Comparative Analysis of Practical Threshold Voltage Extraction Techniques for CMOS. Yu-Hsing Cheng ON Semiconductor October 15, 2018

Similar documents
ECE 342 Electronic Circuits. Lecture 6 MOS Transistors

The Devices. Digital Integrated Circuits A Design Perspective. Jan M. Rabaey Anantha Chandrakasan Borivoje Nikolic. July 30, 2002

Operation and Modeling of. The MOS Transistor. Second Edition. Yannis Tsividis Columbia University. New York Oxford OXFORD UNIVERSITY PRESS

MOS Transistor I-V Characteristics and Parasitics

Device Models (PN Diode, MOSFET )

The Devices. Jan M. Rabaey

The Devices: MOS Transistors

Digital Integrated Circuits A Design Perspective. Jan M. Rabaey Anantha Chandrakasan Borivoje Nikolic. The Devices. July 30, Devices.

EECS130 Integrated Circuit Devices

MOSFET: Introduction

ECE 342 Electronic Circuits. 3. MOS Transistors

ECE 546 Lecture 10 MOS Transistors

Device Models (PN Diode, MOSFET )

CMOS INVERTER. Last Lecture. Metrics for qualifying digital circuits. »Cost» Reliability» Speed (delay)»performance

CMPEN 411 VLSI Digital Circuits. Lecture 04: CMOS Inverter (static view)

Modeling and Computation of Gate Tunneling Current through Ultra Thin Gate Oxides in Double Gate MOSFETs with Ultra Thin Body Silicon Channel

Lecture 4: CMOS Transistor Theory

MOS Transistors. Prof. Krishna Saraswat. Department of Electrical Engineering Stanford University Stanford, CA

The Devices. Digital Integrated Circuits A Design Perspective. Jan M. Rabaey Anantha Chandrakasan Borivoje Nikolic. July 30, 2002

MOS Transistor Theory

CMPEN 411 VLSI Digital Circuits. Lecture 03: MOS Transistor

Today s lecture. EE141- Spring 2003 Lecture 4. Design Rules CMOS Inverter MOS Transistor Model

Design for Manufacturability and Power Estimation. Physical issues verification (DSM)

ECE 497 JS Lecture - 12 Device Technologies

A Compact Analytical Modelling of the Electrical Characteristics of Submicron Channel MOSFETs

A Multi-Gate CMOS Compact Model BSIMMG

Nanoscale CMOS Design Issues

Simulation of the Temperature Influence in IC-EMC

Towards a Scalable EKV Compact Model Including Ballistic and Quasi-Ballistic Transport

EKV MOS Transistor Modelling & RF Application

ECE 415/515 ANALOG INTEGRATED CIRCUIT DESIGN

EEC 118 Lecture #2: MOSFET Structure and Basic Operation. Rajeevan Amirtharajah University of California, Davis Jeff Parkhurst Intel Corporation

CMOS Inverter (static view)

Practice 3: Semiconductors

The Physical Structure (NMOS)

Long Channel MOS Transistors

EE105 - Fall 2005 Microelectronic Devices and Circuits

Analysis of Transconductances in Deep Submicron CMOS with EKV 3.0

Lecture 13 MOSFET as an amplifier with an introduction to MOSFET small-signal model and small-signal schematics. Lena Peterson

EE105 Fall 2014 Microelectronic Devices and Circuits. NMOS Transistor Capacitances: Saturation Region

Simple and accurate modeling of the 3D structural variations in FinFETs

EE 560 MOS TRANSISTOR THEORY

EE 330 Lecture 16. Devices in Semiconductor Processes. MOS Transistors

Lecture 11: MOS Transistor

EE105 - Fall 2006 Microelectronic Devices and Circuits

Fundamentals of the Metal Oxide Semiconductor Field-Effect Transistor

Chapter 4 Field-Effect Transistors

A final review session will be offered on Thursday, May 10 from 10AM to 12noon in 521 Cory (the Hogan Room).

Power Dissipation. Where Does Power Go in CMOS?

Lecture 5: CMOS Transistor Theory

Lecture 04 Review of MOSFET

ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems. Today. Refinement. Last Time. No Field. Body Contact

Integrated Circuits & Systems

Lecture 12: MOSFET Devices

Integrated Circuits & Systems

L ECE 4211 UConn F. Jain Scaling Laws for NanoFETs Chapter 10 Logic Gate Scaling

Scaling Impact on Analog Circuit Performance

! PN Junction. ! MOS Transistor Topology. ! Threshold. ! Operating Regions. " Resistive. " Saturation. " Subthreshold (next class)

MOSFET Model with Simple Extraction Procedures, Suitable for Sensitive Analog Simulations

CMOS Transistors, Gates, and Wires

EE5311- Digital IC Design

Reduction of Self-heating effect in LDMOS devices

ENGR890 Digital VLSI Design Fall Lecture 4: CMOS Inverter (static view)

Lecture 11: MOSFET Modeling

The Intrinsic Silicon

DC and Transient Responses (i.e. delay) (some comments on power too!)

CHAPTER 5 EFFECT OF GATE ELECTRODE WORK FUNCTION VARIATION ON DC AND AC PARAMETERS IN CONVENTIONAL AND JUNCTIONLESS FINFETS

ECE 438: Digital Integrated Circuits Assignment #4 Solution The Inverter

2007 Fall: Electronic Circuits 2 CHAPTER 10. Deog-Kyoon Jeong School of Electrical Engineering

Technische Universität Graz. Institute of Solid State Physics. 11. MOSFETs

Fig. 1 CMOS Transistor Circuits (a) Inverter Out = NOT In, (b) NOR-gate C = NOT (A or B)

ESE 570: Digital Integrated Circuits and VLSI Fundamentals

Investigation of the Thermal Noise of MOS Transistors under Analog and RF Operating Conditions

Part 4: Heterojunctions - MOS Devices. MOSFET Current Voltage Characteristics

Chapter 20. Current Mirrors. Basics. Cascoding. Biasing Circuits. Baker Ch. 20 Current Mirrors. Introduction to VLSI

MOS Transistor Theory

Chapter 2 CMOS Transistor Theory. Jin-Fu Li Department of Electrical Engineering National Central University Jungli, Taiwan

EE115C Winter 2017 Digital Electronic Circuits. Lecture 3: MOS RC Model, CMOS Manufacturing

Chapter 5 MOSFET Theory for Submicron Technology

Contact resistance and TLM measurements

Lecture 15: MOS Transistor models: Body effects, SPICE models. Context. In the last lecture, we discussed the modes of operation of a MOS FET:

CHAPTER 3: TRANSISTOR MOSFET DR. PHAM NGUYEN THANH LOAN. Hà Nội, 9/24/2012

Semiconductor Devices. C. Hu: Modern Semiconductor Devices for Integrated Circuits Chapter 5

Lecture 4: Technology Scaling

Effect of Illumination on Organic Polymer Thin-Film Transistors

Impact of parametric mismatch and fluctuations on performance and yield of deep-submicron CMOS technologies. Philips Research, The Netherlands

Semiconductor Physics fall 2012 problems

Lecture 3: CMOS Transistor Theory

Important! EE141- Fall 2002 Lecture 5. CMOS Inverter MOS Transistor Model

ECEN474/704: (Analog) VLSI Circuit Design Spring 2018

SECTION: Circle one: Alam Lundstrom. ECE 305 Exam 5 SOLUTIONS: Spring 2016 April 18, 2016 M. A. Alam and M.S. Lundstrom Purdue University

Name: Answers. Grade: Q1 Q2 Q3 Q4 Q5 Total. ESE370 Fall 2015

The Devices. Devices

Lecture 15: Scaling & Economics

VLSI Design The MOS Transistor

General I-V and Qb linearization Tuesday, February 28, :34 AM

The drive to make devices smaller and faster

MOS Transistor Properties Review

MODEL MECHANISM OF CMOS DEVICE FOR RELIBILITY ENHANCEMENT

ESE 570: Digital Integrated Circuits and VLSI Fundamentals

Transcription:

Comparative Analysis of Practical Threshold Voltage Extraction Techniques for CMOS and LDMOS Devices in 180 nm Technology Yu-Hsing Cheng ON Semiconductor October 15, 2018

Outline Overview and Background MOSFET Device Id-Vg Characteristics and Threshold Voltage Threshold Voltage Extraction Techniques Comparison of Threshold Voltage Extraction Results Summary and Conclusion 2

Overview and Background

Technology Trends : Moore s Law and More Sample title (First Level) Sample Description Text (second Level) Sample Bullet (third level) Sample Bullet 2 (fourth level) o Sample Bullet 3 (fifth level) Guo Qi Zhang, Alfred Roosmalen, The Changing Landscape of Micro/Nanoelectronics, in More than Moore: Creating High Value Micro/Nanoelectronics, Springer, 2009. The first transistor was invented in 1947 (announced in 1948). The first IC was constructed in 1958. Gordon Moore first published his observation and prediction in 1965. It has become a self-fulfilling prophecy and a yardstick for technology progress. More-than-Moore refers to non-digital functionalities not scaling at the rate of Moore s Law. Trending: Smart manufacturing, virtual metrology, heterogeneous integration 4

MOS Scaling (Dennard Scaling) First described by Robert Dennard et al. in 1974. High leakage current caused the breakdown of Dennard scaling around 2006 (Dark Silicon). Scott E. Thompson et al., IEEE TRANSACTIONS ON SEMICONDUCTOR MANUFACTURING, VOL. 18, NO. 1, FEBRUARY 2005 5

Power Supply and Threshold Voltage Scaling V T and V DD scaling limited by off-state leakage current. Scott E. Thompson et al., IEEE TRANSACTIONS ON SEMICONDUCTOR MANUFACTURING, VOL. 18, NO. 1, FEBRUARY 2005 Yuan Taur, IEEE Spectrum, 1999, Vol. 36, No. 7 6

Driving Energy Efficient Innovations Analog Solutions Group Power Solutions Group Intelligent Sensing Group 7

MOSFET Device Id-Vg Characteristics and V T

MOSFET Device Operation: Inversion Level Yannis Tsividis, Operation and Modeling of the MOS Transistor, 2 nd Edition, 1999 9

Id-Vg Characteristics of MOSFET Devices Sample title Id vs. (First V Level) G Sample Description Text (second Level) Sample Bullet (third level) Sample Bullet 2 (fourth level) o Sample Bullet 3 (fifth level) semi-log plot of Id vs. V G Subthreshold swing is related to kt/q. 5V NMOS W = 10 µm, L = 10,2,1,0.65,0.6,0.55 0.5 µm 10

Meaning of Threshold Voltage V T Threshold voltage is Gate voltage that turns the device on Conceptual level, from the perspective of a user Minimum gate voltage required to induce the channel Conceptual level, from the perspective of device physics Gate voltage at which strong inversion begins Conceptual level, from the perspective of device physics Gate voltage for the inversion of surface to the same degree as the bulk (p- to n- or n- to p-) Conceptual and operational level, from the perspective of device physics o Based on symmetry Gate voltage needed to reach a set value of Ids Operational level, from the perspective of device characterization https://en.wikipedia.org/wiki/threshold_voltage Yannis Tsividis, Operation and Modeling of the MOS Transistor, 2 nd Edition, 1999 11

Measurement and Data Collection Process: 180 nm BCD for high power applications with high efficiency in power conversion. Gate Oxide Thickness: 2.7 nm for 1.8V MOS. 12.5 nm for 5V MOS. TCAD Simulation: East Greenwich, RI Wafer Fab: Gresham, OR Device Evaluation: East Greenwich, RI EG TD Gresham Fab 12

Test Equipment in the Lab Agilent 4156C, 41501B Expander Box, E5250A Switch Matrix Cascade 12K Semi- Automatic probe station Temptronic Temp controller TP 03000, -50C to 200C. Metrics IC/V software package. 13

Test Equipment in the Lab HP4062 Parametric Test System (Agilent 4142 SMU) Electroglas 4080+ Prober Chuck Temp 25C to 130C. HP BASIC programming language 14

Threshold Voltage Extraction Techniques

Threshold Voltage Extraction Techniques Point of Maximum Slope in Id-Vg Curve (from strong inversion) Extrapolation of the Id-Vg curve from maximum gm in the linear region (LEID). Ratio Method: Y-Function (from strong inversion) Extrapolation of Id/ gm in the linear region. This method is considered to be unaffected by mobility degradation and series resistance effects. Subthreshold slope and pre-defined current (from weak inversion) Extrapolation from the subthreshold region to a pre-defined current using subthreshold slope. Constant Current Usually scaled by width and length in the form of Id0 x (W/L) with Id0 dependent on technology and specific device type. Transition Method (transition from weak to moderate and strong inversion) Integration-based: Construct an auxiliary function G1(Vg, Id) = Vg 2 Vg Id dvg. Id 0 G1 is zero if Id is linear with Vg (a measure of nonlinearity). With the transition of Id between diffusion current and drift current, the maximum value of G1 gives V T. 16

Illustration of V T Extraction from Maximum Slope (Gm) Sample title Id = (First W µ L eff Level) Cox Vd (V G -V T - 1 Vd) 2 Sample Description Text (second Level) Sample Bullet (third level) Id vs. V G Sample Bullet 2 (fourth level) o Sample Bullet 3 (fifth level) Gm vs. V G V T + 1 2 Vd 17

Illustration of V T Extraction from Ratio (Id/ gm) Method Sample title Id(First vs. Level) V gm G Sample Description Text (second Level) Sample Bullet (third level) Sample Bullet 2 (fourth level) o Sample Bullet 3 (fifth level) Consider R SD and µ eff in Id expression, and obtain Id gm = k 0Vd (V G -V T - 1 2 Vd) independent of R SD and µ eff where k 0 = W eff L eff µ 0 Cox V T + 1 2 Vd 18

Noise Effect in Id/ gm Ratio Method Id Sample title (First Level) gm vs. V G for NLDMOS devices with short integration time in HP4062 parametric tester Sample Bullet (third level) Sample Bullet 2 (fourth level) o Sample Bullet 3 (fifth level) Sample Description Text (second Level) Id gm vs. V G for NLDMOS devices with medium integration time in HP4062 parametric tester 19

Illustration of V T Extraction from Constant Current The choice of Id0: Id = Id0 x (W/L) Many use Id0 = 100nA. Id0 can be selected from LEID method: Id0= (Id at V T ) / (W/L) In this way, Id0 is not set arbitrarily, and becomes a Technology-Specific Scaled Current (TSSC). Id/(W/L) in log-scale vs. V G V T 20

Illustration of V T Extraction from Transition Method Sample title G1 vs. (First Id Level) Sample Description Text (second Level) Sample Bullet (third level) Sample Bullet 2 (fourth level) o Sample Bullet 3 (fifth level) Construct G1(Vg, Id) = Vg 2 Id 0 Vg Id dvg Id is composed of diffusion current and drift current with different dependencies on Vg. With the transition of Id from weak to moderate and strong inversion, maximum value of G1 gives V T. Integration-based extraction methods present no noise issue as seen in the graph on left. 5V PMOS W = 10 µm, L = 10,2,1.5,1,0.9,0.8,0.7,0.55 µm 21

More Threshold Voltage Extraction Methods Second Derivative (Transconductance Change) Vg at which second derivative of Id vs. Vg (or dgm/dvg) assumes its maximum. Is Dirac Delta function in Level 1 SPICE model: Id=0 for Vg<Vt, Id linear with Vg for Vg>V T. Data could be noisy with two derivatives. gm/id method with its maximum derivative to Vg (from gm/id methodology) Analytical modeling relates this to charge in inversion layer and it occurs when gm/id drops to 2 of its 3 maximum value when Vd approaches zero, and 0.72 of its maximum value when Vd 3~ 4(kT/q). gm/id method with the equality of diffusion current and drift current (from gm/id methodology) Shown by analytical modeling to occur when gm/id drops to 1 of its maximum when Vd is small 2 Point of Maximum Slope in Gm-Vg Curve (Not covered in this study) Extrapolation of the Gm-Vg curve from maximum dgm/dvg in the linear region. Model-based Methods (Not covered in this study) Adjusted Constant Current (ACC), Quasi-Constant Current (QCC), etc. 22

Illustration of V T Extraction with Second Derivative dgm dvg Sample title (First Level) vs. Vg Sample Description Text (second Level) d dvg gm Id vs. Vg Sample Bullet (third level) Sample Bullet 2 (fourth level) o Sample Bullet 3 (fifth level) Resolution of V T extraction is limited by Vg step size 23

Illustration of V T Extraction from gm/id Method Sample title (First (gm/id) Level) vs. V G Sample Description Text (second gmlevel) max Sample Bullet (third level) Id Sample Bullet 2 (fourth level) 0.72 max o Sample Bullet 3 (fifth level) gm Id (gm/id) vs. (Id/(W/L)) max gm Id 0.72 max gm Id 1 max gm 2 Id 1 max gm 2 Id TSSC V T Gm/Id curve covers device operation from weak to moderate and strong inversion. 24

Comparison of V T Extraction Results

Comparison of V T Extraction Results Sample title (First Level) Sample Description Text (second Level) Sample Bullet (third level) Sample Bullet 2 (fourth level) o Sample Bullet 3 (fifth level) 1.8V NMOS W = 10 µm with varying channel lengths. Id0 = 400nA. Vg step size 7.5mV 26

Comparison of V T Extraction Results Sample title (First Level) Sample Description Text (second Level) Sample Bullet (third level) Id0 = 400nA. Vg step size 7.5mV Sample Bullet 2 (fourth level) o Sample Bullet 3 (fifth level) 5V NMOS W = 10 µm with varying channel lengths. 27 Id0 = 300nA. Vg step size 15mV

Comparison of V T Extraction Results for LDMOS Sample title (First Level) Sample Description Text (second Level) Sample Bullet (third level) Sample Bullet 2 (fourth level) o Sample Bullet 3 (fifth level) 45V NLDMOS Id0 = 100nA. Vg step size 50mV 28

Comparison of Technology-Specific Scaled Current Sample title (First Level) Sample Description Text (second Level) Sample Bullet (third level) Sample Bullet 2 (fourth level) o Sample Bullet 3 (fifth level) 1.8V NMOS W = 10 µm with varying channel lengths. 5V NMOS W = 10 µm with varying channel lengths. 29

Discussion of V T Extraction Results Id/ gm ratio (Y-Function) method and transition method both produce V T extraction results in good agreement with V T from LEID method (slightly higher). V T extraction results from subthreshold slope and pre-defined Id is consistent with V T from LEID method (slightly lower). This method is valid only when there is no deviation of subthreshold Id-Vg characteristics from a straight line in semi-log plot (so is the Match Point method in literature). Technology-Specific Scaled Current (TSSC) can be selected from Id at V T from LEID method for each device category (e.g., 1.8V NMOS, 5V NMOS or 45V NLDMOS). Second derivative method produces V T extraction results higher than V T from LEID method. Gm/Id curve covers device operation from weak to moderate and strong inversion. V T from LEID method lies between the two predictions from gm/id methodology. 30

Comparison of V T Extraction Methods Extraction Method Speed Noise Faithfulness Notes LEID Slow Medium High Common practice Ratio Method Slow Medium High Subthreshold Slope and pre-defined current Medium Low Medium Requires calibration TSSC Fast Low High Requires calibration Transition Method Medium Low High Second Derivative Slow High Medium 0.72 (Gm/Id) max Medium Medium Medium 0.5 (Gm/Id) max Medium Medium Medium 31

Summary and Conclusion

Summary Aggressive threshold voltage (V T ) scaling faces challenge on energy efficiency since V T is related to OFF state leakage current. There is no single definition of threshold voltage, and many extraction techniques exist for this key device parameter. Practical threshold voltage extraction techniques tend to avoid noise caused by performing second (or even third) derivative. Several extraction methods produce results close to V T from LEID method, and TSSC can be selected from LEID method. 33

Test Insights from Lab to Fab Technology-Specific Scaled Current (TSSC) Id = Id0 x (W/L) Id0 can be selected from LEID method: Id0= (Id at V T ) / (W/L) Id0 is Technology-Specific Scaled Current (TSSC) that applies to all geometries of the same device category. Id0 information can be used to monitor V T shift in reliability tests. Id0 information can be provided from lab to fab to save time on V T measurements. 34

Acknowledgment Chris Kendrick Thomas Odenwalder Michael Cook 35

Thank You! For questions, please contact: Yu-Hsing Cheng ON Semiconductor Yu-Hsing.Cheng@onsemi.com 36