Alcohol catalyzed alkali metal production

The alcohol catalyzed alkali metal production is an accessible route to obtaining alkali metals, such as sodium and potassium, by reducing their respective hydroxides in a suitable solvent, with magnesium metal, using a tertiary alcohol as catalyst.

This method was first tested on the now defunct Versuchschemie forum in the early 2000s, and was brought to the attention of the general public by NurdRage in late December 2011, when he isolated potassium metal from KOH, using tetralin as high temperature solvent, and t-amyl alcohol as catalyst. Since then many other amateurs have tried their own version, with various success.

Summary

The general procedure requires the use of an inert high boiling point solvent, alkali metal hydroxide, metallic magnesium flakes or powder and a tertiary alcohol as catalyst. The reaction is done at high temperatures, usually around 200 °C and requires several hours for to obtain a good yield.

Procedure and performance for each alkali metal

Lithium

The reaction does not happen at temperatures below 200 °C. At higher temperatures, it appears to work, but the amount of lithium produced with this method is extremely small.

Sodium

In a ground glass flask (round bottom or flat bottom), a stir bar is added, then 40 g sodium hydroxide flakes and 30 g magnesium turnings/powder are added to the flask. Then, 125 ml mineral oil is added to the mixture. A reflux condenser is connected to the flask, and the flask is heated to 230 °C. A tertiary alcohol is used as catalyst, only 1.2 ml are required. The reaction takes 3 days to complete. After completion, the sodium is mixed with magnesium leftovers, and can be separated by melting it in boiling dioxane. The yield of this process is around 30%.

Unlike in the case of potassium, tetralin cannot be used as solvent for this reaction, as sodium rapidly reacts with it. Any hydrocarbon with a double or triple bond that is usually considered inert cannot be used in this reaction.

Potassium

In a flask, 2.5 g of magnesium turnings are added, then 5 g of potassium hydroxide are added in in the flask and the mixture is gently stirred. 20 ml of tetralin is added in the flask, as high boiling point solvent. Tetralin is a good choice, since it's denser than potassium metal, so the resulting potassium will flow on it, allowing better separation. A condenser is connected to the flask. The flask is heated to around 200 °C, and 1.2 ml of tert-amyl alcohol are added in 0.4 ml portions. You can also add the alcohol in smaller portions, multiple times. After a few hours, small spheres of potassium metal appear, and over the course of several hours, more potassium will appear. Slowly, the small spheres will coalesce into larger spheres.

Mineral oil can also be used, and unlike tetralin, it is completely inert to sodium (as long as it's aromatic-free). After one hour of heating the mixture in mineral oil, potassium metal begins to appear as small spheres. 2 hours later, the potassium metal spheres are much bigger. After almost 4 hours since the start, the reaction is completed, and the flask with the potassium metal can be removed from the heating bath.

Rubidium

A similar process to the production of potassium can be employed for rubidium as well, starting from RbOH. This is detailed on the German forum Illumina, which reports a 25% yield, which may be attributed to mishaps during the synthesis. The author states that it is reasonable to assume yields similar to the production of potassium.

Caesium

Not tested so far.

Main challenges

Choice of catalyst

The hypothesis behind this method is that the reaction between alkali metal hydroxide and the tertiary alcohol produces alkoxides that are soluble in the mineral oil or tetralin, and the dissolved alkali alkoxides are reduced by the magnesium metal to sodium metal, and magnesium alkoxide, which further react with the alkali hydroxide to reform the alkali alkoxides, and the process starts again, until the magnesium or hydroxide gets depleted or the alcohol boils off. Since short chain alkoxides are less soluble in organic solvents than longer chain alkoxides, long chain tertiary alcohols are the best for this reaction.

