Olefin Metathesis

OLEFIN METATHESIS

In the last 10 years or so an exciting new strategy has emerged for the formation of carbon–carbon double bonds, namely olefin metathesis. This work grew out of the development of Ziegler–Natta catalysts for the polymerizarion of cyclic olefins. It was found that when 2-pentene was treated with a catalyst prepared from tungsten hexachloride and ethylaluminum dichloride, a mixture of 2-pentene, 2-butene, and 3-hexene was produced in minutes at room tempera-ture (rt)!

It was shown that the mixture was an equilibrium mixture. Thus it appears that the alkenes are being broken apart at the double bonds and the pieces reassembled randomly. This process was termed olefin metathesis because the ends of the carbon–carbon double bonds are being interchanged.

It was subsequently shown that the active catalytic species is a metallocarbene complex which contains a carbon–metal double bond. This catalyst undergoes a 2+2 addition with an alkene to give two metallocyclobutanes 1 and 2 (below) which depend on the regiochemistry of addition. Metallocyclobutane 1 is a non-productive intermediate since cleavage across the ring by A or B gives back the reactants. Metallocyclobutane 2 can cleave across the ring A’ to give the original alkene and metallocarbene or it can cleave in the opposite direction B to give a new alkene and a new metallocarbene. The new metallocarbene can add the starting material to give two metallocyclobutanes 3 and 4. It is seen that 3 is a nonproductive intermediate whereas 4 can give another new alkene by cleavage in the B direction. Since each step is reversible, it can be seen that ultimately the ends of the double bond will be exchanged to give a mixture of olefins.

Olefin metathesis is a unique reaction and is only possible by transition metal catalysis. In fact only complexes of Mo, W, Re, and Ru are known to cat-alyze olefin metathesis. Once it was known that metallocarbenes were the actual catalytic species, a variety of metal carbene complexes were prepared and evalu-ated as catalysts. Two types of catalysts have emerged as the most useful overall. The molybdenum-based catalysts developed by Schrock and ruthenium-based catalysts developed by Grubbs.

Both are stable metallocarbene complexes, but they have very different reac-tivity profiles. The molybdenum catalyst is highly reactive and is effective with sterically demanding olefins. Its drawbacks are that it is not highly tolerant of diverse functional groups and has high sensitivity to air, moisture, and solvent impurities. The ruthenium system, on the other hand, is catalytically active in the presence of water or air, and it exhibits a remarkable functional group tolerance. It is not a reactive as the molybdenum catalyst, particularly toward sterically bulky substrates. However, it is readily available and is the reagent of choice for all but the most difficult substrates.

At present the most synthetically important metathesis process is ring-closing metathesis (RCM) as it can be used for the preparation of a wide variety of medium to large ring compounds from acyclic diene precursors. For such dienes there are two competing processes. The first is RCM, in which the two olefinic units undergo intramolecular reaction to give a cyclic product. The second is intermolecular metathesis, which yields open-chain dimers or oligomers.

The rate of formation of the RCM product is given by a first-order rate law since the metathesis is intramolecular, rate_RCM = k₁ [catalyst][diene]. The rate of dimer formation is a second-order rate law because the reaction is inter-molecular, rate_dimer = k₂ [catalyst][diene]². Although these two rate laws are not strictly comparable because of their different orders, at normal concentrations (0.25–1 M) the difference in rates is mostly due to the difference in the rate con-stants k₁ and k₂. Given that the double-bond energies in five- and six-membered rings are very similar to the bond energy of an open-chain double bond, the ΔH ^± for the two processes are similar. Thus ΔH^± cannot be responsible for a significant difference in the rates of the two reactions.

The rate constant k₁ for the RCM process is greater than the dimerization rate constant k₂ because of a distinct entropic advantage. In RCM the two reacting bonds are present in the same molecule and two molecules (the product and ethylene) are formed from one; thus RCM proceeds with a gain in entropy. The dimerization process causes loss of translational degrees of freedom because one molecule is formed from two and thus occurs with a loss of entropy. As a result the formation of five- and six-membered rings have larger rate constants and thus proceed at faster rates than dimerization under normal conditions.

The use of the Grubbs catalyst is largely restricted to terminal dienes. In these cases the by-product is ethylene. Since this is an equilibrium process, the ethylene diffuses out of solution, which helps drive the reaction to completion. Nevertheless the catalyst is highly tolerant of functional groups and generally gives good yields.

The formation of medium rings presents challenges for this and most ring-forming reactions. There are two major reasons for this. The first is that the molecule must be able to adopt a reactive conformation in which the two double bonds are in close proximity for reaction to occur. For larger ring sizes where the distance between double bonds increases, the population of reactive conforma-tions decreases so the rate slows as well. The second factor is the increased ring strain in medium rings which develops at the transition state and thus raises ΔH^± and slows the rate of the intramolecular process. The rate of the intermolecular dimerization remains roughly the same. Thus the rate of the intramolecular pro-cess for medium rings slows dramatically while the dimerization rate remains the same (for a given concentration).

This behavior is easily demonstrated by the following examples. Formation of a five-membered ring takes place easily under mild conditions. Formation of a similar eight-membered ring from an open-chain precursor gave no RCM product but only dimerization. In contrast bis-allylcatechol gives the eight-membered ring RCM product efficiently. The phenyl ring provides a conformational constraint which keeps the olefins in reasonably close proximity and the lack of C–H bonds in the catechol fragment reduces transannular strain in the eight-membered ring. These factors allow the RCM to be favored over dimerization.

A second way to increase the proportion of RCM is to keep the concentration of the starting diene as low as is feasible. Because dimerization is a second-order reaction, decreasing the concentration of the diene decreases the rate of the second-order reaction much faster than the rate of the first-order RCM reaction decreases. Thus carrying out the reaction at very low concentrations can actually result in the RCM becoming significantly faster than dimerization. There are several ways to do this experimentally. The first is to simply dilute the reaction mixture to the desired concentration and then run the reaction normally. Although simple, this approach requires a lot of solvent which must be paid for and then removed. A simpler process is to use very slow addition of a solution of the diene to a solution of the catalyst. If the diene reacts as it is added, then the concentration of the diene remains very low over the course of the addition without having to use large volumes of solvent to achieve it.

For large rings where ring strain is again minimal, only the effective concen-tration of reactive conformers limits the RCM process. In these cases, running the reaction at high dilution slows the dimerization sufficiently to allow successful formation of large rings. This is truly a remarkable ability since large rings are difficult to access by other reactions.

The effect of conformation on the efficiency of RCM is extremely important. The RCM of the polyether diene shown below gives only a 39% yield under normal conditions. Addition of lithium ions to the reaction mixture causes the ether oxygens to wrap around the lithium cation. This forces the olefinic bonds into close proximity and the RCM product is now formed in >95% yield. Other structural features such as hydrogen bonding, restricted rotation (as in amides), or steric effects which cause particular conformations to be favored are very important in the formation of large ring systems. Such effects can affect RCM either positively or negatively depending on whether the olefinic bonds are held proximal or distal to each other.

The development of RCM has dramatically changed synthetic planning for ring systems, particularly medium and large ring systems. It represents a new paradigm for assembling carbon skeletons of ring compounds. Further advances will only increase its importance as a method for carbon–carbon bond formation.