top of page

Alan Yeung

Ingantec

* All members of the platform can watch the entire presentation.

 

Please register to become a member.

Alan Yeung | Ingantec: Literature red InGaN shifts 50–80 nm with current; how spectrally stable is red InGaN on a four-inch platform?

12:02 - 14:03

Other snippets from this talk

Summary of the clip:

Literature red InGaN shifts 50–80 nm with current; how spectrally stable is red InGaN on a four-inch platform?

The application focus is AI glasses. Work is underway with an industry partner on a four-inch red InGaN chip-on-wafer product intended to support mass transfer and wide bonding, and eventually monolithic RGB. The baseline target moves from lab scale to four-inch, then hopefully six- to eight-inch platforms, with at least 20% EQE at peak wavelength. As current density rises, EQE tapers off and saturates around 5–10%.

The more interesting result is spectral stability. Literature peak wavelength shifts can be dramatic, moving 50–80 nm when current density increases from 0.1 to 10, or from 1 to 100 A/cm². This red InGaN approach has been quite stable, and at some point it shows a bit of a red shift before the blue shift appears. Beyond 100 A/cm², normal blue shifts should appear, but not as dramatic as typical conventional InGaN red.

For monolithic RGB and mass-transfer routes, spectral stability is relevant when red emission is driven across current density. The four-inch chip-on-wafer product is the first step, with six- to eight-inch scale hoped for. The EQE saturating between 5 and 10% at higher current density defines the measured baseline, while the target remains at least 20% EQE at peak wavelength.

In this short video, you can learn:
* A four-inch red InGaN chip-on-wafer product is being developed with an industry partner for AI glasses, mass transfer, wide bonding, and eventually monolithic RGB.
* Literature InGaN red shifts 50–80 nm over current-density increases from 0.1 to 10 or 1 to 100 A/cm², while this approach stays stable and shows a slight red shift before blue shift.
* EQE saturates around 5–10% at higher current density, against a baseline target of at least 20% EQE at peak wavelength, with six- to eight-inch scale hoped for.

📋 **Clip Abstract** Red InGaN on a four-inch chip-on-wafer platform targets AI glasses, mass transfer, wide bonding, and eventual monolithic RGB, with a baseline of at least 20% EQE at peak wavelength. The approach is spectrally stable compared with literature shifts of 50–80 nm, showing a slight red shift before blue shift and less dramatic blue shift beyond 100 A/cm².

About the speaker:
* Speaker: Alan Yeung
* Company: Ingantec
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven

#RedInGaN, #ChipOnWafer, #MonolithicRGB, #SpectralStability, #MicroLED, #ARDisplays

This is a highlight of the presentation:

Red MicroLEDs and Monolithic Full-Color Integration for AR, Medical Imaging & AI Interconnects

MicroLED Connect 2026

AR/VR Connect 2026

16-17 September 2026

High Tech Campus, Eindhoven

Organised By:

Khasha and Ron

Khasha and Ron

More Highlights from the same talk.

04:11 - 05:46

Stacked or planar monolithic RGB: which architecture wins, and where does the three-times size difference land?

Stacked or planar monolithic RGB: which architecture wins, and where does the three-times size difference land?

Blue substrates and wafers are already one of the key processes in some of the smart and AR glasses being used today, and in the next few years to come. The focus is growing very, very spectrally stable red based on indium gallium nitride. Crack that code and you can potentially solve the monolithic RGB problem — which then splits into two distinct architectural routes.

Two architectures exist for monolithic RGB: vertical or stacked, and planar or horizontal. Each carries pros and cons, and the choice really depends on the application and use cases. Stacking brings attenuation challenges, but you do not have to get even three times smaller. The comparison is one micron as opposed to 0.3 micron in size or pitch, and the planar or horizontal route has its own merits, being traditionally what televisions and other displays have been.

Monolithic RGB is also a platform for folding in added functionality. A sensing function can be incorporated along with the emitters, for example in a four-pixel arrangement rather than three. That widens what a single monolithic die can ultimately do. And the famous green gap curve remains the reference point that frames the whole exercise.

In this short video, you can learn:
* Very spectrally stable red grown from indium gallium nitride is the code to crack for monolithic RGB.
* Vertical or stacked integration faces attenuation challenges but avoids getting even three times smaller, where the comparison is one micron against 0.3 micron in size or pitch.
* Planar or horizontal integration is traditionally what televisions and other displays have been, and monolithic RGB can absorb a sensing function into a four-pixel arrangement.

📋 **Clip Abstract** Spectrally stable red grown on indium gallium nitride is the key to a monolithic RGB, building on the blue substrate and wafer processes already used in smart and AR glasses. The architectural split is vertical or stacked versus planar or horizontal, trading attenuation challenges against a three-times size difference between one micron and 0.3 micron, with sensing as an optional fourth pixel.

About the speaker:
* Speaker: Alan Yeung
* Company: Ingantec
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven

#InGaNRed, #MonolithicRGB, #StackedVersusPlanar, #PixelPitch, #ARGlasses, #MicroLEDDisplays

16:40 - 18:49

Can red and yellow microLEDs reach gigahertz modulation bandwidth for AI optical interconnect multiplexing?

Can red and yellow microLEDs reach gigahertz modulation bandwidth for AI optical interconnect multiplexing?

The improved red InGaN approach is said to apply to yellow and green too, not only EQE but modulation bandwidth. In 2016, coworkers in Beijing applied external strain to EQE; changing strain produced improvement by reducing carrier lifetime. A nanoporous DBL, shown in the top curve, gave very fast response when current density was raised in the micro-LED. These results frame strain, lifetime and current density as linked levers for red, yellow and green emitters.

A recent paper from Shanghai, Fudan University, by Lin and coworkers examined ultra-low power—20 amp or 100 amp per square centimeter—and reported pretty good, very exciting 3dB modulation bandwidth performance on both red and yellow. That matters for AI optical interconnect because multiplexing divisions may eventually look at yellow and red. The work links low current density operation to bandwidth, suggesting red and yellow can contribute alongside blue and green bandwidth targets.

The outlook is bright for AI optical interconnect if the strain relaxation problem is solved, EQE improved, and bandwidth raised to respectable levels. The stated general targets are about three and a half to five megahertz for blue, maybe two and a half to three for green, while yellow and red could run between one and two and a half gigahertz. That is presented as a possible path for the monolithic RGB world.

In this short video, you can learn:
* External strain applied to red InGaN EQE in 2016 reduced carrier lifetime, while a nanoporous DBL gave fast response when current density was raised.
* A Fudan University paper by Lin and coworkers reported exciting 3dB modulation bandwidth on red and yellow at ultra-low power of 20 or 100 amp per square centimeter.
* The proposed general targets are roughly 3.5–5 MHz for blue, 2.5–3 for green, and 1–2.5 GHz for yellow and red in monolithic RGB.

📋 **Clip Abstract** Red InGaN improvements for EQE and modulation bandwidth are said to apply to yellow and green, with external strain, nanoporous DBL, and low-current-density results from Beijing and Fudan University. If strain relaxation is solved and EQE improved, AI optical interconnect multiplexing might use yellow and red, with stated bandwidth targets around 3.5–5 MHz for blue, 2.5–3 for green, and 1–2.5 GHz for yellow and red.

About the speaker:
* Speaker: Alan Yeung
* Company: Ingantec
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven

#RedInGaN, #NanoporousDBL, #ExternalStrainEQE, #GigahertzModulation, #MicroLED, #OpticalInterconnect

07:14 - 09:37

Does 10% peak EQE at 2 A/cm² mean anything when AI optical interconnects run at 5,000 A/cm²?

Does 10% peak EQE at 2 A/cm² mean anything when AI optical interconnects run at 5,000 A/cm²?

The red LED problem is not only efficiency at low bias. Spectral stability and blueshift are recurring issues. Many reported light engines do not run at one or five amps per square centimetre, so peak EQE of 3, 5 or 10% at two amps per square centimetre is not that meaningful for industry. Those parameters now feed into the design equations for red LEDs.

In the journey to monolithic RGB microLEDs, high current density and high junction operation temperature change the green gap curve. Blue and green LEDs drop in EQE, and likewise phosphide. In red InGaN, the position is different: comparing target performance with green and blue, red is not that far from its goalposts. With a UCSB partner, past data established target metrics of at least 20% EQE and potentially one to ten amps per square centimetre.

Application current density sets the bar. Optical engines for glasses can reach 40 or 80 amps per square centimetre. Medical diagnostics require 250 to 500 amps per square centimetre. AI optical interconnects now run at 1,000 to 4,000 or 5,000 amps per square centimetre. What was once called 80 amps of high current density had to be re-tuned after talking to AI optical interconnect people.

In this short video, you can learn:
* Spectral stability and blueshift of red LEDs are critical parameters beyond peak EQE.
* High current density and junction temperature shift the green gap curve for blue, green and phosphide LEDs.
* Target red InGaN metrics are at least 20% EQE at one to ten amps per square centimetre, while AI interconnects demand 1,000 to 5,000 amps per square centimetre.

📋 **Clip Abstract** Red microLED reporting must account for spectral stability and blueshift, because light engines do not operate at one or five amps per square centimetre. A UCSB-partnered target of at least 20% EQE at one to ten amps per square centimetre contrasts with medical diagnostics at 250 to 500 amps per square centimetre and AI optical interconnects at 1,000 to 5,000 amps per square centimetre.

About the speaker:
* Speaker: Alan Yeung
* Company: Ingantec
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven

#RedInGaNMicroLED, #MonolithicRGB, #PeakEQE, #HighCurrentDensity, #MicroLEDDisplays, #AIOpticalInterconnects

More Snippets
CONTACT US

KGH Concepts GmbH

Mergenthalerallee 73-75, 65760, Eschborn

+49 17661704139

admin@techblick.com

TechBlick is owned and operated by KGH Concepts GmbH

Registration number HRB 121362

VAT number: DE 337022439

  • LinkedIn
  • YouTube

Sign up for our newsletter to receive updates on our latest speakers and events AND to receive analyst-written summaries of the key talks and happenings in our events.

Thanks for submitting!

© 2026 by KGH Concepts GmbH

bottom of page