My adventures building a Zinc-Bromine battery

danielfp248

Solar Enthusiast
Creating cathodes by soaking CC4 cloth in a 50% TMPhABr solution and then allowing them to dry before putting them into the battery seems to work significantly better than solid TMPhABr layers outside the cathode. These are the results so far charging to 3000 uAh and discharging to 0.5V at 2mA (ZnBr2 3M + 10% PEG):

1602258090726.png
1602258126517.png

Doing this seems to have considerably flattened the discharge curve - more stable discharging voltage - and has allowed the battery to reach CE values above 80% and EE values above 65% at a capacity of 3000 uAh.

As you might have noticed I also increased the PEG concentration from 1% to 10% since I read several research papers that showed the 10-20% range to be necessary to really avoid the formation of zinc dendrites when using Zn metallic anodes.

I am going to cycle this battery until it fails due to Zinc dendrites or until the CE/EE decay considerably. Seems promising so far! 🤓
 

danielfp248

Solar Enthusiast
Latest test uses a 3M ZnBr2 + 10% PEG200 solution with a CC4 cathode that was soaked in 50% TMPhABr and then dried in air. This cell was cycled 43 times at a current of 5mA, charged to 3000 uAh and discharged to 0.5V.

1602604446243.png

Last CE=91.39% EE=68.81%

All charge/discharge curves plotted together:



This battery was very stable, with no shorts after 43 cycles and no substantial drops in CE or EE through the entire experiment. I have now taken this battery apart to inspect how it looked.

A few things are apparent:
  • There is no formation of zinc dendrites through the separator (10% PEG-200 works perfectly to prevent Zn dendrites at this current density without causing a huge increase in internal resistance).
  • The separator did look very yellow (due to diffusion of elemental bromine through the cell)
  • Formation of solid perbromide at the cathode (so the saturated cathode worked, at least to an extent, to sequester perbromide)
The above means that the cathode material is probably not ideal because it cannot "store" or form perbromide quickly enough before the elemental bromine just diffuses through the cell. This is probably why the energy efficiency of the cell seems to be limited to 65-70% and why the self-discharge is probably significant, although I am yet to run experiments dealing specifically with this problem.

Another issue is that there is significant overpotential, because the carbon cloths I have used are just not conductive enough. I have therefore decided to move to the GFE-1 graphitic felt I purchased before for the next experiments (https://www.ceramaterials.com/product/gfe-1-pan-graphite-felt/), since this cathode material has significantly higher surface area, significantly higher conductivity and should lead to lower overpotentials and higher efficiencies.

Expect the results for these experiments in a couple days 🤓
 

danielfp248

Solar Enthusiast
Highest energy efficiency results yet. EE=76% CE=83% (second cycle, charge/discharge 5mA). Using the GFE-1 cathode, pre-treated by soaking in 10% TMPhABr and then drying in air.

1602611845811.png
 

danielfp248

Solar Enthusiast
These results were pretty exciting, so I decided to try doing higher current density and higher capacity. Below you can see a cycle doing 10mA charging to 5000uAh. The results seem pretty amazing! CE=87.45% EE=76.11%. 🤓

1602634875318.png
I will leave this cycling tonight and should have some results of CE and EE as a function of cycle number tomorrow. This puts the specific energy of the battery at 50-55Wh/kg, which puts it at the top of the range of current commercial Zinc-Bromine batteries. After cycling to confirm stability I will try another experiment using a GFE-1 electrode soaked in 50% TMPhABr solution (then air-dried) to see how it compares to this electrode (which was pretreated with a 10% solution).

I am ecstatic about this result! It is 10x the specific energy of my first batteries at 5x the current density :eek:
 
Last edited:

daviddb

New Member
Hi Daniel,

Firstly, congratulations on your results so far - very impressive!

I have been doing research very similar to you. In fact, when I found your blog online and then your posts here I felt like someone was reading my mind!

I come from a physics and electronics background. As part of my electronics work I became involved in rocketry. One of my projects was to construct a perchlorate cell for Ammonium Perchlorate production. Going through that process helped spur my interest in the marvel of electrochemistry!

Regarding my history experimenting with energy storage, I started out with alkaline supercapacitors as getting reasonable performance was trivial. Keeping with alkaline chemistries, I experimented with Ferricyanide catholytes and zinc anodes. I liked the idea of a static flooded cell configuration but membrane leakage and self discharge was always a huge issue.

After reading a paper on Bromine ion intercalation into graphite I became interested in using Bromine for my own experiments. I first tried to use a Zinc Chloride "water-in-salt" electrolyte with zinc bromide / graphite cathode. This actually worked well in terms of power density (> 1A / cm^2). However, the bromine was able to diffuse and self discharge was an issues.

