Recent advances in materials for organic light emitting diodes

The design of highly emissive and stable blue emitters for organic light emitting diodes (OLEDs) is still a challenge, justifying the intense research activity of the scientific community in this field. Recently, a great deal of interest has been devoted to the elaboration of emitters exhibiting a thermally activated delayed fluorescence (TADF). By a specific molecular design consisting into a minimal overlap between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) due to a spatial separation of the electron-donating and the electron-releasing parts, luminescent materials exhibiting small S1–T1 energy splitting could be obtained, enabling to thermally upconvert the electrons from the triplet to the singlet excited states by reverse intersystem crossing (RISC). By harvesting both singlet and triplet excitons for light emission, OLEDs competing and sometimes overcoming the performance of phosphorescence-based OLEDs could be fabricated, justifying the interest for this new family of materials massively popularized by Chihaya Adachi since 2012. In this review, we proposed to focus on the recent advances in the molecular design of blue TADF emitters for OLEDs during the last few years. Introduction Since the pioneering works of Tang and VanSlyke in 1987 [1], organic light emitting diodes (OLEDs) have known major evolutions of their structures, not only of the device stacking but also of the materials composing the different layers [2]. The interest of both the scientific and industrial communities for organic electroluminescent devices is supported by the fact that Beilstein J. Org. Chem. 2018, 14, 282–308. 283 Figure 1: Radiative deactivation pathways existing in fluorescent, phosphorescent and TADF materials. OLEDs have been identified as the key-elements for the fabrication of the next generation display and lighting technology [3]. Notably, lightweight and thin devices can be fabricated onto flexible substrates, favouring the penetration of OLEDs in these two markets. With the aim at reducing the global energy demand on Earth, two parameters govern the power consumption of OLEDs, namely the quantum yield of luminescence of the light emitting material and the device stacking. Indeed, the driving voltage of OLEDs is highly sensitive to the thickness of the different layers, the charge transport ability of the materials but also to their energy levels. By minimizing the energy gaps between adjacent layers and facilitating charge injection from the electrodes, the injection and transportation of holes and electrons can be realized at lower operating voltages. The second parameter concerns the light-emitting ability of the emitter, which is directly related to the nature, and the photoluminescence quantum yield (PLQY) of the emitter. Based on spin statistics, upon electrical excitation, singlet and triplet excitons are formed in a 1:3 ratio [4]. In the case of fluorescent materials, only singlet excitons can be utilized for light emission, limiting the internal quantum efficiency (IQE) of fluorescent OLEDs to 25%. Conversely, phosphorescent materials can both harvest singlet and triplet excitons for emission by intersystem crossing (ISC), enabling to reach a theoretical IQE of 100% for phosphorescent OLEDs [5]. As drawback, triplet emitters are transition-metal complexes mostly based on iridium, platinum and osmium and the scarcity of these metals on Earth, their toxicity and high cost make these materials unsuitable candidates for a mass-production of OLEDs [6]. However, efforts have also been carried out to incorporate emitters comprising less toxic metals, providing mitigate results when tested in devices [7,8]. In 2012, a breakthrough has been obtained by the Adachi group who developed purely organic materials capable to harvest both singlet and triplet excitons for emission [9]. This new family of light emitting materials capable to compete with the well-established triplet emitters and displaying a similar efficiency in devices by developing a new emission mechanism was immediately termed as the third generation of OLEDs emitters that consists of thermally activated delayed fluorescence (TADF) emitters. As specificity, these materials can thermally repopulate the singlet state from the triplet state by reverse intersystem crossing (RISC), leading to an increase of the luminescence intensity. From the OLEDs viewpoint, TADF emitters behave by harvesting both singlet and triplet excitons for radiative transition, excepted that the emission occurs from the singlet state and not from the triplet state (as observed for metal complexes) and that the triplet–triplet annihilation commonly observed with phosphorescent OLEDs [10] can be drastically reduced (see Figure 1). TADF materials can also be metal-free, addressing the fabrication cost and environmental issues. Therefore, TADF emitters retain the high efficiency of the second generation of emitters (triplet emitters), the stability of the first generation of fluorescent materials while eliminating the different problems observed with the two previous generations: triplet–triplet annihilation and low device stability for phosphorescent OLEDs, low IQE for fluorescent OLEDs. To get full-color displays or white-light OLEDs, the combination of the three primary colors red green blue (RGB) is indispensable. At present, highly emissive and stable blue emitters are actively researched [11-16]. Several points justify the low availability of highly efficient blue materials. First, due to their large bandgaps (ΔE > 3 eV), charge injection from the adjacent Beilstein J. Org. Chem. 2018, 14, 282–308. 284 layers is often difficult, requiring devices to be operated at high voltages [17]. Second, and still related to their wide bandgaps, the probability to transfer an electron from the ground to the excited stable state is considerably reduced, providing materials displaying PLQY greatly reduced compared to that observed for the other colors [18,19]. To end, the propensity of blue emitters to rapidly degradate upon device operation is well established, resulting in a fast and irreversible color shift [20,21]. In this context, TADF blue emitters have been identified as promising candidates to address the color purity, quantum efficiency and long-term device stability issues. Due to the enthusiasm of the scientific community for TADF emitters, this research field evolves extremely rapidly. In this review, a summary of the strategies developed during the last years to design organic blue TADF emitters is presented. It has to be noticed that the values of EQEs reported in the different tables correspond to the maximum EQEs, because many articles do not give sufficient data concerning EQE at the practically relevant luminance of 100 cd/m2. Review 1. Molecular design rules to produce a delayed fluorescence The efficiency of OLEDs is intimately related to the ability of the light-emitting materials to convert a maximum of injected charges into photons. To optimize this, the TADF process is the most promising strategy as it allows converting the generated and lost triplet excitons of the classical fluorescent materials into emissive singlets. By efficiently upconverting the triplet excitons from the triplet (T1) to the singlet state (S1), the intrinsic limitation of 25% imposed to fluorescent materials by the 1:3 singlet–triplet ratio can be overcome and an ultimate IQE of 100% can be realized with TADF materials. To promote the endothermic RISC, the energy gap between S1 and T1 (ΔEST) plays a key role and should be as small as possible. From a molecular design viewpoint, ΔEST can be drastically reduced if the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) are spatially separated, what can be obtained by a suitable steric hindrance that introduces an internal twist and interrupts the π-conjugation but also by a sufficient distance between the electron-donating and the electron-accepting moieties [22-25]. In the design of TADF materials, it should be mentioned the major importance of the spin–orbit vibronic coupling, in addition to the small ΔEST. Indeed, a small ΔEST is not sufficient to ensure for a TADF material an efficient RISC which is a vibronically coupled, spin–orbit coupling process with the involvement of the charge transfer state. To remain efficient, the spin–orbit coupling should still have a significant value, even if the separation of the HOMO and LUMO wavefunctions remain a requirement to minimize ΔEST. At present, systematic investigations examining the correlation between the spin–orbit coupling and RISC are stil l scarce [26-29]. Considering that the singlet–triplet energy splitting is one of the key elements for controlling the RISC efficiency, that the dihedral angle between the donor and the acceptor can be difficultly anticipated and that an overlap of both the HOMO/LUMO energy levels could adversely affect the color purity and ΔEST, it has to be noticed that the photophysical properties and the geometry of molecules that are suspected to be TADF emitters are often investigated by theoretical calculations prior to synthesis, optimizing the chance to get suitable energy levels and the desired ΔEST. This strategy was notably applied to the design of TADF blue emitters containing triarylboron accepting units. Besides, as we will see in this review, the design of a good TADF material by optimizing its structure by theoretical calculations is not sufficient to ensure the fabrication of highly emissive OLEDs. As observed for phosphorescent emitters, optimization of the device stacking, an appropriate choice of the host as well as the materials in the adjacent layers, an adequate dopant concentration, and the efficient confinement of excitons within the emissive layer are primordial parameters to elaborate high performance OLEDs while maintaining the color purity [30]. Due to the

Organic light emitting diodes (OLEDs) are at the cusp of becoming the dominant technology for the mobile device and display markets. This is largely due to the concerted efforts over the past thirty years to improve the material design, beginning with the first demonstrated viability of this technology in 1987 by Tang and VanSlyke [1]. Emitters in particular have undergone an evolution in design, from fluorescent compounds to phosphorescent organometallic complexes to organic thermally activated delayed fluorescence (TADF) molecules, the latter driving tremendous recent excitement within the field of organic semiconductor research. This thematic issue of the Beilstein Journal of Organic Chemistry covers novel phosphorescent and TADF materials design and their inclusion as emitters in OLEDs.
Some highlights in this issue include the work of Thanh-Tuân Bui et al., who provide a welcome perspective on blue TADF materials for OLEDs in the form of a review article [2]. Cristina Cebrián and Matteo Mauro review the advances made in platinum(II) complexes for OLEDs [3]. Rebecca Pittkowski and Thomas Strassner report bright blue-to-blue-green phosphorescent platinum(II) complexes employing sterically bulky diketonate ancillary ligands [4]. In addition, Lin Gan et al. describe a new molecular design approach for orange-emitting TADF molecules employing a fluorenone acceptor [5]. In the full research paper by Feng-Ming Xie et al., they disclose two bipolar, high-energy phenothiazine-5,5-dioxide-based host materials conceived to be used for deep blue OLED devices [6].
The articles in this thematic issue provide a window into the design principles used towards the development of next-generation emitter and host materials for OLEDs. I hope these articles will provide inspiration for further research in this exciting area. the rational design of: (i) luminophores for use in organic light emitting diodes (OLEDs) and light-emitting electrochemical cells (LEECs), two types of electroluminescent devices; (ii) sensing materials employed in electro-chemiluminescence; and (iii) photocatalysts employed in photoredox catalytic reactions.

Introduction
Since the pioneering works of Tang and VanSlyke in 1987 [1], organic light emitting diodes (OLEDs) have known major evolutions of their structures, not only of the device stacking but also of the materials composing the different layers [2]. The interest of both the scientific and industrial communities for organic electroluminescent devices is supported by the fact that OLEDs have been identified as the key-elements for the fabrication of the next generation display and lighting technology [3]. Notably, lightweight and thin devices can be fabricated onto flexible substrates, favouring the penetration of OLEDs in these two markets. With the aim at reducing the global energy demand on Earth, two parameters govern the power consumption of OLEDs, namely the quantum yield of luminescence of the light emitting material and the device stacking. Indeed, the driving voltage of OLEDs is highly sensitive to the thickness of the different layers, the charge transport ability of the materials but also to their energy levels. By minimizing the energy gaps between adjacent layers and facilitating charge injection from the electrodes, the injection and transportation of holes and electrons can be realized at lower operating voltages. The second parameter concerns the light-emitting ability of the emitter, which is directly related to the nature, and the photoluminescence quantum yield (PLQY) of the emitter. Based on spin statistics, upon electrical excitation, singlet and triplet excitons are formed in a 1:3 ratio [4]. In the case of fluorescent materials, only singlet excitons can be utilized for light emission, limiting the internal quantum efficiency (IQE) of fluorescent OLEDs to 25%. Conversely, phosphorescent materials can both harvest singlet and triplet excitons for emission by intersystem crossing (ISC), enabling to reach a theoretical IQE of 100% for phosphorescent OLEDs [5]. As drawback, triplet emitters are transition-metal complexes mostly based on iridium, platinum and osmium and the scarcity of these metals on Earth, their toxicity and high cost make these materials unsuitable candidates for a mass-production of OLEDs [6]. However, efforts have also been carried out to incorporate emitters comprising less toxic metals, providing mitigate results when tested in devices [7,8]. In 2012, a breakthrough has been obtained by the Adachi group who developed purely organic materials capable to harvest both singlet and triplet excitons for emission [9]. This new family of light emitting materials capable to compete with the well-established triplet emitters and displaying a similar efficiency in devices by developing a new emission mechanism was immediately termed as the third generation of OLEDs emitters that consists of thermally activated delayed fluorescence (TADF) emitters. As specificity, these materials can thermally repopulate the singlet state from the triplet state by reverse intersystem crossing (RISC), leading to an increase of the luminescence intensity. From the OLEDs viewpoint, TADF emitters behave by harvesting both singlet and triplet excitons for radiative transition, excepted that the emission occurs from the singlet state and not from the triplet state (as observed for metal complexes) and that the triplet-triplet annihilation commonly observed with phosphorescent OLEDs [10] can be drastically reduced (see Figure 1). TADF materials can also be metal-free, addressing the fabrication cost and environmental issues. Therefore, TADF emitters retain the high efficiency of the second generation of emitters (triplet emitters), the stability of the first generation of fluorescent materials while eliminating the different problems observed with the two previous generations: triplet-triplet annihilation and low device stability for phosphorescent OLEDs, low IQE for fluorescent OLEDs.
To get full-color displays or white-light OLEDs, the combination of the three primary colors red green blue (RGB) is indispensable. At present, highly emissive and stable blue emitters are actively researched [11][12][13][14][15][16]. Several points justify the low availability of highly efficient blue materials. First, due to their large bandgaps (ΔE > 3 eV), charge injection from the adjacent layers is often difficult, requiring devices to be operated at high voltages [17]. Second, and still related to their wide bandgaps, the probability to transfer an electron from the ground to the excited stable state is considerably reduced, providing materials displaying PLQY greatly reduced compared to that observed for the other colors [18,19]. To end, the propensity of blue emitters to rapidly degradate upon device operation is well established, resulting in a fast and irreversible color shift [20,21]. In this context, TADF blue emitters have been identified as promising candidates to address the color purity, quantum efficiency and long-term device stability issues. Due to the enthusiasm of the scientific community for TADF emitters, this research field evolves extremely rapidly. In this review, a summary of the strategies developed during the last years to design organic blue TADF emitters is presented. It has to be noticed that the values of EQEs reported in the different tables correspond to the maximum EQEs, because many articles do not give sufficient data concerning EQE at the practically relevant luminance of 100 cd/m 2 .

Review 1. Molecular design rules to produce a delayed fluorescence
The efficiency of OLEDs is intimately related to the ability of the light-emitting materials to convert a maximum of injected charges into photons. To optimize this, the TADF process is the most promising strategy as it allows converting the generated and lost triplet excitons of the classical fluorescent materials into emissive singlets. By efficiently upconverting the triplet excitons from the triplet (T 1 ) to the singlet state (S 1 ), the intrinsic limitation of 25% imposed to fluorescent materials by the 1:3 singlet-triplet ratio can be overcome and an ultimate IQE of 100% can be realized with TADF materials. To promote the endothermic RISC, the energy gap between S 1 and T 1 (ΔE ST ) plays a key role and should be as small as possible. From a molecular design viewpoint, ΔE ST can be drastically reduced if the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) are spatially separated, what can be obtained by a suitable steric hindrance that introduces an internal twist and interrupts the π-conjugation but also by a sufficient distance between the electron-donating and the electron-accepting moieties [22][23][24][25]. In the design of TADF materials, it should be mentioned the major importance of the spin-orbit vibronic coupling, in addition to the small ΔE ST . Indeed, a small ΔE ST is not sufficient to ensure for a TADF material an efficient RISC which is a vibronically coupled, spin-orbit coupling process with the involvement of the charge transfer state. To remain efficient, the spin-orbit coupling should still have a significant value, even if the separation of the HOMO and LUMO wavefunctions remain a requirement to minimize ΔE ST . At present, systematic investigations examining the correlation between the spin-orbit coupling and RISC are still scarce [26][27][28][29]. Considering that the singlet-triplet energy splitting is one of the key elements for controlling the RISC efficiency, that the dihedral angle between the donor and the acceptor can be difficultly anticipated and that an overlap of both the HOMO/LUMO energy levels could adversely affect the color purity and ΔE ST , it has to be noticed that the photophysical properties and the geometry of molecules that are suspected to be TADF emitters are often investigated by theoretical calculations prior to synthesis, optimizing the chance to get suitable energy levels and the desired ΔE ST . This strategy was notably applied to the design of TADF blue emitters containing triarylboron accepting units. Besides, as we will see in this review, the design of a good TADF material by optimizing its structure by theoretical calculations is not sufficient to ensure the fabrication of highly emissive OLEDs. As observed for phosphorescent emitters, optimization of the device stacking, an appropriate choice of the host as well as the materials in the adjacent layers, an adequate dopant concentration, and the efficient confinement of excitons within the emissive layer are primordial parameters to elaborate high performance OLEDs while maintaining the color purity [30]. Due to the difficulty to address simultaneously these different points, numerous light emitting materials have been revisited several times, providing different electrical and optical device characteristics.

