This transcript was generated automatically and lightly edited for readability.

Oliver Strimpel: This is Geology Bites with Oliver Strimpel. The Earth has three main parts: a thin shell called the crust; a much thicker rocky mantle that makes up 84% of the Earth's volume; and a central core with a solid inner part and a liquid outer part, both consisting largely of iron and nickel. The extreme pressures and temperatures of the core make this a very exotic place indeed. You'd think it would be hard to learn much about the nature of this ball of metal that's buried so deeply within our planet.

Bruce Buffett is a professor in the Department of Earth and Planetary Science at the University of California, Berkeley. Using both observations and theory, he has probed the Earth's core to learn a great deal about the motions and heat flow within it. Bruce Buffett, welcome to Geology Bites.

Bruce Buffett: Delighted to be here, Oliver.

Oliver Strimpel: Can we start with some basic facts about the Earth's core?

Bruce Buffett: Sure. The Earth's core occupies roughly the lower 3,000 kilometers of the planet, so this would be roughly half the radius of the planet. It also accounts for about 30% of the mass of the planet, even though the volume is only about an eighth. And so that tells us right away that the material that's in the core is fairly dense. We think that the material in the core is made up primarily of iron and nickel. We also believe that it has some small impurities that cause the density to be a little bit less than what we would infer from the properties of pure iron. The inner part, the inner third of the core, roughly the lower 1,000 kilometers, is solid, and the outer two-thirds then is liquid, with a viscosity that's probably comparable to water. So it's a very, very low-viscosity fluid.

Oliver Strimpel: How do we know all this?

Bruce Buffett: Well, a lot of the information comes from seismology. We have seismic waves that reflect from boundaries in the interior, and this gives us some picture of the geometry. We can also use seismic observations to make inferences about the density, and so this is what we use to compare against the properties of iron and other materials. The other thing that we can use seismology for is to prove, in fact, that the liquid part is indeed liquid. Seismic shear waves are known to not propagate through liquids, and it's been known for quite a long time that shear waves cannot be detected propagating through the Earth's core.

Oliver Strimpel: Just to explain, shear waves are transverse waves, in which the rocks shake sideways, perpendicular to the travel direction of the wave. Another kind of seismic wave is a pressure wave, or sound wave, in which the motion is along the direction of travel of the wave. So no shear waves can get through the liquid outer core to reach the solid inner core. But do we see other waves that can get through and bounce off the solid inner core?

Bruce Buffett: We do, actually. But the types of waves that propagate through the liquid part are purely sound waves. And so these sound waves propagate down, they bounce off the inner-core boundary, and they come back up as sound waves, so we can record those at the surface. But what we don't see is waves that propagate as shear waves, traveling along paths through the fluid outer core.

Oliver Strimpel: I think most of us know that the core is responsible for generating the Earth's magnetic field. How does it do that?

Bruce Buffett: Well, the process involves fluid motions in the liquid core, and these fluid motions lift and twist the magnetic field, and in the process generate electric currents that tend to reinforce the existing magnetic field. Now, the idea that these fluid motions would tend to reinforce the field is far from obvious. But what we've learned is that the influences of planetary rotation have a strong influence on the style of flows in the core, and those styles of flow, we know, are very effective at generating currents that reinforce the magnetic field.

Oliver Strimpel: You said that these lifting and twisting motions in the core fluid serve to amplify an existing magnetic field. So one needs a seed field to get the process started. Since the Earth formed in the magnetic field of the Sun, which in turn formed in the magnetic field of the Milky Way galaxy, that could have provided the initial magnetic field. But then what generates the motions in the core?

Bruce Buffett: So the motions are probably due to convection. The idea that the planet is gradually cooling with time means that heat is being drawn out of the core. The material at the top that cools down becomes dense and sinks down, and this continues to drive flows. Also, the hot regions in the interior are warm and buoyant as a consequence, and they will rise up and produce fluid motions. There's other sources of buoyancy that also drive flow. We know that as the core cools, it gradually solidifies from the inside out. That explains the presence of the solid inner core. And we also know that that solid tends to exclude some of the impurities that are in the liquid, and as these impurities are excluded from the inner core, they can contribute to light fluid, which will rise and also help to drive fluid motions.

