I agree that for many semi-manufacturers, cost of the equipment is not the most critical issue, but it is certainly non-trivial. Periodically, you need to inspect the output from every tool. If you assume a 4 chamber cluster tool, and you want to design a Fab where you never have to remove a wafer from a tool for inspection, then you might include one inspection devise per tool. Each cluster tool would now have a reduced throughput of 25% (only 3 operational chambers), such that you would have to buy 33% more cluster tools. Consequently, you would also have to increase the floorspace in your cleanroom by 33%. Finally, this simple analysis assumes that you are going to use a single inspection devise for that type of tool (ex., just SEM). My impression is that one may use multiple, different inspection tools for examining wafers for defects (This statement is based solely on the observation that KLAC offers multiple, different inspection tools, and, therefore, could well be incorrect.). In this case, you would want to include multiple inspection tools in each cluster, further diminishing throughput.
An alternative strategy that AMAT might be set up to employ would be to include multiple, different inspection tools within a cluster. Now, you could send a wafer through several different inspections without removing it from the tool, and this might be cost effective. |
