Bulky Iridium NHC Complexes for Bright, Efficient Deep‐Blue OLEDs

Four new deep‐blue‐emitting iridium(III) NHC complexes containing sterically demanding ligands are synthesized. The four complexes show bright, deep‐blue emission, with emission maxima between 420 and 427 nm in both acetonitrile solution and 30 wt% doped films in TSPO1; the two meridional isomers showing photoluminescence quantum yields, ΦPL, in doped films of 80% and 89%. The two meridional isomers are used to assess the impact of emitters containing bulky, sterically demanding ligands on the performance of organic light‐emitting diodes (OLEDs). OLEDs employing a stepped doping profile with mer‐Ir(tfpi_tmBn)3 as the emitter produce the highest performing devices in this study, with these devices exhibiting deep‐blue [λEL = 429 nm, CIE = (0.16, 0.08)] emission and a maximum external quantum efficiency (EQEmax) of 14.9%, which decreases to 11.7% at 100 cd m−2. The performance of the OLEDs shows very good efficiencies and moderate efficiency roll‐offs in comparison to reported phosphorescent deep‐blue OLEDs with CIEy ≤ 0.08, as required for commercial displays. The promising results suggest that the design strategy of adding steric bulk to blue emitting iridium complexes containing NHC ligands is a useful strategy for reducing intermolecular interactions between emitters in OLEDs.


Introduction
Organic light-emitting diodes (OLEDs) are becoming the primary display technology in smartphones, smartwatches, and televisions as they exhibit excellent contrast, color reproduction, Tuning the emission color of iridium complexes containing ppy-type ligands through judicious substitution of electrondonating and withdrawing groups is a well-documented strategy, [11] with emission color readily tuned from sky-blue to red. The use of strongly σ-donating NHC ligands results in a destabilized LUMO (relative to ppy-type ligands) and blueshifted emission. The tuning of the emission color of Ir(C^C) 3 complexes is now well understood, where modification of the NHC imidazole with annelated electron-deficient heterocycles can result in a red-shifted emission, as exemplified in fac-Ir(tpz) 3 (λ PL = 480 nm in 2-MeTHF). [7g] Decoration of the phenyl ring with electron-withdrawing or donating groups   in a blue or red shift, respectively, illustrated by the comparison of mer-Ir(CF3pbp) 3 and the related mer-Ir(tbpbp) 3 , an analogous complex that has a t Bu group replacing the electron-withdrawing CF 3 group, with λ PL = 420 nm and 480 nm in DCM, respectively. [3d] Ir(C^C) 3 complexes can exhibit very high Φ PL , with many complexes having Φ PL > 70%. The emission lifetime of most Ir(C^C) 3 complexes are fairly short, with τ PL < 1.5 µs. Complexes with a T 1 of strongly LC character, such as mer-Ir(tfp_tz) 3 , have a longer emission lifetime, in this case τ PL = 4.75 µs.
There are significant challenges in designing suitable device architectures for deep-blue OLEDs. The high triplet energy and shallow LUMO energy of the emitters mean that there are very few suitable hosts, most deep-blue OLEDs use the phosphine oxide-based hosts TSPO1 or DPEPO. [12] There are several issues with the use of DPEPO including poor hole transporting ability and its inherent instability. [13] In the context of being limited to the use of this class of host, in order to improve the performance of deep-blue OLEDs there are a number of device architectures that can be used to partially overcome the deficiencies of the available hosts. For instance, to improve charge balance in the emissive layer (EML) high concentrations of the emitter can be used as the emitter can act as a hole transporting material, this approach can be very successful, although only when the emitter does not undergo aggregation-caused quenching. [3d,7f ] Constraining the excitons within the emissive layer of the OLED is possible with the use of the emitter itself as an exciton-blocking layer. [3b,6] The poor device lifetime of phosphorescent deep-blue OLEDs presents a significant hurdle to their adoption in commercial displays. Unfortunately, device lifetimes are rarely reported in the academic literature; [14] however, the degradation of blue OLEDs has been studied. [15] Several degradation mechanisms have been identified, including via triplet-triplet annihilation (TTA), [16] triplet-polaron annihilation (TPA) [17] and chemical degradation of the emitter. [11c,18] Knowledge of these degradation mechanisms has been used to guide the design and choice of host materials and device structures leading to a steady improvement in device performance. This is exemplified by cross-comparing device lifetimes using the same Ir(cb) 3 emitter ( Figure 1). [9] The first reported OLED using this emitter showed an EQE max of 24.8% with a described short device lifetime. [9a] Jung et al. demonstrated that the device lifetime could be significantly enhanced to LT 50 > 10 000 h at 100 cd m −2 by optimizing the charge balance within the EML by using the bipolar host SiCzTrz. [9c] The blue OLEDs [λ EL = 471 nm and CIE coordinates of (0.12, 0.13)] also showed excellent efficiency and low efficiency roll-off with EQE max of 27.6% that decreased slightly at 1000 cd m −2 to 25.6%.
