Influence of photoresist thinning and underlayer film on e-beam using eP5 for High-NA patterning

Abstract

The introduction of Extreme Ultraviolet Lithography (EUVL) has been an extraordinary achievement in keeping Moore's law alive, as it minimizes the cost, time, and number of patterning steps required for device fabrication. However, as the advanced technology nodes require further scale down of the semiconductor devices, the current state-of-art EUVL tool, equipped with 0.33 Numerical Aperture (NA) and 13.5nm wavelength cannot print the features of interest with an acceptable yield. As a solution, a new generation of EUVL tool has been proposed (High-NA EUVL), having 0.55 NA and 8nm resolution.

As a result of the increased NA, the depth of focus is reduced, thus thinner resist is required. Moreover, we need to restrict the aspect ratio of the features to avoid pattern collapse. For instance, resist thickness of approximately 30nm, which is typically used to print a 16nm half-pitch feature, cannot be applied to print an 8nm half-pitch feature, resist thickness must be lowered significantly to circumvent pattern collapse. However, the use of thin resist may have several negative effects on metrology and defect inspection.

In this paper, we will discuss the influence of resist thickness and underlayer films on Scanning Electron Microscope (SEM) using ASML's eP5. For this study, we prepared a set of wafers with 16nm line-space (LS) patterns at different resist thicknesses using either Metal-Oxide Resist (MOR) or Chemically Amplified Resist (CAR). The case study compares both resist platforms on Spin on Glass (SoG) and Organic Underlayer (OUL). To conduct this study our BKM ADI eP5 SEM condition has been used.

We observed that thinning down film thickness leads to deterioration of SEM image quality as well as loss of defect sensitivity during the inspection. In addition, varying the underlayers greatly affects both image quality and sensitivity. By decreasing the resist thickness, MOR and CAR were observed to behave in the same way, in terms of both image degradation and loss of sensitivity, although image quality did not degrade in the same way in the two cases.