№ 168 · earth science

A volcano runs on pressure, not fire

Nothing burns inside a volcano. Magma is rock that is hot enough to melt, and it glows because it is hot, the way a stove ring glows. What throws it out of the ground is gas. Water and carbon dioxide stay dissolved in magma while it is deep and squeezed, then come out as bubbles as the pressure on it falls. A volcano works like a fizzy drink being opened, not like a fire.

Nothing is burning

A fire is a chemical reaction: fuel joins with oxygen and gives off heat. Magma needs no fuel and no air. The heat it carries up is heat it already had from deep inside the Earth. The temperatures in the models are high: the basalt example in Marco Belan and Enzo Cataldo's conduit model starts at 1,350 K, about 1,080 °C, and the silica-rich magma in a model shown in a 2018 review by Mike Cassidy and colleagues sits at 886 °C. Rock that hot glows red to yellow with no flame at all.

There is no lake of fire underneath either. In a 2025 review, Madeleine Humphreys and colleagues write that much of Earth's magma is stored as crystal mush. Marie Edmonds and colleagues, in a 2019 overview, describe mush as a part of the reservoir with enough crystals to form a semi-rigid framework, with melt between them. They write that the evidence suggests large volumes of melt are not present in the crust, and that melt is spread through mush and drawn out shortly before eruptions. The picture is not settled. In 2025 Pierre Wawrzyniak and colleagues imaged a reservoir of more than 200 km³, 23 km below the island of Mayotte, holding 22–42% melt. They call the mush picture a default model, justified mainly by the lack of clear signs of melt-rich chambers.

Predict first

What mainly drives a violent, explosive eruption?

Check your answer

The third. Cassidy and colleagues write that the acceleration, break-up and explosive eruption of magma are powered by dissolved gases coming out of the melt and expanding once they form bubbles. Nothing is burning, and the weight of rock is what keeps the gas dissolved in the first place.

Gas dissolved in rock

Magma carries only a few per cent of its weight as dissolved gas. Edmonds and colleagues show water in basalt melts ranging from 0.1–0.3% at mid-ocean ridges up to 3–6% under volcanic arcs. How much water a melt can hold depends on its chemistry, its temperature and the pressure on it. For rising magma, the change that matters is the falling pressure: Cassidy and colleagues write that gas comes out as magma rises because the melt can hold less of it at lower pressure.

Modellers often use a simple rule for silica-rich magma: the water a melt can hold grows with the square root of the pressure. Anne Schöpa and colleagues use it with a constant that puts 6% water in the melt at 8 km depth, and note that between 30 and 150 MPa, roughly 1 to 6 km down, it is off from a fuller solubility model by an average of only 0.14% water. As magma rises, the rock above it gets thinner, the pressure falls, and the melt cannot hold all its water any more. The extra water comes out as bubbles of steam. Birnbaum and colleagues put it plainly in a 2025 paper: magma is moved up and erupted by buoyancy, and the buoyancy comes mostly from that gas coming out.

Why it turns violent near the top

A gas takes up more room as the pressure on it drops. So the bubbles do two things at once as the magma rises: more water joins them, and the steam already in them swells. The swelling wins near the surface. In the figure, magma with 4% water starts to form bubbles about 3.6 km down, but the bubbles fill three-quarters of the volume only in the last 420 m or so.

When the bubbles fill most of the volume, the froth can no longer hold together. It breaks into small pieces carried along by the expanding gas. This break-up is called fragmentation, and it is what makes an eruption explosive. Belan and Cataldo use a critical gas fraction they say is generally believed to be close to 0.75, while noting that other values are reported; the model in Cassidy's review uses 85%.

Interactive Magma rising up a volcano's conduit from 8 km down. Set how much water is dissolved in it and whether the gas stays trapped or leaks away. On the left is the conduit, with bubbles drawn as the gas comes out. On the right is the share of the volume that is gas at each depth. Move the marker to follow one parcel of magma up. Every control works with Tab, Enter and the arrow keys.

water dissolved at depth4.0%
bubbles link up at50% gas
follow a parcel at8.0 km
conduit gas share of the volume
Magma with 4% water rising fast.

