(698e) First Principles Membrane Catalyst Co-Design for Low Temperature Ammonia Production

Ammonia production is highly centralized, but recent efforts to produce ammonia at a mass manufactured modular scale faces unique scaling challenges. One approach to modular manufacturing uses membranes with integrated catalysts, such as 2D zeolites, to separate product ammonia streams leading to process intensification. To match membrane conditions, we must design highly active catalysts that can operate under mild conditions (e.g. 300 C, 10 bar). Transition metal nanoparticles supported on metal nitrides (TM-MN) have been shown to enable synergistic mechanisms for ammonia synthesis where nitrogen adsorbs at nitrogen vacancies and is sequentially hydrogenated. This so-called associative mechanism avoids the nitrogen dissociation step and allows for catalytic reactivity at lower temperatures. There are many possible binary and ternary metal nitrides, and here we use a synergistic combination of Density Functional Theory (DFT), microkinetic modeling, and machine learning regression and classification to identify the mechanism and reaction rates for different classes of metal nitrides and screen new hypothetical nitrides. In addition to metal nitrides, we also consider single atom sites directly incorporated into the zeolite membrane for ammonia production.

We have generated a set of binary and ternary transition metal nitrides that significantly differ in their nitrogen formation energy and hydrogenation barriers. This includes nitride perovskites (ABN₃). Recent literature has developed descriptors for predicting transition metal oxide reactivity including the d-band width, electron affinity, as well as centers of the O_2p band that we are adapting to metal nitrides to predict kinetic barriers for ammonia production. Using these approaches, we are screening metal nitrides to develop data-driven descriptor-based models to predict hydrogenation barriers and vacancy formation energies. On a subset of the most kinetically interesting nitrides, we use microkinetic modeling based on the obtained DFT energies to determine turnover rates at relevant process temperatures.