We present an overview of lithography results achieved for materials to support "leave-on-chuck" double-exposure
pitch-division patterning. These materials attempt to make use of a non-reciprocal photoresponse in which the same
number of absorbed 193nm photons can produce different remaining levels of resist, depending upon whether the
photons are received all at once or in two separate exposures. This, in principle, allows for the use of two exposures,
using independent masks and without removing the wafer from the chuck, to produce non-regular patterning down to
one half the pitch limit of the scanner. Such behavior could be produced, for example, by a reversible two-stage
Photoacid Generator (PAG) or other non-reciprocal mechanisms.
Several stages of lithography screening were done on a large number of candidate systems. Initially, thermal stability,
casting behavior, and single-exposure (SE) contrast curves were investigated to determine whether the system behaved
as a usable photoresist. The next stage of testing probed non-reciprocal response, in the form of double-exposure (DE)
contrast curves, typically with an intervening whole-wafer flood exposure at a longer wavelength to enact the nonreciprocity.
The key criterion for the material to pass this stage was to show a shifted contrast curve (difference in
photospeed) for DE vs. SE. Such a shift would then imply that pitch-division imaging would be possible for this
material.
After identifying materials which exhibited this SE vs. DE contrast curve shift, the next step was actual DE patterning.
Since the laboratory tool used for these exposures does not have the precise alignment needed to interleave the two
exposures for pitch division, we employed a technique in which the second exposure is rotated slightly with respect to
the first exposure. This results in a Moiré-type pattern in which the two aerial images transition between overlap and
interleave across the wafer.
One particular PAG + sensitizer did indeed show the desired DE vs. SE contrast curve shift and pitch-divided imaging (k1 = 0.125). This system appears to operate on a scheme based on the creation of a photobase generator between the first and second exposures. Unfortunately, the quality of the pitch-divided images degrades quickly as the pitch is decreased, showing severe LER and bridging defects at a final pitch of 220nm. We postulate that this is caused by the diffusion of one or more key photoproducts. Accompanying papers report on both the photochemical details of the reaction pathways of these materials as well as modeling of the reaction kinetics.
|