Multivariate relationship of methyl- and dimethyl-phenanthrenes to the maturity of organic matter in sedimentary sequences

Abstract

At least thirteen parameters relating phenanthrene and methylated phenanthrenes to the maturity of organic matter have been defined in the literature and a mass of the data have been accumulated. However, these parameters are not always effective in many basins. To explore an effective approach for using these big data, this paper re-studies the relationship of methyl- and dimethyl-phenanthrenes to the maturity of organic matter, mainly using partial least square regression (PLSR) and geochemical stabilities of these phenanthrenes. The samples for this study were taken from the shale and mudstone in the Cretaceous Qingshankou Formation in the Songliao Basin, China. It was found that integrated use of methyl- and dimethyl-phenanthrenes can overcome the limitations inherent in using them separately, and thus can enhance the accuracy of maturity calculations. The multivariate regression equation achieved from PLSR utilizes 95% of the variance information contained in the relative abundances of 1-, 2-, 3- and 9-MPs, 1,7- and 2,7-DMPs, and (methyl-phenanthrene [MP]; dimethyl-phenanthrene [DMP]). The coefficients of these phenanthrenes in this equation are consistent with their geochemical stabilities. The square of the correlation coefficient (R² = .94) of the multivariate regression equation is much higher than those (0.69–0.89) of the univariate regression equations derived from the previously-defined phenanthrene parameters. It is suggested that the multivariate regression approach presented in this paper substitute the previously-defined parameters and the corresponding univariate regression equations when they are not effective.

Keywords

Maturity organic matter methyl-phenanthrene dimethyl-phenanthrene multivariate relation

Introduction

The maturity of organic matter provides important information for studying the organic carbon cycle, is a necessary criterion for recognizing effective oil and gas source rocks, and is a key element of organic geochemistry and geology research (Barenbaum, 2017; Curiale et al., 1989; Galvez et al., 2020; Olah et al., 2011; Radke et al., 1997; Raiswell and Berner, 1987; Xiao et al., 2014; Zhang et al., 2013, 2019; Zhao et al., 2017; Zhou et al., 2021; Kumar et al., 2022, 2025a; Hakimi et al., 2023). Numerous parameters and methods for thermal maturity assessments have been developed. Vitrinite reflectance (%Ro) indicates time-temperature-integrated thermal stress and the thermal maturity level of organic matter (Kumar et al., 2022; Hatem et al., 2023; Sidik et al., 2024). When source rocks cannot be sampled, vitrinite reflectance cannot be measured, but the maturity of organic matter can be estimated using parameters of molecular distributions such as distributions of steranes, hopanes and polyaromatic hydrocarbons in petroleum.

Polyaromatic hydrocarbons are ubiquitous components of recent and ancient sediments. More than thirteen maturity parameters have been defined using phenanthrene homologues in Type III (terrigenous) kerogen samples, including MPI1, MPI2, MPI3, MPR, F1, F2, DPR, DPI, PAI, PP-1, MDR, PHI, log(MPs/P) (Alexander et al., 1986; Dolzhenko and Fomin, 2022; Kvaldheim et al., 1987; Radke et al., 1982a, 1982b; Radke, 1987; Radke et al., 1991; Stojanović et al., 2007; Wang et al., 2016). These parameters are based on conversion reactions resulting in the decrease of the unstable α-type isomers and the increase of the stable β-type isomers with increasing maturity (Kvaldheim et al., 1987; Radke et al., 1982a, 1982b). These phenanthrene parameters can be used across the entire oil generation window, whereas those parameters derived from saturated biomarker stereoisomers cannot (Kruge, 2000). Several equations for calculating maturity have been established using univariate regression analyses of the phenanthrene parameters. These regression equations have generally been applied to calculate %Ro for both type III and type I-II kerogens. However, they are not always effective in determining the maturity of organic matter in rocks and oils in many sedimentary systems, not only for Type III (Boreham et al., 1988) but also for Types I‒II organic matter (Chen et al., 2020; Yang et al., 2018).

Among the existing maturity parameters, the methyl-phenanthrene index 1 (MPI₁ = 1.5 × (2-MP + 3-MP)/(P + 1-MP + 9-MP); where P is phenanthrene, MP is methy-phenanthrene, 3-MP, 2-MP, 9-MP, and 1-MP stand for isomers methylated at the 3-, 2-, 9-, 1-position respectively; Radke et al., 1982a) and distribution fraction parameters (F₁ = (2-MP + 3-MP)/(2-MP + 3-MP + 1-MP + 9-MP) and F₂ = 2-MP/(2-MP + 3-MP + 1-MP + 9-MP); Kvaldheim et al., 1987), as well as their univariate regression equations, are most widely used in maturity determinations (Bao et al., 1992; Boreham et al., 1988; Chen et al., 2020; Luo et al., 2023; Sivan et al., 2008; Xiao et al., 2014; Yang et al., 2018; Zhang et al., 2019; Zhao et al., 2013). However, the maturities calculated using the univariate regression equations of MPI₁, F₁ and F₂ deviate from the actual data of many basins, including eight Australian basins (Type III kerogen; Boreham et al., 1988), the Middle Upper Rhine Graben in Canada (Type III kerogen; Radke et al., 1991), Songliao Basin (Type III kerogen), Bohai Bay Basin (Type II kerogen; Yang et al., 2018), and Yingen-Ejinaqi Basin in China (Type II kerogen; Chen et al., 2020). Even MPI₁ from the Duvernay Formation of the Western Canada Basin does not increase progressively with maturation (Type II kerogen; Requejo, 1994). These studies illustrate that the existing phenanthrene parameters cannot be directly applied in a specific basin. One of the reasons is probably that sedimentary and diagenetic environments may affect the conversion reactions of alkylated phenanthrenes (Hatem et al., 2023; Kumar et al., 2020; 2024a, 2024b, 2025a; Nilankar et al., 2024; Sidik et al., 2024; Singh and Kumar, 2020; Sun, 1998; Zhao et al., 2021, 2023, 2024; Zumberge et al., 2020). Therefore, the maturity parameters need re-calibration for each hydrocarbon system before using them (Dolzhenko and Fomin, 2022; Peters et al., 2005). However, in many areas, the squares of the correlation coefficients $(R^{2})$ between %Ro and these parameters are very low, such as 0.41 for F₁ from Type III kerogen in eight basins in Australia (Boreham et al., 1988), 0.54 for MPI₁, and 0.74 for F₂ from Type II kerogen in the marine carbonate rock in China (Bao et al., 1992). Therefore, these parameters and their univariate regression equations are not universally applicable (Srinivasan et al., 2022), as it is uncertain that the reliable determination of maturity levels can be obtained for either type III or type I-II organic matter.

Studies of organic matter maturity in the world have accumulated a mass of phenanthrene and alkyl-phenanthrene data. How to effectively use these big data for the maturity evaluation of organic matter is an important issue needing prompt solution. Moreover, if the issue in the application of phenanthrene and alkyl-phenanthrene data is addressed via exploration of a new approach, this approach may help to improve the use of the data of other alkylated aromatics for maturity study.

In basins, thermal maturity has different impacts on different aromatic compounds. Multivariate statistical analyses using multiple aromatic compounds may improve the reliability of the maturity calculations. Multivariate statistical methods, such as principal components analysis, nonlinear mapping, and discriminant analysis, have become important data analysis methods in petroleum geochemistry and geology (Budzinski et al., 1995; Kruge, 2000; Kvaldheim et al., 1987; Radke et al., 2000, 2001; Zhang, 1991; Zhang et al., 2014, 2019). However, partial least square regression (PLSR) analysis has not been applied to maturity calculations.

The aim of this study is first to examine the univariate relations of methyl- and dimethyl-phenanthrenes in extracts of rock samples containing Types I‒II kerogen, and then to investigate the multivariate relation of these phenanthrenes to measured maturity (%Ro) using PLSR analysis. In addition, the geochemical stabilities of these compounds, which control the conversion reactions, are integrated into the study. As a result, we suggest that when the previously-defined maturity parameters and their univariate regression equations are not effective, they be replaced by multivariate regression analyses of methyl- and dimethyl-phenanthrenes.

Samples and methods

Samples

The study area of this paper is the Songliao Basin, China. This basin developed mainly during the Early Cretaceous is a super petroliferous basin with high quality of nonmarine petroleum source rocks and considerable hydrocarbon potential. The Cretaceous Qingshankou Formation is one of the important layers of source rocks. This formation is further divided into the Qing-1, Qing-2 and Qing-3 Members from the bottom to top. The source rocks are mainly semi-deep to deep lacustrine sediments, dominated by gray-black shales with black mudstone (Zhou and Littke, 1999). They contain predominantly Type I kerogen with minor Type II₁, and the total organic carbon (TOC) of 1.5wt%～2.5wt% with the maturity Ro from 0.5% to 1.7%.