None of the simple primary or secondary alcohols work for this procedure, though highly branched secondary alcohols such as 3,4,5,6,6-pentamethylheptan-2-ol (also known as Kohinool) or 1-(2,2,6-trimethylcyclohexyl)hexan-3-ol (known as Norlimbanol) will work, although the yield is terrible, at 34% and 26% respectively. This would suggest that having an alcohol with a very branched chain or high steric bulk is important for the performance of the alcohol catalyst. This is was partially proven by using sterically hindered and sterically bulk secondary alcohols, such as menthol and borneol, which work just as good as tertiary alcohol, giving yields of 91% and 87% respectively. This which would suggest that using an alcohol possessing bulky molecule and steric hindrance is critical for the reaction to work. Interestingly enough, the reaction time is very different: for menthol, the reaction took 30 hours, while for borneol it took only 10 hours, even though menthol gave a slightly better yield than borneol. So there needs to be a "sweet spot" in the alcohol's structure, where it has to be bulk enough to limit steric effects but not too bulk to interfere with the reaction mechanism. Different isomers of bulky alcohols don't have any effect of the reaction time and yield.

Carboxylic acids, regardless of their chain length, do not work at all.

Tertiary alcohols lacking a double or triple bond are the only ones that give good yield.

A good cheap catalyst is 4-terpinol, which is the main component of tea tree oil, at 40 % concentration. Raw tea tree oil will work, but the yield is crap, and the resulting sodium metal appears as very small particles. If the 4-terpinol is distilled from the mixture and used as purer compound, the yield is much better, at 94%, very close to that of the method using 3-ethyl-3-pentanol, which had an yield of 96%. Injecting larger amounts of 4-terpinol seems to lower the yield to around 81%, probably due to impurities and side products. However, due to its alkene bond in its structure, 4-terpinol will slowly break down in contact with sodium metal after 24 hours, and a second addition of 4-terpinol is required. Addition of lithium metal as jump start will also improve the yield, up to 90%.

Here are a list of various chemicals tested by amateur chemists as catalysts and the result:

• Primary alcohols: none work; or at least not simple alcohols without steric bulk and hindrance.
• Secondary alcohols: no simple and short chain alcohols worked, nor simple cyclic alcohols like cyclohexanol; however menthol and borneol work very well, while other secondary alcohols, such as 3,4,5,6,6-pentamethylheptan-2-ol (Kohinool) work, but have lower yields.
• Tertiary alcohols: all tertiary alcohols with saturated chain work; tertiary alcohols with at least one double bond work, but they will slowly break down; unrefined patchouli (which contains patchoulol) however, does not seem to work at all. Tetrahydrolinalool, which is made by hydrogenating linalool (lavander oil) will work as well, giving an yield of 95%.
• Carboxylic acids: none work
• Phenols: none work

Glassware destruction and water interference

Sodium hydroxide in the presence of water will etch glass at high temperatures. To solve this problem, a small amount of alkali metal is added to a mixture of sodium hydroxide in mineral oil, to remove the water. Adding sodium is not always needed, you can also add lithium metal from a Li battery. Adding magnesium metal is possible, but this wastes the more valuable magnesium metal.

The small amount of water present in the NaOH and KOH will also lower the yield, as well as slow down the speed of the reaction, as the magnesium metal first has to destroy the water. To solve this problem, a small amount of alkali metal (lithium, sodium, potassium) is added to remove the water, and jump start the reaction. This method speeds up the process, as molten sodium removes water faster than magnesium does, and as a bonus, can increase the yield to above 90% (NurdRage obtained 96%), while also solving the glassware destruction problem, as NaOH and KOH will only attack glass at high temperatures in the presence of water.

Lithium hydroxide does not damage the glassware visibly.

Reaction solvent

Mineral oil is the best choice as reaction solvent, due to its high boiling point and inertness to most reagents. Mineral oil will not react with alkali metals, though if it's impure or has high humidity, the alkali metal will react with said impurities and water. Tetralin can also be used, and works well with potassium, but will slowly react with alkali metals, and while the reaction with potassium metal is very slow, the reaction with sodium metal is fast enough that is poisons the reaction. As such, tetralin is not a good solvent for this method. Heavy tertiary alcohols, like 7-Hexyl-7-tridecanol have been successfully used as both reaction solvent and catalyst. Other similar alcohols can be used the same way.

