A supramolecular microenvironment strategy for transition metal catalysis

See allHide authors and affiliations

Science  04 Dec 2015:
Vol. 350, Issue 6265, pp. 1235-1238
DOI: 10.1126/science.aad3087

Faster elimination inside a cavity

Metals are adept at shuffling molecular bonds. They pry apart two atoms and then pair each one with a different partner. Sometimes the atoms get stuck on the metal, though, and the newly partnered products aren't released. Kaphan et al. designed a strategy for accelerating this elimination process (see the Perspective by Yan and Fujita). A hollow supramolecular capsule captured a gold or platinum complex and induced rapid bond formation between the carbon atoms in methyl groups bound to the metal. Generalization of this strategy could open the door to a wide range of chemical transformations that are currently held up by slow eliminations.

Science, this issue p. 1235; see also p. 1165


A self-assembled supramolecular complex is reported to catalyze alkyl-alkyl reductive elimination from high-valent transition metal complexes [such as gold(III) and platinum(IV)], the central bond-forming elementary step in many catalytic processes. The catalytic microenvironment of the supramolecular assembly acts as a functional enzyme mimic, applying the concepts of enzymatic catalysis to a reactivity manifold not represented in biology. Kinetic experiments delineate a Michaelis-Menten–type mechanism, with measured rate accelerations (kcat/kuncat) up to 1.9 × 107 (here kcat and kuncat are the Michaelis-Menten enzymatic rate constant and observed uncatalyzed rate constant, respectively). This modality has further been incorporated into a dual catalytic cross-coupling reaction, which requires both the supramolecular microenvironment catalyst and the transition metal catalyst operating in concert to achieve efficient turnover.

Supramolecular catalysis was born from biological inspiration. By emulating the transition state binding of enzymatic active sites (1), chemists have developed synthetic noncovalent microenvironment catalysts that are capable of reproducing a range of enzymatic reaction motifs, from proton-catalyzed hydrolases to constrictive terpene cyclases (29). However, due to the circumstances of evolutionary history, nature has selected for enzymes that mediate a narrow subset of chemistries employing Earth-abundant elements. Although the exploration of biomimetic chemistries has dominated supramolecular catalysis thus far, no such limitations are inherent to fully synthetic systems, and as such, microenvironment catalysis is poised for expansion to abiotic mechanistic manifolds.

In many ways, the development of organic chemistry was born out of the same biological impetus, and many triumphs of the field are recapitulations of superior enzymatic systems (consider the excellent stereoselectivity of aldolases or the site selectivity in C-H functionalization by oxidases). In contrast, organotransition metal catalysis was developed in the absence of a preceding biological analog. The proliferation of such methods has relied on the judicious examination of catalyst structure-activity relationships, traditionally exploiting ligand architecture to provide a series of specialized catalyst systems for individual transformations and substrate classes (10). As an alternative strategy, we envisioned a marriage of transition metal and microenvironment catalysts for the expansion of the canon of accessible reactivity. This wedding of biomimetic and anthropogenic chemistries opens the door for a paradigmatic shift in strategic approaches to catalysis.

Successful realization of catalytic cycles requires the careful orchestration of elementary steps: Should any step prove particularly slow, the overall rate will become retarded, and in extreme cases, reactivity can be precluded altogether. Although certain stoichiometric additives have been shown to accelerate elementary organometallic reactions (1115), the catalysis of such reactions is less common (1618). Rather than altering catalyst structure in a way that influences each elementary step, a synthetic microenvironment catalyst could be leveraged to specifically target one step without sacrificing reactivity elsewhere along the catalytic cycle, enabling an otherwise inaccessible catalytic process. One catalytically relevant yet particularly sluggish transformation is reductive elimination of sp3 fragments from transition metals (10). As a result, alkyl-alkyl cross-coupling processes are often plagued by slow turnover and undesired side reactions (19).

In light of the known trends for reductive elimination, we envisioned that the tendency of Ga4L612– (1) supramolecular assembly (Fig. 1A) [L = N,N′-bis(2,3-dihydroxybenzoyl)-1,5-diaminonaphthalene] to encapsulate low-coordinate cationic transition metal complexes was promising as a means to selectively recognize and stabilize the relevant transition state for C-C bond formation (20, 21). The highly anionic, tetrahedral metal-ligand cluster 1 assembles spontaneously in solution and has been shown to catalyze a range of reactions involving neutral and cationic substrates via encapsulation within its hydrophobic interior cavity (22). Thus, we hypothesized that 1 would be competent to act as a microenvironment catalyst for reductive elimination from an appropriate high-valent dialkyl metal complex. Initial realization of this reactivity in a stoichiometric sense would potentially inform the development of a dual catalytic process, wherein a microenvironment catalyst and a transition metal catalyst act in concert to achieve a challenging alkyl-alkyl cross-coupling.

