Photoredox catalysis

Photochemistry of transition metal photocatalysts

The activity of a photoredox catalyst can be described in three steps. First, a molecule of the catalyst in its ground state, where the electrons are distributed among the lowest-energy combination of states, interacts with light and moves into a long-lived excited state, where the electrons are not distributed among the lowest-energy combination of available states. Second, the photoexcited catalyst interacts by an outer sphere electron transfer process to "quench" the excited state and to activate one of the other components of the chemical reaction. Finally, a second single electron transfer occurs to return the catalyst to its original oxidation state and electron arrangement.

Photoexcitation

The first step of this process, photoexcitation, is initiated by absorption of a photon, promoting an electron from the highest occupied molecular orbital (HOMO) of the photocatalyst to another spin-allowed state. Thus, since the ground state of the photocatalyst is a singlet state (a state with no total electron spin), the photon absorption excites the catalyst to another singlet state of higher energy. This excitation is realized as a metal-to-ligand charge transfer, where the electron moves from an orbital centered on the metal (e.g. a d orbital) to an orbital localized on the ligands (e.g. the π* orbital of an aromatic ligand).

While absorption of a photon can occur for any energy corresponding to the difference between two singlet states, the excited electronic state relaxes to the lowest energy singlet excited state through internal conversion, a process whereby energy is dissipated as vibrational energy rather than as electromagnetic radiation. This singlet excited state can relax further by two distinct processes: the catalyst may fluoresce, radiating a photon and returning to the singlet ground state, or it can move to the lowest energy triplet excited state (a state where two unpaired electrons have the same spin) by a second non-radiative process termed intersystem crossing.

Direct relaxation of the excited triplet to the ground state, termed phosphorescence, requires both emission of a photon and inversion of the spin of the excited electron. The spin-forbidden nature of this pathway means that it is a slow process and therefore that the triplet excited state has a substantial average lifetime. For the common photocatalyst tris-(2,2’-bipyridyl)ruthenium chloride (also known as Ru(bipy)₃²⁺), the lifetime of the triplet excited state has been measured to be approximately 1100 ns. This span of time is long enough that other relaxation pathways, specifically electron-transfer pathways, can occur more rapidly than decay of the catalyst to its ground state.

The long-lived triplet excited state accessible by photoexcitation is both a more potent reducing agent and a more potent oxidizing agent than the ground state of the catalyst molecule. In other words, the catalyst can more readily give up one its electrons or accept an electron from an external source. The catalyst excited state is a stronger reductant because its highest energy electron has been excited to an even higher energy state through the photoexcitation process. Similarly, the excited catalyst is a stronger oxidant because one of the catalyst's lowest energy orbitals, which is fully occupied in the ground state, is only singly occupied after photoexcitation and is therefore available for an external electron to occupy. Since organometallic photocatalysts consist of a coordinatively saturated metal complex, i.e. a structure that cannot form any additional bonds, electron transfer cannot take place by an inner sphere mechanism through a direct bond of the metal complex to another reagent. Instead, electron transfer must take place via an outer sphere process, where the electron tunnels between the catalyst and another molecule.

Outer sphere electron transfer

Marcus' theory of outer sphere electron transfer predicts that such a tunneling process will occur most quickly in systems where the electron transfer is thermodynamically favorable (i.e. between strong reductants and oxidants) and where the electron transfer has a low intrinsic barrier.

The intrinsic barrier of electron transfer derives from the Franck–Condon principle, stating that electronic transition takes place more quickly given greater overlap between the initial and final electronic states. Interpreted loosely, this principle suggests that the barrier of an electronic transition is related to the degree to which the system seeks to reorganize. For an electronic transition with a system, the barrier is therefore related to the "overlap" between the initial and final wave functions of the excited electron–i.e. the degree to which the electron needs to "move" in the transition.

In an intermolecular electron transfer, a similar role is played by the degree to which the nuclei seek to move in response to the change in their new electronic environment. Immediately after electron transfer, the nuclear arrangement of the molecule, previously an equilibrium, now represents a vibrationally excited state and must relax to its new equilibrium geometry. Rigid systems, whose geometry is not greatly dependent on oxidation state, therefore experience less vibrational excitation during electron transfer, and have a lower intrinsic barrier. Photocatalysts such as Ru(bipy)3, are held in a rigid arrangement by flat, bidentate ligands arranged in an octahedral geometry around the metal center. Therefore, the complex does not undergo much reorganization during electron transfer. Since electron transfer of these complexes is fast, it is likely to take place within the duration of the catalyst's active state, i.e. during the lifetime of the triplet excited state.

Catalyst regeneration

The final step in the photocatalytic cycle is the regeneration of the photocatalyst in its ground state. At this stage, the catalyst exists as the ground state of either its oxidized or reduced forms, depending on whether it received or gave up an electron. These oxidation states have a strong driving force to return to their equilibrium oxidation state and act as a potent single-electron reductant or oxidant to satisfy that driving force.

To regenerate the original ground state, the catalyst must participate in a second outer-sphere electron transfer. In many cases, this electron transfer takes place with a stoichiometric two-electron reductant or oxidant, although in some cases this step involves a second reagent. The case where the excited state catalyst is first reduced and then oxidized to return to its resting state is known as a reductive quenching cycle. Conversely, the case where the excited state catalyst is first oxidized and then reduced to return to its resting state is known as an oxidative quenching cycle. These two possible cycles can be distinguished by a Stern–Volmer experiment.

Since the electron transfer step of the catalytic cycle takes place from the triplet excited state, it competes with phosphorescence as a relaxation pathway. The Stern–Volmer experiment measures the intensity of phosphorescence while varying the concentration of each possible quenching agent. When the concentration of the actual quenching agent is varied, the rate of electron transfer and the degree of phosphorescence is affected. This relationship is modeled by the equation:

( I 0 I ) = 1 + k q ∗ τ 0 × [ Q ]

Here, I₀ and I denote the emission intensity with and without quenching agent present, k_q the rate constant of the quenching process, τ₀ the excited-state lifetime in the absence of quenching agent and [Q] the concentration of quenching agent. Thus, if the excited-state lifetime of the photoredox catalyst is known from other experiments, the rate constant of quenching in the presence of a single reaction component can be determined by measuring the change in emission intensity as the concentration of quenching agent changes.

Redox potentials

The redox potentials of photoredox catalysts must be matched to the reaction's other components. While ground state redox potentials are easily measured by cyclic voltammetry or other electrochemical methods, measuring the redox potential of an electronically excited state cannot be accomplished directly by these methods. However, two methods exist that allow estimation of the excited-state redox potentials and one method exists for the direct measurement of these potentials. To estimate the excited-state redox potentials, one method is to compare the rates of electron transfer from the excited state to a series of ground-state reactants whose redox potentials are known. A more common method to estimate these potentials is to use an equation developed by Rehm and Weller that describes the excited-state potentials as a correction of the ground-state potentials:

E ∗ 1 / 2 r e d = E 1 / 2 r e d + E 0 , 0 + w r

E ∗ 1 / 2 o x = E 1 / 2 o x − E 0 , 0 + w r

In these formulas, E*_1/2 represents the reduction or oxidation potential of the excited state, E_1/2 represents the reduction or oxidation potential of the ground state, E_0,0 represents the difference in energy between the zeroth vibrational states of the ground and excited states and w_r represents the work function, an electrostatic interaction that arises due to the separation of charges that occurs during electron-transfer between two chemical species. The zero-zero excitation energy, E_0,0 is usually approximated by the corresponding transition in the fluorescence spectrum. This method allows calculation of approximate excited-state redox potentials from more easily measured ground-state redox potentials and spectroscopic data.

Direct measurement of the excited-state redox potentials is possible by applying a method known as phase-modulated voltammetry. This method works by shining light onto an electrochemical cell in order to generate the desired excited-state species, but to modulate the intensity of the light sinusoidally, so that the concentration of the excited-state species is not constant. In fact, the concentration of excited-state species in the cell should change exactly in phase with the intensity of light incident on the electrochemical cell. If the potential applied to the cell is strong enough for electron transfer to occur, the change in concentration of the redox-competent excited state can be measured as an alternating current (AC). Furthermore, the phase shift of the AC current relative to the intensity of the incident light corresponds to the average lifetime of an excited-state species before it engages in electron transfer.

In the following table, redox potentials are quoted as reduction potentials measured in comparison to the saturated calomel electrode (SCE). Thus, the oxidation potential of a chemical species is listed as the reduction potential of its oxidized form. More positive numbers indicate that the catalyst is reduced more readily, i.e. that it is a more strongly oxidizing species. Negative numbers indicate that the electrochemical reaction does not occur spontaneously in the direction written. In such a case, a voltage of the written magnitude must be applied to the catalyst in order to induce reduction. Thus, more negative reduction potentials indicate that the chemical species is more easily oxidized, and is a more strongly reducing species.

Ligand electronegativity

The relative reducing and oxidizing natures of these photocatalysts can be understood intuitively by considering the ligands' electronegativity and the catalyst complex's metal center. More electronegative metals and ligands tend to stabilize electrons better than their less electronegative counterparts. Therefore, complexes with more electronegative complexes are more easily reduced (i.e. be more powerfully oxidizing) than more electropositive complexes. For example, the ligands 2,2'-bipyridine and 2,2'-phenylpyridine are isoelectronic structures, containing the same number and arrangement of electrons.

However, phenylpyridine replaces one of the nitrogen atoms in bipyridine with a carbon atom. Carbon is less electronegative than nitrogen is, so it less tightly holds electrons. Since the remainder of the ligand molecule is identical, this effect is transferred to the structure as a whole: phenylpyridine holds electrons less tightly than bipyridine, i.e. it is more strongly electron-donating and less electronegative as a ligand. Complexes with phenylpyridine ligands are therefore more strongly reducing and less strongly oxidizing than complexes with bipyridine ligands because the phenylpyridine complexes are more electron-rich and hold their electrons with slightly less strength.

In the same way, a fluorinated derivative of phenylpyridine is more electronegative than the corresponding simple phenylpyridine and complexes with fluorine-containing ligands are more strongly oxidizing and less strongly reducing than the unsubstituted phenylpyridine complex. The metal center's electronic influence on the complex is somewhat more complex than the ligand effect. According to the Pauling scale of electronegativity, both ruthenium and iridium have an electronegativity of 2.2. If this was the sole factor relevant to redox potentials, then complexes of ruthenium and iridium with the same ligands should be equally powerful photoredox catalysts.

However, considering the Rehm-Weller equation, the spectroscopic properties of the metal also play a role in determining the redox character of the excited state. In particular, the parameter E_0,0 is related to the emission wavelength of the complex and therefore to the size of the Stokes shift - the difference in energy between the maximum absorption and emission of a molecule.

Ruthenium complexes typically have large Stokes shifts, and therefore have low energy emission wavelengths and small zero-zero excitation energies, when compared to iridium complexes. In effect, this means that although ground-state ruthenium complexes can be potent reductants, the excited-state complex is a much less potent reductant or oxidant than a comparable iridium complex. This makes iridium photocatalysts more favorable for the development of general organic transformations because the stronger redox potentials of the excited catalyst allows the use of weaker stoichiometric reductants and oxidants or the use of less reactive substrates.

Reductive dehalogenation

Reductive dehalogenation is a convenient method for removing halogen atoms that are introduced into a molecule by a method such as halolactonization. However, the standard method for removing halogen atoms requires the use of stoichiometric organotin reagents, such as tributyltin hydride. While this reaction is very powerful and orthogonal to other functional groups commonly present in organic molecules, organotin reagents are highly toxic. The cleavage of activated and reductively labile functional groups including sulfoniums and halogens is the earliest application of photoredox catalysis to organic synthesis, but early attempts were hampered by the need for specific or uncommon substrates or by the preferential formation of dimeric coupling products. More general methods are known. One method employs Ru(bipy)₃²⁺ as the photocatalyst and a stoichiometric amine reductant to reduce "activated" carbon-halogen bonds, such as those with an adjacent carbonyl group or arene. These bonds are considered to be activated, because the radical they produce upon fragmentation is stabilized by conjugation with the carbonyl group or arene, respectively. The stoichiometric reductant present in this reaction transfers an electron to reduce the excited-state catalyst to the Ru(I) oxidation state. The reduced catalyst can then shuttle the transferred electron to the halogenated substrate, reducing the weak C-X bond and inducing fragmentation.

Unactivated carbon-iodine bonds can be reduced using the strongly reducing photocatalyst tris-(2,2’-phenylpyridine)iridium (Ir(ppy)₃²⁺). This updated reaction is mechanistically distinct from the previous transformation of activated bromides and chlorides. The increased reduction potential of this catalyst compared to Ru(bipy)₃²⁺ allows direct reduction of the carbon-iodine bond without first interacting with a stoichiometric reductant. Thus, the iridium complex transfers an electron to the substrate, causing fragmentation of the substrate and oxidizing the catalyst to the Ir(IV) oxidation state. The oxidized photocatalyst is then easily returned to its original oxidation state through interaction with one of the reaction additives.

Just as tin-mediated radical dehalogenation reactions can be used to initiate cascade cyclizations to rapidly generate molecular complexity, the photocatlytic reductive dehalogenation allows access to the same types of complex products. In this work, a radical cascade cyclization that closed two five-membered rings and formed two new stereocenters, with good yield. This reductive dehalogenation protocol is a key step in a total synthesis of the natural product (+)-Gliocladin C.

Oxidative generation of iminium ions

Iminium ions are potent electrophiles useful for generating C-N bonds in complex molecules. However, the condensation of amines with carbonyl compounds to form iminium ions is often an unfavorable process, sometimes requiring harsh dehydration conditions. For this reason, alternative methods for iminium ion generation, particularly by oxidation from the corresponding amine, are a valuable synthesis tool. Iminium ions can be generated from activated amines by the use of Ir(dtbbpy)(ppy)₂PF₆ as a photoredox catalyst. This transformation is proposed to occur by oxidation of the amine to the aminium radical cation by the excited photocatalyst. The photocatalyst is strongly oxidizing due to the electrophilicity of iridium and the electron-poor nature of the fluorinated ligands. This is followed by H-atom transfer to a superstoichimetric oxidant, such as nitromethane (also functioning in the reaction as a nucleophile and as the solvent) or molecular oxygen. Finally, the reactive iminium ion formed by the H-atom transfer is quenched by reaction with nitromethane. Related transformations of amines with a wide variety of other nucleophiles have been investigated, such as cyanide (Strecker reaction), silyl enol ethers (Mannich reaction), dialkylphosphates, allyl silanes (aza-Sakurai reaction), indoles (Friedel-Crafts reaction), and copper acetylides.

Similar photoredox generation of iminium ions has furthermore been achieved using purely organic photoredox catalysts, such as Rose Bengal and Eosin Y.

An asymmetric variant of this reaction utilizes acyl nucleophile equivalents generated by N-heterocyclic carbene catalysis. This reaction method sidesteps the problem of poor enantioinduction from chiral photoredox catalysts by moving the source of enantioselectivity to the N-heterocyclic carbene.

Oxidative generation of oxocarbenium ions

The development of orthogonal protecting group chemistry is a crucial problem in organic synthesis because these protecting groups allow each instance of a common functional group, such as the hydroxyl group, to be distinguished during the synthesis of a complex molecule. One very common protecting group for the hydroxyl functional group is the para-methoxy benzyl (PMB) ether. This protecting group is chemically very similar to the less electron-rich benzyl ether. The usual method for selective cleavage of a PMB ether in the presence of a benzyl ether is through the use of strong stoichiometric oxidants such as 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) or ceric ammonium nitrate (CAN). PMB ethers are far more susceptible to oxidation than benzyl ethers because they are more electron-rich. The selective deprotection of PMB ethers can be achieved through the use of bis-(2-(2',4'-difluorophenyl)-5-trifluoromethylpyridine)-(4,4'-ditertbutylbipyridine)iridium(III) hexafluorophosphate (Ir[dF(CF₃)ppy]₂(dtbbpy)PF₆) and a mild stoichiometric oxidant such as bromotrichloromethane, CBrCl₃. The photoexcited iridium catalyst is reducing enough to fragment the polyhalomethane compound to form trichloromethyl radical, bromide anion and the Ir(IV) oxidation state of the catalyst. The electron-poor nature of the fluorinated ligands means that this iridium complex can be readily reduced: in particular, by an electron-rich arene such as a para-methoxy benzyl ether. After the arene is oxidized, it will readily participate in H-atom transfer with trichloromethyl radical to form chloroform and an oxocarbenium ion, which is readily hydrolyzed to reveal the free hydroxide. This reaction was demonstrated to be orthogonal to many common protecting groups, especially with the addition of a base to counteract the buildup of HBr during the reaction.

Cycloadditions

Cycloadditions and other pericyclic reactions are powerful transforms in organic synthesis because of their potential to rapidly generate complex molecular architectures and particularly because of their capacity to set multiple adjacent stereocenters in a highly controlled manner. However, only certain cycloadditions are allowed under thermal conditions according to the Woodward–Hoffmann rules of orbital symmetry, or other equivalent models such as frontier molecular orbital theory (FMO) or the Dewar-Zimmermann model. Cycloadditions that are not thermally allowed, such as the [2+2] cycloaddition, can be enabled by photochemical activation of the reaction. Under uncatalyzed conditions, this activation requires the use of high energy ultraviolet light capable of altering the orbital populations of the reactive compounds. Alternatively, metal catalysts such as cobalt and copper have been reported to catalyze thermally-forbidden [2+2] cycloadditions via single electron transfer.

The required change in orbital populations can be achieved by electron transfer with a photocatalyst sensitive to lower energy visible light. Yoon demonstrated the efficient intra- and intermolecular [2+2] cycloadditions of activated olefins: particularly enones and styrenes. Enones, or electron-poor olefins, were discovered to react via a radical-anion pathway, utilizing diisopropylethylamine as a transient source of electrons. For this electron-transfer, Ru(bipy)₃²⁺ was discovered to be an efficient photocatalyst. The anionic nature of the cyclization proved to be crucial: performing the reaction in acid rather than with a lithium counterion favored a non-cycloaddition pathway. Zhao et al. likewise discovered that a still different cyclization pathway is available to chalcones with a samarium counterion. Conversely, electron-rich styrenes were found to react via a radical-cation mechanism, utilizing methyl viologen or molecular oxygen as a transient electron sink. While Ru(bipy)₃²⁺ proved to be a competent catalyst for intramolecular cyclizations using methyl viologen, it could not be used with molecular oxygen as an electron sink or for intermolecular cyclizations. For intermolecular cyclizations, Yoon et al. discovered that the more strongly oxidizing photocatalyst Ru(bpm)₃²⁺ and molecular oxygen provided a catalytic system better suited to access the radical cation necessary for the cycloaddition to occur. Ru(bpz)₃²⁺, a still more strongly oxidizing photocatalyst, proved to be problematic because although it could catalyze the desired [2+2] cycloaddition, it was also strong enough to oxidize the cycloadduct and catalyze the retro-[2+2] reaction. This comparison of photocatalysts highlights the importance of tuning the redox properties of a photocatalyst to the reaction system as well as demonstrating the value of polypyridyl compounds as ligands, due to the ease with which they can be modified to adjust the redox properties of their complexes.

Photoredox-catalyzed [2+2] cycloadditions can also be effected with a triphenylpyrylium organic photoredox catalyst.

In addition to the thermally-forbidden [2+2] cycloaddition, photoredox catalysis can be applied to the [4+2] cyclization (Diels–Alder reaction). Bis-enones, similar to the substrates used for the photoredox [2+2] cyclization, but with a longer linker joining the two enone functional groups, undergo intramolecular radical-anion hetero-Diels–Alder reactions more rapidly than [2+2] cycloaddition.

Similarly, electron-rich styrenes participate in intra- or intermolecular Diels–Alder cyclizations via a radical cation mechanism. Ru(bipy)₃²⁺ was a competent catalyst for intermolecular, but not intramolecular, Diels–Alder cyclizations. This photoredox-catalyzed Diels–Alder reaction allows cycloaddition between two electronically mismatched substrates. The normal electronic demand for the Diels–Alder reaction calls for an electron-rich diene to react with an electron-poor olefin (or "dienophile"), while the inverse electron-demand Diels–Alder reaction takes place between the opposite case of an electron-poor diene and a very electron-rich dienophile. The photoredox case, since it takes place by a different mechanism than the thermal Diels–Alder reaction, allows cycloaddition between an electron-rich diene and an electron-rich dienophile, allowing access to new classes of Diels–Alder adducts.

The synthetic value of Yoon's photoredox-catalyzed styrene Diels–Alder reaction was demonstrated via the total synthesis of the natural product Heitziamide A. This synthesis demonstrates that the thermal Diels–Alder reaction favors the undesired regioisomer, but the photoredox-catalyzed reaction gives the desired regioisomer in improved yield.

Photoredox organocatalysis

Organocatalysis is a subfield of catalysis that explores the potential of organic small molecules as catalysts, particularly for the enantioselective creation of chiral molecules. One strategy in this subfield is the use of chiral secondary amines to activate carbonyl compounds. In this case, amine condensation with the carbonyl compound generates a nucleophilic enamine. The chiral amine is designed so that one face of the enamine is sterically shielded and so that only the unshielded face is free to react. Despite the power of this approach to catalyze the enantioselective functionalization of carbonyl compounds, certain valuable transformations, such as the catalytic enantioselective α-alkylation of aldehydes, remained elusive. The combination of organocatalysis and photoredox methods provides a catalytic solution to this problem. In this approach for the α-alkylation of aldehydes, Ru(bipy)₃²⁺ reductively fragments an activated alkyl halide, such as bromomalonate or phenacyl bromide, which can then add to catalytically-generated enamine in an enantioselective manner. The oxidized photocatalyst then oxidatively quenches the resulting α-amino radical to form an iminium ion, which hydrolyzes to give the functionalized carbonyl compound. This photoredox transformation was shown to be mechanistically distinct from another organocatalytic radical process termed singly-occupied molecular orbital (SOMO) catalysis. SOMO catalysis employs superstoichiometric ceric ammonium nitrate (CAN) to oxidize the catalytically-generated enamine to the corresponding radical cation, which can then add to a suitable coupling partner such as allyl silane. This type of mechanism is excluded for the photocatalytic alkylation reaction because whereas enamine radical cation was observed to cyclize onto pendant olefins and open cyclopropane radical clocks in SOMO catalysis, these structures were unreactive in the photoredox reaction.

This transformation include alkylations with other classes of activated alkyl halides of synthetic interest. In particular, the use of the photocatalyst Ir(dtbbpy)(ppy)₂⁺ allows the enantioselective α-trifluoromethylation of aldehydes while the use of Ir(ppy)₃ allowed the enantioselective coupling of aldehydes with electron-poor benzylic bromides. Zeitler et al. also investigated the productive merger of photoredox and organocatalytic methods to achieve enantioselective alkylation of aldehydes. The same chiral imidazolidinone organocatalyst was used to form enamine and introduce chirality. However, the organic photoredox catalyst Eosin Y was used rather than a ruthenium or iridium complex.

