Dramatic expansion of bimodal redox window of indigo by two-electron redox processes

Abstract

Indigo is an extremely popular molecule in dye industry, however, its use in photochemical transformations is surprisingly scarce. This report explores its photocatalytic activity over an unusually wide excited-state redox window, spanning over 5.98 V. The dye molecule exhibits bimodality and proves itself as a simultaneous super-reductant and -oxidant. This extreme bimodal behavior in indigo (IndH₂) originates from the viability of two electron redox processes on the parent architecture. In comparison, major popular photocatalysts (PC) possess singly oxidized/reduced state, limiting the span of such bimodal redox window significantly. In the presence of KO^tBu and white light irradiation, IndH₂ is converted to its tetraanionic form Ind^4- by two-electron reduction and two successive deprotonation steps, exhibiting its reductive power to −3.6 V vs SCE. On the other hand, two-electron oxidized form of IndH₂ forms dehydroindigo, a superoxidant capable of oxidizing substrates up to +2.38 V vs SCE.

Indigo is an extremely popular molecule in dye industry, however, its use in photochemical transformations is surprisingly scarce. This report explores its photocatalytic activity over an unusually wide excited-state redox window.

Introduction

Visible light photoredox catalysis has likely seen a renaissance in recent times and has become a powerful tool for solving challenging chemical transformations that are sometimes even impossible to achieve by thermal pathways^1–5. The utility of these catalysis processes further lies in the selective activation of specific functional groups by appropriate photon energy, keeping other vulnerable groups intact during those transformations. Earlier developments in photocatalysis heavily utilized ruthenium^6,7, iridium^8–10, palladium-based^11,12 catalysts, while there is a clear shift in momentum to discover more organic photocatalysts as they could be more sustainable and environmentally benign. A broad interest in developing new organic photocatalysts (PC) or unraveling a new facet of existing photoactive molecules propels the recent photochemical research^13,14. Typically, a PC absorbs a photon and reaches its electronic excited state which facilitates either energy or electron transfer to different substrate molecules. Different strategies have also been further devised to harness the reducing or oxidizing power of a photocatalyst at its excited state^14–19. Tandem electro-photochemical approach^20–22, sensitization-enhanced electron transfer^17,23, consecutive photoinduced electron transfer (con-PET) are a few of the popular techniques in this direction^{3,5,18,24–27}. In the context of the excited state reducing/oxidizing power of a PC molecule, its ability to operate in both directions is an important attribute to the PC (Fig. 1a)²⁸. Such accessibility of bimodal operation ensures that each oxidative or reductive transformation does not require a tailor-made PC. Some of the widely used transition metal-based photocatalysts exhibit bimodality, however, over a rather narrow range. For example, [Ru(bpy)₃]²⁺ displays the excited state oxidation and reduction potentials at −0.81 V and + 0.77 V (vs SCE), respectively^28,29, spanning an excited state redox window of 1.58 eV. Similarly, fac-Ir(ppy)₃ photocatalyst displays the same potentials at −1.73 V and +0.31 V, respectively, covering a potential range of 2.04 eV (Fig. 1b)^28,29. Organic photocatalysts, often owing to their high excitation energy (E_o,o), afford a sizeable redox window; for example 4CzIPN spans a window of 2.61 eV. However, the excited state oxidation potential for the aforementioned PC, −1.18 V is barely adequate to break a difficult-to-reduce aryl chloride bond (E_red = −2.8 V vs SCE). By the same token, the excited state reduction potential of the same catalyst, +1.43 V, is nowhere close to oxidize the reluctant arenes, since such substrate molecules demand an oxidation potential of + 2 V or even higher³⁰. Henceforth, a single PC molecule that can afford a very large excited state redox window, eliciting the simultaneous super-reductant and super-oxidant trait, is virtually unknown.

Fig. 1. Selective C–C and C–N bond formation via indigo dye as a photocatalyst.

a Conceptual framework to expand the bimodal redox window. b Comparison of bimodality with popular PCs and salient feature of indigo. c Model reactions to demonstrate the oxidative and reductive power of indigo.

