Steven DenBaars | UC Santa Barbara: Can tunnel junctions replace absorbing ITO and stack RGB into one monolithic microLED chip?
07:18 - 09:29
Other snippets from this talk
Summary of the clip:
Can tunnel junctions replace absorbing ITO and stack RGB into one monolithic microLED chip?
Replacing the top indium tin oxide in microLEDs matters because ITO absorbs about 15 to 20% of light. The tunnel junction replaces it with n-type gallium nitride and uses tunneling through a very thin junction gap, leaving no ITO in the device. The resulting structure is an NPN microLED, with full injection achieved by tunneling rather than by a transparent conducting oxide.
Stacking follows by placing a tunnel junction on top, a second tunnel junction, and a third, aiming at one monolithic chip that is red, green and blue at one size. That chip would have four different contact layers and use bump bonding; turning on various contacts gives various colors. Blue and green have already been integrated by removing hydrogen preservation.
Hydrogen removal was the key: data from a Nobel Prize-winning paper shows heating drives hydrogen out of the semiconductor and causes a huge resistance decrease. The team removed hydrogen from the crystal layers by cycling an annealing step with selective area growth. This leaves only about a 0.2 volt tunnel-junction penalty; the blue tunnel-junction device is about 30 to 40% brighter than the ITO chip at these current densities.
In this short video, you can learn:
* ITO on top of microLEDs absorbs roughly 15–20%, so a tunnel junction enables an ITO-free NPN microLED with n-type GaN.
* Tunnel junctions allow LEDs to be stacked with multiple contact layers toward a monolithic red-green-blue chip at one size.
* Cycling an annealing step with selective area growth removes hydrogen, cutting the tunnel-junction penalty to about 0.2 V and giving 30–40% brighter blue devices than ITO chips.
📋 **Clip Abstract** Tunnel junctions replace absorbing ITO with n-type GaN in NPN microLEDs and enable stacked, contact-selectable red-green-blue chips. Removing hydrogen via cycled annealing with selective area growth reduces the tunnel-junction penalty to about 0.2 V, while blue tunnel-junction devices are 30–40% brighter than ITO chips at these current densities.
About the speaker:
* Speaker: Steven DenBaars
* Company: UC Santa Barbara
* Event: MicroLED Connect 2024
* Location: High Tech Campus, Eindhoven, Netherlands
Join the most event MicroLED, AR/VR display and Optical I/O Connect event of the year
Learn more https://www.microledconnect.com/
#TunnelJunctionMicroLED, #ITOFreeNPN, #HydrogenRemoval, #SelectiveAreaGrowth, #MicroLEDDisplays, #IIINitrides
This is a highlight of the presentation:
Recent Advances in III-Nitrides for High Efficiency 1 to 10 micron scale MicroLED Devices
More Highlights from the same talk.
02:24 - 04:20
Can shrinking microLED chips from 20 µm to 1 µm erase the 3× cost disadvantage versus OLED?
Can shrinking microLED chips from 20 µm to 1 µm erase the 3× cost disadvantage versus OLED?
Augmented reality and data-communication centers are two applications that can exploit microLED brightness. OLEDs now reach about 2,000 candelas per square meter, while microLED demonstrations are in the millions; a Display Week result reached 4 million nits. That brightness advantage is the differentiation to pursue, because it is where microLEDs stand apart from other display technologies.
Against OLED, microLEDs currently carry a huge cost disadvantage of about a factor of three. The cost analyses were done at chip sizes of roughly 20 microns to 10 microns. Taking the chip down to one micron makes cost scale as the area squared, so a tenfold or greater reduction may be possible; increasing spacing then leaves free space to use.
The focus must be brightness and efficiency, not only raw output. DenBaars reports 90% lower power consumption for LCD and challenges claims that microLED power consumption is worse, attributing that to the driver and how it is driven. He states microLED efficiencies are now vastly superior to OLED at all colors, with gallium nitride used for blue and green.
In this short video, you can learn:
* MicroLED brightness demonstrations reach millions of candelas per square meter, including 4 million nits at Display Week, versus OLED's roughly 2,000.
* At 10–20 µm chip sizes microLEDs carry about a 3× OLED cost disadvantage, but scaling to 1 µm makes cost fall as area squared, potentially tenfold or more.
* DenBaars attributes reported worse microLED power consumption to the driver, saying microLED efficiencies are vastly superior to OLED at all colors.
📋 **Clip Abstract** MicroLEDs offer a brightness advantage—millions of nits versus OLED's roughly 2,000 candelas per square meter—but carry an approximately 3× cost disadvantage at 10–20 µm chip sizes. Reducing chips to 1 µm can make cost scale as area squared, potentially yielding tenfold or greater reductions while efficiency and driver design determine power consumption.
About the speaker:
* Speaker: Steven DenBaars
* Company: UC Santa Barbara
* Event: MicroLED Connect 2024
* Location: High Tech Campus, Eindhoven, Netherlands
Join the most event MicroLED, AR/VR display and Optical I/O Connect event of the year
Learn more https://www.microledconnect.com/
#MicroLEDBrightness, #ChipAreaScaling, #GalliumNitride, #DriverPower, #MicroLEDDisplays, #IIINitrides
11:15 - 13:29
What still blocks InGaN red microLEDs from projection: efficiency, wavelength, or color purity?
What still blocks InGaN red microLEDs from projection: efficiency, wavelength, or color purity?
Red remains the big problem, whether the emitter is aluminum gallium phosphide or indium gallium nitride. Indium gallium nitride is preferred because it monolithically integrates without changing material system. The approach uses a red indium gallium nitride quantum well: small holes are etched on top so hydrogen can diffuse out and activate the buried tunnel junction, capped with n-type gallium nitride.
The resulting chip is completely ITO-free and needs only one single metallization step — four mask levels in total, reducing microLED process complexity. Because there is no p-contact any more, the same aluminum nickel gold stack serves both the n-contact and the lower contact. The red microLED sits on patterned sapphire and already delivers single emission.
Efficiency now reaches around 6% over a fairly wide range of current densities, but the spectrum is not yet good enough. Emission is still broad, about 40 nanometers full width at half maximum, and should be centered at 630 nanometers; display manufacturers want 20 nanometers centered at 620 to 630. Shrinking the microLED size is the next task.
In this short video, you can learn:
* Etched holes above the red InGaN quantum well let hydrogen diffuse out and activate the tunnel junction, which is what eliminates the p-contact.
* Losing the p-contact collapses the process to four mask levels with a single aluminum nickel gold metallization and no ITO.
* At roughly 6% efficiency the emitter works, but a 40 nanometer linewidth still misses the 20 nanometer, 620 to 630 nanometer display spec.
📋 **Clip Abstract** Indium gallium nitride red microLEDs are being driven toward monolithic integration in the same gallium nitride material system, using a tunnel junction activated by hydrogen escaping through etched holes. The ITO-free, four-mask-level chip reaches about 6% efficiency across a wide current density range, yet a 40 nanometer linewidth and emission short of 630 nanometers leave color purity as the remaining barrier.
About the speaker:
* Speaker: Steven DenBaars
* Company: UC Santa Barbara
* Event: MicroLED Connect 2024
* Location: High Tech Campus, Eindhoven, Netherlands
Join the most event MicroLED, AR/VR display and Optical I/O Connect event of the year
Learn more https://www.microledconnect.com/
#InGaNRedMicroLED, #TunnelJunction, #ITOfreeProcess, #PatternedSapphire, #MicroLEDDisplays, #IIINitrideEpitaxy
04:51 - 06:49
Why do 1 µm microLEDs fail without atomic layer deposition sidewall passivation?
Why do 1 µm microLEDs fail without atomic layer deposition sidewall passivation?
The efficiency droop appears as chip size is reduced from 100 to 10 microns, and the team's solution is an atomic layer deposition coating on the sidewalls. This ALD process, developed over five years, maintains efficiency down to one micron. Without ALD, reference LEDs around 40 to 20 microns produce no light out of the chip.
Traditional plasma-enhanced CVD, used for LED lighting chips, did not give good efficiency at 10 microns. ALD is a very low-temperature, low-damage process, yielding very good uniformity and emission into the 10 and now one-micron range. The droop is particularly evident at low current density: direct-view applications want about 5 A/cm², while projection wants more like a hundred.
ALD passivation addresses the surface: the work shows dangling bonds at the surface were passivated, and the indium tin oxide was protected. This is also evident in leakage current. At low current density, traditional LED technologies were not very efficient, so ALD gives a huge efficiency boost for direct-view operating points where the droop had been most severe.
In this short video, you can learn:
* Atomic layer deposition sidewall coatings maintain microLED efficiency down to one micron, while reference LEDs without ALD produce no light around 40 to 20 microns.
* Plasma-enhanced CVD used for lighting chips failed to give good efficiency at 10 microns, whereas ALD is low-temperature, low-damage and gives uniformity and emission to one micron.
* ALD passivates surface dangling bonds and protects indium tin oxide, reducing leakage current and boosting efficiency at the low current densities used for direct view.
📋 **Clip Abstract** Efficiency droop from 100 to 10 microns was solved by atomic layer deposition sidewall coatings, maintaining emission to one micron where non-ALD references fail. ALD's low-temperature, low-damage passivation protects ITO and dangling bonds, reducing leakage and improving low-current-density direct-view efficiency versus plasma-enhanced CVD.
About the speaker:
* Speaker: Steven DenBaars
* Company: UC Santa Barbara
* Event: MicroLED Connect 2024
* Location: High Tech Campus, Eindhoven, Netherlands
Join the most event MicroLED, AR/VR display and Optical I/O Connect event of the year
Learn more https://www.microledconnect.com/
#AtomicLayerDeposition, #SidewallPassivation, #PlasmaEnhancedCVD, #IndiumTinOxide, #MicroLED, #IIINitrides
About this talk
→ Recent Advances in III-Nitrides for High Efficiency 1 to 10 micron scale MicroLED Devices · Steven DenBaars · MicroLED Connect 2024
Other talks by Steven DenBaars
- Developments in High Efficiency Ultra-Small Micro-LEDs Based on III-Nitrides · Quantum Dots: Material Innovations and Commercial Applications 2022
- Advancements in Ultra-Small InGaN MicroLEDs: Sub-1µm Pixel Efficiency and Native Color Growth · MicroLED Connect 2026
