Sciencemadness Discussion Board


 Pages:  1  2    4  

len1 - 11-9-2007 at 19:50

Thanks Der Alte. The PS I did make a long time ago - when we had that discussion about what amperage wpould be right. I thought at the time, and thats certainly turned out to be the case in this experiment that you need 10A+ for good results. Below about 10A my cell was producing nothing. That could also be due to the fact that to get the current that low the voltage had to be dropped to 3.6V.

The transformer you see on the photo came from a dirt-cheap welder with an adjustable current screw. Welders give 20-30V or so RMS open circuit, so I had to wind down the secondary. I took a tap at 4V 8V and total 12V. About 2.2 V of that is dropped in the diodes. These are 35A full wave bridges connected 4 in parallel to give a total capability of 140A. The PS is capable of about 70A before the cheap chinese transformer (they would the secondary with Aluminium, covered with copper-coloured enamel, to make you think its copper??) starts overheating. The heatsink for the diodes you see is actually a bit of overkill, they hardly warm to the touch in this arrangement at 70A. It would be better to mount the fan next to the transformer. The small red winding you see on the transformer is power for a digital 200mV LCD meter which shows the current you see. I got a 100A shunt from Jaycar across which the LCD meter is connected directly. It is switched with the switch you seeon the panel to show alternately the voltage/current. This is measured through a low pass filter (frequency about 10Hz) so you get to see the mean value, rather than the useless peak.

[Edited on 14-9-2007 by len1]

len1 - 11-9-2007 at 19:54

Originally posted by 12AX7
Huh, then I'm confused about where the sodium comes from.

So the nickel anode bubbles off -- you said Castner, so this is NaOH? -- O2 and H2O, while the cathode is placed in the center of the shield and periodically harvested?


Yes thats absolutely right Tim. Il post some more pics as soon as I find out how to post them all at once. Len

Xenoid - 11-9-2007 at 21:30

Great stuff len1.

I'm going to give this a go myself!

The photos are fantastic, but is there any chance you can produce a simple cross section drawing, showing how everything is arranged. Even just a photo of a pencil sketch with the parts labelled would be a great help!

Regards, Xenoid

len1 - 11-9-2007 at 22:00

Yeah OK Ill try. Ill post a few more pictures first. I am really glad you also want to give this a go.

len1 - 12-9-2007 at 21:12

Here are more pictures of the cell

Na13.JPG - 87kB

len1 - 13-9-2007 at 19:21

I ran the cell for 150 minutes yesterday at an average current of 47A, and the resulting Na ingot can be seen in the picture.

Now the interesting thing is the current yield.

47*150*60 = 432000 Coulombs.

Now 96000 Coulombs at 100% current efficiency give 23 grams of Na. Assuming 1 gram of the weight in the pic is crud (I have tared-out the parafin-soaked paper) that gives a yield of 23.6 grams. The percentage current yield is then

(96000/432000)*(23.6/23) = 22.8%

The maximum theoretical current yield in a castner cell is 50% due to water from the anode diffusing to the cathode (where it generates H2, so necessary to ensure the Na doesnt catch fire). And 18% is a long-term yield quoted for industrial Castner's, so the result is not bad.

Another interesting number is the cost, which depends on the energy efficiency. As stated, and despite whats written in the literature, this 'small' Casnter cell does not need external heating during electrolysis. So the entire energy cost is the electric power. 47A at 6.4V for 150mins assuming 80% transformer efficiency (the diode efficiency has already been accounted for in the voltage drop) is 0.94 kW-Hr. Over where I live a Kw-Hr costs 17 cents. So the cost of Na with this method is AU$6.7/Kg Na. This is VERY good.

Of course I neglected the cost of the heating current in the prior-to-electrolysis phase. Its about 12cents. So if you produce say 100gms Na in a run, it would contribute about an extra $1.20, or $7.90 per Kg Na.

[Edited on 14-9-2007 by len1]

Na16.JPG - 51kB

len1 - 13-9-2007 at 20:00

Now that Ive run the cell several times I have established a regime under which the cell operates perfectly every time.

If anyone repeats this PLEASE WEAR ENCLOSING EYE PROTECTORS. I never approach the cell with the NaOH molten without them. Molten NaOH is one of the best materials for destroying organic tissue. A spray bit landed on my forearm, about a 1mm sphere, and left exactly that cavity on my forearm.

If you dont follow the advice below, or your cell is constructed somewhat differently, there can be an EXPLOSION. This can eject a substantial portion of the bath straight up the Na collector.

If this hits your Cornea - you can kiss it goodbye. And a blind experimental chemist is not a very good one.

Now to the operation.

1) The temperature of the cell must be raised GRADUALLY. In all it took about 2.5 hrs from room temperature till the contents inside the collector were liquid. AT no stage should electrolysis begin until this has occured. Heater elements in radiative heaters are designed to operate at 800C plus, this does not melt the quartz tube in which they are encased, but it will melt the fibreglass. I ran the coil at 50% power. This is provided by a variable duty triac driver. Once the set temperature is reached the triac is turned off. If it is required I can post a suitable circuit. You can basically set the heating, and then go do something else for 2.5hrs.

2) The bath melts at about 350C-360C as displayed by the thermocouple. This is due to the latent heat of NaOH melting as well as an unavoidable T gradient between the thermocouple and the electrodes. Application of electrolysis current at this stage will lead to furious bubbling, and once the Na starts forming small EXPLOSIONS. This is normal, most descdriptions of Castner describe this effect. However, in a large cell such an explosion is easily contained, doesnt eject bath, or ruin the cell. For a small cell, the explosion is of the same force, but is not well contained. It will lead to ejection of some bath. You can see the result of that in the spray present on the cover of the pot in the first picture I posted. I have found that explosions can be entirely avoided, so NO spatter from the cell occurs at all. The procedure is as follows:

3) Once the NaOH has melted turn of the heater. Apply electrolysis current at about 3.3V but no more. The temperature starts dropping, and the cell electrolysis residual water in the bath. It doesnt have enough voltage to electrolyse Na. As this happens the temperature starts to drop.

4) When it reaches 320C raise the mean voltage to 4.6V (you can see the actual variation of voltage across the cell in the scope trace below), the current should approach 50A.

5) If the temperature rises above 325 start dropping the current by screwing in the core shunt to keep it below 330. An alteration of 1 to 2 amps should be sufficient.

6) If the temperature reaches 330C turn the electrolysis off, until its dropped to 325 (again to avoid explosions at higher temperatures due to the fact the the Na is more mobile in the less viscous NaOH melt).

7) If you overcompensated the other way and the temperature drops below 315 turn on the heater.

8) When starting a new run, assuming the cell is already full from the previous run, add some NaOH so bath level remains about 4cm above the cathode, using a funnel, into the NaOH colelctor pipe.

9) Initially a brand-new ladel is a nuisance. The Na wets it perfectly and half of it stays in the ladel which is a waste. To avoid it dip the S/S ladel into the hot NaOH bath for a minute or so, then take it out wash, and dry at 280C. This treatment will cover it with a durable black-brown oxide cover. The Na will not stick to this.

[Edited on 14-9-2007 by len1]

Na15.JPG - 44kB

Magpie - 14-9-2007 at 14:25

Interesting pictures len. I still haven't seen where the gauze is installed, however. Is it placed between the cathode and anode as shown in the industrial design?

Your electricity costs are interesting. If you get hard up for cash you could go into production making Na for the eBay market. There will always be kewls wanting to throw big chunks of it into water. :D

Xenoid - 15-9-2007 at 14:16

Originally posted by len1

If anyone repeats this PLEASE WEAR ENCLOSING EYE PROTECTORS. I never approach the cell with the NaOH molten without them. Molten NaOH is one of the best materials for destroying organic tissue. A spray bit landed on my forearm, about a 1mm sphere, and left exactly that cavity on my forearm.

[Edited on 14-9-2007 by len1]

I'll say amen to that!

One of the reasons, I've never got around to building a Castner cell is my experience as a teenager many, many years ago. I knocked together a rickety cell comprising a steel cooking dish anode and a graphite rod cathode. The NaOH was melted using a gas burner and the electrodes were connected to a car battery. After the NaOH had melted I was lowering the cathode into the dish when it slipped and fell in!.. :o

There was a bright orange flash and molten NaOH was distributed over much of my woodshed "laboratory". Fortunately the dish had also slipped sideways and the contents were deflected away from me. I did however get several drops on the backs of my hands and arms. The scars from which I carry to this day! I had no protective clothing or eye protection, I was very lucky.

Molten NaOH is very nasty, if you are making this cell or something similar, or carrying out manganate fusions, take appropriate precautions, as outlined by len1.

Regards, Xenoid

Eclectic - 15-9-2007 at 15:06

Do you have any plans to try running the cell with potassium hydroxide?

len1 - 16-9-2007 at 20:52

I have no plans for KOH because I really have nothing I need potassium for (and not much I need sodium for either). It is harder to make due to the higher mp of KOH, and the higher reactivity of K at these temperatures. K at 420C is not as docile as Na at 320. But you, or anyone lese is welcome to try.

Magpie, I unfortunately can not give you pics of the gauze inside the collector due to the fact that I did not take any while the cell is being made. You would appreciate that taking the cell apart and removing the caustic is a lot of work (the caustic forms crusts which no longer easily dissolve in water and so the cell needs to be emptied hot) and something I intend doing until the electrolyte is due for replacement.

After having run the cell many times I thought the triac driver circuit is not really needed. That is because it turned out the optimum duty for the element in 50% - and this can be achieved more easily with a diode half-wave rectifier in series with the winding. The diode cannot be a 1N4004, it must be a 5 amp variety at least.

That sound a really unpleasant experience Xenoid, and something I myself have been worried about. Fortunately this cell almost completely encases the caustic, except the Na collector, where the caustic is about 7cm below the surface. I normally put a cap on it, except when harvesting sodium. I have a few sodium burns - the ladel spits when you wash it after harvesting - but they are small and almost healed. PS where is Codfish Island - I grew up in NZ but dont remember a place like that.

I have run the cell many times now, and can say its very reliable, just follow the procedure I outlined. The longest I ran the cell is 7 hours on sunday, and got 63gms Na for my trouble.

Perhaps my calculation of the cost of Na are a funny thing to do, but I have seen it mentioned that it is cheaper to buy Na that make it. The calculations show its not so. The overhead for the cell is perhaps $80 overall (I had the old welder lying around anyway). I would not of course ever engage in the dangerous practise of selling Na. If someone blinds himslef or someone else, you are responsible, and not just financially but morally too. I know that some people get a joy of throwing Na into water. My kids for example - but they are young and I can understand that. Half way through explanations of the elechtrochemical series of metals they go to me 'dad can we throw some more Na into water'.

Na20.JPG - 48kB

Fleaker - 18-9-2007 at 14:45

Very impressive, it is interesting to see someone do the Castner cell! Pretty clean 'ingots' you are getting as well. I've had a shot before with the CaCl2/NaCl electrolysis, but this is much less thermally intensive. Only the molten hydroxide is a pain.

len1 - 19-9-2007 at 21:31

Thanks fleaker. I also tried NaCl/CaCl2 higher up in this thread, because I thought molten NaOH too dangerous, It made some Na but I couldnt get past the pilot run. The trouble was that the melt would expand when cooled, and that would crack the ceramic cell.

The ingots of course dont look like that straight out of the cell. They are ball shaped in the form of the ladel. I coalsce them under the parafin by pressing together. Na has a surface tension which reminds me of mercury (it also looks just like mercury when molten although of course much lighter) so any NaOH crud which stuck to the sodium gets pushed out to the surface, from there i remove it with tweezers, and the finer stuff push to the side.

