3.1 Fieldwork permits

All fieldwork was carried out in compliance with local regulations. Sampling was conducted in accessible outcrops and abandoned/open mining areas, and no specific permits were required for sample collection. No protected sites or species were involved.

3.2 Coal maceral composition and vitrinite reflectance

Coal samples were crushed to ~1 mm, sieved, and cold-mounted in epoxy resin, then polished into coal pellets. Macerals, including vitrinite, inertinite, and liptinite and their subgroups, were identified under a reflected light microscope equipped with a 50 × oil-immersion objective. Homogeneous, crack-free vitrinite was selected for reflectance measurements, and ≥50 points were measured under monochromatic light at 546 nm. The maximum reflectance (Romax) of each point was recorded, and the mean value was calculated.

3.3 Proximate analysis of coal

Air-dried coal samples (<0.2 mm) were analyzed for moisture (dried at 105–110°C until constant weight), ash yield (combusted at 815 ± 10°C until constant weight), volatile matter (heated at 900 ± 10°C under an anaerobic atmosphere for 7 min, converted to dry ash-free basis, daf), and total sulfur (gravimetric sulfate precipitation). Oxygen uptake was determined using a thermogravimetric analyzer by exposing samples to dry O₂ at 30 °C for 48 h and recording the mass gain.

3.4 Ultimate analysis of coal

Air-dried coal samples (<0.2 mm) were analyzed with an elemental analyzer to determine C, H, and N by complete combustion at 1150 °C with chromatographic separation of the products. Total sulfur was measured by high-frequency combustion at 1350 °C in an oxygen-rich atmosphere, and the released SO₂ was quantified using an infrared absorption detector.

3.5 Petrographic analysis of CM rocks

Eight representative samples of CM rocks (including clinker, paralava, and breccia) were selected for thin section preparation and petrographic analysis. Rock slices (~0.05 mm thick, ~ 25 × 15 mm) were cut, mounted on glass slides, and ground using progressively finer abrasives. The polished thin sections were examined and photographed under a Nikon LV100POL polarizing microscope.

3.6 X-ray diffraction (XRD) analysis

Ten CM rock samples with varying metamorphic degrees were ground in an agate mortar to <200 mesh and dried at 105 °C for 2 h. The powdered samples were analyzed using a Rigaku Ultima IV diffractometer equipped with Cu-Kα radiation and a D/teX Ultra detector. X-ray intensities were recorded over a 2θ range of 5–50° at a scanning rate of 20°/min under operating conditions of 40 kV and 40 mA.

3.7 Zircon (U-Th)/He dating

One sample of CM rock was collected, characterized by a reddish to dark-red color with vesicular textures and weak melting features, indicating a relatively high degree of combustion metamorphism. The sample was crushed, and heavy and light mineral fractions were separated using a water-shaking table, followed by heavy liquid and magnetic separation. Zircon crystals were handpicked from the mineral separates under a binocular microscope.

(U-Th)/He analyses were carried out at the University of Melbourne following the protocol of House et al. (2000) for laser He extraction from single zircon grains [66]. Clear, inclusion-free, non-fractured euhedral grains with comparable size ranges were selected, immersed in ethanol, and examined under polarized light to exclude grains with visible inclusions. All dated grains are euhedral zircons with two terminations (Table 2). Grain geometries were imaged and measured for α-ejection corrections [67], and the grains were then loaded into small acid-treated platinum capsules.

Helium was extracted by laser heating at ~12.6 W (~1300 °C) for 20 min, with hot blanks run between samples to ensure complete degassing. Degassed zircons were transferred to Parr bombs, spiked with ²³⁵U and ²³⁰Th, and digested in 0.3–0.5 mL HF at 240 °C for 40 h. Identical spiked standards and unspiked blanks were prepared for calibration. A second digestion step in HCl at 200 °C for 24 h ensured complete dissolution of fluorides. The residues were dried, dissolved in HNO₃, and diluted to 5% acidity with ultrapure water. U, Th, and isotopic ratios were determined using an Agilent 7700X ICP-MS.

Zircon (U-Th)/He ages were calculated and corrected for α- ejection following Farley et al [67]. Analytical accuracy was monitored using Fish Canyon Tuff zircon [68,69]. The zircon (U-Th)/He results are presented in Table 2.

