Oct. 29, 2019 — From the earliest instance of smelting iron ore into metal or weaving fibers into cloth, humans have looked for ways to repurpose materials around us to use for specialized tasks. In recent decades, researchers have increasingly turned to high-performance computing (HPC) to aid in the next frontier of materials science research—in order to imbue materials with new and novel properties, scientists need to be able to understand and manipulate materials at an atomic level.
Recently, researchers led by Prof. Lars Pastewka from the University of Freiburg have been using Gauss Centre for Supercomputing (GCS) resources at the Jülich Supercomputing Centre (JSC) in order to study atomic-level material interactions of nanolaminates, specialized materials that, when combined and structured correctly, exhibit increased strength, hardness, and wear resistance, among other properties.
“Our motivation for studying nanolaminates is focused on trying to develop structural materials with enhanced mechanical properties,” Pastewka said. “These types of materials can have large implications for automotive or aircraft design, but this research is trying to understand their properties at the fundamental level.”
In combination with the experimental groups of Ruth Schwaiger at the Karlsruhe Institute of Technology (now at Forschungszentrum Jülich) and Guang-Ping Zhang at the Shenyang National Laboratory for Materials Science, the team unraveled their mechanical properties at the nanoscale, by combining the high-fidelity experiments with the ability to observe nanosecond, atomic interactions in molecular dynamics simulations. The team’s recent work, focused on nanolaminates made of copper and gold, was published in MRS Communications.
Nanolaminates are a class of composites, meaning they are made up of multiple materials, which exhibit properties that significantly differ from the sum of the individual parts. The interface between the individual nanolaminate layers (in many cases, only several nanometers thick) provide resistance to irreversible deformation of the atomic-level crystal structure. The composite material then exhibits larger strength than the individual components.
In order to design materials with these specialized properties, researchers need to understand how the material reacts to its environment, specifically to external stress. An elegant way of testing the mechanical properties is to manufacture small “pillars” of the material, typically by removing the material surrounding the pillar using a focused ion beam. These pillars are then deformed with a hard flat punch while researchers measure the applied force. When researchers interpret these experiments, though, they don’t often take into account that surfaces—of the pillar and of the flat punch—are never perfectly flat.
While many materials look smooth to the naked eye, at the atomic level, every material exhibits rough, uneven surfaces. Peaks on these surfaces serve as the points of intimate atomic contact in situations such as when pushing down on the pillar. The contact geometry is important to understand where materials are actually touching one another at the atomic scale.
“Roughness has implications in materials science, because the force is transmitted at just the contacting peaks,” Pastewka said. “Thinking about pressure, this means that the local pressure experienced by the surface can be orders of magnitude higher than the apparent applied pressure because the real area of contact is much smaller than we naively think it is. True contact between any two materials happens at the smallest scales.” The point where these interactions happens, the interfaces, is the focal point for researchers.
While the fine details of surface roughness are largely uncontrolled in experiments, computer simulations allow to control the position of every atom in the system and hence creating surfaces that are perfectly flat or that have controlled roughness.
The team applied pressure to different-sized nanopillars with controlled roughness, and then compared the results with experiments. The simulations showed that perfectly flat surfaces lead to homogeneous deformation of the pillar, but introducing roughness induced failure of the pillar through “shear bands”. These shear-bands are also observed in experiments. Shear bands start deforming a structure on a local level and continue deformation along “shear-bands,” which leads eventually to fractures. The simulations revealed that a simple atomic step on the surface is sufficient to induce failure through shear banding. The type of deformation experienced by the material is highly sensitive to the small-scale features of the surface.
In order to simulate pillars large enough to get an accurate representation of the experimental nanolaminates, the researchers needed access to HPC. “We need to do large-scale simulations in our research so we can connect to the experiments,” Pastewka said. “Our largest simulations contain around half a billion atoms and are carried out at the same scale as the experiments, something that can only be simulated on leading supercomputing resources such as those at GCS. Simulation results match the experiments both qualitatively and quantitatively.” While the experiments serve to validate simulation results, the simulations allow monitoring the motion of every individual atom and to control every detail of the virtual reality of the simulation, including surface roughness.
These results are a first step to designing nanolaminate materials that avoid failure through shear-banding. The team’s research helped clarify that shear-banding instability is tied to surface roughness. While surface roughness cannot be avoided, research should focus on designing materials that do not stabilize shear-bands. The team suggests that this could be achieved by looking for nanolaminates with components whose elastic constants match closely.
The team’s nanolaminate research has laid the groundwork for its next focus area—friction and wear. Pastewka indicated that studying friction adds a significant challenge for both experimentalists and computational scientists, but that the team’s nanolaminate research can help inform models used in friction simulations. Nanolaminates are useful model systems for friction research because experimentally the deformation of the initially straight layers can be traced by just looking at them. Compared to the relatively straight, flat, and uniform geometry of the pillar surface, studying friction requires that the researchers focus on spherical objects’ contacts with flat surfaces, a more complicated computational challenge.
That said, HPC can help enable insights into friction models that would otherwise be impossible to observe. “If you run a friction experiment, you can only observe things from the past because the interface is buried,” Pastewka said. “If you have to wait to look at the surface after the experiment, most of the interesting things have already happened. We know that nanolaminates have this straight geometry, and trace while they deform. Friction experiments have shown us patterns and vortices that look almost like cloud formation. Looking at these nanolaminates experimentally has opened up very interesting questions, because we can see phenomena that aren’t well understood, but then we can use computing to try and make sense of pattern formation.”
Source: Eric Gedenk, Gauss Centre for Supercomputing (GCS)