α-Functionalization of Imines via Visible Light Photoredox Catalysis

Alberto F. Garrido-Castro 1 , M. Carmen Maestro 1,* and José Alemán 1,2,* 1 Department of Organic Chemistry, Universidad Autónoma de Madrid, 28049 Madrid, Spain; albertof.garrido@uam.es 2 Institute for Advanced Research in Chemical Sciences (IAdChem), Universidad Autónoma de Madrid, 28049 Madrid, Spain * Correspondence: carmen.maestro@uam.es (M.C.M.); jose.aleman@uam.es (J.A.); Tel.: +34914973875 (J.A.)


Introduction
Visible light photoredox catalysis has been at the forefront of organic chemistry research for over a decade, establishing itself as a sustainable and multifaceted synthetic tool [1]. Irradiation of catalytic amounts of polypyridyl complexes and organic sensitizers under mild conditions has proven to be an excellent activation pathway to access a wide variety of radical intermediates (Figure 1). Spurred by this resurgence, long-standing challenges in the field have been resolved, while a plethora of transformations continue to be developed in an effort to revamp organic synthesis.

23
Visible light photoredox catalysis has been at the forefront of organic chemistry research for over   The reactivity of imines has certainly undergone a complete makeover, as different strategies involving these key building blocks have been developed (Scheme 1). The classical approach still relies Scheme 1. Photocatalytic functionalization of imines: pathway A, alkyl radical addition (left); pathway B, single-electron reduction (right).
Alternatively, the photocatalytic single-electron reduction of imines has emerged as a powerful technology to generate radical anion intermediates which exist as two different resonant forms (pathway B in Scheme 1, right) [3,4]. The α-amino radical species can engage in radical-radical couplings with a large pool of reacting partners (pathway B1 in Scheme 1), while also displaying a complementary nucleophilic behavior to their corresponding electrophilic imine precursors. Indeed, they can be trapped by electron-deficient π-systems, a combination which would not be feasible using polar chemistry (pathway B2 in Scheme 1). Interestingly, the N-centered radical species can be quickly quenched by an H atom donor to yield a stable carbanion capable of reacting with a traditional electrophile in polar fashion (pathway B3 in Scheme 1).
The single-electron reduction event (pathway B in Scheme 1, right) can be a challenging redox process which often requires assistance [4]. While some electron-poor imines can undergo a straightforward photocatalytic reduction (such as N-sulfonylor α-keto-imines), other neutral imines feature a reduction potential which falls out of range of most photocatalysts. The addition of an external Lewis acid can increase the reduction potential of the imine (less negative) through coordination. Moreover, hydrogen-bonding via Brønsted acid can make this reduction a thermodynamically favorable process thanks to proton-coupled electron transfer (PCET), wherein an electron transfer from the photocatalyst to the imine takes place in concert with a proton transfer from the Brønsted acid to the imine.

Racemic Photocatalytic Radical Additions to Imines
Racemic radical additions to imines under visible light photocatalysis began to appear in 2016, when Bode reported the cyclization of silicon amine protocol (SLAP) reagents with an imine moiety (Scheme 2, left) [5]. These α-silyl amine precursors could undergo mild single-electron oxidation to render α-amino radicals, which could then engage with the imine to yield a wide variety of piperazine derivatives. The protocol was expanded further with α-silyl ether and thioether precursors to access morpholines, oxazepanes, thiomorpholines and thiazepanes (Scheme 2, right) [6,7]. It should be noted that, in this case, a Lewis acid was required to activate the imine, and the photocatalytic cycle could start with an initial Lewis acid-assisted single-electron reduction of the imine.

64
when Bode reported the cyclization of silicon amine protocol (SLAP) reagents with an imine moiety 65 (Scheme 2, left) [5]. These α-silyl amine precursors could undergo mild single-electron oxidation to 66 render α-amino radicals, which could then engage with the imine to yield a wide variety of 67 piperazine derivatives. The protocol was expanded further with α-silyl ether and thioether 68 precursors to access morpholines, oxazepanes, thiomorpholines and thiazepanes (Scheme 2, right) 69 [6,7]. It should be noted that, in this case, a Lewis acid was required to activate the imine, and the 70 photocatalytic cycle could start with an initial Lewis acid-assisted single-electron reduction of the

75
The first intermolecular photocatalytic radical addition was published in 2017 by Molander's 76 group [8]. The design of a general and modular approach based on the swift single-electron oxidation 77 of ammonium alkyl bis(catecholato)silicates enabled the alkylation of different N-sulfonyl-and N-78 aryl-imines (Scheme 3, top left) [9]. In addition, Friestad employed these silicon reagents to perform 79 the alkylation of N-acyl hydrazones in the presence of a Lewis acid (Scheme 3, top right) [10]. Alkyl 80 silicates have also been utilized by Kelly and Molander to achieve the synthesis of various saturated 81 N-heterocycles via radical alkylation and subsequent cyclization in a radical polar crossover (RPC) 82 process (Scheme 3, bottom) [11].

