(521eo) Programmable Catalytic Ammonia Synthesis and Its Optimization

Recent experimental capabilities enable the introduction of precisely controlled perturbations in heterogeneous catalytic reactors e.g. in terms of electrodynamic potential [1] or dynamic strain [2]. Systematic application of such perturbations can be viewed as externally enforced “programs” of catalytic activity [3]. It is long known that such programs, when appropriately tuned, have the potential to improve catalytic performance orders of magnitude over the static (or unperturbed) optimum; this has already been demonstrated computationally and experimentally [1]. In this work we computationally discover optimal programs for ammonia synthesis on a ruthenium crystal under oscillating dynamic strain [2]. We use a detailed, high-dimensional, kinetic model which takes into account different types of catalytic sites (such as terrace, upper step, lower step). Using advanced tools from scientific computing (matrix-free time-stepper-based solvers) and active learning (Bayesian Optimization) we demonstrate efficient discovery of optimal dynamic strain programs for ammonia synthesis. Our algorithmic pipeline is specifically designed to deal with detailed kinetic models (high-dimensional state space) and programs of complex shape (high-dimensional parameter space). Furthermore, algorithms such as Bayesian Optimization can be extended to the laboratory and motivate efficient, data-driven real-world applications of dynamic catalysis.

References

[1] M.A. Ardagh, M. Shetty, A. Kuznetsov, Q. Zhang, P. Christopher, D.G. Vlachos, O.A. Abdelrahman, P.J. Dauenhauer, Chem. Sci., 2020,11, 3501-3510

[2] G.R. Wittreich, S. Liu, P.J. Dauenhauer, D.G. Vlachos, Sci. Adv. 2022, 8, eabl6576

[3] Y.M. Psarellis, M.E. Kavousanakis, P.J. Dauenhauer, I.G. Kevrekidis, Chemarxiv, 2023