Exhumation dynamics of high-pressure metamorphic rocks from the Voltri Unit, Western Alps: constraints from phengite Rb–Sr geochronology

Abstract

The Voltri Unit and adjacent Tertiary Piedmont Basin, Western Alps, preserve complementary bedrock and sedimentary archives of Alpine subduction and exhumation. Combined PT modeling and phengite Rb–Sr geochronology of bedrock and clast samples shows that Fe–Ti metagabbros and metasediments preserve a record of protracted high-pressure metamorphism, between ~ 50 and ~ 40 Ma. Bedrock and clast Fe–Ti metagabbros yield similar eclogite-facies peak conditions, between 23 and 25 kbar, 510–530 °C; phengite, zoned in celadonite content, with cores > 3.5 Si c.p.f.u. and rims < 3.3 Si c.p.f.u., constrains the timing of exhumation between 24 and 20 kbar to 45–49 Ma. A single impure quartzite bedrock sample records peak-P conditions of 18–19 kbar, 450–470 °C; partial phengite equilibration in this sample occurred continuously between 19 and 12 kbar, between ~ 45 and ~ 40 Ma. Exhumation-related recrystallization of high-pressure phengite to low-Si phengite in two metasedimentary samples occurred between ~ 33 and ~ 30 Ma, contemporaneous with the onset of deposition in the Tertiary Piedmont Basin and consistent with previous ⁴⁰Ar–³⁹Ar constraints on the timing of greenschist metamorphism. Combined with existing ages from the ophiolite, these data show that peak subduction-related high-pressure conditions, between ~ 18 and ~ 25 kbar, were attained at different times across the Voltri Unit, between ~ 50 and ~ 40 Ma, implying that the Voltri Unit comprises an assembly of discrete lithotectonic units that were juxtaposed prior to erosion and deposition in the Tertiary Piedmont molasse basin. The PTt data reported here support a model in which individual sheets of high-pressure material were detached from the downgoing plate, partially exhumed from peak pressures to blueschist facies conditions, while subduction continued, and were stored for > 10 Myr until subduction ceased on arrival of the European continent into the orogenic wedge. As shear tractions on the plate interface are considered to exceed available buoyancy forces for exhumation, we suggest that syn-subduction exhumation occurred along the wedge-plate interface.

Introduction

High-pressure (HP) metamorphic rocks are the only direct probes of the physical and chemical conditions deep within subduction zones. However, they are anomalous in that they have been returned to the Earth’s surface unlike the majority of subducted oceanic lithosphere. To accurately interpret the petrological information they record, it is, therefore, important to understand how HP metamorphic rocks are detached from the downgoing slab and returned to the Earth’s surface.

Exhumation of HP metamorphic rocks requires either the removal of material that caused the high pressures, or tectonic transport of the HP rocks through the overlying subduction complex. Erosion likely controls the final stages of exhumation, but additional processes are required for exhumation from mantle-depths (e.g., Platt 1993); these include corner flow (e.g., Cloos and Shreve 1988; Gerya, et al. 2002; Shreve and Cloos 1986), buoyancy-driven ascent of metasedimentary and serpentinite-rich bodies (e.g., England and Holland 1979; Hermann et al. 2000) and extension of the accretionary wedge (e.g., Jolivet et al. 2003). Analysis of Pressure–Temperature-time (PTt) data from HP metamorphic terranes shows that HP rocks are exhumed discontinuously, in short-lived episodes (< 15 Myr), during subduction of oceanic lithosphere (e.g., Agard et al. 2009,2018). These observations raise the questions: does exhumation occur concurrently with active subduction, or, are accreted HP materials stored at depth while subduction continues?

Retrograde PTt paths derived from petrological investigations of exhumed HP terranes are considered to be accurate indicators of the timing of exhumation relative to subduction (Ernst 1988). Concurrent exhumation and subduction is expected to result in similar prograde and retrograde PTt paths due to continued underplating of cold oceanic lithosphere; examples of such a PT evolution are the blocks and coherent terranes of the Eastern Franciscan belt (Banno et al. 2000; Kimura et al. 1996; Maruyama and Liou 1988) and the lawsonite metabasalts from the Sivrihisar Massif, Turkey (Davis and Whitney 2006). In contrast, exhumation following the cessation, or pronounced deceleration of subduction, occurs through relaxing isotherms and is expected to result in near-isothermal retrograde PTt paths, as preserved by the Saih Hatat eclogite blocks, Oman (Warren and Waters 2006), Eastern Alpine blueschists and eclogites (Smye et al. 2011) and the Nevado-Filábride Complex, southern Spain (Sánchez‐Vizcaíno et al. 2001).

This paper presents the results of a petrochronological investigation into the PTt evolution of blueschist- and eclogite-facies ophiolitic mafic rocks and metasediments of the Voltri HP Unit, Western Alps. Zoned phengite grains in a suite of HP samples record discrete intervals of PTt information that, collectively, constrain the tectonic history from Alpine subduction to exhumation and erosion. Constraints from phengite Rb–Sr geochronology and metamorphic phase equilibria show that eclogites and metasediments: (1) underwent HP metamorphism at depths between 60 and 80 km, (2) were partially exhumed, accreted and stored within the Alpine orogenic wedge for > 10 Myr, and (3) were finally exhumed following the cessation of subduction. These findings demonstrate the potential importance of storage of slab-top materials under forearc conditions for understanding subduction zone mass transfer.

The Voltri Unit

The Voltri Unit is a dismembered metaophiolite suite, representing the remnants of the Alpine-Tethys oceanic realm located between the southernmost sector of the Western Alps and the northernmost sector of the Apennines. Subduction and exhumation occurred during convergence between the continental Europe and Adria (a promontory of the African continent) plates throughout the Cretaceous-Tertiary period (Polino et al. 1990; Vanossi et al. 1984). After Capponi et al. (2016), the Voltri Unit (Fig. 1) comprises: (1) high-pressure calcareous metasediments and metavolcanics (the Voltri-Rossiglione Unit Auct.) associated with high-pressure serpentinites hosting eclogites, metarodingites, metagabbros and metabasalts (the Beigua Unit Auct.), and (2) variably-serpentinized oceanic mantle peridotite (the Erro-Tobbio Unit Auct.) that underwent Alpine subduction and eclogitization (Scambelluri et al. 1995). Combined, all these lithologies are interpreted to represent portions of the sedimentary cover, crust and mantle lithosphere of subducted and subsequently exhumed Tethyan oceanic lithosphere.

Fig. 1

Tectonic context of the Voltri Unit. Panel a: major tectonic units of the Alpine chain after Schmid et al. (2004,2008); thick red lines represent deformation fronts (solid for exposed, dashed for subsurface) and thin red lines represent major Neogene faults. Box corresponds to region of interest shown in b Panel b: simplified geological map of the Voltri Massif superimposed on topographic relief. Units and tectonic boundaries are from Capponi et al., (2016); white text boxes refer to locations of samples used in this study. Note that the Voltri-Rossiglione Unit of previous workers corresponds to Voltri Unit of this map; similarly, the Beigua and Erro Tobbio Units are included in Voltri Unit

The Voltri Unit is overlain to the north by the Tertiary Piedmont basin that contains molasse deposits with Oligo-Miocene stratigraphic ages (Gelati et al. 1998; Lorenz 1969), including blueschist and eclogite clasts in the basal units. These clasts have been interpreted to derive from the Voltri Unit during denudation of pre-existing Alpine topography (e.g., Federico et al. 2005). To the west, the Unit is juxtaposed against Hercynian basement (the Savona and Arenzano Units), and to the east the high-pressure ophiolitic sequence of the Palmaro-Caffarella Unit (blueschist-facies) and of the Sestri-Voltaggio Zone (pumpellyite-actinolite facies) separates the Voltri Unit from the low-grade flysch of the Northern Apennines.

Bedrock eclogites in the central portion of the Voltri Unit (Vara region, Fig. 1) outcrop as meter-to-decameter-scale pods in a metamorphosed serpentinite matrix. On the basis of major element and REE compositions, the eclogites are similar to modern-day Fe–Ti gabbros (Morten, et al. 1979). Textural characteristics have been used to define two groups of eclogites: (1) massive eclogites containing coarse-grained sodic clinopyroxene rimmed by fine-grained coronas of garnet and rutile; and (ii) foliated eclogites with euhedral garnet embedded in a fine-grained matrix of glaucophane, omphacite and rutile.

