The effects of Ti interlayer on the formation of Ni silicides on Si͑100͒ substrate was investigated. The phase and microstructure of Ni silicides during isochronal annealing treatment are monitored in situ by curvature measurement and characterized ex situ by other measurements. The addition of a thin Ti layer which formed an amorphous layer with the Si substrate acted as a diffusion barrier for the Ni atom in the early phase transformation process. The transformation temperature for the NiSi is significantly increased in an isochronal annealing process. When the NiSi phase is formed, the Ti-related layer moves to the surface of the silicide film. The temperature for the transformation of NiSi to NiSi 2 is relatively irrelevant to the presence of the Ti interlayer. A preferential orientation of ͑200͒ and ͑002͒ was found in the NiSi phase formed via the Ti interlayer, which significantly improved the epitaxial quality of NiSi 2 on Si͑100͒ substrate with a stepwise interface bounded by ͑111͒ and ͑100͒ planes. The top continuous Ti-related layer causes the incomplete accommodation of the thermal stress for the system. Upon disintegration of the Ti-related layer, the thermal stress of the system could be completely accommodated by relaxation of the NiSi layer.Metal silicides have been widely used in ultralarge scale integrated ͑ULSI͒ circuits to reduce the contact resistance of source/ drain in complementary metal-oxide-semiconductor ͑CMOS͒ devicess. Currently, the most commonly used silicides are TiSi 2 and CoSi 2 . For TiSi 2 , the transformation from the high-resistivity C49-TiSi 2 phase to the low-resistivity C54-TiSi 2 phase is nucleation limited, 1-3 causing linewidth dependence of the sheet resistance for gate lines narrower than submicrometer dimension. Therefore, CoSi 2 and NiSi have emerged as the most promising candidates to replace C54-TiSi 2 in self-aligned silicide ͑salicide͒ technology. However, as the lateral dimension continues to shrink, the junction depth of the device also decreases. The requirement of the shallow junction makes the silicon consumption by the silicidation process an important issue. For 0.1 m technology, even CoSi 2 is not a satisfactory material. Hence, NiSi is being investigated as a candidate for future device technologies because the same sheet resistance can be obtained with less Si consumption compared to CoSi 2 . 4 NiSi is not the final stable phase for the Ni/Si reaction system. Once the processing temperature goes higher than 800°C, NiSi transforms into NiSi 2 phase. Realization of the ULSI application for NiSi demands that the thermal stability of the phase be extended as much as possible. 4-7 Recently, great efforts have been made to study the thermal stability of NiSi, reportedly being improved by use of a thin Pt, [8][9][10][11][12][13][14][15]16 and Co 16 interlayer between the Ni film and Si substrate. The sheet resistance of the reaction system remains at the same low level in a wide temperature range. In general, the exploration of different interlayers on a react...

Thermal stability is one of the key issues in phase-change memory. We try to tackle it by developing new compositions based on Ga-Te-Sb system. Thermal stability is exemplified using Ga 18 Te 12 Sb 70 which shows crystallization-temperature (T x ) 248 C and activation energy of non-isothermal crystallization 5.9 eV. Films were isothermally soaked at 5 $ 30 C below T x to estimate the failure-time when electrical resistance dropped to a half of the original. Arrhenius plot attained using logarithm failure-time versus reciprocal temperature were extrapolated to the temperature corresponding to 10-year failure (T 10y ) as 183 C. Pre-crystallization structure upon heating to 2 $ 5 C below T x reflects stable amorphous phase of the alloy up to at least 240 C. Memory-cells made of Ga 18 Te 12 Sb 70 can be set-reset at 20 $ 500 ns with electrical currents around 66% those of our Ge 2 Sb 2 Te 5 cells. We suggest that compositions Ga 18-25 Te 8-12 Sb 67-70 are optimal to ensure T x > 240 C, T 10y > 180 C and with low operation-currents. V C 2012 American Institute of Physics.

No abstract

No abstract

No abstract