A Thermally Activated Delayed Fluorescence Emitter Investigated by Time‐Resolved Near‐Infrared Spectroscopy

Abstract

Emitters for organic light‐emitting diodes (OLEDs) based on thermally activated delayed fluorescence (TADF) require small singlet (S₁)‐triplet (T₁) energy gaps as well as fast intersystem crossing (ISC) transitions. These transitions can be mediated by vibronic mixing with higher excited states S _n and T _n (n=2, 3, 4, …). For a prototypical TADF emitter consisting of a triarylamine and a dicyanobenzene moiety (TAA‐DCN) it is shown that these higher states can be located energetically by time‐resolved near‐infrared (NIR) spectroscopy.

A metal‐free TADF (thermally activated delayed fluorescence) emitter was investigated by time‐resolved near‐infrared spectroscopy. This enabled upper excited states that contribute to spin‐vibronic coupling to be detected.

Introduction

Thermally activated delayed fluorescence (TADF) emitters are designed to convert singlet and triplet excitons efficiently into light. ^[1] They show great potential for the application in organic light‐emitting diodes (OLEDs). ^[2] In OLEDs, excitons are generated by electron‐hole recombination. Due to spin statistics, singlet and triplet excitons are generated in a 1 : 3 ratio, and emitters based on TADF can harvest both excitons to achieve 100 % internal quantum efficiency. ^[3]

Efficient TADF emitters ought to feature small singlet‐triplet energy gaps as well as high rate constants for intersystem crossing (ISC) and reverse intersystem crossing (rISC) between singlet and triplet states. ^[4] The energy gap (ΔE _ST) between the lowest singlet (S₁) and triplet (T₁) excited states ought to be of the order of the thermal energy k _B T. ^[2c] It equals twice the exchange energy, and is small for a small overlap between hole and electron densities. ^[1c] In a molecular orbital picture, the hole density is related to the highest occupied molecular orbital (HOMO) and the electron density to the lowest unoccupied molecular orbital (LUMO) of the emitter. Small singlet‐triplet energy gaps are commonly encountered in compounds with donor and acceptor moieties where the HOMO (LUMO) is mainly localized on the donor (acceptor) moiety. ^[5] As a consequence, the relevant singlet and triplet excited states have a strong charge‐transfer (CT) character. Concerning the kinetics, efficient TADF emitters ought to exhibit high rate constants for ISC and rISC, since the delayed fluorescence relies on the triplet‐to‐singlet energy up‐conversion mechanism. However, between “pure” CT excited states the spin‐orbit coupling (SOC) mediating ISC and rISC is very weak. ^[6] There are two mechanisms, vibrational spin‐orbit and spin‐vibronic coupling, that can lift the forbiddenness of the ISC/rISC transitions. ^[7] With perturbation theory, expressions for the impact on these mechanisms on the ISC/rISC rate constants can be derived. These expressions reveal an explicit and implicit (see, e. g., discussion in ref. [8]) influence of the energetic separation of CT and locally excited (LE) states (Figure 1) on these rate constants. Thus, it seems worthwhile to address the respective energies experimentally. The energic separations are related to S₁→S _n and T₁→T _n (n=2, 3, 4, …) transition energies that ought to be found in the near‐infrared (NIR) region. Respective experiments are presented in the following.

Figure 1.

Scheme showing excited‐state processes with ISC and rISC mediated by spin‐vibronic coupling. Due to vibronic mixing, the lowest excited states of CT character (horizontal stripes) gain some LE character (vertical stripes). The transitions ¹CT→¹LE and ³CT→³LE are located in the NIR region.

A pulsed excitation promotes the emitter to the S₁(¹CT) state and the S₁→S _n transitions can be monitored. After a suitable waiting time the T₁(³CT) state will be populated and the T₁→T _n transitions are accessed. Respective proof of principle experiments will be described here.

