Field evidence for coal combustion links the 252 Ma Siberian Traps with global carbon disruption

With loss of >90% of marine species, the Permian-Triassic extinction was the most severe in Earth history (Erwin, 2006). High-precision geochronology implicates Siberian Traps eruptions in the global environmental changes that caused the extinction (Wignall, 2001; Grasby et al., 2011; Burgess and Bowring, 2015; Burgess et al., 2017) and carbon cycle perturbation, including a sharp negative carbon isotope excursion that is a key feature of the mass-extinction interval (e.g., Payne and Clapham, 2012). This carbon isotope excursion coincides within geochronological uncertainty with the oldest dated rocks from the Norilsk section of the Siberian flood basalts (Burgess and Bowring, 2015).

Siberian Traps magmas were chambered within, and intruded through, the Tunguska sedimentary sequence (Il’yukhina and Verbitskaya, 1976). The Tunguska Basin varies between 3 and 12 km thick, and includes carbonates, evaporites, oil and gas, and coal (e.g., Svensen et al., 2018). Coal strata range in age from Carboniferous to Permian, with a cumulative coal thickness of ∼100 m (Ryabov et al., 2014). Thermal metamorphism and combustion of coal, carbonates, and organic-rich shales produce significant CO₂ and CH₄, as well as carbonate metamorphism producing CO₂, in addition to the gases released by volcanics, all of which would have contributed to global warming (Retallack and Jahren, 2008; Svensen et al., 2009; Iacono-Marziano et al., 2012). However, the magnitude, tempo, and origin of carbon emissions during Siberian Traps magmatism have remained in question despite their critical atmospheric importance (Cui and Kump, 2015; Black et al., 2018).

The earliest volcanic deposits of the Siberian Traps include volcaniclastic rocks that overlie Paleozoic sedimentary rocks and underlie the main lava pile in the southern regions of the province (Naumov and Ankudimova, 1995). The thickest volcaniclastic rocks are near the town of Tura and farther south (Fig. 1). Near Tura, drill cores reveal >600 m of volcaniclastic rocks, grading directly into the earliest lavas of the flood basalts (Levitan and Zastoina, 1985). In the Maymecha-Kotuy region, the Pravoboyarsky Suite basal volcaniclastic sequence reaches a maximum thickness of 200–300 m (Fedorenko and Czamanske, 1997). Near Norilsk, the basal volcaniclastic sequences is typically only several meters thick.

The presence of coal fly ash layers at the end-Permian boundary in Arctic Canada provides tantalizing evidence for coal combustion at that time (Grasby et al., 2011). We examined the organic carbon content of the Siberian Traps rocks with a particular focus on early volcaniclastic rocks spanning the large igneous province (Table 1), from earliest eruptions to latest (here we are referring to the late-stage rocks of the Maymecha-Kotuy region), to provide a comprehensive assessment of organic-matter incorporation during magmatism. Here we give further evidence that Siberian Traps magmas intruded into and incorporated coal and organic material, and, for the first time, give direct evidence that the magmas also combusted large quantities of coal and organic matter during eruption.

We sampled along a traverse north from Ust-Ilimsk along ∼200 km of the Angara River, and a similar distance along the Nizhnyaya Tunguska River centering on Tura (Fig. 1). Almost every outcrop on these rivers consists of thick sequences of volcaniclastic rocks, which have been mapped in direct contact with upper Permian sedimentary rocks (Malich et al., 1974). Carbonized woody fragments as much as 10 cm in length were embedded in a number of outcrops on both the Angara and Nizhnyaya Tunguska Rivers (Black et al., 2015). No exposures of Permian and older coal layers were observed along either river. However, dolerite in a coal quarry near the Angara River in Ust-Ilimsk contains vesicles filled with carbon-rich material, malleable with a fingernail (Fig. 2).

We examined 16 samples of volcaniclastic rocks from the Angara, Nizhnyaya Tunguska, and Podkamennaya Tunguska Rivers for carbon content, along with six samples from northern regions (Table 1; Fig. 1; detailed localities are provided in Figs. S3–S7 in the Supplemental Material¹). These rocks have a range of whole-rock bulk carbon content of as much as 2.7 wt%, and total organic carbon (TOC) content from 0.01 to 1.16 wt% (Table 1; see the Supplemental Material). As context, TOC values in shales of >0.5 wt% have potential as a petroleum source rock (Peters and Cassa, 1994); the volcaniclastic rocks studied here may exceed the carbon threshold for an economically viable petrochemical source.

Samples were prepared as crushed-rock polished pellets and examined under reflected light (Table 1; Table S1 in the Supplemental Material). Of the 22 samples examined, 11 samples that span five geographically separated regions had visible large organic fragments enclosed in the rock matrix (Fig. 2) as well as organic macerals visible under the microscope (Figs. 3A–3N; Table S1). High values of random vitrinite reflectance (R_or) are indicative of higher thermal maturation of organic matter. The thermal maturity of the particles ranged from marginally mature to mature (R_or = 0.56%–0.83%), indicating the varying degree that organic matter was thermally altered by incorporation into the magma. We divided the organic particles into three general maceral types based on morphology and thermal maturation (Table 1).

Type 1 macerals are coal fragments within the volcaniclastic host rock that predominantly consist of vitrinite, with a mean R_or of 0.56% (Fig. S1A). These bituminous coal fragments likely reflect the level of thermal maturity prior to eruption and show devolatilization features such as small vacuoles and desiccation cracks (Figs. 3A–3D).

Type 2 macerals have bright high-temperature char rims surrounding a less-altered interior (Figs. 3G and 3H). These outer chars show R_or values as high as 4%, along with contraction cracks and combustion rims, and likely reflect combusted wood.

Type 3 macerals are cenospheres and char particles embedded in the volcaniclastic matrix (Figs. 3I–3N). Cenospheres are formed by explosive devolatilization of organic matter that was heated rapidly to high temperatures (∼1300 °C; Goodarzi et al., 2008). We recognized two types: isotropic particles (Figs. 3I–3L) with plastic deformation and bright oxidation rims indicative of rapid heating in the presence of air (∼30% of sampled cenospheres), and anisotropic particles (Figs. 3M and 3N) with a fine-grained optical texture typical of combustion byproducts of coal (Goodarzi and Hower, 2008) (Fig. S2).

Of the eight volcaniclastic samples from the most southerly regions (the Podkamennaya Tunguska and Angara Rivers), seven contained coal and combusted organic-matter fragments. Abundant charcoal was also found in an end-Permian crater-lake deposit near Bratsk (Fristad et al., 2017). Only three of the eight samples from the Nizhnyaya Tunguska River, in central Siberia where the southernmost lavas appear, contain coal and organic matter. Farther north, only one of our six analyzed samples from the Norilsk and Kotuy regions contain coal and organic-matter fragments. However, previous work has identified graphite, bitumen, and carbonaceous material within Norilsk lavas and sills, which have been interpreted as evidence for incorporation of hydrocarbons in this area (e.g., Ryabov et al., 2012).

Near Norilsk, in the town of Kaerkan, a large open-pit coal mine contains outcrops where the Ivakinsky, the earliest lava flow, is in contact with coal. The coal appears to have been liquefied and injected into cracks in the cooling lava, leading to more-reducing conditions in the magma (e.g., Ryabov et al., 2014) (Fig. 2). Melenevsky et al. (2008) reported that coal in the broader aureole has been converted to anthracite, indicating heating to ∼200 °C and release of ∼260 kg HC/t organic matter (mass of hydrocarbon [HC] per unit mass of rock in metric tons). The southern volcaniclastics were also capable of burning or coking coal: paleomagnetic data demonstrating unidirectional remnant magnetization among some Angara rocks imply that temperatures exceeded 600 °C during magma emplacement (Black et al., 2015).

A major question is whether these samples record coal heated by magma, or incorporation of charcoal formed previously in wildfires (Grasby et al., 2015; Hudspith et al., 2014). Maceral texture distinguishes these options. We identify both isotropic high-reflectance organic matter (R_or >2% and as high as 11%), which may have resulted from forest fires, as well as high-reflectance chars with anisotropic cenospheres characteristic of coal combustion (Figs. 3M and 3N).

Cenospheres form in present-day coal-burning power plants (Hudspith et al., 2014). These particles are rarely reported in the pre-industrial sedimentary record, but have been observed in sedimentary rocks at the Permian-Triassic boundary in the Sverdrup Basin of Arctic Canada and are interpreted as a signature of magmatic coal combustion (Grasby et al., 2011), consistent with models of Ogden and Sleep (2012). In contrast, char and inertinite particles observed in our samples may be products of forest fires (e.g., Fristad et al., 2017) (Figs. 3I–3L).

The observations presented here are interpreted as evidence that coal and organic-matter combustion, along with forest fires, occurred in response to volcanism. Moreover, we infer that these interactions were widespread, based on the presence of thermally altered and/or burnt coal and organics in volcaniclastic rocks spanning the southern and central Siberian Traps province (Fig. 1).