Petrologic investigations have shown that the Voltri eclogites experienced peak metamorphic conditions of 18–25 kbar and 450 °C–550 °C, followed by a polyphase retrograde evolution through the blueschist, amphibolite, and greenschist facies (Brouwer et al. 2002; Cimmino and Messiga 1979; Ernst 1981; Federico et al. 2007a, b; Liou et al. 1998; Messiga 1987; Messiga and Scambelluri 1991; Messiga et al. 1995; Starr et al. 2020; Vignaroli et al. 2005). Of particular note is the uniformity of recent estimates of peak ecogite-facies conditions, ~ 500–525 °C and 23–25 kbar, recorded by Fe–Ti metagabbros across the Voltri Unit (Starr et al. 2020). Similar peak conditions have been documented in eclogite-facies lherzolites of the Erro-Tobbio Unit Auct. (Scambelluri et al. 1991), whereas PT constraints from the Voltri metasediments are loosely constrained due to pervasive, exhumation-related retrogression. Polyphase retrogression during decompression is evidenced by the occurrence of chloritoid, albite, chlorite, titanite and lower-Si phengite assemblages (e.g., Cimmino and Messiga 1979; Federico et al. 2005).

The tectonic evolution of the Voltri Unit is constrained by a variety of geochronological constraints derived from multiple radiogenic decay systems and from samples pertaining to both bedrock of the Voltri Unit and clasts of the adjacent Tertiary Piedmont Basin (TPB). Combined, these data constrain the timing of peak eclogite-facies conditions across the Voltri Unit to ~ 50–38 Ma. Using bulk garnet Sm–Nd geochronology, Starr et al. (2020) showed that HP conditions were first attained in the northwestern portion of the Voltri Unit at ~ 50 Ma, followed by peak metamorphism in the Voltri metasediments around ~ 47–44 Ma, and ~ 41–38 Ma in Fe–Ti metagabbros in the central portion of the Voltri Unit. These direct constraints on the timing of eclogite-facies metamorphism are supported by ⁴⁰Ar–³⁹Ar ages of zoned high-Si phengite, interpreted by Federico et al., (2005) to represent peak HP conditions at ~ 49 Ma, followed by blueschist and greenschist retrogression at 43–40 and 33 Ma, respectively. Phengite ⁴⁰Ar–³⁹Ar dates obtained from eclogite blocks from the Cascine Parasi Mélange unit are consistent with younger attainment of the eclogite-facies peak at ~ 43 Ma, followed by blueschist facies retrogression at 43–40 Ma. However, these dates are difficult to reconcile with a U–Pb (SHRIMP) baddelyite age of 33.6 ± 1.0 Ma, interpreted to record the timing of eclogite-facies metamorphism in Ti‐clinohumite-bearing, altered Fe–Ti gabbros of the Beigua Unit (Rubatto and Scambelluri 2003).

The stratigraphy of the TPB provides additional constraints on the tectonic evolution of the Voltri Massif. Basal horizons, characterized by continental-to-transitional and shallow marine sediments of the Molare Formation, have an Early Oligocene age of deposition (Blow 1969; Gelati and Gnaccolini 1982; Turco et al. 1994) and contain detrital phengite grains that yield two clusters of ⁴⁰Ar–³⁹Ar dates (Barbieri et al. 2003; Carrapa et al. 2003). The oldest population comprises dates of ~ 45–48 Ma that overlaps the timing of peak HP metamorphism preserved in bedrock units, whereas the younger population is defined by dates of ~ 32–34 Ma, spanning the first depositional stage of the TPB. Apatite (U‐Th)/He and fission track thermochronology dates derived from both Oligo‐Miocene deposits and the Voltri Unit, demonstrate that the basin was exhumed above 4 km by ∼26 Ma (Vignaroli et al. 2010). Combined, these highly dispersed data demonstrate that the age and duration of HP conditions, and rates of subsequent exhumation are spatially variable across the Voltri Unit.

Sample petrography

Samples in this study comprise white mica-bearing eclogites and metasediments collected from bedrock (LPM 09-2, ASL13-2B, VCSC1 and CPSC-2) and basal conglomerates of the TPB (ASL13-17 and ASL13-21). Sample locations are shown in Fig. 1, petrographic characteristics are discussed and representative photomicrographs are presented in Fig. 2. In the text beneath, we refer to phengite as the solid solution series between muscovite, aluminoceladonite and celadonite (Rieder et al. 1998); all mineral abbreviations are after Whitney and Evans (2010).

Fig. 2

Sample petrography. Photomicrographs showing microtextural settings of white mica. Panel (a): close-up, plane polarized light (PPL) image of garnet, phengite and glaucophane in sample LPM 09-2. Red circles draw attention to straight grain boundaries between garnet and glaucophane (eastern red circle) and garnet and phengite (western) interpreted to represent textural equilibrium under peak HP conditions. Note the limited extent of retrograde chlorite growth at the expense of garnet. Panel (b): cross polarized light (XPL) image of phengite flakes in micaschist sample VCSC1. Abbreviations ph1 and ph2 refer to phengite generations that are aligned sub-parallel with and cross-cut the penetrative foliation, respectively. Panel (c): XPL image of penetratively deformed micaschist sample, ASL13-2B. Note the prominent S-C fabric defined by phengite flakes. Panel (d): XPL image of Na-amphibole- and garnet-bearing micaschist sample, CPCS-2. Phengite wraps idioblastic garnet and porphyroblasts of zoned Na-amphibole. Panel (e): XPL image of retrogressed mafic blueschist sample, ASL13-17. Garnet is disaggregated into chlorite and phengite. Straight grain boundaries between relict garnet and amphibole blasts (red circle) imply that these two phases were a part of the peak assemblage; note the absence of clinopyroxene. Panel (f): image (XPL) shows aggregates of phengite in calcschist sample ASL13-21. For all panels, mineral abbreviations are after Whitney & Evans (2010)

Bedrock samples

Sample LPM 09-2 is a mafic eclogite sampled from a meter-scale block in the northwestern sector of the Voltri Unit, along the Erro River, ca. 10 km north of Sassello (Fig. 1; La Pesca locality described by Scambelluri et al. 2016; 44° 33′ 23.15″ N; 8° 26′ 47.22″ E). It contains garnet, glaucophane, omphacite, phengite, rutile, epidote and quartz. Garnet forms mm-scale poikiloblasts with inclusion-rich cores and is wrapped by a penetrative foliation defined by blue amphibole and omphacite; epidote, rutile, quartz and blue amphibole are common as included phases. Multiphase mineral inclusions within the garnet cores consist of block-shaped aggregates of epidote, plagioclase and paragonite that are interpreted to derive from breakdown of former lawsonite. Idiomorphic phengite displays sharp grain boundaries and equilibrium microtextures with cogenetic garnet, omphacite and Na-rich amphibole (Fig. 2a). This sample was collected from the same location as Fe–Ti metagabbro V18-S617C01 reported by Starr et al. (2020).

Sample VCSC1 is a variably-retrogressed micaschist sampled from the central Voltri Unit (44° 28′ 34.48″ N; 8° 42′ 41.58″ E). It contains quartz, white-mica and plagioclase as major phases and carbonate, rutile, chlorite and tourmaline as accessory phases. Two generations of phengite are observed; one predates the penetrative foliation, whereas the other generation cross-cuts microlithons of recrystallized quartz (Fig. 2b). In each case, mica grains are between 100 and 200 μm long.

Sample ASL13-2B (44° 28′ 13.2″ N; 8° 36′ 50.7″ E) is a strongly foliated quartz-mica schist collected from the Vara region of the central Voltri Unit. It contains syntectonic quartz, phengite, carbonate and minor rutile, tourmaline and graphite. Phengite flakes exhibit kink bands and, along with graphite seams, define an S-C fabric; individual mica grains range from 50 to 500 μm length. Quartz grains have undergone recrystallization, show undulose extinction and form aggregates that define microlithons (Fig. 2c).

Sample CPSC-2 (44° 32′ 35″ N, 8° 28′ 56″ E) is a quartz-mica schist collected from the Cascine Parasi mélange (Federico et al. 2007a) and contains quartz, phengite, Na-amphibole, rutile, titanite, allanite-epidote and minor garnet. Phengite wraps euhedral poikiloblasts of zoned Na-amphibole and epidote; garnet is present as small idioblasts (< 1 mm diameter) that occur both within the matrix and as inclusions within Na-amphibole (Fig. 2d).

Clast samples from the Tertiary Piedmont Basin

Sample ASL13-17 is a retrogressed eclogite clast collected from the base of the TPB (44° 37′ 11″ N; 8° 43′ 30″ E) and contains, in order of decreasing abundance glaucophane, epidote, garnet, rutile, phengite and quartz (Fig. 2e). Garnet porphyroblasts are disaggregated and heavily retrogressed to chlorite. Straight grain boundaries between relict garnet and glaucophane blasts indicate that these two phases were a part of the peak assemblage. The conspicuous absence of omphacite implies post-peak retrogression that involved the consumption of clinopyroxene and (re-) equilibration of glaucophane and epidote on exhumation through blueschist facies conditions.

