THE JOURNAL OF CHEMICAL PHYSICS 135, 124505 (2011)

Proton transfer and the mobilities of the H+ and OH− ions from studies
of a dissociating model for water
Song Hi Lee1,a) and Jayendran C. Rasaiah2,b)
1
Department of Chemistry, Kyungsung University, Pusan 608-736, South Korea
2
Department of Chemistry, University of Maine, Orono, Maine 04469, USA
(Received 1 June 2011; accepted 12 August 2011; published online 26 September 2011)

Hydrogen (H+ ) and hydroxide (OH− ) ions in aqueous solution have anomalously large diffusion co-
efficients, and the mobility of the H+ ion is nearly twice that of the OH− ion. We describe molecular
dynamics simulations of a dissociating model for liquid water based on scaling the interatomic po-
tential for water developed by Ojamäe-Shavitt-Singer from ab initio studies at the MP2 level. We use
the scaled model to study proton transfer that occurs in the transport of hydrogen and hydroxide ions
in acidic and basic solutions containing 215 water molecules. The model supports the Eigen-Zundel-
Eigen mechanism of proton transfer in acidic solutions and the transient hyper-coordination of the
hydroxide ion in weakly basic solutions at room temperature. The free energy barriers for proton
transport are low indicating significant proton delocalization accompanying proton transfer in acidic
and basic solutions. The reorientation dynamics of the hydroxide ion suggests changes in the propor-
tions of hyper-coordinated species with temperature. The mobilities of the hydrogen and hydroxide
ions and their temperature dependence between 0 and 50 ◦ C are in excellent agreement with exper-
iment and the reasons for the large difference in the mobilities of the two ions are discussed. The
model and methods described provide a novel approach to studies of liquid water, proton transfer,
and acid-base reactions in aqueous solutions, channels, and interfaces. © 2011 American Institute of
Physics. [doi:10.1063/1.3632990]

I. INTRODUCTION and diffusion of hydrogen ions, with forces computed “on the
fly” using density functional theory (DFT) and the BLYP ex-
Proton transfer is of importance in many chemical re-
change functional. Similar methods were used by Tuckerman
actions and biomolecular processes. The structural diffusion
et al.3, 39 to study hydroxide ions in aqueous solution. The na-
of H+ and OH− ions arises mainly from proton transfer (PT)
ture of the hydrogen ion in aqueous solution, consisting of
between a solvated ion and a neighboring water molecule. It
the Eigen (E) and Zundel (Z) complexes and other interme-
has been studied extensively by ab initio molecular dynamics
diates, and the mechanism of transport that involves proton
(AIMD) (Refs. 1–6) and applications of extended valence
transfer from Eigen to Zundel to Eigen structures (the Eigen-
bond theory (EVB).7–12 A reactive molecular dynamics algo-
Zundel-Eigen (EZE) sequence) driven by solvent fluctuations
rithm has also been proposed to study proton transport.13 The
followed by “presolvation” is well understood,1, 2, 27, 34–36 but
solvation structures of H+ and OH− ions have been probed
there have been conflicting opinions about the structure of the
by neutron14–20 and x-ray diffraction,21 x-ray adsorption,22
hydrated hydroxide ion and the mechanism of transport.3–5, 45
and spectroscopy experiments23–32 that explore the structures
The BLYP functional in DFT calculations favors four-
and excitations (vibrational, rotational, and electronic) of
coordinated hydration of OH− at room temperature,3 while
equilibrium and transition states in proton transfer reactions.
a different functional (PW91) supports a three-coordinated
A theoretical framework that relates microscopic states to the
complex5 formed by hydrogen bonding with water molecules
mechanism and kinetics of PT reactions, and predicts a set
along the three lone pairs of the oxygen atom in OH to pro-
of lifetimes and rates of formation and decay of intermediate
duce a canonical Lewis structure. Four-coordinated hydration,
states that can be verified by time-resolved spectroscopy has
postulated by the BLYP functional, has its origin in a delo-
been proposed.33 Accounts and commentaries describing
calized ring of charge that replaces the directed lone pairs.3
mechanisms and simulations of PT have appeared,34–43 and
The oxygen atom accepts an additional water molecule via
a comprehensive review of structural diffusion in the context
H-bonding to form a roughly square planar arrangement of
of PT, with special reference to the hydroxide ion covering
water molecules on the side opposite to the hydrogen atom of
the field up to mid-2009 was published recently.44
the OH− ion to form OH− (H2 0)4 . The hydrogen atom of this
It is well known that the diffusion coefficients of the hy-
ion is also transiently H-bonded to a fifth water molecule and
drogen and hydroxide ions are anomalously large and the
on average, a hyper-coordinated complex with four H-bonds
mobility of the H+ ion is nearly twice that of the hydrox-
accepted and one donated is postulated with an overall coor-
ide ion. Marx et al.2 employed AIMD to study the solvation
dination number of 4.5.3
The difference in solvation structures postulated by
a) Electronic mail: [email protected]. the BLYP and PW91 functionals, translates into different
b) Electronic mail: [email protected]. mechanisms for structural diffusion; and the PW91 functional
0021-9606/2011/135(12)/124505/10/$30.00 135, 124505-1 © 2011 American Institute of Physics

Downloaded 27 Sep 2011 to 130.111.64.68. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 124505-2 S. H. Lee and J. C. Rasaiah J. Chem. Phys. 135, 124505 (2011)

