At SandBox we help process engineers at semiconductor chip and equipment manufacturers accelerate time to market and reduce the cost of recipe development by quickly and accurately predicting process outcomes using data analytics and advanced physical models. Every application in our software SandBox Studio is designed with process engineers in mind. Using SandBox Analytics, process engineers can accurately simulate, predict, and optimize nanomanufacturing processes. SandBox Oculus’s sharing and collaboration tools allows engineers to securely store, organize, and share all their data—in real-time. In SandBox Architect, engineers can easily create and modify 2D and 3D device stacks. Finally, in SandBox Envision, engineers can visualize their data and quickly recognize process trends and identify optimal process windows.

Our work at SandBox is important because the development of nanomanufacturing processes is oftentimes the critical bottleneck in the advent of new semiconductor technologies. With our technology, we can accelerate process development by a factor of three. This represents tremendous cost savings for semiconductor chip and equipment manufacturers, and its societal impact is immeasurable—by accelerating and pushing the limits of today’s nanomanufacturing processing tools, we enable the next generation of electronic applications to enter the marketplace.

White Paper: Model Based Process Optimization for High Aspect Ratio Trench Etch

We demonstrate the rapid identification of process windows for the plasma etch of high aspect ratio trenches using the software suite SandBox Studio™. The process examined involves a predeposition step to protect the mask and then repeated cycling of etch and deposition steps to the required depth. Here the aim was to identify the process parameters (e.g., pressure, powers, gas flow rates, etc.) to minimize the bow CD of the etch. Model-based inference was used to rapidly identify these process parameters to achieve the target etch. Bayesian experimental design was performed to identify the optimal experiments for calibration of the model of the etch and deposition processes. Synthetic experimental data were created by fixing the model parameters and adding noise. These were then used to calibrate the model. The calibrated model was used to predict the window of process parameters that achieve the targeted range of bow CD. Using SandBox Studio™, the optimal recipe was determined with less than 20 training experiments.

Meghali Chopra, Leandro Medina, Kara Kearney, Bryan Sundahl, Roger Bonnecaze “Model Based Process Optimization for High Aspect Ratio Trench Etch”.

For complimentary access to the full white paper, please request below.