Non-Covalent Interactions in Enantioselective Organocatalysis: Theoretical and Mechanistic Studies of Reactions Mediated by Dual H-Bond Donors, Bifunctional Squaramides, Thioureas and Related Catalysts

Abstract

Chiral bifunctional dual H-bond donor catalysts have become one of the pillars of organocatalysis. They include squaramide, thiosquaramide, thiourea, urea, and even selenourea-based catalysts combined with chiral amines, cinchona alkaloids, sulfides, phosphines and more. They can promote several types of reactions affording products in very high yields and excellent stereoselectivities in many cases: conjugate additions, cycloadditions, the aldol and Henry reactions, the Morita–Baylis–Hilman reaction, even cascade reactions, among others. The desire to understand mechanisms and the quest for the origins of stereoselectivity, in attempts to find guidelines for developing more efficient catalysts for new transformations, has promoted many mechanistic and theoretical studies. In this review, we survey the literature published in this area since 2015.

1. Introduction

The activation of reactants by hydrogen bonding and other noncovalent interactions has enabled the development of an enormous variety of reactions in recent years, and consequently, the production of chiral substances for several applications. This was possible due to the discovery of an array of very efficient chiral bifunctional catalysts, mostly thiourea- or squaramide-based, usually combined with chiral amines or cinchona alkaloids, and more recently also amides, phosphines or sulfides [1,2,3,4]. Urea, thiosquaramide and even selenosquaramide chiral bifunctional catalysts are also starting to appear. They are usually assembled on a modular basis, which allows for easy structural modifications and optimization of the results.

The hydrogen bond-donors differ in duality, rigidity, H-bond spacing, H-bond angle, and pK_a, which influence their catalytic behavior [1,5]. The average distance between the N–H groups in these substances is shown in Figure 1a. Squaramides are distinct from the ureas in their hydrogen bond donor and acceptor character since, upon hydrogen bond formation, aromaticity is enhanced in the latter. The cyclobutadienone ring contains two coplanar carbonyls and two NHs that are almost coplanar and essentially sp² hybridized so that the lone pairs are available for conjugation into the π-system orthogonal plane [6].

Similarly, the ability of thioureas to recognize cations is also more limited [6]. As a result of their ambivalent hydrogen bonding characteristics, squaramides also form dimeric species.

The distance between the N–H groups is larger in squaramides than in thioureas. Due to the geometry of the cyclobutenedione ring, the N–H groups are also oriented towards each other by ~6°, a characteristic, which is absent in the amido/thioamido groups of ureas/thioureas [7]. This property may result in increased linearity in hydrogen bonding with some substrates and provide different binding properties in the transition state.

The determination of the pK_a values has been the subject of some studies. Since, according to the proton affinity–pK_a equalization principle, the strength of hydrogen bonds can be approximated by the pK_a values of the donor and the acceptor species, these values can be used to characterize the catalytic activity and allow predictions to be made that are useful for developing novel catalysts [8]. The higher the acidity of a hydrogen-bonding donor, the stronger is the resulting hydrogen bond, and therefore, introducing stronger hydrogen-bonding donor groups in a catalyst should result in higher catalytic activity. In addition, the formation of a strong hydrogen bond should also help with stereocontrol and enantioselectivity. Figure 1b shows the pK_a values of some well-known thiourea- and squaramide-based catalysts, determined by applying the classical overlapping indicator method in DMSO through UV spectrophotometric titrations [5,8,9,10]. Comparing the pK_a values of related catalysts, it was found that the pK_a values of squaramides are lower than those of their thiourea analogs, with a gap of 0.13–1.97 pK_a units [8]. The actual value itself depends on the nature of other groups attached to the thiourea/squaramide group. However, since squaramides are more acidic than thioureas, they form stronger hydrogen bonds, which may explain why squaramides can often be used at lower catalytic loadings and display higher activity in a range of reactions.

Li, Cheng and coworkers found, from a plot of pK_a values of thioureas versus the pK_a values of the analog squaramides, that there is a good correlation, two good straight lines with R values of 0.995 and 0.932 were obtained, which implies that if desired, the pK_a values of squaramides might be roughly estimated from those of the corresponding thioureas, which have been more studied [8]. In addition, the linear free energy relationship (LEFR) between the acidity and activity of thiourea catalysts was determined with the Michael addition of diethyl malonate to nitrostyrene as a model reaction. Excellent LEFRs were observed for the stereoselectivity correlated with the pKa values of the corresponding thioureas. Although there were significant differences between the catalysts, the slopes of the LFERs between pKa values and enantioselectivities were almost identical, which suggests a similar process for forming the transition states of these reactions, and hence that the catalysts react by similar mechanisms.

Lu and Wheeler demonstrated that thiosquaramides provide even stronger hydrogen-bonding interactions, and since the introduction of chiral bifunctional catalysts containing this motif by Rawal, they are gaining more interest [11,12].

The literature on mechanistic and density functional theory (DFT) studies concerning hydrogen-bonding organocatalysts has been reviewed occasionally [13]. In many cases, the B3LYP method was used since it has become a favorite in organocatalysis, as it yields accurate geometries and energies. Although we are still far from predicting exactly what will be the stereochemical outcome of a reaction, these studies aim to clarify the mode of action of the catalysts, and to reveal the nature of asymmetric induction, the factors responsible for it, so that better and more efficient catalysts can be designed for future applications. In this review, we survey the information published since 2015, when the last general overview of the field was reported [13].

The review is divided into the following sections:

• Introduction;
• Michael and other conjugate addition reactions;
• Cycloaddition reactions;
• Aldol and Henry reactions;
• Miscellaneous reactions involving anion-binding catalysis;
• Miscellaneous;
• Conclusions.

2. Michael and Other Conjugate Addition Reactions

2.1. Additions to Nitroalkenes

The largest number of papers encountered covering studies falling within the scope of this review were developments on the Michael addition reaction and its variants [14,15]. The first report on utilizing a bifunctional thiourea-based organocatalyst in synthesis was by Takemoto and coworkers in 2003 [16]. It involved a Michael addition of malonates (1) to nitroolefins (2) that were achieved with high yields and ees with catalyst C1 in toluene at room temperature (rt) (Figure 2a). Catalytic asymmetric versions of this reaction were known at the time but required either metal catalysis or strict reaction conditions. This reaction opened the way to intensive research, and soon after, this and other thiourea-based bifunctional catalysts were applied in several reactions, e.g., the aza–Henry reaction, dynamic kinetic resolution of aza-lactones, Michael addition to α,β-unsaturated imides, the aldol reaction, sp²-alkylations, spiro-ketal formation and many others [1,2,3,4]. The authors showed through experiments using only either the amine or the thiourea as catalysts that for any significant amount of product 3 to be obtained and for high enantiomeric excesses to be achieved, the simultaneous presence of both units in the catalyst molecule was essential. Kinetic studies to elucidate the reaction mechanism suggested that the active catalyst is a monomeric species, and a bifunctional mode of activation of both reactants was proposed [17]. The catalytic cycle (Figure 2b) begins with the deprotonation of the diester by the tertiary amine moiety of the catalyst and the formation of complex 4. Enolate addition takes place preferentially on the Si face of the nitronate via TS2, affording complex 5. Proton transfer to the nitronate yields the product (3a) and regenerates the catalyst.

Soon after, two theoretical studies were reported with conflicting results. Liu and coworkers identified, through DFT calculations, proton transfer to the nitronate carbon of complex 5 as the rate-determining step in the catalytic cycle [18]. Papái and coworkers identified two possible binding modes, A and B, for TS2, in which either the nitronate bonds to the thiourea moiety and the diester to the tertiary amine (binding mode A) or vice versa (binding mode B) (Figure 3) [19]. It was calculated that binding mode B is energetically favored by 2.7 kcal mol⁻¹ and that carbon–carbon bond formation is the rate-determining step. Other studies have been performed on this reaction and other bifunctional thiourea-based organocatalyzed Michael additions since then, but only the more recent literature is reviewed here. NMR, crystallographic, and computational studies on related systems have revealed that in the lowest energy-binding conformation, the catalyst displays a “syn, anti” configuration of the thiourea functional group and not the traditional thiourea “anti, anti” configuration shown in Figure 4a [20,21,22,23]. Generally, there are three main activation modes possible to explain thiourea-tertiary amine organocatalysis, namely mode A, attributed to Takemoto et al. [17], mode B, attributed to Pápai et al. [19] and mode C, attributed to Wang et al. [24] (Figure 4b). These three possible modes of activation are usually taken into consideration in this field in theoretical studies when a search is done for the transition state, which better describes a certain reaction and the origins of stereoinduction.

A recent study by Hirschi, Vetticatt and coworkers on the Michael addition of malonates to nitro olefins aimed to clarify the mechanism of catalysis by tertiary amine thiourea organocatalysts [25]. The addition of diethyl malonate to trans-β-nitrostyrene catalyzed by Takemoto’s catalyst (C1) was selected as a model system. Kinetic isotope effects (KIEs) were determined for both the malonate and the nitrostyrene using NMR at natural abundance from starting material analysis [26]. At ~80% conversion of the nitrostyrene (−65% in the malonate), the unreacted substances were isolated, and their ¹³C isotopic composition was compared to samples of both substances that had not been subjected to any reaction. KIEs were calculated in the usual manner (Figure 5).

The results revealed a significant primary ¹³C KIE on C1 of 1a (∼3%) and C1′ of 2a, suggesting that both atoms were involved in the rate-determining TS, i.e., that carbon–carbon bond formation is the first irreversible step of the catalytic cycle, as proposed initially by Takemoto and Pápai. DFT calculations were also performed for the transition structures in each step in the catalytic cycle, using the B3LYP method and a 6-31+G** basis set [27]. A polarizable continuum model (PCM) for toluene, as implemented in Gaussian 09, was used to consider solvent effects. The procedure took into account the full system (89 atoms), and it has been shown previously to accurately describe the energetics of other bifunctional thiourea-catalyzed reactions. It was the first time that it was applied to Takemoto’s full system [28]. Values of ¹³C KIEs were also obtained from scaled vibrational frequencies of the respective transition structures using the program ISOEFF98. A Wigner tunneling correction was used. The resulting theoretical predictions at the key carbon atoms of 1a and 2a were found to be in good agreement with the experimentally obtained values, supporting the above conclusions.

The relative energies of the transition state structures leading to the two isomers were calculated. Single point energies were obtained using B3LYLP-D3(BJ)/6-311++G** PCM (toluene). The energies of the lowest energy structures for C–C bond formation, the enantioselectivity determining step, were determined using DFT calculations based upon existing models in the literature and a conformational search using quantum mechanical simulations as implemented in DFTB⁺. The lowest energy structure found, TS2-BMB-S, giving rise to the major enantiomer, is 2.4 kcal mol⁻¹ lower in energy than TS2-BMB-R, leading to the minor isomer (Figure 6) [25]. These values correspond to a predicted ee of 97% (S), consistent with the observed value (93% ee (S)). The origin of enantioselectivity for this reaction may be inferred from an analysis of the interactions between the different components involved. TS2-BMB-S and TS2-BMB-R possess very similar H-bonding networks. The geometry around the forming C–C bond explains the difference in energy between the two transition states. To accommodate the H-bonding stabilizing interactions with the catalyst, the substituents adopt a staggered conformation (dihedral angle of ∼60° between the alkene of 2a and the ester groups of 1a) in TS2-BMB-S during C–C bond formation. On the contrary, transition structure TS2-BMB-R has a more eclipsed configuration for these groups (dihedral angle of ∼30°) when the C–C bond is forming. The result is increased steric interactions between substrates in TS2-BMB-R compared to TS2-BMB-S, and this is very likely the origin of enantioselectivity in this reaction.

A recent study performed by Pliego Jr on the addition of acetylacetone to β-nitrostyrene catalyzed by Takemoto’s catalyst in toluene used the PBE functional for geometry optimization and the DLPNO–CCSD(T) benchmark method for single-point energy [29]. The complete free energy profile calculated for the reaction supported the theory that carbon–carbon bond formation is the rate-determining step. The overall barrier was calculated to be 22.8 kcal mol⁻¹ (experimental value ~20 kcal mol⁻¹). It was found that three transition states with S stereochemistry and one with R stereochemistry contribute to the enantioselectivity, and the ee was calculated as 88% (experimental value = 84–92%). Some functionals were tested for single-point energy using the DLPNO–CCSD(T) method as a reference. The M06-2X functional provided the best results, even compared to double-hybrid functionals. It was also concluded that the additivity approximation of the correlation energy worked well, and it can be used as an alternative method to obtain accurate electronic energies.

In recent years carbohydrates have been used more frequently as chiral modular units for bifunctional organocatalysts [30]. In 2016 Khan, Narula and coworkers developed thiourea-based catalysts incorporating d-glucose as a core unit and tested them on the addition of ketones to nitroalkenes. One of them, C2, provided high ees and excellent stereoselectivities on the addition of cyclohexanone (6) to nitroolefins (7) (Figure 7) [31]. DFT studies, with the Gaussian 09 program package and the B3LYP/6-311G(d,p)//B3LYP/6-31G(d) basic set, were utilized to explain the stereochemical outcome of the reaction leading to products 8. The QM/MM calculations suggested that cyclohexanone plays two roles, that of solvent as well as of reactant, in the rate-determining step, imparting 31.91 kcal mol⁻¹ of energy towards product formation.

A Michael addition of isobutyraldehyde (9) to nitroalkenes (2) was achieved in a highly enantioselective manner by Chinchilla, Nájera and coworkers with primary amine-guanidine C3, derived from trans-cyclohexane-1,2-diamine, as catalyst (Figure 8) [32]. Besides other optimizations, with DMF proving to be the best solvent, the addition of water was found to increase both the reaction rate and the ees leading to 10a (10, Ar = Ph) (at 25 °C, with DMF/imidazole, 99% yield, 48% ee, with DMF/imidazole/water, 99% yield, 70% ee). At lower temperatures, the ee improved (to 80%) with a 10% decrease in yield. DFT theoretical calculations helped to explain the origin of the stereoselectivity observed, as well as the role of water in the reaction [32].

Taking into account intramolecular H-bonding only (TS_H-S vs. TS_H-R), the calculations revealed a preference for the transition state leading to the wrong enantiomer, the (S)-enantiomer, found to be 1.8 kcal/mol lower in energy than TS_H-R. This is the result of having both the nitrostyrene and the guanidine subunit both in the lower face of the enamine (from our point of view), adopting a less strained disposition. In TS_H-R, the structure needs to twist to form the internal H-bonds, adding strain to the transition state structure. In practice, the enantiomer obtained was the (R)-enantiomer, suggesting that intramolecular H-bonding was not present. Another transition state was found (TS_W-S), in which the nitro group was activated by the solvent (implicit water solvent model), 1.7 kcal/mol lower in activation energy than TS_H-S. In fact, water may serve two purposes, to lower the activation energy, since it can solvate better the more polar transition state (TS_W-S vs. TS_H-S), and to form intermolecular hydrogen bonds with the nitro group, disrupting the intramolecular ones.

Although the lack of intramolecular bonding implies much more molecular freedom and a higher number of possible conformations, the two most stable conformations of the reactive enamine could eventually be found (Figure 9a,b). In theory, nitrostyrene may approach each enamine following two different trajectories. In (a), the face leading to the (S)-enantiomer is blocked by the guanidine group and can be ruled out. The remaining three approaches are feasible, but there is not much difference in energy between them. Hence it appears that the (R)-enantiomer is obtained preferentially due to the predominance of the sum of TS_A-R and TS_B-R over TS_B-S, and the non-existence of TSA-S, i.e., the steric effect of the guanidine group seems to be the final reason for the preference observed.

In 2019, Zlotin and coworkers developed a new squaramide-based organocatalyst, which promoted the conjugate addition of β-dicarbonyl compounds (11) and kojic acid derivatives (13) to nitroolefins, to yield products 12 and 14, respectively, in nearly quantitative yields and up to 99% ee, with a catalytic loading of only 1 mol% (Figure 10) [33].

The best catalyst out of a homologous series tried was C4, a C₂-symmetric tertiary amine-squaramide organocatalyst. The reactions with kojic acid derivatives worked well under “green” conditions (in 96% EtOH, or even in pure water). The catalyst could also be easily recovered since it has extremely low solubility in organic solvents, and it was reused up to seven times, making it attractive for applications in the industry.

To ascertain if one or both tertiary amino groups in the catalyst participate in the reaction, the relationship between the ee values of the catalyst and those of the corresponding products were determined in protic and aprotic solvents (96% EtOH, H₂O, and CH₂Cl₂). Irrespective of the solvent, a linear effect was observed for both compounds, suggesting that diastereomeric associates (especially meso forms) between C4 and the reaction products were absent. This can only happen if desymmetrization occurs and only one of the two tertiary amino groups in C4 acts as deprotonating agents in the corresponding transition states TS1 and TS2. These deductions were also confirmed by experiments in which acid was added to the reaction mixture. The addition of 1 mol% of TFA did not have a negative impact on the results (on the yield or ee of 12). However, if enough TFA was added to protonate both amine groups, i.e., 2 mol%, the catalyst was completely deactivated, and no reaction was observed.

In 2017 Rawal introduced bifunctional thiosquaramides as a novel type of organocatalyst [34]. A range of novel thiosquaramide-tertiary amines was tested on the enantioselective conjugate addition reaction of barbituric acid (15) to nitroalkenes (Figure 11). In the same way that changing an oxygen atom in urea to sulfur makes the product thioureas more catalytic active than the ureas, the new thiosquaramide catalysts based on the squaramides also proved to be highly active. The squaramides, originally introduced by Rawal and coworkers in 2008, have found many applications in organic synthesis, particularly as bifunctional organocatalysts, and have become one of the major types of organocatalysts available for developing novel reactions [12]. Despite their popularity, one of their inherent properties is low solubility in nonpolar solvents and a limited ability to modulate the pK_a of the donor hydrogens. These characteristics limit their utility in some applications. The new thiosquaramides are more acidic and significantly more soluble in nonpolar solvents than their oxo-counterparts, showing great promise for future applications. In this work, a series of barbiturates (16) was produced in very high yields and excellent ees, with only 0.5 mol% of catalyst. However, the catalysts could be even more efficient. In a model reaction with nitrostyrene, i.e., 7a, R¹ = Ph and 15a (R² = Ph), when the catalytic loading was lowered to 0.05 mol%, the product (16a) was obtained with 96% ee and the conversion (determined by NMR spectroscopy) was >98% in a 10 h reaction. The corresponding oxosquaramide (C6) at 0.5 mol% afforded identical results to C5 in terms of yield and a similar ee (97 versus 95%, C5/C6). X-ray crystallography revealed the presence of two rotameric forms. The H–H distance of the two rotamers was shown to be slightly smaller than those of the corresponding oxosquaramide by X-ray crystallography. Hartree−Fock calculations for C7 at the 3-21G level of theory gave an H−H distance of 2.60 Å, which suggests that the smaller distance is due to the interactions with the sulfur atoms rather than packing forces or other interactions in the crystal structure. Intermolecular hydrogen bonds to give ladder structures, analogous to those found in the oxosquaramides, were not observed.

The development of an enantio- and diastereoselective reaction provides access to the desired enantiomer of a substance whose properties may be quite distinct from those of the other enantiomer, which is of importance for several applications. For example, in pharmaceuticals, the bioactivity of one enantiomer may be different from that of its antipode, and it is not only the absolute configuration of each chiral centre that may be vital, but also the relative configuration if more than one chiral center is present. It may be desirable to access all enantiomers separately from each other, for example, to use authentic standards to test for purity in a manufacturing process, which can be a very challenging task. The development of new strategies to perform diastereodivergent asymmetric synthesis is, therefore, a subject of great interest [35,36]. The utilization of the same synthetic route for all stereoisomers wanted may be possible if a simple change in catalyst provides alternative isomers as the main product.

In 2020 Wang, Zhao and coworkers reported a catalyst-controlled stereodivergent asymmetric Michael addition of α-azido ketones 17 to nitroolefins 7 (Figure 12) [37]. A squaramide-based catalyst (C9) afforded selectively anti-adducts, while a new bifunctional tertiary amine-phosphoramide catalyst (C10) provided preferentially syn-adducts. The resulting multifunctional tertiary azides (18) were converted to spiro-pyrrolidines with four continuous stereocenters in a one-pot operation. Mechanistic studies were performed to elucidate the reasons for the diastereoselectivity switch, which took place simply by changing the two H-bond donor catalysts.

To elucidate the mode of interaction of the catalysts with the substrates, control experiments were performed. In a model reaction producing 18a (Figure 13), it was observed that in the case of C9, the presence of two NH groups was critical for the yield, although moderate ees could be obtained. Blocking NH(3) with a methyl group caused a reversal in diastereoselectivity. Thus, both NH groups are important for a high yielding and stereoselective anti reaction to take place. In the reaction promoted by C10, both the NH group and the tertiary N are critical for high reactivity and high ees to be obtained. The presence of the ring nitrogen ensures that there is syn selectivity.

A search for nonlinear effects revealed that in the case of C10, a monomeric catalytic species was operating since there was a linear relationship between the catalyst ee value and that of the product syn-18a. No conclusions could be drawn for the reactions catalyzed by C9 due to its very low solubility in CH₂Cl₂. Theoretical studies were performed, and from the lowest energy TS structures, the main interactions responsible for the reactivities observed could be identified (Figure 14). It was found that, while the squaramide-catalyzed reaction proceeded with a TS (TS1-C9) in which both squaramide N–H bonds bind to an enolate intermediate, in the phosphoramide-mediated reaction, an amide N–H bond and an alkylammonium ion formed in situ interact with the nitroolefins, while the enolate is stabilized by nonclassical C–H/O hydrogen-bonding interactions (TS1-C10). This model has no precedence in the literature.

The thiourea-catalyzed Friedel–Crafts alkylation reaction between indoles and nitroalkenes (e.g., 19a and 7b) has been the subject of a few studies since its first report by Ricci and coworkers [38]. The utilization of thiourea-aminoindanol catalyst C11a to promote the reaction afforded the alkylated indole (20a) in 78% yield and 85% ee (Figure 15). A bifunctional mode of action was envisioned, and TS1 and TS2 were postulated as being the TSs responsible for the results obtained. In 2014 Herrera and coworkers reported the results of theoretical studies performed on this reaction to clarify the origin of stereoselectivity [39]. Of interest were previous results that showed that if the configuration of the thiourea-aminoindanol was R,R, i.e., as in C11c, there was hardly any reaction; with C12, missing a hydroxyl group, the product was racemic; and with the analogous catalyst C13, lacking an aromatic ring in the aminoindanol skeleton, the yield was very low and the ee moderate only (Figure 15) [38,40]. Initially, it was envisaged that the hydroxyl group would drive the reaction to one face of the indole in preference to the other, while the configuration of product 20a was dependent on the enantiomer of the catalyst used.

