Recent progress on exciplex-emitting OLEDs

ABSTRACT The thermally-activated-delayed-fluorescence (TADF) characteristics makes exciplexes being the hot subject in the organic light-emitting diode (OLED) research field. The theoretical limit for the efficiency of the conventional fluorescent OLEDs have been leaped by exploiting the triplet harvesting ability of exciplexes. Exciplexes are easily formed by blending electron donor molecules and electron acceptor molecules, which are generally hole transporting materials and electron transporting materials, respectively, resulting in easy access for employing exciplexes to OLEDs. We introduce the photo-physical characteristics of exciplexes derived from the charge-transfer characteristics and their application to OLEDs as emitters. Single exciplex-emitting OLEDs as well as exciplex emission-based white OLEDs are covered in this review.


Excited CT complex
Exciplexes are generated by CT at the excited state between electron donor and electron acceptor molecules, from which the name of exciplex (excited CT complex) originates [7]. The exciplex emission is usually observed in type-II heterojunctions, which consist of a pair of two molecules with the energy levels of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of the donor molecule being higher than those of the acceptor molecule as shown in Figure 1(a). The CT in exciplexes can be considered as the partial CT, and the exciplex wavefunction can be expressed as the mixed state of pure locally ground, locally excited (LE) and CT states as a result of the configuration interaction which is the superposition of the configurations. In this form of wavefunction, the fraction of CT in the exciplex is defined as the square of the (a) Schematic illustration of molecular-orbital energy level for the type-II heterojunction; (b) Measured (black) and simulated (red) PL spectra of exciplexes consisting of DCA and various alkylbenzene donors with different oxidation potentials in cyclohexane, resulting in different average degrees of CT of the exciplexes (ref. [9]); (c) PL spectrum and extinction coefficient of the m-MTDATA:PO-T2T blended film and extinction coefficients of m-MTDATA and PO-T2T films (ref. [11]); (d) Radiative decay rate constants (k f ), nonradiative decay rate constants (k nr ), and ISC rate constants (k isc) of the singlet exciplex states in which TCA and DCA are acceptors and various alkylbenzenes are donors in various solvents as a function of the average emission energy of the exciplexes (ref. [12]). expansion coefficient for the pure CT state [8,9]. Exciplexes in solution and films were reported to exhibit various fractions of CT dependent on the microscopic environment, librational displacement, and geometries of exciplex-forming pairs [9,10].
A significant feature of exciplexes is their photoluminescence (PL) spectra. As the fraction of CT in the exciplex increase, the PL spectra of exciplexes are redshifted from the PL spectra of the constituent molecules, broad, and featureless without the vibrational progression [9,13]. One example is shown in Figure 1b, where exciplexes are formed between 9,10-dicyanoanthracene (DCA) and various alkylbenzenes such as p-xylene, 1,2,4trimethylbenzene, 1,2,3,5-tetramethylbenzene, durene, pentamethylbenzene, and hexamethylbenzene in cyclohexane [9]. DCA and the alkylbenzens are the electron acceptor and electron donors, respectively. The average degrees of CT of each exciplex-forming systems are also given in Figure 1(b). Exciplexes participating in emission have various degrees of CT which are dependent on the microscopic environment and librational displacement. The simulated PL spectra (red curves) of exciplexes are well matched with the experimental ones (black curves). The decrease of the oxidation potentials of the donors corresponding to the increase of the energy levels of the HOMO of the donors leads to the decrease of the energy of the pure CT state. It gives rise to the increase of the average degree of CT of exciplexes, resulting in the reduction of the characteristics from the pure LE state, and the expansion of the characteristics from the pure CT state.
