CO AND CO2 HYDROGENATION BY FISCHER TROPSCH SYNTHESIS PROMOTED WITH IRON BASED HETEROGENEOUS CATALYSIS: DATA COLLECTION AND PROCESS DESIGN

The reduction of CO₂ emissions into the atmosphere has become a global priority, given the crucial role of this greenhouse gas in climate change. A promising approach to mitigate CO₂ is its conversion into fuels and chemicals through processes such as Fischer–Tropsch (FT) synthesis. Traditionally, FT relies on syngas (CO/H₂) over transition-metal catalysts, but the direct hydrogenation of CO₂ has been less explored, particularly with Fe-based heterogeneous catalysts. Iron is an abundant and inexpensive metal, already known for its high activity in conventional FT from CO. This research aims to develop novel Fe-based catalysts effective in both CO and CO₂ derived FT, while also addressing the interconnected challenges of environmental sustainability, operational safety, and predictive process modelling. The ultimate goal is to provide a holistic solution for upgrading CO₂ into hydrocarbons, bridging existing gaps in the development of materials (e.g., Fe-based metallosilicates and hydrotalcites) that have been scarcely investigated to date. To this end, innovative iron-based catalysts were synthesized, tested under CO and CO₂ FT conditions, and evaluated through Life Cycle Assessment (LCA), HAZOP/LOPA safety analysis, and artificial intelligence tools (artificial neural networks) for performance optimization. This integrated approach provides a comprehensive perspective on CO/CO₂ conversion, representing an important step toward the practical implementation of Carbon Capture and Utilization (CCU) technologies. Two main classes of Fe-based heterogeneous catalysts were investigated: (1) mixed oxides derived from hydrotalcites (e.g. Mg-Fe-Cu), synthesized via ultrasound-assisted co-precipitation (US) and, in some cases, subjected to ball milling (BM), and (2) iron catalysts supported on mesoporous metallosilicates, synthesized by non-hydrolytic sol-gel routes. All catalysts were promoted with potassium and copper, both known to enhance activity and selectivity toward heavier FT hydrocarbons. Potassium acts as an electronic promoter, suppressing methane formation and favouring carbon-chain growth, while the Mg-Fe based mixed oxide matrix provides basicity, enhancing the Water Gas Shift (WGS) reaction and thereby benefiting CO₂ conversion. Catalytic testing in a fixed-bed reactor yielded highly promising results. With syngas (CO/H₂), the Fe-based catalysts reached high CO conversion levels (up to ~95–98% at 300 °C) and notable selectivity toward liquid hydrocarbons (C₅⁺ fraction, ~40–50 wt%), while maintaining low methane production (<5%). For instance, a hydrotalcite-derived Mg-Fe-Cu/K catalyst (denoted Fe30) achieved ~96% CO conversion at 300 °C, with ~40% selectivity to the heavy fraction and ~5% to methane. These performances are comparable or superior to those reported for reference systems: Fe30 matched the activity of Fe/K/Cu catalysts supported on conventional oxides (alumina), reaching >80% conversion at lower temperatures, and outperformed analogous catalysts on mesoporous silicates in terms of stability and selectivity toward heavy products. This confirms that the hydrotalcite-derived nanostructure provides excellent dispersion of active sites and a favourable chemical environment which promote FT reactivity while minimizing undesired byproducts. In direct CO₂ hydrogenation, the iron catalysts displayed significant activity, although the reaction pathway is strongly coupled with the reverse WGS reaction (CO₂ → CO), resulting in notable co-production of CO. Accordingly, CO₂ feeds yielded higher selectivity to CO and CH₄ compared with syngas operation. However, introducing cobalt into the catalyst formulation substantially improved liquid hydrocarbon yields: Mg-Fe-Co-Cu/K catalyst (denoted Co45, containing ~42% Co and ~30% Fe) produced a larger C₇⁺ fraction and fewer gaseous byproducts than pure Fe catalysts, particularly at higher temperatures. Thus, Co45 proved more efficient in directly hydrogenating CO₂ into long-chain hydrocarbons, with lower CO selectivity and reduced methane formation. This behaviour is attributed to cobalt’s intrinsic higher activity in FT from CO₂, which remains effective even at 350 °C, whereas iron tends to favour secondary methanation and Boudouard (carbon deposition) reactions under the same conditions. Overall, catalyst composition optimization (Fe ratio, Co incorporation, K promotion) was crucial: the Fe-Co catalyst provided the best balance, maximizing heavy hydrocarbon production while minimizing undesired gaseous emissions, and highlighted the trade-off between catalytic performance and the environmental cost of materials (given cobalt’s higher extraction impact relative to iron). The experimental dataset generated thus constitutes a valuable benchmark for both sustainability evaluation and predictive modeling (see chapter 5). Given the strongly exothermic and high-pressure nature of FT synthesis, safety considerations for the laboratory-scale pilot unit were systematically addressed using industrially derived methodologies. A Hazard and Operability (HAZOP) study was conducted on the FT unit, followed by a Layer of Protection Analysis (LOPA), with the aim of identifying potential hazard scenarios and verifying the adequacy of existing safeguards. This represents an innovative application of such frameworks in an academic setting, motivated by the frequent manual interventions and transient operations typical of laboratory plants. The HAZOP identified several critical deviations from normal operation, such as risks of H₂/CO gas leaks, line blockages, or overheating, particularly in the FT reactor and feeding system, which emerged as the most vulnerable nodes. In some cases, existing protections were found to be insufficient to mitigate the consequences of potential incidents. The subsequent LOPA quantified the level of additional protection required, identifying and prioritizing further independent protection layers. Key recommendations included: increasing automation (to reduce reliance on manual operations), implementing safety interlocks and automatic control loops (e.g., shutdown valves upon leak detection, pressure/temperature control loops), and adding redundancy to critical components. Passive