Article pubs.acs.org/JPCC
Phosphorescent Iridium(III) Complexes of Cyclometalated 5‑Aryl‑1H‑1,2,4-Triazole Ligands: Structural, Computational, Spectroscopic, and Device Studies Kerwin D. Dobbs, Jerald Feldman,* Will J. Marshall, Stephan J. McLain,† Charles D. McLaren, Jeffrey S. Meth, Giang D. Vo, and Ying Wang DuPont Central Research and Development, Experimental Station, 200 Powder Mill Road, Wilmington, Delaware 19803, United States S Supporting Information *
ABSTRACT: Ir(III) complexes of cyclometalated 5-aryl-1H-1,2,4triazole ligands are highly efficient, phosphorescent emitters. We describe herein a series of fac-IrL3 complexes, in which the nature of aryl substituents are shown to strongly affect emission wavelength over the range 453−499 nm. Computational and structural studies indicate that for aryl groups the point of attachment and dihedral angle with respect to the cyclometalated ring influence emission color. Significantly, this degree of color tuning may be achieved without resorting to electron-withdrawing or -donating groups. Photo- and electroluminescence device studies of the different emitters indicate that they are generally highly efficient: photoluminescent efficiencies >90% and external quantum efficiencies of up to 22% are observed.
1. INTRODUCTION Commercial and residential lighting account for about 20% of electricity consumption in the world.1 New lighting technologies such as light emitting diodes (LED), organic light emitting diodes (OLED), and induction fluorescent lighting improve lighting efficiency and save energy. Of these technologies, OLED solid-state lighting (SSL) is attractive to architectural designers due to its pleasing diffuse light, flexible form factors, and high efficiency.2 OLED SSL is also an environmentally safe technology because it does not employ mercury vapor found in fluorescent lighting. In addition, the lure of solution processing to achieve low manufacturing costs attracts significant research efforts.3 These advantages are motivating strong efforts to address current challenges in OLED SSL. These challenges derive from the operational requirements of high power efficiency at medium brightness (3000−5000 cd/ m2), high color rendering index (CRI >80) and high operational lifetimes (>50 000 h vs 6000−15 000 h for fluorescent lamps).4 To achieve high power efficiency, phosphorescent emitters harvest both triplet and singlet excitons, thus allowing for 100% internal quantum efficiency versus 25% for fluorescent counterparts.5,6 To obtain a high CRI, the device needs to emit light across the entire visible spectrum. This emission is accomplished by mixing red, green and blue emitters. Of these three types of emitters, phosphorescent red and green emitters are widely available on the market with excellent operational lifetimes, while blue emitters have the shortest operational lifetimes. Those that emit © 2014 American Chemical Society
sky blue color (>470 nm) are the most stable and afford WOLED devices with the longest lifetimes.7,8 It is of great interest to learn how to fine-tune the color of phosphorescent blue emitters, as this can potentially impact the CRI, color temperature, and lifetime of OLED SSL devices. In this context, phosphorescent Ir(III) complexes containing cyclometalated ligands are of particular interest, owing to their exceptionally high quantum efficiencies. It is well-known that the emission color of these emitters can be manipulated through the use of electron-withdrawing or -donating groups on the ligands.9−14 However, the introduction of functionalities such as fluorine can have unwanted side effects; for example, by adversely affecting device lifetime.15−17 Herein, we describe the synthesis, photoluminescent (PL), and electroluminescent (EL) properties of a series of phosphorescent blue emitters with high external quantum efficiencies (EQE). These emitters are homoleptic iridium(III) complexes of cyclometalated 5-aryl-1H-1,2,4-triazoles and are closely related to deep-blue emitting materials first reported by Burn, Samuel and co-workers.18 Burn and Powell also demonstrated that fluorination of the aryl ring resulted in even deeper blue emission, albeit with reduced efficiency.18,19 In this work, we show how the nature of aryl substituents attached to the cyclometalated aryl ring strongly affects Received: September 23, 2014 Revised: November 3, 2014 Published: November 21, 2014 27763
dx.doi.org/10.1021/jp5096322 | J. Phys. Chem. C 2014, 118, 27763−27771
The Journal of Physical Chemistry C
Article