In contrast to the PL spectra of exciplexes, the absorption spectra of exciplexes are usually not observed [14]. However, the existence of sub-band gap intermolecular CT absorption in exciplex-forming systems with very low intensity was reported employing elaborate measurements such as Fourier transform photocurrent spectroscopy (FTPS) and photothermal deflection spectroscopy (PDS). [11,[15][16][17][18][19][20][21][22]. These new absorption peaks originate from the direct optical excitation from the ground state to the exciplex state [21]. Recently a much simple method combining UV-Vis-NIR spectrophotometer and ellipsometry measurements has been developed to measure the very weak intermolecular CT absorption with comparable sensitivity to FTPS and PDS [11]. The absorbance of an exciplex-forming mixed film obtained from the reflection/transmission measurement using an UV-Vis-NIR spectrophotometer is analyzed using the transfer matrix method combined with ellipsometry measurement to obtain extinction coefficient for the intermolecular CT state as low as 10 −4 . One example is shown in Figure 1c for an exciplex-forming m-MTDATA:2,4, 6-tris [3-(diphenylphosphinyl)phenyl]-1,3,5-triazine (PO-T2T) mixed film. An extra absorption band from the absorption bands of the pristine films appears in the sub-bandgap region of the consisting molecules with the extinction coefficients below 10 −3 , which corresponds to the exciplex absorption band. The low absorption coefficient of the singlet exciplex state is attributed to the CT characteristics of exciplexes which is called the overlap forbiddeness [7].
The transition rate constants in exciplex-forming systems are also influenced by the degree of CT of exciplexes, f CT [9,12,13,[23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39]. The fluorescence rate constants (k f ), nonradiative rate constants (k nr ), and intersystem crossing (ISC) rate constants (k isc ) of the singlet exciplex states between 2,6,9,10-tetracyanoanthracene (TCA) and DCA as electron acceptors and various alkylbenzenes as electron donors in various solvents are plotted as a function of the average emission energy of singlet exciplex states in Figure 1(d) [12]. k f of the singlet exciplex state decreases with f CT due to the decrease of the LE character in the singlet exciplex state. k nr of them decreases with f CT at first due to the decrease of the LE character in the singlet exciplex state, but increases with f CT when f CT is larger than 0.7 due to the decrease of the energy difference between the singlet exciplex state and the ground state. k isc of them increases with f CT due to the decrease of the energy difference between the singlet exciplex state and the triplet LE state. In the exciplex-forming systems, the triplet LE state energy was the lowest triplet state energy of DCA and TCA, which is 13,100 cm −1 [31].
The ISC and reverse intersystem crossing (RISC) are the spin-forbidden transition [7]. Transition probability of them is zero when 0th order approximation. However, state mixing could explain the nonzero probability for the transitions [7]. There are two significant mechanisms for mixing singlet and triplet states of exciplexes. One is the spin-orbit coupling, and the other is the hyperfine interaction. The spin-orbit coupling is caused by the magnetic torque generated by the electron orbital motion which could change the electron spin [7]. The matrix elements of the spin-orbit coupling of the pure CT state are zero based on the one-electron approximation [40]. Thus, the hyperfine interaction plays a major role for the ISC and RISC of the pure CT state (f CT = 1) instead of the spin-orbit coupling [31,41]. The hyperfine interaction is induced by the magnetic torques due to nuclear spin of molecules [7]. It becomes negligible compared to the spin-orbit coupling if the electronic coupling between the singlet and triplet states is not weak, achieved by the configuration interaction of pure CT states with other LE states as well as the small singlet-triplet energy splitting [7,31,42]. In the case of the ISC from the highly ionic singlet exciplex state to the highly ionic triplet exciplex state, the matrix elements of the spin-orbit coupling for the ISC are small because of the high degree of CT of two states. Then, the ISC from the highly ionic singlet exciplex state to the highly neutral triplet exciplex state could be faster than the ISC from the highly ionic singlet exciplex state to the highly ionic triplet exciplex state [40]. It implies that the transition from the highly ionic singlet exciplex state to the highly neutral triplet exciplex state would be governed by the direct transition from the highly ionic singlet exciplex state to the highly neutral triplet exciplex state rather than the transition via the highly ionic triplet exciplex state [31,40]. It indicates that ISC and RISC could be accelerated when energy levels of the highly ionic singlet exciplex state and the highly neutral triplet exciplex state are similar [43].

