Replace the contents of this release, Flash can not be displayed to the display.
Home > Offerings > Product >iDEAS

IDEAS: Yield Rate Improvement Solution

Four steps to yield rate improvement
  1. Off-line management
    Locate root cause of low yield rate ASAP.

  2. Half off-line management
    Continuously accumulate manufacture-related knowledge

  3. Half in-line management
    Adjust process parameters for stabilizing yield rate.

  4. In-line management
    Enhance yield rate for mature processes.
Locating the root cause(s) of low yield rate


Traditional methods

Traditionally, when a low-yield-rate case occurred, it needs an experienced engineer to collect comprehensive data through different steps and to run a cross-examination on various statistical charts and wafer maps in order to locate the root cause(s). This is both time- and labor-consuming, and it usually takes one week to obtain the results. The lead-time may be longer when microscopic observation in the lab is required. This way, it is impossible to immediately feed back the corrective action to the production line.


Solution from Coretech: iDEAS

When a low-yield-rate case occurs, the loader of the iDEAS Engine from Coretech Knowledge will convert data into the byte indexing matrix with the data mining technology. Along with the correlation analysis, feature selection and data mining technologies, the iDEAS can quickly prepare the candidate list of potential root causes for the reference of engineers.

Functions of iDEAS
  • To locate the potential root causes and prepare the candidate list from the EDA/MES database for lot-based low-yield-rate causes.
  • To mine and analyze product manufacturing data.
    Basic information (BS)
    Manufacturing procedures (WIP)
    In-line metrologies (IM)
    CP/WAT Test
  • Objectives
    Help engineers to locate the root cause(s) of defect(s) detected in the process with a list sorted by function within the shortest time.
    Provide product engineers with the list of suspicious products.
    Enhance product quality.
Features of iDEAS

Enterprise Defect Detection System Solution

  • Data Mining Server
  • Provide EQP/CP/WAT/metrology/BS data processing
  • Provide data quality analysis mechanisms
  • Provide data preparation (Combined/Lot-based/Wafer-based data)
  • Provide intelligent decision-making point (decision-making feature/yield rate break point)
  • Provide EQP/symbol/value/statistics mining processing
  • Provide visualized charts
  • Provide cross-function/cross-project result analysis


Example: Defect Detection


  • Target data sets—300 sets, each containing 13,000 attributes (3,900,000 metadata): ONE minute.



  • Experiment: 88% of the 19 cases were ranked within the top five.
  • Online: 63% of the 23 cases were ranked within the top five.
  • Enhancement: Improve data preparation and create domain knowledge to enhance accuracy.
Technical Innovations