Introduction

The Cryogenian Period, spanning the Sturtian (ca. 717–660 Ma) and Marinoan (ca. 650–635 Ma) Snowball Earth glaciations, includes a 10-million-year interglacial interval termed the Cryogenian nonglacial interlude (ca. 660–650 Ma)1,2,3. These two global glaciations and their associated extreme climate fluctuations likely had drastic effects on the evolution of the Earth’s palaeoenvironment-life system3,4,5,6,7. Notably, the recession of glaciers was particularly associated with enhanced continental weathering, increased riverine supply of nutrients, including dissolved phosphate (PO43−) and sulfate (SO42−), and recovery of marine productivity6,7. This transition from icehouse to greenhouse climatic conditions was likely pivotal for subsequent periods of rapid biological innovation, including the shift from a prokaryote-dominated to a eukaryote-dominated biological marine phytoplankton and the prominent rise of complex multicellular eukaryotes6,7,8,9. This climatic shift has traditionally been attributed to the long-term build-up of an exceptionally high atmospheric carbon dioxide (CO2) concentration, estimated to have reached about 350 times the present atmospheric level (PAL), leading to rapid melting of Earth’s ice sheets10. Geochemical data also indicate that active methanogenesis and significant methane (CH4) emissions further intensified the greenhouse effect during the Marinoan deglaciation5,11,12. Methane is a potent greenhouse gas with a global warming potential > 27 times that of CO2, playing a crucial role in the global carbon (C) cycle13. Nevertheless, the role of CH4 and its connection to marine biogeochemical cycles during the Sturtian glaciation remains largely unexplored.

The Sturtian glaciation, the first and longest of the Cryogenian ice ages (ca. 57 My), is also characterized by the resurgence of iron formations (IFs)1,4,14,15. This sudden and conspicuous global reappearance of Sturtian-aged IF deposition, following a nearly 1-billion-year hiatus in the sedimentary record, indicates anoxic deep waters enriched in dissolved Fe2+ 3,4,14. The formation of Sturtian-aged IFs is closely linked to the evolution of glacial systems4,16. These IFs are pivotal in deciphering the Cryogenian ocean chemistry and glacial climate dynamics17. For instance, Sturtian-aged IFs in northern Namibia, south Australia, and California were deposited across a range of glaciomarine environments, from ice-contact to ice-distal settings under sea ice cover4,17,18. However, Sturtian-aged IFs in the Nanhua Basin of the South China Block (SCB) (Fig. 1a) exhibit distinct depositional environments and are thought to have formed during the waning stage of the Sturtian glaciation, exhibiting minimal glacial influence16,19,20,21 (Fig. 1b; Supplementary Fig. 1; Supplementary Information).

Fig. 1: Palaeogeographic setting and stratigraphic correlation of Sturtian iron formations in the South China Block.
figure 1

a Palaeogeographic reconstruction of the South China Block (SCB) during the late Neoproterozoic Era, showing the distribution of Sturtian-aged iron formations (IFs) in the region5,8,21,51,52,54,106; 1 (Sanjiang IF); 2 (Tongdao IF); 3 (Jiangkou IF); 4 (Qidong IF); and 5 (Xinyu IF). The inset shows the geographic locality of the Yangtze Block. The shelf is subdivided into inner and outer regions. Within the broader slope-to-basin environment, an inferred deep-water sub-basin is delineated, reflecting relatively greater water depth (Supplementary Information). This palaeogeographic framework is primarily informed by previous regional interpretations, including stratigraphic, sedimentological, and lithofacies features, and remains provisional in nature (see main text and SI for a detailed discussion of its limitations). Symbols for IFs and submarine uplifts are not to scale and are exaggerated for illustrative purposes. b Representative stratigraphic columns illustrating the correlation of the studied IFs with the waxing and waning cycles of the Sturtian glaciation in the SCB5,8,19,21,51,52,53. The thicknesses of the IFs, manganese formations, and intervening dolostone layers in the Fulu Formation, as well as the pebble-free siltstone, sandstone, and shale layers in the Nantuo Formation, are exaggerated for illustrative purposes and do not accurately represent their true dimensions.

Full size image

IFs are marine chemical sediments rich in iron (Fe), predominantly composed of Fe3+- and Fe2+-bearing minerals such as hematite, magnetite, and Fe-rich carbonates (e.g., ankerite and siderite)22,23. These Fe-bearing minerals are generally interpreted as originating from Fe³⁺ oxyhydroxide (Fe(OH)3) precursors22,23. Their formation is attributed to multiple Fe2+ oxidation pathways, including reaction with molecular oxygen (O2) from oxygenic photosynthesis, photoferrotrophy, and microaerophilic chemolithoautotrophic bacterial oxidation22,24. Hematite is interpreted to have formed either via direct dehydration of a precursor Fe(OH)3 phase during diagenesis or fluid-mediated oxidation and replacement of an Fe2+ phase25. Transformation of Fe(OH)3 precursors at IF deposition sites also drives diagenetic processes that produce magnetite and Fe2+-rich carbonates, which are closely linked to the biogeochemical cycling of Fe and C in anoxic environments26,27,28,29. It is widely accepted that such Fe-bearing mineral transformations were predominantly driven by microbial dissimilatory iron reduction (DIR), wherein Fe(OH)3 reduction was coupled to the oxidation of organic matter, with Fe3+ serving as the terminal electron acceptor22,24,26,28,29,30. However, Konhauser et al.26 proposed Fe3+ reduction might have also been linked to anaerobic oxidation of CH4 (AOM) in anoxic marine sediments.

In the present-day environment, most AOM is sulfate-dependent31,32. Such a process may not have been efficient during the Precambrian, owing to low SO42− concentrations33. In recent years, AOM coupled with the reduction of Fe (OH)3 and Mn oxides (MnO2) has emerged as a globally significant biogeochemical process in O2-depleted environments34,35,36,37,38. Indeed, metal-dependent AOM, where Fe(OH)3 and MnO2 act as electron acceptors, likely played a prevalent role in Precambrian oceans given that much of the global oceans remained ferruginous until at least the early Palaeozoic38,39,40,41,42. These conditions, characterized by high dissolved Fe2+ in porewater, minimal SO42−, elevated CH4 concentrations, and abundant reactive Fe(OH)3 and MnO2, provide a conducive geochemical setting for metal-dependent AOM33,34,35,43. Despite its significance, the exact mechanisms of this CH4-related process in the Precambrian remain poorly understood.