To examine the effectiveness of the previously-defined parameters, the samples should be restricted in one formation in one basin, with little variation in organic matter type. In this study, thirty-seven core samples (100 g for each) were collected from the Qing-1, Qing-2 and Qing-3 Members in the Cretaceous Qingshankou Formation of eleven wells in the Songliao Basin (Table 1). These wells are vertical and were drilled with water-based mud systems. The core samples are conventional, not sidewall, and were taken from Members Qing-1, -2, and -3 in the Cretaceous Qingshankou Formation in this basin, with the depth from 1474.57 to 2419.54 m. Their specific depths, horizon and lithology are listed in Table 1.

Table 1.

Geological information of the samples in this study.

No.	Well	Vertical depth (m)	Member^a	Lithology
1	S2LC	2278.25	K₁qn₂	Gray black shale
2	S2LC	2298.29	K₁qn₁	Gray black shale
3	S2LC	2333.36	K₁qn₁	Gray black shale
4	QL1	2024.55	K₁qn₂ ₊ ₃	Black mudstone
5	QL1	2118.85	K₁qn₁	Black mudstone
6	S3LC	2392.68	K₁qn₂	Laminated shale
7	S3LC	2402.02	K₁qn₂ ₊ ₃	Laminated shale
8	S3LC	2408.93	K₁qn₂	Laminated shale
9	S3LC	2419.54	K₁qn₂ ₊ ₃	Laminated shale
10	SSO3	1980	K₁qn₁	Gray black shale
11	SSO3	2025	K₁qn₁	Black shale
12	SSO3	2056.6	K₁qn₁	Black dolomitic shale
13	S8LC	2319.1	k₁qn₂ ₊ ₃	Gray black shale
14	S8LC	2333.1	k₁qn₂ ₊ ₃	Gray black shale
15	S8LC	2353.52	k₁qn₂ ₊ ₃	Gray black shale
16	S8LC	2365.27	k₁qn₂ ₊ ₃	Gray black shale
17	S8LC	2377.2	k₁qn₂ ₊ ₃	Gray black shale
18	PI85	1880.88	K₁qn₂ ₊ ₃	Black-gray laminated shale
19	PI85	1897.28	K₁qn₂ ₊ ₃	Black-gray laminated shale
20	PI85	1919.11	K₁qn₂ ₊ ₃	Black-gray laminated shale
21	PI85	1943.17	K₁qn₁	Black-gray laminated shale
22	PI85	1956.29	K₁qn₁	Black-gray laminated shale
23	B262-I176	2271.76	K₁qn₂ ₊ ₃	Gray-black mudstone
24	S5LC	2100.84	K₁qn₂ ₊ ₃	Dark-gray mudstone
25	S5LC	2134.27	K₁qn₂ ₊ ₃	Dark-gray mudstone
26	S5LC	2197.09	K₁qn₁	Black-gray laminated shale
27	S5LC	2216.8	K₁qn₁	Black-gray laminated shale
28	S5LC	2236.48	K₁qn₁	Black-gray laminated shale
29	C21	1474.57	K₁qn₂ ₊ ₃	Black mudstone
30	C21	1525.83	K₁qn₂ ₊ ₃	Black mudstone
31	C21	1575.77	K₁qn₂ ₊ ₃	Gray-black mudstone
32	C21	1600.83	K₁qn₁	Black mudstone
33	C21	1625.93	K₁qn₁	Black mudstone
34	C21	1650.98	K₁qn₁	Black mudstone
35	C21	1669.85	K₁qn₁	Black mudstone
36	S36	2106.25	K₁qn₂ ₊ ₃	Gray-black laminated shale
37	S36	2200.5	K₁qn₁	Gray-black laminated shale

K₁qn₁: Qing-1 Member of the Cretaceous Qingshankou Formation; K₁qn₂: Qing-2 Member; K₁qn₂ ₊ ₃: Qing-2 and Qing-3 Members.

Measurements of vitrinite reflectance and phenanthrenes

Reflectance measurements (random, oil immersion) were tested under an oil immersion objective lens of 50× magnification on a Leica DM4500P microscope coupled with CRAIC QDI302 microscope spectrophotometer. The vitrinite reflectance values were calibrated with the reference materials including yttrium aluminium garnet (GBW13402, Ro = 0.90%), and sapphire (GBW13403, Ro = 0.59%). The number of measurements in each sample varies from 30 to 60 to obtain sufficient measurements (Tu et al., 2012; Xiao et al., 2008a). The standard deviations of vitrinite reflectance for individual samples are from 0.03% to 0.06% with the average of 0.05% (see details in the Results section).

The individual core samples were crushed to 100 mesh and subsequently homogenized. The soluble organic matter in each sample was extracted using Soxhlet method (Hu et al., 2021). The aromatic fractions of the extracts were analyzed by gas chromatography (GC, Type Agilent 7890A) using an HP-5MS capillary column with helium as the carrier gas. Carrier gas flow rate is 15–25 cm/s. The vaporizer temperature is 310°C (Xiao et al., 2008b). The relative abundances of the alkylated phenanthrenes were quantified using the normalized areas of the peaks as shown in Figure 1. The analytical uncertainties of the normalized abundances are less than 1.3%.

Figure 1.

Gas chromatogram of phenanthrene and methyl- and dimethyl-phenanthrenes: sample at 2333.1 m of well S8LC (Tables 1 and 2).

Table 2.

Measured Ro, normalized abundances of six phenanthrene isomers, and maturity parameters of samples from the Cretaceous Qingshankou Formation in the Songliao Basin, China.^a

No.	Ro (%)	Ro SD （%）	3-MP (%)	2-MP (%)	9-MP (%)	1-MP (%)	2,7-DMP (%)	1,7-DMP (%)	MPI₁	F₁	F₂
1	1.26	0.04	10.82	12.82	15.61	14.70	14.97	31.08	1.11	0.44	0.24
2	1.28	0.04	8.36	15.14	12.90	15.14	18.02	30.45	1.19	0.46	0.29
3	1.35	0.03	9.71	15.89	12.30	14.44	15.98	31.68	1.39	0.49	0.30
4	1.00	0.05	16.13	16.60	28.80	22.33	4.23	11.90	0.76	0.39	0.20
5	1.07	0.04	14.60	19.33	25.60	22.37	5.57	12.53	0.81	0.41	0.24
6	1.30	0.06	10.11	14.84	12.37	12.72	17.12	32.85	1.43	0.50	0.30
7	1.33	0.05	12.41	20.13	15.19	16.63	13.28	22.36	1.40	0.51	0.31
8	1.34	0.04	7.21	9.35	8.06	7.21	16.40	51.77	0.54	0.52	0.29
9	1.35	0.04	14.33	21.01	16.63	17.53	11.29	19.20	1.39	0.51	0.30
10	0.85	0.05	13.93	14.14	28.09	24.05	2.94	16.84	0.69	0.35	0.18
11	0.92	0.05	12.69	14.08	26.01	19.49	4.94	22.79	0.85	0.37	0.19
12	0.94	0.06	9.10	11.91	25.55	18.69	6.22	28.53	0.69	0.32	0.18
13	1.25	0.04	14.78	20.90	19.17	18.88	8.07	18.21	1.20	0.48	0.28
14	1.27	0.04	16.59	22.54	18.39	18.17	8.10	16.21	1.35	0.52	0.30
15	1.30	0.04	14.48	21.22	17.33	17.50	9.92	19.55	1.44	0.51	0.30
16	1.32	0.05	15.06	21.84	16.42	17.14	10.61	18.93	1.52	0.52	0.31
17	1.34	0.05	11.28	16.75	14.29	17.87	10.23	29.57	1.23	0.47	0.28
18	0.98	0.06	13.45	16.88	26.64	21.13	5.06	16.84	0.86	0.39	0.22
19	1.01	0.04	14.32	18.15	24.42	21.53	5.41	16.17	0.93	0.41	0.23
20	1.03	0.05	14.66	18.25	25.79	20.19	4.77	16.34	0.93	0.42	0.23
21	1.05	0.04	13.46	18.81	22.88	20.65	6.68	17.52	1.02	0.43	0.25
22	1.06	0.04	14.11	19.27	23.48	19.73	6.90	16.51	1.04	0.44	0.25
23	1.14	0.04	15.52	20.68	21.27	22.14	5.76	14.63	0.97	0.45	0.26
24	1.12	0.06	17.33	19.07	28.28	19.49	4.91	10.92	0.87	0.43	0.23
25	1.15	0.06	13.83	18.31	22.56	19.92	7.13	18.25	1.01	0.43	0.25
26	1.23	0.03	11.42	16.30	19.03	16.59	11.76	24.90	1.14	0.44	0.26
27	1.25	0.06	12.76	18.45	17.99	16.45	12.15	22.20	1.30	0.48	0.28
28	1.28	0.05	12.88	18.98	19.39	16.08	12.29	20.38	0.89	0.47	0.28
29	0.68	0.04	14.64	14.98	32.31	25.61	2.03	10.44	0.54	0.34	0.17
30	0.72	0.06	14.35	13.56	31.68	24.81	2.34	13.27	0.52	0.33	0.16
31	0.76	0.06	15.07	15.69	31.76	22.05	2.04	13.39	0.54	0.36	0.19
32	0.78	0.06	14.29	15.24	30.85	20.63	2.86	16.13	0.60	0.36	0.19
33	0.80	0.06	13.49	15.40	30.12	22.00	2.96	16.02	0.62	0.36	0.19
34	0.83	0.05	13.93	15.02	30.41	20.06	2.47	18.11	0.63	0.36	0.19
35	0.85	0.04	13.83	16.30	27.11	21.81	2.81	18.14	0.66	0.38	0.21
36	1.02	0.06	17.12	21.07	26.31	20.52	6.29	8.69	0.74	0.45	0.25
37	1.14	0.05	8.95	14.26	18.21	15.45	14.54	28.61	1.02	0.41	0.25
Min.	0.68	0.03	7.21	9.35	8.06	7.21	2.03	8.69	1.11	0.44	0.24
Mean	1.09	0.05	13.27	17.11	22.25	18.96	8.08	20.32	1.19	0.46	0.29
Max.	1.35	0.06	17.33	22.54	32.31	25.61	18.02	51.77	1.39	0.49	0.30