The onset of the Permo-Triassic mass extinction is marked by a major carbon isotope excursion, which coincides within geochronological uncertainty with the oldest dated rocks from the Norilsk section (Burgess and Bowring, 2015). Emplacement of the organic matter–bearing southern volcaniclastic rocks preceded the main lava sequence (Levitan and Zastoina, 1985), permitting alignment with the carbon isotope excursion. Coincident with the isotope excursion, Siberian Traps magmatism was characterized by emplacement of laterally extensive sill complexes that could have facilitated significant interaction with coal and organic matter–bearing units (Burgess et al., 2017) and production of the volcaniclastic rocks.

Mantle carbon (δ¹³C ≈ −5‰) (Javoy et al., 1986) and carbon in marine limestones (δ¹³C ≈ −2‰ to +2‰) are too isotopically heavy to have caused the end-Permian carbon isotope excursion (e.g., Cui and Kump, 2015; Gales et al., 2020; Payne and Kump, 2007). Consequently, coal or organic matter (δ¹³C ≈ −25‰) (Cui et al., 2013), methane clathrates (δ¹³C ≈ −56‰) (Krull and Retallack, 2000), or petroleum (δ¹³C ≈ −30‰ to −25‰) (Svensen et al., 2009) represent the most plausible sources of light-carbon injection. Assuming equilibrium with an end-Permian dissolved inorganic carbon (DIC) reservoir of 38,000 Gt C with an initial DIC δ¹³C = 0, and a C isotope mass balance in which Δ¹³C = ΔM*δ¹³C_coal/(ΔM + DIC), where Δ¹³C denotes the magnitude of the isotope excursion and ΔM denotes C release from coal, we infer that each 1000 Mt C released from coal or organic matter with δ¹³C_coal ≈ −25‰ would translate to an ∼−0.64‰ shift in ocean-atmosphere δ¹³C (Cui and Kump, 2015). Mass-balance calculations indicate that 6000–10,000 Gt C with δ¹³C ≈ −25‰ could yield a global carbon isotope perturbation with the observed magnitude of −3‰ to −6‰ (Cui and Kump, 2015).

The primary uncertainties for estimating the magnitude of light-carbon release are (1) the total mass of coal and organic matter that interacted with Siberian Traps magmas and (2) the efficiency of carbon release to the atmosphere during these interactions. The cumulative thickness of Carboniferous to Permian coal layers in the Tunguska Basin has been estimated as ∼100 m (Retallack and Jahren, 2008), comprising ∼10⁴–10⁵ Gt C. Thermodynamic and experimental data (Iacono-Marziano et al., 2012) suggest that coal combustion and/or coking takes place only at pressures of several hundred bars or less, and at the onset of Siberian magmatism, Permian coal measures were indeed located near the surface (Retallack and Jahren, 2008). We interpret some of the combusted particles as originating from organic matter rather than mature coal, and the mass of organic matter hosted in organic-rich shales and peats in the Tunguska Basin is much larger than that of coal proper (Svensen et al., 2009), suggesting that ∼10⁴–10⁵ Gt C represents a lower estimate of carbon available, sufficient to drive the observed carbon isotope excursion based on the mass-balance calculations discussed above.

Our findings provide direct field evidence that Siberian Traps magmas incorporated and combusted coal and organic-rich material. This combustion may have been linked to the formation of breccia pipes in the region (Jerram et al., 2016; Ogden and Sleep, 2012; Svensen et al., 2009). The presence of cenospheres and char also provides evidence for ejection of combusted coal ash into the atmosphere, supporting previous suggestions of significant coal fly ash formation at this time in Siberia, carried on global air currents and deposited in the Arctic Canada Sverdrup Basin (Grasby et al., 2011). In addition to carbon released from the mantle (Sobolev et al., 2011) and thermally metamorphosed country rocks, our results show that coal combustion also liberated light carbon, contributing to the global warming and carbon-cycle disruption that characterized the Permo-Triassic mass extinction (e.g., Cui and Kump, 2015).

We thank reviewers Paul Wignall, Henrik Svensen, and Ying Cui, and our team members. Funding was provided by U.S. National Science Foundation (NSF) Continental Dynamics grant EAR-0807585 to Elkins-Tanton, a grant of the Russian Foundation for Basic Research (18-35-20058) to Veselovskiy, and NSF Integrated Earth Systems grant EAR-1615147 to Black.