I have written several blog posts in the past about the potential use of propylene carbonate (PC) as a potential non-aqueous solvent in Zn-Br batteries. However, through my research I have now discovered that this solvent will not work in these devices as a cathode electrolyte, due to the way it interacts with the chemicals that are generated within the cell. In this post I will explain to you the experiments I did and why I reached this conclusion.

The idea with using PC was initially to completely replace the electrolyte within the cell. This was discarded right away due to ZnBr₂ solubility issues and low conductivity issues of the constructed cells. The idea then evolved to using a tetrabutylammonium bromide (TBABr) saturated PC solution (PC-TBABr) as an organic layer in an inverted device, since this layer can rest on top of a Zn-Br solution and can even remain on top after mixing if the Zn-Br solution is concentrated enough (>4M).

The organic layer is completely immiscible with the highly concentrated ZnBr₂ layer and it was my hope that the TBABr3 produced in the cathode would be substantially more soluble in the PC-TBABr compared to the aqueous phase. Given that the PC phase is more than 50% TBABr, it seemed very likely that the produced perbromide would have a significantly higher affinity for the PC-TBABr.

To confirm whether this was happening, I constructed an inverted cell using a glass vial in order to be able to see what was going on (which you can see above). I placed a zinc anode at the bottom, used fiberglass as a separator and placed a GFE-1 cathode saturated with PC-TBABr on top using some C4 carbon cloth as a current collector. Before placing the GFE-1 cathode I filled the cell with a 4.2M ZnBr₂ electrolyte, which makes the PC-TBABr remain less dense even after prolonged mixing. Since my objective was not to measure efficiencies with this device, but just to observe the chemical processes, I did not strive for a reproducible geometry.

After charging for a while with 3 charged AA batteries at more than 3V – just to make the process go fast – I noticed a lot of TBABr₃ forming and precipitating within the cell. Sadly, the perbromide seems to form on the cathode and then immediately migrate out and into the aqueous phase. To my absolute surprise, the TBABr₃ – which is more of a liquid rather than a solid – has higher affinity for the aqueous phase, although it doesn’t solubilize within it but rather forms a suspension with it.

I then proceeded to take the electrolyte out of the cell and perform an extraction using some additional PC-TBABr and surprisingly, all the perbromide stays in the aqueous phase after mixing and just refuses to get into the PC-TBABr. For this reason, PC is not going to work as a cathode electrolyte within the device, as the perbromide just exits it and never returns. This is probably why my devices trying to do this never seemed to work for too long.

During my latest experiments, I have moved to an inverted geometry setup, given that hydrogen evolution is a problem that needs to be eliminated for cells that are expected to last for long periods of time. However, an inverted geometry carries with it the problem of more favorable diffusion of elemental bromine – due to the fact that the cathode is now on top – reason why aggressively sequestering bromine is now a top priority.

In order to do this I have been trying to use a TBABr saturated propylene carbonate (PC) solution (which I will be calling TBABr-PC). My expectation was that by soaking the cathode in this solution I would be able to prevent any elemental bromine formed from escaping. The TBABr-PC behaves a lot like an ionic liquid (it’s > 50% TBABr) so its high conductivity and much higher affinity for elemental Br should allow the battery to work properly while keeping all the Br₂ from reaching the anode or the aqueous electrolyte.

The problem with these initial tests was that the battery seemed to suffer from initially low conductivity and charge retention with substantial changes through time that appeared to improve on these parameters. My guess was that there was a lot of ion migration between the initial TBABr-PC and the ZnBr2 aqueous electrolyte and that the battery was just not stable while these were happening.

To solve this issue I prepared 10mL of 1.5M ZnBr₂, 1% Tween 20 solution and mixed them with 2mL of TBABr-PC. The TBABr-PC was initially above the aqueous electrolyte, as expected from its lower density. After adding them together I then proceeded to mix them vigorously, which lead to the separation of two new phases. The PC phase now became denser, with the aqueous phase resting on top. This shows that there was some transfer of ZnBr₂ into the PC solution, although thankfully the phases do remain immiscible.

I then proceeded to fill an inverted cell with the top solution, saturated a GFE-1 cathode with the bottom solution and placed the saturated GFE-1 cathode on top before compressing the Swagelok cell. The cell had no separator but 4 PTFE o-rings as spacers. Since the PC had proved to have low conductivity before, I decided to cycle this device at 5mA to 15mAh. You can see the result of the first cycle above.

Although the CE and EE are now significantly better than before, there are still big questions about how a cell like this will evolve over time and whether the TBABr-PC is as effective at sequestering elemental bromine as I believe it might be. The fact that the organic phase is now denser also begs the question of whether the organic phase will just pool at the lower half of the cell with time. Hopefully affinity for the GFE-1 cathode is high enough. A potential solution to this problem is to try this experiment again with a 3M ZnBr₂ solution, which is going to have significantly higher density.

I will first cycle this cell for some time to gauge its stability before running a self-discharge experiment to test whether the TBABr-PC does significantly impair self-discharge of the device.

Earlier this month, I wrote an article about the use of non-aqueous solvents in Zn-Br batteries. The only published result I could find was an article dealing with Zn-Br flow batteries using propionitrile as the catholyte solvent but I wanted to avoid the use of propionitrile due to its toxicity and scarcity (hard to find/buy for an individual in the US). However I thought propylene carbonate (PC) could be a suitable replacement, so I bought some to test whether this was true or not.

The first experiments I carried out were to figure out whether PC could be used as the sole solvent within the battery. Sadly the solubility of ZnBr₂ is not high enough – at most in the 0.5-1M range at 20C – and the conductivity of these ZnBr₂ solutions was also not high enough, with very noisy charge/discharge curves with very high charge voltages that retained almost no charge at all.

The solubility of both TMPhABr and TBABr in PC is better, although TBABr is by far the most soluble. With TBABr I was able to achieve saturated solutions with almost 50% of TBABr, giving them a very decent amount of conductivity. Sadly this wasn’t enough to make PC usable as a single electrolyte though, as the bad behavior of the charge/discharge was also apparent when using this as the sole solvent.

The idea then came to use this concentrated PC TBABr solution to saturate the GFE-1 cathode and use this in an inverted cell. It is interesting that although PC is infinitely miscible with water, a 50% solution of PC TBABr is actually not miscible with a 1.5M ZnBr₂ solution in water. This is because the affinity of TBABr for PC is much higher than that of ZnBr₂ and the affinity of ZnBr₂ for water is also significantly higher as well.

This experiment was better behaved with actually measurable charge/discharge curves. I did 4 curves charging/discharging to 15mAh at 15mA – the results are shown above – with the best CE and EE values being 61% and 25% respectively. The charging voltages do show that the internal resistance is significantly higher than when using water so there is likely a lot more of hydrogen evolution at the anode. The generation of elemental bromine at the cathode is also probably significantly slower, given the much higher viscosity and lower conductivity of the PC electrolyte.

Given the higher charge density used, I thought It might be the case that the PC electrolyte is just not able to support as high of a current density as the normal aqueous electrolyte and therefore a much lower charge density needs to be used to use this successfully. I am going to be evaluating this hypothesis within my next few tests.