The overall process flow used to form the polymeric micro and nano complex structure on LED device is described in Figure 1. Details of the fabrication of the LED devices have been shown elsewhere 18. As shown in Figure 1, a micro and nano complex polymer structure was formed on the completed LED structure, which has n and p electrodes and a mesa structure using the nanoimprint lithography process. Since a flexible PDMS stamp was used as an imprint template, a micro- and nanoscale complex polymer structure was uniformly formed on the LED despite the step height between the n- and p-GaN regions. The process details of PDMS replication are available elsewhere 19. The polymer structure on the electrodes of the p- and n-GaN layers was selectively removed by photolithography and RIE. GXR601, which purchased from AZ Electronic Materials (Stockley Park, UK), was used as a positive photo-resist. Since only metal electrodes were exposed and the ITO surface was not exposed to the plasma, plasma-induced damage to the LED device was avoided.

The complicated micro- and nanoscale structure originated from the photochemical etched N-face GaN epitaxial layer. The GaN layer was epitaxially grown on (0001) sapphire substrate with a thickness of a few micrometers and was lifted off using a laser. The N-faced GaN surface was then exposed and etched with 5 M KOH solution at 60°C. To enhance the etching, the solution was continuously stirred and ultraviolet light was illuminated simultaneously onto the surface during the etching process 17.

A high refractive index resin containing TiO₂ nanoparticles was purchased from Brewer Science Inc. (Rolla, MO, USA) and used as an imprint resin to form the micro- and nanoscale complex structures since its n and k values are 2.08 and 0.004, respectively, at 450 nm, the blue LED emission wavelength. The transmittance of the high refractive index is > 90% at the blue LED emission wavelength 20.

The morphologies of micro- and nanoscale complex structures of the high refractive index polymer were analyzed by scanning electron microscopy (SEM) and atomic force microscopy (AFM). The effect of the micro- and nanoscale complex structure on the enhancement of the LED light extraction efficiency was confirmed by electroluminescence measurement. The electrical properties of the LED devices were measured using current-voltage (I-V) characteristics.

The AFM analysis was performed to determine the morphology and height of the micro- and nanoscale complex structure. The AFM images of the micro- and nanoscale complex structure that formed on the N-face n-GaN surface, replicated polymer mold, and LED device are shown is Figure 2a,c, respectively. Figure 2 clearly shows that a continuous array of micro- and nanoscale structures was formed on the LED devices with high fidelity. According to AFM analysis, the height of the micro- and nanoscale complex polymer pattern ranged from 0.18 to 1.2 μm.

Since the actual LED device has a mesa structure, controlling the residual layer was extremely difficult. The residual polymer layer can have a detrimental effect on the transparency of the ITO layer of the GaN LED, so the spin-coating speed of the high refractive index polymer resin was carefully adjusted. As shown in Figure 3a,c, in case of lower spin-coating speeds, high fidelity pattern transfer was achieved and the thickness of the residual layer was also quite high. In cases of higher spin-coating speed, the micro- and nanoscale complex structure was not completely transferred due to the lack of an imprint resin. However, the thickness of the residual layer was drastically decreased. In this study, residual layer thickness control was accomplished by spin-coating speed adjustment rather than by the RIE. With a spin-coating speed of 5,000 rpm, micro- and nanoscale pattern transfer was achieved with high fidelity and the residual layer thickness was minimized.

In order to investigate the effect of the micro- and nanoscale polymer structures on the LED photon extraction efficiency, we chose two identical GaN-based blue LED devices that were grown on PSS and formed the micro- and nanoscale complex polymer structure on one wafer using the nanoimprint lithography process. The cross-sectional SEM micrographs of the LED structure with the micro- and nanoscale complex polymer structure on the PSS are shown in Figure 4a. Figure 4b, c show that the micro- and nanoscale complex polymer structures formed on the metal electrode were clearly removed via the RIE etching process to ensure stable current injection.

We measured the electroluminescence (EL) intensities at 20 mA of injection current LED devices with and without micro- and nanoscale complex polymer structures. PSS were used for both LED devices. The EL measurement was taken from one randomly selected device, measured ten times, and then averaged. As shown in Figure 5a, the EL emission of the LED structure with micro- and nanoscale complex polymer structures increased up to 13%. This increase in photon extraction efficiency is additional; thus, it is very meaningful since the LED structure was grown on a PSS wafer to increase the photon extraction efficiency up to 30% 1112. The microscale surface protrusion pattern on the PSS wafer already compensated the light direction to make it fit inside the escape cone, thus significant light extraction efficiency enhancement of the device was reported. Micro- and nanoscale complex polymer structures allow the photons to be extracted out of the LED structure via the photon scattering effect. In addition, the EL was measured in the other direction (angle) at a 60° tilt as shown in Figure 5c. In this case, similar EL emission intensity enhancement was observed compared with normal EL measurement.

In order to confirm the effect of the nanoimprint patterning process on the electrical performance of the LED devices, I-V measurements were performed for the LED devices, including those with the micro- and nanoscale complex structures. In all cases, identical I-V characteristics were observed and the turn-on voltage and leakage current levels of the LED devices remained unchanged as shown in Figure 6. This finding implies that no electrical degradation was induced by the patterning process.