Jay Barrus, president Rohm Semiconductor Americas, also confirmed that a power module is in development, to be released in 2025 and that the company is developing its fifth generation SiC, which will be available in one to two years’ time.
The company has also announced the acquisition of a fab in Japan to expand SiC capacity, in support of the fab in Germany. Rohm is focusing on yield levels for six-inch wafers. “Current high yields are around 40-50%, with Rohm achieving high 50-60%,” he said. Pressure from the automotive manufacturers in particular means the need to raise yield and reduce cost is acute.
Die and die structure is also significant, added Barrus. “Rohm uses trench technology, which has challenges. “[Our] Gen 5 probably has an RDS benefit but we are focused on yield. Planar is an easier process and design, so it has to come into the equation.” Keng Ly, VP of marketing, Rohm Semiconductor USA, put the drive for the next technology down to the Osbourne Effect, i.e.the desire to always want the next technology. (The social phenomenon is named after Osbourne Computer which pre-announced several models in 1983, all promising to outperform the original. The result was to stunt sales of the only model it had on the market, resulting in bankruptcy the same year.)
“Superjunction does not follow Moore’s Law; it is more like every four years in the power industry for each new power technology,” he sighed.