Fundamentals of Modern VLSI Devices 3rd Edition with Bookmarks Yuan Taur And Tak H. Ning 2024 Scribd Download

Download the full version of the ebook at ebookfinal.
com
Fundamentals of Modern VLSI Devices 3rd Edition

with Bookmarks Yuan Taur And Tak H. Ning
https://ebookfinal.com/download/fundamentals-of-modern-vlsi-
devices-3rd-edition-with-bookmarks-yuan-taur-and-tak-h-ning/
OR CLICK BUTTON
DOWNLOAD EBOOK
Download more ebook instantly today at https://ebookfinal.com

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...
Fundamentals of Multinational Finance 3rd Edition Michael

H. Moffett
https://ebookfinal.com/download/fundamentals-of-multinational-
finance-3rd-edition-michael-h-moffett/
ebookfinal.com
Fundamentals of Liquid Crystal Devices 2nd Edition Deng-Ke

Yang
https://ebookfinal.com/download/fundamentals-of-liquid-crystal-
devices-2nd-edition-deng-ke-yang/
ebookfinal.com
Modern VLSI Design IP Based Design 4th Edition Wayne Wolf
https://ebookfinal.com/download/modern-vlsi-design-ip-based-
design-4th-edition-wayne-wolf/
ebookfinal.com
ESD Circuits and Devices 1st Edition Steven H. Voldman
https://ebookfinal.com/download/esd-circuits-and-devices-1st-edition-
steven-h-voldman/
ebookfinal.com
VLSI Architectures for Modern Error Correcting Codes 1st
Edition Xinmiao Zhang
https://ebookfinal.com/download/vlsi-architectures-for-modern-error-
correcting-codes-1st-edition-xinmiao-zhang/
ebookfinal.com
Chemical Process Safety Fundamentals with Applications 3rd

Edition Daniel A. Crowl
https://ebookfinal.com/download/chemical-process-safety-fundamentals-
with-applications-3rd-edition-daniel-a-crowl/
ebookfinal.com
Twentieth century Chinese Translation Theory Tak
https://ebookfinal.com/download/twentieth-century-chinese-translation-
theory-tak/
ebookfinal.com
Fundamentals of Geometric Dimensioning and Tolerancing 3rd

Edition Krulikowski
https://ebookfinal.com/download/fundamentals-of-geometric-
dimensioning-and-tolerancing-3rd-edition-krulikowski/
ebookfinal.com
Fundamentals of Retiree Group Benefits 1st Edition Dale H.

Yamamoto
https://ebookfinal.com/download/fundamentals-of-retiree-group-
benefits-1st-edition-dale-h-yamamoto/
ebookfinal.com
Fundamentals of Modern VLSI Devices
Third Edition
A thoroughly updated third edition of a classic and widely adopted text, perfect for
practical transistor design and in the classroom. Covering a variety of recent develop-
ments, the internationally renowned authors discuss in detail the basic properties and
designs of modern VLSI devices, as well as factors affecting performance. Containing
around 25% new material, coverage has been expanded to include high-k gate
dielectrics, metal gate technology, strained silicon mobility, non-GCA (Gradual
Channel Approximation) modeling of MOSFETs, short-channel FinFETS, and sym-
metric lateral bipolar transistors on SOI. Chapters have been reorganized to integrate
the appendices into the main text to enable a smoother learning experience, and
numerous additional end-of-chapter homework exercises (+30%) are included to
engage students with real-world problems and test their understanding. A perfect text
for senior undergraduate and graduate students taking advanced semiconductor
devices courses, and for practicing silicon device professionals in the semiconductor
industry.
Yuan Taur is Distinguished Professor of Electrical and Computer Engineering at the

University of California, San Diego, having previously worked at IBM’s T. J. Watson
Research Center, New York. He is an IEEE fellow.
Tak H. Ning is IBM Fellow (retired) at the T. J. Watson Research Center, New York.
He is a fellow of the IEEE and the American Physical Society, and a member of the
US National Academy of Engineering.
Fundamentals of Modern
VLSI Devices
Third Edition
YUAN TAUR
University of California, San Diego
TAK H . N ING
IBM T. J. Watson Research Center (retired)
University Printing House, Cambridge CB2 8BS, United Kingdom
One Liberty Plaza, 20th Floor, New York, NY 10006, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre, New Delhi – 110025, India
103 Penang Road, #05–06/07, Visioncrest Commercial, Singapore 238467
Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.
www.cambridge.org
Information on this title: www.cambridge.org/highereducation/isbn/9781108480024
DOI: 10.1017/9781108847087
© Cambridge University Press 1998, 2009, 2022
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 1998
Second edition 2009
First paperback edition 2013
Eighth printing 2020
Third edition 2022
Printed in the United Kingdom by TJ Books Limited, Padstow, Cornwall
A catalogue record for this publication is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Names: Taur, Yuan, 1946–, author. | Ning, Tak H., 1943– author.
Title: Fundamentals of modern VLSI devices / Yuan Taur, University of California, San Diego,
Tak H. Ning, IBM Fellow (retired).
Description: Third edition. | United Kingdom ; New York : Cambridge University Press, 2022. |
Includes bibliographical references and index.
Identifiers: LCCN 2021029073 (print) | LCCN 2021029074 (ebook) | ISBN 9781108480024 (hardback) |
ISBN 9781108847087 (epub)
Subjects: LCSH: Metal oxide semiconductors, Complementary. | Bipolar transistors. | Integrated circuits–
Very large scale integration. | BISAC: TECHNOLOGY & ENGINEERING / Electronics /
Optoelectronics
Classification: LCC TK7871.99.M44 T38 2022 (print) | LCC TK7871.99.M44 (ebook) |
DDC 621.39/5–dc23
LC record available at https://lccn.loc.gov/2021029073
LC ebook record available at https://lccn.loc.gov/2021029074
ISBN 978-1-108-48002-4 Hardback
Additional resources for this publication at www.cambridge.org/taur3ed
Cambridge University Press has no responsibility for the persistence or accuracy
of URLs for external or third-party internet websites referred to in this publication
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
Contents
Preface to the Third Edition page xiii

Preface to the Second Edition xv
Preface to the First Edition xvii
Physical Constants and Unit Conversions xix
List of Symbols xx
1 Introduction 1
1.1 Evolution of VLSI Device Technology 1
1.1.1 Historical Perspective 1
1.1.2 Recent Developments 3
1.2 Scope and Brief Description of the Book 5
2 Basic Device Physics 9

2.1 Energy Bands in Silicon 9
2.1.1 Bandgap of Silicon 9
2.1.2 Density of States 10
2.1.3 Distribution Function: Fermi Level 12
2.1.4 Carrier Concentration 13
2.2 n-Type and p-Type Silicon 15
2.2.1 Donors and Acceptors 15
2.2.2 Fermi Level in Extrinsic Silicon 17
2.2.3 Degenerately Doped Silicon 20
2.3 Carrier Transport in Silicon 21
2.3.1 Drift Current: Mobility 22
2.3.2 Velocity Saturation 25
2.3.3 Diffusion Current 25
2.3.4 Einstein Relations 27
2.4 Basic Equations for Device Operation 28
2.4.1 Poisson’s Equation: Electrostatic Potential 28
2.4.2 Current–Density Equations 32
2.4.3 Generation and Recombination 35
2.4.4 Current Continuity Equations 38
Exercises 40
v
vi Contents
3 pn Junctions and MetalSilicon Contacts 43

3.1 p–n Junctions 43
3.1.1 Energy-Band Diagrams and Built-in Potential for
a p–n Diode 44
3.1.2 Depletion Approximation 45
3.1.3 Spatial Variation of Quasi-Fermi Potentials 53
3.1.4 The Diode Equation 62
3.1.5 Current–Voltage Characteristics Governed by the Diode
Equation 65
3.1.6 Space-Charge-Region Current 67
3.1.7 Measured Diode Current and Ideality Factor 70
3.1.8 Temperature Dependence and Magnitude of Diode Leakage
Currents 71
3.1.9 Minority-Carrier Mobility, Lifetime, and Diffusion Length 72
3.2 Metal–Silicon Contacts 74
3.2.1 Static Characteristics of a Schottky Diode 74
3.2.2 Current–Voltage Characteristics of a Schottky Diode 82
3.2.3 Ohmic Contacts 87
3.3 High-Field Effects in Reverse-Biased Diodes 89
3.3.1 Impact Ionization and Avalanche Breakdown 90
3.3.2 Band-to-Band Tunneling 93
Exercises 94
4 MOS Capacitors 99
4.1 Energy Band Diagram of an MOS System 99
4.1.1 Free Electron Level, Work Function, and Flatband Voltage 99
4.1.2 Gate Voltage, Surface Potential, and Charge in Silicon 102
4.1.3 Accumulation, Depletion, and Inversion 103
4.2 Electrostatic Potential and Charge Distribution in Silicon 106
4.2.1 Solving Poisson’s Equation 106
4.2.2 Surface Potential and Charge Density as a Function of Gate
Voltage 112
4.3 Capacitance–Voltage Characteristics of MOS Capacitors 114
4.3.1 Measurement Setup 114
4.3.2 Capacitance Components in MOS 114
4.3.3 C–V Characteristics in Different Bias Regions 115
4.3.4 Split C–V Measurement 119
4.3.5 Polysilicon Gate: Work Function and Depletion Effects 121
4.3.6 MOS under Nonequilibrium 125
4.4 Quantum Mechanical Effects in MOS 129
4.4.1 Coupled Poisson–Schrodinger’s Equations 129
4.4.2 Quantum Effect on Inversion-Layer Depth 129
Contents vii
4.4.3 Quantum-Mechanical Solution in Weak Inversion 131

4.5 Interface States and Charge Traps in Oxide 136
4.5.1 Effect of Oxide Charge on Flatband Voltage 137
4.5.2 Interface–State Capacitance and Conductance 138
4.5.3 Distributed Circuit Model for Oxide Traps 147
4.6 High-Field Effects in Oxide and Oxide Degradation 149
4.6.1 Tunneling into and through Silicon Dioxide 149
4.6.2 Injection of Hot Carriers from Silicon into Silicon Dioxide 158
4.6.3 High-Field Effects in Gated Diodes 160
4.6.4 Dielectric Breakdown 162
Exercises 167
5 MOSFETs: Long Channel 171

5.1 MOSFET I–V Characteristics 172
5.1.1 Gradual Channel Approximation 173
5.1.2 Charge Sheet Model 176
5.1.3 Regional I–V Models 178
5.1.4 Non-GCA Model for the Saturation Region 187
5.1.5 pMOSFET I–V Characteristics 192
5.2 MOSFET Channel Mobility 192
5.2.1 Empirical Universal Mobility 192
5.2.2 Strain Effect on Mobility 196
5.3 MOSFET Threshold Voltage 198
5.3.1 Substrate Sensitivity (Body Effect) 198
5.3.2 Temperature Dependence of Threshold Voltage 199
5.3.3 Quantum Effect on Threshold Voltage 201
5.4 MOSFET Capacitance 202
Exercises 203
6 MOSFETs: Short Channel 206

6.1 Short-Channel Effect 206
6.1.1 Threshold Voltage Roll-off 206
6.1.2 Analytic Solutions to 2-D Poisson’s Equation in
Subthreshold 210
6.2 High-Field Transport 219
6.2.1 Velocity Saturation 219
6.2.2 Nonlocal Transport 229
6.3 MOSFET Threshold Voltage and Channel Profile Design 236
6.3.1 Threshold Voltage Requirement 237
6.3.2 Channel Profile Design 241
6.3.3 Nonuniform Channel Doping 246
6.3.4 Discrete Dopant Effects on Threshold Voltage 253
viii Contents
6.4 MOSFET Degradation and Breakdown at High Fields 256

6.4.1 Hot-Carrier Effects 257
6.4.2 Negative-Bias-Temperature Instability 259
6.4.3 MOSFET Breakdown 260
Exercises 262
7 Silicon-on-Insulator and Double-Gate MOSFETs 264

7.1 SOI MOSFETs 265
7.1.1 Long-Channel SOI MOSFETs 265
7.1.2 Short-Channel SOI MOSFETs 271
7.2 Double-Gate and Nanowire MOSFETs 276
7.2.1 Analytic Potential Model for Symmetric DG MOSFETs 277
7.2.2 Short-Channel DG MOSFETs 281
7.2.3 Nanowire MOSFETs 287
7.2.4 Scaling Limits of DG and Nanowire MOSFETs 291
Exercises 292
8 CMOS Performance Factors 295

