You are on page 1of 10

Article

pubs.acs.org/JPCC

Correlating Experimental Photophysical Properties of Iridium(III)


Complexes to Spin−Orbit Coupled TDDFT Predictions
Jarod M. Younker* and Kerwin D. Dobbs
Central Research and Development, E. I. DuPont de Nemours and Co. Inc., RT 141 Henry Clay, Wilmington, Delaware 19880,
United States
*
S Supporting Information

ABSTRACT: Ir(III) complexes are efficient phosphorescent emitters. The


transition dipole moment between the triplet and singlet manifolds is
formally spin forbidden. However, spin−orbit coupling (SOC), induced by
the high angular momentum orbitals in Ir, efficiently mixes the triplet
manifold with higher-energy singlets, increasing the transition probability.
Spin−orbit coupled time-dependent density functional (TDDFT) calcu-
lations within the zero-order relativistic approximation (ZORA) are used to
study nine complexes that have a range of emissions from 450 to 630 nm
and quantum efficiencies of 0.1−0.9. We find that using the singlet ground-
state geometry to calculate radiative rates produces the best correlation with
experiment. We also show that the equal thermal population of the three
sublevels in the triplet manifold is sufficient to understand rates at 300 K. We find that emission energies and radiative rates are
best reproduced at the TD-B3LYP/TZP/DZP//BP86/TZ2P/TZP level of theory, where the larger basis set is supplied for Ir.

■ INTRODUCTION
A key property in organic light-emitting diodes (OLEDs),
A growing number of reports in the literature are using time-
dependent density functional theory (TDDFT) to predict
sensors, probes, imaging agents, and photosensitizers is emission energies and lifetimes (using S0 → T1 vertical
molecular luminescence. Exciton recombination provides the excitation energies), with the goal to understand what controls
required energy for these applications. Unfortunately, some of quantum efficiency and the color of phosphorescence at the
the materials used suffer from low electroluminescence molecular level.6−27 In order to accurately predict the emission
quantum yield due to spin statistics. During exciton energy and lifetime of an iridium complex, the geometry
recombination, two non-geminate electrons couple to give a employed is critical. Both the optimized excited-state triplet and
singlet and triplet spin manifold. Spin statistics limit the ground-state singlet geometries have been advocated by
fluorescent quantum efficiency to a maximum of 25% for purely different groups. It is given that exciton recombination
organic molecules. The remaining energy is lost from the triplet populates the T1 manifold, from which phosphorescent
manifold in the form of heat. However, efficiency can be emission is observed. A review of the literature found a handful
increased to nearly 100% in transition metal complexes of reports where emissive properties were predicted using an
featuring π-conjugated ligands. optimized triplet geometry.8,9,16,18,19,24 Conversely, others have
The presence of the metal center performs two important reported that emission properties are better correlated with
roles. First, spin−orbit coupling (SOC) increases intersystem experiment using the optimized singlet geometry.7,10,12,22 In
crossing from the excited singlet manifolds (Sn, n > 0) to the fact, two reports that looked at emission properties from both
triplet manifold (T1). This process has been referred to as geometries found better experimental correlation when vertical
triplet harvesting and is well documented in the literature.1,2 excitation energies were calculated at the singlet geometry.20,22
Second, the formally spin forbidden T1 → S0 transition is Phosphorescence is a long-lived process (on the order of
activated via SOC.1−3 The matrix element between the ground microseconds), so it is anticipated that the triplet will have time
state (S0) and T1 π−π* states in purely organic systems is to reorganize to a lower-energy geometry. Jansson et al. looked
negligible due to low orbital angular momentum. Transition at the potential energy surface cross section of Ir(ppy)3, along
metals have high orbital angular momentum that couples the vibrational mode corresponding to the geometric change
efficiently to the electron spin, activating the previously between the singlet and triplet geometries.22 They found that
forbidden transitions. The emitter tris(2-phenylpyridine) the triplet occupied a shallow anharmonic potential such that
iridium or Ir(ppy)3 is prototypical of the process of triplet
harvesting.4,5 Ir(ppy)3 emits a green color with ∼90% efficiency Received: October 25, 2013
and has been used as a springboard for similar phosphorescent Revised: November 15, 2013
emitters. Published: November 18, 2013

© 2013 American Chemical Society 25714 dx.doi.org/10.1021/jp410576a | J. Phys. Chem. C 2013, 117, 25714−25723
The Journal of Physical Chemistry C Article

Figure 1. The cyclometalated iridium(III) complexes studied and their abbreviations.

the maximum of the vibrational wave function lies between the structures were optimized, using standard functionals and
S0 and T1 equilibrium positions. basis sets. In addition, the emission energies and lifetimes were
The goal of this investigation is to develop an efficient explored with respect to changes in the basis sets.
computational protocol for reliably predicting phosphorescent
emission energies and lifetimes for organoiridium complexes.
This protocol was validated against the published experimental
■ COMPUTATIONAL MODELS
The excited state(s) leading to phosphorescence are formed by
results for nine Ir complexes with reported phosphorescent electron−hole recombination. Two distance-dependent mech-
wavelengths from 450 to 630 nm and quantum efficiencies of anisms, Dexter and Förster, have been proposed to describe the
0.1−0.9 (Figure 1 and Table 1). In developing the protocol, excitation where we refer the interested reader to the
both ground-state singlet and first excited-state triplet literature.1,32−37 Alternatively, recombination can occur on
the organoiridium complex with hole trapping occurring first.1
Following excitation of the emitter, there are two limiting cases
Table 1. Experimental Emission Energies (λmax),
which may be considered and which are a function of the
Photoluminescence Quantum Yields (ΦPL), and
environment (Figure 2). If the triplet state is long-lived and the
Phosphorescence Lifetimes (τ) for the Complexes Studied
environment reorganization energy is small, the excited emitter
complex λmax (nm) ΦPL τ (μs) will relax on the triplet potential energy surface. However, if
a
Ir(ptz)3 449 0.66 1.08 phosphorescence is faster than reorganization of the triplet
Ir(ppy)2(acac)b 516 0.34 1.6 manifold, the excited emitter would remain near the ground-
Ir(ppy)3c 519 0.9 1.6 state singlet geometry. If the former case is true, emission can
Ir(bzq)2(acac)b 548 0.27 4.5 be described by vertical transitions from the ground-state
Ir(pq)2(acac)b 597 0.1 2 singlet at the triplet geometry into the triplet manifold. In the
Ir(piq)3b 624 0.45 1.25 latter, vertical transitions from the ground-state geometry
Ir(piq)2(ppy)b 627 0.37 1.39 would be adequate. High-resolution results from low-temper-
Ir(piq)(ppy)2b 631 0.4 1.38 ature spectroscopies probably fall into the former category.
Ir(piq)2(acac)c 632 0.2 1.1 Most likely, the emitting complex occupies many local minima
on the potential energy surface and macroscopic properties are
a
In toluene.28 bIn 2-MeTHF.29,30 cIn CH2Cl2.5,31 better described by an ensemble of geometric and electronic
25715 dx.doi.org/10.1021/jp410576a | J. Phys. Chem. C 2013, 117, 25714−25723
The Journal of Physical Chemistry C Article

