SCIENCE & TECHNOLOGY
A TALE OF TWO CATALYSTS Years of effort reveal the mechanisms of two catalytic ASYMMETRIC ALLYLIC SUBSTITUTION reactions STU BORMAN, C&EN WASHINGTON
THE INNER WORKINGS of asymmetric
organometallic catalyst systems are not revealed to chemists every day, every week, or even necessarily every year. Studies to understand the detailed mechanisms of such systems typically take years of effort. The key to success is hard work—many days and nights of research by students and the ingenuity, dedication, and vision of principal investigators. It’s therefore rare for the mechanisms of two related asymmetric organometallic catalysts to be revealed almost simultaneously. But that’s exactly what happened recently, when two independent research groups reported, at about the same time, characterizing the intermediates and mechanisms of two complementary chiral allylic substitution reactions. In such reactions, a group adjacent to a double bond is replaced by another in a highly stereoselective manner. The studies show that organometallic catalytic complexes are indeed tractable, even if they sometimes might seem all but impossible to decipher. They can be analyzed structurally, and how they accelerate reactions can be figured out, albeit with a lot of effort. Guy C. Lloyd-Jones of the University of Bristol, in England; Per-Ola Norrby of the University of Gothenburg, in Sweden; and coworkers carried out one of the two studies, on palladium-based asymmetric catalysis (J. Am. Chem. Soc., DOI: 10.1021/ ja8099757). John F. Hartwig and coworkers of the University of Illinois, Urbana-Champaign (UIUC), did the other, which was on iridium-based chiral catalysis (J. Am. Chem. Soc., DOI: 10.1021/ja902609g). Both studies took years, and it’s extraordinary that they were published only a day apart last month.
mission of asymmetry from catalyst to substrate in these important reactions has lagged.” The new studies “represent significant advances toward demystifying this chemistry,” he adds. And Brian M. Stoltz of California Institute of Technology, who specializes in the synthesis of structurally complex bioactive molecules, says, “These are beautiful studies of mechanism, attempting to understand the very subtle and intricate details of two types of allylic reactions.” The mechanisms proposed in the studies “ultimately will allow one to build new hypotheses that move the field further,” Stoltz continues.
“There aren’t very many catalytic reactions for which people have been able to isolate intermediates and determine the reactivity of the intermediates,” Hartwig says. “There are even fewer catalytic asymmetric reactions for which this is true. IN THE Pd STUDY, mechanistic organomeMany people think of these reactions as a tallic chemist Lloyd-Jones—with coworkers black box. What Guy and I have been trying Craig P. Butts, Emane Filali, David A. Sale, to do with experiments, and Per-Ola with and York Schramm—and Norrby, a specialcomputational methods, is remove the ist in catalyst mechanisms, collaborated black box so you can see the inner workings closely to obtain structural and mechanisof catalytic systems. These are unusual castic information on Trost modular ligands es in which structural information has been (TMLs). Developed by synthetic organic determined by nuclear magnetic resonance chemist Barry M. Trost of Stanford Universispectroscopy or crystallography on interty, TMLs combine with Pd to form catalysts. mediates in a catalytic system, and detailed “Trost introduced these catalysts in information obtained on how asymmetry is 1992, and we started working on their transferred from the intermediates to reacmechanism in 1996, so we’ve been at this tion products.” for about 13 years,” Asymmetric Lloyd-Jones says. catalysis specialist Norrby and coworkAndreas Pfaltz of the ers have been invesO O University of Basel, tigating the mechaNH HN in Switzerland, says nism computationalthe papers “lead to a ly for about the same Flap better understandamount of time. P P ing of two highly TML-Pd catalysts Pd Flap useful, widely used have been successenantioselective fully deployed in L Wall catalytic reactions hundreds of allylic Wall and provide a basis substitution reacfor further catalyst tions to help synFLOPPY CATALYST development.” thesize a wide range It was proposed that in the catalyst Patrick J. Walsh of compounds for formed by TML and Pd, phenyl rings of the University of academic research, act like flaps and walls to control Pennsylvania, whose drug discovery, and stereoselectivity (L = allyl substrate). research focus inindustrial synthesis. cludes asymmetric A common struccatalysis, notes that allylic substitution tural feature of TML-Pd catalysts is the reactions are synthetically very useful presence of two triphenylphosphine units but that “our understanding of the translinked to each other through Pd on one side and through a chiral scaffold on the other. A TML containing the chiral scaffold cyclohexyldiamine, called the Trost standard ligand, is widely used. With the R,R enantiomer of this ligand, the Pdcatalyzed reaction predominantly yields the S product.
