Published on Web Date: July 01, 2010 r2010 American Chemical Society 2154 DOI: 10.1021/jz100479p | J. Phys. Chem. Lett. 2010, 1, 2154–2158 pubs.acs.org/JPCL Effective Monte Carlo Scheme for Multicomponent Gas Adsorption and Enantioselectivity in Nanoporous Materials Titus S. van Erp,* ,† David Dubbeldam, ‡ Tom P. Caremans, † Sofia Calero, § and Johan. A. Martens † † Centrum voor Oppervlaktechemie en Katalyse, K.U. Leuven, Kasteelpark Arenberg 23, B-3001 Leuven, Belgium, ‡ Van't Hoff Institute for Molecular Sciences, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WVAmsterdam, The Netherlands, and § Department of Physical, Chemical, and Natural Systems, University Pablo de Olavide, Ctra. Utrera km 1, 41013 Seville, Spain ABSTRACT We devise an efficient Monte Carlo scheme to study the adsorption of a multicomponent gas in a nanoporous material. The configurational bias move is extended by a novel replica exchange procedure where the configurations of the different simulations describing one particular gas content are being swapped. For chiral mixtures, the efficiency can be further improved using the chiral inversion move. The method is demonstrated for an Ising-type model and a complicated realistic zeolite system. SECTION Statistical Mechanics, Thermodynamics, Medium Effects T he understanding of adsorption of multicomponent gases in nanoporous materials, like zeolites or metallic organic frameworks, is of technological importance for the development of molecular separation techniques. 1 Possibly, the most challenging of these is the separation of chiral compounds. The chemical and physical properties of the left- and right-handed enantiomers are so similar that differentiation becomes highly problematic. Yet, enantiopure compounds are vital to many applications in the pharmaceu- tical and agrochemical industry. Differences in adsorption and diffusion of the two enantiomers inside of a chiral adsor- bent can be exploited for this purpose. 2 However, suitable chiral nanoporous materials are often expensive and only functional for a limited group of chiral molecules. In the past, computer simulations have successfully discovered nontri- vial separation mechanisms between linear and branched alkanes. 3 Recently, we found a new type of enantioselectivity that might ultimately be used for chiral separation without using a chiral adsorbent. 4 An important development that has made these types of simulations feasible is the configurational bias Monte Carlo (CBMC) approach. 5 CBMC grows the mole- cule inside of the framework in successive steps so that the energetically unfavorable overlaps are eliminated. Still, in the aforementioned study, 4 we faced some limitations of the CBMC method, which demanded the development of some additional techniques. Our solution turns out to be very effective and applicable to a wide range of multicomponent gas adsorption studies, specifically those containing chiral compounds. Notwithstanding the prodigious increase of efficiency of CBMC compared to randomized insertions, acceptance rates are still low in dense systems. The situation is often compli- cated further by an extremely slow ergodicity due to the presence of metastable states. These are groups of configura- tions that inhibit strongly connected clusters where molecules of the same type or, on the contrary, of the opposite type are arranged in a specific order. The CBMC move that changes the identity of one molecule will always break up the energetically favorable situation and is therefore rejected as it is unable to flip a whole pair or cluster of molecules. An important class of methods that deals with metasta- bility is parallel tempering, also know as replica exchange (RE). 6 In this method, one performs several simulations in parallel at different temperatures. With a certain frequency and acceptance probability, the configuration at one tempera- ture is being swapped with a lower-temperature simulation. The RE method allows the simulations at low T to hop from one basin to another without having to cross the intermediate barriers physically. This method has proven to be very suc- cessful in various systems. A drawback of this approach is the costly expense of additional simulations that provide little value for a single-temperature experiment. Moreover, at maximal loading conditions, the high T simulations tend to position molecules too close to each other and the crystal framework, so that the swapping move gets rejected. In addition, RE can not overcome entropic barriers. However, the general concept of RE is not limited to temperature. Fukunishi et al. 7 developed the Hamiltonian RE method. Metadynamics 8 and transition interface sampling 9 were con- siderably sped up using RE methods in reaction coordinate space. 10,11 In this Letter, we show the effectiveness of using Received Date: April 13, 2010 Accepted Date: June 3, 2010