4α0 3
kir = k r(S0 , T1i) = ΔES0 − T1i 3 ∑ |M ji|2
3t0 j ∈ {x , y , z} (1)
where α0 is the fine-structure constant, t0 = (4πε0) ℏ /mee4, 2 3

ΔES0−T1i is the transition energy, and Mij is the spin−orbit


coupled S0 → T1 transition moment. Mij is the only non-
vanishing component of the first-order corrected wave
functions:
∞ ⟨S0|μĵ |Sn⟩⟨Sn|Ĥ SO|T1i⟩
M ji = ∑
n=0 E(Sn) − E(T1)
∞ ⟨S0|Ĥ SO|Tm⟩⟨Tm|μĵ |T1i⟩
+ ∑
m=1 E(Tm) − E(S0) (2)
where the electric dipole transition moment from the ith
Figure 2. Limiting case vertical excitations from the optimized singlet sublevel of T1 and S0 is evaluated according to linear response
geometry (green) and optimized triplet geometry (red). Inset: theory and the Cartesian components j ∈ {x, y, z} are used to
Relativistic effects split the T1 manifold into three sublevels that are
separated in energy in the absence of an applied field. This splitting is
represent the spin eigenfunctions. The operators μ̂j and Ĥ SO are
referred to as the zero-field splitting (ZFS). the electric dipole and spin−orbit Hamiltonian, respectively.
The two terms in eq 2 have an inverse dependence on the
energy differences between the spin manifolds; the relative
states. We decided to study the minimum-energy S0 and T1
importance of each term can be qualitatively deduced (see
geometry extremes.22
illustration in Figure 3). Since the energy differences between
The experimental electronic properties are dependent upon
the environment. The experimental numbers we reference in
this paper were obtained from measurements made in toluene,
CH2Cl2, and 2-MeTHF. To minimize the number of response
variables, we decided first to focus on gas-phase calculations,
with the hope that this work could serve as a benchmark to
broadly predict emission properties. Second, we included
environmental effects by implicitly modeling solvent according
to the conductor-like screening model (COSMO).
At the non-relativistic limit, the solution to the Schrödinger
equation is a complex scalar field. However, SOC is a relativistic
property and plays a role in energetics of transition metals, such
as Ir. For electrons treated relativistically, the solution to the
Dirac equation is a four-spinor. In the Dirac wave function, two
spinors represent antimatter and for our purposes can be
neglected. Subsequent application of the Born−Oppenheimer
approximation significantly simplifies the equation while
keeping the most important relativistic corrections. Analytic
expansion of the resulting equation yields the zeroth-order
regular approximation (ZORA) to the Dirac equation, whose
two-component solution is determined self-consistently.38 Figure 3. Relative strength and deconvolution of terms from eq 2 as
Several excellent discussions of relativistic TDDFT calculations predicted by the denominator (e.g., E(Sn) − E(T1)). The wider the
with SOC exist in the literature.7,8,11,22,39−41 solid line, the more important the contribution (red/magenta > blue >
Within the Dirac−Kohn−Sham equations for closed-shell green).
systems, spin dependencies appear only at second order,
allowing them to be treated as a perturbation of the scalar the singlet ground state (S0) and the excited triplet states (Tm)
relativistic or one-component wave functions. We, as well as are much greater than the energy differences between the first
others, have found that the use of an effective single-electron excited triplet state (T1) and the first few excited singlet states
approximation significantly decreases computational cost via (Sn), the second term in eq 2 is much less than the first term (as
elimination of the two-electron spin−orbit integrals.8,41 is the n = 0 element of the first term). As such, Mij is
Treating SOC as a perturbation, following the determination proportional to the product of the singlet transition dipole
of singlet−singlet and singlet−triplet vertical excitation energies moments for n > 0 (⟨S0|μ̂|Sn⟩) and the spin−orbit coupling
and transition moments, is as accurate as the self-consistent element between the emitting triplet state and the first few
two-component calculation. excited singlet states (⟨Sn|Ĥ SO|T1⟩). To further conceptualize
SOC splits the T1 manifold into three sublevels that are this equation to the chemical level, we note that the magnitude
separated in energy in the absence of an applied field. This of the singlet transition dipole moment is proportional to the
splitting is referred to as the zero-field splitting (ZFS, see inset oscillator strength divided by the energy of the respective
of Figure 2). The radiative rate from each triplet sublevel is transition, or in other words the “allowedness” of the singlet−
given by the following equation: singlet transitions between the excited-state singlets and the
25716 dx.doi.org/10.1021/jp410576a | J. Phys. Chem. C 2013, 117, 25714−25723
The Journal of Physical Chemistry C Article

