This transcript was generated automatically and lightly edited for readability.

Oliver Strimpel: This is Geology Bites with Oliver Strimpel. In the traditional picture of mid-ocean ridges, the lithosphere thins to more or less nothing along the spreading axis between two diverging oceanic plates. New oceanic material emerges from volcanoes along the ridge, which are supplied by magma generated by mantle that melts as it decompresses on the way up to the seafloor. But while this seems to fit the observations of some mid-ocean ridges, it is at odds with many others, especially the ridges that are spreading slowly. Instead, the plate thickness on the spreading axis can be as much as seven kilometers, and it is movement along faults in the plate at the spreading axis that accommodates the plate divergence. So what determines whether it is volcanism or faulting that occurs on mid-ocean ridges? Mathilde Cannat is a research director at the Institut de Physique du Globe de Paris. Her research on mid-ocean ridges has fundamentally changed our understanding of the geological processes that create new oceanic crust at these ridges. She also established that these processes vary dramatically between slow-spreading ridges and fast-spreading ridges, and even along the axis of a given ridge. Mathilde Cannat, welcome to Geology Bites.

Mathilde Cannat: Thank you.

Oliver Strimpel: Before we talk about our current understanding of mid-ocean ridges, can you remind us of the traditional view?

Mathilde Cannat: Initially, the ridges were described from the point of view of being volcanic chains. The new plates, or the new lithosphere, that was being formed would then move away from the region and become oceanic plates. The process was mostly volcanism: the space between the two plates as they moved apart would be filled by magma, and this magma would crystallize, depending on the depth, forming different types of rocks, gabbros, or dikes that would feed eruptions at the seafloor.

Oliver Strimpel: If that traditional view was the case, we would expect to see pillow lavas, dikes and so on everywhere, coming out of these mid-ocean ridges, but that's not what we see in all cases.

Mathilde Cannat: We do actually see pillow lavas pretty much everywhere at the seafloor, but what really differs between different ridges, and particularly between fast and slow ridges, is what's underneath: whether these pillow lavas are sitting on a regular system of dikes that have fed the eruptions, pretty much always at the same place, which you could really describe as the ridge axis, like a linear frontier between the two plates, and then a continuous domain of magmatic crystallization, like a magma chamber or a set of successive magma chambers, and then you would go into the mantle. And all of this would occur without much tectonic deformation, without much faulting. What we discovered was really the role of tectonics and faulting in this process. And this role is important. It's not two plates that are separated by a line in which there is no plate. The process of forming the new oceanic lithosphere comprises magmatism, for sure, but it also comprises a process of forming a new plate in between the two diverging ones. And what controls the thickness of a plate, pretty much everywhere, is really the temperature, because rocks behave differently with temperature. The hotter they get, the easier it is to deform them; they can deform under very low stress. The notion of a plate is really a mechanical notion, something that's more rigid, and therefore it's controlled by the distribution of temperature. And so at your ridge, the balance of heat supply and heat extraction is such that your geothermal gradient will be colder or hotter depending on this balance. And if you have a colder geothermal regime, it means that the rocks at a given depth will be cold enough to be more rigid and unable to deform other than by faulting. You said in your introduction that typically at slow ridges, this depth at which the rocks would be brittle would be 7 kilometers. It's actually often much more; it can be up to 15 kilometers. And we know that because we have recordings of microearthquakes along the ridge axis in different regions, and we see that we record microearthquakes down to depths of 15 kilometers. So what it means is that just at the axis, between the two plates that are being formed, you don't have a free zone with very hot material coming up all the way to the seafloor. You've got a ten-kilometer-thick plate that has to be broken if you want to separate the plates and make them diverge. And that's the fundamental new process that you have got to introduce.

Oliver Strimpel: I want to go back to the nature of the seafloor. When we talked about the traditional mid-ocean ridges as being volcanic, that produces one kind of ocean floor, one kind of plate surface. What do we actually observe? Do we actually observe different kinds of seafloor? And did you conduct some of these observations yourself?

Mathilde Cannat: I did. I spent a lot of time on ships; I've been on over 35 cruises. So it's a lot of observation, and a lot of fun also. So we started by saying that we would pretty much see pillow lavas everywhere. It's not completely true, and it's related to this great thickness at the axis. In some situations, the thermal regime at the axis is such that you've got a thick plate, and then you will have to have large faults to separate this thick plate if you want the plates to diverge. And these large faults are large in that they reach down to the base of the brittle lithosphere, but they also have large offsets. That's because they can be active for quite a long time, up to two million years, and with a typical plate tectonic divergence rate of a few centimeters per year, that represents a lot of offset. And these faults therefore take rocks all the way from the mantle and bring them up to the seafloor. But while they are doing that, there is also some volcanism and magmatism involved, and as the rocks are being uplifted by the faults, they're also crosscut and intruded by magmatic products. And therefore, what you get at the end is a lithosphere that is composite. It's made of a tectonic mixture of mantle-derived material, or peridotites, that will be altered by interaction with seawater; the alteration of peridotite is called serpentinization. So you will have serpentinized peridotites mixed with magmatic products, both tectonically and because they are intruded by them. And so underneath this cover of pillow lavas that you see when you explore the seafloor, depending on the thermal regime and also the amount of magma that has been supplied to the ridge axis, you should expect to have a mess, really, of different types of rocks, variably deformed and intruded by magma at different stages as they were uplifted.

Oliver Strimpel: Is that what we actually see, then? You see some pillow lavas interleaved with peridotites, interleaved with other kinds of rocks?

Mathilde Cannat: Yeah. The pillow lavas usually rest peacefully on top of everything, so they're unbroken; they are nice. In the few places where you are able to see what's underneath, you see it's quite complex. By contrast, at fast-spreading ridges, the thermal regime is much hotter. One of the reasons why it's much hotter is very simple: it's just that a given width of axial region at a fast-spreading ridge is going to be much younger, so it will have had much less time to cool than the same width of axis at a slow ridge. That's one of the reasons why the thermal gradient is so much hotter at fast ridges. And so at fast ridges, our estimates of brittle lithosphere thickness right on axis can be as small as 1 or 2 kilometers only. And so there are some faults, but they are very minor, and what takes over is really the magmatic processes.

Oliver Strimpel: So you've been, as you say, on many observational trips to study the seafloor. Can you tell us about the techniques that you use to map the seafloor, and whether you actually get detailed topographic maps, or what the actual product of these observational trips is?

Mathilde Cannat: Usually we combine many techniques, to go from direct observation of seafloor geology with a submersible, and sampling, to geophysical investigations like seismic experiments. Of course, these are not at the same scales. The bathymetry that allows us to map the topography is a unifying method that allows you to have a large map, and then maybe nest more detailed maps, acquired with submersibles near the seafloor, where we can get a much better resolution. And we also use imagery, using the capacity of the seafloor to reflect acoustic signals, because it depends on the type of rocks and on the physical properties of the rocks, so we can infer some geology from that. And we also use gravity and magnetic studies, because it's pretty frustrating when you are a geologist to work on submarine environments, because you can never really get there and walk there and see the landscape.

Oliver Strimpel: The bathymetry, is that using sonar? How do you get the topographic map?

Mathilde Cannat: You get the topographic map by using acoustic signals that you send to the seafloor. They are reflected and they come back, and the time they take to go down and then come back can be used to calculate the depth. This technique also provides you the information to calculate the capacity of the seafloor to reflect the acoustic wave.

Oliver Strimpel: So the better the reflection, the smoother the seafloor?

Mathilde Cannat: No. Actually, the better the reflection, the more chaotic the seafloor, because if the seafloor is very chaotic, there will be many orientations possible to reflect the signal. And if it's very flat, it won't be as efficient.

Oliver Strimpel: Was it the observations of the seafloor that first made you question the traditional view of mid-ocean ridges as chains of volcanoes?

Mathilde Cannat: Yes, it's a combination. It's the observation of the seafloor, because in the early '90s we did a lot of submersible explorations of the seafloor at the Mid-Atlantic Ridge. It was known that there were places where you had outcrops of serpentinized peridotite at the seafloor of slow ridges, but people interpreted these as kind of weird things related to transform faults, or related to diapirism. But we were able to observe that these were in fact very massive rocks, and that they did actually correspond to positive gravity anomalies, and therefore couldn't possibly be diapiric, because a diapir has to be lighter than its surroundings, and they looked like they were actually more massive than their surroundings. But it was also putting together the pieces that had been available for quite a few years, actually, concerning observations of microseismic activity at the Mid-Atlantic Ridge, showing, for example, that you had a seismogenic layer, and therefore probably a brittle layer, that was at least eight kilometers thick. And also, there had been some modeling work done in the late '70s and early '80s, following a great French–American experiment on the Mid-Atlantic Ridge, and following that, some people pointed out that the axial valley of the Mid-Atlantic Ridge had to be a tectonic rift. And in order to make a tectonic rift, you need to have a thick plate. You don't make a rift if you have just no plate at all at the axis. So it was kind of piecing these things together.

