[lesaid]

I'm trying to explore a phenomenon I don't understand with electrolysis using aluminium electrodes in distilled water. The intent was to look at the effect of the surface oxide layer on the conductivity between the electrodes and the water, compared to other materials which don't have such a layer. I was initially curious as to how aluminium can function as a sacrificial anode when it is covered by an impervious and insoluble oxide layer.

My setup has two aluminium foil electrodes, folded into interleaving concertina shapes to get as much surface area as possible with the minimum gap. Using 100 mV, I measured an initial current spike of around 5 uA, decaying away over about a minute to approaching 200 nA. If I disconnected it for a few seconds and reconnected, this behaviour repeated. Same if I reversed polarity.

I chose 100 mV hoping it would be a level too low to induce chemical changes in the electrodes - and hadn't expected a decay curve like that. So I tried with 10 mV and got exactly the same shape, this time starting at around 800 nA, decaying over around 30 seconds to below 10 nA (the minimum resolution of my measuring equipment).

I am putting together a measurement front-end that will allow me to measure currents with pA resolution, and to try with lower voltages still - but this leads me to some questions.

1. Any ideas what this behaviour might be due to? It is behaving almost like a poor capacitor, except that it can't be - the capacitance would have to be way higher than is possible for those electrodes (I think). I have another hypothesis about ion distribution in the water, the rate of natural dissociation of water and the speed of diffusion of ions, but need to take measurements to a greater precision to test it.

2. How can I get my equipment seriously clean using chemicals available to me as a non-professional? Currently, I am cleaning everything thoroughly in hot water with detergent, then rinsing in tap water, then washing in isopropyl alcohol and finally rinsing in distilled water - doing this to both electrodes and beaker - while wearing latex gloves to avoid skin contamination. But in spite of this, I am measuring a voltage developed between the electrodes - which has to mean some kind of contamination. No point in going to more precision and lower voltages if I can't reduce this obfuscating effect!

3. The voltage developed across the electrodes is higher after running a few trials. I thought 100 mV was below the level that would trigger chemistry in the cell - and 10 mV even more so? The behaviour tells me that some chemistry must be happening, but how? I thought over a volt (1.4?) was necessary before aluminium would interact with the water? And there must be energy required to break bonds between the ions in solution and the water molecules - can 10 mV really provide enough to do that?  If I do take this down lower, will there be a voltage at which there really will be no chemistry - in which case, should the current drop to zero, as without chemical interaction, how will the charge carriers (ions) manage to transfer electrons?

At 1V and above there was another weird behaviour, but I'll leave that for the moment lest this post get even longer. I'm trying to get to the bottom of one thing at a time!

I tried a similar experiment, briefly, with copper electrodes - found about three times the current levels and a similar decay curve - but bizarrely, the decay curve was much smaller and went in the opposite direction - was initially low and built up towards a sustained level. That has me even more confused, but I'm parking that one for a bit until I understand more about the behaviour with aluminium.

My apologies if these are daft questions - I'm not really a chemist - but this has me perplexed and fascinated!

[Borek]

Have you checked if the capacity is not that of a double layer?

No idea if that's what you really observe, but the capacitor undoubtedly is there.

[lesaid]

Have you checked if the capacity is not that of a double layer?

Thank you for that suggestion - I hadn't thought about that, and apart from vaguely, didn't really know what a 'double layer' was until you prompted me to look it up.

I don't know either whether this explains the observed behaviour, and that hypothesis is contradicted by the observation I mentioned with copper electrodes when the decay appeared to go the other way. But that was a single set of observations in a rather informal trial and might have been in error, or affected by contamination.

I'm now correlating the voltages, currents and decay times for multiple trials ranging from 10 mV to 1 V (above 1V, the decay is swamped by a linear increase in current over time - which I find equally perplexing but I'll worry about that later!). I'm then going to calculate what capacitance would explain the data and with the aid of a spice electronics model, produce graphs to compare with the data and see if they match. But at first glance, we're talking about at least some hundreds of uF!  Not a clue whether that's plausible or not!

