Version française
Microelectronics & semiconductor manufacturing
The microelectronics and semiconductor manufacturing industries have to deal with several permanent challenges which arise from the need to master the complexity inherent in the manufacturing processes and to adapt to technological evolution.
- Complex fabrication processes :
- Hundreds of process steps are necessary to produce an integrated circuit, each one involving different tools, which must be controlled, and are likely to generate particular defects in products, that must be detected.
- In order to control product quality, high-precision measurements which call upon advanced technologies must be carried out. Because of the cost and complexity of such measurements, their numbers and durations must be minimized.
- While the data collected all along the manufacturing process become more and more numerous and easy to access, their volume makes their exploitation more difficult and new analysis tools are required to extract the relevant information.
- Hundreds of process steps are necessary to produce an integrated circuit, each one involving different tools, which must be controlled, and are likely to generate particular defects in products, that must be detected.
- Permanent technological evolution :
- Technological generations follow one another, contributing, in particular, to maintain Moore’s law in force by the regular reduction of circuits' critical dimensions. As a direct consequence of this evolution, the statistical tools used to analyze process data must be sufficiently generic to adapt to these changes.
- In a competitive market, the learning phase of a new technology is particularly critical. The first functional prototype must be quickly produced (Time-to-Market) and a mass production (Time-to-Volume) characterized by a good yield must then be reached as soon as possible.
- For a given technology, whereas the number of products increases, their lifespan tends to decrease, requiring more frequent adjustments and changes in processes, always requiring more flexibility on behalf of the analysis tools.
- Technological generations follow one another, contributing, in particular, to maintain Moore’s law in force by the regular reduction of circuits' critical dimensions. As a direct consequence of this evolution, the statistical tools used to analyze process data must be sufficiently generic to adapt to these changes.
- Continuous changes in the industry's economic conditions:
- Increased competitiveness and the need to move manufacturing activities from high cost to lower cost areas imposes changes to the way some of the critical data are collected and used.
- The general movement of Electrical Wafer Sort away from WaferFabs has created more attention and interest of the manufacturing community to the integrated Electrical Test/Parametric Test results.
- Increased competitiveness and the need to move manufacturing activities from high cost to lower cost areas imposes changes to the way some of the critical data are collected and used.
Today, the statistical analysis tools used for process control and quality monitoring are often limited to SPC-like, variable per variable analyses. In addition to the fact that this kind of analysis does not take into account the correlations between variables, it requires the set-up and maintenance of control limits.
The statistical learning tools developed by BlueKaizen are complementary to the existing tools. They make it possible to find solutions better adapted to the characteristics of some of the problems encountered thanks to multivariate data modelling (simultaneous treatment of all the relevant variables) and an adaptive approach (models' parameters are adjusted by training from relevant data).
The computing time implied by the use of a model makes most BlueKaizen solutions compatible with real-time, in-line exploitation.
Our offer
BlueKaizen supplies software components with high algorithmic added-value, designed to be integrated in existing software of the micro-electronic and the semiconductor manufacturing industries, such as MES (Manufacturing Executive System), EDA applications (Engineering Dated Analysis), Equipment Integration, SPC or APC platforms (Advanced Process Control).
These components are also exploited in BlueKaizen software packages and can be integrated directly in the information systems of final customers.
MASA BlueKaizen Semiconductor offering is split in 3 product families:
- General Statistical Tools
- BlueKaizen QuEst
Fully Automated Quantile Estimation tool, based on non-parametric density estimation.
- BlueKaizen WaferQuality
Associate an abnormality score to a wafer from a set of – critical or not critical – parametric tests results, for a more accurate detection and diagnosis of parametric problems, a reduction of hold lot processing time and an anticipation of the probe results.
- BlueKaizen WaferClue
Using Parametric Test data, BK WaferClue generates groups of lots that share common signatures (same sets of deviating parameters)making use of non parametric density modeling.
- BlueKaizen TesterWatch
Using Parametric Test data, BK TesterWatch detects drifts and abnormal behaviors on tester cells (Tester/Prober/ProbeCard).
- BlueKaizen LotStatus
BK LotStatus is a container for customers' acceptance rules.
- BlueKaizen WaferFit
Computes reliable estimates of a metrology at un-sampled points using neighboring measurements, for a more accurate process control, reduced metrology costs and more powerful root cause search.
- BlueKaizen WaferProfile
Compute advanced Statistical Process Control (SPC) indicators, which exploit the spatial information carried by measurements co-ordinates or combine the information of several correlated measurements.
BlueKaizen ParametricTest Suite
BlueKaizen ProcessControl Suite