Redox-innocent scandium(III) as the sole catalyst in visible light photooxidations

Abstract

In recent years, the catalytic activity of scandium triflate Sc(OTf)₃ has attracted significant attention due to its robust Lewis acidity and the oxophilicity of Sc³⁺. These features have led to impressive progress in developing diverse organic reactions, including C-C bond formation. The Sc³⁺ also facilitates single electron transfer in photoinduced reactions either by coordination to an organophotoredox catalyst, which modifies its redox reactivity, or by the formation of a scandium–superoxide anion complex after electron transfer from a light-absorbing redox-active compound. The prior consideration of Sc³⁺ as a redox-inactive/innocent metal ion initially hampered the investigation of the possibility of using Sc(OTf)₃ as a sole visible light photoredox catalyst. This work demonstrates the use of Sc(OTf)₃ as a visible light photocatalyst capable of direct and mild aerobic oxidative C-H functionalisation of aromatic substrates by oxidation of the benzylic position and direct cyanation of the aromatic ring.

In recent years, the catalytic activity of scandium triflate Sc(OTf)3 has attracted significant attention due to its robust Lewis acidity and the oxophilicity of Sc3 + . Here, the authors demonstrate different reactivity of the same compound, that Sc(OTf)3 is a visible light photocatalyst capable of direct and mild aerobic oxidative C–H functionalization.

Introduction

Rare-earth metal ions (Sc, Y, lanthanides) have attracted significant attention, mainly due to their Lewis acidity and oxophilicity. In particular, Sc(OTf)₃ has been introduced in recent decades as a promising Lewis acid for C-H functionalisation and C-C bond formation reactions^1,2. In the past 20 years, scandium salts have also effectively assumed a role in photoredox catalysis—a process that has emerged as a full-fledged alternative in organic synthesis, often unlocking transformations that are not attainable in the dark^3–6. Photoredox catalysis relies on an electron transfer process between a highly reactive, light-excited catalyst and a redox-labile substrate (S). The result is a reactive radical intermediate that undergoes further transformations, ultimately resulting in the formation of new bonds under mild conditions.

Photoredox catalysis requires a catalyst that is both photochemically and redox active—typically a compound derived from an organic dye or a complex of a redox-active metal, such as ruthenium or iridium (see Fig. 1a)^7–10. Conversely, redox-inactive/innocent metals, such as scandium, which have a predominantly trivalent redox-stable state, are not directly involved in photoredox reactions^11–13. Nevertheless, these metals, and particularly Sc(OTf)₃, can serve as Lewis acids (LA) that activate substrates/organophotocatalysts by altering their redox characteristics, thereby facilitating their participation in photoredox and photosensitisation reactions (Figs. 1b and 1c)^12–18. For instance, the redox properties of a flavin derivative (Fl) were significantly enhanced via their coordination to Sc(OTf)₃. This alteration enabled the oxidation of some electron-deficient substrates via a SET process between the substrate and flavin in excited Fl-2Sc³⁺ complex (Fig. 1c, right); such reaction cannot be achieved without Sc(OTf)₃^19,20.

Fig. 1. Role of redox-active and redox-inactive metal ions in photochemical transformations.

a Use of metal ions in photoredox catalysis^3,8–10. b Use of Sc³⁺ as a Lewis acid to activate a substrate (S) for reaction with excited photocatalyst (PC*); example shown on right. c Activation of organic photocatalyst (PC) with Sc³⁺ as Lewis acid (LA); example shown on right. d Metal ion-coupled electron transfer (MCET) with oxygen as an acceptor; an example involving Sc³⁺ in a procedure with light absorbing substrate is shown. e Use of Sc³⁺ salt as the sole catalyst in photoredox catalysis for benzylic oxidations and cyanations (this work).

Notably, the oxophilicity of some Mⁿ⁺ cations facilitates their binding to superoxide radical anion (O₂^•−) to form superoxide–metal complexes (Mⁿ⁺–O₂^•−) that have been recognised within the reductive activation of dioxygen by metalloenzymes in numerous biological redox reactions that employ O₂ as an oxidant^21–24. This process has been mimicked in the dioxygen activation cycle during photoredox transformations, commencing with the photoinduced single electron transfer (PET) from a light-absorbing, redox-active compound to O₂ and followed by the generation of Mⁿ⁺-O₂^•− complexes (Fig. 1d, upper part). Binding of Mⁿ⁺ to oxygen enhances its ability to accept an electron. Thus, an electron transfer from donor (D) to O₂, which is thermodynamically infeasible in the absence of metal ions, becomes possible in the presence of Mⁿ⁺. This process is generally called metal ion–coupled electron transfer (MCET)^12,25–29. Scandium³⁺ ion, with its smallest radius among M³⁺, can bind strongly with O₂^•−, making it far more effective than other Mⁿ⁺ ions for providing SET reactions driven by Mⁿ⁺-O₂^•− complex formation^12,30–35.

