Much of the island of Tasmania is covered in basalt, which sometimes shows columnar jointing
There are a few processes in geology that are so fundamental and thoroughly taught that we forget that they are counterintuitive to normal people. One of these is mantle melting. The Earth consists of compositional layers of increasing density from surface to center. Near the surface is the crust, which is about 5-70 km thick. The crust has a complex composition I don’t want to get into now, but it is mostly silicates- complex metal oxides that include at least some silicon. Below that is the mantle. The mantle is mostly magnesium silicates, with some iron substituting for magnesium, and a few less common calcium and aluminum bearing minerals. The core is metallic iron, and is mostly molten. This, in itself, is fairly intuitive.
When the mantle partially melts, the resulting magma is less dense, and rises either to the base of the crust, or through the crust to the surface. For reasons I’ll skip for now, the composition of this partial melt is different to the unmelted material left behind. This mantle-derived magma* is known as basalt, and a lot of the Earth’s crust is made of this material. You can see basalt above, in the post before my last one, in volcanoes like Hawaii. Mot of the ocean floor is covered in basalt. So, when the mantle partially melts, we get a molten rock called basalt. And that is not particularly unintuitive either.
The tricky bit is this: Unlike most of the melting we see in our daily lives, the melting of the Earth’s mantle is never caused by heating it up. Most of the mantle is solid rock, and it melts fairly regularly, but most of the processes that melt it actually cool it, instead of melting it.
We geologists are so used to this that we don’t bat an eye, but for normal people, melting without heating seems a tad unusual. But then, the mantle is quite a different place than the kitchen counter.
The mantle is very hot, fairly dry, and under extremely high pressure. This pressure ranges from a few thousand times atmospheric pressure at the top of the mantle to about 1.3 million times atmospheric at the bottom. And one of the effects of increased pressure is that it also increases the melting point.
In fact, most of the solid mantle is so hot that, if you suddenly released the pressure around it, it would spontaneously melt. It is only the pressure that keeps it solid. The mantle can also flow- at high temperatures and pressures, solids become slightly ductile, and over long periods of time, the mantle can ooze around at speed of a few centimeters each year.
When a very deep, very hot part of the mantle rises close to the surface, if it rises faster than it can cool, it will generally start to melt once the pressure drops to around 15-25 thousand atmospheres. This is called decompression melting, and is the main cause of basalt magmatism in mid-ocean ridges and hotspots.
In some cases, mantle rises so slowly that it cools faster than it rises. When this happens under a spreading center, then you get ocean floor made of mantle, not basalt, because no melt was produced. This is rare, but there are known occurrences, mostly at very slow spreading ridges, such as the Arctic Ocean or the ridge between Africa and Antarctica. An expedition to a newly discovered crustless region was recently summarized by the rockbandit here.
The second main cause of melting is caused by water. When ocean floor sinks down into the mantle, it can carry water with it, which will eventually escape into surrounding mantle. Wet mantle has a lower melting temperature than dry mantle, so the introduction of water into warm dry mantle triggers melting. This is what produces arc volcanism above subduction zones in places like Japan or Chile, although these wet magmas sometimes interact with the overlying crust in ways that changes their composition so that they are no longer basalt.
They key point is that unlike the melting butter or ice at home, melting in the mantle is not caused by the addition of heat. It is caused by lowering the melting temperature of material that is already hot.
*By far the most common one. There are some rare, volumetrically insignificant mantle melts that are not basalts, but we can ignore them for now.