《有机化学》课程教学资源（文献资料）Chiral recognition of ethers by NMR spectroscopy.pdf_大学文库

Review Article Chiral Recognition of Ethers by NMR Spectroscopy HELMUT DUDDECK AND EDISON DI´AZ GO´ MEZ Leibniz University Hannover, Institute of Organic Chemistry, Schneiderberg 1B, D-30167 Hannover, Germany Dedicated to Professor Domenico Misiti on the occasion of his 75th birthday ABSTRACT Enantiomers of chiral ethers and acetals are notoriously difficult to differentiate because their reactivity is low and they are poor donors to any Lewis acid or metal ion. As an exception, epoxides are somewhat better donors. This review describes the properties of ethers, explains NMR methods for their chiral recognition and describes successful examples of ether differentiation. The majority of literature reports deals with chiral lanthanide shift reagents and dirhodium tetracarboxylate complexes, which were used as enantiopure auxiliaries to create diastereomeric adducts with dispersed 1 H and 13C NMR signals. The various methods are compared as to which is best suited for which purpose. Chirality 21:51–68, 2009. VC 2008 Wiley-Liss, Inc. KEY WORDS: 1 H and 13C NMR; enantiodifferentiation; chiral ethers; Pirkle’s reagents; lanthanide shift reagents; dirhodium complexes; cyclodextrins; NMR in nonisotropic phases INTRODUCTION Chirality is one of the most important concepts in nature and science.1–5 An overwhelming majority of molecules in biological systems is chiral, i.e., their chemical structure is not identical with that of their mirror images. Moreover, nature tends to produce most of them in enantiomerically pure form. Organic and pharmaceutical chemistry has made enormous progress during recent decades to find synthetic procedures for enantioselective (asymmetric) syntheses by which the desired enantiomer is the major product whereas the other one is suppressed as far as possible.6–8 This is possible only by interaction with a chiral reference to convert enantiomeric transition states into diastereomeric ones, the latter of which having different energy by principle. Hence, the reaction via the transition state of lower energy will proceed faster resulting in a higher proportion of the corresponding enantiomer in the product mixture. Chemical synthesis has always to be accompanied by effective analytical methods. In the case of the asymmetric syntheses, techniques are required for chiral recognition, i.e., methods which can differentiate between enantiomeric and, thereby, energetically equal forms of molecules.9–11 There are two major objectives: i. Enantiomeric differentiation by which the ratio of the enantiomers can be monitored in their mixture (enantiomeric excess, e.e.). ii. Determination of the absolute configuration (AC) of a given enantiomer isolated from a mixture or present in a mixture. One of the most important techniques for chiral recognition is chromatography at chiral stationary phases at the analytical or preparative scale.12–14 Capillary electrophoresis is playing an increasing role as well.15–17 Historically the oldest approach is monitoring the optical rotation, a method which has been developped later into a variety of chiroptical techniques, such as optical rotatory dispersion (ORD), electronic circular dichroism (ECD) or, very recently, vibrational circular dichroism (VCD).18,19 Many other spectroscopic methods have also been introduced in the past which, however, had to overcome the drawback that they are intrinsically achiral and can be applied only if a chiral auxiliary is added during the experiment. Mass spectrometry can contribute in this field20 but NMR spectroscopy is most prominent because the equipment is available around the world and, in general, an NMR experiment is easy, fast and—if the spectrometer time is no too large—relatively cheap.21–24 NMR auxiliaries may be chiral reagents forming covalent bonds with substrate molecules of interest to convert enantiomers into a pair of diastereomers with the same ratio (chiral derivatizing agent, CDA). In other types of experiments, the auxiliaries may be solvent-like additives (e.g., Pirkle’s chiral solvation agents, CSA),25,26 chiral metal complexes like lanthanide salts Ln31 tris-b-dionates (chiral lanthanide shift reagents, CLSR; Ln 5 Eu, Pr, Yb, etc.),27–29 dirhodium tetracarboxylate complexes,30 or other organometallic systems. Chiral hosting compounds such as cyclodextrins,31–33 calixar- *Correspondence to: H. Duddeck, Leibniz University Hannover, Institute of Organic Chemistry, Schneiderberg 1B, D-30167 Hannover, Germany. E-mail: duddeck@mbox.oci.uni-hannover.de Contract grant sponsor: Deutsche Forschungsgemeinschaft; Contract grant number: DU 98/30 Received for publication 28 February 2008; Accepted 2 May 2008 DOI: 10.1002/chir.20605 Published online 24 July 2008 in Wiley InterScience (www.interscience.wiley.com). CHIRALITY 21:51–68 (2009) VC 2008 Wiley-Liss, Inc

enes34 and others have been employed. Among those many techniques, which is the method of choice depends very much on the structure of the substrate molecules, particularly on the presence or absence of certain functional groups participating in the interaction with the chiral auxiliary. ETHERS—STRUCTURE AND OCCURRENCE The structure of ethers is characterized by the presence of a COC fragment in the molecule. Here, the symbol ‘‘C’’ stands for any organyl residue which might be saturated (sp3 ) or unsaturated (sp2 or sp). There are acyclic and cyclic ethers; related functionalities like acetals C(OC)2, orthoesters C(OC)3, and peroxides COOC are mentioned here as well, although only very few scattered reports on their enantiodifferentiation exist in this field. Some prototypes of the most common basic structures are collected in Scheme 1: dialkyl ethers (1), vinyl ethers (2), aryl ethers (3), oxiranes (4), tetrahydrofuranes (5), dihydrofuranes (6), and furanes (7). The syntheses and chemical reactions of all those compounds are well-documented in textbooks of organic chemistry. Ethers are quite inactive when exposed to nucleophiles and electrophiles; it requires rather drastic conditions to open or substitute a COC arrangement. However, oxiranes (epoxides, 4) are exceptions in that there are specific synthetic pathways (e.g., olefin oxidation) and a variety of facile ring-opening reactions (Scheme 2). The situation is different when a molecule contains more than one ether oxygen atom. Because of entropy reasons, multiple-contact binding may become so strong that kinetically stable adducts and complexes are formed. Typical examples are polyglycol ethers (8), crown ethers (9),35–38 cryptophanes (10),39 substituted cyclodextrins,31–33 ether-linked calixarenes,34 molecular tweezers,40 and related host molecules. Chiral derivatives have been produced in all those compound classes. In general, such molecules are not substrates (to be enantiodifferentiated) but rather chiral auxiliaries or parts thereof.41 Nevertheless, for that purpose they have to be produced and analyzed in enantiomerically pure form so that they are not beyond the scope of this review. Chiral recognition of ethers is of relevance not only for synthetic procedures. Many ethers occur in nature as secondary metabolites; they are or can easily be synthesized by etherification of naturally occurring alcohols, phenols, Scheme 1. Structural prototypes of ethers. Scheme 2. Some prototypes of common polyether structures; polyglycols (8) and crown ethers (9) and cryptophanes (10). Scheme 3. Some representative ether molecules as natural products or derived from natural products: menthofurane 11, 1,8-cineol (12), trimethyl[[(1R,2S,6R)-1-phenyl-7-oxabicyclo[4.1.0]hept-2-yl]oxy]silane (13),43 (1)-cis-lauthisan (14),44 and laddered polyether natural products of marine origin, e.g., CTX-3C (15).45–47 52 DUDDECK AND DI´AZ GO´ MEZ Chirality DOI 10.1002/chir

or polyols (Scheme 3). Moreover, ethers and acetals can act as protecting groups in sugars. For example, benzyl, triphenylmethyl (trityl), and trimethylsilyl ethers, as well as 1,3-dioxolanes or 1,3-dioxanes are popular because they are quite stable over a large pH range but can be easily removed by hydrogenation, acidification, and other mild reactions.42 ETHERS—LEWIS BASE PROPERTIES Oxygen Functionalities and Hard Lewis Acids According to the HSAB concept (hard and soft acid and base),48 oxygen is a hard Lewis base (electron donor) existing in a large number of organic molecules, such as alcohols, ethers, carbonyls (ketones and aldehydes), carboxylic acid and their derivatives, and many others. Consequently, oxygen is expected to form stable complexes with hard Lewis acids. This, however, is not always the case. For example, the relative stability of adducts produced by various neutral Lewis bases (L) with Lewis acids TiX4 (X 5 F, Cl and Br; TiX42 L) has been studied.49 It turned out that the stability of an ether ligand (L 5 Me2O) is about an order of magnitude lower than that of ketone (L 5 Me2C¼¼O). In contrast, adducts with phosphoryl oxygen [MeO3P¼¼O or Cl(MeO2)P¼¼O] are even more stable than Me2C¼¼O. Analogous results have been reported for transition metal complexes.48 In the case of lanthanide shift reagents (Eu31, Yb31, Pr31, etc.), adduct formation effects of oxygen donors on NMR chemical shifts can be categorized as follows:27–29 Strong donors: Water (H2O), alcohols (ROH), sulfonyl (S¼¼O), phosphoryl (P¼¼O). Medium donors: Carbonyls (C¼¼O), esters (RCOOR0 ), amides (RCONR0 2), epoxides ( ). Poor donors: Carboxylic acids (RCOOH), ethers (ROR0 ). Secondary effects due to steric hindrance may perturb this sequence. A number of different molecular properties and/or interactions might be responsible for this large divergence of oxygen functionalities. Properties like nucleophilicity, donor ability, and Lewis basicity of an oxygen functional group depend on changes in the hybridization state of the oxygen orbitals, in steric hindrance against electrophilic attack and in interactions of oxygen electron pairs with orbitals at adjacent atoms.50–52 Such influences will be examined in the following sections. Electrostatic charges. If the oxygen atom binds to a hard Lewic acid, electrostatic attraction is the major source for adduct formation.48,53 So, electrostatic charges at the oxygen atoms may play a significant role in the above sequence; and as shown in Table 1, this trend is fairly reproduced by DFT calculations for most compounds54 [Density functional calculations (B3LYP 6-31G*) using SPARTAN 06 version 1.1.0, Wavefunction, Irvine, CA]. There is a large difference in the electrostatic charges of singly and doubly bonded oxygen atoms. However, oxiranes (epoxides) are not in accordance with that sequence. Although they have proven to be better donors than ethers, this fact is not paralleled by a higher electrostatic charge (see also ‘‘Basicity, ring size, and steric hindrance’’ section). Ionization potential. The donor ability and the Lewis basicity of an oxygen-containing molecule is a consequence of the ionization potential; in a series of aliphatic ethers the ionization potential (in eV) decreases, and the basicity increases: Me2O, 9.94 eV; Et2O, 9.50 eV; (isoPr)2O, 9.32 eV and (tert-Bu)2O, 8.94 eV.50,55 Clearly, hyperconjugation from the alkyl groups influences the donor ability. On the other hand, ionization potentials of alcohols and aldehydes are even larger than those of the comparable ether,55 a sequence opposite to expectation for donor properties. This parameter should be taken into consideration only when a group of compounds with the same oxygen functional group is compared. Charge transfer toward hard Lewis acids. Neutral oxygen bases (O) forming adducts with hard Lewis acids (A) undergo a partial charge transfer from one free-electron pair to the hard Lewis acid (see Scheme 4). It is reasonable to assume that the stability of such an adduct depends on how the oxygen base is able to accommodate this partial positive charge. An alcohol R-OH can easily compensate this charge due to the oxygen-hydrogen bond; in the extreme, a proton is separated. A carbonyl group can delocalize a partial positive charge within the doubly bonded RR0 C¼¼O or R-COOR0 group. An ether group ROR0 , however, has no way to delocalize this partial charge significantly, apart from some hyperconjugaTABLE 1. Density functional (B3LYP6-31G*) calculated54 charges of some oxygen-containing compounds. 16 20.503 17 O1 :20.358 O2 :20.508 18 20.389 19 20.234 Scheme 4. Charge transfer adduct of an oxygen base(O) and a hard Lewis acid (A). CHIRAL RECOGNITION OF ETHERS BY NMR SPECTROSCOPY 53 Chirality DOI 10.1002/chir

tive assistance from the attached alkyl groups (see ‘‘Ionization potential’’ section). This behavior is paralleled by the relative nucleophilicity of those species and by their reactivity, whereas alcohols and carbonyl species are sensitive to a variety of reactants, ethers are rather inert, and catalysis with a strong hard Lewis acid, e.g., BF3, is required for any substitution or fragmentation reaction. Conjugation (vinylic, allylic, and aromatic ethers). Both, oxygen basicity and electrostatic charges are diminished when the oxygen atom is involved in p-bond interactions (Table 2). The aromatic furane (23) is a typical example for an extremely low charge but, in turn, p-system extention can support electrostatic charge accumulation at the oxygen site. Basicity, ring size, and steric hindrance. Threemembered oxirane rings (epoxides, e.g., 19) and, to a lesser extent, four-membered rings in oxetans (e.g., 26) are highly reactive educts or intermediates in a great variety of regio- or stereoselective reaction pathways, which are not viable by other ether compounds.56 This is due to their ring strain and the consequences on the binding state. Although aliphatic acyclic and cyclic ethers with rings larger than four atoms are poor donors, this seems to be different for oxiranes. Whether or how the enhanced donor ability of oxiranes is associated to their unique electronic states is still unclear. It has been determined from various physical and chemical experiments57 that the electron donor abilities and basicities of cyclic ethers decrease in that sequence, i.e., tetrahydrofurane is the best and oxirane is the worst (Scheme 5). Surprisingly, the basicity does not parallel the ability to bind to lanthanide ions in this sequence.27–29 After the failure of the basicity concept, it is plausible to assume that steric effects are responsible for the unique behavior of oxiranes where bond angles COC are much smaller and the distances of oxygen to neighboring CH2 hydrogen atoms are much larger than in any other cyclic ethers. Thus, the oxygen atom of an oxirane (19) is less hindered sterically and, hence, better to approach for Lewis acid atoms (Scheme 6). This argument is confirmed by the observation that 7-oxanorbornane derivatives are comparable to oxiranes in their donor properties; see below 49 and 50 (Scheme 15 and corresponding text). So, it appears that rather steric than electronic properties are responsible for the unique affinity of oxiranes towards lanthanide ions and other Lewis acids. In summary, the stability of hard Lewis acid–ether complexes based on electrostatic dipole–dipole attraction is strongly dependent on the distance r between the binding sites in that Lewis acid–base system (r 26 ). Any steric hindrance in any of the components will increase the distance and thereby weaken the adduct. In the case of ethers, the donor ability of oxygen is influenced by changes in the electrostatic charge at the oxygen atom, by differences in the delocalization of the partial positive charge generated by adduct formation to hard Lewis acids, and by steric effects. These contributions are hard to predict and their simultaneous operation may inverse sequences observed for the individual influences, as for TABLE 2. Density functional (B3LYP6 - 31G* ) calculated54 charges of some oxygen-containing compounds. 18 20.389 20 20.456 21 20.288 22 20.363 23 20.157 24 20.328 Scheme 5. Sequence of donor abilities and basicities of some cyclic ethers 18, 19, 25, and 26. Scheme 6. Spatial orientation of a-CH2 hydrogen atoms with respect to oxygen in some cyclic ethers. 54 DUDDECK AND DI´AZ GO´ MEZ Chirality DOI 10.1002/chir

example, electrostatic charges versus steric effects in oxiranes. Oxygen Functionalities and Soft Lewis Acids Much less literature reports exist for this combination of oxygen bases and soft Lewis acids as adduct partners; all oxygen functionalities are hard bases and, by principle, poor donors when soft Lewis acids are involved. Soft acids are, for example, many heavy transition metals like Rh, Ir, Pd, Pt, Ag, Au, Cd, Hg and, particularly, their low-charge cations. It has been described that two important terms exist in the binding of a Lewis acid–base pair: (i) electrostatic attraction and (ii) electron donation via orbital interaction; mostly HOMO-LUMO interaction.48,53 It is important to note that the latter term is effective only when both partners, acid and base, are soft. Therefore, oxygen functionalities bind to soft bases merely via eletrostatic attraction; the HOMO-LUMO gap is too large for a significant contribution from orbital interaction. Since the charge at the soft base site is generally low, the binding energy is low as well. An instructive experimental result confirms this interpretation: sulfoxides (e.g., Me2S¼¼O) bind to hard acids predominantly by the hard oxygen atoms but to soft acids preferably by the soft sulfur atom,48,58 and this in accordance with the HSAB principle. Nevertheless, oxygen functionalities are able to form complexes quite effectively, particularly carbonyl groups; ethers, alcohols, and water can generally not go beyond some solvent-like complexation. As an example, ether complexes of platinum and palladium are essentially unknown and believed to exist merely in situ. 48 This differential behavior was observed with dirhodium tetracarboxylate complexes59,60 acting as soft Lewis bases as well (see later). ETHERS—CHIRAL RECOGNITION BY NMR SPECTROSCOPY General Principles A fundamental theorem in stereochemistry says that chirality can be recognized only in the presence of a chiral reference. NMR spectroscopy, however, is an achiral method so that only constitutionally heterotopic or diastereotopic nuclei can be discriminated by their different chemical shifts (anisochrony). Enantiotopic nuclei, however, are isochronous and give rise to identical NMR signals. In other words, NMR spectroscopy cannot differentiate enantiotopic nulei or hemispheres. Nevertheless, NMR is most popular in chiral recognition because a variety of techniques has been invented over several decades in which chiral references are introduced. It is a common feature of all these techniques that enantiotopic nuclei are converted into diastereotopic and, thereby, anisochronous nuclei by reaction with or by addition of an enantiopure auxiliary.61 Basically, two different ways exist to reach this target.24 Chiral derivatization. A racemate of a chiral substrate (1)-S/(–)-S is allowed to react with an enantiopure derivatizing auxilliary, either (1)-CDA or (–)-CDA, to produce two diastereomeric derivatives, each in enantiopure form. For example with (1)-CDA eq. 1: ðþÞ-S þ ðÞ-S þ 2ðþÞ-CDA ! ½ðþÞ-S ðþÞ-CDA þ ½ðÞ-S ðþÞ-CDA ð1Þ In the ideal case, these diastereomers exist in the same ratio as the enantiomers before. Principally, every signal in the CDA-derivative is shifted to higher or lower frequencies as compared with those in the isolated components S and CSA, respectively. This is called complexation-induced shift Dd (in ppm). Moreover, every signal appears twice because of different Dd-values of the diastereomeric S–CSA derivatives, although it occurs quite frequently that such differences are undetectably small (incidental isochrony). This difference in the complexation-induced shifts is called diastereomeric resolution or dispersion, and the parameter is DDd (in ppm) or Dm (in Hz). The DDd parameter is independent of the spectrometer frequency. In cases where resolutions/dispersions are low, it may be more convenient to use Dm (in Hz) but one has to keep in mind that the Dm parameter increases (or decreases) proportionally with the external field strength. These parameters must be recalculated if values recorded at different field B0 are to be compared. If the mole fractions of the enantiomers in a nonracemic mixture are significantly different, e.g., 2:1, two sets of dispersed NMR signals exist for the CDA-derivatives, each of which can be assigned to the respective diastereomer by its relative signal intensity. Then, the parameters, DDd or Dm, are no longer absolute values but plus- (1) or minussigns (2) can be attributed to each dispersion according to eq. 2; note that there is no obligatory rule which diastereomer is first and which is second in this differential expressions. DDd ¼ d½ðþÞ-S ðþÞ-CDA d½ðÞ-SðþÞ-CDA ðin ppmÞ ð2aÞ Dm ¼ m½ðþÞ-S ðþÞ-CDA m½ðÞ-SðþÞ-CDA ðin HzÞ ð2bÞ Then, NMR signal integration of the dispersed signals reflects the enantiomeric ratio (1)-S and (2)-S from which enantiomeric excess data (e.e.) and the optical purity can be derived. However, it is indispensable in such experiments to make sure that there was no kinetic resolution in this reaction and both enantiomers react with exactly the same yield, optimally 100%. Of course, any purification by recrystallization or chromatography should be avoided strictly. On the other hand, chiral derivatization is the method of choice when the absolute configuration (AC) of a pure enantiomer, e.g., a natural product isolated, is to be determined.62 Then, the reaction with an enantiopure CDA of known AC affords one diastereomer only. If single crystals CHIRAL RECOGNITION OF ETHERS BY NMR SPECTROSCOPY 55 Chirality DOI 10.1002/chir

of this compound can be produced, X-ray diffraction will easily lead to the correct AC of the original substrate. In many cases, however, such crystallization is not successful so that other well-established methods have to be applied. Among them is NMR spectroscopy which relies on selective anisotropy effects of aromatic moieties inside the S– CDA derivative provided a preferred conformation exists. There is a number of such CDA,22,24,62 for determining AC of alcohols and amines, some of the most prominent and effective ones are Mosher’s methoxytrifluorophenylacetic acid (MTPA; 27) 63–65 and Harada’s 2-methoxy-2-(1-naphthyl)propionic acid (MaNP; 28) 66,67 (Scheme 7). To the best of our knowledge, however, there is no report so far where the AC of an ether has been determined by using a CDA. Chiral salvation. The use of chiral solvating agents (CSA) for AC determination is not as straightforward as using CDA,24,29 because CSA-aggregates (adducts, complexes) often possess a low lifetime and a high conformational flexibility; effects generated in individual conformations may be obscured in the equilibrium. On the other hand, CSA are optimal for chiral resolution, i.e., the determination of the ratio of enantiomers in a mixture via forming diastereomeric adducts.21–23,29 ðþÞ-S þ ðþÞ-CSA ½ðþÞ-S ! ðþÞ-CSA ðKþÞ ð3aÞ ðÞ-S þ ðþÞ-CSA ½ðÞ-S ! ðþÞ-CSA ðKÞ ð3bÞ In most cases, the substrate S and the CSA form kinetically weak adducts (charge transfer interaction, hydrogenbonding, etc.) and exist in fast-exchange equilibria formed by the free components (left side of the equilibrium in eq. 3) and the adduct (right side of the equilibrium in eq. 3); the equilibrium constants for the two diastereomeric adducts are K1 and K2, respectively. Exchange rates are generally high on the NMR time-scale. Therefore, substrate NMR signals are observed as averages of those of the free S and the S component in the adduct. This holds for each diastereomeric adduct. In terms of dynamic NMR, the equilibria are on the fast-exchange or high-temperature side of the coalescence. Generally, addition of a CSA to a substrate in solution gives rise to changes of the NMR signals of all components in the adduct; the nuclei may be shielded or deshielded. This shift is called complexation induced shift Dd (see previous section), which is defined as follows: Dd ¼ dðadductÞ d0ðfree componentÞ; in ppm ð4Þ The Dd-values depend not only on the structural features of the adduct system and the interaction of the molecular components itself but also on the equilibrium constants K1 and K2. They can reach their maximal values, Ddmax, only if the equilibria are predominantly on the adduct side which, however, is not always the case. If needed at all, the Ddmax-parameter and thereby K, can be determined by low-temperature NMR under the condition that the low-exchange regime can be reached, far enough below the coalescence. Fortunately, the success of an enantiodifferentiation experiment is not very sensitive to the K1/K2 ratio. There is no danger of ‘‘thermodynamic induction’’ if both constants are significantly different. On the contrary, when they are different, both equilibria are different, as well, and the same holds of the two complexation induced shifts. In other words, a larger dispersion can be expected in such cases. In contrast to CDA-derivatives, resolution/dispersion effects Dm (or DDd) appear only in that component of a rapidly exchanging adduct, which exists in two enantiomeric forms; generally this is the substrate. Each substrate enantiomer ‘‘sees’’ always the same enantiomer of the CSA, i.e., a CSA-molecule exchange never alters that adduct diastereomer. On the other hand, every CSA-molecule ‘‘sees’’ an average of both (1)-S and (2)-S, which is only dependent on the equilibrium constants K1 and K2 and the molar ratio of (1)-S and (2)-S. Therefore, such NMR signals do not split into two. Nevertheless, CSA signals contain chirality information about the enantiopurity of the substrate, but in another way, the CDA signal (Dd) lies between Dd(1) and Dd(2) , the complexation induced shifts which would result from CSA experiments with enantiopure (1)- S and (2)-S, respectively. If the mixture of enantiomers is ½ðþÞ-S=½ðÞ-S ¼ x=y ðx;y ¼ mole fractionsÞ ð5Þ Dd represents the averaged position between Dd(1) and Dd(2) : Dd ¼ ½x DdðþÞ þ y DdðÞ=2 ð6Þ provided that K1 5 K2. Therefore, if Dd(1) and Dd(2) are known, Dd may reveal the e.e. of the (1)-S/(2)-S mixture. This, however, is not a practical experiment because the two pure enantiomers are generally not available. Anyway, if enantiopure samples are available, analogous optical rotation measurements are easier to perform and more precise. As a conclusion, straightforward determination of enantiomeric ratios by NMR is possible by integrating dispersed NMR signals of the substrate component only. Scheme 7. Structures of Mosher’s acid (MTPA; 27) and Harada’s 2- methoxy-2-(1-naphthyl)propionic acid (MaNP; 28). 56 DUDDECK AND DI´AZ GO´ MEZ Chirality DOI 10.1002/chir

Occasionally, adducts of substrates and a CSA are kinetically stable. Then, the exchange is slow on the NMR time-scale and NMR spectra map the signals of all coexisting species (free components and adducts). In such cases, the CSA acts rather as a CDA than a classical CSA. Alcohols and amines as chiral solvating agents. Parallel to the introduction of CDAs by Mosher’s and other groups, Pirkle developed an alternative kind of NMR auxiliaries, chiral alcohols and amines as solvating agents; see for example, 2,2,2-trifluoro-1-phenylethanol (TFPE; 29) and 1-phenylethylamine (PEA; 30) in Scheme 8; other CSAs have naphthyl or anthryl groups instead of phenyl.25,26 Further chiral CSAs have been reported later.24 In all cases, the differentiation of enantiomers is not produced by conversion into diastereomers but by the creation of relatively weak solvation complexes, mostly via hydrogen bond formation and aromatic ring attraction. It is a crucial point in this method that a single hydrogen-bond alone is generally not sufficient for an effective chiral discrimination. Secondary binding by further hydrogen-bond formation or by charge-transfer between aromatic moieties in both partners (‘‘dibasic solute model’’ and ‘‘three-point interaction’’)24–26 are most desirable. An early example is shown in Scheme 9 with (R)-(1-(1-naphthyl)ethylamine (31) as CSA and both enantiomers of TFPE (29) as solute.68 Theory and application of the CSA method have been described in a variety of reviews,21–26 so a detailed discussion is not required here. Generally, Pirkle’s CSAs and those of other groups have been designed to bind hard Lewis acids and bases which are good donors, and ethers do not fulfill this condition (see earlier). So, only few reports of CSA experiments have documented that chiral ethers can be differentiated. So, Pirkle’s powerful enantiopure 1-(9-anthryl)-2,2,2-tri- fluoroethanol (32) is capable of differentiating the racemic glycol ether derivative, Ph(Me2Si)CH2-O-CH2CH(Ph)- OCH3 (33) (see Scheme 10)69 (Gassmann S, Bienz S, pers comm, 2007); note that a large surplus of CSA was required: 40 mg of 32 to react with 5 mg 33. As an exception, oxiranes (epoxides) have been studied intensively, probably because of their comparatively favorable donor ability (see earlier).21,70,71 Another interesting example is the application of Pirkle’s alcohol 29 to discriminate the enantiomers of the mercurated allyl ether 35 prepared from a chiral allene 34; allenes are unable to react with CSAs.72 This reaction and the dibasic solute model are depicted in Scheme 11. Chiral lanthanide shift reagents. The method of using lanthanide shifts reagents for chiral discrimination has been developed in the early 1970s and ruled out the CSA Scheme 8. Pirkle alcohol 29 and amine 30. Scheme 9. Solute-CSA adduct with secondary binding. Scheme 10. Pirkle’s alcohol 32 (top left), the glycol ether derivative 33 (top central) and a tentative three-point interaction model (top right); section of the 1 H NMR spectrum of 5 mg 33 in 0.6 ml CDCl3 (bottom left); after addition of 40 mg 32 (bottom right) (reproduced with permission from S. Bienz, Organometallics 2001;20:1849–1859). Scheme 11. Pirkle’s alcohol 29 and a mercurated allyl ether (35) prepared from a chiral allene (34); bottom: solvation model. CHIRAL RECOGNITION OF ETHERS BY NMR SPECTROSCOPY 57 Chirality DOI 10.1002/chir

method soon; a number of reviews appeared in the following years.21–24,27–29 Principal features are similar to those of the CSA complexes (see earlier). The most popular reagents are tris(3-trifluoroacetylcamphorato)europium [Eu(tfc)3; 36] and tris(3-heptafluoropropylhydroxymethylene)europium [Eu(hfm)3; 37] but a few others, also with ligands based on the camphor skeleton, have been tested (Scheme 12). In analogy to CSAs, ethers bind weakly with CLSR73 and shift reagents with fluorinated ligands should be used due to their higher Lewis acidity.74,75 Positive results have been reported for tetrahydrofurane (18) 75,76 whereas furane (23) failed, probably because of the electron pair delocalization in the latter (see earlier).76 On the other hand, molecules with several ether oxygen atoms close to each other form chelates producing strong NMR signal resolution. So, constitutions and configurations of such compounds have been studied. For example, ortho-dimethoxybenzene (39) binds well whereas methoxybenzene (38) as well as meta- and para-dimethoxybenzenes (40 and 41, respectively) and methoxynaphthalenes (42 and 43) are poor donors (Scheme 13).77,78 Analogous observations were reported for some glycol ether derivatives79 and and 3,4,5-trimethoxyphenyl-substituted lignans.80 The advantage of chelation has been utilized by Diaz et al.81 for differentiating successfully the enantiomers of 2,20 ,6-trimethoxy-60 -methoxymethyl-biphenyl (44) in the presence of Eu(tfc)3, Pr(tfc)3, and Pr(hfm)3 with an effi- ciency not far away from that of other 2,20 -dimethoxybiphenyl derivatives bearing one or two hydroxyl or carbonyl functions. It should be noted, however, that somewhat higher molar proportions of the CLSR had to be added (Scheme 14). Like in the case of CSA application, oxiranes (epoxides) are generally better donors to CLSR than open-chain ethers or cyclic ethers with more than four ring members, apparently because of the lower steric shielding by ahydrogens (see earlier). Among them are phenyloxirane (45) 82 and 1,2-epoxy-3-methylpentane (46),83 the d- and lenantiomers were discerned from the meso-form of 1,2- dimethyloxirane (47).84 An intriguing series of benzonorbornane derivatives has been described: benzonorbornadiene-exo-oxide (48), 7-oxa-benzonorbornene (49), and 7- oxa-benzonorbornadiene (50).85 Interestingly, the donor ability of the 7-oxa-norbornanes 49 and 50 are similar, if not even better, than that of the structurally related epoxide 48. This is not a surprise when considering the positions of the two next-nearest hydrogens, because of the unique geometry of the norbornane skeleton, all hydrogens are directed away from the bridging oxygen atom so that its steric shielding is less than in any other open-chain or less-strained cyclic ethers; compare Scheme 6 and the pertinent text (Scheme 15). Finally, a solitary report on chiral discrimination of acetals by NMR spectroscopy should be mentioned: synthetic (1)-, (2)- and rac-a-multistriatin (51) were differentiated: a slight C-1 NMR signal splitting could be observed in the presence of the strongly acidic Eu(hfc)3 (37).86 Dirhodium tetracarboxylate complexes. The disadvantage of most chiral axiliaries described earlier is the fact that they are hard Lewis acids (or bases) and, therefore, not effective in binding to soft subtrates. On this background, a dinuclear complex offers an alternative because it is a soft Lewis acid itself.59,60 So, the dirhodium tetracarboxylate complex 52, Rh(II) 2[(R)-(1)-MTPA]4 (MTPA-H 5 methoxytrifluoromethylphenylacetic acid : Mosher’s acid; Scheme 16),87 has been introduced. Indeed, 52 offers excellent chiral resolution of soft Lewis base substrates like sulfides, sulfoxides (complexation at the sulfur atom), selenides, phosphanes, phosphane sulfides, and selenides30 as well as olefins,87 iodides, nitriles, phosphane borane complexes,88 and silanes89 (hydridic hydroScheme 12. Structures of the chiral lanthanide shift reagents Eu(tfc)3 (36) and Eu(hfm)3 (37). Scheme 13. Some methoxylated benzenes and naphthalenes 38–43. Scheme 14. Structure of 2,20 ,6-trimethoxy-60 -methoxymethyl-biphenyl 44. 58 DUDDECK AND DI´AZ GO´ MEZ Chirality DOI 10.1002/chir

gen as binding sites). In that sense, the dirhodium method, i.e., application of 52 for discrimination of chiral substrates, is complementary to the use of CSA and CLSR. In contrast to CLSR, 52 has two rhodium sites for complexation, and the balance of the equilibriation depends on the molar ratio of the components 52 and the ligating substrate molecule S (Scheme 17). With only one exception, namely phosphanes,30 the adducts are kinetically unstable and the ligand exchange rates are fast on the NMR time-scale. Thus, all NMR signals of S are averaged. This situation is analogous to that for CSA and CLSR. In the case of soft ligands, S 5 S, Se, and Te, the equilibria are biassed; free ligand molecules do not exist as long as free rhodium sites are available (large complexation energy). In addition, the energy difference between 1:1- and 2:1-adducts is rather large (in favor of the 1:1-adduct) so that primarily 1:1-adducts are produced until more than one equimolar amount of S is added. On the other hand, hard ligands like oxygen functionalities bind much weaker; the complexation energy is low and so is the energy difference between 1:1- and 2:1- adducts. Moreover, the equilibria are no longer biassed. We found that NMR experiments in the presence of equimolar amount of 52 and hard ligands S provide a predominance of 1:1- over 2:1-adducts (see later). On the basis of thermodynamic and kinetic results, it has been shown that 52 can give good enantiodifferentiation results not only with soft Lewis bases but also with hard oxygen functionalities (category IV ligands).30 In addition, a remarkable selectivity of the dirhodium reagent has been noticed when binding to substrates with more than one heterotopic ligand sites, such as various carbonyls, aromatic iso- and heterocycles, and so forth.90–92 In experiments allowing intramolecular competition, e.g., with the racemic 2,3-trans-dihydro-(4-hydroxy-3-methoxyphenyl)-3-(hydroxymethyl)-1,4-benzodioxin-6-yl]-2(E)-propenoic acid ethyl ester (53; Scheme 18),93,94 we found that a carbonyl oxygen seems always to be the better donor as compared with ether-like oxygen functionalities, as expected.95 Clearly, there is no complexation at the 1,4- dioxane ring moiety or the methoxy group but at the ester carbonyl (Scheme 18, top); whereas signal resolution (dispersion) is found all over the molecule. The definition of a category IV ligand30 implies that the binding energy between such molecule and the dirhodium complex 52 is not far from zero. Then, the complexation constant is in the order of magnetude 100 ; i.e., in an equimolar solution of the ligand S and 52, always a considerable amount of the free components exists beside the 1:1- adduct S?52 (and, to a lesser extent, the 2:1-adduct S?52/S0 ). A low complexation energy DG0 is generally accompanied by a low energy barrier DDG# (Scheme 19) so that it is very difficult to estimate by NMR methods how the equilibrium is balanced, more at the components’ or more at the adduct’s side. We were able to circumvent this dilemma by studying highly symmetrical chiral ether subScheme 15. Structures of the oxiranes 45–47, benzonorbornanes 48–50 and the acetal 51, studied in the presence of CLSR. Scheme 16. Structure of Rh(II)2[(R)-(1)-MTPA]4 (52); MTPA-H 5 methoxytrifluoromethylphenylacetic acid (Mosher’s acid). Scheme 17. Equilibria of 52 (left) with ligand molecules (S) forming 1:1-adducts (center) and 2:1-adducts (right). CHIRAL RECOGNITION OF ETHERS BY NMR SPECTROSCOPY 59 Chirality DOI 10.1002/chir

strates with unusually high barriers, which are due to molecular entropy. Racemic 2,8,12-trioxahexacyclo[8.3.0.03,9.04,6.05,13.07,11]- tridecane (54) 96 and 4,7,11-trioxapentacyclo [6.3.0.02,6. 03,10.05,9]undecane (55) 96 were investigated in the presence of an equimolar amount of 52. This experiment was successful in terms of enantiodifferentiation (Scheme 20).97 In addition, it became apparent that some 1 H and 13C NMR signals were somewhat broadened even at room temperature; see Scheme 20 (bottom right). Therefore, variable-temperature experiments were performed and, surprisingly, the spectra displayed a nearly complete coalescence behavior at moderately lowered temperatures (Scheme 21).98 A comparison with the signals of the free ligand (top traces in Scheme 21) shows that one of the two signal sets of 54, visible at low temperatures, belongs to the free ligand itself, whereas the other one represents the 1:1- adduct 54?52; both exist in approximately equal amounts indicating that DG0 is close to 0 kJ/mol, and the equilibrium constant is K 100 . This is indeed typical for category IV ligands. The energy barrier is DDG# 5 44 6 2 kJ/mol, estimated for the coalescence temperature Tc. The ligand 54, as well as 55 for which an analogous NMR behavior was recognized is a trivalent ether. Two ligand exchange mechanisms are conceivable: (a) an adduct opens its RhO bond, the ligand molecule rotates and forms a new RhO bond with another oxygen of the same ligand molecule (internal rotation), and (b) an adduct opens and another free ligand molecule forms a new adduct (replacement). If mechanism (a) is responsible for the coalescence behavior, each 1 H and 13C signal of the bound 54 should be split into three with relative intensities of 1:1:1 because 54 would loose its C3 symmetry. This, however, is not observed; the signal patterns of the free and the bound 54 are very similar (Scheme 21) as if the bound molecule still would possess C3 symmetry. Thus, the underlying mechanism is replacement (b) whereas internal rotation (a) is a much faster process giving time-averaged signals, which cannot be frozen even at 210 K. (Scheme 22). Interestingly, NMR signals of monoethers like 56–58 do not show an analogous VT-NMR behavior although their NMR signals display similar coordination (Dd) and dispersion (Dm) as well. Apparently, the ligand exchange barrier DDG# (5 DDH# 2 TDDS# ) is much lower for these ethers than for the trivalent ethers 54 and 55. Thus, the question arises: Why is that barrier so exceptionally high for the symmetrical ethers? As mentioned earlier, it is reasonable to assume that activation enthalpies DDH# are generally low for all types of ethers. Thus, it must be the activation entropy DDS# which is responsible. Basically, two factors influence DDS# , namely loss or restriction of symmetry and of mobility during transition state formation. Since adduct formation is the combination of two molecules, DDS# is expected to be negative, but why can its magnitude be so different? A comparison of estimated DDS# contributions, compiled in Table 3, reveals that indeed a much higher energy barrier is expected for a chiral triether than for a monoether; the caged triether 54 and the asymmetric acyclic ether 56 are taken as representatives for the following discussion. Since a quantitative estimate is very difficult and unsafe, we restrict our discussion to general trends. i. Mixing entropy: both 54 and 56 are racemates. So, activation entropy contributions due to the existence of isomers are very similar in both cases and can, therefore, be neglected in the estimation. ii. Symmetry number: the triether 54 loses its C3-symmetry (three indistiguishable orientations in space) in the transition state because it is fixed to a rhodium atom by one and only one oxygen atom (C3 ? C1) creating a large DDS# contribution. In contrast, the monoether 56 Scheme 18. Racemic 2,3-trans-dihydro-(4-hydroxy-3-methoxyphenyl)- 3-(hydroxymethyl)-1,4-benzodioxin-6-yl]-2(E)-propenoic acid ethyl ester (53) investigated by the dirhodium method; top: significant complexation induced shifts Dd (in ppm), bottom: diastereomeric dispersions Dm (in Hz, recorded at 9.4 Tesla).93,94 Values for 13C are in italics; favoured complexation sites are indicated by ‘‘Rh...’’ (symbolizing 52). Scheme 19. Equilibria of the free component molecules S and 52 in the ligation (top) and ligand exchange reaction (below), energy level diagram (bottom); ‘‘TS’’: transition state. 60 DUDDECK AND DI´AZ GO´ MEZ Chirality DOI 10.1002/chir