One approach to improving device performance by reducing bimolecular annihilation processes in blue phosphorescent OLEDs is to increase the steric bulk of the ligands to reduce intermolecular interactions between adjacent phosphorescent complexes. A recent example is mer-Ir(pmp_Bn) 3 (Figure 1), which contains N-benzyl groups. [7f ] This structural change resulted in much improved device performance (EQE max = 24%, L max = 6400 cd m -2 ) compared to the N-Me analog [mer-Ir(pmp) 3 , EQE max = 14.4%)] while retaining deep-blue emission with CIE of (0.151, 0.086). The lifetime of the devices containing mer-Ir(pmp_Bn) 3 was 400 min (LT 50 @500 cd m -2 ), an excellent result for a device showing such deep-blue emission. A similar effect is seen in fac-Ir(tpz) 3 (Figure 1), which contains N-tolyl groups. The sky-blue OLEDs showed excellent brightness (L max = 29 000 cd m -2 ) and minimal efficiency roll-off (EQE max = 18%, EQE 1000 = 16%), although the device lifetime was not investigated.
We had previously synthesized the deep-blue emitting Ir(NHC) complex mer-Ir(tfpmi) 3 with moderate Φ PL of 46.6% in doped films (Figure 1), which produced some of the deepest blue and highest efficiency phosphorescent OLEDs. [3b] In an effort to improve both Φ PL and the corresponding device performance, we have now developed a second generation of deep blue-emitting iridium complexes incorporating bulky groups attached to the imidazole ring of the NHC ligand. We crosscompared the use of the previously reported [3d,7f ] benzyl group (Bn) with the bulkier 2,4,6-trimethylbenzyl (tmBn) group. Optimization of the device structure using the iridium complex as an exciton blocking layer and a stepwise doping scheme led to high-performance deep-blue OLEDs. Our best OLED was fabricated with mer-Ir(tfpi_tmBn) 3 as the emitter in a device with stepwise emitter doping and using a neat film of the same emitter as an electron-blocking layer (EBL). This deep-blue emitting device [CIE(0.16, 0.08)] achieved an EQE max of 14.9% and showed a moderate efficiency roll-off with an EQE 100 of 11.7%. A comparison of device performance (Figure 2) reveals that these are among the best performing deep-blue OLEDs.

Synthesis and Compound Characterization
The synthesis of four new iridium(III) complexes mer-Ir(1-(4-trifluoromethylphenyl)-3-(benzyl)imidazole) 3 [mer-Ir(tfpi_ Bn) 3 ], fac-Ir(1-(4-trifluoromethylphenyl)-3-(benzyl)imidazole) 3 Adv. Optical Mater. 2023, 11, 2201495   (Figure 3). The imidazolium NHC precursors were synthesized in excellent yield by N-alkylation of the imidazole in refluxing acetonitrile [19] with benzyl bromide (L_Bn) or 2,4,6-trimethylbenzylbromide [20] (L_tmBn). The cyclometallation of the iridium proceeded by reaction of the imidazolium salt with [Ir(COD)Cl] 2 in the presence of silver(I) oxide and triethylamine in refluxing 2-ethoxyethanol. This is a simplification of our previously reported procedure that involved the preparation of a silver NHC complex prior to addition of [Ir(COD)Cl] 2 in a 2-step, 1-pot reaction. [3b] The iridium complexes were isolated as a mixture of mer and fac isomers (in a 3.5:1 ratio for the benzyl compounds and a 5.5:1 ratio for the trimethylbenzyl compounds). The isomers could be separated by careful, repeated column chromatography on silica. The mer isomers were isolated in high yield (71 and 58%, respectively for mer-Ir(tfpi_Bn) 3 and mer-Ir(tfpi_tmBn) 3 ). In contrast to our previous work, the fac isomers could be isolated from these reactions. fac-Ir(tfpi_Bn) 3 was obtained in 17% yield; however, it proved exceptionally difficult to purify fac-Ir(tfpi_tmBn) 3 and only a small amount of material, corresponding to a 2% yield, was isolated, which limited its subsequent characterization. Attempts to promote acid-catalyzed mer-to-fac isomerization of mer-Ir(tfpi_tmBn) 3 according to established procedures [7g,21] were not successful.
