Shaoxing's silicon wafers, synthetic sapphire transforming industry

"For manufacturers, size matters. When production yields are comparable, larger crystals allow more usable material to be cut from each ingot, reducing edge losses and unit costs," Zhang said, adding that maintaining a stable thermal field and controlling defects become increasingly difficult as a Kyropoulos crystal grows in volume and weight.

The company's latest annual report said it had achieved scaled production of 750-kg and 1,000-kg sapphire ingots and substrates measuring between 4 and 8 inches. It has also developed substrates of up to 12 inches.

Jingsheng is also applying its expertise in crystal growth and hard-material processing to semiconductor silicon wafers, seeking to supply the equipment needed across the entire manufacturing chain.

"Large silicon wafers form the base material for integrated circuits, while their flatness and surface defects directly affect yields during photolithography," said Li Yangjian, head of Jingsheng's polishing equipment research institute.

For years, the market for high-end 12-inch wafer polishing and edge-polishing equipment was dominated by US and Japanese suppliers, Li said. Chinese wafer manufacturers faced high procurement and maintenance costs while relying on foreign vendors for access to key process parameters and equipment calibration.

Jingsheng has since begun to change that equation. Li said the company independently developed China's first domestically produced 12-inch edge-polishing and final-polishing machines, which are now supplied in batches to major Chinese wafer manufacturers.

More than 95 percent of the core components in its large-wafer equipment are produced domestically. The equipment can reduce overall production-line costs by about 30 percent compared with imported alternatives, Li added.

Localizing the machinery, however, addressed only part of the manufacturing challenge. Wafer polishing involves dozens of constantly changing variables, and engineers previously had to finetune the process through repeated trial and error.

"In the past, optimizing a single set of parameters could take weeks, making it difficult to maintain consistent product quality and constraining the upgrading of China's silicon-wafer industry," Li said.

To shorten that tuning cycle, Jingsheng has incorporated artificial intelligence into the production process. At its Shaoxing plant, 12-inch automated polishing machines use AI-powered vision systems to detect micron-scale scratches and surface irregularities.

The control system analyzes the images and adjusts rotation speed, pressure and polishing-fluid ratios within milliseconds, replacing experience-based manual tuning with real-time, closedloop process control, Li said.

lijiaying@chinadaily.com.cn