Multivalent thermal batteries : concept, development and advances

Abstract

Thermal batteries are single-use energy devices designed to deliver a one-time source of power. They can be adapted to fit the application, such as high currents and pulse loads, making them important in military and space applications where reliable power generation is required. This work has sought to understand and analyse the possibility of producing a thermal battery which does not rely upon traditional lithium chemistry, by finding suitable magnesium and calcium eutectic electrolytes and cell compositions to challenge the traditional thoughts surrounding thermal battery chemistries.

All-magnesium thermal batteries were firstly studied by selection of magnesium-containing halide eutectics, which were analysed in depth through a variety of techniques including PXRD, SEM, wetting and conductivity. Cells were constructed using the well-characterised FeS₂ cathode material to find the optimal chemistries, of which the cells containing eutectic in the anode and eutectic and carbon in the cathode exhibited the best performance with capacities exceeding 400 mA h g⁻¹. The discharge mechanism of the cells was deemed to be unclear, with multiple possibilities examined. An operando neutron powder diffraction experiment was carried out to elucidate a mechanism, and found that MgS, FeS and iron were formed, indicating dissolution and reaction in the eutectic melt as KCl crystallised out. Also, a change in the unit cell of FeS₂ was observed, indicating some solid solution formation.

Experience of the magnesium cell chemistry was transferred over to the calcium analogues. A similar approach was undertaken using the CaCl₂-NaCl eutectic salt, with a wide variety of experiments to explore the properties of the eutectic. The cells constructed for these tests were found to perform most optimally with a pure calcium anode and cathode of FeS₂ mixed with the eutectic with voltages in excess of 2 V and capacities of ~ 200 mA h g⁻¹. As the cells were not able to be optimised to reach their full discharge potential, a mechanism was derived from the PXRD analysis, which found that the conversion of the material proceeded to form CaS, but only reached around ¼ of the total theoretical discharge capacity of FeS₂, which is due to a number of factors. The operando neutron diffraction experiment proved to be less successful but did identify the presence of CaS and FeS during operation, suggesting a similar mechanism as the magnesium derivative.

Finally, several new cathode materials were tested against for these cell chemistries. CoS₂ is an alternative sulfide cathode material with greater temperature resistance, and obtained similar performances to the FeS₂ material, though the discharge mechanisms observed were also affected by the conversion of CoS₂ to CoCl₂ by dissolution in the electrolyte, in both the magnesium and calcium eutectics. ZrS₃ was also synthesised and analysed as a potential cathode material in both the optimised magnesium and calcium systems, which showed some appreciable performance. Cells were then analysed in a mixed-phase by using LiCl-KCl in the magnesium cathode material. The cells performed well and demonstrated a very long discharge plateau of over 300 mA h g⁻¹ capacity, but the discharge mechanism was different to what has been observed in literature. Instead, the formation of other phases was observed instead of the Li₂ZrS₄ spinel structure shown in literature, which is explained in greater detail.

Two phosphate-based materials were synthesised and characterised, Cu₃(PO₄)₂ (CP) and Na₃V₂(PO₄)₂F₃ (NVPF). Both materials were firstly studied against the lithium silicon alloy to understand their discharge mechanisms. The Cu₃(PO₄)₂ material exhibited a large sloping discharge plateau from 2.6 V, with a long discharge plateau at around 1.4 V. The discharge mechanism was deemed to follow literature procedure, converting into Li₃PO₄ and copper with a capacity of > 400 mA h g⁻¹. Substitution of copper for vanadium was found to increase the voltage of the material slightly but was not able to be substituted in significant levels due to the coordination of the copper sites in the material. At higher current densities, most of the capacity was retained and a higher initial voltage, closer to literature, was observed at the start of the discharges when the material was processed by ball milling. Multivalent cells were also explored and deemed to show some, but not ideal performance in terms of capacity and voltage.

Na₃V₂(PO₄)₂F₃ was also synthesised by two separate methods. Both powders showed a pure phase and then one tested against the lithium silicon alloy. The discharge plot identified a plateau at 1.6 V which corresponded to the insertion of 1 lithium into the material. The discharge mechanism of the material was not able to be identified, due to the amorphization of the material during discharge, but it was assumed from the electrochemical data that an insertion mechanism occurred to produce the reduced Na₃LiV₂(PO₄)₂F₃ material. The rate capability of the material was also analysed and was found to perform extremely well at high current densities. A magnesium-based cell showed an expected lower voltage with lower capacity, whilst the calcium cell exhibited similar voltage plateau to the lithium derivative, however with even lower capacity than the magnesium cell. This proved that the multivalent ions were unlikely to favourably react with the material.