Job configuration for semiconductor manufacturing

• US 5,341,302 A
• Filed: 04/23/1992
• Issued: 08/23/1994
• Est. Priority Date: 04/23/1992
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Connors, Daniel P., Yao, David D.
• Primary Examiner(s): Smith, Jerry
• Assistant Examiner(s): OAKES, BRIAN
• Application Number:
  US07/872,878
• Time in Patent Office: 852 Days
• Field of Search: 364/468, 364/402, 364/403, 364/578, 364/554
• US Class Current: 700/102
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling

Abstract

An apparatus and method for configuring semiconductor jobs to achieve a desired level of set serviceability, in either probability or expectation, using a minimal number of wafers. The configuration problem is formulated as a mathematical optimization, where the serviceability level and yield losses, at chip, wafer and job levels, are explicitly considered. The problem is then reformulated by replacing the mathematically intractable service level constraints with tractable lower bounds. The reformulated mathematical optimization is efficiently solved using marginal allocation, giving good, feasible solutions to the original configuration program.

Claims

1. A method of manufacturing a demanded quantity of chip sets at a desired serviceability level, each of the chip sets comprising a demanded quantity of chips of each of a plurality of chip types, the method comprising the steps of:

(a) constructing an initial optimization program having;

(i) decision variable values,

(ii) an objective of minimizing a quantity of wafers used, and

(iii) an initial constraint of producing the demanded quantity of chip sets at the desired serviceability level;

(b) replacing said initial constraint with a lower bound so as to generate a reformulated optimization program, said lower bound being a constraint that is equal to a product of chip type probabilities, each of said chip type probabilities being equal to a probability that a quantity of chips of one of said chip types will be successfully manufactured when jobs are released into a semiconductor manufacturing line according to said decision variable values, each of said chip type probabilities being not less than respective chip demands associated with respective chip types;

(c) determining said decision variable values by solving said reformulated optimization program; and

(d) utilizing said solution of said reformulated optimization program to control a semiconductor manufacturing line by releasing jobs into said semiconductor manufacturing line as indicated by said decision variable values.

2. The method of claim 1, further comprising the step of receiving an input parameter specifying whether a job configuration scenario for the manufacturing is single part jobs, single part wafers, or multiple part wafers.

3. The method of claim 1, further comprising the step of receiving an input parameter specifying whether the desired serviceability level is a probability or an expectation.

4. The method of claim 3, further comprising the steps of:

(a) computing an actual serviceability level given said input parameter; and

(b) reporting said actual serviceability level.

5. The method of claim 1, wherein the desired serviceability level is an expectation and step (b) of claim 1 comprises the step of replacing said initial constraint with a maximum Lai-Robbins lower bound given said decision variable values.

6. The method of claim 1, wherein a job configuration scenario for the manufacturing is single part jobs and the desired serviceability level is a probability, and wherein step (c) of claim 1 comprises the steps of:

(a) initializing said decision variable values to feasible starting quantities;

(b) determining a value of said lower bound given said decision variable values; and

(c) modifying said decision variable values, when said value of said lower bound is less than the desired serviceability level, by

(i) performing marginal allocation to select one of the chip types, and

(ii) incrementing one of said decision variable values to indicate an additional single part job of said selected chip type; and

(d) branching to step (b) of claim 6 when said value of said lower bound is less than the desired serviceability level.

7. The method of claim 1, wherein a job configuration scenario for the manufacturing is single part wafers and the desired serviceability level is an expectation, and wherein step (c) of claim 1 comprises the steps of:

(a) initializing said decision variable values to feasible starting quantities;

(b) determining a value of said lower bound given said decision variable values; and

(c) modifying said decision variable values, when said value of said lower bound is less than the desired serviceability level, by

(i) performing marginal allocation to select one of the chip types, and

(ii) incrementing one of said decision variable values to indicate an additional single part job of said selected chip type; and

(d) branching to step (b) of claim 7 when said value of said lower bound is less than the desired serviceability level.

8. The method of claim 1, wherein a job configuration scenario for the manufacturing is single part wafers and the desired serviceability level is a probability, and wherein step (c) of claim 1 comprises the steps of:

(a) initializing a quantity of jobs to a feasible starting quantity;

(b) constructing an auxiliary program for determining said decision variable values given said quantity of jobs;

(c) solving said auxiliary program given said quantity of jobs;

(d) determining a value of said lower bound given said decision variable values;

(e) incrementing said quantity of jobs when said value of said lower bound is less than the desired serviceability level; and

(f) branching to step (b) of claim 8 when said value of said lower bound is less than the desired serviceability level.

9. The method of claim 8, wherein step (c) of claim 8 comprises the steps of:

(a) initializing said decision variable values to feasible starting quantities;

(b) summing said decision variable values to compute an initial quantity of single part wafers in each of said jobs given said quantity of jobs; and

(c) modifying said decision variable values a number of times equal to the amount by which a required quantity of said single part wafers in each of said jobs exceeds said initial quantity of single part wafers, said modifying comprising the steps of;

(i) performing marginal allocation to select one of the chip types, and

(ii) incrementing one of said decision variable values to reflect an additional single part wafer of said selected chip type.

10. The method of claim 1, wherein a job configuration scenario for the manufacturing is single part wafers and the desired serviceability level is an expectation, and wherein step (c) of claim 1 comprises the steps of:

(a) initializing a quantity of said jobs to a feasible starting quantity;

(b) initializing said decision variable values to feasible starting quantities;

(c) summing said decision variable values to compute an initial quantity of said single part wafers in each of said jobs given said quantity of jobs;

(d) modifying said decision variable values a number of times equal to the amount by which a required quantity of single part wafers in each of said jobs exceeds said initial quantity of single part wafers, said modifying comprising the steps of;

(i) performing marginal allocation to select one of said chip types, and

(ii) incrementing one of said decision variable values to indicate an additional single part wafer of said selected chip type;

(e) computing a maximum quantity of said lower bound given of said decision variable values;

(f) incrementing said quantity of jobs when said maximum quantity of said lower bound is less than the demanded serviceability level; and

(g) branching to step (b) of claim 10 when said maximum quantity of said lower bound is less than the demanded serviceability level.

11. The method of claim 1, wherein a job configuration scenario for the manufacturing is multiple part wafers and the desired serviceability level is a probability, and wherein step (c) of claim 1 comprises the steps of:

(a) initializing a quantity of jobs to a feasible starting quantity;

(b) constructing an auxiliary program for determining said decision variable values given said quantity of jobs;

(c) solving said auxiliary program given said quantity of jobs;

(d) determining a value of said lower bound given said decision variable values;

(e) incrementing said quantity of jobs when said value of said lower bound is less than the desired serviceability level; and

(f) branching to step (b) of claim 11 when said value of said lower bound is less than the desired serviceability level.

12. The method of claim 11, wherein step (c) of claim 11 comprises the steps of:

(a) initializing said decision variable values to feasible starting quantities;

(b) summing said decision variable values to compute an initial quantity of said chip sites on said multiple part wafer;

(c) modifying said values of said decision variable a number of times equal to the amount by which a required quantity of said chip sites on each of said multiple part wafer exceeds said initial quantity of chip sites, said modifying comprising the steps of;

(i) performing marginal allocation to select one of said chip types, and

(ii) incrementing one of said decision variable values to indicate an additional chip of said selected chip type on said multiple part wafer.

13. The method of claim 1, wherein a job configuration scenario for the manufacturing is multiple part wafers and the desired serviceability level is an expectation, and wherein step (c) of claim 1 comprises the steps of:

(a) initializing a quantity of said jobs to a feasible starting quantity;

(b) initializing said decision variable values to feasible starting quantities;

(c) summing said decision variable values to compute an initial quantity of chip sites in each of said multiple part wafers given said quantity of jobs; and

(d) modifying said decision variable values a number of times equal to the amount by which a required quantity of chip sites on said multiple part wafers in each of said jobs exceeds said initial quantity of chip sites, said modifying comprising the steps of;

(i) performing marginal allocation to select one of said chip types, and

(ii) incrementing one of said decision variable values to reflect an additional chip of said selected chip type on each said multiple part wafer;

(e) computing a maximum quantity of said lower bound given of said decision variable values;

(f) incrementing said quantity of jobs when said maximum quantity of said lower bound is less than said demanded serviceability level; and

(g) branching to step (b) of claim 13 when said maximum quantity of said lower bound is less than said demanded serviceability level.

14. An apparatus for manufacturing a demanded quantity of chip sets at a desired serviceability level, each of the chip sets comprising a demanded quantity of chips of each of a plurality of chip types, the apparatus comprising:

(a) means for constructing an initial optimization program having;

(i) decision variable values,

(ii) an objective of minimizing a quantity of wafers used, and

(iii) an initial constraint of producing the demanded quantity of chip sets at the desired serviceability level;

(b) means for replacing said initial constraint with a lower bound so as to generate a reformulated optimization program, said lower bound being a constraint that is equal to a product of chip type probabilities, each of said chip type probabilities being equal to a probability that a quantity of chips of one of said chip types will be successfully manufactured when jobs are released into a semiconductor manufacturing line according to said decision variable values, each of said chip type probabilities being not less than respective chip demands associated with respective chip types;

(c) means for determining said decision variable values by solving said reformulated optimization program; and

(d) means for utilizing said solution of said reformulated optimization program to control a semiconductor manufacturing line by releasing jobs into said semiconductor manufacturing line as indicated by said decision variable values.

15. The apparatus of claim 14, further comprising means for receiving an input parameter specifying whether a job configuration scenario for the manufacturing is single part jobs, single part wafers, or multiple part wafers.

16. The apparatus of claim 14, further comprising means for receiving an input parameter specifying whether the desired serviceability level is a probability or an expectation.

17. The apparatus of claim 16, further comprising:

(a) means for computing an actual serviceability level given said input parameter; and

(b) means for reporting said actual serviceability level.

18. The apparatus of claim 14, wherein the desired serviceability level is an expectation and said replacing means of claim 14 comprises means for replacing said initial constraint with a maximum Lai-Robbins lower bound given said decision variable values.

19. The apparatus of claim 14, wherein a job configuration scenario for the manufacturing is single part jobs and the desired serviceability level is a probability, and wherein said determining means of claim 14 comprises:

(a) means for initializing said decision variable values to feasible starting quantities;

(b) means for determining a value of said lower bound given said decision variable values; and

(c) means for modifying said decision variable values, when said value of said lower bound is less than the desired serviceability level, said modifying means of claim 19 comprising;

(i) means for performing marginal allocation to select one of the chip types, and

(ii) means for incrementing one of said decision variable values to indicate an additional single part job of said selected chip type; and

(d) means for invoking said determining means of claim 19 when said value of said lower bound is less than the desired serviceability level.

20. The apparatus of claim 14, wherein a job configuration scenario for the manufacturing is single part wafers and the desired serviceability level is an expectation, and said determining means of claim 14 comprises:

(a) means for initializing said decision variable values to feasible starting quantities;

(b) means for determining a value of said lower bound given said decision variable values; and

(c) means for modifying said decision variable values, when said value of said lower bound is less than the desired serviceability level, said modifying means of claim 20 comprising;

(i) means for performing marginal allocation to select one of the chip types, and

(ii) means for incrementing one of said decision variable values to indicate an additional single part job of said selected chip type; and

(d) means for invoking said determining means of claim 20 when said value of said lower bound is less than the desired serviceability level.

21. The apparatus of claim 14, wherein a job configuration scenario for the manufacturing is single part wafers and the desired serviceability level is a probability, and wherein said determining means of claim 14 comprises:

(a) means for initializing a quantity of jobs to a feasible starting quantity;

(b) means for constructing an auxiliary program for determining said decision variable values given said quantity of jobs;

(c) means for solving said auxiliary program given said quantity of jobs;

(d) means for determining a value of said lower bound given said decision variable values;

(e) means for incrementing said quantity of jobs when said value of said lower bound is less than the desired serviceability level; and

(f) means for invoking said constructing means of claim 21 when said value of said lower bound is less than desired serviceability level.

22. The apparatus of claim 21, wherein said auxiliary program solving means of claim 21 comprises:

(a) means for initializing said decision variable values to feasible starting quantities;

(b) means for summing said decision variable values to compute an initial quantity of single part wafers in each of said jobs given said quantity of jobs; and

(c) means for modifying said decision variable values a number of times equal to the amount by which a required quantity of said single part wafers in each of said jobs exceeds said initial quantity of single part wafers, said modifying means of claim 22 comprising;

(i) means for performing marginal allocation to select one of the chip types; and

(ii) means for incrementing one of said decision variable values to reflect an additional single part wafer of said selected chip type.

23. The apparatus of claim 14, wherein a job configuration scenario for the manufacturing is single part wafers and the desired serviceability level is an expectation, and wherein said determining means for claim 14 comprises:

(a) means for initializing a quantity of said jobs to a feasible starting quantity;

(b) means for initializing said decision variable values to feasible starting quantities;

(c) means for summing said decision variable values to compute an initial quantity of said single part wafers in each of said jobs given said quantity of jobs;

(d) means for modifying said decision variable values a number of times equal to the amount by which a required quantity of single part wafers in each of said jobs exceeds said initial quantity of single part wafers, said modifying means of claim 23 comprising;

(i) means for performing marginal allocation to select one of said chip types; and

(ii) means for incrementing one of said decision variable values to indicate an additional single part wafer of said selected chip type;

(e) means for computing a maximum quantity of said lower bound given of said decision variable values;

(f) means for incrementing said quantity of jobs when said maximum quantity of said lower bound is less than the demanded serviceability level; and

(g) means for invoking said decision variable initializing means of claim branching to said decision variable initialization means of claim 23 when said maximum quantity of said lower bound is less than the demanded serviceability level.

