At A Glance:

Using cost-of-ownership (COO) has significant benefits for the end user. It is neither complex nor hard to do. With a few significant details about purchase, operation, utilization, and performance, users can determine the life-cycle cost of owning a semiconductor tool. Over the life of the system, equipment reliability, utilization, and yield factors may have a greater impact on COO than initial purchase costs. COO was developed for wafer fabrications tools but can easily be extended to other applications. These new applications are broadening the impact of cost modeling analysis and providing a metric for improvement for the semiconductor industry.

With equipment costs rising every generation and manufacturers increasingly sensitive to the cost per wafer, SEMATECH began developing a cost-of-ownership (COO) model in 1990. Since then, COO standards have been developed with Semiconductor Equipment and Materials International (SEMI), and a commercial COO model has been prepared through a joint development project.

Historically, purchase decisions have been based on initial purchase and installation costs. However, purchase costs do not consider the effect of equipment reliability, utilization, and yield. Over the life of the system, these factors may have a greater impact on cost-of-ownership than initial purchase costs. Lifetime cost-of-ownership per good device or wafer is generally sensitive to production throughput rates, overall tool reliability, and yield. It is relatively insensitive to initial equipment purchase price. While initial COO models were developed for wafer fabrication tools, COO can easily be extended to other applications.

A COO model has several benefits for the end user. First, the model can provide a clear estimate of the cost-of-ownership. The program can also highlight details that might be overlooked. COO provides an objective analysis method for evaluating decisions. Both suppliers and manufacturers can work from hard data to support a purchase plan. The model can also be used to evaluate process and tool design. Finally, the COO model provides communication between equipment suppliers and users. They are able to speak the same language - comparing similar data and costs using the same algorithms and equations.

Basic COO Algorithm

Estimating a tool's cost-of-ownership is neither complex nor hard. With a few significant details, users can determine the life-cycle cost of owning a semiconductor tool. The basic cost-of-ownership algorithm is described by:

CW = (CF + CV + CY)/(TPT * Y * U)

where:

CW = Cost per wafer
CF = Fixed cost
CV = Variable cost
CY = Cost due to yield loss
TPT = Throughput
Y = Composite Yield
U = Utilization

Fixed costs include purchase, installation, and facilities costs that are normally amortized over the life of the equipment. Variable costs such as material, labor, repair, utility and overhead expenses are costs incurred during equipment operation. Throughput is based on the time to meet a process requirement such as depositing or etching a nominal film thickness. Composite yield is the operational yield of the tool and may include breakage and misprocessing. Utilization is the ratio of production time compared to total available time.

Yield loss cost is a measure of the value of wafers lost through operational losses and defects. Yield models are used in COO models for estimating the relationship between contamination and yield loss or scrap. These models relate integrated circuit yield to circuit and process parameters such as device geometry and particle density.

A complete COO model requires information from many sources. To facilitate data collection, the SEMATECH model has input fields for nearly 100 different categories. By referencing the basic COO equations, these input categories may be aggregated into a few significant inputs for initial COO estimates.

The Texas A&M Center of Excellence in Manufacturing Systems Research has suggested the following grouping of COO inputs:

• Equipment cost
• Annual operating cost
• Process scrap yield
• Die scrap yield
• Downtime
• Value of wafer at process step
• Value of completed wafer.

Equipment cost includes all fixed costs such as equipment purchase, installation, and facilities support costs that are normally amortized over the life of the equipment. Annual operating cost includes all of the variable costs such as material, labor, repair, utility and overhead expenses due to equipment operation. Process scrap yield, also referred to as mechanical throughput yield, is the operational yield of the tool. Die scrap yield is the defect limited yield that is recognized at wafer test or probe.

Downtime is the non-production time lost due to scheduled maintenance, engineering usage, standby, and repair. Repair time is estimated from mean time between failures (MTBF) and mean time to repair (MTTR).

Metrology & Inspection Tools

The COO model was developed to address the economic and productive performance of a fabrication tool or specific semiconductor process step. But COO for non-process equipment such as metrology tools is also needed. Equipment cost, annual operating cost, downtime, and wafer values for the metrology tool are considered the same way process tools are. Scrap costs must be more carefully considered. Scrap caused by the inspection methods, such as destructive testing, is part of COO. Scrap identified by the inspection, but caused by previous processing, is part of process tool COO. Thus yield losses caused by the process tool must be clearly separated from yield losses caused by inspection.

Two additional costs of inspection are the cost of discarding a good device and the cost of shipping a bad device. These costs are functions of the inspection sampling plan and metrology characteristics. Minimizing the cost of shipping a bad device is one purpose of inspection. However, if either the inspection sampling plan or the metrology tool are insufficient, bad devices will be shipped. But if the inspection specifications are too restrictive, then good devices may be rejected or reworked. These costs are due to the inspection step and are part of inspection COO.

Cluster Tools and Work Cells

Estimating the cost-of-ownership of cluster tools and work cells requires more than just considering the COO of each component. The interactions among the cluster tool or work cell components must also be considered. A cluster tool is an integrated system of process, transport, and cassette modules mechanically linked together. A work cell is a group of resources that are treated as a single entity.

Cluster tools and work cells can be configured in serial, parallel, or a combination of serial and parallel. For a simple serial system, component failure rates may be added to estimate system failure rate. However, a cluster tool is not a simple serial system. In a multiple chamber cluster tool with a single wafer handling platform, the platform is serial with each chamber. Each chamber has a serial failure mode(interrupts other chambers) and a parallel failure mode (no interruption of other chambers). Each chamber is independent of the other chambers but repair may require interruption of the platform or common process supplies for safety reasons. The simplification of the cluster tool as a serial system allows a first estimate of COO. However, better estimates require more complex analysis such as simulation models.

Throughput rate depends on whether the components of the system are configured in series or parallel. If four tools were configured in series, then product would move from tool A to tool B to tool C and finally to tool D, completing the sequence. In this case the system throughput rate is defined by the bottleneck chamber, but if the tool were configured in parallel where each chamber performed the same task, the system throughput rate is defined by the sum of the chamber throughput rates. If subsequent COO analysis shows that throughput is a significant factor, then better estimates may be obtained using discrete event simulation tools.

Test & Assembly

A major difference between a wafer fab COO model and a test and assembly modelis the unit produced. Thus, by thinking in terms of cost per good unit (die or circuit) instead of cost per good wafer, test and assembly COO can be easily estimated.

The COO conditions for metrology and inspection also apply to electrical and mechanical test. Scrap identified by test but caused by the previous processes is part of process COO and must be clearly separated from yield losses caused by test. The cost of discarding a good device and the cost of shipping a bad device is one purpose of test. However, if either the test methods or specification guard bands are insufficient, bad devices will be shipped. But if specifications are too restrictive, then good devices may be rejected. Many test procedures accept a higher probability of rejecting good devices to reduce the shipment of bad devices.

Waste Disposal

Waste disposal COO includes both disposal costs and the COO of equipment required for waste treatment, disposal or reclamation. Equipment may range from simple storage containers to complex chemical reprocessing facilities. For complex systems, a complete COO model of the waste disposal system should be completed. Waste disposal COO is part of the total life-cycle COO of the process tool generating the waste. Thus waste disposal COO can be strongly influenced by design changes in the process tool and waste disposal COO strongly influences process tool COO.

Reference

1. S.Venkatesh, D.T.Phillips, "The SEMATECH Cost of Ownership Model: an Analysis and Critique," SEMATECH/SRC Contract No.91-MC-506 Final Report ÷Appendices: Cost Analysis Background Material, Reporting Period: 9/1/91 to 9/1/92, Texas SCOE, Texas A&M University.

Back to Top