(613i) Molecular Simulation Investigations into the Root Causes of Bridge Defectivity in Block Copolymer Directed Self-Assembly

In 1965, Intel cofounder Gordon Moore predicted that the areal density of transistors patterned on microelectronic devices would double approximately every 18-24 months. Although this prediction (now commonly referred to as Moore’s Law) was largely realized from the 1960’s through the late 2000’s, the rate of increase has slowed within the past decade as the desired feature sizes have shrunk below ten nanometers. The bottleneck in further feature size reduction is due to the physical limits of optical lithography, which is the iterative process of patterning and subsequently developing device features on a silicon wafer. The microelectronics community is well aware of the significant challenges which lie ahead in terms of developing alternative lithographic techniques that can be used to pattern features with sub-10 nm spacing. A few of the primary technologies being explored as possible extensions to traditional optical lithography include extreme ultraviolet lithography (EUV), nano-imprint lithography, electron-beam lithography, and the directed self-assembly (DSA) of block copolymers (BCPs).

BCPs have emerged as increasingly intriguing materials because integrating them into the semiconductor fabrication process is far more economical than other alternatives such as EUV lithography and complex multiple-patterning schemes. BCP materials have the inherent ability to microphase separate and form highly ordered nanostructures such as spheres, cylinders, gyroids, and lamellae. Cylinders and lamellae are the morphological structures that are of particular interest to lithographers because of their potential to be utilized in contact hole and line-space patterning, respectively. Although these nanostructures are naturally unaligned, directed self-assembly (DSA) techniques such as graphoepitaxy and chemoepitaxy make use of either physical or chemical guiding patterns on an underlayer to order and align these features into periodic, perpendicular structures. Chemoepitaxial processes use a chemically neutral underlayer that is patterned with periodic chemically preferential regions which attract one of the polymer blocks. Experimentally, this is achieved by synthesizing a brush underlayer that has periodic guiding stripes which selectively interact with one of the two BCP blocks; meanwhile, the rest of the underlayer is slightly preferential to the opposing block. Modifying the energetic landscape of the self-assembly process can transform what would ordinarily be a defective fingerprint in a lamellae-forming BCP into a well-ordered line-space type pattern. Ideally, BCPs will coat a surface, phase separate, and leave behind a patterned mask that can be transferred into the silicon substrate via etching. Recent advances in DSA processing have drastically reduced the observed density of defects such as dislocations and disclinations in BCP thin films; however, the so-called bridge defect has become an increasingly problematic defect mode within the past decade.

A great deal of effort has been placed into investigating bridge defectivity in various DSA processes including the SMART™ process, COOL process, and LiNe process. Sato and coworkers used self-consistent field theory to investigate the effects of so-called underlayer affinity defects on bridge defectivity in the COOL process. They found that both the frequency of bridging and bridge thickness increased with increasing affinity defect size; however, the PS-b-PMMA model used in this study had a volume ratio of 11:9 PS:PMMA, which likely also contributed to the observed thickness. Henderson and coworkers investigated underlayer affinity defects in chemoepitaxial process flow and their ability to spawn bridge defects in an overlying BCP film with a 50:50 A:B volume ratio. Although there was a positive correlation between bridge propagation and the size of the underlayer affinity defect, propagation primarily occurred laterally and was limited to the bottom of the BCP film. These results indicate that underlayer affinity defects are not the sole root cause of bridge defectivity. Follow-up studies by Henderson and coworkers also suggested that bridge thickness is not significantly affected by BCP material properties such as χ and N. Further work must be conducted to determine which factors in DSA processing contribute to bridge defectivity in order to mitigate them during fabrication. In this study, coarse-grained molecular dynamics simulations are used to systematically investigate errors in material synthesis and DSA processing and their effects on bridge defectivity. The processing errors studied include polymer dispersity, homopolymer contamination, and varying A:B volume ratios. These results will show whether any one factor can individually induce the propagation of thick bridge defects, or if some combination of these factors must be present.