Oliver Strimpel: I suppose the way the inner core excludes the impurities when it solidifies is a bit like the freezing of seawater to pure ice, in which the salt in the water is excluded. But in this case, the impurities are mainly silicon and oxygen, though some people have also suggested sulfur, carbon and even hydrogen. But how do we know that both these drivers of motion, then, the convection and the solidification and the chemical buoyancy, if you like, how do we know that they're both operating at the same time?

Bruce Buffett: Well, it's really a consequence of the heat that we're taking out of the core. So ultimately, the heat flow from the core into the base of the mantle is the thing that drives everything. And so, as I have mentioned, the fact that the fluid is cooling at the top means that it's going to be dense. But the other point is, as you take heat out, you grow the inner core. There's a direct response to that cooling that requires the inner core to grow by a certain amount, and that will produce a certain amount of chemical buoyancy. So once we know the heat flow at the top, we can predict what the thermal contribution is, and we can predict what the chemical contribution is. And it's really impossible to have one without the other. It's really just the question of what their relative importance is, and that's set by physical properties that we've been able to infer about conditions in the core.

Oliver Strimpel: So, walking backwards from the cooling rate that you say we know, because we can figure out how much heat is coming up through the mantle, does that take us back to a time when in fact there was no solid inner core at all?

Bruce Buffett: That's the case, actually. So we know that to produce the present-size inner core, we would have had to cool the core by roughly 200 degrees. And so the amount of time that it takes to cool the core by 200 degrees would be essentially the lifetime that the inner core has been present in the planet. And so different estimates depend on different values of heat flow, but it's very unlikely that the inner core was present from the very earliest times.

Oliver Strimpel: So roughly how far back do we think it goes?

Bruce Buffett: I guess our best estimates would be probably a billion years. The inner core is growing today at probably something on the order of a millimeter per year.

Oliver Strimpel: So before we had a solid inner core, and therefore when there would have been no compositional buoyancy effect to drive motions in the core, would that mean there was a much weaker magnetic field, or perhaps no field at all?

Bruce Buffett: Well, it doesn't necessarily. We think today that compositional buoyancy through inner-core growth can explain about two-thirds of the power that's required to generate the magnetic field. As we go back in time and the inner core disappears, we lose that source of energy. Then we would be relying entirely on thermal convection to make this happen. And so it's possible that as we go back in time, the rate of heat flow coming out of the core was higher, and therefore the rate of generation of thermal buoyancy would have also been a little bit higher. But there's no question that the field would depend entirely on thermal convection prior to the existence of the inner core.

Oliver Strimpel: But do we know for a fact that there was a field?

Bruce Buffett: We have evidence in the geological record, in rocks going back about 3.5 billion years, and possibly a little bit longer. So at least we know, going back for three-quarters of Earth's history, that the planet has had a magnetic field. And our best estimate is that it's probably had a magnetic field for most of its existence.

Oliver Strimpel: OK. So it seems as if figuring out what motions are going on in the core depends on understanding how the flow of heat coming out of the core is divvied up between, on the one hand, thermal conduction, which does not involve any movement of the core material, and on the other hand, heat transported by upward motion of the core material, whether that motion is driven by convection or by the compositional buoyancy resulting from the ongoing solidification of the inner core. To get a handle on the thermal conduction part, it must be important to know how efficiently the metallic material of the outer core can conduct heat.

Bruce Buffett: So it is very important; you're absolutely right. So if the mantle draws heat from the core, some fraction of that heat can actually be supplied by conduction, and the additional heat can be supplied by convection. And so it turns out, for the thermal conductivities that we typically estimate, at least nowadays, a significant fraction of that heat can be carried by conduction alone.

Oliver Strimpel: So where do those numbers come from?

Bruce Buffett: It's a combination of experiments, and it's also first-principles calculations, essentially quantum mechanical calculations.

Oliver Strimpel: What kind of experiments can we do? I mean, those are pretty exotic conditions down there, right?