24. The apparatus of claim 14, wherein a job configuration scenario for the manufacturing is multiple part wafers and the desired serviceability level is a probability, and wherein said determining means of claim 14 comprises:

(a) means for initializing a quantity of jobs to a feasible starting quantity;

(b) means for constructing an auxiliary program for determining said decision variable values given said quantity of jobs;

(c) means for solving said auxiliary program given said quantity of jobs;

(d) means for determining a value of said lower bound given said decision variable values;

(e) means for incrementing said quantity of jobs when said value of said lower bound is less than the desired serviceability level; and

(f) means for invoking said constructing means of claim 24 when said value of said lower bound is less than the desired serviceability level.

25. The apparatus of claim 24, wherein said solving means of claim 24 comprises:

(a) means for initializing said decision variable values to feasible starting quantities;

(b) means for summing said decision variable values to compute an initial quantity of said chip sites on said multiple part wafer;

(c) means for modifying said values of said decision variable a number of times equal to the amount by which a required quantity of said chip sites on each of said multiple part wafer exceeds said initial quantity of chip sites, said modifying means of claim 25 comprising;

(i) means for performing marginal allocation to select one of said chip types; and

(ii) means for incrementing one of said decision variable values to indicate an additional chip of said selected chip type on said multiple part wafer.

26. The apparatus of claim 14, wherein a job configuration scenario for the manufacturing is multiple part wafers and the desired serviceability level is an expectation, and wherein said determining means of claim 14 comprises:

(a) means for initializing a quantity of said jobs to a feasible starting quantity;

(b) means for initializing said decision variable values to feasible starting quantities;

(c) means for summing said decision variable values to compute an initial quantity of chip sites in each of said multiple part wafers given said quantity of jobs; and

(d) means for modifying said decision variable values a number of times equal to the amount by which a required quantity of chip sites on said multiple part wafers in each of said jobs exceeds said initial quantity of chip sites, said modifying means of claim 26 comprising;

(i) means for performing marginal allocation to select one of said chip types; and

(ii) means for incrementing one of said decision variable values to reflect an additional chip of said selected chip type on each said multiple part wafer;

(e) means for computing a maximum quantity of said lower bound given of said decision variable values;

(f) means for incrementing said quantity of jobs when said maximum quantity of said lower bound is less than said demanded serviceability level; and

(g) means for invoking said decision variable initializing means of claim 26 when said maximum quantity of said lower bound is less than said demanded serviceability level.

27. A method of manufacturing a demanded quantity of chip sets at a desired serviceability level, each of the chip sets comprising a demanded quantity of chips of each of a plurality of chip types, the method comprising the steps of:

(a) constructing an initial optimization program that operates in accordance input parameters, including wafer yield data and job yield data, said initial optimization program defining;

(i) decision variable values,

(ii) an objective of minimizing a quantity of wafers used,

(iii) a wafer loss model for modeling yield loss using said wafer yield data

(iv) a job loss model for modeling yield loss using said job yield data, and

(v) a procedure of using said wafer loss model and said job loss model to satisfy an initial constraint of producing the demanded quantity of chip sets at the desired serviceability level;

(b) replacing said initial constraint with a lower bound so as to generate a reformulated optimization program;

(c) determining said decision variable values by solving said reformulated optimization program; and

(d) using said solution of said reformulated optimization program to control a semiconductor manufacturing line by releasing jobs into said semiconductor manufacturing line as indicated by said decision variable values.

28. An apparatus for manufacturing a demanded quantity of chip sets at a desired serviceability level, each of the chip sets comprising a demanded quantity of chips of each of a plurality of chip types, the apparatus comprising:

(a) means for constructing an initial optimization program that operates in accordance with input parameters, including wafer yield data and job yield data, said initial optimization program defining;

(i) decision variable values,

(ii) an objective of minimizing a quantity of wafers used,

(iii) a wafer loss model for modeling yield loss using said wafer yield data

(iv) a job loss model for modeling yield loss using said job yield data, and

(v) (iv) a procedure of using said wafer loss model and said job loss model to satisfy an initial constraint of producing the demanded quantity of chip sets at the desired serviceability level;

(b) means for replacing said initial constraint with a lower bound so as to generate a reformulated optimization program;

(c) means for determining said decision variable values by solving said reformulated optimization program; and

(d) means for using said solution of said reformulated optimization program to control a semiconductor manufacturing line by releasing jobs into said semiconductor manufacturing line as indicated by said decision variable values.