ASL13-21 is a calcschist clast sampled from the base of the TPB (44° 36′ 11″ N; 8° 36′ 15″ E). It is characterized by the high-variance mineral assemblage: phengite, quartz and carbonate; rutile, graphite, chlorite and tourmaline are each present as accessory phases. Phengite is present as flakes (200–500 μm length) that define a dominant foliation that wraps aggregates of quartz and carbonate. Clots of chlorite are interpreted to have formed from the retrogression of precursor garnet (Fig. 2f).

Analytical techniques

Whole-rock and mineral compositions

Major element compositions of rock-forming mineral phases were determined using a Cameca SX-5 electron microprobe at the Materials Characterization Laboratory, Penn State. The instrument is equipped with four wavelength-dispersive spectrometers; operating conditions comprised a 20 kV accelerating voltage and a 20 nA beam current. The spot size was 2 μm for all analyses and on-peak counting time was 20 s for all elements. Natural and synthetic oxides were used for calibration standards.

Bulk rock major element analysis of selected bedrock and clast samples were performed at the Department of Earth and Environmental Sciences, Franklin and Marshall College, utilizing a PANalytical 2404 X-ray fluorescence (XRF) spectrometer and the methods presented in Boyd and Mertzman (1987).

White-mica Rb–Sr geochronology

Whole-rock samples were crushed using a roller mill and washed with deionized water to remove fine particles. After passing through a 1 mm sieve, a Frantz isodynamic separator was employed to concentrate mica, garnet, and other mineral fractions. An agate mill was then used to reduce the mineral fraction size. Prior to digestion in a HF-HNO₃ solution, mineral fractions were spiked with a mixed ⁸⁷Rb-⁸⁴Sr tracer. Spike isotope concentrations were calibrated using NISTSRM984, RbCl, and NISTSRM987, SrCO₃, gravimetric solutions and checked against the established Rb–Sr standard, NISTSRM607, K-Feldspar (Compston et al. 1969; Nebel and Mezger 2006). Following dissolution, Sr was isolated using the automated Prepfast-MC Sr-Ca separation method (Romaniello et al. 2015); the Prepfast-MC system was also utilized to separate Rb using a 2 mL column filled with AG50Wx8 resin. For samples, ASL13-21, and 17, both Sr and Rb were analyzed on a ThermoFisher Triton Plus TIMS at Pennsylvania State University. The Rb fractionation factor was determined by concurrent analysis of NISTSRM984 assuming an ⁸⁷Rb/⁸⁵Rb = 0.3857. All other Rb measurements were made using MC-ICP-MS (ThermoFisher Neptune, Metal Isotopes Laboratory, Penn State) and correcting for mass bias using concurrent analyses of Zr (Waight, et al. 2002). A nominal uncertainty of 1% (1σ) was assigned to ⁸⁷Rb/⁸⁶Sr values based on the variability of NISTSRM607, measured concurrently. The instrument mean ⁸⁷Sr/⁸⁶Sr for NISTSRM987 was 0.71025 ± 0.00004 between June 2017 and July 2018. An uncertainty of 0.01% was assigned to ⁸⁷Sr/⁸⁶Sr determinations based on the propagated uncertainty of sample ⁸⁷Rb/⁸⁶Sr, typically 0.05%. Isochron parameters were calculated using IsoplotR (Vermeesch 2018) and the ⁸⁷Rb decay constant of Nebel and Mezger (2011).

Mineral chemistry

Major element compositions of white mica were collected for each sample that was dated by Rb–Sr geochronology; the compositions of other rock-forming minerals in samples LPM 09-2, ASL13-17 and CPCS-2 were also collected to support phase equilibria calculations. Mica, amphibole and garnet chemistry and compositional profiles are presented in Figs. 3, 4, 5, respectively. Representative mineral compositions are presented in Table 1.

Fig. 3

Mica chemistry. Figure shows Si against Al c.p.f.u. for white-mica from each of the samples investigated. Adjacent panels show false-color BSE images of individual mica grains. Note the pronounced zoning in average atomic number for VCSC1, ASL13-2 and ASL13-21

Fig. 4

Na, Na–Ca and Ca amphibole chemistry. Panels a–c show modified BSE intensity images of zoned amphibole from samples LPM 09-2, ASL-17 and CPCS-2. Following Schumacher (2007), panel d expresses amphibole compositions in terms of the relative magnitudes of the three dominant amphibole substitution vectors: edenite (Na_A), glaucophane (Na_B) and tschermakite (Al_T-Na_A). Note that in this coordinate scheme, substitution along the richterite vector results in negative tschermakite values, as demonstrated by several analyses from LPM 09-2. Arrow marks the core-to-rim compositional vector from ASL13-17. Panel e shows Na_B versus Si (c.p.f.u.) for all amphibole data. Panels f–h show classification diagrams for Na, Na–Ca and Ca amphiboles after Leake (1978), respectively. The color scheme for all plots is as shown in panel d

Fig. 5

Garnet chemistry. Panels a and b show variation in mole fractions of almandine (blue), spessartine (red) and grossular (green) across individual garnet grains from samples LPM 09-2 and ASL13-17. Gaps in data correspond to inclusions; solid and dashed lines represent different traverses across different garnet crystals. Panel c shows Ca (upper) and Mn intensity maps for garnet from sample ASL13-17. Panel d shows Mn intensity in a garnet poikiloblast from sample LPM 09-2; note the pseudomorphs after lawsonite. Panel e is an Mn intensity map for a garnet grain from sample CPCS-2

Table 1 Major element compositions of rock-forming minerals in samples used for P–T modeling. Oxides are reported in weight %, cations in c.p.f.u. and end-members in molar proportions

Phengite is present in each of the samples; Si contents vary from 3.2 to 3.7 c.p.f.u. Mica grains in samples LPM 09-2, VCSC1, ASL13-2B and ASL13-21 are zoned in celadonite content with decreasing inverted tschermak substitution (MgSiAl_-2) from core to rim. Phengite cores generally exceed 3.4 c.p.f.u. Si, whereas rims are characterized by < 3.3 c.p.f.u. Si (Fig. 3), implying that mica grains partially re-equilibrated during decompression. Paragonite is present as a minor alteration product around the rims of phengite grains in sample ASL13-17.

Zoned amphibole grains are present in samples LPM 09-2, ASL13-17 and CPCS-2. Microstructural evidence allows the distinction of three generations of amphibole in sample LPM 09-2: (1) mineral core domains of glaucophane-to-Mg-riebeckite composition (Na_B > 1.5, (Na + K)_A < 0.5, X_Mg > 0.5, where X_Mg = Mg/(Mg + Fe²⁺)) with minor substitution along edenite and tschermakite vectors (Fig. 4d); these regions exhibit patchy, irregular zonation in X_Mg (values between 0.5 and 0.6; Fig. 4f), visible in BSE (Fig. 4a); (2) mantles of Fe-richterite-to-richerite (Na_B = 1.0–1.5, (Na + K)_A > 0.5, X_Mg 0.4–0.5; Fig. 4g), and (3) narrow (< 50 μm), outermost rims composed of Fe-hornblende or barroisite (Na_B < 1.0, (Na + K)_A < 0.5, X_Mg 0.4–0.5, Si 7–7.5 c.p.f.u.; Fig. 4h). Amphibole in sample ASL13-17 exhibits progressive zonation from Mg-riebeckite-to-glaucophane cores to Fe-glaucophane rims; several analyses of richterite were collected from growth zones proximal to glaucophane core domains. These zoned HP amphibole cores are overgrown by discrete, narrow rims (< 50 μm) of Fe- to Mg-hornblende composition (Fig. 4h). Finally, sample CPCS-2 exhibits two generations of Na amphibole: a central glaucophane core phase overgrown by euhedral rims of Mg-riebeckite (Fig. 4f). The main similarity in amphibole chemistry between each of the samples is the glaucophane-riebeckite cores, interpreted to represent high-pressure conditions. The two mafic samples, LPM 09-2 and ASL13-17, exhibit similar magnitudes of retrograde tschermakite and edenite substitution in hornblende rim domains, consistent with formation during or following decompression from eclogite or blueschist facies conditions. However, it is worth noting that the continuous nature of the core-to-rim zonation observed in sample ASL13-17 is markedly different from the discrete overgrowths observed in sample LPM 09-2 and CPCS-2.