Thermally activated delayed fluorescence
Thermally activated delayed fluorescence (TADF) of exciplexes have become the center of attention since it was reported in the OLED research field. In the past, OLEDs employing pure organic molecules without heavy atoms had the limit of 25% for generating singlet excitons from polarons. Considering that only the singlet exciton is the radiative species in room temperature without heavy atoms, it had served as the obstacles for achieving highly efficient OLEDs over the IQE of 25%. Beginning with the report of the OLEDs exploiting the TADF of exciplexes, however, the theoretical limit for generating singlet excitons from polarons reaches 100% without heavy atoms. The TADF of exciplexes has been the core value for the third generation OLEDs along with intra-molecular CT molecules.
The small singlet-triplet energy splitting of exciplexes causes the TADF. In other words, the transition from the triplet exciplex state to the singlet exciplex state plays a major role in the TADF. The singlet exciplex state undergoes transitions to the ground state by the radiative (fluorescent) and nonradiative ways (internal conversion) as well as the transition to the triplet exciplex state by the nonradiative way (ISC). The triplet exciplex state also undergoes the transitions to the singlet exciplex state by the nonradiative way (RISC) and to the ground state with the radiative (phosphorescent) and the nonradiative ways. Conventional fluorescent molecules have very small RISC rate constant at room temperature due to large singlet-triplet energy splitting, so the lowest triplet excited state has little transition probability to the lowest singlet excited state. On the other hand, exciplexes could have large transition probability for the RISC from the triplet excited state to the singlet excited state at room temperature. It is attributed to small singlet-triplet energy splitting of exciplexes, resulting in strong delayed fluorescence originated from efficient RISC process, followed by the fluorescence process. This is called TADF [3,44,45]. Efficient RISC process could be achieved when energy levels of the highly ionic singlet exciplex state and the highly neutral triplet exciplex state are similar [43].
The state energy diagram of the exciplex-forming system which is typically employed in exciplex-emitting OLEDs is shown in Figure 2(a). The basis states of the exciplex-forming system are denoted as S LE , S CT , T LE , and T CT which are the lowest singlet LE state, the lowest singlet CT state, the lowest triplet LE state, and the lowest triplet CT states, respectively, based on the threestate mixing model for each singlet and triplet states [35]. S CT and T CT which are generated in the exciplexforming system induce the effective state mixing between the states with the same spin multiplicity, resulting in S 1 , S 2 , T 1 , and T 2 states, which are the lowest singlet excited state, the second lowest singlet excited state, the lowest triplet excited state, and the second lowest triplet excited state in exciplex-forming system. They have the both the CT character and the LE character by virtue of the state mixing. The mixing coefficients decide the degrees of CT of the mixed states [9,35]. Exciplexes showing effective TADF characteristics usually have T LE and T CT with the similar energy levels, inducing moderately neutral triplet exciplex state (T 1 ) [43]. The moderately neutral triplet exciplex state exhibits considerable LE character, and effective ISC and RISC could occur by virtue of the state mixing between the singlet and triplet states when their energy levels are close [3,7,31,40,43].
Concentrations of the lowest singlet and triplet states showing TADF characteristics follow the below equations when photoexcitation is applied as a delta function at t = 0 [3].
where [S 1 ] and [T 1 ] are the concentrations of the lowest singlet and triplet excited states, respectively, k S g , k T g , k isc , k risc are the transition rate constants from the singlet state to the ground state and that from the triplet state to the ground state, that from the singlet state to the triplet state, and that from the triplet state to the singlet state, respectively, which are depicted in Figure 2(a). The solution for the differential equations are the below equations when the initial condition is [T 1 ] = 0 for the case of photoexcitation at t = 0 as a delta function. [ where k p and k d are the prompt and delayed fluorescence rate constants, respectively, C p and C d are the prompt and delayed pre-exponential factors for the singlet states, respectively, and C T is the pre-exponential factor for the triplet state [3,[46][47][48]. One example following the equations is depicted in Figure 2(b). The population of the singlet state is usually dominant in the prompt time region, and that of the triplet state is usually dominant in the delayed time region because the lifetime of the singlet states is usually shorter than that of the triplet states. Transient PL intensities of exciplexes follow the black curve in Figure 2 The key transition which could make TADF characteristics is the RISC. As the ISC and RISC quantum yield increases, the delayed part of the transient PL profiles becomes dominant. The high RISC quantum yield indicates the efficient harvesting of the triplet excited states to the singlet excited states (emissive states).