In this study, we examined five well-preserved Sturtian-aged IFs in the SCB (Fig. 1a, b), which host diverse Fe-rich and manganese (Mn)-rich minerals, including hematite, magnetite, ankerite, manganosiderite, and occasionally pyrite (Supplementary Figs. 2, 3). This mineralogical diversity provides a unique opportunity to explore direct evidence of CH4-driven processes. The formation of these minerals associated with CH4 cycling may yield distinct patterns in metal element distributions and isotopic compositions5,42,44,45. These patterns reflect microbial metabolic variability and interactions with trace nutrients or metalloenzymes5,42,45,46. Methanogens and anaerobic methane-oxidizing archaea (ANME), which mediate these processes, rely on metal cofactors, such as nickel (Ni), cobalt (Co), and zinc (Zn), for electron transport and catalysis44,45,47. Furthermore, these microbes can preferentially incorporate isotopically light Ni and selectively accumulate heavy rare earth elements (HREE)5,45,48,49. These distinct biogeochemical signatures may ultimately be concentrated within authigenic Fe-bearing minerals5,45.

We analysed Ni and Fe isotopes, along with bio-essential metals (Ni, Co, and Zn), and REE and yttrium (REY) in hematite and magnetite, as well as the bulk and in-situ C isotopic composition of Mn-rich carbonates. These geochemical proxies hold promise for elucidating the role of CH4 and its connection to the palaeoenvironment-life system during the waning stage of the Sturtian glaciation.

Results

Geological setting and sample description

The Sturtian and Marinoan glacial deposits, along with the notably warm interglacial interval, are well-preserved in the Nanhua Basin of the Yangtze Block3 (Fig. 1b; Supplementary Fig. 1). The Cryogenian stratigraphy of the Nanhua Basin, with well-constrained global chronologies, consists of the Chang’an, Fulu, Datangpo, and Nantuo formations in ascending order, along with their correlative equivalents in the slope-to-basin environment50 (Fig. 1a; Supplementary Fig. 1). The Sturtian glacial deposits are represented by the Chang’an and Fulu formations, while the Nantuo Formation corresponds to the Marinoan glaciation, with the non-glacial, Mn-rich Datangpo Formation separating them3,6 (Fig. 1b; Supplementary Fig. 1).

The Sturtian glacial deposits in the Nanhua Basin comprise multiple diamictite–sandstone/siltstone cycles, likely reflecting dynamic ice sheet fluctuations19,21 (SI). The Chang’an Formation, dominated by massive and stratified diamictites, represents the main waxing phase of the Sturtian glaciation and glacier advance19,21 (Fig. 1b; Supplementary Fig. 1). The overlying Fulu Formation is stratigraphically subdivided into two distinct lithological members (Fig. 1b; Supplementary Fig. 1). The basal member primarily comprises sandstones and muddy siltstones, with occasional dolostone intercalations. Sedimentary structures, lithological characteristics, and geochemical proxies indicate a depositional environment with minimal direct glacial influence (SI). This succession is interpreted as representing the waning stage of the Sturtian glaciation, marking the progression of deglaciation2,19,20,21,51 (Fig. 1b; Supplementary Fig. 1). The upper member of the Fulu Formation is characterized by the reappearance of massive and crudely stratified diamictites, indicating a glacial re-advance. This assemblage is interpreted as recording the second waxing phase of the Sturtian glaciation2,19,20,21,51 (Fig. 1b; Supplementary Fig. 1).

The Sturtian-aged IFs of the Nanhua Basin occur at the base of the first member of the Fulu Formation, conformably overlying the diamictites of the Chang’an Formation (Fig. 1b; Supplementary Fig. 1). These IFs are stratigraphically correlated across the slope-to-basin environment, exhibiting distinct palaeoenvironmental and palaeogeographic patterns52,53 (Fig. 1a, b; SI). Palaeoenvironmentally, the Tongdao and Xinyu IFs were deposited within a broad slope-to-basin environment, whereas the Sanjiang, Jiangkou, and Qidong IFs were formed in an inferred deep-water sub-basin delineated within this broader setting, indicative of relatively deeper depositional conditions8,21,51,53,54 (Fig. 1a; SI). Palaeogeographically, the Xinyu IF was positioned nearest to the oldland, while the Tongdao IF formed on a submarine uplift developed atop a pre-existing structural high, as indicated by isopach patterns of the local Fulu Formation8,21,51,53,54 (Fig. 1a). While the current palaeogeographic reconstruction provides a valuable spatial framework (Fig. 1a), it remains provisional due to limited stratigraphic resolution, incomplete exposure of the Cryogenian strata in certain areas, and the tectonic complexity of the SCB (SI). Critically, these uncertainties do not compromise the sampling strategy and the interpretation of the geochemical data, which are central to this study.

These five studied IFs are interlayered with sandstones and siltstones (Fig. 1b) and have experienced localized low-greenschist facies metamorphism into slate and phyllite52,53. Notably, these IFs lack glacial indicators such as diamictites, faceted or striated clasts, and ice-rafted dropstones within their successions. The absence of clear glacial influence suggests their formation during the waning phase between two Sturtian glacial episodes2,16,19,20,21 (Fig. 1b; Supplementary Fig. 1). This study focuses on distinctly banded IF lithologies, with banding evident in both hand specimens and microscopic observations (Supplementary Figs. 2, 3). Mineralogically, the Sanjiang IF is primarily composed of hematite and quartz (Supplementary Fig. 2a, b). In contrast, the other four IFs (Tongdao, Jiangkou, Qidong, and Xinyu) contain both hematite and magnetite. The proportion of magnetite progressively decreases from the Xinyu IF, through the Tongdao IF, to the Jiangkou and Qidong IFs (Supplementary Figs. 2, 3). Additionally, the four magnetite-containing IFs all contain Mn-rich carbonates, aluminosilicates and apatite (Supplementary Figs. 2, 3). Pyrite is predominantly observed in the Xinyu IF (Supplementary Fig. 3g, i, k). A detailed description of petrological characteristics is provided in the Supplementary Information.

Ni, Fe and C isotopic data

Magnetite from the Tongdao, Jiangkou, Qidong, and Xinyu IFs exhibits a wide range of Ni isotopic compositions, with δ60Ni values varying from −0.01‰ to 2.09‰, predominantly falling within the range of −0.01‰ to 1.24‰ (Figs. 2b and 5d; Supplementary Data 1). A similar variability is observed in δ56Fe values across all selected hematite and magnetite samples, with values ranging from 0.20 to 2.27‰ and −0.20 to 2.25‰, respectively (Figs. 2a and 5e; Supplementary Fig. 7c; Supplementary Data 2). Furthermore, the δ56Fe values of coexisting hematite and magnetite pairs from fifteen IF samples exhibit remarkable consistency, with a difference less than 0.15‰, demonstrating a clear linear correlation (Fig. 2a). Bulk-rock δ13CV-PDB values from the IFs range between −13.2‰ and −6.7‰ (Supplementary Fig. 8c, d; Supplementary Data 3). In-situ δ13CV-PDB values of carbonates exhibit greater variability, indicating small-scale heterogeneity, ranging from −17.5‰ to −4.6‰ (Supplementary Fig. 9; Supplementary Data 4).

