Sodium!

Castner progress this weekend: drilled out the crucible, and fitted the cathode, mesh and collection bell. Schematic below - there is a slight taper to the cathode due to its origin:



A 100-mesh stainless steel mesh was used in place of the iron fabric. The bell is an upturned SS hip flask cup with a ca. 15 mm hole drilled in the bottom. Mesh and bell fixed together with fire cement.

The cathode was positioned through a 44mm hole drilled in base and cemented in place with more fire cement. The mesh-bell assembly was embedded into the cement, over the cathode, and a fire lit in the bowl to set the cement.



Unfortunately, during the firing, the cathode slipped to an angle (see picture, below) so it is no longer central. I'm somewhat tempted to break the whole assembly out and reset it, maybe using a high temperature epoxy rather than awkward fire cement.



Does anyone knows if a functional cell will not be possible with the uneven voltage division and current densities resulting? If it's just a case of reduced efficiency then, since this is just a trial build, I'm probably going to push ahead with it regardless. The main goal is a β-alumina cell, though I may build a more robust Castner in time to determine the optimal parameters.

Poor-man's spot welder success!

Having searched the forum I see that the MOT spot welder is old news, but as a quick aside I just this week came across the hack-a-day when looking for a better way to make the stainless steel mesh cylinder and affix it to the collection bell.

Most of my Sunday consisted of building and using this:



Of anything I've built, it wins the prize for use of the oldest hoarded bits; the transformer having been salvaged 11 years ago, tungsten electrodes bought about 8 years ago and a stapler I've had since 'high school'! It ain't pretty but cost less than £10 for the 8 AWG cable. Bonus points for excessive use of duct tape

Anyway, my welds aren't pretty; I had to judge the pressure by hand and time the weld by switching the power at the socket, but... it's my first foray into welding and it's done the job:



This is so much neater than my original attempt with fire cement, and leaves the embedding of the structure in the base of the pot as the only place where the cement is really required to hold up. Slowly getting there!

Thanks for the words of encouragement Tank! This is one of the more involved projects I've taken on and there are many hurdles to overcome. It would be great to see your Downs cell - have you posted it before? I couldn't find it with a cursory search.

Thoughts of automation have also crossed my mind - some sort of pump (Viton peristaltic, connected to micro bore copper tube?) and a simple sodium level sensor, e.g., copper wire at various heights connected to an appropriate circuit to detect when enough has built up to pump, and when to stop it. Fresh salt could be introduced using a powder moving screw type arrangement... Still, this is far into the future for me - just getting my 'Sodium Badge' will be a great personal achievement...
For now


Progress this bank holiday weekend:

Fixed the cathode and SS mesh / collection bell combo in place
Formed an anode option
Experimented with heating element options
Built the power supply


Cathode etc. installation:



Fire cement used here - the only real weak point where the molten alkali could spell the end for my cell. Having previously fired it using a combination of a hot oven, blow-torches and plain old wood fire, I've elected to use a more controlled process this time.

A U-tube of 8 mm diameter 0.6mm wall copper tube was formed and a tightly wound coil of NiChrome wire fed through, wrapped with a spiral of glass fibre weave for electrical isolation. You can see the firing set-up for setting the mesh in place below - voltage, temp and current (just off shot) being monitored on separate meters. The second shot shows the cathode firing set up, and the now heavily oxidised copper U-tube more clearly. Voltage is controlled accurately by an 8 A variable transformer (off-camera), but could just as easily be controlled by the circuit from linearly variable heat gun on a budget. Ceramic wool is used to thermally insulate the set-up.


Anode construction:

About 5 attempts have been made at this and left me with a fair bit of scrap copper. This is the best result:



Using the same 8 mm copper tubing as above, a section was filled with free-running table salt and the ends hammered shut, allowing bending to be free of major kinks. The tube was bent around the same 1 3/4 inch (44 mm) hole cutting bit used to drill the hole in the bottom of the mortar, as it is, conveniently, a good size to give the correct spacing ratio of the cell electrodes from the mesh. I will get this plated with about 500 microns of nickel.


Now, I know what you're thinking, 'why not run the heating coil for the cell through the anode like for the cement firing?' This is the preferred option, though there are a couple of issues to deal with:

1) How to insulate the nichrome from the copper?

I've already used glass fibre weave successfully, though it only exacerbates the second issue (see below). So to provide a non-bulky electrically insulating layer over the nichrome I've done a little proof of concept work:



Nichrome was salvaged from an old electric heater - thanks to len1 for pointing out this, in my hindsight obvious, source - and for the rest of the Castner learnings frankly! The wire was wound into a short spiral and plated in a bath as described by Cyrus here, though without the thiourea (as I don't have any). The fine layer of copper was oxidised in place by passing current. The layer was tested and found to be electrically insulating.

2) How to feed the heating coil through the anode?

This is a bit trickier. I've tried feeding the heating coil directly into the anode, using different gauges of copper wire to lead it. Also tried string, cotton thread, brass wire, and using a magnet, and vacuum in an attempt to draw these through (with appropriate attachements to the end of the lead-cord).

None of these techniques seem to get it very far - it always gets caught somewhere. I'm thinking that the magnet technique to guide a light-weight thread round will probably still be the best bet, but does anyone have any better ideas?

Fall-back heating option:



Not a lot to report here. I found, and put a 44 mm hole in the bottom of, an old aluminium steamer. At the least, this will provide a container for the insulating material, and should I not be able to fit the anode with a heating coil, it will form a custom heating mantle for the cell.


Power Supply:



Thanks again to len1 for the power supply design he described earlier in the thread. I've built something similar using the MOT from the quick and dirty spot welder with 5 turns of 8 AWG cable and 5 x 25 A bridge rectifiers in parallel - some more 8 mm copper tube was hammered flat and drilled to provide cheap bus bars for this purpose. It puts out almost exactly 5 V. Since taking the photo I've added some proper feet to give the bolts protruding on the bottom some stand-off.

I did try a build with a computer PSU but found this to be unreliable. With suitable cooling (maybe even a case to keep bare wires contained for a change!) the PSU above should be far more robust.


Well, that's all for now, will be getting the anode plated in the next couple of weeks and exploring the heating options as the next priority.
Comments and advice, naturally, welcome!

Quote: Originally posted by Dave Angel

Thanks for the words of encouragement Tank! This is one of the more involved projects I've taken on and there are many hurdles to overcome. It would be great to see your Downs cell - have you posted it before? I couldn't find it with a cursory search.

