Structure, Properties, and Phase Transformations of Water Nanoconfined between Brucite-like Layers: The Role of Wall Surface Polarity [original]

Citation: Tsukanov, A.A.; Shilko, E.V.;

Popov, M. Structure, Properties, and

Phase Transformations of Water

Nanoconfined between Brucite-like

Layers: The Role of Wall Surface

Polarity. Materials 2022,15, 3043.

https://doi.org/10.3390/

ma15093043

Academic Editor:

Edward Bormashenko

Received: 30 March 2022

Accepted: 20 April 2022

Published: 22 April 2022

Publisher’s Note: MDPI stays neutral

with regard to jurisdictional claims in

published maps and institutional affil-

iations.

Licensee MDPI, Basel, Switzerland.

This article is an open access article

distributed under the terms and

conditions of the Creative Commons

Attribution (CC BY) license (https://

creativecommons.org/licenses/by/

4.0/).

materials

Article

Structure, Properties, and Phase Transformations of Water

Nanoconfined between Brucite-like Layers: The Role of Wall

Surface Polarity

Alexey A. Tsukanov 1,* , Evgeny V. Shilko 2,* and Mikhail Popov 3

Center for Computational and Data-Intensive Science and Engineering (CDISE), Skolkovo Institute of Science

and Technology (Skoltech), 121205 Moscow, Russia

2Institute of Strength Physics and Materials Science of SB RAS, 634055 Tomsk, Russia

3Institut für Mechanik, Technische Universität Berlin, 10623 Berlin, Germany; [email protected]

*Correspondence: [email protected] (A.A.T.); [email protected] (E.V.S.)

Abstract:

The interaction of water with confining surfaces is primarily governed by the wetting

properties of the wall material—in particular, whether it is hydrophobic or hydrophilic. The hy-

drophobicity or hydrophilicity itself is determined primarily by the atomic structure and polarity

of the surface groups. In the present work, we used molecular dynamics to study the structure

and properties of nanoscale water layers confined between layered metal hydroxide surfaces with

a brucite-like structure. The influence of the surface polarity of the confining material on the prop-

erties of nanoconfined water was studied in the pressure range of 0.1–10 GPa. This pressure range

is relevant for many geodynamic phenomena, hydrocarbon recovery, contact spots of tribological

systems, and heterogeneous materials under extreme mechanical loading. Two phase transitions

were identified in water confined within 2 nm wide slit-shaped nanopores: (1) at p

= 3.3–3.4 GPa, the

liquid transforms to a solid phase with a hexagonal close-packed (HCP) crystal structure, and (2) at

= 6.7–7.1 GPa, a further transformation to face-centered cubic (FCC) crystals occurs. It was found

that the behavior of the confined water radically changes when the partial charges (and, therefore,

the surface polarity) are reduced. In this case, water transforms directly from the liquid phase to

an FCC-like phase at 3.2–3.3 GPa. Numerical simulations enabled determination of the amount

of hydrogen bonding and diffusivity of nanoconfined water, as well as the relationship between

pressure and volumetric strain.

Keywords:

nanoconfined water; nanopore; interface; surface polarity; phase transformation; layered

metal hydroxide; high-pressure crystallization; molecular dynamics

1. Introduction

Solid and liquid phases confined to a volume of several atomic sizes (nanoconfined

matter) may significantly or even qualitatively differ from the corresponding bulk phases.

