Hydrogen stands out as one of the most promising energy carriers for the future due to its availability and production flexibility. It is essential to develop technologies that utilize renewable feedstocks to enable more sustainable hydrogen production. In this context, the present work focuses on hydrogen production from ethanol using nickel catalysts supported on silica derived from biomass residues, such as rice husk and sugarcane bagasse. To enhance the basicity of the NiSi catalyst derived from rice husk, dolomite was added as a promoter to improve its activity. The catalysts were synthesized via wet impregnation and characterized by X-ray fluorescence (XRF), X-ray diffraction (XRD), N2 physisorption, temperature-programmed reduction (TPR), CO2 temperature-programmed desorption (TPD), thermogravimetric analysis (TGA), differential thermal analysis (DTA), and Raman spectroscopy. The silica-supported catalyst prepared from rice husk and promoted with dolomite exhibited higher ethanol conversion and hydrogen yield during the steam reforming reaction at temperatures between 400 and 600 °C, demonstrating the effect of basicity on catalytic activity. All catalysts showed good stability during 30 h of reaction at 500 °C, although higher coke formation was observed for the rice husk–derived catalyst promoted with dolomite.
In this present study, an Artificial Neural Network (ANN) model was developed to predict and optimize catalytic performance in the CO2-to-methanol hydrogenation process. A comprehensive dataset comprising 1045 experimental data points extracted from 60 scientific publications was used to train and test the neural network, covering a wide range of catalyst compositions and reaction conditions. A three-layer feedforward ANN (60 input neurons, 60 hidden neurons, 2 output neurons) achieved high predictive accuracy, with R2 values of 0.95 (training) and 0.81 (testing), confirming its ability to capture complex nonlinear relationships governing catalytic behavior. Furthermore, a sensitivity analysis using the developed model indicated that Temperature, Fe loading, and calcination temperature were the most important parameters affecting CO2 conversion. The model also revealed that catalysts based on Fe/ZrO2, promoted with Mn, Ce, Zn, or Y, and calcined below 500 °C exhibit optimal performance, maximizing methanol selectivity at temperatures above 250 °C and gas space hourly velocities between 2000 and 6000 mL·g−1·min−1. These results are consistent with experimental findings and demonstrate that ANN-based models can effectively guide catalyst design and process optimization by elucidating the key physicochemical factors driving CO2 conversion. Overall, this study highlights the potential of machine learning as a predictive and interpretive tool to accelerate the development of high-performance catalysts for sustainable methanol synthesis and carbon utilization.
Stacked catalyst bed configurations combining CoMo/Al2O3 and NiMo/Al2O3 represent a practical strategy for deep hydrotreating, yet their kinetic description remains challenging due to competitive adsorption, reaction pathway selectivity, and bed-sequence effects. This work presents a systematic comparative assessment of kinetic models for the simultaneous hydrodesulfurization of dibenzothiophene and hydrodenitrogenation of quinoline over NiMo/Al₂O₃ and CoMo/Al₂O₃ catalysts arranged in two stacked-bed configurations (NiMo+CoMo and CoMo+NiMo). Experiments were conducted in a continuous fixed-bed reactor at temperatures from 250 to 340 °C, weight hourly space velocities of 8–14 h−1, a hydrogen partial pressure of 60 bar, feed nitrogen concentrations of 150–450 mg kg−1, and a feed sulfur concentration of 3500 mg kg−1. Pseudo-first-order power-law (PPO, PPOP) and Langmuir–Hinshelwood-type models (LHIbase, LHIconstrained, and LHImodified) were evaluated through parameter estimation, chi-square tests, parameter significance analysis, and model discrimination via the corrected Akaike Information Criterion (AICc). Apparent activation energies for the direct desulfurization route ranged from 41 to 113 kJ mol−1 across the evaluated kinetic formulations, with values generally within the broad range reported for sulfided CoMo and NiMo catalysts. The modified Langmuir–Hinshelwood formulation (LHImodified) yielded nitrogen adsorption enthalpies of −72 and −175 kJ mol−1 for NiMo and CoMo, respectively, indicating a strong affinity of nitrogen-containing species for the catalytic sites. Beyond parameter fitting, all models were applied to simulate axial concentration profiles along the reactor, and their predictive capability was evaluated against independent validation data, including experimental estimates of inter-bed compositions. Model adequacy was found to depend on catalyst arrangement: PPO showed strong predictive capability for the NiMo + CoMo configuration, whereas LHIconstrained provided an accurate description of the CoMo + NiMo configuration. LHImodified provided a physically consistent and transferable description across both bed arrangements by explicitly representing nonlinear nitrogen inhibition while allowing catalyst-specific adsorption parameters. These results demonstrate that catalyst sequence and nitrogen inhibition must be considered explicitly when selecting kinetic formulations for stacked hydrotreating reactors and provide a quantitative basis for reactor modeling and catalyst-loading strategies in ultra-low-sulfur diesel production.
Ni, Co, and Fe catalysts promoted with lanthanum were synthesized via coprecipitation. In most cases, precursors with a layered double hydroxide (LDH)-type structure were obtained, which, upon calcination, yielded mixed oxides composed of nanometric particles with high homogeneity of the metallic species. The catalysts were evaluated in ammonia decomposition under atmospheric pressure. Among the materials evaluated, Ni-Al-La exhibited superior performance, achieving a hydrogen productivity of 34.2 mmol H2 gcat−1 min−1. This productivity is within the range of values reported for Ru-based catalysts under the same reaction temperature and GHSV. However, direct performance comparisons should be interpreted with caution, as other operating conditions and catalyst characteristics differ among studies. Furthermore, the Ni-Al-La catalyst maintained high stability over 36 h of continuous operation without signs of deactivation. The addition of lanthanum to the Ni-Al catalyst had a significant promoting effect, increasing NH3 conversion from 17.3% to 34.1% at 500 °C, nearly doubling its value. The superior performance of Ni-Al-La is associated with the intrinsic catalytic properties of Ni, combined with high metal loading, nanometric particle size, and homogeneous metal distribution derived from LDH-type precursors. The presence of La increases the catalyst reducibility and promotes a marked Ni surface enrichment, with the surface Ni concentration increasing from 4.7 at% for Ni-Al to 15.9 at% for Ni-Al-La, as revealed by XPS. These changes are consistent with the enhanced catalytic performance observed for Ni-Al-La.