predicts a larger OH− ion mobility than the experimental of the OH− suggests that the proportions of these structures
mobility at room temperature.39 A different solvation struc- change with temperature between 0 and 50 ◦ C. The free
ture (also 4-coordinated) is predicted by the HCTH exchange energy barriers for proton transport are small for both H+ and
functional leading to another mechanism for the structural OH− ions and the proton is more significantly delocalized
diffusion of the OH− ion but with a diffusion coefficient that in the weakly acidic solution as shown previously in path
is smaller than that of liquid water which contradicts the integral simulations.3, 44 A Zundel-like transition state occurs
experimental facts.39 Tuckerman et al.39 also showed that for PT towards the OH− with a lifetime comparable to what
the order of the diffusion coefficients for H+ and OH− is has been reported experimentally. The calculated diffusion
reversed when the PW91 functional is used in DFT/AIMD coefficients as a function of temperature between 0 and 50 ◦ C
simulations with the diffusion coefficient of the hydroxide are in good agreement with experiment.
ion greater than that of the hydrogen ion. Of the three, the This paper is organized as follows. In Sec. II we describe
BLYP functional gives the correct experimental order. scaling of the OSS2 model for water. Section III discusses the
There is experimental evidence in support of both 3- and use of the scaled model (sOSS2) to study proton transfer in
4- (or hyper-)coordinated species with suggestions that their acidic and basic solutions and details of the solvation struc-
proportions vary with temperature such that the 4-coordinated tures during proton transfer. Section IV is about the reorienta-
structure is dominant at room temperature.19, 20, 28 Chandra tion dynamics of the solvated OH− ion, and Sec. V discusses
et al. analyzed the available experimental data using their the- the mobility of the H+ and OH− ions at 0.26 M as a function
oretical framework and found that the experimental lifetimes of temperature from 0 to 50 ◦ C and makes comparisons with
at room temperature were consistent with the predictions experiment. The conclusions are in Sec. VI.
of the BLYP functional supporting the hyper-coordinated
structure of the OH− ion and a mechanism of struc-
II. SCALING THE OSS2 MODEL
tural diffusion (“dynamic hyper-coordination”) based on this
structure.33, 44 To describe the scaling, suppose that T’ is the tempera-
Newer and improved density functionals have been pro- ture at which the equilibrium properties of the OSS2 model
posed for water,46 but the effect of temperature on accuracy match those of real water at room temperature (T = 298.15
of the AIMD simulations of PT using a specific exchange K) and density, where we observe from our simulations that
functional is difficult to predict.47 Ufimstev et al. proposed a T > T. Multiplying the OSS2 potential VOSS2 by the factor
charged ring model to represent the delocalized ring of charge β  = 1/kT in the partition function (k is the Boltzmann’s con-
in the OH− ion.48 The model predicts a distribution of 4- and stant) is equivalent to scaling the OSS2 potential by a factor
3-coordinated complexes and provides the base states of a λ = T/T at the temperature T as in thermodynamic perturba-
multistate EVB force field in their studies of proton transfer tion theory.54, 55 The scaled OSS2 potential is thus defined by
in basic solution. Like other EVB methods, the computations VsOSS2 = λVOSS2 ; where the parameter λ determines VsOSS2
are less demanding than ab initio calculations. from the OSS2 potential.49 Small changes in λ are then made
Here we discuss a different model that uses a single to get the optimal results as described below. The method is
force field to study the dynamics of proton transfer and the deceptively simple but has only a partial quantum mechani-
solvation structure of both hydrogen and hydroxide ions in cal basis since it is obtained by scaling the OSS2 potential
aqueous solution. Our calculations are based on scaling the derived from ab initio simulations at the MP2 level.49
Ojamäe-Shavitt-Singer (OSS) (Ref. 49) potential for water, The OSS2 potential model has been described by Ojamäe
and are easily adapted to conventional molecular dynamics et al.49 and details of our simulation methods for this po-
(MD) simulations even on a large scale. The OSS2 potential tential are summarized in Ref. 53. The same methods were
is a central force model50–52 for water that incorporates used here for the sOSS2 potential. Our MD simulations were
dissociation into hydrogen and hydroxide ions, and was performed in the NVT ensemble with the number of water
derived from ab initio studies of protonated water dimers molecules N ∼ 215 and the temperature controlled by a Nose-
and clusters at the MP2 level. However, the original OSS2 Hoover thermostat.56 The new feature is the scaling factor λ
model is unsuitable for the bulk liquid since it represents that defines the scaled potential VsOSS2. This requires scal-
a supercooled or glassy state with a negligible diffusion ing the charge qi of particle i by the square root of λ (qi 
coefficient.53 Normal diffusion in the bulk phase emerges = λ1/2 qi ) in calculations of the electrostatic energy, while the
at lower densities (∼0.5 g cm−3 ) and higher temperatures induced dipole moment at each oxygen site is obtained self-
(∼500 K) (Ref. 53) suggesting that the OSS2 potential can be consistently in the same way as it was before scaling.49, 53 The
scaled for use under ambient conditions. We accomplish this other three non-electrostatic potential energies (VOH , VOO ,
by exploiting the coupling between the inverse temperature and VHOH ) in the OSS2 potential were also scaled by λ. Ewald
and the potential in the partition function and call it the summations are used in our simulations with the parameter for
scaled OSS2 (or sOSS2) model. The model supports the κ = 5.0/L and the real-space cut distance rcut and Kmax chosen
EZE mechanism of proton transfer in weakly acidic aqueous as 0.5L and 7, respectively, where L is the length of the box
solutions and a mix of hyper-coordination (4 accepted and 1 (∼18.64 Å for 216 water). The double summations in recip-
weakly donated) and four coordination (4 accepted and none rocal space, which cannot be reduced to a single summation
donated or 3 accepted and 1 donated) of the hydroxide ion in due to the cutoff functions, were ignored. This is reasonable
weakly basic solutions at room temperature with an average as the distances in reciprocal space are larger than the length
coordination number of ∼4.6. The reorientation dynamics L of the box. The velocity Verlet algorithm57 was employed

Downloaded 27 Sep 2011 to 130.111.64.68. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 124505-3 Proton transfer, H+ and OH− ion mobilities J. Chem. Phys. 135, 124505 (2011)

15 4 TABLE II. Relaxation times (τ in ps) and activation energies Ea (kJ/mol)

T = 300 K Exp.
for reorientation of the OH bond in the sOSS2 model for water and OH− ion
3.5
T = 450 K
sOSS2 in sOSS2 water between 0 and 50 ◦ C.
T = 540 K

D (10 -5cm2/sec)
10 3
MSD (A )
2

2.5
Temp. Water (OH bond) OH− ion
(◦ C) (ps) (ps)
5 2

1.5 0 4.8 ± 0.5 11.3 ± 0.6

(a) (b)
10 3.9 ± 0.5 6.9 ± 0.6
0 1
0 2 4 6 8 10 0 10 20 o 30 40 50 20 3.0 ± 0.5 4.2 ± 0.4
t (ps) t ( C)
25 2.6 ± 0.5 3.2 ± 0.4
FIG. 1. (a) Mean square displacements (MSD) of the OSS2 water at sev- 30 2.4 ± 0.4 2.9 ± 0.3
eral different temperatures. Solid line: at 300 K, dotted line: at 450 K, and 40 1.9 ± 0.5 1.9 ± 0.4
long-dashed line: 540 K all at a density of 0.9970 g/cm3 . (b) Comparison of 50 1.7 ± 0.4 1.7 ± 0.4
self-diffusion coefficients of the sOSS2 water at several different tempera- Ea (kJ/mol) 16 ± 2 17 ± 2/34 ± 2a
tures. : experimental values (Refs. 60 and 61), and ●: calculated values for
216 water molecules using the sOSS2 potential. The error bar represents a a
The two activation energies for the OH− ion are determined from the 3 lower (30–
standard deviation. 50 ◦ C) and 3 higher (0–20 ◦ C) relaxation times, respectively, see Fig. 2.

for time integration with a time step of 1 fs. The equilibrium In principle λ has to be re-determined at each tempera-
properties are averaged over 60–100 blocks of 100 000 time ture T from a new T , but the change in λ with temperature is
steps and the configuration of all the atoms is stored every small because it is the ratio of two temperatures (λ = T/T )
10 time steps for further analyses. The simulations were first that is less sensitive to changes in temperature than the in-
validated by checking our results against Ojamäe’s work for dividual temperatures. Figure 1(b) and Table I show that the
pure water using the OSS2 model.49 The calculated oxygen- experimental diffusion coefficients60, 61 of pure water can be
hydrogen (O-H) radial distribution function and the hydration predicted to high accuracy between 0 to 50 ◦ C (273 to 323 K)
number n(r) for hydrogen in the 216 molecule pure water sys- using a single λ = 0.530 determined at 298.15 K. The acti-
tem were nearly identical,58 even though Ojamäe et al.49 used vation energy for translational diffusion calculated from the
a different method for the Ewald sums in the calculation of Arrhenius equation D = D0 exp(−Etrans /RT), is 16 kJ/mol for
the induced dipole moment. the sOSS2 model in close agreement with 18 kJ/mol from ex-
The self-diffusion coefficient is readily determined in a perimental data.60, 61
MD simulation from the mean square displacement (MSD), Besides translation, the reorientation of water
and we use it as a probe to identify the temperature T’ at molecules62 plays an important role in the dynamics of
which OSS2 model water has nearly the same properties as breaking and forming hydrogen bonds in liquid water.63–66
liquid water under ambient conditions. The diffusion coeffi- We calculated the relaxation times ((τ ) for reorientation of
cient D for the OSS2 model at room temperature and fluid the sOSS2 water between 0 to 50 ◦ C (273 to 323 K) from the
density ρ = 0.9970 g/cm3 is near zero (see Fig. 1(a)), but at orientation correlation functions P2 [u(t) · u(0)], where P2
540 K the diffusion coefficient is nearly identical to the ex- is the second order Legendre polynomial and u(t) is the unit
perimental value for water (2.26 × 10−5 –2.29 × 10−5 cm2 /s vector along the OH bond of a water molecule (Table II). The
(Refs. 59–61)) at 298.15 K. From our scaling hypothesis λ agreement with NMR reorientation times62 shown in Fig. 2
= T/T = 298.15/540 = 0.552, and we expect the sOSS2 is excellent. The activation energies for reorientation were
model at 298.15 K to have nearly the same diffusion coef- determined from plots of log(τ ) vs 1/T. They are 16 kJ/mol,
ficient (2.30 ± 0.09 × 10−5 cm2 /s) as the OSS2 model at 540 for rotation about the OH bond in water (sOSS2 model) in
K. The diffusion coefficient at 298.15 K from the MSD (2.00 good agreement with 14 kJ/mol obtained from the NMR data.
± 0.09 × 10−5 cm2 /s) is within 12% of the experimental re- Remarkably the same scaling hypothesis holds for the
sult. Further fine-tuning by choosing λ = 0.530 leads to agree- equilibrium structure as shown in Figs. 3(a), 3(b), and 3(c),
ment (D = 2.27 ± 0.07 × 10−5 cm2 /s) to within 1%. where the atom-atom O–O, O–H, and H–H distribution func-

