Answers are grounded in this paper and cite their sources.
Chromatographic method development typically resolves peaks before calibration and quantification. Here, we introduce a quantification-guided closed-loop strategy that selects methods by experimentally demonstrated calibration performance rather than baseline separation. The workflow combines automated gas-chromatographic experimentation, simultaneous multicomponent exponentially modified Gaussian fitting, external- or internal-standard calibration, and constrained large-language-model recommendations. The model proposes bounded parameter changes, while deterministic software enforces operating limits, generates executable conditions, and applies predefined acceptance criteria. Across one external-standard and three internal-standard systems containing two to four analytes, selected methods achieved run times of 2.2−2.7 min and minimum analyte-specific R2 values of 0.994−0.999, respectively. In the five-component system, calibration slope RSDs were below 0.30% across five independently prepared calibration series. All 12 analyte-level predictions in three validation mixtures had absolute relative errors below 9%, and mean spike recoveries were 97.61−103.90%. These results indicate that baseline separation need not be universal when overlapping components remain identifiable, and quantitative performance is experimentally verified. More broadly, quantitative validity as a feedback signal moves chromatography from a passive measurement endpoint toward a bounded autonomous process for self-driving laboratories and compositionally bounded high-throughput workflows.
No takes yet. Share an insight, caveat, or question.
Ge et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: