de Pablo Group

Ludwig Schneider

Ludwig is a computational material designer for polymeric and soft matter using traditional simulations and machine learning. His background is in physics, but he combines it with software engineering, chemistry and data science for various research projects. Ludwig pursued his undergraduate and graduate education with Prof. Müller at the Georg-August University Goettingen in Germany investigating rheology and structure formation in complex polymer melts. After being granted his Ph.D degree he moved to Juan de Pablo's group at the Pritzker School of Molecular Engineering, University of Chicago to enhance the skill set to machine learning and chemistry.

Block Copolymers

Self-assembly of block copolymers allows access to the nano meter scale by controlling the chemistry of the polymeric blocks. Understanding the formation, processing and properties of the resulting morphologies is a challenge with many open questions. The materials have to be tailored for each of the possible applications: battery electrolytes, molecular membranes, or optical meta-materials.

One example research area is battery electrolytes which requires two major properties: i) ion-conductivity and ii) mechanical stability. Diblock copolymer materials offer the opportunity to provide each property with one of the two blocks. The chemistry of each block can be optimized to excel for its job. However, the morphology needs to be bicontinous to provide their functions macroscopically to the bulk, which is at odds with equilibrium phase diagram. I investigate the suitability of kinetically trapped morphologies for their application on engineering length scales up to micrometers with simulation of billions of particles - only possible via HPC GPU simulation with the software SOMA.

Multi-Scale Entangled Polymer Dynamics

Multi-scale Entangled Polymer Dynamics is another challenging field. The fractal universality of long, flexible polymers allows subsuming many atoms into coarse-grained interaction centers, resulting in a Gaussian Chain model. The static polymer conformations are well understood and reproduced by such coarse-grained models. However, the dynamics is not necessarily conserved in this process. Extra models have to be developed to cover dynamics and statics aspects correctly. Coarse-graining the interaction beads of long coarse-grained polymers softens the interactions of the repeat units from strong Pauli repulsion to non-diverging potentials. This procedure reduces the degrees of freedoms dramatically and is necessary to model long polymers with modern compute hardware, even as they are GPU accelerated. Unfortunately, the soft repulsion does not prevent chain crossing anymore. In order to regain the correct entangled dynamics slip-springs can be introduced into the model.

Machine Learning

Machine learning techniques are fascinating tools to explore more physics of interesting systems previously deemed intractable. Here we use machine learning to investigate the vast chemical space to tailor system for application-driven properties. In addition, we use machine learning to accelerate computational simulations with new methods that enable insights into extremely protracted kinetics, e.g. diblock copolymer morphologies in rugged free-energy landscapes.