Method for processing data obtained by raman spectroscopy to determine thermal maturity of organic matter in rock

Abstract

This invention relates to the petroleum industry, and more particularly to the field of reservoir modeling, simulation and evaluation, and discloses a method for processing data obtained by Raman spectroscopy to determine the thermal maturity of organic matter in rocks, the method being calibrated based on a vitrinite reflectance standard using different excitation wavelengths.

Description

Invention Field

[0001] This invention pertains to the petroleum industry, and more specifically to the field of reservoir modeling, simulation, and evaluation. It describes a method for processing data obtained through Raman spectroscopy to determine the thermal maturity of organic matter in rocks, wherein the method is calibrated according to the vitrinite standard using different excitation radiations in a dispersive or interferometric Raman spectrometer. Background of the Invention

[0002] One area of ​​great interest to the petroleum industry is determining the thermal maturity of organic matter in rocks. Vitrin, a microscopic component, is widely used as an indicator of the thermal maturity of organic matter when analyzed using percentage reflectance (%Ro) techniques. Therefore, the study of vitrin reflectance is a key method for obtaining the temperature history of sedimentary basins. This technique can be used to measure the conversion of kerogen to hydrocarbons, and involves fixing, cutting, and polishing the sample to a thickness of 30 micrometers, resulting in high-quality palynological sections suitable for reflectance experimental determination.

[0003] Furthermore, Raman spectroscopy has been widely used to correlate spectral distributions with vitrinite or bitumen reflectance data to estimate the thermal maturity of organic matter (Sauerer et al., 2017, “FAST AND ACCURATE SHALE MATURITY DETERMINATION BY RAMAN SPECTROSCOPY MEASUREMENT WITH MINIMAL SAMPLE PREPARATION”, *International Journal of Coal Geology*, Vol. 173, pp. 150-157; Schmidt et al., 2017, “MATURITY ESTIMATION OF PHYTOCLASTS IN STREW MOUNTS BY MICRO-RAMANSPECTROSCOPY”, *International Journal of Coal Geology*, Vol. 173, pp. 1-8; Henry et al., 2018, “ASSESSING LOW-MATURITY ORGANIC MATTER IN SHALES USING RAMAN SPECTROSCOPY: EFFECTS OF SAMPLE PREPARATION AND OPERATING) "PROCEDURE", *International Journal of Coal Geology*, Vol. 191, pp. 135-151; Henry et al., 2019a, "A RAPID METHOD FOR DETERMINING ORGANIC MATTER MATURITY USING RAMAN SPECTROSCOPY: APPLICATION TO CARBONIFEROUS ORGANIC-RICH MUDSTONES AND COALS", *International Journal of Coal Geology*, Vol. 203, pp. 87-98; Henry et al., 2019b, "RAMAN SPECTROSCOPY AS A TOOL TO DETERMINE THE THERMAL MATURITY OF ORGANIC MATTER: APPLICATION TO SEDIMENTARY, METAMORPHIC AND STUCTURAL GEOLOGY", *Geoscience Review*, Vol. 198, pp. 102-936; Khatibi et al., 2018, "RAMAN SPECTROSCOPY: AN ANALYTICAL TOOL FOR "EVALUATING ORGANIC MATTER", Journal of Petroleum, Natural Gas and Petrochemical Science, Vol. 1, No. 1, pp. 28-33;Schito and Corrado, 2020, “ANAUTOMATIC APPROACH FOR CHARACTERIZATION OF THE THERMAL MATURITY OF DISPERSEDORGANIC MATTER RAMAN SPECTRA AT LOW DIAGENETIC STAGES”, Geological Society of London, Special Publications, Vol. 484, No. 1, pp. 107–119; Wilkins et al., 2018, “THERMAL MATURITY EVALUATION FROM INERTINITES BY RAMAN SPECTROSCOPY: THE 'RAMM' TECHNIQUE”, International Journal of Coal Geology, Vol. 128, pp. 143–152. Compared to traditional techniques for determining vitrinite or bitumen reflectance, Raman spectroscopy is a rapid and non-destructive technique that can be used to supplement conventional methods or independently, providing a method for screening samples before more expensive and destructive analyses.

