Estimate of the catalytic activity of nanoparticles with Active Learning algorithms guided by DFT and DFTB

• Grant number: 25/10814-2
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: December 01, 2025
• End date: November 30, 2027
• Field of knowledge: Physical Sciences and Mathematics - Chemistry - Physical-Chemistry
• Principal Investigator: Ernesto Chaves Pereira de Souza
• Grantee: Walber Gonçalves Guimarães Júnior
• Host Institution: Centro de Ciências Exatas e de Tecnologia (CCET). Universidade Federal de São Carlos (UFSCAR). São Carlos , SP, Brazil
• Associated research grant: 21/12394-0 - Sustainable synthetic methods employing catalysis, benign solvents, safer reagents, and bio-renewable feedstock, AP.TEM

Abstract

Photoelectrocatalysis is a promising application technique, including environmental remediation, carbonic gas conversion, hydrogen production, and green chemical synthesis. Notably, the CO2 is a very stable molecule, making converting the greenhouse gas into other molecules challenging. Specifically, the catalysts can be used in chemical reactions to lower the amount of energy required to form or break chemical bonds and are used in the reverse water gas shift (RWGS) reaction to convert CO2 and hydrogen gas (H2) into CO and water (H2O) and posteriorly the CO reduction through the electrochemical CO2 reduction reaction (CO2RR), in which has been considered a promising carbon-neutrality system. Traditional catalysts in the RWGS reaction are made from precious metals, including platinum, palladium, and gold, limiting the reaction's cost efficiency. Because of this, new catalyst materials and formation methods are developed to increase the practicality of the RWGS reaction as a means of lowering atmospheric CO2 and generating syngas. Developing efficient novel catalysts requires optimizing many parameters, such as composition, size, shape, and operational conditions. Within a myriad of possible nanoparticles, the compounding alloys of the first series of transition metals have proven highly effective due to their high surface area and tunable electronic properties. However, experimental exploration of all possible combinations of nanoparticles and their operating conditions is impractical. Therefore, the purpose of this project is the use of Active Learning (AL) combined with Density Functional Theory with Periodic Boundary Conditions (DFT-PBC) and Density Functional Tight-Binding (DFTB) to efficiently guide the discovery and optimization of catalytic nanoparticles for the pristine systems of iron, nickel, and copper and their alloys. The search of main exposed surfaces, pondering the lower Miller indices into the chemical landscape, is performed by DFT calculations. Thereafter, the nanoparticle is built, considering the Wulff construction in the chemical space, and the simulations are carried out at the DFTB level. The order parameter most relevant to the catalyst efficiency estimate is the activation energy, calculated by the unity bond index-quadratic exponential potential (UBI-QEP). The pre-selected descriptors from DFT/DFTB simulations are re-fed in the AL pipeline. As an output, the best structure nanoparticle for catalytic CO2 conversion is presented, considering the size, composition, and shape of the nanoparticle.