Parallel Computation of Electron-Molecule Collisions

• Authors: Carl Winstead;Howard Pritchard;Vincent McKoy
• Venue: IEEE Computational Science & Engineering
• Year: 1995

In chemical physics, the possibilities of parallel computation have just begun to be tapped. This article describes an application where the promise has already been realized. Furthermore, it contains a caveat emptor list of lurking pitfalls for the unwary newcomer trying to parallelize a code containing many different computational pieces. A central element in the parallelization strategy that eventually proved to be a winner for these authors followed from the realization that critical integrals can be formed with a minimum of interprocessor communication by forming a dense coefficient matrix. Readers contemplating the leap into parallel computing should find the analysis leading to this strategy illuminating, and those who have already taken the plunge may find it useful to compare strategies.The application providing the end goal for this parallelization effort is plasma processing in semiconductor fabrication, the understanding of which requires a large database of electron-scattering cross sections. It is difficult, if not impossible, to measure such cross sections for the short-lived and highly reactive intermediates involved in chemical vapor deposition and plasma etching, and analytical theories cannot treat systems this complex. Hence computational science fills a critical need. Because of the light mass of the electrons, the problem must be treated in a fully quantum-mechanical way, and parallelization makes the quantum-mechanical scattering calculations practical.The authors can be reached in care of Carl Winstead, A.A. Noyes Laboratory of Chemical Physics, California Institute of Technology, Pasadena, CA 91125; e-mail, [e-mail].