Long-lived charge separated states in spiro-compact rhodamine–naphthalenediimide D–A–D systems: synthesis and time-resolved optical and electron paramagnetic resonance spectroscopic studies

Abstract

A series of compact donor–acceptor triads featuring rhodamine (Rho) and naphthalenediimide (NDI) were synthesized using either direct conjugation or a flexible ethylene linker to investigate the charge separation of the compounds. The design enables access to long-lived triplet charge-separated states (CS) via intersystem crossing, offering an alternative to conventional singlet CS pathways. Steady-state absorption spectra showed no significant ground-state interactions between Rho and NDI units. However, fluorescence from the NDI moiety was markedly quenched in the triads compared to the reference compound, suggesting photoinduced electron transfer upon photoexcitation. Nanosecond transient absorption spectroscopy revealed that the ³CS state is formed upon photoexcitation of Rho–NDI–Rho in both non-polar hexane (HEX) and polar acetonitrile (ACN) solvents, with lifetimes of 0.11 µs (HEX) and 0.8 µs (ACN). The ³CS state lifetime is similiar in the case of the directly linked dyad and triads (Rho–NDI and Rho–NDI–Rho) (0.11 µs (HEX)) compared to the corresponding analogues possessing a flexible ethylene linker (Rho–CH–NDI and Rho–CH–NDI–CH–Rho). Specifically, Rho–CH–NDI exhibited CS state lifetimes of 0.83 µs in HEX and 0.05 µs in ACN, while Rho–CH–NDI–CH–Rho showed CS state lifetimes of 0.5 µs in HEX and 0.2 µs in ACN. Femtosecond transient absorption spectroscopy indicated that photoinduced charge separation occurred on an ultrafast timescale in all systems, mostly below the time resolution of the measurement (about 200 fs). Additionally, electron paramagnetic resonance (EPR) spectra of the Rho–CH–NDI system showed weak electronic coupling between the Rho and NDI units. Time-resolved EPR (TREPR) spectra confirmed the formation of ³NDI*, with zero-field splitting (ZFS) parameters |D| = 2150 MHz and |E| = 0 MHz. Furthermore, narrow signals attributed to spin-correlated radical pairs (SCRPs) were observed in the triads, with the electron exchange energy (2J) determined as −100 MHz.

---

Introduction

Achieving long-lived charge separated (CS) states is essential for advancing fundamental photochemistry studies and applications in artificial photosynthesis,^1–3 photovoltaics^4–7 and photocatalysis.^8,9 A major challenge in this field is to extend the CS state lifetimes.^10–12 This is typically described as a scenario where charge recombination (CR) happens thousands of times slower than the CS process.^2,13 The relation between the kinetics and thermodynamics of electron transfer (ET) in organic electron donor–acceptor dyads or triads can be described using the Marcus equation.where ℏ is the reduced Planck's constant, V is the electronic coupling matrix element between the initial and final states in which electron transfer occurs, λ is the reorganization energy, k_B is the Boltzmann constant, and T is the absolute temperature. Several strategies for extending the lifetime of CS states have been developed based on the Marcus equation. Reducing the electronic coupling matrix elements V is advantageous for achieving longer CS state lifetimes. For example, introducing long, saturated linkers between donor and acceptor moieties can successfully reduce electronic coupling between the initial and final states.^14–16 However, this approach has limitations, including the difficulty of synthesis.^15,17–19 Another common strategy for achieving a long CS state lifetime is using the Marcus inverted region effect, where the absolute values is ΔG_CS > λ, which slows down the CR process.^10,15,20 Employing electron donor–acceptor dyad structures or triad with small reorganization energy (λ) is advantageous for this purpose, as the CR will easily occur in the Marcus inverted region.²¹

Recently, the electron spin control method has emerged as a potential strategy for enhancing the CS state lifetime.^12,22–26 Electron spin selectivity dictates the feasibility of CS and CR pathways. Specifically, ³CS → S₀ transitions are electron spin forbidden, whereas ¹CS → S₀ transitions are electronically spin allowed.²³ The relaxation dynamics of electronic states are governed by electron spin conservation, resulting in a longer lifetime of the ³CS state compared to the ¹CS state. Nevertheless, the CS process also exhibits spin selectivity, such that the formation of the ¹CS state is favoured over the triplet ³CS state when originating from a ¹LE precursor. To control the electron spin for forming and prolonging the CS state lifetime, a ³LE state should be generated as the precursor of CS, rather than the ¹LE state. A key challenge in utilizing this strategy lies in ensuring that the electron exchange integral between the unpaired electrons of the radical ion pairs is sufficiently large. If the electron exchange interaction is too weak, it can lead to the formation of radical ion pairs with minimal interaction, such as spin-correlated radical pairs (SCRPs).^15,27,28 In such cases, the electron spin control effect cannot be fully harnessed, and the CS state will have a short lifetime.^15,17,23 Additionally, selecting electron donors or acceptors with efficient intersystem crossing (ISC) is often essential, imposing another limitation to this approach. While some compact electron donor–acceptor dyads have been explored, there remains significant potential for further development in this area.²³

Symmetry-breaking charge separation (SBCS) constitutes a distinct class of electron transfer processes.^29,30 Upon photoexcitation in multichromophoric molecular architectures, SBCS can occur in the excited state, resulting in spatial separation of the electron and hole across different chromophores. Such molecular configurations are of particular significance in the fields of photovoltaics and artificial photosynthesis, as they facilitate rapid charge separation to yield long-lived CS states, while concurrently suppressing back electron transfer or CR.³¹ In nonbiological molecular systems, SBCS between two identical chromophores remains relatively rare, largely due to the infrequent occurrence of a negative free energy change for charge separation (ΔG_CS).³² Nevertheless, an expanding range of molecular systems including anthracene,³³ perylene,³⁴ perylene monoimides and diimides,^31,35,36 dipyrrin derivatives,³⁷ pyrene,³⁸ and naphthalenediimides^39,40 have been reported to exhibit SBCS behaviour.

We recently reported orthogonal closed-form rhodamine (in its lactam form as an electron donor) linked to naphthalenediimide (NDI), perylene (Pery), and naphthalimide (NI) to construct spiro electron donor–acceptor dyads.^41–43 These dyads demonstrated a long-lived ³CS state via an electron spin control approach. However, compact electron donor–acceptor dyads exhibiting a long-lived ³CS state remain rare.⁴⁴

In order to address the above challenges, herein we synthesized a series of compact electron donor–acceptor triads, featuring rhodamine (Rho) as the electron donor and naphthalenediimide (NDI) as the electron acceptor (Scheme 1).

Scheme 1 (a) NH₂NH₂·H₂O, EtOH, reflux, 2 h, yield: 80%; (b) NDA, AcOH, N₂, reflux, 24 h, yield: 27%; (c) ethane-1,2-diamine, EtOH, reflux, overnight, yield: 80%; (d) NDA-1, similar step to (b), yield: 10%; (e) NDA, similar step to (b), yield: 30%; (f) 2-ethylhexylamine, KOH, H₃PO₄, 110 °C, overnight, yield: 35%.

In compounds Rho–CH–NDI and Rho–CH–NDI–CH–Rho, the donor and acceptor units are linked by a flexible ethylene bridge. A key advantage of these compounds is that the increased separation between the donor and acceptor reduces conformational fluctuations in solution. This structural stability is expected to facilitate the formation of a long-lived ³CS state. The photophysical properties of these triads were investigated using UV-vis absorption and fluorescence spectroscopy, femtosecond and nanosecond transient spectroscopy, magnetic spectroscopy, and theoretical computations.

Results and discussion

Molecular structure design and synthesis

The NDI core is one of the most extensively studied chromophores, widely utilized in various fields such as for molecular sensor development,⁴⁵ solar cells,⁴⁶ or artificial photosynthesis.^47–49 Similarly, rhodamine has also been extensively explored in different fields such as chemosensors, molecular devices, and biological applications, gaining significant attention due to its exceptional photophysical properties.^50–53 Recently, we reported that both the distance and the type of linkage between electron donor–acceptor dyads composed of rhodamine (Rho) and naphthalenediimide (NDI) units are critical for achieving a long-lived CS state.⁴³ This was accomplished using an electron spin control approach via the SOCT-ISC mechanism, leading to the formation of the ³LE precursor during the CS process.⁴³

In this work, we synthesized a series of compact donor–acceptor–donor (D–A–D) triads to investigate their ISC behaviours and charge separation/recombination (Scheme 1). Specifically, two Rho moieties were attached to an NDI chromophore using various linkages, incorporating different spacers on the Rho units to explore the influence of molecular geometry on the photophysical properties of the triads, particularly their ISC efficiency and formation of the CS state. Rhodamine was chosen as the electron donor (E_OX = +0.51 V and +0.68 V, vs. Fc/Fc⁺),^41,43 while NDI served as the electron acceptor (E_RED = −1.02 V and −1.55 V, vs. Fc/Fc⁺).^14,54,55 Recently, we investigated CS in a compact Rho–NDI dyad, where the Rho and NDI units are connected by an N–N single bond, and observed a long-lived CS state (0.13 µs).⁴³ In this study, we extended this approach by linking two Rho units to the NDI chromophore through both a direct N–N single bond and a flexible ethylene linker. In Rho–NDI–Rho, the two Rho units are directly attached to the imide positions of the NDI core via an N–N single bond. We then considered the dyad Rho–CH–NDI, incorporating an ethylene flexible linker to increase the spatial separation between the Rho and NDI moieties. Additionally, in Rho–CH–NDI–CH–Rho, two Rho units are connected via an ethylene flexible linker to the NDI core. Both Rho–CH–NDI and Rho–CH–NDI–CH–Rho utilize ethylene linkers to enhance the separation between the donor and acceptor units. The ethylene linker was specifically chosen for its flexibility, to minimize electronic coupling between the units. These structural variations in distance, orientation, and electronic interactions between donor and acceptor units are expected to yield distinct photophysical properties of the compounds.

The triads were synthesized using standard derivatization chemistry of rhodamine and NDI chromophores (Scheme 1). The molecular structures were confirmed through characterization techniques including ¹H NMR, ¹³C NMR, and HR-MS methods. Finally, the structures of some of the triads were validated by single-crystal X-ray diffraction analysis.

Single crystals of Rho–NDI–Rho and Rho–CH–NDI–CH–Rho were successfully obtained using the slow diffusion of the vapor of N,N-dimethylformamide (DMF) into a solution of the compounds in acetonitrile (ACN). The molecular structures of these triads were confirmed through single-crystal X-ray diffraction analysis, as shown in Fig. 1 and detailed in Table S1.

Fig. 1 ORTEP view of the molecular structures of (a) Rho–NDI–Rho and (b) Rho–CH–NDI–CH–Rho determined by single-crystal X-ray diffraction. Hydrogen atoms are omitted for clarity. Thermal ellipsoids are set at 50% probability.

