Etch Overetch Budget

# Etch Overetch Budget

An etch overetch budget compares the additional removal needed to clear required features with the damage or material loss that the rest of the stack can tolerate. It helps determine whether an additional etch interval has a defensible process window.

Overetch commonly follows a main-etch endpoint or a timed clearing estimate. State the time reference and the recipe segment explicitly. A percentage of main-etch time is only meaningful when its denominator and process conditions are defined.

## Clear the target and protect the stack

An endpoint signal can represent a transition in the monitored region or an aggregate signal. It does not automatically establish that every relevant opening has cleared. Local thickness, feature geometry and pattern effects can leave residual target material elsewhere.

At locations that cleared earlier, the underlying layer may already be exposed. Those locations can accumulate loss while remaining target material is removed from slower locations. A stop layer is not assumed to have zero etch rate.

## Define both sides of the budget

For clearing, identify the largest relevant residual thickness and the applicable removal rate during the actual overetch segment. For protection, identify the remaining loss allowance at vulnerable locations and their applicable underlayer etch rate. Include loss that occurred before the chosen time reference when calculating the remaining allowance.

Mask survival, critical dimension, sidewall shape, surface condition and electrical damage can impose additional limits. An interval that meets a thickness budget alone may still be unacceptable.

## Work through a simple example

Use a deliberately simplified model. At the start of overetch, the slow location has 24 nm of target film remaining. The assumed target removal rate is 3 nm/s. It therefore needs at least 24/3 = 8 s to clear.

At a different, already-clear location, the protected layer has 6 nm of remaining allowable loss. Its assumed removal rate is 0.5 nm/s throughout this segment. It can tolerate at most 6/0.5 = 12 s of additional exposure.

Under these assumptions, the thickness-only overlap is 8-12 s. At 10 s, the modeled target-removal capacity is 30 nm and the underlayer loss at the already-clear location is 5 nm. That calculation leaves 1 nm of the stated allowance before any additional uncertainty or other constraint is considered.

The example uses constant rates and excludes transition delays, uncertainty, mask loss and profile limits. It does not recommend a 10 s recipe. The two locations need not have identical exposure histories, and the applicable rates must come from relevant patterned-process evidence.

## Carry variation without hiding the assumptions

A conservative clearing estimate can use a larger residual and a smaller target rate. A conservative protection estimate can use a smaller remaining allowance and a larger underlayer rate. State whether those combinations are valid bounds and whether the quantities are correlated.

If the minimum clearing time exceeds the maximum permitted exposure, increasing the time cannot solve both requirements in that model. The stack, incoming variation, mask, chemistry, selectivity or process sequence needs investigation and qualification.

For changing conditions, total loss depends on removal rate over the actual exposure history. A single blanket-film selectivity number does not establish local patterned-feature loss through a recipe transition.

## Link endpoint behavior to physical results

Record endpoint method, monitored area or signal, trigger definition, delay and fallback behavior. Optical emission and laser-based endpoint methods have different sensitivities and coverage. Establish how the chosen signal relates to the structures that must clear.

Measure residuals and protected-layer loss on representative structures and locations. Keep the exact process state visible: as-etched, after strip or after the specified clean. Subsequent cleaning loss belongs in the overall integration budget even when it is tracked separately from plasma overetch.

## Accept with complete evidence

Retain recipe revision, chamber condition, incoming stack, layout, measurement method and acceptance rule. Verify clearing together with mask condition, profile, underlayer integrity and relevant functional results. Revisit the budget when any of those inputs changes.

A useful overetch interval clears the required target while preserving every applicable limit. The interval is supported by measured process behavior, not by a universal percentage or a nominal endpoint alone.

## Sources and scope

• [Oxford Instruments — Automatic endpoint detectors](https://plasma.oxinst.com/service-support/upgrades/endpoint-detectors): endpoint methods and their process-specific strengths.
• [Lam Research — Tech Brief: All About ALE](https://newsroom.lamresearch.com/Tech-Brief-All-About-ALE): selectivity, underlying-material protection and plasma-induced damage challenges.

Original explanatory model and diagrams. Numeric values are teaching assumptions, not measured capability or a qualified operating recipe.