Intramolecular Dimerization Quenching of Delayed Emission in Asymmetric D-D'-A TADF Emitters.

Understanding the excited-state dynamics and conformational relaxation in thermally activated delayed fluorescence (TADF) molecules, including conformations that potentially support intramolecular through-space charge transfer, can open new avenues for TADF molecular design as well as elucidate complex photophysical pathways in structurally complex molecules. Emissive molecules comprising a donor (triphenylamine, TPA) and an acceptor (triphenyltriazine, TRZ) bridged by a second donor (9,9-dimethyl-9-10-dihydroacridin, DMAC, or phenoxazine, PXZ) are synthesized and characterized. In solution, the flexibility of the sp3-hybridized carbon atom in DMAC of DMAC-TPA-TRZ, compared to the rigid PXZ, allows significant conformational reorganization, giving rise to multiple charge-transfer excited states. As a result of such a reorganization, the TRZ and TPA moieties become cofacially aligned, driven by a strong dipole-dipole attraction between the TPA and TRZ units, forming a weakly charge-transfer dimer state, in stark contrast to the case of PXZ-TPA-TRZ where the rigid PXZ bridge only supports a single PXZ-TRZ charge transfer (CT) state. The low-energy TPA-TRZ dimer is found to have a high-energy dimer local triplet state, which quenches delayed emission because the resultant singlet CT local triplet energy gap is too large to mediate efficient reverse intersystem crossing. However, organic light-emitting diodes using PXZ-TPA-TRZ as an emitting dopant resulted in external quantum efficiency as high as 22%, more than two times higher than that of DMAC-TPA-TRZ-based device, showing the impact that such intramolecular reorganization and donor-acceptor dimerization have on TADF performance.

Introduction

Organic molecules that yield thermally activated delayed fluorescence (TADF) have evolved to become the third-generation materials for organic light-emitting diodes (OLEDs).[1] Unlike phosphorescent emitters, TADF molecules can harvest triplet excitons by a reverse intersystem crossing (rISC) mechanism between the triplet charge-transfer states (3CT) and the singlet charge-transfer states (1CT) mediated by vibrionic coupling between 3CT and a local excited triplet state (3LE) when the energy gap between all three states becomes small, <100 meV.[2−4] One of the key parameters in achieving extremely small singlet–triplet splitting (ΔEST) is the use of conformational twisting between donor (D) and acceptor (A) to minimize the overlap between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO).[5−7] Beyond the subtle manipulations on the degree of intramolecular through-bond charge transfer, recent interest in through-space charge transfer between D and A moieties, mediated by strong intramolecular π–π dipolar interactions between D and A, has been motivated by the potential of increasing the photoluminescence quantum yield.[8,9] In a conjugated system, the D and A can be arranged cofacially, in close proximity, using a noncoplanar molecular scaffold.[10] This approach has also led to the design of nonconjugated polymers with spatially separated pendant D and A.[11,12] Despite the growing interest of intramolecular through-space charge transfer as a means to TADF, there have been few detailed studies of the excited-state dynamics and molecular conformations of such systems. Here, we investigate the excited-state dynamics of new molecules configured with asymmetrical donor–donor′–acceptor (D–D′–A) architectures, where the electron-donor (D) moiety (triphenylamine, TPA) is bridged through a donor’ (D′) such as 9,9-dimethyl-9-10-dihydroacridin (DMAC) or phenoxazine (PXZ) to link to the electron acceptor (A) (triphenyltriazine, TRZ). The former is referred to as DMAC–TPA–TRZ and the latter is PXZ–TPA–TRZ, as shown in Figure . Both DMAC and PXZ moieties are popular electron-donating groups that are used to construct high-performance TADF emitters.[13−18] In a previous study of a linear D–A–D′ system, the near orthogonality of both D and D′ units separated by the A ensured strong decoupling of the electrons on the D and A in the CT state.[19] In this study, the secondary donor TPA and TRZ acceptor are placed in close proximity, which allows us to investigate the detailed excited-state dynamics in the presence of potential through-space D/A interactions. Here, we observe that DMAC–TPA–TRZ exhibits multiple CT states, one of which, between TPA and TRZ, is mediated by strong dipole–dipole coupling between the D and A, which drives an activated cofacially intramolecular dimerization, giving through-space charge transfer resulting in an excited-state intramolecular dimeric CT state. However, we find that this low-energy CT state has a high-energy dimer local triplet state that introduces a large singlet–triplet gap that effectively quenches rISC. In sharp contrast, this type of D/A interaction does not occur in PXZ–TPA–TRZ, which was rationalized as a result of the rigidity of the bridging PXZ moiety.

Figure 1

Chemical structures of the TADF molecules studied.