This is valid both for nanosized particles or droplets and for liquids or solids confined

within nanosized pores or gaps [

–

]. Under conditions of nanoconfinement, many kinetic

and thermodynamic processes—including chemical reactions—demonstrate extraordinary

behavior not observed in bulk conditions [

–

]. This causes both nanoconfined matter and

processes occurring under low-dimensional constraints to be of increasing interest in a

variety of scientific and industrial areas [

]. For example, nanoconfined hydrides are

among the most promising solutions for hydrogen storage [

]. Confined ionic liquids

are used in promising supercapacitors (electrochemical devices that store energy through

the reversible adsorption of ionic species) with enhanced capacitance and the capability

of fast charging and discharging [

]. Nanoconfinement-based methods are used for

the analysis of long biomolecules [

], as well as for studies of hydrocarbon-recovery-

related problems using nanofluidic devices [

]. Matter in nanoconfined conditions is

Materials 2022,15, 3043. https://doi.org/10.3390/ma15093043 https://www.mdpi.com/journal/materials

Materials 2022,15, 3043 2 of 19

found everywhere in geological media, thereby making nanoconfinement effects essential

for the proper understanding of many phenomena in geoscience, including the physical

and mechanical properties of saturated porous media [

], formation of superhydrated

phases of minerals [

], dehydration processes of subducting slabs [

], deformation and

rupture of heterogeneous geomaterials [

], and lubrication effects under high-pressure

conditions [

]. Confinement [

] and nanoconfinement [

–

] strongly influence the

mechanical and frictional properties of lubricants in highly loaded contacts.

Taking into account nanoconfinement-related effects is important in solving many

materials science and nanotechnology problems, including the development of functional

nanomaterials (e.g., molecular sieves and ion-selective membranes [

–

]), bionanosen-

sors [

], highly ordered mesostructured composites with unusual mechanical and struc-

tural properties [

], auxetic materials [

], efficient sorbents [

], etc. [

–

]. Additionally,

in a variety of cases, nanoconfinement provides favorable conditions for the catalysis of

chemical reactions [41–43].

It is well known that confinement of thin layers has a strong effect on the phase state

and phase transformations of a substance. Water confined in slit-shaped nanopores of

7 Å width between two graphite plates shows solid-like behavior, whereas in pores of 9 Å

width or more this effect is not observed [

]. A similar effect (formation of 2D ice crystals)

was recently studied for water bilayers between mica and graphene [

]. Moreover,

the mechanism of nanoconfinement-induced freezing of liquids was experimentally in-

vestigated for cyclohexane, octamethylcyclotetrasiloxane, and toluene [

]. It was found

that when the nanopore width is below a certain critical value—about six molecular layers

of confined liquid—a liquid-like–solid-like phase transition occurs. Another example is

argon; when confined between atomically flat graphene nanosheets, argon does exist in

a solid phase, with mixed hexagonal close-packed (HCP) and face-centered cubic (FCC)

structures, at pressures significantly lower than the crystallization pressure of argon in a

bulk system [48].

Water is among the most interesting subjects for the study of nanoconfined matter, due

to its ubiquitous presence and importance for our planet. Even in a bulk state, water shows

an unusually large variety of structural forms (including many phases of ice [

]), and has

some well-known anomalous features. Confined water plays a key role in many practically

significant processes, such as the functioning of biological molecular machines [

], selective

ion filtration through biological channels [

], functioning of porins, and changes in the

properties of biological components in soft nanoconfinement [52].

The character of the interaction between water and the confining medium is largely

determined by the hydrophobicity/hydrophilicity of the pore walls which, in turn, depends

on both the atomic structure of the surface and the distribution of electric charge between

surface groups of atoms, i.e., surface polarity [53–55].

The present study is devoted to the question of how surface polarity can influence the

network structure and the density of hydrogen bonds in nanoconfined water, as well as

its mechanical properties at extremely high pressures. To carry out the analysis, we used

molecular dynamics (MD) simulation with atom-level models. We considered brucite-type

layered metal hydroxides as the nanopore wall material. This is a widespread natural

mineral class defined by the chemical composition Me

(II)

(OH)

, where Me

(II)

is a divalent

metal [

] or a mixture of metals [

]. The MD model allowed us to study the properties

and structure of water trapped in the interlayers of typical layered hydroxide minerals

under different pressures, and to directly compare the behavior of nanoconfined water

under hydrophilic and hydrophobic conditions in otherwise identical nanopores.

