Ivan-Christophe | Aledia: Can nonpolar nanowire microLEDs reach one gigahertz at only ten amps per square centimetre?
12:15 - 14:21
Other snippets from this talk
Summary of the clip:
Can nonpolar nanowire microLEDs reach one gigahertz at only ten amps per square centimetre?
Typical 2D microLEDs use c-plane emission, where a strong electric field and the quantum confined Stark effect reduce bandwidth at low current density. Nanowires instead grow the active region on their side, producing in-plane, nonpolar emission without that electric field. That structural change is the basis for obtaining high bandwidth at low current density, as measured in the -3 dB bandwidth versus current density data.
Above ten amps per square centimetre, the nanowire microLED already reaches one gigahertz -3 dB bandwidth, which corresponds to about two gigabits per second. At 200 amps per square centimetre, capability rises to 1.5 gigabits per second. This matters for optical interconnect power consumption: operating at lower current density also lowers droop, whereas typical 2D microLED operation sits above 2000 amps per square centimetre because below that the bandwidth is not obtained.
The work is also extending to the full architecture required to enable those data links. The measured link between low current density and high bandwidth comes directly from in-plane nanowire emission, not from c-plane operation. With one gigahertz at ten amps per square centimetre and 1.5 gigabits per second at 200 amps per square centimetre, the data link can be designed around lower-current-density operation for reduced power.
In this short video, you can learn:
* In c-plane 2D microLEDs, a strong electric field and the quantum confined Stark effect reduce bandwidth at low current density.
* Nanowire microLEDs grow the active region on the side of the nanowires, giving nonpolar in-plane emission without that electric field.
* Measured -3 dB bandwidth reaches 1 GHz above 10 A/cm², and capability is 1.5 Gbit/s at 200 A/cm², while typical 2D microLED operation is above 2000 A/cm².
📋 **Clip Abstract** Nonpolar nanowire microLEDs place the active region on the side of the nanowires, giving in-plane emission and avoiding the strong electric field and quantum confined Stark effect of c-plane 2D microLEDs. Measured -3 dB bandwidth is above one gigahertz beyond ten amps per square centimetre, about two gigabits per second, with 1.5 gigabits per second at 200 amps per square centimetre, supporting lower-current-density optical interconnect operation.
About the speaker:
* Speaker: Ivan-Christophe
* Company: Aledia
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven
#NanowireMicroLEDs, #NonpolarInPlaneEmission, #QuantumConfinedStarkEffect, #LowCurrentDensityBandwidth, #MicroLEDDisplays, #OpticalInterconnect
This is a highlight of the presentation:
High‑Performance Nanowire MicroLEDs for AR and Optical Interconnect Applications
More Highlights from the same talk.
03:08 - 04:35
Why do etched 2D microLEDs lose efficiency as pixels shrink, and how does bottom-up nanowire growth avoid it?
Why do etched 2D microLEDs lose efficiency as pixels shrink, and how does bottom-up nanowire growth avoid it?
Conventional micro-LEDs are etched into a 2D GaN wafer. That etching generates sidewall defects, and these defects are what a bottom-up approach avoids. Instead of etching pixels from a planar wafer, the micro-LED is grown directly in the form of nanowires. This produces a self-passivated structure that avoids those sidewall effects. The result is a route to smaller micro-LED pixels without the efficiency loss normally caused by shrinking the pixel.
With nanowire micro-LEDs, pixel size is reduced by reducing the number of nanowires in the pixel, not by losing efficiency as the pixel shrinks. Because the nanowires are grown bottom-up, they can be arranged in a periodic way. That periodic arrangement creates a photonic crystal effect. The photonic crystal effect gives directional emission of light, and it does so without the use of a microlens.
That directional emission is described as very good for the etendue. It also helps optimize the coupling between the microdisplay and the optical waveguide. So the bottom-up nanowire approach addresses two linked problems: sidewall defects from etching a 2D GaN wafer, and light extraction without microlenses. By growing self-passivated nanowires and arranging them periodically, the micro-LED can shrink through fewer nanowires per pixel while enabling photonic-crystal-based directional emission for AR and optical interconnect applications.
In this short video, you can learn:
* Etching 2D GaN into micro-LEDs generates sidewall defects that a bottom-up nanowire growth approach avoids.
* Nanowire micro-LEDs reduce pixel size by reducing the number of nanowires in the pixel without losing efficiency.
* Periodic nanowire arrangement creates a photonic crystal effect for directional emission without a microlens, good for etendue and microdisplay-to-waveguide coupling.
📋 **Clip Abstract** Etched 2D GaN micro-LEDs suffer sidewall defects as pixels shrink, while bottom-up nanowire growth creates a self-passivated structure that avoids those sidewall effects. The nanowires can be arranged periodically to produce a photonic crystal effect, giving directional emission without a microlens, improving etendue and coupling between the microdisplay and the optical waveguide.