Results and Discussion

The synthesis of target molecules PXZ–TPA–TRZ and DMAC–TPA–TRZ is shown in Scheme S1, and the detailed procedures and characterization of all materials are reported in the Supporting Information (SI). To probe the excited states of these complex systems, we replaced triphenylamine (TPA) with a phenyl group (Ph) to make further two compounds, PXZ–Ph–TRZ and DMAC–Ph–TRZ (Figure ), which were synthesized from intermediates PXZ–Ph and DMAC–Ph.[20] All of the newly synthesized molecules were purified by column chromatography followed by vacuum sublimation before characterization and device fabrication. Single crystals of PXZ–TPA–TRZ, PXZ–Ph–TRZ, and DMAC–Ph–TRZ suitable for X-ray diffraction analysis were obtained by a bilayer (CH2Cl2/hexane) diffusion method. Unfortunately, a single crystal of DMAC–TPA–TRZ could not be obtained. Figure depicts the single crystal structures of PXZ–TPA–TRZ, PXZ–Ph–TRZ, and DMAC–Ph–TRZ. Full crystallographic data are summarized in Table S1 in the SI and Figure S1. Ortho-substituted donors of TADF molecules commonly exhibit folded conformations because of the large steric interaction between the ortho-substituted group and acceptor.[21,22] The folded conformation in the electron-donating moiety was also observed in PXZ–Ph–TRZ and DMAC–Ph–TRZ. The X-ray structures of PXZ–Ph–TRZ and DMAC–Ph–TRZ (Figure ) indicate buckled conformations of the donating units PXZ and DMAC. The PXZ of PXZ–Ph–TRZ and DMAC of DMAC–Ph–TRZ exhibit a distorted boat configuration and are folded relative to the phenylene bridge plane, creating dihedral angles of 115.4 and 127.2°, respectively. In addition, the dihedral angles between the phenyl group and its donor of PXZ–Ph–TRZ and DMAC–Ph–TRZ are calculated to be 41.4 and 53.0°, respectively. As observed in the PXZ–Ph–TRZ and DMAC–Ph–TRZ crystals, the boat configuration leads to a highly twisted D–A configuration, where the nitrogen lone-pair electrons of the donor unit align favorably for better π-conjugation with the TRZ rather than with the phenylene ring of the donor unit. Thus, the N–C (TRZ) bond in PXZ–Ph–TRZ and DMAC–Ph–TRZ is shorter than the N–C (PXZ and DMAC) bond, whereas the N–C (TRZ) bond in PXZ–TPA–TRZ is longer than the N–C (PXZ) bond, as indicated in Figure . In the PXZ–TPA–TRZ case, the buckled conformation of the PXZ unit is not observed by X-ray analysis. Instead, the central ring of PXZ in PXZ–TPA–TRZ is practically planar and exhibits a highly twisted D–A conformation with a dihedral angle of 75.6° between PXZ and TRZ core (Figure S1). The TRZ lies in a quasi-equatorial configuration[23,24] with respect to the PXZ, with the N–C bridging bond tilted by ∼36.5° with respect to the PXZ plane, resulting in the staggered arrangement of TPA and TRZ at a shortened C–N distance of 2.931 Å. In addition, the dihedral angle between the triphenylamine and PXZ of PXZ–TPA–TRZ is calculated to be 57.6°.

Figure 2

Molecular structures of (a) PXZ–TPA–TRZ, (b) PXZ–Ph–TRZ, and (c) DMAC–Ph–TRZ with thermal ellipsoids drawn at a 50% probability level.

Figure a shows the normalized absorption and emission spectra of PXZ–TPA–TRZ and PXZ–Ph–TRZ in methylcyclohexane (MCH). Optical transitions at ∼270 and ∼350 nm can be attributed to the TRZ and PXZ moieties, respectively, while TPA exhibits an absorption peak at ∼320 nm. The emission spectra of both PXZ–Ph–TPZ and PXZ–TPA–TRZ are similar except the latter is red-shifted. Figure b depicts the normalized absorption and emission spectra of DMAC–TPA–TRZ and DMAC–Ph–TRZ in MCH. Both absorption edges of PXZ–TPA–TRZ and DMAC–TPA–TRZ are red-shifted with respect to TPA-free molecules, indicating increased conjugation in the π system of PXZ and DMAC (Figure S2). The emission spectra of each material in nonpolar MCH reveal a slightly structured band, suggesting a mixed locally excited (1LE) and 1CT character of the state. The increase in photoluminescence (PL) intensity in degassed solvents (Figure S3) indicates a contribution from the triplet states (Table S2). In polar solvents such as toluene, the emission loses all its structure, indicative of an increased CT character. A strong positive solvatochromic shift of the onset of emission band in polar solvents indicates strong CT character of the transition. The shift of the emission-onset energy for each material between MCH and toluene is largest in PXZ–Ph–TRZ (0.19 eV), followed by PXZ–TPA–TRZ (0.14 eV) and then by DMAC–TPA–TRZ (0.11 eV) with DMAC–Ph–TRZ showing the smallest shift (0.06 eV). Incorporating TPA in PXZ–Ph–TRZ decreases the induced change in dipole moment (degree of charge transfer) compared to that in PXZ–Ph–TRZ, indicating a stronger stabilization of the charge transfer in PXZ–TPA–TRZ in a nonpolar environment. However, it increases the degree of charge-transfer character significantly for DMAC–TPA–TRZ. DMAC–Ph–TRZ is seen to have the weakest CT character supported by it having the smallest integrated intensity ratio between degassed and aerated solvent in MCH, which only increases slightly in toluene (Table S2). Interestingly, in MCH, DMAC–TPA–TRZ singularly exhibits two very distinct emission bands centered at 420 and 525 nm. In toluene, the 525 nm band red-shifts only slightly but the intensity is greatly reduced, indicative of a CT state with a near-complete one-electron transfer from D to A, whereas the emission centered at 420 nm red-shifts far more, compared to its position in MCH concomitant with weak CT in MCH. DMAC–Ph–TRZ in comparison shows this larger red shift of its single band from 430 nm in MCH to 480 nm, indicating stabilization of the DMAC–TRZ CT state with polarity.[25] From this, we make an initial identification of the two CT states in DMAC–TPA–TRZ to be a DMAC–TRZ CT state at 420 nm and a TPA–TRZ CT state at 525 nm (taking the MCH data). The weak emission band between 350 and 400 nm is ascribed to donor 1LE emission of the DMAC/PXZ moieties, which are initially excited with 325 nm (Figure S4).

