We add optical emission or laser interferometry endpoint detection to older plasma etch tools, so etches stop on the film rather than on a timer.
Many older RIE and ICP etchers in research cleanrooms run every step on a timer. That works when film thickness and etch rate are tightly controlled. In a research lab they rarely are: films come from different deposition tools, wafers vary, and the chamber drifts as it is used for different processes. A timer that is safe on one wafer over-etches the next, damaging the layer underneath, or under-etches and leaves residue.
Endpoint detection lets the etch stop when the film clears. It saves test wafers, protects underlying layers and makes results easier to repeat between users. On older tools that never had it, it is one of the most useful upgrades for process control.
Optical emission spectroscopy (OES) watches the light emitted by the plasma. When a film clears, the chemistry changes: reaction products from the film drop and etchant species rise. A spectrometer or a filtered detector at a chosen wavelength picks up that change. OES needs only a view of the plasma and works across the whole wafer, but the signal shrinks as open area gets small.
Laser interferometry shines a laser on the wafer and follows the reflected intensity as the film thins. Each oscillation corresponds to a known thickness change, so it can stop on a remaining thickness as well as at clear. It needs a view of the wafer, ideally from above, and a suitable spot or test structure. It suits small open areas, III-V stacks and transparent films where emission barely changes.
Some tools benefit from both. We recommend a method after the process review, and say clearly where a method is unlikely to give a clean signal.
A typical upgrade covers:
On deep silicon etchers using the Bosch process, endpoint is usually used to detect the stop layer in through-wafer or SOI etches. See the Bosch process explained for how the cycling affects the signal.
Many older controllers were never built to end a recipe step on an external signal. In that case there are two options. The endpoint trace can be shown on an operator display, and the operator stops the step when the signal turns; this works for low-volume research use but depends on the person at the tool. Or the input can be added as part of a control system retrofit, so the recipe stops by itself. The process review states which applies to your tool.
Installation is usually a short planned stop, longer if the chamber needs a new viewport. Developing endpoint settings takes tool time spread over test etches, which can often be fitted around normal use.
You receive the installed sensor, endpoint settings for each agreed step, test etch comparisons and operating notes. The upgrade suits single-wafer etch tools such as the Oxford Plasmalab range and STS deep silicon etchers.
We review your films, stack, open area and the etch steps where a timer causes trouble.
We propose optical emission, laser interferometry or both, and check viewport access on the chamber.
Our engineers fit the sensor and connect its signal to the tool so a recipe step can end on endpoint.
We run test etches, set the endpoint algorithm and compare results with timed etches.
200 mm (8 in) · Late 1990s–2010s
150 mm (6 in) · Mid-1990s–2000s
Optical emission works well when the plasma chemistry changes as the film clears and the open area is not too small. Laser interferometry follows the film thickness directly at one spot and suits small open areas and stacks where emission barely changes.
Optical emission becomes harder as open area drops, because the signal change gets small. Laser interferometry on a test structure, or careful wavelength choice and signal processing, is often needed in that case.
Often, when the tool controller accepts an external endpoint input. On older controllers without that input, the signal can be shown to the operator, or added as part of a control system retrofit.
Optical emission needs a clear view of the plasma, and laser interferometry needs a view of the wafer, ideally close to normal incidence. Many chambers have a usable port; some need a new window or adapter.
How the Bosch process alternates etch and passivation to cut deep, vertical features in silicon, what limits it, and what to check on a used DRIE tool.
A parallel-plate RIE covers most teaching and dielectric work. An ICP-RIE opens up deep, anisotropic and III-V etching. How to decide which to buy used.
Tell us the tool and what you need. We scope the work and send a written offer.