[0004] The parameters typically used to estimate reflectance from Raman spectroscopy are based on the area and location of the D and G bands, which characterize organic matter. However, many difficulties exist regarding data acquisition, such as the choice of excitation line to use, sample fluorescence, laser type, and power. In previous work (e.g., Henry et al., 2019b; Schmidt et al., 2020), factors related to the processing and mathematical handling of spectral data, such as noise smoothing filters, baseline calculation, assignment, and deconvolution, were performed manually and without standardization, making it difficult to correlate the results with vitrinite or bitumen spectra. While this approach is very powerful and has become the standard for mature organic matter research, it is a manual, slow, subjective, and expensive procedure, prone to human error, and requires highly specialized training. Therefore, there is a desire to find alternatives that require shorter analysis times and produce equivalent reflectance values ​​for use as references in exploration, as these can reduce resource consumption in the process.

[0005] In this sense, the present invention enables a method to process Raman spectral results obtained from different excitation lines ranging from the near-infrared region to the ultraviolet, and generate equivalent vitrinite reflectance values ​​within seconds of calibration and software training. This solution can be applied to assess the maturity of organic matter present in rock samples from oil exploration wells. Given that this solution requires no sample pretreatment, it can be applied directly to the exploration area using portable equipment. Compared to vitrinite reflectance studies, this solution offers advantages such as faster results, richer information per sample, and higher procedural standardization, preventing inconsistent results for the same sample due to operator subjectivity and susceptibility to errors caused by the complexity of the analysis.

[0006] The development of this method involves implementing algorithms for processing spectral data that eliminate manual and repetitive procedures by analysts and standardize a set of methods for data processing and analysis. Therefore, the primary objective is to ensure the reliability of results by training these algorithms to reduce analysis time, subjectivity, and potential human error in Raman spectroscopy processing through established and standardized mathematical procedures. Existing technology

[0007] In the following paragraphs, seven (seven) documents will be described, which describe the prior art in terms of the content proposed and the limitations of each solution. Following the description, the differences between the present invention and the prior art documents will be discussed.

[0008] Document US 11029250 B2 describes a method that takes into account the separation values ​​of the G and D1 bands in the Raman spectrum to use an equation To estimate the vitrinite reflectance, where x is the vitrinite reflectance, and c1 and c2 are constants obtained for each instrument.

[0009] Document US 2023118696 A1 aims to obtain Raman spectra of organic matter in geological rock samples for determining kerogen properties. The obtained Raman spectra are quantitatively correlated with thermal maturity through Raman parameters of the G and D bands and their separation. It uses a general description of the rocks, including Type I, II, and III kerogen, asphaltenes, and bitumen. It should be noted that analysis of petroleum samples containing asphaltenes requires prior separation steps, such as gas chromatography, and cannot be performed directly on rock samples, as described in the document. Furthermore, they used a computational model but did not calibrate and validate it using vitrinite. It is noteworthy that vitrinite represents a more reliable indicator of organic matter maturity compared to other carbonaceous materials.

[0010] The document by DeHan et al., entitled "SAMPLE MATURATION CALCULATED USING RAMANSPECTROSCOPIC PARAMETERS FOR SOLID ORGANICS: METHODOLOGY AND GEOLOGICAL APPLICATIONS," relates to the methodology and geological applications of using Raman spectroscopy parameters to calculate the maturity of solid organic matter samples. The carbonized materials processed in this document (e.g., vitrinite, hemivitrinite, hemifilamentite, filamentite, or inertinite in coal microstructures) may exhibit different condensations of aromatic rings, leading to significant differences in Raman spectral distributions and reflectance values. Among these carbonized materials, vitrinite is a more reliable indicator of organic matter maturity (ASTM D7708-23 and ASTM D7708-14 standards).

[0011] Therefore, the analysis, particularly focusing on vitrinite to infer the maturity of organic matter, is extremely important. Unlike DeHan et al., he obtained Raman spectra of different carbides using only a single 532 nm laser line, involving only two Raman parameters: the separation between the G and D bands and the ratio of Raman height to reflectance of the D and G bands. Furthermore, the aforementioned document does not perform data processing such as baseline correction, smoothing, band deconvolution, and intensity normalization.