Figure 3

UV–vis and PL spectra of 10 μg mL–1 in MCH solution of (a) PXZ–TPA–TRZ and PXZ–Ph–TRZ, and (b) DMAC–TPA–TRZ and DMAC–Ph–TRZ.

To elucidate the nature of the DMAC–TPA–TRZ dual emission, time-resolved fluorescence spectroscopy in MCH was carried out using time-gated emission measurements using an iCCD-based spectrometer having a time resolution better than 1 ns.[1] At room temperature, the spectrum is complex. The blue band at ca. 425 nm decays very rapidly, within 10 ns, as shown in Figure a. This band is neither D nor A emission (see Figure S5), confirming it most likely to be a DMAC–TRZ CT state. The short lifetime indicating it being quenched by intramolecular energy transfer to the lower-energy CT state. Inspection of the lower-energy emission band indicates the presence of two emitting CT components, centered at ca. 485 nm and ca. 575 nm, both being present from very early times. The 485 nm component red-shifts over the first 200 ns, reaching 525 nm. The lowest-energy CT band has a much shorter lifetime and shows little or no red shift either in time or with increasing solvent polarity, Figure S3. By deduction, we assigned this lowest-energy state to arise from the TPA–TRZ D A pair but having weak CT character. At 240 and 200 K in Figure b (and Figure S6), this lowest-energy TPA–TRZ species is not observed. Further, cooling the MCH solution below its freezing point suppresses the green 575 nm emission completely and hinders the formation of the 485 nm CT state, as seen in Figure c. Given that the low-energy excited states are not observed at low temperature, the lack of quenching of the 425 nm band at low temperatures indicates that the formation of these low-energy bands requires thermal activation. This would be in line with excited states that form after the large geometric reorganization of the molecule.[26]

Figure 4

Time evolution of the spectral shape of DMAC–TPA–TRZ in MCH (a) at room temperature, (b) at 240 K, and (c) 120 K (frozen). Notice the disappearance of the second peak when the solution is frozen. (d) Normalized integrated intensity at different temperatures.

The disappearance of the lowest-energy TPA–TRZ species component coincides with the appearance of DF in the time-resolved emission of DMAC–TPA–TRZ (Figure d). At room temperature, the decay curve is single exponential, while at 240 and 200 K, DF is also observed. The spectra observed at 120 K, Figure c, are very similar to those found in DMAC–Ph–TRZ, Figure S8, at early times, indicative of the DMAC–TRZ early-time CT state. In frozen solution, it is clear that this state relaxes very slowly, red-shifting over 0.25 eV in 200 ns. This may indicate that this state can relax into another state but the bulky substitution on the DMAC and the confinement in the frozen solution make this process very slow indeed.

Comparing this behavior to our previously reported[27,28] dual emission from axial and equatorial D–A conformations, we propose that the fast decay blue band at 425 nm could be evidence for an axial DMAC–TRZ conformation that is rapidly quenched by the lower-energy TPA–TRZ states.[29]

In zeonex matrix, the lower-energy emission components of DMAC–TPA–TRZ are suppressed as they are in frozen solution (Figure S7), having an integrated emission band very similar to that observed in frozen MCH, indicating that large molecular reorganization is required to form these emissive species. However, the prompt fluorescence during the first few 100 ns (PF) of DMAC–TPA–TRZ in zeonex shows two clear emitting species, one centered at ca. 450 nm and the second at 510 nm, as shown in Figure a. The lowest-energy feature has a fast lifetime of only a few hundred nanoseconds, whereas the higher-energy component gives longtime DF but with significantly reduced intensity. This again points to the lowest-energy TPA–TRZ band having a different character to the higher-energy component.

