26 June 2013 copyright 2013 G450C

Transcription

1 450 mm Equipment Demonstrations at G450C Statistics Used During Tests of the Semiconductor Industry s Latest Fab Equipment Transition Lorn Christal, G450C Program Manager Demonstration Test Execution 26 June 2013 copyright 2013 G450C

2 Global 450 mm Consortium A public/private program based at SUNY College of Nanoscale Science and Engineering in Albany, NY Driving effective industry 450 mm equipment development: Coordinate test wafer capability supporting development Demonstrate equipment/process performance Improve tools with suppliers to ready for customer operations G450C Members: CNSE / Research Foundation GLOBALFOUNDRIES Intel IBM Samsung TSMC 26 June 2013 copyright 2013 G450C Page 2

3 Wafer Size Transition to 450 mm One Key Factor in Moore s Law Every 10 years (on average), the semiconductor industry transitions to a larger wafer size. At the present time, the transition from 300 mm to 450 mm wafers is beginning June 2013 copyright 2013 G450C Page 3

4 Key Areas of G450C s Mission Requiring Statistical Methods Demonstrate equipment/process performance Improve tools with suppliers to ready for chip maker operations To realize these mission goals, G450C will use standard Demonstration Test Methods (DTM) 26 June 2013 copyright 2013 G450C Page 4

5 DTM Statistics Used Statistical Process Control (SPC) Analysis of Variance (ANOVA) Design of Experiments (DOE) Time/Failure Censored Reliability Tests Bayesian Inference 26 June 2013 copyright 2013 G450C Page 5

6 Demonstration Test Method DTM Origins Intel standard burn-in methodology IBM minienvironment test plan SEMATECH Equipment Qualification Plan I300I (300 mm transition) DTM modified for 450 Consensus member test plans USDOD Military Standards 26 June 2013 copyright 2013 G450C Page 6

7 DTM Topology 26 June 2013 copyright 2013 G450C Page 7

8 DTM Flow Chart Equipment Maturity Assessment Test Plan Test Plan Review Go Gauge Studies Baseline Characterization Decision points for further testing No-Go Mechanical Dry Cycle Passive Data Collection Baseline Review No-Go Demonstration Testing Go Sensitivity Analysis/DOE Marathon Test Final Demonstration Test Report Test Close-Out 26 June 2013 copyright 2013 G450C Page 8

9 Major DTM Components 26 June 2013 copyright 2013 G450C Page 9

10 Maturity Assessment/Test Planning Equipment Maturity Assessment Establishes the Test Regime Test Plan Set Detailed Scope and Goals Test Plan Review Approval? 26 June 2013 copyright 2013 G450C Page 10

11 Maturity Based Test Plans A viable process has four phases of maturity: 1. A measurable result is present. 2. Measurable results are repeatable. 3. A stable process has been demonstrated. 4. The process is optimized. Equipment Maturity Characteristics The four process characteristics are complimented with others to form 25 characteristics for maturity assessment. Doing so enables engineers to determine the appropriate demonstration test level 26 June 2013 copyright 2013 G450C Page 11

12 Determination of Test Level Equipment Maturity determines the amount of testing that should be applied Maturity Level 1 Maturity Level 2 Maturity Level 3 Production Test Level 1 Equipment suppliers validate basic functionality Test Level 2 Consortia Test Plan Gauge Study Test Level 3 Consortia Test Plan Gauge Study Test Level 4 IC Makers Mechanical Dry Cycle Mechanical Dry Cycle Process Characterization Process Characterization Sensitivity Analysis Review Reports Sensitivity Analysis Marathon Test Reports 26 June 2013 copyright 2013 G450C Page 12

13 Goal/Statistical Confidence Scaling G450C will optimize its metric goals for demonstration testing: Optimization (goal scaling) ensures: Appropriate test rigor Individual goal scaling Scaling based on maturity and current performance Initial goals for demonstration testing Modification at a later date Four (4) forms of scaling: Test levels e.g., Test Level 3 vs. Test Level 2 Performance e.g., uniformity goal of 3% vs. 2% Statistical Confidence e.g., 80% vs. 90% Bayesian Inference 26 June 2013 copyright 2013 G450C Page 13

14 Test Planning - continued Confidence Intervals for MTBF Metric Productive Time An example showing MTBF and its associated confidence bounds Data Analysis Fails Lower Confidence Limit (LCL) Point Estimate (PE) Upper Confidence Limit (UCL) MTBF p LCL 200 PE UCL 26 June 2013 copyright 2013 G450C Page 14

15 Bayesian Inference Pre-knowledge Prior data on an equipment s stability or manufacturing performance Can be applied when design commonality exists An engineer s judgment can be used, with qualifications if data supporting pre-knowledge is not available 3 types of pre-knowledge Extensive data e.g., performance data (test log) Some data e.g., engineer s logs Engineering judgment e.g., memory of past performance results 26 June 2013 copyright 2013 G450C Page 15

16 Baseline Characterization Baseline Characterization Gauge Studies Mechanical Dry Cycle Passive Data Collection Stability of the metrology system Stability of the wafer handling system Baseline Review Stability of the process 26 June 2013 copyright 2013 G450C Page 16

17 Baseline Characterization cont. Gauge Study Determines stability and variation present in the metrology system Mechanical Dry Cycle Stability test - mechanical wafer handling system Process gases and chemicals turned off Passive Data Collection (PDC) Stability test to determine process control Normal maintenance, no process adjustments 26 June 2013 copyright 2013 G450C Page 17

18 Demonstration Testing Sensitivity Analysis/DOE Marathon Test Optimizes the process Reliability Test Final Demonstration Test Report Reporting Test Close-Out Completion 26 June 2013 copyright 2013 G450C Page 18

19 Demonstration Testing cont. DOE Optimizes the equipment s process prior to Marathon Testing Desire - minimize process excursions during the Marathon test such that any failures encountered are hoped to be due to equipment design Marathon Test Time/Failure truncated reliability test 26 June 2013 copyright 2013 G450C Page 19

20 Demonstration Testing cont. Characterization Experiments Characterization Experiments explore the process response space to identify: Test strategy Response variables of interest Input factors and range of settings Number of replications to include Experiment restrictions Staffing assignments for specific tasks Analysis plan Decisions from Baseline Characterization Review Optimal equipment settings Three Dimensional Response Surface Two Dimensional Contour Plot The team acquires updates on: Time frame for the experiment (include analysis and documentation) Test cost Sampling plan Metrology system stability 26 June 2013 copyright 2013 G450C Page 20

21 Demonstration Testing cont. Marathon Tests Purpose: Marathon tests measure reliability and throughput Simulates a production environment Costs are significant Run 24 hour/7 days per week Wafers required are determined by information needed from the test Preparation is extensive Objective is to determine equipment performance 26 June 2013 copyright 2013 G450C Page 21

22 Demonstration Testing cont. Web Applications to support Marathon/MDCs Web-based applications for test planning, data collection and analysis: RAMCalc Reliability, Availability, and Maintainability (RAM) Calculator for Marathon Test data collection RELKit Reliability toolkit for MDC and Marathon test planning/analysis WaferCalc Marathon test planning calculator to calculate the amount of mechanical and process grade required for a Marathon test 26 June 2013 copyright 2013 G450C Page 22

23 Demonstration Testing cont. Reliability Toolkit Example 26 June 2013 copyright 2013 G450C Page 23

24 Demonstration Testing cont. RAMCalc Example 26 June 2013 copyright 2013 G450C Page 24

25 Statistical Challenges Learning the DTM Learning statistical software Application of Bayesian priors to stability tests Mentoring of process engineers Sea of data 26 June 2013 copyright 2013 G450C Page 25

26 Summary The 450 DTM is useful for characterizing new or existing tools, conversions, or line requalifications. The methodology applies to any type of manufacturing characterization at any phase of the development cycle for equipment or processes. The methodology addresses definitions and requirements, but cannot replace sound management techniques nor engineering experience. Again, the document is not a process or hardware development procedure. 26 June 2013 copyright 2013 G450C Page 26