Table 2. Optimized Gas-Phase Singlet (S0) and Triplet (T1) Ir−C/N/O Bond Distances (Å) at BP86 and B3LYPa
BP86 B3LYP BP86 B3LYP
S0 T1 S0 T1 S0 T1 S0 T1
Ir(ptz)3 Ir(ppy)2(acac)
Ir−C1 2.030 2.034 2.030 2.037 Ir−C1 2.003 1.988 2.006 1.982
Ir−C2 2.029 1.994 2.034 1.995 Ir−C2 2.002 1.987 2.006 1.982
Ir−C3 2.029 2.040 2.034 2.048 Ir−N1 2.041 2.045 2.054 2.055
Ir−N1 2.173 2.193 2.199 2.227 Ir−N2 2.040 2.045 2.054 2.055
Ir−N2 2.173 2.148 2.188 2.145 Ir−O1 2.171 2.162 2.170 2.172
Ir−N3 2.174 2.168 2.196 2.197 Ir−O2 2.172 2.163 2.171 2.171
Ir(ppy)3 Ir(bzq)2(acac)
Ir−C1 2.026 2.016 2.024 2.020 Ir−C1 2.010 1.989 2.015 1.989
Ir−C2 2.027 2.019 2.026 2.041 Ir−C2 2.009 1.988 2.014 1.989
Ir−C3 2.024 2.014 2.025 1.985 Ir−N1 2.049 2.060 2.061 2.072
Ir−N1 2.152 2.153 2.169 2.192 Ir−N2 2.050 2.060 2.062 2.073
Ir−N2 2.152 2.152 2.166 2.176 Ir−O1 2.165 2.152 2.163 2.146
Ir−N3 2.152 2.150 2.166 2.128 Ir−O2 2.167 2.154 2.163 2.146
Ir(piq)3 Ir(pq)2(acac)
Ir−C1 2.024 2.014 2.022 2.015 Ir−C1 1.989 1.979 1.992 1.980
Ir−C2 2.024 2.014 2.023 2.017 Ir−C2 1.989 1.978 1.992 1.979
Ir−C3 2.023 2.013 2.022 2.013 Ir−N1 2.080 2.083 2.101 2.095
Ir−N1 2.153 2.148 2.163 2.153 Ir−N2 2.081 2.083 2.100 2.096
Ir−N2 2.154 2.147 2.163 2.153 Ir−O1 2.178 2.166 2.188 2.168
Ir−N3 2.153 2.148 2.162 2.149 Ir−O2 2.179 2.166 2.187 2.166
Ir(piq)2(ppy) Ir(piq)2(acac)
Ir−C1* 2.029 2.024 2.028 2.023 Ir−C2 1.995 1.977 2.001 1.977
Ir−C2 2.022 2.031 2.021 2.034 Ir−C2 1.995 1.977 2.001 1.977
Ir−C3 2.022 1.999 2.020 2.008 Ir−N1 2.043 2.046 2.061 2.052
Ir−N1* 2.156 2.173 2.171 2.183 Ir−N2 2.043 2.048 2.061 2.052
Ir−N2 2.154 2.152 2.163 2.167 Ir−O1 2.175 2.175 2.172 2.167
Ir−N3 2.147 2.115 2.157 2.102 Ir−O2 2.175 2.176 2.172 2.168
Ir(piq)(ppy)2
Ir−C1* 2.027 2.018 2.026 2.021
Ir−C2* 2.028 2.038 2.026 2.041
Ir−C3 2.020 2.009 2.019 2.008
Ir−N1* 2.153 2.164 2.167 2.178
Ir−N2* 2.155 2.164 2.169 2.177
Ir−N3 2.147 2.105 2.158 2.100
a
Atoms are defined in Figure 4. For heteroleptic complexes, the C/N atoms from ppy are designated with an asterisk (*). Changes in bond
distances, upon going from the singlet to the triplet geometry, are shown in italic (>0.02 Å) and bold-faced (<−0.02 Å) fonts.

ground state. We also note that the coupling matrix element 3


1
increases as the degree of metal d character (χ(3d)) in the T1 kr = ∑ kri
3 (4)
state increases. For metal-to-ligand charge transfer (MLCT), i=1
we further stipulate that, for coupling between the excited The validity of this approximation will be explored in this work.
singlet and triplet states to be allowed, the same π*-orbital must Closed-shell singlet and open-shell triplet geometry opti-
be involved, as well as Ir t 2g orbitals with different mizations with no symmetry constraints were performed with
symmetries.2,3,18,20 the Amsterdam Density Functional package (ADF 2012.01)42
The equation to determine the radiative rate of each triplet using the provided BP86 and B3LYP functionals. These two
sublevel was given in eq 1. Assuming a thermal population functionals are the workhorses of computational chemistry,
distribution governed by Boltzmann statistics of the triplet with each one representative of a broad class of commonly
sublevels, the overall observable rate is given by the following employed methods (generalized gradient approximation and
equation: hybrid). We anticipate that other functionals of these two
classes would perform comparably. An all-electron TZ2P basis
k1r + k 2r exp( −ZFSi,ii /kBT ) + k 3r exp( −ZFSi,iii /kBT ) was used for Ir, while TZP was used for all other atoms. No
kr =
1 + exp( −ZFSi,ii /kBT ) + exp( −ZFSi,iii /kBT ) core electrons were frozen during the SCF calculation. A Becke
(3) grid of “good” quality was used for the numerical integration.
Vibrational frequency calculations at the same level of
where ZFS is the so-called zero-field splittings between each of computation followed to verify that each geometry was an
the three sublevels (i, ii, iii). The ZFSi,iii for phosphorescent energetic minimum. Unless otherwise noted, all optimizations
emitters is typically <200 cm−1.1,2 If kBT ≫ ZFSi,iii, then the and vibrational frequencies were performed for molecules in
previous equation reduces to the gas phase.
25717 dx.doi.org/10.1021/jp410576a | J. Phys. Chem. C 2013, 117, 25714−25723
The Journal of Physical Chemistry C Article

Figure 4. Ball-and-stick structure, calculated closed-shell HOMOs (bottom), and LUMOs (top) of Ir(ppy)3 at singlet and triplet BP86 and B3LYP
geometries. Ir, N, C, and H are colored orange, blue, black, and gray, respectively.

Figure 5. Ball-and-stick structure, calculated closed-shell HOMOs (bottom), and LUMOs (top) of Ir(ppy)2(acac) at singlet and triplet BP86 and
B3LYP geometries. Ir, O, N, C, and H are colored orange, red, blue, black, and gray, respectively.