The quantum-mechanical equation for k risc could be transformed into the Arrhenius equation If the vibrational states are treated classically, which is Equation (5) [3,47,[49][50][51]. where A is the pre-exponential factor, E ST is the singlettriplet energy splitting, k B is the Boltzmann constant, and T is the temperature. The thermal distribution corresponds to the Boltzmann distribution, and the thermal equilibration rate of the vibrational states is much faster than the radiationless quantum transition rate between the singlet and triplet states for Equation (5). E ST of the m-MTDATA: 2-(biphenyl-4-yl)-5-(4-tert-butylphenyl)-1,3,4-oxadiazole (t-Bu-PBD) exciplex in the 50 mol% m-MTDATA:t-Bu-PBD blended film was obtained from Equation (5), which is about 50 meV [3]. Considering k B T ≈ 25 meV at room temperature, E ST = 50 meV is small enough to help k risc being large contribution to exciplex fluorescence. The free energy change between the single singlet state and the single triplet state could be obtained from the relation between the free energy change and the equilibrium constant of the reaction as given in Equation (6) [46,52].
where G ST is equal to the free energy change between singlet and triplet states and a factor of 3 is for the number of the degeneracy of the triplet states. Using Equation (6), 26 meV was obtained for G ST of the 9,9 ,9 -triphenyl-9H,9 H,9 H-3,3 :6 ,3 -tercarbazole (Tris-PCz): 3 ,3 , 3 -(1,3,5-triazine-2,4,6-triyl)tris(([1,1 -biphenyl]-3carbonitrile)) (CN-T2T) exciplex in the Tris-PCz:CN-T2T blended film [46]. The PLQY of the exciplexes is not just the portion of the radiative decay rate constant in the decay process of the singlet excited state when TADF takes place. The PLQY in exciplexes is the sum between the PLQY from prompt fluorescence and that from delayed fluorescence. The prompt PLQY ( p ) is the portion of the radiative decay rate constant in the decay process of the singlet excited state which is k f /(k f + k nr + k isc ). The delayed PLQY ( d ) could be written as the following equation [3].
Where isc and risc are the ISC and RISC quantum yields, respectively. The ISC and RISC quantum yields could be expressed by k isc /(k f + k nr + k isc ) and k risc /(k T nr + k risc ) where k T nr is the nonradiative decay rate constant from the triplet state to the ground state, respectively. Then, the PLQY of the exciplexes could be expressed by the below equation if phosphorescence from exciplexes is negligible compared to fluorescence.
The 100% PLQY could be achieved by two ways according to Equation (8). The one way is when k f k nr , k isc . In that case, the delayed fluorescence becomes negligible in the PL process because the transition from the singlet excited states to the triplet excited states becomes negligible. The other way is when k f k nr and risc = 1. In this situation, triplet loss becomes zero even though the transition from the singlet excited states to the triplet excited states is considerable.

