Johnson Matthey’s (JM) vision is for a world that is cleaner and healthier, today, and for future generations. As global leader in sustainable technologies, we apply science to catalyse the net zero transition for our customers. Through science we enhance life for millions of people, all over the world. We are making it our business to help address the four essential transitions: driving down transport emissions, transforming our energy emissions, decarbonising chemicals production, and creating a truly circular economy. If the transport transition is all about moving people and goods while lowering emissions, the energy transition is about finding sustainable ways to power the world. Hydrogen has a huge role to play because when it is used as fuel, the only by-product is water. Low-carbon hydrogen is essential for a viable hydrogen economy.

Because chemical and materials design for real world applications are a multiscale problem, in JM we have experts in different length scales (from Å to m) and we use state-of-the-art methods, such as high-throughput calculations and machine learning (ML) techniques. We perform 1) Nanoscale: Electronic and atomic modelling, 2) Micro/Mesoscale: Kinetic data and modelling, pore-scale modelling, and 3) Macroscale: Reactor scale modelling (e.g., CFD, process modelling). Theoretical work together with experimental measurements guide the development of real-world catalysts, driving innovation and progress in sustainable technologies. Several use cases will be demonstrated coming from our four business areas: Clean Air (CA), Platinum Group Metal Services (PGMS), Catalyst Technologies (CT) and Hydrogen Technologies (HT) where theoretical and computational tools are used to understand complex materials and guide experiment.

A fundamental study employing Density Functional Theory has been performed to study the deactivation of the industrially relevant Cu/ZnO/Al₂O₃ catalysts for methanol synthesis. It has been shown that the catalyst is impacted strongly by the behaviour of the zinc oxide (ZnO) component, particularly with regards to sintering. Although the copper component also sinters, the overall deactivation observed is largely driven by changes in the ZnO moieties.[1]

The addition of small amounts of a silicon-based promoter[2] has previously been identified as a promising low-cost, readily available, and non-toxic additive that slows down the rate of deactivation.[3] The presence of silicon improves the aging characteristics and slows down the sintering of ZnO. Improving and slowing down the sintering of the standard methanol catalyst is of great industrial relevance. This is particularly important with regards to sustainable methanol production from CO₂, due to the significantly higher quantities of by-product water produced compared to conventional methanol synthesis conditions. Therefore, gaining fundamental understanding by looking at the stability promoter role of Si is important.

In this study, two main questions were addressed: 1) What is the effect of Si in both bulk and surface? and 2) How does water interact with bare ZnO and Si_xZn_y-2xO_y (x=1,2,3)? Finally, the results will be compared to experimental data.

The green transition requires discovery and development of new catalyst materials for sustainable production of chemicals and fuels. However, it is difficult to predict a material, which might have a high catalytic activity for a given reaction, therefore the development of catalysts up until now has been driven mainly by trial and error. It would increase the pace of development, if we could predict a range of promising materials or if we at least could understand the limitations of catalysis. In this context high entropy alloys offer a chemical space of possible materials where the composition can be smoothly varied and where the properties also might vary in a seamless manner. This is good news for catalysis as such a smooth space is easier to explore to determine the interesting regions in composition space. Furthermore, the highly heterogeneous nature of a high entropy alloy surface reveals fundamental effects which are important for chemistry on surfaces in general, but are overlooked in the classic mean field view on catalysis. 

Artificial nitrogen fixation is essential to provide food security. Today ammonia is produced through the Haber-Bosch process from nitrogen and hydrogen gases whereas nitrates are produced through the Ostwald process from ammonia and oxygen. From these, various nitrogen containing fertilizers are produced depending on the application of use. The Haber-Bosch process is, however, not sustainable since it relies on natural gas resources for hydrogen generation, and at the same time highly polluting of CO2 emission. It is therefore necessary to develop alternative routes for ammonia and nitrate synthesis. One of the most attractive solution would be to have a heterogenous electrolytic cell with an aqueous electrolyte that works at ambient conditions, where a nitrogen fertilizer can be produced on-site. There are, however, several factors that make it difficult to accomplish this, mainly because the N2 molecule is inert and difficult to reduce and because of the side reaction, the hydrogen evolution reaction (HER), which usually takes place more easily than the nitrogen reduction reaction (NRR). It has been predicted that all the transition metals will much more easily catalyze HER than NRR.
 