Zoned garnet is present in samples LPM 09-2, ASL13-17 and CPCS-2. Example chemical traverses and maps are presented in Fig. 5, and example chemical analyses are presented in Table 1. Garnet porphyroblasts in sample LPM 09-2 are almandine-rich solid solutions (X_alm > 50 molar %) and preserve at least two distinct growth zones (Fig. 5a, d). The internal zone dominates the volume of each garnet grain analyzed and is characterized by bell-shaped core-rim Mn zonation, implying prograde sequestration of elements according to Raleigh fractionation (yellow shading, Fig. 5d; core composition: Alm_55-67Grs_24-27Prp_24-27Sps_5-20). Chemical maps of Mn zonation reveal an external growth zone that spans the outermost ~ 100 μm radial distance and is characterized by low-Mn and Ca concentrations and an absence of internal zonation (deep purple shading, Fig. 5d; outermost rim composition: Alm_73-80Grs_15-20Prp_5-10Sps_0-1). A tortuous boundary between the internal and external growth zones is consistent with some degree of resorption following growth of the internal zone; however, the boundary is marked by a sharp (< 100 μm) change in spessartine concentration implying short length scales of diffusive modification. Garnet in sample ASL13-17 has a similar compositional range to LPM 09-2 and also has at least two growth zones (Fig. 5b, c), comprising a bell-shaped Mn-rich internal zone (hot shading, lower panel Fig. 5c; core composition: Alm_55-65Grs_20-26Prp_24-27Sps_5-15) surrounded by a Mn-poor external zone (cold shading, lower panel Fig. 5c; rim composition: Alm_73-81Grs_14-16Prp_9-12Sps_0-2). The boundary between these growth zones is defined by a step-like discontinuity in X_sps content that occurs over a length scale of < 100 μm. Garnet in sample CPCS-2 is spessartine-rich and occurs as small (< 100 μm diameter), euhedral and inclusion-free porphyroblasts. Atoll-shaped Mn zonation defines core domains (core composition: Alm_8-12Grs_9-11Prp_1-4Sps_71-79) which are surrounded by narrow rim domains (< 20 μm; rim composition: Alm_15-19Grs_4-9Prp_3-6Sps_65-73) with little discernable internal zonation (Fig. 5e).

Sodic clinopyroxene is present in sample LPM 09-2 as matrix blasts after primary igneous clinopyroxene with an omphacitic composition (~ Jd₂₅Ag₂₀Di₅₅). No discernable core-to-rim chemical zonation was observed in EPMA analyses.

Phengite Rb–Sr geochronology

Statistically valid isochrons were obtained for six of twelve samples initially analyzed (Fig. 6; Table 2); the remaining samples yielded low Rb–Sr ratios (< 1), unsuitable for age determination. Bedrock samples yield Rb–Sr whole rock-phengite isochron dates of 47.37 ± 0.41 (LPM 09-2; n = 5, MSWD = 9), 44.57 ± 1.68 (VCSC-1; n = 4, MSWD = 1.2), 35.76 ± 0.54 (CPCS-2; n = 3, MSWD = 22) and 30.45 ± 0.45 (ASL13-2B; n = 6, MSWD = 0.48). Clast samples yield whole rock-phengite isochron dates of 45.18 ± 0.33 (ASL13-17; n = 4, MSWD = 4.8) and 32.46 ± 0.32 (ASL13-21; n = 4, MSWD = 11). Results of two-point isochron calculations are presented in Table S1.

Fig. 6

White-mica Rb–Sr geochronology results

Table 2 Rb–Sr white-mica geochronology

PT modeling

A suite of PTX pseudosections were calculated to constrain prograde PT paths and peak PT conditions for two mafic samples (LPM 09-2 and ASL13-17) and one impure quartz micaschist sample (CPCS-2). We focused on these samples due to their well-constrained Rb–Sr phengite ages and amenable phase assemblages to constrain the PT conditions of phengite equilibration. All calculations were performed in the simplified (MnO–)Na₂O–CaO–K₂O–FeO–MgO–Al₂O₃–SiO₂–H₂O–TiO₂–Fe₂O₃ ((Mn–)NCKFMASHTO) chemical system using Gibbs free energy minimization as implemented in the Theriak-Domino program (de Capitani and Petrakakis 2010; version 11.03.2020). MnO was considered for the impure quartzite (CPCS-2) and for modeling the growth of garnet-core domains in the two mafic samples. All calculations were performed with H₂O as a saturated component based on: (1) the presence of prograde pseudomorphs after lawsonite (sample LPM 09-2); (2) the prevalence of glaucophane in peak mineral assemblages; (3) the presence of idiomorphic white-mica aligned with peak-P mineral fabrics samples, and (4) elevated XRF Loss On Ignition (LOI) values for samples LPM 09-2 and ASL13-17 (Table 3). It is likely, however, that some fraction of the LOI value reflects post-peak addition of H₂O during exhumation.

Table 3 Bulk-rock XRF compositions used for pseudosection calculations (mole %)

All calculations were performed with the internally consistent thermodynamic database of Holland and Powell (1998; version ds5.5) and the following activity-composition (a–X) models: garnet, biotite (White et al. 2007), plagioclase (Holland and Powell 2003), chlorite, epidote, talc (Holland and Powell 1998), white mica (Coggon and Holland 2002), ilmenite (White et al. 2005), magnetite (White et al. 2002), clinoamphibole and clinopyroxene (Diener and Powell 2012). Quartz/coesite, aluminosilicate, albite, rutile, sphene, zoisite and lawsonite were all treated as pure phases; since we excluded CO₂ from consideration, the fluid phase was also treated as pure H₂O. The Holland and Powell (1998) ds5.5 thermodynamic database and derivative a–X models were preferred over the more recent ds6.2 database (Holland and Powell 2011) for the following reasons: (1) calculations initially performed with the ds6.2 database and the updated amphibole and clinopyroxene models of Green et al (2016) yielded unreasonably high peak pressure estimates for garnet rim compositions (> 28 kbar), within the coesite stability field, for which there is no supporting petrographical evidence; (2) the consideration of K₂O and TiO₂ components in Na and Na-Ca amphibole results in negligible differences in phase relations under the PT conditions of interest, and (3) the ds5.5 database and models accurately predict the stability of lawsonite in garnet core assemblages and yield similar garnet compositions to those observed under non-UHP conditions.

Estimated bulk compositions were based on whole-rock XRF analyses for each of the samples (Table 3). Values of XFe³⁺  = 0.30 and 0.66 (XFe³⁺  = Fe³⁺/(Fe³⁺ + Fe²⁺) in molar proportions) were assumed for the two mafic samples (Fe₂O₃ = 2.11, and 2 mol%, garnet core and rim compositions, sample LPM09-2; Fe₂O₃ = 2.24 and 1.73 mol%, sample LPM09-2) and the impure quartzite sample (Fe₂O₃ = 1.3 mol%, sample CPCS-2), respectively. These estimates were determined by comparison of ferric iron titration measurements—that tend to overestimate XFe³⁺—and estimates of ferric iron concentrations derived by mineral modes and EPMA analyses. We note that a value of XFe³⁺  = 0.3 is agreeable with values reported for altered oceanic crust (e.g., XFe³⁺  = 0.1 – 0.5, Dungan et al. 1978) as well as for other massive Fe–Ti metagabbros from the Voltri massif (e.g., XFe³⁺  = 0.21 – 0.37, Starr et al. 2020).

Growth of porphyroblastic phases, such as garnet, at temperatures lower than required for grain-scale volume diffusion can modify the reactive bulk composition and subsequent prograde phase relations (e.g.,Evans 2004; Konrad-Schmolke et al. 2008; Marmo et al. 2002; Tinkham and Ghent 2005; Zeh 2006). To account for this in our calculations, we followed the approach presented by Evans et al. (2004) that applies a Rayleigh fractionation model based on the measured Mn content of garnet to derive the vectors of crystal fractionation. We applied this approach to samples LPM09-2 and ASL13-17 in which garnet is zoned in almandine, pyrope, grossular and spessartine components and is volumetrically abundant (> 10 vol%); garnet in sample CPCS-2 is present as a minor phase and any such fractionation is assumed to have had a negligible effect on the reactive bulk composition. For both samples, core-rim EPMA traverses were screened to identify the most primitive Mn-rich core compositions, reducing the risk of using an aliased, non-central chemical profile. Core-to-rim Mn transects exhibit classic bell-shaped zonation (Fig. 5), consistent with prograde formation during Rayleigh fractionation. Pseudosections for the formation of garnet cores were constructed using measured XRF bulk-rock compositions, whereas phase relations pertaining to equilibration of garnet rim domains were calculated using fractionated bulk compositions. All bulk compositions are presented in Table 3. Estimates of the PT conditions of equilibration for garnet rims and cores were constrained by linearly-independent combinations of isopleths of almandine, pyrope, grossular and spessartine.