The data used to calculate the maturity parameters are listed in Supplementary Table S1. SD = Standard Deviation; MPI₁ = 1.5 × (2-MP + 3-MP)/(P + 1-MP + 9-MP); P is phenanthrene, MP is methyl-phenanthrene, 3-MP, 2-MP, 9-MP, and 1-MP stand for the respective methyl-phenanthrene isomers, methylated at the 3-, 2-, 9-, 1-position (Radke et al., 1982a). F₁ = (2-MP + 3-MP)/(2-MP + 3-MP + 1-MP + 9-MP); F₂ = 2-MP/(2-MP + 3-MP + 1-MP + 9-MP) (Kvaldheim et al., 1987).

Partial least square regression

Some methyl- and dimethyl-phenanthrenes are correlated to each other (see Results), resulting in the multicollinearity problem that influences multivariate regression analyses and must be solved (Wold et al., 1984, 2001). PLSR is superior to ordinary least square regression in solving this problem (Wang et al., 2006; Wang et al., 2024; Wold et al., 1984, 2001). Both PLSR and principal component analysis (PCA) construct components. PLSR specifically optimizes these components to maximize their correlation with the response variable (e.g., Ro), but PCA cannot. Therefore, the multivariate data analysis software of SIMCA-P (version 11.5) was employed for the PLSR analysis in this study to establish a multivariate regression equation for maturity prediction.

The data for each variable, including Ro, methyl-phenanthrenes (MPs) and dimethyl-phenanthrenes (DMPs), are standardized by subtracting the mean and dividing by the standard deviation. Using the standardized data of descriptors (MPs and DMPs), PLSR analysis establishes the components that aim to capture all available descriptor information while ensuring high correlation to the response (Ro).

After computing the components, the appropriate number of them needs to be determined mainly by using the evaluation parameter $Q^{2}$ computed from leave-one-out cross-validation (Wakeling and Morris, 1993; Wang et al., 2006; Wold, 1978). Then, the regression equation of a standardized response (i.e., standardized Ro) versus the components can be established. From this equation and the equations of the appropriate components versus the descriptors, the multivariate regression equation for calculating the response with the descriptors can be achieved (see details in Appendix A).

A multivariate regression equation needs to be validated using the goodness of fit between measured Ro and calculated Ro (Rc) after establishing. Commonly, the square of the correlation coefficient (R²) between Ro and Rc is used to evaluate a regression equation. Since the samples used to calculate Ro are all involved in building the regression equation, R² mainly reflects regression accuracy but cannot effectively assess its predictive performance on new data. To evaluate the predictive ability of a multivariate regression equation, a cross-validation method in multivariate regression analysis is applied. In this method, Rc for each sample is predicted using a regression equation constructed from all the other samples, not from all the samples. The R² between the measured Ro and the predicted Rc is then calculated and represented by R_p². R_p² derived from this cross-validation method can be used to evaluate the predictive ability of a multivariate regression equation and in this way to validate the equation (Wang et al., 2006).

A multivariate regression equation can also be validated through descriptors. The cumulative proportion of the deviation (relative to the mean) squares of the entire descriptors explained by the model components, denoted as R²X(cum). It can be used to examine the extent to which the descriptor data are utilized and to help to evaluate the effectiveness of a multivariate regression equation (Eriksson et al., 2013). Generally, the higher R²X(cum) is, the better a multivariate regression equation is.

In PLSR, the variable importance in projection (VIP) of a descriptor to the response is computed from the weights of descriptors in components and correlation coefficients of components with response, and quantitatively describes the importance of a descriptor for the response (Ji et al., 2011; Wang et al., 2006). The larger the VIP value, the greater the importance of a descriptor for the response. The descriptors with VIP values larger than one show the above-average importance for predicting the response and are deemed to be relatively important descriptors (Favilla et al., 2013; Wang et al., 2006).

Results

Maturity of the samples

The test results of Ro values of our samples range from 0.68% to 1.35% with a mean value of 1.09% (Table 2), indicating that the organic matter in the samples entered the oil window (Tissot et al., 1987; Mukhopadhyay, 1994). Therefore, the maturity of the organic matter in these samples is suitable for the study of the parameters and methods for maturity calculation of organic matter.

Normalized methyl- and dimethyl-phenanthrenes

The relations between phenanthrenes and maturity were analyzed with the normalized data of relative abundances of phenanthrenes obtained from GC analyses (Kvaldheim et al., 1987; Radke et al., 1982a). In the early studies, phenanthrene was used in the normalization, such as in MPI₁ and MPI₂ (Radke et al., 1982a). However, the subsequent studies found that the linear correlation between Ro and normalized methylated phenanthrenes is weakened if phenanthrene is included in the normalization (Kvaldheim et al., 1987; Telnaes et al., 1987). Therefore, subsequent parameters such as F₁ and F₂ do not use phenanthrene (Kvaldheim et al., 1987; Telnaes et al., 1987). The normalization of methylated phenanthrenes also excludes phenanthrene in this paper. On the other hand, individual peaks containing two or more dimethyl-phenanthrene isomers (Figure 1 and Table 3) were not used either. The reason is that these peaks reflect mixtures of these isomers with different thermal stabilities (see details in Budzinsk et al., 1993). Consequently, the methyl- and dimethyl-phenanthrenes used in this work include 3-MP, 2-MP, 9-MP, 1-MP, 2,7-DMP, and 1,7-DMP. The normalized results of these six compounds are listed in Table 2.

Table 3.

Gc peaks of phenanthrene and its homologs identified from gas chromatogram.^a

Peak number	compounds	Peak number	Compounds
1	P	7	DMP2 (3,6-DMP with 9-, 2- and 1-Et)
2	3-MP	8	DMP3 (2,6- and 3,5-DMP)
3	2-MP	9	DMP4 (2,7-DMP)
4	9-MP	10	DMP5 (2,10-, 1,3-, 3,10- and 3,9-DMP)
5	1-MP	11	DMP6 (1,6- and 2,9-DMP)
6	DMP1(9-, 2- and 1-Et with 3,6-DMP)	12	DMP7 (1,7-DMP)

P: phenanthrene; MP: methyl-phenanthrene; DMP: dimethyl-phenanthrene. Nine dimethyl-phenanthrenes cannot be separated under the conditions of the GC analyses.

The normalized methyl- and dimethyl-phenanthrenes abundances vary significantly. 9-MP ranges from 8.06% to 32.31% with an average of 22.25%, and 1-MP from 7.21% to 25.61% with an average of 18.96%. Their means exceed those of 2- and 3-MPs (β-type) (Table 2 and Figure 2). This phenomenon was also observed in the early stage of thermal evolution (Li et al., 2022; Song et al., 2007).

Figure 2.

