My adventures building a Zinc-Iodine battery

[danielfp248]

• I don't think white vinegar is a good idea. It has a complex composition and contains protic acid which may offer activity H+.
  

Bear in mind that the pH of a 15m ZnCl2 solution is really low (pH < 2), given the pKa of acetic acid is 4.75, it provides no protons under these conditions. Given that ZnCl2 is several orders of magnitude more acidic (very strong Lewis acid), the acidity of acetic acid is not relevant. It does seem to improve ion mobility though.

 

[Qinzheng]

Bear in mind that the pH of a 15m ZnCl2 solution is really low (pH < 2), given the pKa of acetic acid is 4.75, it provides no protons under these conditions. Given that ZnCl2 is several orders of magnitude more acidic (very strong Lewis acid), the acidity of acetic acid is not relevant. It does seem to improve ion mobility though.

Reasonable，my fault.

 

[danielfp248]

Reasonable，my fault.

No worries! I do agree that an organic chemical is an additional complexity that we better avoid if possible. If it gives worse results I'll definitely not use it.

 

[houseofancients]

daniel, seems like black magic to me, how ever love following your experiments, just trying to read up on knowledge fast enough to understand it

keep up the good works !

 

[danielfp248]

A lot of shorting was happening, preventing me from doing any substantial testing. It seems that using the Whatman 42 filter paper as separator was making things even worse. The fact that the paper turns completely black - as in the picture I showed in this thread - demonstrates that it is in fact reacting with some iodine and not entirely "neutral". The problem with using my glass fiber paper was that it has very large pore sizes and carbon fiber from carbon cloths were able to climb through a single layer and short the battery.

However, now that I have carbon filter papers (which don't have any "loose hairs"), I have been able to use a single fiber glass layer as a separator. Fiberglass does remain unchanged through charging/discharging, so hopefully this will either be the end of these dendrite shorts, or at least I'll be able to see clearly where they are happening.

 

[danielfp248]

After doing a lot of reading on water-in-salt electrolytes and Zinc dendrites, it seems clear that Zinc deposits in these electrolytes are only dendrite free under low current densities. In all the published studies using WiS electrolytes, the maximum current density used is usually around 1mA/cm2. Below this density the deposits are compact and highly reversible while above this current they start to have significant dendrite formation.

Additionally, trying to use the fiberglass separator has been complicated due to its very high porosity, I get shorts right from the start in around 1 out of 3 assembled devices, for this reason I am going to continue with the Whatman 42 paper, while I look for other separator options.

Right now, I am testing a device using a 15m ZnCl2 + 5m KI + 5m NaCl in white vinegar (4% v/v Acetic acid)) electrolyte, a zinc anode, W42 separator and a Spectracarb 2050A-1050 cathode, using the Ti electrodes in the Swagelok cell. However, I am charging to 1.6V at 0.5mA (0.38mA/cm2). It will take a long time to get some long term results, but there should be a big increase in capacity. I will post the first cycle information later today.

Total thickness of this device is 0.0454cm with an area of 1.29cm2, volume is 0.0586cm3.

 

[danielfp248]

After doing a lot of reading on water-in-salt electrolytes and Zinc dendrites, it seems clear that Zinc deposits in these electrolytes are only dendrite free under low current densities. In all the published studies using WiS electrolytes, the maximum current density used is usually around 1mA/cm2. Below this density the deposits are compact and highly reversible while above this current they start to have significant dendrite formation.

Additionally, trying to use the fiberglass separator has been complicated due to its very high porosity, I get shorts right from the start in around 1 out of 3 assembled devices, for this reason I am going to continue with the Whatman 42 paper, while I look for other separator options.

Right now, I am testing a device using a 15m ZnCl2 + 5m KI + 5m NaCl in white vinegar (4% v/v Acetic acid)) electrolyte, a zinc anode, W42 separator and a Spectracarb 2050A-1050 cathode, using the Ti electrodes in the Swagelok cell. However, I am charging to 1.6V at 0.5mA (0.38mA/cm2). It will take a long time to get some long term results, but there should be a big increase in capacity. I will post the first cycle information later today.

Total thickness of this device is 0.0454cm with an area of 1.29cm2, volume is 0.0586cm3.

