SimCP3—An Advanced Homologue of SimCP2 as a Solution-Processed Small Molecular Host Material for Blue Phosphorescence Organic Light-Emitting Diodes

We have overcome the synthetic difficulty of 9,9′,9′′,9′′′,9′′′′,9′′′′′-((phenylsilanetriyl)tris(benzene-5,3,1-triyl))hexakis(9H-carbazole) (SimCP3) an advanced homologue of previously known SimCP2 as a solution-processed, high triplet gap energy host material for a blue phosphorescence dopant. A series of organic light-emitting diodes based on blue phosphorescence dopant iridium (III) bis(4,6-difluorophenylpyridinato)picolate, FIrpic, were fabricated and tested to demonstrate the validity of solution-processed SimCP3 in the device fabrication.


Results and Discussion
In our earlier reports [16,24,25], both 9,9′-(5-(triphenylsilyl)-1,3-phenylene)bis(9H-carbazole) (SimCP) and SimCP2 were synthesized in a similar method (Scheme 2). Either chlorotripheylsilane or dichlorodiphenylsilane was reacted with monolithiated 1,3,5-tribromobenzene to furnish 1,3dibromophenyl substituted thriphenylsilane and bis-1,3-dibromophenyl substituted diphenylsilane, respectively. The silane compounds were then brought to a palladium-catalyzed amination reaction with excess amount of carbazole. Both SimCP and SimCP2 were obtained with a reasonable isolation yield of 54% and ~40%, respectively, in the final amination reaction. It is conceivable that the reaction yields will be lower than that of SimCP2 if the higher homologue SimCP3 is prepared by a same reaction sequence shown in Scheme 2. Due to the much more partial amination products formed in the reaction of synthesizing SimCP3, we obtained a series of product mixtures, which were hard to separate by column chromatography. In addition, the purification of SimCP3 required a tedious process in order to remove the excess amount of carbazole used in the reaction. Alternatively, we have been able to synthesize SimCP by using mono-lithiated 9,9′-(5-bromo-1,3-phenylene)bis(9H-carbazole) (BrmCP) to react with chlorotripheylsilane. However, the same synthetic approach has been problematic in the preparation of SimCP2 and SimCP3, even though a very recent literature has reported the same synthetic approach in the preparation of SimCP2 and SimCP3 [23]. We also tried the approach of Grignard reaction (Mg instead of n-butyllithium), but it did not work out well either. It seems to us that either lithiated or magnesiated BrmCP is relatively unstable under the reaction condition. In the case of SimCP2 and SimCP3, the chance of either lithiated or magnesiated BrmCP reacting more than Scheme 1. Chemical structure of SimCP2 and SimCP3. MM2 energy minimized structures (by Chem 3D) are shown on the right, respectively.

Results and Discussion
In our earlier reports [16,24,25], both 9,9′-(5-(triphenylsilyl)-1,3-phenylene)bis(9H-carbazole) (SimCP) and SimCP2 were synthesized in a similar method (Scheme 2). Either chlorotripheylsilane or dichlorodiphenylsilane was reacted with monolithiated 1,3,5-tribromobenzene to furnish 1,3dibromophenyl substituted thriphenylsilane and bis-1,3-dibromophenyl substituted diphenylsilane, respectively. The silane compounds were then brought to a palladium-catalyzed amination reaction with excess amount of carbazole. Both SimCP and SimCP2 were obtained with a reasonable isolation yield of 54% and ~40%, respectively, in the final amination reaction. It is conceivable that the reaction yields will be lower than that of SimCP2 if the higher homologue SimCP3 is prepared by a same reaction sequence shown in Scheme 2. Due to the much more partial amination products formed in the reaction of synthesizing SimCP3, we obtained a series of product mixtures, which were hard to separate by column chromatography. In addition, the purification of SimCP3 required a tedious process in order to remove the excess amount of carbazole used in the reaction. Alternatively, we have been able to synthesize SimCP by using