Trigger-When-Charged: A Technique for Directly Measuring RTN and BTI-Induced Threshold Voltage Fluctuation Under Use-<inline-formula> <tex-math notation="LaTeX">${V}_{dd}$ </tex-math> </inline-formula>

Abstract

Low-power circuits are important for many applications, such as Internet of Things. Device variations and fluctuations are challenging their design. Random telegraph noise (RTN) is an important source of fluctuation. To verify a design by simulation, one needs assessing the impact of fluctuation in both driving current ΔI <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">d</sub> and threshold voltage ΔV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sub> . Many early works, however, only measured RTN-inducedAId. ΔV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sub> was not directly measured because of two difficulties: its average value is low and it is highly dynamic. Early works often estimated ΔV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sub> from ΔI <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">d</sub> /g <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">m</sub> (V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">g</sub> = V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">dd</sub> ), where g <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">m</sub> is the transconductance, without giving its accuracy. The objective of this paper is to develop a new Trigger-When-Charged (TWC) technique for directly measuring the RTN-induced ΔV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sub> . By triggering the measurement only when a trap is charged, measurement accuracy is substantially improved. It is found that there is a poor correlation between ΔI <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">d</sub> /g <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">m</sub> (V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">g</sub> = V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">dd</sub> ) and the directly measured ΔV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sub> (V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">g</sub> = V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sub> ). The former is twice of the latter on average. The origin for this difference is analyzed. For the first time, the TWC is applied to evaluate deviceto-device variations of the directly measured RTN-induced ΔV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sub> without selecting devices.