Fig. 1 Supramolecular catalysis of reactivity at a transition metal center.

(A) Structure of the Ga4L6 catalyst. (B) Catalyzed reductive elimination from gold and halide series. Me, methyl; kobs, observed rate; krel, relative rate. (C) Improved catalysis with a triethylphosphine analog. TON, turnover number. (D) Catalyzed reductive elimination from platinum. (E) Identification of a catalyst-deactivation pathway.

To that end, we evaluated 1 as a catalyst for the elimination of ethane from the dialkyl Au(III) iodide complex 2-I (23, 24). The observed half-life for reductive elimination under a 10 mol% loading of 1 decreased from 20 weeks to just 53 min, corresponding to a 4000-fold acceleration in the observed initial rate (Fig. 1B). Substitution of the compact trimethylphosphine ligand by its more sterically demanding triphenyl congener resulted in no observable acceleration for the reductive elimination in the presence of 1, which is indicative of size exclusion from the internal cavity of the catalyst. Likewise, blocking the interior cavity with the strongly encapsulated Et4P (Et, ethyl) cation eliminated the accelerating effect of 1. Compared to observations for reactions with 2-I, the corresponding chloride and bromide complexes 2-Br and 2-Cl showed similar behavior under cluster-catalyzed conditions, with relative rates of 5.7 and 6.5, respectively.

From the kinetic profile of the reductive elimination from 2-I (fig. S1), it became clear that a catalyst-deactivation pathway was operative at extended reaction times. Although no product inhibition was observed, examination of the reaction mixture by 1H nuclear magnetic resonance (NMR) spectroscopy revealed a strongly encapsulated species, which was identified as the cationic bis(phosphine) complex 3 (25). Its identity was verified by independent synthesis, and Embedded Image was shown to be incompetent as a catalyst for the elimination of ethane from 2-I (Fig. 1E). To disfavor this deleterious pathway, we hypothesized that a phosphine of intermediate steric demand would preclude the formation of the analogous bis(phosphine) cation while remaining sufficiently compact to allow encapsulation by 1. In accordance with this hypothesis, triethylphosphine-ligated complex 4 exhibited rapid and complete reductive elimination in the presence of 1 (Fig. 1C). No catalyst deactivation was observed, and a turnover number in excess of 300 could be achieved. The catalyzed reaction of 4 exhibits a half-life of just 47 s compared with 45 days in the uncatalyzed reaction, corresponding to more than an 80,000-fold increase in the observed rate.

We then endeavored to evaluate whether this catalytic methodology could be generalized to reductive elimination from other dialkyl transition metal complexes bearing a labile X-type ligand. To that end, platinum complex 5 was prepared and subjected to catalysis by 1 (26). The observed half-life for reductive elimination of ethane decreased from 9 days in the absence of 1 to just 6 min in its presence, corresponding to a 2300-fold acceleration of the observed rate (Fig. 1D).

For a better understanding of the catalyzed reductive elimination process, we conducted kinetic experiments to determine the order in each reactant, using the method of initial rates. For the reductive elimination of ethane from complex 2-I, the reaction displayed first-order dependence on catalyst 1, as measured by competitive inhibition with Et4P+ (a linear relation was observed between the rate of reductive elimination and the concentration of the unblocked cluster). Conversely, the rate of reductive elimination was found to be dramatically attenuated in the presence of exogenous iodide (Fig. 2A). The dependence on gold concentration showed saturation behavior, which could be linearized by plotting the double reciprocal of concentration and rate (27). These results are consistent with an overall Michaelis-Menten–type mechanism involving pre-equilibrium halide dissociation followed by a transient and reversible encapsulation of the nascent cationic species and, finally, an irreversible reductive elimination event within the cluster cavity (Fig. 2B).

Fig. 2 Kinetic experiments and implications.

(A) Determination of the order in each reactant in the rate law. (B) Proposed mechanism for catalysis by 1. M, metal; X, halide. (C) Experimentally determined rate law. (D) Measured values for kcat and overall acceleration. Asterisks denote 298 K in MeOH-d4; the dagger symbol indicates 313 K in 9:1 MeOH-d4/D2O. (E) Lineweaver-Burk analysis. All error bars represent 1 SD, based on three replicates.

Accordingly, from the aforementioned double reciprocal (i.e., Lineweaver-Burk) plot, the Michaelis-Menten parameter kcat (Michaelis-Menten enzymatic rate constant) could be assessed (Fig. 2E). The measured kcat for complex 2-I was found to be 3.3 × 10−2 s−1, corresponding to an overall acceleration (kcat/kuncat) of 5.0 × 105 (where kuncat is the rate constant for the uncatalyzed reaction). Gold complex 4 and platinum complex 5 showed analogous kinetic profiles consistent with a Michaelis-Menten mechanism (figs. S10 to S12 and S18 to S20). The corresponding Lineweaver-Burk plot for 5 provided a kcat of 2.3 × 10−2 s−1, corresponding to a similar acceleration of 2.6 × 104 (Fig. 2D).

For complex 4, however, the rapidity of the reaction introduced substantial error into the estimation of kcat by the method of initial rates. Thus, we instead applied reaction progress kinetic analysis to generate a larger data set, affording an estimate for kcat of 3.4 s−1, corresponding to a kcat/kuncat of 1.9 × 107 (28). The data obtained in this way are consistent with the initial rate data (fig. S12) while providing a more robust measurement due to the expanded data set. This rate acceleration is on par with that of many enzymatic processes; for comparison, chymotrypsin has been shown to accelerate amide bond hydrolysis with 107-fold rate accelerations (29).

The extension of this phenomenon to a dual catalytic cross-coupling would represent a proof of principle for our initial hypothesis that a synthetic self-assembled supramolecular cocatalyst could be employed to overcome kinetically prohibitive barriers in otherwise desirable cross-coupling reactions (30). On the basis of the stoichiometric reactivity (see above), we envisioned a cocatalytic cross-coupling of a methyl electrophile with a complementary nucleophilic alkyl metal species. This goal presented several challenges: The supramolecular assembly 1 was found to decompose in the presence of methyl iodide, requiring the implementation of the previously reported analog 6, which bears catechol ligands with diminished electron density (Fig. 3A) (31). Identification of a nucleophile capable of transmetallating to PtII while remaining tolerant of both a protic solvent and the supramolecular catalyst eliminated several typical candidates (32), but stannanes were found to be suitable partners under these criteria. Subsequently, we discovered that the Me3SnI by-product formed upon transmetallation from tetramethyltin was a strong guest for 6, requiring the use of fluoride to generate Me3SnF and prevent catalyst inhibition. Under these conditions, we found that efficient C-C coupling occurred only in the presence of both the platinum and supramolecular catalysts (Fig. 3B). The presence of a radical trap (9,10-dihydroanthracene) did not inhibit the reaction, and an isotopic labeling study employing deuterated iodomethane indicated the incorporation of both coupling partners. An overall mechanism for this process is proposed in which the demonstrated microenvironment catalysis of reductive elimination from complex 5 is incorporated into a traditional organometallic catalytic cycle (Fig. 3C), enabling an otherwise prohibitively slow process. This strategy should prove general; a tailored synthetic microenvironment catalyst could be designed to recognize the rate-limiting transition state for other high-value catalytic processes.

Fig. 3 Demonstration of dual catalysis.

(A) Modified supramolecular assembly with improved stability to electrophiles. (B) Dual catalytic alkyl-alkyl cross-coupling enabled via supramolecular catalysis. (C) Proposed mechanism.

Supplementary Materials

Materials and Methods

Figs. S1 to S42


References and Notes

  1. 2-Br was employed rather than 2 because it was expected to show a higher affinity for the interior of 1, on the basis of previous observations in similar systems (19). The calculated value for kcat is necessarily identical for 2 and 2-Br because of the mechanistic convergence in the encapsulated intermediate.
  2. Acknowledgments: This research was supported by the Director, Office of Science, Office of Basic Energy Sciences and the Division of Chemical Sciences, Geosciences, and Biosciences of the U.S. Department of Energy at Lawrence Berkeley National Laboratory (grant DE-AC02-05CH11231) and NIH National Institute of General Medical Sciences (grant R01 GM073932). D.M.K. was supported by an NSF Graduate Research Fellowship Program (GRFP) (grant DGE 1106400), and M.D.L. was supported by the ARCS Foundation and an NSF GRFP. We thank J. N. Brantley and M. S. Winston for helpful discussions and C. G. Canlas for assistance with NMR experiments.
View Abstract

Navigate This Article