The concept of statistical process control (SPC) is not new to the semiconductor industry. The semiconductor industry has been using SPC for a long time to successfully reduce scrap and improve quality. Using SPC during the testing phase, however, was always a puzzle.
The first step in producing an integrated circuit (IC) begins with making wafers that have several dice. The dice are then electrically tested, assembled, and packaged. Finally, the packaged dice are electrically retested and the integrity of the leads is verified. Since test is the last operation in the chip-making process, it is very important to control this process in order to avoid rejects. Any loss at this stage due to physical damage to the parts can result in loss of money, increased cycle time, or reduced capacity. Today, when the majority of IC manufacturers are restricted due to wafer capacity fabs, they cannot afford to lose the IC at final test.
Cultural change at Motorola
Prior to implementing SPC the final test phase at Motorola, an operator's only responsibility was to test the product electrically and mechanically. There was a quality control person coming behind them to verify the product and point out their mistakes. Thus, the operator did not have total ownership of the process. In order to change the culture and give the operator ownership of the process, the company had to have process control in place to increase the operator's confidence. Also, for this process to be successful and have total buy-in from everyone in the manufacturing unit, a team concept was used.
A cross-functional team was put together which included operators, process technicians, manufacturing technicians, process engineers, group leaders, trainer, and engineering section manager. The goal was set by the team to have the processes under control within 12 months of the team's first meeting date. Although the goal was aggressive, the team accepted the challenge.
The job of an operator in final test is repetitive in nature. Introduction of SPC gave a new meaning to the job. Introduction of SPC at Motorola, however, was not easy, because the majority of veteran workers resisted the change. They viewed it as extra work, more training, and above all, they were afraid of making mistakes and being disciplined. The veteran workers' pride was at stake, because a new process would put them at the same level as new workers. The worker's misgivings were communicated to the team through a suggestion box located in the manufacturing area.
In order to make SPC successful at Motorola, the team decided to pilot it on the midnight shift because it had the most senior workers. Midnight shift personnel were told that they were selected because of their experience. Once they debugged the new methodology, the program would be rolled out on the other shifts. This gave the veteran workers confidence since they would keep their status as more experienced than newer employees. The team knew that the easiest way to implement the cultural change would be by beginning with the senior employees.
Training and implementation
An engineering statistician experienced in process engineering was chosen to train the operators. A ten-hour training class was designed to familiarize midnight shift operators with basic elements of statistics, process control, and SPC. Real life examples were used as the teaching materials so that the students could relate them to the process control techniques that they would be using at work. They became familiar with attributes, variable data, and the SPC charts available to control attributes and variable data. During the class, a manufacturing process was discussed in detail. The class brainstormed for the necessary data or other elements of SPC. SPC charts were designed for each procedure with input from the class. The control limits were derived statistically through the data collected from previous operations. …