- © 2007 E. Schweizerbart’sche Verlagsbuchhandlung, D-70176 Stuttgart
This is the first detailed report on local concentration of chromian spinel in a dunite from an ultraslow spreading ridge, the Southwest Indian Ridge (SWIR). The sample was collected from an outcrop with detailed observations using submersible SHINKAI 6500 of the Japanese Marine Science Technology Center. The dunite occurs as a tabular-shaped layer in a lherzolite host outcrop. Spinel is found as a string of small micropods 2–3 centimeters in size. These spinel micropods make a layer in the middle part of a spinel-poor dunite (< 1 vol % spinel) parallel to the lithological boundary between dunite and enstatite-poor harzburgite. The enstatite-poor harzburgite has relatively high-Cr# spinel (> 0.4) compared with other peridotite samples in the studied area (lherzolites to harzburgite with low-Cr# spinel, typically Cr# ≤ 0.3). The occurrence and chemical compositions of clinopyroxene in the enstatite-poor harzburgite suggest that some clinopyroxenes crystallized from infiltrated interstitial melts. The host peridotites are interpreted as a residue of relatively low degrees of partial melting consistent with a location along the SWIR far from a mantle hot spot. This was then followed by crystallization of clinopyroxene from interstitial melt in the dunite. Irrespective of their small size, the lithological relationships between the spinel micropods and the host peridotites are the same as those for podiform chromitite in ophiolites and orogenic peridotites. The spinel Cr# in the micropods (0.3) is compatible with the lower range of those in basalts from SWIR far from hot spot. The spinel micropods were mainly formed by interaction between relatively depleted peridotite and a locally significant volume of basaltic melt traversing the upper mantle. This study coupled with the previous works on chromitites suggest that podiform chromites occur in every geodynamic setting, though economic concentrations of chromite (Cr-rich spinel) are unlikely to occur in the mantle at ultraslow spreading ridges.
Chromian spinel-rich rocks, i.e. chromitites, are commonly found as lenticular or pod-shaped bodies within tabular to irregular shaped dunite bodies, termed podiform dunite, enclosed in mantle tectonite peridotites in many ophiolites and orogenic peridotite massifs. They are called podiform chromitite (e.g., Thayer, 1964) and have been mined as both a source of chrome and high alumina refractories. Chromitites appear to form by selective crystallization of chromian spinel from melts for a period of time and/or by selective concentration of chromite due to physical mechanisms such as flow differentiation and/or metamorphic differentiation (e.g., Lago et al., 1982; Leblanc & Ceuleneer, 1992). Chromian spinel is, however, a minor phase produced during closed-system cotectic crystallization of basaltic magmas (e.g., Campbell & Murck, 1993). Edwards et al.(2000) suggested that the water content of basaltic melt plays a significant role in the formation of podiform chromitites as hydrous melts have a greater abundance of octahedral sites, which would promote greater chrome solubility. Irvine (1975, 1977) proposed that mixing of Si-rich and primitive basaltic melts would drive melts into the chromite phase field leading to formation of layered stratiform chromitite, another type of chromitite associated with large layered intrusions. This model can be basically applied for the origin of podiform chromitites (Dick & Bullen, 1984; Arai & Yurimoto, 1994; Zhou et al., 1994).
It is generally accepted that Si-rich secondary melts can be produced locally in mantle peridotites by reaction between the primitive basaltic melts and peridotites due to dissolution of pyroxene and precipitation of olivine during melt-rock reaction (e.g., Dick, 1974; Kelemen, 1990). Where (or when) the melt supply is high, such Si-rich secondary melts can subsequently mix with more primitive basaltic magma in melt transport conduits. The origin of podiform chromitites is thus generally explained by crystallization from a hybridized melt because the mixed magma is expected to be oversaturated with chromite (e.g., Arai & Yurimoto, 1994; Zhou et al., 1994).
