In this study we investigate the pattern collapse mechanism of dense patterns with resolution under 60nm printed in Extreme Ultra Violet (EUV-IL) and Electron Beam Lithographies (EBL). Pattern collapse occurs when physical properties of the material can't imbalanced the capillary force exerted on the pattern during the drying of the rinse liquid. In former simulation models, the height of the pattern at which collapse occurs (critical height, Hc) was predicted using either elastic deformation properties, or plasticizing limit value of the resist. Experimental observations of unstuck patterns, lead us to develop 2 new models considering the adhesion properties of the resist film on the substrate. By comparing simulated to experimental results for varying pattern pitchs printed in 2 Chemically Amplified Resists (CARS), we show that pattern collapse behaviour of EUV-IL and EBL patterns is not only ruled by rigidity or strength of the resist but can be perfectly described with equation defining the unsticking of a non bending pattern. Finally by using surfactinated solution on sub-60nm dense patterns, great improvements in Hc values and increase of process window latitude are shown. However, due to larger capillary force, this efficiency decreases with pattern pitch and appears limited on patterns width smaller than 40 nm.
Future lithography tools will have to address the 32 nm node. EUV lithography at 13.4 nm wavelength is the technology that may achieve such resolution if chemically amplified EUV resists show high enough resolution capabilities. However for sub 100 nm line width patterns, the pattern collapse, generated during the drying step of the developing process, becomes a serious limiting phenomenon. We performed ultra high resolution exposures of EUV positive chemically amplified resists using either electron beam lithography (EBL), or EUV interferometry Lithography (IL) produced in a synchrotron. Two theoretical models have been compared with experimental results. One is mainly dealing with adhesion failure and the other with the line deformation. Adhesion failure occurs when capillarity pressure on the pattern become stronger than the attractive Van der Walls forces assuring the pattern adhesion on the substrate. Mechanical failure occurs once the lines deflection exceeds the mechanical breaking resistance of the resist. We highlighted that pattern collapse mode depends on resist thickness. Collapsing of patterns with thickness>100 nm are properly fitted with the deformation model of the resist; whereas for pattern height under 60 nm, experimental results obtained by EUVIL and EBL are properly predicted with the adhesion failure model. To push resolution further and avoid pattern collapse, we targeted to expose sub 100 nm thick resist films. AFM3D measurements on EBL patterns show that reducing the resist thickness increases their top Line Width Roughness (LWR) testifying of physical resist properties variations in the resist interfacial layers. However we pointed out an optimum resist thickness, hence an optimal dilution. By tuning developer normality and puddle time, straight resist profiles were obtained. Finally we reached dense 40/40 nm lines in XP9947W150 resist using both exposure tools and validate the process compatibility with future etching steps by transferring 40/40 nm dense lines patterned with EBL into a metallic hard mask.