Improving Light Extraction Efficiency of InGaN/GaN LEDs Using a Dip‑Drop Fabricated Polystyrene Nanosphere Array

Background

Photonic crystals (PCs) are routinely employed to enhance the performance of optoelectronic devices such as LEDs, solar cells, and photodetectors. PCs create a periodic refractive index modulation at the wavelength scale, which can diffract guided modes into radiative channels. While high‑index contrast PCs can open a full bandgap, practical implementation in LEDs is limited by fabrication complexity and material constraints. An alternative strategy is to use the PC’s periodicity to shift the in‑plane wave vector of guided light into the escape cone, thereby increasing the light extraction efficiency (LEE).

GaN‑based LEDs suffer from a severe light‑trapping problem because the critical angle for total internal reflection (TIR) at the GaN/air interface is only ~23°. Consequently, the external quantum efficiency (EQE) is dominated by LEE. Numerous approaches—including textured sapphire substrates, patterned GaN surfaces, and side‑wall roughening—have been explored, but each faces practical challenges such as chemical damage or limited scalability.

Self‑assembled PS NS coatings offer a promising, low‑cost alternative. They can be deposited over large areas with minimal equipment and do not require high‑temperature processing or aggressive etching. In this work, we systematically study the dip‑drop deposition of PS NS monolayers on ITO, optimize the suspension concentration and dip‑drop speed, and demonstrate a 38 % LEE improvement in InGaN/GaN LEDs.

Experimental

Dip‑Drop Method

The dip‑drop setup consists of a glass reservoir with a controllable outlet valve. A deionized water–PS NS colloidal mixture is stirred to achieve a target concentration, then the ITO‑coated substrate is lowered through the liquid at a prescribed speed. Typical PS NS sizes used were 100, 200, and 500 nm, diluted to concentrations of 4.1×10¹¹, 5.1×10¹⁰, and 3.2×10⁹ spheres cm⁻³, respectively.

After deposition, the substrates are dried at room temperature for ~1.5 h. Oxygen plasma treatment (0–10 s) is employed to render the surface hydrophilic, optimizing PS NS adhesion while avoiding excessive ITO damage.

LED Fabrication

InGaN/GaN blue LEDs were grown on c‑plane sapphire by MOCVD. The structure includes a low‑temperature GaN buffer, heavily Si‑doped n‑GaN, InGaN/GaN MQWs, and Mg‑doped p‑GaN. ITO was deposited on the p‑GaN layer as a transparent conductor. Mesa isolation (300 × 300 µm²) was defined by photolithography, and Ti/Pt/Au contacts were alloyed at 450 °C for 5 min. The completed wafers were then dipped in the PS NS suspension to form the monolayer array.

Results and Discussion

Scanning electron microscopy (SEM) revealed that the PS NS arrangement is highly sensitive to dip‑drop speed. At high speeds (0.05 mL s⁻¹), the particles are dispersed; reducing the speed to 0.01 mL s⁻¹ yields a loosely packed layer, while 0.005 mL s⁻¹ produces a tightly packed monolayer. Larger spheres (200–500 nm) self‑assemble more readily than 100‑nm particles at the same speed due to stronger capillary forces.

Suspension concentration also governs coverage. Concentrations below 4.1×10¹¹ spheres cm⁻³ for 100‑nm PS NSs leave gaps; above 5.4×10¹¹ spheres cm⁻³ the monolayer turns into multilayers, which reduce transmittance. Thus, the optimal conditions for a compact monolayer were 0.005 mL s⁻¹ and 4.1×10¹¹ spheres cm⁻³ (100‑nm), 5.1×10¹⁰ spheres cm⁻³ (200‑nm), and 3.2×10⁹ spheres cm⁻³ (500‑nm).

Finite‑difference time‑domain (FDTD) simulations show that a 100‑nm PS NS monolayer with 100‑nm periods in both directions offers the highest extracted light intensity, improving the simulated LEE by ~40 % relative to a bare LED. The enhancement arises from the diffractive shift of the in‑plane wave vector into the escape cone.

Experimentally, LEDs with the optimized PS NS window exhibited a 38 % increase in light output at 20 mA (from 112.9 mcd to 148.0 mcd). I‑V characteristics remained unchanged, confirming that the deposition process does not alter the electrical behavior. LEDs with disordered or multilayer PS NS arrays performed only slightly better or worse than the bare device, underscoring the importance of a regular monolayer.

Uniformity across the wafer was excellent: the standard deviation of the LEE enhancement was 1.4 %, and variations in forward voltage and light output were below 3 %. Electroluminescence spectra showed a stronger, narrower peak at 465.5 nm for the PS NS‑enhanced LEDs, reflecting reduced re‑absorption and improved out‑coupling.

Conclusion

We have established a simple, scalable dip‑drop protocol to fabricate a periodic PS NS monolayer on ITO, serving as an effective photonic crystal window for InGaN/GaN LEDs. Optimizing dip‑drop speed (0.005 mL s⁻¹) and suspension concentration (4.1×10¹¹ spheres cm⁻³ for 100‑nm PS NSs) yields a 100‑nm array with 100‑nm periods that increases the LEE by 38 % at 20 mA. This method is compatible with standard LED processes, requires no high‑temperature steps, and is suitable for large‑area production.

Abbreviations

EQE

External quantum efficiency

FDTD

Finite‑difference time‑domain

ITO

Indium‑tin‑oxide

I‑V

Current‑voltage

LED

Light‑emitting diode

LEE

Light extraction efficiency

L‑I

Light output intensity‑current

PCs

Photonic crystals

PS NS

Polystyrene nanosphere

SEM

Scanning electron microscopy

TIR

Total internal reflection