Monthly Archives: November 2022

A DIY cation exchange membrane with PVA and cellulose

In previous posts (here and here), I have talked about my goal to create an Fe/Mn flow battery and how to do this I will need to create a cation exchange membrane to use instead of Nafion. In this post I will talk about what I have achieved so far, which is the first iteration of a PVA based cation exchange membrane.

Early on, it became clear that polyvinyl alcohol (PVA) was going to be the easiest polymer choice, as it is readily available and easily to functionalize. Phosphorylation also seemed as the easiest route towards functionalization, as highly concentrated phosphoric acid is easy to get and urea catalyzed phosphorylation reactions of PVA are already well known. The introduction of phosphoric acid esters provides the ability of the membrane to repel anions and allow only cation transport.

Experimental setup to measure the potential across one of the membranes created. The potential between the Ag/AgCl reference electrode and graphite electrode is measured with both electrodes on the same side, then the potential is measure again by placing the reference electrode on the opposite side. The membrane potential is calculated from the difference between same side and opposite side potentials. (a small amount of dye was added to the right side so that you can see how the membrane separates the solutions).

My first problem creating a membrane of this sort had to do with casting PVA films and being able to peel them off. These membranes are extremely sensitive and can easily stick to glass or to themselves, making the fabrication process difficult. I tried casting on glass petri dishes – with mold release – and was unable to remove them without breaking them. A friend suggested casting on Al foil instead, so I will be keeping this for a future experiment.

Furthermore, the few times I was able to successfully peel off films, the films then dissolved quite easily in water. Although I thought the phosphorylation of the PVA would provide some crosslinking, it definitely increases the solubility of the polymer in water, making things actually worse. Using things like aldehydes for crosslinking is not going to be work, but perhaps future experiments with boric acid or citric acid would help with this issue.

A breakthrough came when I realized that cellulose is also known to be phosphorylated with phosphoric acid plus urea and that it could therefore be cross-linked through a phosphoric acid ester with PVA. The cellulose could also provide a support, which would greatly enhance my ability to work with the PVA solids.

Final result of the process mentioned below

My fabrication process was as follows:

  1. To 15mL of ice cold distilled water add 1g of PVA, 1g of Urea and 1mL of 81-85% phosphoric acid. This is solution A.
  2. Place in a fridge for 48 hours, with occasional stirring/shaking. Surprisingly, cold conditions are much better for dissolving PVA because they discourage agglomeration.
  3. Wait till solution A is fully homogeneous, keep longer in fridge and shake/stir as needed.
  4. Dip a filter paper in Solution A. I used Stony Lab 101 but other fine grain filter papers should work just as well. Make sure all excess has dripped off and tap with paper towels to remove any excess.
  5. Place on a hot plate at 180C for 3min
  6. Flip it to the other side for another 3 minute.
  7. Repeat steps 4-6 once.
  8. Place on the hot plate with a petri dish on top (to keep it flat) for 1 hour.
  9. The result should be as shown in the image above.

The process seems to work. The resulting membrane is not soluble in water, is sturdy and easy to manipulate and loses the micro porosity of the filter paper. It is quite brown, which means some oxidation has happened, but reducing the temperature or time leads to membranes that are not properly crosslinked, and dissolve quite easily (leaving just a porous cellulose membrane behind).

To determine whether the above membrane is in fact a cation exchange membrane, I can measure its permselectivity. To do this, I measure the membrane potential between a 0.1M NaCl and a 0.5M NaCl solution (more details about this process on the first image in this post). The membranes produced in this way have permselectivity values between 0.5-0.7, which means that the membrane does in fact act as a cation exchange membrane. However, the membrane is nowhere as good as Nafion, which has a permselectivity >0.95 under these conditions.

I will now try changes in the composition of solution A and optimize the curing temperature to increase the permselectivity of the membrane. So far I think the fabrication process is quite straightforward which allows me to reproducibly fabricate the membranes described in this post.

