News Article

Enabling High-efficiency MicroLEDs


Sidewall treatments empower microLEDs with size-independent efficiency By Matthew Wong, James Speck, Shuji Nakamura and Steven DenBaars from the University of California, Santa Barbara

The commercialization of smartphones and portable electronics has piqued academic and industrial interest in developing better display technologies. It's a sector that is moving fast, progressing from cathode-ray tubes to liquid crystal displays (LCDs) and more recently organic LEDs (OLEDs), with latter already employed extensively in wearables. However, emerging display applications - namely near-eye displays for virtual-reality, augmented-reality and head-up displays - are demanding even better display performances, such as a superior colour gamut and a greater resolution. Efforts at fulfilling these requirements are driving the rapid development of new display technologies.

Among the various candidates for next-generation displays, microLEDs are the most promising, offering a variety of benefits that include a high brightness, a long operating lifetime, excellent stability and outstanding efficiency characteristics. While there is no official cut-off for the device dimensions of the microLED, with some use the term ‘miniLEDs' to refer to bigger microLED, it is reasonable to assume that the dimensions of this device are less than 100 m. The actual value matters a great deal, as it determines the display resolution and the manufacturing cost.

To produce a full-colour display, designers tend to employ red, green, and blue colours that match mankind's visual receptors. So when making a display, designers want to employ microLEDs with these three emission wavelengths. The emitting chips are made from either the InGaN or AlGaInP families of materials. Although InGaN-based LEDs are capable of spanning the entire visible spectrum, red InGaN LEDs are currently limited to the active research stage and not commercially available. Hence, one option for realizing a full-colour microLED display is to employ InGaN microLEDs for the blue and green and AlGaInP variants for the red.

One of key advantages of the microLED display is that it draws on the exceptional performance of microLEDs. In this form of display, every red, green and blue subpixel is modulated independently. This equips the display with a simple architecture that is capable of futuristic applications. These are out of reach of conventional displays based on liquid crystals and OLEDs - they are held back by issues related to bulkiness and stability, respectively.

Scaling issues
As the performance of the microLED display is governed by device performance, these emitters must retain the remarkable optical brightness and efficiency of their larger siblings. For standard-size InGaN-based blue and AlGaInP-based red LEDs, engineers have demonstrated efficiencies greater than 80 percent and 50 percent, respectively.

These values give microLED displays the potential to operate with a much higher efficiency than today's LCDs and OLED displays, and deliver a leap in energy efficiency for display technologies. However, for this to happen, the efficiency of the LED must remain high when its dimensions are reduced. And that's far from a given: early reports show that the maximum efficiency of the microLED plummets when device dimensions are shrunk below 100 m.

The dramatic fall in efficiency is attributed to two factors: sidewall damage and surface recombination. Both are inevitable in conventional device fabrication. Surface recombination results from surface states and dangling bonds of the semiconductor surface, and sidewall damage comes from a plasma process that's employed to define the light-emitting area. Both these issues, which are present in microLEDs and standard LEDs, create undesired Shockley-Read-Hall non-radiative recombination sites. However, the key difference with standard LEDs is that they have a huge light-emitting area compared with the sidewall perimeter, so sidewall damage and surface recombination have minor influences on device performance.

For these devices, other factors have a greater impact on efficiency characteristics, such as device design and material quality. In stark contrast, in small microLEDs sidewall damage and surface recombination can wreak havoc, due to a far higher sidewall perimeter-to-emitting-area ratio.

It is important to note that both these issues are more of a concern in LEDs made from the AlGaInP material system. As this alloy has a longer minority-carrier diffusion length and a higher surface-recombination velocity than the III-nitride material system, the fall in efficiency with scaling is more severe in AlGaInP red microLEDs than it is in the blue and green cousins made from InGaN.

The impacts of sidewall damage and surface recombination are easily illustrated by measurements of optical performance, thanks to the strong coupling of efficiency and light output power characteristics (see Figure 1). Electroluminescence images of microLEDs with dimensions from 100 m by 100 m to 10 m by 10 m show that despite inhomogeneous light emission appearing in devices larger than 40 m by 40 m, the bigger devices are brighter. This indicates that non-radiative recombination has dominating influences in smaller devices.

Building better devices
At the University of California, Santa Barbara, our team is addressing these scaling issues associated with microLEDs by developing post-etch fabrication techniques that either lessen or eliminate the effects of sidewall damage and surface recombination.

An insight into the pitfalls of scaling, and how to overcome them, is provided by measurements of the leakage current of microLEDs (see Figure 2). For devices without sidewall passivation, the leakage current is high for devices of all dimensions. However, the leakage increases as the device gets smaller, due to a rise in the perimeter-to-area ratio.

One well-known and common approach to reducing leakage, which stems from sidewall damage and surface recombination, is to passivate the sidewalls using plasma-enhanced CVD. This technique, also known as physical vapor deposition, is effective in larger devices. However, as the results in Figure 2 show, it fails to suppress leakage current as the device size shrinks.

Figure 1. Electroluminescence images show that as microLEDs get smaller,
their maximum efficiency decreases. This trend is a major issue,
because ultra-small devices with high efficiency are needed to produce
efficient microLED displays. Note that the crosses are the p-contact of
the devices.

We are pioneering a superior alternative for microLEDs, using atomic layer deposition (ALD) to passivate sidewalls. Our results, presented in Figure 2, show that this technique is far better at suppressing the leakage current than plasma-enhanced CVD. The success stems from the excellent dielectric material quality and the uniformity of thickness control provided by ALD, as well as this technique's more effective deposition mechanism that eliminates surface states.

As expected, the benefits of ALD make the biggest difference in AlGaInP microLEDs. However, they still deliver significant gains in those made from InGaN. In both cases, devices with sidewalls passivated by ALD have better forward current-voltage characteristics, a lower leakage current and a lower ideality factor. Optical performance is also better, with ALD sidewall passivation ensuring a uniform light emission homogeneity (see Figure 1), and a significant increase in the light output power. For 20 m by 20 m microLEDs driven at 20 A cm-2, this ALD treatment increases the light output power for blue-emitting InGaN devices by 40 percent, and for the red-emitting AlGaInP siblings by 150 percent. We attribute these gains to suppression of sidewall damage and surface recombination, and the resulting reduction in Shockley-Read-Hall non-radiative recombination.

The passivation of sidewalls by ALD is certainly a step in the right direction, delivering a remarkable improvement in the optical and electrical performance of InGaN and AlGaInP microLEDs. However, on its own it is insufficient to prevent efficiency from holding up when shrinking device dimensions. Additional sidewall treatment techniques are needed to realise miniature microLEDs with a high efficiency.

We have pursued this, combining chemical treatment with ALD sidewall passivation to demonstrate, for the first time, size-independent efficiency performance in InGaN and AlGaInP microLEDs. Our technique has much promise, providing a straightforward, versatile way to enhance the optoelectrical characteristics of the microLED.

Figure 2. Researchers at UCSB have shown that atomic layer deposition
(ALD) and plasma-enhanced CVD cut the leakage current in microLEDs.
These results are for InGaN microLEDs.

The downside of dielectric sidewall passivation is that it merely lessens the influences of sidewall damage and surface recombination. With this form of passivation, sidewall defects still drags down device performance. To eradicate these imperfections, chemicals can be applied that either alter the chemical composition or etch away surface defects, so that pristine material results at the sidewalls.

For InGaN microLEDs, we have combined sidewall passivation with a potassium hydroxide chemical treatment. This is an attractive approach, because we can draw on a wealth of reports that describe the reaction mechanism between potassium hydroxide and the III-nitride material system, and detail how this chemical treatment can be applied to standard LEDs.

Combining potassium hydroxide chemical treatment with ALD sidewall passivation has enabled us to realise size-independent leakage current and efficiency characteristics in InGaN microLEDs with dimensions from 100 m by 100 m to 10 m by 10 m (see Figure 3 for details of efficiency performance). These results reveal that with our approach we can eliminate or greatly diminish the influences of sidewall damage and surface recombination.

We have applied similar principles to AlGaInP devices. This has enabled 20 m by 20 m AlGaInP microLEDs to deliver a size-independent efficiency performance at a high current density (see Figure 4). This suggests that we are now on the brink of realising a size-independent efficiency for AlGaInP microLEDs, a goal that could be reached via optimizations that improve performance at low current densities.
While we are still to demonstrate a size-independent efficiency at low current densities for the AlGaInP devices, we take heart from the significant reduction in the drop in efficiency at 20 A cm-2, thanks to the combination of chemical treatment and ALD sidewall passivation. With unoptimized sidewall treatments, the efficiency reduction in 20 m by 20 m AlGaInP microLEDs is just 20 percent at a current density of 20 A cm-2; in comparison, for devices without sidewall treatments and with just ALD sidewall passivation, reductions at the same current density are 80 percent and 50 percent, respectively.

Figure 3. External quantum efficiency measurements show that treating InGaN microLEDs with potassium hydroxide chemical treatment and ALD sidewall passivation (results for conventional devices are on the left, and treated ones on the right) leads to leakage current and efficiency characteristics that are independent of device size.