Let’s talk about uranium. From its fiery birth within the phenomenal temperatures and pressures of exploding stars (“supernovae”), to its Hollywood role as the radioactive stuff of doomsday trafficked by questionably-accented villains, it certainly seizes the imagination.
Image above: An example of a volcanic ocean island, which can hold clues as to the make-up of the Earth’s mantle: Mount Otemanu, the remains of a volcanic peak on Bora Bora island, French Polynesia. Image courtesy Wikimedia Commons/Sergio Calleja.
But uranium also has a much quirkier aspect, one that the movie-makers don’t seem so interested in. Like other metals, such as rusting iron, uranium corrodes in the presence of oxygen. And the resulting oxidised uranium has a peculiar property: it can dissolve in water. Now, I can practically hear the yawns as I type this; it’s water-soluble, so what? But this odd little fact has had a fascinating impact on Earth’s ancient past.
Rewind around 4.4 billion years back from the present, to a time when the Earth had just finished aggregating from the rubble of the primordial Solar System. Careful, it’s hot! The ferocious heat generated by the collisions that had built the planet, together with heat produced by decay of radioactive elements (including uranium!), meant that this early Earth was largely molten, and – as the Greek in the bath famously observed – fluid systems neatly sort themselves by density. And that’s exactly what happened: the primordial Earth differentiated into a dense, iron-rich core; a less dense, silicon-oxygen-magnesium mantle; and an even less dense silicon-oxygen crust. But as well as bulk density, the Earth’s elements also sorted themselves by their chemical behaviour. Essentially, some elements melt more easily out of a rock than others, and uranium happens to be one of these elements (geochemists term this ‘incompatible’ behaviour, because it sounds fancier than ‘easily melted’). The result? Uranium is physically dense but chemically incompatible, and ended up enriched in the nascent crust, and depleted in the mantle (and probably – although we don’t know for sure – virtually absent from the Earth’s core). And as the crust itself is further fractionated between denser, more magnesium rich oceanic crust and lighter, more silicon rich continental crust, uranium ended up most enriched in the embryo continents.
Now here’s where the story gets interesting. I’ve already mentioned that oxidised uranium is water-soluble. But on Earth, free oxygen is mostly a product of biological photosynthesis, and for a long time there wasn’t any, with early single-celled critters probably preferring to make a living from the chemical products of undersea hydrothermal vents – and such strange ecosystems, albeit more complex, still exist around so-called ‘black smokers’ today. So from the formation of the first continental crust that lasted more than an eyeblink without melting (sometime before 4 billion years ago) up to the evolution of widespread photosynthetic organisms, a lot of uranium was locked up in the continental crust (less the small amounts in the sediment physically pounded off it by erosion). That’s important to our story – keep it in mind.
Eventually, of course, photosynthesising critters did get going sometime around 3.3 billion years ago, such as the famous stromatolites. It took a while, but in a series of pulses around 2.5 – 2.0 billion years ago, and again at around 0.6 billion (600 million) years ago, free oxygen rose to comprise firstly about 3% and finally about 30% of the atmosphere (as covered previously on this blog by Michael Babechuk). So from about 2.5 billion years ago, uranium held in surface rocks had an option never before available.
It could oxidise.
And didn’t it do just that. Suddenly, surface waters were rich in dissolved, oxidised uranium; and the thing about surface water is it’s a fabulous concentrator. Trickles become streams; streams become rivers; rivers dump into the sinks of swamps, lakes, seas, and oceans. As with the transporter, so with the transported; uranium quickly accumulated in the aforementioned sinks. There was so much of it about that a particularly wacky phenomenon became possible: natural nuclear reactors. At Oklo in Gabon, about 1.7 billion years ago, water rich in oxidised uranium percolated into a shallow sandstone aquifer. Here, away from the oxidising conditions at the surface, the uranium was stripped of its oxygen and promptly came out of solution, depositing itself in voids and cracks within the rock. Eventually, a critical mass of uranium was deposited, and a chain reaction of spontaneous nuclear fission began – essentially the same process that occurs in modern atomic reactors. This went on for hundreds of thousands of years, converted several tonnes of uranium to fission products such as caesium and xenon, increased the local ambient heat by a couple of hundred degrees centigrade, and probably did some rather funky things to any of the aforementioned single-celled critters unlucky enough to be in the vicinity. A great write-up of the Oklo natural reactors is available here, but cool as this is, it isn’t the end of our story.
