Photoactivation Properties of Self-n-Doped Perylene Diimides: Concentration-dependent Radical Anion and Dianion Formation.

Perylene diimides (PDIs) have garnered attention as organic photocatalysts in recent years for their ability to drive challenging synthetic transformations, such as aryl halide reduction and olefin iodoperfluoroalkylation. Previous work in this area employs spectator pendant groups attached to the imide nitrogen positions of PDIs that are only added to impart solubility. In this work, we employ electron-rich ammonium iodide or ammonium hydroxide pendant groups capable of self-n-doping the PDI core to form radical anions (R •- ) and dianions (D ••2- ). We observe R •- formation is favored at low concentrations where aliphatic linkers are able to freely rotate, while D ••2- formation is favored at elevated concentrations likely due to Coulombic stabilization between adjacent chromophores in a similar manner to that of Kasha exciton stabilization. Cyclic voltammetric measurements are consistent with steric encumbrance increasing the Lewis basicity of anions through Coulombic destabilization. However, sterics also inhibit dianion formation by disrupting aggregation. Finally, femtosecond transient absorption measurements reveal that low wavelength excitation (400 nm) preferentially favors the excitation of R •- to the strongly reducing doublet excited state 2[R •- ]*. In contrast, higher wavelength excitation (520 nm) favors the formation of the singlet excited state 1[N]*. These findings highlight the importance of dopant architecture, counterion selection, excitation wavelength, and concentration on R •- and D ••2- formation, which has substantial implications for future photocatalytic applications. We anticipate these findings will enable more efficient systems based on self-n-doped PDIs.

Introduction

Perylene diimide (PDI) small organic molecules have become an increasingly relevant class of material in a broad range of applications beyond simple pigments, such as in biological systems, thermoelectrics, optoelectronics, sensing, photocatalysis, and synthetic photochemistry.[1−7] The successful utilization of PDIs in many of these applications requires an effective doping method to produce free carriers in the solid-state or radical anions in solution. For example, PDIs have garnered attention in synthetic photochemistry, such as aryl halide reduction and olefin iodoperfluoroalkylation, wherein the PDI is reduced by a sacrificial reductant under illumination to generate the radical anion R.[8−14]R then absorbs another photon to produce the doublet excited state 2[R]*, which is a strong reducing agent capable of driving challenging reduction reactions.[14,15] Thus, the efficient production of R under irradiation is generally regarded as an essential step in PDI-based photocatalytic applications.

The method of covalently tethering two functionalities into the same catalytic system is an interesting prospect.[16] Self-n-doping is a method of electronic doping in organic semiconductors that has recently risen to prominence.[17] In the solid state, covalently incorporating amine/ammonium moieties into small molecules is an effective strategy to mitigate performance issues that arise from inhomogeneous dopant distribution.[18] Much of this work has focused on thermal activation of dopants,[19,20] with comparably less work on photoactivation in the solid-state or in solution.[21,22] Though underexplored, the method of self-doping could provide unique advantages to solution-based applications. The incorporation of self-doping ammonium salts to the PDI scaffold enhances solubility, and the doping process solely relies on the spatial organization of the molecular components as opposed to a diffusion-mediated process that relies on the shuttling of extrinsic reductants. This may prove to be of importance since it is known that anion reductants tend to form charge—transfer and anion—π interactions with the conjugated backbones of rylene dyes.[23] Moreover, substituents added to the imide nitrogens are not critical to the performance of PDIs in photoelectrochemical transformations since these are not electrically connected to the PDI core where the anion radicals reside.[24] However, there is very little insight regarding the performance of self-dopants in solution, which is critical for future work. A detailed understanding of structure–function relationships in these systems is also critically lacking as acknowledged by the field[25] and serves as motivation to establish self-dopant design principles for new solution-based applications involving organic electronics.

