etch loading budget

# Etch Loading Budget

An etch loading budget accounts for differences in clearing time caused by the patterned wafer, then checks whether the extra exposure is acceptable. It connects layout-dependent removal to mask survival and protected-layer loss.

Microloading already describes local pattern-dependent etching. This entry focuses on the practical time and thickness budget rather than introducing a second definition of that effect.

## Compare like features

Local exposed area can change the balance between reactive-species supply and consumption. Under some conditions a dense region removes more slowly than a sparse region. The rate difference and even its direction depend on chemistry, transport and surface behavior; dense does not always mean slower.

Etch Loading Budgetequal-width openings can clear at different timesSparse region100 nm/minDense region80 nm/minSame duration · same opening width · different local densityIllustrative slower-dense case · rate ordering and magnitude must be measured.

The drawing holds opening width constant and changes local density. Depth and rates are illustrative, not measured results. To test density dependence, compare matching feature dimensions, target film and incoming surface state. Otherwise a difference attributed to loading may instead reflect feature geometry or film variation.

Aspect-ratio-dependent etching concerns the effect of feature depth and width on removal. It can coexist with local loading but should not be treated as an identical mechanism. Wafer-scale exposed area and chamber-scale supply also belong to a broader loading assessment.

## Convert rates into clearing times

Use a deliberately simple constant-rate example: the target is 100 nm thick, the fast region removes at 100 nm/min and the slow region at 80 nm/min. Their nominal clearing times are 100/100 = 1.00 min and 100/80 = 1.25 min.

Clear the Slowest Featurethe faster-clearing landing sees extra exposureFast regionTarget removalSlow regionTarget removal1.00 min1.25 minExtra landing exposure = 0.25 min in the fast region100 nm target at constant rates · no extra endpoint or timing margin included.

Ending at 1.25 min clears both regions in this model. The fast region's landing layer has already been exposed for 0.25 min. If that layer removes at 2 nm/min after exposure, it loses about 0.5 nm during this interval. That is a calculated example with constant local rates; it is not a qualified process limit.

The 20% lower removal rate produces a 25% longer clearing time because time is inversely proportional to rate. Distinguish a rate percentage from a time percentage when specifying an added clearing margin.

## Keep three budgets separate

  • Clearing: does the slowest relevant location remove all the intended target?
  • Mask: does the mask survive the total sequence, including any added time?
  • Landing: does the earliest-clearing location retain enough of its protected material?

For the simplified model, estimate clearing time as target thickness divided by local target rate. Estimate landing loss as the local protected-material rate multiplied by the time from local clearing until etch end. A later stop-layer opening or clean needs its own allowance. If a recipe changes conditions, sum the losses across its actual exposure intervals.

## Qualify across the layout

Use a test matrix that varies density while holding feature width fixed, then vary width separately. Sample array centers, array edges, isolated features and relevant wafer positions. Measure profile and residual target as well as depth. Include the slowest clearing and most exposed landing locations in the same assessment.

Do not assume a global endpoint signal resolves every local feature. Dummy patterns or layout changes may help balance loading, but their placement must be checked against design rules, electrical constraints and the process. Simply extending the etch may clear a slow region while increasing mask loss or landing damage elsewhere.

## Engineering takeaway

A loading budget succeeds when the last feature clears before the most exposed mask or landing exceeds its loss allowance. Validate that window on representative patterns, rather than relying on a single blanket-film rate.

## Sources and scope

  • [Lam Research — Using Dummy Patterning to Solve Etch Uniformity Problems](https://newsroom.lamresearch.com/dummy-patterning-etch-uniformity?blog=true): pattern-dependent removal mechanisms and layout compensation.
  • [VTT — Deep Silicon Etching](https://sarjaweb.vtt.fi/pdf/publications/2005/P559.pdf): local and wafer-scale pattern effects.

Original illustrations and calculated examples; the diagrams show one assumed rate ordering and are not universal process predictions.

Go deeper with CFSGPT

Get AI-powered deep-dives, save terms, and run advanced simulations — free account.

Create Free Account