Fab Layout Design Methodology: Case of the 300 mm Fab
This article covers issues to consider when designing a 300 mm lab.
Doron meyersdorf, Ph.D., and Abolfazl Tazhizadeh, TEFEN USA, Foster City, Calif. -- Semiconductor International, 7/1/1998
| At a Glance | |||
| |||
Perhaps one of the best ways to increase capital utilization is by increasing wafer size while decreasing die size (die shrink). However, introducing wafers with larger diameters translates into more expensive and sophisticated tools. With an average capital cost of $2 billion to build a new state-of-the-art fab, semiconductor manufacturers are realizing that traditional approaches to manufacturing and cost analysis will no longer satisfy future requirements.
One of the major areas neglected in the past is facility layout design. Fabs were traditionally designed around building and construction constraints, safety codes, cleanroom protocols and utility requirements, while little or no attention was paid to the macro and micro analysis of product flow, material handling, ergonomics, human factors and manufacturing goals. Therefore, many existing fabs cannot operate at optimal levels because their floor layouts hamper manufacturing and material handling efficiency. It has been estimated that effective facility layouts can reduce manufacturing operating expenses by at least 10% to 30%.1 Recently, with the introduction of 300 mm wafer sizes, manufacturers have fully realized the importance of these factors in the design and construction of new fabs. This is where industrial engineering and its systematic view of manufacturing environments become vital.
Layout concepts and challenges
| Lead Photo: Virtual reality modeling allows IC makers to "see" how productivity and efficiency can be increased using fab design methodologies employed by TEFEN. |
Semiconductor fabs present unique challenges for the cleanroom layout design team. Some issues that complicate fab layout and present challenges to layout optimization are the following:
- Relatively compact factories with proportionally long process flows
- Multiple process flows sharing equipment
- Capacity mismatching: sequential processes that have batch and process time variability Automation in various technologies and implementation levels
- Capacity boulders: difficulty of adding capacity in discrete units
- Expensive equipment: must seek benefits from adding the lower cost units where practical
- Re-entrant flows: product returns to the same process node multiple times
- Changing processes: process flows revision and update
- Cross contamination concerns
- Existing physical plant constraints
- Ergonomic and safety considerations
| Fig. 1. In a ballroom cleanroom, the production space is open and without bays and maintenance chases. |
Ballroom vs. bay-chase cleanrooms
In a ballroom cleanroom (Fig. 1), the production space is an open space without bays and maintenance chases. Chases might be placed at the perimeter of the room, in which case tools may be facilitated from the floor or from the ceiling. Thus, the fab will require additional floor levels (subfab, basement, penthouse, etc.). A growing number of photo modules utilize a ballroom configuration. Photo tools do not demand intensive maintenance and they require the same type of environment (i.e., lighting, temperature, humidity, etc.).
| Fig.2. The production floor is segmented in several bays in the bay-chase arrangement, with the bays linked by a common WIP corridor. |
| Fig. 3. In the farm (or functional) type tool configuration, the product tends to travel back and forth between bays, creating a "spaghetti-like" flow pattern throughout the fab. |
Pros and cons of ballrooms
Ballroom cleanrooms have the following advantages over bay-chase cleanrooms:
- Higher levels of communication and visibility
- Enhanced supervision and production control
- Ballrooms can have a higher tool density (can "pack" more tools in the available cleanroom space)
- Ballrooms are more adaptable to higher levels of material handling automation
- Minienvironments are more easily implemented in ballrooms
- Ballroom cleanrooms have the following disadvantages over bay-chase cleanrooms:
- High-maintenance tools will be problematic in a ballroom because of contamination issues and interference with other tools.
- It is more difficult to maintain laminar flows in ballrooms, and there are increased chances of turbulence. Raised floors are required and they generally require more than one fab floor level.
- Particle problems at a specific location of the ballroom can spread easily throughout the entire room.
- The layout design of a ballroom will have to take into account the height of all tools so that visual contact and communication is not hindered severely.
- In labor intensive modules, it is important to consider the amount of personnel in the room at any given time. Production areas with many people will pose safety and productivity problems.
- Noise levels are higher in ballrooms since there are no walls to isolate noise.
| |
| Fig. 4. In the cellular tool configuration, the product follows a more organized flow throughout the fab, compared to that shown in Figure 3. |
Farm vs. functional layout
In each of the above concepts, tools can be arranged in different configurations: functional, group technology or a combination of both. In a farm or functional layout, the tools are grouped according to their function. Each machine group (or pool) is shared among all product types that require that tool type. In a cellular or group technology layout, tools with different functions are grouped together to perform a set of operations on the product(s).
