Resolving uncertainties in the application of zircon Th/U and 1 CL gauges to interpret U-Pb ages: a case study of eclogites in 2 polymetamorphic terranes of NW Iberia

. Zircon crystal texture and Th/U ratio have been used as a watertight argument when interpreting U- 23 Pb ages. The wide, and sometimes indiscriminate, use of those gauges could result into misinterpretation of the 24 geological meaning of U-Pb data. A case study is presented here where zircons from a controversial 25 polymetamorphic eclogite unit were analyzed with SHRIMP. Both U-Pb and trace element (TE) data were 26 collected for each point. The combination of TE and structural arguments indicates that zircon was part of the 27 eclogite facies mineral assemblage at 390 Ma. However, using Th/U ratio and CL textures lead to a different interpretation. Our results suggest that in complex orogenic scenarios and extreme environments well-known techniques (CL) and geochemical relationships (Th/U) must be used in combination with TE data and structural relationships as provenance/process gauges. While geochronology provides accurate isotope relationships, their temporal dimension must rely on structural and petrological evidence.

Dating metamorphic rocks using the U-Pb isotopic system in zircon can be a challenging task owing to 35 the ability of this mineral to grow in a variety of geological conditions and its relative resistance to metamorphic 36 processes. When the evolution of a rock results in complex textures in zircon, the combination of the high 37 spatial resolution provided by the SIMS (secondary ionization mass spectrometry) instruments together with 38 cathodoluminescence (CL) or backscattered (BS) images has turn out to be very convenient in most cases to 39 decipher this intricate history (see Corfu et al., 2003). As most of the geological processes result in a specific set 40 of zircon textures under CL or BS, this methodology strongly relies in our ability to recognize the origin of 41 zircon based on those textures, so we can link the obtained ages to specific geological processes. For example, 42 the most frequent texture in metamorphic zircon is homogeneous zoning found in discordant rims (Rubatto and 43 Gebauer, 2000), patchy zoning is commonly found in eclogitic zircon (Tomaschek et al., 2003), soccer-ball 44 zoning appears in high-grade metamorphic rocks (Fernández Suárez et al., 2007), subrounded and truncated 45 internal areas are considered inherited zircon (xenocrystic cores), and oscillatory zoning is typical of magmatic 46 zircon (Corfu et al., 2003). However, there is a lack of understanding of the zircon growth process, precluding in 47 some cases a straightforward distinction between magmatic and metamorphic zircon (e.g., Harley and Black, 48 1997;Corfu et al., 2003;Kelly and Harley, 2005). Furthermore, experimental data are scarce and technically 49 challenging (e.g., Ayers et al., 2003), limiting our interpretation of growth ages. This is particularly true when 50 high pressure and high temperature conditions are explored, or when we are dealing with suspected 51 polymetamorphic terranes. Metamorphic growth of zircon may occur not only during the thermal peak, but also 52 along the prograde and retrograde path (Roberts and Finger, 1997;Liati and Gebauer, 1999;Vavra et al., 1999; 53 Hermann et al., 2001). Moreover, it is commonly accepted that Th/U ratios lower than 0.1 indicate zircon 54 growth under metamorphic conditions, whereas higher ratios are found in magmatic environments (Williams et 55 al., 1997).

57
However, some of these one-to-one correspondences have been defied in a few cases; whether it be 58 metamorphic zircon with high Th/U ratios (see Harley et al., 2007 and references therein) or the unconventional 59 correspondence between oscillatory zoning and an eclogitic origin for zircon (Gebauer et al., 1997;Rubatto et 60 al., 1998;Bingen et al., 2001;Corfu et al., 2002;cited by Corfu et al., 2003). In such cases, the problem is 61 solved falling back on previous geochronological studies to interpret the obtained age, but the distinctive 62 composition of zircon grown under eclogitic conditions can be used as well to determine its origin (e.g. Young

116
Mineral separation was carried out at the Universidad Complutense (Madrid) and it involved an initial 117 concentration of heavy minerals using a Wilfley table, the sieving of the resulting sample below 0.2 mm, the 118 separation of the magnetic minerals with a Franz isodynamic magnetic separator, and the final concentration 119 with methylene iodide (MEI). A significant amount of heavy minerals was obtained, mainly rutile and 120 sulphides, whereas the zircon yield was poor (hardly 50 grains out of 30 kg of sample). Zircon grains are usually 121 fragments, typical of populations extracted from mafic rocks (Corfu et al., 2003), varying in size from 0.1 to 0.2 122 mm across. Still, in a few grains it is possible to recognize some crystal faces. Zircon is colorless with scarce 123 mineral or fluid inclusions.

124
The zircon grains were mounted on glass slides with a double-sided adhesive in parallel rows together 125 with some grains of zircon standard R33 (Black et al., 2004) and set in epoxy resin. After the resin was cured, University. An O 2− primary ion beam varying from 4 to 6 nA generates secondary ions from the target spot with 149 a diameter of ~20 µm and a depth of 1-2 µm. As we assumed a Paleozoic age, the counting time for 206 Pb is 150 increased to improve counting statistics and precision of the 206 Pb/ 238 U age. Concentration data were normalized 151 against zircon standard CZ3 (550 ppm U, Pidgeon et al., 1994), and isotope ratios were calibrated against R33 152 (419 Ma, Black et al., 2004). Data reduction followed the methods described by Williams (1997)

155
Twenty analyses were performed in 17 zircon grains. The youngest result has high common Pb content 156 and it will not be further considered in the discussion of the age (analysis #3.2). The following youngest result is 157 obtained from a non-luminescent rim in grain #2 (369 Ma) and the seventeen remaining analyses are evenly 158 distributed between 382 and 403 Ma (Fig. 3). The weighted mean obtained from eight analyses is 390.4±1.2 159 Ma, with a mean square of weighted deviation (MSWD) of 0.65. Finally, one analysis taken in a core yields the 160 oldest age (474 Ma) and, in spite of its high common lead, its significance will be discussed later. diameter (about 15 µm) and a less energetic O -2 beam (between 1 and 2 nA) permitted that the analyses were 165 conducted in a volume adjacent to that analyzed for isotopic compositions. The primary standard MAD is a gem 166 quality crystal from Madagascar that has been extensively characterized in-house and found to be very 167 chemically homogeneous (Mazdab and Wooden, 2006). The secondary zircon standard is CZ3 (see previous 168 section). These standards were analyzed every ten unknowns over multiple analytical sessions to establish 169 precision of the trace element analyses. The procedure to obtain concentrations from raw counts is described in 170 Schwartz et al. (2010). Precision for Y at 2σ is ±6%; for the measured rare earth elements (REE, excluding La), 171 Hf, Th, and U, 2σ precision ranges from ±8 to 18%; the precision for La is ±30%.

172
Even though we performed thirty-six trace element (TE) analyses in 19 zircon grains, in this work we 173 are only reporting those analyses adjacent to a U-Pb spot (Table DR-

207
Dissolution and precipitation can be used to explain the texture and isotopic data from the only grain with a 208 core-rim feature (grain #3 Fig. 2, Table DR-1). In spite of its high common Pb content, the age obtained in the 209 core (~474 Ma) is equivalent to other ages found in the literature for the eclogitic protoliths (Bernard Griffiths et 210 al., 1985;Peucat et al., 1990;Ordoñez Casado et al., 2001), whereas the rim age (~360 Ma) is probably affected 211 by lead loss. In any case, the absence of xenocrystic cores in the rest of the grains suggests that dissolution and 212 precipitation was subordinate and other zircon-forming processes were active in this eclogite.
Even though zircon recrystallization during metamorphism usually disturbs the former igneous zoning, 214 Schaltegger et al. (1999) report a U-loss process during zircon recrystallization that results in a weakening of 215 CL intensity, without losing the oscillatory zoning. However, this annealing is usually coupled with partial U-Pb 216 resetting. In our case, data are tightly grouped and they are equivalent to other ages obtained for the HP-HT 217 metamorphism in adjacent areas (e.g., Ordoñez Casado et al., 2001), making this argument unsound.

218
Other possibility is that new zircon grew during eclogite metamorphism. Direct crystallization from a melt or a 219 fluid has been invoked in the few studies where oscillatory zoning was found in eclogitic zircon (Gebauer et al., 220 1997; Rubatto et al., 1998). However, crystallization from a fluid can be discarded as zircons grown that way 221 usually have Th/U ratios lower than 0.1 (Rubatto et al., 1998). On the other hand, zircon crystallization from a 222 melt generated during eclogitic metamorphism could show Th/U ratios higher than 0.1. In that case, the system 223 would be open, making HREE available for both garnet and zircon (Rubatto, 2002). Nevertheless, the REE

239
The combination of the trace elements (TE) signature, CL images and a high-resolution ion probe emerges as an 240 excellent approach to overcome regional uncertainty in the studied case as previously stated in similar context 241 (e.g. Lotout et al., 2018). The positive correlation of TE evolution and metamorphic assemblage let us connect 242 textural information and regional evidence with geochronology, which was not considered in previous works.

243
Due to the correlation between metamorphic evolution and TE data, we suggest that the analyzed zircons 244 represent part of the eclogite facies assemblage, so that a HP event about 390 Ma is favored. It should be noted 245 that the existence of a previous (pre-400 Ma) HP event is not dismissed but specific experiments need to be 246 conducted to figure out its geological meaning (e.g. Lotout et al 2018).

248
We have dated zircons from an eclogitic block-in-matrix with a combination of high-resolution ion probe, CL-249 image and TE data. Strengths and weakness of those techniques have been discussed and correlated with