The crystal structure of Rho–NDI–Rho (Fig. 1a) belongs to the monoclinic system, with the space group P21/n1. The dihedral angle between the xanthene and NDI units is 82.7°, and the centroid-to-centroid distance between the xanthene and NDI units is 6.3 Å. Rho–CH–NDI–CH–Rho crystallizes in the triclinic system with the space group P1̄. The dihedral angle between the electron-donating xanthene unit, which is connected to the NDI units via a flexible ethylene linker, is 85.5°, and the centroid-to-centroid distance between these components is considerably larger, 8.6 Å (Fig. 1b). Compared to the directly linked Rho–NDI–Rho unit, the structure with a flexible ethylene linker Rho–CH–NDI–CH–Rho exhibits greater spatial separation and a more twisted orientation between the xanthene and NDI units, as reflected in the dihedral angle and inter-unit distance (Fig. 1b). Compared to our previously reported Rho–NDI dyad, where the dihedral angle between the xanthene and NDI units was 52.9° and the centroid-to-centroid distance was 5.8 Å,⁴³ the dihedral angle and distance in the current triads differ significantly. Moreover, when compared with NI–PTZ dyads incorporating flexible linkers (14.5–17.5 Å),⁵⁶ both Rho–NDI–Rho (6.3 Å) and Rho–CH–NDI–CH–Rho (8.6 Å) enforce substantially shorter centroid-to-centroid distances, underscoring the intrinsic rigidity of these architectures.

UV–vis absorption spectra

The UV–vis absorption spectra of the compounds were examined (Fig. 2a and Fig. S10). The NDI unit displays a distinct absorption band within the 300–400 nm range, characterized by a significant vibrational fine structure, which is consistent with previous reports.⁴³ For Rho–Ph, absorption bands centered at 326 and 343 nm were observed. The absorption spectra of the compounds Rho–NDI–Rho, Rho–CH–NDI and Rho–CH–NDI–CH–Rho closely resemble the sum of the individual NDI and Rho–Ph moieties. The introduction of an ethylene flexible linker in Rho–CH–NDI and Rho–CH–NDI–CH–Rho does not lead to any shifts in the absorption wavelength. Our results align with observations in previously reported 1,8-naphthalimide-linker-phenothiazine dyads NI–PTZ,⁵⁶ 1,8-naphthalimide-linker-phenothiazine dyads,⁵⁷ 1,8-naphthalimide-linker-phenothiazine dyads NI–L–PTZ⁵⁸ and bis(phenylethynyl)anthracene-phenothiazine BPEA–PT⁵⁹ dyads containing long, flexible polymethylene linkers, where no changes in absorption wavelengths were detected upon introducing flexible linkers. This similarity indicates that there is no significant electronic coupling between the donor and acceptor units in the ground state for Rho–NDI–Rho, Rho–CH–NDI, and Rho–CH–NDI–CH–Rho.

Fig. 2 (a) UV–vis absorption spectra of the compounds in n-hexane (HEX). (b) UV–vis absorption spectra of Rho–CH–NDI in methanol (MeOH) with an increasing amount of trifluoroacetic acid (TFA) added. (c) UV-vis absorption spectra of Rho–CH–NDI in MeOH with the addition of TFA (c = 3.0 × 10⁻⁴ M) or triethylamine (TEA) in MeOH (c = 1 × 10⁻⁵ M). (d) Kinetics of lactam-opened amide transformation of the Rho moiety of the triads, monitored at 557 nm upon addition of TFA (c = 5.0 × 10⁻⁴ M) in MeOH, c = 1.0 × 10⁻⁵ M, 20 °C.

The rhodamine moiety is well-known for its reversible structural transformation between the spirolactam and amide forms, characterized by ring-opening and ring-closing isomerization processes, which are induced under acidic and basic conditions.^60–63 We investigated this reversible behaviour in these triads and observed the formation of two distinct structures (Fig. 2b). For Rho–CH–NDI and the related triads, an absorption band at 556 nm appeared in the presence of trifluoroacetic acid (TFA) (Fig. 2b and Fig. S11), corresponding to the ring-opened amide form of the rhodamine derivative. Upon adding the triethylamine (TEA) base, the absorption band of the ring-opened rhodamine form decreased (Fig. 2c), confirming the reversibility of the process.⁴³ The apparent transformation rate constants (k) were determined for Rho–NDI (0.21 s⁻¹), Rho–NDI–Rho (0.18 s⁻¹), Rho–CH–NDI (0.75 s⁻¹), and Rho–CH–NDI–CH–Rho (1.52 s⁻¹), showing that fully rigid rhodamine derivatives exhibit faster closed-ring to open-ring interconversion kinetics. These derivatives demonstrate significantly faster kinetics than previously reported rhodamine derivatives such as RB–NI (1.52 s⁻¹),⁶²Rho–NDI (1.74 s⁻¹), Rho–Ph–NDI (4.6 s⁻¹) and Rho–PhMe–NDI (0.22 s⁻¹),⁴³ making them promising candidates for developing stimuli-responsive molecular systems.^53,64,65

Fluorescence emission spectra

The fluorescence spectra of the triads were analyzed in various solvents (Fig. 3). In hexane (HEX, ε = 1.88), NDI shows a fluorescence band centered at 430 nm (fluorescence quantum yield Φ = 0.4%) (Fig. 3a). The fluorescence of Rho–NDI–Rho, Rho–CH–NDI, and Rho–CH–NDI–CH–Rho is markedly quenched compared to the native NDI, with optically matched solutions ensuring reliable comparisons (Fig. 3a). In acetonitrile (ACN, ε = 37.5), the emission spectra of the triads display two overlapping bands: a vibrationally structured band peaking near 415 nm and a broader band with maxima beyond 450 nm, attributed to relaxation from the locally excited (LE) state and charge transfer (CT) states, respectively (Fig. 3b). This attribution is further verified by measuring the emission spectra in solvents of varying polarities. In the nonpolar solvent HEX, the red shifted CT band is relatively weak compared to the emission from the LE state. However, as solvent polarity increases, the intensity of the redshifted CT band grows, reaching its maximum in ACN (Fig. 3b). This result aligns well with a previous report on solvent polarity-dependent CT emission.^66–68

Fig. 3 Fluorescence emission spectra of the compounds in (a) HEX, (b) ACN. λ_ex = 320 nm, c = 1.0 × 10⁻⁵ M, 25 °C.

Furthermore, we compared the emission properties of the triad Rho–NDI–Rho with those of the Rho–NDI dyad.⁴³ The Rho–NDI–Rho triad exhibited distinct emission bands and intensities in ACN, as shown in Fig. 3b. These results support the occurrence of CT in polar solvents for the triad. In contrast, the previously reported Rho–NDI dyad in ACN predominantly displays a LE emission band, indicating a different excited-state behaviour compared to the Rho–NDI–Rho system. Notably, the Rho–CH–NDI–CH–Rho triad shows an emission profile comparable to that of Rho–NDI–Rho under similar conditions, further substantiating the involvement of CT states in these extended donor–acceptor systems. Additionally, we compared the emission properties of the current compact flexible linkers, Rho–CH–NDI and Rho–CH–NDI–CH–Rho, whose results align with previous findings on 1,8-naphthalimide-linker-phenothiazine dyads NI-PTZ,⁵⁶ 1,8-naphthahlimide-linker-phenothiazine dyads, and bis(phenylethynyl)anthracene-phenothiazine BPEA-PT.⁵⁷ Similar to previously reported systems containing long, flexible linkers, fluorescence quenching of BPEA arises from efficient photoinduced electron transfer and formation of a long-lived CS state.⁵⁹ Derivatives of 1,8-naphthalimide-linker-phenothiazine (NI-L-PTZ) dyads have been reported to exhibit red-shifted emission profiles in polar solvents.⁶⁹ In protic environments, the emission of 1,8-naphthalimide–linker–phenothiazine (NI-L-PTZ) dyads arises predominantly from aggregation and excited-state interactions.⁵⁸

To preliminarily assess the ISC efficiency, the singlet oxygen quantum yield (Φ_Δ) of the compounds was measured in various solvents (Table 1 and Table S2). The pristine NDI exhibited moderate singlet oxygen production ability (Φ_Δ = 14.7%) while Rho–NDI showed low singlet oxygen efficiency. The primary triads, such as Rho–NDI–Rho and Rho–CH–NDI–CH–Rho, exhibited no singlet oxygen production in both polar and non-polar solvents. This observation supports the occurrence of ISC, likely due to the ³CS state being stabilized below the ³LE state in both polar and non-polar media, thereby inhibiting ¹O₂ production.

Table 1 Photophysical properties of the compounds in hexane

Compounds	λ abs (ε^b)		λ F	Φ F	Φ Δ	τ F	K r	K nr	λ p	τ p	Φ cs
	Lactam form	Open form^c
a Maximal UV-vis absorption wavelength, c = 1.0 × 10^−5 M. b Molar absorption coefficient, in 10^4 M^−1 cm^−1. c In MeOH. d Maximal fluorescence emission wavelength in nanometers. e Fluorescence quantum yields, in %. f Singlet oxygen quantum yield with [Ru(bpy)3]^2+ used as the standard (ΦΔ = 57% in DCM). g Fluorescence lifetimes. h Radiative decay rate constant, in 10^7 S^−1. i Nonradiative decay rate constant, in 10^9 S^−1. j Maximal phosphorescence emission wavelength in MeTHF, in nm, at 77 K. k Phosphorescence lifetimes at 610 nm. l CS quantum yields, in %. m Not applicable. n Not observed. o Literature values.^43 p Not observed. q Literature values.^43 r Not observed. s Not observed. t Not applicable.
NDI	374 (2.2)^m	—^m	430	0.4	14.7	0.29/3.22^o	1.04^q	2.07^q	609^q	58.4^q	—^t
Rho–NDI	373 (2.0)	556 (1.3)	380	0.2	—^n	0.09/2.91^o	0.59^q	2.94^q	—^s	—^s	25
Rho–NDI–Rho	371 (1.5)	558 (1.17)	413	0.1	—^n	—^p	—^r	—^t	—^s	—^s	23
Rho–CH–NDI	372 (1.7)	554 (1.8)	397	0.3	0.6	—^p	—^r	—^t	—^s	—^s	48
Rho–CH–NDI–CH–Rho	370 (1.3)	554 (5.2)	398	0.1	—^n	—^p	—^r	—^t	—^s	—^s	37

---

Redox properties and UV–vis absorption of the radical anion and radical cation of compounds

To evaluate the Gibbs free energy changes of the photo-induced electron transfer and the energy of the CS states, cyclic voltammetry measurements were performed for the triads and the reference compounds (Fig. 4a). The reference compound Rho–Ph displayed two reversible oxidation waves at +0.51 V and +0.68 V (vs. Fc/Fc⁺), with no reduction wave observed within the potential range applied in the study. In contrast, NDI exhibited reversible reduction waves at −1.02 V and −1.55 V (vs. Fc/Fc⁺), without any oxidation wave. These results confirm that the NDI moiety functions as the electron acceptor, while the Rho moiety acts as the electron donor. For Rho–NDI, two reversible oxidation waves at +0.51 V and +0.68 V, along with reversible reduction waves at −1.07 V and −1.54 V, are observed, consistent with previously reported values.⁴³ A reduction wave at −1.05 V was observed for Rho–NDI–Rho, differing from that of Rho–NDI (−1.07 V and −1.54 V vs. Fc/Fc⁺), which can be attributed to the additional Rho unit in the triad. Rho–NDI–Rho showed a single reversible oxidation wave at +0.51 V.

Fig. 4 (a) Cyclic voltammograms of the compounds in deaerated dichloromethane (DCM). Ferrocene (Fc/Fc⁺) was used as the internal reference (set as 0 V in the cyclic voltammograms). Conditions: in deaerated DCM solvent containing 0.010 M Bu₄N[PF₆] as the supporting electrolyte and Ag/AgNO₃ as the reference electrode. The scan rate is 100 mV s⁻¹ and c = 1.0 × 10⁻³ M, 25 °C. (b) UV-vis absorption spectra of Rho–CH–NDI reduced with tetrabutylammonium fluoride (TBAF) to generate NDI⁻˙ in deaerated DMF, c = 3.0 × 10⁻⁵ M, 25 °C.

In contrast, the reference compounds Rho–NDI and the Rho–NDI–Rho triad without a flexible ethylene linker showed variations when compared to Rho–CH–NDI and Rho–CH–NDI–CH–Rho. The difference in reduction waves at −1.12 V and −1.11 V for Rho–CH–NDI and Rho–CH–NDI–CH–Rho, respectively, is due to the presence of the flexible ethylene linker. Similarly, Rho–CH–NDI exhibited a single reversible oxidation wave at +0.51 V, while Rho–CH–NDI–CH–Rho showed one reversible oxidation wave at +0.53 V. We observed that the addition of the Rho part and a flexible linker also influenced the electrochemical behaviour. The redox potentials are reported in Table 2.

Table 2 Electrochemical redox potentials, Gibbs free energy changes of the charge separation (ΔG_CS) and the energy of the charge separated state (E_CSS) of the compounds in different solvents^a

	E Ox (V)	E Red (V)	ΔG^0CS (eV)/ECSS (eV)
			Hex	TOL	DCM	ACN
a Cyclic voltammetry in N2-saturated solvents containing 0.10 M Bu4N[PF6]; Pt electrode as the counter electrode, glassy carbon as the electrode working electrode, ferrocene (Fc/Fc^+) as the internal reference (set as 0 V in the cyclic voltammograms), and Ag/AgNO3 couple as the reference electrode in DCM; E00 = 3.28 eV for Rho–NDI, Rho–NDI–Rho, Rho–CH–NDI and Rho–CH–NDI–CH–Rho. E00 is the approximated energy with the cross point of UV–vis absorption and fluorescence emission of NDI after normalization in Hex. b Not observed. c Not applicable.
NDI	—^b	−1.1, −1.55	—^c	—^c	—^c	—^c
Rho–Ph	+0.51, +0.68		—^c	—^c	—^c	—^c
Rho–NDI	+0.51, +0.68	−1.07, −1.54	−1.59/1.68	−1.72/1.55	−2.08/1.19	−2.19/1.08
Rho–NDI–Rho	+0.51	−1.05	−1.38/1.89	−1.55/1.72	−2.02/1.25	−2.15/1.12
Rho–CH–NDI	+0.51	−1.12	−1.35/1.92	−1.48/1.79	−1.98/1.40	−1.98/1.29
Rho–CH–NDI–CH–Rho	+0.53	−1.1	−1.37/1.90	−1.50/1.77	−1.98/1.39	−1.98/1.29

---

The Gibbs free energy of the photo-induced electron transfer (ΔG⁰_CS) in the triads was calculated using the Rehm–Weller equation (equation 2),²⁰ showing that in all cases the electron transfer is thermodynamically allowed (Table 2). Nanosecond transient absorption spectroscopy further corroborates this observation (see the later section).

Furthermore, we calculated the CS state energy (E_CSS) using eqn (5). The energy of the ³NDI state was estimated to be approximately 2.04 eV.^19,43,70 These results demonstrate that the CS state energy of Rho–NDI is less than 2.0 eV in HEX. From this observation, we conclude that the low-lying triplet state is a CS state rather than a ³NDI* state, which aligns well with previous findings.⁴³ For Rho–NDI–Rho, the energy of the CS state in HEX is 1.89 eV. For Rho–NDI–Rho, the calculated CS energy value differs slightly from that of the Rho–NDI (1.66 eV in HEX) dyad due to the presence of an additional Rho unit.⁴³ To investigate the impact of a flexible ethylene linker on the CS state energy, we examined Rho–CH–NDI and the Rho–CH–NDI–CH–Rho triad. The results indicate that the CS state energy in HEX is 1.92 eV for Rho–CH–NDI and 1.90 eV for Rho–CH–NDI–CH–Rho. Comparing these findings with systems without the flexible ethylene linker, such as Rho–NDI (1.66 eV) and Rho–NDI–Rho (1.89 eV) in HEX, suggests that the slight variations in energy arise due to the presence of the flexible linker. For all triads, the results show that the CS state energy decreases in polar solvents compared to non-polar solvents.

ΔG⁰_CS = e[E_OX − E_RED] − E₀₀ + ΔG_S (2)

ΔG⁰_CR = −(ΔG⁰_CS + E₀₀) (4)

E_CSS = e(E_OX − E_RED) + ΔG_S (5)

In order to support the assignment of the CS state of the triads in the transient absorption spectra, we also measured the UV-vis absorption spectra of the radical anion of the triads using a chemical reduction method employing tetrabutylammonium fluoride (TBAF) as an external electron donor. Upon the addition of TBAF, Rho–NDI exhibited a reduction in its absorption band at 360 nm, while new absorption bands emerged at 471 nm, 606 nm, 708 nm, and 788 nm, corresponding to the NDI⁻˙ radical anion species (Fig. S14).^43,55 In the case of Rho–NDI–Rho, the gradual addition of TBAF led to a decrease in the 360 nm absorption band, accompanied by the appearance of new absorption bands centered at 468 nm, 607 nm, 734 nm, and 823 nm. Compared to previously reported absorption for the radical anions of the Rho–NDI dyad (471 nm, 606 nm, 708 nm, and 788 nm), the absorption spectrum of the Rho–NDI–Rho triad is notably different due to the presence of an additional Rho unit. Similarly, for Rho–CH–NDI, the absorption band at 355 nm decreased, while new absorption bands at 474 nm, 608 nm, 684 nm, and 758 nm appeared upon reduction (Fig. 4b). Rho–CH–NDI–CH–Rho also exhibited absorption bands characteristic of the NDI⁻˙ radical anion at 476 nm, 610 nm, 685 nm, and 765 nm (Fig. S14). These findings highlight significant differences compared to the Rho–NDI dyad and Rho–NDI–Rho triad lacking a flexible linker. Furthermore, we evaluated the absorption of the radical anion of the triads using spectroelectrochemical analysis and the results are well matched with those described above (Fig. S15). We investigated the absorption characteristics of the radical cation of triads using spectroelectrochemical analysis. The radical cation absorption spectra of Rho–NDI, Rho–NDI–Rho, Rho–CH–NDI, and Rho–CH–NDI–CH–Rho all exhibited a distinct and sharp absorption band centered at 565 nm (Fig. S16).

Femtosecond transient absorption (fs-TA) spectroscopy

The photoinduced dynamics of both the triads and dyads have been investigated using femtosecond transient absorption spectroscopy (fs-TA). To examine the influence of solvent polarity on CS and ISC processes, experiments were carried out in two solvents, n-hexane and acetonitrile. Additionally, the fs-TA spectra of reference compound NDI and the previously analyzed Rho–NDI dyad were recorded.⁴³

Firstly, the fs-TA spectra of the reference compound NDI were measured (Fig. S17). Upon photoexcitation, strong positive excited-state absorption (ESA) bands were observed at 497 nm and 580 nm, corresponding to the T₁ and S₁ states, respectively. A rapid ISC process, facilitated by spin–orbit coupling (SOC-ISC), was identified, with the lifetime determined to be tens of nanoseconds. These results indicate the presence of S₁/T_n equilibrium and are consistent with findings reported in previous studies.

In the case of the Rho–NDI dyad, global analysis of the time resolved data reveals the occurrence of very fast charge separation upon photoexcitation both in ACN and HEX, as noted by the rise of an absorption band peaked at about 470 nm, which, based on the spectroelectrochemistry measurements, can be assigned to the NDI⁻˙ radical anion. Additional ESA bands are also noticed at about 610 nm and 730 nm, assigned to Rho⁺. The anion/cation bands rise in about 200 fs in HEX, while in ACN charge separation occurs even faster, below the time resolution of the measurements. The charge separated state recombines in about 185 ps in HEX (Fig. S18b). In ACN the absorption band attributed to the NDI⁻˙ anion is substituted by a different broader absorption band with a maximum at 470 nm and a small shoulder at 500 nm, which rises in 1.4 ps. On the same timescale we also notice the formation of a negative band at 560 nm, which, in agreement with previous findings, is attributed to the emission of the Rho moiety undergoing a ring opening reaction in this polar environment.⁷¹ The signal intensity strongly decreases in about 24 ps. On this time scale the Rho stimulated emission decays almost completely, leaving an extremely small residual signal with a long lifetime of about 14 ns (Fig. S18d).

We then moved to investigate the excited state behavior of the triad, starting from Rho–NDI–Rho. The Evolution Associated Difference Spectra (EADS), shown in Fig. 5b and c, evidence a different behavior in polar and non-polar solvents.

Fig. 5 Femtosecond transient absorption spectra of (a) Rho–NDI–Rho and (d) Rho–CH–NDI in HEX. The related evolution associated difference spectra (EADS) obtained from global analysis with a sequential model in (b) HEX and (c) ACN for Rho–NDI–Rho and (e) HEX and (f) ACN for Rho–CH–NDI.

In HEX the behavior of the Rho–NDI–Rho triad is substantially similar to that observed for the Rho–NDI dyad. Photoinduced electron transfer producing the charge separated state Rho⁺-NDI⁻ occurs on an ultrafast timescale, below the time resolution of the measurement. The CS state undergoes a solvent induced relaxation process in about 2 ps and partially recombines in about 53 ps, as shown by the substantial intensity decrease of the transient signal observed within this time scale. A residual signal is still observed on the nanosecond timescale, possibly attributable to a triplet state localized on the NDI molecule, with the precursor of the CS triplet state being observed on a longer timescale through ns-TA spectroscopy. In ACN a different process occurs after the ultrafast formation of the charge separated Rho⁺-NDI⁻ state. Indeed, as already observed for the Rho–NDI dyad, the lactone form of the rhodamine undergoes a fast ring opening to form the zwitterionic form of the molecule, as shown by the development of the negative band peaking at 560 nm within 1.9 ps, attributed to the stimulated emission of the Rho open form. The transient signal recovers almost completely in 24 ns (Fig. 5c). The very small residual signal has a lifetime of a few nanoseconds.