I started a thread but as I recall, it turned into a pissing contest due to my sloppyness. I made some modifications and improvements to the cell but ultimately stuck with the KIS,S approach. I've since performed a ton of experiments with different salts, anodes, cathodes, methods of collecting sodium, etc..

To be honest, if I need some sodium I just fire up the power supply and "roast some marshmallows" (my neighbor's words). I sit there with a long scoop and when enough sodium forms on the melt, I scoop and dump then refine.

I'll look through my old pics and post some images of my makeshift Downs cell projects when time permits. I salute you for being a better planner and thinking your design through (unlike me) heh.

Tank

Quote: Originally posted by m1tanker78

I started a thread but as I recall, it turned into a pissing contest due to my sloppyness.

I find it disappointing when this happens - as amateurs, it is worth encouraging each other in the simpler / inventive / resourceful approaches, particularly as our resources become restricted by endless wars on terror/drugs/'chemicals'.

Heh, 'roasting marshmallows' brings really vivid images to mind... I'd not likely be the only one to find interesting your successful blends of salts for your Downs cell, not to mention configuration and materials - I'll look forward it when you get the chance.

Finally, I've got a quick update - a shiny anode back from the platers today. Like the cathode, ca. 500 microns of nickel on this:



I'm still slightly in awe of what electroplating does, no matter how commonplace it may be.

I'll take a leaf out of Tank's 'KIS,S' book and stick with external heating for this first cell, saving the heated anode idea for a future build once I've ironed out the basics.

Success!

I'm pleased to report that I've made sodium with my Castner cell

Summary of recent work:

I built some feet into the base of the old steamer with threaded rod and nuts, wrapped it in loft insulation and lined the inside with a crucible form-matching packing of ceramic wool. A coil of fresh, 22 SWG nichrome was pressed into the wool to provide the initial heating.



The crucible itself had the anode coil fitted using fire cement and an additional, ca. 15 mm, 'puck' of cement was also added to the base for extra containment should the base dissolve through. A sheet of twill weave fibreglass mat was placed over the nichrome coil and the crucible lowered into place. Temperature probes were inserted in two places: solid probe at the crucible wall and flexible probe through the anode for a core measurement.



The set up took about 1.5 Kg of molten alkali (OTC drain cleaner variety) to bring it to a working capacity, and probably took about 45 minutes to come to temperature (max. 330 °C) - I cannot report accurate figures as I was more concerned with safety and successful operation than time keeping. I had an exciting moment with sodium catching fire (argon blanket then installed), and I was on the clock given the effects the molten alkali was having on the cement...



Length of operation estimated at 2.5 hours given the timestamps on photos. Current is unknown as the actively cooling case I built around my PSU prevents access to the appropriate wires; a built-in ammeter for the next upgrade, I think. On the topic, the cooling was achieved by encasing the PSU in a long rectangular wooden box, open at both ends, with a microwave fan wired in at one end - very simple and effective.

Results:



Sodium was collected with disposable pasteur pipettes, which worked about half the time; the other half of the time, the sodium froze in the glass. The first photo is the metal I managed to pipette out, prior to purification. The second photo was taken during purification of the material which froze in the pipettes, these having been crushed and heated under oil.


Findings and learnings:

I still don't like sodium hydroxide, particularly not molten and especially not at these volumes. I can echo the sentiments and findings of others; it fumes and spits, gets everwhere, and is generally unpleasant to work with. I'd previously oven baked about 1 Kg of hydroxide to remove any water, but had to add another 0.5 Kg, un-dried, to fill the cell. Some significant foaming during the first 15 minutes of operation could probably have been avoided had the entire quantity of salt been thoroughly dried.

My cell was not insulated well enough / delivering enough current to maintain temperature without external heating. Unfortunately the nichrome broke about half-way through the run; a stray piece of hydroxide lost during filling having acted on it I suspect. Still, the cell carried on whilst very slowly losing heat, and actually seemed to become more 'sedate' (if one can apply that word to such an experiment) at lower temperatures, i.e., just above melting point.

Coalescence of the liquid sodium during purification under oil was far easier with the broken glass present, and I observed that lumps of alkali impurities appeared to help more so. Perhaps a simple physical effect of helping to break the surface tension - could be replicated with unglazed ceramic maybe?


Next steps:

Yield:
For this I intend to coalesce the entirety of the metal into one ingot under a suitable solvent; I have mixed xylenes on hand, with b.p. 140 °C, so figure this could work under reflux, providing the beads will come together.

Disassembly:
I've washed the cell out and it's going to come apart so I can fully assess the damage from the operating conditions.

Downs cell:
I intend to trial this option using my, now empty, argon canister as a crucible. I likely have enough sodium to build a β-alumina cell but would be more comfortable with an additional batch, and I'd like to have personal experience with all three methods to make a fair comparison.

Further update soon to follow...

[Edited on 23/9/2012 by Dave Angel]

