Even as the first 90nm devices hit volume production, the industry is hard at work developing 65 and 45nm technology. The insatiable desire to make things smaller is driven by economical and technical needs; the smaller the die size, the lower the die cost and the more transistors on the die, the higher the functionality. Because of this, and despite the technical challenges, there will continue to be innovation in IC manufacturing. But, by all accounts, it isn’t going to be easy.
Ludo Deferm, vice president of IMEC, explained that for some independent device manufacturers (IDMs) such as Philips Semiconductors, STMicroelectronics and Freescale – as well as the leading pure play foundries like TSMC – there isn’t a choice; they will have to continue to invest in ever smaller processes, at least for the next five years. During that time, closer alliances between IDMs and pure play foundries will be forged because, as Deferm pointed out, whilst foundries provide manufacturing capability, IDMs generally provide the technology.
As the industry progresses from 65 to 45 and, ultimately, to 32nm, there will be more alliances similar to the one formed between Motorola (Freescale), ST and Philips at Crolles, because both the cost and complexity of developing the next process will be too great a burden for one company. Although Deferm is confident that 45nm will happen and is actively researching 32nm, he admits some manufacturers will draw a line at 65nm, either because they can’t afford to scale any further, or because they simply don’t need to. This isn’t new – the industry ‘sweet spot’ is still between 0.18 and 0.13µm – but Deferm predicts it will become more widespread.
|