2. Materials and Methods

2.1. Molecular Dynamics Model

Molecular dynamics simulation with atomic-level description is a convenient tool for

studying matter under nanoconfinement. To investigate the structure and properties of

nanoconfined water, as well as the influence of wall–water electrostatic interaction on water

Materials 2022,15, 3043 3 of 19

behavior under high-pressure conditions, two molecular dynamics models of a slit-shaped

nanopore filled with water were developed. Each model consisted of an approximately

2 nm thick layer of water confined by two parallel metal hydroxide nanosheets located in

the x–y plane (Figure 1). Periodic boundary conditions (PBCs) were applied along the xand

yaxes.

Materials 2022, 15, x FOR PEER REVIEW 3 of 20

2. Materials and Methods

2.1. Molecular Dynamics Model

Molecular dynamics simulation with atomic-level description is a convenient tool

for studying matter under nanoconfinement. To investigate the structure and properties

of nanoconfined water, as well as the influence of wall–water electrostatic interaction on

water behavior under high-pressure conditions, two molecular dynamics models of a

slit-shaped nanopore filled with water were developed. Each model consisted of an ap-

proximately 2 nm thick layer of water confined by two parallel metal hydroxide

nanosheets located in the x–y plane (Figure 1). Periodic boundary conditions (PBCs) were

applied along the x and y axes.

Figure 1. Initial configuration of the modelled system’s periodic boxes 36[2Me

(II)

(OH)

·7H

equilibrated at the pressure 0.1 GPa: perspective view (a), top view (b), x–z view (c), and y–z views

(d).

In the first model, the nanopore walls are representative of a class of naturally oc-

curring metal hydroxides Me

(II)

(OH)

. The structural parameters of divalent iron hy-

droxide Fe(OH)

with near-medial lattice constant a [56] were chosen as representative

parameters for layered hydroxides of divalent metals within the trigonal crystal system

(Supplementary Table S1 and Supplementary Figure S1).

The second model is equivalent to the first one, except for modified partial atomic

charges (PACs), whose values were reduced by a factor of 10 compared to the first model

(Supplementary Table S2). Henceforth, we will refer to the first model as the “Naturally

charged host model” and the second model as the “Low-charged host model”. It should

be noted that both models have the overall composition 36[2Me

(II)

(OH)

·7H

O], and that

there is no net electric charge. In other words, we studied two “host-guest” systems with

the same atomic structure of the nanopore walls and the same number of “guest” water

Figure 1.

Initial configuration of the modelled system’s periodic boxes 36[2Me

(II)

(OH)

2·

O] equili-

brated at the pressure 0.1 GPa: perspective view (a), top view (b), x–z view (c), and y–z views (d).

In the first model, the nanopore walls are representative of a class of naturally occur-

ring metal hydroxides Me

(II)

(OH)

. The structural parameters of divalent iron hydroxide

Fe(OH)

with near-medial lattice constant a[

] were chosen as representative parameters

for layered hydroxides of divalent metals within the trigonal crystal system (Supplementary

Table S1 and Supplementary Figure S1).

The second model is equivalent to the first one, except for modified partial atomic

charges (PACs), whose values were reduced by a factor of 10 compared to the first model

(Supplementary Table S2). Henceforth, we will refer to the first model as the “Naturally

charged host model” and the second model as the “Low-charged host model”. It should

be noted that both models have the overall composition 36[2Me

(II)

(OH)

2·

O], and that

there is no net electric charge. In other words, we studied two “host-guest” systems with

the same atomic structure of the nanopore walls and the same number of “guest” water

molecules. The only difference was that the confining walls were hydrophilic in the first

model, and hydrophobic in the second.

