Maturation grade of coals as revealed by Raman spectroscopy: Progress and problems

https://doi.org/10.1016/j.saa.2005.02.015Get rights and content

Abstract

The present study questions the sensitivity and the accuracy of Raman spectroscopy as a tool for determining the maturity of natural organic matter (NOM). It focuses on the definition of optimized experimental parameters in order to maximize the quality of the Raman signal and control the accuracy and reproducibility of measurements. A series of 11 coals has been investigated, sampling a wide maturity range (2–7% vitrinite reflectance VR). The role of experimental parameters is first investigated. An excitation wavelength of 514.5 nm gives better results than 457.9 and 632.8 nm, minimizing the fluorescence background observed in the spectra of low-rank coals. Both Raman and fluorescence spectra were investigated with time-resolved experiments in air and argon. These data show that fluorescence and Raman spectra are sensitive to acquisition time and laser power parameters, and reveal a physicochemical instability of the samples under laser irradiation, mostly due to photo-oxidation processes. These data clearly show that the experiments, especially in air, should be performed with strictly constant acquisition parameters. In addition, the results of a whole series of coal measurements performed in air under constant experimental conditions show that Raman spectroscopy is definitely sensitive to the maturity of coal samples with VR > ∼1%. The most sensitive spectral maturity tracers are the width of the D-band (FWHM-D), the ratio of the peak intensities of the D- and G-bands (ID/IG), the normalized ratio of the band integrated intensities AD/[AD + AG] for the maturity range VR = 3–7% and the width of the G-band (FWHM-G) for VR = 1–5%. However, the accuracy and reproducibility are definitely weaker in such measurements compared to the standard VR. Future work must solve the problem of sample stability under laser irradiation, and greatly increase the number of samples to improve the statistical significance of the results.

Introduction

The chemical composition and structural order of natural organic matter (NOM), such as C-rich solids (e.g. coals) or materials dispersed in a host rock (e.g. kerogen, meteoritic OM), are determined by geological processes affecting the thermal history (time–temperature pair), physicochemical parameters acting as co-factors (pressure, presence of water, etc.) and thermodynamically irreversible reactions. The degree of advancement of these processes, referred to as the NOM maturity is, therefore, a useful indicator of the thermal history of the OM and/or the host rock, with very important applications for instance in petroleum research, geology, and of special interest here, planetology.

Many techniques to determine the maturity of natural organic matter are currently available: optical reflectance (vitrinite reflectance for coals), transmission electronic microscopy (TEM) (imaging the extent of the polyaromatic structure and its degree of stacking), Rock–Eval pyrolysis (Tmax), elemental analysis, fluorescence, fluorescence alteration and Raman spectrometry. These techniques are generally well suited in a restricted range of maturity, and vary in terms of the difficulty and time required for implementation. Raman micro-spectrometry offers a number of advantages: (1) it allows measurements at a micrometric scale and thus requires only a very small amount of material; (2) it does not require organic/mineral separation by acid attack or the use of polished sections; (3) it is fast and easy-to-implement; (4) it provides abiogenic spectral tracers particularly useful for planetological studies.

Wopenka and Pasteris [1] first demonstrated that Raman micro-spectrometry was suitable for determining the maturity of NOM by performing a survey over a wide series of natural samples. For poorly ordered materials, such as low-rank coals and kerogens, they first suggested the D-band width as a sensitive maturity indicator. Unfortunately, no correlation analysis was performed between the spectral information contained in the G- and D-first-order carbon bands and the maturity of samples quantified by independent tracers (e.g. vitrinite reflectance). Note that for such poorly ordered carbonaceous materials, the classic determination of the coherent domain length (La) using the intensity ratio of the G- and D-bands is not suitable. These authors also questioned measurement reproducibility and heterogeneity, but the quality of their data was insufficient to draw valuable conclusions. Since then, the sensitivity of Raman spectrometers has greatly improved and spectra can now be acquired in a few of minutes or even seconds, over a wide spectral range and with a high signal-to-noise ratio [2], [3], [4]. Therefore, recent studies over large series of samples demonstrate the high sensitivity of Raman spectroscopy to the maturity of coals and kerogens [3], [4]. However, these studies have been performed using different experimental procedures and different data processing techniques, making it difficult to compare their results. Moreover, they definitely do not provide a reliable procedure that can be used by other operators and, in particular, the question of the accuracy and reproducibility of measurements has not been addressed for the case of coals and kerogens, although some studies have investigated this point for graphitic materials [5]. A reliable and pertinent inter-laboratory technique is still required, i.e. a procedure capable of producing comparable results even when implemented in different laboratories. Defining such a procedure requires: (1) clearly defined experimental conditions; (2) quantification of spectral information in the Raman spectra by a stable and accurate numerical fitting procedure; (3) controlled and quantified sensitivity and accuracy of the measurements and of the numerical fitting procedure.

