Micro-LED Sidewalls Need Vacuum Passivation
Micro-LED sidewall passivation is no longer a finishing detail once pixels shrink. Etching creates a larger edge-to-active-area burden, so leakage, surface states, and non-radiative recombination can consume a growing share of device performance. A recent 5 µm blue GaN result makes the manufacturing implication clear: the unit operation is not simply “passivate the sidewall,” but repair the etched surface and keep it clean through the handoff to dielectric deposition.
Researchers reported on August 7 that neutral nitrogen/hydrogen (N/H) radical treatment followed directly by vacuum-interconnected atomic-layer deposition (ALD) raised peak external quantum efficiency (EQE) of a 5 µm blue device from 13.76% to 18.52%—a 34.6% relative increase. It is a research demonstration, not a qualified display process. The mechanism still matters for any route toward smaller emitters.
Why the perimeter becomes the device
A conventional LED can tolerate a modest sidewall contribution because much of its active region sits far from an etched boundary. In a micro-LED, perimeter grows quickly relative to emitting area as dimensions fall. Plasma etching can leave damaged material, dangling bonds, and chemical residue. Those defects add leakage paths and sites where carriers recombine without light.
Size dependence is the useful clue. The 5 µm device saw the largest EQE benefit; 10 µm and 20 µm devices also improved, but by less. Reverse leakage at −5 V fell by as much as five orders of magnitude across the reported sizes. That pattern fits a process attacking a surface-dominated loss mechanism rather than only shifting the optical stack.
The sequence is the engineering result
The team used neutral N/H radicals to reduce sidewall damage, then transferred the wafer under vacuum to deposit a 30 nm aluminium-oxide layer by ALD. The no-air-exposure handoff is central. An etched surface can quickly collect oxygen, moisture, and organic contamination; a later passivation step then seals a less controlled interface.
According to the authors, hydrogen helps remove fluorine and chlorine residues from patterning and mask removal, while nitrogen helps address under-coordinated gallium sites. The ALD layer supplies electrical and chemical passivation. This is not proof that one chemistry suits every micro-LED flow. It is evidence that surface repair, contamination control, and dielectric deposition should be developed as one linked process module.
What production teams still need to prove
The experiment used a specific GaN epitaxial structure, 5–20 µm mesas, and laboratory conditions. A production decision needs more than peak metrics:
- Uniformity — radical exposure and ALD coverage across the target wafer size and pixel array.
- Compatibility — preservation of current-spreading layers, contacts, colour-conversion integration, and any transfer flow.
- Reliability — low leakage and higher EQE after thermal cycling, humidity stress, and display operating conditions.
- Throughput — vacuum transfer can improve interface quality, but tool integration and cycle time must be costed against yield.
Those checks decide whether the approach is an enabling production module or a high-performing laboratory step.
The broader technology signal
Micro-LED roadmaps are often framed around mass transfer, colour conversion, and backplane integration. Those remain essential. This result highlights a more local constraint: once pixels are small enough, the etched edge can dictate the optical and electrical budget. A clean vacuum-linked repair-and-passivation flow is a candidate scaling lever—not because it removes every micro-LED challenge, but because it addresses a loss mechanism that intensifies as pixel dimensions fall.
Takeaway: Treat sidewall repair and passivation as a vacuum-linked process module. Peak EQE gains are the research signal; wafer uniformity, reliability, and tool throughput decide whether the module scales into production.