Wir verwenden Cookies

Wir nutzen Cookies auf unserer Website. Einige von ihnen sind essenziell, während andere uns helfen, diese Website und Ihre Erfahrung zu verbessern.



Statistics for Chip Production

The production of semiconductors is complex and subject to certain fluctuations. Simulations absorb these instabilities.

The production of semiconductors is subject to minimal, but notable fluctuations. Nonetheless, all framework conditions specified by the design must be observed in detail in semiconductor production. Time and cost-intensive simulations are performed in order to keep newly developed chips robust – i.e. functional – on account of these production fluctuations. JOANNEUM RESEARCH is developing a statistical model within the context of the EU project “eRamp” with which up to a fifth of these elaborate simulations can be omitted. 

Depending on requirements, technical parameters are described so they can be fulfilled in production, which means that chip design must be envisaged so that the chip remains functional even when minor fluctuations occur in production, resulting in elaborate calculations. The greatest fluctuation width in production is described in so-called corner points, which result from more than a hundred thousand measurements. A team from the POLICIES research group “Statistical applications” has now developed a statistical model that can guarantee a consistent quality in chip production with few corner pints, which reduces cost and time-intensive simulations by approximately 20 percent. “In our calculations, deviations from the actual design must be concurrently considered for many parameters. In the so-called corner model, we describe how data depend on each other as we calculate the corner points. The number of these points is essential in the semiconductor industry since a simulation is performed for each corner point in order to check the functionality of the chip for the maximum fluctuation width of a certain number of – e.g. geometric – parameters,” explains statistician Thomas Riebenbauer. The research group “Statistical applications” specialises in highly complex data analyses.