Synthetic Molecular Photoelectrochemistry: New Frontiers in Synthetic Applications, Mechanistic Insights and Scalability

Abstract Synthetic photoelectrochemistry (PEC) is receiving increasing attention as a new frontier for the generation and handling of reactive intermediates. PEC permits selective single‐electron transfer (SET) reactions in a much greener way and broadens the redox window of possible transformations. Herein, the most recent contributions are reviewed, demonstrating exciting new opportunities, namely, the combination of PEC with other reactivity paradigms (hydrogen‐atom transfer, radical polar crossover, energy transfer sensitization), scalability up to multigram scale, novel selectivities in SET super‐oxidations/reductions and the importance of precomplexation to temporally enable excited radical ion catalysis.


Introduction
Thef ield of single-electron transfer (SET) in organic synthesis and in the late-stage functionalization of complex molecules has expanded remarkably in the past two decades. Among this field, photoredox catalysis (PRC) and synthetic organic electrochemistry (SOE) are highly attractive to synthetic chemists due to 1) their abilities to afford reactive intermediates under mild conditions and 2) their use of light and electricity as sustainable energy sources to drive reactions. [1,2] As powerful and generally applicable as PRC and SOE are,t hese chemistries present some issues on af undamental level. PRC reactions are limited by the redox window of photocatalysts,which is in most part defined by the energy of visible photons (ca. 1.8-3.1 eV). [3] UV photons that access higher energy limits come with penalties of less energy-and cost-efficient reactors,s afety and the direct excitation of substrate molecules leading to deleterious pathways.V isiblelight PRC reactions utilizing multiphotonic processes therefore came to the fore for overcoming the "redox energy limit". [4] To turnover PRC reactions,i ncluding these multiphotonic paradigms,l arge excesses of sacrificial oxidants or reductants can be required to generate active forms of the catalyst or to turnover "spent" photocatalyst. In SOE reactions,e lectrodes generally do not distinguish between different reaction components beyond inherent thermodynamic redox potentials of reaction components.O ver-oxidations and over-reductions as well as solvent redox processes can plague reactions.
In this respect, engineering controls like tiny interelectrode distances, [5] alternating potential, [6] continuous flow manifolds [7] or electrode surface modifications, [8] have shown promise in imparting selectivity in recent years.The merger of PRC and SOE, "synthetic photoelectrochemistry", tackles these issues and has come to the forefront of methods for SET chemistry. [9] By channeling electrochemical and photochemical energy through ah omogeneous (or heterogeneous) catalyst, intriguing new reactivity and selectivity opportunities are manifested. Thee ssence of this emerging field has been captured by highlights [10] as well as afull review [9] and is partially covered in other reviews. [11] However,t he field has expanded dramatically in the last couple of years in terms of its synthetic opportunities,reactivity concepts and mechanistic understanding. In the super-redox chemistry of electrogenerated radical ion (open-shell) photocatalysts (Scheme 1), curious selectivities and reactivities unattainable by PRC or SOE chemistries have emerged, together with first evidences of catalyst-substrate precomplexation to rationalize otherwise time-forbidden excited-state reactivity.While the use of HATasareactivity paradigm is well established in PRC and SOE, [12,13] its recent combination with PEC signals ab road scope of future synthetic applications.F inally,i n( closedshell) photocatalyst electro-recycling, methods to upscale PEC to ap roductive level (multigram!multigram h À1 )a re emerging,i ncluding various continuous photoflow and undivided cell batch setups.
In the following sections,wecategorize the new examples as follows:1 )super-oxidations/reductions;2 )photocatalyst electro-recycling;3 )photoelectrochemical HATr eactions. Interfacial photoelectrochemistry [9] is not covered. [14] Nomen-Synthetic photoelectrochemistry (PEC) is receiving increasing attention as anew frontier for the generation and handling of reactive intermediates.PEC permits selective single-electron transfer (SET) reactions in amuch greener wayand broadens the redox windowofpossible transformations.Herein, the most recent contributions are reviewed, demonstrating exciting new opportunities,n amely,the combination of PEC with other reactivity paradigms (hydrogen-atom transfer,radical polar crossover, energy transfer sensitization), scalability up to multigram scale,n ovel selectivities in SET superoxidations/reductions and the importance of precomplexation to temporally enable excited radical ion catalysis.
clature and abbreviations herein are as defined in our introductory review. [9]

Electrochemically Mediated PhotoRedox Catalysis for Selective Super-Oxidations or Super-Reductions
Thet erm "electrochemically mediated photoredox catalysis" (e-PRC) was coined as ab lanket term to describe the category of PEC involving an intimate relationship of electroand photochemical steps within the same catalytic cycle,a s subsequent steps. [9] This broadly separates into two subcategories," radical ion e-PRC" and "recycling e-PRC" (Section 2) which, although technically are mechanistically identical, are conceptually different. Radical ion e-PRC involves electrogenerated radical ions which are photoexcitated to yield super-oxidants or super-reductants.T he key advantage is that radical ion e-PRC can engage compounds beyond the redox windows accessible to photoredox catalysis or synthetic electrochemistry alone (À3.0 Vt o+ 3.0 V), since photoexcited radical ion catalysts are oftentimes coloured species that possess higher redox energy (ca. % 0.5-1.5 eV) than conventional neutral (closed-shell) photocatalyst excited states.S eminal contributions of Moutet and Reverdy [15] first demonstrated this concept in the oxidations of diphenylethene and benzyl alcohol by an electrogenerated photoexcited phenothiazine (PTZ)r adical cation. These reports have been reviewed previously. [9] Lund and Carlsson first demonstrated the reductive paradigm:d echlorination of chlorobenzene by electrogenerated photoexcited pyrene radical anions. [16] Inspired by this,t he reductive dechlorination of chlorobiphenyls using photoexcited 9,10-diphenylanthracene radical anion was studied by Rusling. [17] While these reports established the e-PRC concept, their focus was not synthetic and their conditions not applicable beyond af ew catalytic turnovers.The field lay dormant for several decades until its recent renaissance driven by acombination of:1)an increasing interest in PRC and SOE, [9] 2) the realization of chemically generated photoexcited radical ions as potent redox agents in synthesis [4a-d] and 3) recent seminal efforts in e-PRC in organic synthesis [18] as reviewed previously. [9] Akey platform that has underpinned rapid uptake of PRC by the synthetic community is the compilation of photocatalyst structures with their photophysical and redox properties, enabling chemists to logically plan reactions and providing best chances of success.F igure 1d epicts structures of historical and contemporary radical ion (pre)photocatalysts, while Table 1s ummarizes their spectroscopic and redox properties as well as reported synthetic applications. Figure 2s ummarizes redox properties of radical ion photocatalysts compared to conventional photocatalysts and example substrates.While the focus of this section is on e-PRC,this compilation of properties should equally assist practitioners of conPET photocatalysis in planning of reactions.
Comparison of conPET and e-PRC: While conPET and e-PRC methods are analogous in their active catalyst excited state and their scope of SET reactions,t heir downstream chemistry is fundamentally distinct and complementary.T he reactivity paradigms are compared in the context of superreductions ( Figure 3). In both cases,f ollowing SET from the photoexcited radical ion catalyst, C-X bond cleavage affords ar adical intermediate.I nc onPET,t his radical intermediate, typically an aryl C(sp 2 )r adical, can undergo 1) hydrogenatom transfer (with solvent or with by-products derived from sacrificial reductants in the radical ion catalyst generation step), 2) addition to ah eteroatom trapping agent or 3) addition to an unsaturated partner (usually an electron-rich aromatic) in aC ÀCc oupling reaction. In the case of (3), the subsequent radical either reacts with aH AT agent ("ZC" ! olefination) or is oxidized (by "QC + ") and deprotonated, in both cases reforming the unsaturation. TheHAT agent ("ZC") or oxidant ("QC + ") is the by-product from the sacrificial electron donor that was required to form the radical anion photocatalyst. Forexample,the a-amino radical or N-radical cation of atrialkylamine electron donor.
In e-PRC,d ifferent fates await the radical intermediate following initial SET and C-X bond cleavage,byvirtue of the absence of sacrificial chemical reductants and their byproducts.T he radical can undergo the same radical trapping reactions as with conPET,but instead of HATwith solvent its electroreduction to ac arbanion occurs in ar adical polar crossover (RPC) fashion. Theresulting carbanion intermediate can undergo 1) protonation, 2) elimination of an a-leaving group or 3) electrophilic trapping.T he radical intermediate following CÀCc oupling can undergo further reduction and subsequent protonation, or can engage in radical trapping depending on applied potential. Thereby,e -PRC offers the unique advantage of au ser-potentiocontrollable mechanism.
While radical ion photocatalysts are ac ommon theme in conPET and e-PRC chemistries,the focus of this review now turns to recent synthetic applications of the latter. Elegant, synthetic conPET applications are reviewed elsewhere. [2e,4i-k, 11b] C(sp 2 )ÀNb ond formations: In the oxidative direction, e-PRC has witnessed impressive synthetic applications in C(sp 2 ) À Nbond formations.T he Buchwald-Hartwig coupling is an important, widespread reaction for forging C(sp 2 ) À N bonds,b ut relies on prefunctionalization of the C(sp 2 )containing arene (an aryl halide or pseudohalide) for oxidative addition of aP dc atalyst. Ap owerful alternative is the direct C(sp 2 )-H activation of arenes by SET oxidation to their radical cations,w hich undergo S N Ar-type attack by Ncontaining nucleophiles.T he Nicewicz group first demonstrated this concept using PRC with the Fukuzumi catalyst, an acridinium salt (Mes-Acr + )w ith ah igh excited-state oxidation potential (*E 1/2 =+1.88 Vvs. SCE). Electron-rich arenes Figure 1. Radical ion/radical precatalystsg rouped by molecular architecture:photooxidant precatalysts (red background) and photoreductant precatalysts (blue background).Shading corresponds to precatalyst family and is indicative of the oxidizing/reducing power of radical ion forms of the family but not specific cases (Table 1). )-X cleavage [17] such as anisole underwent direct C(sp 2 )À Nbond formations with azoles. Elegant work by the Lambert group exemplified the power of e-PRC to breach the upper redox limit of singlephoton PRC [18a] which has been reviewed elsewhere. [9] They disclosed at risaminocyclopropenium cation (TAC + )a sa ne lectroactivated photoredox catalyst (e-PRCat) whose excited state could oxidize benzene,c hlorobenzene and up to dichlorobenzenes as arene partners.T rifluorotoluene gave no product, defining the upper limit of the excited stateso xidative power. While its scope of applications was broad, the exotic structural architecture of the catalyst does not seem amenable to modifications.Structural modification of aphotoredox catalyst core to afamily of derivatives with different photophysical and redox properties is ac oncept underpinning much of the success of field of PRC with transition metal or organic photocatalysts. [2c,e] In the context of e-PRC,the Barham group disclosed tri(p-substituted) arylamines (TPAs) as at unable class of e-PRCat for super-oxidations (Figure 4A). [20] The TPAsa re easily accessed and customized in as ingle step by Suzuki or Ullmann coupling reactions.Aubiquitous structural motif in hole-transport functional materials (OLEDs,p hotovoltaics) [23] and as mediators in electrolysis, [24] the photophysical and redox properties of TPAsa nd their bench-stable (isolable) radical cations [25] have arich history of characterization [26] that precedes even that of ruthenium polypyridyl complexes. [27] In their e-PRC reactions, TPAsf irstly undergo anodic oxidation to their corresponding strongly coloured radical cations at ac onstant cell potential of + 1.4 to 1.8 V( Figure 4B). Altering para-substituents on the TPA allows facile access to radical cation photocatalysts with different oxidative powers.U se of am oderate-power TPA (TpBPA) allowed C(sp 2 )-H amination of alkylbenzenes with high selectivity;aldehydebearing pyrazoles and arene benzylic positions were tolerated without oxidation. Introduction of a para-cyano group on the peripheral aromatic ring gave ah igher-power TPA (TCBPA)w hich allowed CÀHamination of benzene and chlorobenzene in good yields and toler-  [19k] however,the lifetime of NpMIC À itselfwas successfully determined in arecent studyin0.1 M n Bu 4 N·PF 6 in DMAc. [19r] [p] This study [19q] did not assign the measured lifetime to aspecificexcited stateand the excited statewas thus denoted "ES 1 ". [q] Aquartet statewas tentatively proposed as acandidate for thelong-lived species. Since this state was found to be catalytically inactive in the PETstep due to no quenchingofthe excited stateinthe presence of substrate, an excitedstate potential is not given. ated free carboxylic acids known to undergo decarboxylation in PRC [28] and Kolbe oxidation pathways in SOE. [29] Using the most powerful TPA (TdCBPA), the oxidative SET C À H aminations of dichlorobenzenes,f luorobenzene and even acetophenone were achieved. SET activations of very electron-poor trifluorobenzene and trifluorotoluene were achieved, leading to S N Ar reactivity ( Figure 4C). Theu tility of triarylamines as af amily of e-PRCats that could be tuned to target substrates was successfully demonstrated.
Ak ey mechanistic question comes to mind when discussing the photochemistry of excited radical ions (doublet states), given that these species are well known to exhibit ultrashort (< 1ns) lifetimes that prohibit their diffusioncontrolled photochemistry.I th ad been tentatively hypothesized in previous studies involving conPET reduction that ap recomplexation of the substrate and radical ion photocatalyst occurs to rationalize otherwise temporally forbidden photochemistry. [4a,g] Ak ey advantage of TPAsa se lectroactivated photocatalysts is that they can be easily oxidized and isolated as their bench-stable radical cations. [25c,d] This allowed the authors to measure *TPAC + lifetimes as short as < 10 ps by transient absorption spectroscopy,c learly ruling out diffusion-controlled photochemistry.P recomplexation between TPAC + sa nd arene substrates was proposed as the key to overcoming ultrashort (ps) lifetimes ( Figure 4B). It was postulated that steric effects in the precomplex formation rationalized both 1) the reactivity trend of xylene and dichlorobenzene isomers,s ince preparative reaction yields increased in the order:1 ,4-< 1,2-disubstituted arenes;a nd 2) the lack of reactivity of iodobenzene.The increasing order of E p ox :1 ,4-< 1,2-disubstituted arenes [18a] and the highly accessible redox potential of iodobenzene meant that such behaviour was unexpected and contra-thermodynamic.  Table 1, all catalyst redox potentials are excited-state half-wave potentials *E 1/2 taken from the literature. [2c,e] Unless the excited-state redox potential is explicitly claimed in the associated citation, solid lines (c c)represent either the upperboundary excited doublet state redox potential (D n )and assignments are guided by the redox scope of accessible substrates.D otted lines (a a) represent upper or lower redox bounds for the excited doublet states (D 1 or D n ). Substrate redox potentials taken from the literature are irreversiblep eak redox potentials E p/2 [4c, 18a,b, 19q, 21,22] or lie beyond the solvent redox window (> À3.0 V; > + 3.0 V). [9]  Thep resence of precomplexes was identified spectroscopically via changes in the UV-vis spectra and EPR spectra of isolated TPAC + si nt he presence of arene substrates.T he most striking comparison was when 1,2-or 1,4-dichlorobenzene substrates were added to TCBPAC + .T he former "reactive" substrate perturbed the EPR signal to a" triplet shape", confirming spin density localization on the Natom in the precomplex. Thel atter "unreactive" substrate perturbed the signal to a" broad singlet", indicating another type of precomplex with spin density delocalization away from the N atom, which may stabilize the radical cation and decrease its excited state reactivity to SET.D FT calculations (wB97X-D or B3LYP functionals) found optimized structures involving T-p or p-p interactions.R elative binding energies revealed "reactive" substrates favoured T-p geometries and "unreactive" substrates favoured p-p geometries ( Figure 5). In chlorobenzenesp recomplex with TCBPAC + ,t he calculated spin density changed when the Cl atom faced "in" but did not change when facing "out", providing ac lue about the preferred geometry of precomplexes involving unsymmetrical haloarenes ( Figure 6). [30] Although TPAC + sabsorb strongly in the visible region (ca. 600-900 nm) corresponding to their first (doubly degenerate) excited states (D 0 !D 1 ), only shorter wavelengths (400 nm) gave reactivity,s uggesting anti-Kasha photochemistry.T he ability to access higherorder doublet photoexcited states of TPAC + sw ith 400 nm provides as high as *E p ox =+4.4 Vv s. SCE as an upper boundary, [31] and would rationalize the oxidation of arenes so electron poor that they exceed the measurable solvent window of cyclic voltammetry.T D-DFT calculations,w hich agreed well with UV-vis spectra of TPAC + s, [32] revealed the near-IR transition (D 0 !D 1 )i nvolved a p-p*t ransition localized at the core aromatic rings (Figure 7). In contrast, the transition closest to % 400 nm (D 0 !D n )i nvolved a p-p* transition at the peripheral aromatic rings.T his is precisely the binding location of substrate arenes as found by geometry optimizations.Precomplexation may rationalize synthetically productive anti-Kasha photochemistry by the localization of hole density at the binding site of the substrate,p riming the

Angewandte Chemie
Reviews precomplex for ultrafast SET (faster than internal conversion;D n !D 1 ).