The crystal structure obtained for deuterated iron(II) hydroxide Fe[O(H

0.86

0.14

)]

by Parise et al. [

] was used for the nanosheet wall model. This material was chosen

because it has a crystal structure similar to many brucite-like metal hydroxides (such

as Ni(OH)

, Mg(OH)

, Co(OH)

, Zn(OH)

, Mn(OH)

, Cd(OH)

, Ca(OH)

), and has an

Materials 2022,15, 3043 4 of 19

intermediate lattice constant among them (see comparison in Supplementary Table S1 and

Supplementary Figure S1).

The force field parameters that describe the nanopore walls (including the PACs,

Lennard-Jones, and covalent bond parameters) were assigned in the naturally charged host

model in accordance with the CLAYFF force field [

]. The only exception was that the

Lennard-Jones parameters for hydrogen atoms were parameterized as described in [

with non-zero values r

= 0.449 Å and

= 0.1925 kJ/mol [

], aiming for better compatibility

with the TIP3P water model. Partial atomic charges of walls in the low-charged host model

system were 10 times lower than in the first model. The PAC values used in the present

study are listed in Supplementary Table S2.

Both models contain a constant number of water molecules (N

= 7

36 = 252)

confined between two parallel nanosheets of metal hydroxide. Each nanosheet contains 36

(II)

(OH)

units. The entire system can therefore be described as 36[2Me

(II)

(OH)

2·

O].

The TIP3P model was used to parameterize the water molecules [

]. The covalent

bonds and angles of the model are described in Supplementary Figure S2 and Supplemen-

tary Table S3.

The dimensions of the simulation box—X = 16.965 Å and Y = 19.590 Å—remained

unchanged during the runs. The size of the system along the zaxis and, therefore, the

width of the nanopore, was self-adjusting depending on external load. The simulation for

each applied pressure was 3 ns long.

Larger systems were additionally prepared by replication of the base models twice

along both the xand yaxes. The dimensions of the enlarged system were X = 33.93 Å,

Y = 39.18 Å, with an overall formula 144[2Me

(II)

(OH)

2·

O]. These models contained 1008

water molecules. They were used in the series of additional verifying simulations near the

phase transformation points. The duration of each additional simulation was 5 ns.

The cutoff distance for pairwise interactions was 10 Å, with switching to a smooth

damping function starting at 8 Å. The long-range term of the electrostatic interaction was

evaluated with the PPPM (particle–particle–particle–mesh) method, with a relative error of

−4

[

]. Temperature was constant and equal to 310 K in all simulations. The temperature

was maintained by a Nose–Hoover thermostat [64]. The time step was 1 fs.

The traditional approach in molecular dynamics to the implementation of confinement

of matter is the use of rigid [

] and elastic [

] walls. We used the rigid wall approximation,

with all heavy (non-hydrogen) atoms of the Me

(II)

(OH)

being part of a single rigid body

(one rigid body per nanosheet). Thus, the metal and oxygen atoms of the walls were able

to move only in the vertical direction z. Surface hydrogen atoms, being covalently bonded

with oxygen atoms, had no constraints applied.

The model samples were subjected to uniaxial compressional loading along the z-axis

(see Figure 1). The coordinates of the heavy atoms of the lower wall were fixed during

simulation, whereas the upper wall behaved as a piston, providing the specified load on

the confined water along the zaxis (Figure 1). The load was applied by an external force

-f

, which was uniformly distributed over all non-hydrogen atoms of the upper wall piston.

The force was directed downward; its magnitude was calculated as the product of the

specified pressure and the area of the wall–piston surface. The external (applied) pressure

pvaried in the range of 0.1–10 GPa. This range particularly corresponds to tectonic plates

at the characteristic depth interval from 3 km (upper crust) to 300 km (upper mantle) [20],

or to highly stressed contact spots in heavily loaded tribological couples [67].