This of course is a far from ideal way of purifying the Na. I have yet to think up of a better way. Certainly evaporating in a vacuum would be ideal, but that would require appartus of the same level of complexity as the cell itself regards Len

Twospoons - 19-9-2007 at 21:59

Maybe, if you got really ambitious, you could combine the Castner cell and vacuum distillation into one. No more ladling out the molten Na - just vacuum distill straight into a chilled receiver, giving you your pure Na in one step.
That would be one hell of an achievement.

garage chemist - 19-9-2007 at 22:36

I frequently purify Na and K from their crusts by putting them in oil in a test tube (bp > 100°C) and adding a few drops of isopropyl alcohol.
This causes the crust to fall off and, upon heating to the melting point of the metal, coalescence of the metal to a shiny drop.
With Na/K alloy, which I use frequently for drying Et2O and THF for use in Grignards, this is an extremely useful technique. Without it, it is difficult to get a useful amount of alloy since once the metal has split into drops, it wont coalesce again without the IPA addition.
Of course, you'll lose a slight bit of metal to alcoholate formation, but this is negligible.

len1 - 19-9-2007 at 22:42

Thanks garage chemist, that (and your link) is extremely useful. Twospoons - youre not serious?

garage chemist - 19-9-2007 at 23:45

Here are some pictures of this purification:
Na/K alloy preparation with use of IPA

The first pic is a piece of potassium with a lot of crust on it under petroleum, the second pic is after melting it.
The third and fourth pic are after addition of IPA! The potassium flows together by itself.
Then some sodium was added and the alloy transferred into an ampoule as seen in the next two pictures.
The last two pics are of the alloy being used to dry Ether, you can see how the stirring bar separates it into a fine suspension.

[Edited on 20-9-2007 by garage chemist]

12AX7 - 20-9-2007 at 01:14

Mmmm, alkali metal flux! :D

not_important - 20-9-2007 at 01:35

Originally posted by Twospoons
Maybe, if you got really ambitious, you could combine the Castner cell and vacuum distillation into one. No more ladling out the molten Na - just vacuum distill straight into a chilled receiver, giving you your pure Na in one step.
That would be one hell of an achievement.

It certainly would be, if you consider the following.

The vapour pressure of sodium at 440 C is about 1 mmHg. That's roughly 100 C higher than Len is running at. Going to take some fairly hard vacuum to get much distilling.

And then all the time the sodium is being made, O2 and H2 are also coming off the cell. To handle those gases while maintaining the low pressure needed, Len's going to need a bank of turbopumps that'll overheat the substation he's connected to.

And while that sodium is boiling off, it's mixed with those gases. Even if you use a twin vacuum system to keep the O2 and H2/Na streams separate, the sodium and hydrogen may react.

Twospoons - 20-9-2007 at 15:42

It was only a wild "blue-sky" idea. You could use an additional boiler element where the sodium collects to push the temp to 600 C - then the vapour pressure rises to around 10kPa. Go to 880 and its boils at atmospheric.

So its not impossible - just not easy.

garage chemist - 20-9-2007 at 16:57

At those temperatures the sodium immediately reacts with the NaOH, forming Na2O. That distillation approach is unrealistic for a NaOH cell.

The distillation method would be more appropriate for a Downs cell that uses pure NaCl as the electrolyte! You could have the temperature high enough that no vacuum at all would be necessary.

jimmyboy - 24-9-2007 at 19:55

I still don't quite understand your cell -- maybe a schematic/drawing would help? -- it sounds like you are keeping the sodium metal separated from the hydroxide melt with the mesh and fine temperature control -- but I'm not sure.. (has anyone else made one yet?) very impressive sodium ingot by the way..

len1 - 25-9-2007 at 22:34

Drawing the cell takes quite some time. Ive already spent a lot of time making it and taking pictures. The pictures are quite detailed I believe. If after that you still have problems ask me specific questions, and Ill talk you thru it. Most people are not serious so I want to see that some effort has been made by someone before spending my time.

To understand where the S/S gauze goes look at the picture marked anode and Na Collector. The Na collector is the internal pipe with slots. A cylindrical S/S mesh (bent to shape of course, fits snugly inside that and separates the inside of the collector from the anode. I have now taken the cell apart and can say the gauze has hardly corroded with the many runs. There are no issues with cell construction I could find.

Magpie - 25-9-2007 at 23:10

Len1 says:


A cylindrical S/S mesh (bent to shape of course, fits snugly inside that and separates the inside of the collector from the anode.

Did you mean cathode (instead of anode)? Does it keep the sodium from migrating through the slots?

len1 - 26-9-2007 at 21:01

Originally posted by Magpie
Len1 says:


A cylindrical S/S mesh (bent to shape of course, fits snugly inside that and separates the inside of the collector from the anode.

Did you mean cathode (instead of anode)? Does it keep the sodium from migrating through the slots?

Well I meant anode, but you could obviously also use cathode in that sentence and be equally correct. The slotted collector separates the spiral copper cathode inside the collector from the cylindrical nickel anode outside. The S/S gauze fits inside the collector and stops the Na migrating thru the slots as you say.

len1 - 1-10-2007 at 19:14

I am posting this exerpt from a U2U in case it might be of help to others.

Im indeed in Australia, and will be pleased to help you if you are going to make an Na cell.

You ask a very pertinent question about nickel. We are one of the biggest producers but on of the smallest consumers of nickel, and finding it was a heck of a job. However, the Nickel Institute of Australia, I think its called, or a very similar name, does have some off-cuts. Its not cheap though, $100+.

A microwave transformer is no good. Thats because its step-up, not step down. It gives 1000+ secondary volts at low current. You need to get a current-adjust welder. There is a great deal on precisely this in thrifty-link stores at the moment, $95 for an adjustable 140A welder. Rip out the transformer and the fan, and use the leads as the high current leads.

The current adjusting screw, is essential, voltage control at high currents is very difficult (unless done on the primary). There is a screw on the transformer which adjusts the position of some shunt metal in the core, to reduce or increase the amount of magnetic flux from the primary passing thru the secondary. I have found that without such adjustment the procedure for getting sodium just doesnt work well. The cell easily overheats with all the attendant consequences.

The holes in the collector are not critical, make them as large as you can cosnistent with structural integrity (and of course containment of the Na at the top of the pipe). I made mine by milling 4 uniformly spaced straight channels 10mm wide by 60mm long. The collector is not floating, and is electrically connected to the cell body at the top, at the bottom theres about a 10mm gap from the bottom of the cell, which is covered with gauze, to allow good circulation.

Anode material.

Xenoid - 2-10-2007 at 00:12

Hi len1,

I was going to ask you about the nickel myself, as to where in Australasia you obtained it. You note it is by far the best material for an anode, do you have any feeling for the difference in corrosion rates between nickel and SS in this situation. If thicker (say 2mm) SS was used for the anode, how long would it last, 1 hour, 10 hours etc. How is your 1mm nickel holding up, how many hours has it done!

Regards, Xenoid

JohnWW - 2-10-2007 at 00:30

Nickel? I have a hoard of the old cupronickel coins that were removed from circulation in New Zealand last year, which were identical in size to the ones still used in Australia (and also Fiji and Samoa). They were replaced by much smaller coins, made of a cheaper alloy with nickel plating or cladding. I am holding on to them mostly for their Cu and Ni content, or as alternatives to using stainless steel washers where required.

garage chemist - 2-10-2007 at 00:33

Could one use a nickel plated SS or maybe Cu anode? Nickel salts can be made from readily available nickel carbonate (online pottery supplier) and used to plate nickel onto another metal.

12AX7 - 2-10-2007 at 00:38

Canadian nickels used to be pure nickel. And I think their cents, too. Might be worth importing some. Do beware, they switched their coin metal like every five or ten years... the Canadian Mint's listing boggles the mind...

Edit: Don't forget your old friend, the welding store. Besides graphite, they should also have nickel rods, 55 and 95% or so IIRC, used for welding cast iron. Use as-is or pound flat (with heat if necessary).


[Edited on 10-2-2007 by 12AX7]

len1 - 2-10-2007 at 20:42

I have indeed made some investigation of corrosion rates of different anodes in NaOH. The anode in this cell is the point most subject to attack because of 1) the positive anode voltage creates oxidizing conditions which tend to dissolve the anode material in the bath 2) the O2 and H2O evolved at the anode tend to oxidize the anode material. I am more concerned in the contamination and shortening of bath life by the presence of anode cations in the bath, rather than the dissolution of the anode per-se, although, using nickel in the anode has allowed me to spot weld the anode material, something I would not have been able to do if it was subject to rapid dissolution.

I have taken the cell apart now after about 30hrs operation and 200gms of Na, and the nickel anode shows no signs of corrosion whatsover. Zilch. I think it can be regarded as a permanent item of the cell. The copper cathode shows more corrosion, but not really significat. Copper can not be used in the anode, it dissolves and contaminates the bath rapidly. S/S I have used in mock-ups before. High Ni SS (which can be detected by the fact that they are non-magnetic) I believe these types are called austenitic, seem to stand up us anodes the best. Their dissolution rate just as a very rough estimate is less than about 100microns/hr at the current densities of this cell. The best material, without any doubt however, is the original Ni chosen by Castner. Another source I havent mentioned is ebay. Now and then Ni appears - though sometimes wat is claimed as nickel is actual Ni SS. As garage chemist says you can plate Ni using a sacrificial anode and NiCl2 or NiSO4 bath. I thought of that, but youd want a thick layer, and rather than hassle with that getting the sheet seemed a better deal. Len

Tacho - 3-10-2007 at 03:37

Congratulations len1, your cell is a fantastic job.

About nickel anodes: A good electroless plating on a ceramic substrate should be a cheap option. I have obtained good electroless nickel plating in the past fairly easily, but hypophosphites are hard to find and even controlled in some countries.

I remeber obtaining a good, smooth, shiny nickel plating on carbon using nickel chloride and ascorbic acid as electrolyte. Ascorbic acid (vitamin C) is the key here. Low voltages are required.

len1 - 3-10-2007 at 22:21

Thanks for your kind words Tacho. I have not heard of electroless plating onto ceramics using hypophosphites. Can you explain how it works or give me a reference? Thanks Len

Tacho - 4-10-2007 at 04:08

Here is a specific thread with links:

Here are some related threads:

Nickel ions get reduced by hypophosphite to produce a solid layer on non-conductive substrates, similar to that produced by silver in the mirror making process.

I have obtained solid shiny (not silver shiny, more like dull aluminum shiny) platings on my very first crude attempts, so it's not not a very difficult procedure.

I gather that hypophosphites are controlled in many countries because it has uses in drug making.

len1 - 9-10-2007 at 18:25

Thanks Tacho for the info.

Further on how to coalesce/purify the Na ingots obtained from the cell (from oxide and bath crusts). I have tried the isopropyl alcohol method sugegsted by garage chemist and unfortunately it doesnt work with the large-size ingots obtained from the cell. Its addition to the Na in parafin causes vigorous bubbling but no coalescence or freeing from the bath crust, it also introduces a light-brown precipitate/impurity which floats on the Na globule (can be seen in the picture). Perhaps it operates better with small amounts in glass capiliaries.

I have found the best method for coalesceing the globules is as follows. Once the Na has melted below the parafin surface the temperature needs to be brought up to 140C+ to decrease the surface tension of the Na. The small globules can be sucked up by a 5ml pipete on the picture, and injected into the larger globules. It is best if some parafin is sucked up first to reduce oxidation of the exposed Na surface at the top of the pipete. The action needs to be performed quickly to avoid the Na solodifying in the pipete. Once one big globule is obtained, large bits of crust, which due to the surface tension are ejected from the Na to the surface can be picked off with tweezers. When all the large bits are gone a fork can be used to trawl thru the molten ingot several times, this will collect all the oxide and small crusts to the side where it can be easily picked off. The end result is an effectively pure Na ingot as the accompanying pictures show. Purification by distillation is considered inappropriate for an amatuer set-up since the apparatus needs be completely evacuated.

