In recent years, transition-metal-catalyzed carbon-carbon bond-forming chemistries have become essential tools for the development of agrochemical and pharmaceutical active ingredients and other fine chemicals. Advances achieved in the 1990s have made the reactions cost-effective and scalable. Nevertheless, significant challenges remain, and the field continues to be fertile for research.
Perhaps the most widely used of these chemistries is Suzuki coupling, named after Akira Suzuki, a chemistry professor at Kurashiki University of Science & the Arts, in Japan. The reaction is also called Suzuki-Miyaura coupling to recognize the contributions of Norio Miyaura, a chemistry professor at Hokkaido University, in Japan, and a former member of Suzuki's research group.
In this palladium-catalyzed reaction, the coupling partners are an electrophile, usually an aryl halide, and a nucleophile, usually an arylboronic acid, and the product usually is a biaryl. Biaryls are common substructures in natural products and bioactive compounds. Commercial examples include the antihypertensive drug valsartan, from Novartis, and the fungicide boscalid, from BASF.
Suzuki coupling is now widely used in the discovery of bioactive compounds. With a library of boronic acids and a library of aryl halides, "you can very quickly generate a whole range of drug candidates," says Peter Jackson, vice president for pharmaceutical products at Avecia. "This technology has helped to fuel a lot of the expansion of synthesis in medicinal chemistry."
"The more it is used in discovery, the more it will eventually be used in development and production," says Fredrik Cederbaum, a group leader for fungicide chemistry at Syngenta. Although no Syngenta product at the moment uses a Suzuki coupling in production, some candidates in the pipeline, if successful, would require it as a key synthetic step, he notes.
Suzuki coupling also is used to produce advanced materials. Dow Chemical's Lumation light-emitting polymers are assembled through Suzuki coupling. So are the monomers for Dow's low-dielectric-constant polymer called SiLK, which is used as an insulator for electronic chips. Similarly, Sigma-Aldrich carries out metric-ton-scale Suzuki couplings to prepare compounds used in organic light-emitting diodes.
THE REACTION is powerful because it generally tolerates the presence of functional groups in the coupling partners, says Shaun R. Stauffer, a senior research medicinal chemist at Merck Research Laboratories. Older methods of joining aryl groups typically require reagents that are incompatible with molecules bearing polar groups such as amino and hydroxyl groups. "We get a lot of customer requests" related to Suzuki coupling, says Thomas H. Riermeier, senior R&D manager at Degussa Homogeneous Catalysts. Demand for catalysts is high, in amounts ranging from lab scale to production scale, to make building blocks and intermediates.
The reaction requires relatively mild reaction conditions, notes Johannes G. de Vries, principal scientist for homogeneous catalysis at DSM Pharma Chemicals. Because it can be used without protection and deprotection of functional groups in the reagents, it significantly shortens synthetic routes.
That Suzuki coupling yields a biaryl in one step makes it highly desirable, says Anita Schnyder, a catalysis and process development scientist at Solvias AG. For most other coupling reactions, she explains, alternatives in standard organic chemistry exist, but not for aryl-aryl coupling.
Even when the desired end product is not a biaryl, the aryl-aryl coupling is effective. For example, in a project for Pharmacia (now part of Pfizer), Dowpharma constructed a chiral 3-arylpiperidine by Suzuki coupling of an aryl bromosulfone and a pyridyl borane, followed by selective reduction of the pyridine ring in the 3-arylpyridyl product.
That this construction could be accessed through a Suzuki coupling "was not necessarily obvious," says Robert B. Appell, a process research leader at Dowpharma. "But after considering other routes using the fully hydrogenated heterocyclic ring, it was obvious that the coupling of two aromatic partners and reduction of one of them was the most straightforward."
THE SUZUKI methodology used to be limited in scope. Because catalysts were not optimal, reactions required high catalyst loadings and were effective only with the more reactive but more expensive aryl iodides or bromides and not with the less reactive but less expensive aryl chlorides. Boronic acids also were not widely available. At present, many palladium catalysts are practical for large-scale operations. Electrophilic reagents include aryl iodides, bromides, and chlorides as well as aryl triflates and tosylates. A wide range of organoboron compounds are available from fine chemicals producers, including BASF, Clariant, FMC Lithium, Sigma-Aldrich, and Strem Chemicals.
