Rational design of dynamic ammonium salt catalysts towards more flexible and selective function.

This review focuses on the development of dynamic ammonium salt catalysis for selective organic transformations conducted in our laboratory since 2002. Several important concepts in designing of catalysts are described with some examples. In particular, the practical synthesis of chiral 1,1'-binaphthyl-2,2'-disulfonic acid (BINSA) and its application in chiral ammonium salt catalysis for the enantioselective direct Mannich-type reaction are described.

Introduction

The acidity of catalysts can be controlled based on acid–base combination chemistry.[1),2)] For example, if p-toluenesulfonic acid is too strong as an acid catalyst, pyridinium p-toluenesulfnonate may be chosen. p-Toluenesulfonic acid is a classical single-molecule catalyst, and its ammonium salt behaves as a dynamic complex in solution. Ammonium sulfonates are equilibrated with free amines and free sulfonic acids in solution. Enzymes also function through dynamic complexation between acidic molecules and basic molecules in vivo. It might be possible to create an artificial catalyst taking advantage of the conformational and dynamic flexibility of such salts expecting enzymatic functions such as induced-fit, molecular recognition, cooperative effect, allosteric effect, etc. The key to designing dynamic complexes is non-bonding interactions such as hydrogen-bonding, electrostatic, ionic, π-π, cation-π, cation-n, hydrophobic, hydrophilic interactions, etc. Recently, we developed a new type of catalyst which function through dynamic aggregation of acids and bases. In this article, the potential of dynamic salt complexes as highly functional catalysts is demonstrated with several examples that were developed in our laboratory.

Scheme 1 shows the equilibrium in a solution of an equimolar mixture of a sulfonic acid and a tertiary amine. In this case, a 1:1 complex is expected to be the major species.

Scheme 1.

Dynamic ammonium salts of sulfonic acid with tertiary amine.

Scheme 2 shows the equilibrium in a solution of an equimolar mixture of a sulfonic acid and a secondary amine. In this case, linear or cyclic n:n salt complexes may be formed because secondary ammonium cations have two acidic protons. The selectivity of complexation can be controlled by steric repulsions or other non-bonding interactions.

Scheme 2.

Dynamic ammonium salts of sulfonic acid with secondary amine.

Scheme 3 shows the equilibrium in a solution of an equimolar mixture of sulfonic acid and primary amine. In this case, more complicated n:n salt complexes may have to be considered because primary ammonium cations have three protons. Therefore, it is not easy to control the selectivity of complexation. Nevertheless, a primary ammonium cation may be useful as a building block to construct a three-dimensional structure.

Scheme 3.

Dynamic ammonium salts of sulfonic acid with primary amine.

Thus, dynamic salt complex catalysts can be prepared in situ from acidic and basic small molecules which are elaborately designed based on acid–base combination chemistry as shown in Schemes 1–3.

The first two examples are the dehydrative condensation reactions catalyzed by dynamic salt complexes of sulfonic acids with secondary amines. These salt catalysts are aggregated like a reverse micelle in non-polar hydrocarbons, and their catalytic activities are much higher than those of the corresponding sulfonic acids themselves due to local hydrophobic function around the ammonium protons of their salts.

The third example is the enantioselective Mannich-type reaction catalyzed by dynamic salt complexes of chiral disulfonic acids with tertiary amines. Interestingly, although ammonium cations are achiral, high enantioselectivity is induced through the chiral counter anion. Chiral dynamic salt complexes can be rapidly optimized to induce high enantioselectivity by a combinatorial approach.

The last three examples are the enantioselective Diels–Alder reactions of dienes with α-(acyloxy)- and α-(N,N-diacylamino)acroleins and the enantioselective [2+2] cycloaddition reactions of alkenes with α-(acyloxy)acroleins catalyzed by dynamic salt complexes of achiral Brønsted acids with chiral amines. A conformationally flexible chiral triamine 24 derived from H-L-Phe-L-Leu-N(CH2CH2)2 is very effective as a amine component of catalysts.

Bulky diarylammonium pentafluorobenzenesulfonates as mild and extremely active dehydrative esterification catalysts

A great deal of research has been focused on more environmentally benign alternatives to esterification processes, which are in great demand by the chemical industry.[3)] In general, the dehydrative condensation reaction of carboxylic acids with alcohols is catalyzed by Brønsted acids such as HCl, H2SO4, p-TsOH, etc. for acid-resistant substrates. For acid-sensitive substrates, weak Brønsted acids such as pyridinium p-toluenesulfonate (PPTS) are usually used. However, their catalytic activities are somewhat low, and the reactants that can be used are rather limited. In 2000, Tanabe et al. reported that N,N-diphenylammonium trifiate (1) (1–10 mol%) efficiently catalyzed the esterification reaction of carboxylic acid with equimolar amounts of alcohols under heating at 80 °C without the removal of water (Scheme 4).[4)] In 2006, Tanabe et al. reported that pentafluoroanilinium trifiate (2) was more active than 1.[5)] However, it is difficult to apply these methods to sterically demanding and acid-sensitive alcohols because 1 and 2 are still strongly acidic salts of a superacid and weak bases. In addition, its turnover is much lower than those of hafnium(IV) and zirconium(IV) catalysts.[6)–11)] In 2005, we found bulky N,N-diarylammonium pentafluorobenzenesulfonates such as 4a, 5a and 6a, which are much milder acids than the corresponding ammonium trifiates, as extremely active esterification catalysts.[12)–15)] The hydrophobic effect of the bulky ammonium sulfonates effectively promotes the dehydrative condensation reaction, and their steric bulkiness suppresses the dehydrative elimination of secondary alcohols to produce alkenes.

Scheme 4.

Anilinium arenesulfonates.

The catalytic activities of various arenesulfonic acids, dimesitylammonium arenesulfonates, and diarylammonium pentafluorobenzenesulfonates for the esterification reaction of 4-phenylbutyric acid with cyclododecanol in heptane are shown in Figs. 1–3.

Fig. 1

Catalytic activity of arenesulfonic acids.

Fig. 3

Catalytic activity of N,N-diarylammonium pentafluorobenzenesulfonates for the esterification in Fig. 1.

The catalytic activities of alkanesulfonic acids were almost independent of their counter anions (Fig. 1). For example, TfOH (pKa (CD3CO2D) = −0.74, Ho =−14.00), TsOH (pKa (CD3CO2D) = 8.5) and pentafluorobenzenesulfonic acid (pKa (CD3CO2D) = 11.1, Ho =−3.98) exhibited similar catalytic activities. These experimental results suggest that alkanesulfonic acid may function as a specific acid catalyst.

In contrast, the catalytic activities of N,N-diarylammonium alkanesulfonates depended on the structures of ammonium cations and alkanesulfonates (Figs. 2 and 3). These experimental results suggest that ammonium sulfonates function as general acid catalysts. In the case of N,N-dimesitylammonium arenesulfonates (Fig. 2), ammonium tosylate (4b, squares), pentafluorobenzenesulfonate (4a, circles), and mesitylenesulfonate (×) were more active as dehydration catalyst but less acidic than ammonium trifiate (+) and 2,4,6-trichlorobenzenesulfonates (triangle).

Fig. 2

Catalytic activity of N,N-dimesitylammonium alkane-sulfonates for the esterification in Fig. 1.

In the case of N,N-diarylammonium pentafluorobenzenesulfonates (Fig. 3), N,N-diphenylammonium pentafluorobenzenesulfonate (3a, squares) showed lower catalytic activity than C6F5SO3H (circles) due to its weaker acidity. More bulky N,N-diarylammonium pentafluorobenzenesulfonates showed higher catalytic activities than 3a. Surprisingly, N-(2,6-diphenylphenyl)-N-mesitylammonium pentafluorobenzenesulfonate (5a, triangles) exhibited higher catalytic activity than C6F5SO3H. These experimental results suggest that the hydrophobic effect due to bulky N-aryl groups and the S-pentafluorophenyl group of 5a, which surround NH2+ of the catalyst, may synergistically accelerate the dehydrative condensation reaction, and the hydrophobic effect may be more important than the strong acidity of NH2+ in promoting the dehydrative reaction.

Comparative experiments using catalysts 1 and 6a have been performed using the esterification reaction of 4-phenylbutyric acid with 6-undecanol in hexane under reflux conditions without the removal of water and under azeotropic reflux conditions with the removal of water (Fig. 4). While the reaction catalyzed by 1 was slightly decelerated under reflux conditions without the removal of water, the reaction catalyzed by 6a proceeded efficiently without the influence of water.[12)–15)]

Fig. 4

Esterification reaction of 4-phenylbutyric acid with 6-undecanol. The catalytic activities of 1 and 6a under reflux conditions without the removal of water (solid lines) and under azeotropic reflux conditions (broken lines) were compared. The proportions of 6-undecanol (circles), 6-undecyl 4-phenylbutyrate (squares), and 5-undecene (triangles) in the reaction mixture over time were evaluated by 1H NMR analysis.

Representative results for the esterification catalyzed by 4a (1 mol%) at 80 °C in heptane are shown in Scheme 5. 2-Unsubstituted carboxylic acids, 2-monosubstituted carboxylic acids, and sterically demanding 2,2-disubstituted carboxylic acids were smoothly condensed to produce the corresponding esters. α,β-Unsaturated carboxylic acids and benzoic acids were also transformed into the corresponding esters. 2-Alkoxycarboxylic acids were very reactive probably due to favorable chelation between the substrates and 4a. 4-Oxopentanoic acid was selectively esterified without protecting the ketone moiety. 4a is useful for acid–sensitive alcohols such as benzyl alcohol, allylic alcohols, propargylic alcohols, and secondary alcohols. In particular, esterification with the sterically demanding alcohol 6-undecanol gives the desired esters in good yield with less than 5% of alkenes. Although Lewis-acidic metal salts such as Hf(IV) and Zr(IV) were not adapted to 1,2-diols due to tight chelation with metal ions,[6)–11)] these diols were esterified in high yield by 4a. Less-reactive aryl alcohols and 1-adamantanol were also esterified in high yields.

Scheme 5.

Generality and scope of the dehydrative esterification reaction catalyzed by 4a.

Dehydrative condensation reactions of carboxylic acids with 1.1 equivalents of more-reactive primary alcohols proceeded even at room temperature without any solvents in good yield in the presence of 1 mol% of 4a (Scheme 6). Water was separated as a second phase during the reaction. This is an ultimate atom-economical esterification process.[15)]

Scheme 6.

Dehydrative esterification without solvents at room temperature.

Dehydrative cyclocondensation catalysts

Bulky N,N-diarylammonium pentafluorobenzenesulfonates promote the dehydrative cyclization of 1,3,5-triketones to γ-pyrones much more effectively than the dehydrative esterification reaction, since 1,3,5-triketones are generally less polar than carboxylic acids and alcohols.[16)] The local hydrophobic environment created around the ammonium protons in ammonium sulfonates is the key to the unusual acceleration of dehydration reactions. We have investigated the relationship between the catalytic activity and the steric and/or stereoelectronic factors of N,N-diarylammonium arenesulfonate catalysts for the dehydrative cyclization of 1,3,5-triketones, and have discussed the microscopic hydrophobic environment created in aggregated ammonium sulfonates based on a consideration of their X-ray single-crystal structures.

The catalytic activities of 6a and C6F5SO3H in the dehydrative cyclization of 4,6-dimethylnonan-3,5,7-trione (7a) have been compared under heating without the removal of water and under azeotropic reflux conditions with the removal of water (Fig. 5). While the reaction catalyzed by C6F5SO3H at 80°C without the removal of water gave γ-pyrone 8a in 74% yield after 8 h (circles, graph A), the reaction under azeotropic reflux conditions in cyclohexane (bp. 80.7 °C) with the removal of water gave 8a in 96% yield after the same reaction time (squares, graph A). The reaction proceeded more smoothly in perfluoromethylcyclohexane, which was a water-repellent solvent, without the influence of water, and gave 6a quantitatively after 8 h (rhombuses, graph A).[17)] Fluorous media appear to release the water produced from the active site of the catalyst. In contrast, the reaction catalyzed by 6a at 80 °C gave 8a quantitatively regardless of the above three conditions (graph B). Very importantly, 6a exhibited much higher catalytic activity than C6F5SO3H under heating conditions without the removal of water despite the weaker acidity of 6a. These experimental results show that the use of 6a gives the same rate-accelerating effect as the use of a Dean–Stark apparatus or hydrophobic perfluoromethylcyclohexane.

Fig. 5

Dehydrative cyclization of 7a in solvent. The yield of γ-pyrone 8a was evaluated by HPLC analysis.

The solvent effect for the dehydrative cyclization of 7a catalyzed by 6a or C6F5SO3H without the removal of water has also been investigated (Table 1). When the reaction catalyzed by 6a was conducted in heptane at 80 °C for 8 h, 8a was obtained in quantitative yield (entry 1). On the other hand, this reaction gave 8a in 56–64% yield in more polar solvents such as toluene, 1,4-dioxane, and propionitrile (EtCN) (entries 2–4). Similarly, the reaction catalyzed by C6F5SO3H proceeded better in heptane than in toluene, 1,4-dioxane, or EtCN (entry 1 versus entries 2–4). Interestingly, 6a and C6F5SO3H had similar catalytic activities in toluene, 1,4-dioxane, and EtCN (entries 2–4), while 6a showed much higher catalytic activity than C6F5SO3H in heptane (entry 1). In less polar solvents such as heptane, 6a should form a stable ion pair, in which the ammonium cation is tightly surrounded by two bulky N-aryl groups and an S-pentafluorophenyl group. It is conceivable that due to the hydrophobic environment created by these bulky aryl groups, the generated water is rapidly released from the active site of 6a and the hydrophobic wall prevented polar water from gaining access to the active site of 6a. On the other hand, it is suggested that 6a can not form a stable ion pair in polar solvents such as toluene, 1,4-dioxane and EtCN because of no rate-accelerating effect.

Table 1

Dehydrative cyclization of 7a to 8a catalyzed by 6a or C6F5SO3H

entry	solvent	yield [%] of 8ab
		6a	C6F5SO3H
1	heptane	100	74
2	toluene	56	55
3	1,4-dioxane	64	58
4	EtCN	64	61

Reactions were carried out with 0.2 mmol of 7a and 5 mol % of catalyst in 4 mL of solvent at 80 °C for 8 h without the removal of generated water.

Determined by HPLC analysis.

The catalytic activities of C6F5SO3H, TsOH and their N,N-diarylammonium salts for the dehydrative cyclization of 4,6,9-trimethyldecan-3,5,7-trione (7b) to γ-pyrone 8b have been examined in heptane at 80 °C without the removal of water (Fig. 6). Compound 7b has a methyl group at its 9-position, and is less reactive than 7a because of the steric hindrance of the methyl group. As in the ester condensation reactions, N,N-diarylammonium tosylates showed slightly lower catalytic activities than TsOH in the dehydrative cyclization of 7b (graph C, specific acid catalysis).[18)] In contrast, the catalytic activities of N,N-diarylammonium pentafluorobenzenesulfonates strongly depended on the structures of the N,N-diarylamines (graph D, general acid catalysis).[18)] TsOH (pKa (CD3CO2D) = 8.5) is a stronger acid than C6F5SO3H (pKa (CD3CO2D) = 11.1).[12)–15),19),20)] Compared with C6F5SO3H (circles), 3a showed lower catalytic activity due to its weaker acidity and slight hydrophobicity (triangles), while 5a (rhombuses) and 4a (squares) exhibited significantly higher catalytic activities than C6F5SO3H. In particular, the most bulky catalyst 5a had the highest catalytic activity due to its efficient creation of a local hydrophobic environment. Very interestingly, 5a showed higher catalytic activity than TsOH despite the much weaker acidity of 5a (circles, graph C versus rhombuses, graph D). These experimental results suggest that two bulky N-aryl groups and an S-pentafluorophenyl group, which surround the active site (+NH2) of 5a, synergistically accelerate the dehydration reactions. Based on the results in Figs. 5 and 6 and Table 1, the rate-accelerating effect on the 5a-catalyzed dehydration reaction can be attributed to the local hydrophobic environment in 5a. A similar tendency has been observed in the ester condensation reaction of carboxylic acids with alcohols as well as the dehydrative cyclization of 1,3,5-triketones.[13)] However, the latter reaction was promoted much more effectively than the former reaction.

Fig. 6

Dehydrative cyclization of 7b catalyzed by N,N-diarylammonium tosylates (graph C) and N,N-diarylammonium pentafluorobenzenesulfonates (graph D). The yield of γ-pyrone 8b was evaluated by HPLC analysis. rhombuses: [(2,6-Ph2C6H3)MesN+H2]; squares: [Mes2N+H2]; circles: H+; triangles: [Ph2N+H2].

X-ray single-crystal structures of the N,N-diarylammonium sulfonates suggest that a hydrophobic environment in their aggregates may play an important role. Crystal 9 was obtained by the recrystallization of 5a, which was a 1:1 molar mixture of N-(2,6-diphenylphenyl)-N-mesitylamine and C6F5SO3H in CHCl3–hexane (Scheme 7). Surprisingly, X-ray crystallographic analysis revealed that 7 is a supramolecular complex composed of two diarylammonium cations, four pentafluorobenzenesulfonate anions and two oxonium cations (Fig. 7).

Scheme 7.

Formation of 9 by crystallization of 5a.

Fig. 7

X-ray single-crystal structures of 9, dimeric 4a, dimeric 4b and [PhN+H3]n[−O3SC6F5]n. Upper, ORTEP drawing; lower, space-filling drawing. F: green; N: blue; O: red; S: yellow.

Two ammonium cations and two oxonium cations in 9 are surrounded by 12 hydrophobic aryl groups, like reverse micelles. Furthermore, the cyclic ion pair is thermodynamically and conformationally stabilized by not only four “HN+−H•••O=SO2−” and six “H2O+−H•••O=SO2−” intermolecular hydrogen bonds but also two intermolecular π-π interactions between mesityl groups and pentafluorophenyl groups, two intermolecular π-π interactions between phenyl groups and pentafluorophenyl groups and two intra-molecular π-π interactions between mesityl groups and phenyl groups: the distance between the mesityl and pentafluorophenyl groups is 3.6–3.8 Åand that between the mesityl and phenyl groups is 3.0–3.6 Å. The extremely high catalytic activity of 5a in the ester condensation and dehydrative cyclization of 1,3,5-triketones may be ascribed to the hydrophobic environment around ammonium protons in 7, which includes carboxylic acids or 1,3,5-triketones in place of water.

When crystal 9 was used as a catalyst instead of a 1:1 molar mixture of C6F5SO3H and N-(2,6-diphenylphenyl)-N-mesitylamine, the ester condensation reaction of 4-phenylbutyric acid with cyclododecanol proceeded more slowly. An equilibrium mixture of 9 and N-(2,6-diphenylphenyl)-N-mesitylamine probably exists in a 1:1 molar solution of C6F5SO3H and N-(2,6-diphenylphenyl)-N-mesitylamine in heptane. The above experimental results suggest that the ratio of 9 in a 1:1 molar solution of C6F5SO3H and N-(2,6-diphenylphenyl)-N-mesitylamine is much higher than that in a 2:1 molar solution.

The X-ray single-crystal structure of 4a is a dimeric complex composed of two N,N-dimesitylammonium cations and two pentafluorobenzenesulfonate anions (Fig. 7). The ammonium cation moiety is surrounded by six aryl groups, and the cyclic ion pair is also stabilized by two intermolecular π-π interactions. The distance between the N-mesityl and pentafluorophenyl groups is 3.5–3.6 Å. N,N-Dimesity-lammonium tosylate (4b) also forms a complex composed of two N,N-dimesitylammonium cations and two p-toluenesulfonate anions. In contrast to 9 and 4a, 4b does not exhibit intermolecular π-π interactions between the N-mesityl and tolyl groups. In contrast to the N,N-diarylammonium sulfonates, anilinium pentafluorobenzenesulfonate did not form a cyclic ion pair structure. Therefore, N,N-diarylamine structures are important for the formation of cyclic ion pairs in which the ammonium cation moieties are surrounded by aryl groups.

Aggregated cyclic ion pairs 9 and dimeric 4a may be similar to real active species in the dehydration reactions such as ester condensation and dehydrative cyclization of 1,3,5-triketones. They are stabilized by not only intermolecular hydrogen bondings but also intermolecular π-π interactions between the mesityl and pentafluorophenyl groups. Moreover, the use of a less polar solvent such as heptane also promotes tight aggregation between diarylamines and C6F5SO3H, which is less acidic and less polar than TsOH. In contrast, 4b is less active as a dehydration catalyst because of the instability of the cyclic ion pair due to the absence of intermolecular π-π interaction and the more acidic and more polar nature of TsOH. Therefore, it seems that the catalytic activity of 4b is mainly due to its strong acidity. The formation of a stable cyclic ion pair, in which the ammonium protons are located in the local hydrophobic environment, is considered to be crucial for the excellent catalytic activity.

Water molecules produced at the active site of bulky N,N-diarylammonium pentafluorobenzenesulfonates are easily exchanged for less polar substrates such as carboxylic acids and 1,3,5-triketones. Once water molecules are released from the ammonium cation moiety, the hydrophobic wall prevents polar water molecules from gaining access to the active site of the catalysts, leading to the inhibition of inactivation of the catalysts by water. In contrast, less polar substrates can easily approach the active site through the hydrophobic wall and is efficiently activated. Thus, bulky N,N-diarylammonium pentafluorobenzenesulfonates exhibit remarkable catalytic activities for the dehydration reactions without any loss of catalytic activities even under conditions without the removal of water. In contrast, sulfonic acids interact with water more strongly than with less polar substrates. Therefore, sulfonic acids are inactivated under conditions without the removal of water.

A proposed mechanism for the dehydrative cyclization of 1,3,5-triketones is shown in Scheme 8. Bulky ammonium catalyst 5a coordinates more preferentially with the carbonyl oxygen at the 5-position of 1,3,5-triketones 7 than with that at the 1-position, to avoid the steric hindrance of the substituent group indicated by R (step 1). However, enol intermediate 11 is less stable than 10 because of steric hindrance between the R group and methyl group. Therefore, 7b, which has an isobutyl group, is less reactive than 7a. Compound 11 is reversibly converted to cyclic hemiacetal 12 (step 2). Coordination of 3a with the hydroxy group of 12 makes the hydrophilic hemiacetal moiety of 12 labile in the hydrophobic environment, and the dehydration of 12 (step 3) is promoted. Dehydration is also promoted by the steric hindrance of 5a, and thus 12 is easily converted to the corresponding γ-pyrones 8. The generated water is rapidly released from the active site of 5a and easily exchanged for less polar 7, due to the hydrophobic environment (step 1). Thus, compound 5a exhibits remarkable catalytic activities without being affected by the generated water.

