The ophiolite complex in the island of Cyprus, known as the Troodos Massif, lies in the Tethyan ophiolite zone. It is one of the least deformed and most extensively studied ophiolites. The complex has an annular outcrop pattern, with the basal part exposed at the central region (Fig. 5.6). The pattern indicates an overall domal structure, which is believed to be the result of late Tertiary differential uplift and subsequent erosion (Gass 1980). The complex has an aggregate thickness of about 6-7 km consisting of a typical ophiolite sequence (Fig. 5.7) that includes both mantle tectonite (harzburgite) and crustal layered cumulates, separated by a transitional dunite zone. The lower part of the sequence comprises two crosscutting plutonic suites, which are believed to have been derived from different magma sources. The older suite, which includes tectonized harzburgite and dunite with chromite, probably formed as a large solid-state mantle diapir; the younger suite is less deformed, consists of ultramafic and mafic rocks, and displays cumulate layering . Based on the age of the overlying sediments and sparse K- Ar dates, the age of the Troodos ophiolite is considered to be Upper Cretaceous (pre- Maestrichtian, most likely Campanian) (Malpas & Robinson 1987).
The tectonic environment of the Troodos ophiolite has been discussed extensively in the literature. The sheeted dike complex strongly suggests a spreading center and for a long time a mid-ocean ridge setting was the generally accepted model (Coleman 1977). Such a setting, however, cannot be reconciled with the fact that the Troodos ophiolite contains two distinct magma suites: an early, relatively high-Ti02 basaltic suite encompassing the upper gabbros, the sheeted dike complex, and the andesitic to rhyodacitic lower pillow lavas (LPL); and a late, low-Ti02 basaltic andesite suite constituting the lower ultramafic to gabbroic cumulates and the upper pillow lavas (UPL). Studies of isotope and trace-element geochemistry of the rock package suggest a subduction-related spreading-center setting for the ophiolite emplacement (Pearce 1975, 1980, Gass 1980, McCulloch & Cameron 1983, Moores et al. 1984, Thy & Moores 1988). Synthesizing the available data, Thy & Moores (1988) proposed that the Troodos ophiolite probably developed in a short-lived spreading basin tectonically and petrologically in an arc position above a subducting oceanic crust but without the development of a mature arc, a setting analogous to the present-day plate-tectonic configuration in the Andaman Sea of the eastern Indian Ocean as suggested by Moores et al. (1984). Other Mideast ophiolite complexes may also have formed in a similar manner along short spreading segments separated by transform faults above a subduction zone.
Chromite mineralization in the complex ranges from accessory chromite as disseminations (<5% chromite) to pods of massive chromitite (>90% chromite), all believed to be of magmatic origin and subjected to variable degrees of deformation. Accessory chromite is ubiquitous in all rocks of ultramafic composition, but most chromite segregations occur either in the dunite close to the harzburgite contact (64% of deposits) or in the tectonite harzburgite, within 1 km beneath the lowermost cumulates, as pods and lenses enclosed in dunite (33% of deposits); the dunite cumulates contain only a small fraction (3%) of the chromite deposits (Greenbaum 1977, Malpas & Robinson 1987). The chromite bodies show all gradations between cumulate textures (polygonal, net, occluded silicate, nodular and orbicular textures) and deformation textures (schlieren and pull-apart textures).
Chromite has been mined in the Troodos area from a number of prospects since 1924, but no mines are currently in production. Most of the production in recent years came from the Chrome Mine situated near the summit of Mt. Olympus, where massive podiform ore occurs in dunite, close to the harzburgite-dunite contact. More than 0.5 million tonnes of ore was mined from this deposit; typical production-grade ore after concentration contained 47% Cr203 with a Cr:Fe ratio of about 2.7 (Greenbarum 1977).
Chromite (FeCr204) belongs to the spinel group of minerals. Natural chromite (chrome spinel) is a solid solution having the general formula (Mg,Fe2+)(Cr,Al,Fe3+)204 and its composition may be represented adequately within a six-component triangular prism and three of its faces as shown in Figure 5.8. The face corresponding to Mg/(Mg+Fe2+) versus Cr/(Cr+Al) is particularly important because most chromian spinel compositions plot close to that face. The Mg:(Mg+Fe2+) ratio of chromite (referred to as magnesium ratio) is largely temperature dependent and, considered in conjunction with the Mg:(Mg+Fe2+) ratio of coexisting olivine or pyroxene, is useful for estimating the temperature of magmatic crystallization or metamorphic reequilibration (Irvine 1967, Jackson 1969, Evans & Frost 1975, Henry & Medarisl980). The variation in the Cr:(Cr+Al) ratio of chromites (referred to as chrome ratio) is due to Cr-Al substituion and it is controlled partly by variation in total pressure of crystallization. The chrome ratio is also affected by partitioning of Al3+ into into cumulus Al-silicate phases, which causes a relative increase of Cr in the residual melt. The other binary graph, Fe3+/(Cr+Al+Fe3+) versus Mg/(Mg+Fe2+), incorporates the variation in Fe3+:Fe2+ ratio of chromian spinels and provides an approximate indication of the oxygen fugacity (Irvine 1965, 1967), a controlling parameter for magmatic crystallization of chromite (Osborn 1959, Hill & Roeder 1974, Ulmer 1969). Available experimental data and thermodynamic formulations bearing on compositional variations in natural chromites have been discussed by Irvine (1965, 1967) and Haggerty (1991).
As has been discussed by many authors (Thayer 1964, Irvine 1967, Duke 1983, Dickey 1975, Leblanc et al. 1980, Stowe 1994), despite some overlap, the compositional trends of stratiform chromites are quite distinct from those of the podiform chromites (see Table 5.4). Stratiform deposits are characterized by a large variation in the magnesium ratio (=0.2-0.70) relative to the chrome ratio (=0.6-0.8) (Fig. 5.9), low Cr:Fe(total) ratio (<2.5), generally high Fe3+:Fe2+ ratio (up to about 1), and Ti content that may reach up to 1%. Stratigraphic variations generally follow fractionation trends, but may be complicated because of fluctuations in temperature, total pressure, fO2, or co-crystallization of Fe-Mg-Al silicate minerals. The general upward decrease in the magnesium ratio and a slight upward decrease in the chrome ratio are usually attributed to decreasing Mg2+ and Cr in the residual melt during fractional crystallization. The Cr:Fe(total) ratio also generally decreases upward because of Cr depletion and Fe increase in the residual melt. For example, in the Bushveld Complex, the Upper Group chromites have a distinctly lower Cr:Fe(total) ratio compared with the Lower Group chromites (Table 5.3, Fig. 5.10). An overall upward decrease in the Cr.Fe(total) ratio of chromite seams (up to 3.9; Prendergast 1987) is also present in the Great Dyke, but in the Stillwater Complex the Cr:Fe(total) ratio of chromitite layers first increases upward and then decreases toward the top (Jackson 1968).
The composition of chromites in the ultramafic cumulate section of ophiolites is broadly similar to that of stratiform chromites. The contrasting compositional features of podiform chromites in the mantle tectonite include a larger range of chrome ratio (=0.2-0.9) for a relatively smaller variation in the magnesium ratio (=0.4-0.7) (Fig. 5.9), higher Cr:Fe(total) ratio (=2.4-4.6), lower Fe3+:Fe2+ ratio (commonly <0.5), and
PODIFORM [Ultramfic Cumulate]
STRATIFORM 1 (Layered | Complexes] I
< 0.5 i
^ 0.4 &
0.2 0.3 0.4 0.5 0.6 0.7 Mg/(Mg + Fe2+)
Figure 5.9. Compositional fields of stratiform and podiform chromites on the Mg/(Mg+Fe2+) versus Cr/(Cr+Al) projection of the spinel compositional prism. Stratiform deposits (layered complexes): Bushveld, Great Dyke, and Stillwater. Podiform deposits (ophiolites): Cyprus (Troodos), Oman (Semail), New Caledonia (Tidbaghi and Massif du Sud), Canada (Thetford), and Pakistan (Zhob Valley). Data for chromites in Archean layered complexes and Archean greenstone belts are excluded for the sake of clarity. (Source of data: compilation and lower Ti (<0.1%). There is a distinct inverse relationship between Cr and Al, but in contrast to the stratiform deposits the relationship is independent of the fractionation trend and so cannot be ascribed to Cr depletion in the residual melt fraction. Two additional features of podiform chromite composition deserve mention. One is the wide variation of the chrome and Cr.Fe(total) ratios even in individual deposits. For example, the chrome and Cr/Fe(total) ratios in podiform chromites hosted by mantle tectonite range from 0.1 to 0.8 and from 0.8 to 3.9, respectively, in the Semail (Oman) ophiolite (Brown 1980, Auge 1987) and from 0.2 to 0.8 and from 1.4 to 4.6, respectively, in the New Caledonia ophiolite (Leblanc et al. 1980, Leblanc 1987). The other is an apparent bimodal distribution of the chrome ratio in the tectonite-hosted chromites (Leblanc & Violette 1983), with Al-rich chromite pods (chrome ratio typically between 0.4 and 0.6) distributed along the harzburgite-dunite transition zone not far from the base of the ultramafic cumulates and the Cr-rich pods (chrome ratio typically >0.6) located in deeper peridotites. As will be discussed later, the compositional variation provides important clues to the origin of podiform chromites.
Investigations of the geochemistry of platinum-group elements in chromitites have shown that podiform deposits hosted in Paleozoic and Mesozoic ophiolite complexes have distinctly different chondrite-normalized PGE profiles compared with those hosted by layered complexes (Fig. 5.11). A reason for the marked enrichment of stratiform
chromitites in Pt and Pd is the higher original abundance of PGE-enriched sulfides in these chromitites (Naldrett & von Gruenewaldt 1989); a contributing factor may be the difference in relative proportions of PGE in the source magmas for layered intrusions and ophiolites. Chromitites of the Sukinda and Nausahi districts in Orissa (India), interpreted earlier as stratiform deposits on the basis of chemical composition and texture (Chakraborty & Chakraborty 1984), have been reinterpreted as ophiolite-hosted podiform deposits based on their PGE profiles (Page et al. 1985).
5.4.4. STRATIFORM DEPOSITS
From the textures of stratiform chromite deposits and their association with ultramafic cumulates, it is quite clear that the chromite-rich layers represent segregation of chromite crystals that crystallized from a basaltic magma. It is also evident that, because of the very low solubility of chromium in basaltic magmas (e.g., a maximum of =1,000 ppm Cr in a magma containing 13% MgO; Barnes 1986), the accumulation of Bushveld-type chromitite layers must have involved the processing of tremendous volumes of magma. The features which have remained controversial and need to be addressed are: (a) the formation of chromitite layers (in which chromite is the only cumulus phase); (b) the great lateral extent of chromitite layers in some deposits; and (c) the presence of many such layers in a given layered intrusion. A good summary of the various hypotheses has been presented by Duke (1983).
The liquidus phase relations in the (Mg,Fe)2Si04-Si02-(Mg,Fe)Cr204 system (Fig. 5.12a) provide a reasonable representation of chromite crystallization from basaltic magmas at low pressures. The abundance of olivine cumulates at the lowermost parts of layered intrusions suggests that the parent magma compositions should plot in the olivine liquidus field, such as point a in Figure 5.12a. Fractional crystallization of olivine from such a magma drives the liquid composition away from the olivine comer toward the olivine-chromite cotectic line. Crystallization of chromite begins at point b and co-precipitation of olivine and chromite continues as the liquid composition moves along the cotectic. Crystallization of both olivine and chromite terminates at point c, a reaction or distribution point, and orthopyroxene becomes the sole crystalline phase as the liquid composition moves along cd. The sequence of cumulus minerals for this crystallization path (a => b => c => d) is: olivine, olivine + chromite, orthopyroxene. Note that chromite becomes an early-crystallizing phase despite the very small amount of dissolved chromite in the parent magma, but only a small proportion of chromite (about 2 modal%) co-precipitates with olivine (Irvine 1965). The olivinexhromite ratio crystallizing at any point on the cotectic line is given by the intersection of the tangent to the cotectic line with the olivine-chromite join.
The crystallization sequence described above explains the commonly observed chromite disseminations in ultramafic cumulates, but not the formation of chromitite layers. Settling of chromite crystals may be possible when chromite crystallizes alone from a magma supersaturated in chromium, but effective gravitative separation of the very small amount of chromite from a cotectic chromite-silicate mixture is considered unlikely. To obtain monomineralic chromite layers, some perturbation in the system must lead to an interval when chromite is the sole crystallizing phase. Postulated mechanisms by which chromite-silicate cotectic crystallization may be supplanted by crystallization of chromite unaccompanied by silicate minerals include: (a) change in magma composition by silica assimilation or mixing of magmas, (b) increase in the oxygen fugacity, and (c) increase in the total pressure of crystallization.
Ulmer (1969) and Cameron and Desborough (1969) proposed that the “chromitic intervals” in the Critical Zone of the Bushveld Layered Series formed due to periodic increase in the oxygen fugacity of the magma. Experimental investigations (Ulmer, 1969, Hill & Roeder 1974) do suggest a significant expansion of the liquidus chromite field with increasing oxygen fugacity. However, because of the internal buffering capacity of a large body of magma, it is difficult to visualize the rapid but spatially uniform fluctuations in oxygen fugacity implied by the repetition of chromitite layers of great continuity. Subsequently, Cameron (1980) abandoned this hypothesis in favor of chromitite formation in response to tectonically induced changes in total pressure in the magma chamber. An expansion of chromite stability field with increasing pressure should be expected, as has been demonstrated by Osborn (1978) for Mg-Fe-Al spinel in a simplified basaltic system, and a pressure increase also increases the stability of pyroxene relative to plagioclase (Lipin 1993). Moreover, variations in total pressure due to, for example, tectonic activity or addition or withdrawal of large batches of magma, are likely to be laterally uniform. However, the magnitude of pressure change required to shift the spinel-silicates cotectic surface sufficiently so as to form a 1 m- thick chromitite layer can be qualitatively estimated to be unrealistically large, and the direct effect of pressure change on mineralogy, in general, has been shown to be trivial (Hatton and von Gruenewaldt 1987). In a chamber as large as the Bushveld Complex the roof could not have been rigid, but merely floating on the magma. Hence, mechanisms which would increase the pressure at the base of the chamber are difficult to envisage (Cawthorn & Lee 1998).
Irvine (1975) proposed that chromitite layers in the layered intrusions were formed on occasions when the mafic parental magma of the intrusion was extremely contaminated with granitic material derived from sialic roof rocks. The principle of this hypothesis is illustrated in Figure 5.12b. The curvature of the olivine-chromite cotectic is such that addition of Si02 to a liquid on the cotectic, such as at point e, drives the liquid composition to some point / inside the liquidus chromite field, resulting in the crystallization of chromite alone. Chromite continues to be the only crystallizing phase until the liquid composition reaches the cotectic at point g, at which point the normal crystallization path is resumed. Variations of this model with different amounts or episodes of silica assimilation could account for the formation of the more common stratigraphic sequences involving chromitite layers in the Muskox, Stillwater, Great Dyke, and Bushveld complexes (see Irvine 1975, Figs. 10 and 11). Alapieti et al. (1989) have concluded that the silica contamination hypothesis offers the best explanation for the main chromitite layers in the Kemi Complex, where evidence for silica contamination is found in the form of small silicate inclusions in chromite grains. The inclusions are rich in alkalis and apparently represent trapped droplets of the contaminant sialic melt. As the sulfur solubility in a mafic magma decreases with an increase in its silica activity, the silica contamination hypothesis offers a possible explanation also for the occurrence of sulfide deposits in some mafic-ultramafic complexes (see Ch. 6).
Although sound in principle, the above mechanism would require geologically improbable amounts of silica contamination to account for the Bushveld chromitites. Moreover, the marked Fe-enrichment trend of differentiation in the Bushveld Complex suggests a tholeiitic melt relatively uncontaminated by alkali-rich granitic melt. The alternative hypothesis proposed by Irvine (1977) is based on the same phase relations, but calls for influxes of a more primitive magma at different stages of fractionation, actually an extension of the scheme believed to have been responsible for the formation of cyclic units. For example, as illustrated in Figure 5.12c, the mixture of a more evolved magma of composition d with a less evolved magma of composition c would result in a magma composition represented by a point such as h in the liquidus field of chromite (provided that points on the mixing line lie above the liquidus surface), inducing the crystallization of chromite alone. Murck and Campbell (1986) have shown that the solubility of Cr in a basalt magma in equilibrium with chromite decreases more rapidly per unit fall in temperature at higher temperatures (=1,400°C) than at lower temperatures (=1,200°C). Because of the resulting concave-upward curvature of the Cr solubility curve, the mixing of two magmas, both saturated (or nearly saturated) in chromite but at different temperatures, would place the hybrid above the saturation curve, suggesting that point h in Figure 5.12c would likely lie above the liquidus. In the case of the Bushveld Complex, such mixing adequately explains the chromitite layers (LG-1 to LG-4) associated with olivine cumulates, but the thicker layers associated with orthopyroxene only (LG-5 to LG-7) or with orthopyroxene and plagioclase (MG-2 to MG-4 and UG-1, UG-2) are better explained by mixing of two quite distinct magmas (Hatton & von Gruenewaldt 1987).
As mentioned earlier, the crystallization of the Rustenburg Layered Suite involved the mixing of derivatives of an ultramafic (U) magma and an anorthositic (A) magma. The U liquid is believed to have been the main carrier of Cr, the Cr content decreasing rapidly with decreasing normative olivine content of the liquid. From melting experiments on Bushveld chilled margin rocks, Sharpe and Irvine (1983) showed that each of these melts at its liquidus would be saturated in chromite and a silicate mineral
5.12.olivine in the case of U melt and plagioclase in the case of A melt — and yield cumulates containing only accessory amounts of chromite, but hybrids of the two would have the capability to form chromitite by virtue of being saturated only in the oxide mineral. Chromite would remain the sole liquidus phase over almost the entire region of intermediate composition, even for relatively low f02 conditions under which the U melt has only olivine on the liquidus and the A melt only plagioclase (Irvine & Sharpe 1986).
A problem with the magma-mixing hypothesis discussed above is that the temperature interval over which chromite is the sole crystallizing phase is small (=20°C), after which a silicate mineral begins to crystallize; this should dilute the chromite to accessory concentrations or even terminate its crystallization through a liquidus reaction relationship. Experiments with (7, and A, melt compositions by Sharpe and Irvine (1983) indicated that the hybrid melts would yield 0.02-0.04 wt% chromite by this process, or a little more (=0.05 wt%) with a more primitive U liquid
certainly not enough to produce a Bushveld-type chromitite layer. For example, to form a 1 m-thick chromitite layer like LG-6 by this process would require a layer of parent liquid of equivalent lateral extent and at least 1 km thick! To circumvent this geologically unrealistic situation, Todd et al. (1982) and Irvine et al. (1983) proposed a “double-diffusive convection magma mixing model” for the solidification of large layered intrusions such as the Bushveld and the Stillwater. The essence of this model is that the U and A liquids differ sufficiently in density to form separate layers in the magma chamber and their crystallization and mixing are controlled by double-diffusive convection. Whole sequences of cumulates form concurrently by down-dip accretion from a column of density-stratified liquid layers that are separated primarily by diffusive interfaces. Each cumulate layer accumulates from one or more liquid layers and the lower cumulates, because of their higher crystallization temperatures, grow in advance of the upper cumulates. Thus, chromitite layers grow (prograde) continuously by the down-dip accretion process, a little bit at a time and at particular levels on the basin- form intrusion floor, while silicate cumulate layers are prograding concurrently from
liquid layers of appropriate compositions at other levels (Irvine & Sharpe 1986).
Campell et al (1983) have discussed the application of hydrodynamic theory for modeling the mixing process when a fresh primitive silicate magma is introduced into a magma chamber that is stratified by composition and density. The form and efficiency of mixing depends on the density contrast between the two magmas and on the velocity and volume of the input magma. If the input magma is more dense than the lowest magma layer in the chamber, the initial momentum of the input magma may carry it in the form a turbulent fountain, mixing and entraining with the resident magma, but eventually it will fall back forming a dense layer at the top of the cumulate pile. If the resident magma is denser, then the input magma is likely to rise as a turbulent plume, entraining and mixing with a high proportion of the resident magma on the way, and the hybrid, when it reaches its own density level, will spread out as a turbulently convecting layer. Barnes and Naldrett (1985) have estimated that the density of a mafic magma should decrease by the fractional crystallization of olivine, remain about the same by the crystallization of orthopyroxene, but tend to increase once plagioclase becomes a liquidus phase. Thus, a turbulent plume is most likely to develop if the influx of primitive magma is preceded by significant plagioclase crystallization in the magma chamber. The Bushveld Upper Group chromitites probably formed in such an environment that was conducive to efficient mixing; the Lower Group chromitites of the Bushveld Complex and those of the Stillwater Complex appear to have formed before the onset of plagioclase crystallization and, therefore, in an environment of less efficient mixing (Naldrett & von Gruenewaldt 1989). Other problems with the magma mixing model include the systematic vertical variation in chromite composition, the lateral uniformity of the chromitite layers, and the absence of cryptic variations.
Despite uncertainties in the details of the mechanism, mixing of magmas of different composition and density is currently the favored model for explaining the igneous stratigraphy of large layered complexes as well as for the formation of stratiform chromitite layers hosted by them. Magma mixing is also a popular model for the PGE-enriched sulfide mineralization in layered complexes (see Ch. 7).