Discovery and characterization of an acridine radical photoreductant.

Photoinduced electron transfer (PET) is a phenomenon whereby the absorption of light by a chemical species provides an energetic driving force for an electron-transfer reaction1-4. This mechanism is relevant in many areas of chemistry, including the study of natural and artificial photosynthesis, photovoltaics and photosensitive materials. In recent years, research in the area of photoredox catalysis has enabled the use of PET for the catalytic generation of both neutral and charged organic free-radical species. These technologies have enabled previously inaccessible chemical transformations and have been widely used in both academic and industrial settings. Such reactions are often catalysed by visible-light-absorbing organic molecules or transition-metal complexes of ruthenium, iridium, chromium or copper5,6. Although various closed-shell organic molecules have been shown to behave as competent electron-transfer catalysts in photoredox reactions, there are only limited reports of PET reactions involving neutral organic radicals as excited-state donors or acceptors. This is unsurprising because the lifetimes of doublet excited states of neutral organic radicals are typically several orders of magnitude shorter than the singlet lifetimes of known transition-metal photoredox catalysts7-11. Here we document the discovery, characterization and reactivity of a neutral acridine radical with a maximum excited-state oxidation potential of -3.36 volts versus a saturated calomel electrode, which is similarly reducing to elemental lithium, making this radical one of the most potent chemical reductants reported12. Spectroscopic, computational and chemical studies indicate that the formation of a twisted intramolecular charge-transfer species enables the population of higher-energy doublet excited states, leading to the observed potent photoreducing behaviour. We demonstrate that this catalytically generated PET catalyst facilitates several chemical reactions that typically require alkali metal reductants and can be used in other organic transformations that require dissolving metal reductants.

Our lab, as well as others, have published numerous examples highlighting the diverse reactivity of acridinium salts, such as Mes-AcrBF, as photooxidation catalysts in the excited state (*Mes-Acr; Figure 1A).[13] Upon absorption of visible light, the corresponding excited state of the acridinium salt is populated and may be quenched via electron transfer from an electrochemically-matched substrate, resulting in the formation of an acridine radical (Mes-Acr•; Figure 1A). In past work utilizing acridinium photoredox catalysts, this radical is typically oxidized to regenerate the parent acridinium and close a catalytic cycle. During previous mechanistic studies conducted by our laboratory, it was noted that solutions of Mes-Acr• generated via reduction of Mes-AcrBF with cobaltocene were indefinitely stable under oxygen-free conditions and possessed two major absorption features (350–400 nm and 450–550 nm; Figure 1B).[14] These observations led us to explore the photophysical behavior of this radical, with a focus on identifying potential PET behavior. Previous studies have detailed the in situ generation and excitation of stable cation radical and anion radical species and their use in catalytic reactions,[15-19] indicating the potential feasibility of this strategy and prompting our studies of the photophysical behavior of Mes-Acr•.

Figure 1:

Mechanistic studies on Mes-Acr radical.

a, Reduction potential of various elemental alkali metals compared to peak reducing potential of Mes-Acr•, b, Absorbance and emission (excitation = 400 nm) profiles for Mes-Acr• in MeCN (5 mM, 1 mm path length) c, SOMO and LUMO +1 visualizations for Mes-Acr•. d, Transient absorption spectra of Mes-Acr• (2.5 mM, THF, 1 mm path length) collected with a 250 fs pump pulse centered at 400 nm e, Excited state energies calculated using SRSH-PCM/TD-DFT method for Mes-Acr• and frontier orbital plot showing donor and acceptor density for TICT excited state. f, Debromination reaction of brominated acridinium derivative giving circumstantial evidence for TICT state. Mes = mesityl; DIPEA = N,N-diisopropylethylamine; t-Bu = tert-butyl.

Upon investigation of the excited state dynamics of Mes-Acr•, we found that there are two main excited states, tentatively assigned as a lower energy doublet (D) and higher energy twisted intramolecular charge transfer state (TICT). The excited state energy for the doublet excited state of Mes-Acr• is estimated by averaging the energies of the lowest energy absorption maximum and the highest energy emission observed upon 484 nm excitation. The energy of the proposed higher order excited state is estimated by averaging the energies of the emission maximum near 490 nm and the maximum of the corresponding excitation spectrum monitored at this wavelength (see SI for full details of excited state energy calculations). Estimation of excited state energies in this fashion gives values of 2.31 eV for the energy of the proposed D1 excited state and 2.76 eV for the corresponding higher energy excited state (Figure 1E). Using the known electrochemical potential of Mes-Acr•[20] the excited state oxidation potentials of these states were estimated to be −2.91 V vs. SCE and −3.36 V vs. SCE, respectively. To our knowledge, these values represent some of the most negative excited state oxidation potentials reported for an organic molecule.