If this is the explanation and if my 'lay' understanding is correct, than I think the capacitance should be unchanged by the electrode separation, though the effective series resistance would obviously change. But the capacitance should drop with increasing voltage (increasing thickness of the 'insulating' layer)?

[lesaid]

While starting to look at the effect of electrode oxide coatings on the conductivity between them and water, I came across a strange effect that I cannot explain (superficially looks like an impossibly high capacitance between the electrodes). I'm now planning a series of more rigorous experiments to establish whether it is repeatable, and if so, exactly how it behaves.

To do which, I want to get a range of electrodes as free of contaminants as possible, using materials available to me in the household. What I did for the initial trials was simply to wash the electrodes in hot water and detergent, then rinse them in isopropyl alcohol and then in distilled water. Can I do better than this? As may be obvious, I am not a chemist, though I have a little basic knowledge. Physics is more my 'thing'

I don't want to abrade them, because I want to be able to calculate the surface area of the electrodes in contact with the water, and scoring / scratching on the surface might make that impossible.

Electrodes I would like to try and would need to be able to clean include aluminium, copper, lead, gold (gold leaf on some rigid backing) and carbon. I would say platinum, but I don't think wire wouldn't give me a large enough surface area in contact with the water, and anything else would be too expensive!

Any ideas?

For interest, the experimental setup for my first informal trial used aluminium foil electrodes, several cm² in area, placed close to each other in distilled water.

Voltages used ranged from 10 mV to 1V, and currents ranged from 10 nA to 1 uA.

Below about 200 mV, conductivity settled after a few tens of seconds to a consistently low value. Above 400 mV, after initially starting to settle, conductivity rose steadily for many minutes, indicating (I believe) that some unwanted chemistry was going on. It's the behaviour at low voltages where (I hope) there is no chemistry happening, that interests me.

A quick trial with copper seemed to show very different behaviour - need to repeat under more controlled conditions. I suspect contaminants are playing a large part.

[billnotgatez]

@lesaid
I want to make sure of your process.
Are you just using distilled water and the electrodes with no catalyst.

[lesaid]

that is correct - distilled water with nothing else. Presumably there'll be dissolved gasses in the water - that is something which I guess I might have to look at, but for now, I just want to repeat what I have already done, under much more strictly controlled conditions, and get rid of all the unknowns that I can. Most of which are around

contaminants
area of contact of electrodes with water
geometry of electrodes with each other
temperature
experimental procedure (timing, data logging etc.)
calibration of measurement devices
general repeatability

Most of these I can do something about, though it will take a little time to plan and set up - but not sure how to remove contaminants!


Then I'll see where that takes me!

[Enthalpy]

Aluminium is the wrong choice as electrodes. Its behaviour depends fully on the oxide layer it always develops, and this layer depends on everything. Forget it.

Zinc won't give consistent results, for the same reason. Titanium, niobium, tantalum are some metals known for that behaviour. What you need are precious or semi-precious metals, among which copper, tin, nickel are affordable - but often participate in the reaction. You should consider graphite. A thin gold layer is an option, yes; the connectors of computer RAM modules have a gold layer, just saw the rest off.

The electrode area is not measureable. Metals are not flat. If they sem flat at human scale, it's because their corrugations are smaller than we can see, but they still change the area, since the effect depends on the corrugations' angle, not depth. Depending on what you measure, small corrugation can be negligible, but if you measure for instance a capacitance to the electrolyte, they're paramount.

I clean printed circuits, hence copper, with the hard back of a sponge. It desoxides copper perfectly and isn't as abrasive as sand paper. If a metal is dirty, you may try an eraser first.

[lesaid]

Thanks for the reply 

This whole thing started out as an attempt to find out how materials with an oxide layer compared with those that don't, following a discussion in the pub about sacrificial anodes, and wondering how an 'impervious' oxide layer allowed the anode to pass a current through it. And if it isn't as impervious to water as it seems, how does it protect reasonably well against corrosion in damp air?

