by Tom Connolly and Monty Stranski, IDC
The impact of 300-mm tools, automated material handling systems (AMHS) and future semiconductor processing technologies on fab layout methodology.
The manufacturing era that consisted of clean hoods, beaker wet processing and operator garb consisting of a smock and hair net is long gone. Not much, if any, consideration was given to the layout of the processing area, which was little more than a large lab, in many cases.
With the advent of more complex products and processes, the manufacturing stage saw the onset of larger wafer size, automation, shorter development cycles, shorter time between technology releases, and increased pressures on time to market, productivity, costs, etc. These factors have had significant impact on process tool vendors, automation, both hardware and software, and the facility design itself. We have come a long way from the clean hood room to the megafab of today.
The layout of a manufacturing line has long been impacted by various considerations; process complexity and mix, levels of automation available, contamination requirements, even the personal preferences of significant process or manufacturing personnel.
Traditional concepts
First, it is important to have a view of common practices and some of the rationale that has been applied for having gotten there. The idea of tool farms, a collection of like tools in function and type in close proximity, has manifested itself in many factory layouts.
The farm concept evolved as a result of the need for high-volume production and tool availability. Because tool reliability had the most adverse impact to availability, having parallel tools ready in near proximity allowed manufacturing to continue.
This provided high tool utilization but increased cycle time and work in progress (WIP). However, tool reliability steadily progressed and as processes got more complicated, and tooling began to “cluster” multiple processes in a single tool, demands for shorter cycle times and lower WIP drove many to reconsider the proliferation of large tool farms. The combination of both a farm and some distributed tool approach is widely accepted as a partial solution.
Current environment
Today, 300-mm processing has raised a whole set of additional concerns. Along with the continued advances in process tool technology and factory automation, the need for stricter contamination control and separation of process steps and tools can play heavily in the layout design process.
The perceived need for automated materials handling system (AMHS) in 300-mm production has raised considerable layout issues. The ergonomics of manually transporting and tracking a full 25-wafer 300-mm front-opening universal pod (FOUP) has driven the more pervasive inclusion of full automation. The desire to separate copper processing from other processes has influenced the design of the facility itself, the strategy for dedicating process tools and the handling of product and FOUPs. Air handling scenarios need revisiting as well as the location and operation of support spaces.
Viewing this as an opportunity to revisit and test the validity of entrenched paradigms, the IDC Industrial Engineering department developed a methodology for better understanding how the trends in process, tool and manufacturing technology impact the facility, layout and factory operation in the future.
A prime purpose for employing this method was to determine if there was a significant difference in laying out the factory in a farm concept versus a work cell centered on a process sequence when specifically addressing 300-mm production. The data gathered here were to serve as input for a prototype 300-mm factory designed by IDC which would address the facility and manufacturing needs of a new generation of high-volume fabs.
The methodology
 Figure 1. 300-mm factory design process |
A systematic approach was developed as shown in Figure 1.
The analysis starts with a specific process definition. This process completely defines the process step sequence from which tool type requirements are derived. A static model is then exercised which combines various tool parameters, including availability, throughput, utilization, space, etc. The manufacturing requirements are added to the model and a first-order tool list is generated and available for use.
To exercise the method and evaluate various scenarios, the Sematech 0.15-micron copper process for 300-mm wafers was selected. The process contains seven levels of metal using a spin on low-k dielectric. The wafer start level was 20,000 wafers per month. IDC designers created the necessary CAD information to start the physical layout process and a facility requirement matrix to assess overall building service needs.
The current effort concentrated on a process-sequence-oriented work cell approach to the floor layout. The process was laid out step by step and recurring and re-entrant process loops were determined. This served to establish the starting point for determining the tooling content of the work cells. A macro analysis of the exact process flow established the preferred configurations of bays and open areas and defined the adjacent process functional areas. In addressing the most conservative approach to copper contamination control, the process was divided into a front end of line (FEOL) portion and a back end of line (BEOL) portion. Separation of the tools, AMHS and the clean air stream can be satisfied by this concept.
Results
 Figure 2. Layout of prototype 300-mm fab |
The resulting layout (see Figure 2) embodies a center spine approach along which an interbay AMHS resides. The interbay system connects the various process bays.
Equipment is located in either bays or large ballroom areas off the center aisle. Intrabay AMHS provides wafer movement between equipment in the bay/ballroom areas. Stockers are strategically located at the bay ends to interface the interbay and intrabay systems.
Once the major directions mentioned above were set, tool distribution began. Tools identified within re-entrant process loops were placed in the same or adjacent bays. The static model initially determined the tool counts. Tools were added as needed to maintain process flow or to provide metrology capability within the work cell.
Upon completion of the first order layout, the tool list and utility matrix were updated to the new tool counts. The next step was to look at the AMHS design and determine the best combination of stockers, system designs and transport types (OHT, AGV and PGV) and assess the impact of the cell approach to the performance of the system.
 Figure 3. AMHS interbay statistics |
The results of this analysis helped determine stocker size and location; types and number of product vehicles required and provided a first-pass look at the AMHS design, transport times and overall system performance. Figure 3 contains the statistics of the modeled layout.
Future work
A first-pass model for the dynamic simulation of the cell approach is currently in design. This is the next step in determining the manufacturing and operational characteristics of the assumptions incorporated in the layout of the IDC factory. Future work will center on issues of increased flexibility, product mix and the effects on the factory. Migration of technology on a running factory (e.g. aluminum to copper), lithography ground rule changes, FOUP and reticle management are all factors to be addressed in future activity.
AMHS integration into the methodology, while incorporated now, needs closer association with the modeling effort. Further work is required using other more complicated process descriptions. IDC intends to meld tool utility data into the model to assess service requirements as a function of processes, tooling and factory loading.
Conclusion
The advent of 300-mm wafer processing has provided both the opportunity and the need to introduce a structured methodology into the factory layout and design process. IDC has constructed such a methodology and is currently using it to determine various alternatives to optimizing the facility and the manufacturing floor.
It has focused on a process-centric work cell approach and incorporates the requirements of handling automation and contamination control desired for 300-mm processes. Initial results are highlighting areas of benefit, opportunity and further study.
This paper was presented at the SEMICON China 2000 Technical Symposium. It has been reprinted with permission of Semiconductor Equipment and Materials International.
Thomas Connolly is the manager of Industrial Design Corp.'s (IDC) Advanced Technology department. IDC is a global leader in the design of semiconductor facilities.
Monty Stranski is a member of IDC's Industrial Engineering department. He has worked on a many projects involving full project cycles from concept through implementation and startup.
300 mm: Rethinking contamination control design
By Michael O'Halloran and Larry Hennessy
The advent of 300-mm technologies has prompted revised thinking on how cleanrooms should operate—as well as how cleanroom design should be executed.
Contamination issues are particularly critical at the molecular and atomic levels. For example, the use of copper at the interconnect processing level is causing significant concern related to atomic level copper contamination. New photoresist processes are also vulnerable to airborne molecular contamination.
Big changes have been made in airflow management strategies, tool layout and automation techniques to separate processes that are sources of contamination from processes that are vulnerable to contamination. This requires dynamic give and take between industrial engineers doing layout and mechanical engineers developing airflow systems.
The standardization of the front opening universal pod (FOUP) allows fab and automation layout designers to generate design configurations that are more flexible to the contamination control requirements of these new processes. The organization of tool clustering and both interbay and intrabay automation can be tailored to the present and future process requirements of the fab.
Many owners are implementing or anticipating future separation of front end of line (FEOL) and back end of line (BEOL). Of course, separation strategies will change over time as new technologies continue to be introduced to the fab environment. Issues related to airflow, fab layout, automation and operational requirements are in a state of continuous dynamic evolution.
It's not adequate to simply build a fixed wall between these two types of processes, because the wall will have to move as the technology evolves. Building future flexibility into fab design is of paramount importance.
Providing isolation of contaminating processes without sacrificing flexibility is one of the most difficult challenges. It can only be solved through a multi-discipline approach that closely integrates consideration of all fab design and construction elements. Locations of tools and WIP location and automation paths may need to be altered. Transfer of product between cleanliness zones may change in volume and location.
Accommodation of gown rooms is another area of opportunity for contamination control. Maintenance technicians who work around copper-contaminated tools shouldn't come into contact with non-copper tools or with other operators and maintenance technicians assigned to non-copper areas of operation.
This is both an operations protocol issue and a facility design issue. Owners often need assistance in the areas of both facility design and operations protocol to resolve contamination issues associated with copper “cross-contamination.”
Many owners, particularly DRAM manufacturers, have discovered the importance of selecting “clean” cleanroom construction materials. Proper attention to the out-gassing properties of cleanroom construction materials has resulted in some all-time best startup yields for recently constructed facilities.
Many logic manufacturers in the Western world are more attuned to copper-related contamination issues, while many DRAM manufacturers in Asia are more focused on airborne molecular contamination concerns. The two forms of contamination threat, atomic and molecular, will need to be addressed in single fabs as more diverse manufacturing processes converge in single fabs as a result of such new technology drivers as process-shrink and system-on-a-chip.
The good news is that while the industry's latest-generation fab technologies bring increased contamination concerns, they also offer a broader range of tools with which to address those concerns.
Michael O'Halloran is director of technology and Larry Hennessy is AMHS technical manager, industrial engineering department at IDC, a global design and construction firm based in Portland, OR.