The effect of heavy atoms on the thermally activated delayed fluorescence properties of naphthalimide–phenoselenazine electron donor–acceptor dyads: electron transfer and intersystem crossing

Abstract

We prepared two electron donor–acceptor dyads based on naphthalimide (NI) as the electron acceptor and 10H-phenoxazine (PXZ) or 10H-phenoselenazine (PSeZ) as the electron donor. Both NI–PXZ and NI–PSeZ exhibit thermally activated delayed fluorescence (TADF). The purpose of this study is to unravel the effects of heavy atoms on the TADF properties, which are believed to enhance the reverse intersystem crossing (rISC); thus, more dark ³CS and ³LE states are transformed into the emissive ¹CS state, so that the TADF properties are enhanced by the presence of heavy atoms such as Se in the molecular structure. Steady state UV-vis absorption and fluorescence studies show negligible interaction between the donor and acceptor at the ground state (S₀), yet charge transfer (CT) emission was observed, indicating interaction between the radical anion and the radical cation of the CS state. We didn’t observe shortened delayed fluorescence lifetime for NI–PSeZ (τ_PF = 14.9 ns, τ_DF = 91.1 µs) as compared to the analogues of NI–PXZ (τ_PF = 24.3 ns, τ_DF = 57.8 µs) and the previously reported NI–PTZ (τ_PF = 11.9 ns, τ_DF = 82.1 µs); therefore the heavy atom effect for the rISC is not significant. Moreover, femtosecond transient absorption spectroscopy only captures the ¹CS state as the final species and the charge separation and recombination become faster in polar solvetns. Nanosecond transient absorption spectra show the presence of a low-lying ³NI state in non-polar solvents (τ_T = 41.3 µs), an admixture of the ³NI state and the ³CS state (τ_T = 23.5 µs) in solvents with intermediate polarity, and only the ³CS state (τ_T = 0.8 µs) in polar solvents. These observations support the spin-orbit charge transfer ISC (SOCT-ISC) mechanism. The results further show that the rISC is not enhanced by the heavy atom effect, for NI–PSeZ, the k_rISC = 7.0 × 10⁴ s⁻¹, in comparison, k_rISC = 1.3 × 10⁵ s⁻¹ for NI–PXZ, and k_rISC = 1.4 × 10⁷ s⁻¹ for NI–PTZ. Time-resolved electron paramagnetic resonance (TREPR) spectral studies show that the localized triplet state (³NI) is the last triplet state for the dyads in a frozen solution at 80 K, and based on the selective population of the three sublevels of the T₁ state, SOCT-ISC may contribute to the formation of the triplet states.

---

Introduction

Electron donor–acceptor compounds may undergo charge separation to form a charge-separated state upon photoexcitation,^1–7 or charge recombination-induced intersystem crossing (ISC) to form a localized triplet state (³LE, LE: localized excited).^8–11 These materials play a pivotal role in photovoltaics,^12–14 photocatalysis,^15,16 and organic light emitting diodes (OLEDs), such as the thermally activated delayed fluorescence (TADF) emitters,^17–26 as well as in fundamental photochemistry studies.^27,28 One of the conventional goals in studying these dyads is to achieve long-lived charge separation (CS) states, for mimicking the natural photosynthesis center.^29–37 The long-lived CS states can function as the oxidant (the radical cation part) or the reductant (the radical anion part); thus the long-lived CS state can drive intermolecular electron transfer to perform photocatalytic reactions.^{3,15,16,38,39} In the conventional electron donor–acceptor compounds, several strategies were used to prolong the CS state lifetimes, such as long separation distance between the donor and acceptor to reduce the electronic coupling (V_DA),⁴⁰ or use of the Marcus inverted region effect (eqn (1)):^27,29,41,42where ΔG⁰_ET denotes the Gibbs free energy change of the electron transfer, h and k_B represent the Planck and Boltzmann constants separately, V is the electronic matrix element and λ is the total reorganization energy.

The potential of the traditional electron donor–acceptor compounds has been almost fully exploited after several decades of investigation.^2,38,43,44 New strategies are desired for accessing the long-lived CS states in electron donor–acceptor compounds, for instance, in compact electron donor–acceptor compounds having a simple molecular structure, e.g. a short single bond as a linker between the donor and acceptor, instead of the long, saturated linker used in the synthetically demanding traditional electron donor–acceptor compounds. With this seemingly simple alteration of the molecular structure, the photophysics of the dyads will be substantially changed. First, the V_DA increases substantially for these compact dyads, and thus both the CS and the charge recombination (CR) processes will be accelerated.^8,9,43 Second, the molecular orbital overlap integral will increase significantly, leading to an increased electron exchange integral (J) of the two electrons in the overlap of singly occupied molecular orbitals (SOMO) of the radical anion and cations; thus, a larger energy gap between the ¹CS and ³CS states (2J) is expected. This large 2J value is in sharp contrast to the negligible 2J value in conventional non-compact D–A compounds. In the conventional electron donor–acceptor compounds, the spin–spin interaction of the radical anion and cation of the CS state is very weak, and the CS state exists in spin correlated radical pair (SCRP), the ¹CS and ³CS state share almost the same energy (2J can be down to a few MHz) and the interchange between the two species is fast.⁴³ Because the CR from the ¹CS state to the ground state (S₀) is spin-allowed, the CS state should have a short lifetime. In contrast, in the compact electron donor–acceptor compounds, however, the 2J is large, thus the interconversion between the ¹CS and ³CS state is inhibited, and a long-lived ³CS state can be formed because the CR from the ³CS state to the S₀ state is electron spin-forbidden. Therefore, this large 2J value will affect the CS state profoundly.^8,43 This so-called electron spin control method has been used to attain a long-lived ³CS state in compact electron donor–acceptor compounds.⁴³ The formation of a ³CS state, instead of the SCRP, has been unambiguously confirmed via pulse laser excited time-resolved electron paramagnetic resonance (TREPR) spectroscopy.^45–48

The photophysics of the compact electron donor–acceptor compounds is rich and interesting.⁴⁹ For instance, if the ³LE state of the electron donor–acceptor molecule shares a similar energy with that of the CS state, then TADF properties can result. However, due to the complexity of the photophysical processes of these electron donor–acceptor compounds, the detailed mechanism of the electron donor–acceptor TADF emitters has not been fully unveiled. For instance, the experimental evidence for the three-state model (the spin–vibronic coupling mechanism) of the TADF emitters is rare,^50–55 and controversial experimental evidence has been found for the recently proposed heavy atom effects on the TADF properties of the compounds.^56–59 Notably, the heavy atom effect on TADF has been well investigated in traditional chromophores with heavy atom substitution (e.g., Br/I), where spin–orbit coupling (SOC) is effectively enhanced to accelerate the ISC.⁵⁹ However, there is controversy regarding compact D–A TADF systems with short linkers between donor and acceptor.⁵⁷ In such compact systems, the strong electronic orbital overlap and large 2J energy gap and spin–orbit charge transfer ISC (SOCT-ISC) mechanism may alter the role of the heavy atom effect, but this hypothesis lacks direct verification by model molecular systems.

Herein we prepared two electron donor–acceptor compounds as model compounds to study the photophysics of the donor–acceptor TADF emitters, especially the heavy atom effect on the TADF photophysics. Steady state and femtosecond/nanosecond transient absorption and photoluminescence were used in the study. The dynamics of the transient paramagnetic species, such as the ³LE state, as well as the ³CS state, were studied via pulse laser excited TREPR spectroscopy.

Results and discussion

Molecular structure design and properties

Previously we prepared NI–PTZ dyads, with the N-phenyl PTZ attached at the imide position of the NI moiety and TADF properties were observed.⁶⁰ It is interesting that a phenyl linker is used between the NI and the phenothiazine (PTZ) moiety, yet electronic coupling between the two units exists, and the ¹CS state is emissive (without coupling between the donor and acceptor, the CS state would be a dark state). We also prepared NI–PXZ dyads in which the PXZ unit is connected at the 4-position of the NI moiety and TADF properties were observed.⁶¹ Herein in order to study the heavy atom effect on the photophysical properties of the dyads, we prepared a NI–PSeZ dyad, in which the heavy atom Se is embedded in the molecular skeleton, which is different from the molecular structures in which the bromo atoms are attached. Thus, NI–PSeZ was prepared (Scheme 1). As a reference compound, NI–PXZ was also prepared, the difference in the molecular structure as compared to that of NI–PSeZ and the previously reported NI–PTZ being only a one atom variation.⁶⁰ This is a unique method to modify the molecular structure with minimal perturbation, leaving many interactions between the two units in the dyad intact, which is beneficial to unveil the effect of a specific factor on the photophysics of the dyads.⁶²

Scheme 1 (a) 1,2-bromoiodobenzene, tBuONa, Pd₂(dba)₃, 1,1-bis(diphenylphosphino)ferrocene (DPPF), toluene, 120 °C, 12 h, under N₂, yield: 81%; (b) CuI, NaI, N,N′-dimethylethylenediamine, dioxane, 110 °C, 10 h, under N₂, yield: 79%; (c) Se, KOH, DMSO, 120 °C, 12 h, under N₂, yield: 70%; (d) 4-bromoaniline, acetic acid, 120 °C, 9 h, under N₂, yield: 46%; (e) phenoxazine, Pd(OAc)₂, tri-tert-butylphosphine tetrafluoroborate, K₂CO₃, toluene, 120 °C, 5 h, under N₂, yield: 15%; (f) phenoselenazine, similar to step (e), yield: 18%.