A review of previous reports on the role of Sc(OTf)₃ in photoinduced electron transfer (PET) transformations involving the reductive activation of dioxygen confirms that a light-absorbing electron donor and additional stabilisation are necessary. For example, Fukuzumi³⁰ recorded the EPR spectrum of the (HMPA)₃Sc³⁺-O₂^•− complex stabilised by three equivalents of hexamethylphosphoramide (HMPA) ligand, formed via PET from the singlet excited state of 1-benzyl-1,4-dihydronicotinamide dimer¹[(BNA)₂]*, to O₂. This procedure was accompanied by fast cleavage of the C-C bond in the formed (BNA)₂^•+ (Fig. 1d, bottom part).

In this work we demonstrate the activity of Sc(OTf)₃ functioning as a sole photoredox catalyst. Sc(OTf)₃ is shown to mediate aerobic oxidation of a benzylic C-H bond and direct oxidative cyanation of arenes as examples of C-H activation and C-C coupling reactions (Fig. 1e). These procedures dispel the usual notion of scandium salts as redox-inactive species useful mostly as Lewis acids and expand their possible application as simple, readily available photocatalysts. In general, photoredox processes might be a future opportunity for scandium, which is abundantly found on earth but remains an underutilised metal³⁶.

Results

Development of Sc(OTf)₃-based aerobic photooxidations

Upon the surprising observation of the photocatalytic activity of Sc(OTf)₃, we directed our attention towards the aerobic oxidation of toluene (1a; E_ox = +2.36 V vs SCE³⁷) as a benchmark. The oxidation was performed with 5 mol% of Sc(OTf)₃ (E_red < –1.0 V vs SCE, see ref. ³⁸ and our experiments below; i.e. not strong enough in ground state to oxidise 1a) in O₂–saturated acetonitrile (MeCN) under blue light irradiation (400 nm LED) for 24 h. Encouragingly, this experiment afforded benzoic acid (2a) as the sole product at an 80% yield (Fig. 2a, entry 1). Under anaerobic conditions, or in the absence of light or Sc(OTf)₃, the reaction did not proceed (entries 2–5). This system demonstrated significantly decreased effectiveness with 450 nm LED illumination (Entry 6). An assessment of Lewis acids revealed that Sc(OTf)₃ was the optimal selection, probably due to its superior ability to bind strongly with O₂^{•− 35}. By contrast, employing Mg(OTf)₂, Zn(OTf)₂, La(OTf)₃, and Ba(OTf)₂ under identical experimental conditions resulted in the complete cessation of the reaction (see Supplementary Section S2).

Fig. 2. Sc(OTf)₃ can act as a sole photocatalyst.

a Control experiments showing Sc(OTf)₃-based aerobic photocatalytic oxidation of toluene. b Screening of solvents. c Time-dependent NMR yield profile d UV-Vis absorption and fluorescence spectra (λ_ex = 400 nm) of Sc(OTf)₃ in MeCN (c = 0.05 M) and a range of LED emission; an image of the solution in a 1 cm cuvette (inset). ^a Standard reaction conditions: toluene (0.140 mmol), Sc(OTf)₃ (5 mol%), MeCN (0.25 mL), 45 °C, O₂ (balloon), blue LED (400 nm), and 24 h. ^b Yields were determined from the ¹H NMR spectra. ^c Product was not detected.

As depicted in Fig. 2b, the choice of solvent was pivotal in determining the reaction progression. A switch from nitrile solvents, such as MeCN or EtCN, to MeNO₂ resulted in some product, but at a decreased yield compared to nitrile solvents. A mixture of MeCN and MeOH (9:1) resulted in a significant decrease in the reaction yield, with complete cessation in MeOH or water, as examples of protic solvents. The use of polar aprotic solvent (DMSO or DMF) or nonpolar solvents also resulted in complete inhibition of the reaction.

An investigation of catalyst loading (Fig. 2a, entries 7 and 8 and Supplementary Section S2) also revealed that using a smaller amount of Sc(OTf)₃, at 2.5 mol%, slightly reduced the yield of benzoic acid (2a), whereas employing 10 mol% of the catalyst notably achieved a full conversion to 2a. Moreover, the time-dependent yield profile showed an almost constant rate of toluene (1a) oxidation with increasing reaction time (Fig. 2c). Consequently, the plot of turnover frequency (TOF) values over the reaction time (4–20 h) remained stable, signifying maintenance of the photocatalytic activity of Sc(OTf)₃ (see Supplementary Section S2).

The unexpected finding of the photocatalytic activity of Sc(OTf)₃ prompted us to check our experiments thoroughly multiple times. We ensured that no organic impurities were driving the reaction. We also tested Sc(OTf)₃ from different suppliers, but we found no significant variations except for reduced activity caused by traces of oxygen atom–containing solvents like MeOH, attributable to synthesis residues. This reduced activity could be improved by recrystallising Sc(OTf)₃ from its solution in MeCN and using chloroform for precipitation.

