The Effects of High ‐ Grade Metamorphism on Cr ‐ Spinel from the Archean Sittampundi Complex, South India

: We investigated the crystal and structural behavior of Cr ‐ bearing spinels from the Ar ‐ chean chromitites of Sittampundi (India), which had been subjected to very high ‐ grade metamor ‐ phism. The structural data show that their oxygen positional parameters are among the highest ever recorded for Cr ‐ bearing spinels with similar Cr# and Mg# and very similar to those found for other Archean occurrences. The general agreement between electron microprobe and Mössbauer data in ‐ dicates that the analyzed spinels are stoichiometric. It is therefore most likely that the P H2O and P total values as well as both the oxygen fugacity and the temperature reached during high ‐ grade meta ‐ morphism inhibited the possibility of the non ‐ stoichiometry of chromites, contrary to what can hap ‐ pen in ophiolites, where non ‐ stoichiometry has recently been documented.

Cr-spinel is useful as an indicator of the crystallization condition in a wide range of geological environments. It also records modifications induced during subsequent prograde or retrograde metamorphism of host rocks. The chemical durability of the chromite makes it useful as a petrogenetic indicator in ultramafic rocks in which metamorphic recrystallization has obliterated other primary fingerprints [8][9][10]. Changes in the composition and structure of chromite from ultramafic complexes during metamorphic modification have been studied by several researchers [11][12][13][14]. The relationships between compositional and structural parameters in natural Cr-spinels have been considered by several authors, providing a new approach to the understanding of mechanisms of spinel crystallization in magmatic and metamorphic environments [15][16][17][18][19][20][21][22][23]. The intracrystalline distribution of cations in spinels is a function of the thermodynamic parameters, including temperature, pressure and composition. Natural samples exhibit a non-equilibrium distribution of cations that depends on the thermodynamic parameters controlling the equilibrium order and the pressure-temperature-time path the crystals experienced. The ordering process involves short-range diffusion of cations, along with the breaking of bonds and a large kinetic barrier [24]. The equilibrium temperature corresponding to the observed order of a sample is called its closure temperature, TC [25]; TC provides insight on (1) a lower limit of the maximum temperature that the rock experienced and (2) a comparative cooling rate. Although the geothermometer modelled by [26] is mainly used for Mg-Al spinels, [27] found out that is suitable for Cr-rich compositions as well. The commonly used intercrystalline thermometers are those by [28,29], using olivine-Cr-spinel pairs, even if [30] showed that they do not always show an appropriate equilibration temperature.
In this study, we report the results of a structural-chemical investigation of chromite crystals extracted from the chromitite seams hosted by meta-anorthosite from the Sittampundi complex. We have taken up this work (a) to increase the database for Cr-spinels in chromitites of high-alumina continental layered intrusions, as crystal-chemical studies on spinels from layered chromitites are limited to a few places only, such as the Bushveld, Stillwater, Rum and Amsaga complexes [31][32][33][34] and Fiskenaesset [35] and (b) to verify whether relationships exist between the structural and compositional parameters of the spinels and a high grade of metamorphism of the chromitites.

Geology and Petrography
The Sittampundi complex belongs to the metamorphosed Archean layered anorthosite complex. The study area forms a part of the granulite terrane of south India. Major rock types are chromitite-bearing meta-anorthosite, amphibolite, basic granulite, two-pyroxene granulite, leptynite, biotite gneiss and pink granite. Subramanian [2] carried out a detail systematic study of the Sittampundi complex ( Figure 1). Sajeev et al. [6] reported retrogressed eclogite within the orthopyroxene free metagabbro, with the garnet of the rock hosting needles of rutile and omphacite. They suggested the assemblage as the peak eclogite facies condition of the rock. However, several authors have suggested that the high-grade metamorphism of the Sittampundi complex culminated at 11-12 Kbar pressure and 700-800 °C [37][38][39]. The peak metamorphic conditions were later overprinted by fluid influx and decompression-related retrograde a ssemblages. Although there is local preservation of primary igneous textures, almost all the primary minerals recrystallized during the high-grade metamorphism and obliterated the igneous textures and minerals. Felsic magma intruded the Sittampundi layered Complex at about 2.51 Ga, and, successively, the rock sequence underwent granulite facies (2450-2500 Ma) and amphibolite facies (500-720 Ma) metamorphism. The amphibolite facies metamorphism caused the modification of some chromites to Cr-poor green spinel [40]. Chromitite occurs exclusively within meta-anorthosite as discontinuous layered bands/lenses of varying thickness, with a maximum of 5 m within anorthosites and consisting of more than 60% chromite grains.
Earlier workers reported that the chromitites contain FeAl-rich chromites concentrated in layers between amphibole-rich layers, with a dominant mineralogy of amphibole − spinel − plagioclase ± sapphirine [7,36,41]. These authors suggested the existence of original highly calcic plagioclase (>An95), FeAl-rich chromite and magmatic amphibole, which are consistent with a derivation from a parental magma of hydrous tholeiitic composition.
Samples used for crystal-chemical investigation were culled from chromitites (chromite + rutile + calcic amphibole  anthophyllite  clinochlore). The samples studied were collected from conformable chromitite lenses within meta-anorthosite ( Figure 2a) on a foot track in the western part of Karungalpatti, Salem District, Tamil Nadu. According to [40], chromite compositions in Sittampundi show large variations in their chemical parameters, such as Cr/(Cr + Al) (Cr# = 0.26-0.60), Mg/(Mg + Fe 2+ ) (Mg# = 0.40-0.45) and Fe 3+ /(Fe 3+ + Al + Cr) (Fe 3+ # = 0.20-0.24), depending on textural type and nature of the associated phases, and occur within discontinuous bands within clinopyroxenite and highly calcic anorthosite. Previous Mössbauer studies on some chromites from Sittampundi, developed by [42], show that Fe 2+ and Fe 3+ occur at both tetrahedral and octahedral sites and that the Fe 3+ /ΣFe ratio is in the range 0.45-0.48. The oxygen positional parameters reported in that paper, calculated through chemical and Mössbauer analyses, are 0.2561 and 0.2603. While the first one is rather unreliable, being very similar to those of completely inverted spinels such as magnetite, the second value is rather close to that of 0.2607 recorded in an oxidized Cr-spinel recovered from a terra rossa soil [43]. Another study on Cr-spinel from Sittampundi is the one by [44], which, following those of [45][46][47] among the others, studied the Raman shifts of these samples.
Chromite occurs as granular aggregates in close association with amphiboles in the meta-anorthosite hosted chromitites. The rock shows a schistose texture, displaying the preferred orientation of amphiboles. Chromite, amphibole, chlorite and plagioclase form the major components. Chromite is also present as inclusions within the amphibole (Figure 2c). Chromite abundance is between 40% and 75% in the studied samples. The original igneous texture of the chromitite, such as chromite cumulate texture, igneous layering, etc., is preserved. The chromite grains are euhedral to subhedral (Figure 2c,d). Euhedral chromite grains are in common with well-defined grain boundaries and fracture zones. Some chromite grains of pseudohexagonal grain boundaries indicate their recrystallization ( Figure 2b).

Electron Microprobe Analysis
Ten to fifteen spot analyses were performed on the same Cr-spinels used for X-ray data collection, using a CAMECA-SX50 electron microprobe (EPMA) (CAMECA, Gennevilliers Cedex, France) at IGG-CNR (Istituto di Geoscienze e Georisorse-Consiglio Nazionale delle Ricerche,Padua, Italy), operating at 15 kV and 15 nA. A 20 s counting time was used for both peak and total background. Synthetic MgCr2O4 and FeCr2O4 spinels were used for Mg, Cr and Fe determination, Al2O3 for Al, MnTiO3 for Ti and Mn, NiO for Ni and sphalerite for Zn. The following diffracting crystals were used: TAP for Al and Mg, PET for Ti and Ti, and LIF for Cr, Fe, Mn, Ni, V and Zn. Raw data were reduced by a PAP-type correction software provided by CAMECA. The mineral chemical analyses are reported in Table 1.

Cation Distribution
According to [52,53], several different procedures may be adopted to determine cation distribution; satisfactory results can generally be obtained by combining data from single-crystal X-ray structural refinements and electron microprobe analyses. In fact, this approach simultaneously takes into account both the structural and chemical data, reproducing the observed parameters by optimizing cation distributions. Differences between measured and calculated parameters are minimized by a function F(x), taking in consideration different observed quantities such as a0, u, T-and M-m.a.n. (mean atomic number), atomic proportions, and constraints imposed by the crystal chemistry (total charges and T and M site populations). Several minimization cycles of F(x) were performed up to convergence. A summary of the procedure can be found in [54]. The obtained cation distributions are listed in Table 1a.

Mössbauer Spectroscopic Analysis
Optical examination of the samples show minute amounts of opaque phases (magnetite) at the grain boundaries, which could be magnetically separated after crushing each to a fine powder under acetone.
Clean chromite crystals were selected under an optical microscope. About 50 mg of each sample was used, distributed as 28 mg/cm 2 of sample in the sample holder, as calculated via RECOIL 1.04 computer software [55,56]. Room temperature Mössbauer measurements were carried out on clean chromite separates with a 25 mCi 57 Co source in a Rh matrix, which was driven at constant acceleration in a triangular mode. The Doppler energies from the 14.4 keV γ-rays were detected with a YAlO3:Ce scintillation counter. Mössbauer measurements were repeated on replicas of all samples to check the consistency of our measurements.
MossA (ver. 1.01a), a computer program developed by Prescher et al. [57], utilizing a full transmission integral with a normalized Lorentzian source line-shape, was used to fit the Lorentzian lines of the folded data. A fitting model consisting of three doublets assigned to Fe 2+ and one doublet assigned to Fe 3+ , as used by Lenaz et al. [58], was adopted to fit the spectra. In addition, to obtain accurate quantification of Fe 3+ /Fe for the chromite spectra, correction for the difference between the recoil-free fractions was applied to the chromite doublets for Fe 2+ and Fe 3+ at the tetrahedral and octahedral sites, respectively, at room temperature, based on previous studies [59][60][61][62], and calculated from the sample composition and the procedure enumerated.
For accurate quantification of Fe 3+ /Fe for the chromite spectra, we used the following unified equation, which incorporates the recoil free fractions, 0.687 and 0.887, for the observed Fe 2+ and Fe 3+ , respectively: where R is the (Fe 3+ /Fe 2+ ) ratio of the observed Fe-cations in the chromite doublets.
Mӧssbauer parameters obtained for centroid shift δ, quadrupole splitting, ΔEQ and the full width at half-maximum Γ, together with the averages of (Fe 3+ /∑Fe) ratios for the samples and their respective replicas, are shown in Table 2. The identification of criteria for Fe-cation sites is based on the relative centroid shift -values, following previous reports by [62][63][64]: The site occupancy was determined from the area ratios of the doublets. Table 2. Mössbauer parameters and corrected (Fe 3+ /ΣFe) ratios obtained from (a) measurements of samples and (b) repeated measurements of three replica samples, respectively. δ = centroid shift (mm/s), ΔEQ = quadrupole splitting (mm/s), Γ = full width at half-maximum (mm/s). The averages of (Fe 3+ /∑Fe) ratios for the samples and their respective replicas are reported in this study.
The structural parameters considered are the cell edges that range between 8.2364 (4) and 8.2477 (4) Å and the oxygen positional parameter, u, between 0.2633 (1) and 0.2637 (1). This last parameter is related to the oxygen packing distortion within the lattice. The ideal cubic-close packed structure shows u = 0.25, but it is observed that u is higher than 0.25 for all four samples of Cr-bearing spinels. The observed distortion is a consequence of similar M-O and T-O bond distances (u = 0.2625 when distances are equal) [65]. Previous studies suggested that 0.2625 and, consequently, the T-O and M-O bond distances are related to the cooling history experienced by the mineral. Slow cooling allows the cations to accommodate at their preferred site, i.e., trivalent cations at the octahedral M site and divalent cations at the tetrahedral T site. On the contrary, rapid cooling causes a disordered cation distribution between the two sites, with trivalent and divalent cations in both T and M sites. As can be seen in Tables 1 and 2, the values found for these Cr-spinels are among the highest recorded u values, thereby pointing to very slow cooling rates. The degree of inversion, i, i.e., the amounts of trivalent cations at T site or divalent cations at M site, is in the range of 0.066-0.092 atoms per formula unit. Figure 3 shows spectra of samples 5B and 30B_rep, as typical examples of Mössbauer spectra, for a four-doublet fitting model, which provided a good  2 -value for the fitting of the Sittampundi chromite spectra. As the spectra show limited resolution, which is typical for this type of chromite, the assignment of individual doublets to specific sites is uncertain. However, the Fe 3+ /ΣFe ratios obtained are considerably more reliable, as the Fe 3+ absorption is well constrained by the asymmetry of the two main high-and low-velocity components in the measured spectra, where the Fe 3+ contribution occurs entirely within the low-velocity component.  57 Fe Mössbauer spectra obtained at room temperature for samples 5B and 30B rep. Fitted absorption doublets assigned to Fe 2+ and Fe 3+ are indicated in blue and red colors, respectively. Diamond symbols denote measured spectra, and the black curves represent summed fitted spectra. Table 2 shows the room temperature Mössbauer data for (a) all samples and (b) the repeated runs on replicas. The calculated average of their respective corrected (Fe 3+ /Fe) ratios in the Tables are, 0.13, 0.17 and 0.10, respectively.
According to microprobe analyses, structural refinement and Mössbauer analyses data, the values of (Fe 3+ /ΣFe) ratios in the Sittampundi chromite-spinels are in the range 10-18%. However, on the basis of the values of the quadrupole splittings of the Fe 3+ doublets, it appears all Fe 3+ ions preferentially occupy the T-site.

Discussion
In terms of mineralogy and field relations, the Sittampundi chromitites are remarkably similar to anorthosite-hosted chromitites in the Neoarchean Fiskenaesset anorthositic complex, Greenland. In Figure 4, we compare the structural parameters of the studied Crspinels with those of other occurrences that are similar in age. In fact, there are some other Archean occurrences, such as the Fiskenaesset and Ujragassuit (Greenland) [35], Amsaga (Mauritania) [34], Inyala and Rhonda (Limpopo belt, Zimbabwe) [35], detrital Cr-spinels from Banavara [66], as well as the Bushveld and Stillwater Complexes [31,33]. Moreover, the figure comprises the Cr-spinels in seams entrapped within anorthosites from the Paleogene Rum Layered Complex [32].
In Amsaga, the chromites can be found associated with different matrix minerals, according to the metamorphic grade that affected the rocks. Those with the higher u values are metamorphosed in amphibolite facies and are very close to the parameters exhibited by Sittampundi Cr-spinels. Fiskenaesset chromitites, even if their paragenesis also includes pargasitic amphiboles, as in Sittampundi, have to be considered as primary and not metamorphosed [67,68]. In order to estimate the intracrystalline closure temperatures of the Cr-spinels, we applied to our samples the geothermometer of [26]. This geothermometer takes into account the Mg and Al content of the Cr-spinels and their distribution amongst the octahedral and tetrahedral sites. The presence of other cations is accounted for by coefficients present in the equation of the geothermometer. The calculated intracrystalline closure temperature (Tc) for the Sittampundi Cr-spinels is in the range of 715-965 °C. The computed intracrystalline closure temperature is more or less in agreement with the temperature obtained by topological constraints and quantitative geothermometry of the area [39,41]. In Figure 5, we compare these temperatures with those of Rum, Fiskenaesset and Amsaga complexes. It is possible to see that those from Rum are the lowest (521-653 °C), while the others are more or less comparable. Black circles: Cr-spinels from Sittampundi (this study); red circles: detrital Cr-spinels from Banavara [66]; grey circles, squares and star: Cr-spinels from Amsaga (Mauritania) in amphibole, chlorite and talc-serpentine matrix, respectively [34]; yellow, purple and green circles: Cr-spinels from Fiskenaesset and Ujragassuit (Greenland) and Zimbabwe, respectively [35]; blue squares: Cr-spinels from Rum (Scotland) [32]; green field: Cr-spinels from Bushveld and Stillwater complexes [31,33]; yellow field: Cr-spinels from Nuggihalli komatiites [69].
It is interesting to note that Amsaga Cr-spinels record an intracrystalline closure temperature higher than that of the amphibolite facies; thus, it is possible they still retain the primary closure temperature after crystallization. Sittampundi Cr-spinels underwent a granulite facies metamorphism. Sunder Raju [70] noticed that the pristine composition of the chromites is preserved despite intense metamorphic and tectonic processes. In this case, the highest temperatures are comparable with those of the metamorphism, but they extend down to about 700 °C. It is pertinent to remember that the intracrystalline closure temperature is the temperature recording the last Mg-Al exchange between T and M sites and not the temperature of crystallization. Thus, they suffered a high temperature metamorphism and then slowly cooled until their actual Tc was achieved.
In the past few years, some papers have dealt with the possibility of non-stoichiometry in Cr-spinels from different occurrences. This was determined using a combination of techniques, including structural refinement, electron microprobe and Mössbauer analysis. The authors [18,71,72] found some Cr-spinels from ophiolites to be non-stoichiometric, i.e., showing structural vacancies in the lattice due to the oxidation of Fe 2+ to Fe 3+ . Furthermore, some authors [73,74] also documented the existence of minor vacancies in Cr-spinels from the Bushveld layered complex and some Antarctica mantle xenoliths, respectively. However, when the same combined approach was used for other ophiolites (Shetland, UK, [75]; Leka, Norway, [76]) and the Amsaga complex in Mauritania ( [34], no vacancies were detected. Using both the conventional-and point-57 Co-source methods, [58] showed how, in some cases, there can be large discrepancies in the (Fe 3+ /ΣFe) ratios calculated via structural refinement, electron microprobe and Mössbauer analyses. The reason, for example, is that the larger samples used in conventional Mössbauer techniques include several different grains, which could have a range of very different degrees of oxidation. Here, we can see that the values of Fe 3+ /ΣFe obtained for the Sittampundi Crspinels using various experimental techniques are very similar, thus allowing us to conclude that these samples of spinel are stoichiometric. Figure 5. Range of intracrystalline closure temperatures (°C) for the spinel composition investigated and other occurrences of Cr-spinels from chromitites in anorthosites used for comparison (Fiskenaesset, unpublished data; Amsaga [34]; Rum [32]).
Although chemical modifications related to sub-solidus re-equilibration and metamorphic hydrothermal processes can significantly influence the primary high-T composition of Cr-spinel, the composition of chromite in chromitite bodies can still be used as a petrogenetic and geotectonic indicator [77]. Suita and Streider [78] showed that the core of Cr-spinel grains in massive chromitites conserves the primary chemical composition, regardless of the metamorphic changes. Notably, the degree of metamorphic re-equilibration in metamorphosed chrome-spinels depends on the ratio of PH2O and Ptotal. According to [79], when PH2O is about 25% of Ptotal, a complete metamorphic re-equilibration occurs; however, where it is lower, relict igneous cumulate textures are preserved. In fact, [80] suggested that in high pressure conditions, there is no resetting of the primary igneous composition of chromites. The composition of mantle fluid phases attending magmatic and metasomatic processes strongly depends on oxygen fugacity (fO2). So, in the C-H-O system, the fluid phase consists mainly of CO2 and H2O in relatively oxidizing conditions close to the fayalite-magnetite-quartz (FMQ) solid buffer, H2O and CH4 in moderately oxidizing conditions between the FMQ and iron-wüstite (IW) buffers, and CH4 below IW [81]. For Albanian ophiolites, [82] found that the pre-oxidation Δlog(fO2)FMQ values range from -3.2 to +2.3 log units, with most values close to or higher than the FMQ buffer, while the post-oxidation Δlog(fO2)FMQ values are systematically the highest, ranging from +0.5 to +2.9 log units. A report by [62] showed that there are two groups of spinels in ophiolites from Oman, with different oxygen fugacities. The calculated oxygen fugacities for the low Fe 3+ /ΣFe ratio group are between 1.9 and 2.7 log units above the FMQ buffer, while samples from the high Fe 3+ /ΣFe ratio group yield higher oxygen fugacities, between 3.2 and 3.4 log units above the FMQ buffer. It is therefore most likely that the PH2O and Ptotal values, as well as both the oxygen fugacity and the temperature reached during different processes, inhibits the possibility of non-stoichiometry of chromites from amphibolite-granulite-eclogite facies, as in Amsaga and Sittampundi, contrary to what can happen in ophiolites.

Conclusions
According to the combined results from structural refinement, electron microprobe and Mössbauer analyses of the Sittampundi Cr-spinels, the following can be concluded: (i) Oxygen parameters of the studied Cr-spinels are close to the Amsaga chromites. (ii) The computed intracrystalline closure temperature of the Cr-spinels is comparable to the reported temperatures obtained by topological constraints and quantitative geothermometry. (iii) Estimated values of Fe 3+ /Fe ratios obtained by using different techniques are similar and establish that the studied Sittampundi Cr-spinels are stoichiometric. (iv) Structural refinement study of Cr-spinels could be important in discriminating the various tectonic domains.

Conflicts of Interest:
The authors declare no conflict of interest.