A hydrazine-free photoredox catalytic synthesis of azines by reductive activation of readily available oxime esters

Abstract

Herein, we present a novel, hydrazine-free photoredox catalyzed platform for azine synthesis using mild, simple reaction conditions. While previous energy transfer activations of oxime esters lead to decarboxylation of the O-auxiliary and radical combination with the iminyl radical, the reductive electron transfer strategy herein affords high yields of azines using only a triarylamine organophotocatalyst and no additives. Scale up was readily achieved by means of a continuous flow reactor. Mechanistic studies indicate a preassembly of photocatalyst and substrate is key to achieving efficient and selective N–N iminyl radical coupling.

---

Introduction

Azines (N–N linked diimines) are versatile building blocks in organic and pharmaceutical chemistry.¹ They can undergo unusual [1 + 3] dipolar cycloaddition with dienophiles, but also participate in [2 + 3] cycloadditions as an ene fragment to give valuable nitrogen containing heterocycles.² Additionally, they are precursors to isoquinoline derivatives via Rhodium catalyzed oxidative annulation with alkynes.³ Azines show interesting optical,⁴ biological,⁵ and physical properties;⁶ and find applications as liquid crystals⁷ or in the synthesis of covalent organic frameworks.⁸ Classically, azine synthesis is achieved by condensation of two carbonyl compounds with hydrazine (or via an isolated hydrazone intermediate). To form azines from aldehydes and dialkyl ketones, such condensations can be performed under mild reaction conditions. On the other hand, aryl ketones need higher reaction temperatures, up to 100 °C, for their condensations with hydrazine.⁹ This leads to safety issues, as hydrazine is explosive at higher temperatures and is known to auto ignite at 24 °C on rusty iron, and at 240 °C on a glass surface.¹⁰ Additionally, hydrazine is highly carcinogenic and highly toxic to humans and the environment.¹¹

Therefore, extensive research has prioritized synthetic access to azines under milder and non-toxic reaction conditions. For example, Zanotto and co-workers developed an elegant Pt-catalyzed homodimerization of diazo compounds under mild reaction conditions (Fig. 1A).¹² Subsequently, Guan's group developed a neat Cu-catalyzed azine synthesis from hydroxylamines, using Boc anhydride to generate oxime esters in situ (Fig. 1B).¹³ However, in the context of green chemistry drawbacks remained, such as the need for high temperatures and stoichiometric reductants, as well as the use of transition metals (T.M.s) and unstable/hazardous diazo compounds.

Fig. 1 (A) Platinum catalyzed homocoupling of diazo compounds to give azines.¹² (B) Copper catalyzed homocoupling of hydroxylamines with an external oxidant.¹³ (C) EnT [2 + 2]-cycloaddition mediated under air by 4-fluorobenzoyl as auxiliary, as inspiration.²³ (D) This work.

Since the turn of the century, light-mediated reactions have offered mild and environmentally-friendly alternative reaction conditions to thermal or T. M.-catalyzed processes, that access target compounds by different reactive intermediates. Over the last decade, numerous new synthetic applications involving photocatalysis have emerged, using either energy transfer (EnT) or single electron transfer (SET) redox paradigms.¹⁴ Due to their weak N–O bond¹⁵ and facile preparation from ketones, oxime esters are ideal targets for PRC¹⁶ or EnT mediated iminyl radical generations.^17,18 Particularly, Glorius¹⁷ and Cho¹⁸ pioneered oxime ester activations for EnT N–O bond activations by homolytic cleavages. Upon N–O cleavage, generated N-centered radicals (NCRs) are intercepted by C-centered radicals derived from decarboxylation of the cleaved O-auxiliary. In these reports, azines were detected only as low-yielding side products (up to 11% in the former report¹⁷), while DFT calculations by the groups of You and Cho suggest that N–N bond formation is thermodynamically unfavorable.^18a

Recent years have revealed that the design of an organophotocatalyst (orgPC) bears critical importance on its stability, reactivity and interactions/aggregations with itself or the target substrate. Two urgent aspects of the field are often overlooked; (i) that the ‘photocatalyst’ added to the reaction is often a ‘precatalyst’ that transforms to the active photocatalyst under the reaction conditions;¹⁹ (ii) how orgPC self-aggregation, or assembly with a target substrate prior to photoexcitation, can divert excited state photochemical pathways to afford different products.²⁰ For example, triarylamine (TPA)-type orgPCs are known to interact non-covalently with target substrates via their neutral forms (N lone pair) or via their radical cations, with crucial impacts on reactivity and selectivity.²¹ Elsewhere, enforced aggregation with polymeric, micellular and π-stacking microstructuring profoundly impacts the selectivity and efficiency of photochemical EnT [2 + 2]-cycloaddition reactions, even allowing reactions to occur under air.^22,23 For example, we reported the 4-fluorobenzoyl group as a light harvesting auxiliary which effected catalyst-free self-[2 + 2]-photocycloadditions efficiently under air (Fig. 1C).²³ Exclusive diastereoselectivity was rationalized by non-covalent π-stacking self-aggregation. Based on the above observations, we questioned if different catalyst designs or oxime O-auxiliaries could non-covalently assemble oxime ester precursors or iminyl radicals to selectively afford azines in synthetically useful yields, either via an EnT pathway or via a reductive SET pathway. In this work, we present a novel, photocatalytic access to azines from simple oxime ester precursors using a cheap, compact orgPC with light as the only reagent (Fig. 1D).

