Hydrothermal Chromitites from the Oman Ophiolite: The Role of Water in Chromitite Genesis

: The role of water ‐ rich solutions in the formation of chromitites has been the matter of controversy. We found small chromite concentrations (chromitites) in diopsidites, precipitated from high ‐ temperature hydrothermal fluids, in the mantle to the crust of the Oman ophiolite. Here, we present petrologic characteristics of the hydrothermal chromitites to understand their genesis. In the chromitites, the chromite is associated with uvarovite in the crust and diopside + grossular in the mantle. They are discriminated from the magmatic podiform chromitite by dominance of the Ca ‐ Al silicates in the matrix. The fluids responsible for chromite precipitation are possibly saline, being derived from the seawater circulated into the mantle through the crust. The saline fluids precipitate chromite to form chromite upon decompression and cooling, and transport platinum ‐ group elements (especially Pt and Pd). The fluids obtain Ca and Al from the crustal rocks and Cr from the mantle rocks during circulation. Saline fluids are also supplied from the slab to the mantle wedge, and can metasomatically precipitate chromite and pyroxenes within peridotites. They re ‐ distribute Cr and chromite in peridotites along with circulation of saline fluids in the mantle wedge.


Introduction
Chromian spinel or chromite has been recognized a typical high-T (temperature) igneous mineral, precipitating at an early stage from fractionating magmas or residing in refractory peridotites after magma extraction on partial melting at the mantle [1]. Chromitites, concentrates of chromite or chromian spinel, are therefore considered as plutonic rocks formed from high-T magmas in the upper mantle or within the lower crust [2]. The chromitite in the upper mantle typically occurs as pod-like or irregular forms with a dunite envelope within the mantle peridotite, especially harzburgite [3,4], and are grouped as podiform chromitites. The chromitite in the lower crust typically occurs as layers or seams with ultramafic or anorthositic rocks in layered intrusions, mainly of Precambrian age [5][6][7], and are grouped as stratiform chromitites.
The stratiform chromitite is formed by crystal accumulation from magmas that filled crustal magma chamber [5,6]. As for the podiform chromitite formation, an igneous origin during a harzburgitemagma reaction in the mantle has been proposed and appears widely accepted [4,8]. The essential role of aqueous solutions in the formation of podiform chromitites in the mantle and other chromitites has been the subject of controversy [4], and the essential involvement of water in chromitite formation  [28,30,31], and a panoramic view of the uppermost mantle to lower crustal section across Moho along Wadi Fizh (c). The location is represented by star on the geological map. The two diopsidites are about 70 m apart in stratigraphical level across Moho (red broken line).
The Wadi Fizh area is located around a boundary of segments of a spreading ridge responsible for the Oman ophiolite formation, where the Moho transition zone (MTZ) is very thin (~10 m) [29,32], i.e., a lowermost gabbro-dominant portion quickly grades to the harzburgite-dominant mantle section via the dunite-dominant MTZ. The uppermost part (~2 m in thickness) of the MTZ is composed of wehrlite with gabbroic bands and their network [29,32]. We tentatively call the base of the gabbrodominant portion Moho after Reference [29] in this article. The gabbroic layers just above the Moho have wehrlitic screen, which sharply decreases in amount upward in the lowermost crust (gabbro) section [29,32]. Gabbroic bands decrease in frequency downward in the MTZ to the harzburgite section. They are parallel with the layering of the crustal gabbros and the foliation of the mantle harzburgite (N70°N, 40°E on average [29]). Details of petrologic and geochemical characteristics of the deep-seated rocks around the Moho in this area are described and discussed in detail by Reference [29,33,34].
The crustal diopsidite is located in the layered gabbros located around 20 m above the Moho [24] (Figure 1). The diopsidite gradually changes to surrounding gabbro, which is mainly composed of augitic clinopyroxene and plagioclase (Figure 2a,b). The transitional rock is composed of gabbroic relics in fine-grained aggregates of anorthite and diopside [24]. The diopsidite contains anorthite-rich part (anorthosite). The diopsidite is composed of mainly diopside and anorthite, with a subordinate amount of uvarovite. Minor minerals are chlorite, tremolite, pumpellyite, epidote, and titanite. Minute aqueous fluid inclusions with daughter minerals (calcite or anhydrite) were found in titanite and anorthite [25]. Chromite forms small black clots within the uvarovite-rich part of the diopsidite (Figure 3a). The mantle diopsidite is located in a harzburgite-dunite portion in the mantle section around 50 m stratigraphically below the Moho [22,25] (Figure 1). It forms a flattened network almost parallel with the gabbro-peridotite boundary upsection [29,33] within the mantle peridotite ( Figure 2c). The thickest part reaches 80 cm and has a whitish aureole composed of tremolite [21,22]. The mantle diopsidite is mainly composed of diopside with subordinate grossular instead of uvarovite. The diopside is varied in color from white to pale green, and contains black chromite dots within the greenish part (Figures 2 and 3). Chromite is concentrated as small pod-like to thin segregation-like bodies, less than several centimeters in thickness, in diopsidite, with ragged boundaries with the wall diopsidite [21] (Figures 2 and 3). We recognized a thin chromitite veinlet branching from the main segregations at a high angle (Figure 3b).

Petrography of Chromitites (Chromite Concentrations)
The chromite clots (chromitite mini-pods) in the crustal diopsidite ( Figure 3a) are fine-grained, mainly composed of chromite, uvarovite, anorthite, and chlorite [21,25] (Figure 4a,b). The minerals are mostly less than 100 microns across. The chromite is euhedral to anhedral and opaque in thin section. It shows optical homogeneity by reflected light under the microscope (Figure 4b). Some euhedral grains are skeletal at their marginal part, where chlorite is closely associated (Figure 4a,b). Chlorite was also observed as minute rounded inclusions in chromite (Figure 4b). The chromites in the small chromite aggregates (Figure 3a) are similar in appearance to disseminated ones in diopsidite reported by Arai and Akizawa [21]. No PGM (platinum-group minerals) have been found under the microscope. The chromitite segregations in the mantle diopsidite [20] show a peculiar appearance ( Figures  3b-d and 4c). Aggregates of fine black chromite grains (mostly less than 0.1 mm across) fill the interstice of euhedral coarse pale-greenish diopside (less than 1 cm in dimension) (Figures 3b,c and 4c). The chromite is associated with grossular in the diopside interstice (Figure 4b-f). Individual chromite grains are opaque in thin section and rounded to euhedral in shape (Figure 4c,h). Some of them are in part hollow inside to various extents; the hollow part is composed of grossular or chlorite with or without vermicular chromite (Figures 4h,I and 5d,e). At an extreme case, chromite shows vermicular forms: chromite strings apparently trace euhedral chromite forms within grossular or chlorite (Figures 4h,I and 5e). The grossular matrix is faintly birefringent and oscillatory-zoned in a backscattered image (Figure 5d,e). The chromite grains contain abundant minute inclusions of grossular (Figure 4i). No PGM have been recognized in a thin section, even though carefully searched for. Instead, heazlewoodite is sometimes associated with chromite (Figure 4k,l). Chalcopyrite occasionally fills interstices of the matrix euhedral grossular (Figure 4m). It is interesting to note that the mineral assemblage of the hydrothermal chromitites and surrounding diopsidites from Oman is essentially the same as that of some low-temperature (<600 °C) metamorphic rocks which contain relic chromite from regional metamorphic belts [35][36][37][38]. At these metamorphic conditions, relic chromites are associated with Cr-bearing silicates, such as clinopyroxene (diopside to kosmochlor), uvarovite, chlorite, tremolite, pumpellyite, and epidote.

