Get email delivery of the Cadence blog featured here
Modern fabs use 300mm (12") wafers. Older fabs have used 200mm (8") wafers since the early '90s, starting with IBM Fishkill's pilot with Siemens to build 64Mb DRAM in 1990. Before that, we had 5" wafers (150mm), and if you go back far enough in history, you'll find 4", 2", and even 1" when integrated circuits started. 300mm started in 2000 with Siemens/Motorola (and two beta lines in the US at l300l and Japan at Selete).
For several years there has been a program called G450C, working on the feasibility of 450mm (18") wafers. There seems to be a full equipment flow for this, but there is no enthusiasm on the part of the IDMs and foundries to make the massive investment necessary to build a 450mm fab. So it looks like 300mm will be the standard for some time. The big issue is that the equipment industry is unclear how they will ever recover their investment if the transition takes place since it takes many process generations. Fabs for 10nm are being, or have already been, built so the earliest process that could use 450mm would be 7nm. An old SEMATECH study from 1997 showed that a wafer size remains in production for approximately 24 years so that everyone can recover their investment.That would take us to 2040 or beyond, way past any visibility we have on semiconductor processes.
To prove 450mm is really a thing, the picture to the left is from SEMICON West last year, me holding a 450mm wafer. A little bit of fab trivia for you, too. The orange tape on the plastic shell containing the wafer shows that the wafer has copper on it. In the days when interconnect was built using aluminum, or titanium/tungsten, copper was considered a horrible contaminant that would never be allowed anywhere near a fab. But then we switched to copper interconnect, so we had to work out how to do that safely. One minor change was that once a wafer has copper on it, it is marked with orange tape so it can be handled safely without causing contamination. I think once it has orange tape on it, it is never again allowed in the FEOL where the transistors are manufactured. Of course, in volume manufacturing the wafers are handled entirely automatically, but during technology development, single wafers are often processed and handled manually.
At the SEMI/Gartner symposium on the Monday of this year's SEMICON West, Christian Dieseldorff presented 200mm Fabs Awaken: A Glimpse into the Past, Present and Resurgence of 200mm Fabs. Christian works for SEMI and is the person who maintains the fab database. He knows every fab that is being built, being decommissioned, or expanded. A few years ago there was basically no market for 200mm equipment since nobody was building any new 200mm fabs nor expanding existing ones (except to convert to 300mm). Installed capacity at 200mm fell 7% between 2006 and 2015, but it is forecast to increase by 8% by 2019.
The US and Japan both closed 14 200mm fabs each between 2008-14, which makes this trend even more surprising. Another 15 fabs were converted from 200mm to 300mm. However, since 2010 there have also been 19 new 200mm fabs built and another 17 are expected to begin operations in the 2015 to 2019 period. The 200mm fabs that closed have some interesting stories among them. Two changed back to 4" for LED (Samsung line 4 and 5). Two Fujitsu fabs have been repurposed into hydroponic agriculture to grow salad greens. The former Qimonda fab in Richmond, Virginia, has been converted to a datacenter. A Western Digital fab in Phoenix is being turned into restaurants, retail, and housing.
The combination of these closings and openings is that by 2018, we will be back to 2006 levels of 200mm capacity. Since 300mm came online, a new 200mm fab would typically be equipped with used equipment since it was cheap and available. But now there is a shortage, and so new fabs are having to purchase new equipment. You can see the total 200mm fabs declining until last year, and then growing fast again. This graph excludes LED, pure BEOL fabs (metalization), and R&D fabs.
The driver for all of this is not leading-edge digital processes. Nobody is going to build a 10nm fab with anything other than 300mm equipment. But power devices, MEMS, and IoT do not require leading-edge processes and don't really have the volume of wafers that makes 300mm attractive. To be economical, 300mm fabs have to be large (50,000 wafer starts per month, say). 200mm capacity passed 150mm capacity in 1998 and reached its peak in 2007. In 2008, installed 300mm capacity passed 200mm capacity. However, 200mm capacity has also been increasing slowly and by 2019 should be almost back to its 2007 peak. In fact, between 2015 and 2018, 200mm capacity will increase by 274,000 wafer starts per month.
Where are all these new fabs? The table below shows that they are all over the world (except Korea). The big growth is six new fabs in China. Christian said it is actually one more than that since he only just found out about another fab. For the first time ever, China will be the second biggest market for semiconductor equipment, behind Taiwan at #1. Of course, this includes 300mm fabs, too.
The result of this (and closures, migration to 300mm) encompasses the worldwide capacity changes shown in the graph below. In blue are "existing" fabs, which means fabs that began operation before 2010, and in red are new fabs that began operation in 2010 or later.
If this is not already more than you want to know about 200mm fabs, SEMI is producing a 200mm Fab Outlook available some time later this quarter. Information about the fab database, which covers all fabs and all wafer sizes, is on the SEMI website.
Previous: Economic Uncertainty, the Global Economy and Semiconductors