Method and apparatus for statistical process control of machines and processes having non-constant mean of a response variable

• US 5,987,398 A
• Filed: 04/30/1998
• Issued: 11/16/1999
• Est. Priority Date: 04/30/1998
• Status: Expired due to Fees

• Current Assignee: Sony Corporation (Sony Group Corp.), Sony Electronics Inc. (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.), Sony Electronics Inc. (Sony Group Corp.)
• Inventors: Halverson, Galen Dean, Chou, Youn-Min
• Primary Examiner(s): Barlow, John
• Assistant Examiner(s): Bui, Bryan
• Application Number:
  US09/069,868
• Time in Patent Office: 565 Days
• Field of Search: 702/81, 702/84, 702/179, 702/182-183, 364/148.01, 364/148.06, 364/151-152, 364/164, 364/184, 706/23, 706/903, 706/906
• US Class Current: 702/179
• CPC Class Codes: G05B 19/41875 characterised by quality surveillance of production
• CPC Class Codes: G05B 2219/32201 Build statistical model of past normal proces, compare with actual process
• CPC Class Codes: Y02P 90/02 Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Abstract

A method and apparatus for analyzing a process or machine having a non-constant mean of a response variable using statistical process control techniques. Several measurements of a response variable are taken while the monitored machine is known to be operating in control. Through statistical regression techniques, a mathematical formula is derived to predict the value using indicator variables at various periods in the process. Then, while the process or machine is operating on an ongoing basis, actual measurements of the response variable are taken and compared to the predicted values using indicator variables calculated from the derived formula. Statistical process control rules are applied to the measured values of the response variable and the predicted value using indicator variables to determine if the machine or process has fallen out of control.

Claims

1. A method for analyzing a process or machine having a non-constant mean value of a response variable using statistical process control techniques comprising the steps of:

identifying an indicator variable which varies in correlation with said mean value of said response variable;

measuring said response variable while the process or machine is operating in control;

deriving a mathematical formula correlating said indicator variable to said mean value of said response variable for predicting said mean value based on a value of said indicator variable;

measuring actual values of said response variable during ongoing operation of the process or machine; and

applying statistical process control rules by comparing said measured actual values to said predicted mean value of said response variable based on a current value of said indicator variable.

2. The method of claim 1 wherein said deriving step comprises utilizing statistical regression analysis to correlate said indicator variable and said non-constant mean value of said response variable.

3. The method of claim 1, further comprising the step of adjusting the process or machine if said application of statistical process control rules indicates that the process or machine has fallen out of control.

4. The method of claim 2, further comprising the step of adjusting the process or machine if said application of statistical process control rules indicates that the process or machine has fallen out of control.

11. The method of claim 1, wherein said process is the deposition of aluminum by sputtering on a semiconductor wafer, said response variable is the thickness of aluminum deposited on a wafer and said indicator variable is the number of wafers on which aluminum has been deposited using a single aluminum source target.

5. An apparatus for analyzing a process or machine having a non-constant mean value for a response variable and an indicator variable which varies with said mean value of said response variable, the apparatus using statistical process control techniques and comprising:

circuitry for measuring said response variable while the process or machine is operating in control;

circuitry for correlating said indicator variable with said non-constant mean value of said response value for predicting an expected mean value of said response variable based on a current value of said indicator variable;

circuitry for measuring actual values of said response variable during ongoing operation of the process or machine; and

circuitry for applying statistical process control rules by comparing said measured actual response values to an expected mean value of said response variable based upon said indicator variable.

6. The apparatus of claim 5, further comprising circuitry for adjusting the process or machine if said application of statistical process control rules indicates that the process or machine has fallen out of control.

7. A method for analyzing output of a process or machine having a non-constant mean value of a response variable, wherein an indicator variable varies in correlation to said mean value of said response variable, the method comprising comparing actual measured values of said response variable to an expected means value of said response variable calculated based on said indicator variable using said correlation between said indicator variable and said non-constant mean value of said response variable, wherein said actual values of said response variable are measured during on going operation of the process or machine.

8. The method of claim 7, wherein said comparing further comprises determining if said actual measured values of said response variable are within one standard deviation of said expected mean value of said response variable calculated based on said indicator variable.

9. The method of claim 8, wherein, if 4 of 5 of said actual measured values of said response variable equal or exceed on standard deviation of said expected mean value of said response variable calculated based on said indicator variable, said method includes the step of adjusting said process or machine.

10. The method of claim 8, wherein said process is the deposition of aluminum by sputtering on a semiconductor wafer, said response variable is the thickness of aluminum deposited on a wafer and said indicator variable is the number of wafers on which aluminum has been deposited using a single aluminum source target.

12. An apparatus for analyzing a process or machine having a non-constant mean value for a response variable, said process or machine also having an indicator variable which varies in correlation with said mean value of said response variable, the apparatus comprising:

circuitry for measuring actual values of said response variable while the process or machine is operating;

circuitry for comparing said measured actual values of said response variable to an expected mean value of said response variable which is calculated based upon said correlation between said non-constant mean value of said response variable and said indicator variable.

13. The apparatus of claim 12, wherein said circuitry for comparing applies statistical process control rules to determine whether said apparatus is operating in control based on a coincidence of said measured actual values of said response variable and said expected mean value.

14. The apparatus of claim 13, wherein said circuitry for comparing determined whether said measured actual values of said response variable meet or exceed a standard deviation of said expected mean value.

15. The apparatus of claim 12, wherein said process is the deposition of aluminum by sputtering on a semiconductor wafer, said response variable is the thickness of aluminum deposited on a wafer and said indicator variable is the number of wafers on which aluminum has been deposited using a single aluminum source target.