8.1 MOSFET Scaling 295
8.1.1 Constant-Field Scaling 295
8.1.2 Nonscaling Factors 297
8.2 Basic CMOS Circuit Elements 298
8.2.1 CMOS Inverters 299
8.2.2 CMOS NAND and NOR Gates 309
8.2.3 Inverter and NAND Layouts 313
8.3 Parasitic Elements 316
8.3.1 Source–Drain Resistance 317
8.3.2 Parasitic Capacitances 321
8.3.3 Gate Resistance 324
8.3.4 Interconnect R and C 326
8.4 Sensitivity of CMOS Delay to Device Parameters 332
8.4.1 Propagation Delay and Delay Equation 333
8.4.2 Delay Sensitivity to Channel Width, Length, and Gate Oxide
Thickness 339
8.4.3 Sensitivity of Delay to Power-Supply Voltage and Threshold
Voltage 343
8.4.4 Sensitivity of Delay to Parasitic Resistance and Capacitance 344
8.4.5 Effect of Transport Parameters on CMOS Delay 348
8.4.6 Delay of Two-Way NAND Gates 349
8.5 Performance Factors of MOSFETs in RF Circuits 352
8.5.1 Small-Signal Equivalent Circuit 353
8.5.2 Unity-Current-Gain Frequency 354
8.5.3 Power Gain Condition of a Two-Port Network 354
8.5.4 Unity Power-Gain Frequency 355
Exercises 358
Contents ix
9 Bipolar Devices 361

9.1 Basic Operation of a Bipolar Transistor 365
9.1.1 Modifying the Simple Diode Theory for Describing Bipolar
Transistors 365
9.2 Ideal Current–Voltage Characteristics 370
9.2.1 Intrinsic-Base Resistance and Emitter Current Crowding 371
9.2.2 Collector Current 375
9.2.3 Base Current 378
9.2.4 Current Gains 382
9.2.5 Ideal IC–VCE Characteristics 384
9.3 Measured Characteristics of Typical n–p–n Transistors 385
9.3.1 Effect of Emitter and Base Series Resistances 386
9.3.2 Effect of Base–Collector Voltage on Collector Current 389
9.3.3 Collector-Current Falloff 392
9.3.4 Excess Base Current Associated with Extrinsic-BaseEmitter
Junction 396
9.4 Base Transit Time 400
9.5 Diffusion Capacitance in an EmitterBase Diode 401
9.5.1 Small-Signal Current in a Forward-Biased Diode 401
9.5.2 Low-Frequency [ωτpE < 1 and ωtB < 1] Diffusion
Capacitance 405
9.5.3 Diffusion Capacitance at High Frequencies [ωτpE > 1] 406
9.6 Bipolar Device Models for Circuit Analyses 407
9.6.1 Basic Steady-State Model 407
9.6.2 Basic ac Model 409
9.7 Breakdown Voltages 416
9.7.1 Common-Base Current Gain in the Presence of Base–Collector
Junction Avalanche 417
9.7.2 Saturation Currents in a Transistor 418
9.7.3 Relation between BVCEO and BVCBO 419
9.7.4 Breakdown Voltages of Symmetric Lateral Bipolar Transistors
on SOI 421
Exercises 421
10 Bipolar Device Design 425

10.1 Design of the Emitter of a Vertical Bipolar Transistor 425
10.1.1 Diffused or Implanted-and-Diffused Emitter 426
10.1.2 Polysilicon Emitter 427
10.2 Design of the Base Region of a Vertical Bipolar Transistor 427
10.2.1 Base Sheet Resistivity and Collector Current Density 429
10.2.2 Ion-Implanted versus Epitaxially Grown Intrinsic Base 430
10.2.3 General Expression for Base Transit Time 433
10.3 Design of the Vertical Bipolar Transistor Collector Region 434
10.3.1 Collector Design for Low-Injection Operation 435
10.3.2 Collector Design for High-Injection Operation 436
x Contents
10.4 SiGe-Base Vertical Bipolar Transistors 437

10.4.1 SiGe-Base Vertical Transistors Having Linearly Graded Base
Bandgap 438
10.4.2 Base Current When Ge Is Present in the Emitter 443
10.4.3 Transistors Having a Trapezoidal Ge Distribution in the
Base 447
10.4.4 Transistors Having a Constant Ge Distribution in the Base 451
10.4.5 Some Optimal Ge Profiles 454
10.4.6 Base-Width Modulation by VBE 459
10.4.7 Reverse-Mode IV Characteristics 462
10.4.8 Heterojunction Nature of a SiGe-Base Vertical Bipolar
Transistor 465
10.4.9 SiGe-Base Vertical Bipolar Transistor on Thin SOI 467
10.5 Design of Symmetric Lateral Bipolar Transistors on SOI 468
10.5.1 Relationship Governing Emitter-to-Collector Spacing
and Base Width 470
10.5.2 Analytic Model for Collector and Base Currents 471
10.5.3 Analytic Ebers-Moll Model Equations 473
10.5.4 Early Voltage and EmitterCollector Spacing 475
10.5.5 Analytic Model for the Transit Times 475
10.5.6 On the Fabrication of Thin-Base Symmetric Lateral
Transistors 476
10.5.7 SiGe-on-Insulator Symmetric lateral n–p–n Transistors 477
10.5.8 Symmetric Si-Emitter/Collector SiGe-Base Lateral HBT 478
Exercises 479
11 Bipolar Performance Factors 485

11.1 Figures of Merit of a Bipolar Transistor 485
11.1.1 Cutoff Frequency 486
11.1.2 Maximum Oscillation Frequency 488
11.1.3 Logic Gate Delay 489
11.2 ECL Circuit and Delay Components 489
11.2.1 Transit-Time Delay Component 491
11.2.2 Intrinsic-Base-Resistance Delay Component 492
11.2.3 Parasitic-Resistance Delay Components 492
11.2.4 Load-Resistance Delay Component 492
11.2.5 Diffusion-Capacitance Delay Component 493
11.3 Speed-versus-Current Characteristics of Bipolar Transistors 493
11.3.1 fT and fmax as a Function of Collector Current 493
11.3.2 Logic Gate Delay as a Function of Collector Current 495
11.4 Vertical-Transistor Optimization from Data Analyses 496
11.5 Bipolar Device Scaling for Logic Circuits 498
11.5.1 Vertical-Transistor Scaling for ECL 498
11.5.2 Symmetric-Lateral-Transistor Scaling for Logic Circuits 499
Contents xi
11.5.3 Power-Dissipation Issues with Resister-Load Bipolar Logic

Circuits 500
11.6 Vertical-Transistor Design Optimization for RF and Analog Circuits 502
11.6.1 The Single-Transistor Amplifier 502
11.6.2 Maximizing fT of a Vertical Transistor 503
11.6.3 Minimizing rbi of a Vertical Transistor 504
11.6.4 Maximizing fmax of a Vertical Transistor 505
11.6.5 Maximizing VA of a Vertical Transistor 505
11.6.6 Examples of Vertical-Transistor RF and Analog Design
Tradeoffs 505
11.7 Symmetric-Lateral-Transistor Design Tradeoffs and Optimization
for RF and Analog Circuits 507
11.7.1 Calculated Low-Injection fT and fmax of Symmetric
Lateral npn 507
11.7.2 Fin-Structure Symmetric Lateral Transistors for
fmax > 1 THz 509
11.7.3 Noise Reduction with Substrate Bias 510
11.8 Unique Opportunities from Symmetric Lateral Bipolar Transistors 511
11.8.1 Symmetric Lateral Bipolar Transistor as a High-Drive-Current
Device 511
11.8.2 Revisit Integrated Injection Logic Circuits and SRAM 513
11.8.3 Complementary Bipolar Logic Circuits 514
11.8.4 Performance-On-Demand Designs with I2L or CBipolar
Circuits 516
Exercises 517
12 Memory Devices 521

12.1 Static Random-Access Memory 523
12.1.1 CMOS SRAM Cell 523
12.1.2 Other Bistable MOSFET SRAM Cells 532
12.1.3 Bipolar SRAM Cell 533
12.2 Dynamic Random-Access Memory 541
12.2.1 Basic DRAM Cell and Its Operation 541
12.2.2 Device Design and Scaling Considerations for a DRAM Cell 545
12.3 Nonvolatile Memory 546
12.3.1 MOSFET Nonvolatile Memory Devices 547
12.3.2 Flash Memory Arrays 554
12.3.3 Devices for a NOR Array 559
Exercise 564
References 565
Index 587
Preface to the Third Edition
It has been twenty-four years since the first edition of this book was published, thirteen
years since the second edition. Both editions have been translated into Japanese. The
second edition has also been translated into Chinese. During this time period, the
VLSI integrated-circuit industry marches on. The minimum feature size has shrunk by
more than 10 to now reaching <10 nm. The transistor count has well exceeded ten
billion in the densest populated IC chips. New technology thrust areas include double-
gate MOSFETs (known as FinFETs), thin-film silicon-on-insulator devices (ET-SOI),
and 3-D integration in nonvolatile memory chips. Meanwhile, SOI wafers suitable for
partially depleted SOI CMOS have found new applications in the integration of
npn and pnp lateral bipolar, offering the interesting possibility of bipolar as
a complement to CMOS for VLSI.
The purpose of writing the third edition is to update the book with additional
material developed after the completion of the second edition. Key topics added
include high-κ gate dielectrics, metal gate technology, strained silicon mobility,
non-GCA (Gradual Channel Approximation) modeling of MOSFETs, and lateral
bipolar transistors on SOI. Furthermore, the chapters of the book have been reorgan-
ized to consolidate the discussions of the various subjects to the main chapters with
no appendices.
We would like to take this opportunity to thank all the friends and colleagues who
gave us encouragement and valuable suggestions for improvement of the book; in
particular, Professor Sorin Cristoloveanu of the University of Grenoble Alpes and Dr.
Kangguo Cheng of IBM on SOI device technology, and Professor Scott Thompson of
the University of Florida on strained silicon mobility.
Tak Ning would like to thank many of his colleagues at IBM, particularly Dr. Jin
Cai (now at TSMC), Dr. Jeng-Bang Yau, and Dr. Ghavam Shahidi for their contribu-
tions to the SOI lateral bipolar project. Yuan Taur would like to thank many of his
students at the University of California, San Diego, in particular Chuyang Hong, Qian
Xie, and Bo Yu, for their help with the completion of the third edition. He would also
like to thank Katie and Ko Taur for their love and support during the course of
the work.
xiii
Preface to the Second Edition
Since the publication of the first edition of Fundamentals of Modern VLSI Devices by
Cambridge University Press in 1998, we received much praise and many encouraging
reviews on the book. It has been adopted as a textbook for first-year graduate courses
on microelectronics in many major universities in the United States and worldwide.
The first edition was translated into Japanese by a team led by Professor Shibahara of
Hiroshima University in 2002.
During the past ten years, the evolution and scaling of VLSI (very-large-scale-
integration) technology has continued. Now, sixty years after the first invention of the
transistor, the number of transistors per chip for both microprocessors and DRAM
(dynamic random access memory) has increased to over one billion, and the highest
clock frequency of microprocessors has reached 5 GHz. In 2007, the worldwide IC
(integrated circuits) sales grew to $250 billion. In 2008, the IC industry reached the
45-nm generation, meaning that the leading-edge IC products employ a minimum
lithography feature size of 45 nm. As bulk CMOS (complementary metal–oxide–
semiconductor field-effect transistor) technologies are scaled to dimensions below
100 nm, the very factor that makes CMOS technology the technology of choice for
digital VLSI circuits, namely, its low standby power, can no longer be taken for
granted. Not only has the off-state current gone up with the power supply voltage
down scaled to the 1V level, the gate leakage has also increased exponentially from
quantum mechanical tunneling through gate oxides only a few atomic layers thick.
Power management, both active and standby, has become a key challenge to con-
tinued increase of clock frequency and transistor count in microprocessors. New
materials and device structures are being explored to replace conventional bulk
CMOS in order to extend scaling to 10 nm.
The purpose of writing the second edition is to update the book with additional
material developed after the completion of the first edition. Key new material added
includes MOSFET scale length theory and high-field transport model, and the section
on SiGe-base bipolar devices has been greatly expanded. We have also expanded the
discussions on basic device physics and circuits to include metal–silicon contacts,
noise margin of CMOS circuits, and figures of merit for RF applications. Furthermore,
two new chapters are added to the second edition. Chapter 9 is on memory devices and
covers the fundamentals of read and write operations of commonly used SRAM,
DRAM, and nonvolatile memory arrays. Chapter 10 is on silicon-on-insulator (SOI)
devices, including advanced devices of future potential.
xv
xvi Preface to the Second Edition
We would like to take this opportunity to thank all the friends and colleagues who
gave us encouragement and valuable suggestions for improvement of the book. In
particular, Professor Mark Lundstrom of Purdue University, who adopted the first
edition early on, and Dr. Constantin Bulucea of the National Semiconductor
Corporation, who suggested the treatment on diffusion capacitance. Thanks also go
to Professor James Meindl of the Georgia Institute of Technology, Professor Peter
Asbeck of the University of California, San Diego, and Professor Jerry Fossum of the
University of Florida for their support of the book.
We would like to thank many of our colleagues at IBM, particularly in the areas of
advanced silicon-device research and development, for their direct or indirect contri-
butions. Yuan Taur would like to thank many of his students at the University of
California, San Diego, in particular Jooyoung Song and Bo Yu, for their help with the
completion of the second edition. He would also like to thank Katie Kahng for her
love, support, and patience during the course of the work.
We would like to give special thanks to our families for their support and under-
standing during this seemingly endless task.
Preface to the First Edition
It has been fifty years since the invention of the bipolar transistor, more than forty
years since the invention of integrated-circuit (IC) technology, and more than thirty-
five years since the invention of the MOSFET. During this time, there has been
tremendous and steady progress in the development of IC technology with a rapid
expansion of the IC industry. One distinct characteristic in the evolution of IC
technology is that the physical feature sizes of the transistors are reduced continually
over time as the lithography technologies used to define these features become
available. For almost thirty years now, the minimum lithography feature size used
in IC manufacturing has been reduced at a rate of 0.7 every three years. In 1997, the
leading-edge IC products have a minimum feature size of 0.25 μm.
The basic operating principles of large and small transistors are the same. However,
the relative importance of the various device parameters and performance factors for
transistors of the 1 μm and smaller generations is quite different from those for
transistors of larger-dimension generations. For example, in the case of CMOS, the
power-supply voltage was lowered from the standard 5 V, starting with the 0.6 to 0.8
μm generation. Since then CMOS power supply voltage has been lowered in steps
once every few years as the device physical dimensions are reduced. At the same time,
many physical phenomena, such as short-channel effect and velocity saturation, which
are negligible in large-dimension MOSFETs, are becoming more and more important
in determining the behavior of MOSFETs of deep-submicron dimensions. In the case
of bipolar devices, breakdown voltage and base-widening effects are limiting their
performance, and power dissipation is limiting their level of integration on a chip.
Also, the advent of SiGe-base bipolar technology has extended the frequency capabil-
ity of small-dimension bipolar transistors into the range previously reserved for GaAs
and other compound-semiconductor devices.
The purpose of this book is to bring together the device fundamentals that govern
the behavior of CMOS and bipolar transistors into a single text, with emphasis on
those parameters and performance factors that are particularly important for VLSI
(very-large-scale-integration) devices of deep-submicron dimensions. The book starts
with a comprehensive review of the properties of the silicon material, and the basic
physics of p–n junctions and MOS capacitors, as they relate to the fundamental
principles of MOSFET and bipolar transistors. From there, the basic operation of
MOSFET and bipolar devices, and their design and optimization for VLSI applica-
tions are developed. A great deal of the volume is devoted to in-depth discussions of
xvii
xviii Preface to the First Edition
the intricate interdependence and subtle tradeoffs of the various device parameters
pertaining to circuit performance and manufacturability. The effects which are par-
ticularly important in small-dimension devices, e.g., quantization of the two-
dimensional surface inversion layer in a MOSFET device and the heavy-doping effect
in the intrinsic base of a bipolar transistor, are covered in detail. Also included in this
book are extensive discussions on scaling and limitations to scaling of MOSFET and
bipolar devices.
This book is suitable for use as a textbook by senior undergraduate or graduate
students in electrical engineering and microelectronics. The necessary background
assumed is an introductory understanding of solid-state physics and semiconductor
physics. For practicing engineers and scientists actively involved in research and
development in the IC industry, this book serves as a reference in providing a body
of knowledge in modern VLSI devices for them to stay up to date in this field.
VLSI devices are too huge a subject area to cover thoroughly in one book. We have
chosen to cover only the fundamentals necessary for discussing the design and
optimization of the state-of-the-art CMOS and bipolar devices in the sub-0.5-μm
regime. Even then, the specific topics covered in this book are based on our own
experience of what the most important device parameters and performance factors are
in modern VLSI devices.
Many people have contributed directly and indirectly to the topics covered in this
book. We have benefited enormously from the years of collaboration and interaction
we had with our colleagues at IBM, particularly in the areas of advanced silicon-
device research and development. These include Douglas Buchanan, Hu Chao, T. C.
Chen, Wei Chen, Kent Chuang, Peter Cook, Emmanuel Crabbé, John Cressler, Bijan
Davari, Robert Dennard, Max Fischetti, David Frank, Charles Hsu, Genda Hu,
Randall Isaac, Khalid Ismail, G. P. Li, Shih-Hsien Lo, Yuh-Jier Mii, Edward
Nowak, George Sai-Halasz, Stanley Schuster, Paul Solomon, Hans Stork, Jack Sun,
Denny Tang, Lewis Terman, Clement Wann, James Warnock, Siegfried Wiedmann,
Philip Wong, Matthew Wordeman, Ben Wu, and Hwa Yu.
We would like to acknowledge the secretarial support of Barbara Grady and the
support of our management at IBM Thomas J. Watson Research Center where this
book was written. Finally, we would like to give special thanks to our families –
Teresa, Adrienne, and Brenda Ning, and Betty, Ying, and Hsuan Taur – for their
support and understanding during this seemingly endless task.
Physical Constants and Unit Conversions
Description Symbol Value and unit
Electronic charge q 1.6 1019 C