Boron-containing TADF emitters
Boron-containing molecules have been extensively investigated in organic electronics [31] as these materials are characterized by a remarkable electron mobility resulting from the presence of a vacant p-orbital on the boron atom [32,33]. Triarylboron compounds are also strong electron acceptors, justifying that numerous groups developed TADF emitters using triarylboron moieties as acceptors. As possible donors, diarylamines have often been proposed (carbazole, triphenylamine, carbazole/triphenylamine hybrids, 9,9-dimethyl-9,10-dihydroacridine), as exemplified in Figure 2 [34][35][36]. In B1 and B2, isolation of the two parts was obtained by linking the 10H-phenoxaborin unit or the dimesitylphenylboron moiety to the 9,9-dimethyl-9,10-dihydroacridine part through a phenylene bridge substituted at the 1,4-positions. By mean of steric repulsions occurring between the hydrogen atoms of the aromatic π-bridge and those of the neighbouring electron-donating and accepting parts, an effective spatial separation of the HOMO and LUMO levels could be obtained, resulting in the rotation of the two end-groups relative to the plane of the central aromatic ring. A dihedral angle of 51.8° was found between the phenylene and the 10H-phenoxaborin unit in B1, increasing to 88.4° for the dihedral angle between the phenylene and the 9,9-dimethylacridane unit in B2. ΔE ST values of 0.013 eV and 0.041 eV were experimentally de- termined for B1 and B2, respectively, calculated from the difference existing between the onset of the fluorescence and the phosphorescence emission. The decay time of the delayed component of luminescence was determined as being 2.36 μs and 6.71 μs for B1 and B2, respectively. When evaluated in multilayered OLEDs, a blue electroluminescence (EL) peaking at 466 nm and 479 nm, an external quantum efficiency (EQE) of 15.1% and 16.0% were obtained for B1 and B2, respectively, indicating the substantial contribution of the triplet excitons to the luminescence.
Interestingly, compared to B2, the introduction of two additional methyl groups in the phenyl part (B3) resulted in a clear bathochromic shift of the EL, OLEDs emitting a green light peaking at 502 nm [37]. A blue shift of the emission and skyblue OLEDs could only be obtained with this acceptor by replacing the electron-donating 9,9-dimethyl-9,10-dihydroacridinyl unit of B3 by a bis(diphenylamino)carbazole group in B4 or a diphenylaminocarbazole unit in B5. The outstanding EQE of 21.6% could be attained for the sky-blue B4-based devices. Still based on the combination of acridan and 10Hphenoxaborin units, a complete isolation of the two units could be realized in B6 by directly functionalizing the 10H-phenoxaborin core with a spiro-type acridan group [38]. Using this strategy, pure blue OLEDs exhibiting an EQE of 19.0% and Commission Internationale de l'Eclairage (CIE) coordinates of (0.14, 0.16) were obtained with B6. Comparable performances were determined for B7 (20.1%, (0.14, 0.16)), comprising the sterically demanding tetramethylcarbazole. In these two structures, a large dihedral angle arising from steric repulsions between hydrogen atoms in the peri-position of B6 and from the presence of methyl groups at the 3,6-positions of 1,3,6,8-tetramethylcarbazole in B7 could be obtained. In fact, the substitution at the 3,6-position of carbazole could maintain a large dihedral angle in B7 whereas the two methyl groups at the 1,8-positions were introduced for a higher electrochemical stability of the carbazole donor. Finally, by modifying the connectivity between the donor and acceptor in B8, a record-high EQE of 24.1% could be realized for pure-blue OLEDs (0.139, 0.150) close to the National Television Standards Committee standard (NTSC) blue values of (0.14, 0.08) [36]. Upon ortho-substitution of the dimesitylphenylboron acceptor with a carbazole, a mutual steric hindrance could be exerted between the donor and the acceptor resulting in the large dihedral angle of 72.6°. A S 1 -T 1 energy splitting of 0.13 eV could be also experimentally determined for B8. Interestingly, the outstanding EL characteristics of B8-based devices were assigned to the large contribution of the delayed fluorescence (61%) in the overall luminescence decay of B8. A pure blue emission could also be realized by totally blocking the structure, what was done with B9 and B10 in which two of the three aromatic rings of triphenylamine were connected to the boron center [39]. By elongating the π-conjugation of the electron-donating group in B10 compared to B9, a more delocalized HOMO level could be generated, resulting in a greater intramolecular charge transfer and an increase of the oscillator strength. As a result, EQE of corresponding OLEDs increased from 13.5% (459 nm, (0.13, 0.09)) for B9-based devices to 20.2% (467 nm, (0.12, 0.13)) for B10based devices. If the electron-to-photon conversions are remark-able, none of the OLEDs could reach the brightness of 1000 cd/m 2 owing to a dramatic efficiency roll-off. Precisely, the efficiency roll-off determined for B9-and B10-based devices was determined as originating from an imbalanced charge transportation and the presence of bimolecular quenching processes occurring at high current density such as triplet-triplet annihilation and exciton-polaron annihilation.

Diphenylsulfone-based emitters
Concerning the design of blue TADF emitters, diphenylsulfone is the third most widely studied acceptor in the literature, followed by triarylboron and triazine derivatives. In this field, the contribution of the Adachi's group is remarkable. The first report mentioning a pure blue emission with a diphenylsulfone derivative was reported in 2012 [40]. By a careful control of the π-conjugation length between the donor and the acceptor, D3-based OLEDs producing a deep blue emission with CIE coordinates of (0.15, 0.07) were fabricated (see Figure 3). Examination of the phosphorescence spectra of D1-D3 at 77 K revealed their T 1 states to be 3 ππ* states centred on their electron-donating parts. ΔE ST values of 0.54, 0.45 and 0.32 eV were, respectively, determined for D1-D3. Changes in ΔE ST were explained as follow: By introducing tert-butyl groups on the diphenylamine unit, the electron donating ability in D2 was reinforced compared to D1, red-shifting the charge transfer (CT) band and lowering the CT energy as well as ΔE ST . By replacing the diphenylamine unit of D1 by a tert-butyl-substituted carbazole unit in D3, the 3 ππ* state was considerably destabilized, raising its energy level and decreasing ΔE ST . Parallel to this, a greater separation of the HOMO and LUMO orbitals was evidenced by theoretical calculations for D3, as a result of a larger dihedral angle (49° instead of 32° for D1 and D2), resulting in a smaller energy difference between the singlet and triplet excited states. As expected, the contribution of the slow decay component in the luminescence of D1-D3 decreased with increasing ΔE ST , almost disappearing for D1. While using D1-D3 as dopants for multilayer OLEDs, maximum EQEs of OLEDs coincide the order previously determined for the proportion of the delayed component in the total emission of D1-D3 with the EQE (D1) < EQE (D2) < EQE (D3) (2.9%, 5.6% and 9.9% for D1-D3, respectively). If D3 displayed the best EQE for the series, a dramatic efficiency rolloff at high current density was observed, as the result of a long TADF lifetime (270 μs). This issue was addressed with D4 [41]. By replacing the tert-butyl groups of D3 by methoxy groups in D4, a significant decrease of ΔE ST was obtained (0.21 eV instead of 0.32 eV for D3), reducing the TADF lifetime and efficiency roll-off. More precisely, the higher electron-donating ability and the longer conjugation length of the 3,6-dimethoxycarbazole compared to the 3,6-di-tert-butylcarbazole lowered the S 1 state and to a greater extend the T 1 state of D4, furnishing in turn a molecule with a smaller ΔE ST than D3. Jointly, due to the reduction of ΔE ST , a TADF lifetime of 93 μs was determined for D4, far from the value measured for D3 (270 μs). When tested in a similar device structure than that previously used for D3, a maximum EQE of 14.5% and a smaller efficiency roll-off was evidenced for D4-based devices, attributed to the smaller ΔE ST and the shorter TADF lifetime. Recently, a thermally cross-linkable and solution-processable version of D4, i.e., D5 was reported in the literature [42]. If the strategy is appealing, the final EL performances of D5-based OLEDs were far from that obtained with vacuum-processed OLEDs and a maximum EQE of only 2.0% could be reached.
Following the basic rule of molecular design consisting in maximizing the dihedral angle to minimize ΔE ST , substitution of diphenylsulfone by 9,9-dimethyl-9,10-dihydroacridine resulting in an almost orthogonality of the two groups in D6 as a dihedral angle as large as 89° could be determined between 9,9dimethyl-9,10-dihydroacridine and the connected phenyl ring of the diphenylsulfone unit [43].
A significant reduction of the TADF lifetime (≈7 μs) and a small ΔE ST of 0.08 eV were measured for D6, favorable to the fabrication of highly emissive blue OLEDs. Devices fabricated with D6 furnished a maximum EQE of 19.5% and maintained the high EQE of 16% at 1000 cd/m 2 with a satisfactory color purity of coordinates (0. 16, 0.20). Recently, high-performance TADF based hybrid WOLEDs employing D6 as the blue emitter were successfully fabricated [44]. Interestingly, WOLEDs showed excellent device characteristics with an EQE of 23.0%, a current and power efficiency of 51.0 cd/A and 51.7 lm/W, respectively. These performances are among the highest values reported to date for hybrid WOLEDs using a TADF material as the blue emitter. Derivative D6 was also examined in the context of undoped OLEDs [45]. Undoped OLEDs are more attractive than their doped analogues due to an easier fabrication process, a higher reproducibility and reliability. With regards to the highly twisted structure of D6 and the presence of methyl groups on the 9,9-dimethyl-9,10-dihydroacridine units, this molecule proved to be also nearly insensitive to the concentration, showing an emission maximum for the neat film at 470 nm which is almost similar to that obtained for a 10 wt %-doped mCP film (462 nm where mCP stands for m-bis(N-carbazolyl)benzene). Parallel to this, the fluorescence and TADF lifetime were almost the same for both the doped and undoped film, making D6 a candidate applicable for the design of undoped OLEDs. Trilayered undoped OLEDs fabricated with D6 displayed a sky-blue emission peaking at 480 nm, with an EQE of 19.5% at a luminance of 100 cd/m 2 , slightly red-shifted compared to the emission observed for doped OLEDs. Clearly, the specific design of D6 and its highly twisted structure efficiently weakened the π-π-stacking interactions, providing a general design rule for the elaboration of TADF emitters insensitive to the concentration. Belonging to the same family of structure than D6, D7 that derives from the 9,9-dimethylthioxanthene-S,S-dioxide structure provided a better color purity (465 nm, (0.16, 0.24) for D7 instead of 480 nm for D6) and a higher EQE (22.4% for D7 instead of 19.5% for D6) than D6 by optimizing the architecture of the doped EML [29]. By selecting the host of appropriate polarity, the combination of D7 with the correct host could minimize the RISC barrier, optimize the RICS rate and thus maximize the TADF efficiency. While combining the blue TADF emitter D7 with a green and an orange TADF emitter, all-TADF white OLEDs with 16% EQE could be fabricated [30].

Triazine-pyrimidine based emitters
Among possible electron acceptors, another structure has been extensively regarded as an adequate electron acceptor for the design of blue TADF emitters and this structure is the triazine unit. When combined with the azasiline donor, OLEDs displaying the unprecedented EQE of 22.3% were obtained [46]. As specificity, azasiline is a 6-membered heterocycle comprising a silicon atom introduced instead of a carbon atom to enlarge the HOMO-LUMO gap and lower the HOMO level. Due to the sp 3 hybridization of the silicon atom, two phenyl rings can be introduced on the silicon-bridged structure providing bulkiness and rigidity to the donor. Intermolecular interactions are thus efficiently prevented and the conformation disorder drastically reduced. When used as electron donor in T1, a ΔE ST of 0.14 eV was determined experimentally, with a TADF lifetime of 25.4 μs and a 13:87 ratio between the prompt and delayed fluorescence. OLEDs fabricated with T1 and a mCP:TSPO1 cohost (with TSPO1 = diphenyl-4-(triphenylsilyl)phenylphosphine oxide) furnished a blue emission peaking at 463 nm, with CIE coordinates of (0.149, 0.197) and a low efficiency roll-off. Another key and general design rule for obtaining a small ΔE ST consists in the physical separation of the donor and the acceptor by elongating the spacer that couples the two partners. Following this recommendation, an additional phenyl ring was introduced between the donor and the acceptor in T2, providing the extended version of T1 (see Figure 4) [47]. As expected, the phenyl ring increased the separation of the HOMO and LUMO orbitals, such that ΔE ST decreased. A value as low as 0.04 eV was experimentally determined for T2. In doped devices, T2 demonstrated an EL efficiency of 4.7% with a deep blue emission (0.151, 0.087) approaching the NTSC blue standard (0.14, 0.08). However, a comparison with the previous EL performance evidenced that EQEs obtained with T2 are 5 times lower than that obtained with T1, despites the more favorable S 1 -T 1 energy splitting. This problem is commonly observed if the isolation of the electron-donating and electronaccepting parts is obtained upon extension of the distance between the two moieties. Indeed, as a consequence of this strategy, a weaker intramolecular charge transfer takes place and a reduction of the oscillator strength in the D-A diad is observed, resulting in a drastic reduction of the PLQY and thus of the external quantum efficiency. In the same study, authors examined the case of two TADF emitters based on a donor-acceptor-donor (D-A-D) structure, i.e., T3 and T4, where azasiline was used as the donor and diphenylsulfone or benzophenone as the acceptors. Here again, the higher twisted molecular structure of T4 was beneficial in terms of ΔE ST , color purity and EL performances. Thus, the higher internal torsion of T4 furnished OLEDs with a deeper blue emission (0.154, 0.107) than devices fabricated with T3 (0.174, 0.310). Even if the EQE of T4-based devices was lower than that of T3-based devices (2.3% for T4-based OLEDs instead of 11.4% for T3-based devices), it is attributable to the higher color purity of T4-based devices and not to differences of ΔE ST (0.07 eV and 0.06 eV for T3 and T4, respectively). Azasiline is a promising electron donor but examples of blue TADF emitters are still scarce. The opposite situation is found for carbazole, which has long been considered as an excellent donor and a large variety of blue TADF emitters have been designed on the basis of this scaffold. At least 19 examples of blue TADF emitters can be cited, the molecules differing by the strategy used to connect the donor(s) to triazine. However, contrarily to azasiline that possesses a six-membered central ring, carbazole only possesses a five-membered central ring, inducing a deviation of the two adjacent aromatic rings. As a result, carbazole is not capable to induce the same encumbrance as that of azasiline by inducing smaller steric effects and the substitution of the 1,8-positions is often required to maintain a large dihedral angle.
As interesting design rules, Adachi determined that the extension of the electronic delocalization of both the HOMO and LUMO energy levels could greatly increase the rate of the radiative decay by inducing a large oscillator strength while lowering ΔE ST , even for emitters for which only a small overlap between the two wavefunctions is observed [48]. These findings constitute a second guideline for the molecular design of TADF emitters that can address the distance and the reduction of the oscillator strength issue previously mentioned. To establish this, a series of molecules T5-T8 with varying length of the π-conjugated system for the donating part was investigated. Thus, for T5 and T6, a similar ΔE ST value of 0.09-0.12 eV was experimentally determined for the two emitters. However, significant differences were determined for their PLQYs and values of 0.1 and 0.7 were measured for T5 and T6, respectively. By theoretical calculations, the oscillator strength of T6 was found to be 13.6 times greater than that of T5, supporting the enhanced luminescence of T6 by the higher delocalization of its HOMO level. This trend was confirmed by keeping the acceptor constant in T6-T8. An increase of ΔE ST while reducing the possible electronic delocalization over the electron-donating part was clearly evidenced going from T6 to T8. In OLEDs, EL performances followed the same trend, with the highest EQE obtained with T6 (EQE = 20.6%) and the lowest one with T8 (EQE = 14.6%). A lower color purity was obtained for T6-based devices (λ EL = 487 nm) compared to T7 and T8 (λ EL = 478 and 477 nm, respectively) [22]. A worse result was obtained for T5 that produced a blue-green EL at 506 nm. Recently, an extensive work was devoted to examine the degradation mechanisms in blue TADF OLEDs and T7 was revisited in this context [49]. The synergy of an electro-oxidation process together with a photo-oxidation was determined as playing a critical role in the degradation of blue TADF emitters. In fact, a parallel can be easily done with the treatment of wastewater, where pollutants are removed from water by combining a photochemical and an electrochemical process [50]. During this study, the localization of the triplet spin density was found determinant for the stability of blue TADF emitters. To evidence this, four emitters (T7, T9-T11) exhibiting the same S 1 and T 1 energy levels, the same TADF lifetimes but differing by the distribution of the triplet spin densities were examined (see Figure 4 and Figure 5). Notably, for T9, the triplet spin density was found to be mainly localized on the bicarbazole donor, whereas for T7 and T10, the triplet spin density is localized on their acceptor fragment. To end, the triplet spin density of T11 is delocalized over the entire structure. While examining the device lifetime, T9-based devices had the longest device lifetime (32 hours), far from T10-, T7-and T11-based OLEDs (1.4 h, 2.8 h and 0.9 h, respectively), demonstrating the higher stability of the emitters with a triplet spin density centered onto the donor unit. In another study, an analogue of T9, i.e., T12, differing by the removal of a phenyl ring between the carbazole and the triazine units proved once again the crucial role of the oscillator strength in the photophysical properties [51]. Notably, major differences in the separation of their HOMO and LUMO energy levels were determined by theoretical calculations. An overlap of the two electronic wavefunctions was detected for T9 whereas the two orbitals are strongly localized in the case of T12. Resulting from this marked localization in T12, a smaller variation of the electronic density upon excitation is expected, reducing the oscillator strength and the PLQY. When tested in devices, only a green-blue emission was obtained with T12 (see Figure 5) [52]. The Influence of the oscillator strength on OLEDs characteristics could also be evidenced while comparing T13 and T14 [53]. Molecular orbital calculations performed on T13 and T14 showed the two molecules to exhibit a similar electronic distribution, what was confirmed by UV-visible and photoluminescence (PL) spectroscopy. Only a slight red shift of the absorption was detected for T14 as the result of the strengthened donating ability of the dicarbazolylphenyl moieties. Similarly, almost identical ΔE ST were determined with values of 0.25 and 0.27 eV for T13 and T14, respectively). As it could be anticipated, T14 furnished slightly better EL performances (18.9%) compared to that measured for T13 (17.8%), due to its more extended donating part but also owing to its higher PLQY. Conversely, the color purity was higher for T13-based devices (λ EL = 459 nm) instead of 467 nm for T14-based devices. However, a remarkable device stability was demonstrated for T14-based OLEDs, 80% of the initial luminance being retained after 52 hours. This value was reduced to only 5 hours for T13-based OLEDs. A comparison established with an iridium complex, i.e., tris [1-(2,4- 3 ) evidenced the relevance of the TADF approach, as a device lifetime of only 18 hours was found while operating OLEDs in the same conditions. The spatial separation of the electron-donating part from the electron-accepting moiety by elongating the spacer has already been discussed and the drawbacks evoked.
Minimization of the electron density overlap can also be realized by means of an ortho-phenyl linkage, enabling to maintain the donor in proximity of the acceptor.
In this situation, one aromatic ring of the donor and/or the acceptor is substituted at the 1,2-positions, generating a highlytwisted structure. Five blue TADF emitters T15-T19 were designed on this basis (see Figure 6). By increasing the number of carbazoles in T16 compared to T15, a decrease of ΔE ST was logically observed (0.06 eV for T15 and 0.03 eV for T16) [54]. A large torsion angle of 66° and 67° were, respectively, determined by theoretical calculations for T15 and T16, favorable to the separation of the two orbitals. In devices, a remarkable enhancement of the EL performances was realized by increasing the number of carbazole units. Thus, a maximum EQE of 12.2% was realized with T15, whereas an EQE of 16.5% was determined for T16-based devices.
This enhancement can also be attributable to an increase of the oscillator strength from T15 to T16, the number of donors being increased. The low efficiency roll-off of T16-based devices was assigned to the specific design of the emitter, with the triazine acceptor being totally surrounded by carbazoles. As a result, triplet-triplet annihilation by the Dexter mechanism could be efficiently prevented, enabling to maintain high efficiencies at high current density. Although the number of carbazole units increased, no modification of the EL position was detected, the emission peaking at 467 and 468 nm for T15and T16-based devices. In the same spirit, other authors examined the possible impact of the substitution pattern of the carbazole unit on the photophysical properties.
While maintaining the same number of carbazole units on the emitter and by varying the substitution pattern of the carbazole core, only a weak influence on the EL characteristics was evidenced [55]. In fact, performances only varied by their differences of PLQYs (16.7%, 50.5% and 43.0% for T17, T18 and T19, respectively), the three molecules exhibiting similar photophysical properties (ΔE ST , emission wavelength and decay times of the delayed emission). Recently, a potential alternative to the ortho-substitution of the triazine acceptor by carbazole moieties was examined, consisting in introducing methyl groups in the proper position of the triazine or the carbazole moieties  [56]. By changing the methyl group positions, optical properties of T20-T23 were not significantly modified, contrarily to their ΔE ST (see Figure 7). In fact, the authors evidenced the introduction of methyl groups at the 1,8-positions of carbazole to be harmful for producing a deep-blue emission whereas the substitution of the central phenyl ring by methyl groups could provide the same molecular twist than the 1,8-substitution of carbazole while maintaining a large optical bandgap. In fact, dihedral angles of 49.9, 86.8, 71.4 and 82.4° were determined by density functional theory (DFT) calculations between the donor plane and the acceptor plane for T20-T23, respectively. Due to its lesser twisted structure and based on the design rule previously evoked (orthogonality between the donor and the acceptor is researched to isolate the two groups), T20 showed the higher ΔE ST of the series. Theoretical calculations clearly evidenced for T20 the HOMO level to extend to the neighbouring phenylene bridge, adversely affecting ΔE ST . Conversely, the large dihedral angle of T21-T23 contributed to spatially separate the HOMO from the LUMO orbitals. By electrochemistry, an appreciable reduction of the oxidation potential was detected (+0.87 V) for T21 which is substituted at the 1,8-positions of the donor whereas T20, T22 and T23 exhibited the same oxidation potentials (+0.97 V). By PL, T 1 states of T20, T22 and T23 proved to be 3 LE states whereas a 3 CT state was found for T21.
To evidence this, examination of the phosphorescence spectra of T20-T23 in a frozen toluene matrix at 77 K revealed for T20, T22 and T23 to exhibit well-resolved vibrational structures, demonstrating their T 1 states to be 3 LE states. Conversely, only a broad spectrum was obtained for T21, and its triplet state was assigned to a 3 CT state. Precisely, by its large dihedral angle, T21 differs from T20, T22 and T23 by the order of its orbitals, 3   opposite, prompt and delayed fluorescence components were clearly evidenced for T21-T23. Lifetimes of the delayed components for T21-T23 were 3.5, 13.0 and 10.3 μs, respectively. Due to the inability of T20 to upconvert its electrons from the triplet to the singlet state, T20-based device could only reach an EQE of 7.2%. On the opposite, maximum EQEs of 22.0, 19.2 and 18.3% were obtained for T21-T23-based devices, with CIE coordinates of (0.148, 0.098) and (0.150, 0.097) for T22-and T23-based devices, respectively. As anticipated, a lower color purity was obtained for T21-based devices resulting from its lower oxidation potential. Recently, a significant enhancement of blue OLED performances was obtained by replacing the triazine acceptor by a 2,4,6-triphenylpyrimidine unit in donoracceptor-based TADF emitters [57]. Considering that the electron acceptor is not symmetrical anymore, positions of the nitrogen atoms will significantly influence the distribution of the electronic cloud and potentially the overlap with the HOMO level. Examination of the electronic properties of T24 revealed the HOMO and the LUMO levels are located on both the donor and acceptor part, respectively, without any contribution of the phenyl linker. Another situation was found for T25 and T26 since the LUMO predominantly extends on both the acceptor and the phenyl ring which is between the donor and the 4,6diphenylpyrimidine fragment. Due to the smaller overlap of the two wavefunctions, a weaker intramolecular charge transfer was attended, enabling to provide an emission in the blue or skyblue region. Optical properties were evaluated in solution confirming this trend, with an emission at 455, 476, and 496 nm for T24-T26, respectively. Major differences could be found in the contribution of the delayed component in the luminescence decay. Following the trend determined for the intramolecular charge transfer, a regular increase of the prompt component in the overall decay of the three emitters was found, evidencing the up-conversion of the triplet excitons to the singlet ones. The best EQE was obtained for T26-based devices (22.8%) consistent with the higher delocalization of its electron-donating part, its smaller ΔE ST and the higher contribution of the delayed component in the overall luminescence decay. A regular decrease of the EQE was observed for T25-based devices (18.6%) and T24-based devices (11.8%), confirming the absence of delayed fluorescence for the last emitter and the reduction of the strength of ICT interactions. Interestingly, the EQE reported for T26-based devices is among the best so far reported for blue OLEDs. Attesting the interest of the community for this new acceptor, other authors developed quasi-simultaneously a struc-ture-performance relationship with T24, T25 and T27-T28 (see Figure 8) [58]. The choice of pyrimidine as the electron acceptor was notably justified by authors due to the easier synthesis of the central core and a versatile peripheral substitution. Additionally, compared to triazine, the LUMO level of pyrimidine is slightly destabilized, facilitating the access to wide bandgap materials. In this work, a more intriguing behaviour was found even for T24 and T25 that have just been discussed above since mechanochromic properties were evidenced for the four emitters. Based on photophysical investigations, the presence of two different packing modes in the solid state were proven. When tested in OLEDs, no clear conclusions could be deduced as results of opposite trends were detected. Thus, if the EQE of T24-based OLEDs was lower than that determined for T27-based OLEDs (7.2% and 11.8%, respectively), the opposite trend was found with T25 and T28 (12.6% and 11.8%, respectively). Only the influence of the symmetrical or the unsymmetrical substitution of the pyrimidine acceptor by the donor was evidenced, following the conclusions of previous authors. Finally, two D-A-D triads comprising the 9,9-diphenyl-9,10dihydroacridine donor were reported in 2016 [59]. Here again, existence of relatively large dihedral angles of 82-87° between the donor unit and the nearby phenylene linker for T29 and T30 was confirmed by quantum chemical calculations. Resulting from the almost perfect orthogonality, a good confinement of the electronic density of the two orbitals was obtained with a HOMO level predominantly located on the donor and a distribution of the LUMO over the central pyrimidine acceptor core and the adjacent phenylene linkers small ΔE ST were determined (0.16 and 0.15 eV for T29 and T30, respectively), indicative of reduced electronic correlations between frontier orbitals and accounting for their high performance. Indeed, EQEs of 19.0 and 20.8%, an EL at 468 and 472 nm were, respectively, determined for T29 and T30. However, the efficiency roll-off was quite severe and this drawback was assigned to the relatively long exciton lifetimes of T29 and T30 in doped films (330 and 210 μs, respectively). Recently, an original strategy to combine the electron-donating 9,9-dimethyl-10-phenylacridan with the electron-accepting 2,4,6-triphenyl-1,3,5-triazine was reported under the form of random copolymers derived from a polystyrene (T31-T34, see Figure 9) [60]. Contrarily to the classical TADF materials in which the electron donor is connected to the acceptor, interactions between the two moieties occur by mean of a through-space charge transfer (TSCT). Polystyrenes of different compositions T31-T34 were examined, varying by the acceptor content (5 or 50 wt % of acceptor) and the donor units, i.e., 9,9-dimethyl-10-phenylacridan or 9,9bis(3,5-di-tert-butylphenyl)-10-phenylacridan. Precisely, effect of the steric hindrance on TADF properties of the polymers was investigated by introducing a steric hindrance on the electron donor. Use of polystyrene to generate EL materials is counterintuitive due to its inherent insulating character, but EL polymers substituted with iridium complexes have previously been studied in the literature, evidencing the pertinence of the strategy [61].
In this case, charge transport properties are provided by the substituents attached to the polymer chain. As main finding of this work, the detrimental effect of the steric hindrance was demonstrated, no TSCT effects and no TADF features were detected for T33 and T34. Conversely, for the less hindered polymers, a delayed fluorescence could be evidenced for the two polymers T31 and T32, with a ratio for the prompt/delayed component of 13/87, respectively. ΔE ST values of 0.019 (T31) and 0.021 eV (T32) were also determined by examining the fluorescence and phosphorescence spectra. Interestingly, the bluest EL emission (472 nm) was obtained for the polymer only containing 5 wt % of acceptor T31, with an EQE peaking at 12.1% for these solution-processed OLEDs, what is remarkable. Conversely, a less blue emission was obtained for T32, the emission peaking in the blue-green region (497 nm).

