Zircon ages in granulite facies rocks: decoupling from geochemistry above 850 °C?

Granulite facies rocks frequently show a large spread in their zircon ages, the interpretation of which raises questions: Has the isotopic system been disturbed? By what process(es) and conditions did the alteration occur? Can the dates be regarded as real ages, reflecting several growth episodes? Furthermore, under some circumstances of (ultra-)high-temperature metamorphism, decoupling of zircon U–Pb dates from their trace element geochemistry has been reported. Understanding these processes is crucial to help interpret such dates in the context of the P–T history. Our study presents evidence for decoupling in zircon from the highest grade metapelites (> 850 °C) taken along a continuous high-temperature metamorphic field gradient in the Ivrea Zone (NW Italy). These rocks represent a well-characterised segment of Permian lower continental crust with a protracted high-temperature history. Cathodoluminescence images reveal that zircons in the mid-amphibolite facies preserve mainly detrital cores with narrow overgrowths. In the upper amphibolite and granulite facies, preserved detrital cores decrease and metamorphic zircon increases in quantity. Across all samples we document a sequence of four rim generations based on textures. U–Pb dates, Th/U ratios and Ti-in-zircon concentrations show an essentially continuous evolution with increasing metamorphic grade, except in the samples from the granulite facies, which display significant scatter in age and chemistry. We associate the observed decoupling of zircon systematics in high-grade non-metamict zircon with disturbance processes related to differences in behaviour of non-formula elements (i.e. Pb, Th, U, Ti) at high-temperature conditions, notably differences in compatibility within the crystal structure.


Introduction
Zircon is an accessory phase found in a large variety of rock types. While it is stable over a wide range of metamorphic conditions, modelling predicts growth mostly in restricted parts of a P-T path (e.g. Yakymchuk and Brown 2014;Kohn et al. 2015). For new metamorphic zircon to form in hightemperature (HT) metamorphic rocks, either the presence of a hydrous fluid or melt phase is required (e.g. Watson and Harrison 1983;Fraser et al. 1997;Tomaschek et al. 2003;Tichomirowa et al. 2005). In the absence of fluids, metamorphic zircon may form via solid-state (partial) recrystallisation (e.g. Schaltegger et al. 1999;Hoskin and Black 2000;Tomaschek et al. 2003;Rubatto et al. 2006) or by the breakdown of Zr-rich minerals (Fraser et al. 1997;Degeling et al. 2001;Bingen et al. 2004;Ewing et al. 2013;Pape et al. 2016).
Several studies on zircon from (ultra-)high-temperature (UHT) terrains show that decoupling of zircon ages from textures and/or geochemistry can occur (e.g. McFarlane et al. 2005;Flowers et al. 2010;Kusiak et al. 2013a, b;Whitehouse et al. 2014). The conditions and mechanisms that lead to the observed decoupling are far from being fully understood: metamictisation can contribute but does not appear to be a requirement (Whitehouse et al. 2014), and nm-sized Pb (and trace elements) clusters or patchy Pb distributions have been reported (Kusiak et al. 2013a, b;Whitehouse et al. 2014;Regis et al. 2017). So far, decoupling has been studied in zircon of Paleo-Proterozoic age only. Here we report decoupling in relatively young (~ 280 Ma) crystalline zircon from high-grade samples (> 850 °C) in the Ivrea Zone.
The Ivrea Zone (Fig. 1) is regarded as a section through the mid-to lower continental crust preserving a HT metamorphic imprint established during the late Palaeozoic (e.g. Berckhemer 1968;Mehnert 1975;Fountain 1976;Quick et al. 1995). Contrary to interpretations of early studies (e.g. Fountain 1989;Voshage et al. 1990;Sinigoi et al. 1991;Quick et al. 1994;Schnetger 1994;Henk et al. 1997), the now-accepted view is that the intrusion of mafic magmas into the lower crust was not the main cause for HT metamorphism (e.g. Barboza et al. 1999;Redler et al. 2012). There is no consensus, however, whether the regional HT metamorphism was first established during Variscan (collisional) metamorphism (e.g. Zingg et al. 1990;Schmid 1993) or caused by late Palaeozoic crustal thinning (e.g. Barboza and Bergantz 2000).
This study investigates the age and trace element variability in HT metamorphic zircon, with an emphasis on unravelling the origin of the large scatter of U-Pb dates and their decoupling from textures and trace element geochemistry in zircon from the highest grade (> 850 °C) granulite facies part of the Ivrea Zone. The continuous metamorphic field gradient from mid-amphibolite (650 °C) to granulite facies conditions (950 °C) exposed in the Ivrea Zone allows us to systematically study the evolution of zircon under increasing high-temperature conditions. Twelve samples collected along the metamorphic field gradient and four samples close to the contact of the mafic intrusion complex ( Fig. 1) were investigated in detail. We analysed zircon core-rim proportions, internal textures, U-Pb ages, and trace element characteristics: Th/U ratios, REE, and Ti concentration.

