Coal samples with different thermal maturation degree were collected from Ningdong mine, Jiulong mine, and Ge Mudi mine. The raw coal samples were hand-selected to obtain vitrinite-rich samples and inertinite-rich samples (Figure 1). According to the Chinese standard GB/T 15,588-2013 and GB/T 8899-2013, the obtained samples were observed under a microscope (Figures 1A, B), and the samples with 90% or higher inertinite purity were used as the precursor for the simulation experiments.

Coal component quantification, vitrinite reflectance measurement, industrial analysis, and elemental analysis were performed as basic tests and analyses on the experimental samples. The basic information of the samples is presented in Table 1.

The sample was crushed and sieved through a 200-mesh (75 μm) standard sieve. Approximately 15 g of coal powder was weighed and placed in a plastic beaker, it was then mixed with 80 mL of hydrochloric acid (36% mass fraction) solution (with proportional increase in acid quantity when the sample weight increased). The mixture was stirred for 4 h at a constant temperature of 60 °C in a thermostatic water bath. Subsequently, the HCl solution was filtered out, and 80 mL of hydrofluoric acid (40%) solution was added to the coal sample. The same water bath and acid washing process were repeated. Finally, the acid-washed coal sample was rinsed with ultrapure water until no precipitation appeared in the filtrate, and then filtered through filter paper. The filtered sample was placed in a oven and dried at 60°C for 24 h to obtain the demineralized coal samples.

The demineralized samples were assigned unique numbers and divided into six categories for high-temperature thermal simulation experiments (Table 2).

The specific experimental procedure for high-temperature thermal simulation is as follows: Take a 10 g sample in a crucible and place it in the NTG-SML-60W integrated laboratory graphitization furnace. Prior to heating, evacuate the furnace to a vacuum level of 5 Pa and replace the gas once. Throughout the experiment, the furnace is protected by a continuous flow of argon gas at a rate of 10 L/min, maintaining a pressure inside the furnace 20–30 KPa above standard atmospheric pressure. The experiment uses a segmented heating method. Initially, in the temperature range of room temperature to 1,000°C, the heating rate is 5°C/min. Once it reaches 1,000°C, it is held for 60 min, and then the heating rate is changed to 10°C/min until the target temperature is reached, followed by a 90-min dwell time at the target temperature. Subsequently, natural cooling occurs. In this graphite thermal simulation experiment, the temperature points are set starting from 1800°C with intervals of 300°C, and the highest temperature point is set at 3,000°C. A total of 5 temperature points are set, and 30 sets of high-temperature thermal simulation experiments are conducted.

The experiments utilized the Tecnai G2 F30 field emission transmission electron microscope (TEM) with an acceleration voltage of 300 kV. The point resolution was 0.20 nm, the line resolution was 0.10 nm, and the information resolution was 0.14 nm. The magnification ranged from 3,000x to 500,000x.

The specific procedure involved grinding the sample to 300 mesh and then dispersing it in ethanol using ultrasonic treatment. The dispersed sample was then dropped onto a copper mesh. Subsequently, the sample was searched for on the holes of the copper mesh, and particles that could represent the majority of the particle characteristics within the sample were selected for multiscale observation. High-resolution images were captured during the observation process.

The XRD instrument selection specifications are: Smart Lab-9kW, copper target, acceleration voltage 45 kV, current 200 mA, scanning range 2θ from 5° to 70°, scanning rate 2°/min, X-ray wavelength 0.15418 nm. Two diffraction peaks on the XRD pattern (2θ ranges of 20°–30° and 40°–50°) correspond to the positions of the 002 and 100 peaks in the standard graphite XRD diffraction.

The experimental instrument used is the Jobin-Yvon Labram HR Evolution high-resolution micro-Raman spectrometer. The experiment measurements employed an argon ion laser as the excitation light source, with an excitation wavelength of 532 nm and laser power of 100 mW. The scanning range was from 800 cm^-1 to 3,500 cm^-1, and the exposure time was 10 s.

The obtained Raman spectra were analyzed using Origin 8.0 software, utilizing Lorentzian functions for fitting and processing. The Raman spectrum exhibited two bands: the first-order Raman (700 cm^-1 to 2000 cm^-1) and the second-order Raman (2000 cm^-1 to 3,000 cm^-1).