Results and discussion

Synthetic results

Initially, we envisaged using 4-fluorobenzoyl-capped oximes in a photocatalyst-free EnT activation mode that might encourage π-stacking self-aggregation of oxime esters²³ to promote iminyl radical dimerization by close proximity. The catalyst-free photochemical reaction of oxime ester 1 gave desired azine 2 in only 20% yield (50% conversion) after 40 h (Table 1, entry 1). While addition of an exogenous, commercially available EnT photocatalyst thioxanthone (TX) to transfer energy to 1 increased conversion (full conversion after 16 h), 2 was formed in a similar yield (24%, entry 3). Although azine formation was more efficient here than in previous EnT reports,^17,18 this experiment revealed that the 4-fluorobenzoyl auxiliary was not suitable for a controlled EnT cleavage of oxime esters. Coupling of the iminyl radical with the aryl radical of fluorobenzene was never observed, yet acetophenone was detected; suggesting iminyl radicals may slowly undergo HAT reactions and then hydrolyse upon work-up. Taken together, this confirms the EnT activation mode is unsuitable for azine formation and that side reactions of derived iminyl radicals prevail.

Table 1 Photocatalyst screening for formation of azines

Entry	PC (mol%)	Time (h)	λ (nm)	Yield of 2 ^a (%)	Conversion of 3 ^a (%)
All reactions performed on a 0.1 mmol scale in dry DCE (0.5 mL), under N2.a Conversions/yields determined by ^1H NMR of the reaction mixture using CH2Br2 as an internal standard.
1	—	40	395	20	50
2	TX (5)	16	405	10	100
3	TX (5)	16	395	24	100
4	Ir(ppy) 3 (2)	40	450	16	45
5	PTZ-1 (10)	40	405	22	68
6	Phenox-1 (5)	40	450	26	81
7	Phenox-2 (5)	40	450	37	96
8	TpAA (5)	40	405	45	86
9	TpAA (20)	40	405	59	98
10	pOMe-TPA (20)	40	405	16	50
11	TPA (20)	40	405	6	50
12	TBPA (20)	40	405	5	81

---

Therefore, we changed strategies to target a reductive SET activation. However, despite its sufficiently reducing excited state, T.M.-based catalyst Ir(ppy)₃ gave low conversion and only a 16% yield of 2 (entry 4). We were inspired by the achievements of orgPCs with potent reducing excited states like N-phenyl phenothiazine (PTZ)²⁴ and phenoxazines developed by Miyake and co-workers,²⁵ as well as bespoke cyanated derivatives developed by our group.^19c While N-phenyl phenothiazine was unsuccessful (entry 5), phenoxazine-based orgPCs (entries 6 and 7) gave slightly increased yields of 2 (26–37%).

Gratifyingly, an even simpler and easily prepared tris(p-anisyl)amine (TpAA) was similarly efficient, and the yield of 2 increased to 59% (entries 8 and 9), albeit at the expense of a higher catalyst loading. A less electron-rich derivative pOMe-TPA gave a worse result (entry 10), while electron-neutral/poor TPAs like triphenylamine and TBPA gave low yields (<6%) and lower conversions (entries 11 and 12). Since following the SET reaction the ester moiety is cleaved off anyway, we reasoned for the sake of atom economy that the 4-fluorobenzoyl of 1 could be replaced by a simple acyl group (Table 2, entry 1). To our delight, the yield of azine 2 from (O-acyl) oxime ester 3 was only slightly lower (50%) under identical reaction conditions. Increasing polarity of the aprotic solvent by using MeCN benefitted reaction efficiency – increasing the yield to 68% (entry 2); while less polar, aprotic solvents such as DMA, THF and DCE gave products in lower yields (entries 5–8). Gratifyingly, although performance was inferior to MeCN, DMC could be used as a biodegradable ‘green’ solvent,²⁶ giving the product in an acceptable (51%) yield (entry 9).

Table 2 Optimisation studies for the reaction with 3 and TpAA

Entry	TpAA loading (mol%)	Solvent ([M] of 3)	Additive (equiv.)	Yield of 2 ^a (%)
All reactions carried out with 0.1 mmol of 3 in dry solvents and under N2. See ESI† for further optimization studies.a Conversions/yields determined by ^1H NMR of the reaction mixture using CH2Br2 as an internal standard.b Reaction was performed without light irradiation.c Reaction was performed open to air. Solvents: dichloroethane (DCE), dimethylacetamide (DMA), dimethyl carbonate (DMC).
1	20	DCE (0.2)	—	50
2	20	MeCN (0.2)	—	68
3	20	MeCN (0.1)	—	86
4	20	MeCN (0.07)	—	76
5	20	PhMe (0.1)	—	48
6	20	DCE (0.1)	—	56
7	20	THF (0.1)	—	29
8	20	DMA (0.1)	—	60
9	20	DMC (0.1)	—	51
10	10	MeCN (0.1)	—	65
11	25	MeCN (0.1)	—	75
12	—	MeCN (0.1)	—	0
13^b	20	MeCN (0.1)	—	0
14	20	MeCN (0.1)	Et3N (3)	28
15	20	MeCN (0.1)	DABCO (3)	0
16	20	MeCN (0.1)	TEMPO (1)	23
17	20	MeCN (0.1)	TEMPO (2)	2
18^c	20	MeCN (0.1)	O2	63

---

Several concentrations of substrate 3 in MeCN (see ESI†) were tested and 0.1 M of MeCN gave 2 in the highest yield of 86% (entry 3). Consequently, different catalyst loadings were tested, but lower or higher catalyst loadings decreased the yield (entries 10 and 11); presumably the latter being due to increased shielding of light. Control experiments confirmed the necessity of light and catalyst (entries 12 and 13). We noticed TpAA was degrading during the reaction and only 53% recovery could be quantified from the model reaction conditions (entry 3). Therefore, we postulated a sacrificial amine might enhance the yield of 2 or the recovery of the catalyst, inspired by the amine electron shuttle concept demonstrated by Gilmour and co-workers in an SET N–O bond cleavage of Weinreb amides.²⁷ However, addition of either Et₃N or DABCO (3 equiv.) was detrimental (entries 14 and 15). While typical radical reactions are shut down by the presence of TEMPO or air, a reaction involving close-proximity radical generation and coupling would tolerate radical quenchers or O₂ to some extent. Consistent with this, 1 equiv. of TEMPO was insufficient to fully shut down the reaction and the reaction under air gave 63% of 2 (entries 16–18).

Next, the scope of azine synthesis was evaluated (Table 3). Halogen substituents para- to the oxime ester were well-tolerated giving moderate to very good (49–81%) yields of 2b–d. Both electron-poor 3e and electron-rich 3g oxime esters efficiently led to the desired azines (2e, 2g) in high yields (83% and 70% respectively); the electron-rich ester being slightly less efficient, as expected for a more challenging reductive SET. A more hindered meta-dimethylaryl-substituted oxime ester 3f gave its respective azine 2f in moderate yield (57%). A longer α-alkyl chain-bearing oxime ester 3h gave the product 2h in the same (86%) yield as the model product (2a), while 1-tetralone-derived oxime ester 3i and cyclohexyl ring-fused 3j gave lower yet satisfactory yields of azines (60–63%), suggesting that conformational rigidification disfavors the selectivity-determining non-covalent (π-stacking) assembly of 3/iminyl radicals with the orgPC. Benzophenone-derived oxime esters 3k–m gave the desired azines 2k–m in moderate to good yields (40–63%), with electron-richer derivatives performing better. Although oxime ester 3p led to no azine product (no conversion was observed, presumably due to the lack of π-systems needed for the assembly with TpAA), we were pleased to find that oxime esters 3n and 3o bearing both electron-rich and electron-poor heterocycles were equally well-tolerated, giving decent yields of products 2n (59%) and 2o (62%).

Table 3 Scope of the photocatalytic azine synthesis

a All reactions were performed using 0.1 mmol of oxime ester in dry MeCN (1 mL) and under N₂. Reactions were irradiated with a 405 nm LED for 40 h. b Reaction time 72 h. Unless otherwise stated, isolated yields are given. Yields in parenthesis determined by ¹H NMR of the reaction mixture using CH₂Br₂ as an internal standard. c EMY values, see ESI† for further information.

---

Subjecting two different oxime esters 3a and 3m (or, 3b and 3g) together to the reaction conditions gave no selectivity for a cross-coupled product could be observed. These reactions gave a mixture of homo-coupled and cross-coupled azines (see ESI† file for details).

Mechanistic studies

To probe for NCRs, substrate 3q was subjected to the reaction conditions (Fig. 2A). Observation of 5-exo-trig radical cyclisation (37% yield of 4) confirmed the intermediacy of the iminyl radical as one half of the reductive SET cleavage. Concerned that the acyl group might lead to byproducts that are volatile/lost in work-up, to confirm the fate of the oxime ester O-auxiliary, we subjected 1a to the optimal reaction conditions for 16h (Fig. 2B) and directly measured the ¹⁹F NMR of the reaction mixture without work-up. A peak corresponding to 4-fluorobenzoic acid was detected, confirmed by a spiking experiment (see ESI† for details). No fluorobenzene was detected. A peroxide strip test was negative, ruling out presence of a benzoyl peroxide. This confirms the role of *TpAA as a reductive photoredox catalyst that reductively cleaves the oxime ester to an iminyl radical and a carboxylate anion.

Fig. 2 (A) Iminyl radical cyclisation. (B) ¹⁹F NMR study of the fate of the O-auxiliary of 1a. (C) Control reaction to probe for addition–elimination mechanism. (D) Conversion of TpAA to CabZ under the reaction conditions. Conversions/yields determined by ¹H NMR of the reaction mixture using CH₂Br₂ as an internal standard.

On the other hand, TpAA radical cation can transform into 3,6-dimethoxy-9-(4-methoxyphenyl)-9H-carbazole (CabZ) following photochemical cyclization, deprotonation and further SET oxidation steps. This reactivity is known in literature, albeit under more complex or harsher reaction conditions.²⁸ We consistently observed CabZ in the NMR of the crude reaction products at the end of synthetic reactions, and a control experiment without oxime ester 3q present irradiating with 405 nm gave a 65 : 35 ratio of TpAA : CabZ (Fig. 2C). As N-phenyl carbazoles have been reported as potent reducing photocatalysts,²⁹ we speculated whether TpAA is a precatalyst to CabZ as the main photocatalyst. Therefore, we prepared CabZ authentically and tested the model reaction with it as the input catalyst (Table 4, entry 1). After 16 h, 2a was formed in 59% yield, while 3a was fully converted. Given the typical ratio of TpAA : CabZ (65 : 35) that we observed at the end of the synthetic reactions and in the control reaction (Fig. 2D), we tested the model reaction with various TpAA : CabZ ratios (entries 2–6) and found that a ratio of 50 : 50 gave the highest yield (75%) after 16 h. For comparison, the model reaction with TpAA when run for 16 h gave only a 54% yield of 2a after 63% conversion.

Table 4 Catalytic efficiencies of the model reaction with CabZvsTpAA

Entry	TpAA : CabZ	Time (h)	Yield of 2a (%)	Conversion of 3a (%)
All reactions were carried out following the standard procedure on a 0.1 mmol scale of oxime ester 3a in MeCN (1 mL). The mixture of TpAA and CabZ corresponds to a combined catalyst loading of 0.02 mmol. Conversions/yields determined by ^1H NMR of the reaction mixture using CH2Br2 as internal standard.a Reaction was performed without light.
1	0 : 100	16	59	100
2	20 : 80	16	63	98
3	35 : 65	16	72	98
4	50 : 50	16	75	98
5	65 : 35	16	69	90
6^a	65 : 35	16	—	—
7	80 : 20	16	68	96
8	90 : 10	16	53	100
9	100 : 0	16	54	63
10	100 : 0	40	86	100

---

A control reaction using TpAA : CabZ (65 : 35) in the dark did not afford product. Taken together, these results suggest the TpAA catalyst and the CabZ to which it transforms establish a cooperative photocatalytic mechanism in which TpAA delivers more selective conversion to the product (than CabZ), while CabZ leads to a higher rate of reaction.

Although the homodimerization of iminyl radicals is known to occur in poor yields in other studies,^17,30 we considered an alternative mechanism to arrive at the azine product. This involves addition of an iminyl radical to the N atom of another oxime ester to form a benzylic radical, which fragments a carboxyl radical by reforming the C=N bond (i.e. addition–elimination mechanism). To test for this, oxime ester 3a and 3p were subjected to reaction conditions (Fig. 2C). While 3a was fully converted to form 2a, no conversion of 3p was observed (when using any of 1, 5 and 10 equiv. of 3p). Additionally, to further rule out an addition elimination and fragmentation pathway, an oxime ester bearing a cyclopropyl group as a radical clock was tested under our reaction conditions (see ESI file†). This led to a 50% yield of the corresponding azine homocoupled product, and we did not detect any ring-opened products. The above two experiments rule out an addition–elimination mechanism.

