2021, Vol.50, No.5

Harvesting excited spin-triplet states as light is essential to realize highly efficient electroluminescence (EL) in organic light-emitting devices. In recent years, thermally activated delayed fluorescence (TADF) has attracted much attention as a novel electronic transition process, since it enables harvesting electrically generated triplet energy as EL without the utilization of rare metals such as iridium and platinum. When the energy gap between the excited spin-triplet and spin-singlet states in molecules is small enough to be compared to the environmental thermal energy at room temperature, they exhibit an intense state mixing between them, resulting in highly efficient reverse intersystem crossing from the spin-triplet to the spin-singlet due to the spin allowed transition, and successive light emission as delayed fluorescence from the singlet excited-state. Using molecules exhibiting TADF, internal EL quantum efficiencies of nearly 100%, which is the theoretical limit, have been realized with sophisticated molecular design. Here, we briefly review recent developments of TADF molecules along with the current understanding of spin-flip mechanisms in purely organic molecular systems.

In organic light-emitting diodes (OLEDs), the excited states are formed in emitter layers (EML) containing organic emissive molecules through carrier recombination of both injected electrons and holes, and the successive light emission occurs when the excited molecules undergo the electronic transition from the excited states to the ground state. By maximizing the unique abilities of light-emitting and semiconducting properties of organic materials, high performance OLEDs have been realized.

The very first electroluminescence (EL) phenomenon under AC bias was reported by A. Bernanose in 1955 using a cellophane film containing an organic dye exhibiting fluorescence,1 and successively similar research using organic fluorescence dyes was also carried out by S. Nanba et al. in the late 1950s.2 In 1963, M. Pope et al. firstly demonstrated clear carrier injection type EL from anthracene single crystals upon application of DC bias.3 Following this report, W. Helfrich and W. G. Schneider performed a detailed analysis of the EL process from carrier injection, exciton formation and radiative decay in anthracene single crystals,4,5 providing the basis of the present OLED research. However, due to the several millimeters thickness of the anthracene single crystals, high voltage over 1000 V was required to observe EL. Thus, the research on EL in organic solids remained only an academic interest and was far from commercial use.

During the 1970s and the early 1980s, aiming for low driving voltages, EL research was focused on thin film devices such as vacuum deposited films, spin-coated polymer films and even Langmuir-Blodgett films.6,7 Especially, a sophisticated multilayer structure reported by C. W. Tang and S. A. VanSlyke in 1987 launched the modern age of OLEDs, since the device showed high luminance with satisfactory low driving voltage and rather long device lifetime.8 The multilayer structures composed of donor and acceptor layers could well confine both electrons and holes, leading to efficient exciton formation. Here we note that in the early 1990s the combination of some donor and acceptor molecules suggested the possibility of thermally activated delayed fluorescence (TADF), but no one was aware of this critical issue which was covered with the low efficiency and redshifted emission.9

In developing high-performance OLEDs, maximizing the internal EL efficiency is fundamental, i.e., the fraction of injected charge carriers converted into photons. Since there are virtually no free charge carriers in organic solids without intentional chemical doping, electrons and holes must be injected into the organic layers from external electrodes, i.e., carrier injection. The injected electrons and holes meet each other and recombine on molecules to form excited states also known as excitons. Since the spin-states of injected carriers are random, excited spin-singlet states (singlet excited state: S) and excited spin-triplet states (triplet excited state: T) are directly generated in a ratio of 1:3. In conventional organic molecular systems, the radiative decay transition process from the lowest T state (T1) to the ground state (S0), i.e., phosphorescence, is a spin-forbidden transition due to its small spin-orbit coupling. As a result, 75% of excitons formed as spin-triplet states on the molecules are consumed as heat instead of producing light.

One of the critical selection rules in the electronic transition process in organic molecular systems is the spin conservation law. Electron transitions between states with different electron spin states are strictly forbidden in principle, as mentioned earlier. However, some electronic transitions, such as intersystem crossing and phosphorescence between the spin-singlet and the spin-triplet states, can be observed in many common molecular systems. This indicates that transitions between different spin states are not “strictly forbidden” but rather “partially allowed.”

