Cross-production collaboration
Today’s advanced manufacturing equipment is capable of recording hundreds of different metrics, simultaneously generating vast amounts of data – which has tremendous value for OEMs (Original Equipment Manufacturers).
OEMs can leverage this abundance of information to provide new or better services to their clients, such as benchmarking against other users of the same equipment, advanced monitoring and calibration, or preventative maintenance support. OEMs can also leverage the real world operational performance and monitoring data to further improve their product design, knowing that real world operational conditions will always be different from the conditions in which equipment is designed and tested.
Yet, sharing data from semiconductor factories – also known as ‘fabs’ - is usually a no go. Detailed production data is considered commercially and IP sensitive, and fabs will therefore not allow any data to leave their premises. This makes deep collaboration between OEMs and their fab clients complicated.
Closing the loop
In an ideal scenario, fabs enable OEMs to run analysis on the data generated by their equipment, but without the ability to see the raw input data. This is possible with Roseman Labs.
To illustrate, the OEM would be able to compare and benchmark certain metrics across its clients (different fabs) without being able to see the specific values of individual machines or fabs. Additionally, the OEM would be able to monitor certain parameters, without seeing what that parameter is; the only trigger being if the parameter exceeds a certain threshold.
With these capabilities, OEMs and their clients can set up a highly controlled environment in which the OEM is able to extract insights from the data, without the need to see the commercially or IP sensitive details. This delivers value for both the OEM and the fabs, though higher uptime, better yield and ultimately better product development for the next generation of equipment - even more tailored to the real production conditions.