Following geometry optimization, a one-component ZORA For the TDDFT calculations, the following basis sets were
TDDFT calculation, which included SOC perturbatively, tested: TZ2P/TZP, TZP/TZP, TZP/DZP, and TZP/DZ,
followed using the B3LYP functional. The 20 lowest scalar where the two basis sets separated by the backslash are for
relativistic singlet and triplet excitations were included. In order the Ir and main group elements, respectively. Unless otherwise
to verify SOC as a perturbation to the first-order wave noted, all calculations were performed in the gas phase.
functions is satisfactory in predicting radiative rates, self-
consistent two-component spin−orbit ZORA TDDFT calcu-
lations were done for Ir(ppy)3 and Ir(piq)(ppy)2 (optimized
■ RESULTS AND DISCUSSION
Structures. The chemical structures of the molecules
BP86 singlet and triplet). Others have reported that the studied are shown in Figure 1. Optimized gas-phase geometries
radiative rates using the perturbative spin−orbit approach give were obtained for both the ground-state singlet and first
results to within 15% of the full relativistic treatment.41 Our excited-state triplet states at the BP86/TZ2P/TZP and hybrid
results agree. B3LYP/TZ2P/TZP levels of theory. Other literature reports
25718 dx.doi.org/10.1021/jp410576a | J. Phys. Chem. C 2013, 117, 25714−25723
The Journal of Physical Chemistry C Article

employed a double-ζ quality basis set for the main group


elements, with polarization on the heavy atoms, and an effective
core potential for the transition metal (e.g., LANL2DZ or
SDD).6−24 The geometries obtained with the all-electron basis
set are consistent with previous reports. Optimized Ir−O, Ir−
N, and Ir−C bond lengths are given in Table 2.
The general trend is for a nearly symmetric singlet ground-
state geometry to become distorted upon going to the first
excited-state triplet geometry. This distortion is consistent with
previous reports.6,12,16,18,19,22,24 We observe the following upon
going from the optimized singlet geometry to the triplet: (1)
Ir(ptz)3, one ptz ligand is pulled closer to the metal, while the
N of a second ligand is pushed away; (2) Ir(ppy)3, for B3LYP
only is one ppy ligand pulled closer to the metal, while the N of
a second ligand is pushed away; (3) Ir(piq)3, all ligands are
pulled in toward the metal; (4) Ir(piq)2(ppy) and Ir(piq)-
(ppy)2, one piq ligand is pulled in closer to the metal; (5)
Ir(ppy)2(acac), Ir(bzq)2(acac), and Ir(piq)2(acac), the Ir−C
bond distances shorten; (6) Ir(pq)2(acac), the Ir−O bond
distances shorten.
Frontier molecular orbitals are shown for Ir(ppy)3 and
Ir(ppy)2(acac) in Figures 4 and 5, and are qualitatively
representative of all molecules studied. The highest-occupied
molecular orbitals (HOMOs) show density primarily on the
metal, as well as on the phenyl ring of the ligand. For Ir(ppy)3,
the lowest-unoccupied molecular orbitals (LUMOs) have no
density on the metal, only on the ligands, with the exception of
the B3LYP triplet geometry, which has a small percentage of d-
character (1.6%, Figure 4). For Ir(ppy)2(acac), LUMOs are
primarily of π* character, with a small contribution from the
metal d-orbitals. These observations are consistent with the
MLCT nature of the S0 → T1 transition (intraligand π → π*
transitions also contribute). It is well-known that difficulties
arise when using DFT to describe charge transfer, but these
difficulties can be alleviated by including exact exchange, as
found in the hybrid functional B3LYP.43−47
Emission Energies. At the BP86 singlet geometries, the Figure 6. Correlation of calculated (TD-B3LYP) and experimental
emission energies were calculated as the average vertical emission energies for the BP86-optimized singlet (S0) and triplet (T1)
geometries as a function of basis set: TZP/DZ (open circles), TZP/
excitation from the singlet to the three lowest-energy triplet DZP (solid squares), TZP/TZP (open squares), and TZ2P/TZP
sublevels (see Figure 6 for correlated emission energies). The (solid circles). Linear fits can be found in the Supporting Information.
TD-B3LYP emission lines at this geometry featured a high
degree of correlation across the entire spectrum of compounds
studied (R2 = 0.95−0.97). The emission energy increased as Radiative Rates. Both radiative and non-radiative processes
additional polarization functions were included, both on the contribute to the photoluminescence quantum yield (ΦPL):
main-group elements and the iridium.
kr
A similar trend and high degree of correlation was observed ΦPL = = τk r
for the emission lines at the B3LYP singlet ground-state k + k nr
r (5)
geometries (R2 = 0.96, see Figure S1 in the Supporting r nr
where k and k are the radiative and non-radiative rates,
Information). The full gamut of basis sets was not studied at respectively, and τ is the lifetime of the phosphorescent state.
the B3LYP geometry, as the data were found to be consistent The most efficient emitter has a quantum yield near a
with the BP86 singlet geometries. Returning to the conceptual maximum value of 1. In purely organic molecules, the T1 →
models explained in the Computational Models section, the S0 transition is spin forbidden, and as such, non-radiative rates
results at the singlet geometries reflect the limiting case where are much faster than phosphorescence. Strong SOC, introduced
phosphorescence closely follows exciton recombination and/or by a transition metal like Ir, increases radiative rates via mixing
energy transfer to the emitter, such that the emitting molecule of the T1 with higher energy excited singlet states, increasing
does not undergo geometric relaxation on the triplet potential ΦPL. To determine the experimental radiative rate of an
energy surface. organoiridium compound, the phosphorescence is fit to a first-
Alternatively, the other limiting conceptual model involves order decay, yielding the lifetime of the T1 state, τ, which is
relaxation of the excited triplet emitter on the potential energy typically on the order of a few microseconds for efficient
surface. As anticipated from Figure 2, this relaxation redshifts emitters of interest. A simple rearrangement of eq 5 gives the
the emission energy. Poorer correlation is found for transitions experimental radiative rates, as shown in Table 1.
from the BP86 and B3LYP triplet geometries (R2 = 0.81−0.93,
see Figure 6 and Figure S1 in the Supporting Information). k r = ΦPL /τ (6)

