Origin and Nature of Parental Magma and Sulfide Segregation of the Baixintan Magmatic Ni–Cu Sulfide Deposit, Southern Central Asian Orogenic Belt (CAOB), NW China: Insights from Mineral Chemistry of Chromite and Silicate Minerals

: The mineral chemistry of chromite and silicate minerals in the Baixintan magmatic Ni-Cu sulfide deposit in the Northern Tianshan, southern Central Asian Orogenic Belt (CAOB) are reported here. Two types of chromite were identified in mafic-ultramafic rocks. Type I chromite occurs as inclusions encased in olivine and has a primary and magmatic origin and homogeneous composition with Cr# values (49–66). It is characterized by high Ti contents (0.33–1.52 wt%) and small quantities of ZnO (0–0.21 wt%), MnO (0.28–0.45 wt%), and NiO (0.06–0.24 wt%) contents. In contrast, type II chromite with interstitial phase and larger compositional variations has significantly higher TiO 2 (up to 6.2 wt%) and FeOt contents (up to 69.3 wt%) and slightly lower Al 2 O 3 (minimum 3.0 wt%) and MgO contents (minimum 0.53 wt%). It is considered to crystallize from a more evolved and fractionated melt and suffers from post-magmatic alteration, such as serpentinization and chloritization. The olivine has forsterite values (Fo) varying from 76.8 to 85.6. The parental magma is characterized by high temperature (1389 °C), high pressure (3.8 Gpa), and high Mg content (11.4 wt%) with oxidized (FMQ + 1.6) and hydrous nature based on compositions of primary chromite and olivine–chromite pairs. The intrusion originated from high - degree partial melting of depleted mantle that had been modified by crustal components and metasomatized by subduction fluid in a post-orogenic extensional setting. Two stages of sulfide segregation have been recognized. Early segregation led to the depletion of platinum group elements (PGE), and disseminated sulfide mineralization was the product of later segregation. The assimilation of crustal Si and S components played more important roles on sulfide segregation rather than fractional crystallization. and pyroxene were altered, and thus, only fresh grains with crystal integrity were selected for analyses. The standards used were olivine for Mg and Si, garnet for Al, diopside for Ca, omphacite for Na, ilmenite for Ti, chromite for Cr, fayalite for Fe, niccolite (NiAs) for Ni, sphalerite for Zn, phlogopite for K, and manganese oxide for Mn.

Abundant early Permian (275-285 Ma) mafic-ultramafic intrusions associated with Ni-Cu sulfide deposits occur in the Northern Tianshan, southern CAOB [23][24][25]. These deposits are mainly located in the eastern segment; however, there are some breakthroughs of Ni prospecting in the western segment in the past several years [26][27][28][29]. One example is the Baixintan deposit, a mediumsized deposit with 6.49 × 10 4 tons of Ni and 8.43 × 10 4 tons of Cu [30]. Preliminary petrogenetic studies have been conducted in the newly discovered deposit, such as geology, geochronology, geochemistry, and Sr-Nd-Hf-O-S isotopic studies [29,[31][32][33]. The sulfide mineralization has also been discussed, such as sulfur saturation, quantitative modeling of sulfide segregation, and regional Ni-Cu sulfide prospectivity [26,29,32]. These studies are beneficial for understanding petrogenesis and mineralization; however, mineralogical studies, especially detailed chromite study, have not yet been conducted though it is widespread as accessory mineral in ultramafic-mafic rocks.
In this paper, systematical mineralogical studies of chromite and silicate minerals are conducted. First, two types of chromite have been identified according to mineral texture and occurrence. Then, parental magma composition and physicochemical conditions, including temperature, pressure, and oxygen fugacity, are estimated based on a new dataset of mineral chemistry. Their implications for tectonic setting and mantle source of mafic-ultramafic rocks are consequently presented. Finally, timescale and trigger factors of sulfide segregation are discussed based on olivine crystallization model and magma evolution.