Fig. 2: Fe and Ni isotope systematics in magnetite and hematite from the studied Sturtian iron formations.
figure 2

a Variations in Fe isotope compositions between coexisting hematite and magnetite, highlighting distinct trends that delineate diverse pathways of magnetite formation30. The Fe isotope fractionations for equilibrium hematite and magnetite formation are illustrated in blue30,107. The gray line labeled “1:1” represents the origin of magnetite through the in-situ dissimilatory iron reduction (DIR) of the common precursor Fe(OH)322,27. b, c Cross-plots of δ60Ni versus δ56Fe values for magnetite samples. Plot (c) displays average δ60Ni and δ56Fe values. The δ60Ni and δ56Fe values of magnetite exhibit a positive correlation. Error bars for the δ56Fe and δ60Ni data represent two standard deviations (2 SD).

Full size image

Variations and correlations of elements and isotopes

The δ56Fe and δ60Ni values in magnetite exhibit a positive correlation (Fig. 2b, c). Negative correlations were observed between the contents of Ni, Co, Y/Ho, Mn, and both δ60Ni and δ56Fe values in the studied magnetite samples (Figs. 3a, b, e, f and 4a–c; Supplementary Fig. 6a). In contrast, the Dy/YbPAAS ratios display a positive correlation with both δ60Ni and δ56Fe values in magnetite (Figs. 3d, 4d). Additionally, an anticorrelation was observed between Zn content and δ60Ni values in magnetite (Fig. 3c). Nickel exhibits positive correlations with Co, Zn, and Y/Ho (Fig. 3h, i; Supplementary Fig. 6b), but an inverse relationship with Dy/YbPAAS in magnetite (Fig. 3g). Magnetite from the Tongdao and Xinyu IFs displays significantly higher Ni, Co, and Y/Ho levels (Fig. 5a, b; Supplementary Fig. 7b), but lower Dy/YbPAAS ratios, δ56Fe, and δ60Ni values, compared to the Jiangkou and Qidong IFs (Fig. 5c–e). Magnetite from the Xinyu IF exhibits the highest levels of Ni and Co, yet the lowest Dy/YbPAAS ratios, δ56Fe, and δ60Ni values, except for a few outliers.

Fig. 3: Correlations among Ni isotopes, metal elements, and REY patterns in magnetite from the studied Sturtian iron formations.
figure 3

a–c, e, f Cross-plots of Ni, Co, Zn, and Mn contents, as well as Y/Ho ratios, versus δ60Ni, illustrating anticorrelations between these element contents and ratios and the δ60Ni values. d Plot of Dy/YbPAAS versus δ60Ni, demonstrating a positive correlation between Dy/YbPAAS ratios and δ60Ni values. Here, Dy/YbPAAS represents the ratio of middle rare earth elements (MREE) to heavy rare earth elements (HREE). Rare earth element data are normalized to post-Archean Australian shale (subscript PAAS). g Plot of Ni contents versus Dy/YbPAAS. An anticorrelation exists between Ni contents and MREE/HREE ratios. h, i Plots of Ni contents versus Co and Zn contents, showing positive correlations among the variables. Error bars for the δ60Ni data represent two standard deviations (2 SD).

Full size image
Fig. 4: Correlations of metal elements and REY patterns with Fe isotope compositions in magnetite from the studied Sturtian iron formations.
figure 4

a–c Plots of Ni, Co, Mn contents versus δ56Fe values, illustrating anticorrelations between the parameters and δ56Fe values. d Plot of Dy/YbPAAS ratios versus δ56Fe values. A positive correlation exists between the ratios of MREE to HREE and δ56Fe values. Error bars for the δ56Fe data represent two standard deviations (2 SD).

Full size image
Fig. 5: Distributions and mineral-specific differences of metal elements, REY patterns and isotope compositions from the studied Sturtian iron formations.
figure 5

a–e Box plots illustrating the distribution of metal elements, REY, as well as Ni and Fe isotopes in magnetite. a Ni contents (n = 34), b Co contents (n = 34), c Dy/YbPAAS ratios (n = 34), d δ60Ni values (n = 24), and e δ56Fe values (n = 34). Boxes represent the interquartile range (IQR = Q3 − Q1), encompassing the central 50% of the data. The mean is shown as a colored solid circle, and the median (Q2) as a colored solid line within the box. Outliers are indicated by light gray circles. Extreme outliers, defined as exceeding double the IQR from Q2, are excluded from the visualization. f–h Comparative plots of different IFs were generated to investigate the relationships between Ni (f) and Co (g) contents, as well as the Dy/YbPAAS (h) ratios in coexisting hematite and magnetite.

Full size image

It is noteworthy that nearly half of the analysed magnetite shows high molar Ni/Fe ratios, ranging from above 0.0001 to nearly 0.0006 (Supplementary Fig. 10; Supplementary Data 5). Wherever hematite is present alongside magnetite, the latter exhibits more enrichment in Ni and Co (Fig. 5f, g). Additionally, most magnetite samples have higher Y/Ho ratios and Zn contents than hematite (Supplementary Fig. 7d, e), but lower Dy/YbPAAS ratios (Fig. 5h). The carbonates typically have elevated Mn contents (Supplementary Fig. 8a, b; Supplementary Data 6), following a trend similar to Ni, Co, and Y/Ho in magnetite, but contrasting with δ56Fe and δ60Ni trends in magnetite. Carbonates in the Xinyu IF exhibit the highest MnO contents and the highest corresponding bulk-rock δ13CV-PDB values (Supplementary Fig. 8b–d). In-situ δ13CV-PDB values of calcite, ankerite, and manganosiderite correlate positively with their respective MnO contents (Supplementary Fig. 11a). Interestingly, an inverse correlation is observed between the bulk-rock δ13CV-PDB values of Mn-rich carbonates and the δ60Ni values of magnetite (Supplementary Fig. 11b, c).

Discussion

The presence of finely-crystalline and platy hematite, characterized by distinct banding and hematite inclusions within ankerite (Supplementary Figs. 2b, f–h, l–n and 3c, e, j), in the Sturtian-aged IFs of the Nanhua Basin suggests its formation through the diagenetic transformation of Fe(OH)3 precursors22,30,55. These precursors likely precipitated in the upper water column via oxidation of dissolved Fe2+, subsequently settling on the seafloor as precursor sediments for IFs18,22 (Fig. 6a, b). Their transformation involved dewatering and silica release (e.g., 2Fe(OH)3 → Fe2O3 + 3H2O)16,25, and/or diagenetic replacement of Fe(OH)318 within sediment porewater, ultimately forming hematite (Fig. 6b).