[0012] Henry et al.'s paper, "A RAPID METHOD FOR DETERMINING ORGANIC MATTERMATURITY USING RAMAN SPECTROSCOPY: APPLICATION TO CARBONIFEROUS ORGANIC-RICHMUDSTONES AND COALS," estimates vitrinite reflectance using parameters obtained from Raman spectroscopy. It correlates the %Ro values ​​of standards obtained via optical microscopy with values ​​estimated by Raman spectroscopy and applies the method to other wells with coal and shale samples. Raman data were obtained using only a 514.5 nm laser excitation. Several variables were obtained from the Raman spectra, such as the half-width-half-height (FWHM) of the G and D bands, the interval between these two bands (RBS), the intensity ratio R1, and the area ratio SSA. Using these parameters, curves were obtained to estimate the %Ro for each individually, with the FWHM of the G band inferred as the optimal parameter. However, it does not estimate the spectral parameter values ​​after band deconvolution, raising questions about the reproducibility of the method.

[0013] Henry et al.'s paper, "RAMAN SPECTROSCOPY AS A TOOL TO DETERMINE THE THERMALMATURITY OF ORGANIC MATTER: APPLICATION TO SEDIMENTARY, METAMORPHIC AND STRUCTURAL GEOLOGY," is a review of several works in the literature that use Raman spectra of standard vitrinite samples to obtain spectral parameters that allow for the estimation of %Ro and the determination of organic matter thermal maturity. This work mentions the need for standardized band nomenclature and the obtained Raman parameters, but achieves the best correlation for the %Ro range above 3.0, suggesting that this work generally uses these calibration methods for highly mature organic matter.

[0014] The document by Sauerer et al., entitled "FAST AND ACCURATE SHALE MATURITY DETERMINATION BY RAMAN SPECTROSCOPY MEASUREMENT WITH MINIMAL SAMPLE PREPARATION," analyzed sedimentary rock samples containing organic matter classified as type II kerogen. Raman spectral parameters were obtained through deconvolution, and they emphasized that RBS was the only parameter sufficiently correlated with %Ro. They were not calibrated using %Ro according to standards.

[0015] The document by Wilkins et al., entitled "Thermal maturity evaluation from inertinites by Raman spectroscopy: The 'RaMM' technique," describes the possibility of determining the equivalent %Ro from inertinites and vitrinite samples. The reflectance studied ranged from 0.4%Ro to 1.2%Ro. Raman spectra were collected using a 488 nm wavelength excitation line, and the spectral parameters were correlated with the %Ro of the vitrinite and inertinite microstructures based on deconvolution using only two bands. However, they did not use samples calibrated according to standards. The authors themselves recommend analyzing calibrated samples to make the method more reliable.

[0016] As seen in paragraphs 008 to 015, existing techniques for characterizing organic matter in rocks using Raman spectroscopy employ conventional methods, resulting in non-standardized data processing and thus varying across the mentioned literature, subject to analyst subjectivity. However, none propose the possibility of establishing procedures for processing Raman spectral data collected with different excitation lines, involving the use of machine learning to train predictions of %Ro based on calibrations of vitrinite standard samples analyzed according to ASTM D7708-23 and ASTM D7708-14 standards. Furthermore, it is noted that some studies use matrices other than petroleum, such as coal. Therefore, a complete calibration and training process is required. Although the principles are similar, the organic matter in petroleum and coal is at very different stages of maturity.

[0017] Although both this invention and the prior art discussed in paragraphs 008 to 015 above involve the use of Raman spectroscopy measurements, significant differences exist that make the method proposed in this application innovative. First, in this application, the Raman spectrum of vitrinite is obtained from a sample in the form of a 30-micrometer-thick flake of pollen phase, produced by an industrially known method, containing a previously selected fragment with reflectance values ​​ranging from 0.46 to 2.72, measured according to ASTM D7708-23 and ASTM D7708-14 standards, which was used for different method calibrations at different excitation wavelengths to obtain Raman spectra. This allows for verification of the vitrinite reflectance values ​​predicted by Raman spectroscopy relative to values ​​obtained using internationally recommended methods.