Based on a perturbation theory, it can be understood that the wavefunction for the spin-singlet is not a purely zero-order singlet state but in reality a mixture of a spin-triplet state through perturbation such as spin-orbit coupling (SOC). In other words, the spin-state of organic molecules that we observe and think of as a spin-singlet (or spin-triplet) is partially mixed with the component of the spin-triplet (or spin-singlet). For the sake of simplicity, up to the first-order perturbation, we can describe the wavefunction as:10

(1)
The numerator in the formula represents the magnitude of the SOC element matrix, and the denominator represents the energy difference between the considered excited spin-singlet and spin-triplet states. In other words, to enhance spin-state mixing in a molecular system, one should 1) increase the SOC matrix element and 2) decrease the energy difference between the spin states, i.e., ΔEST. For example, an organo-metal complex containing a rare metal such as iridium acquires a large SOC matrix element by its internal heavy atom effect and can emit phosphorescence with high photoluminescence quantum yield (PLQY) even at room temperature. Since the excited spin-singlet state of such phosphorescent materials efficiently flips to the excited spin-triplet state by aid of the strong SOC, it is possible to convert all injected charge carriers into electro-phosphorescence.1113 Therefore, most current OLED displays have been manufactured using phosphorescent emitters for colors other than blue. However, in addition to phosphorescent materials being heavy metal complexes requiring expensive rare metals, the molecular design of materials that emit stable blue light is still challenging.

Recently, as an alternative mechanism for exploiting T1 energy as EL, we pioneered an advanced organic luminescent molecular system exhibiting highly efficient thermally activated delayed fluorescence (TADF), a process in which electron spin conversion occurs from the T1 to the lowest singlet excited state (S1) with the help of sufficiently small ΔEST (Figure 1b). Thus, in a molecule exhibiting TADF, the T1 energy can be extracted as fluorescence from S1.

Since TADF proceeds through the result of multiple cycles between intersystem crossing (ISC) from S1 to T1 and the reverse ISC (RISC) from T1 to S1, a long-lived “delayed” fluorescence (DF; emission lifetime on the microsecond scale) is observed after the short-lived “normal” fluorescence (prompt fluorescence; emission lifetime on the nanosecond scale) (Figure 1c). Utilizing this unique process, the energy of electrically generated excited spin-triplets can be harvested as TADF from the S1. Therefore, in principle, all the electrically generated excitons can be used as “electrofluorescence.” Further, since strong SOC induced by the internal heavy atom effect is unnecessary, expensive rare metal atoms such as iridium and platinum, which are required for room-temperature phosphorescent materials, are not needed and the degree of freedom of molecular design is dramatically increased. Therefore, the development of TADF emitters is being carried out worldwide as a critical material for next-generation OLEDs.

In this review, recent progress in molecular design strategies for developing high-performance TADF emitters aimed for OLED applications and the understanding of the spin-flip mechanism in TADF processes are discussed.

Here, we briefly review the history of the discovery and research development of TADF processes. In the 1920s and 1930s, the French physicists Jean and Francis Perrin and the Polish physicist A. Jablonski proposed an excited state energy level diagram for organic molecules, i.e., now called Jablonski or Perrin-Jablonski diagram, to explain long-lived emission and/or energetically isolated emission in organic molecules.1416 For example, Jablonski suggested that the “metastable states,” which are energetically different from the excited state for fluorescence, form another energy level for phosphorescence (It is referred to as “phosphorescence” in ref. 16. However, note that phosphorescence was not yet recognized as radiative decay from the excited spin-triplet state at that time. The term phosphorescence may have meant long-lived emission at that time: refs. 17, 18). It is noteworthy that the TADF process is already implied in these diagrams. They pointed out that if temperature is sufficiently high, the state transition from the “metastable state” to the fluorescent state, followed by the emission from the fluorescent state. Thus, they expected the possible electron transition process of S0 → S1 → T1 → S1 → ‘phosphorescence’.

The early experimental observation of TADF behavior in scientific history appeared in reports on fluorescence with long duration reported by F. Perrin in the 1920s and the room-temperature “phosphorescence” of Eosin Y in glycerol solution reported by S. Boudin in 193019 (Note again that phosphorescence was not yet recognized as radiative decay from the excited spin-triplet state at that time). Then, 10 years later, G. N. Lewis et al. reported a temperature dependence of the fluorescence intensity of fluorescein dispersed in a boric acid glass matrix. They found that fluorescein can emit delayed fluorescence that appears with elevating temperature with an activation energy of about 8 kcal mol−1.20

About 30 years after S. Boudin reported her findings, C. A. Parker and C. G. Hatchard re-examined her report and experimentally proved that the “true phosphorescence” of Eosin Y is identified at a wavelength different from the “phosphorescence” she had observed and that the “phosphorescence” she had observed was TADF.21 They also found an activation energy of about 10 kcal mol−1 for the TADF of Eosin Y. Since TADF was firstly observed in Eosin Y in this way, it is also called E-type delayed fluorescence (E-type DF) to distinguish it from delayed fluorescence originating from triplet-triplet annihilation (TTA).