Na21.bmp - 938kB

len1 - 9-10-2007 at 18:29

The solidified purified ingot

Na22.bmp - 529kB

kilowatt - 15-10-2007 at 05:26

Very nice work len1! As elusive as alkali metal production has been for all the rest of us amateurs it is hard to believe that potassium was first isolated around 200 years ago with very limited material selections. It's quite inspiring to see someone else has succeeded in this endeavor.

I have always thought the Castner Cell to be too touchy for my taste, requiring a very narrow operating temperature range. Not to mention you can get a 40lb bag of NaCl for about the same price as a 1lb jar of NaOH, and making pure NaOH electrolytically is relatively involved requiring either fractional crystallization from a membrane cell, or an amalgam cell which has relatively low production rate and requires a good amount of mercury.

I have been working for some years now on a 500A cell similar to a downs cell which liquefies the chlorine off the anode by means of a cascade cooler which chills the condenser coils to -50C and runs it into a tank. Meanwhile the sodium would run into an argon filled hopper. It will run eutetic CaCl2 and NaCl and has a riser pipe just like an industrial downs cell. After the process, the chlorine would be brought up to vapor pressure at room temperature and fed into a reactor to produce anhydrous ferric chloride. The project is still in a highly incomplete stage and I am still waiting on a special refractory piece made from cut pieces of ceramic tiles with low-alkali borosilicate glaze. I have at least built the steel part of the cell and the central graphite anode. I have also partially built the chlorine liquefier system and a set of casting trays for the sodium metal. Production should be about 1lb/hour at 500A. I still hope to finish the thing sometime, but it's a major project and it doesn't look like it's coming as soon as I had hoped. There are several safety measures that need put in place before this cell can be operated, too. It will contain something like 60MJ of thermal energy plus a large amount of liquid sodium and hot chlorine when operating. In the mean time I have made a few other (all failed) attempts at alkali metals.

Lithium should be much easier to make than sodium as lithium chloride melts at 605C while a eutetic mixture of that and KCl melts at something like 400C if I remember right. I once tried to make lithium metal in a small nickel plated crucible as a cathode, with a central graphite anode and a glass tube with a stainless steel screen around it dipping into the bath as a divider. I had a continuous flow of argon over the cathode area of the cell to shield the produced metal. It was all held in a single piece of caved firebrick. The soda-lime glass of the divider tube ended up being relatively conductive in the bath at that temperature, which I didn't know at the time, and it destroyed it. The stainless steel screen was completely dissolved, too. No metal could be collected, even though early in the electrolysis I could see some shiny metallic blobs form briefly. Once the divider dissolved away there was no trace of metal produced and it all vanished.

Later I tried to make a very small tubular sort of cell out of a solid piece of graphite for electrolyzing eutetic NaCl/CaCl2. It had 3 holes drilled into the block from the top, and one hole bridging them all inside which was plugged at the end with a bolt. It was operated manually with a graphite anode, while the cathode was a little ceramic feedthrough with a nickel plated brass rod coming though. The cathode chamber was liquid tight, and a tube ran out the side and into an oil filled jar. I made a rather large error in designing the thing and any sodium produced would have shorted the cathode to the cell body. It ended up that the cell body acted more like a cathode than the cathode did as a result, and any sodium developed at the anode compartment immediately combusted in the air and chlorine there. When I opened the cathode chamber, there was some sodium/calcium carbide formation (evidenced by the release of acetylene when I was cleaning it out), and brass cathode and the nuts on it had thoroughly alloyed with sodium/calcium and it was spongy and all stuck together, despite the nickel plate.

Here is my latest endeavor, with which I hope to electrolyze either eutetic NaCl/CaCl2 or LiCl. It looks almost halfways promising so far. I was almost able to finish the thing this weekend but I didn't quite have time and I didn't have a good way to vent out or otherwise dispose of the chlorine gas. Now it will probably be at least a few weeks before I can get home and work on it more.

The cell body is made from one of those 16oz propane canisters cut roughly in half. It is relatively thick steel. That will be suspended by bolts running through the firebrick on either side of the setup. I still need to braze a wire across the thing in the other direction to hang the stainless steel screen that will divide the cell in half. Both electrodes are held in a graphite block by a set screw, and surrounded by a tube made of fused quartz.

On the left is the graphite anode. A small hole in the top of the graphite block is to be connected to the chlorine vent tube. The fused quartz tube around the anode is held in by a set screw, which will no doubt need tightened as the cell warms up, and then loosened again before it can be allowed to cool down. There isn't anything more to the anode assembly really.

On the right is the cathode assembly, a little more complex. The cathode itself is a hollow stainless steel tube with a couple holes drilled in it above the cell liquid level so sodium can go inside it. The fused quartz tube around the cathode is sealed into the graphite block with boric oxide. It seemed to have no real problems with contraction when this assembly was cooled from something like 500C where I cast the boric oxide in down to room temperature. Nothing cracked, anyhow. The top of the tube and the graphite block will operate at a much lower temperature than the cell, and much lower than the melting point of boric oxide, because stainless steel is a poor thermal conductor and most of the outer tube will be filled with argon gas. Since the top of the assembly should be air tight, the sodium should not be able to rise higher than the hole where it enters the cathode itself. Obviously if the sodium did rise up into the top of the assembly it would reduce the boric oxide and screw everything up. The cathode has an outlet there above the graphite block, where it empties into a heated oil filled jar, which is in turn connected to a liquid filled tube with a valve at the bottom. That tube will be used to provide suction on the cathode tube until the piping is all primed with sodium, so it can run out into the collection jar under gravity. Obviously this will take some care since if any salt is sucked up into the tubing it will freeze and clog the thing. The cell will be rather well surrounded by refractory ceramic and firebrick to get it up to 600C. It is fired with propylene gas and the electrolysis runs on a large battery charger. Due to the rather small, distant, and otherwise inefficient electrode placing, I expect the full 12V of the battery charger will be dropped across the cell, and even then it will probably need additional heat from the torch. So far my major concerns are how well the thing will hold together under more temperature cycling, and how quickly liquid sodium might dissolve the brazed joints used here and there in the cathode assembly.

Edit: I think I will run sodium nitrate in this cell. Melting at only 307C, it should yield sodium at the cathode and NO2 + O2 at the anode according to our favorite online sodium texts thanks to BromicAcid. Those anodic gasses will then be fed through a low pressure bubbler with water to form nitric acid.

[Edited on 18-10-2007 by kilowatt]

len1 - 28-10-2007 at 18:46

Working with NaCl is much more tricky I have found. For a start Na at that temperature attacks glass (see my posts above), also I have found that iron rusts really quickly at these temperatures and with Cl present. Cost wise, the price of NaOH is about 3 times that of NaCl here, but when you include CaCl2, the costs are the same per kg of bath.

I dont think NaNO3 is a good candidate, this is nitre saltpetre. It starts decomposing as soon as it melts, to NaNO2 and Na20, which absorbs H2O and CO2 from the atmosphere. The NO2 is highly acidic and will attack the anode, and percolate to the cathode to unite with the sodium. I think some Na can be obtained this way, but very little and its not worth the hassle. My advise is you attack the problem in the easiest possible way, and then build in complications. regards Len

kilowatt - 28-10-2007 at 19:51


Working with NaCl is much more tricky I have found. For a start Na at that temperature attacks glass (see my posts above)

Yes, molten chlorides attack ordinary glass as well as borosilicate to some degree at those temperatures, and they are conductive enough to draw the ions through, making it a useless barrier. However, the molten salts will not attack fused quartz which is pure silica and contains no free ions (like Na+ and Ca2+ found in ordinary glass). I have used fused quartz tubes for this latest cell.

I have decided to use eutectic LiCl and KCl for the first run of this mini cell, as lithium production was its original purpose anyway. I will probably need to heat the pickup tube from the outside for lithium to remain molten, where I would not have had to with sodium. The collection jar will simply be filled with argon since lithium floats on any would-be barrier liquid anyway, and an oil layer will be used in the suction tube. This cell should also work with eutectic NaCl/CaCl2, but there could be difficulties with calcium clogging the riser pipe. The alkali metal nitrates would probably prove too problematic for this design, plus NaNO3 requires something like a 15V cell drop. The Darling Cell, which actually does electrolyze NaNO3 to get sodium metal and NO2/O2, is a nice design, but what I have here is probably not too suitable for one.

nitroglycol - 18-11-2007 at 16:01

Originally posted by 12AX7
Canadian nickels used to be pure nickel. And I think their cents, too. Might be worth importing some. Do beware, they switched their coin metal like every five or ten years... the Canadian Mint's listing boggles the mind...

There's a link here that goes into detail about Canadian nickels and pennies. The nickel was 99.9% nickel from 1946 to 1951 and again from 1955 to 1981. The quarter and dime aren't covered in this article, but IIRC they were pure nickel from the early 1960s until 2000; those made after 2000 are nickel-plated steel. The outer ring of the toonie is also virtually pure nickel, and the loonie is bronze-plated nickel.

Canadian readers should take note that destruction of Canadian coins is technically illegal here.

LSD25 - 6-1-2008 at 15:07

Could I plate lithium metal out of the LiI:KI eutectic (MP around 260C)? If so, what would be the best material for the electrode? Could one use an all glass setup if one was doing this? Also, is it possible to use a glass coated electrode - ie. could I expect sufficient electrical current to go through the glass (such electrodes are used for some purposes - just unaware if they could be used in this endeavour - the idea of course being that glass is one of the very few materials which cope well with iodine/iodide/iodate/hydrogen iodide/etc.

Alternatively, if one used a sacrificial Al anode, would one not get AlI3 out? If that was so, it may be worthwhile considering the use of Hydrogen gas as the blanket for the Li metal - which upon reaction with the AlI3 should give LiAlH4 should it not? This would be a nice target to work toward, the LAH would be useful (to say the least) while the side product of LiI would be fed back into the modified Downs cell.

This is merely the start of an idea, it will need a lot of research, development and consideration.


Of course, on consideration, glass would be a poor choice of material for lithium (it apparently attacks it).

Len, where does the Na collect in the NaOH cell (top or bottom)?

[Edited on 6-1-2008 by LSD25]

kilowatt - 7-1-2008 at 11:46

The bath will definitely attack normal glass; I have tried using glass in such a cell before. Lithium is somewhat soluble in the bath and will alloy too easily with most electrode metals except iron and the like. The bath is also well above the melting point of lithium. Obtaining a decent lithium plating this way would be difficult or impossible; I'm not even sure if it will wet to an iron electrode.

I would suggest just dipping a brass piece into molten lithium metal if you want a lithium plating, or electroplate it out of an exotic cold electrolyte solution like discussed in the unconventional sodium thread if you want to obtain it as an electrowon solid.


franklyn - 9-1-2008 at 03:42

The direct combination of 2 mols CO2 with 2 mols of sodium metal at 360 ºC
forms Na2C2O4 . Sodium oxalate itself will melt with decomposition above 250 ºC.
It seems a promising prospect to subject a melt of sodium oxalate to electrolysis
driving off carbon dioxide to recover the pure metal.


