 1/n!$ that would give the correct answer for an ideal gas
497: `solvent.' This model has the additional and heuristic advantage of being
498: significantly broader. Our experience has been that these maximum
499: entropy fitting procedures work better when the default model is
500: broader than the distribution sought. The results, illustrated in the
501: middle panel of Fig.~\ref{inf:fig}, show an improved fit. Here the
502: chemical contribution to the free energy is -23~kcal/mol, yielding a
503: net absolute hydration free energy of -68~kcal/mol when the same Born
504: formula is used to estimate the outer sphere contributions.
505:
506: A third fitting possibility was based on a suggestion from a previous
507: `ab initio' molecular dynamics calculation on K$^+$(aq): that the
508: innermost {\em four} water molecules have a special
509: status\cite{ramaniah:98}. In fact, the quasi-chemical approximation above
510: and the fitting of the upper panel of Fig.~\ref{inf:fig} suggests also
511: that the $x_n$ results for n$\le$4 and for n$\ge$5 display
512: different behaviors. The radial distribution
513: function of Fig.~\ref{gor:fig}, somewhat better resolved than
514: heretofore, is relevant to this issue and, in contrast, doesn't
515: directly support a hypothesis of two populations of water molecules in
516: the inner shell. Nevertheless, that g(r) does not rule out the possibility
517: that the structures might become more flexible as the inner shell
518: nears maximum capacity with lower incremental binding energies.
519:
520: To clarify these possibilities, we reduced the radius defining the
521: inner sphere to R=2.68~\AA, for which $<$n$>$ is close to 4 (see
522: bottom panel of Fig.~\ref{gor:fig}) and reanalysed the `ab initio'
523: molecular dynamics trajectory to extract the appropriate alternative
524: moment information. Again using the Gibbs default model, we obtained
525: the results shown in the lowest panel of Fig.~\ref{inf:fig}. The
526: inferred chemical contribution is $RT\ln x_0 \approx$~-13~kcal/mol.
527: Using again the Born approximation for the outer sphere
528: contribution, this time with R=2.68~\AA, we obtain an absolute
529: hydration free energy estimate of -65~kcal/mol.
530:
531: The insensitivity of these latter results to choice of inner sphere
532: radius deserves emphasis and further discussion. From a formal point
533: of view, the inner sphere radius R serves only a bookkeeping role;
534: the left side of Eq.~\ref{gqca} should be strictly unaffected by
535: changes in R. Nevertheless, the terms on the right side of that
536: equation are individually affected by changes in R. Thus, we might
537: take the insensitivity of the sum of those individual terms as an
538: indication that the inevitable approximations are reasonably balanced.
539: Values of R for which the sum Eq.~\ref{gqca} is insensitive are
540: pragmatic values given the approximations made.
541: Fig.~\ref{twomcut:fig} illustrates these points and establishes the
542: pragmatic value R$\approx$3.06\AA\ for the current application. The
543: similarity of this value with the radius of the inner shell suggested
544: by Fig.~\ref{gor:fig} (3.12\AA) is encouraging. The value -68~kcal/mol
545: is then suggested for the hydration free energy in the absence of any
546: account of packing or van der Waals interactions.
547:
548: \begin{figure}[tb!]
549: \begin{center}
550: \leavevmode
551: \includegraphics[scale=0.7]{twomcut.eps}
552: \end{center}
553: \caption{Variation of hydration free energy contributions with changes
554: in radius R defining the inner sphere. The upper curve is
555: the chemical contribution obtained with an information theory fit
556: using a Gibbs default. The middle curve is the outer sphere
557: contribution approximated by the Born formula for a spherical ion with
558: unit charge at its center. The bottom curve is the sum of the other
559: two. The slope of the bottom curve is zero at R$\approx$3.06\AA. As
560: discussed in the text, this value identifies a R-region for which the
561: approximations used are pragmatically balanced. The curve takes the
562: value -68~kcal/mol in that region.}
563: \label{twomcut:fig}
564: \end{figure}
565:
566:
567: \section{Conclusions} The `ab initio' molecular dynamics simulation
568: predicts the most probable occupancy of the inner shell of Na$^+$(aq)
569: to be 5 and the mean occupancy to be 4.6 water molecules at infinite
570: dilution, T=344~K, and a nominal water density of 1~g/cm$^3$. The
571: simulation produces both a satisfactory Na-O radial distribution
572: function and self-diffusion coefficient for Na$^+$, but these
573: satisfactory results required more care with thermalization and
574: averaging time than is most common with these demanding calculations. Recently,
575: this point has been separately emphasized in the context of `ab initio'
576: simulation of water\cite{sorenson:00}.