While the simple computational model of precomplexes is consistent with experimental observations,d eeper investigations with other DFT theory levels, [33] different (non-DFT) [34] theories and different precomplex candidates [35] are warranted for am ore holistic picture.N onetheless,t his study evidenced how radical ion-substrate precomplexes not only circumvent ultrashort lifetimes of doublet excited states,b ut even circumvent internal conversion, allowing ag reater proportion of the photon energy to be harnessed. Moreover, the study indicated how radical ion-substrate precomplexes give rise to different selectivities than SET chemistry involving diffusion-controlled SOE or PRC. C(sp 3 ) À Nb ond formations: In an impressive synthetic application of e-PRC,t he Lambert group used TAC + for vicinal diamination of alkylarenes (4, Figure 8A). [36] Depending on the electrolyte,a lkylated arenes undergo vicinal diamination to afford either 3,4-dihydroimidazoles (5)o r oxazoline derivatives (6 and 6' '). In the anodic chamber of ad ivided cell, TAC + is oxidized to its coloured dication radical TACC 2+ .U pon photoexcitation, the highly oxidative *TACC 2+ (*E 1/2 =+3.33 Vv s. SCE) engages alkylarenes like cumene 4a in SET ( Figure 8B). Upon loss of ap roton from 4aC + and SET oxidation, benzylic carbenium ion 7 is generated and nucleophilic attack of MeCN (as solvent) leads to Rittertype amidation ( Figure 8C). Ther esulting acetamide (8) undergoes acid-catalyzed elimination to a-methylstyrene (9). Subsequent e-PRC oxidation of 9 by *TACC 2+ affords radical cation 9C + and MeCN solvent adds.F urther oxidation and addition of at hird equivalent of MeCN was proposed to afford dihydroimidazole 5 in aRitter-type fashion. [37] Control reactions using a-methylstyrene (9)that led to polymerization seemed to contradict this proposal. However, an equilibrium between 8 and 9 would generate small quantities of 9 in situ, mitigating polymerization. LiClO 4 electrolyte presumably alters the stability of cationic intermediates and the addition of H 2 Ot o9C + or 10 a affords 6' ' and 6,respectively.
When applied to secondary alkylbenzenes,h alide substitution was well-tolerated, while alkyl para-substitution on the arenes led to formation of Ritter-type benzylic by-products (5j,F igure 9A). Cyclic systems were successfully functionalized, with preference for 4-phenylimidazoles over 5-phenylimidazoles.Inprimary alkylbenzenes,the regioselectivity was inverted. Most impressively,the simple change of electrolyte from Et 4 N·BF 4 to LiClO 4 diverted the mechanism to form oxazolidines in an overall oxyamination. Them ethodology was amenable to late-stage functionalizations of pharmaceutical compound analogues ( Figure 9C). Either dihydroimidazole or subsequent 1,2-diamine scaffolds could be accessed by as imple modification of the workup procedure,w hile bamino alcohols derived from oxazolidine hydrolysis,signaling abroad and industrially relevant scope of future applications for this chemistry.
In the context of C À Ha minations,i ti si mportant to mention that DDQ as aneutral (closed-shell) photocatalyst [38]   can achieve C(sp 2 )-H aminations of electron-deficient arenes up to dichlorobenzenes,e ither under PRC [39] with ac ooxidant or under recycling e-PRC (see Section 2). [40] However,l arger,s ubstoichiometric quantities (10-20 mol %) are typically required and the evolution of hydrogen cyanide upon contact of this catalyst with moisture presents safety concerns. [41] DDQ is arather potent ground-state oxidant that can react with N-nucleophiles, [39] while TAC + and TPA catalysts do not suffer such issues.
C À Obond formations: TheLambert group demonstrated the acetoxyhydroxylation of aryl olefins (11)byradical ion e-PRC using TAC + .T ransition-metal-catalyzed dioxygenation reactions of olefins are well-known, but the authors noted the need for transition-metal-free approaches to obviate the expense and toxicity of certain metals.While electrochemical approaches using "cation pool" strategies are attractive, [42] in dioxygenations they oftentimes lead to the cleavage of deoxygenated products to carbonyl-and acetal-derived byproducts (Figure 10 A). [43] Direct electrolysis employing DMSO and DMF as nucleophiles to attack electrogenerated olefin radical cations has successfully afforded dihydroxylated olefins. [44] TheLambert groupsuse of radical ion e-PRC and acetic acid led instead to acetoxyhydroxylations,p roviding amild platform to access products such as 12 a in 71 %yield.
Direct electrolysis likely occurs to an extent under the relatively high potential of + 2.0 V; control reactions without light or TAC + did afford product 12 a but in low yields.Direct electrolysis at + 3.0 Vincreased the yield of 12 a to 40 %b ut led to over-oxidation and cleavage to aldehyde,k etone and acetal by-products.I nt he mechanism, 11 a is oxidized by *TACC 2+ to its radical cation, which undergoes nucleophilic attack by AcOH to afford benzylic radical 11 a (Figure 10 B). Oxidation of 11 a,e ither by anodic potential or by *TACC 2+ , induces intramolecular cyclization to 11 a",primed for attack by H 2 Ot oa fford 12 a.C yclic olefins were acetoxyhydroxylated to give 12 b-12 i in good to excellent (50-82 %) yields.A striking feature is the selectivity of this method compared to prior chemical and electrolytic reports.B enzylic methyl groups,a lcohols and aldehydes were all tolerated (12 j-12 l). Af ree alkylsulfide,aBpin ester, ap roduct-bearing styrene and electron-rich heterocycles were all tolerated in the syntheses of 12 m-12 s (31-78 %y ields). Yields of furan-and thiophene-containing 12 q and 12 r were low (31 %), but primarily due to lack of conversion. Selectivity was further exemplified in the late-stage acetoxyhydroxylations of various arylolefin-conjugated amino acids and complex molecules (Figure 11 A). Other acids were tolerated, affording 13 a-13 e in modest to good (38-55 %) yields (Figure 11 B). Interestingly,regioselectivity was inverted in these cases.Acrylic acid as an ucleophile underwent oligomerization in the synthesis of 13 e,which may be rationalized by Kolbe-type oxidation of acrylic acid and radical addition, [45] yet authors deemed this unlikely in the absence of base.Finally,multigram amounts of products were achieved via ar ecirculated flow setup (Figure 11 C). Electro-activation of TAC + was done in ab atch  undivided cell and the reaction mixture was recirculated through three CFL-irradiated coils with atotal residence time (R T )o f3min, affording up to 8.4 go f12 a and 3.7 go f12 w without appreciable yield losses after 20-36 h.
C(sp 2 )-X bond cleavages: On the reduction side,Lambert and Lin disclosed 9,10-dicyanoanthracene (DCA)a sa n electroactivated photoreductant for super-reductions of aryl halides, [18b] which was previously reviewed. [9] Simultaneously, the Wickens group reported an N-arylmaleimide (NpMI)a s ac atalyst that achieves the same transformation (Figure 12). [18c] Inspiration was drawn from seminal work of the Kçnig group on perylene diimide (PDI-b)p hotocatalysts, which are known to form stable radical anions (PDI-bC À )that can be photoexcited to reduce electron-poor aryl halides in ac onsecutive photoelectron transfer (conPET) mechanism. [4a] e-PRC reactions with DCA and NpMI catalysts extended the scope of aryl halide partners to electron-neutral and electron-rich, and conditions provide substantial advantages for such challenging SET reductions over traditional Birch-type conditions which are practically undesirable (dissolving alkali metals in liquid ammonia).