Results and Discussion

Here, we report the photoactivation of a series of progressively more sterically encumbered self-n-doped PDIs with iodide and hydroxide counterions, whose structures are shown in Figure A. The general naming scheme for this family of PDIs used throughout this work denotes the degree of steric encumbrance in alphabetical order, followed by the carbon spacer length, and counterion selection (e.g., A2I, A2OH, etc). All samples are soluble in water, dimethylformamide (DMF), and dimethyl sulfoxide (DMSO), and are moderately soluble in methanol and ethanol. For this work, samples were prepared and handled in the dark in 10–60 μM DMF solutions (details available in the Supporting Information). Figure B shows the absorption spectra of A2I solutions kept in the dark. The S0 → S1 perylene core absorption band at 526 nm (0–0) is accompanied by Franck–Condon vibronic transitions 0–1 (490 nm), 0–2 (459 nm), and 0–3 (430 nm) that are characteristic of neutral PDI photoexcitation in solution (N → 1[N]*). The peak ratio 0–0/0–1 is often used to measure the degree of aggregation, with higher values (∼1.7) being indicative of monodispersity and lower values (∼1) indicating H-type aggregation.[26] In all the iodide samples measured, the peak ratios decrease from ∼1.7 to 1 with increasing concentration due to interactions between neighboring perylene chromophores. Irradiation of A2I with a 405 nm lamp for 90 min induces a photochromic change from fluorescent orange to a dull green (Figure S1) due to an intramolecular self-n-doping process that forms R•–, as shown in Figure C. R•– has six characteristic absorption peaks from ∼680 to 950 nm, with the 950 nm feature corresponding to the D0 → D1 transition accompanied by a complex vibronic structure.[27] All samples bearing iodide counterions exhibit comparable behavior to A2I upon irradiation (Figures S2–S5). Samples with hydroxide counterions are strongly doped to the degree that A2OH includes a mixture of radical anions and dianions before any light exposure (Figure D). The singlet–singlet transitions of D are blueshifted from the doublet–doublet transitions of R and have five major peaks from ∼530–720 nm. Following photoexcitation, a portion of the radicals is converted to dianions, and the perylene core achieves a doping density up to approximately 130%, as shown in Figure E. The absorption spectra for all the compounds with hydroxide counterions are shown in Figures S6–S9. An immediately discernible feature is that C2OH and D2OH do not form dianions at all, and their presence is minimal in B2OH in the dark. Thus, steric encumbrance is found to slow or inhibit dianion formation in these systems. We also note that photoactivation of PDIs is also observed to a lesser degree in DMSO, though not at all in water where R species are immediately quenched.

Figure 1

(A) Chemical structures of the self-doped PDIs investigated in this work. The naming scheme denotes the degree of steric encumbrance in alphabetical order, followed by the carbon chain length, and finally the counterion (e.g., A2I, A2OH, etc). (B) Absorption spectra of A2I measured in the dark and (C) after 90 min of irradiation with a 405 nm lamp. (D) Absorption spectra of A2OH measured in the dark and (E) after 90 min of irradiation with a 405 nm lamp. Sample concentrations are denoted in the legend.

Since the absorption characteristics of reduced PDIs are well defined, the relative components of N, R, and D in each spectrum can be integrated and quantified in terms of their percentage abundance (Figure ). However, spectral overlap between R and D requires fitting to evaluate abundances. The absorption spectra (450 nm < λ < 1000 nm) for each solution was fit, using linear regression, to a linear combination of R and D molar attenuation coefficient spectra, which have been reported previously in DMF.[27] Using Beer’s law, the molar concentration of R and D were then computed for each solution. For the iodide samples, the concentration of R varies from 65–95% following irradiation, as shown in Figure A. The formation of R is favored in solutions of lower concentration (10–30 μM) both before and after irradiation, as well as during the overall radical conversion process (Figure S10). This trend is in stark contrast to the formation of dianions in the samples containing hydroxide counterions, which preferentially form in solutions of higher concentration both before and after irradiation (Figure B). To understand this trend, we first discuss the photoinduced electron transfer mechanism in PDIs.[28−30] In Figure , ground state N is photoexcited to 1[N]*, and R then forms when 1[N]* is reductively quenched by the anion. The difference between this and previous reports is that, in the case of self-n-doping, the process is not diffusion-limited but instead mediated by chain rotation as the dopant folds back toward the imide acceptor moieties. Figure offers a graphical representation of this concentration dependence of radical anion and dianion formation. At low concentrations (Figure A), chain rotation is relatively uninhibited, and doping proceeds readily. At elevated concentrations (Figure B), chain rotation is slowed by the presence of adjacent PDI structures. However, dianion formation is favored under these conditions, most likely due to a Coulombic stabilization effect that arises from electronic coupling between adjacent chromophores similar to that of exciton stabilization in Coulombically coupled PDI chromophores.[31,32] This view is supported by the fact that dianions do not form in the presence of tethered bulky dopants, which inhibit intermolecular attractions between adjacent chromophores. However, the precise arrangements of the various components during dianion formation are yet to be determined.