Eucalyptol (1,8-cineole) has also been tried as reaction solvent, due to its high boiling point of 176 °C. Unfortunately, eucalyptol will slowly react with sodium metal, possibly by cleaving the ether bond in its molecule, though the reaction does not produce tar, and the sodium remains shiny, unlike in the case of tetralin, which suggests that a different reaction happens here, possibly a condensation of sorts. This is further confirmed by the fact that this side reaction does not appear to interfere with the sodium production, and when using menthol as catalyst for the sodium production along with sodium jumpstart, the reaction time is almost halved, at 14 hours compared to 30 hours when using mineral oil (though complete hydrogen bubbling only stops after around 24 hours), and the yield of this process is 94%, which is actually better than in the case when mineral oil was used, where it was 91%. This would suggest that the reaction of eucalyptol with metallic sodium is very slow and while the amount impurities obtained appear significant, they're actually very small.

Reaction time

The reaction time decreases the lower you go in the periodic table:

• Lithium: >7 days (if any)
• Sodium: 10-40 hours
• Potassium: 4-8 hours

If the trend is kept, then the reaction time for rubidium and caesium should be even shorter.

Reducing agent

Magnesium is the choice reducing agent for this reaction, and it's used as powder or turnings. Other metals, like aluminium, do not seem to work.

Separation of alkali metal

The alkali metal obtained from the mixture appears as small globules, and separating them from the slag is somewhat problematic, as it's easy to lose them during the separation process. A good method is to slow the stirring over the course of an hour, to allow the small globules to coalesce into larger spheres. The stirring is then stopped and the heat is turned off to allow the molten sodium metal to solidify. To separate the sodium metal from the slag, the reaction product is poured through a sieve or kitchen strainer, and then washed with a hydrocarbon or ether to remove the magnesium oxide impurities. The resulting sodium metal still has some magnesium metal impurities, which can be removed by melting the sodium metal in boiling dioxane.

By refluxing the alkali metals in dioxane, they will coalesce into metallic spheres, separating them from the leftover magnesium metal and other impurities. No other solvent seems to work like dioxane, and it would appear that dioxane can separate the alkali metals due to its cyclic structure, which has chelating effects. However, as seen with eucalyptol, simply being a cyclic ether is not enough to separate the metals, meaning it requires a specific chelating effect.

Stirring

While stirring is not required for potassium, when making sodium metal it's absolutely crucial to stir the mixture, otherwise the reaction will take forever to give any results. Although PTFE stir bars can be used, molten sodium will attack it. A better choice would be glass coated stirrers, which resist molten sodium.

Although it hasn't been tried yet, an iron or a magnetic stainless steel rod could also be used as stirrer, as iron alloys don't react with hot alkali hydroxides or molten alkali metals at such low temperatures.

Catalysts performance in the production of alkali metals

Alcohol used	Lithium	Sodium	Potassium	Rubidium	Caesium
2-Methyl-2-octanol
3-Ethyl-3-pentanol		Reaction time: 30-40 h; Yield: 33%
4-Terpineol		Yield: 94%
7-Hexyl-7-tridecanol		Works both as catalyst and as reaction solvent
Borneol	No significant Li metal if any was produced after 7 days	Reaction time: 10 h; Yield: 87%
Cyclohexanol		Failure
Dimetol		Reaction time: 20 h; Yield: 95%
Eucalyptol
Kohinool		Reaction time: 10 h; Yield: 34%
Menthol		Reaction time: 30 h; Yield: 91%
Norlimbanol		Reaction time: 10 h; Yield: 26%
Patchoulol		Impure patchoulol doesn't work
tert-Amyl alcohol		Reaction time: 15-30 h
tert-Butanol			Reaction time: 30-40 h; Yield:	Confirmed
Tetrahydrolinalool	No significant Li metal if any was produced after 7 days	Reaction time = 20 h; Yield: 95%

See also

• Thermochemical dioxane approach

References

Relevant Sciencemadness threads

• Make Potassium (from versuchschemie.de)
• unconventional sodium