On the crest of a wave

Marcello DiMarea and Gerard Mignanib of Rhodia Pharma Solutions describe some of the latest developments in palladium-catalysed reactions.

Palladium-catalysed bond-forming reactions are a very active area of academic research that has been fully embraced by medicinal chemists. These reactions have implications for drug development as the leads synthesised using them are entering development pipelines.This article explores some of the issues surrounding the scale-up of palladium-catalysed reactions, a new class of ligands for palladium-catalysed amination reactions, and a palladium-catalysed synthesis of arylhydrazines, a key precursor to indoles.

Palladium-catalysed bond-forming reactions are the subject of attention in a large number of academic laboratories because of their specificity and mildness. Complex molecules can be assembled from elaborate fragments with little concern for competing reactions involving their functionality and with very high specificity for the desired bond formation. This has proven a boon to medicinal chemists by facilitating structure-activity relationship (SAR) studies, which require the synthesis of many analogues, by allowing rapid access to complex molecules.

Some idea of the scope of these reactions is given by the number of named reactions involving C-C bond-forming reactions. These include Suzuki-Miyaura, Stille, Hiyama, Negishi, Kumada-Corriu, Sonogashira, Tsuji-Trost, and Heck. However, this is only a fraction of the overall scope, as there are also palladium-catalysed cyanation, amination, carbonylation, cyclisation, and etherification reactions that have been discovered and exploited.

The wide use of palladium-catalysed transformations in drug discovery has had ramifications: drug candidates are entering development pipelines relying on palladium catalysis for their assembly, to the consternation of process chemists. This consternation is due to the different pressures on discovery and process chemists. Medicinal chemists are measured by the quantity and novelty of molecular candidates produced, while process chemists are measured by the quality of and ultimately the cost of drug substance produced. While the advantages of palladium catalysis to medicinal chemists are clear, there are disadvantages from the perspective of process chemists.

With respect to quality, there are strict regulatory limits as to residual palladium levels in drug substance, typically single digit ppm values. These can prove difficult to attain. With respect to cost, despite palladium being used in catalytic quantities, it is very expensive, as are the ligands required for catalytic activity. It is not unusual for the extra processing of a palladium recycle to be considered as a means of reducing overall costs. The substrates for the reactions need to be considered as well. For example, Sonogashira coupling reactions often require aryl iodides, which are more expensive than their corresponding chlorides.

So process chemists can be forgiven for wanting to 'design out' a palladium-catalysed bond-forming reaction from a synthesis. There have been a number of recent advances, however, that diminish these quality and cost concerns.

The palladium itself isn't solely responsible for the fascinating array of chemistries observed. For example, the ligand is vital for modifying the electronic characteristics of the palladium, maintaining the correct oxidation state, and preventing the precipitation of palladium. It is in this area that there has been a revolution.

A good example of this revolution is the Buchwald-Hartwig reaction - the palladium-catalysed amination of aromatic halides or their equivalent (figure 1).1,2 The Buchwald group at the Massachusetts Institute of Technology (MIT)3 has demonstrated three different generations of ligands for this reaction. Rhodia Pharma Solutions has licence to this and related technologies from MIT and has collaborated to accelerate their development.

The first generation of ligands was based on compounds such as tri(o-tolyl)phosphine, but the overall catalytic system required high temperatures and high catalyst loadings among other limitations. The second generation of ligands demonstrated by the Buchwald group were chelating bisphosphine compounds such as Xantphos and BINAP. These have the advantage of increased reactivity; for example, electron-rich aromatic halides become effective amination substrates.

third generation

The third generation of ligands is electron-rich, mono-dentate ligands such as DavePhos,4 MePhos,5 and XPhos (figure 2). These are all crystalline, air-stable solids that are available from Rhodia Pharma Solutions in multi-kg quantities using a three-step sequence conducted in a single vessel.6 The chemistry is unusual, relying on the addition of a Grignard reagent to transiently form benzyne.

