Technical Insight
Research Review: Electroluminescence exposes phase separation in AlInGaN
Scientists at West Virginia University have obtained experimental evidence of phase separation in AlInGaN layers with a few percent aluminum and indium.
Their findings contradict earlier theoretical studies by a team of Brazilian researchers that suggested that InAlGaN films – which form the active region in UV LEDs – are random alloys when aluminum and indium concentrations are very low.
Xian-An Cao and Yi Yang identified nanoscale phase separation in the AlInGaN material system through a series of electroluminescence measurements.
This duo studied an MOCVD-grown LED that comprised a sapphire substrate, a 3 μm-thick AlGaN template, a silicon-doped Al0.15Ga0.85N cladding layer, an active region with 3 nm thick Al0.06In0.02Ga0.92N quantum wells and 10 nm-thick Al0.1Ga0.9N barriers, and magnesium-doped Al0.25Ga0.75N cladding and GaN capping layers.
Electroluminescence measurements were recorded at temperatures ranging from 5K to 300K, and various drive currents. These studies revealed that emission is dominated by peaks at 3.47 eV and 3.59 eV at low temperatures and currents. Heat up the device, or driver it harder, and the electroluminescence profile switches to a single peak at 3.39 eV. Cao and Yang claim that these results indicate that the active region is composed of GaN and aluminumrich and indium-rich nanoclusters (see figure). Indium-rich quantum dot-like clusters form potential wells that are 0.2 eV deep, which are each surrounded by an aluminumrich region that acts as a raised rim.
At low temperatures and low currents luminescence from the GaN and AlGaN phases dominates, because most injected carriers fall into the energy bands of GaN and AlGaN. Emission from InGaN is negligible, because this phase accounts for just a fraction of the active region. Either increasing the temperature of the LED beyond 150K or cranking up the current leads to the injection of carriers that overcome the energy associated with the aluminum-rich rim and reach the InGaN phases. There they recombine radiatively. This switch in the distribution of carriers accounts for the red-shift from 3.47 eV and 3.59 eV emission to a peak at 3.39 eV.
The phase separation in ultraviolet LEDs that Cao and Yang have uncovered is undesirable, because it reduces the device’s internal quantum efficiency. But this can be suppressed through strain reduction and optimization of the growth recipe.
“We plan to work with LED growers to improve the performance of UV LEDs based on quaternary alloys,” says Cao. “Two specific tasks are: to design and grow lattice matched templates and heterostructures by tuning quaternary compositions; and determining the optimal growth temperature to allow for aluminum/indium incorporation, while maintaining good structural quality.”
X. A. Cao et al. Appl. Phys .Lett. 96 151109 (2010)
