Photoinduced Nickel-Catalyzed Selective N-Demethylation of Trialkylamines Using C(sp²)–Bromides as HAT Reagents

Abstract

N-Demethylation of trialkylamines is a useful transformation, but typically requires harsh reaction conditions and stepwise procedures, as well as judicious protection of labile functional groups. Herein we report a mild, catalytic approach for the demethylation of trialkylamines by utilizing photoinduced nickel catalysis wherein C(sp²)–bromides serve as hydrogen-atom transfer (HAT) reagents. This method achieves direct demethylation of trialkylamines with wide functional group compatibility making it highly suitable for late-stage derivatization of complex molecules. Mechanistic investigations provide evidence that C(sp²) radicals generated via photoinduced Ni─C(sp²) bond homolysis are involved in hydrogen atom abstraction from trialkylamines. Utilizing steric control of the C(sp²)–bromides, our HAT approach achieves demethylation with excellent site selectivity in the presence of benzylsubstituted amines, which is complementary to the selectivity of classical approaches that afford debenzylation product instead.

Graphical Abstract

Trialkylamines are important motifs in alkaloid natural products and pharmaceutical compounds.¹ The removal or replacement of N-methyl substituents is a useful transformation that has been employed in drug synthesis and in the development of tools for pharmacological and drug metabolic studies.^2,3 The classical and frequently employed approaches for N-demethylation of tertiary amines include the von Braun Reaction, the use of chloroformate reagents (Scheme 1A), and the Polonovski reaction, which are two-step protocols involving a cyanamide, carbamate, or N-oxide intermediate respectively, and require harsh reaction conditions (high temperature, employment of toxic reagent such as cyanogen bromide, or a strong oxidant), with limited functional group tolerance and poor site-selectivity.^4,5 Other approaches such as biochemical,^6,7 photochemical,^8-11 electrochemical,^12,13 and transition-metal catalyzed^14,15 methods have also been reported, albeit are restricted to specific structures and lack broad applicability.

Scheme 1.

Reaction Design

Over the past decade, the merger of photoredox and nickel catalysis has enabled novel disconnection approaches and new reaction pathways, especially broadening cross-coupling reactions.¹⁶ Previously, our group developed a nickel/photoredox dual catalytic platform for selective C(sp²)─C(sp³) N─Me arylation of trialkylamines under mild conditions, including in the context of late-stage functionalization of pharmaceutical compounds.¹⁷ In an effort to expand the scope of late-stage derivatization of trialkylamine scaffolds, we questioned whether it would be possible to employ C(sp²)–bromides as hydrogen-atom transfer (HAT) reagents instead of coupling partners. In recent years, photoelimination (photoinduced ligand dissociation) of a Ni─Cl or Ni─Br bond has been applied in C─H cross-coupling reactions,^18,19 where C(sp²)–chlorides^20-23 or C(sp²)–bromides^24-26 serve as both coupling partners and the chlorine/bromine radical source for hydrogen atom abstraction. Inspired by Doyle’s²⁷ and Hadt’s²⁸ recent mechanistic investigations on photoinduced Ni─aryl bond homolysis, we envisioned the use of aryl and vinyl bromides as a C(sp²) radical source for hydrogen abstraction from the α-nitrogen position followed by an oxidation event with nickel to achieve N-demethylation of trialkylamines (Scheme 1B). This HAT approach for demethylation would allow complementary site selectivity compared to the classical chloroformate approach, where substrates such as benzyl amines are not tolerated since chloride attacks the most electrophilic benzylic position (Scheme 1A, bottom). In contrast to Ready’s recent functionalization of alkyl/benzyl amines via photoredox chemistry governed by differences in acidity and bond dissociation energy (BDE),²⁹ in our catalytic system, the relative ambiphilicity of aryl radicals^30,31 would allow abstraction of hydrogen from the most accessible methyl position instead, and the selectivity for demethylation could be further improved via steric control by tuning the sterics of C(sp²)-bromides.