[0018] A calibration and subsequent estimation of vitrinite reflectance based on Raman spectra of these standard samples was developed, allowing the collection of Raman spectra of organic matter present in rocks using excitation lines of different wavelengths to estimate the maturity of organic matter in the analyzed rocks after standardized data processing. For this purpose, for each excitation wavelength, baseline adjustment was performed and wavenumbers in the range of 1100 cm⁻¹ were used. -1 Up to 1750 cm -1 The spectral data is processed by deconvolving the six spectral bands between them to standardize the data.

[0019] The method for constructing training algorithms for data processing via deconvolution involves managing the intensity and width of spectral bands for standardization and implementation of the analysis. Mathematical prediction methods using linear fitting are employed to weight Raman spectral parameters such as band center, intensity, full width at half maximum (FWHM), and exponential and logarithmic functions of all these variables to establish a correlation with the reflectance values ​​of the vitrinite sample.

[0020] Therefore, a total of 72 variables were used to train the vitrinite reflectance prediction using machine learning techniques. Subsequently, LASSO (Least Absolute Contraction and Selection Operator) regression analysis was used, which removes contributions that do not contribute to improving correlation and retains only the variables with the highest correlation to the %Ro value.

[0021] The variable with the highest weight was found to be the FWHM of the G-band; however, other variables also had significant weights, such as the logarithm and exponential of D1. This model avoids different values ​​obtained by different analysts using different methods in data processing, thus reducing the possibility of analytical errors due to user subjectivity.

[0022] In this way, a robust and comprehensive relationship was obtained, involving vitrinite spectral parameters and the maturity of organic matter, which is more in-depth than that reported in the aforementioned documents.

[0023] Therefore, this invention represents an innovative contribution to the use of vitrinite Raman spectroscopy in calibration to determine the maturity of organic matter in rocks, far exceeding those proposed in the various previous documents discussed in paragraphs 008 to 015.

[0024] Furthermore, it is noteworthy that the use of Raman spectroscopy in the method proposed in this invention enables the direct use of the parent rock without pretreatment, except for grading to a size suitable for the irradiation site, or even using lithofacies or palynological sections. This invention, developed for the use of Raman spectroscopy, also allows data to be acquired using different excitation laser lines calibrated in a suitable manner, enabling the extension of data to other types of microscopic components and carbon-based compounds. Finally, the data can be processed in a computer application accessible to any user. Such applications can be implemented using various computational methods, such as selecting excitation lines, baseline adjustment methods, band deconvolution, and spectral regions of interest.

[0025] Therefore, the detailed description of the proposed method involving spectral data processing makes the process robust because it considers all variables relevant to the prediction of vitrinite reflectance, but through parameterization, it produces an objective analysis and is rarely affected by user subjectivity. Summary of the Invention

[0026] This invention aims to develop a method for processing Raman spectroscopy results and generating equivalent vitrinite reflectance values ​​within seconds. This solution can be applied to assess the maturity of organic matter present in rock samples from oil exploration wells. Since the solution does not require sample pretreatment, it can be applied directly to the exploration area using portable equipment. Compared to vitrinite reflectance studies, this solution offers advantages such as faster result acquisition, richer information per sample, and higher procedural standardization. Attached Figure Description

[0027] Figure 1 Vitrin reflectance values ​​are shown, covering a range of hydrocarbon generation of interest to the petroleum industry, including geological processes of diagenesis, plutonicity, and metamorphism. Element A, highlighted, indicates the interval encompassing the established calibration curves.

[0028] Figure 2 Raman spectra of organic matter present in the same pollen phase layer are shown using excitation lines at wavelengths of 514.5 nm, 532 nm, 632.8 nm, and 785 nm, highlighting the variations in band distribution.

[0029] Figure 3 Wavenumbers are used to show the analytical range that enables the characterization of organic matter.

[0030] Figure 4 The application of spectral smoothing methods is shown, which minimizes noise present in the initial analysis and does not interfere with the evaluation of the data.