In addition to Eosin Y and fluorescein, C. A. Parker et al. also reported similar TADF behavior in benzil and benzophenone.22,23 At that time, the term “Activation-controlled DF” was used to explain the behavior. It is also noteworthy that Parker et al. mentioned the possibility of TADF to explain long-lived emission from the aspect of charge-transfer (CT) complexes in ref. 22, which had been reported by S. Nagakura et al. at the same time.24

Although delayed fluorescence via TTA had been extensively studied as an EL process for OLEDs to harvest the triplet energy as EL,25 TADF was rarely considered. This was due to the low TADF efficiency of compounds known at the time. The first report regarding electro-TADF was by A. Endo et al. in 2009.26 They used a tin porphyrin complex (SnF2OEP) as an emitter for OLEDs and confirmed the TADF behavior even under electrical excitation. This result implies the possibility of realizing highly efficient OLEDs by TADF. However, the reported external quantum efficiency (EQE) was relatively low at less than 1%.

A notable turning point for TADF-OLEDs was the report of highly efficient electro-TADF with a metal-free TADF molecule, namely PIC-TRZ, developed by A. Endo et al. in 2011.27 In an OLED using PIC-TRZ as an emitter, the maximum EQE reached 5.3%, which exceeded the theoretical limit of 5% for conventional fluorescent OLEDs assuming a light outcoupling efficiency of 20%. Thus, the advantages of the TADF process for achieving high EL efficiency by harvesting excited spin-triple state were suggested. Successively, an internal EL quantum efficiency of nearly 100% was achieved in cyanobenzene-carbazole derivatives by H. Uoyama et al. in 2012.28 This is a breakthrough in luminescent materials innovation and such ultimate internal EL efficiency in purely organic compounds attracted much attention as the third generation emitter molecule for OLEDs following room-temperature phosphorescent emitters.

In this section, we review the fundamental design guidelines for molecules exhibiting efficient TADF. The basis of material design to obtain efficient TADF is to achieve a small ΔEST between the two considered excited spin states, as evidenced by eq (1). In other words, if the activation energy for RISC from T1 to S1 is sufficiently small, a large RISC rate constant (kRISC) relative to the nonradiative decay rate constant from T1 to S0 (knrT) should be obtained, and the spin-triplet state will be more likely upconverted to the spin-singlet state than non-radiatively decay. Note that the nonradiative decay channel from T1 to S0 would also be a slow process because of the large energy gap between them, i.e., weak spin-state mixing.

Here, kRISC can be described theoretically as follows:

(2)
where A is the frequency factor, kB is Boltzmann's constant and T is absolute temperature. Also, ΔEST is determined by the exchange integral J, which is described as follows:
(3)
where r1 and r2 denote the coordinates of electrons in a highest occupied molecular orbital (HOMO) and a lowest unoccupied molecular orbital (LUMO). This is the result of the expected Coulomb repulsion of two electrons, which means that the ΔEST will be small if the distribution of the HOMO and LUMO is spatially separated.

An effective method for spatially separating the HOMO and LUMO in a molecule is the formation of a CT-type excited state for both spin states. When an electron donor unit (D) and an electron acceptor unit (A) coexist in a molecule, the HOMO is strongly localized on the D and the LUMO on the A. When this molecule is excited, the electron in D can transfer to A, resulting in the formation of a CT excited state for both spin states naturally. To reduce the overlap integral between the HOMO and LUMO in a molecule further, it is also useful to increase the dihedral angle of the bond between D and A units, or to cleave the π conjugation between the units by introducing a spacer unit. In addition to the formation of the CT excited state, note that the energy level of the lowest excited spin-triplet state of each unit must be higher than the CT spin-triplet state or energetically close enough to degenerate to confine the triplet energy to the CT excited state.

