The Molecular Origin of Anisotropic Emission in an Organic Light

Letter pubs.acs.org/NanoLett

The Molecular Origin of Anisotropic Emission in an Organic LightEmitting Diode Thomas Lee,† Bertrand Caron,† Martin Stroet,† David M. Huang,‡ Paul L. Burn,*,†,§ and Alan E. Mark*,† †

School of Chemistry & Molecular Biosciences, §Centre for Organic Photonics & Electronics, The University of Queensland, St. Lucia Campus, Brisbane 4072, Australia ‡ Department of Chemistry, School of Physical Sciences, The University of Adelaide, Adelaide 5005, Australia ABSTRACT: Atomistic nonequilibrium molecular dynamics simulations have been used to model the induction of molecular orientation anisotropy within the emission layer of an organic light-emitting diode (OLED) formed by vapor deposition. Two emitter species were compared: racemic fac-tris(2-phenylpyridine)iridium(III) (Ir(ppy) 3 ) and trans-bis(2phenylpyridine)(acetylacetonate)iridium(III) (Ir(ppy)2(acac)). The simulations show that the molecular symmetry axes of both emitters preferentially align perpendicular to the surface during deposition. The molecular arrangement formed on deposition combined with consideration of the transition dipole moments provides insight into experimental reports that Ir(ppy)3 shows isotropic emission, while Ir(ppy)2(acac) displays improved efficiency due to an apparent preferential alignment of the transition dipole vectors parallel to the substrate. The simulations indicate that this difference is not due to differences in the extent of emitter alignment, but rather differences in the direction of the transition dipoles within the two complexes. KEYWORDS: Organic light-emitting diodes, thin films, molecular simulation, vapor deposition, transition dipole orientation, outcoupling efficiency

O

introduced by the aliphatic acac ligand, which they proposed would orient toward the vapor phase during deposition.5 They also implied that molecules with high chemical symmetry, such as Ir(ppy)3, orient isotropically. On the other hand, Kim et al. have proposed that the emitters become aligned because of interactions with the horizontally aligned host molecules, with the degree of TDV alignment dependent on the strength of these interactions.4 In more recent work, Moon et al. have used molecular dynamics simulations to model the orientational distribution of individual iridium(III) complexes placed on a surface prepared by annealing a slab of host material at 500 K.7 Some insight into the degree of TDV alignment can be gained through measurement of the spectral radiant intensity the emission intensity as a function of viewing angle and emission wavelength. The spectral radiant intensity can be fitted to an optical simulation model using the fraction of light emitted from horizontal components of the TDVs, ΘH, as a free parameter.8 Microscopically, for a film containing N TDVs, each with a magnitude p, the fraction of light emitted from the horizontal components of the TDVs can be expressed in terms of the sum of the squares of the horizontal components of each individual TDVi, pHi, via:

rganic light-emitting diodes (OLEDs) are increasingly used in electronic displays and lighting. A major factor that affects the external quantum efficiency of OLEDs is the loss of photons due to poor outcoupling from the devices. The emission layer in many of the most efficient OLEDs is composed of a semiconducting host matrix doped with a phosphorescent organometallic guest species. These guests emit light perpendicular to their transition dipole moment vectors (TDVs) such that outcoupling efficiency is optimized when the TDVs are aligned horizontally (parallel to the surface).1 For example, devices containing the homoleptic emitter fac-tris(2-phenylpyridine)iridium(III) (Ir(ppy)3, Figure 1a) emit light isotropically2−4 and consequently have an outcoupling efficiency 12% lower (23% compared to 26%) than devices containing the closely related heteroleptic emitter transbis(2-phenylpyridine)(acetylacetonate)iridium(III) [Ir(ppy)2(acac), Figure 1b],2−6 which emit a larger fraction of light along the vertical axis (perpendicular to the surface). The increased efficiency of devices containing Ir(ppy)2(acac) has been attributed to a net horizontal (parallel to the substrate) alignment of the TDVs. The origin of this net alignment has not been determined experimentally, as the amorphous nature of the emission layers limits direct observation of the orientation of the emitters at an atomic level. Based on observations of a series of related emitters, Jurow et al. proposed that the horizontal alignment of the TDVs of Ir(ppy)2(acac) arises from the chemical asymmetry © 2017 American Chemical Society

