Lava spattering from the Halema`uma`u lava lake at Kilauea on May 12, 2012. Image: HVO/USGS. I get a lot of questions here at Eruptions, but one of the more common themes is the properties of rocks - and specifically why they melt where they melt to produce magma? There are a lot of misconceptions out there about the interior of the Earth, namely that the tectonic plates that we make our home (both the continental and oceanic kinds) are sitting on a "sea of magma" that makes up the mantle. As I've said before, the mantle of the Earth, that layer of silicate rocks that starts at ~10-70 km depth and goes down to the outer core at ~2900 km depth that constitutes a large volume of the planet, is not molten, but rather a solid that can behave plastically. This means it can flow and convect, which is one of the ways that geologists have theorized that plate motion is started and sustained. However, as we know, rocks are found entirely molten within the Earth, so how can so much of the planet be solid but then some parts of it melt as well?
This sketch illustrates why rocks melt on Earth. The geotherm (solid line) would suggest that rock shouldn't melt as it never cross the dry mantle solidus (the point where mantle rock would melt merely by heating it). Adding water moves the solidus to the wet mantle solidus (short dashed line). Decompressing mantle at constant temperature allows for the mantle to cross the solidus as the mantle rises (thick solid line). See text for more details. Image: Erik Klemetti It starts with the question "how do you melt a rock"? The most straightforward way that might pop into you head is "raise the temperature!". That is what happens with ice -- it is solid water that melts when the temperature exceeds 0ºC/32F. However, when it comes to rocks, we run into a problem. The Earth actually isn't really hot enough to melt mantle rocks, which are the source of basalt at the mid-ocean ridges, hotspots and subduction zones. If we assume the mantle that melts is made of peridotite*, the solidus (the point where the rock starts to melt) is ~2000ºC at 2o0 km depth (in the upper mantle). Now, models for the geothermal gradient (how hot it gets with depth; see above) on Earth as you go down through the crust into the upper mantle pegs the temperature at 200 km at somewhere between 1300-1800ºC, well below the melting point of peridotite. So, if it is cooler as you head up, why does this peridotite melt to form basalt? Well, that is where you need to stop thinking about how to heat the rock to melting but rather how to change the rock's melting point (solidus). Think about our ice analogy. During the winter, there are a lot of times where you'd like to get rid of that
Sketch illustrating melting at a subduction zone. Water from the downgoing slab is released at depth as it heats, causing the mantle above the slab to partially melt, forming basalt. Image: Erik Klemetti ice but the ambient temperature is below the air temperature. So, what do you do? One solution is to get that ice to melt at a lower temperature by disrupting the bonding between the H2O molecules -- thus, halting the formation of rigid ice. Salts are a great way to disrupt this, so throw some NaCl or KCl on ice and it will melt at a lower temperature than 0ºC. For a rock, water behaves as its salt. Add water into a mantle peridotite and it will melt at a lower temperature because the bonds in the minerals that make up the rock will be disrupted by the water molecule (we call it a "network modifier"). In a subduction zone (like the Cascades or the Andes), where an oceanic plate slides down under another plate, that downgoing slab releases its water as it heats up. That water then rises up into the mantle above it, causing it to melt at a lower temperature and, bam! Basalt is produced in the process called
Sketch illustrating decompression melting at the mid-ocean ridge. Warm, fertile mantle rises, partially melts to form basalt, then moves laterally away from the ridge as it cools. Image: Erik Klemetti Wait! The largest volcanic system on Earth is the mid-ocean ridge system, where you don't have any subduction to bring water down into the mantle to help melting along. Now, why do you get basalt there? This time we have to use another method to melt that peridotite - we need to decompress it at constant temperature. This is called adiabatic ascent. The mantle is convecting, bringing hot mantle from depth up towards the surface and as it does so, the mantle material stays hot, hotter than the surrounding rocks. The melting point (solidus) of peridotite changes with pressure, so the 2000ºC melting point at 200 km is only ~1400ºC at 50 km. So, keep that mantle material hot and decompress it and you get melting to form basalt! So, underneath mid-ocean ridges (and at hotspots like Hawaii), the mantle is upwelling, causing decompression melting to occur. Let's review: Under normal conditions, mantle rock like peridotite shouldn't melt in the Earth's upper mantle -- it is just too cool. However, by adding water you can lower the melting point of the rock. Alternatively, by decompressing the rock, you can bring it to a pressure where the melting point is lower. In both cases, basalt magma will form and considering it is hotter and less dense than the surrounding rock, it will percolate towards the surface ... and some of that erupts! *The mantle is definitely not homogenous, but for our purposes, we're interested in what we call "fertile mantle" -- that is, mantle that hasn't experienced melting before and can produce basaltic liquid.