mono-lithiated 9,9′-(5-bromo-1,3-phenylene)bis(9H-carbazole) (BrmCP) to react with chlorotripheylsilane. However, the same synthetic approach has been problematic in the preparation of SimCP2 and SimCP3, even though a very recent literature has reported the same synthetic approach in the preparation of SimCP2 and SimCP3 [23]. We also tried the approach of Grignard reaction (Mg instead of n-butyllithium), but it did not work out well either. It seems to us that either lithiated or magnesiated BrmCP is relatively unstable under the reaction condition. In the case of SimCP2 and SimCP3, the chance of either lithiated or magnesiated BrmCP reacting more than Due to the much more partial amination products formed in the reaction of synthesizing SimCP3, we obtained a series of product mixtures, which were hard to separate by column chromatography. In addition, the purification of SimCP3 required a tedious process in order to remove the excess amount of carbazole used in the reaction. Alternatively, we have been able to synthesize SimCP by using mono-lithiated 9,9 -(5-bromo-1,3-phenylene)bis(9H-carbazole) (BrmCP) to react with chlorotripheylsilane. However, the same synthetic approach has been problematic in the preparation of SimCP2 and SimCP3, even though a very recent literature has reported the same synthetic approach in the preparation of SimCP2 and SimCP3 [23]. We also tried the approach of Grignard reaction (Mg instead of n-butyllithium), but it did not work out well either. It seems to us that either lithiated or magnesiated BrmCP is relatively unstable under the reaction condition. In the case of SimCP2 and SimCP3, the chance of either lithiated or magnesiated BrmCP reacting more than once with dichlorophenylsilane or trichlorophenylsilane is rather low. In addition, the reaction is quite sensitive to the purity of dichlorophenylsilane or trichlorophenylsilane, which is prone to hydrolysis and generates hydrochloric acid that is detrimental to the reaction. Finally, we were able to synthesize and isolate pure SimCP3 in multi-gram scale via reaction with sodium metal, i.e., a Würtz-Fittig-type coupling reaction between chlorosilane and aryl halide (Scheme 3) [26,27]. Instead of a carbon anion of BrmCP, a more feasible silyl anion of trichlorophenylsilne is involved in the coupling reaction. once with dichlorophenylsilane or trichlorophenylsilane is rather low. In addition, the reaction is quite sensitive to the purity of dichlorophenylsilane or trichlorophenylsilane, which is prone to hydrolysis and generates hydrochloric acid that is detrimental to the reaction. Finally, we were able to synthesize and isolate pure SimCP3 in multi-gram scale via reaction with sodium metal, i.e., a Würtz-Fittig-type coupling reaction between chlorosilane and aryl halide (Scheme 3) [26,27]. Instead of a carbon anion of BrmCP, a more feasible silyl anion of trichlorophenylsilne is involved in the coupling reaction. To facilitate the SimCP3 preparation, we improved the synthesis of BrmCP with a significantly better isolation yield of >90% [12,28]. Regarding the Würtz-Fittig-type coupling reaction, SimCP3 was furnished in only modest yields around 30% due to the competition reaction of the potential silicon-silicon bond formation of oligosilane or polysilane [29,30]. However, the desired product was much easier to isolate from the product mixture. In this reaction, we also observed silanol side products, although they were readily removed by column chromatography. In addition, there was no excess amount of carbazole involved in the reaction; hence, no extra purification process was required for SimCP3. These are the main advantages of the Würtz-Fittig reaction over other synthetic methods of preparing SimCP3. Several spectroscopic and physical methods were used to characterize the homogeneity of thin film and the necessary optoelectronic properties of SimCP3 as a solution-processable host material for FIrpic OLEDs.