The identity and purity of each of the complexes were confirmed by a combination of NMR spectroscopy, mass spectro metry, melting point determination, and X-ray crystallography. Coordination of the ligand to the iridium center was confirmed by 1 H NMR spectroscopy, which showed the disappearance of the downfield imidazolium resonance, loss of one resonance from the Ph-CF 3 and an upfield shift in the vinylic resonances. Upon coordination to the iridium the symmetry of the ligand decreased, exemplified by the splitting of the signal from the two benzylic protons on each ligand (Figure 4). The mer/fac isomers were readily identified by NMR spectroscopy, with the 1 H and 19 F NMR spectra of the C 3 -symmetric fac isomer revealing a single ligand environment, while the 1 H and 19 F NMR spectra of the mer isomer showed three different ligands, this is exemplified in the 19 F NMR spectra where the fac isomers show a single signal for the CF 3 groups while the mer isomers show three signals ( Figure 4).
The thermal properties of mer-Ir(tfpi_Bn) 3 , fac-Ir(tfpi_Bn) 3 and mer-Ir(tfpi_tmBn) 3 were studied by TGA and DSC (see Figure S27 and S28, Supporting Information). All compounds are thermally stable, with 5% mass loss occurring above 380 °C for all complexes. The complex mer-Ir(tfpi_Bn) 3 has a low melting point of 123-124 °C, the DSC trace for this complex showed a small endothermic process at 87 °C followed by a broad endothermic process centered at 130 °C corresponding to the melting of the sample. The DSC trace for the other complexes showed no heat flow until an endothermic melting process at 313 °C for fac-Ir(tfpi_Bn) 3 and 364 °C for mer-Ir(tfpi_tmBn) 3. Heat flow beyond 380 °C results from the decomposition of the material, as observed by the mass loss in the TGA.
Crystals suitable for single-crystal X-ray diffraction analysis were grown for all compounds by diffusion of n-pentane or n-hexane into a dichloromethane or chloroform solution of the compound. The structures of the four iridium complexes are shown in Figure 5, while the structures of the ligands are shown in Figure S19 (Supporting Information).
All four iridium complexes exhibit a distorted octahedral environment about the iridium center. The bond lengths and angles are consistent with previously reported mer/fac-Ir(C^C) 3 complexes ( Table S2, Supporting Information). [3b,5,7e,g,22] The IrC Ar bond lengths lie within the range 2.068(2) to 2.121 (14) Å and the IrC NHC bond lengths lie within the range 2.022(2) to 2.069(11) Å. There is a weak structural trans effect observed for these complexes. IrC NHC bonds that are trans to another IrC NHC bond are on average 0.015 Å shorter than those IrC NHC bonds trans to an IrC Ar bond; the IrC Ar bonds trans to an IrC NHC bond are on average 0.028 Å shorter than the IrC Ar bonds trans to another IrC Ar bond. The C Ar IrC NHC angles within the bidentate ligands lie within a range of 77.89(9) to 78.7(6)° and the imidazole and aryl ring are almost co-planar, suggesting there is little structural distortion of the ligands within the complexes.

Theoretical Modeling
The electronic properties of the four iridium complexes along with the previously reported mer-Ir(tfpmi) 3 [3b] and its facial isomer fac-Ir(tfpmi) 3 were investigated using density functional theory (DFT) and time-dependent DFT (TD-DFT) calculations. All calculations were performed using the B3LYP [23] functional and the 6-31G(d,p) [24] basis set for non-metal atoms and the SBKJC VDZ ECP [25] basis set for the Ir atom, using the conductor-like polarizable continuum model (CPCM) for acetonitrile solution. [26] This level of theory has previously been shown to provide accurate predictions of energy levels for iridium complexes.
[3b] The optimized ground state structures match those determined crystallographically (Table S2, Supporting Information). All six complexes show remarkably similar electronic density distribution, where the addition of Bn and tmBn groups in mer-Ir(tfpi_Bn) 3 and mer-Ir(tfpi_tmBn) 3 have minimal impact on the calculated properties of these complexes compared to those of mer-Ir(tfpmi) 3 . The calculated frontier orbitals and excited states of the fac isomers are very similar to those of the mer isomers (Figure 6 and Table 1  and Ir(pmpz) 3 ] in which the mer isomer is calculated to have a destabilized HOMO, stabilized LUMO and lower excited state energies in comparison to the fac isomer. [6,7c,g] The HOMO is distributed across the iridium center and the C^C ligands, while the LUMO is localized only on the C^C ligands. The bulky Bn and tmBn substituents are not involved in the frontier MOs, the energy of the orbitals and excited states are similar to those of the previously reported mer/fac-Ir(tfpmi) 3 . [3b] The T 1 state is of a predominantly ligand-centered character mixed with some metal-to-ligand charge transfer character (LC/MLCT). A large degree of MLCT character in the excited states would be revealed in the reduced singlet-triplet energy gap (ΔE ST ) between the states. The large ΔE ST of 0.45 to 0.48 eV for these complexes supports the assignment of significant LC character to the T 1 excited state. The singlet and triplet state mixing is thus mostly the result of the high level of SOC due to the heavy metal in these iridium complexes. [10d] Optimization of the triplet state geometry of the complexes revealed only minimal structural changes compared to their ground state structures, with only a notable contraction in the IrC Ar distance for one of the ligands by between 0.040 and 0.077 Å. Visualization of the triplet spin density distribution ( Figure S20, Supporting Information) reveals that it is concentrated on the same C^C ligand that contains the shortened IrC Ar bond. TD-DFT calculations show that there are three nearly degenerate triplet energy levels, with T 1 and T 3 separated by between 27 and 74 meV. The localization of T 1 on a single ligand and the three very close triplet states support the assignment of a strong LC character to the excited states, with each of T 1 to T 3 associated with a LC state on each of the three C^C ligands.
To investigate the nature of the structured emission (vide infra), simulation of the PL was performed for mer-Ir(tfpmi) 3 Adv. Optical Mater. 2023, 11, 2201495  and fac-Ir(tfpmi) 3 as these two complexes were identified to act as computationally accessible models of the bulky complexes synthesized in this study. Within the framework of the Frank-Condon principle, the calculation of the overlap between different vibrational modes in the ground and excited state allows the calculation of the vibrationally coupled emission spectra. [27] This framework has now been expanded to calculate spin-forbidden phosphorescent emission, [28] and has been used to study the emission from a growing number of phosphorescent platinum, iridium, and gold complexes. [29] There is good agreement between the simulated and experimental PL spectra (Figure 7). The highest energy (0-0) peak in the emission spectrum does not correspond to the most intense peak, highlighting the significant contribution of vibrational modes to the shape of the PL spectra in these complexes.

Electrochemical Characterization
The electrochemical properties of the complexes were studied using cyclic voltammetry (CV) and differential pulse voltammetry (DPV) in acetonitrile and the voltammograms are shown in Figure 8 and the data are tabulated in Table 2. All four complexes exhibited a reversible oxidation wave, while no reduction processes were observed within the solvent window. The oxidation process can be assigned as a mixed metalcentered and C^C ligand-centered oxidation as corroborated by the DFT calculations. The oxidation potentials (E ox from DPV) lie within a narrow range of between 0.99 and 1.03 V versus SCE. These E ox values are very close to the E ox of mer-Ir(tfpmi) 3 (E ox = 0.97 V vs SCE). [3b] The experimental HOMO values for the complexes lie between -5.41 and -5.45 eV, in good agreement with the calculated energy levels. In contrast, the experimental LUMO values (calculated using the optical energy gap, E opt , due to the absence of an observed reduction wave in the electrochemical experiments) were significantly stabilized in comparison to the theoretical values.