Geological setting
The Ivrea Zone is a slice of Permian intermediate to lower continental crust of the Adriatic margin, today tilted by 90° and exposed in the Southern Alps of northern Italy (e.g. Zingg et al. 1990;Handy et al. 1999). The slice (ca. 120 by 14 km) exposes magmatic (Mafic Complex) and metamorphic rocks (Kinzigite Formation). To the northwest, the Ivrea Zone is bounded by the Insubric Line that separates it from units of the Western and Central Alps (Schmid et al. 1987). To the southeast, the Ivrea Zone is tectonically separated from an upper crustal block, the Serie dei Laghi/ Strona-Ceneri Zone (Handy 1987(Handy , 1999Boriani et al. 1990;Mulch et al. 2002a, b).
The Mafic Complex is an intrusive body consisting mainly of gabbro, norite and diorite with subordinate granite and tonalite (e.g. Quick et al. 2003;Rivalenti et al. 1975Rivalenti et al. , 1981Sinigoi et al. 1996). It was formed by magmatic underplating during the Permian extensional regime between ~ 290 and 280 Ma (e.g. Peressini et al. 2007;Sinigoi et al. 2011). The main intrusive phase has been dated on zircon by Peressini et al. (2007) at 288 ± 4 Ma, with locally earlier magmatic activity at ~ 295 Ma (Sinigoi et al. 2011) and 314 ± 5 Ma (Klötzli et al. 2014). In the northern part of the Ivrea Zone, the Finero mafic-ultramafic body is characterised by a range of Permian U-Pb zircon dates  as well as younger zircon intrusion ages of 232 ± 3 Ma and late fluid-related recrystallisation between ~ 220 and 200 Ma (Zanetti et al. 2013;Schaltegger et al. 2015;Langone et al. 2017).
The Kinzigite Formation mainly consists of metasedimentary rocks interlayered with metabasic rocks (e.g. Bertolani 1968). A continuous metamorphic field gradient from mid-amphibolite to granulite facies conditions is best exposed along Val Strona di Omegna, in the central Ivrea Zone (e.g. Zingg 1978;Schmid 1993;Henk et al. 1997;Redler et al. 2012;Kunz et al. 2014). The metamorphic field gradient is thought to reflect protracted heating during the late Palaeozoic, with a HT period in the Carboniferous/ Permian (Schmid 1993;Henk et al. 1997). Owing to a pervasive thermal overprint during Permian regional high-grade metamorphism little information is preserved in the Ivrea Zone of its previous history. Polyphase deformation predating the Permian mafic intrusion (Rutter et al. 2007) and the presence of locally preserved kyanite relics (Bertolani 1959;Capedri 1971;Boriani and Sacchi 1973;Handy 1986) have been related to collision during the Variscan orogeny.
Based on phase equilibrium modelling in Val Strona di Omegna, P-T estimates for peak metamorphic conditions range from ~ 3.5-6.5 kbar at 650-730 °C in mid-amphibolite facies samples to 10-12 kbar at > 900 °C in the highest grade samples (Redler et al. 2012). Temperatures of 900-950 °C for peak conditions in the granulite facies have been derived from Zr-in-rutile thermometry (Luvizotto and Zack 2009;Ewing et al. 2013). Several studies have presented field and petrological evidence that the Mafic Complex was not the heat source for regional granulite facies metamorphism (Barboza et al. 1999;Barboza and Bergantz 2000;Redler et al. 2012). A local contact aureole of 1-3 km around the mafic intrusion is observed in Val Sesia (Fig. 1), overprinting earlier high-grade mineral assemblages (Barboza et al. 1999;Redler et al. 2012). Furthermore, the northeastern contact of the intrusion crosscuts earlier metamorphic isograds in the surrounding country rocks (e.g. Zingg et al. 1990;Barboza et al. 1999), suggesting a previous regional metamorphic event, possibly related to Variscan crustal thickening (Zingg et al. 1990;Schmid 1993). Geochronological studies show that the earliest phase of regional metamorphism is coeval with the first intrusive phase of the Mafic Complex at c. 315 Ma (Ewing et al. 2013;Klötzli et al. 2014). However, the range of HT metamorphic zircon dates (316-258 Ma; e.g. Vavra et al. 1999;Ewing et al. 2013) exceeds the range  Zingg (1980), Bigi and Carrozzo (1990), and Rutter et al. (2007) ? of magmatic zircon (315-280 Ma, e.g. Peressini et al. 2007;Klötzli et al. 2014) by about 20 Myr. Previous studies by Vavra et al. (1999) and Ewing et al. (2013) interpreted the youngest zircon ages as possible resetting pre-existing Variscan (?) regional metamorphic ages. However, the driving mechanism behind this resetting process remains unclear, and we note that the young generation of ages is the most abundant fraction of the zircon age population.

Sample material
Previous studies have shown how bulk rock composition can control the amount and possible timing of zircon crystallisation/dissolution (e.g. Rubatto et al. 2006;Kelsey et al. 2008;Kelsey and Powell 2011;Yakymchuk and Brown 2014;Kohn et al. 2015). Therefore, to understand the evolution of zircon growth along the continuous metamorphic field gradient in Val Strona di Omegna and avoid misinterpretations due to compositional differences, we collected nine aluminous metapelitic samples (Figs. 1, 2). The samples were collected from the same outcrops or in close vicinity of the sample locations of the metapelites modelled by Redler et al. (2012Redler et al. ( , 2013. This allowed us to infer approximate P-T conditions for our samples (Table 1). Two samples (IZ 419 and IZ 420) did not yield zircon during mineral separation and, therefore, are not further discussed. For comparison, three metapsammite samples (Fig. 2) from low-to highgrade were also sampled and analysed. In Val Sesia, where the intrusion of the Mafic Complex caused a narrow contact aureole, two samples were collected directly at the contact and two at some distances (~ 500 m) from the intrusive contact ( Fig. 1).
Seven metapelite samples were collected in Val Strona di Omegna (Fig. 2, S1 Online Resource 1). Along the metamorphic field gradient the metapelites evolve from foliated micaschists to migmatites (Fig. S1 Online Resource 1). Sample IZ 407 originates from the Rosarolo shear zone and shows a mylonitic foliation. The mineral assemblages ( quartz, plagioclase, garnet, K-feldspar and minor biotite. Accessory phases in the metapsammites are the same as in the metapelites. Zircon in all samples occurs abundantly in the matrix, in biotite, and sillimanite as well as occasionally in garnet. In samples IZ 405, IZ 406a and IZ 406b and IZ 407, rutile is present as fresh grains (up to 500 µm diameter) included in garnet, sillimanite, and in the matrix.

