Kinetic resolution

The first reported kinetic resolution was achieved by Louis Pasteur. After reacting aqueous racemic ammonium tartrate with a mold from Penicillium glaucum, he reisolated the remaining tartrate and found it was levorotatory.[3] The chiral microorganisms present in the mold catalyzed the metabolization of (R,R)-tartrate selectively, leaving an excess of (S,S)-tartrate.

Kinetic resolution by synthetic means was first reported by Marckwald and McKenzie in 1899 in the esterification of racemic mandelic acid with optically active (−)-menthol. With an excess of the racemic acid present, they observed the formation of the ester derived from (+)-mandelic acid to be quicker than the formation of the ester from (−)-mandelic acid. The unreacted acid was observed to have a slight excess of (−)-mandelic acid, and the ester was later shown to yield (+)-mandelic acid upon saponification. The importance of this observation was that, in theory, if a half equivalent of (−)-menthol had been used, a highly enantioenriched sample of (−)-mandelic acid could have been prepared. This observation led to the successful kinetic resolution of other chiral acids, the beginning of the use of kinetic resolution in organic chemistry.[4][5]

Kinetic resolution is a possible method for irreversibly differentiating a pair of enantiomers due to (potentially) different activation energies. While both enantiomers are at the same Gibbs free energy level by definition, and the products of the reaction with both enantiomers are also at equal levels, the ${\displaystyle \Delta G^{\ddagger }}$, or transition state energy, can differ. In the image below, the R enantiomer has a lower ${\displaystyle \Delta G^{\ddagger }}$ and would thus react faster than the S enantiomer.

The ideal kinetic resolution is that in which only one enantiomer reacts, i.e. k_R>>k_S. The selectivity (s) of a kinetic resolution is related to the rate constants of the reaction of the R and S enantiomers, k_R and k_S respectively, by s=k_R/k_S, for k_R>k_S. This selectivity can also be referred to as the relative rates of reaction. This can be written in terms of the free energy difference between the high- and low-energy transitions states, ${\displaystyle \Delta \Delta G^{\ddagger }}$.[6]

${\displaystyle s={\frac {k_{R}}{k_{S}}}=e^{\Delta \Delta G^{\ddagger }/RT}}$

The selectivity can also be expressed in terms of ee of the recovered starting material and conversion (c), if first-order kinetics (in substrate) are assumed. If it is assumed that the S enantiomer of the starting material racemate will be recovered in excess, it is possible to express the concentrations (mole fractions) of the S and R enantiomers as

${\displaystyle [S]={\frac {(1+ee)(1-c)}{2}}}$

${\displaystyle [R]={\frac {(1-ee)(1-c)}{2}}}$

where ee is the ee of the starting material. Note that for c=0, which signifies the beginning of the reaction, ${\displaystyle S_{0}=R_{0}={\frac {1}{2}}}$, where these signify the initial concentrations of the enantiomers. Then, for stoichiometric chiral resolving agent B*,

${\displaystyle {\frac {d[S]}{dt}}=-k_{S}[S][B^{*}]\implies \log[S]=-k_{S}[B^{*}]t+\log S_{0}}$

Note that, if the resolving agent is stoichiometric and achiral, with a chiral catalyst, the [B*] term does not appear. Regardless, with a similar expression for R, we can express s as

${\displaystyle s={\frac {k_{R}}{k_{S}}}={\frac {\log[R]-\log R_{0}}{\log[S]-\log S_{0}}}={\frac {\log[(1-c)(1-ee)]+\log {\frac {1}{2}}-\log R_{0}}{\log[(1-c)(1+ee)]+\log {\frac {1}{2}}-\log S_{0}}}={\frac {\log[(1-c)(1-ee)]}{\log[(1-c)(1+ee)]}}}$

If we wish to express this in terms of the enantiomeric excess of the product, ee", we must make use of the fact that, for products R' and S' from R and S, respectively