577:
578: The complementary calculation framed in terms of quasi-chemical theory
579: based on electronic structure results for ion-water clusters, the
580: harmonic approximation for cluster motion, and a dielectric continuum
581: model for outer sphere contributions underestimates the inner shell
582: water molecule occupancies for Na$^+$ in liquid water. Maximum entropy
583: fitting of the inner shell occupancy distribution shows that the
584: ion-water cluster results yield a distribution significantly narrower
585: than that obtained from the simulations. For this reason, naive
586: inference of the absolute hydration free energy of Na$^+$(aq) based on
587: the cluster electronic structure results and utilizing information
588: gleaned from the `ab initio' molecular dynamics was unsuccessful. The
589: electronic structure calculations found minimum energy
590: Na(H$_2$O)$_7{}^+$ clusters only with obvious outer sphere placements
591: of some water molecules though hepta-coordinate inner sphere clusters were
592: observed in the `ab initio' molecular dynamics with the most natural
593: cluster definition. These results suggest that the anharmonicities and
594: large amplitude motion are serious concerns, particularly for the
595: larger clusters, and that the approximate theory utilized for outer
596: sphere contributions must treat cluster conformations differently
597: from minimum energy structures of the isolated clusters.
598:
599: Abandonment of the cluster electronic structure results in favor of a
600: broader default model improved the modeling of the $x_n$ distribution on
601: the basis of the information extracted from the simulation. A sequence
602: of more aggressive fits eventually suggested the value -68~kcal/mol for the
603: hydration free energy at this somewhat elevated temperature on the
604: basis of the quasi-chemical perspective of inner sphere occupancies but in the
605: absence of any account of packing or van der Waals interactions.
606:
607: We acknowledge helpful discussions of many related issues with Gerhard
608: Hummer and Joel Kress. This work was supported by the US Department of
609: Energy under contract W-7405-ENG-36 and the LDRD program at Los
610: Alamos.
611:
612:
613: \begin{thebibliography}{10}
614:
615: \bibitem{kchannel:98}
616: D.~A. Doyle, J.~M. Cabral, R.~A. Pfuetzner, A.~L. Kuo, J.~M. Gulbis,
617: S.~L. Cohen, B.~T. Chait, and R. MacKinnon,
618: \newblock {\em Science} 280 (1998) 69--77.
619:
620: \bibitem{guidoni:99}
621: L. Guidoni, V. Torre, and P. Carloni,
622: \newblock {\em Biochem.} 38 (1999) 8599--8604.
623: %md simulation; hydration
624:
625: \bibitem{laio:99}
626: A. Laio and V. Torre,
627: \newblock{\em Biophys. J.} 76 (1999) 129--148.
628: %langevin equ and Kramer rxn rate theory; hydration
629:
630: \bibitem{caminiti:80}
631: R. Caminiti, G. Nacheri, G. Paschina, G. Piccaluga, and G. Pinna,
632: \newblock{\em J. Chem. Phys.} 72 (1980) 4522--4528.
633:
634: \bibitem{skipper:89}
635: N. T. Skipper and G. W. Nielson,
636: \newblock{\em J. Phys. Condens. Matter} 1 (1989) 4141--4154.
637:
638: \bibitem{Michaellian:78}
639: K.~H. Michaellian and M. Moskovits,
640: \newblock {\em Nature} 273 (1978) 135--136.
641:
642: \bibitem{Ohtaki:93}
643: H. Ohtaki and T. Radnai,
644: \newblock {\em Chem. Rev.} 93 (1993) 1157--1204.
645:
646:
647: \bibitem{Heinzinger:79}
648: K. Heinzinger, and G. P{\'a}link{\'a}s,
649: \newblock {\em in THE CHEMICAL PHYSICS OF SOLVATION}
650: \newblock Elsevier, Amsterdam, 1985; pp. 313.
651:
652: \bibitem{Mezei:81}
653: M. Mezei and D.~L. Beveridge,
654: \newblock {\em J. Chem. Phys.} 74 (1981) 6902--6910.
