Magmatic volatile release to the atmosphere can lead to climatic changes and substantial environmental degradation including the production of acid rain, ocean acidification and ozone depletion, potentially resulting in the collapse of the biosphere. The largest recorded mass extinction in Earth’s history occurred at the end of the Permian, coinciding with the emplacement of the Siberian large igneous province, suggesting that large-scale magmatism is a key driver of global environmental change. However, the source and nature of volatiles in the Siberian large igneous province remain contentious. Here we present halogen compositions of sub-continental lithospheric mantle xenoliths emplaced before and after the eruption of the Siberian flood basalts. We show that the Siberian lithosphere is massively enriched in halogens from the infiltration of subducted seawater-derived volatiles and that a considerable amount (up to 70%) of lithospheric halogens are assimilated into the plume and released to the atmosphere during emplacement. Plume–lithosphere interaction is therefore a key process controlling the volatile content of large igneous provinces and thus the extent of environmental crises, leading to mass extinctions during their emplacement.
Large igneous provinces (LIPs) are the product of rapid eruptions of large volumes of magma over short geological timescales. The Permo-Triassic Siberian flood basalts (SFB) erupted approximately 4 × 106 km3 of basalt in less than 1 Myr1. The eruption of the SFB is contemporaneous with the main stage of the end-Permian crisis and is hypothesized to have contributed to environmental changes that resulted in loss of >90% of all marine, and >70% of all terrestrial species2,3. The end-Permian mass extinction has been attributed to sharp fluctuations in global temperatures and/or increased levels of ultraviolet (UV) radiation resulting from extensive ozone depletion, both of which are associated with the magmatic release of volatiles to the atmosphere2,4,5,6,7. Yet the amount of volatiles expected to be released from the SFB (assuming conventional plume source magmatism) is insufficient to account for the environmental degradation and climatic fluctuations that occurred during the end-Permian crisis, requiring an additional source of volatiles to be released during SFB emplacement8,9. To reconcile the missing SFB volatiles, it has been variously argued that large quantities of volatiles were released via contact metamorphism of a sedimentary sequence1,5, by melting of recycled eclogite within the mantle plume10 or by melting of the cratonic lithosphere7. However, the source of volatile species responsible for climatic fluctuations and ozone depletion during the end-Permian crisis remains unknown.
Here we report the first detailed halogen (Cl, Br and I) data for peridotite xenoliths from two Siberian kimberlites: one (Udachnaya, 360 million years ago (Ma)) emplaced before, and the other (Obnazhennaya, 160 Ma) after the SFB eruption approximately 250 Ma (refs. 11,12; see also Supplementary Fig. 1). The Udachnaya xenoliths (n = 9) represent melt extraction from the depleted cratonic mantle; whereas Obnazhennaya xenoliths (n = 6) contain both Archaean cratonic lithosphere and melt residues generated from the SFB plume (Supplementary Information)13. Determining the halogen composition of the cratonic sub-continental lithospheric mantle (SCLM; Udachnaya) and the SFB plume residues (Obnazhennaya) provides a means to estimate the composition of the SFB before eruption, and to quantify the contribution of the lithospheric mantle to the halogen budget of the SFB.
Lithospheric mantle as a reservoir for halogens
Owing to its isolation and non-convective nature, the SCLM retains geochemical heterogeneities introduced through interactions with mantle, crustal and subduction-related sources14,15. Metasomatic components infiltrating the SCLM are sampled by mantle xenoliths. Rapidly transported to the surface during kimberlite volcanism, these xenoliths provide a window into the composition and origin of SCLM volatiles16,17,18.
The halogen and noble gas composition of the Udachnaya and Obnazhennaya xenoliths was determined using neutron-irradiated noble gas mass spectrometry (Methods). The results are summarized in Supplementary Tables 2 and 3 and displayed in Fig. 1. The range of Cl, Br and I concentrations within the Udachnaya and Obnazhennaya xenoliths are distinct, with the average concentrations from crushing and stepped-heating consistently higher in Udachnaya samples, indicating that they originate from different domains within the SCLM (Supplementary Information). Halogen-bearing fluids present in the samples, as indicated by release during crushing experiments, have a similar range of Br/Cl and I/Cl values in both Udachnaya and Obnazhennaya xenoliths (Fig. 2a). During stepped-heating of crushed residues, the xenoliths show evidence for distinct endmember halogen compositions (Fig. 2b). Udachnaya retains similar Br/Cl and I/Cl values to those measured during crushing, whereas Obnazhennaya samples have more mantle-like Br/Cl and I/Cl values (Supplementary Table 3).
