Functionalized SBA-15-based catalysts for energy-efficient CO2 desorption: Bridging experimentation and machine learning to enhance amine sorbents regeneration
Yingjie Niu, Shiying Zou, Haonan Liu, Minyue Hu, Jinjun Cai, Chao’en Li, Kathryn A. Mumford, Masood S. Alivand, Francesco Barzagli, Rui Zhang
Abstract
Improving the energy efficiency of CO 2 desorption remains a key challenge in post-combustion carbon capture using amine solvents. In this work, a series of solid acid catalysts based on sulfonated mesoporous SBA-15 and functionalized with phosphotungstic acid (HPW) were synthesized via post-synthetic modification and applied to the catalytic regeneration of monoethanolamine (MEA). The catalysts, featuring both Brønsted and Lewis acid sites, were thoroughly characterized and tested in desorption experiments using CO 2 -loaded MEA solutions. Among the synthesized materials, HPW-SBA-15-SO 3 H-1 exhibited the best performance, achieving significantly higher CO 2 desorption rate and cyclic capacity compared to the uncatalyzed system, while reducing the relative heat duty by up to 37.7 %. This catalyst also demonstrated good stability over 20 absorption–desorption cycles, maintaining its structural integrity. In addition, machine learning was employed to correlate the catalysts' physicochemical features with their desorption performance, highlighting the role of acidity and porosity and supporting the proposed mechanism by which acidic sites promote carbamate decomposition and MEAH + deprotonation. These findings underscore the potential of functionalized mesoporous silica as efficient solid catalysts for lowering the regeneration energy penalty in MEA-based carbon capture systems. • SBA-15 functionalized with HPW and SO 3 H to improve catalytic CO₂ desorption. • Up to 37.7 % energy reduction in MEA regeneration using HPW-SBA-15-SO 3 H-1. • HPW-SBA-15-SO₃H-1 maintained high desorption efficiency for 20 consecutive cycles. • SHAP analysis highlights Brønsted acidity and porosity as key performance drivers. • Mechanistic insights guide catalyst design through ML-structure correlation.