The Roles of Various Types of Crustal Contamination in the Genesis of the Jinchuan Magmatic Ni-Cu-PGE Deposit: New Mineralogical and C-S-Sr-Nd Isotope Constraints

Magmatic sulfide deposits are major sources of nickel and platinum group elements (PGEs) for the world (Barnes and Lightfoot, 2005; Song et al., 2009; Naldrett, 2010; Li et al., 2019). Early sulfide saturation in mantle-derived ultramafic and mafic magmas is critical in the formation of high Ni tenor sulfide ores (Naldrett, 1999; Arndt et al., 2005; Ripley and Li, 2013). Experimental studies have demonstrated that the sulfur concentration at sulfide saturation (SCSS) in magma is controlled by intrinsic melt properties such as temperature, total pressure, oxygen fugacity (⁠$fO2$⁠), and oxide contents, especially FeO (Haughton et al., 1974; O’Neill and Mavrogenes, 2002; Jugo et al., 2005a, b, 2010; Li and Ripley, 2009). Due to the negative effect of total pressure on SCSS, a sulfide-saturated magma from the mantle will become undersaturated during magma ascent (O’Neill and Mavrogenes, 2002). Thus, other processes that can change magma composition significantly are required to induce sulfide saturation in the magma again (Naldrett, 2010; Ripley and Li, 2013). These include assimilation of siliceous crustal rocks (Lightfoot and Hawkesworth, 1997; Li and Naldrett, 2000; Ripley and Li, 2013) and addition of externally derived sulfur (Lambert et al., 1998; Ripley et al., 2002; Keays and Lightfoot, 2010). If the mantle-derived magma is highly oxidized (⁠$fO2$ >QFM+1.0), such as some arc basaltic magmas, reduction of the magma by addition of organic matter from country rocks to the magma may be required to induce sulfide saturation in the magma (e.g., Tomkins et al., 2012; Xue et al., 2021).

The cause of sulfide saturation in the parent magma of the Jinchuan magmatic sulfide deposit, one of the single largest magmatic Ni-Cu deposits in the world, has been debated for a long time (see summary in Li and Ripley, 2011). Many researchers suggested that it was triggered by fractional crystallization that took place prior to the final emplacement of the magma as well as contamination with siliceous and sulfidic crustal materials (Chai and Naldrett, 1992a, b; Li et al., 2004; Ripley et al., 2005; Song et al., 2007, 2009, 2012; Duan et al., 2016). In contrast, Lehmann et al. (2007) proposed that sulfide saturation in the Jinchuan magma resulted from magma interaction with marble wall rocks. Their idea is that addition of CO₂ to the magma during decarbonation in the contact zone increased the $fO2$ of the magma, thereby decreasing the FeO/Fe₂O₃ ratio in the magma and ultimately triggering sulfide saturation in the magma (Tang et al., 2018; Ding et al., 2021), because SCSS decreases with decreasing FeO content in the magma (O’Neill and Mavrogenes, 2002; Li and Ripley, 2005, 2009; Liu et al., 2007; Ariskin et al., 2013; Wykes et al., 2015). To evaluate the competing genetic models, we have carried out a detailed mineralogical study of the contact zone and hybrid rocks, as well as collected additional Sr-Nd-C-S isotope data and S/Se for the intrusive rocks (nonhybrid ultramafic rocks), hybrid rocks, and different types of country rocks. In addition, we have used the olivine-sulfide oxygen barometer to determine the change of oxidation state of the magma after its interaction with marbles. Based on the new results, we can rule out marble assimilation as the main cause of sulfide saturation in the Jinchuan magma. The isotope data strongly support the notion that contamination with siliceous and sulfidic crustal materials in the magma plumbing system beneath the Jinchuan intrusion played a critical role in triggering sulfide saturation in the magma.

The Jinchuan magmatic Ni-Cu-PGE sulfide deposit is located in the eastern part of the elongate Longshoushan thrust belt at the southern margin of the Alxa block, the westernmost part of the North China craton (Fig. 1a). The main landmark in southern Alxa is a NE-trending Longshou mountain range, which is commonly referred to as the Longshoushan terrane (Fig. 1b). The Longshoushan terrane is bounded by two E-W–trending regional thrust faults, and to the south, it is bounded by the North Qilian orogenic belt. The terrane is mainly composed of Proterozoic metamorphic rocks from greenschist to amphibolite facies (mainly migmatites, gneisses, and marbles) overlain by Late Mesoproterozoic sedimentary rocks that have experienced greenschist-grade metamorphism, Paleozoic metasedimentary rocks (slates and sandstones), and Paleozoic granitoids (Fig. 1b). Previous studies have shown two episodes of mafic magmatism in the Longshoushan terrane. The older episode occurred at ~830 Ma, as represented by the Jinchuan ultramafic intrusion (Li et al., 2005; Zhang et al., 2010), and the younger one occurred at ~420 Ma as represented by the Xijing mafic intrusion and mafic dikes (Duan et al., 2016).

The Jinchuan deposit contains more than 500 million metric tons of ore with 1.1 wt % Ni and 0.7 wt % Cu. The ore-hosted ultramafic intrusion was emplaced as a sill-like body with length of ~6,500 m, width of ~300 m, and vertical downward extension >1,100 m in its central part. This intrusion is divided into four segments by a series of NE-trending strike-slip faults. These segments were numbered as III, I, II, and IV from west to east by the mine geologists based on the order of mine development (Fig. 2). Segment I is exposed and located to the west of segment II. Segment II has the largest surface exposure and is located in the middle portion of the Jinchuan intrusion. The unexposed segments III and IV are emplaced at the western and eastern ends of the Jinchuan intrusion, respectively. The Jinchuan intrusion is dominated by ultramafic rocks such as lherzolite, plagioclase lherzolite, dunite, harzburgite, and olivine websterite. The ultramafic rocks in segments III and IV contain more abundant plagioclase than those that occur in segments I and II. Sulfide orebodies occur over a range of depths in the ultramafic intrusion from surface down to the maximum ~1,200 m below surface.