The TADF emitter TAA‐DCN studied here consists of a triarylamine (TAA) donor and an 1,4‐dicyanobenzene (DCN) acceptor moiety (Scheme 1). Dissolved in toluene and at room temperature, the compound was shown to be TADF active with a prompt fluorescence lifetime of 21 ns and a delayed one of 30 μs. ^[9] According to temperature‐dependent measurements its energy gap ΔE _ST was estimated to be 980 cm⁻¹ (0.12 eV). ^[9]

Scheme 1.

Structure of 4′‐(diphenylamino)‐2′‐methyl‐[1,1′‐biphenyl]‐2,5‐dicarbonitrile (TAA‐DCN). The dihedral angle α is indicated.

Here, femto‐ and nanosecond transient absorption measurements of TAA‐DCN covering the UV/Vis to NIR range are reported on. S₁→S _n and T₁→T _n (n=2, 3, 4, …) transitions were observed and compared favourably with high‐level quantum chemical computations. As these computations point to a non‐Condon effect in the S₁ emission respective experiments were also conducted.

Results and Discussion

Quantum chemical computations

To assess the spectral range in which the lowest excited state transition S₁→S _n and T₁→T _n (n=2, 3, 4, …) are to be expected, quantum‐chemical computations were performed. A previous TD‐DFT (B3‐LYP) study ^[9] gives the first estimates for the vertical excitation energies. It places the S₁→S₂ transition at 1.01 eV and the T₁→T₂ one at 0.63 eV at the optimized ground state geometry. In the NIR experiments vertical energies with respect to the S₁ and T₁ geometries will be recorded. Furthermore, TD‐DFT is known to have deficits when it comes to describe CT excitations in conjunction with standard hybrid density functionals such as B3‐LYP. ^[10] The CT states do not only appear at too low energies in the TD‐DFT spectrum, also the twist angle of the donor and acceptor moieties is typically too large. To avoid these problems, geometry optimizations of the S₀, S₁, and T₁ states were performed with Gaussian16 ^[11] employing an optimally tuned range‐separated hybrid density functional ωB97X‐D ^[12] (ω=0.15 Bohr⁻¹, details about the optimal tuning procedure in Figure S1 in the Supporting Information) in combination with a def2‐TZVP basis set. In these computations, the solvent toluene (permittivity ϵ_r=2.38 ^[13] ) was considered implicitly using the polarizable continuum model (PCM) implemented in Gaussian16. Two rotamers were identified for TAA‐DCN in the S₀ state (Figure S2). They differ in the dihedral angle α defined by the orientation of the m‐(Ph₂N)‐toluene and acceptor moieties (Scheme 1). The rotamer lowest in energy features an angle α of 70.2°. The second one (α=112.3°) is about 0.02 eV higher in energy. Thus, both rotamers should be present at room temperature with a slight preference for α=70.2°. The respective profile for the S₁ state (Figure S3) shows that this rotamer remains the one with the lowest energy. In the T₁ state (Figure S4), the minima are shifted to α=40° and 150°. In this state, a barrier of ∼0.35 eV separates the rotamers. This should preclude an interconversion on the timescales relevant here. Thus, to simplify the analysis and discussions, only the preferred rotamer, lowest in energy, with α=70.2° will be considered further. In addition, the computed spectra signatures of the two conformers differ only slightly (Tables S1–S3). Vertical singlet excitation energies with respect to the optimized S₀ geometry were computed with the DFT/MRCI‐R2016 ^[14] approach. The energies and the respective oscillator strengths are summarized in Table 1 and Figure 2. The energies are by ∼0.5 eV higher than the ones computed previously, ^[9] a behaviour already expected due to the different methods applied. The vertical S₀→S₁ energy computed here is in better agreement with the experiment (see below). The respective oscillator strength $f$ amounts to 0.069. The colour bar code plotted in Figure 2 highlights the CT character of the excited states (see Tables S1–S3 for difference densities ^[15] ).

Table 1.

Vertical excitation energies [eV] and oscillator strengths for S₀, S₁, and T₁ equilibrium geometry of TAA‐DCN in toluene.

