A yen for better chips

Following Japan’s success in developing a new concept in chip production, a UK team checked the coping strategy to see how it could be applied here. Jon Excell reports.

A group of UK engineers recently returned from Japan armed with knowledge that some claim could reinvigorate the UK’s semiconductor industry.

The Japanese semiconductor industry, which has had a tough time in recent years, put in place a series of initiatives four years ago to develop a chip production method for the 21st century. So a crack team of UK industry experts was duly dispatched under the DTI ‘global watch’ umbrella to check out Japan‘s coping strategy and see how it could be applied in the UK.

Japan’s high-volume manufacturing plants have to produce millions of chips to justify their £1.5bn construction costs, a situation that is at odds with today’s shorter product lifecycles and the requirement to produce lower numbers of higher-volume chips for fashion-led products such as mobile phones.

So in 2001, 14 of Japan’s top chip companies — including Toshiba, Toyota and Tokyo Electron — set about developing HALCA (highly agile line concept advancement) a manufacturing model for low volume and flexible production of silicon chips in so-called minifabs. These are defined as facilities where the wafer throughput is in the range of 2,500 8in wafers a month. The other extreme is the ‘megafab’ which deals with 25,000 wafers a month.

Ian Burnett, external affairs director at JEMI UK, the mission’s co-ordinating body, said that these small, agile production units overcome many of the industry’s problems and reduce the cost of wafer processing through addressing the depreciation of typical wafer fab facilities as well as cutting the energy costs.

He explained that the key to HALCA has been minimising the number of machines involved in making devices by developing ones that can do, say, three jobs.

For instance, while a UK fab will typically have separate systems for etching in oxide, silicon or polysilicon, HALCA has seen the development of machines that can do all three. ‘If you only need one machine, it needs just one electricity supply. Plus, since the standby mode is quite energy consumptive it’s better to run it and have it generating revenue than it is to have it standing by,’ said Burnett.

While the impossibility of competing with super-fabs in developing countries means that mini-fabs could be one way of regenerating UK manufacturing, Burnett suggested that the chief promise of the approach for the UK lies in new industries. ‘There’s no chance of super fabs in the UK, but this could be a good way of developing a hi-tech industry sector. The concept could be used in MEMS or lab-on-a-chip type devices, where there are issues of what you are doing possibly being biologically or chemically incompatible with what might be going on in the fab area of another device.’

Burnett added that while MEMs products are mostly low-volume, they could still have huge markets. He claimed that the manufacture of an RFID chip for everyone on the planet would occupy the capacity of an 8in wafer fab for just three weeks. ‘It might well be that these emerging production areas are what drives this,’ he said.