Direct β-arylation of saturated aldehydes and ketones can be effected through the combination of photoredox and organocatalytic methods. The previous method to accomplish direct β-functionalization of a saturated carbonyl consists of a one-pot consists of a two-step process, both catalyzed by a secondary amine organocatalyst: stoichiometric reduction of an aldehyde with IBX followed by addition of an activated alkyl nucleophile to the beta-position of the resulting enal. This transformation, which like other photoredox processes takes place by a radical mechanism, is limited to the addition of highly electrophilic arenes to the beta position. The severe limitations on the arene component scope in this reaction is due primarily to the need for an arene radical anion that is stable enough not to react directly with enamine or enamine radical cation. In the proposed mechanism, the activated photoredox catalyst is quenched oxidatively by an electron-deficient arene, such as 1,4-dicyanobenzene. The photocatalyst then oxidizes an enamine species, transiently generated by the condensation of an aldehyde with a secondary amine cocatalyst, such as the optimal isopropyl benzylamine. The resulting enamine radical cation usually reacts as a 3 π-electron system, but due to the stability of the radical coupling partners, deprotonation of the β-methylene position gives rise to a 5 π-electron system with strong radical character at the newly accessed β-carbon. Although this reaction relies on the use of a secondary amine organocatalyst to generate the enamine species which is oxidized in the proposed mechanism, no enantioselective variant of this reaction exists.

The development of this direct β-arylation of aldehydes led to related reactions for the β-functionalization of cyclic ketones. In particular, β-arylation of cyclic ketones has been achieved under similar reaction conditions, but using azepane as the secondary amine cocatalyst. A photocatalytic "homo-aldol" reaction works for cyclic ketones, allowing the coupling of the beta-position of the ketone to the ipso carbon of aryl ketones, such as benzophenone and acetophenone. In addition to the azepane cocatalyst, this reaction requires the use of the more strongly reducing photoredox catalyst Ir(ppy)₃ and the addition of lithium hexafluoroarsenide (LiAsF₆) to promote single-electron reduction of the aryl ketone.

Additions to olefins

The use of photoredox catalysis to generate reactive heteroatom-centered radicals was first explored in the 1990s. Ru(bipy)₃²⁺ was found to catalyze the fragmentation of tosylphenylselenide to phenylselenolate anion and tosyl radical and that a radical chain propagation mechanism allowed the addition of tosyl radical and phenylseleno- radical across the double bond of electron rich alkyl vinyl ethers. Since phenylselenolate anion is readily oxidized to diphenyldiselenide, the low quantities of diphenyldiselenide observed was taken as an indication that photoredox-catalyzed fragmentation of tosylphenylselenide was only important as an initiation step, and that most of the reactivity was due to a radical chain process.

Heteroaromatic additions to olefins include multicomponent oxy- and aminotrifluoromethylation reactions. These reactions use Umemoto's reagent, a sulfonium salt that serves as an electrophilic source of the trifluoromethyl group and that is precedented to react via a single-electron transfer pathway. Thus, single-electron reduction of Umemoto's reagent releases trifluoromethyl radical, which adds to the reactive olefin. Subsequently, single-electron oxidation of the alkyl radical generated by this addition produces a cation which can be trapped by water, an alcohol, or a nitrile. In order to achieve high levels of regioselectivity, this reactivity has been explored mainly for styrenes, which are biased towards formation of the benzylic radical intermediate.

Hydrotrifluoromethylation of styrenes and aliphatic alkenes can be effected with a mesityl acridinium organic photoredox catalyst and Langlois' reagent as the source of CF₃ radical. In this reaction, it was found that trifluoroethanol and substoichiometric amounts of an aromatic thiol, such as methyl thiosalicylate, employed in tandem served as the best source of hydrogen radical to complete the catalytic cycle.

Intramolecular hydroetherifications and hydroaminations proceed with anti-Markovnikov selectivity. One mechanism invokes the single-electron oxidation of the olefin, trapping the radical cation by a pendant hydroxyl or amine functional group, and quenching the resulting alkyl radical by H-atom transfer from a highly labile donor species. Extensions of this reactivity to intermolecular systems have resulted in i) a new synthetic route to complex tetrahydrofurans by a "polar-radical-crossover cycloaddition" (PRCC reaction) of an allylic alcohol with an olefin, and ii) the anti-Markovnikov addition of carboxylic acids to olefins.