655:
656: \bibitem{Impey:83}
657: R.~W. Impey, P.~A. Madden, and I.~R. McDonald,
658: \newblock {\em J. Phys. Chem.} 87 (1983) 5071--5083.
659:
660: \bibitem{Chandrasekhar:84}
661: J. Chandrasekhar, D.~C. Spellmeyer, and W.~L. Jorgensen,
662: \newblock {\em J. Am. Chem. Soc.} 106 (1984) 903--910.
663:
664: \bibitem{bounds:85}
665: D.~G. Bounds,
666: \newblock {\em Mol. Phys.} 54 (1985) 1335--1355.
667:
668: \bibitem{wilson:85}
669: M.~A. Wilson, A. Pohorille, and L.~R. Pratt,
670: \newblock{\em J. Chem. Phys.} 83 (1985) 5832--5836.
671:
672: \bibitem{Zhu:91}
673: S.~B. Zhu and G.~W. Robinson,
674: \newblock {\em Z. Naturforsch. A} 46 (1991) 221--228.
675:
676: \bibitem{Heinzinger:93}
677: K. Heinzinger,
678: \newblock in M. U. Palma, M. B. Palma-Vittorelli, and F. Patak (Eds.),
679: {\em WATER-BIOMOLECULE INTERACTIONS}
680: \newblock SIF, Bologna, 1993, pp. 23--30.
681:
682: \bibitem{lee:94}
683: S.~H. Lee and J.~C. Rasaiah,
684: \newblock {\em J. Chem. Phys.} 101 (1994) 6964--6974.
685:
686: \bibitem{toth:96}
687: G. Toth,
688: \newblock {\em J. Chem. Phys.} 105 (1996) 5518--5524.
689:
690: \bibitem{obst:96}
691: S. Obst and H. Bradaczek,
692: \newblock {\em J. Phys. Chem.} 100 (1996) 15677--15687.
693:
694: \bibitem{Koneshan:98b}
695: S. Koneshan, J.~C. Rasaiah, R.~M. Lynden-Bell, and S.~H. Lee,
696: \newblock {\em J. Phys. Chem. B} {\bf 1998}, 102, 4193--4204.
697:
698:
699: \bibitem{schwegler:00}
700: J.~A. White, E. Schwegler, G. Galli, and F. Gygi,
701: \newblock {\em personal communication} {\bf 2000}.
702:
703: \bibitem{marx:99} D. Marx, in C. Caccamo, J.-P. Hansen, and G. Stell (Eds.),
704: \newblock{in NEW APPROACHES TO PROBLEMS IN
705: LIQUID STATE THEORY}, Kluwer, Dordrecht, 1999, 439--458.
706:
707: \bibitem{alfe:2000}
708: D. Alf\`{e}, G.~E. De~Wijs, G. Kresse, and M.~J. Gillan,
709: \newblock{\em Int. J. Quant. Chem.} 77 (2000) 871-879.
710:
711: \bibitem{rempe:99}
712: S.~B. Rempe, L.~R. Pratt, G. Hummer,J.~D. Kress, R.~L. Martin, and A.
713: Redondo,
714: \newblock {\em J. Am. Chem. Soc.} 122 (2000) 966--967.
715:
716: \bibitem{pratt:98}
717: L.~R. Pratt and R.~A. LaViolette,
718: \newblock {\em Mol. Phys.} 94 (1998) 909--915.
719:
720: \bibitem{feature}
721: G. Hummer, L.~R. Pratt, and A.~E. Garc{\'{\i}}a,
722: \newblock {\em J. Phys. Chem. A} 102 (1998) 7885--7895.
723:
724: \bibitem{martin:97}
725: R.~L. Martin, P.~J. Hay, and L.~R. Pratt,
726: \newblock {\em J. Phys. Chem. A} 102 (1998) 3565--3573.
727:
728: \bibitem{pratt:99}
729: L.~R. Pratt, and S.~B. Rempe,
730: \newblock in L. R. Pratt and G. Hummer (Eds.), {\em SIMULATION AND
731: THEORY OF ELECTROSTATIC INTERACTIONS IN SOLUTION}
732: \newblock AIP, New York, 1999, 172--201.
733:
734: \bibitem{G98}
735: M.~J. Frisch, {\em et al}.
736: \newblock {\em Gaussian 98 (Revision A.2)}.
