trunk/iceWater2/iceWater3.tex

%% PNAStwoS.tex
%% Sample file to use for PNAS articles prepared in LaTeX
%% For two column PNAS articles
%% Version1: Apr 15, 2008
%% Version2: Oct 04, 2013

%% BASIC CLASS FILE
\documentclass{pnastwo}

%% ADDITIONAL OPTIONAL STYLE FILES Font specification

%\usepackage{PNASTWOF}
\usepackage[version=3]{mhchem}
\usepackage[round,numbers,sort&compress]{natbib}
\usepackage{fixltx2e}
\usepackage{booktabs}
\usepackage{multirow}
\usepackage{tablefootnote}

\bibpunct{(}{)}{,}{n}{,}{,}
\bibliographystyle{pnas2011}

%% OPTIONAL MACRO DEFINITIONS
\def\s{\sigma}
%%%%%%%%%%%%
%% For PNAS Only:
%\url{www.pnas.org/cgi/doi/10.1073/pnas.0709640104}
\copyrightyear{2014}
\issuedate{Issue Date}
\volume{Volume}
\issuenumber{Issue Number}
%\setcounter{page}{2687} %Set page number here if desired
%%%%%%%%%%%%

\begin{document}

\title{The different facets of ice have different hydrophilicities:
  Friction at water / ice-I\textsubscript{h} interfaces}

\author{Patrick B. Louden\affil{1}{Department of Chemistry and Biochemistry, University of Notre Dame, Notre Dame,
IN 46556}
\and
J. Daniel Gezelter\affil{1}{}}

\contributor{Submitted to Proceedings of the National Academy of Sciences
of the United States of America}

%%%Newly updated.
%%% If significance statement need, then can use the below command otherwise just delete it.
\significancetext{Surface hydrophilicity is a way of describing the
  interaction strength between a solid surface and liquid water.  Our
  simulations show that the solid that is thought to be most
  hydrophilic (ice) displays different behavior depending on which
  crystal facet is presented to the liquid. This behavior is
  potentially important in geophysics, in recognition of ice surfaces
  by anti-freeze proteins, and in understanding how the friction
  between ice and other solids may be mediated by a quasi-liquid layer
  of water.}

\maketitle

\begin{article}
  \begin{abstract}
    We present evidence that the prismatic and secondary prism facets
    of ice-I$_\mathrm{h}$ crystals possess structural features that
    can reduce the effective hydrophilicity of the ice/water
    interface. The spreading dynamics of liquid water droplets on ice
    facets exhibits long-time behavior that differs for the prismatic
    $\{1~0~\bar{1}~0\}$ and secondary prism $\{1~1~\bar{2}~0\}$ facets
    when compared with the basal $\{0001\}$ and pyramidal
    $\{2~0~\bar{2}~1\}$ facets.  We also present the results of
    simulations of solid-liquid friction of the same four crystal
    facets being drawn through liquid water, and find that the two
    prismatic facets exhibit roughly half the solid-liquid friction of
    the basal and pyramidal facets.  These simulations provide
    evidence that the two prismatic faces have a significantly smaller
    effective surface area in contact with the liquid water. The ice /
    water interfacial widths for all four crystal facets are similar
    (using both structural and dynamic measures), and were found to be
    independent of the shear rate.  Additionally, decomposition of
    orientational time correlation functions show position-dependence
    for the short- and longer-time decay components close to the
    interface.
\end{abstract}

\keywords{ice | water | interfaces | hydrophobicity}
\abbreviations{QLL, quasi-liquid layer; MD, molecular dynamics; RNEMD,
reverse non-equilibrium molecular dynamics}

\dropcap{S}urfaces can be characterized as hydrophobic or hydrophilic
based on the strength of the interactions with water. Hydrophobic
surfaces do not have strong enough interactions with water to overcome
the internal attraction between molecules in the liquid phase, and the
degree of hydrophilicity of a surface can be described by the extent a
droplet can spread out over the surface. The contact angle, $\theta$,
formed between the solid and the liquid depends on the free energies
of the three interfaces involved, and is given by Young's
equation~\cite{Young05},
\begin{equation}\label{young}
\cos\theta = (\gamma_{sv} - \gamma_{sl})/\gamma_{lv} .
\end{equation} 
Here $\gamma_{sv}$, $\gamma_{sl}$, and $\gamma_{lv}$ are the free
energies of the solid/vapor, solid/liquid, and liquid/vapor interfaces,
respectively.  Large contact angles, $\theta > 90^{\circ}$, correspond
to hydrophobic surfaces with low wettability, while small contact
angles, $\theta < 90^{\circ}$, correspond to hydrophilic surfaces.
Experimentally, measurements of the contact angle of sessile drops is
often used to quantify the extent of wetting on surfaces with
thermally selective wetting
characteristics~\cite{Tadanaga00,Liu04,Sun04}.

Nanometer-scale structural features of a solid surface can influence
the hydrophilicity to a surprising degree.  Small changes in the
heights and widths of nano-pillars can change a surface from
superhydrophobic, $\theta \ge 150^{\circ}$, to hydrophilic, $\theta
\sim 0^{\circ}$~\cite{Koishi09}. This is often referred to as the
Cassie-Baxter to Wenzel transition.  Nano-pillared surfaces with
electrically tunable Cassie-Baxter and Wenzel states have also been
observed~\cite{Herbertson06,Dhindsa06,Verplanck07,Ahuja08,Manukyan11}.
Luzar and coworkers have modeled these transitions on nano-patterned
surfaces~\cite{Daub07,Daub10,Daub11,Ritchie12}, and have found the
change in contact angle is due to the field-induced perturbation of
hydrogen bonding at the liquid/vapor interface~\cite{Daub07}.