Photophysics
The UV-vis absorption spectra of the four complexes were recorded in acetonitrile solution (Figure 9 and Table 3). The absorption spectra of the four complexes show similar profiles, with strong absorption bands at high energy (λ abs < 230 nm, ε > 60 000 M −1 cm −1 ) attributed to ligand-centered (LC) π-π* transitions. The low energy shoulder around 300 nm (ε ≈ 15 000 M −1 cm −1 ) is assigned to a mixed LC/MLCT transition, consistent with the DFT calculations. The absorption spectra are similar to those observed for mer-Ir(tfpmi) 3 . [3b] All four iridium complexes show deep blue emission in acetonitrile solution. The emission is structured, with emission maxima, λ PL , observed at 425-427 nm for the mer isomers

Figure 7.
Comparison of simulated and experimental spectra for a) mer-Ir(tfpi_Bn) 3 and b) fac-Ir(tfpi_Bn) 3 . Simulated vibronic spectra were calculated using the VH implementation of the FC framework at the B3LYP/6-31G(d,p)(non-metal atoms)/SBKJC VDZ ECP(Ir atom) level of theory, spectra were calculated on complexes containing N-Me ligands to reduce the cost of the calculations. Two experimental spectra are plotted, one of which is offset to match the PL maximum, highlighting the similar spectral features. PL spectra were recorded in degassed acetonitrile solution at 298 K, λ exc = 320 nm).
in addition to a high energy shoulder at 415 nm. The emission spectra of the fac complexes are blue-shifted by ca. 5 nm (35 meV), with λ PL of 420-421 nm, consistent with the moderate increase in the calculated triplet energies. The mer isomers have the same λ PL and structured emission spectra as that observed for mer-Ir(tfpmi) 3 , indicating that the pendant bulky groups have little impact on the emission energy or character of the emissive T 1 state of the complexes. This small difference in emission energies between the facial and meridional isomers is in contrast to the significant (40-60 nm) red-shift of the emission of the meridional isomer compared to the facial isomer in previously reported isomeric pairs of complexes (e.g., fac/mer-Ir(pmp) 3 , [6] fac/mer-Ir(pmpz) 3 , [7g] f-timpz/m-timpz, [7i] f-tpb1/m-tpb1 [7h] and f-1tBu/m-1tBu).
The absence of solution-state solvatochromism of the emission spectra ( Figure S22, Supporting Information) is also consistent with a T 1 state of LC character. This is contrast to other recently reported deep-blue Ir(C^C) 3 emitters mer-Ir(pmp_Bn) 3 and fac-Ir(dbfmi) 3 that showed significant positive solvatochromism.
[7a,f ] The similar emission spectra in MeCN and DCM solutions, and doped films suggest that there is little solvatochromism experienced by these emitters, supporting the assignment of emission from a predominantly LC excited state.
