Addressing the Challenges in Multiscale Modeling

A challenge for such systems is how to construct an accurate chemical mechanism without in-process concentration measurements being available. 

Most processes in which control of events at the molecular scale is important have only one or two manipulated variables available during processing. For example, the applied potential is the only manipulated variable during the electrodeposition of copper to form an interconnect. A challenge for constructing predictive models for such systems is how to excite the dynamics during model identification experiments when so few manipulated variables are available. 

This section summarizes our views on how to address the challenges posed in the previous section. Substantial unknowns in a complex heterogeneous mechanism can 

The questions ofhowto construct a chemical mechanism withouthaving in-process concentration measurements and how to excite the dynamics during model identification while having few manipulated variables can be addressed by a combination of three methods  

One question that arises with such an approach is how well the model parameters associated with surface diffusion and the chemical and electrochemical reactions can be extracted from the current, potential and ex situ surface morphology data, given the complex nature of the interactions of the additives with the surface (e.g. see Table 4.3). A key point is that current and potential curves and the surface morphology are very sensitive to changes in the experimental inputs (shown in Table 4.2), indicating that 

---