Due to the Beer-Lambert relationship,i ne -PRC SET takes place in bulk solution where the electrogenerated catalysts is irradiated, not at the electrode surface.T his leads to immediate deactivation of the photoexcited state;t hus e-PRC exhibits ak ey benefit in harnessing the reactivity of aryl(sp 2 )radicals.Direct electrolytic reduction of aryl halides that takes place at the electrode surface suffers the issue that aryl(sp 2 )r adicals are thermodynamically easier to reduce than their aryl halide precursors,leading inevitably to overall dehalogenation. [21a] Wickens and co-workers found that diimide catalyst architectures (PDI and naphthalene diimide, NDI)were ineffective catalysts for the reduction of electronneutral aryl halides, [18c] likely because the high stabilization of the radical anion prohibits super-reductive chemistry in its photoexcited state.I nt heir optimized e-PRC reaction conditions,t he new catalyst N-(2,6-diisopropylphenyl)naphthalene monoimide (NpMI)i sf irst reduced to its coloured radical anion by acathodic constant current of 0.8 mA. Upon photoexcitation with blue LEDs,p hotoexcited radical anion *NpMIC À engages aryl chlorides as challenging as 4-chloroanisole (15 a)i nS ET reduction to aryl radicals (Figure 12  Notably higher preparative yields were obtained compared to direct electrolysis (which gave noticeable decomposition), demonstrating the key selectivity benefit of radical ion e-PRC.W hen the debromination of 4-bromobiphenyl (E p red % À2.43 Vvs. SCE) [46] was used to optimize reaction conditions, ab is-N-(2,2',6,6'-diisopropyl)naphthalene diimide (NDI-d) precatalyst afforded dehalogenated product 21 a in al ower yield than NpMI did (Figure 13 A). Bardagi and co-workers recently reported conPET and e-PRC reductions of 4bromobenzonitrile (E p red = À1.95 Vvs. SCE) using amodified naphthalene diimide precatalyst (NDI-c). [19m] Thea ryl(sp 2 ) radical was trapped by an excess of benzene and afforded desired products (such as 21 b)a lbeit in low yields (< 20 %) (Figure 13 B). Though attention is needed to increase conversion and yields,t his represents ap otential alternative, milder set of conditions than transition-metal-free arylations of haloarenes requiring KOtBu and organic additives.S uch chemistry requires high temperatures to form electron donors in situ that initiate ab ase-assisted homolytic aromatic substitution (BHAS) chain reaction. [47] C(sp 3 )-O bond cleavages: Inspired by previous photocatalytic generations of carbanions [48] and direct electrolytic reductions of phosphinates in overall deoxygenations, [49] Barham, Kçnig and co-workers reported the first e-PRC reductive cleavage of C(sp 3 ) À Ob onds to access sp 3 -radicals  and sp 3 -centered carbanions. [19q] Phosphinates of aliphatic alcohols successfully underwent e-PRC reduction and C-O cleavage when N-(para-butoxyphenyl)naphthalene monoimide ( n BuO-NpMI)w as employed (Figure 14 A). Following cathodic reduction to its radical anion and photoexcitation, *[ n BuO-NpMIC À ]e ngages phosphinates (E p red %À2.4 to À2.6 Vv s. SCE) in SET.T hen, C(sp 3 )ÀOb ond cleavage of 22C À forms aC(sp 3 )radical, proposed to undergo further SET reduction likely by cathodic current (or by *[ n BuO-NpMIC À ]) to afford aC (sp 3 )carbanion. In the presence of an a-leaving group (chloride,b romide), elimination occurs in an overall reductive olefination (Figure 14 B).
Since phosphinate substrates derive from a-chloroketones and not aldehydes,the method has adifferent starting point to classic (Wittig-type) olefinations that can be leveraged to access cyclic and hindered olefins (Figure 14 C). Compared to acid-catalyzed or base-dependent eliminations of alcohols, the method proceeds at ambient temperature,tolerates basesensitive functionality and allows user control over the site of olefination. Forexample,asubstrate containing afree amide proton (23 i)was tolerated, as were esters (23 p-r). Formation of terminal olefin 23 c demonstrates the benefit over an acidcatalyzed elimination of at ertiary alcohol, which typically affords the most substituted olefin. In the absence of an aleaving group,o verall deoxygenation occurred as am ild and tin-free alternative to the Barton-McCombie reaction (Figure 14 B). Thep resence of ac arbanion intermediate was confirmed when ap hosphinate with a b-chloro atom led to cyclopropanation.
Stilbene 23 l could be accessed from the standard phosphinate 22,b ut also from ac yclic phosphinate derived from ad iol in ap hotoelectrochemical Corey-Winter-type olefination reaction that avoids high temperatures and hazardous reagents normally associated with this reaction.
Unsymmetrical stilbenes could also be readily accessed by this method, surprisingly with generally high Z-selectivities when n BuO-NpMI was employed (Figure 14 D). Olefinations and deoxygenations of phosphinates derived from benzylic or allylic alcohols were successful, while those derived from nonbenzylic/allylic aliphatic alcohols did not react despite almost identical redox potentials.R emarkably,a nd in stark contrast to the previous report of Wickens, [18c] aryl halides (chlorides and bromides) were tolerated under these conditions,despite their similar redox potentials to phosphinates (E p red (PhCl) = À2.78 V; E p red (PhBr) = À2.44 vs.S CE). Ap henol-derived phosphinate was also tolerated, contrasting with ap revious reports on C(sp 2 )-O bond cleavage by PRC with ap henothiazine catalyst. [50] 4-Vinyl benzoates of terpene natural products that are liquid crystals or fragrance compounds could be prepared by late-stage e-PRC olefination from the phosphinates derived from the 4-acetylbenzoate esters of the terpenes,g iving rise to potential monomers for polymerization (Figure 14 E). Here,t erminal olefination using base risks hydrolysis or E 2 elimination, while direct esterification is problematic due to thermal instability of 4-vinylbenzoic acid or its formulation with ar adical stabilizer (BHT).
To probe the mechanism behind stilbene E/Z-isomerism, control reactions with E-stilbene as an input revealed the critical importance of e-PRCat, light and potential on the E/ Z-isomerism. Luminescence spectroscopy revealed an anosecond-lived emitting state from [ n BuO-NpMIC À ]* that was catalytically inactive in the initial SET reduction step (its lifetime was not quenched by 22 a), but is likely responsible for E/Z-isomerism. One candidate for this emitter is aquartet state, 4 [ n BuO-NpMIC À ]*, that results from intersystem crossing from ah igher energy doublet state.T he energy of this emitting state (E 0-0 )w as identical to transition metal photocatalyst triplet energies known to effect E/Z-photoisomerism of olefins by triplet-triplet energy transfer, [51] and was within range of the triplet energies of stilbenes. Theauthors sought to determine why n BuO-NpMI was an effective catalyst for all (benzylic/allylic) substrates attempted, while NpMI was ineffective for most substrates, despite the identical ground-state reduction potentials (E 1/2 = À1.3 Vv s. SCE) and UV-vis properties of both e-PRCats,a s their neutral or radical anion forms.Byacombination of CV, EPR and computational investigations examining C(sp 3 )ÀO bond dissociation free energies (BDFEs), the authors found that the C(sp 3 ) À Ob ond cleavage was likely the reactivitydetermining step,since the initial SET step was successful for both e-PRCats.M irroring their study on TPAC + s, [20] irradiation of the near-IR UV-vis bands of the catalyst gave no conversion, suggesting anti-Kasha photochemistry from ah igher-order excited doublet state.G iven 1) the known ultrashort lifetimes of similar excited-state radical anions (*PDIC À ,* NDIC À ,T able 1) that prohibit their diffusioncontrolled photochemistry,2 )the fact that catalyst architecture was able to influence aC ÀOb ond cleavage step and 3) the involvement of higher-order excited doublet states,the authors proposed ap reassembly of phosphinate substrates with n BuO-NpMIC À .H owever,i nc ontrast to the authors earlier study on TPAC + s, [20] no spectroscopic (UV-vis/EPR) changes were detected when electrogenerated n BuO-NpMIC À was mixed with phosphinate 22 a.T he authors rationalized that precomplexation may occur at the N-aryl moiety,which is not spectroscopically detectable since the spin density and chromophore of the radical anion are localized on the naphthalene moiety,o rthogonal and electronically disconnected from the N-aryl moiety.I ns upport of this proposal, as trong correlation was found between decreasing steric hindrance at the N-aryl moiety ortho-positions of the e-PRCats and increasing reactivity of 22 a as amodel substrate ( Figure 15). Va riations in the naphthalene moietyse lectronics (e-PRCats 25 a,b)did not improve the activity compared to NpMI.A na dditional two alkoxy substituents at the metapositions (25 c)d ecreased activity relative to n BuO-NpMI. DFT calculations (wB97X-D) found several candidate preassemblies between 22 a and both NpMIC À and n BuO-NpMIC À . In all cases,converged structures resembled a"pincer" where two aromatic groups of 22 a interact with the N-aniline moiety by a T-p and a p-p interaction ( Figure 16). Regardless of the converged candidate structure,binding free energies (DG bind ) were always more favourable for n BuO-NpMIC À than for NpMIC À .F inally,e xcited-state calculations (DFT-MCRI), in good agreement with experimental UV-vis of n BuO-NpMIC À , revealed that the transition at % 430 nm (D 0 !D n )i nvolved ac harge transfer from the naphthalene moiety to the Naniline moiety (Figure 17). Localization of electron density at the N-aniline in this excited state (D n )i st hus exactly where required for rapid intra-assembly SET.Amore intimate precomplexation of 22 a and n BuO-NpMIC À likely promotes C(sp 3 )-O bond cleavage.