One would expect the interfaces of ice to be highly hydrophilic (and
possibly the most hydrophilic of all solid surfaces). In this paper we
present evidence that some of the crystal facets of ice-I$_\mathrm{h}$
have structural features that can reduce the effective hydrophilicity.
Our evidence for this comes from molecular dynamics (MD) simulations
of the spreading dynamics of liquid droplets on these facets, as well
as reverse non-equilibrium molecular dynamics (RNEMD) simulations of
solid-liquid friction.

Quiescent ice-I$_\mathrm{h}$/water interfaces have been studied
extensively using computer simulations. Hayward and Haymet
characterized and measured the widths of these
interfaces~\cite{Hayward01,Hayward02}.  Nada and Furukawa have also
modeled the width of basal/water and prismatic/water
interfaces~\cite{Nada95} as well as crystal restructuring at
temperatures approaching the melting point~\cite{Nada00}.

The surface of ice exhibits a pre-melting layer, often called a
quasi-liquid layer (QLL), at temperatures near the melting point.  MD
simulations of the facets of ice-I$_\mathrm{h}$ exposed to vacuum have
found QLL widths of approximately 10 \AA\ at 3 K below the melting
point~\cite{Conde08}. Similarly, Limmer and Chandler have used the mW
water model~\cite{Molinero09} and statistical field theory to estimate
QLL widths at similar temperatures to be about 3 nm~\cite{Limmer14}.

Recently, Sazaki and Furukawa have developed a technique using laser
confocal microscopy combined with differential interference contrast
microscopy that has sufficient spatial and temporal resolution to
visualize and quantitatively analyze QLLs on ice crystals at
temperatures near melting~\cite{Sazaki10}. They have found the width of
the QLLs perpendicular to the surface at -2.2$^{o}$C to be 3-4 \AA\
wide.  They have also seen the formation of two immiscible QLLs, which
displayed different dynamics on the crystal surface~\cite{Sazaki12}.

% There is now significant interest in the \textit{tribological}
% properties of ice/ice and ice/water interfaces in the geophysics
% community.  Understanding the dynamics of solid-solid shearing that is
% mediated by a liquid layer~\cite{Cuffey99, Bell08} will aid in
% understanding the macroscopic motion of large ice
% masses~\cite{Casassa91, Sukhorukov13, Pritchard12, Lishman13}.

Using molecular dynamics simulations, Samadashvili has recently shown
that when two smooth ice slabs slide past one another, a stable
liquid-like layer develops between them~\cite{Samadashvili13}. In a
previous study, our RNEMD simulations of ice-I$_\mathrm{h}$ shearing
through liquid water have provided quantitative estimates of the
solid-liquid kinetic friction coefficients~\cite{Louden13}. These
displayed a factor of two difference between the basal and prismatic
facets.  The friction was found to be independent of shear direction
relative to the surface orientation.  We attributed facet-based
difference in liquid-solid friction to the 6.5 \AA\ corrugation of the
prismatic face which reduces the effective surface area of the ice
that is in direct contact with liquid water.

In the sections that follow, we describe the simulations of
droplet-spreading dynamics using standard MD as well as simulations of
tribological properties using RNEMD.  These simulations give
complementary results that point to the prismatic and secondary prism
facets having roughly half of their surface area in direct contact
with the liquid.

\subsection{Construction of the Ice / Water interfaces}
To construct the four interfacial ice/water systems, a model of a
proton-ordered, zero-dipole crystal of ice-I$_\mathrm{h}$ was created.
This crystal structure was cleaved to expose each of the four facets
under investigation, and these were solvated in boxes of liquid water.
The procedure for creating the interfaces and equilibrating with a
liquid phase is described in more detail in the SI. The dimensions as
well as the number of ice and liquid water molecules contained in all
of the simulations are shown in Table S1.

The SPC/E water model~\cite{Berendsen87} has been extensively
characterized over a wide range of liquid
conditions~\cite{Arbuckle02,Kuang12}, and its phase diagram has been
well studied~\cite{Baez95,Bryk04b,Sanz04b,Fennell:2005fk}. With longer
cutoff radii and careful treatment of electrostatics, SPC/E mostly
avoids metastable crystalline morphologies like
ice-\textit{i}~\cite{Fennell:2005fk} and ice-B~\cite{Baez95}.  The
free energies and melting
points~\cite{Baez95,Arbuckle02,Gay02,Bryk02,Bryk04b,Sanz04b,Fennell:2005fk,Fernandez06,Abascal07,Vrbka07}
of various other crystalline polymorphs have also been calculated.
Haymet \textit{et al.} have studied quiescent Ice-I$_\mathrm{h}$/water
interfaces using the SPC/E water model, and have seen structural and
dynamic measurements of the interfacial width that agree well with
more expensive water models, although the coexistence temperature for
SPC/E is still well below the experimental melting point of real
water~\cite{Bryk02}. Given the extensive data and speed of this model,
it is a reasonable choice even though the temperatures required are
somewhat lower than real ice / water interfaces.

