Catalytic Enantioselective Synthesis of Axially Chiral Imidazoles by Cation-Directed Desymmetrization

Abstract

Axially chiral five-membered heterobiaryls synthesized by enantioselective catalysis typically feature large ortho-substituents or a heteroatom in the chiral axis to maintain a stable configuration. Herein we report a cation-directed catalytic enantioselective desymmetrization method that enables rapid access to axially chiral imidazoles with the basic nitrogen at the ortho position and efficiently integrates π-stacking moieties to ensure a stable axial configuration for further applications. The process is operationally simple, highly enantioselective, and can be performed on the gram scale. The majority of the products are obtained in >90% ee, but interestingly even those with only moderate ee can readily be enriched to near optical purity by selective racemate crystallization. Together with a mild phosphine oxide reduction method, axially chiral imidazoles such as StackPhos and its derivatives, are readily prepared in high yield and excellent enantioselectivity on the gram scale. The method also enables the preparation of new chiral non-phosphine-bearing imidazoles.

Graphical Abstract

Introduction

Due to the prevalence of axially chiral heteroaromatic scaffolds in bioactive molecules¹ and chiral ligands,² methods to construct different types of axially chiral heteroaromatics are highly desirable. Pioneering studies were focused on the synthesis of axially chiral [6.6] aryl-heteroaryls³ and a variety of catalytic enantioselective methodologies to access these biaryls have been developed.⁴ Only recently, significant progress has been made towards the design and synthesis of axially chiral compounds containing five-membered heterocycles ([5.6]-biaryls).⁵ Inclusion of a five-membered heterocycle leads to reduced steric interactions between ortho-substituents around the chiral axis. Consequently, with the labile conformation, axially chiral [5.6] aryl-heterobiaryls are more difficult to prepare than [6.6] aryl-heteroaryls.⁶ To maintain a stable configuration, axially chiral five-membered heterobiaryls typically require four ortho-substituents around the chiral axis⁷ or a relatively short C–N bond⁸ to provide sufficient steric interactions (Figure 1A). In most cases two ortho-substituents on the five-membered heterocycle are needed to maintain the axial chirality.^5a,9 Direct access to axially chiral five-membered heterobiaryls beyond these current structural limitations remains elusive.¹⁰

Figure 1.

Overview of axially chiral imidazole synthesis

Imidazoles show unique functions in biological processes and synthetic chemistry (Figure 1B). The imidazole group in histidine serves as a coordinating ligand in metalloproteins, as a catalytic residue in enzymes, and as a proton shuttle in biological systems.¹¹ Many elegant imidazole ligands and organocatalysts with point chirality have been developed and have found broad applications in asymmetric catalysis;¹² however, axially chiral imidazoles have been rarely studied due to their configurational lability. One recent strategy was to lock the conformation with another chiral group enabling the preparation of a few highly interesting axially chiral 2,2’-biimidazoles such as 6.¹³ In contrast, we previously reported an axially chiral imidazole based P,N–ligand, StackPhos 7, and demonstrated that the axially chiral configuration incorporating a five-membered heterocycle can be stabilized through π-stacking.¹⁴ This strategy obviates the need for the incorporation of large ortho-substituents and makes it possible to utilize the basic nitrogen on the imidazole as a coordination site. StackPhos has exhibited high enantioselectivity and reactivity in Cu-catalyzed alkyne addition reactions¹⁵ that is complementary to the six-membered heteroaryl P,N-ligands. Enantioenriched StackPhos and congeners can be obtained through a dynamic process whereby deracemization of the racemic ligand is achieved by employing a stoichiometric amount of a chiral Pd-complex. Other state-of-the-art P,N-ligands such as QUINAP typically utilize these chiral Pd-complexes for resolution, but regardless of the mode of reactivity, use of a sacrificial Pd-complex is highly limiting with regard to the types of compounds that can be prepared and adds undue cost to ligand preparation, especially on scale. Although our deracemization strategy works quite well and provides these ligands in >99% ee,^15d it is limited to phosphines and is therefore not applicable to the synthesis of other axially chiral imidazoles. Thus, an efficient method to construct axially chiral imidazoles is needed to overcome the current synthetic limitations and expand the availability of axially chiral imidazoles.

Results

To address this problem, the axially chiral imidazole core would ideally be constructed through a catalytic enantioselective transformation. We hypothesized that 1H-imidazole 8 (Figure 1C) could be deprotonated to generate the achiral anion 9 containing enantiotopic nitrogen atoms that could be differentiated by a suitable chiral reagent or catalyst. Subsequent installation of the R group on the imidazole to form 10 should increase the rotational barrier around the aryl-imidazole bond through our π-stacking strategy or potentially steric hindrance, thus setting the stereochemistry of the axially chiral imidazole.

Inspired by the recent reports on phase-transfer catalysis in atropisomer synthesis,^9b,9d,16 we envisioned that a chiral cation-directed desymmetrization of 2-aryl imidazoles should be possible, albeit unprecedented. Initial attempts with the phosphine 1H-imidazole led to low enantioselectivity (~10% ee) and substantial amount of chiral StackPhos oxide formation (28% ee). We were pleased to find that direct use of phosphine oxide 11a led to the formation of 13a with moderate enantioselectivity, 48% ee (Table 1, entry 1). Additional experiments (shown here and in the SI) confirmed that the phase-transfer catalyst structure has a significant impact on enantioselectivity with electron deficient N-substituents giving higher selectivity (12c, 12d, entries 3–4). Further improvement to 80% ee was observed with free phenolic catalyst 12e. Finally, combining these structural features in 12f led to the highest selectivity (83% ee) with these aqueous NaOH conditions. Further optimization with the use of solid K₂CO₃ improved enantioselectivity to 92% ee but only gave 60% conversion (Table 1, entry 7). With K₃PO₄,¹⁷ the StackPhos oxide 13a was formed with both excellent yield and enantioselectivity (Table 1, entry 8), and these were deemed the optimal conditions.

Table 1.

Reaction Optimization

entry	catalyst	base	ee (%)
1	12a	NaOH (aq.)	48
2	12b	NaOH (aq.)	46
3	12c	NaOH (aq.)	57
4	12d	NaOH (aq.)	74
5	12e	NaOH (aq.)	80
6	12f	NaOH (aq.)	83
7	12f	K2CO3 (s)	92a
8	12f	K3PO4(s)	92b

10 mol% catalyst, 3.0 equiv C₆F₅CH₂Br, 72 h. See SI for full details;

^a60% conversion instead of full conversion;

^b95% yield.

With the optimized conditions established, the reaction scope was explored. Initial studies were aimed at varying the imidazole substituents (Table 2A) and those with electron rich aryl (13b, 13d) and heteroaryl substituents (13g) as well as bulky aryl groups (13f) were formed in excellent enantioselectivity, while F-substituted aryls resulted in 13c and 13e in very good, but slightly lower enantioselectivity (90% ee and 83% ee, respectively). Different ortho-substituents on the naphthalene ring of the imidazoles are also tolerated in this reaction (Table 2B). Electron rich (14a, 14f), alkyl substituted aromatic phosphine oxides (14b, 14g, 14h), and heteroaryls such as furan (14i) and thiophene (14j) were all benzylated in excellent enantioselectivity. Interestingly, the 4-CF₃C₆H₄ product 14e is formed with only modest selectivity. The P=S product (14k) was generated in 93% ee with a longer reaction time. Various benzylating reagents were also tested (Table 2C). Simple benzyl bromide gave the N-benzylated imidazole 15a in 91% ee. Benzylating reagents with 3- and 5-substituents, including both electron withdrawing (15b, 15c, 15d) and donating groups (15e, 15f) showed excellent enantioselectivity. Many other substitution patterns were also tolerated and generated the corresponding products in high enantioselectivity and yield.

Table 2.

Reaction Scope

^aReaction time 7 days

^bNaOH (aq.) conditions from Table 1.

^cReaction conducted at room temperature. See the SI for full details

At this stage, it was also highly desirable to have access to the enantiomeric series of products for further applications. The use of diastereomeric phase-transfer catalysts are known to act as pseudoenantiomers¹⁸ and studies were undertaken to form ent-13a. As can be seen in Table 2, catalyst 12f’ afforded the desired compound in 90% yield and 87% ee.

With access to both enantiomers, we envisioned that this method may provide access to a versatile library of imidazole-based axially chiral P,N-ligands; however, reduction to the phosphine would be needed as well as further enantioenrichment to reach the contemporary standard of 99% ee required for single enantiomer ligands. The difficulty of this task lies in breaking the P=O bond without racemizing the atropisomer under the harsh conditions typically required.⁶ Indeed, preliminary work in our laboratory demonstrated that heating the reaction to 150 °C was needed for the reduction of racemic StackPhos with HSiCl₃; and, in separate studies, significant racemization (loss of 24% ee) was observed at 75 °C after 10 hours.¹⁴ After an extensive survey, most of the reported phosphine oxide reduction methods were not successful.¹⁹ Alternatively, the chiral StackPhos sulfide 14k was successfully reduced with Raney-Ni to give the product without any loss of enantiopurity at room temperature (see SI for full details).²⁰ While this strategy worked well, it should be noted that lower enantioselectivities were obtained when other phosphine sulfides were exposed to the enantioselective phase-transfer catalysis reaction, and hence we were resolute in identifying more widely applicable P=O reduction conditions.