Mineral Chemistry
Mineralogical characteristics of diopsidite with or without chromite are described in detail by Arai and Akizawa [21], Akizawa and Arai [22], and Akizawa et al. [25], and summarized below. Analysis was made with a JEOL microprobe (JXA-8800R Superprobe) at Kanazawa University, Japan. Refer to Arai and Akizawa [20] for details of the analysis. Cr#, Mg#, and YFe are Cr/(Cr + Al), Mg/(Mg + Fe 2+ ), and Fe 3+ /(Cr + Al + Fe 3+ ) atomic ratios, respectively. Fe 2+ and Fe 3+ in chromite were calculated assuming spinel stoichiometry. All Fe in Ca-rich garnets and epidotes was treated as Fe 3+ , and as Fe 2+ in the other silicates. Selected analyses are listed in Table 1. New analysis was made only for the mantle hydrothermal chromitite here.

Chromites
Chromitites or chromite concentrations in diopsidites are essentially similar in mineral chemistry to the host diopsidites [21]. The chemical characteristics of chromite are summarized in Figure 5. It is interesting to note that the chromite of hydrothermal origin is high in Cr# and relatively low in YFe, similar in Cr-Al-Fe 3+ ratios to chromites in ordinary podiform chromitites from the MTZ/mantle from the Oman ophiolite and other ophiolitic complexes [4,8,21,25,39] (Figure 5a). The Mg# is lower than 0.5, clearly lower than in the ordinary deep-seated podiform chromitites, where the Mg# is mostly > 0.5 [21] (Figure 5b). TiO2 content of chromite is low, <0.5 wt%, in all the hydrothermal chromites from Wadi Fizh [21,25] and as low as that in the ordinary podiform chromitites (Figure 5c).
The chromites in the mantle diopsidite shows clear chemical heterogeneities [21,25]. Their core shows very complicated shapes, suggesting embayment by corrosion (Figure 5d of [25]), and is relatively low in Cr# and high in Mg#, around 0.5-0.6 and 0.3-0.5, respectively [21] (Figure 5a,b). It is similar in composition to chromites in chromitites from the MTZ to the upper mantle section of Wadi Fizh of the Oman ophiolite [21] (Figure 5a,b). The mantle to marginal part around the core exhibits higher Cr#, 0.7 to 0.8, and lower Mg#, 0.3 to 0.2 [21,25]. The vermicular chromite and the chromite forming the chromitite veinlet ( Figure 3b) are especially high in Cr# (>0.8) and low in Mg# (<0.2) (Figure 5a,b). The high-Cr# (>0.7) chromite tends to be high in MnO (up to 2 wt%) ( Table 1). Those "peripheral" chromites are low in TiO2 (Figure 5c). It is noteworthy that the veinlet chromite ( Figure  3b) is similar in chemistry to the chromite in the crustal diopsidite ( Figure 5).
The uvarovite garnets in the crustal diopsidite show the Cr# higher than 0.5, which is especially high (~0.8) when coexisting with chromite [21] (Figure 6). Uvarovite component in the grossular garnets in the mantle diopsidite is up to 30 mol% [21,25] (Figure 6). The grossular in the mantle hydrothermal chromitite shows weak oscillatory zoning in terms of Cr# (Figure 5d,e). All the garnets examined are low in andradite component (<10 mol%). Both the uvarovites and grossulars examined by us show around 100 wt% oxide totals, suggesting they are almost anhydrous [21,25].
All the chlorites analyzed are classified as clinochlore, according to Hey [40]. They are highly magnesian; the Mg# is 0.92 to 0.93 in the crustal diopsidite and 0.95 to 0.97 in the mantle diopsidite [25]. The Cr2O3 content of chlorite is around 2 wt% in the crustal diopsidite and from 3 to 5 wt% in the mantle diopsidite [24].

