Case Study - Yield
Wireless wafer-like vibration sensor for diffusion furnaces
When a manufacturer has yield problems sometimes finding the cause can be as challenging as the solution. Allyn Jackson, Field Application Engineer at CyberOptics Semiconductor discusses how a wireless wafer-like vibration sensor helped a diffusion furnace team identify vibration sources and subsequently increase their yield.
20 A
200mm fab diffusion furnace group with a total of 13 tools was fighting
low wafer yield for months. While some tools performed to standard, others were missing the mark, with four tools in particular experiencing low yields due to the following problems:
Random broken wafers Scratched wafers Abnormally high wafer defect rates downstream
Excessive particle counts
To resolve these issues, the diffusion team initially conducted various trouble shooting steps such as parts replacement, complicated and time-consuming partitioned particle checks, and various tool parameter adjustments. These randomly conducted, trial-and-error efforts, however, provided unverifiable or inconsistent results that contributed little to finding the primary cause of tool yield issues.
Next, the team sought to test a theory that
differing and/or indiscriminate vibrations and/or misaligned tools for wafer handling leveling were affecting performance with some tools. Traditional wafer vibration and level test methods such as attaching wired accelerometers to wafers or running dummy wafers through a tool and listening with stethoscopes for abnormal noises proved difficult, inconsistent and ultimately inaccurate as such crude methods
are not repeatable. The diffusion team wanted a more accurate, comprehensive and repeatable test method that could assess the whole tool while providing reliable data to quantify results with the ultimate goal of documenting and incorporating new maintenance procedures to prevent such problems in the future. A wireless, wafer-like vibration sensor
offered the capability to travel through tools to monitor three-axis accelerations and vibrations and transmit real-time vibration exposure data for analysis. By filtering out acceptable vibration frequencies produced by regular, slow-moving equipment or high-frequency noises between 1 to 200 Hz, the sensor could help identify vibration anomalies during wafer processing. As the sensor operates in unison with its own vibration monitoring software, data could be viewed and manipulated to identify problems and predict equipment failures to improve process yield and cycle times.
Diffusion tool characterization goals The fab chose to use the auto vibration sensor with an auto levelling sensor to characterize all 13 of its tools to determine if wafer handling and excessive vibrations were a contributing factor to these yield issues. A comprehensive test plan was formulated with the following primary goals: Establish a baseline levelling standard for all wafer transport or process positions in the tool. Of particular note were wafer
www.euroasiasemiconductor.com Issue IV 2011
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40