Previously published helium isotopic data also vary between the xenoliths suites, with the average 3He/4He value of Udachnaya (0.4 ± 0.3 RA, RA = 3He/4He of atmosphere = 1.38 × 10−6) consistently lower than Obnazhennaya (4.2 ± 0.9 RA), which has a maximum 3He/4He value (8.4 ± 0.3 RA) that is similar to mid-ocean-ridge basalt (MORB)18. 3He/4He, Br/Cl and I/Cl values seem to be coupled (Fig. 3), showing that fluids within the Obnazhennaya xenoliths represent a mixture between a component rich in radiogenic 4He, Br and I, and a component with mantle-like 3He/4He and halogen compositions. The lower 3He/4He and elevated Br/Cl and I/Cl values characteristic of Udachnaya xenoliths are considered representative of the ancient metasomatized section of the SCLM (metasomes) that was present before major influence of the SFB mantle plume. In contrast, Obnazhennaya helium and halogen data (Fig. 3) suggest aplume-like volatile origin that has subsequently mixed with SCLM-derived volatiles. The observation of a plume-like volatile source in Obnazhennaya is supported by similar rare earth element (REE) patterns and melt extraction ages between Obnazhennaya xenoliths andthe SFB13.
The initial inventory of halogens within the metasomatized section of Siberian SCLM before plume impingement can be estimated using the composition of Udachnaya xenoliths. The Siberian SCLM transitions from depleted harzburgite and lherzolites to predominantly metasomatized peridotites at 180–190 km depth19. Assuming that the Udachnaya peridotite xenoliths are representative of metasomatized peridotites in the lower 30 km of the SCLM, and taking the surficial area of the Siberian Craton (4 × 106 km2), then the metasomatized portion of the Siberian SCLM contains approximately 0.6–1.5 × 1019, 1.6–2.7 × 1017 and 0.5–1.1 × 1014 kg of Cl, Br and I, respectively (Supplementary Table 1 and Supplementary Information). The metasomatized Siberian SCLM is therefore enriched in Cl, Br and I by factors of up to 125, 675 and 100 times, respectively, relative to the depleted MORB mantle (DMM)20. Thus, the SCLM is a notably larger and more heterogeneous halogen reservoir than previously considered, and may impart a major influence on global volatile cycles21.
Release of lithospheric halogens during LIP emplacement
The comparatively large quantity of halogens residing in the base of the Siberian SCLM means that even small proportions released to the surface will have important consequences for the global halogen cycle. The eruption of halogens into the stratosphere catalyses ozone-destroying reactions, raising surface levels of biologically damaging UV radiation22,23. Transit of the SFB plume through the SCLM could potentially have liberated major amounts of halogens and other volatiles to the atmosphere, contributing to species decline and extinction during the end-Permian crisis.
Udachnaya xenoliths formed deeper (>50 km difference)17 in the lithosphere compared with Obnazhennaya. The identification of Udachnaya-like metasomatic signatures in Obnazhennaya indicates that volatiles residing in the metasomatized basal SCLM were mobilized and ascended to shallower parts of the SCLM. Obnazhennaya xenoliths have trace element signatures that fall within the range of previously reported values of the SFB17, strong P–platinum-group element depletions that are uncharacteristic of cratonic lithosphere and Os isotopic compositions that are consistent with a formation age similar to the time of plume impingement13. These characteristics indicate that the part of the lithosphere sampled by the Obnazhennaya kimberlite represents the melt residue of the SFB plume (Fig. 4d)13. The identification of metasomatized SCLM signatures within the Obnazhennaya xenoliths therefore suggests that as the SFB plume impacted the base of the lithosphere, the resulting melts incorporated volatiles mobilized from the deeper metasomatized SCLM, before being erupted at the surface or stalling in the lithosphere. The contribution of SCLM-derived volatiles to the SFB plume can therefore be estimated using differences in the halogen and noble gas signatures between the Udachnaya (metasomatized SCLM) and Obnazhennaya (SFB + metasomatized SCLM) xenoliths.
