The notch occurrence factor in polysilicon etching was investigated using a newly designed test mask pattern. This pattern varied the space width, the line width of the line-and-space structure, and the pad perimeter which connected the line-and-space structure. By using this test pattern, it was found that the notch depth increases as the “perimeter ratio”, (i.e. the ratio of the pad perimeter to the notch line perimeter), increases. Moreover, when the perimeter ratio was very small, the notch depth was quite small. Therefore, it is considered that the notch is caused by electron supply from the periphery of the pad which collects the electrons from plasma. As a result of this electron conduction, in the case of the non-connected lines, the notch occurs only outside the line of the line-and-space structure, and in the case of the connected lines, the notch occurs at all of the connected lines. The notch depth difference between the non-connected lines and connected lines is explained by the difference in perimeter ratio.

Characteristics of the surface-wave plasma generated by a radial-line slot antenna (RLSA) have been studied by both direct plasma probe measurements and numerical simulations. Some unique characteristics have been found, including excellent critical radial plasma uniformity, low electron temperature under various pressure conditions, the main plasma generation area of RLSA being limited in the plasma surface, and few high-energy electrons existing in the wafer region. Numerical simulations are implemented to reveal the more essential difference in plasma generation between the RLSA and the other plasma sources, where the superiority of RLSA plasma has been confirmed. The features of high plasma uniformity and low electron temperature lead to free plasma damage in our associated etching process.

Plasma is extensively used for the etching/ashing of low-dielectric (low-k) films. However, since low-k films, such as SiOC films, are vulnerable to plasma irradiation, they are severely damaged during plasma processes, such as the extraction of methyl groups from low-k films. As a result, plasma irradiation increases the dielectric constant of low-k films and reduces the reliability of Cu/low-k interconnects. In previous work, the authors achieved highly selective and low-damage etching processes for low-k films by using their developed neutral beam process instead of the conventional plasma process. They have now investigated the damage mechanism in low-k films (porous SiOC films) during plasma processes by clarifying the effects of ions, radicals, and photons in plasma. First, they compared the damage in SiOC films etched by the conventional plasma process and the neutral beam process. Their results show that plasma processes change the structure of the SiOC film deeply within the film (over 100nm in depth) and increase the film’s dielectric constant, whereas the neutral beam process maintains the structure of the SiOC film. Additionally, they found that when a combination of the neutral beam process and a 172nm excimer lamp is used, photon irradiation enhances the extraction of methyl groups from the SiOC film by breaking Si–C bonds in the film. These results show that photon irradiation plays a very important role in the damage mechanism in low-k films during plasma processes.

This work demonstrated a process for the atomic-scale etching of SiO2 films, consisting of alternating nanometer-thick fluorocarbon film deposition with O2 plasma irradiation in a capacitively coupled plasma reactor. Ar plasma etching after fluorocarbon film deposition tends to suffer from nanometer- or subnanometer-thick carbon films deposited on the SiO2 surface and chamber walls. These carbon films cause various problems, such as reductions in the etching rate per cycle and degradation of the SiO2 quality. In contrast, in our two-step process, O2 plasma removes carbon atoms in such fluorocarbon films. This process therefore allows the atomic scale etching of SiO2 films without any residue or surface contamination. Additionally, since the etching rate per cycle plateaus as both the etching time and deposition time are extended, it is unnecessary to uniformly deposit a fluorocarbon film over the wafer.

No abstract