F-Blog

The previous post described some of the efforts in the Bannwarth and Tiller labs regarding the use of fluorous Pd catalysts on solid supports. They are just one of many groups actively looking for novel methods by which to immobilize fluorous catalysts. Two 2008 Organic Letters publications from the Vallribera group at Universitat Autònoma de Barcelona describe their efforts.

The first letter described their work in forming phosphine-free Pd nanoparticles immobilized on fluorous modified silica gel. The researchers first formed a Pd catalyst using fluorous triazine A as the ligand. Essentially this is the same approach that Bannwarth used back in 2002, except that they used fluorous phosphines as the Pd ligand. Vallribera and co-workers demonstrated that they could conduct Heck reactions using the supported catalyst with 0.1% Pd loading and that recycling could be conducted up to about four times without appreciable loss of product. In addition, they analyzed the crude mixtures by ICP-MS and found about 2-7 ppm residual Pd. They also tried using reverse phase silica gel instead of the fluorous silica gel and found that the second run did not work as well, which would seem to support the idea that the Pd nanoparticles are indeed adsorbed onto the fluorous silica gel.

The second paper describes a different approach where they covalently bound the triazine ligand B to the silica gel then form the Pd nanoparticle. This covalently-bound material was prepared since the previous material using the fluorous silica gel was not compatible with microwave heating due to leaching of both Pd and the fluorous ligand. The researchers first had to develop some modified conditions for attaching the ligand to the silica gel, due to the insolubility of the fluorous ligand under conditions normally used, which they successfully accomplished. Incorporation of the stabilizer to the silica gel was confirmed by elemental analysis and solid state NMR. The Pd nanoparticles were then formed and examined by electron microscopy. The catalyst was then used to conduct a microwave-mediated Heck reaction between iodobenzene and butyl acrylate. The catalyst could be recovered by filtration and recycled six times without appreciable loss of conversion. This catalyst system proved to be more stable over the previous noncovalently bound catalyst under microwave conditions. Examination of the crude reaction mixtures also indicated less than 6 ppm residual Pd.

There are still several unanswered questions such as the exact nature of the catalytic species and substrate generality. With the ligand covalently bound to the silica it also raises the question as to whether the fluorous chains are even necessary! Interestingly, an alkyl version of B was prepared using dodecylthiol, but there was no mention of forming and testing Pd nanoparticles using this non-fluorous analog.

It should be interesting to see future studies using these or related catalysts.

The Journal of Fluorine Chemistry has published online an upcoming paper from Profs Jörg Tiller (Universität Dortmund) and Willi Bannwarth (Universtät Freiburg) describing ring closing metathesis using a fluorous Grubbs-Hoveyda catalyst immobilized on an amphiphilic polymer co-network (APCN).

First what is an amphiphilic polymer co-network (APCN)? An APCN is a polymer that contains covalently bound sections of two different phases, i.e. hydrophilic and hydrophobic phases. They are different from coblock polymers and other similarly related materials that may contain elements of the two phases in that the phases are still immiscible as seen by atomic force microscopy. These types of polymers have found utility in a number of appications including tissue engineering, drug delivery, and catalyst supports. One of the characteristics of such materials is that the different phases of the co-network swell to differing degrees depending on the nature of the solvent and that this is a dynamic and reversible process. For example, in water, the hydrophilic portion of APCN would swell while the hydrophilic would collapse and vice versa in a solvent such as hexane. Prof. Tiller has utilized this property to conduct enzymatic reactions in organic solvents using a hydrophobic/hydrophilic APCN. The figure below depicts the concept.