Measurements were then repeated on the second dyad, Rho–CH–NDI, where the donor and acceptor are connected through an ethylene bridge. The transient data have been fitted using global analysis, obtaining the EADS reported in Fig. 5e and f.

As noticed in Fig. 5d, the different bridge clearly influences the photodynamics of the Rho–CH–NDI dyad, although ultrafast charge separation is still occurring in both solvents. In HEX the positive band peaking at 470 nm and assigned to the NDI⁻˙ anion is visible at a very short timescale, inferring that charge separation occurs on a timescale comparable with the time resolution of the measurement. The charge separated state undergoes a relaxation process within 4.8 ps. On the following 60 ps timescale the transient spectrum evolves significantly. Looking at the evolution between the second and third EADS (Fig. 5e) and comparing the spectral shape of the final EADS with that of the triplet state of the isolated NDI molecule, it is concluded that charge recombination and triplet state formation occur within about 60 ps, with the formation of a triplet state localized on the NDI molecule. Such a state is the precursor of a ³CT state observed on a longer timescale through nanosecond transient absorption measurements (see the ns-TA section). The formation of the NDI localized triplet state is also observed in ACN. In this polar solvent, however, the initially formed charge separated state recombines on a shorter 4.2 ps timescale (Fig. 5f). Upon recombination the CS state populates an intermediate state presenting an ESA band extending between 470 and 525 nm (the red line in Fig. 5f), whose exact nature is not clear, which successively evolves towards the NDI localized triplet state within ca. 70 ps.

The behavior of the Rho–CH–NDI–CH–Rho triad, whose EADS are reported in Fig. S20, is qualitatively similar to that of the other triad. Charge separation occurs on an ultrafast timescale both in HEX and ACN, and the interconversion between the spiro and zwitterionic forms of rhodamine is observed in ACN also for this compound. The main difference between the photophysics of the two triads is that a slightly slower evolution is observed in the case of Rho–CH–NDI–CH–Rho if compared to Rho–NDI–Rho, attributed to the presence of the longer linker and the increased donor–acceptor distance and that partial ring opening is observed in this case also in HEX.

Nanosecond transient absorption (ns-TA) spectroscopy

We used ns-TA spectroscopy to study the long-lived species formed in the triads after photoexcitation. For the reference compound NDI, we observed a strong absorption band at 470 nm, corresponding to the ³NDI* state,^19,43 upon excitation at 355 nm in deaerated HEX (Fig. S22). Comparing these results with chemical reduction, we conclude that the low-lying triplet state of the reference compound NDI is the ³NDI* state, with a measured lifetime of 18.7 µs. This is in good agreement with the previously reported literature, where the lifetime of the ³NDI* state was determined to be 23.6 µs.^19,43

For the triad Rho–NDI–Rho, positive absorption bands were observed at 470 nm and 600 nm upon pulsed nanosecond laser excitation corresponding to the radical anion NDI⁻˙ and the radical cation Rho⁺˙, respectively,⁴³ indicating the formation of a CS state in HEX (Fig. 6a). The observed features closely resemble the absorption bands of NDI⁻˙ and Rho⁺˙, as corroborated by reduction and spectroelectrochemical studies (Fig. S15 and S16). Accordingly, this species is assigned to the triplet CS state. We propose that this represents a case in which the ³CS state is generated from the triplet ³LE state,⁴³via an electron spin control mechanism, thereby facilitating the formation of a long-lived ³CS state. The triplet state exhibits a biexponential decay with lifetimes of 0.11 µs (91%)/0.65 µs (9%) (Fig. 6b). To our knowledge, the observation of a long-lived CS state in compact electron donor–acceptor–donor triads is rare, and its lifetime is similar to the previously reported lifetimes of 0.13–0.63 µs in similar systems.^42,43,72 The short-lived component is assigned to the CS state, while the long-lived component corresponds to the intermolecular CR process. This interpretation is validated by recording ns-TA spectra in a viscous polar solvent, such as dimethyl silicone oil 500, where diffusion is restricted, inhibiting intermolecular electron transfer. In this solvent, the decay traces observed at 490 nm exhibit biexponential behaviour with lifetimes of 0.3 µs (96%)/3.0 µs (4%), respectively (Fig. S23). These findings confirm the occurrence of the intermolecular CR process.

Fig. 6 Nanosecond transient absorption spectra of Rho–NDI–Rho in deaerated solvents of (a) HEX, (c) TOL, and (e) ACN. The decay traces of Rho–NDI–Rho in deaerated solvents of (b) HEX, (d) TOL, and (f) ACN at 470 nm, λ_ex = 355 nm, c = 3.0 × 10⁻⁵ M, 25 °C.

Interestingly, the lifetime of the CS state of this Rho–NDI–Rho triad is similar to that of the Rho–NDI dyad, presenting a lifetime of 0.13 µs.⁴³ In deoxygenated TOL, similar transient absorption bands were observed, including a prominent band at 470 nm and a weaker shoulder at 600 nm, again attributed to NDI⁻˙ and Rho⁺˙, confirming the formation of the CS state in TOL (Fig. 6c). The triplet state ³CS of the lifetime in TOL was determined to be 0.07 µs (57%)/6.2 µs (33%) for Rho–NDI–Rho (Fig. 6d). In a polar solvent such as deoxygenated ACN, a positive absorption band was observed at 480 nm (Fig. 6e), with a lifetime of 0.8 µs (64%)/8.7 µs (36%) (Fig. 6f). Additionally, we evaluated the lifetime of the ³CS state in aerated solvents such as HEX, TOL, and ACN. Monoexponential decay of NDI⁻˙ in Rho–NDI–Rho was observed, with lifetimes shortened to 66.6 ns, 85.1 ns, and 69.8 ns, respectively (Fig. S27). It is well established that the ³CS state can be quenched by O₂.⁴³

Photoexcitation in multichromophoric molecules can induce SBCS when the CS state is more stable than the LE state. Previous studies have reported SBCS in dyads and triads, including a CS lifetime of 0.16 ps for perylenediimide linked with naphthalenediimide (PDI–NDI–PDI).⁷³ Recently, the CS state lifetime of perylenemonoimide derivatives was determined as 0.58 µs.⁷⁴ A CS state lifetime of 1.41 µs in acetone was recently observed for angular perylenediimide dimers.⁷⁵

To the best of our knowledge, imide-based triads directly linked by an N–N single-bond showing SBCS are rare. In this study, SBCS was detected in the Rho–NDI–Rho triad, demonstrating a CS state lifetime of 0.11 µs, which is notably shorter than those reported in previous studies.

To further confirm the formation of a CS state, an NDI⁻˙ quenching experiment was conducted by introducing N,N′-dimethyl-4,4′-bipyridinium (MV²⁺) in MeOH as an electron acceptor and recording the ns-TA spectra (Fig. 7). Upon adding MV²⁺ to the deoxygenated ACN solution of Rho–NDI–Rho, the absorption peak at 470 nm, corresponding to NDI⁻˙, decreased while new peaks appeared at 390 nm and 605 nm (Fig. 7a), attributed to the reduced electron acceptor MV˙⁺.⁷⁶ Our results agree with previous studies that examined the addition of MV²⁺ to TnNDI₂ and DPAF-Ptn-NDI, revealing the appearance of new peaks at 395 nm and 605 nm.^76,77 Additionally, the lifetimes measured in the presence of MV˙⁺ were 0.2 µs for NDI⁻˙ and 5.8 µs (32%)/12.4 µs (68%) for MV˙⁺, respectively. This finding further confirms the formation of the ³CS state in the Rho–NDI–Rho triad.

Fig. 7 (a) Nanosecond transient absorption spectra of quenching of the radical anion of Rho–NDI–Rho by MV²⁺ in ACN, c = 3.0 × 10⁻⁵ M (MV²⁺ as an electron acceptor, c = 6.0 × 10⁻⁴ M), (b) decay trace of the radical anion at 470 nm, and (c) evolution trace of the radical MV˙⁺ anion at 605 nm, λ_ex = 355 nm.

The ns-TA spectra of Rho–CH–NDI in HEX were also measured, revealing ESA bands at 470 nm and 600 nm, similar to those observed for the Rho–NDI–Rho triad (Fig. 8a). The lifetime of the CS state is 0.83 µs (85%)/4.45 µs (15%) (Fig. 8b), which is longer than a recently reported compact Rho–NDI dyad (0.13 µs in HEX).⁴³ This result can be rationalized by the relatively large distance between the Rho and NDI units in the Rho–CH–NDI dyad and the weak coupling between the donor and acceptor, which is expected to contribute to a long-lived CS state. This is particularly significant, as, to the best of our knowledge, the presence of a long-lived CS state in dyads with flexible linkers is rare. The observed lifetime significantly exceeds the previously reported 0.3 µs for 1,8-naphthalimide-linker-phenothiazine (NI–L–PTZ)⁵⁶ and 0.36 µs for bis(phenylethynyl)anthracene-phenothiazine (BPEA–PT).⁵⁹ Similar ns-TA spectral features were observed in deaerated TOL, where the CS state lifetimes were 0.13 µs (45%)/13.17 µs (55%) (Fig. 8d). In the polar solvent ACN, the transient absorption spectra remained comparable, with CS state lifetimes of 0.05 µs (56%) and 0.4 µs (44%) (Fig. 8f).

Fig. 8 Nanosecond transient absorption spectra of Rho–CH–NDI in deaerated solvents of (a) HEX, (c) TOL, and (e) ACN. The decay traces of Rho–CH–NDI in deaerated solvents of (b) HEX, (d) TOL, and (f) ACN, at 470 nm, λ_ex = 355 nm, c = 3.0 × 10⁻⁵ M, 25 °C.

To further investigate the long-lived component in the transient decay of Rho–CH–NDI in TOL, ns-TA spectra were recorded in deaerated dimethyl silicone oil 500, a highly viscous solvent with a dielectric constant (ε = 2.75) comparable to that of TOL (ε = 2.38). A biexponential decay profile of the NDI⁻˙ species in Rho–CH–NDI was observed, with the lifetime determined as 0.53 µs (86%)/14.8 µs (14%) (Fig. S23). This suggests that the long-lived species arises from a closely lying CS state and the triplet excited state of NDI (³NDI*), rather than from diffusion-controlled intermolecular charge recombination, consistent with the behaviour observed in the Rho–NDI–Rho triad. This interpretation is supported by the comparable energy levels of the CS state (1.92 eV, obtained from electrochemical measurements) and the ³NDI* state (2.04 eV). Additionally, the formation of the CS state in the Rho–CH–NDI dyad was confirmed through quenching experiments using MV²⁺ (Fig. S24). These findings are consistent with those for the Rho–NDI–Rho triad.