Sales of fine chemicals related to Suzuki coupling have "grown aggressively" in the past five years at Sigma-Aldrich, says Geoffrey Irvine, the company's business development manager for fine chemicals. In particular, sales of organoboron compounds have grown at a double-digit pace. Because these compounds are more likely to be used in a Suzuki coupling than in any other reaction, they are good indicators of how the reaction has grown in use, he adds.
According to Chemical Abstracts Service, the term "Suzuki reaction" first appeared in 1989 in the basic index of Chemical Abstracts. Since then, the number of publications has climbed steadily. The area "got very hot around 1999 or 2000," says Matthew Toussant of CAS's editorial group.
That hot period came shortly after breakthroughs in the labs of chemistry professors Stephen L. Buchwald and Gregory C. Fu at Massachusetts Institute of Technology in 1998. Buchwald and Fu had found systems that catalyze the coupling of aryl chlorides under mild conditions. People got excited because the systems are more general than any before, Fu says.
In the Buchwald system, the ligand is an electron-rich phosphine-bearing biphenyl, a motif that characterizes what are now referred to as the Buchwald ligands. In the Fu system, the ligand is tri-tert-butylphosphine. Both electron rich and bulky, these ligands catapulted bulky phosphines to the forefront of palladium-catalyzed coupling chemistry. Proof of practicality is that Buchwald's chemistry has been licensed exclusively by Rhodia Pharma Solutions. Fu's chemistry is not patented and is widely emulated.
The versatility of the Buchwald chemistry and the accessibility of Buchwald's catalysts by practical synthesis were key factors in licensing this technology, says Michel Spagnol, vice president for sales and marketing at Rhodia Pharma Solutions. The company is commercializing the chemistry not only for carbon-carbon bond formations but also for carbon-nitrogen (see page 62) and carbon-oxygen bond formations.
Other researchers and companies have joined the quest for new, bulky, electron-rich ligands with an eye not only for activity but also for commercializability. That means ligands should be easy to prepare in large scale and also be tunable--that is, they should be easy to modify structurally for particular applications. And unlike the quintessential bulky phosphine tri-tert-butylphosphine, which is pyrophoric, the ligands should be reasonably easy to handle.
SOLVIAS BEGAN developing catalysts with such properties about 10 years ago. In the beginning, the focus was on activating aryl chlorides, Schnyder says. Aryl chlorides are the preferred electrophiles not only because they are more widely available but also because their reaction produces sodium chloride, a relatively innocuous by-product, compared with sodium iodide or bromide formed with other aryl halides. R&D yielded the proprietary ligand SK-CC01-A. Degussa's proprietary Suzuki-coupling ligands--cataCXium A and cataCXium P--also are easy to handle and highly tunable. Riermeier points out that Suzuki coupling is sensitive to small changes in reaction components. Although high-throughput screening can identify optimum conditions for a given catalyst, tunable ligands provide another avenue for optimization.
"There is not one ligand that will work with every coupling reaction," says Matthias Beller, director of the Leibniz Institute for Organic Catalysis at the University of Rostock, in Germany, and Degussa's collaborator in developing the cataCXium catalysts. Ligands must be optimized for particular reactions; that's why the tunability is important, he adds.
Johnson Matthey has several nonproprietary catalysts in its portfolio. The company develops routes for unpatented palladium complexes, counting on its ability to make the best quality catalysts at the best price, says Mark Hooper, a development scientist.
The latest additions to the nonproprietary portfolio are two ferrocenylphosphine-based catalysts that catalyze the coupling of aryl halides at very low metal loadings and relatively mild conditions. One of them--1,1´-bis(di-tert-butylphosphino)ferrocene palladium dichloride--is already being used by a customer to make active pharmaceutical ingredients (APIs), according to Thomas J. Colacot, a senior development scientist. Customer feedback, he says, indicates that the catalyst is up to 20 times more active than an earlier analogous palladium complex in the Johnson Matthey portfolio.
Recently, Johnson Matthey licensed Q-Phos--pentaphenylferrocenyl di-tert-butylphosphine--a ligand developed by Yale University chemistry professor John F. Hartwig. Its palladium complex is reactive with aryl chlorides. It also allows a general procedure for coupling aryl halides with not only arylboronic acids but also alkylboronic acids under relatively mild conditions. The catalyst is effective for carbon-nitrogen and carbon-oxygen bond-forming couplings, too.
