Example And Discussion Of Implied Cpk in IC Assembly Line Defect Projections

# Example And Discussion Of Implied Cpk in IC Assembly Line Defect Projections

Publish Date:2017-08-19 17:06:30 Clicks: 205

Figure 4.4 is an example of a portion of an IC fabrication line. Only a few operations are shown in order to demonstrate the utility of using Cpk-based analysis for the line. This analysis can be used to determine defect projections for all different IC types that are made by the line, based on the number of manufacturing steps required by the IC for each operation. Note that by using the Cpk approach, the 1.5 a shift of the average to the specification nominal is not considered in the defect calculations. For each operation, several attributes are shown by rows in Figure 4.4 to classify their quality:

• The process specification. Each operation is characterized by one- or two-sided implied specifications that cause defects to occur when they interact with the variability of the process. This information is required to make the decision when back-calculating the Cpk from the defect data. The specifications are assumed to be either single (one-sided) or double (two-sided) .The Cpk for each operation. This Cpk is calculated from previous historical data when the process capability of each operation was determined. They are recorded as the current quality level of that operation. Note that in the last operation, solder reflow has achieved six sigma quality of Cpk = 2.

The next two attributes convert this Cpk number to the more familiar DPU number for defect measurement in PPM. The DPU number could alternately be used to record the quality instead of the Cpk number.

2 is the variable from the standard normal distribution, derived from Cpk by Equation 2.13 (z = 3 • Cpk). The next line is the/*(-2) to determine the one-sided probability of defects that can be found directly.

DPU (PPM) is the defect rate of the operation. It is derived from the f{-~z) and then multiplied by 1,000,000. If the implied specifications of the operation in this section are two-sided, then the defect rate is multiplied by two. The DPU can be used as a substitute for defining the quality of the operation, instead of the Cpk if so desired.

AT is the number of operations required for the IC being assessed for quality. In this case, the IC has to undergo 183 epoxy dispense operations. NDPU, or total defects for producing the IC in this operation, is calculated by multiplying N by the DPU to produce NPDU for that IC.

The operation FTY is calculated by subtracting the NDPU from 1 for each operation.

When all of the data for each operation have been determined, then the total line information can be calculated. Depending on the goals set for the IC manufacturing line, three indicators can be determined for each IC type that is produced on that line:

1. Total line NDPU—the total manufacturing defects for the line resulting from making a particular IC. This is calculated by adding the defects (NDPU) from each operation.

2. Total line FTY—the total yield for a particular IC made in the line. It can be calculated either by multiplying the yield of each operation or from subtracting the total NDPU from 1.

3. Total line Cpk—the quality index for the IC being made in the line. This is back-calculated from the defect rate, assuming two- sided specifications and no process average shift.c

The information gathered from this example can be used by different parts of the organization, helping them achieve their individual goals. Management can use this information to document the production lines progress toward six sigma. Test engineers can use this information to plan for test and troubleshooting stations. Production and process engineers can use this information to focus on which manufacturing operations most need quality improvements. In this example, the ribbon bonding operation has the lowest Cpk and highest DPU, and therefore should be the first operation to be targeted for quality improvements.

label: CPK