Garnet growth in a geological blink of an eye

The timing and rates of metamorphic reactions, and the pressure (P) and temperature (T) conditions they represent, are central to constraining the evolution of tectonic processes. These reactions also control the release of fluids, which are linked to various processes of societal relevance, such as the transfer of precious metals and the occurrence of intermediate-depth seismic events (e.g., Hacker et al., 2003; Viete et al., 2018; Giuntoli et al., 2024). Interpretation of metamorphic processes and their rates rely on the translation of chemical and isotopic changes in rocks into P-T conditions and their time (t) of occurrence. Garnet is a Rosetta Stone in this regard: stable and ubiquitous throughout the lithosphere, garnet tends to grow as P-T conditions increase and provides a unique record of the reactions and chemical changes that affect metamorphic rocks during their tectonic cycle. Garnet is generally considered to grow over millions of years, slowly recording changes in P-T conditions and reactive mineral assemblages. However, whether this is actually the norm is not well known.

Zoned garnet dating using Rb-Sr or Sm-Nd chronology indicates that garnet growth may indeed take millions of years, comparable with time scales of tectonic processes (e.g., Christensen et al., 1989; Vance and O'nions, 1990; Pollington and Baxter, 2010; Dragovic et al., 2015) but also, when precision allows, significantly faster growth has been identified (e.g., Dragovic et al., 2012, 2015; Tual et al., 2022; Farrell et al., 2024), consistent with diffusion modeling (Viete et al., 2018; Gaidies and George, 2021). These time scale variabilities could be caused by overstepping and kinetic factors related to nucleation and element supply, which are known key parameters in garnet growth (Wilbur and Ague, 2006; Carlson et al., 2015; Konrad-Schmolke et al., 2023). Garnet growth may occur outside equilibrium conditions, with its zoning reflecting rapid chemical shifts during sudden completion of overdue reactions, rather than significant changes in P-T conditions. Many examples illustrate such phenomena, demonstrating the key role of kinetically controlled processes in the evolution of garnet-bearing metamorphic assemblages and its significance in accurately constraining geological processes (Austrheim, 1987; Pattison et al., 2011; Spear, 2017; Carlson et al., 2015; Spear and Wolfe, 2019; Nagurney et al., 2021).

The emergence of trace-element mapping in garnet has significantly increased our ability to read its compositional record and interpret the underlying geological processes, as they are generally robust to diffusional re-equilibration and sensitive to subtle geochemical changes occurring during metamorphism (e.g., Raimondo et al., 2017; George et al., 2018; Konrad-Schmolke et al., 2023; Smit et al., 2024).

Zoned garnet chronology (e.g., Pollington and Baxter, 2010) provides an unrivaled opportunity to constrain “time-resolved” processes from garnet grains, but common micro-sampling methods by Micromill® required for high-precision garnet chronology do not provide the necessary high spatial resolution to match that achieved with trace-elements mapping in garnet. Micro-sampling using laser cutting and high-resolution Lu-Hf analysis (Tual et al., 2022) allow such combined high-precision garnet dating of individual growth zones. In this study, we use this approach together with major- and trace-element mapping to tease apart the growth history of a typical Barrovian amphibolite-facies mica schist. The analyses show that cm-garnet in this rock nucleated out of equilibrium and grew in short fluid-boosted bursts that lasted a hundred thousand years or less.

The sample analyzed in this study (HG01) is a garnet carbonate-rich mica schist from the Bunte Randserie, which is situated close to the Austrian-Italian border (Fig. 1). The area hosts the Schneeberg Fault zone (SFZ), the tectonic boundary between the Schneeberg Complex to the southeast and Ötztal-Stubai Complex to the northwest. The tectonic significance of the SFZ has been interpreted as either (1) an Eo-Alpine greenschist to upper-amphibolite facies crustal shear zone that accommodated the exhumation and/or burial of the Schneeberg and the eclogite-bearing Texel Complexes, or (2) a reactivated Variscan contact between the Texel-Schneeberg and Ötztal-Stubai Complex that allowed the final tectonic juxtaposition of the various units at ca. 80 Ma (e.g., Klug and Froitzheim, 2022; Montemagni et al., 2023). The SFZ underwent protracted metamorphism between 105 and 95 Ma with peak metamorphism reaching 550–600 °C and 0.8–1.0 GPa (Konzett and Hoinkes, 1996; Sölva et al., 2005). The analyzed sample (HG01) comprises cm-sized euhedral porphyroblasts of garnet in a largely recrystallized, fine-grained matrix of ankerite-dolomite, chlorite, muscovite, plagioclase, biotite, calcite, ilmenite, rutile, and zircon. These matrix minerals also occur as (polyphase) inclusions in garnet (Figs. 2A–2C and S1; Table S1 in the Supplemental Material¹). Millimeter-sized kyanite, rutile, and rare remnants of staurolite occur at a high angle to the fine-grained matrix (Figs. S1C–S1E).