The Jinchuan ultramafic intrusion intruded into a Proterozoic metamorphic rock suite (Fig. 2). The suite in the region is characterized by interbedded amphibolites, granitic gneisses, schists, and marbles. From lower to upper parts, the suite in the west portion of the Jinchuan intrusion (profile A-A’) is plagioclase amphibolite → marbles → plagioclase amphibolite + graphite-bearing quartz-schist → granitic gneiss → marbles → calc-schist → granitic gneiss → marbles. The suite in the middle portion of the intrusion (profile B-B’) is plagioclase amphibolite → granitic gneiss + graphite-bearing quartz schist → marbles → plagioclase amphibolite + quartz-schist → quartz schist → granitic gneiss → marbles → graphite-bearing calc-schist.

The northern margin of the ultramafic intrusion is in contact with marbles, which are preserved as layers of variable thickness (Fig. 2). Interaction between magma and marbles at the contact is marked by a 2- to 3-m-wide zone of hybrid rocks (Lehmann et al., 2007; Ding et al., 2021). New drill cores at Jinchuan have revealed a large hybrid rock zone up to 12 m in width in places (Fig. 2). The hybrid rocks are generally characterized by mafic minerals mixed with carbonate patches or carbonate-rich matrices. The abundances of carbonate minerals in the hybrid rocks increase toward the marble side. Some hybrid rocks from the intrusion side contain semimassive sulfides, disseminated sulfides, or sulfide veins, whereas those from the marble side generally contain fine sulfide grains surrounded by carbonate-rich matrix. Marble xenoliths are commonly developed along the contact zone and appear sporadically throughout the intrusion (Ding et al., 2021).

The metamorphic sequence to the south of the Jinchuan intrusion mainly consists of granitic gneisses and plagioclase amphibolites, which contain varying amounts of pyrite grains (Fig. 2). These two types of country rocks commonly host layers and lenses of schists. The schists are calcareous or felsic, and locally contain disseminated, fine-grained sulfides. Graphite-bearing schist embedded in gneiss and schist is less well defined on the surface, possibly due to its thin layer, whereas it is well preserved in the drill cores and exhibits clear stratigraphy expressed by regular and continuous bedding units within gneiss and quartz-schist (Fig. 2).

The marbles are classified as pure marble, olivine marble, and serpentine marble (Fig. 3a-d). The dominant minerals of pure marble are calcite with a small amount of dolomite, ankerite, allanite, actinolite, and andradite. The olivine marble is characterized by a porphyritic-like texture of olivine crystals hosted in carbonate matrix (Fig. 3b-d). Olivine is sub- to euhedral with variations in grain size, which is the product of high-degree, upper amphibolite-facies metamorphism of impure, chert- or quartz-bearing dolomitic limestones (Lehmann et al., 2007). In addition to the unaltered olivine marble, partial or bulk replacement of olivine by serpentine is common in olivine marble (Fig. 3b, d), which is referred to as serpentine marble in this study. Carbonate minerals in marbles comprise a large amount of calcite with minor dolomite and ankerite.

The amphibolite is composed of amphibole and plagioclase, with a small amount of biotite and pyrite (Fig. 3e, f). This rock exhibits an equigranular texture of euhedral amphibole surrounded by sub- to anhedral plagioclase (Fig. 3f). A few amphibolite samples contain plagioclase up to 40 vol %. Large euhedral pyrite grains and rare fine-grained chalcopyrite grains are present with amphibole and plagioclase (Fig. 3e, g).

The gneiss suite is dominated by granitic gneiss with a granulitic texture (Fig. 3h). It is composed of quartz, biotite, hornblende, K-feldspar, and minor pyrite (Fig. 3i). Pyrite commonly occurs in the interstitial spaces as small anhedral grains (Fig. 3j).

The schists in the Jinchuan area are classified as quartz schist and calcareous schist, composed of variable amounts of quartz, plagioclase, muscovite, and biotite (Fig. 3k, l). Both types of schists are locally graphite and sulfide bearing. The sulfide-bearing schists are present as sulfide-rich layers with thicknesses varying from 0.2 to 1 m. Flake and fibrous graphite occurs in the interstitial spaces between felsic minerals. Pyrite occurs as anhedral crystals with variable grain sizes and commonly as patchy aggregates (Fig. 3m).

Samples used in this study were collected from seven drill cores in segment III (ZK603, ZK804, and ZK404) and segment II (ZK48-2, ZK22-1, ZK1-1, and ZK8-1) of the Jinchuan intrusion. The samples encompass representative ultramafic rocks, hybrid rocks, and all major country rocks of the Jinchuan deposit. In particular, samples from drill core ZK603 record significant interaction between magma and marbles, with a thick zone (~12 m) of hybrid rocks.

Mineral mappings were obtained on carbon-coated thin sections using a MIRA3 scanning electron microscope (SEM) equipped with two energy-dispersive X-ray spectrometers (EDS), an EDAX Element and a TESCAN Integrated Mineral Analyzer (TIMA), at the State Key Laboratory of Continental Dynamics, Department of Geology, Northwest University, Xi’an, China. An acceleration voltage of 25 kV and a beam current of 9 nA were used. The current and backscattered electron (BSE) signal intensity were calibrated on a platinum Faraday cup using an imbedded automatic procedure. The EDS intensities were then normalized using an Mn standard. The samples were also scanned using a TIMA liberation analysis module.

The chemical compositions of olivine grains were determined by wavelength dispersive X-ray analysis using a JEOL JXA8100 electron microprobe in the Institute of Geology and Geophysics, Chinese Academy of Sciences (IGGCAS), Beijing. The analytical conditions were as follows: 20-nA beam current, 15-kV accelerating voltage, 5-μm beam diameter, and peak counting time of 30 s. The detection limit for Ni under these conditions was ~400 ppm.

