Towards a Higher Comparability of Geothermometric Data obtained by Raman Spectroscopy of Carbonaceous Material. Part I: Evaluation of Biasing Factors

Abstract

Raman spectroscopy of carbonaceous material (RSCM) is frequently used to determine metamorphic peak temperatures from the structural order of carbonaceous material enclosed in metasediments. This method provides a quick, robust and relatively cheap geothermometer. However, the comparability of the RSCM parameter is low as there are at least three major sources of biasing factors. These sources are the spectral curve-fitting procedure, the sample characteristics itself and the experimental design including the used Raman system. To assess the impacts of the biasing factors on RSCM, a series of experiments was performed. The experiments showed that curve-fitting is strongly influenced by individual operator-bias and the degrees of freedom in the model, implying the need for a standardised curve-fitting procedure. Due to the diversity of components (optics, light detection device, gratings, etc.) and their combinations within the Raman systems, different Raman instruments generally give differing results. Consequently, to estimate comparable metamorphic temperatures from RSCM data, every Raman instrument needs its own calibration. This demands a reference material series that covers the entire temperature calibration range. Although sample heterogeneity will still induce some variation, a reference material series combined with standardised curve-fitting procedures will significantly increase the overall comparability of RSCM data from different laboratories.

Assessment of Thermal Maturity Trends in Devonian–Mississippian Source Rocks Using Raman Spectroscopy: Limitations of Peak-Fitting Method

“…dependent. These G-band peak positions agree with frequencies tabulated in the literature (Table 6) (Wang et al, 1990;Spotl et al, 1998;Matthews et al, 1999;Castiglioni et al, 2001;Beyssac et al, 2002;Quirico et al, 2005;Guedes et al, 2012;Lünsdorf et al, 2014;Li et al, 2015;Lünsdorf, 2016); however, using only one peak to fit the D-band resulted in a shift in the peak maximum location, as shown in Table 2. Limiting the fit to two peaks permitted the bands to be evaluated as a whole, without over-complicating the analysis, and also without needing to add further peaks when the true number of vibrational modes that contribute to the spectral region between ~1,200 and 1,500 cm −1 is unknown.…”

“…While a two-peak fit elucidated some basic correlations with VRo, it was apparent that additional peaks comprised the spectra, both from the experimental data and the literature (Wang et al, 1990;Matthews et al, 1999;Castiglioni et al, 2001;Beyssac et al, 2002;Jehlička et al, 2003;Quirico et al, 2005;Guedes et al, 2010Guedes et al, , 2012Lünsdorf et al, 2014;Li et al, 2015;Lünsdorf, 2016). Additional peaks were added to the fitting process and were centered between 1,263 and 1,286 cm −1 and 1,430 and 1,466 cm −1 , depending on the sample.…”

“…Raman spectra of shale reveal two characteristic peaks: the disordered, D-or D1-band, and the graphitic, or G-band (Wang et al, 1990). These features arise near ~1,310-1,370 and ~1,580-1,610 cm −1 , respectively; however, the exact frequency is dependent on the sample and the spectral acquisition parameters (Wang et al, 1990;Spotl et al, 1998;Matthews et al, 1999;Castiglioni et al, 2001;Beyssac et al, 2002;Quirico et al, 2005;Guedes et al, 2012;Lünsdorf et al, 2014;Li et al, 2015;Lünsdorf, 2016). Other D-vibrational modes have been characterized and are referred to as D2 (~1,620 cm −1 ), D3 (~1,500 cm −1 ), and D4 (~1,250 cm −1 ).…”

“…Interestingly, the G-D-band separation metric extended to the full 0.2-2.7% maturity range. Factors that could bias the correlations generated from peak-fitting the Raman spectra of carbonaceous samples were assessed and included baseline-correction technique, the curve fitting strategy (type of function, peak start position, position, and slope of baseline), intrinsic sample characteristics, the instrumental configuration, and operator interpretation (Lünsdorf et al, 2014). A linear baseline subtraction exhibited lower relative SDs for several samples, when contrasted to polynomial baseline-correction.…”

“…While studies comparing Raman spectral parameters to intrinsic properties like thermal maturity have uncovered some basic trends, they have employed a multitude of different experimental and spectral processing techniques to develop the correlations (Spotl et al, 1998;Quirico et al, 2003Quirico et al, , 2005Rahl et al, 2005;Guedes et al, 2010Guedes et al, , 2012Hinrichs et al, 2014;Lünsdorf et al, 2014;Wilkins et al, 2014;Zhou et al, 2014;Bonoldi et al, 2016;Lünsdorf, 2016;Lünsdorf and Lünsdorf, 2016;Schito et al, 2017). These methods differed in the number of samples evaluated, the acquisition parameters used to generate the data (excitation wavelength, spectral resolution, acquisition time, number of unique and random points measured per sample), data processing (i.e., baseline correction) and peak-fitting metrics (number of peaks, wavenumber, maximum peak width).…”