Exciplex-emitting OLEDs
OLEDs are optoelectronic devices using the electroluminescence (EL). The first step of the EL is the encounter of a positive polaron (free hole) and a negative polaron (free electron). Free polarons have the same probability of the up-spin and the down-spin. The initial formation ratio of singlet states and triplet states by the encounter of a positive polaron and a negative polaron would be 1:3 consisting of S and T + , T 0 , and T − states. Conventional fluorescent emitters with the large singlet-triplet energy splitting cannot exploit triplet excitons as photons in anyway because the nonradiative transition rate from T 1 to S 1 is much smaller than that from T 1 to S 0 . As the singlet-triplet energy splitting increases, the RISC rate usually decreases. As the energy gap between T 1 and S 0 decreases, the nonradiative transition rate from T 1 to S 0 usually increases [53]. On the other hand, exciplexes with the small singlet-triplet energy splitting, resulting in the increase of k isc and k risc . The large k risc could enhance the transformation ratio of the triplet exciplex state to the singlet exciplex state, resulting in high efficiency of the OLEDs using exciplexes as emitters.
The triplet harvesting ability of exciplexes makes EQEs of OLEDs employing exciplexes increase compared to conventional fluorescent OLEDs. The equation for EQE of OLEDs employing exciplexes as emitting species can be written as Equation (9) if the intensity of phosphorescence is negligible [3].
where η OC is the light-out coupling efficiency which is considered to be 25-30% without any light-out coupling ways when the emitting transition dipole moment vectors are randomly distributed, γ is the recombination efficiency of electrons and holes in OLEDs, and PLQY P is the PL quantum yield of the exciplexes in microcavity structures of OLEDs. PLQY P of the exciplexes is the same with Equation (8) but k f changes into Fk f where F is the Purcell factor. The EQE of OLEDs increases when risc or PLQY P of exciplexes increases. The increase of risc helps to improve the triplet harvesting ability, and the increase of PLQY P helps to improve the transition ability of the energy of singlet excited species to photons. The high risc and PLQY P of the exciplex systems are necessary for highly efficient exciplex-emitting OLEDs. The maximum EQE of OLEDs using exciplexes as emitters can be achieved when risc = 1. risc = 1 leads to the PLQY P to be Fk f /(Fk f + k nr ), and PLQY P = 1 if Fk f k nr . Then, the maximum EQE becomes η OC . If the singlet-triplet energy splitting is large or T 1 state is the highly ionic exciplex state, they would have small risc . When risc = 0, Equation (9) becomes the case for OLEDs using conventional fluorescent emitters.

Emitting structure composed of a donor and an acceptor
In with the PLQY of 62% in the blended film even though exciplex was thought to be inefficient emitting species because of the low transition dipole moment [54]. They fabricated the OLEDs with the organic trilayer composed of NPB, PPSPP, and 2,5-bis-(2 ,2 -bipyridin-6-yl)-1,1dimethyl-3,4-diphenylsilacyclopentadiene (PyPySPyPy). The EL emission was originated from the interface exciplex between NPB and PPSPP with the EL peak of 495 nm and the EQE of 3.4%. The EQE was lower than theoretical limit of the conventional fluorescent OLEDs because the TADF characteristics of the exciplexes could not employed due to the low T 1 state of NPB, resulting in the large singlet-triplet energy splitting.
In 2012, K. Goushi et al. reported an OLED with the higher EQE than the theoretical EQE limit of the conventional fluorescent OLEDs [3]. The emitting layer of the OLED is a blended layer of m-MTDATA/t-Bu-PBD of which structures are shown in Figure 3(a). The redshifted featureless emission of the blended film implies that they form the exciplex as shown in Figure 3(b). Even though the PLQY of the blended film was 20%, the EQE of the OLED was 2%, of which emitting layer is the blended layer of m-MTDATA and t-Bu-PBD as shown in Figure 3(c). Considering that the theoretical limit for the EQE of conventional fluorescent OLEDs with the PLQY of 20% of the emitting layer is around 1.0 ∼ 1.5%, the triplet harvesting process occurs in the OLED, and its mechanism is TADF. The transient PL and EL both followed the two-exponential decay form consisting of prompt and delayed decays with similar decay constants as shown in Figure 3(d). The singlet-triplet energy splitting of the exciplex was 50 meV obtained by the analysis of the transient PL profiles at various temperatures, which is close to the thermal energy at the   Figure 4(a) [5,6]. They form the exciplex considering that the blended film emits a different emission band different from molecule emissions as shown in Figure 4(b). The PLQY of the blends was 36%, and the maximum EQE of the device was 3.1%. It indicates that the triplet harvesting of the exicplex operates in the device. They also reported that the EQE of the device increases as the temperature decreases due to the decrease of the nonradiative decay to the ground state with temperature in the exciplexforming system leading to the increase of the PLQY of the blends as shown in Figure 4(c,d). The PLQY of the blends reaches 99% at 35 K and the EQE of the device reaches 10% at 195 K at which the PLQY of the blends was 87% when the current density was 0.1 mA/cm 2 . Their reports open up possibility that the increase of EQEs using exciplexes could be a key for efficient fluorescent OLEDs with the EQE over 30% like phosphorescent OLEDs.