Before we proceed to discuss the calculated excited state energies, we consider the key orbitals involved in the low-lying excited states. We find that the singly occupied molecular orbital (SOMO) density is localized on acridine core and the lowest unoccupied MO (LUMO+1) is localized on the N-phenyl ring of Mes-Acr• (Figure 1C). Based on this observation of small spatial overlap between these two orbitals, we expect to find a relatively low lying excitation of an intramolecular CT (charge transfer) state. To further probe the excited state behavior of Mes-Acr•, transient absorption experiments were performed (Figure 1D). At early pump-probe delay times in THF, the ground state of Mes-Acr• is bleached (ΔA < 0) and excited state absorbance resonances (ΔA > 0) with maxima at 550 nm and ~650 nm are observed. Aromatic radical anions are known to exhibit broad absorbances in the 600–800 nm range as is aqueous solvated electron.[21-23] The observed excited state absorbance signal at ~550 nm also matches the absorbance profile expected for a general acridine exitonic structure. Simple first-order decay to the ground state occurs after ~100 ps, matching well with previously reported values for excited state lifetimes of organic radicals. TD-DFT calculations indicate that other red-shifted absorptions present are well matched with energies calculated for a general acridinium structure. These spectral features support the formation of a charge transfer state possessing both aromatic radical anion and acridinium features, as expected for the proposed TICT state.

To better understand the effect of rotation of the N-phenyl ring on the excited state energetics of the acridine radical, we employ the recently reported polarization consistent TD-DFT based framework for obtaining excited state energies of solvated molecular systems. The approach addresses dielectric polarization consistently by invoking the same dielectric constant in the screened range separated hybrid (SRSH) functional parameters and in the polarizable continuum mode (PCM). SRSH-PCM was benchmarked well in calculating CT state energies of solvated donor-acceptor complex and in analyzing the spectral trends of several pigments with increased accuracy where conventional TD-DFT calculations fail to reproduce the observed trends (see SI for full computational details).[24-27]

The calculated doublet excited state energies for Mes-Acr• agree very well (within 0.1 eV) with values estimated via spectroscopic measurements for both the absorption and emission spectra (Figure 1E). The lowest energy calculated D1 state, possessing an excited state energy of 2.29 eV, agrees with the experimentally determined D1 value of 2.31 eV. Additionally, two excited states possessing significant charge transfer character were identified and the corresponding energies were calculated to be 2.75 and 2.78 eV, matching closely with estimated spectroscopic values for the proposed TICT state energy of 2.76 eV. As such, the identified D1 (2.29 eV) state is assigned as an untwisted exitonic state, while the calculated 2.78 eV state is assigned as a TICT state. These excited state energy values also correspond well with previously reported excited state energies for neutral radical species.[9] Additionally, visualizations of the geometries of the corresponding TICT state indicate significant rotation of the N-phenyl ring (36 degrees) relative to the more planarized geometry of the D1 state, providing further evidence of the profound effects of N-phenyl rotation on excited state energy.

With the electronic and excited state behavior of Mes-Acr• elucidated, we sought to utilize this species as a catalytic reductant in a photoredox manifold. Previous work in reductive photoredox catalysis has established the reduction of aryl halides as a common benchmark reaction for these types of transformations.[15,28-30] Furthermore, the extremely potent reducing behavior of the acridine radical should enable the reduction of a wide range of electronically diverse substrates. Diisopropylamine (DIPEA) was identified as a suitable single electron reductant for the generation of Mes-Acr• from Mes-AcrBF in situ. Following excitation, Mes-AcrBF undergoes single electron reduction via electron transfer from DIPEA, generating the desired radical Mes-Acr•. To chemically probe the possibility of charge transfer to the N-phenyl ring, brominated acridinium 1 was prepared. In the presence of 3 eq. DIPEA, 1 was completely converted to a mixture of debrominated acridinum 1a and hydroacridine 1b following irradiation at 390 nm for 18 hours (Figure 1F). As aryl halide radical anions are known to quickly fragment to yield the corresponding aryl radicals, this experiment is indicative of the formation of radical anion character localized on the N-phenyl ring during excitation.