The photocatalytic activity could be ascribed to the weak absorption of Sc(OTf)₃ in MeCN (this extended to the visible region around 400 nm), and it was also indicated by blue fluorescence with a distinct band at 505 nm observed upon excitation with 400 nm light (Fig. 2d).

Substrate scope of Sc(OTf)₃-based aerobic photooxidation

Using our optimised conditions, we evaluated various substrates. Both electron-donating and electron-deficient aromatic compounds (Fig. 3a) revealed the promising photocatalytic activity of Sc(OTf)₃ in most cases. Toluene (1a) yielded 80% benzoic acid (2a) and p-chlorotoluene (1b) (E_ox = +2.21 V vs SCE³⁷) underwent even complete conversion to its respective carboxylic acid 2b. As anticipated, the electron-deficient 4-(trifluoromethyl)toluene (1c), with the highest oxidation potential (E_ox = +2.61 V vs SCE³⁷), displayed considerable resistance to oxidation, resulting in trace product formation. Conversely, the less electron-deficient ester group-containing derivative 1 d (E_ox = +2.45 V vs SCE³⁹) was effectively oxidised. Substrates with higher oxidation potentials than 2.5 V vs SCE seem beyond the capabilities of Sc(OTf)₃–catalysed oxidation. However, we observed that combination of high temperature (65 °C) and the addition of 0.2 equivalent of trifluoroacetic acid allows the oxidation of even such a demanding substrate as 4-(trifluoromethyl)toluene (1c) (see Fig. 3). It should be noted that stronger trifluoromethanesulfonic acid or the weaker acetic acid did not cause such a significant increase in oxidation conversion (see Supplementary Section S2 for details).

Fig. 3. Scope of Sc(OTf)₃-based aerobic photocatalytic oxidations.

a Benzylic oxygenations/oxidations of alkyl/alkenyl/aryl arenes. b Oxidations of primary and secondary benzyl alcohols and related substrates. ^a Standard reaction conditions: substrate (0.140 mmol), Sc(OTf)₃ (5 mol%), MeCN (0.25 mL), 45 °C, O₂ (balloon), blue LED (400 nm), and 24 h. Yields were determined by ¹H NMR spectra unless otherwise stated. ^b Isolated yield (in magenta); experiments were performed at a 1 mmol scale (see Supplementary Section S3 for details). ^c in propionitrile at 65 °C with CF₃COOH (0.2 equiv.), 48 h. ^d Only 2 f as the sole product. ^e 48 h. ^f 59% of 13b and 16% of 2b, 24 h.

Unexpectedly, 4-methylanisole 1e (E_ox = +1.51 V vs SCE³⁷), despite being the most electron-rich substrate, gave only a 4% yield. This could be attributed to the back-electron transfer (BET) process during the oxidation of 1e. Biphenyl 3 (E_ox = +1.95 V vs SCE⁴⁰) underwent efficient oxidation, providing a complete yield of benzoic acid (2a). Oxidation of p-xylene (1 f) afforded a mixture of 4-methylbenzoic acid (2 f) (67%) and terephthalic acid (2 g) (33%) after 24 h of irradiation. Ethylbenzene derivatives 4a-c provided reasonable yields of benzoic acid derivatives according to the electronic effect, although 4-ethyltoluene (4 d) was oxidised to a mixture of 4-acetylbenzoic acid (2 h) and 2 f. By contrast, cumene (5) gave a poor yield of benzoic acid upon oxidation. Diphenylmethane (6) was transformed into a mixture of the corresponding ketone 7 and benzoic acid (2a). 9H-fluorene (8) was converted into fluorenone (9) as a sole product with excellent yield, while stilbene (10) yielded benzoic acid via oxidative C = C cleavage.

Further investigation of the oxidation of various primary and secondary aryl alcohols, as depicted in Fig. 3b, revealed that all the tested primary benzyl alcohols, including electron-rich and highly electron-deficient ones 11a-d and o-chlorobenzyl alcohol (11e), were oxidised to the corresponding benzoic acids 2 in quantitative yields. The secondary alcohol 1-(4-chlorophenyl)ethan-1-ol (12b) provided 4-chloroacetophenone (13b) in a quantitative yield, while the electron-poor 1-[4-(trifluoromethyl)phenyl]ethanol (12c) produced a low yield of 4-(trifluoromethyl)acetophenone (13c), as anticipated. Surprisingly, the oxidation of 1-phenylethanol (12a) resulted in a low conversion to the corresponding ketone 13a, for a yield of only 23%. A mixture of benzophenone (15) and benzoic acid was obtained following the oxidation of diphenylmethanol (14). By contrast, cinnamyl alcohol (16) underwent a C = C bond cleavage during oxidation to generate a quantitative yield of benzoic acid. Notably, 4-methoxybenzyl methyl ether (17) yielded benzoic acid and not the corresponding ester.

