Disturbing Revelations From TSMC's EUV Hydrogen Management Patent

TSMC's understanding of hydrogen blistering in EUV systems should raise concerns

A recent patent filed by Taiwan Semiconductor Manufacturing Company Ltd. (TSMC) disclosed an apparatus for reducing hydrogen permeation of a mask when using extreme ultraviolet (EUV) radiation. The application for the patent was started in 2021, when TSMC was beginning 4nm risk production, and was continued in 2023, with the filing of what is now US Patent #12025922. The patent’s description of the problem being addressed raises concerns for different reasons.

“To suppress particles or contaminant from accumulating on the reticle or the reticle holder, a gas flow, such as a hydrogen gas flow, is provided to the reticle holder. During an extreme ultraviolet (EUV) lithography process, however, the hydrogen flow drifts towards a surface of the reticle due to its light molecular weight. The drifted hydrogen accumulates/deposits at an edge of the reticle, permeating into the space between the reticle and a covering film. This hydrogen permeation (also called a “blister problem”) onto the reticle results in bubbles at the edge of the reticle and causes the covering film to peel off. Undesirable particles generated by the peeled-off film can interfere with the further processing steps. Therefore, it is beneficial to prevent the undesirable particles caused by hydrogen permeation as a part of the lithographic process.”

Firstly, there is the issue of the bubble or blister formation at the edge of the reticle, i.e., the EUV mask, which causes the covering film to peel off. The hydrogen blistering issue in EUV multilayers is not new, having been mentioned over a decade ago and also observed by Samsung in 2017.

What is new in this patent’s description is the indication that it starts toward the edge, implying hydrogen entry from the edge. This led to the attribution to the flow of hydrogen toward the edge.

Secondly, there is the reference to the lightness of hydrogen. This is used to explain why the hydrogen drifts upward as it flows across the EUV mask. However, this assumes there is air in the system:

“During the EUV lithography process, the hydrogen flow 930 ejected by the Y-nozzle 920 along the y-axis drifts towards a surface of the reticle due to its light molecular weight. Because the hydrogen is lighter than air, it rises in a vertical direction relative to the air.“

It is well-known that there cannot be air in an EUV lithography system, as it would absorb too much of the EUV light.

The patent’s proposed solution is to flow another, heavier gas above the hydrogen flow, so as to prevent the upward drift. Such a gas, however, would also contribute excess EUV absorption.

Some other details, which could draw questions, come from the figures. Figure 2 shows an EUV lithography system that is missing the EUV mask placement as well as 5 instead of 6 mirrors. The figures showing the EUV mask do not show the pellicle; could this be an indication that a pellicle is not being used?

In short, this patent serves to remind us that the hydrogen permeation issue in EUV masks still exists, but it also presents a bigger concern that a correct fundamental understanding of the EUV system and its workings is not generally present at TSMC.