Using the described models, a comprehensive analysis of nanoconfined water structure

and properties at different pressures was carried out. The analysis covered:

1. Water density and compressibility;

Spatial arrangement of water molecules, including the radial distribution functions

(RDFs);

3. The number of hydrogen bonds;

4. The self-diffusion coefficient.

Materials 2022,15, 3043 5 of 19

2.2. Advantages of the Model

As previously mentioned, the choice of the Fe(OH)

structure with a lattice constant

of a= 3.265 Å [

] as the basis of the model allowed us to extend the results of the study to

the general class of brucite-like layered hydroxides of divalent metals, or at least to those

of Ni (3.117 Å), Mg (3.147 Å), Co (3.173 Å), Zn (3.194 Å), Mn (3.316 Å) and, probably, Cd

(3.499 Å), due to the similar values of their lattice constants [44].

2.3. Assumptions of the Model

We assumed that the utilized force field parameters were sufficiently accurate to model

the studied systems at high pressures, and in particular that the TIP3P model was suitable

for describing water behavior up to a pressure of 10 GPa

We also assumed that the mineral did not change its structure under loading and,

therefore, that the rigid model was applicable (except for surface hydrogen atoms). This

assumption is partially supported by data obtained by Hermann and Mookherjee [

According to a computationally derived thermodynamic phase diagram for brucite at

~300 K [

], the brucite Mg(OH)

retains its layered structure up to a pressure of ~20 GPa.

The rigid model of the wall structure is consistent with these data.

2.4. Limitations of the Model

The modeled systems had a fixed ratio of water molecules N

to the number of metal

atoms (7:2), and only the case of constant dimensions along the xand yaxes was considered.

3. Results

3.1. Structure and Properties of Nanoconfined Water

The simulations revealed two solid high-pressure phases of water confined between natu-

rally charged Me

(II)

(OH)

sheets (in addition to the low-pressure liquid phase). These phases can

be clearly seen in all numerically obtained dependencies (see Figures 2,3,4and 5a,c). Two phase

transitions at p

~3.1 GPa and p

~6.7 GPa were determined within the studied pressure range for

the base model 36[2Me

(II)

(OH)

2·

O] (Figure 2a, black curves). More accurate estimates were

obtained using the fourfold-enlarged model 144[2Me

(II)

(OH)

2·

O] in prolonged simulations

(Figure 4, red curves in the left and central panels): p1~3.3–3.4 GPa, and p2~6.7–7.1 GPa.

In the first phase transition at the pressure p

, the random arrangement of water

molecules became ordered (Figure 3a), and the diffusion coefficient of water decreased

by three orders of magnitude (Figure 5c, solid black curve) from D~3.24

−10

/s to

4.33

−13

/s. It is worth mentioning here that our estimate of the self-diffusion

coefficient at p= 0.1 GPa for nanoconfined water was 3.1

−9

/s, which is of the same

order of magnitude as that of bulk water Dbulk~2.299 ×10−9m2/s [69].

According to generally accepted data, ordinary water in the bulk state crystallizes

into ice VI [

] at a phase transition pressure of about 1.0–1.17 GPa (at temperatures of

300–310 K) (see Supplementary Equations (1)–(3)). However, experimental evidence of the

possibility of metastable supercompressed bulk liquid water at higher pressures has also

been reported [

]. Dolan et al. [

] were the first to show the formation of cubic ice (ice

VII) under shock compression with GPa peak pressures. They particularly showed that the

solid phase becomes more stable than the liquid phase at applied pressures above 2 GPa,

and that the phase transformation takes place within nanoseconds instead of seconds

under ambient conditions. Later, Dolan et al. [

] demonstrated a practical limit for the

metastable liquid phase. They showed that bulk water under isentropic shock compression

can homogeneously transform into a high-pressure phase (ice VII) at a pressure of 7 GPa

(a practical limit for the metastable liquid phase) [

–

]. The abovementioned values

1 GPa and 7 GPa bound the pressure range in which crystallization of water is possible

at ambient or moderately elevated temperatures. The specific value of crystallization

pressure depends (at least for microscale, and even more so for nanoscale volumes of water)

on interface hydrophilicity—that is, on hydrogen bonding between the wall surface and

water [73,74].

Loading more pages...