Oliver Strimpel: Okay. So you said that, in addition to volcanism, the main alternative mechanism that goes on at these mid-ocean ridges is faulting, and I presume you mean extensional faulting, because these things are spreading.

Mathilde Cannat: Yeah.

Oliver Strimpel: But just naively speaking, when you have a normal extensional fault going on, you'd expect the lithosphere to become thinner, following from conservation of volume or mass. How does that work?

Mathilde Cannat: I think the key is to consider the faulting and the plate thickness as two separate things. The faulting is there because we have two plates that want to move apart and that are linked by a plate, so this plate has to be faulted. And the thickness of the plate is not related to faulting; it's primarily related to the thermal regime at the ridge axis. It's like a continually renewed plate that you have to fault. So if you have a certain spreading rate, your domain stays at the ridge axis for a given length of time, and you have a certain effective cooling rate, if you want, that will be a product of how much heat you get and how much you remove. Most of the heat will be provided by magma, so it will be related to the magma influx, and most of the cooling is by hydrothermal circulation. So I have this cooling rate for my ridge, and this cooling rate tells me how thick my plate is going to be, and it will remain so, because it will continuously be cooling. So in other words, your fault takes material up, but it doesn't take the isotherms up, because the position of the isotherms is a function of the cooling rate at the ridge axis. It's one of the things that differentiates a divergent plate boundary from a diverging region somewhere else on Earth.

Oliver Strimpel: So that's fascinating. This is the crucial mechanism that you really discovered, which is like a continuous process. The extensional fault removes the crust as it extends, but the rising mantle below underplates the bottom of the lithosphere in a continuous process. So the more you remove, the more new rock you attach to the bottom of the lithosphere, so it just keeps on going forever.

Mathilde Cannat: Yeah. You attach it because you have uplifted it. Because you were removing material with the fault, it has attracted new material to fill the void, if you want. And when it gets there, it's being cooled, because the ridge has a given thermal state.

Oliver Strimpel: It's like a factory with a conveyor belt moving newly minted material along the footwall of the fault.

Mathilde Cannat: Yeah, exactly. It's a conveyor belt.

Oliver Strimpel: So could extension along a given fault proceed forever?

Mathilde Cannat: Observation shows that it doesn't. These large-offset faults are typically active for one to two, in some instances three, million years, but no more, and a new fault is created to take over. For a long time, it has been thought that this was due to flexure of the footwall of the fault, so the compartment underneath, which is being uplifted. As it's being uplifted, the weight of the overlying, overriding compartment is being removed, and therefore there is a kind of flexure, and this flexure kind of pushes on the fault surface, so it makes faulting more difficult. And at some point, mechanically, it becomes easier to just break through and start a new fault than to continue. So, to answer your question, we don't know for sure, but what we know is that these faults don't last more than two to three million years, which corresponds to offsets, depending on the spreading rate, of 10 to 25 kilometers.

Oliver Strimpel: So what you've really come up with is that there are two mechanisms, fundamentally, that accommodate the spreading. There's the volcanism, which would be the actual volcanic injection of a dike, maybe, and then there's the faulting, where you get the plates essentially sliding in a brittle fashion past each other, and extending. So what determines whether a given mid-ocean ridge will extend by faulting as opposed to magma injection and volcanic activity?

Mathilde Cannat: It's not really whether it's going to be one or the other; it's whether there is more of a role for faulting and less for magmatic activity, or the opposite. You have all the different shades between the two extremes. The main factor is melt supply, because if there is a lot of melt being supplied, it will have two effects. The melt will contribute to filling this void between the two plates that's opened by the plate divergence, and therefore you won't need to have so much displacement on faults. So that's one factor. And the second factor is that the melt will carry heat, and if you bring more melt, you will therefore have a hotter thermal regime, and therefore your plate will be thinner, and therefore your faults won't reach as deep. The controlling factor is definitely melt supply, but it does it through two different mechanisms.