4.1 Coal characteristics

4.1.1 Maceral composition and vitrinite reflectance.

The organic constituents of the coal are dominated by inertinite, averaging 69.52%, 70.05%, and 50.73% in the B₂, B₁, and B₀ seams, respectively. Vitrinite is secondary, accounting for 25.76%, 26.43%, and 37.89% in the same seams, with minor liptinite present. The maximum vitrinite reflectance (R^omax) of the B₂ seam ranges from 0.30% to 0.55% (mean 0.40%); that of the B₁ seam ranges from 0.27% to 0.65% (mean 0.39%); and that of the B₀ seam ranges from 0.49% to 0.50% (mean 0.50%). Coals from the B₂, B₁, and B₀ seams of the Xishanyao Formation are classified as low-rank. Among the inorganic constituents, clay minerals are overwhelmingly predominant (Table 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Statistical results of coal petrographic identification of the B-coal seams in the Xishanyao Formation ().

https://doi.org/10.1371/journal.pone.0348116.t001

4.2 Distribution and petro-mineralogical characteristics of CM rocks

CM rocks are mainly distributed within the Shazhang Fold-Thrust Belt, occurring in the southern segment of the Huoshaoshan Anticline and on both flanks of the axis of the Xidagou Syncline, with a general SW–NE strike. The combustion metamorphic zone extends over approximately 10 km², with the depth of combustion increasing progressively from the northwest and west toward the southeast and east. The regional topography gently slopes from north to south, forming a landscape of hills and depressions (Fig 3 a-c), with elevations ranging from 487 to 538 m and a north–south relief of 51 m.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Field photographs of CM rocks.

(a-c) Hills and depressions formed by CM rocks. (d) Hill formed by CM rocks, more than 10 m high. (e) Brick-red clinker. (f) Upper part composed of brick-red clinker and lower part of grayish-black paralava. (g) Columnar joints developed in CM rocks. (h) Gypsum infilling within fractures of CM rocks.

https://doi.org/10.1371/journal.pone.0348116.g003

CM rocks are high-temperature metamorphic products formed by coal seam combustion. During spontaneous combustion, the adjacent sandstones or mudstones were subjected to the high-temperature baking of the burning coal seams, which altered their original characteristics and led to dehydration or oxidation, thereby modifying their petrographic, mineralogical, and geochemical properties. Based on the degree of combustion metamorphism, CM rocks can be divided into three types: baked rocks, clinker, and paralava. The most distinctive petrographic feature of CM rocks is their characteristic coloration. Clinker is typically brick-red, whereas some may also exhibit light yellow or gray-green hues (Fig 3 d-f). Paralava usually exhibits brown or grayish-black colors (Fig 3f). Due to the high-temperature roasting during coal seam combustion, CM rocks show distinct structures and textures, mainly including relic structures, vesicular structures, columnar jointing (Fig 3g), and combustion-induced fractures (Fig 3h).

Petrographic observations show that in red baked rocks, quartz grains develop numerous microcracks due to the intense thermal stress generated by rapid heating and cooling; hematite adhering to clay minerals is predominantly red (Fig 4a). The thermal stress associated with combustion metamorphism significantly enhances the undulatory extinction of quartz, making it more widespread, pronounced, and complex, with broader and more irregular extinction bands (Fig 4b). In ochre-red clinkers, the edges of quartz grains are partially melted, becoming rounded and smooth, and in some cases exhibiting embayed contours; iron oxides appear opaque, with transparent edges showing dark brown coloration (Fig 4c). In dark gray paralava, tridymite is observed, and the glassy matrix appears black due to the presence of iron minerals, with vesicles present within the glass (Fig 4d). Additionally, mullite and cristobalite form radially oriented symplectitic aggregates, and dendritic magnetite and cordierite are visible (Fig 4 e, f).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Thin-section micrographs of CM rock.

