How old is that rock? Part 3: ⁴⁰Ar/³⁹Ar dating

In the previous blog I described how we measure the amount of ⁴⁰K, ⁴⁰Ar and ³⁶Ar to calculate how old a sample is.

All good yeah? Unfortunately, not quite.

While there are some samples and situations where this K-Ar dating technique works really well, it isn’t perfect. The technique uses a few key assumptions that are not always true. These assumptions are:

1. The ⁴⁰K and ⁴⁰Ar* are homogenously distributed in the sample, so it doesn’t matter that the K and Ar measurements are carried out on different aliquots (sub-sample) of the sample using different techniques.
2. When we measure the Ar content, we manage to release ALL of the Ar from the sample – we need absolute concentrations of ⁴⁰K and ⁴⁰Ar because they are measured with different techniques on different aliquots.
3. All of the ⁴⁰Ar in the sample is either from radioactive decay of ⁴⁰K (i.e. ⁴⁰Ar*) or from the atmosphere (⁴⁰Arₐ).

Assumption 1 shouldn’t cause too many problems for old rocks where the concentration of ⁴⁰Ar* is high and it is easy to analyse small samples (10s of milligrams). But the younger the rock, or the lower K-content of the rock, the less ⁴⁰Ar* there is and larger samples need to be analysed to release enough gas to measure. The larger the sample, the greater chance of sample inhomogeneity – and that means there is a bigger chance that the two aliquots analysed for K and Ar don’t quite match.

Assumption 2 can cause problems when analysing certain minerals, especially a mineral called sanidine. This is a kind of K-rich feldspar that forms at high temperatures and has a very disordered crystal lattice. This disordered crystal lattice makes it more difficult for Ar to diffuse out of the sample during analysis, and the high melting temperature makes it difficult to completely melt the sample to release the all of the gas. This means you might end up underestimating the amount of ⁴⁰Ar* and getting an age that is too young.

Assumption 3 can be a problem in various situations. Sticking with the simple volcanic eruption scenario from the last blog, if the magma chamber is quite new and forms in old continental crust, there might be a lot of ⁴⁰Ar in the magma that has come from radioactive decay of ⁴⁰K in the crust.  This might become trapped in a crystal we want to date, either just by being in equilibrium with the gas, or the crystal might trap tiny pockets of magma or hydrothermal fluid as it grows – we call these magmatic or fluid inclusions. This kind of ⁴⁰Ar did form by radioactive decay, but in a different system to the one we are trying date (it formed in the host crust, rather than in the crystals we want to date), so for ⁴⁰Ar/³⁹Ar dating we stop referring to this as ⁴⁰Ar* and start calling it excess-argon, or ⁴⁰Arₑ.

Fortunately, in the late 1960s, 2 scientists called Craig Merrihue and Grenville Turner discovered that if you put a K-bearing sample into a nuclear reactor and bombarded it with neutrons, some of the ³⁹K changed into ³⁹Ar. This meant that you could measure the ³⁹Ar on a noble gas mass spectrometer, at the same time as the usual ⁴⁰Ar and ³⁶Ar and calculate ⁴⁰K/⁴⁰Ar* from ³⁹Ar/⁴⁰Ar*.

With this new technique, the amount of ³⁹Ar produced depends on how much neutron irradiation the sample received, which is difficult to directly measure. To get around this, a sample of well-known age (an “age standard” or “fluence monitor”) is included in the irradiation and the Ar-isotopic composition of this standard is used, along with its age,  to calculate something called the J-value, which is a proxy for neutron dose. This J-value is then used to help calculate the age of our samples.

This new technique dealt with any problems associated with assumption 1 of the K-Ar technique. It also dealt with assumption 2, because it no longer mattered if you didn’t release all of the Ar from the sample – it is the ratio of the parent to daughter isotopes that is important, rather than the absolute concentrations.

Being able to measure both the parent and daughter isotope at the same time also opened up a whole new level of gas-release technique that helped to address any problems associated with assumption 3. Ar could be released from samples by stepwise heating (heat the sample a little bit and analyse the gas released, and then increase the temperature – repeat until there is no more gas left)- this helps in two ways. Firstly, any separate reservoirs of ⁴⁰Arₑ, like the fluid inclusions I mentioned above, tend to be released from crystals at lower temperatures than the ⁴⁰Ar* stored in the crystal lattice. That means that stepwise heating can identify different reservoirs of Ar in a sample, and we can use this information to identify which heating steps can be used to calculate an age.  Secondly, multiple measurements from the same sample (either stepped heating, or multiple analyses of single crystals) can be plotted on isotope correlation diagrams and these can be used to calculate mixing lines between different end-member isotopic compositions, making it possible to interpret complex data.

In the next blog I will explain how some of these diagrams and data-analysis techniques work. But I want to finish this post with a brief summary of the capabilities and challenges of ⁴⁰Ar/³⁹Ar dating.

• The ⁴⁰Ar/³⁹Ar technique can potentially date rocks and minerals between a few thousand, and a few billion years old;
  • The first ⁴⁰Ar/³⁹Ar dates produced in the late 1960s and early 1970s were on meteorites and lunar rocks recovered from the Apollo missions, which are all between 3 and 4.5 billion years old. The technique is still routinely used to date old, extra-terrestrial material.
  • With the right samples, it is also possible to date relatively young rocks – back in 1997 the ⁴⁰Ar/³⁹Ar lab at the Berkeley Geochronology Centre successfully dated the 79AD eruption of Vesuvius that wiped out the town of Pompeii.
• The easiest samples to work on contain a lot of K (typically between 5 and 15% K₂O), but with increased sensitivity of mass spectrometers, changes to the gas extraction systems and improvements in sample preparation, it is now possible to date samples that have a K-content of less than 0.5%, even for quite young samples. This means that the range of material that can be analysed to give an ⁴⁰Ar/³⁹Ar age is massive.
• ⁴⁰Ar/³⁹Ar dating gives cooling ages, so it can date not just volcanic eruptions, but igneous intrusions and metamorphism. It can even been used to date fault movement and meteorite impacts.
• ⁴⁰Ar/³⁹Ar dates need to be calibrated against a standard of known age. This means that the accuracy and precision of the ages it produces is always going to be limited by the accuracy and precision of the age of the standards. This is being addressed as part of an international project called EARTHTIME that aims to improve the precision of various dating techniques.

So, in short, the technique covers a massive date range and it can date a wide range of materials to give age information on lots of different kinds of geological events.

I was fortunate enough to do my PhD in the Ar-dating lab at The University of Manchester, using the MS-1 – the mass spectrometer that was built by Grenville Turner and produced the first Ar-dates. Grenville retired the year I started at Manchester, and the lab is now run by Prof. Ray Burgess, who was my PhD supervisor.  But here is a cheesy photo of me and Grenville at my graduation in 2005.

Grenville recently wrote an article giving a bit of the history of the MS-1 mass spectrometer, which you can read here.

And here are a couple of other interesting articles about Grenville and this history of his research:

Todmorden news

Floreat Domus (pdf – go to p. 34)