In a closer scrutiny, the bottleneck directly hints at the operational mode of these PC. Almost invariably, the PC in its excited state releases one electron to reductively cleave a target bond, or it accepts an electron to conduct oxidative transformation to the substrate molecule. So, the span of its excited state potential encompassing both oxidative and reductive ends is limited by single electron redox event. This limitation primarily stems from the inherent nature of the PC, where two-electron redox events are not accessible. Intuitively, if a PC undergoes two-electron redox processes to generate an active catalyst along both directions without any detrimental bond cleavage, there is a strong possibility that it will virtually double its operational redox window. However, the chemical transformation will be dictated by one-electron redox processes. To illustrate, a PC molecule can act simultaneously as both super-reductant and super-oxidant once it can manage two-electron redox transformations onto it. We identify a popular dye, indigo that possesses a cross-conjugated p-quinone and a p-phenylene diamine-type architecture^31,32. We posit such redox motifs will allow to perform two-electron oxidation and two-electron reduction discretely so that the excited state window can reach a massively large value. Herein, it is further demonstrated that simple in situ chemical modification of indigo by an extremely mild reductant and a mild oxidant can prepare the highly reducing and oxidizing states, respectively so that their super-redox trait can be harnessed over an unusually wide potential window of 5.98 eV (Fig. 1a).

Results and discussion

Photophysical study of catalytic intermediates

Indigo’s prolific use as a dye element to cloths, fabrics are age old^33–35. The deep blue color and immense photostability of the indigo under sunlight are perhaps the reasons behind its enormous popularity as a dyeing agent. Its tremendous photostability even sparked research interest to find means to decompose such a molecule in the presence of a catalyst. Surprisingly, such a robust photocatalyst has not been examined widely for steering photochemical reactions. Parent indigo molecule, IndH₂ strongly absorbs in the visible region (620 nm) and results in dark blue color appearance (Fig. 2a). To augment the reductive prowess of indigo, we treated IndH₂ with a mild reductant, KO^tBu and was able to reduce it in full under white light excitation by a LED lamp. Our primary plan is motivated by con-PET mode, so that one-electron-reduced IndH₂ is further photoexcited and becomes strongly reducing^3,18. However, a higher equivalent of KO^tBu completely deprotonated the starting species, parallel to its double reduction, generating Ind⁴⁻³⁶. The color of the putative species dramatically changes to red-violet from the starting blue color of parent IndH₂ (see supplementary information for details, Section 4.2). This Ind⁴⁻ species is extremely reducing in character and reductively cleaves a very demanding aryl chloride as well as aryl fluoride bond (vide infra)²⁰. The putative tetraanion intermediate is distinctly different from doubly-reduced leuco indigo, which is widely used in the dye industry to make the indigo water soluble³⁷. The present active catalyst Ind⁴⁻ is doubly deprotonated in tandem with two-electron reduction. To ascertain that double deprotonation of IndH₂ is required on top of two-electron reduction, we performed a control reaction (see supplementary information for details, Section 4.4.c). We used a strong reductant, metallic sodium (2 equiv.) to completely reduce IndH₂ and authenticated the leuco form by the presence of the N–H proton resonating at 10.5 ppm. Such a doubly reduced, yet protonated leuco-indigo species was employed to break a recalcitrant aryl chloride bond. Notably, the leuco form was unable to perform such a challenging reduction, strongly suggesting photoexcitation of Ind⁴⁻ is the requisite stage to elicit its super-reducing behavior. The Ind⁴⁻ form is sensitive to oxygen; however, the molecule can remain stable for weeks in an inert atmosphere of a glove box. To gauge the extreme reducing power of this form, we at first measured its ground state oxidation potential by cyclic voltammetry. In the electrochemical time scale, it displays an irreversible wave at a peak potential of −1.2 V vs SCE (see supplementary information for details, Section 4.3.a). Notably, under the electrochemical conditions applied for this measurement, Ind⁴⁻ is sufficiently stable so that all the redox waves were reproducible for at least three consecutive runs (see supplementary information for details, Section 4.3.a). The tetraanionic molecule shows a broad absorption from 400-550 nm, with a maximum around 478 nm (Fig. 2b). When excited at 478 nm, Ind⁴⁻ fluoresces at 540 nm. The excitation energy E_o,o is calculated to be 2.4 eV from the intersecting point of its absorption and emission spectra. Using the Rehm-Weller equation, the excited state potential can be estimated as E* = −1.2−2.4 V = −3.6 V vs SCE (see supplementary information for details, Section 5)³⁸. Since we are accounting for the peak potential of the anodic wave, rather than a truly reversible potential, we may take this potential as the upper limit of its reducing power. Nevertheless, such a large negative redox potential completely justifies its super-reducing behavior of the in situ-generated Ind⁴⁻ upon visible light excitation condition. The electron transfer from such a strongly reducing Ind⁴⁻ to aryl chloride is extremely favorable by a thermodynamic drive (ΔG) of 16.1 kcal mol⁻¹. To investigate further the excited state electron transfer from the putative species to aryl chlorides, a Stern-Volmer quenching experiment was conducted. Along this direction, a gradual increase in chlorobenzamide concentration resulted in sequential decrease of fluorescence intensity of Ind⁴⁻ at 540 nm. Once the ratio of fluorescence intensities (in the absence and in the presence of quencher) was plotted with increasing chlorobenzamide substrate concentration, a clear straight line with an K_SV value of 785 M⁻¹ was derived (see supplementary information for details, Figs. S12 and S13). Apparently, the administered base KO^tBu plays a dual role of doubly reducing the indigo, and simultaneously doubly deprotonating to generate the most reactive Ind⁴⁻ form. To examine the fate of this reductive catalyst, we prepared the Ind⁴⁻ species in situ by the addition of 4.5 equiv of KO^tBu to a DMSO solution of IndH₂ under visible light and added one equivalent of substrate chlorobenzene. The radical anion, Ind³⁻ was generated instantaneously upon visible light excitation that was clearly detected by X-band EPR spectroscopy at room temperature. A broad signal at g = 1.92 emerges without any discernible hyperfine couplings from the nitrogen at room temperature. However, the broad nature of the signal likely testifies to the delocalized nature of the radical generated at the indigo backbone upon an electron transfer from the PC (see supplementary information for details, Section 7). We will be showing through a plethora of reactivity studies that this Ind⁴⁻ promptly transfers electron via photoinduced electron transfer (PET) pathway to reductively demanding substrate molecules.