Parallel to our synthetic mechanistic experiments, we sought to probe the mechanism spectroscopically to detect quenching of *TpAA by 3a. The lifetime of *TpAA was not previously reported, and so was determined by time-resolved single photon counting (TCSPC, λ_ex 370 nm; λ_em 395 nm). Its excited state decay consisted of two components, τ₁ = 2.3 ns and τ₂ = 8.2 ns; agreeing well with the nanosecond lifetimes reported previously for acyclic TPAs in MeCN.^21a Although this value is suitable for diffusion-controlled quenching, quenching of the steady-state emission of *TpAA by 3a was not possible to observe (see ESI file†). To the contrary, the steady-state emission of *TpAA increased in intensity upon repeated measurements using the excitation source of λ_ex 370 nm (see Fig. 3A). Such behaviour, which was irreversible in character, would not permit reliable Stern–Volmer analyses since the quenching effect of 3a was being countered by the increasing emission of a photoactive decomposition product of TpAA (i.e.CabZ).

Fig. 3 (A) UV-vis overlay of TpAA (black), TpAA after 370 nm for 185 min (blue), CabZ UV-vis (red). (B) Overlay of isoconcentration emission spectra of TpAA and CabZ. (C) Emission spectra of TpAA recorded over 185 min during irradiation (λ_ex = 370 nm) within the spectrometer.

After the kinetic analysis, the UV-vis spectrum of the sample (input TpAA) clearly showed new features at 310 nm, 355 nm, and 370 nm, while the original feature of TpAA at 295 nm persisted. Clearly, this is a mixture of TpAA and CabZ (Fig. 3A, blue trace) as compared to the authentic standards (TpAA: black trace, CabZ: red trace). Since CabZ is a stronger emitter than TpAA at the same excitation wavelength and concentration (see Fig. 3B), this explains the increase in emission intensity upon irradiation of TpAA over time.

Next, we elected to monitor emission kinetically in the presence and absence of 3a over a period of 185 min (Fig. 3C). Interestingly, in the presence of 3a, the emission intensity jumped rapidly within the first ∼20 min and then stabilized/increased incrementally thereafter (see ESI file† for 3D kinetic plots, of which Fig. 3C is a cross-section). Notably, in the presence of 3a, the emission of the sample upon irradiation did not increase to the same level as when 3a was absent, suggesting presence of 3a limits the maximum concentration of CabZ formed in the reaction after an initial promotion. Consistent with this, we recovered TpAA (57%) and CabZ (20%) after the synthetic reaction of 3a, compared to complete decomposition of TpAA when it was irradiated on its own at λ_ex = 370 nm. These data corroborate an interaction between 3a (or the generated iminyl radical) and TpAA, that stabilizes TpAA during the reaction.

Proposed mechanism

Synthetic mechanistic experiments strongly indicate that a reductive SET based mechanism is in operation. This contrasts to – and explains the diverted reactivity compared to – the earlier works of the groups of Glorius and Cho under an EnT manifold.^17,18 We propose the mechanism shown in Fig. 4. Initially, a major proportion of TpAA participates in an assembly with substrate 3,^21d,e shielding photochemical transformation of the former to CabZ. A minor proportion of non-assembled TpAA is photoexcited and undergoes known photochemical transformation to CabZ.^28b Upon photoexcitation of the assembly (TpAA---3), oxidative quenching occurs. Radical anion 5 either undergoes unproductive back electron transfer (BET) with TpAA˙⁺, or productive mesolytic cleavage to afford 6 and iminyl radical 7, stabilized to HAT/degradation by assembly with the TpAA˙⁺.^21a–c The stabilized iminyl radical 7 then undergoes selective homocoupling, assisted by the assembly with TpAA˙⁺/CabZ˙⁺. As there is no stoichiometric reductant present in this net-reductive process, we propose that carboxylate anion 6 provides the electron to close the cycle. Redox processes that are uphill by ∼500 mV can occur in the ground state and although a direct redox reaction between TpAA˙⁺ and the carboxylate would be too endergonic by redox potentials (E_1/2TpAA = +0.61 V (see ESI†); E_p/2ⁿBu₄NOBz = +1.40 V; vs. SCE³¹), CabZ˙⁺ would be able to shuttle electrons between these species (E_1/2CabZ = +0.91 V vs. SCE).³² This furnishes carboxyl radical 8, that engages solvent in HAT to afford carboxylic acid 9.

Fig. 4 Proposed SET reductive cleavage mechanism and co-operative catalysis.

