PREDICTING how materials will behave and perform is central to the success of major industries, the manufacturing of a vast range of consumer goods, and scientific research programs like those at Lawrence Livermore. For years, scientists have longed for computer simulations that could track changes in material properties on scales ranging from the atomic to the engineering. Such simulations would accurately predict material performance. But until recently, such simulations have been relatively crude or even impossible.
Now, because of the extraordinary computational capabilities of Department of Energy parallel-processing supercomputers, that has all changed. For the first time, we are successfully simulating the evolution of mechanical and chemical changes in materials and the performance of these materials in a variety of environments and conditions. These simulations, described in the article beginning on p. 4, are showing us in unprecedented detail how materials form and how they react under different environments. In so doing, the simulations suggest how we can change formulations or manufacturing methods to improve materials.
The advanced simulations are based on software that exploits the capabilities of thousands of small processors working together. Multiprocessing helps scientists solve some of the most perplexing problems in materials science, including understanding how and why materials crack, age, and ultimately fail; how semiconductors can be manufactured to greater tolerances for better performance; and how the three-dimensional structures of biological materials determine their function.
Because of the very nature of materials science, the simulation research is multidisciplinary. Livermore physicists, engineers, computational scientists, materials scientists, and bioscientists team up to develop the new simulations. The research is also based in part on laboratory experiments that serve to both motivate and validate them. We are thus confident that the simulations faithfully recreate the physical processes at work in materials.
One of the most significant attributes of our pioneering work is the vast range of the length and time scales of the simulations, from nanometers to meters and nanoseconds to tens of years. In this way we can trace, for example, the consequences of atomic-scale defects as they accumulate in time and space and eventually affect the performance of a vital part. Computer codes that successfully bridge different scales are essential because changes in material properties, such as radiation damage, can depend on phenomena occurring over many length and time scales.
The most immediate payoff for the simulations is stockpile stewardship-the Department of Energy's program to assure the safety and reliability of the nation's nuclear stockpile. The new simulations are also applicable to a large number of other Livermore research efforts, such as modeling optics damage for the National Ignition Facility, a 192-beam laser.
The potential of computational materials science has become obvious to American industry. The new generation of simulations will allow companies in almost every field to reduce the development time of new materials and more quickly ascertain the causes of material failures.
Clearly, we are entering an exciting era for materials science. Advanced computer simulations amount to a revolution in the field, one that will forever change this centuries-old discipline. Increasingly, computational modeling will complement both laboratory experiments and development of materials science theory. We even foresee the simulations minimizing the role and reducing the number of some traditional experiments that are largely based on trial and error. I am proud that Livermore scientists are among the leaders in the national computational materials science effort.

Back to December 2000