Bruce Buffett: So the experiments are quite extreme. To actually reproduce conditions in the core, you would need pressures on the order of several hundred gigapascals, and you would need temperatures on the order of 4,000 degrees. So these are very extreme.

Oliver Strimpel: Like, how many atmospheres would that be?

Bruce Buffett: So it's going to be, what, 100,000 atmospheres, something like that. So it's pretty high pressures. And so the way that these experiments are typically done is in something known as a diamond anvil cell. These are two gem-quality diamonds, and a little sample chamber is created right at the tips of these two diamonds. And because the area is so small, you can exert a relatively small total force that produces a very large pressure. And because the diamond is transparent, we can shine laser beams and heat the sample up with lasers to get simultaneous high pressures and high temperatures. The point is that these have to be done on very, very small samples, and to do thermal conductivity experiments, or comparably electrical conductivity experiments, is a very challenging effort. People have gotten in their experiments to temperatures of about 4,000 degrees, and people have gotten to several hundred gigapascals, but it's just very challenging to do finesse-type experiments under those conditions.

Oliver Strimpel: And how do we manage to get the composition right?

Bruce Buffett: So the initial estimates really looked at just pure iron, and sort of that would be an end member. And we know that impurities are likely to cause a decrease in the thermal conductivity. And so I think we're currently at the stage of moving beyond pure iron materials and looking at some of these alloys. And so the influence of these impurities is still to be determined.

Oliver Strimpel: Moving back up to the surface and the magnetic field. If we approximate the Earth's magnetic field to that of a bar magnet, the north pole of this magnet comes out somewhere near the geographical North Pole, but not coincident with it exactly. And in fact, it drifts around quite fast, as much as a few tens of kilometers a year. Why is it so variable?

Bruce Buffett: Well, the same process that generates the magnetic field, fluid motions and convection, can also sweep the magnetic field across, over the surface of the core, and the magnetic field will tend, as a first approximation, to be transported by the flow. And so if you have strong magnetic field at the North Pole which is being swept around, the direction of the dipole can also be swept around. The fact that it's on the order of tens of kilometers per year is telling us also that the fluid motions that are responsible for those changes are also on the order of tens of kilometers per year.

Oliver Strimpel: Apart from the wandering magnetic poles, do we see other variations in the Earth's field?

Bruce Buffett: We do. We see much shorter-period fluctuations, periodic variations that have periods anywhere from really 10 years, perhaps up to 60 or 70 years. What's interesting is that with the recent development of satellite observations, we have a record now going back for about 20 years, so we're getting a long enough record that we can actually look at some of these short-period variations, relatively short-period variations, in more detail, and get some greater insight into what the origin of these features is.

Oliver Strimpel: And do we have any current theories to explain it?

Bruce Buffett: Well, certainly one idea is that these fluctuations are due to waves. And in particular, waves with periods of several decades can, in fact, be obtained if we have a layer of stratification at the top of the core. The stratified fluid has a different dynamic and supports a variety of different types of waves. And there's some indication that the fluctuations that we're seeing could in fact be due to these particular types of waves.

Oliver Strimpel: So just to be clear, by waves here, you mean physical bodily waves, if you like, of the material in the stratified outer layer of the core. And the stratification would consist of layers of different temperature, but also perhaps chemical layering, with lighter elements at the top. And why would the core get stratified?

Bruce Buffett: So there are a variety of reasons. You could imagine that as we draw heat from the core, the contribution to that heat can come from two places. It can come from conduction, by virtue of the temperature gradients in the core. And so the temperature gradient that's implicit in the radial structure of the planet will carry heat toward the surface, and it turns out that that contribution of heat is a significant fraction of the total that the mantle takes. The rest of the heat is carried by convection; this is part of the thermal convective process. And so it turns out that if the heat flow that the mantle takes is greater than the value you can conduct, then you need to supply convection to make up the difference. But if the mantle were to take slightly less than can be supplied by conduction alone, that heat will simply accumulate at the top of the core and create a stratified layer, which these waves could then propagate through.

Oliver Strimpel: So I suppose in the extreme case, you might imagine that the mantle would be a perfect thermal blanket and no heat whatsoever would get out. Then that would shut down both convection and the chemical buoyancy, because you wouldn't be losing any heat, so that there wouldn't be any freezing, if you like, of the center of the core, and then you just have a kind of stratification by density that would just go all the way through.