Field and geochemical evidence suggests that melt-mantle interactions are common in the upper mantle. Podiform chromitites are, however, more commonly found in harzburgite-dominated ophiolite and orogenic peridotites than in lherzolite-dominated ophiolite and orogenic peridotites (Boudier & Nicolas, 1985; Noller & Crater, 1986; Leblanc & Temagoult, 1989; Nicolas & Al Azri, 1991; Roberts & Neary, 1993). Dunite crosscutting layering and/or foliation plane of the host peridotites is commonly documented in the mantle sections and is interpreted to be a result of melt-rock interactions (e.g., Dick, 1976; 1977a, b; Dick & Sinton, 1979; Quick, 1981; Piccardo et al., 2007). The crosscutting dunites are not always associated with chromitites (Quick, 1981; Nicolas & Al Azri, 1991). Even if podiform chromitite exists in lherzolite-dominant ophiolite and orogenic peridotites, it is usually found closely associated with harzburgite (Zhou et al., 1996; Morishita et al., 2006).
Although podiform chromitites are common in supra-subduction zone mantle environments represented by ophiolites, their occurrence in other tectonic settings is still in debate (Roberts, 1988; Yumul, 1992; Arai, 1995, 1998, Arai & Yurimoto, 1995; Prichard et al., 1996; Robinson et al., 1997; Schiano et al., 1997; Zhou & Robinson, 1997; Zhou et al., 1998). This is partly because few chromitites have been collected from well-constrained tectonic settings. Several authors have suggested that major oceanic ridges are unlikely places for chromite mineralization to take place (Roberts, 1992; Zhou & Robinson, 1997; Arai, 1997a). Most abyssal peridotites, however, have been sampled at transform faults far from regions where focused melt flow likely occurs in the shallow mantle beneath ocean ridges (Dick, 1989). The average abyssal peridotite is close to harzburgite, while dunite, away from transform faults at the Gakkel, Southwest Indian Ridge and Mid-Atlantic Ridges, constitutes from 4 to 10 % of all peridotite sampled, and commonly contains small chromitite segregations, suggesting that podiform chromitites may be common in the shallow abyssal mantle at all spreading rates (Dick, Tivery & Tucholke, G3, in revision).
We report here the first detailed description of the petrology and geochemistry of a small dunite spinel segregation and host peridotite sampled from an ultraslow spreading ocean ridge representing a key end-member for seafloor spreading, crustal accretion and magma genesis. The sample, from the Atlantis II Fracture Zone at the Southwest Indian Ridge, despite its small size, effectively mirrors the formation of podiform dunites and chromitites commonly found in ophiolite peridotite massifs.
Geological background and sample description
The Southwest Indian Ridge (SWIR) is an ultraslow spreading ridge with a 14 mm/year full-spreading rate (Hosford et al., 2003). The Atlantis II Fracture Zone is a transform-valley, a 199 km offset of the SWIR (Fig. 1⇓). Hole 735B of the Ocean Drilling Program (Legs 118 and 176) succeeded in recovering 1.5 km section of gabbroic rocks at Atlantis Bank which is located 100 km south of the SWIR rift valley on the crest of the eastern traverse ridge of the Atlantis II Fracture Zone (e.g., Dick et al., 2000) (Fig. 1⇓).
Many peridotite samples were collected from outcrops exposed on the eastern rift valley wall of the Atlantis II Fracture Zone by the ABCDE cruise of the Japan Marine Science Technology Center (JAMSTEC) using submersible SHINKAI 6500 (Fig. 1⇑). Dive 643 collected only peridotites from outcrops or on outcrops from 3396 m to 2564 m depth, whereas other 5 peridotites and 2 gabbros were collected from recent talus. Another dive (no. 649), which was conducted along an extended track from Dive 643, confirmed that gabbros are only exposed of 2561-2493 m depth. Accordingly, we describe only samples collected from outcrops and on outcrops (Table 1⇓). Ten of 13 samples are lherzolite, two samples are clinopyroxene-bearing harzburgite (643R06, 643R07) and one is dunite (643R15) (Fig. 2⇓). These indicate that mainly fertile peridotites crop out along the dive track, consistent with a slow-spreading ridge where degree of partial melting is expected to be relatively low often leaving lherzolite or clinopyroxene-rich harzburgite residues.