prepared solution of quinine sulfate, for which the PLQE was assumed to be 0.57 (57%). The lifetime of the triplet state was measured using a Horiba TemPro 01 time-correlated, single photon counting (TCSPC) unit, with excitation set at 343 nm. The phosphorescent lifetime was measured at the highest energy peak in the emission spectrum. OLED devices were fabricated by a combination of solution processing and thermal evaporation techniques. Patterned indium tin oxide (ITO) coated glass substrates from Thin Film Devices, Inc. were used. These ITO substrates are based on Corning 1737 glass coated with ITO, having a sheet resistance of 30 ohms/square and 80% light transmission. The patterned ITO substrates were cleaned ultrasonically in aqueous detergent solution and rinsed with distilled water. The patterned ITO was subsequently cleaned ultrasonically in acetone, rinsed with isopropanol, and dried in a stream of nitrogen. Immediately before device fabrication, the cleaned, patterned ITO substrates were treated with UV ozone for 10 min. Immediately after cooling, an aqueous dispersion of hole injection material, HIJ-1, was spin-coated over the ITO surface and heated to remove solvent. After cooling, the substrates were then spin-coated with a hole transport solution, and then heated to remove the solvent. The substrates were masked and placed in a vacuum chamber. The emissive layer, the electron transport layer, and the antiquenching layer were deposited by thermal evaporation, followed by a layer of CsF. Masks were then changed in vacuo, and a layer of Al was deposited by thermal evaporation. The chamber was vented, and the devices were encapsulated using a glass lid, desiccant, and UV curable epoxy. The OLED samples were characterized by measuring their (1) current−voltage (I−V) curves, (2) electroluminescence luminance versus voltage, and (3) electroluminescence spectra versus voltage. All three measurements were performed simultaneously under computer control. The current efficiency of the device at a certain voltage is determined by dividing the electroluminescence luminance of the LED by the current density needed to run the device, with units of cd/A. The color coordinates were determined using a Minolta CS-100 m. The electroluminescence spectra were measured using a calibrated Ocean Optics spectrometer. The external quantum efficiency (EQE) was calculated from the cd/A efficiency and the electroluminescence spectrum, assuming a Lambertian distribution of output light.
emission wavelength over the range 453−499 nm. Computational and structural studies indicate that for aryl groups the point of attachment and dihedral angle with respect to the cyclometalated ring influence emission color. Moreover, this degree of color tuning is accomplished without electronwithdrawing or -donating groups. The steric and electronic effects of substituents on the emission color and quantum efficiency are elucidated via density functional theory, X-ray crystal structure analysis, PL, and EL device studies.
2. COMPUTATIONAL DETAILS All calculations were performed with the density functional theory (DFT) methods within the Gaussian 09 suite of programs.20 The structures of the iridium complexes were optimized for the ground-state singlet molecules in the gas phase, assuming a C3 symmetry constraint. Subsequently, these structures were verified as equilibrium ones by performing analytic vibrational frequency calculations at the same level of computation. For these calculations, the keyword−option combination, “int=ultrafine”, was used to specify the integration grid in order to avoid imaginary frequencies arising from low frequency modes. The level of computation for the geometry optimizations and vibrational frequencies utilized the pure BP86 functional,21,22 which has enhanced computational performance gains compared to the hybrid functional, B3LYP.23,24 In addition, the 6-31G25 basis sets were used for the main-group atoms, while the Stuttgart pseudopotential was used for Ir (specifically, the MWB60 keyword for the pseudopotential parameters and triple-ζ quality valence basis sets). The optimized Ir structures then underwent time-dependent DFT (TDDFT) computations, using the same basis sets but with the B3LYP functional. Previous computational reports on Ir complexes have predicted satisfactory excitation energies when using this functional. 18,26−39 These excited-state computations involved the lowest seven triplet energy transitions; only the first triplet transition energies are reported later. Single-point energy calculations were then performed for these structures to obtain the energy values for both the highest-occupied molecular orbitals (HOMOs) and the lowestunoccupied molecular orbitals (LUMOs), using the B3LYP functional with the larger 6-31+G(d)40,41 basis sets for the main-group atoms (the Ir pseudopotential remains the same since it includes a diffuse s function and a diffuse d function in the valence set). These larger main-group basis sets include extra diffuse and polarization functions, which have proven economical and efficient to obtain acceptable HOMO and LUMO values compared to experiment.42,43