Distribution of the average values of the normalized methylated phenanthrene contents in the samples from the Cretaceous Qingshankou Formation in the Songliao Basin, China.

The normalized abundance of 1,7-DMP ranges from 8.68% to 57.77%, and that of 2,7-DMP from 2.03% to 18.02% with the average at 8.08%. The average value of 2,7-DMP (ββ-type) is lower than that of 1,7-DMP (αβ-type; Table 2 and Figure 2), which was also observed in the Mahakam Delta in Indonesia (Budzinski et al., 1993) and the Keras Basin in Indonesia (Budzinski et al.,1995).

Correlations between phenanthrenes

2- and 3-MPs belong to β-type and are correlated to each other (Figure 3(a)). 1- and 9-MPs are α-type isomers and are correlated too (Figure 3(b)). However, 1-MP (α-type) is not correlated to 2-MP (β-type), and 9-MP (α-type) not to 3-MP (β-type; Figure 3(c) and (d)). Although α- and β-types of methyl phenanthrenes can form via methylation of phenanthrene and conversion reactions among methyl phenanthrenes, more β-type methyl phenanthrenes can derive from the conversion reactions than α-type ones (Alexander et al., 1995; Behar et al., 1999; Kvaldheim et al., 1987; Nomoto et al., 2000; Radke et al., 1982a, 1982b; Smith et al., 1995; Voigtmann et al., 1994). This is probably the reason why α-type methyl phenanthrenes are not correlated to β-type methyl phenanthrenes.

Figure 3.

Cross plots of methyl- and di-methyl phenanthrenes in the samples from the Cretaceous Qingshankou Formation in the Songliao Basin, China.

ββ-type isomers are the most stable among all the types of DMPs. 2,7-DMP is a ββ-type isomer. Although 1,7-DMP is αβ-type, it is the most stable among αβ-type and αα-type isomers (Budzinski et al., 1993) and can be a product of the conversion reactions among methylated phenanthrenes (Behar et al., 1999). Figure 3(e) shows that 1,7- and 2,7-DMPs are roughly correlated to each other. In Figure 3(f), 3-MP (β-type) and 2,7-DMP (ββ-type) are roughly correlated possibly because they are all stable compounds.

Discussion

Univariate relations of maturity parameters to Ro

One of the widely used maturity parameters is MPI₁ (MPI₁ = 1.5 × (2-MP + 3-MP)/(P + 1-MP + 9-MP)), as mentioned in the Introduction. The vitrinite reflectance has been calculated using the univariate regression equation established with the measured Ro and MPI₁ values of 16 samples of type III kerogen from the Western Canada Basin (Radke et al., 1982a). This equation is shown in Figure 4(a). However, it does not accurately predict the actual Ro values of the samples of Types I‒II₁ kerogen from the Cretaceous Qingshankou Formation in the Songliao Basin, China (Figure 4(a)). The difference between the equation and actual data was also observed in the Bohai Bay Basin in China (Type-II kerogen; Yang et al., 2018), eight Australian basins (Type-III kerogen; Boreham et al., 1988), the Canada middle Upper Rhine Graben (Type-III kerogen; Radke et al., 1991) and Songliao Basin in China (Type III kerogen). The difference between the equation and actual data is probably due to source control on phenanthrene distributions (Radke et al., 1986). Therefore, the univariate regression equation of MPI₁ (Radke et al., 1982a) is definitely not effective in any basins, and cannot be directly used before it is validated.

Figure 4.

Relations of MPI1(a), F1 (b), and F2 (c) to Ro of the samples from the Cretaceous Qingshankou Formation in the Songliao Basin, China. MPI₁ = 1.5 × (2-MP + 3-MP)/(P + 1-MP + 9-MP), F₁ = (2-MP + 3-MP)/(2-MP + 3-MP + 1-MP + 9-MP) and F₂ = 2-MP/(2-MP + 3-MP + 1-MP + 9-MP). The blue dashed lines represent the maturity regression equations in Radke et al. (1982a) and Kvaldheim et al. (1987), while the red solid lines are defined by refitting new equations to the data.

The difference between the equation and actual data may be solved by refitting the data. However, the square of the correlation coefficient obtained from refitting the data from the Cretaceous Qingshankou Formation in the Songliao Basin is only 0.69 (Figure 4(a)). The low correlations between maturity and MPI₁ were also reported by studies for type III kerogen in eight Australian basins ( $R^{2}$ = .71; Boreham et al., 1988), Canada middle Upper Rhine Graben ( $R^{2} = .44$ ; Radke et al., 1991) and Songliao Basin ( $R^{2} = .45$ ), and for type II kerogen in Chinese marine carbonate rock ( $R^{2} = .54$ ; Bao et al., 1992). Figure 5(a) and (b) shows that Ro is significantly correlated to the depth $(R^{2} = .92)$ , but MPI₁ is not $(R^{2} = .63)$ . Therefore, MPI₁ may be flawed.

Figure 5.

Relations of Ro(a), MPI1 (b), F1 (c), and F2 (d) to the vertical depth of the samples from the Cretaceous Qingshankou Formation in the Songliao Basin, China. The definitions of MPI₁, F₁ and F₂ in the caption of Figure 4.

To improve the calculation of maturity (%Ro), more parameters were proposed (Alexander et al., 1986; Dolzhenko and Fomin, 2022; Kvaldheim et al., 1987; Radke, 1987; Radke et al., 1982b; Radke et al., 1991; Stojanović et al., 2007; Wang et al., 2016). In the maturity parameters of phenanthrenes proposed after MPI₁, the distribution fractionation parameters F₁ and F₂ are representative [F₁ = (2-MP + 3-MP)/(2-MP + 3-MP + 1-MP + 9-MP) and F₂ = 2-MP/(2-MP + 3-MP + 1-MP + 9-MP); Kvaldheim et al., 1987; Telnaes et al., 1987]. As mentioned in the Results, phenanthrene was excluded in these two parameters. Two univariate regression equations for maturity calculation were set up with F₁ and F₂, respectively, using 15 samples with Type-III kerogen from England, Norway, West Africa, and other regions (Kvaldheim et al., 1987). However, these two equations still do not accurately predict the Ro values of the samples from the Cretaceous Qingshankou Formation in the Songliao Basin, China (Figure 4(b) and (c)). Similar lack of accuracy was also reported in the studies of type II kerogen from the Yingen-Ejinaqi Basin and marine carbonate rock in China (Bao et al., 1992; Chen et al., 2020). Therefore, the univariate regression equations of F₁ and F₂ are not effective in any basins, and also cannot be directly used before they are validated.

The data from the Cretaceous Qingshankou Formation in the Songliao Basin were refitted also with F₁ and F₂, Figure 4(b) and (c) indicate that by excluding phenanthrene in the normalization of phenanthrenes, the correlations of Ro with methylated phenanthrenes were improved. However, the squares of the correlation coefficients are still lower than 0.9. Poor correlations of these two parameters with Ro are also documented in the studies for both type II ( $R^{2} = .78$ for F₁and $R^{2} = .74$ for F₂; Bao et al., 1992) and Type-III kerogens ( $R^{2} = .41$ for F₁; Boreham et al., 1988). In Figure 5(c) and (d), correlation coefficients of these two parameters with the depth are also low ( $R^{2} = .73$ for F₁ and $R^{2} = .77$ for F₂). Although F₁ and F₂ are better than MPI₁, the problems with the maturity parameters have not been completely solved. In this study, other previously-defined maturity parameters (Alexander et al., 1986; Dolzhenko and Fomin, 2022; Kvaldheim et al., 1987; Radke, 1987; Radke et al., 1982a, 1982b; Radke et al., 1991; Stojanović et al., 2007; Wang et al., 2016) were also tried, including MPI₂(mod) [MPI₂(mod) = 1.89 × (2-MP + 3-MP)/(P + 1.26 × (1-MP + 9-MP)], MPI₂ [MPI₂ = 3 × 2-MP/(P + 1-MP + 9-MP); $R^{2}$ =.72], MPI₃ [MPI₃ = (2-MP + 3-MP)/(1-MP + 9-MP); $R^{2} = .83$ ], MPR (MPR = 2-MP/1-MP; $R^{2} = .79$ ), DPR [DPR = (DMP3 + DMP4)/(DMP5 + DMP6); $R^{2} = .69$ ], DPI [DPI = 4 × (DMP1 + DMP2 + DMP3 + DMP4)/(P + DMP5 + DMP6 + DMP7), $R^{2} = .44$ ], PAI-1 [PAI-1 = (2-MP + 3-MP + 1-MP + 9-MP)/P; $R^{2} = .09$ ), PP-1(PP-1 = 1-MP/(2-MP + 3-MP); $R^{2} = .69$ ], MDR (MDR=∑MP/∑DMP; $R^{2} = .65$ ), PHI (PHI = 2-MP/P; $R^{2} = .14$ ), log(MPs/P) (log(MPs/P)= log[(2-MP + 3-MP + 1-MP + 9-MP)/P), $R^{2} = .22$ ]. Their data are listed in Supplemental Table S1. Although the maturity parameters have been continuously studied with phenanthrenes, the significant correlations of them with Ro cannot be achieved in the Songliao Basin, as shown in Supplemental Figure S1.

