Catalytic Asymmetric Aza-Darzens Reaction with a Vaulted Biphenanthrol Magnesium Phosphate Salt

Abstract

Conditions for a catalytic asymmetric aza-Darzens aziridine synthesis mediated by a vaulted biphenanthrol (VAPOL) magnesium phosphate salt is described. Using simple substrates this methodology explores the scope and reactivity of a new magnesium catalyst for an aziridination reaction capable of building chirality and complexity simultaneously.

The Darzens aziridine synthesis, first proposed by Deyrup,¹ is a pathway to prepare complex aziridines, with considerable synthetic utility, from common synthetic building blocks. Examples where such aziridines are used as versatile intermediates are mainly through their ring opening reactions for the preparation of chiral amines.² Clearly new catalytic asymmetric methods of aziridine synthesis via the aza-Darzens reaction would be interesting due to this synthetic potential. In two reports from 1999 and 2000, a breakthrough by Wulff in the area of catalytic asymmetric aziridination of imines was reported.³ The catalyst for this reaction, while initially believed to be a boron Lewis acid, was later proven to be a chiral Brønsted acid after intense mechanistic study.4 Additionally, in 1999 independently both Sweeney⁵ and Davis⁶ reported asymmetric versions of the aza-Darzens reaction using stoichiometric amounts of chiral reagents. In other early studies, Johnston and coworkers published a triflic acid-catalyzed aza-Darzens variant.⁷ More recently a number of new Brønsted acid variants similar to these reactions were subsequently published, with highly enantioselective conditions described.⁸ We believed that despite the above efforts, the use of typical aza-Darzens type nuceophiles⁹ like α-halo-1,3-dicarbonyl compounds could lead to functionalized aziridines using chiral phosphoric acids as potential catalysts.¹⁰

After some limited initial success while focusing on chiral phosphoric acids, we turned our attention to chiral phosphate metal salts as viable mediators for this reaction. We were attracted to these catalysts after a report by Ishihara on similar Mannich reactions showed that metal phosphate salts were superior catalysts when screening versus the free phosphoric acids.¹¹ These investigations were supported by the findings of List,¹² who showed that phosphoric acids used directly after column chromatography can be the neutralized mixtures of phosphate salt impurities and acid, rather than just the pure acid.

We were encouraged by recent reports that successfully demonstrated a variety of chiral phosphate salts for catalytic asymmetric induction.¹³ Using chiral phosphate salts as catalysts, we wish to report an asymmetric aza-Darzens reaction that employs α-chloro-1,3-diketones as competent nucleophiles for addition to N-benzoyl imines, allowing access to tri-substituted aziridines with good enantioselectivity.

As we began our screening we found that common phosphoric acids (Table 1, entries 1–3) were relatively poor catalysts for this reaction. However, these catalysts did indicate, based on enantioselectivity, that the reaction carried a strong preference for P3 (VAPOL), a phosphoric acid derived from (R)-2,2′-Diphenyl-3,3′-(4-biphenanthrol) (Figure 1, entry 3). When P3 was used directly after purification on silica gel (entry 4), the enantiomeric excess was significantly higher than the re-acidified phosphoric acid (entry 3); this lead to the obvious pursuit of a phosphate salt catalyst with chirality from VAPOL. To explore this new chiral phosphate system we matched P3 with commerically available alkali or alkaline salts. Group 1 elements were not promising: lithium showed excellent efficiency leading to a high yield but low enantioselectivity (entry 5), while sodium prevented the aza-Darzens reaction from occurring at all (entry 6). While all three of the group 2 alkalines promoted the reaction with some enantioselectivity, the most selective example was the pairing of P3 with magnesium (entries 7–9). From the comparison of the yields and selectivity’s of the various salts, it became clear that magnesium VAPOL phosphate salt impurities originating from the silica gel were most likely catalyzing the reaction in entry 4. To our knowledge this is the first example of a magnesium phosphate salt being the optimal catalyst for an enantioselective Mannich type reaction.

Table 1.

Optimization of aza-Darzens aziridination reaction.^a

entry	catalystb	cat. loading (mol %)	yield (%)c	ee (%)d
1	H[P1]	5	48%	0%
2	H[P2]	5	63%	−7%
3	H[P3]	5	32%	52%
4e	H[P3]	5	65%	80%
5	Li[P3]1	5	83%	4%
6	Na[P3]1	5	0%	-
7	Ca[P3]2	2.5	57%	44%
8	Ba[P3]2	2.5	49%	49%
9	Mg[P3]2	2.5	74%	80%
10	Mg[P3]1	5	63%	70%
11	Mg[P3]4	1.25	14%	75%
12	Mg[P1]2	2.5	57%	−4%
13	Mg[P2]2	2.5	65%	−5%

^aGeneral reaction conditions: THF was removed before step 2 while 1.5 equiv of 1e, 1.0 equiv of 2 and 2.0 equiv of DMAP were added.

^bSee the Supporting Information for details on the preparation of the catalysts.

^cIsolated yield.

^dDetermined by chiral HPLC analysis.

^ePurified on silica gel without re-acidification.

Figure 1.