A 1.2-cm-diameter garnet grain was extracted from the sample, mounted in epoxy, and cut in two halves through its geometric center (Figs. 2A and 2B). Both halves (HG01A and HG01B) were characterized through in situ major element analysis by electron probe micro-analyzer (EPMA) and X-ray major-element mapping and the distribution of minor- and trace-elements mapped by laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS). An ~1.5-mm-thick wafer was cut from each garnet half, and five chemically distinct zones were cut out from one of these wafers using an Oxford A-Series micromachining system fitted with a nanosecond-pulse diode-pumped solid-state laser (λ = 355 nm) (method of Tual et al., 2022). Each zone was halved to allow for two analyses per growth zone (17–50 mg/aliquot). Details of the analytical methods are provided in Text S1.

Garnet zones 1–4 (Figs. 2D, S2, and S3) show a bell-shaped Mn content (Xsps_6→3), with associated increase in pyrope (Xpy_8→17), decrease in almandine (Xalm_74→69), and relatively constant grossular (Xgrs_13→15; Figs. 2F and S2C–S2E). The core (zone 1) is enriched in heavy rare earth elements (HREE) and Y (Figs. 2E and S3) and shows µm-scale oscillations in HREE, particularly conspicuous in Yb_N/Dy_N. The HREE concentrations broadly decrease throughout zone 2, are even, but consistent, in zone 3, and increase in zone 4. Zone 5 shows a 1%–2% increase in Xalm and Xgrs and a decrease in Xpy and Xsps compared to zone 4, a sharp decrease in Y (Y_{200→50 µg/g}) and HREE (e.g., Yb_{40→5 µg/g}; Figs. 2D, 2E, and S3), and a slight increase in Ti (Ti_{250→400 µg/g}; Fig. 2E). Trails of micro-inclusions of zircon are homogeneously distributed across garnet (Fig. 2E).

Garnet contains less than 0.1 ppm Hf (Table 1), and Lu concentrations are highest in zone 1 (~30 ppm) and lowest in zones 2 and 5 (~0.7 ppm). The whole-rock analyses, done by both high-pressure autoclave digestion and table-top dissolution to monitor potential Hf contribution from inherited zircon inclusions, differ in their Hf concentration but have identical isotope compositions. The Lu-Hf analyses of individual garnet zones (n = 10) and the whole-rock (n = 2) together yield an isochron regression with an age of 98.3 ± 0.2 Ma (Fig. 3; Table 1). The mean square of weighted deviates (MSWD) of 2.0 indicates residual scatter, which is caused by a resolvable age difference between zones 1–4 and zone 5 (Figs. 3B–3D). Separate regressions provided 98.4 ± 0.1 Ma for zones 1–4 (n = 10; MSWD = 0.45; Figs. 3B and 3C) and 97.8 ± 0.3 Ma for zone 5 (n = 4; MSWD = 0.72; Fig. 3D).

The spatially resolved, high-precision age data obtained reveal that the duration of garnet growth was remarkably short (0.6 ± 0.4 m.y.; Fig. 3A). Within this time frame, two even shorter growth pulses occurred. The first of these produced ~85% of the total garnet volume and occurred at 98.4 ± 0.1 Ma, with the uncertainty limit of this weighted mean age setting a maximum of 200 k.y. on the total process. The second growth pulse is represented by zone 5, which shows no significant Lu zoning but is chemically distinct from the rest of the grain, yielding a resolvably younger age of 97.8 ± 0.3 Ma. Garnet growth associated with zones 1–4 and zone 5 was separated by a possible short hiatus of 0.6 ± 0.4 m.y.

The presence of continuous inclusion trails and Mn zoning across zones 1–4 is consistent with Rayleigh fractionation during a continuous prograde growth (Hollister, 1966); µm- (zone 1) to mm-scale (zone 1–4) HREE oscillations indicate, however, that a parallel process was operating. Such fluctuations may be (1) thermodynamically controlled through oscillations in P–T (e.g., Viete et al., 2018) or phase breakdown (e.g., Raimondo et al., 2017), or (2) kinematically controlled by variation in garnet growth rate (e.g., George et al., 2018) or matrix diffusion limitations (Skora et al., 2006), typically imposed by fluid availability and the matrix permeability (e.g., Konrad-Schmolke et al., 2023; Kulhánek and Faryad, 2023). Here, three lines of evidence indicate that these euhedral, harmonic µm-scale fluctuations are likely caused by the latter: (1) the continuous inclusion trails are consistent with fast-growing garnet and preclude significant changes in growth rate (Fig. 2E); (2) sharp peak correlation among HREE across garnet implies an even supply of REE and no mass fractionation (Figs. 2 and S4); and (3) the rapid garnet growth constrained here rules out oscillations in P–T, as this would require nearly instantaneous, recurrent rock-wide (re-)equilibration. Ca-bearing phases are abundant in the matrix; however, given that Ca in garnet broadly correlates only with the mm-scale oscillations and remains constant overall, garnet growth may reflect Ca-free reactant phases, e.g., chlorite, with no evidence for aCO₂ changes, which could affect scales of equilibrium (e.g., Carlson et al., 2015).

