Overall equipment effectiveness (OEE) for fab equipment is the product of three factors (availability, performance and quality) expressed as a single percentage that captures how much of a tool’s potential productive output is actually realised. SEMI E10 supplies the underlying equipment-state model and the reliability metrics, including MTBF and MTTR, that make that number consistent and comparable across tools and sites.
This article defines the SEMI E10 six-state model, works through an OEE calculation for a legacy 200mm etcher, separates MTBF from MTTR, gives realistic benchmarks for legacy fabs, and sets out a tracking template an engineer can put into use immediately. The aim is a measurement system that survives audit and drives uptime decisions, not a vanity figure.
What is SEMI E10?
SEMI E10 is the SEMI standard titled Specification for Definition and Measurement of Equipment Reliability, Availability, and Maintainability (RAM) and Utilization. It is the canonical reference that fabs use to define equipment states and to calculate reliability, availability and maintainability metrics in a way that is consistent between equipment owners and suppliers. Where uptime numbers are quoted in a purchase agreement, a service contract or an internal scorecard, SEMI E10 is usually the definition behind them.
The standard does two things. First, it partitions every hour of a tool’s life into mutually exclusive states, so there is no ambiguity about whether a given period counts as downtime. Second, it defines the metrics (mean time between failures, mean time to repair, availability and utilisation among them) that are derived from the time spent in each state. SEMI E10 is the granddaddy of the SEMI productivity standards; the related SEMI E79 standard builds on it to define overall equipment efficiency itself, and SEMI E116 adds a lighter real-time tracking model.
The SEMI E10 six-state model
SEMI E10 divides total calendar time into six mutually exclusive equipment states. Every minute the tool exists falls into exactly one of them, which is what makes the metrics built on top reconcilable. The states form a hierarchy: total time splits into scheduled and non-scheduled, scheduled splits into productive, standby, engineering and downtime, and downtime splits into scheduled and unscheduled.
| Equipment state | Definition | Counts as |
|---|---|---|
| Productive time | Tool is processing wafers, including regular production, work, engineering runs done while producing, and rework. | Up, producing |
| Standby time | Tool is available and able to run but is not processing, no operator, no wafers, or no upstream demand. | Up, idle |
| Engineering time | Tool is used for engineering experiments rather than production, such as process characterisation. | Up, non-production |
| Scheduled downtime | Planned non-productive time: preventive maintenance, setup, scheduled wet cleans, qualification. | Down, planned |
| Unscheduled downtime | The failed condition, repair, waiting for parts, waiting for an engineer, unplanned corrective work. | Down, unplanned |
| Non-scheduled time | Time the tool is not scheduled to be available: unworked shifts, holidays, installation, long-term storage. | Out of scope |
The state that matters most for uptime work is unscheduled downtime, the standard’s definition of a failed tool. It is the numerator of most reliability problems and the first place a maintenance programme should attack. Distinguishing it cleanly from scheduled downtime is the single most common measurement error: counting a planned wet clean as a failure inflates the apparent fault rate and corrupts MTBF.
The three factors of OEE for fab tools
Overall equipment effectiveness is the fraction of potential good output a tool actually delivers, calculated as the product of three independent factors. Each factor isolates a different class of loss, so the breakdown tells an engineer where the productivity is leaking rather than only that it is leaking.
OEE = Availability × Performance × Quality.
Availability
Availability is the fraction of scheduled time during which the equipment is up and able to run. It is the factor most directly governed by SEMI E10, because it is computed from the equipment-state times: availability is uptime divided by the scheduled time the tool was meant to be operable. A tool that is down for repair or waiting on a part is, by definition, not available, which is why unscheduled downtime hits this factor hardest.
For legacy 200mm tools, availability is typically in the 85–95% band when the install base is well supported, and falls below it when obsolescence starts to bite, a failed subsystem with no qualified spare can turn a four-hour repair into a four-week wait, and that wait is pure unscheduled downtime. Availability is therefore the factor where parts supply and repair turnaround show up directly in the number.
Performance
Performance, sometimes called the efficiency or speed factor, is the ratio of actual throughput to the throughput the tool would achieve at its rated cycle time. It captures the losses that do not stop the tool but slow it: minor stoppages, reduced process rates, ageing components running below specification, and recipe inefficiencies. A pump losing conductance or an RF generator de-rating with age erodes performance without ever registering as a failure, which is why a tool can be highly available yet still under-deliver.
Quality
Quality is the fraction of processed wafers that meet specification, good wafers out divided by total wafers processed. Scrapped wafers, rework and yield loss attributable to the tool all reduce this factor. In semiconductor manufacturing the quality factor is usually the highest of the three, often above 99%, because a tool producing out-of-spec wafers is normally taken down and requalified rather than left running. A persistent quality problem is frequently the first visible symptom of a developing hardware fault, such as a degrading electrostatic chuck or a drifting mass-flow controller.