SODIUM CELL

This is from Synthesis of Laboratory Reagents by LEONID LERNER

Setup and Operation
A 500–1000-mL nickel crucible with 10–12 cm top diameter is filled with 800–1000
g 99% commercial NaOH, placed inside a crucible oven, and heated to a temperature
of 360°C–380°C, as indicated by a thermocouple placed inside the melt. It is best to
protect the thermocouple by a close-fitting copper sheath, which can be made of a
sealed section of 3/16-in. copper tubing. This is best located in the anode compartment
directly behind the nickel separator where the temperature registered is not prone to
spurious variation due to intermittent contact with molten sodium. The NaOH takes
about 2 h to melt fully, whereupon 40 g of dehydrated Na2CO3 is added (this serves to
reduce the sodium diffusion rate, α, and the melt is held for an additional ½–1 h at
that temperature to almost complete dehydration. Thereupon, the oven is turned off
and a nickel separator, about 7 cm in diameter lowered to extend about 1–2 cm below
the melt surface. When the electrolyte has cooled to 318°C, an overhead cathode, at
a voltage of 12–15 V with respect to the nickel crucible anode, is lowered so that its
tip just touches the electrolyte, and current begins to flow. The high voltage produces instantaneous heating, preventing immediate formation of an insulating crust on the
electrode.
As soon as the current exceeds 20 A, the voltage is wound back to an average
operating level of 6.3 V at 318°C. There upon the cell voltage is computer-controlled
to counteract thermal runaway. This is achieved by decreasing the cell voltage linearly
as a function of temperature, so that at 316°C the cell voltage rises to 6.7 V,
with a corresponding decrease for a temperature variation in the other direction.
This control does not need to be precise timewise, and a digital output fed through
a low-pass filter with a 10-sec. time constant has proved adequate. During this time
the cell current should vary in the range 25–55 A.
After a period of 10–15 min, more sodium starts dissolving than is reacting with
the diffusing water, and hydrogen evolution moves outside the cathode compartment.
This is accompanied by a sudden surge in current above 60 A, which is used to trigger
an alarm indicating sodium needs to be removed from the cell. The cathode is
lifted out, and a strainer (mesh 30/32 is adequate) is used to lift the pool of sodium
and empty it into a container of paraffin. This method is used in the original Castner
patent, and works because sodium has a reasonably high surface tension compared
to the electrolyte. Any electrolyte frozen on the surface of sodium globules can be fitted with a 1–2-cm steel cylinder on which the mesh rests. In this case, any residual
NaOH in the strainer freezes to the walls of the cylinder when the sodium is poured.
Although a small amount of sodium is lost this way, the recovered sodium is clean
and ready for use (Figure€2.6).
The sodium should be removed from the cell as quickly as possible (the whole
operation should take no more than about 30 sec, so that the melt surface does not
have time to freeze), and the cathode tips are cleaned off any adhering solidified
electrolyte by pressing between the jaws of small pliers so that the solidified melt
flakes off. The cathode can now be reinserted, touching it against the electrolyte to
commence the flow of current. When a globule of sodium has formed after about 30
sec or so, the cathode is raised 1–2 mm, and the whole process repeated. It should not
now be necessary to raise the voltage above the normal level on reinsertion because
the cathode is still hot. To prevent excessive freezing of the electrolyte surface during
sodium removal, which can lead to spitting and small explosions upon recommencement
of electrolysis, the oven is set to turn on when the bath temperature dips below 314°C, which occurs only for a few minutes following sodium removal. After 80–100
g of sodium has been collected, fresh sodium hydroxide needs to be added (sodium
carbonate is not added as it is not electrolyzed).
Although it is often suggested that prior to use sodium be cleaned from the
adhering oxide/hydroxide crust by remelting and rolling under xylene [1], the evaporation
rate of xylene above the mp of sodium has been found to be too high, and
xylene is not very effective in removing the oxide crust. On the other hand, the bp
of xylene is still high enough to present some removal problems, and it is unsuitable
for long-term sodium storage because sodium rapidly oxidizes under it. In this
respect, paraffin oil is superior. Not only does sodium retain its luster for months
when stored under paraffin, but by heating the paraffin to above 120°C a light
crust cover disappears in a matter of minutes, due to saponification of the paraffin
and dissolution. To remove thicker crusts, the sodium surface can be trawled with
fine stainless steel mesh. The paraffin adhering to the sodium surface cannot be
volatilized as it has a very high boiling point, and some of its constituents carbonize
without vaporizing even under a vacuum of 10−3 torr. However, paraffin can be
fairly effectively removed by wiping the sodium with an absorbent lint free cloth,
which still leaves a very thin layer of paraffin adhering to the sodium surface,
preventing rapid oxidation in air during the wiping. The sodium is then shaken
for several minutes with a large quantity of dry ether, which removes the last thin
paraffin coat.

References
1. Furniss, B. S., Hannaford, A. J., Smith, P. W. G., and Tatchell, A. R., Vogel’s Textbook of
Practical Organic Chemistry, 5th ed. London: Addison Wesley Longman, 1989.
2. Birch, A. J., Reduction by dissolving metals. Part I. J. Chem. Soc. 117: 430–436, 1944.
3. Moody, C. J. ed., Synthesis: Carbon with two attached heteroatoms with at least one carbon-
to-heteroatom multiple link, pp. 27–8. Vol. 5 of Comprehensive Organic Functional
Group Transformations, edited by Katritzky, A. R., Moody, C. J., Meth-Cohn, O., and
Rees, C. W. New York: Pergamon, 1995.
4. Mordini, A., Sodium and potassium. In Main Group Metal Organometallics in Organic
Synthesis, edited by McKillop, A. New York: Pergamon 2002; Jenkins, J. W., Nalo, L.
L., Guenther, P. R., and Post, H. W., Studies in silico-organic compounds. VII. The
preparation and properties of certain substituted silanes. J. Org. Chem. 13(6): 862–866,
1948; see also Chapter 10.
5. Davy, H., On some new phenomena of chemical changes produced by electricity, Phil.
Trans. Roy. Soc. 98: 1–44, 1808.
6. Castner, H. Y., Process of Manufacturing Sodium and Potassium. U.S. Patent No.
452030, May 12, 1891.
7. Wallace, T., The Castner sodium process, Chem. Ind. 876–882, 1953.
8. Lorenz, R. and Clark, W., Über die Darstellung von Kalium aus Geschmolzenem Ätzkali,
Zeit. Elektrochem. 9: 269–71, 1903.
9. Allmand, A. J. and Ellingham, H. J. T., Principles of Applied Electrochemistry, 2nd ed.,
pp. 498–502. London: Edward Arnold & Co., 1924.
10. Thomson, G. W. and Garelis, E., Sodium: Its Manufacture, Properties, and Uses,
pp.€20–23. New York: Reinhold Pub. Corp., 1956.
© 2011