Regional Geology
The CAOB, set between the Siberian and the Tarim-North China cratons, is one of the largest Phanerozoic accretionary belts on the globe and hosts numerous polymetallic deposits [34]. It is formed by multiple subduction-accretion of island arcs, oceanic plateaus, seamounts, ophiolites, and microcontinents that occurred from the late Proterozoic to the Mesozoic [35]. A series of Permian mafic-ultramafic rocks and related Ni-Cu deposits are located in the southern margin of the CAOB from NW China to NE China (Figure 1a, [36]). The Northern Tianshan and Beishan orogens are situated northeast to the Tarim craton and represent the southernmost segments of the CAOB in eastern Xinjiang, NW China ( Figure 1b). The Northern Tianshan is separated from Tuha Basin to the north and the Middle Tianshan by the Aqikkuduk fault to the south and comprises the Dananhu-Harlike island arc, the Kangguer intra-arc basin, and the Yamansu arc [29,[37][38][39].  [34]); (b) schematic division of massifs and the location of the Northern Tianshan, NW China (modified after Ruan et al. [36]); and (c) a regional geological map showing the distribution of early Permian mafic-ultramafic complexes in Northern Tianshan (modified after Deng et al. [29]).
The Dananhu-Harlike arc is composed of Ordovician-Carboniferous arc volcanic rocks, siliceous slate, and limestone [29]. Precambrian inherited zircons have been found in volcanic rocks from the Kalatage area [37,39]. The Late Carboniferous pillow basaltic rocks may form in a back-arc tectonic setting [40,41], whereas the Permian bimodal volcanism in the Bogda zone is considered to have formed in a post-collisional extensional environment caused by slab break-off [42,43]. Various types of arc-related ore deposits occur in this area, including the Silurian Huangtan volcanogenic massive sulfide (VMS) Cu-Au deposit, the Silurian Hongshan epithermal deposit, the Middle Devonian Xierqu skarn Fe deposit, the Middle Devonian Yudai porphyry Cu deposit, and the Early Carboniferous Tuwu porphyry Cu deposit [39].
In addition to widespread intermediate-mafic volcanic eruptions, more than 30 mafic-ultramafic intrusions have been identified along the east-trending Kangguer-Huangshan deep fault [26,28,44]. Available age data show that they were emplaced in early Permian with a major peak at approximately 280 Ma due to the interaction between the metasomatized lithospheric mantle and the ascending asthenospheric mantle, as a result of slab break-off or an upwelling mantle plume [23,24,43,[45][46][47][48]. A few intrusions are known to host magmatic Ni-Cu sulfide deposits, such as evidenced by the Huangshan, Huangshandong, Tulaergen, and the newly found Yueyawan deposits. Most of them are located to the eastern segment of Northern Tianshan, and the Ni mineralization belt was previously considered not to extend to the western segment. However, the recent Ni prospecting breakthroughs in the Baixintan and Lubei mafic-ultramafic intrusions have extended the metallogenic belt to the west for 180 km (Figure 1c) [26,28,29].