[0031] Figure 5 This demonstrates how to apply the two-point baseline adjustment method to disregard background radiation that may exist in the Raman spectrum from different sources.

[0032] Figure 6 This demonstrates that normalizing the intensity using the strongest spectral band makes the relative intensity the focus of the analysis, and the calculated area is comparable across different spectra.

[0033] Figure 7 The deconvolution of the Raman spectra of six bands obtained from pollen phase thin sections using the excitation line at 514.5 nm is shown, and the sum of these bands perfectly reproduces the original distribution.

[0034] Figure 8 The flowchart illustrates the parameters obtained from the Raman spectra used in the model to predict the percentage of vitrinite reflectance (%Ro): for each band obtained in the deconvolution, the position of the band center, the full width at half maximum (FWHM) of the band, and the height of each band are used as variables; other variables are obtained by applying logarithmic and exponential functions to each of the initial variables; this set of variables is also applied to calibration data for standard vitrinite samples, whose reflectance has been measured and correlated with the thermal maturity of the organic matter; finally, machine learning techniques enable the selection and weighting of each variable to predict the thermal maturity of the sample based on the variable that contributes most to the final analysis. Detailed Implementation

[0035] This invention relates to a method for processing data obtained through Raman spectroscopy to determine the thermal maturity of organic matter in rocks, wherein the method is calibrated according to a vitrinite standard using different excitation radiations. The method includes the following steps:

[0036] (a) Select the wavenumber and collect Raman spectra from a standard vitrinite sample with a known reflectance value;

[0037] (b) Smooth Raman spectrum;

[0038] (c) Baseline correction / subtraction of the spectrum;

[0039] (d) Normalization of the relative intensity of the Raman spectrum;

[0040] (e) Deconvolve the Raman spectra to obtain the model parameters;

[0041] (f) Calibrate the vitrinite reflectance prediction model; and

[0042] (g) Collect Raman spectra of organic matter present in the rock and predict equivalent reflectance values ​​based on previous calibration.

[0043] To begin calibration, a sample containing reflectance data generated by another technique must be used. For analysis, the Raman scattering excitation laser must be located in a region of organic matter. Raman spectra are obtained from different points, thus the spectra truly represent the diversity of the material. Each laser incident produces a Raman spectrum.

[0044] Using the same vitrinite standard sample, calibration can be extended to different Raman excitation lines, which will enable the identification of problem samples that show better results in Raman spectra across excitation lines in the ultraviolet, visible, or near-infrared regions.

[0045] Raman spectra were recorded in text files x and y, each containing two columns (wavenumber and Raman intensity) and thousands of rows, with values ​​for each variable in each column. Files were manually selected and loaded into memory via the software's graphical interface. A 1100 cm⁻¹ depth was selected for each spectrum. -1 Up to 1750cm -1 The wavenumber range is defined. A Savitzky-Golay filter is applied to the intensity of each spectrum sorted by wavenumber, where the window size is 21 and the polynomial degree is 2. Baseline correction is calculated for the intensity of each spectrum. Each spectrum is normalized. Deconvolution is performed using a parameter estimation procedure representing the G and D curves (D1, D2, D3, D4, and D5) using a combined model of the Voigt and Gaussian functions.

[0046] The estimated parameters are the center, FWHM, and intensity of each curve. These estimated parameters are combined and a nonlinear function is applied to generate new descriptors. Specifically, the half-width at half-maximum (FWHM) of each band and a function representing exponential decay are applied to the parameters in this step. Finally, using these descriptive parameters and reflectance data, LASSO (Least Absolute Shrinkage and Selection Operator) regression is manually applied.

[0047] As a result, a mathematical model with linear terms in the parameters and variables was generated. The model was manually recorded in a file containing the parameter values ​​from the original sequence presented by descriptors. To begin predictions, it was necessary to use organic-containing rock samples that had not yet been characterized by reflectance analysis.

[0048] These samples underwent the same spectral generation steps and mathematical procedures as described above, except for the final step, the application of LASSO technology. In this case, a file containing the mathematical model parameter values ​​must be selected and loaded into memory. Finally, an enhanced set of descriptive parameters of the uncharacterized sample spectra is used as input to the mathematical model for performing vitrinite reflectance prediction calculations.