To obtain carbonate C isotope compositions, the crushed samples of marble, hybrid rocks, and intrusive rocks were pretreated for 60 to 120 min with pure H₃PO₄ under vacuum using a GasBench-II workstation to extract CO₂ for further analysis. To obtain noncarbonate C isotope composition, the crushed samples of intrusive rocks, graphite-bearing schist, amphibolite, and gneiss were pretreated with hydrochloric acid to remove carbonate minerals from the samples. The residues were then transferred to a high-temperature (980°C) reactor consisting of a heated quartz tube filled with Cr₂O₃, reduced copper (Cu) wire, and silvered CoO-Co₂O₃, where the samples were processed in the presence of O₂ by flash combustion to produce CO₂ for further analysis. The CO₂ extracted from the splits of the samples by different methods was used to determine C concentrations and C isotope compositions of carbonate or noncarbonate mineral phases, respectively. The isotope measurements were performed using a Finnigan Delta-V mass spectrometer at the Institute of Geology and Geophysics, Chinese Academy of Sciences (IGGCAS). The analytical procedures are given in Xie et al. (2016) and Xue et al. (2022). All carbon isotope data are reported in delta notation relative to Vienna-Pee Dee Belemnite (V-PDB). The laboratory standards IAEA-603 (2.46‰), IAEA-CO-8 (–5.764‰), and GBW04416 (1.61‰) are used as carbonate standards for value calibration, while USGS64 (–40.81‰), GBW04407 (–22.43‰), and USGS66 (–0.67‰) are used as noncarbonate standards for value calibration. The reported analytical precision is <5% for carbon content and better than ±0.1‰ for the carbon isotope ratio.

Whole-rock Sr-Nd isotopes were analyzed using a multicollector-inductively coupled plasma-mass spectrometry (MC-ICP-MS) analytical instrument (Neptune Plus) in the laboratory of Wuhan Sample Solution Analytical Technology Ltd., Wuhan, China, following the procedures described in Li et al. (2012). The solutions were prepared by complete digestion of powdered rock samples in HF + HNO₃ in Teflon bombs at 190°C for >24 h. The total procedural blanks contain 0.04 ppb Rb, 0.3 ppb Sr, 0.02 ppb Sm, and 0.05 ppb Nd. Mass fractionation corrections for Sr and Nd isotope ratios were based on the values of ⁸⁶Sr/⁸⁸Sr = 0.1194 and ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219. The measured values for the NIST-987 Sr standard and the GSB 04-3258-2015 Nd standard are ⁸⁷Sr/⁸⁶Sr = 0.710244 ± 8 (2σ, n = 7) and ¹⁴³Nd/¹⁴⁴Nd = 0.512441 ± 7 (2σ, n = 7), which are identical within error to their published values of 0.710241 ± 12 (Zhang and Hu, 2020) and 0.512439 ± 10 (Li et al., 2017), respectively. In addition, the United States Geological Survey (USGS) standards BCR-2 (basalt) and RGM-2 (rhyolite) were analyzed together with our samples. The measured ⁸⁷Sr/⁸⁶Sr and ¹⁴³Nd/¹⁴⁴Nd ratios for BCR-2 and RGM-2 standards are all similar to the recommended values (Li et al., 2012).

The concentrations of S and Ni-Cu-Co in the sulfide ores and different types of country rocks were measured using a LECO C-S Analyzer and an inductively coupled plasma-atomic emission spectroscopy (ICP-AES) analytical instrument, respectively, in the ALS Chemex laboratory in Guangzhou. The detection limits are ~0.01 to 60 wt % for S and ~0.005 to 30 wt % for Ni, Cu, and Co. The concentrations of Se, As, Sb, and Bi were determined using a Perkin Elmer Elan 9000 inductively coupled plasma-mass spectrometer (ICP-MS) at the National Research Center for Analysis (NRCA), Beijing. The final solution was prepared by adding 1 mL internal standard solution containing 1 ppm indium to monitor the element measurement. The detection limits are 0.1 ppm for Se, Sb, and Bi and 1 ppm for As. The analytical precisions are ~8% for S, ~3% for Ni, Cu, and Co, and ~10% for Se, As, Sb, and Bi.

In situ S isotopes were analyzed on pyrrhotite, pentlandite, chalcopyrite, and pyrite in different types of sulfide ores from the Jinchuan deposit and pyrite in the country rocks, using an MC-ICP-MS analytical instrument (Nu Plasma II) equipped with a J-100 femtosecond laser ablation sampling system (Applied Spectra, USA). Samples were measured at the NRCA, Beijing, following the procedures described in Zhu et al. (2016, 2017). Sulfide minerals were ablated using a spot size of 30 mm in diameter and a fluence of 1.02 J/cm² at 6 Hz. The measurements of the pyrite standard (HN) produced a mean δ³⁴S_{Vienna-Canyon Diablo Troilite (V-CDT)} value of 16.8 ± 0.37‰ (2σ, n = 21), which agrees well with the previous analytical result determined by gas mass spectrometry (δ³⁴S_V-CDT = 16.9 ± 0.27‰, 2σ). All data reduction for the MC-ICP-MS analysis of S isotope ratios was conducted using “Iso-Compass” software (Zhang et al., 2020).