25719 dx.doi.org/10.1021/jp410576a | J. Phys. Chem. C 2013, 117, 25714−25723


The Journal of Physical Chemistry C Article

One-component scalar relativistic ZORA TDDFT calcula- Both the singlet ground-state and excited triplet state B3LYP
tions with SOC included as a perturbation were used to geometries yielded extremely poor correlations with experiment
calculate radiative rates (eq 1). Rates determined for a subset of (R2 < 0.15, see Figure 7 and Figure S2 in the Supporting
molecules with two-component relativistic TDDFT-SOC are Information). The radiative rates at the BP86 triplet geometry
not appreciably different from the one-component solution. are poor although significantly better than B3LYP (T1, R2 =
Assuming that all three triplet sublevels are equally populated at 0.29−0.40, see Figure S2 in the Supporting Information).
300 K, a simple average of the individual rates for each sublevel In summary, the dependence of the radiative rates on basis
results in the overall radiative rate (eq 4; individual radiative set is more pronounced than for the emission energies.
rates for each sublevel are included in the Supporting Whereas a DZ basis set for the main group elements was
Information). Experimental errors of 5−15% are associated sufficient for transition energies, polarization functions were
with radiative rates of the molecules included in this required to converge the calculated rates. The calculated rates
study.5,28−31 The errors are illustrated via vertical error bars are predicted to within 40% of the experimental results for all
in subsequent figures. complexes except Ir(bzq)2(acac) and Ir(pq)2(acac) at TD-
Moderate correlation between the calculated and exper- B3LYP/TZP/DZP//BP86/TZ2P/TZP (S0 geometry).
imental rates at the BP86 singlet geometry was found (R2 = The use of eq 4 is only valid with the assumption that kBT ≫
0.81−0.85, see Figure 7). A comparison of rates calculated ZFSi,iii; otherwise, the thermal population of the sublevels is not
upon going to higher basis sets converged. The addition of an equal. At 300 K, kBT = 208 cm−1. Experimental and calculated
extra set of polarization functions on the Ir (TZP → TZ2P) ZFS values are shown in Table 3. It is unclear that the previous
does little to alter the rate; however, addition of polarization
functions on the main-group elements (DZ → DZP) has a Table 3. Experimental and Calculated Zero-Field Splittings
large effect upon the rate, lowering it to be more inline with for the Complexes Studied
experiment. complexes ZFSexpa (cm−1) ZFScalcb (cm−1)
Ir(ptz)3 121
Ir(ppy)3 1352 113
Ir(ppy)2(acac) 161
Ir(bzq)2(acac) 81
Ir(pq)2(acac) 186
Ir(piq)3 572 113
Ir(piq)2(ppy) 652 121
Ir(piq)(ppy)2 10730 194
Ir(piq)2(acac) 962 129
a b
Experimental. TD-B3LYP/TZP/DZP//BP86/TZ2P/TZP (S0 ge-
ometry).

assumption is valid for all the complexes studied, at least from a


computational standpoint. For example, only the theoretical
ZFS data for Ir(ppy)2(acac), Ir(pq)2(acac), and Ir(piq)(ppy)2
appear to be on par with kBT. Therefore, we also correlated the
Boltzmann-weighted radiative rates (eq 3) with experiment in
order to increase the accuracy of the predictive model. These
results are shown in Figures S3 and S4 in the Supporting
Information. The linear correlation value R2 increased slightly
from 0.85 to 0.89 for the BP86 S0 geometry and from 0.40 to
0.50 for the BP86 T1 geometry by accounting for the thermal
population distribution. We find that the three complexes that
have ZFS values on par with kBT are now predicted to have
lower radiative rates, better in line with experiment. The large
ZFS decreases the thermal populations of the two higher
energy sublevels (see inset of Figure 2). Radiative transitions
from these two higher sublevels are at least an order of
magnitude faster than the lowest energy sublevel. A detailed
comparison of our calculated radiative rates to experiment, as
well as the few theoretical literature reports, are given in Table
4. We find that our results are in good agreement with
experiment, as well as being comparable to previous work using
spin−orbit coupled TDDFT.9,12,19
Figure 7. Correlation of calculated (TD-B3LYP) and experimental
A major premise of this paper was to design a computational
radiative rates (average) for the BP86- and B3LYP-optimized singlet protocol which could predict general trends in radiative rates
(S0) geometries as a function of basis set: TZP/DZ (open circles), for organoiridium compounds. Experimentally, rates follow the
TZP/DZP (solid squares), TZP/TZP (open squares), and TZ2P/ trend Ir(ptz)3 > Ir(ppy)3 > Ir(piq)3 > Ir(piq)(ppy)2 >
TZP (solid circles). Experimental errors are shown via the vertical Ir(piq) 2(ppy) > Ir(ppy)2(acac) > Ir(piq) 2 (acac) > Ir-
error bars. Linear fits can be found in the Supporting Information. (bzq)2(acac) > Ir(pq)2(acac). The proposed protocol predicts
25720 dx.doi.org/10.1021/jp410576a | J. Phys. Chem. C 2013, 117, 25714−25723
The Journal of Physical Chemistry C Article