[0049] It should be mentioned that Raman spectroscopy processing is performed in an optimal manner and can be converted to a spectrum obtained by any Raman spectrometer and in any excitation radiation used.

[0050] In addition, deconvolution generates a large number of variables, which are actively evaluated using the LASSO regression method to obtain the variables that are most correlated with the determined %Ro.

[0051] This method can be transferred to different samples without the need for palynological slices or any special treatment, except for grading to introduce the sample chamber into the Raman apparatus.

[0052] Invention Examples

[0053] To construct calibration curves using Raman parameters measured based on known %Ro values, nine palynological sections (%Ro values ​​of 0.46, 0.64, 0.70, 0.76, 0.86, 0.85, 2.07, 2.61, and 2.72, respectively) were used. The vitrinite reflectance values ​​of the nine sections cover the hydrocarbon generation range of interest to the petroleum industry, from 0.46% to 2.72%, encompassing geological processes such as diagenesis, plutonicization, and metamorphism. Figure 1As shown. Each thin section studied contained approximately 10 to 20 organic fragments classified as plant debris, and its Raman spectrum was obtained, with each thin section having a specific %Ro value specified by organic petrography according to the D7708-23 standard. In obtaining the Raman spectra, nine thin sections were analyzed on a Horiba Scientific LabRAM HR Evolution instrument with an excitation line wavelength of 514.5 nm, using 50x, 3.2%, 600 (500 nm) objectives, a cumulative time of 45 s, and a range of 100 cm⁻¹. -1 Up to 1900 cm -1 .

[0054] The spectrum obtained through the excitation line at a wavelength of 632.8 nm was acquired on a Bruker dispersive spectrometer (modelSENTERRA) with a spectral resolution of 3 cm⁻¹. -1 Up to 5 cm -1 The magnifying objective is 50x ULWD (NA = 0.51) and the confocal aperture is 50 μm.

[0055] Subsequently, noise correction filters, baseline correction, and normalization were used to analyze a specific region (1750 cm⁻¹) of the pollen leaf spectrum. -1 Up to 1100 cm -1 Mathematical processing was performed to obtain Raman parameters (D1 band, D3 band, D4 band, D5 band, D6 band, and G band), using Origin 9.0 and FityK 1.3.1 software.

[0056] Calibration curves were also constructed for two different excitation lines (514.5 nm and 632.8 nm) using exponential adjustments to the G-FWHM parameters, which showed the best correlation with %Ro. Good correlations were shown between the measured and predicted values. These parameters were used as initial input data for method calibration.

[0057] By establishing best practices for data processing, standardization, and parameterization, a method for obtaining %Ro values ​​from Raman spectra was planned. LASSO (Minimum Absolute Contraction and Selection Operator) regression analysis was used for calibration because it has the advantage of reducing the number of input variables in the model, retaining only those most relevant to the prediction. These variables are parameters estimated and calculated from the spectral curves, as well as linear and nonlinear applications and combinations of these parameters.

[0058] Although only the center, sigma, and amplitude parameters are estimated, several other parameters of the function can be calculated, such as the full width at half maximum (FWHM), height, and area under the curve. As an example, Equation 1 gives the Gaussian function and its main parameters.

[0059]

[0060] Where x is the wavenumber of the spectrum, and A, μ, σ, FWHM and h are the amplitude, center, standard deviation, half width and height of the curve, respectively.

[0061] Table 1: Reference for estimating curve parameters.

[0062]

[0063] Note: I k It corresponds to the intensity at the center of the curve.

[0064] Table 2: Search intervals used to estimate curve parameters.

[0065]

[0066] Note: I k It corresponds to the intensity at the center of the curve.

[0067] The proposed scheme relates to the generation of a vitrinite reflectance prediction model or the calculation of its predicted values. As already explained, the algorithm used is LASSO (Least Absolute Shrinkage and Selection Operator), which has the advantage of reducing the number of input variables in the model, retaining only those variables most relevant to the prediction. The algorithm has input parameters that balance the number of non-zero parameters (the smaller the better) and predictive performance (the larger the better). The value of this parameter is set to 0.05. The input variables of the model are the estimated and calculated parameters of the spectral curve, as well as the linear and nonlinear applications and combinations of these parameters, as shown in Equation 2 below.

