Blog Projects

How Old is Earth’s Inner Core?

This post summarizes a publication in the journal Nature (note paywall):

Biggin, A.J., Piispa, E.J., Pesonen, L.J., Holme, R., Paterson, G.A., Veikkolainen, T., and Tauxe, L., 2015, Palaeomagnetic field intensity variations suggest Mesoproterozoic inner-core nucleation: Nature, v. 526, no. 7572, p. 245–248, doi: 10.1038/nature15523.


Earlier this month, an article about the age of Earth’s inner core made a big splash on both traditional and social media. The early history of Earth’s core – and particularly the inner core – aren’t things that scientists understand particularly well. Nonetheless, the question of when the inner core formed is a part of bigger questions about how Earth got here and what makes it unique.  We think that the inner core began to form once Earth had cooled a certain amount after its fiery infancy [1]. The inner core, then, is a sort of a thermometer for the early Earth: knowing exactly when the inner core started to form gives us some insight into just how hot Earth was shortly after its formation [2]. And I’m talking really hot here – magma oceans, not global warming. Other aspects of Earth that we consider more or less unique would have developed once Earth became cool enough. Plate tectonics and life are two examples.

Geochemists have fairly strong evidence from isotopes of the elements Hafnium and Tungsten that Earth’s metallic core must have formed in the first 50 million years of our planet’s history [3]. Earth has an unusually large core for a planet of its size. We think this is at least partially due to a Mars-size rock that plowed into the Earth early in the Solar System’s history. Computer simulations of this planetary collision indicate – and geochemistry confirms – that the event would have spun off some rocky material, becoming the Moon. The denser, metal-rich stuff left behind in the crash would have migrated toward the center of the developing Earth, forming our oversized core.

The origin of the inner core presents a bigger problem. Right now, we know about the inner core from the patterns of seismic waves that travel from one side of the Earth to the other, crossing different parts of the core. But seismic waves are ephemeral, dissipating soon after we record them. There are no seismic records from Earth’s past.

Earth’s magnetic field, generated in the fluid outer core (the geodynamo), might give us some clues about when the solid inner core formed. It takes some energy to generate Earth’s magnetic field. In the modern Earth, we think that the growing inner core powers the geodynamo: as the iron-nickel inner core crystallizes, light elements are spit out into the fluid outer core. The density difference between the sinking iron and nickel and the rising light elements, along with the temperature difference between the hot core and relatively cold mantle, cause the outer core to churn like a boiling pot of soup on the stove. This vigorous churning – thermal and compositional convection – produces a strong magnetic field relative to those of most other terrestrial planets. Without the inner core to drive the compositional part of the convection, the core would need to be very much hotter if we wanted it to produce a magnetic field like the one we have today.

Unlike seismic waves, we do have a record of magnetic fields from very old rocks. Since about the 1980s, geophysicists have been trying to identify patterns in the waxing and waning of Earth’s magnetic field over the billion-year timescales that we would need to look at planetary evolution. If we can identify a time when Earth’s strong and relatively stable magnetic field switched on, the reasoning goes, we might be able to tell when the inner core began powering the geodynamo.

I hedge my bets here about the magnetic field-inner core connection because of two main problems: first, it’s hard to measure the strength of Earth’s magnetic field in the past, and second, Earth’s magnetic field has varied through time due to a number of factors besides the inner core.