	TD‐DFT[a]		DFT/MRCI[b]
Geometry	S0 vertical	Oscillator Strength	S0 vertical	Oscillator Strength	S1 vertical (adiabatic)	Oscillator Strength	T1 vertical (adiabatic)	Oscillator Strength
S0	0	–	0	–	−2.50	0.089	−2.01	–
S1	2.42	0.073	3.06	0.069	0 (2.89)	–	–	–
S2	3.43	0.335	3.83	0.205	1.22	0.009	–	–
S3	3.65	0.006	3.97	0.222	1.38	0.034	–	–
S4	3.74	0.016	4.26	0.251	1.54	0.220	–	–
T1	2.32	–	2.90	–	–	–	0 (2.53)	–
T2	2.95	–	3.32	–	–	–	0.89	0.024
T3	3.11	–	3.46	–	–	–	1.17	0.295
T4	3.24	–	3.52	–	–	–	1.39	0.001

[a] TD‐DFT vertical electronic transitions of TAA‐DCN in toluene modelled according to the polarizable continuum model (PCM) for the ground state molecular geometry at the B3LYP/6‐31G(d,p) level of theory from ref. [9]. [b] DFT/MRCI vertical electronic transitions of TAA‐DCN in toluene modelled according to the polarizable continuum model (PCM) for the S₀, S₁, and T₁ equilibrium geometry at OT‐ωB97X‐D/def2‐TZVP level of theory.

Figure 2.

Computed vertical excitation energies of TAA‐DCN in toluene and the CT character at the respectively optimized geometry for the relevant excited states. The vertical energies refer to the S₀, S₁, and T₁ equilibrium geometry. The CT character of the excited sates is depicted by the coloured bars (with the corresponding values in brackets); it was obtained from a fragment‐based analysis of the transition density matrix.

Starting with the S₀ geometry, TAA‐DCN was geometry optimized in its S₁ state. The optimized S₁ structure of the predominant rotamer exhibits a dihedral angle α of 65.9°. The structural relaxation lowers the S₁ energy from 3.06 to 2.89 eV and the computed vertical emission energy amounts to 2.50 eV. Interestingly, this relaxation goes along with a slight increase of the oscillator strength f to 0.089. This is due to the more parallel arrangement between donor and acceptor moieties resulting in a more local excitation character. So, a dependence of the transition dipole moment on the nuclear coordinates, that is, a non‐Condon effect, ^[16] is predicted. Vertical excitation energies with respect to the S₁ geometry and the oscillator strengths of the spin‐ and electric dipole‐allowed transitions are shown in Table 1 and Figure 2 (see Table S2 for the other rotamer). The computation places the lowest of these transitions (S₁→S₂) at 1.22 eV. The value is similar to the rough estimate given above and indicates that in the experiment we ought to search for this transition at ∼1000 nm.

The T₁ state was also geometry optimized. The optimized T₁ structure exhibits a dihedral angle α of 41.4°. Its computed adiabatic energy amounts to 2.53 eV. This translates into a singlet‐triplet gap ΔE _ST of 0.36 eV. This is roughly a factor of three higher than the earlier experimental value derived from kinetic data. ^[9] The computed vertical excitations with respect to the T₁ geometry and the oscillator strengths of the spin‐ and electric dipole‐allowed transitions are displayed in Table 1 and Figure 2 (see Table S3 for the other rotamer). According to these computations, the lowest transition (T₁→T₂) ought to be found at 0.89 eV or ∼1400 nm. These quantum chemical predictions will now be compared with spectroscopic measurements.