Back in 2006 his group reported the first preparation and examination of APCNs containing a fluorophilic phase and either a hydrophilic or hydrophobic phase. Meanwhile, Prof. Bannwarth published in 2005 the immobilization of a fluorous Grubbs-Hoveyda catalyst (shown to the left) on fluorous modified silica gel. In that work, they conducted the reaction in CH2Cl2, which presumably releases the catalyst from the fluorous silica gel, then performed a solvent switch to 80:20 MeOH:H2O which readsorbs the catalyst onto the silica gel. By now replacing the fluorous silica gel with a fluorous APCN they hope to conduct the same chemistry in an aqueous solvent. So much like the enzyme in the figure above, the fluorous catalyst enters the APCN upon exposure in a fluorophilic solvent and is entrapped there upon drying. Exposure to water then results in swelling of the hydrphobic phase allowing the substrate to react with the catalyst.

The first thing the researchers did was evaluate the catalyst in various aqueous solutions, namely actone/water and DME/water mixtures. They found that the catalyst worked although perhaps not quite as well as the normal Grubbs-Hoveyda catalyst under the same conditions. They then tested the catalyst once immobilized on the fluorous APCN. In this instance it worked best in 100% water at 60 degrees. The catalyst loading was very high and the recycling experiments were disappointing in that the second run provided lower conversions than the first run. Whether this is due to catalyst leaching, catalyst decomposition, change in the APCN is unclear.

It’s also somewhat difficult to guage the effect of the APCN on the overall process, since the exact reaction conditions (60 degrees in water for 2h) were not conducted in the absence of the APCN. This is an important reaction in my opinion since the catalyst contains three fluorous chains and is probably quite insoluble in water at room temperature. Given that in Prof. Bannwarth’s previous report the use of fluorous silica gel or regular silica gel had negligible effect on the amount of residual Ru in the product, it’s possible that the recovery of the catalyst in this instance and in the previous report was entirely due to precipitation of the catalyst and that the support, silica or APCN, does not play an appreciable role in either case.

In any event, while the overall results were not spectacular, the concept and the materials used are very interesting. What remains is to understand the role of the APCN better in order to find ways to improve the catalyst and the process.

In our last email newsletter (link), we detailed our Summer Sorbent Promotion:

All orders of selected FluoroFlash® separation media (HPLC Columns, FSPE Cartridges, FTLC Plates, Bulk silica gel) that are placed before July 31st receive 10% off.  We don’t normally discount these products, so this is a great opportunity to get a good deal.

Additionally,  any reagent order placed with a sorbent order will receive a free coffee mug while supplies last.

For details on how to order, please click here.

A recently available manuscript from Prof. Julien Legros and co-workers at the Univeersité Paris-Sud describes their efforts in making a fluorous version of DMAP for the acylation of alcohols. As nearly every organic chemist knows, DMAP is very effective at catalyzing acylations through the formation of an acylated pyridinium salt which is then attacked by the nucleophile, usually an amine or an alcohol. It’s also generally used in small amounts and can easily be washed away with an acid extraction, so you wouldn’t think separation after the reaction would be that big of a problem. Unfortuantely, that’s not always the case, especially if you have other nitrogen heterocycles in your structure as is often the case. In fact quite a bit of effort has been put into coming up with supported DMAP analogs which are easier to separate. Amongst them are the usual suspects; polymer-supported, silica supported, magnetic beads, etc. Not to be left out, Legros and co-workers decided to try a fluorous based method.

Their initial plan was to take 4-amino-pyridine and alkylate it with RfCH2CH2I. As noted an earlier F-Blog post they may have been better off trying the propyl spaced halide rather than the ethyl spaced due to the possibility of elimination in the ethyl spaced, but in this instance they report observing various alkylated species including alkylation of the pyridine. In any event they decided to try a radical addition approach to the diallyl compound instead to provide a symmetrical f-DMAP. In this reaction the major product observed ended being a radical cyclization product resulting from iodine transfer. Not too surprising in retrospect given how fast radical 5-exo-cyclizations are. Treatment with DBU then afforded the exo-methylene compound A. This is where the researchers decided to make chicken salad.