Fig. 6: Conceptual illustration of biogeochemical cycling during the waning stage of the Sturtian glaciation in the South China Block.
figure 6

a The studied Sturtian-aged IFs were deposited during deglaciation, with minimal glacial influence, allowing for open-water conditions. This created favorable conditions for direct O₂ exchange between the atmosphere and ocean, enhanced nutrient (PO43⁻ and SO42⁻) influx from continental weathering, sustained oxygenic photosynthesis and primary productivity recovery, and facilitated ocean stratification, with an anoxic, ferruginous deep layer and an oxygenated surface where the IFs formed. Meanwhile, CH4 emissions into the atmosphere likely contributed to ice sheet melting or sustained the waning phase between Sturtian glacial episodes. b, c Under these conditions, metal-driven anaerobic oxidation of methane (AOM), dissimilatory iron reduction (DIR), and dissimilatory manganese reduction (DMR) occurred within anoxic sediment porewater (Eqs. 1–4). These processes involved the reduction of Fe(OH)3 and MnO2 formed in the upper oxic water column, coupled with varying degrees of methanogens and anaerobic methane-oxidizing archaea (ANME) activity, as well as organic matter oxidation, syntrophic interactions with dissimilatory Fe3+ and Mn4+-reducing bacteria, leading to the release of dissolved Fe²⁺ and Mn²⁺ into porewater. Subsequently, the reaction of Fe(OH)3 with Fe2+ led to the formation of magnetite. During these processes, ANME and methanogens preferentially absorbed Ni, Co, Zn, HREE, and light Ni isotopes, which were later released into porewater and incorporated into magnetite. Meanwhile, biologically produced HCO3⁻ mixed with seawater-derived HCO3⁻, interacting with Ca, Mg, Fe, and Mn cations, led to the formation of 13C-depleted Mn-rich carbonates. The positions of the IFs are schematic. The palaeogeographic framework follows Fig. 1a and reflects previously discussed provisional constraints (see Fig. 1a caption, main text and SI).

Full size image

Several lines of evidence seemingly suggest that the studied magnetite, ankerite, and manganosiderite originated from in situ DIR and dissimilatory manganese reduction (DMR) of Fe(OH)3 and MnO2 precursors within sediment porewater during early diagenesis. Here, DIR and DMR are specifically referred to as pathways for Fe and Mn cycling, where microbes utilize organic carbon as an electron donor to reduce Fe3+ and Mn4+ as terminal electron acceptors, representing potentially ancient respiratory processes56. These microbial activities facilitate the release of aqueous Fe2+ and Mn2+ into anoxic sediment porewater via the following reactions22,26,28,29,30 (Eqs. 1 and 2; Fig. 6b, c):

$${{{{\rm{CH}}}}}_{3}{{{{\rm{COO}}}}}^{\mbox{-}}+8{{{\rm{Fe}}}}{({{{\rm{OH}}}})}_{3}\to {8{{{\rm{Fe}}}}}^{2+}+{2{{{{\rm{HCO}}}}}_{3}}^{\mbox{-}}+15{{{{\rm{OH}}}}}^{\mbox{-}}+{5{{{\rm{H}}}}}_{2}{{{\rm{O}}}}$$
(1)
$${{{{\rm{CH}}}}}_{3}{{{{\rm{COO}}}}}^{\mbox{-}}+4{{{\rm{MnO}}}}2+{3{{{\rm{H}}}}}_{2}{{{\rm{O}}}}\to {4{{{\rm{Mn}}}}}^{2+}+{2{{{\rm{HCO}}}}}_{3}^{\mbox{-}}+{7{{{\rm{OH}}}}}^{\mbox{-}}$$
(2)

Subsequently, the reaction between Fe(OH)3 and microbially produced Fe2+ facilitated magnetite formation (e.g., 2OH− + Fe2+ + 2Fe(OH)3 → Fe3O4 + 4H2O)26,29, while Mn2+ − from a precursor MnO2 phase that also formed in the upper water column − was primarily incorporated into carbonates57,58,59.

This interpretation is further supported by petrographic textures and C and Fe isotopic data, which collectively indicate in situ formation of magnetite and Mn-rich carbonates during early diagenesis. First, the magnetite displays euhedral to subhedral morphologies, with sizes ranging from tens to hundreds of microns, and is located within hematite-rich bands (Supplementary Figs. 2f, g, n–p and 3e, j). These bands are oriented parallel to or cross-cutting the layering, consistently aligning with sedimentary bedding, suggesting an authigenic origin for the magnetite. Experimental studies have demonstrated that finely-crystalline magnetite produced through DIR, typically a few hundred nanometers or smaller upon formation, can grow to tens of microns or larger29. Second, the occurrence of disseminated, micro-sized hematite inclusions within euhedral Mn-rich ankerite rhomboids (Supplementary Fig. 2h) likely reflects in-situ DIR and DMR during the early stage of IF formation28. The studied Mn-rich carbonates exhibit low δ13CV-PDB values, ranging from −17.5‰ to −4.6‰. These values have traditionally been attributed to mixing of dissolved inorganic carbon (DIC) (δ13CV-PDB = ca. 0‰) with remineralized organic carbon, which typically has δ13CV-PDB values of ca. −20‰ to −30‰22,28. Third, the consistent Fe isotopic signatures of coexisting hematite and magnetite suggest a common precursor. During early diagenesis, Fe(OH)3 transforms into hematite within the sediments while retaining its original δ56Fe values. In the δ56FeHematite – δ56FeMagnetite diagram30 (Fig. 2a), it is evident that magnetite and hematite did not achieve Fe isotopic equilibrium. However, the δ56Fe values of magnetite exhibit a strong positive correlation with those of co-existing hematite, aligning along a 1:1 line (Fig. 2a). This relationship indicates that the synthesis of magnetite by the reaction between hydrothermal Fe2+ and Fe(OH)3 is unlikely, as this scenario would be expected to plot along a line with a slope of two-third in Fig. 2a, where one-third of Fe in magnetite would come from hydrothermal sources (δ56Fe = 0‰). The most parsimonious explanation is that magnetite formed in the sediments via in-situ DIR of Fe(OH)3 precursors, predominantly inheriting the Fe isotope composition of the original Fe(OH)327,30. The pronounced Mn enrichment in 13C-depleted authigenic carbonates likely indicates the DMR in the Sturtian-aged IFs of the Nanhua Basin38,57,58,59,60.

