The objective of this paper is to present an algorithm for planning optimal time-varying trajectories in chemical vapor deposition (CVD) processes and to discuss the potential benefits of doing so. The traditional strategy for CVD is to control at fixed set points, which are established to provide the best overall operating conditions. While this approach has generally worked well in practice, it certainly represents a highly restricted view of process control. There is potentially great value in time-varying processing conditions. However, realizing the benefits requires the ability to understand and control the interactions among strongly nonlinear fluidmechanical and chemical phenomena.What are some of the potential benefits of transient processing? In thin film growth there is reason to believe that the optimal conditions for film initiation and grain nucleation are different from those that provide the best mature growth. 1,2 If a single process is to operate in both regimes, there must be a strategy to transition from one to the other. Another example where transient processing shows promise is in the filling of vias or trenches in semiconductor manufacture. Throughput could be improved if the process itself varied throughout the course of the fill. For example, a higher wafer temperature could be used early in the process when the features are relatively open and have low aspect ratios, with lower temperatures required to maintain good step coverage as the features fill and the aspect ratios increase. 3,4 Growing functionally graded materials, such as compound semiconductors, is another example where timevarying process conditions are required to produce through-thickness compositional variations in the film. Finally, rapid thermal processing for applications such as film deposition, oxide growth, and annealing rely on transient reactor conditions to achieve increasingly stringent processing goals. 5,6 Developing optimal processing strategies requires at least four essential elements. First is a quantitative measure of the relative benefits and costs associated with any particular processing trajectory, i.e., an objective or cost function. For CVD processes, the value of the film could be measured in terms of chemical composition, morphology and microstructure, and uniformity. The cost to achieve the value might be measured in terms of reagent consumption, energy costs, and throughput. A quantitative measure of both the value and the cost to achieve the value can be cast in terms of an objective function that is to be minimized by the optimal processing strategy. There may also be constraints or boundaries that a particular process or system must obey. For example, heaters or mass flow controllers have physical limitations on the rates at which they can respond. Or, the operational range of the process controllers can be restricted so as to keep processing conditions within certain limits. In some cases, film properties could also be viewed as constraints. For example, if a crystalline film is required, th...