For these reasons, the utilization of intramolecular CT excited states has become a common strategy for guiding molecular design to achieve highly efficient TADF in recent years. In the intramolecular CT excited state, the CT excited level can be determined by the HOMO energy level of the D unit and the LUMO energy level of the A unit. In fact, many molecules exhibiting TADF in wavelengths ranging from ultraviolet to visible and even near-infrared (NIR) have been reported by selecting appropriate D and A units.2932

The simplest way to reduce ΔEST is to connect the D unit and the A unit and to increase the dihedral angle between the units in the orthogonal direction, as mentioned previously. This strategy has been used in the earliest designs for TADF materials, and many TADF molecules are being developed based on this design guideline. For example, Cz-TRZ1 (Figure 2), composed of phenyl-carbazole (PhCz) as a D unit and triphenyltriazine as an A unit, possesses weak TADF activity because of the small dihedral angle between the units. However, the introduction of methyl groups into the central phenyl or the PhCz unit dramatically increased the dihedral angle due to the steric repulsion between them, resulting in efficient TADF.33

A completely different molecular design strategy for TADF was reported by T. Hatakeyama et al., in 2016 (Figure 3).34 By including electron donor nitrogen and electron acceptor boron in the π-conjugated system, the HOMO can be localized to three of the six carbon atoms of the benzene ring, and the LUMO can be localized to the other three. This is called the “multiple resonance effect”, providing a small ΔEST (Figure 4). In fact, DABNA derivatives possess a small ΔEST (∼0.20 eV) and showed high PLQY with efficient TADF properties.3436 Further, since the fused structure significantly suppresses the structural relaxation, i.e., molecular vibrational mode, the spectral width of DABNA derivatives (<30 nm) is quite narrow compared to typical fluorescent materials (∼40–60 nm), phosphorescent materials (∼60–90 nm), and conventional TADF materials (∼100 nm). This is a critical parameter for future display applications.

Note that PLQY, i.e., a radiative decay rate constant from S1 to S0, depends on the oscillator strength (f) between S1 and S0. Theoretically, the f value is proportional to the square of the transition dipole moment (Q), and Q becomes large with an increase of the HOMO and LUMO overlap while the spatial separation of the HOMO and LUMO is useful for reducing ΔEST. In other words, if the overlap between the HOMO and LUMO is completely zero in order to degenerate both spin-state, i.e., ΔEST = 0 eV, the radiatively decay channel from S1 to S0 should shut off. Therefore, to obtain high PLQY while achieving small ΔEST, it is necessary to ensure the overlap of the HOMO and LUMO to some extent by adjusting the degree of the dihedral angle between the D and A units with the presence or absence of a π conjugation spacer. To develop materials that show strong TADF activity, exquisite and precise molecular design is required to satisfy two contradictory factors of small ΔEST and large oscillator strength simultaneously.

As a method for adjusting the orbital overlap, in addition to introducing a π-conjugated spacer, the D or A unit can be further modified with an additional donor or acceptor moiety, respectively. By extending the donor or acceptor moieties, the HOMO or LUMO distribution becomes larger in the molecule and the overlap of orbitals becomes smaller relative to the whole orbitals. For example, DACT-II,37 which enhanced donor properties by introducing additional diphenylamine at the β-position of carbazole, has a very small ΔEST (<0.01 eV) despite the relatively small dihedral angle between the D and A units and exhibits high internal EL quantum efficiency, reaching 100% due to their strong TADF activity.

Although various TADF materials have been reported, it has become clear that the mechanism of TADF processes is not as simple as initially thought. For example, the plots of kRISC against ΔEST clearly depend on ΔEST (Figure 5). However, the comparison of compounds with similar ΔEST values reveals an intriguingly broad range of kRISC, i.e., ∼103 order of magnitude. Further, some TADF materials show very high TADF efficiency with high kRISC even though ΔEST is more than ten times larger than the thermal energy at room temperature. This means that ΔEST alone does not determine kRISC. This is because, as can be seen from eq (1), the small ΔEST is not a necessary and sufficient condition to achieve high kRISC and the SOC matrix element has to be carefully considered. Hence, theoretical calculations based on quantum chemistry have been used to reproduce and predict kRISC of molecules exhibiting TADF.

The L·S term (the inner product of angular momentum and spin magnetic momentum) in the perturbed Hamiltonian for the SOC should play an essential role in TADF molecules because a large internal heavy atom effect cannot be expected. That is, in theory, if the spin quantum number changes by ±1, i.e., a spin-flip event, the magnetic quantum number must change by ∓1 simultaneously to keep the total spin number unchanged. This is the well-known El-Sayed selection rule in organic molecular systems.38 Thus, researchers pointed out the importance of the orbital selection rule and the second-order perturbation including non-Born-Oppenheimer effect for the spin-flip process in these molecular systems.3944