The ns-TA spectra of Rho–CH–NDI–CH–Rho were recorded (Fig. S25). Due to the flexible linker, the CS state energy is expected to be higher than that of Rho–NDI and the Rho–NDI–Rho triad. In HEX, the CS state exhibited lifetimes of 0.5 µs (92%) and 2.9 µs (8%) (Fig. S25a). In TOL, the corresponding lifetimes were 0.09 µs (86%) and 4.4 µs (14%) (Fig. S25c) while in ACN, a monoexponential CS state lifetime of 0.20 µs was observed (Fig. S25e). Remarkably, these CS state lifetimes are longer than those reported for other flexible-linker-based dyads, such as NI–O–PTZ (0.3 µs),⁵⁶NI–L–PTZ (0.8 µs in p-xylylene/CH₃CN),⁶⁹ phenothiazine-bridged pyromellitimide (0.1 µs),⁷⁸ and phenothiazine-linker Pt–naphthalene-monoimide systems (less than 1 ns).⁷⁹ This finding is notable, as the large Rho–CH–NDI–CH–Rho separation and weak coupling in the triad favour a long-lived CS state. To confirm CS state formation in Rho–CH–NDI–CH–Rho, we conducted experiments in ACN with MV²⁺ as the electron acceptor (Fig. S26), showing ESA peaks at 390 and 605 nm attributed to MV˙⁺, consistent with Fig. 7.

CW-EPR and time-resolved electron paramagnetic resonance (TREPR) spectroscopy

To investigate the spin density distribution in the radical anions of the compounds, we performed continuous-wave EPR (CW-EPR) experiments on the triads and their reference compounds (Fig. 9). EPR spectroscopy unambiguously confirmed the formation of the NDI⁻˙ radical anion, achieved through reduction in degassed DMF using a strong Lewis base, such as TBAF. The hydrogen atoms on the aromatic core of NDI exist in equivalent environments because NDI is symmetrically N,N-disubstituted. The EPR spectrum of NDI⁻˙ displayed four equivalent protons (Table 3) and two equivalent nitrogen atoms (Table 3), consistent with previously reported literature values (Fig. 9).⁴³ These spectral features were accurately reproduced through simulation using the EasySpin software.⁸⁰ The EPR spectrum of Rho–NDI revealed well-defined hyperfine splitting patterns, resulting from interactions with two pairs of equivalent nitrogen nuclei (Table 3) and protons (Table 3), consistent with previously reported values (Fig. 9).⁴³ For Rho–CH–NDI, the asymmetrical N,N-disubstitution of the NDI unit gives rise to two magnetically equivalent environments for the aromatic protons, as reflected in its EPR spectrum. Similar to Rho–NDI, Rho–CH–NDI exhibited hyperfine splitting associated with two pairs of equivalent nitrogen atoms (Table 3) and protons (Table 3) (Fig. 9). The slight differences observed between the two spectra can be attributed to the flexible ethylene linker present in Rho–CH–NDI, which likely reduces the electronic coupling between the Rho and NDI units. This diminished interaction affects the delocalization of the unpaired electron and alters the spin distribution within the molecule. Furthermore, minor variations in the g-values of both triads, relative to those of the NDI radical anion (NDI⁻˙), were observed and are summarized in Table 3. Additionally, DFT-calculated hyperfine coupling constants and experimental parameters are provided in Table S4. For Rho–NDI–Rho and Rho–CH–NDI–CH–Rho triads, exceeding 100 atoms, direct calculation of hyperfine coupling constants was not feasible; hence we used the simulated values from the Rho–NDI and Rho–CH–NDI dyads instead.

Fig. 9 CW-EPR spectra of NDI, Rho–NDI, Rho–NDI–Rho, Rho–CH–NDI and Rho–CH–NDI–CH–Rho generated NDI⁻˙ by chemical reduction upon the addition of TBAF in deaerated DMF. The black lines represent the experimental data, while the red lines correspond to the simulated spectra. c[sample] = 1.0 × 10⁻⁴ M, c[TBAF] = 0.05 M and 25 °C. Microwave frequency = 9.41 GHz, microwave power = 0.796 mW, and modulation amplitude = 1 G.

Table 3 Isotropic hyperfine splitting parameters for the EPR spectra of NDI⁻˙ for the compounds^a

		NDI	Rho–NDI–Rho	Rho–CH–NDI	Rho–CH–NDI–CH–Rho
a NDI^−˙ of the triads was triggered by TABF in deaerated DMF. b HFC constants. c The number of equivalent atoms; [sample] = 1.0 × 10^−4 M, [TBAF] = 0.05 M and 25 °C. Microwave frequency = 9.40 GHz, microwave power = 0.796 mW and modulation amplitude = 1 G.
g		2.004	2.0039	2.0038	2.004
α/G^b	^14N (n^c)	0.99 (2)	1.03 (1)/0.98 (1)	0.91 (1)/0.90 (1)	1.03 (1)/0.98 (1)
	^1H (n^c)	1.86 (4)	1.98 (2)/1.82 (2)	1.80 (2)/1.69 (2)	1.80 (2)/1.69 (2)
Linewidth/G	—	0.103	0.100	0.101	0.101

---

To elucidate the nature of the transient paramagnetic species and the associated electron spin dynamics in dyads and triads upon photoexcitation, we recorded pulsed laser time-resolved electron paramagnetic resonance (TREPR) spectra on the triads and reference compounds (Fig. 10). TREPR spectroscopy is an effective technique for identifying transient paramagnetic species, including triplet excited states and radical pairs formed upon photoexcitation.^81,82 While TREPR has been widely used to characterize these species, the direct investigation of the spin multiplicity of CS states, particularly triplet CS states, which are fundamentally distinct from SCRPs, has been comparatively rare. This study aims to address this gap by exploring the formation and spin properties of triplet CS states in the investigated systems.^22,43,72,83

Fig. 10 TREPR spectra of (a) NDI, (b) Rho–NDI, (c) Rho–NDI–Rho and (d) Rho–CH–NDI. The laser excitation wavelength is 355 nm, in TOL:2-MeTHF (3 : 1, v/v), c = 3.0 × 10⁻⁵ M at 80 K. The black lines represent the experimental data, while the red lines correspond to the simulated spectra using the EasySpin package. The simulated spectra b and c are sum of the two spectra as described in the text.

The reference compound NDI exhibited broad dominant signals (Fig. 10a), which are attributed to NDI localized triplets formed through common spin orbit coupling ISC (SOC-ISC), consistent with previous reports.^43,70,84,85 The observed electron spin polarization (ESP) phase pattern, (e, e, e, a, a, a), where e represents emission and a denotes enhanced absorption (Fig. 10a), is indeed characteristic of triplet states formed via the SOC-ISC mechanism.^70,85 The zero-field splitting (ZFS) parameters |D| and |E| are 2150 MHz and 0 MHz, respectively (Table 4). The population ratio of the three sublevels of the T₁ state in NDI is P_X : P_Y : P_Z = 0 : 1 : 0.

Table 4 Zero-field splitting parameters (|D| and |E|) and relative populations P_x, P_y, and P_z of the zero field spin states obtained from simulation of the triplet state TREPR spectra of the compounds^a

	|D| (MHz)	|E| (MHz)	P x	P y	P z
a Obtained from simulation of the triplet-state TREPR spectra of the indicated molecules in polar (TOL/2-MeTHF, 1/3, v/v) glass at 80 K.
NDI	2150	0	0	1	0
Rho–NDI	2050	0	0	1	0
Rho–NDI–Rho	2050	0	0	1	0
Rho–CH–NDI	2150	0	0	1	0
Rho–CH–NDI–CH–Rho	2150	0	0	1	0

---

For Rho–NDI, two spectra were observed: a narrow one and a broader one (Fig. 10b). The broad spectrum, also observed in the native NDI reference compound (Fig. 10b), is attributed to triplet states localized on the NDI moiety, formed via SOC-ISC. This assignment is consistent with the previously reported behaviour of NDI-based systems.⁴³ The mid-range narrow signal is assigned to the SCRP state, in contrast to previous findings. The ZFS parameters extracted for the triplet state localized on the NDI moiety of the Rho–NDI dyad is |D| = 2050 MHz and |E| = 0 MHz (Table 4). These values differ slightly from those previously reported for Rho–NDI (|D| = 2075 MHz and |E| = 34 MHz), which may be attributed to variations in molecular geometry, spin–spin interactions, or local environments within the different systems.⁴³

A detailed analysis of the TREPR spectra of the Rho–NDI–Rho triad (Fig. 10c) shows a notable difference compared to those of the NDI and the Rho–NDI dyad (Fig. 10b). The triad displays a distinctly narrower and more intense signal, whereas the spectrum corresponding to the localized triplet state of NDI remains at the noise level. Interestingly, a similar sharp central signal has been reported for the naphthalimide–phenyl–naphthalenediimide system, which also corresponds to a SCRP.⁸⁶ The phase pattern of this narrow signal is strongly time dependent, as clearly shown in Fig. 10b and c. The superposition of the narrow and triplet spectra persists up to 2000 ns; however, the triplet contribution in the TREPR spectrum of Rho–NDI–Rho is significantly less than that for Rho–NDI. Beyond this timescale, the triplet spectrum attains thermal equilibrium, while the narrow signal remains observable. This observed phase reversal is most likely attributed to repopulation of the charge-separated states, since two processes are involved in the population of the SCRP states: ¹NDI* → ¹CS → ³CS and ¹NDI* → ³NDI* → ³CS. A comparable phase reversal shift has also been reported for the naphthalene-1,8 : 4,5-bis(dicarboximide) dyad.⁸⁷ The TREPR spectrum of the Rho–NDI–Rho triad was simulated, yielding ZFS parameters |D| = 2050 MHz and |E| = 0 MHz. The |D| value is significantly larger than that of the previously reported NI-PTZ dyad (|D| = 900 MHz), which features a shorter bond linkage.⁴⁴ The spectra, supported by simulations, indicate that the population ratio of the three sublevels of the T₁ state in the Rho–NDI–Rho triad follows P_x : P_y : P_z = 0 : 1 : 0 (Table 4). For the Rho–NDI–Rho triad, the electron spin dipolar and exchange interactions are −100 MHz in comparison to the reported 128 MHz for the BDX–ANI–NDI system incorporating naphthalene-1,8-dicarboximide, both of which are assigned to SCRPs.⁸⁸ Moreover, when the TREPR spectra of the Rho–NDI–Rho triad were recorded at various delay times after laser excitation in a slightly more polar solvent mixture (DCM/2-MeTHF), a different polarization pattern was observed, and no triplet signal was detected (Fig. 11b). The spectrum, assigned to the radical-pair states, was simulated using EasySpin by considering the 2JS₁S₂ exchange interaction, dipole–dipole interaction and difference in the Zeeman interaction between two radicals. The simulation showed good agreement with the experimental data for comparable magnitudes of exchange and dipole interactions of order −100 MHz, consistent with a dipole–dipole interaction within a radical pair. The dipolar interaction is −100 MHz in comparison to the reported −128 MHz for the BDX–ANI–NDI system incorporating naphthalene-1,8-dicarboximide, both of which are assigned to SCRPs.⁸⁸ The spectral shape is highly dependent on the population of the four spin states. Initially, S and T₀ states are predominantly populated; however, with increasing delay times (900 ns and 2000 ns), population redistribution occurs, resulting in corresponding spectral changes. Similar to the behaviour observed for Rho–NDI–Rho in a less polar solvent (TOL/2-MeTHF), the contribution of the ¹NDI* → ³NDI* → ³CS process becomes more prominent over extended timescales.