The deep blue emission of mer-Ir(tfpi_Bn) 3, fac-Ir(tfpi_Bn) 3 and mer-Ir(tfpi_tmBn) 3 is conserved in the evaporated thin films. At a doping concentration (c D ) of 30 wt%, both meridional complexes showed a λ PL at 426 nm, accompanied by high and low energy shoulders at 404 nm and 450 nm, respectively (Figure 9g). The full width at half maximum (FWHM) of the emission spectra of both complexes is 62 nm (0.42 eV). In the neat film, the emission broadened slightly to FWHM = 70 nm (0.47 eV). The facial complex fac-Ir(tfpi_Bn) 3 showed shorter wavelength emission than the other materials (λ PL = 420 nm) in 6 wt% doped film in TSPO1, consistent with the trend observed in acetonitrile solution. In neat film, the facial isomer showed significantly broadened emission (FWHM = 113 nm), suggesting greater aggregation-induced broadening of the emission spectrum for this isomer. The meridional isomers showed high Φ PL in doped films, with Φ PL of 68% and 75%, respectively, for mer-Ir(tfpi_Bn) 3 and mer-Ir(tfpi_tmBn) 3 at c D of 6 wt%, increasing to 78% and 89% at c D of 16.5 wt%, reaching 78% and 79% at c D of 30 wt%. In neat films, the Φ PL dropped to 19% and 33%, respectively. Complex fac-Ir(tfpi_Bn) 3 showed lower Φ PL , with Φ PL of 53%, 80%, 65%, and 12% at c D of 6 wt%, 16.5 wt%, 30 wt% and in neat films, respectively. The doping concentration-dependent Φ PL of the Ir complexes is summarized in Figure 9h. The lifetimes of the three iridium complexes in 16.5 wt% evaporated films in TSPO1 (Table 4 and Figure S23, Supporting Information) ranged from 3.14 to 4.18 µs and are similar to those measured in acetonitrile.
The combination of broadened PL spectra and lower Φ PL in the neat films compared to the doped films suggests that there   6 in MeCN using a Pt disc electrode, platinum counter electrode and a Ag/Ag + reference electrode. Potentials are referenced to a standard calomel electrode (SCE) using ferrocene/ferricenium as an internal standard (Fc/Fc + = 0.38 V vs SCE). [30] [31] where E ox is reported vs Fc/Fc + ; c) calculated by E LUMO = E HOMO + E opt ; d) measured from the intersection of the normalized absorption and emission spectra; e) remeasured for this study in acetonitrile. [3b] is some aggregation-caused quenching (ACQ) of the emission in the neat films (Figure 9g,h and Figure S22, Supporting Information). The higher Φ PL and reduced spectral broadening of mer-Ir(tfpi_Bn) 3 in the neat film compared to fac-Ir(tfpi_Bn) 3 suggest the meridional isomer is less susceptible to ACQ, which makes sense as one side of a facial isomer is relatively open and therefore able to aggregate at high concentration. [33] The higher Φ PL of mer-Ir(tfpi_tmBn) 3 versus mer-Ir(tfpi_Bn) 3 and the less pronounced reduction of the Φ PL in the neat film reflects a greater effectiveness for the bulkier trimethylbenzyl group to inhibit intermolecular interactions and aggregation that contribute to the quenching of the emission. The minimal spectral changes in the PL spectra and Φ PL between 6 and 30% w/w doping in TSPO1 suggest that there is effectively no deleterious aggregation in this doping regime. The Φ PL of the two meridional isomers in the 30 wt% doped film (78 and 79%) is higher than the less bulky N-methyl analog mer-Ir(tfpmi) 3 , which showed a Φ PL of 38%. [3b] The Φ PL of the two meridional isomers are slightly lower than other recently reported complexes containing bulky N-benzyl substituents, mer-Ir(pmp_Bn) 3 and mer-Ir(CF 3 pbp) 3 , which have Φ PL of 95 and 89%, respectively, in doped (20 wt% and 50 wt%, respectively) TSPO1 films. Considering the higher Φ PL , reduced aggregation, and their greater availability, the meridional isomers were explored as emitters in OLEDs.

OLEDs
Exciton confinement and bipolar carrier transport should be carefully controlled for achieving high-performance deep blue OLEDs. The basic device structure used was ITO/HATCN    [32] ; d) measured in degassed MeCN at 298 K using TCSPC, λ exc = 379 nm; e) photophysics in DCM solution from ref.