Table 4. Experimental and Theoretical Radiative Rates kr emission energies and phosphorescent lifetimes with exper-
(s−1) for the Complexes Studieda imental values for nine complexes (in which we have a high
degree of confidence of the characterization) having a variety of
computational
different ligands, emission energies, and radiative rates.
complex experimental average kBT-weighted We find that vertical excitation properties calculated at the
Ir(ptz)3 (6.1 ± 0.8) × 105 b 5.9 × 105 f,g 4.7 × 105 f,g BP86 S0 geometry correlate the best with experiment. We
4.3 × 105 f,h acknowledge that the long-lived triplet state should have time
Ir(ppy)2(acac) (2.1 ± 0.5) × 105 c 1.8 × 105 f,g 1.4 × 105 f,g to relax and that phosphorescence from such a geometry is
Ir(ppy)3 (5.6 ± 0.4) × 105 d 4.0 × 105 f,g 3.3 × 105 f,g physical. However, correlations using the triplet geometry are
4.8 × 105 f,i poor (both with BP86 and B3LYP). An extensive study of
6.1 × 105 e Ir(ppy)3 by Jansson et al. concluded that the true geometry of
Ir(bzq)2(acac) (0.6 ± 0.1) × 105 c 1.2 × 105 f,g 1.0 × 105 f,g the emitting species is intermediate between the singlet ground
Ir(pq)2(acac) (0.5 ± 0.1) × 105 c 1.8 × 105 f,g 1.0 × 105 f,g and excited triplet states, as the triplet state exists in a shallow
Ir(piq)3 (3.6 ± 0.4) × 105 c 2.9 × 105 f,g 2.4 × 105 f,g anharmonic potential energy well.41 Although the “true”
3.5 × 105 e geometry of the emission state was not determined in this
Ir(piq)2(ppy) (2.7 ± 0.3) × 105 c 2.7 × 105 f,g 2.2 × 105 f,g work, time-dependent calculations from the S0 geometry are
2.2 × 105 e sufficient to produce results in agreement with experiment, and
Ir(piq)(ppy)2 (2.9 ± 0.3) × 105 c 2.7 × 105 f,g 1.6 × 105 f,g thus satisfying the stated intent of our research: to develop a
2.7 × 105 e protocol which is sufficient to reproduce and understand
Ir(piq)2(acac) (1.8 ± 0.4) × 105 d 1.1 × 105 f,g 1.1 × 105 f,g experimental trends. The use of the singlet geometry is also
a
Theoretical rates were calculated using eqs 4 (average) and 3 (kBT- advocated by Minaev et al., where they employed SOC-
weighted). bIn toluene.28 cIn 2-MeTHF.29,30 dIn CH2Cl2.5,31 eIn quadratic response theory to predict photophysical properties
THF.19 fGas phase. gThis work, TD-B3LYP/TZP/DZP//BP86/ of several iridium complexes.25−27
TZ2P/TZP S0 geometry. hFrom ref 9. iFrom ref 12. We have also presented results on the basis set dependence
of the time-dependent calculation of the emission energies and
the general trend of radiative rates, based on the average of the radiative rates. The former can be adequately calculated without
three triplet sublevels, with one exception: Ir(ptz)3 > Ir(ppy)3 > including polarization functions on the main group elements,
Ir(piq)3 > Ir(piq)(ppy)2 ∼ Ir(piq)2(ppy) > Ir(ppy)2(acac) ∼ whereas, for the latter, such higher angular momentum
Ir(pq)2(acac) > Ir(piq)2(acac) > Ir(bzq)2(acac). The one functions are necessary. We propose that photophysical
exception is Ir(pq)2(acac), which is predicted to have a higher properties for cyclometalated Ir complexes be calculated at
rate than observed experimentally. As discussed earlier, the the TD-B3LYP/TZP/DZP level of theory, using the S0
average predicted rate for Ir(pq)2(acac) is high due to the large geometry optimized at BP86/TZ2P/TZP, where the larger
ZFS value. Taking into an account the thermal populations of basis set is used for Ir. We have also analyzed the response of an
the triplet sublevels lowers the rate for Ir(pq)2(acac) to be implicit solvent (both on the vertical excitations and the
more inline with experiment. geometry) and have found that the predicted radiative rates
Finally, we explored the effect of solvent on the emission differ from experiment.
energies and radiative rates. Solvent (toluene, THF, or CH2Cl2) We also show that, at 300 K, the radiative rate can be
was implicitly included via the conductor-like screening model described by a simple average of the rates from the three T1
(COSMO) within ADF. First, the response of the implicit sublevels. Nevertheless, a slightly better correlation is achieved
dielectric was included within the TDDFT calculation (TD- when accounting for the thermal population of the sublevels
B3LYP/TZP/DZP) using the gas-phase-optimized geometries. according to Boltzmann statistics. This unequal distribution of
Second, the organoiridium complexes were optimized within states is needed for complexes which have a large calculated
the implicit solvent (BP86/TZ2P/TZP), with subsequent ZFS. The lowest T1 sublevel has a radiative rate an order of
excitation properties calculated at the level given in the prior magnitude or more smaller than the other two sublevels. By
sentence. The resulting data can be found in the Supporting using a simple average of rates, the predicted values may be
Information (Figure S5 and Table S5). In summary, somewhat faster.
correlations to emission energies were excellent (R2 = 0.96− This work represents the first step toward a general protocol
0.97), with predicted emissions blue-shifted in relation to the suitable for the study of photophysical properties of cyclo-
gas-phase calculations. However, correlation between the metalated iridium. The universality of our approach, and
predicted and experimental radiative rates was poor (R2 = subsequent conclusions, will be validated by the growing size of
0.40−0.59). A large increase in radiative rates was predicted the literature, wherein SOC-TDDFT is used to study similar
systems.


upon including the solvent response. We also compared
electronic properties predicted when including the solvent
response in the TDDFT calculation using either the gas-phase- ASSOCIATED CONTENT
or solvent-optimized geometries. Emission energies and *
S Supporting Information
radiative rates predicted with both geometries were nearly Tables of emission energies and radiative rates of individual
identical, signifying that the gas-phase geometry is adequate to sublevels and linear fits. This material is available free of charge
predict electronic properties. via the Internet at http://pubs.acs.org.

■ CONCLUSIONS
In order to develop an efficient and reliable computational
■ AUTHOR INFORMATION
Corresponding Author
protocol to predict phosphorescent properties for cyclo- *E-mail: Jarod.M.Younker@usa.dupont.com. Phone: (+1) 302-
metalled iridium complexes, we have correlated theoretical 695-7376. Fax: (+1) 302-695-9873.
25721 dx.doi.org/10.1021/jp410576a | J. Phys. Chem. C 2013, 117, 25714−25723
The Journal of Physical Chemistry C Article

Notes (17) De Angelis, F.; Santoro, F.; Nazeruddin, M. K.; Barone, V. Ab


The authors declare no competing financial interest. Initio Prediction of the Emission Color in Phosphorescent Iridium-


(III) Complexes for OLEDs. J. Phys. Chem. B 2008, 112, 13181−
ACKNOWLEDGMENTS 13183.
(18) De Angelis, F.; Belpassi, L.; Fantacci, S. Spectroscopic
We would like to acknowledge support from the developers at Properties of Cyclometallated Iridium Complexes by TDDFT. J.
SCM (Erik van Lethe and Stan van Gisbergen), as well as Mol. Struct.: THEOCHEM 2009, 914, 74−86.
support from the high-performance computing group at (19) Liu, Y.; Gahungu, G.; Sun, X.; Qu, X.; Wu, Z. Effects of N-
DuPont (Alistair J. McGhie and Douglas Cloud). Substitution on Phosphorescence Efficiency and Color Tuning of a

■ REFERENCES
(1) Yersin, H., Ed. Highly Efficient OLEDs with Phosphorescent
Series of Ir(III) Complexes with a Phosphite Tripod Ligand: A DFT/
TDDFT Study. J. Phys. Chem. C 2012, 116, 26496−26506.
(20) De Angelis, F.; Fantacci, S.; Evans, N.; Klein, C.; Zakeeruddin, S.
M.; Moser, J.-E.; Kalyanasundaram, K.; Bolink, H. J.; Grätzel, M.;
Materials; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim,
Nazeeruddin, M. K. Controlling Phosphorescence Color and
Germany, 2008.
Quantum Yields in Cationic Iridium Complexes: A Combined
(2) Yersin, H.; Rausch, A. F.; Czerwieniec, R.; Hofbeck, T.; Fischer,
Experimental and Theoretical Study. Inorg. Chem. 2007, 46, 5989−
T. The Triplet State of Organo-Transition Metal Compounds. Triplet
Harvesting and Singlet Harvesting for Efficient OLEDs. Coord. Chem. 6001.
Rev. 2011, 255, 2622−2652. (21) Abedin-Siddique, Z.; Ohno, T.; Nozaki, K.; Tsubomura, T.
(3) Rausch, A. F.; Homeier, H. H. H.; Yersin, H. Organometallic Intense Fluorescence of Metal-to-Ligand Charge Transfer in [Pt(0)-
Pt(II) and Ir(III) Triplet Emitters for OLED Applications and the (binap) 2] [binap = 2,2′-Bis(diphenylphosphino)-1,1′-binaphthyl].
Role of Spin−Orbit Coupling: A Study Based on High-Resolution Inorg. Chem. 2004, 43, 663−673.
Optical Spectroscopy. Top. Organomet. Chem. 2009, 29, 193−235. (22) Jansson, E.; Minaev, B.; Schrader, S.; Ågren, H. Time-
(4) Finkenzeller, W. J.; Yersin, H. Emission of Ir(ppy)3. Temperature Dependent Density Functional Calculations of Phosphorescence
Dependence, Decay Dynamics, and Magnetic Field Properties. Chem. Parameters for fac-tris(2-Phenylpyridine) Iridium. Chem. Phys. 2007,
Phys. Lett. 2003, 377, 299−305. 333, 157−167.
(5) Hofbeck, T.; Yersin, H. The Triplet State of fac-Ir(ppy)3. Inorg. (23) Liu, Y.; Gahungu, G.; Sun, X.; Su, J.; Qu, X.; Wu, Z. Theoretical
Chem. 2010, 49, 9290−9299. Study on the Influence of Ancillary and Cyclometalated Ligands on
(6) Hay, P. J. Theoretical Studies of the Ground and Excited the Electronic Structures and Optoelectronic Properties of Hetero-
Electronic States in Cyclometalated Phenylpyridine Ir(III) Complexes leptic Iridium(III)) Complexes. Dalton Trans. 2012, 41, 7595−7603.
Using Density Functional Theory. J. Phys. Chem. A 2002, 106, 1634− (24) Qu, X.; Liu, Y.; Godefroid, G.; Si, Y.; Shang, X.; Wu, X.; Wu, Z.
1641. Theoretical Study on Cationic Iridium(III) Complexes with a
(7) Li, X.; Minaev, B.; Ågren, H.; Tian, H. Density Functional Theory Diphosphane Ligand - Geometry, Electronic Properties, and
Study of Photophysical Properties of Iridium(III) Complexes with Application for Light-Emitting Electrochemical Cells. Eur. J. Inorg.
Phenylisoquinoline and Phenylpyridine Ligands. J. Phys. Chem. C Chem. 2013, 2013, 3370−3383.
2011, 115, 20724−20731. (25) Minaev, B.; Ågren, H.; De Angelis, F. Theoretical Design of
(8) Smith, A. R. G.; Burn, P. L.; Powell, B. J. Spin-Orbit Coupling in Phosphorescence Parameters for Organic Electro-Luminescence
Phosphorescent Iridium(III) Complexes. ChemPhysChem 2011, 12, Devices Based on Iridium Complexes. Chem. Phys. 2009, 358, 245−
2429−2438. 257.
(9) Smith, A. R. G.; Riley, M. J.; Burn, P. L.; Gentle, I. R.; Lo, S. C.; (26) Minaev, B.; Minaeva, V.; Ågren, H. Theoretical Study of the
Powell, B. J. Effects of Fluorination on Iridium(III) Complex Cyclometalated Iridium(III) Complexes Used as Chromophores for
Phosphorescence: Magnetic Circular Dichroism and Relativistic Organic Light-Emitting Diodes. J. Phys. Chem. A 2009, 113, 726−735.
Time-Dependent Density Functional Theory. Inorg. Chem. 2012, 51, (27) Li, X.; Zhang, Q.; Tu, Y.; Ågren, H.; Tian, H. Modulation of
2821−2831. Iridium(III) Phosphorescence via Photochromic Ligands a Density
́
(10) Swiderek, K.; Paneth, P. Modeling Excitation Properties of Functional Study. Phys. Chem. Chem. Phys. 2010, 12, 13730−13736.
Iridium Complexes. J. Phys. Org. Chem. 2009, 22, 845−856. (28) Lo, S.-C.; Shipley, C. P.; Bera, R. N.; Harding, R. E.; Cowley, A.
(11) Perumal, S.; Minaev, B.; Ågren, H. Triplet State Phosphor- R.; Burn, P. L.; Samuel, I. D. W. Blue Phosphorescence from
escence in tris(8-Hydroxyquinoline) Aluminum Light Emitting Diode Iridium(III) Complexes at Room Temperature. Chem. Mater. 2006,
Materials. J. Phys. Chem. C 2013, 117, 3446−3455. 18, 5119−5129.
(12) Li, X.; Minaev, B.; Ågren, H.; Tian, H. Theoretical Study of (29) Lamansky, S.; Djurovich, P.; Murphy, D.; Abdel-Razzaq, F.;
Phosphorescence of Iridium Complexes with Fluorine-Substituted Kwong, R.; Tsyba, I.; Bortz, M.; Mui, B.; Bau, R.; Thompson, M. E.
Phenylpyridine Ligands. Eur. J. Inorg. Chem. 2011, 2011, 2517−2524. Synthesis and Characterization of Phosphorescent Cyclometalated
(13) Zhang, T.-T.; Jia, J.-F.; Ren, Y.; Wu, H.-S. Ligand Effects on Iridium Complexes. Inorg. Chem. 2001, 40, 1704−1711.
Structures and Spectroscopic Properties of Pyridine-2-aldoxime (30) Deaton, J. C.; Young, R. H.; Lenhard, J. R.; Rajeswaran, M.;
Complexes of Re(CO)3+: DFT/TDDFT Theoretical Studies. J. Phys. Huo, S. Photophysical Properties of the Series fac-and mer-(1-
̂ ̂
Chem. A 2011, 115, 3174−3181. Phenylisoquinolinato-NC2′)x(2-Phenylpyridinato-NC2′)3−x Iridium-
(14) Polosan, S.; Chow, T. J.; Tsuboi, T. Density Functional Theory (III) (x= 1−3). Inorg. Chem. 2010, 49, 9151−9161.
Analysis of a Mixed-Ligand Iridium Compound for Multi-Color (31) Su, Y. J.; Huang, H. L.; Li, C. L.; Chien, C. H.; Tao, Y. T.; Chou,
Organic Light-Emitting Diodes. J. Phys. Org. Chem. 2008, 21, 315− P. T.; Datta, S.; Liu, R. S. Highly Efficient Red Electrophosphorescent
320. Devices Based on Iridium Isoquinoline Complexes: Remarkable
(15) Gunaratne, T. C.; Gusev, A. V.; Peng, X.; Rosa, A.; Ricciardi, G.; External Quantum Efficiency Over a Wide Range of Current. Adv.
Baerends, E.-J.; Rizzoli, C.; Kenney, M. E.; Rodgers, M. A. J. Mater. 2003, 15, 884−888.
Photophysics of Octabutoxy Phthalocyaninato-Ni(II) in Toluene: (32) Andrews, D. L. A Unified Theory of Radiative and Radiationless
Ultrafast Experiments and DFT/TDDFT Studies. J. Phys. Chem. A Molecular Energy Transfer. Chem. Phys. 1989, 135, 195−201.
2005, 109, 2078−2089. (33) Wu, C.; Djurovich, P. I.; Thompson, M. E. Study of Energy
(16) Song, M.-X.; Hao, Z.-M.; Wu, Z.-J.; Song, S.-Y.; Zhou, L.; Deng, Transfer and Triplet Exciton Diffusion in Hole-Transporting Host
R.-P.; Zhang, H.-J. Efficient Blue-Emitting Ir(III) Complexes with Materials. Adv. Funct. Mater. 2009, 19, 3157−3164.
Phenyl-Methyl-Benzimidazolyl and Picolinate Ligands: A DFT and (34) Jeon, W. S.; Park, T. J.; Kim, S. Y.; Pode, R.; Jang, J.; Kwon, J. H.
Time-Dependent DFT Study. Int. J. Quantum Chem. 2013, 113, Ideal Host and Guest System in Phosphorescent OLEDs. Org. Electron.
1641−1649. 2009, 10, 240−246.

25722 dx.doi.org/10.1021/jp410576a | J. Phys. Chem. C 2013, 117, 25714−25723


The Journal of Physical Chemistry C Article

(35) Suljovrujic, E.; Ignjatovic, A.; Srdanov, V. I.; Mitsumori, T.;


Wudl, F. Intermolecular Energy Transfer Involving an Iridium
Complex Studied by a Combinatorial Method. J. Chem. Phys. 2004,
121, 3745−3750.
(36) Reineke, S.; Lindner, F.; Schwartz, G.; Seidler, N.; Walzer, K.;
Lüssem, B.; Leo, K. White Organic Light-Emitting Diodes with
Fluorescent Tube Efficiency. Nature 2009, 459, 234−238.
(37) Kawamura, Y.; Brooks, J.; Brown, J.; Sasabe, H.; Adachi, C.
Intermolecular Interaction and a Concentration-Quenching Mecha-
nism of Phosphorescent Ir(III) Complexes in a Solid Film. Phys. Rev.
Lett. 2006, 96, 017404.
(38) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. Relativistic
Regular Two-Component Hamiltonians. J. Chem. Phys. 1993, 99,
4597.
(39) Van Gisbergen, S.; Snijders, J. G.; Baerends, E. J.
Implementation of Time-Dependent Density Functional Response
Equations. Comput. Phys. Commun. 1999, 118, 119−138.
(40) van Lenthe, E.; Snijders, J. G.; Baerends, E. J. The Zero-Order
Regular Approximation for Relativistic Effects: The Effect of Spin−
Orbit Coupling in Closed Shell Molecules. J. Chem. Phys. 1996, 105,
6505.
(41) Jansson, E.; Norman, P.; Minaev, B.; Ågren, H. Evaluation of
Low-Scaling Methods for Calculation of Phosphorescence Parameters.
J. Chem. Phys. 2006, 124, 114106.
(42) te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Guerra, C. F.;
Gisbergen, S. J. A. V.; Snijders, J. G.; Ziegler, T. Chemistry with ADF.
J. Comput. Chem. 2001, 22, 931−967.
(43) Yang, W. T.; Cohen, A. J.; Mori-Sanchez, P. Insights into
Current Limitations of Density Functional Theory. Science 2008, 321,
792−794.
(44) Dreuw, A.; Weisman, J. L.; Head-Gordon, M. Long-Range
Charge-Transfer Excited States in Time-Dependent Density Func-
tional Theory Require Non-Local Exchange. J. Chem. Phys. 2003, 119,
2943−2946.
(45) Vlček, J. A.; Zális, S. Modeling of Charge-Transfer Transitions
and Excited States in d6 Transition Metal Complexes by DFT
Techniques. Coord. Chem. Rev. 2007, 251, 258−287.
(46) Adamo, C.; Jacquemin, D. The Calculations of Excited-State
Properties with Time-Dependent Density Functional Theory. Chem.
Soc. Rev. 2013, 42, 845−856.
(47) Jacquemin, D.; Perpète, E. A.; Ciofini, I.; Adamo, C. Assessment
of Functionals for TD-DFT Calculations of Singlet-Triplet Tran-
sitions. J. Chem. Theory Comput. 2010, 6, 1532−1537.

25723 dx.doi.org/10.1021/jp410576a | J. Phys. Chem. C 2013, 117, 25714−25723

You might also like