Space experiments are key to better crystal-growth modeling

WEST LAFAYETTE, Ind. – Information gleaned from space experiments is enabling a Purdue University engineering professor and his graduate students to design new software aimed at manufacturing superior crystals for electronics and other industrial applications.

Better crystals would improve the quality of electronic hardware, including computers, and enable engineers to design superior alloys for a wide range of applications.

Because gravity masks the fine details of how crystals form, making them in space in the near-absence of gravity is uncovering information critical to designing better crystals and alloys. The earth-grown crystals are fundamentally subjected to the same molecular effects as those grown in space, but the effects can only be studied precisely in the low gravity of space. Those details are then incorporated into mathematical models that might ultimately be used to manufacture tailor-made crystals perfectly suited for specific applications.

“The goal is to be able to grow materials to exact specifications,” says Suresh Garimella, an associate professor of mechanical engineering at Purdue. Garimella has designed the new software with the help of data from French-built crystal-growth experiments flown on four space shuttle flights.

“Our model predicts the experiments very well, ” says Garimella, who will discuss new findings about the work during an international conference this month. The ultimate goal, he says, is to be able to custom-design better materials without needing elaborate space-based research.

“It would be wonderful if you never had to do any experiments and you could just use computer programs to predict everything,” Garimella says. “But to develop such computer programs, you need to first establish confidence in them by comparing their predictions to the actual results seen in experiments, and that’s the stage where we are now.”

Findings from his research will be detailed in a scientific paper to be presented on Aug. 23, during the Ninth Conference on Modeling of Casting, Welding and Advanced Solidification Processes. The conference, in Aachen, Germany, is sponsored by the United Engineering Foundation, a non-profit organization based in New York that provides grants for the advancement of engineering.

Liquids, such as water or molten metals, become crystalline solids as they are cooled. Garimella is using revelations about the physics of that crystal “growth” to write computer programs for new models based on mathematical equations. His most recent model has been shown to accurately predict the final characteristics of crystals, given certain processing conditions such as the temperature at which the crystallizing liquid is cooled, how fast it is cooled and various material properties, including its melting point.

The Purdue engineer uses data from a French-designed experiment, called MEPHISTO, an acronym for Materiel pour L’Etudes des Phenomenes Interessant la Solidification sur Terre et en Orbite. The English translation is Materials for the Study of Interesting Phenomena of Solidification on Earth and in Orbit.

“We couldn’t do these experiments on Earth because gravity is so strong,” Garimella says. Gravity causes liquids to flow in convection currents, in which lighter materials rise and heavier materials fall. The movement makes it difficult to analyze how liquid molecules turn into solid, crystalline materials.

“All you see are the effects of gravity, essentially,” Garimella says. “What is helpful is if you can take out gravity and study how a liquid molecule becomes solid in the absence of gravity. The Holy Grail is to be able to predict what kind of solid will form, given certain processing conditions.”

During the past four years, Garimella has been developing mathematical models to do just that. The complex models show how crystals grow at the microscopic level.

The research has revealed a surprising detail: Scientists have known that the crystalline surface where solid first starts forming from liquid has a curving, concave shape. In crystals grown on Earth, that curvature is much more pronounced, and Garimella has found that the nature of the curvature has a strong effect on the subsequent growth of the crystalline solid, which can then be predicted accurately with mathematical models.

“We are in the forefront of developing those models and making them better,” Garimella says, noting that models currently available to industry do not capture enough of the physics of crystal formation. “We model it from a fundamental physics point of view,” he says.

“Obviously, industry makes pretty good silicon crystals now, otherwise computer technology wouldn’t be where it is today,” he says. “However, to go from 99 percent purity, to 99.999 percent purity, we need to know a lot more. The MEPHISTO effort was designed to understand some of the things that will help us make even purer crystals.”

After the crystals were grown during space shuttle missions, they were studied by scientists at the University of Florida, who provided raw data for the models developed at Purdue.

Source: Suresh Garimella, (765) 494-5621, sureshg@ecn.purdue.edu

Writer: Emil Venere, (765) 494-4709, evenere@uns.purdue.edu

Purdue News Service: (765) 494-2096; purduenews@uns.purdue.edu

NOTE TO JOURNALISTS:   A copy of the research paper referred to in this news release is availble from Emil Venere, (765) 494-4709, evenere@uns.purdue.edu.

An experimental and numerical study of the horizontal Bridgman growth of pure succinonitrile (SCN) has been performed to explore the effect of melt convection on the interface propagation and shape. Measurements obtained both under conditions of no-growth and for a 40 mm/s growth rate included interface shape and location in the horizontal and vertical planes, melt velocities and temperature boundary conditions. The growth front was stable and non-dendritic, but was significantly distorted by the influence of convection in the melt and, for the growth case, by the moving temperature boundary conditions along the ampoule. Three-dimensional numerical simulations were performed for the no-growth and growth cases, using a primitive-variables approach; for the growth case, the simulation was fully transient. Temperatures throughout the phase change material and ampoule as well as melt velocities were obtained from the simulations. The predicted interface shapes and melt velocities agree well with experimental results. In ongoing work, this approach is being applied to numerical simulations of the Bridgman growth of a binary alloy.