A tiny, invisible crack in the aluminum-alloy skin of an airplane can sometimes grow into a major crack that causes the skin to tear open in flight. A new research program at Cornell University aims to understand how this happens, starting at the level of atoms and working up. Eventually this understanding could lead to improved alloys and designs to prevent cracking and failure in many materials.

The National Science Foundation's Knowledge and Distributed Intelligence Initiative has announced a $1.5 million, three-year grant to the Cornell Theory Center (CTC) to support part of this research. The project being funded, called Multiscale Modeling of Defects in Solids, will enable researchers to create computer simulations that show how defects at the atomic level can lead to changes at increasingly larger scales, up to the visible cracks we can see and measure in the everyday world. CTC is Cornell's high-performance computing and communications arm.

"We're trying to understand how existing cracks grow and what kinds of defects cause little cracks to begin. At the beginning is microscopic physics," says James P. Sethna, Cornell professor of physics, who is the principal investigator for the work. "The goal is to understand how metals and other solids form defects as they age and fatigue; understanding this might allow us to make the material different so cracks don't form as quickly, to make the material different so cracks don't grow as fast."

Multiscale modeling, Sethna says, is one aspect of a larger project called the Digital Material, a software "environment" where researchers can build simulations of materials on many levels. It is an interdisciplinary project that involves, along with Sethna, Cornell investigators Christopher R. Myers, senior research associate at CTC, Paul R. Dawson, professor of mechanical and aerospace engineering, and Anthony R. Ingraffea, the D.C. Baum Professor of Engineering in the School of Civil and Environmental Engineering and associate director of CTC.

Members of the team have been looking at materials on several different scales, Sethna says. He himself is interested in the behavior of atoms as they line up in crystal lattices and how a defect in a crystal -- such as a missing atom or a missing half-sheet of atoms -- can multiply and spread through the crystal.

Dawson works on a scale that is still microscopic by human standards but gigantic compared to atoms. The metals that combine to form an alloy don't mix at the atomic level; rather, the alloy consists of tiny grains of the different metals mixed together. Many grains interact to produce a higher level called "texture," which has to do with how the grains line up.

Finally, Ingraffea looks at the macroscopic world where cracks become visible. He sometimes illustrates his work by displaying pictures of airplanes whose skins have torn open in flight.

There are several other levels researchers use for different purposes. There are computer simulations for all of them, but up to now there has been no way to relate one to another.

"Dawson wants to connect textures up to cracks and down to grains," Sethna says. "My mission is to connect atoms upward to defects and texture; Ingraffea's is to get texture linked to cracks. We want to build on the honesty of the low levels to get honest descriptions in the high levels."

Myers, the leader of the group's software development efforts, plans to do this using a computer language called Python (named for the British TV comedy show), which will supply the "glue" to allow computer simulations to communicate with one another. It is a "scripting" language that can be used to control programs written in many different computer languages for many different purposes.

"We are trying to develop tools for piecing together lots of different software components," Myers says. "One of the problems in these big simulations is that you find yourself trying to build one huge application that does everything. We want to have something more modular that lets a researcher drag in smaller components as needed."

Several ongoing projects will funnel into the Digital Material work. Ingraffea and Dawson are working with Cornell engineering professors Matthew Miller and Mircea Grigoriu on probabilistic descriptions of fatigue cracking, with funding from the Air Force Office of Sponsored Research. In addition, Ingraffea and Cornell computer science faculty members Keshav Pingali and Steve Vavasis are developing tools to simulate crack propagation on TeraFLOPS computers, with NSF funding.

The Digital Material environment can be used to simulate many materials other than those used in aircraft skins, Sethna adds. One possibility is to study defects that can develop in wires on computer chips. Another, which Dawson is investigating, is the development of texture in the Earth's mantle.

"This program is not only about organizing information from different length scales, but also to allow people from different areas to be able work together in a focused effort," Myers says.