Assuming that the melt residues in the Obnazhennaya lithosphere had a starting composition similar to the SFB plume (12.7 RA and mantle-like Br/Cl and I/Cl values)24,25, then the amount of assimilation from the SCLM (Udachnaya) can be estimated from the extent of mixing between the two sources (Figs. 2 and 3, Supplementary Information). Reconciling He, Br and I systematics between the SFB plume and the SCLM component represented by Udachnaya requires that up to 70% of volatiles in Obnazhennaya are derived from the SCLM (Fig. 3a,b, Supplementary Fig. 2). Furthermore, any potential overprinting related to crustal assimilation affecting the halogen composition of the melt can be excluded as the rapid transport of xenoliths to the surface via kimberlite volcanism limits interaction with the surrounding crust26,27.
Taking the volume of Cl degassed as calculated from the SFB melt inclusions (8.7 × 1015 kg)28, the total fluxes of Br and I to the atmosphere are estimated to be 2.3 × 1013 kg and 9.6 × 1010 kg, respectively. This calculation assumes that the melt had Br/Cl and I/Cl values similar to Obnazhennaya and considers that halogens are not fractionated during degassing29. Explosive eruptions inject reactive HCl and HBr gases into the lower stratosphere (~12–25 km) and deplete ozone levels, whereas effusive eruptions lead to soluble HCl being washed out before reaching the stratosphere24. Considering only explosive events (20–30% of the SFB)30, a rate of stratospheric injection of roughly 75%31 and the amount of Cl measured within the SFB28, the mass of Cl released to the stratosphere over the main eruptive phase of the SFB (two-thirds of the total eruptive volume over 300 kyr)32 is the equivalent to 0.5–1.0 Pinatubo (1991–1992) eruptions (which caused a 15–20% reduction in global ozone33) every year for 300 kyr. Models of ozone depletion during the SFB eruption that use estimated stratospheric HCl fluxes predict a 30–55% reduction in ozone over the same eruptive timeframe4. These stratospheric HCl flux estimates are five times lower than that predicted from the SFB melt inclusions (8.7 × 1015 kg)28. Furthermore, these estimates do not take into account the consequences of Br degassing on ozone depletion. The large release of Br to the stratosphere during the emplacement of the SFB, as indicated by the high Br/Cl value of the Siberian SCLM, probably further exacerbated ozone depletion. Bromine has a much greater capacity for depleting ozone (~45 times more effective23) and could have reduced ozone levels by a further 20% during the SFB eruption. Although there are several uncertainties in the rate and magnitude of volatile degassing during SFB magmatism, the scale of halogen degassing fluxes presented here is sufficient to incur a near to total loss of global ozone during the end-Permian crisis.
Melt inclusions within the SFB contain between 0.01–0.33 wt% Cl10,28, which is an order of magnitude higher than the maximum measured Cl concentrations in other LIPs8,28. Inclusions with high Cl concentrations are found to be equally enriched in other volatile species such as fluorine (1.95 wt%) and sulphur (0.51 wt%)28. Creating such high volatile contents within these melts, from an initial DMM-like composition, would require a very low degree of partial melting, or the assimilation of volatiles from another unknown reservoir. Low degrees of partial melting can concentrate volatiles in the melt fraction, however, the high Mg contents measured within the melt inclusions preclude low degrees of partial melting, suggesting that the assimilation of volatile-rich material is most likely cause of the high volatile contents in the SFB.
The composition of the Obnazhennaya xenoliths, assumed to represent plume melt residues, can be used to estimate the pre-eruptive SFB melt composition and establish whether the SCLM is the potential source of volatile enrichment in the SFB. Obnazhennaya xenoliths require ~30% melt extraction to account for the elevated olivine Fo >92 (forsterite content, Fo%: molar Mg/(Mg + Fe) × 100)13 and the Cl composition of this melt can therefore be estimated using a batch melting model34. Experimentally determined partition coefficients for Cl between olivine (DClOl/Melt = 1.9 × 10−2) and pyroxene (DClPyx/Melt = 1.5 × 10−2) were used to calculate the Cl concentration of the melt at 1,500 °C, before eruption35. Using the range of Cl concentrations in the olivine and pyroxene minerals from Obnazhennaya xenoliths yields Cl concentrations of 0.1–0.2 wt% in the melt. These estimates are considered upper limits given the potential for an unknown proportion of intact fluid inclusions to remain following crushing, thus Cl data based on stepped heating are likely to overestimate the Cl abundance within the minerals. However, it is notable that the melt Cl estimates are consistent with published values for the eruptive melt composition (0.01–0.33 wt% Cl)10,28 providing confidence in the assumptions that we have made and confirming that the SFB melt was already enriched in Cl before eruption. These arguments therefore constitute further (albeit indirect) support for a SCLM origin for the majority of halogens in SFB melts.
