For the most part fluorous domains used in fluorous biphasic systems have been linear perfluoroalkyls, such as perfluorooctyl moieties. Branched perfluoralkyls such as perfluorinated t-butyl groups have also been used as tags to facilitate fluorous based separations. In the 2004 “Handbook of Fluorous Chemistry”, the authors define a fluorous tag as the “portion of domain of a colecule that is rich in sp3 carbon-fluorine bonds and exerts primary control over the separability characgteristics of the molcule in fluorous separation techniques.” So within that definition does lie moieties that are not necessarily perfluoroalkyls. These would include perfluoropolyethers, generally in the form of perfluorinated oligoethyleneglycols.
At FTI we have in the past done some cursory examination into the fluorous separation behavior of perfluoropolyethers as fluorous tags and compared them to perfluoroalkyl tags. What we found, if I recall correctly, was that the oxygens basically functioned much like a CF2, so that a CF2OCF2CF2OCF3 tag had a similar F-HPLC retention time as a C6F13. This was somewhat surprising since we always used fluorine count as a first approximation of fluorophilicity. What the F-HPLC experiments told us, however, is that perhaps we should consider the size of the fluorous domain instead, in this case a 6 atom chain that is “rich in sp3 carbon-fluorine bonds”.
A just available paper from Kvicala et al at the Institute of Chemical Technology in Prague attempts to explain the fluorous behavior of perfluoropolyethers, at least as tags on NHC ligands in silver complexes. The authors first prepared a series of substituted NHC ligands, either with traditional perfluoroalkyl tags or with perfluoropolyether tags. They then formed the silver salts of these ligands. Fluorous partition coefficients were then measured for the ligands and the complexes by partitioning between toluene and perfluoromethylcyclohexane. The results were fairly striking as seen in the table below.
Compounds 9a-h were all perfluoroalkyl tagged compounds which essentially had fairly equivalent partition coefficients, something in the single digits. All of these compounds had a total of 12-16 CF2 or CF3 groups. Compounds 12b, c, and f, however, were polyfluoroether tagged and possessed dramatically higher partition coefficients. 12b contains only 8 perfluorinated carbons, 12c has 12, and 12f has 10, so the overall fluorine content for compounds 12 is less than 9 yet they are more fluorophilic. So what’s going on? Well, if you consider the FTI results mentioned earlier and regard the oxygens in the perfluoropolyether as CF2′s than you get a pseudo-fluorous count for 12b, c, and f of 12, 18, and 15. That’s now in the range of compounds 9, but still doesn’t account for the greater fluorous partition coefficient.
The authors then turned to computational chemistry. DFT calculations were conducted on both the NHC ligands and the imidazolium salts. What these calculations showed was that the most stable conformation of the perfluoroalkyl chain had the chain directed straight out from the imidizole ring and beasically planar to it. The perfluoropolyether chains on the other hand were twisted at the carbon-oxygen bond resulting in the chains being out of plane with the heterocyclic ring. The authors hypothesize that fluorophobic ionic center of the imidizolium salts and the silver complexes are more shielded with the perfluoropolyether chains resulting in a significantly more fluorophilic compounds. It’s certainly a reasonable hypothesis, particularly when one considers that fluorophilicity is based on solvophobicity more than on any attractive forces.
It’s an interesting paper which shows how structural elements can have profound effects on fluorophilicity. Something we should certainly keep in mind more often.