Dunite 643R15, the focus of this paper, has a distinctive ~1 cm spinel-rich layer a few centimeters in size with numerous aligned small irregular micropods of spinel. The dunite has a contact with a clinopyroxene-rich harzburgite (Enstatite-poor harzburgite hereafter) (643R15-4 in Table 1⇑) roughly parallel to the spinel-rich layer at the edge of the sample (Fig. 3⇓). Olivine textures in the dunite are masked by extensive serpentinization and alternation while, based on pyroxene morphology, the Enstatite-poor harzburgite has a protogranular texture. As the dunite is in contact with the harzburgite and occurs in an outcrop with massive granular peridotite, it is likely a crosscutting tabular body typical of many dunites in mantle peridotites. A ~1–1.5 cm gabbro vein also cuts irregularly through the sample, crosscutting the spinel-rich layer (Fig. 3⇓) with a sharp contact with dunite (Morishita et al., 2004), representing a late intrusion crystallized from basaltic melt. The gabbro vein contains a small amount of titanium and/or zirconium oxide minerals such as rutile, ilmenite, zircon and srilankite (Morishita et al., 2004). Although the gabbroic vein is extensively altered, anhedral granular-shaped orthopyroxene is the main primary igneous phase remaining. Some orthopyroxenes in the gabbro vein, however, are interpreted to be formed by reaction between SiO2-component in the melts and olivine in the peridotite hosts (Morishita et al., 2004). Many carbonate veins ranging from a few millimeters to a centimeter in thickness then cut all lithologies.
The dunite mainly consists of olivine (now completely serpentinized and/or altered) with minor amount of spinel and clinopyroxene. Clinopyroxene usually occurs with spinel as a symplectitic mineral aggregate (cpx-spl symplectite hereafter) (Fig. 4a⇓). Only a minor amount of very fine-grained clinopyroxene ( < 30 μ m) is found as isolated grains in olivine matrix. Rounded anhedral spinel also occurs as discrete grains ( < 5 mm across). The dunite, excluding the spinel-rich layer, is spinel-poor ( < 1 vol %) (Table 1⇑). Mineral inclusions in the spinel (e.g., pyroxenes, pargasite and phlogopite), which are common in ophiolites and orogenic peridotites (e.g., Johan et al., 1980; Talkington et al., 1986; Augé, 1987; Lorand & Ceuleneer, 1989; McElduff & Stupfl, 1991; Matsumoto et al., 1995; Schiano et al., 1997; Matsukage & Arai, 1998; Ahmed & Arai, 2002, Morishita et al., 2006), were not found. The protogranular Enstatite-poor harzburgite at the contact consists of olivine with small amount of pyroxenes and spinel. However, cpx/(cpx+opx) ratio is high similar to the other typical lherzolites (Fig. 2⇑). Clinopyroxene in the Enstatite-poor harzburgite has several modes of occurrence; coarse discrete grain (Fig. 4b⇓), interstitial grain (Fig. 4c⇓) and rimming of large orthopyroxene grain (Fig. 4d⇓) (referred to as discrete, interstitial and rimming clinopyroxenes hereafter). This textural variability likely indicates multiple origins as discussed below.
Other peridotite samples collected from the same dive sites have also undergone variable degrees of serpentinization and/or alteration, but have protogranular to weak porphyroclastic textures. Some peridotites are intruded by gabbroic vein (643R07) or clinopyroxenite (649R02).
Major-element compositions of minerals were determined with a JEOL JXA-8800 Superprobe at the Center for Cooperative Research of Kanazawa University. The analyses were performed with an accelerating voltage of 15–20 kV and a beam current of 15–20 nA using a 3 μ m diameter beam. Details of the analysis are shown in Morishita et al., (2004). Spinel and clinopyroxene in the dunite near the gabbro vein have a distinctively TiO2-enriched character compared to those far from the gabbro vein (Fig. 5⇓) (Morishita et al., 2004). The mineralogy of the gabbros suggests that the TiO2-rich melt was required for the formation of the gabbro. However ilmenite grains in the gabbro vein are distinctively high in MgO compared with those in both oxide-poor olivine gabbros and Fe-Ti-rich oxide gabbros collected from the studied area (Bloomer et al., 1989; Dick et al., 2000). Furthermore, the Cr2O3 content in rutile in the gabbro vein is also high. Based on these features, Morishita et al., (2004) suggested that the gabbroic vein was formed from an in-situ highly fractionated melt from a MORB-type melt during ascent in the upper mantle. The TiO2-enriched minerals would result from the interaction with the TiO2-rich fractionated melts responsible for the formation of the gabbro vein after the formation of spinel-rich layer. In order to avoid the metasomatic effects of the TiO2-rich melts on samples, we basically discuss chemical characteristics of minerals far from the gabbro vein. It is, however, difficult to completely avoid the metasomatic effects of the TiO2-rich melts because of the size of the sample.