TABLE I. Diffusion coefficients (10−5 cm2 /s) of H2 O, H+ , and OH− ion between 0 ◦ C and 50 ◦ C. Simulations results
for sOSS2 model for 0.26 M HCl or NaOH compared with experiment.

Temp. (◦ C) (Density (g/cc)) H2 O sim (expt.) H+ ion sim-0.26 M (expt.)a OH− ion sim-0.26 M (expt.)a

0 (0.9999) 1.22 ± 0.08 (1.129) 4.95 ± 1.34 (6.00) 2.77 ± 0.70 (3.15)
10 (0.9997) 1.60 ± 0.08 (1.536) 6.03 ± 1.59 (7.34) 3.44 ± 0.91 (3.95)
20 (0.9982) 2.05 ± 0.08 (2.023) 7.10 ± 1.57 (8.66) 4.19 ± 1.14 (4.84)
25 (0.9970) 2.27 ± 0.12 (2.290) 7.62 ± 1.59 (9.31) 4.56 ± 1.29 (5.30)
30 (0.9957) 2.55 ± 0.12 (2.590) 8.09 ± 1.66 (9.94) 4.87 ± 1.31 (5.76)
40 (0.9922) 3.10 ± 0.15 (3.238) 9.05 ± 1.68 (11.2) 5.54 ± 1.47 (6.69)
50 (0.9881) 3.70 ± 0.26 (3.956) 10.0 ± 1.78 (12.3) 6.24 ± 1.46 (7.61)
Ea (kJ/mol) 16 (18) 10.3 (10.6) 12.0 (13.0)
a
Numbers in parenthesis are the experimental results for H+ and OH− ions at infinite dilution (Ref. 77).

Downloaded 27 Sep 2011 to 130.111.64.68. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 124505-4 S. H. Lee and J. C. Rasaiah J. Chem. Phys. 135, 124505 (2011)

12 For the system containing a single HCl, we have 215 O2- , 431
10 H2O, NMR H+ , and a single Cl− ion. Although O2− is polarizable in the

Relaxation time (ps)

H2O, MD
-
OSS2 model, we treat the Cl− ion as a nonpolarizable anion
8 OH , Exp
OH -, MD
that contributes electrostatic and Lennard-Jones interactions
6 to the potential energy of the system with = 0.5292 kJ/mol,
4
σ Cl-O = 3.823 Å,69 and σ Cl-H = 2.225 Å assuming σ H-H = 0.

2
III. PROTON TRANSFER USING THE sOSS2
0
3 3.2 3.4 3.6 WATER MODEL
1000/T (K -1)
We find that the hydrogen ion exists as a range of struc-
FIG. 2. The orientation relaxation times of water and OH− ion calculated tures in liquid water and PT occurs from a central proton in
from P2 [u(t) · u(0)] as function of 1000/T. The solid line is the experimental the Eigen complex H3 O+ (H2 O)3 through a Zundel intermedi-
NMR result for water (Ref. 62) fitted to τ NMR = 0.34 [(T-223)/223]−1.83 and
the symbols (䉬) are MD calculations for sOSS2 water with the unit vector ate preceded and followed by solvent reorganization as noted
u(t) along the OH bond. The relaxation times for the OH− are from MD previously in AIMD simulations by Marx et al.2 and theoret-
calculations for the ion in sOSS2 water (●), and from CTTS experiments ical discussions by Agmon34, 35 and others.36, 40–43 The Eigen
(Ref. 28) (◦) with the unit vector u(t) along the OH bond.
cation H3 O+ (H2 O)3 (Fig. 4(a)(i)), present before PT occurs,
has a second solvation shell around H3 O+ hydrogen-bonded
to the primary shell. Thermal fluctuations break the H-bond
tions for the sOSS2 potential (VsOSS2 = 0.530 VOSS2 ) are of a water molecule in this shell, and prepare the under-
compared with the experimental results for liquid water from coordinated water molecule in the first shell with the oxygen
neutron and x-ray diffraction data,67 and distribution func- atom designated as O^ from which it was detached, to receive
tions for the rigid SPC/E model for water68 at 298.15 K. The a proton from H3 O+ characterized as “presolvation.”2, 44 This
peak heights are lowered in moving from the OSS2 to the feature is also reproduced in our simulations using the sOSS2
sOSS2 model potential at room temperature but the positions model. Characterizing the oxygen in H3 O+ as O*, the O* and
are nearly the same (not shown). The distribution functions O^ atoms are linked by a hydrogen bond to form the Zundel
of the sOSS2 model are in good agreement with experiment complex [H2 O*-H*-O^H2 ]+ in which the O*-O^ distance has
and are comparable or superior to the SPC/E model, except shrunk from 2.67 Å to 2.36 Å. This “most active” H* shut-
that the peaks corresponding to dissociation are absent in the tles between two O’s (Fig. 4(a)(ii)) in an essentially barrier-
SPC/E model, and the rest are shifted to slightly longer dis- less transition as discussed by Agmon34, 35 and Marx et al.2
tances by ∼0.05–0.06 Å. These results taken together suggest If the O^ returns to its former state by forming a hydrogen
that the sOSS2 model represents the equilibrium structure and bond with an extraneous water molecule the most active H*
transport properties of water quite accurately at and near am- atom goes back to O* and PT does not occur (not shown).
bient temperatures. Alternatively, if the O* atom of the original Eigen complex,
We used the same sOSS2 model, with the predetermined now present in the Zundel form, becomes four coordinated by
scaling of λ = 0.530 for liquid water, to study the mobility of forming a H-bond with a fourth water, the active H* within
H+ and OH− ions in 215 water molecules in the presence of the Zundel complex moves to O^ and PT occurs to form a
either a stationary chloride (Cl− ) or sodium (Na+ ) counterion new H3 O+ ion that is stabilized in the next step by the forma-
to maintain electro-neutrality. For the system containing a tion of an Eigen complex or undergoes one or more successive
single NaOH, we have 216 O2− , 431 H+ , and a single Na+ PTs. A single PT is completed and stabilized with the config-
ion. The H+ ion is not polarizable in the OSS2 model,49 and uration shown in Fig. 4(a)(iii).
we consider the Na+ ion as an additional (432th) cation (like Following Marx et al.2 we select the displacement coor-
H+ ) that contributes electrostatic and Lennard-Jones (LJ) 6- dinate δ = RO*H* -RO^H* to identity the solvated complexes
12 interactions to the total potential with ε = 0.6034 kJ/mol, and determine the free energy profile for PT. Here RO*H* and
σ Na-O = 2.773 Å,69 and σ Na-H = 1.175 Å assuming σ H-H = 0. RO^H* are the respective distances of the shared proton H*