The optimised conditions were subsequently used at a preparative scale, with 1 mmol of the selected substrates (see Fig. 3) to confirm the practicality and effectiveness of our aerobic photocatalytic oxidation approach employing Sc(OTf)₃ (see Supplementary Section S3 for details). The reaction progression was monitored using ¹H NMR analysis and the time was optimised to achieve the maximal conversion of reactants to products. Interestingly, p-xylene (1 f), a substrate with two potential oxidation sites, was oxidised chemoselectively to 2 f at the preparative scale after appropriately adjusting the reaction time.

Sc(OTf)₃-based photocatalytic oxidative cyanation of arenes

A further aim of this study was to demonstrate the utility of Sc-based photoredox catalysis. We chose arenecarbonitriles, which serve as crucial structural scaffolds in synthesising bioactive compounds, alongside their substantial role as intermediates in organic synthesis⁴¹. We particularly focused on the direct photooxidative cyanation of the aromatic C-H bond, as this reaction remains a formidable challenge for the chemical community because of the need for prefunctionalised starting material and the lack of regioselectivity. Moreover, photocatalytic methodologies for direct C-H cyanation are limited and still require further exploration^42–48.

We first used diphenyl ether (18a) (E_ox = 1.88 V, vs. SCE)⁴⁹ as a benchmark substrate and performed the cyanation using TMSCN as a nucleophilic cyanating agent (12 equiv.) in oxygen–saturated MeCN under blue light irradiation (400 nm) for 24 h. Our preliminary investigations (Fig. 4a, entry 1) showed that a mixture of 4-phenoxybenzonitrile (19a) and 2-phenoxybenzonitrile (19b) could be obtained in a good yield of 74%, with notable regioselectivity towards the para-substituted isomer. Control experiments demonstrated that the cyanation of 18a ceased in the absence of light or Sc(OTf)₃, or under anaerobic conditions (entries 2-5). Raising the wavelength of the irradiation source from 400 nm to 450 nm resulted in only trace yields of the desired product (entry 6). For explanation, the cyanating agent TMSCN was used in a significant excess (12 equiv.) to avoid competitive photooxidation reactions (see Supplementary Section S4).

Fig. 4. Experiments investigating Sc(OTf)₃-based photocatalytic oxidative cyanation of arenes under aerobic conditions.

a Condition optimisation and control experiments. b Reaction scope for arenes. ^a Standard reaction conditions: Substrate (0.140 mmol), Sc(OTf)₃ (5 mol%), Me₃SiCN (12 equiv., 1.68 mmol), MeCN (0.25 mL), 45 °C, O₂ (balloon), blue LED (400 nm), and 24 h (unless otherwise specified); Yields were determined by ¹H NMR unless otherwise stated. ^b Isolated yield (in magenta); experiments were performed at a 1 mmol scale (see Supplementary Section S6 for details). ^c Product was not detected. ^d 48% of 19a, and 8% of 19b. ^e total yield of a mixture of 19e and 19 f (4:1).

Similarly, as was observed for the benzylic oxidations, switching Sc(OTf)₃ to Mg(OTf)₂, Zn(OTf)₂, La(OTf)₃, or Ba(OTf)₂ resulted in the cessation of cyanation under otherwise identical experimental conditions (see Supplementary Section S4). Switching from MeCN to MeOH or DMSO also completely inhibited the reaction, whereas the reaction in MeNO₂ occurred with remarkable conversion. Reaction in dioxane gave small amount of product (Fig. 4a, entries 7-10). Reducing the amount of the Sc(OTf)₃ catalyst (2.5 mol%) lowered the yield, whereas doubling the amount (10 mol%) gave a slightly higher yield (entries 11 and 12).

Using the optimised conditions, we next investigated the direct cyanation of various arenes, including both electron-rich and electron-poor substrates, as depicted in Fig. 4b (see Supplementary Section S6 for details). Cyanation of 1,3,5-trimethoxybenzene (18b) and 1,3-dimethoxybenzene (18c) furnished the corresponding carbonitriles 19c and 19 d, respectively, as sole products in quantitative yields. Cyanation of 3-chloroanisole (18 d) resulted in a mixture of 4-chloro-2-methoxybenzonitrile (19e) and 2-chloro-4-methoxybenzonitrile (19 f) in excellent yields, with 3.9:1 regioselectivity. Efficient cyanation of biphenyl (18e) occurred, yielding 4-phenylbenzonitrile (19 g). Cyanation of butoxybenzene (18 f) led to the formation of 4-butoxybenzonitrile (19 h) in a quantitative yield; however, the efficiency of this system in the cyanation of electron-poor substrates was only modest. Cyanation of methyl 2-methoxybenzoate (18 g) and 1-methoxy-2-(trifluoromethyl)benzene (18 h) gave poor yields of the desired products 19i and 19j, respectively. The highly electron-poor substrates, such as chlorobenzene (18i), methyl benzoate (18j), and trifluorotoluene (18k), did not yield any cyanation products. Subsequent application of the optimised conditions using 1 mmol of the selected substrates 18a and 18 d led to cyanation products in good yields, thereby confirming the practicality of our cyanation approach on a preparative scale.