4. RESULTS AND DISCUSSION Ligand and Ir(III) Complex Synthesis. To synthesize the 1,3,5-substituted-1H-1,2,4-triazole ligands presented in this work, we utilized a convenient one-pot method described recently by Castanedo and co-workers.44 The room temperature reaction of an aromatic carboxylic acid with acetimidamide in the presence of the coupling agent 2-(7-Aza-1Hbenzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HATU) and diisopropylethylamine (DIPEA) occurred to form the intermediate benzoyl amidine. After the addition of acetic acid and either cyclohexylhydrazine hydrochloride (compounds 1−4) or methyl hydrazine (compound 5) and heating at 90 °C followed by purification, 5-arylsubstituted 1H-1,2,4-triazoles were obtained in good yields (1, 71%; 2, 74%; 3, 76%; 4, 55%) (Scheme 1). However, the yield of compound 5 was only 21% due to competing formation of the primary amide [1,1′:3′,1″-terphenyl]-3-carboxamide, which
3. EXPERIMENTAL SECTION The photophysical properties of complexes 6−10 were characterized by UV−vis absorption, luminescence, photoluminescent quantum efficiency (PLQE), and photoluminescent lifetime. Solutions of the compounds were prepared in spectroscopic grade toluene at concentrations of ∼10 μm. The solutions were degassed by four rounds of freeze−pump−thaw. The absorption spectra were recorded using an Ocean Optics Chem2000-UV−vis unit. The luminescence spectra were measured with a Fluorolog (ISA Jobin-Yvon-Spex) with excitations at 300, 320, 340, and 360 nm. Emission spectra were independent of exciting wavelength. The photoluminescent quantum yield (PLQE) was measured relative to a freshly 27764
dx.doi.org/10.1021/jp5096322 | J. Phys. Chem. C 2014, 118, 27763−27771
The Journal of Physical Chemistry C
Article
or recrystallization to afford complexes 6−10 as yellow, crystalline solids (Figure 1). 1H NMR spectroscopic analyses of these compounds, and X-ray studies of 9 and 10, were consistent with facial geometry for complexes 6-10. X-ray Crystallographic Studies. X-ray quality crystals of complexes 9 and 10 were grown from dichloromethane/ pentane and toluene, respectively. Complex 9 was obtained as the dichloromethane solvate and 10 as the bis(toluene) solvate. The solid-state structures of these compounds are shown in Figures 2 and 3. As noted above, these are fac-isomers. The IrC and IrN bond distances for both complexes are shown in Table 1; not surprisingly, the bond distances do not differ significantly between the complexes. Table 2 provides values for the dihedral angles, ϕ, formed between the iridium-bound aryl rings and the attached phenyl or o-ethylphenyl rings of 9 and 10, respectively. The average value of ϕ for the three cyclometalated ligands in 9 is 47°, whereas for 10 it is 69°. As we shall see below, the greater ϕ in 10 leads to a significant blue-shift in its emission relative to 9. Computational Studies. An important goal of the present study was to develop an efficient protocol for reliably predicting molecular structures, phosphorescent emission energies, and HOMO/LUMO energy values for cyclometalated Ir(III) emitters. Previous studies have reported better correlation between experimental and computational emission properties when vertical excitation energies were calculated at the groundstate singlet geometry rather than the first excited triplet geometry.27,28,39 This general finding for excitation energies from previous studies has led to the computational protocol used in the current investigation. For iridium complexes similar to those in Figure 1, previous literature reports employed double-ζ quality basis sets for the main-group elements, with polarization on the heavy atoms,
Scheme 1. Synthesis of Aryl-Substituted 1H-1,2,4-Triazole Ligands
had to be separated by multiple chromatographies and extraction with hot hexane. For complexation to Ir(III), we used a procedure similar to that reported by Burn, Samuel, and co-workers.18 Excess ligand was allowed to react with IrCl3(H2O)3 in a refluxing 2ethoxyethanol/water (3:1) mixture. This reaction formed the cyclometalated L2Ir(III) chloro-bridged dimer, which precipitated from the reaction mixture after cooling and the addition of more water. The chloro-bridged dimer was isolated by filtration and characterized by 1H NMR spectroscopy. It was then dissolved in dichloromethane and allowed to react with silver trifluoromethanesulfonate. The AgCl that was formed was removed by filtration, and the filtrate was concentrated to afford the corresponding Ir(III) triflate complex. Finally, the triflate complex was allowed to react with excess ligand in refluxing 2-ethoxyethanol to form the tris-cyclometalated Ir(III) complex. The products were purified by chromatography and/
Figure 1. Ir(III) complexes of cyclometalated 1H-1,2,4-triazole ligands. Note that complex 11 was not synthesized and is only treated computationally in this study. 27765
dx.doi.org/10.1021/jp5096322 | J. Phys. Chem. C 2014, 118, 27763−27771
The Journal of Physical Chemistry C
Article
Figure 2. X-ray crystal stucture of complex 9.
Figure 3. X-ray crystal structure of complex 10.