K.-H. Kim et al. demonstrated the exciplex-emitting OLED with the EQE of 25.2% at 150 K. The emitting layer of the device was blends of TCTA/4,6-bis[3,5(dipyrid-4-yl)phenyl]-2-methylpyrimidine (B4PYMPM) of which the PL peak wavelength and singlet-triplet energy splitting were 509 nm and 8.5 meV, respectively [49]. The EQE and PLQY were 11% and 60%, respectively, at room temperature. As the temperature decreases, the EQE and PLQY increase, leading to the EQE of 25.2% and the PLQY of 100% at 150 K. The calculated outcoupling efficiency of the device at 150 K was 26.6% implying that the almost all triplet excited states in the device were harvested at 150 K. The highest EQE of the reported exciplex-emitting OLEDs at room temperature with the emitting layer of the donor-acceptor blended layer is 20.9%, which was reported in 2019 [73]. The emitting layer of the device was the blends of 4,4 -(diphenylsilanediyl)bis(N,N-diphenylaniline) (TSBPA)/ PO-T2T. Their structures are shown in Figure 5(a) and the EL spectra peak was at 528 nm. The redshifted featureless emission spectrum and delayed emission of their blends indicate that they form the exciplex as shown in Figure 5(b,c). The structure of the device was ITO/NPB/TSBPA/TSBPA:PO-T2T/PO-T2T/LiF/Al of which EQE curve versus current density is shown in Figure 5(d). The PLQY of the exciplex-forming blends was 100%, and the high EL efficiency can be achieved. They also reported blue exciplex-emitting OLED with the EL spectra peaking at 480 nm of which the Commission Internationale de L'Eclairage (CIE) coordinates was (0.16, 0.28). The emitting layer was the blends of 1,3-bis(N-carbazolyl)benzene (mCP)/PO-T2T of which the PLQY was 55% [73]. The maximum EQE was 16% as shown in Figure 5(d). It is the same efficiency with the blue exciplex-emitting OLED with the emitting layer of CN-Cz2:PO-T2T mixtures which was reported in 2018 [72].

Emitting structure composed of a donor, an acceptor, and a functional molecule
Generally, exciplexes formed at bilayers between a donor layer and an acceptor layer or donor-acceptor blends have been employed for exciplex-emitting OLEDs. There are reports that a trilayer consisting of a donor layer, a spacer layer, and an acceptor layer was employed for emitting structure of the exciplex-emitting OLEDs [82,83]. H. Nakanotani et al. reported the exciplex-emitting OLED with an emitting structure of the trilayer consisting of a m-MTDATA layer, a 3,3-di(9H-carbazol-9-yl)biphenyl (mCBP) layer, and a 2,4,6-tris(biphenyl-3-yl)-1,3,5triazine (T2T). m-MTDATA, mCBP, and T2T layers act as a donor layer, a spacer layer, and an acceptor layer, respectively [82]. m-MTDATA and T2T form exciplexes exhibiting fluorescent luminescence with a peak wavelength of ∼ 550 nm in the trilayer structure. The device structure employing the trilayer structure was ITO/m-MTDATA/mCBP/T2T/tris(8-hydroxyquinolinato) aluminum (Alq 3 )/LiF/Al, and the layer thickness of mCBP was varied from 0 to 15 nm. The EL spectra of the devices show exciplex emission only when the thickness of the spacer layer is under 3 nm, and molecule emissions only when the thickness of the spacer layer is over 10 nm. The maximum EQE of the devices was 2.5% when the spacer thickness is 5 nm, which is higher than the EQE of the OLEDs employing exciplexes formed at the bilayer between a donor layer and an acceptor layer or donor-acceptor blends which are 0.5% and 0.9%, respectively. The reduction of the nonradiative decay of the triplet state to the ground state of the exciplex in trilayer structure would help to increase the efficiency of the device.
Exciplex-emitting OLEDs with donor-acceptor-diluter blends were also reported by M. Colella et al. different from typical exciplex-emitting OLEDs with donoracceptor blends [84]. They employed TSBPA and PO-T2T as the electron donor and acceptor, respectively. The diluter was 1,3-bis(triphenylsilyl)benzene (UGH-3) whose structure is shown in Figure 6(a). As the concentration of the diluter increases, the PL spectra of the exciplex in the donor-acceptor-diluter blends show blue-shift because of the increase of the donor-acceptor separation as shown in Figure 6(b), and the PL intensity from the molecules increases. The molecule/exciplex ratio of the PL intensity was larger than 0.1 when the concentration of UGH-3 was larger than 70 vol% due to the increase of the donor-acceptor separation. The PLQY of the donor-acceptor blends was 58%, and the maximum PLQY of the donor-acceptor-diluter blends was 80% when the concentration of UGH-3 was 50 vol% by virtue of the decrease of concentration quenching of the exciplex as shown in Figure 6(c). The increase of the PLQY of the exciplex in the blends leads to the increase of the EQE of the OLED exploiting donor-acceptor-diluter blends. The maximum EQE of the devices was 19.2% when the donor-acceptordiluter blends with 50 vol% UGH-3 was employed as the emitting layer of the device as shown in Figure 6(d), which is higher value than 14.8% which is the EQE of the device with emitting layer of the donor-acceptor blends.

TTA-dominant exciplex-emitting OLEDs
TADF is always not the dominant way for triplet harvesting process in OLEDs employing exciplexes as emitting species [6,50,85,86]. The triplet-triplet annihilation (TTA)was proposed as another mechanism for triplet harvesting process [85,86]. The encounter of two T 1 's can generate S 1 and S 0 in TTA process, resulting in the theoretical maximum IQE of TTA OLEDs of 62.5%. V. Jankus et al. reported the exciplex-emitting OLED with the emitting layer of the blends of NPB/1,3,5-tri(1-phenyl-1Hbenzo[d]imidazol-2-yl)phenyl (TPBi) whose structures are shown in Figure 7(a) [86]. The PLQY of the blends was 28%, and the EQE of the device was 2.7% exceeding the theoretical EQE limit of conventional fluorescent OLEDs as shown in Figure 7(b). They, however, proposed that the triplet harvesting process in the device arose from the TTA process of the NPB T 1 states. The S 1 and T 1 energy of NPB were 3.1 and 2.38 eV, respectively. When TTA process is dominant, the transient PL profiles of delayed emission shows power law decay whose slope changes from −1 to −2. When energetically randomly  generated triplet states relax towards the tail of the density states resulting in the time-dependent rate constant for the TTA process, the slope of the power law decay is −1. After the energetic relaxation of the triplet states, the rate constant for the TTA process becomes timeindependent leading to the power law decay of delayed emission with slope of −2 [86]. The delayed emission of the neat NPB film follows the power law decay whose slope changes from −1 to −2, and the delayed PL emission of the blends and EL emission also follow the power law decay as shown in Figure 7(c). It implies that the TTA process takes place for the NPB T 1 state in the NPB:TPBi blends. The TTA process is the bimolecular reaction so that the intensity of the delayed emission arising from the TTA is linearly proportional to the square of concentration of the T 1 state compared to the TADF process of which delayed emission intensity is linearly proportional to the concentration of the T 1 state [50,86]. Therefore, the delayed emission intensity follows a quadratic dependence on the excitation intensity if the TTA process is dominant. The delayed emission intensity of the blends follows the quadratic dependence on the excitation intensity as shown in Figure 7(d). It implies that the TTA process is dominant for the delayed emission of the NPB:TPBi blends and the triplet harvesting process. The characteristics of the OLEDs described in section 4 are summarized in Table 1.