Received: August 17, 2017 Published: September 11, 2017 6464

DOI: 10.1021/acs.nanolett.7b03528 Nano Lett. 2017, 17, 6464−6468

Letter

Nano Letters

derived ligands.4 For comparison, experimental studies of the related complex Re(ppy)(CO)4 in a crystalline state suggest that the TDV lies at an angle of 18.5° to the Re−N bond.11 The emission properties of these complexes depend strongly on the environment.12 Thus, the precise TDV directions from calculations or measurements in vacuum or crystal cannot be directly applied to emitters in amorphous guest−host films. Atomistic nonequilibrium molecular dynamics simulations were used to mimic the vapor deposition of emission layers containing Ir(ppy)3 and Ir(ppy)2(acac) to determine the detailed morphology of the layers. These morphologies have been used to examine the distribution of the molecular orientations with respect to the symmetry axis and TDV orientations to provide insight into the origin of the difference in the outcoupling efficiency of the layers. The systems consisted of approximately 6 wt % of either Ir(ppy)3 or Ir(ppy)2(acac) in a host matrix of 4,4′-bis(N-carbazolyl)biphenyl (CBP, Figure 1c). The guest and host molecules were randomly deposited onto a graphene substrate as illustrated in Figure 2. The protocol and conditions used for

Figure 1. Chemical structures of the racemic emitter molecules (a) fac-tris(2-phenylpyridine)iridium(III) (Ir(ppy)3) and (b) trans-bis(2phenylpyridine)(acetylacetonate)iridium(III) [Ir(ppy)2(acac)] and (c) the matrix molecule 4,4′-bis(N-carbazolyl)biphenyl (CBP). In the 3D representations, atoms of iridium, carbon, nitrogen, and oxygen are gray, green, blue, and red, respectively. Hydrogens attached to aromatic rings are included in the simulation model but are not drawn here. The cyan, orange, and magenta arrows indicate the degenerate transition dipole moments, p, of the molecules, illustrated as lying close to the Ir−N bonds. The red arrow indicates the principle symmetry axis vector, m, of the molecules. N

ΘH =

∑i pH2i Np2

1 = N

⎛ pHi ⎞2 ∑⎜ ⎟ p ⎠ i ⎝ N

(1)

Perfect vertical and horizontal alignment of the TDVs correspond to ΘH = 0 and 1, respectively. Because light emitted along the vertical axis originates from the horizontal components of the TDVs, larger values of ΘH would lead to a greater outcoupling efficiency. However, the underlying cause of differences in TDV alignment cannot be inferred from ΘH. For example, ΘH = 2/3 is equally consistent with the distribution of the emitter TDVs within the layer being isotropic or the TDVs of all emitters in the layer being aligned either vertically or horizontally with a 1:2 ratio. An additional complication is that many emitter complexes contain multiple TDVs. For example, Ir(ppy)3 and Ir(ppy)2(acac) have one TDV associated with each ppy ligand, indicated by the colored arrows in Figure 1. Note that the precise directions of the TDVs within Ir(ppy)3 and Ir(ppy)2(acac) are uncertain. Timedependent density functional theory (TDDFT) calculations in vacuum generally agree that the TDV associated with the phenylpyridine ligand in an iridium complex lies in the C−Ir− N plane and projects between the Ir−N and the Ir−C bonds.4,7,9,10 From the first moment of the transition density distribution of the T1 to S1 transition of Ir(ppy)3 in vacuum calculated using TDDFT including spin−orbit coupling by Gonzalez-Vanquez et al.,10 we have determined that the TDV lies at an angle of 49° to the Ir−N bond (equivalently, 87° to the C3 symmetry axis). Similarly, recent calculations by Moon et al. report an angle of 45° to the Ir−N bond.7 The small difference in the TDV direction from TDDFT calculations could arise from the different functionals or basis sets used in the two sets of calculations. A similar calculation carried out for Ir(ppy)2(acac) by Heil et al.9 found the TDV at an angle of 16° to the Ir−N bond. Comparable TDV directions have been reported for other iridium complexes with phenylpyridine-

