LIQUID METAL BATTERIES
Mid 2013, F. Stefani (HZDR) presented the work of his group on the Tayler instability and saw it as a potential source of hazard in liquid metal batteries (LMBs). I had never heard of this instability nor of these batteries before, but found the idea to do some more applied MHD studies, very appealing.
LMB basics
A liquid metal battery is composed of three layers of fluid with different densities that lay on top of each other. In the top electrode, we find a light metal A often in pure form. Right under this layer sits a heavier salt that is chosen so that it can dissolve and transport A-ions. The heaviest, lower electrode is a liquid metal alloy B(A) with high affinity for the metal A.
The chemical processes involved in charging and discharging of LMBs are reversible redox reactions. When discharging, metal A dissolves in ionic form in the salt, liberating an electron. In the salt, the A-ions move towards the bottom layer, where they alloy with the metal B and consume an electron that moved through the external circuit. The entire process is reversed during charging.
LMB technology was invented in the 1960s and the initial purpose was to design cells that could be recharged by distillation. The technology was recently brought back to daylight, by Prof. D. Sadoway’s group at MIT, who saw in these batteries an interesting candidate for energy storage on power-grid scale. A that large scale, we need
1. Large battery systems with a simple design
2. Robust battery systems that are durable in time.
3. Cheap battery systems based on Earth abundant materials that are cheap and will remain so.
With their simple self-assembling structure (gravity), LMBs can virtually be scaled up in size without making the design more complex. Material interfaces being liquid, LMBs can in principle be charged/discharged for an infinite number of times. Finally, one can build them with cheap materials such as Li, Na, Pb, Sn as shown in Prof. D. Sadoway’s group.
To minimize the internal resistance of the battery, it is important to keep the badly conducting molten salt layer as thin as possible. We expect that 1 to 5 mm is a suitable height for the salt-layer. Another limiting factor is the maximal electrical current density J that can pass through the cell. Mg or Li -based LMBs, were efficiently cycled with maximal current densities (i) J=3 kA/m^2, (ii) J=10 kA/m^2. More exotic LMBs with chemistries based on non-Earth abundant materials have reached values as high as (iii) J=100 kA/m^2. For comparison, a copper wire with section 1mm^2, carrying a 1A current has a current density of J = 1 MA/m^2.
Flows in LMBs
As they are all liquid, flows naturally occur inside LMBs. Flows are desirable in the lower alloy layer since they help in mixing the alloy and avoid the formation of intermetallic phases. Flows will not be desirable when they become so intense that the layered structure of the battery gets compromised.
Since 2014 we study different types flows can be driven inside these batteries using the numerical code SFEMaNS. During his Phd, L. Capannera adapted this code to simulate multiphase MHD flows in LMBs. We use a momentum-based, phase field approach, that is explained here. In 2020, L. Cappanera further improved the multi-phase method implemented in SFEMaNS and details of this method will soon be published.
Click on the figures below to access pages that discuss our work on different types of flows.
Visit page on Tayler instability
Just sending a vertical electrical through a liquid metal is enough to destabilize a flow. We investigate if this can happen in LMBs. We derive scaling laws for the intensity of the flow and estimate the electrolyte layer deformation that is causes. Tayler instability may occur in moderate size LMBs, but electrolyte pinches are not immediately expected, not even in very large (several meters) cells.
Visit page on metal pad roll instability
Just as in Hall-Héroult cells, the presence of a background magnetic field can allow the metal pad roll instability in LMBs. This was shown by simulations done by N. Weber, in which we collaborated. These simulations made me realize that a precise theoretical model, including dissipation, was needed. Our two-layer model is explained and a benchmark of SFEMaNS with OpenFOAM was done on a small set-up. We can use the theory and numerical simulations to demonstrate the feasibility of a small metal pad roll experiment using gallium and mercury or GaInSn eutectic alloy as working fluids.
