(395ad) Mechanism of Adsorption Hysteresis in Mesoporous Silica With Surface Roughness

The siliceous mesoporous materials such as MCM-41 and SBA-15 have attracted much attention because of their potential use as catalyst supports, separation, and drug delivery, etc. These materials are also regarded as the most suitable model adsorbent for a fundamental study of capillary condensation. Vapor-liquid phase transitions of fluids in the mesoporous silica materials have been extensively studied by experiments, theory, and molecular simulations, and these studies reveal that adsorption hysteresis loop depends on pore diameter and temperature of the system. However, the mechanism of the adsorption hysteresis has still been a missing piece of the puzzle since MCM-41 appeared in 1992, because cylindrical pore models with smooth and structureless wall have been used to understand the capillary condensation behavior. Recent theoretical and simulation studies have shown that surface roughness of pore wall affects adsorption and capillary condensation of fluids in the siliceous mesoporous materials [1,2]. We therefore constructed an atomistic silica pore model mimicking MCM-41, which has molecular-level surface roughness, with the aid of the electron density profile (EDP) of MCM-41 obtained from X-ray diffraction data, and performed the molecular simulations to understand the mechanism of the adsorption hysteresis.

Molecular dynamics simulation was used for constructing a bulk fused silica with the BKS type potential. The system was equilibrated in the canonical ensemble at 4000 K, and subsequently, the fused silica was quenched from 4000 K to room temperature. To obtain an atomistic MCM-41 model, the amorphous silica block was carved out to reproduce the experimentally determined EDP of MCM-41 [3]. Adsorption isotherms of Lennard-Jones argon and nitrogen on the atomistic MCM-41 model at 75 K, 80 K, and 87 K were obtained by the grand canonical Monte Carlo (GCMC) method and the gauge cell Monte Carlo (MC) method [4]. Adsorption isotherms of argon and nitrogen on MCM-41 samples (pore diameter: 3.2 nm and 4.0 nm) at 75 K, 80 K, and 87 K were also measured by an adsorption apparatus consisting of a cryostat with a helium closed-cycle refrigerator and BELSORP-max (BEL, Japan) to compare with the simulation results.

The GCMC isotherms of argon and nitrogen on the atomistic MCM-41 model at 75 K, 80 K, and 87 K were in good agreement with the experimental data, which suggests that our model provide a good description of the surface roughness of MCM-41. Then, the simulated equilibrium capillary condensation pressures give close agreement with the experimental desorption branches, which indicates that the experimental desorption branch comes from thermodynamic equilibrium vapor-liquid transition. However, a large difference in the adsorption branch between the experiment and simulation was observed for all the cases. This suggests that the energy fluctuation of the system to overcome an energy barrier for the vapor-liquid transition is different between the experiment and the GCMC simulation. We therefore calculated canonical work function [5] by integrating a sigmoid adsorption isotherm obtained from the gauge cell MC method. The canonical work function provides a work required for a state change from a multilayer adsorption state to a capillary condensed state at a constant chemical potential. The obtained canonical work function shows that the energy barrier for the capillary condensation becomes small with increasing pressure, and if the system cannot overcome the energy barrier at the thermodynamic equilibrium condition, spontaneous capillary condensation should occur at a higher pressure than the equilibrium transition pressure. The energy barrier obtained from a comparison with the experimental adsorption branch of argon for MCM-41 (pore diameter 4.0 nm) at 75 K was 0.156 kT per adsorbed molecule. We assumed it as an energy fluctuation of the system at the spontaneous capillary condensation pressure, and predicted the capillary condensation pressure of argon on the atomistic MCM-41 model at 80 K and 87 K. The predicted spontaneous condensation pressures are in good agreement with the experimental data. We also determined the spontaneous condensation pressures of argon and nitrogen for the other atomistic MCM-41 model at all the temperatures, by setting the energy fluctuation of the system of 0.156 kT per adsorbed molecule, and succeeded in obtaining excellent agreement with the experimental adsorption branches for MCM-41 (pore diameter 3.2 nm).

[1] B. Coasne et al., Langmuir 22, 194 (2006).

[2] N. Muroyama et al., J. Phys. Chem. C 112, 10803 (2008).

[3] H. Tanaka et al., Adsorption 19, 631 (2013).

[4] A. V. Neimark and A. Vishnyakov, Phys. Rev. E 62, 4611 (2000).

[5] A.Vishnyakov and A. V. Neimark, J. Phys. Chem. B 110, 9403 (2006).