A local predictive modelling approach for enhanced separation of critical volatile compounds with flow-modulated comprehensive two-dimensional gas chromatography

Abstract

In recent years, the application of flow-modulated comprehensive two-dimensional gas chromatography (FM-GC × GC) has significantly increased, particularly for profiling complex food volatilomes. However, the full potential of this technique is often hindered by the complexity of instrumental optimisation, which is critical for achieving high-resolution separation across a wide variety of analytes in diverse samples. This work addresses this challenge by developing a predictive strategy to optimise FM-GC × GC conditions for the separation of a wide variety of volatile compounds. The efficiency of a local model based on peak parameters was tested across four different column setups, monitoring its correlation to the separation of 22 critical coeluting pairs. Using a Doehlert experimental design and a General Linear Model (GLM) approach, robust predictive models (R² > 0.80) were established and validated. Although the predictive models are confined to the tested experimental space, the underlying strategy and observed general behaviours provide a transferable framework for other setups. The use of a multiresponse optimisation strategy coupled to a global resolution metric, the Fraction of Resolved Peaks (FRP), successfully improved the separation of coelutions, particularly for specific column configurations. The effectiveness of this workflow was ultimately demonstrated by its successful application to the complex volatilomes of wine, honey, cascara coffee tea, and masala tea.