As mentioned earlier, the S1 and T1 of D-A type molecules exhibiting TADF are often a CT excited state having a “similar” molecular orbital. As a result, the spin-flip between pure CT excited states is expected to be unlikely. Therefore, it has been considered that excited spin-triplet states with different molecular orbitals to that of S1 enable the spin-flip events to proceed (Figure 5a). For example, a local excitation (LE) state corresponding to π-π* transition is well-known as an excited state having a property different from that of the CT excited state. As an example, P. Data et al. have pointed out that, in dibenzophenazine derivatives, the RISC process occurs between the 1CT and triplet LE (3LE) of the dibenzophenazine unit.42 Also, P. K. Samanta et al. reported that the theoretically calculated SOC matrix element of the 1CT → 3LE transition is about two orders of magnitude larger than that of the 1CT → 3CT transition.43 Further, using transient absorption measurements, T. Hosokai et al. have experimentally demonstrated that the energy difference ΔEST(LE) between the 1CT and 3LE levels was an essential factor for the improvement of TADF efficiency in benzonitrile-carbazole derivatives.45

In recent years, H. Noda et al. have reported that spin-flip events for not only RISC but also forward ISC proceed endothermically in 4CzIPN (Figure 5b). They also pointed out that these processes efficiently proceed between S1 and the high-ordered hybrid local-CT (HLCT) state that originates from the partial molecular structure of the D-A type molecules with multiple donors and acceptors.46 It was also noted that molecular vibration plays an essential role in bypassing the high-ordered excited spin-triplet state. In other words, it is necessary to consider not only the simple energy relationship between 1CT and 3LE states but also the transition process by taking into account the molecular vibration and high-ordered excited states with different molecular orbitals.

N. Aizawa et al. have also reported that it is possible to evaluate kRISC by quantum chemical calculations by determining the molecular structure at the intersection of the potential surface of each excited state in consideration of the high-ordered excited spin-triplet states.47 They applied the calculation method to about 20 existing TADF materials and theoretically confirmed that the RISC proceeds efficiently through the conical intersection between S1 and Tn, and not between S1 and T1 directly. This suggests that the ISC/RISC cycles are mediated by the upper Tn states. On the other hand, Aspuru-Guzik et al. reported a high-speed protocol that enables high-efficiency identification of molecules exhibiting efficient TADF by using high-throughput computing and machine learning algorithms to perform high-speed screening of more than 1.6 million molecules by quantum chemical computing.48 In the future, materials informatics combined with machine learning is expected to further improve OLED characteristics through efficient virtual screening of TADF materials.

Although the spin conversion mechanism in TADF processes has been gradually unveiled, discussion is still ongoing, and there are still open questions in detail. However, there is no doubt that molecular design including not only S1 and T1 but also multiple high-ordered excited states for both spin-triplet and spin-singlet is necessary for efficient upconversion. By forming multiple excited states that are capable of participating in the spin-flip process, the acceleration of kRISC is expected (Figure 5a). Indeed, L.-S. Cui et al. have reported that kRISC of up to 107 s−1 can be achieved by 5Cz-TRZ, which possesses multiple electron donors and achieves a dense triplet-state manifold.49 OLEDs using 5Cz-TRZ as an emitter exhibit a maximum external quantum efficiency (EQE) of over 29%, and their EL efficiency roll-off characteristics are also well suppressed, i.e., less than a 2.3% decrease against the maximum EQE even in the high luminance region (1,000 cd m−2). In the 5Cz-TRZ based OLED, the accumulated triplet density during device operation is expected to be reduced by the large kRISC. Therefore, the reduction of triplet density surely contributes to suppress exciton deactivation processes even under a high current density region.

Independently of L.-S. Cui et al., H. Kaji et al. reported that TpAT-tFFO with dimethylacridan and diphenyltriazine introduced onto a triptycene core could achieve large kRISC of the order of 107 s−1.50 In TpAT-tFFO, dimethyl acridan and diphenyl triazine are spatially arranged at an inter-unit distance (4.72 Å) through the triptycene core. The energy difference of the excited spin state related to the ISC/RISC process, i.e., CT and LE, is minimized according to quantum chemical calculations. Also, since the electron transition between the excited spin states is strictly prohibited when the orientations between the units become perfectly parallel, a dihedral angle between the units has settled to 10°. Such precise molecular design achieved the remarkably large kRISC of 1.2 × 107 s−1.

