Chemical Composition and Genesis Implication of Garnet from the Laoshankou Fe-Cu-Au Deposit, the Northern Margin of East Junggar, NW China

: In order to reveal the formation mechanism of different garnets and its implications for the ﬂuid evolution in the Laoshankou Fe-Cu-Au deposit active incorporation of Fe 3+ and REE, caused by long ﬂuid pore residence and continuing inﬁltration metasomatism. Type 3 andradite formed by oxidizing magmatic ﬂuid and external high-salinity and Ca-rich basin brine with highest ƒO 2 in a mildly acidic open system, which is very different from typical skarn Fe-Cu deposit.

V I I I +1 , without the evaluation of the creation of site vacancy. The compositional variations from Type 1 to Type 3 garnet indicate significant differences of fluid compositions and physicochemical conditions, and can be used to trace the fluid-rock interaction and hydrothermal evolution of garnet. Type 1 grossular was formed by magmatic fluid under low water-rock ratios and ƒO 2 , and neutral pH environment by diffusion metasomatism in a nearly closed system with the preferential incorporation into the grossular of HREE. As the long fluid pore residence and continuing infiltration metasomatism under nearly closed-system conditions, fluids with high water/rock ratios were characterized by increased ƒO 2 , more active incorporation of Fe 3+ and REE, and formed Type 2 Al-rich andradite. In contrast, Type 3 garnet formed by oxidizing magmatic fluid under a mildly acidic environment with highest ƒO 2 and water-rock ratios, and was influenced by externally derived high salinity and Ca-rich fluids in an open system. Thus, the geochemical features of different types and generations of garnets in the Laoshankou deposit can provide important information of fluid evolution, revealing a transition from neutral magmatic fluid to oxidizing magmatic fluid with addition of external non-magmatic Ca-rich fluid from the Ca-silicate stage to the sulfide stage. The above proved the fluid evolution process further indicates that the Laoshankou deposit prefers to be an IOCG-like (iron oxide-copper-gold) deposit rather than a typical skarn deposit.
The Central Asian Orogenic Belt (CAOB; Figure 1a; [23]), situated between two major Precambrian cratons (the Siberian to the north and the North China-Tarim to the south), is the largest Paleozoic to Mesozoic accretionary orogeny in the world. The northern margin of the East Junggar terrane (Figure 1b; [24]), as the important metallogenic belt in the CAOB [25], is highly prospective for Paleozoic Fe-Cu-Au mineralization targeting. The Laoshankou Fe-Cu-Au deposit (41 km southwest of Qinghe City, Xinjiang, NW China) is one of the important magnetite deposits in the northern margin of East Junggar (Figure 1c; [26,27]). The previous speculation of various genetic types for the formation of the Laoshankou deposit include submarine volcanic-type [28][29][30] and skarn-type [31,32]. Recently, based on the microthermometry of fluid inclusions and stable isotopic analyses, an iron-oxide copper gold (IOCG)-like type with skarn alteration was proposed [26]. The confusion of defining genetic types highlights the importance of detailed hydrothermal fluid evolution and mineralization processes. On the basis of megascopic and microscopic texture relationships and mineral assemblages, significant two-stage mineralization existed in the Laoshankou deposit, e.g., Fe and Cu-Au mineralization [26,33]. Garnet in the Laoshankou deposit formed in both two mineralization stages, which show significantly different features [26]. Despite of many previous studies on geological descriptions, the age of the intrusion and mineralization, fluid inclusions and C-H-O-S stable isotopes for the Laoshankou deposit [26,31,32], few studies paid attention to the calc-silicate minerals, such as garnet, and the hydrothermal fluid evolution of pH and oxygen fugacity during the formation of garnets.