Bulk-Rock PGE + Re Geochemistry
Bulk-rock concentrations of PGE (Os, Ir, Ru, Pt, and Pd) and Re were determined with isotope dilution-mass spectrometry following a protocol of Ishikawa et al. [41]. The sample aliquots were digested and diluted into a 2 mL 1% HCl with spike solutions following a flow chart of Ishikawa et al. [41]. Osmium concentrations were determined by a Thermo Scientific Triton plus negative thermal ionization mass spectrometer (N-TIMS) at Tokyo Institute of Technology, Tokyo, Japan, whereas the other elemental concentrations were determined by a Thermo Scientific Element XR high resolutioninductively coupled plasma-mass spectrometer (HR-ICP-MS) at the University of Tokyo, Tokyo, Japan. The average total procedural blank compositions with one standard deviation are 0.30 ± 0.01 pg Os, 0.9 ± 0.4 pg, 4.8 ± 0.2 pg Ir, 29.9 ± 17.0 pg Pt, 2.5 ± 1.3 pg Pd, and 2.0 ± 0.2 pg Re (number of analyses is 3). All the analyses were blank corrected. The analytical results are given in Table 2.
The hydrothermal chromitite (= chromite-rich part of diopsidite) from the mantle section ( Figure 1) shows a PGE pattern flat to declining to Pd (Figure 7). It is similar in PGE pattern and level to the peridotites from the MTZ and mantle, as well as to a concordant podiform chromitite [42,43] (Figure 7). It is, however, relatively depleted in Pd (Figure 7). The chromite-rich diopsidite from the lower crust ( Figure 1) shows a fractionated PGE pattern, increasing from Os to Pt and Pd (Figure 7). This pattern is similar to that of the lower crustal gabbro (Figure 7) and of almost all terrestrial magmas [44].  [42]. Chondrite values are after Fischer-Gödde et al. [45]. Abbreviations: MD = chromite-rich mantle diopsidite, Hz = harzburgite, D-Hz = dunitic harzburgite, CD = chromite-rich crustal diopsidite, and Ga = gabbro.

PGE Mineral Chemistry
As stated above, no PGM have been discovered under the microscope even though carefully looked for. The bulk total PGE contents of the hydrothermal chromitites are almost equivalent to those of a concordant podiform chromitite in the Oman mantle exposed at Wadi Hilti, which contains PGM (mainly laurite) grains visible under the microscope (Figure 7) [42,43,46]. PGE in chromite were qualitatively analyzed by LA-ICP-MS (Agilent 7500s and Microlas Geolas Q-plus) at Kanazawa University [47] to understand the distribution of PGE (or PGM) in the chromitites (Figure 8). Analyses were performed by using a spot size of 150 μm and repetition rate of 10 Hz of laser ablation with energy density of 8 J cm −2 (Figure 8). We measured 8 isotopes of PGE ( 101 Ru, 103 Rh, 105 Pd, 189 Os,191 Ir,192 Os,193 Ir and 195 Pt). 29 Si, 34 S, 47 Ti, 61 Ni, 62 Ni, 63 Cu, and 65 Cu were also monitored for evaluation of the PGE measurement. Integrated time of 40 msec for each PGE isotope was used. Ablation for 35 s was started after measurement of the background signal for 30 s. Representative profiles of time-resolved signal count (cps: count/second) at each target spot are shown in Figure 9 and Figure 10. At first, we examined signal profiles of PGE isotopes in chromite in the concordant podiform chromitite (see Reference [42,43]) ( Figure 9). Although signal intensities of the PGE isotope are very low (20-150 cps) as indistinguishable from the background intensity (0-25 cps) (Figure 9a), remarkable spike-like peaks of PGE isotopes were observed in the profile of some analytical spots (Figure 9b,c). Small peaks of PGE were detected with an increase in 63 Cu intensity in the signal profile (Figure 9b). In the ICP-MS analysis of PGE isotope, mass interference caused by oxides (MO + ) and argides (MAr + ) should be evaluated because of production of 61 Ni 40 Ar + , 63 Cu 40 Ar + , and 65 Cu 40 Ar + interferes with 101 Ru, 103 Rh, and 105 Pd, respectively [48]. In our measurements, the argide production rate was very low as less than 5 × 10 −4 estimated from 101 Ru/ 61 Ni of NIST 610 ( Figure 9d). As shown in Figure 9c, 103 Rh peaks can be attributed to an increase of Cu. However, 61 Ni is not related to an increase of 101 Ru. 192   Similar spike-like peaks of the four PGE isotopes, such as 101 Ru, 103 Rh,191 Ir,and 192 Os, are characteristically seen in the signal profile of the chromitite segregations in the mantle diopsidite ( Figure 10). In contrast to the profiles from the concordant chromitite (Figure 9), PGE peaks are not related with fluctuation of Cu ( Figure 10). While 95 Pt and 105 Pd intensities are low (10-10 2 cps) and constant, the highest peak intensities of the 4 PGE isotopes are exceeding 10 4 cps, which is almost equivalent to intensity of 63 Cu.