That is what got me interested in comparing aluminium electrodes with carbon (graphite) and gold. I tried copper simply because I had some copper sheet to hand - and found the behaviour fundamentally different (not just in scale) from aluminium - hence my interest in that.

One specific thing I want to find out now is whether what I'm seeing really is a capacitance (by finding out whether energy really is being stored), and if so, whether it is due to a more resistive oxide layer or whether it is a 'boundary effect' as was suggested in a previous thread. Or a combination of both. That's why comparing the behaviour of aluminium to that of gold or carbon is interesting to me.

I'm also wondering about iron/steel, which has an oxide layer that is very much not impervious!

I take your point about the smoothness of the electrodes. Perhaps I could do a kind of 'sensitivity analysis' of similar electrodes, one pair of which has been roughly sandpapered and the other not, to establish how significant smoothness is to whatever I am seeing.

I'm also hoping that by keeping the voltage as low as possible (planning on at most, 10 mV for the next round of experiments, though hoping for much lower still), I will avoid unwanted chemistry confusing the results.

I know this is probably a naïve investigation to a professional chemist who probably would know the answers already, but it is interesting exploring it - and also an interesting challenge to develop as rigorous an experimental setup as possible.

I hadn't though of cleaning with something as simple as a hard-backed sponge!! I wouldn't try that on gold leaf though - so a cleaning regime based on chemistry rather than mechanical action would still be attractive!

[Enthalpy]

The rationale is clearer now.

A good oxide layer takes an alloy very resistent to corrosion. Aluminium foil for food packaging is a candidate (it's about 99.5% pure), but aluminium for capacitors would be much purer. Titanium works the same, and tantalum and niobium are also used to make capacitors.

As a teenager I used sodium bicarbonate in banal aluminium (drug tube) to make a capacitor. It takes few hours with a modest current, you get a true capacitor of very few µF. You can leave the bicarbonate for further use of the capacitor.

One idea is that the voltage across the oxide layer transports the ions that make more oxide. This stops as the layer gets thick enough for the applied 'forming" voltage, chosen about 50% bigger than the "operating" voltage. Consequently:

• You get a very uniform thickness, just as much as is needed, thinner than a plastic film can be and with a bigger permittivity, making a big capacitance.
• An electrochemical capacitor has a polarization. Reverse voltage, even a fraction of the operating voltage, destroys it. But 100mV is fine; usual tantalum capacitors accept reverse 1V.
• Room temperature has about 0.026V force to move ions, so an oxide layer made with 5V or 500V is impervious to humidity - if the alloy makes a good oxide.
• "Anodization", essentially the same process that makes capacitors, serves to protect mechanical aluminium parts against rain.
• The electrolyte is resistive, so electrochemical capacitors work properly at low frequency only - say, under 20kHz.
• Without forming the capacitor first, the oxide layer will be so thin that you'll see the series resistance of the electrolyte - and anyway, the temperature would move a few ions through it, in the first line protons, which are not corrosive but transport current.

So your attempt has excellent chances to work. Bicarbonate in good amount, aluminium foil compared with copper as minus electrode candidate, graphite from a 1.5V battery (mind your fingers, corrosive electrolyte!) or iron as the plus electrode, possibly 50mA*24h for 1dm². You can watch the voltage increase regularly up to the supply voltage where the current stops or drops.

[lesaid]

Thanks for that - you've given me some interesting ideas and some more questions!

I wasn't actually trying to make a capacitor at all - though I might now, out of curiosity. I was trying to understand the phenomena that I'd noticed while looking at conductivity between different electrodes and water.

So far, at voltages from 10-100 mV or so, I've seen what superficially looks like a very high capacitance - based on the shape of the charge curve, could even be in the hundreds of microfarads, with maybe 200-300K parallel (leakage) resistance. Not sure what the series resistance would be though - my original data capture arrangements couldn't catch the height of the initial charging spike. But it cannot be more than 10-15k - probably significantly less. With an electrode surface area of maybe 15-20 square cm, and a separation of a few mm, I think this is of the right order of magnitude for distilled water.

