Reducing the analysis time for contaminated crude oil datasets using pixel-based GC×GC-TOFMS and multivariate models

Contamination of crude-oil samples by drilling fluids undermines accurate geochemical characterization. Here, we combine comprehensive two‑dimensional gas chromatography coupled to time‑of‑flight mass spectrometry (GC×GC‑TOFMS) with a pixel‑based approach and statistical models to accelerate interpretation, in the context of contamination, of large datasets. Eighteen samples were analyzed: 15 reservoir crude oils collected during field development, one reference crude oil, and two simulated samples prepared by spiking the reference crude oil with representative synthetic drilling fluids (Br‑Mul - paraffinic; Ole - olefinic). Total ion chromatograms (TICs) from GC×GC‑TOFMS were imported to MATLAB and structured into an augmented data matrix (18 × 587,400), which was aligned using the icoshift algorithm, normalized, mean‑centered, and submitted to principal component analysis (PCA). PCA analysis (PC1 + PC2 = 53.56% cumulative variance) revealed clear groupings, with contaminated samples tending to cluster separately from uncontaminated ones. PC1 primarily discriminated contamination type (paraffinic versus olefinic) and contamination level, while PC2 captured overall contamination intensity and aided separation of mildly contaminated oils. Simulated samples (#S1, #S2) provided interpretative anchors: Br‑Mul‑spiked sample clustered with paraffinic‑influenced oils, and the Ole‑spiked sample clustered with olefinic‑influenced oils. The pixel‑reconstructed loading maps highlighted chromatographic “hotspots” responsible for sample grouping; back‑projection to original GC×GC‑TOFMS chromatograms enabled targeted peak‑by‑peak identification of diagnostic analytes via deconvoluted mass spectra and NIST library matching. This targeted strategy reduced the scope of manual interpretation while preserving structural and compositional resolution. Detailed GC×GC‑TOFMS chromatographic comparison identified paraffinic contamination signatures—enrichment in n‑alkanes with a maximum at n‑C₁₄ and in branched and cycloalkane regions—for oils aligned with Br‑Mul. Oils associated with Ole displayed pronounced enrichment in olefinic regions (mainly C_14, C₁₆ and C₁₈ linear alpha olefins and C₁₅/C₁₇ isomers) alongside contributions from n‑alkanes in the C₁₄–C₁₆ interval. Olefinic contamination was more clearly distinguished in the score plot than paraffinic contamination, likely due to compositional overlap between paraffinic drilling fluids and the native aliphatic fractions of crude oil, combined with the near-absence of alpha-olefins in uncontaminated crude oils. Overall, the integration of GC×GC‑TOFMS with a pixel‑based preprocessing approach followed by PCA analysis provided a rapid, robust workflow for preliminary screening and discrimination of drilling‑fluid contamination in crude oils, thereby mitigating the risk of misleading geochemical interpretations in reservoir studies.