In the analysis of the chromitite in the mantle diopsidite, spots mixed between chromite and garnet (grossular) were also measured. The wide range of Si intensity (defined by 29 Sinet) is mainly due to contamination of garnet ( Figure 10). Assuming that the 29 Si intensity of chromite in the concordant chromitite (3-10 × 10 4 cps) corresponds to < 0.01 wt% of SiO2, spot #88213 of the mantle diopsidite corresponds to < 3 wt% of SiO2 based on the 29 Si intensity (9 × 10 6 cps) (Figure 10a). On the other hand, #88202 and #88204 are low as equivalent to chromite in the concordant podiform chromitite in terms of Si intensity (10 4 -10 5 cps) (Figure 10c,d). As seen in signal profiles of #88213 and #88204 (Figure 10a,c), interestingly, 195 Pt was constantly detected. The intensity is clearly high (10 2 -10 3 cps) relative to the background count (0-50 cps). As mentioned above, HfO + interferes not only with 195 Pt but also with 193 Ir. Even if Hf is accommodated in the garnet and it causes interference by HfO + , #8813 and #88216 should be 300 and 33 times higher in 195 Pt than #88204, respectively. Accordingly, high 195 Pt at these analytical spots is not due to mass interference, and these spot data reflect that Pt is heterogeneously distributed in the hydrothermal chromitite in the mantle diopsidite.
The profile of #88215 also shows a similar spike-like peak of PGE isotopes (Figure 10f). 61 Ni and 65 Cu are considerably fluctuated through the analysis and they are increasing at the PGE peak. Sulfur was characteristically detected with the peak. On the other hand, high 105 Pd interval was remarkably observed after the peak. Because this interval is not dependent on the high-Cu interval, interference of 65 Cu 40 Ar + can be excluded. Therefore, this indicates that Pd was detected in the analytical spot.
In summary, in the hydrothermal chromitite in the mantle diopsidite the so-called IPGE (Irsubgroup; Os, Ir, and Ru) may be present as micro-nugget of alloys in chromite, which are not visible under the microscope. In contrast, Pt and Pd (especially Pt) are more evenly distributed, either as micro-nuggets or in chromite structure, in some chromite grains independently of the IPGE micronuggets.  Figure 8. Note that wide range of 29 Sinet (= average (intensity in ablation-background count)) is due to garnet (grossular) contamination at each analytical spot.).

Role of H2O in the Podiform Chromitite Formation
Chromitites in peridotitic rocks, such as podiform chromitites, are most commonly suffered from severe selective hydration. This has led us to make misunderstanding of the role of water in chromitite genesis. As stated in Arai [16], the selective hydration around chromitite is due to a highly magnesian character of olivine in the chromitite and its dunite envelope. If the chromitite is cooled with successive water supply, the dunitic matrix of chromitite and surrounding dunite, of which olivines have been enriched in Mg via Mg-Fe redistribution with chromite, will be preferentially altered at higher temperature than ambient peridotites [50] (Figure 11). This apparently selective settling of water in and around chromitite is one of subsolidus phenomena due to high concentrations of chromite, but the water supply is not essentially related to the primary concentration of chromite [16] (Figure 11). This can occur in a microscopic way in peridotites, where the olivine has been selectively hydrated around chromian spinel ( Figure 12). This is in part due to a reaction between chromian spinel and olivine to form chlorite at some higher temperatures prior to initiation of serpentinization [16]. In conclusion, the selective serpentinization or chloritization in and around chromitite is not the evidence for involvement of water in primary formation of chromitite.
