Parallelization and fast processors on Eagle quicken a key calculation from 38.9 days (projected on a local workstation) to just 5 hours 45 minutes. details (pdf)

Cerium oxide (ceria) is a critical and multi-functional component of automotive emissions catalysts. The presence of hydrogen in ceria has been probed by infrared apsorbtion spectroscopy (IR), and characteristic bands have been identified, but a theoretical prediction of the hydrogen vibrational frequencies is needed to positively assign the bands. Unfortunately, the same property that makes ceria so useful in automotive catalysis -- the fact that cerium has two stable oxidation states -- also renders it a very challenging system to treat computationally. With Eagle, the computational challenge of computing vibrational frequencies for hydrogen in ceria with accurate first-principles methods has been met.

The parallel implementation of the vibrational analysis scheme was first tested on a simpler problem, hydrogen vibrations in a hydrogen-doped alumina, which had been carried out successfully on a local workstation. The single-processor workstation took 44 hours of CPU time to complete the calculation, Eagle turned the problem around in 1 hour 50 minutes by the wall clock! The results were in excellent agreement with experiments, thus validating not only the methodology, but also the parallel implementation.