Among the pretreatment methods which are performed just after the lithographic process to minimize the roughness increase of 193nm photoresist during the subsequent etching processes, an in situ plasma pretreatment is the most cost effective. A HBr plasma pretreatment has proven quite effective and a few papers have described the mechanism. In an effort to understand further, the authors evaluated four plasma pretreatments using HBr, Ar, H2, or Cl2 gases and compared their results. Fourier transform infrared (FTIR) spectroscopy was performed for the investigation of the chemical changes effected by the plasma pretreatments. Cross-section scanning electron microscope (SEM) images were used to measure the photoresist film thickness, while top-down SEM images and an off-line program were used to determine linewidth roughness (LWR) changes for 70 and 80nm line features. They found two different types of roughness. The first type is a low-frequency roughness, which repeats about every 400nm and increases the LWR value substantially. The second type is a high-frequency roughness, which appears about every 100nm and causes a moderate increase in the LWR value. From the top-down SEM images, they recognize that the low-frequency roughness is caused by collapse of the 193nm photoresist during the following bottom antireflective coating and hard-mask etching processes. The no plasma and the Ar plasma pretreated samples show this low-frequency roughness and produce the worst LWR values of about 11nm at the 70nm linewidth features after ashing processes. The HBr and the H2 plasma pretreated samples, which mainly show the high-frequency roughness, result in the best LWR values of about 6nm at the 70nm linewidth features after ashing processes. The FTIR analysis shows that both the HBr and H2 plasma pretreatments reduce the CO content substantially, down to about 20%–40% of the original CO content of the 193nm photoresist as-coated film. On the other hand, the Ar plasma pretreated photoresist film still has about 60% of the CO content of the pristine 193nm photoresist. The authors conclude that the low-frequency roughness has a critical relationship with the CO content in the 193nm photoresist. They also find that the elements being incorporated into the 193nm photoresist during the plasma pretreatment are important for their impact on the LWR. Especially, the Cl2 plasma pretreatment, which eliminates about the same amount of the original CO content in the photoresist as both the HBr and H2 plasma pretreatments, deteriorates the LWR notably just after the pretreatment and produces the most severe deformation after etching processes. Of the plasma pretreatments evaluated in this work, the HBr plasma pretreatment is the best in view of both the LWR and the application. The H2 plasma pretreatment, which shows the same lowest LWR value as the HBr plasma pretreatment, reduces the photoresist thickness substantially. However, even the HBr plasma pretreatment has one critical disadvantage because it generates the high-frequency type of roughness, which is not found in the case of the inert Ar plasma pretreatment. The H and Br radicals react with the 193nm photoresist during the HBr pretreatment and appear to cause some side reactions and generate the high-frequency type of roughness during subsequent plasma processes. To minimize both the low- and high-frequency deformations simultaneously, we propose an inert gas plasma pretreatment process of which process parameters such as pressure and power are optimized to reduce the CO content of the 193nm photoresist less than 40% of the original CO content as coated.
Read full abstract
Jun 13, 2021
Erliang Liu + 2
Open Access