Metric · Fab Ops, Yield & Metrology
Cycle time
The elapsed time a lot takes from wafer start to wafer out, including all queueing, not just processing. Fab cycle time is usually quoted in days per mask layer or as total days for a technology.
- Also known as
- Turnaround time, TAT
- Type
- Metric
- Domain
- Fab Ops, Yield & Metrology, Fab operations
Notes
For leading-edge logic, total fab cycle time commonly runs three to four months before test and packaging.
Overview
Cycle time in a wafer fab is the elapsed time a lot takes to travel from wafer start to wafer out, counting every hour spent waiting in queues as well as the time actually spent inside process tools. It is one of the main measures of how well a fab runs, because it sets how quickly new products, engineering experiments and customer orders come out of the line. Fabs commonly express it in days per mask layer, so that processes with different numbers of layers can be compared.
How it works
Only part of a lot's time in the fab is spent being processed; the rest goes on waiting for tools and transport, engineering holds, rework and measurement. Raw process time, sometimes called theoretical cycle time, is how long the lot would take if it never waited, and the ratio of actual cycle time to that figure is known as the X-factor. Days per mask layer divides the total cycle time by the number of lithography masking layers in the process.
Queueing is usually the largest component. INFICON, whose FabTime software tracks cycle time in fabs, names three drivers at tool level: utilisation, variability in process and arrival times, and the number of tools qualified to run a step. As a tool group approaches full utilisation, waiting time rises steeply, so a fab that loads its tools harder gains output but lengthens cycle time.
Cycle time also grows with process complexity. Each extra mask layer, and each multipatterning step used in place of a single exposure, adds more passes through lithography, etch and deposition. Little's law links the quantities: average work in progress equals throughput multiplied by average time in the system, so for a given output, cutting cycle time means carrying less work in progress.
Examples
- A planner reporting a fab's weekly average in days per mask layer alongside its X-factor, so trends can be compared across products.
- The first wafers of a new product run as hot lots that jump the queues, so first silicon arrives sooner at the cost of slowing other work.
- An industrial engineer qualifying a second tool for a bottleneck lithography step to cut the queue in front of it.
- A 2017 industry analysis that put a typical fab at about 1 to 1.5 days per layer and the best fabs near 0.8 days.
Why it matters
Shorter cycle time means faster learning during a process ramp, quicker response when demand changes and less money tied up in work in progress. Production planners, fab industrial engineers, WIP and dispatch engineers and capacity planners manage it day to day, while product teams and customers experience it as lead time.
Related terms
Part of the Fab Ops, Yield & Metrology glossary, with every term in the topic and the lifecycles, tools and standards that use them.
Where it appears on the Hub
Used on 31 pages of the Hub. In each group, the pages that mention it most come first.
Lifecycle stages and gates
Roles
Suggest an improvement
Suggest an improvement to this entry
A clearer definition, an example, a correction or a source an editor can check. Editors review every suggestion before the entry changes.
Last reviewed . Written by The Semiconductor Hub editorial team. The full glossary