We investigate two types of internal light-extraction layer structures for organic light-emitting diodes (OLEDs) that consist of silica nanoparticles (NPs) embedded in high-refractive-index TiO2 matrices. The composite of silica NPs and TiO2 matrices was coated on the glass substrate and fabricated with and without a SiO2 planarization layer. An increase in the optical out-coupling efficiency by a factor of 2.0 was obtained at a high luminance of 3,000 cd/m2 from OLEDs containing the silica NPs embedded in TiO2 matrices between glass substrates and Zn-doped In2O3 (IZO) electrodes after additional planarization processes. This is consistent with the analytical result using the finite-difference time-domain (FDTD) method. Randomly distributed silica NPs acting as scattering centers could reduce the optical loss when extracting light. By using additional planarization processes with a PECVD-derived SiO2 layer, one can assure that smoother surfaces provide higher out-coupling efficiency, which attain 100% and 97% enhancements in power (lm/W) and current (cd/A) efficiencies, respectively.
© 2014 Optical Society of America
Organic light-emitting diodes (OLEDs) have attracted a great deal of scientific interest due to their unique characteristics. OLEDs are one of the most promising devices for lighting and display applications . Though they have countless benefits such as high contrast ratio, high CRI (color rendering index), flexibility, and low power consumption, OLEDs have a fatal shortcoming of low out-coupling efficiency . Typically, the device out-coupling efficiency of OLEDs cannot be made greater than 20% due to the losses from the glass mode (30%) and the wave-guided mode (50%).
In order to improve the light-extraction efficiency of OLEDs, internal and/or external light out-coupling layers are introduced, where these layers are respectively placed between the transparent conducting oxide (TCO) and glass substrate and on the substrate. Since the external light out-coupling layer can be attached physically to the substrate without inhibiting device performance, research on the various external layers has led to development of commercial products. On the other hand, the internal light out-coupling layer can affect the change of current movement, so that could pose challenges to achieving the high power efficiency and luminance, which are properties desirable for mass production of devices with such out-coupling layers. However, since internal light-extraction layer surfaces that are rough or wavy can be advantageous for preventing the lateral propagation of light in the wave-guided mode, these types of layers have been the subjects of frequent research. However, rough or wavy surfaces can deteriorate device performance and shorten lifetimes because of the accelerated aging, which is not a suitable feature for mass production. Bocksrocker and his associates  fabricated a WOLED on top of a monolayer of SiO2 microspheres, and Koo et al.  showed that defective hexagonal close-packed SiO2 spheres could be effective as an internal out-coupling layer. There has been a noticeable improvement in ordered arrays, that is, “photonic crystals.” [5–8] Although their use has been reported to increase light out-coupling efficiency by a factor of 1.5−2.1, OLEDs using photonic crystals show an emissive spectrum changes. Among the various kinds of light-extraction technologies using NPs, high-refractive index NPs such as TiO2 surrounded with low-refractive index polymer matrices have exhibited an outstanding enhancement in light out-coupling efficiency by a factor of 2.04 . Therefore, OLEDs using low-index metal oxide nanoparticles with high-index metal oxide matrices not only can improve light extraction, but also can be free from deformation and discoloration typically when exposed at high temperatures during the device fabrication process. Despite its notable durability and reliability, research on the internal light-extraction technique that utilize high index metal oxide matrices embedded with low index metal oxide nanoparticles in OLED devices has not been reported yet.
In this study, we have used silica NPs and prepared TiO2 sol-gel solutions to develop a highly efficient light-extraction layer composed of only metal oxides as a final product that can make the device fabrication process and the device lifetime more robust to variations in the device formulation. We demonstrated the effect of the composite of silica and TiO2 matrices located in between the glass substrate and the anode on light out-coupling, based on experimental and theoretical results.
