Advanced optical modeling of thin metals for improved robustness and accuracy of scatterometric models

The majority of scatterometric models used in production control assume constant optical properties of the materials included into the film stack. Only dimensional parameters are assumed as the degrees of freedom. This assumption negatively impacts model precision and accuracy (especially with the trend of scaling down the critical dimensions). In this work we focus on the modeling of Cu and TaN/Ta optical properties in back-end-of-line applications and consider the impact of Cu optical properties modifications in the trenches and as a substrate. We also consider the Cu transparency threshold when Cu acts as a substrate in the film stack. In the case of ultrathin Cu substrate the model output becomes invalid. Quite frequently this fact is not reflected in the goodness of fit. We show that accurate optical modeling of Cu is essential to achieve the required scatterometric model quality for automatic process control in microelectronic production. As a result, we obtain appreciably better matching with electrical data. Therefore, electrical performance can be predicted early in production flow. The modeling methodology presented here can be applied for all technology nodes and also other thin metals such as Co and Ru.