In this study, we conducted an integrated study of garnet mineralogy as well as major and trace element compositions of garnet from different stages to discuss (1) trace elements and REE substitution mechanism, (2) change in garnet composition at different stages, and (3) hydrothermal fluid evolution processes of the Fe-Cu-Au deposit.  [23]). (b) Simplified geologic map of North Xinjiang (modified from [24]). (c) Regional geologic and mineralization map of the SE Chinese Altay and NE East Junggar (modified from [26,27]

Regional Geology
The Chinese Altay and East Junggar located along the boundary between the Siberian and Kazakhstan-Junggar terranes [34], is an important part of the Central Asian Orogenic Belt (CAOB; Figure 1a). The northern margin of the East Junggar terrane is dominated by the Dulate Paleozoic active island-arc belt, which is separated from the Chinese Altay terrane to the north by the Erqis fault ( Figure 1b). Major structures in the region comprise mainly NW-trending and NNW-trending fault systems, as represented by the regional Erqis and Fuyun fault zones, respectively (Figure 1c; [26,27]). The Erqis Fault (or Irtysh Fault), striking NW-SE (290 • -300 • ), dipping 75 • to the NE, and extending 400 km, is one of the largest trans-current faults of Asia [35,36]. Previous research indicated that the Erqis Fault is a sinistral strike-slip fault, and was initiated from 290 to 275 Ma [36][37][38][39]. The Fuyun Fault, which has strikes of 342 • (NNW) with dips of approximately 70 • to the east, and extends 200 km, crosscuts the Erqis Fault and was considered to be dextral strike-slip fault [40,41].
The lithostratigraphy in the northern margin of East Junggar is mainly composed of the Upper Paleozoic volcanic-sedimentary sequences with minor exposed Upper Ordovician pyroclastic rocks, including shallow marine medium-fine grained sandstone to limestone of the Late Ordovician Jiabosaer Formation, marine pyroclastic and sedimentary rocks of the Early Devonian Tuoranggekuduke Formation, coastal-shallow marine intermediate-basic volcanic rocks of the Middle Devonian Beitashan Formation, shallow marine-paralic continental pyroclastic and volcanic rocks of the Middle Devonian Yundukala Formation, coastal-continental sandstone to limestone of the Late Devonian Kaxiweng Formation, pyroclastic-continental clastic rocks of the Late Devonian-Early Carboniferous Jiangzierkuduke Formation, continental-coastal volcanic-sedimentary rocks of the Early Carboniferous Nanmingshui Formation, multi-cyclic continental-rift bimodal volcanic rocks of the Middle Carboniferous Batamayineishan Formation, and continental pyroclastic rocks of the Permain Zhaheba Formation from the base upward [26,32,42].
Two stages of Paleozoic magmatism at 390-370 Ma and 320-270 Ma have been conformed in the northern margin of East Junggar terrane [43][44][45][46]. The 390-370 Ma I-type calc-alkaline intrusions were closely related to a series of bipolar subductions [47][48][49] and have been recently documented to form during basin inversion in a volcanic arc setting [26,50,51]. Coeval mineralization dominated by the arc-related Fe-Cu (-Au) and porphyry copper deposits generally occurred in the northern margin of East Junggar. The 330-280 Ma A-type alkaline intrusions were interpreted to have formed in a postcollision/intraplate extension setting [50,52], and genetically linked to deposits such as the Suoerkuduke skarn Cu-Mo deposit, the Xilekuduke porphyry Cu-Mo deposit and Aketasi hydrothermal-vein Au deposit.

Deposit Geology
The Laoshankou Fe-Cu-Au deposit has an estimated metal reserve of 3.26 Mt Fe at 33.5-36.42%, 9.8 kt Cu at 0.19-0.41%, and 0.14 t Au at 0.49-1.31 g/t [29,53]. The orebodies are hosted in the Middle Devonian Beitashan Formation (380.5 ± 2.0 Ma; [54]) that contains mainly basaltic/and esitic breccias and tuffs intercalated with fossiliferous limestone (Figure 2b). The basaltic breccias are mainly located in the footwall of the orebody, while the andesitic breccias are located in the hanging-wall of the orebody (Figure 3). Monzodiorite and diorite porphyry have close temporal and spatial relationship with Fe-Cu-Au mineralization in the Laoshankou deposit, with zircon U-Pb ages of 379.2 ± 4.4 Ma and 379.7 ± 3.0 Ma, respectively [45,55], which usually crosscut the earlier Beitashan volcanic rocks. There are two large NW-trending faults in the deposit, i.e., the Fuyun Fault in the north (dips at 60 • -70 • ) and the Shanqian Fault in the south (dips at~70 • , subparallel to Fuyun Fault) (Figure 2a; [29]). A series of subsidiary E-W-striking faults (i.e., F3, F4, F5 and F6) are present between Fuyun and Shanqian faults, and divided the ore district into a number of rhombic sectors.