Mechanistic investigations

Additional control experiments (see Figs. 2a and 4a for the first set of control experiments) were first conducted to unravel the mechanisms underlying Sc(OTf)₃-based photooxidative transformations. Toluene oxidation was clearly impeded in the presence of TEMPO (2,2,6,6-tetramethylpiperidine-1-oxyl) as a radical scavenger. Atmospheric pressure chemical ionization mass spectrometry (APCI-MS) data confirmed the formation of a TEMPO-benzyl adduct (Fig. 5a), indicating the involvement of a radical pathway in this oxidation process. Similarly, the cyanation of diphenyl ether was almost completely halted in the presence of TEMPO (see Supplementary Section S7 for details).

Fig. 5. Mechanistic studies.

a Formation of a TEMPO-benzyl adduct during photooxidation of toluene using Sc(OTf)₃. b Oxidation with isotopically labelled oxygen (see Supplementary Section S8 for details). c Quenching of fluorescence of Sc(OTf)₃ solution in MeCN (c = 0.05 M, λ_ex = 400 nm) with toluene and the corresponding Stern-Volmer plot (inset). d Comparison of the theoretical absorption spectra of [Sc(CH₃CN)₃(O₂)]³⁺ obtained by quantum chemical calculations at PBE0/ Def2-TZVPD level of theory (DFT calculations were performed in Orca software package, release 6.0.1., ref. ⁵⁶) with experimental spectra of Sc(OTf)₃ solution in MeCN (inset). e Cyclic voltammogram at Glassy Carbon microdisc working electrode of oxygen-saturated acetonitrile before and after addition of Sc(OTf)₃. f The EPR spectrum of an oxygen-saturated MeCN solution of toluene (c = 2 M) and Sc(OTf)₃ (c = 0.1 M) upon irradiation with a 400 nm LED at room temperature; kinetic trace after stopping the irradiation (inset).

Oxidation of toluene using isotopically labelled oxygen ¹⁸O₂ gave mainly benzoic acid (2a) with two incorporated ¹⁸O atoms and almost no product containing two ¹⁶O atoms (Fig. 5b). This result supports the hypothesis that molecular oxygen is the source of the oxygen atom in the product. Some amount (30 %) of product containing one unlabelled ¹⁶O atom and one ¹⁸O was also observed which is explained by enhanced residual moisture causing oxygen atom exchange in benzaldehyde intermediate by oxo –hydrate form equilibria which is supported by Sc³⁺ (Lewis acid) catalysis (see Supplementary Section S8 for details).

Stern-Volmer quenching experiments revealed that toluene, diphenyl ether and 1,3,5-trimethoxybenzene quenched the fluorescence of a Sc(OTf)₃ solution in acetonitrile with Stern-Volmer constants K_s of 3, 22 and 42 L mol^-1, respectively (see example in Fig. 5c). This signified SET from the tested substrates to an excited scandium complex (see Supplementary Section S7 for details).

As mentioned earlier, the UV-Vis absorption spectrum of Sc(OTf)₃ in acetonitrile showed an elongated tail in the visible region around 400 nm and a fluorescence emission spectrum with a sharp band at 505 nm (Fig. 2d). ESI-MS measurements aimed at elucidating the structure of the scandium complex responsible for this absorption in a solution of Sc(OTf)₃ in MeCN. It suggested the formation of a cluster of complexes, including [Sc(OTf)₂(MeCN)₂]⁺, as the primary species. However, these particles do not absorb in the visible region but below 260 nm, as indicated by the theoretical spectra (see Supplementary Section S12 and Supplementary Data 1 for Cartesian coordinates). By contrast, particles containing coordinated molecular oxygen seems to absorb visible light. For example the [Sc(MeCN)₃((η²-O₂)]³⁺ particle in S₀ state, a complex with three MeCN molecules and coordinated oxygen molecule could absorb in the visible region (400 – 442 nm) as documented by the theoretical spectrum, consistent with our experimental results (Fig. 5d). Moreover, calculated emission of this species (522 nm) corresponds to experimental fluorescence (around 506 nm; see Supplementary Section S12).

Indeed, the coordination of oxygen molecule is ascribed to the oxophilicity of Sc³⁺ ions and is further supported by our EPR and cyclic voltammetry measurements (see below). It should be noted that the isomeric [Sc(MeCN)₃(η¹-O₂)]³⁺ complex (in T₁ state) is more stable by 0.950 eV (91.67 kJ mol⁻¹) at DLNPO-CCSD(T)/aug-cc-pVTZ level of theory (see Supplementary Section S12 for further information on excited state energies) compared to the S₀ state [Sc(MeCN)₃((η²-O₂)]³⁺, nevertheless, [Sc(MeCN)₃(η¹-O₂)]³⁺ is proposed to not absorb in the visible region (absorption below 300 nm, see Supplementary Section S12). We believe the small barrier can be overcome thermally.

