To facilitate accelerated catalyst design, a combined computation and experimental workflow based on machine learning algorithms is proposed, which detects key performance-related descriptors in a CO2 to methanol reaction, for In2O3-based catalysts.
The environmental impact of unsustainable CO2 emissions calls for immediate action. One of the main methods for large‐scale reduction of CO2 emissions is conversion of carbon dioxide to valuable feedstocks like energy carriers or chemicals. Realization of this goal requires catalysts showing high‐performance characteristics under the relevant industrial conditions. In recent years, In2O3‐based catalysts have been discussed as target materials for conversion of CO2 to methanol in the presence of hydrogen. Optimization of this catalytic system via conventional routes requires massive resources in terms of extensive testing, synthesis, and characterization. In our study, we take an alternative approach and exploit high‐throughput computation to create a database for oxygen vacancy formation energy and explore a large number of candidates, which motivates the selection of a subset of materials for the synthesis of bulk and supported catalysts, to be tested in hydrogenation of CO2 to methanol. The method shows the impact, as results show improved performance for selected candidates, compared to the reference In2O3‐system. This confirms the potential of the selected descriptors as criteria for a new way of targeted design of alternative catalysts.
Ethanol is a high‐value chemical that can be produced from syngas derived from fossil or renewable resources. Silica‐supported catalysts based on Rh are reported to show the best performance so far. We report a comprehensive investigation for influence of catalyst composition on performance in syngas conversion to ethanol. We evaluate 53 different silica‐supported catalysts with a varying degree of Rh, Mn, Ir, Li and Ti, via high‐throughput testing. Correlation analysis reveals that Li and Ir tend to increase the selectivity to C2+ oxygenates, and catalysts with a 0.95 molar ratio of Rh : Mn show a tendency for high CO conversion. The composition 2.8Rh, 0.45Mn, 1.3Ir, 0.04Li, where the numbers indicate weight percentage, is identified to yield one of the highest C2+ oxygenate values reported in the literature so far. Moreover, combining this catalyst with a hydrogenation catalyst and converting C2+ oxygenates to ethanol, provides one of the highest ethanol yields reported so far in literature.
The direct conversion of synthesis gas could open up economically viable routes for the efficient production of ethanol. RhMn/SiO2 represents one of the most active systems reported thus far. Potential improvements were reported by added dopants, i.e., Ir, Ti, and Li. Yet, combining these elements leads to contradicting results, owing to the complexity of the interactions in a multi-promoted system. This complexity is often encountered in heterogeneous catalysis. We report a systematic data-driven approach for the assessment of complex multi-promoter interactions based on a combination of design-of-experiment, high-throughput experimentation, statistical analysis, and mechanistic assessment. We illustrate this approach for the system RhMn/SiO2 promoted with Ir, Li, and Ti. Using this approach, we investigate the impact of promoters’ interactions on a mechanistic level. Our analysis depicts the means to learn hidden correlations in the performance data and, additionally, high performance for ethanol yield for the RhMnIr/SiO2 catalyst. The method presented outlines an efficient way to also elucidate co-promoter interactions in other complex environments.
Converting CO2 with renewable hydrogen requires high-value products to be economically viable due to its inherent energy intensity and associated renewable energy costs. Direct hydrogenation of CO2 via exothermic reactions is appealing given volatile energy prices. Methanol is valuable due to its wide use in the chemical industry and as sustainable fuel component. Indium oxide-based catalysts have been actively researched for several years; however, the scarcity of indium and the need for platinum group metals drive the search for alternatives using more abundant and non-noble metals and metal oxides. Employing a protocol which unites Artificial intelligence boosted atomic scale modelling approaches, advanced synthesis, high-throughput catalytic testing and state of the art characterization, we have developed a Cu-Al-Ga system that produces methanol and dimethyl ether (in different ratios) at highly competitive performance characteristics. The elemental composition of the Cu-Al-Ga catalysts was optimized towards a low gallium content to favor economic feasibility. The final lead composition with around 1 wt.% Ga content showed high activity while the ratio between MeOH and DME can be precisely adjusted by the reaction conditions in terms of temperature and syngas composition. Even in CO2 rich syngas flows, the catalyst remains active and selective towards the primary products. In addition, we introduce an advanced simulation protocol for studying potential surface changes under reactive condition to shed light on potential dynamical active site formations and their impact on reaction mechanisms.