Achieving a closed carbon cycle requires capturing unavoidable CO2 emissions and converting them into valuable chemicals. Ethanol is of particular interest as it is both a platform chemical and a sustainable fuel. Currently, only two complex, multi-step industrial processes produce ethanol from CO, which itself could be obtained from CO2 via the rWGS reaction. No industrially established processes exist for either direct CO2 to ethanol conversion or indirect conversion via CO. Thermodynamic analysis shows that converting pure CO2 feeds is not industrially viable: both direct and indirect routes yield low conversions and broad product distributions, including unconverted CO2 and H2 as well as CO, water, alcohols, and hydrocarbons. Methane is formed most selectively among hydrocarbons, yet its formation must be minimised because it serves as an H2 sink. Because CO2 and CO can be recycled, the catalyst feed becomes mixed CO/CO2 (COX). Although extensive literature exists for pure CO or pure CO2 feeds, mixed COX feeds are scarcely studied. This work, therefore, systematically investigates ethanol-synthesis catalysts under COX conditions, using pure CO and pure CO2 feeds as references. Additionally, new catalyst concepts are developed to improve the performance in COX feeds and suppress methane formation. Because CO to ethanol is the state-of-the-art in literature, selected catalysts from those studies were evaluated under comparable conditions and in COX feeds. Rh0.57Mn0.17Ir0.14Li0.12/SiO2 (2.69 wt% Rh) showed the highest ethanol selectivity (37.4%) but also a high methane selectivity (29.5%), with a stable performance in mixed feeds. The performance was strongly reduced in pure CO2 (5.8% ethanol). CoCuMnOx gave a high conversion but produced mainly methane. Cs-CZA performed better than CZA, reaching 4.2% ethanol selectivity, but the selectivity decreased in CO2 containing feeds. Overall, the rhodium catalyst remains the most selective, although a high methane selectivity and a high Rh loading limit its industrial potential. These challenges were tackled in a project with BASF and hte GmbH to develop rhodium catalysts with a lower rhodium content and reduced methane selectivity. Screening the RhMnLi composition range showed that high rhodium contents gave the best ethanol selectivity, with Rh0.33Mn0.33Li0.33/SiO2 (1.60 wt% Rh) as a lead candidate. Lithium suppressed conversion and methane selectivity, shifting products toward methanol. To improve the performance in CO2 containing feeds, promoters were screened using ML-guided QM calculations and high-throughput testing. The resulting Rh0.25Mn0.25Li0.25Mo0.25/SiO2 (1.00 wt% Rh) catalyst showed a favourable product spectrum in CO and remained stable in CO2. Thus, resulting in 7.4% ethanol, 23.3% methanol, and 22.3% methane selectivity as the best performance in pure CO2 in this work, combined with a reduction of the rhodium content by 63% compared to Rh0.57Mn0.17Ir0.14Li0.12/SiO2. FeCoCu catalysts were studied as a non-precious alternative to rhodium. Iron is essential for ethanol formation, and multimetallic systems perform better, though lowering the reaction temperature has a stronger effect. The trimetallic catalyst was stable over 585 h TOS in CO feeds but produced no ethanol from pure CO2. Strategies such as a lower calcination temperature (350 °C), Cu support on ZrO2, and Ge or Na promotion improved selectivity and enabled ethanol formation in some cases. Despite these advances, FeCoCu remains less selective than rhodium catalysts. This work showed that direct CO2 to ethanol conversion is less favourable than the indirect route via CO. Combining the rWGS and ethanol synthesis in one reactor is therefore a promising process solution, allowing heat integration while requiring identical operating conditions for both reactions. NiCu rWGS catalysts were tested under ethanol synthesis conditions, revealing that the lower temperature limits conversion, producing CO2-rich feeds. Copper was most selective but deactivated rapidly, while Ni stabilised the catalyst. Based on these findings, a tandem system combining the Ni25Cu75/Al2O3 rWGS catalyst with the Rh0.25Mn0.25Li0.25Mo0.25/SiO2 ethanol catalyst was successfully demonstrated. To accelerate catalyst development, literature data from 447 sources were used to test an ML-based workflow. An LGB model was trained on the sparse, heterogeneous dataset to identify key features for catalyst performance. For ethanol selectivity, the most relevant parameters were a rhodium content around 1 wt%, an electronegativity difference of 0.4–0.9 between elements, and low reaction temperatures. Similar analyses were carried out for higher-alcohol selectivity and ethanol STY. While the model successfully highlighted such trends and offers a complementary approach to literature analysis, it cannot perform detailed feature optimisation due to limited and inconsistent data. This demonstrates the need for higher-quality, standardised data and metadata. In summary, this work applied a broad literature-guided approach. A variety of catalysts was tested under unified reaction conditions, and their performance in mixed feeds was assessed. The results highlight that multimetallic systems offer advantages not achievable with simpler catalysts. The study also explored tandem catalysis with low-temperature rWGS as an improved process route. Overall, these insights move CO2 to ethanol conversion closer to industrial feasibility.
Franziska Thimm (Thu,) studied this question.