Nevertheless, some geochemical features observed in the studied magnetite and Mn-rich carbonates cannot be explained by DIR and DMR alone, suggesting the involvement of additional processes. In particular, DIR and DMR processes fail to account for the considerable variability in δ60Ni values (which can drop to −0.01‰), the inverse relationships between Ni, Co, Zn, Mn and δ60Ni values, the inverse correlation between Ni and Dy/YbPAAS, and the positive correlation between Dy/YbPAAS and δ60Ni values in the studied magnetite (Figs. 2b, c, 3a–g and 5d). Additionally, they cannot adequately explain the positive correlation between the in-situ δ13CV-PDB values and MnO contents in carbonates (Supplementary Fig. 11a), nor the inverse correlation between bulk-rock δ13CV-PDB values of Mn-rich carbonates and magnetite δ60Ni values (Supplementary Fig. 11b, c).

Recent studies have highlighted the role of Fe- and Mn-dependent AOM in driving the formation of authigenic minerals in ferruginous and methanic sedimentary environments. Authigenic magnetite, coexisting with Fe(OH)3 and authigenic carbonates, has been increasingly identified in Fe-rich marine sediments characterized by high CH4 fluxes at CH4 seeps61,62. The spatial association between methanogens and magnetite indicates that methanogenic archaea mediate Fe reduction, promoting magnetite formation61,62. Fe(OH)3 reduction coupled to AOM has also been observed in the methanic zone of ferruginous modern marine sediments, lacustrine environments, and culture-based experiments26,34,40,46,62,63. Furthermore, methanogens in the sediments of the methanic zone have been identified as key catalysts for in-situ Fe-driven AOM, highlighting their crucial role in this geochemical process63,64,65. These findings contribute to the understanding of magnetite formation mechanisms in Fe-rich environments and help explain the global increase in Fe2+ concentrations in methanic marine sediments64,66. Additionally, Mn-dependent AOM is proposed as a major CH4 sink in CH4 seep sediments of SO42--depleted early Earth oceans, particularly during the Marinoan deglaciation, facilitating the deposition of Mn-enriched, 13C-depleted carbonates38,67.

Methanogenesis represents one of the most metal-dependent enzymatic pathways in biology44. Nickel, an essential metal in this process, is a crucial component of the porphyrin ring in methyl-coenzyme M reductase (MCR), a key enzyme in methanogenesis, highlighting its unique metal dependence within this enzyme class48,49,68. Cobalt is present in cobamides, which play a role in methyl group transfer, while Zn serves as a structural atom in several enzymes45,69. AOM by ANME proceeds through a reverse methanogenesis pathway, as ANME possess analogous enzymes to those in the methanogenic pathway44,45,48,49. Similar to methanogens, ANME require significant quantities of Ni and Co, with lesser amounts of Zn42,44. Furthermore, continental weathering induces Ni isotope fractionation, preferentially releasing isotopically heavier Ni into solution70. Consequently, modern global average seawater exhibits relatively heavy Ni isotopic values, with a δ60Ni value of ca. 1.44‰71,72. In contrast, both ANME and methanogens preferentially utilize lighter Ni isotopes, fractionating δ60Ni as low as −1.46‰49,68. To date, there is no evidence that other types of microbes are capable of substantially fractionating Ni isotopes5,49,68,69.

The integration of Ni, Co, and REY abundances, along with Ni isotopes, has recently been applied to study CH4-related biogeochemical cycles. For example, Zhao et al.5 observed a negative correlation between Ni contents and δ60Ni values, as well as between middle REE (MREE)/HREE ratios and Ni contents in pyrite of the Nantuo Formation, suggesting active methanogenesis during the Marinoan deglaciation. Similarly, Chen et al.45 suggested that enrichment of Co and Ni, along with low δ60Ni values in pyrite from modern marine sediments, serves as indicator of AOM. Thus, metal-driven AOM likely played a crucial role in transforming Fe(OH)3 and MnO2 into magnetite and Mn-rich carbonates of the Sturtian-aged IFs examined here. Fe(OH)3 and MnO2 have been identified as potential electron acceptors for AOM, where diagenesis induced by the upward migration of CH4 could have occurred through the following reactions26,38,42,60,67 (Eq. 3 and 4; Fig. 6b, c):

$${{{{\rm{CH}}}}}_{4}+{8{{{\rm{Fe}}}}({{{\rm{OH}}}})}_{3}+{15{{{\rm{H}}}}}^{+}\to {{{{\rm{HCO}}}}}_{3}^{\mbox{-}}+{8{{{\rm{Fe}}}}}^{2+}+{21{{{\rm{H}}}}}_{2}{{{\rm{O}}}}$$
(3)
$${{{{\rm{CH}}}}}_{4}+{4{{{\rm{MnO}}}}}_{2}+{7{{{\rm{H}}}}}^{+}\to {{{{\rm{HCO}}}}}_{3}^{\mbox{-}}+{4{{{\rm{Mn}}}}}^{2+}+{5{{{\rm{H}}}}}_{2}{{{\rm{O}}}}$$
(4)

The low δ60Ni values and enrichment of Ni, Co, and Mn in the analysed magnetite and carbonates may be complicated by abiotic Fe(OH)3 and MnO2 reduction and dissolution during metal-driven AOM. Fe(OH)3 precipitation has been shown to preferentially incorporate light Ni isotopes and MREE5,73,74,75, while MnO2 serves as an efficient sorbent for Co, Ni, and MREE, with a preference for adsorbing holmium (Ho) over yttrium (Y) from seawater60. However, reported Ni isotopic fractionation during Fe(OH)₃ uptake typically reaches up to −0.77‰75. If this process played a significant role in the present study, the minimum expected δ60Ni values for magnetite would be ca. 0.67‰, which cannot account for the significantly lower δ60Ni values observed in this study. Moreover, the analysed magnetite lacks the characteristic “MREE-bulge” pattern, defined by Dy/YbPAAS ratios >1, which is typically associated with the reductive dissolution of Fe(OH)3 and MnO2 and the preferential release of MREE into porewater60. Instead, the studied magnetite exhibits relative HREE enrichment over MREE, as reflected in their low Dy/YbPAAS ratios (Fig. 5c; Supplementary Data 5). Significantly, magnetite typically exhibits lower Dy/YbPAAS ratios, higher Ni and Co contents, and higher Y/Ho ratios compared to coexisting hematite (Fig. 5f-h; Supplementary Fig. 7e). Furthermore, magnetite and carbonates in the Xinyu IF show the highest Mn enrichment, along with significant HREE enrichment and the highest Y/Ho ratios. Collectively, these observations do not support the release of these elements and isotopes from Fe(OH)3 and MnO2.