${\displaystyle ee''={\frac {[R']-[S']}{[R']+[S']}}={\frac {ee(1-c)}{c}}\implies ee=ee''{\frac {c}{1-c}}}$

From here, we see that

${\displaystyle 1-ee={\frac {1-c-cee''}{1-c}}}$

${\displaystyle 1+ee={\frac {1-c+cee''}{1-c}}}$

which gives us

${\displaystyle (1-c)(1-ee)=1-c(1+ee'')}$

${\displaystyle (1-c)(1+ee)=1-c(1-ee'')}$

which, when we plug into our expression for s derived above, yield

${\displaystyle s={\frac {\log[1-c(1+ee'')]}{\log[1-c(1-ee'')]}}}$

Additionally, the expressions for c and ee can be parametrized to give explicit expressions for C and ee in terms of t. First, solving explicitly for [S] and [R] as functions of t yields

${\displaystyle {\frac {d[S]}{dt}}=-k_{S}[S]\implies S={\frac {1}{2}}e^{-k_{S}t}}$

which, plugged into expressions for ee and c, gives

${\displaystyle ee={\frac {[S]-[R]}{[S]+[R]}}={\frac {e^{-k_{S}t}-e^{-k_{R}t}}{e^{-k_{S}t}+e^{-k_{R}t}}}}$

${\displaystyle c=1-{\big (}[S]-[R]{\big )}=1-{\frac {e^{-k_{S}t}+e^{-k_{R}t}}{2}}}$

Without loss of generality, we can allow k_S=1, which gives k_R=s, simplifying the above expressions. Similarly, an expression for ee″ as a function of t can be derived

${\displaystyle ee''={\frac {1-c}{c}}ee={\frac {e^{-t}+e^{-st}}{2-e^{-t}-e^{-st}}}\left({\frac {e^{-t}-e^{-st}}{e^{-t}+e^{-st}}}\right)}$

Thus, plots of ee and ee″ vs. c can be generated with t as the parameter and different values of s generating different curves, as shown below.

As can be seen, high enantiomeric excesses are much more readily attainable for the unreacted starting material. There is however a tradeoff between ee and conversion, with higher ee (of the recovered substrate) obtained at higher conversion, and therefore lower isolated yield. For example, with a selectivity factor of just 10, 99% ee is possible with approximately 70% conversion, resulting in a yield of about 30%. In contrast, in order to get good ee's and yield of the product, very high selectivity factors are necessary. For example, with a selectivity factor of 10, ee″ above approximately 80% is unattainable, and significantly lower ee″ values are obtained for more realistic conversions. A selectivity in excess of 50 is required for highly enantioenriched product, in reasonable yield.

This is a simplified version of the true kinetics of kinetic resolution. The assumption that the reaction is first order in substrate is limiting, and it is possible that the dependence on substrate may depend on conversion, resulting in a much more complicated picture. As a result, a common approach is to measure and report only yields and ee's, as the formula for k_rel only applies to an idealized kinetic resolution. It is simple to consider an initial substrate-catalyst complex forming, which could negate the first-order kinetics. However, the general conclusions drawn are still helpful to understand the effect of selectivity and conversion on ee.

With the advent of asymmetric catalysis, it is necessary to consider the practicality of utilizing kinetic resolution for the preparation of enantiopure products. Even for a product which can be attained through an asymmetric catalytic or auxiliary-based route, the racemate may be significantly less expensive than the enantiopure material, resulting in heightened cost-effectiveness even with the inherent "loss" of 50% of the material. The following have been proposed as necessary conditions for a practical kinetic resolution:[6]

• inexpensive racemate and catalyst
• no appropriate enantioselective, chiral pool, or classical resolution route is possible
• resolution proceeds selectively at low catalyst loadings
• separation of starting material and product is facile

To date, a number of catalysts for kinetic resolution have been developed that satisfy most, if not all of the above criteria, making them highly practical for use in organic synthesis. The following sections will discuss a number of key examples.

Dynamic kinetic resolution (DKR) occurs when the starting material racemate is able to epimerize easily, resulting in an essentially racemic starting material mix at all points during the reaction. Then, the enantiomer with the lower barrier to activation can form in, theoretically, up to 100% yield. This is in contrast to standard kinetic resolution, which necessarily has a maximum yield of 50%. For this reason, dynamic kinetic resolution has extremely practical applications to organic synthesis. The observed dynamics are based on the Curtin-Hammett principle. The barrier to reaction of either enantiomer is necessarily higher than the barrier to epimerization, resulting in a kinetic well containing the racemate. This is equivalent to writing, for k_R>k_S,

${\displaystyle \Delta G_{rac}<\Delta G_{R}^{\ddagger }<\Delta G_{S}^{\ddagger }}$

A number of excellent reviews have been published, most recently in 2008, detailing the theory and practical applications of DKR.[38][39][40]

Noyori asymmetric hydrogenation

The Noyori asymmetric hydrogenation of ketones is an excellent example of dynamic kinetic resolution at work. The enantiomeric β-ketoesters can undergo epimerization, and the choice of chiral catalyst, typically of the form Ru[(R)-BINAP]X₂, where X is a halogen, leads to one of the enantiomers reacting preferentially faster. The relative free energy for a representative reaction is shown below.[41][42] As can be seen, the epimerization intermediate is lower in free energy than the transition states for hydrogenation, resulting in rapid racemization and high yields of a single enantiomer of the product.

The enantiomers interconvert through their common enol, which is the energetic minimum located between the enantiomers. The shown reaction yields a 93% ee sample of the anti product shown above. Solvent choice appears to have a major influence on the diastereoselectivity, as dichloromethane and methanol both show effectiveness for certain substrates. Noyori and others have also developed newer catalysts which have improved on both ee and diastereomeric ratio (dr).

(S)-SYNPHOS

Genêt and coworkers developed SYNPHOS, a BINAP analogue which forms ruthenium complexes, which perform highly selective asymmetric hydrogenations.[43] Enantiopure Ru[SYNPHOS]Br₂ was shown to selectively hydrogenate racemic α-amino-β-ketoesters to enantiopure aminoalcohols, as shown below utilizing (R)-SYNPHOS.[44] 1,2-syn amino alcohols were prepared from benzoyl protected amino compounds, whereas anti products were prepared from hydrochloride salts of the amine.

Fu acylation modification

Ruthenium catalyst used by Fu for racemization of secondary alcohols

Recently, Gregory Fu and colleagues reported a modification of their earlier kinetic resolution work to produce an effective dynamic kinetic resolution.[45] Using the ruthenium racemization catalyst shown to the right, and his planar chiral DMAP catalyst, Fu has demonstrated the dynamic kinetic resolution of secondary alcohols yielding up to 99% and 93% ee, as shown below. Work is ongoing to further develop the applications of the widely used DMAP catalyst to dynamic kinetic resolution.

Enzymatic dynamic kinetic resolutions

A number of enzymatic dynamic kinetic resolutions have been reported.[46] A prime example using PSL effectively resolves racemic acyloins in the presence of triethylamine and vinyl acetate as the acylating agent.[47] As shown below, the product was isolated in 75% yield and 97% ee. Without the presence of the base, regular kinetic resolution occurred, resulting in 45% yield of >99% ee acylated product and 53% of the starting material in 92% ee.

Another excellent, though not high-yielding, example is the kinetic resolution of (±)-8-amino-5,6,7,8-tetrahydroquinoline. When exposed to Candida antarctica lipase B (CALB) in toluene and ethyl acetate for 3–24 hours, normal kinetic resolution occurs, resulting in 45% yield of 97% ee of starting material and 45% yield of >97% ee acylated amine product. However, when the reaction is allowed to stir for 40–48 hours, racemic starting material and >60% of >95% ee acylated product are recovered.[48]

Here, the unreacted starting material racemizes in situ via a dimeric enamine, resulting in a recovery of greater than 50% yield of the enantiopure acylated amine product.

Chemoenzymatic dynamic kinetic resolutions

There have been a number of reported procedures which take advantage of a chemical reagent/catalyst to perform racemization of the starting material and an enzyme to selectively react with one enantiomer, called chemoenzymatic dynamic kinetic resolutions.[49] PSL-C was utilized along with a ruthenium catalyst (for racemization) to produce enantiopure (>95% ee) δ-hydroxylactones.[50]

Bäckvall's ruthenium catalyst for racemization in chemoenzymatic dynamic kinetic resolution of secondary alcohols

More recently, secondary alcohols have been resolved by Bäckvall with yields up to 99% and ee's up to >99% utilizing CALB and a ruthenium racemization complex.[51]

A second type of chemoenzymatic dynamic kinetic resolution involves a π-allyl complex from an allylic acetate with palladium. Here, racemization occurs with loss of the acetate, forming a cationic complex with the transition metal center, as shown below.[52] Palladium has been shown to facilitate this reaction, while ruthenium has been shown to affect a similar reaction, also shown below.[53]

In parallel kinetic resolution (PKR), a racemic mixture reacts to form two non-enantiomeric products, often through completely different reaction pathways. With PKR, there is no tradeoff between conversion and ee, as the formed products are not enantiomers.[54][55] One strategy for PKR is to remove the less reactive enantiomer (towards the desired chiral catalyst) from the reaction mixture by subjecting it to a second set of reaction conditions that preferentially react with it, ideally with an approximately equal reaction rate. Thus, both enantiomers are consumed in different pathways at equal rates. PKR experiments can be stereodivergent, regiodivergent, or structurally divergent.[56] One of the most highly efficient PKR's reported to date was accomplished by Yoshito Kishi in 1998; CBS reduction of a racemic steroidal ketone resulted in stereoselective reduction, producing two diastereomers of >99% ee, as shown below.[57]

PKR have also been accomplished with the use of enzyme catalysts. Using the fungus Mortierella isabellina NRRL 1757, reduction of racemic β-ketonitriles affords two diastereomers, which can be separated and re-oxidized to give highly enantiopure β-ketonitriles.[58] Highly synthetically useful parallel kinetic resolutions have truly yet to be discovered, however. A number of procedures have been discovered that give acceptable ee's and yields, but there are very few examples which give highly selective parallel kinetic resolution and not simply somewhat selective reactions. For example, Fu's parallel kinetic resolution of 4-alkynals yields very enantioenriched cyclobutanone in low yield and slightly enantioenriched cyclopentenone, as shown below.[59]

In theory, parallel kinetic resolution can give the highest ee's of products, since only one enantiomer gives each desired product. For example, for two complementary reactions both with s=49, 100% conversion would give products in 50% yield and 96% ee. These same values would require s=200 for a simple kinetic resolution. As such, the promise of PKR continues to attract much attention. The Kishi CBS reduction remains one of the few examples to fulfill this promise.

• Chiral auxiliaries
• Chiral pool synthesis
• Chiral resolution
• Enantioselective synthesis

• Dynamic Kinetic Resolutions. A MacMillan Group Meeting. Jake Wiener Link
• Dynamic Kinetic Resolution:A Powerful Approach to Asymmetric Synthesis. Erik Alexanian Supergroup Meeting March 30, 2005 Link
• Dynamic Kinetic Resolution: Practical Applications in Synthesis. Valerie Keller 3rd-Year Seminar November 1, 2001 Link
• Kinetic Resolution. David Ebner Stoltz Group Literature Seminar. June 4, 2003 link
• Kinetic Resolutions. UT Southwestern Presentation. link