Methods
After sample disintegration using high-voltage pulse power fragmentation in a Selfrag lab system, zircon grains were separated by magnetic and heavy liquid mineral separation. Substantial amounts of good quality (idiomorphic, transparent) zircon were separated from the 250-64 µm grain size fraction of all but two samples (IZ 419 and IZ 420). Zircon grains were handpicked, mounted into acryl discs, and polished to expose their near-equatorial grain sections. To analyse internal textures and target potential locations for LA-ICP-MS spots, cathodoluminescence (CL) images were produced using a VPSE detector on the ZEISS EVO 50 scanning electron microscope at the Institute of Geological Science in Bern.
U-Pb ages, Ti and REE concentrations were measured by LA-ICP-MS at the University of Bern with a GeoLas-Pro 193 nm ArF Excimer laser system (Lambda Physik) and an Elan DRC-e quadrupole mass spectrometer (Perkin Elmer). Ablation was achieved using an energy density of 2.5 J/cm 2 with a repetition rate of 9 Hz using a spot size of 44 or 32 µm. Aerosol transport was attained with a mixture of He (1 l/min) and H 2 (0.08 l/min). Instrument tuning for U-Pb measurements was optimised for heavy masses; oxide production (ThO + /Th + ) was always maintained below 0.5%. For trace element measurements the instrument was optimised to achieve equal sensitivity over all measured masses. GJ-1 ( 206 Pb/ 238 U 602.1 ± 0.7 Ma; Jackson et al. 2004) was used as primary standard for U-Pb ratios and NIST SRM 612 was used for quantification of element concentrations (U and Th). Stoichiometric SiO 2 = 32.8 wt% in zircon was used as internal standard. Plešovice ( 206 Pb/ 238 U 337.13 ± 0.37 Ma; Sláma et al. 2008) was used as a secondary standard for accuracy and long-term reproducibility control. Population ages for Plešovice were within error of the accepted reference age in all sessions ( Fig. S2 Online Resource 2). The long-term reproducibility of Plešovice 206 Pb/ 238 U dates was 1.4% (2 S.D., standard deviation) for a period of 17 months and on average ~ 2.0% (2 S.D.) for individual analytical days. U-Pb data were processed offline using the software Iolite 2.5 (Paton et al. 2010(Paton et al. , 2011 and the data reduction scheme Visual ages (Petrus and Kamber 2012). Error propagation was carried out using Iolite (Paton et al. 2010). Only concordant (i.e. error ellipse intersects Concordia) dates are reported ( Fig. S3 Online Resource 2), and no common Pb correction was applied. All U-Pb dates reported in this publication are 206 Pb/ 238 U dates. The term 'date' in this study refers to an individual U-Pb zircon spot analysis, while the term 'age' is used for a number of pooled dates, which are regarded as having geological significance. Errors on individual dates and isotope ratios are reported as 2 standard error (S.E.). U-Pb and trace element LA-ICP-MS analyses measure a 3D volume, while textural characterisation and spot selection were based on 2D CL-images. Mixing by accidentally drilling into relic cores and/or multiple rim generations at depth is a general concern. To minimise this risk we only targeted growth zones wide enough to fit our laser spot without the risk of drilling into other growth zones visible in CL-images. Furthermore, during data reduction we monitored and carefully evaluated each signal individually, using only flat signal parts and avoiding inclusions. Analyses that did not fulfil this criterion or showed obvious signs of mixing have been discarded.
Data reduction for trace element concentration was done with the MATLAB-based program SILLS 1.3.2 (Guillong et al. 2008). SRM 612 (primary standard) and SRM 610 (secondary standard) were used to standardise trace element data (reference values from Pearce et al. 1997 were used). Accuracy for trace elements with relatively high concentration is within ~ 5%, while trace elements close to detection limit have accuracy of ~ 10%.

Zircon textures
To study the internal textures of zircon by CL-images and make representative descriptions for each sample, we aimed to have at least 100 individual zircon grains per sample (Table 1); overall our study investigated 2576 zircon grains. Based on our observation, four types of core-rim proportions were distinguished in metapelitic zircon from Val Strona di Omegna, as schematically shown in Fig. 3a: Type1 is dominated by a detrital core with a very narrow or no metamorphic rim; Type2 shows a prominent detrital core with a variable (10-60 µm) metamorphic rim; Type3 is dominated by the metamorphic rim with a small or relic core; Type4 shows metamorphic zircon only, with no core visible.
Within metamorphic zircon, a sequence of texturally different metamorphic rims is distinguished based on their CL-appearance (Fig. 3b): from internal to external: Rim1 is bright in CL, has an irregular shape and contains many inclusions. Rim2 is dark in CL, the internal texture is either uniform or shows sector zoning or growth banding or fir-tree zoning. Rim3 displays oscillatory zoning and an intermediate grey CL-colour, and Rim4 is bright and featureless in CL. Individual zircon grains rarely show all four types of rims, usually various subsets of these rims are observed (Fig. 4).
In mid-amphibolite facies metapelites, the examined zircon grains show an elongate or stubby crystal habit, commonly reflecting the shape of a detrital core. Towards the upper amphibolite and granulite facies, grains commonly show a 'soccer ball' shape, roundish with many crystal facets. In metapelite samples from Val Strona di Omegna, the proportion of preserved detrital or inherited core to metamorphic rim decreases with increasing metamorphic grade (Fig. 4). In the mid-amphibolite facies samples, Type1 and Type2 dominate the zircon population. In the upper amphibolite facies, Type2 and Type3 are most abundant, while in the granulite facies samples Type3 and Type4 predominate. With increasing grade, from mid-amphibolite facies towards upper amphibolite facies, Rim1 increases in width, reaching its maximum thickness (up to 30 µm, Fig. 4) in sample IZ 422. Towards higher grade, Rim1 becomes less prominent, and generally only a thin (< 5 µm) relic remains (Fig. 4, IZ 407, IZ 406 a, b, IZ 405). Overall, Rim2 is the most abundant rim and it is observed across the mid-amphibolite to granulite facies. In the mid-amphibolite facies, Rim2 is narrow (< 5 µm), whereas it starts to show significant width (> 30 µm) in the upper amphibolite facies (Fig. 4, IZ422). At upper amphibolite facies conditions, Rim2 frequently is rich in sillimanite inclusions. Rim3 (Fig. 4, IZ 406a, b) occurs only in leucosome-rich samples of the granulite facies, while the occurrence of Rim4 (1-50 µm) is independent of metamorphic grade, though it is not present in every grain (Fig. 4).
Although abundant zircon grains were recovered from the metapsammites, the overall quality of the crystals was poor (i.e. broken grains, internal cracks or other forms of disturbance, Fig. 5). Independent of metamorphic grade, zircon grains are mostly dominated by detrital cores with narrow rims (< 5-10 µm). The granulite facies sample (IZ 424) shows substantial metamorphic rim growth (up to 50 µm). The same four rim types as for the metapelites could be identified (Fig. 5). Rim1 is present predominantly in the amphibolite facies (IZ 417), Rim2 is the most abundant rim type, Rim3 occurs only in the granulite facies (IZ 424) and Rim4 occurs independent of metamorphic grade (Fig. 5).