83
More radical precursors have also been deployed in an attempt to expand the synthetic prowess The first intermolecular photocatalytic radical addition was published in 2017 by Molander's group [8]. The design of a general and modular approach based on the swift single-electron oxidation of ammonium alkyl bis(catecholato)silicates enabled the alkylation of different N-sulfonyland N-aryl-imines (Scheme 3, top left) [9]. In addition, Friestad employed these silicon reagents to perform the alkylation of N-acyl hydrazones in the presence of a Lewis acid (Scheme 3, top right) [10]. Alkyl silicates have also been utilized by Kelly and Molander to achieve the synthesis of various saturated N-heterocycles via radical alkylation and subsequent cyclization in a radical polar crossover (RPC) process (Scheme 3, bottom) [11].
More radical precursors have also been deployed in an attempt to expand the synthetic prowess of this transformation. For instance, Hanna, Jr. and Molander disclosed the photocatalytic activation of alkyl trifluoroborates, enabling the radical alkylation of non-activated imines (Scheme 4) [12,13] Alkyl carboxylic acids hold a preferred position among radical precursors due to their versatility and ubiquity. Indeed, these alkylating agents can be implemented into mechanistically distinct photoredox pathways. Deprotonation of the acid can render a carboxylate species which can then undergo single-electron oxidation and subsequent decarboxylation to afford the alkyl radical intermediate. Alternatively, these acids can be activated with N-hydroxyphthalimide (NHPI) or its tetrachlorinated derivative (TCNHPI) through a simple esterification process to provide redox-active esters (RAEs). In this case, single-electron reduction can deliver the alkyl radical intermediate. This flexible behavior has been exploited by several research groups attempting to perform the alkyl radical addition to imines (Scheme 5). Weng and Lu reported the decarboxylative benzylation process following the oxidative pathway (Scheme 5, left) [14,15], while Mariano and Wang published a reductive version. In this later case, the decarboxylative glycosylation of imines was featured, although a Hantzsch ester (HEH) derivative was needed as a stoichiometric photosensitizer (Scheme 5, right) [16,17]. Notably, the radical fluoroalkylation of imines had remained inaccessible in the field until Maestro and Alemán recently reported the direct difluoromethylation of imine moieties (Scheme 6) [18]. This general procedure was predicated on the single-electron oxidation of readily available zinc difluoromethane sulfinate (DFMS) in the presence of an organophotoredox catalyst (Rhodamine 6G). Lastly, hydrogen atom transfer (HAT) has also been used in order to perform the C-H activation of different alkyl radical precursors and perform the desired alkylation reaction with activated imines (Scheme 7). Lu and Gong reported the α-oxyalkyl radical addition of 1,3-dioxolane to fluoroalkyl imines (Scheme 7, top) [19,20], while Dilman managed to install different alkyl and acyl radicals into N-sulfonyl imines (Scheme 7, bottom) [21][22][23].

Stereoselective Photocatalytic Radical Additions to Imines
In the field of asymmetric photocatalytic additions to imines, Knowles first described in 2013 an elegant intramolecular example in which a hydrazone trapped a ketyl radical intermediate-generated by PCET-in enantioselective fashion thanks to the chiral induction exerted by a chiral phosphoric acid (Scheme 8) [24].   No further reports were published on this topic until Maestro and Alemán disclosed in 2017 an asymmetric intermolecular radical alkylation of imines based on the use of chiral sulfoxides (Scheme 9) [25]. The photocatalytic reduction of NHPI-derived RAEs delivered the alkyl radical, which then engaged with the enantiopure N-sulfinimine in diastereoselective fashion to afford α-branched benzyl amine derivatives.    Moreover, Gong's research group developed a series of transformations in 2018 and 2019 based on chiral Lewis acid-catalyzed radical alkylations of different imine scaffolds (Scheme 10, top) [26,27]. When using redox-active alkyl trifluoroborates and silanes, the Cu-BOX complexes acted as bifunctional chiral photocatalysts, performing both the asymmetric induction and the single-electron oxidation of the radical precursors, while suppressing the need for an external photocatalyst. In the latest report published by Gong, an HAT-photocatalyst (5,7,12,14-pentacenetetrone, PT) was required to perform the C-H activation of benzyl and allyl positions, as well as non-activated alkanes (Scheme 10, bottom) [28].   The generality observed throughout this section noticeably stands out, wherein an assortment of radical precursors has been inserted into mechanistically similar protocols based on their photocatalytic activation to render the nucleophilic radical intermediate. Most notably, the range of imine building blocks employed in these reactions is quite impressive, as both activated and non-activated substrates have proven to be suitable acceptors to the different radical additions.

Racemic α-Amino Radical-Radical Couplings-Pathway B1
The generation of α-amino radical intermediates derived from imine building blocks was first reported by Kisch (Scheme 11, top) [29,30]. By means of a family of heterogeneous photocatalyst semiconductors, their group developed a series of transformations involving the coupling of α-amino radicals and allyl radicals (pathway B1 in Scheme 1) [31]. CdS powder as well as supported versions on SiO 2 , ZnS, and Al 2 O 3 can behave as the photocatalyst semiconductor which features a surface with the ability to engage in interfacial electron transfer (IFET). Upon visible light absorption, the semiconductor can generate an electron-hole pair-essentially reducing and oxidizing surface centers. These sites can then perform IFET with the adsorbed substrates, delivering the two radical intermediates that eventually afford the recombination product (known as semiconductor photocatalysis B). Later on, Pu and Shen used CdSe/CdS core/shell quantum dots (QDs) as photocatalysts for the transfer hydrogenation of diaryl imines with a thiophenol as H atom donor (Scheme 11, bottom) [32,33].

159
The generation of α-amino radical intermediates derived from imine building blocks was first   In the field of homogenous photocatalysis, MacMillan first reported the coupling of α-amino radicals with α-oxybenzyl and β-enaminyl radicals, giving access to pinacol-type products and the formal β-Mannich reaction, respectively (Scheme 12, top) [34,35]. Notably, the generation of the radical reacting partners required elegant multicatalytic approaches. The pinacol-type coupling reaction relied on the initial photocatalytic oxidation and deprotonation of methyl thioglycolate to produce a thiyl radical. Then, this S-centered radical could perform the H atom abstraction from the benzyl ether to afford the α-oxybenzyl radical and regenerate the thiol catalyst. On the other hand, the β-enaminyl radical could be accessed following: i) initial condensation of a cyclic ketone with a simple aminocatalyst (azepane), ii) subsequent oxidation of the catalytic enamine, and iii) final allylic deprotonation (Scheme 12, bottom).
Following the development of these transformations, different variants of the radical-radical coupling of α-amino radicals began to appear. Sudo successfully employed an organophotoredox catalyst in a similar symmetrical coupling [38], whereas Gilmore obtained vicinal primary diamine products through in situ formation of the imines with aldehydes and ammonia [39]. Regarding unsymmetrical adducts, Wang reported the coupling of imine-derived α-amino radicals with tetrahydroisoquinoline-derived α-amino radicals [40]. In most cases, the imine reduction is believed to be assisted by coordination to an external acidic species.  [Ir(ppy) 2  A unique example based on the reactivity of these α-amino radicals was reported by Opatz featuring a four-component reaction which gave access to structurally diverse products (Scheme 14) [41]. The protocol involved the simultaneous construction of three new bonds: C-N (via in situ formation of the imine), C-S (via sulfonyl radical addition to styrene) and C-C (via α-amino radical-benzyl radical coupling). tetrahydroisoquinoline-derived α-amino radicals [40]. In most cases, the imine reduction is believed 198 to be assisted by coordination to an external acidic species.

199
A unique example based on the reactivity of these α-amino radicals was reported by Opatz  Arylation reactions, which had remained elusive in the context of radical chemistry with imines, were achieved by Xia and Lehnherr and Rovis (Scheme 15) [42,43]. Through generation of stabilized aryl radical intermediates, the radical-radical coupling became feasible with 1,4-dicyanobenzene and 4-cyanopyridine precursors, respectively.

Racemic α-Amino Radical Additions to Activated Olefins-Pathway B2
The development of the single-electron reduction of imines to produce new radical-radical couplings has undoubtedly revamped this area. Furthermore, the polarity reversal displayed in this redox process served as a platform upon which an even greater challenge could be tackled. The new nucleophilic character of the α-amino radical was exploited in a series of Giese radical additions (pathway B2 in Scheme 1) reported by several groups. Indeed, Chen [44], Dixon [45,46], Ngai [47], and Rueping [48] independently worked on the addition of these intermediates to different activated olefins (Scheme 16).

