In the proposed OCAP, as shown in Fig. 6, the main causes that contribute to Probe Test Low Yield are presented. This is done by analyzing the occurrence frequency of probe tester equipment repair history data accumulated over one year. The modified proposed OCAP in Fig. 6 suggests two OCAP items: Open or Short failures of 1% or more, and probe low yield lower than 85%. These two items exhibit the highest distribution in probe tester equipment repair occurrences.
After implementing all necessary facility measures and conducting verifications, a process buy-off is performed. As part of the process buy-off actions, the number of probe touch downs is reduced from the current process's 10 times to 3 times in the optimized process. Analyzing existing data reveals that if the first and second touch downs during the process buy-off have no issues, there is generally a 95% likelihood that all 10 touch downs will proceed without problems. Hence, the buy-off condition in the optimized process is determined to be 3 times.
Table 3 compares the AS-IS OCAP process flow versus the proposed OCAP process flow for low yield lots. The estimated equipment downtime for each low-yield item is summarized in the current OCAP process flow. It shows that there are around 2 to 3 hours of equipment downtime per low-yield item, resulting in an estimated total wafer probing tester downtime of approximately 16 hours. In the case of Fig. 6, the optimized OCAP process flow for low yield lots, the expected equipment downtime for each low-yield item differs from the typical recovery time of maintenance and process engineer equipment for each low-yield item. However, overall, it takes approximately 0.5 to 3 hours of equipment downtime per low-yield item, resulting in a total wafer probing tester downtime of around 5.5 hours. Compared to Fig. 5, the current OCAP process flow for low yield lots, Fig. 6, the optimized OCAP process flow, is expected to reduce equipment downtime by 10.5 hours by further simplifying the conditions for equipment shutdown.
Table 3
The AS-IS OCAP Process flow vs the proposed OCAP Process flow
Items
|
AS-IS OCAP Process Flow
|
|
Proposed OCAP Process flow
|
|
Summary for saving on machine downtime
|
|
Action Items
|
Estimated Tester Down
|
Action Items
|
Estimated Tester Down
|
|
Line pattern
|
Verify Line pattern
|
3hrs
|
No control
|
0
|
3hrs
|
Site issue
|
Verify Site issue
|
3hrs
|
No control
|
0
|
3hrs
|
Open or Short failures
|
Verify Open, Short failure
|
2hrs
|
Verify Open, Short > = 1%
|
2hrs
|
0hrs
|
Low Probe Low Yield
|
Verify Low Yield issue
|
3hrs
|
Verify Low Yield issue
|
3hrs
|
0hrs
|
Perform Buy-off (Probe Test condition check)
|
10times touch down
|
2.5hrs
|
3times touch down
|
0.5hrs
|
2hrs
|
Check for H/W Problem
|
Verify H/W Problem
|
2.5hours
|
No control
|
0
|
2.5hrs
|
Total
|
|
16hours(960mins)
|
|
5.5hours(330mins)
|
10.5hrs(630mins)
|
In order to demonstrate the effectiveness of the proposed OCAP, one production facility of the case company is selected to implement the proposed OCAP. The downtime of the wafer probing test facility, the overall equipment effectiveness (OEE), and the mass production wafer throughput quantity were evaluated. A total of 13 lots, equivalent to 325 wafer probe data, were considered. To compare the results with the ones using the AS-IS OCAP, a random selection of 12 lots, approximately 300 wafers, was made from the existing facilities currently undergoing probe tests. The downtime, OEE, and mass-production wafer throughput of this facility were compared to the lots operating under the proposed OCAP process flow conditions. Figure 7 shows the wafer probing yield of the selected facility using the AS-IS OCAP and the proposed OCAP. The horizontal axis in Fig. 7 represents the lot number used in the experiment, while the vertical axis represents the wafer probing yield. The left side of the graph, separated by the red reference line in-between the two OCAPs, organized the wafer probing yield under the AS-IS OCAP while the right side of the graph organized the ones using the proposed OCAP process flow conditions. It is found that the average test yield using the proposed OCAP is 1.1% higher than the one using the AS-IS OCAP.
Figure 8 illustrates the weekly machine downtimes of the selected facilities while using the AS-IS OCAP and the proposed OCAP. It is obvious that, from Fig. 8, the average machine downtime per week under the proposed OCAP process flow conditions was shortened by 1450 minutes, which is more than 24 hours, when compared to the one under the AS-IS OCAP. In other words, the machine downtime is improved significantly using the proposed OCAP process flow conditions. Note that machine downtime plays a crucial role in increasing output and improving OEE.
Furthermore, Fig. 9 displays a graph comparing the OEE and pack out of one of the existing facilities undergoing a probe test with the OEE and pack out of lots operating under optimal OCAP process flow conditions. The X-axis of the graph represents the work weeks code and time duration, while the Y-axis represents the wafer probe test pack out. The red reference line serves as the axis, with the left side representing the OEE and pack out of the facility undergoing probe tests under the existing OCAP process flow mass-production conditions, and the right side representing the OEE and pack out of lots with optimal OCAP process flow conditions.
By comparing the total number of wafers in more than 300 instances of OEE and pack out, it becomes evident that the wafer probe test pack out of lots operating under optimal OCAP process flow conditions are over 23 wafers per week higher than the wafer probe test pack out of lots operating under the existing OCAP process flow mass-production conditions. In other words, the substantial increase in probe test pack out quantities in lots with optimal OCAP process flow conditions significantly contributes to facility efficiency, leading to production growth and improved OEE.
According to the summarized results in Table 3, it is evident that the probe test yield, machine downtime, and throughput under optimal OCAP process flow conditions outperform the existing OCAP process flow mass-production conditions in all scenarios.
Table 3
The best combination of effectiveness summary
|
Average Wafer Probing Yield
|
Average machine Downtime
|
Average throughput
|
Using the AS-IS OCAP
|
95.58%
|
1980mins
|
114wafers
|
Using the proposed OCAP
|
96.67%
|
580mins
|
137wafers
|
As shown in Table 3, the most significant improvement is on the average of 1.1% increase in wafer probing yields. For the case company, the total number of chips per wafer is about 15,000, and the quantity corresponding to a 1.1% increase in probe yield is approximately 165 chips (varying per product). The unit cost of the product used for this effectiveness verification is USD 0.5. Therefore, the total value used for verification corresponds to 325 wafers amounts to USD 26.8K.
Additionally, when considering machine downtime and pack out, the facility downtime has decreased by 24 hours per week compared to the AS-IS OCAP process flow conditions, while the wafer production volume has increased by approximately 23.