The Shepherd: Root Cause Analysis

To better understand Operation 2700, the Shepherd looks at the operation-level cycle time by tool group, averaged across the entire time period displayed previously. Operation 2700 was performed on five different tool groups during the period, with approximately the same average process time for each tool group, but a range in the queue times observed. 

Further investigation reveals that the Nitride Dehydrate tools are dedicated, with two tools in each of the five sub-groups shown above. All five sub-groups can perform Operation 2700, but for different products. So, when a lot of a certain product type arrives, ready to be processed at Operation 2700, that lot must be processed on one of two tools in one of the five sub-groups, rather than on any one of the ten nitride dehydrate tools in the fab. This makes lot queue times highly sensitive to the tool states of the nitride dehydrate tools. As shown below, the nitride dehydrate tools as a whole exhibit considerable variability in availability. The chart below shows the average time in each SEMI E-10 state, across the 10 nitride dehydrate tools, in each shift. 

The tool dedication, combined with significant variability in scheduled downtime, unscheduled downtime, and engineering time on the ten tools, leads to variability in moves and operation-level cycle times, as shown on the next page.