\section{Droplet Simulations}
Ice surfaces with a thickness of $\sim$~20~\AA\ were created as
described above, but were not solvated in a liquid box. The crystals
were then replicated along the $x$ and $y$ axes (parallel to the
surface) until a large surface ($>$ 126 nm\textsuperscript{2}) had
been created.  The sizes and numbers of molecules in each of the
surfaces is given in Table S1.  Weak translational restraining
potentials with spring constants of 1.5~$\mathrm{kcal\
  mol}^{-1}\mathrm{~\AA}^{-2}$ (prismatic and pyramidal facets) or
4.0~$\mathrm{kcal\ mol}^{-1}\mathrm{~\AA}^{-2}$ (basal facet) were
applied to the centers of mass of each molecule in order to prevent
surface melting, although the molecules were allowed to reorient
freely. A water droplet containing 2048 SPC/E molecules was created
separately. Droplets of this size can produce agreement with the Young
contact angle extrapolated to an infinite drop size~\cite{Daub10}. The
surfaces and droplet were independently equilibrated to 225~K, at
which time the droplet was placed 3-5~\AA\ above the surface.  Five
statistically independent simulations were carried out for each facet,
and the droplet was placed at unique $x$ and $y$ locations for each of
these simulations.  Each simulation was 5~ns in length and was
conducted in the microcanonical (NVE) ensemble.  Representative
configurations for the droplet on the prismatic facet are shown in
figure \ref{fig:Droplet}.

\section{Shearing Simulations (Interfaces in Bulk Water)}
To perform the shearing simulations, the velocity shearing and scaling
variant of reverse non-equilibrium molecular dynamics (VSS-RNEMD) was
employed \cite{Kuang12}. This method performs a series of simultaneous
non-equilibrium exchanges of linear momentum and kinetic energy
between two physically-separated regions of the simulation cell.  The
system responds to this unphysical flux with velocity and temperature
gradients.  When VSS-RNEMD is applied to bulk liquids, transport
properties like the thermal conductivity and the shear viscosity are
easily extracted assuming a linear response between the flux and the
gradient.  At the interfaces between dissimilar materials, the same
method can be used to extract \textit{interfacial} transport
properties (e.g. the interfacial thermal conductance and the
hydrodynamic slip length).

The kinetic energy flux (producing a thermal gradient) is necessary
when performing shearing simulations at the ice-water interface in
order to prevent the frictional heating due to the shear from melting
the crystal. Reference \citealp{Louden13} provides more details on the
VSS-RNEMD method as applied to ice-water interfaces.  A representative
configuration of the solvated prismatic facet being sheared through
liquid water is shown in figure \ref{fig:Shearing}.

The exchanges between the two regions were carried out every 2 fs
(i.e. every time step). Although computationally expensive, this was
done to minimize the magnitude of each individual momentum exchange.
Because individual VSS-RNEMD exchange moves conserve both total energy
and linear momentum, the method can be ``bolted-on'' to simulations in
any ensemble.  The simulations of the pyramidal interface were
performed under the canonical (NVT) ensemble.  When time correlation
functions were computed, the RNEMD simulations were done in the
microcanonical (NVE) ensemble.  All simulations of the other
interfaces were carried out in the microcanonical ensemble.

\section{Results}
\subsection{Ice - Water Contact Angles}
To determine the extent of wetting for each of the four crystal
facets, contact angles for liquid droplets on the ice surfaces were
computed using two methods.  In the first method, the droplet is
assumed to form a spherical cap, and the contact angle is estimated
from the $z$-axis location of the droplet's center of mass
($z_\mathrm{cm}$).  This procedure was first described by Hautman and
Klein~\cite{Hautman91}, and was utilized by Hirvi and Pakkanen in
their investigation of water droplets on polyethylene and poly(vinyl
chloride) surfaces~\cite{Hirvi06}. For each stored configuration, the
contact angle, $\theta$, was found by inverting the expression for the
location of the droplet center of mass,
\begin{equation}\label{contact_1}
\langle z_\mathrm{cm}\rangle = 2^{-4/3}R_{0}\bigg(\frac{1-cos\theta}{2+cos\theta}\bigg)^{1/3}\frac{3+cos\theta}{2+cos\theta} ,
\end{equation}
where $R_{0}$ is the radius of the free water droplet. 

A second method for obtaining the contact angle was described by
Ruijter, Blake, and Coninck~\cite{Ruijter99}, and is described in more
detail in the SI.  Both methods provided similar estimates of the
dynamic contact angle, although the method described above is
significantly less prone to noise, and is the method used to compute
the contact angles in table \ref{tab:kappa}.

Because the initial droplet was placed above the surface, the initial
value of 180$^{\circ}$ decayed over time (See figure S1 in the SI).
Each of these profiles were fit to a biexponential decay, with a
short-time contribution ($\tau_c$) that describes the initial contact
with the surface, a long time contribution ($\tau_s$) that describes
the spread of the droplet over the surface, and a constant
($\theta_\infty$) to capture the infinite-time estimate of the
equilibrium contact angle,
\begin{equation}
\theta(t) = \theta_\infty +  (180-\theta_\infty) \left[ a e^{-t/\tau_c} +
  (1-a) e^{-t/\tau_s}  \right]
\end{equation}
We have found that the rate for water droplet spreading across all
four crystal facets, $k_\mathrm{spread} = 1/\tau_s \approx$ 0.7
ns$^{-1}$. However, the basal and pyramidal facets produced estimated
equilibrium contact angles, $\theta_\infty \approx$ 35$^{\circ}$, while
prismatic and secondary prismatic had values for $\theta_\infty$ near
43$^{\circ}$ as seen in Table \ref{tab:kappa}.

These results indicate that by traditional measures, the basal and
pyramidal facets are more hydrophilic than the prismatic and secondary
prism facets, and surprisingly, that the differential hydrophilicities
of the crystal facets is not reflected in the spreading rate of the
droplet.

% This is in good agreement with our calculations of friction
% coefficients, in which the basal
% and pyramidal had a higher coefficient of kinetic friction than the
% prismatic and secondary prismatic. Due to this, we believe that the
% differences in friction coefficients can be attributed to the varying
% hydrophilicities of the facets. 

