Glossary¶

For: both audiences — developers without a semiconductor background, and app users who meet an unfamiliar term on screen.

Quick-reference definitions for semiconductor and library terms used in @wafertools/wafermap. The Library mapping notes on each entry are developer-facing; app users can ignore them.

Wafer¶

A circular silicon substrate, typically 200 mm or 300 mm in diameter, on which hundreds or thousands of identical chips are fabricated simultaneously. The library models a wafer as a circle with a defined diameter, edge exclusion zone, and orientation mark. Library mapping: WaferConfig (buildWaferMap input).

Die¶

A single chip instance on the wafer, identified by its integer grid position (x, y). Each die is the atomic unit of test data — it carries a bin result, optional parametric test values, and probe metadata. Library mapping: DieResult (input), Die (output).

Hard bin (hbin)¶

The physical sort category assigned by the handler after testing — it determines which physical bin tray a part is dropped into. Hard bins are the coarsest classification and are what final yield and disposition decisions are based on. Library mapping: DieResult.hbin.

Soft bin (sbin)¶

The logical classification assigned by the test program, representing the specific failure mode detected (e.g. "leakage too high", "continuity fail"). Many soft bins typically roll up into one hard bin. Soft bins are useful for failure analysis; hard bins drive physical handling. Library mapping: DieResult.sbin.

Pass bin / passBins¶

The bin number(s) considered a passing result for yield calculation. The library defaults to [1] (hard bin 1 = pass), but this must match your actual test program — some programs use a different bin for pass, or multiple passing bins. Setting this incorrectly will produce wrong yield numbers. Library mapping: WaferMapInput.passBins.

Yield¶

The fraction of tested dies that pass, expressed as a percentage. By default the library divides passing dies by the dies that count toward yield — edge-excluded and partial dies are left out of both numerator and denominator. Library mapping: YieldSummary.yieldPercent; see also gross die yield.

Filter¶

Conditions that keep only some wafers and dies: one split, bin 1, the outer ring, a range of a test, wafers below a yield. One filter applies to the gallery, a single map and every Insights tab; a wafer condition drops whole wafers from the gallery, a die condition draws the dropped dies faded on the maps, and every figure is analysed again over what is kept. Library mapping: PlotFilter (insights.filter); saved by name as SavedFilter (insights.savedFilters) and moved as a wafermap-filters file (readFiltersFile / writeFiltersFile).

Edge exclusion¶

A ring of dies around the wafer perimeter that are excluded from yield calculations. Dies near the edge are mechanically stressed during handling and may be partially outside the lithography field, making their results unreliable for lot disposition. The exclusion width is specified in millimetres. Library mapping: WaferConfig.edgeExclusion.

Notch / flat¶

The orientation mark cut into the wafer edge so that handlers, probers, and inspection tools can align the wafer consistently. A notch is a small V-shaped cut; a flat is a longer straight edge (older convention). The library uses this to orient the wafer map correctly on screen. Library mapping: WaferConfig.notch, WaferNotch.

Prober¶

The machine that moves the wafer under a probe card and makes electrical contact with each die in sequence to run the test program. The prober's controller outputs step coordinates for each die contact — these are what you supply as x/y in your die data. See also: prober step coordinates.

Prober step coordinates¶

The integer grid positions output by the prober controller, one per die contact. These are dimensionless step counts, not millimetres. The library accepts them directly and internally converts to physical geometry for rendering — you do not need to pre-convert to mm. Library mapping: DieResult.x, DieResult.y.

Test value¶

A continuous numeric measurement recorded for one die on one parametric test — for example, threshold voltage, leakage current, or ring oscillator frequency. Test values are stored keyed by test number so a die can carry results from many tests simultaneously. Library mapping: DieResult.testValues: { [testNumber]: number }.

TestDef¶

Metadata describing one parametric test: a stable integer ID (testNumber), a human-readable name, an SI unit string, and optional test and spec limits. TestDef entries drive tooltip labels, the plot-mode dropdown, and limit colouring in the rendered map. Library mapping: TestDef, WaferMapInput.testDefs.

Test limit¶

The range a parametric test value must fall in to pass (limitLow, limitHigh — STDF's LO_LIMIT and HI_LIMIT, labelled Lo limit and Hi limit on charts). A value exactly on a limit passes unless the test says otherwise (limitLowInclusive/limitHighInclusive: false, which STDF and ATDF files state per test). In value plot mode a die outside the test limits keeps its colour on the value gradient and is marked with a triangle: ▽ below limitLow, △ above limitHigh. Switch to solid pass/fail colours — green pass, blue below, red above — with passFailDisplay: 'spec' in WaferViewOptions, or "Limit pass/fail" in the toolbar. Library mapping: TestDef.limitLow, TestDef.limitHigh, WaferViewOptions.passFailDisplay.

Spec limit¶

The product specification for a parametric test (specLow, specHigh — STDF's LO_SPEC and HI_SPEC, labelled LSL and USL). Test limits are often set inside the spec limits as a guard band. Process capability (Cp/Cpk/Pp/Ppk) is measured against the spec limits when a test has both, and against its test limits otherwise; the chart and report say which. Library mapping: TestDef.specLow, TestDef.specHigh, TestCapability.limitBasis.

Functional test¶

A test with no measured value — only a recorded pass/fail outcome (a continuity check, boundary scan, or any other go/no-go test). Set testType: 'F' on the test's TestDef (default 'P', parametric) and record the outcome per die in testPass, keyed by testNumber like testValues. Selecting a functional test as the active test always displays as Test pass/fail — there is no value to put on a gradient. Functional tests are excluded from every parametric statistic (per-test stats, capability, correlation, distribution charts) and instead get pass-rate analysis: stats.functionalYield, a "Functional Tests" summary-panel table, and regional pass-rate findings. Library mapping: TestDef.testType, DieResult.testPass, getTestPassStatus().

Pass/fail display¶

The solid, categorical colouring shown in value plot mode in place of the continuous gradient — set via WaferViewOptions.passFailDisplay. 'spec' judges dies against the active test's test limits, limitLow and limitHigh (green pass / blue fail-low / red fail-high). 'test' colours dies by the tester's own recorded verdict (DieResult.testPass) instead — green pass / red fail, undirected — and is what a functional test (no measured value) always renders as, regardless of the requested display. Library mapping: WaferViewOptions.passFailDisplay.

Process capability (Cp / Cpk / Pp / Ppk)¶

A statistical measure of how well a parametric test's values fit within its limits (spec limits where it has both, its test limits otherwise) — the Insights tab's Distributions sub-tab plots one box per test, normalised so limitLow = 0 and limitHigh = 1 where both limits are defined, worst Ppk first. Cp/Cpk ("potential"/short-term capability) use the pooled within-wafer standard deviation, treating each wafer as the natural short-term subgroup; Cp ignores how centered the distribution is, Cpk penalizes an off-center mean. Pp/Ppk ("performance"/long-term capability) use the plain standard deviation across every die instead. Higher is better; ≥1.33 is a common (but process-specific) threshold for "capable." Cp/Cpk are omitted when no wafer contributes at least two values (no within-subgroup variance is computable). Tests missing one or both limits are not omitted from the chart — they still appear (muted, dashed, no capability indices), normalized onto their own observed range instead, and sorted after tests with limits by most-variable-first, since a lot with sparse limit coverage would otherwise render an all-but-empty chart. Library mapping: StatsSummary.stats.capability and LotStatsSummary.stats.capability (TestCapability), from analyzeWaferMap/analyzeWaferLot with computePerTestStats.

Welch test, Kruskal–Wallis and q-value¶

How Insights asks whether groups of wafers (the splits of a lot) really differ. Each wafer is one value, the mean of its dies, so n is wafers. Welch's t / ANOVA compares group means without assuming equal spreads; Kruskal–Wallis asks the same of ranks, so one odd wafer cannot drive it; the q-value is the Welch p adjusted (Benjamini–Hochberg) for how many tests were compared. A small p or q says a difference is unlikely to be chance, not that it matters or why it exists. See the user guide, section 10.2. Library mapping: stats/groupCompare.ts (internal); shown in the Distributions comparison table and in plot footnotes.

Robust sigma¶

A spread estimate that outliers cannot inflate: the interquartile range divided by 1.35 (the interquartile range of a normal distribution in standard deviations), around the median rather than the mean. Part average testing sets its limits from it. See the user guide, section 10.12.

Part average testing (PAT)¶

Screening that flags a die whose value on a test is far from the rest of its wafer even though it passes the test limits. Here the limits are dynamic, per wafer and test: the median ± 6 robust sigma of the passing dies, never wider than the test's own limits. A PAT outlier passes its bin and every test limit but is outside the PAT limits of at least one test. This is analysis of collected data, not a production screen, and it does not claim compliance with AEC-Q001 Rev D, whose method it follows (no static limits, no check for non-normal data). See the user guide, section 10.12. Library mapping: waferPatLimits, waferPatVerdicts (internal), the plot field { builtin: 'pat' }.

Western Electric rules¶

Four tests for a control chart that say a run of points is unlikely to be chance: one point beyond 3σ; two of three beyond 2σ on one side; four of five beyond 1σ on one side; eight in a row on one side of the centre. Each is flagged at the point that completes it. See the user guide, section 10.7. Library mapping: chart: 'control'.

Probability plot¶

Each group's sorted values against their cumulative share, on an axis stretched by the normal distribution so normal data falls on a straight line. A shift, a different spread, a second population and outliers each look different. See the user guide, section 10.4. Library mapping: chart: 'probability'.

Reticle¶

The rectangular exposure field used in photolithography, typically containing a fixed grid of die sites. One reticle field is stepped across the wafer repeatedly to pattern the full surface. The library can overlay the reticle grid and attribute findings to specific reticle positions, which is useful for identifying systematic defects tied to a particular mask location. Library mapping: WaferMapInput.reticleConfig, ReticleConfig.

Validity limit¶

The range a real measurement of a test lies in (validLow, validHigh; tsmap's lvl/uvl). A tester that runs out of range records its rail, such as a current held at the compliance clamp or an open-circuit value, and that number is not a reading. Validity limits are a third kind beside the test and specification limits: those judge good against bad, these judge a reading against a clamp, and a clamped value is neither. Library mapping: TestDef.validLow, TestDef.validHigh.

Excluded value¶

A test value that lies outside the limit set the value filter uses (the validity limits by default). It is treated as missing for that test on that die, in the map, the statistics and every chart, and it is counted wherever a population is shown. Bins and each die's recorded pass/fail are the tester's own and are unchanged. Library mapping: WaferMapInput.valueFilter, WaferMapResult.valueFilter, CapabilityDatum.excluded.

Reticle cell¶

A die's position inside the reticle: which of the field's dies it is, such as column 1, row 3 of a 4 × 3 field. The same cell recurs in every shot across the wafer, so a cell that is bad everywhere points at the mask or lens rather than at the wafer. Shown as Reticle cell (1, 3) and available once a reticle is set. Library mapping: getReticleCell, DIE_REGIONS.reticleCell.

Reticle shot¶

Where on the wafer the reticle was placed for a die: one exposure of the mask, numbered by its column and row among the exposures (the one containing the anchor die is (0, 0)). Each shot is a different place on the wafer, so a shot that is bad points at something that varies across the wafer, such as focus, tilt or the edge. Not to be confused with the reticle cell, which is a position inside the mask. Shown as Reticle shot (-1, 2). Library mapping: getReticleShot, DIE_REGIONS.reticleShot.

Lot¶

A batch of wafers processed together through the fabrication line, typically 25 wafers. All wafers in a lot share the same process conditions and are usually tested together. The library's lot stack feature lets you combine multiple wafers from a lot into a single aggregated map. Library mapping: WaferMapInput.lotStack, LotStackConfig.