The data set of phenanthrene and methylated phenanthrenes in this study is restricted in one formation in one basin, with the same organic matter type. The reason why the existing maturity parameters cannot work well may not be due to source influence. It may lie in their definitions and methods for their application. They were mainly defined as ratios of stable compound(s) to unstable compound(s) or to the sum of unstable compounds and stable compound(s) (Alexander et al., 1986; Kvaldheim et al., 1987; Radke, 1987; Radke et al., 1982a, 1982b; Srinivasan et al., 2022; Zumberge et al., 2020). The relations between Ro and the parameters were established with univariate regression analyses. This kind of univariate methods can only use the maturity information carried by the ratios of the finite compounds (denominators and numerators). The maturity information carried by more compounds cannot be utilized synthetically. If the maturity information carried by the finite compounds cannot sufficiently reflect maturity, these ratio parameters cannot work well. This is probably the reason why the univariate regression equations cannot achieve significant correlations of maturity with the previously defined maturity parameters, in the Songliao Basin and many other basins. Consequently, it is uncertain that the existing parameters and the equations obtained from univariate regression analyses can be successfully applied to maturity calculation for all kerogen types.

Multivariate relation of phenanthrenes to Ro

Multivariate regression analysis can simultaneously exploit more information contained in the distributions of multiple phenanthrenes to characterize Ro, compared with univariate regression analysis. In the PLSR analysis of the data from the Cretaceous Qingshankou Formation in the Songliao Basin, the six phenanthrenes were used as descriptors, and the measured Ro as the response. After standardizing the data, the components were computed (see Appendix B). To determine the appropriate number of the components, their Q² values (Table S2 in the Supplementary Data) were calculated and plotted in Figure 6(a). When the first two components are selected, Q²_cum (Table S2 in the Supplemental Data) becomes maximal and reaches 0.94 (Figure 6(b)), indicating that the regression equation with two components is both reliable and robust with excellent predictive ability (see details in Appendix A, Eriksson et al. (2013)). Consequently, the number of the appropriate components was determined to be two.

Figure 6.

Q2 (a) and Q²(cum) (b) of components in PLSR with the measured Ro and six phenanthrenes from the Cretaceous Qingshankou Formation in the Songliao Basin, China.

The regression equation of the two components for the standardized Ro was computed (see Appendix B). From this regression equation and the equations for the first two components, the equation for calculating thermal maturity with the normalized abundances of the six phenanthrenes was obtained (see details in Appendix B):

\begin{aligned} R_{c} = & 0.0049 \times [3 - M P] + 0.0260 \times [2 - M P] - 0.0119 \\ \times [9 - M P] - 0.0138 \times [1 - M P] + 0.0131 \\ \times [2, 7 - D M P] + 0.0016 \times [1, 7 - D M P] + 0.97 \end{aligned}

(1)

where Rc is the calculated Ro; $[3 - M P]$ , $[2 - M P]$ , $[9 - M P]$ , $[1 - M P]$ , $[2, 7 - D M P]$ , and $[1, 7 - D M P]$ represent their normalized abundances.

The multivariate regression equation (1) was validated following the procedure described in Section Partial least square regression. Firstly, Figure 7 shows a cross-plot of measured Ro values against Rc values calculated using equation (1). The R² value is 0.94, indicating that the regression equation is accurate. Secondly, a cross-validation method was applied to the 37 samples. The Rp² obtained from this cross-validation is 0.93, and suggests that the multivariate regression equation is reliable with strong predictive ability when applied to new data excluded in building the regression. Thirdly, from the perspective of variable descriptors, the R²X(cum) value of the multivariate regression equation (1) were computed. It reaches 0.95, indicating that equation (1) captures most (95%) of the variance information from the six phenanthrene parameters. All the high values of R², Rp² and R²X(cum) manifest that the multivariate regression equation (1) can effectively utilize the information contained in MPs and DMPs, toreliablly predictthermal maturity within the oil window (0.68%‒1.35% Ro).

Figure 7.

Correlation of the measured Ro with the Rc calculated with PLSR using six phenanthrenes in the samples from the Cretaceous Qingshankou Formation in the Songliao Basin, China.

Moreover, the relative importance of the MPs and DMPs for maturity prediction and their relationships with geochemical stabilities were analyzed. In PLSR, the VIP (variable importance in the projection) index reflects the relative importance of the descriptors in the predictions (see Appendix A). The VIP values of the descriptors in the regression equation computed from the standardized data are listed in Table 4. If VIP ≥1, the corresponding descriptor is significant in predicting the response (Eriksson et al., 2013; Mahieu et al., 2023; Wang et al., 2006; Wold et al., 1993). The VIP values of 2,7-DMP, 9-MP, and 1-MP exceed one (Table 4 and Figure 8), and that of 2-MP is close to one, which indicate that these four phenanthrenes are significant variables for predicting the maturity. In contrast, the VIP values of 1,7-DMP and 3-MP are much less than one. Their contribution to the maturity prediction in the Songliao Basin is less significant. These VIP values of phenanthrenes are corresponding to their stabilities (see the next subsection), which also supports that the multivariate regression equation (1) is effective and reliable.

Figure 8.

Values of the VIPs of the phenanthrenes in the multivariate equation established with the samples from the Cretaceous Qingshankou Formation in the Songliao Basin, China.

Table 4.

The VIPs in the regression equation computed from the standardized data.

Var.ID	3-MP	2-MP	9-MP	1-MP	2,7-DMP	1,7-DMP
VIP	0.73	0.97	1.24	1.03	1.13	0.82

Thermal stabilities and effects of phenanthrenes in the model

The variations in normalized abundances of alkyl-phenanthrenes with increasing maturity are mainly controlled by their thermal (thermodynamic) stabilities, characterized by increases in stable (e.g., β type) isomers and decreases in unstable (e.g., α type) isomers. This pattern likely results from rearrangement reactions of methyl groups from α-positions to β-positions to reduce steric hindrance (Budzinski et al., 1993; Garrigues et al., 1990; Radke et al., 1982b). As shown in equation (1), Rc is correlated positively to stable 2- and 3-MP (β-type) and negatively to unstable 9- and 1-MP (α-type), which corresponds to the stability trends.

For dimethyl-phenanthrenes, ββ-type isomers are the most stable. Correspondingly, 2,7-DMP (ββ-type) has both a high VIP value (Table 4) and a positive coefficient in equation (1). 1,7-DMP is the most stable isomers among the αβ- and αα-type isomers (Budzinski et al., 1993). As a result, 1,7-DMP possesses a positive coefficient. But its VIP value is lower than that of 2,7-DMP due to the higher stability of the latter. All these results coincide with the thermodynamic stabilities of both compounds.

As mentioned above, the studies of phenanthrene parameters for maturity calculation have focused primarily on methyl-phenanthrenes. Although a few of dimethyl-phenanthrene parameters (e.g., DPI and DPR; Radke et al., 1982a, b) were proposed, they have not been widely used. Notably, the VIP value of the 2,7-DMP (the most stable DMP) exceeds that of 2-MP (the most stable MP, Behar et al., 1999). This seemingly indicates that effective parameters for maturity can be established with dimethyl-phenanthrenes. But in numerous studies (Garrigues et al., 1988; Radke et al., 1986; Wang et al., 2020; Zdravkov et al., 2023), data of most dimethyl-phenanthrenes, especially αα-type isomers, have not been accumulated, as they cannot be identified and tested from gas chromatograms. The lack of sufficient data has negatively impacted the development of reliable DMP parameters for maturity calculation. However, integrated use of methyl- and dimethyl-phenanthrenes can compensate for the limitations of the methods using methyl- and dimethyl-phenanthrenes separately.

Dimethyl-phenanthrenes as potential maturity variables

As mentioned in Subsection Thermal stabilities and effects of phenanthrenes in the model, dimethyl-phenanthrenes have been underutilized in maturity assessment compared with methyl-phenanthrenes. This study demonstrates that incorporating dimethyl-phenanthrenes (1, 7- and 2,7-DMPs) into a multivariate regression equation can significantly improve maturity prediction (see Subsection Multivariate relation of phenanthrenes to Ro). DMPs have a greater number (25) of isomers than MPs. The differences in stability among the different isomers (Budzinski et al., 1993) give DMPs the potential to characterize thermal maturity.