Deposit Geology
The Baixintan complex occurs along the secondary Dacaotan fault and has an exposed area of 5 km 2 . It extends to depths of more than 400 m with a flat funnel shape. The immediate country rocks are middle Ordovician dacite, basaltic andesite, volcanic tuff, and sandstone clastic rocks, and later monzogranite and granodiorite are situated in the south. The complex generally consists of olivine gabbro, olivine websterite, and lherzolite. The olivine gabbro occurs in the north, and mineralized lherzolite appears in the southwestern region. The olivine websterite is found between lherzolite and olivine gabbro. The lherzolite and olivine websterite are enveloped by olivine gabbro in the cross section ( Figure 2b). The lherzolite is dark black and composed of olivine (50-60%); pyroxene (25-30%); amphibole (10-15%); and a little bit of plagioclase, chromite, and phlogopite (Figure 3a,d). The olivine websterite is grey black and contains pyroxene (50-70%) olivine (20%), plagioclase (5-10%), and accessory amphibole and chromite (Figure 3b). Significant alteration has been identified in these ultramafic rocks, including serpentinization, iddingsitization, chloritization, and talcization. The olivine in ultramafic rocks are 0.2-3 mm, idiomorphic, and surrounded by clinopyroxene and serpentine. The gabbro and olivine gabbro exhibit a mosaic texture and predominantly consist of plagioclase (approximately 50%), pyroxene (approximately 40%), and occasional olivine (5-10%) (Figure 3c). The ore bodies occur as lenticles and funnels at depths of 0-200 m (Figure 2b). The sulfide mineralization appears mainly at the bottom of lherzolite with Ni content ranging from 0.27 wt% to 5.45 wt% (average 0.41 wt%) and Cu content ranging from 0.24 wt% to 1.54 wt% (average 0.60 wt%), respectively. They are dominantly disseminated sulfide, and vein-type sulfide is also recently discovered at these depths. Sulfide minerals, including chalcopyrite, pyrrhotite, pentlandite, pyrite, minor bornite, and violarite, occur as interstitial phases between olivine grains (Figure 3e,f).

Mineral Occurrence
Chromite, usually the first phase to crystallize from mafic-ultramafic magmas, occurs as an accessory mineral in the Baixintan complex and is more common in ultramafic rocks than in mafic rocks. It has sizes ranging from 20 to 200 μm. Two types of chromite have been identified based on their morphology and occurrence. The first type occurs as euhedral-subhedral inclusions in olivine and pyroxene and has straight borders and a bright white color (Figure 3g). The second type is more common and refers to anhedral or irregular interstitial grain surrounded by alteration minerals, such as serpentine and chlorite (Figure 3h). They usually have rough boundaries produced by later dissolution and alteration. Some of them have bright rims in BSE images.
The olivine dominantly occurs as cumulate grains in lherzolite or is occasionally enclosed by pyroxene and amphibole. The cumulate grains have larger size (0.3-3 mm) and are more fragmented. The enclosed ones are round and fresh with small size (0.2-0.5 mm). Pentlandite and pyrrhotite are found filled among olivine grains in lherzolite. The orthopyroxene mainly occurs as short columnar and granular in olivine websterite and lherzolite, while clinopyroxene is common in various rock types. Alteration minerals are generally found surrounding pyroxene, such as uralite, chlorite, and talc. Plagioclase is characterized by short columnar, perfect cleavage, and polysynthetic twin. Amphibole is common in various rock types and is characterized by uralitization and talcization. Hydrous minerals, including amphibole and phlogopite are widely found in the intrusion.

Materials and Methods
Thirty samples of lherzolite, websterite, olivine gabbro, and gabbro were collected from drill cores in profile 7 ( Figure 2) and outcrops for petrographic observations and mineral composition analysis. Chromite and silicate minerals, including olivine, pyroxene, plagioclase, and amphibole, in different rock types have been analyzed. The thin sections were carefully observed by transmitted and reflected optical microscopy as well as by environmental scanning electron microscope (ESEM) under backscattered electron (BSE) mode prior to quantitative electron probe microanalysis (EPMA). The latter were performed using a Cameca SX 50 electron microprobe instrument at Wuhan Sample Solution Analytical Technology Co., Ltd., China and the Materials Research and Testing Center, Wuhan University of Technology.
The standards used for EPMA of chromite were spinel for Mg and Al, diopside for Si, ilmenite for Ti, chromite for Cr and Fe, manganese oxide for Mn, niccolite (NiAs) for Ni, sphalerite for Zn, albite for Na, and phlogopite for K. Peak and background counting times were set at 30 and 15 s, respectively. The beam diameter was 3-5 μm, and spot mode was used for tiny particles or minerals with rough surfaces. The experimental conditions were 20 kV of accelerating voltage and 10 nA of beam current. Detection limit was 0.1 wt.%. Iron was determined as total iron (FeOt), and Fe 2+ and Fe 3+ were calculated by assuming an R 2+ R 3+ 2O4 formula and by balancing RO: R2O3 = 1 for chromite [49].
The same experimental conditions were applied to silicate minerals. Most olivine and pyroxene were altered, and thus, only fresh grains with crystal integrity were selected for analyses. The standards used were olivine for Mg and Si, garnet for Al, diopside for Ca, omphacite for Na, ilmenite for Ti, chromite for Cr, fayalite for Fe, niccolite (NiAs) for Ni, sphalerite for Zn, phlogopite for K, and manganese oxide for Mn.
The chromite compositional comparison between the Baixintan and other mafic-ultramafic intrusions is illustrated in Figure 5. Only Chr I was used for comparison since it represents the primary chromite while Chr II crystallized from a more evolved and fractionated melt (see the discussion in Section 6.1). The Baixintan chromite as well as the chromite in the Northern Tianshan and Beishan display similar Cr# values to the middle ocean ridge basalt (MORB), which is slightly lower than that in island arc-related intrusions, boninites, hot spot, and large igneous provinces (LIPs), while their Mg values are slightly higher than MORB (Figure 5a,b). The Baixintan chromite is characterized by higher Fe 3+ than others and some plot in the fields of Alaskan-type intrusions and island-arc tholeiite (Figure 5c,d).

