Last Monday, STMicroelectronics announced its plans to develop GaN-on-Silicon technology for power. The fact that this major player in power electronics is just entering does raise some questions – especially about the timescale.
When International Rectifier first unveiled its GaN-on-silicon platform for power semiconductors in November 2008, I was on site. Today, almost exactly ten years later, this technology is almost ready for mass production and is available from a number of semiconductor companies - actually off the shelf at foundries like TSMC. And this is where STMicroelectronics enters, a major player in the field of power semiconductors, and also wants to jump onto the GaN train.
So the question arises why STMicroelectronics did not get involved earlier is justified. Probably the responsible people have now realized that electromobility will be the driving force behind wide-bandgap semiconductors, i.e. gallium nitride and silicon carbide. And losing ground in the automotive sector would be devastating for the company; actually the automotive sector contributed 3 billion US dollars in fiscal year 2017, some 36 percent to the company’s overall revenue. And like many of those who come too late, STMicroelectronics has picked an experienced development partner. The GaN-experienced microelectronics research institute Leti has the task of developing the process technology.
Infineon proves that things can be different and that you do not have to develop everything from scratch. Despite the acquisition of GaN pioneer International Rectifier in January 2015, the market leader came a bit too late for GaN transistors for higher voltages. Two months later, in March 2015, it licensed the mature GaN technology from Panasonic and transferred it to its own plant in Villach, Austria. At the end of 2018, almost three years later, mass production of CoolGaN is to start there.
But let’s get back to STMicroelectronics. What does all this mean for the timescale we are talking about? Mass production is by no means an issue in the efforts now started. According to their statement, only a 200 mm pilot line will be up and running in 2020. There is still a long way to go before starting mass production! With Infineon as the benchmark, another three to four years will go by, meaning that mass production cannot be expected before 2023. And if an automotive qualification – as I have learned from well-informed sources at Infineon – takes another five years, fully qualified automotive GaN components from STMicroelectronics will be manufactured in large quantities not before 2028!
Therefore, the company is likely to miss the first big wave of demand for GaN-on-silicon. But as a second-source supplier, this major player in power semiconductors will certainly have its place. But others will make the big margins. Because if you come too late, you only get what's left.