Figure 2

(A) Percentage radical anion abundance in iodide samples of varying concentration after 90 min of irradiation with a 405 nm lamp. (B) Percentage dianion abundance in A2OH and B2OH both before and after irradiation with a 405 nm lamp.

Figure 3

Graphical depiction of the doping mechanism. First, the neutral perylene core N is photoexcited to 1[N]*. The anion then transfers an electron to 1[N]*, thereby reducing it to generate R•–.

Figure 4

Graphical representations of concentration-dependent doping (A) in the low and (B) high concentration regimes.

To probe the limit of self-n-doping in solution, we irradiated samples for 24 h. We observe a dramatic decrease in the absorption intensity of the samples with iodide counterions (Figures S2–S5). While the features of N broaden, R is almost completely quenched, and a broad feature centered at ∼700 nm persists, which we assign to a charge-transfer state between iodide counterions and the PDI core.[23] In the samples with hydroxide counterions, we observe dianion quenching, and the spectrum of R slightly broadens (Figures S6–S9). Thus, all samples are prone to photodegradation due to the strongly reducing character of 2[R]* and D. The samples with hydroxide counterions degrade more slowly. A comparison of photocatalytic activity between both sample types could provide insight into the disputed catalytic mechanism of 4′-bromoacetophenone reduction.[33] We also observe that particles begin to precipitate at these longer timescales (Figure S11), and Mie scattering off of these particles contributes to spectral broadening.

Self-doped PDIs exhibit electrochromic color changes in response to changing oxidation state via cyclic voltammetry (CV).[34] Cyclic voltammograms were produced for each compound by first measuring in the dark and then scanning each hour during photoirradiation for 24 h. Voltammograms of A2I and A2OH are presented in Figure A,B, respectively. As the carboximide core substituents are electron-poor, PDIs generally share two reversible reduction waves at ∼−0.9 and −1.1 V versus ferrocene/ferrocenium (Fc0/Fc+) attributed to R and D, respectively.[35] Incorporating two iodide counterions (Figure A) results in an I–/I2 redox process that occurs in two steps[36]

Figure 5

Cyclic voltammograms of (A) A2I and (B) A2OH in the dark and after various photoirradiation timepoints indicated in the legend. Each solution was measured at 10 μM in DMF and 100 mM tetrabutylammonium hexafluorophosphate (TBAPF6) as a supporting electrolyte with ferrocene as an external standard reference potential, since a Ag/Ag+ quasi-reference electrode was used. Scan rate: 100 mV/s.

These are at roughly twice the intensity of the oxidation waves for the perylene core due to the improved solubility of iodide versus hydroxide counterions. The irreversible reduction wave observed at −2.2 V versus Fc0/Fc+ is related to the photodoping process as it only appears and becomes more intense after doping (Figures S12–S14). This could be a reducible photodegradation product or the product of a complex reaction between triiodide, which forms as a product of photoactivated doping and A2I. The precise nature of this peak is unclear but may be related to the particles that begin to crash out of the solution over time. The first iodide oxidation also shifts and becomes more electronegative in sterically encumbered samples, C2I and D2I, which we attribute to the weaker ion-pairing between ammonium and iodide, making them better electron donors. The iodide counterions in C2I and D2I thus act as stronger Lewis bases, but steric encumbrance also prevents aggregation, which is important for a high degree of doping. The voltammograms for the samples with hydroxide counterions are more challenging to interpret (Figure B and S15–S17). As shown in the voltammograms, two quasi-reversible oxidation waves are observed, which we attribute to

The two reversible reduction waves at ∼−1.5 and −2 V versus Fc0/Fc+ are more challenging to assign, though they are related to electronic doping, similar to the iodide counterion samples. For example, the peak currents of the −2 V versus Fc0/Fc+ feature are relatively larger in A2OH and B2OH, which form dianions, and virtually non-existent in C2OH and D2OH, which do not. These peaks could arise from a photodegradation product, though we contend this is unlikely given that the observed peaks are generally more pronounced in the dark than after irradiation. Instead, we attribute these peaks to the reduction of H2O2 and O2.

