Substrate Structure Metrology Using Two-Target Reference Imaging

The patent divides the measurement task into segments: a large-scale reference structure is divided into multiple local features, and the measurement is performed in stages. First, a low-resolution overview captures the entire field of view including the reference structure and target feature. Then, high-resolution measurements are performed on specific local regions (the reference structure and the target feature) separately. This segmentation allows the system to combine the large field of view capability with nanometer-scale measurement precision.

The patent introduces a reference structure as an intermediary element that bridges the gap between large-scale and small-scale measurements. The reference structure has features at multiple scales: a large overall dimension that can be captured in a low-resolution overview, and smaller local features that can be measured with high precision using SEM. By measuring the reference structure's local features with high precision and using the known relationship between these features and the large-scale dimension, the system can indirectly determine large dimensions with nanometer precision without requiring the entire field of view to be visible at high resolution.

The patent performs preliminary action by pre-defining a reference structure with known dimensional relationships before the actual measurement of the target feature. The reference structure is designed and fabricated with precise, pre-calibrated features that establish a measurement scale. During the measurement process, instead of performing complex image stitching and coordinate transformations across multiple images, the system simply compares the target feature against the pre-established reference structure. This preliminary preparation eliminates the time-consuming image stitching process while maintaining nanometer-scale accuracy.