The depth at which mantle materials melt depends on the elements present in the rock, particularly volatile elements such as carbon and hydrogen, and in turn impacts how much carbon is released during subduction and volcanic eruptions.
Even small variations in volatile concentrations can change the temperature of mantle rock melting, lowering it by hundreds of degrees at a given pressure. In a carbon-bearing mantle, melting should begin at much greater depth than it would without carbon. It is this corrosive effect of volatiles that Tobias Keller and Richard Katz (University of Oxford, UK) thought might have interesting implications for melt transport in the mantle. They published new modeling results recently in the Journal of Petrology .
In the laboratory, scientists probe the melting behavior of mantle rocks by exposing samples to pressures and temperature relevant to real conditions in Earth. Although these experiments are difficult, they are now standard, and show how small concentrations of volatiles can have significant effects on the melting temperature of mantle rock.
What happens to mantle melt after it is produced at depth? Laboratory experiments alone cannot address this, because the Earth processes at play happen at enormous scale and over very long timescales; melt transport in the mantle traverses about 100 kilometers over thousands of years. Laboratory experiments, by necessity, are much smaller in size and cannot recapitulate such vast time scales.
To understand the consequences of the corrosive nature of carbonated magma in the mantle, Keller and Katz extended existing physical models of melt transport in the mantle and developed new numerical simulations. These simulations, carried out on the Brutus supercomputer at ETH Zürich, Switzerland, model mantle melting and melt-transport in the presence of volatiles, in particular carbon dioxide and water. In each of the around 10,000 time steps it took to complete a simulation, ten equations were solved in each of the 65,000 pixels on the model.
The results show that volatiles have an important dynamic effect: their corrosivity carves high-flux channels into the mantle, through which much of the melt is transported. Inside these channels, melt flow rates and melting reactions are strongly enhanced, which has important implications for geochemical transport: melt on the inside of channels has a distinctly different chemical signature than expected from geochemical calculations assuming simple, distributed melt flow.
The image above shows a representative output from the simulation. The localized carbon dioxide flux transported in melt channels is shown in panel (a). Volatile-induced channeling is most effective around 80-50 km depth. The phase diagram in panel (b) shows the solidus and liquidus temperatures on top of the three-component space used to model the carbonated mantle. The three components are (i) dunite, the olivine-rich residue of mantle melting, (ii) mid-ocean ridge basalt (MORB), the product of carbon-free mantle melting, and (iii) carbonated (cMORB), the product of carbon-rich mantle melting, containing 20% CO2. Any rock or melt composition is modelled as a mixture of these three.
One of the key advances needed to perform these calculations is a thermochemical model of mantle melting that is: (a) complex enough to capture the chemical interactions of carbon or water, and mantle rocks; and (b), simple and robust enough to be incorporated into tectonic-scale fluid-dynamics simulations. To address this challenge, Keller and Katz introduced a method called the Reactive Disequilibrium-driven Multi-Component method (R_DMC, pronounced “Run Dee Em See”). This method combines ideal solution theory with linear kinetics, providing a simple framework for multi-component thermodynamics, a combination that Katz says is, “not bad meaning bad but bad meaning good."
The method scales to an arbitrary number of chemical components, but its strength is that it can capture complex mantle thermochemistry with only a few components. Keller and Katz used it to capture the sharp depression of the mantle melting temperature with carbon content. But, thanks to the flexible and accessible nature of the method, it could address many other problems related to mantle melting.
The paper describes the simulation results and the R_DMC method, paving the way for more sophisticated carbon transport models at mid-ocean ridges and subduction zones. Keller, Katz and collaborators are working on these, with a new manuscript on mid-ocean ridges promising “exciting new results on carbon degassing” according to Keller. As for subduction, Katz reports on progress but notes, “It’s tricky.”
Visit the lab website for more.