A sound approach to cutting costs

The harnessing of acoustic energy promises to enable substrate re-use during the production of wide bandgap power electronics.

BY ARNO MERKLE FROM CRYSTAL SONIC

There’s no doubt that compound semiconductor devices are on a long-established upward trajectory. Even as far back as the end of the twentieth century they were playing a number of crucial roles in our lives: they were a key ingredient in telecommunication networks; they were providing the optical source for the CD player; and they performed two essential roles in handsets – lighting keypads and amplifying RF signals. During the last two decades sales of these devices have continued to climb, due in part to deployment in lightbulbs and chargers. And growth is sure to continue, as revenues ramp through the electrification of transportation, the roll-out of 5G and the launch of countless satellites.

However, despite great performance advantages that have driven the success of compound semiconductor devices, their potential is yet to be fully realised. Holding this back are three related constraints in the manufacturing supply chain: a large demand-supply gap, particularly for wide bandgap devices based on SiC and GaN; high cost; and waste. Let’s analyse these three issues, one by one.

Figure 1. Sonic Lift-off workflow, enabling substrate re-use for wide bandgap materials.

First, there is a large and increasing gap between the demand for wide bandgap devices and their supply. For example, the Canaccord Genuity Group estimates that the demand for SiC-based power devices, driven by thirst for the likes of electric mobility, charging stations, industrial motors and solar inverters, is leading to approximately a three-fold shortage in SiC wafer supply over the course of this decade.

Second, wide bandgap materials are costly to produce, due to the inherent difficulty required to produce high-quality wafers with limited crystalline defects. Unlike silicon, crystals of GaN and SiC cannot be grown from the melt, leading to the need for a far higher energy input and greater challenges for process control. Another factor at play is that the industrial know-how for producing such materials, while broadening somewhat, is still relatively concentrated, especially for GaN and AlN. Due to these difficulties, it’s not surprising that the substrate prices for GaN, AlN and SiC are at least one or two orders of magnitude higher than those for silicon. As a result, their cost can account for about half of total device manufacturing cost – and this is not expected to appreciably change. Forecasts by PGC Consultancy indicate that wide bandgap substrates will remain the dominant manufacturing cost contributor, bottoming out at around 35-45 percent of SiC die cost in the coming decade as wafer production ramps up.