Bruce Buffett: Yep, that's exactly right. If the heat flow is zero, there would be no magnetic field; there would be no convection. And in fact, something analogous to that may be happening on Venus. But the idea with these stratified layers is that the heat flow's not zero, but it's some small number. And the point, I guess, is that if the heat flow begins to drop lower and lower and lower, you start with no stratified layer, and it becomes thicker and thicker and thicker until you get to the state that you've described, where that layer could extend across the entire core. We think on the basis of numerical models that if the layer exists, it can't be too deep and too thick, probably not much more than 400 kilometers. And the reason is that we start to lose small-scale features that we can presently observe in the magnetic field, and these would be absent in a stratified layer which was that thick. So layers perhaps on the order of 100 kilometers or so are perfectly reasonable.

Oliver Strimpel: OK. So we can explain these short-timescale, say 20- to 60-year, variations using periodic motions in these layered parts of the upper core. But there's another amazing periodicity that has been discovered, actually, for some time, which is in the Earth's spin. But the Earth actually varies the length of its day by three to four milliseconds over a 60-year period. Could the core be involved in that?

Bruce Buffett: Most people would say that it is. We do know, for example, that atmospheric fluctuations on seasonal timescales can affect the rotation of the Earth. As the winds spin up or become faster, the mantle will slow down to conserve angular momentum. And so in principle, the atmosphere could in fact be contributing to the length of day on a variety of timescales. But observations of the atmosphere tell us that the variations in the atmosphere on 60-year timescales are just not strong enough to explain the amplitude of the variations that we see at 60 years. And so the only other alternative is to be having flows in the core which are varying on sort of 60-year timescales, which are exchanging angular momentum with the mantle and producing these variations in the length of day.

Oliver Strimpel: How would they exchange angular momentum? Would that be like friction at the core-mantle boundary?

Bruce Buffett: It could be some sort of friction. It could be some sort of pressure effects acting on topography at the core-mantle boundary. There are a variety of mechanisms that could account for this, and in fact, I would say at the present time there's no clear consensus. But what we do know is that observations of flow in the core, based on variations in the magnetic field, do suggest that the flows in the core speed up and slow down, and in fact speed up and slow down by enough to accommodate and account for oscillations in the length of day that we see.

Oliver Strimpel: So continuing on with these temporal variations of the magnetic field: not only does the pole wander, and we get these 20- to 60-year fluctuations on smaller scales, but we actually get complete field reversals of the Earth's dipole from time to time. And in fact, that helped provide some of the clinching evidence for plate tectonics, when magnetic stripes reflecting these reversals through geological time were found in the seafloor on either side of the spreading ridge in the middle of the Atlantic. So what's happening when a reversal takes place?

Bruce Buffett: Yeah, that's a great mystery. I had mentioned earlier that the way in which the dynamo works is that fluid motions lift and twist magnetic fields, and that they produce electric currents that tend to reinforce the existing magnetic fields. If we did a thought experiment and just simply kept the convective motions the same, but changed the sign of the magnetic field, then those convective motions would still lift and twist the magnetic field, and they would also produce currents which tend to regenerate the magnetic field in the opposite, or negative, polarity. And so the point is, the fluid motions are fairly indifferent to the orientation of the field. Given a field, they will tend to act to amplify it and increase it. And so you could imagine that if you could reverse the magnetic field in some part of the core, and the fluid motions are amplifying that field in the opposite polarity, if that region of reverse generation could grow and essentially take over all of the core, then the magnetic field would in fact reverse. And something analogous to that must be going on inside the core.

Oliver Strimpel: Do we expect that kind of process to be chaotic, because these reversals aren't regular?

Bruce Buffett: No, they're quite irregular. We get them on average four times per million years, but the last one was 780,000 years ago. So they're quite irregular, and they're consistent, or at least compatible, with a process which is completely stochastic and random.

Oliver Strimpel: Since we're pretty confident that we know the basic physics of what's going on in the core, can't we just plug in the well-known equations of fluid motion and electromagnetism, which have been known for centuries, into a numerical model and just get a computer to generate the structure of the flows and the magnetic fields for us?