First, as shown in Figure 1, SimCP3 is very thermally stable with a thermal decomposition temperature (Td) as high as 620 °C, where 5% weight loss was determined by thermogravimetric analysis (TGA) under a nitrogen atmosphere. Second, SimCP3 exhibits amorphous or semi-amorphous characteristic based on the thermograms of differential scanning calorimetry (DSC). As shown in Figure 1, SimCP3 has melting temperatures (Tm), crystallization temperatures (Tc), and glass transition temperatures (Tg) of 330, 223, and 174 °C, respectively. Two very weak exothermic signals around 290 and 330 °C may be the inhibited crystallization process of thermally annealed SimCP3. The Tm and Tc of SimCP3 disappeared, and Tg was observed after the first heating scan, indicative of the amorphous tendency of SimCP3. To facilitate the SimCP3 preparation, we improved the synthesis of BrmCP with a significantly better isolation yield of >90% [12,28]. Regarding the Würtz-Fittig-type coupling reaction, SimCP3 was furnished in only modest yields around 30% due to the competition reaction of the potential silicon-silicon bond formation of oligosilane or polysilane [29,30]. However, the desired product was much easier to isolate from the product mixture. In this reaction, we also observed silanol side products, although they were readily removed by column chromatography. In addition, there was no excess amount of carbazole involved in the reaction; hence, no extra purification process was required for SimCP3. These are the main advantages of the Würtz-Fittig reaction over other synthetic methods of preparing SimCP3. Several spectroscopic and physical methods were used to characterize the homogeneity of thin film and the necessary optoelectronic properties of SimCP3 as a solution-processable host material for FIrpic OLEDs.
First, as shown in Figure 1, SimCP3 is very thermally stable with a thermal decomposition temperature (T d ) as high as 620 • C, where 5% weight loss was determined by thermogravimetric analysis (TGA) under a nitrogen atmosphere. Second, SimCP3 exhibits amorphous or semi-amorphous characteristic based on the thermograms of differential scanning calorimetry (DSC). As shown in Figure 1, SimCP3 has melting temperatures (T m ), crystallization temperatures (T c ), and glass transition temperatures (T g ) of 330, 223, and 174 • C, respectively. Two very weak exothermic signals around 290 and 330 • C may be the inhibited crystallization process of thermally annealed SimCP3. The T m and T c of SimCP3 disappeared, and T g was observed after the first heating scan, indicative of the amorphous tendency of SimCP3. once with dichlorophenylsilane or trichlorophenylsilane is rather low. In addition, the reaction is quite sensitive to the purity of dichlorophenylsilane or trichlorophenylsilane, which is prone to hydrolysis and generates hydrochloric acid that is detrimental to the reaction. Finally, we were able to synthesize and isolate pure SimCP3 in multi-gram scale via reaction with sodium metal, i.e., a Würtz-Fittig-type coupling reaction between chlorosilane and aryl halide (Scheme 3) [26,27]. Instead of a carbon anion of BrmCP, a more feasible silyl anion of trichlorophenylsilne is involved in the coupling reaction. To facilitate the SimCP3 preparation, we improved the synthesis of BrmCP with a significantly better isolation yield of >90% [12,28]. Regarding the Würtz-Fittig-type coupling reaction, SimCP3 was furnished in only modest yields around 30% due to the competition reaction of the potential silicon-silicon bond formation of oligosilane or polysilane [29,30]. However, the desired product was much easier to isolate from the product mixture. In this reaction, we also observed silanol side products, although they were readily removed by column chromatography. In addition, there was no excess amount of carbazole involved in the reaction; hence, no extra purification process was required for SimCP3. These are the main advantages of the Würtz-Fittig reaction over other synthetic methods of preparing SimCP3. Several spectroscopic and physical methods were used to characterize the homogeneity of thin film and the necessary optoelectronic properties of SimCP3 as a solution-processable host material for FIrpic OLEDs.