Thinking about a membrane for my Fe/Mn flow battery

To build an Fe/Mn flow battery we need a cation exchange membrane to separate the catholyte and anolyte chambers of the device. In this post I want to talk about my initial thoughts about how to create a DIY membrane for this purpose.

Chemical representation of PVA (Polyvinyl alcohol) not to be confused with polyvinyl acetate (what PVA glue is made of).

Commercial cation exchange membranes do exist. Nafion membranes are the most commonly used, but their cost is too high. Just a small 10cm x 10cm square of Nafion can cost upwards of 50 USD, depending on the type of Nafion used. Lower cost membranes (like SPEEK based membranes) have been tested in the literature, but I cannot find any place that actually sells these “lower cost” membranes at a truly lower cost than Nafion.

To be able to make a viable DIY flow battery we need a membrane that we can make, that is lower cost. The requirements of a cation exchange membrane for the Fe/Mn system would be as follows:

  1. Not dissolve in water at neutral pH.
  2. Made from readily available, low cost materials.
  3. Mechanically stable.
  4. No reaction with any of the redox species in solution.
  5. Contain anionic groups (which makes it selective to cations)
  6. Have high conductivity

I looked at potential materials to build this membrane and PVA has become the most prominent base material. It is a polymer with OH functional groups, which I can use to react with readily available chemicals to create a functionalized polymer. My first experiments will involve using phosphoric acid, urea and potassium silicate to create functionalized membranes.

I will prepare 10% w/w solutions of PVA in distilled water, then add different amounts of the above mentioned additives to determine which compositions cast best and have the best properties. I will be casting the films in petri dishes, as this seems to be the most common method in the PVA membrane literature. I will also possibly anneal the membranes by heating them at different temperatures after they have settled.

Double chamber electrochemical cell I bought (haven’t received it yet)

I have also bought a double chamber electrochemical cell to perform experiments using these membranes. The idea is to measure if there is any crossover across the membranes and possibly also measure the ionic conductivity of the membrane.

To measure crossover of ions I can setup one side with the Fe salt and another with the Mn salt, then carry out cyclic voltammetry measurements on the Mn side as a function of time, to measure the appearance of the Fe peak (if there is any crossover). I can compare times between membranes as well. I can also test microporous membranes and non-functionalized PVA membranes, to obtain some baseline measurements. If I setup one side with just NaCl and the other with Fe, I can likely obtain more sensitive measurements (as I will have no current from reactions with Mn species).

Additionally if I use Fe-EDDHA I could sample the solution and measure the appearance of the Fe-EDDHA visible absorption peak near 500nm, which is highly sensitive given the chelate’s very high molar extinction coefficient. Although for this I would near to purchase a Uv-Vis spectrometer, which would cost me 500-1000 USD.

I can also measure ion diffusion by setting up distilled water on one side and a 3M NaCl on the other side and measuring conductivity as a function of time on the distilled water side. This will allow me to compare different membranes and see which ones transport ions faster. If I add Fe chelate to the NaCl I could perhaps measure both ion transport and selectivity simultaneously.

It will be a very interesting journey!

The best low cost Fe/Mn flow battery: Some perspectives about solubility and chelates

I have previously discussed my project to create a DIY flow battery using Fe/Mn chemistry. On this post I want to expand on the potential limits of this chemistry and some modifications that should enhance our ability to increase its energy density and performance.

My first idea is to attempt to create a flow battery using an NaFe(EDTA) solution as anolyte and an Na2Mn(EDTA) solution as catholyte. This battery would have a potential of around 0.74V, as I measured by cyclic voltammetry (CV) of the species involved. I commented on how the limit of solubility of these chemicals – without any additives – is limited to at best around 0.5M, which limits the battery power density to around 10 Wh/L.

This image shows some NaFe(EDDHA)

However, it is interesting to note that the solubility of these EDTA salts increases aggressively with pH, such that both can be dissolved above 1M at a pH of 7. I confirmed that the solubility increases aggressively as a function of pH, being able to create a solution that was around 1M for both compounds with 3M NaCl supporting electrolyte. To do this I used potassium carbonate to increase the pH gradually to the 7-7.5 range. I also confirmed that the reversibility of the electrochemistry was unaffected through CV, although both standard half-cell potentials are shifted negatively by around 50mV.