U-Pb zircon geochronology
An overview of U-Pb dates for each sample is shown in Table 2, individual analyses for each sample are listed in Online Resource 4 Tables S1 and S4. Both rock types (metapelite and metapsammite) will be described and discussed together according to their metamorphic grade. In samples from Val Strona di Omegna we analysed 169 detrital cores and 386 metamorphic rims (Table 2). Overall, the detrital core dates range from ~ 2640 to 352 Ma (Table 2), whereas the metamorphic rim analyses cover a range from 316 to 211 Ma ( Granulite facies: Samples from the granulite facies show a wide spread in U-Pb dates (Fig. 6c). Zircon rims in the metapsammite IZ 424 yielded a range of dates from 303 Ma to 256 Ma. Two distinct portions of metapelite IZ 406 were analysed separately; the melanosome (IZ 406a) and the leucosome (IZ 406b) show a similarly large spread of dates (Fig. 6c). Sample IZ 406a yields a range of dates from 280 Ma to 250 Ma while sample IZ 406b yields a range between ~ 300 and ~ 250 Ma. The highest grade sample IZ 405 shows a broad peak in the probability density plot (Fig. 6c), with U-Pb dates ranging from ~ 290 to 240 Ma. U-Pb dates were obtained for Rim2, Rim3 and Rim4; however, no relationship between rim type and metamorphic dates has emerged ( Fig. 8; S6a, b Online Resource 3; Table S1 Online Resource 4).

Trace element geochemistry of metamorphic zircon rims
Th/U ratios, Ti-in-zircon concentration as well as other trace elements (Hf, P, Y, REE) were measured for the same growth zones, close to the spots used for U-Pb dates. Table 2 gives an overview of the Th, U, Ti concentrations and the Th/U ratio for each sample. Individual analyses of trace elements are listed in Table S3 and S5 Online Resource 4. Trace element data are presented for metamorphic zircon. Since no systematic differences were evident between the data from different rim texture types defined above, the trace element results are described together below.
Metamorphic zircon rims in mid-amphibolite facies samples (IZ 418, IZ 422) are characterised by Th/U < 0.01 (Fig. 7a). In the upper amphibolite facies, Th/U ratios range from ~ 0.01 to 0.1 in most samples (IZ 409, IZ 407). Granulite facies samples show mostly high Th/U ratios between 0.1 and 1.0, but in samples IZ 405 the Th/U ratios cover a wide range (0.002-0.847, Table 2). In samples with scattered U-Pb dates we observe no correlation between Th/U ratios and U-Pb dates (Fig. 8a), except in sample IZ 407 where we see an increase in Th/U ratio with decreasing dates (Fig. 7c). Ti concentrations in metamorphic zircon rims are low and close to detection limit (~ 0.5-3 ppm) in the mid-amphibolite facies samples (Fig. 7d). In the upper amphibolite and granulite facies, minimum values are around 2 ppm (close to detection limit), while the maximum concentration (up to 38 ppm) and the scatter in the data increase towards higher grade ( Fig. 7b; Table 2). Metamorphic zircon in the granulite facies samples yield a large spread in Ti concentration, with no correlation to their respective U-Pb dates (Figs. 7d,  8b). To calculate Ti-in-zircon temperatures, the calibration of Watson et al. (2006) was used, assuming a Ti activity of a TiO2 =1 (see Discussion for more details). Ti-in-zircon temperatures ( Fig. 7b; Table 2) in the mid-amphibolite facies range from 540 to 620 °C. In the upper amphibolite facies, zircon yielded temperatures in the range of 620-780 °C, and in the granulite facies of 630-880 °C. While we observed an increase of scatter in Th/U ratios and in Ti concentration for granulite facies zircon, there is a correlation between these two variables (Fig. 8c, d). Generally speaking, independent of metamorphic grade of the sample, all zircon with low Ti concentration also have low(er) Th/U ratios (Fig. 8c, d). Metamorphic zircon rims are characterised by REE patterns depleted in LREE with respect to HREE; chondritenormalised (McDonough and Sun 1995) REE plots are shown in Fig. S4 Online Resource 3. A strong negative Eu anomaly 1 and slightly positive HREE patterns are seen in the mid-amphibolite facies zircon samples. In the upper amphibolite facies, the Eu anomaly (Eu/Eu* 0.03-0.27) is more variable, while HREE slopes are overall flat in sample IZ 409 (Lu N /Gd N 0.7-2.8) and flat to slightly negative in sample IZ 407 (Lu N /Gd N 0.1-5.2). In sample IZ 407 we observe a decrease in Hf and Yb concentration from older to younger ages ( Fig. S5 Online Resource 3). The range in Eu anomaly in granulite facies samples is similar as at the upper amphibolite facies (Eu/Eu* 0.03-0.29), but the HREE patterns show much variability in the granulite facies, ranging from negative to positive (Lu N /Gd N 0.03-10.17). Sample IZ 405 shows large variability in REE composition, but a significant proportion of analyses show low HREE concentrations that are not observed in other samples.

Contact zone with Mafic Complex (Val Sesia)
In Val Sesia a narrow sliver of the Kinzigite Formation is exposed next to the Mafic Complex (Fig. 1), far less extensive here than in Val Strona di Omegna. The metamorphic grade of the country rocks in Val Sesia is comparable with the area around the Kfs/Ms isograd in Val Strona di Omegna (Redler et al. 2012). Samples were collected at various distances from the intrusive contact to the Mafic Complex (see Fig. 9a). Sample IZ 411 and IZ 412 were selected next to the intrusive contact, while samples IZ 415 and IZ 416 were taken some 500 m away from that contact (Fig. 9a). As in Val Strona di Omegna, a limited range of bulk rock compositions (metapelite, metapsammite, leucosome) was studied  to investigate the bulk rock composition influence in the behaviour of zircon.