Steady‐state spectroscopy

TAA‐DCN in toluene exhibits a lowest‐energy absorption band peaking at around 382 nm with an absorption coefficient ϵ_max of 3844 M⁻¹cm⁻¹ (Figure 3). An experimental value for the oscillator strength f was determined by first plotting the absorption coefficient ϵ($\tilde{v}$ ) as a function of wavenumber. To isolate the lowest energy transition, a Gaussian fit was performed (Figure 3a). From the integral of this Gaussian, f ^[17] was derived. It amounts to 0.09 and is close to the quantum chemical one of 0.07. Experimental excitation energies were determined from absorption and fluorescence spectra converted into the transition dipole representation ^[17] (Figure 3b). Equating the respective absorption maximum with the vertical excitation energy for S₀→S₁ yields a value of 25 830 cm⁻¹ (3.20 eV) in good agreement with the computation. From the intersection of the normalized and redrawn absorption and emission spectra we deduce a 0–0 energy of 22 676 cm⁻¹ (2.81 eV). This is close to the computed adiabatic energy of 2.87 eV including zero‐point vibrational energy corrections. The maximum of the emission is located at 19 790 cm⁻¹ (2.45 eV), again in good agreement with the computation. Absorption and emission maxima translate into Stokes shift of 6040 cm⁻¹.

Figure 3.

Absorption coefficient and fluorescence spectra of TAA‐DCN in toluene. a) Comparison of measured absorption spectra (ϵ($\tilde{v}$ ), black line) and the calculated transitions from the S₀ state (sticks). The filled area represents the region used to determine the experimental oscillator strength. b) Spectra in the transition dipole representation. Absorption data are reported as absorption coefficient as a function of wavenumber (ϵ($\tilde{v}$ ) and were rescaled according to ϵ($\tilde{v}$ )/$\tilde{v}$ , and fluorescence spectrum (F($\tilde{v}$ )) according to F($\tilde{v}$ )/$\tilde{v}$ ³; then the rescaled spectra were normalized (for details see the Experimental Section). For the acquisition of the fluorescence spectrum, λ_ex=380 nm.

Time‐resolved spectroscopy

The S₁ state of TAA‐DCN was characterized by femtosecond UV/Vis and nanosecond NIR absorption spectroscopy. In the femtosecond experiment, a TAA‐DCN solution was excited with 400 nm laser pulses (Figure 4).

Figure 4.

Femtosecond transient absorption of TAA‐DCN in toluene (c=1.9 mM) after excitation at 400 nm. In the contour representation, the difference absorption ΔA is shown as a function of the detection wavelength λ and the delay time t is colour coded. Selected transient spectra of up to 10 ps highlight the change in stimulated emission with time. The spectral region in which stimulated emission is expected is marked by the dashed lines. The arrows indicate the increase in the stimulated emission during solvent relaxation.

Around time zero, the excitation generates a broad excited state absorption (ESA) signature covering the complete detection range (360–730 nm). Maxima around 720, 435, and <350 nm are discernible. A pronounced minimum is located around 500 nm. With reference to the fluorescence spectrum (Figure 3), this is attributed to stimulated emission (SE). Note that in a transient absorption experiment SE gives a negative signal contribution. Here, this negative contribution “cuts into” a strong and broad positive one. Up ∼10 ps changes of the SE contribution are observable (see also top panel in Figure 4). The SE spectrum shifts to the red (dynamic Stokes shift) and gains in signal strength. To quantify these changes, the SE contributions to the transient difference spectra were identified by overlaying the transient signal with the spontaneous emission (Figure 5a). From these contributions, the average emission wavenumbers [$⟨\tilde{\nu }⟩$ , Eq. 1]:

$${\displaystyle ⟨\tilde{\nu }⟩=\frac{\underset{{\tilde{\nu }}_{\mathrm{m}\mathrm{i}\mathrm{n}}}{\overset{{\tilde{\nu }}_{\mathrm{m}\mathrm{a}\mathrm{x}}}{\int }}\frac{{\Delta A}_{SE}(\tilde{\nu },t)}{\tilde{\nu }}\tilde{\nu }\mathrm{d}\tilde{\nu }}{\underset{{\tilde{\nu }}_{\mathrm{m}\mathrm{i}\mathrm{n}}}{\overset{{\tilde{\nu }}_{\mathrm{m}\mathrm{a}\mathrm{x}}}{\int }}\frac{{\Delta A}_{SE}(\tilde{\nu },t)}{\tilde{\nu }}\mathrm{d}\tilde{\nu }}}$$ (1)

Figure 5.