HRTEM allows the observation of the microcrystalline carbon layer structure and in-plane defect states (Oberlin and Terriere, 1975; Sharma et al., 2018; Zhu et al., 2021). It can be observed that there is difference in graphitization ability between vitrinite and inertinite, and the degree of extension and stacking of graphite layers increases significantly with increasing processing temperature.

Under the condition of 1800°C, the NDV sample exhibits carbon layers stacked to approximately 3 layers, with an extension length of less than 5 nm. The predominant structure is a flexibly wrinkled graphite structure with curved and randomly oriented carbon layers, and there is no apparent trend of three-dimensional graphite structure development (Figure 2A). Small-scale flat graphite layers have formed in the NDI sample, indicating a better degree of development than the NDV sample (Figure 2B). However, the internal development is uneven, with coexistence of curved and flexibly wrinkled graphite and well-developed flat graphite under the microscope (Figure 2B). In the JLV sample, a large number of onion-like concentric structures can be observed. This structure is a transitional structure where flexibly wrinkled graphite unfolds towards the development of large-sized graphite structures (Figure 2C). A large number of well-developed large-sized graphite layers can be observed in JLI, with a stacking thickness of more than 20 layers, about 10 nm (Figure 2D). Through observation, it can be noted that under the condition of 1800°C, the development of graphite structures in inertinite is superior to that in vitrinite under the same conditions.

As the thermal simulation temperature increased to 3,000°C, all samples formed large-sized graphite structures. The graphite carbon layers in NDV and NDI samples exhibited relatively disordered orientations (Figures 3A, B). In the NDV sample, smaller-scale and more curved graphite structures were predominant. The development of graphite structures in the NDI sample was better than that in the vitrinite-rich samples, showing the presence of large-scale graphite layers. With an increasing degree of coal metamorphism, the carbon layers in JLV and JLI samples became flatter, more ordered in orientation, and formed larger and well-developed graphite layers (Figures 3C, D). In the JLI sample, the development of graphite layers was more advanced than that in the vitrinite-rich JLV sample. It exhibited highly ordered three-dimensional graphite structures with neat arrangement and distinct planar orientation, while the JLV sample still contained a significant amount of disordered carbon and defect structures, with higher curvature in the formed graphite layers.

When comparing the results of high-temperature thermal simulation experiments between GMV and GMI, which had higher initial degrees of coal metamorphism, more pronounced changes in development characteristics were observed under the microscope (Figure 4). Under the 1800°C and 2,100°C condition, both GMV and GMI samples exhibited predominantly flexibly wrinkled graphite structures with varying directions of microcrystalline stripes and high curvature (Figures 4A–D). However, the GMI sample showed the formation of higher elongation graphite layers at the edges. When the temperature increased to 2,700°C, the GMV sample still consisted mainly of flexibly wrinkled graphite structures, but their elongation and stacking layers significantly increased. In contrast, the inertinite-rich GMI sample at the same conditions already developed large-scale graphite layers (Figures 4E, F). At 3,000°C, the internal graphite structures of the GMV sample gradually exhibited more pronounced orientation and developed flat graphite layers, approaching the degree of structure development observed in the GMI sample under the same conditions (Figures 4G, H). Comparing the products at the same temperature points, the microcrystalline stripes in the GMI sample (Figures 4B, D, F, H) exhibited a more ordered orientation compared to the GMV sample (Figures 4A, C, E, G), indicating a higher degree of structural rationalization in the inertinite-rich sample.

XRD is widely used in the study of coal and coal-based graphite materials (Ruland, 1964), and the following methods can be used to calculate the interlayer spacing d ₀₀₂ and microcrystalline parameters (La and Lc) from the fitted XRD spectra.

(1) The interlayer spacing d ₀₀₂ reflects the average interlayer distance of microcrystals within the scanning range, with units in nm. It can be calculated using the Bragg equation:

λ is the X-ray wavelength (0.154,056 nm) and θ ₀₀₂ is the diffraction angle of the (002) peak

(2) La represents the average size of microcrystalline planes. It can be calculated using the Scherrer formula:

(3) Lc is the average stacking thickness in the vertical direction, with units in nm. The calculation formula is as follows: L c = 1.05 λ / β 002 cos θ 002 (3) where β ₀₀₂ is the FWHM of the (002) peak.

