In hindsight, continental drift should have been obvious. That the Earth moves up and down had been known for centuries, shown by the layered beaches of Sweden, the seashells of the Himalayas, or the sinking harbours of the Med. The drowned and resurfaced Pillars of Pozzuoli became famous as the frontipiece of Charles Lyell’s opus magnus, Principles of Geology (1833). If up and down is possible, why not sideways? The close relation between the coasts of Africa and South America was obvious to everyone. But science, ruled by conservative geologists, blocked this obvious step.
Don’t judge these senior scientists too harshly. The physics of continental drift is still far from straightforward. There was no known source of sufficient energy within the Earth. Convection in solid rock was unimaginable. The data spoke loudly, but in an unintelligible language. It needed translating and interpreting. Science stuck to things it understood. It kept its feet firmly on solid ground.
Only in the 1960’s did the tide change. The mid-oceanic ridges had been discovered, and the oceanic crust dated. The ages left no doubt that new crust was being created. Even the most conservative geologists could no longer deny the flood of evidence. Continental drift became accepted, 200 years delayed. Let it move.
We now know of 15 major plates, each moving independently from the others. Most of them contain a continent (the North America plate covers more than one) but some are purely oceanic. In between the plates are the spreading ridges and subduction regions. All active spreading ridges are oceanic, even if some (such as the Red Sea spreading ridge) started on land. The answer to how continental drift works lies underneath the sea.
See it move
The spreading rate of a typical mid-oceanic ridge is 5 cm/yr. As every book on the topic says, it is the rate at which finger nails grow. Nowadays we can measure this directly. Radio telescopes measure the exact direction to distant galaxies. Over the years, this direction changes as the ground underneath them moves. Compare two telescopes, and you can measure how the distance between them changes. This is accurate to millimeters. The figure shows the increasing width of the Atlantic, measured between two telescopes. You can see the slow, smooth movement, with an annual fluctuation. The rate of increase seems slow, but this is because the baseline crosses the spreading ridge at an angle and does not see the full movement.
But what is it that makes the continents move? The main action is clearly in the oceanic plates. But what force can make the surface of the Earth drift? Physics tells us that there are four different processes at work. In increasing order of importance, these are:
Magma injection The magma injected into a spreading ridge can push the two sides apart, and set the plates in motion.
Convection currents The convection currents in the mantle can themselves induce horizontal motion, because upwelling and downwelling occurs in different regions. The flow in the mantle can drag the crust along.
Gravity Spreading ridges are pushed up by several kilometers. Subduction occurs in deep trenches. Gravity causes material to move from high (spreading ridge) to low (the trench). This is called ‘ridge pull’. (It is also called ‘ridge push’ but this is misleading as there is no push from behind.)
Subducting plates As the slab sinks, it pulls the rest of the plate with it. This is called ‘trench pull’ or ‘slab pull’.
Slab pull is now considered the most important effect, followed by ridge pull. Horizontal convection currents in the mantle would affect the continents most, since they have deep keels sticking into the mantle, but this is not observed. Instead the continents seem to respond mainly to the oceanic crust. Ridge pull is important, but magma injection by the rising mantle, which would seem to be the most obvious way to drive plates apart, does not work as a force.
Spreading ridges and plumes
Mantle plumes are different from spreading ridges. They produce ‘hot spots’, seen in places such as Hawaii. They are deep, rising, convective cells, which reach to the surface and by their heat, push it up. Magma forms, and a tall volcano appears. It grows tall not just from the lava deposited on top, but also from the push from below. Once the volcano drifts away from the hot spot (the curse of continental drift: it is hard to stay in one place), the mountain quickly sinks again. This will be the fate of Hawaii. Hot spots do not form spreading ridges: Hawaii did not break the Pacific plate. This shows how ineffective magma injection is in setting plates in motion. However, sometimes a hot spot does break a continent, forming a triple junction which can develop into a new spreading centre. Africa has been susceptible to this. But it is rare, as shown by the fact that neither the Yellowstone hotspot nor the Deccan Traps broke their host continent.
