⁴⁰Ar/³⁹Ar dating of basaltic rocks and the pitfalls of plagioclase alteration

Abstract

⁴⁰Ar/³⁹Ar geochronology is one of the most important techniques for constraining the timing of basaltic events due to the paucity of suitable minerals in basalts for other geochronological techniques such as U–Pb (e.g., zircon, baddeleyite). Among a variety of materials from basaltic rocks that have been used for ⁴⁰Ar/³⁹Ar dating, plagioclase is the most important due to its common presence in basalts as a primary crystallizing phase, and its transparency so that fresh grains can be selected during sample preparation. However, plagioclase ⁴⁰Ar/³⁹Ar geochronology has often been compromised by alteration (e.g., sericitization by hydrothermal events), which, in practice, is difficult to identify using a petrographic microscope when the amount of alteration is low (e.g., <1%).

We used laboratory step-heating experiments and theoretical simulations to characterize the ⁴⁰Ar/³⁹Ar age and Ca/K spectra of altered plagioclase so that ⁴⁰Ar/³⁹Ar dating results on altered samples can be identified and better interpreted. The step-heating experiments and theoretical simulations yielded consistent results, and show that with the presence of even a tiny amount of sericite (~0.01% for K-poor samples and ~0.1% for K-rich samples), the plagioclase samples yielded alteration plateau ages that are 3%–4% younger than the crystallization age. The difference between the alteration age of sericitized plagioclase and its crystallization age is primarily controlled by the time lapse between the crystallization and sericitization events, but also by the Ca/K ratios of the plagioclase. For plagioclase samples that experienced the same alteration event, the higher the Ca/K ratio is, the more sensitive the ⁴⁰Ar/³⁹Ar age is to alteration. We propose that the alteration signatures of plagioclase can be effectively identified through inspecting the ⁴⁰Ar/³⁹Ar age spectra, the Ca/K spectra, and the degassing curves. We also investigated the effect of sericitization of plagioclase microliths in basaltic groundmass and modelled the ⁴⁰Ar/³⁹Ar age and Ca/K spectra of altered groundmass samples. We validate our approach by revisiting published ⁴⁰Ar/³⁹Ar dating results for large igneous provinces, and showed that these dates should have been interpreted as alteration ages (minimum eruption ages) rather than crystallization ages. Finally, we demonstrate that with high degrees of alteration (~50% for K-poor and >70% for K-rich plagioclase samples), the age of hydrothermal alteration can be successfully dated.

Introduction

Precise and accurate dating of basaltic terrestrial rocks is a prerequisite for understanding a large variety of geological and geodynamical phenomena, such as the link between mantle plumes, supercontinent breakups, and intraplate alkaline magmatism (e.g., Dalrymple and Clague, 1976, Turner et al., 1994, Marzoli et al., 1999, Olierook et al., 2019); the relationship between large igneous provinces (LIPs), impact events, and mass extinction and oceanic anoxic events (e.g., Jourdan et al., 2014, Burgess et al., 2015, Renne et al., 2015, Schoene et al., 2019); and the formation mechanisms of volcanic-hosted ore deposits (e.g, Camprubí et al., 2003, Barrote et al., 2020) on Earth. Understanding the early history and evolution of the Moon, Mars, and asteroids, such as their collisional/impact history, timing of volcanic activity, and metamorphic/cooling history, also depends on accurate and precise age constraints on basaltic rocks (e.g., Cohen et al., 2001, Iizuka et al., 2019, Thiessen et al., 2019, Jourdan et al., 2020, Merle et al., 2020), which constitute the most abundant crustal material in the solar system.

Materials from basalt that can be used for ⁴⁰Ar/³⁹Ar dating include plagioclase, pyroxene, groundmass (or whole rock for aphyric basalt), biotite and hornblende. Despite a few recent cases of successful dating on pyroxene separates in terrestrial basalt (Ware and Jourdan, 2018, Konrad et al., 2019, Zi et al., 2019, Jiang et al., 2021c), ⁴⁰Ar/³⁹Ar dating of pyroxene is still challenging due to its very low K content and reliance on the absence of thick exsolution lamellae which prevent irradiation-induced recoil redistributions (Ware and Jourdan, 2018). Hydrated minerals such as biotite and hornblende are rich in K and have the ability to yield high-precision ⁴⁰Ar/³⁹Ar ages (e.g., Reichow et al., 2009, Ware et al., 2018, Sahoo et al., 2020). However, these minerals are rarely present in mafic rocks and therefore plagioclase and groundmass are still the most commonly used materials in dating low-K basalt (e.g., Koppers et al., 2000, Renne et al., 2015, Gomes et al., 2020).

There is a growing number of publications demonstrating that groundmass of relatively old samples (e.g., >30 Ma) do not always provide reliable crystallization age for the basaltic rocks despite generating concordant plateau ages (Hofmann et al., 2000, Baksi, 2007a, Jourdan et al., 2007, Renne et al., 2015, Merle et al., 2019, Jiang et al., 2021a). This is because groundmass samples are prone to be affected by: (i) K leaching due to hydrothermal alteration and/or ³⁹Ar recoil loss (Jourdan et al., 2007, Renne et al., 2015), which can lead to older apparent plateau ages, and/or (ii) ⁴⁰Ar* (radiogenic Ar) loss and/or recrystallization from cryptic hydrothermal alteration (Hofmann et al., 2000), which will lead to younger apparent plateau ages (e.g., Merle et al., 2019). For groundmass, even though potential signs of alteration (e.g., brown, red or yellow discolouration) can be avoided during sample preparation, cryptic alteration cannot be easily observed since groundmass is not transparent. Plagioclase is transparent and thus, contrary to groundmass, fresh plagioclase grains can be separated from altered ones under a binocular petrographic microscope and, if only fresh plagioclase can be isolated, can yield reliable crystallization ages for basaltic rocks.

Nevertheless, the often very low K₂O abundance (<0.1 wt.%) of plagioclase makes ⁴⁰Ar/³⁹Ar dating on plagioclase extremely sensitive to the alteration of plagioclase, as the latter process is accompanied with neo-crystallization of sericite, a fine-grained variety of muscovite with ~10 wt.% of K₂O. The sericitization of plagioclase refers to the replacement of plagioclase (NaAlSi₃O₈–CaAl₂Si₂O₈) by sericite [KAl₃Si₃O₁₀(OH,F)₂] through filling in microfractures within the plagioclase crystal and/or mineralogical replacement when hydrothermal fluids percolate through the plagioclase at low-temperature (100–350 °C) and low-pressure conditions (e.g., Morad et al., 2010, Li et al., 2013). The neo-crystallized sericite is a newly added phase and thus, does not retain any ⁴⁰Ar* from the plagioclase and starts its own K/Ar clock at the time of formation. Furthermore, the process of alteration can be accompanied by ⁴⁰Ar* diffusive loss associated with the temperature rise caused by the hydrothermal fluids. These processes would cause the loss of ⁴⁰Ar* without a concomitant loss of K, and lead to lowered apparent age compared to the crystallization age. As a result, the ⁴⁰Ar/³⁹Ar ages provided by partially or fully sericitized plagioclase can only be regarded as strict minimum eruption ages (e.g., Rohde et al., 2013, Olierook et al., 2015).

Although one can exclude sericite grains and partially sericitized plagioclase grains, and select fresh plagioclase grains for ⁴⁰Ar/³⁹Ar dating during sample preparation processes, the sericitization of plagioclase below 1% is hard to visually detect using a petrographic microscope (Fig. 1). It is possible to measure the compositions of separated grains by using in situ analytical methods such as Scanning Electron Microscopy (SEM), Secondary Ion Mass Spectrometry (SIMS), or Electron probe micro-analyzer (EPMA), in order to ensure the purity and freshness of plagioclase grains, however, these analyses require the grains to be mounted in epoxy, which could contaminate the plagioclase with K and/or Cl. Alternatively, thin sections can be used for these analyses, however, it is impossible to assess the proportion of sericite in the very grains that are separated for dating. Furthermore, in order to obtain a precise age, several mg to tens of mg (i.e., hundreds to thousands of crystals) of plagioclase are needed, so measuring the grains individually to detect possible sericitization is impractical. Ultraviolet laser ablation can be used to measure a spot ⁴⁰Ar/³⁹Ar age from a single plagioclase grain (e.g., Kelley et al., 1994, Mark et al., 2008), but this technique is too imprecise for most applications related to LIPs, impact events, and extraterrestial rocks. Therefore, assessing the effect of alteration on plagioclase ⁴⁰Ar/³⁹Ar geochronology is very important in order to determine whether a given ⁴⁰Ar/³⁹Ar age truly reflects the crystallization age of the dated material or is biased by alteration.

Previously, Verati and Jourdan (2014) used theoretical modelling to demonstrate how the K/Ar and ⁴⁰Ar/³⁹Ar total fusion age can be lowered by different degrees of sericitization of plagioclase with various Ca/K ratios (proxy for K abundance). In practice however, one uses ⁴⁰Ar/³⁹Ar age spectra (⁴⁰Ar/³⁹Ar apparent ages plotted against cumulative percentage of ³⁹Ar released in a step-heating experiment) rather than total fusion ages to date plagioclase. This is because for ⁴⁰Ar/³⁹Ar plateau ages, it is possible to assess if the K/Ar system stayed closed since formation, by assessing the reproducibility of the ages through robust statistical tests (e.g., MSWD and/or p; Baksi, 2007b); and to detect the presence of excess ⁴⁰Ar by testing whether the trapped ⁴⁰Ar/³⁶Ar ratio is atmospheric, or if not, using an inverse isochron approach (McDougall and Harrison, 1999).

In this study, we illustrate the behaviour of ⁴⁰Ar/³⁹Ar age and Ca/K spectra of plagioclase crystals that have been affected by various degrees of sericitization, by using a theoretical simulation and step-heating experimental approach on mixtures of plagioclase and muscovite (an ideal proxy for sericite or fully sericitized plagioclase both chemically and regarding diffusion kinetics). The laboratory step-heating experiments were conducted on mixtures with known proportions of plagioclase and muscovite that aimed to reproduce the physical process of a plagioclase–sericite mixture, and its observable effect on ⁴⁰Ar/³⁹Ar age spectra. The theoretical simulations aim to model the geological processes that occur in nature, including the ⁴⁰Ar* accumulation since the crystallization of the plagioclase, the ⁴⁰Ar* loss associated with the increase of temperature due to hydrothermal fluids, and the Ar degassing behaviour from theoretical plagioclase and sericite in a step-heating experiment. The results are then used to discuss how ‘fresh’ the plagioclase needs to be in order to produce reliable crystallization ages, how sericite signatures can be identified in ⁴⁰Ar/³⁹Ar age and Ca/K spectra, and how ‘altered’ the sericitized plagioclase needs to be to yield the age of hydrothermal alteration events.