Silicate Minerals
The olivine in lherzolite has the highest MgO content (41.5-46. Pyroxene is characterized by tholeiite to calc-alkaline series. Orthopyroxene is classified as enstatite with Mg# = 85.0. Clinopyroxene is more common than orthopyroxene and is classified as augite and diopside. Clinopyroxene in lherzolite displays high Mg# = 86.5 and Wo = 43.7. Its compositional variation seems to be more related to a rift cumulated trend and is different from that of typical Alaskan-type complexes ( Figure S1 and Table S3).
The anorthite proportions of plagioclase (An numbers) are from 67.4 to 80.8 for lherzolite, from 72.5 to 82.5 for olivine gabbro, and from 69.1 to 86.1 for gabbro. Except for a few grains of labradorite, all plagioclases are classified as bytownite ( Figure S2 and Table S4).
The amphiboles in lherzolite and olivine gabbro are classified as pargasite and endenite, respectively, while that in websterite has larger compositional variations. Amphibole in lherzolite and websterite has higher Mg# (Mg#/(Mg# + Fe 2+ )) than that in olivine gabbro. They almost all have a mantle origin and similar compositions with other amphibole in ultramafic rocks ( Figure S3 and Table S5).

Primary and Magmatic Chromite
Chromite is sensitive to hydrothermal alteration and prograde metamorphism, which may modify its composition [50,53,54]. Type I chromite mainly occurs as inclusions in olivine and, rarely, in silicate interspaces. They were crystallized earlier and captured by olivine crystals, which had avoided reaction with interstitial evolving liquid. It is noteworthy that magmatic type I chromites are outside any specific field of MORB, IAT, and Alaskan-type intrusions. This may be caused by a possible Fe-Mg subsolidus exchange equilibrium, in which process Fe 2+ could incorporate into chromite crystals as an isomorphism of Mg, resulting in lower Mg# values than MORB (Figure 5b). The incorporation of Fe 3+ and loss of Al lead to some samples plotting in the Alaskan-type intrusions and IAT (Figure 5c,d). Despite compositional changes, the extent of modification could be evaluated by Mg# ratios and MnO content. Significant subsolidus exchange between chromite and olivine commonly gives rise to low Mg# ratios (<15) and high MnO content (up to 2 wt%) because Fe 2+ and Mn could replace Mg during subsolidus exchange [55][56][57]. Type I chromite has Mg# ranging from 20 to 44 and low MnO content (0.28-0.45 wt%) and plots the "filter polygon" for magmatic chromite ( Figure 6) [55,58]. The subsolidus exchange may have an effect on type I chromite composition but at a much slower rate. Type I chromite can thus represent magmatic chromite, while type II chromite cannot be treated as primary chromite due to its interstitial phase and large compositional variations. It is considered to crystallize from a more evolved and fractionated melt [20,58]. Some samples have low MgO, Al2O3, and Cr2O3 contents and high FeOt and TiO2 contents, indicating far greater modification than type I chromite ( Figure 4). Particularly, Fe-rich chromite rims in zoned chromite formed during serpentinization and chloritization [14,20,54,59]. The loss of Al, Mg, and Cr may be caused by chloritization, which involved the reaction of chromite, olivine, and magnetite to produce Fe-rich chromite rim and chlorite [20].
The crystallization temperature and pressure of olivine, pyroxene, and amphibole are also respectively estimated. The initial crystallization temperature of olivine with the highest Mg# value is estimated to be 1309 °C using the empirical formula: T (℃) = 1066 + 12.067Mg# + 312.3(Mg#) 2 [62]. The clinopyroxene has a temperature of 1206 °C and a pressure of 0.62 Gpa using the thermobarometer proposed by Thompson [63]. Amphibole in the Baixintan intrusion characterizes low pressure (≤0.8 Gpa) and relatively high temperature (≥950 °C, Figure 7), with the reference of experimentally synthesized amphibole in various temperature-pressure conditions [64].