\subsection{Solid-liquid friction of the interfaces}
In a bulk fluid, the shear viscosity, $\eta$, can be determined
assuming a linear response to a shear stress,
\begin{equation}\label{Shenyu-11}
j_{z}(p_{x}) = \eta \frac{\partial v_{x}}{\partial z}.
\end{equation}
Here $j_{z}(p_{x})$ is the flux (in $x$-momentum) that is transferred
in the $z$ direction (i.e. the shear stress). The RNEMD simulations
impose an artificial momentum flux between two regions of the
simulation, and the velocity gradient is the fluid's response. This
technique has now been applied quite widely to determine the
viscosities of a number of bulk fluids~\cite{Muller99,Bordat02,Cavalcanti07}.

At the interface between two phases (e.g. liquid / solid) the same
momentum flux creates a velocity difference between the two materials,
and this can be used to define an interfacial friction coefficient
($\kappa$),
\begin{equation}\label{Shenyu-13}
j_{z}(p_{x}) = \kappa \left[ v_{x}(liquid) - v_{x}(solid) \right]
\end{equation}
where $v_{x}(liquid)$ and $v_{x}(solid)$ are the velocities measured
directly adjacent to the interface.

The simulations described here contain significant quantities of both
liquid and solid phases, and the momentum flux must traverse a region
of the liquid that is simultaneously under a thermal gradient.  Since
the liquid has a temperature-dependent shear viscosity, $\eta(T)$,
estimates of the solid-liquid friction coefficient can be obtained if
one knows the viscosity of the liquid at the interface (i.e. at the
melting temperature, $T_m$),
\begin{equation}\label{kappa-2}
\kappa = \frac{\eta(T_{m})}{\left[v_{x}(fluid)-v_{x}(solid)\right]}\left(\frac{\partial v_{x}}{\partial z}\right).
\end{equation}
For SPC/E, the melting temperature of Ice-I$_\mathrm{h}$ is estimated
to be 225~K~\cite{Bryk02}.  To obtain the value of
$\eta(225\mathrm{~K})$ for the SPC/E model, a $31.09 \times 29.38
\times 124.39$ \AA\ box with 3744 water molecules in a disordered
configuration was equilibrated to 225~K, and five
statistically-independent shearing simulations were performed (with
imposed fluxes that spanned a range of $3 \rightarrow 13
\mathrm{~m~s}^{-1}$ ).  Each simulation was conducted in the
microcanonical ensemble with total simulation times of 5 ns. The
VSS-RNEMD exchanges were carried out every 2 fs. We estimate
$\eta(225\mathrm{~K})$ to be 0.0148 $\pm$ 0.0007 Pa s for SPC/E,
roughly ten times larger than the shear viscosity previously computed
at 280~K~\cite{Kuang12}.

The interfacial friction coefficient can equivalently be expressed as
the ratio of the viscosity of the fluid to the hydrodynamic slip
length, $\kappa = \eta / \delta$. The slip length is an indication of
strength of the interactions between the solid and liquid phases,
although the connection between slip length and surface hydrophobicity
is not yet clear. In some simulations, the slip length has been found
to have a link to the effective surface
hydrophobicity~\cite{Sendner:2009uq}, although Ho \textit{et al.} have
found that liquid water can also slip on hydrophilic
surfaces~\cite{Ho:2011zr}. Experimental evidence for a direct tie
between slip length and hydrophobicity is also not
definitive. Total-internal reflection particle image velocimetry
(TIR-PIV) studies have suggested that there is a link between slip
length and effective
hydrophobicity~\cite{Lasne:2008vn,Bouzigues:2008ys}. However, recent
surface sensitive cross-correlation spectroscopy (TIR-FCCS)
measurements have seen similar slip behavior for both hydrophobic and
hydrophilic surfaces~\cite{Schaeffel:2013kx}.

In each of the systems studied here, the interfacial temperature was
kept fixed to 225~K, which ensured the viscosity of the fluid at the
interace was identical. Thus, any significant variation in $\kappa$
between the systems is a direct indicator of the slip length and the
effective interaction strength between the solid and liquid phases.

The calculated $\kappa$ values found for the four crystal facets of
Ice-I$_\mathrm{h}$ are shown in Table \ref{tab:kappa}. The basal and
pyramidal facets were found to have similar values of $\kappa \approx
6$ ($\times 10^{-4} \mathrm{amu~\AA}^{-2}\mathrm{fs}^{-1}$), while the
prismatic and secondary prism facets exhibited $\kappa \approx 3$
($\times 10^{-4} \mathrm{amu~\AA}^{-2}\mathrm{fs}^{-1}$). These
results are also essentially independent of the direction of the shear
relative to channels on the surfaces of the facets.  The friction
coefficients indicate that the basal and pyramidal facets have
significantly stronger interactions with liquid water than either of
the two prismatic facets.  This is in agreement with the contact angle
results above - both of the high-friction facets exhibited smaller
contact angles, suggesting that the solid-liquid friction (and inverse
slip length) is correlated with the hydrophilicity of these facets.

\subsection{Structural measures of interfacial width under shear}
One of the open questions about ice/water interfaces is whether the
thickness of the 'slush-like' quasi-liquid layer (QLL) depends on the
facet of ice presented to the water.  In the QLL region, the water
molecules are ordered differently than in either the solid or liquid
phases, and also exhibit distinct dynamical behavior.  The width of
this quasi-liquid layer has been estimated by finding the distance
over which structural order parameters or dynamic properties change
from their bulk liquid values to those of the solid ice.  The
properties used to find interfacial widths have included the local
density, the diffusion constant, and both translational and
orientational order
parameters~\cite{Karim88,Karim90,Hayward01,Hayward02,Bryk02,Gay02,Louden13}.