SODIUM CELL

This is from Synthesis of Laboratory Reagents by LEONID LERNER

Setup and Operation
A 500–1000-mL nickel crucible with 10–12 cm top diameter is filled with 800–1000
g 99% commercial NaOH, placed inside a crucible oven, and heated to a temperature
of 360°C–380°C, as indicated by a thermocouple placed inside the melt. It is best to
protect the thermocouple by a close-fitting copper sheath, which can be made of a
sealed section of 3/16-in. copper tubing. This is best located in the anode compartment
directly behind the nickel separator where the temperature registered is not prone to
spurious variation due to intermittent contact with molten sodium. The NaOH takes
about 2 h to melt fully, whereupon 40 g of dehydrated Na2CO3 is added (this serves to
reduce the sodium diffusion rate, α, and the melt is held for an additional ½–1 h at
that temperature to almost complete dehydration. Thereupon, the oven is turned off
and a nickel separator, about 7 cm in diameter lowered to extend about 1–2 cm below
the melt surface. When the electrolyte has cooled to 318°C, an overhead cathode, at
a voltage of 12–15 V with respect to the nickel crucible anode, is lowered so that its
tip just touches the electrolyte, and current begins to flow. The high voltage produces instantaneous heating, preventing immediate formation of an insulating crust on the
electrode.
As soon as the current exceeds 20 A, the voltage is wound back to an average
operating level of 6.3 V at 318°C. There upon the cell voltage is computer-controlled
to counteract thermal runaway. This is achieved by decreasing the cell voltage linearly
as a function of temperature, so that at 316°C the cell voltage rises to 6.7 V,
with a corresponding decrease for a temperature variation in the other direction.
This control does not need to be precise timewise, and a digital output fed through
a low-pass filter with a 10-sec. time constant has proved adequate. During this time
the cell current should vary in the range 25–55 A.
After a period of 10–15 min, more sodium starts dissolving than is reacting with
the diffusing water, and hydrogen evolution moves outside the cathode compartment.
This is accompanied by a sudden surge in current above 60 A, which is used to trigger
an alarm indicating sodium needs to be removed from the cell. The cathode is
lifted out, and a strainer (mesh 30/32 is adequate) is used to lift the pool of sodium
and empty it into a container of paraffin. This method is used in the original Castner
patent, and works because sodium has a reasonably high surface tension compared
to the electrolyte. Any electrolyte frozen on the surface of sodium globules can be fitted with a 1–2-cm steel cylinder on which the mesh rests. In this case, any residual
NaOH in the strainer freezes to the walls of the cylinder when the sodium is poured.
Although a small amount of sodium is lost this way, the recovered sodium is clean
and ready for use (Figure€2.6).
The sodium should be removed from the cell as quickly as possible (the whole
operation should take no more than about 30 sec, so that the melt surface does not
have time to freeze), and the cathode tips are cleaned off any adhering solidified
electrolyte by pressing between the jaws of small pliers so that the solidified melt
flakes off. The cathode can now be reinserted, touching it against the electrolyte to
commence the flow of current. When a globule of sodium has formed after about 30
sec or so, the cathode is raised 1–2 mm, and the whole process repeated. It should not
now be necessary to raise the voltage above the normal level on reinsertion because
the cathode is still hot. To prevent excessive freezing of the electrolyte surface during
sodium removal, which can lead to spitting and small explosions upon recommencement
of electrolysis, the oven is set to turn on when the bath temperature dips below 314°C, which occurs only for a few minutes following sodium removal. After 80–100
g of sodium has been collected, fresh sodium hydroxide needs to be added (sodium
carbonate is not added as it is not electrolyzed).
Although it is often suggested that prior to use sodium be cleaned from the
adhering oxide/hydroxide crust by remelting and rolling under xylene [1], the evaporation
rate of xylene above the mp of sodium has been found to be too high, and
xylene is not very effective in removing the oxide crust. On the other hand, the bp
of xylene is still high enough to present some removal problems, and it is unsuitable
for long-term sodium storage because sodium rapidly oxidizes under it. In this
respect, paraffin oil is superior. Not only does sodium retain its luster for months
when stored under paraffin, but by heating the paraffin to above 120°C a light
crust cover disappears in a matter of minutes, due to saponification of the paraffin
and dissolution. To remove thicker crusts, the sodium surface can be trawled with
fine stainless steel mesh. The paraffin adhering to the sodium surface cannot be
volatilized as it has a very high boiling point, and some of its constituents carbonize
without vaporizing even under a vacuum of 10−3 torr. However, paraffin can be
fairly effectively removed by wiping the sodium with an absorbent lint free cloth,
which still leaves a very thin layer of paraffin adhering to the sodium surface,
preventing rapid oxidation in air during the wiping. The sodium is then shaken
for several minutes with a large quantity of dry ether, which removes the last thin
paraffin coat.

References
1. Furniss, B. S., Hannaford, A. J., Smith, P. W. G., and Tatchell, A. R., Vogel’s Textbook of
Practical Organic Chemistry, 5th ed. London: Addison Wesley Longman, 1989.
2. Birch, A. J., Reduction by dissolving metals. Part I. J. Chem. Soc. 117: 430–436, 1944.
3. Moody, C. J. ed., Synthesis: Carbon with two attached heteroatoms with at least one carbon-
to-heteroatom multiple link, pp. 27–8. Vol. 5 of Comprehensive Organic Functional
Group Transformations, edited by Katritzky, A. R., Moody, C. J., Meth-Cohn, O., and
Rees, C. W. New York: Pergamon, 1995.
4. Mordini, A., Sodium and potassium. In Main Group Metal Organometallics in Organic
Synthesis, edited by McKillop, A. New York: Pergamon 2002; Jenkins, J. W., Nalo, L.
L., Guenther, P. R., and Post, H. W., Studies in silico-organic compounds. VII. The
preparation and properties of certain substituted silanes. J. Org. Chem. 13(6): 862–866,
1948; see also Chapter 10.
5. Davy, H., On some new phenomena of chemical changes produced by electricity, Phil.
Trans. Roy. Soc. 98: 1–44, 1808.
6. Castner, H. Y., Process of Manufacturing Sodium and Potassium. U.S. Patent No.
452030, May 12, 1891.
7. Wallace, T., The Castner sodium process, Chem. Ind. 876–882, 1953.
8. Lorenz, R. and Clark, W., Über die Darstellung von Kalium aus Geschmolzenem Ätzkali,
Zeit. Elektrochem. 9: 269–71, 1903.
9. Allmand, A. J. and Ellingham, H. J. T., Principles of Applied Electrochemistry, 2nd ed.,
pp. 498–502. London: Edward Arnold & Co., 1924.
10. Thomson, G. W. and Garelis, E., Sodium: Its Manufacture, Properties, and Uses,
pp.€20–23. New York: Reinhold Pub. Corp., 1956.
© 2011

Quote: Originally posted by BromicAcid

Love it Dave! My full scale Downs cell was a veritable disaster, glad to see it work out to some extent for another. Looking forward to your future updates.

So instead, I've today invested in some baby oil (Johnson's). This contains mineral oil for the most part, and some isopropyl palmitate and fragrance, though likely not much given how stingy such companies are with their ingredients. Anyway, this will suffice until I can get some pure mineral oil.