Two ore zones have been delineated at Laoshankou, including the upper ore zone dominated by magnetite orebodies with minor Cu-Au mineralization, and the lower ore zone containing mainly Cu-Au orebodies ( Figure 3) [29,31]. Ore minerals mainly include magnetite, chalcopyrite, and pyrite, with minor chalcocite and hematite, while gangue minerals include garnet, diopside, epidote, chlorite, hornblende, actinolite, plagioclase, K-feldspar, quartz, and calcite, with rare apatite and sphene. Based on the mineral assemblages and ore textures, the ores in magnetite orebodies are mainly massive and banded magnetite-epidote-sulfide, whilst the Cu-Au orebodies are mainly characterized by massive and disseminated chalcopyrite [26].  Pervasive pre-mineralization Na-Ca silicate alteration were preserved at the Laoshank ou deposit, and two-stage mineralization of Fe and Cu-Au were precipitated in a multiphase paragenetic sequence. Three hypogene hydrothermal alteration and mineralization stages (Figure 4; [26]) include the Ca-silicate alteration (Stage I), amphibole-epidote-magnetite alteration/mineralization (Stage II) and pyrite-chalcopyrite mineralization (Stage III). The Ca-silicate alteration (Stage I) is recorded mainly by garnet with lesser amounts of pyroxene/diopside and scapolite in the wall rocks adjacent to orebodies. The amphibole/actinolite-epidote-magnetite alteration/mineralization (Stage II) is the main mineralization stage characterized by abundant magnetite, amphibole/actinolite, epidote with minor prior garnet. Abundant sulfide minerals are present in the pyrite-chalcopyrite mineralization (Stage III) with two major mineral associations, i.e., pyrite-epidote-quartzgarnet and chalcopyrite-amphibole/actinolite-chlorite/ripidolite-pycnochlorite (-garnet). Garnets formed in all three stages.

Electron Microprobe Analysis
Quantitative mineral analyses for garnet in situ major elements geochemistry and back-scattered electron (BSE) images were performed using a JEOL JXA-8100 at the Key Laboratory of Mineral and Metallogenic, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences (GIGCAS), Guangzhou, China. The analytical parameters used an accelerating voltage of 15 kV, a beam current of 20 nA, and 1 µm beam diameter. Natural minerals and metals were used as standards. Analysis data were treated with the conventional ZAF program, and the number of ions of garnet spots were calculated on 12 oxygens and with Fe 2+ /Fe 3+ calculated assuming full site occupancy [56]. The major element analyses of garnet by EMPA and the calculated number of ions for garnet are listed in Table 2.

LA-ICP-MS Analysis
Trace elements of garnet were conducted with a pulsed RESOlution M-50 Laser Ablation system (Resonetics, USA) coupled with an Agilent 7500a ICP-MS at the Key Laboratory of Mineralogy and Metallogeny, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences (GIGCAS), Guangzhou, China. Detailed operating conditions for the LA-ICP-MS instrument and data reduction have been described by Liu et al. (2008) [57]. Helium was applied as a carrier gas and argon as a makeup gas mixed with the carrier gas via a T-connector before entering the ICP. Each analysis includes a background acquisition of approximately 20 s for a gas blank, followed by 40 s data acquisition from the sample. Laser spots were 40 µm in diameter with successive pulses at 4 Hz. The external standards were BCR-2G and SRM612, analyzed twice every five to six sample analyses to correct the time-dependent drift of sensitivity and mass discrimination as quality control. Offline data reduction was performed with ICPMSDataCal v.10 [57], including selection and integration of background and analysis signals, time drift correction, and quantitative calibration. Trace elements and EPMA measurements were made in the same place in the minerals. For most trace elements (> 0.5 ppm), the accuracy is better than 5% of relative deviation with precision of 10%. All the analyses of garnet samples and standards are provided in Table 3.