"To achieve the most competitive process, you need the most active catalyst to work on the most cost-effective substrate," says Fred Hancock, product manager for Johnson Matthey Catalysis & Chiral Technologies. "That's where we think Q-Phos and other ferrocenylphosphines take off. They are straightforward to synthesize. Monosubstituted ferrocenes are commercially available, and you get the phosphine in one or two steps."
THOUGH WIDELY practiced, Suzuki coupling is far from perfect. A major issue is palladium. It is an expensive metal, costing about $7,400 per kg, according to Christopher J. Woltermann, an organic chemistry group leader at FMC Lithium. Because of this and environmental regulations, immobilizing the metal so that it can be recovered and recycled is highly desirable. Metal immobilization also reduces product contamination. Metal residues in APIs are strictly regulated. For products used in electronics applications, tolerance for metal residues is even lower than for APIs. And for agrochemicals, although no official threshold for palladium residues exists yet, "in practice you try to keep it as low as possible, for cost reasons and likely future regulations," Syngenta's Cederbaum says.
Various ways to immobilize palladium are available. In Johnson Matthey's FibreCat catalysts, palladium is held in place by coordination to a phosphine ligand that is covalently tethered to an insoluble polymer support. After reaction, the catalyst is recovered through filtration. Recently, chemists at Abbott Laboratories developed an efficient Suzuki-coupling protocol using FibreCat catalysts in conjunction with microwave heating.
Polium Technologies offers something similar with its Rexalyst system, launched last May. Here also, a polymer-linked ligand immobilizes the metal by coordination. However, the polymer is soluble at the reaction temperature so that coupling occurs in a homogeneous system, explains Hamilton J. Lenox, business development manager. At the end of the reaction, when the system cools, it separates into two layers. Because the polymers are designed specifically for particular reaction systems, the polymer pulls the ligand and coordinated metal into one layer, leaving the product in the other layer. The catalyst is recovered by liquid extraction.
Rexalyst was developed in collaboration with David E. Bergbreiter, a chemistry professor at Texas A&M University, College Station. Suzuki-coupling catalysts are one of the first applications. Catalyst developers and suppliers, including Degussa, Engelhard, FMC Lithium, and Digital Specialty Chemicals, have expressed interest in it, Lenox says.
Pd Encat, Avecia's immobilized catalyst technology, is based on encasing palladium in highly cross-linked polyurea beads. The technology, developed in collaboration with Cambridge University chemistry professor Steven Ley, won the British Chemical Industries Association's Innovation of the Year Award in 2004.
In Pd Encat, metal and ligand occupy the spaces between the polymer chains; the reagents diffuse into the polymer and react, and the products diffuse out, Jackson explains. Recycling experiments show that only about 2% of the catalytic activity and less than 0.1% of palladium are lost after 20 rounds. Because the polyurea matrix itself activates the metal, the amount of phosphine ligand required to form an active catalyst is reduced, in some cases by up to 50%.
Phosphine contamination can be a big problem with pharmaceuticals, Jackson points out. Phosphines tightly associated with APIs can be difficult to remove. And emulsions formed by phosphines in conventional coupling reactions can be hard to clean up, requiring multiple washes with multiple solvents.
Pd Encat has been effective in solving problems encountered by clients, Jackson says. For example, one client had optimized a heterogeneous Suzuki coupling catalyzed by palladium on carbon but could get only a 94% yield of the desired product. Studies by Avecia showed that the desired reaction was occurring with dissolved palladium but that competing reactions were taking place with palladium on the metal surface. Pd Encat avoids those competing reactions because palladium is always in a controlled environment in the polymer, he explains. Even with a loading of 0.1 mol %, the crude product formed via Pd Encat passes specifications easily; it's 99% pure without recrystallization.
DSM IS TAKING a different tack to managing palladium, as well as phosphine, contamination--by using extremely low loadings of palladium and eliminating the ligand. This approach is based on the fact that the active form of palladium, Pd(0), tends to cluster and eventually precipitates. "A lot of people have observed this phenomenon. Halfway through a Suzuki, they get this black goo, and the reaction stops," DSM's de Vries says. At a loading of 0.01 to 0.1 mol %, palladium will not precipitate and does not require a ligand. DSM has tested the low-loading method in a scale of several hundred grams. The technology is ready for commercial application, he adds.
