Yield Loss Pareto
A yield loss Pareto chart ranks every identified yield-loss mechanism by its actual impact on total yield, from biggest contributor to smallest, following the broader Pareto principle that a small fraction of causes typically accounts for the large majority of the total effect. In a fab's yield-engineering context, this almost always means a small handful of specific mechanisms — one lithographic hot spot, one particle source, one parametric spec — turn out to be responsible for the bulk of all yield loss, while a long tail of minor contributors adds up to comparatively little even though each individual one still technically costs some yield.
Building a yield loss Pareto starts with categorizing every detected yield-loss event into a specific mechanism bucket — random particle defects, a specific systematic pattern failure, a specific parametric spec miss — because the chart's entire diagnostic value depends on those categories actually being meaningful and consistently applied, not generic catch-all labels. Inspection data, electrical test bin codes, and failure analysis results all feed this categorization, and the granularity matters: lumping every systematic failure into one undifferentiated "systematic" bucket hides exactly the kind of specific, addressable mechanism that a Pareto chart is supposed to surface.
Once categorized, each mechanism's loss contribution gets quantified in a common unit — typically yield impact in percentage points, or equivalently die count — so that mechanisms affecting different process steps and detected by different tools can be directly ranked against each other on a single chart. This common-unit conversion is what makes a litho hot spot costing 3 yield points directly comparable to a particle-defect source costing 2 yield points, even though one was found by a lithography inspection tool and the other by a particle scanner — without that normalization, ranking mechanisms against each other wouldn't be meaningful at all.
The resulting bar chart, ranked largest to smallest with a cumulative percentage curve overlaid, routinely shows that two or three mechanisms account for the clear majority of total yield loss while a long tail of minor contributors adds up to comparatively little — which is exactly the pattern the 80/20 rule predicts and exactly why the chart is worth building in the first place. Seeing this pattern explicitly, rather than working from an unranked list of equally-weighted yield issues, is what lets a yield-engineering team recognize that fixing the top bar or two will move total yield far more than spreading the same engineering effort evenly across every item on the list.
A yield loss Pareto is never a one-time chart — it gets rebuilt regularly as mechanisms get fixed and the ranking shifts, because today's dominant mechanism won't stay dominant forever once it's addressed, and some previously minor item will inevitably rise to take its place at the top. This recurring rebuild is exactly what connects the Pareto chart to the concept of a [yield limiting factor](yield-limiting-factor) — the single current top-ranked item on the chart at any given time, and the specific target that an ongoing yield-improvement program should be focused on fixing next.