How is OEE calculated for fab tools?
OEE is calculated by measuring each of the three factors over a defined observation period and multiplying them. The worked example below uses a single 200mm etch chamber over a 168-hour week, with the tool scheduled to be available for the full period.
| Quantity | Value | Source |
|---|---|---|
| Scheduled time | 168 h | Full week, tool scheduled up |
| Scheduled downtime (PM, wet clean) | 12 h | Planned |
| Unscheduled downtime (one RF fault + wait for part) | 14 h | Failure |
| Standby (no wafers) | 10 h | Idle but available |
| Productive time | 132 h | Processing |
| Rated throughput | 30 wafers/h | Tool specification |
| Actual wafers processed | 3, 500 | Counted |
| Wafers meeting specification | 3, 468 | Counted |
The three factors follow directly:
- Availability = (scheduled time − all downtime) ÷ (scheduled time − scheduled downtime). Uptime is 168 − 12 − 14 = 142 h; the operable base after planned downtime is 168 − 12 = 156 h. Availability = 142 ÷ 156 = 91.0%. (Where a fab defines availability against total scheduled time, the figure would be 142 ÷ 168 = 84.5%; the basis must be stated.)
- Performance = actual throughput ÷ rated throughput over productive time. Rated output across 132 productive hours is 132 × 30 = 3, 960 wafers. Performance = 3, 500 ÷ 3, 960 = 88.4%.
- Quality = good wafers ÷ total processed = 3, 468 ÷ 3, 500 = 99.1%.
OEE = 0.910 × 0.884 × 0.991 = 79.7%. The breakdown is the useful part: availability and performance are the two leaks, and the 14 hours of unscheduled downtime, much of it a wait for a part, is the largest single recoverable loss. That is a parts-and-repair problem, not an operator problem, and the metric points straight at it.
MTBF versus MTTR
SEMI E10 defines the two reliability and maintainability metrics that sit underneath availability. They answer different questions and are improved by different actions, so they should never be collapsed into one number.
| Metric | Definition | What it measures | Improved by |
|---|---|---|---|
| MTBF | Mean time between failures, productive time divided by the number of failures in the period. | How often the tool fails (reliability). | Better PM, qualified parts, root-cause fixes |
| MTTR | Mean time to repair, total unscheduled repair time divided by the number of repairs. | How long each failure takes to clear (maintainability). | Parts on hand, trained engineers, documented procedures |
MTBF is a reliability metric: a higher MTBF means failures are rarer. SEMI E10 also defines mean cycles between failures (MCBF) and mean wafers between failures (MWBF) for tools where a wafer or cycle count is a more honest denominator than clock hours, a useful refinement for etch and deposition chambers whose wear tracks RF hours or wafer throughput rather than calendar time. MTTR is a maintainability metric: a lower MTTR means each failure is cleared faster. The two combine into availability through the relationship availability ≈ MTBF ÷ (MTBF + MTTR), which makes the trade-off explicit. A tool with an MTBF of 300 hours and an MTTR of 6 hours has an inherent availability of about 98%; let parts scarcity push the effective MTTR to 60 hours and the same tool drops to about 83%. On legacy platforms, the MTTR side is where obsolescence does its damage, because the repair clock keeps running while the fab hunts for a discontinued part.
Utilisation and the difference from availability
Utilisation is the fraction of available time the tool actually spends producing, productive time divided by total time, in the SEMI E10 framing. It is distinct from availability: a tool can be 95% available yet only 65% utilised because there are no wafers to run, which is a planning or demand condition, not an equipment fault. Confusing the two leads to the wrong fix; chasing equipment reliability when the real loss is standby time wastes maintenance budget. For legacy fabs, utilisation typically sits in the 60–80% band, and the gap between availability and utilisation is standby time that maintenance cannot recover.
Benchmarks for legacy and 200mm fabs
Benchmarks give the raw OEE number meaning. The figures below reflect commonly cited targets; what matters more than hitting a published number is measuring consistently against the same SEMI E10 basis over time so trends are real.
| Factor / metric | Bottom quartile | Industry average | World-class target |
|---|---|---|---|
| Availability | ~80% | ~88% | ≥95% |
| Performance | ~85% | ~90% | ≥95% |
| Quality | ~98% | ~99% | ≥99.9% |
| OEE (product) | ~65% | ~75% | ≥85% |
Most fabs sit at 65–75% OEE, and the gap to world-class is dominated by availability loss, unplanned downtime, rather than performance or quality. For a legacy 200mm tool, that availability loss is disproportionately a parts and repair-turnaround problem: the failure rate may be acceptable, but each failure costs too many hours because the supply route for a qualified spare is slow or uncertain. A world-class OEE of 85% is, in practice, the Nakajima TPM target of 90% availability, 95% performance and 99% quality carried into the fab.