[3b]; f) the radiative (k r ) and nonradiative (k nr ) rate constants were calculated as k r = Φ PL /τ PL and k nr = (1−Φ PL )/τ PL . , and LiF are the hole injection, hole transporting, electron blocking, hole blocking, electron transporting, and electron injection layers, respectively. Two doping strategies were explored and compared: stepwise doping and uniform doping. For the uniform doping, a doping concentration of 16.5 wt% in a 24 nm thick EML was chosen due to the optimized Φ PL as shown in Figure 9h. As to the stepwise doping, a gradual 30, 20, 10 to 6 wt% doping profile with each sublayer of 6 nm was deployed, which was used to guarantee the same dopant content in both devices to permit a reasonable comparison. The device optimization consisted of a crosscomparison of two wide bandgap phosphine oxide hosts, DPEPO (E T = 2.99 eV) and TSPO1 (E T = 3.36 eV), [12] three different electron/exciton blocking layers, mCP, CzSi, and self-blocking with the iridium emitter, and two different doping strategies (i.e., uniform doping and stepwise doping) to improve the bipolar carrier transport. Firstly, using the device structure with theuniformly doped EML, i.e., ITO/HAT-CN (5 nm)/NPB (45 nm)/TCTA (10 nm)/ mCP (10 nm)/mer-Ir(tfpi_tmBn) 3 (5 nm)/EML (24 nm)/TSPO1 (6 nm)/TmPyPB (20 nm)/LiF (0.5 nm)/Al (100 nm), as shown in Figure S24 (Supporting Information), the device performance as a function of host choice was compared. Although both TSPO1 and DPEPO could well confine the triplet exciton on emitters within the EML, resulting in similar EQE max , the device with TSPO1 showed higher maximum luminance (L max ) and lower efficiency roll-off than that with DPEPO, which we attributed to the better carrier transport ability of TSPO1. To improve the device performance further by suppressing the exciton/electron leakage into the HTL, a blocking layer with high triplet energy level and shallow LUMO level was needed.
Based on the device structure shown in the inset of Figure 10b, with a 10 nm thick single-layer EBL structure, devices with mCP showed higher current density and luminance than that with CzSi at the same voltage (Figure 10a). The EQE max of devices with mCP reached over 10%, which was twice that with CzSi. However, the efficiency roll-off was significant with mCP EBL-based devices. The poor device performance of the single EBL devices was attributed to either triplet exciton leakage into the EBL or carrier imbalance induced by the low hole mobility of CzSi. In addition, both mCP and CzSi have LUMO energy level of around 2.4 eV, which is not sufficient to block electrons transported by either mer-Ir(tfpi_tmBn) 3 or TSPO1 whose LUMO levels are 2.11 eV and 2.50 eV, respectively. To address these issues, a double-layer EBL structure was investigated, including a 5 nm thick mCP or CzSi and a 5 nm thick selfblocking layer consisting of mer-Ir(tfpi_tmBn) 3 . Consequently, the device performance was greatly improved as shown in Figure 10b. Not only did EQE max increase to 14.2% and 7.3% for mCP/mer-Ir(tfpi_tmBn) 3 EBL and CzSi/mer-Ir(tfpi_tmBn) 3 EBL devices, respectively, but their efficiency roll-off were reduced as well with EQE at 100 cd m −2 (EQE 100 ) of 11.0% and 6.5%, respectively. We contend that the self-blocking strategy effectively reduced the triplet exciton diffusion and electron leakage into the HTL, especially in the case when an efficient carrier blocking layer with a sufficiently high triplet energy level is unavailable.
Finally, the device performance was improved further by optimizing the doping profile to a stepwise doping strategy, which was expected to lead to more balanced bipolar carrier transport and wider light emission zone. [34] To investigate the bipolar carrier transport, single carrier devices were fabricated firstly, in which c D was varied from 0, 10, 20, to 30 wt%. The    Figure S25 (Supporting Information).
The neat TSPO1 single carrier devices showed dominant electron conduction. [35] With increasing c D from 10 wt% to 30 wt%, there was a monotonic increase of the hole current in the HODs, which was attributed to the trapping state-assisted hole transport of dopants because of their shallower HOMO level (5.4 eV) than that of TSPO1 (6.7 eV). [36] By contrast, a decrease  of the electron current was observed with c D increasing from 0 to 20 wt% in the EODs, implying a large carrier scattering effect [36b,37] of dopants under the electron transport condition, which is consistent with their energy level alignment (i.e., a shallower LUMO of 2.09-2.11 eV for the emitters than that of TSPO1, 2.5 eV). With a further increase of the c D to 30 wt%, an increase of the electron current was observed. Considering the dependence of both bipolar carrier conduction and Φ PL (Figure 9h) on c D , a stepwise doping profile of EML was designed, which comprised four sub-layers of the same thickness of 6 nm. The doping concentration of the sub-EML was highest at 30 wt% near the HTL, which then decreased gradually with distance to 20 wt%, 10 wt%, and finally 6 wt%. The total EML thickness and averaged doping concentration across the EML were the same in both the uniform doping and stepwise doping devices, the latter shown in the inset of Figure 11b.