There are no apparent differences in the chemistry of spinel between spinel-rich layer and discrete grains in the dunite except for Mg# (Table 2⇓). The Cr# and TiO2 contents of spinel are 0.3 and 0.1 wt. %, respectively (Fig. 5⇑ and 6⇓). Those of spinel near gabbro reach 0.5 and 1.5 wt. %, respectively (Morishita et al., 2004) (Fig. 5⇑ and 6⇓). The Mg# of chromite is 0.74 and 0.70 for spinel-rich layer and discrete grain in dunite, respectively. The chemical compositions of clinopyroxene are different reflecting the differences in occurrences. The symplectite clinopyroxenes are higher in the Cr2O3 and Na2O contents (1 and 0.6 wt. %, respectively) than discrete clinopyroxenes (0.2–0.4 and < 0.3 wt. %, respectively) (Fig. 7⇓ and Table 2⇓).
The chemical compositions of spinel in the Enstatite-poor harzburgite 643R15-4 have variations in the Cr# and TiO2 contents ranging from 0.4 to 0.6, and from 0.1 to 0.7 wt. %, respectively (Fig. 5⇑ and 6⇑). The high Cr# spinel is also high in TiO2 content. These grains were also affected by interaction with the TiO2-rich melt for the formation of the gabbro vein. The chemical compositions of clinopyroxene between discrete grains and interstitial grains are similar to each other in the core, whereas those of rimming grains are slightly different from the former two types. The Al2O3 and Na2O contents are lower and higher in the rimming clinopyroxene than in the other two clinopyroxenes (Fig. 7⇑ and Table 2⇑). The Al2O3, Cr2O3 and TiO2 contents of the core of orthopyroxene porphyroclast are 3.5, 0.5 and 0.1 wt. %, respectively (Table 2⇑).
The spinel Cr# in the remaining peridotites collected from the same dive sites ranges from 0.2 to 0.35, mostly around 0.2 (Fig. 6⇑), except 643R07 (0.4–0.6) which has a gabbroic vein. Spinel TiO2 contents are low, < 0.1 wt. %, except 643R07 (0.4–0.8 wt. %). The peridotite spinel chemical compositions (except 643R07) are consistent with those in mantle residue of a low-degree of partial melting predicted for slow-spreading ridges (Dick & Bullen, 1984; Arai, 1994a,b). The forsterite content of olivine is 90 in lherzolite and 91 in harzburgite (Table 5⇓). The Al2O3, Cr2O3 and Na2O contents of the core of clinopyroxene porphyroclast are 6–7 wt. %, < 1.3 wt.%, and < 0.3 wt. % in lherzolite, and 5, 8, and 0.6 wt. % in harzburgite, respectively (Fig. 7⇑ and Table 5⇓). The Al2O3, Cr2O3 and TiO2 contents of enstatite porphyroclast cores are 5, 1 and <0.04 wt. % respectively in lherzolite, and 4, 1 and 0.1 wt. % in harzburgite (Table 5⇓).
Origin of the Enstatite-poor harzburgite
Mineral textures in the peridotites are typical of granular peridotites emplaced by solid-state flow to the base of the crust at ocean ridges, and these rocks appear little affected by higher-grade ductile deformation associated with detachment faulting and emplacement seen elsewhere. In particular, the dunite crosscuts these peridotites, and the spinels are irregular aggregates that show little visible evidence of deformation. Thus, the layering consisting of the spinel micropod layer, spinel-poor dunite and the adjoing Enstatite-poor harzburgite are a primary magmatic feature rather than a mechanical accumulation of minerals due to physical mechanisms such as flow differentiation and/or metamorphic differentiation.