The very low reactivity and selectivity of catalysts C12 and C13 are quite noticeable. The lack of a hydroxyl group to direct the attack of an incoming nucleophile can explain the lack of selectivity, but not the low reactivity, which suggests that the hydroxyl group may play yet another role in the reaction. In C11c and C11d, the hydroxyl group occupies a trans orientation, and hence the results obtained suggest the importance of a cis relationship in order for the hydroxyl group to play its role in directing the attack of the indole in a stereoselective manner. Theoretical calculations were performed to clarify the origin of the enantioselectivity. The calculations were carried out at the PCM(CH₂Cl₂)/M06-2X/6-311G(d,p) level. A simplified system was studied initially, but the complete system was also studied with catalyst (1R,2S)-C11a, indole (19a) and nitrostyrene (7b). In all the transition states obtained for the Friedel–Crafts alkylation reaction using a catalyst (1R,2S)-C20a, the hydroxyl group of the catalyst interacts with the NH group of the indole through H-bonding, i.e., in an H–O⋯H–N interaction, leading the nucleophilic attack as predicted (Figure 16). The small difference in activation energy for indole attack (TS1, 0.0 kcal mol⁻¹) for the Si face and (TS2, 2.1 kcal mol⁻¹) for the Re face explains the differences in ee obtained (85% ee). A product with (R)-configuration is expected, as observed. In all structures except in TS2, there is bidentate coordination of the nitroalkene, as observed previously by Etter and coworkers [41], which helps to create a more rigid transition state involving all three molecules. The higher stability of TS1 may be the result of a less hindered packaging, minimizing repulsive forces, with the relative orientation of the indole-nitroalkene playing a crucial role in the selectivity. An interesting interaction was found between the hydroxyl group and one of the oxygen atoms of the nitroalkene (O–H⋯O–N=O) (TS1, TS2, TS5 and TS6), which supports the lack of reactivity observed when an OH group is not present in the catalyst, or it is not placed in the correct position (Figure 16).

In 2017 Fan and Kass achieved high yielding and highly stereoselective Friedel–Crafts alkylations of indoles (19) with nitroalkenes using as catalysts a series of methylated and octylated pyridinium- and quinolinium-containing thiourea salts with a chiral 2-indanol substituent, e.g., C14 (Figure 17) [42]. The new organocatalysts are positively charged analogs of the privileged bis(3,5-trifluoromethyl)phenyl-substituted thioureas. They were shown to be much more active catalysts in many cases, despite the absence of an additional hydrogen bond donor or acceptor site (i.e., the presence of a heteroatom−hydrogen bond or a heteroatom with free lone pairs). Yields in the order of 30% were observed with indoles bearing an electron-withdrawing group, i.e., 5-Cl. Otherwise, they were higher than 75% in reactions conducted at–35 ^°C. The temperature was kept low since the ee was largely influenced by temperature.

Kinetic studies showed that the reaction is first-order in both indole and trans-β-nitrostyrene. A plot of the second-order rate constants versus the square of the catalyst mol% is linear, which suggests that the active catalytic species of the thiourea is dimeric. It was postulated as having structure C14a. In the ¹H NMR spectra of the catalyst recorded from 5.0 to 20.0 mM, it was observed that both NH signals moved downfield linearly with increasing concentration, consistent with the occurrence of a rapidly occurring monomer/dimer equilibrium. These results suggest that the association constant is small and that the resting state for the catalyst is largely monomeric.

2.2. Additions to Enones and Unsaturated Esters

The first cinchona-thiourea catalyzed the Michael addition of nitroalkanes to enones was reported independently by four different groups in 2005 [43,44,45,46]. Given the importance of this C–C bond-forming reaction, it has been the subject of a few computational studies, but the mechanism still remains unclear, and it continues to attract attention. Three activation modes are possible, mode A, mode B and mode C, as shown in Figure 4b [17,19,24].

Since none of these studies explained the origin of selectivity, Grayson undertook a new work in 2017 (Figure 18) [47]. Quantum mechanical calculations were performed using Gaussian 09 (Revision D.01). Mode A, B and C complexes are expected to exist in rapid equilibrium, and therefore, their relative thermodynamic stabilities do not determine the course of the reaction (Curtin−Hammett conditions). In this work, the relative stabilities of the C−C bond-forming TSs (the rate-determining step) were determined to find the favored activation mode and explain the origin of stereoselectivity. Catalyst C16 was used. DFT calculations showed that the preferred mode of activation is B, in which the electrophile is activated by the catalyst protonated amine and the nucleophile through hydrogen bonds with the thiourea moiety. The TSs that lead to the major and minor products following mode B of activation have s-cis and s-trans enone conformations, respectively, which differ from each other only by rotation of the quinoline ring by ∼180° (TS_thio-Bcis-(major) and TS_thio-B_trans-(minor), Figure 19). Previous works have shown that these conformations are strongly preferred over all other possibilities [48]. The other two possible TSs may be obtained if the quinoline ring methoxy group is rotated by ∼180° so that the methyl group is oriented away from the thiourea moiety, but the new conformations were found to be disfavored by at least 2.2 kcal mol⁻¹ for the syn and anti-open conformations of TS_thio-B_cis-(major) and were disregarded.

The lowest energy TS, TS_thio-B_cis-(major), leads to the major product observed experimentally (Figure 19). Unfavorable interactions between the enone’s phenyl substituent and the catalyst’s quinuclidine ring raise the energy of TS_thio-B_cis-(minor) relative to TS_thio-B_cis-(major). TS_thio-B_cis-(major) is also stabilized by π-stacking interactions between the enone’s phenyl group and the catalyst’s aromatic ring, which does not happen in TS_thio-B_cis-(minor). In the other two conformations examined, TS_thio-B_trans-(major) and TS_thio-B_trans-(minor), the enone has the less favorable s-trans conformation and are, therefore. They are higher in energy. They also have no favorable π-stacking interactions. The s-cis conformation of the enone on its own was calculated to be favored over the s-trans by 1.0 kcal mol⁻¹. All mode A TSs are disfavored relative to TS_thio-B_cis-(major) because in mode B TSs stabilize the enone’s developing alkoxide to a greater extent than mode A TSs. In the first case, there is proton transfer from the quinuclidinium ion. In the second, there is hydrogen bonding from the thiourea. Attempts to find TSs corresponding to mode C, optimized instead to mode A.

The reaction was also studied with catalysis by cinchona squaramide C17. Mode B was again favored over mode A. For generality, the Michael addition of nitromethane to cyclohexenone was also studied. The calculations showed that in the case when the reaction is catalyzed by squaramide C17 mode B is also favored, by 2.0 kcal mol⁻¹, when the two lowest energy transition states found for each mode of activation are compared, confirming the generality of the reaction.

Furthermore, in 2017 Song and coworkers reported the addition of nitromethane to α,β-unsaturated ketones 24 containing pyridine [49]. A series of cinchona-thiourea-based catalysts were evaluated for the reaction, and finally, high yields and ees could be obtained with C19 or ent-C19. The origins of enantioselectivity were probed through computational studies. DFT calculations showed that in the most stable Cat–Nu complex, there were hydrogen bonding interactions between the oxygen on the nucleophile and an N–H group on the thiourea as well as with the protonated quinine amine, in line with Wang´s mode C of activation (Figure 4b) [19]. The reaction was postulated to proceed via TS1 (Figure 20). It was also shown that the products were highly active against rice bacterial leaf blight, with the S-enantiomer displaying higher activity than the R-enantiomer (for (S)-25a, b, (a: R = 2-FC₆H₄, b: R = 4-ClC₆H₄) at 100 μM/mL, 100% inhibition).

In 2011 Rodriguez and coworkers reported the first organocatalyzed enantioselective Michael addition of β-ketoamides (26) to enals and enones (e.g., 27) (Figure 21) [50]. Products 28, containing one all-carbon quaternary stereogenic center, are useful synthetic intermediates. The reaction was subsequently the subject of further experimental, spectroscopic and theoretical studies [51]. Tertiary β-ketoamides did not react, showing that the presence of amide hydrogen is crucial for the reaction to occur. Neither did an enone containing an internal double bond, i.e., chalcone. NMR studies with 26a (R = Ts) showed the absence of an enol form of the substrate in a reagent/catalyst mixture. The presence of hydrogen bonding was suggested by a strong downfield shift of the signals caused by H1 (δ = 0.5 ppm) and by H7 (δ = 0.7 ppm) of the catalyst. The signals due to the aromatic protons of both the catalyst and the enone were also significantly shifted, indicating a possible interaction between the corresponding functional groups. When the ketoamide was mixed with the catalyst only, fast deprotonation was observed. From a kinetic analysis, among other results, it was found that the reaction was zero-order for the β-ketoamide. A significant primary KIE (KH/D = 1.15) was also observed in a reaction with >80% deuterated β–ketoamide, consistent with rapid deprotonation leading to the active form.

Computational studies were performed with the hybrid meta-GGA DFT functional M06-2X,34 with the 6-31G(d) basis set. Relative Gibbs free energies were determined for the TSs along the reaction coordinate. Based on the results, as well as on the experimental observations, a detailed catalytic cycle was postulated (Figure 22). It seems that amide hydrogen abstraction by the organocatalyst takes place initially, leading reversibly to an off-cycle unreactive form. An equilibrium back to the starting β-ketoamide and subsequent enolate methine hydrogen abstraction would be the preferred path. Subsequent coordination of the enone to the amide hydrogen leads to the direct formation of the most stable enolate intermediate. Nucleophilic addition takes place next, a thermodynamically unfavored process because the resulting intermediate lies higher in energy than the ternary complex deprotonated β-ketoamide-protonated catalyst-enone (ΔΔG = 3.9 kcal mol⁻¹). Enantiodifferentiation takes place in this step.

In the most favored transition states, the β-ketoamide takes up a perpendicular orientation concerning the thiourea. The TS leading to the R adducts was found to be lower in energy by 2.9 kcal mol⁻¹ over the TS leading to the S adducts (e.g., 28a), in agreement with the experimentally observed 99% ee value. The ease of proton transfer from the amide to the enolate, and the high stability of the resulting intermediate, are probably responsible for the equilibrium shift in this thermodynamically unfavorable C–C bond-forming step. This probably explains why the acidity of the amide proton is crucial for success. Furthermore, of interest is the fact that the deprotonated amide–ammonium complex was found to be thermodynamically more stable than the separated products (ΔΔG = 4.6 kcal mol⁻¹), in complete agreement with the experimentally observed inhibition of the catalyst by the product.

An enantioselective organocatalytic conjugate addition of α,β-unsaturated ketones 29, promoted by a bifunctional organocatalyst, cinchona-squaramide C17, was described by Fernández-Salas, Alemán and coworkers in 2019 (Figure 23) [52]. The attractiveness of this method lies in its environmentally friendly nature since employing TMSN₃ (30) as a nucleophile avoids the use or preformation of hydrazoic acid, a dangerous and highly explosive substance. No acidic additives were required either. The formation of a chiral carbon at the carbon C–N bond occurs in a highly enantioselective manner, and the product β-azido ketones 31 were obtained with good yields. It was found that the presence of water was important, causing a significant enantioselectivity enhancement (from 53 to 93% ee), and an initial controlled amount was used to ensure reproducibility issues. The best results were obtained in the presence of 0.5 equivalent of water.