Figure 5

Time-resolved spectra of (a) PF and (b) DF of DMAC–TPA–TRZ at room temperature in zeonex matrix. (c) PL spectra with different excitation wavelengths in MCH and (d) photoluminescence excitation (PLE) and overlay with absorption curve in MCH and toluene.

The CT bands of PXZ–TPA–TRZ and DMAC–Ph–TPZ in zeonex display an apparent spectral shift of ca. 0.12 eV over 100 ns, whereas PXZ–Ph–TPZ shows no spectral shift in time at all (Figure S8). This spectral shift is, however, an effect of D–A dihedral angle heterogeneity giving large lifetime dispersion of the CT emission.[30,31] The higher-energy CT states decay faster than the lower-energy species due to their increased LE character, giving an apparent time-dependent spectral shift. The lack of emission shift in PXZ–Ph–TPZ demonstrates that it has a very rigid structure that narrows the structural inhomogeneity. This rigidity can be traced back to the large steric interaction between the phenyl group and the donor. The apparent spectral shifts in the DMAC–Ph–TRZ are larger than those in PXZ–Ph–TRZ, implying greater flexibility of the DMAC moiety. The presence of apparent spectral shift over time for PXZ–TPA–TRZ also implies the presence of structural inhomogeneity in a disordered medium. In DMAC–TPA–TRZ, we observe a similar shift in the 485 nm band (MCH) at early times but not the 575 nm band, indicating that the latter has far less heterogeneity and a more rigid configuration. The very large shift of the DMAC–TRZ CT band in frozen solution may indicate very large heterogeneity in this molecule in frozen solution, whereas the TPA–TRZ CT band shows little or no such heterogeneity, Figures a and S6, indicating a much more rigid and well-defined state.

DMAC–TPA–TRZ in zeonex again shows multiple distinct CT emissions occurring on different time scales, as illustrated in Figure b. The prompt emission shows an initial 450 nm band that decays within 50 ns, leaving a low-energy feature centered at 510 nm. This low-energy band does appear to grow in as the high-energy feature decays, again indicating that the species giving rise to the 450 nm band is quenched by the species yielding the 510 nm band. The delayed fluorescence, Figure b, clearly shows two distinct low-energy emissive species, the 510 nm band decaying in microseconds, leaving a band at 480 nm. The emission at 2 μs is the superposition of two bands, but at 200 ns, there is no emission at 480 nm, which may indicate that the low-energy species converts to the higher-energy species which then emits at 480 nm.

The CT bands of DMAC–TPA–TRZ are found to be excitation-energy-dependent, but counterintuitively, the low-energy CT emission bands disappear as the excitation energy decreases, Figure c. Surprisingly, lower-energy excitation gives rise only to the higher-energy blue emission, whereas the lower-energy green emission component requires excitation at higher energy. Such behavior is observed in MCH, toluene, and 2-methyltetrahydrofuran solutions (Figure S9). This indicates that these low-energy species require excess energy to overcome a formation energy barrier. This then explains their greatly reduced contribution in solid zeonex. Photoluminescence excitation (PLE) spectra from DMAC–TPA–TRZ, monitoring the emissions at 445 and 505 nm in MCH and at 450 and 525 nm in toluene were measured. These wavelengths correspond to the peak emission of each band in the respective solvent. Figure d shows this PLE and the absorbance of DMAC–TPA–TRZ in both solvents. The absorbance edge red-shifted with increasing solvent polarity from MCH to toluene while the locations of the peak absorbance are the same. The PLE curves for the blue and green components are very different. The PLE of the green component follows the shape of the absorbance, whereas the blue emission has a red-shifted PLE spectrum in both solvents. The PLE for all other materials (Figure S10) follows the absorbance curves almost exactly. The PLE of both components in DMAC–TPA–TRZ is red-shifted in MCH compared to that in toluene, consistent with the absorbance spectra, indicative of strong n−π* character. From the shape and position of the blue component PLE, which is located at the lower energy side of the absorbance, we assume that it arises from a direct CT absorption, i.e., from an n−π* transition.[30]

We performed a theoretical conformer search and found that DMAC–TPA–TRZ exists in two distinct conformational groups, having either a distorted boat or a planar DMAC configuration. Within the distorted boat configuration, the separation between TPA and TRZ is larger than the planar configuration (Figure S11). Both distorted and planar conformations exist in almost the same proportion, as their Gibbs free energies are the same. For PXZ–TPA–TRZ, the next lowest conformational structure of PXZ–TPA–TRZ differs by 0.6 kcal mol–1, corresponding to a larger separation between TPA and TRZ. We believe that such conformational structures exist in an amorphous host medium, as evidenced in the time-resolved photoluminescence spectroscopy (Figure S8a)