Fig. 2. Mechanistic studies.

a UV-vis absorption spectra of neutral indigo and its tetraanionic form b Juxtaposition of absorption and emission profile of Ind⁴⁻. c X-band EPR spectrum of Ind^3−•. d Excited state lifetime of Ind⁴⁻, as measured by TCSPC. e Control reaction to prove Ind⁴⁻ as the active catalyst. f Cyclic voltammetry of Ind⁴⁻. g Plausible mechanism for isoindoline synthesis.

We reiterate that Ind⁴⁻ is the in situ-generated active species that elicits such a strong reducing power. To verify that simple di-deprotonated indigo Ind²⁻ is not capable of directing reductive cleavage under such a demanding potential, we prepared the latter by deprotonating both -NH protons of IndH₂ by Cs₂CO₃. It has been proved earlier that Cs₂CO₃ is sufficiently basic to carry out the complete deprotonation³⁹. However, this doubly deprotonated form under visible light irradiation converted the test substrate 2-Chloro-N,N-bis(1-methylethyl)benzamide to isoindolinone in mere 11% yield (Fig. 2e). To substantiate our hypothesis further, that double deprotonation is required in parallel to two-electron reduction, we synthesized a N,N’-diphenyl indigo, replacing two N–H protons by phenyl groups (see supplementary information for details, Section 4.4.e)^39,40. Logically, upon photoexcitation, N,N’-diphenyl indigo can only be doubly reduced. In strong agreement with our conjecture, such a modified indigo molecule was not highly reducing. Taking p-chloroanisole as a model substrate, a cross-coupling reaction was performed in benzene to obtain the C–C cross-coupled product in a poor 25% yield (see supplementary information for details, Section 4.4.f). This lack of extreme reductive power in Ind⁻² is also in alignment with CV data. The electrochemical analysis proves that Ind⁻⁴ is approximately 1 V more reducing in its ground state than Ind⁻² (see supplementary information for details, Section 4.3.b). Taken together, all these experiments strongly suggest that double deprotonation and simultaneous two-electron reduction are the requisite to prepare the super-reducing species. Very importantly, all these processes happen in the presence of KO^tBu under visible light excitation.