As far as the first problem goes, we have to make a lot of assumptions when using old rocks to look at Earth’s magnetic field. We assume that the rocks recorded Earth’s magnetic field reliably in the first place, through processes that we understand well. We assume that rocks are able to hold onto their magnetization for billions of years. We assume that the same rocks have not been remagnetized or demagnetized throughout their long histories. Although we have ways to test some of these assumption, our ability to do the tests depends on geological circumstance – whether the rocks contain certain dateable minerals, whether they have been folded or broken up and re-cemented together, or whether they are crosscut by more recent rocks, for instance. Even for young rocks, reading the strength of Earth’s magnetic field is much harder than looking at the direction of the ancient magnetic field (more on the specifics in a later post). Add to that the fact that rocks that have been around longer are generally more likely to have been re-magnetized, and you have a lot of reasons to doubt the paleomagnetic record of multi-billion-year-old rocks. For this reason, Andy Biggin and his co-authors on the recent Nature paper are choosy about which published data they use for their analysis [4]. Unfortunately, because of the problems mentioned here, and because there just aren’t many old rocks to choose from, reliable data on Earth’s magnetic field from the first four billion years of Earth’s history are few and far between. This is not the fault of the paper’s authors: reliable as the data may be, there are a lot of gaps. There are almost no useful data from the time interval between 300 million and one billion years ago. If you consider only the most reliable ancient field data from very old rocks – as Biggin et al. do – it is still difficult to draw a convincing long-term trend, since the estimates of Earth’s magnetic field strength vary widely for any particular range of ages. Nonetheless, in the set of data discussed in October’s Nature article, the average magnetic field estimate for the time period between  1.4 and 2.4 billion years ago is lower than that of the chunk of time between 0.5 and and 1.3 billion years ago. So perhaps we should be looking at rocks between 0.5 and 2.4 billion years old for other indications that the inner core formed. We also need to work at sorting out long-term trends from short-term variations for these very old rocks.

The second problem with the magnetic field-inner core connection is that Earth’s magnetic field is sensitive to things besides the growth of the inner core. For example, we think that the geographic pattern of hot and cold spots at the boundary between the mantle and the core might influence geodynamo activity [5]. We see this effect in both the intensity and in the reversals of Earth’s magnetic field. The hot and cold spots are related to convective motion in the mantle and perhaps to old, cold chunks of tectonic plates that have sunk over hundreds of millions of years. This would mean that the strength of Earth’s magnetic field may change on the few-hundred-million-year time scales associated with plate tectonics. Biggin et al. average data over long time periods to get rid of these fluctuations – akin to averaging weather variations to look at climate – but they acknowledge that especially long-lived patterns of heating and cooling may influence their averages. The further we look back in time, the fewer assumptions we can make about plate tectonic activity and its consequences. A electrically conductive molten layer of rock that may have blanketed the core early in Earth history (a “deep magma ocean”) could also have affected the early geodynamo [6]. More importantly, although geophysicists have made incredible strides over the past decade simulating core behavior, it’s not entirely clear how the average intensity of the magnetic field is related to the power driving the geodynamo. My impression (as a non-modeler) is that modelers agree that if the power available to move stuff around in the outer core drops below a certain lower limit, the geodynamo would shut off. But does more power necessarily mean a stronger field? The intensity of Earth’s field is not the only indicator of geodynamo activity. We might seek clues from other aspects of Earth’s magnetic field, such as how often the North and South magnetic poles switch, the rate at which the poles move, or the degree to which Earth’s field is similar to that of a bar magnet (Biggin et al. do explore the latter issue to a certain extent).

All of this is to say that, while the Nature article represents the best of our knowledge right now, there’s still a log way to go before we can conclusively say when the inner core formed. We are much closer than we were when I was a graduate student (when I graduated in 2003, only about 36% of the studies used in this paper had been published [7]). This is an exciting time to be a paleomagnetist!


[1] Rubie, D.C., Nimmo, F., and Melosh, H.J., 2007, Formation of Earth’s Core, in Treatise on Geophysics, Elsevier, p. 51–90.

[2] The Earth may have got rid of some of its initial thermal energy before the core formed.

Stevenson, D.J., 2007, Earth Formation and Evolution, in Treatise on Geophysics, Elsevier, p. 1–11.

[3] Basically, radioactive Hafnium-182 preferentially goes into molten metals, whereas its decay product, Tungsten-182, stays in rocks. Comparison between Tungsten-182 concentrations in meteorites and the Earth indicates that the radioactive Hafnium-182 must have separated from the mantle about 50 million years after Earth initially formed.