Exciplex emission-based WOLEDs
Exciplex emission has been employed for WOLEDs as the emission of conventional fluorescence and phosphorescence does. At the early stage, the broad featureless emission bands of exciplexes was considered as the only advantage for WOLEDs with good color quality.
As exciplexes were reported to have triplet harvesting ability, however, exciplexes have become a more important candidate for emitters of WOLEDs with good color quality and efficiency. The emission types of WOLEDs employing exciplex emission are various. Various combinations among exciplex emission, molecule emission, dopant emission, and excimer emission were employed for exciplex emission-based WOLEDs.

Emitting layer of a single donor-acceptor blend
There are various types of structures for WOLEDs. The simple one is a single donor-acceptor blend employed as the emitting layer for WOLEDs [87][88][89][90]. M. Mazzeo et al. reported a WOLED with the emitting layer of blends of N,N -bis(3-methyl-phenyl)-N,N -diphenylbenzidine (TPD) and 2,5-bis(trimethylsilyl thiophene)-1,1-dioxide (STO) [87]. Exciplex formation between TPD and STO was incomplete in the blends, and white EL emission was comprised of blue emission from the TPD molecule and orange emission from the TPD:STO exciplex. The CIE coordinates of EL spectrum were x = 0.34, y = 0.38, which are close to the equal-energy white point. The single emitting layer of the donor-acceptor blends does not always emit the combination of a molecule emission band and an exciplex emission band. J. Kalinowski et al. demonstrated white EL emission comprised of exciplex emission, molecule emission, and excimer emission as shown in Figure 8(a) [89]. The three different emissions came from the blends of m-MTDATA/platinum [methyl-3,5-di-(2-pyridyl) benzoate] chloride (PtL 2 Cl). m-MTDATA and PtL 2 Cl form the exciplex. However, the exciplex formation was incomplete, and the phosphorescent moleculeand excimer emission from PtL 2 Cl were also implemented in the WOLED whose structure is shown in Figure 8(b) along with the molecuar structures. The CIE coordinates were x = 0.46, y = 0.45 and the color rendering index of CRI was 90 which was a high value given that CRI of 100 is for ideal white light. The maximum EQE was 6.5% as shown in Figure 8(c). The broad featureless emissions from the excimer and exciplex as well as the emissions from three different sources with a different peak wavelength contribute to the good color quality of the device.

Conclusion
In this review paper, the basic principles of exciplexes coming from its CT character and the characteristics of OLEDs using exciplexes are explained. Exciplexes show slower radiative transition rate than conventional fluorescent molecules due to their small magnitude of transition dipole moments caused by their CT character, which typically induce the low PLQY. In spite of low PLQY of exciplexes, OLEDs employing exciplexes as emitters have been studied actively due to their potential of achieving the IQE of 100% by virtue of the large RISC efficiency in exciplex-forming systems. The large RISC efficiency could be achieved when the singlet-triplet splitting is small, inducing effective state mixing between singlet and triplet states. Even though the PLQY of the exciplex was reported to be 100%, the performance of OLEDs employing exciplexes is still lacking compared to phosphorescent OLEDs with the EQE over 30%. The more study would be needed to understand the transition rates in exciplexforming systems from molecular structures. Then, the molecular design for the exciplex with the high PLQY will be available, and it will acceleratethe improvement on the performance of the exciplex-emitting OLEDs.
KAST in 2013, and Excellence in Research Award from Seoul National University in 2017.