Alternatively, metal-driven AOM in the Sturtian-aged IFs of the Nanhua Basin was due to a complex microbial-mineral interaction network, requiring Fe-reducing microorganisms and reactive Fe(OH)3 and MnO2 in sediments34,36,42,46,49,64,66,76,77. ANME archaea have been recognized as critical microbial agents of metal-driven AOM in the methanic zones of marine sediments63. These ANME groups may have contributed to Fe(OH)3 and MnO2 reduction during CH4 oxidation, typically engaging in syntrophic interactions with dissimilatory Fe3+ and Mn4+-reducing bacteria34,42,49,77. Furthermore, methanogens not only produce CH4 but they play two pivotal roles in sedimentary biogeochemical cycles. First, they can directly reduce Fe(OH)376, facilitating a rapid metabolic transition from methanogenesis to Fe(OH)3 reduction64,66. Second, methanogens produce intermediate compounds essential for ANME, thus contributing significantly to the AOM process64,78. Thus, metal-AOM in this context was likely driven by a diverse microbial community, including ANME, methanogens, and Fe3+ and Mn4+-reducing bacteria in anoxic, ferruginous environments63,64,66. This is supported by observations of concurrent methanogenesis and metal-driven AOM in active CH4 seepage areas42,79.

During metal-driven AOM, varying degrees of ANME and methanogen biomass degradation facilitate the reduction of Fe3+ to Fe2+ and Mn4+ to Mn2+, mobilizing adsorbed Ni, Co, Zn, HREE, and light Ni isotopes into porewater (Fig. 6b, c). Subsequently, the precipitation of authigenic magnetite in sediment porewater incorporates these elements and isotopic signatures, resulting in δ60Ni values as low as −0.01‰, negative correlations between Ni, Co, Zn, Mn contents and δ60Ni values, and a positive correlation between Dy/YbPAAS ratios and δ60Ni values in the studied magnetite. Additionally, Mn2+ was primarily incorporated into the analysed 13C-depleted authigenic carbonates through microbial MnO2 reduction. These Mn-enriched carbonates, with MnO contents ranging from 0.25 wt.% to 21.48 wt.%, exhibit a wide range of δ13CV-PDB values from −17.5‰ to −4.6‰. These C isotope signals are considered to reflect a mixing of multiple DIC sources, including seawater DIC (δ13CV-PDB = ca. 0‰), AOM-derived DIC (δ13CV-PDB = ca. −60‰), and residual DIC pool in methanogenic zones, which can exhibit extreme 13C-enrichment (up to 50‰ to 70‰) due to CO2 fractionation during archaeal methanogenesis28,67,79,80,81,82. Despite the generally low total organic carbon (TOC) contents (<1.0 wt.%) in the Fulu Formation and its equivalents, the contribution of organic matter-derived DIC (δ13CV-PDB = − 20‰ to −30‰) cannot be entirely ruled out21,83. In low SO42− anoxic environments, organic matter degradation occurs via microbial methanogenesis, DIR, and DMR, with methanogenesis being roughly three times more prevalent than the latter processes combined40,44,84. While the exact proportions are uncertain, these mixed DIC sources likely infiltrated sediment porewater and were incorporated into authigenic carbonates, as supported by the positive correlation between in-situ δ13CV-PDB values and MnO contents in carbonates and the inverse correlation between magnetite δ60Ni values and bulk-rock δ13CV-PDB values of carbonates.

Our findings indicate that CH4-related process played significant role in the precipitation of Sturtian-aged IFs in the Nanhua Basin. However, spatial heterogeneity in CH4 distribution and the intensity of metal-driven AOM activity likely varied significantly across different IF sites. These variations are thought to be significantly influenced by the bioavailability of Ni, Co, and Zn44. Accordingly, the Xinyu IF likely exhibited the highest CH4 levels and most extensive AOM activity, followed by the Tongdao IF. In contrast, the Jiangkou and Qidong IFs appear to have experienced lower CH4 levels and weaker AOM. This is supported by the highest Ni and Co contents, along with the lowest Dy/YbPAAS ratios and δ60Ni values observed within the Xinyu IF, an intermediate status in the Tongdao IF, and contrasting patterns in the Jiangkou and Qidong IFs (Figs. 3 and 5a–d). While no clear trend is observed for Zn, it is most abundant in the Xinyu IF (Supplementary Fig. 7a). At Xinyu, the most extensive CH4-related activity explains the highest δ13CV-PDB values and Mn contents in carbonates, along with the lowest δ60Ni values in magnetite. This correlation arises as methanogenesis can enrich the DIC pool in δ13C79,80. However, the lack of magnetite and carbonates in the Sanjing IF suggests that CH4-related processes, along with DIR and DMR, were either absent or insignificant.

Despite spatial heterogeneity, methanogenesis was active in the Nanhua Basin during the Sturtian period. Sedimentological and geochemical evidence indicates that the studied IFs were deposited during deglaciation (SI). Given CH4’s potent greenhouse properties, a key question is whether it contributed to deglaciation—a possibility that hinges on two critical factors.

First, was CH4 sufficiently abundant to influence climate? Mass-balance models of global negative δ13C excursions indicate that CH4 hydrate reservoirs, and their destabilization, were more extensive during the Neoproterozoic than at any other time in Earth’s history11,12, potentially acting as positive feedback during Marinoan deglaciation5,11,12. Although direct estimates of CH4 fluxes during the Sturtian deglaciation are lacking, elevated molar Ni/Fe ratios in magnetite—nearly half of which exceed 0.0001 and peaking at ~0.0006 (Supplementary Fig. 10; Supplementary Data 5)—suggest enhanced biogenic CH4 production, comparable to levels inferred for Archean IFs85. Consistent with this, the Precambrian ocean-atmosphere system is widely considered to have contained much higher CH4 levels than present33,34,35,38,82. Furthermore, high CH4 fluxes are also required to sustain metal-driven AOM in CH4-rich marine sediments, where the availability of terminal electron acceptors such as Fe(OH)3 and MnO2 is relatively limited34,61,62,63.

Second, could sufficient CH4 have bypassed oxidation to reach the atmosphere? In CH4-rich marine sediments, metal-driven AOM oxidizes CH4 at only ~2% the rate of sulfate-driven AOM34,63. Nevertheless, even today—where sulfate-driven AOM dominates—between 6 and 12 Tg/yr of CH4 escapes to the atmosphere, accounting for 1–10% of natural emissions13,86. Given the inherently low oxidation efficiency of metal-driven AOM and abundant coexisting hematite and magnetite in the studied IFs, it is plausible that a significant fraction of CH4 reached the atmosphere. If marine CH4 levels were indeed elevated during the Neoproterozoic, such emissions could have contributed to ice sheet melting or sustained the waning phase between Sturtian glacial episodes.