Zircon textures
The metapelitic samples IZ 415 and IZ 416 from Val Sesia show internal zircon textures similar to those from the upper amphibolite facies (IZ 422 and IZ 409) in Val Strona (Fig. 9b, c.f. Fig. 4). In sample IZ 416 well-developed Rim1 and Rim2 overgrow detrital cores, while in IZ 415 Rim1 is rarely preserved and overgrown by up to 50 µm Rim2 (Fig. 9b). In both samples detrital cores show features of partial resorption. Sample IZ 412 is a garnet-rich leucosome that retains abundant evidence of a detrital origin of many zircon cores. The rim is either uniformly dark in CL or shows oscillatory zoning. Several grains show evidence for inward migrating reaction fronts (see Fig. 9b). Zircons in metapsammite sample IZ 411 are dominated by detrital cores with a narrow metamorphic Rim2 (Fig. 9b). Some metamorphic rims show zoning that curves inwards, and detrital cores show evidence of cloudy and/or turbulent overprinting of previously developed textures (Fig. 9b).

CL-images of zircon textures in metapelites from Val
Strona di Omegna show that detrital grains/cores were progressively resorbed with increasing metamorphic grade, whereas new metamorphic zircon grew (Fig. 4). Although Rim2 at upper amphibolite facies conditions shows discrete age peaks and uniform trace element compositions (Figs. 6,7,8), towards higher grades the scatter in the zircon U-Pb dates increases significantly, as do the Th/U ratio and Ti concentrations (Figs. 7, 8). The chemical complexity of the zircons in individual samples, especially those in the granulite facies, indicates additional processes compared to those operating at lower metamorphic grade.

Zircon textures
Core-rim proportions in zircon from metapelites evolve continuously from mid-amphibolite facies (dominated by detrital core) to granulite facies (dominated by metamorphic rim) (Fig. 4). Previous studies have related such observations to the increasing degree of partial melting leading to resorption of detrital grains into the melt, which ultimately resulted in the growth of a thicker rim (e.g. Rubatto et al. 2001;Kelsey et al. 2008;Yakymchuk and Brown 2014). According to our observations, we suggest that the four types of metamorphic rim textures recognised in zircon (Fig. 4) can be explained by different growth environments and/or mechanisms: • Rim1 and detrital cores, with bright-CL intensity, numerous inclusions, and cloudy zoning have the characteristics of zircon grown by a fluid-assisted dissolution-reprecipitation process (e.g. Tomaschek et al. 2003;Geisler et al. 2007;Vonlanthen et al. 2012;Rubatto 2017 (Fig. 4). c, core; r1, Rim1; r2, Rim2; r3, Rim3; r4, Rim4. c Probability density plot of U-Pb zircon ages from Val Sesia for Rim1 could not be obtained due to their narrowness, generally very low U-contents and abundant inclusions. • Rim2 has textural features typical for anatectic growth at HT conditions in the presence of a melt, i.e. dark-CL emission (U-rich), uniform in CL-appearance or sector zoned (e.g. Vavra et al. 1996Vavra et al. , 1999Schaltegger et al. 1999;Rubatto and Gebauer 2000;Corfu et al. 2003). This rim type is the most abundant and occurs in amphibolite facies samples as well as samples of restite or melanosome at granulite facies. • Rim3 only occurs in leucosome-rich granulite facies samples (Fig. 4, IZ 406 a, b). Its fine-scale oscillatory zoning or planar growth banding suggests crystallisation from an anatectic melt (e.g. Hoskin 2000; Corfu et al. 2003). • Rim4 shows features typical for late-stage fluid-induced recrystallisation, as reported by Vavra et al. (1999). The enhanced CL-emission in this rim type may be related to a purification process (trace element rejection) in the zircon crystal structure (e.g. Schaltegger et al. 1999;Vonlanthen et al. 2012).
Zircon from metapsammitic samples shows much less change with metamorphic grade. The zircon textures (Fig. 5; e.g. blurred zoning, 'ghost' zoning, convolute zoning, or inwards curved recrystallisation fronts) are typical features for solid-state recrystallisation, as described by Hoskin and Black (2000). Metapelites and metapsammites experienced the same P-T path (Redler et al. 2012), and both contained abundant detrital zircon in their precursor, but during metamorphism zircon evolved differently. In metapsammites (Fig. 5), resorption of cores and growth of new metamorphic rims were limited, as compared to metapelites (Fig. 4), indicating that the presence of a fluid or melt is a decisive factor, as shown by other studies (e.g. Rubatto et al. 2006;Kelsey et al. 2008).