Liberated by atmospheric oxygen, concentrated by surface waters, the uranium of the crust was on its way home to the mantle, via the oceans, incorporation from seawater into oceanic crust by fluid alteration, and ultimately subduction back into the mantle as a passenger on that same oceanic crust. But as mentioned before, oxidation of the Earth’s atmosphere and its oceans was a process, not an event, lasting from roughly 2.5 to 0.6 billion years ago. The result of this? It took a long time for circulating oxygen, dissolved at the sea surface from the atmosphere, to fully oxidise the oceans. So during this time, uranium was being released from the continental crust by the partially-oxidised atmosphere into the partially oxidised ocean. So our uranium would have undergone a lot of cycles of oxidation and reduction (the opposite process to oxidation). And here’s where a subtle quirk of uranium’s physical properties kicks in. Natural uranium occurs in two main flavours, called isotopes: U-235 and U-238 (the values refer to the number of neutrons and protons that comprise the nucleus of the atom, but don’t worry about this). For decades, it was thought that the ratio of these two isotopes was fixed at any given point in time. That is, it was known that the ratio changed in a predictable way over time, because although both isotopes are radioactive, U-235 decays faster than U-238 and thus has become relatively less abundant over time (and as a matter of fact, formation of the Oklo natural reactor was only possible due to this greater past abundance of U-235), but at any given moment the ratio of the two should be the same globally. However, over the last decade, high-precision measurement of this ratio has shown small variations, thought to be caused by…cycles of oxidation and reduction! It seems that, at low temperatures, U-235 is ‘slightly’ (about 0.001 times) easier to oxidise than U-238, resulting in small but detectable enrichment of U-238 in any system subject to such cycles.
What does this mean? Prior to about 2.5 billion years, any uranium reaching the oceans (and thus eventually the mantle, via subduction) was transported only as an undissolved component of eroded sediment, with no U-238/U-235 fractionation. And after about 0.6 billion years ago, surface uranium generally stays oxidised once eroded from its host rock, allowing for very little isotopic fractionation. But between these two ages, during the oxidation of Earth’s atmosphere and oceans, uranium liberated from surface rocks would have encountered a whole range of oxygen levels on its way back to the mantle, with plenty of oxidation-reduction cycles and plenty of fractionation. The result? Oceanic crust subducted back into the mantle at different times in Earth’s history had different U-238/U-235 ratios. The subduction input to the mantle is ‘tagged’ with a distinct uranium isotope value, depending on when it was exposed to dissolved surface uranium prior to subduction.
Now this is important. The mantle is hot and plastic (due to both the radiogenic and leftover primordial heat from Earth’s formation mentioned earlier), and therefore is capable of flowing and mixing. The question is, to what extent? The mantle, after all, is about 2900 kilometres thick; does it wholly convect like a hot pot of soup on a stove, or is mixing more limited, confined within set levels? And if whole-mantle convection does take place, with lower and upper levels merrily mixing away, how long does that take? These questions matter, because mantle convection, whether whole-mantle or largely confined to the upper mantle, drives plate tectonics and thus shapes the world we live in. And the existence of isotopically distinguishable inputs to the mantle over time allows these questions to be answered. Whole-mantle convection – the “pot of soup” hypothesis – would homogenise the uranium isotopic signature of the mantle over time; partial mantle convection, in which the upper mantle readily convected but the lower mantle convected only slowly or even not at all, would result in isotopically distinct layers within the mantle. All that is needed is a way to sample mantle uranium compositions at depth.
Fortunately, nature makes this possible, and the authors of a recent paper in the journal Nature do just that*. By comparing the uranium isotope signature of a volcanic rock called basalt formed at mid-ocean ridges (which sample the uppermost mantle) and isolated volcanic islands such as Hawaii, (which sample the mantle at much greater depths), they found that the two basalts are drawing on two isotopically different mantle sources.
The ridge basalts had a similar composition to modern subducting ocean crust, suggesting that the uranium isotope signature of the upper mantle developed since 0.6 billion years ago; however, the ocean island basalts were enriched in U-238, requiring that the upper and lower mantle be differentiated. This implies that the lower and upper mantle largely convect separately; in particular, these results suggest that the lower mantle has not mixed with subducting ocean crust since about 2.5 – 0.6 billion years ago. This would suggest either agonisingly slow convection times for the whole mantle, or that something was different about the subduction process during 2.5 – 0.6 billion years; perhaps cooling of the mantle since this time has reduced the density contrast with sinking slabs of subducted ocean crust, inhibiting them from entering the lower mantle. Whatever the reason, it is a fascinating result. Starting with a small, apparently trivial observation – that uranium is soluble in water only when oxidised – we can begin to infer the dynamics of the Earth’s mantle. That’s science, folks!
Postdoctoral Research Fellow in the Department of Geology, Trinity College Dublin
* Andersen, M., et al., 2015, The terrestrial uranium isotope cycle, Nature, v.517, p.356-359.