The XRD diffraction patterns of the high-temperature simulated samples are shown in Figure 5, and the lattice parameters such as d ₀₀₂, La, and Lc are calculated (Tables 3–5).

For the same precursor, as the processing temperature increases, the (002) characteristic peak around 2θ=26.6° moves closer to the position of the standard graphite peak (2θ=26.6°), and the peak shape becomes sharper and more symmetrical (Figure 5). The overall trend of the calculated microcrystalline parameters shows a decrease in d ₀₀₂ and an increase in L_a and L_c, indicating an increase in the graphitization degree of the samples with the increase in experimental temperature.

The Raman spectra of the high-temperature simulated samples are shown in Figures 6–8. Various structural characterization parameters, such as the intensity of the D1 peak and the G peak (I_D1 and I_G), the intensity ratio between the D1 peak and the G peak (R1), the peak shift difference between the D1 peak and the G peak (ΔP), the full width at half maximum of the G peak (FWHMG), the defect peak area ratios (R2 and R3), and others, were calculated and are listed in Table 6. Among these parameters, R2 is referred to as the “in-plane defect parameter” representing the proportion of in-plane defects (D1) to the total area (A_D1/(A (_G+D1+D2))). This parameter is particularly useful for evaluating samples with a higher degree of evolution, fewer defects, and primarily D1-type defects (semi-graphite to graphite). R3 is called the “total defect parameter,” representing the proportion of all types of defect peaks to the total area (R3=A_{(D1+D2+D3+D4)}/A_{(D1+D2+D3+D4+G)}). It is more effective in characterizing samples with a lower degree of evolution and more defects (anthracite to semi-graphite).

By analyzing the Raman spectra and data, it can be observed that with increasing simulated temperature, the G peak, which represents ordered graphite structures, gradually strengthens and shifts towards 1,580 cm^-1, while the intensity of the D peak, representing disordered structures, weakens. Among the defect peaks, the D1 peak is predominant, and the D2 peak may also appear in samples with a higher degree of graphitization. However, the D4 peak completely disappears under high-temperature treatment conditions. When the simulated temperature exceeds 1800°C, the D3 peak, representing other heterocyclic defects, almost does not develop, and only a few samples may exhibit weak D4 defect peaks. When the temperature exceeds 2,400°C, the D4 peak no longer develops. Additionally, with increasing temperature, in the second-order Raman spectra, the S1 peak gradually separates into S1’ and S1” peaks, while the S2 peak gradually disappears, indicating the gradual formation of a three-dimensional structure.

By plotting the variation curve of d ₀₀₂ with temperature for all samples in high-temperature thermal simulation, with the high-temperature thermal simulation temperature as the x-axis and the measured d ₀₀₂ values as the y-axis (Figure 9), the following evolution characteristics of d ₀₀₂ during the maceral graphitization process can be summarized.

1) In comparison to vitrinite, inertinite has inherent structural advantages, with lower initial d ₀₀₂ values and stronger structural ordering. In the graphitization process under high-temperature thermal simulation, except for JLV at 2,100°C, where the d ₀₀₂ value is lower than that of JLI (Figure 9B), the evolution of graphite structure of vitrinite is generally lower than that of inertinite under the same conditions. Throughout the entire temperature range, the d ₀₀₂ of vitrinite is higher than that of inertinite, indicating a lower degree of graphitization compared to inertinite (Figures 9A,C).

Using the simulated temperatures as the x-axis and the R2 and R3 values of the simulated samples as the y-axis, the curves of R2 and R3 as a function of temperature for all the simulated samples were plotted (Figure 10).

The R2 and R3 curves of vitrinite and inertinite exhibit similar evolution trends, showing a continuous decrease with increasing temperature, indicating the gradual healing of lattice defects. However, the forms of the lattice defect evolution curves for vitrinite and inertinite are not identical.

The R2 and R3 curves of vitrinite are consistently below the curves of inertinite, and in most cases, the slope of vitrinite curves is higher than that of inertinite curves. This indicates that under the same simulated temperature conditions, vitrinite has fewer lattice defects than inertinite, and the healing rate of defects in vitrinite (reflected by the decrease in R2 and R3 values) is also higher than that in inertinite.

