Today's leading-edge semiconductor chips are made on thin wafers measuring 300 mm (12 in.) in diameter. Before the advent of wafers of this size, automation of container handling and transport in semiconductor wafer fabs was virtually non-existent.
But gone are the days when workers carried cassettes of wafers from one tool to the next. Full 300 mm adoption in the early 2000s brought with it a bevy of automation tools incorporating front-opening unified pods (FOUPs), carrying 25-wafer lots; equipment front-end modules (EFEMs), which shuffle wafers from the FOUPs to equipment process chambers; standardized load ports; and automated overhead transport vehicles.
Along with the 300 mm era came a substantial increase in factory size and wafer throughput, placing a heavier burden on automation systems; increased travel distance coupled with more FOUP transactions per hour has a direct effect on overall factory cycle time and automation complexity.
Discussions about moving to 300 mm wafer sizes began in 1994, but the industry didn't truly have manufacturing-ready systems until the 2002-04 timeframe. In 2005, just when we had mastered 300 mm manufacturing, industry organization SEMI began holding meetings about a move to 450 mm wafers to keep productivity on a Moore's Law curve. A potential die area that's 2.25 times larger than the 300 mm die area and a goal of maintaining the current tool throughput rates provide at least a twofold increase in productivity.
There has been skepticism from machine builders about the expense of moving to 450 mm platforms vs. improving 300 mm productivity, but standards finally are in place this year that allow companies to develop 450 mm platforms with confidence, with pilot production aimed for the 2017-21 timeframe. To get down the learning curve in terms of process tool development, metrology and automation, work must start immediately. One of the first building blocks needed by process tool manufacturers is the wafer-handling equipment.
In developing some of the first 450 mm wafer-handling platforms, we needed to innovate in several areas, particularly in relation to scale-up size and weight. For example, the new load ports have to be able to handle much heavier FOUPs—in the 50-60 lb range for 450 mm, compared with 20 lb for 300 mm. That not only makes them completely unmanageable for human loading, but also requires a much larger port door opening and increased rigidity needed to maintain wafer-mapping accuracy.
The load port itself has gone from weighing 70-100 lb for 300 mm wafers to more than 300 lb for 450 mm. Whereas load ports today are typically moved away from the equipment for maintenance, the new load port will need a mechanical assist or multiple people to move it, which is not desirable when maintenance is called for service at 2 a.m. to get a tool up and running again. To address this, we've transitioned from the typical welded tube frames to a unified load port frame designed to be serviced through removable modules.
Another innovation relates to the increased distance that must now be traveled across the EFEM, which can hold four load ports (or more). The majority of EFEMs today use a heavy SCARA-type robot, which is centrally mounted to reach up to four load ports. We've trended instead to track-type robots that are lighter, and can slide back and forth on linear rails, bringing the robot to the load port. It's a unique design that uses a folded Z drive to enable a lighter and faster robot in a smaller space.
Concurrently with the change in wafer size, the industry is going through other significant technology changes that add automation challenges, including a general rise in the number of process and metrology steps while maintaining—or reducing—total factory cycle time. Also, there is growing interest in increasing molecular-level environmental control by nitrogen-purging containers and EFEM mini-environments.