Sample LPM 09-2 (bedrock mafic schist)

The PT phase diagram calculated using the garnet core composition (Fig. 7a) is dominated by amphibole- and omphacite-bearing tri- and quadravariant assemblages. Garnet core compositions (composition used to constrain PT coordinates of isopleth intersection: Alm_54-55Grs_26-28Sps_19-21; Table 1) are calculated to be stable between 17–23 kbar and 460–510 °C in the phase assemblage field: grt-chl-omp-gln-lws-qz-(rt-ph-H₂O), in broad agreement with the presence of paragonite + clinozoisite pseudomorphs after lawsonite in garnet poikiloblasts (Fig. 5d). The ~ 5 kilobar pressure range obtained for garnet core growth reflects the poor level of fit between garnet isopleths. Conditions of outermost garnet rim equilibration (Fig. 7b; Alm_73-75Grs_14-16Prp_8-11; Table 1) are estimated to have occurred at higher pressures to core growth, around 24–25 kbar and 510–530 °C in the grt-omp-gln-tlc-qz-(rt-ph-lws-H₂O) assemblage field. This assemblage generally agrees with the observed mineral paragenesis with the notable exception that lawsonite is not preserved. It is plausible that lawsonite was indeed a constituent of the peak assemblage and underwent decomposition during retrograde, exhumation-related metamorphism. Support for this contention is provided by the presence of pseudomorphs in garnet rim domains of sample LPM09-2 (Fig. 5d). An alternative explanation is that the rock was undersaturated in H₂O under peak metamorphic conditions and that the lawsonite stability field calculated assuming H₂O saturation in Fig. 7b does not accurately reflect the metamorphic system. Fig. S1 displays TX phase relations for bulk compositions containing 2 to 20 mol percent H₂O between 500 and 600 °C at 24.5 kbar. Inspection of the plot reveals that bulk-rock H₂O concentrations of < 3 mol percent are required to stabilize lawsonite-absent equilibria and that under such H₂O-undersaturated conditions, garnet isopleths are strongly dependent on bulk-rock H₂O and shift to lower-T with decreasing mole percent H₂O.

Fig. 7

Phase equilibria modeling for samples LPM09-2 (panels a–b), ASL13-17 (c–d) and CPCS-2 (e–f). Pressure–temperature fields are shaded according to assemblage variance. Dashed lines correspond to observed ranges of garnet (mol fractions of almandine, pyrope, grossular and spessartine) and phengite (c.p.f.u. Si) compositions as determined by EPMA analyses. Beige polygons reflect best-fit conditions of garnet core and rim equilibration determined by intersection of isopleths. For samples LPM09-2 and ASL13-17 phase diagrams on the left (a and c) were calculated using bulk rock XRF compositions to model garnet core growth, whereas diagrams on the right (b and d) correspond to conditions of garnet rim growth. For sample CPCS-2 garnet core fractionation was not considered and panels e and f show phase relations and isopleths calculated using the bulk rock XRF composition, respectively. See text for discussion

The observed range of Si-in-phengite (3.4–3.6 c.p.f.u..) corresponds to pressures < 23–24 kbar at peak temperatures of garnet rim equilibration (light blue lines, Fig. 7b), implying that phengite equilibrated during decompression from peak pressures.

Sample ASL13-17 (retrogressed eclogite clast)

The topology of the garnet core PT pseudosection calculated for sample ASL13-17 is characterized by expansive tri- and pentavariant amphibole- and chlorite-bearing assemblages (Fig. 7c). Garnet core compositions (Fig. 7c; Alm_58-60Grs_24-26Sps_8-10; Table 1) correspond to a region of PT space between 17 and 19 kbar, and 480 and 510 °C spanning the low-P limit of clinopyroxene stability in the grt-omp-chl-ep-gln-lws-qz(-rt-ph-H₂O) assemblage field. The garnet rim PT pseudosection (Fig. 7d) shares a similar topology to bedrock sample LPM09-2 (Fig. 7b), characterized by quadra- and pentavariant amphibole- and omphacite-bearing stability fields. Conditions of garnet rim equilibration are tightly constrained by measured garnet compositions (Alm_73-74Grs_14-16Prp_8-11; Table 1) to between 23–25 kbar and 515–530 °C, in the grt-omp-gln-lws-qz(-rt-ph-H₂O) assemblage stability field. The observed peak mineral assemblage (gln-ep-grt-ph-rt-qz) is notably devoid of clinopyroxene and lawsonite, both of which are predicted to be stable under peak PT conditions (~ 5 and ~ 15 vol.% omphacite and lawsonite, respectively, at 24 kbar and 525 °C). Epidote is predicted to be stable at pressures less than ~ 20 kbar, forming as a product of lawsonite and omphacite breakdown (e.g., Fig. 7d) and implying that the relict peak assemblage contained both of these phases and was similar to sample LPM09-2. Finally, the observed range of peak Si-in-phengite values (3.4–3.6 c.p.f.u.) occurs between ~ 20 and ~ 24 kbar at the peak temperature obtained from garnet rim isopleths, implying that phengite analyses with Si contents close to 3.6 c.p.f.u. preserve a record of peak-P conditions.

Sample CPCS-2 (bedrock metasediment)

The PT pseudosection calculated for impure quartz-mica schist CPCS-2 is dominated by tri- and quadravariant garnet- and amphibole-bearing assemblages (Fig. 7e). Due to the Mn-rich bulk composition (MnO, 1.45 mol.%; Table 3), garnet stability is predicted to extend to low-grade conditions < 5 kbar and < 400 °C. Garnet core compositions (Alm_10-11, X_Mg = 0.17 – 0.18; Table 1) correspond to PT conditions of 430–450 °C and 12–14 kbar, in the g-gl-chl-sph-ru-hem(-ep-ph-q-H₂O) stability field, whereas rim compositions (Alm_17-18, X_Mg = 0.21 – 0.22; Table 1) occur at higher pressures, between 18 and 19 kbar, at 450–470 °C, in the grt-gln-chl-rt-ilm(-ep-ph-qz-H₂O) stability field. The peak assemblage of grt-gln-rt-ep-ph-qz-H₂O is in broad agreement with the predicted phase assemblage with the exception that chlorite was not identified in thin section. These peak PT conditions are in good agreement with observed Si-in-phengite values (3.4–3.6 c.p.f.u.), demonstrating that phengite with the highest Si contents equilibrated at pressures around 18–19 kbar; conversely, the lowest Si-content phengite corresponds to ~ 12 kbar at peak temperatures, similar to the pressures at which the observed replacement of rutile by sphene is predicted to occur.

Interpretation of Rb–Sr white-mica dates

Combined, the PT modeling and Rb–Sr white-mica dates constrain segments of the prograde, peak and early-retrograde PTt paths for three samples. However, as four of the six samples analyzed here contain zoned phengite grains and the whole-grain Rb–Sr technique produces a volume-average of intracrystalline isotopic zonations, it is important to first consider the effect of intragranular chemical, and thus temporal, zonation on Rb–Sr isochron statistics.

Figure 8 shows a suite of isochrons calculated with synthetic Rb–Sr isotopic data that reflect different mixtures of high- and low-Si phengite (X = volume fraction high-Si phengite). The data were generated assuming uniform distributions of equilibration ages between 45 and 50 Ma for phengite cores and 30–34 Ma for low-Si phengite overgrowths, as shown by ⁴⁰Ar–³⁹Ar data from Voltri white micas (Federico et al. 2005). The calculations also assume that ⁸⁷Rb/⁸⁶Sr varies between 10 and 40 for both mica generations and an initial ⁸⁷Sr/⁸⁶Sr value of 0.70970; these parameters were chosen to reflect the range in Rb–Sr isotopic compositions determined for the metasedimentary samples analyzed (VCSC1, CPCS-2, ASL13-2B, ASL13-21; Table 2), which exhibit evidence for low-Si phengite overgrowth on high-Si phengite cores.

