Let's break down some sodium chloride. Adding a few volts, chlorine is evolved at the anode and sodium at the cathode. Both initially react with the H2O solvent, the sodium liberating hydrogen, making the vicinity of the cathode very basic; while the chlorine dissolves and partially dissociates to hydrochloric and hypochlorous acids (Cl2 + H2O ↔ 2H+ + Cl- + OCl-). As the anolyte and catholyte diffuse together, bleach is formed. The chloride cycles back, making more hypochlorite. If you raise the cell temperature, hypochlorite self-oxidizes to chlorate, ClO3-, yielding still more chloride. Now this can all be done without voltage by bubbling chlorine gas through a hot soda solution, but 5/6ths of the soda remains as chloride - yields are terrible! If you decompose that chloride with voltage, you can acheieve 90% yields (more than 90% isn't possible because the chlorate begins to oxidize to perchlorate, which also destroys graphite anodes quite quickly!). The actual chemical process is a bit more involved than this but I'll let someone else handle that. ;-) That website is also a good resource, perhaps the best. That said, I won't cover any more, I'll just get to my own cell.
This is my latest cell. For electrodes, I purchased three 18" long, 1½" square graphite bars from a company in Oregon, over e-Bay, for $30 shipping included. After some dusty, blackening(!) hacksawing, I had the eight slabs seen here. Add 5V from an old computer power supply and we're rollin'. I don't know efficiency and I wasn't able to measure current draw; my guess is 10A at maybe 60-80% efficiency. In a week's run time I produced maybe 600 grams total, not much. At the 30A the anodes are rated for (given 300mA/cm^2 I think), it should've made several kilos. Most of the time the cell was draped over with a garbage bag, to control the mist. The hydrogen bubbles make a horrible mess of salt spray if you don't; after my first run there was a thin crust of salt on everything within twenty feet and the power supply, well let's not go there! I had to soak it out with distilled water, although it only went so many more days before just pooping out altogether.
After a run, I boiled the solution (somewhat anyway) and added potassium chloride (from Wal-Mart, they sell it 50lbs sack for $6 for sodium-free water softening, to be used with the zeolite type ion-exchange water softeners), cooled (stuck it in the freezer actually) and filtered the crystals. Then I put the yield in a bucket of ice water, to wash the chloride salts out, and filtered again. Good yield from washing. After drying, I ground the cake for use with uh, whatever. The filter/drying paper ends up nice and flammable, in fact you can strike it like a match (due to the same reaction as normal matches, which have KClO3 in them) on the red strike pad of a matchbook.
A small test batch of sucrose + KClO3 was prepared, 40/60 I think is the CO-stoichiometry. On ignition it burned heartily with a purple flame, indicative of potassium; no sign of sodium whatsoever. (FYI, this mixture, and many others using chlorates, are ignitable with the magic touch of a stick wetted with concentrated sulfuric acid.) Apparently it is very hard to remove sodium from some compounds (enough that the sodium doesn't color the flame), but after just one washing I don't see a problem. Odd?
(6/8/06) Ah yes, a new year, another run of the cell. Here's my most recent crop, from running the above cell for nearly two weeks on a 4V bench supply. The total yield was somewhere near a pound (that would be as KClO3), rather pitiful.
This time I went all out. I started with a kilo of salt, then after all that electrolysis, I boiled the solution in a Pyrex bowl (which, despite "pyr" in the name, has "microwave use only" printed on it?!) to kill hypochlorite, reduce volume and remove salt, then I cooled the solution to crystallize sodium chlorate, reduced more and so on, then recrystallized the sodium chlorate. It should be pretty good stuff. Not sure what the slight tan color is, seems to be left over from the liquor. Some I later precipitated with potassium chloride.
The precipitate was washed and dried, then ground, dried some more, ground a second time, dried and canned. There may be some moisture left in the product, but I can accept that it is simply so fine that it wants to clump. It really is powdery. Like my barium nitrate, the color is slightly tan due to rust on the grinding surface. Come to think of it, I haven't tested color yet, but it does burn nicely with powdered charcoal.
(4/21/07) New cell running this time, and months ahead of schedule no less! I'm reusing the batches of now quite dried salt from previous runs, and have discovered that I actually made a quite good yield from the last run! Funny story: so, I had a gallon ice cream bucket mostly full of crystallized sodium and potassium chlorides (not many nice crystals, what a shame), plus a bunch of KCl briquettes ("tabs", from a 50lb. sack of Wal-Mart KCl, sold for low-sodium water softening), and some green crud (chrome or iron impurity) and plenty of black smudge (worn graphite that I hadn't disposed of). Well after dissolving the chunks apart, I was left with those tabs. I soon realized that hey, these things are dissolving awfully slowly. In fact, not only were they appearing to dissolve slowly, the centers were dissolving! It turns out the excess chlorate ion in solution reacted at the surface of the briquettes. Boiling this batch of salt with water dissolved out the KClO3 and gave me a good pound yield straight away!
This cell is built for the long haul. I'm not going to set any records with the low surface area, but the bulky electrode will take months to erode away. The cathode/casing is a chunk of stainless steel rectangular tube, 4 x 4", 1/4" thick and about a foot tall. I welded a 1/8" mild steel plate to the bottom (with only two pinhole leaks after the first pass!). To keep the anode from shorting, I cut a bunch of bits of PVC pipe and spread a layer on the bottom to hold it up off the bottom.
The electrode is connected to the supply by a copper strap held in place with a spring clamp. A garbage bag in the annulus between the anode and cathode condenses salt spray and chlorine, keeping odor and dust suprisingly low. The black liquid in front is hot, freshly electrolysed liquor. Thanks to the steel casing, I can boil in situ by pointing my torch (pictured) at the thing and waiting a few hours for the hypochlorite to decompose. I do this under voltage to prevent corrosion, and to, well why stop if I don't need to?
The next step is boiling the liquor with mixed salt (I'm still going through old admixtures here) until fully saturated, then pouring off the liquid and cooling it. So far, the salt has been rich in potassium, giving me a total of several pounds over the last few weeks (since about 3/25, it would seem ...wow, it's been a month already).