C(sp 2 )-O and C(sp 3 )-+ NR 3 bond cleavages: Wickens and co-workers recently extended their aryl halide reduction methodology to the reduction of aryl pseudohalides. [19t] Phosphonated phenols and anilinium salts could be reduced by e-PRC in an overall hydro-defunctionalization reaction as the main theme of the study (Figure 18 A). Alternatively,the aryl radical could again be intercepted by triethylphosphite, N-methylpyrrole or B 2 pin 2 (Figure 18 B), inspired by the authors previous work [18c] and that of the groups of Larionov [50] and Kçnig, [52] who demonstrated reductive cleavages of strong bonds and borylation of aryl radicals via photocatalysis involving proton-coupled electron transfer [50] or an EDA complex. [52] Reactions were conducted in divided H-cells under ac onstant potential, with 10 mol %e -PRCat and 2equiv.ofEt 3 Npresent in the cathodic chamber as aterminal reductant. As an example synthetic application, aphenol was used to direct the Friedel-Crafts reaction of 30 with 31 followed by e-PRC hydro-defunctionalization to 28 e and global O-demethylation to afford tricyclic resorcinol cannabinoid agonist 32 in 14 %y ield over three steps.T his   application followed the work of Makriyannis and co-workers [53] but substituted dissolving Li metal reduction with e-PRC.A lthough isophthalonitrile structures are widely employed in PRC,this report, [19t] together with aconcurrently reported conPET variant, [19s] constitute the first reports of isophthalonitrile radical anions in reductive catalytic transformations. 4-DPAIPN structurally resembles cyano-substituted triarylamines used in oxidative e-PRC; [20] it is interesting to find the acyclic triarylamine architectural family can be used in both reductive and oxidative e-PRC reactions. Current-based analysis of the reaction rate over time evidenced the enhanced stability of 4-DPAIPN compared to previously reported catalyst (NpMI). NpMI decomposed over time under the reaction conditions,d ecreasing conversion rate.Asthe reaction progressed, the rate increased again, meaning ad ecomposed form of the catalyst is also catalytically (albeit less) active.T his concurs with Nocera and coworkers recent report [19r] which analysed the electro-decomposition product of NpMI (irreversible CV wave at À2.3 V). Electrolysis at U cell = À3.0 Vp rovided as pecies absorbing at (l max =)480-500 nm and luminescing at (l max =)560-580 nm, which matched the spectra of as pecies formed when NpMI was chemically treated with NaBH 4 (in DME) or TBAF (in DMAc). XRD of the species chemically formed with NaBH 4 revealed ahydride adduct 33 (Scheme 2). Irradiating 33 with 440 nm in the presence of stoichiometric 4-methylchlorobenzoate (15 f)a nd excesses of P(OEt) 3 or N-methylpyrrole led to detection of 17 f and 19 f by 1 HNMR spectroscopy, confirming that 33 can serve as ap hotoreductant. Nocera questioned the previously proposed involvement of radical anion photocatalysts, [18c] claiming that NpMIC À is too shortlived (t = 24 ps) to permit diffusion-controlled photochemistry,y et the electro-decomposed emitting species assigned as 33 is sufficiently long-lived to do so (t = 20 ns). However, photoreductive activity of NpMIC À is not ruled out since Wickens kinetic analysis clearly evidenced the presence of multiple (at least two) active catalysts during e-PRC reactions. [19t] Although the excited state of 33 was effectively quenched by activated aryl chloride 15 f,PhCl was ineffective (radical trapping of PhCl was not reported). Adduct 33 or related species may be acandidate for the emitting state (ES 1 ) in Barham, Kçnig and co-workers study, [19q] but comparisons are still questionable due to 1) UV-vis absorptions (l max )o f NpMIC À agreeing within AE 10 nm despite different solvents (DMAc [19r] vs.MeCN) in the two studies, [19q] yet ES 1 differing in peak shape and l max (by ca. 40 nm); [19q] 2) notably lower cell potentials for electrolysis (U cell = À1.6 V) used both in spectroscopy and reactions where U cell = À3.0 Vl ed to intractable complex mixtures. [19q] Thep roposal of diffusioncontrolled SET photochemistry of adducts like 33, [19r] while intriguing,c annot rationalize 1) clear structure/activity relationships at the N-aniline of NpMI-type catalysts, [19q] 2) reduction of non-activated aryl chlorides like PhCl/4-chloroanisole (15 a), [18c] 3) clear quenching of PRCatC À sUV-vis/EPR signals upon irradiation in the presence of substrates [4a, 19q, s] and 4) absence of quenching of longer (ns or ms)-lived emitting species derived from radical anion PRCatCsi no ther reports. [4g,19q] Closed-shell anionic species may be reservoirs/ precursors to radical anion photocatalysts as proposed by Miyake and co-workers. [4g] Future perspectives: Radical ion e-PRC is ar apidly expanding field that offers 1) new opportunities in synthetic transformations,i ncluding complex molecule functionalizations,2 )access to ultrahigh-energy redox processes under exceedingly mild conditions to cleave or form strong bonds, and 3) new opportunities for selectivity that differ from conventional parameters in SOE and PRC (redox potentials). While radical ion catalysts and their excited-state behaviours are likely interchangeable concepts between conPET and e-PRC manifolds,d ownstream chemistry of at arget substrate following its initial SET differs.H ere,e -PRC offers the attractive and unique benefit of user-potentiocontrollable radical polar crossover.W ith the key importance of preassembly in radical ion photocatalysis established, aparticularly exciting prospect is leveraging factors in preassembly formation to guide 1) SET chemoselectivity and/or 2) following bond cleavages/formations;s imilar to the "lock-and-key" concept of enzyme catalysis.Indeed, noncovalent interactions (dispersion, p-p stacking) historically viewed as "weak interactions" are receiving increasing attention in catalysis. [54] Another emerging theme is access to excited states higher than the first excited state in anti-Kasha photochemistry. [4g, 19q, 20] This phenomenon in itself corroborates substrate/photocatalyst preassembly to rationalize SET taking place more rapidly than internal conversion of the higherorder excited state (D n !D 1 ). Thereby,anti-Kasha radical ion photochemistry harnesses the full visible photon redox energy,w here in conventional PRC much is lost to internal conversion. Consequentially,t he redox window of transformations is dramatically expanded-substrates beyond typical solvent windows (MeCN/DMF) are engaged. The ultrafast timescale within which SET must occur shields the bulk reaction mixture from extreme redox potentials generated in situ-a clear benefit compared to high cell potentials constantly applied across reaction mixtures in direct electrolysis.
In theory,p reassembly and anti-Kasha photochemistry should raise photochemical reaction (quantum) efficiency.In practice,l ong reaction times often plague radical ion e-PRC reactions.Agreater understanding of factors involved in the preassembly is essential for the field to achieve higher quantum efficiencies.W hile it is straightforward to assign quantum yields and lifetimes to closed-shell excited states, interrogation of open-shell doubled excited states is ac hallenging endeavour demanding more sophisticated spectroscopic and theoretical techniques.T his said, solvated electrons [4g] and decomposition of radical ion PRCats to closedshell PRCats as in *DCAC À and *NpMIC À[19p,r] must be probed as alternative mechanisms as the field continues to evolve. Finally,limited reports of scalability in radical ion e-PRC are likely due to practitioners requiring 1) divided cell configurations to spatially mitigate nonproductive half-reactions and 2) constant potential desirable for selective e-PRCat activation.

Photocatalyst Electro-recycling
Thes econd subcategory of e-PRC,d ubbed "recycling e-PRC", involves the turnover of ap hotocatalyst that is ak nown photoredox catalyst (PRCat) in PRC,a nd that is acolored species in its ground, neutral state. Figure 19 depicts the structures of PRCats used in recyclinge -PRC;t heir photophysical properties are thoroughly detailed elsewhere. [2c,e, 38, 55] In recycling e-PRC,t he available "redox window" is no wider than it is for PEC.I nstead, the key benefit is the replacement of sacrificial oxidants or reductants in the photocatalyst turnover with electrochemistry ( Figure 20), which can (or whose by-products can) interfere with downstream chemistry and/or complicate separation of desired products. [9,56,57] This is not to say that sacrificial oxidants or reductants are completely avoided;t hey can be required by the counter-electrodesr eaction. However,h ere protons or water typically serve as much milder,a tomeconomical sacrificial oxidants or reductants (respectively). Moreover,adivided cell configuration provides the opportunity to spatially separate sacrificial redox agents from the desired catalytic reaction in the product-forming chamber.