3 3
3
Exp. Exp. Exp.
SPC/E SPC/E SPC/E
sOSS2 2 sOSS2 2 sOSS2
2
gOO(r)

gOH(r)

gHH(r)

1 1 1

(a) (b) (c)

0 0 0
2 4 6 8 0 1 2 3 4 5 1 2 3 4 5
r (A) r (A) r (A)

FIG. 3. (a) O–O radial distribution functions at 298 K. Solid line: the experimental result (Ref. 67), dotted line: the SPC/E model (Ref. 68), and dashed line:
the sOSS2 model. (b) O–H radial distribution functions at 298 K. (c) H–H radial distribution functions at 298 K.

Downloaded 27 Sep 2011 to 130.111.64.68. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 124505-5 Proton transfer, H+ and OH− ion mobilities J. Chem. Phys. 135, 124505 (2011)

(a) 1.82 (b)

1.00 1.63
O^ 1.74 H* 1.18
1.83
O* 0.98
2.22 0.99

2.67 2.73
3.15 |δ|>0.5 |δ|<0.1

(i) (ii)

(iii)

(c) (d)
2.81

1.83 1.00 1.72 H H* 1.21

0.99 0.98

O^ 2.73 O* 2.43 |δ|>0.5 |δ|<0.1

1.81

(i) (ii)

(iii)

FIG. 4. (a) Representative configurations showing the mechanism of single proton transfer leading to the transport of a hydrogen ion in a simulation of one
H3 O+ ion in 215 water molecules using the sOSS2 potential. (i) H3 O+ ion stabilized as an Eigen complex H3 O+ (H2 O)3 showing three coordination of H3 O+
and a water molecule in the second solvation shell. (ii) Loss of water in the second solvation shell in (i) and the formation of the Zundel intermediate H5 O2 + .
(iii) Completion of proton transfer via the Zundel intermediate to form a new H3 O+ ion. (b) The radial distribution functions gO*H and gO*O and corresponding
coordination numbers nO*H and nO*O with respect to the oxygen atom O* of the hydrogen ion H3 O* + in 215 water molecules for configuration in which the
displacement coordinate |δ| > 0.5 Å (left panel) and |δ| < 0.1 Å (right panel). (c). Representative configurations of OH− ion showing the mechanism of single
proton transfer leading to the transport of the hydroxide ion in a simulation of one OH− ion in 215 water molecules using the OSS2 potential. (i) O*H’− (H2 O)4.6
complex showing four coordination of O*. (ii) OH− (H2 O)3 intermediate with H’ between O* and O^ showing three coordination of O*. (iii) Completion of
proton transfer via OH− (H2 O)4.6 intermediate (see the text). (d) The radial distribution functions gO*H and gO*O and corresponding coordination numbers nO*H
and nO*O with respect to the oxygen atom O* of the hydroxide ion O*H− in 215 water molecules for configuration in which the displacement coordinate |δ|
> 0.5 Å (left panel) and |δ| < 0.1 Å (right panel).

from O* and O^. The sequence of events leading to proton tion numbers nO*-H of 2 and 3, respectively, consistent with
transfer can be understood by examining the conditional dis- the presence of two chemical bonds and one hydrogen bond
tribution functions for two sets of configurations for which |δ| associated with the O* atom of the Zundel intermediate. (see
is small (|δ| < 0.1 Å) or large (|δ| > 0.5 Å). A small |δ| (right configuration (ii) of Fig. 4(a)).
panel of Fig. 4(b) and configuration (ii) of Fig. 4(a)) indicates The free energy profile F((δ) shown in Fig. 5(a), for
a potential pathway for proton transfer via a Zundel complex proton transfer within the Zundel state in acidic solution
and a large |δ| (left panel of Fig. 4(b) and configurations (i) of 0.26 M (Figs. 4(a) and 6(a)), was constructed from the
and (iii) of Fig. 4(a)) represents states before and after PT. normalized probability distributions P(δ) of the displacement
The running coordination number nO*H = 3 after the first coordinate δ and the relation F((δ) = −kT ln P(δ). The barrier
O*-H intra-molecular peak in gO*H at 0.99 Å in the left panel height is 0.14 kcal/mol compared to 0.13 kcal/mol reported in
of Fig. 4(b) (|δ| > 0.5 Å), confirms the presence of three H- Ref. 44 and is small relative to the average thermal energy of
atoms bonded to O* in the Eigen complex H3 O*+ . The peak 0.59 kcal/mol at 300 K. The low barrier and the absence of a
in gO*O at 2.67 Å for large |δ| > 0.5 Å reflects the presence well at or near the top suggests that the proton is significantly
of an oxygen atom in the first coordination shell of the Eigen delocalized in the Zundel complex, as suggested earlier by
complex (see the left panel of Fig. 4(b)). When |δ| is small Marx et al.,2 and the energy difference between the Eigen
(right panel of Fig. 4(b)) this peak is split into two, a sharp and Zundel forms is small. The lower panel of Fig. 5(b)
one at 2.36 Å equal to the O*-O^ distance in the Zundel com- depicts the proton jump distances every ten time steps over
plex and running coordination number nO*O = 1 and a sec- an interval of 100 000 time steps or 100 ps.
ond peak identifying the oxygen atom of a water molecule in In our study using the sOSS2 model, the hydroxide
the first solvation shell of O* which has receded to 2.73 Å. ion at 0.26 M exists mainly as hyper-coordinated structures
The peak in the conditional pair correlation function gO*H at at 298 K before PT occurs. The solvation of the OH− ion
0.99 Å for δ > 0.5 Å is also split into two peaks at 0.98 Å and by water molecules involves four H-bonds accepted by the
1.18 Å, respectively, when δ < 0.1 Å with running coordina- oxygen atom in a roughly square planar arrangement and

Downloaded 27 Sep 2011 to 130.111.64.68. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 124505-6 S. H. Lee and J. C. Rasaiah J. Chem. Phys. 135, 124505 (2011)

(a) (b)
(a)

(i) (ii)

Jump dsitance(A)
(iii) (iv)

FIG. 5. (a) Free energy profile at 298 K along the proton transfer coordinate (vi)
δ of the H+ and OH− systems. (b) Jump distances of H* of H+ ion (lower
(v)

panel), H* of OH− ion (middle panel), and O* of OH− ion (lower panel)
versus the time step* = time steps/10 where one time step = 1 fs.

one donated by the H atom of the OH− ion adding up to

an overall average coordination number of ∼4.6 as noted
previously by Tuckerman et al.3 from AIMD simulations (vii) (viii)

of an ion and 31 water molecules (∼1.7 M). (see Figs. 4(c)

and 6(b)). This structure is supported by neutron and x-Ray (b)
diffraction,19–21 x-ray adsorption,22 ultrafast and 2D infrared
spectroscopy,29, 31 and core-level photoelectron emission
studies.32 In contrast to the AIMD simulations of Tuckerman (i) (ii)

et al.,3 we find that the hyper-coordinated species exists

before the initial PT step. The weaker hydrogen bond donated
to the oxygen atom of a neighboring water remains during
PT and becomes stronger and part of the first solvation
shell of a neighboring water molecule after PT has taken
(iv)
place. (iii)

