Ge (001) surface was modeled as a molecular cluster. The several conditions of the cluster model were varied by the Grid application, generating the multiple choices of the model. These models were benchmarked on the Grid system. After that, the potential energy surface of acetylene movement over the model Ge (001) surface was scanned, in order to study the adsorption reaction. The each point of the potential energy surface was calculated on the distributed node of the Grid system. These results show that the concept of high throughput computing can be successfully adopted to computational chemistry with Grid. The Grid system for the presented work and its simple architecture were developed.