Jun 1, 2020

[paper] Device Scaling for 3-nm Node and Beyond

Opportunities in Device Scaling for 3-nm Node and Beyond:
FinFET Versus GAA-FET Versus UFET
U. K. Das and T. K. Bhattacharyya
in IEEE TED, vol. 67, no. 6, pp. 2633-2638, June 2020, 
doi: 10.1109/TED.2020.2987139

Abstract: The performances of FinFET, gate-all-around (GAA) nanowire/nanosheet, and U-shaped FETs (UFETs) are studied targeting the 3-nm node (N3) and beyond CMOS dimensions. To accommodate a contacted gate pitch (CGP) of 32 nm and below, the gate length is scaled down to 14 nm and beyond. While going from 5-nm node (N5) to 3-nm node (N3) dimensions, the GAA-lateral nanosheet (LNS) shows 8% reduction in the effective drain current (Ieff) due to an enormous rise in short channel effects, such as subthreshold slope (SS) and drain-induced barrier lowering (DIBL). On the other hand, 5-nm diameter-based lateral nanowire shows an 80% rise in Ieff. Therefore, to enable future devices, we explored electrostatics and Ieff in FinFET, GAA-FET, and UFET architectures at a scaled dimension. The performances of both Si- and SiGe-based transistors are compared using an advanced TCAD device simulator.

Fig: Transistor architectures for future technologies. (a) FinFET device
(in {001} substrate plane, and sidewalls are in {110} planes) with crosssectional
fin channel (5 nm thin). (b) Fin is changed into a four-stacked
GAA-LNWs. (c) GAA- LNS having 20-nm width (W). (d) UFET structure.

Acknowledgment: The authors would like to thank Dr. Bidhan Pramanik, IIT Goa, India, Dr. KB Jinesh, IIST, Trivandrum, India, and Dr. Geert Eneman, IMEC, Leuven, Belgium, for their valuable technical support.

URL: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9078841&isnumber=9098120

#US putting $37bn into #semiconductors. The aim is to match #China in its state investment in semiconductors. https://t.co/jJ6adDEiSE #paper https://t.co/pHv5mcyi6m


from Twitter https://twitter.com/wladek60

June 01, 2020 at 08:59AM
via IFTTT

May 31, 2020

#DISLIN Graphics Library to plot S- and Y-parameters on Smith Charts, and to create arbitrary Cartesian plots of S-parameter data along with Polar plots of radiation patterns. https://t.co/Fj7CEHywUv #paper https://t.co/A0cRlYw3uH


from Twitter https://twitter.com/wladek60

May 31, 2020 at 09:55PM
via IFTTT

[paper] ReRAM: History, Status, and Future

ReRAM: History, Status, and Future
Y. Chen, Member, IEEE
Western Digital Corporation, Milpitas, CA
in IEEE TED, vol. 67, no. 4, pp. 1420-1433, April 2020
doi: 10.1109/TED.2019.2961505.


Abstract: This article reviews the resistive random-access memory (ReRAM) technology initialization back in the 1960s and its heavily focused research and development from the early 2000s. This review goes through various oxygen/oxygen vacancy and metal-ion-based ReRAM devices and their operation mechanisms. This review also benchmarks the performance of various oxygen/oxygen vacancy and metal-ion-based ReRAM devices with general trend drawn. Being a semiconductor memory and storage technology, the commercialization attempts for both stand-alone mass storage/storage-class memory and embedded nonvolatile memory are also reviewed. Looking toward the coming era, the potential of using ReRAM technology to improve machine learning efficiency is discussed. 
Fig: General category of resistive switching memory technologies
with ReRAM highlighted as the review focus

Acknowledgment: Sincere acknowledgment to people who ever contribute to ReRAM technology development and understanding.

URL: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8961211&isnumber=9046113