The VSS-RNEMD simulations impose thermal and velocity gradients.
These gradients perturb the momenta of the water molecules, so
parameters that depend on translational motion are often measuring the
momentum exchange, and not physical properties of the interface.  As a
structural measure of the interface, we have used the local
tetrahedral order parameter, which measures the match of the local
molecular environments (e.g. the angles between nearest neighbor
molecules) to perfect tetrahedral ordering.  This quantity was
originally described by Errington and Debenedetti~\cite{Errington01}
and has been used in bulk simulations by Kumar \textit{et
  al.}~\cite{Kumar09} It has previously been used in ice/water
interfaces by by Bryk and Haymet~\cite{Bryk04b}.

To determine the structural widths of the interfaces under shear,
tetrahedrality profiles for each of the interfaces were computed (see
the SI for more details). The lower panels in figures S2-S5 in the SI
show tetrahedrality profiles (in circles) for each of the four
interfaces.  These tetrahedrality profiles were fit using a hyperbolic
tangent function (see Eq. 6 in Reference \citealp{Louden13}) designed
to smoothly fit the liquid-to-ice transition while accounting for the
weak thermal gradient.  This fit provides an estimate of
$d_\mathrm{struct}$, the structural width of the interface.
 
We find the interfacial width to be $3.2 \pm 0.2$ \AA\ (pyramidal) and
$3.2 \pm 0.2$ \AA\ (secondary prism) for the control systems with no
applied momentum flux. This is similar to our previous results for the
interfacial widths of the quiescent basal ($3.2 \pm 0.4$ \AA) and
prismatic systems ($3.6 \pm 0.2$ \AA).

Over the range of shear rates investigated, $0.4 \rightarrow
6.0~\mathrm{~m~s}^{-1}$ for the pyramidal system and $0.6 \rightarrow
5.2~\mathrm{~m~s}^{-1}$ for the secondary prism, we found no
significant change in the interfacial width. The mean interfacial
widths are collected in table \ref{tab:kappa}. This follows our
previous findings of the basal and prismatic systems, in which the
interfacial widths of the basal and prismatic facets were also found
to be insensitive to the shear rate~\cite{Louden13}.

The similarity of these interfacial width estimates indicate that the
particular facet of the exposed ice crystal has little to no effect on
how far into the bulk the ice-like structural ordering persists. Also,
it appears that for the shearing rates imposed in this study, the
interfacial width is not structurally modified by the movement of
water over the ice.

\subsection{Dynamic measures of interfacial width under shear}
The spatially-resolved orientational time correlation function,
\begin{equation}\label{C(t)1}
  C_{2}(z,t)=\langle P_{2}(\mathbf{u}_i(0)\cdot \mathbf{u}_i(t))
  \delta(z_i(0) - z) \rangle,
\end{equation}
provides local information about the decorrelation of molecular
orientations in time. Here, $P_{2}$ is the second-order Legendre
polynomial, and $\mathbf{u}_i$ is the molecular vector that bisects
the HOH angle of molecule $i$.  The angle brackets indicate an average
over all the water molecules and time origins, and the delta function
restricts the average to specific regions. In the crystal, decay of
$C_2(z,t)$ is incomplete, while in the liquid, correlation times are
typically measured in ps.  Observing the spatial-transition between
the decay regimes can define a dynamic measure of the interfacial
width.

To determine the dynamic widths of the interfaces under shear, spatial
profiles of the orientational time correlation functions were computed
(see the SI for more details). Figures S6-S9 in the SI show the
$z$-coordinate profiles for the three decay constants, $\tau_{short}$,
$\tau_{middle}$, and $\tau_{long}$.

We have estimated the dynamic interfacial width $d_\mathrm{dyn}$ by
fitting the profiles of all the three orientational time constants
with an exponential decay to the bulk-liquid behavior (see SI).  These
values are shown in table \ref{tab:kappa}.  Because the spatial
granularity must be quite wide to obtain reasonable profiles of
$C_2(z,t)$, the error estimates for the dynamic widths of the
interface are significantly larger than for the structural widths.
However, all four interfaces exhibit dynamic widths that are
significantly below 1~nm, and are in reasonable agreement with the
structural widths described above.

\section{Conclusions}
In this work, we used MD simulations to measure the advancing contact
angles of water droplets on the basal, prismatic, pyramidal, and
secondary prism facets of Ice-I$_\mathrm{h}$.  Although we saw no
significant change in the \textit{rate} at which the droplets spread
over the surface, the long-time behavior predicts larger equilibrium
contact angles for the two prismatic facets.  

We have also used RNEMD simulations of water interfaces with the same
four crystal facets to compute solid-liquid friction coefficients.  We
have observed coefficients of friction that differ by a factor of two
between the two prismatic facets and the basal and pyramidal facets.
Because the solid-liquid friction coefficient is directly tied to the
inverse of the hydrodynamic slip length, this suggests that there are
significant differences in the overall interaction strengths between
these facets and the liquid layers immediately in contact with them.

The agreement between these two measures have lead us to conclude that
the two prismatic facets have a lower hydrophilicity than either the
basal or pyramidal facets.  One possible explanation of this behavior
is that the face presented by both prismatic facets consists of deep,
narrow channels (i.e. stripes of adjacent rows of pairs of
hydrogen-bound water molecules).  At the surfaces of these facets,
the channels are 6.35 \AA\ wide and the sub-surface ice layer is 2.25
\AA\ below (and therefore blocked from hydrogen bonding with the
liquid).  This means that only 1/2 of the surface molecules can form
hydrogen bonds with liquid-phase molecules.