I then tried to do something similar in a flooded configuration with different Bromide electrolytes. I have achieved ~35 Wh/L (total internal volume of cell) with good power densities. However, again, self discharge was a problem. This was partially inspired by the work done at Princeton, but with a supporting electrolyte.

I then came across the same Chinese paper you did regarding tribromide confinement with a complexing agent. I ordered myself a Kg of TPABr as used in the paper. I did some back of the envelope calculations as you did and realised the energy density values reported were misleading (par for the course it seems). Even so, I thought it worth experimenting with.

There are multiple issues I'm facing now:
1) Solubility of TPA Br in the presence of Zinc ions
2) Zinc corrosion
3) Dendrites

At the moment, I'm experimenting with water-in-salt electrolytes, aprotic solvents, thixotropic modification and complexing agents.

Ok, so long post, but I felt it important to share. Lastly, thank you so much for putting your work out there in a blog and on this forum. I literally refresh many times a day waiting for updates!

- David
 

danielfp248

Solar Enthusiast
Ok, so long post, but I felt it important to share. Lastly, thank you so much for putting your work out there in a blog and on this forum. I literally refresh many times a day waiting for updates!

Thanks for your comments David! It's always great to find someone working on the same thing 🤓 Your development history is also very interesting. Could you share some charge/discharge curves with us? I would be very interested to look at the types of efficiencies and capacities you have been able to produce with your current battery configuration and anode/cathode materials.

I then came across the same Chinese paper you did regarding tribromide confinement with a complexing agent. I ordered myself a Kg of TPABr as used in the paper. I did some back of the envelope calculations as you did and realised the energy density values reported were misleading (par for the course it seems). Even so, I thought it worth experimenting with.
I have not ordered TPABr, given that TMPhABr is less symmetric, less aliphatic and more polar I thought it would be a better candidate in terms of having higher solubility while its perbromide - given the melting points between TPABr3, TMPhABr3 and TBABr3 - seems to be the most stable and therefore probably the one with the lowest solubility and highest "sequestering efficiency". In the literature sequestering agents containing aromatic groups tend to do better at capturing more bromine atoms per mole (since they seem to be able to stabilize larger polybromides).

1) Solubility of TPA Br in the presence of Zinc ions

I also faced this wall with TMPhABr, which I thought would at least be able to reach 1M+1M ZnBr2. Sadly I don't think it's possible to get significantly higher than 0.5M+0.5M ZnBr2.

Complexing agents have the problem of also increasing the over-potential of Zn deposition, mainly because you require more energy to get the Zn out of the chelate and into the electrode, these might be interesting however if the over-potential cost is not large and you are able to get >3M of TMPhABr in the presence of the chelated Zn salt. Experiments with Zn(EDTA)Na2 were on my mind but I have chosen to delay them because of the very low mass efficiency of this chelate.

Aprotic solvents are very interesting, but then the problem becomes the perbromides which are soluble in most of these solvents (so the self-discharging problem becomes large again). I haven't found a solvent yet that would solve these issues, propylene carbonate - which is a great candidate - greatly solubilizes the perbromides and so do many other solvents (like DMF, DMSO, etc). Do tell me if you find a solvent that doesn't!

So far the best solution I have found to this problem is to use a very conductive and highly porous cathode and just "load" it by presoaking it in a concentrated solution of the quaternary ammonium salt (10-50%) (allowing it to air-dry before use). This seems to allow for the formation of the insoluble perbromides within the cathode while only allowing for a small amount of bromine to escape. I am yet to measure self-discharge yet but energy and coulombic efficiencies are quite high, even at high capacities. I have not tried higher current densities though, because of how strongly higher densities lead to both hydrogen evolution and additional Zn dendrites.

2) Zinc corrosion

This is not something I have experienced, probably because my Bromine migration has always been very limited or my testing not long enough. My zinc anodes don't seem corroded upon examination of my batteries. At least when doing 10-30 cycles at 5-10 mA (my electrodes are circular, with a diameter of 0.5 inches). Limiting your bromine migration should solve this problem.

3) Dendrites
For me 10% PEG-200 seems to have solved this issue, at least at the current densities I have used. You can probably go up to 20-30% PEG while still having a working battery if you want to avoid dendrites, although the cost in energy efficiency might be significant.

At the moment, I'm experimenting with water-in-salt electrolytes, aprotic solvents, thixotropic modification and complexing agents.

Those are very cool experiments! I would love to see the curves you get and the design characteristics of the cells you are building. 🤓

Thanks again for commenting and I look forward to further conversations on the topic :cool:
 

danielfp248

Solar Enthusiast
It seems my specific energy values were too optimistic. A couple of substantial things change within the cell when using the GFE-1 cathode material:
  1. It is heavier (70-100mg per loaded cathode depending on the TMPhABr concentration)
  2. It is very porous, so it will hold a large amount of electrolyte (100-200uL)
The cell I measured before at CE=87.45% EE=76.11% to 5000 uAh has a total weight of 720mg (all materials), so its actual specific energy was 11 Wh/kg. The zinc anode I use is pretty heavy (0.2mm) so changing to a 0.02mm anode could reduce this weight to around 530mg without any change in performance, which would put the specific energy at ~15 Wh/kg.

Due to the larger weight we therefore need to push the capacity even further if we want to reach the 50-55 Wh/kg mark, probably up to 15mAh, this shouldn't be much of a problem since we seem to have the "storage space" to hold the perbromides produced within the more porous GFE-1 cathode. I have ran a cell with a cathode pretreated with 50% TMPhABr and tried to run it to 10mAh at 10mA:

CE = 94.52%, EE=69.29%

1602706993638.png
I don't know the mass of this full battery yet - won't know till I finish experiments with it and open it - but overall the specific energy is probably around 30 Wh/kg. I might need to change back to a cathode treated with 10% TMPhAbr - resistance seems to increase too much with a cathode treated with a 50% solution - but I do hope we can get to the 45-55 Wh/kg region pretty soon :)
 

daviddb

New Member
Thank you for your reply Daniel.

Many of my preliminary tests have been either prismatic, between graphite plates or within 5 mL pill containers. Its not ideal by any means, but it's a simple way to get a visual idea of dendrites, cross-diffusion etc.

Building a practical bromine battery is going to require a fairly large electrode spacing for the current collector volume to be made negligible. One benefit of aqueous electrolytes is of course higher ionic conductivity, which allows for this.

I have been testing with graphite felt - 10mm (I got excited and bought a 5m roll from Alibaba) as the cathode. I'm plating directly on to a graphite anode. My standard testing cell is about 30 mm in diameter and 10mm deep (I can set the depth). The graphite felt cathode is joined directly onto another graphite end stop with graphite / nitrocellulose binder.

I have found that infusing the graphite felt with activated carbon (nitrocellulose binder in acetone) drastically increases the capacity of TPABr/ZnBr cells. Without this, it seems the surface area is too limited to support a reasonable charging current.

I don't have good measurements for efficiency as I'm not measuring total charge in. My measuring device is a datalogger measuring voltage during discharge over various resistors. My main concern is total stored Wh/L (and stability).

Regarding corrosion, I actually meant hydrogen evolution. I would be very interested to know if you are having any issues there.

I will post some discharge curves and a photo or two as soon as I can!

PS: Your latest results with a higher charge is impressive! I would love to see discharge curves with a delay included after charging!
 

danielfp248

Solar Enthusiast
Thanks for the reply David! :)

I have been testing with graphite felt - 10mm (I got excited and bought a 5m roll from Alibaba) as the cathode. I'm plating directly on to a graphite anode. My standard testing cell is about 30 mm in diameter and 10mm deep (I can set the depth). The graphite felt cathode is joined directly onto another graphite end stop with graphite / nitrocellulose binder.

What's the conductivity and surface area of the felt you're using? The GFE-1 graphitic felt I am using has very high conductivity, so I haven't had to use any additional conditioning to improve its characteristics but I'm interested on the properties/cost of your felt.

My main concern is total stored Wh/L (and stability).

In terms of Wh/L I have achieved around ~20 Wh/L. Given the volume of electrolyte I use this makes me think I should perhaps be pushing my capacities even further. I might do a test pushing capacity as far as I can before the battery breaks, see how much I can actually get. I am definitely far from the ~35 Wh/L you have achieved.

Regarding corrosion, I actually meant hydrogen evolution. I would be very interested to know if you are having any issues there.
Provided your potential stays below 1.9V the hydrogen evolution should be pretty limited, at least this has been my experience using Zn anodes.

PS: Your latest results with a higher charge is impressive! I would love to see discharge curves with a delay included after charging!
I'll definitely focus on this type of experiment as soon as I get to a value of Wh/L and Wh/kg that seems good enough!
 

daviddb

New Member
What's the conductivity and surface area of the felt you're using? The GFE-1 graphitic felt I am using has very high conductivity, so I haven't had to use any additional conditioning to improve its characteristics but I'm interested on the properties/cost of your felt.
Surface resistance as quoted from the manufacturer: Ω/cm2 0.145-0.183. That number seems fairly generous, but I can confirm that the conductivity is very high. The main issue is that it is not activated, so the surface area is low. I don't have a good way to measure it, but based off super-capacitor measurements, its orders of magnitude lower than regular activated carbon. My latest tests have used this felt with a coating of activated carbon.

I think it would be better to just used activated felt like you have. I believe I can activate my felt by placing it in a furnace at 500 deg. for an hour or so.

Provided your potential stays below 1.9V the hydrogen evolution should be pretty limited, at least this has been my experience using Zn anodes.
With the TPABr it is fairly low, but its definitely still there and it will make sealing a cell tricky. I don't think hydrogen recombination is going to help here.

In terms of Wh/L I have achieved around ~20 Wh/L
Is that based on the electrolyte volume, or the total cell inner volume, including separators etc? So, the 35 Wh/L I achieved was certainly no free lunch! The self discharge was high. I didn't measure it, but at least 50% after 24 hours. There were some dendrites, but this wasn't too much of an issue as the electrode spacing was >4 mm.

So yeah, this is definitely a tradeoff game.
 
Last edited:

danielfp248

Solar Enthusiast
With the TPABr it is fairly low, but its definitely still there and it will make sealing a cell tricky. I don't think hydrogen recombination is going to help here.
Recombination is certainly not practical. Given that PEG-200 or some other additive is likely to be necessary to avoid dendrites, I think you might want to test this to see how it affects your hydrogen evolution as well. Given my Coulombic efficiency values, I do believe I might be losing some amount of charges to H2 production, especially in cells where my potential reaches close to 2V.

Is that based on the electrolyte volume, or the total cell inner volume, including separators etc? So, the 35 Wh/L I achieved was certainly no free lunch! The self discharge was high. I didn't measure it, but at least 50% after 24 hours. There were some dendrites, but this wasn't too much of an issue as the electrode spacing was >4 mm.
This is based on total volume of the cell (all materials). If I used only electrolyte volume, it would be like >45Wh/L, but that wouldn't be very indicative of practical energy density values. Using 8 layers of the fiberglass separator I used I'm usually at 2mm of electrode spacing, but I have been doing some tests at 16 layers, so that I can tolerate a bit more of Zinc dendrites.

I am running tests to see how far up I can take capacity, but this takes forever due to the current densities I am testing. Hopefully I'll have some results before the week-end :)

Still hoping to see some of your curves and photos! 🤓
 

danielfp248

Solar Enthusiast
I have used a GFE-1 cathode (pretreated with 10% TMPhABr), in conjunction with a 10% PEG-200 + 3M ZnBr2 electrolyte, using the graphite electrode as anode (so no metallic Zn anode), 16 layers of fiberglass separator. To see if zinc dendrites would be an issue quickly I decided to charge to only 15mAh, but at a current density of 15mA (much larger than what I used before (5-10 mA)). The larger current density would facilitate the growth of dendrites. The results are very good so far. The following are charge/discharge curves as a function of cycle number:

1603036799803.png
1603036811524.png

At the last cycle the cell is at 12.87mAh of stored charge (discharging to 0.5V) which gives it an energy efficiency of 65.26% with an energy density of 30.11 Wh/L (area is 1.29cm2 and height - estimated from previous devices with the same layout - is 0.53cm, volume of ~0.68mL, I will get more precise values when I open the cell but they will be very close to these). This is total volume of cell (including cathode, separators, etc). Internal resistance has dropped dramatically as well - I could never charge at 15mA at these potential values - this is probably because the Zinc anodes were not perfectly flat and therefore appeared to not make perfect contact with the graphite electrode (so there were regions of higher electric field) this meant higher dendrite production and higher internal resistance.

I will continue to cycle to see if dendrites show up, but given the current density and capacities I've been charging to, I think it's pretty safe to say that the dendrite problem is at least 10x better than with the previous devices. The edge effects of the 0.2mm Zinc anodes were apparently a big part of the problem. If we reach >100 cycles without dendrites I will proceed to perform some self-discharge tests to see how the battery retains capacity as a function of time. Steady progress so far 🤓
 

danielfp248

Solar Enthusiast
It's amazing how everytime I post about something, the next cycle I start seeing issues ☹️ This time zinc dendrites killed the battery just one cycle after:

1603061459902.png
 

svetz

Works in theory! Practice? That's something else
Staff member
Moderator
So, the battery is always the most coulombic just before the dendrite? ;)
 

danielfp248

Solar Enthusiast
This is a new experiment, using a GFE-1 cathode treated with 10% TMPhABr, 16 fiberglass separator layers, electrolyte is 3M ZnBr2 + 20% PEG-200. Charging to 20 mAh at 10mA, discharging to 0.5V. So far, 4 cycles without dendrites shorting the battery 🤞

1603123863662.png
Energy density at this point is ~38.77 Wh/L. Internal resistance is higher than when using 10% PEG-200 - as expected - but hopefully this fully prevents dendrite development at this current density.

How long will it last? Probably just 5 cycles since I just posted this ;)
 
Top