Visible Light-Mediated, Diastereoselective Epimerization of Morpholines and Piperazines to More Stable Isomers

Abstract

We report a photocatalyzed epimerization of morpholines and piperazines that proceeds by reversible hydrogen atom transfer (HAT) and provides an efficient strategy for editing the stereochemical configurations of these saturated nitrogen heterocycles, which are prevalent in drugs. The more stable morpholine and piperazine isomers are obtained from the more synthetically accessible but less stable stereoisomers, and a broad scope is demonstrated in terms of substitution patterns and functional group compatibility. The observed distributions of diastereomers correlate well with the relative energies of the diastereomer pairs as determined by density functional theory (DFT) calculations. Mechanistic studies, including luminescence quenching, deuterium labeling reactions, and determination of reversibility support a thiyl radical mediated HAT pathway for the epimerization of morpholines. Investigation of piperazine epimerization established that the mechanism is more complex and led to the development of thiol free conditions for the highly stereoselective epimerization of N,N’-dialkyl piperazines for which a previously unrecognized radical chain HAT mechanism is proposed.

Graphical Abstract

INTRODUCTION

The introduction of stereogenic centers is one of the ultimate challenges in organic synthesis.¹ To access different stereoisomers of a complex structure, independent synthesis routes are often required. Alternatively, direct site-selective stereochemical modification could enable the single-step interconversion of one stereoisomer to another. Isomerization by C–H bond cleavage and re-formation at stereogenic centers provides a powerful approach for stereochemical modification.² Pioneering studies by Roberts^2c and Bertrand,^2d and recent examples by Chen,^2e demonstrated radical-mediated epimerization through reversible hydrogen abstraction.

Recently, photoredox catalysis has emerged as a platform for the selective isomerization of stereoisomers.^{3, 4} The generally mild photocatalytic conditions enable stereochemical transformations in late-stage settings. Out-of-equilibrium products can be obtained by relying on the kinetic susceptibility of radical intermediates,⁵ as demonstrated by seminal examples from Knowles and Miller^5d and Wendlandt.^5e In addition to kinetic isomerization, photocatalysis can also promote isomerization to more stable stereoisomers,⁶ as demonstrated by the highly diastereoselective epimerization of piperidines^6a and lactams^6d reported by our lab, and selective epimerization of diols reported by the MacMillan^6b and Wendlandt^6c groups. The photocatalytic stereoselective epimerization of piperidines is most relevant to the current work (Scheme 1a). The transformation proceeds by thiyl radical mediated, reversible hydrogen atom transfer (HAT) at the α-amino site of piperidines. This approach has immediate practical benefits because it enables rapid access to more stable piperidine isomers from stereoisomers that are more easily synthesized. We hypothesized that this approach could be applied to other important classes of saturated nitrogen heterocycles.

Scheme 1.

Light-mediated epimerization of saturated nitrogen heterocycles

Morpholines and piperazines are two important classes of saturated nitrogen heterocycles that feature heavily in natural products and pharmaceuticals with diverse biological properties.⁷ The development of robust and versatile synthetic methods to access these compounds has been intensively pursued.⁸ However, many of the stereoselective routes for the preparation of morpholines⁹ and piperazines¹⁰ containing two or more stereogenic centers provide the less stable stereoisomers. In contrast, rapid access to the thermodynamically favored isomers is often challenging. By employing a reversible photocatalytic epimerization approach, the more stable morpholine and piperazine diastereomers theoretically should be accessible from more readily prepared but less stable stereoisomers.

Herein, we report efficient and synthetically useful protocols for accessing the more stable stereoisomers of morpholines and piperazines by a reversible HAT approach (Scheme 1b). A range of readily accessible morpholines and piperazines bearing various functionalities and substitution patterns were shown to undergo stereo-interconversion to afford their more stable counterparts with high selectivity. Notably, while mechanistic studies indicate that morpholine epimerization proceeds by a thiyl radical mediated, reversible HAT pathway that had been previously reported for piperidine epimerization, the epimerization of piperazines is more complex. Based on these investigations, conditions were developed for highly stereoselective thiol-free epimerization of N,N’-dialkyl piperazines. Mechanistic studies for thiol-free epimerization support a previously unrecognized radical propagation pathway after photocatalytic oxidation of the piperazine to an amine radical cation.

RESULTS AND DISCUSSION

One of the most robust and versatile synthetic routes to highly substituted morpholines is accomplished by first coupling commercially available amino alcohols with α-chloroacyl chlorides (Scheme 2).^{9a, 9b} Treatment of the amide intermediate with base results in epimerization at the α-chloro stereogenic center with selective cyclization to the syn morpholinone as rigorously determined by NMR analysis.^9a.9b Reduction then affords the syn morpholine. In all examples, only the all-syn stereoisomer is obtained and is isolated in excellent yield for the three-step sequence.

Scheme 2.

Stereoselective synthesis of morpholine derivatives

We applied our lab’s previously identified reaction conditions for piperidine epimerization to morpholines (1 mol % of [Ir{dF(CF₃)ppy}₂(dtbpy)]PF₆¹¹ and 1 equiv of PhSH in methanol under blue LED irradiation). The enantiopure isomer (2R,5R)-1a-syn was readily epimerized to (2R,5S)-1a-anti in high yield and with high diastereoselectivity (Table 1). Other HAT mediators and solvents were explored, but PhSH in MeOH was optimal (Table S1 in the Supporting Information). In addition, for 1a-syn, only a minimal reduction in yield and diastereoselectivity was observed when 10 mol% of PhSH was utilized (entry 3, Table S1). We established that the product had been obtained in enantiomerically pure form by epimerizing its antipode, (2S,5S)-1a-syn, followed by chiral HPLC analysis. Formation of the anti isomer without racemization established that selective HAT mediated epimerization occurs at the α-amino C–H bonds without any epimerization at the tertiary α-alkoxy site.

Table 1.

Epimerization of morpholines^a

^aIsolated yields on 0.3 mmol scale; dr was determined by ¹H NMR analysis. Crude yields and dr are noted in parentheses and were determined by ¹H NMR analysis with an internal standard.

^b5 equiv PhSH instead of 1 equiv.

^cThe crude yield and dr is 60%, 88:12 dr when using 1 equiv PhSH.

^d5 equiv of CySH instead of PhSH.

^e10 equiv of methyl thioglycolate instead of PhSH.

^fCySH instead of PhSH.

^g70 h reaction time.

^hMeCN as solvent.

ⁱ4-CzIPN instead of [Ir{dF(CF₃)ppy}₂(dtbpy)]PF₆.

^j3 equiv CySH instead of PhSH.

^k[Ir(ppy)₂(dtbbpy)]PF₆ instead of [Ir{dF(CF₃)ppy}₂(dtbpy)]PF₆.

^l40 h reaction time.

^m10 equiv of CySH instead of PhSH.

Next, a range of morpholine derivatives was evaluated. Derivatives with electron-deficient and electron-rich aromatic rings were effective substrates, providing the anti stereoisomers in high yields and diastereoselectivities (1c-anti-1h-anti). However, epimerization of the highly electron-deficient morpholine 1f-anti under the standard reaction conditions only gave moderate yield and stereoselectivity due to competitive side reactions (see footnote c in Table 1). The decomposition pathways can be minimized by employing excess thiol, presumably because a higher thiol concentration increases the rate of HAT relative to undesired reaction pathways.^2d Additionally, the corresponding N-methyl (1g) and benzyl (1h) morpholine derivatives epimerized in high yields and selectivities without detectable decomposition.

Highly stereoselective epimerization could also be achieved for morpholines with α-alkyl amine substituents with varying steric demands (1i-anti-1k-anti). Consistent with epimerization of alkyl substituted piperidines,^6a CySH, with a stronger S–H bond than PhSH,¹² is necessary for polarity matching with the stronger α-amino C–H bond of alkyl substituted morpholines. The tert-butyl (1i-anti) and cyclohexyl (1j-anti) substituted morpholines were converted to the anti products in excellent yields and stereoselectivities. However, epimerization of the less sterically encumbered isobutyl-substituted morpholine (1k) was slower and not quite as selective. Similar to α-aryl substituted morpholines, the enantiomerically pure α-alkyl substituted morpholines epimerized without racemizing the tertiary α-alkoxy stereocenter based upon deuterium incorporation studies (vide infra).

Morpholines with N-alkyl substituents were effective substrates, affording the more stable isomers in excellent yields and diastereoselectivities (1l-anti-1n-anti). The epimerization of N-phenyl substituted morpholine proceeded slowly to afford 1o-anti with only moderate stereoselectivity, consistent with previous results for N-phenyl substituted piperidine derivatives.^6a

In addition to the 2,5-substitution pattern, epimerization was also effective for 2,3-substituted morpholines with aryl (1p) and alkyl (1q) substituted morpholines isomerizing to the anti products in good yields and selectivities. Additionally, the α-methyl substituted morpholine 1r epimerized with moderate anti selectivity. However, some racemization of stereocenters at the 2-position likely occurs for 2,3-substituted morpholines due to a reversible enamine formation pathway (vide infra).

To demonstrate the generality of this methodology, we next investigated piperazine derivatives. One of the most popular and efficient routes for preparing piperazines proceeds by synselective hydrogenation to provide the less stable diastereomer (Schemes 3).^10b

Scheme 3.

Efficient synthesis of piperazine derivatives

The readily prepared but less stable α-phenyl (2a) and α-cyclohexyl (2b) syn piperazines epimerized to the more stable anti products in high yields and good stereoselectivities (Table 2). However, for piperazine derivatives, methyl thioglycolate was a slightly more effective HAT reagent than PhSH. Piperazines with para-methoxy (2c) and meta-amino (2d) electron-donating aromatic ring substituents were effective substrates, providing the anti isomers with high selectivities though in moderate yields. The 2,3,5-trisubstituted piperazine 2e was also efficiently converted to the anti isomer in 99% yield and 94:6 dr. However, 2,3,5,6-tetrasubstituted piperazines were not explored because efficient methods for their synthesis are not available.

Table 2.

Epimerization of piperazines^a

^aSee footnote a in Table 1.

^b10 equiv of 1-BuSH instead of methyl thioglycolate.

^c40 h reaction time.

^dPhSH instead of methyl thioglycolate.

^e70 h reaction time.

^f10 equiv of thiol instead of 1 equiv.

^gMeCN (0.04 M) as reaction solvent.

^hp-Toluenethiol instead of methyl thioglycolate.

ⁱ3 equiv of thiol instead of 1 equiv.

A number of N,N’-dimethyl piperazines were also explored (2f-2q). N,N’-Dimethyl piperazines with bulky substituents were efficiently epimerized to the more stable isomers as demonstrated for isopropyl (2g) and ortho-tolyl (2h) substituted derivatives. Additionally, for N,N’-dimethyl piperazines, various electron-withdrawing functionalities, such as ester (2l), secondary (2m) and tertiary (2n) amides, and an unhindered pyridyl group (2o) were compatible with the standard epimerization conditions. In contrast, as observed for the highly electron-deficient morpholine derivatives (see footnote c, Table 1), decomposition was seen for N-unsubstituted piperazines with electron-deficient aryl substituents (data not shown). Even with excess thiol, decomposition outcompetes the desired HAT epimerization pathway. The anti product was also obtained in high yield and diastereoselectivity for 2,3-diphenyl substituted piperazine (2p). In addition to 2,3-substitution patterns, 2,5-substituted N,N’-dimethyl piperazine 2q is an effective substrate, providing the more stable isomer in high yield and good selectivity.

A piperazine with N,N’-diphenyl substituents (2r) was also evaluated, affording the anti product in high yield and moderate diastereoselectivity, consistent with that previously observed for N-phenyl substituted morpholine 1o and piperidines.^6a

Along with the syn piperazines, synthetically accessible but less stable anti diastereomers were also investigated as potential substrates for epimerization. Anti 2,6-substituted piperazine derivatives were readily prepared by stereoselective α-C–H difunctionalization of piperazines¹³ and epimerized to the more stable isomers 2s-syn and 2t-syn in high yields and good stereoselectivities.

Racemic piperazines 2 were evaluated in our epimerization studies. However, we postulate that epimerization of enantiomerically pure derivatives of 2 would result in racemization because the stereocenters are inherently located α- to one of the nitrogens in the piperazine, and therefore, should be capable of undergoing epimerization. This conclusion is supported by the deuterium incorporation experiments shown in Figure 3b (vide infra).

Figure 3.

Mechanistic studies on piperazine 2a. (a) Luminescence quenching for 2a-syn. (b) Deuterium labeling experiment of 2a.

We further extended the substrate scope of epimerization by evaluating a N-Boc protected piperazine (Figure 1). The N-Boc, N’-methyl substituted piperazine 2u-syn epimerizes but with poor conversion to the anti isomer (Figure 1a). Interestingly, epimerization of the diastereomerically pure anti isomer 2u-anti provided an anti:syn ratio of 35:65 (Figure 1b). We believe the preference for the syn isomer for N-Boc protected piperazine 2u is due to unfavorable A^1,3-strain interactions between the N-Boc group and the α-phenyl substituent when it is placed in the equatorial position.¹⁴

Figure 1.

Epimerization of N-Boc piperazines. The yield and dr of the reactions were determined by ¹H NMR analysis of the unpurified product with an internal standard. (a) Epimerization of 2u-syn. (b) Epimerization of 2u-anti.

To investigate the mechanism of epimerization, we first computed the relative free energies of select morpholine and piperazine derivatives using density functional theory (DFT) calculations (Table 3). The observed distributions of diastereomers for the overall transformation correlate with the calculated relative stabilities of the diastereomers, indicating that under the photocatalytic reaction conditions, a thermodynamic ratio of products is generally obtained.

Table 3.

Experimental and calculated relative energies

entrya	substrate	exptl. dr anti/syn	exptl. ΔGsyn − ΔGanti (kcal/mol)	calcd. ΔGsyn − ΔGanti (kcal/mol)
1	1a	96:4	1.9	2.4
2	1j	97:3	2.0	1.3
3	1l	95:5	1.8	2.5
4	1p	99:1	>2.5	3.9
5	2a	95:5	1.8	2.7
6	2b	84:16	1.0	2.0
7	2e	94:6	1.6	3.7
8	2f	97:3	2.1	2.8

^[a]ωB97X-D/def2-TZVPP, SMD (MeOH)//ωB97X-D/def2-SVP, SMD (MeOH). exptl. = experimental; calcd. = calculated

Upon re-subjecting the stereoisomerically pure piperazine 2b-anti to the epimerization conditions (Scheme 4), the mixture of syn and anti isomers obtained was in close agreement with the distribution obtained starting with 2b-syn. This result supports that epimerization is a reversible process that provides an equilibrium mixture of diastereomers under the reaction conditions.

Scheme 4.

Thermodynamic equilibrium demonstration

A series of mechanistic studies were next carried out, first with an investigation of morpholine 1a (Figure 2). Luminescence quenching experiments revealed that the mixture of both PhSH and morpholine 1a-syn quenched the photocatalyst excited state 51-fold faster than either the thiol or substrate alone (Figure 2a). In addition, deuterium labeling studies were performed in methanol-d₄, which results in rapid exchange of PhSH to PhSD given the large excess of methanol-d₄. Under these conditions, complete deuterium incorporation at the 5-position of morpholine 1a-anti was observed (Figure 2b), suggesting HAT occurs solely at the weakest α-amino C–H bond.¹⁵ In contrast, for tert-butyl substituted morpholine 1i-anti where CySH with a stronger S–H bond was necessary for epimerization (Figure 2c), complete deuterium incorporation still occurred at the 5-position along with significant deuterium incorporation at the 3-position. These results reflect the more modest difference in α-amino C–H bond strengths at the 5- and 3-positions of this substrate. Collectively, these results are consistent with previous investigations on piperidine epimerization^6a and support a mechanism involving the quenching of a photocatalyst excited state by the in situ generated thiol anion and reversible HAT between a thiyl radical¹⁶ and the saturated nitrogen heterocycle (Figure 2d). It should be noted that deuterium incorporation occurred at the 6-position of morpholines 1a-anti and 1i-anti (Figure 2b–c). This result supports a reversible enamine formation pathway (Figure 2d), which had been previously documented for piperidines through a series of mechanistic experiments.^6a

Figure 2.

Mechanistic studies and proposed mechanism for morpholines. (a) Luminescence quenching for 1a-syn. (b) Deuterium labeling of 1a. (c) Deuterium labeling of 1i. (d) Proposed mechanism.

Mechanistic studies were next performed on piperazine 2a (Figure 3). Luminescence quenching experiments for piperazine 2a-syn revealed that the mixture of thiol and substrate quenched the photocatalyst excited state only 8-fold faster than for piperazine 2a-syn alone, which suggested appreciable competitive direct oxidation of the piperazine under the reaction conditions (Figure 3a). When deuterium labeling studies were performed in methanol-d₄, piperazine 2a-anti was obtained with only moderate deuterium incorporation at the two tertiary α-amino sites along with ~11% deuteration at the methyl site (Figure 3b). Moreover, MS analysis of the deuterium labeled product established that >50% of the product was not deuterated with a significant percentage of product multiply deuterated. We believe that the thiyl radical mediated HAT mechanism shown in Figure 2d occurs. However, the appreciable direct piperazine oxidation established by luminescence quenching along with the significant percentage of nondeuterated product relative to the observed 93:7 dr of the 2a-anti, suggest that an additional epimerization pathway must be operative.

Due to the interesting results for N-unsubstituted piperazine 2a, we next conducted mechanistic studies with N,N’-dimethyl piperazine 2f (Figure 4). Luminescence quenching experiments for N,N’-dimethyl piperazine 2f-syn established that 2f-syn efficiently quenched the photocatalyst excited state¹⁷ (kq = 1.17 × 108 M-1 s-1) (Figure 4a), presumably by electron transfer between *Ir^III and 2f-syn to generate an amine radical cation.¹⁸ Notably, in the presence of both thiol and N,N’-dimethyl piperazine 2f-syn, the quenching kinetics (kq = 1.73 × 10⁸ M⁻¹ s⁻¹) were only slightly faster than for the N,N’-dimethyl piperazine 2f-syn alone. Next, deuterium labeling of N,N’-dimethyl piperazine 2f was carried out in methanol-d₄ resulting in 39% and 51% deuteration at the two tertiary α-amino sites along with lower amounts of deuteration at multiple other sites (Figure 4b). MS analysis indicates that ~25% of the product is not deuterated, which given the excellent conversion to the 2f-anti (>99%, 97:3 dr), is inconsistent with a mechanism that proceeds solely by a thiyl radical mediated HAT.

Figure 4.

Mechanistic studies on N,N’-dimethylpiperazine 2f. (a) Luminescence quenching for 2f-syn. (b) Deuterium labeling of 2f.

We hypothesized that due to the efficient photocatalytic oxidation of N,N’-dimethyl piperazine, thiol independent epimerization might be possible. Indeed, in the absence of thiol, comparably high yields and stereoselectivities were achieved for N,N’-dimethyl piperazine 2f and its derivatives bearing tolyl (2h) and pyridyl (2o) substituents (Table 4). It is noteworthy that compared to the standard conditions, eliminating the thiol reagent simplifies the reaction setup and purification.

Table 4.

Epimerization of N-methyl piperazine in the absence of thiol.