Atlantis II transform residual peridotite spinel compositions have a limited local variation with low Cr# in residual mantle rocks (Dick, 1989; Johnson & Dick, 1992). The previous study on abyssal peridotites suggests that spinel Cr# is a good indicator for the degree of partial melting of plagioclase-free-and vein-free peridotitic residues (Dick & Bullen, 1984; Arai, 1987; Hellebrand et al., 2001). The fertility of most SWIR peridotite away from mantle hotspots is the result of the slow spreading rate combined with the cooling effect of the transform faults (Hellebrand et al., 2002). Locally, however, spinel in the Enstatite-poor harzburgite is relatively high (Cr# = 0.4–0.6) compared with those in other peridotites collected from the same site (this study) as well as previous work from the Atlantis II Fracture Zone (Dick, 1989; Johnson & Dick 1992) (Cr# = 0.15*–0.35) (Fig. 6⇑). Higher spinel Cr# generally correlates well with depletion and inferred degree of mantle melting in abyssal peridotites (Dick & Bullen, 1984; Dick et al., 1984; Hellebrand et al., 2002). On the other hand, TiO2 and Na2O contents of clinopyroxene in the Enstatite-poor harzburgite are all higher than predicted by partial melting (Fig. 7⇑). This combined with textural observations on clinopyroxene suggests that some or most of clinopyroxene in the Enstatite-poor harzburgite might be not a simple residual origin but have partially re-equilibrated with later infiltrating MORB melt. Similar crystallization of clinopyroxene from infiltrating interstitial melt has been found for other slow-spreading ridge peridotites (Seyler et al., 2001). Furthermore, higher Na2O and TiO2 contents of rimming clinopyroxene were probably locally formed by the high-TiO2 melt for the formation of the later gabbro vein. Some spinels with high Cr# was also probably affected by the high-TiO2 melt. This would also explain the anomalously high cpx/(cpx+opx) ratio (Fig. 2⇑). Intrusion of the gabbro vein was fracture controlled rather than due to large scale permeable flow, the influence of the melt producing the gabbro vein is limited to the immediate vicinity of the vein. Since low range of the Cr# of spinel in the Enstatite-poor harzburgite is still higher than the others at our dive site, then, could reflect higher degrees of melting than elsewhere along the Atlantis II Transform. In conclusion, the Enstatite-poor harzburgite is interpreted to be a partial melting residue later infiltrated by a MORB-like melt and the TiO2-rich melt for the formation of the gabbro vein. Although little shallow melt transport is generally expected near transforms due to focusing of melt flow towards the mid-point of the adjoining magmatic segment (Dick, 1989), dunite is generally accepted as the product of melt-rock reaction stripping pyroxene from the host peridotite by ascending melt in a zone of focused flow (e.g., Dick, 1976a; Quick, 1981; Nicolas, 1989; Kelemen, 1990). Thus, it is clear that at least some melt was transported through the shallow mantle beneath the transform zone at the Atlantis II Fracture Zone.
Origin of the spinel micropods in slow-spreading ridge dunites
It should be emphasized that the spinel-rich layer consists of several spinel micropods within spinel-poor dunite matrix (Fig. 3⇑). Irrespective of its small size, the spinel-rich layer shares the same petrographical characteristics with podiform chromitites in ophiolites and orogenic peridotites as the product of late channelized melt transport and melt-rock reaction in the shallow mantle.
In contrast to the Enstatite-poor harzburgite, the spinel-poor dunite has little clinopyroxene and is nearly orthopyroxene-free. The boundary between the spinel-poor dunite and the Enstatite-poor harzburgite is sharp (Fig. 3⇑). Spinel TiO2 and Cr#, and discrete clinopyroxene TiO2 and Na2O contents in the dunite host are distinctively lower than that of the Enstatite-poor harzburgite (Fig. 5⇑–7⇑⇑), but very similar to that found in remaining peridotites. These characteristics are typical of residual peridotites that have undergone fractional melting. The last melt increments produced during such melting are ultra-depleted with very low TiO2, Na2O and other incompatible elements (Dick & Natland, 1996). Thus, if the spinel-rich layer and dunite were originally the product of melt rock reaction with a MORB melt, they were subsequently modified by re-equilibration with an infiltrating ultra-depleted melt at the end of mantle melting. Physical evidence of the latter can be seen in the highly depleted composition of the discrete clinopyroxene in the dunite which likely crystallized from this cryptic melt.
When primitive basaltic melts generated by partial melting of peridotite migrate upward, they selectively react with orthopyroxene in peridotite wall-rock to produce SiO2-rich secondary melts, resulting in the formation of a dunite (e.g., Dick, 1974; 1976; Kelemen, 1990). While it is clear that a late ultra-depleted high silica melt was present at the end of melting (Sobolev & Shimizu, 1993), and infiltrated the dunite as well, this cannot by itself explain the spinel-rich layer with its concordant contact with the host peridotite. Thus, the present composition of the spinel-rich layer reflects this event, but its crystallization could likely have been triggered by mixing of the ultra-depleted melt with a primitive MORB melt in a dunite conduit. The permeability of dunite is higher than that of the host peridotite (Toramaru & Fujii, 1986; Zhu & Hirth, 2003), and thus during late stage melt transport ultra-depleted melt present in the host peridotite at the end of melting will be drawn into the conduit to mix with whatever melt is passing through, and may later entirely re-equilibrate with olivine and spinel in the conduit – erasing the initial mineral compositions signature of the original melt migrating through the dunite that produced it. The hybridized melt is formed by the mixing of primitive basaltic melts traversing the dunite conduit with the SiO2-rich secondary melts drawn in from the host peridotite. This would likely draw the hybrid melt into the spinel stability field resulting in excess the local concern spinel precipitation and the local concentration of spinel in the micropods in this study, as suggested for the origin of chromitites (Irvine, 1975, 1977; Dick & Bullen, 1984; Arai & Yurimoto, 1994; Zhou et al., 1994).
Arai (1997a, b) emphasized that the chemistry of the host peridotites controls the formation of podiform chromitites, particularly high-Cr and/or large chromitites in ophiolite/orogenic peridotites. Arai & Abe (1995) and Arai (1997a, b) suggested that moderately refractory harzburgite with intermediate Cr#-spinel (Cr# = 0.4–0.6) is the most suitable host for large-scale podiform chromitites because orthopyroxene in lherzolite is too low in Cr# to concentrate chromite (as opposed to more aluminous spinel) due to meltperidotite interactions. In fact, even in lherzolite-dominant ophiolites and orogenic peridotites, podiform chromitite is usually found closely associated with harzburgite (Zhou et al., 1996; Morishita et al., 2006). In these cases, gradual lithological changes from dunites to lherzolites through harzburgite have been described around these chromitite pods (Zhou et al., 1996; Morishita et al., 2006). Given the similarity of the spinel in the dunite to that in the surrounding peridotites (excluding the Enstatite-poor harzburgite), we conclude that the studied spinel micropods were locally formed due to meltperidotite interactions in a region and/or time with a relatively high rate of magma supply near the transform.
Implications for petrogenesis of podiform chromitites
Podiform chromitites are ubiquitous in podiform and tabular dunites found in orogenic and ophiolitic peridotites. They occur in deposits up to many thousands of tons of massive ore, and range down in size to micro-deposits such as we find in our dunite 643R15. Podiform chromitites include both high-alumina and high-chromium ore deposits, and are significant source of both chrome ore and high-alumina spinel furnace refractories. The Cr# in our spinel-rich clots, however, is lower than the large majority of podiform chromitites in ophiolites and orogenic peridotites (0.19–0.8) (Dick & Bullen, 1984; Arai, 1997a, b), the very few oceanic chromitites reported to date (0.5–0.6) (Arai & Matsukage, 1998), and chromitite xenoliths beneath Japan Arc (0.6–0.8) (Arai, 1978; Arai & Abe, 1994), and is similar to the lowest end of the low-Cr chromitites of Robinson et al., (1997) (Fig. 6⇑).
The range of spinel Cr# in our residual lherzolite and harzburgite is similar to that from the Isabela ophiolite in Philippines (Andal et al., 2005) and overlap that of the Luobusa ophiolite in Tibet (Zhou et al., 1997): both lherzolite-dominated ophiolites (Fig. 6⇑). However, the Cr# of spinel in chromitites is much lower in our sample than in these chromitites from those ophiolites. These differences in spinel Cr# between them likely reflect the differences in the geochemical characteristics of melts responsible for the formation of chromitites.
Chrome generally has a very low solubility in basaltic melts, 330–520 ppm in MORB (Roeder & Reynolds, 1991). Water in melts may have a strong influence on increasing the solubility of Cr, however, because it will depolymerize the melt silica network (Edwards et al., 2000). Hydrous silicate minerals such as amphibole and phlogopite within chromite are common in chromitite (Talkington et al., 1986; Augé, 1987; McElduff & Stumpfl, 1991; Matsumoto et al., 1995; Ahamed et al., 2001, Ahamed & Arai, 2002; Morishita et al., 2006). We find no such direct evidence for hydrous conditions in the formation of our spinel micropods, which appear to be inclusion free.
Chromites in genetically related basalts and mantle peridotites have similar composition ranges (Dick & Bullen, 1984). Despite a data base well over a thousand samples, extremely few peridotites and basalts with chromite with Cr# > 0.6 are reported for in-situ ocean crust and mantle, whereas high-Cr# spinel ( > 0.7) is commonly found in arc-related magmas (Dick & Bullen, 1984; Arai, 1992, 1994a, Arai, b; Robinson, et al., 1997). The Cr# of spinel in the spinel micropods is similar to the lower range of those in mid-ocean ridge basalts collected from around the Atlantis II Transform (Fig. 6⇑). High-Cr magmas, probably related to subduction zone magmatism, then are likely required to form high-Cr chromitites. This study suggests that local concentration of spinel can be found in every geodynamic settings, but economically important large high-Cr chromite deposits appear to be restricted to supra-subduction zone settings.
We present the first detailed petrology and mineralogy of a local concentration of spinel from the slow-spreading Southwest Indian Ridge. Spinel is locally concentrated in small pods with centimeter scale in spinel-poor dunite associated with an enstatite-poor harzburgite host occurring in an outcrop area largely dominated by spinel lherzolite. Petrographical and mineralogical data suggest that the Enstatite-poor harzburige is a residue of mantle melting with subsequent crystallization of clinopyroxene from infiltrating interstitial melts. The Cr# of spinel in the spinel micropods is lower than in typical podiform chromitites from ophiolites and orogenic peridotites. Studies of REE and major element chemistry of MORB in the Atlantis II Fracture Zone region show that these were produced by relatively low degree of mantle melting (Robinson et al., 1996). The low Cr# of spinel in the spinel micropods in this study, which is similar to that in some MORB’s, then, likely reflects the geochemical characteristics of mid-ocean ridge basalts derived by this low-degree of partial melting. The spinel micropods were formed as a result of meltperidotite interactions. This study suggests that local concentration of spinel occurs in every geodynamic settings, but economically important chromite are likely restricted to supra-subduction zone where both moderately depleted peridotite and high-Cr magmas are expected to be coexisting.
We are grateful to Captain Ishida and the crew of the Yokosuka and the Shinkai Team who contributed to the success of the cruise. We also thank M. Cheadle, B. John, Y. Otomo, A. Kavassnes, G. Baines, A. Hamadate, M. Imamura, E. Miranda and J. Warren, for their help in collecting the data and discussions on board. T.M. deeply thanks S. Arai for his daily discussions on origin of chromitites. The manuscript was improved by valuable comments from Marja Lehtonen and three anonymous reviewers. This study is partly supported by a Grant-in-Aid for Scientific Research of the Ministry of Education, Culture, Sports, Science and Technology of Japan (No. 17740349) to T.M.
- Received 3 December 2006.
- Modified version received 28 August 2007.
- Accepted 3 September 2007.