Oxygen Fugacity
The magma oxygen fugacity corresponding to the olivine-chromite pairs is calculated as FMQ + 1.1 to FMQ + 2.1, with an average of FMQ + 1.6, using the method proposed by Ballhaus et al. [15] ( Table S1). The oxygen fugacity exhibits a generally elevated trend with fractionation of magmas. Specifically, the olivine has relatively higher oxygen fugacity (FMQ + 2.1) than lherzolite (FMQ + 1.5) and olivine websterite (FMQ + 1.7). The result is similar to the other mafic-ultramafic intrusions in Northern Tianshan, such as the Tulaergen and Huangshandong intrusions, and higher than those in the Beishan area, for which the parental magma commonly have higher MgO content (Figure 8).  [65], the mantle fO2 of intrusions in the southern CAOB [22], Huangshannan and Tulaergen [22], Poyi [21], and Hongnieshan (unpublished data). MORB: middle ocean ridge basalt.

Tectonic Setting
The tectonic setting of early Permian mantle-origin rocks in the Northern Tianshan has been a disputed topic. Some research considered that they are the Alaskan-type complexes emplaced in an island-arc environment because of widespread hydrous minerals with no alteration origin and island-arc signature of trace elements [69,70]. However, the presence of orthopyroxene in these complexes in the Northern Tianshan, which is extremely scarce in the Ural-Alaskan-type intrusions [71][72][73], argues against the Alaskan-type origin. Furthermore, a post-orogenic extensional setting has gained growing recognition in the past several years due to increasing studies of geochronology and metallogeny. The Paleo-Asian ocean subducted southward until later Carboniferous (310 Ma) [74][75][76], which predates the early Permian mafic-ultramafic complexes (275-285 Ma) in the Northern Tianshan. The arc-related deposits in this area, such as porphyry-type and VMS deposits in this area, mainly formed from Silurian to Carboniferous rather than in Permian [37,39]. Our mineralogical studies in the Baixintan intrusions also give new constraints on the origin and support the postorogenic extensional setting.
The Cr# of Baixintan chromite (49)(50)(51)(52)(53)(54)(55)(56)(57)(58)(59)(60)(61)(62)(63)(64)(65)(66) as well as chromite in other early Permian intrusions in the northern Tianshan and Beishan are significantly lower than that in island arc-related rocks (Cr# 75-90, Figure 5a). Chromite from island-arc tholeiite (IAT) is characterized by low Al content and significant positive relationship between Al and Ti content [8], whereas chromite in early Permian intrusions in Northern Tianshan and Beishan displays a negative relationship and relatively high Al content. Their compositions are mainly plotted in the mid-ocean ridge basalt (MORB) and ocean island basalt (OIB) area rather than in the IAT area (Figure 5d). Clinopyroxene in this area is mainly augite and exhibits a rift cumulate trend, while clinopyroxene in the Alaskan-type complexes is mainly diopside and exhibit an arc cumulate trend ( Figure S1). In addition, a non-arc setting is also indicated by plagioclase composition. Plagioclase from early Permian mafic-ultramafic rocks in Northern Tianshan and Beishan has higher CaO and Al2O3 contents and lower SiO2 and alkaline content than that from typical Alaskan-type intrusions ( Figure S4).