First, as shown in Figure 1, SimCP3 is very thermally stable with a thermal decomposition temperature (Td) as high as 620 °C, where 5% weight loss was determined by thermogravimetric analysis (TGA) under a nitrogen atmosphere. Second, SimCP3 exhibits amorphous or semi-amorphous characteristic based on the thermograms of differential scanning calorimetry (DSC). As shown in Figure 1, SimCP3 has melting temperatures (Tm), crystallization temperatures (Tc), and glass transition temperatures (Tg) of 330, 223, and 174 °C, respectively. Two very weak exothermic signals around 290 and 330 °C may be the inhibited crystallization process of thermally annealed SimCP3. The Tm and Tc of SimCP3 disappeared, and Tg was observed after the first heating scan, indicative of the amorphous tendency of SimCP3.  Similarly, amorphous or semi-amorphous characteristics are evident from the X-ray diffraction (XRD) spectrum (Figure 2a) of the solution-casting thin film of SimCP3. The XRD spectrum is basically a featureless halo, indicative of an amorphous nature. Macroscopically, the homogeneity of the solution spin-coated SimCP3 thin film is evident from the photoluminescence (PL) images of the neat film and FIrpic-doped SimiCP3 thin films (Figure 2b  The most important property of the host material for a blue phosphorescence dopant such as FIrpic is ET. We estimated the ET of SimCP3 as a neat thin film with time-gated acquisition PL spectra at low temperatures (Figure 4a,b). From spectra with either variable delayed time or variable low temperature, it can be confirmed that the emission band with a λmax around 400 nm is the delayed fluorescence and a λmax around 525 nm is the phosphorescence. By taking the onset wavelength of the high energy edge of phosphorescence bands, we obtain ~461-468 nm (2.69-2.65 eV) as the optical energy gap for ET. From our measurements, the ET of SimCP3 is sufficiently larger than that of FIrpic, ensuring triplet state exciton on the phosphorescence dopant rather than its host SimCP3. The UVvisible absorption and fluorescence spectra of SimCP3 in solution are illustrated in Figure 4c.  The most important property of the host material for a blue phosphorescence dopant such as FIrpic is ET. We estimated the ET of SimCP3 as a neat thin film with time-gated acquisition PL spectra at low temperatures (Figure 4a,b). From spectra with either variable delayed time or variable low temperature, it can be confirmed that the emission band with a λmax around 400 nm is the delayed fluorescence and a λmax around 525 nm is the phosphorescence. By taking the onset wavelength of the high energy edge of phosphorescence bands, we obtain ~461-468 nm (2.69-2.65 eV) as the optical energy gap for ET. From our measurements, the ET of SimCP3 is sufficiently larger than that of FIrpic, ensuring triplet state exciton on the phosphorescence dopant rather than its host SimCP3. The UVvisible absorption and fluorescence spectra of SimCP3 in solution are illustrated in Figure 4c. The most important property of the host material for a blue phosphorescence dopant such as FIrpic is E T . We estimated the E T of SimCP3 as a neat thin film with time-gated acquisition PL spectra at low temperatures (Figure 4a,b). From spectra with either variable delayed time or variable low temperature, it can be confirmed that the emission band with a λ max around 400 nm is the delayed fluorescence and a λ max around 525 nm is the phosphorescence. By taking the onset wavelength of the high energy edge of phosphorescence bands, we obtain~461-468 nm (2.69-2.65 eV) as the optical energy gap for E T . From our measurements, the E T of SimCP3 is sufficiently larger than that of FIrpic, ensuring triplet state exciton on the phosphorescence dopant rather than its host SimCP3. The UV-visible absorption and fluorescence spectra of SimCP3 in solution are illustrated in Figure 4c. Knowing an optical energy gap of 3.57 eV (347 nm shown in Figure 4c) and the HOMO energy level of 5.95 eV (Figure 5a), which was acquired by a low-energy photoelectron spectrometer Riken-Keiki AC-2, the LUMO energy level