Version française
Industrial Process Control and Quality Monitoring
Process Control engineers have been using simple and robust statistical algorithms, such as control charts, for several decades. These algorithms have already proven their value in many traditional industries such as glass fabrication, steel or industrial chemistry.
However, these statistical tools show their limits when the system under observation become complex: high dimensionality of the inputs, non-linearity or unstability of the system's behavior. In more recent industries like microelectronics and semiconductors, characterized by the complexity of their ever-evolving fabrication processes, the traditional tools are often inadequate or insufficient to properly perform process control and quality monitoring.
BlueKaizen Machine Learning, BlueKaizen's statistical learning toolkit, offers a large range of modern statistical learning algorithms allowing us to develop a novel approach to process control and quality monitoring. In addition to the features offered by the standard statistical tools, the most innovative aspects provided by our tools are:
- Multivariate data processing: it allows to detect changes in the structure of the correlation between variables.
- Non parametric model construction: no a priori hypothesis about the form of the data distribution is needed.
- Unsupervised learning: models can be built without the need to associate labels to the elements of the learning set - a fastidious and sometimes unrealizable task.
- Automatic model selection: just a few parameters must be tuned by the user. Most free parameters are automatically fit by the learning process itself.
- Possibility of on-line use: the models are often very compact and fast to evaluate, which makes real-time applications possible.
Our offer
BlueKaizen solutions are dedicated to fault detection and classification for manufactured products, and to abnormal behavior detection (drifts, outliers) for fabrication equipments. For each of these two distinguished areas, BlueKaizen offers a generic solution:
- BlueKaizen Product Quality: for product quality monitoring and global yield improvement;
- BlueKaizen Tool Condition: for statistical process control and predictive maintenance of fabrication tools.
The following specific offers have been defined for the microelectronics and semiconductor industries :
- BlueKaizen WaferQuality provides both global and detailed monitoring of parametric tests (PT);
- BlueKaizen WaferFit reconstructs a surface, exploiting the spatial information provided by metrology data;
- BlueKaizen DeviceReliability optimises the trade-off between reliability and yield;
- BK Tool Pattern models an equipment behavior's degradation based on the equipment's monitoring parameters;
- BlueKaizen WaferProfile characterizes product defects and improves statistical process control (SPC).