UCLA

This work addressed the processing challenges associated with the phase-change material GeSbTe (GST) for use in phase-change memory (PCM) devices. As the GST composition dictates its crystallization temperature, a careful control of the composition is essential to ensure the targeted phase-change characteristics. This work focused on studying the effect of surface states and plasma chemistries on the compositional control of GST. The study of surface states focused on the effect of ambient exposure on changing the GST composition and reactivity. Ambient oxidation of GST was shown to be most strongly influenced by water vapor in the atmosphere, which resulted in a greater extent of oxidation than those exposed to N2 or O2. To restore the GST composition by wet chemical processing, the effectiveness depends significantly on whether the native oxides were removed, as the difference in polarity between metallic Ge, Sb, Te bonds and Ge-O, Sb-O, Te-O bonds made the oxides more susceptible to removal by polar chemistries. For example, citric acid resulted in significantly greater change in composition on the oxidized GST than on sputter-cleaned GST. For plasma etching of GST, thermochemical analysis was used to assess the viability of etch chemistries by quantifying the volatility of the potential etch products, followed by plasma etching experiments. First, halogen chemistries including Cl2 and SF6/Ar were evaluated based on thermochemical analysis, predicting that GST could be etched by both chemistries, with Ge being preferentially removed, followed by Sb then Te. Plasma etch experiments demonstrated an etch rate of GST by Cl2/Ar was 13.1 nm/min and that by SF6/Ar was negligible. The extent of halogen incorporation on the etched films followed the increasing bond energies between individual GST elements with the halogen species. SF6/Ar resulted in the smaller composition change and less crystallization temperature change, but the significant halogenation was problematic. For H2 plasma, thermochemical analysis predicted that atomic hydrogen could etch all elements, but molecular hydrogen could not. The effect of H2 plasma was examined first by etching each GST element separately. It was found that hydrogen plasmas etched Sb the fastest (with SbHx as dominant etch products) and Ge the slowest (with no observed gas phase products). The etch rate of GST in a H2 plasma was 32.7 nm/min when etched directly but was reduced to zero in a downstream configuration. Sb and Te were preferentially removed by H2 plasma. This observed preferential removal of Sb and Te over Ge followed the order of the measured etch rates of Ge, Sb, and Te, suggesting that the preferential removal was due to the reaction kinetics rather than the desorption of the etch products from the surface. H2 plasma resulted in greater composition change than SF6/Ar plasma, less composition change than Cl2/Ar plasma, and greater crystallization temperature change than either halogen plasma. For the CH4 plasmas, only thermodynamic data for Ge was available, and it suggested that CH3 could etch Ge while CH4 could not. CH4 plasmas were found to etch Ge the fastest, Te moderately, but deposited on Sb (no etch). Consistent with these observations, in etching GST, preferential removal of Ge and Te was observed, leaving the surface enriched with Sb, with an etch rate of 16.8 nm/min. The addition of N2 to CH4 plasma led to the formation of HCN and NHx, which reduced the etch rate to 8.4 nm/min., likely due to the dilution of CH4 and scavenging of atomic H. Among the CH4-containing plasmas, CH4/N2 plasma resulted in the smallest composition and crystallization temperature changes in GST. In summary, considering the effects of surface states and plasma chemistries on GST composition and crystallization temperature, the removal of the surface oxides followed by CH4/N2 or H2 plasma processing was suggested to be most suitable for etching GST while maintaining the target composition and crystallization temperature.