Figure 4 shows the textures and modal compositions of two ultramafic intrusive samples from the Jinchuan deposit. Detailed TIMA petrographic observations show that the intrusive rocks are sulfide-bearing, ultramafic cumulates with varying degrees of alteration. In Figure 4a, olivine is the dominant cumulus phase in excess of 50 vol % of the rock, most of which is replaced by serpentine. Sulfides in this sample are commonly intercumulus, consisting of pyrrhotite, pentlandite, and chalcopyrite (Fig. 4a). High-temperature calcite has been observed in association with sulfide minerals in the sample. Fine-grained irregular monticellite is dispersed within the calcite and olivine. Diopside occurs in the interstitial spaces between cumulus olivine crystals (Fig. 4a). Minor secondary magnetite veins also occur in the sample and are commonly interleaved in the sulfides (Fig. 4a). In Figure 4b, diopside in this sample is present in irregular aggregate surrounded by actinolite. Sulfides are soft-walled veins and dominated by pyrrhotite. Secondary pyrite is present as grains and enclosed in pyrrhotite (Fig. 4b), reflecting a replacement of pyrrhotite by pyrite during postmagmatic hydrothermal alteration. Secondary pyrite also occurs as veins that crosscut silicate and sulfide assemblages in the Jinchuan deposit (Ripley et al., 2005).

Figure 5 shows hand specimens, textures, and modal compositions of two representative sulfide-bearing hybrid samples. The hybrid rocks are mainly composed of calcite, diopside, serpentine, biotite, chlorite, wollastonite, and varying amounts of sulfides. The sample shown in Figure 5a displays a semimassive sulfide assemblage dominated by pyrrhotite, pentlandite, and pyrite, with minor chalcopyrite and violarite. More than half of the pyrrhotite is replaced by secondary pyrite (Fig. 5b-c). Almost all the calcites occur intermingled with sulfide. In Figure 5d, the hybrid sample is characterized by finely disseminated interstitial sulfides within calcite matrix. The sulfides in this sample are dominated by pyrrhotite, which has been mainly replaced by secondary pyrite (Fig. 5e-f).

Figure 6 shows hand specimens, textures, and modal compositions of two representative serpentine marble samples with different degrees of skarn alteration from the contact zone. Figure 6a shows an example of serpentine marble, which is characterized by slight skarn alteration and occurrences of abundant brucite and serpentine (Fig. 6b-c). The sample shown in Figure 6d displays more severe skarn alteration. The primary minerals have been changed into serpentine and magnetite, plus negligible brucite (Fig. 6e). In this brucite-barren serpentine marble sample, nearly all the calcite has been replaced by skarn minerals such as diopside, actinolite, biotite, garnet, and orthoferrosilite. The two marble samples contain disseminated sulfides mainly composed of pyrrhotite, chalcopyrite, and pentlandite. Sulfide minerals occur as patches and interstitial networks within the calcite matrix or skarn minerals (Fig. 6b, e).

The C contents and C isotope ratios of the ultramafic intrusive rocks, hybrid rocks, and country rocks are listed in Appendix Table A1. The carbonate C contents and isotopes were determined on calcite-bearing and calcite-barren intrusive rocks, hybrid rocks, and marbles. The marbles contain 1.22 to 8.51 wt % carbon, whereas the intrusive and hybrid rocks have much lower carbon of 0.02 to 0.12 and 0.45 to 1.12 wt %, respectively. Our data and previous data (Ding et al., 2021) show that the calcite-bearing intrusive rocks have higher carbonate δ¹³C_carb values (–4.7 to –2.9‰) than those of calcite-barren intrusive rocks (–9.3 to –8.0‰), but are similar to the hybrid rocks and marbles (–3.9 to 0.8‰; Fig. 7).

The C contents and isotopes of noncarbonate phases were determined on the intrusive rocks and immediate country rocks (amphibolites, gneisses, graphite-bearing calc-shist and quartz-schist) with pretreatment to remove carbonate minerals. The pretreated intrusive rocks contain 280 to 450 ppm C with δ¹³C between −28.9 and −26.7‰, significantly lower than mantle values (δ¹³C from −7 to −5‰; Fig. 7). The pretreated graphite-bearing calc-shist and quartz-schist show higher δ¹³C values, varying from −13.0 to −7.6‰, and contain much higher C content (0.04–7.98 wt %) than the pretreated intrusive rocks and other immediate country rocks. The pretreated amphibolites in the vicinity of the Jinchuan deposit contain 400 to 1,700 ppm C with δ¹³C of −19.6 to −6.6‰, whereas the pretreated gneisses contain 800 to 3,000 ppm C with δ¹³C of −9.3 and −5.5‰. Overall, the δ¹³C values of the pretreated intrusive rocks are much lower than the pretreated immediate country rocks (Fig. 7).

The Rb-Sr and Sm-Nd isotope data for intrusive rocks and immediate country rocks from the Jinchuan deposit are presented in Appendix Table A2 and illustrated in Figures 8 and 9. The zircon U-Pb age of 831.8 Ma from a previous study (Zhang et al., 2010) is used to calculate the (⁸⁷Sr/⁸⁶Sr)_i ratios and $εNdt$ values for the intrusive rocks. The calculations based on our data and those from literature (Duan et al., 2016; Tang et al., 2018) yield (⁸⁷Sr/⁸⁶Sr)_i ratios varying from 0.7066 to 0.7232 and negative $εNdt$ values from –5.7 to –10.5 (Fig. 8, 9), showing isotopically enriched signatures (Fig. 9a). At 831.8 Ma, the $εNdt$ values of different types of country rocks range from –10.0 to –17.8, which are generally lower than those of the Jinchuan ultramafic intrusive rocks (Fig. 8). The (⁸⁷Sr/⁸⁶Sr)_i values of serpentine marble (0.7078–0.7161), plagioclase amphibolite (0.7079–0.7177), quartz-schist (0.7064–0.7096), and calc-schist (0.7228–0.7363) are comparable to the Sr isotope compositions of the Jinchuan intrusive rocks, whereas values of sulfide-bearing marbles (0.7124–0.7209), marble (0.7309–0.7401), and granitic gneiss (0.7420–0.7504) are generally higher than those of the intrusive rocks (Fig. 8).