Figure a shows the possible ground-state conformations for DMAC–TPA–TRZ optimized by density functional theory (DFT) using a tuned range-separated LC-ωPBE* functional at cc-pVDZ basis set level (Figure S12). This tuned range-separated LC-ωPBE* functional is used to minimize localization/delocalization error for CT states and was found to be a better predictor for vertical excitation (VE) energy of the singlet state (EVE(S1)) for CT molecules despite the fact that it tends to overestimate EVE(S1);[32,33] for conformer 1 (Figure a), the TRZ is quasi-equatorial with respect to the DMAC,[24,34] with θt tilted by 34° with respect to the DMAC plane, while the tilting angle for TPA is less than 5° in the opposite direction as if TPA and TRZ are cofacially “skewed” away from each other. Conformer 1 has a continuous conjugation, while conformer 2, which is distorted, has a bending angle of 138°. Furthermore, for conformer 2, the TRZ is quasi-axial with respect to the DMAC[24,32] with TRZ and TPA “forked” away from each other with TRZ “pointing” downward by an angle of 68°. The interchromophic distance between TPA and TRZ is far shorter in conformer 1 than 2 with d1 = 3.28 Å, d2 = 5.62 Å (conformer 2) and d1 = 2.89 Å, d2 = 4.56 Å (conformer 1). Figure b shows the HOMO and LUMO levels of conformers 1 and 2 of DMAC–TPA–TRZ. One distinct difference is that there is no continuous electron delocalization for the HOMO level for the case of distorted DMAC. Unlike the LUMO levels of conformer 1, the N–C bridging bond in TRZ is not involved in the frontier orbitals in conformer 2, having a strong effect on the reverse intersystem crossing (rISC). Figure c shows the calculated singlet vertical excitation energy EVE(S1) along with the oscillatory strength of the different conformers, reduced by 0.25 eV from the vacuum values, estimated using the Tamm–Dancoff approximation on time-dependent DFT at the LC-ωPBE*/cc-pVDZ level. Conformer 1 having a flat DMAC contributes significantly to CT states, which correspond to the absorption band edge states, while the first absorbance peak is mainly contributed by the distorted DMAC (conformer 2 and conformer 3). Conformer 3 is a variant of conformer 2 with a smaller interchromophic distance between the TRZ and TPA. Natural transition orbitals (NTOs) are used to shed additional light on absorption and emission.[35] The corresponding NTOs for the five lowest-energy states are given in Figure d. For conformer 1, the highest occupied natural transition orbital (HONTO) and the lowest unoccupied natural transition orbital (LUNTO) are clearly dominated by the transition between HOMO and LUMO, in contrast with conformer 2 where the VE transition is not a CT state. If the TPA and TRZ are closer, as in conformer 3, the contribution from the frontier orbitals of the TPA increases, resulting in a red shift in the 1LE state.

Figure 6

(a) DFT-optimized ground-state geometry of DMAC–TPA–TRZ conformers, (b) frontier orbitals of corresponding conformers, (c) calculated vertical excitation energy and respective oscillator strength for different conformers reduced by 0.25 eV overlaying with the absorption curve of DMAC–TPA–TRZ in MCH, and (d) HONTO and LUNTO of the lowest vertical excitation singlet states for different conformers.

From Figure , it is clear that the 1CT state, coming from conformer 1, results in the emission band at ∼450 nm, whereas excitation in the 1LE states of conformer 2 results in the 1CT-like emission band at ∼500 nm with a large Stoke shift, indicating pronounced geometric distortion in the excited adiabatic geometry. The question then arises, how does the vertical excitation of the lowest 1LE state transform into the lowest-energy emissive state of the system having a geometric distortion?

To elucidate this transformation, we mixed equal amounts of TPA and TRZ (units) by weight in toluene. We observed a new emission band occurring at ∼500 nm, not present in pure solutions of either molecule, which decreased in intensity with decreasing concentration, as shown in Figure a, indicating emission from a TPA/TRZ excited-state complex. Normally, exciplexes do not readily form in solution as the intermolecular distance is large compared to the solid state.[36] The formation of a TPA/TRZ complex in solution thus indicates a high stabilization energy[37−39] between TPA and TRZ. The emission peak also shifts with increasing solvent polarity, indicating a state with large excited-state dipole moment (Figure S13). Interestingly, whereas the emission from TPA is completely suppressed in chloroform, diluting a mixture of TPA/TRZ to 10 μg mL–1 in chloroform has no effect in suppressing the complex emission, suggesting that very strong stabilization can be conferred in chloroform. We also measured the PLE and absorption of the diluted TPA/TRZ mixture. The absorption curve is not a superposition of the absorption curves of TPA and TRZ; rather, a new absorption band appeared at 350–400 nm, peaking at 365 nm, corresponding to the maximum PLE emission, as illustrated in Figure b. This indicates that the interaction between TPA and TRZ occurs in the ground state, i.e., a physical heterodimer state, not an exciplex.[40]

Figure 7

(a) PL of TPA/TRZ mixture in toluene with different concentrations, (b) PLE and absorption of the dilute mixture in chloroform, (c) reduced density gradient mapping for conformer 1 of DMAC–TPA–TRZ at ground state, (d) optimized excited geometries with d1 = 2.81 Å, d2 = 3.73 Å in conformer 2 of DMAC–TPA–TRZ along with their HONTO and LUNTO as compared to those of conformer 1 for vertical fluorescence.

