About every three years, a new DRAM generation requires a new lithogra
phy system capable of imaging 3x more pixels, with a substantial impro
vement in resolution. This results in a 2-5x-3x increase in the cost o
f lithography lenses used for each generation and portends an increasi
ng cost per square centimeter of each layer of patterned silicon. Opti
cal lithography may be the least risky patterning technology through t
he gigabit DRAM generation, but DOF will be a major concern, and vario
us types of wave-front engineering will be needed to improve process r
obustness. Steppers based on 1x, mostly reflecting optics may reduce t
he escalation in cost of ownership. Such systems appear capable of mee
ting the lithography requirements of the 256-Mbit DRAM and, with evolu
tionary improvements, the 1-Gbit DRAM. Cost improvement is achieved wi
th an incremental change in current technology and, therefore, holds m
ore promise than new, more revolutionary technologies.