This increase in solubility is interesting, as it increases the power density of the battery substantially. If the compounds can be dissolved at 2M, then it would give the battery a density closer to lead acid, at 40Wh/L.

Sadly there are no published studies that show the solubility of EDTA salts as a function of pH, however one of the few published studies of Mn-EDTA in flow batteries (here) shows that you can dissolve Na2MnEDTA at concentrations past 1M. I have bought some additional Mn-EDTA to perform my own solubility experiments, I will let you know what I find out.

Image from this study, using a Zn/Mn flow battery at slightly acidic pH.
Image from this study using Fe-EDDHA at a slightly basic pH.

Another interesting note is to look at other Fe chelate candidates. While EDTA is the lowest cost chelate, the Fe-EDDHA chelate is interesting, as it has a significantly more negative potential Vs Ag/AgCl (-0.6V instead of -0.1V for Fe-EDTA). Recent literature of Fe-EDDHA chelate characterization and its use in flow batteries already shows its practical application (here and here). This increases the potential of an Fe/Mn battery from 0.74V to around 1.2V, which is a decent potential to achieve within the stable window of water at pH 7.

This means that, if using Fe-EDDHA, we could potentially achieve a power density of up to 80Wh/L at a solubility of 2M. If the solubility limit is around 1M, then it should still allow us to get to 40 Wh/L. With this in mind, the Fe/Mn chemistry should match lead acid power density and be a strong competitor against Vanadium based chemistries. This is especially given the fact that Fe/Mn are super abundant and this battery is based on already commercially available chemicals in water, at a neutral pH.

As you can see above, the anolyte and catholyte I propose have been tested, so this is definitely a system that can be built in a rather straightforward manner.

Towards a DIY Manganese/Iron flow battery. First experiments using cyclic voltammetry.

Flow batteries are a great approach for large scale energy storage. While in a battery the amount of energy is constrained by the mass of the anode and the cathode, in a flow battery the cathode and anode are stable electrodes (most commonly graphitic foams) and the energy is stored in solutions that are pumped through these electrodes.

General diagram of a flow battery.

Many of the lowest cost approaches to the chemistry of a flow battery are unable to fully take advantage of this, because they reduce a metal to its solid form on the anode. Approaches using Fe and Zn where this happens are common. The deposition of a solid metal then creates additional issues with both passivation and with dendrites, which can end up shorting the flow battery down the line.

To solve these issues, we need a chemistry where both the oxidation and the reduction happen in solution (with no solid formation on the electrodes). Additionally both of the half-reactions need to be reversible. From a DIY perspective, they should ideally happen under mild conditions and, to make things even more difficult, we need materials that are low cost and that can be easily purchased.

Relative abundance of elements in the earth’s crust

Manganese (Mn) and Iron (Fe) are some of the most common elements on the Earth’s crust, so they fulfill the cost issue. However, when building a flow battery with Mn, we find that the oxidation reactions that Mn is involved in generally involve the formation of insoluble Mn oxides. This happens because Mn3+ is generally unstable in solution and reacts with water to create Mn2+ plus an Mn4+ oxide or hydroxide.

However, a few papers have been published on the use of Na2Mn(EDTA) in flow batteries. This chelate – a commonly available fertilizer – protects the Mn3+ from reacting with water and enhances the reversibility of the reaction. Given the potential where the oxidation of Mn(EDTA)-2 happens, I thought it could certainly be coupled with the reduction of Fe3+ to create a flow battery. Additionally NaFe(EDTA) is also a low cost highly available fertilizer we can use.

On a previous post, I spoke about a setup for electrochemistry that I created, which allows me to carry out several measurements in solution. Using cyclic voltammetry of a solution containing Na2Mn(EDTA) and NaFe(EDTA) I was able to characterize the system and obtain half reaction potential values for the Fe and Mn reactions mentioned above.

Half reactions and half reaction potentials measured Vs an Ag/AgCl reference electrode. For Mn the reduction is shown.
Cyclic voltammetry used to obtain the E1/2 measurements. The Fe reaction happens to the left while the Mn reaction happens to the right.

If we add the two potentials above, we can obtain the expected potential for our battery, which would be 0.74V. This is not very high, which means that the energy density of our flow battery system is going to be low. If we consider the solubility of both compounds, then we expect the power density of this battery to be around 10Wh/L. This means that you would need 100L of 0.5M NaFe(EDTA) and 100L of 0.5M Na2MnEDTA to get a 1kWh battery. This means 18.3kg of the Fe salt and 19.5kg of the Mn salt. You will also need around 35kg of NaCl as a supporting electrolyte.

At retail you can find both Fe and Mn salts for a price of around 6-15 USD/kg (if you buy 25-55lb bags). On the low end this means the cost would be 226.8 USD/kWh and on the high end 567 USD/kWh at a retail price point. In pallet amounts, the cost for both is around 2 USD/kg, so the cost goes down to 75.6 USD kWh. Note that this is only for the Fe and Mn salts.

The challenge is now to create a small electrochemistry setup with two electrochemical chambers separated by an ion exchange membrane where we can carry out some initial charge/discharge measurements and measure the cross-over of ions without the need to do any sort of pumping. This is also going to involve the design of some DIY low cost membranes, since Nafion membranes would be extremely expensive. Additionally, since the conditions are so mild (pH 5-6), we can use some modified PVA or cellulose cation transport membranes that can be produced for very low cost.

A home setup for cyclic voltammetry

With my DIY potentiostat/galvanostat, I can perform many types of experiments. One of the most useful – especially for its link to battery chemistry – is the cyclic voltammetry (CV). In a CV experiment, the potential between a working and reference electrode is changed within a range and the current at each potential measured. The shape of the plot measured, gives important information about the chemistry in the solution. You can tell if anything is oxidized or reduced, at which potentials these processes happen and you can also get an idea about the reversibility of the reactions taking place.

My home cyclic voltammetry setup

A CV setup has several components. The experiment is carried out in an electrochemical cell, which is a glass vessel that can hold all of the electrodes in place, without any risk of the electrodes ever touching each other and causing a short. Although such setups can be built at home, relatively low cost high quality solutions are indeed available.

The cell has a three electrode setup. The first is the working electrode (WE) which is the electrode where the relevant electrochemical processes will happen. This electrode is usually made of an inert material – glassy carbon, platinum and gold are most common – with a high polish and a limited surface area. This is because we want the surface of the electrode to be reproducible and to always provide us with the same measurements.

Image showing the WE, RE and CE of my home CV setup.

The reference electrode (RE) provides us with a potential based on a chemical reaction that is always happening at a very defined potential. For best results, these are usually 1 electron reactions that happen at a very small scale – which means a current draw below pico amps – and normally involve an equilibrium with an insoluble solid. Popular choices are calomel and Ag/AgCl electrodes.

Finally the counter electrode (CE) is the electrode that is connected to ground and has a polarity opposite to that of the working electrode. It is meant to complete the circuit. This is generally made of an inert material and should have a very high surface area compared with the WE, such that reactions are never limited by it. Popular material choices are graphite rods and palladium and platinum wires.

First successful CV experiment using my home setup. This was a transition metal mix (including Fe, Mn, Cu and Zn in 15% phosphoric acid). This was simply to ensure everything worked as expected.

Although the above might all sound complex and expensive – I got quotes of over 1000 EUR for the above with some EU supplier companies – I was able to find everything on ebay for relatively low prices from Chinese suppliers:

The total price for the entire setup including shipping – not counting the DIY potentiostat – was around 162 USD. I am very satisfied with the quality of all the components that I have received. I have also successfully performed my first CV experiment (showed above).

My first idea with this CV setup is to explore manganese chemistry and measure the reversibility of Mn2+ oxidation reactions in concentrated sulfuric acid solutions. Highly reversible Mn+2/M+3 reactions are very important for Mn based flow batteries.