In as eminal report that was previously reviewed elsewhere, [9,10] Xu and co-workers reported recycling e-PRC using Fukuzumiscatalyst Mes-Acr + for aMinisci-type coupling of alkyltrifluoroborates with heteroarenes in an undivided cell. [56] Inspired by this report, different applications of recycling e-PRC have emerged in recent years. [40,57] Recycling e-PRC benefits from the use of well-characterized and longlived (nanosecond to millisecond) closed-shell excited states. As opposed to radical ion e-PRC where divided cells are generally employed, recyclinge-PRC typically uses undivided cells or modular batch/recirculated flow setups.I nf act, most

Angewandte Chemie
Reviews reports in recycling e-PRC easily achieve gram-to multigramscale reactions-indicating the practical accessibility and robustness of this synthetic photoelectrochemical technology.
C(sp 2 )-H trifluoromethylation: In an elegant example of photocatalyst electro-recycling,A ckermann and co-workers reported the PEC C(sp 2 )-H trifluoromethylation of arenes and heteroarenes under anodic current and with Langlois reagent, CF 3 SO 2 Na (34) ( Figure 21). [57a] Upon visible-light photoexcitation of the Fukuzumi-type organic dye, Mes-Acr + (catalyst a)the authors proposed that the excited-state * Mes-Acr + engages in SET with the CF 3 SO 2 anion 34 to furnish the reduced acridinyl radical form of the catalyst and the CF 3 SO 2 radical 34'.L oss of SO 2 generates the active trifluoromethyl radical 36,w hich attacks arene substrate 1a to form arene radical 37.SET oxidation of 37,either by Mes-Acr + or by the anode,f orms cation 37 + which loses ap roton to form the trifluoromethylated product 35 a.T he protons generated could undergo cathodic reduction to form H 2 to complete the circuit. Meanwhile,ground-state Mes-Acr + photocatalyst is then regenerated by anodic oxidation of its acridinyl radical form at the C (felt) anode under constant current conditions (4.0 mA). LiClO 4 was selected as an electrolyte which eases the separation of products,w hile another set of conditions were developed using Ru(bpy) 3 2+ (catalyst b). This method enabled the C(sp 2 )-H trifluoromethylation of ar ange of arenes including electron-rich and electron-poor arenes and various heteroarenes (affording products such as 35 b-35 e, Figure 21 D). As apioneering example of the transfer of PEC to ac ontinuous flow setup,a ne lectrochemical flow coil consisting of ac arbon felt (C felt )a node and an ickel cathode was employed prior to ap hotochemical fluoropolymer coil added in sequence (Figure 21 B). Recirculation of the reaction mixture over 12 ha fforded 35 a in 76 %y ield using catalyst a (conditions employing catalyst b were inferior). Speaking to the facile integration of continuous flow with process analytical technologies (PATs), [58] the authors used an in-line NMR spectrometer to monitor the reaction and observed the Wheland-type arenium cation 37 + as al onglived intermediate via 19 Fa nd 1 HNMR spectroscopy.T his demonstrated ak ey opportunity for continuous flow in the investigation of SET reaction mechanisms by its ability to provide asteady-state concentration of reactive species.
C(sp 2 )-C(sp 3 )c oupling: Building upon their previous work in photoelectrochemical C(sp 2 )-C(sp 3 )M inisci-type coupling of heteroarenes 30 with alkyl trifluoroborates, [56] and as the first synergy of PEC with cerium photocatalysis, [59] Xu and co-workers reported direct decarboxylative CÀH alkylation of heteroarenes 38 using an RVCa node and CeCl 3 ·7H 2 Oa sap hotocatalyst precursor ( Figure 22). [57b] Initial anodic oxidation of Ce III to Ce IV occurs (Figure 22    demonstrated by the late-stage functionalizations of various N-heteroarenes including bioactive molecules such as 40 e (fasudil), aR ho-associate protein kinase inhibitor (Figure 22 C). [60] In the same report, Xu and co-workers further exploited decarboxylative radical formation in the PEC carbamoylation of heteroarenes using an RVCa node and a 4CzIPN photocatalyst (Figure 23). [57b] Upon photoexcitation of 4CzIPN with blue LEDs,S ET oxidation of oxamate substrate 45' by 3 4CzIPN*f orms 4CzIPNC À and, upon decarboxylation, carbamoyl radical 47 which adds to protonated substrate 38-H resulting in radical cation 48 (Figure 23 B). Reduction of 48,e ither by the Pt cathode or by 4CzIPNC À ,a ffords intermediate 49.A nodic oxidation of 49 with loss of H 2 was proposed to result in aromatization to protonated product 46-H. Alternatively,the authors proposed deprotonation of 48 would afford radical 50 which could be oxidized by ground-state 4CzIPN or by the anode.T he substrate scope featured various examples of oxamic acids (bearing primary,secondary and tertiary N-substituents) and various electron-deficient N-heteroarenes (affording products such as 46 a-e), including the late-stage functionalization of antihistamine Loratadine [61] to afford 46 e (Figure 23 C). Finally,X ua nd co-workers extended their method to alkyl oxalates as precursors to alkyl radicals in the absence of at ransition metal catalyst ( Figure 24). H owever,C -(sp 2 ) À Oand C(sp 2 ) À Nbond formations can also be achieved under recyclinge -PRC. DDQ as an eutral photocatalyst is ap owerful oxidant in its long-lived triplet excited state (+ 3.18 Vv s. SCE). [38] Despite its absorbance maxima at % 400 nm, DDQ is successfully photoexcited with longer wavelength blue (455 nm) light. Prior to any contemporary reports of recycling e-PRC,K çnigsg roup achieved the photocatalytic oxidation of electron-deficient arenes by 3 DDQ*i nt he presence of tert-butyl nitrite and molecular oxygen, [39] which was reviewed previously. [9] Inspired by this work, and by their earlier work on recycling e-PRC for the S N Ar-type arene azolations [62] (reviewed previously), [9] the Lambert group recently reported arene acetoxylation using DDQ as photocatalyst under recycling e-PRC ( Figure 25). [40a] Benzene (1e)w as hydroxylated to phenol (54 a)i nt he optimization study.Control reactions confirmed the necessity of light, catalyst and potential in the reaction (Figure 25 A). Direct electrolysis at ap otential of U cell =+3.2 Vg ave no product after 48 h( presumably,d ecomposition occurred under direct electrolysis,aspreviously reported in the radical ion e-PRC azolation of benzene), [18a] confirming the superiority of photoelectrochemistry in this reaction.
In the mechanism, SET oxidation of arene 1 by 3 DDQ* was proposed (Figure 25 B), followed by nucleophilic addition of heteroatom partner 53 to radical cation 1C + . DDQC À formed after SET is protonated and engages 55 in HATt oa fford product 54.Arange of electron-deficient arenes were hydroxylated and acetoxylated to afford products 54 a-g in modest to very good (30-76 %) yields (Figure 25 C). Remarkable selectivities were observed:a liphatic alcohols,t erminal alkynes and benzylic positions were all tolerated, where these positions would not likely tolerate direct electrolysis.A minations were also possible with free amides,c arbamates and ureas,a ffording products 56 a-d in satisfactory to high (49-80 %) yields,a lthough the scope of amination partner was relatively narrow in each of these classes.I na ni ntriguing competition experiment (Figure 26 A), benzene (1e)w as selectively hydroxylated while anisole (1k)a nd trifluoroto-

Angewandte Chemie
Reviews luene (1j)were untouched by the recycling e-PRC conditions. This highlights the importance of matching excited-state PRCat redox potentials to the substrate.T rifluorotoluene is beyond the scope of 3 DDQ* (and *TACC 2+ , > 3.3 Vv s. SCE, but could be engaged by potent *TdCBPAC + ,upto+ 4.4 Vvs. SCE). On the other hand, anisole is easily accessible and likely oxidized by 3 DDQ*.I ti sp lausible that back electron transfer (BET) is rapid enough to prohibit downstream chemistry of 1kC + . [63] In general, this recycling e-PRC was superior to the preceding PRC-only report [39] where cheap constant potential obviated the need for the sacrificial oxidant (molecular oxygen) and expensive tert-butyl nitrite,b oth hazardous reactants that require considerable safety considerations for scaling up. [64] In this respect, the Lambert group successfully scaled the benzene-to-phenol hydroxylation reaction in ar ecirculated continuous flow setup (Figure 26 B). Electroregeneration of DDQ was achieved in abatch undivided cell, while the reaction mixture was recirculated through blue LED-irradiated coil with aresidence time (R T )of3min "per pass". By extending reaction time and adding additional photocoils,t he reaction was successfully scaled from 0.4 to 15 mmol without appreciable loss in the yield of 54 a.Thereby, recycling e-PRC benefitted safety and cost-efficiency.