As mentioned above, only one set of reduction conditions seemed promising and after adaptation of Lemaire’s remarkable conditions to our system, it was found that reduction with tetramethyldisiloxane (TMDS) and Ti(OiPr)₄ at 60 °C yielded StackPhos 7 with only 2% ee loss and in 85% yield (Table 3).²¹ Other chiral phosphine oxide derivatives were also selectively reduced with this mild method. Interestingly, while 7a, a phosphine oxide with bulky substituents on the imidazole, was reduced with minimal loss of optical purity, less bulky substituents (7b and 7c) led to a more substantial loss of enantiopurity. Substituents on the diarylphosphino-groups, impacted both selectivity and reactivity. With 7d, only 1% ee loss was observed under the standard conditions. The reduction of electron-rich phosphine 7e required higher temperature (70 °C) to achieve good conversion and it was found that 7f could be formed in high enantiopurity with a shorter reaction time (6 h). In contrast to the enantiospecific reduction of 13a, only racemic 7-H₅ was obtained from 15a under the identical reaction conditions (eq 1).

Table 3.

Scope of Chiral StackPhos Oxide Reduction

^a70 °C in heptane 16 h

^b6 h. See SI for full details.

Although the highly enantioenriched compounds could now be obtained directly from the reduction, for use in enantioselective catalysis, optical purities would need to be further increased to >99% ee. Initial studies were focused on recrystallization and traditional resolution methods, but these attempts were not successful. In fact, the racemates seemed to selectively crystallize and this observation led to a fairly general method of enantioenrichment (Table 4). Using selective racemate crystallization, a variety of phosphine oxides and phosphines could be obtained in excellent ee. In a very straightforward procedure, the phosphines are simply dissolved in hexanes and allowed to stand while the racemate crystallizes.

Table 4.

Selective Racemate Crystallization

^ahexanes/toluene at 0 °C.

^bhexanes/DCM at 0 °C.

With the enantioselective desymmetrization, reduction, and further enantioenrichment protocols established, a concise synthesis of enantiopure StackPhos was undertaken. As can be seen in Scheme 1, the condensation of 16, benzil, and ammonium acetate, followed by triflation gave 17 in 80% yield on a >10 g scale. The triflate 17 was then coupled with diphenylphosphine oxide to yield phosphine oxide 11a, which smoothly underwent enantioselective desymmetrization on the gram scale to produce S-StackPhos oxide 13a in 94% yield and 92% ee. The reduction of 13a led to S-StackPhos 7 in 93% yield and 90% ee. Selective racemate crystallization was achieved by treating 90% ee S-StackPhos 7 with hexanes at −20 °C, to provide >1 g of optically pure product.

Scheme 1.

Five-step synthesis of enantiopure S-StackPhos on gram scale

As mentioned above, one of the goals was to provide a method that did not require Pd-complexes formation for enantioenrichment to expand the availability of these axially chiral imidazoles. To test this, the simple 2-iodo- and 2-phenylnaphthalenes 14l and 14m, respectively, were prepared under the conditions listed in Table 2. Although the ee’s were modest, it was found that both 14l and 14m could be enantioenriched to >90% ee by selective racemate crystallization and it is likely that the catalytic enantioselective phase transfer reaction could be further optimized to address compounds of this type. The absolute stereochemistry of 14m was determined by X-ray crystallography. The π-stacking between F₅-phenyl group and the naphthalene ring of this non-phosphino-imidazole 14m was also elucidated by this crystal structure.

Although the method could be expanded to a broad scope, one observation made during the optimization seemed curious. More specifically, we were surprised that catalysts with the free phenolic hydroxyl group (e.g. 12f) performed much better than those without (e.g. 12a), perhaps suggesting that a pentafluorophenyl-benzylated catalyst might be the actual catalyst. To probe the catalyst behavior, the reaction was monitored by high resolution mass spectrometry. Interestingly, no change in the catalyst structure was observed under the standard potassium phosphate conditions (Figure 2). However, the generation of a mono- and di-benzylated catalyst was detected by HRMS when using the initially employed and less selective aqueous NaOH conditions (Figure 2). Over the course of the reaction, full conversion of the catalyst to the di-benzylated catalyst is achieved and it is the sole catalyst detected in the mixture.

Figure 2.

Mass spectrometry analysis of catalyst. R=C₆F₅CH_2⁻

This change in catalyst structure could be related to many factors affecting the rate of catalyst benzylation. Under the standard conditions, representative of the K₃PO₄ HRMS experiments (Figure 2), 13a is obtained in 92% ee (Table 5, entry 1). With aqueous NaOH conditions, representative of the NaOH HRMS experiments, 13a is obtained in 83% ee (entry 2). A control experiment with KOH was employed to probe the simple switch from Na+ to K+ under aqueous conditions (entry 3) and a similar ee was obtained, 86% ee. Interestingly, no reaction is observed with solid Na₃PO₄ (entry 4). These results clearly indicate that more is at play under these conditions. The impact of using K₃PO₄ in phase transfer reactions has been studied previously.¹⁷

Table 5.