2. Experimental setup
A colloidal mixture of NPs (20 vol%) that was dispersed into ethanol, ethylene glycol, and poly (4-vinyl phenol) was first used to coat the glass substrate. After the first layer was prepared, TiO2 sol-gel solutions diluted with ethanol were successively over-coated to fill the gaps between silica particles and, thus, silica NPs (refractive index of 1.45) were placed directly on the glass substrate, resulting in low surface roughness. Each coating mentioned above was prepared by drying for 10 min at 100°C and annealing for 30 min at 500°C. Two different samples were prepared to secure the surface roughness by post-processing such as chemical-mechanical planarization (CMP) and deposition of SiO2 layer (refractive index of 1.45) using plasma-enhanced chemical vapor deposition (PECVD). The TiO2 layer (refractive index of 2.4) with embedded SiO2 for device B contain 50% more SiO2 content (15 vol% in the composite of SiO2 and TiO2) than that of device A (10 vol%). The SiO2 planarization layer deposited by PECVD and CMP were applied only to device B. For specifying optical characteristics, the total and diffuse transmittance of the scattering layer is measured by spectrophotometer (PerkinElmer Lambda 950) and atomic force microscopy (AFM)
To fabricate the OLEDs, the following layers were thermally deposited on a 150-nm-thick IZO anode; A 10-nm-thick hole injection layer (HIL) mixed with HTL as ratio of 1:1 followed by a 80-nm-thick HATCN as a hole transport layer (HTL) followed by a 40-nm-thick emitting layer (EML) with 10 wt% Iridium derivative doped green phosphorescent followed by a 25-nm-thick electron transport layer (ETL) mixed with EIL as ratio of 1:1 followed by a 1-nm-thick electron injection layer (EIL) and finally a 100-nm-thick aluminum cathode. And then, electrical characteristics were recorded with a source meter (Keithley 2400) and structural characteristics were measured using transmission electron microscopy (TEM) at 200 keV and scanning electron microscopy (SEM) at 15 kV. The CMP process after deposition of SiO2 provides a flat surface for comparison with the surface as coated. For reference, conventional OLED structures without a light-scattering layer were also prepared on the same glass substrates. The I–V–L characteristics of the fabricated OLED devices were measured under normal atmosphere using a spectroradiometer (Konica Minolta CS1000A) in the dark box. For the theoretical analysis, three-dimensional FDTD calculations were performed using commercial software (Lumerical FDTD Solutions 8.0). The simulation space has a volume of15 × 15 × 5 μm3 and the perfectly matched layer (PML) boundary condition was applied to all the simulation domain boundaries. A dipole emitter was positioned in the middle of the EML layer and a flux monitor was set near the glass/air interface to detect light intensity coming out from the OLED structure through the glass substrate, finally calculating enhancement ratio. Assuming emitter molecules are isotropically distributed in the EML layer, each emitter case with three different dipole orientations were independently calculated and their results were then averaged out.
3. Enhancement of light-extraction efficiency
The schematic drawings for two different kinds of samples (Fig. 1) and their optical and surface characteristics are summarized in Table 1.Deposition of 1.5-μm SiO2 PECVD decreases the Rpv (peak-to-valley roughness) of 213 nm to 53 nm and Ra (average roughness) of 28 nm to 21.6 nm, and CMP flattens the SiO2-deposited surface to an Rpv of 15.3 nm. As-coated layer (Device A without OLED stack) shows a total and diffuse transmittance of 80.5 and 5.2% while post-processed layer (Device B without OLED stack) shows 76.9 and 15.6%, respectively. Though the diffuse transmittance (Tdiffuse) or haze is not directly proportional to light-extraction enhancement ratio, the measurement of the total and diffuse transmittance is very important step since a high degree of scattering is very significant for more light-extraction enhancement [10,11].
The cross-sectional images of devices A and B including the internal light out-coupling layers, as imaged by TEM, are shown in Fig. 2.Figures 2(a)–2(f) depict the details of structural characteristics for devices A and B such as the stacks of NPs, TiO2 matrix, air voids near silica particles, and deposited SiO2 planarization layer (for device B). The interfacial structure between the Zn-doped In2O3 (IZO) anode and the TiO2 matrix is shown in Fig. 2(b). Since the surface of the layer is not smooth enough, the interface between two layers is quite blurry. On the other hand, one can confirm the clear interface between the IZO and the PECVD-deposited SiO2 layer of device B, as shown in Fig. 2(e), owing to the CMP-processed flat surface. The difference in the interface characteristics could influence the current movement and thus possibly the efficiency enhancement. Among the various layers, the most outstanding region is the TiO2 matrix layer with embedded silica. To eliminate the angular dependence of the extracted light spectrum due to the ordered array of NPs, silica NPs are intentionally distributed randomly in the layer for devices A and B, and then the TiO2 matrix layer is over-coated to generate both a considerable difference in refractive index and a smooth surface. However, coating the TiO2 matrix layer once in device B, rather than twice in device A, causes device B to develop a bumpier surface than device A. The coatings consisting of a mixture of silica NPs with TiO2 sol-gel solutions contain unexpected voids of various sizes and shapes that can play the role of light-scattering centers with a refractive index of 1.0, as shown in Figs. 2(c) and 2(f). As shown in Fig. 3, air voids are observed over a large area, and one can clearly see that device B has more silica and air voids contents thandevice A. The shapes of the voids are not uniform, and the voids near the silica have large volume. These larger voids might have occurred due to the difference of the thermal expansion coefficient during the drying and annealing process of the light-extraction layers. Some of the voids seem to be interconnected while others are isolated. At first glance, it appears that the void formation resulted from the proximity of silica NPs since the voids are observed mainly in the space between two silica NPs. However, meticulous observation helps us find small voids that exist without the presence of nearby silica NPs. It should be noted that device B includes more voids than device A, and the formation of more voids is likely to be caused by the higher SiO2 content in device B, though the total volume of voids is not exactly proportional to the SiO2 concentration in the layer.
Figure 4 represents the J–V characteristics of the reference OLEDs and the two devices. Device A represents a higher leakage current level than device B since the surface of the internal light-extraction layer of device A is rougher. On the other hand, device B and the two reference devices show quite low current leakage, probably due to the CMP process and the flat surfaces of the glass substrates, respectively. The correlation between the current leakage and ITO roughness has been reported by various researchers. Gil et al.  and Fukushi et al.  have reported that the surface state of ITO directly influences the OLED current leakage and lifetime, respectively. Specifically, Tak et al.  focused on the types of surface roughness that cause unacceptably high current leakage of OLED, and thus reported that the Rpv of the substrate is the most crucial factor affecting the device performance, when compared with Rrms (root mean square roughness) and Ra (average roughness).
The typical luminance (cd/m2), power (lm/W), and current efficiency (cd/A) are plotted as functions of current density for devices A and B and their reference devices (Ref. A and Ref. B). In Figs. 5(a) and 5(b), one can confirm the specific current densities (6.8, 6.5, 3.1, and 6.2 mA/cm2 for device A, Ref. A, device B, and Ref. B, respectively) that show a luminance of 3,000 cd/m2. Device A and Ref. A show similar current-density–luminance curves up to 3,000 cd/m2, and then device A represents a luminance that is a bit higher than that of Ref. A. On the other hand, the current density of Ref. B is twice as high as that of device B throughout the whole current-density range. This means that device B attains the same luminance while only consuming half the current. In Fig. 5(c), the shapes of the power and current efficiency curves of device A are quite different from those of Ref. A, and thus device A has a current leakage in the range of low voltages (0−3 V). Specifically, the power and current efficiencies for device A are 44 lm/W and 55.2 cd/A at an applied current density of 6.8 mA/cm2, respectively, while those for Ref. A are 28 lm/W and 42.4 cd/A at 6.5 mA/cm2, respectively. The power and current efficiencies for device B are 78 lm/W and 94.8 cd/A at 3.1 mA/cm2, respectively, whereas those for Ref. B are 39 lm/W and 48.0 cd/A at a current density of 6.2 mA/cm2, respectively, as depicted in Fig. 5(d). Device B shows 100% and 97% enhancements in the power and current efficiencies compared to Ref. B, and the enhancement ratios as a function of applied current density are depicted in Figs. 5(e) and 5(f). Photographs of the OLEDs emitting light without and with light-scattering layers are shown in the insets of Fig. 5(f). One can clearly observe the wave-guided light at the edges of the glass substrate for Ref. B (left), and a considerable portion of light is significantly extracted from the substrate due to the internal light-scattering layer, though the optical loss is not completely eliminated (right).
As seen in Fig. 6, the angular distributions of emitted light for devices A and B similarly exhibit Lambertian emission, but represent slightly different angular emission patterns from each of the references. Each reference has a relatively low intensity at the normal angle and a high intensity at around ± 45°. This indicates that the light-scattering composite of silica NPs and the TiO2 matrix provides more uniform angular distribution due to the random scattering of light. However, the tendency based on the viewing angle would be discernible, and a light-scattering layer decreases the angular dependence of the emitted light. From the point of view of measured angles, these measurements were made at −50°− + 50° because the light-emitting area is small (2 mm × 2 mm), and then the enhancement ratio of each reference was normalized to 1.0.
The power efficiencies of the fabricated devices in this study and of the reference devices as well as the simulated and experimentally measured enhancement values are summarized in Table 2.There are minor discrepancies from the power efficiencies of the reference device since each device was fabricated from a different batch. The power efficiencies in Table 2 were each measured at an applied voltage that supplies a luminance of 3,000 cd/m2 for the baseline. Device B shows a higher enhancement in power efficiency than device A, which is not consistent with the FDTD simulation results. The FDTD simulation of individual devices based on the model structures, reflecting the surface status and the distribution and extent of the particles as well as the voids, estimates that devices A and B extract, at maximum, 136% and 97% more light than the reference, respectively. These findings may indicate that the addition of a low-refractive-index SiO2 planarization layer in device B could be detrimental to the light extraction, even though a higher volume fraction of silica and air voids as well as rough surfaces of the top layer beneath the IZO could increase the efficiency.
Figures 7, 8 , and 9present the results of numerical studies on the effect of bead fraction, a low index planarization layer, surface roughness, and voids on light extraction. For the peak-to-valley surface roughness of TiO2 layers shown in Figs. 7(a) and 7(c), the AFM measurements of 213 nm and 200 nm is used as an input value for the FDTD calculations, respectively. The light-extraction efficiencies of OLED devices with and without low-refractive index SiO2 planarization layer are simulated in Fig. 7. The insertion of a low-refractive index SiO2 planarization layer reduces the light-extraction efficiency significantly, as expected. It is not also surprising that device A containing a higher volume fraction of silica beads in the TiO2 matrix would further increase the light extraction, as demonstrated in Fig. 8. That figure shows a light-extraction-enhancement contour map as a function of silica vol% and the refractive index of the matrix, as determined from FDTD analysis. In the simulated structure, single-sized silica NPs are immersed into a TiO2 matrix, but an additional SiO2 layer for planarization on top of the TiO2 matrix is not considered for this simple case. The FDTD analysis sets the light-extraction layer thickness and the diameter of the silica NPs to 600 nm and 200 nm, respectively. The absorption coefficients of the emitting layers are ignored for all cases. According to the results, using a high-refractive-index matrix such as anatase TiO2 with 30 vol% of silica will lead to a maximum in light out-coupling enhancement. The effects of surface roughness and voids on the light-extraction enhancement were additionally investigated, as indicated by the simulated results shown in Fig. 7 and Fig. 9, respectively. From Fig. 9, one may believe that the high void content up to nearly 9 vol% is modestly advantageous for light extraction, though the void content is difficult to be precisely controlled. For example, approximately 3 vol% of air voids estimated to exist in Device B can provide more enhancement of light extraction up to 10% than 1 vol% of voids. However, air voids over 9 vol% may deteriorate the light-extraction enhancement effect, probably due to the decrease in the effective refractive index of the TiO2 matrix containing the large quantity of air voids. In addition to that, a rough TiO2 matrix surface (Fig. 7(c)) shows light out-coupling boosts in comparison with a smooth surface (Fig. 7(b)). With all the same boundary conditions except for the rough TiO2 surface having an Rpv of 213 nm, the model structure of Figs. 7(a) and 7(b) shows light out-coupling enhancements of 63% and 97%, respectively. In contrast with these numerical estimates, the relatively low measured enhancement of the light-extraction efficiency of device A, as seen in Fig. 4 and Fig. 5, is probably attributed to high current leakage. Higher current leakage of the device would lead to a decrease in the current that is convertible to light and thus to a reduction in light-extraction efficiency. Therefore, one can conclude that smooth, flat surfaces not generating current leakage would be beneficial not merely for device durability but also for device efficiency.
In summary, variations of the structure of a light-extraction layer based on embedding silica NPs in a TiO2 matrix, using different post-processing approaches, are investigated. By using silica NPs in a high-refractive-index TiO2 matrix and an additional planarization layer, 100% enhancement of the out-coupling power efficiency, which is a fairly outstanding result, was achieved. Additionally, the angular dependence based on the input wavelength is improved by randomly distributing the silica NPs. Moreover, one should note that the fabricated OLEDs in this study show very high luminance and efficiency. It is demonstrated that the efficiency enhancement is associated with the volumetric ratio of light-scattering centers (silica particles and voids) and surface roughness of the layer beneath the IZO. The current leakage is also likely to play a significant role in the device performance as well as in the device lifetime. Consequently, a smooth internal light-extraction layer composed of a high-index matrix embedded with a high volume fraction of low-index light scatterers would be desirable for enhancement in light-extraction efficiency and could provide practical solutions for numerous applications such as OLED displays and lighting.
Authors acknowledge Dr. Sangyoon Lee and Ho-Suk Kang in Emerging Material Research Center at SAIT (Samsung Advanced Institute of Technology) for their support for the device fabrication.
References and links
1. C.-H. Chang, H.-C. Cheng, Y.-J. Lu, K.-C. Tien, H.-W. Lin, C.-L. Lin, C.-J. Yang, and C.-C. Wu, “Enhancing color gamut of white OLED displays by using microcavity green pixels,” Org. Electron. 11(2), 247–254 (2010). [CrossRef]
2. C. W. Tang and S. A. VanSlyke, “Organic electroluminescent diodes,” Appl. Phys. Lett. 51(12), 913–915 (1987). [CrossRef]
3. T. Bocksrocker, J. Hoffmann, C. Eschenbaum, A. Pargner, J. Preinfalk, F. Maier-Flaig, and U. Lemmer, “Micro-spherically textured organic light emitting diodes: A simple way towards highly increased light extraction,” Org. Electron. 14(1), 396–401 (2013). [CrossRef]
4. W. H. Koo, W. Youn, P. Zhu, X.-H. Li, N. Tansu, and F. So, “Light extraction of organic light emitting diodes by defective hexagonal-close-packed array,” Adv. Funct. Mater. 22(16), 3454–3459 (2012). [CrossRef]
5. M. Fujita, K. Ishihara, T. Ueno, T. Asano, S. Noda, H. Ohata, T. Tsuji, H. Nakada, and N. Shimoji, “Optical and electrical characteristics of organic light-emitting diodes with two-dimensional photonic crystals in organic/electrode layers,” Jpn. J. Appl. Phys. 44(6A), 3669–3677 (2005). [CrossRef]
6. Y. R. Do, Y.-C. Kim, Y.-W. Song, and Y.-H. Lee, “Enhanced light extraction efficiency from organic light emitting diodes by insertion of a two-dimensional photonic crystal structure,” J. Appl. Phys. 96(12), 7629–7636 (2004). [CrossRef]
7. Y. R. Do, Y.-C. Kim, Y.-W. Song, C.-O. Cho, J. Jeon, Y.-J. Lee, S.-H. Kim, and Y.-H. Lee, “Enhanced light extraction from organic light-emitting diodes with 2D SiO2/SiNx photonic crystals,” Adv. Mater. 15(14), 1214–1218 (2003). [CrossRef]
8. K. Ishihara, M. Fujita, I. Matsubara, T. Asano, S. Noda, H. Ohata, A. Hirasawa, H. Nakada, and N. Shimoji, “Organic light-emitting diodes with photonic crystals on glass substrate fabricated by nanoimprint lithography,” Appl. Phys. Lett. 90(11), 111114 (2007). [CrossRef]
9. H.-W. Chang, K.-C. Tien, M.-H. Hsu, Y.-H. Huang, and C.-C. Wu, “OLEDs integrated with internal scattering structure for enhancing optical outcoupling,” SID Int. Symp. Dig. Tec. 41(1), 50–53 (2010).
10. H.-W. Chang, J. Lee, T.-W. Koh, S. Hofmann, B. Lüssem, S. Yoo, C.-C. Wu, K. Leo, and M. C. Gather, “Bi-directional organic light-emitting diodes with nanoparticle-enhanced light outcoupling,” Laser Photonics Rev. 7(6), 1079–1087 (2013). [CrossRef]
11. Y. H. Kim, J. Lee, W. M. Kim, C. Fuchs, S. Hofmann, H.-W. Chang, M. C. Gather, L. Müller-Meskamp, and K. Leo, “We want our photons back: simple nanostructures for white organic light-emitting diode outcoupling,” Adv. Funct. Mater.in press.
12. T. H. Gil, C. May, S. Scholz, S. Franke, M. Toerker, H. Lakner, K. Leo, and S. Keller, “Origin of damages in OLED from Al top electrode deposition by DC magnetron sputtering,” Org. Electron. 11(2), 322–331 (2010). [CrossRef]
13. Y. Fukushi, H. Kominami, Y. Nakanishi, and Y. Hatanaka, “Effect of ITO surface state to the aging characteristics of thin film OLED,” Appl. Surf. Sci. 244(1–4), 537–540 (2005). [CrossRef]
14. Y.-H. Tak, K.-B. Kim, H.-G. Park, K.-H. Lee, and J.-R. Lee, “Criteria for ITO (indium-tin-oxide) thin film as the bottom electrode of an organic light emitting diode,” Thin Solid Films 411(1), 12–16 (2002). [CrossRef]