This mechanism consists with the known propensities of triarylamines and their radical cations to form assemblies with substrates and reactive intermediates, by N lone-pair—C=O interactions or π–π stacking interactions.²¹ It explains the need for CabZ in the reaction, and consists with the observation that greater proportions of CabZ catalyst lead to faster reactions. This is presumably due to the final endergonic SET step with the carboxylate, which is likely rate-determining. Further mechanistic investigations are ongoing.

Scale up in flow

The reaction of model substrate 3a was successfully scaled up in batch to a 1.00 mmol scale (74% yield of 2a) but required 90 h reaction time, translating to 0.97 mg h⁻¹ of nominal productivity. To further challenge the scalability of this reaction, flow experiments were performed using a Vapourtec UV-150 flow reactor fitted with an 8.2 mL coil. The standard reaction conditions at a flow rate of 0.02 mL min⁻¹ (R_T 410 min) gave 80% conversion of 3a and a 61% yield of 2a (Table 5, entry 1), translating to a productivity of 8.6 mg h⁻¹ (already 8.4× that in batch). Increasing the flow rate by 2.5× was detrimental to the conversion and yield (entry 2). Interestingly, at an even faster flow rate of 0.1 mL min⁻¹, the reaction was perfectly selective with yield matching conversion (34%), and productivity was highest yet (entry 3). Recirculating three times under these conditions increased the conversion and yield to 58% (entry 4), at the compromise of productivity. Lower reaction concentrations increased the yield (entries 5–6 vs. 3). Finally, the conditions of entry 4 were employed on a gram-scale (6.95 mmol) reaction of 3a, affording after 70 h 0.66 g (40%) of the product (entry 7); although the yield was ∼half that of the 1.00 mmol batch reaction, nominal productivity was ∼10× greater. Overall, a larger/longer reactor would allow higher product yields, selectivity and productivity.

Table 5 Continuous flow experiments

Entry	Scale (mmol)	Concentration (M)	Flow rate (mL min^−1)	R T (min)	Conversion of 3a (%)	Yield of 2a ^a (%)	Productivity (mg h^−1)
All reactions were carried out with (E)-1-phenylethan-1-one-O-acetyl oxime 3a (1 equiv.) in MeCN (10 mL) under irradiation with a 405 nm LED.a After completion, the solvent was simply removed under reduced pressure and yield was determined by ^1H NMR of the reaction mixture using CH2Br2 as an internal standard.b Mixture circulated through the reactor continuously under irradiation for three passes.c Isolated yield.d Reaction performed with 100% CabZ as the catalyst.
1	1	0.1	0.02	410	80	61	8.6
2	1	0.1	0.05	164	64	37	13.1
3	1	0.1	0.1	82	34	34	24.1
4^b	1	0.1	0.1	246	58	58	13.2
5	0.75	0.075	0.1	82	39	39	20.7
6	0.50	0.05	0.1	82	54	49	17.3
7	6.95	0.1	0.1	(70 h)	—	40 (0.66 g)^c	—
8^d	1	0.1	0.1	82	67	43	37.1

---

Conclusions

In conclusion, we demonstrate a hydrazine-free, reductant-free and non-CO₂ extruding photocatalytic azine synthesis, by a reductive SET pathway involving a triarylamine organophotocatalyst. The reaction proceeds under conditions that are simple, safe and green with light as the only reagent; and can be accelerated and scaled up via continuous flow. Mechanistic experiments corroborate (i) a reductive SET mechanism for engaging the N–O bond cleavage, instead of EnT and (ii) a preassembly between the oxime ester and the catalyst, that promotes iminyl radical coupling in close proximity, avoiding other pathways and accessing synthetically useful yields of azines.

Author contributions

J. S. discovered the reaction, performed the majority of the synthetic work, data analysis and wrote the manuscript with guidance from J. P. B., D. C. synthesized several oxime esters, contributed to the substrate scope and robustness of the reaction, performed all the flow experiments and wrote the ESI file with support from J. S.† S. S. contributed to starting material synthesis. W. H. performed all spectroscopic investigations. J. Ž. contributed to synthesis of catalysts. J. P. B. conceptualized and led the project, supervised all co-workers in their contributions, edited the manuscript and ESI file† and dealt with peer review.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

J. S., S. S., W. H., J. Ž. and J. P. B. thank the Alexander von Humboldt Foundation, provided within the framework of the Sofja Kovalevskaja Award endowed by the German Federal Ministry of Education and Research to J. P. B., for funding the main investigation. D. C. and J. P. B. acknowledge financial support of D. C. by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – SPP2370 – “Nitroconversion”. W. H. thanks Regina Hoheisel and Julia Zach for support and training in electrochemical and luminescence measurements. D. C. and J. P. B. are members of the Elite Network of Bavaria Doctoral College: “IDK Chemical Catalysis with Photonic or Electric Energy Input”. J. P. B. is a associated member of DFG TRR 325 “Assembly Controlled Chemical Photocatalysis” (444632635). We thank other members of the TRR and IDK for insightful discussions.