However, the application of dimethyl-phenanthrenes has been limited due to they cannot be sufficiently separated by conventional GC analysis. Subsequantly-developed two-dimensional gas chromatography (GC × GC) may separate them sufficiently and may quantitatively analyze more dimethyl-phenanthrene isomers. Its combination with vacuum ultraviolet spectroscopy (VUV) has shown that it can effectively distinguish aromatic hydrocarbon isomers. But, this combination remains rarely applied in maturity research. With the advancement of the method for DMP analysis, big data of dimethyl-phenanthrenes can be derived. These data can more sufficiently remedy the limitations of MPs. The integrated application of MPs and DMPs with the PLSR method in this paper will become more accurate and effective at maturity assessment.

Implications of multivariate analysis for maturity evaluation

The thermal evolution of buried organic matter is an important section of the organic carbon cycle. If deeply-buried rocks with organic matter cannot be sampled, thermal evolution levels cannot be determined with rock samples. Nevertheless, maturity calculation can help to solve this issue. In the petroleum industry, the maturity calculation of discovered oils can also help to relate the oils to their source rocks. However, it is not suggested to directly use the multivariate regression equation (1) for maturity calculation in the other areas. It is suggested that this equation be validated with the actually data from the other areas and even multivariate regression equations be set up by using the PLSR approach established in this study.

Recently, many aromatic hydrocarbons can be tested by using GC-MS or GC-MS-MS analysis (GC: gas chromatography; MS: mass spectrometry). For example, alkyl aromatics with 9–19 carbons haven been tested. A series of alkyl-substituted naphthalene ratios were constructed for maturity estimation (Wang et al., 2021; Zumberge et al., 2020). In addition, a huge number of polycyclic and heterocyclic aromatic compounds measured using Fourier transform-ion cyclotron resonance-mass spectrometry (FT-ICR-MS). The corresponding maturity parameters were defined as the ratios of the lighter compounds with higher double bond equivalents (C_20–50) to the heavier (C_30–60) (Noah et al., 2020). All these maturity parameters are defined as the ratios of stable compound(s) to unstable compound(s) or to the sum of unstable compound(s) and stable compound(s). It has been illustrated in the first subsection of Discussion that this kind of parameters and their univariate regression equations cannot sufficiently utilize the maturity information carried by these compounds. Comparatively, PLSR can utilize the maturity information carried by these compounds more sufficiently and synthetically. Therefore, PLSR may become a practical approach for effective use of plenty of these data to evaluate organic matter maturity.

Furthermore, this multivariate approach may improve the maturity prediction across different kerogen types. This needs to be demonstrated with its application to the areas where type III kerogen is developed. But the shale and mudstone in the Qingshankou Formation mainly contains type I kerogen, with a small amount of type II₁ kerogen. It is suggested that the PLSR approach be experimentally applied to the data obtained from type III organic matter.

Conclusion

The existing maturity parameters defined with phenanthrenes and their univariate regression equations can only use the maturity information carried by finite compounds. The multivariate regression equation established with PLSR can comprehensively utilize the maturity information carried by more compounds including methyl and dimethyl phenanthrenes. This multivariate equation overcomes the limitations of univariate equations of the existing maturity parameters and thus enhances maturity calculations. It utilizes 95% of the variance information of methylated phenanthrenes, and its R² is .94.

The coefficients of the six phenanthrenes in the multivariate regression equation correspond well with their geochemical stabilities. The stable phenanthrenes have positive coefficients and unstable ones have negative coefficients. The most significant methylated phenanthrenes for maturity calculation are (in order) 9- and 1-MP (α-type), 2,7-DMP (ββ-type) and 2-MP (β-type).

The multivariate regression equation of maturity with methyl- and dimethyl-phenanthrenes outperforms univariate regression ones of the previously defined parameters. For the application in other basins, it is suggested that this multivariate regression equation be examined with the local data. If R² < .9, a new multivariate regression equation needs to be set up. The PLSR-based approach developed in this study is demonstrated to be a powerful tool for constructing multivariate regression equations with of methyl- and dimethyl-phenanthrenes. Furthermore, if maturity for Types I and II₁ kerogens needs to be estimated with other aromatic hydrocarbons, and if maturity for the other types of kerogen needs to be estimated, this approach is recommended for establishing multivariate regression equations.

Supplemental Material

sj-docx-1-eea-10.1177_01445987251408084 - Supplemental material for Multivariate relationship of methyl- and dimethyl-phenanthrenes to the maturity of organic matter in sedimentary sequences

Supplemental material, sj-docx-1-eea-10.1177_01445987251408084 for Multivariate relationship of methyl- and dimethyl-phenanthrenes to the maturity of organic matter in sedimentary sequences by Xiaotian Huang, Liuping Zhang, Bangjun Liu, Kun Zhang and Zhaoyang Li in Energy Exploration & Exploitation

Supplemental Material

sj-docx-2-eea-10.1177_01445987251408084 - Supplemental material for Multivariate relationship of methyl- and dimethyl-phenanthrenes to the maturity of organic matter in sedimentary sequences

Supplemental material, sj-docx-2-eea-10.1177_01445987251408084 for Multivariate relationship of methyl- and dimethyl-phenanthrenes to the maturity of organic matter in sedimentary sequences by Xiaotian Huang, Liuping Zhang, Bangjun Liu, Kun Zhang and Zhaoyang Li in Energy Exploration & Exploitation

Footnotes

Acknowledgments

We are grateful to Drs. Ziyi Wang and Shanshan Zhou who kindly assisted in the data processing. This research was supported by China National Key Science and Technology Project (grant number 2025ZD1405004), and China National Petroleum Corporation-Peking University Strategic Cooperation Project of Fundamental Research (grant number 22-2-1). This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

ORCID iDs

Xiaotian Huang

Liuping Zhang

Bangjun Liu

Zhaoyang Li

Authors’ contributions

Xiaotian Huang contributed to methodology, formal analysis, visualization, and writing‒original draft. Liuping Zhang contributed to conceptualization, methodology, formal analysis, writing–original draft, and supervision. Bangjun Liu contributed to formal analysis, and writing–original draft. Kun Zhang contributed to methodology, formal analysis. Zhaoyang Li contributed to formal analysis, writing–original draft.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability

The data that support the findings of this study are available in the Supplementary Material.

Supplemental material

Supplemental material for this article is available online.

Author biographies

Xiaotian Huang received her Master degree from China University of Petroleum (2017-2020). Since 2022, she has been pursuing her Doctoral degree at the Key Laboratory of Deep Oil and Gas Theory and Intelligent Exploration and Development, Institute of Geology and Geophysics, Chinese Academy of Sciences. Her interests are in shale oil geochemistry and geology.

Liuping Zhang received BSc degree in geochemistry from Beijing University (1979-1983) and PhD degree in geochemistry from China University of Geosciences (1997-2000). His interests include mechanism for petroleum accumulation in reservoirs with low permeability, distribution prediction of gas, condensate, light oil and heavy oil in shale, and In-situ conversion of shale with low maturity.

Bangjun Liu received his Doctoral degree from Goethe University Frankfurt, Germany in 2020. His research focuses on organic geochemistry and stable isotopes of coal-bearing strata.

Kun Zhang received her Master degree from the Qinghai Institute of Salt Lakes, Chinese Academy of Sciences (2006-2009). Her interest focuses on experimental studies of hydrocarbon accumulation.

Zhaoyang Li received his Doctoral degree from the Institute of Geology and Geophysics, Chinese Academy of Sciences (2010-2017). His interests are in shale oil geochemistry and geology.

Appendix

References

Alexander

Bastow

Fisher

, et al. (1995) Geosynthesis of organic compounds: II. Methylation of phenanthrene and alkylphenanthrenes. Geochimica et Cosmochimica Acta 59(20): 4259–4266.

Alexander

Strachan

Kagi

, et al. (1986) Heating rate effects on aromatic maturity indicators. Organic Geochemistry 10(4–6): 997–1003.

Bao

Wang

Zhou

, et al. (1992) The relationship between methylphenanthrene ratio and thermal evolution of organic matter. Journal of Jianghan Petroleum Institute 14(4): 8–13.

Barenbaum

(2017) Oil origin and age. Georesursy 19(1): 30–37.

Behar

Budzinski

Vandenbroucke

, et al. (1999) Methane generation from oil cracking: Kinetics of 9-methylphenanthrene cracking and comparison with other pure compounds and oil fractions. Energy & Fuels 13(2): 471–481.

