SUPPLEMENTARY INFORMATION

Transcription

1 Correction notice White organic light-emitting diodes with fluorescent tube efficiency Sebastian Reineke, Frank Lindner, Gregor Schwartz, Nico Seidler, Karsten Walzer, Björn Lüssem & Karl Leo Nature 459, (2009) In the version of the Supplementary Information originally posted online, three lines of the Figure 1 legend were missing. This has been corrected in the new version of the Supplementary Information; see Supplementary Information Table of Contents for details.

2 Colour stability of white OLEDs as a function of brightness. Figure 1 Exemplarily, the electroluminescence (EL) spectrum of Device LI is shown at various brightness. From 100 to 5,000 cd/m 2, spanning the complete range of application relevant brightness, no significant change in the shape of the spectra is observed. This clearly indicates that the exciton distribution within the EML does not change as a function of current density, as it is responsible for feeding the spatially separated, differently emitting phosphors. This is a result of the optimized EML architecture with reduced energetic barriers. 1

3 Picture of the pyramidal outcoupling structure. Figure 2 A photograph of the high refractive index (glass type: Schott N-LAF 21 with n=1.78) outcoupling structure is shown. The pyramids have a base length of 0.50 mm and a height of 0.25 mm. The insets show close-ups of this structure; bottom: the structure when applied ontop of a yellow test OLED without any index matching fluid. 2

4 Angular emission pattern of selected white OLEDs. Figure 3 Integrated emission intensity as a function of the angle of observation at 1,000 cd/m 2. Here, 90 represents the direction normal to the sample surface. The dashed line shows the calculated distribution, assuming Lambertian emission. Both samples, Device LI and HI-3, show a deviation from the cosine law, which supports the necessity to measure in an integrating sphere. For the device with thick electron transport layer (ETL) (Device HI-3), the emission is more direct. 3

5 Calculated radiance of single emitter devices as a function of electron transport layer (ETL) thickness. Numerical calculations were carried out to obtain a better understanding of the optical effects taking place when moving the ETL thickness for a white device to the second emission maximum. Using the thin film optics simulation software Etfos (version 1.3, Fluxim AG, Winterthur, Swiss), the following layer sequence (i.e. Devices HI-x) is used to model the forward emitted radiance: Layer sequence: Glass (n=1.8) / ITO (90 nm) / MeO-TPD:NDP-2 (45 nm) / NPB (10 nm) / TCTA:Ir(MDQ) 2 (acac) (6 nm) / TCTA (2 nm) / TPBi:FIrpic (4 nm) / TPBi (2 nm) / TPBi:Ir(ppy) 3 / TPBi (10 nm) / Bphen:Cs (x nm) / Ag (100 nm). Input parameters for each layer are optical constants (n (refractive index) and k (absorption coefficient) values), as derived from ellipsometric measurements, and the photoluminescence spectrum of each emitter system. The thickness of the ETL (i.e. Bphen:Cs) is varied from 20 to 250 nm. The results are displayed in Fig. 4. 4

6 Figure 4 Calculated emitted radiance to forward direction as a function of ETL thickness (normalized to the second antinode). While keeping the layer sequence identical, only one emitter is allowed to emit light for each set of data (blue, green and red solid lines). Additional, the product of all components is displayed (black solid line). To investigate the influence of the optical system to each emitting phosphor, three calculations were performed, where each time only one emitter is allowed to emit while the emission of the other two phosphor-doped layers is suppressed. While the first antinode is observed almost at the same thickness for each colour, a difference of roughly 60 nm in ETL thickness can be seen between the second maximum in emission for blue and red. In order to obtain emission from all emitting phosphors, one has to locate the ETL thickness between the optimum for blue and red, which then inevitably closely fulfils the interference condition for green. Switching to thicker ETLs calls for an independent optimization of the complete OLED architecture, as it demands a different internal colour balance in order to improve the chromaticity, which we cannot 5

7 cover in this study. It is apparent from this calculation that is becomes more difficult to reproduce the white emission spectrum, as it is obtained for thin ETLs. Either, the balance between the different phosphors must be newly addressed or, alternatively, it is possible to switch to even higher orders. In this very case, the 5 th and 4 th order maxima of blue and red, respectively, fall together at roughly 580 nm, while green emits about 60 % of its 5 th maximum (630 nm) at this thickness. This, of course causes higher material consumption, which, in turn, reduces the cost-efficiency of organic LEDs. Small efficiency roll-off at high brightness. Generally, the external quantum efficiency (EQE) as a function of current density is used to discuss this roll-off. For that, the critical current density j c is introduced, defining the current density at which the EQE drops to half of its initial value at very low brightness. It is a measure for the overall bimolecular annihilation taking place at high brightness. 1 Taking the consumed power into account, the power efficiency (PE) of a device is the relevant measure in everyday life, which, compared to the EQE, drops faster with increasing current, as it additionally includes ohmic losses. The latter can be drastically reduced with the use of doped transport layers. 2 At 5,000 cd/m 2 and for all devices, the EQE still is about (80±2) % of the efficiency at very low brightness. Due to our novel emitter layer design, the critical current density j c is approximately (100±20) ma/cm 2, which is about one order of magnitude higher than the current needed for 5,000 cd/m 2. As a consequence, still outstandingly high PEs of 74 and 73 lm/w are obtained at 5,000 cd/m 2 for Devices HI-2 and HI-3, respectively, using the pyramidal outcoupling structure. 6

8 Lifetime of white and blue reference devices. To really establish the OLED technology for general lighting, one serious issue is the limited device stability that is not fully solved for the blue emitters. Despite the fact that lifetime investigations of FIrpic-based devices are rare in literature, it is well known that such devices are accompanied with fast degradation. We have measured the lifetime, i.e. the time it takes until the initial luminance is reduced to 50 % of its value, of Device LI and a reference FIrpic-based device with the following structure: Glass (n=1.51) / ITO (90 nm) / MeO-TPD:NDP-2 (25 nm) / NPB (10 nm) / TCTA:FIrpic (5 wt%) (5 nm) / TPBi:FIrpic (15 wt%) (20 nm) / TPBi (10 nm) / Bphen:Cs (50 nm) / Al (100 nm). For both devices, short operation lifetimes between 1-2 hours are obtained at an initial forward luminance of 1,000 cd/m 2. The electroluminescence of Device LI additionally shows a shift in the CIE colour space of Δx = and Δy = , which can be directly appointed to a decrease of the blue FIrpic emission. Alternative materials for p-doped transport layers. As mentioned in the Methods section, we have used NDP-2 as a dopant for the chemically doped hole transport layer (HTL), which we purchased from Novaled AG, Dresden. To assure that these results can be reproduced by interested researchers, we will provide data (cf. Fig. 5) to show that 2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquinodimethane (F 4 -TCNQ) can be used as an freely available alternative. Other studies show that both HTL systems have electically identical properties. 3 In our study, we have used the commercial material since it avoids possible vacuum tool contamination issues. 7

9 Figure 5 This plot compares the current density-voltage characteristics of two devices comprising either NDP-2 or F 4 -TCNQ as p-dopant in the hole transport layer. Both materials have an approximate concentration of 4 mol% in MeO-TPD. The OLED device structure is similar to Device LI. Additionally given are the external quantum efficienies (EQEs) of both devices, which show nearly identical values. 8

10 References 1 Baldo, M. A., Adachi, C. & Forrest, S. R. Transient analysis of organic electrophosphorescence: II. Transient analysis of triplet-triplet annihilation Phys. Rev. B 62, (2000). 2 He, G. et al. High-efficiency and low-voltage p-i-n electrophosphorescent organic light-emitting diodes with double-emission layers Appl. Phys. Lett. 85, (2004). 3 Schwartz, G., Ke, T. H., Wu, C. C., Walzer, K., & Leo, K. Balanced ambipolar charge carrier mobility in mixed layers for application in hybrid white organic lightemitting diodes. Appl. Phys. Lett 93, (2008). 9