For all of the real progress and the hype around 450mm, there is one thing that is for certain, that is, you can’t develop and test a new 450mm tool and its process without patterned wafers. This has been the Achilles’ heel many OEMs have been concerned about, namely, access to patterned wafers for process development and testing.
With 450mm intercepting the ~10nm technology node in 2018 (at least, that seems to be the latest projection), it is important for the various process tools to be proven out long beforehand. A stop-gap means of developing processes on a 450mm-wafer form-factor was to insert a 300mm single-crystal wafer into the middle of a sintered 450mm wafer. The sintered wafer has very similar mechanical properties to single-crystal wafer. While I’m sure this had some value for early adopters in developing various aspects of their tools, there is truly nothing like the real thing.
In fact, it takes many tests with “real-wafers” to fully develop a process step, and then many more to statistically validate that the process really works at the desired node. A few billion dollars were spent on test wafers in the 300mm transition, and though the G450C wafer loan and reuse program may help throttle the cost this time, the test wafer cost for 450mm has been estimated to be up to $5B US.
But perhaps one of the biggest, looming issues has been just getting patterned wafers into the hands of equipment manufactuers. Some of that concern was significantly diminished with the Intel’s announcement of the first 450mm patterned wafer last week. Bob Bruck, who is corporate vice president and general manager of Technology Manufacturing Engineering (TME), Intel Corporation, and Mario Abravanel, Intel 450mm Equipment Program Manager, displayed a fully patterned 450mm silicon wafer while at the Industry Strategy Symposium—the wafer was patterned with 26nm features using nano-imprint lithography from Molecular Imprints which is located in Austin, Texas. Thousands of similar wafers will be made available to suppliers engaged G450C.
So, is this just another, necessary, stop-gap on the way 450mm until EUV or quad-patterned lithography is available? Possibly. However, there are a few things that make me think there might, eventually, be more nano-imprint lithography in store for 450mm down the road.
First of all, by being the first company to pattern 450mm wafers in any significant volume, Molecular Imprints will be obtaining a tremendous number of learning cycles on their tool and process. This will give them a significant head-start when it comes to understanding issues with 450mm and with their own system. Second, depending on the pricing specifics (it will be much less expensive than EUVL), volume of resist used, tool throughput, defects, and tool footprint it could very well be that nano-imprint lithography may have cost-of-ownership that bears consideration for production use in 2018.
Granted, I’ll need to canvass friends in the industry to get their thoughts, however, unlike some previous announcements circa 2009 where aluminum and sintered mock-up 450mm wafers were shown at ISS and other conferences, it looks like this announcement is the real deal. And just in time. There’s a lot of development and testing that needs to be done on 450mm, and soon.