and an effective core potential for the transition metal (e.g., LANL2DZ or SDD).26−38 A more recent literature article reported optimized structures obtained with all-electron basis sets which were triple-ζ and included polarization.39 No matter the size of the basis sets, the predicted metric parameters were essentially the same. For the current investigation, the basis sets for the main-group atoms are of double-ζ quality but without added polarization functions. The iridium metal was represented by a pseudopotential for the core electrons and a
triple-ζ basis set for the valence electrons. For the six iridium complexes represented in Figure 1, optimized gas-phase geometries were obtained for the singlet ground states, assuming C3 symmetry and utilizing the BP86 functional. A selection of optimized bond lengths, bond angles, and dihedral angles are given in Table S1 (Supporting Information, SI). As expected, the metric parameter values local to the metal center are consistent with previous literature reports. The very minor differences in the computed metric parameters among these six 27766
dx.doi.org/10.1021/jp5096322 | J. Phys. Chem. C 2014, 118, 27763−27771
The Journal of Physical Chemistry C
Article
one may use H−Lgap values as a guide to determine which substituent groups on similar molecules would red-shift or blueshift the emission energies. Additional computations were carried out as a minor test of this trend. In Table 3, the numbers in parentheses are the computed electronic properties for complex 11, in which the amino-methyl group in the triazole of complex 8 was replaced with an amino-cyclohexyl group. Since the H−Lgap of complex 11 is identical to that of 7, it is expected that the computed T1 energies would be identical, which is the case. Of course, this computational trend needs to be verified with larger sets of similar molecules. There are two important structure−property relationships within this set of metal complexes that we examined computationally. The first is the result of adding a phenyl ring to two different positions of the phenyl which is bonded to the iridium (6 → 7 or 9). In both cases, this addition leads to a decrease of the triplet emission energy (red-shifting) compared to 6. Since molecular orbital delocalization on the ligand is greater in the LUMO than in the HOMO, adding a conjugating phenyl ring to 6 lowers the LUMO energies for 7 and 9 compared to 6, whereas the HOMO energies for all three complexes are essentially identical. This lowering of LUMO energies results in smaller H−Lgap values for 7 and 9 and concomitant red-shifting of the triplet emission energies compared to 6. Another important structure−property relationship relates to twisting phenyl rings out of conjugation and how this affects emission triplet energy. We discussed above the dihedral angle ϕ formed between the cyclometalated aryl rings and the attached phenyl or ortho-ethyl phenyl rings of 9 and 10. The respective experimental (average values derived from the X-ray studies) and computed dihedral angles are in Table 4. The average experimental dihedral angle of 47° (34° computationally) in 9 is increased to 69° (51°) in 10 because of the orthoethyl group in the latter molecule. Increasing the dihedral angle from 9 to 10 results in a blue-shifting of the computed triplet energy by 0.08 eV (14 nm). From a comparison of the LUMO pictures in Figure 4 for these two molecules, one can see that increasing the dihedral angle from 9 to 10 essentially eliminates the π-bonding character for the phenyl−phenyl CC link, thereby raising the LUMO energy level from −1.53 to −1.41 eV. This orbital energy change results in a larger H−Lgap and a higher triplet energy for 10 compared to 9. Thus, our calculations suggest that by strategically adding a sterically
Table 1. IrC and IrN bond distances in complexes 9 and 10 IrC bond distances (Å)
complex 9
10
expt.
comp.
Ir(1)C(29) 2.021(4) Ir(1)C(50) 2.025(4) Ir(1)C(8) 2.026(4) Ir(1)C(8) 2.013(3) Ir(1)C(54) 2.018(3) Ir(1)C(31) 2.019(3)
2.04
2.04
IrN bond distances (Å) expt. Ir(1)N(7) Ir(1)N(1) Ir(1)N(4) Ir(1)N(1) Ir(1)N(7) Ir(1)N(4)
comp. 2.133(3) 2.140(4) 2.141(4) 2.119(2) 2.123(2) 2.125(2)
2.15
2.15
iridium complexes reflect a similar molecular environment local to the metal center, an observation already made for the X-ray crystal structure IrC and IrN bond distances for complexes 9 and 10. Table 1 contains the corresponding computed bond distances for these complexes, indicating a very good agreement between the crystal structure and computational values for the local structure around the iridium center. For complexes 6−10, computed HOMO and LUMO energy values and associated differences, H−Lgap, are reported in Table 3 along with the first excited triplet (T1) transition energies. With just a handful of similar molecules, determining firm trends in the electronic properties is challenging, but a few have become identifiable. To help understand these trends, the frontier molecular orbital pictures of complexes 9 and 10 are presented in Figure 4; these are qualitatively representative of all molecules studied. The HOMO and LUMO are of primary importance since the TDDFT calculations reveal that the first triplet (S0 → T1) excited states primarily involve the electronic transition between these two orbitals. The HOMOs have electron density primarily on the metal and a small amount delocalized on the cyclometalated ring in all three ligands; an even smaller amount of electron density appears localized on two of the nitrogens in the triazole ring of each ligand. In contrast, the LUMOs have no electron density on the metal, only on the ligands, and show primarily π-bonding character for the phenyl-triazole CC link. These HOMO and LUMO descriptions mirror what has been previously reported in the literature for homoleptic iridium complexes.26,39 The importance of the HOMO and LUMO for the electronic transitions is readily observed in the ordering of the H−Lgap values in Table 3; namely, 9