Over the last few years we have been searching, using density functional theory (DFT) calculations, for alternative materials that can catalyze NRR while suppressing HER. The class of materials we have investigated are transition metal ceramics of e.g. nitrides, oxides, sulfides, carbides, oxynitrides and carbonitrides. Several promising candidates are predicted within each class of materials and we have tested several of them experimentally. There, we grow the catalysts in thin-films using magnetron sputtering, which are then tested in a micro reactor for electrocatalytic performance. The electrochemical micro reactor is connected in-line with the ammonia detection unit, preventing any possible contamination which makes the results reliable and robust. Experiments are done both in N2 saturated electrolyte and in Ar saturated electrolyte and isotope labelled 15N2 is used to proof catalysis. In this presentation, I will discuss both the theoretical predictions and the experimental performance of several candidates for NRR.

Electrocatalytic systems are crucial in various renewable energy conversion and storage technologies, forming a foundational basis for our sustainable future. Realizing their full potential requires advancements e.g., in catalytic materials to achieve better catalytic efficiencies, higher stability, and lower costs. This necessitates an atomic-level understanding of electrocatalytic systems, particularly the complex electrocatalyst-electrolyte interface, which involves numerous components and processes. Moreover, the interface properties can vary substantially depending e.g. on solvent and electrode potential and the variations can, in turn, have direct impact on electrocatalytic behaviour. The theoretical and computational methods are pivotal, as they can offer atomic level insight into interface chemistry even under realistic reaction conditions, but this calls constant development of methods and approaches.

The grand-canonical ensemble (GCE) DFT calculations [1] offer a robust framework for modelling electrochemical interfaces and reactions at the atomic level, while maintaining fixed electrode potentials. In my presentation, I will cover our recent developments in GCE-DFT [2], which make the method applicable to systems beyond the reach of the standard GCE-DFT approach. The examples of GCE-DFT calculations to be presented include computing Pourbaix diagrams for metals under realistic reaction conditions [3,4], demonstrating how pH and potential can strongly influence the state of the catalyst, and N₂ reduction to ammonia on graphene-based material, highlighting the potential dependency of reaction thermodynamics and kinetics and the role of explicit water molecules in these calculations [5]. Finally, the advantages and limitations of this method will be discussed and compared to standard DFT calculations.

Electrochemistry holds the promise to be a cornerstone for the sustainable production of fuels and chemicals. However, these catalytic reactions are increasingly complex to understand and hereby also improve. In particular, reactions that suffer from selectivity challenges such as CO₂ reduction

In this talk, I will discuss how experiments and computational simulations can support each other. I will focus on electrochemical reduction of NO_x, CO₂, N_2, and the combinations. Importantly, all these reactions share a direct competition with hydrogen, and furthermore, several products are formed from each reactant of these reactants. I will give minimalistic models that do not overfit or over-interpretate experimental data. Examples are:

- Electrochemical CO₂ reduction show multiple different products depending on metal catalyst [1]. I will show why copper is unique as catalyst with a multiple-carbon product distribution [2]. Following I will discuss data analytics on copper facets to steer the product distribution [3].

- Electrochemical NO_x reduction, also multiple products are formed; N₂O, N₂ and NH₃[4]. Several catalyst enable reductions to ammonia amongst them copper [5] and recently also Co and Fe based catalysts close to their reduction potential [6].

- Electrochemical N₂ reduction to ammonia (NH₃) at ambient conditions is burgeoning [7-8]. Most interesting in aqueous there is not a “copper” catalyst [9]. While in non-aqueous, the univocally working system is a Li-mediate system [10]. For this system, I will show that varying multiple experimental parameters display similar performance characteristics [11] and I will discuss systems beyond lithium.

