As manufacturing facility costs escalate and device prices erode, the semiconductor industry faces economic and technological challenges that demand new methods of achieving continuous productivity improvements. With the cost of building and equipping a fab approaching $2 billion, OEMs must adopt a new design philosophy that will enable them to keep pace with this rapid development cycle.
According to the 1997 SIA Technology Roadmap, productivity increases historically were derived from design innovations, device shrinks, wafer-size increases, and yield improvements. But with today’s sophisticated process technology and high yields, we have reached a productivity plateau where these achievements are being offset by increased design complexity and rising facility capital and operating costs.
Working in concert with OEMs, device manufacturers can design a new facility or improve an existing one based on the particular requirements of the fab. For the future, key productivity drivers include:
• Integrated equipment replacing large, single-function machines.
• Improved process and equipment control to increase efficiency.
• Automated wafer and material-handling systems.
• Improved floor layout and facility design using modeling and simulation software.
• Coordination of the production manufacturing flow with operator control of complex reconfigurable equipment.
Market forces are converging to demand productivity increases which can be achieved through automation, standardization, and integration. Incorporating these changes constitutes a fundamental shift in equipment design and function, requiring time, effort, and education for OEMs and device manufacturers alike.
At the first Test Assembly Packaging Automation and Integration Conference last year, one study showed that high-level fab automation combined with integration nearly halves product cycle time—delivering in 34 days what used to take 64 (Figure 1).1
While automation improves equipment use and standardization achieves sought-after interchangeability, integration increases the complexity of equipment; decreases the use of ancillary, specialized equipment within the fab, and results in a more refined, simplified process. Integration has the greatest potential to provide the most pervasive influence on productivity.
In the past, fabs were equipped on a piecemeal basis by separate groups evaluating individual equipment, usually selecting one for its particular benefits without completely considering its relationship to others. Future fabs will formulate a clear strategy that encompasses all aspects of manufacturing.
On the back end of semiconductor manufacturing, integration will take the form of test and assembly workcells. Automated workcells will combine previously distinct equipment into synchronized modules. This holistic approach is expected to yield improvements in throughput and overall cycle time while optimizing floor space, power, and peripherals.
Designing Workcell Technology
Designing a workcell system begins with an understanding of the structure of the building, the equipment layout, and the dimensions of the material-loading alleys. This forms a roadmap for the physical dimensions of the overall process.
Presently, test-floor configurations usually are based on what is most effective for the tester. As illustrated in Figure 2, the traditional test-floor layout measures 6 m to 10 m × 7 m, permitting only four testers and probers. The configuration is severely restricted by the size of the tester, which extends toward the columns and wastes floor space. Integrating probers and handlers to directly interface with the tester and continuing the integration throughout the back-end process—incorporating adjacent functions such as lead inspection, marking, and mark inspection—define the workcell.
The workcell minimizes equipment footprint and maximizes the space available. Given the same restrictions as presented in Figure 2, a workcell configuration has a radically different impact.
Figure 3 shows how workcells optimize the space perpendicular to the loading alley, allowing six integrated systems in the same space where the traditional method accommodated only four. Moving more wafers within an improved footprint translates directly to higher productivity per square meter.
A wafer-probe workcell may consist of a test head submerged into the prober, eliminating the requirement for any manipulator, either stand alone or dedicated. This single feature reduces floor space, decreases equipment costs, and removes an entire process step. Ring carriers (the plate assembly supporting the electromechanical interface between the tester and the probe card that must be customized for every tester/prober combination), computer monitors and peripherals are other elements rendered practically useless by the wafer-probe workcell.
With the test head’s weight resting completely on adjustable support points within the prober’s structure, vibrations—a critical test concern with shrinking geometries—are diminished, directly improving yields. The only component between the tester pins and the device-under-test is the probe card, which can be changed by a front-loading semi- or automatic probe card changer to minimize operator intervention.
To optimize the test-floor layout, the workcell’s width must not extend beyond the existing width of the prober (approximately 1 m). Any other system-supporting assemblies, such as servers, monitors, and manipulators, must be positioned behind the prober. Unless there is full-scale automation in the facility, their height must not extend above the ergonomic height at which an operator changes probe cards and loads the cassette with wafers.
Calibrating the tester and the prober module as a complete unit without disturbing the setup is a key function of the workcell. Additionally, software components allow an operator to control the workcell from one point, as opposed to overseeing multiple stations using multiple monitors.
In the case of the final test workcell (handler integration), the challenge is to provide a flexible, dedicated solution that accommodates a multitude of device packages while minimizing changeovers.
Featuring direct hard docking, the final test workcell may combine an advanced tester with a pick-and-place handler, enabling preprogrammed hardware to be software selectable. To reduce changeovers, the workcell’s multiple sites could be clustered for sequential testing of varying package types.
Lead inspection, marking, marking inspection, lead repair, and packaging can be linked with the parts-handling system, creating a modular, multifunctional workcell. Then the entire final test process can be programmed at the beginning of the production cycle, synchronizing the production flow to accommodate large volumes in optimized batches.
Advantages of the Workcell
Traditionally, equipment was evaluated using a process model rather than an operational one. Implementing workcell technology calls for extensive simulation and modeling to help clarify manufacturing complexities while defining priorities and objectives. Clearly, advance planning substantially decreases room for error before production begins.
Workcells limit the number of equipment suppliers per fab, saving precious time currently wasted waiting for service from different suppliers. Operators spend substantially less time on setup, maintenance, and oversight of an integrated tool than on a number of distinct tools.
In traditional environments, the productivity of the individual tools is not equal to the productivity of a workcell. Given the lack of a seamless interface, traditional machines actually can result in compromising each other’s results. By engineering design, integrating equipment into a workcell eliminates weak links found in the individual tools, providing manufacturers with higher reliability and fewer machines.
The beauty of using workcells to configure manufacturing flow is their modularity, which allows them to be assembled any time and any place to replicate proven manufacturing processes.
Productivity = Competitive Advantage
With the advent of workcell technology, cost of ownership will be supplanted by the cost of productivity as the manufacturer’s metric. The productivity of integrated tools is better than that of individual tools. This advantage is further enhanced by improved labor productivity and cycle time, decreased overhead, and increased capital for device manufacturers and the subcontract test and assembly market.
Applying workcell technology requires close cooperation between suppliers and customers to control the risks inherent in new system implementation. For ease of migration, the recommended approach is to use workcell islands in tandem with the existing process and to have a controlled buffer process allowing a continuous, focused improvement implementation program.
Given the tremendous economic and technological pressures facing device manufacturers, they no longer can afford to be saddled with yesterday’s piecemeal approach to tomorrow’s integrated process environment. To successfully enter the productivity zone, device manufacturers need to transform their way of thinking from “What are we good at, and how can we get better?” into “What must we be good at tomorrow, and how can we stay competitive in the next decade?”.
Carson, D. and Mencino, S., “Integration of Fab, Assembly and Test: A Cycle Time Perspective,” Tap Automation and Integration Conference, Feb. 18-19, 1997, Mesa, AZ.
About the Author
Andrei Berar is the vice president of Interface and Integrated Systems Technologies at Credence. He joined the company in 1996 after tenuring at Teradyne and Electroglas. Mr. Berar earned an M.S. degree in mechanical engineering from Politechnical Institute in Bucharest, Romania, and an M.B.A. degree from Pepperdine University. Credence Systems, 215 Fourier Ave., Fremont, CA 94539, (510) 657-7400, www.credence.com.
Copyright 1998 Nelson Publishing Inc.