Implications for the end-Permian extinction
Ozone depletion during the end-Permian crisis is considered to have led to the decline in the dominant terrestrial plant species at the time, followed by the rapid expansion of opportunistic lycopsids36. The global distribution of preserved microspores from these emerging lycopsids exhibits features indicative of a failure in the normal development process of the spores. The global dispersion of these mutagenic spores suggests that this was a reaction to global stress factors that are unlikely to be related to changes in global temperature from the release of gases such as SO2 and CO2 during SFB emplacement4. Experiments on the effects of end-Permian UVB regimes on modern conifers led to fivefold increases in both the occurrence of mutagenic malformations and complete sterilization37. This would have caused widespread deforestation and the collapse of the terrestrial biosphere, indicating that ozone depletion was a major contributing factor in the end-Permian mass extinction event4,37.
The peak occurrence of mutagenic spores occurs before the rapid negative shift in δ13C in end-Permian carbonates that is attributed to the extinction of calcified marine life36. The δ13C excursions coincide with a change from predominantly extrusive to intrusive eruptions of the Siberian LIP1. The emplacement of sills into volatile-rich sediments was considered to have released vast quantities of volatiles including CO2 and halocarbons gases to the atmosphere, leading to rapid climate change and ozone depletion1,5. However, from the palynological evidence36 it is clear that a reduction in terrestrial biodiversity was occurring before the onset of marine extinction. Furthermore, evidence for reduced sedimentation rates prior to the Permo-Triassic boundary indicates that there was global eustatic sea-level regression, potentially caused by falling global temperatures and the onset of glaciation38. The rapid decrease in temperatures has been linked to the emission of SO2 to the atmosphere during the eruptive phase of the SFB7.
The concurrent timing of the eruptive phase of the SFB and the palynological evidence for ozone depletion is not consistent with the idea that degassing of sedimentary brines during later intrusive phases of igneous activity were the primary source of halogens causing ozone destruction. As we have shown in this study, the majority of halogens in the SFB were added during plume–lithosphere interaction, followed by their subsequent release to the atmosphere during explosive eruptions. Sulphur enrichments, co-existing with halogens in SFB28, may also have been derived from the SCLM (Fig. 4c). Evidence for a decline in terrestrial species before the Permo-Triassic boundary therefore suggests that the release of halogens and gaseous sulphur species, and the subsequent decrease in ozone and global temperatures, respectively, were the predominant factors in initiating the end-Permian mass extinction. The change in eruptive phase from explosive to intrusive may have played a role in extending the extinction from a mainly terrestrial phenomenon to a global event.
Subducted origin of volatiles in the Siberian lithosphere
The high concentrations of halogens in the Udachnaya xenoliths indicate that the Siberian SCLM has been enriched in volatiles by metasomatic processes. Udachnaya xenoliths have Br/Cl and I/Cl values that are similar to fluids trapped within minerals in the altered oceanic crust (AOC), suggesting that the metasomatism of the Siberian SCLM was driven by subduction-derived fluids (Fig. 2)39. Combined with the noble gases (Fig. 3; Supplementary Figs. 3 and 4) the Udachnaya xenoliths show an evolution from seawater-like 3He/4He, Br/Cl and I/Cl values, to values with increasingly radiogenic 3He/4He values and enriched Br/Cl and I/Cl values (Fig. 3), further suggesting that the metasomatic fluid originated as seawater but subsequently evolved, during subduction or within the SCLM, due to halogen fractionation and the production of 4He from U–Th decay18. Eclogite xenoliths from the Udachnaya kimberlite exhibit δ18O values of up to +7.7%40, which lie outside the normal mantle range (+5.4 ± 0.2%)41 and indicate that they originated as oceanic crust that underwent low-temperature alteration40. Eclogites formed from the subduction of oceanic crust have been shown to retain the halogen and oxygen isotopic signatures of the oceanic crust protolith during metamorphism, providing a mechanism for the delivery of subduction-derived halogens to the Siberian SCLM42,43.
