In order to reduce the current costs associated with carbon capture technologies, novel materials such as zeolites and metal-organic frameworks that are based on microporous networks are being studied. We have developed a GPU code that can characterize and screen a large database of zeolite structures and help identify the most e cient structures for carbon capture. The interactions between the atoms that constitute the zeolite structures and the gas molecules such as carbon dioxide and methane are described by the Lennard-Jones and Coulomb potentials. We have also developed a CPU algorithm that identifies inaccessible regions inside the zeolite structures based on their energy profiles. Putting all this together, we can compute the selectivity for separating carbon dioxide molecules from other flue gases in each of the zeolite structures.