In API names — analyzeWaferLot, LotStatsSummary, lotStack, level: 'lot' — "lot" means a set of several wafers, which may span more than one physical lot or record no lot ID at all. What the user sees is stricter: a panel, report or finding says "lot" only when every wafer records the same lot ID ("Lot LOT123 · 13 wafers", "lot median"); otherwise it names the wafers ("26 wafers from 2 lots", "median of all wafers").

Wafer attribute¶

A fact about a whole wafer that came with the data or was assigned afterwards: its lot, product, test program, temperature, operator, test date, slot, or a label such as a process split. A wafer attribute separates wafers; it is what Group by pools them on. A split (TT, FF, SS…) is one wafer attribute, not a separate mechanism. Library mapping: WaferMetadata, the attributes render option, attributeLabel.

Die region¶

A category a die falls into from where it sits, calculated rather than recorded: its ring, quadrant, reticle cell or reticle shot. A die belongs to every one of them at once, which is why a region separates the dies of a chart (Compare by) and is never a Group by choice. Library mapping: DIE_REGIONS, availableRegions.

Verdict¶

The outcome of a go/no-go: a functional test's pass or fail, or a derived test that gives a yes/no. A verdict has no value to plot, so the plot builder offers it as a category (Verdicts, shown as Scan Chain (pass/fail)) to colour or compare by. Library mapping: getTestPassStatus, DieResult.testPass.

Group by¶

Pools whole wafers by a wafer attribute (lot, split, temperature) for every chart on the Insights tab, so each chart draws one group per value. It takes wafer attributes only. To separate the dies within each wafer, use Compare by.

Compare by¶

What a single plot separates its marks by: its Colour, and on a bar or box its Categories. It takes a wafer attribute or a die region, a bin, a site or a verdict. Where Group by pools wafers for the whole tab, Compare by acts on one plot.

Lot stack¶

An aggregated wafer map computed from multiple individual wafers, showing a per-die statistic across the lot (mean, median, bin count, etc.). Lot stacking reveals systematic spatial patterns — for example, a recurring hot spot at the same reticle position across all wafers — that would not be visible on any single wafer. Library mapping: LotStackConfig, WaferViewOptions.plotMode: 'stackedValues' | 'stackedBins' | 'stackedSoftBins'.

Retest / retest policy¶

Some probers contact a die more than once, either due to contact failures (a retry) or deliberate multi-touch sequences. The retest policy controls which result the library keeps when a die has multiple records: 'last' (default, keeps the final probe), 'first' (keeps the earliest probe), 'best' (keeps the passing result or the best bin), or 'worst'. Library mapping: WaferMapInput.retestPolicy, Die.retestCount.

STDF¶

Standard Test Data Format — the binary file format output by most ATE systems after a wafer test run. STDF encodes die bin results, parametric test values, and lot/wafer metadata in a compact binary record structure. The library does not parse STDF itself: map the records to DieResult rows yourself (die position, bins and test results), or open the file in tsmap, which reads STDF and ATDF directly.

ATE¶

Automatic Test Equipment — the tester (e.g. Teradyne, Advantest) that executes the test program and records pass/fail and parametric results for each die. The ATE produces the STDF file and drives the prober. It is the upstream source of all data the library visualises.

Gross die yield¶

Yield with edge-excluded dies counted in the denominator but never as passes, so edge losses show up in the figure instead of being excluded from it — the usual way to quantify yield lost to edge effects. The library reports it alongside ordinary yield when edgeDieYieldMode: 'denominator-only' is set. Library mapping: YieldSummary.yieldPercentGross, WaferMapInput.edgeDieYieldMode.

Gross die per wafer¶

The number of complete die sites that fit on a wafer — the ceiling on how many dies a wafer can yield, and the denominator in many cost-per-die estimates. A die straddling the edge is not counted. Library mapping: buildWaferMap({ layout: true, waferConfig, dieConfig }).dies.length.

See also: guide.md for worked examples, api.md for full type and option reference.