Finally, I will discuss how one can use these insights to establish predictive schemes for products beyond the typical reduction reaction products, hereunder synthesis of urea [12].

[1] Hori et. al., Journal of the Chemical Society, Faraday Transactions, 1989, 85, 2309-2326.

[2] Bagger, Ju, …, Strasser, Rossmeisl, ChemPhysChem, 2017, 18, 3266–3273.

[3] Bagger, Ju, …, Strasser, Rossmeisl, ACS Catalysis, 2019, 9, 7894−7899

[4] Rosca, Duca, Groot, Koper, Chem. Rev. 2009, 109, 2209–2244

[5] Wan, Bagger, Rossmeisl, Angewandte Chemie, 2021, 133 (40), 22137-22143

[6] Carvalho, …, Stoerzinger, JACS 2022, 144, 14809 14818.

[7] Andersen et al. Nature, 2019, 570, 504-508.

[8] Lazouski et al, Nature Catalysis, 2020, 3, 2520-1158.

[9] Bagger, Wan, Stephens, Rossmeisl, ACS Catalysis, 2021, 11 (11), 6596-6601

[10] Tsuneto et al. Journal of Electroanalytical Chemistry (1994)

[11] Spry, Westhead, Tort, Titirici, Stephens, Bagger, ACS Energy Letters, 2023, 8 (2), 1230-1235

[12] Wan, Wang, Tan, Filippi, Strasser, Rossmeisl, Bagger, ACS Catalysis, 2023, 13, 1926-1933

Understanding selectivity trends is a crucial hurdle in the developing innovative catalysts for generating hydrogen peroxide through the two-electron oxygen reduction reaction (2e-ORR). The adsorption free energy of O* and OOH* intermediate and the degree of O2's adsorption to the surface and have been suggested as selectivity descriptors for 2e-ORR.[1] These approaches have been the main guide for understanding and predicting selectivity for 2e-ORR catalysts over the past decade. Yet, none of them has yielded an appropriate selectivity descriptor capable of quantifying selectivity, thereby serving as a metric for describing trends in selectivity. To resolve this issue, we identify a thermodynamically derived selectivity parameter (ΔΔG) based on computational hydrogen electrode (CHE) model [2] which allows to quantify selectivity using predicted adsorption free energies of ORR intermediates (OOH* and O*) and free energy of H2O2 (Figure 1). [3] We validate the efficacy of this parameter, across a wide spectrum of reported binary alloys [4] and demonstrate that only a small number of binary alloys with a single active site that have been reported to have high activity are selective for 2e-ORR. [5] These findings highlight the potential of ΔΔG as a selectivity parameter for identifying high-performance 2e-ORR catalysts. It also demonstrates the significance of concurrently considering both selectivity and activity trends. This holistic approach is crucial for obtaining a comprehensive understanding in the identification of high-performance catalyst materials for optimal efficiency in various applications.

The two-electron oxygen reduction reaction (2e^- ORR) is an appealing alternative to produce hydrogen peroxide (H₂O₂) for isolated communities, where water treatment infrastructure is rudimentary or non-existent [1]. Notwithstanding, to efficiently carry the 2e^- ORR, stable and selective electrocatalysts are needed that circumvent the complete reduction to water (4e^- ORR). As pure noble metals and their alloys generally display the best performance, affordable and active materials are intensively sought after [2,3].

Computational models to design ORR electrocatalysts extensively rely on DFT-calculated adsorption energies of key intermediates, such as *OOH and *OH. To avoid the ill-defined energy of O₂, water is used as reference, which is suitable for the 4e- ORR. However, when applied to the 2e^- ORR, it is often overlooked the fact that H₂O₂ is seriously misdescribed by density functional theory calculations, potentially harming the conclusions of widely used routines to design improved electrocatalysts [4,5].