Boreham

Crick

Powell

(1988) Alternative calibration of the Methylphenanthrene Index against vitrinite reflectance: Application to maturity measurements on oils and sediments. Organic Geochemistry 12(3): 289–294.

Budzinski

Garrigues

Radke

, et al. (1993) Thermodynamic calculations on alkylated phenanthrenes: Geochemical applications to maturity and origin of hydrocarbons. Organic Geochemistry 20(7): 917–926.

Budzinski

Garrigues

Connan

, et al. (1995) Alkylated phenanthrene distributions as maturity and origin indicators in crude oils and rock extracts. Geochimica et Cosmochimica Acta 59(10): 2043–2056.

Chen

Zhang

Niu

, et al. (2020) Applicability of aromatic hydrocarbon parameters in the evaluation of maturity of lacustrine hydrocarbon source rocks: A case study of mesozoic hydrocarbon source rocks in the Yinchuan-Ejinaqi Basin. Acta Petrolei Sinica 41(8): 928–939.

10.

Curiale

Larter

Sweeney

et al. (1989) Molecular thermal maturity indicators in oil and gas source rocks. In: Naeser

McCulloh

(eds) Thermal History of Sedimentary Basins: Methods and Case Histories. New York: Springer-Verlag, 53–72.

11.

Dolzhenko

Fomin

(2022) Informativeness of the phenanthrene indicators of organic matter maturity at late mesocatagenesis and apocatagenesis: An example of material from superdeep hole Srednevilyuiskaya 27 in East Siberia. Geochemistry International 60(1): 33–42.

12.

Eriksson

Byrne

Johansson

et al. (2013) Multi- and Megavariate Data Analysis: Basic Principles and Applications, 3rd ed. Umeå: Umetrics Academy.

13.

Favilla

Durante

Vigni

, et al. (2013) Assessing feature relevance in NPLS models by VIP. Chemometrics and Intelligent Laboratory Systems 129: 76–86.

14.

Galvez

Fischer

Jaccard

, et al. (2020) Materials and pathways of the organic carbon cycle through time. Nature Geoscience 13(8): 535–546.

15.

Garrigues

De Sury

Angelin

, et al. (1988) Relation of the methylated aromatic hydrocarbon distribution pattern to the maturity of organic matter in ancient sediments from the Mahakam delta. Geochimica et Cosmochimica Acta 52(2): 375–384.

16.

Garrigues

Oudin

Parlanti

, et al. (1990) Alkylated phenanthrene distribution in artificially matured kerogens from Kimmeridge clay and the Brent Formation (North Sea). Organic Geochemistry 16(1–3): 167–173.

17.

Hakimi

Kumar

Singh

, et al. (2023) Geochemistry and organic petrology of the Bituminite shales from the Kapurdi mine, Rajasthan of NW India: Implications for waxy oil generation potential. Journal of Petroleum Exploration and Production Technology 13(2): 505–521.

18.

Hatem

Mustapha

Hakimi

, et al. (2023) Preliminary study of crude oils from Shabwah depression in the Central Sab'atayn Basin (Yemen): Revealing oil-oil correlation with organic matter input and maturity based on geochemical properties of Maltene and Asphaltene fractions. Geoenergy Science and Engineering 226(2023): 211754.

19.

Huang

et al. (2021) Determination of the extract content of rock. National Energy Administration SY/T 5118-2021: 1–4.

20.

Yang

You

(2011) PLS-based gene selection and identification of tumor-specific genes. IEEE Transactions on Systems, Man, and Cybernetics. Part C Applications and Reviews 41(6): 830–841.

21.

Kruge

(2000) Determination of thermal maturity and organic matter type by principal components analysis of the distributions of polycyclic aromatic compounds. International Journal of Coal Geology 43(1–4): 27–51.

22.

Kumar

Dubey

Singh

, et al. (2025a) A petrological approach to understand the signatures of palaeowildfire activities during the formation of Permian coal: A case study of Bastacolla Mine, Jharia Coalfield, India. Journal of the Palaeontological Society of India 70(1): 156–166.

23.

Kumar

Hakimi

Singh

, et al. (2022) Geochemical and petrological characterization of the early Eocene carbonaceous shales: Implications for oil and gas exploration in the Barmer Basin. Northwest India. ACS Omega 7(47): 42960–42974.

24.

Kumar

Mustapha

Singh

, et al. (2024a) Evaluating the hydrocarbon generation potential of the Paleocene–Eocene carbonaceous rocks in the Barmer and Bikaner-Nagaur basins, Western Rajasthan, India. China Geology 8(1): 77.

25.

Kumar

Singh

Paul

, et al. (2020) Evaluation of hydrocarbon potential with insight into climate and environment present during deposition of the Sonari Lignite, Barmer Basin Rajasthan. Energy and Climate Change 1: 100006.

26.

Kumar

Singh

Christanis

(2024b) Paleodepositional environment and hydrocarbon generation potential of the paleogene lignite in the Barmer Basin, Rajasthan, India. Journal of Asian Earth Sciences 259: 105892.

27.

Kvalheim

Christy

Telnæs

, et al. (1987) Maturity determination of organic matter in coals using the methylphenanthrene distribution. Geochimica et Cosmochimica Acta 51(7): 1883–1888.

28.

Huang

George

(2022) Unusual occurrence of alkylphenanthrenes in upper Eocene to Oligocene sediments from the western margin of Tasmania, Australia. Marine Geology 450: 106859.

29.

Luo

Liu

Bai

(2023) Geochemical characteristics of aromatic hydrocarbons in crude oil from oil fields in the southern part of the Turkmenistan Turan Plateau. Complex Hydrocarbon Reservoirs 16(3): 256–263.

30.

Mahieu

Qannari

Jaillais

(2023) Extension and significance testing of Variable importance in projection (VIP) indices in partial least squares regression and principal components analysis. Chemometrics and Intelligent Laboratory Systems 242: 104986.

31.

Mukhopadhyay

(1994) Vitrinite reflectance as maturity parameter. Petrographic and molecular characterization and its applications to basin modelling. ACS Symposium Series 570: 1–24.

32.

Nilankar

Singh

, et al. (2024) Evolution of pore morphometry and compositional changes in immature Raniganj Shale under heated environment: Implications for enhanced gas recovery. Energy & Fuels 38(22): 22128–22145.

33.

Noah

Horsfield

Han

, et al. (2020) Precise maturity assessment over a broad dynamic range using polycyclic and heterocyclic aromatic compounds. Organic Geochemistry 148: 104099.

34.

Nomoto

Hagiwara

Nakano

, et al. (2000) A new parameter for maturity determination of organic matter in sediments based on the clay-catalyzed thermal isomerization of monomethylphenanthrenes. Bulletin of the Chemical Society of Japan 73(6): 1437–1443.

35.

Olah

Prakash

Goeppert

(2011) Anthropogenic chemical carbon cycle for a sustainable future. Journal of the American Chemical Society 133(33): 12881–12898.

36.

Peters

Walters

Moldowan

(2005) The Biomarker Guide: Volume 1, Biomarkers and Isotopes in the Environment and Human History. Cambridge: Cambridge University Press.

37.

Radke

(1987) Organic geochemistry of aromatic hydrocarbons. In: Brooks

Welte

(eds) Advances in Petroleum Geochemistry, Volume 2. London: Academic Press, 141–207.

38.

Radke

Horsfield

Littke

et al. (1997) Maturation and petroleum generation. In: Welte

Horsfield

Baker

(eds) Petroleum and Basin Evolution: Insights from Petroleum Geochemistry, Geology and Basin Modeling. Berlin, Heidelberg: Springer, 169–229.

39.

Radke

Vriend

Ramanampisoa

(2000) Alkyldibenzofurans in terrestrial rocks: Influence of organic facies and maturation. Geochimica et Cosmochimica Acta 64(2): 275–286.

40.

Radke

Vriend

Schaefer

(2001) Geochemical characterization of lower Toarcian source rocks from NW Germany: Interpretation of aromatic and saturated hydrocarbons in relation to depositional environment and maturation effects. Journal of Petroleum Geology 24(3): 287–307.

41.

Radke

Welte

Willsch

(1982a) Geochemical study on a well in the Western Canada Basin: Relation of the aromatic distribution pattern to maturity of organic matter. Geochimica et Cosmochimica Acta 46(1): 1–10.

42.

Radke

Welte

Willsch

(1982b) Aromatic components of coal: Relation of distribution pattern to rank. Geochimica et Cosmochimica Acta 46(10): 1831–1848.

43.

Radke

Welte

Willsch

(1986) Maturity parameters based on aromatic hydrocarbons: Influence of the organic matter type. Organic Geochemistry 10(1–3): 51–63.