Garnet Petrography
Based on geology, transmitted-reflected light microscopy, BSE and EPMA studies, garnets from the Laoshankou deposit can be divided into three types with changes in color. There is no obvious contact interface between the different garnets. (1) Type 1, the brown garnet in stage I occurs in the volcanic wall rocks (Figure 5a,b) with allotriomorphic texture (Figure 6a-c), which is optically dark and homogeneous in BSE images (Figure 6c). Type 1 garnet is generally replaced by the subsequent Type 2 garnet, magnetite, epidote, albite, and calcite (Figures 5b and 6a,b) with preservation of their original textures. (2) Type 2, the dark red garnet in the early sub-stage of stage II forms adjacent to the orebody (Figure 5c) or cross-cuts the diorite porphyrite near the orebody (Figure 5d) with fine-grained texture (Figure 6d-f), which coexists with K-feldspar and is replaced by subsequent epidote, calcite and chlorite (Figure 6d,e). Two forms of Type 2 garnet occur, the Type 2a garnet is irregular and replaces the Type 1 garnet (Figure 6c

Major Elements of Garnet by EMPA
Electron microprobe analyses show that the andradite content increases steadily for garnets from stage I to stage III in the Laoshankou deposit, ranging in composition from Adr 44 Grs 53 to almost pure andradite Adr 100 (Figure 7) with almandine, spessartine, pyrope and uvarovite collectively less than 5% ( Table 2)      The normalized multi-element diagram shows that all garnets in Laoshankou are generally depleted in large ion lithophile elements (LILEs, e.g., Rb, Sr, and Ba) relative to the average primitive mantle [59]. However, high field strength elements (HFSE) show a more consistent variation with major element geochemistry and petrographic observations. Both Type 1 and Type 2 garnets have higher U, Th, Hf, Y, and Ti, compared to the average primitive mantle (Figure 10). In contrast, Th, Nb, Hf, Y, and Ti compositions of Type 3 garnet are lower than the average primitive mantle (Figure 10).

Rare Earth Elements
Chondrite-normalized REE patterns of different types of garnets show distinct differences. Both Type 1 and Type 2 garnets are strongly depleted in LREEs, and have distinctly low ΣLREE/ΣHREE and (La/Yb) N ratios, as well as positive Eu anomalies (Figure 11a,b). ΣLREE/ΣHREE ratios (0.09-0.88) and ΣREE contents (8.14-32.8 ppm) of Type 1 garnet are lower than these (0.81-2.82; 29.2-124 ppm) of Type 2 garnet (Figure 11a,b). Type 3 garnet has a flat to moderately downward sloping (LREE relatively enriched and HREE-depleted) trend with obviously positive Eu anomalies (Figure 11c). Compared to Type 1 and Type 2 garnets, the andradite Type 3 garnet is characterized by the highest (La/Yb) N ratios (11.0-61.9) and positive Eu anomalies (Figure 11d and Table 2).