This shows the strong possibility of such an interaction between the TPA and TRZ in DMAC–TPA–TRZ and PXZ–TPA–TRZ if enough geometric rearrangement can occur in the molecule. Given that this state is only observed in DMAC–TPA–TRZ where the DMAC can be distorted to allow large geometric rearrangement and that frozen solution or a solid host matrix severely hinder the formation of this state, we can now ascribe the lowest emissive state in DMAC–TPA–TRZ as an intramolecular TPA–TRZ dimer, having partial CT character.

Noncovalent interactions can be used to visualize such an interaction in three-dimensional real space using a reduced density gradient, as shown in Figure c.[41,42] It is clear that there is a van der Waals interaction between the connecting aromatic rings of TPA and TRZ to the DMAC. The same behavior can be seen in PXZ–TPA–TRZ with the interaction strength reduced when the TPA and TRZ are further away from each other (Figure S14). Minimizing the excited-state geometry of conformers 1 and 2 of DMAC–TPA–TRZ gives a completely different geometry with respect to the ground state. The θt is reduced from 34° at ground state (Figure a) to 7.6° at the adiabatic geometry for conformer 1, while θt changes from 68° to less than 1° in conformer 2. In conformer 2, the TPA and TRZ moieties become cofacial with each other, as shown in Figure d. The large screwing motion for conformer 2, to bring TPA and TRZ to be cofacial with each other, results in the large Stoke shift in emission and dimerization forming new emission band at lower energy. The HONTO and LUNTO are now located predominantly at TPA and TRZ, respectively.

Figure a shows the optimized ground-state geometry of PXZ–TPA–TRZ. The arrangement of TPA and TRZ is very similar to conformer 1 of DMAC–TPA–TRZ, including the frontier electron distribution. However, the TPA and TRZ moieties are not cofacial in the adiabatic excited-state geometry, as shown in Figure b, and the θt is reduced from 56 to 15°. We rationalized this through the rigidity of the PXZ moiety. The HONTO and LUNTO now located on PXZ and TRZ, respectively, as if TPA does not participate in the CT emission. Although the HOMO and LUMO of DMAC–Ph–TRZ and PXZ–Ph–TRZ are spatially separated (Figure S15), from the NTO, the vertical fluorescence state of DMAC–Ph–TRZ shows strong mixing of 1LE state, hence reducing its CT character, as seen in Figure c. This is consistent with the small energy onset shift in DMAC–Ph–TRZ observed between MCH and toluene, an indicative of a strong 1LE character. This is expected to increase the electron exchange energy, introducing a larger energy gap between 3CT and 3LE, while PXZ–Ph–TPZ shows a clear CT character (Figure d) and it is expected to have a small energy gap.[2]

Figure 8

(a) DFT-optimized ground-state geometry of PXZ–TPA–TRZ and the frontier orbitals, (b) adiabatic geometry of PXZ–TPA–TRZ and the HONTO and LUNTO of the lowest vertical fluorescence. HONTO and LUNTO of vertical fluorescence for (c) DMAC–Ph–TRZ and (d) PXZ–Ph–TRZ.

To correlate these theoretical results with the molecular photophysics and TADF behavior, time-resolved emission measurements at room temperature in zeonex were again carried out, as shown in Figure a. Both PXZ–TPA–TRZ and PXZ–Ph–TRZ display clear transition from prompt to DF emission at ∼100 ns with the prompt emission having two decay lifetimes of 2.93 ns (1LE from donor) and 7.83 ns (1CT) for PXZ–TPA–TRZ and a single prompt exponential decay of 11.6 ns (1CT) for PXZ–Ph–TRZ, while the delay CT lifetimes are 10.6 and 7.99 μs for PXZ–TPA–TRZ and PXZ–Ph–TRZ, respectively (Figures S16 and S17). This is not surprising as their ΔEST are very small ∼0.03 eV or less. The rISC rates for PXZ–TPA–TRZ and PXZ–Ph–TRZ are determined[5] to be 7.64 × 104 and 1.48 × 105 s–1, respectively (Table S3). However, DMAC–TPA–TRZ shows virtually no DF. The PF decay from DMAC–TPA–TRZ and DMAC–Ph–TRZ can be fitted with two exponential decays, i.e., 6.90 ns (1CT) and 35.4 ns (dimer), and 18.4 ns (1CT) and 73.6 ns (very weak component), respectively. The DF exponential decay of DMAC–Ph–TPZ and DMAC–TPA–TRZ cannot be fitted reliably.

Figure 9

(a) Room-temperature time-resolved fluorescence decay curves in zeonex host and (d) phosphorescence spectra at 80 K for the compounds under study.

To determine the lowest-energy 3LE state of each material, phosphorescence spectra were obtained, as shown in Figure b. Phosphorescent emission from PXZ–TPA–TRZ cannot be seen even at 20 K, implying that the energy gap between 1CT and 3LE is very small and rISC very efficient. The phosphorescence spectra of DMAC–Ph–TRZ closely resembled those of the TRZ moiety, indicating this to be the lowest local 3LE state (ET = 2.76 eV), whereas, for PXZ–Ph–TRZ, the lowest 3LE state originates from the PXZ, as seen in Figure S18. We also determined the triplet state of the TPA/TRZ dimer (1%) in a zeonex matrix. Indeed, the dimer has triplet energy (3DE) at 3.06 eV, while the DMAC–TPA–TRZ has an 3LE of 2.76 eV, as shown in Figure b. The 3DE spectrum is clearly not a superposition of 3LE states of TRZ and TPA (Figure S19). Hence, the dimer (conformer 2) is expected to display no TADF behavior. In solution, the molecules can transform from one conformer to another; thus, the emission from the DMAC–TRZ CT state will be quenched rapidly by the TPA–TRZ dimer state, as can be seen in Figure a,b. As temperature decreases, the viscosity of the solvent increases, increasing the energy barrier for large reorganization required for the dimer to form, so reducing the quenching of the TPA–TRZ CT state, giving rise to weak TADF behavior that can be seen in Figure d.[40] In a restricted space such as in a zeonex matrix, the large reorganization required by conformer 2 is significantly suppressed, resulting in emission dominated by conformer 1, which has a gap between 3CT and 3LE of 0.2 eV, giving rise to weak TADF. The detailed Jablonski diagrams for DMAC–TPA–TRZ along with the summarized relevant energy levels of the materials for DMAC–TPA–TRZ, PXZ–TPA–TRZ, DMAC–Ph–TRZ, and PXZ–Ph–TRZ are given in Figure .

Figure 10

Jablonski diagram for DMAC–TPA–TRZ for conformers 1 and 2 and (b) energy levels of the materials PXZ–TPA–TRZ, PXZ–Ph–TRZ, and DMAC–Ph–TRZ.

These materials clearly indicate that conformational states play a crucial role in rISC and TADF processes. From the photophysics of DMAC–Ph–TRZ, it exhibits a weak CT state having strong 1LE character. Upon incorporating TPA, two new states are created, a TPA–TRZ CT state and as a result of strong dipole–dipole attraction between TPA and TRZ, a ground-state cofacial dimeric species having a partial through-space CT character. These two states are also identified in theoretical calculations as DMAC–TPA–TRZ conformers 1 and 2. From PLE measurements, it is clear though that in the DMAC–TPA–TRZ molecule, the dimer state formation is activated by excess energy to overcome the stiffness of the DMAC. This dimeric state, however, has a high-energy local triplet state that quenches DF emission. Thus, the use of asymmetric donors (or acceptors) in a potential TADF molecular design has to be made with care to avoid this possibility of excited-state reorganization driven by dimer/exciplex forces between D and A. Interactions of this sort can be hindered by a rigid bridging moiety such as PXZ or increasing the intramolecular distance between donor and acceptor moieties.

