--> The Use of Multivariate Analysis and Raman Spectroscopy for the Assessment of Thermal Maturity

47th Annual AAPG-SPE Eastern Section Joint Meeting

Datapages, Inc.Print this page

The Use of Multivariate Analysis and Raman Spectroscopy for the Assessment of Thermal Maturity


Vitrinite reflectance (VRo) is a standard petrographic method for evaluating thermal maturity (rank). The vitrinite reflectance protocol, however, requires significant petrographic proficiency, can be labor-intensive, and may be biased on an analyst-to-analyst basis. Correlations between thermal maturity and Raman spectra present an enticing option that can remove some of the weaknesses in the VRo protocol. Our previous research, and other published studies have shown that traditional peak-fitting methodologies for quantifying metrics from Raman spectra can be correlated to thermal maturity, however, these approaches also are inhibited by analyst subjectivity which can affect correlations between analyte and spectral properties. In this study, we have combined Raman spectroscopy with multivariate analysis (MVA) to create calibration models for the prediction of coal rank using VRo values and atomic O/C ratios. MVA techniques eliminated the ambiguity prevalent in peak-fitting Raman data by evaluating the full Raman spectrum, and identifying the spectral regions relevant to the construction of accurate thermal maturity models. Partial least squares (PLS) regression models were created using Raman spectra of 68 geographically diverse coal samples and VRo values (0.23 to 5.23%) or atomic O/C ratios (0.027-0.206) for 39 samples. The calibration set was validated by reserving half of the samples to serve as “unknowns” thereby assessing the model's predictive accuracy. Both models exhibited linear correlations, with coefficients of determination (R2) for the validation set of 0.99 (VRo) and 0.93 (atomic O/C), despite the geographic and rank diversity of the samples. This study demonstrated the applicability and power of using PLS models for thermal maturity prediction calibrated to complete Raman spectra as opposed to Raman-derived parameters. More recently, our focus has turned to developing a predictive model for evaluating the thermal maturity of shale. This quantitative MVA protocol provides a Raman-based alternative to the VRo industry benchmark for coal rank that mitigates the limitations and subjectivity of peak-fitting methods.