Solid‐Phase Conversion of Four Stereoisomers into a Single Enantiomer

Abstract Viedma ripening is an emerging method for the solid‐phase deracemization of mixtures of enantiomers. Up to now, the scope of the method has remained limited to molecules with a single stereocenter. We show here that this method can be extended to obtain a single enantiomer from a mixture of stereoisomers with two different stereocenters. In addition, we show that by using tailor‐made chiral additives, the conversion time can be reduced by a factor of 100.

and contain multiple stereocenters.C onversion of such compounds into as ingle enantiomer using such grinding experiments is obviously more challenging.T his is due to the process involving 2 n chiral compounds (with n the number of stereocenters), instead of only two.O ft hese 2 n compounds, one set of enantiomers will be the thermodynamically most favorable one,w hereas the other diastereomers will have ah igher energy (or higher solubility). Sakamoto et al. used this difference in stability for ar elated experiment. [7] They studied as ystem with two stereocenters,o fw hich one was enantiopure and could not epimerize,w hile the second one could epimerize in solution, resulting in as olid-phase that contained only as ingle diastereomer.O ther crystallization methods that lead to the formation of as ingle enantiomer exist as well. [8] In the case of Viedma ripening,itis important that the thermodynamically more stable set of enantiomers crystallizes as aracemic conglomerate ( Figure 1).
Thec rystallization behavior of the less stable diastereomers is expected to be less important, since these compounds will eventually be eliminated from the solid phase.M ore challenging is the interconversion (racemization) of the two enantiomers,s ince this requires epimerization of all chiral centers.T his can be achieved if the centers epimerize in a( near) identical way,o ri ft he conditions for the different epimerization pathways are compatible.A saf irst step towards multiple stereocenters,weherewith show asuccessful demonstration on two molecules with two different stereocenters to which these conditions apply.F rom at otal of four diastereomers,asingle one was obtained using grinding experiments.T ot he best of our knowledge,t his is the first example of conversion of astereoisomeric mixture of acompound with multiple stereocenters into only one enantiomer using such grinding experiments.S of ar,e xperiments by Hachiya et al. approximate this goal most closely,b ut using total spontaneous resolution instead. [9] They succeeded in partially converting molecules with two identical stereocenters (meaning their system consisted of only three stereoiso- Figure 1. In order to successfully convert acompound with two stereocenters into asingle stereoisomer using grinding experiments, epimerization of both stereocenters, as well as crystallization of the most stable pair of enantiomersa saracemic conglomerate, is required. mers (one pair of enantiomers (1)a nd the corresponding achiral meso compound) instead of four stereoisomers (two pairs of enantiomers), Figure 2). Compounds 2 and 3 are examples of molecules with two different stereocenters,w hich thus exhibit four distinguishable stereoisomers.T hey belong to the class of succinimides, some of which exhibit anticonvulsant properties.T hey are structurally closely related to phensuximide (imide 4), adrug used to treat epilepsy. [10] Related methylated succinimides are known for their antifungal activity,o fwhich the enantiopure (R,R)-configured succinimides are the most effective ones. [11] Obtaining such succinimides in enantiopure form is therefore highly relevant.
In case of compounds 2 and 3,t he trans diastereomers (3R,4R and 3S,4S)crystallize as aracemic conglomerate.F or compound 2,t he cis isomers (3R,4S and 3S,4R)f orm ar are example of as olid solution. [12] It is envisioned that the trans diastereomers are more stable,since this configuration is less sterically hindered. NMR experiments confirmed this hypothesis.
Epimerization of both stereocenters can be achieved by adding DBU (1,8-diazabicyclo-[5.4.0]undec-7-ene)a sabase. When exposing as olution of compound 2 to epimerization conditions,o nly 2% was present as the cis isomer at room temperature as observed in the 1 HNMR spectrum. This corresponds to the conglomerate trans form being more stable by 2.5 kcal mol À1 (in solution), thus making this system potentially suitable for Viedma ripening ( Figure 3).
Since the synthesis of compound 2 solely yielded the cis diastereomer (through hydrogenation of the corresponding maleimide derivative), this isomer was used as the starting point for the grinding experiments.U pon addition of the racemization catalyst (DBU), the dissolved cis diastereomers were epimerized into the trans form. Further dissolution and subsequent epimerization of cis-2 resulted in supersaturation and consequent crystallization of trans-2.T his process happened relatively fast in at ime span of several minutes ( Figure 4).
Using temperature-dependent selective exchange spectroscopy measurements,itwas found that the transition-state barrier for the conversion of cis to trans was only DG = 17.0 kcal mol À1 .T his results in ah alf-life of approximately one second (using the Eyring equation). Thelimiting factor in this process is thus the dissolution, rather than the epimerization of the cis compound. Next, Viedma ripening conditions resulted in deracemization of the crystals to provide only one of the two trans enantiomers.R acemization of one trans isomer into the enantiomeric trans isomer happens at one center at atime,through sequential epimerization of the two chiral centers.T he cis isomer is thus an intermediate in the racemization pathway (Figure 3). Theracemization barrier of this process was calculated to be 19.5 kcal mol À1 (see the Supporting Information). Thed eracemization process took several days,s howing the exponential behavior typical for Viedma ripening.T he long deracemization time is largely caused by the long "dead time" before symmetry breaking occurs.Deracemization proceeds stochastically,meaning that different enantiomers are obtained as the crystalline product in different experiments.Insimilar grinding experiments,the production of asingle enantiomer could also be achieved for Figure 2. Deracemization of compounds of type 1 was previously studied by Hachiya et al. [12] This study reports the conversion of compounds 2 and 3 into single enantiomers, both of which are structurally closely related to the anticonvulsant drug phensuximide (4). Figure 3. Epimerization and racemization of compound 2 takes place via reversible deprotonation of the stereocenters using DBU as acatalyst. In this case, the two epimerization rates are nearly identical. The indicated Gibbs free energies were determined using temperaturedependent selective exchange spectroscopy.

Angewandte Chemie
Communications amixture of diastereoisomers of imide 3 (see Figure S3 in the Supporting Information).
Since it is synthetically more useful to be able to direct the outcome of aV iedma ripening experiment towards the desired enantiomer,w ee xplored the possibility of using chiral additives,astrategy that was previously successfully applied to achieve this goal (alternatively,asmall amount of enantiopure product might be added to achieve this same goal). [13] As uitable chiral additive with high affinity for only one of the two crystal forms may act as as elective growth inhibitor. [14] In practice,m olecules that closely resemble one of the two enantiomers are selected for this purpose. Importantly,t he chiral additive should not alter its stereochemical configuration during the experiment.
In this study,ac ombination of four enantiopure chiral additives was synthesized, [15] all of which closely resemble the same enantiomer of compound 2 ( Figure 5). All additives contain three chiral centers,o fw hich one (carrying the hydroxy group) cannot epimerize under the influence of DBU.T he locked chirality of this stereocenter also ensures the desired chirality of the other two centers (since epimerization of these centers would result in sterically unfavored cis orientations;s ee Supporting Information). Using these tailor-made additives,t he outcome of the grinding experiments always proceeded in the desired direction. The combination of the four additives based on (R,R)-2 always resulted in (S,S)-2 as the product, while (S,S)-based additives resulted in (R,R)-2.T his phenomenon of opposing chirality has been extensively studied and is known as the Lahav rule of reversal. [13b,16] Since ac ombination of four additives was used in all experiments, [17] we could not determine which of the four was most effective.T he deracemization curve shows linear behavior, which is typical for the use of chiral additives. [18] Not only could the desired enantiomer be obtained using this approach, but the time required for deracemization was also significantly reduced ( Figure 6). When starting from the racemic trans diastereomers,t he deracemization time was reduced from 70 to 2.5 hours.When starting from cis-2,t he conversion time was even reduced to less than one hour, in other words ar eduction of almost afactor 100 was achieved.
This increased conversion speed is induced in the initial phase of the process.F ast epimerization of the cis diastereomer results in as upersaturated solution of the trans enantiomers.S ince the rate of cis-to-trans epimerization is higher than that of racemization, this results in crystallization of both trans enantiomers in the absence of an additive.I nt he presence of the additive,h owever,c rystallization of the enantiomer similar to the additive is hampered, thereby resulting in preferential crystallization of the other enantiomer. [19] This results in as olid phase that already has ah igh starting ee (around 25 %a fter adding the racemization catalyst) when the Viedma ripening commences.T he time required to obtain as ingle enantiomer is therefore even shorter than with the use of an additive starting from trans-2.
In conclusion, we have shown that the formation of one single isomer out of four stereoisomers using isomerization and subsequent Viedma ripening is possible.T he requirements are similar to those for compounds with only as ingle stereocenter but can in practice be difficult. Thedesired pair of enantiomers should be the thermodynamically most stable pair and should also crystallize as ar acemic conglomerate. However,w hen the most stable pair of enantiomers crystallizes as ar acemic compound, derivatization [20] or salt formation [21] can additionally be used to convert it into ac onglomerate.R acemization can be more challenging,s ince epimerization of all chiral centers should proceed under the same conditions.These principles hold not only for amolecule with two stereocenters,b ut also for more stereocenters.A s long as these requirements are met, Viedma ripening may be used for the deracemization of any compound with multiple stereocenters.

Conflict of interest
Theauthors declare no conflict of interest.  Curves showing the conversion of compound 2 into asingle enantiomer,using 5mol %o fthe chiral additives 5a-d (ee stands for enantiomeric excess; + 100 % ee implies only RR while À100 % ee means only SS). Experiments were started from either the racemic trans-(R,R/S,S)o rcis-(R,S/S,R)diastereomers. Within aminute after adding the racemization catalyst, all experiments contained only the trans diastereomer. When additives were used, 5a-d were always used as acombination.P lease note the two different time scales on the xaxis.