[billnotgatez]

The link was only meant to give you the general idea.
I think you can mostly stay with detergent, tap water, distilled water and acetone.
The Isopropyl alcohol is good as well.
Anything more may be over the top, unless you think your electrodes have some sort of coating on them. And, from what you say it appears not to be the case.

By the way, you can get some solvents like acetone at your hardware / paint store.

I am eagerly awaiting your data and analysis.

[lesaid]

Thank you for your comments and interest

I have searched on the web and not found a lot about these sorts of behaviours, so that makes it particularly interesting to me.

However, don't expect data and results next week - I've got quite a bit of preparation to do before I'll be ready for the next set of runs - including a significant amount of programming and some electronics, as well as the physical setup. And the work is having to fit around a lot of other things in my life so it's kind of a background activity!

So I welcome any more comments/ideas in the meantime, and I'll gladly share whatever comes out of this on the forum as soon as I've got something. Hopefully by the end of the year.

[billnotgatez]

For those following these posts

We have combined the older threads with the newer threads for readability.

Below are the combined titles-
Cleaning electrodes for electrolysis of distilled water.
Perplexed by electrolysis of distilled water with aluminum electrodes.

[lesaid]

An update on the experiments described in this thread, and some questions at the end .

I've now set up a more controlled experimental setup to explore this behaviour, and am rehearsing my procedures and software with a couple of aluminium foil electrodes. I intend to fine-tune things based on the results, and then run this again 'properly' with rigid, measured electrodes of various metals.

The experimental setup consists of

• a plastic container sitting in a temperature controlled water bath
  two aluminium foil electrodes dangling into distilled water, with roughly 10x10 cm immersed in the water, and about 5-10 mm separation.
  
• The electrodes are connected in series with a 10K 'current sensing' resistor to a D-to-A convertor, allowing software control of the voltage applied to the electrodes.
  
• The voltage drop across the resistor is also captured by analog 'input' channels on the same DAQ device, allowing the current through the resistor to be graphed interactively and logged.
  

The software is set up to run a series of repeated test cycles, each of which applies a voltage alternately in opposite directions, with 'recovery' periods in between when the differential voltage is held to zero (effectively, a short circuit).

Each time a voltage change is made, it is held for around 20 minutes to settle completely, before the next change.

Each cycle is repeated identically five times

Each group of cycles is repeated for a range of voltages, starting at 1.2 mV, and increasing progressively up to 2V.

The whole procedure takes several days to run and (thankfully) runs unattended, while collecting a large dataset.


My first run is still in progress, and initial results appear to show a true capacitance that is independent of voltage (at least, up to 250 mV) and is around 350 uF.

The series resistance is also independent of voltage, at about 5.5 k

The parallel resistance however, changes dramatically over time and/or as the voltage increases. On the first four repetitions at 1 mv (in the case, with five minutes in each direction and five minutes 'rest' time between the changes), the parallel resistances measured rose from an initial 10 k up to 81 k.

A later run at 250 mV shows the same capacitance and series resistance, but a parallel resistance of over 1M


If we assume that this system is made up of two capacitors in series (one for each electrode) with a building layer of aluminium oxide as the dialectric, the each capacitor would have to be around 700 uF, and assuming a dialectric constant of 10 (from wikipedia), the oxide layer would be around 1.2 nm thick.

Guessing from the atomic radii of aluminium and oxygen that the diameter of an oxide molecule might be around 300-350 pm, then there this layer would be around 3-4 molecules thick.

Given that the capacitance does not change over the period being measured, the thickness of this layer must stay about the same, even though its electrical resistance went up by a factor over 100 to over 1 megohm (both electrodes in series - that would be 500k each).


So - some specific questions .

• Do these results make sense?  Is that thickness of oxide layer reasonable?
  
• What happens to the oxide layer to increase its resistance if its thickness doesn't change?
  
• Is it reasonable that a voltage as low as 1 mV would be enough to cause the layer to change like this?   Note that the aluminium was left disconnected in the water for a day before the experiment started.
  