In this presentation, I will show that the DFT energies of O₂ and H₂O₂ entail large errors for several exchange-correlation functionals and that these errors are correlated. I will also explain how this prevents the calculation of accurate equilibrium potentials and distorts the free-energy diagram of the ideal catalyst, even when water is used as a reference. Finally, I will detail how Sabatier-type plots are affected when incorrect energies of molecules are used for the 2e^- ORR, emphasizing that experimental trends of real materials are matched when both O₂ and H₂O₂ energy are rectified.

References

[1]        S. Brueske, C. Kramer, A. Fisher, Bandwidth Study on Energy Use and Potential Energy Saving Opportunities in U.S. Chemical Manufacturing, (2015). https://www.osti.gov/biblio/1248749.

[2]        S.C. Perry, D. Pangotra, L. Vieira, L.-I. Csepei, V. Sieber, L. Wang, C. Ponce De León, F.C. Walsh, Electrochemical synthesis of hydrogen peroxide from water and oxygen, Nat Rev Chem 3 (2019) 442–458. https://doi.org/10.1038/s41570-019-0110-6.

[3]        S. Siahrostami, A. Verdaguer-Casadevall, M. Karamad, D. Deiana, P. Malacrida, B. Wickman, M. Escudero-Escribano, E.A. Paoli, R. Frydendal, T.W. Hansen, I. Chorkendorff, I.E.L. Stephens, J. Rossmeisl, Enabling direct H2O2 production through rational electrocatalyst design, Nature Mater 12 (2013) 1137–1143. https://doi.org/10.1038/nmat3795.

[4]        M.O. Almeida, M.J. Kolb, M.R.V. Lanza, F. Illas, F. Calle‐Vallejo, Gas‐Phase Errors Affect DFT‐Based Electrocatalysis Models of Oxygen Reduction to Hydrogen Peroxide, ChemElectroChem 9 (2022) e20220021 (1-7). https://doi.org/10.1002/celc.202200210.

[5]        R. Urrego-Ortiz, M. Almeida, F. Calle-Vallejo, Error awareness in the volcano plots of oxygen electroreduction to hydrogen peroxide, ChemSusChem (2024) e202400873. https://doi.org/10.1002/cssc.202400873.

Improving the efficiency of electro-catalysts for the Oxygen Evolution Reaction (OER) is key for the energy transition. RuO2 and IrO2 are considered the gold standards for OER. In recent years, it has been suggested that the OER mechanism on these oxides could involve the formation of unconventional intermediates, -O-H and -OO-H, formed by the direct interaction of -O and -OO species on a coordinatively unsaturated metal site and a proton bound to a surface oxygen [1,2,3]. These species are competitive with the classical -OH and -OOH OER intermediates. Solvation is a key ingredient to describe interfacial electrochemical processes and can affect the relative stability of these intermediates. Here we present a comparative study on the nature of key intermediates of OER on TiO2, RuO2, and IrO2 (110) surfaces, by means of Density Functional Theory (DFT) calculations in conjunction with Ab-Initio Molecular Dynamics (AIMD). We first rationalize the nature of the species and the relative stability trends in vacuum. Then, we discuss the effect of including water solvation by means of static solvation schemes. The results indicate that -OO-H is preferred in place of -OOH for all surfaces considered, and -OH is preferred over -O-H except for RuO2. Finally, we investigated the nature of the catalyst/water interfaces as well as the interaction of intermediates with liquid water based on AIMD. On RuO2 -OH and -O-H display a very different interaction with water, resulting in distinct hydrogen bond networks [4]. Interestingly, -OO-H is quite rigid on RuO2, while it has a dynamic behavior on IrO2 as the proton is shared between the -OO species and a surface oxygen atom. This study provides critical insights on the role of solvation to the nature of key intermediates of OER, a key aspect for providing a fundamental understanding of OER on catalytic surfaces.

