Thin film technology is currently used extensively in many applications including microelectronics, optics, magnetic,corrosion resistant coatings, and micromechanics. The technology also plays an important role in semiconductor fabrication industries. The goal of the deposition process is to selectively and controllably deposit thin films on silicon wafers. This paper presents the design and development of a parallel computational tool for deposition that allows process engineers to optimize the deposition process as well as to predict the optimal sticking coefficients that are used to control the uniformity and surface growth, for arbitrary input conditions. In order to study the process, several inputs such as the type of gas, initial pressure. density, substrate temperature, reactor size and substrate size, can be parameterized. Our work focuses on the deposition processes in two-dimensional geometries. The simulation tool is developed based on the neutral flow Direct Simulation Monte Carlo(DSMC). As sited in previous literature,DSMC has a high computational cost when applied to a large substrate size. Therefore, a parallel and distributed programming technique is employed in order to keep the computational cost within an acceptable limit. The tool is designed to be executed in a gird environment where high computational power is readily available. Our parallel implementation utilizes the message passing technique as the communication paradigm. The problem is partitioned using the domain decomposition method. The simulation domain is divided into several cell-grids and each gird is assigned to a separate processor. Particle transport and collision are computed independently in each processor. The inter-processor communications only occur when particles move out of the cell-grid bound. When the steady state is reached, the output data from the simulation are collected, i.e., density, velocity, and temperature of gas molecules. These values are then used to calculate the sticking coefficient based on the Langmuir model. The model describes ideal mechanical absorption and is used to study the surface growth. Our simulation produces the result in the previously published theoretical ranges. Finally, the parallel performance and scalability are observed. From our experiment, the average processing time decreased as more computing nodes are added to the computation therefore it can be concluded that our simulation tool is fast, efficient, and accurate