216
The development of the single-electron reduction of imines to produce new radical-radical 217 couplings has undoubtedly revamped this area. Furthermore, the polarity reversal displayed in this 218 redox process served as a platform upon which an even greater challenge could be tackled. The new 219 nucleophilic character of the α-amino radical was exploited in a series of Giese radical additions 220 (pathway B2 in Scheme 1) reported by several groups. Indeed, Chen [44], Dixon [45,46], Ngai [47],

221
and Rueping [48] independently worked on the addition of these intermediates to different activated

226
In addition, Dixon's lab described two procedures built on these radical additions to   In addition, Dixon's lab described two procedures built on these radical additions to electrophilic partners followed by cyclization events, thus granting access to molecules of higher structural complexity (Scheme 17) [49,50]. Notably, the addition of the α-amino radical to vinyl sulfones led to a reverse polarity Povarov reaction (Scheme 17, top), while bridged 1,3-diazepanes could be prepared via α-amino radical addition to the 4-position of a quinoline core and subsequent ring closure at the 2-position (Scheme 17, bottom).  Ring Closure [2-position] Radical Addition [4-

Racemic α-Amino Carbanion Nucleophilic Attacks-Pathway B3
After thorough evaluation of the results outlined above, it could be assumed that the single-electron reduction of imine scaffolds usually delivers a radical anion that can exist as two different resonant forms. As mentioned previously, the N-radical resonant form can engage in an HAT to yield a stable carbanion (pathway B3 in Scheme 1). This behavior has been exploited by Yu and Fan and Walsh to achieve polar nucleophilic attacks on numerous electrophiles, such as CO 2 and aldehydes, as well as the direct hydrolysis of these anionic intermediates to afford the formal reduction to benzyl amines (Scheme 18) [51][52][53][54].

Stereoselective Photocatalytic α-Amino Radical Reactivity via Single-Electron Reduction of Imine Derivatives-Pathway B
Controlling the stereochemical outcome of transformations involving the single-electron reduction of imines has proven to be an outstanding challenge. An exceptional solution to this problem was developed by Ooi's lab in 2015 and 2016 (Scheme 19) [55,56], wherein the radical anion derived from this redox process was ion-paired with a chiral phosphonium salt, rendering an asymmetric radical-radical coupling with α-amino radicals (pathway B1 in Scheme 1). Moreover, Jiang's lab reported two transformations involving the use of traditional organocatalysts (Scheme 20) [57,58]. In the first one, the formal reduction of activated ketimines was achieved via coordination of Takemoto's urea catalyst to the radical anion and ensuing asymmetric HAT (pathway B1 in Scheme 1). In the second one, a chiral Brønsted acid managed to activate vinyl pyridines-acting as acceptors in a Giese radical addition-while exerting stereocontrol through H-bond interactions (pathway B2 in Scheme 1).
Finally, Ward and Wenger published an interesting asymmetric reduction of cyclic imines in which enzymatic catalysis played a fundamental role (Scheme 21) [59]. Initial photocatalytic reduction and HAT would deliver a racemic mixture of both amines (pathway B1 in Scheme 1), yet, in the presence of monoamine oxidase MAO-N-9, only one enantiomer could undergo subsequent re-oxidation (enantiomer recycling). Therefore, the combination of photocatalysis and biocatalysis yielded an elegant dynamic kinetic resolution (DKR) en route to chiral amines.
In this section, the pool of radical precursors employed in reactions following pathway B1 may not be as diverse when compared to conventional radical additions (pathway A). However, thanks to the nucleophilic character of the imine-derived radical anion, pathways B2 and B3 have widened the scope of reacting partners since they no longer require photocatalytic activation. Remarkably, common electrophiles used in polar methodologies find a smooth transition into these new radical processes thanks to the inspired polarity reversal enforced upon the imine substrates.

265
Finally, Ward and Wenger published an interesting asymmetric reduction of cyclic imines in 266 which enzymatic catalysis played a fundamental role (Scheme 21) [59]. Initial photocatalytic 267 reduction and HAT would deliver a racemic mixture of both amines (pathway B1 in Scheme 1), yet, yielded an elegant dynamic kinetic resolution (DKR) en route to chiral amines.

271
In this section, the pool of radical precursors employed in reactions following pathway B1 may