Throughput

Notes

EUV scanner throughput in wafers per hour has been the single most watched equipment metric of the last decade.

Overview

Throughput is the rate at which a tool, a line or a whole fab completes work: wafers per hour for a piece of equipment, or wafer outs per week or month for the factory. A fab's throughput is set by its bottleneck, the most heavily loaded step, rather than by the average tool, so improving a step that is not the bottleneck does not raise output.

How it works

A tool's throughput depends on its processing time per wafer, how many wafers it handles at once, the time spent loading, unloading and changing recipes, and how often it is down for maintenance or waiting for material. Manufacturers quote nominal throughput for a recipe, but the figure a fab sees is lower once availability, set-ups and idle time are counted. SEMI standards define how to measure equipment states and productivity, including overall equipment effectiveness, which multiplies availability, performance and quality.

At factory level, throughput, work in progress and cycle time are linked: by Little's law, the average work in progress equals throughput multiplied by average cycle time. Running a fab closer to full capacity raises throughput but makes queues, and so cycle time, grow quickly. Fabs therefore plan capacity with some slack on bottleneck tools, and industrial engineers use simulation to decide how many of each tool to buy.

Lithography scanners are often the bottleneck, because they are very expensive and every patterned layer passes through them. Their throughput depends on exposure dose, so a slower but more sensitive resist or a higher dose for better patterns has a direct capacity cost.

Examples

• A scanner rated at a certain number of wafers per hour for a standard exposure dose.
• A fab finding that one etch step has become the bottleneck after a product mix change.
• Overall equipment effectiveness tracked for every tool group using SEMI definitions.
• Batch furnaces processing many wafers at once, so their throughput depends on how full each load is.
• An assembly line measured in units per hour at the wire bonders.

Why it matters

Throughput sets how many wafers a fab can ship and therefore its revenue and cost per wafer. Industrial engineers, capacity planners, equipment engineers and fab managers watch it daily, and equipment suppliers compete on it.

In the chip lifecycle, throughput assumptions shape fab and tool purchase plans before production, constrain how quickly a new product can ramp, and are tracked and improved throughout volume manufacturing.

It matters most to: Backend Industrial Engineer, Equipment Engineering Manager, Fab Equipment Engineer, Fab Industrial Engineer, OSAT Capacity Planner

Common confusions

Not the same as yield

Yield is the share of good units among those made. Throughput is how many units are made in a period. Throughput vs yield

Not the same as capacity

Capacity is the maximum throughput a tool or fab could reach. Actual throughput is usually lower.

Not the same as data throughput

In networking and computing, throughput means data transferred per second. In a fab it means wafers or parts processed per unit of time.

Related standards

• SEMI E10 (equipment reliability, availability, maintainability and utilisation)
• SEMI E79 (definition and measurement of equipment productivity)

Sources

1. SEMI E79: Specification for Definition and Measurement of Equipment Productivity (opens in a new tab), SEMI, 2026-10-22
2. SEMI E10: Specification for Definition and Measurement of Equipment Reliability, Availability, and Maintainability (RAM) and Utilization (opens in a new tab), SEMI, 2026-10-22
3. Little's law (Wikipedia) (opens in a new tab), Wikimedia Foundation, 2026-10-22
4. Overall equipment effectiveness (Wikipedia) (opens in a new tab), Wikimedia Foundation, 2026-10-22

Related terms

• Cycle time
• OEE
• Capacity utilisation

Where it appears on the Hub

Used on 76 pages of the Hub. In each group, the pages that mention it most come first.