By borrowing from the burgeoning field of evolutionary biology, scientists have developed computational techniques to predict the structure of nano-catalysts. Researchers at the Center for Atomic Level Catalyst Design (CALCD) demonstrated – for the first time – that the structure and location of metal nano-clusters confined to a porous scaffold can be reliably predicted by coupling a genetic algorithm with computer simulations. Accurately modeling these confined metal clusters is a significant step forward to guiding experimentalists in the synthesis of highly efficient catalysts for processes such as the sustainable production of fuels.
Tailored metal nano-catalysts, which atoms where? Catalysts are essential for processes such as the synthesis of liquid fuels from biomass and the production of hydrogen from water driven by sunlight. Catalysts often consist of metal clusters a few nanometers in size (that is, 3 million times smaller than a coffee bean) confined to porous supports. While catalyst discovery has relied traditionally on a trial-and-error approach, the rational design and synthesis of catalysts with atomic precision is considered a "holy grail" of science. Ideally, scientists would like to comprehend how atoms are arranged in metal clusters confined to porous frameworks and how reactant molecules will interact with them. The final aim is to identify the optimal cluster configurations that will selectively produce the desired products, e.g., clean fuels, to pursue their synthesis.
"Our calculations open the door to developing new catalytic materials in a more predictive way," said David Sholl, who led this research at CALCD.
Simulation of Metal Clusters Inside Porous Structures: Quick Search for The Best Trait: Metal clusters in real catalysts contain up to several thousand atoms. For a number of reactions, the optimum cluster size is roughly 200-300 atoms. Modeling metal clusters with more than 5-10 atoms, particularly when they are hosted in a porous structure, is beyond current computational methods. The group, led by Scholl, shows how to overcome these limitations. They have coupled their calculations to a genetic algorithm, enabling the reliable simulation of gold, palladium, and gold-palladium clusters in a porous framework. A genetic algorithm is a collection of mathematical operations that mimics the process of natural selection to determine which beneficial traits persist across generations in a population.
This tool allows researchers to automatically generate the large number of possible candidate structures for their metal clusters, and their location in the porous framework scaffold, and to quickly select the most feasible ones. "Starting with hundreds of nanocluster configurations means that we could draw firm conclusions about materials that would be simply too complex to be treated with more classical approaches," said Sholl.
Do Small Metal Clusters Stay Small? The tendency of metal clusters to aggregate and grow into larger clusters is pernicious in catalysis because the cluster growth limits the proportion of atoms available to catalyze the reaction and therefore the catalyst deactivates. Profiting from their innovative approach, Sholl's group at the CALCD has also studied the mobility of different metal clusters located inside the channels of the metal-organic framework.
"The genetic algorithm can deal with so much data that we have been able to understand not only the binding but also the diffusion of metal clusters inside the framework. This outcome was a wonderful bonus that we didn't anticipate at the outset of our work," indicates Sholl.
This approach has allowed the team to identify the candidates less prone to grow into larger, less efficient metal clusters.