There are cases where a hot spot coincides with an existing spreading ridge. Iceland is one example, the Azores and Amsterdam/St Paul are others. Are these accidental? If so, the hot spot should rather quickly drift away from the spreading ridge. But this is not seen. Iceland is at least 20 million years old, and it is still largely centred on the continental divide. The reason may be gravity pull. The hot spot pushes a region up. Gravity pull begins to act. If there already was a plate break nearby, the plate may now begin to break at this point, letting the old break nearby stop and heal. In this way, a hot spot can pull a spreading ridge along, at least for a while.
Hot spot volcanoes are common on Mars and Venus, as they are on Earth. But spreading ridges and continental drift only occur on Earth. We are unique.
The mantle extends from 100 km depth to almost 3000 km. It is solid material, not liquid, although it can move around as well as up and down. If this seems strange, think how sand can do the same. The mantle rock changes at a depth of 410 km, and again at 660 km. The higher one corresponds with a change from olivine to spinel, and the lower one a change to perovskite.
Subducting plates are cooler than the surrounding rock, and go through the first phase change a bit earlier. This increases their density while the surrounding rock is still of lower density olivine. This increases the downward pull on the subducting slab. Most of the slab-pull force comes from this phase change. Subducted plates often, but not always, stay above the deeper discontinuity: there is a ‘plate graveyard’ at this depth.
The phase changes can stop convecting cells from rising. Only strong convective cells break through them from below. If a cell gets stopped, it develops a broad plume head, creating a larger region of warmer material. So there may be two layers of convection in the mantle, separated by the discontinuity, with only a few hot, deep plumes being able to rise through both.
What causes the convection in the mantle? Convection forms when there is a large temperature gradient. The hot material at the bottom has lower density, and the buoyancy causes it to rise. Convection can be triggered either by heat from below, or cooling at the top. Mantle convection probably has both effects. The heat from below can come from the core (from the slow solidification of the inner core), or from radioactive elements in the mantle itself. Cooling at the top can come from subducting plates, or from water cooling of oceanic plates. The convection can be deep, from the core, or be in the upper mantle only, driven from the 410km phase discontinuity. Underneath continents, sideways convection can occur, from hot regions near the centre of the continents (where mantle heat is trapped) to the cooler edges. Perhaps these also happen from the warm mid-oceanic ridges.
Let’s try to put some numbers in.
The total length of the spreading ridges on Earth is a staggering 40,000 km. The oceanic crust is about 8 km thick. For an average spreading rate of 5 cm/yr, we need to create 16 billion cubic meter of basalt every year (length times width times thickness). The density of basalt is 2400 kg/m3, so that 38 trillion kg of basalt is added to the oceanic plates each year. Obviously, the same amount must be going down in the subduction zones each year, to keep the Earth in net balance.
The new basalt coming up from the mantle is hot. It cooled down while coming up but it is still hotter than the normal oceanic crust. Because it is warmer, it has lower density. This buoyancy pushes the mid-oceanic ridges up, to 3-4 kilometer above the ocean floor.
Let’s assume that basalt within the spreading ridge in the ocean has a temperature of perhaps 1300 C. As it moves away from the ridge, pulled in by slab-pull and by gravity, it cools and grows denser. Let’s assume that once it reaches 800 C, it begins to subduct. (The numbers are approximate but not far off. Obviously the surface is much cooler and these numbers are for the bottom of the crust.) The heat capacity of basalt is 0.84 kJ/(kg K): in other words, to cool 1 kilogram of rock by 1 degree centigrade releases 840 Joule of energy. Cooling by 500 degree centigrade releases 0.4MJ per kg of basalt, between upwelling and subduction. This is a lot of energy. For 1 square meter of ocean floor, and 8 km of crustal depth, the total energy to be lost is 8 TeraJoule (8 x 1012 J).
Rock insulates very well and it gives up this heat very reluctantly. The k value of basalt is 1.69 W/ (m K), which means that for each degree of temperature difference, a 1 meter thick rock layer lets out 1.69 W per square meter surface area. A 1 kilometer thick rock reduces the leakage to 1.69 milli-Watt. So how long does it take oceanic crust to cool down from 1300C to 800C? Take the average thickness as 4 km, and the temperature difference with the surface 800 C (the deep ocean being 4C). A quick calculation shows that the ocean floor near the ridge will radiate about 0.34 W per square meter.
