Materials Science: The Modern Philosopher’s Stone

The Philosopher’s Stone is a legendary substance that was believed to have the ability to turn base metals into gold. While the stone really is just a legend, we are using materials science and high performance computing (HPC) resources to create novel materials more valuable than gold!

Imagine the applications of materials that can withstand extreme heat or resist cracking or reduce weight while retaining strength? Manufacturers in automotive, aerospace, consumer electronics, power generation, plastics, construction (improved concrete formulations), and military equipment--any manufacturing industry that cares about the durability, strength, weight, and even appearance of their products--can benefit from the use of materials science.

What is materials science?

Materials science is the study of the properties that materials exhibit under various conditions, such as bending, cracking, corrosion, phase transitions (melting, etc.), and even combustion/explosion when subjected to loads, shocks, or extreme temperatures. These studies can (and must) be done at various temporal and spatial scales, ranging from picoseconds (one trillionth of a second) to days. They must also consider spatial scales of nanometers to meters. Depending on the specific properties, materials and scales under study, methods ranging from the quantum level, atomistic level, molecular level, particle level, and continuum level are used.

Researchers like me and others at SAIC apply molecular dynamics (MD) to study materials. MD is the computer simulation of a material done at a level which models the motions of individual particles (typically thousands to millions of atoms or molecules) as they interact with each other by inter-particle forces. One of the goals is to improve upon current insensitive munitions. With recent significant performance gains, we are at the cusp of being able to use the developed software to go beyond just the basic science.

HPC-enabled materials study: unlocking secret knowledge

The mythical Philosopher’s Stone was reported to unveil mystic secrets, which isn’t far from the abilities of materials science. If you want to see what is happening inside the combustion cloud behind a rocket without melting like a villain in an "Indiana Jones" movie, use HPC. Simulations in HPC enable the study of materials in conditions that are hard to reproduce experimentally, due to expense, hazards, or the inability to observe inside the material in a real experiment. As the mathematical models of materials are developed and improved, HPC simulations can greatly expand and accelerate the search for novel material properties based on variations in composition, microstructure and material heterogeneity (e.g., composites) without having to verify each individual simulation against physical experiments. This is analogous to the use of computational fluid dynamics to replace the bulk of wind tunnel tests for modern aircraft design.

FURTHER READING: High-performance data analytics uses high-performance computing and AI to tackle extreme data sizes

 

Why wait?

Gaining these insights (which would be impossible or prohibitively expensive by other means) is great, but if we can’t deliver insights in a timely manner, they are useless to decision-makers. DoD researchers can’t wait a full year for a model to run to find out how a new composite for a turbine is going to perform under extreme cold. That’s where computational scientists, who understand HPC and materials science, enter the story.

SAIC researchers apply HPC resources to solve materials science problems. Although applying thousands of processors with ultra-high bandwidth communications to the problem is the normal use of HPC by materials scientists, we don’t rest there. We employ performance engineering methods to materials science codes that run on these HPC resources. For instance, recent advances in simulation models for energetic materials in combination with code optimizations have enabled Army Research Laboratory scientists to collect data about material properties of an explosive in a few hours of simulation that previously required many years of simulation on the same supercomputer. This 12,000X gain in performance over atomistic simulation techniques is allowing ARL scientists to qualitatively change what questions they can ask and hope to answer about energetic materials.

You can imagine the importance that government and industry place on this discipline, as they look to enhance their systems. While those unfamiliar with these concepts might gloss over the mention of molecular dynamics, materials science, and HPC, in the hands of skilled practitioners, they become the Philosopher’s Stone of the future.