Garnet Substitution Mechanisms
The general formula of garnet is X 3 Y 2 Z 3 O 12 , where X site is divalent cations (Ca, Mg, Mn, or Fe 2+ ) in eight-fold coordination, Y site is trivalent cations (Fe 3+ , Al, and Cr) in octahedral coordination, and Z site is mainly Si in tetrahedral coordination [12]. Based on the ionic radii and garnet crystal radii for elements relevant in garnet chemistry, the only possible incorporation of U, REE 3+ , and yttrium is substitution of X site divalent cations (such as Ca 2+ ) in dodecahedral site [11,12,60]. For Eu 2+ , the substitution is isovalent and does not imply any charge imbalance by replacing the X 2+ site. However, the U, REE 3+ , and yttrium have different charge from the X 2+ site [12]. The charge imbalance of substituting U, REE 3+ and yttrium must be compensated either through (1) coupled substitution, like substitution of a trivalent cation (e.g., Fe 3+ or Al 3+ ) into the  (Figure 12a), and negative ∑REE 3+ vs. total Al (Figure 12b) correlations of Type 1 and 2 garnets in Laoshankou are not in favors of a YAG-type scheme. Furthermore, the Type 2 Al-rich andradite has higher ΣREE than the Type 1 grossular, which is also not consistent with the YAG-type equation. Therefore, the YAG-type is not the REE substitution mechanism for Type 1 and 2 garnets. This is completely different to most garnets from global skarn deposits [7]. Neither the total Al or Fe 3+ contents correlate with the REE 3+ contents (Figure 12a,b) in the Type 3 garnet, suggesting that the YAG-type is not the substitution mechanism for Type 3 garnets, either. All garnets in the Laoshankou deposit do not show obvious negative ΣREE 3+ vs. Fe 2+ (Figure 12c), and positive ΣREE 3+ vs. Mg (Figure 13a) correlations, implying that the incorporation of REEs of all garnets in Laoshankou is also not favorable to the menzerite-type substitution mechanism [10,60].  ; [64,65]). Owing to the low REE contents in all garnets from Laoshankou, the mere presence of Na in the garnet structure is in itself a proof of the intervention of the corresponding Na substitution scheme, even the Na contents of all Laoshankou garnets are far below those of Na-bearing garnet (up to 0.30% Na 2 O; [62]). However, the Na contents do not display a liner relationship with the REE 3+ contents for Type 1 and 2 garnets (Figure 13b), indicating that REEs in Laoshankou Type 1 and 2 garnets did not undergo Na + -REE 3+ coupled substitutions, but possible the creation of X 2+ (e.g., Ca) site vacancy. Type 3 garnet in Laoshankou show different REE correlations from those of Type 1 and 2 garnets, which may imply a different substitution mechanism for Type 3 garnet. A positive correlation between Na and REE 3+ implies that the key substitution mechanism for Type 3 garnet may be the Na cooperating substitution mechanism. However, the mechanisms involving the creation of structural vacancies are difficult to evaluate [66] for Type 3 garnets in Laoshankou.
In spite of the above controlled condition of crystal chemistry (e.g., coupled substitution mechanism), previous studies proved that incorporation of REEs into garnet can also be largely controlled by external factors, such as fluid chemistry, physicochemical parameters (e.g., pH, ƒO 2 ), and water-rock ratios [2,7,12,17,67]. Yttrium, Ca, U and REE 3+ in garnets from the Laoshankou deposit do not display obvious linear relationships (Figure 13c-e), indicated that incorporation of trace elements and REEs into the Laoshankou garnets can be also strongly governed by fluid-rock chemistry and physicochemical parameters.

Oxygen Fugacity (ƒO 2 )
Uranium as the redox-sensitive element can occurs in different valence stages (U 4+ and U 6+ ), which is significantly affected by ƒO 2 [68], making U to be an effective tool to estimate ƒO 2 of hydrothermal fluids. As the ionic radius of U 4+ is more comparable to that of Ca than U 6+ , U 4+ is more likely to substitute Ca 2+ [12,[68][69][70], meaning the higher U contents of garnet may indicate the lower ƒO 2 of hydrothermal fluids. Type 1 and Type 2 garnets have higher U contents (0.41-9.32 ppm) than these of Type 3 garnet (0.01-1.77 ppm), which partly indicate that Type 3 garnet formed under higher ƒO 2 . However, any conceivable substitution scheme (such as [Ca 2+ ] ) and increase of U solubility (caused by increasing ƒO 2 of the fluid system; [69]) may also cause these changes in U content, more evidence is discussed below.
In addition to the influence of U content, previous studies suggested that increased ƒO 2 of fluid favors radite precipitation [14,71], as the activity of Fe 3+ with respect to Al 3+ increased. The electrovalence of iron is sensitive to oxygen fugacity. Ferrous and the Fe 2+bearing minerals tend to be composed in a reduced condition, and the ferrous minerals could be converted to ferric minerals (such as Fe 3+ -bearing andradite and magnetite) if the reduced condition changed to oxidized, and vice versa [12,72]. The end member formula of Adr 91-100 Grs 0-8 Sps 0-2 Prp 0-1 Uv 0-1 for Type 3 garnet (Figure 7) indicated the highest ƒO 2 condition, which may explain the significant differences from typical skarn Fe-Cu deposit, e.g., the existence of magnetite at sulfide stage (stage III) of the Laoshankou deposit [26]. However, a variation in the relative content of andradite and grossular can also be due to the alumina activity of the fluid, or reaction with epidote and other alumina-bearing minerals. Fortunately, variation in the Fe 3+/ Fe T ratios could be explained in terms of variations in ƒO 2 . The relationships between the Fe 3+/ Fe T ratios and the ∑REE (Figure 12d) shows the high ƒO 2 for Type 3 garnet in stage III. Moreover, the clear drop in V content from stage II magnetite to stage III magnetite at Laoshankou also suggests that the oxygen fugacity at stage III increased [73]. As oxygen fugacity increased, the partition coefficient of V in magnetite decreased [74].
The ephemeral presence of hematite (change to mushketovite; [26,73]) at early amphi bole-epidote-magnetite stage (stage II) records the increased ƒO 2 from stage I to stage II ( Figure 4). The compositional variation from Type 1 grossular to Type 2 Al-rich andradite (Figure 7) partly indicates the slight increase of ƒO 2 from Type 1 to Type 2 garnet. However, the minimal variation of Fe 3+/ Fe T ratios between Type 1 and 2 garnets (Figure 12d) indicates that the change of ƒO 2 from stage I to stage II is indistinctive, which may explain the rapid transformation from hematite to mushketovite in stage II [26,73].

pH, Temperature and Salinity
Previous studies demonstrate that the pH of hydrothermal fluid can significantly affect the REE fractionation [2,75], for example, the fluids are commonly LREE-enriched and HREE-depleted with positive Eu anomalies under mildly acidic condition [75], whereas HREE-enriched and LREE-depleted with negative or no Eu anomalies in the nearly neutral condition [67,69,70]. Divalent Eu should predominate in skarn system at temperatures above 250 • C [76]. Especially, under mildly acidic conditions, the REE patterns are significantly controlled by the presence of Cl -, which can enhance the stability of soluble Eu 2+ (the dominant species being EuCl 4 2− ) with respect to REE 3+ and favor the transportation of Eu 2+ in hydrothermal fluid, forming distinctly positive Eu anomalies [12,77,78]. In Laoshankou, the isotopic thermometer and fluid inclusion microthermometry showed that all garnets formed at relative higher temperatures (> 250 • C with pressure of 0-3 kbar) [26], meaning Eu mainly occurred as Eu 2+ in the hydrothermal fluid during the formation of different types of garnets. Type 1 and 2 garnets in Laoshankou show LREE-depletion, HREE-enrichment, and a relatively weak positive Eu anomaly (Figure 11a,b), implying that they crystallized from a nearly neutral fluid. In contrast, the Type 3 garnet shows LREE-enriched, HREE-depleted, and a pronounced positive Eu anomaly (Figure 11c), indicating that Type 3 garnet formed in a mildly acidic and Cl − -rich fluid.