I had been using '3-in-1' brand oil for purification, but this seems to lose its lighter fractions and thickens up considerably, plus I'm not sure what sort of additives might be in it. Additionally, I stored the sodium under kerosene (Bartoline brand 'paraffin', from B&Q) overnight and observed a layer of red powder over the exposed material this morning, so clearly something is reacting - again, unsuitable.

Coalesence and purification complete:



The ingot weighs ca. 14 g (I don't trust the decimal point on my balance - it's time for a new one) so I'm quite pleased with that for a first attempt. I actually found (obvious in hindsight) that the coalescence works best as the sodium is solidifying; holding the beads at melting point and then cooling whilst gently stirring brings it all together nicely. Its behaviour reminds me of solder...

I've also partly dismantled the cell. The anode coil came away from its fixing as soon as the cell was washed out, and it's evident that it was in contact with the gauze at some point - I imagine towards the end of the run, when there seemed to be something amiss, with some popping / small explosions. This, along with the temperature drop, is what ultimately led me to end the run. Additionally, the alkali has managed to (as feared) eat quite heavily into the fire cement at the base - it did not take much to remove the gauze, so it looks like 2 - 3 hours is about as much as one can get out of a reasonable quantity of the cement used (KOS brand, black). Also, the plating on the cathode seems to have been breached, as evidenced by the rusty marks on the tip, but I have not yet attempted to clean this up to see if it's just iron oxide that's been splashed on the top, or actually from the cathode itself - the plater did warn me that the coat would be thinner there.



With some refitting using the learnings above, this cell could be made sturdier and deliver a significant run. However, given the unpleasant experience that is working with molten alkali, I'm shelving this design for now. Looking at the Downs cell, the first thing that comes to mind is the need for a PSU rewind for higher voltage; 7 to 8 V for an effective process vs. my 5 V used for Castner... Already started working on the cell's melting pot but that's a post for another day

[Edited on 23/9/2012 by Dave Angel]

The shiny ball of sodium metal speaks volumes to the time and effort you invested in your cell. You ran into some problems but you gained experience and you now have a landmark, if you will, for future designs and material selection.

Nice work Dave!

http://www.sciencemadness.org/talk/viewthread.php?tid=16225

^This thread is an idea dump for cleaning up sodium and potassium and sort of outlines my, umm, adventures with reclaiming and cleaning sodium scraps. You mentioned that broken glass helped the smaller sodium beads to coalesce..

Tank

[Edited on 9-24-2012 by m1tanker78]

Quote: Originally posted by Magpie

Your solution to a truly robust design is now the proper selection of materials of construction. And perhaps you feel that some tweaking of the design is needed to improve safety and control.

Quote: Originally posted by m1tanker78

http://www.sciencemadness.org/talk/viewthread.php?tid=16225 ^This thread is an idea dump for cleaning up sodium and potassium and sort of outlines my, umm, adventures with reclaiming and cleaning sodium scraps. You mentioned that broken glass helped the smaller sodium beads to coalesce..

Heh, is that a hint to go add to that thread?

To be honest, I'm not sure if the broken glass is what was doing it. I tried prodding and scratching with a piece of broken ceramic insulating material from the old electric heater and it did nothing. Same with a broken pipette.

When I next make sodium I'll be prepared with some more experiments, but frankly I found that forcing them together (vortex, pliers, pokey-stick etc.) as they approach freezing point is the simplest way to get the beads to coalesce. Perhaps it was cooling at just the time I observed the glass interacting with it, and I drew to hasty a conclusion. I'll blame the late hour and the cold!

Dave, I found some pics of my early experiments after I abandoned NaOH and switched to NaCl. I blacked out the surroundings to mitigate distractions.

This was my first 'improvement' after much experimenting with different salt mixtures all in an open crucible. Indeed, sodium collected in the tube but presented considerable difficulty to remove. I could tell when the tube was filled to capacity when I'd see 'fireflies' on the melt. The only way to collect the sodium was to shut down the power supply and wait until it all cooled to ~ RT. The block was broken and sodium scooped out of the tube. Assuming all went as planned, yields were usually between 15g and 23g. BTW, what you see here is the pre-heat cycle to get everything up to temp (beyond, actually).



This is simply a slightly scaled-up version of the above design. Same difficulties too! I ran into some problems with the larger parts sinking too much heat away but got around that by pouring more salt on top of the melt once everything was positioned. This served to insulate and exclude air from the melt. The anode is the hottest part of the cell so it never really 'seals' - good thing too! This is the cell during normal operation...




As you can see, it's all very parts bin-friendly. At the time of these pics I was using a DC welder for electrons. I'm now the proud owner of a beefy programmable TCR power supply.

Tank

Hey Tank, I like the parts bin use, have enjoyed playing 'spot the part' with your set up - it looks like a case from a D-cell battery on the top of the tube in the first picture, though I'm not sure as the Dr Pepper can in the second would put the D-cell at a larger scale than it should be, unless perspective is playing tricks on me...

I'm actually using a galvanised steel bucket for my cell insulation container. It looks like you're not using any external heating before or during electrolysis - is it melted and kept that way simply by the anode-cathode current? Anyway, I'm rapidly running out of money for builds at the moment due to my habits of buying specific parts, but hope to make some steady progress over the next couple of months.

By the way, it's an absolute travesty: I've identified that pretty much any molten salt physical data report is available from the NIST website (via Google) except those volumes reporting on single salts and on binary mixtures of fluorides and those of chlorides! Would be nice to have the whole collection, and the data most relevant to home chemists - molten chloride mix conductivities would help plan the cell voltage and electrode spacing etc.

Maybe it's a conspiracy against us

Quote: Originally posted by Dave Angel

it looks like a case from a D-cell battery on the top of the tube in the first picture

Quote: Originally posted by Dave Angel

though I'm not sure as the Dr Pepper can in the second would put the D-cell at a larger scale than it should be, unless perspective is playing tricks on me...

Quote: Originally posted by Dave Angel

It looks like you're not using any external heating before or during electrolysis - is it melted and kept that way simply by the anode-cathode current?

Quote: Originally posted by Dave Angel

[...]molten chloride mix conductivities would help plan the cell voltage and electrode spacing etc.

Quote: Originally posted by m1tanker78

I never got around to trying any fluoride salts in my cells. Chlorine is bad enough! Using calcium chloride has always caused me problems so I stay away from the stuff.

Quote: Originally posted by m1tanker78

Plan to pump ~35V through your cell in order to reach and maintain the ideal temperature. I had no problems adjusting the current on the fly with my cheapo inverter welder.

Status:

I'm currently working on an Mk. II spot welder for this build at the moment and will post that in an appropriate thread once done. This one doesn't involve radioactive materials, mostly just honest-to-god copper And duct tape, naturally!

Power supply rewind is putting out over 8 volts, so we'll see if that's sufficient - fitting voltmeter / ammeter so I can diagnose the cell properly this time.

A cross-section top-down schematic of the cell I'll be building is shown below:



The central cathode is mild steel, 12 mm dia and the outer anodes are graphite, 5 mm dia., qty 24. The length over which they are in opposition will be ca. 90 mm, with a gap of ca. 18 mm between them. 100 mesh SS gauze will be placed at 12 mm from the cathode, anode distance from gauze therefore 6 mm.


Mathematical aside (comments from electronics/maths experts would be helpful!):

Consider a simplification - say the anodes are 5 by 90 mm flat plates facing 5 by 90 mm flat plates on the cathode at a distance of 18 mm, then this gives a 1.8 cm long resistor of cross-sectional area 4.5 cm², of which we have 24.

Using the inverse of conductivity (taken for NaCl @ 1090 K, see below) to give resistivity and then putting this into the relevant equation:

Resistance = Resistivity x (length / area) = (1/3.663) x (1.8 / 4.5) = 0.11 ohm

Now, we have 24 of these identical in parallel effectively, so:

R = R/24 = 0.0046 ohm

This seems awfully low to me - maybe that's just the case, but my maths could quite easily be wrong, or it could be an over-simplification - there's only limited cathode surface to go round. Perhaps each 'resistor' has to be considered a wedge rather than a 1.8 cm long plate of 4.5 cm² X-sec area, and that's more maths than I fancy at this time of night.

I'm really just looking for a ballpark figure - the only way to know for sure will be to fire the finished cell up and see what the resistance across it is.


Data:

I've dug out the binary chloride salt reference which happens to have the 100 mol % NaCl data as part of the NaCl-CaCl₂ data. Some examples posted to save on awkward attachments, J. Phys. Chem. Ref. Data 4 871 (1975):

Quote: Originally posted by m1tanker78

Let the adventure continue...

Quote: Originally posted by tetrahedron

Rsingle = integral01.8cm resistivity / {9cm * [x * 0.5cm / 1.8cm + (1 - x / 1.8cm) * (1.2cm * pi / 24)]} dx =~ 0.5058 * resistivity * cm-1 =~ 0.1381ohm

Quote: Originally posted by 12AX7

Keep in mind also the resistivity of metals generally goes way up with temperature. Copper has a bad enough tempco that, by the time it's melting, it looks like iron.

National Bureau of Standards publication 260-90 gives some data for SS up to just over 1000 K; ca. 120 x10^-8 ohm m at what appears to be 1100 K - this from page 42:



My trusty 'Nuffield Advanced Science Book of Data' states (page 133) that iron has a resistivity of 122 x10^-8 ohm m at 1473 K. I can't find any data for mild steel, but even its resistivity is double these figures or even, anomalously, say 2 to 3 powers of 10 higher, I'm not particularly worried.

Quote: Originally posted by 12AX7

How about graphite's chlorine overpotential (if any)

I found a snip from Google Books - Industrial Electrochemistry (D. Pletcher, F.C. Walsh), page 178:



It refers to the chlor-alkali process rather than molten salt electrolysis, but states a Cl₂-graphite overpotential of 500 mV, which points us in the right direction. The graphite-chlorine overpotentials here are shown to drop in going from 25 to 70 °C, so we might assume a greater beneficial result at significantly elevated temperatures - relevant image from the site attached to save everyone a tab:




Given the data gathered so far, I'm fairly confident that a little over 8 V across the cell as specced should suffice. A major area that we have no immediate data on is the reduction potentials at elevated temperatures - let's hope it's a favourable effect as you've described Tim, until data is found - or generated

Before you all get too carried away with the math...

Remember that chlorine, the product generated at the anode, is a gas and as such, impedes the reliable flow of current. Unfortunately, molten NaCl whose temperature range falls anywhere above NaCl MP and below Na BP is considerably viscous. I suppose this isn't such a big problem if the bath is heated externally in that the NaCl bath will remain molten regardless of the current density, etc.

In any case, once a certain anode current density is exceeded, a fearsome hiss will ensue and efficiency will go through the floor. This is actually an equilibrium between chlorine gas generation (and NaCl displacement from the anode) and the electrolyte flowing back to the anode (as chlorine bubbles off). The result is that the tip(s) of the anode(s) will carry more average current and will erode much faster than the superior part of the anode(s).

When the current density is decreased below that bitter sweet spot, I get 'burbling' at the anode. Sodium and chlorine are red/ox'ed rapidly. This is the Downs Cell sweet spot as far as I'm concerned.

Sorry, I got a bit off topic. Anyway, how does one account for gaseous product(s) when planning out electrode configuration, resistances, etc?? Surely the viscosity of the electrolyte must be taken into consideration??

Tank

What's "considerably viscous"? My experience with molten salt has been a very watery consistency. Of course, I'm not about to mix water with it for a side-by-side test!

Quote: Originally posted by 12AX7

...Of course, I'm not about to mix water with it for a side-by-side test!

Recently I did some experiments with molten salts, using my new electric furnace. This was basically just me "playing around"
with various salt mixtures, so sorry about being short on pictures this time around. I wanted to report these results, however,
because I thought some of them were interesting.

Using the information from this phase diagram here:


I made a 50/50 mole% Li2CO3/Na2CO3 melt of about 17 grams total. The crucible was mild steel. I actually just wanted to
watch it melt, but once it did I decided to try electrolyzing it to see what would happen.

I figured that no alkali metals would form at the cathode, because for one thing, lithium is so reactive at those temperatures
that it would immediately react with the electrolyte if it was formed at all. So I used a copper wire as a cathode, and a graphite
rod as an anode, and electrolyzed the melt at 0.5-1.0A.

