More Information about using Inorganic Salt Complexes to simply Purify Alcohols at Scale

kilomentor | 16 May, 2011 08:19

Barry Sharpless et al. were the first to teach the separation of alcohol mixtures and alcohols from non-alcohols using calcium and manganese salt complexes. The original J. Org. Chem. paper [J. Org. Chem. 40(9) 1252=-1257 (1975)] unluckily recommended as a most preferred modality, using one mole of anhydrous calcium chloride per two moles of alcohol. This turns out to have been a good pressumption based on theory, but not such a good practical recipe. Soon after Sharpless’s publication, Cook et al. [Biochemical and Biophysical Research Commun. 68(1) 143-148 (1976)] showed in a crystal structure that 2:1 alcohol:calcium would be theoretically expected. Mentally extrapolating from what the Cook paper says about the probable structures, strings of calcium chloride ions are interspersed by layers of hydrophobic organic substructures where the thickness of these hydrophobic layers is somewhat related to the length and shape of the hydrocarbon substructures. How fast and how purely these clay-like layers form probably depends upon how rapidly the alcohols get into the layers and how quickly they sort themselves out between layers. This is however conjecture. In any case, in the light of more experience, this 2:1 stoichiometry recommendation turned out to be poor advice, Professor Sharpless has told me. Large excesses of calcium chloride, generally, work best.

It is not just some alcohols that complex inorganic salts such as calcium chloride, calcium bromide, manganese chloride, lithium bromide etc. ; all alcohols do but there are differences in affinities and separation is only possible where precipitate/crystallize occurs. This formation of solid complexes by stirring the substrate mixture, and the inorganic salts together with an appropriate organic liquid (most often hexane) depending upon the specic situation may be driven by relative rates or thermodynamics.

Alcohols are not the only functional groups that form complexes with such inorganic salts, many other functional groups do, but not as strongly and they don’t often precipitate. The interaction between these other functional groups do cause the inorganic salts to dissolve as some sort of soluble complexes or miscelles but precipitation/crystallization is not typically observed.

Whether an insoluble complex forms depends also on the character of the salt chosen to combine with it. L-menthol for example formed complexes with calcium bromide and manganese chloride but apparently not with calcium chloride.

Lower alcohols seem to enhance the rates of complexation perhaps by getting the inorganic salts dissolved as complexes in the organic liquid although there must be something more than this because ethanol catalyst was used in the exchanges of Table 1 in the J. Org. Chem. paper even though the starting complex would be somewhat soluble in the hexane liquid. Yet still catalysis is found to be useful.

Based on this preliminary understanding, certain questions that affect the ruggedness and the practical breadth of the method come to mind.

Are 1,2- or 1,3-diols very strongly and dependably complexed and precipitated? It would be useful to have an almost generally useful method to drop out such compounds from a reaction mixture.

Will the presence of other complexed functional groups in the compounds to be separated tend to screw up such a complexing/ precipitating scheme?

For example, the geraniol/citronellol mixture works since the other functional groupt here (alkene) does not form complexes; but what if both molecules contained alcohols and say ketones? Ketones do form complexes but the complexes don’t seem to precipitate (although cholestenone does form an isolable complex with calcium bromide). Actually, not all ketones appear to form complexes. The solvents methyl isobutyl ketone (MIBK) and methyl amyl ketone can be used as the liquid phases for forming complexes and so must participate very poorly (or it would swamp the effect).
Other examples of complexes between inorganic salts and organic compounds containing an alcohol group were known. In BE1555968 reference is made to isolating sterols by heating in hydrocarbon solvents with 14-16 molar equivalents of anhydrous zinc chloride and isolating the solid. This is referenced to in DE827199. Also in BE1164769, there is described a process for isolating sterols from mixtures by dissolving the mixture in a hydrocarbon, adding to the solution an aqueous solution of a metal salt suitable for complex formation, removing the water to a great extent by azeotropic distillation, and after cooling the mixture, isolating and splitting up the precipitated adduct in the usual way.

In the patent BR1555968, 3-hydroxy steroids and 3 oxo-steroids in various natural oil mixtures were separated by treatment with calcium bromide dehydrate in MIBK or n-amyl methyl ketone. It is unclear why the 3-oxo-steroids gave complexes under these conditions since all other literature suggests that ketones form weak complexes and/or these complexes are soluble and do not precipitate.

Kilomentor thanks Professor Sharpless for answering questions about this technique and for directing me to the Cook et al. paper.