measures were also emphasized, such as improved ventilation (exhaust hoods, fume extraction) and enhanced operator training for correct handling during operation and emergencies. The adoption of these strategies will minimize accident risks and unplanned downtimes, bringing the laboratory-scale setup closer to industrial safety standards. In conclusion, the combined HAZOP–LOPA approach in an academic environment proved effective in raising safety standards and in creating a replicable framework for managing operational risks in complex experimental projects like FT synthesis. To evaluate the overall sustainability of the process and developed materials, a Life Cycle Assessment (LCA) was carried out. The LCA covered both the production phase of catalysts (from raw material extraction through Fe-based catalyst synthesis) and the operational phase of CO₂ FT conversion into hydrocarbons. Three representative catalysts were compared: Fe30, Fe40, and Co45 (all Cu/K-promoted, with varying Fe/Co ratios) the same tested experimentally, assessing their environmental footprint under optimal operating conditions (350 °C) for the production of 1 kg of C₂⁺ hydrocarbons. The analysis was conducted with SimaPro software (CML-IA method, using IPCC GWP100 and CED indicators), evaluating impact categories such as Global Warming Potential (GWP), Cumulative Energy Demand (CED), and toxicity/ecotoxicity metrics. Results clearly indicated that catalytic efficiency strongly dictates environmental performance: the less efficient Fe40 catalyst generated the highest impacts across nearly all categories. For instance, Fe40 exhibited the largest carbon footprint (~2.39 kg CO₂-eq per kg of product), compared with ~1.31 kg for Fe30 and ~0.98 kg for Co45. In other words, for the same quantity of synthetic fuel, Fe40 contributed roughly double the greenhouse gas emissions of the cobalt-containing catalyst. Similarly, Fe40 showed higher impacts in other categories, e.g., ~67% greater abiotic depletion relative to Co45 and ~41% greater than Fe30. This disparity is due to Fe40’s lower catalytic efficiency, which demands higher energy inputs and produces more byproducts per unit of fuel synthesized. Conversely, Co45, despite incorporating cobalt, a critical and environmentally costly metal, compensated through its high activity in operation, leading to lower per-product impacts. In summary, the LCA highlighted a performance vs. environmental impact trade-off: improving catalytic activity (e.g., through Co promotion) can significantly reduce lifecycle impacts per product, even when catalyst production has higher initial burdens. These findings underscore the importance of integrating environmental assessment early in catalyst and process design. The Fe/Co combination emerged as the most promising pathway, suggesting directions for future development of FT catalysts that are both efficient and sustainable. In parallel with experiments, machine learning (ML) techniques were applied to model and predict FT process performance. Specifically, an Artificial Neural Network (ANN) was trained on the experimental dataset to correlate input variables (operating parameters such as temperature, pressure, GHSV, H₂/CO₂ ratio, and catalyst composition) with performance outputs (CO₂ conversion, CH₄ selectivity, C₂–C₄ fraction, C₅⁺ fraction, CO selectivity). The ANN architecture was optimized to minimize prediction error and capture nonlinear relationships, thus providing an alternative predictive tool to conventional kinetic models that require detailed mechanistic knowledge. ANN results showed good accuracy in predicting certain key aspects (e.g., CO₂ conversion and C₂-C₄ selectivity), consistent with trends reported in the literature. Sensitivity analysis confirmed the positive influence of higher Fe content and K promotion in increasing liquid hydrocarbon yields and conversion efficiency. However, limitations emerged: the ANN struggled to accurately predict heavy hydrocarbon (C₅⁺) formation and CO selectivity, likely due to the limited size and diversity of experimental data. This suggests that future models should incorporate larger datasets and potentially more advanced architectures (e.g., hybrid AI or deep learning). To complement the ANN, a Random Forest (RF) model was also applied on the same dataset. RF proved particularly robust with limited data, mitigating overfitting risks through ensemble averaging. While RF predictions were comparable in accuracy to ANN, it offered a distinct advantage in interpretability: feature-importance analysis identified which operating conditions and catalyst properties most strongly affected outcomes, aligning well with chemical intuition (e.g., temperature and catalyst composition as dominant factors). Thus, ML models demonstrated their potential as decision-support tools, enabling rapid virtual screening of catalyst/process combinations and accelerating the optimization cycle. In conclusion, this thesis develops a comprehensive framework for the sustainable conversion of CO₂ into fuels, combining innovation in catalyst design, process optimization, and integrated assessments of safety and environmental impacts. Results demonstrate that Fe-based catalysts, suitably promoted, can achieve high CO₂ FT performance, while Co incorporation further enhances liquid fuel yields and reduces per-product impacts. In parallel, rigorous HAZOP/LOPA studies ensure that such processes can be conducted safely even at laboratory scale, and LCA provides quantitative guidance to steer technological choices toward genuinely sustainable solutions. Finally, the application of AI-based predictive tools offers new perspectives for accelerating catalyst and process development, uncovering hidden correlations, and directing future experimental efforts more effectively. The synergy of these disciplines: industrial chemistry, safety engineering, environmental analysis, and data science, constitutes a holistic and highly innovative approach. This integrated vision facilitates the transition of CO₂-to-fuel FT technology from laboratory to industry, contributing to circular economy goals and decarbonization. Ultimately, the work lays the foundation for CO₂-to-fuels systems that are more efficient, safe, and sustainable, outlining strategies with tangible impact in the context of the ongoing energy and ecological transition.