Figure 3 shows a farm (or functional) type tool configuration in a bay-chase fab layout. In this configuration, the product tends to travel back and forth between bays, creating a "spaghetti-like" flow pattern throughout the fab.
Figure 4 shows a cellular tool configuration with a typical flow patterns. In this type of configuration, the product follows a more organized flow throughout the fab.
The decision of specific tool configuration inside each bay is typically based on the number and nature of process flows (product types) and the product mix. When the number of process flows is larger than one or two, the flows may be significantly different, and there is no dominance of a single process flow (or a family of similar flows). In this case a farm layout would result in better performance. On the other hand, if there is a single process flow or a dominant family of similar process flows, a cell layout would be more appropriate.
It is common to incorporate both types of layout concepts in a design. The two main reasons are that high-volume products have their own cells, while other products share a farm-type layout, and expensive equipment might be grouped in a farm-type layout, while inexpensive tools can be located in several areas to reduce material handling.
| |
| Fig. 5. A facility layout design methodology achieved in four phases. |
Layout design methodology
This section presents a structured modular methodology that focuses on planning and designing the facility in several phases. During each phase, data are collected, analyzed and processed by analytical models and distinct tools. The output of one phase is required input for the following phase (Fig. 5). A description of the three phases of the design follows.
Phase 1 -- macro layout design
Macro layout design refers to the design at the facility at the module level. The analysis focuses on fab clusters, or functional areas (e.g., photo, etch, etc.) and the interactions between them. The final macro layout design outlines the optimal relative location of these functional areas according to the design's constraints and objectives. There are some major issues to be considered during this phase:
- WIP flow(s) between functional areas (interbay transactions)
- Relationship between different areas
- Automated material handling system (AMHS) (critical in the case of 300 mm wafer size)
- Capacity of each area at macro level
- Space requirement including stockers
- Safety and cleanliness considerations
- Facilities/building constraints
These issues are analyzed using a set of qualitative and quantitative tools, such as area transaction matrix, capacity analysis and discrete-event simulation (Fig. 6). After the analysis is completed, several layout alternatives are generated. Each alternative is then evaluated against a set of criteria defined by the layout design team, such as quality, cycle time, maintenance, safety, operational flexibility, WIP management, wafer start, mix flexibility, etc.1 The best alternative is chosen as the basis for the micro layout design.
| |
| Fig. 6. Phase 1 in designing a facility layout should include such considerations as capacity analysis and discrete-event simulation. |
When starting a layout design, it is important to first define the layout design criteria and their priorities. This set of criteria serves a dual purpose. First, it provides the design team with the necessary direction to generate a layout design that matches the manufacturing objectives of the organization. Second, the design and evaluation of floor layout alternatives should be classified under the area of "multiple criteria decision making" (MCDM) -- i.e., the general class of problems that involve multiple attributes, objectives and goals. This perspective provides a framework in which alternative layout designs can be evaluated objectively through the use of pro and con analysis and through the implementation of MCDM techniques.1
Phase 2 -- micro layout design
Micro layout design refers to facility design within each fab module. It applies to the design of a functional area (photo, for example), as opposed to the macro layout methodology, which applies to the block design of large functional areas and/or the rough design of the entities within each functional area.
The micro layout involves a more detailed analysis in which individual equipment sets are analyzed based on process flow and capacity. Metrology and engineering equipment usage is analyzed, as well as safety, ergonomics and maintenance clearances. WIP management, material handling automation requirements and operator walking distances are optimized at this phase. Other issues addressed during micro layout design are the following:
- Tool-specific data
- Facility constraints and design assumptions
- Layout design criteria and objectives
- Intrabay AMHS alternatives and their interfaces with interbay AMHS systems
- AMHS alternatives and tool interfaces (SMIF) (where applicable)
- Safety standards and cleanliness guides
- Effects of minienvironments
- WIP storage requirements
- Maintenance requirements
- Manpower requirements
- Effects of work procedures
- Workstation design
- Analysis of flow of materials in operational unit (e.g., bay)
- Micro capacity analysis
- Analysis of space requirements and mini stocker allocations
- Comparison of space requirements against available space
- Activity relationship analysis
The above factors are carefully analyzed during the course of a micro design project. Of significant importance is WIP flow analysis within each area and between tools (depending on the layout concept under development). Bottleneck tools should be studied carefully to ensure that layout design features will not cause any production interruption or create long cycle times. The AMHS and its material control system (MCS) should also be analyzed to ensure that it will not act as a bottleneck to the fab. The appropriate number of vehicles, traveling paths, location of stockers and I/O ports should be determined using analytical models.