Even though they didn’t get the molecule they originally wanted, they presumed that A would work given that it had all the necessary elements; a pyridine ring, a tertiary 4-amino substituent, and a fluorous chain. And they were right. Compound A did behave very similarly to DMAP. According to the manuscript they ran a few acylations, essentially without solvent, then added hexanes to partially precipitate A, which the recovered and reused. Unfortunately, the recovery wasn’t very good so the recycling experiments were not very good either. They tried using perfluoromethylcyclohexane or perfluorodecalin, but found that A wasn’t soluble at all and correctly assumed that it was due to the overall polarity of molecule. They concluded the paper by stating that the recovery problem might be solved either by using FSPE or by adding more fluorous chains to the molecule and that efforts in both of these directions are underway.

There is a third option and that’s fluorous solvent tuning. At FTI we recognized some time ago that perfluorocarbons were overall pretty lousy solvents, exactly what Legros and co-workers observed, which is why people always had to load up on the number of fluorous chains. We started looking at hydrofluoroethers as more polar fluorous solvents and found that by tuning both the organic and fluorous layers that one could achieve high levels of separation for fluorous molecules with as little as two fluorous chains. The Mizuno group has applied these principles successfully to fluorous based oligosaccharide synthesis. We have applied solvent tuning to the recycling of a fluorous CBS reagent with the added twist of conducting the actual reaction in the fluorous phase then extracting the product away from the reagent with an organic solvent. If Prof. Legros can make a double tagged f-DMAP, there’s a good chance that they could do the same thing.

Halogen bonding is the non-covalent interaction of a halogen atom, acting as a Lewis acid, with an electron donor such as a heteroatom. Conceptually, it’s very similar to the much more well-known effect called hydrogen bonding. Halogen bonding has gained some prominence in crystal engineering and other disciplines where supramolecular structures are important, such as protein interactions with other molecules. It has not been used, or invoked, that much in small molecule synthesis. A recent publication from Prof. Carsten Bolm’s group at Aachen does exactly that.

They report the use of perfluoroalkyl halides as catalysts in the reduction of quinolines to tetrahydroquinolines using Hantzsch esters as the stoichiometric reductant. Presumably, the first step is activation of the C=N bond by halogen bonding with the perfluoroalkyl halide. 13C and 19F NMR studies would seem to support this as discernible shifts in the chemical shifts of the quinoline and the perfluoroalkyl halide. Once the C=N bond has been activated, a transfer hydrogenation from the Hantzsch ester occurs resulting in the formation of the tetrahydroquinoline and the corresponding pyridine from oxidation of the Hanzscth ester. There was no mention of any dihydroquinoline being formed. According to the experimental the desired tetrahydroquinoline was isolated by flash chromatography.

It’s a curious reaction in that it works at generally much lower temperatures and at ambient pressure while most reductions of quinolines require hydrogenation at elevated pressures and/or temperatures. In general it was found that the perfluoroalkyl iodides out performed the perfluoroalkyl bromides with perfluorooctyl iodide being found to be the best. While only a limited number of substituents were reported and only at the 2 and 7 positions, the reaction did work well in each of those cases.

From a fluorous perspective, the use of the perfluoroalkyl halides as an activation agent is interesting, especially since it could be removed after reaction using various fluorous separation techniques. In fact, this could be very nice reaction system for fluorous biphasic catalysis, since the perfluoroalkyl halides are very soluble in perfluorocarbons such as FC-72. A bigger problem, however, might be how to remove the stoichiometric amount of pyridine that’s formed as a by-product. The use of Hantzsch esters as an organoreductant actually has a rich history owing to its use as a NADH mimic. Many of the same reductions mediated by NADH can be conducted using Hantzsch esters. It would be quite easy to make some fluorous Hantzsch esters and utilize either a fluorous LLE or fluorous solid phase extraction process to remove the pyridine by-product. This is a little more interesting since there’s a lot more of it to remove, and also in light of the work in the last several years by Rueping, List, MacMillan, and others using Hantzsch esters in the reduction of imines.

Just a Hantzsch, but fluorous may have a role here beyond just halogen bonding.