The first fully integrated SOI device using 42nm-pitch directed self-assembly (DSA) process for fin formation has been demonstrated in a 300mm pilot line environment. Two major issues were observed and resolved in the fin formation process. The cause of the issues and process optimization are discussed. The DSA device shows comparable yield with slight short channel degradation which is a result of a large fin CD when compared to the devices made by baseline process. LER/LWR analysis through the DSA process implied that the 42nm-pitch DSA process may not have reached the thermodynamic equilibrium. Here, we also show preliminary results from using scatterometry to detect DSA defects before removing one of the blocks in BCP.
Implementation of Directed Self-Assembly (DSA) as a viable lithographic technology for high volume manufacturing
will require significant efforts to co-optimize the DSA process options and constraints with existing work flows. These
work flows include established etch stacks, integration schemes, and design layout principles. The two foremost
patterning schemes for DSA, chemoepitaxy and graphoepitaxy, each have their own advantages and disadvantages.
Chemoepitaxy is well suited for regular repeating patterns, but has challenges when non-periodic design elements are
required. As the line-space polystyrene-block-polymethylmethacrylate chemoepitaxy DSA processes mature,
considerable progress has been made on reducing the density of topological (dislocation and disclination) defects but
little is known about the existence of 3D buried defects and their subsequent pattern transfer to underlayers. In this
paper, we highlight the emergence of a specific type of buried bridging defect within our two 28 nm pitch DSA flows
and summarize our efforts to characterize and eliminate the buried defects using process, materials, and plasma-etch
optimization. We also discuss how the optimization and removal of the buried defects impacts both the process window
and pitch multiplication, facilitates measurement of the pattern roughness rectification, and demonstrate hard-mask open
within a back-end-of-line integration flow. Finally, since graphoepitaxy has intrinsic benefits in terms of design
flexibility when compared to chemoepitaxy, we highlight our initial investigations on implementing high-chi block
copolymer patterning using multiple graphoepitaxy flows to realize sub-20 nm pitch line-space patterns and discuss the
benefits of using high-chi block copolymers for roughness reduction.
Wet chemical slimming of resist can enable a resist mandrel for sidewall-image transfer (SIT) by decreasing the mandrel width and smoothing the mandrel sidewalls. This would reduce the cost of the SIT process. Several key metrics are used to compare the traditional etched mandrel and the slimmed resist mandrel, including: process window, critical dimension uniformity, and defectivity. New resists are shown to have larger process windows after slimming than an etched mandrel process while maintaining comparable critical dimension uniformity. The major challenge to the resist mandrel is the profile post-slim.
The patterning capability of the directed self-assembly (DSA) of a 42nm-pitch block copolymer on
an 84nm-pitch guiding pattern was investigated in a 300mm pilot line environment. The chemoepitaxy
guiding pattern was created by the IBM Almaden approach using brush materials in
combination with an optional chemical slimming of the resist lines. Critical dimension (CD)
uniformity, line-edge/line-width roughness (LER/LWR), and lithographic process window (PW) of
the DSA process were characterized. CD rectification and LWR reduction were observed. The
chemical slimming process was found to be effective in reducing pattern collapse, hence, slightly
improving the DSA PW under over-dose conditions. However, the overall PW was found to be
smaller than without using the slimming, due to a new failure mode at under-dose region.
Directed Self-Assembly (DSA), as an extension of current state-of-the-art photolithography, has demonstrated the
capability for patterning with resolution and cost effectiveness beyond the capability of other techniques. Previous
studies of DSA have reported encouraging benchmarks in defect density and throughput capability for the patterning
step, and such results provide a foundation for our ongoing efforts to integrate the DSA patterning step into a robust
process for fabricating device layers. Here we provide a status report on the integration of two chemoepitaxy DSA
patterning methods for the fabrication of 28nm pitch Si fin arrays. In addition to the requirements for a robust pattern
transfer process, it is also important to understand the pattern design limitations that are associated with DSA. We
discuss some of the challenges and opportunities associated with developing efficient device designs that take advantage of the capabilities of DSA.