Scheme 8.

Proposed mechanism of the dehydrative cyclization of 1,3,5-triketones 7 (C6F5SO3− is omitted for clarity).

In summary, bulky N,N-diarylammonium pentafluorobenzenesulfonates show unusual rate-accelerating effect for dehydration reactions such as the esterification and the cyclization of 1,3,5-triketones. In particular, the most bulky and hydrophobic catalyst 5a shows much higher catalytic activity than C6F5SO3H even though 5a is a weaker acid. It is conceivable that the local hydrophobic environment created by the tight aggregation of 5a in less polar solvent efficiently promotes dehydrative reactions. The X-ray crystallographic analysis of 9, which may be the real active species, suggests that stabilization of the cyclic ion pair by intermolecular π-π interactions between hydrophobic bulky aryl groups is crucial for the creation of the hydrophobic environment.

Chiral ammonium salts as asymmetric Mannich-type catalysts

A chiral organic salt which consists of a Brønsted acid and a Brønsted base is one of the most promising catalysts in modern asymmetric syntheses. [1),2),21),22)] In general, acid–base combined salts have several advantages over single-molecule catalysts, with regard to the flexibility in the design of their dynamic complexes. Chiral ammonium salts of chiral amines with achiral Brønsted acids are typical examples of these organocatalysts with enantioselective function. 2,2′-Disubstituted 1,1′-binaphthyl is one of the most popular chiral auxiliaries of asymmetric catalysts.[23)–29)] However, bulky substituents at the 3,3′-positions of 1,1′-binaphthyl are often required to achieve high enantioselectivity in asymmetric catalyses. In sharp contrast, chiral 1,1′-binaphthyl-2,2′-disulfonic acid (BINSA, 13)[30)–32)] is a promising chiral Brønsted acid, since both the Brønsted acidity and bulkiness can be easily controlled by complexation with achiral amines without substitutions at the 3,3′-positions in a binaphthyl skeleton (Scheme 9). However, despite this potential, there had been no reports on the application of chiral 13 to asymmetric catalyses since the first synthesis of rac-13 in 1928 by Barber and Smiles.[33)–35)] In 2008, we developed a practical synthesis of chiral 13 from inexpensive 1,1′-bi(2-naphthol) (BINOL) and efficient enantioselective catalysis in direct Mannich-type reactions using 13–2,6-diarylpyridine (14) combined salts as chiral Brønsted acid–base organocatalysts in situ.[36)]

Scheme 9.

Dynamic complexation of BINSA (R)-13 with tertiary amine.

The method used to prepare (R)-13 from (R)-BINOL via the oxidation of dithiol (R)-17 is shown in Scheme 10. Thermolysis in the Newman–Kwart rearrangement of (R)-15 to (R)-16 has been dramatically improved by using a microwave technique at a lower temperature (200 °C).[37),38),40)] The oxidation of thiols (RSH) to sulfonic acids (RSO3H) is usually accompanied by the generation of disulfides (RS–SR) via intermolecular reactions.[41),42)] In particular, it seems that dithiol (R)-17 bearing two SH groups at the 2,2′-positions in a binaphthyl skeleton may be suitable for the formation of an oxidative S–S bond, intramolecularly.[37),38)] However, the oxidation of (R)-15 proceeded smoothly in 82% yield without epimerization under 7 atm of O2/KOH in HMPA. The chemical structure of the potassium salt of (R)-13 had been ascertained by X-ray diffraction analysis of its single crystal. Compound (R)-13 was isolated by ion-exchange. Thus, (R)-13 was prepared in 51% yield over five steps from (R)-BINOL, or in 82% yield in one step from commercially available (R)-17.

Scheme 10.

Asymmetric synthesis of (R)-13 and the X-ray structure of potassium salt of (R)-13 (K+ is omitted).

The enantioselective direct Mannich-type reaction[27)–29)] has been examined using (R)-13 as a chiral Brønsted acid catalyst (Table 2). Since the reaction between N-Cbz-phenylaldimine (18a) and acetylacetone (19a) proceeds without catalysts in dichloromethane at 0 °C, the slow addition of 19a is the key to preventing the achiral pathway. However, despite such care, the enantioselectivity of 20a was low (17% ee) when 5 mol % of (R)-13 was used (entry 1). Next, chiral (R)-13 (5 mol%)–achiral amine (10 mol%) combined salts prepared in situ were examined as chiral Brønsted acid–base catalysts. Pyridine, 2-phenylpyridine, and 2,6-lutidine gave (R)-20a in low yield due to the insolubility of the corresponding salts (entries 2–4). In sharp contrast, 2,6-di-tert-butylpyridine improved the enantioselectivity up to 76% ee (entry 5). Moreover, (R)-13 with 2,6-diphenylpyridine (14a), which led to a homogeneous catalyst in situ, was found to be highly effective, and (R)-20a was obtained in 74% yield with 92% ee (entry 6). N-Boc-phenylaldimine (18b), which had been reported as a sole protecting group by Terada and co-workers using pioneering chiral phosphoric acids (Scheme 11),[27)–29)] was compatible with the present reaction conditions using (R)-13•14a2, and the corresponding adduct (R)-20b was obtained in 83% yield with 85% ee (entry 7). In their catalytic enantioselective reactions, acetylacetone[27)] is the sole nucleophile, and N-Boc protection in aldimines is essential for achieving high enantioselectivities. (S)-1,1′-Binaphthyl-2,2′-dicarboxylic acid•14a2 (5 mol%) showed low catalytic activity and low enantioselectivity (8% ee) under the same conditions as in entry 6 (entry 8).[26)]

Table 2

Ammonium salts of (R)-13 as tailor-made catalysts

entry	18	amine	yield [%]	ee [%]
1	18a	–	81	17
2	18a	C5H5N	8	5
3	18a	2-Ph-C5H4N	11	10
4	18a	2,6-Me2-C5H3N	19	0
5	18a	2,6-t-Bu2-C5H3N	32	76
6	18a	2,6-Ph2-C5H3N (14a)	74	92
7	18b	14a	83	85
8b	18a	14a	85	8 (S)

Acetylacetone 19a was added at 0 °C over 1 h, and the resultant mixture was stirred for 30 min.

(S)-1,1′-Binaphthyl-2,2′-dicarboxylic acid was used instead of (R)-13. After being stirred at 0 °C for 30 min, the reaction mixture was further stirred at room temperature for 4 h.

Scheme 11.

Example of the enantioselective direct Mannich reaction using a single-molecule catalyst.

The molar ratio of 14a (0–15 mol%) to (R)-13 (5 mol%) has been optimized for the above direct Mannich-type reaction of 19a with 18a (Table 3). Interestingly, the enantioselectivities of 20a were dramatically improved when a molar ratio of (R)-13:14a was 1:≥0.75 (entries 4–9 vs. entries 1–3). As a result, a 1:1.5 to 1:2.5 ratio of (R)-11:12a was effective for achieving both a high yield and a high enantioselectivity (entries 6–8). Probably, the wide range of suitable ratios for (R)-13:14a is due to the dynamic structure of the catalysts (Scheme 9).

Table 3

Effect of the ratio of (R)-13:14a

entry	13• 14an	yield [%]	ee [%]
1	13	81	17
2	13•14a0.25	82	17
3	13•14a0.5	83	34
4	13•14a0.75	81	79
5	13•14a	82	84
6	13•14a1.5	84	90
7	13•14a2	74	92
8	13•14a2.5	76	95
9	13•14a3	68	86

Fortunately, (R)-20a was obtained in 91% yield with 90% ee with the use of 1 mol% of (R)-13•14a2 in the presence of MgSO4, which prevented the decomposition of 18a (1.5 equiv) due to adventitious moisture (Table 4, entry 1). Under these optimized conditions, N-Boc-Mannich product (R)-20b was obtained in 99% yield with 84% ee (entry 2). From 19a and a variety of N-Cbz-arylaldimines bearing electron-donating or electron-withdrawing groups in the aryl or heteroaryl moiety, the corresponding adducts (20c–j) were obtained in excellent yields with high enantioselectivities (entries 3–8). When other diketones such as 3,5-heptanedione (19b) and 1,3-diphenylpropane-1,3-dione (19c) were reacted with 18a, 20i and 20j were obtained with 95% ee and 84% ee, respectively (entries 9 and 10).

Table 4

Catalytic enantioselective direct Mannich-type reaction

entry	18 (R, Ar)	19	20	yield [%]	ee [%]
1	18a (Cbz, Ph)	19a	20a	91	90 (R)
2	18b (Boc, Ph)	19a	20b	99	84 (R)
3	18c (Cbz, o-MeC6H4)	19a	20c	99	96
4	18d (Cbz, m-MeC6H4)	19a	20d	99	89
5	18e (Cbz, p-MeOC6H4)	19a	20e	95	96
6	18f (Cbz, p-BrC6H4)	19a	20f	92	98 (R)
7	18g (Cbz, 1-Naph)	19a	20g	99	96
8	18h (Cbz, 3-Thionyl)	19a	20h	98	98
9	18a (Cbz, Ph)	19b	20i	95	95
10	18a (Cbz, Ph)	19c	20j	>99	84

The absolute stereochemistry of the products 20a and 20b has been determined by following Terada’s procedure, which includes Baeyer–Villiger oxidation. [27)] However, unexpected tertiary alcohols 21[43)] were obtained exclusively instead of the Baeyer–Villiger products when Mannich adducts 20 were oxidized under the same reaction conditions as reported by Terada.[27)] Compound 21f was determined by X-ray analysis (Scheme 12).

Scheme 12.

Unexpected oxidation of 20f and X-ray analysis of 21f.

Moreover, cyclic 1,3-diketone 19d could also be used, and the corresponding adduct 20k with a quaternary carbon center was obtained in 98% yield with a syn/anti diastereomer ratio (dr) of 83/17 and high enantioselectivity (91% ee and 96% ee, respectively) (Scheme 13).

Scheme 13.

Enantio- and diastereoselective direct-Mannich-type reaction.

A suitable chiral ammonium salt is easily tailor-made for a ketoester equivalent such as 3-acetoacetyl-2-oxazolidinone (22) (Scheme 14). Chiral ammonium salt (R)-13•14a2, which was optimized for the reaction of diketones 19 with 18, was not effective, and the desired product 23 was obtained in 86% yield with low diastereo- and enantioselectivities. In contrast, the enantioselectivity of 23 increased to 93% ee when 2,6-dimesitylpyridine (14b) was used in place of 14a. In this way, tailor-made salts (R)-13•142 make it possible to avoid preparing single-molecule catalysts in advance and offered a quick solution to this type of optimization problem.

Scheme 14.

Enantio- and diastereoselective direct Mannich-type reaction between 18a and 22.

Compound 23 could be easily transformed to β-amino carbonyl compound 24 via deprotection of the oxazolidinone moiety without a loss of enantioselectivity (Scheme 15).

Scheme 15.

Deprotecion of the oxazolidinone moiety.

In summary, BINSA (R)-13 is a highly effective chiral Brønsted acid that can be combined with an achiral Brønsted base. The combination of the achiral bulky 2,6-diarylpyridine 14 with the simple disulfonic acid (R)-13 circumvents the trouble of having to build bulky substituents at the 3,3′-positions, as is normally required in analogous binaphthyl phosphoric acid catalysts. In the presence of 1 mol% of (R)-13 and 2 mol% of 14, highly enantioselective direct Mannich-type reactions of a variety of 1,3-diketones and a 1,3-ketoester equivalent with arylaldimines proceed smoothly with high enantioselectivities. BINSA 13 is a powerful chiral auxiliary like BINOL, BINAP (2,2′-bis(diphenylphosphino)-1,1′-binaphthalene), BINAM, etc., and is expected to trigger a new frontier in acid–base chemistry in asymmetric catalyses.[44)]

Chiral ammonium salt catalysts for the enantioselective Diels–Alder reaction with α-(acyloxy)acroleins

The first enantioselective organocatalytic Diels–Alder reaction of dienes with α-unsubstituted acroleins was reported by MacMillan et al. (Scheme 16).[45),46)] Their organocatalysts, which are chiral ammonium salts of HCl or HClO4 with cyclic secondary amines derived from l-phenylalanine, are almost inactive for the Diels–Alder reaction with α-substituted acroleins, due to poor generation of the corresponding iminium ions.

Scheme 16.

MacMillan’s Diels–Alder catalysis.

In 2005, we succeeded in the first enantioselective Diels–Alder reaction of dienes with α-substituted acroleins catalyzed by chiral ammonium salts of acyclic primary amines derived from l-phenylalanine and Brønsted acids.[30)] α-Substituted acroleins are readily activated through the corresponding aldimines by catalytic amounts of primary amines and Brønsted acids. For example, the ammonium salt of (S)-N1-benzyl-3-phenylpropane-1,2-diamine (10 mol%) with 2,4-dinitrobenzenesulfonic acid (20 mol%) catalyzed the Diels–Alder reaction of cyclopentadiene with methacrolein to afford the corresponding exo-adduct with 52% ee in 73% yield. The enantioselectivity was further increased to 79% ee by using chiral triamine (24) derived from H-L-Phe-L-Leu-N(CH2CH2)2 instead of the diamine. The increase in enantioselectivity can be explained by the preference for the cis-iminium transition state due to the steric bulkiness of R as shown in Scheme 17.

Scheme 17.

Design of chiral ammonium salt catalysts for the enantioselective Diels–Alder reaction with α-methacrolein.

α-Haloacroleins are outstanding dienophiles in the asymmetric Diels–Alder process because of their high reactivity and the exceptional synthetic versatility of the resulting adducts.[31),32),47)–49)] However, α-haloacroleins are difficult to handle because they are irritants and are unstable even at ambient temperature. In contrast, our chiral ammonium salt catalyst 24•2.75C6F5SO3H was highly effective for the enantioselective Diels–Alder reaction of dienes with α-(p-methoxybenzoyloxy) acrolein (25a), which were synthetic equivalents of α-haloacroleins. Representative results are shown in Scheme 18. The Diels–Alder reaction of not only cyclic but also acyclic dienes gave the Diels–Alder adducts with high enantioselectivities. Interestingly, THF was more suitable than nitroethane as solvent for the Diels–Alder reaction of cyclopentadiene, while nitroethane was more suitable than THF for the Diels–Alder reactions of cyclohexadiene and acyclic dienes. The endo/exo selectivity had the similar tendency with that of Lewis acid-catalyzed Diels–Alder reaction with other α-substituted acroleins such as methacolein and α-bromoacrolein.

Scheme 18.

Diels–Alder reaction with 25a.

To improve the catalytic activity and the enantioselectivity for the Diels–Alder reaction of cyclopentadiene, we developed a superior asymmetric catalyst, an ammonium salt of (S)-2,2′-diamino-1,1′-binaphthyl (26) with trifluoromethanesulfoimide, both of which are commercially available, for the enantioselective Diels–Alder reaction of cyclic dienes with 25 (Scheme 19).[50),51)] The Diels–Alder reaction of α-(cyclohexanecarbonyloxy)acrolein (25b) with cyclopentadiene proceeded in EtCN at −75 °C in the presence of 5 mol% of 26•1.9HNTf2 to give the adducts in 88% yield with 92% exo and 91% ee. In a similar manner, the Diels–Alder reaction of cyclohexadiene in nitroethane gave (2S)-endo-adduct as a major diastereomer (>99% de) in 92% yield with 91% ee.

Scheme 19.

Enantioselective Diels–Alder reaction with 25b.

According to an X-ray structural analysis of 25a, the plane of the s-trans acrolein moiety and the plane of the acyloxy group can be distorted (Fig. 8, top). The transition-state (TS) structures 27 and 28 have been proposed based on a 1H NMR study and the X-ray structure of 25a, as shown in Fig. 8 (bottom). In each TS, one of the amino groups of 26 forms an aldimine with 25b and the other amino group forms an ammonium salt with Tf2NH. In TS-27, the aldimine is activated by the other molecule of Tf2NH to be an active dication intermediate. Moreover, the acyloxy group should be activated by linear intramolecular hydrogen bonding with a proton of the ammonium group in the same molecule. In TS-27, the diene approachs the si-face of the dienophile from the less-hindered side to give the (2S)-exo adduct. On the other hand, in TS-28, both the aldimine and the acyloxy group are activated by the ammonium protons in the same molecule. However, the two intramolecular hydrogen bondings of the nitrogen of aldimine with a proton of the ammonium group (N•••H−N) and the carbonyl oxygen of the acyloxy group with a proton of the ammonium group (O•••H−N) are not linear; therefore, these hydrogen bondings are weak, and TS-28 is conformationally unstable. Moreover, the aldimine of TS-28 is activated by the weakly acidic ammonium group, while the aldimine of TS-27 is activated by superacidic Tf2NH. Therefore, it is suggested that the present Diels–Alder reaction proceeds via TS-27.

Fig. 8

ORTEP plot of 25a (top) and proposed TS assemblies (bottom). Counteranions (Tf2N−) are omitted for clarity.

Chiral ammonium salt catalysts for the enantioselective [2+2] cycloaddition of unactivated alkenes with α-(acyloxy)acroleins

In 2007, we found that 24•2.6HNTf2 catalyzed the enantioselective [2+2] cycloaddition reaction of unactivated alkenes 28 with 25, to give optically active 1-(acyloxy)cyclobutanecarbaldehydes (29), and the subsequent ring expansion of 29 gave optically active 2-hydroxycyclopentanone derivatives 30 and 31 (Scheme 20).[52)] To the best of our knowledge, there are only three previous examples of catalytic enantioselective [2+2] cycloaddition reactions for the synthesis of optically active cyclobutanes or cyclobutenes. [53)–57)] The previous methods are limited to the [2+2] cycloaddition of highly nucleophilic alkenyl or alkynyl sulfides[53)–56)] and sterically demanding alkenes such as norbornene derivatives.[57)]

Scheme 20.

Enantioselective [2+2] cycloaddition reaction of 28 with 25.

First, the [2+2] cycloaddition reaction of 2,4-dimethylpent-2-ene (28a) with α-(benzoyloxy)acrolein (25c) has been examined in the presence of chiral amines (10 mol%) and Brønsted acids (x mol%) in nitroethane (Table 5). Although 24•2.75C6F5SO3H and 24•2.75TfOH were inert at 0 °C (entries 1 and 2), more acidic 26•1.9HNTf2 catalyzed the cyclo-addition even at −78 °C (entry 3). However, the enantioselectivity was moderate (64% ee for major diastereomeric cycloadduct 29ac). Fortunately, the enantioselectivity was increased to 80% ee with the use of 24•2.6HNTf2 at 0 °C. Moreover, the enantio-selectivity was increased to 85% ee when the reaction temperature was lowered to −20 °C (entry 5). The absolute and relative stereochemistry of 29ac, which was obtained in the experiments shown in Table 5, was determined to be a (1S,3R)-anti configuration based on the X-ray crystal analysis of (1′S)-camphanyl ester derived from 29ac.

Table 5

Enantioselective [2+2] cycloaddition of 28a with 25c

entry	amine•HXb	conditions [°C, h]	yield [%]	syn: anti	ee [%]c
1	24•2.75HX1	0, 24	N.R.	–	–
2	24•2.75HX2	0, 24	N.R.	–	–
3	26•1.9HX3	−78, 36	24	8 : 92	64
4	24•2.6HX3	0, 24	71	10 : 90	80
5	24•2.6HX3	−20, 48	64	8 : 92	85

28a (2 equiv) and 25c (1 mmol, 1 equiv) were used in EtNO2 (0.3 mL).

HX1 = C6F5SO3H, HX2 = TfOH, HX3 = HNTf2.

The ee value of anti-29ac was determined by chiral HPLC.

Next, several acyloxy groups of 25 have been screened for the enantioselective [2+2] cycloaddition of 28a in nitroethane at room temperature in the presence of 10 mol% of 24•2.6HNTf2 (Table 6). The result indicates that the electronic and steric effects of the acyloxy group of 25 do not have greatly influence the enantioselectivity or reactivity. Nevertheless, when α-(2,6-difluorobenzoyloxy)acrolein (25e) was used in place of 25c, the ee value was slightly improved from 73% to 84%.

Table 6

Enantioselective [2+2] cycloaddition of 28a with 25 to 29

25, R2	time [h]	29 yield [%]	syn: anti	eeb [%]b
25c, Ph	7	29ac, 74	14 : 86	73
25b, c-C6H11	6	29bc, 72	13 : 87	78
25a, p-(MeO)C6H4	7	29ac, 69	13 : 87	75
25d, p-FC6H4	18	29dc, 73	13 : 87	71
25e, 2,6-F2C6H3	12	29ec, 80	11 : 89	84

The reaction of 28a (2 equiv) with 25 (1 mmol, 1 equiv) was carried out in the presence of 24•2.6HNTf2 (10 mol%) in EtNO2 (0.3 mL) at room temperature.

The ee value of anti-29 was determined by chiral HPLC. (1S,3R)-anti-29 was the major enantiomer.

To explore the generality and scope of the 24•2.6HNTf2-induced enantioselective [2+2] cycloaddition with 25, structurally diverse alkenes 28 have been examined (Table 7). In most cases, cycloadducts 29 were obtained in slightly better yield when the reaction was performed in nitropropane, which was less polar than nitroethane. Cyclic and acyclic trialkylethenes 28a–f were reacted with α-(fluorobenzoyloxy)acroleins 25e–g to give 29 in moderate to good yield with high ee. In contrast, 1,1-, and 1,2-dialkylethenes showed no reactivity. However, the cycloaddition of a 1,1-disubstituted styrene derivative such as 28g with 25c gave 29cg with 80% ee (entry 12). The absolute and relative stereochemistry of 29ge, which was obtained in the experiment (entry 10), was also determined to be a (1S,2S,3R)-anti configuration based on an X-ray crystal analysis.

Table 7

Enantioselective [2+2] cycloaddition of 28 with 25 to 29

entry	28	25	conditions [°C, h]	29, yield [%]	syn:anti	ee [%]b
1[c,d]	Me2C=CH-i-Pr, 28a	25e	−20, 48	29ea, 63	7 : 93	95
2e	Me2C=CH-i-Bu, 28b	25gf	−20, 48	29gb, 63	6 : 94	89
3[d,e]	Me2C=CH-c-C6H11, 28c	25e	0, 48	29ec, 89	8 : 92	82
4[e,g]		25e	−20, 48	29ec, 61	9 : 91	85
5e	Me2C=CH-t-Bu, 28d	25fh	−20, 60	29fd, 67	7 : 93	91
6[e,i]		25gf	−20, 30	29gd, 49	7 : 93	91 87
7e		25e	0, 24	29ee, 63	5 : 95	82
8e		25fh	0, 48	29fe, 77	5 : 95	83
9e		25fh	−10, 48	29fe, 57	4 : 96	89
10[e,g,i]		25gf	−20, 72	29ge, 24	4 : 96	90
11[e,i]		25e	−10, 24	29ef, 37	17 : 83	87
12[c,d,i,j]	PhMeC=CH2, 28g	25c	0, 6	29cg, 20	16 : 84k	80

Unless otherwise noted, 25 (1.0 mmol, 1.0 equiv) and 28 (1.2 equiv) were used in PrNO2 (1.0 mL) in the presence of 24•2.6HNTf2 (10 mol%).

The ee value of anti-29 was determined by chiral HPLC.

EtNO2 was used.

28 (2.0 equiv) was used.

24•2.6HNTf2 (20 mol%) was used.

25g (R2 = C6F5).

PrNO2 (2.0 mL) was used.

25f (R2 = 2,4,6-F3C6H2).

1,1,2,2,3,3-Hexafluoropropane-1,3-disulfonimide was used in place of HNTf2.

Water (2.0 equiv) was added.

The relative stereochemistry of 29cg is unknown

A possible stepwise mechanism that accounts for the observed absolute and relative stereochemistries of cycloadducts 29ca and 29gd is shown in Scheme 20 and Fig. 9. The possibility of a concerted [π2s+π2s] cycloaddition and the possibility of a folded transition state for the initial Michael addition step are forbidden by orbital symmetry considerations. The possibility of a concerted [π2s+π2a] pathway is also excluded because of the steric hindrance of substrates. Initially, the enantioselective Michael addition of alkenes to a (Z)-iminium intermediate, which is generated from 25 and 24•2.6HNTf2, occurs through enantiofacial approach between the re-face of electron-rich alkenes and the si-face of the electron-deficient (Z)-iminium intermediate in an extended TS assembly 32. The (Z)-iminium isomer of 32 is expected to be more stabilized by intramolecular hydrogen-bonding interactions between R2-C=O or o-F substituents in R2 and H-N+(CH2CH2)2. Subsequently, the resulting tertiary carbocation intermediate is intramolecularly cyclized through a folded TS-33. The high anti-selectivity of cycloadducts may also be achieved by an intramolecular hydrogen-bonding interaction in 32 and 33.

Fig. 9

Possible TS-32 and TS-33 for the present [2+2] cycloaddition.

To demonstrate the synthetic utility of cycloadducts 29, 29ea was expanded to 2-(acyloxy)cyclopentanone 31ea by treatment with AlCl3 (1.2 equiv) through successive 1,2-shifts of a tertiary alkyl group and a hydride (Eq. 1).[58),59)] On the other hand, 29ge was expanded to 2-hydroxycyclopentanone 30e in 95% yield with 64% ds by treatment with Bu4NF•3H2O (2 equiv) through hydrolysis and the subsequent 1,2-shift of a tertiary alkyl group (Eq. 2). It is expected that 30e may become a new chiral common intermediate candidate in the enantioselective total syntheses of 4a-methylhydrofluorene diterpenoids such as (–)-taiwaniquinol B.[60)–63)]

In summary, we have developed a novel and useful formal [2+3] cycloaddition of 28 with 25 by an organocatalytic enantioselective [2+2] cycloaddition and subsequent ring expansion to give optically active 30 or 31 with high ee.

Chiral ammonium salt catalysts for the enantioselective Diels–Alder reaction of dienes with α-(N-acylamino)acroleins

Optically active α-amino acids as well as α-hydroxy acids are valuable chiral synthons that bear two functional groups. Chiral ammonium salts 24•2.75HX and 26•1.9HNTf2 activate α-(acyloxy)-acroleins 25 as an aldiminium cation intermediate to react with dienes or monoalkenes to provide cyclo-aliphatic α-quaternary α-hydroxy acid equivalents with high enantioselectivity. In contrast, to the best of our knowledge, there has been only one example of the enantioselective Diels–Alder reaction with α-(N-acylamino) acrolein derivatives: in 1991, Cativiela et al. reported the Diels–Alder reaction of cyclopentadiene with methyl α-(N-acetylamino)acrylate promoted by 50 mol % of chiral titanium(IV) Lewis acid (64% yield, 78% exo, 70% ee (exo)).[64)] In 2008, we developed the catalytic and highly enantio-selective Diels–Alder reaction of dienes with α-(N,N-diacylamino)- or α-(N-acylamino)acroleins to give optically active cyclic α-quaternary α-amino acid precursors.[65)] Conformationally constrained α-amino acids are valuable in biochemistry as modified peptides, enzyme inhibitors, and ligands for probing receptor recognition.[64),66)–70)]

In an initial investigation, the Diels–Alder reaction of 2,3-dimethylbutadiene with α-(N-benzoylamino) acrolein (34)[71)] has been examined in nitroethane in the presence of 20 mol% of 24•2.75 C6F5SO3H. The reaction was slow even at room temperature, and stirring for 24 h led to the desired cycloadduct with 81% ee in 59% yield (Table 8, entry 1). α-Phthalimidoacrolein (35a) was also examined instead of 34 under the same conditions as above. Both the reactivity and the enantioselectivity were increased, and stirring at room temperature for 4.5 h led to the desired cycloadduct (36a) with 92% ee in 97% yield (entry 2). The solvent effect have been investigated (entries 2–7): most aprotic polar solvents except for DMF were suitable, and the best result was observed with nitroethane. Brønsted acids were also examined as HX of 24•2.75HX (entries 2, 8–11): most sulfonic acids were effective, but on the other hand trifluoroacetic acid and superacidic triflylimide were not suitable. Another candidate 26•1.9HNTf2 did not catalyze the Diels–Alder reaction with 35a because 26 irreversibly reacted with 35a even at −78 °C in the presence of triflylimide (entry 12). 24•1.9C6F5SO3H did not catalyze the Diels–Alder reaction with 35a at −78 °C (entry 13), and did not induce high enantioselectivity at room temperature.

Table 8

Diels–Alder reaction of 2,3-dimethylbutadiene with 34 or 35 catalyzed by 24•2.75HX or 26•1.9HNTf2

entry	dienophile (R1, R2)	catalyst	conditions solvent [°C, h]	product
				yield [%]b	ee [%]c
1d	34 (Bz, H)	24• 2.75C6F5SO3H	EtNO2, 0, 36 to rt, 24	59	81
2	35a (phthal)	24• 2.75C6F5SO3H	EtNO2, rt, 4.5	97	92
3	35a (phthal)	24• 2.75C6F5SO3H	MeNO2, rt, 4.5	77	89
4	35a (phthal)	24• 2.75C6F5SO3H	MeCN, rt, 4.5	86	89
5	35a (phthal)	24• 2.75C6F5SO3H	THF, rt, 4.5	71	93
6	35a (phthal)	24• 2.75C6F5SO3H	DME, rt, 4.5	74	94
7	35a (phthal)	24• 2.75C6F5SO3H	DMF, rt, 4.5	41	74
8	35a (phthal)	24• 2.75CF3CO2H	EtNO2, rt, 84	43	42
9	35a (phthal)	24•2.75ArSO3He	EtNO2, rt, 4.5	91	90
10	35a (phthal)	24• 2.75TfOH	EtNO2, rt, 4.5	90	89
11	35a (phthal)	24• 2.75HNTf2	EtNO2, rt, 3	<5f	–
12	35a (phthal)	26•1.9HNTf2	EtCN, −78, 31	<5f	–
13	35a (phthal)	26•1.9C6F5SO3H	EtNO2, −78, 24	0	–

Unless otherwise noted, the reaction of 2,3-dimethylbutadiene (0.6 mmol) with 34 or 35 (0.5 mmol) was carried out in a solvent (0.5 mL).

Isolated yield.

Determined by chiral HPLC analysis.

2,3-Dimethylbutadiene (1.0 mmol) was used in EtNO2 (156 μL).

ArSO3H = 2,4-(NO2)2C6H3SO3H.

A complex mixture was obtained.

The absolute configuration of cycloadduct 36a, which was obtained as a major enantiomer in Table 8, was determined to be (S) by X-ray crystallographic analysis, as shown in Fig. 10.

Fig. 10

ORTEP illustration of (S)-36a with thermal ellipsoids drawn at the 50% probability level (flack parameter = 0.1228).

α-Phthalimidoacrolein 35a, which was a novel compound, was prepared by a one-pot procedure of dehydrative condensation between 2-amino-1,3-propanediol and phthalic anhydride, and subsequent oxidative dehydration under Swern conditions (Scheme 21).[71)] Other substituted α-phthalimidoacroleins 35 were synthesized in the same manner as 35a.

Scheme 21.

One-pot synthesis of 35.

To explore the generality and scope of the 24•2.75C6F5SO3H-induced enantioselective Diels–Alder reaction with 35a, representative dienes have been examined with 2.5∼10 mol % catalyst loading in nitroethane at −10 °C∼rt (Table 9). The enantioselectivity in the initial model reaction of 2,3-dimethylbutadiene with 35a was further increased to 96% ee (entry 1). Isoprene, 2-phenylbutadiene, myrcene and (E)-β-farnesene also reacted smoothly with 35a to give the desired 4-alkyl-substituted cyclohex-3-enecarboxaldehydes 37a–40a with >99% regioselectivity and 88∼94% ee (entries 3–6). Although butadiene was less reactive than substituted ones, Diels–Alder adduct 41a was obtained in 55% yield with 83% ee (entry 7). The reaction of cyclopentadiene with α-(tetrafluorophthalimido)acrolein (35b) gave endo-formylbicycloadduct 42b with 72% ds and 90% ee (entry 9). Anthracene, which was much less reactive, was also usable as a diene, although the chemical yield and enantioselectivity were moderate (entry 10).

Table 9

Enantioselective DA reaction of dienes with 35

entry	diene	35	cat. [mol %]	conditions [M, °C, h]	yield [%]b	product
						endo:exo	ee [%]c
1		35a	10	0.7, 0, 32	36a, 82	–	96 (S)
2		35a	2.5	1, rt, 48	36a, 91	–	92 (S)
3		35a	10	0.7, 0, 48	37a, 82	>99:1d	94
4e		35a	10	1, 0, 48	38a, 80	>99:1d	88
5		35a	10	0.7, 0, 48	39a, 73	>99:1d	94
6		35a	10	1, 0, 48	40a, 89	>99:1d	94
7		35a	10	1, RT, 48	41a, 55	–	83
8e		35a	10g	1, 0, 36	42a, 86	62:38	87h
9e		35b	10	1, −10, 84	42b, 73	72:28i	90 (2S)[h,i]
10		35a	10	1, RT, 11j	43a, 52	–	67

Unless otherwise noted, the reaction of diene (0.6 mmol) with 35 (0.5 mmol) was carried out in EtNO2.

Isolated yield.

Ee of major diastereomer determined by chiral HPLC analysis.

Ratio of 4- and 3-alkyl isomers. The stereochemistry of 38a was determined by X-ray diffraction analysis.

THF was used instead of EtNO2.

Diene (2 equiv) was used.

H-L-[3-(2-Naph)Ala]-L-Leu-N(CH2CH2)2-reduced triamine was used instead of 24.

54% ee (exo-42a), 57% ee (exo-42b).

The relative and absolute stereochemistries of major diastereomer of 42b were determined by X-ray diffraction analysis after its derivation to (R)-N-phenylethylamide.

11 days.

Phthalimido groups of Diels–Alder adducts 36–43 were deprotected in high yield by the treatment with hydrazine after conversion to the corresponding methyl esters (Scheme 22). Norbornene derivatives are particularly valuable as important optically active synthetic intermediates of bioactive alkaloids such as norbornene-2-amino-2-methanol derivatives[72), 73)] and (–)-altemicidin.[74)]

Scheme 22.

Deprotection of the phthalimido group of 42a.

Considering that 24•2C6F5SO3H is much less active than 24•2.75C6F5SO3H as a catalyst for the Diels–Alder reaction of dienes with 35 as well as α-(acyloxy)acroleins,[59),66)] we assume that 24•3C6F5SO3H may be a real active catalyst, which activates 35b as aldiminium salt (46) with 24•3C6F5SO3H. In our previous papers,[30),52)] it was assumed that (Z)-aldiminium salt derived from 24•2HX and 25 would be a key intermediate. However, an aldiminium salt derived from 26•3HX and 25 may be more favorable. The (Z)-isomeric preference of 46 is supposed based on theoretical calculations of geometries of its analogous aldiminium salt (47) derived from α-(maleimido)acrolein and 24•3HCl. The geometries of 47 have been optimized at DFT calculations with B3LYP using the 6-31+G(d,p) basis set (Fig. 11). The relative energy of (Z)-47 is 2.8 kcal/mol lower than that of (E)-47, and the re-face of the enimide moiety of (Z)-47 is sterically shielded by the benzyl substituent.

Fig. 11

The relative energy difference between (E)- and (Z)-geometries of aldiminium salt 47 based on theoretical calculations.

Cyclopentadiene approachs enantioselectively the si face of the electron-deficient enimide moiety to give endo-(2S)-42b as a major isomeric product. Thus, as well as the (Z)-isomeric preference of 47, it is expected that (Z)-TS-48 is preferred to (E)-TS-48 (Fig. 12).

Fig. 12

Possible (Z)-TS-48 and (E)-TS-48.

In summary, we have developed a catalytic and highly enantioselective Diels–Alder reaction of dienes with 35 to provide cyclic α-quaternary α-amino acid precursors for the first time. Chiral triamine 24, which is conformationally more flexible than 26, could be used as a catalyst ligand for cycloadditions for not only 25 but also 35.

Conclusion and prospect

The dynamic ammonium salt catalysis for selective organic transformations conducted in our laboratories is reviewed. Bulky N,N-diarylammonium pentafluorobenzenesulfonates show unusual rate-accelerating effect for dehydration reactions such as the esterifictaion and cyclization of 1,3,5-triketones. The role of the locally created hydrophobic environment is reasonable for the observed high conversion reactions. Currently our efforts are focused on exploring this overall mechanism in greater detail. Chiral ammonium salts of achiral amines with chiral BINSA are effective as asymmetric Mannich-type catalysts. Chiral ammonium salts of chiral amines with achiral Brønsted acids are effective as asymmetric [2+4] and [2+2] cycloaddition catalysts. Further studies on the exploration and applications of dynamic ammonium salts as functional catalysts are actively underway in our laboratory.