The reaction mechanism was studied by ¹H and ²⁹Si NMR spectroscopy and DFT calculations. The role of water was studied through analysis of the reaction side-products. It was found that after a catalytic run, even if water was not added to the system, two new silylated species had been produced, trimethyl silanol (δ 15.82 ppm (²⁹Si-NMR)) and its corresponding siloxane (δ 6.57 ppm), as observed using ²D-HMBC{¹H,²⁹Si}. The addition of excess water accelerated the reaction but deteriorated the ee. These results appear to suggest that activation of TMSN₃ proceeds via hydrolysis, although, in the absence of catalyst, water could not activate TMSN₃. DFT calculations showed that proton transfer is involved in the rate-determining step (in good agreement with experimental kinetic data). In control experiments, it was observed that TMSN₃ hydrolyzes in hexafluorobenzene in the presence of an acid additive but not in the presence of water alone. Therefore, no NH₃ is formed in situ during the reaction, but it appears that there is proton transfer from the water molecule to the nitrogen atom of the quinuclidine unit followed by a nucleophilic attack of the resulting hydroxide to the silicon atom and N–Si bond cleavage. The theoretical results suggested that these processes take place in an asynchronous concerted manner.

Organocatalyzed sulfa-Michael reactions have been known since 1977, when Winberg and coworkers showed that aromatic thiols could add to cycloalkenones in an enantioselective manner in the presence of cinchona alkaloids [53]. These findings represent a milestone in organocatalysis, and since then, many cinchona alkaloid derivatives have been developed to promote a wide range of asymmetric reactions [54,55]. Wynberg proposed a dual-mode of activation in which the thiol, after deprotonation by the quinuclidine nitrogen of the organocatalyst, would remain coordinated to the tertiary amine center, while the enone hydrogen, bonded to the hydroxyl group, was placed in a good orientation for the reaction to proceed in an enantioselective manner. Catalysis by cinchona alkaloid-squaramide catalysts was introduced by Chen and coworkers in 2010 [56]. A theoretical study on this reaction was performed by Guo and Wong recently to investigate the mode of bifunctional activation and the origin of stereoselectivity [57]. The authors aimed to disclose guidelines that could be used for the rational development of more efficient bifunctional catalytic systems based on hydrogen-bonding activation. Geometry optimizations of equilibrium and transition state structures were performed using the M06-2X density functional with the standard 6-31G(d) basis set. In a similar manner to catalysis by bifunctional thiourea-based organocatalysts (Figure 4b), it is also possible to distinguish a few modes of activation with squaramide-based organocatalysts. For the formation of the C–S bond in the sulfa-Michael reaction between 21 and 22, four distinct modes of activation (A–D) are possible (Figure 24). Mode A is Takemoto’s mode of bifunctional activation also observed for bifunctional thiourea-based catalysts; in a similar manner, mode B is Pápai’s; mode C was postulated in a joint experimental/theoretical study on a bifunctional cinchona alkaloid-thiourea catalyst by Wang and coworkers, although they also considered it for squaramide-catalyzed reactions [24]; and finally, in mode D, the electrophile interacts with the alkylammonium and one of the N−H groups of the Brønsted acid, while the nucleophile does so with the distal N−H of the acid. All these modes of activation were investigated by Guo and Wong as possibilities for activation of the sulfa-Michael addition reaction between benzyl thiol 32 and trans-chalcone 21, leading to 33. It has been shown that, since the initial protonation of the catalyst, as well as its final deprotonation in the catalytic cycle, is facile, the C–S bond-forming step is the rate-determining step. The conformations of the protonated catalyst were investigated first. The two N−H groups of the squaramide have a syn alignment in the active catalytic conformation. The vinyl group on the quinuclidine moiety has been shown to be anti concerning C2−C3 bond in a different study, and it was adopted as such in this study [58]. Since there is free rotation around the C8−C9 and C4′-C9 bonds, six main conformations are possible, anti-open, syn-open, anti-closed, syn-closed, plus two hindered conformations, which were ignored since cooperativity between the three N−H functionalities of the protonated catalyst is not possible in those cases, and therefore, they are not catalytically active. Previous studies also showed that the protonated catalyst CatH⁺ favors the open conformations by at least 16 kJ/mol over the closed ones, and therefore, closed conformations were also not examined in this study. The energies of these conformations were calculated, and the lowest energy one was found to have the anti-open form. The next lowest conformation is the syn-open conformer, which is less stable by 7.0 kJ mol⁻¹. These results agree with a report by Sharpless of epiquinine in solution, for which the anti-open form was the only conformer found to be present [59].

In this conformation, the squaramide ring is coplanar to the 3,5-bis(trifluoromethyl)phenyl group, and this orientation is stabilized by an attractive C−H···O interaction between one carbonyl oxygen of the squaramide moiety and the acidic ortho-proton of the aromatic ring. The 6′-methoxy group of the quinoline moiety is also coplanar to the quinoline ring.

Transition states for the C–S bond-forming step were calculated for each activation mode, with the anti-conformation of the protonated catalysts (Figure 25). The lowest energy TS was found to be in mode B, TS-B-R, which gave rise to the major (R)-configured product, in agreement with the experimental findings. Both TS’s of mode A (TS-A-R and TS-A-S) are less stable by more than 13 kJ mol⁻¹. The TSs with syn-open arrangements were much higher in energy (by at least 26 kJ mol⁻¹), and those related to mode C were even higher in energy. Hence, they were all disregarded. TS’s for mode D could not be located. To determine the origin of stereo-induction, a distortion−interaction analysis was carried out. The factors that make mode B the preferred conformation are a greater stabilization of the developing alkoxide by the quinuclidinium ion rather than hydrogen-bonding interactions. The excellent stereo-selectivity observed with mode B stems from the more favorable cooperative noncovalent interactions: hydrogen bond, π-stacking, C−H···π interaction, and C−H···F interactions, between the catalyst and the substrates in the major transition state. NCI plots show that the 3,5-bis(trifluoromethyl)phenyl moiety is capable of forming multiple attractive interactions with the substrates to help in differentiating the stereo-controlling transition states. These results are in agreement with Houk’s Brønsted acid-hydrogen bonding model for the sulfa-Michael addition reaction with enones, catalyzed by cinchona alkaloids, in which the enone is activated by the alkylammonium ion, while the thiolate electrophile coordinates with the catalyst’s hydroxyl group [60].

Grayson and Houk recently published DFT studies on the asymmetric conjugate addition of aromatic thiols, e.g., 35 to cycloalkenones (34) catalyzed by cinchona alkaloid-derived ureas C20/C21, a reaction first reported by Singh and coworkers and reported in 2010 [48,61]. This was the first detailed mechanistic study of a cinchona urea-catalyzed reaction [61]. In recent reports on other reactions catalyzed by cinchona-ureas, it was generally found that when comparison data were available, cinchona urea catalysts yielded similar or better levels of enantioselectivity than their thiourea counterparts [48,62], which prompted this study. Since urea is a much weaker acid than thiourea (pK_a = 26.9 and 21.1, respectively, in DMSO), it may happen that cinchona urea and thiourea catalysts operate by different reaction mechanisms.

In these studies, the rate-determining C−S bond-forming step was investigated to determine the preferred mode of catalyst activation and to explain the observed selectivity. The activation modes considered were the same as those described above for thiourea catalyzed reaction, modes A, B and C (Figure 4b) [17,19,24]. The TSs were located at the M06-2X/def2-TZVPP-IEFPCM(toluene)//M06-2X/6-31G(d)-IEFPCM(toluene) level of theory using Gaussian 09, and a methyl group was used instead of the vinyl group on the quinuclidine ring to simplify the calculations. Of all the major conformations possible for the protonated catalyst, the anti-open conformation was found to have the lowest energy, and it was considered for further study (Figure 26). A total of 78 unique pre-reaction complexes were located [48]. In complex 113, found to have the lowest energy, the thiolate is closely associated with the alkylammonium ions, and there are hydrogen-bonding interactions from the urea group to the thiolate. The alkylammonium ion interacts with the enone too. It was found in this study that the transition states of lowest energy, giving rise to the major and minor products observed experimentally, were obtained via mode B, i.e., TS1 for the major product, Figure 26.

From 37, the free energy barrier to TS-B(major) is 10.9 kcal mol⁻¹. The nucleophilic attack takes place axially on the half-chair cyclohexenone, in preference to equatorial attack by 3.3 kcal mol⁻¹. TS-B(minor), which leads to the minor product via mode B, is destabilized relative to TS-B(major) by 3.9 kcal mol⁻¹. No obvious steric interactions were found to contribute to this difference in energy. However, the geometry adopted by enone 34 in TS-B(minor) results in a longer and less directional interaction from the quinuclidinium ion to the enone oxygen relative to TS-B(major) (N−H···O distance = 1.59 and 1.65 Å and N−H···O angle = 151° and 143° in TS-B(major) and TSB(minor), respectively). It was also found that in TS-B(major), there is a staggered conformation about the developing C−S bond, whereas, in TS-B(minor), there is an eclipsed conformation (CSCH dihedral = 69° and 6°, respectively). In TS-B(major) there is only one NH···S interaction is present, but the other urea NH hydrogen interacts with the π system of the thiolate instead. It was also found in this study that the quinoline ring methoxy group has a strong effect on the TS energy, an effect, which has not been considered in previous mechanistic studies of reactions catalyzed by cinchona alkaloids.

In 2020 Vetticatt, Hirschi, Seidel and coworkers described a new type of bifunctional catalyst, a selenourea-thiourea Brønsted acid catalyst (C22), which catalyzed the conjugate addition of cyclic amines 39 to inactivated α,β-unsaturated esters 38 [63] (Figure 27 and Figure 28). β-amino esters 40 were obtained with high ees. DFT calculations and ¹³C kinetic isotope effect studies helped to throw some light on the reaction mechanism. The catalyst could also be successfully used for the kinetic resolution of cyclic amines.

¹³C kinetic isotope effects (KIEs) were determined for benzyl crotonate using Singleton’s ¹³C NMR methodology for starting materials at natural abundance. Benzyl crotonate and piperidine were reacted and taken as 75 ± 2% and 79 ± 2% conversion concerning the ester. Unreacted benzyl crotonate was reisolated, and the ¹³C isotopic composition was compared that in samples of crotonate not subjected to the reaction conditions. From the changes in relative isotopic composition and the fractional conversion, ¹³C KIEs were determined (in duplicate) in the standard way. The values obtained are shown in Figure 28. For confirmation, ¹³C values of KIEs were also predicted from the scaled vibrational frequencies using the program ISOEFF98 for both R-TSC−N and R-TSC--prot, then applying a Wigner tunneling correction to all values. The most interesting results are the unity KIE observed at the β-carbon and the normal KIE of ∼1.5% on the α-carbon. If C−N bond formation was the first irreversible step in the catalytic cycle, a normal KIE was expected on the β-carbon, not unity, which implies reversible C−N bond formation. A normal KIE on the α-carbon (∼1.5%) suggests that α-protonation is likely to be the first irreversible step in the catalytic cycle. The calculated values are in agreement with these postulates. Therefore, C–N bond formation is a reversible event, and the rate- and enantio-determining step occurs after. These results suggested that the origin of enantioselectivity would be better understood if the transition state analysis were made for the α-carbon protonation step. Since there is no external Brønsted acid, the most acidic proton would be one of the thiourea NHs, which would then be involved in α-protonation. Two situations are possible: an α-carbon−catalyst-mediated tautomerization or catalyst-mediated direct C-protonation. An investigation using DFT calculations showed that catalyst-mediated tautomerization of the enol is prohibitively high in energy (ΔG⧧ = 48.6 kcal mol⁻¹), whereas catalyst-mediated C-protonation is energetically accessible (ΔG⧧ = 18.2 kcal mol⁻¹) (Figure 29). In the lowest energy TS leading to the R-enantiomer, it was found that there is strong intramolecular H-bonding between the protonated piperidine and the enolate oxygen (1.81 Å). One thiourea NH is loosely bound (2.35 Å), while the other (more acidic) is transferred to the ester α-carbon. Selenium forms a weak nonconventional CH···Se interaction (2.76 Å) with one of the α-CHs of the piperidine moiety. There have been a few scanty reports on the involvement of a thiourea catalyst as a Brønsted acid in mechanisms, and this related behavior of Se represents a new mode of catalysis, which could be used as a guide in future developments.

2.3. Cascade Reactions Involving Conjugate Addition

In 2014 Carrillo, Vicario and coworkers reported a stereoselective and diastereo-divergent procedure to synthesize functionalized cyclohexanes containing four stereocenters through an asymmetric Michael-initiated ring closure (MIRC) cascade reaction [64]. Two different (R,R)-configured bifunctional squaramide-based organocatalysts (C23 and C24) made possible the conversion of the same starting materials into products with different absolute configurations. This was possible by the introduction of subtle structural differences in the two catalysts. A wide range of nitroalkenes 7 and α-nitro-δ-oxo esters 41 reacted in a highly diastereo- and enantioselective Michael/Henry cascade giving high yields of products 42 (Figure 30). A gram-scale synthesis was also possible without loss of stereoselectivity.

Kinetic experiments were performed to find the rate-determining step, and DFT calculations were performed at the B3LYP/6-31G(d) level of theory using the PCM model to explain the reaction process and the origin of stereo-divergence. The initial Michael reaction, in which the initial two stereocenters are formed, was considered as the stereo-defining step. It is in this step that the diastereo-divergency of the reaction occurs when the two different C-1 epimers are formed with each catalyst. It was observed that the reaction was first-order for all reagents: both catalysts, nitronate 41a (117, R¹ = Et, R² = Me), and nitrostyrene. The fact that the rate of the reaction was strongly dependent on the electronic nature of the aryl substituent of the nitroalkene also suggested that the initial Michael reaction is the rate-limiting step. Among others, an experiment was performed with a preprepared Michael intermediate, which, when stirred with either catalyst, provided the same product in comparable yields. This is a strong indication that although the configuration of the two stereocenters generated during the initial Michael reaction step is controlled by the catalyst, the diastereoselectivity of the intramolecular Henry reaction is strictly governed by the substrate. Hence the DFT analysis was performed for each catalyst on a simplified Michael reaction between 44 and 7b. The optimized transition state structures (TS1 and TS2) found by FDT analysis leading to the major isomer formed by each catalyst are shown in Figure 31.

From the DFT experiments, it could be concluded that the mode of activation of the catalysts involves activation of the nucleophile by the squaramide unit through multiple H-bonding interactions, while the nitroalkene stays bound to the ammonium salt (formed after the initial deprotonation of the pronucleophile) via a single H-bond. The different simple diastereo-selections provided by each catalyst depend on their ability to form two different H-bonded complexes with the nitroacetate enolate, with the nitronate moiety exposing a different reactive prochiral face depending on the catalyst used.

Meninno, Lattanzi and coworkers reported the first enantioselective catalytic approach to cis- and trans-2,3-diaryl-substituted 1,5-benzothiazepines in 2018 [65]. The one-pot two-step sulfa-Michael/lactamization sequence between compounds 47 and 48, promoted by thiourea-based bifunctional catalyst C25 and p-toluenesulfonic acid, afforded products 49 in good to high yields and very high ees (Figure 32). Mechanistic studies were performed on the reaction. An uncatalyzed reaction monitored by ¹H NMR in toluene-d₈ over time revealed that small amounts of cis-Michael adduct 50, as well as cyclized product 49a (R¹ = R² = Ph, R³ = H), are formed during the reaction, which contributes to lower the ee of the cis-benzothiazepine. A similar experiment in the presence of the catalyst showed the formation of an increasing amount of Michael adduct over time, but not a cyclic product. It was found that the organocatalyst promoted a competitive 1,2-addition/elimination reaction, confirmed by detecting 30% of free pyrazole at the end of the reaction. The mixture of cis/trans-49a was recovered with the same level of diastereo- and enantioselectivity of the starting adduct 50, which suggests that a retro-sulfa Michael/sulfa-Michael process does not happen since that would lead to epimerization/racemization. A racemic mixture of adduct treated with the catalyst afforded <10% of the trans-benzothiazepine after two days, showing that the lactamization reaction is negligible within the overall reaction times for the method developed. Overall, although the uncatalyzed Michael reaction, lactamization, as well as catalyst promoted lactamization, may influence the diastereo- and enantioselectivities observed at the end, they were nevertheless predicted to be poorly effective during the overall reaction times taken by both sequential steps.

Recently, Liu, Li, Song, Ban and coworkers described a thiosquaramide (C26)-catalyzed cascade Michael−Henry elimination reaction for synthesizing cyclopentenes [66]. In this reaction, phenacylmalononitriles 51 and nitroolefins, 7 reacted with very high diastereoselectivity, giving high yields of products 52 with moderate to high ees, which could nevertheless be as high as 98% (Figure 33). Previous approaches to cyclopentenes from the same substrates were reported previously, but in the racemic series, the latest one using TBAF as catalyst [67]. The products contain a chiral quaternary center as well as an internal double bond, with many possibilities for derivatization. Based on experimental data and preliminary theoretical analysis (Hartree−Fock calculations), a mechanism was proposed involving an organocatalyzed asymmetric Michael addition (with TS1) and catalyst-assisted E2 elimination (with TS3) (Figure 34). Although an alternative path exists via non-catalyzed E₁ elimination (path A), this route would give rise to two isomers, the product observed, 52a and E. However, E was not observed. Therefore, it was concluded that the reaction proceeds via the catalyzed E₂ elimination path.

3. Cycloaddition Reactions

Another class of reactions that have been shown to benefit from developing hydrogen-bonding catalysis with bifunctional catalysts is cycloaddition reactions. [4 + 2] Cycloadditions as well as [3 + 2], i.e., formal [3 + 2], formal [3 + 3], [5 + 2], 1,3-dipolar cycloadditions and Tamura cycloadditions have been successfully achieved. This topic was reviewed by Held and Tsogoeva in 2016 [68]. Some interesting kinetic and theoretical studies have been reported recently. Wang, Wang and coworkers described a formal thio [3 + 3]-cyclization catalyzed by a thiourea-based bifunctional catalyst (C27) in 2014 [69]. Starting from 2-benzylidenemalononitrile (54) and indoline-2-thiones 55, optically active thiopyrano-indole annulated heterocyclic compounds 56 were obtained in high yields and excellent ees. The authors also attempted to react the less acidic indolin-2-ones 57 with 2-benzylidenemalononitrile in a similar way, but these attempts were unsuccessful, and no reaction was observed. To explain the differences in reactivity between the two indoline substrates, DFT calculations were performed at the B3LYP/6-311++G(d,p) level using the CPCM solvent model. Based on previous transition-state models for thiourea-catalyzed reactions, it was assumed that both reactants are activated simultaneously by the catalyst, as shown in Figure 35. According to the theoretical results obtained, enolization of the indolin-2-one (57) is much more difficult than that of indoline-2-thione (55a). Catalyst protonation by the enol forms is facile according to the Pápai and Wong models. The enantioselectivity obtained is dependent on the way the substrates coordinate to the catalyst. Complex S-M1 (3.75 kcal mol⁻¹) is more stable than O-M1 (8.80 kcal mol⁻¹), in agreement with the experimental observations that 55 is much more reactive than 57. The energy barrier for C–C bond formation is 1.34 kcal mol⁻¹ lower for the indoline-2-thione, which also explains why the thiopyrano-indoles are easier to obtain.

In 2017 Jacobsen and coworkers introduced a new concept in H-bond donor catalysis [70]. To overcome the fact that small-molecule dual hydrogen-bond donors, such as ureas, thioureas, squaramides, and guanidinium ions are only weakly acidic and, therefore, favor reactions with highly reactive electrophilic substrates, they developed a way to activate them through by associating them with Lewis acids. Silyl triflates (58) were found to be ideal for this purpose since they are readily available, have broad applicability in organic synthesis, and when they coordinate with weakly coordinating anionic ligands, e.g., disulfonimides, their reactivity is increased [71,72]. Association with a chiral H-bond donor catalyst, as in 59, can give rise to a charge-separated complex, by anion abstraction, with higher Lewis acidity than that of the silyl triflate alone (Figure 36a).

One of the systems utilized to investigate the H-bond donor–silyl triflate cooperativity was a [4 + 3] cycloaddition reaction, which gives rise to functionalized seven-membered carbocyclic frameworks, a reaction which had proved previously to be challenging to perform in an enantioselective manner. Silyl enol ethers capable of acting as oxyallyl cation precursors (60) were reacted with furans (61) to generate bicyclic [4 + 3] cycloadducts (62) in good yield and high ees, as single diastereomers in some cases in the presence of C28a (Figure 36b).

Kinetics and theoretical studies were performed on this reaction. For the kinetics studies, Burés’ method was utilized [73]. A first-order kinetic dependence on the acetal, zero-order dependence on the furan and a first-order dependence on the catalyst were observed, as well as saturation kinetics concerning triethylsilyl trifluoromethanesulfonate (TESOTf). The data are consistent with a pre-equilibrium formation of a resting-state complex between the catalyst and TESOTf, and a rate-limiting reaction of this complex with the acetal. It was found that there was a 1:1 binding interaction between the simpler squaramide C28b and different triflate sources, and the bonding constants were determined from titration experiments by ¹H NMR spectroscopy. TESOTf was found to bind 4000 times as tightly as NBu₄OTf, which suggests that there is the simultaneous binding of both the triflate and the trialkyl silyl component in the complex formed. Taking into account the combined results, it appears that there is a formation of an unexpectedly strong 1:1 complex between TESOTf and 5 as the resting state of the catalyst in the reactions described.

The kinetics and DFT studies suggest that in this resting state, the complex acts as a potent Lewis acid that promotes acetal ionization. Post-rate-determining reaction of the oxyallyl cation intermediate with the furan gives rise to the [4 + 3] cycloadduct. Since similar ees were obtained with different trialkyl silyl triflate promoters tried, it appears that the enantioselectivity-determining step occurs after the formation of the oxyallyl cation and involves the reaction with furan. Single-point calculations at the M062X/6-31+G(d,p) level of theory could reproduce both the sense and magnitude (ΔΔE^‡_calc = 1.28 kcal mol⁻¹) of enantioinduction determined experimentally. The structure corresponding to the lowest-energy TS (TS1) for the first, enantioselectivity-determining C–C bond-forming step, towards the major enantiomer of product, is shown in Figure 36. Besides a network of hydrogen-bonding interactions between the different reacting components, there seems to be a stabilizing interaction between the furan and the catalyst in the TS leading to the major enantiomer, absent in the TS leading to the minor enantiomer, which can be deduced from the position of the furan ring close to the aromatic substituent of the catalyst. This may be a key factor responsible for enantioselectivity.