“Our understanding of the transmission of asymmetry from catalyst to substrate” in allylic substitution reactions “has lagged.” WWW.CEN-ONLINE.ORG
31
JUNE 15, 2009
SCIENCE & TECHNOLOGY
Trost has proposed a mechanism that celerating attack on one carbon instead of ing nucleophile and thereby controls the accurately predicts the catalyst’s stereoblocking attack on the other,” Lloyd-Jones stereoselectivity of nucleophilic attack. selectivity based on the hypothesis that says. He believed it might be an activating The other amide group projects back onto in the reaction intermediate, four of the ligand rather than a deactivating one and the other face of the chiral scaffold and phosphorus-linked that the ligand is not is thus remote from the nucleophile. The phenyl rings control just controlling the team believes the H-bonding acts like a Racemic substrate the nucleophile’s atreaction’s stereoselectargeting system to set up the nucleophile X X tack on the substrate, tivity but is also makto attack the pro-S carbon. Pro-S attack to which is linked to Pd. ing it go faster. make the S product predominates experiThe first step in a The conventional mentally with the R,R form of the Trost – X Nu– + TML-Pd-catalyzed view of the geometry standard ligand. Pd Nucleophile allylic substitution reof the reaction com“Our explanation predicts the same attacks here Pd catalyst action is the removal plex formed with the stereoselectivity as the Trost hypothesis of the substrate’s Trost standard ligand but shows how the ligand can be much Intermediate leaving group. With a is that the substrate more active because that H-bond switches Nu substrate that’s asymis appended below Pd things on,” Lloyd-Jones says. First, the liHigh selectivity for S enantiomer metric at the allylic and between the four gand helps pull away the substrate’s leaving position, detaching wall and flap phenyl group to generate the intermediate, he exthe leaving group from rings. That view sugplains, and then it “helps bring in a nucleoHIGH STEREOSELECTIVITY either enantiomer gests that the substrate phile via H-bonding, lowering the barrier to The R,R form of a TML-Pd catalyst yields the same inis located far from the nucleophilic attack.” The TML-Pd catalyst converts a racemic substrate to a termediate, with one cyclohexyldiamine “is very efficient because it assists in both predominantly S product (X = pro-R and one pro-S chiral scaffold and its steps,” he adds. – leaving group, Nu = nucleophile). carbon. Nucleophilic two amide groups. “This is an intriguing, almost enzymeattack on the pro-S But “there’s been no like mechanism,” the University of Basel’s carbon in the next step creates an S substistructural evidence to support that” view, Pfaltz says. “The Pd atom holds the allyl tution product, and a pro-R attack yields Lloyd-Jones says. “The majority of our efsystem in place, and the functional groups the R enantiomer. fort for the past 13 years has been trying to present in the chiral ligand guide the nuWhether the nucleophile attacks primardetermine the structure of the complex and cleophile in the right direction.” ily a pro-R or pro-S carbon is controlled by therefore identify where the four of the phosphorus-linked phenyl rings substrate is located in relation – on the ligand, according to Trost’s proposal. to the ligand.” O X M+ Two of these are oriented equatorially In the new study, LloydHN N and are like bent-out flaps, and the other Jones, Norrby, and coworkers Nu– H Nu two are axially oriented and point straight demonstrate that the subO + MX down, like walls. One wall tends to block strate is not between those P Pd+ P the nucleophile from attacking the pro-R four phenyl rings; instead, it is carbon, and because of the blockage, the twisted up on one side of Pd, High selectivity for S enantiomer R product isn’t formed very often. On the bringing it close to one of the other side, a flap leaves the pro-S carbon chiral scaffold’s two amide open for reaction, which forms an S prodgroups. “We’ve spent a long PLAN OF ATTACK Experimentally determined uct, the result obtained most of the time. time using NMR methods to structure of reaction intermediate suggests that Trost’s mechanism is a deactivation prove that,” Lloyd-Jones says. H-bonding between the nucleophile and the chiral model in that it From this scaffold’s amide group actually controls proposes that hard-won strucChiral stereoselectivity (X– = counterion to Pd; M+ = scaffold stereoselectivity tural informacounterion to nucleophile, Nu–; curved black is controlled by tion, the team arrow = pro-S nucleophilic attack). Wall the tendency of derived an alterFlap one of the walls native mechaPd Flap to block the disfanism of the catalytic process. Why did it take Lloyd-Jones’s group 13 Wall vored R pathway. “It took Guy and me sitting years to obtain the structure of the TMLBut “based on exdown together for a very inPd-substrate complex? That’s because Pro-R Pro-S perience in my lab tense week last year to finally most chelating ligands for asymmetric and reports from make sense of this and come catalysis form complexes with 5-, 6-, or DEACTIVATING WALL a lot of people up with a working selectivity 7-membered rings, whereas the TML-PdIt was originally proposed that who had used the model,” Norrby says. substrate complex has a 13-membered ring. a phenyl wall tends to block ligand that it was They propose that the “This ring is very flexible, which is key to attack at one carbon position, exceptionally amide group that’s close to its success, making it possible to put the Pd and a phenyl flap generally reactive, we susthe substrate forms a hydronear the chiral scaffold,” Lloyd-Jones says. allows it to occur at the other. pected it was acgen bond with the incom“But this flexibility also causes the ligand to WWW.CEN-ONLINE.ORG
32
JUNE 15, 2009
form oligomers, and in solution the oligomeric form is favored. That’s made study of the monomer extremely difficult because it’s such a vanishingly small component in the mixture. And yet it’s that vanishingly small component that’s the active catalyst. So to deduce its structure by NMR, we’ve had to use special techniques—such as deuterium labeling and special counterions to control oligomer-monomer ratios—and that’s taken us a long time.” The effort was worth it, though, he says. With the new model, researchers can rationalize not only the ligand’s enantioselectivity but also an important effect of counterions on selectivity. “It’s been known for a long time that as you change the M+ counterion of the nucleophile, the enantioselectivity changes a lot,” Lloyd-Jones says. “People have never been able to satisfactorily explain that. Our model explains it because the more dissociating you make M+—as you go from Li+ to Na+ to K+ to Cs+—the better the nucleophile can H-bond and therefore the higher the enantioselectivity. That’s exactly what’s observed experimentally. It means you can now rationally optimize conditions to improve selectivity, rather than doing it via empirical screening.” The new mechanism is important because “there are cases where the ligand is not ideal, and it helps to see why that is so,” Lloyd-Jones adds. “It shows you how to get the best from that ligand.” IN THE SECOND study, transition-metal-
catalyst specialist Hartwig and UIUC coworkers Sherzod T. Madrahimov and Dean Markovic structurally characterized and analyzed Ir-based, allylic substitution catalysts, which accelerate the formation of products with atomic connectivity complementary to that formed by Pd systems. Hartwig’s group began developing this class of catalysts in 2002. Ir catalysts, together with the Pd ones, allow researchers to dial in the product they will get with allylic substitution. With monosubstituted allylic ester substrates, Ir catalysts promote substitution reactions that produce branched products, and Pd-catalyzed substitutions give linear products.
According to Pfaltz, “Iridium-catalyzed When they obtained the crystal strucallylation is still a relatively young reaction, ture, they were able to see the structure of but it has become a very powerful method the whole complex for the first time in the for C–C and C-heteroatom bond formaseven years they had been working on it. tion. What makes it so useful is that it leads “We had no idea how the chirality of the to branched chiral substitution products. ligand would transfer to the chirality of the But compared with Pd-catalyzed allylation, product,” Hartwig says. “This structure the mechanism was much less understood. gave us a way to view how that transfer proElucidation of the three-dimensional cess occurs.” structure of the crucial allyl intermediate represents a breakthrough that gives THE CATALYST complex consists of a fiveimportant insights into the mechanism of membered ring that includes Ir, a chiral this reaction and will allow a more rational binaphthyl group, a cyclooctadiene, and the approach to catalyst development.” bound allyl substrate. Hartwig and coworkHartwig says he and his coworkers reers find that a stereocenter in the five-memcently worked out a synthesis that will make bered ring controls the absolute stereothe Ir catalysts “more widely available to chemistry of the product by causing the the research community and to industry neighboring nitrogen-linked phenylethyl groups, but we have not yet scaled it up.” group to point toward the allyl substrate in Johnson Matthey Catalysts has licensed the the complex. They believe the phenylethyl Ir catalysts. group forces the allyl group to orient itself Now, after years of effort, Hartwig and with its substituent toward the outside of coworkers have identified the intermedithe catalyst to avoid a steric clash. ate that forms in Ir-catalyzed reactions. In “That controls the orientation at which addition, they have structurally analyzed it the allyl group is bound and the catalyst’s and have proposed a catalytic mechanism. enantioselectivity,” Hartwig says. “The They generally use allylic esters as subnucleophile adds to the carbon that has the strates in Ir-catalyzed reactions, which substituent, on the side of the molecule far convert them into substitution products. from the controlling elements of the ligand. To identify the inter“It’s pretty intricate mediate and obtain its how the catalyst’s + structure, they had to asymmetry gets transfind a way to capture a ferred from one point R nonreactive version of to another,” Hartwig Ir O O P the catalyst-substrate continues. “Without a N complex. structure, there would By using allylic chlobe no way to envision Ph Ph rides instead of allylic how that could happen. esters as substrates and There are certainly FROZEN COMPLEX then removing the chlolimitations in reaction Researchers believe the phenylride from the system, scope that we would ethyl group in this Ir complex they were able to freeze like to overcome, and dictates enantioselectivity by the catalyst-substrate by knowing how the orienting the allyl group (Ph = complex in the absence atoms are interacting in phenyl, R = allylic substituent). of any nucleophile, and the structure—where this enabled them to sucthere are steric clashcessfully crystallize the complex. “That was es—we are hoping we’ll be able to make by no means easy to do,” Hartwig says. “It changes to the structure to allow us to do took a lot of experimental skill to figure out the reaction with substrates that so far we the right reagents and conditions that would haven’t had success with.” allow the system to crystallize. Dean worked Hartwig adds that “people often throw up out the conditions to form the intermediate, their hands and say catalysis is too compliand Sherzod obtained single crystals.” cated, that you never know what the active species is, and that asymmetric catalysis is another level less decipherable” than achiral catalysis. “But I think the collaborative work by Guy and Per-Ola and by our group, among others, shows this is not true. We can figure out how these systems work and can learn from the information gained.” ■
The mechanisms proposed in the studies “ultimately will allow one to build new hypotheses that move the field further.” WWW.CEN-ONLINE.ORG
33
JUNE 15, 2009