Phenoxaphosphine oxide and phenoxathiin dioxide derivatives
Recently, phenoxaphosphine oxide and phenoxathiin dioxide have gained interest as electron acceptors since the first report mentioning their use as acceptors was published by Lee et al. in 2016 [62]. Prior to this work, phenoxaphosphine oxide derivatives were mostly studied for the design of flame-retardants [63] or as chiral molecules for fullerene recognition [64][65][66]. Similarly, the scope of applications of phenoxathiin dioxide ranged from antimicrobial activity [67] to the use as inhibitor for Hepatitis C virus infection [68]. Here, in the context of OLEDs, Lee et al. reported two blue TADF emitters, P1 and P2 (see Figure 10), containing a phenoxaphosphine oxide or a phenoxathiin dioxide acceptor covalently linked to a dimethylacridan donor.
Theoretical calculations predicted the two molecules to adopt in their optimized molecular geometries a highly twisted conformation, what is a requirement for a spatial separation of the HOMO and LUMO energy levels. As attended, the LUMOs of P1 and P2 are localized on the acceptor moieties whereas their HOMOs are mostly distributed on the donor. Separation of the frontier orbitals lead to ΔE ST values of 0.02 (P1) and 0.10 eV (P2), which are in perfect accordance with the experimental data: ΔE ST = 0.03 and 0.06 eV for P1 and P2, respectively. Interestingly, theoretical calculations also showed the higher electron-accepting ability of the phenoxathiin dioxide moiety compared to that of the phenoxaphosphine oxide group owing to the stronger electron-withdrawing properties of the sulfone group, with a theoretical LUMO level at −1.52 and −1.24 eV for P2 and P1, respectively. In multilayered devices, remarkable CIE coordinates could be realized with P1-and P2-based OLEDs ((0.15, 0.14) with P1 and (0.16, 0.26) with P2), combined with high EQEs (12.3% and 20.5%, respectively). Additionally, for P2-based devices, the efficiency roll-off could be remarkably suppressed and an EQE as high as 13% could be maintained at the luminance of 1000 cd·m −2 .

CN-Substituted pyridine and pyrimidine derivatives
In 2015, Liu et al. constructed a novel blue TADF emitter CN-P1 comprising a carbazole donating moiety connected to a pyridine-3,5-dicarbonitrile accepting group (see Figure 11) [69]. The choice of pyridine-3,5-dicarbonitrile as acceptor was notably motivated by the outstanding charge-transport ability and the remarkable electrochemical stability of this group [70,71]. Thus, CN-P1 had a small singlet−triplet splitting (ΔE ST = 0.04 eV), fairish PLQY in doped films (49.7%), and a delayed decay lifetime of 46.6 μs, which suggests that it could be a promising candidate as emitter. EL performance of CN-P1 was investigated in OLEDs with different CN-P1 doping concentrations in mCP as the emitting layers. The highest EQE (21.2%) of devices was obtained at 13 wt % doping conditions. It was found that the maximum EQEs are enlarged along with the increase of doping concentration, which can be mainly attributed to the more efficient exciton utilization with a higher emitter concentration. However, EQEs decreased with the further concentration increase of CN-P1 due to the strong interaction and aggregation between CN-P1 molecules at high doping concentration in the emitting layer. Authors obtained EL spectra red-shifting from sky-blue (λ max = 475 nm, CIE = (0.18, 0.26)) to greenish-blue (λ max = 510 nm, CIE = (0.24, 0.40)) emissions by varying the doping concentration from 5 to 50 wt %. Such red shift is clearly caused by the interaction between CN-P1 molecules at high dopant concentrations. Parallel to this, CN-P1 molecules can also increase the polarity of the EML, thus introducing a solvatochromaticity-like shift comparable to that observed in solutions while varying the solvents polarity. The optimized device exhibited a maximum current efficiency of 47.7 cd·A −1 , and a maximum power efficiency of 42.8 lm·W −1 without any light outcoupling structures, indicating that nearly 100% of excitons are harvested for light emission. Such high performance should not only be attributed to the fairish PLQY and the efficient RISC process from T 1 to S 1 of CN-P1 emitter, but also owed to the reasonable high T 1 , good charge mobility, and well-matched PL spectrum of the mCP host with the CN-P1 absorption spectrum. Still based on pyridine derivatives, Pan et al. prepared a series of twisted D-π-A type emitters based on the dimethylacridan and different CN-substituted acceptors (pyridine, pyrimidine, and benzene, see Figure 11) [72]. Theoretical calculations showed the different emitters to adopt a nearly orthogonal conformation between the donor and the central aromatic ring, interrupting the π-conjugation and localizing the HOMO level on the acridan moiety and the LUMO level on the central accepting group. The calculations also predicted a more planar phenyl-pyrimidine/ phenyl-pyridine conformation (i.e., a smaller dihedral angle) in CN-P5/CN-P4 and a more twisted phenyl-pyrimidine/ phenyl-pyridine conformation (i.e., a larger dihedral angle) in CN-P3/CN-P2. All the DFT-optimized data were in perfect accordance with single crystal X-ray diffraction analyses. The results showed that the molecular conformations (twist angles in D-spacer-A diads) could be easily tuned by controlling the orientation of the nitrogen atom(s) in the heteroaromatic rings relative to the donor plane. In fact, two main groups of molecules were identified. Thus, CN-P3, CN-P5 and CN-P6 are characterized by a relatively small ∆E ST of 0.032-0.090 eV, show the most pronounced contribution of the delayed component in PL with emission quantum yields for the delayed component of luminescence in the 38-44% range.
These molecules also exhibit high reverse intersystem crossing rates (k RISC > 15 × 10 4 s −1 ). Conversely, CN-P2 and CN-P4 show larger ∆E ST (0.180 − 0.190 eV) than CN-P3, CN-P5 and CN-P6 and lower TADF contributions in PL with smaller quantum yields for the delayed component of luminescence (19-23%). Smaller RISC were also determined (k RISC of < 8 × 10 4 s −1 ). Finally, TADF contribution on the total luminescence of CN-P7 and CN-P8 was the weakest of the series (≤1%) as a result of their extremely large ∆E ST (>400 meV). Due to the weak contribution of the TADF process, these emitters could be nearly assimilated to conventional fluorescent emitters. All light-emitting materials show lifetimes for the prompt decay component of luminescence in the 6.5-27 ns range whereas the lifetimes for the delayed decay component varied from 1.9 to 19 μs. All compounds were tested in OLED and all devices exhibited a relatively low turn-on voltage (≈2.5 V) and a low operation voltage (≈3.5-4 V for a brightness of 100 cd·m −2 ). Devices using high-PLQY emitters (PLQY = 90-100%) exhibited rather high EQEs of up to 23.1-31.3%, while CN-P7 and CN-P8 having the lower PLQYs gave inferior EQEs of 5.7% and 1.6%, respectively. Noticeably, emitters showing the most pronounced TADF characteristics (i.e., CN-P6, CN-P3, and CN-P5) furnished the remarkable EL efficiencies of 29.2% (96.3 cd·A −1 , 105.5 lm·W −1 ), 31.3% (104.5 cd·A −1 , 117.2 lm·W −1 ), and 30.6% (103.7 cd·A −1 , 116.3 lm·W −1 ), respectively. On the opposite, CN-P2 and CN-P4 showing the less pronounced TADF characteristics exhibited similarly high PLQYs (90-92%) but lower EQEs (23-24%). Finally, CN-P8, in which the TADF contribution is almost inexistent, furnished the low EQE of 5.7% (this is also the material exhibiting the lowest PLQY (36%)), yet such an EQE is still significantly higher than it can be expected from a conventional non-TADF fluorescent emitter of similar PLQY (i.e., EQE can be estimated to be ≈2.5-3% at most), suggesting therefore the contribution from the delayed fluorescence in the overall EL process. Although CN-P6, CN-P5, and CN-P3 could reach high maximum EQEs, different efficiency roll-off behaviours could be evidenced with the following order: CN-P6 < CN-P5 < CN-P3. Such a trend for the efficiency roll-off correlate well with the order of their delayed fluorescence lifetimes and their RISC decay rate values in the host film: CN-P6 < CN-P5 < CN-P3 for the delayed fluorescence lifetimes and CN-P6 > CN-P5 > CN-P3 for k RISC . Such correlation is also observed for CN-P4 and CN-P2 devices. It has been rationalized that a small delayed fluorescence lifetime (and thus effective RISC) is beneficial for faster triplet-to-singlet conversion, for reducing the triplet exciton population at higher brightness/ current, and thus for reducing associated quenching mechanisms (e.g., triplet-triplet annihilation, etc.). This year, Sasabe et al. reported high efficiency blue OLEDs using isonicotinonitrile-based fluorescent emitters comprising 9,10-dihydro-9,9dimethylacridine(s) as donor unit(s) [73]. The chemical structures of the two emitters CN-P9 and CN-P10 is given in Figure 12. While evaluating the optical and photophysical properties of the different materials, all compounds showed reasonably high PLQYs (71-79%) in the host films, with a sky-blue emission located at 489 and 495 nm for CN-P9 and CN-P10, respectively. Delayed luminescence lifetimes of 453.7 µs and 116.9 µs, sufficiently small ∆E ST of 0.30 eV and 0.28 eV to allow a RISC were also determined for CN-P9 and CN-P10, respectively. Performances of the two sky-blue emitters CN-P9 and CN-P10 were then evaluated in OLEDs. CN-P9-based devices showed a sky-blue emission with CIE chromaticity coordinates of (0.19, 0.36), a low turn-on voltage of 3.1 V and an EQE of 15%. In contrast, CN-P10-based devices showed still a sky-blue emission with CIE coordinates of (0.22, 0.45), a low turn-on voltage of 2.9 V but an EQE peaking at 22%, resulting from its smaller ∆E ST . Considering the EQE values overcoming the 5% EQE limit for fluorescent materials, contribution of a TADF process in the overall emission of these two emitters was clearly demonstrated.

