+44 (0)24 7671 8970
More publications     •     Advertise with us     •     Contact us
 
Technical Insight

Design is the key to optimal performance (Cover Story - GaAs Manufacturing)

Building a fab from scratch is a very risky and complicated business. Oded Tal of MAX International Engineering Group looks at the best way to create a complete design and construction process to ensure a better-performing fab at a reasonable cost.
Choosing to design and build a new fab is one of the most important and complex decisions a company will have to face. A working fab is more than a building, a set of tools and systems and human operators: it s the sum of all these parts working together. In today s uncertain trading environment, designing and building a facility that will only be fully operational a year and a half later could be seen as a real gamble. Nevertheless, if the process of fab design is done properly from paper flow to finished layout it will yield an agile facility that is geared toward changes in market behavior. This article will discuss the best way to start the fab design process and create the right toolset from a paper flow, incorporating different market scenarios. The process also involves modeling the capacity requirements of the different scenarios to finalize the tools quantity and selection, designing the right physical layout to support the manufacturing process and understanding the pros and cons of different layouts with respect to management goals. Following this complete fab design process will help organizations to make better choices. Designing the process There are four main objectives of a good design process. First, it is necessary to forecast performance demands as accurately as possible, and understand the resources needed to meet them. Then, we need to generate the best possible fab design within a budget. As an overall target, we need to get from the initial decision to capacity on line as fast as possible. As part of this, it is necessary to build, install and qualify all tools on time and on budget. gives an overview of the design process. Designing any fab must start with a detailed process flow or flows. A good process flow is detailed down to the level of individual steps and describes each step from both process and manufacturing standpoints. The fab s physical layout should be based on the predominant flows, and from these a composite flow is superimposed. This flow should capture 80% of the products or processes. The process flows dictate the tool list and the size and relationships of the functional areas. Of course, since different sections of the process flow may have varying levels of confidence, your toolset confidence will mirror these variations. But even in a very uncertain flow, it is possible to identify problematic areas and look for the right tools to fix them. A thorough analysis of the steps allows the decision about tool choice to be delayed until the last moment when the flow is more mature. Cleanroom space is not the most expensive resource in a fab, and fab design is not simply a matter of layout and tool locations. To truly understand fab operations, all the available resources must be modeled. All the available tool types must be included and fully understood. The flow of materials and products must be analysed to create a flow model. Furthermore, the allocation of operators, technicians and support staff should be based on a detailed resource model. Finding a balance Obviously, management s role is to demand the best fab for the least money. But while budget constraints are easily determined, the best criteria for fab design are harder to define. Clearly, performance should be optimized, and people typically aim for either the best cycle time or the highest capacity. However, closer scrutiny shows that these are not mutually exclusive: both are important to optimize performance, and we suggest a composite of the two. (To help the actual design process, the criteria in should be studied.) The key performance indicators of any manufacturing environment are throughput, cycle time and inventory level. Throughput (TPT) is the number of units a toolset can process in a given time, typically measured in wafers per week. Cycle time (CT) is the time it takes to process a single wafer from start to finish, typically given in weeks or days. Inventory or work in process (WIP) is the average number of products in the line at any stage of production. For any toolset it is desirable to maximize TPT, minimize CT and make sure that the WIP level is as small as possible. Figure 3 shows the relationship between the three values. Drawing from Little s law (WIP = CT TPT) there is an optimal WIP level for each toolset that allows for the best CT and TPT composite (a WIP level of 4000 wafers for the example in ). The quality of the fab design can be measured by the ability to sustain manufacturing at that level. Our experience shows that understanding the capabilities of the toolset and building the fab and layout to support them by providing efficient flow and a controlled WIP will allow the manufacturer to achieve the best CT and TPT composite (see ). From concept to construction Once the design guidelines are set and the resources are understood, the next stage is to create a conceptual layout by locating the different functional areas. In this stage, it is important to understand the weights of the different functional areas in the fab, and to analyse the transactions and the number of wafers going between the areas. At this point, we know how big the functional areas are, as well as the movements between them, so it is possible to start designing the shell, based on a block diagram of the type shown in . By designing the building to accommodate the functional areas, it is possible to begin construction without