To better understand this simplified reaction, a series of mechanistic studies were performed for N,N’-dimethyl piperazines in the absence of thiol (Figure 5). First, deuterium labeling of N,N’-dimethyl piperazine 2f-syn was carried out in methanol-d₄ without thiol (Figure 5a). No deuterium (<5%) was detected at any site in 2f-anti after epimerization. The lack of deuterium indicates that after electron transfer with *Ir^III, the resulting amine radical cation is unlikely to epimerize by reversible deprotonation and re-protonation of the acidic α-amino C–H bond, in stark contrast to the previously proposed mechanism for the epimerization of N-arylated piperidines.^4d Rather, the deuterium labeling results suggest that another species is acting as a hydrogen atom shuttle for epimerization of the of N,N’-dimethyl piperazines. To investigate if this alternative epimerization pathway proceeds by a radical chain process,^2c,19 the quantum yield (Φ) was calculated to be 20 (see Supporting Information) and is consistent with a product-forming chain mechanism. Additional control experiments were performed to understand the necessity of light (Figure 5b). No reaction was detected when the epimerization was run in the dark. When the reaction mixture was irradiated for 2 min, a 14% yield of 2f-anti was observed. Interestingly, when the reaction was irradiated for 2 min, then kept in the dark for 18 h, 57% conversion to 2f-anti was achieved. This outcome establishes that the reaction continues in the dark, providing further support for a radical chain mechanism. However, the moderate epimerization efficiency suggests that the radical chain is not so efficient, and therefore, constant light irradiation is necessary to achieve complete epimerization.²⁰

Figure 5.

Mechanistic studies and proposed mechanism for N,N’-dimethylpiperazine in the absence of thiol. (a) Luminescence quenching for 2f-syn. (b) Deuterium labeling experiment of 2f. (c) Proposed mechanism for epimerization of N,N’-dimethyl piperazines without thiol.

Based on these collective results, we propose the following mechanism for epimerization of N,N’-dimethyl piperazines in the absence of thiol (Figure 5c): after light irradiation, the photocatalyst excited state is reduced by N,N’-dimethyl piperazine, providing the amine radical cation. Then, deprotonation of a greatly acidified α-amino proton affords the α-amino radical, which initiates a radical propagation cycle by undergoing HAT with another equivalent of piperazine to provide the more stable isomer and another α-amino radical that in turn further propagates the radical chain to drive the epimerization process. We postulate that HAT occurs between the piperazine substrates and not with the solvent methanol due to its strong C–H and O–H bonds.¹² In support of this hypothesis, when methanol-d₄ was used as solvent, no deuterium was incorporated (see Figure 5a). Lastly, the amine radical cation oxidizes Ir^II to complete the catalytic cycle.

Compared to thiol mediated transformations, epimerization without thiol has a more limited scope. The isopropyl substituted N,N’-dimethyl piperazine 2g-syn did not undergo epimerization (Table 5), suggesting that the steric profile of the reactant is an important factor for epimerization in the absence of thiol. Moreover, while the piperazine without nitrogen substituents, 2a-syn, effectively quenched *Ir^III with a quenching constant of 6.09 × 10⁷ M⁻¹ s⁻¹ (see Figure 3a), decomposition occurred rather than epimerization when thiol was not present. Finally, because morpholine 1a-syn is inefficiently oxidized to the amine radical cation due to its very low quenching rate (Figure 2a), it is not surprising that no reaction was observed for 1a-syn in the absence of thiol. Consistent with this result, a morpholine with an N-methyl substituent (1l-syn) did not undergo epimerization in the absence of thiol.

Table 5.

Limitations of epimerization in the absence of thiol.

CONCLUSION

In summary, we report the first example of visible light-mediated epimerization of morpholine and piperazine derivatives, which greatly expands upon the previously reported photocatalyzed epimerization of piperidines. With this approach, the more stable morpholine and piperazine diastereomers can be readily prepared from their more synthetically accessible isomers, providing a powerful tool for editing the stereochemical configurations of these saturated nitrogen heterocycles. Mechanistic studies revealed that two HAT pathways are plausible for N,N’-dimethyl piperazines. In addition to a thiyl radical mediated HAT pathway, the epimerization can proceed in the absence of a thiol by a previously unrecognized radical chain propagation cycle. Key to the success of this pathway are the dual functions of the N,N’-dimethyl piperazine substrate as both a single electron donor and hydrogen atom shuttle. The epimerization approaches reported here significantly expand the toolbox for the stereoselective synthesis of morpholines and piperazines and should be broadly applicable to additional classes of saturated nitrogen heterocycles.