Fig. 8

Effect of intragrain zonation on Rb–Sr isochron systematics. Panels a–d show synthetic data generated for different proportions (X = vol. fraction phengite) of HP phengite (equilibration dates between 45 and 50 Ma, ⁸⁷Rb/⁸⁶Sr between 10 and 40, ⁸⁷Sr/⁸⁶Sr_i = 0.70970) and low-Si phengite overgrowth (equilibration dates between 30 and 34 Ma, ⁸⁷Rb/⁸⁶Sr between 10 and 40, ⁸⁷Sr/⁸⁶Sr_i = 0.70970). Nominal uncertainties of 1% and 0.01% (1σ) assumed for ⁸⁷Rb/⁸⁶Sr and ⁸⁷Sr/⁸⁶Sr ratios, respectively. Isochron statistics were computed using IsoplotR (Vermeesch 2018)

Two principal conclusions can be drawn from the calculations. First, the computed isochron MSWD values exhibit a strong positive correlation with the volume fraction of low-Si phengite overgrowth, demonstrating that MSWD is potentially a useful indicator for the degree of isotopic heterogeneity in the analyzed mica population (e.g., Glodny et al. 2008; McIntyre et al. 1966; Wendt 1993). Second, for mixtures dominated by high-Si phengite (i.e., X > 0.5), isochron dates represent minimum estimates of the true age of high-Si phengite equilibration and conversely for mixtures dominated by low-Si phengite (X < 0.5), isochron dates represent maximum dates of the lower-P equilibration. Of course, these calculations are simplified and do not account for further complications associated with variations in ⁸⁷Sr/⁸⁶Sr_i between mica generations and progressive overgrowth of low-Si phengite. Nevertheless, these principles provide a qualitative framework for the interpretation of the Voltri white-mica Rb–Sr. In addition to the complete multi-point Rb–Sr isochrons shown in Fig. 6, two-point isochrons connecting individual white-mica analyses with the whole rock fraction were calculated and are shown in Table S1. These two-point isochrons provide further constraints on the effect of mixing multiple mica generations on Rb–Sr isochron systematics and the timing of HP phengite growth.

White mica in the bedrock eclogite sample, LPM 09-2, is dominated by phengite and individual grains exhibit minor Si zonation, between 3.6 and 3.4 c.p.f.u., from core to rim domains. The pseudosection calculations (Fig. 7a, b) show that these phengite compositions correspond to a P difference of ~ 4 kbar recorded within individual mica grains, at temperatures corresponding to maximum metamorphic pressures (510–530 °C). Although the observed phengite Si contents are also calculated to be stable during a plausible prograde PT trajectory (Fig. 7b), the observation that phengite grains share straight grain boundaries with peak mineral phases, garnet, glaucophane and omphacite (Fig. 2a), supports the interpretation that phengite equilibrated close to peak PT conditions (24–25 kbar), likely during the initial stages of decompression at < 23 kbar. The elevated MSWD of the full Rb–Sr isochron (47.37 ± 0.41 Ma; MSWD = 9) and the presence of minor near-rim paragonite on some phengite grains implies that this date represents an estimate of the minimum age of HP metamorphism. Accordingly, we interpret the pair of two-point, whole rock-white mica, isochron dates that yield dates of ~ 48 ± 1 Ma as the best constraint on the age of HP phengite growth (Table S1).

White mica from the retrogressed eclogite clast ASL13-17 also exhibits minor variations in Si content, between 3.4 and 3.6 c.p.f.u., corresponding to an absolute P range of 4–5 kbar (Fig. 7d). Microtextural relations show that phengite equilibrated with chlorite during decompression-related consumption of garnet (Fig. 2e); this is consistent with phengite equilibration between ~ 20 and 24 kbar during the early stages of exhumation from peak PT conditions (23–25 kbar and 515–530 °C), notably similar to the interpretation proposed for bedrock mafic eclogite LPM09-2. The low MSWD (4.8) of the full Rb–Sr isochron (45.18 ± 0.33 Ma) supports the interpretation that this age is minimally impacted by retrograde white mica generations and represents the timing of HP phengite growth. We thus interpret that the sample underwent early decompression from peak conditions around ~ 45 Ma, consistent with the oldest two point isochron date of 45.86 ± 0.92 Ma (Table S1).

In contrast to the other metasediment samples investigated, phengite from micaschist CPCS-2 appears to preserve a restricted range in Si contents, between 3.4 and 3.6 c.p.f.u, overlapping the range of Si contents observed in phengite from the mafic samples. Phengite is present as a matrix phase and wraps porphyroblasts of garnet and Na-amphibole (Fig. 2d), implying that high-Si phengite was reoriented and underwent partial chemical equilibration (to Si 3.4 c.p.f.u.) during exhumation-related deformation. Pseudosection calculations show that the observed range in Si contents corresponds to a range of pressures between ~ 12 and 19 kbar at peak metamorphic temperatures of 450–470 °C. These PT conditions are in broad agreement with the wide range of peak PT conditions (~ 15–20 kbar and ~ 400–550 °C) reported by Federico et al (2007a, b) for mafic and metasedimentary blocks at the same locality. The elevated MSWD (22) of the Rb–Sr isochron for this sample is at odds with the limited range of phengite Si contents observed and implies that the phengite EPMA analysis did not fully characterize the diversity of phengite chemistry present in the sample. Thus, we interpret that the full isochron age (35.76 ± 0.54 Ma) reflects the analysis of phengite fractions with significant retrograde components and we, therefore, consider the oldest two-point isochron date (42.93 ± 2.65 Ma, Table S1) as a minimum age constraint on the timing of phengite growth under retrograde blueschist facies conditions.

Quantitative PT constraints for the remaining three metasediments, ASL13-2B, ASL13-21 and VCSC1 were not derived due to their high variance mineral assemblages. However, each of these samples preserve relict high-Si phengite cores, overgrown to various extents by low-Si phengite rims (Fig. 3), and a broad positive correlation exists between Rb–Sr white mica date and the minimum mica Si-content observed (Fig. 9). This correlation, combined with the low MSWD of each of the isochrons (< 11), enables qualitative interpretations of the Rb–Sr phengite dates. The well-preserved high-Si phengite cores in sample VCSC1 (Fig. 3) are consistent with the interpretation that this sample records HP conditions at 44.57 ± 1.68 Ma. In contrast, samples ASL13-2B and ASL13-21 exhibit a greater volume fraction of phengite grains with a progressive spread in Si contents between ~ 3.6 and 3 c.p.f.u., indicating that the Rb–Sr isochrons are likely dominated by low-Si phengite and represent upper bounds to the timing of white mica equilibration during exhumation, between 30 and 33 Ma.

Fig. 9

Relationship between white-mica Rb–Sr date and Si-content. Individual data points correspond to EPMA spot analyses

In summary, the analyzed samples constrain the timing of phengite equilibration under eclogite and blueschist facies conditions to between ~ 49 and ~ 40 Ma. This age range is derived from: (1) the upper bound to the two-point isochron date for sample LPM09-2 (48 ± 1 Ma), and (2) the lower bound to the two-point isochron for sample CPCS-2 (42.93 ± 2.65 Ma); excluding uncertainties reduces the range to ~ 48–43 Ma. Recrystallization of HP phengite to low-Si phengite occurred between ~ 33 and ~ 30 Ma. These dates overlap both the range of phengite ⁴⁰Ar–³⁹Ar dates reported by Federico et al (2005), who obtained eclogite and blueschist facies phengite dates between 40 and 49 Ma, and a greenschist facies overprint between 32 and 34 Ma, and the Sm–Nd garnet dates from Starr et al (2020) who reported ages of ~ 40 Ma and ~ 50 Ma from the central and northwestern Voltri Unit, respectively.

Tectonic evolution of the Voltri Unit

Diachronous HP metamorphism

Combined with existing datasets, the PTt data presented here allow for discrimination between competing structural models for the metamorphic evolution of the Voltri Unit. Contiguous PTt histories from across the unit would suggest that the Voltri Unit behaved as a single, coherent terrane during subduction and exhumation; rather, disparate and diachronous PTt histories would suggest an amalgamation of discrete subunits.

