A Stand-Alone, Physics-Based, Measurement-Driven Model and Simulation Tool for Random Telegraph Signals Originating From Experimentally Identified MOS Gate-Oxide Defects

Abstract

Investigating random telegraph signals (RTS) observed in MOS devices is important for studying the gate-oxide defect characteristics and developing simulation and modeling tools in highly scaled devices. In this paper, we are presenting a comprehensive, variable-temperature, single-to-multitrap scalable RTS model and a simulation tool (RTSSIM) based on the first principles, and supported by experimental data. The physical and electrical characteristics of the actual oxide defects are considered, such as trapping barrier energy, capture cross section, trap-binding enthalpy, entropy change with emission, Coulombic scattering effect, and the trap location. Experimentally obtained time-domain switching data and RTS traces reconstructed by the developed simulation tool are compared for assessing the accuracy of the modeling. In this paper, we identified the trap species responsible for the RTS as an unrelaxed neutral oxygen deficiency center (V 0 ODC II) with measured and theoretically verified relaxation energy of 1.11, 1.25, and 0.60 eV. RTSSIM is a tool used to mimic or reproduce a given noise measurement on a particular device. The simulation tool shows over 92% accuracy in replicating the RTS and trap characteristics. The model also represents the first realistic simulation of multilevel RTS.Index Terms-Complex random telegraph signals (RTS), gate-oxide traps, RTS simulation, RTS, SiO 2 oxygen vacancies.

Anomalous random telegraph noise in nanoscale transistors as direct evidence of two metastable states of oxide traps

“…Therefore, based on the relationship 16 , 17 where T IL is the thickness of the SiO 2 interfacial layer and T HK the thickness of the HfO 2 high-k layer, ε SiO2 and ε HK are the dielectric constants for SiO 2 and the HfO 2 high-k dielectric, respectively, k B is the Boltzmann constant and T is the temperature, the vertical distance x to the oxide/channel interface can be attained. It is worth mentioning that since the RTN measurement region is above the threshold voltage in this work, the effect of ∂ψ S /∂V GS is small and thus is ignored for simplicity 18 . As shown in Fig.…”

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>

“…The increased number of devices per chip also leads to larger statistical spread [1,2] and high data transmission rate requires tight control of fluctuations [3]. Fluctuations have become a major concern for circuit design and have attracted many attentions recently [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20]. It has been reported that current fluctuation in some fresh devices can be over the typical device lifetime criterion of 10% [5].…”

“…average. The discrepancy originates partly from the device-todevice variation (DDV) of relative local current density beneath a trap at Vg=Vth [16][17][18][19] and partly from the charge-induced mobility degradation [20]. Some deeply scaled devices have analyzable RTN signals in terms of extracting mean capture/emission time [11], while others can have a complex within-a-device-fluctuation [12].…”

On the Accuracy in Modeling the Statistical Distribution of Random Telegraph Noise Amplitude

“…There have been many efforts to model RTN, both in the frequency domain [1], [19]- [21] and in the time domain [1]- [6]. It is widely accepted that RTN originates from trapping/detrapping charge carriers from/to the conduction channel [1]- [21]. The number of traps per device follows the Poisson distribution [3]- [6].…”