• Am I right in assuming that aluminium oxide would be the main constituent of the layer?  I was wondering whether it might be a mix of aluminium oxide and aluminium hydroxide, given that I'm reversing the polarity every few minutes. But aluminium hydroxide has a much lower dialectric constant (2 as opposed to 10), so if that is part of the layer, the thickness must be less, down to a single molecule or two?
  

Thoughts/comments?


What I'm doing now, is waiting for this set of tests to complete, and I'm writing a program to analyse and interpret the logged data automatically, rather than having to repeat the tedious manual procedure over and over again!

I'm also tuning the software that runs the tests, and planning what I want to get out of subsequent runs.

Will probably be a little while before I'm ready to start the next set of trials, but a good bit more to get out of these first results yet!

[Borek]

Sounds like you have reinvented http://en.wikipedia.org/wiki/Cyclic_voltammetry and probably rediscovered http://en.wikipedia.org/wiki/Double_layer_(interfacial).

[lesaid]

I had never heard of cyclic voltammetry, and I don't think I'm quite there yet!  I'm working with voltage steps and settling times rather than continuous variation. But one thing which does intrigue me, is how fast the ion equilibrium in water re-establishes itself after being upset - e.g. by an electrode sucking away electrons. I was wondering whether, as I try to capture the peak current after a voltage change, whether momentary unavailability of ions (until more form or diffuse in from farther away) might be affecting the shape of the curves that I see. At an initial quick skim of your internet link, that looks like that approach might be interesting.  I've tried to come up with a rough guess as to how fast equilibrium might be restored, based on a Wikipedia comment that in one model, it was found that a water molecule might on average expect to exist for around ten hours before undergoing a collision with enough energy to cause disassociation.  Relating that to the ion concentration in water makes me think the answer is a few milliseconds, but so far, that's little more than guesswork.  If true though, that is slow enough that I might be able to detect the effect.

On the double layer - following your earlier post (ages ago) suggesting this possibility, I've been thinking about it. If I understand things correctly, a double layer effect would only last for as long as the electrodes are kept charged. So if it's a double layer effect that I'm seeing, and I were to switch everything off and go back to my initial run at 1 mV, I should find everything should go back to the original state, with a series resistance of around 10 k. But if it's an oxide layer, the series resistance would persist at 1 M or more.

If that presumption is right, then data so far would point to a persistent oxide layer, but it will be easy to test specifically when the current run ends.  Perhaps it's a combination of both, but I'm guessing the oxide layer is the dominant effect?

[Borek]

one thing which does intrigue me, is how fast the ion equilibrium in water re-establishes itself after being upset

This is not a trivial question, as it can be answered on several levels and there are several different levels of equilibrium. In general these processed are limited by the electrolyte transport, which can be limited by diffusion or mixing or by by a combination of both. I am afraid there are thick books on the subject, the only one I can think of ATM is Fundamentals of Electrochemical Analysis by Zbigniew Galus, published by Ellis Horwood in 1994. I am sure there are other books I just don't know them.

On the double layer - following your earlier post (ages ago) suggesting this possibility, I've been thinking about it. If I understand things correctly, a double layer effect would only last for as long as the electrodes are kept charged.

Again - this is not trivial. First chapter of the book I listed is devoted just to the most important (but not all!) effects of the double layer. As a first approximation it behaves just like a capacitor, but it is not necessarily completely static, it can change during the experiment.

[lesaid]

thanks for those references - and I'm not afraid of 'thick books' (except perhaps their price!!). However, I have to watch how much time I spend on this as I am a 2nd year undergraduate (mature) student in mathematics and physics - and need to spend most of my available time on study for that.  This is one reason why it is taking me quite a while to do these 'electrolysis' experiments.

So I won't right now be launching off into an in-depth study of electrochemical analysis, but my interest has definitely been triggered and I've noted your reference to explore when I'm next able to devote real time to it. Perhaps I should be studying physical chemistry rather than straight physics !  And I think some of the topics I'll be studying in the next few months will directly pertain to all of this.