Zircon geochronology and geochemistry
The U-Pb zircon dates from metasediments range between ~ 316 and 240 Ma, a spread of > 60 Myr. In the midamphibolite facies sample, metamorphic rims were too thin (5-30 µm) to be dated by LA-ICP-MS, hence it was not possible to confirm their age. The two upper amphibolite facies samples in Val Strona di Omegna and the samples from Val Sesia that represent similar upper amphibolite facies grade, each have a single, well-defined age generation, obtained in Rim2, with homogenous trace element patterns (Figs. 6, 7, 8, S4-S6a Online Resource 3).
Close to the upper amphibolite/granulite facies transition, two age generations are present, and these are distinguished by Th/U ratios (Fig. 7c) as well. In the granulite facies samples, U-Pb zircon dates scatter broadly over > 40 Myr (Fig. 6). Similarly broad age ranges in granulite facies zircon population have been described in several studies from other terrains (e.g. Rubatto et al. 2001;Diener et al. 2013). Despite the scatter of dates in the granulite facies samples, a trend (Fig. 6) towards on average younger ages (c. 260 Ma) is visible in our samples, compared to those in the upper amphibolite facies. Younger ages in deeper crustal levels suggest either (1) later onset of zircon crystallisation or (2) disturbance of zircon at high-grade conditions. Given the broad scatter in dates, it seems unlikely that the data can be simply explained by later onset of zircon crystallisation. As suggested by other studies (e.g. Taylor et al. 2016), a combination of disturbance and episodic or continuous growth could be a more plausible reason for the scatter. In the present case, Pb loss could result in a spread of dates along Concordia instead of apparently discordant ellipses due to the relatively young metamorphic ages. While we cannot completely rule out this possibility, we see no correlation between U (+ Th) and U-Pb dates in any of our samples; therefore, we do not consider this to have a major effect on our data or the observed decoupling.
Except for five outliers, mid-and upper amphibolite facies samples generally have Th/U ratios < 0.1 (Fig. 7a), while granulite facies samples mostly show Th/U ratios > 0.1 (Fig. 7a). This reflects well-documented trends, i.e. Th/U ratios < 0.1 in metamorphic zircon (Rubatto and Gebauer 2000). However, in some granulite facies (> 900 °C) and (U) HT terranes (e.g. Rogaland: Möller et al. 2003;Napier Complex: Kelly and Harley 2005; Eastern Ghats belt: Korhonen et al. 2013), metamorphic zircon with Th/U ratios > 0.1 are well known. The trend towards higher Th/U ratios in zircon with increasing metamorphic grade reflects both an increase in Th and a decrease in U concentrations. Th and U availability are controlled by the abundance of other major or accessory phases and their partitioning with zircon. However, our present data are from zircon grain separates, so the textural context required to further investigate this is lacking.
Ti-in-zircon temperatures (540-640 °C) for amphibolite facies samples (IZ 418, IZ 422) are considered minimum values, due to lowered TiO 2 activity in the absence of rutile (e.g. Watson et al. 2006). In the other studied samples, rutile is abundantly present as large grains (~ 30-500 µm), and temperatures are considered to relate to zircon growth between 620 and 880 °C. Ti-in-zircon temperatures ( Table 2) are clearly lower than the inferred peak temperature (Table 1) for the Ivrea Zone (up to 950 °C, Luvizotto and Zack 2009;Ewing et al. 2013), such discrepancy has been observed in other studies as well (e.g. Baldwin et al. 2007;Ewing et al. 2013). For cases with such discrepancy it has been proposed that Ti-in-zircon temperatures do represent zircon growth, which occurred below peak temperature (e.g. Roberts and Finger 1997).
To explain the observed increase in maximum Ti concentration with increasing metamorphic grade and the increased scatter in granulite facies samples (Fig. 7b, d), we consider (1) zircon growth in a non-buffered system, or (2) zircon growth over a range of temperatures. Looking at these possibilities in turn: (1) in quartz-bearing rocks, such as metapelites, SiO 2 was surely buffered (a SiO2 =1). While rutile is present in the mineral assemblages of all but two samples, it is possible that the TiO 2 activity (relative to rutile) was lower at the time and/or in the reaction volume in which zircon grew; if so, the calculated temperatures represent minima.
(2) If the system was buffered at a i =1 for both SiO 2 and TiO 2 , the calculated Ti-temperatures delimit the interval in which zircon actually grew. This may imply zircon growth at different stages along the same metamorphic path (e.g. prograde, peak and retrograde). The range of Ti-in-zircon concentrations in our highest grade samples is remarkably large (4-38 ppm; 680-880 °C); however, there is no correlation between Ti concentration and U-Pb ages, limiting, in our samples, the use of the Ti-in-zircon thermometer as a petrochronological tool. Furthermore, heterogeneous Ti distribution at sub-micron scale (e.g. Hofmann et al. 2009;Kusiak et al. 2013a, b) may cause a spread in the apparent Ti-in-zircon temperatures but, given the small extent (< 1 µm) of such patches, it remains unclear to what degree they might influence our trace element data (analyses made with a spot size of 32 µm).
In summary, our data show that in granulite facies zircon there is no correlation between U-Pb dates and internal textures, Th/U ratio, or Ti-in-zircon concentration (Fig. 8,  S6a, b Online Resource 3); however, a positive correlation between Th/U ratio and Ti-in-zircon concentration has been identified (Fig. 8a, b). This correlation is observed when looking at all metapelitic samples from Val Strona di Omegna as well as in sample IZ 405 where the biggest spread in Th/U ratios and Ti concentration is observed ( Fig. 8; S6a, b Online Resource 3). This possibly indicates that zircon grew at different stages along the P-T path of IZ 405, but owing to the decoupling of the U-Pb ages from Th/U ratios and Ti concentration such an interpretation is uncertain.

Heterogeneities between and within individual samples
Granulite facies samples are characterised by younger ages, higher Th/U ratios and Ti-in-zircon concentration, matched with a large scatter in the data (Figs. 6, 7). What caused the age difference between zircon in upper amphibolite and granulite facies samples? More specifically, how does the difference in age relate to the difference in metamorphic grade, and why are the data so scattered in the granulite facies? Did granulite facies samples experience several phases of zircon growth, while upper amphibolite facies samples only experienced one? Is zircon (more) affected by disturbance at granulite facies conditions?
We suggest that possible explanations for the scattered granulite facies zircon data fall into two main categories: (1) prolonged or episodic growth of zircon, and (2) isotopic or geochemical disturbance of (primary) zircon during a protracted HT history (post-crystallisation).
• In granulite facies samples we observe no trends between U-Pb dates, Th/U ratios, and Ti-in-zircon temperatures (Fig. 7c, d). However, there is a trend between Th/U ratios and Ti concentration, indicating zircon growth at different stages. While we cannot rule out that some of the observations in the granulite facies zircon population are the result of several growth episodes or of protracted growth during the prolonged HT metamorphism, such evidence has been obscured by subsequent processes that affected zircon or may be obscured due to analytical limitations in our data. • Isotopic and/or geochemical disturbance of zircon by various mechanisms-at times occurring in combination-has been suggested by other authors (e.g. Vavra et al. 1999;Ewing et al. 2013). Previous studies have found that amphibolite facies samples are less prone to disturbance than granulite facies samples, but no particular process has been accepted to account for this phenomenon (e.g. Schiøtte et al. 1989;Williams 1992;Vavra et al. 1999). In a previous study Vavra et al. (1999) interpreted the spread of granulite facies zircon dates in the Ivrea Zone as the result of zoningcontrolled alteration (ZCA). ZCA can cause rejuvenation in a process that preferentially occurs along primary textural features such as sector boundaries, inclusions or growth banding. While some grains in our granulite facies sample do show textures similar to those described by Vavra et al. (1999), we found no correlation between younger dates and these textures.
Furthermore, the strongest heterogeneity is evident in the highest grade sample IZ 405 (Fig. 10). We observe grains that show the same date (262 ± 5 and 262 ± 9 Ma), but clearly different textures (oscillatory zoning and sector zoning, Fig. 10a), as well as the reverse, i.e. the same texture (fir-tree zoning) but different dates (265 ± 7 Ma and 242 ± 6 Ma, Fig. 10b). Similar observations also pertain to the date and Ti concentration (Fig. 10c, d), and to the Th/U ratio (Fig. 10e, f). These observations clearly indicate decoupling of zircon chemical and isotopic signatures in the granulite facies samples from Val Strona di Omegna.
Based on our data, we exclude several potential processes as the cause for the observed decoupling of zircon dates from trace element compositions: 1. Given the low U concentration (< 1000 ppm) in granulite facies zircon and relatively short accumulation time (< 300 Ma), metamictisation and annealing of metamict domains are not among the possible disturbance mechanisms. In fact, no textures characteristic of metamict zircon-such as suppressed CL (Nasdala et al. 2002)have been observed in any of our samples. The average α-doses calculated for zircons from our study (see Table S1 Online Resource 4) are 0.4 × 10 15 α/mg, clearly below the first percolation point (Stage 1: 2.2 × 10 15 α/ mg; Pidgeon 2014) of radiation damage. We also do not observe any correlation between U-Pb dates and α-dose (see Fig. S7 Online Resource 3). 2. Fluid alteration can be excluded for Rim2 and Rim 3 in the majority of our samples from Val Strona di Omegna, since characteristic CL-textures are lacking, such as bright-CL domains, inward-penetrating reaction fronts or convolute zoning (e.g. Vonlanthen et al. 2012). Rim1, Rim4, and some detrital cores may have been affected by fluid alteration, based on their CL-appearance, but apart from that textural evidence, the respective geochronological and geochemical data are inconclusive. The majority of our data showing decoupling were obtained in Rim2 or Rim3, where CL-textures are not disturbed. 3. Crystal plastic deformation is unlikely to be responsible for the bulk of our observations. Element redistribution under dry conditions without a chemical or structural sink (Piazolo et al. 2016) usually occur on the sub-micrometre scale (e.g. Valley et al. 2014) and are unlikely to have a dominant influence on our LA-ICP-MS data obtained with spot sizes of 44 to 32 µm. Sample IZ 407, taken inside a shear zone, is the only sample for which we observe a trend between U-Pb dates and geochemical data ( Fig. 7 and Fig. S5 Online Resource 3).
The more intense thermal imprint (higher T and longer duration) experienced by the granulite facies samples remains a most likely factor responsible for the observed decoupling. Similar propositions have been put forward by other studies that reported decoupling of zircon characteristics (e.g. Flowers et al. 2010). Decoupling may be explained by different behaviour and retention of certain elements in the zircon structure. In particular, the contrasting fit and diffusion behaviour of U and Pb in the zircon crystal lattice has been proposed as a possible cause for partial resetting of U-Pb ages (e.g. Cherniak and Watson 2003;Hoskin and Schaltegger 2003;Harley et al. 2007;Rubatto 2017). Rubatto and Gebauer (2000) reported that internal zircon textures shown by CL-images are mostly caused by trace elements such as Dy, Gd, Tb and U, so the preservation of CL-structures does not indicate preservation of the U-Pb growth age (Cherniak and Watson 2000).
Experimental results (Cherniak and Watson 2000) show that at ca. 750 °C it would take several billion years for a crystalline zircon with 10 µm radius to lose 1% of its lead. However, at 900 °C the same size zircon can lose more than 25% of its Pb in 1 million years or 50% in 10 million years. The highest grade samples in the Ivrea Zone experienced temperatures above 900 °C, the duration of these HT conditions remains uncertain despite many studies. While it seems possible that volume diffusion of Pb could be a cause for the observed decoupling, it would take further investigations (e.g. systematic in situ Pb profiling around zircon grains that show decoupling) to confirm its effectiveness on the observed length and timescale.
Other trace elements such as Ti or Th are considered similarly compatible in zircon as U, i.e. more unlikely to be disturbed than Pb (e.g. Hoskin and Schaltegger 2003). Patchy Pb and Ti distribution has been reported from (U) HT zircon, but in the case of Ti it remains unclear how common this is in (U)HT zircon and whether it is a primary or secondary feature (Kusiak et al. 2013a, b;Valley et al. 2014;Whitehouse et al. 2014). In the case of Pb, a study by Valley et al. (2014) showed that in their case, at the micrometre scale spots typically used in SIMS (or LA-ICP-MS) analysis, Pb nanoclusters did not have an influence on their ages. By contrast, Kusiak et al. (2013a, b) and Regis et al. (2017) reported that 207 Pb/ 206 Pb ages measured with ~ 20 µm SIMS spots were affected by Pb heterogeneity in their samples. Without investigating the Pb distribution and heterogeneity, one cannot be sure whether patchy Pb distribution caused the observed decoupling.
In our samples, given the observed correlation between Th/U ratios and Ti-in-zircon concentration, we surmise that the patterns observed in the Ti contents and Th/U ratios do reflect different growth periods, but that the U-Pb dates appear to have been disturbed. However, the detailed mechanisms responsible remain speculative until further investigations are made.
Similar decoupling between zircon ages and textures or Pb mobility in granulite facies zircon have been found in other areas, e.g. the Tasiuyak Gneiss, Labrador (e.g. McFarlane et al. 2005), in the western Canadian shield (Flowers et al. 2010), the Napier Complex, Antarctica (e.g. Kusiak et al. 2013a, b), the Kerala Khondalite Belt, South India (Whitehouse et al. 2014), and the Snowbird Tectonic Zone, Canada (Regis et al. 2017). All of these studies were conducted on Proterozoic rocks, whereas our study concerns granulite facies samples from the late Palaeozoic. Although the mechanism for disturbance is not known, high temperatures seem to be an essential factor, for decoupling of zircon characteristics is observed in granulite facies (> 850 °C) or UHT samples only.