Based on previous studies and the analyses in this paper, two explanations for the difference in defect elimination capacity between vitrinite and inertinite are proposed: (1) The formation of molecular configurations on the carbon layer is closely related to the integrity of the carbon layer (Ruland, 1964). Vitrinite has a lower proportion of aromatic structures, and the aromatic fragments are relatively smaller, making it easier to form carbon layers with fewer defects during stacking. In contrast, inertinite has relatively larger aromatic structures, making it more prone to the formation of carbon layers with vacancies during stacking (Figure 11). (2) Vitrinite produces gaseous carbon atoms during the evolution process, which contribute to defect repair.

The structural evolution during the graphitization process can be characterized by two aspects: the construction of three-dimensional graphite structure, represented by the decrease in interlayer spacing (d ₀₀₂) and an increase in microcrystalline size (La and Lc); and the elimination of lattice defects, which can be measured using R2 and R3. By analyzing the correlation between microcrystalline morphology parameters (d ₀₀₂, La, and Lc) and lattice defect parameters (R2) of high-temperature thermal simulation samples, the development process of graphite structure in inertinite and vitrinite under high-temperature thermal simulation conditions and their differences can be revealed.

Analyzing the d ₀₀₂ -R2 scatter plot (Figure 12A), the following observations can be made: ① The evolution of vitrinite shows a near-linear relationship between d ₀₀₂ and R2, indicating a relatively uniform evolution path for vitrinite during the graphitization process. The scatter plot of inertinite follows a reverse “S" curve, exhibiting a three-stage transition characteristic. The two transition points are located at the starting point of coal-derived graphite-smokeless coal boundary (d ₀₀₂ = 0.344 nm) and the boundary between semi-graphite and graphite (d ₀₀₂ = 0.338 nm) (Kwiecińska and Peterson, 2004; Cao et al., 2021), with d ₀₀₂ ranging from 0.344 nm to 0.338 nm, and the rate of decrease in R2 is relatively low. ② The evolution curve of vitrinite is always below or to the right of inertinite, indicating that, under the same interlayer distance (d ₀₀₂ ), vitrinite has fewer lattice defects than inertinite; or under the same defect level, inertinite has a smaller interlayer distance than vitrinite.

The La-R2 and Lc-R2 scatter plots (Figures 12B, C) exhibit similar characteristics: ① Microcrystalline size parameters (La and Lc) are negatively correlated with the lattice defect R2, indicating simultaneous growth of graphite microcrystals during the healing of lattice defects. ② The data points of vitrinite fit into an exponential function curve (y = ax, 0 < a <1), showing clear regularity: in the R2 > 0.35 range, La slowly increases from 5 nm to 10 nm, and Lc slowly increases to 5 nm; in the R2 < 0.35 range, La rapidly increases from 10 nm to 40 nm, and Lc rapidly increases to 30 nm. The data points and fitted curve for inertinite are more scattered and roughly exhibit three-stage characteristics, indicating the complexity of graphite microcrystal development in inertinite. ③ In the La-R2 and Lc-R2 scatter plot, most data points of vitrinite are located below or to the left of inertinite, implying that vitrinite has fewer lattice defects than inertinite under the same microcrystalline size (La and Lc).

The d ₀₀₂ -Lc and d ₀₀₂ -La scatter plot fitted curve (Figures 12D, E) follows a typical exponential function curve (y = a^x, 0 < a <1), roughly consisting of three stages: ① In the d ₀₀₂ > 0.344 nm range, La and Lc show a low-value and nearly horizontal distribution, with La <10 nm and Lc < 5 nm. ② In the 0.344 nm < d ₀₀₂ < 0.338 nm range, Lc and La slowly increase to around 15 nm and 20 nm, respectively. ③ After d ₀₀₂ < 0.338 nm, Lc and La rapidly increase, indicating the rapid formation of three-dimensional ordered graphite structure. The evolution path of microcrystalline parameters in vitrinite and inertinite during high-temperature thermal simulation is generally the same, but there are differences in microcrystalline size. For the same d ₀₀₂ interlayer distance, the La and Lc of inertinite are often larger than those of vitrinite, especially in the d ₀₀₂ < 0.338 nm range, where the difference in Lc is particularly evident.

The fitted curve of the La-Lc scatter plot (Figure 12F) shows that the increase in Lc occurs only when La reaches a certain level. When La <10 nm, Lc increases weakly, while after La >10 nm, Lc increases rapidly.