Phosphine oxide derivatives
Blue thermally activated delayed fluorescence (TADF) dyes are basically combinations of strong acceptors and weak donors. In their recent work, Duan et al. employed a weak acceptor group to construct a series of weak acceptor−strong donor (WASD)type emitters with a phenoxazine donor [74]. The molecular structures of these fluorescent compounds, namely 4-(10Hphenoxazin-10-yl)phenyl)diphenylphosphine oxide (PO-1), bis(4-(10H-phenoxazin-10-yl)phenyl)phenylphosphine oxide (PO-2), and tris(4-(10H-phenoxazin-10-yl)phosphine oxide (PO-3) are given in Figure 13. Similar absorption spectra were measured in dilute solutions for all compounds, with three characteristic bands detected around 370, 320, and 240 nm. The first one was assigned to a n→π* transition from the phenoxazine group to the triphenylphosphine oxide group whereas the second and the third peak was attributed to π→π* transitions of the phenoxazine and the phenyl moities, respectively. A relation of proportionality was demonstrated in the intensities of the band, directly related to the number of phenoxazine groups per molecule. Almost identical PL spectra were determined for these molecules, proving the insulating character of the phosphine oxide group and the pertinence of the WASD strategy to preserve the emission color. Consistent with TD-DFT results, ΔE ST decreased from 0.26 to 0.19 and finally 0.11 eV for PO-1, PO-2 and PO-3, respectively. Relatively high PLQYs were also determined (45%, 57%, and 65%, for PO-1, PO-2 and PO-3, respectively). PLQY of PO-3-based films were determined as 67%, higher than the values determined for PO-2-and PO-1doped films. The prompt fluorescence lifetimes of PO-1, PO-2, and PO-3 are gradually increasing from 8 to 13 to 20 ns. In contrast, the respective order of the delayed fluorescent lifetimes is reversed, at 95, 31, and 17 μs, accompanied by a gradual increase of the quantum yields of 36%, 45%, and 51%, respectively. PO-1-based OLED achieved EL emissions with peaks at 448 nm and CIE coordinates of (0.16, 0.12), corresponding to a deep-blue light. PO-2-based devices displayed a blue emission peaking at 460 nm and CIE coordinates of (0.16, 0.20). OLEDs fabricated with PO-3 produced a pure-blue EL emission peaking at 464 nm, an EQE up to 15.3%, a low efficiency roll-off and CIE coordinates of (0. 17, 0.20). With aim at simplifying the device fabrication, other authors tried to develop emitters PO-4-PO-9 specifically designed for the fabrication of non-doped OLEDs (see Figure 14) [75]. To reach this goal, the electron-transport diphenylphosphine oxide group was attached to pyrene moieties, providing molecules with good filmforming abilities. High performance of OLEDs was assigned to the judicious combination of an enhanced charge transport ability due to the presence of the diphenylphosphine oxide group, the formation of pyrene excimers in the solid state and the assistance of the TADF property. More precisely, a contribution of a TADF process to the overall EL emission of OLEDs is suggested by the presence within the emissive layer of both pyrene and pyrene excimers, resulting in the presence of closelying singlet and triplet states for the two forms. Besides, if a blue emission of the pyrene excimer assisted by TADF is suggested by the authors, no clear evidence of TADF is provided.
To support the presence of a TADF effect in the devices, the authors tentatively assigned the existence of the delayed component of fluorescence by the presence of close-lying singlet and triplet states in both pyrene derivatives and excimers, favorable to a reverse intersystem crossing giving rise to a delayed fluorescence. Multilayered OLEDs fabricated with PO-4-PO-9 showed interesting efficiencies, with EQEs ranging from 7.2 to 9.1%. The contribution of the diphenylphosphine oxide group to the electron mobilities of these emitters was clearly evidenced by fabricating OLEDs using PO-4-PO-9 as electron-carriers. By comparing with a reference electron-transport material, i.e., Alq 3 , a two-fold enhancement of EQEs could be determined while using these materials as electron-transport layers, evidencing their higher electron mobilities compared to that of tris(8-hydroxyquinoline)aluminum Alq 3 . Best OLEDs were obtained with PO-8, EQE peaking at 9.1%.

Benzonitrile derivatives
In the search for new acceptors, benzonitrile was identified as a promising candidate capable to contribute to the design of deep blue TADF emitters. Precisely, the cyano moiety is a group limiting the size of electron acceptor moiety by its compacity while remaining one of the strongest electron-accepting groups at disposal for chemists. By combining benzonitrile with two or three carbazole units, and due to the planarity of the two structures (carbazole, benzonitrile), a sufficient steric hindrance could be induced to provide the highly twisted structures BN-1-BN-4 (see Figure 15) [76]. The four carbazolyl benzonitrile derivatives BN-1-BN-4 were easily prepared in a onestep approach through aromatic nucleophilic substitution. Encouraging results were obtained with the four emitters while using high-triplet-energy hosts with favorable carrier injection/ transporting abilities. The best performance was obtained with BN-2, endowing blueemitting devices with a maximum EQE of 21.5%, which is among the highest values reported for blue TADF devices with an emission peak located at 470 nm. Another possibility could be to increase the number of carbazole units around the benzonitrile moiety. A benzonitrile derivative substituted by five carbazoles (BN-5) was synthesized and characterized by the Adachi team [77]. The OLEDs displayed a light-blue emission and a maximum EQE of 14.8%. Still based on this approach, the group of Hyuk Kwon went even further by introducing a nitrogen atom in the donor, furnishing the carbazole-derived αand δ-carboline where the nitrogen heteroatom is introduced at the α-and δ-position respective to the central nitrogen atom (BN-6 and BN-7, respectively, see Figure 16) [78]. Incorporation of carbolines in these two structures is justified by the fact that this group has recently been identified as an electron-transport material exhibiting a high triplet energy [79][80][81][82]. Even if the introduction of heteroatoms in aromatic compounds can increase the molecular relaxation, the bandgap and the triplet energies will simultaneously increase, consequently dimin-ishing ΔE ST . Effectiveness of the strategy was clearly evidenced by the blue emission produced by OLEDs containing BN-2 as the emitter (CIE coordinates of (0.19, 034), EL at 486 nm) and the high EQE of 22.5% attested of the TADF characteristics of the emitter. In contrast, BN-1-based devices demonstrated a low EQE of 4.2% resulting from its low PLQY (37% contrarily to 93% for BN-2) and the poor contribution of the delayed component to the overall emission (7% contrarily to 45% for BN-2). As a positive point, the EL spectrum of BN-1based devices was blue shifted at 473 nm. Therefore, undeniably, it can be concluded that the effect of the heteroatom position in the carboline donor moiety is essential. Notably, for the two materials, the HOMO and LUMO energy levels of BN-1 and BN-2 are isolated from each other, but a partial overlap exists in BN-1 due to the weaker donating ability of the α-carboline moiety. Jointly, theoretical calculations evidenced a larger bond length change between the ground and excited states for BN-1 (0.048 Å vs 0.041 Å for BN-2 between the carboline and the phenyl group). As a result of this, the higher molecular relaxation in BN-1 is expected to favour the non-  radiative processes, adversely affecting the EL performance. Another study revealed the importance of the donor moiety position compared to benzonitrile for high EL efficiency. In an effort to maximize the TADF process, Adachi developed a series of four highly twisted molecules BN-8-BN11 consisting of the combination of 9,9-diphenylacridane donor unit(s) connected to a benzonitrile central core (see Figure 16) [83]. As first conclusions extracted from the theoretical calculations, the predicted ΔE ST values were similar for all molecules (0.03 eV), suggesting that the substitution position has no effect on the up-conversion properties. Parallel to this, examination of the PL spectra of BN-8-BN-11 showed the PL maximum to be located at 454 and 441 nm for BN-8 and BN-9, respectively, whereas the emission was detected at 433 and 428 nm for the metasubstituted BN-10 and para-substituted BN-11, respectively.
It was thus concluded that the π-conjugation was maximized upon ortho-substitution and the introduction of two donor units on BN-8 optimized the delayed emission intensity so that BN-8 was the only one to be tested in devices. OLEDs fabricated using BN-8 as an emitter showed a blue emission at 463 nm (with CIE coordinates of (0.16, 0.16)) that coincides the PL emission maximum together with the high EQE of 15.9%. However, examination of the chemical stability of an encapsulated film of BN-8 evidenced the emission intensity of the film to decrease in less than 5 min upon photoexcitation. Theoretical calculations pointed out the ortho-substitution to enhance the TADF efficiency because of the optimized steric hindrance but also to decrease the bond dissociation energy as a value of only 0.94 eV for the C-N bond was determined, much lower than the singlet and triplet energies of the molecules (2.75 eV and 2.73 eV, respectively).

Benzoylpyridine and di(pyridinyl)methanone-carbazole derivatives
Emitters displaying efficient RISC and high PLQY are promising candidates for OLEDs and molecules comprising phenyl(pyridin-4-yl)methanone as the acceptor moiety are one of those. As first approach, the two carbazole donors were intro-duced at the orthoand meta-positions of the phenyl ring of the acceptor (see Figure 17, BP-1 and BP-2) [84]. Very small ΔE ST of 0.03 and 0.04 eV and very high PL efficiencies of 88.0 and 91.4% were, respectively, determined for BP-1 and BP-2 in codoped films. These values are higher than that determined in solution for the two molecules (4.4 to 14.2% depending of the solvent for BP-1, 2.8 to 34.0% depending of the solvent for BP-2), demonstrating the suppression of the collisional and the intramolecular rotational quenching in thin films. However, the substitution pattern of carbazole drastically modified the emission wavelengths and a red-shift of approximately 20 nm was observed upon introduction of tert-butyl substituents on BP-2. Conversely, a higher electrochemical stability was determined for BP-2 upon repeating CV scans, the two reactive C 3 and C 6 sites in para-position relative to the nitrogen atom of the carbazole being blocked by the tert-butyl groups. In multilayered devices, the bluer emitter BP-1 provided efficiencies comparable to those obtained with iridium-based phosphorescent OLEDs at similar EL wavelength [85,86]. Notably, skyblue BP-1-based OLEDs reached a maximum efficiency of 24% for the light peaking at 488 nm. The same year (2016), the same authors changed their strategy and combined all electron donors together, replacing the former D-A-D triads by D-A diads [87]. To tune the electron donating ability, carbazoles were introduced at the outer position of a carbazole unit, at the 3 and 3,6-conjugated positions of the first carbazole, resulting in donors composed in total of one to three carbazole groups. Comparison established with this series of emitters evidenced a clear decrease of ΔE ST upon expending the size of the donating part and the number of carbazole units per donor. Thus, ΔE ST decreased from 0.29 eV for BP-3 to 0.07 eV for BP-4 and 0.05 eV for BP-5, consistent with a higher spatial HOMO and LUMO separation and a more extended molecular HOMO orbital distribution.
Unfortunately, despites these favorable features, a significant red-shift of the emission was evidenced for BP-4 and BP-5 as a result of a dual emission, one corresponding to a carbazolecentered π-π* transition at high energy and an additional but unexpected intramolecular charge transfer only observed for BP-4 and BP-5 at lower energy. A clear shift of the emission maximum was notably evidenced in toluene, the maximum emission wavelength shifting from 440 nm for BP-3 to 480 nm for BP-4 and 482 nm for BP-5. Therefore, only blue devices could be fabricated with the mono-substituted emitter BP-3 and a comparison was established with BP-6 differing from BP-3 by the substitution pattern of the unique carbazole. Once again, a red-shift of the emission was observed upon incorporation of tert-butyl groups on carbazole, the emission in toluene being detected at 467 nm. Evaluation of the potential of BP-3 and BP-6 as new developed emitters for OLEDs confirmed the trend observed by PL and BP-3 furnished a more blue OLED than BP-6, with an external efficiency peaking at 9.4%. By optimizing the device structure [88], the same authors could drastically increase the EQE of BP-3-based devices up to 18.4%, even if a non-negligible red-shift of the emission wavelength could be observed: 474 nm, (0.16, 0.25) for this study [88] contrarily to the previous emission detected at 452 nm, (0.13, 0.16) [87]. Inspired by the structure of BP-2, the same authors developed a series of three fluorescent molecules by varying the position of the nitrogen atom of the pyridine moieties BP-7-BP-9 [89]. All molecules are characterized by high PLQYs in thin films, ranging from 92 to 97%, and small ΔE ST varying from 0.01 eV for BP-7 to 0.05 eV for BP-8 and 0.02 for BP-9. Despites these appealing photophysical characteristics, positions of EL peaks appeared at 490, 476 and 490 nm for BP-7-BP-9-based devices, respectively, therefore in the bluegreen region. While comparing with the standard triplet emitter Firpic, a clear enhancement of the EL performance was observed, EQE of Firpic-based OLEDs peaking at 18.7% whereas EQEs of 2.1, 24.6 and 28.0% could be, respectively, realized with the three TADF emitters BP-7-BP-9 (see Figure 18). Here again, the ability of TADF emitters to outperform the standard phosphorescent emitters was demonstrated. Finally, the key to produce a pure blue emission with pyridine-based emitters seems to have been found with the di(pyridinyl)methanone electron-accepting core that could furnish a superior pure blue emission compared to emitters based on the benzoylpyridine core [90]. By introducing two pyridines in bis(6-(3,6-di-tert-butyl-9H-carbazol-9-yl)pyridin-3-yl)methanone (BP-10), a nearly planar molecule could be obtained, favouring the horizontal molecular orientation of the molecule within the co-doped emissive layer. By this specific arrangement in the EML, a perfect stacking of the molecules parallel to the substrate was determined, providing an isotropic orientation of the transition dipole moment. Finally, OLEDs fabricated with BP-10 with a classical device structure furnished a record-breaking EQE of almost 32% with a relatively low dopant concentration (7 wt %) and an emission located at 464 nm.

Triazole derivatives
3,4,5-Triphenyl-4H-1,2,4-triazole is a good electron acceptor but also a remarkable electron-transport material used for the design of numerous OLED materials ranging from charge-transport materials to light-emitting materials [91][92][93]. Logically, combination of 3,4,5-triphenyl-4H-1,2,4-triazole with the electron-donor phenoxazine could provide emitters with TADF properties if conveniently associated and such assemblies were reported for the first time in 2013 (see Figure 19) [94]. Comparison of the diad Trz-1 and the triad Trz-2 evidenced in the absence of oxygen the triad Trz-2 to be more luminescent than the diad Trz-1 (29.8 and 43.1% for Trz-1 and Trz-2, respectively). This trend was confirmed with the design of another series of diad/triads comprising an oxadiazole as the central electron acceptor. This characteristic is opposite to the trend classically reported in the literature where the molecules with a large oscillator strength show a high PLQY [95]. In the present case, the opposite situation was found as the more luminescent materials Trz-2 showed the smaller oscillator strength, evidencing that the order of the PLQYs was not only controlled by the oscillator strength, but also by a competition with vibronic couplings responsible from nonradiative deactivation pathways. The fabrication of OLEDs with the most luminescent Trz-2 furnished sky-blue OLEDs reflecting its PL spectrum in thin doped films (λ EL = 456 nm, EQE = 6.4%).

Triphenylamine derivatives
Triphenylamine is a remarkable electron-donating group that found applications in numerous research fields ranging from OLEDs to organic photovoltaics [96]. In the context of TADF blue emitters, an original strategy to tune the emission wavelength consisted in solely changing the sulfur atom valence state of the thioxanthone core, enabling the emission color to shift from blue to yellow [97]. Even if several connecting modes for the triphenylamine moieties onto the thioxanthone core was envisioned, a blue PL was only detected for TPA-1 by introducing the two triphenylamine groups at the para-positions of the carbonyl group in 9H-thioxanthen-9-one (see Figure 20). Because of this specific substitution, a minimal HOMO/LUMO overlap was evidenced by theoretical calculations. Despites the symmetrical substitution of TPA-1 and the reduction of the oscillator strength in the triad, the PLQY remained high, reaching 35% regardless doped or neat films under air conditions. In a standard device stacking, highly efficient emission could be realized as a maximum EQE value of 23.7% was obtained for OLEDs comprising an emissive layer with a doping concentration of 1 wt % and CIE coordinates of (0.139, 0.280).
In 2017, more blue OLEDs were obtained by using malononitrile as the electron acceptor [98]. The molecular orientation of the emitting material is essential to optimize the EL characteristics and an increase of the external efficiency by up to 46% can be achieved if the molecules are perfectly aligned horizontally by giving rise to light-outcoupling effects [99][100][101]. In this work, TPA-2 and TPA-3 share a similar ΔE ST and similar PL characteristics but major differences were found upon fabrication of OLEDs with these two materials. Notably, the current efficiency of OLEDs elaborated with TPA-3 as dopant was approximately 9 times higher than that determined for TPA-2based OLEDs (12.6 and 1.4 cd/A, respectively). To explain these differences, the perfect horizontal orientation of TPA-3 in doped films contrarily to the weak crystallinity and random orientation of TPA-2 resulted in an improvement of the light extraction for TPA-3-based devices, justifying the enhanced performance.

