Category Archives: Battery research

Zinc Bromine Batteries: Why an inverted configuration is likely more practical

Static Zn-Br batteries have gained popularity during the past 4 years due to their apparent simplicity and their theoretical ability to last thousands of cycles without deterioration at a very low cost (although this is often extrapolated from Zn-Br flow batteries). The particular configuration that has become popular among some researchers and DIY enthusiasts is the one shown below, where the Zn-Br battery is assembled with the cathode at the bottom of the battery and either a graphite or Zinc metal anode at the top, with the electrolyte in between both of them.

Battery configuration that has been popularized during the past few years.

In this configuration for the battery, elemental Bromine (or sequestered bromine if any sequestering agents are used) will tend to accumulate at the bottom. This happens because Bromine is both denser than water and attracted to the carbon felt electrode. As bromine is insoluble in water and prefers to interact with the felt. The Zinc is then deposited at the top electrode and the cell only appears to be limited by the self-discharge of the process, caused by Bromine diffusion from cathode to anode.

However, experimentally – as you can see extensively in my work in this blog – at practical current densities (>15 mA/cm2), even in devices with anode/cathode distances of only 2-3mm, there is a substantial evolution of hydrogen in the anode due to the overpotential required to overcome the internal resistance of the device. This is true for ZnBr2 concentrations of 1.5-3M. This means that a lot of hydrogen is produced, which is then either trapped against the anode – reducing its surface – or then leaves the device at the expense of making the solution more alkaline, both processes which inevitably kill the device as a function of time. Trying to increase the conductivity further also leads to other, worse problems, such as robust Zinc dendrite formation.

Inverted device configuration with the carbon felt cathode at the top of the device.

To deal with the above means that you either need to periodically replace the electrolyte or treat it in some manner. This might not be economical if additives like sequestering agents are used but it is definitely not desirable as you will have to deal with a lot of left-over bromine containing solution. A potential solution might be to replenish the battery by adding HBr – in an analogous way to how you replenish the sulfuric acid in a lead acid battery, but this solution is not likely to be very practical due to the makeup of this battery. This is because adding excess HBr makes the hydrogen evolution problem much worse, so careful titration of the solution with the HBr is required in order to arrive at just the right pH, very impractical for users.

A more permanent solution is to use an inverted architecture, where the cathode is placed at the top and the anode at the bottom. Any hydrogen gas created then reacts with the solution and cathode, regenerating the electrolyte in the process. This sadly decreases the Coulombic and Energy efficiency value of the device, because Bromine diffusion is substantially aided by its tendency to sink into a water solution. Experimentally the CE drops from around 90% to 70% and the EE from 70% to 50% at the same current densities (see here). However this is likely where we need to start if we really want a Zn-Br architecture that can be used for a long amount of cycles in practice.

This fundamental problem was also recognized by the author’s of the Princeton minimal architecture paper. In the image above – taken from the publicly available supporting information of the paper – they also show how this sort of inverted architecture would work. They see even worse decreases in CE and EE as I have found experimentally.

In order to make this architecture viable we would need to increase the affinity of the Br2 for the felt and prevent it from sinking into the water solution, a very challenging proposition but one we can work towards if we start from a solid base. Soaking the felt in an organic phase that is conductive enough is the first things I am going to try to get to this goal.

Zinc Bromine Batteries: Why internal resistance was increasing

For the past week I have been trying to find the reason why the internal resistance of my batteries seems to be increasing linearly as a function of the cycle number. I sought to evaluate several different hypothesis, starting with the potential deterioration of the graphite cathode as a cause for this. Using titanium as a cathode generated similar results though plus a substantial deterioration of the cathode as a function of time. The graph below shows you a Titanium cathode and how it was pitted aggressively by the oxidative conditions of the cathode side of the battery.

A certified Ti-6Al-4V 0.5 cathode after a few cycles in a Zn-Br battery using a 1% Tween 20, 1% PEG-200, 3M ZnBr2 electrolyte. Black spots are actually holes in the titanium electrode.

The above results showed that deterioration of the graphite cathode was not the main cause of the increase in internal resistance, plus, it also showed that titanium – at least in this form – is not a suitable replacement for the graphite electrode in the Swagelok cell.

One interesting thing about the internal resistance problem was that opening up the cell, taking out the anode and putting the anode back in, seemed to regenerate the device to close to its initial conditions. This pointed to the process that was causing the deterioration to be somewhat reversible, at least in that it depended on the state of the device and was therefore likely not caused by the deposition of some insoluble film in the cathode or the unwanted precipitation of something within the electrolyte.

Thinking about this, I realized that a significant amount of hydrogen was being generated inside the battery at these potentials, which, given the flat anode geometry, was unable to escape the device. This trapped H2 gas, against the anode created zones of high resistance which accumulated with time up to the point where they basically prevented the electrode from working correctly at all.

To test this hypothesis I assembled an inverted geometry cell with a GFE-1 cathode treated with 10% TMPhABr on top and a graphite anode at the bottom. The cell used no solid separator, but a spacer made from 3 PTFE o-rings. Sadly I was running out of ZnBr2 and made a mistake in the solution preparation as well, so the electrolyte used is quite strange with a concentration of 1.5M ZnBr2 + 5% PEG-200 + 5% Tween 20. I meant to prepare 0.5% but made a mistake in my calculations.

Test cell using an inverted geometry
EE and CE as a function of time for an inverted cell geometry
Evolution of charge and discharge potentials relative to mean charge or discharge potential of the first cycle

As you can see in the curves above, an inverted device appears to behave in a completely opposite way to my previous devices with larger cycle numbers. The charge and discharge potentials seem to be evolving favorably as a function of time, with the charge potential decreasing and the discharge potential increasing. This means that the internal resistance of the device is actually becoming lower and lower as time goes on. The hydrogen gas is no longer a problem since, as a low density gas, it will tend to go up the cell, into the electrolyte and cathode, where it reacts with perbromides or elemental bromine in solution which regenerates the chemistry (as the hydrobromic acid formed then diffuses and reacts with any ZnO formed in the anode).

The Coulombic and Energy efficiencies of this device are quite low though, which is likely a consequence of the bigger amount of PEG-200 and Tween-20 used, plus a lower concentration of ZnBr2. I will be reproducing this setup with a freshly and properly prepared electrolyte as soon as I get new ZnBr2 this week but I thought these results were interesting enough to share with you. I will also continue cycling the above device to see if it eventually fails due to dendrites or some other reason.

Zinc Bromine Batteries: Increases in resistance when using TW20+PEG-200 containing solutions

As I showed in my last article on Zn-Br batteries using a Tween20+PEG-200 electrolyte, the batteries had no issues with dendrite production on the zinc anode but faced substantial deterioration of both the charging and discharging voltages as a function of time. I repeated the experiment using a Zinc anode to see if this phenomenon was consistent. The tests below were carried out for a battery that used a GFE-1 cathode pretreated with 10% TMPhABr and an electrolyte containing 1% PEG-200, 1% Tween 20 and 3M ZnBr2. The cell used a 0.2mm Zinc anode and graphite electrodes within the Swagelok cell.

Charge/discharge curves, charged to 15mAh at 15mA, discharged to 0.5V.
CE and EE as a function of the cycle. The cell was run for 8 cycles.

As you can see in the images within this post, although the CE and EE values of the battery remain fairly constant through the test, the mean charge potential increases and the mean discharge potential decreases as a function of the cycle number. This is showing that there is an overall increase in the internal resistance of the cell as a function of time, which appears to be linear and doesn’t seem to become smaller with time. Other cells that were charged to even larger numbers of cycles continued to show this deterioration, up to the point where the EE and CE of the cell started to decay and the cell started to fail.

This behavior points to some irreversible process happening within the device that is fundamentally affecting internal resistance. This has to be either a surface modification of the electrodes or an irreversible loss of conductivity in the solution. The first can happen due to bromide/perbromide interactions in the cathode, zinc oxide deposits in the anode caused by a local increase in pH due to hydrogen evolution or even the appearance of some insoluble polymers in the cathode due to electrochemical reactions of the additives.

Mean charge potential as a function of the cycle number
Mean discharge potential as a function of the cycle number

There is some evidence that cathode structure does deteriorate with time under the cell conditions. There is a significant amount of graphite that can be wiped off the Swagelok cell electrode after opening up the cells – while the graphite electrode was sanded and completely clean upon cell fabrication – which might point to the electrode itself or the GFE-1 cathode material degrading. I am currently running an experiment using a titanium electrode with a GFE-1 cathode to rule this out.

About the stability of the additives and sequestering agent, it might be worth it to carry out some cyclic-voltammetry (CV) experiments to investigate the stability of the organic additives being used to see if any of them can fundamentally degrade under these conditions. The TMPhABr could be especially susceptible, given its aromatic amine character. If the sequestering agent can deteriorate under these conditions, then we might be unable to use it as an effective agent and we might need to resort back to TBABr or TPABr. The Tween 20 might also be electrochemically vulnerable, so this additive also needs to be studied in this manner.

Zinc Bromine Batteries: What about non-aqueous solvents?

As my avid reader Giancarlo pointed out in the comment section a few posts back, many of the big problems of the Zn-Br battery system seem to be caused by the use of an aqueous electrolyte. Hydrogen evolution and voltaic losses due to the use of insoluble or immiscible bromine sequestering agents are some of the biggest issues that are inevitably related with water. Changing to a non-aqueous solvent can help solve some problems, although some others are created.

An organic solvent to replace water in Zn-Br batteries would need to be aprotic and to allow for the creation of substantially conductive solutions using ZnBr2. The issue with these organic solvents is that they are also extremely friendly to bromine, most of them being infinitely miscible with elemental bromine. This means that a battery built using these highly polar, aprotic solvents, would discharge significantly faster, as bromine would be significantly more likely to migrate to the anode. Sequestering agents would not be usable as these agents and their tribromides are incredibly soluble in these solvents as well.

propionitrile - Wikidata
Model representation of propionitrile, a highly polar and aprotic organic solvent.

However, this property can be exploited to solve part of the problems of the Zn-Br battery, at least the part dealing with voltaic losses related with the sequestering agents in water. This paper from 1988 shows how propionitrile can be used within a battery to sequester bromine and prevent its migration through the cell. In this device, an aqueous anolyte is used with an organic catholyte to trap bromine near the cathode. Since bromine has such a high affinity for propionitrile it will tend to stay in the organic section of the device, preventing movements towards the aqueous layer and providing more efficient confining and higher conductivity compared with some common sequestering agents. This is the only paper I could find that discusses the testing of Zn-Br batteries with a non-aqueous solvent that is not an ionic liquid. There are however, no papers I could find where the anolyte is replaced by an ionic liquid as well.

Propionitrile – and nitriles in general – are nasty solvents though. They are significantly toxic and we wouldn’t want to use them for DIY home batteries due to these issues, especially if we’re going to be experimenting and having close contact with their solutions. For this reason I will avoid their use and will instead use the non-toxic aprotic polar solvent, propylene carbonate. This solvent can also be bought pretty easily on the internet (I got mine here).

Propylene Carbonate - What's in This?
Model representation of a molecule of propylene carbonate. This is a low toxicity, highly polar, aprotic solvent.

Propylene carbonate can form highly conductive solutions with some salts, it is reasonable to predict that both the quaternary ammonium salt I have (TMPhABr and TBABr) will be soluble in it, as well as ZnBr2. I have no idea about whether Br2 or perbromides are soluble in it though, but it is reasonable to expect Br2 to be highly soluble in it because of how other, similar solvents, behave. There is some precedent for a battery fully constructed with a metal bromide using entirely propylene carbonate, see this paper, so it might actually be possible to use propylene carbonate by itself, although this is probably not possible without a suitable membrane separator.

We know the diffusion coefficient of Br2 in propylene carbonate to be 3.41×10−6 cm2 s−1 (1) while the coefficient in water is 1.18×10−5 cm2 s−1 (2), this is in part because of the higher viscosity of the organic solvent. This means that diffusion of Br2 in propylene carbonate will be more than 5x slower, which might be enough to create a functional battery with limited self-discharge, especially if this coefficient can be reduced further with the addition of the quaternary ammonium salts.

Even if the creation of a battery using entirely propylene carbonate is not possible, it might be the case that a highly concentrated TMPhABr or TBABr solution in this solvent will be immiscible with a 3M solution of ZnBr2 in water. If this is the case then a split approach with a bottom organic solvent and a top aqueous solvent might be possible, although this will depend largely on what the final density values actually are.

In any case, I have now ordered some propylene carbonate and will be carrying out some tests with it during the next couple of weeks.

Zinc Bromine Batteries: PEG-200 plus Tween 20 to eliminate dendrites, first public results ever!

In the past I have discussed zinc dendrites as one of the most important issues to deal with when creating Zn-Br batteries. While the effect of dendrites can be attenuated by using tall cells with large distances between the electrodes, these setups create high electric resistance that greatly diminishes energy efficiency. A high energy efficiency Zn-Br battery will therefore have the ability to reduce or eliminate zinc dendrites, such that a large number of cycles can be achieved without shorting the battery.

I have studied the use of PEG-200 quite extensively within this blog and although PEG-200 does reduce the formation of zinc dendrites, it also increases the internal resistance of the battery, to the point where the voltaic losses become unacceptable. At useful energy density values (>30 Wh/L) and acceptable charging currents (>10 mA/cm2), the maximum PEG-200 concentration for a 3M ZnBr2 solution is therefore restricted to around 1-3%.

Battery configuration used in the experiments discussed in this post.

Looking at other potential low cost solutions to eliminate zinc dendrites, this article using PEG and Tween 20 in alkaline batteries drew my attention. Although the article used PEG-600, it is reasonable to expect a similar effect with PEG-200, given that this has also been shown to reduce Zinc dendrites in alkaline batteries in multiple publications. It is particularly interesting that they can achieve this with a 0.5% PEG-600 + 0.5% Tween 20 solution, as this would be of practical use within Zn-Br batteries.

To investigate this, I bought some USP grade Tween 20. It is a very safe , non-ionic surfactant commonly used commonly for cosmetics. I then prepared a solution using ~1% PEG-200 and ~1% Tween-20 with 3M ZnBr2. I then assembled a battery as shown above. Note that although I have been using a separator-less setup during the last couple of weeks, I decided to try a fiber-glass based separator setup first, since this setup in the past suffered from dendrites at the edges that I believe might have been caused by surface tension issues with the solution. This problem is likely to be solved by the use of the Tween 20.

First eleven cycles, lighter plots are earlier cycles. Final CE and EE values shown. Charging was done to 15mAh at 15mA, discharge was done to 0.5V.
Coulombic and Energy efficiency evolution as a function of the charge/discharge cycle number.

The solution was easy to prepare and the Tween 20 did not generate any solubility issues. The assembly of the cell was quite interesting, while a 3M ZnBr2 solution (with or without PEG-200) normally takes around a minute to fully wick into the fiberglass separator and the GFE-1 cathode, this time the wicking was almost instantaneous, probably thanks to the use of the Tween 20, that greatly reduced the surface tension of the mixture.

The cycling of the cell is going on without any issues. After around 3 cycles, the magnitude of changes in the shape of the curves and capacity started to become smaller and smaller, with the battery currently settling at a CE ~ 92% and an EE ~ 68%. The total amount of charge extracted is around 13.8 mAh with an average potential of 1.46V, putting the energy density of the battery at this current density at around 30 Wh/L. I am so far amazed at the stability of this battery configuration with few aberrations showing in the charge/discharge curves and no signs of dendrites (so far). The above are the first ever published results – as far as I know – for a Zn-Br battery containing both PEG-200 and Tween 20.

An important early sign of dendrites is a decrease in the charging potential with time – as the zinc dendrites effectively enhance the surface area of the anode before shorting the battery – an effect that I haven’t observed after 11 cycles. Although still too soon, the above results are certainly encouraging, hopefully the synergistic effect between PEG and Tween 20 applies to the Zn-Br system as well.