Constraints on Garnet Formation and Fluid Evolution
Trace elements in garnets, especially REEs, can be used to trace the fluid-rock interaction and hydrothermal evolution of garnet [16,68,79,80]. Normalized trace elements and REE patterns for different types of garnets in Laoshankou show distinct differences ( Figures 10 and 11), which indicate that the compositions of the fluid undergo an obvious change during the formation from Type 1 and 2 garnets to Type 3 garnet. The ∑REE contents increase from Type 1 grossular to Type 2 Al-rich andradite, but obviously decrease in Type 3 andradite ( Figure 12). Previous studies suggested that the low ∑REE contents in garnet may be accommodated by the loss of REE or dilution of relatively immobile REE or mineral/mineral REE partitioning [7,16]. In stage III, REE-bearing minerals as apatite and sphene are not common, even epidote (a REE-bearing mineral) forms as veins coexisting with quartz and calcite [26]. REE decreasing in Type 3 garnet via this method cannot be explained. Loss of REE caused by mixing with a fluid with lower REE content maybe the best way to account for the low REE pattern in skarn garnet, rather than dilution of REE in the hydrothermal system [17]. Long fluid residence or high water/rock ratios (such as intense infiltration metasomatism) can significantly change the REE contents [75,81]. So, the significant decrease of REE concentrations from Type 1 and 2 garnets to Type 3 garnet must be related to different hydrothermal fluid compositions and metamorphic processes.
Most previous studies have shown that magmatic hydrothermal fluids have in general very low REE contents (ΣREE = 10-100 ppm; [82]) and are LREE-enriched and HREEdepleted with a general positive but variable Eu anomaly [11,83], which contrast with the REE patterns of Type 1 grossular and Type 2 Al-rich andradite. However, for the magmatic fluid, HREE were preferentially incorporated into grossular [7,12], which can result Type 1 grossular in Laoshankou having a low REE concentration (mainly 1-100 ppm), but enrichment in HREE (Figure 11a). The positive correlation between Y and total REE indicates that grossular crystallized at equilibrium in a nearly closed system by diffusive metasomatism (Figure 14; [16]). Diffusive metasomatism means solute diffusion in a stagnant fluid system, which does not involve fluid movement permeating the rocks, but needs long pore fluid residence under nearly closed-system conditions [84]. Fluids produced by diffusive metasomatism have near neutral pH [7], which is consistent with the pH for Type 1 garnet. So, we infer that the Type 1 grossular formed by magma-derived fluids under relatively low water-rock ratios in a nearly closed system through a process of diffusion metasomatism ( Figure 15). Previous studies about fluid inclusions and stable isotopes (δ 18 O water values of 9.2‰-9.7‰) by Liang et al. (2019) [26] also indicated the addition of high temperature, medium-high-salinity, and Mg/Fe-rich magmatic fluid, which may exsolve from the diorite porphyry.  The REE pattern in Type 2 garnet is similar to that of Type 1, but there are changes of ∑REE contents. During the formation of Type 2 garnet in the early sub-stage of amphiboleepidote-magnetite, the neutral and Cl − -poor fluid has slightly increased ƒO 2 and more active Fe 3+ , which resulted in high Fe content in fluid and formed magnetite and the Al-rich Type 2 andradite replacing the Type 1 grossular. With an increase of iron-rich mineral compositions, the skarn-related fluid system changes from more diffusive (reaction skarn) metasomatism to more advective (infiltration) metasomatism [3,85]. Advective (infiltration) metasomatism will require the fluid flow. The similar Y and total REEs relationship, and REE pattern between Type 2 and Type 1 garnets show that Al-rich andradite also formed in a relatively Cl − -poor fluid system under nearly closed conditions. However, the mediumsalinity (~16 wt.%), and more Mg/Fe-rich features of fluid inclusions and higher δ 18 O water values (9.6‰-10.7‰) in stage II [26] indicate the influence of 18 O-rich wall-rocks (the Beitashan Formation). So, for the same magmatic fluid, an increase in water/rock ratios, continuing infiltration metasomatism and increasing of ƒO 2 promoted incorporation of REE into the Type 2 Al-rich andradite, resulting the increasing ∑REE contents from Type 1 grossular to Type 2 Al-rich andradite ( Figure 15).