About the speaker:
* Speaker: Ivan-Christophe
* Company: Aledia
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven
#NanowireMicroLEDs, #BottomUpGaN, #SidewallDefectAvoidance, #PhotonicCrystalEffect, #ARDisplays, #OpticalInterconnect
08:59 - 11:27
Why does copper hit a wall above one metre, and where can microLEDs beat VCSELs below two picosecond per bit?
Why does copper hit a wall above one metre, and where can microLEDs beat VCSELs below two picosecond per bit?
Copper is reaching its limits. The copper wall appears above about one meter, and that distance becomes shorter when bandwidth increases. A second constraint is the power wall: racks now use more than 100 kilowatts. To keep increasing bandwidth, power consumption must fall. Reaching below one picosecond per bit would help. The challenge is enabling massive bandwidth scaling while lowering the picosecond per bit below one picosecond per bit.
For optical interconnect, VCSELs and lasers are options, but both use a chipset. That chipset consumes power, and they have a threshold. It is difficult to reduce those optical interconnects below two picosecond per bit, with efficiency below two picosecond per bit. Micro LED could have a good playground in short distance, which the speaker places below ten meters. Because there is no chipset for those optical links and no threshold, theoretically those links could go below two picosecond per bit.
In the full architecture, micro LED emitters are required. The comparison is already set: copper cannot do the job above about one meter. VCSELs and lasers face chipset and threshold limits below two picosecond per bit. Micro LED links avoid chipset and threshold. Their target playground is short distance below ten meters, with potential below two picosecond per bit.
In this short video, you can learn:
* Copper hits a wall above about one metre, and the distance shortens as bandwidth rises.
* VCSELs and lasers use a chipset and have a threshold, making below two picosecond per bit difficult.
* Micro LED optical links avoid chipset and threshold, targeting short distances below ten metres with theoretical operation below two picosecond per bit.
📋 **Clip Abstract** Copper interconnect cannot support rising bandwidth beyond about one metre, and racks above 100 kilowatts expose a power wall that demands lower picosecond per bit. VCSEL and laser links face chipset and threshold penalties that hinder operation below two picosecond per bit, while micro LED emitters offer a no-chipset, no-threshold short-reach playground below ten metres with a theoretical path below two picosecond per bit.
About the speaker:
* Speaker: Ivan-Christophe
* Company: Aledia
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven
#CopperWall, #VCSELs, #MicroLEDEmitters, #PicosecondPerBit, #OpticalInterconnect, #DataCenterOptics
07:31 - 08:27
Can monolithic RGB microLEDs be grown in one shot with 2.5 micron subpixels and 47% DCI-P3?
Can monolithic RGB microLEDs be grown in one shot with 2.5 micron subpixels and 47% DCI-P3?
The demonstrators on show run on passive backplanes. They are monolithic green, blue and red micro-LEDs at a 2.5 micron size. Smaller pixels could have been done, but 2.5 micron is what was used for the first demonstrator. The monolithic RGB uses native red, green and blue, and the three colours were obtained in a single growth.
The monolithic RGB demonstrator was built with a 2.5 micron subpixel size, and a passive demonstrator emits the three colours at the same time. For this first demo, 47% of DCI-P3 was obtained. The three colours came out of the single growth, which is the point that matters: red, green and blue are grown together, not assembled afterwards.
What was actually demonstrated: a monolithic stack emitting green, blue and red; monolithic RGB with native red, green and blue; and a passive, backplane-driven demonstrator where all three colours are emitted simultaneously. The first demo reached 47% of DCI-P3. The team says it could have gone to smaller pixels than 2.5 micron, and that 2.5 micron is simply the geometry of this first pass. Monolithic RGB emission in a single growth is the result claimed.
In this short video, you can learn:
* Monolithic green, blue and red micro-LEDs were demonstrated at a 2.5 micron size on passive backplanes.
* The monolithic RGB stack uses native red, green and blue at 2.5 micron subpixels, and 2.5 micron was a choice rather than a limit.
* A passive demonstrator emitted all three colours simultaneously and reached 47% of DCI-P3 from a single growth.
📋 **Clip Abstract** A monolithic RGB micro-LED demonstrator emits native red, green and blue at 2.5 micron subpixels from one growth. The passive-backplane demo fires all three colours at once and reaches 47% of DCI-P3.
About the speaker:
* Speaker: Ivan-Christophe
* Company: Aledia
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven
#MonolithicRGB, #NativeRGB, #PassiveBackplane, #DCIP3, #MicroLED, #EmissiveDisplays