Figure 2. Snapshots from the simulations during (left) and immediately following (right) the deposition of a CBP−Ir(ppy)2(acac) emission layer. The Ir(ppy)2(acac) molecules and graphene substrate are shown in a space-filling representation, with Ir(ppy)2(acac) colored using the same scheme as the models in Figure 1. The graphene is colored black. The CBP molecules are shown in gray. Hydrogen atoms are not drawn. The definition of ϕ with respect to the plane containing the substrate (xy) and the principle symmetry axis, m, is also illustrated.

the deposition were identical to those previously reported by Tonnelé et al.13 Unlike Tonnelé et al., the films were not annealed prior to the initial analysis of the orientations. All simulations were performed using GROMACS (Version 4.6).14 New molecules were inserted 2 nm from the top of the layer every 20 ps with a random orientation and an initial velocity toward the surface to ensure they reach the growing film. The net velocity was randomly selected from a normal distribution with a mean of 0.05 nm/ps and a standard deviation of kBT /m where m is the mass of the molecule. The species to be deposited was randomly selected, with a target Δguest:Λ-guest:host ratio of 1:1:42 (approximately 6 wt % of the racemic emitter) where Δ and Λ refer to the two possible optical isomers of the guest species. Layers containing 1000 molecules were deposited (thickness ≈ 10 nm). Twenty independent deposition simulations were performed for each emitter species. Charges, van der Waals interactions, and bonded interactions for Ir(ppy)3 were identical to those used by Tonnelé et al.13 Parameters for the acac ligand were derived from those of the enol tautomer of acetylacetone taken from the Automated Topology Builder (ATB)15 (molid 35990). The 6465


Letter

Nano Letters atom type of the oxygens was reassigned to the atom type “OA” as defined in the GROMOS 54A7 force field.16 Charges for the acac ligand were derived from acetylacetone by discarding the proton and symmetrizing the charges on the remaining atoms. The methyl and CH groups of the acac ligand were represented as united atoms. The charge on the discarded proton was added to the iridium atom to maintain the neutrality of the complex. Bonded interaction parameters for Ir(ppy)2(acac) were assigned by the ATB using the Ir(ppy)2(acac) crystal structure as input.15,17 The distribution of emitter orientations in layers containing either Ir(ppy)3 or Ir(ppy)2(acac) is shown in Figure 3. Here, ϕ

with the plane of the substrate. Thus, the atomic-level morphology of the emission layer can be related to the outcoupling efficiency. The efficiency depends on the proportion of light emitted along the vertical axis relative to the substrate, which originates from the horizontal components of the TDVs, as determined by spectral radiance measurements. The fraction of light emitted from the horizontal components of the TDVs, ΘH, can be expressed in terms of the relative horizontal components of the individual TDVs, pH/p. Here, the distribution of TDV orientations is expressed in terms of the angle between the TDV and the plane of the substrate, ψ. This angle is related to the horizontal component of the TDV by cos ψ = pH/p (Figure 4a). Following from eq 1, the fraction of

Figure 4. Diagram illustrating (a) the angle between the TDV and the substrate, ψ, and (b) the angular offset of the TDV from the Ir−N bonds, δ. The cosine of ψ can be expressed in terms of the horizontal component of the TDV, pH, and the magnitude of the TDV, p, i.e., cos ψ = pH/p.

light emitted from the horizontal components of each TDVi can be expressed in terms of sin ψi, which is used below to describe the TDV orientation distributions:

Figure 3. Distribution of sin ϕ for Ir(ppy)3 (red) and Ir(ppy)2(acac) (blue). Values of sin ϕ = −1, sin ϕ = 0, and sin ϕ = 1 correspond to downward, horizontal, and upward directions of the principal symmetry axis vectors, respectively, relative to the substrate. The dashed line represents an isotropic distribution. Error bars represent the standard errors of the results of 20 independent simulations.