Oliver Strimpel: So the more melt supply, the more volcanism?

Mathilde Cannat: Yeah, the more melt supply, the more volcanism.

Oliver Strimpel: And what determines how much melt supply comes to a particular spreading axis?

Mathilde Cannat: So that's another of these questions that still need to be the focus of work. We know that at slow-spreading ridges, there is clearly a tendency for the melt supply to vary along the axis of the ridge. If you go along the ridge over a distance of typically 50 kilometers, you will go from a region where the melt supply is lower to a region where the melt supply is higher. It shows that the melt supply is focused to locations along the ridge axis that are typically 50 kilometers apart. The origin of this is not really well understood. One possibility is that it comes from the melting process in the mantle, and the other possibility is that it's actually a loop. If the axial plate is thicker in places where you receive less melt, because your thermal regime is colder, and it becomes thinner towards the places where you receive more melt, then you create a topography at the base of the plate along the ridge axis. And the base of the plate is a change in rheological behavior. And so we expect that any melt coming underneath this topography, because it's lighter than the mantle, goes up, and it will be easier for it to go up the slope towards the center of these magmatic regions than to go into the less magmatic areas. So it's also possible that the supply of melt from the mantle is completely homogeneous along axis, but that the melt is focused by the changing lithospheric thickness, in which case it would really be a loop, because the changing lithospheric thickness is a result of the melt supply.

Oliver Strimpel: You distinguish between so-called slow ridges and fast ridges. First of all, just tell me what spreading rates correspond to these types of ridges, and could you give me an example of each type?

Mathilde Cannat: The only fast ridge we have on Earth that is a big ridge, aside from the back-arc basins in the western Pacific, is the East Pacific Rise. The spreading rates are typically 9 to 16 centimeters per year, and that's definitely fast. The Mid-Atlantic Ridge is our best example of a slow ridge. There, the spreading rate is between less than 2, if you go all the way north of Iceland, and about 4, if you get to the equatorial Atlantic, so that would be slow. And then we have ridges like the Southwest Indian Ridge. We call them ultraslow, because it's less than 2 centimeters per year.

Oliver Strimpel: What determines how fast a particular ridge is spreading? Is it processes that are happening locally?

Mathilde Cannat: No, it's definitely due to other factors. The spreading rate is really linked to far-field stresses acting on each of the separating plates. There are many types of forces that act on the plates, but most likely the reason why the East Pacific Rise is fast is because the Pacific plates are bounded by subduction zones on both the west and the east. One of the main driving forces of oceanic plate tectonics is the weight of the subducting plates, because they are colder than the hot mantle they get into, and therefore they are heavier, and this weight pulls the plates faster.

Oliver Strimpel: So the plates are pulled apart by subduction zones that are at the opposite end of the plate, where the plate is being destroyed.

Mathilde Cannat: Exactly.

Oliver Strimpel: So thousands of kilometers away.

Mathilde Cannat: Yeah, so that's why we talk about far-field forces.

Oliver Strimpel: Is this another factor that determines the spreading mode? Does the speed of spreading affect whether you see a predominance of the volcanic type of spreading as opposed to the faulting type of spreading?

Mathilde Cannat: The answer is yes, definitely. It's really related to the age of the axial region. When the plates spread very fast, your axial region, if it's 10 kilometers wide, for example, is going to be much younger on average, so it will have had much less time to cool than at a slow-spreading ridge. And because this cooling rate is the controlling factor for the plate thickness, the axial plate thickness at fast-spreading ridges is much less than at slow ones. And so the frontier between what we call fast and slow most likely corresponds to the spreading rate at which, given the melt supply, the plate is so thin on axis that you don't need to have faulting, I think.

Oliver Strimpel: So generally speaking, the faster the spreading, the thinner the plates, because they don't have time to cool, and the more volcanism you will get, as opposed to faulting.

Mathilde Cannat: Yeah.

Oliver Strimpel: You mentioned that hydrothermal cooling plays an important role in these mid-ocean ridge processes, because it's the primary way in which heat can be removed from the region at the lower end of the forming plate. But there may also be a wider role in the interaction between the crust and mantle and the ocean itself, and indeed the biology that's occurring at black smokers. Can you talk a bit about the hydrothermal processes and how they couple all these systems together?