Xu and co-workers also reported arene heteroamination using DDQ under recycling e-PRC conditions (Figure 27 A). [40b] Compared to the Lambert groupsr eport, although conditions employed ah igher loading of DDQ (20 mol %), loadings of electrolyte (0.1 equiv.) and amination partner (2 equiv.) were markedly lower, possibly due to the use of constant current (2 mA) to drive the reaction. The focus was on aminations;h ydroxylations and acetoxylations were not investigated. Themechanism was as aforementioned (Figure 27 B). Am uch broader scope of amination partners was reported, including azolations,a ffording products in modest to high (36-70 %) yields (Figure 27 C). Ag ram-scale batch reaction worked (Figure 27 D). Like Lamberts report, [40a] anisole was unreactive. 3 DDQ*d id not engage methyl benzoate-the upper redox limit was dihalobenzenes.
Although these conditions [40b] were unable to engage very electron-deficient arenes (radical ion e-PRC with TPAswere able to engage up to acetophenone), [20] the yields of azolated dihaloarenes were higher than in radical ion e-PRC reports [18a, 20] and, notably,a rene 1 was the limiting reagent as opposed to the requirement of excess arene in previous oxidative radical ion e-PRC (typically 1mL) [18a,20] and conPET (typically 8mL) [4] reports.The excess arene required in radical cation e-PRC is likely due to the requirement of precomplexation between ground-state radical cation and arene for successful photochemistry,whereas 3 DDQ*islonglived enough to engage in outer-sphere diffusion-controlled photochemistry.Byincreasing the electrode surface area and applying ah igher constant current (52 mA), Xu and coworkers scaled the reaction up to produce gram quantities of 56 m deriving from the amination of benzene (Figure 27 D) without any flow setup.T oshed light on the regioselectivity of nucleophilic addition, Xu and co-workers performed DFT calculations on the radical cations of haloarenes.C alculated LUMOs showed, in all cases,t hat positions para to the halogens had larger orbital coefficients than other positions, rationalizing for the first time regioselectivity for the nucleophilic addition in these radical cation S N Ar reactions of electron-deficient arenes.
Scalable synthesis of acridinium salts: Finally,recycling e-PRC was used by Xusg roup to synthesize al ibrary of 3/6substituted acridinium PRCats from an acridinium core 60  Impressively,d ifferent trifluoroborate salts can be employed at each step to furnish unsymmetrical 3,6-disubstituted acridinium salts.I nt he photochemical step (step 1) of the proposed mechanism (Figure 29 A), photoexcitation of 60, followed by SET reduction of *60 by the alkyl trifluoroborate salt, affords reduced form 64.A ddition of trifluoroboratederived alkyl radical to 64 affords 65.I nt he electrochemical step (step 2), TEMPO is oxidized to 66 which engages 65 in oxidation, affording TEMPO-H and radical 61.T EMPO-H undergoes reduction at the cathode to liberate H 2 and TEMPO À ,t he latter of which is transformed back to TEMPO by SET at the anode.B yp assing a0 .05 Ms olution of 60 (R = Ph) through af low photocoil into an electro-chemical batch reactor for the first functionalization, then washing with KPF 6 before repeating for the second functionalization (see reference for details), 10.9 g( 80 %o ver both steps) of N-Ph Mes-Acr (structure in Figure 1) was successfully synthesized. In the first example of an end-to-end, semicontinuous homogeneous synthetic photoelectrochemical flow process,X ua nd co-workers transformed 2.0 go f60 (R = Ph) into 1.4 g( 51 %o ver both steps) of N-Ph Mes-Acr (Figure 29 B). Here,t he authors found that Et 3 Nw as necessary to improve conversion in the electromediated dehydrogenation of 65.H owever,E t 3 Nw as detrimental to the photochemical step,s ob ases were neutralized in situ by TfOH before the subsequent photocoil. Acollection flask was required after the first electrochemical flow reactor in order to purge gas bubbles (H 2 ).
To demonstrate the value of their 3,6-disubstituted product acridinium salts (62), Xu and co-workers compared av ariety of established photocatalysts in the photocatalytic decarboxylative conjugate addition of 67 to 68 (Figure 29 C). While unsubstituted Mes-Acr + (60,R = Me), an iridium photocatalyst and 4CzIPN gave only trace products,t he 3,6di-tert-butyl-substituted acridinium salt (N-Ph Mes-Acr)w as effective and novel catalyst 62 b gave anear-quantitative yield of 69.I nterestingly,t he nature of substituents at the 3,6-  positions was found to dramatically influence the lifetime. Thelifetime of N-Ph Mes-Acr (T 1 )is% 6.1 ns (Table 1), while the lifetime of 62 c was much longer (t = 30.7 ns) and is the longest lifetime ever reported for an acridinium salt. Enhanced lifetimes may provide arationale for the increased activity of acridinium PRCat derivatives synthesized herein.
Future perspectives: Recycling e-PRC continues to pave the way to enhanced sustainability and safety in photocatalysis,b yr eplacing sacrificial redox agents with cheap, non-hazardous electrochemistry.I nc ontrast to radical ion e-PRC which typically uses 0.1 Mo fs upporting electrolyte, ar ecurring theme in recycling e-PRC is the ability to use lower electrolyte loadings,typically < 1toseveral equivalents, likely as ar esult of an undivided cell setup with shorter interelectrode distances.A pplicability of the technology is clear:f rom the late-stage functionalizations of pharmaceutically relevant molecules to the synthesis of novel acridinium scaffolds representing attractive functional materials and photocatalysts.R ecent advances demonstrate how recycling e-PRC in recirculated or semicontinuous flow systems enables scalability of reactions,a tl east to multigram scales. However,t he use of recycling e-PRC in the reductive direction is yet to be explored. Moreover,i nm ost reports of oxidative recycling e-PRC,hydrogen gas is evolved and is not utilized downstream. Harnessing the by-product at the counter-electrode within as ynthetic transformation (for example,u tilizing H 2 in as ubsequent catalytic hydrogenation, [7b, 65] isolating it, or conducting ap aired electrolytic reaction [66] )w ill improve Faradaic efficiencya nd enhance sustainability further. In ag eneral sense,t he elimination of sacrificial cathodic or anodic processes in favour of ap aired electrolytic system is deemed necessary to encourage uptake of electroand photoelectrochemistry in process chemistry. [67]
Lambert and co-workers reported aphotoelectrochemical HATactivation of C(sp 3 ) À Hbonds of alkyl ethers (70,75). [70a] Here,the trisaminocyclopropenium ion (TAC + )developed by the group for radical ion e-PRC super-oxidations was employed as ac atalyst. Upon anodic oxidation of TAC + and photoexcitation of TACC 2+ ,t he authors proposed that the Figure 29. A) Proposed mechanism for each step. B) End-to-end continuous flow photoelectrochemical synthesis of functionalizeda cridinium salts. C) Utility of 3,6-disubstituted acridinium salts in photocatalysis. excited radical dication *TACC 2+ engaged C(sp 3 )ÀHb onds of ethers in HATt oa fford sp 3 radical 79 (Figure 31 D), which either 1) engaged in aM inisci-type reaction with quinoline derivatives like 71 to afford products 72-74 (Figure 31 A); 2) underwent 1,4-addition to electron-deficient alkenes or alkynes 76 to afford products 77 a-d (Figure 31 B);o r 3) underwent further oxidation to an oxocarbenium ion and then nucleophilic azolation with N-heteroarenes 2 to afford products 78 a-d (Figure 31 C). Yields were generally satisfactory to excellent for all downstream transformations (31-89 %). Following the proposed HATs tep, TAC-H 2+ releases aproton which is reduced to hydrogen at the cathode.
Theo bservation of ak inetic isotopic effect (KIE, k H/D = 3.0) confirmed rate-determining CÀHc leavage.D espite the apparent bulkiness of *TACC 2+ as aH AT agent, the authors proposed that high selectivity for secondary C(sp 3 )-H > tertiary C(sp 3 )-H corroborates HATa st he mechanism. However,the lifetime of *TACC 2+ is unknown, and as adoublet excited state,itislikely ultrashort-lived such that HATmight require preassembly of ethers and TACC 2+ .