After exploring the reductive end of the indigo molecule, we set out to explore its oxidative prowess. We anticipate, the presence of p-phenylene diamine architecture will help to reach dehydroindigo (Ind), upon 2e⁻/2H⁺ removal from parent IndH₂³¹. The Ind form is the most oxidized state of indigo, and will likely be a super-oxidizing species upon photoexcitation. With this goal, we identify potassium peroxodisulfate to be the desired oxidant that oxidizes IndH₂ completely. This oxidation also happens under visible light excitation, which cleanly generates the dark brown solid Ind in quantitative yield. The chosen oxidant works very efficiently, given the hydrogen atom abstraction ability of SO₄^•−, generated from S₂O₈²⁻ promoting oxidation. The importance of visible light initiation was investigated further, and it was proved that the oxidation did not proceed at all under dark condition. The fully oxidized Ind is a very stable molecule which can be purified by column chromatographic separation to isolate the targeted dark brown granular solid. The nature of the isolated molecule was confirmed by its absorption, emission, electrochemical behavior, and high-resolution ESI-mass spectrometry. Gratifyingly, the spectroscopic signature of the molecule corroborates well with prior literature report³¹. The molecule in DMSO solution absorbs at 397 nm, which is compatible with the observed color for the molecule (Fig. 3a). The ground state reduction of the molecule occurs at −0.43 V vs SCE, which can be translated to its excited state reduction potential as +2.38 V vs SCE (see supplementary information for details, Section 5). This value further promises that in situ-generated Ind will be a super-oxidant and will activate many reluctant substrates in an oxidative fashion. In this context, an oxidation of p-fluoroanisole by Ind will be thermodynamically propelled by a driving force of 13.4 kcal mol⁻¹.

Fig. 3. Oxidative cleavage.

a Absorption spectrum of Ind. b Emission spectrum of Ind, when excited at λ = 445 nm. c In situ generation of Ind. d Cyclic voltammetry of Ind. e Control experiments. f Substrate scope for the C–N bond formation reaction. g Representative example for C–H amination reaction.

Substrate scope

To evaluate the synthetic potential of indigo to reductively cleave aryl halide bonds, we started our foray with an array of α-chloro amide substrates^41,42. The standard reaction conditions were optimized to involve 3 mol% indigo loading in the presence of 3 equivalent of base KO^tBu. The reaction mixture was irradiated with a white LED (see supplementary information for optimization details, Section 2.1). Under this condition, the α-chloro amide 1 afforded the isoindolinone in 93% yield. Interestingly, it was observed that the reactions finished in 4 h, attributable to the highly reactive nature of the super-reductant Ind⁴⁻. We do not suspect the dimsyl anion to be responsible for this electron transfer catalysis despite the use of DMSO solvent in the presence of KO^tBu. The reducing ability of the photoexcited dimsyl anion is limited and can only cleave aryl-iodide or aryl-bromide bonds⁴³. Substrates with N,N’-ethyl, -isopropyl groups selectively went for 1,5-HAT from the isopropyl motif to result in isoindolinone 2 in 87% yield (Fig. 4). Varying the substitutions at the amide nitrogen to methyl and benzyl were well tolerated and the reaction smoothly converted these substrates to the respective isoindolinones 3 and 4 in 86–89% yields. A variety of substitutions at the phenyl ring, including -Me at the meta or para position, survived well the reaction condition to furnish isoindolinones (5–6, Fig. 4) in 77–88% yields. Heterocyclic ring pyridine also afforded the final isoindolinone product 7 in 76% yield. Furthermore, a set of sophisticated isoindolinones (8–9, Fig. 4) were also prepared by this methodology in 78-86% yields. Spirocyclic isoindolinone can also be synthesized under this method, as isoindolinone 10 was prepared in high, 87% yield. After that, the starting substrate was varied to N-alkylated α-chloro anilides. The chosen anilide responded equally well under the optimized reaction conditions to afford the oxindole 11 in 84% yields. Electron-donating methoxy, methyl groups in the chloro anilide furnished products 12–14 in 71–78% yields. On the other hand, strongly electron-withdrawing functional groups such as trifluoromethyl, nitro, cyano responded well to the reaction conditions to furnish 15–17 in 68–88% yields. Furthermore, a spirocyclic oxindole 18 was synthesized under this developed methodology in 81% yield. Next, N-benzyl acetamide was examined, which smoothly converted to the N-benzyl oxindole product 19 in 68% yield. On the other hand, α-halo benzamides 20 assembled the corresponding phenanthridinone in high yield (81%). Interestingly, phenanthridione 21, a known precursor of an antitumor agent was synthesized in 79% yield. Finally, using this protocol, a bioactive alkaloid phenaglydon, 22, was assembled in 83% yield, starting from an aryl bromo precursor molecule⁴⁴.

