semiAIfoundry

Fabrication / Repeated operations · Atlas v1.3.0

Plasma and wet etch: selectivity and profile

Removes selected material while preserving the feature and surrounding stack.

Explore in AtlasOptical field to nanometer pattern; material states distinguished

Learning objectiveSeparate etch selectivity, directionality, endpoint and damage.

The mechanism

Etching converts temporary masking geometry into a material feature. Wet etching relies on reactions in a liquid and can depend strongly on crystal orientation and chemistry. Plasma etching combines reactive species with energetic particle effects that can help control directionality. Neither is simply “cutting downward”: sidewall reactions, transport, charging and mask erosion influence the result. Selectivity compares removal of the target material with loss of another exposed material. A successful transfer must satisfy depth, profile, roughness, residue and damage requirements together. This explains why an acceptable resist image does not automatically become an acceptable transistor or via.

  1. Identify the exposed target, mask and stop/protected materials before choosing the route.
  2. Deliver reactive species and, for a plasma route, control the relevant energetic interactions.
  3. Remove reaction products while transport changes along a deepening feature.
  4. Monitor endpoint or timing while allowing for wafer and pattern-density variation.
  5. Measure depth, sidewall profile, residues and damage after transfer and cleaning.

Science and assumptions

Mask-consumption estimate

Δhmask≈htargetS\Delta h_{\mathrm{mask}}\approx\frac{h_{\mathrm{target}}}{S}

Mask thickness consumed is approximately target etched depth divided by target-to-mask selectivity.

Symbols and units
Δh_mask
Mask thickness consumed

nm

h_target
Target material etched depth

nm

S
Target-to-mask etch-rate selectivity

dimensionless rate ratio

htargeth_{\mathrm{target}} is etched target depth and S target-to-mask selectivity; lengths use matching units.

Assumptions Constant-rate approximation without loading, faceting, endpoint overetch or spatial variation; use it only as an initial material-budget check.

Engineering example

Synthetic budget: etching 200 nm with target-to-mask selectivity 10 consumes about 20 nm of mask in the simple model. A 20 nm starting mask has no allowance for nonuniformity or overetch.

Distinctions and tradeoffs

High selectivity is not the same as vertical sidewalls or low defectivity.

More directional or faster removal can trade against selectivity, damage and uniformity; the best compromise depends on the complete stack.

Fin shapes, contacts, vias and TSVs rely on accurate material removal that leaves neighboring structures electrically and mechanically usable.

Specifications and interfaces

Selectivity
Target removal rate / protected-material removal rate

Dimensionless ratio measured under the same process conditions.

Profile
Depth, sidewall angle and bottom shape

Name the material boundary and measurement stage.

Interfaces

Patterned mask and stack Input
Exposed materials, feature geometry and initial surface state
Transferred structure Output
Etched profile, residues and possible surface damage

Check your understanding

A process preserves the mask well but leaves tapered holes. Which property is still inadequate?

  1. Atomic number of silicon
  2. Only the test software
  3. Profile control, despite acceptable mask selectivity
Answer and reasoning

Profile control, despite acceptable mask selectivity

Selectivity compares material removal; it does not uniquely determine the spatial removal profile.

Evidence and scope

No plasma settings or wet-chemical composition is given. Schematic etch routes are not production recipes or proof of a defect cause.

Use the manufacturing investigation to trace compatible inputs, assembly and operating constraints; no current standalone calculator models this lesson’s full mechanism.

Sources

Published lesson data · Atlas releases and licensing

Continue exploring

Foundations

Related lessons