Effects of random number and location of the nanosized metal grains on the threshold voltage variability of silicon gate-all-around nanowire n-type metal-oxide-semiconductor field-effect transistors

Deep Learning Algorithms for the Work Function Fluctuation of Random Nanosized Metal Grains on Gate-All-Around Silicon Nanowire MOSFETs

“…In our prior work [17], it has been observed that the highκ/metal-gate (HKMG) technology with GAA Si NW MOSFETs is considered to be a more effective technology and the magnitude of electrical characteristics induced by the WKF depend on two factors: the random number of MGs and the random position of MGs. The most crucial electrical characteristics induced by low and high WKs is the variability of the threshold voltage (VTH) defined by:…”

“…From the prior work [16], three-dimensional (3D) statistical device simulation depicted that the RDF in GAA Si NW MOSFETs greatly affects the device variability. In [17], characteristic fluctuation induced by the WKF on GAA Si NW MOSFETs was studied; and, the results have shown that the reduction of variation of the threshold voltage (VTH) affects the reduction of variation of the frequency. Similarly, in [18], the timings and the power fluctuations were determined by considering the various random discrete dopants (RDDs) on GAA Si NW complementary metal-oxide-semiconductor (CMOS); and, it concluded that the timing fluctuation and the power consumption in CMOS are directly dependent on the variation of the VTH.…”

“…To examine the effect of the variability of the WKF on emerging MOSFETs, the cuboid grain method and the averaged WKF method, etc., [52] have been proposed and proven remarkable. In [17], the cuboid grain method was adopted to study the effect of the metal-grain number (MGN) and metal-grain location (MGL) with the low and high work functions on the variability of the VTH. However, with the increase of MGN, the requirement for a large number of samples of WK also increases.…”

Work-Function Fluctuation of Gate-All-Around Silicon Nanowire n-MOSFETs: A Unified Comparison Between Cuboid and Voronoi Methods

“…To compare the cuboid and Voronoi methods, first, the main advantages of the cuboid method are (i) grains generation is very simple without any acute angle by C++ code and (ii) it be modelled analytically [20] with equal sub-area of metal grains. Moreover, the effects of MGN and random location of metal grain can be further analyzed by the cuboid method [21], where the MGN is equal to (Lg/GS) × (Weff/GS). One possible disadvantage of the cuboid method is that it may yield a residual error when the metal-gate region cannot be totally covered by cuboids.…”

“…Thus, for the two methods with merely 150 samples, according to (1), the calculated values of Vth will be biased and then result in errors between the two methods. Physically, it can be attributed that the random location effect of metal grains [21,24]; to further analyze the physical mechanism of random location effect of metal grains for the case of MGN = 16. Both Case 1 (Vth = 246.8 mV) and Case 2 (Vth = 211.9 mV) for the cuboid method, and both Case 3 (Vth = 234.5 mV) and Case 4 (Vth = 216.7 mV) for the Voronoi method are selected for the same number of high WK (number = 6), as shown in Figs.…”