IVWorks strengthens GaN epitaxy
IVWorks, a Korean semiconductor materials startup, is accelerating the shift from automated production which follows fixed procedures to an autonomous process in which AI judges the current state and decides the next action on its own. The shift is built on the company's in-house AI epitaxy platform, DOMM, used in its GaN epiwafer production.
GaN epitaxy requires precise control of multiple process conditions to keep crystal growth stable, making it essential to continuously monitor and respond to even small changes during production. To address this, IVWorks has collected RHEED data from its MBE process together with time-series data such as equipment temperature and pressure since 2019. To date, the company has accumulated more than 330 million time-series data points and over 30TB of RHEED data.
The accumulated data is also used in the actual mass production process. AI models analyze RHEED and equipment data to assess growth conditions and determine whether there are any abnormalities, while the automation framework MRA (MBE Run Automation) is responsible for carrying out the actual process, from wafer loading and growth to unloading. Although the two systems have different roles, by working together, IVWorks says it has been able to gradually reduce human intervention and build a foundation for improving production stability.
Building on this experience, IVWorks says it is now advancing its manufacturing system to use data not just for after-the-fact quality checks, but to monitor process conditions and guide production in real time. As a next step, the company plans to develop AI agents that combine equipment status, production history and quality data to recommend maintenance timing and production plans on their own, enabling uninterrupted production even at night and on weekends. The goal is to raise both quality and productivity, strengthening IVWorks' competitiveness in GaN epitaxy production.
Through this effort, IVWorks aims to raise equipment uptime from roughly 84 percent to 89 percent, while streamlining staffing from about two operators per machine to 1.5.
In recognition of this in-house technology capability, IVWorks was recently selected to take part in a program run by Korea's Ministry of SMEs and Startups to build AI-specialised smart factories. Through the program, the company plans to further expand the use of its accumulated manufacturing data and AI-based automation across its production environment.
Hongkyun Noh, CIO of IVWorks, said: “GaN epitaxial wafers are a key material for next-generation semiconductors, including 5G and 6G communications, AESA radar, and optical interconnects. As mass production of GaN epitaxial wafers begins in earnest, combining data-driven autonomous processes will be essential to securing competitiveness in quality, delivery, and cost simultaneously.”
Pictured above: As part of the AI-driven automated production process, an engineer remotely monitors equipment operation and thin-film growth in real time from the control room