12AX7 - 9-1-2008 at 05:32

Won't that disproportionate to carbonate and CO?

franklyn - 16-1-2008 at 22:22

That's my understanding, but it's not instantaneous. Maintaining the temperature well
below 360 ºC where Oxalate forms from Sodium and CO2, under pressure by metering
the excape of CO2 with a spring loaded pintle, will inhibit the normal decomposition at
atmospheric pressure into carbonate and monoxide at it's more moderate melting point
of around 260 ºC. These reactions are temperature driven , the applied electric field
should bias decomposition into CO2. Think of it as CO2 solvation. Sodium melts at just
98 ºC and floats on the oxalate of density 2.34, a hollow cathode insulated on the outside
will collect the sodium on the interior shielding it from the gas products that form in
the reaction container. That's why it's called Technochemistry.


The prospect of an oxalate anion decomposing into 2 mols CO2 seems a reach.
It looks more likely that some part of the melt will convert into Na2CO3 as a final
product. NaCO3 melts at 858 ºC so in this low temperature scheme it could only
be solvated if that, otherwise with a 2.54 density, being greater than the oxalate
it would precipitate out. The decomposition into Na2O and CO2 could not occur.

2 mols of NaHCO3 decomposes at above 50 ºC into Na2CO3 + CO2 + H2O vapor
but here the anhydrous bicarbonate anion cannot produce water but only CO2.

Electrolytic cell bias
2 NaC2O4 -> 2 Na(+) + 2 NaC2O4(-)

Cathode reaction
2 Na(+) + 2e- -> 2 Na

Anode reaction(s)

2 NaC2O4(-) -> 2 NaCO3(-) + 2 CO

2 NaCO3(-) -> 2 NaO(-) + 2 CO2

2 NaO(-) + CO -> Na2CO3 + 2e-

Final anode products
Na2CO3 + CO + 2 CO2

From this we see the occurence of two reactions,
1. Decomposition of half of the melt into Na2CO3 and CO.
2. Decomposition of the other half of the melt by circuitous reactions
into 2 Na + 2 CO2

[Edited on 14-2-2008 by franklyn]

LSD25 - 11-2-2008 at 04:17


Why does the use of the vessel as the anode preclude using nickel as the anode? Wouldn't it be better just to plate the vessel with nickel? IIRC Nickel plating solutions, kits, etc. are still available online in Oz....

Assuming that I may or may not have just purchased a cheap, portable Arc Welder rated at a max of 150A, how exactly should I modify this (or can I simply attach it via jumper leads to an external voltage reducer/regulator) to give no more than 5V? Quite simply, I'd like to set the welder up as a high amperage PSU for arc-furnaces, electrolysis, etc. Is this workable, or is it a pipedream which should be discarded?

PS I am at home with turning, cutting, milling and welding (inc. threads, etc.) - but electricity is well and truly outside my comfort zone. Please dumb it down if you do answer this.

kilowatt - 13-2-2008 at 09:29

You really want more than 5V for a Downs cell; about 8 is good. Not sure about a Castner cell. I would see how much voltage the welder puts out under the load of such a cell, by building the cell first and then measuring the voltage drop across it, but you really can't lower this without physical access to lower voltage taps on the transformer. The welder is more or less a constant current source, so the lower the resistance you put on it the lower the output voltage will be. It can't hurt the cell to drop more voltage than you might like either; it will just generate more heat and may need external cooling. If it's really too high then your best bet would be to put more cells in series. Industrial downs cell banks run on 240V DC and have about 16 cells in series. A 150A cell is gonna be pretty small; the industrial ones run about 25kA.

I think I'm going to try my little cell with NaCl/KCl eutectic next time I get home. This way I can still run at a fairly low temp (not as low) but won't have to deal with crystallizing calcium, and I know potassium will reduce sodium out of the bath.

[Edited on 13-2-2008 by kilowatt]

DJF90 - 13-2-2008 at 12:50

How about electrolysis of molten sodium thiosulphate? It melts at about 50C and decomposes at its boiling point, which I cannot find. The sodium metal wouldnt be molten at this temperature and so I'm not really sure what would happen. I just thought I would mention it incase it had been overlooked. If sodium metal can be produced this way then it wouldnt be the most economical method of production but it isn't too bad, thiosulphate is reasonably priced and electricity bills would be cut compared to the other electrolysis methods?

kilowatt - 13-2-2008 at 16:22

Like any complex ion electrolysis (for example the Darling electrolysis cell which uses sodium nitrate) it will likely go through a process where it is first reduced to thiosulfite, then to sulfide, and a gradient of concentrations of these will be established in the cell. Sodium sulfide melts at 950°C and may freeze up around the cathode. It depends on what the gradient is like.

len1 - 13-2-2008 at 19:18

Originally posted by LSD25

Why does the use of the vessel as the anode preclude using nickel as the anode? Wouldn't it be better just to plate the vessel with nickel? IIRC Nickel plating solutions, kits, etc. are still available online in Oz....

Assuming that I may or may not have just purchased a cheap, portable Arc Welder rated at a max of 150A, how exactly should I modify this (or can I simply attach it via jumper leads to an external voltage reducer/regulator) to give no more than 5V? Quite simply, I'd like to set the welder up as a high amperage PSU for arc-furnaces, electrolysis, etc. Is this workable, or is it a pipedream which should be discarded?

PS I am at home with turning, cutting, milling and welding (inc. threads, etc.) - but electricity is well and truly outside my comfort zone. Please dumb it down if you do answer this.

Well having the cell walls acting as anode does not literally preclude the anode being nickel at the same time - only if money is an object. If you check the price for nickel you will see that proposition very expensive. I could only get nickel in sheet form (several hunderd dollars worth) which I utilised as in the cell. To get it welded to act as the container as well is possible but would be more expensive. In addition the anode is a consumable part, so having the cell walls consumed is not a long-term solution.

Plating is a long-winded option, you want a thick good-quality (non-porous) nickel plate and thats difficult and long-winded to achieve.

You can use a welder transformer for the cell provided its of the adjustable current variety as I described. You measure the AC voltage on the secondary open circuit, and establish how many turns the secondary carries. That gives you how many turns per volt. Next you want at least two taps - 6V AC and 8V AC open circuit. So you know how many turns you need to wind down.

Alternatively you can modify the 5V output of a PC power supply as described in Silicon Chip. Its simple - but you have to know more electronics than with a welder. Most PC PS's can supply 60A @5V which is suited to the castner I decsribe.

The NaCl cell I would not recommend - its harder to get going than the Castner - and its not going to be nearly as long-term judging by my experiences with the oxidation of materials at 700C Len

LSD25 - 22-2-2008 at 18:26

Right, thanks Len that is great...

I bought one of the CIG welders on special, but I do have an old PC handy as well (a spare portable welder would be a useful thing to keep if possible). You wouldn't happen to have an issue number/page for the Silicon Chip article you referred to would you, that would be very helpful (although I am all at sea with electronic theory, I can certainly follow the step-by-steps they provide)?

The only reason I suggested plating is because I have found it fucking near impossible to find/purchase nickel sheet or pipe/tube. This is absurd in fact, especially given that I used to live next to and work at a bloody nickel refinery.

Is any other material going to be comparable (even if greatly lesser so as long as it would not fuck the product or the electrolyte too quickly)?

len1 - 22-2-2008 at 19:17

The SC issue in question in July 2004 I believe. That article is only useful to show how an AT PS works (it actually shows how to step up the voltage).

You need at 350W PS which is available for the newer PCs. The older tend to be 250W, which is not enough power, given you need 49Amps at 4.6V and accounting for diode drop and converion efficiency. If you post high resolution pics of both sides of a 350W PS board in most likelyhood I can tell you what mods you need to make.

Instead of Ni you can use SS sheet as second best. Its best if this is the nonmagnetic (high Ni) variety (dull in surface finish). Len

watson.fawkes - 16-8-2008 at 12:21

I've quite enjoyed reading through all the collected material here. I'll add a few notes.

Welders (as a rule) come in CV (constant voltage) and CC (constant current). CV is used for stick and MIG (GMAW); CC is used for TIG (GTAW). CC is better for controlling electrolysis generally, since it will automatically adjust the voltage to maintain steady current, that is, constant molar flow of electrons.

The boiling points of ammonia and chlorine are almost identical. Since refrigerators generally operate somewhat below the boiling point of the refrigerant, a liquid ammonia loop should readily condense the gaseous chlorine coming off a NaCl cell.

Nickel sheet can be readily spot welded. A homemade spot welder can be built from an old microwave-oven transformer. Here's one such <a href="">home-built spot welder</a>.

12AX7 - 16-8-2008 at 17:22

Close, stick (SMAW) is also CC. TIG machines generally have finer control and more freatures than a stick machine; both come in AC or DC, and obviously we want DC for this purpose (if not available, AC can be rectified with cheap components).


watson.fawkes - 17-8-2008 at 09:10

Originally posted by 12AX7
Close, stick (SMAW) is also CC.

Yes, that's right; my brain fart.

I've got an old (== non-inverter, 2' cube of transformer) CC/CV welding supply, but I'm hesitant to use it for prototyping since the smallest graticule on the ammeter is 10A.

Picric-A - 19-9-2008 at 07:52

I have thought of a easy and quick way to make sodium via electrolosis of NaOH.
I have yet to test this method as i am currently away from my lab but i feel the theary works.
Basically it is electrolosis but the electrode where the sodium is produced is a hollow graphite tube. The Na is produced and collected in the tube and is protected form oxidation. Some Na may form on the outside of the tube but it can simply be scraped off.
At the end of elecrolosis the electrodes are removed and immedetly plunged into mineral oil and the electrodes are heated to remove the molten sodium. Alterativly the electrodes can be broken up to remove the tube of Na inside.

[Edited on 19-9-2008 by Picric-A]

not_important - 19-9-2008 at 17:12

Your image is about twice as large as it should be, please resize it.

A problem with your concept is that most of the sodium will form on the outside of the tube(s). The current path is short, less voltage drop, there's sort of a Faraday cage effect going on. You'd have to encapsulate your tubes in insulators to get around that.

Picric-A - 28-9-2008 at 06:41

Sealing them in glass would work but i agree it would be better to just find a way to tap it off...
Sorry bad idea :P

len1 - 21-4-2009 at 21:32

Quote: Originally posted by Organikum  
I was irritated all time because of Marvins remark that the molten sodium would give a shortening of the electric current between the "bell" and the central electrode and make the design unworkable.

But I had done it and it worked.
How could this be?

Solution: The iron bell works is cooled by the air. At the edges inserted in the molten NaOH the NaOH solidifies and forms such an isolator on the bell. This works because the temperatur of the NaOH is only slightly above the melting temperature of NaOH and the superior heattransfer properties of iron compared to NaOH.

Try it!
Just put something from iron into just molten NaOH and you will see the solidified NaOH forming a stable protection layer on the iron. And this layer wont dissolve also not after times.

Nothing shortens here.

Hope this answers your question Marvin?

This reasoning doesnt work, because Na apart from being an excellent electrical conductor is also a great thermal conductor (used as medium in nuclear power stations). Any sodium reaching from the electrode to the bell, and touching the frozen NaOH insulator coat, will rapidly melt it. That it must do because one end of the Na pool is in contact with the electrode where the temperature is above melting by definition, and theres your electrical contact. If this overhead electrode method worked for the author, his apparatus did not produce sufficient Na to reach the bell.