Control Experiments^a

entry	catalyst	base	ee (%)
1	12f	K3PO4 (s)	92
2	12f	NaOH (aq.)	83
3	12f	KOH (aq.)	86
4b	12f	Na3P04 (s)	N/A
5	12f-OMe	K3PO4 (s)	85
6	12f-OMe	NaOH (aq.)	65

^aConditions: 10 mol% catalyst, 3.0 equiv C₆F₅CH₂Br, 72 h, full conversion (>95% yield).

^bNo reaction was observed.

To specifically assess the importance of each hydroxyl group in the catalyst, we prepared the phase-transfer catalyst 12f-OMe. Compared to the optimal conditions (Table 5, entry 1), the use of 12f-OMe led to a decrease in enantioselectivity to 85% ee (Table 5, entry 5). This finding suggests that the phenolic OH group may play a substantial role in the control of enantioselectivity. A further decline in enantioselectivity was noted when aqueous NaOH was used (Table 5, entry 6), presumably due to the benzylation of the remaining hydroxyl group on 12f-OMe. It is unlikely that the K₃PO₄ base would cause benzylation of the catalyst based on the mass spectrometry analysis results in Figure 2. Therefore, it can be concluded that both hydroxyl groups in the catalyst 12f are critical for achieving high enantioselectivity in this reaction.

A Hammett correlation between the aryl phosphine oxide substitution and the enantiomeric ratio was also noted (Figure 3) whereby more electron-rich phosphine oxides led to higher enantioselectivities. Other than electrostatic interactions, hydrogen bonding interactions have also been proposed in the different mechanistic models of enantioselective phase-transfer catalysis.²² One possible explanation is that the electronic effect of phosphine oxides could come from the hydrogen bonding interaction between phosphine oxides and the catalyst hydroxyl groups. These observations could be valuable for the future development of a detailed mechanistic model and further studies aimed at understanding the important selectivity parameters are underway in our laboratory.

Figure 3.

Hammett correlation between phosphine oxide aryl substitution and enantiomeric ratio

While the pentafluorobenzene π-stacking stabilization is crucial for the preparation of enantiopure axially chiral imidazole-based phosphines, it seems less significant for phosphine oxides, since phosphine oxides without the pentafluorobenzene π-stacking can be prepared with high enantioselectivity via phase-transfer desymmetrization. To gain a better understanding of the role of the π-stacking effect on configuration stabilization, rotational barriers of representative compounds were measured (Table 6). Surprisingly, the rotational barrier of 15a is only 26.4 kcal/mol, showing 0.2 kcal/mol increase from 7-H₅.¹⁴ The half-life for racemization of 15a at room temperature is only about 14 days based on its rotational barrier, and indeed, a 24% ee loss of 15a was observed after 24 hours in 1,2-dichloroethane (DCE) (eq 2). However, when stored at −20 °C as a solid, 15a can retain >90% ee for up to a year. To the best of our knowledge, very few compounds with such a low rotational barrier have been synthesized by enantioselective catalysis in an enantioenriched form (>90% ee).^7d The additional challenges posed by these products without pentafluorobenzene π-stacking stabilization further underscore the mildness and effectiveness of this enantioselective phase-transfer desymmetrization method.

Table 6.

Rotational Barriers of Axially Chiral Imidazoles

The impact of π-stacking becomes more apparent when considering the differences in racemization half-life at room temperature. 13a has a racemization half-life of over 10 years which is in stark contrast to the configurational lability of 15a, due to the 3.3 kcal/mol pentafluorobenzene π-stacking stabilization effect. A similar increase in the rotational barrier due to the pentafluorobenzene π-stacking, 3.0 kcal/mol, is observed between 7-H₅ and 7. These results highlight the significant role of π-stacking and further emphasize the importance of incorporating π-stacking moieties enantioselectively.

Conclusions

In conclusion, we have reported the catalytic enantioselective synthesis of axially chiral imidazoles via cation-directed desymmetrization and mechanistic studies verify the importance of hydroxyl groups in the optimized catalyst. The method is simple, easily scalable and enables the preparation of new chiral non-phosphine, 2-aryl imidazoles in high enantioselectivity. Incorporating π-stacking moieties appears to be the key to stabilizing these 2-aryl imidazoles for further applications. Overcoming the challenging preparation of configurationally stable axially chiral five-membered heteroaryls should unlock the potential of these heterocycles in catalysis and other applications by enabling the design and enantioselective synthesis of these molecules. Further studies of this ilk are currently underway in our laboratories and will be reported in due course.