To study the effects of the possible TPA–TRZ excited state in DMAC–TPA–TRZ-based devices, all of these compounds were made into OLED devices to measure their electroluminescent (EL) properties. This was achieved through using a common device architecture of glass substrate/indium tin oxide anode/MoO3 (1 nm)/TAPC (40–50 nm)/mCP (10 nm)/mCPCN doped with TADF dopants (8 wt %; 20 nm)/3TPYMB (50–55 nm)/LiF (1 nm)/Al (100 nm). Thermogravimetric (Figure S20) and differential scanning calorimetry (Figure S21) analyses were conducted to probe the thermal and morphological properties of DMAC–TPA–TRZ, DMAC–Ph–TRZ, PXZ–TPA–TRZ, and PXZ–Ph–TRZ. In addition, cyclic voltammetry (Figure S22) was employed to study the electrochemical property and determine the HOMO/LUMO energy levels. These data were summarized in Table S4. DMAC–TPA–TRZ-, DMAC–Ph–TRZ-, PXZ–TPA–TRZ-, and PXZ–Ph–TRZ-doped mCPCN films exhibit photoluminescence quantum yields (PLQYs) of 50, 45.3, ∼100, and ∼100%, respectively. The lower PLQY for DMAC–Ph–TRZ compared to PXZ–Ph–TRZ could be the result of an increased nonradiative loss channel due to a larger 3CT and 3LE gap, which reduces the TADF behavior. EL characteristics of OLEDs incorporating these emitters are shown in Figure a–d, with their EL performance parameters being summarized in Table . The EL spectra, ranging from deep blue to green emission, are similar to corresponding PL spectra in both solution and doped films except DMAC–TPA–TRZ. In the latter, in line with the early-time emission measured in zeonex film, a substantial contribution of the emission comes from the dimeric TPA–TRZ excited state, Figure a, along with a contribution from the DMAC–TRZ CT state at ca. 480 nm, i.e., on the blue edge of the EL, as well. The EL is very different compared to DMAC–Ph–TRZ. The effect of the different host material, mCPCN, is shown in Figures S23 and S24, and data are summarized in Table S5. It can be seen that the PL spectra in mCPCN and zeonex are very similar as well. This is in fact in line with other TADF emitters because in dense film, “solid-state solvatochromism” is not possible[43,44] because the host material cannot reorganize around the excited guest molecule to stabilize the excited-state dipole. Thus, the dimer state plays a critical role in the device performance of the DMAC–TPA–TRZ-based devices, along with the large singlet–triplet gap. The devices in general exhibit a relatively low turn-on voltage of ∼2.5 V and low operation voltage. Under a similar device architecture, devices adopting high-PLQY emitters PXZ–TPA–TRZ and PXZ–Ph–TRZ exhibit rather high external quantum efficiencies (EQEs) of up to ∼22–23.3%, while emitters DMAC–TPA–TRZ and DMAC–Ph–TRZ having lower PLQYs give inferior EQEs of 10.5 and 4.7%, respectively. Noticeably, emitters showing the most pronounced TADF characteristics (i.e., PXZ–TPA–TRZ and PXZ–Ph–TRZ) give significantly higher EL efficiencies of up to (22%, 70.2 cd A–1, 78.8 lm W–1) and (23.3%, 72.8 cd A–1, 76.2 lm W–1), respectively. On the other hand, DMAC–TPA–TRZ (that show less pronounced TADF characteristics) exhibit a lower PLQY of 50% and a lower EQE of 10.5%. Finally, although the device based on blue emitter DMAC–Ph–TRZ gives a lower EQE of 4.7% due to its relatively low PLQY of 45.3%, such an EQE indeed is still significantly higher than one would expect from a pure fluorescent emitter of similar PLQY.[45] This suggests a contribution from DF in the EL process even in this compound.

Figure 11

(a) EL spectra, (b) I–V–L characteristics, (c) EQEs, and (d) luminous efficiencies for OLEDs adopting different emitters.

Table 1

Summary of OLED Characteristics of Various Compounds

emitter	ηexta (%) max., 100, 1000 cd m–2	ηcb (cd A–1) max., 100, 1000 cd m–2	ηpc (lm W–1) max., 100, 1000 cd m–2	CIEd	φ (%)
DMAC–TPA–TRZ	10.5, 3.8, 2.1	24.2, 8.8, 4.6	27.2, 5.1, 1.9	(0.20, 0.35)	50
DMAC–Ph–TRZ	4.7, 2.9, 1.2	5.6, 3.5, 1.5	5.9, 1.9, 0.5	(0.16, 0.16)	45.3
PXZ–TPA–TRZ	22.0, 17.1, 10.6	70.2, 54.9, 33.2	78.8, 39.5, 15.7	(0.35, 0.54)	100
PXZ–Ph–TRZ	23.3, 19.3, 12.0	72.8, 59.8, 36.6	76.2, 45.3, 18.0	(0.30, 0.53)	100

External quantum efficiency.

Current efficiency.

Power efficiency.

1931 Commission Internationale de I’Eclairage (CIE) coordinates.

Conclusions

In summary, we developed asymmetrical donor–donor′–acceptors as novel TADF emitters. We show that multiple emission observed in DMAC–TPA–TRZ is the result of two distinct TPA–TRZ conformers and a DMAC–TRZ CT state. Due to the flexibility of sp3 carbon bond in DMAC, excited-state conformational relaxation driven by strong TPA–TRZ dipole–dipole attraction results in the formation of an intramolecular dimer that gives rise to weak through-space charge transfer, in stark contrast to the rigid PXZ, which prevents these states forming in PXZ–TPA–TRZ. This intramolecular dimer state, however, has a high-energy local triplet state that prevents DF, as the singlet–triplet gap (>0.2 eV) is too large to support rISC. By using the more rigid PXZ donor as a bridge, which prevents the dimeric conformation, strong TADF from the PXZ–TRZ CT state yields excellent OLED performance with EQE as high as 23%. Thus, we show that the flexibility of D–A bridging units is very important in emitter design, allowing multiple CT excited states to form. We also find that where strong dipole–dipole attraction between D and A units exists, stable intramolecular dimer species can form (formation may be activated when large molecular reorganization is required), and although such cofacial dimer states can facilitate through-space charge transfer, care must also be taken because the local triplet states of these dimeric species can be rather higher than expected, leading to quenching of rISC because the local triplet cannot easily couple to the dimer CT states.