We also investigated the absorption and luminescence behavior of our PDI series using both femtosecond transient absorption (TA) and time-resolved photoluminescence (TRPL) spectroscopies. TA was used to probe whether the photophysical properties of the perylene core were altered by the presence of the dopant moieties. Additionally, both R and D are non-luminescent due to non-radiative relaxation, and thus total fluorescence and TRPL were used to probe N and 1[N]*, respectively. TA of PDIs generally involve an S0 → S1 ground-state photobleach of N, often accompanied by stimulated emission, concomitant with the growth of the excited singlet state 1[N]* photoinduced absorption transient S1 → S (n > 1), and 1[N]* quickly relaxes (within a few ns) back to the ground state N. The absorption of R and 1[N]* appear in the same spectral region due to the fact that both excitations occur at similar energies, both involve the LUMO of N, and the energy of the LUMO is not dramatically altered when populated by an electron.[27] TA maps and spectra of iodide samples excited at 400 nm are shown in Figures S18–S21. Photobleaching occurs at 490–540 nm accompanied by stimulated emission at 560 nm and a photoinduced absorption feature centered at 700 nm, which decays after 7 ns. The overall intensity of these features progressively decreases after irradiation, and the total fluorescence brightness also decreases (Table S1) as N is converted to R. The fluorescence lifetimes also vary slightly after irradiation due to aggregate formation among the remaining neutral PDIs, which can be observed in the 0–0/0–1 peak ratios of the N absorption bands (Figures S2–S5). Contrastingly, the TA maps of A2OH depend on the excitation wavelength. When the sample is excited at 520 nm near the ground-state absorption energy, N photobleaches and 1[N]* rises in the same manner as iodide samples, as shown in Figure A. The overall TA intensity is quite low, which is unsurprising given the low concentration of N in A2OH, as evidenced by the total measurable fluorescence (Table S2). Interestingly, when the same sample is excited at 400 nm, the features are inverted, as shown in Figure B. At this excitation wavelength, R is preferentially excited to 2[R]*, and N is excited to 1[N]*. Electronic diagrams for each of these pathways are illustrated in Figure C,D. These findings demonstrate that 2[R]* can be preferentially selected at lower excitation wavelengths, which is an important finding for future photocatalytic work.

Figure 6

(A) Femtosecond transient absorption maps and spectra for A2OH pumped at 520 nm and (B) 400 nm and were not exposed to any other light sources. (C) Energy level diagrams depicting the two pathways activated by the different excitation wavelengths. Pumping at 520 nm (left) bleaches the S0 → S1 transition of N accompanied by stimulated emission and a S1 → S absorption transient of 1[N]* centered at 700 nm. In contrast, pumping at 400 nm (right) bleaches the D0 → D1 transition of R and is accompanied by the transient absorption of the ground-state excitation.

Conclusions

In summary, we have presented a series of soluble self-n-doped perylene diimides with increasingly sterically encumbered dopant moieties for use in photocatalytic applications. Anion doping proceeds readily when samples dissolved in DMF are excited with a 405 nm lamp, achieving anywhere from ∼65–130% doping depending on counterion selection. Anion doping generally follows Lewis basicity strength, with hydroxide being the stronger dopant. We observed radical anion formation is favored at low concentrations where associated ion pairs move more freely, but dianion formation is favored at high concentrations due to the stabilizing interactions of adjacent chromophores. Additionally, while steric encumbrance was found to increase the Lewis base strength of the anions by weakening counterion interactions, sterics also disrupt aggregation, which is essential for dianion formation. Finally, we found irradiation wavelength drives R excitation; lower wavelengths far from the ground-state absorption band promote the formation of 2[R]*, which helps to explain why these excitation wavelengths are able to drive synthetic transformations. Future work to establish the catalytic ability of self-n-doped PDIs is forthcoming and more groundwork to promote the synthetic modulation of organic materials to maximize their doping efficiency is needed. Specifically, the use of divalent anions may drive forward reactions more efficiently than monovalent anions due to their improved electron-donating ability.[37]