References

1. J. Kishore, S. Thiyargarajan and C. Gunathan, Chem. Commun., 2019, 55, 4542.
2. R. Cohen, B. Rybtchinsky, M. Gandelman, L. J. W. Whimon, L. Martin and D. Milstein, Angew. Chem., Int. Ed., 2003, 42, 1949–1952.
3. W. Han, G. Zhang, G. Li and H. Huang, Org. Lett., 2014, 16, 3532–3535.
4. W. Tan, Y. Xiang and A. Tong, J. Org. Chem., 2009, 74, 2163–2166.
5. Y. A. Kuznetsova and V. B. Romakh, Appl. Biochem. Biotechnol., 1996, 61, 205–209.
6. M. Revanasiddappa, T. Suresh, S. Khasim, S. C. Raghavendra, C. Basavaraja and S. D. Angadi, E-J. Chem., 2008, 5, 395–403.
7. E. C. Kesseln, W. B. Euler and B. M. Foxman, Chem. Mater., 1999, 11, 336–340.
8. V. S. Vyas, F. Haase, L. Stegbauer, G. Savasci, F. Podjaski, C. Ochsenfeld and B. V. Lotsch, Nat. Commun., 2015, 6, 8508.
9. Examples: (a) W. Han, G. Zhang, G. Li and H. Huang, Org. Lett., 2014, 16, 3532–3535; (b) J. Safari and S. Gandomi-Ravandi, RSC Adv., 2014, 4, 46224–46249.
10. Hydrazine (ICSC: 0281), IPCS INCHEM, WHO, 2013.
11. G. Choudhary and H. Hansen, Chemosphere, 1998, 37, 801–843.
12. R. Bertani, M. Biasiolo, K. Darini, R. A. Michelin, M. Mozzon, F. Visentin and L. Zanotto, J. Organomet. Chem., 2002, 642, 32–39.
13. M. Zhao, H. Liang, Z. Ren and Z. Guan, Synthesis, 2012, 1501–1506.
14. Selected reviews: (a) C. K. Prier, D. A. Rankic and D. W. C. MacMillan, Chem. Rev., 2013, 113, 5322–5363; (b) N. A. Romero and D. A. Nicewicz, Chem. Rev., 2016, 116, 10075–10166; (c) L. Marzo, S. K. Pagire, O. Reiser and B. König, Angew. Chem., Int. Ed., 2018, 57, 10034–10072; (d) F. Strieth-Kalthoff, M. J. James, M. Teders, L. Pitzer and F. Glorius, Chem. Soc. Rev., 2018, 47, 7190–7202; (e) F. Julia, T. Constantin and D. Leonori, Chem. Rev., 2021, 122, 2292–2352.
15. (a) D. S. Lee, V. K. Soni and E. J. Cho, Acc. Chem. Res., 2022, 55, 2526–2541; (b) C. Pratley, S. Fenner and J. A. Murphy, Chem. Rev., 2022, 122, 8181–8260.
16. (a) X. Yu, J. Chen, P. Wang, M. Yang, D. Liang and W. Xiao, Angew. Chem., Int. Ed., 2018, 57, 738–743; (b) J. Chen, P. Wang, B. Lu, D. Liang, X. Yu, W. Xiao and J. Chen, Org. Lett., 2019, 21, 9763–9768; (c) P. Wang, Y. Gao, J. Chen, X. Huan, W. Xiao and J. Chen, Nat. Commun., 2021, 12, 1815; (d) C. Chen, J. Zhao, X. Shi, L. Liu, Y. Zhu, W. Sun and B. Zhu, Org. Chem. Front., 2020, 7, 1948.
17. T. Patra, S. Mukherjee, J. Ma, F. Strieth-Kalthoff and F. Glorius, Angew. Chem., Int. Ed., 2019, 131, 10624–10630.
18. (a) V. K. Soni, S. Lee, J. Kang, Y. K. Moon, H. S. Hwang, Y. You and E. J. Cho, ACS Catal., 2019, 9, 10454–10463; (b) D. S. Lee, V. K. Soni and E. J. Cho, Acc. Chem. Res., 2022, 55, 2526–2541; (c) Q. H. Nguyen, H. S. Hwang, E. J. Cho and S. Shin, ACS Catal., 2022, 12, 8833–8840.
19. Selected examples: (a) K. Donabauer, M. Maity, A. L. Berger, G. S. Huff, S. Crespi and B. König, Chem. Sci., 2019, 10, 5162; (b) Y. Kwon, J. Lee, Y. Noh, D. Kim, Y. Lee, C. Yu, J. C. Roldao, S. Feng, J. Gierschner, R. Wannemacher and M. S. Kwon, Nat. Commun., 2023, 14, 92; (c) J. Žurauskas, S. Boháčová, S. Wu, V. Butera, S. Schmid, M. Domański, T. Slanina and J. P. Barham, Angew. Chem., Int. Ed., 2023, 62, e202307550.