Active cooperation of process engineers, facilities, manufacturing, maintenance and management with the design team is essential to the success of the project. Several alternative micro layouts are developed based on the results of the micro analysis, and each is evaluated vs. the design criteria. This set of criteria is essentially the same as the one developed during the macro layout design (Phase 1). However, it is more detailed and includes micro elements not considered in the macro phase (e.g., intrabay WIP flow, walking distances, cell and work center flexibility, support tools configuration, etc.)
| |
| Fig. 7. Phase 2 of a facility layout is the micro layout of fab modules and involves a more detailed analysis of process flow and capacity. |
Phase 3 -- detailed operational design
Detailed operational design gives an even finer analysis to the micro layout. It involves detailed storage analysis and determination of operational methods for each cluster, providing detailed workstation design including optimal work methods for each tool and wall elevation drawings. The analysis also looks at labor requirements, support tool requirements, SMIF and minienvironment, and detailed transportation methods. Figure 7 summarizes the detailed operational design project methodology.
300 mm wafer fabs
Technological advances in semiconductor fabrication and growing market demand have forced the device manufacturer to invest heavily in fab construction. Existing fabs will soon run out of capacity if global demand continues to grow at its current growth rate. On the other hand, demand for more sophisticated devices has been the drive for the development of better and more advanced tools to support 0.25 µm (250 nm) and below geometry while maintaining a high throughput.
Although the unit cost of a manufactured 300 mm wafer is projected to be 45% higher than a similarly manufactured 200 mm wafer, the unit die cost will be 40% lower because of the much higher die count of a 300 mm wafer. This savings is critical to DRAMs and other products with low profit margins.2 However, the new 300 mm fabs have some unique features that need to be considered:
- Capital intensive: Because of more expensive equipment, advanced automation and other technical necessities, the cost of a new 300 mm fab will exceed $2 billion.
- High automation level: Three hundred millimeter wafers will be more expensive and should be handled with care. However, only 13 wafer lots can be hand carried. Because of very high start rates, manual material handling will require a fleet of runners to carry production lots through the fab. Future fabs will be equipped with interbay and intrabay AMHSs, such as automated guided vehicles (AGVs) and overhead transport (OHT) that will deliver lots to the right manufacturing tools at the right time. This requires standardization of carriers, load ports, communication protocols, equipment protocol and on-equipment buffering to allow for continuous wafer processing.
- High product quality: New fabs should manufacture high-quality products and should operate at even higher line and die yield. Since 300 mm wafers are expensive (they currently cost 10 times more than 200 mm wafers3), risk of contamination, scratches, misprocessing, scrap and rework should be minimized. This is accomplished through advanced computer integrated manufacturing (CIM), MCSs, AMHSs, standard mechanical interfaces (SMIF) and minienvironments.
- High production volume: Future fabs need to have a high throughput to make investments cost-effective. As for the size of these new 300 mm lines in terms of production volume, pilot lines and low-volume lines with wafer starts per month ranging from 500 up to 2000 will be seen beginning in 1998. As the industry moves along the learning curve, medium fab lines running 10,000 wafer starts per month will be in existence by 1999. By the year 2000, high-volume lines with 20,000 wafer start per month will be running.3 The product mix of these lines may also consist of several devices with different process flows. To reduce and prevent equipment waste and improve throughput productivity, a throughput evaluation of overall equipment efficiency (OEE) and periodical review and update of OEE is essential.
- High flexibility: With demand change a reality in the semiconductor manufacturing business, 300 mm fabs should be able to react quickly to changing market demand to manufacture new products. Their learning and time-to-market periods should be as minimal as possible. This requires layout and automation flexibility. Layout should facilitate tool move-in, future expansions and tool reconfigurations. Automation (especially AMHS) should support layout flexibility through extra transport capacity, ease of track rerouting and modification and interface standardization.
- Short cycle time: To stay competitive, new fabs should minimize manufacturing cycle time. This is achievable through effective AMHS, better scheduling and dispatching software, elimination of non-value-added activities and better bottleneck and WIP management techniques. Because of high production volume, WIP management systems become critical to fab cycle time. Excess WIP in the fab will increase the cycle time of the products and jeopardize the time-to-market advantage of the company. On the other hand, the 300 mm wafers are bigger and heavier. SEMATECH suggests reducing the lot size of 25 wafers per lot to 13 wafers per lot. A smaller lot means more cassettes on the production floor. When cassettes increase, it is important to manage the stocking and tracking of these cassettes to improve control through automated tracking systems.3
Currently, international effort is focused on developing standards to address many issues of future fabs, including compatibility and inter-operability of tools from different vendors. The International 300 mm Initiative (I300I), in conjunction with the Japan 300 mm (J300), has published initial standards for 300 mm semiconductor fabs. The two consortia agreed to use SEMI to administer the international standards to be developed out of their discussion.4
300 mm fab layout design implications
The layout of a 300 mm fab plays an important role in the achievement of the goals outlined above. We can no longer rely on traditional methodologies to design a modern, cost-effective fab. Certain aspects of the new fabs, particularly material handling systems and automation, affect the development and implementation of layout design projects. Some of the factors to be considered during the layout design process are as follows:
- Larger tools with more support equipment: Three hundred millimeter fabs require tools with larger footprints that need more space. They also require more room for support equipment and facilities. This makes the ballroom concept more appealing since all support tools are installed in the sub-fab, and cleanroom construction costs are reduced. Tool manufacturers are currently trying to decrease the space requirements of their tools and support equipment while increasing throughput, although the actual result is yet to be seen.
- Use of minienvironment and SMIF: Minienvironments are popular solutions for quality improvement and contamination reduction in semiconductor fabs. A fully integrated minienvironment protects the wafers from particulate contamination and process environment composition. However, it adds to the cost and footprint of the tool and requires more space in the already expensive cleanroom. This problem can be partially resolved through integration of a minienvironment with the process tool. Both minienvironments and SMIF increase the bay width requirements that should be considered in macro and micro layout design phases.
- AMHSs: Interbay (bay-to-bay) and intrabay (tool-to-tool inside each bay) material handling systems significantly affect the layout of a fab. There are different automation techniques with different space requirements and footprints such as robots, rail robots, track vehicles, AGVs, stockers and monorails. In general, track vehicles (or rail-guided vehicles (RGVs)), AGVs and overhead monorails seem to be the best alternatives for 300 mm wafer fabs, with overhead monorails requiring the least footprint and cleanroom space.4 When making decisions about the type of AMHS to use, several factors should be considered, such as initial cost, cost per vehicle, capacity of the system, reliability, space requirement, flexibility, modularity, and load and unload time. However, floor space requirement is the most important factor that determines the cleanroom space necessary to accommodate the AMHS and its components (such as stokers, rails, etc.).
Summary
With an average capital cost of $2 billion to build a new state-of-the-art fab, semiconductor manufacturers are realizing that traditional approaches to manufacturing and cost analysis will no longer satisfy future requirements. One of the major areas neglected in the past is the facility layout design. Fabs were traditionally designed around building and construction constraints, safety codes, cleanroom protocols and utility requirements, while little or no attention was paid to the macro and micro analysis of product flow, material handling, minienvironment, ergonomics, human factors and most important, manufacturing goals.
Therefore, many existing fabs cannot operate at optimal levels because their floor layouts hamper manufacturing and material handling efficiency. It has been estimated that effective facility layouts can reduce manufacturing operating expenses by at least 10% to 30%. Recently, with the introduction of 300 mm wafer sizes, manufacturers have fully realized the importance of these factors in the design and construction of new fabs. This is where industrial engineering and its systematic view of manufacturing environments become vital.
In this paper, a brief description of macro and micro layout methodology was presented. The basic characteristics of a 300 mm fab and their effect on fab layout design were discussed. The implementation of the proposed methodology in the design of dozens of fabs in the past decade has shown significant reduction in both start-up cycle time and total installed cost (TIC) of the new facilities.
References
| Doron Meyersdorf, Ph.D. is corporate vice president for R&D with TEFEN Ltd. Industrial Engineering and System Analysis. Phone: (650) 577-8097 E-mail: doron@tefen.com |
| Abolfazl Taghizadeh is a TEFEN's project manager for the central region of the United States. Phone: (602) 753-0531 E-mail: abol@tefen.com |