Boltzmann’s constant k 1.38 1023 J/K
Vacuum permittivity ε0 8.85 1014 F/cm
Silicon permittivity εsi 1.04 1012 F/cm
Oxide permittivity εox 3.45 1013 F/cm
Velocity of light in vacuum c 3 1010 cm/s
Planck’s constant h 6.63 1034 J-s
Free-electron mass m0 9.1 1031 kg
Thermal voltage (T ¼ 300 K) kT/q 0.0259 V
Angstrom Å 1Å ¼ 108 cm
Nanometer nm 1 nm ¼ 107 cm
Micrometer (micron) μm 1 μm ¼ 104 cm
Millimeter mm 1 mm ¼ 0.1 cm
Meter m 1 m ¼ 102 cm
Electron-volt eV 1eV ¼ 1.6 1019 J
Energy ¼ charge voltage E ¼ qV Joule ¼ Coulomb Volt

Charge ¼ capacitance voltage Q ¼ CV Coulomb ¼ Farad Volt
Power ¼ current voltage P ¼ IV Watt ¼ Ampere Volt
Time ¼ resistance capacitance t ¼ RC second ¼ Ω (ohm) Farad
Current ¼ charge/time I ¼ Q/t Ampere ¼ Coulomb/second
Resistance ¼ voltage/current R ¼ V/I Ω (ohm) ¼ Volt/Ampere
A word of caution about the length units: Strictly speaking, MKS units should be used
for all the equations in the book. As a matter of convention, electronics engineers often
work with centimeters as the unit of length. While some equations work with lengths
in either meters or centimeters, not all of them do. It is prudent always to check for
unit consistency when doing calculations. It may be necessary to convert the length
unit to meters before plugging into the equations.
xix
Symbols
Symbol Description Unit
A Area cm2
AE Emitter area cm2
α Common-base current gain None
α0 Static common-base current gain None
αF Forward common-base current gain in the Ebers–Moll model None
αR Reverse common-base current gain in the Ebers–Moll model None
αT Base transport factor None
αn Electron-initiated rate of electron–hole pair generation per unit cm1
distance
αp Hole-initiated rate of electron–hole pair generation per unit distance cm1
BV Breakdown voltage V
BVCBO Collector–base junction breakdown voltage with emitter open circuit V
BVCEO Collector–emitter breakdown voltage with base open circuit V
BVEBO Emitter–base junction breakdown voltage with collector open circuit V
β Current gain None
β0 Static common-emitter current gain None
βF Forward common-emitter current gain in the Ebers–Moll model None
βR Reverse common-emitter current gain in the Ebers–Moll model None
c Velocity of light in vacuum (= 3 1010 cm/s) cm/s
C Capacitance F
Cd Depletion-layer capacitance per unit area F/cm2
Cd MOS gate depletion-layer capacitance per unit area F/cm2
CdBC Base–collector diode depletion-layer capacitance per unit area F/cm2
CdBC,tot Total base–collector diode depletion-layer capacitance F
CdBE Base–emitter diode depletion-layer capacitance per unit area F/cm2
CdBE,tot Total base–emitter diode depletion-layer capacitance F
Cdm Maximum MOS gate depletion capacitance per unit area F/cm2
CD Diffusion capacitance F
CD MOS gate depletion-layer capacitance F
CDn Diffusion capacitance due to excess electrons F
CDp Diffusion capacitance due to excess holes F
CDE Emitter diffusion capacitance F
Cfb MOS capacitance at flat band per unit area F/cm2
CFC Capacitance between the floating gate and the control gate of a F
MOSFET nonvolatile memory device
Cg Intrinsic gate capacitance per unit area F/cm2
CG Total gate capacitance of MOSFET F
Cinv MOS capacitance per unit area in the inversion region F/cm2
Ci Inversion-layer capacitance per unit area F/cm2
xx
List of Symbols xxi
(cont.)
Cit Interface trap capacitance per unit area F/cm2

CJ Junction capacitance F
CL Load capacitance F
Cin Equivalent input capacitance of a logic gate F
Cinv MOSFET capacitance in inversion per unit area F/cm2
Cmin Minimum MOS capacitance per unit area F/cm2
Cout Equivalent output capacitance of a logic gate F
Cov Gate-to-source (-drain) overlap capacitance (per edge) F
Cox Oxide capacitance per unit area F/cm2
Cp Polysilicon-gate depletion-layer capacitance per unit area F/cm2
Csi Silicon capacitance per unit area F/cm2
Cw Wire capacitance per unit length F/cm
Cπ Base–emitter capacitance in the small-signal hybrid-π F
equivalent-circuit model
Cμ Base–collector capacitance in the small-signal hybrid-π F
equivalent-circuit model
d Width of diffusion region in a MOSFET cm
dsi Depth in silicon in non-GCA model cm
Dit Interface-state density per unit area per unit energy 1/cm2-eV
Dn Electron diffusion coefficient cm2/s
DnB Electron diffusion coefficient in the base of an n–p–n transistor cm2/s
Dp Hole diffusion coefficient cm2/s
DpE Hole diffusion coefficient in the emitter of an n–p–n transistor cm2/s
ΔVt Threshold voltage rolloff due to short-channel effect V
ΔEg Apparent bandgap narrowing J
ΔEgB Bandgap-narrowing parameter in the base region J
ΔEg,max Maximum bandgap narrowing due to the presence of Ge J
ΔEg,SiGe Local bandgap narrowing due to the presence of Ge J
ΔL Channel length modulation in MOSFET cm
ΔQtotal Total charge stored in a nonvolatile memory device C
E Energy J
Ec Conduction-band edge J
Ev Valence-band edge J
Ea Ionized-acceptor energy level J
Ed Ionized-donor energy level J
Ef Fermi energy level J
Eg Energy gap of silicon J
EgB Energy gap of base region of a bipolar transistor J
EgE Energy gap of emitter region of a bipolar transistor J
Ei Intrinsic Fermi level J
Efn Electron quasi-Fermi level J
Efp Hole quasi-Fermi level J
E Electric field V/cm
Ec Critical field for velocity saturation V/cm
Eeff Effective vertical field in MOSFET V/cm
Eox Oxide electric field V/cm
Es Electric field at silicon surface V/cm
xxii List of Symbols
(cont.)
Ex Vertical field in silicon V/cm

Ey Lateral field in silicon V/cm
ε0 Vacuum permittivity (¼ 8.85 1014 F/cm) F/cm
εi Permittivity of gate insulator F/cm
εBOX Permittivity of buried oxide layer in SOI F/cm
εsi Silicon permittivity (¼ 1.04 1012 F/cm) F/cm
εox Oxide permittivity (¼ 3.45 1013 F/cm) F/cm
fD Probability that an electronic state is filled None
f Frequency, clock frequency Hz
fmax Unity power gain frequency (also called maximum oscillation Hz
frequency)
fT Unity current gain frequency (also called cutoff frequency) Hz
FI Fan-in None
FO Fan-out None
ϕ Barrier height V
ϕbg Work function of the back gate in SOI V
ϕox Silicon–silicon dioxide interface potential barrier for electrons V
ϕms Work-function difference between metal and silicon V
ϕn Electron quasi-Fermi potential V
ϕp Hole quasi-Fermi potential V
ϕBn Schottky barrier height for electrons V
ϕBp Schottky barrier height for holes V
g Number of degeneracy None
gds Small-signal output conductance per unit width A/V-cm
gm Small-signal transconductance per unit width A/V-cm
GE Emitter Gummel number s/cm4
GB Base Gummel number s/cm4
Gd Equivalent conductance per unit area in MOS inversion 1/Ω-cm2
Gn Electron emission rate (also called electron generation rate) 1/cm3-s
Gp Hole emission rate (also called hole generation rate) 1/cm3-s
γ Emitter injection efficiency None
h Planck’s constant (¼ 6.63 1034 J-s) J-s
i Time-dependent current A
iB Time-dependent base current in a bipolar transistor A
ib Time-dependent small-signal base current A
iC Time-dependent collector current in a bipolar transistor A
ic Time-dependent small-signal collector current A
iE Time-dependent emitter current in a bipolar transistor A
I Current A
IB Static base current in a bipolar transistor A
IC Static collector current in a bipolar transistor A
IE Static emitter current in a bipolar transistor A
IS Switch current in an ECL circuit A
Ig Gate current in a MOSFET A
I0 MOSFET current per unit width to length ratio for threshold A
definition
Idsat MOSFET saturation current A
Ion MOSFET on current A
List of Symbols xxiii
(cont.)
Ioff MOSFET off current A

IN/w nMOSFET current per unit width A/cm
IP/w pMOSFET current per unit width A/cm
IN nMOSFET current A
IP pMOSFET current A
Ids Drain-to-source current in a MOSFET A
Isx Substrate current in a MOSFET A
Ids,Vt MOSFET current at threshold A
IonN/w nMOSFET on current per device width A/cm
IonN nMOSFET on current A
IonP/w pMOSFET on current per device width A/cm
IonP pMOSFET on current A
λ MOSFET scale length cm
J Current density A/cm2
JB Base current density A/cm2
JC Collector current density A/cm2
JCF Collector current density injected from emitter A/cm2
JCR Collector current density injected from collector A/cm2
Jn Electron current density A/cm2
Jp Hole current density A/cm2
k Boltzmann’s constant (= 1.38 1023 J/K) J/K
κ Scaling factor (>1) None
l Mean free path cm
L Length, MOSFET channel length cm
LD Debye length cm
LE Emitter-stripe length of a bipolar transistor cm
Ln Electron diffusion length cm
Lp Hole diffusion length cm
Lmin Minimum channel length of MOSFET cm
Lw Wire length cm
m MOSFET body-effect coefficient None
m Ideality factor of current in a Gummel plot None
m0 Free-electron mass (= 9.1 1031 kg) kg
m* Electron effective mass kg
M Avalanche multiplication factor None
ml Electron effective mass in the longitudinal direction kg
mt Electron effective mass in the transverse direction kg
μ Carrier mobility cm2/V-s
μeff Effective mobility cm2/V-s
μn Electron mobility cm2/V-s
μp Hole mobility ccm2/V-s
n Density of free electrons cm3
n0 Density of free electrons at thermal equilibrium cm3
ni Intrinsic carrier density cm3
nie Effective intrinsic carrier density cm3
nieB Effective intrinsic carrier density in base of bipolar transistor cm3
nieE Effective intrinsic carrier density in emitter of bipolar transistor cm3
nn Density of electrons in n-region cm3
xxiv List of Symbols
(cont.)
np Density of electrons in p-region cm3

Na Acceptor impurity density cm3
Nd Donor impurity density cm3
Nb Impurity concentration in bulk silicon cm3
Nbt Density of oxide traps per volume per energy cm3 eV1
Nc Effective density of states of conduction band cm3
Np Doping density of polysilicon gate cm3
Nv Effective density of states of valence band cm3
NB Base doping concentration cm3
NC Collector doping concentration cm3
NE Emitter doping concentration cm3
N(E) Density of electronic states per unit energy per volume 1/J-m3
p Density of free holes cm3
p0 Density of free holes at thermal equilibrium cm3
pn Density of holes in n-region cm3
pp Density of holes in p-region cm3
P Power dissipation W
Pac Active power dissipation W
Poff Standby power dissipation W
q Electronic charge (= 1.6 1019 C) C
Q Charge C
QB Excess minority charge per unit area in the base C/cm2
QB,tot Total excess minority charge in the base C
QBE Excess minority charge per unit area in the base–emitter space-charge C/cm2
region
QBE,tot Total excess minority charge in the base–emitter space-charge region C
QBC Excess minority charge per unit area in the base–collector space- C/cm2
charge region
QBC,tot Total excess minority charge in the base–collector space-charge C
region
QDE Total stored minority-carrier charge in a bipolar transistor biased in C
the forward-active mode
QE Excess minority charge per unit area in the emitter C/cm2
QE,tot Total excess minority charge in the emitter C
QpB Hole charge per unit area in base of n–p–n transistor C/cm2
Qs Total charge per unit area in silicon C/cm2
Qd Depletion charge per unit area C/cm2
Qi Inversion charge per unit area C/cm2
Qg Charge on MOS gate per unit area C/cm2
Qt Interface-state trapped charge per unit area C/cm2
Qox Equivalent oxide charge density per unit area C/cm2
r, R Resistance Ω
rb Base resistance Ω
rbi Intrinsic base resistance Ω
rbx Extrinsic base resistance Ω
rc Collector series resistance Ω
re Emitter series resistance Ω
List of Symbols xxv
(cont.)
r0 Output resistance in small-signal hybrid-π equivalent-circuit model Ω

rπ Input resistance in small-signal hybrid-π equivalent-circuit model Ω
RL Load resistance in a circuit Ω
Rs Source series resistance Ω
Rd Drain series resistance Ω
Rn Electron capture rate (also called electron recombination rate) 1/cm3-s
Rp Hole capture rate (also called hole recombination rate) 1/cm3-s
Rsd Source–drain series resistance Ω
Rch MOSFET channel resistance Ω
Rw Wire resistance per unit length Ω/cm
RSbi Sheet resistance of intrinsic-base layer Ω/□
Rsw Equivalent switching resistance of a CMOS gate Ω
Rswn Equivalent switching resistance of nMOSFET pulldown Ω
Rswp Equivalent switching resistance of pMOSFET pullup Ω
ρ Resistivity Ω-cm
ρsh Sheet resistivity Ω/□
ρch Sheet resistivity of MOSFET channel Ω/□
ρsd Sheet resistivity of source or drain region Ω/□
ρc Specific contact resistivity Ω-cm2
ρnet Volume density of net charge C/cm3
S MOSFET inverse subthreshold current slope V/decade
Sp Surface recombination velocity for holes cm/s
σn Capture cross-section of electron traps cm2
σp Capture cross-section of hole traps cm2
t Time s
tB Base transit time s
tBE Base–emitter depletion-layer transit time s
tBC Base–collector depletion-layer transit time s
tBOX Thickness of buried oxide in SOI cm
ti Thickness of gate insulator cm
tinv Equivalent oxide thickness for inversion charge calculations cm
tox Oxide thickness cm
tr Transit time s
tw Thickness of wire cm
tsi Thickness of silicon film cm
T Absolute temperature K
τ Lifetime s
τ Circuit delay s
τb Buffered delay s
τint Intrinsic, unloaded delay s
τF Forward transit time of bipolar transistor s
τn Electron lifetime s
τn nMOSFET pulldown delay s
τnB Electron lifetime in base of n–p–n transistor s
τp Hole lifetime s
τp pMOSFET pullup delay s
τpE Hole lifetime in emitter of n–p–n transistor s
xxvi List of Symbols
(cont.)
τR Reverse transit time of bipolar transistor s

τw Wire RC delay s
τE Emitter delay time s
τB Base delay time s
τBE Base–emitter depletion-region delay time (= tBE) s
τBC Base–collector depletion-region delay time (= tBC) s
U Net recombination rate 1/cm3-s
n Velocity cm/s
v Small-signal voltage V
nth Thermal velocity cm/s
nd Carrier drift velocity cm/s
nsat Saturation velocity of carriers cm/s
vT Thermal injection velocity at MOSFET source cm/s
V Voltage V
V Quasi-Fermi potential along MOSFET channel V
VA Early voltage V
Vapp Applied voltage across p–n diode V
V0 app Applied voltage appearing immediately across p–n junction (smaller V
than Vapp by IR drops in series resistances)
VBE Base–emitter bias voltage V
VBC Base–collector bias voltage V
VCE Collector-to-emitter voltage V
VCG Control gate voltage in a nonvolatile memory device V
VFG Floating gate voltage in a nonvolatile memory device V
Vbg Back gate bias voltage in SOI V
Vdd Power-supply voltage V
Vds Source-to-drain voltage V
Vdsat MOSFET drain saturation voltage V
Vfb Flat-band voltage V
Vox Potential drop across oxide V
Vg Gate voltage in MOS V
Vgs Gate-to-source voltage in a MOSFET V
Vbs MOSFET body bias voltage V
Vt Threshold voltage (2ψB definition) V
Von Linearly extrapolated threshold voltage V
Vin Input node voltage of a logic gate V
Vout Output node voltage of a logic gate V
Vx Node voltage between stacked nMOSFETs of a NAND gate V
Vt,high The higher threshold voltage of a nonvolatile memory device V
Vt,low The lower threshold voltage of a nonvolatile memory device V
W Width, MOSFET width cm
Wn nMOSFET width cm
Wp pMOSFET width cm
WB Intrinsic-base width cm
Wd Depletion-layer width cm
WdBE Base–emitter junction depletion-layer width cm
WdBC Base–collector junction depletion-layer width cm
List of Symbols xxvii
(cont.)
Wdm Maximum depletion-layer width in MOS cm

WE Emitter-layer width (thickness) cm
WEC Emitter-to-collector spacing of a lateral bipolar transistor cm
WS Source junction depletion-layer width cm
WD Drain junction depletion-layer width cm
ω Angular frequency rad/s
xj Junction depth cm
xc, xi Depth of inversion channel cm
ψ Potential V
ψB Difference between Fermi potential and intrinsic potential V
ψbi Built-in potential V
ψf Fermi potential V
ψi Intrinsic potential V
ψp Band bending in polysilicon gate V
ψs Surface potential in MOS V
ψs,min Minimum surface potential in short-channel MOSFET V
1 Introduction

Since the invention of the bipolar transistor in 1947, there has been an unprecedented
growth of the semiconductor industry, with an enormous impact on the way people
work and live. In the last forty years or so, by far the strongest growth area of the
semiconductor industry has been in silicon very-large-scale-integration (VLSI) tech-
nology. The sustained growth in VLSI technology is fueled by the continued shrink-
ing of transistors to ever smaller dimensions. The benefits of miniaturization – higher
packing densities, higher circuit speeds, and lower power dissipation – have been key
in the evolutionary progress leading to today’s computers, wireless units, and com-
munication systems that offer superior performance, dramatically reduced cost per
function, and much reduced physical size, in comparison with their predecessors. On
the economic side, the integrated-circuit (IC) business has grown worldwide in sales
from $1 billion in 1970 to $20 billion in 1984 and has reached $439 billion in 2020.
The electronics industry is now among the largest industries in terms of output as well
as employment in many nations. The importance of microelectronics in economic,
social, and even political development throughout the world will no doubt continue to
ascend. The large worldwide investment in VLSI technology constitutes a formidable
driving force that will all but guarantee the continued progress in IC integration
density and speed, for as long as physical principles will allow.
1.1 Evolution of VLSI Device Technology
1.1.1 Historical Perspective

An excellent account of the evolution of the metal–oxide–semiconductor field-effect
transistor (MOSFET), from its initial conception to VLSI applications in the mid-
1980s, can be found in a paper by Sah (1988). Figure 1.1 gives a chronology of the
major milestone events in the development of VLSI technology. The vertical bipolar
transistor technology was developed early on and was applied to the first integrated-
circuit memory in mainframe computers in the 1960s. Vertical bipolar transistors have
been used all along where raw circuit speed is most important, for bipolar circuits
remain the fastest at the individual-circuit level. However, the large power dissipation
1
2 1 Introduction
One-transistor
First bipolar DRAM cell
transistor invented
(1947) (1968)
First
MOSFET
(1960) VLSI era
1940 1950 1960 1970 1980 1990 2000 2010 2020

CMOS
invented
(1963)
IC First micro-
invented processor
(1958) (1971)
Figure 1.1 A brief chronology of the major milestones in the development of VLSI
of vertical bipolar circuits has severely limited their integration level, to about 104
circuits1 per chip. This integration level is very low by today’s VLSI standard.
The idea of modulating the surface conductance of a semiconductor by the appli-
cation of an electric field was first envisioned in 1930. However, early attempts to
fabricate a surface-field-controlled device were not successful because of the presence
of large densities of surface states which effectively shielded the surface potential
from the influence of an external field. The first MOSFET on a silicon substrate using
SiO2 as the gate insulator was fabricated in 1960 (Kahng and Atalla, 1960). During the
1960s and 1970s, n-channel and p-channel MOSFETs were widely used, along with
bipolar transistors, for implementing circuit functions on a silicon chip. Although the
MOSFET devices were slow compared to the bipolar devices, they had a higher layout
density and were relatively simple to fabricate; the simplest MOSFET chip could be
made using only four masks and a single doping step. However, just like vertical
bipolar circuits, single-polarity MOSFET circuits suffered from large standby power
dissipation, hence were limited in the level of integration on a chip.
The major breakthrough in the level of integration came in 1963 with the invention
of CMOS (complementary MOS) (Wanlass and Sah, 1963), in which n-channel and p-
channel MOSFETs are constructed side by side on the same substrate. A CMOS
circuit typically consists of an n-channel MOSFET and a p-channel MOSFET con-
nected in series between the power-supply terminals, so that there is negligible
standby power dissipation. Significant power is dissipated only during switching of
the circuit (i.e., only when the circuits are active). By cleverly designing the “switch
activities” of the circuits on a chip to minimize active power dissipation, engineers
have been able to integrate billions of CMOS transistors on a single chip and still have
the chip readily air-coolable. Until the minimum feature size of lithography reached
180 nm, the integration level of CMOS was not limited by chip-level power
1
ECL circuits, discussed in Section 11.2.
dissipation, but by chip fabrication technology. Another advantage of CMOS circuits

comes from the ratioless, full rail-to-rail logic swing, which improves the noise margin
and makes a CMOS chip easier to design.
As CMOS scaling reached the 0.5-μm level in the early 1990s, the performance of
high-end computers built using CMOS started to approach those built using bipolar, due
to the much higher integration level of CMOS chips. Designers of high-end computer
systems were able to meet their performance targets using CMOS instead of bipolar
(Rao et al., 1997). Since then, CMOS has become the technology for digital circuits,
and vertical bipolar is used primarily in radio-frequency (RF) and analog circuits only.
Advances in lithography and etching technologies have enabled the industry to
scale down transistors in physical dimensions, and to pack more transistors in the
same chip area. Such progress, combined with a steady growth in chip size, resulted in
an exponential growth in the number of transistors and memory bits per chip. The
technology trends up to 2020 in these areas are illustrated in Figure 1.2. Traditionally,
dynamic random-access memories (DRAMs) have contained the highest component
count of any IC chips. This has been so because of the small size of the one-transistor
memory cell (Dennard, 1968) and because of the large and often insatiable demand for
more memory in computing systems. It is interesting to note that the entire content of
this book can be stored in one 64-Mb DRAM chip, which was in volume production
in 1997 and has an area equivalent to a square of about 1.2 1.2 cm2.
One remarkable feature of silicon devices that fueled the rapid growth of the
information technology industry is that their speed increases and their cost decreases
as their size is reduced. The transistors manufactured in 2020 were 10-times faster and
occupy less than 1% of the area of those built 20 years earlier. This is illustrated in the
trend of microprocessor units (MPUs) in Figure 1.2. The increase in the clock
frequency of microprocessors is the result of a combination of improvements in
microprocessor architecture and improvements in transistor speed.
1.1.2 Recent Developments

Since the publication of the second edition of this book in 2009, there have been major
developments in the VLSI industry. Several fabrication technologies emerging at the
time have taken hold, enabling the continued chip-level density improvements,
resulting in continued reduction of cost per transistor and cost per memory bit.
These in turn have driven the continued growth of the semiconductor industry.
These recent developments include the following.
Immersion lithography has been adopted for volume IC manufacturing (Lin, 2004).
Immersion lithography is a photolithography resolution enhancement technique
where the usual gap between the final lens and the wafer surface is replaced with a
liquid medium having a refractive index greater than one. The resolution
enhancement is equal to the refractive index of the liquid used. With immersion,
deep ultraviolet (DUV) lithography systems remain the work horse for
semiconductor manufacturing today.
4 1 Introduction
1E+12 3
FLASH 256Gb
1E+11
CMOS
16Gb 8Gb 1
1E+10
Minimum feature size (µm)

1Gb 5.2GHz
# of transistors per chip
1E+9
0.3
5GHz
1E+8
1E+7 4Mb 0.1

200MHz
DRAM
1E+6
MPU 0.03
1E+5 Lithography
5MHz
1E+4 4Kb 0.01
2MHz
1E+3
1970 1980 1990 2000 2010 2020
Year
Figure 1.2 Trends in lithographic feature size, number of transistors per chip for DRAM and
MPU, and number of memory bits per chip for Flash. The transistor count for DRAM is
computed as 1.5-times the number of bits on the chip to account for the peripheral circuits. Data
points represent announced leading-edge products
Driven by the need for low-power and light-weight data storage in battery-operated
personal systems, NAND flash (the highest density version of the electrically
programmable and erasable nonvolatile memory) development has been on an
exceptionally steep trajectory since the mid-1990s. In the past decade or so, NAND
flash has overtaken DRAM as the IC chip with the highest component count, as
shown in Figure 1.2 (Kim, 2008). Since then, the combination of 3D NAND
process technology, where upwards of 100+ layers of NAND flash memory
cells are stacked on top of another to form a three-dimensional IC chip, and multi-
bit-per-cell design has dramatically increased the chip-level bit density of
NAND flash.
For a long time, a common practice in designing a scaled CMOS device was to
allow its off current to increase, by reducing the device threshold voltage, in
order to increase the on current to achieve the targeted performance of the scaled
device. As a result, the off current of scaled CMOS devices had been increasing
from one generation to the next. By the time scaling reached the 65 nm node, the
off current of scaled CMOS device had reached 100 nA/μm, the maximum level
acceptable to designers of high-end microprocessors. Since then, high-
performance CMOS devices have been designed with a nominal off current of
100 nA/μm (Kuhn et al., 2012). Capping the tradeoff between on current and off
current in designing scaled CMOS devices severely limits the speed (clock
frequency) of microprocessors. Today, the highest speed microprocessors run at
5.2 GHz (Berry et al., 2020), practically the same as those in 2009 (see
Figure 1.2).
Without the ability to increase the device off current, CMOS designers turned to
device structures having fully depleted device body, which enables the subthreshold
swing of the device to reach the ideal 60 mV/decade at room temperature. Today,
FinFET CMOS, where the depleted device body is shaped like a fin, planar ETSOI
(extremely thin silicon-on-insulator) CMOS, which is basically fully depleted SOI
CMOS, as well as traditional planar bulk CMOS are in volume manufacturing.
Partially depleted (PD) planar SOI CMOS ran its course over a span of about fifteen
years, with the first PD SOI CMOS microprocessor fabricated in the 220 nm node
in 1999 (Shahidi et al., 1999), and the last in the 22 nm node in 2015 (Freeman
et al., 2015). PD SOI CMOS is judged not scalable to smaller dimensions.
However, SOI wafers suitable for PD SOI CMOS have found new applications in
complementary (integration of n p n and p n p) lateral bipolar (Cai et al.,
2011), offering interesting possibilities in bipolar for VLSI.
1.2 Scope and Brief Description of the Book
In writing this book, it is our goal to address the factors governing the performance of
modern VLSI devices in depth. This is carried out by first discussing the device
physics that goes into the design of individual device parameters, and then discussing
the effects of these parameters on the performance of small-dimension modern
transistors at the basic circuit level. A substantial part of the book is devoted to in-
depth discussions on the interdependency among the device parameters and the subtle
tradeoffs in the design of modern CMOS and bipolar transistors.
This book contains sufficient background tutorials to be used as a textbook for
students taking a graduate or advanced undergraduate course in microelectronics. The
prerequisite is one semester of either solid-state physics or semiconductor physics. For
the practicing engineer, this book provides an extensive source of reference material
that covers the fundamentals of CMOS and bipolar technologies, devices, and circuits.
It should be useful to VLSI process engineers and circuit designers interested in
learning basic device principles, and to device design or characterization engineers
who desire more in-depth knowledge in their specialized areas.
New topics and materials in the third edition include an expanded chapter on ETSOI
and FinFETs, non-GCA (Gradual Channel Approximation) model for MOSFETs, and
the relatively recent development of symmetric lateral bipolar transistors on SOI. Also
added are sections on high-κ gate dielectrics, metal gates, strain effect on mobility, and
interface–state models. Much of the materials in the second edition have been restruc-
tured by consolidating all the appendices into the main chapters for a more focused
coverage of the various subjects. Here is a brief description of each chapter.
Chapter 2: Basic Device Physics

Chapter 2 covers the appropriate level of basic device physics to make the book self-
contained, and to prepare the reader with the necessary background on device
operation and material physics to follow the discussion in the rest of the book.
6 1 Introduction
Starting with the energy bands in silicon, Chapter 2 introduces the basic concepts of
Fermi level, carrier concentration, drift and diffusion current transport, and Poisson’s
equation. Also addressed in this chapter are generation and recombination, minority
carrier lifetime, and current continuity equation.
Chapter 3: p–n Junctions and Metal–Silicon Contacts

Chapter 3 covers the basic physics and operation of p n junctions and Schottky
diodes as well as metal silicon contacts in general. p n junctions are basic building
blocks of bipolar transistors and key components of MOSFETs. Basic knowledge of
their characteristics is a prerequisite to further understand the operation of bipolar
devices and for designing MOSFETs. And basic knowledge of Schottky diodes is
prerequisite to understanding metal silicon contacts in general and for designing
ohmic contacts with low contact resistance. The chapter ends with a discussion of
high-field effects in reverse-biased diodes.
Chapter 4: MOS Capacitors

Chapter 4 covers the fundamentals of MOS capacitors – a prerequisite to MOSFET
transistors. Starting with the basic concepts of free electron level and work function,
the chapter proceeds to the solution of charge and potential in silicon, followed by a
full description of the C–V characteristics. Quantum mechanical effects, important for
MOS capacitors of thin oxides, are then discussed. Added in the third edition is a new
section on interface states and oxide traps. Lastly, the high field section covers
tunneling currents, high-κ gate dielectrics, and gate oxide reliability.
Chapter 5: MOSFETs: Long Channel

Chapter 5 describes the basic characteristics of MOSFET devices, using n-channel
MOSFET as an example for most of the discussions. It deals with the more elementary
long-channel MOSFETs, with sections on the charge sheet model, regional I–V
models, and subthreshold current characteristics. A recently developed non-GCA
model gives insights to the saturation region behavior while clarifying the misleading
term of “pinch-off” in most standard textbooks. In the section on channel mobility, the
strain effects, both biaxial and uniaxial, on electron and hole mobilities are discussed.
The last section addresses the body effect, temperature effect, and quantum effect on
the long-channel threshold voltage.
Chapter 6: MOSFETs: Short Channel

This chapter deals with the more complex short-channel MOSFETs. Most circuits are
built with short-channel devices because of their higher current and lower capacitance.
Among the main topics are short-channel effects, scale length model, velocity satur-
ation, and non-local transport. A ballistic MOSFET model is described on the current
limit of a MOSFET. Next considered are the major device design issues in a CMOS
technology: choice of threshold voltage based on the off-current requirement and on-
current performance, power supply voltage, design of nonuniform channel doping,
and discrete dopant effects on threshold voltage. The last section discusses high-field
effects in a short-channel MOSFET.
Chapter 7: Silicon-on-Insulator and Double-Gate MOSFETs

Chapter 7 deals with fully-depleted SOI and double-gate MOSFETs. A general,
asymmetric double-gate model is applied to long channel SOI MOSFETs. For sym-
metric double-gate MOSFETs – the generic form of FinFETs, an analytic potential
model is described that covers all regions of operation continuously. The scale length
model first introduced in Chapter 6 for bulk MOSFETs is modified for short-channel
DG MOSFETs. Nanowire MOSFET models, both long and short channel, are also
discussed. The last section examines the scaling limits of DG and nanowire MOSFETs
based on quantum mechanical considerations.
Chapter 8: CMOS Performance Factors

This chapter begins by reviewing MOSFET scaling – the guiding principle for achiev-
ing density, speed, and power improvements in VLSI evolution. The implications of the
non-scaling factors, specifically, thermal voltage and silicon bandgap, on the path of
CMOS evolution are discussed. The rest of the chapter deals with the key factors that
govern the switching performance and power dissipation of basic digital CMOS
circuits. After a brief description of static CMOS logic gates, their layout and noise
margin, Section 8.3 considers the parasitic resistances and capacitances that may
adversely affect the delay of a CMOS circuit. These include source and drain series
resistance, junction capacitance, overlap capacitance, gate resistance, and interconnect
capacitance and resistance. In Section 8.4, a delay equation is formulated and applied to
study the sensitivity of CMOS delay to a variety of device and circuit parameters such as
wire loading, device width and length, gate oxide thickness, power-supply voltage,
threshold voltage, parasitic components, and substrate sensitivity in stacked circuits.
The last section addresses the performance factors of MOSFETs in RF circuits, in
particular, the unity-current-gain frequency and unity-power-gain frequency.
Chapter 9: Bipolar Devices

The basic components of a bipolar transistor are described in Chapter 9. Both vertical
bipolar transistors, including SiGe-base transistors, and symmetric lateral bipolar
transistors on SOI are covered. The discussion focuses on the vertical n–p–n transis-
tors, since they are the most commonly used. The difference between n–p–n vertical
transistors and symmetric lateral n–p–n transistors are pointed out where appropriate.
The basic operation of a bipolar transistor is described in terms of two p–n diodes
connected back to back. The basic theory of a p–n diode is modified and applied to
8 1 Introduction
derive the current equations for a bipolar transistor. From these current equations,
other important device parameters and phenomena, such as current gain, early voltage,
base widening, and diffusion capacitance, are examined. The basic equivalent-circuit
models relating the device parameters to circuit parameters are developed. These
equivalent-circuit models form the starting point for discussing the performance of a
bipolar transistor in circuit applications.
Chapter 10: Bipolar Device Design

Chapter 10 covers the basic design of a bipolar transistor. The design of the individual
device regions, namely the emitter, the base, and the collector, are discussed separ-
ately. Since the detailed characteristics of a bipolar transistor depend on its operating
point, the focus of this chapter is on optimizing the device design according to its
intended operating condition and environment, and on the tradeoffs that must be made
in the optimization process. The physics and characteristics of vertical SiGe-base
bipolar transistors are discussed in depth. The design of symmetric lateral bipolar
transistors on SOI is also covered, including the development of analytical models for
the device parameters, base and collector currents, and the transit times.
Chapter 11: Bipolar Performance Factors

The major factors governing the performance of bipolar transistors in circuit applica-
tions are discussed in Chapter 11. Several of the commonly used figures of merit,
namely, cutoff frequency, maximum oscillation frequency, and logic gate delay, are
examined, and how a bipolar transistor can be optimized for a given figure of merit is
discussed. Sections are devoted to examining the important delay components of a
logic gate, and how these components can be minimized. The scaling properties of
vertical bipolar transistors for high-speed digital logic circuits are discussed.
A discussion of the optimization of bipolar transistors for RF and analog circuit
applications is given. The chapter concludes with a discussion of the design tradeoff
and optimization of symmetric lateral bipolar transistors for RF and analog circuit
applications. Finally, several unique opportunities offered by symmetric lateral bipolar
transistors, some of them beyond the capability of CMOS, are discussed.
Chapter 12: Memory Devices

In Chapter 12, the basic operational and device design principles of commonly used
memory devices are discussed. The memory devices covered include CMOS SRAM,
DRAM, bipolar SRAM, and several commonly used in nonvolatile memories. Typical
read, write, and erase operations of the various memory arrays are explained. The
issue of noise margin in scaled CMOS SRAM cells is discussed. A brief discussion of
more recent developments of NAND flash technologies, including multi-bit per cell,
3D NAND, and wear leveling is given.
2 Basic Device Physics

2.4 Basic Equations for Device Operation 28
This chapter reviews the basic concepts of semiconductor device physics. It covers
energy bands in silicon, Fermi level, n-type and p-type silicon, electrostatic potential,
drift and diffusion current transport, and basic equations governing VLSI device
operation. These will serve as the basis for understanding more advanced device
concepts discussed in the rest of the book.
2.1 Energy Bands in Silicon
The starting material used in the fabrication of VLSI devices is silicon in the crystalline
form. The silicon wafers are cut parallel to either the <111> or <100> planes (Sze, 1981),
with <100> material being the most commonly used. This is largely due to the fact that
<100> wafers, during processing, produce the lowest charges at the oxide–silicon
interface as well as higher mobility (Balk et al., 1965). In a silicon crystal each atom has
four valence electrons to share with its four nearest neighboring atoms. The valence
electrons are shared in a paired configuration called a covalent bond. The most important
result of the application of quantum mechanics to the description of electrons in a solid
is that the allowed energy levels of electrons are grouped into bands (Kittel, 1976). The
bands are separated by regions of energy that the electrons in the solid cannot possess:
forbidden gaps. The highest energy band that is completely filled by electrons at 0 K is
called the valence band. The next highest energy band, separated by a forbidden gap from
the valence band, is called the conduction band, as shown in Figure 2.1.
2.1.1 Bandgap of Silicon

What sets a semiconductor such as silicon apart from a metal or an insulator is that, at
absolute zero temperature, the valence band is completely filled with electrons, while
the conduction band is completely empty, and that the separation between the
conduction band and valence band, or the bandgap, is on the order of 1 eV. On the
one hand, no electrical conduction is possible at 0 K, since there are no electrons in
the conduction band, whereas the electrons in the completely filled valence band
cannot be accelerated by an electric field and gain energy. On the other hand, the
bandgap is small enough that at room temperature a small fraction of the electrons are
excited into the conduction band, leaving behind vacancies, or holes, in the valence
9
10 2 Basic Device Physics
Figure 2.1 Energy-band diagram of silicon
band. This allows limited conduction to take place from the motion of both the
electrons in the conduction band and the holes in the valence band. In contrast, an
insulator has a much larger forbidden gap of at least several electron volts, making
room-temperature conduction virtually nonexistent. Metals, on the contrary, have
partially filled conduction bands even at absolute zero temperature, so that the
electrons can easily move into states of higher energy in response to an applied
electric field. This makes them good conductors at any temperature.
As shown in Figure 2.1, the energy of the electrons in the conduction band increases
upward, while the energy of the holes in the valence band increases downward. The
bottom of the conduction band is designated E c , and the top of the valence band Ev .
Their separation, or the bandgap, is Eg ¼ E c E v . For silicon, E g is 1.12 eV at room
temperature or 300 K. The bandgap decreases slightly as the temperature increases, with
a temperature coefficient of dE g =dT 2:73 104 eV=K for silicon near 300 K.
Other important physical parameters of silicon and silicon dioxide are listed in Table 2.1
(Green, 1990).
2.1.2 Density of States

The density of available electronic states within a certain energy range in the conduc-
tion band is determined by the number of different momentum values that can be
acquired by electrons in this energy range. Based on quantum
mechanics,
there is one
allowed state in a phase space of volume ðΔx Δpx Þ Δy Δpy Δz Δpz ¼ h3 , where
px , py , pz are the x-, y-, z-components of the electron momentum, respectively, and h is
Planck’s constant. If we let N ðE ÞdE be the number of electronic states per unit volume
with an energy between E and E þ dE in the conduction band, then
dpx dpy dpz
N ðE Þ dE ¼ 2g , (2.1)
h3
Table 2.1. Physical properties of Si and SiO2 at room temperature (300 K)
Property Si SiO2
Atomic/molecular weight 28.09 60.08

Atoms or molecules/cm3 5:0 1022 2:3 1022
Density (g/cm3) 2.33 2.27
Crystal structure Diamond Amorphous
Lattice constant (Å) 5.43 –
Energy gap (eV) 1.12 89
Dielectric constant 11.7 3.9
Intrinsic carrier concentration (cm3) 1:0 1010 –
Carrier mobility (cm2/V-s) Electron: 1,430 –
Hole: 480 –
Effective density of states (cm3) Conduction band, Nc: 2:9 1019 –
Valence band, Nv: 3:1 1019 –
Breakdown field (V/cm) 3 105 >107
Melting point ( C) 1,415 16001700
Thermal conductivity (W/cm- C) 1.5 0.014
Specific heat (J/g- C) 0.7 1.0
Thermal diffusivity (cm2/s) 0.9 0.006
Thermal expansion coefficient ( C1) 2:5 106 0:5 106
where dpx dpy dpz is the volume in the momentum space within which the electron
energy lies between E and E þ dE, g is the number of equivalent minima in the
conduction band, and the factor of two arises from the two possible directions of
electron spin. The conduction band of silicon has a sixfold degeneracy, so g ¼ 6. Note
that MKS units are used here (e.g., length must be in meters, not centimeters).
If the electron kinetic energy is not too high, one can consider the energy–
momentum relationship near the conduction-band minima as being parabolic and
write
p2x p2y p2
E Ec ¼ þ þ z , (2.2)
2mx 2my 2my
where E Ec is the electron kinetic energy, and mx , my , mz are the effective masses.
The constant energy surface in momentum space is an ellipsoid with the lengths of the
symmetry axes proportional to the square roots of mx , my , and mz . For the silicon
conduction band in the <100> direction, two of the effective masses are the
transverse mass mt ¼ 0:19m0 , and the third is the longitudinal mass ml ¼ 0:92m0 ,
where m0 is the free electron mass. The volume of the ellipsoid given by Eq. (2.2)
in momentum space is ð4π=3Þð8mx my mz Þ1=2 ðE Ec Þ3=2 . Therefore, the volume
dpx dpy dpz within which the electron energy lies between E and E þ dE is
4πð2mx my mz Þ1=2 ðE E c Þ1=2 dE. Thus, Eq. (2.1) becomes
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi
8πg 2mx my mz pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 8πg 2m2t ml pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
N ðE ÞdE ¼ E E c dE ¼ E E c dE: (2.3)
h3 h3
Figure 2.2 Schematic plots of density of states, Fermi–Dirac distribution function, and their
products versus electron energy in a band diagram (after Sze, 1981)
The 3-D electron density of states in an energy diagram is then a parabolic function
with its downward apex at the conduction-band edge, and vice versa for the hole
density of states in the valence band. These are shown schematically in Figure 2.2
(Sze, 1981).
2.1.3 Distribution Function: Fermi Level

The energy distribution of electrons in a solid is governed by the laws of Fermi–Dirac
statistics. For a system in thermal equilibrium, the principal result of these statistics is
the Fermi–Dirac distribution function, which gives the probability that an electronic
state at energy E is occupied by an electron,
1
fD ðE Þ ¼ : (2.4)
1 þ eðEEf Þ=kT
Here k ¼ 1:38 1023 J=K is Boltzmann’s constant, and T is the absolute tempera-
ture. This function contains a parameter, E f , called the Fermi level. The Fermi level is
the energy at which the probability of occupation of an energy state by an electron is
exactly one-half. At absolute zero temperature, T ¼ 0 K, all the states below the Fermi
level are filled ( fD ¼ 1 for E < E f ), and all the states above the Fermi level are empty
(fD ¼ 0 for E > E f ). At finite temperatures, some states above the Fermi level are
filled as some states below become empty, in which case the probability distribution
fD ðE Þ makes a smooth transition from unity to zero as the energy increases across the
Fermi level. The width of the transition is governed by the thermal energy, kT. This is
plotted schematically in Figure 2.2, with a Fermi level in the middle of the forbidden
gap (for reasons that will soon be clear). It is important to keep in mind that the
1
thermal energy at room temperature is 0.026 eV, or roughly 40 of the silicon bandgap.
In most cases when the energy is at least several kT above or below the Fermi level,
Eq. (2.4) can be approximated by the following:
fD ðEÞ eðEEf Þ=kT for E > Ef (2.5)
and
fD ðE Þ 1 eðEf EÞ=kT for E < Ef : (2.6)
Equation (2.6) should be interpreted as stating that the probability of finding a hole
(i.e., an empty state not occupied by an electron) at an energy E < Ef is eðEf EÞ=kT .
The last two equations follow directly from Maxwell–Boltzmann statistics for clas-
sical particles, which are good approximations to Fermi–Dirac statistics when the
energy is at least several kT away from E f .
Fermi level plays an essential role in characterizing the equilibrium state of a
system. Consider two electronic systems brought into contact with Fermi levels E f 1
and E f 2 , and corresponding distribution functions f D1 ðEÞ and f D2 ðEÞ. If Ef 1 > E f 2 ,
then f D1 ðEÞ > f D2 ðE Þ, which means that at every energy E where electronic states
are available in both systems, a larger fraction of the states in system 1 are occupied
by electrons than those in system 2. Equivalently, a larger fraction of the states in
system 2 are empty than those in system 1 at energies where electronic states exist.
Since the two systems in contact are free to exchange electrons, there is a higher
probability for the electrons in system 1 to redistribute to system 2 than vice versa.
This leads to a net electron transport from system 1 to system 2, i.e., current flows
(defined in terms of positive charges) from system 2 to system 1. If there are no
power sources connected to the systems to sustain the Fermi level imbalance,
eventually the two systems will come to an equilibrium and E f 1 ¼ E f 2 . No further
net electron flow takes place once the same fractions of the electronic states in the
two systems are occupied at every energy E. Note that this conclusion is reached
regardless of the specific density of states in each of the two systems. For example,
the two systems can be two metals, a metal and a semiconductor, or two semicon-
ductors of different doping or different composition. When two systems are in
thermal equilibrium with no current flow between them, their Fermi levels must
be equal. A direct extension is that, for a continuous region of metals and/or
semiconductors in contact, the Fermi level at thermal equilibrium is flat, i.e.,
spatially constant, throughout the region. The role of Fermi level at the contacts
when there is an applied voltage driving a steady-state current is further discussed in
Section 2.4.2.
2.1.4 Carrier Concentration