After running the cell for ten minutes or so, the power was shut off and the electrodes removed. The anode showed noticeable
erosion, but it was a "clean" erosion; it wasn't disintegrating or spalling, but wearing smoothly. The cathode was encased in a
hard, black, material that looked like carbon sponge. It was quite hard, but voluminous considering the amount of energy
that had passed through the cell. Upon submerging this cooled material in water, no gas evolution was noted. The black material
was then crushed underwater with pliers, and still no gas was produced.

Just for fun, I decided to add NaCl to the melt. After doing so, the molar composition of the molten bath was split evenly three
ways between NaCl, Li2CO3, and Na2CO3. Upon applying the same amount of current as before for about the same amount
of time, I noticed small "fireflys" appearing around the cathode. Tiny little orange flames would appear from time to time on
the surface of the electrolyte. No chlorine production was noted, although there was quite a bit of gas coming out from around
the anode. When the cathode was removed, the same hard black solid was present around the wire, but this time it reacted with
water, vigorously gassing for a couple of seconds. Every time the gas evolution slowed down, it could be restored by slightly
crushing the solid with pliers underwater. One time I had trouble breaking open a particularly hard piece, so I lifted it out of
water to crush it. Upon breaking open, the moist solid erupted into a 2cm yellow fireball, with a whisp of white smoke curling
away.

I tried the same experiment again, but this time with a copper anode that I had electroplated with nickel. The nickel was
suspended in the furnace exposed to the air for about an hour before it was submerged in the electrolyte. The results
at the cathode were the same. I was hoping that the anode erosion would be less, but it was inconclusive this time. The nickel
plating appearing to be loosening up, and I probably need to use a solid piece of nickel for this.

Afterward I was hunting around online to see if I could find some information that would corroborate what I had just done.
I think this is a relevant resource here, complete with pictures:

Attachment: 1209.3512.pdf (636kB)
This file has been downloaded 1485 times

In one case the electrolysis of molten Na2CO3/K2CO3 was noted as not leaving a coating of carbon on the cathode. Upon
adding Li2CO3, or CaCO3, a hard carbon deposit would begin to form at the cathode. As far as I know, no mention was made of
the presence or absence of alkali metals formed in any of their experiments.

As a result of my few tests, I'm not sure of the temperature at which Sodium reacts with a carbonate, but it appears that it can
coexist in the molten carbonate bath at about 650C....or at least it can when it is co-deposited into a solid conglomeration with
carbon.

In a way this is what I'm going for: Intercalating (or alloying) alkali metals under cell conditions into a solid at the cathode.
This solid in turn will (hopefully) have similar reactivity to the pure alkali metal itself, while making it easier to collect the product.
The carbonate has a much lower decomposition potential at high temperatures than the equivalent chloride does. The oxide
requires even less. So really, no chlorine is formed unless one blasts the snot out of the anode.

The next experiment will probably start from this system:



with Li2CO3 added incrementally until I begin obtaining an adherent deposit of carbon at the cathode. The current density was
probably too high for that little piece of wire last time, so I'll adjust it downward until sodium stops floating to the top of the bath.

Over time more carbonate will have to be added. In a way, CO2 is being co-electrolyzed using lithium as a catalyst
of sorts. According to the reference mentioned previously CaCO3 can be used to add CO2 back into the system, as CaCO3
is soluble in the melt, but the resulting CaO precipitates out.

Now, for a bit of fun with calcium I picked up some fluorspar (CaF2) from the pottery shop, and some CaCl2 from the grocery
store. The minimum melting point of this system is about 650C, which is much better than pure CaCl2 alone (772C). At 650C it
turns out that CaO is quite soluble in this melt, so this should allow me to avoid halogen gas production, as well as to use
a nickel/nickel oxide anode. Calcium metal is a solid at those temperatures, so its collection--especially at the lower melt
temperature--shouldn't be too difficult. I think at the melting point of CaCl2, calcium dissolves quite readily into the molten salt.
Anyway, if i get around to that, I'll post it in one of the "calcium" threads.

Quote: Originally posted by WGTR

The carbonate has a much lower decomposition potential at high temperatures than the equivalent chloride does. The oxide requires even less. So really, no chlorine is formed unless one blasts the snot out of the anode. The next experiment will probably start from this system: with Li2CO3 added incrementally until I begin obtaining an adherent deposit of carbon at the cathode. [...] Over time more carbonate will have to be added. In a way, CO2 is being co-electrolyzed using lithium as a catalyst of sorts. According to the reference mentioned previously CaCO3 can be used to add CO2 back into the system, as CaCO3 is soluble in the melt, but the resulting CaO precipitates out. Now, for a bit of fun with calcium I picked up some fluorspar (CaF2) from the pottery shop, and some CaCl2 from the grocery store. The minimum melting point of this system is about 650C, which is much better than pure CaCl2 alone (772C). At 650C it turns out that CaO is quite soluble in this melt, so this should allow me to avoid halogen gas production, as well as to use a nickel/nickel oxide anode. Calcium metal is a solid at those temperatures, so its collection--especially at the lower melt temperature--shouldn't be too difficult. I think at the melting point of CaCl2, calcium dissolves quite readily into the molten salt. Anyway, if i get around to that, I'll post it in one of the "calcium" threads.

Perhaps a 'molten salt experiments' thread should be opened?

If you want to recover your alkali metal(s) then forget about trying to form oxides. Oxide of any alkali/alkaline earth metal will spoil your experiment (the melt will freeze). With that in mind, carbonates are to be avoided as much as possible because they eventually decompose to oxide and greatly increase the needed heat input/current to keep the electrolyte molten. I've melted(!) a few carbon steel and stainless steel cathodes trying to 'fight the freeze' in the past.

Also, beware that a molten salt electrolyte that employs carbonate does produce some CO. I imagine highly toxic nickel carbonyl could easily form if one uses a nickel cathode under certain conditions.

A few 'typical' observations I made when sodium carbonate was included in the recipe:

One interesting side reaction I've observed is the formation of Fe3(CO)12 in a certain temperature range while the solid electrolyte is cooling. IIRC, I posted a pic in the 'pretty pictures' thread.

Warm sodium puts off an eerie blue flame in a CO2 atmosphere (i.e. decomposing sodium carbonate/bicarbonate). Sodium reduces CO2 to CO which in turn burns in air to produce a lovely blue flame. In slightly harsher conditions, sodium will reduce CO2 to carbon. Under very harsh conditions, sodium will reduce CO2 and form diamonds from the resultant carbon. I don't have the equipment to personally confirm that last one but I suppose you could just start with graphite.