Single atom catalysts (SACs) have emerged as a new class for the development of active and selective catalysts. These materials are commonly based on anchoring a noble transition metal to some kind of carrier. In the present lecture, we demonstrate that MXenes — two-dimensional materials with application in energy storage and conversion — spontaneously form SAC sites under anodic polarization conditions, using the applied electrode potential as a probe to transform the surface into an SAC-type structure. Combining ab initio molecular dynamics simulations and electronic structure calculations in the density functional theory framework, we demonstrate that only the SAC sites rather than the basal planes of MXenes are highly active and selective for the oxygen evolution or chlorine evolution reactions, respectively. Our findings may simplify synthetic routes toward the formation of active and selective SAC sites and could pave the way for the development of smart materials by incorporating fundamental principles from nature into materials discovery: while the pristine form of the material is inactive, the application of an electrode potential activates the material by the formation of active and selective single-atom sites.

A key strategy for decarbonising our economy is electrifying industrial syntheses using 'power-to-X' technologies.[1,2] These methods, marrying electrocatalysis with renewable energy sources such as solar or wind power, show great promise for generating platform chemicals and high-value products. For instance, the increasing demand for Nylon-6 has intensified the need for greener routes to synthesize cyclohexanone oxime, a precursor to this polymer. Current methods rely on hydroxylamine derived from environmentally demanding processes, often involving harsh conditions, acidic or alkaline environments, and hazardous reagents. To address these issues, we developed a novel one-pot electrochemical synthesis of cyclohexanone oxime using aqueous nitrate as the nitrogen source under ambient conditions.[3]

Our approach utilizes Zn-Cu alloy catalysts to drive nitrate electroreduction, forming a hydroxylamine (*NH₂OH) intermediate that subsequently reacts with cyclohexanone in the electrolyte to yield the oxime. Optimal performance was achieved with a Zn₉₃Cu₇alloy, which reached a 97% yield and a 27% Faradaic efficiency at 100 mA/cm². Mechanistic insights from in situ Raman spectroscopy and density functional theory (DFT) calculations highlighted the importance of binding strength of key reaction intermediates in controlling product selectivity within the electrochemical-chemical (EChem-Chem) tandem process. Specifically, weak surface adsorption (e.g., pure Zn) requires higher potentials for the EChem step, whereas stronger binding (e.g., pure Cu) facilitates this process at lower potentials but leads to the complete reduction of *NH₂OH to NH₃ rather than oxime formation.

This presentation will focus on our computational DFT studies, which provided detailed insights into the resting states of the electrocatalysts, the reaction mechanisms, and how surface interactions dictate catalytic activity and selectivity. Overall, this work introduces a sustainable pathway for organonitrogen synthesis via electrochemical nitrate reduction, demonstrating how tuning surface properties enables selective production. These findings are expected to inspire novel approaches for nitrate utilization and the development of other EChem-Chem processes for environmentally friendly organic synthesis.

Understanding molecular adsorption behavior on electrode surfaces is crucial for optimizing electrocatalytic processes. Since electrode reactivity is significantly influenced by varying experimental conditions, there is a pressing need for adaptable theoretical frameworks that provide in-depth insights into these phenomena. Modeling surface coverage offers a good balance between accuracy and computational cost in capturing the complexity of the solid-liquid interface [1,2].

In this oral presentation, I will introduce an automated workflow developed in Python for the high-throughput analysis of key molecular adsorbates—specifically hydrogen, oxygen, and hydroxide—that may be present during electrochemical oxidation and reduction reactions in aqueous electrolytes. This approach enables predictive assessments of the resting state of metallic electrode surfaces at different applied potentials and pH values by leveraging the Computational Hydrogen Electrode (CHE) method, which is essential prior to any reactivity study.

This framework is driven by a machine learning calculator trained on DFT data, generated using the Cluster Expansion method [3]. It can predict the surface coverages of pristine Cu, Ag, Au, Ni, and Pt metal surfaces [4]. Future work will explore mixed coverages and extend the model to different surface facets and metal alloys. The model's extrapolating capabilities, powered by a representative featurization of the systems, should allow for the screening of the electrocatalyst resting state across the entire composition range of alloying elements. If the predictions prove accurate, this automated scheme could serve as a rapid Pourbaix diagram generator for any metallic electrode surface.