However, I've not so far been able to prove definitively that energy is being stored - that will come in the next round of trials.

So far also, the behaviour seems completely symmetrical - no evidence of anything forming that lasts long enough to influence a subsequent trial a few seconds later. And the very first time I try this with fresh electrodes gives exactly the same behaviour as when those electrodes have been in use for hours. So long as the voltage never rises above maybe 200 mV.

If this is caused by a pre-existing aluminium oxide layer, then trying this with gold leaf, or graphite should presumably not show this effect.

But what really has me intrigued is that with copper instead of aluminium, the behaviour is completely different and looks nothing like a capacitor. On connection, instead of an initially high (maybe as much as a microamp or so) and quickly decaying 'charge' current, I get a fairly low initial current that rises to approach a higher value!  However, I need to repeat all these measurements with more care and precision and a better experimental setup before I'm going to try to read any more into these results.

I suspect contamination is affecting things - hence my original question and my keenness to find good ways of getting the electrodes as clean as possible.

[billnotgatez]

For the Aluminum, does one electrode exhibit more oxide than the other?

[lesaid]

I've seen no visible evidence of any changes to either electrode, so long as the voltage used stays below 100 mV or so. There is a slight asymmetry - the currents measured are slightly different in opposite directions, but that asymmetry was there on the very first time the electrodes were used, and did not change throughout all the trials. I am convinced that was not due to the measurement equipment however, and put it down to another issue with contaminants!

Just to clarify - my experimental procedure for each voltage level was to

• give plenty of time for the distilled water to reach room temperature before starting the first trial
  
• short the electrodes together for a while - measuring current - to check I didn't inadvertently have a 'battery'
• Apply my test voltage (something between 10 mV and 200 mV), filming a digital display of current
• Monitor the current until it settled at a new value (two or three minutes)
• Reverse the polarity and monitor again until it settled
• Repeat the above four steps twice more to check the results were repeatable

This procedure was repeated for multiple voltages. I then applied 100 mV in one direction and left it for an hour or two - before repeating the 10mV cycle, to see if anything long term was forming on the electrodes. This final test gave similar results to the first one, so I conclude no long term changes.

Finally, trying it at higher voltages (up to 1V) showed initially a similar charge curve, but after an initial drop, the current rose again, linearly, until levelling out at around the level that I estimate corresponds to the conductance of the distilled water. Above 1V, that initial drop is swamped completely by the subsequent quick rise. At 200 mV, the first indications of that rise are barely perceptible.

I have been hindered by the experimental setup - this is what I'm now improving. Problems included

• the electronics to provide the voltage source were thrown together on a 'breadboard' - the electrical contacts were mechanical and dodgy when measuring to nA and mV resolutions
• Swapping polarity meant moving wires about, or operating a switch on the breadboard, both of which caused mechanical vibration disrupting the physical circuit
• The slightest vibration to the electrolytic cell destroyed the measurements
• the measurement device had a refresh rate of about three per second - I filmed it with a camcorder and recorded the readings afterwards - not fast enough to capture the initial charge spike
• Running each trial took a long time and was very labour intensive

Changes I am making to the experimental technique over the next few weeks include

• using rigid electrodes, firmly clamped into measurable positions, rather than fabricating them out of foil
• putting the cell into a controlled water-bath so I can know the temperature will be consistent when running trials on different days - and can experiment with different temperatures if I wish
• building the electronics properly, so no more breadboard connections
• Using a USB-attached data acquisition interface with a PC to allow me to control the applied voltage and monitor currents (and temperature) with a computer - that also gives me a much faster measurement refresh rate, better current resolution and automatic data logging.
• running the whole thing on a bench separate from the PC to avoid physical disturbance of anything during or between trials
• Automating the test cycle so I can set it running and leave it unattended and undisturbed for several hours, cycling through different voltage and polarity trials (I'm writing the program for that at the moment)

[lesaid]

Aside from the experimental side of it, here are one or two 'chemistry' questions that have sprung to mind ...

