 |
|
Terahertz Spectroscopic / Imaging Analysis System
|
|
|
|
Electronic Measuring Instruments
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Best Solution for Small-Lot-Multiple-Type Production Lines
|
|
|
|
|
|
|
Changes at IC Device Production Sites |
|
|
|
|
Having entered the age of system LSIs, IC makers are changing. If we make the assumption that the number of ICs produced from the beginning to the end of production is the lifetime production quantity, Figure 1 shows the lifetime production quantity on the horizontal axis and the percentage of models produced at plants on the vertical axis. In the heyday of general-purpose memory, several million units of the same product were produced. For most system LSIs models, however, the range is in the tens of thousands and does not exceed one hundred thousand units. Models that ended with a production volume of no more than 100 account for a rather high percentage if we include those models that, for whatever reason (includes models in their development phase), did not reach the commercial production stage.
Our current age of system LSIs has confronted chip-makers with many changes. Figure 1 shows lifetime production quantities (the total number of units produced) graphed against the number of different models produced. In the heyday of general-purpose memory, a typical product had a lifetime production quantity of several million units. But for most modern system LSI models, the lifetime production quantity does not exceed 100,000 units. Models with a total production volume of no more than 100 units account for a rather high percentage if we include those models that did not make it out of development or for some other reason never went into commercial production.
Every time a new generation of miniaturized devices appears, the cost of masks with their indispensable phase shifting and optical proximity effect correction technologies is seen to double. As an example, assume that one set of masks for 130nm rule devices costs 100 million yen, and 2.3 redesigns are needed. If a given device has a lifetime production quantity of 10,000 units, the cost of a mask for one chip then runs to 23,000 yen. At these costs, the system LSI business is not economically viable.
|
|
|
|
 |
|
Figure 1 System LSIs production volumes
|
|
|
|
|
Comparison in COO with an ArF Model |
|
|
|
|
Figure 2 compares the COO (Cost Of Ownership) of an EBDW model and an ArF immersion lithography model.
Graphing lithography costs against lifetime production quantities, we see that when a chip-maker produces 1 ~ 99,999 units of a given model, EBDW technology offers better COO, even if the EBDW equipment has a throughput of just 1 wafer every 2 hours. This advantage is prediced to grow in future with the widespread adoption of block exposure and multibeam technologies.
|
|
|
|
 |
|
Figure 2 COO comparison for EB direct-write model and ArF model
|
|
|
|