Analysis of the early behaviour of the stimulated emission. a) Comparison of the transient absorption spectrum with the steady‐state fluorescence spectrum. An example transient absorption spectrum at a long (24 ps) delay time compared to the relaxation times is shown. For comparison, the fluorescence spectrum F($\tilde{v}$ ) was redrawn according to F($\tilde{v}$ )/$\tilde{v}$ ³, and the transient absorption ΔA($\tilde{v}$ ) according to ΔA($\tilde{v}$ )/$\tilde{v}$ to arrive at the transition dipole representation. Then, the rescaled fluorescence (green line) was inverted and shifted to overlay the transient absorption spectrum. The marked area is proportional to the integral I(24 ps). b) The spectrally integrated SE signal [Eq. (3)] is plotted as a function of time (shown in black). This signal (I(t)) was fitted with the trial function (grey line) shown in Equation (4); the instrument‐response function was approximated by a Gaussian function (with FWHM∼0.15 ps, plotted in blue). The evolution of the average emission wavenumber (⟨$\tilde{v}$ ⟩, dark red data points) is also shown with the respective double exponential fit (red line).

as well as the spectral integrals [Eq. (2)] were computed:

$${\displaystyle I\left(t\right)=\underset{{\tilde{\nu }}_{\mathrm{m}\mathrm{i}\mathrm{n}}}{\overset{{\tilde{\nu }}_{\mathrm{m}\mathrm{a}\mathrm{x}}}{\int }}\frac{{\Delta A}_{SE}(\tilde{\nu },t)}{\tilde{\nu }}\mathrm{d}\tilde{\nu }}$$ (2)

The integration range ${\tilde{\nu }}_{min}-{\tilde{\nu }}_{max}$ was selected in such a way that only the SE contributes to the integral (Figure 5a).

The time dependencies of these quantities are plotted in Figure 5b. As a biexponential fit shows, the average emission wavenumber ($⟨\tilde{\nu }⟩$ ) approaches its steady state value with a time constant of 0.13 ps and 2.07 ps. Values of similar magnitude were determined in studies on dielectric relaxation of toluene. ^[18] Thus, this shift seems to be, at least in parts, due to dielectric relaxation. The structures changes seen in the quantum chemical calculations are also expected to contribute to the dynamic shift and might also occur on this timescale. The plot of integral (I(t)) reveals an increase of the SE signal with time. As the SE integral is proportional to the oscillator strength this implies a non‐Condon behaviour. To separate the additional rise due to non‐Condon behaviour from the instrumentally limited one, the time trace was fitted with the following trial function [Eq. 3]:

$${\displaystyle S\left(t,{\tau }_{\mathrm{c}\mathrm{c}}\right)=\mathrm{I}\mathrm{R}\mathrm{F}\left({\tau }_{\mathrm{c}\mathrm{c}}\right)\otimes \lfloor A_1{e}^{\left[-\frac{t}{{\tau }_1}\right]}+A_2{e}^{\left[-\frac{t}{{\tau }_2}\right]}\rfloor }$$ (3)

IRF(τ _CC)$\otimes$ stands for the convolution with the instrumental response function which was approximated by a gaussian function with τ _CC=0.15 ps (FWHM), A ₁<0 is the amplitude associated with the rise time τ ₁, and A ₂>0 is the signal after termination of the rise with the time constant τ ₂ that was set to infinity, A ₁+A ₂ equals the initial signal. The fit yields a rise time of 0.6 ps, an initial signal A ₁+A ₂ of 2.3 and final one A ₂ of 4.2. Thus, an increase in oscillator strength by a factor of around 2 is measured. The quantum chemical computation predicted only a smaller increase (a factor of 1.3), although it depends on the configuration selected (for α=123.9° a factor of 2.1 was estimated).

From ∼10 ps until ∼3 ns, the UV/Vis transient absorption signal is essentially constant. This is in line with the reported fluorescence lifetime of 21 ns. ^[9] So, from ∼10 ps onwards signals stem from the relaxed S₁ state of TAA‐DCN. A spectral signal recorded after ∼10 ps thus constitutes the short wavelength part of the S₁→S _n spectrum. According to the quantum chemical computations described above, the long wavelength part of this spectrum ought to be centred around 1000 nm. This region is accessible by inspection of the early part of a nanosecond NIR experiment (Figure 6).

Figure 6.

Nanosecond transient absorption of TAA‐DCN in deoxygenated toluene (c=0.2 mM) for the early‐nanosecond range after excitation at 355 nm covering the NIR. The contour plot in the centre gives an overview. Time traces at indicated detection wavelengths are plotted on the left. Transient difference spectra at early and later times are shown on the right.

Around time zero, a difference absorption is observed which gradually increases from 1600 to 1000 nm. Below 1000 nm the signal sharply rises pointing to an absorption band peaking around 900 nm (not covered by our instrument). On the timescale of 10 ns the signal is seen to arise throughout most of the spectral region covered. The timescale of the rise matches the S₁ lifetime. We therefore assign the time zero spectrum to the S₁ state. By combining the “late” femtosecond UV/Vis and the “early” nanosecond NIR signals the experimental S₁→S _n spectrum plotted in Figure 7 was obtained. For comparison, the quantum chemical stick spectrum (oscillator strengths as a function of wavelength) were convoluted with Gaussians. There is good agreement between measured and computed spectra. In particular, the lowest four transition S₁→S_2,3,4 can be identified. The S₁→S₂ and S₁→S₃ transitions are among CT states and the S₁→S₄, centred at around 800 nm, is a CT→LE transition.

Figure 7.

Comparison of the measured and computed S₁→S _n spectra of TAA‐DCN in toluene.

To obtain the T₁→T _n spectrum, nanosecond UV/Vis spectroscopy and NIR probing were conducted. Hereby, the focus was laid on the microsecond time range, as the T₁ lifetime was reported to be 30 μs. ^[9] In the respective experiment, a broad excited state absorption extending from 400 to 1600 nm is observed at time zero (Figure 8). Peaks are discernible at 460, 710, and 1200 nm. The signal decays to zero on the timescale of 10 μs. A global fit analysis yields a time constant τ _T of 19 μs for this decay. This is shorter than the value of 30 μs derived from the decay of the delayed fluorescence. ^[9] We, thus, investigated the dependence of the lifetime τ _T on the TAA‐DCN concentration (see in Figure S5). The experiment reveals a linear increase of 1/τ _T with [TAA‐DCN]. Such “self‐quenching” is often observed for triplet states. A linear fit of the behaviour yields a quenching constant k _sq of 9.7 ⋅ 10⁷ M⁻¹s⁻¹ and a lifetime $\tau \genfrac{}{}{0pt}{}{\mathrm{T}}{0}$ for infinite dilution of 30.1 μs. This value is very close the one reported early ^[9] indicating that in both experiments the same state, namely, the T₁ one is monitored.

Figure 8.

Nanosecond transient absorption of TAA‐DCN in deoxygenated toluene (c=0.2 mM) after excitation at 355 nm covering from visible to NIR for the microsecond range. The contour plot in the centre gives an overview. Time traces at indicated detection wavelengths are plotted on the left. Transient difference spectra at early and later times are shown on the right.

The experimental spectrum of this state is compared with the quantum chemical prediction in Figure 9. Again, a good agreement between measured and computed spectra is observed. Also, the lowest four transitions T₁→T_2,3,4 can be identified. The T₁→T₂ transition is among CT states and the T₁→T₃, centred at around 1100 nm, is a CT→LE transition.

Figure 9.

Comparison of the measured and computed T₁→T _n spectra of TAA‐DCN in toluene.