The concentrations of S, Se, Ni, and Cu in the sulfide ores and the immediate country rocks of the Jinchuan deposit are given in Appendix Table A3. The calc-schist and quartz-schist have the highest sulfur contents (0.11–1.75 wt %) among the different types of country rock. The amphibolites and gneisses contain sulfur in the range of 0.08 to 0.47 wt % and 0.09 to 0.26 wt %, respectively. The bulk S/Se ratios of the sulfide ores vary from 2,150 to 8,080 (Fig. 10), half of which are greater than the average value of the mantle Se/S ratio of ~3,150 (Lorand et al., 2003), 3,300 (Hattori et al., 2002), or 3,333 (McDonough and Sun, 1995). In comparison, immediate country rocks in the Jinchuan area have a wider range of S/Se ratios, from 2,200 to 58,800 (Fig. 10).

Sulfur isotope compositions of sulfide minerals (pyrrhotite, pentlandite, chalcopyrite, and secondary pyrite) in the Jinchuan deposit and pyrite in different types of country rocks are listed in Appendix Table A4. A total of 224 point analyses were measured on 224 sulfide grains in 32 polished thin sections from 32 drill core samples taken from different depths in segments II and III. Sulfur isotope values of sulfide minerals from intrusive rocks and hybrid rocks fall in the range from −7.6 to 8.7‰ (−1.7‰ on average; n = 175; Fig. 11a), which are generally lower than the values considered normal for uncontaminated, mantle-derived mafic magmas (0 ± 2‰), and have over 50% of the samples out of the mantle range (Fig. 11a). The new data significantly expand the δ³⁴S range of previous results for this deposit (−2.6 to 8.0‰, over 80% of 106 samples within the range from −2 to 2‰; Ripley et al., 2005; Duan et al., 2016). The data show no systematic variation of sulfur isotopes between intrusive rocks (mostly between −7.6 and 2.9‰) and hybrid rocks (−5.2 to 2.7‰), as well as between primary pyrrhotite-pentlandite-chalcopyrite assemblages (−7.2 to 2.7‰) and secondary pyrite grains (−7.6 to 2.9‰; Fig. 11b), except the pyrite-carbonate veins (–27 to –7.2‰: Ripley et al., 2005).

The δ³⁴S values of pyrite in sulfide-bearing immediate country rocks (n = 57) show a wide range of δ³⁴S values from −4.0 to 11.3‰ (Fig. 11c). The results are characterized by a positive average value of 2.9‰, which is higher than both the mantle value and the average value of the Jinchuan deposit.

Previous geochemical studies have indicated that the hybrid rocks are characterized by anomalous O-Mg isotopes and high CaO/SiO₂ ratios, reflecting obvious magma-carbonate interaction at the contact zone (Lehmann et al., 2007; Tang et al., 2018; Ding et al., 2021). Within the Jinchuan intrusion, calcite xenocrysts and high-Ca minerals such as diopside and monticellite are recognized in association with sulfide minerals in some intrusive rocks but sporadically occur throughout the intrusion (Fig. 4). Our new carbon isotope results show that the calcite-bearing intrusive rocks have δ¹³C_carb values (−4.7 to −2.9‰) higher than calcite-barren intrusive rocks (−9.0 to −8.3‰) but generally comparable to or slightly lower than hybrid rocks and marbles (−3.9 to 0.8‰), indicating assimilation of marbles into the Jinchuan intrusion (Fig. 7). One of the major processes of magma-carbonate interaction is identified as chemical mixing between mafic magma and contaminant in a previous experimental study (Deegan et al., 2010). Chemical mixing can be reflected in the Sr and C isotope systematics of the melts. The mixing calculation of (⁸⁷Sr/⁸⁶Sr)_i and δ¹³C_carb values between mantle melt and marble indicate that the most contaminated intrusive rocks are a mixture of 3% marble-derived ⁸⁷Sr/⁸⁶Sr and δ¹³C_carb, whereas the hybrid rocks are a mixture of up to 40 wt % marble. This suggests that marble assimilation at Jinchuan took place primarily at the magma-marble interface along the contact zone and is very limited within the intrusion.

High pressure-temperature experiments indicate that the principal process of carbonate assimilation is carbonate dissociation; that is, the breakdown of the CaCO₃ molecule into its component parts, CaO and CO₂ (Deegan et al., 2010; Carter and Dasgupta, 2018). In mafic magma systems, potential oxidative behavior of the CO₂ that is dissolved in the magma can be simplified as CO₂ + 2FeO = CO + Fe₂O₃, where FeO is consumed and carbon monoxide is produced. For this reason, assimilation of marble country rocks has been suggested to increase the oxygen fugacity of the Jinchuan magma and lower the FeO content, and eventually leads to the segregation of immiscible sulfides in magma (Lehmann et al., 2007; Ding et al., 2021). The validity of this hypothesis is evaluated below.

Based on experiments, Simakin et al. (2012) suggested that CO₂ flashing increases $fO2$ in magma, whereas Mollo et al. (2010) proposed the opposite effect. The different results could be due to different compositions of the bulk experimental systems plus different configurations of the experiments. In theory, the effect of carbonate-magma interaction on the redox of the magma depends on the nature of the system, such as open versus closed system, total pressures and/or depths, and initial contents of CO₂ and FeO in the magma, which collectively control the fate of CO₂ produced in the contact zone. The amount of CO₂ that can be added to the magma is also governed by the solubility of CO₂ in the magma, which is very low at low pressures or shallow depths (Holloway and Blank, 1994). As pointed out by Iacono-Marziano et al. (2009) and Deegan et al. (2010), CO₂ generated in the contact zones of magma chambers in the upper crust will escape from the reaction zones immediately upon production, leaving none to trace amounts of such gas to be dissolved in the magma, resulting in no to negligible effect on the redox of the magma.