With both self-blocking and stepwise doping strategies, the mer-Ir(tfpi_tmBn) 3 device showed improved performance compared with the uniformly doped device, with EQE max of 14.9%, implying nearly 100% exciton utilization efficiency. The EQE at 100 cd m −2 and 500 cd m −2 were 11.7%, and 7.8%, respectively. In addition, PE max was increased to 6.3 lm W −1 , double that reported for the device with mer-Ir(tfpmi) 3 which employed a uniform doping strategy. [3b] The device showed deep blue emission with λ EL at 424 nm, FWHM of 66 nm, and CIE coordinates of (0.16, 0.08). The device with the mer-Ir(tfpi_Bn) 3 emitter showed a slightly lower device performance with EQE max of 12.2%, PE max of 5.1 lm W −1 , at the same CIE coordinates of (0.16, 0.08). The cross-comparison of the performance of the mer-Ir(tfpi_Bn) 3 based device with that of the mer-Ir(tfpi_tmBn) 3 device is shown in Table 5 and Figure S26 (Supporting Information). The OLEDs with mer-Ir(tfpi_Bn) 3 and mer-Ir(tfpi_tmBn) 3 both satisfy the National Television System Committee (NTSC) standard for the primary blue (CIE y of 0.08). Our devices show among the best performance of deep blue OLEDs (CIE y < 0.1) with Ir(C^C) 3 complexes as emitters (cf. Figure 2). The severe efficiency roll-off behavior beyond 500 cd m −2 is commonly observed in deep blue OLEDs as evidenced in the studies documented in Figure 2. Besides the well-studied intrinsic exciton loss channels such as triplet-triplet annihilation, and tripletpolaron quenching, which are mainly governed by the exciton lifetime and exciton density, our results highlight three important extrinsic factors in the device architecture. In other words, the device performance under high current density is not only emitter property limited, but more dependent on: 1) the host material as shown in Figure S24 (Supporting Information). By replacing DPEPO host with TSPO1, the maximum luminance was increased from less than 300 cd m −2 to over 600 cd m −2 , which we attribute to the more balanced carrier transport and wider exciton recombination zone; 2) The blocking layer as shown in Figure 10. The significant improvement in efficiency roll-off after inserting a self-blocking layer is due to the suppressed electron and exciton leaking into the HTL; 3) Doping profile of the EML as shown in Figure 11. The step-doping is effective to expand the exciton distribution and improve the power efficiency. This aggregate evidence illustrates how device performance can be improved further with a combination of improved functional materials and doping strategies, both of which are current objects of research under investigation.

Conclusions
A family of four new homoleptic iridium complexes containing bulky benzyl and trimethylbenzyl groups on the imidazole group of the NHC ligand have been prepared and fully characterized. The complexes exhibit deep blue emission with high photoluminescence quantum yields in both solution and TSPO1 as a host matrix. Two of the complexes, mer-Ir(tfpi_Bn) 3 and mer-Ir(tfpi_tmBn) 3 were used to fabricate deep blue OLEDs. The device structure was thoroughly investigated by comparing different host materials, adopting step-doping EML, depositing a self-blocking layer, and optimizing carrier transport layer thickness. With the optimized device structure, deep blue emission with λ EL of 429 nm and CIE coordinates of (0.16, 0.08) was observed, which is close to the BT.2020 standard. The EQE max (EQE 100 ) is 14.9% (11.7%), and 12.2% (11%) for the mer-Ir(tfpi_tmBn) 3 and mer-Ir(tfpi_Bn) 3 OLEDs, respectively. These results underscore the value of incorporating bulky N-Bn or N-tmBn groups to the ligands in deep-blue emitting iridium NHC complexes.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.