The hydroxide ion moves in a direction opposite to

that of proton transfer towards the oxygen of OH− and
occurs in the sequence depicted in Figs. 4(c)(i) to 4(c)(iii).
PT of the hydroxide ion starts from the hyper-coordinated
OH− (H2 O)4.6 and proceeds through three-coordinated in-
termediates by losing a water molecule from the roughly (v) (vi)

square planar arrangement followed by the formation of a

Zundel-like transition state [HO*-H-O^H]− involving one of
the three remaining water molecules hydrogen-bond to O*,
the oxygen of OH− , to which PT may eventually occur. This
transition state has been observed recently in spectroscopic
studies by Roberts et al.29, 31 (vii) (viii)
The sequence of events in PT to an OH− is readily
followed by selecting two sets of configurations for which
|δ| is small (|δ| < 0.1 Å) or large (|δ| > 0.5 Å) where the FIG. 6. (a) Representative configurations showing the mechanism of succes-
displacement coordinate δ = RO*H -RO^H . Here RO*H and sive proton transfers leading to the transport of a hydrogen ion in a simulation
RO^H are the respective distances of the shared proton from of one H3 O+ ion in 215 water molecules using the sOSS2 potential. (i) H3 O+
ion stabilized as a three-coordinated Eigen complex H3 O+ (H2 O)3 with a wa-
O* and O^ the oxygen of the water molecule to which O* ter molecule in the second solvation shell. (ii) Loss of water in the second
is hydrogen bonded to before PT occurs to O^. Several solvation shell in (i). (iii) The formation of the Zundel intermediate H5 O2 + .
transitions of the intervening H atom occurred back and forth (iv) Completion of proton transfer via the Zundel intermediate to form a new
H3 O+ ion. Repetition of (iii)-(iv) in (v)-(vi) and (vii)-(viii). (b) Representa-
between these two oxygen atoms were observed before PT
tive configurations of OH− ion showing the mechanism of successive proton
is complete. A newly formed four coordinated hydroxide transfer leading to the transport of the hydroxide ion in a simulation of one
ion on O^ is produced by hydrogen bonding with a water OH− ion in 215 water molecules using the sOSS2 potential. (i) OH− (H2 O)4.6
molecule in the vicinity (seen in Fig. 4(c)(iii)) when PT ends. complex showing four coordination of OH− and an additional H-bonded wa-
ter molecule accepted 60% of the time by H of OH− (ii) OH− (H2 O)3 in-
Alternatively, successive proton transfers proceed through termediate showing three coordination of OH− by loss of water. (iii) The
three-coordinated intermediates accompanied by solvent formation of the Zundel-like transition state H3 O2 − . (iv) Completion of pro-
reorganization that must occur before proton transfer (see ton transfer via OH− (H2 O)3 intermediate to form new OH− ion. Repetition
Fig. 6(b)). During a chain of successive PTs, the three coor- of (iii)-(iv) in (v)-(vi) and (vii)-(viii). The final structure of OH− in (viii) is
OH− (H2 O)4.6 complex.

Downloaded 27 Sep 2011 to 130.111.64.68. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 124505-7 Proton transfer, H+ and OH− ion mobilities J. Chem. Phys. 135, 124505 (2011)

dinated OH− is dominant. PT ends when the final structure is 50

(a) (b)
the stable hyper-coordinated complex. For the hydroxide ion, o
40 0 Co
successive PT begins and ends with the hyper-coordinated 25oC

Probability
-
complex. The mechanism is in broad agreement with pre- 30 O*(OH ) 50 C
vious studies AIMD studies at room temperature using the
20
BLYP functional.3, 33, 39
+
The evidence for hyper-coordination comes from left 10 H*(H )
panel of Fig. 4(d) which shows that for large |δ| > 0.5, the
0
oxygen O* of the hydroxyl ion accepts on average four hy- 0 5 10 15 20 25 30
Number of consecutive PT
drogen bonds and donates one 60% of the time (coordination
number nO*H ∼ 4.6). Visual observations (configurations (i)
and (iii) in Fig. 4(c)) show that the four hydrogen atoms are FIG. 7. (a) Probability distributions of consecutive PT jumps within 1 ps
vs. their number at 298 K. The inset shows the distributions for consecutive
nearly in a plane slightly below the oxygen atom of the hy- jumps within the average time between jumps (0.72 ps for H+ and 2.52 ps for
droxyl ion. When |δ| < 0.1 (right panel of Fig. 4(d)) the OH− OH− ions, respectively, see text). (b) Probability distributions of consecutive
ion is in a Zundel-like transition state, the first peak in the con- PT jumps within 1 ps vs. their number as a function of temperature.
ditional pair correlation function gO*O present at 2.73 Å when
δ > 0.5 Å is split into two peaks, one at 2.43 Å which is equal
to the O*-O distance the Zundel-like complex and the other at is governed by this time delay; the average time between two
2.81 Å due to the presence of an oxygen atom in the first sol- PT events at room temperature is 0.72 ps in the solution con-
vation of O*. This splitting does not seem to have been ob- taining a single H+ ion, and 2.52 ps in the solution with a
served in the AIMD study of Tuckerman et al. (see Fig. 2(b) single OH− ion. Over the same time interval more successive
of Ref. 3). The first intra-molecular peak in the O*-H pair cor- PT events occur in the solution containing an excess H+ ion
relation function gO*H at 0.99 Å for δ > 0.5 Å also splits into than in the solution with an extra OH− ion. The probabilities
two peaks at 0.98 Å and 1.21 Å, respectively; the second peak of a single isolated PT are 0.062 and 0.418, respectively, for
corresponding to the OH distance in the Zundel-like transition the H+ and OH− ions at 298 K, demonstrating that successive
state. This splitting is observed in AIMD simulations of Tuck- PTs are a more likely to occur in the diffusion of the H+ ion
erman et al.3 than in OH− ion diffusion in aqueous solution. The dynamical
The free energy profile F((δ) for proton transfer within asymmetry of PT transfer underlying the structural diffusion
the Zundel-like states for the basic solutions at 0.26 M of H+ and OH− ions in water is clearly visible in the proba-
was constructed from the normalized probability distributions bility distributions of consecutive PT jumps each within 1 ps
P(δ)) and F(δ) = −kT ln P(δ) shown in Fig. 5(a) with the cor- (Fig. 7(a)). When the 1 ps time delay is replaced by the av-
responding profile for the H+ ion. The free energy barrier is erage time between jumps (0.72 ps for the H+ ion and 2.52
0.23 kcal/mol compared to 0.34 kcal/mol reported in Ref. 44; ps for the OH− ion), the probability distribution functions are
the barrier for the H+ ion, as noted earlier, is lower and about nearly coincident (inset of Fig. 7(a)). This suggests that the
0.13 kcal/mol. The barrier height for PT transfer for the OH− PT step is similar for both ions, but that it is modulated by
is quite still small indicating considerable proton delocaliza- different structural rearrangements of the solvated ions and
tion. the solvent between successive PT steps.
In spite of the slightly larger barrier height and longer tra- The structural rearrangements and associated time de-
jectory (∼0.3 Å) for proton transfer across Zundel-like com- lays are different for H+ and OH− ions, and are reflected
plex of the OH− ion, the shuttling frequency is greater for the in the large difference in their mobilities. Successive proton
OH-ion than for the H+ (compare middle and lower panels of transfers shown in Fig. 6(a) suggest a concerted mechanism
Fig. 5(b)). Since the underlying mechanism of structural dif- for H+ ion transport involving a reduction in the coordina-
fusion is proton transfer, fewer trajectories are successful in tion number of a water molecule in the solvation shell of
eventually leading to proton transfer in the basic solution (PT H3 O+ before proton transfer occurs (panels (iv) and (v) in
to the OH− ion) than in the acidic solution (PT from the H3 O+ Fig. 6(a)).1, 34, 70, 71
ion). Comparison of the middle and upper panels of Fig. 5(b) The average time between successive PT events changes
shows that the frequency of successful PT events for the OH− with the temperature; it is larger (0.89 ps at 0 ◦ C) at lower
ion is less than the frequency of proton shuttling across the and smaller (0.62 ps at 50 ◦ C) at higher temperatures than at
Zundel-like complex. This implies that the rate of structural 25 ◦ C (0.72 ps) for a H+ ion in 215 water molecules. The
diffusion in OH− ion is determined by a factor besides the corresponding delay times for a OH− ion in solution are
proton transfer rate across the Zundel-like complex. One such 3.04 ps at 0 ◦ C and 2.23 ps at 50 ◦ C, respectively, in con-
factor is the structural reorganization that must take place for trast to 2.52 ps at 25 ◦ C, following the same trends as the H+
PT to occur. ion. Since the probabilities of consecutive PTs increase with
Large systems present no special difficulties in our simu- the temperature, the probabilities of a single PT are smaller
lations, and we observed several successive PT events in our at higher temperatures and larger at lower temperatures
study of the diffusion of H+ and OH− in 215 water molecules, (Fig. 7(b)). When the 1 ps time delay is replaced by the av-
in which each PT step, except the last, takes place after a time erage time between jumps, the jump probability distributions
delay for solvent rearrangement and proton rattling.33 Since are nearly the same and similar to the inset of Fig. 7(a) at
the barriers are low, the dynamics of PT in aqueous solution 298 K.