Conclusions

rISC and ISC transitions in TADF emitters can be mediated by vibronic mixing with higher excited singlet (S _n>1 ) and triplet (T _n>1 ) states into lower‐energy excited states (S₁ and T₁). According to perturbation theory, the energy gaps between the lower‐energy excited states and higher ones ought to affect this mixing. ^[8] Here, it has been shown that, for a prototypical TADF emitter, the respective gaps are accessible experimentally by time‐resolved NIR spectroscopy. The respective transition energies and oscillator strengths were well reproduced by quantum chemical computations. In the singlet, the lowest ¹CT→¹LE (S₁→S₄) transition was located at around 800 nm and in the triplet, the lowest ³CT→³LE (T₁→T₃) is at 1100 nm. In the future, we will shift these transitions by suitable substitutes and/or solvents and investigate the impact of these shifts on the rate constants k _ISC and k _rISC.

Experimental Section

Sample and common conditions: The synthesis of TAA‐DCN has been described previously, ^[9] however, here we have transposed the BLEBS sequence (bromo‐lithium‐exchange‐borylation‐Suzuki) ^[19] providing a higher‐yielding access to the target molecule.

4‐Bromo‐3‐methyl‐N,N‐diphenylaniline (TAA‐Br)) was converted by bromine‐lithium exchange at −78 °C in THF to the lithiated derivative, which was reacted with trimethylborate to give the corresponding boronate complex. By subsequent Suzuki cross‐coupling the boronate complex was reacted with 2‐iodoterephthalnitrile (DCN‐I) in the presence of potassium tert‐butoxide and catalytic amounts of Pd(PPh₃)₄ as a catalyst at 80 °C for 18 h to give after isolation and purification TAA‐DCN (Scheme 2).

Scheme 2.

Synthesis of TAA‐DCN by BLEBS in a one‐pot fashion.

TAA‐DCN solutions were prepared in toluene (≥99.7 %, from Sigma‐Aldrich). All measurements were carried out at room temperature (20 °C). For the steady‐state absorption and fluorescence measurements fused silica cells (from Hellma analytics) of 1 cm path length were used. To avoid contributions of photo‐products the sample solutions were pumped through fused silica flow cell in the femtosecond transient absorption measurements (1 mm path length) and in the nanosecond transient absorption measurements (5 mm path length in pump and 10 mm path length in probe direction). For the nsTA measurements all solutions were deaerated by purging with nitrogen (Air Liquide), and to prevent changes in the concentration, gases were saturated with the solvent.

Steady‐state spectroscopy: Steady‐state absorption was carried out with Lambda 19 spectrometer from PerkinElmer. Steady‐state fluorescence was performed on FluoroMax‐4 from Horiba Scientific. All fluorescence spectra were corrected for the solvent background and for the spectral sensitivity of the instruments. For fluorescence measurements the excitation was tuned to 380 nm, and the solutions were prepared to have an absorption below 0.05 in 1 cm cell at the excitation.

Femtosecond transient absorption spectroscopy: The setup was described in detail elsewhere. ^[20] The pump pulses of 400 nm were obtained from the output of a Ti : Sa laser amplifier system (Coherent Libra, with 1 kHz of repetition rate and pulse duration of 100 fs (FWHM)) by second harmonic generation. The energy per pulse amounted to ≈1 μJ. The probe pulse was obtained by supercontinuum generation in CaF₂. At the sample location, the diameter of the pump beam was about 160 μm, and the probe 100 μm. The relative polarization of pump and probe pulses was set to the magic angle. Raw data were corrected for the chirp and the solvent contribution. Absorptions of the sample solutions at 400 nm were adjusted to 0.7 in 1 mm cell.

Nanosecond laser flash photolysis: Nanosecond transient absorption data were performed with a laser flash photolysis spectrometer LP980 from Edinburg Instruments. The excitation (pump pulse) was obtained by frequency tripling (355 nm) of the output of a Nd:YAG laser (Spitlight 600, InnoLas, Germany) with 5 Hz of repetition rate and 12 ns (FWHM) of pulse duration. The average pulse energy amounted to ≈5 mJ. The probe light was obtained from a pulsed xenon flash lamp (Osram XBO 150 W/CROFR). After passing the sample cell at a right angle geometry with respect to the pump, the transmitted probe light was dispersed by a grating monochromator and detected by two different detectors to cover the UV/Vis (photomultiplier Hamamatsu PMT‐900) and the NIR (photodiode Hamamatsu InGaAs) spectral range. To obtain the transient signals, time traces were collected to cover the visible and near‐infrared spectral range in different steps and averages to gain the best signal to noise ratio. Absorptions of the sample solutions at 355 nm were adjusted to 0.7 in 1 cm cell.