Single crystal X-ray diffraction analysis

The molecular structures of NI–PXZ and NI–PSeZ were determined via single crystal X-ray diffraction (Fig. 1). Single crystals of NI–PXZ and NI–PSeZ were obtained via slow diffusion of n-hexane (HEX) liquid into a chloroform (CHCl₃) solution of the compound. The geometry of NI–PSeZ adopts a quasi-equatorial (eq) conformation, in contrast to the quasi-axial (ax) conformation previously observed in PSeZ-containing D–A TADF emitters.⁵⁷ The torsion angles between the NI moiety and the phenyl unit are 86.7° (NI–PXZ) and 58.7° (NI–PSeZ) (Fig. 1a and b). The dihedral angle between the PXZ and phenyl unit is 78.3°, which is similar to that in NI–PSeZ (72.0°). The detailed parameters and the molecular structure of NI–PXZ and NI–PSeZ are presented in Table S1 (SI).

Fig. 1 ORTEP view of the molecular structures determined using single crystal X-ray diffraction: (a) NI–PXZ and (b) NI–PSeZ. Hydrogen atoms are omitted for clarity. Thermal ellipsoids are at 50% probability.

UV-vis absorption and fluorescence emission spectra

The UV-vis absorption spectra of the compounds were studied (Fig. 2). For PXZ and PSeZ, broad, structureless absorption bands in the range of 300–350 nm were observed. For NI–Br, a stronger, structured absorption band in the slightly red-shifted range was observed. For NI–PXZ and NI–PSeZ, stronger absorption bands in a similar range were observed, and the absorption profile of the dyads is virtually the sum of the electron donor (PXZ or PSeZ) and the electron acceptor parts (NI unit); thus, the electronic interaction of the electron donor and acceptor is negligible at the ground state. This is reasonable since the electron donor and acceptor part are isolated by an orthogonal phenyl linker.^60,63 In other words, the π-conjugation between the donor and the acceptor is excluded because the π-plane of phenyl linker adopts an orthogonal geometry against the NI part. This is supported by the lack of the S₀ → ¹CT absorption band, which was observed for an analogue NI–PTZ dyad, with the PTZ unit attached at the 4C position of the NI moiety (not the imide position as in NI–PXZ and NI–PSeZ).^64,65

Fig. 2 UV-vis absorption spectra of the compounds in cyclohexane (CHX), c = 1.0 × 10⁻⁵ M, 20 °C.

The fluorescence properties of the compounds were studied (Fig. 3). First, the fluorescence emission of the dyads in air and in a N₂ atmosphere were compared (Fig. 3a and b). For NI–PXZ, a broad, structureless emission band centered at 587 nm was observed under a N₂ atmosphere (Fig. 3a). This is a typical CT state emission.⁶⁴ However, the emission was quenched to a large extent in an aerated solution, and only a minor emission band at 535 nm was observed. Similar results were observed for NI–PSeZ (Fig. 3b). The sensitivity of the fluorescence emission intensity to O₂ is one characteristic of TADF.⁶⁶ Thus, we envision the TADF properties for these dyads. We propose that the minor emission band centered at 535 nm in an aerated solution originated from a CT state which has slightly higher energy than the major emissive CT state, giving an emission band centered at 587 nm. This is different from the previously reported NI–PTZ analogue.⁶⁰

Fig. 3 Fluorescence emission spectra of (a) NI–PXZ and (b) NI–PSeZ in CHX under different atmospheres, (c) NI–PXZ and (d) NI–PSeZ in different solvents under a N₂ atmosphere. Optically matched solutions were used and all solutions showed the same absorbance at the excitation wavelength, λ_ex = 340 nm, A = 0.10, 20 °C.

The fluorescence spectra of the two dyads in solvents with different polarities were studied (Fig. 3c and d). The results show that the fluorescence of the dyads is strongly quenched in polar solvents as compared to non-polar solvents. This result indicates that the emissive S₁ state has significant charge transfer character.⁶⁷ This is in agreement with the broad, structureless emission band of the compounds, as well as the TADF mechanisms, in which the ¹CS state is the emissive state (³CS and ³LE states are dark states).⁶⁰

Fluorescence lifetime studies

In order to confirm the TADF properties of the compounds, the fluorescence lifetimes of the dyads were studied (Fig. 4). Under a N₂ atmosphere, NI–PXZ shows a biexponential decay for the fluorescence (Fig. 4a), the faster decaying component has a lifetime of 24.3 ns (86%), and the slower decaying component has a lifetime of 57.8 µs (14%). Note that this delayed fluorescence lifetime is much shorter than that of the previously reported analogue containing a PTZ electron donor (τ_DF = 82.1 µs).⁶⁰ This result leads to the inference that a minor variation of the molecular structure by only one atom may alter the excited state properties of the compound significantly. Note the triplet state lifetimes have important implications in OLED devices because roll-off of the efficiency may result in a long triplet excited state lifetime.^68,69 Triplet–triplet annihilation (TTA) may also result and it may affect the OLED device performance. On the other hand, for the dyad NI–PSeZ, the prompt and delayed fluorescence lifetimes are τ_PF = 14.9 ns (91%) and τ_DF = 91.1 µs (9%), respectively. These values are longer than those for the previously reported dyad of NI–PTZ (τ_PF = 11.9 ns/τ_DF = 82.1 µs).⁶⁰ Under an air atmosphere, the luminescence lifetimes are shortened, especially the delayed fluorescence lifetimes.

Fig. 4 Fluorescence decay traces at 600 nm of (a) NI–PXZ and (b) NI–PSeZ; the short lifetime was measured with an EPL picosecond laser, and another component with longer lifetime was monitored using a microsecond flash xenon lamp. λ_ex = 340 nm, c = 5.0 × 10⁻⁵ M in deaerated CHX, 20 °C.

As compared to the luminescence lifetimes of NI–PTZ, NI–PSeZ shows longer luminescence lifetimes, i.e. there is no so-called heavy atom effect for the TADF process (accelerated rISC rate constant thus shortened triplet state lifetimes in the series analogues). The previously reported NI–N-PTZ dyad containing a C–N bond as the linker between the donor and acceptor also exhibits TADF properties, with a luminescence lifetime of 22.6 ns (96.6%)/2.6 µs (3.4%) in HEX.⁶⁵ The currently reported delayed fluorescence lifetime is much longer, for instance, for NI–PXZ (τ_PF = 28.7 ns/τ_DF = 43.1 µs) and NI–PSeZ (τ_PF = 14.7 ns/τ_DF = 64.5 µs) in HEX. This difference may be attributed to the different energy level orders of the ³CS state and the ³LE state. For NI–PXZ and NI–PSeZ, the energy of the ³LE state is relatively lower than that of the ³CS state. Therefore, the energy storage effect of the ³LE state results in a longer delayed fluorescence lifetime. In contrast, the ³CS state of NI–N-PTZ exhibits a lower energy, consequently leading to a shorter delayed fluorescence lifetime. The photophysical parameters of the compounds are summarized in Table 1. It was found that the fluorescence quantum yields are usually low. On the other hand, as an approximation of ISC quantum yields, the singlet oxygen quantum yields (Φ_Δ) of the compounds are high (Φ_Δ = 41% or 49%), which are slightly higher that reported previously for the NI–PTZ analogues (Φ_Δ = 38%)⁶⁰ and NI–N-PTZ analogues (Φ_Δ = 16%).⁶⁵

k_ISC = (1 − Φ_p)k_p (2)

k^T_nr = k_d − Φ_pk_rISC (4)

Table 1 The photophysical properties of compounds in CHX

	λ abs	ε	λ em	Φ Δ	τ F	Φ F	τ T
a Maximal UV-vis absorption wavelength in CHX, c = 1.0 × 10^−5 M, in nm. b Molar absorption coefficient at absorption maxima, in 10^4 M^−1 cm^−1. c Maximal fluorescence emission wavelength in CHX, λex = 340 nm, in nm. d Singlet oxygen quantum yields in CHX, Ru(bpy)3[PF6]2 was used as standard compound (ΦΔ = 57% in DCM), λex = 340 nm, in %. e Fluorescence lifetime in CHX under N2, λex = 340 nm. f Absolute fluorescence quantum yields under N2, determined using an optical integration sphere, λex = 340 nm, in %. g Triplet excited-state lifetimes in deaerated CHX, in µs. h Not observed.
PXZ	313	0.8	—^h	—^h	—^h	—^h	—^h
PSeZ	309	0.4	—^h	—^h	—^h	—^h	—^h
NI–Br	332/348	1.5/1.3	377	12	—^h	—^h	44.2
NI–PXZ	329/345	2.7/2.1	610	49	24.3 ns/57.8 µs	2.0	53.5
NI–PSeZ	327/342	2.3/1.9	616	41	14.9 ns/91.1 µs	1.7	41.3

---

For compounds exhibiting TADF behavior, the rate constants for ISC (k_ISC), reverse intersystem crossing (k_rISC), and triplet non-radiative decay (k^T_nr) may be derived from equations 2–4 based on the luminescence lifetimes and luminescence quantum yields.^70,71k_p and k_d represent the decay rate constants of prompt fluorescence and delayed fluorescence, respectively, where p and d denote the photoluminescence quantum efficiencies of prompt fluorescence and delayed fluorescence, respectively. For NI–PXZ, the rate constants k_ISC, k_rISC and k^T_nr were obtained as 4.1 × 10⁷ s⁻¹, 1.3 × 10⁵ s⁻¹ and 1.7 × 10⁴ s⁻¹, respectively. For NI–PSeZ, the rate constants were 6.7 × 10⁷ s⁻¹, 7.0 × 10⁴ s⁻¹ and 8.5 × 10³ s⁻¹, respectively. The key photophysical parameters for NI–PXZ, NI–PSeZ and the previously reported NI–PTZ clearly shows that the incorporation of the heavy atom Se in PSeZ does not enhance the rISC process: the k_rISC of NI–PSeZ (7.0 × 10⁴ s⁻¹) is lower than that of NI–PXZ (1.3 × 10⁵ s⁻¹) and far lower than that of NI–PTZ (1.4 × 10⁷ s⁻¹), accompanied by a longer delayed fluorescence lifetime (τ_DF = 91.1 µs) of NI–PSeZ compared with NI–PXZ (57.8 µs) and NI–PTZ (82.1 µs), which is in contrast to the expected acceleration of rISC by the heavy atom effect.⁵⁹

We also measured the Φ_Δ values of the compounds in solvents with different polarities (Table 2). The results show that for NI–Br, the Φ_Δ values increase in polar solvents as compared to those in non-polar solvents. For the two dyads, however, the Φ_Δ values became smaller in polar solvents, and no singlet oxygen photosensitizing ability was observed for the dyads in polar solvents, such as tetrahydrofuran (THF), DCM and ACN. These results show that in polar solvents, the CS state energy decreases, and becomes lower than in the ³NI state. Note the CS state energy is highly dependent on the solvent polarity, whereas the ³LE state energy is almost independent of the solvent polarity. Given that the ³CS state is the lowest state for the dyads upon photoexcitation, the singlet oxygen photosensitizing will be non-efficient because there is no triplet energy transfer to the ground state ³O₂ to produce ¹O₂. These postulations were confirmed by the nanosecond transient absorption spectral studies (see a later section).

Table 2 The singlet oxygen quantum yield (Φ_Δ/%) of the compounds^a

Compounds	CHX^b	HEX^c	TOL^d	THF^e	DCM^f	ACN^g
a Singlet oxygen quantum yield (ΦΔ) of dyads with Ru(bpy)3[PF6]2 as the standard (ΦΔ = 0.57 in DCM) in different solvents, λex = 340 nm. b E T (30) = 30.9 kcal mol^−1. c E T (30) = 31.0 kcal mol^−1. d E T (30) = 33.9 kcal mol^−1. e E T (30) = 37.4 kcal mol^−1. f E T (30) = 40.7 kcal mol^−1. g E T (30) = 45.6 kcal mol^−1. h Not observed.
NI–Br	12	6	25	19	44	61
NI–PXZ	49	43	12	—^h	—^h	—^h
NI–PSeZ	41	30	27	—^h	—^h	—^h

---

Electrochemical and spectroelectrochemical properties

In order to study the thermodynamics of the electron transfer in the dyads upon photoexcitation, the redox potentials of the compounds were studied (Fig. 5a). For both PXZ and PSeZ, similar reversible oxidation waves at +0.21 V (vs. Fc⁺/Fc) were observed (Fig. 5a and Table 3). For the dyads, similar oxidation potentials were observed (Fig. 5a and Table 3), indicating that the electronic coupling between the electron donor and acceptor is negligible. For NI–PXZ and NI–PSeZ, similar reversible reduction waves at −1.80 V (vs. Fc/Fc⁺) were observed. Compared to the similar structure NI–PTZ,⁶⁰ the oxidation potential of NI–PSeZ was 0.06 V higher than that of NI–PTZ, but both have similar reduction potentials. The difference in the reduction and oxidation potentials of NI–PSeZ and NI–PXZ is small.

Fig. 5 (a) Cyclic voltammogram of the compounds studied in deaerated DCM. Ferrocene (Fc) was used as the internal reference (set as 0 V in the cyclic voltammograms). 0.10 M Bu₄NPF₆ was used as the supporting electrolyte. Scan rates: 50 mV s⁻¹, c = 1.0 × 10⁻³ M, 20 °C. Spectroelectrochemical studies of NI–PSeZ: evolution of the UV-vis absorption spectra with reduction and oxidation potentials applied, (b) upon reduction under −1.72 V, c = 5.0 × 10⁻⁵ M and (c) upon oxidation under +0.71 V, c = 1.0 × 10⁻⁴ M. An Ag/AgNO₃ reference electrode was used. The spectra were recorded in situ with a spectroelectrochemical cell (1 mm optical path), in deaerated DCM, 20 °C.

Table 3 Redox potentials, Gibbs free energy changes of the charge separation (ΔG_CS) and charge separation states energy (E_CS) of the compounds in different solvents^a

Compounds	E RED/V^b	E ox/V^b	ΔGCS/eV					E CS/eV
			HEX	CHX	TOL	DCM	ACN	HEX	CHX	TOL	DCM	ACN
a Cyclic voltammetry in N2-saturated DCM containing a 0.10 M Bu4NPF6 supporting electrolyte; Pt electrode was used as the counter electrode; the working electrode is the glassy carbon electrode; Ag/AgNO3 couple is the reference electrode. E00 = 3.52 eV. E00 is the energy level of the singlet excited state localized on the NI moiety (^1NI*) approximated via the crossing point of UV-vis absorption and fluorescence emission spectra after normalization. b The value was obtained by setting the oxidation potential of Fc^+/Fc as 0. c Not observed or not applicable.
NI–PXZ	−1.79	+0.33	−0.86	−0.96	−1.05	−1.57	−1.71	2.66	2.56	2.47	1.95	1.80
NI–PSeZ	−1.81	+0.36	−0.82	−0.91	−1.00	−1.52	−1.66	2.70	2.61	2.52	2.00	1.86
NI–Br	−1.78	—^c	—^c	—^c	—^c	—^c	—^c	—^c	—^c	—^c	—^c	—^c

---

Based on the Weller eqn (5)–(7), the Gibbs free energy changes of the intramolecular electron transfer (ΔG_CS) were calculated (Table 3).^27,72,73 The results show that electron transfer is thermodynamically allowed even in the non-polar solvent cyclohexane and it becomes more exothermic in polar solvents. Thus, electron transfer should be observed, especially in polar solvents. It should be noted that what was observed in transient absorption spectra may not necessarily follow this trend, because the electron transfer can be outcompeted kinetically by other photophysical processes, such as ISC to produce the ³LE state. Moreover, determination error may exist for the relative energy levels of the CS state and the ³LE state.

Moreover, it should be noted the CS state energies of NI–PXZ and NI–PSeZ are different, and the ³LE states localized on the NI moiety should be virtually the same. Thus, besides the possible heavy atom effect on the photophysics of NI–PSeZ, the effect of the variation of CS/³LE state energy gap should not be neglected.⁵⁹ Previously we found similar variation of the CS state energy of brominated electron donor–acceptor TADF emitters.⁵⁹ Thus, one should be careful in interpretation of the so-called heavy atom effect in the electron donor–acceptor TADF emitters. The ³NI state energy was determined as 2.25 eV,⁷⁴ and thus in polar solvents such as DCM and ACN, the CS state should be the lowest state, which should be observed in the nanosecond transient absorption spectra (see a later section).

ΔG_CS = e[E_OX − E_RED] − E₀₀ + ΔG_s (6)

E_CS = e[E_OX − E_RED] + ΔG_s (7)

In order to study the UV-vis absorption of the radical anions and radical cations, and thus to aid the analysis of the femtosecond transient absorption spectral data, spectroelectrochemical characterization of the dyads was performed (Fig. 5b and c). For NI–PSeZ, when a negative potential of −1.72 V (vs. Fc/Fc⁺) was applied, the original absorption band centered at 335 nm decreased, concomitantly, and two new absorption bands centered at 420 nm, 490 nm and 830 nm were observed, which were attributed to the NI⁻˙ radical anion.⁶⁰ On the other hand, when a positive potential of +0.71 V (vs. Fc⁺/Fc) was applied, the original absorption band at 330 nm showed only a minor decrease, which is reasonable, because the NI unit should be intact, and only the PSeZ part should be oxidized. Indeed, new absorption bands centered at 525 nm and 835 nm were observed, which were attributed to the PSeZ⁺˙ radical cation.⁵⁹ This information will be useful for assignment of the CS state in the femtosecond or nanosecond transient absorption spectra (see later sections). It should be noted that upon photoexcitation, CS states may be formed. In these CS states, both the radical anion and cation exist simultaneously, and thus there is coulombic interaction between the ions. Such interaction does not exist in the electrochemically generated radical anions or cations. However, this interaction may impose negligible effects on the UV-vis absorptions of the CS states.

Femtosecond transient absorption (fs-TA) spectra

In order to study the excited kinetics of the dyads, the fs-TA spectra of the compounds were studied (Fig. 6). For NI–PXZ, positive absorption bands centered at 410 nm and 540 nm were observed, then red-shifted absorption bands developed at longer delay times, and the absorption band centered at ca. 410 nm became sharper. In order to analyse the transient species in detail, evolution associated difference spectra (EADS) were obtained via global fitting and target analysis, based on the sequential model (Fig. 6b). The first spectrum shows positive absorption bands centered at 410 nm and 490 nm, which are assigned to the ¹NI state.⁶⁰ The absorption profile is slightly different from the ¹NI state with PXZ attached at the 4-position of the NI chromophore via a C–N bond.⁶¹ Then after 1.4 ps, the second spectrum developed, which shows drastically different absorption bands centered at 420 nm and a broad absorption band centered at 550 nm. These absorption bands are in agreement with the absorption of NI⁻˙ radical anions and the PXZ⁺˙ radical cations, respectively (Fig. S16). Thus, the second spectrum is attributed to the CS state, and the CS takes 1.4 ps.

Fig. 6 Femtosecond transient absorption spectra of (a) NI–PXZ and (b) NI–PSeZ in CHX, and the relative EADS obtained from global analysis of the transient absorption data of (c) NI–PXZ and (d) NI–PSeZ upon excitation at 350 nm.

This CS process is slightly slower than for the previously reported NI–PTZ analogue, in which the CS takes 0.8 ps.⁶⁰ After 12.5 ps, the third spectrum developed, which is very close to the second spectrum. Thus, the third spectrum is assigned to the solvation relaxed CS state. Then after ca. 542.9 ps, the system evolves to another state, the final state with a long lifetime is a CS state, and we propose that it is a ¹CS state. The geometry relaxation may take 542.9 ps. The final spectrum is a CS state. This result indicates that the CR takes longer than 1.5 ns. This CS in cyclohexane is in agreement with the study of the thermodynamics of the CS (Table 3), which shows the CS from the ¹LE state is thermodynamically allowed. Note the excited state dynamics is different from the NI–PXZ analogue with the PXZ unit attached at the 4-position of the NI moiety.⁶¹ Due to the simple molecular structure, we did not observe any upper intermediate states,⁷⁵ or delocalized excited states.⁷⁶

Similarly, we studied the excited state dynamics of the dyad NI–PSeZ (Fig. 6c and d). The CS takes 2.6 ps, which is two times longer than that of NI–PXZ. Then similar solvent and geometry relaxation were observed, and the final state was relaxed ¹CS state. This assignment is supported by the long prompt fluorescence lifetime of 14.9 ns (Fig. 3b). These results show that the CR from the ¹CS state to produce the ³NI state is slow.

The CS and CR of NI–PSeZ and NI–PXZ in TOL and ACN were also studied (Fig. S18 and S19, SI). With the increase in solvent polarity, for NI–PSeZ in toluene, ¹NI was first observed to reach the ¹CS state via rapid CS, and within 313.1 ps, the ¹CS state relaxed to the lowest vibrational excited ¹CS state through vibrational relaxation. A similar process was observed for NI–PXZ in toluene: it reached the ¹CS state within 1.4 ps, followed by relaxation to the lowest vibrational excited ¹CS state at 153.7 ps. Unlike the compound NI–PXZ reported in the literature,⁶¹ this study observed the long-lived ³CS state, likely due to differences arising from the varying positions at which the PXZ chromophore is attached to NI. In the highly polar ACN solvent, NI–PXZ and NI–PSeZ exhibit no TADF properties, and no triplet CS state was observed in fs-TA spectra (Fig. S18d and S19d). NI–PXZ and NI–PSeZ undergo CS at 770 fs and 448 fs, respectively, to generate the ¹CS state, and subsequently undergo CR to the ground state at 59.8 ps and 265.6 ps, respectively. The signal decayed completely within the time window of the femtosecond measurement, indicating a negligible proportion of long-lived components. This is consistent with the photophysical process of NI–PXZ reported in the literature in which the PXZ unit is connected at the 4-position of the NI moiety.⁶¹ The charge recombination process in this compound occurs within 10.3 ps, which is faster than the rates observed in NI–PXZ and NI–PSeZ. Furthermore, due to the rapid CR of NI–PXZ and NI–PSeZ in ACN, triplet states cannot be generated via ISC. Consequently, no CS state signal was detected in ACN in longer-timescale ns-TA spectra.

Nanosecond transient absorption (ns-TA) spectra

In order to study the long-lived species formed in the dyads upon photoexcitation, the nanosecond transient absorption (ns-TA) spectra of the dyads were studied (Fig. 7). For NI–PSeZ in CHX, a prominent positive absorption band centered at 460 nm was observed (Fig. 7a), which is attributed to the triplet excited state absorption (ESA) of the NI T₁ state (i.e. T₁ → T_n transitions), this is supported by the ns-TA spectra of the reference compound NI–Br (Fig. S20). The triplet state lifetime of the ³NI state (³LE, i.e. localized excited state) was determined as 41.3 µs (Fig. 7d). In this case, the CS state has higher energy (2.61 eV, Table 3) than the ³NI state (E_T1 = 2.25 eV).⁷⁴

Fig. 7 Nanosecond transient absorption spectra of NI–PSeZ in different deaerated solvents of (a) CHX (c = 4.0 × 10⁻⁵ M), (b) TOL (c = 4.0 × 10⁻⁵ M), and (c) THF (c = 3.0 × 10⁻⁵ M). The corresponding decay traces are presented in (d) 470 nm (c = 2.0 × 10⁻⁵ M), (e) 470 nm (c = 5.0 × 10⁻⁶ M), and (f) 410 nm (c = 1.5 × 10⁻⁵ M), λ_ex = 355 nm, 20 °C.

In a more polar solvent TOL, the prominent absorption band centered at 470 nm is persistent, and moreover, two shoulder absorption bands centered at 410 nm and 540 nm were observed (Fig. 7b), these shoulder bands are attributed to NI⁻˙ and PSeZ⁺˙, respectively, based on the spectroelectrochemical studies (Fig. 7b and c). The transient decays had a lifetime of 23.5 µs, which is attributed to the intramolecular CS.

The decay kinetics monitored at three wavelengths (410 nm, 470 nm, and 540 nm) give the same decay rate constants, which suggests that the ³CS state and the ³LE state are in good equilibrium. This is in agreement with the spin–vibronic coupling mechanism of the TADF process. We also measured the ns-TA spectra of NI–PSeZ in a more polar solvent THF (Fig. 7c). The results show that the absorption bands centred at 410 nm and 500 nm are more prominent, and thus the CS state is dominant. The lifetime is shortened to 0.8 µs. These results show that even with small variation of the ³CS/³LE states energy gap, the photophysical properties of the electron donor–acceptor TADF emitter will change substantially.⁷⁷ In the polar solvent ACN, no triplet CS state signal of NI–PSeZ was detected, which is consistent with the results from fs-TA. We found the CS state lifetime of NI–PSeZ in TOL (23.5 µs) is much longer than that of the reference dyad NI–PXZ (13.8 µs) and the previously reported analogue of NI–PTZ (ca. 0.03 µs).⁶⁰ These results indicate that there is no simple heavy atom effect to enhance the rISC to shorten the triplet states lifetimes. Other factors, such as the energy gaps of the states may also play a role in determination of the photophysics of the electron donor–acceptor dyads. Moreover, the property of the current NI–PSeZ is also different from the NI–N-PTZ analogue with the PTZ moiety attached at the 4-position of the NI moiety,⁶⁵ which shows a lifetime of 2.6 µs in hexane, and the ³CS state is dominant. In HEX, NI–PSeZ exhibits a ³LE state with the triplet-state lifetime of 43.7 µs, longer than that of NI–N-PTZ.

In order to study the ns-TA spectra of NI–PXZ, measurements were repeated in solvents with different polarities (Fig. S22). In the non-polar solvents CHX and HEX, identical triplet signals (³LE state) were detected, accompanied by relatively long triplet lifetimes of 53.5 µs and 25.7 µs (Fig. S22d and S21c), respectively. As the polarity of the solvent increased, the ns-TA signals were distinctly different. No TADF was observed in toluene, likely due to reduced CS energy levels caused by increased polarity. The ns-TA signals shifted to 410 nm and 540 nm, revealing a mixture of ³LE and ³CS states and a shortening of the triplet lifetime to 13.8 µs (Fig. S22e). In the more polar THF solvent, the triplet signal gradually evolved into a pure ³CS signal. The ³NI* signal at 470 nm completely disappeared, and the lifetime was shortened to 4.7 µs (Fig. S22f). Similar to NI–PSeZ, no signal for NI–PXZ was observed in ACN. Moreover, we found that NI–PXZ, which contains a phenyl linkage between the donor and acceptor moieties, exhibits longer triplet lifetimes than the previously reported NI–PXZ analogue when tested under identical solvent conditions.^61,78 But the ³LE state lifetime in CHX is significantly shorter compared to NI–PTZ (ca. 274.7 µs).⁶⁰

To confirm the formation of the CS state in polar solvents for NI–PSeZ and NI–PXZ, radical anion quenching experiments were performed. Electron acceptor 4-isopropyl-4′-methyldiphenyliodonium tetrakis(pentafluorophenyl)borate (DPI) was added to the solution and the change of ns-TA spectra was monitored (Fig. 8).^39,79 In the presence of DPI, we observed the disappearance of the NI⁻˙ absorption band, whereas the PSeZ⁺˙ band remained unchanged.

Fig. 8 (a) Nanosecond transient absorption spectra of NI–PSeZ (3.0 × 10⁻⁵ M) before and after addition of DPI (5.0 × 10⁻⁵ M) in deaerated THF. (b) The decay traces of NI–PSeZ (1.5 × 10⁻⁵ M) at 410 nm upon incremental addition of DPI and (c) the corresponding Stern–Volmer plots for quenching of the NI–PXZ and NI–PSeZ³CS states, λ_ex = 355 nm. All solutions used in measurements were deaerated with N₂, 20 °C.

This is reasonable because the quencher, DPI, is an electron acceptor and only the radical anion NI⁻˙ can be quenched, and thus the radical cation (PSeZ⁺˙) cannot be quenched by DPI and remains intact (Fig. 8a). By incrementally adding the DPI to the solution, the lifetimes of the species become shorter (Fig. 8b). For the NI⁻˙ state, which absorbs at 410 nm, the lifetime became shorter upon the addition of DPI. For instance, the lifetime monitored by the trace at 410 nm changes from 0.85 µs (0 eq. DPI added), to 0.66 µs (1.67 eq. DPI) and 0.44 µs (2.58 eq. DPI). Accordingly, based on the Stern–Volmer eqn (8), K_SV is calculated to be 2.8 × 10⁴ M⁻¹ (Fig. 8c). Similar behavior was observed for NI–PXZ, for which the K_SV value is calculated to be 4.0 × 10⁴ M⁻¹.

Time-resolved electron paramagnetic resonance (TREPR) spectra

Both NI–PXZ and NI–PSeZ show TADF properties, thus, it is interesting to analyze the existence of the transient paramagnetic species for these TADF emitters, for instance, the localized triplet state (³LE) and the triplet CS state (³CS), as well as their relative energy level order. Electron spin dynamics, as well as electron spin polarization (ESP), offer information about the ISC mechanism, etc.^53,80,81 TREPR spectroscopy is an ideal tool.^82–84 First, we studied the triplet state TREPR spectra of the electron donors PXZ and PSeZ (Fig. 9). For PXZ (in frozen solution in TOL/2-MeTHF = 3 : 1, v/v), a typical TREPR spectrum for the triplet state of a powder sample (randomly oriented in the magnetic field) was observed, with an ESP pattern of (e, e, e, a, a, a) (e: emission; a: enhanced absorption). Half-field transition at 155 mT was observed, demonstrating the signal is due to the triplet excited state. The population rates of the three sublevels of the T₁ state were determined as P_x : P_y : P_z = 0.88 : 0.58 : 0.06, based on spectral simulation. Spectral analysis gives the zero-field splitting (ZFS) |D| = 3738 MHz and |E| = 386 MHz (Table 4). These data obtained with the powder sample are similar to the previous report with single crystals for the triplet state of phenoxazine, |D| = 0.099 cm⁻¹ (2970 MHz), |E| = 0.015 cm⁻¹ (450 MHz).⁸⁵ Similarly, we also recorded the triplet state TREPR spectrum of the PSeZ (Fig. 9). The ESP phase pattern of the spectrum is (e, e, e, a, a, a), half-field transition was also observed. Spectral simulation gives the population rates of the three sublevels as P_x : P_y : P_z = 0.45 : 0.58 : 0.06, and ZFS |D| = 3190 MHz and |E| = 494 MHz. These data are similar to the previous report for the triplet state of PTZ, measured with single crystal, ZFS |D| = 0.123 cm⁻¹ (3690 MHz) and |E| = 0.014 cm⁻¹ (420 MHz).⁸⁵ These large D values are due to the small π-conjugation size of these chromophores.

Fig. 9 TREPR spectra of compounds measured using a X-band EPR spectrometer excited with a 355 nm laser with energy 2 mJ per pulse in TOL/2-MeTHF (3 : 1, v/v) at 80 K. The simulation (red curve) was performed using the EasySpin package, c = 1.0 × 10⁻⁴ M.

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 simulations of the triplet-state TREPR spectra for compounds^a

Compounds	|D| (MHz)	|E| (MHz)	P x  : Py : Pz	ΔP^b
a Obtained from simulation of the triplet-state TREPR spectra of the indicated molecules in frozen solution (TOL/2-MeTHF, 3/1, v/v) at 80 K. b ΔP = |Px − Py|/|Pz − Py|. c The second species (relative intensity of 30%) exhibit the same polarization state, attributed to the formation of some molecular clusters due to the high concentration used because of the low intensity of PSeZ.
NI–PXZ	2478	126	0.45 : 1 : 0.60	1.38
NI–PSeZ	2461	148	0.54 : 1 : 0.43	0.81
NI–Br	2474	127	0.50 : 1 : 0.46	0.93
PXZ	3738	386	0.88 : 0.58 : 0.06	0.58
PSeZ	3190	494	0.45 : 0.58 : 0.06	0.25
	2461^c	148^c	0.45 : 0.58 : 0.06	0.25

---

Next, we studied the TREPR spectrum of the NI–Br. The bromophenyl group is not a typical electron donor, and so we presume that the ISC of this reference compound should be the SOC–ISC. The TREPR spectrum of this compound shows an ESP pattern of (e, e, a, e, a, a), which is the same as that observed with the NI derivative (with n-butylamine at the imide position).⁸⁶ Spectral simulations show that the population rates of the three sublevels are P_x : P_y : P_z = 0.5 : 1 : 0.46. The ZFS |D| was determined as 2473 MHz and |E| was determined as 127 MHz, which is similar to the previous report.⁸⁶

Next, we studied the TREPR spectra of the electron donor–acceptor dyads. For NI–PXZ, a TREPR spectrum with an ESP phase pattern of (e, e, a, e, a, a) was observed, which is similar to that of NI–Br. Spectral simulations show the population rates of the three sublevels of the T₁ state P_x : P_y : P_z = 0.45 : 1 : 0.6, and the ZFS to be |D| = 2478 MHz, |E| = 126 MHz. These values are similar to that of NI–Br. The slightly different population rates of the three sublevels of NI–PXZ and NI–Br may be attributed to the different ISC mechanisms. For NI–PXZ, the charge recombination induced ISC may contribute to the formation of the triplet state. In the fs-TA spectral study, we found CS and CR in CHX, although the process may be inhibited to some extent in a frozen solution at 80 K. No ³CS state was observed, which should show a much smaller |D| value than the ³LE state.⁸⁷ This result shows that ³LE has a lower energy than the ³CS state for NI–PXZ, in non-polar solvents. This conclusion is supported by the ns-TA spectral studies (Fig. 9). The results of NI–PXZ are different from the previously studied analogue of the dyad with the PXZ unit attached at the 4-position of the NI chromophore, which has the ³LE state and the ³CS state showing similar energy, and a fast equilibrium between the two states, which leads to a much smaller |D| value of 1484 MHz.⁶¹ The elevated ³CS state energy for the current NI–PXZ is attributed to the larger distance between the NI and the PXZ units (a phenyl linker between the two units is used, for the previously reported analogue, a C–N single bond was used as the linker).⁶¹ For NI–PSeZ, similar results were observed as compared to those for NI–PXZ. For both dyads, only the ³NI state (the ³LE state) was observed. These results show that the ³CS state energy is elevated to be above that of the ³LE state. Previously with analogues with a dyad based on NI and PTZ units, a weak radical pair signal was observed,⁶⁰ which is probably due to the slightly more polar solvent used in that case as well as the stronger electron donating ability of the PTZ unit.⁶⁰ These results show that a TADF emitter can have a ³LE state as the lowest state,^51,80 although it was shown that a good TADF emitter should show a more significant ³CS state in the TREPR spectrum.^65,88 This information is useful for understanding the photophysics of the TADF emitters.

Theoretical calculations

DFT computations were performed on the compounds to rationalize the photophysical properties. First, the ground state geometry of the dyads was optimized using the DFT method (Fig. 10a). For NI–PXZ, the dihedral angles between the PXZ and the NI moieties against the intervening phenyl linker are 87.9° and −69.5°, respectively. However, geometry fluctuation is possible in fluid solution at room temperature. Similar results were observed for NI–PSeZ (Fig. 10b).

Fig. 10 Optimized ground state geometry and energies of the selected frontier orbitals (isovalue = 0.02) of (a) NI–PXZ and (b) NI–PSeZ, based on the results of optimized ground state geometry. Calculated at CAM-B3LYP/6-31G(d) level with Gaussian 16.

The frontier molecular orbitals of the dyads were examined. For NI–PXZ, the HOMO is confined on the PXZ part, and LUMO is confined on the NI moiety. Thus, the PXZ part is the electron donor and NI is the electron acceptor. Similar results were observed for NI–PSeZ (Fig. 10b). Moreover, a slight distribution of the electron density on the phenyl linker is observed, thus electronic coupling between the donor and acceptor exists, this is critical for the emissive character of the ¹CS state, otherwise the ¹CS state will be a dark state.⁴³

The electron spin density surface of the T₁ state of the dyads under different conditions was studied (Fig. 11) to unveil the electronic character of the T₁ states. First, the T₁ state spin density of the dyads in the gas phase was studied (this condition can be treated as an approximation of the situation in non-polar solvents). For NI–PXZ, the spin density is distributed on both the NI and the PXZ moieties. However, the density on the two parts is unequal, and higher spin density was observed for the NI part. Based on the spin density distribution of the triplet state of the NI moiety, and the PXZ⁺˙, we conclude that in the gas phase, the T₁ state of NI–PXZ has significant ³LE character. Similar results were observed for NI–PSeZ (Fig. 11b). These results are in agreement with the ns-TA spectral results, in which a ³LE state was observed.

Fig. 11 Triplet spin density distributions of (a) NI–PXZ and (b) NI–PSeZ in the gas phase, (c) NI–PXZ and (d) NI–PSeZ in THF, isovalue = 0.0004. Calculations were performed by DFT theory at CAM-B3LYP/6-31G(d) level with Gaussian 16.

However, in the polar solvent THF, we found the electron spin density distributions on the two parts are more equal. This is in agreement with the character of a ³CS state. This postulation is supported by the spin density distribution pattern on NI, and is different from that in Fig. 11a. Moreover, the spin distribution pattern on the NI unit is similar to that of NI⁻˙ (Fig. S26). This is in agreement with the ns-TA spectral studies. Thus we conclude a ³CS state is formed for NI–PXZ. Similar results were observed for NI–PSeZ. Previously we found a similar scenario for the electron donor–acceptor compounds,⁵⁹i.e. the PTZ unit adopts a bent geometry with the dyad populating the ³NI state, whereas the PTZ units change to a plane geometry in the CS state of the dyad (i.e. in its radical cation form). Interestingly, we found that the PSeZ unit adopts different geometries in the dyad in the gas phase and in THF solvent, i.e. the unit exhibits a twisted geometry in the gas phase and a planar geometry in THF. That means the neutral and the cationic forms of this chromophore adopt a different geometry, which is similar to the scenario of PTZ.^89,90 In contrast, the configuration of PXZ is identical in the gas phase and in the THF solvent, both adopting a planar geometry.

We calculated the spin orbit coupling matrix elements (SOCMEs) of the possible ISC pathways of the compounds using the TDDFT method in the ORCA 6.0 program.⁹¹ First, we checked the reference compounds. For NI, a large SOCME of 14.9 cm⁻¹ was obtained for the S₁ → T₃ ISC (the two states have energy of 3.67 eV and 3.57 eV, respectively), which is in good agreement with the previous study,⁹² and it should be responsible for the ISC of this chromophore. For PSeZ, large SOCMEs were also obtained, for example, 70.7 cm⁻¹ for S₁ → T₂ and 20.9 cm⁻¹ for S₁ → T₃, respectively, which are due to the heavy atom effect of Se atoms.^93–95 For PXZ, however, much smaller SOCMEs were obtained (<0.05 cm⁻¹). For the D–A dyads, much smaller SOCMEs were obtained. For NI–PXZ, the SOCMEs of the S₁ → T₁ and S₁ → T₂ are negligible. The optically transition is S₀ → S₇ (3.655 eV, oscillator strength is 0.325), the SOCMEs is 0.60 cm⁻¹ for the ISC of S₇ state to the triplet states sharing similar energy or lower energy. For NI–PSeZ, this value is 11.8 cm⁻¹. However, the fs-TA study shows that upon photoexcitation, the ¹LE state undergoes ultrafast CS to form the ¹CS state with a time constant of 1.4 ps (NI–PXZ) and 2.6 ps (NI–PSeZ), and this CS process is kinetically far faster than the direct ISC from the LE state (the rate of direct ISC is much lower than that of CS based on the calculated SOCMEs). Thus, the direct ISC from the LE state is not the dominant photophysical process and makes a negligible contribution to the overall excited state dynamics of the dyads, so we do not discuss this process further.

The ZFS D tensor of the triplet state of the compounds was preliminarily studied using the ORCA program.⁹¹ The ZFS D of the NI triplet state was calculated as 0.036 cm⁻¹, which is in reasonable agreement with the experimental results of 0.082 cm⁻¹.⁸⁶ The calculation also shows that the SOC contribute (0.057 cm⁻¹) more significantly than the spin–spin coupling (SSC, −0.02 cm⁻¹). For the compound PXZ, the calculated ZFS D is 0.036 cm⁻¹, which is smaller than the experimental value of 0.12 cm⁻¹ (with single crystals, the value was reported to be ±0.099 cm⁻¹).⁹⁶ In this case the SOC contribution (−0.035 cm⁻¹) is smaller than the SSC (0.070 cm⁻¹). For PSeZ, however, the ZFS D is largely overestimated (−9.85 cm⁻¹), compared to the experimental value (0.11 cm⁻¹). The SOC contributed substantially (−9.86 cm⁻¹) more than the SSC (0.015 cm⁻¹). For NI–PXZ, the ZFS D of the triplet state was calculated as −0.037 cm⁻¹, which is close to the experimental value of 0.083 cm⁻¹ (note that the sign of the ZFS D cannot be determined by experiments). However, for NI–PSeZ, the ZFS D (2.40 cm⁻¹) is largely overestimated, compared to the experimental value (0.082 cm⁻¹). This result could be due to the overestimated SOC part (2.39 cm⁻¹).

The photophysical processes of NI–PSeZ in different solvents are summarized in Scheme 2. In the nonpolar solvent CHX, NI–PSeZ is excited to generate the ¹LE state. This is followed by a rapid CS (2.6 ps) to generate the ¹CS state. Driven by the SOCT-ISC mechanism, CR (386.2 ps) occurs, producing the ³LE localized on the NI moiety. It is known that the energy gap between ¹CS and the ³CS is generally small. Under this condition, ¹CS, ³LE, and ³CS share close energy levels. The spin–vibrational coupling between ³LE and ³CS facilitates rISC, which regenerates the ¹CS state and emits delayed fluorescence. However, with increasing solvent polarity, the energy level of the CS state gradually decreases, the energy gap between ³LE and ³CS increases, inhibits the occurrence of spin–vibrational coupling and makes the rISC process difficult. These results indicate that the TADF of electron D–A dyads follows the three-state model (energy level equilibrium among ¹CS, ³LE, and ³CS).

Scheme 2 Photophysical process of NI-PSeZ in (a) CHX and (b) TOL.^{a a}The ¹LE and ³LE state energy is calculated using TDDFT at the CAM-B3LYP/6-31G(d) level using Gaussian 16. The ¹CS state energy is obtained by electrochemical calculation. J is the electronic exchange energy, and the energy difference between ¹CS and ³CS is 2J. The number of the superscript designates spin multiplicity.

Conclusions

In summary, we prepared two electron donor–acceptor dyads as thermally activated delayed fluorescence (TADF) emitters in order to study the heavy-atom effect on the intersystem crossing (ISC) and reverse ISC (rISC) of these TADF emitters. The dyads are based on naphthalimide (NI) as the electron acceptor and 10H-phenoxazine (PXZ) or 10H-phenoselenazine (PSeZ) as the electron donor, with the N-phenyl PXZ or PSeZ attached at the imide position of the NI moiety (dyads of NI–PXZ and NI–PSeZ, respectively). The linker between the electron donor and acceptor is a phenyl moiety. Interestingly, the luminescence lifetime studies show that the delayed fluorescence lifetime of NI–PSeZ containing the heavy atom Se (τ_PF = 14.9 ns, τ_DF = 91.1 µs) is longer compared to the analogues of NI–PXZ (τ_PF = 24.3 ns, τ_DF = 57.8 µs) as well as the previously reported NI–PTZ (τ_PF = 11.9 ns, τ_DF = 82.1 µs). Thus, apparently the rISC is not accelerated by the presence of heavy atoms in the molecular structures of the currently investigated compounds. In femtosecond transient absorption spectroscopy, the ¹CS state was observed in a non-polar solvent (CHX) with no evidence of a charge recombination (CR) process, consistent with the results of fluorescence lifetime analysis. As long-lived species are undetectable in the femtosecond time window, nanosecond transient absorption spectroscopy was used. Nanosecond transient absorption spectra show the low-lying ³NI state in a non-polar solvent (CHX, τ = 41.3 µs), an admixture of the ³NI state and ³CS state (τ = 23.5 µs) in a solvent with intermediate polarity (TOL), and only the ³CS state (τ = 0.8 µs) in polar solvents (THF). Combined with femtosecond and nanosecond results, these observations confirm the SOCT-ISC mechanism. These experimental observations are consistent with the Gibbs free energy changes of the electron transfer and the charge separation state energy. Furthermore, the CS state was observed in polar solvents, while no TADF phenomenon was detected, which also supports the spin-vibronic coupling theoretical model of TADF. Pulsed laser excited electron paramagnetic resonance (TREPR) spectral studies show that the localized triplet state (³NI) is the last triplet state for the dyads in frozen solution at 80 K, and based on the selective population of the three sublevels of the T₁ state, spin orbit charge transfer ISC may contribute to the formation of the triplet states.

Author contributions

J. Z.: acquired the funding, conceived the research study, directed the data analysis and the writing of the manuscript; P. J.: synthesized the compounds, collected and analyzed the data and modified the manuscript; Y. L. and Y. P.: revised parts of the manuscript; H. B. and Y. W.: recorded the femtosecond transient absorption spectra; A. T. and A. B.: recorded and simulated the TREPR spectra and analyzed the data.

Conflicts of interest

There are no conflicts to declare.

Data availability

All relevant data are presented in the main text and/or the supplementary information (SI). The supplementary information includes experimental details, molecular structure characterization (¹H/ ¹³C NMR, HRMS, single crystal X-ray diffraction with CCDC 2476452 and 2476453), photophysical measurements (UV−vis absorption/fluorescence spectra, fluorescence lifetimes), femtosecond/nanosecond transient absorption spectra, spectroelectrochemical data, and theoretical calculation results (DFT, SOCMEs, ZFS parameters). See DOI: https://doi.org/10.1039/d6cp00066e.

CCDC 2476452 and 2476453 contain the supplementary crystallographic data for this paper.^97a,b

Acknowledgements

J. Z. thanks the NSFC (22473021, U25A20619 and W2521100), the National Key Research and Development Program of China (the Ministry of Science and Technology, 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 for financial support. A. Barbon acknowledges the Italian Ministry of Foreign Affairs (Progetti Grande Rilevanza 2024, CN24GR01). Funding by the European Union – Next Generation UE (PRIN2022 PNRR PHOTOCORE) is gratefully acknowledged.

Notes and references

1. D. M. Guldi, Chem. Commun., 2000, 321–327.
2. J. W. Verhoeven, H. J. van Ramesdonk, M. M. Groeneveld, A. C. Benniston and A. Harriman, ChemPhysChem, 2005, 6, 2251–2260.
3. S. Fukuzumi, Pure Appl. Chem, 2007, 79, 981–991.
4. S. Suzuki, M. Kozaki, K. Nozaki and K. Okada, J. Photochem. Photobiol. C: Photochem. Rev., 2011, 12, 269–292.
5. O. S. Wenger, Chem. Soc. Rev., 2011, 40, 3538–3550.
6. E. Vauthey, ChemPhysChem, 2012, 13, 2001–2011.
7. H. Imahori, Bull. Chem. Soc. Jpn., 2023, 96, 339–352.
8. Y. Hou, X. Zhang, K. Chen, D. Liu, Z. Wang, Q. Liu, J. Zhao and A. Barbon, J. Mater. Chem. C, 2019, 7, 12048–12074.
9. M. A. Filatov, Org. Biomol. Chem., 2020, 18, 10–27.
10. D. J. Gibbons, A. Farawar, P. Mazzella, S. Leroy-Lhez and R. M. Williams, Photochem. Photobiol. Sci., 2020, 19, 136–158.
11. X. Zhang, Z. Wang, Y. Hou, Y. Yan, J. Zhao and B. Dick, J. Mater. Chem. C, 2021, 9, 11944–11973.
12. 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.
13. A. Devižis, J. De Jonghe-Risse, R. Hany, F. Nüesch, S. Jenatsch, V. Gulbinas and J.-E. Moser, J. Am. Chem. Soc., 2015, 137, 8192–8198.
14. E. Salvadori, N. Luke, J. Shaikh, A. Leventis, H. Bronstein, C. W. M. Kay and T. M. Clarke, J. Mater. Chem. A, 2017, 5, 24335–24343.
15. M. Xiang, Q.-Y. Meng, X.-W. Gao, T. Lei, B. Chen, C.-H. Tung and L.-Z. Wu, Org. Chem. Front., 2016, 3, 486–490.
16. S. M. Sartor, Y. M. Lattke, B. G. McCarthy, G. M. Miyake and N. H. Damrauer, J. Phys. Chem. A, 2019, 123, 4727–4736.
17. H. Uoyama, K. Goushi, K. Shizu, H. Nomura and C. Adachi, Nature, 2012, 492, 234–238.
18. Y. Tao, K. Yuan, T. Chen, P. Xu, H. Li, R. Chen, C. Zheng, L. Zhang and W. Huang, Adv. Mater., 2014, 26, 7931–7958.
19. Y. Im, S. Y. Byun, J. H. Kim, D. R. Lee, C. S. Oh, K. S. Yook and J. Y. Lee, Adv. Funct. Mater., 2017, 27, 1603007.
20. Z. Yang, Z. Mao, Z. Xie, Y. Zhang, S. Liu, J. Zhao, J. Xu, Z. Chi and M. P. Aldred, Chem. Soc. Rev., 2017, 46, 915–1016.
21. X. Cai and S.-J. Su, Adv. Funct. Mater., 2018, 28, 1802558.
22. T. Huang, W. Jiang and L. Duan, J. Mater. Chem. C, 2018, 6, 5577–5596.
23. F. Ni, N. Li, L. Zhan and C. Yang, Adv. Opt. Mater., 2020, 8, 1902187.
24. Y. Xu, P. Xu, D. Hu and Y. Ma, Chem. Soc. Rev., 2021, 50, 1030–1069.
25. L. Chen, W.-C. Chen, Z. Yang, J.-H. Tan, S. Ji, H.-L. Zhang, Y. Huo and C.-S. Lee, J. Mater. Chem. C, 2021, 9, 17233–17264.
26. Y.-Z. Shi, H. Wu, K. Wang, J. Yu, X.-M. Ou and X.-H. Zhang, Chem. Sci., 2022, 13, 3625–3651.
27. 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.
28. H. Imahori, Y. Kobori and H. Kaji, Acc. Mater. Res., 2021, 2, 501–514.
29. 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.
30. 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.
31. J. Hankache and O. S. Wenger, Chem. Commun., 2011, 47, 10145–10147.
32. N. L. Bill, M. Ishida, Y. Kawashima, K. Ohkubo, Y. M. Sung, V. M. Lynch, J. M. Lim, D. Kim, J. L. Sessler and S. Fukuzumi, Chem. Sci., 2014, 5, 3888–3896.
33. V. M. Blas-Ferrando, J. Ortiz, K. Ohkubo, S. Fukuzumi, F. Fernández-Lázaro and Á. Sastre-Santos, Chem. Sci., 2014, 5, 4785–4793.
34. S.-H. Lee, C. T.-L. Chan, K. M.-C. Wong, W. H. Lam, W.-M. Kwok and V. W.-W. Yam, J. Am. Chem. Soc., 2014, 136, 10041–10052.
35. T. Higashino, T. Yamada, M. Yamamoto, A. Furube, N. V. Tkachenko, T. Miura, Y. Kobori, R. Jono, K. Yamashita and H. Imahori, Angew. Chem., Int. Ed., 2015, 55, 629–633.
36. 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.
37. K. B. Daisymol and K. R. Gopidas, J. Phys. Chem. Lett., 2023, 14, 977–982.
38. K. Ohkubo and S. Fukuzumi, Bull. Chem. Soc. Jpn, 2009, 82, 303–315.
39. Y. Pei, X. Chen, Y. Hou, J. Zhao, Y. Li and S. Ji, Polym. Chem., 2025, 16, 2600–2610.
40. K.-Y. Chen, C.-C. Hsieh, Y.-M. Cheng, C.-H. Lai, P.-T. Chou and T. J. Chow, J. Phys. Chem. A, 2006, 110, 12136–12144.
41. E. H. Yonemoto, G. B. Saupe, R. H. Schmehl, S. M. Hubig, R. L. Riley, B. L. Iverson and T. E. Mallouk, J. Am. Chem. Soc., 1994, 116, 4786–4795.
42. 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.
43. J. W. Verhoeven, J. Photochem. Photobiol., C, 2006, 7, 40–60.
44. H. Kotani, K. Ohkubo and S. Fukuzumi, Faraday Discuss., 2012, 155, 89–102.
45. Y. Kobori, M. Fuki and H. Murai, J. Phys. Chem. B, 2010, 114, 14621–14630.
46. R. Carmieli, A. L. Smeigh, S. M. M. Conron, A. K. Thazhathveetil, M. Fuki, Y. Kobori, F. D. Lewis and M. R. Wasielewski, J. Am. Chem. Soc., 2012, 134, 11251–11260.
47. H. Yonemura and M. D. E. Forbes, Photochem. Photobiol., 2015, 91, 672–677.
48. Y. Oka and K. Inoue, Phys. Chem. Chem. Phys., 2024, 26, 16444–16448.
49. X. Chen, X. Zhang, X. Xiao, Z. Wang and J. Zhao, Angew. Chem., Int. Ed., 2023, 62, e202216010.
50. J. Gibson, A. P. Monkman and T. J. Penfold, ChemPhysChem, 2016, 17, 2956–2961.
51. P. K. Samanta, D. Kim, V. Coropceanu and J. L. Brédas, J. Am. Chem. Soc., 2017, 139, 4042–4051.
52. F. B. Dias, T. J. Penfold and A. P. Monkman, Methods Appl. Fluoresc., 2017, 5, 012001.
53. E. W. Evans, Y. Olivier, Y. Puttisong, W. K. Myers, T. J. H. Hele, S. M. Menke, T. H. Thomas, D. Credgington, D. Beljonne, R. H. Friend and N. C. Greenham, J. Phys. Chem. Lett., 2018, 9, 4053–4058.
54. I. Kim, K. H. Cho, S. O. Jeon, W.-J. Son, D. Kim, Y. M. Rhee, I. Jang, H. Choi and D. S. Kim, JACS Au, 2021, 1, 987–997.
55. M. Hempe, N. A. Kukhta, A. Danos, M. A. Fox, A. S. Batsanov, A. P. Monkman and M. R. Bryce, Chem. Mater., 2021, 33, 3066–3080.
56. B. H. Drummond, G. C. Hoover, A. J. Gillett, N. Aizawa, W. K. Myers, B. T. McAllister, S. T. E. Jones, Y. J. Pu, D. Credgington and D. S. Seferos, J. Phys. Chem. C, 2020, 124, 6364–6370.
57. S. Goto, Y. Nitta, N. O. Decarli, L. E. de Sousa, P. Stachelek, N. Tohnai, S. Minakata, P. de Silva, P. Data and Y. Takeda, J. Mater. Chem. C, 2021, 9, 13942–13953.
58. Y. X. Hu, J. Miao, T. Hua, Z. Huang, Y. Qi, Y. Zou, Y. Qiu, H. Xia, H. Liu, X. Cao and C. Yang, Nat. Photonics, 2022, 16, 803–810.
59. Y. Pei, A. A. Sukhanov, X. Chen, A. Iagatti, S. Doria, X. Dong, J. Zhao, Y. Li, W. Chi, V. K. Voronkova, M. Di Donato and B. Dick, Chem. Eur. J., 2024, 31, e202403542.
60. X. Xiao, Y. Yan, A. A. Sukhanov, S. Doria, A. Iagatti, L. Bussotti, J. Zhao, M. Di Donato and V. K. Voronkova, J. Phys. Chem. B, 2023, 127, 6982–6998.
61. X. Zhang, X. Liu, M. Taddei, L. Bussotti, I. Kurganskii, M. Li, X. Jiang, L. Xing, S. Ji, Y. Huo, J. Zhao, M. Di Donato, Y. Wan, Z. Zhao and M. V. Fedin, Chem. Eur. J., 2022, 28, e202200510.
62. M. K. Etherington, J. Gibson, H. F. Higginbotham, T. J. Penfold and A. P. Monkman, Nat. Commun., 2016, 7, 13680.
63. H. Tanaka, K. Shizu, H. Nakanotani and C. Adachi, J. Phys. Chem. C, 2014, 118, 15985–15994.
64. S. Sasaki, K. Hattori, K. Igawa and G.-I. Konishi, J. Phys. Chem. A, 2015, 119, 4898–4906.
65. 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.
66. H. Tanaka, K. Shizu, H. Miyazaki and C. Adachi, Chem. Commun., 2012, 48, 11392–11394.
67. M. Baruah, W. Qin, C. Flors, J. Hofkens, R. A. L. Vallée, D. Beljonne, M. Van der Auweraer, W. M. De Borggraeve and N. Boens, J. Phys. Chem. A, 2006, 110, 5998–6009.
68. R. Wang, Z. Li, T. Hu, L. Tian, X. Hu, S. Liu, C. Cao, Z.-L. Zhu, J.-H. Tan, Y. Yi, P. Wang, C.-S. Lee and Y. Wang, ACS Appl. Mater. Interfaces, 2021, 13, 49066–49075.
69. Y. H. Lee, Y.-S. Shin, T. Lee, J. Jung, J.-H. Lee and M. H. Lee, Chem. Eng. J., 2021, 423, 130224.
70. K. Goushi, K. Yoshida, K. Sato and C. Adachi, Nat. Photonics, 2012, 6, 253–258.
71. R. J. Vázquez, J. H. Yun, A. K. Muthike, M. Howell, H. Kim, I. K. Madu, T. Kim, P. Zimmerman, J. Y. Lee and T. Goodson III, J. Am. Chem. Soc., 2020, 142, 8074–8079.
72. R. Ziessel, B. D. Allen, D. B. Rewinska and A. Harriman, Chem. Eur. J., 2009, 15, 7382–7393.
73. 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.
74. B. Ventura, A. Bertocco, D. Braga, L. Catalano, S. d'Agostino, F. Grepioni and P. Taddei, J. Phys. Chem. C, 2014, 118, 18646–18658.
75. H. Noda, X.-K. Chen, H. Nakanotani, T. Hosokai, M. Miyajima, N. Notsuka, Y. Kashima, J.-L. Brédas and C. Adachi, Nat. Mater., 2019, 18, 1084–1090.
76. T. Hosokai, H. Matsuzaki, H. Nakanotani, K. Tokumaru, T. Tsutsui, A. Furube, K. Nasu, H. Nomura, M. Yahiro and C. Adachi, Sci. Adv., 2017, 3, e1603282.
77. P. L. dos Santos, M. K. Etherington and A. P. Monkman, J. Mater. Chem. C, 2018, 6, 4842–4853.
78. J. Tang, X. Liu, X. Zhang, J. Zhao and Y. Wan, New J. Chem., 2023, 47, 22418–22429.
79. Y. Toba, Y. Usui, M. M. Alam and O. Ito, Macromolecules, 1998, 31, 6022–6029.
80. T. Ogiwara, Y. Wakikawa and T. Ikoma, J. Phys. Chem. A, 2015, 119, 3415–3418.
81. S. Yumoto, J. Katsumata, F. Osawa, Y. Wada, K. Suzuki, H. Kaji and K. Marumoto, Sci. Rep., 2023, 13, 11109.
82. S. Weber, eMagRes, 2017, 6, 255–270.
83. T. Biskup, Front. Chem., 2019, 7, 10.
84. S. Richert, C. E. Tait and C. R. Timmel, J. Magn. Reson., 2017, 280, 103–116.
85. J. M. Lhoste, A. Haug and M. Ptak, J. Chem. Phys., 1966, 44, 648–654.
86. K. Chen, I. V. Kurganskii, X. Zhang, A. Elmali, J. Zhao, A. Karatay and M. V. Fedin, Chem. Eur. J., 2021, 27, 7572–7587.
87. L. Lv, K. Yuan, T. Zhao, H. Li and D. Wang, New J. Chem., 2023, 47, 18972–18982.
88. B. H. Drummond, N. Aizawa, Y. Zhang, W. K. Myers, Y. Xiong, M. W. Cooper, S. Barlow, Q. Gu, L. R. Weiss, A. J. Gillett, D. Credgington, Y.-J. Pu, S. R. Marder and E. W. Evans, Nat. Commun., 2021, 12, 4532.
89. 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.
90. T. Okamoto, M. Kuratsu, M. Kozaki, K. Hirotsu, A. Ichimura, T. Matsushita and K. Okada, Org. Lett., 2004, 6, 3493–3496.
91. F. Neese, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2025, 15, e70019.
92. P. K. Samanta and N. J. English, J. Phys. Chem. C, 2020, 124, 8178–8185.
93. V. V. Patil, H. J. Jang and J. Y. Lee, J. Mater. Chem. C, 2023, 11, 8084–8090.
94. D. Zhong, S. Liu, L. Yue, Z. Feng, H. Wang, P. Yang, B. Su, X. Yang, Y. Sun and G. Zhou, Chem. Sci., 2024, 15, 9112–9119.
95. Z. Cheng, X. Wu, X. Zhang, S. Gao, H. Tong and L. Wang, Adv. Opt. Mater., 2025, 13, 2500517.
96. J.-M. Lhoste, M. Ptak and D. Lexa, J. Chim. Phys., 1968, 65, 1876–1888.
97. (a) CCDC 2476452: Experimental Crystal Structure Determination, 2026 DOI:10.5517/ccdc.csd.cc2p3ykb; (b) CCDC 2476453: Experimental Crystal Structure Determination, 2026 DOI:10.5517/ccdc.csd.cc2p3ylc.

---

Footnote

† These authors contributed equally to this work.

---