Interaction of Sc³⁺ ions with molecular O₂ was also supported by cyclic voltammetry experiments (Fig. 5e). Addition of Sc(OTf)₃ into oxygen-saturated acetonitrile was characterised by formation of a new peak (E_cp = –0.95 V vs SCE) with less negative potential compared to the peak of oxygen in acetonitrile (E_pc = –1.16 V vs SCE). The new signal is attributed to oxygen coordinated to Sc³⁺ ions. The same signal can be achieved by an inverse experiment. Cyclic voltammogram of Sc(OTf)₃ under argon contains peak at –1.83 V vs SCE corresponding to Sc³⁺ reduction. After bubbling the solution with oxygen gas, two peaks attributed to coordinated and “free” oxygen appeared (see Supplementary Section S9).

In subsequent experiments, we investigated the mechanism of Sc-based photooxidative catalysis using toluene oxidation as the representative example. Irradiation of an oxygen-saturated acetonitrile solution containing Sc(OTf)₃ (0.1 M) and toluene (2 M) with a 400 nm LED at room temperature resulted in a distinct isotropic EPR signal (Fig. 5f). The pronounced eight-line isotropic spectrum was attributed to the formation of a complex of superoxide radical anion with scandium ion, Sc³⁺-O₂^•−. This resulted in superhyperfine splitting (a_Sc = 4.28 G) due to the 7/2 scandium nuclear spin and featured an isotropic g value, g_iso = 2.0160. This observation is consistent with reported EPR spectra in the literature³⁰.

Interestingly, while previous studies required a low temperature to observe Sc³⁺-O₂^•− by EPR, we were able to obtain well-defined spectra at room temperature. The kinetic recordings showed stability of the Sc³⁺-O₂^•− complex for about 2 min after the cessation of irradiation (Fig. 5f, insight). Importantly, no significant EPR signal appeared upon irradiation of a blank solution containing no toluene. Irradiation of the acetonitrile solution of Sc(OTf)₃ and toluene under anaerobic conditions also did not reveal any EPR signals. Replacing Sc(OTf)₃ with La(OTf)₃ (which showed no catalytic activity in photooxidative procedures) led to a very faint EPR signal, even after an extended period of irradiation (see Supplementary Section S10). These observations provide evidence that all species — toluene as a substrate and electron donor, oxygen as a terminal electron acceptor and scandium as mediator — are necessary for efficient photoinduced electron transfer and Sc³⁺-O₂^•− formation.

Transient absorption spectroscopy

We utilised ultrafast transient absorption (TA) spectroscopy to unravel the nature of the excited state species involved. We performed measurements of Sc(OTf)₃ in MeCN under both O₂ and N₂ atmospheres (purging for 30 minutes, followed by gassing with N₂ in a purity of 99.99%). Initially, measuring a solution of Sc(OTf)₃ in MeCN under an inert N₂ atmosphere resulted in a strong bleach signal at approximately 500 nm, accompanied by weak excited-state absorption (ESA) features (Fig. 6a). This bleach rapidly converted into a positive signal over a slower time scale of a few nanoseconds. The positive signal then continuously grew over the microsecond time scale until reaching a maximum at around 1 microsecond, after which it faded slowly, as depicted in Fig. 6b-c. In the presence of O₂, similar kinetic behaviour was found at the early stage, but with a feeble TA signal, and only during the long timescale (microseconds, the decay was somewhat faster than this in the N₂ atmosphere) (see Fig. 6d). The initial bleach signal at 500 nm was assigned to the initial fluorescence of the Franck-Condon (FC) state, which quickly relaxed to some non-emissive state (S₁ relaxed); this took 300 fs. This relaxed state populated another state, the triplet state, at a slow rate (10 ns). The triplet state was then slowly quenched over a microsecond lifetime.

Fig. 6. Mechanistic studies by transient absorption spectroscopy.

a Extracted kinetic trace @ 500 nm for the excited Sc(OTf)₃ in MeCN under N₂ (from fs to ns). b Extracted spectra for the excited Sc(OTf)₃ in MeCN under N₂ at longer time scale (µs), each spectrum is assigned to the population of an excited state. c Extracted kinetic trace from Sc(OTf)₃ in MeCN under N₂ at longer time scale (µs). d Comparison between extracted kinetic traces under N₂ and O₂ for Sc(OTf)₃ in MeCN at longer time scale. e Extracted kinetic trace for Sc(OTf)₃ in toluene under O₂ at short time scale (fs-ns). f Evolving spectra for Sc(OTf)₃ in toluene under O₂ at longer time scale (µs). g Extracted kinetic trace for Sc(OTf)₃ in toluene under O₂ at longer time scale (µs). h Comparison between extracted kinetic traces under N₂ and O₂ for Sc(OTf)₃ in toluene.

