In early 2018, Infineon signed a long-term delivery agreement with Wolfspeed for SiC raw wafers. Nevertheless, Infineon has now acquired Siltectra. Why?
Dr. Peter Wawer: Silicon carbide is strategically important for Infineon. However, at the moment the availability of raw wafers is the bottleneck in the supply chain. In this context, the supply agreement with Wolfspeed is of course very important: It secures our long-term supply of raw wafers. But of course we don't know how long the shortage will last. Not only we, but also many analysts believe that the demand for silicon carbide will continue to grow strongly. The bottleneck could therefore persist much longer.
And how does Siltectra fit into this picture?
This clearly has to do with the difficult situation on the material side. Of course, we don't want to enter the materials business by growing up crystals ourselves and sawing raw wafers from them. But during our market research, we came across various interesting methods that can help us with the supply issue. One of these processes is Siltectra’s Cold Split process. This start-up company from Dresden was founded back in 2010, in a phase when the market - in that case for silicon - was empty due to the photovoltaic boom. At that time, the idea of splitting silicon wafers using a laser was born.
But that time is long gone!
You are right. But the researchers at Siltectra found out that this process is also suitable for other materials, such as silicon carbide. And so they further refined it for the much more expensive silicon carbide. A standard silicon carbide wafer is about 350 micrometers thick. A laser is used to introduce energy into the wafer so that it breaks in a controlled manner. By varying the supplied energy and focusing, it is possible to adjust the depth at which this splitting takes place.
According to the claim »Making two out of one«, by this process we can create two wafers with half the thickness of a standard wafer - with the larger wafer surface for twice as many microchips. This obviously provides some economies of scale, and the bottleneck regarding the availability of silicon carbide isn't that tight for us any longer.
How should the process look like eventually?
In the beginning, a standard silicon carbide wafer is processed normally. At a certain stage it is then split by that Cold Split process. The part that has already begun is then processed to the end, while the split part is reintroduced into the production process at the front-end. This process also saves a large part of the thin grinding required for power semiconductors.
Is this process also suitable for separating raw wafers from the original ingot?
Of course this is also possible. This process could therefore be very interesting for the manufacturers of the raw wafers.
Would you then also make this process available to such companies?
Yes, of course. We are willing to offer this process to raw material manufacturers as a service for the so-called »boule split«. We would like to work together with our suppliers. However, we have not yet reached that stage, and after the takeover we first have to enter into further development and corresponding discussions.
That would certainly be of interest to them, because of the kerf-loss when sawing off wafers, wouldn't it?
Correct. The rule of thumb with silicon carbide is that about half of the material is lost during sawing. This material is considerably more expensive than silicon due to the complex manufacturing process using a gas phase. And if you know that a 150 millimeter silicon carbide wafer is currently traded on the spot market for well over one thousand dollars, then this is quite expensive sawdust.
Is the Cold Split process already technically mature?
On lab scale: yes! However, the process is not yet sufficiently developed to be used in volume production. It will take some time and investment before Cold Split runs stable with a correspondingly high yield on a large scale. But we are confident that this can be achieved - because we have looked very closely at the process in advance and have been processing thin wafers ourselves for decades. This gives us a high level of expertise in this very area.
When do you think Cold Split will be ready for mass production?
The sooner the better, of course. Realistically, we expect to integrate the new cutting process into volume production within five years. Since we are still in the start-up phase it means that we still have to develop machines that will work in our manufacturing. We already have laser machines that are suitable for our purposes in mass production, and we already have quite clear ideas on how to implement them. But we still face a number of process steps at Cold Split where we still have to work hard. There is still a lot of engineering work to be done.
Dr. Wawer, thank you very much for the interview.
The interview was conducted by Ralf Higgelke.