Optimal Dynamic Regimes for CO Oxidation Discovered by Reinforcement Learning

Metal nanoparticles are widely used as heterogeneous catalysts to activate adsorbed molecules and reduce the energy barrier of the reaction. Reaction product yield depends on the interplay between elementary processes: adsorption, activation, desorption, and reaction. These processes, in turn, depend on the inlet gas composition, temperature, and pressure. At a steady state, the active surface sites may be inaccessible due to adsorbed reagents. Periodic regime may thus improve the yield, but the appropriate period and waveform are not known in advance. Dynamic control should account for surface and atmospheric modifications and adjust reaction parameters according to the current state of the system and its history. In this work, we applied a reinforcement learning algorithm to control CO oxidation on a palladium catalyst. The policy gradient algorithm was trained in the theoretical environment, parametrized from experimental data. The algorithm learned to maximize the CO2 formation rate based on CO and O2 partial pressures for several successive time steps. Within a unified approach, we found optimal stationary, periodic, and nonperiodic regimes for different problem formulations and gained insight into why the dynamic regime can be preferential. In general, this work contributes to the task of popularizing the reinforcement learning approach in the field of catalytic science.

A transient flow reactor for rapid gas switching at atmospheric pressure

Herein, we present a design for a transient flow reactor system with high detection sensitivity and minimal dead volume, such that it is capable of sub-second switching of the gas stream flowing through a catalytic bed. We demonstrate the reactor's capabilities for step transient, pulse, and stream oscillation experiments using the model system of CO oxidation over Pd catalysts, and we find that we are able to precisely model step transients for CO oxidation using a pseudo-homogenous-packed bed reactor model. The design principles leading to minimal gas hold-up time and increased sensitivity that are described in this paper can be implemented into existing flow reactor designs with minimal cost, providing a readily accessible alternative to the existing transient instrumentation.

Comparative Evaluation of Light-Driven Catalysis: A Framework for Standardized Reporting of Data

Light-driven homogeneous and heterogeneous catalysis require a complex interplay between light absorption, charge separation, charge transfer, and catalytic turnover. Optical and irradiation parameters as well as reaction engineering aspects play major roles in controlling catalytic performance. This multitude of factors makes it difficult to objectively compare light-driven catalysts and provide an unbiased performance assessment. This Scientific Perspective highlights the importance of collecting and reporting experimental data in homogeneous and heterogeneous light-driven catalysis. A critical analysis of the benefits and limitations of the commonly used experimental indicators is provided. Data collection and reporting according to FAIR principles is discussed in the context of future automated data analysis. The authors propose a minimum dataset as a basis for unified collecting and reporting of experimental data in homogeneous and heterogeneous light-driven catalysis. The community is encouraged to support the future development of this parameter list through an open online repository.

The concept of active site in heterogeneous catalysis

Catalysis is at the core of chemistry and has been essential to make all the goods surrounding us, including fuels, coatings, plastics and other functional materials. In the near future, catalysis will also be an essential tool in making the shift from a fossil-fuel-based to a more renewable and circular society. To make this reality, we have to better understand the fundamental concept of the active site in catalysis. Here, we discuss the physical meaning - and deduce the validity and, therefore, usefulness - of some common approaches in heterogeneous catalysis, such as linking catalyst activity to a 'turnover frequency' and explaining catalytic performance in terms of 'structure sensitivity' or 'structure insensitivity'. Catalytic concepts from the fields of enzymatic and homogeneous catalysis are compared, ultimately realizing that the struggle that one encounters in defining the active site in most solid catalysts is likely the one we must overcome to reach our end goal: tailoring the precise functioning of the active sites with respect to many different parameters to satisfy our ever-growing needs. This article ends with an outlook of what may become feasible within the not-too-distant future with modern experimental and theoretical tools at hand.