The weak invariance between the N₂ and O₂ atmosphere was attributed to the O₂ that was strongly bound to the Sc³⁺ complex and present even under the N₂ atmosphere because of strong Sc oxophilicity³⁵. A role for the triplet state of O₂ in the population of the Sc³⁺ complex triplet state was highly expected.

When measured in toluene/MeCN (Fig. 6e-h), the transient absorption data were changed to a greater extent than those in pure MeCN, but remained similar under O₂ and N₂ atmospheres. For instance, the initial fluorescence from FC was missing, but a fast lifetime decay of 500 fs (due to the relaxation state) was present, followed by a lifetime rise of 20 ps that was assigned to the early formation of the triplet state (see Fig. 6e). Within 20 ns, the bleach signal around ca. 500 nm started to evolve again, followed by an increase in the overall ESA features across the visible region over 90 ns. The appearance of the bleach signal was assigned to the delayed fluorescence of the Sc complex, which can occur due to the reversible ISC (r-ISC) process from the triplet to the singlet state⁵⁰, as depicted in Fig. 6f. Due to the presence of toluene molecules, an ET was expected to occur from toluene to the Sc complex on a time scale of 90 ns, leading to the disappearance of the fluorescence peak and formation of S₁^– state, as evident from Fig. 6f-g. The recombination of this reduced state to the S₀ state occurred at a slow rate > 1.2 microseconds.

The absence of toluene remarkably affected the Sc complex behaviour by delaying the triplet state formation in MeCN vs in the MeCN/toluene mixture. The insensitivity of the Sc complex towards the O₂ atmosphere also highlighted the role of strongly bound O₂ molecules in extending the lifetime of the observed excited states. For instance, the slow decay of the S₁^– state was expected to be due to structural rearrangements within the Sc complex, leading to the desorption of the reduced O₂ into the reaction mixture for further use in product formation pathways.

Mechanism of Sc(OTf)₃-based photocatalysis

Drawing from the outcomes of control experiments, mechanistic studies using EPR and transient absorption measurements (TA), and considering previous reports, a plausible mechanism for the Sc(OTf)₃-catalysed photooxidation reactions is postulated (Fig. 7). Initially, the Sc³⁺ complex can be excited upon coordination with molecular oxygen, as demonstrated by the quantum chemical calculations (see Fig. 5d and Supplementary Section S12). This type of complex is proposed to involve acetonitrile ligands and the participation of triflates either as counterions or ligands. Upon excitation, the scandium complex forms a singlet excited state, which can subsequently undergo an intersystem crossing (ISC) process, as elucidated by TA, to generate a triplet excited state of the scandium complex. Nevertheless, according to TA measurements, in the presence of toluene, a delayed-singlet excited state of the Sc³⁺ complex (with oxygen involved) seems to be responsible for the productive reaction pathway that undergoes SET from toluene during the production of a toluene radical cation and the Sc³⁺-O₂^•− complex, as verified by the eight-line EPR spectrum arising from the I = 7/2 ⁴⁵Sc nucleus. Subsequently, the superoxide ion, O₂^•− abstracts a proton (proton transfer = PT) from the benzyl radical cation to produce a hydroperoxy radical, HOO^• and a benzyl radical. Subsequently, the interaction between the benzyl radical and O₂ produces the peroxyl radical, followed by benzaldehyde formation. Further oxidation of benzaldehyde yields the desired product, benzoic acid (2a), as well as H₂O₂, as confirmed by iodometry (see Supporting Section S11 for details). Alternatively, the aryl radical cation can react with a cyanide source (TMSCN) to generate a cyclopentadienyl radical and produce a cyanoarene by hydrogen atom transfer (HAT). The final transformation using the hydroperoxy radical is likely the one proposed by Nicewicz⁴⁸ but an alternative participation of oxygen is not excluded.

Fig. 7. Proposed mechanism of aerobic oxidative transformations catalysed by Sc³⁺ complexes in acetonitrile.

The mechanism involves excitation of Sc³⁺ species upon oxygen coordination, intersystem crossing (ISC), singlet electron transfer (SET) from a substrate (toluene, 1a) and transformation of thus formed radical cation either (i) by proton transfer (PT) followed by the reaction with oxygen to form oxygenation product 2a, or (ii) by the reaction with cyanide source followed by hydrogen atom transfer (HAT) to form benzonitrile 19.