Data analysis: The oscillator strength was determined according to ref. [17], in which the integral covered the lowest absorption coefficient band. To this end, the absorption spectrum was decomposed in Gaussian components and only the lowest in energy entered in the analysis. To obtain the 0–0 excitation energy (E ₀₀) and Stokes shifts, fluorescence spectra were converted from constant wavelength (λ) to constant wavenumber ($\tilde{v}$ ) bandpass by multiplying with λ². ^[21] Then, the absorption coefficient was rescaled according to ϵ($\tilde{v}$ )/$\tilde{v}$ , and fluorescence spectrum according to F($\tilde{v}$ )/$\tilde{v}$ ³ to arrive at the transition dipole representation. ^[17] Finally, the normalized corrected absorption and fluorescence spectra were plotted and the E ₀₀ was obtained from their intersection.

The transient absorption spectra (ΔA(λ,t)) obtained here are function of the probe wavelength (λ) and the time delay (t) between pump and probe. To retrieve the time and wavelength dependencies for these measurements two approaches were used. In the first one, the stimulated emission spectra signal was analysed for the early times. For that, the transient absorption ΔA($\tilde{v}$ ,t)/$\tilde{v}$ was compared to the steady‐state fluorescence. The first step herein was to rescale the fluorescence spectrum according to F($\tilde{v}$ )/$\tilde{v}$ ³ to arrive at the transition dipole representation. ^[23] For comparing both spectra, the rescaled fluorescence was then flipped and shifted. The spectral range in which both spectra overlaps were used latter to define the limits of the integrals analysed. And in the second approach, the nanosecond transient absorption data were analysed by global multi‐exponential fit function [Eq. 4]: ^[24]

$${\displaystyle \Delta A(\lambda ,t)=IRF\otimes \sum _{i=1}^n\Delta A_i\left(\lambda \right)e^{-\frac{t}{{\tau }_i}}}$$ (4)

which was convoluted with the response function (IRF) of the instrument. Here, the IRF was approximated by a Gaussian with a FWHM of 12 ns. The fit yields time constants τ_i and the respective decay associated difference spectra ΔA _i .

Quantum chemical computations: Electronic ground‐state geometries of the TAA‐DCN emitter were optimized with DFT at the ωB97X−D/def2‐TZVP level of theory[ ¹² , ²⁵ ] with ω=0.15 Bohr⁻¹ (after optimal tuning procedure in vacuo) and including implicit toluene solvation through the polarizable continuum model (PCM) ^[26] implemented in Gaussian16. ^[11]

Time‐dependent DFT (TDDFT) ^[27] was used for the optimization of the excited states (Tamm‐Dancoff approximation (TDA) for excited triplet states ^[28] ). Analytic harmonic vibrational frequencies were computed with Gaussian16.

Vertical and adiabatic excitation energies as well as optical electronic properties were calculated using the DFT/MRCI method.[ ^10b , ^14a , ^14b ] Up to 20 excited states in the singlet and triplet manifold (in case of ESA spectra 40 roots) employing closed‐shell BH‐LYP ^[29] orbitals as the one‐particle basis. The parametrization of the Hamiltonian reported by Lyskov et al. ^[14c] (DFT/MRCI‐R2016) was employed for the tight configuration selection threshold of 0.8 E_h, which is specially designed for large multichromophore systems.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Haselbach W., Kaminski J. M., Kloeters L. N., Müller T. J. J., Weingart O., Marian C. M., Gilch P., Nogueira de Faria B. E., Chem. Eur. J. 2023, 29, e202202809.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.