[0068]

[0069] Therefore, the algorithm performs adjustments (calibration) using dozens of input variables and one output variable (vitrinite reflectance value) in each spectrum-%Ro pair. It's worth noting that many points (spectrum-%Ro pairs) are needed for adjustment to achieve statistical significance. At the end of this adjustment, the algorithm provides a calibrated model that can be used to predict %Ro based on the new spectrum. This model contains a reduced set of non-zero parameters (e.g., less than 10) that correspond to the input variables that contribute most to the prediction of %Ro. Equation 3 shows an example of the generated prediction model.

[0070]

[0071] in denoted as the predicted value of vitrinite reflectance (%Ro), and p0, p1, p2, and p3 are parameters of the prediction model, while FWHMG, hD1, and σD2 are model input variables, which are obtained by processing the spectrum of the organic matter maturity to be determined.

[0072] As mentioned above, this step also involves using a model to obtain a prediction of %Ro from the Raman spectrum, which must be used with... Figure 8 The same steps are used in the scheme shown.

[0073] This method can be easily installed on a computer using an installer file (.exe) and an installation wizard, in which the installation location must be selected. Furthermore, it can be associated with any Raman device, different types of lasers, and can be used for calibration or direct estimation of vitrinite or pitch reflectivity.

[0074] However, this method performs all stages of the spectral analysis within seconds; that is, it reads files containing Raman spectra (“.txt” and “.dpt”), which are easily viewed in a text editor and may or may not have a header, and may have two or more variables, the first being the wavenumber and the others being the Raman intensity. Raman spectra of organic matter in rocks can be recorded using the excitation lines at wavelengths of 514.5 nm or 632.8 nm mentioned in this example, or any other excitation line used for Raman scattering.

[0075] This method identifies an indicator region containing organic matter, which includes an area of ​​1100 cm. -1 Up to 1750cm -1 The wavenumbers between these values ​​can be interpreted as input parameters for the algorithm. Then, based on what was established during previous training using standard samples, it applies mathematically parameterized processing: filtering, baseline correction, and band deconvolution. This approach avoids human subjectivity and standardizes all mathematical processing applied to the input data.

[0076] Furthermore, this method employs more variables in its mathematical processing, thus possessing novelty and demonstrating a development that could not be manually reproduced in previous work. For each processed dataset, the results are presented graphically, containing all files for a given measurement selected for the analysis, where the final %Ro values ​​of organic matter present in the analysis, deconvolution, identification, and reference samples can be visualized. Training these algorithms revealed a significant reduction in analysis time, achieving higher accuracy and standardization in the analysis.

[0077] Some general aspects of the method proposed in this invention can be observed, which together give it novelty and inventiveness:

[0078] • Raman spectra of organic matter present in rock samples and standard samples containing vitrinite fragments in the form of lithofacies thin sections were collected. The samples were pre-selected and the reflectance values ​​were measured in the range of 0.46 to 2.72 according to ASTM D7708-23 and ASTM D7708-14 standards. This allowed the method to be calibrated with values ​​of high confidence.

[0079] • Use different excitation lines ( Figure 1 The recorded Raman spectra make this method flexible and adaptable to different needs for characterizing organic matter;

[0080] • Use wavenumber range of 1100 cm -1 Up to 1750 cm -1 The six spectral bands between them are processed through smoothing, baseline adjustment, and deconvolution of the spectral data to ensure the robustness and ease of transfer of the obtained results;

[0081] • By imposing constraints on intensity and bandwidth, a systematic approach is adopted for the acquisition and processing of data obtained through managed deconvolution, to standardize and implement predictive models for determining reflectivity. Figure 2 );