44.

Radke

Willsch

Leythaeuser

, et al. (1991) Distribution of alkylated aromatic hydrocarbons and dibenzothiophenes in rocks of the Upper Rhine Graben. Chemical Geology 93(3–4): 325–341.

45.

Raiswell

Berner

(1987) Organic carbon losses during burial and thermal maturation of normal marine shales. Geology 15(9): 853–856.

46.

Requejo

(1994) Maturation of petroleum source rocks—II. Quantitative changes in extractable hydrocarbon content and composition associated with hydrocarbon generation. Organic Geochemistry 21(1): 91–105.

47.

Sidik

Mustapha

Kumar

(2024) Geochemical characteristics of crude oils and source rocks in the Malaysia-Thailand Joint Development Area(MTJDA), North Malay Basin. Marine and Petroleum Geology 170(2024): 107135.

48.

Singh

Kumar

(2020) Assessment of thermal maturity, source rock potential and paleodepositional environment of the paleogene lignites in Barsingsar, Bikaner–Nagaur basin, Western Rajasthan, India. Natural Resources Research 29(2): 1283–1305.

49.

Sivan

Datta

Singh

(2008) Aromatic biomarkers as indicators of source, depositional environment, maturity and secondary migration in the oils of Cambay Basin, India. Organic Geochemistry 39(11): 1620–1630.

50.

Smith

George

Batts

(1995) The geosynthesis of alkylaromatics. Organic Geochemistry 23(1): 71–80.

51.

Song

Jin

Liu

, et al. (2007) Distribution of methylphenanthrene in hydrocarbon source rocks and its impact on maturity parameters. Petroleum Geology & Experiment 29(2): 183–187.

52.

Srinivasan

Jacobi

Atwah

, et al. (2022) Integration of methyldibenzothiophene and pyrolysis techniques to determine thermal maturity in sulfur-rich Type II-S source rocks and oils. Organic Geochemistry 163: 104333.

53.

Stojanović

Jovančićević

Vitorović

, et al. (2007) Hierarchy of maturation parameters in oil-source rock correlations. Case study: DRMNO depression, Southeastern Pannonian Basin, Serbia and Montenegro. Journal of Petroleum Science and Engineering 55(3–4): 237–251.

54.

Sun

(1998) Influences of secondary oxidation and sulfide formation on several maturity parameters in Kupferschiefer. Organic Geochemistry 29(5–7): 1419–1429.

55.

Telnaes

Bjørseth

Christy

, et al. (1987) Interpretation of multivariate data: Relationship between phenanthrenes in crude oils. Chemometrics and Intelligent Laboratory Systems 2(1–3): 149–153.

56.

Tissot

Pelet

Ungerer

(1987) Thermal history of sedimentary basins, maturation indices, and kinetics of oil and gas generation. AAPG Bulletin 71(12): 1445–1466.

57.

Qiu

et al. (2012) Method of determining microscopically the reflectance of Vitrinite in sedimentary rocks. National Energy Administration SY/T 5124-2012: 1–6.

58.

Voigtmann

Yang

Batts

, et al. (1994) Evidence for synthetic generation of methylphenanthrenes in sediments. Fuel 73(12): 1899–1903.

59.

Wakeling

Morris

(1993) A test of significance for partial least squares regression. Journal of Chemometrics 7(4): 291–304.

60.

Wang

Meng

(2006) Partial Least Squares Regression: Linear and Nonlinear Methods (Refined Edition). Beijing: National Defense Industry Press, pp. 97–123.

61.

Wang

Huang

Zheng

(2021) Thermal maturity parameters derived from tetra-, penta-substituted naphthalenes and organosulfur compounds in highly mature sediments. Fuel 288: 119626.

62.

Wang

Dong

Jin

, et al. (2024) Efforts to untie the multicollinearity knot and identify factors controlling macropore structures in shale oil reservoirs. Advances in Geo-Energy Research 11(3): 194–207.

63.

Wang

, et al. (2016) Novel maturity parameters for mature to over-mature source rocks and oils based on the distribution of phenanthrene series compounds. Heliyon 2(3): e00093.

64.

Wold

(1978) Cross-validatory estimation of the number of components in factor and principal components models. Technometrics 20(4): 397–405.

65.

Wold

Jonsson

Sjörström

, et al. (1993) DNA and peptide sequences and chemical processes multivariately modelled by principal component analysis and partial least-squares projections to latent structures. Analytica Chimica Acta 277(2): 239–253.

66.

Wold

Ruhe

Wold

, et al. (1984) The collinearity problem in linear regression. The partial least squares (PLS) approach to generalized inverses. SIAM Journal on Scientific and Statistical Computing 5(3): 735–743.

67.

Wold

Sjöström

Eriksson

(2001) PLS-regression: A basic tool of chemometrics. Chemometrics and Intelligent Laboratory Systems 58(2): 109–130.

68.

Xiao

Liu

Zhang

, et al. (2014) Conflicting sterane and aromatic maturity parameters in Neogene light oils, eastern Chepaizi High, Junggar Basin, NW China. Organic Geochemistry 76: 48–61.

69.

Xiao

Zhang

et al. (2008a) Analytical Method of Hydrocarbons in Petroleum and Sediment by Gas Chromatography. National Development and Reform Commission SY/T 5779-2008: 1–12.

70.

Xiao

Zhang

(2008b) Method of determining microscopically the reflectance of Vitrinite in coal. Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of China GB/T 6948-2008: 1–10.

71.

Yang

Lai

, et al. (2018) Methylphenanthrene index and methylphenanthrene ratio of organic matter in lacustrine hydrocarbon source rocks and their relationship with maturity. Journal of Yangtze University (Natural Science Edition 15(19): 12–17.

72.

Zdravkov

Groß

Bechtel

, et al. (2023) Depositional settings of the Eocene Suhostrel bituminous coal, SW Bulgaria, inferred from organic petrology and molecular proxies. International Journal of Coal Geology 276: 104319.

73.

Zhang

(1991) A study of the method for oil source correlation with trace elements in petroleum. Petroleum Exploration and Development 18(S1): 96–106.

74.

Zhang

Bai

Zhao

, et al. (2019) Oil-source correlation in the slope of the Qikou Depression in the Bohai Bay Basin with discriminant analysis. Marine and Petroleum Geology 109: 641–657.

75.

Zhang

Bai

Zhao

(2014) Data-processing and recognition of seepage and microseepage anomalies of acid-extractable hydrocarbons in the south slope of the Dongying depression, eastern China. Marine and Petroleum Geology 57: 385–402.

76.

Zhang

Wang

, et al. (2013) A novel molecular index for secondary oil migration distance. Scientific Reports 3(1): 2487.

77.

Zhao

Qin

Shen

, et al. (2023) Significant influence of different sulfur forms on Sulfur-containing polycyclic aromatic compound formation in high-sulfur coals. Fuel 332: 125999.

78.

Zhao

Qin

Zhao

, et al. (2021) Origin and geological implications of super high sulfur-containing polycyclic aromatic compounds in high-sulfur coal. Gondwana Research 96: 219–231.

79.

Zhao

Qin

, et al. (2024) Effect of Sulfur contents on polycyclic aromatic compounds in low-rank bituminous coals. Polycyclic Aromatic Compounds 45(4): 697–722.

80.

Zhao

Chen

Guo

, et al. (2013) Geochemical characteristics of aromatic hydrocarbons in crude oil and source rocks from the Naiman Depression. Kelu Basin. Geochimica 42(3): 262–273.

81.

Zhao

Zhang

Jin

, et al. (2017) Hydrocarbon charging and accumulation history in the Niudong buried hill field in the Baxian depression, eastern China. Marine and Petroleum Geology 88: 343–358.

82.

Zhou

Bai

Zhang

, et al. (2021) Identifying oil sources in the Wen'an Slope of the Baxian depression, the Bohai Bay Basin, North China. Marine and Petroleum Geology 128: 104938.

83.

Zhou

Littke

(1999) Numerical simulation of the thermal maturation, oil generation and migration in the Songliao Basin, Northeastern China. Marine and Petroleum Geology 16(8): 771–792.

84.

Zumberge

Rocher

Zumberge

et al. (2020) From here to maturity: Estimating thermal maturity of crude oils. In: Proceedings of the AAPG Hedberg Conference, the Evolution of Petroleum Systems Analysis. American Association of Petroleum Geologists Search and Discovery Article 42487. Available at: https://www.searchanddiscovery.com/documents/2020/42487zumberge/ndx_zumberge.pdf (accessed 15 May 2024).

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.01 MB

0.12 MB