Compositional signatures of CM, CO, CV, and CK chondrites: Insights from Micro-FTIR spectroscopy and machine learning tools

M. Yesiltas, T.D. Glotch

Icarus, In Press, Journal Pre-proof, Available online 16 February 2026

LINK

“Highlights

  • Micro-FTIR mapping of 65 CM, CO, CV, and CK chondrites reveals class-specific spectral–compositional signatures at micrometer scales.
  • Quantitative analyses reveal systematic variations in chemical composition and alteration-related features across the chondrite classes.
  • Machine-learning classifiers (RF, PLS-DA, k-means) trained on pixel-level spectra achieve high classification performance and accurately distinguish carbonaceous chondrite classes.
  • The integrated spectroscopic–machine-learning framework provides a rapid, non-destructive tool for classifying carbonaceous chondrites.”

“Carbonaceous chondrites preserve diverse records of nebular and parent-body processes. Classification within the CM-CO-CV-CK chondrite systems may sometimes be uncertain because secondary processes can overprint primary textures and mineralogy. In this study, we apply micro-Fourier transform infrared (micro-FTIR) spectroscopy, combined with machine learning, to 65 meteorites representing these four groups to evaluate their spectral relationships and re-examine problematic classifications. Mid-infrared hyperspectral maps and spectra were analyzed using Random Forest (RF), Partial Least Squares (PLS), and k-means clustering algorithms. All models independently reproduce the established group structure with accuracies of 80–89%, confirming that micro-FTIR spectra contain diagnostic information distinguishing hydrated, anhydrous, and thermally metamorphosed carbonaceous chondrites. However, several specimens display systematic spectral and statistical deviations from their formal groups. In particular, LAR 12002 and MAC 02528 (cataloged as CV3) exhibit CK3/4-like signatures, NWA 16889 (CK3) aligns with CV3 chondrites, and MET 01077 (CM2) shows CO3-like spectral properties consistent with dehydration or heating. Additional CO chondrites fall along the CO-CV interface, suggesting that some meteorites have spectral properties intermediate between these groups. These findings demonstrate that micro-FTIR spectroscopy, coupled with statistical and machine-learning analyses, provides a quantitative analytical framework for assessing spectral heterogeneity, distinguishing among carbonaceous chondrite groups.”