Fig. 4. Scope of cyclic C–C bond formation reactions.

Reaction conditions—PC (3 mol%), KO^tBu (3 equiv), White light, DMSO, 4 h, rt. ^a24 h.

After that, we exploited the power of indigo as a strongly reductive photocatalyst in a series of C–C, C–B, and C–P bond formation reactions^3,20. For this set of reactions, unsubstituted aryl halides were chosen as substrates, which were very difficult-to-break via single electron transfer pathway⁴⁵. In this manner these substrates can truly gauge the reductive capability of the PC, as unsubstituted or electron-rich aryl chlorides require a potential of ca −2.9 V vs SCE to promote reductive cleavage⁴⁶. Following this methodology, a series of aryl chloride molecules were utilized as an aryl radical precursor and cross-coupled with a different heteroarenes such as N-methyl pyrrole, thiophenes or furans. The C–C cross-coupling products (23–26, Fig. 5) were isolated in 55-64% yields. The reaction was performed in a relatively redox-inert solvent DMSO; however, the moderate yield of the cross-coupled product may originate from super-reducing Ind⁴⁻, which makes the DMSO a vulnerable molecule towards reduction. Furthermore, α-naphthyl chloride was reductively cleaved to generate the corresponding naphthyl radical which was coupled to N-methyl pyrrole to result 27 in 59% yield. Similarly, the p-trifluoromethyl chlorobenzene was examined with furan to synthesize 28 in 65% yield. The m-methyl, p-methoxy, p-cyano, 3,5-dimethyl chlorobenzene were reductively cleaved and cross-coupled to thiophene to furnish products 29–32 in 57–71% yields. Naphthyl chloride and 5-methoxy naphthyl chlorides were similarly cross-coupled to thiophenes to provide 33–34 in 58–64% yields. Analogously, a photo-Arbuzov reaction was attempted with aryl chloride to prepare aryl phosphonates, which resulted in the preparation of 35 in 52% yield⁴⁷. Furthermore, a series a borylation reaction was investigated starting from aryl chloride substrate in the presence of pinacol borane. Gratifyingly, five different aryl boronates 36–40 were synthesized in 49–64% yields following our developed protocol. A series of C–C cross-coupled products were assembled using benzene as the reaction partner. Aryl chlorides with electron-withdrawing trifluoromethyl, cyano groups resulted in cross-coupled product 41–42 with benzene in 78–81% yields. The electron-rich aryl chloride such as p-methoxy aryl chloride afforded 58% yield of the product 43. The same product can also be attempted from even more challenging substrates like aryl fluoride, however, the product forms in poor yield^4,45. We further attribute the lower yield to competitive reduction of DMSO at such a demanding potential, at which the Ind⁴⁻ is capable of delivering an electron towards reductive cleavage reaction. Different naphthyl chloride/bromides were successfully cross-coupled to furnish 50, 51, 53 in 62–81% yields. Such cross-coupling reactions can also be stretched to polycyclic rings effectively. Accordingly, 1,4- and 1,3-dibromobenzene can be easily converted to terphenyl 54 and 56 in 54–57% yields. Encouragingly, several heterocyclic aryl halides such as pyridyl or thiophenyl bromides were tested further, which can furnish the desirable C–C cross-coupled products 57–60 in moderate to good yield. Additionally, the strong reducing ability of indigo was utilized for the N–O bond cleavage of Weinreb amides. Two such substrates readily underwent the reductive condition to smoothly cleave the N–O bond resulting N-methyl benzamides 61–62 in 89–93% yields⁴⁸. Detosylation of an amide is also a reductively demanding reaction. Our model reaction on N-methyl,N’-tosyl aniline was successful to isolate the tosyl-free, N-methyl aniline 63 in 58% yield¹⁶. Despite all these successful reactions, we were unable to apply such a super-reductant for photochemical Birch reduction of naphthalene. The proton source required for the isolation of the reduced product poisons the reactive Ind⁴⁻ and makes such attempts futile.

