1 Introduction
Catalysis is playing an irreplaceable role in developing a sustainable future [
1]. Heterogeneous thermal catalysis currently serves as a critical facilitator in modern industrial processes, while electrocatalysis may play a prominent role in the future since it can be in principle powered by green electricity. Nevertheless, the understanding and design of a cost-effective catalyst represent a formidable challenge. The majority of successful catalyst discoveries have historically relied on a time- and resource-consuming trial-and-error process.
Back in the 1920s, the
Sabatier principle was proposed, which revolutionized how scientists evaluated the performance of a catalyst [
2]: A good heterogeneous catalyst should neither bind the adsorbate too strongly, nor too weakly, to reach the optimal activity (Fig.1). In the 2000s, Jens K. Nørskov [
3], a pioneer of modern catalysis theory development, successfully quantified and visualized the Sabatier principle by volcano activity models coupling the identified scaling relations and microkinetic modeling based upon density functional theory (DFT) calculations [
4–
6]. This type of models can depict the intrinsic activity optima and trends of a catalytic reaction as a function of the reaction descriptors, i.e., the bonding strengths of key reaction species, which can reduce the high-dimension catalytic problem into one- or two-dimension. Over the past two decades, tremendous follow-up works have corroborated that as long as it is judiciously derived (e.g., with rational computational methods and parameters, well-defined mechanism, and precise linear scaling relations), a catalytic volcano model can be precisely developed and can exhibit strong concordance with experimental measurements [
7–
16]. Therefore, when designing an active catalytic site, one can only provide the information about the bonding strength of the key adsorbate(s) over the site, and the model can predict the corresponding activity in the form of turnover frequency (TOF),
Gmax (i.e., the maximum free energy needs to be overcome along the favorable reaction pathway), current density, or overpotential [
10,
17,
18]. This offers a substantial advantage for researchers by obviating the need for a resource-intensive experimental trial-and-error process, thereby enabling a more streamlined approach to catalyst design [
1].
Nonetheless, a significant challenge arises in that many of these volcano activity models are derived via a complicated process, which could be a hard task for beginners in catalysis theory research and experimentalists. Although there are some available packages for microkinetic modeling, e.g., the CatMap package [
19], these tools often necessitate a degree of proficiency in programming languages (e.g., Python), thereby requiring users to undergo preliminary training prior to effective utilization. This significantly impedes the broad adoption of a systematic approach to catalyst design and understanding, in both pre- and post-experimental investigations.
Besides, because electrochemical conditions will induce water activation or generation (which would lead to the formation of HO*/O*/H* or anion vacancy on the surface), analyzing the electrochemistry-induced surface state (i.e., the coverage of an electrocatalyst under electrochemical conditions) is essential [
20]. In an electrochemical environment, various phenomena may occur, including the poisoning of reactive sites, alterations in the surface’s electronic structure, or the generation of new active sites. These transformations yield a mechanistic understanding that diverges significantly from the analysis of activity over a stoichiometric pristine surface. However, this is an often-dismissed process in electrocatalyst design and understanding. To provide a more precise understanding of an electrocatalyst, Surface Pourbaix Diagram analysis [
21,
22] is an important tool to theoretically probe the surface state under electrocatalytic conditions, before the analysis of the activity [
14,
23].
Motivated by the current stage, we developed a user-friendly online platform for volcano activity model development—the CatMath—to provide the catalyst community with the predictive volcano activity models of industrially significant thermal + electrocatalysis and electrocatalytic Surface Pourbaix Diagrams. All the model generation processes are based on cloud computing, and thus users are not required to implement a special package for model generation. This online platform is dynamically updated by the Hao Li Laboratory (the Digital Catalysis Laboratory, DigCat), which currently contains the volcano activity models of six thermal catalysis (four 1D and three 2D volcano models), and eight electrocatalysis (seven 1D and five 2D volcano models). Specifically, the thermal catalysis contained in the current version of CatMath includes CO oxidation, ethylene epoxidation, nitrite reduction, Hg oxidation, and NO oxidation, while the available electrocatalysis includes hydrogen evolution, oxygen evolution, oxygen reduction, CO2 reduction, water oxidation, and electrolytic propylene epoxidation. The scaling relations behind each model and the applied energy references of each binding energy can be found on the CatMath website. Details of the reaction models are listed in Tab.1.
2 Results and discussion
The illustration of the main interface of CatMath in the current version is shown in Fig.2. From the interface, users can use the “Surface Pourbaix Diagrams” module to generate a Surface Pourbaix Diagram based on their calculated energetic values (Fig.2(a)). The “Catalytic Volcano Models” module can be used to access the available catalytic reactions included in the up-to-date version (Fig.2(b)).
Some examples of the CatMath are also presented here as shown in Fig.3. For the Surface Pourbaix Diagrams module, we used ZnCr2O4 as a typical example (Fig.3(a)). In terms of the Catalytic Volcano Models, herein, we take the ethylene epoxidation and CO2 reduction reaction (CO2RR) as the examples of thermal catalysis and electrocatalysis, respectively (Fig.3(b) and Fig.3(c)). Users can use the built-in drop-down menu to choose the reaction, set the required and optional parameters (e.g., temperature, pressure, and applied potential), and submit requests to the cloud. After cloud computing, the volcano model can be generated (Fig.3(b) and Fig.3(c)), respectively for ethylene epoxidation and CO2RR. Meanwhile, for those reactions involving significant selectivity competition, their selectivity plots can also be displayed. A video showing how the CatMath works can be found in the Electronic Supplementary Material (video).
3 Conclusions
Herein, we have developed an online predictive platform through cloud computing, for analyzing thermal + electrocatalysis (namely, the CatMath). This platform is dynamically updated. We expect that this user-friendly tool can provide researchers with a convenient way to understand the reaction trend of a catalysis and make fast prediction on their materials of interest as a potential catalyst.
PDF
(1181KB)