The problem is to produce continuous production, not a demonstration amount, which has been done by Davy in 1808. This has been well explained by Castner, but by no one else. (My lesson in life is that people, and I mean professionals, often dont understand the essence of what they are talking about if they have no first hand experience). The key claim he is making is in having discovered

1) That Na can be produced only over a narrow temperature range by Davys method

2) An inverted electrode must be used

3) The method works for Na and K

Note however an interesting thing: the patent claims both Na and K, however the former is described in great detail - in particular the temperature range 300-320 degrees is mentioned, while no such detail is given for K at all - when this happens you immediately suspect the author is not terribly familiar with that particular process. I greatly suspect the following

a) Castner never produced any substantial amounts of K with his method - he only claimed it as cover lest someone else should succeed. Even great inventors it seems are prone to falsification

Also note the H2O vent is on the non-active side of the Ni anode. Either the anode is not circular, or the drawing is wrong - (the patent maintains its a drawing of an actual apparatus)

Broken Gears - 17-5-2009 at 11:55

I have been working on a Castner cell for some time now. I´ve welded a bottom on a stainless tube (ø130mm) with a hole in the bottom for the cathode.... well you know the setup.
But I cant find Ni anywhere to make a good cathode and I remembered some of you guys talking about, that copper wasn't a good choice for elektrodes. Since Im not much of a chemist but more a machinist :(, I dont really understand the basic chemisrty of even this simple setup.
Can I just use Iron for both elektrodes? My 2nd choice would be copper or graphite, but I dont have much available.

len1 - 17-5-2009 at 15:12

Yes the Castner cell is all about machining - I remember spending several weeks in the workshop.

You really need Ni for the anode, copper is no good. Iron just passes the test as a very distant second choice. The cell does will not produce sodium long with iron, because the erosion will rapidly foul the bath.

Europe should be OK for Ni though. Id suggest looking for metal stockists on the net.

You dont have to be much of a chemist to understand Casners cell though - sodium hydroxide melts to give positive sodium and negative hydroxide ions. The positive ions are attracted to the negative copper electrode where the gain a negative charge and become sodium metal.

[Edited on 17-5-2009 by len1]

Broken Gears - 18-5-2009 at 04:48

Okay so the center elektrode (cathode) sould be made of copper and the rind (anode) of Ni? So the copper elektrode will be "producing" the Na.
Can Titainium be used as either elektroede?

Hydragyrum - 18-5-2009 at 05:39

Quote: Originally posted by len1  
... The method works for Na and K

Note however an interesting thing: the patent claims both Na and K, however the former is described in great detail - in particular the temperature range 300-320 degrees is mentioned, while no such detail is given for K at all - when this happens you immediately suspect the author is not terribly familiar with that particular process. I greatly suspect the following

a) Castner never produced any substantial amounts of K with his method - he only claimed it as cover lest someone else should succeed. Even great inventors it seems are prone to falsification...

Well, I disagree that this means that he is falsifying results: a patent is not the same as a scientific paper - had he made unsubstantiated claims in a published paper then one could accuse him of falsification. However, the purpose of a patent is quite different, and frankly I'm surprised that he did not claim his method for all of the alkali metals (unless certain examples had been shown not to work prior to his publication? - I haven't read it firsthand so ... :P ). A scientific paper is written by scientists for scientists; a patent can be considered as being written by lawyers for lawyers.

A patent is to claim an idea backed up by actually doing it - but he only needs to demonstrate it in one example - in this case for M = Na - and can still claim the case for M = K as his idea too.

One thing I've noticed about a lot of patents is that they are not always so very easy to reproduce; I suspect that they write the bare minimum with the deliberate intention of being obscure enough to keep the "competition" guessing - just my own personal opinion, nothing more. Procedures given in papers seem much more reproducible.

vulture - 18-5-2009 at 09:58


You need at 350W PS which is available for the newer PCs.

PC PSUs are available in ranges to atleast 1kW nowadays and perhaps even more. They tend to be somewhat expensive though and won't start up unless they have proper ATX connection with a motherboard. So that would probably require some hotwiring. The upside is that for single rail PSUs you can get 80 amps at 12V.

len1 - 18-5-2009 at 21:30

Patent examiners are bound not to approve patents which they know, or have reasonable grounds to suspect, are false, misleading, or do not work.

Scientific publications also contain falsehoods - more so nowadays - in then past a scientists word indeed carried much weight than that of a patent owner. Nowadays from my experience many papers published contain falsehoods. It is indeed the policy of many journals not to correct them. The falsehoods are generally not at the level of out-and-out lies. More a case of youve done 10 experiments, 9 times didnt get anything publishable, tenth you did. But cant reproduce it. So you publish it as a typical result.

I have not seen 1kW PSU on the market. At 12V they would be useless for electrolysis - at 5V they would be OK. But remember that many switch supplies listing a 50A current mean a peak current beyond which protection circuits operate. The rectified supply I used provides a peak current about pi/2 greater than the average value quoted.

len1 - 24-5-2009 at 06:27


The liquid Na wets almost anything including glass, which it attacks also, hence the black coating you can see. The coating - whose composition I dont know, does not seem to conduct electricity, and once formed protects the glass, which was just superficially corroded. I would be interested to know if you or anyone else has any references to this effect of liquid Na on glass. Using a metal tube to collect the Na (which you could also do in NaOH) would overcome this, but it cant be used around the negative electrode.

Placing the -ve elctrode in the center of the tube has the Na wetting the tube, crawling around it, and being electrically connected to the -ve, starts formation of globules outside the tube.

I have now realised what the black stuff is - its Si. At 600C Na reduces SiO2 in glass

Formula409 - 11-7-2009 at 00:49

Quote: Originally posted by vulture  

You need at 350W PS which is available for the newer PCs.

PC PSUs are available in ranges to atleast 1kW nowadays and perhaps even more. They tend to be somewhat expensive though and won't start up unless they have proper ATX connection with a motherboard. So that would probably require some hotwiring. The upside is that for single rail PSUs you can get 80 amps at 12V.

In order to start your typical compuer PSU, it's just a simple matter of grounding the green wire :)

Good link here:
Pinout of ATX connector:

I've been interested in making some Sodium for a very long time, however in Australia, I believe it is on the highest regulation list whereby possession of a small amount such as 2g results in a massive sentence. Len can you comment on this?


[Edited on 11-7-2009 by Formula409]

len1 - 11-7-2009 at 05:14

I think its deportation ..

len1 - 5-8-2009 at 00:59

This is what I wrote a few months ago.

Note however an interesting thing: Casnter's patent claims both Na and K, however the former is described in great detail - in particular the temperature range 300-320 degrees is mentioned, while no such detail is given for K at all - when this happens you immediately suspect the author is not terribly familiar with that particular process. I greatly suspect the following - Castner never produced any substantial amounts of K with his method

Now I found my suspicions shared by German researchers at the beginning of this century who attempted to 'repeat' Casners finding, stating bluntly that he did not do what he said:


Castner hat seine Versuche offenbar nur mit Natron-, jedenfalls nicht mit Kalischmelzen angestellt; wir halten nach unseren Erfahrungen fur ausgeschlossen, dass Kalium aus reinem, geschmolzenem Kali nach dem Castnerschen Verfahren industriell hergestellt werden kann.

Formatik - 5-8-2009 at 11:28

Looks like a classical case of "if it works with A", it will (really an implied hope for should) 'also work with B, C, D, E, E ,F, J' that one typically finds prodding around in the patent literature.

Sodium by electrolysis through glass

franklyn - 21-9-2009 at 12:16

From - 1926

A remarkable paper , not so much for production of sodium , but for the
disclosure that glass normally thought of as an impermeable barrier can
serve as a membrane.

The scheme reveals that by using an ordinary lightbulb immersed in
molten sodium nitrate and the application of ordinary household electricity,
sodium is diffused into it and collects inside of the bulb.

A modest amount of perhaps a gram but sufficient for drying chemicals.
given lightbulbs are as cheap as they are this really bears investigating.


S.C. Wack - 21-9-2009 at 13:52

He had an article published a year before, mentioned in a JCE article [34, 289 (1957)] that I uploaded 5 years ago. A quote from it interested me, but I never found the journal cited:

"The use of the hot filament of an incandescent vacuum electric light bulb in place of the cold cathode in a Geissler tube as a source of electrons for combination with the mobile sodium ions from the glass, was first suggested by A. W. Hull of the General Electric Company. In 1925 R. C. Burt [J. Opt. Soc. Amer., 11, 87 (1925)] used the Hull principle to produce commercially acceptable sodium photoelectric cells."

But 257 mg. does not sound commercially acceptable, for the production of sodium. But given another quote:
"Sodium ions from the fused salt electrolyte thus continuously replace those from the glass which have been reduced to atoms, and the weight of metallic sodium obtained inside the evacuated glass bulb is proportional to the number of coulombs used.",
I have to wonder just how much sodium one can obtain from a single 100 watt bulb.

[Edited on 21-9-2009 by S.C. Wack]

densest - 21-9-2009 at 17:45

Quote: Originally posted by S.C. Wack  

I have to wonder just how much sodium one can obtain from a single 100 watt bulb.
[Edited on 21-9-2009 by S.C. Wack]

Assuming a mostly spherical bulb 10cm in diameter, with the filament centered, leaving 1/2 of the radius as space between the filament and the surface of the sodium, pi*h2*(r-h/3) is about 82cm3 assuming r=5 and h=2.5. Also assuming the cheapest possible light bulb (but stock up, they're disappearing soon!) it's maybe $0.70 for 60+g? The paper seemed to imply that it just kept coming through...

For the truly insane, one could string the filament from a quartz-halogen lamp inside a beta-alumina tube, seal & pump, dip in molten salts & have a fine time... It's really sad that long fluorescent tubes have to have mercury in them because they have a -lot- of surface area.

Poasssium can be had via Potassium glass

dann2 - 22-9-2009 at 07:52

Procedures in Experimental Physics (book) page 536 describes the process. A mixture of Sodium Nitrate and Nitrite are headed in an Iron tub with the bulb in it.
The Sodium condenses up high on the bulb where a stream of cold air is blown onto it..........................
Potassium can be prepared by a similar method using K Nitrate and Nitrite and using Potassium glass.

Will copy and post later.

Where would one obtain Potassium glass? Make from fused quartz and K compound?



chemoleo - 22-9-2009 at 13:51

Cold air on Na condensation? How cold is cold?

Would be interesting to hear more!

watson.fawkes - 22-9-2009 at 16:30

Quote: Originally posted by dann2  

Procedures in Experimental Physics (book) page 536 describes the process.[...]
Where would one obtain Potassium glass? Make from fused quartz and K compound?
Oh, you've referenced one of my most favorite books.

Basically yes. I don't think potassium glasses are readily available in commercial trade. Carbonates are typically used in glass manufacture because the liberated gas bubbles help mix the melt and drive dross to the surface. Standard flint glass is also made with calcium, whence the lime in soda-lime glass. So raw materials are silica of some form (the purity of fused quartz might not be necessary), potassium carbonate, and lime (or the carbonate). Other additions, if you want a more refractory glass, can include alumina and borax. After that, you'll need to work it into a vessel, which more or less means blowpipe work. Not exactly a quick project.

Then again, neither is the procedure in the book. You first seal on a tube to a light bulb and evacuate argon with your trusty vacuum rig. Then you use a power supply that provided both (1) arc voltage from one side of the filament to the bulb immersed in the sodium salt bath and (2) a DC bias from the other side, which provides your electrolysis current. Electrically, this device is much like a tube diode with the heater filament and cathode combined. You heat the bath to melt the sodium salts and turn the glass into an ionic conductor. You blow air on the top of the bulb to make, essentially, a reflux condenser that's cold enough so that the reflux solidifies. To answer the question above, it seems that it's ambient air, but in rather high volume.

The reference given is "Burt, R.C., J. O. S. A., 11, 87 (1925)", but I don't know offhand what journal that is.

dann2 - 22-9-2009 at 16:39

I photographed some pages of the book and attached.