20. M. J. P. Mandigma, J. Kaur and J. P. Barham, ChemCatChem, 2023, 15, e202201542.
21. (a) S. Wu, J. Žurauskas, M. Domański, P. S. Hitzfeld, V. Butera, D. J. Scott, J. Rehbein, A. Kumar, E. Thyraug, J. Hauer and J. P. Barham, Org. Chem. Front., 2021, 8, 1132–1142; (b) A. Kumar, P. Malevich, L. Mewes, S. Wu, J. P. Barham and J. Hauer, J. Chem. Phys., 2023, 158, 144201; (c) L. Wylie, J. P. Barham and B. Kirchner, ChemPhysChem, 2023, 24, e202300470; (d) A. Dewanji, L. van Dalsen, J. A. Rossi-Ashton, E. Gasson, G. Crisenza and D. J. Procter, Nat. Chem., 2023, 15, 43–52; (e) D. J. Castillo-Pazos, J. D. Lasso, E. Hamzehpoor, J. Ramoz-Sánchez, J. M. Salgado, G. Cosa, D. F. Perepichka and C.-J. Li, Chem. Sci., 2023, 14, 3470–3481.
22. (a) T. Dünnebacke, K. K. Kartha, J. M. Wahl, R. Q. Albuquerque and G. Fernández, Chem. Sci., 2020, 11, 10405–10413; (b) J. C. G. Kürschner, L. Brüss and L. Næsborg, Chem. – Eur. J., 2023, 29, e202300627.
23. S. Yakubov, W. J. Stockerl, X. Tian, A. Shahin, M. J. P. Mandigma, R. M. Gschwind and J. P. Barham, Chem. Sci., 2022, 13, 14041.
24. Selected examples: (a) S. Jin, H. T. Dang, G. C. Haug, R. He, V. D. Nguyen, V. T. Nguyen, H. D. Arman, K. S. Schanze and O. V. Larionov, J. Am. Chem. Soc., 2020, 142, 1603–1613; (b) H. Wang and N. T. Jui, J. Am. Chem. Soc., 2018, 40, 163–166; (c) F. Seyfert and H.-A. Wagenknecht, Synlett, 2021, 582–586.
25. R. M. Pearson, C. Lim, B. G. McCarthy, C. B. Musgrave and G. M. Miyake, J. Am. Chem. Soc., 2016, 138, 11399–11407.
26. (a) P. Tundo and M. Selva, Acc. Chem. Res., 2022, 35, 706–716; (b) V. Klopfer, R. Eckl, J. Floß, P. M. C. Roth, O. Reiser and J. P. Barham, Green Chem., 2021, 23, 6366–6372.
27. J. Soika, C. McLaughlin, T. Nevesely, C. G. Daniliuc, J. J. Molloy and R. Gilmour, ACS Catal., 2022, 12, 10047–10056.
28. (a) D. T. Breslin and M. A. Fox, J. Org. Chem., 1994, 59, 7557–7561; (b) A. C. Hernandez-Perez, A. Caron and S. K. Collins, Chem. – Eur. J., 2015, 21, 16673–16678.
29. (a) R. Matsubara, T. Yauta, U. M. Idros, M. Hayashi, F. Ema, Y. Kobori and K. Sakata, J. Org. Chem., 2018, 83, 9381–9390; (b) T. D. Weinhold, N. A. Reece, K. Ribeiro, M. L. Ocasio, N. Watson, K. Hanson and A. R. Longstreet, J. Org. Chem., 2022, 87, 16928–16936.
30. D. Griller, G. D. Mendenhall, W. Van Hoof and K. U. Ingold, J. Am. Chem. Soc., 1974, 96, 6068–6070.
31. H. G. Roth, N. A. Romero and D. A. Nicewicz, Synlett, 2016, 714–723.
32. The half potential of CabZ is reported to be +360 mV higher than TpAA under comparable voltammetry conditions, see: W. Lamm, F. Pragst and W. Jugelt, J. Pract. Chem., 1975, 317, 995–1004.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental procedures, reaction optimization, electrochemistry, spectral and X-ray data. CCDC 2288426. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d4gc00804a

---