Yield Limiting Factor
A yield limiting factor is whichever single process step, defect mechanism, or design marginality currently sits at the top of the yield loss Pareto — the one specific bottleneck that, if fixed, would move total yield further than fixing anything else on the list right now. It's inherently a moving target: once today's limiting factor is actually addressed, the next item down the Pareto ranking automatically becomes the new limiting factor, whether or not anyone intervenes on it directly.
Pareto ranking is the most direct way a yield limiting factor gets identified in practice — it's simply whichever mechanism sits at the very top of the current [yield loss Pareto](yield-loss-pareto) chart, the single bar contributing the largest share of total yield loss at this moment. Because the Pareto chart already ranks every known mechanism by its yield impact, the limiting factor is almost definitionally just that chart's current #1 item — there's no separate identification step needed beyond keeping the Pareto chart itself current and accurate.
Kill ratio analysis offers a complementary way of finding the limiting factor, specifically useful when a defect type's sheer frequency doesn't obviously match how damaging it actually is — some defect types kill a die almost every time they occur, while others are frequent but mostly harmless. A defect type that occurs relatively rarely but kills the die nearly every single time it does occur can easily be a bigger yield limiting factor than a far more frequent defect type that only occasionally actually causes failure. Kill ratio — the fraction of occurrences of a given defect type that actually result in a failing die — captures this distinction directly, which raw defect-count frequency alone cannot.
Spatial signature analysis helps confirm a candidate limiting factor by checking whether its failures cluster in the specific location pattern that mechanism should produce — a repeating edge-of-wafer pattern pointing toward an etch or CMP uniformity issue, a specific reticle-field location pointing toward a lithographic hot spot. This spatial confirmation step matters because two different mechanisms can sometimes produce superficially similar yield-loss magnitudes in aggregate numbers while showing completely different spatial signatures — and only the spatial pattern reliably distinguishes which mechanism is actually the one driving the loss, pointing engineering effort toward the genuinely correct fix.
Because fixing today's limiting factor automatically promotes whatever was previously the second-biggest contributor into the new limiting-factor position, yield improvement is inherently an iterative, moving-target process rather than a project with one single fix and a final finish line. A yield-engineering team that successfully eliminates the current top Pareto item hasn't finished the job — a different mechanism, previously overshadowed by the one just fixed, immediately becomes the new constraint on further yield improvement. This is exactly why effective yield engineering treats limiting-factor identification as a continuously repeated step, re-running the Pareto ranking after every fix, rather than as a one-time analysis performed only at the start of a yield-improvement effort.