Note that the delayed fluorescence lifetimes of 5Cz-TRZ (4.1 µs) and TpAT-tFFO (1.3 µs) are still on the order of µs, despite the large kRISC. This is closely related to the balance of three rate constants of radiative decay from S1 (kr), ISC (kISC), and kRISC in order to shorten the triplet exciton lifetime. In fact, kRISC of TMCz-BO proposed by J.-U. Kim et al. is 1.9 × 106 s−1, which is about one order of magnitude smaller than those of 5Cz-TRZ and TpAT-tFFO, but the delayed fluorescence lifetime of TMCz-BO is as short as 750 ns.51 Recently, multiple D-A molecules containing multiple types of electron donor or electron acceptor units have also been proposed,50 which is expected to be an effective molecular design for forming dense excited spin states. Especially, it is noteworthy that the accelerated RISC process by the introduction of multiple types of electron donor in 5CzBN significantly improved OLED lifetime.52

Conversely, a slow spin-flip can be achieved by taking into account the alignment of the high-ordered excited spin-triplet states. H. Noda et al. designed p-2Cz2BMe using phenyl-dimesitylboron as the A unit and phenyl-carbazole as the D unit and reported the very small kRISC of 2.1 × 103 s−1 while the PLQY was as high as 89%.53 The slow spin-flip can be ascribed to the large energy difference between the 3LE and the 1,3CT states in the p-2Cz2BMe, i.e., weak SOC. Small kRISC is undesirable for OLED applications because slow spin-flip events cause a large accumulation of triplet excitons, but it may be useful for other applications such as security inks and so on. Note that large kRISC is not always necessary to obtain high TADF efficiency.

According to the active research and development of TADF molecules and the devices based on them, the internal EL quantum efficiency of TADF-OLEDs has reached the theoretical limit, i.e., 100%. TADF-OLEDs are thus now recognized as “third-generation OLEDs” following OLEDs based on fluorescence (first-generation) and room-temperature phosphorescence (second-generation). However, there are still some critical problems that need to be solved before TADF-OLEDs can be used for practical applications.

For example, to use such OLEDs in display applications, it is necessary to increase the chromaticity of the EL spectra. Most TADF materials exhibit a broad emission spectrum with a rather large full width at half-maximum (FWHM) of ∼100 nm because they emit from a CT excited state. On the other hand, conventional fluorescent materials, which emit through a π-π* transition, have a narrower spectrum. Further, in the current stage of R&D in the OLED industrial field, it is a prerequisite to realize an internal EL quantum efficiency of 100%, and it is essential to realize a device that achieves both high EL efficiency and high device durability simultaneously. In phosphorescent-OLEDs and TADF-OLEDs, since high energy and long-lived excited spin-triplet states are used as an energy source for EL, high densities of triplet excitons accumulate during device operation. This excess accumulation of triplet excitons can cause exciton annihilation processes and unexpected chemical reactions, which can trigger OLED degradation processes, especially in blue OLEDs.54 Hence, to realize OLEDs with high operational stability, triplet excitons generated during operation must be consumed quickly.

As mentioned in the previous section, in TADF-OLEDs, to quickly consume the triplet excitons, it is necessary not only to increase kRISC but also to extract the singlet excitons as delayed fluorescence quickly, i.e., a large kr. However, small ΔEST is an essential factor for increasing kRISC, and increasing kr while keeping ΔEST small is a contradictory relation. Fortunately, this conundrum can be solved by providing an additional electron transition pathway.

The most unique feature of the TADF process is the ability to convert the generated excited spin-triplet states into the spin-singlet states with different spin multiplicities regardless of whether emission happens on the TADF molecule or not. Thus, TADF molecules can also be used as an energy source for “fluorescent molecules” rather than as “emitter.” In 2014, H. Nakanotani et al. and D. Zhang et al. demonstrated the concept of “TADF-sensitized fluorescence” almost simultaneously and independently.55,56 Figure 6 illustrates a conceptual energy diagram of the proposed energy transfer process (TADF-Assisted Fluorescence: TAF). Triplet excitons are generated directly by charge carrier recombination on a TADF molecule. The electrically generated triplet excitons are then converted to excited spin-singlet states via the RISC process on the TADF molecule. In the TADF process, kr and the kISC compete with each other. Therefore, the ISC and RISC processes cycle until the exciton undergoes radiative decay from S1.