Since fD ðE Þ is the probability that an electronic state at energy E is occupied by
an electron, the total number of electrons per unit volume in the conduction band is
given by
Z ∞
n¼ N ðEÞfD ðE ÞdE: (2.7)
Ec
Here the upper limit of integration is taken as infinity because the top of the
conduction band is far above Ec . Both the product N ðE ÞfD ðE Þ and n, p are shown
schematically in Figure 2.2. Equation (2.7) with the full Fermi–Dirac distribution
function, Eq. (2.4), is discussed in Section 2.2.3. For nondegenerate silicon with a
Fermi level at least 3kT/q below E c , the Fermi–Dirac distribution function can be
approximated by the Maxwell–Boltzmann distribution, Eq. (2.5). Equation (2.7) then
becomes
pffiffiffiffiffiffiffiffiffiffiffiffiffi Z
8πg 2m2t ml ∞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðEEf Þ=kT
n¼ E Ec e dE: (2.8)
h3 Ec
With a change of variable, the integral can be expressed in the form of a gamma
function, Γð3=2Þ, which equals π 1=2 =2. The electron concentration in the conduction
band is then
n ¼ N c eðEc Ef Þ=kT , (2.9)
where the pre-exponential factor is defined as the effective density of states,

qffiffiffiffiffiffiffiffiffiffi
2πkT 3=2
N c ¼ 2g mt ml
2 : (2.10)
h2
A similar expression can be derived for the hole density in the valence band,
p ¼ N v eðEf Ev Þ=kT , (2.11)
where N v is the effective density of states of the valence band, which depends on
the hole effective mass and the valence band degeneracy. Both N c and N v are
proportional to T 3=2 . Their values at room temperature are listed in Table 2.1
(Green, 1990).
For an intrinsic silicon, n ¼ p, since for every electron excited into the conduction
band, a vacancy or hole is left behind in the valence band. The Fermi level for intrinsic
silicon, or the intrinsic Fermi level, E i , is then obtained by equating Eqs. (2.9) and
(2.11) and solving for E f :

Ec þ E v kT Nc
Ei ¼ Ef ¼ ln : (2.12)
2 2 Nv
By substituting Eq. (2.12) for E f in Eq. (2.9) or Eq. (2.11), the intrinsic carrier
concentration, ni ¼ n ¼ p is obtained:
pffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffi
ni ¼ N c N v eðEc Ev Þ=2kT ¼ N c N v eEg =2kT : (2.13)
Since the thermal energy, kT, is much smaller than the silicon bandgap, E g , and
ln ½N c =N v is not a large number, the intrinsic Fermi level is very close to the
midpoint between the conduction band and the valence band. In fact, E i is some-
times referred to as the midgap energy level, since the error in assuming E i to be
ðEc þ E v Þ=2 is only about 0.3 kT. The intrinsic carrier concentration, ni, at room
temperature is 1:0 1010 cm3 , as given in Table 2.1. This is very small compared
with the atomic density of silicon.
Equations (2.9) and (2.11) can be rewritten in terms of ni and Ei :
n ¼ ni eðEf Ei Þ=kT , (2.14)
p ¼ ni eðEi Ef Þ=kT : (2.15)
These equations give the equilibrium electron and hole densities for any Fermi level
position (not too close to the band edges) relative to the intrinsic Fermi level at the
midgap. In Section 2.2, we will show how the Fermi level varies with the type and
concentration of impurity atoms in silicon. Since any change in Ef causes reciprocal
changes in n and p, a useful, general relationship is that the product
pn ¼ n2i (2.16)
in equilibrium is a constant, independent of the Fermi level position.
2.2 n-Type and p-Type Silicon
Intrinsic silicon at room temperature has an extremely low free-carrier concentration;

therefore, its resistivity is very high. In practice, intrinsic silicon hardly exists at room
temperature, since it would require materials with an unobtainably high purity. Most
impurities in silicon introduce additional energy levels in the forbidden gap and can be
easily ionized to add either electrons to the conduction band or holes to the valence
band, depending on where the impurity level is (Kittel, 1976). The electrical con-
ductivity of silicon is then dominated by the type and concentration of the impurity
atoms, or dopants, and the silicon is called extrinsic.
2.2.1 Donors and Acceptors

Silicon is a column-IV element with four valence electrons per atom. There are two
types of impurity in silicon that are electrically active: those from column V, such as
arsenic or phosphorus, and those from column III, such as boron. As is shown in
Figure 2.3, a column-V atom in a silicon lattice tends to have one extra electron
loosely bonded after forming covalent bonds with other silicon atoms. In most cases,
the thermal energy at room temperature is sufficient to ionize the impurity atom and
free the extra electron to the conduction band. Such types of impurity are called
donors; they become positively charged when ionized. Silicon material doped with
column-V impurities or donors is called n-type silicon, and its electrical conductivity
is dominated by electrons in the conduction band. On the other hand, a column-III
impurity atom in a silicon lattice tends to be deficient by one electron when forming
covalent bonds with other silicon atoms (Figure 2.3). Such an impurity atom can
also be ionized by accepting an electron from the valence band, which leaves a
Figure 2.3 Three basic bond pictures of silicon: (a) intrinsic Si with no impurities, (b) n-type
silicon with donor (phosphorus), (c) p-type silicon with acceptor (boron) (after Sze, 1981)
Figure 2.4 Energy-band diagram representation of (a) donor level E d and Fermi level E f in n-type
silicon, (b) acceptor level Ea and Fermi level E f in p-type silicon
free-moving hole that contributes to electrical conduction. These impurities are called
acceptors; they become negatively charged when ionized. Silicon material doped with
column-III impurities or acceptors is called p-type silicon, and its electrical conduct-
ivity is dominated by holes in the valence band. It should be noted that impurity atoms
must be in a substitutional site (as opposed to interstitial) in silicon in order to be
electrically active.
In terms of the energy-band diagrams in Figure 2.4, donors add allowed electron
states in the bandgap close to the conduction-band edge, while acceptors add allowed
states just above the valence-band edge. Donor levels contain positive charge when
ionized (emptied). Acceptor levels contain negative charge when ionized (filled). The
Figure 2.5 Donor and acceptor levels of various impurities in silicon. Numbers next to the level
indicate ionization energies E c Ed (donors) or Ea E v (acceptors) in electron volts (after Sze,
1981)
ionization energies are denoted by E c E d for donors and E a E v for acceptors.

Figure 2.5 shows the donor and acceptor levels of common impurities in silicon and
their ionization energies (Sze, 1981). Phosphorus and arsenic are commonly used
donors, or n-type dopants, with low ionization energies on the order of 2kT, while
boron is a commonly used acceptor or p-type dopant with a comparable ionization
energy. Figure 2.6 shows the solid solubility of important impurities in silicon as a
function of annealing temperature (Trumbore, 1960). Arsenic, boron, and phosphorus
have the highest solid solubility among all the impurities, which makes them the most
important doping species in VLSI technology.
2.2.2 Fermi Level in Extrinsic Silicon

In contrast to intrinsic silicon, the Fermi level in an extrinsic silicon is not located at
the midgap. The Fermi level in n-type silicon moves up toward the conduction band,
i.e., E c E f decreases, consistent with the increase of electron density in Eq. (2.9). On
the other hand, the Fermi level in p-type silicon moves down toward the valence band,
i.e., E f E v decreases, consistent with the increase of hole density in Eq. (2.11).
These cases are depicted in Figure 2.4. The exact position of the Fermi level depends
on both the ionization energy and the concentration of dopants. For example, for an
n-type material with a donor impurity concentration, N d , the charge neutrality condi-
tion in silicon requires that
n ¼ Nþ
d þ p, (2.17)
where N þ
d is the density of ionized donors given by
!
1
Nþ
d ¼ N d ½1 fD ðE d Þ ¼ N d 1 , (2.18)
1 þ 12 eðEd Ef Þ=kT
since the probability that a donor state is occupied by an electron (i.e., in the neutral
state) is fD ðEd Þ. The factor 12 in the denominator of fD ðE d Þ arises from the spin
degeneracy (up or down) of the available electronic states associated with an ionized
Figure 2.6 Solid solubility of various elements in silicon as a function of temperature (after
Trumbore, 1960)
donor level1 (Ghandhi, 1968). In other words, while a neutral donor atom has only one
electron to lose, a positively charged donor atom can recapture an electron in one of its
two possible states. Substituting Eqs. (2.9) and (2.11) for n and p in Eq. (2.17) obtains
Nd
N c eðEc Ef Þ=kT ¼ þ N v eðEf Ev Þ=kT , (2.19)
1 þ 2eðEd Ef Þ=kT
which is an algebraic equation that can be solved for Ef . In n-type silicon, electrons
are the majority current carriers, while holes are the minority current carriers, which
means that the second term on the right-hand side (RHS) of Eq. (2.19) can be
neglected. For shallow donor impurities with low-to-moderate concentration at room
temperature, ðN d =N c Þ exp½ðE c E d Þ=kT 1, a good approximate solution for E f is
1
Detailed study showed that there is no other degeneracy with the electronic ground state in a donor except
for the spin (Ning and Sah, 1971).

Nc
E c E f ¼ kT ln : (2.20)
Nd
In this case, the Fermi level is at least a few kT below E d and essentially all the donor
levels are empty (ionized), i.e., n ¼ N þ d ¼ Nd.
It was shown in Eq. (2.16) that, in equilibrium, the product of majority and
minority carrier densities equals n2i , independent of the dopant type and Fermi level
position. The minority hole density in n-type silicon is then given by
p ¼ n2i =N d : (2.21)
Likewise, for p-type silicon with a shallow acceptor concentration, N a , the Fermi level
is given by

Nv
E f Ev ¼ kT ln , (2.22)
Na
the hole density is p ¼ N

a ¼ N a , and the minority electron density is
n ¼ n2i =N a : (2.23)
Figure 2.7 plots the Fermi level position in the energy gap versus temperature for a
wide range of impurity concentration (Grove, 1967). The slight variation of the silicon
bandgap with temperature is also incorporated in the figure. It is seen that as the
temperature increases, the Fermi level approaches the intrinsic value near midgap.
When the intrinsic carrier concentration becomes larger than the doping concentration,
Figure 2.7 The Fermi level in silicon as a function of temperature for various levels of impurity
concentration (after Grove, 1967)
the silicon is intrinsic. In an intermediate range of temperature including room

temperature, all the donors or acceptors are ionized. The majority carrier concentration
is then given by the doping concentration, independent of temperature. For tempera-
tures below this range, freeze-out occurs, i.e., the thermal energy is no longer
sufficient to ionize all the impurity atoms even with their shallow levels (Sze,
1981). In this case, the majority-carrier concentration is less than the doping concen-
tration, and one would have to solve Eq. (2.19) numerically to find E f , n, and p
(Shockley, 1950).
Instead of using N c , N v and referring to E c and E v , Eqs. (2.20) and (2.22) can be
written in a more useful form in terms of ni and E i defined by Eqs. (2.12) and (2.13):

Nd
E f Ei ¼ kT ln (2.24)
ni
for n-type silicon, and

Na
E i Ef ¼ kT ln (2.25)
ni
for p-type silicon. In other words, the distance between the Fermi level and the
intrinsic Fermi level near the midgap is a logarithmic function of doping concen-
tration. These expressions will be used extensively throughout the book.
2.2.3 Degenerately Doped Silicon

For heavily doped silicon, the impurity concentration N d or N a can exceed the
effective density of states N c or N v , so that E f > E c or Ef < E v according to Eqs.
(2.20) and (2.22). In other words, the Fermi level moves into the conduction band for
nþ silicon, and into the valence band for pþ silicon. Under these circumstances, the
silicon is said to be degenerate. For degenerately doped silicon, Boltzmann approxi-
mation [Eqs. (2.5) and (2.6)] is no longer valid. The full Fermi–Dirac distribution
function must be used in Eq. (2.7) for the electron density:
pffiffiffiffiffiffiffiffiffiffiffiffiffi Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
8πg 2m2t ml ∞ E Ec
n¼ dE: (2.26)
h E c 1 þ eðEE f Þ=kT
3
This integral cannot be carried out analytically. In terms of the Fermi–Dirac integral
defined by
Z ∞ pffiffiffi
y
F 1=2 ðuÞ dy, (2.27)
0 1þe
yu
Eq. (2.26) takes the form of

2 Ef Ec
n ¼ pffiffiffi N c F 1=2 , (2.28)
π kT
100
eu
2
10
1
F1/2(u)
0.1
0.01
0.001
0.0001
–10 –5 0 5 10
u
Figure 2.8 Fermi–Dirac integral of the order ½. The dotted line depicts Boltzmann approximation
valid for u << 1
where the effective density of states N c is given by Eq. (2.10). A numerical plot of the
Fermi–Dirac integral is shown in Figure 2.8. Its asymptotic approximations are
(Blakemore, 1982)
8 pffiffiffi
>
> π u
< e for u 1
2
F 1=2 ðuÞ (2.29)
>
>
: 2 u3=2 for u 1
3
. Ec Ef
Note that in the non-degenerate limit where kT, Eq. (2.28) is reduced to
Eq. (2.9).
Another effect with heavily doped silicon is bandgap narrowing. When the
impurity concentration is higher than 1017 cm3 , the donor (or acceptor) levels start
to broaden into bands. This results in an effective decrease in the ionization energy
until finally the impurity band merges with the conduction (or valence) band and the
ionization energy becomes zero. Bandgap narrowing has a nonnegligible effect in
bipolar devices and is discussed in more detail in Section 9.1.1.2.
2.3 Carrier Transport in Silicon
Carrier transport which gives rise to current flow in silicon is driven by two different
mechanisms: (a) the drift of carriers, caused by the presence of an electric field, and (b)
the diffusion of carriers, caused by an electron or hole concentration gradient in
silicon. They are discussed in this section.
2.3.1 Drift Current: Mobility