The presence of hydrous silicate inclusions in chromite is possibly not related to concentration of chromite for chromitite. The chromite-hosted silicate inclusions, which represent trapped melt rich in water and Na, are also commonly found in dunites, troctolites, and related rocks from the current ocean floor [51][52][53], in addition to chromitites [51,54,55]. The formation of troctolites and dunites from mantle harzburgite is closely associated with dissolution of mantle orthopyroxene via reaction with an invading magma [51,56]. The reaction between the mantle orthopyroxene and exotic melt from deeper part produces chemically heterogeneous secondary melt, of which some part is rich in silica and alkalis [57]. Peridotite-magma reaction is also one of the important elements of podiform chromitite formation [8]. We suggest that the presence of chromite-hosted hydrous inclusions is not directly related with the chromite concentration process. And, again, it does not provide any evidence for the involvement of hydrothermal fluids in podiform chromitite formation.  [26]. No petrographical and textural evidence for precipitation of those high-Cr, -Fe chromites directly from hydrothermal fluids.

Precipitation and Concentration of Chromites from Hydrothermal Fluids
Both dissolution and precipitation of chromites from hydrothermal fluids were recognized in the Oman ophiolite by Arai and Akizawa [21], Akizawa and Arai [22], and Akizawa et al. [25]. Concentration of the hydrothermal chromites was also clearly shown by those authors. The temperature of the hydrothermal chromitite formation of Oman is possibly 700-800 °C, according to the critical mineral assemblage (anorthite-diopside-chlorite) and mineral chemistry [21,23,25]. Some of the minerals in the diopsidites and related chromities are possibly of retrograde origin.
The seam-like segregation of chromite at the interstice of coarse euhedral diopside in the mantle diopsidite indicates that a chromite-oversaturated fluid got formed by fractional crystallization of diopside ( Figure 2). The dendritic or skeletal form of chromite ( Figure 3) in chromitite suggests chromite oversaturation after fast precipitation of large amounts of diopside from the fluid. After the diopside production, Cr-bearing grossular is co-precipitated with chromite [22]. The thin chromitite veinlet branching from the main segregations (Figure 3b The core of chromite grains of the hydrothermal chromitite in the mantle diopsidite is similar in chemistry to chromites in podiform chromitites or peridotite from Oman (Figure 5a-c). Combined with the textural character of the core chromite (Figures 4 and 5d,f), it possibly represents relics of mantle chromites from podiform chromitite or peridotite. This is consistent with the similar PGE pattern between the chromite-rich part of mantle diopsidite and chromitite (concordant)/peridotite (Figure 7). This further means that the mantle chromite was incorporated in the fluid before or during metasomatic precipitation of diopsidite and subsequently modified/digested in part to give rise to formation of Cr-rich fluids, which precipitated chromite during cooling. The mantle chromite was most probably incorporated from a magmatic chromitite seam in the uppermost mantle just beneath MTZ (Figure 1).
The micro-pods of chromitite in the crustal diopsidite exist in the uvarovite-rich part [21] (Figure  3a), which is enveloped by anorthosite-rich part (or anorthosite) within crustal layered gabbros. The field relation indicates that the fluid invading the gabbro first crystallized anorthite via reaction with the wall gabbro (Figure 2a,b). Cr-rich diopsidite was formed within the fluid conduit lined with anorthite-rich rocks (Figure 2a). The Cr# of the fluid fluctuated in the conduit and precipitated chromite and uvarovite when it was high and low, respectively [21]. The fluid precipitated both chromite and high-Cr# uvarovite when it was intermediate in Cr# [21]. When it was very high in Cr#, the degree of chromite-oversaturation was so enhanced that skeletal or dendritic chromite was crystallized and aggregated to form the chromitite micro-pods (Figure 3a).