We began our investigation of N-demethylation of trialkylamines with the introduction of water as a nucleophile to the photoredox/nickel catalytic system. Gratifyingly, after optimization we observed exclusive N-demethylation product 2, alongside reduced arene 3, the product that would form after aryl radical formation and HAT. Further screening revealed the optimized conditions to include 4-bromobenzonitrile Br1 (1.1 equiv), NiCl₂(H₂O)₆ (5 mol%), organic photocatalyst 5CzBN (1 mol%),^32,33 NaOPiv (1.0 equiv) as base, and H₂O (36 μL, 10 equiv) in acetonitrile (0.1 M) under blue LEDs irradiation (427 nm), providing 2 in 66% yield (Table 1, entry 1). Using excess of aryl bromide (2 equiv) decreases the yield, with homocoupling of ArBr observed as side pathway (Table 1, entry 2). Changing aryl bromide into vinyl bromide Br2 affords comparable yield, generating propene as a low-molecular weight byproduct, thus increasing atom efficiency (Table 1, entry 3). Subsequent one-pot Boc protection was performed for facile isolation of the secondary amine, affording a 61% isolated yield of the product (Table 1, entry 3). Base was shown to be unnecessary but promotes the reaction (Table 1, entry 4). The reaction also works well with Ni(cod)₂ in the absence of chloride sources (Table 1, entry 5), eliminating the necessity of Cl radical as HAT reagent in our system.¹⁸ Control studies showed that nickel, light, photocatalyst, and aryl bromide are all necessary, as their removal results in no desired product (Table 1, entries 6-9).

Table 1.

Optimization and Control Studies^a

Entry	Deviation	Yield 2 (%)b	Ar-H 3 (%)b
1	none	66	92
2	2 equiv Br1	58	99
3	Br2 instead of Br1	63 (61)c	/
4	no base	27	48
5	Ni(cod)2 instead of NiCl2(H2O)6	59	86
6	no Ni/dtbbpy	0	21
7	in the dark	0	0
8	no photocatalyst	0	2
9	no aryl bromide	0	/

^aOptimizations were performed on a 0.2 mmol scale.

^bGC-MS yield with 1,3,5-trimethoxybenzene as external standard.

^cYield in bracket refers to isolated amine products after subsequent one-pot Boc protection with NEt₃ (2 equiv) and Boc₂O (1 equiv) followed by purification using flash chromatography.

After the development of N-demethylation conditions, we then examined the site selectivity of our method with benzylamines as substrates. The classical approach for demethylation has poor tolerance of substrates such as benzylsubstituted amines, affording benzyl chloride as the debenzylation product exclusively in 83% yield instead of the demethylation product when methyl chloroformate is employed (Table 2A).³⁴ In contrast, our method provides demethylation of N,N-dimethylbenzylamine 4a as the major product, achieving selectivity of 4:1 to the debenzylation product (benzaldehyde) when using bromobenzonitrile Br1, or 2:1 when using the smaller vinyl bromide Br2 (Table 2B). The change in selectivity depending on size of the C(sp²)–bromide used inspired us to utilize steric control³⁵ to further improve selectivity for N-demethylation. Gratifyingly, bulky TRIP-Br (2,4,6-triisopropylbromobenzene) Br3 furnishes the demethylated amine with 7:1 selectivity. Considering the slow oxidative addition with sterically-hindered aryl bromides, we premixed Br3 with Ni(0) source Ni(cod)₂ prior to irradiation (with a noticeable color change to orange-brown and a characteristic peak at δ 9.33 (d, J = 6.0 Hz, 1H) in ¹H NMR as evidence for NiBr(TRIP)(dtbbpy) formation),³⁶ which increases the yield to 51%, and interestingly, also improves selectivity to >20:1. Next, we demonstrated the generality of our approach with a representative panel of benzylamines. In most cases the bulky TRIP-Br Br3 leads to better site selectivity of demethylation compared to the smaller vinyl bromide Br2. In general, electron-deficient benzylamines (7-9) show higher selectivity than electron-rich benzylamines (5, 6, and 10), which we postulate is due to more favorable abstraction of weaker and more hydridic benzylic C(sp³)─H bonds when benzylic positions are more electron-rich.¹⁸ In addition, benzylamine 13 possessing a cyclohexyl group also provides selective demethylation even in the presence of a weaker tertiary C(sp³)─H bond. Furthermore, Br3 combined with Ni(cod)₂ affords excellent selectivity of >20:1 and good yield for all the examples, except benzylamines 11 and 12 bearing sterically-hindered benzylic positions which give higher or comparable yield with less sterically encumbered Br2, with no loss of enantiopurity at the benzylic center.

Table 2.