Following Starr et al (2020), a compilation of available geochronological data for the Voltri Unit is presented in Fig. 10. Inspection of the data shows a diachronous distribution of HP mineral dates across the Voltri Unit. Peak HP metamorphism is oldest in the northwest Voltri Unit, where eclogite facies conditions were attained ~ 50 Ma as evinced by similar Sm–Nd garnet (sample V18-S617C01, Starr et al (2020)) and high-Si phengite Rb–Sr dates (sample LPM-09-2, 47.37 ± 0.41 Ma, this study). Eclogitic and metasedimentary blocks from the Cascine Parasi mélange structure yield distinctly younger ⁴⁰Ar–³⁹Ar and Rb–Sr phengite dates, between ~ 45 and 40 Ma (Federico et al., (2007a, b) and the oldest two-point isochron date, 42.93 ± 2.65 Ma, sample CPCS-2, this study). Together with the disparate range of exotic HP-LT blocks preserved in the mélange, these dates imply a different origin and/or tectonic evolution to the central and northwestern portions of the Voltri Unit. In contrast to the northwest Voltri Unit, Sm–Nd garnet dates from the central portion of the Voltri Unit (Vara region) indicate that peak conditions (23–25 kbar, 500–525 °C) were attained ~ 10–12 Myr later, between 41 and 38 Ma (Starr et al. 2020). These garnet dates indicate that the U–Pb baddeleyite age of 33.6 ± 1 Ma obtained from a Ti-clinohumite vein in the Vara region by Rubatto and Scambelluri (2003) most likely represents post-peak crystallization and not HP conditions as originally interpreted. Peak conditions in metasediments of the central and eastern portion of the Voltri Unit occurred between 47 and 44 Ma (sample VCSC-1, this study; sample FL34 of Federico et al (2005), estimated PT conditions of > 15–16 kbar and 400–500 °C), implying that such metasediments experienced a distinct PTt evolution relative to neighboring metagabbros in the Vara region of the central Voltri massif. In agreement with previous investigations (Federico et al. 2005, 2007b; Starr et al. 2020), these data collectively show that peak HP conditions, between ~ 18 and ~ 25 kbar (~ 60–90 km; ρ = 2.9 g/cc), were attained at different times across the Voltri Unit, between ~ 50 and ~ 38 Ma, consistent with previous interpretations that the Voltri Unit comprises a sequence of discrete lithotectonic units as opposed to a structurally coherent unit. It is worth noting, however, that the current density of PTt sampling across the terrane precludes precise identification of the position and nature of the inferred tectonic boundaries that must separate the various sub-units.

Fig. 10

Geochronology of the Voltri Unit. Compilation of available geochronological data. Horizontal bars correspond to 2σ errors. Data sources: U–Pb titanite and zircon (Vignaroli et al. 2010); U–Pb baddeleyite (Rubatto and Scambelluri 2003); ⁴⁰Ar–³⁹Ar mica (Barbieri et al. 2003; Federico et al. 2005); Sm–Nd garnet, (Starr et al. 2020), and Rb–Sr white-mica, this study. Age of onset of TPB sedimentation from Turco et al (1994) and Gelati and Gnaccolini (1998)

Despite the ~ 12 Myr difference in age of HP metamorphism between the northwest and central Voltri Unit, the PT conditions of HP metamorphism define a restricted range of thermal gradients, between ~ 6 and ~ 8 °C/km (ρ = 2.9 g/cc). This observation implies that an active subduction regime was operating from the Early Eocene to the Eocene/Oligocene boundary.

The geochronological compilation also shows that the Rb–Sr white-mica dates presented here span a similar date range to ⁴⁰Ar–³⁹Ar white-mica dates reported previously for both clasts and bedrock samples that range from ~ 32 to ~ 65 Ma (Barbieri et al. 2003; Carrapa et al. 2003; Federico et al. 2005, 2007b). This similarity between datasets is expected if extraneous ⁴⁰Ar contamination is negligible. Furthermore, there is no systematic difference in the distribution of ⁴⁰Ar–³⁹Ar and Rb–Sr white mica dates for bedrock and clast samples, implying that the HP units of the Voltri Unit were juxtaposed prior to surface exposure and the timing of basal deposition in the adjacent TPB. The timing of basal sedimentation in the TPB (34–30 Ma) is also contemporaneous with several exhumation-related ⁴⁰Ar–³⁹Ar and Rb–Sr white mica dates (~ 30–33 Ma, samples ASL13-2B and ASL13-21, this study). This implies that (1) greenschist facies metamorphism was concomitant with erosion and sedimentation, and (2) at least the final stages of exhumation of HP units involved significant erosion at rapid rates.

Exhumation of the Voltri Unit

Preservation of a protracted history of subduction in the Voltri Unit raises the question: was exhumation continuous over this time interval? As a precedent to the ensuing discussion, two end-member tectonic scenarios for the exhumation of the Voltri Unit are introduced: (1) exhumation occurred concurrently with subduction resulting in the continuous return of HP material to shallow crustal depths (e.g., Shreve and Cloos 1986), or (2) HP slices were accreted to the forearc under blueschist and eclogite facies conditions before abrupt exhumation and juxtaposition during the cessation of subduction (e.g., Bayet et al. 2018).

Using the conditions and timing of HP metamorphism and the onset of basal deposition in the TPB as constraints, time-averaged rates of exhumation are between 0.4 and 0.54 cm/year for the northwest Voltri Unit (calculated using durations of 16–20 Myr, P between 23 and 25 kbar and ρ = 2.9 g/cc) and between 0.73 and 1.74 cm/year in central Voltri Unit, respectively (durations of 5–11 Myr). In sufficient PT data are available to calculate exhumation rates for the eastern metasediments (e.g., sample VCSC-1). Such time-averaged rates are comparable to exhumation rates derived from other Alpine oceanic HP units, including the Schistes Lustres (e.g.,Agard et al. 2002) and the Gran Paradiso massif (e.g.,Manzotti et al. 2018). While affording a constraint on the total duration over which exhumation may have operated, these time-averaged exhumation rates do not account for: (1) potential storage of the tectonic slices that comprise the Voltri Unit during on-going subduction, and (2) the differential exhumation that must have occurred to facilitate tectonic juxtaposition prior to erosion as a coherent package that explains the mixture of HP detritus preserved in the TPB.

In the northwestern sector of the Voltri Unit, garnet Sm–Nd and phengite Rb–Sr ages constrain peak pressure conditions to ~ 50 Ma (Starr et al. 2020), followed by phengite growth ~ 2–3 Myr later (this study; sample LPM 09–2, 49–47 Ma). This observation implies that exhumation initiated within 1–3 Myr of the attainment of peak eclogite-facies conditions, while subduction remained active. Assuming, (1) pressure differences between 1 and 4 kbar, and (2) a duration of between 1 and 3 Myr separate the conditions of garnet-rim and phengite equilibration results in rates of 0.12–1.40 cm/year (ρ = 2.9 g/cc) for this early stage of syn-subduction exhumation.

Several previous investigations have documented post-peak HP metamorphic overprints of the Voltri Unit that support the interpretation that a phase of HP exhumation occurred contemporaneous with active subduction. These include: (1) development of a post-eclogitic assemblage comprising Na-amphibole, phengite and garnet (Federico et al. 2005); (2) coexisting glaucophane and winchite idioblasts that cross-cut eclogite-facies foliations (Messiga and Scambelluri 1991), (3) formation of lower-pressure syn-tectonic blueschist facies assemblages (Vignaroli et al. 2005), and (4) reported retrograde blueschist-facies phengite growth at 8–15 kbar and 400–500 °C in eclogite clasts from the TPB at 45.2–41.8 Ma (Federico et al. 2005). Collectively, these constraints, when combined with the Rb–Sr phengite data, are consistent with the interpretation that individual sheets of HP material were detached from the downgoing plate, partially exhumed from peak pressures into the blueschist-facies (~ 8–20 kbar) and stored for durations between ~ 10 (central Voltri Unit) and ~ 15 Myr (northwest Voltri Unit) within the Alpine orogenic wedge. Final exhumation was likely facilitated by the cessation of subduction that accompanied arrival of the European continent into the subduction zone, ~ 32–34 Ma (e.g., Rubatto and Hermann 2001).

Comparison with circum-Alpine HP units

The metamorphic history of the Voltri Unit shares common features with several of the circum-Alpine HP terranes. In particular, the Zermatt-Saas and Monviso ophiolites of the Western Alps represent coherent slices of Tethyan oceanic lithosphere that preserve a diachronous record of HP metamorphism that is remarkably similar to the Voltri Unit. In the Zermatt-Saas ophiolite, constraints on timing of eclogite-facies conditions span two ranges, the first between c. 52–46 Ma, interpreted to date prograde and peak metamorphism in the uppermost structural levels of the ophiolite, and the second, between 42 and 39 Ma, derived from samples in the Saas Fee and St. Jacques regions (Amato et al. 1999; de Meyer et al. 2014; Lapen et al. 2003; Skora et al. 2015). In the case of the Monviso ophiolite, peak eclogite-facies conditions were attained between 51 and 45 Ma in the upper, Lago Superiore Unit, in contrast to the speculative ~ 40 Ma age proposed for the underlying Monviso unit (Angiboust and Glodny 2020; Duchêne et al. 1997; Garber et al. 2020; Rubatto and Angiboust 2015; Rubatto and Hermann 2003). The shared diachronous nature of these Alpine ophiolite units implies that they comprise an amalgamation of tectonic slices that underwent distinct PTt evolutions prior to juxtaposition.