Functional materials are important for applications in the fields of catalysis and renewable energy. Specific functionalities include charge transport through electronic material components as well as catalytic reactivity on material surfaces. In the talk, I will advocate that there is a relation between charge transport efficiency and reactivity, and therefore developing novel algorithms that calculate both are important for better understanding of intrinsic material limitations. We cover our latest results in developing and using charge transport calculation methods and demonstrate them on catalytic materials. Our home code is developed on a user-friendly GUI and enables to use widely available Density Functional Theory results as input. The methodology is demonstrated on several generic materials such Fe2O3 on top of graphene, a well-studied water oxidation catalyst. We have calculated the cumulative probability of a charge to reach hematite's surface using a wave propagation simulator. Graphene supported hematite has higher cumulative probability of charge transfer than bare hematite. Graphene supported hematite having carbon vacancies in graphene shows higher cumulative probability than its pristine counterpart. These indicatives for improved carrier transport and catalysis are beneficial for water splitting. 

The electrochemical reduction of CO₂ into industrially valuable chemicals like ethylene and ethanol is one of the most promising strategies to mitigate the greenhouse gas effect and tackle global energy challenges. Currently, the production of these crucial C₂ molecules is only achieved using Cu-based electrocatalysts. However, obtaining high selectivity and efficiency towards specific reduction products remains challenging due to the complex reaction pathways involved. In this regard, the adsorption of CO onto the catalyst surface represents a crucial step since it corresponds to a key intermediate for the formation of C-C bonds.

In this work, we first employ machine learning models (ML) to predict the adsorption energies of CO on Cu/M alloy surfaces, where M indicates different types of metal atom impurities. These ML models, trained on a dataset of adsorption energies calculated via Density Functional Theory (DFT) calculations, help us understand the interaction of CO with various alloy compositions and identify potential candidates for high-performance catalysts. Secondly, we investigate the kinetics and thermodynamics of CO dimerization on these alloy surfaces using DFT simulations at constant potential, by including an explicit water layer wetting the catalyst surface. Our results enable the identification of scaling relations for this critical reaction step by varying the metal species in the alloy, which in turn facilitates the rational design of more efficient and selective copper-based catalysts.

The structure–activity relationship is a cornerstone topic in catalysis, which lays the foundation for the design and functionalization of catalytic materials. Of particular interest is the catalysis of the hydrogen evolution reaction (HER) by palladium (Pd), which is envisioned to play a major role in realizing a hydrogen-based economy. Designing Pd-based catalysts with optimal activity and selectivity relies on a thorough understanding of the surface structure under reaction conditions. Herein, first-principles density functional theory calculations are employed to investigate the stability of Pd-hydride/Pd interfaces under electrochemical conditions and the effect of Pd-hydride formation on the HER activity. Based on calculated Pourbaix diagrams, we can identify the relevant regions close to the equilibrium electrode potential and pH for the HER, where the Pd surfaces start to be covered by hydrogen adatoms, and when the electrode potential is decreased, there are clear thermodynamic indications for more and more subsurface hydride layers [1-2]. The formation of subsurface hydride results in a compressive strain that lowers the magnitude of the H adsorption free energy on Pd surfaces, thereby increasing the HER activities. Our results reveal an activity trend following Pd(111) > Pd(110) > Pd(100) and that the formation of subsurface hydride layers causes morphological changes and strain, which affect the activity for proton electroreduction and HER, as well as the nature of active sites [3]. Further, computational approaches to elucidate the reconstruction processes on these low-index Pd surfaces during proton electroreduction will be discussed and insights corroborated by experimental electrochemical scanning tunneling microscopy and on-line electrochemical inductively coupled plasma mass spectrometry during cyclic voltammetry. Reconstruction phenomena, creation of defects, phase transformations, and dislocations within the Pd subsurface layers upon the hydride formation, will be discussed in detail on the basis of molecular dynamics simulations. Summarizing, significant insights into the role of hydride formation on the structure–activity relations toward the design of efficient Pd-based nanocatalysts for HER.