Effects of thermal annealing and chemical abrasion on ca. 3.5 Ga metamict zircon and evidence for natural reverse discordance: Insights for U-Pb LA-ICP-MS dating

We present a microstructural and Usingle bondPb systematics study comparing pristine, thermally annealed (TA) and chemically abraded (CA) ~ 3500 Ma zircon from a quartz-dioritic gneiss, with the aim to improve pre-analytical workflows for more accurate and precise LA-ICP-MS Usingle bondPb dating of ancient zircon. Four zircon domains are identified: i) low- to medium-U concentric-oscillatory zoned cores, ii-iii) two porous high-U outer alteration domains, and iv) low-U narrow inward growing recrystallization rims. Raman spectroscopy on the pristine zircon reveals a positive correlation between increasing structural damage and U content. The porous high-U outer domains show a drastic increase in non-formula Ca above estimated amorphous fractions of ~ 0.8, which we ascribe to hydrothermal alteration of high-damage zircon characterized by percolating networks of amorphous areas. Upon treatment (TA, CA), hyper-spectral CL and Raman spectroscopy suggest structural recovery of point defects, but not full repair of high-damage amorphous areas. Calculated Raman radiation damage ages suggest that natural annealing affected all domains at ~ 500 Ma, consistent with negligible laser ablation matrix effects comparing pristine and treated cores. We show that reverse discordance in pristine and TA cores is not an analytical artifact. High-U alteration domains are normal discordant and were partially reset at ca. 3410 Ma. Upon CA, Usingle bondPb discordance and scatter are reduced in the cores, yielding intercepts of 3503 ± 14 Ma and 560 ± 550 Ma (MSWD = 1.0). The lower intercept matches the timing of Pb-loss recorded in the alteration domains and the Raman radiation damage age. We argue that short distance Pb redistribution within the cores, which led to reverse discordance, was controlled by Usingle bondPb systematics recording both the crystallization age and the timing of Pb redistribution. Some excess Pb was redistributed from the partially re-set alteration domains into the cores, leading to mixed Usingle bondPb systematics and further reverse discordance. Removal of the latter excess Pb upon CA suggests that radiogenic Pb from the alteration domains accumulated within distinct damage sites accessible for CA leaching. We conclude that standardization of pre-analytical treatment is not recommended; we propose to investigate the structural state of unknown specimen by means of Raman spectroscopy, in combination with commonly applied CL imaging. During LA-ICP-MS analyses, co-measurement of non-formula and selected trace elements is deemed highly valuable in detecting ablations for use in reliable Usingle bondPb age determination