Downloaded 27 Sep 2011 to 130.111.64.68. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 124505-8 S. H. Lee and J. C. Rasaiah J. Chem. Phys. 135, 124505 (2011)

IV. REORIENTATION DYNAMICS OF THE OH− ION 40

12 H+ sim.
− H* (H+) -
The relaxation times for the reorientation of an OH ion OH sim.
H+ exp.
30 O* (OH-)
10

D (10 -5cm2/sec)
-
are difficult to determine because of PT events that occur on H2O OH exp.

MSD (A )
2
almost the same time scale, which is about 2.5 ps at room tem- 8
20
perature. However, there is a distribution of times scales for
PT, and we can compute the orientation correlation function 6
10
P2 [u(t) · u(0)] for the unit vector u(t) along the OH bond of 4
(a) (b)
the hydroxide ion from trajectories for which no PT events
0 2
occur over times long enough to enable the relaxation time 0 2 4 6 8 0 10 20 o 30 40 50
t (ps) t ( C)
to be calculated. The relaxation times for reorientation are
plotted as a function of the inverse temperature in Fig. 2 and FIG. 9. (a) Mean square displacement of H* of H+ ion, O* of OH− ion and
coincide with the relaxation times for the OH bond of wa- water at 298.15 K. (b) Diffusion coefficient of the H+ and OH− ion as a
ter at high temperatures, but deviate sharply from them below function of temperature (i) calculated from the sOSS2 model (◦ for H+ and
290 K, in agreement with charge-transfer-to-solvent (CTTS) ♦ for OH− ) at 0.26 M, (ii) experimental results (Ref. 77) at infinite dilution
(● for H+ and 䉬 for OH− ) and (iii) experimental results (Refs. 74 and 76)
experiments28 and a MD simulation study72 of the OH− ion for 0.26 M HCl and NaOH solutions at 298.15 K ( for H+ and  for OH− ).
in a non-dissociating model for water. There are then two ac-
tivation energies (Table II), suggesting at least two different
structures for the solvated hydroxide ion dominant at higher
for proton transfer events in 215 water molecules. We average
and lower temperatures respectively. This agrees with CTTS
over 60 and 100 blocks, respectively, for H* and O* and plot
experiments that probe OH− ions in water,28 and is interpreted
the MSD as a function of time in Fig. 9(a). The plots are linear
as arising from an increase in probability of OH− (H2 O)3
and it is obvious that the slope for the hydroxide ion is smaller
structure and a corresponding decrease in the probabilities of
than that for the hydrogen ion in agreement with experiment.
hyper-coordinated structures OH− (H2 O)4.6 with rise in tem-
The diffusion coefficients calculated from the slopes are 7.62
perature. This shift would also affect the rate of proton trans-
× 10−5 cm2 /s for the hydrogen ion and 4.56 × 10−5 cm2 /s
fer and the mobility of the hydroxide ion.
for the hydroxide ion at 298.15 K. The concentration of a sin-
gle H+ or OH− ion in 215 water molecules is 0.26 M. The
V. THE MOBILITY OF H+ AND OH− IONS IN WATER experimental diffusion coefficients of the H+ and OH− at in-
To monitor the sequence of PT events in the calculation finite dilution are 9.31 × 10−5 cm2 /s and 5.30 × 10−5 cm2 /s,
of the diffusion coefficients, the index numbers of the trans- respectively, at 298 K.59
ferred proton H* in the solution with the excess proton, and We can also estimate the diffusion coefficients D from the
the oxygen O* of the hydroxide ion to which the proton was average PT jump distances (d) shown in Fig. 8 and the aver-
transferred in the solution with an excess hydroxide ion were age time (τ ) between successive jumps using D ∼ d2 /6τ .73
recorded every 0.01 ps during each 100 ps block of time steps For H* we find D ∼ (1.78 Å)2 /6 (0.72 ps) = 7.33 × 10−5
in our MD simulations (see Fig. 8). In the solution with the cm2 /s and for O* we obtain D ∼ (2.68 Å)2 /6 (2.52 ps) = 4.75
excess OH− ion, O* moves in a direction opposite to that of × 10−5 cm2 /s. These numbers are close to our calculations of
proton transfer. In Fig. 5(b) we have shown the jump distances the diffusion coefficients from the MSD in Fig. 9(a).
of O* and H* plotted against the time step. The smaller jump The agreement between our simulations and the exper-
distances of H* (∼0.25 Å) or O* (∼0.06 Å) correspond to imental results at infinite dilution is already very good but
no change in index number and includes rattling while the the comparison should be with the diffusion coefficients of
larger jump distances of ∼1.78 Å for H* and ∼2.68 Å for O* H+ and OH− ion in 0.26 M HCl and NaOH, respectively,
imply a change in the index number indicating net structural since the sodium and chloride ions at this concentration were
diffusion. the counter-ions in our simulations. The ion diffusion coef-
The diffusion coefficients of the hydrogen and hydroxide ficients were calculated from the experimentally determined
ions were calculated from MSD for H* and O*, respectively, equivalent conductances of HCl and NaOH at 0.26 M (37.8
mS m2 /mol and 21.2 mS m2 /mol (Refs. 74 and 75), respec-
tively) using Di = i RT/zi 2 F2 ( i = ti : equivalent ion con-
500 250 ductance, zi : valence, F: Faraday constant) after multiplying
(a) (b)
for each electrolyte by the corresponding transport num-
400 200
bers (tH+ = 0.82 and tOH− = 0.8 at 298 K) which are nearly
constant.75, 76 The experimental diffusion coefficients are 8.24
H* index

