Radial CDO Non-Uniformity Causing Wafer Edge Speed Loss

Showerhead radial compensation

Your idea about different hole sizes is very relevant.

The showerhead is not only a gas distributor. In PECVD, it also influences plasma structure, local precursor delivery, oxidant delivery, residence time, and sometimes RF coupling. Patents and PECVD studies show that showerhead hole pattern, hole density, radial zones, and edge-flow design are used exactly to improve center-to-edge film uniformity and film properties

Best concept:

Multi-zone showerhead: center zone and edge zone with independently controlled flows.

For example:

1. Center zone: normal CDO precursor + oxidant ratio
2. Edge zone: slightly higher carbon precursor / lower oxidant ratio
3. Outer ring purge: inert gas curtain to control edge plasma and residence time

This is better than changing only the global precursor ratio.

Larger showerhead / extended edge coverage

Increasing showerhead diameter can help if the current showerhead creates edge starvation or edge boundary distortion.

If the gas distribution ends too close to the wafer edge, the edge region behaves like a boundary zone rather than a normal deposition zone. A larger showerhead, or an edge-compensated faceplate, can make the wafer edge “feel” more like the wafer center.

Concept:

Do not allow the wafer edge to see the end of the gas-distribution system. Move the gas-distribution boundary outside the wafer.

This can reduce radial discontinuity in precursor flux, oxidant flux, and plasma density.

Edge gas ring/compensation ring

A very practical concept is an independent edge gas ring around the showerhead or chamber wall.

It could supply:

1. inert gas to dilute edge plasma,
2. carbon-containing precursor to compensate carbon depletion,
3. lower oxidant concentration near the edge,
4. purge gas to reduce edge residence time.

This gives a powerful control knob:

Use the edge ring to tune only the last 10–20 mm of the wafer radius.

This is useful because the edge problem is often localized.

Radial temperature compensation

If the edge is hotter, the CDO may become denser, more oxide-like, less porous, and therefore higher-k.

Possible solutions:

1. improve ESC thermal uniformity,
2. tune backside He cooling zones,
3. reduce edge-ring heating,
4. modify edge ring material/emissivity,
5. add multi-zone chuck control,
6. reduce plasma heating near the edge.

For CDO, temperature is important because it can affect carbon incorporation, densification, porogen behavior, moisture sensitivity, and final k.

Strong idea:

If edge k is higher because edge film is denser, compensate by lowering edge temperature or reducing edge plasma power density.

RF/plasma radial compensation

The edge often sees a different plasma sheath and radical flux. If the edge plasma is more aggressive, it may increase oxidation, remove carbon-containing groups, densify the film, or damage porous CDO.

Possible directions:

1. tune RF frequency/power split,
2. change electrode spacing,
3. redesign edge ring,
4. add plasma confinement ring,
5. adjust showerhead-to-wafer gap,
6. use pulsed plasma,
7. reduce ion bombardment during the carbon-sensitive part of deposition.

A very good process concept:

Separate deposition and densification.

First deposit carbon-rich low-k film gently. Then apply controlled UV / plasma treatment more uniformly.

This reduces the chance that the edge is over-treated during deposition.

Chemistry ratio compensation: carbon/oxygen balance

If edge k is high, one likely mechanism is:

Edge receives relatively more oxidizing environment → lower carbon content → denser Si–O network → higher k

So compensation could be:

1. lower O₂ / N₂O / CO₂ fraction globally or locally,
2. increase organosilicon precursor at edge,
3. increase porogen or carbon-rich precursor at edge,
4. reduce oxidant dissociation near edge,
5. use a less oxygen-aggressive chemistry.

But global change may damage center performance. Therefore:

Local chemistry compensation is better than global chemistry compensation.

Pressure, gap, and pumping compensation

Edge k can also come from different residence time and exhaust behavior.

Try:

1. change chamber pressure,
2. change showerhead-to-wafer spacing,
3. tune throttle valve/pumping symmetry,
4. add flow straighteners,
5. use symmetric exhaust or compensated edge exhaust,
6. reduce lateral gas flow along the wafer surface.

A useful principle:

If the edge composition is different, check whether the edge gas composition is different before it reaches the surface.

Post-treatment compensation

For porous CDO, UV cure or plasma cure can also create radial k variation.

If edge k increases after cure, not during deposition, then the compensation should be in the cure:

1. radial UV intensity correction,
2. wafer rotation during cure,
3. edge masking / reduced edge exposure,
4. temperature compensation during cure,
5. optimized cure time to avoid edge over-densification.

So first we need to know:

Is edge k already high after deposition, or only after cure / integration?

This is critical.