Evaluation of the use of mandelate derivatives to determine the enantiomeric purity and the absolute configuration of secondary cyclohexenols

The use of mandelate derivatives to determine the enantiomeric purity and the absolute configuration of isomeric mixtures of bicyclic and monocyclic secondary cyclohexenols was investigated. Synthesis, NMR and conformational analyses of the derivatives were performed and Gauge-Independent Atomic Orbital (GIAO) 1 H NMR Boltzmann weighted average chemical shifts were computed. The studied methodology proved to be of practical value for most of the systems under study.


Introduction
NMR spectroscopy is one of the most useful and widely used methods for the determination of the optical purity and the absolute configuration of enantiomerically pure compounds. 1The methodology relies on the derivatization of the studied compound with a chiral derivatizing agent (CDA) and comparison of the chemical shifts of the resulting diastereoisomers.The mandelate derivatives of secondary alcohols can be conveniently prepared from the readily available O-substituted mandelic acids. 2 In particular, the use of O-acetyl analogues has been shown to be straightforward and give reliable results.2d Our interest in the development of asymmetric Diels-Alder reactions of boron-activated dienophiles has led us to examine the properties of the mandelate derivatives of different bicyclic and monocyclic secondary cyclohexenol products. 3Cyclohexenols are versatile building blocks for the synthesis of natural and pharmaceutical products.In a previous communication we reported that 1 H NMR spectroscopy of the O-acetylmandelate and mandelate derivatives could be efficiently used to determine the optical purity and to predict the absolute configuration of endo-and exo-norborn-5-en-2-ol 1, both as a mixture and also as separate diastereomers. 4The signals of the bridgehead protons attached to C1 of the four diastereomeric O-acetylmandelates 2 were nicely resolved (Scheme 1).In addition, the signals of C3-H were clearly separated, and the same was observed for the olefinic protons attached to C6 for the endo isomer.Subsequent selective hydrolysis of the acetate group gave the mixture of mandelates 3 without apparent epimerization.The spectra of the mandelates showed s ( = R -S) even higher than those for the O-acetylmandelate analogues (Scheme 1).The conformational properties of these derivatives were analyzed using theoretical methods, showing that all major conformers were synperiplanar, as anticipated by the empirical model proposed for secondary Omethylmandelates.The methodology was finally validated by computing GIAO 1 H NMR Boltzmann weighted average chemical shifts, which were in good agreement with the experimental  and  values.

Scheme 1
As an extension of our previous work, in this paper we wish to report the results of the application of this experimental/theoretical protocol to determine the enantiomeric ratios and to predict the absolute configuration of other bicyclic and monocyclic secondary cyclohexenols. 5

Results and Discussion
Synthesis of mandelate derivatives and NMR analysis (a) Bicyclo[2.2.2]oct-5-en-2-ol.We first studied the synthesis and use of the O-acetylmandelates of endo-and exo-bicyclo[2.2.2]oct-5-en-2-ol 4. Reaction of racemic bicyclo[2.2.2]oct-5-en-2-ol (endo/exo 73:27) with (S)-O-acetylmandelic acid, 6 N,N′-dicyclohexylcarbodiimide (DCC) and catalytic 4-(dimethylamino)pyridine (DMAP) in dichloromethane (DCM) gave a mixture of diastereomeric O-acetylmandelates 5 in 70% yield (Scheme 2). 7Integration of the 1 H NMR spectrum indicated that the 73:27 endo/exo and 50:50 R/S ratios were maintained.We were disappointed to note that none of the signals of the four diastereoisomers of O-acetylmandelates 5 were well resolved in the 1 H NMR spectrum.Therefore, O-acetylmandelates derivatives were found to be unsuitable to quantify the molar fraction of each compound in the mixture.Subsequent selective hydrolysis of the acetate group with potassium carbonate gave the diastereoisomeric mixture of mandelates 6 without noticeable epimerization.Fortunately, in this case the signals of the bridgehead protons attached to C1 for the 6NR and 6XR diastereoisomers were baseline resolved (Scheme 2).In addition, the signal of the C6-H corresponding to 6NS was clearly separated.Finally, the molar fraction of compound 6XS could be computed by subtracting the integral of the C1-H signal of 6XR (2.79-2.72 ppm) to the value corresponding to the protons attached to C2 of both exo diastereoisomers (4.80-4.68ppm).Alternatively, the signal appearing at 5.20-5.13ppm for C-10H of 6XR and 6XS could be used.We then applied the same methodology to monocylic cyclohexenol 2,5-diphenyl-cyclohex-3-en-1-ol (endo/exo 47:53) 7. The diastereoisomeric mixture of O-acetylmandelates 8 was quantitatively prepared using standard conditions (Scheme 3).Analysis of the 1 H NMR spectra showed that the C2-H signals of 8NR and 8XS and the C5-H signal corresponding to 8XR were all well resolved.The molar fraction of diastereoisomer 8NS could be computed by subtracting the integral value corresponding to the C2-H signal of 8NR (3.99-3.91) to the integral value corresponding to the protons attached to C1 for both endo isomers (5.46-5.26ppm).Attempts to conduct the partial hydrolysis of O-acetylmandelates 8 to the mandelates using potassium carbonate were unsuccessful.(c) 3-and 4-Methyl-3-cyclohexen-1-ol.Finally, we synthesized the O-acetylmandelates of a 34:66 mixture of regiosiomers 3-and 4-methyl-3-cyclohexen-1-ol 9 with 93% yield (Scheme 4).However, only the ratio of the major para isomers could be quantified in the mixture.The C3-H signals in the 1 H NMR spectrum were well resolved for 10PR and 10PS (5.28-5.20 ppm and 5.20-5.12ppm respectively).To our regret, use of the mandelates obtained after partial hydrolysis of mixture 10 proved unsuccessful too since none of the signals of the corresponding four isomers were well resolved.

Theoretical calculations: conformational analysis and GIAO NMR calculations
To validate these experiments, we performed theoretical DFT calculations as done in our previous study for the mandelate analogues of endo-and exo-norborn-5-en-2-ol 1. 4 Conformational searches were run to locate the minimum energy conformers of the studied derivatives.Initially, a large number of geometries were generated using the conformational search module of Hyperchem 8 with the MM+ method.Selected structures were then successively reoptimized at the RHF/AM1, RHF/3-21G and B3LYP/6-31G* levels of theory using Gaussian 03. 9 Normal coordinate analyses were carried out to confirm the nature of the stationary points and to evaluate the thermochemical properties at 1 atm and 298.15K including zero-point energies (ZPEs) without scaling.Finally, GIAO NMR calculations at the B3LYP/6-31G* level of theory were performed for all significantly populated conformers of each diastereoisomer.This is in good agreement with our previous results for the derivatives of norborn-5-en-2-ol (1). 4 For each compound, we found two synperiplanar conformers of similar energies, having H-C2-O-C9 torsion angles () of ca.40º and -40º.In some cases, antiperiplanar conformations were also located, but the relative energies of these structures were much higher so that they do not contribute to the population.Consequently, they were not considered in the GIAO NMR calculation.As expected, all structures showed a H-bond interaction between the free hydroxy group of the mandelate and the carbonyl oxygen.
The Boltzmann weighted average 1 H NMR chemical shifts were computed using both synperiplanar conformers for each compound and are gathered in Table 1, together with the corresponding experimental values.As can be seen, the calculated and experimental and values correlate very well. 12The most interesting results arise from the shielding of H-6 for 6NS and the deshielding of H-1 for the endo and exo diastereosiomers having the R absolute configuration at the carbinolic proton (6NR and 6XR) relative to their diastereoisomer having the opposite configuration, which is correctly predicted by the calculations.The simple observation of the molecular models in Figure 1 evidences the greater spatial proximity of the phenyl ring to H-1 in the 6NS and 6XS counterparts and also to H-6 in 6NS.Although the endo and exo protons attached to C-3 are affected by the anisotropy generated by the aromatic ring too, the corresponding signals appear in a complex region of the NMR spectrum and are overlapped by the signals of H-7 and H-8 so they cannot be used for the purpose of this study.2 (for all conformers, see the Supplementary Material).Again, they all have a synperiplanar arrangement of the carbinolic proton, the carbonyl and the oxygen of acetate group, as expected from the empirical model.2a,2c However, contrary to the mandelates of bicyclo[2.2.2]oct-5-en-2-ol 6, in this case only one synperiplanar conformation was found for each compound.In addition, conformations with antiperiplanar arrangements of the carbonyl of the mandelate and the oxygen of the acetate contributed to the population of O-acetylmandelates 8 to different extents (up to 32%).For instance, two antiperiplanar conformations of similar energy that represent 21% and 11% of the population were located for 8XS.As has been previously observed for other systems, 4 this corroborates that O-acetylmandelates are more flexible than mandelates, which are locked in synperiplanar conformations by the H-bond interaction between the carbonyl oxygen and the free hydroxyl group.It is interesting to note that the O-acetylmandelate substituent adopts a pseudoequatorial position in the endo isomers, while it occupies a pseudo-axial position in the exo analogues.This might have consequences on the success of the application of the proposed methodology, since the phenyl group of the O-acetylmandelate moiety is correctly positioned to influence the chemical shifts of other nuclei in the cyclohexene ring in exo structures, while in their endo counterparts it might be too far away. 13

8XS 8XR
Since antiperiplanar conformations are significant for O-acetylmandelates 8, they had to be considered for computing the Boltzmann weighted average 1 H NMR chemical shifts.Gratifyingly, the experimental and  values were correctly reproduced, as can be observed from Table 2. 12 In this case, the shielding of H-5 in 8XR and of H-2 in 8XS suggested by the models depicted in Figure 2 is correctly predicted by the calculations.Unfortunately, the signals of the olefinic hydrogens attached to C-3 are overlapped with those for H-4 and H-8, while all C-6 methylenes appear in the same region of the spectrum.(c) 3-and 4-Methyl-3-cyclohexen-1-ol. Figure 3 shows the optimized geometries of the major conformers of regioisomeric O-acetylmandelates 10-M and 10-P, respectively derived from 3and 4-methyl-3-cyclohexen-1-ol (for all conformers, see the Supplementary Material).Once more, in all global minima the carbinolic proton and the carbonyl and acetate oxygens are synperiplanar. 11In this case, two synperiplanar conformations of similar energy were found for each compound, with H-C1-O-C8 torsion angles () of approximately 35º and -35º, in which the O-acetylmandelate groups occupy pseudo-equatorial positions.These conformations account for ca.65% of the population.In addition, the synperiplanar conformations with pseudo-axial O-acetylmandelate groups represent around 30% of the population for each system.Again, the phenyl is better located to shield other nuclei in the cyclohexene ring in conformations where the O-acetylmandelate group occupies a pseudo-axial position.For these regioisomeric systems both diastereoisomers have contributing structures of this type.Antiperiplanar conformations were located too, but their contribution to the population is relatively small (less than 10%).These results reinforce the idea that O-acetylmandelates can adopt more conformations than H-bond locked mandelates. 4ll conformations that contribute more than 5% to the Boltzmann distribution were considered to calculate the GIAO 1 H NMR chemical shifts.Key experimental and calculated chemical shifts shown in Table 3 indicate that the correlation is very good. 12However, only the signals of olefinic protons of the para isomer (H-3 of 10P) were of practical value since the ones corresponding to the meta analogue (H-4 of 10M) superimpose and all the other signals appear in a very narrow region of the spectrum centered around 2 ppm.These results indicate that this methodology cannot be used to determine the optical purity of 3-methyl-3-cyclohexen-1-ol, at least when forming part of a regioisomeric mixture with 4-methyl-3-cyclohexen-1-ol.

Conclusions
Synthesis, NMR and conformational analyses of the mandelate derivatives of isomeric mixtures of bicyclic and monocyclic secondary cyclohexenols were performed and GIAO 1 H NMR Boltzmann weighted average chemical shifts were computed.Synthesis of O-acetylmandelates was trivial for all systems and these compounds were found to be suitable to determine the enantiomeric ratio and the absolute configuration of monocylic cyclohexenols 2,5-diphenylcyclohex-3-en-1-ol (endo/exo 47:53) and 4-methyl-3-cyclohexen-1-ol (meta/para 34:66).The mandelates derived from bicyclo[2.2.2]oct-5-en-2-ol (endo/exo 73:27) were obtained by partial hydrolysis of the acetate group without epimerization and could be successfully used for the studied purpose.The proposed methodology only failed for 3-methyl-3-cyclohexen-1-ol since none of the signals were well resolved.These results, together with those derived from our previous investigations for norborn-5-en-2-ol, suggest that mandelic acids can be used as CDAs for bicyclic and monocyclic secondary cyclohexenols with different patterns of substitution.To reach reliable conclusions, it is highly recommended that experimental results are complemented with thorough conformational analyses followed by GIAO NMR calculations.

Experimental Section
General.All non-aqueous reactions were performed in oven dried glassware under positive argon pressure.All reagents and solvents were used directly as purchased or purified according to standard procedures.DCM was distilled from calcium hydride.Methanol was purchased from Cicarelli in a pro analysis grade and used directly.Analytical thin layer chromatography was carried out using commercial silica gel plates (Merck, Silica Gel 60 F254) and visualization was effected with short wavelength UV light (254 nm) and a p-anisaldehyde solution (2.
a  =

Table 3 .
Experimental and calculated 1 H NMR and values (in ppm) for selected nuclei of O-acetylmandelates 10 a a  = R -S.b Syn and anti are relative to the O-acetylmandelate group in C-1.