At this rate, the cooling will take a million years. This sounds like a lot but in geological terms it is fast. However, although the initial cooling is fast, it slows down a lot later, when the temperature difference is much less. Cooling continues for typically 100 million years.
As the rock cools, it contracts and becomes denser, and the ocean floor begins to sink because of the contracting rock: the sea grows deeper over time. The approximate relation is that the depth of the ocean is 2.5km, plus 0.35 times the square root of the age in million of years. At zero age (i.e. the spreading ridge) the depth is 2.5 km. After 1 million years it is 2.85 km, and after 100 million years, the ocean is about 6km deep. After this it increases only very slowly. Therefore the oceans are least deep at the spreading ridge, and after about 1 million years, when the floor has moved by about 50km, it rapidly becomes deeper. The spreading ridge is therefore typically 100 km across.
The plate is actually much thicker where it forms, and thins further out as the movement speeds up. Eventually the oceanic crust cools so far that it become denser than the pressured mantle (aestenosphere, to be precise) below. Once this happens, the ocean plate begins to sink, like blobs in an old-fashioned lava lamp. Let subduction begins.
Subduction reheats the sinking plate. It comes in an area where temperatures are much higher and sucks up this heat. The sinking and the higher pressure itself also generate heat. After 100 million years of solitude, the plate can get cozy again in the homely warmth of the upper mantle, returning to its Macondo, and dissolve in the plate graveyard.
Volcanoes occur in two part of this cycle. The spreading ridge itself is effectively a 40,000 km long volcanic rift, producing gigantic amounts of basalt. Images of the ocean floor clearly show a rift, situated in the centre of the ridge. Most of the basalt is injected far below the surface; only a small fraction is erupted on the surface. Sometimes a ridge becomes overactive and floods the sea floor with thick layers of basalt, as a LIP (or Large Igneous Province). The largest such event known is Ojong Java Nui, which covered as much as 1.2% of the Earth, and was emplaced below the sea 125 million years ago. Ridge volcanic activity is well behaved, with stuff oozing out. The pressure of kilometers of seawater above is too high to allow for explosive ejection. These are the only types of volcanic eruptions which attract rather than destroy life.
This volcanic activity quickly dies down as the plate cools, and the plate becomes inert apart from the occasional hot spot or warm region it traverses on its journey. Where this happens, a volcanic island may form. All deep sea islands (apart from Australia) are volcanic in origin.
The second phase of volcanic activity occurs during subduction. The oceanic plate is solid. As it subducts, both temperature and pressure rise and a race begins. The melting point increases with pressure. If the temperature increases faster than the pressure, a bit of melting will begin. In practice, this may happen at certain depths, and the plate becomes solid again as it goes deeper still. The subducting plate probably contains water (the upper range of the oceanic plate becomes pretty wet, as you may imagine) and this lowers the melting temperature. Once rock melts, it becomes less dense and buoyancy pushes it upwards. The result is volcanic.
Let the dance commence
Therefore, continental drift begins with a cool surface slab descending into the mantle (‘the enemy’s gate is down’). This pulls oceanic crust with it, and some distance away the crust tears in response. The tear creates a gap, and the underlying mantle floats up to fill the gap, forming a new hot surface. Below, more material comes up from the mantle, carrying its heat with it. All activity happens below the sea. The continents are floating on the oceanic crust, passive and driven by the whim of the oceans.
Spreading ridges thus form by tearing crust asunder. Convective mantle cell are not needed. The mantle does have convective cells, forming hot spot volcanoes. As a rule, hot spot volcanoes have magma that comes from very deep in the mantle, pristine material that has not been to the surface before. Mid oceanic ridges have magma from shallow mantle sources, that has been through several cycles, lacking some elements that ended up in the continents. Subduction takes the plates down to the graveyard, but only rarely to the deeper mantle. The shallow mantle is thoroughly processed, mixed, baked, cooled and wetted. The deeper mantle, source of the hot spots, has had a more sheltered life.
What we see on the surface is the stately dance of the continents. Now we know that this is just a veneer, driven by the far more rousing dance of the oceans which pulls and pushes the continents along. In turn, this is part of a lively upper mantle. Our solid ground is floating on a sea of trouble. There is danger in the dance.
Geolurking, bless him, pointed out the real origin of continental drift