Etch Microtrenching Budget

# Etch Microtrenching Budget

An etch microtrenching budget defines the allowed local recess near a feature's sidewall feet and the evidence needed to verify it. It keeps the deepest local feature visible rather than relying on center depth or an average bottom profile.

The budget is specific to the structure and integration. It is not a universal recess-depth limit or a recipe setting that guarantees a flat floor.

## Identify the local geometry

Microtrenching describes locally enhanced removal near the sidewall feet relative to the central floor. A feature can reach its intended center depth while the corners extend farther into the underlying material. Local recesses can be unequal on the two sides.

The schematic is a geometric description, not a quantitative process model. It does not imply that all bottom defects are microtrenches. Distinguish a vertical local recess from lateral notching, footing, roughness or incomplete clearing using defined profile descriptors.

## Define the depth reference and metric

State the top reference surface and how the central floor and local minimum surface are identified. In this guide, recess depth is local corner depth minus central-floor depth, with depth positive into the material.

Define the lateral search region, image resolution and treatment of roughness. If the central floor slopes or curves, specify the reference construction instead of choosing whichever point gives a favorable result. Keep left and right values where asymmetry matters.

The invented example uses a common top reference. The center is 800 nm deep, the left local depth is 824 nm and the right is 818 nm. Their recesses are 24 nm and 18 nm. The maximum observed recess in this section is 24 nm.

With an illustrative 20 nm upper limit, the left recess exceeds the limit by 4 nm. Averaging the two sides would discard the local maximum and would not satisfy a requirement defined for each local recess. The numbers are teaching values, not a qualified specification.

## Connect recess to the actual stack

A local recess can consume a protected-layer thickness allowance or change the geometry needed by a subsequent step. Evaluate the material and interface actually reached by the feature rather than assuming that a differential depth alone establishes integrity.

A central-floor depth and a local recess can share a measurement reference, but they answer different questions. Include actual remaining layer thickness, profile and relevant functional evidence where the integration requires them.

No universal recess fraction or absolute depth establishes acceptability across different materials, stop layers and devices. Derive the limit from the intended integration and its supporting evidence.

## Investigate mechanism with controlled evidence

Primary studies have demonstrated microtrenching associated with reflection of energetic particles from sidewalls in defined plasma etch systems. Sidewall geometry, particle transport and local surface conditions can affect where enhanced removal occurs.

Those results do not establish a unique mechanism for every observed recess. Charging, passivation, mask shape and other process variables may also need investigation in the actual structure. A cross-section alone cannot isolate the cause.

Use controlled comparisons when evaluating recipe changes. A change that reduces recess can also change rate, selectivity, critical dimension or clearing. Qualify those tradeoffs together rather than treating lower recess as the only objective.

## Measure representative features

Sample relevant widths, depths, pattern densities and wafer locations. Cross-section placement and orientation matter, especially for three-dimensional openings. One section may not contain the deepest local point around the full perimeter.

Record preparation artifacts, calibration, resolution and uncertainty. If an observed recess approaches a limit, apply the defined measurement decision rule. A smooth-looking image or a small sample does not rule out deeper local features elsewhere.

State whether the result is as-etched or after strip and cleaning. Subsequent processing can alter the surface and the remaining protected-layer allowance.

## Close the budget with local acceptance evidence

Retain recipe revision, incoming stack, mask geometry, chamber condition and measurement locations with the results. Specify whether acceptance governs each feature, a sampled maximum or another justified statistic.

If a local limit is exceeded, preserve the full profile and material-integrity evidence. Do not hide the excursion by reporting only a mean depth or by changing the reference after seeing the result.

A useful microtrenching budget measures the local recess and checks the material it reaches. Center depth supports the analysis; it cannot replace the local requirement.

## Sources and scope

• [Hoekstra et al. — Microtrenching resulting from specular reflection during chlorine etching of silicon](https://cpseg.eecs.umich.edu/pub/articles/jvstb_16_2102_1998.pdf): primary modeling and experimental comparison for localized etching near sidewall bases.
• [Dai et al. — Study on mechanism of etching in low pressure radio-frequency plasmas](https://impact.ornl.gov/en/publications/study-on-mechanism-of-etching-in-low-pressure-radio-frequency-pla/): primary profile-evolution simulation including particle reflection and charging considerations.

Original diagrams and numeric teaching example. No universal mechanism, recess limit or operating recipe is implied.