of SimCP3 can thus be calculated as ~2.4 eV. Together with the HOMO and LUMO energy levels and ET of TAPC (hole transporting or electron blocking material) [31,32], FIrpic [32], and TmPyPB (electron transporting or hole blocking material) [33], the energy level alignment of materials involved in OLEDs is depicted in Figure 5b. Therefore, TAPC, SimCP3, and TmPyPB all have HOMO and LUMO energy levels that encompass those of FIrpic, and FIrpic has the smallest ET among all materials involved in the OLED fabrication. This is a good set of materials enabling the confinement of triplet exciton on the FIrpic phosphorescence dopant. Accordingly, we fabricated three types of FIrpic OLED having SimCP3 as a solution-processed host material: a singlelayer device (SL)-ITO/PEDOT:PSS/SimCP3:FIrpic/CsF/Al-a bilayer device (BL)-ITO/PEDOT:PSS/ SimCP3:FIrpic/TmPyBP/CsF/Al-and a trilayer device (TL)-ITO/PEDOT:PSS/TAPC/SimCP3:FIrpic/ TmPyBP/CsF/Al. After optimization, the dopant concentration of FIrpic was 4 wt % in all three devices. Except for TmPyBP and CsF/Al cathode, each thin film layer of such FIrpic OLEDs was fabricated by solution process, i.e., the spin-coating method.  Knowing an optical energy gap of 3.57 eV (347 nm shown in Figure 4c) and the HOMO energy level of 5.95 eV (Figure 5a), which was acquired by a low-energy photoelectron spectrometer Riken-Keiki AC-2, the LUMO energy level of SimCP3 can thus be calculated as~2.4 eV. Together with the HOMO and LUMO energy levels and E T of TAPC (hole transporting or electron blocking material) [31,32], FIrpic [32], and TmPyPB (electron transporting or hole blocking material) [33], the energy level alignment of materials involved in OLEDs is depicted in Figure 5b. Therefore, TAPC, SimCP3, and TmPyPB all have HOMO and LUMO energy levels that encompass those of FIrpic, and FIrpic has the smallest E T among all materials involved in the OLED fabrication. This is a good set of materials enabling the confinement of triplet exciton on the FIrpic phosphorescence dopant. Accordingly, we fabricated three types of FIrpic OLED having SimCP3 as a solution-processed host material: a single-layer device (SL)-ITO/PEDOT:PSS/SimCP3:FIrpic/CsF/Al-a bilayer device (BL)-ITO/PEDOT:PSS/SimCP3:FIrpic/TmPyBP/CsF/Al-and a trilayer device (TL)-ITO/ PEDOT:PSS/TAPC/SimCP3:FIrpic/TmPyBP/CsF/Al. After optimization, the dopant concentration of FIrpic was 4 wt % in all three devices. Except for TmPyBP and CsF/Al cathode, each thin film layer of such FIrpic OLEDs was fabricated by solution process, i.e., the spin-coating method. Knowing an optical energy gap of 3.57 eV (347 nm shown in Figure 4c) and the HOMO energy level of 5.95 eV (Figure 5a), which was acquired by a low-energy photoelectron spectrometer Riken-Keiki AC-2, the LUMO energy level of SimCP3 can thus be calculated as ~2.4 eV. Together with the HOMO and LUMO energy levels and ET of TAPC (hole transporting or electron blocking material) [31,32], FIrpic [32], and TmPyPB (electron transporting or hole blocking material) [33], the energy level alignment of materials involved in OLEDs is depicted in Figure 5b. Therefore, TAPC, SimCP3, and TmPyPB all have HOMO and LUMO energy levels that encompass those of FIrpic, and FIrpic has the smallest ET among all materials involved in the OLED fabrication. This is a good set of materials enabling the confinement of triplet exciton on the FIrpic phosphorescence dopant. Accordingly, we fabricated three types of FIrpic OLED having SimCP3 as a solution-processed host material: a singlelayer device (SL)-ITO/PEDOT:PSS/SimCP3:FIrpic/CsF/Al-a bilayer device (BL)-ITO/PEDOT:PSS/ SimCP3:FIrpic/TmPyBP/CsF/Al-and a trilayer device (TL)-ITO/PEDOT:PSS/TAPC/SimCP3:FIrpic/ TmPyBP/CsF/Al. After optimization, the dopant concentration of FIrpic was 4 wt % in all three devices. Except for TmPyBP and CsF/Al cathode, each thin film layer of such FIrpic OLEDs was fabricated by solution process, i.e., the spin-coating method.  