Semiconductor fabs are leveraging data, artificial intelligence (AI) and machine learning(ML) to improve quality, yield, and profits. The fabs generate terabytes of data each day from the manufacturing process that is used for predictive analysis to support the manufacturing process. Quality data on most of the materials used in the manufacturing process is not readily available to them. There is limited data available to them in the form of the Certificate of Analysis for the materials, which is usually restricted to a handful of parameters. The rest of the manufacturing data for the materials is held by the supplier and is not easily available. Athinia and others are creating secure platforms to facilitate the sharing of data between the suppliers and the fab. However, even if this effort is successful the only data that will be available will be the quality metrics on the as produced materials and not on the as used materials. The materials are delivered in containers and move through the supply chain over a period and the materials could be used many months after they are produced. It is generally accepted that there is interaction between the material and the container surfaces that has the potential to add contaminants to the material. Thus, it would be advantageous for the fabs to have data on the quality of the materials as they are being delivered to the fabrication process. We have developed an automation platform that can measure, in real time, critical quality parameters of liquid process chemicals that are used in the fabrication process. The platform can move samples from up to 40 locations of different chemicals within the fab to a central detector that can make quantitative measurements to obtain quality metrics at the point of use. The central detector can be up to 500m from the sampling locations. The central detector is an analyzer specific to measuring trace metals and nanoparticles or organic contaminants. The automation system can measure quality metrics of the liquid chemicals from the POU sampling points throughout the fab at predetermined time intervals or on demand from a specific sample point. The system can also sample from the ISO container delivering the product to ensure that out of control product is not connected to the distribution system. The results are available in SECS-GEM format and can be easily integrated with the fab wide monitoring system to improve performance. Timely availability of material quality metrics also enables the fab to avoid excursions and avoid costly downtime for the manufacturing line. We will present data for the measurement of trace metals, with sub ppt detection limits and nanoparticles from sampling points that are 500m away from the central detector. We will also present data on the measurement of organic contaminants like phthalates and phosphates in solvents from sample points that are 500m away. High molecular weight organics present in solvents have the potential to be left behind as a residue on wafer surfaces creating issues with further processing of the wafers. The ability to measure the high molecular weight organics at ppt levels will ensure that any contamination issues are caught before the solvents are used for wafer processing.