Mantle Source
A depleted mantle source modified by a subducted slab has been indicated by Hf-Sr-Nd-O isotope and trace element geochemistry. Our oxygen fugacity calculation provides new evidence to support the conclusion. The parental magma in the Northern Tianshan is relatively oxidized because its fO2 value is clearly higher than those of MORBs (FMQ − 1 to FMQ + 0.5 [77]) and the mantle fO2 of intrusions in the southern CAOB (FMQ − 1 to FMQ + 0.5 [22]), implying incorporation of minor slabderived oxidized components into to mantle source. The extent of oxidized component involvement in the Northern Tianshan is slightly higher than that in the Beishan (Figure 8). A few chromite samples exhibit subduction signatures, such as plotting in the field of Alaskan-type intrusions and IAT (Figure 5c,d). Chen et al. [28] used the silicate-liquid model pMELTS to simulate magmatic evolution of the adjacent Lubei intrusion, and the result showed that the mantle was hydrous (approximately 0.29 wt % H2O). The hydrous nature may be ascribed to the mantle source having been metasomatized by subduction fluid. The partial melting degree can be estimated by the composition of primary chromite enclosed within olivine according the formula: F = 10 * ln (Cr#) + 24 [5]. The result ranges from 17.0% to 18.9% with average 18.3% and shows a high degree of partial melting.
To sum up, the mafic-ultramafic rocks originated from high-degree partial melting of depleted mantle that had been modified by crustal components and metasomatized by subduction fluid in a post-orogenic extensional setting.

Early Segregation
High Mg tholeiite magma caused by high-degree mantle partial melting is enriched in the chalcophile element, especially Ni, Cu, and PGE [78,79]. The Baixintan intrusion has a high partial melting degree (18.3%), which is sufficient to eliminate sulfide retention in mantle source [79]. Besides, the oxidized nature (FMQ + 1.6, Figure 8) could significantly increase the sulfur solubility [22,80]. The parental magma should be enriched in Ni, Cu, and PGE, which are positive factors to form magmatic Ni-Cu-PGE sulfide deposits. However, the mineralization mainly occurs as a disseminated sulfide with a relatively low Ni-Cu tenor, and the PGE is depleted in the sulfide ore [26,29]. Similar PGE depletion had also been confirmed in other Ni-Cu deposits in the Northern Tianshan with a high degree of partial melting and oxidized parental magma (Table S6), and it was considered to be caused by early sulfide segregation at depths.
Early segregation could remove a majority of PGE because of their higher partition coefficients between sulfide and silicate (10 3 and 10 5 ) than that for Ni (300-1000) and Cu (10 2 -10 3 ) [81]. A slightly negative relationship between Cu/Pd and Pd is illustrated by the Baixintan deposits as well as the adjacent Lubei and Huangshandong deposits (Figure 10a), and this may be the effect of early segregation, since the Cu/Pd ratio could increase rapidly when Pd is strongly partitioned into a sulfide liquid during early segregation [82]. The evolution of sulfides formed by early segregation and by PGE-depleted melt is also depicted by the plot of Pd vs. Cu (Figure 10b). Lightfoot et al. [83] illustrated that an assumed initial PGE-undepleted melt of approximately 80 ppm Cu and approximately 10 ppb Pd produced PGE-and Cu-rich sulfide ores (red line). Conversely, sulfides that precipitated from a PGE-depleted melt (approximately 45 ppm Cu and approximately 0.278 ppb Pd) after early segregation would be Cu-rich but PGE-poor (blue trend line). Almost all the sulfide ores plot along the blue trend line, suggesting that they formed from a PGE-depleted melt due to early segregation (Figure 10b). Moreover, the 100% sulfide calculation of PGE in the Baixintan and Lubei deposits suggests that the primary magma underwent 0.007% early sulfide removal at depth and lost much chalcophile content. Then, the sulfide was segregated from PGE-depleted parental magma in a shallow chamber under R factors (R = silicate/sulfide, mass ratio) of 100 to 800 [29]. Though trigger factors needs further study, the early segregation has been widely recognized in the Northern Tianshan [26,29,44,84].  [83]). The numbers are sulfide percentages removed from the initial magma system. The yellow star represents the composition of the primitive mantle. (b) Pd vs. Cu. Data sources: Baixintan [26,29], Lubei [29], and Huangshandong [44].