How low can loadings go? All the way to zero, according to Nicholas E. Leadbeater, an assistant professor of chemistry at the University of Connecticut, Storrs. His group has run Suzuki-type couplings in water with microwave heating in the absence of palladium. He has taken great care to check for metal, analyzing 32 metals in reagents and in the postreaction mixture down to detection limits. He suggests that a metal-free pathway to the coupling of aryl halides and arylboronic acids exists. His group is probing the reaction in more detail.
Leadbeater's results have been received enthusiastically despite obvious questions. For example, how far can scale-up proceed with microwaves? Huge equipment won't work because microwaves penetrate only about half a centimeter from the surface of the material that's being zapped. A flow system through the microwave equipment might be a solution, Leadbeater says. Another is to use conventional heating, which also works but is limited in scope.
Another issue is that water, which must be the reaction medium, is a poor solvent for many Suzuki reagents and promotes side reactions. But if those problems are overcome, water is a great solvent to work with, being cheap, nonflammable, and nontoxic.
Is the system really metal free? DSM's de Vries is skeptical. "Really, very tiny amounts of metals without any ligands" could catalyze the reaction, he says. Such amounts could be released by the reaction vessel or the stir bar while the reaction is running and then return to the vessel or stir bar when the reaction is over. Leadbeater should measure metal content during the reaction and at sub-parts-per-million sensitivity, de Vries suggests.
Meanwhile, at the University of Tokyo, catalyst immobilization and very low catalyst loadings converge in the work of chemistry professor Shu Kobayashi on immobilization technology called polymer incarceration.
Here, the metal is first encapsulated in a polymer. Then the microcapsules are cross-linked. Kobayashi has shown that incarcerated palladium is highly active in hydrogenation and suggests that it could also work with Suzuki couplings and other palladium-catalyzed reactions.
Kobayashi says the high activity is due to small clusters of palladium atoms inside the polymer. "Now there is a movement to synthesize palladium nanoclusters. The smallest so far measures 1 nm," he says. "We have seen subnanometer clusters, containing only seven atoms."
After the catalyst, the organoboron reagent is perhaps the next most important consideration in Suzuki coupling. A subindustry based on supplying organoboron compounds has grown, making huge numbers available for mixing and matching with the coupling partner. But if the required compound has never been made before, someone has to prepare it first.
Synthesis of boronic acids is not trivial. Typically, an aryl compound is treated with butyllithium to form a lithiated aryl intermediate. Boronic acid then replaces the lithium, forming an arylboronic acid.
Companies with expertise in this chemistry have grown their custom synthesis business from the ability to supply customer-specific boronic acid reagents. An example is FMC Lithium, which has many years of experience in the organometallic chemistries used to produce boronic acids. It is now applying its expertise to the phosphine ligands, many of which are prepared also via organometallic chemistries, Woltermann says. With these complementary capabilities, the company hopes that customers will ask it to run the coupling reaction as well.
BUT THE butyllithium route to boronic acids has deficiencies in cost, flexibility, and yield, according to Andreas Meudt, R&D manager for Clariant's pharmaceuticals business. Also, some boronic acids are not accessible by the usual chemistries or are unstable and easily lose boron. For these reasons, alternative routes and reagents have been developed. For example, Clariant prepares arylboronic acids with an alkyllithium prepared in situ from an alkyl chloride and lithium metal. In the presence of an aryl substrate, the alkyllithium immediately transfers lithium to the aryl compound, which is then replaced by a boronic acid. The method is flexible because the alkyl chloride can be varied. Compounds that could not be prepared with butyllithium become accessible through other alkyllithiums. However, the reaction requires cryogenic conditions.
Beyond arylboronic acids, Clariant has expanded its boron-based technology to aliphatic and vinylic boronic acids, based on chemistry codeveloped with Victor A. Snieckus, a chemistry professor at Queen's University, Kingston, Ontario. The move reflects the increasing interest in extending Suzuki coupling beyond only aryl partners to all types of carbon-carbon bond formations.
The centerpiece of the Snieckus chemistry is the hydroborating reagent di(isopropylprenyl)borane, which adds to the terminal carbon atom of terminal alkenes or terminal alkynes. When the intermediate is treated with aqueous formaldehyde, an alkyl- or vinylboronic acid is formed in yields of up to 86%, compared with only 25% by previous methods. All of the reactions--from preparing the reagent to generating the alkyl- or vinylboronic acids--take place in one pot. "The reaction looks difficult and complex on paper but is really easy and economical in practice," Meudt says.
Sometimes, boronic esters are used instead of boronic acids. They are preferred when the corresponding boronic acid is unstable or poorly soluble in the reaction solvent or when functional groups in the substrate are not compatible with organometallic reactions, explains Karl Matos, R&D manager for BASF's inorganics business unit in Evans City, Pa.
Among the reagents used to prepare arylboronic esters are pinacolborane, isopropyl pinacol borate, and bis(pinacolato)diboron. Boronic esters that deliver alkyl or alkenyl groups also can be prepared with reagents such as catecholborane and 9-borabicyclo[3.3.1]nonane. These and others are available from BASF and Sigma-Aldrich, among others. Preparing boronic esters by these routes does not have special requirements, such as cryogenic conditions. A customer requiring total secrecy can do everything in-house, notes Scott Starr, an account manager at BASF.
The importance of boron reagents for Suzuki coupling is underscored by Sigma-Aldrich's continuing development of new compounds that expand the reaction's scope. For example, diborane esters offer a route to arylboronic esters or acids with sensitive functional groups, and potassium trifluoroborates are potential alternatives to boronic acids. Recently, Sigma-Aldrich added to its portfolio three stable boronic acid derivatives that deliver vinyl groups.
DESPITE ITS commercial success, Suzuki coupling still poses many challenges. For example, catalyst costs remain high. Some of the newer, more active catalysts are too expensive, Syngenta's Cederbaum notes. For most applications, catalyst loadings still need to come down. The problem of high catalyst loading is often encountered when drug candidates are moved from medicinal chemistry to process development. In drug discovery, catalyst loadings of 10–20 mol % can be tolerated, but that is a lot for large-scale production, says Gerjan J. Kemperman, a senior development scientist for process chemistry at Organon. If the Suzuki coupling in the medicinal chemistry route cannot proceed at lower loadings, he says, "we have to consider other routes."
Catalysts for certain specific couplings still are inadequate. Beller notes that the coupling of an aryl halide with a terminal olefin cannot be carried out selectively using current methods. "It is possible with styrene and acrylic acid derivatives. These are model systems that everybody uses. But if you use a simple aliphatic olefin, such as 1-hexene, you observe scrambling of the double bond and you get different regioisomers. In these cases, there are no selective catalysts," he explains.
Catalysts with lower environmental impact still are needed. On this front, Kevin H. Shaughnessy, an assistant professor of chemistry at the University of Alabama, Tuscaloosa, has been developing water-soluble catalysts to make processes more environmentally benign than currently practiced. Water-soluble catalysts would also be easier to separate, reducing metal contamination of the organic product. In fact, synthesis of the first commercial-scale pharmaceutical intermediate requiring a Suzuki coupling--Clariant's process for o-tolylbenzonitrile--uses the water-soluble tris(3-sulfanatophenylphosphine) trisodium (TPPTS) ligand.
Shaughnessy has been seeking ligands with better activity than TPPTS and has found two that couple aryl bromides and boronic acids in excellent yield under mild conditions, with turnover numbers of up to 734,000.
One of the ligands, called t-Bu-Amphos, has been applied to the preparation, at room-temperature, of the nonsteroidal anti-inflammatory drug diflunisal using only water as the reaction solvent. In this case, the product precipitates and is separated by filtration. But if a product were soluble, it would be separated from the aqueous mixture by extraction with ethyl acetate, with no observable leaching of the catalyst into the organic phase, Shaughnessy says. The catalyst could be recycled up to four times before activity is lost, he adds.
Catalysts that extend Suzuki coupling beyond aryl-aryl sp2-sp2 carbon-carbon bond formation still are limited in scope. Researchers are working to make Suzuki coupling applicable to all types of coupling partners. On this front, MIT's Fu is developing methodology for alkyl-alkyl coupling. "A stereoselective and high-yielding method to generate sp3-sp3 carbon-carbon bonds would change the way synthetic chemists approach retrosynthetic analysis," he says. Any carbon-carbon bond can then be assembled from coupling partners. And when sp3 centers are involved, stereochemistry becomes another dimension. "Controlling that adds to the potential utility of the process, as well as to the challenge of method development," he says.
Finally, catalysts that enable the use of a C–H rather than a C–X electrophile (X = I, Br, or Cl) would be highly desirable. Doing the chemistry without the halide is a huge challenge, Leibniz Institute's Beller says. Significant advances are likely within the next 10 years, he believes.