a five‐stage approach for enhancing yield and assuring ZD of a manufacturing process is proposed. This five‐stage approach involves RD, ramp‐up, and MP phases. Observing the left portion of Figure 1.13, the RD and ramp‐up phases cover the first two stages; while the right portion of Figure 1.13 contains the last three stages for the MP phase.
The production line of the bumping process shown in Figure 1.14 consists of two sub‐processes, i.e. Re‐Distribution Layer (RDL) and Under Bump Metallurgy (UBM). In the following, UBM is selected as the illustrative manufacturing process, as depicted at the bottom of Figure 1.14. The UBM bumping process contains the following stations: Sputtering Deposition, Photoresist (including Positive Photoresist Coating, Edge Bead Remover, Exposure, and Developing), Cu Plating, Striping, Etching, Ball Mount, Reflow, and Flux Clean.
Figure 1.14 Production line of the bumping process.
The proposed actions at Stages 1 and 2 and associated challenges are described below.
Stage 1: developing a yield enhancement system (YES) to cope with the ‘p>>n’ problem to find and exclude the root cause affecting yield
In the illustrative bumping process, there are roughly 25 production stations, each station has 10 tools, each tool has 4 chambers, and each chamber has 100 sensors. Thus, there are totally about 100,000 parameters affecting the yield of the bumping process. If the information of tool maintenance and different material sources is also considered, the number of yield‐affecting parameters, i.e. p, is even higher. In the RD phase of the product life cycle, the number of samples, i.e. n, is relatively small, thereby leading to a challenge of finding the root causes of poor yield under the condition ‘p>>n.’ This is the so‐called high dimensional regression problem [61]. Thus,the developed YES should be able to promptly find the root causes affecting yield from the enormous number of parameters (p) under the constraint of small number of samples (n) and exclude them so as to effectively enhance the yield in the RD phase.
Stage 2: developing an equipment prognosis system (EPS) to find the aging features of tool failures and perform tool matching
While the YES at Stage 1 can be used to identify the problematic tool affecting yield, an equipment prognosis system (EPS) shall be developed to facilitate assuring the capability of the tool. Specifically, by creating the cause‐effect relationship of failure and prognosis model of equipment, the developed EPS should be able to observe the variation trend of key aging features and further estimate the remaining useful life (RUL) of equipment. Accordingly, the problematic tool can be maintained at proper time before it fails. Consequently, the possibility of tool’s abnormality can be reduced, and the yield in the RD phase can be enhanced. Moreover, after building a successful pilot production line, a tool matching scheme shall be applied for rolling out the pilot production line to multiple lines.
The right portion of Figure 1.13 shows the last three‐stage (Stages 3, 4, and 5) actions for assuring good yield in the MP phase. The proposed actions at Stages 3, 4, and 5 and associated challenges are described below.
Stage 3: conducting a fab‐wide deployment of AVM to achieve the goal of Total Inspection and to perform workpiece‐to‐workpiece (W2W) control with AVM
The AVM system is capable of converting off‐line sampling measurement into on‐line and real‐time Total Inspection of all workpieces to timely detect abnormalities during production. Also, the sampling rate of real measurements can be reduced by applying AVM. Accordingly, fab‐wide AVM applications can effectively reduce the production cost and achieve the goal of nearly ZD of all the deliverables in the MP Phase. In addition, due to the ability of achieving Total Inspection of all the workpieces, the outputs of AVM can be applied to support W2W control for fulfilling the goal of enhancing manufacturing process capability.
Stage 4: constructing the IPM system to perform tool health monitoring and execute tool RUL prediction
At this stage, the IPM system is constructed to detect abnormality on key components of all the manufacturing tools and to predict the RULs of all the key components so as to improve the tool availability and prevent unscheduled down of all the manufacturing tools.
Stage 5: developing the IYM system to promptly find the root causes of yield losses
In the MP phase, the IYM system is developed to promptly find out the root causes which affect the yield so as to reduce the trouble shooting time and improve the yield. As such, the goal of nearly ZD of all products can be achieved in the MP phase.
1.4 Conclusion
The evolution of automation is surveyed in this chapter, including e‐Manufacturing and Industry 4.0. Then, the importance of ZD, which is the vision of Industry 4.1 is presented. Finally, the five‐stage strategy of yield enhancement and ZD assurance is proposed. This five‐stage strategy is the guideline for developing the Intelligent Manufacturing System with ZD. As a result, an Intelligent Factory Automation (iFA) System Platform is designed and elaborated in Chapter 6 to realize the proposed five‐stage approach of yield enhancement and ZD assurance.
Appendix 1.A ‐ Abbreviation List
| APC | Advanced Process Control |
| ATP | Available‐to‐Promise |
| AVM | Automatic Virtual Metrology |
| BDA | Big Data Analytics |
| CASD | Capacity‐Allocated‐Support Demand |
| CIM | Computer‐Integrated Manufacturing |
| CMfg | Cloud‐based Manufacturing |
| CORBA | Common Object Request Broker Architecture |
| CPS | Cyber‐Physical Systems |
| EC | Engineering Chain |
| ECMS | Engineering‐Chain‐Management System |
| EECs | Equipment Engineering Capabilities |
| EE | Equipment Engineering |
| EES | Equipment Engineering System |
| EPS | Equipment Prognosis System |
| ERP | Enterprise Resource Planning |
| ESCM | Electronic Supply Chain Management |