Application for GCIP Model to Non-volatile Memory in HK/MG NMOSFET

碩士 === 國立臺北科技大學 === 機電整合研究所 === 104 === The MOSFETs oxide thickness is sustained decreases, MOSFETs leakage current be-comes bigger, so use HK / MG replace SiO2 or SiON as the current mainstream method; Leakage current not only cause unnecessary power consumption of the MOSFETs, The memory device to...

Full description

Bibliographic Details
Main Authors: Li-Fu Yang, 楊理夫
Other Authors: Shea-Jue Wang
Format: Others
Language:en_US
Published: 2016
Online Access:http://ndltd.ncl.edu.tw/handle/9n2rp7
id ndltd-TW-104TIT05651087
record_format oai_dc
spelling ndltd-TW-104TIT056510872019-05-15T23:00:43Z http://ndltd.ncl.edu.tw/handle/9n2rp7 Application for GCIP Model to Non-volatile Memory in HK/MG NMOSFET GCIP Model 應用於HK/MG NMOSFET之非揮發性記憶體 Li-Fu Yang 楊理夫 碩士 國立臺北科技大學 機電整合研究所 104 The MOSFETs oxide thickness is sustained decreases, MOSFETs leakage current be-comes bigger, so use HK / MG replace SiO2 or SiON as the current mainstream method; Leakage current not only cause unnecessary power consumption of the MOSFETs, The memory device to read and write is also cause significant interference. Because there are few studies of GCIP and CHC to program on logic device, so this is our point of the thesis. In our experiment, the manufactory UMC 28 nm advance process which channel length is between 0.15 μm to 0.24 μm sample would be used. The gate terminal use the HfxZryOz ( HfOx/ZrOx/HfOx ) dielectric with atomic layer deposition (ALD) technology. For our re-search, the behavior of memory program have two methods. The first behavior is programed by CHC. As MOSFET operate in saturation region, this condition will be affected by high electric field that induced not only impact-ionization but also FN tunneling which increase opportunity for the electric enter in the oxide layer. The other behavior use Gate Current Induced Punch-through Model ( GCIP Model ) that source current access to the reverse direction and generate electron-hole pair to the program resource. The gate terminal provide bias that electric can get away from the excitation situation and program to the gate. Electric will charge to the dielectric layer trap as the behavior same to the program process from memory device. In the results of this thesis, writing speed of GCIP Model is shower than CHCI. Because GCIP Model initial current is lower than CHCI; But the GCIP program with small gate voltage bias is less than CHCI, Those appearance that the GCIP Model have slower program speed but low energy consuming. In the future, I think we can improve GCIP Model program speed such as increase gate voltage to strengthening perpendicular Electric Fields to attract electrons. In-crease the drain voltage that lead the range of GCIP increase. Shea-Jue Wang Heng-Sheng Huang 王錫九 黃恆盛 2016 學位論文 ; thesis 0 en_US
collection NDLTD
language en_US
format Others
sources NDLTD
description 碩士 === 國立臺北科技大學 === 機電整合研究所 === 104 === The MOSFETs oxide thickness is sustained decreases, MOSFETs leakage current be-comes bigger, so use HK / MG replace SiO2 or SiON as the current mainstream method; Leakage current not only cause unnecessary power consumption of the MOSFETs, The memory device to read and write is also cause significant interference. Because there are few studies of GCIP and CHC to program on logic device, so this is our point of the thesis. In our experiment, the manufactory UMC 28 nm advance process which channel length is between 0.15 μm to 0.24 μm sample would be used. The gate terminal use the HfxZryOz ( HfOx/ZrOx/HfOx ) dielectric with atomic layer deposition (ALD) technology. For our re-search, the behavior of memory program have two methods. The first behavior is programed by CHC. As MOSFET operate in saturation region, this condition will be affected by high electric field that induced not only impact-ionization but also FN tunneling which increase opportunity for the electric enter in the oxide layer. The other behavior use Gate Current Induced Punch-through Model ( GCIP Model ) that source current access to the reverse direction and generate electron-hole pair to the program resource. The gate terminal provide bias that electric can get away from the excitation situation and program to the gate. Electric will charge to the dielectric layer trap as the behavior same to the program process from memory device. In the results of this thesis, writing speed of GCIP Model is shower than CHCI. Because GCIP Model initial current is lower than CHCI; But the GCIP program with small gate voltage bias is less than CHCI, Those appearance that the GCIP Model have slower program speed but low energy consuming. In the future, I think we can improve GCIP Model program speed such as increase gate voltage to strengthening perpendicular Electric Fields to attract electrons. In-crease the drain voltage that lead the range of GCIP increase.
author2 Shea-Jue Wang
author_facet Shea-Jue Wang
Li-Fu Yang
楊理夫
author Li-Fu Yang
楊理夫
spellingShingle Li-Fu Yang
楊理夫
Application for GCIP Model to Non-volatile Memory in HK/MG NMOSFET
author_sort Li-Fu Yang
title Application for GCIP Model to Non-volatile Memory in HK/MG NMOSFET
title_short Application for GCIP Model to Non-volatile Memory in HK/MG NMOSFET
title_full Application for GCIP Model to Non-volatile Memory in HK/MG NMOSFET
title_fullStr Application for GCIP Model to Non-volatile Memory in HK/MG NMOSFET
title_full_unstemmed Application for GCIP Model to Non-volatile Memory in HK/MG NMOSFET
title_sort application for gcip model to non-volatile memory in hk/mg nmosfet
publishDate 2016
url http://ndltd.ncl.edu.tw/handle/9n2rp7
work_keys_str_mv AT lifuyang applicationforgcipmodeltononvolatilememoryinhkmgnmosfet
AT yánglǐfū applicationforgcipmodeltononvolatilememoryinhkmgnmosfet
AT lifuyang gcipmodelyīngyòngyúhkmgnmosfetzhīfēihuīfāxìngjìyìtǐ
AT yánglǐfū gcipmodelyīngyòngyúhkmgnmosfetzhīfēihuīfāxìngjìyìtǐ
_version_ 1719138693232459776