When homophthalic anhydrides 63 are reacted with imines 64, chiral lactams 65 can be obtained in a highly enantioselective manner in the presence of bifunctional squaramide–based catalysts (Figure 37). This was discovered by Vetticatt and Seidel, who proposed a novel ion-pairing mechanism to achieve a formal [4 + 2] cycloaddition [74]. Although this reaction had been previously used in synthesis and even been applied to alkaloid synthesis, no asymmetric versions had yet been developed [75,76]. The new mode of activation, an anion-binding/ion-pairing approach, was based on the concept that the hydrogen-bonding catalyst, which itself remains neutral throughout the reaction, can convey enantioselectivity by interacting with an anionic nucleophile and a cationic electrophile at the same time. The interaction of the hydrogen-bonding thiourea with the anhydride increases the acidity of the substrate, facilitating ion-pair formation. The result is an increased equilibrium concentration of the ion pair intermediate. The iminium ion generated can then interact with another binding site on the catalyst, which results in creating a well-defined ion pair, which then undergo the enantioselective Mannich reaction. Catalyst C29 was found to be the best out of a range tried. The reaction mechanism was investigated, and it was observed that there were no nonlinear effects, which suggested that the rate-limiting step involves only one catalyst unit. The nature of the transition state of the rate- and stereo-determining step, the Mannich addition for the reaction between 63 and N-phenyl benzimine catalyzed by C29 was investigated using B3LYP-GD3[22]/6-311+G** PCM [23] (diethyl ether)//B3LYP/6-31G* calculations as implemented by Gaussian 09. The lowest-energy transition structures found leading to the major and minor enantiomers of the product are shown in Figure 37.

The results show an early C–C bond formation step. In the TS structure leading to the major enantiomer (SS-TS), the enolate is bound to the catalyst through two strong H-bonds via the thiourea NH groups (Figure 37). The protonated imine is directed to the Re face of the enolate through a strong H-bond with the carbonyl oxygen of the amide moiety of the catalyst. Similar H-bonding interactions exist in TS leading to the minor enantiomer (RR-TS), but this TS is higher in energy by 2.0 kcal mol⁻¹, which agrees with the experimentally observed value of ee.

Thiourea-based bifunctional catalysis was also responsible for an enantioselective 1,3-dipolar cycloaddition of salicylaldehyde (66)-derived azomethine ylides with nitroalkenes 67 that proceeded with excellent exo/endo selectivity [77]. An intramolecular hydrogen bond involving an o-hydroxy group allowed the reaction with azomethine ylides bearing only one activating group. In methods previously developed, two activating groups (two EWGs or one EWG and one aryl group) were required at the dipole to increase the acidity of α-protons. As a result, the substitution pattern of the final products was restricted.

In this work, the activation of the azomethine ylide by intramolecular bonding, as well as the intermolecular hydrogen bonding with the catalyst, was demonstrated by DFT calculations to be crucial for the reaction. For these studies, β-nitrostyrene, Takemoto’s catalyst C1 and azomethine ylide 69, formed by deprotonation of 66 by the catalyst, were considered as the reacting species. Three different coordination points of the reactants with the catalyst with both endo and exo approaches were considered, according to the Takemoto, Pápai and Zhong’s models discussed above. The pre-association complex, the intermediates, and the catalyst-bound products were computed, as well as the corresponding TSs. The first TS, TS1, gives rise to the formation of the C–C bond, whereas the second, TS2, to the intramolecular attack of the imine leading to ring closure. Calculations for the Zhong´s model (Figure 38) showed that the energy first barrier (TS1 = 3.35 kcal mol⁻¹) was higher than in the case of Pápai’s model (TS1 = 1.84 kcal/mol), but the second lower (TS2 = 14.48 and 16.47 kcal mol⁻¹, respectively). The intermediate and the catalyst-bound product, according to Zhong´s model, were thermodynamically more stable. Altogether the results suggest that Zhong’s mechanism is in operation. An intermolecular hydrogen bond between the thiourea and the OH group of ylide is observed only in Zhong’s model. While it provides additional stabilization to the PAC, it also increases the reactivity of the azomethine ylide, and it is one of the factors that favor Zhong’s approach to the others.

Cinchona thiourea catalysts have also been used to induce axial chirality. In a recent work described by Yan and coworkers, an atroposelective intramolecular [4 + 2] cycloaddition was used to convert in situ generated vinylidene ortho-quinone methides (VQM), formed by the intramolecular reaction of 2-ethynylphenol derivatives with alkynes, into axially chiral heterobiaryls, i.e., naphthalenylpyran 71 [78]. Several products could be obtained in very high yields and ees (96–>99% ee) at rt. Subsequently, Yan, Bai, Lan and coworkers performed theoretical studies, i.e., DFT calculations, on the reaction shown in Figure 39 to investigate the mechanism and origin of enantioselectivity [79]. All DFT calculations were performed with the Gaussian 09 series of programs, and the B3-LYP functional with the standard 6-31G(d) basis set was used for geometry optimization. The results obtained led to the proposal of the mechanism shown in Figure 39. The reaction starts with deprotonation of the naphthol moiety by the quinuclidine, facilitated by the formation of two hydrogen bonds with the thiourea moiety, which enhances the acidity of the alkynyl naphthol. Conjugation with the naphthalenoate enhances the basicity of position C2 in the alkyne moiety, facilitating protonation. Intramolecular proton transfer of ammonium naphthalenolate salt B gives rise to the VQM intermediate C-R with dearomatization. This process is endergonic by 15.3 kcal mol⁻¹. The conjugative allene and carbonyl moieties act as an enophile in the ensuing [4 + 2] hetero-Diels–Alder cycloaddition. The axial chirality of the VQM intermediate is created in the protonation step and the enantioselectivity of the reaction. The enantioselectivity is controlled by steric repulsions with the cinchonine framework, leading to an R-axial chiral VQM as the major intermediate. The enantioselectivity for the axial chirality of the product is controlled in the cycloaddition step. An additional hydrogen bond between the naphthalenol and quinuclidine moieties makes the generation of an S-axial chiral naphthopyran the most favorable (Figure 40). The calculated energy barrier for this step is only 6.8 kcal mol⁻¹, whereas the energy barrier for the formation of its enantiomer was calculated to be 13.1 kcal mol⁻¹. The difference in energy (6.3 kcal mol⁻¹) predicts >99% ee for the S-axial enantiomer, which corresponds with the experimental observations.

The Tamura reaction, the cycloaddition between activated alkynes or alkenes and enolizable anhydrides, such as homophthalic anhydride, was studied by Connon and coworkers. They reported the first examples of asymmetric Tamura cycloaddition reactions involving singly activated alkenes in 2016 [80]. Homophthalic anhydride (63) was reacted with α-methyl nitrostyrenes (72) in the presence of squaramide-based catalyst C30 to produce bicyclic aromatic ketones 73/74 with three new stereocenters in 12–99% ee. It was found that the products were susceptible to equilibration, 74 equilibrated to 73, which was observed even during evaporation of the solvent during purification. To obtain products under reproducible conditions, the reaction conditions described in the equation in Figure 41 had to be followed carefully, and also the purification, with a concentration of the solvent performed at low temperatures (not higher than 30 °C). It was also observed with 73a (73, R = Ph) that when the pure diastereoisomer was subjected to epimerization conditions in the presence of CD₃OD, i.e., heating to 55 °C for 55 h in MTBE (0.1 M), that there was 60% incorporation of deuterium at the ester’s α position. It was also found that 73a epimerizes faster than 74a and that 74a is the more stable isomer in the presence of MeOH. This experiment showed that even the more stable diastereoisomer undergoes proton transfer. This unusual equilibration, which occurs in a methanolic solution in the absence of a recognizable base via proton transfer at the α-carbon of the ester, was subsequently investigated computationally. These calculations, as well as XRD data, indicated that the nitro group and the proton involved in the epimerization are ca. 2.74 Å apart, and hence it was proposed that proton transfer is mediated by the proximal nitro functional group. However, such mechanism is thermodynamically unfavorable (relative energy >50 kcal mol⁻¹ with respect 74a), and another mechanism was considered solvent-assisted proton transfer. The involvement of two MeOH molecules was found to be kinetically more favorable than one only. It was, therefore, proposed that two molecules of MeOH interact with the product; subsequently, a shuttle (concerted) deprotonation/protonation takes place with the participation of both MeOH molecules bound to both each other and to the substrate, leading to intermediate 75. The energy barrier for this process was calculated to be 18.7 kcal mol⁻¹. The corresponding reaction involving the other diastereoisomer (74) has a slightly higher energy barrier (21.5 kcal mol⁻¹). 74·2MeOH was found to be 1.6 kcal mol⁻¹ more stable than 73a·2MeOH, consistent with the findings from deuteration experiments, that 73a epimerizes faster than 74a, and that 74a is the more stable diastereoisomer.

The Connon group also showed that a wide range of aldehydes undergoes cycloaddition with homophthalic anhydride (63) in the presence of organocatalyst C31 to yield substituted lactones 77 in excellent yield and ees (Figure 42) [81]. In 2016 Rozas, Connon and coworkers reported the first DFT mechanistic study on the cycloaddition of benzaldehyde (76a) with homophthalic anhydride [82]. The reaction was found to follow a mechanism in which the catalyst binds to the anhydride, a highly energetically favorable process (ca. 40 kJ mol⁻¹), which then deprotonates it, leading to a squaramide-bound enolate. This reaction is thermodynamically favorable, with the catalyst-bound enolate found to be 21.1 kJ mol⁻¹ more stable than the substrate–catalyst complex. This reaction is followed by C–C bond formation. An atoms in molecules (QTAIM) analysis revealed the existence of a web of HB interactions between enolate oxygen atoms not involved in squaramide binding and both the catalyst’s quinoline ring and the C-2 phenyl substituent, shown in previous studies to be beneficial for product dr and ee. Therefore it seems that this unit facilitates face-selective binding of the enolate, which then adds to the aldehyde, which is at the same time being activated by the catalyst’s ammonium ion. In the C–C bond-forming step, in which the stereocenter is established, the binding interaction between C31 and the anhydride is mostly unaffected by the aldehyde. This implies that as a result of the location of the ammonium ion, only one enolate face is available to attack the aldehyde, which leads to the high enantioselectivity observed. It also appeared from the results that the face-selective addition to the aldehyde was primarily directed by the HB to the ammonium ion. A computational analysis of all available pathways revealed that the pathway leading to trans-(R,R) was clearly favored. C–C bond formation is followed by acylation to give the lactone in a concerted and general-base/hydrogen bonding catalyzed step. The web of attractive interactions between the catalyst’s quinoline/quinuclidine ring and the anhydride O-atoms was found to be absent in the TSs/intermediates leading to minor stereoisomers. The calculated ee value agreed well with the experimental results.