EL characteristics of SL, BL, and TL devices are summarized in Figure 6 and Table 1. All three devices exhibit typical sky-blue emissions of FIrpic with a λ max of~468 nm, although a variation of emission shoulder band around 500 nm (Figure 6a) was observed. Since both BL and TL devices have a similar intensity of the shoulder band and are both more intense than that of the SL device, it can be logically inferred that TmPyPB effectively blocks the triplet exciton of FIrpic, confining the charge recombination mostly in the SimCP3:FIrpic light-emitting layer (i.e., farther away from the cathode surface) [19]. Therefore, the same reason can be ascribed to the low EL efficiency of the SL device (Figure 6b,c), where the triplet exciton of FIrpic generates in an area too close to the cathode surface and hence is prone to being quenched. Similarly, the TAPC of the TL device plays a role of inhibiting triplet exciton from contacting the anode of the device. Therefore, the EL efficiency of the TL device is promoted further when compared with the BL device. The maximum external quantum efficiency (EQE), current efficiency (CE), and power efficiency (PE) of the TL device are~9%, 20 cd/A, and 12 lm/W, respectively. However, the TL device has the lowest current density and the highest turn-on voltage-around 5 V at 1 cd/m 2 (see Figure 6d)-among the three devices. This can be attributed to the two exciton blocking layers in the device, which limits the current density and elevates the driving voltage of the device. Nevertheless, all three devices exhibited relatively mild efficiency roll-off. Even at the high brightness of 3000 cd/m 2 (about 38, 30, and 19 mA/cm 2 of the SL, BL, and TL devices), an EQE of 3.6%, 4.5%, and 7.4% were achieved by the SL, BL, and TL devices, which are 95%, 62%, and 81% of the maximum EL efficiency, respectively (Figure 6c). Compared with those of vacuum-thermal-evaporation OLEDs [2,4], the efficiency roll-off observed for the three devices is relatively modest. EL characteristics of SL, BL, and TL devices are summarized in Figure 6 and Table 1. All three devices exhibit typical sky-blue emissions of FIrpic with a λmax of ~468 nm, although a variation of emission shoulder band around 500 nm (Figure 6a) was observed. Since both BL and TL devices have a similar intensity of the shoulder band and are both more intense than that of the SL device, it can be logically inferred that TmPyPB effectively blocks the triplet exciton of FIrpic, confining the charge recombination mostly in the SimCP3:FIrpic light-emitting layer (i.e., farther away from the cathode surface) [19]. Therefore, the same reason can be ascribed to the low EL efficiency of the SL device (Figure 6b,c), where the triplet exciton of FIrpic generates in an area too close to the cathode surface and hence is prone to being quenched. Similarly, the TAPC of the TL device plays a role of inhibiting triplet exciton from contacting the anode of the device. Therefore, the EL efficiency of the TL device is promoted further when compared with the BL device. The maximum external quantum efficiency (EQE), current efficiency (CE), and power efficiency (PE) of the TL device are ~9%, ~20 cd/A, and 12 lm/W, respectively. However, the TL device has the lowest current density and the highest turn-on voltage-around 5 V at 1 cd/m 2 (see Figure 6d)-among the three devices. This can be attributed to the two exciton blocking layers in the device, which limits the current density and elevates the driving voltage of the device. Nevertheless, all three devices exhibited relatively mild efficiency roll-off. Even at the high brightness of 3000 cd/m 2 (about 38, 30, and 19 mA/cm 2 of the SL, BL, and TL devices), an EQE of 3.6%, 4.5%, and 7.4% were achieved by the SL, BL, and TL devices, which are 95%, 62%, and 81% of the maximum EL efficiency, respectively (Figure 6c). Compared with those of vacuum-thermalevaporation OLEDs [2,4], the efficiency roll-off observed for the three devices is relatively modest.     