Bruce Buffett: That would be great to be able to do, for sure, and you're absolutely right. We completely agree on the physics. There's no ambiguity about what the equations are that we should be solving. The problem is that when it comes to putting in physical properties that are appropriate for the core, we simply can't accommodate or put in realistic values. And the reason, particularly for viscosity, is that it allows very, very small-scale motions to be generated, and we're just simply not able to resolve those small scales in our codes. And so what we do instead is use viscosities which are substantially higher, maybe comparable to warm asphalt instead of liquid water. And we use those models with the hope that some of the physical processes will be quite similar, and that we can draw useful insights from those models. But they are still very, very far away from being realistic.

Oliver Strimpel: Part of the problem of small scales arises from turbulence, where you get a cascade of structures moving down on all scales, including really, really small scales. And I think some of your research really focuses on trying to model what's happening with that turbulence right now.

Bruce Buffett: That's exactly right. Because the viscosity of the fluid core is so low, this cascade of fluid motions down to smaller and smaller scales can go to very, very small scales before viscosity starts to play a role and suppress those motions. So we have essentially two options that we can proceed with. We can either wait until the computers are big enough to solve the problem that we want, or we can think about ways of representing those small scales, representing their influence, without actually having to include them in our numerical calculations. And that's certainly an activity that we've been fairly active in within the last few years.

Oliver Strimpel: I've seen some beautiful pictures of results of the computer simulations we've been able to do today of the magnetic field. In fact, one of them is on the website. How much can we trust what we see in those?

Bruce Buffett: Yeah, it's a perennial question. I think there are many aspects of the dynamics that we are probably getting right. And maybe a point to remember is that as the viscosity gets smaller and smaller and smaller, at some point it will just simply not matter. And so at that point, we'll be pretty confident that the results are reliable and trustworthy. At the present time, we're moving in that direction. We're getting closer, but it's not really clear how much further we have to go to be close enough to be good enough. I would say that many people think that the physics that we're capturing with these models is good enough to make inferences about what the properties, the behaviors, might be at realistic conditions. There are certainly other people that are more skeptical, and more "just prove to me that this could actually be true."

Oliver Strimpel: In the rendering that I saw, there were coarse structures right at the boundary of the solid and the liquid core, maybe 50 to 100 of them around the circumference, and then getting finer and finer till they were almost too fine to see at the core-mantle boundary. Now, is that general picture something that we can believe?

Bruce Buffett: I think that that's a reasonable picture. There's reasons for the structures that we see in those models. And I think what we would imagine as we go to more and more realistic conditions is that the full range of scales will become essentially much, much smaller. So you'll see all of those structures that you saw at the lower resolutions, but then they would just continue down to much, much finer scales. And so the hope is that we capture the part of the flow which is most important for generating the magnetic field, and that perhaps the influences of the smaller scales will be less important. And this does come back a little bit to the turbulence modeling, because there are consequences of those small-scale flows that are sort of separate from the magnetic field generation that still matter. And so it's not like we can completely ignore them. But perhaps if we can resolve the ones that are most important and account for the influences of the ones that are less important, perhaps that's a good way forward.

Oliver Strimpel: For my final question: if I were to give you unlimited funds and resources to pursue your research, what would you use them for?

Bruce Buffett: Well, certainly, one of the things that I would love to see is a numerical calculation that could actually get close enough to Earth-like conditions that we can be confident that it's the right result. So perhaps a computer that hasn't been invented yet, or engineered yet. And maybe a second thing would be that the satellite observations that we've been collecting for the last 20 years or so are an extraordinary window into the dynamics of the core, and extending those observations for longer, to make better insights about the dynamics, in particular dynamics at even longer timescales, would be fantastic. So that would be the other place that I would throw infinite resources at the problem.

Oliver Strimpel: Bruce Buffett, thank you very much.

Bruce Buffett: Thanks very much for having me.

Oliver Strimpel: For more about Geology Bites, as well as pictures and diagrams that illustrate this podcast, you can go to geologybites.com.