EUV 리소그라피 공정시 발생하는 마스크의 카본 오염물 및 나노 입자의 흡착 및 제거 거동

Abstract

EUV lithography for next generation lithography (NGL) has remarkable differences with conventional lithography technology. The main difference is the use of photons of 13.5 nm wavelength for ultra-fine patterning. Since photons are heavily absorbed by all materials including air at this wavelength, the use of pellicle is forbidden and hence, imaging system operates in reflection by mirror in vacuum atmosphere is preferred. EUVL mask surface, consisting of Ru capping layer and absorber layer, is not only exposed to outside and easily to be contaminated during processing but also easily to be damaged by etching based wet cleaning chemicals. Hence, the selection of EUVL mask cleaning method and process conditions are being great significance. In the present work, the carbon contaminant removal and particle removal from the Ru capping layer were investigated in detail. A possible candidate for carbon contaminant removal in Ru capped EUVL mask is ozone dissolved water (DIO3). However, the use of DIO3 leaves reflectivity loss and serious surface damages on Ru capping layer because of its high oxidation potential. Thus, as a first step, the critical oxidation reduction potential (ORP) values above which the Ru surface undergoes corrosion were estimated from Pourbaix diagram. Then, the effect of addition of feed gases such as O2, CO2 and N2 of various concentrations to DIO3 generation on the ORP values was observed experimentally. From these theoretical and experimental investigations, the critical condition at which the Ru surface does not get corroded is identified and the cleaning experiment was performed at this conditions. However, carbon contaminant cleaning was ineffective at this critical DIO3 cleaning condition. To enhance the cleaning efficiency, 1MHz cone type megasonic is irradiated during DIO3 process. Surprisingly, the results show that not only carbon contaminant removal efficiency was increased but also reflectivity loss was suppressed. This is mainly due to the fact that the megasonic irradiation during DIO3 process is increasing the cleaning efficiency only by enhancing indirect oxidation rather than direct oxidation. To understand the loss of reflectivity during DIO3 cleaning, simulation was performed based on experimental results. The effect of reflectivity changes by surface roughness and pit damage was analyzed by S-Litho and EM suit software, respectively. The simulation results show that the effect of both roughness and pit damage on reflectivity loss was insignificant in sub nanometer scale. However, even small change in the ruthenium oxidation rate and thickness leads to the significant reflectivity change. Therefore even ozone process is effective for carbon contaminant cleaning, the use of ozone based cleaning solution have to be deliberated. As a second part of the work, the silicon nitride and gold particle removal from Ru capping layer was investigated. First, the zeta potential measurements were carried out to understand the interaction between the Ru substrate and the particles. The results show that the zeta potential is around zero for Ru for all pH values investigated and for tantalum nitride, the values were increased as pH value increases but the range was very small when compared with silicon. Similarly, for silicon nitride and gold particles, the values were increased as pH increases but the range was smaller than silica. From the zeta potential values, the total interaction force is calculated based on DLVO theory. The results show that the interaction force of gold and silicon nitride on ruthenium and tantalum nitride is attractive in the pH range investigated which means the particle cleaning on EUV mask is more difficult than on silicon wafer. For the particle cleaning, conventional SC-1 solution was used in EUV mask without theoretical concerning with its high particle cleaning performance in silicon wafer cleaning area. Therefor the characteristics of conventional cleaning solution SC-1 were studied on ruthenium capping layer. It was observed from these results that ruthenium surface was not efficiently etched and oxidized in SC-1, although the major particle cleaning mechanism of SC-1 is by surface etching and passivation. Moreover, in SC-1 cleaning, the pit type defect was detected as similar to DIO3 process. From all these results it is very obvious that the advanced particle removal process is essential for EUV mask. In summary of the characteristics of EUV mask, it has not only flat type zeta potential and repulsive force against particles in all pH ranges investigated but also the thickness of ruthenium capping layer has to be seriously controlled. Fortunately, EUV mask pattern has very high and enough collapse fore to use of megasonic. Here, the effective particle cleaning process must be etch free and should have high repulsive force and high removal force. By considering these conditions, megasonic induced surfactant cleaning process was suggested for the advanced particle cleaning on EUV mask. The characteristics of typical surfactants such as nonionic (TRITON X100), zwitterionic (CHAPS), anionic (SDS), cationic (CTAB) surfactant and its mixture (TRITON X100 & SDS) were analyzed and finally particle removal test was performed. Among these surfactants, only SDS shows repulsive force in zeta potential analysis. Particle removal test was performed with those surfactants with megasonic in neutral and alkali pH. The particle removal efficiency (PRE) in megasonic induced SDS (anionic surfactant) at pH 9 shows the highest PRE which was even higher than conventional SC-1 on EUV mask.