A novel and simple process for the exploitation of metal-gate-induced stress in the Si channel region of a FinFET is reported. TaN metal-gate electrode was employed. The use of a silicon nitride capping layer, which covered the metal gate during the source and drain anneal, led to a large stress being developed as a result of the thermal-expansion coefficient mismatch between the gate and the capping layer. The resulting strain was retained and transferred to the Si channel. The stress introduced by the gate stressor was found to enhance the performance of the n-channel FinFETs and is believed to be tensile in the sourceto-drain direction. This approach may be applicable to other metal-gate materials having a mismatch in the thermal-expansion coefficient with surrounding capping materials.

Low temperature crystallization of high permittivity Ta oxide using an Nb oxide thin film for metal/insulator/metal capacitors in dynamic random access memory applications J. Vac. Sci. Technol. B 23, 80 (2005); 10.1116/1.1829060Effect of growth temperature and postmetallization annealing on the interface and dielectric quality of atomic layer deposited HfO 2 on p and n siliconThe formation of tungsten nanocrystals ͑W-NCs͒ on atomic-layer-deposited HfAlO/ Al 2 O 3 tunnel oxide was demonstrated for application in a memory device. It was found that the density and size distribution of W-NCs are not only controlled by the initial film thickness, annealing temperature, and time, but also by the metal/tunnel oxide interface structure. Well-isolated W-NCs with an average diameter of 5 nm and a surface density of 5 ϫ 10 11 cm −2 were obtained by applying a thin Al 2 O 3 wetting layer onto HfAlO tunneling oxide. A large flatband voltage shift of 5.7 V was observed from capacitance-voltage measurement when a bias voltage up to ±4 V was applied.

For the first time, we perform a systematic investigation of the fast and slow components of dynamic NBTI (DNBTI)in p-MOSFET with SiON gate dielectric. The new findings are: (1)The recent debate in the slow DNBTI component measured by conventional DC method [1-5] is clarified. We show evidence that the slow DNBTI is due to interface traps N it generation and passivation. The conventional methods used over the past years seriously underestimate N it due to passivation of N it during measurement. (2) The fast DNBTI component measured by the fast method [7] is due to trapping and de-trapping of hole traps N ot in SiON. The accumulative degradation increases with increasing stress frequency. Model simulations are in excellent agreement with all experiments. (3)We re-evaluate the impact of DNBTI on device lifetime and circuit applications in the light of this new finding. Introduction: DNBTI is a critical reliability issue for CMOS transistors with silicon oxynitride SiON gate dielectric [1-6].Using the fast measurement method developed in [7], the DNBTI degradation in ultra-thin SiON gate dielectrics is re-investigated. Device fabrication and measurement: Devices were fabricated using 0.11 �m CMOS technology. The gate oxides with EOT of 1.3nm were grown by thermal oxidation followed by decoupled plasma nitridation and post-deposition thermal annealing. The fast V th measurement method [7] is illustrated in Fig.1. Pulses are inserted in gate stress and I d -V g is measured at falling or rising edge of the pulse (Fig. 1 inset). Fig.2 shows that 50us measurement time is fast enough. Interface trap density is measured by an improved DCIV method [1]. What DC and fast methods measure in DNBTI: Fig.3 shows the DNBTI [1,7] response of a p-FET, measured by conventional DC and fast methods respectively. DC method :Figs.4-6 clearly demonstrate that the slow components in the DNBTI measured by the DC method is mainly due to generation and passivation of the interface traps N it [1-4], and not due to trapping and detrapping in SiON hole traps N ot as proposed in [5]. Figs.4,5 show that the transient result depends on the order of measurements of N it and V th , indicating both measured N it and �V th are underestimated due to passivation of N it during measurement. Fig.6 demonstrates the misleading interpretation by [5] in the passivation phase due to the wrong curve alignment. Fast method: In Fig.8, using the pulse falling and rising edges to measure points S f and P in Fig.3, the transient amplitude can be measured. In Fig.9, rising edge measures the accumulative degradation under DNBTI. Figs.10-12 show the static and dynamic NBTI measured by fast and DC methods respectively, showing different time, V g and temperature dependences. The slope of time dependence of the fast method result is much smaller than the value of 0.25 anticipated in interface trap diffusion-reaction model [2]. Actually the fast NBTI component, defined as the difference between �V th measured by fast and slow method, can be interpreted by trapping and de...

No abstract

No abstract