In the basal plane, the surface features are shallower (1.3 \AA), with
no blocked subsurface layer.  The pyramidal face has much wider
channels (8.65 \AA) which are also quite shallow (1.37 \AA).  These
features allow liquid phase molecules to form hydrogen bonds with all
of the surface molecules in the basal and pyramidal facets.  This
means that for similar surface areas, the two prismatic facets have an
effective hydrogen bonding surface area of half of the basal and
pyramidal facets.  The reduction in the effective surface area would
explain much of the behavior observed in our simulations.

\begin{acknowledgments}
  Support for this project was provided by the National
  Science Foundation under grant CHE-1362211. Computational time was
  provided by the Center for Research Computing (CRC) at the
  University of Notre Dame.
\end{acknowledgments}

\bibliography{iceWater}
% *****************************************
% There is significant interest in the properties of ice/ice and ice/water 
% interfaces in the geophysics community. Most commonly, the results of shearing
% two ice blocks past one 
% another\cite{Casassa91, Sukhorukov13, Pritchard12, Lishman13} or the shearing 
% of ice through water\cite{Cuffey99, Bell08}. Using molecular dynamics 
% simulations, Samadashvili has recently shown that when two smooth ice slabs 
% slide past one another, a stable liquid-like layer develops between 
% them\cite{Samadashvili13}. To fundamentally understand these processes, a 
% molecular understanding of the ice/water interfaces is needed.     

% Investigation of the ice/water interface is also crucial in understanding
% processes such as nucleation, crystal 
% growth,\cite{Han92, Granasy95, Vanfleet95} and crystal 
% melting\cite{Weber83, Han92, Sakai96, Sakai96B}. Insight gained to these
% properties can also be applied to biological systems of interest, such as 
% the behavior of the antifreeze protein found in winter 
% flounder,\cite{Wierzbicki07,Chapsky97} and certain terrestial
% arthropods.\cite{Duman:2001qy,Meister29012013} Elucidating the properties which
% give rise to these processes through experimental techniques can be expensive,
% complicated, and sometimes infeasible. However, through the use of molecular 
% dynamics simulations much of the problems of investigating these properties
% are alleviated.

% Understanding ice/water interfaces inherently begins with the isolated
% systems. There has been extensive work parameterizing models for liquid water,
% such as the SPC\cite{Berendsen81}, SPC/E\cite{Berendsen87}, 
% TIP4P\cite{Jorgensen85}, TIP4P/2005\cite{Abascal05}, 
% ($\dots$), and more recently, models for simulating
% the solid phases of water, such as the TIP4P/Ice\cite{Abascal05b} model. The 
% melting point of various crystal structures of ice have been calculated for 
% many of these models
% (SPC\cite{Karim90,Abascal07}, SPC/E\cite{Baez95,Arbuckle02,Gay02,Bryk02,Bryk04b,Sanz04b,Gernandez06,Abascal07,Vrbka07}, TIP4P\cite{Karim88,Gao00,Sanz04,Sanz04b,Koyama04,Wang05,Fernandez06,Abascal07}, TIP5P\cite{Sanz04,Koyama04,Wang05,Fernandez06,Abascal07}), 
% and the partial or complete phase diagram for the model has been determined
% (SPC/E\cite{Baez95,Bryk04b,Sanz04b}, TIP4P\cite{Sanz04,Sanz04b,Koyama04}, TIP5P\cite{Sanz04,Koyama04}). 
% Knowing the behavior and melting point for these models has enabled an initial 
% investigation of ice/water interfaces. 

% The Ice-I$_\mathrm{h}$/water quiescent interface has been extensively studied 
% over the past 30 years by theory and experiment. Haymet \emph{et al.} have 
% done significant work characterizing and quantifying the width of these 
% interfaces for the SPC,\cite{Karim90} SPC/E,\cite{Gay02,Bryk02}, 
% CF1,\cite{Hayward01,Hayward02} and TIP4P\cite{Karim88} models for water. In 
% recent years, Haymet has focused on investigating the effects cations and 
% anions have on crystal nucleaion and 
% melting.\cite{Bryk04,Smith05,Wilson08,Wilson10} Nada and Furukawa have studied
% the the basal- and prismatic-water interface width\cite{Nada95}, crystal 
% surface restructuring at temperatures approaching the melting 
% point\cite{Nada00}, and the mechanism of ice growth inhibition by antifreeze 
% proteins\cite{Nada08,Nada11,Nada12}. Nada has developed a six-site water model
% for ice/water interfaces near the melting point\cite{Nada03}, and studied the 
% dependence of crystal growth shape on applied pressure\cite{Nada11b}. Using 
% this model, Nada and Furukawa have established differential 
% growth rates for the basal, prismatic, and secondary prismatic facets of 
% Ice-I$_\mathrm{h}$ and found their origins due to a reordering of the hydrogen 
% bond network in water near the interface\cite{Nada05}. While the work 
% described so far has mainly focused on bulk water on ice, there is significant
% interest in thin films of water on ice surfaces as well.  

% It is well known that the surface of ice exhibits a premelting layer at 
% temperatures near the melting point, often called a quasi-liquid layer (QLL).
% Molecular dynamics simulations of the facets of ice-I$_\mathrm{h}$ exposed
% to vacuum performed by Conde, Vega and Patrykiejew have found QLL widths of
% approximately 10 \AA\ at 3 K below the melting point\cite{Conde08}.
% Similarly, Limmer and Chandler have used course grain simulations and 
% statistical field theory to estimated QLL widths at the same temperature to
% be about 3 nm\cite{Limmer14}.
% Recently, Sazaki and Furukawa have developed an experimental technique with 
% sufficient spatial and temporal resolution to visulaize and quantitatively
% analyze QLLs on ice crystals at temperatures near melting\cite{Sazaki10}. They 
% have found the width of the QLLs perpindicular to the surface  at -2.2$^{o}$C 
% to be $\mathcal{O}$(\AA). They have also seen the formation of two immiscible
% QLLs, which displayed different stabilities and dynamics on the crystal 
% surface\cite{Sazaki12}. Knowledge of the hydrophilicities of each
% of the crystal facets would help further our understanding of the properties 
% and dynamics of the QLLs.
  