The bulk-rock PGE characteristics suggest that Pd was released from the initial podiform chromitite during reaction with the fluid that dissolved chromite because of the slight drop of the PGE pattern at Pd for the mantle hydrothermal chromitite (Figure 7). The mantle hydrothermal chromitite may be representative of a residual character of a podiform chromitite when reacted with hydrothermal fluids, which can dissolve its chromite. The highly fractionated Pt-, Pd-rich pattern of the crustal hydrothermal chromitite (chromite-rich diopsidite), which is a cumulate from the hydrothermal fluid, is due to enrichment of Pt and Pd in the fluid [59][60][61]. The distribution of PGE in the hydrothermal chromite (Figures 8 and 10) suggests a high heterogeneity in PGE (especially Pt) in the hydrothermal fluid that is a reaction product between the circulated fluid and the primary mantle chromite. The chemically heterogenous fluid was separated and homogenized to produce an as-a-whole high-Pt (-Pd) fluid, which could precipitate chromite with high Pt/Os ratio upsection (Figure 7).
It is noteworthy that the main Cr-containing mineral is diopside after chromite in the mantle diopsidite but is uvarovite (garnet) in the crustal diopside. The Cr-rich diopside coexists with grossular in the mantle section, and the uvarovite, with anorthite in the crustal section. These assemblages may be related to each other by a reaction, CaCr2SiO6 (Ca-Cr tschermakite) + Ca3Al2Si3O12 (grossular) + SiO2 = CaAl2Si2O8 (anorthite) + Ca3Cr2Si3O12 (uvarovite).
The fluid involved in formation of the crustal diopsidite is higher in silica content than that for the mantle diopsidite. The crustal fluid was formed from the fluid passing through the mantle through a kind of crystallization differentiation via precipitation of the minerals of the mantle diopsidite because the two diopsidites are spatially very close to each other (about 70 m apart in stratigraphic level).

Origin of the Fluid
The fluid responsible for formation of the diopsidite + chromitite was possibly originated from the seawater, as Python et al. [23] suggested ( Figure 13). Detection of Na-bearing H2O-rich fluid from inclusions in titanite in the crustal diopsidite [25] is consistent with this idea. Chlorine and other anions may form complexes with chromium that are soluble in water-rich fluids [21,62,63]. Saline fluids are also expected to be released from the slab to the mantle wedge [64][65][66][67]. If we carefully look peridotite xenoliths derived from the mantle wedge, we frequently observe some textures suggesting accumulation of fine chromite or chromian spinel grains ( Figure 14). They are mostly within olivine and sometimes form secondary inclusions trails associated with fluids and sulfide ( Figure 14). We suggest that the chromite or chromian spinel of this occurrence are precipitated from aqueous fluids or highly hydrous melts released from the slab. Precipitation of chromian spinel is observed in some peridotite xenoliths of mantle wedge origin from the Avacha volcano, Kamchatka [68]. Metasomatic addition of pyroxenes, including diopside, is also recognized in highly metasomatized peridotite xenoliths from Avacha [69]. Those mean that some of the fluids derived from the slab possibly precipitate chromite or chromian spinel and Ca-rich silicates in the mantle. The fluids originated from the crust obtain calcium and aluminum when they pass through the crustal rocks, and chromium when they pass through the mantle rocks.

Conclusions
Hydrothermal chromitites can be formed from water-rich fluids, both in the crust and in the mantle. The aqueous fluids are saline, being derived either from the seawater or from the subducting slab. Chromites are precipitated from such Cr-rich saline aqueous fluids on transportation upward because of drop of Cr solubility on decompression and/or cooling of the fluids [63], possibly similar to the magmatic chromite precipitation in magma chamber [70]. If a sudden decrease in oxygen fugacity is available in the saline fluids [71], chromite is also possibly concentrated from the fluids. The hydrothermal chromitites are characterized by the association of Ca-Al silicates, such as diopside and Ca-garnets. They are in contrast to magmatic podiform chromitites, of which silicate matrix is mainly composed of olivine. The fluid possibly redistributes chromium and chromite (or chromian spinel) within the mantle wedge significantly, where saline aqueous fluids are possibly circulating [67]. It can mobilize Pt and Pd in the mantle wedge and crust significantly, although we need more data and information on this subject.