AI-guided simulations rank two zeolites differently as CO₂ pressure changes
Source check: Source checked. Not independent scientific confirmation.
Finding: RaspAssistant compared three literature-based adsorption models against existing MFI zeolite measurements, selected the one with the lowest aggregate error and used it to simulate CO₂ adsorption in 197 pure-silica structures. JBW had the highest simulated uptake at 1 kPa, while GIS led at 10 and 100 kPa and had the highest calculated working capacity from 1 to 100 kPa. These are model-dependent rankings of known structures. The reference measurements were for MFI.
Technology: RaspAssistant; GPT-5.2; GPT-5.4
Family GPT-5: profile · models page
Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.
- AI contribution
- The GPT-5.2 supervisor and GPT-based specialists compared candidate force fields with reference measurements, selected the model for the larger screen, coordinated simulations and analyzed pressure-specific rankings.
- Test and result
- Three force-field combinations were compared against existing MFI adsorption isotherms at three temperatures. The selected model was used for 2,955 adsorption calculations on 197 structures. The outcome is relative simulated uptake and working capacity in the specified screening study.
- Limitation / still unknown
- Quantitative transfer from MFI to other frameworks was not established. The screened candidates were not retested experimentally in this case study, and reported descriptors could not identify a pore-scale mechanism. Prior work already considered GIS and JBW for carbon capture.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://arxiv.org/abs/2610.27572 · https://arxiv.org/html/2610.27572
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: linked separately.
Review date: 21 Nov 2026. Review scope: Original arXiv v1 abstract and submission metadata; HTML §§1, 2.4, 3–4 and Supplementary Table S5.
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 1/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.