ΘH =

1 N

=1−

is the angle between the xy plane of the substrate and the principle symmetry axis, m [C3 and C2 for Ir(ppy)3 and Ir(ppy)2(acac), respectively]. The dashed line represents the probability density function for an isotropic distribution of the principle symmetry axes of the emitters. When sin ϕ is used as the order parameter, the isotropic distribution corresponds to a constant value of 0.5. The analysis was based on the final frame of each deposition simulation. Both emitters display a net vertical alignment of the principle symmetry axis (sin ϕ ≈ −1 or sin ϕ ≈ 1). The Ir(ppy)3 emitter aligns with either the pyridyl or phenyl face toward the vapor phase with equal likelihood. In contrast, the Ir(ppy)2(acac) emitter preferentially aligns with the acac ligand oriented toward the vapor (sin ϕ ≈ 1) rather than the substrate (sin ϕ ≈ −1). Moon et al. recently used molecular dynamics simulations to characterize the orientation of single Ir(ppy)3 molecules deposited on a surface prepared by annealing a slab of CBP at at 500 K, then 300 K.7 They found no preferential orientation of the symmetry axis of Ir(ppy)3. The alignment of the symmetry axis of Ir(ppy)3 observed in our work therefore suggests that the preferred orientation of the emitter originates from the presence of the surrounding molecules with which it is co-deposited. The preferential alignment of Ir(ppy)2(acac) is consistent with the proposal of Jurow et al.5 With the molecular orientation of the emitters being known, it is possible to analyze the degree of alignment of the TDVs

N

⎛ pHi ⎞2 1 ⎟ = N ⎝ p ⎠

∑⎜ i

1 N

N

∑ cos2 ψi i

N

∑ sin 2 ψi i

(2)

To generate the distribution of TDV orientations from the atomic morphology of the deposited layer, an assumption regarding the relative directions of the TDVs within the emitter molecules is required. Given that the TDV lies approximately in the plane of the phenylpyridine ligand, the relative direction of the TDV can be expressed as an angular offset from the Ir−N bond, denoted δ, as illustrated in Figure 4b. As described above, the value of δ and its dependence on the local environment is not known precisely. Given this uncertainty, the distribution of TDV orientations was calculated assuming an offset of δ = 0°, reflecting molecular orbital calculations showing that the lowest unoccupied molecular orbital is located on the pyridyl moiety, as well as an additional value based on TDDFT calculations of Ir(ppy)3 and Ir(ppy)2(acac) in vacuum. For Ir(ppy)3, an offset of δ = 49° was calculated based on the transition density calculated by TDDFT in ref 10. For Ir(ppy)2(acac), an angle of δ = 16° was used.9 In this analysis, it is assumed that all TDVs within a molecule contribute equally to the total emission. In reality, it is also possible that a particular TDV may be favored due to the symmetry-breaking effect of the environment.12 To determine the importance of this effect would require high-level quantum mechanical calculations to be performed for each individual emitter within its local environment while accounting for thermal motions. 6466


Letter

Nano Letters

at the solid−vapor interface as they are deposited. The alignment induced at the interface is maintained as the emitters are buried providing that the temperature is below the glass transition temperature. The simulations suggest a net vertical alignment of the principal symmetry axes of Ir(ppy)3 and Ir(ppy)2(acac) within blended emission layers. The TDVs of Ir(ppy)3 orient isotropically, despite the net molecular alignment, due to the relative directions of the TDVs within the emitter. In contrast, the net molecular alignment of Ir(ppy)2(acac) leads to a net alignment of the TDVs. Thus, the simulations provide a clear explanation for the higher outcoupling efficiency of OLEDs containing Ir(ppy)2(acac). The work demonstrates the importance of understanding the relative directions of the individual TDVs within the emitter molecules in order to optimize the efficiency of the device. The results suggest that the TDVs of Ir(ppy)3 and Ir(ppy)2(acac) lie closer to the Ir−N bond than indicated by calculations in vacuum. This highlights the need to better understand the influence of the solid-state environment on emission properties. Overall, the work shows that atomistic molecular simulations can be used to mimic the vapor deposition of blended phosphorescent emissive layers containing either homoleptic or heteroleptic complexes. Such deposition simulations also have the potential to provide insight into the impact on OLED efficiency of other aspects of the emission layer morphology. For example, energy losses due to triplet−triplet annihilation depend strongly on emitter aggregation.13,19 Beyond OLEDs, similar simulations could be used to understand the morphology and properties of wide variety of organic thin films, such as those used for photovoltaics, biosensors, and organic electronics.

The distribution of TDV orientations in the Ir(ppy)3 and Ir(ppy)2(acac) layers immediately following the deposition is shown in Figure 5. Assuming an offset δ = 0°, such that the

Figure 5. Distribution of the transition dipole orientation |sin ψ| for Ir(ppy)3 (red) and Ir(ppy)2(acac) (blue) for two choices of TDV angular offset from the Ir−N bond (illustrated above the plot): δ = 0° for both emitters (left) and δ = 49° and 16° for Ir(ppy)3 and Ir(ppy)2(acac), respectively (right). Light emitted along the vertical axis originates from the horizontal components of the TDVs. The dashed line represents an isotropic distribution. Error bars represent the standard errors of the results of 20 independent simulations.

■

TDV lies along the Ir−N bond, the distribution of TDV orientations of the Ir(ppy)3 layer is relatively close to the isotropic case, with the probability density ranging between 1.1 and 0.8 (Figure 5, left). In contrast, the Ir(ppy)2(acac) layer displays a net horizontal TDV alignment (probability density approximately equal to 1.2 and 0.6 when |sin ψ| ≈ 0 and |sin ψ| ≈ 1, respectively). Based on these distributions, the fractions of light emitted from the horizontal component of the TDVs were 3 2(acac) calculated to be ΘIr(ppy) = 0.68 ± 0.01 and ΘIr(ppy) = 0.74 H H ± 0.01. These are in close agreement with the experimental 3 2(acac) measurements ΘIr(ppy) = 0.69 ± 0.02 and ΘIr(ppy) = 0.77 ± H H 2 0.02. Despite the clear preference for the principle symmetry axis of Ir(ppy)3 to align vertically, the orientation of the TDVs is essentially isotropic assuming δ = 0°. This is because the three Ir−N bonds in Ir(ppy)3 are nearly orthogonal, as illustrated in Figure 5 (top-left). Assuming δ = 49°, the TDVs are no longer orthogonal. This results in a small but significant net horizontal alignment of Ir(ppy)3 TDVs, with 3 ΘIr(ppy) = 0.71 ± 0.01 (Figure 5, right). This result is still within H the experimental uncertainty. Conversely, the net horizontal alignment of Ir(ppy)2(acac) TDVs is reduced by using a larger 2(acac) δ. Using δ = 16°, ΘIr(ppy) = 0.72 ± 0.01. This comparison H between the simulations and published experimental results suggests that the TDVs of Ir(ppy)3 and Ir(ppy)2(acac) lie closer to the Ir−N bond than indicated by TDDFT calculations on isolated emitters in vacuum. To understand the effect of the deposition process on the alignment of the emitter molecules, the layers were annealed at 400 K for 10 ns, well above the glass transition temperature of the CBP host (335 K).18 Neither emitter displayed a net alignment after annealing. This indicates that the alignment of these emitters results from the asymmetry in the environment

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]; phone: +61 (0)7 3365 4180. *E-mail: [email protected]; phone: +61 (0)7 3365 3872. ORCID

David M. Huang: 0000-0003-2048-4500 Paul L. Burn: 0000-0003-3405-3517 Alan E. Mark: 0000-0001-5880-4798 Funding

This work was supported by computational resources provided by the Australian Government through the National Computational Infrastructure under the National Computational Merit Allocation Scheme. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge funding from the Australian Research Council (ARC) grant DP150101097. A.E.M. is a University of Queensland Vice-Chancellor’s Research Focused Fellow. P.L.B. is an ARC Laureate Fellow (FL160100067).

■

REFERENCES

(1) Barnes, W. L. J. Mod. Opt. 1998, 45, 661−699. (2) Liehm, P.; Murawski, C.; Furno, M.; Lüssem, B.; Leo, K.; Gather, M. C. Appl. Phys. Lett. 2012, 101, 253304. (3) Graf, A.; Liehm, P.; Murawski, C.; Hofmann, S.; Leo, K.; Gather, M. C. J. Mater. Chem. C 2014, 2, 10298−10304.

6467


Letter

Nano Letters (4) Kim, K.-H.; Lee, S.; Moon, C.-K.; Kim, S.-Y.; Park, Y.-S.; Lee, J.H.; Woo Lee, J.; Huh, J.; You, Y.; Kim, J.-J. Nat. Commun. 2014, 5, 4769. (5) Jurow, M. J.; Mayr, C.; Schmidt, T. D.; Lampe, T.; Djurovich, P. I.; Brütting, W.; Thompson, M. E. Nat. Mater. 2015, 15, 85−91. (6) Flämmich, M.; Frischeisen, J.; Setz, D. S.; Michaelis, D.; Krummacher, B. C.; Schmidt, T. D.; Brütting, W.; Danz, N. Org. Electron. 2011, 12, 1663−1668. (7) Moon, C.-K.; Kim, K.-H.; Kim, J.-J. arXiv:1706.00172. (8) Flämmich, M.; Gather, M. C.; Danz, N.; Michaelis, D.; Bräuer, A. H.; Meerholz, K.; Tünnermann, A. Org. Electron. 2010, 11, 1039− 1046. (9) Heil, A.; Gollnisch, K.; Kleinschmidt, M.; Marian, C. M. Mol. Phys. 2016, 114, 407−422. (10) Gonzalez-Vazquez, J. P.; Burn, P. L.; Powell, B. J. Inorg. Chem. 2015, 54, 10457−10461. (11) Vanhelmont, F. W. M.; Strouse, G. F.; Güdel, H. U.; Stückl, A. C.; Schmalle, H. W. J. Phys. Chem. A 1997, 101, 2946−2952. (12) Hofbeck, T.; Yersin, H. Inorg. Chem. 2010, 49, 9290−9299. (13) Tonnelé, C.; Stroet, M.; Caron, B.; Clulow, A. J.; Nagiri, R. C. R.; Malde, A. K.; Burn, P. L.; Gentle, I. R.; Mark, A. E.; Powell, B. J. Angew. Chem., Int. Ed. 2017, 56, 8402−8406. (14) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. J. Chem. Theory Comput. 2008, 4, 435−447. (15) Malde, A. K.; Zuo, L.; Breeze, M.; Stroet, M.; Poger, D.; Nair, P. C.; Oostenbrink, C.; Mark, A. E. J. Chem. Theory Comput. 2011, 7, 4026−4037. (16) Schmid, N.; Eichenberger, A. P.; Choutko, A.; Riniker, S.; Winger, M.; Mark, A. E.; van Gunsteren, W. F. Eur. Biophys. J. 2011, 40, 843−856. (17) Lamansky, S.; Djurovich, P.; Murphy, D.; Abdel-Razzaq, F.; Kwong, R.; Tsyba, I.; Bortz, M.; Mui, B.; Bau, R.; Thompson, M. E. Inorg. Chem. 2001, 40, 1704−1711. (18) Tsai, M.-H.; Hong, Y.-H.; Chang, C.-H.; Su, H.-C.; Wu, C.-C.; Matoliukstyte, A.; Simokaitiene, J.; Grigalevicius, S.; Grazulevicius, J. V.; Hsu, C.-P. Adv. Mater. 2007, 19, 862−866. (19) Reineke, S.; Baldo, M. A. Phys. Status Solidi A 2012, 209, 2341− 2353.

6468


The Molecular Origin of Anisotropic Emission in an Organic Light

Recommend Documents