Source Rock Py-GC-MS Data Deconvolution via an Open-Source Python GUI for MCR-ALS

Pyrolysis coupled to gas chromatography and mass spectrometry (Py-GC-MS) allows petroleum source rocks to be analyzed directly, without solvent extraction or sample pretreatment, providing information on thermodesorbed and pyrolytic products in one run. However, the resulting signals are highly complex and difficult to interpret. As compounds are released as a function of furnace temperature, severe co-elution and mass spectral overlap can occur, and a given region of the analytical thermogram may contain thermodesorbed hydrocarbons or pyrolysis products from kerogen or bitumen. Reliable interpretation therefore depends on resolving these overlapping contributions. Multivariate curve resolution with alternating least squares (MCR-ALS) can recover pure-component concentration and spectral profiles from multicomponent signals. Free graphical user interfaces (GUIs) were developed by Jaumot et al. and later extended as MCR-ALS GUI 2.0. However, their dependence on the proprietary MATLAB environment can be a barrier for laboratories with restricted licensing, and no open-source equivalent is currently available. A new open-source, general-purpose, Python-based GUI for MCR-ALS was developed and applied here to the processing of Py-GC-MS data. It reproduces all constraints available in MCR-ALS GUI 2.0 for single- and multiset-matrix analyses, while introducing new features to streamline practical workflows. The GUI workflow employed three components, selected by Singular Value Decomposition (SVD), and used a spectral input as initial estimate. AsLS baseline correction was applied to all sub-matrices, with non-negative constraint to both concentration and spectrum and unimodality for concentration. The local outlier factor (LOF) and coefficient of determination (R²) were used, with LOF convergence set at 0.1%. The processed data corresponds to Tremembé Formation source rock (Type I kerogen, Brazil) analyzed by Py-GC-MS at sequential pyrolyzer temperatures from 100 to 500 °C using a stepwise method. Because pyrolysis under an inert carrier gas produces characteristic alkenes that are not generated by thermodesorption alone, the co-elution of n-alkanes, n-alkenes, and same-carbon-number branched isomers contains information about the underlying thermal process, if these contributions are resolvable. Submatrices corresponding to specific carbon-number regions of the chromatogram were concatenated as temperature-dependent multisets and processed using the new GUI. The method resolved the contributions of n-alkanes, n-alkenes, and branched isomers, allowing the temperature-dependent emergence of olefins to be used as a chemometric marker to distinguish thermodesorption (≲300 °C, alkene-poor) from pyrolysis (≳350 °C, alkene-rich). Short-chain alkene formation was most pronounced between 250 and 350 °C, with a second significant contribution near 450 °C, whereas long-chain alkene formation occurred predominantly within the main pyrolysis window of 325–425 °C.