O* index

300 150
× 10−5 cm2 /s for H+ ion and 4.51 × 10−5 cm2 /s for the OH−
200 100
ion at 0.26 M in excellent agreement with our simulations
100 50 (7.62 × 10−5 cm2 /s for the H+ ion and 4.56 × 10−5 cm2 /s
for the OH− ion) using the sOSS2 model (see Fig. 9(b)). We
0 0
0 20 40 60 80 100 0 20 40 60 80 100 note however that nuclear quantum effects are not explicitly
t (ps) t (ps)
included in our study.
FIG. 8. Change in index numbers of (a) H* of H+ ion and (b) O* of OH−
Repeating the calculations at temperatures from 0 to
ion as a function of time. 50 ◦ C, the diffusion coefficients of both ions were found to

Downloaded 27 Sep 2011 to 130.111.64.68. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 124505-9 Proton transfer, H+ and OH− ion mobilities J. Chem. Phys. 135, 124505 (2011)

increase almost linearly with temperature in agreement with (MEST) (NRF-2010-0023062) and J.C.R. was supported by a
the trends shown by experimental results extrapolated to infi- National Science Foundation (NSF) Grant No. CHE 0549187.
nite dilution77 (see Fig. 9(b)). The difference in mobilities of
H+ and OH− ions persists and our calculations are in good 1 M. E. Tuckerman, K. Laasonen, M. Sprik, and M. Parrinello, J. Chem.
agreement with experiment, assuming that the corrections for Phys. 103, 150 (1995).
2 D. Marx, M. E. Tuckerman, J. Hutter, and M. Parrinello, Nature (London)
the concentration (0.26 M) remain small. This suggests that
397, 601 (1999).
the sOSS2 potential is an excellent model for the hydrated 3 M. E. Tuckerman, D. Marx, and M. Parrinello, Nature (London) 417, 925
H+ and OH− ions as well as for bulk water in this tempera- (2002).
ture range. 4 B. Chen, J. M. Park, I. Ivanov, G. Tabacchi, M. L. Klein, and M. Parrinello,

The activation energies for the diffusion of H+ and J. Am. Chem. Soc. 124, 8534 (2002)
5 D. Asthagiri, L. R. Pratt, J. D. Kress, and M. A. Gomez, Proc. Natl. Acad.
−
OH ions, calculated from the Arrhenius equation D = D0 Sci. U.S.A. 101, 7229 (2004).
exp(−Etrans /RT), are 10.3 kJ/mol and 12 kJ/mol, respec- 6 D. Bucher, A. Gray-Weale, and S. Kuyucak, J. Chem. Theory Comput. 6,

tively, in excellent agreement with the experimental values of 2888 (2010).

7 A. Warshel and R. M. Weiss, J. Am. Chem. Soc. 102, 6218 (1980).
10.6 kJ/mol and 12 kJ/mol, respectively, calculated from the 8 R. Vuilleumier and D. Borgis, J. Chem. Phys. 111, 4251 (1999).
experimental data at infinite dilution between 0 and 50 ◦ C 9 J. Lobaugh and G. A. Voth, J. Chem. Phys. 104, 2056 (1996).
(Refs. 60 and 61), see Table I. 10 U. W. Schmidt and G. A. Voth, J. Phys. Chem. B. 102, 5547 (1998).
11 T. J. F. Day, A. V. Soudakov, M. Cuma, U. W. Schmidt, and G. A. Voth, J.

Chem. Phys. 117, 5839 (2002).

12 I. S. Ufimstev, A. G. Kalinichev, T. J. Martinez, and R. J. Kirkpatrick, Phys.

VI. CONCLUSIONS Chem. Chem. Phys. 11, 9420 (2009).

13 M. E. Selvan, D. J. Keffer, S. Cui, and S. J. Paddison, J. Phys. Chem. C 14,

Classical MD simulations using a scaled dissociating 11965 (2010).

14 F. Bruni, M. A. Ricci, and A. K. Soper, J. Chem. Phys. 114, 8056
model potential (sOSS2) for liquid water provide unambigu-
(2001).
ous and clear pictures of the hydrated OH− and H+ ions and 15 A. Botti, F. Bruni, S. Imberti, M. A. Ricci, and A. K. Soper, J. Chem. Phys.
a computationally fast, accurate, and robust method of study- 119, 5001 (2003).
16 A. Botti, F. Bruni, S. Imberti, M. A. Ricci, and A. K. Soper, J. Chem. Phys.
ing PT reactions in aqueous solutions. The simulations do not
120, 10154 (2004).
require large amounts of computational time or resources and 17 S. Imberti, A. Botti, F. Bruni, G. Cappa, M. A. Ricci, and A. K. Soper, J.
can therefore be easily extended to large systems. Chem. Phys. 122, 194509 (2005).
The model supports the EZE mechanism of proton trans- 18 A. Botti, F. Bruni, S. Imberti, M. A. Ricci, and A. K. Soper, J. Mol. Liq.

fer in acidic solutions and the transient hyper-coordination 117, 77 (2005).

19 A. Botti, F. Bruni, S. Imberti, M. A. Ricci, and A. K. Soper, J. Mol. Liq.
of the hydroxide ion in weakly basic solutions at room
117, 81 (2005).
temperature.1–6, 34, 35, 38–44 The proportions of three-, four-, and 20 S. E. McLain, S. Imberti, A. K. Soper, A. Botti, F. Bruni, and M. A. Ricci,

hyper-coordinated solvation of the hydroxide ion change with Phys. Rev. B 74, 0942201 (2006).
21 T. Megyes, S. Balint, T. Grosz, T. Radnai, L. Bakó, and P. Sipsos, J. Chem.
temperature. Large numbers of successive PT events were ob-
Phys. 128, 044501 (2008).
served, which enabled us to determine accurately the relative 22 C. D. Cappa, J. D. Smith, B. M. Messer, R. C. Cohen, and J. Saykally, J.
and absolute mobilities of hydrogen and hydroxide ions in Phys. Chem. A 111, 4776 (2007).
23 K. S. Asmis, N. L. Privonka, G. Santambrogio, M. Brummer, C. Kaposta,
aqueous solution, and their temperature dependence between
0 and 50 ◦ C. Differences between the mobilities of these ions D. M. Neumark, and L. Wöste, Science 299, 1375 (2003).
24 M. Rini, B. Z. Magnus, E. Pines, and E. T. J. Nibbering, Science 301, 493
are traced to differences in the solvation structures and to (2003).
different time delays for solvent reorganization of hydrated 25 O. F. Mohamed, D. Pines, J. Dreyer, E. Pines, and E. T. J. Nibbering,

H+ and OH− ions before PT occurs, thereby breaking the Science 310, 83 (2005).
26 J. M. Headrick, E. G. Diken, R. S. Walters, N. I. Hammer, R. A. Christie,
hole-particle symmetry of the dynamics of proton transfer in
J. Cui, E. M. Myshakin, M. A. Duncan, M. A. Johnson, and K. D. Jordan,
H+ and OH− ions.50, 51 Nuclear quantum effects2, 3, 44 are ne- Science 308, 1765 (2005).
glected in our study and future work could involve optimiza- 27 S. Woutersen and H. J. Bakker, Phys. Rev. Lett. 96, 138305 (2006).
28 J. Thøgersen, S. K. Jensen, C. Petersen, and S. Keiding, Chem. Phys. Lett.
tion of scaling to include these effects.
The sOSS2 model potential for liquid water and compu- 466, 1 (2008).
29 S. T. Roberts, P. B. Petersen, K. Ramasesha, A. Tokmakoff, I. S. Ufimtsev,
tational methods for PT discussed here could be adapted to and T. J. Martinez, Proc. Natl. Acad. Sci. U.S.A. 106, 15154 (2009).
study acid-base chemistry and large scale simulations of pro- 30 S. G. Olesen, T. L. Guasco, J. R. Roscioli, and M. A. Johnson, Chem. Phys.