This link:
says television tube glass is Potassium glass. A large 24 inch TV tube should be large enough of a 'bulb' (at least to get started :P)



Attachment: (1.3MB)
This file has been downloaded 973 times

[Edited on 23-9-2009 by dann2]

S.C. Wack - 22-9-2009 at 16:52

Quote: Originally posted by chemoleo  
How cold is cold?

With the NaNO3 being 350C, not very. I'm unaware of any incandescent light bulbs under vacuum; which seems to be a requirement, for sublimation if nothing else.

watson.fawkes - 22-9-2009 at 16:54

Quote: Originally posted by dann2  
This link [...] says television tube glass is Potassium glass.
It says that potassium goes into the making of glass for television tubes. There's potassium oxides in ordinary container glass, as a rule. So while we might expect that this TV tube glass is higher in potassium content that other glasses, that doesn't mean it's going to be free of sodium. Then again, this question really only addresses the purity of the final product. It's fully likely that you'll get ionic substitution in the glass and sodium contamination, but that may not matter, depending on what you're aiming for.

Nevertheless, if you're serious about this, I would recommend that you first duplicate the sodium electrolysis, since there are already enough moving pieces to get right, even with a known-good example.

12AX7 - 22-9-2009 at 18:14

I wonder if the sodium and potassium ions melt at different temperatures.

(Notice ion mobility == phase change of just one part (not the silica lattice), in the same way that many ceramics are liquid phase sintered while retaining their overall structure.)


vulture - 23-9-2009 at 13:27


I have not seen 1kW PSU on the market. At 12V they would be useless for electrolysis - at 5V they would be OK.

You can order them from any decent supplier nowadays. PC PSU's will also provide 5V and 3.3V with quite beefy amperages, so you can choose.

That model is capable of delivering peak currents of 40A on both 5 and 3.3V

[Edited on 23-9-2009 by vulture]

hinz - 23-9-2009 at 14:04

Look for the HP 6260B PSU at E-bay It can supply 100A DC at 0-10V. Probably better than any PC PSU, cause the voltage can be regulated to the needs of the cell. Futhermore these beasts are no Switching PSUs, they're based on a tyristor preregulated transformer, so ther're no MOSFETs to go bad.

There're also other HP 62xx PSUs around, some of theulra heavy ones can supply even more than 100A DC.

watson.fawkes - 23-9-2009 at 14:43

Quote: Originally posted by 12AX7  
I wonder if the sodium and potassium ions melt at different temperatures.
An excellent question. Each exists in glass as the oxide. I'd have to say, if I were pressed to guess, that K2O melts first, simply because K has a free 4s electron with a lower ionization energy than Na with its 3s one. It seems to me that the activation energy would be lower in this case. It's likely, though, that there's no discrete phase change associated with these states, having a solidus/liquidus range where there are blended phases.

In this case, however, I'd guess that it's not phase transition that dominates, but instead ion mobility. Here smaller ions have a definite advantage. Potassium is going to move through any ionic conductor less readily that sodium. Offhand, I'd say that the phase partition between K and Na, even if it favors K, isn't enough to overcome the Na size advantage.

Having said all this, it seems far more interesting to try this, not with potassium, but with lithium. Lithium is going to be a better ionic conductor even than oxygen, and second only to hydrogen. The special glass would be made of silica-lithia-calcia, that is, lithia-lime glass. If temperature of the salt bath were controlled appropriately, it's possible that this makes a decent electrowinning process for refined, but not purified, lithium salts. There's some optimum rejection temperature above that where Li(+) becomes significantly mobile but below that where Na(+) does.

densest - 23-9-2009 at 17:53

Well, Google turned up an 1887 receipt book containing the instructions to grind up lepidolite, roast it for some hours, then melt it and use it as a glass... KLi2Al(Al, Si)3O10(F, OH)2, Potassium lithium aluminum silicate hydroxide fluoride. A violet to pink mica, Notable Occurrences include Brazil; Ural Mountains, Russia; several African localities and California, USA. If I can get hold of some I will try to blow some bubbles; dunno what melting point would be. If it worked, then one would have both Li and K glass in one go. The 1925 article did say that the bulb needn't be evacuated if I read it correctly - dunno what gases might have been used for fill then - more history to look up.

There were a number of lithium-containing glass formulas showing up in the search but all of them had large amounts of sodium in them which might or might not be bad. It would be tempting to substitute lithium into the formula for 7720 glass, but the article says that borosilicate glass didn't work. Pity, a low-thermal-expansion glass is much easier to work.

Now, how to find a chunk of lepidolite which is sufficiently pure to give a useful glass but not so beautiful as to be a sacrelige to melt?

watson.fawkes - 23-9-2009 at 18:18

Quote: Originally posted by densest  
If it worked, then one would have both Li and K glass in one go. The 1925 article did say that the bulb needn't be evacuated if I read it correctly - dunno what gases might have been used for fill then - more history to look up. [...] There were a number of lithium-containing glass formulas showing up in the search but all of them had large amounts of sodium in them which might or might not be bad.
As for making glass, look up "batch glass" in the art glassblowing world. It's typical for those folks to mix up their own glass compositions. There's plenty of information on it. Note: lithia is an aggressive alkali in molten form, meaning it will attack crucibles, particularly those with high silica content, readily. I'm not familiar with materials compatibility here, but I'm sure there are issues. And in any case, definitely do a tiny melt before a large one.

As far as evacuation, you need the right gas pressure to sustain a discharge. Now I said arc discharge before, but that's probably not right. Glow discharge is likely the right regime. The book mentioned sodium vapor discharge, which would certainly happen once the electrolysis and sublimation started. It also mentioned argon discharge, which may be what happens at the beginning. Whatever the actual pressure is, it interacts strongly with what the power supply has available to ionize gas. The supply supply should have an appropriate ballast circuit to limit current once the discharge starts.

Edit: I found the original article

Journal of the Optical Society of America, volume 11, issue 1
by ROBERT C. BURT (of Cal Tech)

[Edited on 24-9-2009 by watson.fawkes]

densest - 23-9-2009 at 20:10

@watson.fawkes - yes, I'm at least theoretically familiar with glass production - I do flameworking and have made gold-ruby and silver-multicolor glass in the flame from clear boro and metal/metal compounds. The biggest problem is crucibles; no matter what you use, molten glass eats it. I have some BN powder spray which seems to be inert up to about 1600C which might protect a small crucible as long as the atmosphere was not strongly oxidizing. The most promising idea from my point of view (small batches, occasionally) is an induction furnace since once hot, glass becomes sufficiently conductive to absorb electromagnetic energy strongly. That has the advantage that the crucible is relatively cool and the batch self-insulates. One would have to initiate heating with a torch to melt a blob in the center of the crucible and let that heat the rest from the inside out. I can make high power electronics from parts on hand. Crucibles I have to buy and I'm on a very very tight budget for new acquisitions. I have some alumina crucibles of the right size which would probably get eaten out during the first melt if they were heated from the outside. They might hold up if the highest temperature molten glass didn't touch them.

The curse of the "new age" has hit yet another mineral. Lepidolite is purported to have all sorts of wonder peaceful principles so it is in demand for charms and fetishes. Still, there's a place which sells it for $12/pound with the pictures appearing to be of reasonably pure mica, so that's a possibility. Reade Materials advertises that they sell it - anyone ever dealt with them? A railroad car load is a bit more than is necessary.

If I -can- make a melt of this, I can probably make 5-8 cm spherical bubbles - would anyone be interested in one if I make them? Doing just one or two seems to be a waste of prep time... I'll also try melting some Corningware (tm) which is at least partially Li (probably a lot of Na too, don't have the formula.)

All this is moot if anyone finds prefabricated useful shapes of Li glass...

not_important - 23-9-2009 at 21:15

Just buy the oxides/carbonates needed for the glass at a pottery supply, don't try to find lepidolite.

watson.fawkes - 23-9-2009 at 22:06

I've had something of an insight about this. It's possible that you don't need to start with a glass of any particular composition if (big if) you are willing to run it through an initial period and/or tolerate some contamination at the beginning of the campaign for the apparatus.

The electrical current in this system travels through three media: the gas fill from the cathode to the inner surface of the glass, the glass itself, and the salt bath. In the gas, the current carriers are free electrons and the alkali ions in plasma. In the salt bath, they are electrons and alkali ions in solution. In the glass, however, I believe that major negative current carriers are oxygen ions, with some small portion of electrons. While the net model is that of alkali ion diffusion through the glass, an individual alkali ion would travel only slowly through the viscous, hot glass. The alternate activity is that the alkali ions are combining with oxygen at the outer surface of the glass and dissociating from it (reducing) at the inner surface. It's easier for an oxygen atom to move from a neutral atom to a newly arrived alkali than for that alkali ion to move under the influence of the electric field. On the other side, free electrons reduce alkali atoms in their oxide, making space for new alkali to diffuse to the surface, tugging on a diffusion chain for the alkali current. So the oxygen current corresponds to this hopscotch kind of net motion. There will also be some direct diffusion because of the electric field; this corresponds to the electronic current.

The upshot of all this is that, if the composition of the alkali salt bath doesn't match the alkali composition of the glass, then the composition of the glass will change over time by ionic replacement. The most easily reduced ions will come off the surface first, so potassium before sodium before lithium. Among other things, it means that if you use it for lithium, you'll need to purge the glass of its free sodium (which may be all the sodium; I don't know).

If this idea is true, it should be easy enough to test by running an ordinary soda-lime glass bulb in a lithium bath and running an assay in the initial product. There should be a mixed product of sodium and lithium, with sodium initially dominating.

Crown ethers?

Helgoland - 25-5-2010 at 17:14

Sorry to revive an old thread, butI've been thinking a bit. If I annoy you, I'll stop.
Crown ethers are cyclic ethers that can form complexes with cations and so act as PTCs.
I was thinking water at the anode, aprotic solvent with crown ether at the cathode.
I couldn't find anything on their electrolysis though.
Crown ethers are quite expensive, I think. Maybe polyethylene glycol (with methyl groups at the ends, called podands) might do the job as well.
What do you think?


EDIT: Here's another good link (in german) on host-guest-systems in general.

[Edited on 26-5-2010 by Helgoland]

An interesting case of soil electrolysis of sodium

The WiZard is In - 26-5-2010 at 13:10

Noted in passing. You will forgive if this is not new, I am
not going to read 7-years of posts ......

The Journal of the Society of Chemical Industry
No.5 Vol. XXVI,. March 15, 1907.

Liverpool Section.

Meeting held at the University, on Wednesday, December 12,1906.




Now that electricity is so largely used for lighting and power purposes, and wires
conveying the current traverse the ground, especially in towns, in all directions,
the effects of wandering earth currents, due to leaks from the wires, and earth
returns on tramway systems, occasionally attract attention. These earth currents
usually manifest themselves by causing the corrosion of metal objects, such as
pipes, tramway rails, &c., placed in the ground. In such cases the object attacked
acts as anode, the surrounding earth being the electrolyte. In other cases, where
we are dealing with the opposite end—that is to say, the cathode—of the huge
Voltaic cell thus formed, alkaline deposits are found.

The following case, which was brought to my notice some months ago, may be
of interest, as it shows the magnitude of the chemical processes which may be
caused by such earth currents. On March 6th a bad " earth " occurred at the
Brownlow Hill Workhouse, Liverpool. The leak occurred just under a small
roadway, and although the cable had probably been failing there for some time,
the leak was only discovered from the fact that horses became restive at that
spot owing to shocks which they received, and which, in fact, one could get by
touching certain places. The cable affected was one supplying power to a motor
at 460 volts. The current was carried by two thick insulated copper cables (a
negative and a positive). These were laid about an inch apart on wooden
bridges, in a wooden trough, which was then filled with bitumen, a wooden lid
being finally laid over the bitumen. This trough was buried in porous sandstone,
the top being about 18 inches below the surface of the ground. The leak had
possibly been started by a heavy cart causing a crack in the bitumen, through
which water had penetrated to the cable. Only the negative cable (which would
be at 230 volts below the potential of the earth) was affected. This cable had
been joined at the place where the leak occurred, the join having been bound
round, perhaps not very carefully, with waterproofed cloth.