Our geological and geochemical analyses further substantiate the intricate links between CH4-related processes, Sturtian deglaciation, and Neoproterozoic environmental changes. Our Fe isotope data suggest that the studied Sturtian-aged IFs formed in a stratified ocean, characterized by an anoxic, ferruginous deep layer beneath an oxygenated surface (Fig. 6a; SI). The δ56Fe trends in hematite and magnetite reflect variable Fe2+ oxidation, suggesting spatially heterogeneous seawater redox conditions (SI). The exceptionally low δ56Fe values— approaching zero in the Xinyu IF—point to extensive Fe²⁺ oxidation under relatively more oxidizing conditions (SI). The studied magnetite exhibits a positive correlation between δ60Ni, Dy/YbPAAS, and δ56Fe values, while Co, Ni, and Mn contents show an inverse relationship with δ56Fe values (Figs. 2b, c, 4). Notably, the Xinyu magnetite shows the lowest δ60Ni, Dy/YbPAAS, and δ56Fe values, along with the highest Co, Ni, and Mn contents. These geochemical patterns suggest a strong interplay between intense methanogenesis, metal-driven AOM, and elevated O2 levels in shallow seawater. It has been proposed that the O2 in shallow seawater during the Sturtian glaciation, responsible for oxidizing Fe2+ to Fe3+, may have originated from subglacial meltwater O2 supply4 or vertical transport processes within glacial ice87. However, during Sturtian deglaciation, minimal glacial influence allowed open-water conditions, at least in the Nanhua Basin, enabling direct O2 exchange between the atmosphere and ocean (Fig. 6a). This shift is supported by evidence of enhanced continental weathering and an active hydrological cycle, as indicated by high Chemical Index of Alteration (CIA) values (65–75), consistently positive δ53Cr (0.64–1.22‰), and elevated aluminosilicate contents in the Nanhua Basin IFs16,20,88. The impact of continental weathering was particularly pronounced at Xinyu, as indicated by the highest Mn contents in magnetite and carbonates, significant pyrite abundance originating from bacterial SO42− reduction associated with organic matter decomposition52, and the highest magnetite-to-hematite ratio among the five studied IFs. This is likely due to Xinyu’s palaeogeographic proximity to the continental margin, which facilitated the accumulation of weathering products. Along with the widespread abundance of euhedral to subhedral apatite in the sedimentary layers of the Tongdao, Jiangkou, Qidong, and Xinyu IFs (Supplementary Figs. 2h–j, p, 3d, l, m), these features suggest an enhanced influx of continental weathering-derived PO43- and SO42- into seawater. These nutrients likely fostered favorable conditions for photosynthetic organisms, supporting primary productivity recovery and O2 production3,5,6 (Fig. 6a). Notably, PO43- is widely recognized as the primary limiting nutrient controlling marine primary productivity over extended geological periods89. Further supporting evidence for sustained oxygenic photosynthesis comes from the detection of various steranes—biomarkers of photosynthetic eukaryotes—in Sturtian and Marinoan deposits from the Nanhua Basin90.

In conclusion, our study reveals a CH4-related biogeochemical mechanism for Fe-bearing mineral transformation in Sturtian-aged IFs. While exact CH4 fluxes remain uncertain, our findings underscore its complex interplay with glaciation dynamics, palaeoenvironmental shifts, and biospheric evolution during the Sturtian Snowball Earth, highlighting the need for further investigation.

Methods

Sample preparation

The weathered surfaces and obvious later-stage veins of the studied five IFs samples were initially removed, followed by division into three fragments. One fragment was prepared as polished thin section for petrographic analysis, chemical composition determination, and in-situ C isotope analysis of minerals using light microscopy, EPMA (electron probe microanalyzer)-EDS (energy dispersive X-ray spectroscopy), and a RESOlution 193 nm laser ablation system coupled to a Nu Plasma II MC-ICP-MS. A second fragment was crushed in a stainless-steel mortar and then pulverized in an agate mill to approximately 200 mesh. The resulting powders were then utilized for bulk-rock C and O isotopic analyses. In the third fragment, hematite and magnetite grains were selected. After extraction, grains were rinsed three times with deionized water, crushed into a fine powder using an agate ring mill, and subsequently utilized for analyses of major and trace elements, as well as Fe and Ni isotopic analysis.

Electron microprobe analysis (EPMA) of carbonate minerals

The quantitative analysis of in-situ major elements in carbonate minerals was conducted using an electron probe microanalyzer at the Wuhan Sample Solution Analytical Technology Co., Ltd in Wuhan. The analysis was conducted using a JEOL JXA-8230 electron microprobe with a 10μm beam diameter, operated at 15 kV accelerating voltage and 5 nA beam current. Rhodonite, dolomite, and hematite were employed as standards for Mn, Ca, Mg, and Fe, respectively. The detection limit for selected elements was set at 0.01 wt. %.

In-situ C isotopic analyses on carbonates

In-situ C isotopic analyses (n = 106) of calcite, ankerite, and manganosiderite were conducted using a RESOlution 193 nm laser ablation system coupled to a Nu Plasma II MC-ICP-MS at the State Key Laboratory of Geological Processes and Mineral Resources (GPMR), China University of Geosciences (Wuhan). The instrumental operating conditions and analytical procedures for in-situ C isotope measurements of calcite, dolomite, and siderite are detailed in Chen et al.91 and Lu et al.92,93. A standard-sample bracketing (SSB) method was used to correct instrumental mass bias, with SXD8 calcite, DOL8 dolomite, and SD-5 siderite serving as external standards for the studied calcite, ankerite, and manganosiderite, respectively. Both samples and standards were ablated using a 193 μm laser beam diameter, an 8–10 Hz repetition rate, and an energy density of 3 J/cm². The analytical uncertainty for C isotopes is better than 0.5‰ (2 SD), based on repeated analyses of SXD8 calcite, DOL8 dolomite, and SD-5 siderite91,92,93. For high-Mg calcite, dolomite, and siderite, corrections were applied for critical spectral interferences of Mg2+93. Data quality was monitored through repeated analyses of the internal laboratory standards OkA153 calcite, DOL9 dolomite, and SD-1 siderite after every 10 samples, yielding an analytical precision better than 0.5‰ (2 SD). The C isotope data are reported in standard δ-notation as δ13C values, referenced to Vienna Pee Dee Belemnite (V-PDB).

Bulk-rock C and O isotopic analyses

Thirty-six bulk-rock IF samples were used for C and O isotope analyses. Approximately 1 mg of powder was reacted for more than 72 h at 72 °C with an excess of 100% H3PO4 under vacuum conditions. The evolved CO2 was sampled using a Thermo Finnigan Gas Bench II, and the isotopic ratios were measured using a MAT 253 mass spectrometer at the Beijing GeoAnalysis Technology Co., Ltd in Beijing. The isotopic measurements were calibrated against the Chinese national standards (GBW04405, GBW04406, GBW04416, and GBW04417), with analytical errors of less than 0.10‰ and 0.20‰ for δ13C and δ18O, respectively. The C and O isotope data are reported in the standard δ-notation as δ13C and δ18O values, referenced to Vienna Pee Dee Belemnite (V-PDB).

