CVD (chemical vapor deposition) is a process which grows a thin film on a substrate, and can be applied to producing many important engineering devices such as microelectronic circuits, optical and magnetic recording media and high performance cutting⧹grinding tools [1] . The thickness of the thin film ranges from a few nm to tens of μm according to the objective of a product. One of the most important measures of quality of this thin film is uniformity of the thickness and is crucial to the performance characteristics : device behavior, optical reaction, etc. To attain uniform growth rate, various attempts have been made and reported in the literature. Controlling the flow path of the reactant gas can be effective by either altering the chamber shape or setting up baffles inside the chamber [2] . Making the susceptor rotate [3–5] can also improve the growth rate and its uniformity as the centrifugal force enhances the inflow of the new reactant gas by throwing the used gas outward. Although the aforementioned studies can be useful in designing an efficient CVD chamber, they are mostly qualitative and very little effort has been devoted to the optimizing of the chamber shape or operating conditions. Optimization of the key parameters is a very complex problem as these variables are intricately related and thus cannot be controlled independently. As a first step toward this goal, we devised an optimization procedure to determine the inlet concentration distribution of the reactant gas, that results in the most uniform growth rate, and carried out an analysis to find an optimal concentration profile at the inlet for various flow rates. This is a relatively simple case since the concentration field is passive to the flow and temperature fields. The task, however, is by no means trivial as shall be seen later. The optimized profile may or may not be of the shape that can be easily produced. It is proper to mention here that the focus of the paper is how to obtain the optimal profile rather than discussing its economic feasibility. The procedure, if found successful, can serve as the basis for more complex and general optimization problems, such as the shape optimization, in the future.