Implications for the interpretation of the metamorphic evolution in the central Ivrea Zone
Our study demonstrates that parts of the lower Ivrea Zone experienced an extended period (~ 60 Myr) of HT metamorphism during the late Carboniferous to Permian times, as suggested by other studies (e.g. Vavra et al. 1999;Peressini et al. 2007;Ewing et al. 2013). Our survey of U-Pb zircon and monazite ages for metasedimentary samples from the different localities in the Ivrea Zone indicates that the main range of preserved metamorphic ages in the Kinzigite Formation and paragneiss-bearing belt is 280-260 Ma ( Fig. S9 Online Resource 3). In Val Strona di Omegna, amphibolite facies samples show two narrowly constrained metamorphic age peaks in the Permian, at 280 ± 2 Ma and 272 ± 1 Ma, while the granulite facies samples span a wide range from 300 to 240 Ma. A previously reported U-Pb zircon metamorphic age at 316 ± 3 Ma (Ewing et al. 2013) has not been confirmed to be of regional occurrence in the metapelites from the Kinzigite Formation. Whether this is owed to the local preservation or to isolated growth of this age generation remains unclear. Our ages from amphibolite facies samples are within error of the 276 ± 4 Ma zircon recrystallisation age reported by Ewing et al. (2013). Our zircon ages for amphibolite facies samples are only slightly younger than the main intrusion phase of the Mafic Complex at 288 ± 4 Ma (Peressini et al. 2007) and might thus represent a contact metamorphic imprint. However, the studied samples from Val Sesia closest to the intrusion of the Mafic Complex have the same age (IZ 415; 278 ± 3 Ma) as the samples from Val Strona di Omegna. If indeed a contact metamorphic event was responsible for the younger growth generation of amphibolite facies zircon, we would expect to see the age to depend on the distance from the intrusion contact. This is not observed.
In the northeastern and southwestern parts of the Ivrea Zone, some samples show an age generation < 220 Ma (Zanetti et al. 2013;Schaltegger et al. 2015;Langone et al. 2017) we have found no evidence for that in Val Strona di Omegna (Fig. S9 Online Resource 3).
The cause of the long-lasting regional HT conditions in this deep crustal section remains an important controversy. Petrological studies found convincing evidence that the HT metamorphic field gradient was established prior to the intrusion of the Mafic Complex (e.g. Zingg et al. 1990; Barboza et al. 1999; Barboza and Bergantz 2000;Redler et al. 2012). The present study, together with literature data, indicates a duration of ~ 60 Myr for the HT conditions in the Kinzigite Formation (Vavra et al. 1999;Ewing et al. 2013). Thermal modelling of intrusion-related thermal disturbances contend that thermal anomalies usually subside after < 10 Myr, independent of crustal parameters used (e.g. Annen and Sparks 2002;Bodorkos et al. 2002). Therefore, the long duration (~ 60 Myr) of HT conditions in the Kinzigite Formation is unlikely to have been caused by any single intrusion such as the Mafic Complex. The thermal history in the Ivrea Zone requires additional heat sources to raise temperatures to granulite facies conditions and sustain them for some 60 Myr. Considering the tectonic situation of the Ivrea Zone during Permian times, i.e. incipient rifting at a passive margin (e.g. Handy et al. 1999), the most likely source for long-lasting heat input into the crust is asthenospheric upwelling of the mantle during crustal thinning (e.g. Schaltegger et al. 2002). The idea that Permian HT metamorphism in the Ivrea Zone was caused by a largescale thermal anomaly involving the mantle is supported by the large spatial and temporal extent of the evolution: abundant evidence for Permian HT metamorphism and magmatism is found in Adria-derived units all along the Alpine chain (e.g. Hermann and Rubatto 2003;Schuster and Stüwe 2008;Kunz et al. 2018). The abundant Permian mafic and felsic intrusions (e.g. Monjoie et al. 2007;Peressini et al. 2007;Schaltegger and Brack 2007) together with a thermal input from the mantle account for the HT conditions in the lower and middle crust, as has been shown for other regions by thermal modelling (e.g. Bodorkos et al. 2002). We thus conclude that regional-scale Permian HT metamorphism in the Adriatic margin was caused by a combination of asthenospheric upwelling, magmatic underplating and magmatic bodies intruding to different crustal levels.

Conclusion
The present study is the first work reporting decoupling in young (~ 280 Ma) non-metamict (average α-dose 0.4 × 10 15 α/mg) zircon from a HT (> 850 °C) metamorphic terrain. Decoupling has been previously described, but only for zircon from Proterozoic rocks, where it was noted in radiationdamaged areas of zircon grains and was tentatively linked to Pb mobility during (U)HT metamorphism.
Here we document the evolution of zircon chemistry and ages along the Val Strona di Omegna metamorphic gradient: two amphibolite facies samples (< 750-800 °C) show a metamorphic rim (Rim2) of homogenous trace element geochemistry and well-defined U-Pb age, respectively, of 280 ± 2 Ma and 272 ± 1 Ma. Zircon from the highest grade samples (> 850 °C) are affected by HT disturbance that caused decoupling of U-Pb dates from their trace element geochemistry, while preserving the correlation between trace element characteristics (Th/U ratios and Ti concentration). Such decoupling hampers the ability to link and interpret the observed scatter in these data to different stages along the metamorphic evolution of the samples. Details of the process(es) that caused this disturbance remain uncertain; since lower grade samples were not affected, HT processes are inferred. As the zircon characteristics in samples closest to this intrusive body in Val Sesia show no unusual scatter, these processes are not directly related to the intrusion of the Mafic Complex.