4. Aldol and Henry Reactions

In 2017 an organocatalytic decarboxylative aldol reaction of β-ketoacids with α-ketophosphonates allowed the enantioselective synthesis of tertiary α-hydroxyphosphonates was described by Chowdhury and coworkers for the first time (Figure 43) [83]. α-Hydroxyphosphonates have been shown to exhibit many kinds of useful biological activities, including in reports by one of us, and they are the object of much research [84,85,86]. Recent accounts on the synthesis of tertiary α-hydroxyphosphonates include cross aldol reactions at low temperatures, but long reaction times are required. Up to this report from the Chowdhury group, cross aldol reactions of α-ketophosphonates (79) with acetone or acetaldehyde were reported, but reactions with aromatic ketones, e.g., acetophenones, not. In this report, β-aryl β-ketoacids 78 were employed as surrogates of the acetophenone enolate to fill this gap. In this quinidine-derived urea (C32)-catalyzed process, the products 80 were obtained in high yields and ees.

Control experiments were performed to investigate the mechanism. The reaction was thought to proceed via the formation of an intermediate of type 81, which can lose CO₂ to yield the desired product. ³¹P NMR studies of a reaction between α-ketophosphonate 79a, benzoylacetic acid 78a and the catalyst showed that, soon after 78a and the catalyst are added to a solution with 79a, a few signals appear in the spectrum, due to the product, to two diastereoisomers of intermediate 81 and also to dimethylphosphite (82), due to the decomposition of 79a. An additional signal at 22.54 ppm, possibly due to a complex formed between 79a and the catalyst C32, was also observed. After some time, the intensity of the signal due to product 80a increased, while all those of the signals at δ 25.46 ppm and 25.1 ppm decreased, which is consistent with the mechanism proposed.

The Henry or nitroaldol reaction, mechanistically similar to the aldol reaction, is another important tool for carbon–carbon bond formation. Several studies have shown that cinchona alkaloids bearing a hydrogen bonding group at the C-6′ position can catalyze the Henry reaction in high yields and ees [87]. The first theoretical studies aiming to clarify the mode of action of thiourea-cinchona alkaloid bifunctional catalysts (e.g., C33) were reported by Himo and coworkers in 2007 [88]. In the reaction between nitromethane and benzaldehyde (76a), with simplified catalyst C34, the results supported a mechanism involving initial nitromethane deprotonation and aldehyde complexation (Figure 44a). Two pathways with comparable energy barriers are then possible for C–C bond formation. They differ in the way the reactants bind to the catalyst. Either nitromethide binds both to the quinuclidinium and one thiourea NH, while the aldehyde binds to one thiourea NH only (path A), or nitromethide binds to both thiourea NH hydrogens and benzaldehyde to the quinuclidinium unit (path B). A small preference for path A was found to exist. Experiments have shown that the results are strongly dependent on the nature of the solvent and not simply a polar, nonpolar distinction, often observed in some organocatalyzed reactions. This type of correlation is not possible with the Henry reaction. Ees correlate roughly with the Lewis basicity of the solvent, so that the higher the Lewis basicity, the higher the ee. This suggests that the energies leading to both TSs are similar. In this study, TSs states leading to both enantiomers were located for the two pathways (Figure 44b); for path A, three TS rotamers were found for each enantiomer, corresponding to the different Newman projections for C–C bond formation, as shown in Figure 44b. For path B, since there is bidentate coordination between both NH groups of the thiourea and the nitro group, only one TS is possible for the formation of each enantiomer (R and S). It was found that the lowest energy TS (TS-S) corresponds to the S-product as observed experimentally, and the calculated energy difference between TS-S and TS-R was quite small, only 1.1 kcal mol^−1, which is also in agreement with the experimental observations from solvent effects. The energies calculated for the six lowest-lying transition states are shown in Figure 44b. Slight differences in the interaction of the solvent with the TSs leading to the S and R-enantiomers cause small differences in energy, which result in the changes in ee observed.

More recently, Otevrel and Bobal developed a novel multifunctional C2-symmetric biphenyl-based diamine-tethered bis(thiourea) organocatalyst (C35), which was tested in the Henry reaction and also applied with success to the catalytic enantioselective syntheses of the enantiopure drug (S)-econazole and (R)-mirabegron, a late-stage intermediate (Figure 45) [89]. Under optimized conditions, the catalyst provided excellent yields and ees, particularly with electron-deficient aromatic and heterocyclic substrates, although moderate drs. Preliminary kinetic and spectroscopic experiments were also performed to study the reaction. Among other parameters, the effect of temperature was probed. A plot of temperature vs. stereoselectivity of nitroaldol 86a exhibited a linear trend intersecting at a value of the reciprocal inversion temperature (T_inv = −5 °C). The syn diastereomers prevailed only at reaction temperatures lower than T_inv. This unusual temperature effect on the ee and dr values was attributed to catalyst-mediated reversibility of the reaction at T > T_inv. On the contrary, temperature variations were less pronounced on the formation of 86b, which displayed a simple temperature effect. No nonlinear effects were found, suggesting that the active form of the catalyst is monomeric. Kinetic studies revealed first-order kinetics concerning the model aldehyde, 2-nitrobenzaldehyde, and the catalyst. Similar experiments to determine the reaction order of the nitroalkane were not possible due to solubility problems and volatility. Instead, the role of nitromethane was investigated by isotopic substitution with MeNO₂-d₃. Proton abstraction of the nitroalkane was identified as the rate-limiting (or a partially rate-limiting) step in the overall reaction.

In 2017 Herrera and coworkers performed a study to test the accuracy of different combinations of DFT methods and basis sets in squaramide catalysis [90]. The Henry reaction between 4-cyanobenzaldehyde and nitromethane was selected as a model reaction. This was a complex reaction of about 100 atoms and many diverse non-covalent interactions. Eighteen different computational approaches were compared to find an accurate one that required the least possible amount of calculation time. Functional wB97X-D provided the best results when used with different versions of the 6-311 basis sets. Highly accurate calculations of the outcomes were obtained when the method was tried on nine aldehydes with different structural characteristics. It was also found that for relatively large systems, such as this, using a split-valence triple-zeta basis saves much time compared to using larger basis sets sometimes employed in organocatalytic studies, as, for example, the TZV and Def2TZV basis set families.

5. Miscellaneous Reactions Involving Anion-Binding Catalysis

Besides the synthesis of chiral lactams described by Vetticat and Seidel (Figure 37, Section 3), other examples of anion-binding catalysis have been described [74]. Those which do not fit in the categories covered in the previous sections have been grouped in this section. Anion-binding catalysis by dual H-bond donors may take place via different activation modes (Figure 46): there may be an association of the H-bond donor to a preformed ion pair, as in A; there may be Brønsted acid co-catalysis, with the H-bond donor enhancing the acidity of a weak acid by stabilizing its conjugate base, as in B or they may be anion abstraction from a neutral substrate as in C [91]. Mode C has been used to explain many enantioselective reactions, but until 2016 it remained still a mechanism poorly understood. In 2016 Jacobsen and coworkers published a study on the amido-thiourea (C36)-catalyzed alkylation of racemic α-chloroisochroman 87 with silyl ketene acetal 88, which sheds some light on this topic [92]. This study followed mechanistic studies performed on this reaction by the group, which suggested that an unusual cooperative mechanism was in operation, with the catalyst resting as a dimeric aggregate under typical reaction conditions (0.1 M in the substrate, 10 mol% of catalyst). For the catalyst to react with the substrate and drive the rate-determining C−C bond formation step, deaggregation was necessary. This can limit efficiency. Catalytic loadings of 5–20 mol% and long reaction times (>24 h) are often required by H-bond donor-catalyzed reactions, and in many cases, they have the limitation of working better under dilute reaction conditions (≤0.1 M in the substrate). These studies aimed to help the design of more efficient catalysts.

Epimerization rate data alone did not allow a distinction to be made between an S_N2 and an S_N1 mechanism. However, whereas racemization of 89 through an S_N1 mechanism should be inhibited by exogenous chloride sources, racemization via an S_N2 mechanism should be promoted by an increased concentration of an exogenous chloride nucleophile. SIR experiments with the addition of tetraoctylammonium chloride or hydrogen chloride (such that [Cl−] = [C36]T) were performed. It was found that epimerization was completely suppressed in both cases, consistent with an S_N1 mechanism for substrate racemization. Ionization promoted by the catalyst affords the oxocarbenium−chloride ion pair. Substrate epimerization and alkylation share the same kinetic dependence on catalyst and α-chloroether, which means that the two processes have a common transition-state stoichiometry. Comparing the rate laws reveals that epimerization catalyzed by C36 or C37 is 101–103 times faster than alkylation, which would be needed for alkylation via either dynamic kinetic resolution or a stereoablative S_N1 mechanism. Since the oxocarbenium−chloride ion pair is generated in the epimerization, it must also be an intermediate in the alkylation. Alkylation proceeds via an anion-abstraction ion-pairing mechanism, as shown in Figure 47, involving the cooperative action of two catalyst molecules. The racemization is catalyst-mediated and Cl⁻ inhibited.

Catalyst aggregation is counterproductive. In further studies by the Jacobsen group, a linkage was introduced between two catalyst monomers, aiming to improve stabilization of the dominant TS complex and hence to improve catalytic activity [93]. Although preliminary DFT analysis did not distinguish whether catalyst activation proceeds by a 4H- or a 2H-Cl⁻-binding geometry, i.e., via I or II (Figure 48), in the solid-state, a 4H-geometry was observed, a fact, which guided the choices of structural changes [90,91]. A methyl substituent, shown previously to minimize competing reaction pathways and improve reactivity and the ees, was introduced [94]. The bisthioureas were also linked in a way that disfavored aggregation while still resembling the structures of the untethered monomers. Of all the catalysts prepared, the best results were obtained with ent-C38, which, when applied to the enantioselective alkylation of racemic isochroman 87 (Figure 47a), afforded the product in 96% yield, 92% ee, in 3 h at −78 °C with a catalytic load of 0.1 mol% only, i.e., with 100 times lower catalyst loading, 8 times shorter reaction time and a 5 times more concentrated solution.

The kinetic resolution of racemic amines via acylation and other methods is an important means to obtain chiral amines for various applications. The utilization of small-molecule catalysts for resolution is still challenging, in contrast to enzymatic methods, and only recently have solutions started to appear, which nevertheless, are not as advanced as the corresponding methods for kinetic resolution of alcohols [95,96]. The kinetic resolution of various classes of amines has been achieved by the Seidel group, with asymmetric nucleophilic catalysis, using an in situ generated chiral acylating reagent [97]. Three components are involved: 4-dimethylaminopyridine (DMAP, a nucleophilic catalyst), an achiral acylating reagent (91), and a chiral anion receptor/hydrogen bonding (HB) catalyst (Figure 49). DMAP, which exists in equilibrium with the corresponding acylpyridinium salt (e.g., ion-pair I), is converted into a chiral species through the interaction of the counteranion with a chiral anion receptor, e.g., a thiourea catalyst, forming ion-pair II. This has an impact on the DMAP/acylpyridinium salt equilibrium. Since ion-pair II is more electrophilic and/or soluble than ion-pair I, reactions with amines should occur more readily with ion-pair II than with ion-pair I, a fact that has been applied in the kinetic resolution of amines. In a recent study, the Seidel group investigated the role of ion-pair II and other reaction components in the enantioselectivity of the kinetic resolution shown in Figure 49a as a model reaction [90]. Computational studies at the M06/6-31G(d,p) level of theory, including solvent modeling utilizing a polarized continuum model, provided new information on the nature of the ion pair and revealed a range of important secondary interactions responsible for enantiodiscrimination. Catalyst C39 was found to be the most efficient. Toluene was the best solvent, and it has been generally observed to be a privileged solvent for enantioselective reactions preceding via catalyst−substrate ion pairing, which agrees with the fact that the nature of the ion pair is influenced by the reaction medium, with nonpolar solvents favoring tight (contact) ion pairs and the more polar ones leading to solvent-shared or solvent-separated ion pairs. NMR studies at different concentrations suggested that the catalyst is present in a nonaggregated form at the concentrations used for catalysis. A 1:1 binding stoichiometry of benzoate by the catalyst was inferred by titration studies with tetrabutylammonium benzoate (TBAB) monitored by NMR and UV-vis titration. This information confirms the idea that the key ion pair involved in catalysis consists of a catalyst with a thiourea subunit binding to the benzoate counteranion of the DMAP-derived N-benzoylpyridinium salt. Deprotonation of the catalyst and involvement of a neutral acylating agent in the reaction could be ruled out by comparison of NMR data with spectra obtained when the catalyst was treated with different bases. Evaluation of two diastereomeric quaternary complexes helped to rationalize the experimental observation that the R-enantiomer of 92 is benzoylated preferentially over the corresponding S-enantiomer. In the case of both the bis-thiourea C39 and the amide-thiourea catalyst C40, dual intramolecular hydrogen bonding to the sulfur atom of a thiourea subunit was found to occur, and the result was activation of this moiety for anion binding since the two N−H bonds become more acidic.

A recent example of anion binding catalysis involves the nucleophilic addition of N-tert-butylhydrazones to isoquinolinium ions (from 95) (Figure 50). This reaction, which leads to the dearomatization of the isoquinoline, proceeded in a highly enantioselective manner in the presence of bifunctional thiourea-based catalyst C41. Two chiral centers were formed in the reaction, and the diastereoselectivity was >20:1 in all cases studied [98]. If the reaction proceeds by anion-binding catalysis, it is expected that the geometry, size and coordination ability of the counteranion should have a marked effect on the results. As a test, anion exchanges (Cl!TFA, BF₄, BARF) were performed before the addition of hydrazone 97a. Poorer reactivities and ees were observed in all cases, as expected when the spherical, small chloride anion was replaced by the new anions. In addition, an experiment with pyrrolidine-derived hydrazone 97b yielded low amounts of product, but racemic, which confirms that the chloride-binding ability of the hydrazone is essential. Further, the evidence available pointed to the formation of a 1:1 complex: the absence of a significant nonlinear effect suggests that a 2:1 complex may not be operating, information, which was also confirmed by the method of continuous variation, for which the data obtained fits with a 1:1 complex.

Computational studies revealed that in the most favored TS, the azomethine carbon approaches the Si face of the isoquinolinium C(1)=N bond (Figure 51A). A difference of 10.5 kcal mol⁻¹ was found concerning the most stable transition structure leading to the R enantiomer of 97c. NCI analysis of the lowest energy transition structure showed the presence of stabilizing Cl–π, CH–π and π–π interactions [99]. A new tool for NCI quantitative analysis showed that noncovalent interactions of the chloride ion with the NH’s of the thiourea moiety and hydrazone, as well as with the azomethine hydrogen atom, are stronger in the preferred TS (Figure 51B, bottom square). These interactions fix the orientation of the reagents in a highly ordered [120]-Cl-hydrazone complex as the key intermediate, from which there is the preferential approach of the azomethine carbon to the Si face of the isoquinolinium C(1)=N bond. This mode of reaction is responsible for the high stereoselectivity observed, as well as the S,S absolute configuration of the product.

6. Miscellaneous

Enantioselective transfer hydrogenation is another reaction that has benefited from developing bifunctional dual hydrogen donor catalysts [100]. The reduction of nitroalkenes is an example. In 2015 Benaglia and coworkers developed a procedure for the reduction of β-trifluoromethyl-substituted nitroalkenes 99 with Hantzsch esters 100 as hydrogen sources (Figure 52) [101]. A multifunctional thiourea-based (S)-valine derivative (C42a) was found to be a good catalyst, affording very good yields and ees of amines in the case of trisubstituted olefins. Tetrasubstituted olefins reacted with lower stereoselectivity. Computational studies were performed to investigate the reaction further. The transition states leading to the formation of both (R)- and (S)-101a (101, R¹ = Ph), TS-R and TS-S, for the reaction with catalyst C42b, located at a B3LYP/6-31G(d) level of theory are shown in Figure 52; finer electronic energies were obtained increasing the basic setup to 6/311+(2df,2pd) with two different functionals: B3LYP and M062X. A methyl-substituted Hantzsch ester was used instead of 100 for simplification. A preliminary conformational analysis revealed a very rigid structure, and only three conformations were found. It was assumed that there was the coordination of the nitro group to the thiourea moiety and of the Hantzsch ester NH group to the catalyst carboxyamide group. At a first inspection, it may be seen that the pyrrole moiety plays an important role. In TS-S, there is a repulsive interaction between the pyrrole ring and phenyl group of the nitroalkene, which disfavors the formation of the S enantiomer. In addition, with the substrate coordinated to the thiourea, the nitroolefin C2 electrophilic carbon atom in TS-S is closer to the bulky diaminocyclohexane moiety than in TS-R. This forces the Hantszch ester, in releasing the hydride, to a hindered area with the ester group repulsively interacts with the 2,5-methyl pyrrole substituents, thus further favoring TS-R.

Qiu, Zhao and coworkers described a Morita–Baylis–Hillman reaction of N-alkyl isatins 102 with acrylates 103 catalyzed by a bifunctional phosphine–squaramide (C43) in 2015 (Figure 53) [102]. The catalyst was part of a series synthesized from inexpensive and commercially available β-amino alcohols. The products, chiral 3-substituted 3-hydroxy-2-oxindoles (104) bearing an exocyclic double bond, were obtained in very high yields and ees and the reaction mechanism was probed by means by ³¹P NMR, ESI-MS and KIE studies. KIE studies with N-methylisatin and butyl acrylate revealed a carbon isotope effect on the 3-carbon when the ¹³C ratio of recovered N-methyl isatin was compared to that of the original sample (C3 = 1.021). This suggests that C–C bond-formation is the rate-limiting step of the MBH reaction. A mechanism was proposed, as shown in Figure 53, with the support of MS since some of the intermediates could be detected.

The Pictet–Spengler reaction allows the efficient construction of important heterocycles of interest for the pharmaceutical industry. In nature, this reaction gives rise to monoterpene indole alkaloids, as well as many secondary metabolites. The reaction mechanism was investigated by Jacobsen and coworkers recently [103]. Initially, there is a condensation of a β-arylethylamine with an aldehyde, C–C bond formation by addition of the pendant arene to the resultant iminium ion and subsequent rearomatization (via deprotonation). In tryptamine derivatives, the C−C bond-formation may take place by direct electrophilic aromatic substitution at C2 or by alkylation at the more nucleophilic C3 position, followed by C−C migration. Although both pathways had been shown to be viable, some questions remained unknown, such as how rearomatization occurred in the nonpolar media commonly used in asymmetric synthesis, and the studies by the Jacobsen group aimed to clarify those issues. Based on a system previously shown by the group to have broad applicability, with 105 and 76, and to afford chiral tetrahydro-β-carboline products 106 with high ees thanks to the cooperative action between a bifunctional chiral thiourea-based catalyst (C44) and weak achiral Brønsted acids, such as benzoic acids, new studies were undertaken (Figure 54) [104]. It was found by applying kinetic studies, isotope effects, structure−enantioselectivity relationships, and computational analyses that rearomatization (the deprotonation) is the rate- and enantioselectivity-determining step. The indole N-H was found to be important for enantioinduction since the reaction of a methylated substrate yielded a nearly racemic product. In addition, it was also observed that the co-catalysts play key roles in several steps leading up to and, including the deprotonation step. Coordination to the thiourea increases the acidity of benzoic acid leading to substrate protonation. Anion-binding to the conjugate base helps to stabilize the pentahydro-β-carbolinium ion intermediate. These interactions contribute to rate acceleration. Enantiodifferentiation is due to differential π···π and C−H···π interactions in the TSs leading to the two enantiomers, which are organized by hydrogen-bonding interactions. On the whole, catalyst C44 displays many of the features of enzymatic Pictet−Spengler catalysts.

In 2018 Zhong and coworkers did a related study on a cinchona alkaloid squaramide-catalyzed asymmetric Pictet–Spengler reaction [105]. DFT calculations were also performed to clarify the reaction mechanism. The products were obtained with very high yields and moderate to very high ees when quinidine-derived squaramide catalyst C44 was employed to promote the reaction. In this process, there was no need for the addition of a co-acid catalyst. The ees were found to be highly dependent on the nature of the aldehyde used. They were as high as 99%, but generally, when a 2- or 4-electron-withdrawing substituent was present in an aryl aldehyde, the ees were low.

Bifunctional dialkyl sulfide catalysts are still rare. However, in 2018 Yamanaka, Shirakawa and coworkers described the synthesis of some novel catalysts with these features that could promote regio-, diastereo-, and enantioselective bromolactonization of 107 to yield compounds 109 [106]. C45 was the best. The presence of the urea functionality was found to be critical since an analogous catalyst bearing a methyl group instead of SBu afforded racemic product only. Hence, was the nature of the brominating agent 108. The results for a few regents 108 tried are shown in Figure 55. DFT calculations were performed to provide some insight into the origin of stereoselectivity. An associative TS model involving monocoordinated succinimide anion fitted best. The structures of the most promising TS models are shown in Figure 56. A 4.3 kcal mol⁻¹ difference between TS_major and TS_minor agrees with the experimental results. The succinimide anion is found to be coordinated with both the urea and the sulfide moieties in TS_major, but only with the urea moiety in TS_minor, and additional NH/π interaction between a portion of the NH residue of the urea moiety and the naphthyl moiety in TS_major account for the higher stability obtained.

A recent development in this field is using multivariate linear regression analysis to help the design of new catalysts. One of these works was recently published by Werth and Sigman [107]. By connecting catalyst and substrate structural features, the proposed transition states and the noncovalent interactions responsible for asymmetric induction, using an iterative statistical modeling process, this research aimed to provide a basis for extrapolation to new reactions and/or catalysts so that more accurate predictions could be made in future developments rather than utilizing a totally empirical approach.

7. Conclusions

Dual H-bond donor bifunctional catalysts have become in recent years a very popular and efficient means to perform both new and old reactions in a highly stereoselective manner. Understanding the mechanisms of these reactions and the origin of stereoinduction can help the rational development of new powerful catalysts for future applications. Computational tools and DFT calculations have helped to elucidate mechanisms and reveal nonbonding interactions, which are responsible for the stabilization of transition states and selectivity. Additional studies, such as the determination of acidity scales and kinetic isotopic effects, have had important contributions too. Catalytic loadings are becoming lower, with successful transformations being achieved with even less than 1 mol% of catalysts of interest for large-scale applications. Given the easy-to-assemble modular nature of these catalysts, the large number of reactions that they have made possible with high stereoselectivities, and the mild reaction conditions under which they operate, a bright future is expected for further developments in this area.