Conclusion
To conclude, a wide range of strategies are currently developed to produce a blue TADF emission. Among the different findings that can constitute a guideline for the molecular design for blue TADF emitters, it can be cited: 1) The interruption of the π-conjugation by introducing an orthogonality between the donor and the acceptor to minimize the coupling between the two parts, 2) the fact to maintain the donor close to the acceptor to prevent a complete isolation of the donor and the acceptor, 3) the extension of the π-conjugated system of the donor and/or acceptor to maximize the oscillator strength and thus to increase the PLQY, 4) a minimization of ∆E ST to optimize the rate constant of the reverse intersystem crossing, 5) the elaboration of light emitting materials with lifetimes of the delayed component of luminescence as short as possible to address the excited states annihilation issue, 6) a careful selection of the connectivity introduced between the electron donor/acceptor moieties as exemplified by the difference of the EL performance for materials differing by the substitution (ortho-, metaand para-position of aromatic rings). The different results and observations reported in this review have clearly evidenced that a great deal of efforts has still to be done to produce a deep blue EL, as evidenced in Figure 21. At present, the bluest emitters reported in the literature, i.e., emitters with CIE x-coordinate below 0.16 and CIE y-coordinate below 0.10 only four are known: D3 (0.15, 0.07) [29], reported in 2012, T22 and T23 (0.15, 0.10) [45], reported in 2017, and finally CN-P8 (0.16, 0.06) [59], reported in 2016. D3, T22 and T23 are all based on carbazole, but carbazole is certainly not the best candidate for the design of highly stable deep blue emitter because of the photo-assisted electrochemical degradation processes it can initiate. Since 2016, a great deal of efforts has been done to investigate new structures issued from communities other than Organic Electronics and electron donors such as phenoxaphosphine oxide or phenoxathiin dioxide and electron acceptors such as αand δ-carbolines that have historically been used for the design of biologically active molecules are now commonly used during the elaboration of light emitting materials. Blue and stable emitters that will be developed in the future will certainly comprise such unprecedented moieties. Recently, another aspect of crucial importance to increase the EL performance concerns the molecular alignment of the emitter molecules in OLEDs as this can have an important effect on the outcoupling efficiency; this point warrants more systematic investigations in the future.
We herein present the synthesis and photophysical properties of two new C^C* cyclometalated platinum complexes. Both are based on the original 3-methyl-1-phenylimidazolium (MPIM) ligand system which together with the acac auxiliary ligand showed only a very low quantum yield of 7%. We introduced sterically demanding aryl substituted β-diketonate auxiliary ligands to further examine their influence on the emission properties of the resulting platinum(II) complexes.

Results
The mesityl-and duryl-substituted 3-methyl-1-phenylimidazole complexes 2, Pt(MPIM)(mes) and 3, Pt(MPIM)(dur), were synthesized from 3-methyl-1-phenylimidazolium iodide (1) according to a modified literature procedure (Scheme 1) [41,42]. The starting imidazolium salt 1 was prepared from phenylimidazole by addition of methyl iodide as previously described [43]. Complexes 2 and 3 were obtained as yellow solids in isolated yields of 5% and 18%, respectively (Scheme 1). They were characterized by standard methods, NMR techniques ( 1 H, 13 C, and 195 Pt) as well as mass spectrometry (ESIMS). The purity of all compounds was verified by elemental analyses. Additionally we could unequivocally determine the structural parameters of 3 by a solid-state structure ( Figure 1). Details of the structure determination are given in Supporting Information File 1, Table S1.
The absorption spectra ( Figure 2) were measured in dichloromethane solution at ambient temperature. The complexes show almost identical absorption behavior with only minor deviations in the absorption intensity. Both complexes exhibit a strong absorption in the ultraviolet spectral region with an intense shoulder at 241 nm. Two weak and one more intense absorption bands are additionally located at 280 nm, 293 m, and 313 nm, respectively.
Photoluminescence spectra ( Figure 3) were measured at ambient temperature in a PMMA matrix (2 wt % complex) and at 77 K in 2-MeTHF (0.5 mM). The room-temperature emission spectra of both complexes exhibit one broad, structurally unresolved band in the sky-blue spectral region.
The low-temperature emission maxima of both complexes display only a minor hypsochromic shift compared to the emission at ambient temperature: 5 nm for complex 2 and 8 nm for     (2) and 73% (3) at ambient temperatures as well as short decay times around 3 μs (Table 1) were measured. The complexes show no aggregation behavior at higher concentrations (10 wt % in PMMA and 100% amorphous film measurements, see Figures S1, S2 and Tables S2, S3 in Supporting Information File 1), which can be assigned to the steric demand of the aryl-substituted diketonate counter ligand.
Cyclic voltammograms of complexes 2 and 3 were measured in DMF with ferrocene as an internal reference. For both compounds, one irreversible oxidation wave was measured (Figure 4), which is commonly found for platinum(II) complexes [16,44]. Irreversibility of the measured signals was confirmed by variation of the scan rate (30 mV/s to 1 V/s). The peak potential of the oxidation is located at 0.69 V vs ferrocene for both complexes. No reduction was observed for both complexes in the electrochemical window of the solvent. Thus, the electrochemical behavior of the newly synthesized substances is comparable.
The higher emission efficiency is accompanied by a red shift in emission color of about 40 nm ( Figure 6). An improved quantum yield of Φ = 30% (5 wt % in PMMA) has already been reported for a 3-methyl-1-phenylimidazolium cyclometallated platinum(II) complex by the introduction of a sterically demanding ancillary ligand (α-duryl substituted acac) in the central position of the acetylacetonate between the two C=O groups [35]. Besides an increased quantum yield, the complex displayed a small red shift (λ exc = 467 nm) compared to Pt(MPIM)(acac) and a decay time of 8.7 μs. When mesityl or duryl groups replace both methyl groups of the acetylacetonate, the quantum yield is further enhanced. Such a severe influence of the mesityl-and duryl-substituted auxiliary ligands on the quantum yield is unprecedented, although enhanced quantum yields have been reported for both ligands [18,[39][40][41]. Additionally, the decay times of Pt(MPIM)(mes) and Pt(MPIM)(dur) are shorter compared to the phosphorescence decay of the α-durylsubstituted complex (8.7 µs at 77 K in 2-MeTHF).
The observed red shift in emission color is also in agreement with the results of the DFT calculations (Supporting Information File 1, Table S4) of the predicted emission wavelength, ac- cording to a previously published procedure [60]. The bathochromic shift in emission color of complexes 2 and 3 can be assigned to the delocalization of electron density on the arylsubstituted auxiliary ligands.

Conclusion
As shown above, we observed an unprecedented enhancement of the quantum yield for platinum(II) complexes with 3-methyl-1-phenylimidazole as C^C* cyclometalating ligand by changing the ancillary ligand from acetylacetonate (R = CH 3 ) to sterically demanding aryl-substituted β-diketones (R = 2,4,6-trimethylphenyl, 2,3,5,6-tetramethylphenyl). The drastically increased quantum yield was accompanied by a shift in the emission color from the deep-blue to the sky-blue spectral region. Besides a very efficient phosphorescent emission, the two newly synthesized complexes also exhibit very short decay times of less than 3 μs. The profound impact of the counter ligand on the complexes' emission properties originates from the diketonate ligand, which was also confirmed by DFT calculations.

Experimental
Both complexes were characterized by 1 H, 13  The product was obtained following the general procedure reported for 2 using 1-methyl-3-phenyl-1H-imidazol-3-ium iodide ( maximal external quantum efficiency (EQE max ) of 5%, when assuming the out-coupling efficiency to be 20%. On the other hand, phosphorescent materials could utilize triplet excitons in electroluminescence processes to achieve 100% IQE max [2,3]. However, the utilization of metals like iridium and platinum, which are expensive and nonrenewable, inevitably increase the cost of the final OLEDs. Alternatively, a thermally activated delayed fluorescence (TADF) material is a kind of noble-metalfree fluorescent material able to transform triplet excitons into singlet excitons through reverse intersystem crossing (RISC) to achieve 100% IQE max in theory [4].
On the basis of the previous considerations, for TADF materials, the energy difference (ΔE ST ) between the first singlet excited state (S 1 ) and the first triplet excited state (T 1 ) must be small enough to enable the RISC process with the activation of environmental thermal energy [5]. To achieve this, electron donors (D) and electron acceptors (A) are introduced into the molecule to form an intramolecular charge transfer (ICT) state with a large twisting angle between the donor and the acceptor to achieve the separation of highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) [6], which is the key to reduce the ΔE ST . Therefore, D-A-type or D-A-D-type molecules are the most classical TADF molecular structures [7].
Although there have been numerous TADF materials synthesized and reported [8,9], to the best of our knowledge, orange and red TADF materials are still rarely reported in comparison with blue and green TADF materials [10,11]. It is difficult to achieve TADF in orange and red fluorescent materials not only because red TADF materials require a strong ICT state, which strongly facilitates nonradiative transition processes, but also because the energy gap law generally results in a low radiative rate constant (k r ) to compete with a large nonradiative rate constant (k nr ) [12]. The increasing nonradiative transition processes and large k nr play a role in competition with RISC and radiative transition processes and seriously restrict the development of orange and red TADF materials [5]. Therefore, further attempts and new designs towards orange and red TADF materials are necessary.
In this work, we designed and synthetized two novel D-A-Dtype orange TADF materials, namely 2,7-bis(9,9-dimethylacridin-10(9H)-yl)-9H-fluoren-9-one (27DACRFT, 1) and 3,6bis(9,9-dimethylacridin-10(9H)-yl)-9H-fluoren-9-one (36DACRFT, 2, Scheme 1). The compounds are isomers with different donor-accepter bonding positions, where the fluorenone unit is a strong electron acceptor, which has not been reported in the field of TADF materials before, while acridine, one of the most commonly used donors in TADF materials, has strong electron-donating and hole-transport ability. The combination of the strong acceptor and strong donor can give a narrow energy gap and thus longer wavelength emission. Compounds 1 and 2 were thoroughly characterized by 1 H NMR, 13 C NMR and electron ionization (EI) mass spectrometry. Both of them show TADF behavior with orange emission color according to the photoluminescence spectra and time-resolved transient photoluminescence decay measurement. EQEs of 2.9% and 8.9% were achieved for the OLED devices based on 1 and 2, respectively, which are higher than the theoretical efficiency of the OLEDs based on conventional fluorescent materials.

Results and Discussion
27DACRFT 1 and 36DACRFT 2 have similar thermal properties according to thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) measurements. They have high decomposition temperatures (T d , corresponding to a 5% weight loss) of 361 and 363 °C, respectively. In addition, no glass-transition temperature (T g ) was found according to their DSC curves. Thanks to their amorphous characteristics, the stability of their morphology and chemical composition can be expected during the evaporation processing fabrication of OLEDs.
In order to characterize their electrochemical properties, cyclic voltammetry (CV) measurements were conducted to measure their oxidation potentials (E ox ) and reduction potentials (E red ). Ionization potential (IP) and electron affinity (EA), which approximate to their HOMO and LUMO energy levels, are calculated from E red and E ox . Compounds 1 and 2 have similar HOMO and LUMO energy levels due to the same donor and acceptor in the molecules (Table 1). The molecular geometry of 1 and 2 in the ground state and excited state were simulated by density functional theory (DFT) and time-dependent density functional theory (TD-DFT) calculations, respectively. The ground state (S 0 ) geometries were optimized on B3LYP/6-31G* level in gas phase, while the lowest triplet excited state (T 1 ) energy levels and the singlet excited state (S 1 ) geometries of those molecules were optimized by TD-DFT on m062x/6-31G* level based on the optimized ground state geometries. The optimized geometries of S 0 and S 1 are shown in Figure 1.
The optimized geometries in S 0 are shown in Figure 1a, and all the data are summarized in Table 2. Large twisting angles (θ) of 89.33° and 88.80° between the donor units and the accepter units were estimated for compound 1 and 2, respectively. As shown in Figure 1b, HOMOs and LUMOs are mainly located on the acridine unit and the fluorenone unit, respectively, which contribute to small ΔE ST . The existence of a very small overlap of HOMOs and LUMOs is advantageous to retain high photoluminescence (PL) quantum yields [13][14][15]. The calculated ΔE ST of 1 and 2 are 0.33 and 0.27 eV, which are small enough to achieve TADF behavior.
As shown in Figure 1c, the twisting angle (θ') of 1 in S 1 is 63.74°, which is much smaller than its θ in S 0 , meanwhile, the conformation of the acridine units in 1 is also changed in S 1 as a result of vibrational relaxation and internal conversion (IC), which means the S 0 geometry of 1 becomes unstable when the molecule is excited and the wave function distribution is changed. The different twisting angles between S 0 and S 1 may reduce its PL property according to the energy gap law [16] as vibrational relaxation and intersystem crossing (IC) processes can consume the energy in S 1 , leading to increased nonradiative deactivation [17], reduced PL quantum yield, and thus reduced singlet exciton utilization. On the contrary, the geometry of 2 is hardly changed when excited. Thus, compound 2 shows more potentiality in the application of OLEDs for its better configuration. Ultraviolet-visible (UV-vis) absorption and PL spectra in dilute solutions of 1 and 2 (10 −5 M) are presented in Figure 2. Both compounds 1 and 2 have similar absorption peaks at around 345 and 456 nm. The peaks at around 456 nm result from their ICT states from the donor to the acceptor, while the absorption below 380 nm is caused by their short π-conjugation. It is obvious that 2 has not only a higher oscillator strength (f) than 1 from its transition of charge-transfer states, but also a weaker oscillator strength from its local excited (LE) states. It could be considered that 2 has a better configuration, which is advantageous to intramolecular charge transfer compared with 1, which coincides with the conclusion from DFT calculation.
The PL spectra of the materials in different solvents were also measured. However, no emission was observed in the dilute solutions of dichloromethane (DCM) and tetrahydrofuran (THF) because vibrational relaxation and internal conversion are promoted to reduce the PL intensity. Both compounds 1 and 2 show almost the same PL spectra in dilute solutions of toluene and n-hexane. The photoluminescence spectra of the n-hexane solutions show a peak at 517 nm with a shoulder at 545 nm, which can be considered as the radiative transition of 1 LE states. Noticeably, the charge-transfer process is limited in n-hexane because of its lower polarity. Only one peak at 593 nm was observed for the dilute toluene solutions of both molecules with the typical PL spectra from the radiative transition of ICT states, which could be the evidence of the existence of strong ICT states of both molecules. More importantly, both materials achieve orange luminescence in a dilute solution of toluene, which could be attributed to the strong electron-withdrawing ability and excess conjugation length of fluorenone plane compared with conventional benzophenone acceptor [18].
In addition, low temperature photoluminescence (LTPL) spectra of the materials in toluene at 77 K were measured. The energy levels of S 1 and T 1 were determined from the onset of the prompt and delayed emission peaks, respectively. As shown in Figure 3, both T 1 states of the materials could be confirmed as 3 CT character from their delayed photoluminescence spectra without any well-defined vibronic structure [7]. The ΔE ST of 1 and 2 are 0.19 and 0.09 eV, respectively, indicating that compound 2 may have a much more efficient RISC process than 2 [19,20] (Table 3).
To gain a further understanding of the photophysical properties of 1 and 2 in solid state, two doped films in 4,4'-dicarbazolyl-1,1'-biphenyl (CBP) were vacuum co-deposited at a concentration of 8 wt % for photoluminescence quantum yield (PLQY) and time-resolved transient photoluminescence decay measurements. The concentration of the doped films was optimized to ensure complete energy transfer between the host and the guest. PLQY measurements of 1:CBP and 2:CBP are 7% and 26%, respectively. The PLQY measurements of the doped films with lower concentration show varying degrees of deviation due to the incomplete energy transfer and the obvious luminescence from CBP (PLQY of 1 and 2 doped in CBP with 1 wt % are 2%  a Ultraviolet-visible absorption spectra and photoluminescence spectra measured in toluene; b photoluminescence (PL) spectra measured in n-hexane; c photoluminescence spectra and PL quantum yields measured in doped film 8 wt % in CBP; d energy gap (E g ) calculated from the empirical formula: E g = 1240/λ abs-onset , where λ abs-onset is the onset of ultraviolet-visible absorption spectra. e ΔE ST is calculated from the onset of photoluminescence spectra at 77 K. and 10%, respectively). As shown in Supporting Information File 1, both PL spectra of the doped films of 1:CBP and 2:CBP show red-shift from their PL spectra in n-hexane, which could be considered as the influence from aggregation. As mentioned above, 1 and 2 show nearly the same PL spectra in their dilute toluene solution. However, the PL spectrum of 2 is slightly blue-shifted from its PL spectrum in toluene, while 1:CBP shows alike spectra with 1 in toluene. It could be considered as the solid-state solvation effect [21], as 2 and 1 have different dipole moment of 1.814 D and 3.501 D, respectively from DFT calculation, owing to their different configurations.
The doped film 2:CBP shows a typical TADF behavior as shown in Figure 4b, according to the time-resolved transient photoluminescence decay measurement. The proportion of delayed fluorescence increases rapidly with improved temperature from 77 to 250 K and slowly by acceleration of the nonradiative transition rate when the temperature is higher than 250 K. On the other hand, 1:CBP hardly shows a TADF behavior when the temperature is below 300 K, as shown in Figure 4c.
The signals are characterized by noise rather than delayed fluorescence when the temperature is lower than room temperature due to its low PLQY. Delayed fluorescence can be only observed when the temperature is above 300 K. This could be attributed to the large ΔE ST and low PLQY of 1 which requires more energy to achieve RISC process from T 1 to S 1 . According to the integration and the lifetime of the prompt and delayed components of the time-resolved transient PL decay curves at room temperature, the PLQY of their respective components and rate constant of different kinetic processes were calculated, as shown in Table 4.
(1)  where k r , k nr , k isc , and k risc represent the rate constant of radiative, nonradiative, intersystem crossing and reverse intersystem crossing, respectively; Φ, Φ PF , Φ TADF , τ PF and τ TADF represent the photoluminescence quantum yield, quantum yield of the prompt component, quantum yield of the delayed component, and lifetimes of the prompt and delayed components, respectively. As shown in Table 4, 2 has a significantly larger k nr than 2, which is consistent with the DFT simulation. On the other hand, a much lower k risc and longer τ TADF was acquired by 1:CBP than 2:CBP, as a result of the blocked reverse intersystem crossing and the large ΔE ST . Further, the existence of strong IC and vibrational relaxation processes of 1 is proved by its large k nr and low PLQY. In contrast, owing to the relatively small ΔE ST , k risc of 2 is higher and τ TADF is relatively shorter than 1. The short τ TADF not only signifies efficient utilization of singlet excitons, but is also advantageous in reducing the triplet exciton concentration and efficiency roll-off in the OLED devices.
TAPC and TmPyPB also play the role of exciton blocking layer at the same time because of their high T 1 energy level. Carriers will also be trapped by the emitter directly because of the energy level difference between CBP and the emitter, which makes it possible for the OLEDs with such a low emitter concentration to achieve complete energy transfer. The performance of the fabricated devices is summarized in Table 5 while the J-V-L (current density-voltage-luminance) and EQE-current density characteristics of the devices are shown in Figure 6.
A significantly higher performance was observed from the device based on 2 with a maximal current efficiency (CE max ) of

21
.84 cd/A, maximal power efficiency (PE max ) of 19.11 lm/W and maximal external quantum efficiency (EQE max ) of 8.92%, which is higher than the theoretical maximal external quantum efficiency of the OLEDs based on conventional fluorescent emitter. Meanwhile, the device based on 1 shows poor performance due to its low PLQY and nonobvious TADF behavior. Moreover, the efficiency roll-off of the device based on 2 was reduced compared with the 1-based device. The EQE of the 2-based device is still over half of its EQE max at a brightness of 100 cd/m 2 , while the EQE of 1 at the same brightness is only about 22% of its EQE max . According to the previous study, triplet-triplet annihilation (TTA) might be the main cause of efficiency roll-off in the TADF-OLEDs when the triplet exciton concentration increases with brightness and current density [23,24]. The efficiency roll-off caused by the TTA process of TADF-OLEDs could be analyzed by the TTA model using Equation 5 [25,26] below: (5) where η 0 represents the EQE without the influence of TTA, and J 0 represents the current density at the half maximum of the EQE; η and J represent the EQE with the influence of TTA and the corresponding current density, respectively. As shown in Figure 7, both devices show good agreement with the TTA model fitted curves at low current density because TTA process is the leading factor to the efficiency roll-off of TADF-OLEDs when the exciton concentration is low. With the increase of exciton concentration, singlet-triplet annihilation (STA), singlet-polaron annihilation (SPA) and triplet-polaron annihilation (TPA) may also have serious impact to the efficiency rolloff, which cause the TTA model fitted curves to deviate from the actual value. The device based on 2 shows a better agreement with the fitted curve in higher current density while the device based on 1 does not. In addition, 2 has a better triplet exciton utilization ability to reduce the efficiency roll-off, which comes to the same conclusion with the analysis of their photophysical properties.

Conclusion
In summary, two novel D-A-D-type orange-emitting TADF materials, namely 2,7-bis(9,9-dimethylacridin-10(9H)-yl)-9Hfluoren-9-one (27DACRFT, 1) and 3,6-bis(9,9-dimethylacridin-10(9H)-yl)-9H-fluoren-9-one (36DACRFT, 2), with the fluorenone unit as acceptor and the acridine as donor, were synthetized. Compounds 1 and 2 are isomers but show greatly different performance in terms of both photoluminescence and electroluminescence. It has been shown that the fluorenone unit is a promising acceptor for orange TADF materials, which aids in the design of the TADF behavior and luminescence color of 1 and 2. Owing to the strong electron-withdrawing ability and extended conjugation length of fluorenone unit, the emission peaks of both materials show obvious red-shifts from other TADF materials based on carbonyl acceptor [27,28]. According to the DFT and TD-DFT simulation and photophysical characterization, 2 shows a smaller singlet-triplet energy difference (ΔE ST ) and a larger radiative rate constant (k r ) to give reduced internal conversion, promoted RISC process, and thus a better triplet exciton utilization ability. Maximum EQE values of 8.9% and 2.9% were achieved for the OLED devices based on 2 and 1, respectively. Efficiency roll-off, which is considered to be the result of TTA, is also reduced more effectively for the OLEDs based on 2. coupling (SOC), and obtain nearly 100% of the internal quantum efficiency (IQE). Thus, most of researchers have focused on phosphorescent organic light-emitting diodes (PhOLEDs) all over the world [7,8]. With the deepening of this research, the performance of red and green electroluminescent devices has been able to meet the commercial requirements, but the blue electroluminescent devices have several weaknesses such as low efficiency and poor stability and so on, which hinder its development. It has been proved that the selection of proper materials for each layer is very important for achieving highly efficient PhOLEDs. In particular, the design of host materials plays a critical role in the determination of the devices performance. Therefore, it is beneficial to develop new blue phosphorescent host materials with high-performance for blue PhOLEDs [9][10][11][12].

Experimental
Generally, ideal host materials are required to fulfill several requirements [13,14]: i) the triplet energy level (E T ) should be higher for efficient energy transfer to the guest; ii) suitable energy levels appropriately aligned with those of the neighboring active layers for efficient charge carrier injection to achieve a low operating voltage; iii) good and balanced charge carrier transport properties for the hole-electron recombination process; iv) good thermal and morphological stability for the vacuum deposition method to prolong the device operational lifetime.
Carbazole groups are widely used in host materials because of their high triplet energy levels and high hole mobility [15]. The Lee group [16] linked carbazolyl groups to diphenyl phosphoramines to design asymmetric (9-phenyl-9H-carbazole-2,5diyl)bis(diphenylphosphine oxide) (PCPOs) with a higher triplet energy level (2.80 eV) and a glass transition temperature (140 °C). The maximum external quantum efficiency (EQE) of PhOLEDs was 31.4%, which was prepared by PCPO as a bipolar host material. Kim et al. [17] reported that the bipolar host material 9-(4-(9H-pyrido [2,3-b]indol-9-yl)phenyl)-9H-3,9'bicarbazole (pBCb2Cz) has a high triplet energy level (2.93 eV), which is the main material of blue PhOLEDs, and the EQE of the device is 23.0%. The Suh group [18] reported that the EQE of the prepared device of the bipolar host material N-(3,5-di(9H-carbazol-9-yl)phenyl)-N-(pyridin-2-yl)pyridin-2amine (DCPPy) based on carbazole group is 21.6%. The Wang group [19] designed a host material with symmetrical structure based on phenothiazine-5,5-dioxide. But the host materials with asymmetric structure based on the phenothiazine-5,5-dioxide were rarely reported. For obtaining a high triplet energy level and good stability, herein, with phenothiazine-5,5-dioxide as acceptor (A) and carbazole as donor (D), and introducing an alkane chain group to the host materials for better film-forming properties, two novel blue phosphorescent host materials, CEPDO and CBPDO, were synthesized. At the same time, the photophysical properties, electrochemical properties and their thermal stability were studied and the expected results were obtained.

Synthesis and theoretical calculations
The synthesis route for CEPDO and CBPDO is shown in Scheme 1. The detailed synthesis procedures and characterizations are given in Supporting Information File 1.
In order to further understand the structural properties of the materials and the possibility of charge transfer from donor to acceptor on electronic excitation, the electronic structure of the materials were analyzed by density functional theoretical (DFT) calculations using the Gaussian 09 program package. The electron density distributions and energy levels of the HOMO and LUMO are displayed in Figure 1.  The HOMOs of CEPDO and CBPDO are mainly distributed over the electron-donating carbazole moiety and slightly extended to the phenyl ring. The LUMOs are mostly localized on the phenothiazine-5,5-dioxide, based on the DFT calculation.
There is a small degree of spatial overlap between the HOMO and LUMO in these two molecules. The separated HOMO and LUMO resulted from the strong electron-donating nature of the carbazole unit and electron-withdrawing ability of the phenothiazine-5,5-dioxide unit, thus realized the orbital separation of hole and electron transport in the same molecule. This indicated CEPDO and CBPDO have bipolar characteristic. Figure 2 presents the UV-vis absorption, photoluminescence and phosphorescence (77 K) spectra of CEPDO (a) and CBPDO (b) in solution, respectively. Obviously, the strong absorption peak at 236 nm can be ascribed to the π→π* transition of carbazole moiety of the molecules, and the weaker absorptions around 295 nm assign to the n→π* transition of the conjugation of the whole molecule.

Photophysical properties
The optical bandgap (E g ) of the two substances were all calculated to be 3.32 eV from the UV-vis absorption spectra of CEPDO and CBPDO. Upon photoexcitation of 330 nm at room temperature, both CEPDO and CBPDO exhibited a FL spectrum with peaks at 408 nm and emitted blue fluorescence. The fluorescence quantum efficiencies (Φ) of CEPDO and CBPDO were 62.5% and 59.7%, respectively, by using quinine sulfate as a reference [20]. Compared with CEPDO, the longer alkyl chain of CBPDO led to a corresponding increase in Φ value. Therefore, CEPDO and CBPDO are promising photoelectric materials. To obtain the triplet energy level, their low-temperature Phos spectra were measured in a 2-methyltetrahydrofuran solution at 77 K which are also shown in Figure 2. The phosphorescence emission peaks were at 440 nm and 439 nm, respectively. According to the onset of their phosphorescence spectra, the calculated triplet energy levels (E T ) of CEPDO and CBPDO were identical at 2.82 eV, which matched with the blue phosphorescent guest material FIrpic (2.65 eV), dark blue phosphorescent guest material FCNIrpic (2.74 eV) and FIr6 (2.73 eV). The high E T was attributed to the insulated carbazole moieties. Hence, both CEPDO and CBPDO are expected to be applied to PhOLED as a blue phosphorescent host material.

Electrochemical properties
The electrochemical properties of CEPDO and CBPDO were studied by cyclic voltammetry (CV) measurements in deoxygenated DCM solution with 0.1 M tetrabutylammonium hexafluorophosphate as the supporting electrolyte. The cyclic voltammograms are shown in Figure 3.   [21]. There-fore, we successfully synthesized two novel bipolar host materials with higher triplet energy by choosing suitable donor and acceptor units.

Thermal properties
The thermal properties of CEPDO and CBPDO were determined by thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) under nitrogen atmosphere at a scanning rate of 10 °C/min and the results are shown in Figure 4. Both CEPDO and CBPDO show very high thermal stability with decomposition temperatures (T d ) of 409 and 396 °C and glass transition temperatures (T g ) of 167 and 138 °C, respectively. Compared to the introduction of ethyl groups, the introduction of normal butyl groups to the phenothiazine-5,5-dioxide moieties appears to decrease the T d and T g of CEPDO by 13 °C and 29 °C, respectively, relative to those of CBPDO. The reason may be the n-butyl chain is longer. It reduces the polarity and intermolecular forces of molecules. The high thermal values ensure high thermostability and that the amorphous structure can form homogeneous and stable films by vacuum deposition to improve the lifetime of the PhOLEDs. The photophysics, electrochemical and thermal properties of CEPDO and CBPDO are summarized in Table 1.

Conclusion
In summary, we have designed and synthesized two bipolar host materials CEPDO and CBPDO. CEPDO and CBPDO not only have a high triplet energy but also show a bipolar behavior. Moreover, their fluorescence emission peaks are blue fluorescence at 408 nm and the fluorescence quantum efficiency (Φ) of CEPDO and CBPDO are 62.5% and 59.7%, respectively. Both CEPDO and CBPDO show very high thermal stability with T d of 409 and 396 °C, T g of 167 and 138 °C, respectively, and also appear suitable HOMO and LUMO energy levels. Hence, a The absorption maximum of the UV-vis spectrum; b estimated from the onset of the UV-vis spectrum; c emission fluorescence maximum at room temperature; d phosphorescence emission peak at 77 K; e fluorescence quantum yield; f first oxidation peak potential; g E g = 1240/λ onset ; h E T = 1240/ λ phos ; i E HOMO : measured from the oxidation potential in 10 −3 M DCM solution by cyclic voltammetry, E LUMO = E g + E HOMO ; j decomposition temperature (T d ) with 5% loss, glass transition temperature (T g ).
CEPDO and CBPDO are two promising blue phosphorescent host materials for PhOLEDs.

Supporting Information
Supporting Information File 1 Experimental part and copies of NMR spectra.

License and Terms
This is an Open Access article under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The license is subject to the Beilstein Journal of Organic Chemistry terms and conditions: (https://www.beilstein-journals.org/bjoc) The definitive version of this article is the electronic one which can be found at: doi:10.3762/bjoc.14.73 Abstract Phosphorescent organometallic compounds based on heavy transition metal complexes (TMCs) are an appealing research topic of enormous current interest. Amongst all different fields in which they found valuable application, development of emitting materials based on TMCs have become crucial for electroluminescent devices such as phosphorescent organic light-emitting diodes (PhOLEDs) and light-emitting electrochemical cells (LEECs). This interest is driven by the fact that luminescent TMCs with longlived excited state lifetimes are able to efficiently harvest both singlet and triplet electro-generated excitons, thus opening the possibility to achieve theoretically 100% internal quantum efficiency in such devices. In the recent past, various classes of compounds have been reported, possessing a beautiful structural variety that allowed to nicely obtain efficient photo-and electroluminescence with high colour purity in the red, green and blue (RGB) portions of the visible spectrum. In addition, achievement of efficient emission beyond such range towards ultraviolet (UV) and near infrared (NIR) regions was also challenged. By employing TMCs as triplet emitters in OLEDs, remarkably high device performances were demonstrated, with square planar platinum(II) complexes bearing π-conjugated chromophoric ligands playing a key role in such respect. In this contribution, the most recent and promising trends in the field of phosphorescent platinum complexes will be reviewed and discussed. In particular, the importance of proper molecular design that underpins the successful achievement of improved photophysical features and enhanced device performances will be highlighted. Special emphasis will be devoted to those recent systems that have been employed as triplet emitters in efficient PhOLEDs.

Introduction
Photoactive TMCs have attracted enormous attention in the last two decades because of their peculiar photophysical and rich redox properties, which make them appealing from both fundamental research and technological applications points of view. Nowadays, several research groups have devoted much effort in exploring a large variety of classes of luminescent TMCs with closed-shell d 6 , d 8 and d 10 electronic configurations [1][2][3][4][5]. The concomitant presence of a heavy metal ion and coordinated π-conjugated chromophoric ligands enriches the photophysical features displayed by TMCs when compared to classical organic luminophors. Indeed, apart from ligand centred (LC) and intraligand charge transfer (ILCT) states, admixing of the metal and ligand orbitals close to the frontier region results in excited states featuring a certain degree of metal contribution. In particular, metal-to-ligand charge transfer (MLCT), ligand-to-metal charge transfer (LMCT), ligand-to-ligand charge transfer (LLCT) and metal centred (MC) states actively contribute to the richer photophysical and photochemical features of TMCs and to their resulting properties, also in terms of electrochemistry. Additionally, the presence of a heavy metal atom induces spinorbit coupling (SOC) effects to such an extent that intersystem crossing (ISC) processes become thus competitive over other radiationless deactivation pathways owing to relaxation of spin rules. In this way, long-lived and low energy lying excited states with triplet (T n states) character are accessible and can be efficiently populated. The subsequent deactivation from the lowest lying T 1 state into the electronic ground state (S 0 ) through radiative channels, T 1 → S 0 , occurs with decay kinetics between hundreds of nanoseconds to several microseconds, constituting a formally spin-forbidden transition (phosphorescence). Structural modification of the TMCs and proper tailoring of coordinated ligands can independently act on the nature, energy and topology of frontier orbitals. In fact, a fine modulation is achieved through a precise energetic positioning and mixing of different excited states, as well as tuning of the energetic band gap between S 0 and the lower-lying singlet and triplet manifold excited states. This approach did successfully yield phosphorescent TMCs with an emission wavelength tuneable over the entire visible spectrum and beyond; together with compounds with photoluminescence quantum yield (PLQY) approaching unity. These peculiar features have greatly fuelled the still growing interest in luminescent TMCs for its potential employment in applications and real-market technology including photocatalysis [6], bio-imaging [7,8], and solar-energy conversion [9], just to cite a few.
Thompson and Forrest reported in 1998 on the first example of a phosphorescent emitter, namely 2,3,7,8,12,13,17,18-octaethyl-21H,23H-porphyrin platinum(II) (Pt(OEP)), used as dopant for the fabrication of an efficient (external quantum efficiency, EQE, ca. 4%) OLED device [10]. Since that pioneering work, an impressive amount of research effort has been devoted in the last two decades to seeking for TMCs that display better device performances. In this respect, iridium(III) and platinum(II) derivatives undoubtedly play leading roles as electro-active materials in light-emitting devices. Their outstanding photophysical and electrochemical features enabled fabrication of PhOLEDs and LEECs [11] with enhanced device performances in terms of efficiency, operating lifetime and colour purity. In electrophosphorescent devices, the triplet nature of excited states localized on the active TMCs allows harvesting of both singlet and triplet electro-generated excitons through either direct trapping or energy transfer processes. As a consequence, the theoretical internal quantum efficiency rises from 25%, which corresponds to purely fluorescent-based devices from a first approximation spin statistics, up to 100%. Nonetheless, EQEs are typically upper limited to values of ca. 20-25% owing to differences in the refractive index of organic materials commonly employed and suboptimal light outcoupling. In spite of that, highly performing vacuum-processed devices with record EQEs up to 54% have been reported to date for PhOLEDs based on Ir(III) with optimized light outcoupling [12]. On the other hand, an impressive EQE value as high as 38.8% [13] and 55% [14] have been recently achieved in platinum(II)-based OLEDs without and with outcoupling elements, respectively, via engineering of transition dipole moment orientation in the device active matrix.
Owing to the enormous interest they are currently attracting, the scope of the present review article is to highlight the current trends and achievements in the field of phosphorescent platinum complexes for PhOLEDs with a special emphasis on the most recent advances. It should be noted that this contribution is not indented to be comprehensive and readers are invited to refer elsewhere for previous examples of platinum emitters [15][16][17][18]. In particular, we will focus our attention on recently reported Pt(II) complexes by breaking down the different classes into those containing monodentate, bidentate, tridentate and tetradentate chromophoric ligands, in order to put in context and compare their photophysical and electroluminescent properties. Finally, some very recent and interesting examples of Pt(IV) compounds as triplet emitters in OLEDs, a class of compound that has been much less explored, will also be reviewed. PhOLED performances of devices comprising the examples reviewed herein are summarised in Table 1.

Review Platinum(II) complexes
Platinum complexes bearing mono-dentate ligands Platinum(II) complexes bearing monodentate ligands are likely to have very poor luminescent properties. In these complexes, the molecular flexibility as a consequence of the low denticity favors efficient thermal deactivation via MC excited states and other nonradiative relaxation pathways. Schanze and co-workers have demonstrated, however, that it is possible to obtain satisfactory photo-and electroluminescence from transplatinum(II) complex 1 bearing only monodentate ligands (Figure 1) [19]. In this derivative, the MC states were efficiently destabilized by selecting strong σ-donating NHC and -C≡C-R ligands. The presence of the two bulky cyclohexyl substituents on the imidazolylidene moiety contributed to rigidify the structure, as well as avoid detrimental intermolecular interactions. Though being weakly emissive in THF solution, the compound exhibited a narrow deep blue photoluminescence (CIE = 0.14, 0.12) with a PLQY of 0.30 in PMMA films. Multilayer vacuum-processed OLEDs were fabricated to test the electroluminescence performance of this complex. A remarkable value of 8% of EQE was attained, but a severe rolloff efficiency was observed with an EQE value dropping to 2% at a practical brightness of 500 cd m −2 . Nevertheless, this work opens the door for a novel design of highly efficient deep-blue phosphors. More complex structures based on dinuclear platinum(II) complexes have also been recently described [20]. Upon using two 1,3,4-oxadiazole-2-thiol as bridging ligands coordinating two Pt(II) centers in a monodentate fashion, Zhu and co-workers have reported on dimeric structures, namely 2 and 3, exhibiting an interaction between the two metallic centres (Pt···Pt distance of ca. 3 Å) (Figure 2). The appearance of a triplet metal-metalto-ligand charge transfer ( 3 MMLCT) transition led to NIR emission with PLQY of ca. 0.31. These bimetallic compounds were tested as dopants in solution-processed PLED, achieving EQE values up to 5.2% at 100 mA cm −2 , even though with relatively high turn-on voltages of 10.4-14.6 V. However, molecular aggregation was observed at dopant concentrations above 12 wt %.
Although (hetero-)metallic clusters are beyond the scope of this review, it is worth to mention some recent reports from Chen and co-workers on trimetallic systems based on PtAu 2 [21,22] and PtAg 2 [23,24] core. Motivated by very high PLQYs in doped films, OLED devices were fabricated showing remarkable efficiency attaining EQE of 21.5% at a luminance of 1029 cd m −2 with small roll-off [21]. These performances are the best reported so far for such a practical luminance.

Systems based on bidentate ligands
In the past, the most common synthetic strategy to obtain luminescent platinum(II) complexes has been the use of π-conjugated chelating ligands with a bidentate motif bearing π-accepting (hetero)aromatic units. Compared to monodentate ligands, the more rigid structure of the bidentate motif is expected to reduce excited-state molecular distortion and access to quenching channels to some extent. On the other hand, the appearance of new low-lying excited states associated to the π molecular orbitals typically results into efficient emission due to their larger radiative decay rates [25].
Though limited in the 1980s by their sensitive synthesis via lithiated species, archetypical luminescent platinum(II) complexes were based on 2-phenylpyridine (ppy) and its derivatives. The combination of the strong σ-donor effect of the phenylate and the π-accepting character of the pyridine ring results in a high ligand-field for the coordinated metal, thus raising the energy of quenching d-d states while lowering emissive MLCT and LC excited states. Alternatively, the use of N-deprotonable azole units has also been largely explored due to the fact that it can exert similar effects to ppy-like ligands [16]. Nevertheless, easier deprotonation of the N-H site in comparison with ppy chelates notably widens applicability and increases the chemical structure diversity of the final luminophors, e.g., for complexating metal ions less prone to undergo cyclometallation reactions. Extensive work based on azolatetype of ligands has been developed by the group of Chi [16] who has recently described a series of neutral platinum(II) complexes bearing isoquinolylpyrazolates, complexes 4-7 in Figure 3 [26]. Control on the intermolecular interactions was exerted through the substitution pattern, yielding solids that exhibited mechano-and solvatochromic properties. Indeed, bathochromic shifts in the emission energy were observed upon either grinding or incrementing solvent polarity. This emission was attributed to a radiative transition with triplet metal-metalto-ligand charge transfer character ( 3 MMLCT), which ultimately strongly depends on the platinum···platinum intermolecular distance. These compounds were also suitable OLED dopants, achieving high EQE of 8.5-11.5%. Nevertheless, the electroluminescence was slightly broader than the corresponding photoluminescence due to incomplete suppression of the intermolecular interactions.
Taking advantage of the easy generation of anionic ligands from azoles, the same group described the preparation of neutral platinum(II) complexes resulting from the combination of dianionic with neutral chelates (Figure 4) [27]. Compounds 8 and 9 were weakly emissive in solution. Nevertheless, the solid-state emission of these particular heteroleptic complexes was switched on notably. Apart from reduced geometry distortions within a rigid environment, the presence in some cases of interligand H-bonding interactions further contributed to efficiently suppress nonradiative decay channels. More important-ly, these supplementary interactions reinforced the ligand-metal bond, which explains well the remarkable phosphorescence efficiency obtained in solid-state thin films being PLQY of 0.  On the other hand, strong σ-donor NHC carbenes ligands could be regarded as the neutral variant of phenylate-like counterparts [28][29][30]. Apart from the strong σ-donor ability, the great Figure 6: Cyclometalated thiazol-2-ylidene platinum(II) complexes with different acetylacetonate ligands [37].
interest for these ligands relies on the robustness that they confer to the resulting complexes, upon coordination onto both early [31] and late transition metals [32,33]. In this regard, the group of Chi employed carbene-based chelates as neutral imine substitutes in an attempt to further improved the stability and the performances of their N···H-C stabilized phosphors ( Figure 5) [34,35]. Either when one, compound 10 [34], or two, compound 11 [35], carbene moieties were used, the resulting platinum compounds were basically nonemissive in solution.
On the contrary, they became strong emitters in the solid state owing to the switching of the nature of the excited state that becomes 3  The beneficial effect of carbene moieties on the photophysical features of the dopant was also shown by Strassner and co-workers [36][37][38]. Compared with previously reported imidazolylidene and triazolylidene acetylacetonate (acac) platinum(II) complexes, complexes 12 bearing 1,3-thiazol-2- Figure 5: Selected neutral platinum(II) complexes from bipyrazolate and carbene-based chelates [34,35].
ylidene carbenes outperformed the former when evaluating the photophysical properties ( Figure 6) [37]. The intermolecular interaction was finely tuned as a function of the steric hindrance of the acac-type ancillary ligand, which had a profound impact on the emission quantum yield. Characterization of the electroluminescence performances of these complexes in mixed-matrix OLED led to EQE values as high as 12.3%, CE of 37.8 cd A −1 and PE of 24.0 lm W −1 at 300 cd m −2 for complex 12f.
In spite of typical TTA processes at high concentrations for phosphorescent dopants, azolate-containing platinum(II) complexes have recently shown great potentiality for the fabrication of non-doped OLEDs. In fact, Wang and collaborators reported a red-emitting device based on Pt(fppz) 2 [39], where fppz is 3-(trifluoromethyl)-5-(2-pyridyl)-1H-pyrazolate, that attained remarkable EQE of 31% [40] (see Figure 7 for the chemical structure of the complex). With the aim of correlating molecular structure, photophysical properties and OLED performances, Chi, Kim and co-workers analyzed the X-ray structures of Pt(fppz) 2 (13) and other related platinum(II) complexes 14 and 15 in both single crystal and thin film samples (Figure 7) [13]. They observed different degrees of crystallinity as a function of the substrates, though the crystal pattern of the investigated compounds was not affected. More interestingly, upon analysis of angle-dependent emission intensities at various wavelengths along with the birefringence of the films, the authors concluded that the arrangement of the complexes within  The beneficial effect of the emitting dipole orientation on the light outcoupling efficiency was further illustrated in a following work by the group of Chi [14]. Exploiting the strong tendency to form ordered structures, a new series of platinum(II) bearing fluorinated 2-pyrazinylpyrazoles was developed, namely complexes 16-18 in Figure 8. Upon aggregation, very efficient NIR emission arising from a 3 MMLCT excited state with PLQY as high as 0.81 was obtained. As aforementioned, the perpendicular molecular arrangement, together with a highly ordered structure, allowed the exciton to diffuse over long distances with minimal vibrational relaxation to the ground state. Among these dopants, incorporation of 16 into an optimized planar non-doped OLED structure with archi-tecture as follows ITO (100 nm , led to an EQE of 24 ± 1%. This result was even improved when a light outcoupling hemisphere structure was employed, achieving outstanding values of EQE up to 55 ± 3%. This performance is the highest reported so far for a NIR OLEDs. Therefore, these works nicely showed how both crystallinity and molecular orientation are key parameters that can make great differences for the resulting thin-film optoelectronic performances. Apart from display applications, general lighting efficiency currently constitutes a main concern of our society and white-emitting OLEDs (WOLEDs) represent a valuable alternative because of their energy-saving potential. In this regard, development and improvement of white-light emitting devices attracts considerable interest. Nowadays, two main fabrication strategies seemed to be the most promising ones such as i) including either three (RGB) or two emitting components (skyblue-orange); ii) using a phosphorescent material to partially down-convert UV or blue light from a LED source; the latter seems a promising option to date. The group of Sicilia has recently applied some cyclometallated platinum(II) complexes bearing NHC ligands to develop WOLEDs, whose chemical structure is sketched in Figure 9 [41]. Depending on the π-conjugation of the NHC-based bidentate ligand, emitting complexes with luminescence varying from blue (19 and 20) to yellow (21) were obtained. Several devices were prepared following a remote phosphor configuration, which places the phosphors spatially separated from the LED source. The associated values of correlated colour temperature (CCT), colour rendering index (CRI) and luminous efficacy of the radiation (LER) were acceptable, proving the suitability of these systems for lighting applications. Nevertheless, a fast degradation of the emission was observed under device operation.

Systems based on tridentate ligands
During the last two decades, platinum(II) complexes bearing tridentate ligands have been extensively investigated as well. Compared to their mono-and bidentate counterparts, a threefold chelating motif imposes higher geometrical rigidity, which is expected to further decrease molecular distortions. The overall stability of the resulting compound is increased, thus helping to greatly suppress nonradiative deactivation pathways. Although 2,2':6',2"-terpyridines showed widespread use in coordination chemistry [42,43], the bite angle of such class of tridentate ligands is not ideal for a square-planar geometry, leading to longer bond lengths when compared with their bidentate congeners. As a consequence, ligand-field is reduced and the presence of low-lying d-d excited states provide easy access to nonradiative deactivation channels [25,44].
Nevertheless, the use of multidentate chromophoric ligands that are able to provide metal-ligand bonds with higher covalent character, as for instance cyclometalating ligands, has proven to be a successful strategy for improving the luminescence proper-ties due to the energetic destabilization of quenching MC states [45,46].

Complexes based on C^N^N ligands
Following the seminal work of von Zelewski [47,48] on platinum(II) complexes bearing C-deprotonated 2-phenylpyridines (C^N), the development of tridentate analogues has received a great deal of attention in the recent past. Early reports were based on 6-phenyl-2,2'-bipyridine, namely C^N^N [49,50]. In spite of the strong ligand field exerted by the cyclometalating moiety, this type of complexes resulted to be rather weakly emissive due to large structural distortion of the emitting triplet excited state. Nevertheless, Che and co-workers demonstrated that extending the π-conjugation of the cyclometalated ligand led to enhanced phosphorescence quantum yields [51,52]. Indeed, the increased conjugation resulted in a modification of the frontier molecular orbitals and prevention of Jahn-Teller distortions.
Recently, Che and co-workers reported a series of asymmetric tridentate C^N^N platinum(II) complexes with π-extended moieties, compounds 22 ( Figure 10) [53]. Depending on the ancillary ligand, these complexes showed emission arising from several contributions, being 3 MLCT and 3 ILCT, together with 3 XLCT or 3 LLCT, where XLCT is a halogen-to-ligand charge transfer, with PLQY values approaching unity for some derivatives. Different structural isomers were synthesized, including a π-conjugated fragment attached at different positions of the employed tridentate ligand. The best results were obtained when the azine moiety isoquinolin-3-yl was used due to the minimization of the repulsions within the tridentate scaffold as well as with the ancillary ligands. Based on these initials results, new structural variations were investigated at both the cyclometalating and the ancillary ligands. As for the former, a clear impact on the emission colour was observed due to the participation of the cyclometalating unit to the HOMO frontier orbital. Thus, an emission ranging from green to yellow and finally to red was obtained going from phenyl, thiophene and benzothiophene cyclometalating rings, respectively. On the contrary, the ancillary ligand had a remarkable effect on the emission efficiency. In the case of pentafluorophenylacetylide, the change in the nature of the emitting excited state led to an almost negligible k nr value, which resulted in an outstanding PLQY close to unity. The most promising complexes were selected by the authors as dopants for OLED fabrication and their chemical structure is displayed in Figure 10.

Complexes based on N^C^N ligands
Although formally bearing similar coordinating units, platinum(II) complexes bearing symmetrical N^C^N ligands resulted in better emitters than those bearing the corresponding C^N^N motif. For instance, while [Pt(C^N^N)Cl] (C^N^N = 6-phenyl-2,2'-bipyridine) possess a rather low emission (PLQY = 0.025) in degassed CH 2 Cl 2 solution at room temperature [50], [Pt(N^C^N)Cl], where N^C^N is a bis-cyclometalating 2,6-dipyridylbenzene type of ligand (complexes 23), displays a much higher PLQY reaching 0.60 in similar conditions, as for instance compound 23a [54]. The chemical structure of complexes 23 is shown in Figure 11. These distinct results can be interpreted as follows. A shorter Pt-C bond length was observed for the N^C^N-containing complex, revealing a stronger interaction with the metallic ion. As a consequence, a higher d-d splitting could be foreseen, thereby reducing the possibility of a non-radiative deactivation channel of the emitting excited-state. On the other hand, [Pt(N^C^N)Cl] displayed a metal-perturbed 3 π-π* emission as also demonstrated by the relatively high radiative rate constant value. The combination of these two factors explained well the aforementioned good emission efficiencies. As a result, N^C^N-coordinated complexes have found numerous applications as emitting materials in areas such as emitters in PhOLEDs [55,56] and luminescent probes in bio-imaging [57][58][59]. Noteworthy, NIRemitting OLED were fabricated by using complexes 23g and 23h, which presented a π-delocalized substituent at the 5-position of the central phenyl ring. As the parent complex 23a, excimer formation via metal-metal interactions was observed for both derivatives at high concentrations or in neat films. Nevertheless, the increased conjugation within the chromophoric ligand led to a lower emission energy, which fell into the NIR region. The structure of the optimized vacuumprocessed OLED was as follows: ITO (120 nm)/Mo 2 O x (2 nm) / TCTA (80 nm) /23g or 23h (15 nm)/TPBi (25 nm)/LiF (0.5 nm)/Al (100 nm). Complex 23g attained remarkable performances for this class of Pt(II)-based compounds, with an EQE of 1.2% at a current density of 10 mA cm −2 and an electroluminescence intensity of about 10 mW cm −2 at 9 V. Figure 11: Chemical structure of platinum(II) complexes 23 bearing bis-cyclometalating 2,6-dipyridylbenzene type of ligands [54][55][56].
Due to the triplet character of typical platinum(II) complex emission, these metal-based dopant phosphors are typically dispersed in high triplet energy hosts to suppress energy transfer processes onto the host matrix that detrimentally affect the final performances [60]. Alternatively, development of emissive complex incorporated in a dendritic structure allows controlling both charge transport and light emission in a single material [61]. In this regard, Yam and co-workers reported on a series of dendritic carbazole-based alkynylplatinum(II) complexes with cyclometalated 2,6-bis(N-alkylbenzimidazol-2'-yl)benzene (bzimb) as the N^C^N tridentate ligand [62]. These complexes were found to be highly emissive with PLQYs of up to 0.80 in Figure 12: Molecular structure of dendritic carbazole-containing alkynyl-platinum(II) complexes 24a-d [62].
solid-state thin films. Contrarily to other alkynylplatinum(II) complexes, their emission was ascribed to an admixture of 3 IL/ 3 MLCT since no influence of the dendrimeric ancillary ligand was observed. Nevertheless, upon increasing the dopant concentration in thin films up to 50 wt %, a new low-energy band was observed that was attributed to the formation of excimeric species. Nonetheless, it is worth to note that this excimeric emission was reduced on increasing the generation of the ancillary ligand, highlighting the importance of this molecular design strategy towards highly efficient dopants. The interesting photophysical properties of these compounds prompted the evaluation of their electroluminescence performances in OLED devices. Solution-processed green-emitting PhOLEDs were prepared with the structure of ITO/poly(ethylenedioxythiophene):poly(styrene sulfonic acid) (PEDOT:PSS 70 nm)/mCP:24 5-50 wt % (60 nm)/SPPO13 (30 nm)/LiF (0.8 nm)/Al (100 nm), where SPPO13 is 2,7-bis(diphenylphosphoryl)-9,9′-spirobifluorene. For all devices, emission similar to those recorded in solution was obtained independently of the doping concentration. Moreover, the decreasing driving voltages measured were ascribed to better charge transport properties in the emissive layer upon increasing the dendron generation. However, the best hole-electron current balance was achieved for a platinum(II) complex with the second generation dendrimeric structure (Figure 12), yielding a maximum CE and EQE of 37.6 cd A −1 and 10.4%, respectively. This enhanced Figure 13: Molecular structure of bipolar alkynyl-platinum(II) complexes 25 bearing carbazole and electron-accepting phenylbenzimidazole or oxadiazole moities [63].
performance highlights the beneficial effect of employing emitters with a dendrimeric design. Indeed, these results were among the best values ever reported for PhOLEDs based on metal-containing dendrimers, and even compared well with vacuum-deposited devices of non-dendritic structurally-related platinum(II) complexes.
As a further development of the work, the same group reported very recently another family of platinum(II) complexes containing both electron-donor and electron-acceptor moieties embedded within the dopant structure ( Figure 13) [63]. This bipolar character was intended to reduce the TTA phenomena commonly experienced at high current density that leads to severe rolloff efficiency of OLEDs [64]. In particular, carbazole-based donor moieties and either phenylbenzimidazole (PBI) or oxadiazole (OXD) accepting units were selected as the hole-transporting and electron-transporting moiety, respectively. Two linkage fashions were explored between these donor-acceptor groups, namely metaand para-substitution. In another study from the group of Yam, the bipolar design was conceived to finely tune the emission energies of the compounds [65]. Two series of platinum(II) alkynyl (compounds 26) and carbazoyl (compounds 27) complexes were reported, which included different donor and/or acceptor groups on the ancillary ligand ( Figure 14). As expected, their emission behavior was strongly dependent on the nature of this latter, displaying different combinations of π-π* and charge-transfer triplet excited states, together with a broad emission ranging from the green to the red portion of the spectrum. Interestingly, a solution-processed OLED fabricated with a complex bearing a carbazoyl ancillary ligand showed concentration-dependent electroluminescence. In addition, a change in nature of the emission from 3 IL to 3 MLCT/ 3 LLCT character was observed upon increasing doping concentration from 5 to 20 wt %. Moderate performances were attained at this latter concentration, with CE of 24.0 cd A −1 and EQE of 7.2%. Alternatively, these compounds were successfully employed in the fabrication of organic memories, which demonstrates the great versatility of this class of platinum(II)-containing materials.

Complexes based on bis-anionic C^N^C and N^N^N ligands
In an attempt to further destabilize the d-d excited states, doubly cyclometalating 2,6-diphenylpyridine [66,67] and their extended π-conjugated analogues have been employed as C^N^C tridentate ligands for platinum(II) complexes. Nevertheless, the resulting complexes resulted to be almost nonemissive in solution at room temperature in spite of the stronger ligandfield exerted. Similar to the case of C^N^N type of ligands, a significant structural distortion is the main factor that accounts for this low emission efficiency. However, Che and co-workers demonstrated that extension of the π-conjugation at the tridentate ligand, together with the use of heterocyclic moieties such as thiophene or carbazole, clearly favours the luminescence properties of these type of platinum(II) complexes [68].  As aforementioned, N-deprotonable azole units constitute a compelling alternative to C-cyclometalating ligands [16]. In this regard, dianionic tridentate N^N^N ligands bearing pyrazolate [69], triazolate [69][70][71][72][73] or tetrazolate [74] units have been used to successfully prepare highly luminescent neutral platinum(II) complexes in dilute solution and/or as aggregated state. Due to their promising emitting features, these complexes have also been employed as phosphors in optoelectronic devices [71,72]. Neutral platinum(II) complexes with an asymmetrical triazolate-and tetrazolate-containing tridentate ligand, complexes 28, were also reported [75] (Figure 15). These green emitters were used to fabricate solution-processed PhOLEDs, displaying performances as high as their vacuum-processed structurallyrelated analogues, with a maximum PE of 16.4 lm W −1 , CE of 15.5 cd A −1 and EQE of 5.6% obtained for derivative 28b. These performances are amongst the highest EQE values for solution-processed platinum-based OLEDs.

Systems based on tetradentate ligands
Tetradentate ligands have attracted an increased attention due to the even higher rigidity of the chromophoric scaffold that helps to suppress nonradiative decay pathways induced by large distortions around the metal atom [76,77].
Following on their strategy of employing rigid N^C^C^N and C^C^C^N ligands bearing either methyl-2-phenylimidazole or phenylpyrazole moieties [78], Li and co-workers recently reported on a series of tetradendate platinum(II) complexes 29-32 that displayed narrow emission spectral bandwidth ( Figure 16) [79]. In such derivatives, the introduction of an electron-donating moiety, such as a tert-butyl group, onto the pyridyl ring of the tetradentate scaffold induces a larger energy separation between the carbazolepyridine and the phenylpyrazolate moieties.
In consequence, spectra are narrowing and a higher colour purity can be achieved by reducing vibronic sideband contributions to the overall emission spectrum. Variation of the emissive moiety from the methylimidazole or phenylpyrazole to the 4-phenylpyridyl carbazole afforded compound 33 ( Figure 17). This complex displayed an emission maximum at 602 nm in CH 2 Cl 2 arising from an excited state with strong 3 MLCT character with PLQY of 0.34 ( Figure 17) [80]. OLEDs were fabricated with device architecture as follows: ITO/HATCN(10 nm)/NPD(40 nm)/TrisPCz (10 nm Seeking for stable and efficient blue emitter for OLED devices and following the previous work on the red-emissive compound 33 and the green-emissive derivative 36 that showed a peak EQE of 14.3% [83], Li and co-workers developed a novel blue-emitting tetradentate platinum complex, namely 37. The excited state of this compound was raised by breaking the π-conjugation of the carbazole moiety upon introduction of 9,10-dihydro-9,9-dimethylacridine moiety, where the two methyl groups were introduced to minimize oxidation of the benzyl carbon under device operation ( Figure 17) [84]. Compound 37 exhibited a maximum of emission at 486 nm with a spectrum characterized by vibronic features, most likely due to the increased flexibility of the acridine moiety that imparted a more distorted excited state geometry compared to the carbazole-based counterpart. Upon device optimization, 37 resulted to be a rather efficient sky-blue triplet emitter. In particular, OLEDs with the following architecture ITO/HATCN A similar strategy based on the rupture of the π-conjugation in a cyclometalating ligand was employed by the same authors to achieve blue emission in symmetric tetradentate platinum(II) complexes 38 bearing six-membered pyridyne-carbazole chelating rings [85]. This latter compound showed modest (PLQY = 0. 31 [86] or bis(1,2,4-triazolylphenyl) ligands [87]. Examples of the former class, namely complexes 39 and 40, are displayed in Figure 18.
In particular, these complexes were designed to reduce excitedstate distortions by bearing a macrocyclic chelating ligand and either ether, methylene or carbonyl bridging units. The derivatives showed bright blue phosphorescence centred at λ em ca. 448-470 nm depending on the bridging unit. Such blue emission was retained when the complexes were embedded in PMMA rigid matrix. Interestingly, macrocyclic derivatives possessed higher PLQY in solution with values of 0.58-0.62 when compared to non-macrocyclic counterparts that was attributed to the enhanced structural rigidity imposed by the cyclic structure. By employing complex 39 as emissive material OLED devices with the following architecture were fabricated: ITO/NPB (50 nm)/mCP (10 nm)/9,9′-(4,4′-(phenylphosphoryl)bis(4,1-phenylene))bis(9H-carbazole) (BCPO):complex 39 x wt % (20 nm)/bis [2-(diphenylphosphino)phenyl] ether oxide (DPEPO) (10 nm)/TPBi (30 nm)/LiF (1 nm)/Al (100 nm) with doping level x of 2, 5 and 10 wt %. EL spectra showed an emission peak at λ EL = 452 nm that did not show any dependency on the doping concentration and a rather low turn-on voltage of 3.2 V. The best EL performances were recorded for the OLED device at 10 wt % doping level that showed peak brightness, CE and PE of 10680 cd m −2 , 11 cd A −1 and 10.8 lm W −1 , respectively, and EQE value of 9.7%. In a second set of deep-blue OLED devices, maximum EQE of 15.4% were achieved at brightness of 490 cd m −2 . Figure 18: Chemical structure of the macrocyclic tetradentate platinum complexes reported by Wang and co-workers [86].
Other classes of tetradendate platinum(II) complexes bearing N^C^C^N chromophoric ligands have been recently reported by Fan and coworkers [88,89]. In order to prevent detrimental intermolecular interactions which might largely affect colour purity and emission efficiency in a condensed state, as well as increase solubility of the complex, the authors developed a series of (2-phenylbenzimidazole)-based tetradentate Pt(II) complexes bearing a diisopropylphenyl group, which is orthog- Figure 19: Molecular structure of complex 41-46 [88,89].
onally oriented with respect to the molecular plane [88]. The three complexes featured 2-pyridylcarbazole (41), 2-thiazolylcarbazole (42) and 2-oxazolylcarbazole (43) moieties employed as the luminophoric motifs that were linked to the 2-phenylbenzimidazole unit through an ether bridge ( Figure 19). The three complexes exhibited high thermal stability since thermogravimetric analysis (TGA) showed a weight loss of only 5% at temperatures in the range 436-463 °C. An intense and structured emission in the green region with λ em = 500-507 nm and PLQY = 0.6-0.78 was recorded when the complexes were used as dopant in PMMA thin-film. DFT calculations helped to ascribe the nature of the frontier molecular orbitals as being carbazole/phenoxy and phenylbenzimidazole for HOMO and LUMO, respectively.
In a following study, a second series of tetradentate platinum complexes bearing a pyrazolo[1,5-f]phenanthridine moiety and with a general coordination motifs of the type N pyridine^Cphenyl^Cphenyl^Npyrazole was reported by the same group, namely complexes 44-46 ( Figure 19) [89]. The complexes showed moderate to intense sky-blue emission with PLQY in the range 0.2-0.7 and high thermal stability. Unfortunately, going from dilute solution to neat solid-state samples, PLQY values dramatically dropped to values as low as 0.10-0.02 that might point to strong intermolecular interaction and TTA phenomena. The tendency toward aggregation for complex 44 and 46 in condensed phase was also evidenced in the EL spectra. Although its shape was independent from the doping ratio, a bathochromically shift was observed along with a featureless emission profile. In sharp contrast, compound 45 displayed an EL emission maximum similar to that observed for the solution sample, indicating a much less pronounced aggregation. OLED devices were fabricated with the following con- The same authors have recently reported on another class of asymmetric [90] platinum complexes featuring tertiary arylamine motifs and their chemical structure is displayed in Figure 20. Such complexes, whose structure is derived from the parental symmetric systems previously reported by Huo and co-workers [91], bear a 3-methylindole, a carboxylic and a dangling phenoxy moiety, complex 47, 48 and 49, respectively, resulting in a general ligand structure with general formula being either C^N^N^C or C^N^N^O.
The compounds displayed moderate emission in the greenyellow portion of the visible spectrum with λ em maximum peaking at 504-513 nm and PLQY of 0.27-0.47, attributable to an excited state with main LC character as suggested by the vibronic profile of the spectrum, repectively. Employment of these complexes as triplet emitters in OLEDs with configuration ITO/HATCN (10 nm)/TAPC (40 nm)/TCTA/mCP: platinum complex 10 wt %/TmPyPb (40 nm)/Liq (2 nm)/Al (120 nm) afforded electroluminescent devices with peak EQE of 13.3% and 13.6% for 48 and 49, respectively. Even a higher peak EQE value of 16.3% was achieved for devices fabricated with 47 at similar doping level, although colour purity of the device resulted to be affected due to the fact that the EL emission resembles the PL spectra recorded in doped PMMA thin films rather than solution sample. This spectral broadening and shift is most likely due to the establishment of intermolecular interactions at such high doping level.
Indeed, platinum(II) complexes are well known to show both ground state aggregation phenomena including formation of metallophilic d 8 ···d 8 interactions and/or π-π stacking of the coordinating ligands [67,92] as well as excited-state interactions such as formation of excimers [93,94]. Although they may be usefully employed to shift both absorption and emission spectra, obtain long-range ordered luminescent supramolecular architectures and fabricate white-light emitting devices, aggregation phenomena of luminophors is typically considered detrimental due to the TTA and aggregation cause quenching (ACQ) processes that might take place. Thus, several strategies have been employed to date to avoid platinum emitters in close proximity, including introduction of bulky groups such as adamantyl [71] and spiro moieties [95]. By introducing on N^C^N^O tetradentate motifs both tert-butyl and spiro groups, Fan and co-workers recently reported on two platinum complexes, 50-51, bearing a phenylpicolinate moiety. Their chemical structure is sketched in Figure 21 [96]. The complexes displayed structured luminescence with moderate PLQY (ca. 0.2) and relatively long lived-excited state lifetime in the range 8.4-11.6 μs. It is worth to notice that the presence of several bulky groups successfully suppressed aggregation as demonstrated by the similar PL spectra recorded in dilute CH 2 Cl 2 and solid-state samples. Upon host material and doping ratio optimization, OLED devices achieved maximum EQE of 22.9% for complex 50 with relatively low roll-off efficiency that is attributed to the reduced quenching processes at high current density imparted by the bulky groups. Spirofluorene and spiroacridine groups were also employed by Chi and co-workers on azolate-based tetradendate platinum complexes bearing either N trz^Npy^Npy^Ntrz (52) and N pz^Npy^Npy^Npz type (53 and 54) of ligands where trz and pz and py is a trifluoromethyltriazolate, trifluoromethylpyrazolate and pyridine ring, respectively [97] (Figure 22). This strategy has proven to enhance solubility and processability during device fabrication as demonstrated for a related Os(II) compound [98].
Photophysical characterization showed that complexes 52, 53 and 54a exhibited a structured and intense (PLQY = 0.58-0.8) blue emission with emission maxima at 452-465 nm. Complex 54b was characterized by a large solvatochromic effect as a consequence of the large variation of the transition dipole moment from S 0 to T 1 states of 29.33 D. Indeed, while a structured phosphorescence ascribed to a 3 LC/ 3 MLCT transition has been observed in cyclohexane, a much broader and featureless profile is recorded in CH 2 Cl 2 and ethanol, which underlies involvement of an emitting excited state with sizeable ILCT character becoming stabilized in such more polar solvents. The two derivatives displaying the highest PLQY among the series, namely 53b and 54b, were employed as triplet emitters in OLED device with architecture comprising an enlarged carrier recombination zone, such as ITO/TAPC (40 nm)/mCP:platinum complex 8 wt % (17 nm)/DPEPO platinum complex 8 wt % (3 nm)/TmPyPB (50 nm)/LiF (0.8 nm)/Al (150 nm). Devices fabricated with complex 54b showed the highest peak efficiency of 15.3% with lower roll-off that was attributed to the better charge transport ability of compound 54b. Furthermore, by combination of sky-blue emitter 53b and 54b and a red emitting osmium complex reported elsewhere [99], WOLED with a sandwiched recombination zone blue/red/blue emitters displayed warm-white emission with peak EQE of 12.7, CRI of 64 and CIE coordinate of 0.365, 0.376 at 1000 cd m −2 .
Achieving efficient electroluminescence into the deep red and NIR region represents a challenging research topic of current interest, and only few examples are reported up to now showing remarkable performances [14]. Such challenge mainly arises from the intrinsic increase of the nonradiative rate constant upon decreasing the energy gap between excited and ground state that follows an exponential law known as energy gap law (EGL) [100]. In this respect, Su, Zhu and co-workers reported on two series of salophen-based tetradentate platinum(II) complexes decorated with donor-acceptor moieties such as triphenylaminophenazine [101] and triphenylaminobenzothiadiazole [102] and their chemical structure is shown in Figure 23.
All the complexes displayed long-lived red-to-NIR emissions in both solution and solid-state samples. The deepest red maximum was recorded for complex 57 with a maximum centred at λ em = 697 nm arising from a triplet excited state with admixed MLCT/ILCT character as a consequence of the large donor-acceptor character of the ligand [102]. By employing complex 57 as triplet emitter in solution-processed OLED featuring a single-emissive layer, devices with architecture ITO/ PEDOT (40 nm)/PVK:OXD-7:Pt complex 1-4 wt % (50 nm)/ TPBI (30 nm)/Ba (4 nm)/Al (100 nm) were fabricated showing emission maximum λ EL = 703 nm and peak EQE of 0.88% with relatively low roll-off efficiency upon increasing current density.

Platinum(IV) complexes
The first examples of luminescent platinum compounds with +IV oxidation states were reported by Balzani and von Zelewski back in the late 80s [103]. The complexes contained bis-cyclometalating (C^N) ligands of the general formula Pt(C^N) 2 (CH 2 Cl)Cl and were prepared by a photooxidative addition of CH 2 Cl 2 onto the corresponding bis-cyclometalated Pt(II) parental complexes. Although Pt(IV) complexes have attracted great attention in cancer therapy [104][105][106], only in the very recent past they are receiving increasing interest as luminescent compounds [107,108]. Such derivatives are characterized by long-lived triplet-manifold π-π* excited states with either 3 LC or 3 ILCT nature. Most of the so far reported examples of octahedral Pt(IV) derivatives are based on heteroleptic and homoleptic systems containing phenyl-pyridine-type cyclometalating (C^N) ligands, reaching PLQY up to ca. 0.80 [109]. To date, only two examples of Pt(IV) derivatives, namely 58 and 59, have been reported to be employed as active compounds in polymer-based OLEDs and their chemical structure is reported in Figure 24 [110]. The compounds contain a biscyclometalating tetradentate ligand scaffold based on phenylisoquinoline moiety decorated with hole-transporting triphenylamine groups, and two chlorine ancillary ligands in trans geometry. The complexes showed NIR luminescence (λ em ca. 750 nm) in dilute 2-methyltetrahydrofuran solution and longlived excited states with lifetime in the order of 0.7 μs.   To explore the potentiality of such phosphorescent Pt(IV) compounds as active materials in electroluminescent devices, solution-processed OLEDs with the following architecture ITO/ PEDOT (40 nm)/PVK:complex (50-60 nm)/TPBi (30 nm)/Ba (4 nm)/Al, where PVK is poly(9-vinylcarbazole), were fabricated with dopant concentration adjusted in the range 1-8 wt % and their EL performances investigated. The devices showed interesting NIR emission similar to the PL spectra with emis-sion maximum at λ EL of about 750 nm for both compounds. Maximum radiant intensity and EQE of 164 μW cm −2 and 0.85% were recorded for compound 59 with relatively low rolloff at higher current densities.

Conclusion
In conclusion, we have here reviewed the most recent trends in the field of phosphorescent platinum complexes, and their use  as phosphors in light-emitting optoelectronic devices such as OLEDs. Indeed, such class of luminescent complexes still represents a fascinating research topic of enormous current interest, in particular in the case of derivatives with oxidation state +II. This is because these systems possess excellent photophysical properties that can be tuned by judicious molecular design through ligand modification. Seeking for emitters with improved features, interesting examples with great structural variety have been reported to date that are based not only on bidentate and tridentate moieties, but recently also on tetradentate scaffolds. Differently from many other transition metals, square planar platinum(II) complexes bearing π-conjugated ligands also possess a peculiar tendency to establish weak intermolecular interactions, such as metallophilic and π-π interactions. These additional features could further widen the already available chemical toolbox for designing highly efficient electrophosphorescent solid-state materials in the near future.
Overall, design efforts have allowed the achievement of impressive OLED performances for devices embedding platinumbased triplet emitters with EQE above 30%. Such results have been achieved thanks to the combination of molecular and dipole moments orientation engineering in the electroactive thin film. Finally, recent reports on platinum(IV) derivatives demonstrate that this type of complexes do also possess interesting photophysics and therefore, further growing interest in their use as emitters in OLEDs could be also foreseen.