To evaluate the competency of this radical species as a catalytic reductant, conditions for the reductive dehalogenation of aryl halides were developed (Figure 2A). A variety of both electron-rich (6–13) and electron-poor (14, 15) aryl bromides afforded the desired hydrodebrominated products in good to excellent yields (NMR yields of products taken using HMDSO as an internal standard). It is of note that reductively recalcitrant aryl chlorides also participated efficiently in this reaction, in contrast to previously reported methods which are only effective for electron-poor (under visible light irradiation) or moderately electron rich aryl chlorides (under UVA irradiation).[31,32] A variety of both electron-donating (16–20) and electron-withdrawing (21–24) substituents were tolerated, with only slightly reduced yields in the case of electron-poor substrates. Substrates bearing ketone (30), carboxylic acid (31), and alcohol (28) functionalities all afforded the desired hydrodechlorinated products in good to excellent yield. Medicinally relevant pyridine (25, 26) and aryl carbamate (27) derivatives were also efficient substrates for this transformation. When substrate (23), which bears a trifluoromethyl substituent, was subjected to the reaction conditions, partial hydrodefluorination (5%) to yield the corresponding difluoromethyl derivative was observed in addition to hydrodechlorination. In all other examples, no Birch-type products resulting from overreduction are detected. The bis-reduction of polyhalogenated compounds (9a) and (9b) gave the corresponding bis-hydrodebromination (9) and bis-hydrodechlorination products in 58% yield and 46% yield, respectively. For compound 9b, 49% yield of the product resulting from mono-hydrodechlorination (9c) was observed in addition to the fully dechlorinated product.

Figure 2:

Reductive dehalogenation of aryl halides enabled by acridine radical photoredox catalysis

a, Reaction scope. b, proposed mechanism. †0.3 M reaction concentration. ‡0.5 M reaction concentration.

Based on prior work in reductive dehalogenation, the following mechanism is proposed (Figure 2B). Following excitation, Mes-AcrBF engages in single electron transfer with the tertiary amine reductant (DIPEA), generating Mes-Acr• and the corresponding amine cation radical. Mes-Acr• is then excited by 390 nm light, populating a combination of highly reducing Dn/TICT excited states, and undergoes electron transfer with an electronically matched aryl halide generating an arene radical anion and reforming Mes-AcrBF. The arene radical anion then fragments yielding an aryl radical. The resulting aryl radical abstracts a hydrogen atom from the amine cation radical, yielding the desired product as well as the corresponding iminium salt. Deuterium labeling studies confirmed the amine cation radical as the primary source of hydrogen atoms in this system (see SI – no deuterium incorporation in MeCN-d3).

The reductive detosylation of amines was identified as another possible transformation, which may be facilitated by Mes-Acr• (Figure 3). Typically, strong acid, dissolving metal (Li/Mg), or low-valent transition metal reductions are employed in detosylation reactions.[33-35] A variety of electronically diverse tosylated aniline derivatives were smoothly converted to the desired free anilines in moderate to excellent yield. Interestingly, substrates containing aryl halides were tolerated under the reaction conditions. As this reaction is conducted at a significantly lower concentration of substrate compared to the reductive dehalogenation method (0.1 M vs. 0.5 M), the observed lack of aryl halide reduction may be a function of concentration. Esters (43, 73), free carboxylic acids (44), ketones (48), and free alcohols (58) were tolerated under the reaction conditions, showcasing the high functional group tolerance of this method relative to methods relying on harsh dissolving metal conditions. Benzylic (52) and secondary alkyl amines (45, 53, 65–68) were efficient substrates for this transformation as well. Medicinally-relevant heterocycles, including pyridines (59), indoles (58), pyrroles (62), pyrrolidines (67), indazoles (63), benzimidazoles (64), and morpholines (65) were deprotected in good to excellent yields, with no reduction of the aromatic system observed in all cases. Of note is the ability of this method to chemoselectively and efficiently deprotect tosyl-amines over mesyl protected-amines, as shown by the reaction of substrate 51, yielding the desired detosylation product in 61% yield with no observed cleavage of the mesyl-amine. Additionally, the reaction performed well on 1.28 g scale, with substrate 64 undergoing the desired detosylation in 92% yield when conducted in a standard round bottom flask irradiated with LED lamps (see SI for experimental details).

Figure 3:

Scope of reductive detosylation catalyzed by Mes-Acr•.

1H NMR yields using DMSO or pyrazine as internal standard. †48 h reaction time. ‡0.5 M reaction concentration. Ts = p-toluenesulfonyl

In conclusion, an acridine radical generated in situ from single electron reduction of an acridinium derivative may act as potent single electron reductant upon excitation with 390 nm light. Spectroscopic and computational investigations indicate the formation of at least two distinct excited states, one of which may be characterized as a TICT state. The development of chemoselective dehalogenation and desulfonylation reactions utilizing Mes-Acr• compliment the well-known oxidative chemistry associated with acridinium salts and highlight the potential for the development of other types of reactions based on excitation of organic radicals.