Ooze or blast

Every rising magma loses pressure, so why do some volcanoes pour out lava quietly? The answer is whether the gas can get out of the magma before it fragments. For silica-rich magma, Cassidy and colleagues name the main controls as how sticky the magma is, how easily gas escapes, and the shape of the conduit. Together these set how fast the magma rises and how much gas it loses on the way.

Gas escapes once enough bubbles touch to form connected paths. Cassidy and colleagues give that threshold as somewhere between 30% and 78% bubbles in magma that is forming them, and Birnbaum and colleagues found in a conduit-shaped lab set-up that shearing the foam could connect the bubbles below 20%. Slow rise gives the gas time to leak away. Fast rise, faster than about 0.1 m/s, keeps the gas trapped and almost always ends in an explosion. Slow magma can still explode, though. Cassidy and colleagues describe Mount St Helens in 1980. Magma rose slowly into the mountain for two months, then the flank gave way. Taking that weight away suddenly dropped the pressure at the top of the magma, and a wave of falling pressure ran down into it, setting off runaway gas expansion.

In short

A volcano is powered by stored heat and dissolved gas, not by burning. Deep magma holds its water because it is under pressure. As it rises, the water comes out as bubbles and the bubbles swell, most of all near the surface. If the gas stays trapped, the froth breaks apart and the eruption is explosive. If it leaks out on the way up, the magma arrives as lava. Cassidy and colleagues called working out which will happen, and when, one of the biggest challenges in assessing volcanic hazard.

Where this comes from

  1. Shear-enhanced permeability development of magma vesiculating in cylindrical conduits reuse permitted with attribution
    J. Birnbaum, J. Schauroth, J. Weaver, J. E. Kendrick, A. Lamur, Y. Lavallée · arXiv:2505.10103 · 2025
    arxiv.org/abs/2505.10103
  2. A model for steady flows of magma-volatile mixtures linked only, not reproduced
    Marco Belan, Enzo Cataldo · arXiv:1207.4506 · 2012
    arxiv.org/abs/1207.4506
  3. Controls on explosive-effusive volcanic eruption styles (Nature Communications 9, 2839) linked only, not reproduced
    Mike Cassidy, Michael Manga, Kathy Cashman & Olivier Bachmann · 2018
    pmc.ncbi.nlm.nih.gov/articles/PMC6053376/
  4. Architecture and dynamics of magma reservoirs (Phil. Trans. R. Soc. A 377, 20180298) linked only, not reproduced
    Marie Edmonds, Katharine V. Cashman, Marian Holness & Matthew Jackson · 2019
    pmc.ncbi.nlm.nih.gov/articles/PMC6335485/
  5. Crystal mush processes and crustal magmatism (Nature Reviews Earth & Environment) linked only, not reproduced
    Madeleine C. S. Humphreys, Olivier Namur, Wendy A. Bohrson, Pierre Bouilhol, George F. Cooper, Kari M. Cooper, Christian Huber, C. Johan Lissenberg, Eduardo Morgado & Frank J. Spera · 2025
    escholarship.org/uc/item/2xk0n5hk
  6. Magnetotelluric evidence for a melt-rich magmatic reservoir beneath Mayotte (Nature) linked only, not reproduced
    Pierre Wawrzyniak, Fabrice Gaillard, Sophie Hautot, Joan Andújar, Pascal Tarits, Laurent Arbaret et al. · 2025
    pubmed.ncbi.nlm.nih.gov/41162718/
  7. Magma emplacement rates and porphyry copper deposits: thermal modeling of the Yerington batholith, Nevada (Economic Geology 112, 1653–1672; accepted manuscript) linked only, not reproduced
    Anne Schöpa, Catherine Annen, John H. Dilles, R. Stephen J. Sparks & Jon D. Blundy · 2017
    research-information.bris.ac.uk/ws/files/141670497/Schoepa2017Yerington_Word.pdf