Fig. 11 TREPR spectra of (a) Rho–NDI, (b) Rho–NDI–Rho and (c) Rho–CH–NDI. The laser excitation wavelength is 355 nm, DCM : 2-MeTHF (1 : 1, v/v), c = 3.0 × 10⁻⁵ M at 80 K. The black lines represent the experimental data, while the red lines correspond to the simulated spectra. The simulation details are given in the text.

The TREPR spectrum of the Rho–CH–NDI dyad (Fig. 10d) exhibits an unusual pattern compared to the reference compounds: the Rho–NDI dyad and the Rho–NDI–Rho triad (Fig. 10c). The flexible linker in Rho–CH–NDI increases the donor–acceptor separation, resulting in weaker electronic coupling relative to the directly connected systems, the Rho–NDI dyad and the Rho–NDI–Rho triad. No radical pair signal was detected for this Rho–CH–NDI dyad in TOL/2-MeTHF. Spectral simulations provided ZFS parameters |D| = 2150 MHz and |E| = 0 MHz (Table 4). The triplet state displayed an ESP pattern (e, e, e, a, a, a) (Fig. 10d). Weak central spectral features of Rho–CH–NDI were successfully reproduced using a SCRP model with weak electronic coupling. When we measured Rho–CH–NDI in a slightly more polar solvent mixture (DCM/2-MeTHF), a similar polarization pattern was observed and a weak central signal was detected (Fig. 11c). Simulation of this spectrum, assigned to a radical-pair state, yielded J and dipole–dipole interaction values of the same order of magnitude as those observed for the Rho–NDI–Rho triad.

We measured the TREPR spectrum of the Rho–CH–NDI–CH–Rho triad, which contains a flexible linker (Fig. S29). The spectral features exhibited a central weak narrow signal, similar to those observed for Rho–NDI and Rho–NDI–Rho (Fig. 10c). The incorporation of a flexible linker in this triad, analogous to the Rho–CH–NDI dyad, increases the spatial separation between the electron donor and acceptor, thereby diminishing the electronic coupling relative to the directly connected donor–acceptor systems Rho–NDI and Rho–NDI–Rho. Simulation of the TREPR spectrum yielded ZFS parameters |D| and |E| as 2150 MHz and 0 MHz, respectively (Table 4). The simulations indicate that the population ratio of the three sublevels of the T₁ state in the Rho–CH–NDI–CH–Rho triad is P_x : P_y : P_z = 0 : 1 : 0 (Table 4). This trend was also noted for 1,8 : 4,5-naphthalenediimide (NI) or pyromellitimide (PI) acceptors, which show ESP with |D| = 757 MHz and |E| = 4.7 MHz.¹⁴ For comparison with other systems, the bicyclo[2.2.2]octane–phenothiazine–anthraquinone dyad, the TREPR spectra revealed that the ion pair could be attributed to a SCRP.¹⁶ In the case of a µ-oxo-bridged porphyrin heterodimer, the TREPR spectra indicated a ³CS state, showing narrow width and polarization with D = −650 MHz and E = 0 MHz.⁸⁹ The TREPR spectrum of PTZ₃-B-AQ, containing a bicyclo [2.2.2] octane linker, demonstrated four narrow signals (e, a, e, and a) characteristic of a SCRP.⁹⁰ The TREPR spectra of the Rho–CH–NDI–CH–Rho triad in frozen solution are informative for evaluating the spin–spin interaction between the radical anion and cation in the triad. The simulated spectrum of the SCRP featuring an electron spin–spin coupling approximately 2J ∼ −100 MHz aligns with the weak narrow spectrum observed for the Rho–CH–NDI–CH–Rho triad. By comparison, phenothiazine-methyl viologen systems with alkyl linkers exhibit much stronger J values (5.6–2800 MHz),⁹¹ whereas MTA–Pt–MNDI triads display markedly weaker coupling (0.14 MHz).⁹² Likewise, in the perylene-3,4:9,10 bis(dicarboximide)–phenothiazine dyad with a phenylene spacer, the J is similarly weak (0.14 MHz).¹⁵ Moreover, when the TREPR spectrum of the Rho–CH–NDI–CH–Rho triad was measured in a slightly more polar solvent mixture (DCM/2-MeTHF), no TREPR signal was detected (i.e. no polarized radicals are obtained). In summary, the ³CS signals observed in the Rho–NDI dyad and the SCRP signals observed in the Rho–NDI–Rho triad are significantly more intense than those in the Rho–CH–NDI dyad and Rho–CH–NDI–CH–Rho triad. This difference is attributed to the presence of longer and more conformationally flexible linkers in the Rho–CH–NDI and Rho–CH–NDI–CH-Rho compounds, which likely diminish electron spin interactions between the NDI and Rho units. The TREPR spectral analyses presented herein unequivocally demonstrate that photoexcitation induces transition from the singlet to the triplet SCRP state in both the dyad and triad systems. This finding introduces a novel conceptual framework for elucidating the underlying spin dynamics in electron donor–acceptor–donor molecular assemblies, thereby advancing the mechanistic understanding of charge separation and spin-selective processes in such systems.

DFT calculations

We finally complemented our experimental investigation with DFT computations. Initially, we optimized the ground-state geometries of the compounds using DFT calculations (Fig. 12). For Rho–NDI–Rho, the dihedral angles between the two xanthene moieties and the NDI unit were calculated to be 87.6° and 87.9°. These values exhibit slight deviations compared to those determined by single-crystal X-ray diffraction (Fig. 1a), likely due to molecular packing effects in the crystal structure. For Rho–CH–NDI, the dihedral angle between the xanthene unit and the NDI was calculated as 93.5°. Interestingly, in Rho–CH–NDI–CH–Rho, the dihedral angles between the two xanthene moieties and the NDI were found to be 91.7° and 93.6°, respectively, which are larger compared to the other triads. These DFT-optimized values also slightly differ from those obtained via single-crystal X-ray diffraction (Fig. 1b), further reflecting the influence of molecular packing effects in the crystalline state. The center-to-center distances between the xanthene and NDI units in the triads Rho–NDI–Rho, Rho–CH–NDI, and Rho–CH–NDI–CH–Rho were measured as 6.3 Å, 6.8 Å, and 6.8 Å, respectively. In comparison, the previously reported dyad Rho–NDI exhibited a center-to-center distance of 6.0 Å and a dihedral angle of 63.6°.⁴³ The increased center-to-center distances and larger dihedral angles in these triads may result in distinct photophysical behaviors.

Fig. 12 Optimized ground state (S₀) geometries of (a) Rho–NDI–Rho, (b) Rho–CH–NDI, and (c) Rho–CH–NDI–CH–Rho in the gas phase. The selected dihedral angles are presented. Calculation was performed at the B3LYP/6-31G(d) level using Gaussian 16.

The highest occupied molecular orbital (HOMO) of the compounds is primarily localized on the rhodamine chromophore, while the lowest occupied molecular orbital (LUMO) is entirely confined to the NDI moiety (Fig. 13). Notably, there is negligible delocalization of the molecular orbitals (MOs) between the two moieties.

Fig. 13 Selected frontier molecular orbitals of Rho–CH–NDI, Rho–NDI–Rho, and Rho–CH–NDI–CH–Rho, based on the optimized ground state geometries. In the gas phase, calculation was performed at the CAM-B3LYP/6-31G(d) level using Gaussian 16.

The electron spin density of the T₁ state of the triads was investigated in the gas phase and in ACN (Fig. 14). For all the triads, the electron spin density in the T₁ state is delocalized across both the rhodamine and NDI moieties in ACN. These findings are consistent with the ns-TA spectral results, which indicate the presence of a CS state for all compounds (Fig. 6 and Fig. 8)

Fig. 14 Isosurfaces of spin density distribution at the optimized triplet state triads in the gas phase and in ACN. Calculation was performed at the CAM-B3LYP/6-31G(d) level using Gaussian 16.

Based on the results, simplified energy diagrams were constructed to illustrate the photophysical processes of the triads (Scheme 2). The singlet state energy levels of the triads were determined using spectroscopic data, while the triplet excited state energies were obtained through DFT calculations. Photoexcitation predominantly populates the singlet excited state of the triads, localized on the NDI moieties. Quite rapidly the ¹CS state is obtained through electron transfer (200 fs) and subsequently intersystem crossing (ISC) occurs (60–70 ps), finally populating the charge-separated (³CS) state. The ³CS state lifetimes of Rho–NDI–Rho were measured as 0.11 µs (91%)/0.65 µs (9%) in Hex, 0.07 µs (57%)/6.2 µs (33%) in TOL, and 0.08 µs (64%)/8.7 µs (36%) in ACN. For Rho–CH–NDI, the ³CS state had lifetimes of 0.83 µs (85%)/4.45 µs (15%) in Hex, 0.13 µs (45%)/13.17 µs (55%) in TOL, and 0.05 µs (56%)/0.4 µs (44%) in ACN. Similarly, for Rho–CH–NDI–CH–Rho, the photophysical process resembled that of Rho–CH–NDI, with ³CS state lifetimes of 0.5 µs (92%)/2.9 µs (8%) in Hex, 0.09 µs (86%)/4.4 µs (14%) in TOL, and 0.2 µs in ACN. The TREPR spectra of Rho–NDI–Rho, Rho–CH–NDI, and Rho–CH–NDI–CH–Rho demonstrate the formation of a long-lived CS state, facilitated by spin correlated radical pairs (SCRPs). In comparison to the Rho–NDI–Rho triad, the introduction of a flexible ethylene linker in the Rho–CH–NDI dyad and Rho–CH–NDI–CH–Rho triad reduces the electronic coupling between the donor and acceptor, thereby prolonging the CS state lifetime. Furthermore, the emissive nature of the ion pair suggests a singlet precursor for the CS state.

Scheme 2 Simplified Jablonski diagram representing the photophysical process involved in Rho–CH–NDI in different solvents. The ¹[D–A*] energy levels were estimated using the singlet excited state localized on the NDI unit, determined by the intersection of the normalized UV-vis absorption and fluorescence emission spectra. The ¹CS energy levels are obtained from the electrochemical data in Table 3. The energy levels of the triplet states were taken from TDDFT calculations at the B3LYP/6-31G(d) level using Gaussian 16. J represents the electron exchange energy, and the energy gap between the ¹CS and ³CS states corresponds to 2J. The numbers in the superscript designate the spin multiplicity.

Conclusions

We successfully synthesized three compact electron donor–acceptor dyads featuring rhodamine (Rho) as the donor and naphthalenediimide (NDI) as the acceptor. By varying the connection from either direct conjugation in the Rho–NDI–Rho triad or via a flexible ethylene linker in Rho–CH–NDI and Rho–CH–NDI–CH–Rho we systematically modulated the donor–acceptor distance and electronic coupling. This structural variation enabled comprehensive investigation of the underlying photophysical processes, including charge separation (CS), charge recombination (CR), and intersystem crossing (ISC). UV-vis absorption spectra indicated minimal electronic interaction between Rho and NDI in the ground state, whereas charge transfer (CT) emission bands were observed in the triads. Nanosecond transient absorption spectroscopy confirmed the formation of a ³CS state upon photoexcitation of Rho–NDI–Rho in both non-polar n-hexane (HEX) and polar acetonitrile (ACN) solvents, with lifetimes of 0.11 µs and 0.8 µs, respectively. For the Rho–NDI dyad, the ³CS state lifetimes were 0.13 µs in hexane and 0.03 µs in acetonitrile. The lifetime of the ³CS state for the Rho–NDI–Rho triad (0.11 µs) was comparable to that of the Rho–NDI dyad (0.13 µs) in HEX. Notably, introducing a flexible ethylene linker in the Rho–CH–NDI dyad led to a prolonged ³CS state lifetime of 0.83 µs in HEX and 0.05 µs in ACN, which is significantly longer than that of the Rho–NDI dyad (0.13 µs) and the Rho–NDI–Rho triad (0.11 µs) in HEX. Similarly, Rho–CH–NDI–CH–Rho exhibited ³CS state lifetimes of 0.50 µs and 0.20 µs in these solvents, respectively. Femtosecond transient absorption spectroscopy revealed ultrafast charge separation occurring within or below the instrument time resolution in all the triads, regardless of solvent polarity and quite fast ISC in some cases. Time-resolved EPR (TREPR) spectroscopy revealed broad spectral features characteristic of native NDI, corresponding to NDI-localized triplet states formed via SOC-ISC. In addition, narrow signals were observed, attributed to spin correlated radical pairs (SCRPs) in the triads, with an electron exchange energy (2J) of −100 MHz. Notably, the Rho–NDI–Rho triad exhibits a time-dependent phase reversal of the SCRP signal from e,a to a,e, indicating that the SCRP is initially generated from a singlet state. However, at long delay times (around 2000 ns at 80 k) the contribution of the triplet state born SCRP predominates. This observation strongly supports the role of spin dynamics in stabilizing long-lived charge-separated states through SCRP mechanisms. These findings contribute to the development of new electron donor–acceptor–electron donor triads designed to achieve long-lived CS states and provide insights into their photophysical behavior.

Author contributions

B. T. and A. A. S. contributed equally to this work. B. T.: formal analysis, data curation, validation, writing – original draft; Y. Pei: data curation, visualization, validation; A. A. S. and V. K. V.: measured and simulated the TREPR spectra and analyzed the data; J. Z.: supervision, visualization writing – original draft, writing and editing, project administration, funding acquisition; G. S., T. R. and M. D. D. measured and analyzed femtosecond transient absorption spectra.

Conflicts of interest

There are no conflicts to declare.

Data availability

All relevant data (the molecular structure characterization data, optical spectral, and time-resolved EPR spectral characterization, as well as theoretical computation details) are presented in the main text and supplementary information (SI). See DOI: https://doi.org/10.1039/d5cp03836g.

CCDC 2370986 and 2370987 contain the supplementary crystallographic data for this paper.^93a,b

Acknowledgements

J. Z. thanks the NSFC (22473021 and U2001222), the National Key Research and Development Program of China (No. 2023YFE0197600), the Research and Innovation Team Project of Dalian University of Technology (DUT2022TB10), the Fundamental Research Funds for the Central Universities (DUT22LAB610) and the State Key Laboratory of Fine Chemicals of Dalian University of Technology for financial support. A. A. S. and V. K. V. acknowledge financial support from the government assignment for the FRC Kazan Scientific Center of RAS.