According to the clerk of the works, under whose directions the leak was
repaired, the material surrounding the leak was still so hot that one could only
just touch it, although the current had been switched off for an hour and a half. It
was found that the top of the wooden trough bad been burnt through, and that a
hard substance protruded through the hole thus formed and penetrated to the
wire. This substance was so hard that it had to be removed with cold-chisel and
hammer, and the workmen, to their surprise, caused big sparks during the
operation. They also got rather burnt under the finger nails and on the face by
flying chips of the evidently strongly alkaline material. The hard lump had been
formed round the negative wire, but the metal was not at all corroded, neither
had the insulation of the neighbouring positive wire been affected. The
substance removed from the leak was collected, and the engineer, Mr. A. J.
Wilson, who had been much struck by its curious properties, asked me to
investigate it.

The pieces brought me were dark grey in colour, very hard, very deliquescent,
and strongly alkaline. If treated with a little water, small yellow or lavender flames
were formed by bubbles of hydrogen catching fire. The mass evidently contained
free alkali metal; in fact, on breaking a piece of the hard substance, the crevices
in the interior were found to be filled with bright liquid metallic globules, evidently
a liquid alloy of sodium and potassium. When these globules came in contact
with water they ignited, although the viscous alkaline liquid, with which the lumps
of the grey mass soon became covered on exposure to the air, had very little
effect on them. Analysis of the hard grey substance showed it to consist chiefly
of potassium and sodium hydroxides, with a certain amount of the free metals.
There was also a considerable amount of sand from the surrounding sandstone,
as well as some silica rendered soluble by the action of the alkali, which must
have been nearly fused during the passage of the current. It was quite free from
carbonate, sulphate, chloride, &c., neither did it contain any metals other than
sodium and potassium. A quantitative analysis gave the following results: KOH,
33.37; NaOH, 32.26; K, liquid alloy 1.00 ; Na, liquid alloy, 0.80 ; soluble silica,
4.80 ; sand and earthy matter, 26.36; bitumen and water (by diff.), 1.41; total,
100.00 per cent.

The determination of the free metals was carried out by measuring the amount of
hydrogen evolved when a given weight of the substance was decomposed with
water in a graduated tube over mercury (0.312 grm. substance gave 8.4 c.c.
hydrogen at 19o and 470 mm. pressure). Knowing the total amount of potassium
and sodium in the substance, the amount of the liquid alloy and its composition
can then be calculated. The composition of the alloy, as calculated from the
above analysis, is 55.6 per cent. potassium and 44.4 per cent. sodium.
According to the researches of Kurnalrow [1], such an alloy would be liquid at
temperatures above 7o C.

The formation of alkali by the electrolysis of the salts dissolved in surface water
is, of course, not very surprising, but the quantity formed in the present case is
rather astonishing when it is considered what a very small amount of potassium
and sodium salts are present in soils. I had in my possession pieces of the
alkaline mixture weighing in all 570 grms., while from what I was told probably
not less than 1000 grms. had been formed altogether. The surrounding bitumen
for a short distance was also more or less impregnated with alkali.

If the leak had not been discovered, but had been allowed to continue, as often
happens in such cases, the alkali would have eventually destroyed the insulation
of the neighbouring positive wire and have caused a " short circuit."

The alkali was of course formed by the potassium and sodium first liberated by
the electrolysis reacting with water. This must have caused the evolution of a
large amount of hydrogen. If this gas had been evolved in any position in which it
could have accumulated and have got mixed with air, a violent explosion could
quite easily have been caused, either by a light being brought near or simply by
a globule of the liquid sodium-potassium alloy coming in contact with water and
causing a spark. Many of the explosions in connection with electric light mains,
usually attributed to accumulations of coal gas, are very possibly in reality due to
hydrogen formed by electrolysis--a suggestion which was first put forward by Mr.
Foulger at the Board of Trade inquiry into the explosion which occurred in the
conduit of an electric light main in the Euston Road; London, on Dec. 29, 1895.

It seemed pretty certain that the alkali formed in the present case had come from
the surrounding soil and sandstone, but as it was just possible that it had come
from the bitumen, this was analysed. A piece of the same bitumen as that which
had been used in laying the cable gave on ignition 33.2 per cent. of ash, and the
amount of alkali extracted from this by boiling with concentrated hydrochloric
acid corresponded to 0.18 per cent. potassium oxide and. 0.37 per cent. sodium
oxide in the bitumen. Now as the total weight of bitumen removed during the
repairs, and which might possibly have had its alkali removed, only amounted to
about 1200 grms., 6.7 grms. of mixed alkali is the maximum obtainable from this
source. However, it is much more likely that none of the alkali had come from the
bitumen, for to all appearance this had been very slightly affected. A small
amount of potash might likewise have come from the wood of the trough in which
the cables were laid, but the amount of this would be still more insignificant than
that which could have been derived from the bitumen. It is therefore clear that
the alkali could only have come from the surrounding soil.

The sandstone in which the cable was laid was seen under the microscope to
consist practically entirely of rounded quartz grains, and, as was to be expected,
was found to be very poor in alkali. The alkalis, which were extracted by boiling
with concentrated hydrochloric acid, were determined in a piece of sandstone
which was obtained from a spot at some distance from the Beene of the " earth,"
so as to avoid any possible contamination with the alkali which had been
accumulated there. This sample of sandstone contained only 0.047 per cent. of
potassium oxide, and 0.016 per cent. of sodium oxide. The thin layer of surface
soil contained a slightly higher amount of alkalis, which were also present in
rather different proportions, viz. : 0.056 per cent. of potassium oxide, and 0.055
per cent. of sodium oxide.

Although these quantities are very small, the volume of ground from which they
could accumulate was practically unlimited, and the amount of alkali found round
the " earth " could all have been obtained from about a third of a cubic metre of
the sandstone. Taking the specific gravity of the latter as 2.7, this quantity would
weigh 900,000 grms. and would yield 630 grms. of alkali, whereas about 660
grms. were actually formed in the present case. The same amount of alkali could
have been obtained from about a fifth of a cubic metre of the surface soil. It will
be noticed that the relative proportion of the two alkalis in the surface soil is
exactly the same as in the hard alkaline material formed round the " earth." It is
therefore probable that the alkali accumulated at the latter had been almost
entirely derived from the surface soil, and only to a very slight extent from the
underlying sandstone.

It should be mentioned that a few days before the occurrence of the above "
earth " the weather had been moderately dry with, however, occasional pretty
heavy rain.

Another case of alkali formation.—When I mentioned the above case of
electrolysis to Professor Marchant, he told me that he had recently been having
some trouble with some of the switches on the basement walls of the
electrotechnical buildings of the University of Liverpool. White efflorescences had
been forming on these and causing trouble. Professor Marchant gave me some
of the efflorescence from one of the switches, and an examination showed that it
was simply a mixture of 42.66 per cent. of potassium carbonate, and 12.02 per
cent. of sodium carbonate, together with water and a small quantity of alumina
and silica. It was only the switches on the negative wires which were thus
affected. It is plain that moisture had crept through the walls from the outside
soil, and finally on reaching the switch, electrolysis of the dissolved salts had
occurred. Potassium and sodium hydroxides had thus been slowly formed, and
carbonated by the action of the atmospheric carbon dioxide. The formation of
this efflorescence had been noticed for about two months, and by the end of that
time it had become so bad that the switch had to be removed and replaced by a
new one. It may be mentioned that this switch was placed against a concrete
wall which was below the ground level, but tarred on the side in contact with the
earth. The ratio of potash to soda quite, different to that found in the case of the
workhouse " earth," although the spots at which the two cases of electrolysis
occurred are only about 200 yards apart.

It is well known that the insulation of a positively charged wire tends to improve,
while that of a negatively charged one nearly always deteriorates. [3] This is due
to the phenomenon of " electric endosmose," which drives the moisture (or
electrolyte) away from the positive wire and towards the negative wire. This
action can be observed in an ordinary electrolytic cell. If the cathode is enclosed
in a porous pot, the level of the liquid in the pot gradually rises above the level of
the surrounding liquid. This is what is usually observed, but cases are also
known in which the effect takes place in the opposite direction. The force tending
to drive liquid through such a porous partition is proportioned to the difference of
potential between the two sides of the partition, and may be very great.

Insulated wires placed in the soil are subject to similar forces, and, as mentioned
above, the insulation of a positive wire improves. Any weak spot in the insulation
of a negative wire is, however, soon broken down, and moisture reaching the
wire, a leak is started. The insulating material may be regarded as a very slightly
porous partition, and in some cases it has been found that leaks may occur
without any mechanical break in the insulation, electrolysis also occurring with
formation of alkaline liquids under pressure between the insulation and the wire.
It appears that on account of this " electric endosmose " about 99 per cent. of the
faults which occur on direct current networks occur on the negative wire. At all
negative faults alkaline deposits or solutions are formed, and although it is
known that these usually contain potash and soda, in only one or two cases do
careful analyses of the products appear to have been made. In certain rare
cases also the formation of free alkali metal has been observed.

The first record of such a case that I have been able to find is in a letter to the
"electrical World" of July 6, p. 5, 1889, in which an account of an "earth" in
Boston, U.S.A., is given. The "earth" in many respects was similar to the one
recorded in the present note, for the cable affected was embedded in bitumen
and a hard lump was found at the fault, which flashed when wetted.

Another case of the kind does not appear to have been noticed until 1895. In
February of that year a serious explosion occurred in Euston Road in one of the
conduits of the St, Pancras electric-light system. Major Cardew, during his
investigation of the affair for the Board of Trade, found incrustations on some of
the insulators supporting the bare copper strips by which the current was caried.
These incrustations gave sparks with water, and were found to contain free alkali
metal, and from a statement in an editorial note in the " Electrician," [4] it is plain
that a fluid potassium sodium alloy had been formed, as the metal is described
as occurring in globules almost as fluid as mercury. As a result of Major
Cardew's discovery a committee was formed to investigate the causes of the
formation of these alkaline deposits. When the committee inspected the St.
Pancras mains (not quite at the place the explosion occurred), they found an
incrustation which flamed on treatment with water, but which in this case
contained solid (not liquid) alkali metal. In the Committee's report is an analysis
of the incrustation found on the cables near where the explosion occurred, as
follows :—Sodium hydroxide, 7.69 per cent. ; potassium hydroxide, 4.88 per
cent. ; sodium carbonate, 34.40 per cent. ; potassium carbonate, 52.77 per cent.
; also traces of silica, alumina, and lime. There appears to have been no free
metal in this incrustation at the time of analysis.

Since 1895 numerous cases of alkali metal formation have been noticed, but I
have not been able to find any other mention of an occurrence of the fluid alloy,
nor any other analysis of the alkaline incrustation.