737: \newblock Gaussian, Inc., Pittsburgh PA, 1998.
738:
739: \bibitem{vasp1}
740: G. Kresse and J. Hafner,
741: \newblock {\em Phys. Rev. B} 41 (1993) 558--561.
742:
743: \bibitem{vasp2}
744: G. Kresse and J. Hafner,
745: \newblock {\em J. Phys.: Condens. Mat.} 6 (1994) 8245--8257.
746:
747: \bibitem{hertz:73}
748: H. G. Hertz,
749: in F. Franks (Ed.), {\em WATER: A COMPREHENSIVE
750: TREATISE} Vol.~3, Plenum Press, New York, 1973.
751:
752: \bibitem{hummer:cp:2000}
753: G. Hummer, S. Garde, A.~E Garc\'{i}a, and L.~R. Pratt,
754: \newblock{\em Chem. Phys.} ``New Perspectives on Hydrophobic Effects''
755: (in press 2000), LA-UR-00-222.
756:
757: \bibitem{pratt:NATO:1999}
758: L. R. Pratt, S. Garde, and G. Hummer in C. Caccamo, J.-P. Hansen, and G. Stell (Eds.),
759: \newblock{in NEW APPROACHES TO PROBLEMS IN
760: LIQUID STATE THEORY}, Kluwer, Dordrecht, 1999, 407--420.
761:
762: \bibitem{uematsu:80}
763: E. Uematsu and E.~U. Franck,
764: \newblock{\em J. Phys. Chem. Ref. Data} 9 (1980) 1291--1305.
765:
766: \bibitem{marcus:94}
767: Y. Marcus,
768: \newblock {\em Biophys. Chem.} 51 (1994) 111--127.
769:
770: \bibitem{ramaniah:98}
771: L. Y. Ramaniah, M. Bernasconi, and M. Parrinello
772: \newblock{\em J. Chem. Phys.} 109 (1998) 6839--6843.
773:
774: \bibitem{ramaniah:99}
775: L. Y. Ramaniah, M. Bernasconi, and M. Parrinello
776: \newblock{\em J. Chem. Phys.} 109 (1998) 1587--1591.
777:
778: \bibitem{sorenson:00}
779: J. M. Sorenson, G. Hura, R. M. Glaeser, and T. Head-Gordon,
780: ``What Can X-Ray Scattering Tell Us about the Radial Distribution
781: Function of Water?'' preprint (2000).
782:
783:
784:
785: \end{thebibliography}
786:
787:
788: % If in two-column mode, this environment will change to single-column
789: % format so that long equations can be displayed. Use
790: % sparingly.
791: %\begin{widetext}
792: % put long equation here
793: %\end{widetext}
794:
795: % figures should be put into the text as floats.
796: % Use the graphics or graphicx packages (distributed with LaTeX2e).
797: % See the LaTeX Graphics Companion by Michel Goosens, Sebastian Rahtz,
798: % and Frank Mittelbach for instance.
799: %
800: % Here is an example of the general form of a figure:
801: % Fill in the caption in the braces of the \caption{} command. Put the label
802: % that you will use with \ref{} command in the braces of the \label{} command.
803: %
804: % \begin{figure}
805: % \includegraphics{}%
806: % \caption{}
807: % \label{}
808: % \end{figure}
809:
810: % tables follow here or maybe be put in the text
811: %
812: % Here is an example of the general form of a table:
813: % Fill in the caption in the braces of the \caption{} command. Put the label
814: % that you will use with \ref{} command in the braces of the \label{} command.
815: % Insert the column specifiers (l, r, c, d, etc.) in the empty braces of the
816: % \begin{tabular}{} command.
817: % The ruledtabular enviroment adds doubled rules to table and sets a
818: % nice set of default table settings.
819: % Use the table* environment to get a full-width table in two-column
820: % \begin{table}
821: % \caption{}
822: % \label{}
823: % \begin{ruledtabular}
824: % \begin{tabular}{}
825: % \end{tabular}
826: % \end{ruledtabular}
827: % \end{table}
828:
829: % If you have acknowledgments, this puts in the proper section head.
830: %\begin{acknowledgments}
831: % put your acknowledgments here.
832: %\end{acknowledgments}
833:
834: % Create the reference section using BibTeX:
835: \bibliography{your bib file}
836:
837: \end{document}
838: %
839: % ****** End of file template.aps ******
840:
841: