Growth of zircon during metamorphism is described in a variety of lithologies and can occur at different metamorphic grades, from high- (e.g., Fraser et al., 1997, 2004; Degeling et al., 2001; Möller et al., 2003; Wu et al., 2006; Harley et al., 2007; Zhao et al., 2015) to low-temperature metamorphism, including low-temperature hydrothermal reactions (e.g., Dempster et al., 2004, 2008; Rasmussen, 2005; Hay and Dempster, 2009; Hay et al., 2010; Kohn et al., 2015). Growth and new precipitation of zircon has been traced under temperatures as low as 250 ∘C (Rasmussen, 2005). Metamorphic zircon can (a) precipitate from fluid or melt (e.g., Rasmussen, 2005; Kohn et al., 2015), (b) can result from breakdown of Zr-bearing phases (major or accessory) (e.g., Davidson and van Breemen, 1988; Fraser et al., 1997, 2004; Degeling et al., 2001), or (c) can exsolve from Zr-bearing phases (e.g., Bingen et al., 2001; Tomkins et al., 2007). a.

New metamorphic zircon precipitates during exhumation and cooling of the host rock (Rasmussen, 2005; Kohn et al., 2015) as a result of partial or complete dissolution of preexisting zircon (e.g., Dempster et al., 2008) in partial melt or metamorphic fluid during high-temperature metamorphism (Ewing et al., 2014). Solubility of zircon in most natural fluids is very low (Tromans, 2006); thus, zircon dissolution models mostly describe interactions with the melt (e.g., Harrison and Watson, 1983; Watson and Harrison, 1983). However, under high (prograde) temperatures and in fluids of favorable composition, zircon can be dissolved without melt involvement (e.g., Rubatto et al., 2008; Hay and Dempster, 2009; Hay et al., 2010; Ewing et al., 2014; Kohn et al., 2015).

In this contribution, we investigate unusual zircon textures, such as coronae found in dehydrated metapelitic granulites of the IVZ. We start with a review of the process of zircon precipitation from various Zr-bearing phases, followed by an overview of known examples of zircon coronae. After a short geological background of the unit, we describe the sampled outcrop as well as the sample itself macroscopically. Then a short exposition of applied methods and microscopic description of the studied sample are presented, followed by a detailed depiction of observed zircon microstructures and textures. For the sake of completion, we also include microprobe data of the studied sample. In the discussion, we suggest mineral reactions that could result in the formation of observed zircon coronae textures and then discuss the implications of our findings.

Zircon dissolution and growth during metamorphism are not independent processes but must be coupled with the breakdown–growth of other phases and/or with various mineral–fluid reactions in the host rock (e.g., Tomkins et al., 2007; Austrheim et al., 2008). Metamorphic precipitation of zircon could be a result of the breakdown of and/or exsolution from various Zr-bearing phases (Davidson and van Breemen, 1988) such as garnet, amphibole, (clino)pyroxene and ilmenite (e.g., Fraser et al., 1997; Degeling et al., 2001; Möller et al., 2003; Söderlund et al., 2004; Harley et al., 2007; Kelsey et al., 2008; Morisset and Scoates, 2008), hemo-ilmenite (Morisset et al., 2005), baddeleyite (Bingen et al., 2001; Söderlund et al., 2004), rutile (Harley et al., 2007; Tomkins et al., 2007; Morisset and Scoates, 2008; Kelsey and Powell, 2011; Ewing et al., 2013, 2014; Pape et al., 2016), epidote, titanite (Kohn et al., 2015), chlorite (Fraser et al., 2004), and biotite (Vavra et al., 1996). Zircon coronae have been reported around Martian baddeleyite as a result of shock metamorphism (Moser et al., 2013). In mafic metamorphic rocks, precipitation of zircon is commonly associated with the Fe–Ti oxides (Bingen et al., 2001; Ewing et al., 2013, 2014) due to similar chemical properties of Zr and Ti.

Zirconium and titanium both belong to group 4 in the periodic table, have close chemical properties and are usually regarded as relatively immobile trace elements (e.g., Mohamed and Hassanen, 1996). In the group of incompatible cations, Zr and Ti belong to high field strength (HFS) elements, which are smaller and are highly charged compared with large ion lithophile (LIL) elements. The chemical similarities result in a positive correlation between Zr and Ti for most rock suites and in their ability to replace each other in oxides (e.g., Morisset et al., 2005). The fact that Zr oxides and Ti oxides are spatially related in many rocks confirms chemical similarities between Zr and Ti.

Thus, rutile and ilmenite are the main minerals interpreted to influence the Zr mass balance in metabasites in the absence of other Zr phases (e.g., Ferry and Watson, 2007; Tomkins et al., 2007; Morisset and Scoates, 2008; Ewing et al., 2013). In the absence of zircon, rutile can be the main phase holding Zr and Hf in the absence of zircon (Ewing et al., 2014). Zirconium is a common component of rutile, in which its content can reach 10 000 ppm (e.g., Ewing et al., 2013); thus, Zr distributions generally reflect the formation and decomposition of rutile. The temperature dependence of Zr solubility in rutile can have a fundamental impact on the zircon growth rate (Kohn et al., 2015) and controls zircon stability (Kelsey and Powell, 2011). The zirconium-in-rutile thermometer for the rutile–quartz–zircon system was calibrated by a number of authors (e.g., Watson et al., 2006; Ferry and Watson, 2007; Tomkins et al., 2007; Lucassen et al., 2010; Ewing et al., 2013), who have shown a large temperature- and pressure-dependent solubility of Zr in rutile. Zircon growth is frequently associated with the oxide transition from Zr-rich rutile to ilmenite during late-stage exhumation and cooling under a large variety of P-T conditions (Ewing et al., 2013). Magmatic and metamorphic ilmenite can also contain significant amounts of Zr (Bingen et al., 2001; Morisset et al., 2005; Charlier et al., 2007), up to more than 500 ppm (e.g., Morisset and Scoates, 2008). Consistently, many authors describe zircon precipitation on ilmenite (e.g., Bingen et al., 2001; Austrheim et al., 2008; Morisset and Scoates, 2008) (see below).

In igneous rocks zircon usually forms euhedral elongated single crystals that are shaped by a combination of prismatic and pyramidal faces, whereas metamorphic zircon is characterized by roundish or irregular shapes (Corfu et al., 2003). Rarely, zircon has unusual saccharoidal or needle-shaped morphology or forms coronae (Corfu et al., 2003 and references therein). Mineral–fluid interactions, decomposition of Zr-bearing minerals, and exsolution from Zr-bearing accessory and rock-forming minerals can result in such unusual zircon textures (e.g., Corfu et al., 2003 and references therein; Dempster et al., 2004, 2008; Rasmussen, 2005), even at low metamorphic grades (e.g., Dempster et al., 2008).

In natural samples, there are several documented examples of zircon coronae textures from igneous and metaigneous rocks (e.g., Bingen et al., 2001; Söderlund et al., 2004; Austrheim et al., 2008), as well as from metapelites of the IVZ (Pape et al., 2016). Such textures are found in rocks of different metamorphic grades, ranging from prehnite-pumpellyite to eclogite facies (Austrheim et al., 2008). It has been suggested that coronae textures may evolve in magmatic rocks as a result of slow cooling (Morisset et al., 2005) and in metamorphic rocks due to mineral–fluid reactions or exsolution with fluid-aided diffusion along grain boundaries during progressive metamorphism (e.g., Bingen et al., 2001).

One of the first descriptions of zircon coronae in mafic metaigneous rocks was done by Söderlund et al. (2004). The authors attributed formation of secondary fine-grained zircon to the breakdown of baddeleyite in the presence of silica (saccharoidal zircon) and to consumption of minerals that have trace amounts of Zr, such as ilmenite (coronitic zircon). Both of these textural types of secondary zircon precipitated under prograde heating. Bingen et al. (2001) reported hat-shaped zircon grains and coronae around ilmenite in granulites and amphibolites. Charlier et al. (2007) and Austrheim et al. (2008) reported fine-grained zircon chains around, but at a distance from ilmenite and rutile grains in metagabbros. These authors suggested that zircon chains had grown around primary Fe–Ti oxides and, therefore, trace the former grain boundaries. Fine-grained zircon was reported to frame some rutile grains in the metapelitic septae from the IVZ (Ewing et al., 2013; Pape et al., 2016). Morisset and Scoates (2008) reported 1–100 µm thick zircon coronae around ilmenite in mafic plutonic rocks. They consider it to be a result of Zr diffusion from ilmenite during slow cooling, aided by hydrothermal fluid.

In this study, we report the two textural types of zircon coronae, characterized by various thickness and aspect ratio, occurring within Fe–Ti oxides in granulitic metapelites. We present evidence that these textures formed as a product of the breakdown of Fe–Ti oxides, which helps to understand the initial mineral paragenesis of the host rock and reveals former reaction fronts.

The generally restitic mineralogy of the sample is composed of garnet, biotite and sillimanite with minor amounts of cordierite, ilmenite, rutile, K-feldspar and quartz (Fig. 1c–f). The primary mineralogy indicates prograde–peak mineral paragenesis, which consists of biotite, sillimanite and garnet. The foliation is formed by a fabric of elongated garnet and sillimanite crystals 0.5–1 mm in length that compose 80–90 % of the sample (Fig. 1c–d). The stretching lineation is formed by elongated biotite crystals. Biotite contains numerous micrometer-sized apatite needles and is mostly replaced by chlorite. Primary metamorphic fabric is crosscut by several generations of veins and/or fractures (Fig. 2c–f), which were formed during cataclastic deformation and shear zone development (e.g., Siegesmund et al., 2008). Fractures are filled with post-peak and late hydration mineral assemblages. The earlier generation of veins is mostly composed of Fe–Ti oxides and their intergrowths with quartz (Fig. 1c–d, black material). Fe–Ti oxides form aggregates with lobate boundaries with the primary minerals (garnet and sillimanite) (Fig. 1e–f). The network of veins of the later generation cross-cuts the veins of the earlier generation or follows their contacts. These later veins are more abundant than earlier ones and are composed of fine-grained phyllosilicates, such as chlorite, muscovite and/or phengite, and may also contain K-feldspar patches in the vein cores (Fig. 1c–d, grey-brown material; Table 1). Abundance of phyllosilicates indicates post-metamorphic hydration reactions. Large (up to 2 mm in length and 0.3 mm thick) elongate quartz aggregates generally follow the vein distribution (Fig. 1d, bottom part). Veins and fractures form a conjugated orthogonal network, stretching in at least two directions in a 2-D section.

Accessory minerals are zircon and monazite (e.g., Figs. 1f, 2). Where hosted by garnet, zircon forms roundish elongated crystals with aspect ratios from 1 : 1 to 1 : 3 and lengths from 30 to 100 µm (Fig. 2a). Where forming intergrowths with sillimanite, zircon reveals well-developed faces and forms triple junctions with the adjacent sillimanite grains (Fig. 2b), reflecting equilibration growth with sillimanite during prograde and peak metamorphism. Where hosted by fine-grained material that fills fractures, zircon crystals are elongated, with an aspect ratio from 1 : 2 to 1 : 3; these grains are fractured and fragmented. The fragments have irregular dissolved boundaries and show evidence of crystal-plastic deformation (Fig. 2c). Vermicular- and hat-shaped zircon aggregates and zircon coronae are spatially associated with each other and occur within ilmenite–rutile–quartz or rutile–quartz clusters and/or intergrowths (Figs. 3–5), which fill transgranular fractures and pockets in the gneiss (e.g., Fig. 4a).

Zircon textures reported in this study are coronae, by which we mean thin envelopes or shells in 3-D. Accordingly, in the 2-D plane of a sample they have thread- or worm-like shapes (depending on the thickness and aspect ratio). Zircon coronae in our sample occur as two main textural types. The first type is referred to as vermicular-shaped (coarser-grained) aggregates, which have a thickness ≥ 5 µm and an aspect ratio of 1 : 4 to 1 : 20. The second type is referred to as coronae (finer-grained) zircon aggregates, which have a thickness ≤ 1 µm and an aspect ratio of approximately 1 : 100 (e.g., Figs. 3–4). There is also a third (subordinate) coronae type: hat-shaped aggregates that are the result of zircon coronae overgrowth preexisting (probably detrital) grains.

Mineral electron microprobe data are presented in Tables 1 and 2. These data are used to determine the temperature of metamorphism using the garnet–biotite thermometer and to support the suggestion of possible mineral–fluid reactions. The XFe of garnet is systematically lower in the cores than in the rims and in the smaller fragments. The same applies to the biotite (Table 1). Garnet rims are also systematically enriched in Mn, compared to the cores. Compositions of Fe–Ti oxides (Table 2) demonstrate that rutile is much higher in SiO2 content than ilmenite. Rutile is also slightly enriched in such elements as Al, Cr and Nb, but lower in Mn (Table 2) compared to ilmenite.

Mineral textures in the studied sample provide important information about the reactions that could have enhanced the growth of zircon coronae during metamorphism. We suggest the following reaction sequence: (1) formation of peak metamorphic phases and partial melting of the metapelites; (2) initial resorption of peak metamorphic phases and crystallization of interstitial ilmenite with lobate boundaries in fractures; (3) retrograde metamorphism, further resorption and fracturing of the high-temperature phases, hydration reactions with formation of phyllosilicates and decomposition of ilmenite to rutile; and (4) further cooling and recrystallization of rutile.

Temperature estimations were done using the garnet–biotite thermometer using microprobe data (Table 1). Various calibrations of this geothermometer (Thompson, 1976; Holdaway and Lee, 1977; Ferry and Spear, 1978; Hodges and Spear, 1982; Perchuk and Lavrent'eva, 1983; Bhattacharya et al., 1992) gave temperatures for garnet–biotite preserved cores of 570–700 ∘C, for inner rims 800–860 ∘C, and for outer rims 820–1090 ∘C. Estimations were done for pressures of 0.7 and 1.0 GPa. Pressure variations did not have any significant effect on the resulting temperatures. It is, however, possible, that garnet and biotite rims were affected by diffusion from the host environment during retrograde metamorphism. Mineral textures (Figs. 1e–f, 2c, 4a, 5) and microprobe analyses (Table 1) indicate that the initial granulite-facies garnet, biotite and sillimanite were intensely altered and resorbed. The fragments of garnet and sillimanite have ragged edges and are plastically deformed, dissolved and altered. Mn and Fe, enriched in the rims of large garnet grains and in small garnet fragments, suggest garnet resorption and hydration during retrograde metamorphism (e.g., Tuccillo et al., 1990) after peak metamorphism. Thus, garnet rims do not indicate peak metamorphic temperatures; therefore, the rim temperatures are erroneous. More likely, peak metamorphic temperatures in this IVZ section were between 700 and 860 ∘C (temperatures obtained for the cores and inner rims), in agreement with previous estimations (Sills, 1984; lowermost estimations of Luvizotto and Zack, 2009).

Pockets and fractures in garnet and sillimanite are filled with Fe–Ti oxide aggregates with lobate boundaries. Thus, ilmenite probably crystallized from the partial melt and/or fluid after the formation of peak metamorphic phases and after their resorption was initiated. Occasionally, veins with Fe–Ti oxides are associated with the quartz aggregates; thus, quartz possibly formed in the early generation of veins together with ilmenite. Ubiquitous fracturing of the rock (e.g., Fig. 1c–d) and crystal-plastic deformation of zircon (Figs. 2c, 4f) indicate the extreme conditions of post-peak metamorphism deformation (e.g., Siegesmund et al., 2008).

Further retrograde (greenschist facies) metamorphism led to hydration reactions and formation of veins filled with phyllosilicates and K-feldspar. The following features are regarded as evidence of intensive mineral–fluid reactions in the dry restitic granulite-facies rock (e.g., Rajesh et al., 2013): reaction rims around fragments of granulite-facies minerals (e.g., Fig. 1e), fine-grained phyllosilicate mixture that fills fractures (e.g., Fig. 1c–d, f), quartz veins (Figs. 1d, 4a), alteration of biotite with chloritization and exsolution of apatite needles. Water-rich fluids could have been sourced from the decomposing biotite (e.g., Pape et al., 2016). Rare ilmenite cores are surrounded by rutile rims (Figs. 1e–f, 5a–b, d). Thus, post-peak, trace-element-rich ilmenite was partially or entirely decomposed to rutile, which resulted in the migration of excess Fe into the matrix and into the garnet and biotite rims. Fe from ilmenite and Mg diffusing out of garnet and biotite rims are needed to compensate for the formation of the large volume of Mg–Fe phyllosilicates in the second generation of veins (Figs. 1c–d, 2c; Table 1). K from biotite and Al from sillimanite would allow and/or favor the growth of K-feldspar in the veins (e.g., Fig. 4a). Excess Hf and Zr from breaking down ilmenite are responsible for the formation of zircon intergrowths with rutile (thick vermicular-shaped grains, e.g., Figs. 3b, c, e, 4a–c). Newly formed rutile is enriched in trace elements, possibly due to the decreased volume of Fe–Ti oxides (e.g., Austrheim et al., 2008). This rutile is also enriched in silica (Table 2) that requires sourcing SiO2 from the environment and may indicate solid solution of SiO2 in rutile (Taylor-Jones and Powell, 2015), which would play a role in a further reaction. Excess Si, possibly derived from the fragmentation and dissolution of garnet and sillimanite, would form quartz veins and react with Zr to form zircon coronae. As for the apatite formation, P, F, Cl and OH could be derived either from decomposed biotite or were delivered by the water-rich fluid as components of a water brine from, for example, dissolution of monazite. However, the occurrence of apatite needles inside altered biotite grains points to genetic relationships between these two minerals.

Further cooling caused nonequilibrium recrystallization of rutile. Ewing et al. (2013) described partial replacement of rutile by other phases, characteristic for all granulitic metapelites from the IVZ. In our sample, we observe recrystallization of rutile with the formation of fine rutile–quartz intergrowths and thin zircon coronae around them. This took place during later stages of the rock evolution, when the temperature decreased and caused the Zr solubility in rutile to decrease. Therefore, we suggest the following stylized reaction (R1): RutZr-Hf+SiO2=Rut2+Qtz+Zrncor, in which RutZr-Hf (= SiO2-, Zr- and Hf-bearing rutile) resulted from decomposition of ilmenite and Rut2 (= SiO2-, Zr- and Hf- depleted rutile) forms intergrowth with quartz (Qtz) and zircon coronae (Zrncor). Quartz was exsolved from SiO2-rich rutile and formed thin intergrowths with the newly crystallized, trace-element-depleted rutile (e.g., Figs. 3, 4c). Zircon coronae produced in this way are very thin, suggesting slow reaction rates and a limited solubility of Zr in rutile at a low temperature (Tomkins et al., 2007; Ewing et al., 2013, 2014; Pape et al., 2016).

Zircon grains hosted by garnet, sillimanite and fine-grained phyllosilicate matrix (Fig. 2a–c accordingly) represent detrital grains, enclosed within the main mineral phases during metamorphism. Zircon in garnet shows euhedral shapes and concentric growth zoning (Fig. 2a), indicating capture of detrital grains by metamorphic garnet. Zircon enclosed in sillimanite has detrital cores that are overgrown by metamorphic rims, which are in growth equilibrium with sillimanite (Figs. 2b, 5c). After the peak metamorphic conditions these detrital zircon grains seem to have been mostly inert and are therefore well-preserved. Zircon grains hosted by the fine-grained phyllosilicate matrix in hydration veins are the most deformed and fractured and show dissolved and/or corroded surfaces (Fig. 2c). These latter grains were probably exposed to the post-peak metamorphic fluids. The dissolved material from their surfaces might have been transported with a fluid and serve as a source for the zircon coronae precipitation. We suggest, however, that this was not the main source of Zr for coronae zircon and that coronae mainly precipitated from Fe–Ti oxides.

Vermicular-shaped aggregates of zircon (or thick coronae) and thin coronae (Figs. 3–5) have a different origin than the detrital grains (Fig. 2). It has been shown that zircon can grow from other Zr-bearing phases as a result of mineral reactions and as a mineral response to the changing conditions (Bingen et al., 2001; Söderlund et al., 2004; Austrheim et al., 2008; Ewing et al., 2013, 2014; Kohn et al., 2015; Pape et al., 2016). Metamorphic (coronae) zircon in granulite-facies rocks may not be a product of peak metamorphism, but precipitate during the retrograde evolution (Tomkins et al., 2007). Zircon coronae textures are the evidence of zircon formation due to breakdown of Zr-bearing Fe–Ti oxides (e.g., Davidson and van Breemen, 1988; Fraser et al., 2004; Degeling et al., 2001). However, taking into account complexity of the textures and the fact that they were formed in more than one stage, we do not entirely exclude the possibility of exsolution of zircon from, for example, Zr-bearing rutile or exsolution from ilmenite before its breakdown (e.g., Bingen et al., 2001; Tomkins et al., 2007; Ewing et al., 2013; Pape et al., 2016). The possibility of the two-stage exsolution of Zr from Fe–Ti oxides was suggested by Ewing et al., (2013). Our textural observations are consistent with this idea. The sketch in Fig. 6 shows stages (1) and (2) of zircon coronae formation:

After the peak metamorphic conditions, ilmenite was the main host phase for Zr, together with the primary detrital zircon (Bingen et al., 2001). At the initial cooling stage it partially decomposed to rutile (Ewing et al., 2013). The expelled Zr was not entirely incorporated into the growing rutile and precipitated as new zircon (Fig. 6). Formation of zircon vermicular aggregates preceded brittle and ductile deformation of the rock. Vermicular grains in 3-D volume represent curved envelope-type aggregates (Fig. 4b–e), thus resembling coronae in shape (e.g., Bingen et al., 2001). However, they are thicker than what was previously observed for zircon. Therefore, we interpret vermicular grains as evolved coronae. The thickness of these coronae should be controlled by reaction and cooling rates (Kohn et al., 2015). At comparatively high temperatures and slow reaction rates, zircon coronae grew thick, and formed lamellae-like intergrowths with the newly forming rutile (e.g., Fig. 4a–b). Formation of similar exsolution lamellae was described for many metamorphic minerals (e.g., Zhang and Liou, 2000).

Zircon coronae in our sample are different from those described in Bingen et al. (2001), Charlier et al. (2007), and Morisset and Scoates (2008), who only observed coronae at the boundary of the (former) ilmenite grains. The described textures are also different from the coronae reported by Austrheim et al. (2008) and Pape et al. (2016), in which zircon forms continuous chains or closed contours of small grains. However, zircon coronae in all cases (described in earlier literature and here) represent 3-D shells and/or envelopes around the reacting grains (Bingen et al., 2001). Textures, indicating reaction fronts of rutile recrystallization, have not been found by Pape et al. (2016), even though these authors searched these features. In contrast, in our sample we observe former reaction fronts formed by tangled and split zircon coronae within recrystallized rutile aggregates. Split coronae may show different reaction fronts converging to one point (Figs. 3e, 4e). The reaction fronts moved from rutile–quartz intergrowths towards unreacted rutile, forming rutile–quartz embayments in the latter, rimmed by zircon coronae (e.g., Fig. 4c, arrows show the directions of the reaction front). The chains of small zircon grains are effectively the same as zircon coronae and are similar to those described in Austrheim et al. (2008). The hat-shaped zircon grains are formed by coronae that are connected to the larger zircon grains (Fig. 5a) and thus represent aggregates formed by different zircon generations.

Not all rutile aggregates in our sample are associated with zircon coronae. Similarly, the diversity in appearance of rutile grains from the same sample was described by Pape et al. (2016) for IVZ metapelites. This can be due to (a) thin section cut that does not reveal associated coronae or (b) only local recrystallization of rutile (e.g., due to locally elevated strain or inhomogeneous distribution of fluid), so that the rest of the rutile still contains a significant amount of Zr. In the case of (b), Zr-in-rutile thermobarometry can be applied to the Zr-enriched rutile to estimate the temperature of ilmenite decomposition and coeval formation of vermicular zircon (e.g., Ewing et al., 2013).