TheX ug roup reported an elegant photoelectrochemical pathway to activate C(sp 3 )ÀHb onds of substrates (81)u sing chlorine radicals as HATagents generated in situ within an undivided cell. [70b] Theg enerated C(sp 3 )r adical (83)a gain participated in Minisci-type reaction with heteroarenes (80) (Figure 32 A). In their proposed mechanism, anodic oxidation of Cl À to Cl 2 first occurs.Subsequent light irradiation leads to homolysis of Cl 2 , [73] generating ClC as ap owerful HATagent. Continuous in situ generation of Cl 2 by anodic oxidation and its consumption in the reaction avoids the direct use of toxic Cl 2 gas (Figure 32 B). Since the bond dissociation enthalpy (BDE) of HCl is high (102 kcal mol À1 ), [71] ClC is ap owerful HATagent that successfully engages various C(sp 3 )ÀHbonds  in at hermodynamically favoured process.A sw ell as ethers, radicals were successfully accessed from C(sp 2 )-H positions of formamides and even C(sp 3 )-H positions of hydrocarbons were successfully engaged. Thes ubstrate scope with respect to both radical precursors and heteroarenes was relatively broad, tolerating many sensitive functional groups and affording products like 82 a-f generally in modest to excellent (40-94 %) yields (Figure 32 C). Thea bility to replace photoor electroactivated photocatalysts (TBADT, TAC + )w ith ClC from HCl readily available in all laboratories is ak ey advantage of this method. Without noticeable erosion in the yield of 82 f,abatch reaction was done on a4 5mmol scale. Even a1 22 mmol batch reaction was achieved by recirculating the reaction mixture through areservoir (Figure 32 D). As per previous Minisci-type photoelectrochemical reactions (Section 2), late-stage functionalizations were achieved for complex and bioactive molecules such as dihydroinchonidine, fasudil, roflumilast and even an adenosine analogue (Figure 32 C). While this process involved the photoelectrochemical generation of chlorine radicals,itisworth noting that the photoelectrochemical generation of iodine radicals for HAT at benzylic positions was reported earlier by Stahl and coworkers, [74] and was reviewed previously. [9] It is well established that the photoexcited state of tetra-nbutylammonium decatungstate, TBADT (( n Bu 4 N) 4  TBADT is regenerated from TBADT-H by anodic oxidation at av ery low applied potential (DE WE-RE =+0.15 V) in ad ivided cell (Figure 33 B), such that the reaction could even be driven successfully by two AAA batteries.Laser flash photolysis confirmed the quenching of 3 TBADT* by 81 and by ad ihydro-precursor of 86,b ut not by the LiNTf 2 electrolyte.
Lei and co-workers group reported ap hotoelectrochemical oxidative azidation of C(sp 3 ) À Hbonds of substrates (81) in an undivided cell, in which electrochemistry plays multipole roles (Figure 34 A). [70d] Upon photoexcitation of an aromatic ketone catalyst (Cat.)a nd intersystem crossing, the long-lived triplet excited state abstracts aC(sp 3 )-H atom from either activated benzylic positions (when Cat. = 9fluorenone or bis(4-methoxyphenylmethanone) or unactivated hydrocarbons (when Cat. = DDQ), affording an C(sp 3 ) radical 83 which reacts with aMn III -azide complex species to form the azidated product 88.T his Mn III -azide complex is electrogenerated in situ from its precursor Mn II -azide complex. Thea uthors proposed that, upon photoexcited HAT, reoxidation back to the carbonyl is simultaneously mediated by anodic oxidation (Figure 34 B), while liberated protons are reduced to hydrogen at the cathode.Benzylic and unactivated hydrocarbons were azidated to afford ab road scope of products such as 88 a-e in modest to excellent (31-99 %) yields (Figure 34 C). Reactivities and selectivities were generally higher for tertiary benzylic C(sp 3 )-H azidation compared to secondary benzylic azidations.T he azidation of acumene derivative was successfully achieved on gram scale after extended reaction time (71 %y ield after 72 h). Latestage azidation of bioactive molecules such as ad ifferin precursor and ibuprofen methyl ester (affording 88 d,e)w as accomplished when DDQ was used as acatalyst.
In their mechanistic studies,t he authors found by cyclic voltammetry that the oxidation of NaN 3 occurred at al ower potential than substrates and photocatalysts.This led them to  suggest the azide radical was preferentially formed under the anodic conditions of the reaction. TEMPO was found to inhibit the reaction and by EPR the authors were able to detect the azide radical. In their optimization, authors found that the reaction did proceed appreciably in the dark without catalyst (45 %y ield), although the rate of reaction was markedly accelerated by irradiation (61 %y ield). To rationalize progress of the reaction in the dark, they suggested that the azide radical can also engage in HAT.

Reactor Platforms
Of key importance to the discovery of new PEC reactions and their uptake in industry is the requirement for robust reactor platforms that deliver reproducible chemistry.T othis end, a3 D-printed photoreactor accommodating two interchangeable high-power commercial LED lamps and up to six vials was reported by Schiel and co-workers at Boehringer Ingelheim (Figure 35 A). [75] Thereactor gave excellent reproducibility,b oth vial-to-vial and of literature PRC reaction yields.B ye xchanging the vial holder module,t he reactor accommodated an undivided cell photoelectrochemical reactor vial driven externally by abenchtop commercial potentiostat. [76] Ther ecycling e-PRC heteroarene carbamoylation [57b] was successfully reproduced, affording 46 f in 83 %y ield (Figure 34 B). Since 3D printing has previously converted ac ommercial benchtop potentiostat into ac ompact flow electrochemical reactor, [77] it is likely that commercial 3Dprinted photoelectrochemical batch and flow systems will soon be available to practitioners,b oth in divided and undivided modes.
Ak ey challenge in the discovery of molecular photoelectrochemical reactions is the number of variables that arise when combining SOE with PRC.L ight wavelength, light intensity,c urrent, potential, electrode materials,d ivided vs.
undivided cell, temperature,c atalyst choice and loading, electrolyte choice and loading all influence the reaction outcome.The critical influence of light intensity and reaction temperature in PRC has been recently highlighted. [78] Highthroughput screening is apowerful tool for reaction discovery, allowing multiple variables to be simultaneously explored. [79] Lin, Lehnherr,Kalyani and co-workers developed acompact, high-throughput microscale electrochemical reactor that was successfully applied to screen up to 24 conditions at once in ar adical ion e-PRC reaction under constant potential ( Figure 36), including control reactions. [80] Ther eactor increased the reaction rate threefold, likely due to increased transmission of light on the microscale and LED optical power. Vial-to-vial reproducibility was high (identical reactions with an average 75 %y ield gave a5%s tandard deviation) confirming the robustness of the system for discovery.

Summary and Outlook
Synthetic photoelectrochemistry (PEC) involving in situ generated homogeneous photocatalysts is ar apidly growing research frontier in single-electron transfer chemistry and organic synthesis.R adical ion electrochemically mediated photoredox catalysis (e-PRC) pushes the boundaries of the redox window further than ever before,a chieving unprecedented oxidations and reductions in acontrolled, selective manner that can be leveraged to construct, or cleave,strong bonds.While controversy continues to surround the operating mechanisms of reactions proposed to involve radical ion  photocatalysts,substrate-catalyst preassembly provides aconvincing interpretation of reactivity patterns and offers exciting new opportunities for selectivity control that challenge conventional parameters like the thermodynamic redox potential. Recycling electrochemically mediated photoredox catalysis forges ahead of conventional photoredox catalysis in improving sustainability,scalability,safety and cost-efficiency of reaction conditions,w ith sacrificial chemical redox additives being substituted for cheap,b enign electricity.B oth subcategories of e-PRC leverage the general selectivity benefits of substrate engagement with ap hotoexcited state in bulk solution, mitigating against over-reduction or overoxidation processes and harnessing reactive intermediates that may normally undergo further redox chemistry at electrode surfaces or lead to grafting/passivation. Thecombination of PEC and HATo pens new opportunities for synthesis,i ncluding new photoexcited HATa gents or electro-recycling established ones,aswell as providing new access to ground-state HATa gents from inexpensive,a bundant precursors.F inally,i na ddition to promising initial efforts in the scale-up of these chemistries in recirculated or continuous flow,new reactor platforms designed for high reproducibility, control of reaction variables and high-throughput experimentation pave the way to photoelectrochemistry becoming atool accessible to both academic and industrial chemists alike.