So I decided to not run this experiment, as getting long term cycling results of batteries at low current densities is not practical, it would take months to get hundreds of cycles. The lowest current that I believe is practical to gather results for me is around 2.0mA/cm2. This means I will need to start testing solutions to alleviate the dendrite problem in the WiS electrolyte. Here are some of the additives that have been used in the literature to reduce dendrites with some success (some which I tried with the Zn-Br cells with some success):

1. Boric Acid
2. Thiourea
3. Cetrimonium bromide
4. Sodium dodecyl sulfate
5. PEG-200
6. Tween-20

I have cetrimonium chloride, which has been tested to reduce dendrites in lithium but not zinc batteries, so I will try this one first. I also have Tween-20, which I used in Zn-Br batteries successfully, so that will be my next try.

The first attempt will be a cell with a 15m ZnCl2 + 5m KI + 0.00375m cetrimonium chloride (around 1200ppm) in distilled water. The cell uses a zinc anode, W42 separator and a Spectracarb 2050A-1050 cathode, using the Ti electrodes in the Swagelok cell. Experiment will be to charge to 1.3V at 2.5mA, discharge to 0.5V.

 

[danielfp248]

So I decided to not run this experiment, as getting long term cycling results of batteries at low current densities is not practical, it would take months to get hundreds of cycles. The lowest current that I believe is practical to gather results for me is around 2.0mA/cm2. This means I will need to start testing solutions to alleviate the dendrite problem in the WiS electrolyte. Here are some of the additives that have been used in the literature to reduce dendrites with some success (some which I tried with the Zn-Br cells with some success):

1. Boric Acid
2. Thiourea
3. Cetrimonium bromide
4. Sodium dodecyl sulfate
5. PEG-200
6. Tween-20

I have cetrimonium chloride, which has been tested to reduce dendrites in lithium but not zinc batteries, so I will try this one first. I also have Tween-20, which I used in Zn-Br batteries successfully, so that will be my next try.

The first attempt will be a cell with a 15m ZnCl2 + 5m KI + 0.00375m cetrimonium chloride (around 1200ppm) in distilled water. The cell uses a zinc anode, W42 separator and a Spectracarb 2050A-1050 cathode, using the Ti electrodes in the Swagelok cell. Experiment will be to charge to 1.3V at 2.5mA, discharge to 0.5V.

This cell short circuited after only 3 cycles, on opening there seemed to be a lot of salt precipitation. The cetrimonium chloride seems to cause a loss of solubility of the Zinc salts, at least at this concentration. If I try cetrimonium chloride again, it will likely be only at 500ppm.

I will now try a cell with a 15m ZnCl2 + 5m KI + 1000 ppm Tween 20 in distilled water. The cell uses a zinc anode, W42 separator and a Spectracarb 2050A-1050 cathode, using the Ti electrodes in the Swagelok cell. Experiment will be to charge to 1.3V at 5mA, discharge to 0.5V.

 

[danielfp248]

All the experiments were showing shorts quickly after running at either 2.5 or 5 mA. I tried several different concentrations of Tween-20 and cetrimonium chloride with no difference in shorting. However I had previously ran cells with 15m ZnCl2 + 5m KI in distilled water for long times without any dendrites, so it seemed strange that now I was having trouble running even 10 cycles.

I decided to change back to graphite electrodes in the Swagelok cell. I ran a battery with a 15m ZnCl2 + 5m KI in distilled water electrolyte, W42 separator, Spectracarb 2050A-1050 cathode and the bare graphite electrode of the cell as anode. The cell was able to run for 27 cycles without any dendrites at 2.5mA. I stopped the test because of a loss in capacity, but this does reproduce my previous results under these conditions. The larger losses in capacity are mainly due to the lack of reversibility in the Zn anode formation when a bare graphite electrode is used.


Long story short, it seems that the Titanium electrodes in the Swagelok cell were somehow responsible for the fast formation of zinc dendrites in the battery's anode. I will go back to graphite electrodes from now on and retest the effects of Tween-20 and cetrimonium chloride.

 

[danielfp248]

So both cetrimonium chloride and Tween-20 gave similar results when using graphite electrodes, both had shorts after only a few cycles. Interestingly the cetrimonium chloride at 300ppm enhanced the capacity of the device significantly (allowing me to go near 40 Wh/L at 2.5mA) but dendrites formed a lot faster than with the standard WiS electrolyte. I was only able to complete one cycle before dendrites shorted the battery.

I found a really interesting method to modify Zn anodes (https://www.mdpi.com/2079-4991/11/3/764) which involves depositing a thin film of Cu over sanded Zn anodes using a 0.1M solution copper sulfate. Since I have copper sulfate I tried this out, a black copper film was deposited by placing 100uL over freshly sanded zinc anode disks with a 0.5 inch diameter, left for 3 minutes, then washed with distilled water and dried in the air. I prepared 4 anodes in this way.

I am now cycling a cell with one of these anodes, a 15m ZnCl2 + 5m KI in distilled water electrolyte with a W42 separator and a Spectracarb 2050A-1050 cathode. I am charging to 1.3V and discharging to 0.5V at 2.5mA.

 

[adrian1702]

Daniel, thank you for sharing your experience and I keep my fingers crossed for finding the optimal solution. Are you thinking about other types of cells?

 

[danielfp248]

Daniel, thank you for sharing your experience and I keep my fingers crossed for finding the optimal solution. Are you thinking about other types of cells?

Thanks for following my progress! It takes a long time to understand these devices and especially to do quantitative experiments. I seek to understand the Zn-I chemistry deeply before moving onto any other device types. However, it might take a long time for that to happen given all that there is to explore!

 

[danielfp248]

So both cetrimonium chloride and Tween-20 gave similar results when using graphite electrodes, both had shorts after only a few cycles. Interestingly the cetrimonium chloride at 300ppm enhanced the capacity of the device significantly (allowing me to go near 40 Wh/L at 2.5mA) but dendrites formed a lot faster than with the standard WiS electrolyte. I was only able to complete one cycle before dendrites shorted the battery.

I found a really interesting method to modify Zn anodes (https://www.mdpi.com/2079-4991/11/3/764) which involves depositing a thin film of Cu over sanded Zn anodes using a 0.1M solution copper sulfate. Since I have copper sulfate I tried this out, a black copper film was deposited by placing 100uL over freshly sanded zinc anode disks with a 0.5 inch diameter, left for 3 minutes, then washed with distilled water and dried in the air. I prepared 4 anodes in this way.

I am now cycling a cell with one of these anodes, a 15m ZnCl2 + 5m KI in distilled water electrolyte with a W42 separator and a Spectracarb 2050A-1050 cathode. I am charging to 1.3V and discharging to 0.5V at 2.5mA.

This battery lasted for 14 cycles when dendrite issues started showing up. Charged to 1.35V at 2.5mA. Results are below:

 

[danielfp248]

I have been battling a lot of stability problems in the devices. As you can see in the tests in #93 and #89, whenever I have tried to go to capacities >20Wh/L using graphite electrodes in the Swagelok cell, I have had decreasing capacity values when charging to constant potential. These decreases are dramatic, with devices never lasting more than 25-30 cycles and capacity decreasing dramatically from start to finish.

The current architecture of the Swagelok cell is shown below (not to exact scale). Total cathode+anode+separator width is usually around 0.4mm.



As you can see, the graphite electrodes compress the battery structure. They are also in contact with the wet cathode and occasionally there is solution that drips from the separator onto the graphite electrodes (as the separator drips a little bit upon compression), if there is an electrical path, then this solution also becomes electrochemically active and formation of elemental iodine or zinc happens outside of the intended battery structure.

I believe some of the instability of the battery can be attributed to these reactions happening outside of the intended battery space, as well as Iodine being absorbed irreversibly by the graphite electrode on the cathode side.

To solve these problems, I have changed again to Ti electrodes, but this time I am using Cu tape to perfectly cover the face of the Ti electrode on the anode side, to avoid having any exposed Ti surface where Zn dendrites could form. I have also added Teflon tape to the Ti electrodes in order to completely fill the airgap and prevent any solution from dripping and reacting on the sides of the Ti electrodes. The new configuration is shown below:


I am now trying this configuration with the standard 5m KI + 15m ZnCl2 solution, using a W42 filter paper separator and a Spectracarb 2050A-1050 cathode. Hopefully this time we will get more stable results. I am trying charging to a constant capacity of 1.5mAh, discharging to 0.5V at a current density of 2.5mA/cm2.

 

[danielfp248]

These are the results of building a battery with 5m KI + 15m ZnCl2 solution, using two layers of a W42 filter paper separator and a Spectracarb 2050A-0850 cathode. This is with a copper anode using Titanium electrodes in the Swagelok cell. Failures due to dendrites start at around cycle 25. It was charged to 1.6V and discharged to 0.5V, at 5mA (3.87mA/cm2). This battery is thicker (double the separator) so the max energy density is around 21 Wh/L.


So far I haven't been able to build a battery with no dendrite issues with area capacity values above 1 mA/cm2. Doubling the separator definitely helps, as dendrites need to travel a lot further, but the issue is still clearly present. Any suggestions are always welcome!

 

[danielfp248]

This is an experiment using a symmetric cell with Spectracarb 2050A-0850 as both anode and cathode, using 2 layers of W42 filter paper as separator and a 5m KI + 15m ZnCl2 electrolyte. Cell was charged to 1.6V at 5mA/cm2 and discharged to 0.5V. Cell ran for 76 cycles, although some dendrite issues did show up, the battery did not completely die because of it. Capacity was stable until dendrites showed up.


Next cell is exactly the same, except with only one layer of W42 separator as filter paper and an additional 3000 ppm of cetrimonium chloride in the electrolyte. This is because wetting of the carbon electrodes was poor, so I want to test if better wetting improves the energy efficiency of this configuration.

 

[danielfp248]

After running several different tests, I continue to have dendrite issues that kill cells within the first 100 cycles. This happens regardless of additives I have tried - sometimes even faster with some additives - and regardless of any anode modifications (I have tried both copper and tin plating over Zn and also changing to carbon anodes or copper anodes). It also happens even with increases in separator width. This tells me that the process of dendrite formation is extremely fast.

It is however quite mysterious, because even when using fiberglass separators, I see no clear visual evidence of dendrite formation within the separators, except perhaps for some formation near the edges. However, no one in the literature seems to be having problems with edge-related dendrites in Swagelok cells, which makes this an unlikely source of the problem. I have also tried covering the edges of the anode with Teflon, with no improvement at all. It seems to me that these dendrites are therefore quite small and not easily visible to the naked eye. Their formation also seems inevitable with this electrolyte and happens with both filter paper, copy paper and fiberglass separators.

I have also tried separator-less cells, where dendrite formation doesn't seem to be an issue and it is clear that the zinc deposits are quite compact on the anode. However, separator-less setups are much thicker (1.6mm vs 0.5mm) so the energy density is reduced very substantially. They are also harder to assemble and test.

Right now I am unsure which way to go.

 

[svetz]

...It is however quite mysterious,... no one in the literature seems to be having problems...unsure which way to go.

Email one of the researchers to get their thoughts and take a stroll in a nearby park to get your mind off things for a bit?
Wish I had more to offer.

Possibly you're too focused on dendrites and its something else or something really crazy (e.g., dendrites dissolve when no current is flowing but have an ion-path so they can reform instantly when current is applied).

 

[danielfp248]

Email one of the researchers to get their thoughts and take a stroll in a nearby park to get your mind off things for a bit?
Wish I had more to offer.

Possibly you're too focused on dendrites and its something else or something really crazy (e.g., dendrites dissolve when no current is flowing but have an ion-path so they can reform instantly when current is applied).

Thanks for the suggestions

There is no other thing I can think of that can cause shorts on charge/discharge besides dendrites, they also respond as you would expect from them (form faster at higher current densities and larger over-potentials for example).

I will follow both of your pieces of advice though!

Note that the researchers from the paper I've been following never looked at current densities this high. If the electrolyte easily forms dendrites, then this would explain why. It is also not uncommon for research in Zn batteries to implement some aspect that circumvents dendrites, for example using low current densities, using very large separator thickness or anodes with really large surface areas. People in research often want to just publish something interesting, not so much in building a battery that's practical.

 

[danielfp248]

I believe I have been unsystematic with the current experimentation. There has been a lot of impulse to try a lot of things, wanting to observer interesting phenomena and learn more about the subject, but this has actually caused some confusion and a lack of direction in how to proceed. For example, I have not studied variables by themselves judiciously and have erred by making multiple changes simultaneously. I have also failed to properly study variables such as additive concentration in a systematic manner.

I have therefore decided to become more systematic in this research, such that I can arrive at more robust conclusions and avoid driving myself into "experimental corners" where I have done so many experiments that I no longer even remember all of them.

Being systematic is a lesson that life has taught me time and time again. Seems I didn't learn it well enough from my Ph.D. I guess sometimes I just get too excited

A big part of doing more systematic research is recording and analyzing the research in a more judicious manner. For this reason I have now created a shared folder in my dropbox that anyone can access (https://www.dropbox.com/sh/67hdgpm5lijt5s1/AACS0gjDSYHZny_tLSfXXmfza?dl=0), see which experiments I am running, which ones I have ran, analyze the raw data and take a look at the analyzed results of any previous experiments.

I have decided to fall back and start studying the ZnCl2 15m + KI 5m system in more detail. Since the filter paper separator is clearly not inert (gets turned black by reactions with Iodine quite quickly), I have decided to use exclusively fiberglass, even though this demanded making the Ti electrodes in the Swagelok cell much flatter and being way more careful in the assembly of the devices.

Exp1 was done using filter paper though and you can clearly see the dendrite failure facilitated by the use of this separator material.