(a) Red clinker, quartz grains with microcracks, clay minerals coated by hematite (PPL). (b) Red clinker, quartz grains showing undulatory extinction (XPL). (c) Reddish-brown clinker, quartz grains with embayed margins, iron oxides coating or disseminated within the glassy matrix, appearing opaque (PPL). (d) Dark-gray paralava, tridymite, glass, vesicles (PPL). (e) Paralava, dendritic magnetite, radiating/fibrous intergrowths of cristobalite and mullite (PPL). (f) Paralava, rhombohedral hematite, tridymite, cordierite (PPL). Quartz, Qz; Hematite, Hem; Tridymite, Trd; Cordierite, Crd; Mullite, Mul; Glass, Gl; Vesicles, Ves.

https://doi.org/10.1371/journal.pone.0348116.g004

XRD results show that in red clinker, the dominant minerals are quartz, hematite, and ferrocordierite (Fig 5a). In paralava, the main minerals include quartz, microcline, albite, phosphoquartz, and pseudobrookite (Fig 5b). Some of these minerals are products of high-temperature metamorphic reactions during coal seam combustion.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. XRD diagrams of CM rocks.

(a) clinker, (b) Paralava. Quartz, Qtz; Tridymite, Trd; Hematite, Hem; Ferrocordierite, Fe-Crd; Microcline, Mc; Albite, Ab; Pseudobrookite, Psbr.

https://doi.org/10.1371/journal.pone.0348116.g005

4.3 (U-Th)/He ages

Mineral separation was conducted on red clinker sample, and five zircon grains were selected for (U-Th)/He dating. Three zircons yielded similar ages, ranging from 0.56 ± 0.03 to 0.70 ± 0.04 Ma, with a weighted mean age of 0.63 ± 0.19 Ma. Two additional grains yielded markedly older ages of 8.7 ± 0.5 Ma and 87.7 ± 5.4 Ma (Table 2). All obtained ages are younger than the depositional age of the host strata, suggesting that the samples experienced post-depositional thermal annealing. An eU-age comparison shows no simple systematic negative correlation between age and eU, suggesting that radiation damage alone is insufficient to explain the observed dispersion.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Single grain zircon (U-Th)/He ages of CM rocks.

https://doi.org/10.1371/journal.pone.0348116.t002

5.1 Causes of coal seam spontaneous combustion

Coal seam spontaneous combustion is controlled by multiple factors, which can be broadly categorized into three aspects: the intrinsic propensity of the coal to spontaneous combustion, the thickness and occurrence of the coal seam, and external environmental conditions [70–73].

The intrinsic propensity of coal to spontaneous combustion is considered an internal factor, mainly governed by its rank, maceral composition, moisture, ash yield, volatile matter, total sulfur, and oxygen absorption capacity [74–77]. The tendency of coal to self-ignite is generally inversely proportional to its rank, with low-rank coals being more susceptible to spontaneous combustion than high-rank coals [78–80]. Inertinite, characterized by a loose structure and strong oxygen affinity, enhances the susceptibility of coal seams to self-heating [81]. In addition, high volatile matter, high moisture, elevated contents of carbon, hydrogen, and nitrogen, and reduced ash yield collectively contribute to an increased tendency for spontaneous combustion [73,78]. Smaller coal particle sizes further promote self-heating, as their larger specific surface area enhances both oxygen accessibility and heat transfer, thereby increasing the risk of self-ignition [82–84].

In the study area, the spontaneous combustion tendency of coal seams was investigated. The Jurassic Xishanyao Formation B-seam coals are dominated by inertinite, followed by vitrinite and liptinite. The vitrinite reflectance (R^omax, %) ranges from 0.39 to 0.50 (Table 1), indicating a low-rank coal with a high propensity for spontaneous combustion. In addition, the B-seam coals of the Xishanyao Formation are characterized by high moisture, high volatile matter, large specific surface area, and low ash yield. The elemental compositions of carbon, hydrogen, nitrogen, and sulfur also show a strong linear relationship with the spontaneous combustion tendency index (Table 1). Therefore, the B-seam coals of the Xishanyao Formation exhibit a pronounced tendency toward spontaneous combustion. The thickness of a coal seam provides the material basis for spontaneous combustion. Thicker seams generate more heat during oxidation, and the heat is less likely to dissipate, thereby increasing the likelihood of spontaneous combustion [50,72]. In the study area, the coal seams of Member 3 of the Xishanyao Formation have an average thickness of 2–26 m, corresponding to medium to thick seams, which favors spontaneous combustion.

In addition to intrinsic factors such as the spontaneous combustion tendency, thickness, and occurrence of coal seams, external conditions mainly include ignition sources (e.g., lightning strikes or wildfires), arid climatic conditions, and the availability of oxygen [35,85]. Previous studies have further shown that coal seam spontaneous combustion occurs more frequently under arid to semi-arid climatic conditions, where low precipitation, sparse vegetation, and windy environments favour long-term exposure of coal seams and sustained oxidation and heat accumulation. In contrast, during more humid climatic stages, higher groundwater levels and thicker overburden tend to suppress coal seam spontaneous combustion to some extent [35,36,78,86].

Nevertheless, the most critical external factor controlling the initiation and persistence of coal seam spontaneous combustion is the availability of sufficient oxygen, which constitutes the prerequisite for coal self-ignition [87,88]. Regional tectonic uplift and denudation can raise deeply buried coal seams to shallow levels or even expose them at the surface, thereby providing the oxygen required for combustion. Consequently, such geological processes play a decisive role in triggering coal seam spontaneous combustion [38,44,89]. In the study area, tectonic uplift and erosion since the Late Cenozoic have progressively brought coal seams to shallow depths or surface exposure. Meanwhile, faults and joints generated by tectonic activity cut through the coal seams, enhancing downward oxygen migration from the surface into the coal-bearing strata, promoting oxidation reactions and ultimately initiating spontaneous combustion.

5.2 Coupling between tectonic uplift events and the formation of CM rocks

Tectonic uplift and denudation are the key external processes that bring coal seams into shallow, oxygen-rich conditions and thereby enable spontaneous combustion. Since the Mesozoic, the eastern orogenic belt of the Junggar Basin has undergone multiple phases of uplift and erosion [19,22,90]. Within this framework, the zircon (U-Th)/He ages of CM rocks provide direct chronological information on the timing at which the Xishanyao coal seams approached or reached near-surface conditions in the study area.

The dominant zircon (U-Th)/He ages of CM rock samples obtained in this study range from 0.56 ± 0.03 to 0.70 ± 0.04 Ma, with a weighted mean age of 0.63 ± 0.19 Ma, is interpreted as the principal age of coal-seam combustion and CM rock formation in the Middle Pleistocene (Table 2). This result indicates that the Jurassic Xishanyao Formation B coal seams were uplifted and eroded to near-surface levels during the Middle Pleistocene, subsequently undergoing spontaneous combustion and forming CM rocks, during which the zircon (U-Th)/He system in most grains was fully reset. In contrast, the two older single-grain ages (8.7 ± 0.5 Ma and 87.7 ± 5.4 Ma) do not define coherent additional age populations. Given their single-grain occurrence, large deviation from the clustered population and the absence of a systematic eU-age trend, these two grains are more conservatively interpreted as incompletely reset or inherited pre-combustion thermochronologic components rather than as evidence for separate combustion episodes.

This interpretation should nevertheless be regarded as cautious rather than definitive. The dated grains are all euhedral and free of optically visible inclusions under binocular and polarized-light inspection, but no cathodoluminescence imaging or internal U-Th zoning analyses were carried out. Accordingly, crystal-scale heterogeneity cannot be fully excluded. Likewise, because the dataset is limited to five usable single-grain ages from one sample, it is not sufficiently robust for inverse thermal-history modelling, and the distinction between partial resetting and inheritance cannot be resolved uniquely. Future studies based on larger single-grain datasets, together with zircon imaging and complementary thermochronometers, will be important for refining the interpretation of the anomalously old grains.

The spatial distribution and geological context of the CM rocks nonetheless support a close coupling between uplift and combustion. CM rocks occur in the central part of the Shazhang Fold-Thrust Belt, specifically on the southern segment of the Huoshaoshan Anticline and along both flanks of the Xidagou Syncline axis. The overall stratigraphic thickness of the Xishanyao Formation shows a decreasing trend near the Xidagou Syncline and the axis of the Huoshaoshan Anticline, indicating that the anticline had already developed prior to coal seam deposition and subsequently experienced uplift. During the Late Permian, nearly east–west compressive stress modified the region, giving rise to a series of incipient north–south-trending folds. In the Middle Jurassic, during the deposition of the Xishanyao Formation, the intense Yanshanian orogeny drove north–south plate convergence, resulting in uneven uplift and subsidence in the Shazhang area [91,92]. In the Late Jurassic, nearly east–west shortening further intensified the north–south folds formed in the Late Permian, eventually establishing their final structural pattern and producing an angular unconformity between the Cretaceous and Jurassic strata [12,92]. Subsequently, the study area underwent multiple episodes of uplift and denudation, including during the Late Cretaceous–Paleocene and since the Miocene [20]. These earlier structural developments set the stage for later exhumation of the coal-bearing strata.

Critically, the rapid uplift and denudation phases since the Miocene are temporally linked with the age signals identified in the CM rocks. During this period, the Bogda Mountain and Kalamaili Mountain experienced intense uplift due to north–south compression triggered by the far‐field effects of the collision between the Indian and Eurasian plates. Liu (2023) suggested that this stage was characterized by the greatest magnitude of uplift, the most intense tectonic activity, and the strongest deformation of strata along the basin margins [19]. Apatite fission-track dating of volcanic rocks near Tianchi on the northern flank of the Bogda Mountains indicated an uplift–denudation phase during 12–7 Ma, as revealed by quantitative thermal history modeling [9]. Similarly, apatite fission-track dating of tuffaceous sandstones from the northern Bogda region demonstrated a rapid uplift–denudation phase since 11 Ma, with an estimated uplift rate of 190.6 m/Ma [23]. In the Kelameili Mountains, studies have shown that the most pronounced phase of rapid uplift–denudation occurred since ~20 Ma, with an uplift rate of approximately 71 m/Ma [22,93]. Thermal history modeling of samples from the basin margins further indicates that nearly all of them recorded a rapid uplift phase since the Late Miocene (ca. 10 Ma) (Fig 6).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Statistical summary of Mesozoic-Cenozoic tectonic activity timings [2,3,9,14,17,22,23,93,94] and CM rock ages in the orogenic belts around the eastern Junggar Basin.

https://doi.org/10.1371/journal.pone.0348116.g006

In this regional context, the ~ 8.7 Ma zircon (U-Th)/He age is compatible with retention of an older uplift-related cooling signal in a grain that was not completely degassed during the later combustion event. Such partial resetting and preservation of inherited zircon (U-Th)/He ages are common in transient high-temperature systems, including CM rocks [38,45]. The much older 87.7 Ma grain is best regarded as a residual pre-combustion age component that has no direct significance for the timing of coal burning at the study site.

Continued tectonic activity and erosion after the Late Miocene eventually culminated in the Middle Pleistocene exposure of the Jurassic coal seams. The clustered ~0.63 Ma ages therefore record the stage at which the coal seams actually reached near-surface, oxygenated conditions and spontaneous combustion was initiated. In this sense, the Late Miocene to Pliocene uplift prepared the structural and geomorphic framework for combustion, whereas Middle Pleistocene exhumation provided the final condition necessary for ignition. The present study thus suggests a sequential relationship between regional tectonic exhumation and subsequent CM rock formation, with a measurable lag between broader orogenic uplift and local combustion metamorphism.

In addition, the main faults in the study area include the Shaxi Fault, Huodong Fault, and Zhangdong Fault. All of them belong to basement-involved triangle shear faults, which initiated in the late Permian and were finally formed at the end of the Cretaceous. The fault terminations only extend into the Lower Jurassic strata and do not cut through the Jurassic coal seams [11]. Therefore, we infer that these faults did not provide favorable conditions for coal seam spontaneous combustion. The combustion ages therefore appear to represent a lagged response to regional uplift and erosion rather than direct fault-controlled ignition. Such a lag is reasonable because tectonic stress was first accommodated by uplift of the surrounding mountain belts and only later translated into sufficient local exhumation to expose coal seams at the surface or near surface.

In summary, the formation ages of the CM rocks record a sequential coupling between regional tectonics exhumation and local coal-seam combustion. The dominant Middle Pleistocene age of ~0.63 Ma tightly constrains the timing at which continued uplift and erosion exposed the Jurassic coal seams to near-surface, oxygen-rich conditions, thereby triggering spontaneous combustion and CM rock formation. The isolated older ages are interpreted more cautiously as incompletely reset or inherited components that may retain information on earlier exhumation history, but they are not used as direct evidence for separate coal-fire events in the study area.