A SIMPLIFIED METHOD FOR DETERMINATION OF THERMAL MATURITY BY RAMAN SPECTROSCOPY OF CARBONACEOUS MATTER

There have been many attempts at using Raman spectroscopy of organic matter as a paleothermometer in shales but no standard method has emerged. The difficulty is that there is no agreement on how to best process the raw Raman data or on which properties of the spectra best reflect thermal maturity. Most of the proposed methods involve the fitting of theoretical peak functions to the measured spectra followed by calculation of some quantitative property of these peaks that has a correlation with other known paleothermometer such as vitrinite reflectance (%VRo). The problem is that as the shape of the spectra change with maturity, workers have used different number of fitted peaks in different locations and varied the fitted theoretical function. This has made it hard to reproduce results (Lupoi et al. 2018). We propose a method using minimal processing, that avoids both peak-fitting and baseline correction, and can be done with widely available software that yields excellent correlation to known %VRo values in coals and black shales of the Appalachian basin. We obtained Raman spectra from vitrinite slides, coal chips, and shale chips using a 532 nm green laser under 50X magnification and 5% laser power, integrating 5 scans per sample. After testing several spectral parameters such as position of the D1 and G peaks, G/D1 ratio, height of the saddle between D1 and G, and prominence of the D1 and G peaks, we found that spacing between the D1 and G peaks, known in the literature as Raman Band Shift (RBS), is easy to measure with minimal processing and has a very high correlation coefficient with known %VRo (R²=0.96) in the %VRo range from 0.7 and 3.5. The linear correlation does break down both below and above this range. The procedure consists of: 1) inspecting the spectra for extraneous secondary peaks in the 1200 to 1900 cm^-1 range, 2) smoothing of the spectra to remove noise by running three passes of a 9-point centered running average, 3) determining the location of D1 and G peak maxima, and 4) calculation of RBS. There is no need to baseline the spectra except in cases of very high fluorescence where the routine for finding the D1 maximum may fail. With our current calibration dataset, calculated VRo%=0.0452*RBS-8.79 with a standard deviation of 2% in RBS which translates to an average difference between measured and calculated %VRo of +/- 0.12.