NXE:3100 full wafer imaging performance and budget verification

Abstract

With the introduction of the NXE:3100 NA=0.25 exposure system a big step has been made to get EUV lithography ready for High Volume Manufacturing. Over the last year, 6 exposure systems have been shipped to various customers around the world, active in Logic, DRAM, MPU and Flash memory, covering all major segments in the semi-conductor industry. The integration and qualification of these systems have provided a great learning, identifying the benefits of EUV over ArF immersion and the critical parameters of the exposure tool and how to operate it.In this paper we will focus specifically on the imaging performance of the NXE:3100 EUV scanner. Having been operational for more than a year a wide range of features were evaluated for lithographic performance across the field and across wafer. CD results of 32nm contact holes, 27nm isolated and dense lines, 27nm two-bar, 22nm dense L/S with Dipole, as well as several device features will be discussed and benchmarked against the current ArF immersion performance. A budget verification will be presented showing CD and contrast budgets for a selection of lithographic features. The contribution of the resist process and the mask will be discussed as well.The litho performance optimization will be highlighted with the 27nm twobar and isolated lines features that are sensitive to the illuminator pupil shape and projection lens aberrations.We will estimate the amount of resist induced contrast loss for 27 and 22nm L/S based on measurements of Exposure Latitude and the contributors from the exposure system.We will further present on the impact of variations in the mask blank and patterned mask on imaging, with several new contributors to take into account compared to traditional transmission masks.Finally, the combined results will be projected to the NXE:3300 NA=0.33 exposure system to give an outlook for its imaging performance capabilities.

EUVL mask performance and optimization

“…The impact of absorber height on CD is demonstrated recently in the experiments performed on NXE:3100 [7] . A CD increase of approximately 0.6 nm per 1 nm absorber height was observed for 27 nm dense lines.…”

“…The CD of 27 nm dense lines having -2 nm HV bias on the mask -were measured on the wafer and the resulting CD shadowing profiles were determined. The solid lines in Figure 21 represent the experimental shadowing profiles averaged over several scanners (see also [7]). The shadowing profiles were simulated using the optical constants provided by the blank vendor and the CXRO optical constants.…”

Experimental approach to EUV imaging enhancement by mask absorber height optimization

“…In contrast, the CDU budget of dense features is mostly determined by mask contribution [1] and the Mask Error Enhancement Factor (MEEF). MEEF typically decreases if the image contrast is increased, but for larger absorber heights, wafer CD is more sensitive to mask CD variations.…”

“…It was shown experimentally [1] that it impacts CDU of 27 nm dense lines with a sensitivity of about 1.8 nm / nm. This sensitivity can be reduced if the absorber height is chosen in a minimum or a maximum of CD swing curve [5], [2] .…”

“…A high-contrast image is also less sensitive to variations of scanner parameters such as aberrations, dynamics etc., therefore the Critical Dimension Uniformity (CDU) of a high contrast image is also better. This is especially important for (semi-) isolated features as their CDU is mostly scanner driven [1] .…”