Notes and references

1. H. Levanon and J. R. Norris, Chem. Rev., 1978, 78, 185–198.
2. J. W. Verhoeven, H. J. van Ramesdonk, M. M. Groeneveld, A. C. Benniston and A. Harriman, ChemPhysChem, 2005, 6, 2251–2260.
3. O. S. Wenger, Chem. Soc. Rev., 2011, 40, 3538–3550.
4. S. Fukuzumi, Pure Appl. Chem., 2007, 79, 981–991.
5. Y. Wang, S. Li, S. V. Kershaw, F. Hetsch, A. Y. Y. Tam, G. Shan, A. S. Susha, C.-C. Ko, V. Wing-Wah Yam, K. K. W. Lo and A. L. Rogach, ChemPhysChem, 2012, 13, 2589–2595.
6. S. Suzuki, Y. Matsumoto, M. Tsubamoto, R. Sugimura, M. Kozaki, K. Kimoto, M. Iwamura, K. Nozaki, N. Senju, C. Uragami, H. Hashimoto, Y. Muramatsu, A. Konno and K. Okada, Phys. Chem. Chem. Phys., 2013, 15, 8088–8094.
7. H. Song, H. Zhao, Y. Guo, A. M. Philip, Q. Guo, M. Hariharan and A. Xia, J. Phys. Chem. C, 2020, 124, 237–245.
8. L. Shi and W. Xia, Chem. Soc. Rev., 2012, 41, 7687–7697.
9. D. Ravelli, M. Fagnoni and A. Albini, Chem. Soc. Rev., 2013, 42, 97–113.
10. K. Ohkubo, H. Kotani, J. Shao, Z. Ou, K. M. Kadish, G. Li, R. K. Pandey, M. Fujitsuka, O. Ito, H. Imahori and S. Fukuzumi, Angew. Chem., Int. Ed., 2004, 43, 853–856.
11. F. D’Souza, R. Chitta, K. Ohkubo, M. Tasior, N. K. Subbaiyan, M. E. Zandler, M. K. Rogacki, D. T. Gryko and S. Fukuzumi, J. Am. Chem. Soc., 2008, 130, 14263–14272.
12. B. Geiß and C. Lambert, Chem. Commun., 2009, 1670–1672.
13. H. J. van Ramesdonk, B. H. Bakker, M. M. Groeneveld, J. W. Verhoeven, B. D. Allen, J. P. Rostron and A. Harriman, J. Phys. Chem. A, 2006, 110, 13145–13150.
14. G. P. Wiederrecht, W. A. Svec, M. R. Wasielewski, T. Galili and H. Levanon, J. Am. Chem. Soc., 2000, 122, 9715–9722.
15. Z. E. X. Dance, Q. Mi, D. W. McCamant, M. J. Ahrens, M. A. Ratner and M. R. Wasielewski, J. Phys. Chem. B, 2006, 110, 25163–25173.
16. A. Karimata, H. Kawauchi, S. Suzuki, M. Kozaki, N. Ikeda, K. Keyaki, K. Nozaki, K. Akiyama and K. Okada, Chem. Lett., 2013, 42, 794–796.
17. E. A. Weiss, M. A. Ratner and M. R. Wasielewski, J. Phys. Chem. A, 2003, 107, 3639–3647.
18. E. A. Weiss, M. J. Ahrens, L. E. Sinks, A. V. Gusev, M. A. Ratner and M. R. Wasielewski, J. Am. Chem. Soc., 2004, 126, 5577–5584.
19. M. T. Colvin, A. B. Ricks, A. M. Scott, D. T. Co and M. R. Wasielewski, J. Phys. Chem. A, 2012, 116, 1923–1930.
20. D. I. Schuster, P. Cheng, P. D. Jarowski, D. M. Guldi, C. Luo, L. Echegoyen, S. Pyo, A. R. Holzwarth, S. E. Braslavsky, R. M. Williams and G. Klihm, J. Am. Chem. Soc., 2004, 126, 7257–7270.
21. D. M. Guldi, Chem. Commun., 2000, 321–327.
22. S. Fukuzumi, H. Kotani, K. Ohkubo, S. Ogo, N. V. Tkachenko and H. Lemmetyinen, J. Am. Chem. Soc., 2004, 126, 1600–1601.
23. J. W. Verhoeven, J. Photochem. Photobiol., C, 2006, 7, 40–60.
24. M. Murakami, K. Ohkubo, T. Nanjo, K. Souma, N. Suzuki and S. Fukuzumi, ChemPhysChem, 2010, 11, 2594–2605.
25. J. Hankache and O. S. Wenger, Chem. Commun., 2011, 47, 10145–10147.
26. V. M. Blas-Ferrando, J. Ortiz, K. Ohkubo, S. Fukuzumi, F. Fernández-Lázaro and Á. Sastre-Santos, Chem. Sci., 2014, 5, 4785–4793.
27. Y. Kobori, M. Fuki and H. Murai, J. Phys. Chem. B, 2010, 114, 14621–14630.
28. R. Carmieli, A. L. Smeigh, S. M. Mickley Conron, A. K. Thazhathveetil, M. Fuki, Y. Kobori, F. D. Lewis and M. R. Wasielewski, J. Am. Chem. Soc., 2012, 134, 11251–11260.
29. A. F. Coleman, M. Chen, J. Zhou, J. Y. Shin, Y. Wu, R. M. Young and M. R. Wasielewski, J. Phys. Chem. C, 2020, 124, 10408–10419.
30. N. Kaul and R. Lomoth, J. Am. Chem. Soc., 2021, 143, 10816–10821.
31. Y. Wu, R. M. Young, M. Frasconi, S. T. Schneebeli, P. Spenst, D. M. Gardner, K. E. Brown, F. Würthner, J. F. Stoddart and M. R. Wasielewski, J. Am. Chem. Soc., 2015, 137, 13236–13239.
32. A. Weller, Z. Phys. Chem., 1982, 133, 93–98.
33. S. Das, W. G. Thornbury, A. N. Bartynski, M. E. Thompson and S. E. Bradforth, J. Phys. Chem. Lett., 2018, 9, 3264–3270.
34. V. Markovic, D. Villamaina, I. Barabanov, L. M. Lawson Daku and E. Vauthey, Angew. Chem., Int. Ed., 2011, 50, 7596–7598.
35. J. M. Giaimo, A. V. Gusev and M. R. Wasielewski, J. Am. Chem. Soc., 2002, 124, 8530–8531.
36. M. W. Holman, P. Yan, D. M. Adams, S. Westenhoff and C. Silva, J. Phys. Chem. A, 2005, 109, 8548–8552.
37. L. Estergreen, A. R. Mencke, D. E. Cotton, N. V. Korovina, J. Michl, S. T. Roberts, M. E. Thompson and S. E. Bradforth, Acc. Chem. Res., 2022, 55, 1561–1572.
38. J. Wega, K.-F. Zhang, J. Lacour and E. Vauthey, J. Phys. Chem. Lett., 2024, 15, 2834–2840.
39. N. Banerji, A. Fürstenberg, S. Bhosale, A. L. Sisson, N. Sakai, S. Matile and E. Vauthey, J. Phys. Chem. B, 2008, 112, 8912–8922.
40. N. Sakai, M. Lista, O. Kel, S.-I. Sakurai, D. Emery, J. Mareda, E. Vauthey and S. Matile, J. Am. Chem. Soc., 2011, 133, 15224–15227.
41. D. Liu, A. M. El-Zohry, M. Taddei, C. Matt, L. Bussotti, Z. Wang, J. Zhao, O. F. Mohammed, M. Di Donato and S. Weber, Angew. Chem., Int. Ed., 2020, 59, 11591–11599.
42. M. Hu, A. A. Sukhanov, X. Zhang, A. Elmali, J. Zhao, S. Ji, A. Karatay and V. K. Voronkova, J. Phys. Chem. B, 2021, 125, 4187–4203.
43. X. Xiao, I. Kurganskii, P. Maity, J. Zhao, X. Jiang, O. F. Mohammed and M. Fedin, Chem. Sci., 2022, 13, 13426–13441.
44. G. Tang, A. A. Sukhanov, J. Zhao, W. Yang, Z. Wang, Q. Liu, V. K. Voronkova, M. Di Donato, D. Escudero and D. Jacquemin, J. Phys. Chem. C, 2019, 123, 30171–30186.
45. V. K. Gawade, R. W. Jadhav, V. R. Chari, R. V. Hangarge and S. V. Bhosale, Anal. Methods, 2023, 15, 3727–3734.
46. M. Al Kobaisi, S. V. Bhosale, K. Latham, A. M. Raynor and S. V. Bhosale, Chem. Rev., 2016, 116, 11685–11796.
47. M. P. Debreczeny, W. A. Svec and M. R. Wasielewski, Science, 1996, 274, 584–587.
48. M. Borgström, N. Shaikh, O. Johansson, M. F. Anderlund, S. Styring, B. Åkermark, A. Magnuson and L. Hammarström, J. Am. Chem. Soc., 2005, 127, 17504–17515.
49. F. Chaignon, F. Buchet, E. Blart, M. Falkenström, L. Hammarström and F. Odobel, New J. Chem., 2009, 33, 408–416.
50. P. Mahato, S. Saha, E. Suresh, R. Di Liddo, P. P. Parnigotto, M. T. Conconi, M. K. Kesharwani, B. Ganguly and A. Das, Inorg. Chem., 2012, 51, 1769–1777.
51. Y. Li, Z. Ma, A. Li, W. Xu, Y. Wang, H. Jiang, K. Wang, Y. Zhao and X. Jia, ACS Appl. Mater. Interfaces, 2017, 9, 8910–8918.
52. Z. Ma, Y. Ji, Y. Lan, G.-C. Kuang and X. Jia, J. Mater. Chem. C, 2018, 6, 2270–2274.
53. X. Wang, L. Guo and L. Feng, New J. Chem., 2019, 43, 4181–4187.
54. Y. Mori, Y. Sakaguchi and H. Hayashi, J. Phys. Chem. A, 2002, 106, 4453–4467.
55. N. Pearce, E. S. Davies, R. Horvath, C. R. Pfeiffer, X.-Z. Sun, W. Lewis, J. McMaster, M. W. George and N. R. Champness, Phys. Chem. Chem. Phys., 2018, 20, 752–764.
56. D. W. Cho, M. Fujitsuka, A. Sugimoto, U. C. Yoon, P. S. Mariano and T. Majima, J. Phys. Chem. B, 2006, 110, 11062–11068.
57. D. W. Cho, M. Fujitsuka, K. H. Choi, M. J. Park, U. C. Yoon and T. Majima, J. Phys. Chem. B, 2006, 110, 4576–4582.
58. D. W. Cho, M. Fujitsuka, A. Sugimoto, U. C. Yoon, D. W. Cho and T. Majima, Phys. Chem. Chem. Phys., 2014, 16, 5779–5784.
59. C. V. Suneesh and K. R. Gopidas, J. Phys. Chem. C, 2009, 113, 1606–1614.
60. L. Yuan, W. Lin, Y. Xie, B. Chen and S. Zhu, J. Am. Chem. Soc., 2012, 134, 1305–1315.
61. H. Li, H. Guan, X. Duan, J. Hu, G. Wang and Q. Wang, Org. Biomol. Chem., 2013, 11, 1805–1809.
62. X. Cui, J. Zhao, Z. Lou, S. Li, H. Wu and K.-L. Han, J. Org. Chem., 2015, 80, 568–581.
63. J. Sun, B. Hua, Q. Li, J. Zhou and J. Yang, Org. Lett., 2018, 20, 365–368.
64. C. Cui, X. Gao, X. Jia, Y. Jiao and C. Duan, Inorg. Chim. Acta, 2021, 520, 120285.
65. M. Mathivanan, B. Tharmalingam, O. Anitha, C.-H. Lin, V. Thiagarajan and B. Murugesapandian, Mater. Chem. Front., 2021, 5, 8183–8196.
66. Y. Dong, A. A. Sukhanov, J. Zhao, A. Elmali, X. Li, B. Dick, A. Karatay and V. K. Voronkova, J. Phys. Chem. C, 2019, 123, 22793–22811.
67. A. Chatterjee, J. Chatterjee, S. Sappati, T. Sheikh, R. M. Umesh, M. D. Ambhore, M. Lahiri and P. Hazra, J. Phys. Chem. B, 2021, 125, 12832–12846.
68. M. Imran, H. Cao, J. Zhao and G. Mazzone, J. Phys. Chem. A, 2023, 127, 4856–4866.
69. D. W. Cho, M. Fujitsuka, U. C. Yoon and T. Majima, J. Photochem. Photobiol., A, 2007, 190, 101–109.
70. X. Chen, A. A. Sukhanov, M. Taddei, B. Dick, J. Zhao, V. K. Voronkova and M. Di Donato, J. Phys. Chem. Lett., 2022, 13, 8740–8748.
71. J. Karpiuk and Z. R. Grabowski, AIP Conf. Proc., 1996, 364, 91–98.
72. X. Chen, A. A. Sukhanov, Y. Yan, D. Bese, C. Bese, J. Zhao, V. K. Voronkova, A. Barbon and H. G. Yaglioglu, Angew. Chem., Int. Ed., 2022, 61, e202203758.
73. K. Bhardwaj, M. Ahuja, S. K. Saini, M. Kumar and R. Kumar, J. Mol. Struct., 2024, 1301, 137338.
74. A. Khan, N. A. Tcyrulnikov, R. Roy, R. M. Young, M. R. Wasielewski and A. L. Koner, J. Phys. Chem. C, 2024, 128, 10474–10482.
75. A. Mazumder, K. Vinod, A. C. Thomas and M. Hariharan, J. Phys. Chem. Lett., 2025, 16, 4819–4827.
76. C. Liao, J. E. Yarnell, K. D. Glusac and K. S. Schanze, J. Phys. Chem. B, 2010, 114, 14763–14771.
77. A. L. Jones, M. K. Gish, C. J. I. V. Zeman, J. M. Papanikolas and K. S. Schanze, J. Phys. Chem. A, 2017, 121, 9579–9588.
78. Y. Mori, Y. Sakaguchi and H. Hayashi, Bull. Chem. Soc. Jpn., 2002, 74, 293–304.
79. M. Delor, T. Keane, P. A. Scattergood, I. V. Sazanovich, G. M. Greetham, M. Towrie, A. J. H. M. Meijer and J. A. Weinstein, Nat. Chem., 2015, 7, 689–695.
80. S. Stoll and A. Schweiger, J. Magn. Reson., 2006, 178, 42–55.
81. S. Richert, C. E. Tait and C. R. Timmel, J. Magn. Reson., 2017, 280, 103–116.
82. C. Hintze, U. E. Steiner and M. Drescher, ChemPhysChem, 2017, 18, 6–16.
83. D. R. Subedi, H. B. Gobeze, Y. E. Kandrashkin, P. K. Poddutoori, A. van der Est and F. D'Souza, Chem. Commun., 2020, 56, 6058–6061.
84. G. P. Wiederrecht, W. A. Svec, M. R. Wasielewski, T. Galili and H. Levanon, J. Am. Chem. Soc., 1999, 121, 7726–7727.
85. D. L. Meyer, R. Matsidik, M. Sommer and T. Biskup, Adv. Electron. Mater., 2018, 4, 1700385.
86. K. Hasharoni, H. Levanon, S. R. Greenfield, D. J. Gosztola, W. A. Svec and M. R. Wasielewski, J. Am. Chem. Soc., 1995, 117, 8055–8056.
87. S. Shaakov, T. Galili, E. Stavitski, H. Levanon, A. Lukas and M. R. Wasielewski, J. Am. Chem. Soc., 2003, 125, 6563–6572.
88. P. J. Brown, Y. Qiu, E. I. Latawiec, B. T. Phelan, N. A. Tcyrulnikov, J. R. Palmer, M. D. Krzyaniak, S. M. Kopp, Y. Huang, R. M. Young and M. R. Wasielewski, J. Phys. Chem. A, 2024, 128, 9371–9382.
89. N. Zarrabi, B. J. Bayard, S. Seetharaman, N. Holzer, P. Karr, S. Ciuti, A. Barbon, M. Di Valentin, A. van der Est, F. D’Souza and P. K. Poddutoori, Phys. Chem. Chem. Phys., 2021, 23, 960–970.
90. A. Karimata, S. Suzuki, M. Kozaki, K. Kimoto, K. Nozaki, H. Matsushita, N. Ikeda, K. Akiyama, D. Kosumi, H. Hashimoto and K. Okada, J. Phys. Chem. A, 2014, 118, 11262–11271.
91. H. Yonemura and M. D. E. Forbes, Photochem. Photobiol., 2015, 91, 672–677.
92. S. Suzuki, R. Sugimura, M. Kozaki, K. Keyaki, K. Nozaki, N. Ikeda, K. Akiyama and K. Okada, J. Am. Chem. Soc., 2009, 131, 10374–10375.
93. (a) CCDC 2370986: Experimental Crystal Structure Determination, 2025 DOI:10.5517/ccdc.csd.cc2kl6ft; (b) CCDC 2370987: Experimental Crystal Structure Determination, 2025 DOI:10.5517/ccdc.csd.cc2kl6gv.

---

Footnote

† These authors contributed equally to this work.

---