Major and trace element analyses of hematite and magnetite

Analyses of major and trace elements were conducted on thirty-four magnetite and seventeen hematite samples at Qingdao Sparta Analysis & Test Co., Ltd in Qingdao. Major element concentrations were determined using a Thermo Fisher Scientific iCAP 6300 inductively coupled plasma optical emission spectrometer (ICP-OES). Trace elements such as Ni, Co, Zn, rare earth elements and yttrium (REY) were analyzed using inductively coupled plasma mass spectrometry (ICP-MS) with a Thermo Fisher Scientific iCAP RQ instrument. The dissolution procedure was as follows: ~0.04 g of powdered sample was digested in a sealed Teflon beaker with 1.5 ml HF and 1.5 ml of a mixed acid solution of HNO3 and HCl in a 1:3 (v/v) ratio. The mixture was heated at 180 °C in an electric oven for ~12 h. After cooling, the solution was evaporated to incipient dryness on a hot plate at 150 °C. Subsequently, 1 ml of HNO3 and 3 ml of H2O were added, and the sample was further digested in a sealed beaker at 180 °C for ~5 h. Analysis of major and trace elements utilized internal calibrations against reference materials (GBW07315, GBW07316, BHVO-2, and BCR-2).

Fe isotope analysis of hematite and magnetite

Iron isotope ratios of twenty-six magnetite and twenty-two hematite samples were analyzed using a Neptune Plus Multi-Collector Inductively Coupled Plasma Mass Spectrometer (MC-ICP-MS) at the Key Laboratory of Submarine Geosciences and Prospecting Techniques, Ocean University of China, Qingdao. The details of sample dissolution, column chemistry, and instrumental analysis can be found in the works of Dauphas et al.94, Chen et al.95 and Li et al.96. Following complete dissolution, the solution was evaporated to dryness and then fluxed with 6 N HCl for column chromatography. Fe was purified using 1 ml of pre-cleaned and 6 N HCl-conditioned AG1-X8 resin (a Bio-Rad anion resin), with elution of Fe performed using 0.4 N HCl after a 10 ml 6 N HCl wash95. The Fe solutions were diluted with 2% HNO3 to a concentration of 2 ppm and bracketed with 2 ppm IRMM14. The associated error for each datum is 2 standard deviations (2 SD), with a long-term external precision of ± 0.06‰ for δ56Fe. In this study, Fe isotopes were analyzed in geological reference materials, including the basalt standard JB-2 (0.09 ± 0.02‰) and andesite standard JA-2 (0.14 ± 0.04‰). The δ56Fe values for all reference materials agree with published data97. Fe isotope compositions of the remaining eight magnetite and three hematite samples were analyzed using a Neptune Plus MC-ICP-MS at Wuhan Center, China Geological Survey. The analytical procedures are detailed in Sossi et al.98 and Hu et al.99. Fe was separated using Bio-Rad AG-MP-1M strong anion exchange resin with a recovery rate exceeding 95%. The long-term external reproducibility (2 SD) of Fe isotope analysis is better than ±0.08‰ for δ56Fe, as demonstrated by repeat analyses of multiple Fe isotope standard solutions. The measured Fe isotope compositions of international basalt standards BCR-2, with a δ56Fe of 0.04 ± 0.03‰, and BHVO-2, with a δ56Fe of 0.13 ± 0.06‰, are in line with recommended values, considering analytical uncertainties97,100. Fe isotope compositions are reported using conventional delta notation relative to the IRMM-14 standard: δ56Fe (‰) = [(56Fe/54Fe) sample/(56Fe/54Fe) IRMM-014 – 1] ×1000.

Ni isotopic analysis of magnetite

Nickel isotopic analyses were conducted on twenty-four magnetite samples at the Isotope Geochemistry Laboratory of China University of Geosciences, Beijing. Around 5–100 mg of magnetite powders were digested using a concentrated mixture of HF-HNO3-HCl acid to ensure full dissolution. Aliquots of sample solutions containing 500–800 ng Ni were spiked with a 61Ni – 62Ni double spike to reach an optimal spike-sample ratio of 64:36. The mixtures were subsequently refluxed on hotplates to guarantee sample-spike equilibration prior to column chemistry. Nickel separation from matrices was successfully accomplished through a four-stage chromatography procedure101,102,103. Briefly, the first column used AG1-X8 resin in a media of 6 M HCl to separate Ni from Fe and other matrix elements. The second column applies a mixture of 15 vol.% 10 M HCl and 85 vol.% acetic acid using cation exchange resin Bio-Rad AG50W-X8 to separate Ni from elements such as Mg, Al, Ca, and Ti. In the third separation step, a column was packed with 0.4 ml of AG1-X8 anion exchange resin (Bio-Rad Laboratories, 200 – 400 mesh size) to separate Ni from Na, K, Mn, Cr, V, and other matrix elements103. Finally, the first separation step is repeated to further purity the sample if necessary. The total column procedural blank for Ni was <3 ng, which is negligible compared to 500–800 ng Ni from samples that were processed through the columns.

Nickel isotopic data were determined on a Thermo Scientific® Neptune Plus MC-ICP-MS in medium resolution mode with an Aridus II introduction system. The Neptune MC-ICP-MS was operated under a medium mass resolution mode to separate polyatomic interferences such as 40Ar18O+ and 40Ar17OH+ on 58Ni. The four Ni isotopes (62Ni, 61Ni, 60Ni, 58Ni) together with 57Fe were measured simultaneously on separate Faraday cups (H2, H1, Axial, L1, and L3). The measurement of 57Fe was used to correct for interference from 58Fe on 58Ni. No detectable peak for polyatomic interferences such as 40Ar18O+ and 40Ar17OH+ was observable on masses 58 or 57. The background Ni signal for 60Ni was <10−3 V, which is negligible relative to the sample signals of ~7–8 V (for 100 ppb solutions). Each sample analysis was bracketed by measurements of a double-spiked NIST SRM 986 nickel solution at similar concentration. The raw data were processed using double-spike equations to correct for isotope fractionation resulting from instrument-induced mass bias and imperfect column yields. The Ni isotopic ratio is presented in the delta (δ) notation as per mil deviation (‰) relative to SRM 986: δ60∕ 58Ni (‰) = (60Ni/58Ni) sample∕(60Ni/58Ni) SRM⋅986 − 1. When processed alongside the samples, reference materials NOD-A-1 and NOD-P-1 yield δ60Ni values averaging +1.06 ± 0.04‰ (n = 9; 2 SD) and +0.34 ± 0.06‰ (n = 9; 2 SD), respectively, consistent with data from previous studies70,71,104,105. During the course of this study, the external precision, as determined by replicate analyses of the reference materials, was better than 0.06‰ (2 SD).