Lee, D.-C., and Halliday, A.N., 1995, Hafnium–tungsten chronometry and the timing of terrestrial core formation: Nature, v. 378, no. 6559, p. 771–774, doi: 10.1038/378771a0.

Rubie, D.C., Nimmo, F., and Melosh, H.J., 2007, Formation of Earth’s Core, in Treatise on Geophysics, Elsevier, p. 51–90.

[4] The Nature paper that I’m discussing here is a meta-anlaysis, meaning that it analyzes data collected in many studies to draw a broad conclusion. The authors rate estimates of magnetic field strength from 53 studies on a scale (QPI) according to how many assumptions have been verified in the experiment and field work of the original study. Most studies published these days get about a 3 out of 6 on the QPI scale. More about the types of experiments – called paleointensity experiments – in a future post.

[5] Although there has been more recent work on this topic, here’s the original reference:

Glatzmaier, G.A., Coe, R.S., Hongre, L., and Roberts, P.H., 1999, The role of the Earth’s mantle in controlling the frequency of geomagnetic reversals: Nature, v. 401, no. 6756, p. 885–890, doi: 10.1038/44776.
[6] Ziegler, L.B., and Stegman, D.R., 2013, Implications of a long-lived basal magma ocean in generating Earth’s ancient magnetic field: Geochemistry, Geophysics, Geosystems, v. 14, no. 11, p. 4735–4742, doi: 10.1002/2013GC005001.
[7] Just for my own interest, I plotted up the number of studies and the number of magnetic field intensity estimates from the dataset that Biggin and co-authors use. The graph shows how many of these studies (and field estimates) of different quality indices have been published each year. There’s a definite shift from publishing papers with few assumptions justified (QPI 1 or 2) to those with more checks (QPI 3 or above)! I do wonder how many more rock units are available for study – sounds like a useful GIS project!
Composition of Biggin et al. dataset by QPI and publication year. The graph on the left counts the fraction of studies included in the Biggin et al. dataset; the graph on the right counts the fraction of paleointensity estimates. The year of my Ph.D thesis (2003) is indicated in red for comparison with today (2015). Note the large fraction of the QPI 5 and 6 studies and estimates published since 2003. Note that Biggin et al. average paleointensity estimates to counteract the over-representation of certain studies with large numbers of paleointensity estimates in the database.
Composition of Biggin et al. dataset by QPI and publication year. The graph on the left counts the fraction of studies included in the Biggin et al. dataset; the graph on the right counts the fraction of paleointensity estimates. The year of my Ph.D thesis (2003) is indicated in red for comparison with today (2015). Note the large fraction of the QPI 5 and 6 studies and estimates published since 2003. Note that Biggin et al. average paleointensity estimates to counteract the over-representation of certain studies with large numbers of paleointensity estimates in the database.


Blog Orientation

Reversals, Part 3

Two lava flows and their magnetic directions

Blog Orientation

Reversals, Part 2

I’m finally getting back to the blog after about a week of frantic magnetometry (we discovered a bug in our magnetometer software, because of which we had to measure lots of stuff all over again!) and report-writing. Here is another in my reverse-color series on magnetic reversals.

Why do we call them magnetic reversals? Because the way some lavas are magnetized, Earth's magnetic NORTH pole would have had to be where the SOUTH pole currently is.

Blog Orientation

Basics of Magnetism 4: Reversals Part 1

Earth has a magnetic field, which is what keeps your compass lined up with the North Pole [1]. The Earth’s outer core generates that magnetic field. You may have heard before that Earth’s magnetic field has, in the past, switched its North and South Poles. This is true, and kind of amazing and mysterious, but useful at the same time. This is the first in a series of picture-posts – not quite comics – that discusses magnetic reversals, and why and how we use them. I owe Maxwell Brown for this one.

A picture of a volcano with a Roman temple on top, and a story about building stones and magnetism


[1] Previous relevant posts are under the paleomagnetism tag.