Whether garnet growth was thermodynamically or kinetically controlled, the total garnet growth was extremely fast. These time scales are much shorter than those required for tectonic forces to drive the P–T changes that are generally required to drive widespread reaction and may only apply if fluids effectively catalyzed rapid T equilibration (e.g., Beinlich et al., 2020). The growth of zones 1–4, and perhaps even zone 5, must then have occurred under (near-)isobaric and isothermal conditions and provides only a short snapshot of the Eo-Alpine tectonic history of the region.

The formation of garnet in discrete pulses of ultrafast growth points to reaction overstepping (e.g., Spear, 2017). The rate of nucleation and growth depends on the degree of temperature overstepping, the equilibrium volume and the absolute temperature of the system, with overstepping at high temperature resulting in faster crystal nucleation and growth due to more efficient element transport to the nucleation site (e.g., Ridley and Thompson, 1986; Waters and Lovegrove, 2002; Kelly et al., 2013; Nagurney et al., 2021). Significantly overstepped reactions are common in metamorphic rocks, and to some extent necessary for garnet growth reactions (e.g., Wilbur and Ague, 2006; Pattison et al., 2011), and high growth rates of 1–2 mm m.y.⁻¹ in rocks lithologically comparable to HG01 have been attributed to this phenomenon (Gaidies and George, 2021). Relatively fast garnet growth has been observed in other zoned garnet chronology studies (e.g., Dragovic et al., 2012; Tual et al., 2022; Farrell et al., 2024). These studies, however, focus on garnet from subduction zones, where T is low and reactions are commonly kinetically limited and directly associated with brief intervals of fluid-rock interaction. The data obtained here provide one of the first direct measurements of garnet growth rates in a typical Barrovian mica schist (see also Pollington and Baxter, 2010), perhaps the most commonly used rock for determining P-T histories of tectonic terranes. The growth rates we determine for garnet are also extremely high. Garnet in zones 1–4 formed at an average rate of at least ~4.9 cm³ m.y.⁻¹ and at an average radial growth rate of ~6.2 cm m.y.⁻¹ or higher. Even when corrected for grain size, these minimum rates are estimated at least twice as high as those estimated through geospeedometry (Gaidies and George, 2021), but consistent with similar rates of ~10 cm³ m.y.⁻¹ (Pollington and Baxter, 2010). The growth rates also provide constraints on the pace at which pelitic rocks release their fluids. Garnet in HG01 likely formed through the continuous reaction: 3 Chl + 1 Ms + 3 Qtz → 4 Grt + 1 Bt + 12 H₂O (e.g., Pattison et al., 2011). At the rates calculated for zones 1–4 and representative end-member compositions, this reaction would produce 1.6 g H₂O m.y.⁻¹ grain⁻¹. At the observed garnet modal abundance of about one grain per every two cubic centimeters, that would indicate that the garnet-forming reaction allowed the production of H₂O at a rate of 0.81 g H₂O m.y.⁻¹ cm⁻³ of rock during the short bursts when this reaction was ongoing. That is significantly higher than previous estimates of 0.003–0.012 g H₂O m.y.⁻¹ cm⁻³ (Dragovic et al., 2018). This volume of H₂O far exceeds the pore space of any metamorphic rock and thus would instigate pulsed, but overall, very fast high-flux fluid escape following recurring pore fluid pressure build-up (e.g., Konrad-Schmolke et al., 2023; see Text S1). The migration of fluids within the middle and lower crust thus may be a direct consequence of the critical overstepping and pulsed nature of metamorphic reactions, which results in punctuated high-flux fluid flow during the brief episodes when delayed reactions are allowed to occur.

E. Czech, S. Gorgani, K. Gordon, and M. Kielman are thanked for technical assistance. Funding was provided by the Swedish Research Council (International Postdoc grant 2018-00200); European Union's Horizon 2020 (Marie Sklodowska-Curie grant 899546) and Geo-Ocean to Tual; the Natural Sciences and Engineering Research Council of Canada (Discovery Grant RGPIN-2015-04080); the Canadian Foundation of Innovation; and the British Columbia Knowledge Development Fund (joint infrastructure grant 229814) to Smit. NordSIMS-Vegacenter is funded by the Swedish Research Council as a national research infrastructure (#2017-00671). F. Stuart (editor), E. Baxter, R. Tamblyn, and B. Dragovic are thanked for their reviews that helped improve the clarity of this manuscript. This is Vegacenter publication #082.