Fig. 5. Scope of C–C, C–P, C–B bond formation reactions, Weinreb amide reduction and Detosylation of amine.

^aP(OMe)₃ (2 mmol), DMSO (1.5 mL). ^bB₂pin₂ (2 mmol), DMSO (1.5 mL). ^cBenzene (2 mL). Reaction conditions- PC (3 mol%), KO^tBu (3 equiv), White light, DMSO, 24 h, rt.

Mechanistically, Ind⁴⁻ is converted to Ind^3−• by transferring a single electron to a substrate aryl chloride during its reductive cleavage. The resulting aryl radical intermediate II from the C(sp²)–Cl bond cleavage performs a 1,5-hydrogen atom transfer to generate an isopropyl radical. Further ring cyclization followed by deprotonation forms a highly reducing radical anion III that can cleave further aryl chlorides (Fig. 2g). We believe a chain propagation cycle starts, which heavily assists the aryl radical generation from aryl chloride. As strong evidence for the chain propagation step, we measured the reaction quantum yield for the isoindolinone formation to be 19 (see supplementary information for details). In addition, III can likely reduce Ind^3−• to bring Ind⁴⁻ back, only when the former remains at the photoexcited state. Although the step is difficult, invoking diffusion-limited bimolecular collision of two reactive radicals, product yields of specific substrates suggest contribution from catalytic loop. Alternatively, beyond the highly reducing initiator Ind⁴⁻, Ind^3−• also participates in cleaving C(sp²)-Cl bonds and contributes effectively toward aryl radical generation. Similarly, engendered aryl radical from reductive cleavage of aryl chlorides leads to a C-C  cross-coupling with (hetero)arenes. The C–C cross-coupling reaction follows a pathway involving base-promoted homolytic aromatic substitution reaction^49,50. The generation of arylcyclohexadienyl radical anion is critical in this reactions which assists significantly in the product formation step by creating a chain propagation loop.

After successfully expanding the substrate scope at the reductive end (Ind⁴⁻), we sought to explore the oxidative prowess of the catalyst (Ind). Along this goal, we chose a nucleophilic aromatic substitution (S_NAr) reaction. If oxidatively demanding fluoroarenes can be oxidized, a nucleophile can attack and, through further electron transfer, may furnish the C–N coupled product. When indigo in the presence of 2 equiv. of K₂S₂O₈ was used under photoexcitation, p-methoxy fluoroarene was conveniently converted to the C–N coupled product 64 in 72% yield (see supplementary information for optimization details, Section 2.2). A variety of nucleophiles, including triazole, benzimidazole and benzotriazole were examined for the S_NAr reaction, successfully affording the desired products 65, 66, 68 in 21–60% yields (Fig. 3). Furthermore, the ortho-substituted electron-rich systems readily underwent S_NAr C-F substitution with 4-iodopyrazole and 4-carboxylatepyrazole to afford the products 67 and 69 in 60–67% yields. Methyl substituted fluorobenzene also showed promising reactivity under this protocol when the nucleophile was selected to 4-carboxylatepyrazole resulting into the desired products 70–71 in 59–62% yields. Finally, the nucleophile was changed to 2H-1,2,3-triazole, and it was reacted with 3,4-dimethoxyfluorobenzene to synthesize 72 in 50% yield. Given the super-oxidizing ability of Ind, we further posit that direct C–H amination of an arene can also be viable. Indeed, such aminations from direct C–H activation has been valued as an important C–N bond formation technique. A model reaction with 3,5-dimethoxy benzene in the presence of pyrazole nucleophile cleanly converted to the desirable C–N coupled product, 1-(2,4-Dimethoxyphenyl)−1H-pyrazole 73 in 78% yield⁵¹. In the S_NAr reaction on fluorobenzene, the in situ-generated Ind oxidizes the former by one electron, to generate the radical cation of fluorobenzene. The high oxidizing power of Ind at its photoexcited state makes this redox process feasible. Upon this activation, the nucleophilic pyrazole attacks the radical cation and forms a Meisenheimer intermediate^30,52,53. Very likely, the resulting Ind^•− upon photoexcitation reduces the Meisenheimer intermediate to release the nucleofuge fluoride anion, and the product is formed (see details in Supplementary Information, Section S9).