For the Type 3 andradite at the sulfide stage, the fluid system has changed to mildly acidic fluid. As discussed above, Eu mainly occurred as Eu 2+ in the fluid of Laoshankou. The increasingly positive Eu anomalies in Type 3 garnet from stage III show high W/R ratios and enrichment in Eu 2+ in the sulfide-related ore-forming hydrothermal fluid, which also contained abundant Cl − and metallic elements. That is consistent with the high salinity (CaCl 2− NaCl system, 14-24 wt.%) of fluid inclusions in stage III [26]. The highest ƒO 2 , obvious decrease in ∑REE contents and enrichment in LREE for Type 3 garnet indicated a significant different fluid from Type 1 and 2 garnets. The trace element contents (U, Sc, Ti, V, Co, Ga, As, Hf, Th, Zn, and Au) also indicate differences in fluid composition between Type 1, Type 2 and Type 3 garnets. The REE pattern of LREE-enriched and HREEdepleted with pronounced positive Eu anomalies in Type 3 garnet perfectly matches the characteristics of oxidizing magmatic fluid [11,69,70,82,83]. However, previous studies of fluid inclusions and stable isotopes [26] proved a Ca-rich basin brine. The highest Ca content and salinity (CaCl 2− NaCl system) also suggest the influence of Ca-rich fluid in stage III. The significant changes in fluid properties, increase of secondary fluid inclusions in Type 3 garnet crystal and lack of correlation between Y and the total REEs all indicate an open system ( Figure 14) and significant influence of externally derived fluids with lower REE content. That was referred to as advective metasomatism [12,86], matching with the presence of non-magmatic external fluid by previous studies [26] (Figure 15). So, the type 3 garnet may be issued from the oxidizing magmatic fluid with addition of Ca-rich basin brine. The lowest contents of Co, Ni, Cu, and Zn in Type 3 garnet (Figure 9) may be caused by the continuous and massive sulfide crystallization.
Hence, a mixing process of different fluids is suggested in the Type 1, Type 2, and Type 3 garnets from different stages of Laoshankou deposit. The high-temperature, mediumhigh-salinity, and Mg/Fe-rich neutral magmatic fluid is dominated in stages I and the early substage of stage II with a control by increase of water/rock ratios and ƒO 2 . Furthermore, such fluids would have been replaced by oxidizing magmatic fluid with addition of highsalinity, Ca-rich basin brine in stage III, forming Type 3 garnet. Fluids in sulfide stage (stage III) as well as the product of magnetite are much different from a typical skarn Fe-Cu deposit, but have similarities with IOCG-type deposits (especially Central Andes and East Tianshan), which partly proves the particularity of Fe-Cu-Au deposits in the northern margin of East Junggar (such as Laoshankou).

Conclusions
The major findings of this study can be summarized as follows: (3) Type 1 and 2 garnets show higher total REE contents, LREE-depletion, HREEenrichment, and positive Eu anomaly, while Type 3 garnet shows lower total REE contents, LREE-enriched, HREE-depleted, and strongly positive and variable Eu anomaly.
(4) Type 1 grossular formed by magmatic fluid under low ƒO 2 and nearly neutral environment by diffusive metasomatism in a nearly closed system. Fluids forming Type 2 Al-rich andradite are neutral and Cl --poor with increased ƒO 2 , water-rock ratios, and more active incorporation of Fe 3+ and REE, caused by long fluid pore residence and continuing infiltration metasomatism. Type 3 andradite formed by oxidizing magmatic fluid and external high-salinity and Ca-rich basin brine with highest ƒO 2 in a mildly acidic open system, which is very different from typical skarn Fe-Cu deposit.
Author Contributions: Conceptualization, P.L., Y.Z. and Y.X.; Data curation, P.L.; Funding acquisition, P.L. and Y.X.; Writing-original draft, P.L.; Writing-review and editing, Y.Z. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement:
The data presented in this study are available on request from the corresponding authors.