% Presented here is the follow up to our previous paper\cite{Louden13}, in which
% the basal and prismatic facets of an ice-I$_\mathrm{h}$/water interface were
% investigated where the ice was sheared relative to the liquid. By using a 
% recently developed velocity shearing and scaling approach to reverse 
% non-equilibrium molecular dynamics (VSS-RNEMD), simultaneous temperature and 
% velocity gradients can be applied to the system, which allows for measurment
% of friction and thermal transport properties while maintaining a stable 
% interfacial temperature\cite{Kuang12}. Structural analysis and dynamic
% correlation functions were used to probe the interfacial response to a shear, 
% and the resulting solid/liquid kinetic friction coefficients were reported.
% In this paper we present the same analysis for the pyramidal and secondary 
% prismatic facets, and show that the differential interfacial friction 
% coefficients for the four facets of ice-I$_\mathrm{h}$ are determined by their 
% relative hydrophilicity by means of dynamics water contact angle
% simulations. 

% The local tetrahedral order parameter, $q(z)$, is given by
% \begin{equation}
% q(z) = \int_0^L \sum_{k=1}^{N} \Bigg(1 -\frac{3}{8}\sum_{i=1}^{3}
% \sum_{j=i+1}^{4} \bigg(\cos\psi_{ikj}+\frac{1}{3}\bigg)^2\Bigg)
% \delta(z_{k}-z)\mathrm{d}z \Bigg/ N_z
% \label{eq:qz}
% \end{equation}
% where $\psi_{ikj}$ is the angle formed between the oxygen sites of molecules
% $i$,$k$, and $j$, where the centeral oxygen is located within molecule $k$ and 
% molecules $i$ and $j$ are two of the closest four water molecules
% around molecule $k$. All four closest neighbors of molecule $k$ are also
% required to reside within the first peak of the pair distribution function
% for molecule $k$ (typically $<$ 3.41 \AA\ for water). 
% $N_z = \int\delta(z_k - z) \mathrm{d}z$ is a normalization factor to account 
% for the varying population of molecules within each finite-width bin.


% The hydrophobicity or hydrophilicity of a surface can be described by the
% extent a droplet of water wets the surface. The contact angle formed between
% the solid and the liquid, $\theta$, which relates the free energies of the 
% three interfaces involved, is given by Young's equation.
% \begin{equation}\label{young}
% \cos\theta = (\gamma_{sv} - \gamma_{sl})/\gamma_{lv}
% \end{equation} 
% Here $\gamma_{sv}$, $\gamma_{sl}$, and $\gamma_{lv}$ are the free energies
% of the solid/vapor, solid/liquid, and liquid/vapor interfaces respectively.
% Large contact angles ($\theta$ $\gg$ 90\textsuperscript{o}) correspond to low 
% wettability and hydrophobic surfaces, while small contact angles 
% ($\theta$ $\ll$ 90\textsuperscript{o}) correspond to high wettability and 
% hydrophilic surfaces. Experimentally, measurements of the contact angle
% of sessile drops has been used to quantify the extent of wetting on surfaces
% with thermally selective wetting charactaristics\cite{Tadanaga00,Liu04,Sun04},
% as well as nano-pillared surfaces with electrically tunable Cassie-Baxter and 
% Wenzel states\cite{Herbertson06,Dhindsa06,Verplanck07,Ahuja08,Manukyan11}. 
% Luzar and coworkers have done significant work modeling these transitions on 
% nano-patterned surfaces\cite{Daub07,Daub10,Daub11,Ritchie12}, and have found 
% the change in contact angle to be due to the external field perturbing the 
% hydrogen bonding of the liquid/vapor interface\cite{Daub07}.

% SI stuff:

% Correlation functions:
% To compute eq. \eqref{C(t)1}, each 0.5 ns simulation was 
% followed by an additional 200 ps NVE simulation during which the 
% position and orientations of each molecule were recorded every 0.1 ps.


\end{article}

\begin{figure}
\includegraphics[width=\linewidth]{Droplet}
\caption{\label{fig:Droplet} Computational model of a droplet of
  liquid water spreading over the prismatic $\{1~0~\bar{1}~0\}$ facet
  of ice, before (left) and 2.6 ns after (right) being introduced to the
  surface.  The contact angle ($\theta$) shrinks as the simulation
  proceeds, and the long-time behavior of this angle is used to
  estimate the hydrophilicity of the facet.}
\end{figure}

\begin{figure}
\includegraphics[width=2in]{Shearing}
\caption{\label{fig:Shearing} Computational model of a slab of ice
  being sheared through liquid water.  In this figure, the ice is
  presenting two copies of the prismatic $\{1~0~\bar{1}~0\}$ facet
  towards the liquid phase.  The RNEMD simulation exchanges both
  linear momentum (indicated with arrows) and kinetic energy between
  the central box and the box that spans the cell boundary.  The
  system responds with weak thermal gradient and a velocity profile
  that shears the ice relative to the surrounding liquid.}
\end{figure}

% \begin{figure}
% \includegraphics[width=\linewidth]{ContactAngle}
% \caption{\label{fig:ContactAngle} The dynamic contact angle of a
%   droplet after approaching each of the four ice facets.  The decay to
%   an equilibrium contact angle displays similar dynamics.  Although
%   all the surfaces are hydrophilic, the long-time behavior stabilizes
%   to significantly flatter droplets for the basal and pyramidal
%   facets.  This suggests a difference in hydrophilicity for these
%   facets compared with the two prismatic facets.}
% \end{figure}

% \begin{figure}
% \includegraphics[width=\linewidth]{Pyr_comic_strip}
% \caption{\label{fig:pyrComic} Properties of the pyramidal interface
%   being sheared through water at 3.8 ms\textsuperscript{-1}. Lower
%   panel: the local tetrahedral order parameter, $q(z)$, (circles) and
%   the hyperbolic tangent fit (turquoise line).  Middle panel: the
%   imposed thermal gradient required to maintain a fixed interfacial
%   temperature of 225 K. Upper panel: the transverse velocity gradient
%   that develops in response to an imposed momentum flux. The vertical
%   dotted lines indicate the locations of the midpoints of the two
%   interfaces.}
% \end{figure}

% \begin{figure}
% \includegraphics[width=\linewidth]{SP_comic_strip}
% \caption{\label{fig:spComic} The secondary prismatic interface with a shear 
% rate of 3.5 \
% ms\textsuperscript{-1}. Panel descriptions match those in figure \ref{fig:pyrComic}.}
% \end{figure}

% \begin{figure}
% \includegraphics[width=\linewidth]{Pyr-orient}
% \caption{\label{fig:PyrOrient} The three decay constants of the
%   orientational time correlation function, $C_2(z,t)$, for water as a
%   function of distance from the center of the ice slab. The vertical
%   dashed line indicates the edge of the pyramidal ice slab determined
%   by the local order tetrahedral parameter. The control (circles) and
%   sheared (squares) simulations were fit using shifted-exponential
%   decay (see Eq. 9 in Ref. \citealp{Louden13}).}
% \end{figure}  

% \begin{figure}
% \includegraphics[width=\linewidth]{SP-orient-less}
% \caption{\label{fig:SPorient} Decay constants for $C_2(z,t)$ at the secondary 
% prismatic face. Panel descriptions match those in \ref{fig:PyrOrient}.}
% \end{figure}


\begin{table}[h]
\centering
\caption{Structural and dynamic properties of the interfaces of
  Ice-I$_\mathrm{h}$ with water.\label{tab:kappa}}
\begin{tabular}{r|cc|cc|cccc}  
\toprule
\multirow{2}{*}{Interface} & \multicolumn{2}{c|}{Channel Size} &\multicolumn{2}{c|}{Droplet} & \multicolumn{4}{c}{Shearing\footnotemark[1]}\\
  & Width (\AA) & Depth (\AA) & $\theta_{\infty}$ ($^\circ$)  & $k_\mathrm{spread}$  (ns\textsuperscript{-1}) &
$\kappa_{x}$  & $\kappa_{y}$ & $d_\mathrm{struct}$ (\AA) &  $d_\mathrm{dyn}$ (\AA) \\ 
\midrule
Basal  $\{0001\}$                    & 4.49 & 1.30 & $34.1(9)$ &$0.60(7)$
& $5.9(3)$ & $6.5(8)$ & $3.2(4)$ & $2(1)$  \\
Pyramidal  $\{2~0~\bar{2}~1\}$       & 8.65 & 1.37 & $35(3)$ &  $0.7(1)$ &
$5.8(4)$ & $6.1(5)$ & $3.2(2)$ & $2.5(3)$\\
Prismatic  $\{1~0~\bar{1}~0\}$       & 6.35 & 2.25 & $45(3)$ & $0.75(9)$ &
$3.0(2)$ & $3.0(1)$ & $3.6(2)$ & $4(2)$ \\
Secondary Prism  $\{1~1~\bar{2}~0\}$ & 6.35 & 2.25 & $43(2)$ & $0.69(3)$ &
$3.5(1)$ & $3.3(2)$ & $3.2(2)$ & $5(3)$ \\ 
\bottomrule
\end{tabular}
\begin{flushleft}
\footnotemark[1]\footnotesize{Liquid-solid friction coefficients ($\kappa_x$ and
  $\kappa_y$) are expressed in 10\textsuperscript{-4} amu
  \AA\textsuperscript{-2} fs\textsuperscript{-1}.} \\
\footnotemark[2]\footnotesize{Uncertainties in
  the last digits are given in parentheses.} 
\end{flushleft}
\end{table}

% Basal  $\{0001\}$                    & 4.49 & 1.30 & $34.1 \pm 0.9$ &$0.60 \pm 0.07$
% & $5.9 \pm 0.3$ & $6.5 \pm 0.8$ & $3.2 \pm 0.4$ & $2 \pm 1$  \\
% Pyramidal  $\{2~0~\bar{2}~1\}$       & 8.65 & 1.37 & $35 \pm 3$ &  $0.7 \pm 0.1$ &
% $5.8 \pm 0.4$ & $6.1 \pm 0.5$ & $3.2 \pm 0.2$ & $2.5 \pm 0.3$\\
% Prismatic  $\{1~0~\bar{1}~0\}$       & 6.35 & 2.25 & $45 \pm 3$ & $0.75 \pm 0.09$ &
% $3.0 \pm 0.2$ & $3.0 \pm 0.1$ & $3.6 \pm 0.2$ & $4 \pm 2$ \\
% Secondary Prism  $\{1~1~\bar{2}~0\}$ & 6.35 & 2.25 & $43 \pm 2$ & $0.69 \pm 0.03$ &
% $3.5 \pm 0.1$ & $3.3 \pm 0.2$ & $3.2 \pm 0.2$ & $5 \pm 3$ \\ 


\end{document}
Revision:	4268
Committed:	Wed Dec 31 17:25:13 2014 UTC (10 years, 10 months ago) by gezelter
Content type:	application/x-tex
File size:	44029 byte(s)
Log Message:	JCP version now that PNAS said no