As we have shown in this study, up to 70% of the volatile content of the Siberian plume originated from assimilation of metasomatized lithospheric material. The composition of the SCLM therefore plays an integral role in controlling the volatile content of LIPs, and as such the overall effect that they have on the global environment. Based on the evolution of the Br/Cl, I/Cl and 3He/4He values of the xenoliths from seawater- to AOC-like values, coupled with the AOC-like δ18O values within eclogite xenoliths from Udachnaya, it seems that the Siberian SCLM volatiles originate from the subduction of a seawater-derived component within the AOC. Enrichment of seawater-derived volatiles in the Siberian SCLM provided the plume with an abundant supply of halogens, which were released to the atmosphere during eruption and resulted in globally extensive reductions in ozone levels and the decline of the biosphere. The SCLM is also a major repository for other subducted volatile species, including sulphur and carbon44,45, which can also contribute to environmental degradation during plume–lithosphere interaction and LIP emplacement7. The SCLM can therefore store subducted volatiles that can periodically be mobilized and released to the Earth’s surface and atmosphere during deep-seated melting and volcanism, leading to devastating impacts on the global environment.
Neutron irradiation noble gas mass spectrometry
Olivine and clinopyroxene mineral separates from Udachnaya and Obnazhennaya peridotite xenoliths were selected for heavy halogen (Cl, Br and I) and K analysis using neutron-irradiated noble gas mass spectrometry47,48,49,50,51,52. Samples weighing between 0.015 and 0.062 g were first cleaned with deionized water in an ultrasonic bath for 20 min, then cleaned for a further 5 min in acetone. Samples were then dried under a heat lamp at 100 °C, wrapped in Al foil and sealed in evacuated fused-silica tubes together with the Hb3gr, scapolite and Shallowater meteorite standards to monitor noble gas proxy production from K and halogens51.
Samples were irradiated in the GRICIT (MN2014b) facilities of the TRIGA Reactor, Oregon State University, for a few hours each day between 22 April 2014 and 1 July 2014, giving a total irradiation time of 205 h. Irradiation details and monitor values for this irradiation have been reported previously51.
Noble gas proxy isotopes (38ArCl, 80KrBr, 128XeI and 39ArK) formed during irradiation were measured on the MS1 mass spectrometer52, alongside natural Ar, Kr and Xe isotopes. A subset of samples and repeat analyses were also performed on a Thermo Fisher Scientific ARGUS VI mass spectrometer51. Noble gases were first extracted from trapped fluid inclusions by loading samples into hand-operated modified Nupro valve crushers53 (MS1 mass spectrometer). For bulk sample analysis, powders from crushing analyses were loaded into in a tantalum resistance furnace (MS1) and stepped-heated using four temperature steps of 600 °C, 1,000 °C, 1,400 °C and 1,600 °C to release halogens contained within the mineral matrix. Halogens from four samples (UV33, UV88, UV357 and O129-74) were extracted using a 10.6-μm-wavelength CO2 laser (Teledyne CETAC Fusions CO2 ARGUS VI). To test that both extraction methods gave the similar results, sample O97-12 was analysed using both the furnace and laser; the resulting Br/Cl and I/Cl values varied by less than 20% and 35%, respectively, between laser and furnace extraction, suggesting that halogens were released in similar proportions using both methods. Halogens abundances were then calculated using the well-defined conversion standards with known halogen concentrations (Hb3gr, scapolite and Shallowater meteorite)50,51, which monitor the efficiency of noble gas production through thermal and epithermal neutron reactions.
Air calibrations and blanks were analysed daily to check the sensitivity and background of the spectrometers, with maximum furnace blank values at 1,600 °C on the MS1 being 1.65 × 10−10 cm3 STP 40Ar, 2.92 × 10−13 cm3 STP 84Kr and 3.54 × 10−14 cm3 STP 132Xe and ARGUS VI blanks being 5.76 × 10−12 cm3 STP 40Ar and 1.41 × 10−15 cm3 STP 132Xe, with Kr blanks below the detection limit. The analytical procedures for noble gas purification for the MS1 and ARGUS VI mass spectrometers and the data reduction procedures have been documented previously52,53. External precision is reported at 3% (1σ) for Cl and 7% (1σ) for Br and I determinations.
All data pertaining to this study are presented in the paper and Supplementary Information.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work is dedicated to L.A.T, who passed away in 2017. L.A.T. devoted his life to science and teaching, serving as an excellent mentor to P.H.B. during his time at University of Tennessee. This work was financially supported though a NERC studentship NE/J500057/1 (to M.W.B.) and NERC (NE/M000427/1) and ERC (ERC-267692 NOBLE) grants to C.J.B. and R.B. P.H.B. was funded by an NSF fellowship (EAR-114455) to investigate the geochemical signatures in these samples.