**6.1 Opportunities**

The most significant opportunity of employing MD simulations for obtaining gas diffusivity in MOFs lies in areas where experiments for transport property of interest (transport diffusivities, energy barrier to diffusion) are challenging, not in reiterating properties that have already been addressed experimentally. Measuring diffusivity at a wide range of loadings in the pores at extreme conditions such as infinite dilute loading and/or saturation loading is experimentally difficult. MD simulations can provide information about gas diffusion in MOFs' pores under these conditions. Getting diffusivity data as a function of gas loading is crucial to design membranes, adsorbents, catalysts from MOFs that will work under a wide range of operating conditions.

As addressed in Section 5, the development of quantitative information about mixture diffusion in MOFs is just beginning. Since performing mixture MD simulations for MOFs with large frameworks and for gas mixtures at high adsorbed loadings are computationally demanding, theoretical correlations that predict mixture diffusion based on single component diffusion data are very useful. Recent research search showed that these models yield accurate results for at least simple chemical mixtures in MOFs. Testing and validation of theoretical correlations for predicting gas diffusivity in various subclasses of MOFs will be useful to widely utilize these correlations for different structures.

A great advantage of using MD simulations is to test hypothetical MOF structures for particular applications if the metric describing the performance of a material for the application can be directly calculated. For example, Düren and coworkers used GCMC simulations to design materials with large adsorption capacities for CH4.(Düren et al., 2004) In a similar way, MD simulations can be used to design materials with slow diffusivities for CH4 and fast diffusivities for CO2 to identify materials that will be promising in kinetic separation of CO2 from CO2/CH4 mixtures.

### **6.2 Challenges**

The development of accurate classical interatomic potentials for describing gas diffusion in MOFs remains challenging. From the modeling perspective, it is important to use experimental diffusion data from a broad range of conditions to parameterize interatomic potentials whenever this is practical. However, as discussed in Section 4, the number of experimental data on gas diffusion in MOFs is very limited. Furthermore, developing potentials specific to a MOF structure is not the solution since hundreds of different MOF structures are available. Therefore, efforts to test and improve the transferability of potentials among related families of MOFs will have a great value. One of the major challenges in using MD simulations for MOFs was addressed in Section 2.2: absence of fully flexible force fields. Rigid framework assumption creates tremendous savings in computational effort. A handful of studies used flexible force fields to include the lattice dynamics effects on gas diffusivity in MOFs. These studies showed that there can be orders of magnitude difference between the diffusivity data from MD simulations using rigid framework and the one using flexible framework, specifically for large adsorbates. This issue indeed turns to be related with the challenge listed above, having accurate flexible interatomic potentials which can be applied to a family of MOF structures in a computationally meaningful time scale.

Another major challenge, especially in diffusivity simulations of CO2 and N2, is the choice of method to assign partial charges to MOF atoms. The QM calculations were used to define partial point charges in the literature, however there is no unique way to accomplish this task and different charge decomposition methods can give rather different results. Studies have shown that charge effects are important especially for computing diffusivities at low loadings. Careful studies that establish reliable approaches in charge assignment will be very useful in employing MD simulations for diffusion of polar and quadrupolar molecules in MOFs.

To date MD simulations have been used to compute the transport rates of adsorbates in MOFs. One remaining challenge is to predict the long term stability of MOFs since this is a serious issue in practical applications of these materials. Although stability issue sounds to be most likely addressed by experimental studies, one recent MD study which investigated the mechanism of water induced decomposition of IRMOF-1(Greathouse&Allendorf, 2006) showed that molecular simulations can be also helpful in this area.