[0082] • A mathematical prediction method was used to obtain vitrin reflectance values ​​through parametrically weighted linear adjustments, employing Raman spectroscopy-adjusted parameters (such as band center, band intensity, band area, and full width at half maximum (FWHM)) and exponential and logarithmic functions of all these variables. A total of 72 variables were used to train the vitrin reflectance prediction using machine learning techniques. Subsequently, LASSO (Minimum Absolute Shrinkage and Selection Operator) regression analysis was used, which removed useless contributions and retained only variables with some correlation to the %Ro value. The variable with the largest weight was found to be the FWHM of the G band; however, other variables also had significant weights, such as the exponential and logarithmic values ​​of D1. The model works with multiple inputs to produce a single output, namely the inference of the %Ro value of the problem sample. Figure 3 );

[0083] • Different calibrations can be performed using vitrinite standard samples, which can be applied to any type of sample that shows a signal in Raman spectra and can be performed using any Raman device, where the excitation line can be in the ultraviolet, visible, or near-infrared regions, provided that the vitrinite standard sample is used for the initial calibration performed via machine learning.

[0084] The method proposed in this invention can be directly applied to the study of the thermal maturity of organic matter. Currently, this procedure is destructive on the sample and requires not only highly trained personnel but also specific thin sections for the estimation. Therefore, as a rapid, non-destructive procedure, this invention has immediate effects. Furthermore, it can be used with any type of sample and any Raman device that displays a signal in Raman spectroscopy, provided that a vitrinite standard sample is used for initial calibration via machine learning.

[0085] One area of ​​great interest to the petroleum industry is determining the thermal maturity of organic matter in rocks. Vitrin, a microscopic component, is widely used as an indicator of the thermal maturity of organic matter when analyzed using percentage reflectance (%Ro) techniques. Raman spectroscopy has been widely used to estimate the reflectance of vitrin or bitumen, thereby determining the thermal maturity of organic matter.

[0086] Compared to conventional techniques used to determine the reflectance of vitrinite or pitch, Raman spectroscopy is a rapid and non-destructive technique that can be used to supplement conventional methods or independently, providing a means of screening samples before more expensive and destructive analyses. In previous work (e.g., Henry et al., 2019b; Schmidt et al., 2020), factors related to the processing and mathematical handling of spectral data, such as noise smoothing filters, baseline calculations, identification, and deconvolution, were performed manually and without standardization, making it difficult to correlate the results with vitrinite or pitch spectra.

[0087] Since these procedures directly interfere with the determination of the thermal maturity of organic matter in a more accurate, reliable, and reproducible manner, the purpose of this patent is to establish an innovative method for processing Raman spectra.

Claims

1. A method for processing data obtained by Raman spectroscopy to determine the thermal maturity of organic matter in rocks, characterized in that... Includes the following steps: (a) Select the wavenumber and collect Raman spectra from a standard vitrinite sample with a known reflectance value; (b) Smooth the Raman spectrum; (c) Baseline correction / subtraction of the spectrum; (d) Normalize the relative intensity of the Raman spectrum; (e) Deconvolve the Raman spectrum to obtain the parameters of the model; (f) Calibrate the vitrinite reflectance prediction model; and (g) Collect Raman spectra of organic matter present in the rock and predict equivalent reflectance values ​​based on previous calibration.

2. The method according to claim 1, characterized in that, Select 1100 cm⁻¹ in each spectrum -1 Up to 1750 cm -1 Wavenumber ranges between.

3. The method according to claim 1, characterized in that, The Savitzky-Gore filter is applied to the intensity of each spectrum sorted by wavenumber, where the window size is 21 and the polynomial degree is 2.

4. The method according to claim 1, characterized in that, Combine the following parameters: center of each curve, FWHM, and intensity.

5. The method according to claim 1, characterized in that, Using these descriptive parameters and reflectance data, LASSO (Least Absolute Shrinkage and Selection Operator) regression techniques were manually applied.

6. The method according to claim 1, characterized in that, An enhanced set of descriptive parameters of the spectra of uncharacterized samples is used as input to a mathematical model for performing vitrinite reflectance prediction calculations.

7. The method according to claim 1, characterized in that, Using the same vitrinite standard sample, the calibration was extended to different Raman excitation lines located in the ultraviolet, visible, or near-infrared regions.

Images