Selective N-Demethylation over N-Debenzylation via Steric Control of C(sp²)-Bromides^a,c

^aReactions were run on a 0.2 mmol scale at ambient temperature in MeCN (0.1 M) using 5CzBN (1 mol%), Ni (5 mol%), dtbbpy (8 mol%), C(sp²)-Br (1.1-1.5 equiv), NaOPiv (1 equiv) and H₂O (36 μL) under 8-30 h of blue LEDs irradiation (427 nm) until complete conversion of the amine substrate. Yields of secondary amines and ratios were determined by GC-MS with 1,3,5-trimethoxybenzene as standard. Yields in parentheses refer to isolated amine products after subsequent one-pot Boc protection with NEt₃ and Boc₂O.

^bBr2 and Ni(cod)₂ was used.

^cTrace side products such as primary amine and imine were observed as a result of over-oxidation, along with unreacted substrate. Driving the reaction to completion also results in overoxidation by-products.

Having identified conditions that enable high selectivity for N-demethylation, we then sought to study the reaction mechanism. Direct excitation with 390 nm UV light affords the demethylation product without the participation of photocatalyst, which demonstrates that an oxidation/deprotonation pathway¹⁷ to the α-amino radical is not dominant (Scheme 2B). On the other hand, we noted that the reaction features a strong dependence on sodium pivalate as the base to achieve efficient demethylation. Indeed, a minor byproduct observed is C─O coupling between the aryl bromide and pivalate, suggesting that pivalate displacement of bromide is operative in the reaction. This led us to employ sodium adamantane-1-carboxylate instead of sodium pivalate to track the possible formation of a tertiary carbon-centered radical that could mediate HAT. However, no admantane is detected, rendering the possibility of carboxyl radical generation via Ni–carboxyl bond homolysis unlikely (otherwise adamantane should be observed as the rate of decarboxylation (k ≈ 10⁹ s⁻¹)³⁷ is orders of magnitude faster than HAT (k ≈ 10⁷M⁻¹s⁻¹)³⁸). Furthermore, an aryl radical is trapped by N-methylpyrrole with nickel catalyst under 390 nm light, and control studies showed that light is necessary for the generation of aryl radical (Scheme 2C).

Scheme 2.

Mechanistic Investigations

To eliminate the potential involvement of bromine radicals generated via photoelimination of Ni(II)─Br bond as HAT reagents,^24,25 we perfomed a control study without base using benzylamine 5, and achieved debenzylation instead of demethylation providing benzaldehyde 5b as the major product (Scheme 2D), which is a result of Br radical (BDE (H─Br) = 87.5 kcal/mol) abstracting the weak benzylic C─H bond (BDE = 83 kcal/mol). For comparison, in the presence of sodium pivalate, demethylation is provided as the major product with site-selectivity of 14:1. It is possible that pivalate displacement of a halide ligand on nickel is important for blocking halogen radical photoelimination pathways, instead liberating aryl radicals selectively, and that the involvement of aryl radical is responsible for the excellent site-selectivity of demethylation. Therefore, we propose a mechanistic pathway where an α-amino radical is generated upon hydrogen atom abstraction from the trialkylamine involving aryl radical (Scheme 2A), which is formed via photoelimination of Ni(II)─Ar bond of Ni(II)ArOPiv complex (Ni-3).^27,28 Interception of the α-amino radical by Ni(I)-OPiv intermediate (Ni-4) affords a Ni(II) intermediate (Ni-5), which is in equilibrium with the iminium ion (Ni-6),^17,39 providing N-demethylation product and formaldehyde upon hydrolysis.

Next, we investigated the role of photocatalyst under 427 nm visible light. Employment of photocatalysts that are less oxidizing than 5CzBN (*E_red = 1.31 V vs SCE; E_T = 65.3 kcal/mol),³² including Ir(ppy)₃ (*E_red = 0.31 V vs SCE; E_T = 55.2 kcal/mol) and Ir(dFppy)₃ (*E_red = 0.77 V vs SCE; E_T = 63.5 kcal/mol) also improves the yield compared to using no photocatalysts under 427 nm light (Scheme 2E), which lends additional evidence to an energy transfer (EnT) pathway proceeding via triplet sensitization of Ni(II)ArOPiv to a ³d-d excited state (Ni-3*), instead of single electron transfer (SET), although we cannot rigorously exclude the latter.