ton transfer reactions in solution, at interfaces40, 78 and in wa- Lett. 509, 89 (2011).
31 S. T. Roberts, K. Ramasesha, P. B. Petersen, A. Mandal, and A. Tokmakoff,
ter filled nanopores.40, 79–81 The method of scaling the OSS2 J. Phys. Chem. A 115, 3957 (2011).
potential could also be refined and applied to other inter- 32 E. F. Aziz, N. Ottosson, M Faubel, I. V. Hertel, and B. Winter, Nature

atomic potentials derived from ab initio quantum mechanical (London) 455, 89 (2008).
33 A. Chandra, M. E. Tuckerman, and D. Marx, Phys. Rev. Lett. 99, 145901
studies of a few molecules for use in large scale simulation
(2007).
studies of the bulk phase. 34 N. Agmon, Chem. Phys. Lett. 244, 456 (1995).
35 N. Agmon, Isr. J. Chem. 39, 493 (1999).
36 J. Hynes, Nature (London) 397, 565 (1999).
37 N. Agmon, Chem. Phys. Lett. 319, 247 (2000).
ACKNOWLEDGMENTS 38 R. Ludwig, Angew. Chem., Int. Ed. 42, 258 (2003).
39 M. E. Tuckerman, A. Chandra, and D. Marx, Acc. Chem. Res. 39, 151
S.H.L. was supported by the National Research Foun- (2006).
40 G. A. Voth, Acc. Chem. Res. 39, 143 (2006).
dation of Korea Grant funded by the Korean Government

Downloaded 27 Sep 2011 to 130.111.64.68. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 124505-10 S. H. Lee and J. C. Rasaiah J. Chem. Phys. 135, 124505 (2011)

41 J. M. J. Swanson, C. M. Maupin, H. Chen, M. K. Petersen, J. Xu, Y. Wu, 62 D. A. Turton and K. Wynne, J. Chem. Phys. 128, 154516 (2008).
and G. A. Voth, J. Phys. Chem. B 111, 4300 (2007). 63 A. Luzar and D. Chandler, Nature (London) 379, 55 (1996).
42 H. Lapid, N. Agmon, M. K. Petersen, and G. A. Voth, J. Chem. Phys. 122, 64 D. Laage and J. T. Hynes, Science 311, 832 (2006).

014506 (2005). 65 D. Laage and J. T. Hynes, Proc. Natl. Acad. Sci. U.S.A. 104, 11167
43 O. Markovitch, H. Chen, S. Izvekov, F. Paesani, G. A. Voth, and N. Agmon, (2007).
J. Phys. Chem. B 112, 9456 (2008). 66 D. Laage and J. T. Hynes, J. Phys. Chem. B 112, 14230 (2008).
44 D. Marx, A. Chandra, and M. E. Tuckerman, Chem. Rev. 110, 2174 (2010). 67 A. K. Soper, Chem. Phys. 258, 121 (2000).
45 D. Asthagiri, L. R. Pratt, J. D. Kress, and M. A. Gomez, Chem. Phys. Lett. 68 H. J. C. Berendsen, J. R. Grigera, and T. P. Straatsma, J. Phys. Chem. 91,

380, 530 (2003). 6269 (1987).

46 E. E. Dahlke and D. G. Truhlar, J. Phys. Chem. B 109, 15677 (2005). 69 S. Koneshan, J. C. Rasaiah, and L. X. Dang, J. Chem. Phys. 114, 7544
47 J. VandeVondele, F. Mohamed, M. Krack, J. Hutter, M. Sprik, and (2001).
M. Parinello, J. Chem. Phys. 122, 014515 (2005). 70 M. E. Tuckerman, K. Laasonen, M. Sprik, and M. Parrinello, J. Phys.
48 I. S. Ufimstev, A. G. Kalinchev, T. J. Martinez, and R. J. Kirkpatrick, Chem. Chem. 99, 5749 (1995).
Phys. Lett. 442, 128 (2007) 71 T. C. Berkelbach, H.-S. Lee, and M. E. Tuckerman, Phys. Rev. Lett. 103,
49 L. Ojamäe, I. Shavitt, and S. J. Singer, J. Chem. Phys. 109, 5547 (1998). 238302 (2009).
50 H. L. Lemberg and F. H. Stillinger, J. Chem. Phys. 62, 1677 (1975). 72 X. Sun, S. Yoo, S. S. Xantheas, and L. X. Dang, Chem. Phys. Lett. 481, 9
51 F. H. Stillinger and C. W. David, J. Chem. Phys. 73, 3384 (1980). (2009).
52 T. A. Weber and F. H. Stillinger, J. Chem. Phys. 77, 4150 (1982). 73 Z. Luz and S. Meiboom, J. Am. Chem. Soc. 86, 4768 (1964)
53 S. H. Lee and J. C. Rasaiah, Mol. Simul. 36, 69 (2010). 74 R. H. Stokes, J. Phys. Chem. 65, 1242 (1961).
54 G. Stell, “Fluids with long-range forces,” in Statistical Mechanics Part A, 75 C. H. Hamann, A. Hamnett, and W. Vielstich, Electrochemistry, 2nd ed.

Equilibrium Techniques, Modern Theoretical Chemistry Vol. 5, edited by (Wiley VCH, Berlin, 2007).
B. J. Berne (Plenum, New York, 1977), Chap. 2. 76 H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolyte Solu-
55 G. Stell, J. C. Rasaiah, and H. Narang, Mol. Phys. 27, 1393 (1974). tions, 3rd ed. (Reinhold, New York, 1958).
56 W. G. Hoover, Phys. Rev. A 31, 1695 (1985). 77 T. S. Light, S. Licht, A. C. Bevilacqua, and K. R. Morash, Electrochem.
57 W. C. Swope, H. C. Andersen, P. H. Berens, and K. R. Wilson, J. Chem. Solid-State Lett. 8, E16 (2008).
Phys. 76, 637 (1982). 78 C. J. Mundy, I.-F. W. Kuo, M. E. Tuckerman, H.-S. Lee, and D. J. Tobias,
58 S. H. Lee, Bull. Korean Chem. Soc. 22, 847 (2001). Chem. Phys. Lett. 481, 2 (2009).
59 P. Atkins and J. D. Paula, Physical Chemistry, 7th ed. (Freeman, New York, 79 J. C. Rasaiah, S. Garde, and G. Hummer, Ann. Rev. Phys. Chem. 59, 713

2002), p. 1104. (2008).

60 M. Holz, S. R. Heil, and A. Sacco, Phys. Chem. Chem. Phys. 2, 4740 80 G. Hummer, J. C. Rasaiah, and J. P. Noworyta, Nature (London) 414, 188

(2002). (2001).
61 A. J. Easteal, W. E. Price, and L. A. Woolf, J. Chem. Soc., Faraday Tans. 1 81 C. Dellago, M. M. Naor, and G. Hummer, Phys. Rev. Lett. 90, 105902

85, 1091 (1985). (2003).

Downloaded 27 Sep 2011 to 130.111.64.68. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions