China is the most important single market for Infineon. Nevertheless, in spring 2018 the company decided to build a brand new 300-millimeter fab in Villach, Austria. This is an important sign for Europe, isn't it?
Prof. Leo Lorenz: Absolutely! And I am very pleased, that high-tech like this will help to strengthen Europe as competitive location. To operate such a wafer fab entails a very high level of expertise in the manufacturing processes, the equipment, the test procedures and eventually also in the education at the universities.
It is also an important message to power semiconductors in general. Processing 300 millimeter wafers is completely different from processing smaller diameters. It’s no longer possible to grow the crystal in the usual way, and other processes to manufacture such wafers are needed. Defect densities, doping profiles and so on play an important role. Many things are different.
Initially, no one had taken Infineon for serious. However, the Dresden plant has been manufacturing large quantities of these large diameter wafers for quite some time now. And now other manufacturers are also following suit. STMicroelectronics has announced to manufacture on twelve inches. There was a meeting on this subject in Japan. And the Chinese also want to jump on that bandwagon.
It's incredibly difficult to handle and process such large thinned wafers, isn't it?
Correct. The transfer of the wafer diameter from 150 to 200 millimeters was relatively simple, because the process technology remained the same. But the leap from 200 millimeters to 300 millimeters – that’s a completely different world.
Imagine a pizza that is 100 micrometers thick. This is like aluminum foil, which has to be processed precisely in fractions of a micrometer! Infineon really did a pioneering job in this area, doing the preliminary work and developing the entire manufacturing and handling technology together with the equipment manufacturers (see ; editor's note).
The fact that the competitors also now want to produce on 300 millimeters is, of course, very pleasing to Infineon. It is much more difficult to convince wafer suppliers to produce wafers in such large diameters, if there is only one customer.
Which special sessions are scheduled at PCIM 2019?
In the advisory board we agreed on four special sessions, all very important future topics: DC networks, solid-state transformers, smart power electronics and functional safety in drive technology.
There is already an important application for DC power grids today: server farms. These are lower voltages – currently 24 volts, but tomorrow 48 and 380 volts. However, the development will not stop there; we are also discussing medium voltages up to maybe six kilovolts. There are already several pilot projects.
Despite all their advantages, such DC grids pose their own challenges, which are still largely unsolved; these include disconnection in the event of a fault and other safety-related issues. With alternating current, this is much easier, because the spark is automatically extinguished, when the current crosses zero. By its nature, this is not the case with direct current. Therefore, hybrid approaches are discussed, combining a mechanical switch with a semiconductor switch – the semiconductor for the dynamic phase, the mechanical switch for the steady state.
Wasn’t there also a special session on solid-state transformers last year, too?
Prof. Leo Lorenz: You are right, but up to now this topic has been perceived too negative, especially by the utilities, which are very conservative. Of course they are afraid of blackouts, and a 50 Hertz transformer is unrivalled cheap and robust – even a lightning strike is usually no problem. There is now a pilot project in Scotland, where a »smart« grid is to be set up parallel to the existing power grid. The 7,800 solid-state transformers to be installed by 2030 are estimated to save 62 million Pounds Sterling [approx. 71 million Euros, editor’s note] and 523 kilotons of carbon dioxide. By 2050 the figures are projected to be 528 million pounds and 2032 kilotons of carbon dioxide, when 36,270 solid-state transformers are to be connected to the grid.
Another very good application for such smart transformers are trains. There – depending on the individual country – the mains frequency is only 162/3 Hertz. So the transformer fills almost the whole locomotive. A corresponding solid-state transformer is only one tenth as big. This would allow the drive and the power electronics to be distributed in the train, and the locomotive is gone. But also there the life cycles are very long. So it will take some time, until smart transformers are actually used in trains.
Generally speaking, one could say, that in today’s AC grids, solid-state transformers have little prospects because of the missing proof of robustness and their costs. In DC grids however, they are already required today. One example is data centers for banking or other enterprises. Most of these are powered directly from the medium-voltage grid, and solid-state transformers offer significant advantages there. Additionally, I can imagine a very large number of applications in the factories of the future, shopping centers and the entire charging infrastructure for e-mobility.
One special session is named smart power electronics. What do you mean by that?
Let’s take a standard inverter as an example. The actual power electronics, in other words the transistors and the topology, remain almost the same, as does the control and drive circuitry. What is new is communication – both to the inside as well as to the outside.
Regarding communication to the inside, for example, the focus is on protecting the inverter, optimization and diagnostics. Communication to the outside world is about maintenance, process optimization and lifetime prediction. Especially when predicting the lifetime, it is necessary to record and process ageing relevant parameters and make them available to the operator.
Mesago PCIM GmbH