The improved selectivity of demethylation vs. debenzylation observed with Ni(cod)₂ compared to NiCl₂ also captured our interest. Introducing chloride into the Ni(cod)₂ system through the addition of exogenous tetrabutylammonium chloride (TBACl, 10 mol%) results in excellent selectivity of demethylation similar to the system with Ni(cod)₂, rather than the NiCl₂(H₂O)₆ system (Scheme 2F). We reasoned that an off-cycle pathway could be operative where Cl radical acts as a competitive HAT reagent via a Ni(I)/Ni(III) cycle when a Ni(II) source NiCl₂(H₂O)₆ is employed (Scheme 2G). This pathway can be induced by the reduction of NiCl₂ with the reduced state of the photocatalyst while trialkylamine serves as a sacrificial reductant, generating Ni(I)-Cl (Ni-7). Oxidative addition of ArBr leads to a Ni(III) intermediate (Ni-8), followed by photoelimination to afford Cl radical. In addition to selective abstraction of methyl C(sp³)─H bond enabled by the C(sp²) radical (Scheme 2A), the smaller and much more electrophilic Cl radical abstracts hydrogen from benzylic positions competitively, thus lowering the selectivity for demethylation when NiCl₂(H₂O)₆ is employed. The high bond strength of HCl (BDE = 102 kcal/mol), in addition to the stoichiometry of abstractable α-amino C─H bonds (2 benzylic: 6 methyl) could also be consistent with the involvement of chlorine radicals in HAT.²²

With rationalizations of the reactivity and selectivity in hand, we anticipated that our method would not only be applicable to derivitization of simple benzylamines but also to a broader scope of substrates. Vinyl bromide Br2 was employed for higher atom economy, and in the same pot the amine was Boc-protected at the conclusion of the reaction (Table 3). Trialkylamines bearing α-cyclohexyl and α-cyclopentyl substituent(s) provide selective demethylation (15-17). Cyclic amines such as piperidine decorated with an ethyl ester undergo the reaction in moderate to good yield (18-19), and benzoate is also well tolerated (20).

Table 3.

Scope of Selective N-Demethylation and Late-Stage Applications^a

^aReactions were performed with 0.2 mmol of substrate at ambient temperature in MeCN (0.1 M) using 1 mol% PC (5CzBN), NiCl₂(H₂O)₆ (5 mol%), dtbbpy (8 mol%), Br2 (1.5 equiv), NaOPiv (1 equiv) and H₂O (36 μL) under 12-30 h of blue LED light irradiation (427 nm), followed by adding NEt₃ (2 equiv) and Boc₂O (1 equiv). Yields refer to isolated products after purification by flash chromatography. Regioselectivities were determined by isolated yield.

^bTroc-Cl (1 equiv) was used in place of Boc₂O.

^cTRIP-Br Br3 (1.1 equiv) and Ni(cod)₂ was used in place of Br2 and NiCl₂(H₂O)₆ for improved site selectivity.

^dThe ratios in parentheses refer to site selectivity between the methyl position and position(s) highlighted in grey determined by crude ¹H NMR.

Additionally, we performed late-stage functionalization on commercially available drugs and natural products containing trialkylamine scaffolds and other potential reactive sites (21-30), providing moderate to good yields with excellent selectivity.⁴⁰ Functional groups such as alcohol (22) are tolerated with our method, while the traditional chloroformate approach requires extra protection and deprotection steps,⁴¹ enabling novel approaches to protecting group strategies in complex target synthesis. Furthermore, our approach opens up previously inaccessible secondary amine functionalities on biologically active scaffolds for various late-stage transformations such as N-arylation, amide coupling, or reductive amination, as well as provides new exit vectors for the synthesis of various novel therapeutic modalities such as protein degraders, drug-conjugates, and fluorescence labeling.⁴²

In conclusion, we have developed a photoinduced Ni-catalyzed approach to the N-demethylation of trialkylamines using C(sp²)–bromides as hydrogen-atom transfer (HAT) reagents. The reaction displays excellent selectivity for N-methyl C(sp³)─H bonds, and mechanistic investigations indicate a C(sp²) (vinyl or aryl) radical, rather than a halogen radical, mediated HAT to provide the excellent selectivity observed. As such, our approach offers an opportunity to complement traditional protocols for demethylation under milder conditions and with better functional group compatibility, and we anticipate that this strategy can be used to rapidly diversify amine analogs in library synthesis.