Several of the circum-Alpine HP units also exhibit evidence for an initial phase of exhumation that occurred, while subduction was active. In the Eastern Alps, the Tauern Eclogite Zone underwent peak HP metamorphic conditions, 24–28 kbar, ~ 560 °C, at ~ 34 Ma (Glodny et al. 2005; Smye et al. 2011), followed by a retrograde epidote blueschist facies overprint at 400–450 °C, 10–15 kbar that occurred prior to collision and underplating of the leading edge of the European margin at ~ 30 Ma (e.g., Gleissner et al. 2007; Glodny et al. 2005). In the Western and Northwestern Alps, the Zermatt-Saas, Rocciavre and Monviso ophiolitic terranes also preserve evidence for overprinting of eclogite-facies peak assemblages under epidote blueschist conditions at 10–15 kbar and 400–450 °C (Angiboust et al. 2012; Schwartz et al. 2000); this retrograde event occurred between 39 and 35 Ma (Angiboust and Glodny 2020) contemporaneous with, or just preceding, deep subduction of the Internal Crystalline Massifs, such as the Dora Maira UHP continental slice (36–33 Ma) (e.g., Duchêne et al. 1997; Gebauer et al. 1997; Rubatto and Hermann 2001). This orogen-wide sequence of metamorphic events requires exhumation from depths of ~ 70–90 km to ~ 40–60 km, while subduction was active and before entry of continental basement into the Alpine subduction zone. As noted by Angiboust and Glodny (2020), this implies that additional forces, aside from the positive buoyancy force exerted by burial of continental crystalline material, are required to facilitate deep exhumation of the Alpine HP units.

Implications for exhumation dynamics

Whereas shallow exhumation following continental collision can plausibly be driven by denudation of overburden (e.g., Platt 1993), the mechanisms responsible for deep exhumation during, or shortly after the cessation of active subduction are less certain. In the absence of return flow, two principal forces act on material along the subduction interface: shear stresses that operate along the interface and body forces resulting from the density contrast between slab-top and adjacent crustal or mantle rocks. Subduction occurs if shear forces exceed buoyancy forces. Constrained by forearc heat-flow measurements, estimates of shear stresses at the maximum depth of thrust-faulting earthquakes along the plate-plate interface of modern subduction zones vary from ~ 20 to ~ 125 MPa (e.g., England 2018; Gao and Wang 2014). These estimates are consistent with the magnitudes of shear stress required to support topography above plate-plate interfaces (e.g., Davis et al. 1983; Lamb 2006; Suppe 2007). Along the plate-wedge interface estimates of the magnitude of shear stresses are typically < 20 MPa. These lower values of shear stress satisfy the requirement that the interface be strong enough to drive corner flow beneath the mantle wedge (Kelemen et al. 2003; McKenzie 1979) and paleo-piezometric estimates from exhumed HP rocks (e.g., Platt et al. 2018; Stöckhert et al. 1997).

Following England and Holland (1979), the magnitude of the buoyancy force ($h\Delta \rho gsin\theta$, where $h$ is interface width, $\Delta \rho$ is density contrast, $g$ is gravitational acceleration and $\theta$ is slab dip) relevant to typical slab-top materials is 1–5 MPa (assuming a slab dip of 20°, a density contrast of 300 kg/m³ and an interface width of 1–3 km). Figure 11 shows the magnitude of the buoyancy force calculated for a wide range of parameters plotted against the estimates of interface shear stress described above. Shear stresses significantly exceed buoyancy forces along the plate-plate interface for reasonable values of interface width (< 10 km; Abers 2005), whereas along the plate-wedge interface, shear and buoyancy forces are of comparable magnitude (< 20 MPa). Given that the strengths of typical slab-top minerals, including antigorite (e.g., Hilairet et al. 2007), exceed the strength of the interface required to facilitate subduction (< 2 MPa for interface widths of 1–5 km; England and Holland 1979), this simplified analysis indicates that: (1) exhumation by buoyancy forces alone during active subduction is unlikely along the plate-plate interface, and (2) that buoyancy-driven exhumation along the wedge-plate interface is plausible during subduction when shear stresses are < 10 MPa and channel widths are < 10 km.

Fig. 11

Force balance on the subduction interface. Plot shows the magnitude of buoyancy force ($h\Delta \rho gsin\theta$, where $h$ is interface width, $\Delta \rho$ is density contrast, $g$ is gravitational acceleration and $\theta$ is slab dip) for a range of channel widths and subduction angles and ($\Delta \rho$ = 300 kg/m³). Superimposed on the y-axis are estimated ranges of shear stress along the plate-plate (blue) and wedge-plate (red) interfaces. See text for discussion

Whether the Voltri HP rocks were exhumed from the wedge-plate interface is unclear, but inferences can be made from the fact that the maximum depth of thrust faulting earthquakes in active subduction zones is between 30 and 70 km (England 2018). Assuming that the maximum depth of thrust faulting earthquakes constrains the depth of the plate-plate-to-wedge-plate transition, it is plausible that some of the Voltri HP rocks were exhumed from the wedge-plate interface in response to low shear stresses.

The arguments developed in the preceding sections demonstrate that slab-top HP materials of the Voltri Unit were initially exhumed from > 70 km, potentially along the wedge-plate interface, then stored and refrigerated at depths of 30–70 km for > 10 Myr, while subduction was ongoing, before final exhumation on cessation of subduction. Several other studies provide supporting evidence for protracted forearc storage of HP rocks during active subduction. Bayet et al (2018) reported that a 4–5 km-thick package of clastic sediment, exposed in the southern Tianshan, underwent accretion and storage at ~ 80 km and ~ 540 °C for at least 2–5 million years. Eclogites from the Franciscan terrane preserve a record of ~ 7 Myr storage under blueschist facies conditions (Anczkiewicz et al. 2004; Mulcahy et al. 2009) and xenoliths from the Colorado Plateau yield zircon dates from 81 to 33 Ma, consistent with storage for ~ 50 Myr (Usui et al. 2003). Similar scenarios have also been proposed for protracted (> 40 Ma) storage of HP metamorphic rocks in Caribbean serpentinite mélanges (e.g., Blanco-Quintero et al. 2011; Krebs et al. 2008; Lázaro et al. 2009).

Conclusions

• 1. Phengitic white-mica hosted in bedrock metasediments and metagabbros of the Voltri Unit and detrital clasts of the adjacent Tertiary Piedmont Basin is zoned in Si-content. Mineral core domains in phengite from six samples exceed Si 3.5 c.p.f.u., whereas rims are characterized by significantly lower values, Si < 3.3 c.p.f.u., implying that phengite grains partially re-equilibrated during decompression from HP conditions.

• 2. PT pseudosection calculations for bedrock and clast Fe–Ti metagabbros yield remarkably similar conditions for garnet-rim equilibration: 23–25 kbar and 510–530 °C. Phengite cores equilibrated during the early stages of exhumation, ~ 20 and ~ 24 kbar. An impure quartzite bedrock sample preserves a record of lower-grade HP metamorphism with garnet-rim equilibration at 18–19 kbar and 450–470 °C. Core-to-rim celadonite contents in phengite corresponds to pressures between 19 and 12 kbar, respectively.

• 3. Interpreted as mixtures of high- and low-Si phengite, Rb–Sr phengite-whole-rock dates from the two eclogitic metagabbros constrain the timing of HP exhumation to between ~ 45 and ~ 50 Ma, marginally older than the age of HP phengite equilibration in the impure quartzite, 42.93 ± 2.65 Ma and a bedrock micaschist, 44.57 ± 1.68 Ma.

• 4. Exhumation-related recrystallization of HP phengite to low-Si phengite in two metasedimentary samples occurred between ~ 33 and ~ 30 Ma, contemporaneous with the onset of deposition in the TPB and consistent with previous ⁴⁰Ar−³⁹Ar constraints on the timing of greenschist metamorphism.

• 5. These PTt data collectively show that peak HP conditions, between ~ 18 and ~ 25 kbar were attained at different times across the Voltri Unit, between ~ 50 and ~ 40 Ma, consistent with the interpretation that the Voltri Unit comprises an assembly of discrete lithotectonic units that were juxtaposed prior to erosion and deposition in the TPB molasse basin.

• 6. Combined with existing garnet Sm–Nd and phengite ⁴⁰Ar–³⁹Ar mineral dates and reported metamorphic mineral textures, the Rb–Sr phengite PTt data reported here support a model in which individual sheets of HP material were detached from the downgoing plate, partially exhumed from peak pressures to 8–20 kbar, and stored for > 10 Myr within the Alpine orogenic wedge, until subduction ceased on arrival of the European continent into the subduction zone.