This study focuses on the definition of optimized experimental parameters in order to maximize the quality of the Raman signal and control the accuracy and reproducibility of measurements. First, the effect of excitation wavelength is explored in order to minimize the fluorescence background that perturbs the Raman spectra in most low-rank samples. Secondly, time-resolved experiments at the same laser spot location are performed in order to study sample stability under laser irradiation. The results of this work are applied by performing systematic measurements over a series of natural coals. Sensitive spectral tracers of maturity are then proposed for an excitation wavelength of 514.5 nm. Finally, the use of Raman spectroscopy as a possible sensitive, accurate, inter-laboratory technique is discussed.

Section snippets

Experimental setup

Raman experiments were performed at Laboratoire de Sciences de la Terre (ENS-Lyon, France), with a JOBIN-YVON LABRAM spectrometer. The laser beam was focused by a microscope equipped with a 50× lens, leading to a spot diameter of approximately 2–3 μm. The spectral resolution was ∼1 cm−1 (1800 l/mm grating). The power at the sample surface was carefully measured and ranged generally between 500 and 600 μW, except for special tests. Two sets of experiments were performed. The first consisted in

Quantification of Raman spectra

The accurate quantification of Raman spectra by a systematic procedure is necessary to derive reliable spectral parameters that can be then compared between different operators and/or equipment setups. For pre-graphitic and graphitic NOM, Raman spectra exhibit no or negligible fluorescence background and baseline subtraction is, therefore, straightforward. For less mature materials, the fluorescence background is significant and the fluorescence intensity increases as maturity decreases.

Effects of excitation wavelength

Three visible excitation wavelengths were used in an attempt to reduce or eliminate the fluorescence background in the spectra: 632.8, 514.5, and 457.9 nm. Measurements were carried out on three coals with maturities VR = 1.16, 1.86, and 5.45%. None of the wavelengths fully removed the fluorescence background from any of these coals. The lowest fluorescence intensity was obtained with the 514.5 nm wavelength and the highest with the red 632.8 nm wavelength (Fig. 2). These results are not consistent

Conclusion

The main results of this work are the following:

  • Raman spectroscopy is definitely sensitive to the maturity of coal samples with VR > ∼1%. Using strictly constant experimental conditions with a 514.5 nm excitation, a LBWF fitting procedure and a very simple sample preparation procedure, coal maturity can be assessed using spectral tracers based on the spectral parameters of the two fit profiles. The most sensitive tracers are FWHM-D and ID/IG for the maturity range VR = 3–7%, and FWHM-G for VR = 1–3%.

Acknowledgements

We greatly thank the Centre National de Spectrométrie Raman (Laboratoire des Sciences de la Terre ENS-Lyon, France), where all Raman experiments were performed. We are also grateful to O. Brissaud (LPG – Grenoble) who designed and built the argon cell and Hervé Cardon (LST-ENS-Lyon) for technical assistance during the experiments. We kindly acknowledge Dr. M. Vandenbroucke (Institut Français du Pétrole – Rueil Malmaison, France) for numerous stimulating discussions on sedimentary organic

References (17)

  • O. Beyssac et al.

    Spectrochim. Acta Part A

    (2003)
  • J.-N. Rouzaud et al.

    Fuel Process. Technol.

    (2002)
  • J.-N. Rouzaud et al.

    Thin Solid Films

    (1983)
  • B. Pradier et al.

    Adv. Org. Geochem.

    (1990)
  • B. Pradier et al.

    Org. Geochem.

    (1992)
  • W.E. Kilby

    Int. J. Coal Geol.

    (1988)
  • B. Wopenka et al.

    Am. Miner.

    (1993)
  • O. Beyssac et al.

    Metamorph. Geol.

    (2002)
There are more references available in the full text version of this article.

Cited by (197)

View all citing articles on Scopus
View full text