During the formation of metal-organic frameworks (MOFs), metal centres can coordinate with the intended organic linkers, but also with solvent molecules. In this case, subsequent activation by removal of the solvent molecules creates unsaturated 'open' metal sites known to have a strong affinity for CO 2 molecules, but their interactions are still poorly understood. Common force fields typically underestimate by as much as two orders of magnitude the adsorption of CO 2 in open-site Mg-MOF-74, which has emerged as a promising MOF for CO 2 capture. Here we present a systematic procedure to generate force fields using high-level quantum chemical calculations. Monte Carlo simulations based on an ab initio force field generated for CO 2 in Mg-MOF-74 shed some light on the interpretation of thermodynamic data from flue gas in this material. The force field describes accurately the chemistry of the open metal sites, and is transferable to other structures. This approach may serve in molecular simulations in general and in the study of fluid-solid interactions.
Most energy scenarios project a significant increase in the role of renewable energy sources 1 . These scenarios also predict an even higher increase in our energy needs. As a consequence, although the relative consumption of fossil fuels will be decreasing, in absolute terms we will continue to burn more coal. In such a scenario, carbon capture and sequestration will be one of the only viable technologies to mitigate CO 2 emissions 1,2 . At present the cost associated with the capture of CO 2 from flue gas is one of the bottlenecks in the large-scale deployment of this technology. Of particular concern is that the efficiency of a coal-fired power plant decreases by as much as 30-40% (ref. 3) because of the energy required to separate and compress CO 2 . The aim of decreasing this parasitic load has motivated the search for novel materials 4,5 .A promising class of materials is metal-organic frameworks (MOFs) 4,6 . MOFs are crystalline materials that consist of metal centres connected by organic linkers. These materials have an extremely large internal surface area and, compared to other common adsorbents, promise very specific customization of their chemistry. By changing the metal and the linker, one can in principle generate many millions of possible materials. In practice, however, we can synthesize only a very small fraction of these materials, and it is important to develop a theoretical method that supports the experimental efforts to identify an ideal MOF for carbon capture. A key aspect is the ability to predict the properties of a MOF before the material is synthesized. At present it is possible to carry out accurate quantum chemical calculations on these types of systems 7 . State-of-the-art density functional theory (DFT) calculations provide important insights into the energetics and siting of CO 2 at zero Kelvin 7 . The separation of flue gas, however, requires thermodynamic information (for example, adsorption isotherms) at flue-gas conditions (40...