Seoul Viosys mass-produces rare-earth-free white micro LEDs down to 10µm

Seoul Viosys says it has begun mass production of a white micro LED that produces white light without rare-earth elements, breaking a dependency that has defined white LED manufacturing for roughly three decades. Aju Press reports that the company announced the development on 28 October, with Semiconductor Today publishing its report a day later. Aju Press's machine-translated article renders the company name as "Seoul Biosys", while Semiconductor Today calls it Seoul Viosys, a subsidiary of Seoul Semiconductor.

What changed after three decades

For about thirty years, white LEDs have been made by layering rare-earth phosphors such as yttrium aluminium garnet (YAG) over a blue LED chip. According to Semiconductor Today, that approach required a package holding the phosphor, which is why white chips could not shrink below 1000µm even as other colours went micro-scale. Seoul Viosys and Seoul Semiconductor had already used their WICOP architecture — Wafer Integrated Chip on PCB, which mounts chips directly onto a substrate without separate wires or packaging, as Aju Press describes it — to mass-produce ultraviolet, blue, green, red and infrared LEDs smaller than 100µm. WICOP 3, the third generation, extends the same approach to white light generated from gallium nitride (GaN) alone, with no rare earths, no package and none of the silicone resins, plastics or lead frames found in conventional LEDs. The company claims this is the first white opto-semiconductor to achieve white light in a micro LED under 100µm, with chips as small as 10µm.

Performance and cost

In the first mass-production run, a chip measuring 25µm x 10µm produced 0.015 lumens of white light, with chromaticity inside the CIE standard white region, according to both sources. Luminous efficacy is close to 100 lm/W, with a target of 200 lm/W within two years; Semiconductor Today notes that a conventional 1000µm x 1000µm white chip achieves 50 lm. Seoul Viosys claims manufacturing costs of roughly half those of a conventional white LED, along with lower environmental impact, since rare earths are a major source of pollution and the new chip contains no synthetic organic materials. The company also argues the chips can last for decades. The commercial stakes are sizable: white LEDs account for about $10bn of the $15bn global market for packaged LEDs, Semiconductor Today reports.

From lighting to AR and AI interconnects

According to Aju Press, Seoul Semiconductor founder Lee Jeong-hoon said the project began "out of curiosity about why rare earth elements were necessary to create white light", and that the company now aims to expand into AR and AI optical communication markets. Semiconductor Today reports that WICOP arrays — 1,000 chips of about 10µm arranged within 1mm² — can transmit data ten times faster than copper links at one-tenth the power, a pitch aimed at AI-era data loads where electrical interconnects slow down, generate heat and consume more power. Mass production is scheduled to begin in the fourth quarter of 2026, with supply prioritised for long-standing customers; Semiconductor Today phrases this as clients who have respected the company's intellectual property. The two reports also cite different Omdia figures: Semiconductor Today says Omdia expects this technology and AR together to create a $60bn market within five years, while Aju Press cites Omdia projecting micro LED display revenue growing from $10.54m this year to about $6.8bn by 2032.

Why it matters

White LEDs have been tied to rare-earth supply chains for a generation, and rare earths are a concentrated, geopolitically sensitive commodity. A GaN-only white chip removes that dependency and, if the cost and efficiency claims hold, applies pricing pressure across a $10bn market. Sub-100µm white pixels are also a prerequisite for the AR displays the company is chasing, and the optical-interconnect claims — faster data at a tenth of the power — target one of the most expensive bottlenecks in AI infrastructure. The performance numbers are company claims at this stage, and independent testing will show whether WICOP 3 delivers; the divergent market forecasts across the two reports suggest observers are still sizing up the opportunity.