On the comparison of different barostat implementations for the prediction of the breathing behavior in MIL-53 frameworks

Sven Rogge, Louis Vanduyfhuys, Toon Verstraelen, Guillaume Maurin, and Veronique Van Speybroeck

Metal-organic frameworks (MOFs) are microporous crystalline materials, compromising metallic clusters or chains connected by organic linkers, forming a scaffold-like network with cavities and/or channels, and were first synthesized in 1999 [1]. Some of these frameworks, such as MIL-53, may undergo reversible, large-amplitude structural deformations. These transitions can be induced by applying a mechanical pressure [2], paving the way for mechanical energy storage applications.

However, these applications require a clear understanding of the response of the material on the applied pressure, especially the pressure at which the transition is induced. We determine this transition pressure with the aid of Molecular Dynamics (MD) simulations in the NPT ensemble, allowing for anisotropic volume transformations. Since the time needed for these materials to undergo these large-amplitude transformations is on the order of several hundreds of picoseconds, a complete quantummechanical treatment of the forces is not yet feasible. Therefore, force fields, specifically aimed for these MOFs and developed at the Center for Molecular Modeling [3], are used in conjunction with a simulation package that focuses on using these force fields [4].

To model the NPT ensemble, we have implemented the Nosé-Hoover chain thermostat [5-7], while the barostat implementation of Martyna, Tobias and Klein is used [8]. The simulation results are compared to the Berendsen [9] and Langevin barostats [10], as well as to experimental results from our partners in Montpellier [11].

From these MD simulations, it is observed that pressure fluctuations are significant, and may be up to four orders of magnitude larger than the average applied pressure. As a consequence, one may expect that transitions can be driven by these fluctuations, rather than by the mean applied pressure. We propose a method to distinguish between both possibilities, allowing to successfully determine the transition pressure. Moreover, the effect of the Berendsen and Langevin barostat on these pressure fluctuations is quantified for the MIL-53 type frameworks.

[1] H. Li, M. Eddaoudi, et al. Nature 402, 276 (1999).
[2] I. Beurroies, M. Boulhout, et al., Angew. Chem. Int. Edit., 49, 7526 (2010).
[3] L. Vanduyfhuys, T. Verstraelen, et al., J. Chem. Theory Comput., 8, 3217 (2012).
[4] T. Verstraelen, L. Vanduyfhuys, et al., http://molmod.ugent.be/software/
[5] S. Nosé, J. Chem. Phys., 81, 511 (1984).
[6] W. G. Hoover, Phys. Rev. A, 31, 1695 (1985).
[7] G. J. Martyna, M. L. Klein, et al., J. Chem. Phys., 97, 2635 (1992).
[8] G. J. Martyna, D. J. Tobias, et al., J. Chem. Phys., 101, 4177 (1994).
[9] H. J. C. Berendsen, J. P. M. Postma, et al., J. Chem. Phys., 81, 3684 (1984).
[10] G. Bussi, M. Parrinello, Phys. Rev. E, 75, 056707 (2007).
[11] P. G. Yot, Z. Boudene, et al., Chem. Commun., 50, 9462 (2014).