Mathilde Cannat: We've learned that the very vigorous hydrothermal circulation that you have at black smokers is characteristic of places along the ridges where you are currently in a phase of intense magmatic activity. In other words, you get these black smokers because the seawater that gets in through the fractured upper crust or lithosphere is able to reach very hot magmatic rocks, or even magma. And it's important because, as you said, hydrothermal cooling is key to the thermal regime at the ridge axis, but it's also key to all sorts of interesting biology, and also key to very important heat and chemical exchanges between the solid Earth and the ocean.

Oliver Strimpel: So you mentioned the variation in space along the axis of a given ridge: maybe over 25 or 30 kilometers, you will get a variation from primarily volcanic to primarily fault-dominated. But in a given location, does it vary between the two modes? Do you think there might be a migration of the spreading mode along the ridge axis?

Mathilde Cannat: It does definitely vary. There are some places that have been fault-dominated and become magma-dominated. The cause of this evolution, we don't really understand. But the other objective is to understand the composition of the oceanic lithosphere at large, and what proportion of it may be fault-dominated. And it also has an impact on the recycling of the plates into subduction zones, because our planet has been tectonically active, with plate tectonics and ridges and subduction, for at least 2.5 billion years. It means that there have been several cycles of subduction, and therefore the mantle of the planet is made of a lot of relics of old oceanic lithosphere that has been recycled into the mantle.

Oliver Strimpel: In order to understand the composition of the mantle, you need to know how much of the material from the sediments and from the water is recycled back into the mantle at subduction zones.

Mathilde Cannat: Yeah, and also, if you have a subducted piece of oceanic lithosphere that had been created in a fault-dominated environment, you will have a lot of serpentinized peridotites all over the place, and this is not at all the same composition as magmatic rocks, and it hasn't exchanged matter in the same way with hydrothermal fluids. So it affects the composition of the recycled material.

Oliver Strimpel: I see. So just to explain: when water at high pressure and high temperature reacts with peridotite, you get serpentine, and that's a mineral that has a different chemical composition from the original peridotite, so you would expect to see different chemistry.

Mathilde Cannat: The composition is not that different; mostly you add water. But it's mostly that this serpentine is there at crustal levels. If you model your subducted material as being made of a crust with basaltic products, and in fact it was made mostly of mantle-derived material, the composition is very different, because the mantle-derived material is just a residue from melting, and the magmatic material is the product of melting, so it has concentrated all sorts of what we call incompatible elements. So the composition of the recycled material will be very different.

Oliver Strimpel: The standard oceanic lithosphere has a structure that's been studied in ophiolites, for example in Oman, where oceanic lithosphere has been emplaced onto continental lithosphere. It consists of pillow lavas on the seafloor, below which is a dense network of dikes, referred to as sheeted dikes, below which are gabbros; then you get the Moho transition zone, and then a mantle sequence of various kinds of peridotite. Do the spreading mechanisms that you've just discussed generate this structure at fast ridges?

Mathilde Cannat: Yes. We've never been able to really look at the crust formed at the East Pacific Rise, because there are very few faults, but what we've been able to see looks very much like the Oman or Newfoundland ophiolite sequence. And it makes a lot of sense, because in order to create such a little-faulted sequence, you need to be in a spreading environment with a very thin axial lithosphere.

Oliver Strimpel: What are some of the most important questions about mid-ocean ridges that confront us today?

Mathilde Cannat: Definitely the one that concerns the mechanism that modulates the melt supply to the ridge axis; that is still very much an open question. And then there are many questions remaining to understand the interplay between faulting, volcanism and hydrothermal circulation, I would say more with a view to shorter timescales. So far, we've been talking mostly about rather long timescales, like a few hundred thousand years or 1 million years, but a volcanic eruption, even a major one, can occur in a matter of a few days or a few weeks or a few months. And hydrothermal cooling, the vigorous black-smoker type of hydrothermal circulation, can extract enormous amounts of heat very fast. These are processes that can operate over very short timescales compared to plate spreading, and we know very little about the time variability of magmatic processes.

Oliver Strimpel: Mathilde Cannat, thank you very much.

Mathilde Cannat: You're welcome. It was a pleasure.

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