The second case of electroysis described in the present note is rather a striking
illustration of the respective behaviour of a positively and negatively charged
insulated wire in presence of moisture. It was explained how the insulation of the
negative switch broke down owing to alkali formation. Now, although the positive
switch was placed in an exactly similar position to the negative one, no signs of
electrolytic action were shown by it, and although at only a short distance from
the negative switch, the leak from the latter did not follow, as might have been
expected, the short direct path to the positive terminal, but went a roundabout
way to the soil, and then, probably, to some tramway rail several hundred yards


The CHAIRMAN said it must always be borne in mind that rubber was a very
perishable substance, and that, whilst new rubber might be absolutely
impervious to water, even under the influence of the current, partially perished
rubber might behave like a porous substance. The phenomenon of electric
endosmose possessed very great interest. It had found practical application in
the patented methods of drying peat by aid of the electric current, and, from what
Mr. L. Hargreaves had told him, it was not without influence in the successful
working of the Hargreaves diaphragm.

Prof. E. W. MARCHANT said that the general character of these faults had been
known for a considerable time. In Paris and St. Pancras, faults almost identical
with those Dr. Bassett had described had come to light, but they occurred in
mains laid in earthenware troughs, not in mains laid in bitumen, on what was
known as the solid system. The case named by Dr. Bassett appeared to be the
first one that had been described with mains laid in bitumen. A recent experiment
by F. Fernie, which showed the way in which such faults might be produced, was
this:—One of the highly-glazed earthenware tubes that were used for carrying
conductors was buried in the ground, the lower end embedded in bitumen, and
the urper end about 2 ins. above the level of the ground; a coil of wire was round
the outside of the tube, and another round the inside. Water was first poured on
the ground round the tube in large quantities, and the tube left for 14 days. The
inside remained quite dry showing that the tube was perfectly water-tight. The
two wires were then connected to a lighting circuit of 230 volts pressure the
inside being made negative to the outside wire. After an hour or two, drops of
moisture appeared on the inner side of the tube, and after two days the tube was
half-full of water. On reversing the polarity of the wires, the tube ultimately
became quite dry again inside.

That gave a very good idea of tre actual force due to endosmotic pressure,
which tended to produce faults on a negatively-charged main ; it was astonishing
that the force was as big as it was.

Of course, the greater number of faults did occur on the negative main. The ratio
of faults was something like 100 to 1 ; for example, in comparing continuous
current networks with alternating current networks, wherever there were faults on
a d.c. system, they almost always occurred on the negative side. As to the cause
of the explosions which sometimes occurred in conduits containing electric
mains, he did not think they were often due to the electrolytic production of gas.
The amount of leak from ordinary gas-pipes was enormous. The efficiency of a
gas system was very low. If the meters of the consumers registered 80-90 per
cent. of the amount registered at the station, the gas company considered that
they had done pretty well. If there were a spark from any cause independent of
electricity, which would ignite the gas, an explosion would occur. In laying
cables at the present time, the greatest care was taken to make the pipes as
tight as possible, to prevent any gas leaking into them. It was only where there
were drawn in mains with leaky service-boxes that there was risk of ordinary
town gas leaking in. With reference to the rubber cables, he had never come
across a case in which there was an accumulation of water inside a cable, nor
had he ever known a perfectly insulated rubber cable act as a porous substance.

Dr. BASSETT, in reply, pointed out that the second case of electrolysis referred
to in the paper showed rather clearly the difference between the insulation of a
positive and a negative wire, brought about by electric endosmose. The switch
affected was the negative one, and a good deal of electrolysis had taken place
on it. The positive switch, although placed only a short distance away in an
exactly similar position, was absolutely unaffected, and the leak which had
caused the formation of alkali on the negative one had taken a very roundabout
way, and not the short, direct path between the two switches, as might have
been expected.

1 J. Russ. Phys. Chem. Soc. (1901), 33, 588; Zeits. anorg. Chem. (1902),
2 See The Electrician, 34, 308 (1895).
3 In this connection see an interesting paper, by F. Fernie, In The Electrician, 57,
125 (1906). Several of the statements in the next few lines have been borrowed
from this article.
4 The Electrician, 34, 563 (1895).

condennnsa - 14-10-2010 at 11:08

I found a place that sells nichrome (80% Ni, 20% Cr) heating elements real cheap, about $2 for a 2000watt element, each one is about 30 grams in weight. I figured, could I use a bath of about 30% H2SO4 and coil say 10 of these elements together, use them as anode, and plate it on a piece of stainless or whatever? Just like the way copper is electrorefined. That would give on heck of a nickel electrode.

Magpie - 12-1-2012 at 13:40

Is there an unofficial conspiracy against the selling of sodium to individuals? I am finding it increasingly hard to find. It seems eBay banned it some years ago and now one of my trusted suppliers just announced that they no longer sell it.

Polverone - 12-1-2012 at 13:58

I have seen members here recommend They still appear to carry sodium and sell to anyone.

Magpie - 13-1-2012 at 10:27

Thanks Polverone. Galliumsource even states that sodium is becoming increasingly hard to find. It seems that in the not-too-distant future we may all be faced with constructing our own sodium making machines.

Someone is always wanting to start his own chemical manufacturing business in his garage. For them, making sodium may soon become a business opportunity. It's already priced somewhere between filet mignon and truffles.

β-Alumina Method

Dave Angel - 12-7-2012 at 10:24

Really pleased to see the success had with Castner here, and it would be great to see the other hot electrochemical sodium methods successfully performed at a decent scale by us.

The β-alumina method is where I'll be focussing my efforts for now. If you haven't yet done so, you can download ziqquratu's attachment from page 11; the paper makes for a good read.

To summarise, we have a liquid sodium cathode, and liquid NaAlCl4 occupies the adjoining chamber with a graphite or RVC anode. The two are separated by a sodium-type β-alumina ceramic which selectively allows only the passage of sodium ions. Current causes said ions to flow through the alumina to the sodium cathode, where it is reduced to more sodium.

Chlorine is produced at the anode leaving AlCl3 which regenerates NaAlCl4 with excess NaCl present.

In a patent ( a similar method is described that involves addition of excess metallic Al rather than NaCl to the NaAlCl4 melt in order to generate AlCl3, making two useful industrial chemicals at once.

For amateur use, I propose both NaCl and Al be present in the NaAlCl4 melt, perhaps with a sacrificial Al anode instead of, or in addition to, the loose Al. My chemistry is rusty, but would this not then result in generation of AlCl3 at the anode (or in the mix if loose Al used) which would then regenerate the NaAlCl4? This set up would neatly minimise the need to scrub Cl2 or capture AlCl3 from the cell.

Anything I've overlooked in the chemistry? Worst case: the chemistry doesn't work and a Cl2 scrubber is needed.

Dave Angel - 18-7-2012 at 11:51

I have ordered the main components for a β-alumina cell, and additionally most parts to build a small Castner cell to produce the initial sodium required for the former cell type.

I'm proposing that, for the initial NaAlCl4, a simple HCl generator of excess NaCl with NaHSO4 is used, drying and potentially heating the evolved HCl before passing it into a heated mixture of dry, recrystallised and powdered NaCl (from dishwasher salt) and aluminium powder (250 mesh, filler grade).

At an elevated temperature (say >200°C - with NaAlCl4 melting at 185°C) molten AlCl3 is generated, and in situ with the NaCl forms molten NaAlCl4.

Sound reasonable?

If successful, it should be feasible to adapt the entire method to potassium production by use of a K-type β-alumina and KCl...

name of book?

mineralman - 25-7-2012 at 16:35

Quote: Originally posted by Organikum  
NaOH sodium electrolysis is done with this:

The trick is the iron net between the electrodes (cathode - copper, anode - nickel) which are only 2cm apart. This is a very tight net (100/per cm*cm) and yes it divides also the voltage of about 4V.
If interest I can post more data on this.

Hi, I just started reading this thread and read the words "The original process is so old, that little of the precise method/set up details are available" ?? roughly.

I instantly thaught back to a book that was passed down to me by my great uncle Harry Driver (Boffin @ Royal (warren) Arsenal, Woolwich Arsenal).
It had a detailed history w/pix, of all the pioneers in chemistry, from there humble beginings to the Industrialisaton of there processes/partnerships (castner/kellner for eg), polution containment and the realisation that one factorys pollution is anothers gold. but I digress

The image you have there is almost identicle to the ones in that book. Unfortunatly the police took it when they thaught I was making meth. lol, most days I have trouble making my mind up about something, making meth would be a nightmare.

The book in question was a green hardback,small A5ish size w/black print on cover, would that image be from a similar book? and if so, could you let me know the ISBN or its equivalent.
It's a long shot I know, but that book ment the world to me on both, it's sentimental & educational levels. MM

mineralman - 25-7-2012 at 17:07

Quote: Originally posted by Magpie  
Is there an unofficial conspiracy against the selling of sodium to individuals? I am finding it increasingly hard to find. It seems eBay banned it some years ago and now one of my trusted suppliers just announced that they no longer sell it.

The way in which they changed the drug precurser & general chemical buying laws, has turned both to becoming weekly shopping list chemists, and slowley but surely all those items you could just buy off the shelf will be replaced with modified versions, (inhibited acids,namebrand multiple ingrediant fertilizers instead of single chemical ingrediant & having to sign or show ID to purchase them), they should have invested more time/money in the keeping tabs on, and allowing the sales of those original precursers & chems, they would at least have known to whom and where they were going, now they havn't got a clue as everyone is building castner cells & mixing wood ash water to the concentrated scrapings of white bacterial matter from there backyard shit pile :D

So both good & bad, good that only the ones pepared to put the time/effort in to learning the wonders of alchemy.
Bad that something so usualy readily available takes days to weeks to make.
But as for conspiracy, that would mean this conversation never happened, as we would be unaware of anything going on. MM

mineralman - 26-7-2012 at 06:17

Quote: Originally posted by not_important  
Just buy the oxides/carbonates needed for the glass at a pottery supply, don't try to find lepidolite.

Any decent rock shop will have Lepidolite, but a shop that specalises in mineral specimens will have some real quality lithium content Lepidolite. MM

[Edited on 26-7-2012 by mineralman]

LEPIDOLITE 004.jpg - 135kBLEPIDOLITE 002.jpg - 127kB

Dave Angel - 26-7-2012 at 09:49

Regarding mineralman's comment of chemicals becoming more difficult to aquire OTC or otherwise, we have to face the fact that, for the most part, modern society will never understand our hobby. I've posted before about even university chemistry teachers questioning my motives, and it was only that which surprised and disappointed me.

One way or another we learn to live with it, but then I've always considered it as much a challenge as it is an annoyance...

Anyway, I'm getting off topic.

To celebrate 10 years of science madness, I thought it would be rather appropriate to construct a Castner cell out of this:

CIMP.jpg - 31kB

Look familiar? I think it's quite apt :)

For an off the shelf item, this product is near-perfect for the build; all cast iron, the pestle even has a screw thread for the SS palm rest which, removed, is the perfect place for a connection to be screwed in.

So far I've sliced the grinding head off the pestle, filed the resulting rod and worked it through 6 grits (down to 600) of alumina / wet&dry paper (used in the latter 'mode') and had it plated with >500 microns of nickel - nice and shiny:

Cathy.JPG - 38kB

Sure, the plating wasn't absolutely necessary but, costing surprisingly little, it was a no brainer.

I'm looking forward to a weekend working on this, the intention being to strip the paint from the mortar, put a hole in the bottom and fix the modified pestle upright as the cathode using fire cement. How this will hold up to (near) molten NaOH is uncertain, so I have a few ideas, building on what I've read:

a) Keep the bottom cool so that solid NaOH provides a barrier between the molten alkali and the cement.

b) Fuse some NaCl and fill a small recess above the fire cement with this to offer a protective layer that will not melt at the operating temperatures.

Any idea as to whether molten NaOH will attack solid NaCl at ca. 300°C? I know they have a common ion but I'm concerned about exchange of the anions resulting in erosion of the solid.

Any other suggestions for the build are, of course, welcome.

 Pages:  1  2    4