Use of Optical Emission Spectroscopy to Predict Silicon Nitride Layer Properties

In this work, we use in-situ plasma characterization during SiNx passivation layer deposition in an industrial scale PECVD tool. Optical emission spectroscopy (OES) allows us to determine emission intensity of different species contained in plasma. We correlate these spectra peaks recorded by OES measurements with obtained layer properties: deposition rate, refractive index and bond densities. We select the appropriate OES features to be fitted to predict the layer characteristics using a large amount of OES data at different process conditions. We predict SiNx layer properties with maximum deviation of 10.3% for thickness, 1.7% for refractive index and 35.8% for bond densities. Based on these results, we propose OES as a powerful tool not only for process monitoring, but also for statistical modeling. Furthermore, using the same OES-based model, we realize process optimization for SiNx deposition by calculating a set of processing conditions giving layers with similar properties.