June 9, 2022 Laura Matz
If we have learned one thing from the chip shortage, it is that more actionable data shared among partners and suppliers across the semiconductor industry is needed.
The chip shortage and supply chain issues show how fragile the breakdown in data sharing is. The European Chips Act and the CHIPS for America Act demonstrate how solving the chip shortage is a matter of outmost importance for many nations.
As an industry, we need to respond to this challenge with every company across the supply chain stepping up to help facilitate more sharing of data across the ecosystem.
How did we get here?
The volume of semiconductors has increased such that the semiconductor value chain is very much stretched. As the sheer number of devices required increases and node sizes shrink, the pressure to produce with zero defects and innovate faster is rapidly growing. To succeed, manufacturers and their suppliers need to do things differently.
The industry as a whole has seen greater material sensitivity as new technology nodes have been introduced during the past 10 years. This has created stress on the supply chain as new interactions between the materials and fab processes have been identified.
The sensitivity of materials is getting greater and greater. If you look at what happens within the materials value chain, starting with the material supplier, raw material suppliers need to procure each of the key raw materials. If there is one disruption where there’s a concern over a specific batch of material, this can cause a delay in the entire supply chain, requiring the entire process to start again, including going back to the raw material suppliers and trying to get new inventory.
This value chain runs normally in a streamlined manner, but with the increase in utilization and capacity, and even more emphasized by the chip shortage, every batch of material made becomes very important to deliver on time to the device manufacturer.
Being able to identify parameters for raw materials can improve the design process to form-fit the raw material to a specific process. From a device manufacturer, this can have a huge impact. It can enable the yield ramp to move much faster with higher yields overall, less variation, and higher throughput.
Data becomes a true differentiator
The semiconductor industry collects vast amounts of data on nearly everything, but this valuable data is rarely shared between companies. While one-to-one data analysis has provided many valuable solutions in the past, the opportunities become endless when companies from across the value chain start participating and benefitting from this data sharing.
High-volume, high-velocity data from multiple sources has a huge potential to revolutionize the industry. For example, working together, we can take previously isolated datasets, aggregate them, and analyze them to reveal important and perhaps unforeseen insights and interdependencies among different process parameters or different materials. This could lead to unexpected outcomes and improvements that we would not have expected from those interdependencies.
An industry-wide ecosystem that aggregates data from across multiple companies to pinpoint the hidden parameters that impact quality is paramount so that the industry can focus time and effort on what matters most. This needs to be done by guaranteeing data security to ensure that customers maintain full ownership of their information and IP.
IP security is a key component and gives competitors and various companies confidence to come onto a shared collaborative platform. It also enables companies to pool learnings and advance materials quality so that we can minimize supply disruptions within the industry. To avoid IP contamination, data sharing needs to be facilitated by coding and normalizing all the data so each player can learn across the ecosystem and leverage the data and key trends without confidential information being shared.
Building semiconductor chips is a very complex manufacturing process, and access to a broader dataset and applying advanced analytics help us decode some of that complexity and find innovative solutions faster. By collectively analyzing more data, identifying patterns, and looking for relationships among discrete process steps or materials interactions, optimizations can be realized that lead to maximizing yield and reducing costs.
It is time for the industry to join us in creating a new standard in quality, one based on a data ecosystem that allows for the secure and continuous sharing of data between many companies.
Source: https://www.eetimes.eu/a-call-for-collaborative-data-sharing-across-the-semiconductor-industry/