However, if an energy-accepting fluorescent molecule is present in the matrix, the excited spin-singlet energy of the TADF molecule can rapidly transfer to the spin-singlet state of the fluorescent molecule based on the dipole-dipole mechanism. This mean that the energy initially generated as TTADF is converted to the STADF via RISC, and the transition from STADF to SFluorescence becomes a spin allowed transition. Furthermore, the rate constant for the energy transfer (kET) is generally much larger than kr_TADF and kISC_TADF. Therefore, the probability of the electron transition of STADF to TTADF becomes very small. As a result, the energy of the excited spin-singlet of fluorescence molecules can be released from S1_Flu as EL. Major features of this TAF mechanism are (1) the easy tuning of an EL spectrum by using a wide variety of fluorescent molecules and (2) the reduction of triplet exciton density accumulated in OLEDs by the quick energy transfer process.

In fact, by using TADF molecules as energy donors for conventional fluorophores, ideal EQE up to ∼20% can be achieved with EL that originates from the fluorophore.57 This EQE means that the internal EL quantum efficiency has reached a theoretical limit (100%), indicating that all electrically generated excitons can be used as EL from the fluorophore. Furthermore, since the EL spectrum is derived from a conventional fluorophore, the color purity of the EL is also remarkably improved. For example, although 4CzIPN is well-known as a highly efficient green TADF emitter, FWHM of the EL spectrum is as wide as >50 nm, and the Commission Internationale de l'Eclairage coordinates (CIE) of the 4CzIPN based OLED (0.35, 0.58) are not enough to satisfy ITU-R Recommendation BT. 2020. Recently, X. Song et al. reported that by using boron-dipyrromethene derivatives (tPhBODIPY) (Figure 7), which are known to exhibit high PLQY and quite narrow emission spectrum while not showing TADF activity, as energy acceptor molecules for 4CzIPN, the color purity could be dramatically improved by efficient energy transfer from 4CzIPN while maintaining the high EQE (maximum EQE = 19%).58 The FWHM of the EL spectrum in the device is only 28 nm, and the CIE chromaticity coordinates are (0.26, 0.67) (Figure 7). Although further improvement of purification of chromaticity is required to satisfy ITU-R Recommendation BT. 2020, this may be achieved by improving the device structure, such as by utilizing optical interference effects.59

Here, we would like to point out the other advantages of TADF sensitization for OLED performance. The ability to manufacture high-performance OLEDs with long device operational lifetimes is one of the most critical challenges for scientific and industrial development as mentioned before. The device operational stability of TAF-OLEDs is improved compared to normal TADF-OLEDs while maintaining a high EQE. Figure 8 showed the EL intensity of TADF-OLEDs and TAF-OLEDs as a function of time operated at a constant current starting at an initial luminance of L0 = 1,000 cd m−2. The LT50 (i.e., continuous operation time until the initial luminance decays by 50%) of TADF-OLEDs with 4CzIPN-Me as an emitter was about 1,500 hours, whereas the LT50 of the TAF-OLEDs was more than 3,500 hours.57 This improvement in the operational stability can be attributed to reduction of the triplet exciton density accumulated in the OLED.

By utilizing the energy transfer process from TADF materials to conventional fluorophores, it is possible to realize an ideal EL quantum efficiency reaching the theoretical limit even in an OLED using a conventional fluorophore with high operational stability. TAF-OLEDs with high EQE not only in the visible region but also in the NIR region60 have been reported by combining appropriate TADF molecules with fluorophores. Further, remarkable improvements in the device operational stability of TAF-OLEDs have been reported on the industrial side.61 Thus, this OLED architecture is a promising candidate for future practical applications.

When electrons and holes accumulate at the interface between an electron transport material and a hole transport material, “exciplexes” are often observed (Figure 9). An exciplex is defined as a radiative decay process from the intermolecular excited CT state, and their emission efficiency is generally believed to be quite low. Therefore, it has been thought that exciplexes should be eliminated as much as possible in OLEDs.9

In 2012, K. Goushi et al. disclosed that an exciplex exhibiting TADF can be obtained by using an electron transport material and a hole transport material having a much higher T1 than that of the exciplex.62 Exciplexes are generated by the formation of intermolecular excited CT states, so their electron and hole densities are strongly localized on each molecule. Therefore, since each of the molecular orbitals is spatially separated, the ΔEST in an exciplex system should become so small that it can be regarded as degenerating. The realization of TADF-type exciplexes is a very important discovery for achieving high-performance OLEDs. In particular, when a mixed film of an electron transport material and a hole transport material is used as an emissive layer for an OLED, the driving voltage of the OLED can be remarkably reduced, which greatly contributes to a reduction in power consumption of the OLED with the harvesting ability of the energy of excited spin-triplet state.63