Later Segregation
The later segregation is represented by disseminated sulfide occurring as the interstitial phase between olivine grains (Figure 3e,f), which indicates that sulfide segregation is posterior to olivine crystallization. The Fo-Ni variation during olivine crystallization has been modelled in the Baixintan deposit ( Figure 11 and Table S7). As a whole, the Ni content decreased as olivine crystallization. A few olivine samples in lherzolite have high Ni contents (approximately 2500 ppm) and plot along with the S-unsaturated curve, indicating that no sulfide segregation had occurred in the initial crystallization, while most olivine samples plot below the S-unsaturated line, especially for that in mineralized lherzolite and olivine websterite. The calculated result shows that the Ni depletion begun after about 3% olivine crystallization; then, the Ni content significantly dropped to approximately 1900 ppm, and the missing Ni was partitioned into segregated sulfide, which was evidently indicative of later segregation.  Table S7.
The contamination of crustal Si and S components are more important factors for sulfide segregation rather than fractional crystallization. The assimilation had been indicated by trace elements and Hf-O-Sr-Nd isotopic compositions [26,29,30,33]. In addition, the lower Ordovician volcanics are potential source of external S since a series of Cu-polymetallic deposits developed in volcanics, such as the Huangtupo and Huangtan VMS deposits, the Hongshan epithermal deposit, and the Heicaotan volcanic hydrothermal deposit [39]. Though the sulfide in the Baixintan deposit is near zero (δ 34 S = −0.7 to + 1.1) [22], the sulfide in volcanic strata also displayed a mantle signature and the possibility of incorporation of external S could not be excluded. Furthermore, assimilation of the Si component from southern granitoid is also a positive factor for S oversaturation. The fact is different from Ni deposits in the Beishan, for which immediate country rocks consist marble, and the contamination of CaO would hinder S oversaturation [25]. In addition, the relatively higher oxygen fugacity in Northern Tianshan can dissolve more sulfide than the Beishan (Figure 8). They may be key reasons that the Ni mineralization in Beishan is poorer than that in Northern Tianshan.

Conclusions
The mineralogical studies of chromite and silicate minerals in the Baixintan magmatic Ni deposit in Northern Tianshan (southern CAOB) have given us the following conclusions: (1) Two types of chromite were identified in mafic-ultramafic rocks. Type I chromite occurs as inclusions enclosed in olivine and has a primary and magmatic origin. Type II chromite with an interstitial phase crystallized from a more evolved melt and suffered from post-magmatic alteration.
(2) The parental magma is characterized by high temperature (1389 °C), high pressure (3.8 Gpa), and high Mg content (11.4 MgO wt%) with oxidized (FMQ + 1.6) and hydrous nature. (3) The mafic-ultramafic rocks originated from high-degree partial melting of a depleted mantle that had been modified by crustal components and metasomatized by subduction fluid in a postorogenic extensional setting. (4) Two stages of sulfide segregation have been recognized. The assimilation of crustal Si and S components played more important roles on sulfide segregation rather than fractional crystallization.