The choice of geochronological methodology

The processes operating on Earth occur on a variety of different timescales and have taken place at different moments in the deep time. For instance, major plate tectonic processes such as mountain building, plate motion, subduction, underplating, sedimentary basin evolution, all occur on long timescales of several tens of millions of years. Other processes take place on a much shorter timescale, i.e., in the order of tens of 10⁴-10⁵ years.

For instance, the Milankovitch cycles operate at 10⁴ to 10⁶ years’ timescales, biogeochemical cycles such as biological pump also operate on a similar timescale, ocean acidification occurs on a 100 kyr (Honisch et al., 2012), whereas globally warm and cool periods lasted for 100’s kyr time scales. Similarly, biological evolutionary rates and extinction also occur in on a multi-millennial scale. To fully understand the exact rate at which these processes take place, it is of paramount importance that the geochronological tool provides sufficient temporal resolution to decipher these processes on a timescale at which they have happened.

The most widely used geochronometres to date Earth processes are U-Pb and Ar-Ar methodologies. The U-Pb systematics has always had an advantage over Ar-Ar-Ar-Ar systematics because of its ability evaluate the accuracy of its dates based on the concordance between two independent radioactive decay systems, from ²³⁸U and ²³⁵U to stable ²⁰⁶Pb and ²⁰⁷Pb, respectively. The lack of concordance between the two parent-daughter pairs (²⁰⁶Pb /²³⁸U and ²⁰⁷Pb/²³⁵U) denotes open system behaviour (Tilton, 1960;

Wasserburg, 1963). As such, the U-Pb isotope system has a built-in cross-check mechanism to evaluate accuracy of its dates. However, open system behaviour is a known problem within U-Pb systematics and is attributed to post-crystallisation loss of radiogenic Pb from the mineral lattice (Holmes, 1954; Krogh, 1982; Tilton, 1960). It is a known fact that U-Th-rich minerals such as zircon are damaged by α decay of U and Th (Meldrum et al., 1998;

Nasdala et al., 1996). The high energy decay rates result in a build-up of crystal lattice defects, which results in amorphous domains within the crystals that facilitate the release of radiogenic Pb during post-crystallisation processes (Nasdala et al., 1998). Pb does not

favourably substitute the major chemical constituents in zircon minerals due to its large radius, contrarily to Hf⁴⁺, U⁴⁺, or heavy REE³⁺, which are replacing Zr⁴⁺. Zircon tends to lose Pb²⁺ much easier via diffusion than the 3+ or 4+ cations, because they fit the lattice better (Kober, 1987). Ultimately, the loss of radiogenic Pb will result in biased Pb/U ratios and thus younger U-Pb ages.

Over the last 10 years, the CA-ID-TIMS methodology has seen an unprecedented development towards more reliable and precise results. The most important one was to mitigate the effects of Pb-loss. The most successful of which is the chemical abrasion (CA) technique, which was developed by Mattinson, (2005) and empirically calibrated by Widmann et al., (2019). Broadly, the technique consists of two steps: 1) a high temperature annealing step of the zircon grains in an attempt to restore the parts of the crystal lattice that have undergone weak damage by radioactive decay; 2) a partial dissolution step that involves submitting single grains to high temperature and high pressure partial dissolution under hydrofluoric acid. The latter step aims to preferentially remove the domains affected by radiation damage and thus affected by Pb-loss. As a result, the chemical abrasion technique aims to remove the effects of Pb-loss and yield more accurate dates (Please see section 2.2.2.3 for further detail).

Currently, there are three analytical approaches that are routinely used in U-Pb geochronology: TIMS, SIMS, and LA-ICP-MS. The latter two are often referred to as high-spatial resolution or in-situ geochronology and use a micro-beam that allow minute parts of the analysed material to be dated. However, high-spatial resolution methodologies have limited use in dating processes on the 10⁴-10⁵ year timescale because of their inherently low analytical precision. Precision of U-Pb dates from these methodologies is a function of the stability of the analyte signal, the number of detected ions, the sensitivity of the collectors, and corrections to the measured ratios from reference standards, and correction for common Pb (Ireland and Williams, 2003; Košler and Sylvester, 2003; Schaltegger et al., 2015). The analytical instrumentation of LA-ICP-MS and SIMS displays large ionisation efficiencies (up to 100%) compared to TIMS (10%). On the other hand, the ion transmission is much lower in in-situ techniques.

Figure 1.1 – Illustration of a cross section showing the relative volume of zircon sizes which each mass spectrometry technique.

Modified after Schaltegger et al. (2014)

In the case of Pb, the ion transmission is in the order of 10% for TIMS (Gerstenberger and Haase, 1997), whereas it is 0.39-1.72% in LA-ICP-MS (Amelin and Davis, 2006; Black et al., 2003; Thirlwall, 2002). Moreover, the thermal ionisation technique delivers a much longer-lasting beam (up to 4h) when compared to LA-ICP-MS (1-3 min) and SIMS (20 min). Consequently, the precision in TIMS is improved by sheer counting statistics.

Correction of measured isotope ratios against a naturally occurring reference material such as the Temora (Black et al., 2003) or Plešovice zircons (Sláma et al., 2008) is a major source of uncertainty; mainly because naturally occurring materials are not homogeneous at the scale of analysis and even less so on a grain-to-grain scale. The poor geochemical characterization of naturally occurring standards leads to corrections in the measured ratios that comprise accuracy and precision of the final dates. Conversely, the high-precision U-Pb geochronology community has developed a high purity tracer solution with an accurately and precisely calibrated concentration of each Pb and U isotopes needed.

The tracer solution is a gravimetrically weighed solution that contains precise and accurate molar quantities of U and Pb (Condon et al., 2015; Krogh and Davis, 1975; Parrish and Krogh, 1987). The tracer solution is calibrated against metrological standards and traceable to the units of the SI system. As such, ID-TIMS U-Pb ages are considered to be absolute whereas in-situ techniques are considered relative ages. Finally, the use of a double-isotope tracer solution for both Pb and U, such as the EARTHTIME ²⁰²Pb-²⁰⁵Pb-²³³U-²³⁵U spike allows for accurate correction of mass fractionation thus further improving accuracy (Condon et al., 2015; Parrish et al., 2006). Moreover, a jointly calibrated and world-wide distributed tracer solution contributes to minimize the inter-laboratory bias which is now at the level of 0.1%. Control over other chemical parameters are also of importance in high-precision U-Pb geochronology. A precise and accurate re-determination of the natural

238U/²³⁵U (Condon et al., 2010; Hiess et al., 2012) has contributed to the accuracy of the methodology as well.

The correction for common Pb (Pbc) present in the analysis is a major challenge in U-Pb geochronology. Since Pb²⁺ is not readily incorporated into the zircon lattice during crystallisation, the Pbc present is due to contamination (“blank”). High-precision geochronology laboratories have consistently lowered the Pbc contamination, i.e., that amount of laboratory contamination added to the sample, which has been a major factor in improving precision while analysing evermore smaller samples. Currently the amount of laboratory blanks in high-precision U-Pb laboratories is in the order of 0.1 to 0.8pg per sample. Furthermore, the Pb isotopic composition of laboratory contamination is precisely measured in high-precision geochronology unlike with the U-Pb techniques, contributing largely to the superior precision and accuracy of high-precision U/Pb dating using ID-TIMS. As a result, these combined efforts lead to age uncertainties below 0.1% in

206Pb/²³⁸U that permit the resolution of processes taking place on timescales of hundreds of thousands of years. Contrarily, in-situ techniques such as SIMS or LA-ICP-MS are not capable of quantifying ²⁰⁴Pb present in the sample, therefore not offering a full isotopic analysis of Pbc present in the sample necessary for precise and accurate correction

In addition to issues compromising precision, in-situ techniques lack accuracy because of its inability to address Pb-loss. Theoretically, zircons should be concordant and the source of uncertainty should solely derive from analytical issues (counting statistics), and precision could be improved by repeated analysis of a large zircon population.

However, small proportions of Pb-loss are camouflaged by the low precision (1-5%). As a result, high-spatial resolution techniques are often inaccurate and imprecise.

The standardisation of laboratory procedures between different laboratories (such as, e.g., for the chemical abrasion suggested by Widmann et al., (2019)) is a continuous effort under the EARTHTIME initiative to reduce inter-laboratory bias, and the long-term reproducibility and repeatability of the dates produced. Additionally, the standardization of the statistical algorithms and error propagation by the use of the same open-source software (Tripoli and Redux) for reporting, reducing and treating U-Pb data (Bowring et al., 2011;

McLean et al., 2011; Schmitz and Schoene, 2007) is major advancement.

The mitigation of a systematic off-set in the order of 1% between U-Pb and

40Ar/³⁹Ar ages on the same rock has become a major target of the high-precision geochronology community (Min et al., 2000; Schmitz and Bowring, 2001). The Ar-Ar chronometre relies on radioactive decay of ⁴⁰K to ⁴⁰Ar; however, the inaccurate knowledge of the ⁴⁰K decay constant (Min et al., 2000; Renne et al., 2010), as well as other physical constants such as the ⁴⁰Ar/⁴⁰Ca branching ratio and the natural ⁴⁰K/Ktot ratio contribute to this offset. Additionally, as with in-situ techniques, the Ar-Ar methodology relies on correction of the measured isotope ratios to inhomogeneous natural reference material, which, as with in-situ U-Pb techniques, is a major contributor to the overall uncertainty (Renne et al., 2009; Villeneuve et al., 2000). The widespread use of the Ar-Ar chronometre to date Earth processes is partially due to the widespread presence of K-bearing minerals such as plagioclase, sanidine, biotite, micas, and amphiboles in crustal rocks.

Intercalibration between U-Pb and Ar-Ar ages have been tried through direct comparison

of U/Pb vs. Ar/Ar age pairs (Renne et al., 2010; Villeneuve et al., 2000) and through intercalibration via astronomical cycles (e.g., Kuiper et al., 2008)

In this thesis, I used exclusively high-precision U-Pb geochronology using CA-ID-TIMS techniques applied to zircon from volcanic layers within the sedimentary record, in order to obtain a precision and accuracy to study processes at the 100 kyr timescales, which only this methodology can deliver.