Frontiers Archives | Contact the Editor | Argonne National Laboratory
Frontiers2004
Click on image for larger view.
Sea urchin

New computer cluster quickly reconstructs three-dimensional or cross-sectional images of a series of X-rays. This sea urchin spine image provides researchers insight into its unique properties.


X-ray source focused on expanding research

The Advanced Photon Source at Argonne continues to develop innovative accelerator physics and technology and state-of-the-art X-ray research instrumentation, while providing a highly reliable source of X-ray beams to a rapidly growing user community. As the experiment hall space fills, emphasis is being placed on growing the community of general users—researchers who can benefit from the APS X-ray beams, but who are not affiliated with an established user group.

The Advanced Photon Source, or APS, is a Department of Energy Office of Basic Energy Sciences research facility that produces the most brilliant X-ray beams in the Western Hemisphere. Researchers from around the world use the APS for research in materials science, chemistry, biology, physics, earth and planetary science, and environmental science. The 1,104-meter circumference APS accelerator complex, large enough to encircle a baseball stadium, houses a complex of machines and devices that produce, accelerate and store a beam of subatomic electrons that is the source of APS X-ray beams.

The APS is a highly successful partnership between government, academia and industry. The Department of Energy provides the operating budget. The University of Chicago operates APS and Argonne for DOE. Academic and industrial partners build the beamlines, using funds from federal and state governments as well as industry and private universities. Users who perform the experiments represent universities, research labs and companies from virtually every U.S. state and several foreign countries.

Real-time results
Among the advances in instrumentation is a new computer cluster, which offers the efficiency and flexibility of real-time results in a field where researchers generally wait hours, days or weeks for analyzed data.

“The goal is to give users immediate feedback,” said Francesco De Carlo of Argonne’s Experimental Facilities Division, one of the system’s developers. “Now, by the time the user has prepared the next sample, he or she has an image.”

The new computer system was developed by De Carlo and Brian Tieman as an integral part of beamline instrumentation for a process called “X-ray computed tomography.” The non-invasive process digitally reconstructs three-dimensional images or cross-sectional slices of chemical and biological samples using compiled information from a series of X-ray projections.

At Argonne, X-ray computed tomography has been used to investigate the quality of protective coatings for airplane wing turbine blades, the unique physical properties of a sea urchin spine and the presence of tiny cracks in artificial inserts medical doctors use to repair bone. Such projects require a great deal of computation. De Carlo and Tieman introduced the new computer cluster, which is able to process the massive amounts of data in minutes.

The new cluster’s 16 dual processors offer a total of 33 gigahertz of processing power and 2,400 gigabytes of disk space. Previously, raw data from the experiments had to be saved to disks and processed and analyzed later.

Now, users can look at the experiment results after just a few minutes, perhaps while the next sample to be studied is being prepared. If necessary, a user might even modify an experiment already under way.

“If we can see what’s going on, instead of just taking data, we can watch and make more intelligent decisions about how to run the next sample,” said Stuart Stock, an APS beamline user from Northwestern University. Stock, who has been doing tomography for more than 15 years, said he expects the computer system to change the way he plans his experiments.

To take full advantage of these expanded computing capabilities, researchers may design experiments that require fewer samples to obtain the same amount of usable data.

(continue to page 2)

Back to top