Etch Notching Budget

# Etch Notching Budget

An etch notching budget defines the allowed local lateral removal at a feature base or interface and the resulting dimensions that must remain. It links the notch measurement to the material, structure and functional requirement that the etch must preserve.

Notching is a geometric observation. Its appearance alone does not prove a charging mechanism or establish whether the affected device is acceptable.

## Distinguish lateral notching from a floor recess

The illustrated opening widens locally near its base as material is removed sideways beneath the upper sidewall reference. That is the lateral dimension tracked here. Microtrenching instead describes locally enhanced depth near the sidewall feet relative to the central floor. Both may need measurement, and they should not share an undefined depth metric.

State the material interface, upper-wall reference, notch search region and the direction of the measurement. Some literature calls the lateral penetration notch depth; report explicitly that the dimension is lateral to avoid confusing it with vertical etch depth.

The schematic does not show every possible notch shape. Real profiles can be asymmetric, rounded or variable around the perimeter of a three-dimensional opening.

## Track what remains, not only what was removed

The second diagram is a separate width-accounting example for a line, not the same opening as the first diagram. Its reference width is 100 nm. Lateral losses of 12 nm on one side and 18 nm on the other leave 100 - 12 - 18 = 70 nm at the affected section.

With an illustrative 75 nm minimum remaining width, that section is 5 nm too narrow. The example assumes the losses apply to the same cross-section and reference width. It does not recommend either a line width or an allowed notch size.

A single-sided limit and a minimum remaining-width limit are different constraints. Check both when they apply. Averaging the side losses or measuring only the upper line width can hide a local minimum.

## Keep the interface and exposure history visible

Notching can become important when a process reaches an interface and continues through a clearing or overetch segment. Record the incoming stack, actual exposed material and relevant recipe transitions rather than assigning all lateral removal to one nominal step.

Local charging of insulating surfaces has been studied as a contributor to notching in specified polysilicon-on-insulator etch systems. Particle trajectories, scattering, surface reactions, mask geometry and electrical connectivity can interact. Those studies do not establish that every lateral recess has the same cause.

Changes intended to reduce a notch can also affect clearing, selectivity, profile and electrical stress. Evaluate the complete integration rather than assuming that a smaller geometric notch guarantees less electrical damage.

## Define the acceptance rule

Specify whether the requirement applies to maximum lateral penetration, remaining line width, notch height, interface integrity or another defined metric. A structural dimension and an electrical reliability requirement should remain separate unless a validated relationship connects them.

Derive limits from the intended device and the subsequent process steps. There is no universal acceptable notch size across different films, dimensions and functions.

Carry measurement uncertainty and incoming dimensional variation into the decision rule. A nominal subtraction does not establish the distribution of remaining width or the worst three-dimensional location.

## Measure representative local sections

Include relevant feature widths, pattern densities, electrical connections and wafer locations. Identify where the section crosses the feature and which process state is being measured: as-etched, after strip or after cleaning.

Cross-section preparation, image calibration and edge identification can affect a notch measurement. A single section may miss the greatest lateral loss around an opening or along a line. State the sampling scope before treating a measured maximum as a broader bound.

Use suitable electrical or functional evidence where integrity is part of the requirement. A good-looking image does not establish every device property, and a failed electrical test does not uniquely identify notching as the cause.

## Close the budget with traceable evidence

Retain the reference geometry, local measurements, recipe revision, chamber condition and accepted scope. Preserve asymmetric and out-of-limit results rather than replacing them with a favorable mean.

When evaluating a process change, compare appropriate controls and verify all affected requirements. Keep the measured observation separate from the proposed mechanism until the evidence supports the causal explanation.

A useful notching budget checks the lateral loss and the material that must remain. The local requirement is established by the integration, not by a nominal center depth or an upper-width measurement.

## Sources and scope

• [Hwang and Giapis — On the origin of the notching effect during etching in uniform high density plasmas](https://authors.library.caltech.edu/records/ac5xn-zqb88): primary simulation and comparison with measured notch behavior in a defined integration.
• [Hwang and Giapis — Charging damage during residual metal overetching](https://authors.library.caltech.edu/records/8nw07-a4z30): primary simulation illustrating tradeoffs between charging damage and notch formation.

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