CO₂ solubility in mafic magma is very strongly pressure and depth dependent (e.g., Holloway and Blank, 1994). The thickness of the overlying strata of the Jinchuan deposit at the Neoproterozoic is about 8 to 10 km (Sixth Geological Team of the Gansu Bureau of Geology and Mineral Resources, 1984), which corresponds to the pressure of approx. 0.3 GPa. This is consistent with the estimated total pressure (approx. 0.3 GPa) using MELTS modeling and clinopyroxene thermobarometer (Chen et al., 2009). A basaltic magma at 0.3 GPa and 1,200° to 1,400°C can only dissolve several hundred ppm CO₂ (Mattey, 1991; Blank and Brooker, 1994). This value represents the upper limit of the external CO₂ that can be added to the Jinchuan magma. The lower limit is zero if the magma was already saturated with CO₂ upon emplacement at Jinchuan. Apparently, such a trace amount of external CO₂ can be immediately swamped by the much greater mass of FeO in the Jinchuan parental magma (estimated to be ~13 wt%; Chen et al., 2009), resulting in little change to the magma redox state. The reason is that the redox state of mafic magma is defined by the ratio of FeO/Fe₂O₃, which is very hard to change by addition of only trace amounts of CO₂ to the magma. Moreover, the equilibrium constant (K = [c(Fe₂O₃)/c²(FeO)]/[c(CO₂)/c(CO)]) for the reaction (CO₂ + 2FeO = CO + Fe₂O₃) is constant at given temperature and pressure (Monazam et al., 2014), which means that such reaction will quickly stop as the reaction reaches equilibrium.

The actual effect of carbonate-magma interaction on the redox of the Jinchuan magma can be assessed by the variation of the oxygen fugacities between the calcite-bearing intrusive rocks near the intrusion-marble contact and calcite-barren intrusive rocks away from the marble country rock and marble xenoliths. We have done so using the olivine-sulfide oxybarometer of Barnes et al. (2013), which is based on the relationships between $fO2$⁠, Ni tenor, and the olivine-sulfide liquid Fe-Ni exchange coefficient (K_D = (X_NiS/X_FeS)^sulfide/(X_NiO/X_FeO)^olivine), and is calibrated using the experimental results of Fleet and MacRae (1988), Fleet and Stone (1990), Gaetani and Grove (1997), Brenan and Caciagli (2000), and Brenan (2003). Below is the equation for calculation:

where a, b, and c are constants of 13.95 (±1.96, 2σ), 0.00037 (±0.00002, 2σ), and −4.1 (±0.98, 2σ), respectively, and C_Ni is the Ni concentration in the sulfide melt. The calculated $fO2$ values are given as supplemental materials in Appendix Table A5. In this study, the two types of intrusive rock samples that have similar olivine Fo contents are selected for comparison. The $fO2$ values of calcite-bearing and calcite-barren intrusive rocks were estimated to be QFM+0.7 (±0.4, 1σ) and QFM+0.6 (±0.4, 1σ), respectively. Their comparable $fO2$ values indicate that the consumption of ferrous iron via CO₂-induced oxidative reaction did not happen in a large volume of magma. The marble dissociation processes did not sufficiently oxidize the Jinchuan magma nor result in elevation of $fO2$ in the magma. Therefore, our findings support that assimilation of marbles played a negligible role in decreasing the SCSS needed to produce in situ sulfide saturation in the Jinchuan magma.

In addition to the insignificant oxidation effect, transport and deposition of immiscible sulfide liquid during lateral magma flow is also against the marble assimilation inducing a large volume of in situ sulfide segregation in the dynamic Jinchuan magmatic system (Barnes et al., 2016; Barnes and Robertson, 2019; Yao and Mungall, 2021, 2022). Barnes and Robertson (2019) argue that lateral transport of sulfide droplets is so fast compared with the rate at which the sulfide reacts with the magma that ore deposition must be happening many kilometers from the site of sulfide liquid generation. This model has been taken further recently by Yao and Mungall (2021), who demonstrate that the growth of an endogenous sulfide from its initial size after nucleation to the mm scale requires a long time and needs a long-distance transport to kinetically upgrade the initially metal-poor sulfide melts. Hence, if sulfide segregation at Jinchuan was triggered by in situ marble country rock assimilation, sulfides are supposed to be small in size and having PGEs depleted relative to Ni in the contact zone, which is contrary to the observed mmscale sulfides in the hybrid rocks and sulfide-bearing marbles (Figs. 5 and 6), and the Pd enrichment relative to Ni in these disseminated sulfide ores (Su et al., 2008; Song et al., 2009; Chen et al., 2013). Overall, we conclude that the interaction between magma with marble country rocks has little genetic connection with the mineralization of the Jinchuan deposit.

Generally, siliceous crustal contamination will lower the SCSS in mafic magma mainly due to lowering the temperature and FeO content in the contaminated magma (Irvine, 1975; Li and Ripley, 2005). Petrochemical and Re-Os isotope studies indicate extensive crustal contamination related to the Jinchuan deposit (Song et al., 2007; Yang et al., 2008). The Sr-Nd-Mg-Hf isotope data do not provide a clear answer to the source of the siliceous crustal materials contaminating the Jinchuan magma (Li and Ripley, 2011; Duan et al., 2016; Tang et al., 2018), with some researchers suggesting the upper crust (Duan et al., 2016) and other researchers proposing the lower crust (Tang et al., 2018). The ambiguity may be due to insufficient isotope data for the potential contaminants, especially for the country rocks of the deposit.

A systemic Sr-Nd isotope comparison of intrusive rocks and different types of country rocks is illustrated in Figure 8. The marble samples with extremely low concentrations of Nd from the literature are not included in this study because the Nd isotope ratios yielded from such samples are not as reliable as those with higher Nd concentrations during the measurement. The intrusive rocks have highly negative $εNdt$ values and high initial ⁸⁷Sr/⁸⁶Sr ratios, implying the involvement of a long-term enriched subcontinental lithosphere mantle source and/or assimilation of ancient continental crust (Xu et al., 2008; Li and Ripley, 2011; Arndt, 2013). Different types of country rocks have variable, distinct Sr-Nd isotope compositions but collectively show more enriched Nd isotope signatures relative to the Jinchuan intrusion (Fig. 8). Using each type of the country rocks as contaminants, binary mixing calculations indicate that the observed Sr-Nd isotopes of the Jinchuan intrusion cannot be explained by a single type of immediate country rock contamination (Fig. 8a). In the scenario of using a mixture of country rocks as contaminants, mixing calculations show that variable degrees of in situ contamination can account for the observed variations of the Sr-Nd isotope composition of the Jinchuan intrusion if the amounts of the contaminants are involved up to 50 to 80 wt % (models ^#1–4 in Fig. 8b-d). Such unusually high degrees of contamination are clearly at odds with the heat budget and the fact that the Jinchuan ultramafic intrusion is dominated by olivine cumulates. Only the involvement of calc-schist is estimated to be a reasonable degree of contamination (up to 30 wt %, model ^#5 in Fig. 8d). However, this contaminant plays an insignificant role in modifying the Sr-Nd isotopes of the intrusive rocks because (1) the volume of calc-schist in the Jinchuan area is small (Fig. 2), and (2) it is not responsible for producing the intrusive rocks with relatively low initial ⁸⁷Sr/⁸⁶Sr ratios (Fig. 8d). Therefore, deeper crustal contaminants with extremely low $εNdt$ are largely required.

At ~830 Ma, the lower crust of the North China craton is characterized by highly negative $εNdt$ values (−27 to −16.9) and varying initial ⁸⁷Sr/⁸⁶Sr ratios (0.704–0.730) (Jahn et al., 1999; Huang et al., 2004; Liu et al., 2004). About 15 to 20 wt % of lower crustal contamination is required to explain the observed (⁸⁷Sr/⁸⁶Sr)_i and $εNdt$ values for the Jinchuan intrusive rocks (Fig. 9). Mixing calculations using Sr-Nd isotopes (Fig. 9a) and Sr-C isotopes (Fig. 9b) also indicate that the variable (⁸⁷Sr/⁸⁶Sr)_i with rather constant and high $εNdt$ values of some intrusive rocks can be explained by the presence of variable proportions of lower crust and minor amounts (≤3 wt %) of marble country rocks, consistent with our petrographic observation of calcite xenocrysts in the intrusive rocks (Fig. 4). In addition to the Sr-Nd isotopes, en route crustal contamination is also indicated by the distinct δ¹³C values (−28.9 to −26.7‰) of the ultramafic intrusive rocks compared to those of the immediate country rocks (–19.6 to –5.5‰; Fig. 7). Based on the different lines of evidence described above, it is reasonable to conclude that the immediate country rocks have supplied some materials to the parental magma, but in situ crustal contamination only played a limited role, if any, in triggering sulfide saturation in the Jinchuan magma. Current results support the interpretation that the predominant assimilation process of the Jinchuan magma related to the sulfide segregation took place at the lower crust during magma ascent.

The addition of externally derived sulfur is considered to be the most viable mechanism for producing sulfide well above the cotectic proportion in mafic/ultramafic magmas and for generating sulfide-rich, magmatic Ni-Cu-(PGE) deposits (Keays, 1995; Naldrett, 1997; Ripley et al., 2002; Ripley and Li, 2013; Wang et al., 2018). An isenthalpic modeling proposed by Xue et al. (2019) suggests that sulfide saturation triggered by fractional crystallization with simultaneous siliceous crustal assimilation (AFC) is commonly achieved later than the actual timing of sulfide saturation in the ore-forming system. For this reason, addition of sulfur from crustal rocks is most likely required for the ore formation. As one of the largest magmatic sulfide deposits in the world, addition of significant external sulfur and the sources of external sulfur are still not clearly recognized in the Jinchuan deposit for the following reasons:

1. The immediate country rocks of the Jinchuan deposit are considered to be poor in sulfur (Ripley et al., 2005; Lehmann et al., 2007; Song et al., 2012), but in the last decade, the drilling program at Jinchuan has revealed that the immediate country rocks such as amphibolites, gneisses, and schists are variably sulfide-bearing and locally contain up to ~1.8 wt % sulfur (Fig. 3).

2. Previous sulfur isotope data show that most of the sulfide ores at Jinchuan are characterized by mantle-like δ³⁴S values (0 ± 2‰; Ripley et al., 2005), though a few samples with anomalous Δ³³S values suggest an involvement of external sulfur from the Archean sedimentary rocks (Duan et al., 2016).

Sulfur/selenium ratios have long been used to trace the sources of sulfur in magmatic sulfide ore-forming systems due to the affinity of selenium with sulfur (D_{sulfide/magma Se} = 226–2339; Barnes and Ripley, 2016, and references therein) and the distinct S/Se ratios between mantle rocks and crustal rocks (e.g., Ripley, 1990; Thériault and Barnes, 1998; Holwell et al., 2007; Sharman et al., 2013; Queffurus and Barnes, 2014, 2015; Brzozowski et al., 2020; Deng et al., 2022). Mantle-derived magma has low S/Se ratios of 2,850 to 4,350, with an average value of 3,150 (Eckstrand et al., 1989; Lorand et al., 2003; Lissner et al., 2014; Smith et al., 2016), whereas continental crust has highly variable S/Se ratios with an average value of 8,060 (Rudnick and Gao, 2004). Hence, S/Se values greater than mantle values have been interpreted to reflect the contamination of S-rich country rocks (e.g., Lesher et al., 2001; Deng et al., 2022). Due to the chalcophile nature of Se, S/Se values of sulfide ores are also strongly affected by variations in R factor (silicate to sulfide liquid ratio) (Queffurus and Barnes, 2015; Brzozowski et al., 2020, 2021), which can be modeled iteratively using the closed-system model of Ripley and Li (2003) (Fig. 10b):

