Diffusive Dynamics of Water inside Hydrophobic Carbon Micropores Studied by Neutron Spectroscopy and Molecular Dynamics Simulation

TL;DR

This study investigates how water molecules behave inside nanometer-sized hydrophobic carbon pores, revealing slowed diffusion and the influence of pore size on water dynamics through neutron spectroscopy and molecular simulations.

Contribution

It provides new insights into water dynamics in sub-12 Å pores, highlighting the pore size effect over shape and comparing experimental results with molecular dynamics simulations.

Findings

Water mobility slows down as temperature decreases.

Pore size, not shape, primarily influences water dynamics.

Molecular simulations show reasonable agreement at low Q values.

Abstract

When water molecules are confined to nanoscale spacings, such as in the nanometer size pores of activated carbon fiber (ACF), their freezing point gets suppressed down to very low temperatures ($\sim$ 150 K), leading to a metastable liquid state with remarkable physical properties. We have investigated the ambient pressure diffusive dynamics of water in microporous Kynol\texttrademark ACF-10 (average pore size $\sim$11.6 {\AA}, with primarily slit-like pores) from temperature $T=$ 280 K in its stable liquid state down to $T=$ 230 K into the metastable supercooled phase. The observed characteristic relaxation times and diffusion coefficients are found to be respectively higher and lower than those in bulk water, indicating a slowing down of the water mobility with decreasing temperature. The observed temperature-dependent average relaxation time $<\tau>$ when compared to previous findings indicate that it is the size of the confining pores - not their shape - that primarily affects the dynamics of water for pore sizes larger than 10 {\AA}. The experimental observations are compared to complementary molecular dynamics simulations of a model system, in which we studied the diffusion of water within the 11.6 {\AA} gap of two parallel graphene sheets. We find generally a reasonable agreement between the observed and calculated relaxation times at the low momentum transfer $Q$ ($Q\le 0.9$ \AA${^{-1}}$). At high $Q$ however, where localized dynamics becomes relevant, this ideal system does not satisfactorily reproduce the measurements. The best agreement is obtained for the diffusion parameter $D$ associated with the hydrogen-site when a representative stretched exponential function, rather than the standard bi-modal exponential model, is used to parameterize the self-correlation function $I(Q,t)$.