Plasma etching may be the most important process in semiconductor manufacturing, with three stages of development:

1. Determine what etchant, gas, auxiliary layer, etc. are needed to perform etching;

2. Demonstrate the performance of completely removing thin films within the specification range and achieving uniformity on a single wafer.

3. Determine how to repeat the process on thousands of chips in HVM in a high yield and small drift manner.

Usually, skilled etching and integration engineers handle the first two stages of development. The third stage will once again utilize engineering expertise - machine learning and data analysis, which can access a large amount of data and understand one million interacting things. The third stage used thousands of wafers - at least an order of magnitude larger than the wafers used in the first and second stages. Obtain concept validation from the nominal process flow and layout, and develop one or more work devices on the wafer, then transfer the POC to the product development team at the wafer factory to expand the process and increase production. This may create a huge gap with profitability. Process window modeling narrows this gap by introducing changes in the wafer factory into the early stages of R&D pathfinding.

In the etching process, there are multiple parameters that affect the etching rate, profile, and selectivity. A key factor is temperature. When controlling the etching rate, selectivity, and etching profile, the influence of thermal effects in the etching process can be seen. The etching process depends on the surface temperature of the chip, which depends on several heat fluxes, including heat conduction, ion impact energy, surface reaction, and ionizing radiation heat flux. Therefore, the plasma model needs to combine all these physical characteristics to accurately describe the temperature changes on the chip surface. Process simulation software can model a series of etching attributes, which can achieve better etching results faster and accelerate customers' ability to increase or optimize production.

As design rules shrink, many etching processes have shifted towards very fast plasma etching process steps that require highly precise control of all reaction inputs: power, pressure, chemistry, and temperature. There is also a trend towards optimizing plasma pulse behavior, which is to generate specific ion to neutral ratios and then eliminate by-products. Advanced modeling of this situation is crucial for further expanding equipment scale.

For some time, manufacturers of etching systems have been using modeling software to accelerate the development or ramp production of the next node. When developing the next node technology, there is simply not enough time or enough wafer to perform all possible process experiments. The number of combinations of etching equipment settings can reach millions or even billions, making it impossible to develop powerful wafers using all process possibilities. Of course, all good models are validated on actual chips. An accurate model should be predictive and solve targeted problems that users want to solve. Tool suppliers are also developing advanced etching processes to integrate production lines more tightly and transform the previous two mask level processes (two photolithography steps) into one process, thereby simplifying the process and reducing costs.

The most critical etching steps in the etching process include pseudo gate etching, anisotropic column etching, isotropic spacer etching, and channel release steps. The contour etching through alternating layers of silicon and SiGe is anisotropic and uses fluorination chemistry. The internal spacer etching (indentation) and channel release steps are optimized to remove SiGe with extremely low silicon loss.

The culprit in etching is fluoride gas with high global warming potential (GWP). Chip manufacturers and material suppliers are seeking alternative chemicals to reduce carbon emissions. The reason why PFAS has problems is because its molecules are very stable, and the light or chemical reactions in the atmosphere are not sufficient to decompose it. Many alternative gas mixtures with higher oxygen content are more prone to dissociation and have lower global warming potential. But sustainability is not a particular etching or deposition challenge. From lithography to packaging, this is a comprehensive industry challenge, and the impact of new materials can affect the entire device processing.

This article is reprinted from the internet. If there is any infringement, please contact to delete it. Thank you!