Common phosphoric acids and their phosphate salt equivalents

Initially we assumed that a ratio of two P3 to one magnesium would be optimal; to confirm, a screening of other ratios of VAPOL phosphate and magnesium was performed. A significant drop in yield and enantioselectivity was observed compared with the initial ratio (entries 10–11). Lastly we wanted to determine if other common phosphoric acids could be as effective as the P3 catalyst. After conversion to the corresponding magnesium phosphate salt, both catalysts derived from P1 and P2 showed little enantioselectivity for the aza-Darzens reaction (entries 12–13). These results confirm that VAPOL magnesium phosphate is ideal for promoting the aza-Darzens reaction efficiently and with selectivity.

With the optimal conditions identified we proceeded to examine various substituted imines derived from their corresponding benzaldehydes (Scheme 1). These aza-Darzens aziridine products were formed with moderate yields and high enantioselectivities.

Scheme 1.

Substrate scope.a

^a General reaction conditions: 2.5 mol % catalyst loading, THF was removed before step 2 while 1.5 equiv of 1, 1.0 equiv of 2 and 2.0 equiv of DMAP were added. All yields are isolated and enantiomeric excess was determined by chiral HPLC analysis.

In example 3a when the starting material is fully consumed, the mass balance of the reaction can be accounted for by incomplete conversion in the second, ring closing step. Various substituents at the aryl ring were evaluated in the exploration of the substrate scope. Both methoxy and methyl groups showed analogous tolerance and enantioselectivities regardless of their position on the arene ring system (3b–3f). However, as the steric environment increased around the center of chirality the enantioselectivity decreased, clearly the result of the substituent on the aryl ring. This is evident when observing the enantioselectivities of 3c and 3f compared to 3b and 3e, respectively. The presence of electron withdrawing halogens in the para position (3g–3i) gave comparable results with the electron donating para substituted substrates.

While the active catalytic species involved in these phosphate salt-catalyzed reactions is not yet definitively proven, we moved forward with a theoretical study using two biphenanthrol phosphate ions coordinated to one magnesium cation. Catalyst structures were geometry optimized using the Q-Chem ab initio package¹⁴ and employing the B3LYP¹⁵ functional and 6–31+G* basis set.¹⁶ Figure 2 represents the lowest energy structure found.¹⁷ Interestingly the dihedral angle of the phenanthrene rings was found to be approximately 60°; the axial C₂-symmetry undoubtedly controls enantioselectivity of the products formed.

Figure 2.

Calculated structures of VAPOL magnesium phosphate.

Two alternative catalyst conformations were located; these were higher in energy, relative to (Figure 2, top), by 10 and 24 kcal/mol respectively. Different coordination patterns of magnesium can partially account for these energy differences (vide infra). In the lowest energy structure, (Figure 2, top) the magnesium is coordinated to the four terminal phosphoryl oxygens. In contrast the higher energy conformers, respectively, had magnesium bound to either one or two phosphoryl ethers in conjunction with three or two terminal oxygens. From the NBO¹⁸ analysis energetics of these interactions indicate that an approximate 5 kcal/mol penalty exists when binding phosphoryl ether compared to terminal oxygen. However, this energetic penalty is compensated for by increased stabilization via interactions with the adjacent terminal oxygen. Measuring the magnesium-oxygen distances highlights this as the magnesium-ether-oxygen distance is 2.1 Å compared to 1.9 Å for the magnesium-terminal-oxygen (Figure 2, middle).

In contrast, the lowest energy structure (Figure 2, top) has magnesium-oxygen distances of approximately 2.1 Å for all interactions. Clearly the orbital interactions cannot explain the energetic differences observed, however, large structural changes are evident based on the interaction patterns. Therefore, a combination of steric repulsion, strain, and hydrophobic / hydrophilic effects likely dictate the final structure.

Using this structural information, we propose a possible transition state and mechanism to explain the high enantioselectivity observed. Figure 3 shows coordination of the magnesium to the carbonyls of the imine and diketone’s enol form. Additionally, the enol can hydrogen bond to the oxygen’s on the catalyst. The catalyst can simultaneously stabilize the nucleophile and electrophile, while providing the chiral environment for asymmetric induction.¹⁹

Figure 3.

Proposed transition state.

We surmised, therefore, that the enantioselectivity is a consequence of steric interactions between the phenanthrene rings and the aromatic substituents of the imine, allowing for facial selectivity. As the imine coordinates with the catalyst its aryl rings align perpendicular to the plane of the catalyst while crossing the center to minimize the steric interactions with the phenathrenes. The validity of the proposed mechanism can be supported by its ability to predict the absolute configuration of the major enantiomer (see below, and Supporting Information for details). The absolute configuration was determined by HPLC comparison of the dechlorinated β-amino carbonyl compound to the literature.²⁰ The absolute configuration corresponds with that of the S enantiomer for 3a validating the chirality predicted by the transition state.²¹

In conclusion, we have demonstrated the utility of a chiral VAPOL magnesium phosphate salt catalyst and described its first use in the enantioselective aza-Darzens reaction. This was accomplished while simultaneously showing a new approach to the aza-Darzens reaction through the addition of α-chloro-1,3-dicarbonyl compounds. These chiral phosphate salts represent a powerful step forward in our ability to tailor optimized catalytic systems to specific reactions. Further experiments to evaluate the utility of VAPOL derived phosphate salts are currently underway in our laboratory.