It’s always interesting times for the semiconductor industry

Creating the features on these chips is becoming a challenge. As New Electronics has observed, conventional lithography techniques are reaching the end of their useful life, although the use of double patterning and beyond is staving off what seems to be the inevitable. The solution has been identified as extreme ultraviolet lithography (EUV). Instead of the 193nm light used in current systems, EUV generates radiation at 13.5nm – around the same size as the features that designers would like to create. The problem is that EUV remains tantalisingly out of reach. Originally planned for use in 2007, EUV is still work in progress. The stumbling block is the amount of power delivered at the wafer; the more power, the faster wafers can be processed and the better the economics. Ideally, manufacturers would like more than 200W at the wafer; today's prototypes have yet to generate anything like that. However, significant progress has been claimed by Japanese company Gigaphoton. Earlier this year, it announced it had generated a light power of 43W; now, it has increased that to 92W. And yet. The cost of making chips at the leading edge is growing rapidly and EUV will only add to that cost – apart from the huge price tags envisaged for EUV steppers, there is the small matter of the tens of kW each stepper needs to operate. If the industry wants to follow Moore's Law, then it has to make even more chips on even larger wafers. But that route is the preserve of a handful of manufacturers – probably three – and not many more customers. What about the rest? Will the industry's focus shift from the relentless pursuit of smaller features to something like system in package, despite the inherent difficulties of through silicon vias? Semiconductor executives always live in interesting times; but the decisions being made now seem to put us in times more interesting than ever.