knowing the final location of each tool. There is plenty of time to finalize tool location since some tools have lead times of longer than a year. But by starting the construction in parallel with the detailed design stage, it is possible to save two to six months. Once the high-level design has been processed and construction has begun, the next steps are the detailed layout and the architectural and engineering design. Each tool s location is pinpointed, together with the support infrastructure that it needs. After the detailed layout is finalized, the execution process begins when it is time to take control of the actual project management, i.e. construction and tool installation. Phasing the project The cost and complexity of building a new fab, coupled with the desire to have capacity on line as soon as possible, dictates populating and qualifying the tools in more than one phase. We suggest breaking the project into the following phases. Early toolset This stage is crucial for proving fab capabilities. With a new technology, a new product or even a proven part, the fab owner needs to be sure that the new facility can sustain quality manufacturing before committing all resources. To save time, only one tool of each different type is installed and qualified at this point. There are two typical milestones to execute the early toolset stage: tool qualification, which involves installation and qualification of each individual tool by both the vendor and the process engineers; and process qualification, where the process is certified by running short loop cycles for different blocks, and by running a full loop process and analysing yields. Redundancy toolset Once the process is running, the second tool of each tool type is installed. If the second source is not physically installed in this location, a performance test is in order. This phase should achieve no single point of failure across the line. Ramp steps Here the fab goes to full-blown capacity in steps, e.g. 100 wafers per week, then 200 wafers per week, etc. The available resources and complexity of the installation influence the ramp slope. Realistically, the average capability should not increase by more than 500 wafers per week, from month to month. Reducing the fudge factor To make sure that the gains achieved by aggressive design schedules are not wasted in the execution of the project, the building construction and tool install/ qualification phases must be managed carefully. We recommend the "critical chain" philosophy of project management. To implement the critical chain you need to identify all the activities, get their real duration and reduce all the "fudge factors" across the schedule. As can be seen in , this process is done in three stages. In stage 1, all activities and relationships are defined on a Gantt chart. In stage 2, the real duration of each activity is determined, and the excess time is compiled into a project buffer. In the third stage, the time saved is divided by three; a third goes to the activity owners, a third remains in the central project buffer, and a third is subtracted from the overall duration of the project. It is important to keep in mind that once the building is ready and all the tools are hooked up, the facility still might not qualify as a "best of breed" fab. A whole slew of business transactions still need to be defined and it is necessary to map out all these business processes ahead of time and create standard operating procedures (SOPs) to cover them. Typical processes include shift management, maintenance practices, tool and recipe SOPs, and WIP management. Once all the business processes are defined, the right systems must be put in place to support them. Many organizations do not follow a defined process when selecting a system, which should consist of identifying all the relevant systems needed, analysing them and then looking for vendors who can fulfill most of those needs. Although it is best to stick to off-the-shelf systems, in many cases these are not available. The most common information systems found in a fab are manufacturing execution systems, computerized maintenance management systems, enterprise resource planning, facilities control systems and visual management, i.e. cleanroom personnel and management updates. Future trends in GaAs fabs The GaAs industry is experiencing high demand for commercial products, particularly for wireless applications, and this is driving high-volume manufacturing. Market forces are dictating that processes need to become mature, and that product mixes need to be reduced. Yields and cycle time will be the focus of all the major players. Along with shorter times to market, these will force the modernization of fabs and raise operating efficiency towards that seen in the silicon industry. We expect the same trend to emerge in optoelectronics and lasers. Designing and building a new facility in the semiconductor industry cost a great deal of money. The challenge for every design is to find a balance between the desire to build a facility with the best possible performance and the need to minimize costs. Going through the right fab design and project management process reduces overall time to capacity, brings down the total cost of operations and will ultimately result in a better-performing fab for less initial outlay.
×
Search the news archive

To close this popup you can press escape or click the close icon.
×
Logo
×
Register - Step 1

You may choose to subscribe to the Compound Semiconductor Magazine, the Compound Semiconductor Newsletter, or both. You may also request additional information if required, before submitting your application.


Please subscribe me to:

 

You chose the industry type of "Other"

Please enter the industry that you work in:
Please enter the industry that you work in: