(709f) Detailed Mesoscale Molecular Dynamics Simulation of Block Copolymer Phase Separation: Probing the Fundamentals of Directed Self-Assembly Processes

Directed self assembly (DSA) of block copolymers, which relies on the use of surface features to guide the phase separation of the block copolymer into structures with long range order and precise registration, is a very promising technique for producing sub-30 nm pitch regular patterns. These patterns could be used as an enhancement to current lithographic techniques for manufacturing integrated circuit features below the current minimum pitch of ~80nm provided by leading edge 193nm lithography tools.  In order to speed the development of such techniques and to understand the behavior and limitations of such processes, computer simulation of such methods are proving invaluable. A variety of computational  modeling approaches have already been used to explore the behavior of block copolymer phase separation and directed self-assembly, including techniques based on mean field approaches and Monte  Carlo techniques.  While these methods provide great computational expedience, in general they have suffered from a number of disadvantages, such as the use of less accurate potentials and the requirement that a subset of dynamic moves be intelligently chosen, that can affect the accuracy of the predicted outcomes from such simulations.  Conversely, molecular dynamics combined with realistic potentials for polymer behavior can potentially provide more accurate simulations of the inherent polymer behavior,  dynamics, and equilibrium states without a need to guess modes of molecular movement and without oversimplifying interatomic interactions.  In fact, in theory, atomistically detailed molecular dynamics could be used to carefully  study such DSA processes, but the large spatial size of the domains of  interest in DSA processes cause the number of atoms required in such an atomistic simulation to be infeasible practically at the current moment.  In addition, the number of  molecular dynamics steps required to study block copolymer phase separation under realistic simulation conditions  (i.e. realistic temperatures and interatomic potentials) in fully atomistic simulations is also currently not practical.  In this work, the number of atoms required to simulate a useful size of polymer during the DSA process has been reduced by coarse graining the atomistic polymer models into a mesoscale model form based on Kuhn segment beads to represent the polymer chain.  Such mesoscale models have been developed and carefully parameterized to reproduce the behavior of potentially useful DSA block copolymers (e.g. PS-b-PMMA).  The techniques used for this model development and validation will first be discussed.  These mesoscale models have been implemented into MD simulations using GPU-based computing based on a modified form of the HOOMD-Blue software package.  In performing bulk polymer phase separation simulations in periodic boundary conditions, there are geometric constraints to the allowed pitches for the resulting phase separated materials and thus the simulation box conditions must be carefully chosen and varied in order to properly sample the configurational space and determine the true equilibrium states for the phase separated polymer morphology.  A simple rule and analysis methodology to guide the selection of such periodic simulations will be discussed.  Finally, use of the simulations to probe several interesting questions related to block copolymer phase separation and DSA will be discussed.  First, the scaling of the pitch of phase separated patterns with the polymer degree of polymerization and chi value will be analyzed.  This scaling is an important issue for those designing polymers to achieve selected target pitches for integrated circuit manufacturing.  Second, it will be shown that the simulations faithfully reproduce the expected order-disorder transition (ODT), and the effect of polymer polydispersity on the ODT will be elucidated through simulation.  Finally, the concept of an effective ODT for thin film block copolymer DSA processes will be introduced, and this effective ODT will be discussed in terms of relevant material and DSA process parameters (e.g. polymer chi value, polymer molecular weight, film thickness, etc.).  It will be shown in general that such molecular dynamics simulations are greatly aiding in the practical understanding and design of such DSA materials and processes.