The reactivity of the catalyst systems derived from these third generation ligands is exceptional. Aryl chlorides, once barely acceptable as substrates, are aminated at room temperature even with primary amines, which is a very difficult combination. Similarly, aryl tosylates are tolerated as substrates. The high reactivity also translates to lower catalyst loadings. It is not unusual to achieve useful reaction rates using less than 0.1 mol % palladium and 0.2 mol % ligand. In some instances, palladium loadings of 0.01 mol% have been obtained.

Low catalyst loadings translate to much better economics for palladium-catalysed amination reactions, diminishing one of the major barriers to scaling palladium-catalysed bond-forming reactions. Indeed, Rhodia has conducted amination chemistry at tens of kg scale and invested significant resource in better understanding these transformations.

Arylhydrazines have long been of pharmaceutical interest as intermediates towards pyrazoles, and, most importantly, towards indoles using the Fischer indole synthesis (figure 3). Arylhydrazines are typically accessed by tin(II) or sulfite reduction of diazonium salts, but these are low yielding transformations that generate large volumes of waste.

An alternative approach is based on palladium-catalysed hydrazonation of aryl halides, a variant of amination chemistry, using benzophenone hydrazone as a hydrazine equivalent.7-8 Hydrolysis then reveals the desired arylhydrazine or the hydrazone product itself can be subjected to Fischer indole synthesis conditions.7,9 Rhodia has investigated the model system of p-bromotoluene or p-chlorotoluene combined with benzophenone hydrazone to give the corresponding N-p-tolyl benzophenone hydrazone or p-tolylhydrazine hydrochloride, some of the results of which are shared here.

A screening of palladium(II) sources (MePhos, ligand; t-BuONa, base) indicated no great differences, so cost effective palladium acetate (Pd(OAc)2) was selected for further attention. Interestingly, differences in quality between the various sources of palladium acetate were evidenced by varying reaction rates. This was traced to differing levels of crystallinity, as confirmed by X-ray analysis, indicating that due care should be taken during initial screening work.

Two sets of conditions were identified based on third generation ligands, while other classes, and earlier generations, of ligands were ineffective. The first was similar to those reported by Buchwald involving refluxing toluene with sodium tert-butoxide as base. The second was a novel combination of solid sodium hydroxide in refluxing tert-butanol or tert-amyl alcohol.

Interestingly, ground sodium hydroxide was significantly better than pellets or flaked forms. The base selected for the hydrazonation reaction was critical. Too weak a base led to no reaction, while too strong a base led to competing Wolff-Kishner reaction of the benzophenone hydrazone to afford diphenylmethane. On optimisation, less than 0.1 mol % palladium and 0.2 mol % ligand were needed.

The exothermicity of the hydrazonation reaction (ca. -100 kcal/mol) required either controlled dosing of benzophenone hydrazone or aryl halide. The order of addition is important: adding the aryl halide to the other reactants gives a stalled reaction, while adding benzophenone hydrazone does not. It was also demonstrated that benzophenone hydrazone forms a stable palladium(II) complex. That some substrates are competing ligands for palladium must be taken into consideration.

Probably the oldest of all methods for reducing residual palladium levels in the products of palladium-catalysed bond-forming reactions was very effective in this instance. After screening various activated carbon sources, several were found that could control residual palladium to less than 5ppm in the N-p-tolyl benzophenone hydrazone. This hydrazonation chemistry shows the power of recent advances in at least one area of palladium-catalysed bond-forming reactions: low loadings of palladium and ligand can be used on relatively inexpensive substrates with excellent control of residual palladium levels.

Palladium-catalysed bond-forming reactions are increasingly common in development pipelines, but this should not be feared when scale-up is needed. New generations of ligands are emerging that greatly diminish the barriers previously encountered. Clearly, the wave of palladium-catalysed bond-forming reactions is cresting, but thankfully the tools and understanding are at hand to make this manageable.