The power law dependence of the threshold voltage shift (∆V th ) on stress time for high-k nMOSFETs is studied using a single-pulse I d −V g technique. The power law exponent value is found to be strongly affected by fast transient charge detrapping during the stress interruption time, which may result in an inaccurate lifetime prediction. A new analysis method that eliminates the impact of the stress interruption time is proposed to evaluate bias temperature instability in n-type high-k devices.Index Terms-Lifetime prediction, relaxation, single pulse, transient charge.T HE STRESS-TIME dependence of the threshold voltage shift (∆V th ) is commonly used to predict device lifetime. This dependence is usually described by the simple power law relationship (∆V th ∝ t n ). However, it has been observed that ∆V th relaxes during stress interruption for sense measurements in high-k nMOSFETs under bias temperature instability (BTI) stress [1]. This phenomenon is attributed to the fast charge-relaxation (detrapping) process that occurs within microseconds once the stress bias is removed [1]- [4]. Such fast charge relaxation raises questions about the validity of the conventional V th monitoring methodology during BTI stress and the subsequent accuracy of lifetime predictions of high-k nMOSFETs.To address the issue of charge relaxation during sense measurements in high-k devices, it has been suggested to monitor drain current (I d ) degradation during stress without interruption [5]. However, monitoring I d degradation at the stress voltage (i.e., in the I d saturation regime) does not directly provide ∆V th values, while sensing I d in the linear regime is inevitably accompanied by a delay associated with switching the gate bias from the stress voltage to the sense voltage. To minimize charge trapping/detrapping during sense measurements, the singlepulse I d −V g technique has been developed [6], [7] and applied in BTI studies [8]. The sense-measurement time (pulse time) as well as charge relaxation during measurement is dramatically reduced using this single-pulse I d −V g technique, compared to the conventional dc I d −V g method, as shown in Fig. 1. In this letter, the single-pulse I d −V g measurement with different pulse times is used to study the effect of charge relaxation Manuscript Fig. 1. Comparison of stress sense time scales using the conventional dc and the single-pulse I d −Vg measurement methods.during sense measurements on the measured ∆V th in the high-k devices. A new analysis method is then proposed to extract the intrinsic stress-time dependence of ∆V th measured using both the single-pulse and conventional dc I d −V g techniques.The devices used in this letter were fully processed nMOSFETs, fabricated using a standard CMOS process with a 1000-• C 10-s dopant activation anneal. The gate dielectric stack included a 3-nm atomic-layer-deposited (ALD) HfO 2 dielectric deposited on a 1-nm thermal SiO 2 . The gate electrode was formed using CVD TiN capped with a 100-nm poly-Si layer. The details of the proc...