Sodium formate??? Affirmative.

Sodium hydride??? Maybe. Hard to prove and impossible for me to isolate from other salts. Evidence gives a nod but who knows?


A few scars and scares can be summed up with the following:

Safety should go beyond potential thermal burns and mass ejections of molten salt and alkali metal (been there!) I strongly recommend doubling up on eye protection. Goggles AND face mask should be worn even with small quantities. Some conditions favor emissions of toxic gasses and/or corrosive dust.

I can dig up some references if anyone is interested..

Tank

If considering chlorides, I suggest KCl, as it has a lower melting point, and potassium is more desired in my opinion.

NaK alloy production; shielding with wax

30g NaOH and 30g KOH were melted together in a 100 mL Ni crucible, then ~10A of current were passed through (voltage controlled with variac and rectifier to keep current more or less constant). Unfortunately, the metal burned off as soon as it formed.

Then, I realized that, because the temperature wasn't too high, I could melt paraffin wax on top of the hydroxide to exclude air. In this way, I obtained a pea-sized amount of NaK alloy, liquid at room temperature.



[Edited on 12-7-2015 by Cheddite Cheese]

Nice focus on your arm there, CC!

Quote: Originally posted by Cheddite Cheese

Then, I realized that, because the temperature wasn't too high, I could melt paraffin wax on top of the hydroxide to exclude air. In this way, I obtained a pea-sized amount of NaK alloy, liquid at room temperature.

Caution: wax is flammable!

I was doing another run, trying to make a good video (so I had my camera with me). A little bit of NaK must have come into contact with air despite the wax. It set the wax, which was heated past its flash point, on fire. I put it out by covering up the cell with a metal lid and waiting a few minutes for it to cool down.

A couple of questions, if I may,

Of late, my father decided for some unfathomable reason to order some carbon crucibles. Why? I really don't know. Said he just wanted to play around with them, only, he's shown not the slightest sign of doing anything of the sort. Actually my suspicion is that he bought them expecting that I'd be the one using them (which he gave pernission to do) having realized that currently (pun intended) one of my latest projects is having yet another crack (or quite a lot of cracks, would be more like it, one after another after another after another, accompanied by plenty of flying molten sodium metal)

I have currently a power supply that will deliver either 14A@15V or 35A@5v and I've managed to make some Na and Na/Ca alloy with this, primarily using the higher amperage terminal of the power supply.

I've been concentrating on hydroxides as the melt, because of the reactivity issues with Cl2 salts, although adding a small amount of anhydrous CaCl2 to lower the melting point has been of use. And of course there is the issue of the intermetallic compound formed by fused hydroxide dissolving Na into it.

Now, the crucibles my old man ordered are of the black carbon type material, amorphous carbon AFAIK, I invested myself in some more crucibles of various kinds, steel/iron ones for plating in chromium or in nickel, I did treat myself to an all nickel crucible with lid through which a short length of glass or ceramic tubing will be inserted through the top as an insulating ring, not contacting the melt itself directly, but serving to merely insulate the incoming electrode wire(s) from the metal crucible body and lid.

The ones my old man bought are of the black, dense, quite hard carbon allotrope/polymorph, whilst one of the carbon ones I myself ordered is actual graphite, softer, silvery and unlike the black carbon type, which are reasonably difficult although not challengingly so to scratch with metal implements, the graphite ones are much softer, although the one that arrived today is perhaps a quarter inch thick.

Metal electrodes being used are either copper, or made out of a nickel alloy, the kind used for making guitar frets, since there is a large bucket chock full of fret-wire offcuts about 2'' long, and this alloy, whatever the composition, seems to stand up to fused NaOH very well, even when a blob of caustic is allowed to set on the end of one and that end, holding the other in pliers, is subjected to a roaring flame from a propane torch.

And whilst the black solid carbon (same stuff as is in battery carbon rods) crucibles seems to stand up to a fair lot of abuse, is there a significant difference in the resistance of this allotrope of carbon and of actual bona fide graphite per se, the true, softer (against indentation/scratching mechanically with hard metal or carbide tipped cutterss etc.) to fused alkali hydroxide melts at an equal degree of heat and time course, equal current/voltage and other variables, the variable changing being solely that of graphite vs black (non-vitreous) carbon?

And lastly, I seem to remember that a melt of NaOH and KOH, when fused, acts as an eutectic and in the right proportions, can result in the lowering of melting point of the mixed melt bath to as little as 200 degrees 'C.

Going to have a shot at preparing some NaK tonight now that today, Mr.Postie dropped off not only that large, thick graphite crucible but a nicely weighty sack of KOH, a few kg at least

What is the ideal eutectic proportion in an NaOH/KOH melt? and can it be further lowered with addition of a little anhydrous carbonate or bicarbonate?

Lastly, I have also potassium hydrogen phosphate. Two questions I have regarding this. Has it any role it may serve in the formulation of molten salt eutectics, and also, whether fused, or in aqueous solution, be reduced to phosphorous acid or hypophosphite salts? or for that matter, H3PO4 to these acids as is, without presence of counterion, since the presence of the hydrogen ion species should be sufficient to permit conductivity.

Making sodium metal from sodium hidroxide

This process uses electrolysis as a means of synthesizing sodium metal. . . Using a heat-resistant vessel that does not conduct electricity, place sodium hydroxide in it and heat it to 320 degrees Celsius. When it reaches such temperature the compound will be liquid and therein place the positive and negative poles of the electrolysis cell and begin the process Of decomposition of the element, soon after a few hours will see white balls in one of the poles, this is the metal sodium, remove it from the container and put in a bowl with mineral oil and store, hopefully have helped. . . You may be confused by what I said, well I do not speak fluent English.

Congratulations! You just described the Castner process.
Although in practise it's a little more difficult.

[Edited on 15-7-2017 by Jstuyfzand]

Making sodium metal from sodium chloride

Quote: Originally posted by metalresearcher

Add CaCl2 to the NaCl and the mp will lower to 640 C.

Quote: Originally posted by Jstuyfzand

Quote: Originally posted by metalresearcher   Add CaCl2 to the NaCl and the mp will lower to 640 C. I am aware of that, but 640 is still a bit high and unpractical. A eutectic made with NaOH would be sub 330°C, so I was wondering whether a NaOH NaCl mixture would reach these lower temperatures.