From mechanistic point of view, our method using redox-innocent Sc³⁺ cation represents another mode of metal ion functioning in oxidative catalysis and brings an alternative to (i) Ir or Ru complexes with high oxidation power^8–10, and (ii) to the recently developed systems based on ligand-to-metal charge transfer employing halides of metals like Ce, Fe, or Bi. The latter method uses halogen or alkoxy radical generation which causes hydrogen atom transfer^51–55. Method with Sc³⁺ is based on MCET and as such, it has some limitations given by electron transfer thermodynamics. Nevertheless, on the example of 4-(trifluoromethyl)toluene (1c) oxidation, we have demonstrated possible overcoming this limit by the presence of a proton source (trifluoroacetic acid).

Discussion

In conclusion, all the results obtained during this study confirm that Sc(OTf)₃ possesses noteworthy photocatalytic activity. Hence, we introduce an aerobic oxidation approach for benzylic substrates with Sc(OTf)₃ as the sole photocatalyst. This system revealed considerable efficiency with a wide range of benzylic substrates, including alcohols, toluene derivatives, and methylene-containing substrates. A straightforward Sc(OTf)₃-based photocatalytic approach for the oxidative cyanation of arenes under aerobic conditions was also developed as an example of direct aromatic C-H derivatization resulting in C-C bond formation. The photooxidative cyanation method, despite some limitations (e.g., the cyanation of electron-poor substrates), demonstrated an interesting efficiency with a notable regioselectivity across the tested scope of arenes.

Our results on benzylic photooxidation and oxidative cyanation mediated by Sc(OTf)₃ as sole photocatalyst break the long-held belief that Sc³⁺ salts serves only as a potent Lewis acids for activating organophotocatalysts or substrates. We expect that Sc³⁺ salts can find applications as photocatalysts also in other oxidative organic reactions, some of which are now under investigation in our laboratory. From a general perspective, our examples could pave the way for the exploration of other redox-innocent metals in photoredox catalysis.

Methods

General procedure for photooxidation of benzylic substrates

Small scale experiments (A)

A vial was charged with a mixture of the substrate (140 μmol) and Sc(OTf)₃ (5 mol%, 7 µmol) in MeCN (250 μL). The reaction mixture was bubbled with oxygen (2 min) and then stirred at 45 °C under irradiation with 400 nm LEDs. After selected time, the reaction mixture was diluted with DMSO-d₆ and the yield was determined by ¹H NMR.

Preparative experiments (B)

A mixture of the substrate (1 mmol) and Sc(OTf)₃ (5 mol%) in MeCN (5 mL) was bubbled with oxygen (2 min). It was then stirred at ambient temperature in a 50 mL Schlenk tube under irradiation with 400 nm LEDs under aerobic conditions (oxygen balloon). The reaction progress was monitored by ¹H NMR. When the reaction was completed, the solvent was evaporated, and the crude product was purified either by column chromatography on silica gel or by extraction (see Supplementary Section S3 for details).

General procedure for photocatalytic cyanation of arenes

Small scale experiments (C)

A vial was charged with a mixture of the substrate (140 μmol), Sc(OTf)₃ (5 mol%, 7 µmol) and TMSCN (12 equiv., 1.68 mmol) in MeCN (250 μL). The reaction mixture was bubbled with oxygen (2 min) and then stirred at 45 °C under irradiation with 400 nm LEDs. After selected time, the reaction mixture was diluted with DMSO-d₆ and the yield was determined by ¹H NMR.

Preparative experiments (D)

A mixture of the substrate (1 mmol), Sc(OTf)₃ (5 mol%) and TMSCN (12 equiv., 12 mmol, 1501 μL) in MeCN (5 mL) was bubbled with oxygen (2 min). It was then stirred at ambient temperature in a 50 mL Schlenk tube under irradiation by 400 nm Luxeon LEDs under aerobic conditions (oxygen balloon). The reaction progress was monitored by TLC and ¹H NMR. When the reaction was completed, the solvent was evaporated, and the crude product was purified by column chromatography on silica gel (see Supplementary Section S6 for details).

Author contributions

A.H.T. and R.C. conceived the idea. A.H.T. was responsible for designing and conducting experimental procedures and measurements, analysing the data, and interpreting the results. A.E.-Z. and J.I.K. performed transient absorption experiments. A.E.-Z. contributed to discussion of mechanism. J.S. performed the DFT calculations. J.K. and K.L. contributed to EPR measurements and their evaluation. J.C. performed MS measurements. M.P. performed cyclic voltammetry measurements. E.S. provided data managements and investigation and the corrections of the supplementary information and provided measurements with ¹⁸O₂. A.H.T., A.E.-Z., and R.C. collaborated in drafting the manuscript. R.C. directed the project.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data supporting the results of the article, including optimization studies, experimental procedures, compound characterization, mechanistic studies, and details on quantum chemical calculations are provided within Supplementary Information and Supplementary Dataset. All information (raw, processed and visualized data) about prepared compounds, techniques used for characterisation of prepared compound, NMR, MS, spectral and other used methods are available in this dataset: 10.5281/zenodo.14328398. Data supporting the findings of this manuscript are also available from the corresponding author upon request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-63233-4.