We showcase that indigo, an unexplored photocatalyst, possesses tremendous potential in directing a plethora of photochemical transformations. The presence of a quinone-type and aminoquinol-type cross-conjugated motif in indigo gives access to two-electron reduction, as well as two-electron oxidation. Furthermore, the presence of two deprotonable N–H groups affords double deprotonation to prepare a super-reductant motif. This architectural pattern facilitates accessing simultaneous super-reductant and -oxidant behavior over a colossal redox window of 5.98 V. Compared to this value, a large number of photocatalysts are limited to avail a potential window of ~3 V, likely stemming from their ability to be oxidized or reduced by single electron only. We believe this study is a great entry point for utilizing indigo as a photocatalyst, and to show what design principles can elicit simultaneously strong reductive and oxidative chemistry from a single molecule.

Methods

General procedure for C–C cross-coupling reactions

An oven-dried 15 mL pressure tube was charged with indigo (3 mol%), aryl halide (0.1 mmol), and KO^tBu (3 equiv). After that, dry benzene (1 mL) and dry DMSO (1 mL) was added to the tube and the tube was closed properly. The reaction mixture was stirred under visible light irradiation for 24 h at room temperature. Upon completion, the reaction mixture was worked up in ethyl acetate and brine mixture. Then the resulting solution was dried over MgSO₄, filtered, and concentrated under reduced pressure. Subsequent purification by column chromatography with hexane as the eluent afforded the corresponding product.

Preparation of the Ind⁴⁻ intermediate

To prepare this key intermediate, indigo (1 equiv) and KO^tBu (4.5 equiv) were placed in a pressure tube, followed by the addition of DMSO. The reaction mixture was sealed under an inert atmosphere and irradiated with white LED light for 10 min with continuous stirring. During irradiation, the solution gradually turned purple, indicating the formation of the Ind⁴⁻ species.

Preparation of the Ind intermediate

To prepare Ind, indigo (1 equiv) was combined with K₂S₂O₈ (2.5 equiv, in DMSO) in a reaction tube. The reaction mixture was stirred for 30 min under blue LED irradiation. The reaction mixture was extracted with brine solution and ethyl acetate. After that, the organic layer was separated, and dried under vacuum to isolate a dark brown solid.

Author contributions

M.R. and D.A. conceived the idea. M.R., S.M., and D.A. designed, performed, and analyzed the experimental data. S.M. explored the substrate scope and performed some mechanistic experiments. V.S. explored the substrate scope of several cyclic C–C bond-forming reactions, while D.D. investigated the substrate scope of cross-coupling reactions with different heteroarenes. D.A. supervised the work and was also responsible for the fund acquisition. M.R. and D.A. prepared the manuscript with the input from other authors.

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 authors declare that the data supporting the findings of this study are available within the paper and its Supplementary Information file. The experimental procedures, control experiments, and characterization of all new compounds are provided in the Supplementary Information. Supporting data for the 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/s41467-025-66186-w.