Thickness/strength of aluminium oxide layer
===========================
Enthalpy said "• Room temperature has about 0.026V force to move ions, so an oxide layer made with 5V or 500V is impervious to humidity - if the alloy makes a good oxide"

How does an oxide layer 'remember' what voltage was used to create it? I would have thought the imperviousness (if that's a word) of the oxide layer would depend on the thickness and strength of the bonds. Would not the bond strengths be the same regardless of the voltage present when they were created, and the thickness be related to how long the layer was formed over?  Or is it that the thickness of the layer is limited by the voltage used, because a thicker layer means a lower voltage gradient across it?  Which would imply that the thickness of the layer and its electrical resistance would be directly proportional to the applied voltage, if applied long enough for the layer to reach maximum thickness? And 500V or more would result in a really thick layer!

Sacrificial anodes and impervious oxide layers
=============================
How do sacrificial anodes work, if an impervious oxide layer develops over time in response to the current flow? This is the question that triggered all this experimentation in the first place!  Is it that in practice, cathodic protection is never used in a distilled water environment, and that other ions present in the water prevent such an impervious coating from forming? 

Minimum voltage capable of passing current through water
=====================================
If I were to reduce the applied voltage indefinitely, in theory, would I reach a point where the voltage would be insufficient to exchange electrons with the ions in the water, and so the conductivity would suddenly and dramatically drop?  If so, any idea what kind of voltage we are talking about?

I'm also curious as to whether I could probe that behaviour - though the limiting factor on low voltage is probably chemical contaminants on the electrodes. My measurement kit is theoretically capable of resolving currents down to 1 fA, but in practice, I've found a resolution of around 20 fA the absolute limit (albeit with very long settling times), after taking extreme precautions with shielding (from air currents and mechanical vibration as well as electrical). However, I'm pretty sure the actual limiting factor on low-level measurements will be contaminants on the electrodes again, creating potential differences independent of the voltages I'm applying. I think I'm seeing the first signs of this already at the nA level (responsible I believe for the slight asymmetry in my measurements). Unless I can get the electrodes much cleaner, I don't think I'd get much sense from voltages below a few mV.

It's back to the original question again - how to get super-clean electrodes?

[billnotgatez]

Does this help

http://www.ehs.wisc.edu/chem/GlasswareCleaning.pdf

You may never remove all the stuff you do not want on and in the electrode. but you can get the lions share.
That should leave enough surface to give reasonable results
In the end you may need to do statistical analysis on several trials.

[lesaid]

That link certainly gives food for thought. Most of it though is out of my reach as I don't have access to a chemistry lab or unusual chemicals. I have no chemistry equipment here, and few chemicals here beyond those found in ordinary households, and a few things for electronics (isopropyl alcohol, and some etchant for making PCBs - maybe some ferric chloride somewhere).

I was hoping to find a way of doing something imaginative with ordinary household materials - be it vinegar or drain un-unblocker, but I'm happy to go get anything that is available to me, so long as it can be used safely without a chemistry lab!

I'm guessing though that the contaminants won't be anything too esoteric - mostly grease through handling and general grime!   I'm not too bothered if a bit of the surface is not in contact with the water, so long as it is a small proportion. What bothers me more is anything that will end up reacting with ions in the water and producing a spurious potential to mess up my measurements. Right now, I think I'm on the edge of that problem, and I'd like to push it further back!

One idea - for copper and perhaps zinc at least, would a short dip in PCB etchant (sodium persulphate) to take off a thin surface layer (along with hopefully any contaminants on the surface), followed by a rinse in distilled water sound sensible?

I also came across this link which suggests a combination of ferric chloride and sodium persulphate to etch aluminium?

 http://www.google.co.uk/patents/US3281293


Actually - that would presumably also tend to even out the worst of any microscopic scoring that would confuse measurements of electrode area, as Enthalpy mentioned?

Do those ideas have any merit?   Or is dissolving the surface of the electrodes going 'over the top'?