Rocks are made up of minerals, which in turn are made up of elements. But we now know that individual minerals found in many rocks do not always share the same history. Rather, the host rocks are an assortment of minerals, often from different sources, that have been picked up and mixed together during formation of the rock. This is most obvious in sediments, some of which are literally just grains of sand cemented together. But it is also true of many igneous rocks which are formed when a molten magma from deep within the Earth rises up through the crust, cools and crystallises. The different minerals found in such igneous rocks may have crystallised from different melts at different times. Crushing the rock up and then analysing it – the routine technique for many decades – thus obliterates any information available from individual minerals. Since precise age information is the basis for the time-scale of geological events, and hence the history of this and other planets, it is important to know exactly what is being analysed and dated.

Radiogenic isotopes are produced by radioactive decay and so they are used to determine the ages of rocks and minerals, and to investigate when major events happened in the history of the Earth. Isotopes that had relatively short half-lives, and are therefore now extinct, have been used to investigate the first few million years of the solar system. Others, with longer half-lives, are used to date the age of the Earth and to chart its evolution, such as when and how the continental crust we live on was generated.

The oldest ages on Earth are from zircons that are 4.4 billion years old

However, a particular problem faced by geologists is that most of the Earth’s crust was originally generated more than two billion years (Ga) ago and yet today 50 per cent of the rocks in the crust are less than 300 million years (Ma) old. This is because the process of plate tectonics recycles rocks in the crust over and over again. Each time this happens the geological clock becomes reset, giving younger and younger ages. It is therefore necessary to find a time archive that is robust enough to survive these events and so provide a record of the history of the early crust. That archive is the mineral zircon.

Zircons occur in tiny amounts, but they are common in rocks of the upper continental crust. Given enough time, the magma they crystallised from will become exposed at the surface and be eroded. But because zircons are extremely difficult to destroy they have a good chance of being preserved and becoming deposited in a sediment. The oldest ages on Earth are from such zircons that date back to 4.4 billion years ago – which is within 150 million years of when the Earth first formed. These zircons are preserved as grains in sediments deposited around 3.8 billion years ago. Over time, a zircon lying around in such a sediment can become buried to depths at which the temperature is sufficient to melt the sediment, but not the zircons. Over billions of years, this cycle can happen many times and each time the original zircon acquires a new growth zone (rather like a tree ring) that reflects the changing compositions of the melt around it.

We have a new way to see back to old events

So to be sure of what we are analysing we need to be able to investigate tiny amounts of an individual mineral and be able to see the exact position of that analysis, when looking down a microscope. Using this technique, different portions of the zircon just a few tens of microns across, can be dated directly using uranium (U) and lead (Pb) isotopes, and analysed for the isotopic ratios of elements like hafnium (Hf). As we have explained, –many rocks such as granites are generated by the melting of older rocks in the crust. Hf isotopes are of interest because they allow us to ‘see’ back through time to when new continental crust was first generated by melting in the mantle. The ages we obtain from Hf isotopes are thus called crust formation ages. So by analysing different zircons for U and Pb we can determine when the individual zircon formed, while the Hf isotopes tell us when the original crust formed. The crystallisation ages of the zircons reflect discrete melting events within the crust. Their Hf isotopes indicate whether the magmatic events were associated with the generation of new crust from the mantle or, if not, when their crustal precursors were derived from the mantle. We have recently been involved in such a study from south-east Australia.

There we found that large numbers of zircons have ages of around 500 and 1,000 million years, which means they crystallised from magmas at those times. These dates therefore represent major thermal episodes of melting and crystallisation. In contrast, the Hf crust formation ages of the zircons define peaks at 1.9 Ga and 3.3 Ga. Thus they are much older than the ages of most of the zircons. From these data we can conclude that (a) the major periods in which new continental crust was generated from the mantle in this area was in two episodes at 1.9 Ga and 3.3 Ga, (b) the major periods of granite generation – at 500 and 1000 Ma – involved the melting of pre-existing continental crust, and (c) these granites do not represent periods when new continental crust was generated from the mantle. This highlights the point made earlier that most rocks (and zircons) are geologically young, but most continental crust was generated a long time ago. Now, however, we have a new way to see back to these old events.

This work was supported by the Natural Environment Research Council and reported in Nature, volume 439.

Chris Hawkesworth/Earth Sciences