Although PLQY of exciplexes is generally considered to be low, K.-H. Kim et al. reported that the suppression of nonradiative decay processes of the exciplex system in a low-temperature environment resulted in PLQY of up to 100% for a TADF-type exciplex.64 Therefore, even when the intermolecular CT excited state is used, it is possible in principle to obtain highly efficient EL. Further, by doping the fluorophore into the emissive layer, an internal EL quantum efficiency of up to 100% can be achieved by utilizing the TADF-sensitized fluorescence process.6568

The TADF process could also lead to a breakthrough toward lowering the lasing threshold in organic semiconductor lasers (OSLs).69,70 OSLs have great potential not only as low-cost but also color-tunable laser sources. However, the lasing threshold current density in OSLs is expected to be quite high because electrically generated spin-triplets generally cannot contribute to lasing. Furthermore, it is expected that the accumulated triplet excitons will stop the lasing due to exciton annihilation and light propagation loss arising from absorption by the excited spin-triplet states. Therefore, conventionally, the triplet exciton density has been restricted by deactivating the triplet excitons using a triplet quencher such as an anthracene derivative. However, such a strategy would result in an exciton generation efficiency of at most 25% in OSLs. On the other hand, if TADF materials, which can use triplet excitons as delayed fluorescence, can be used as a laser medium, all electrically generated excitons may contribute to lasing.

As an example, the TAF system using ACRXTN as a TADF molecule and C545T as a laser dye exhibited amplified spontaneous emission (ASE) under optical pumping.71 The ASE threshold has been estimated to be 0.8 µJ cm−2, which is equivalent to the threshold of the sample not containing ACRXTN. The current density required for ASE is expected to decrease from 451 A cm−2 to 83 A cm−2 assuming no other exciton annihilation processes. However, since the kRISC of ACRXTN is as low as 105 s−1, the density of triplet excitons that can contribute to ASE is expected to be extremely small. Recently, it has been also reported that molecules exhibiting TADF themselves exhibit ASE or lasing (Figure 10).7275 Whether or not the singlet excitons generated via RISC can contribute to stimulated emission is just beginning to be studied. However, it is expected that a TADF laser dye will be realized by comprehensive material design and screening in the future.76

Here, we briefly reviewed recent progress and advances in designing molecules exhibiting TADF and the understanding of the TADF process, mainly in terms of their application to OLEDs. The internal EL quantum efficiency of TADF-OLEDs has reached the theoretical limit of ∼100%, making them a promising candidate for a low-cost and high-performance OLEDs, and we will see a bright future for TADF materials in OLED applications. In particular, because of the high degree of freedom in designing molecules, the realization of high-performance blue TADF-OLEDs with high EL efficiency and high operational stability is strongly desired. To realize blue OLEDs, with a complete understanding of the electron spin conversion mechanism in the TADF process by further study, one of the next research milestones is to realize a short DF lifetime that is comparable with typical fluorescence.

Reports of TADF molecules have increased rapidly since a Nature report in 2012. Since it was not possible to cover all of the reports regarding the TADF topics in this review, we are concerned that some important papers are missing. There are many other exciting reviews2932 that can help to fill the gaps.

Chihaya Adachi obtained his doctorate in Materials Science and Technology in 1991 from Kyushu University and held positions as at the Chemical Products R&D Center at Ricoh Co., the Department of Functional Polymer Science at Shinshu University, the Department of Electrical Engineering at Princeton University, and Chitose Institute of Science and Technology before returning to Kyushu University as a professor. Adachi’s research combines the areas of chemistry, physics, and electronics to advance the field of organic light-emitting materials and devices from both the materials and device perspectives.

Hajime Nakanotani received his Ph.D. degree in 2010 from Kyushu University and held positions as at the R&D Center at Ricoh Co., and Institute of Systems, Information Technologies and Nanotechnologies (ISIT). His academic position began at the Department of Applied Chemistry, Kyushu University as an associate professor (2015–). His current research interests focus on photochemistry and physics in organic light-emitting materials to design high-performance organic light-emitting devices.

Youichi Tsuchiya is a research fellow special appointment as an associate professor of Kyushu University’s Center for Organic Photonics and Electronics Research (OPERA). He obtained his doctorate in 2004 from Nara Institute of Science and Technology (NAIST) and held positions as research fellow at National Institute of Advanced Industrial Science and Technology (AIST), Riken, and Institute of Systems, Information Technologies and nanotechnologies (ISIT) before working at Kyushu University. His current research interests focus the photochemistry and physics of functional organic materials and its applications.