Where $Csul.Ri$ is the concentration of element i in the sulfide liquid after it interacted with silicate melt, C_sil and C_sul are the initial concentrations of element i in the silicate melt and the sulfide liquid, respectively, D is the sulfide liquid/silicate melt partition coefficient for element i, and R is the R factor. Similarly, S isotopes are highly associated with the R factor because the higher R factor causes the sulfide ores to have S isotope signatures more similar to the mantle-derived magma (Taranovic et al., 2022). Variations in δ³⁴S of sulfide ores as a function of variable R factor can be modeled following the open-system equation of Ripley and Li (2003) (Fig. 10b):

where δ³⁴S_sul.R is the S isotope composition of the sulfide after isotope exchange with the silicate melt, δ³⁴S_sil and δ³⁴S_sul are the initial S isotope compositions of the silicate and sulfide melts, respectively, Δ is the fractionation factor between sulfide liquid and silicate melt, and R⁰ = C_sil*R/C_sul.

Our new sulfur isotope data for sulfides from the deeper parts of segments II and III indicate a larger range of δ³⁴S values (−7.6 to 3.0‰) than previously reported (Fig. 11). About half of the δ³⁴S values are distinct from those considered normal for uncontaminated, mantle-derived sulfur (0 ± 2‰), clearly suggesting that crustal sulfides have been incorporated into the mineralization. These sulfide ores are also characterized by a wide range of S/Se ratios from 2,150 to 8,080, which fall in the range of typical magmatic sulfide deposits in the world (Fig. 10a), while the immediate siliceous country rocks at Jinchuan that contain variable S contents have high and variable S/Se ratios of 3,929 to 58,824 (Fig. 10a). Using crustal sulfide with different δ³⁴S values as contaminants, the modeled variations in the δ³⁴S and S/Se values of sulfides as a function of variable R factor show that sulfide ores in the Jinchuan deposit experienced varying R factors between 800 and 100,000, which are comparable to the estimated R factors obtained by PGE geochemistry (Song et al., 2009), but only a few data points fall in the δ³⁴S range of the immediate country rocks between −4.0 to 11.3‰ (Fig. 10b), suggesting limited sulfur contribution of immediate country rock contamination to the mineralization in the Jinchuan deposit. This is consistent with an obvious distinction of δ³⁴S values between the sulfide ores characterized by a negative average δ³⁴S value (−1.7‰) and the immediate country rocks with a positive average value (2.9‰) (Fig. 11a, c). Hence, the variations of δ³⁴S and S/Se values observed in the Jinchuan sulfide ores are better explained by assimilation of crustal sulfides at depth, which have high S/Se ratios and negatively variable δ³⁴S values, followed by progressive dilution of the contaminated δ³⁴S-S/Se signature (Fig. 10b), generally in agreement with the enroute crustal contamination reflected by the Sr-Nd-C isotopes of the Jinchuan intrusion and immediate country rocks. Therefore, we conclude that addition of crustal sulfur plus significant siliceous crustal materials assimilation at depth is responsible for the important sulfide mineralization of the Jinchuan Ni-Cu-PGE deposit.

The detailed mineralogical study of the contact zone and hybrid rocks from Jinchuan reveals that in addition to serpentinization and decarbonation in the contact zone during magma emplacement, chemical dissolution of carbonate minerals from the carbonate xenoliths also took place, resulting in the occurrence of calcite xenocrysts from the highly contaminated magma in limited places. This is further supported by similar δ¹³C_carb between the intrusive rocks containing calcite and the hybrid rocks. With few exceptions, the δ¹³C_carb of these two types of rock varies between the mantle and carbonate country rock values, which corresponds to significant and limited carbonate contamination by the Jinchuan magma at the contact zone and within the intrusion, respectively. Less than several hundred ppm of the CO₂ generated by carbonate breakdown could be added to the Jinchuan magma, which was rapidly swamped by the much greater mass of FeO in magma and played a negligible role in increasing the magma $fO2$⁠. Using the olivine-sulfide oxygen barometer, our calculations show that the calcite-barren and calcite-bearing intrusive rocks have similar $fO2$⁠, confirming that the effect of carbonate-magma interaction on the redox of the Jinchuan magma is negligible. These new findings support that sulfide saturation in the Jinchuan magma was not triggered by such process. Meanwhile, the δ³⁴S, Se/S, $εNdt$⁠, initial ⁸⁷Sr/⁸⁶Sr, and δ¹³C values all support the interpretation that the Jinchuan magma experienced significant siliceous crustal contamination coupled by addition of crustal sulfur prior to its final emplacement at Jinchuan. Therefore, we conclude that sulfide saturation in the Jinchuan magma took place at depths below the carbonate strata, due to contamination with sulfur-rich, siliceous crustal materials.

This work was funded by the National Key Research and Development Project of China (Grant 2020YFA0714802), the National Natural Science Foundation of China (Grant 42172076 and 41802076), and the Fundamental Research Funds for the Central Universities (Grant 265QZ2022008 and 2-9-2019-043). We thank Yongcai Wang and Yanning Wang for assistance in fieldwork and sampling and Chao Li, Lihui Jia, Hongwei Li, and Hongfang Chen for guidance in chemical analysis. We would also like to thank Larry Meinert, Stephen Barnes, and Matthew Brzozowski for their detailed and constructive reviews that greatly improved the manuscript.

Shengchao Xue is an associate professor at the State Key Laboratory of Geological Processes and Mineral Resources, China University of Geosciences, Beijing, China. He completed his Ph.D. degree at the Institute of Geology and Geophysics, Chinese Academy of Sciences. His research currently focuses on magmatic sulfide deposits of intraplate and convergent tectonic settings in the Central Asian orogenic belt, East Kunlun orogenic belt, and North China craton, causes of sulfide saturation in mafic magmas, controls on the compositions of base metals and PGEs in sulfide ores, and models that can be used to assess magmatic sulfide potential.