An OEE and reliability tracking template
A workable tracking system does not require an MES integration to start. The minimum dataset, logged per tool per shift, is enough to compute every metric in this article and to expose the obsolescence-driven losses that hurt legacy fabs. SEMI E116, the equipment-performance-tracking standard, provides the lighter Busy/Idle/Blocked state model many fabs use for this real-time logging, mapping back to the SEMI E10 states for reporting.
- Log state time. Record minutes in each of the six SEMI E10 states per shift. This single discipline makes availability and utilisation calculable and forces the scheduled-versus-unscheduled distinction.
- Count failures and repairs. Each entry into unscheduled downtime is one failure; record its start, end and a fault code. Productive time ÷ failure count gives MTBF; repair time ÷ repair count gives MTTR.
- Count wafers. Total processed and total good, per shift, yield the performance and quality factors.
- Tag the wait. Within each unscheduled-downtime event, separate hands-on repair time from waiting time (for a part, for an engineer). This is the obsolescence signal: rising wait time against flat repair time means the supply chain, not the tool, is the constraint.
- Trend, do not snapshot. Plot OEE and its factors weekly. A single number is noise; a declining availability trend with rising MTTR is a tool heading toward an unsupported failure.
Step four is the lever most fabs miss. When tracking shows that effective MTTR is being driven by part-wait rather than wrench time, the corrective action is not more maintenance training, it is securing a qualified second-source or refurbished supply route before the next failure, so the repair clock stops being held hostage by a discontinued OEM part. A structured uptime programme of this kind is covered in the semiconductor equipment repair guide.
Frequently asked questions
What is SEMI E10?
SEMI E10 is the SEMI standard for the definition and measurement of equipment reliability, availability, maintainability (RAM) and utilisation. It defines six mutually exclusive equipment states, productive, standby, engineering, scheduled downtime, unscheduled downtime and non-scheduled time, and the metrics derived from them, including MTBF, MTTR, availability and utilisation. It is the common basis fabs and equipment suppliers use so that uptime figures mean the same thing on both sides of a contract.
How is OEE calculated for fab tools?
OEE is calculated as Availability × Performance × Quality over a defined observation period. Availability is uptime divided by the operable scheduled time; performance is actual throughput divided by rated throughput over productive time; quality is good wafers divided by total wafers processed. Multiplying the three gives a single percentage, while the individual factors show whether the loss is downtime, speed or scrap.
What is the difference between MTBF and MTTR?
MTBF (mean time between failures) measures how often a tool fails and is a reliability metric, improved by better preventive maintenance and qualified parts. MTTR (mean time to repair) measures how long each failure takes to clear and is a maintainability metric, improved by having parts on hand and trained engineers. They combine into availability through availability ≈ MTBF ÷ (MTBF + MTTR).
What is a good OEE for a semiconductor fab?
Most fabs operate at 65–75% OEE, with the world-class target around 85%, built from roughly 90% availability, 95% performance and 99% quality. The gap to world-class is usually dominated by availability loss from unplanned downtime. For legacy 200mm tools, that availability gap is frequently a parts-supply and repair-turnaround problem rather than a reliability problem in the tool itself.
What is the difference between availability and utilisation?
Availability is the fraction of scheduled time the tool is up and able to run; utilisation is the fraction of time it is actually producing. A tool can be highly available yet poorly utilised because there are no wafers to run, which is a demand or planning issue, not an equipment fault. Separating the two keeps maintenance effort on real reliability losses rather than standby time.
How do you measure OEE on a legacy tool without an MES?
Log the six SEMI E10 state times per shift, count failures and repairs to get MTBF and MTTR, and count total and good wafers to get performance and quality. SEMI E116 offers a simpler Busy/Idle/Blocked tracking model for real-time logging. Within each downtime event, separating hands-on repair time from part-wait time exposes the obsolescence-driven losses that distinguish legacy fabs, all without an automated data system.
Improving the number that matters
Measuring OEE and the SEMI E10 metrics is the first step; acting on what they reveal is where uptime is won. When the data shows that unscheduled downtime and a rising MTTR are driven by waiting for discontinued parts, the fix is a qualified supply route, not more diagnostics. ESCT Fab Support helps owners of legacy LAM, TEL and AMAT tools defeat obsolescence with qualified second-source parts, repair and refurbishment that shorten the repair clock and protect availability. To turn a declining availability trend into a recovery plan, talk to the team about parts and repair support for your platform.
REQUEST A QUOTE
Fill out the form below to request a quote for the part/s you require, making sure to include the OEM part number.
If the part you are looking for is not listed, get in touch and we will try to source it for you through our wide network of partners and suppliers.