AFM measurements were carried out with XE-100 from Park Systems (Suwon, South Korea) using a non-contact mode to obtain topography images of the thin films, which were prepared by a spin-coating chlorobenzene solution of SimCP3 or SimCP3:FIrpic on a PEDOT:PSS-coated ITO-glass substrate. UV-visible absorption spectra were recorded on a Hewlett-Packard 8453 diode array spectrophotometer (Palo Alto, CA, USA). Room temperature fluorescence spectra were recorded on a Hitachi fluorescence spectrophotometer F-4500 (Tokyo, Japan). The ionization potentials (or HOMO energy levels) of SimCP3 were determined with a low energy photo-electron spectrometer (Riken-Keiki AC-2, Riken Keiki Co., Ltd., Tokyo, Japan). LUMO energy levels were estimated by subtracting the optical energy gap (E g ) from HOMO energy levels. E g was determined by the on-set absorption energy from the absorption spectra of the materials as thin film. To measure the triplet energy gap (E T ) of a thin film sample, we established a system with a temperature control of Model 350 (LakeShore Company, Westerville, OH, USA) as low as 10 K by Model 22C/350C Cryodyne Refrigerators (Janis Research Company, Woburn, MA, USA) and a tunable laser with excitation wavelengths of 213, 266, 355, 532, and 1064 nm (Brilliant B laser, Quantel Company, Newbury, Berkshire, UK), which was coupled with a time-delay controller of LP920 flash photolysis spectrometer (Edinburgh Instrument, Kirkton Campus, Livingston, England). We also acquired the triplet energy gap of SimCP3 in frozen 2-Me-THF by using Fluorolog III photoluminescence spectrometer (Kirkton Campus, Livingston, UK), which was equipped a xenon lamp as excitation light source.

Device Fabrication and Performance Measurements
The ITO substrate was purchased from Buwon Precision Sciences (Taoyuan, Taiwan with a sheet resistance around 25 Ω/sq and a thickness of 100 nm. The hole transport layer-PEDOT:PSS (CH8000)-was purchased from Sigma-Aldrich (St. Louis, MO, USA). The blue phosphorescence dopant FIrpic, the electron transport layer of TmPyPB, and the hole transport layer of TAPC were obtained commercially from Lumtec (Hsinchu, Taiwan). The indium tin oxide (ITO) substrates were pre-coated with the PEDOT:PSS hole transporting layer (HTL) and baked in the air at 150 • C for 10 min. Blends of SimCP3 and FIrpic in chlorobenzene were spin-coated on ITO/PEDOT:PSS. The active layer was then annealed at 90 • C for 30 min. After spin-coating the active single-layer, the TmPyPB (50 nm), an ultrathin CsF (2 nm) interfacial layer, and then an aluminum cathode (100 nm) were vacuum-thermal deposited. In the triple-layer device, a thin layer (~20 nm) of TAPC was spin-coated first on PEDOT:PSS. In this case, TAPC layer was thermally annealed at 90 • C for 10 min before spin-coating SimCP3:FIrpic. A surface profiler (Dektak 150, Veeco Instruments, Plainview, NY, USA) was used for calibrating the thickness of HTL and the active layer. After the deposition of the cathode, the devices were hermetically sealed with glass and UV-cured resins in a glove box (O 2 and H 2 O concentration below 0.1 ppm). The device's active area was 0.04 cm 2 and was defined by the self-made shadow mask applied in the cathode deposition. Current density and voltage characteristics were measured by a dc current-voltage source meter (Keithley 2400, Tektronix, Beaverton, OR, USA), and the device brightness (or electroluminance, cd/m 2 ) and EL spectra were monitored and recorded with a spectrophotometer (PR650; Photo Research, Syracuse, NY, USA). The EQE of OLEDs was calculated from the luminance, current density, and the EL spectra, assuming the EL of the OLED is isotropic, i.e., a Lambertian emission [34].

Conclusions
In summary, we have overcome the synthetic difficulty of SimCP3, an advanced homologue of previously known SimCP2. We have examined the amorphous nature and the spectroscopic properties (HOMO and LUMO energy levels as well as the triplet energy gap) of SimCP3. Three sky-blue phosphorescence OLEDs have been fabricated with a SimCP3:FIrpic light-emitting layer, which was spin-coated from solution. A trilayer device exhibits the highest EL efficiency of~9%,~20 cd/A, or~12 lm/W, demonstrating the viability of SimCP3 as a solution-processed host material for sky-blue phosphorescence OLEDs.