DETRITAL ZIRCON GEOCHRONOLOGY WITH N=1000 (Invited Presentation)

There has been considerable recent discussion about the number of analyses that should be conducted from a detrital zircon sample, and to what degree these analyses represent the true distribution of ages in the sample. It is currently common to analyze between 60 and 120 grains from each sample and acknowledge that the proportions of different age groups may not be reliable, and that some age groups may have been missed entirely. But improvements in hardware (e.g., static multi-collection with ion counters, ablation cells with rapid response) and software (e.g., fully automated analysis routines and total count data reduction) now make it possible to conduct U-Pb analyses much more rapidly while still retaining reasonable (~2%) precision and accuracy. To evaluate the impact of these improvements, 1000 analyses were conducted from a single sample known to contain ~8 different age groups. Total analysis time was ~6 hours. The Probability Density Plot (PDP) for these analyses lacks the “noise” characteristic of n=100 distributions, and is nearly identical to the distribution generated by a Kernel Density Estimate (Vermeesch, 2012). But n=1000 is clearly not necessary for samples with only a few age groups, or with age groups of subequal abundance. The number of analyses to be conducted should accordingly be different for each sample, depending on the number of different age groups present, the abundance of each age group, and the desired confidence level (e.g., minimum number of analyses from each age group). Complexities such as Pb loss or inheritance will inevitably compromise some ages and require additional analyses. A significant improvement for detrital zircon geochronology (currently under development in collaboration with EarthChem, EARTHTIME, and CIRDLES) is a data reduction/analysis system that allows an operator to evaluate ages, complexities in ages, age groups, population abundances, PDPs, and KDEs real-time as the analyses are conducted, such that analyses can be conducted until the observed age distribution approaches the predicted age distribution and each population has been characterized to the degree required.