Auto-metrology on TEM images of LED epitaxial layers

Yield in semiconductor manufacturing is directly affected by the performance of metrology equipment. The metrology tools in the nano-scale regime need to provide precise data. However the error creeps in data collection, analysis, interpretation and presentation due to the manual or semi-auto data processing systems adopted in the semiconductor industries. The accurate measurement with least uncertainty and quantification of the local change in the width of the epitaxial layers including multiple quantum wells (MQWs) and super lattices in a LED device is critical for its performance and optimization of the device. Manual measurement of nano-scale features on TEM images of LED epitaxial layers is not only a time consuming and tedious task but also erroneous owing to visual judgement. Here, an auto-metrology approach is presented to extract the measured values with higher precision and accuracy minimizing the uncertainty in the manual measurement. Firstly, TEM image of epitaxial layers is processed through an edge detecting algorithm to reveal the edge profile precisely. Finally, an algorithm calculates out the required geometrical data relevant to the useful parameters and presents it in a data table or graph based on the purpose of data interpretation. This auto-metrology approach is expected to be adopted by academia or industry for proper data analysis and interpretation with higher precision and efficiency.