Under thermal equilibrium, electrons possess an average kinetic energy proportional
to kT. They move in random directions through the silicon crystal with an average
thermal velocity vth . At room temperature, vth is of the order of 107 cm=s. In the
absence of an electric field, the net velocity of electrons in any particular direction is
zero, since the thermal motion is completely random. When an electric field E is
applied, the carriers are accelerated and acquire a drift velocity superimposed upon
their random thermal motion. The magnitude of the acceleration is qE =m∗ , where m∗
is the effective mass of carriers. The drift velocity of carriers, however, does not
increase indefinitely under field acceleration because carriers scatter frequently with
the lattice (phonons) and ionized impurity atoms. After each collision event, carrier
velocities become randomized again and the drift velocity is reset to zero.
Consequently, the drift velocity only builds up during the mean free time τ between
collisions to a value
vd ¼ qE τ=m∗ : (2.30)
τ is related to the mean free path l, the average distance carriers travel between
collisions, through τ ¼ l=vth . Typically, l 10 nm, τ 0:1 ps. This establishes that,
at low electric fields, the drift velocity vd is proportional to the electric field strength E
with a proportionality constant μ, defined as the mobility, i.e.,
vd ¼ μE , (2.31)
with
qτ ql
μ¼ ¼ : (2.32)
m∗ m∗ vth
Electron and hole mobilities in silicon at low impurity concentrations are listed in
Table 2.1. The electron mobility is approximately three times the hole mobility, since
the effective mass of electrons in the conduction band is much lighter than that of
holes in the valence band.
Figure 2.9 plots the electron and hole mobilities at room temperature versus n-type
or p-type doping concentration. At low impurity levels, the mobilities are mainly
limited by carrier collisions with the silicon lattice or acoustic phonons (Kittel, 1976).
As the doping concentration increases beyond 1015 1016 =cm3 , collisions with the
charged (ionized) impurity atoms through Coulomb interaction become more and
more important and the mobilities decrease. In general, one can use Matthiessen’s rule
to include different contributions to the mobility:
1 1 1
¼ þ þ , (2.33)
μ μL μI
where μL and μI correspond to the lattice- and impurity-scattering-limited components

of mobility, respectively. At high temperatures, the mobility tends to be limited by
lattice scattering and is proportional to T 3=2 , relatively insensitive to the doping
Exploring the Variety of Random
Documents with Different Content
prevented its descent: “it is impossible to conceive that it is borne
downwards.”[37] Thus all the vulgar notions as to Hades were in
contradiction with Stoic psychology, a point to which we will return
in treating of the nether world.[38] These philosophers do indeed
speak of Hades but, faithful to their habits, while they use traditional
terms they give them a new meaning. “The descent into Hades” is for
them simply the departing from life, the transference of the soul to
new surroundings. Thus Epictetus, who uses this expression
(κάθοδος εἰς Ἄιδου), clearly states in another passage, “There is no
Hades, no Acheron, no Cocytus, no Pyriphlegethon, but all is filled
with gods and demons.”[39] These gods and demons were, however,
no more than personifications of the forces of nature.[40]
The true Stoic doctrine is, then, that souls, when they leave the
corpse, subsist in the atmosphere and especially in its highest part
which touches the circle of the moon.[41] But after a longer or less
interval of time they, like the flesh and the bones, are decomposed
and dissolve into the elements which formed them.
This thought, like Epicurean nihilism, often appears in epitaphs,
and shows how Stoic ideas had spread among the people. Thus on a
tombstone found in Moesia we read first the mournful statement
that there is neither love nor friendship among the dead and that the
corpse lies like a stone sunk into the ground. Then the dead man
adds:[42] “I was once composed of earth, water and airy breath
(πνεῦμα), but I perished, and here I rest, having rendered all to the
All. Such is each man’s lot. What of it? There, whence my body came,
did it return, when it was dissolved.” Sometimes there is more
insistence on the notion that this cosmic breath, in which ours is
gathered up, is the godhead who fills and rules the world. So in this
epitaph: “The holy spirit which thou didst bear has escaped from thy
body. That body remains here and is like the earth; the spirit pursues
the revolving heavens; the spirit moves all; the spirit is nought else
than God.”[43] Elsewhere we find the following brief formula, which
sums up the same idea: “The ashes have my body; the sacred air has
borne away my soul.”[44] Very characteristic is an inscription inspired
by verses of a Greek poet, on the tomb of a Roman woman: “Here I
lie dead and I am ashes; these ashes are earth. If the earth be a
goddess, I too am a goddess and am not dead.”[45]
These verses express the same great thought in various forms:
death is disappearance into the depths of divine nature. It is not for
the preservation of an ephemeral personality that we must hope. Our
soul, a fleeting energy detached from the All, must enter again into
the All as must our body: both are absorbed by God,
“When that which drew from out the boundless deep
Turns again home.”[46]
The fiery breath of our intelligence is gathered, as are the matter and
the humours of our organism, into the inexhaustible reservoir which
produced them, as one day the earth and the heavens will be
gathered thither also. All must be engulfed in one whole, must lose
itself in one forgetfulness. When man has reached the term of his
fate, he faints into the one power which forms and leads the
universe, just as the tired stars will be extinguished in it, when their
days shall be accomplished. Resistance to the supreme law is vain
and painful; rebellion against the irresistible order of things is
impious. The great virtue taught by Stoicism is that of submission to
the fatality which guides the world, of joyous acceptance of the
inevitable. Philosophic literature and the epitaphs present to us,
repeatedly and in a thousand forms, the idea that we cannot strive
against omnipotent necessity, that the rule of this rigid master must
be borne without tears or recriminations. The wise man, who
destroys within himself desire of any happenings, enjoys even during
this existence divine calm in the midst of tribulations, but those
whom the vicissitudes of life drive or attract, who let illusions seduce
or grieve them, will at last obtain remission of their troubles when
they reach the tranquil haven of death. This thought is expressed by
a distich which often recurs on tombs, in Greek and in Latin. “I have
fled, escaped. Farewell, Hope and Fortune. I have nothing more to do
with you. Make others your sport.”[47]
Stoic determinism found support in the astrology which originated
in Babylonia and was transplanted to Egypt, and which spread in the
Graeco-Latin world from the second century B. C. onwards,
propagating its mechanical and fatalistic conception of the universe.
According to this pseudo-science, all physical phenomena depended
absolutely, like the character and acts of men, on the revolutions of
the celestial bodies. Thus all the forces of nature and the very energy
of intelligence acted in accordance with an inflexible necessity.
Hence worship had no object and prayer no effect. In this way the
sidereal divination, which had grown up in the temples of the East,
ended in Greece, among certain of its adepts, in a negation of the
very basis of religion.[48] It is noteworthy that in the writings left to us
there is hardly an allusion to the immortality of the soul. When they
speak of what comes after death there is question only of funerals
and posthumous glory. We never find in them a promise to the
unfortunate, weighed down by misadventure and infirmities, of
consolation or compensation in the Beyond. The systematic astrology
of the Greeks limits its horizon to this world, although traces of the
belief in Hades subsist in its vocabulary and its predictions and
although this same astral divination inspired in the mysteries certain
eschatological theories, as we shall see later.[49]
The rationalistic and scientific period of Hellenic thought which

began, as we have said, with Aristotle, filled the Hellenistic period
and continued until the century of Augustus. Towards the end of the
Roman Republic faith in the future life was reduced to a minimum
and the scepticism or indifference of the Alexandrians was carried
into Italy. The mocking verses of an epigram of Callimachus, a man
of learning as well as a poet, is well known.[50] “Charidas, what is
there down below? Deep darkness. But what of the journeys
upwards? All lies. And Pluto? A fable. Then we are lost.” Catullus was
to say as much, less lightly, with a deeper feeling. “Suns can set and
rise again, but we, when our brief light is extinguished, must sleep
for an eternal night.”[51] The religions belief in retribution in the
Beyond was shaken, as all the others were, not only in literary and
philosophic circles but among a large section of the population. The
old tales of the Elysian Fields and Tartarus no longer found credence,
as convincing testimony will show us.[52] Those who sought to
preserve them could do so only by using a daring symbol which
altered their character. But the idea of conscious survival after death
was itself no longer looked upon as sure. Many who did not go so far
as to deny it brutally were firmly agnostic. When we turn over the
pages of the thick volumes of the Corpus inscriptionum, we are
struck by the small number of the epitaphs which express the hope of
immortality. The impression received is quite the contrary of that
given by going through our own graveyards or surveying the
collections of Christian epitaphs of antiquity. On by far the larger
number of the tombs the survival of the soul was neither affirmed
nor denied; it was not mentioned otherwise than by the banal
formula Dis Manibus—so bereft of meaning that even some
Christians made use of it. Or else the authors of funereal
inscriptions, like the contemporary writers, used careful phrases
which showed their mental hesitations: “If the Manes still perceive
anything.... If any feeling subsist after death.... If there be reward for
the righteous beneath the ground.”[53] Such doubting propositions
are most frequent. The same indecision made people return to an
alternative presented by Plato in the Apology,[54] before his ideas had
evolved, and repeat that death is “an end or a passage,”—mors aut
finis aut transitus,—and no choice is made between the two
possibilities: the question is left open. The future life was generally
regarded as a consoling metaphysical conception, a mere hypothesis
supported by some thinkers, a religious hope but not an article of
faith. The lofty conclusion which ends Agricola’s eulogy will be
remembered. “If,” says Tacitus, “there be an abode of the spirits of
virtuous men, if, as sages have taught, great souls be not
extinguished with the body, rest in peace.” But side by side with the
supposition thus hazarded, the historian expresses the assurance
that Agricola will receive another reward for his merits. All that his
contemporaries have loved and admired in his character will cause
the fame of his deeds to live in men’s memory through the eternity of
ages.
We here see how the perplexity in which men struggled, when they
thought of psychic survival, gave earthly immortality a greater value
in the eyes of the ancients. It was for many of them the essential
point because it alone was certain. Not to fall into the abyss of
forgetfulness seemed a sufficient reward for virtue. “Death is to be
feared by those for whom everything is extinguished with their life,
not by those whose renown cannot perish.”[55] That the
commemoration of our merits may not cease when the short time of
our passage here below has ended, but may be prolonged for as long
as the sequence of future generations lasts—this is the deep desire
which stimulates virtue and excites to effort. Cicero, when
celebrating in the Pro Archia[56] the benefits wrought by the love of
glory,—from which he was by no means exempt himself,—remarks
shrewdly that even philosophers, who claim to show its vanity, are
careful to place their names at the beginning of their books, thus
showing the worth they attach to that which they exhort others to
despise. Even more than today, the hope of a durable renown, the
anxiety that their fellows should be busy about them even after their
departure, the preoccupation lest their life should not be favourably
judged by public opinion, haunted many men, secretly or avowedly
dominated their thought and directed their actions. Even those who
had played only a modest part in the world and had made themselves
known only to a narrow circle, sought to render their memory
unforgettable by building strong tombs for themselves along the
great roads. Epitaphs often begin with the formula Memoriae
aeternae, “To the eternal memory,” which we have inherited,
although the idea it represents no longer has for most of us any but a
very relative value.
In antiquity it was first connected with the old belief in a
communion of sentiments and an exchange of services between the
deceased and their descendants who celebrated the funeral cult.
When the firm belief in the power of the shades to feel and act ceased
to exist, offerings were made with another intention: survivors liked
to think that he who had gone had not entirely perished as long as
his remembrance subsisted in the hearts of those who had cherished
him and the minds of those who had learnt his praises. In some way,
he rose from the grave in the image made of him by the successors of
those who had known him. Epicurus himself stipulated in his will
that the day of his birth should be commemorated every month,[57]
and under the Roman Empire his disciples were still piously
celebrating this recurring feast. Thus this deep instinct of
preservation, which impels human beings to desire survival, showed
itself even in him who contributed most of all to destroy faith in
immortality.
It is always with difficulty that men resign themselves to dying

wholly. Even when reason admits, nay when it desires, annihilation,
the subconscious self protests against it; our personality is impelled
by its very essence to crave the persistence of its self. Besides, the
feelings of survivors rebel against the pain of an unending
separation, the definite loss of all affections. In the troubled times
which marked the end of the Roman Republic, at a moment when
changing fortune periodically turned all the conditions of existence
upside down, there grew up a stronger aspiration to a better future, a
search, to use the words of the ancients, for a sure haven, in which
man, tossed by the storms of life, might find quiet. Thus in the first
century B. C. the birth was seen, or rather the rebirth, of a mystic
movement which claimed to give by direct communication with God
the certainties which reason could not supply. The chief
preoccupation of philosophers began to be those capital questions as
to the origin and end of man which the schools of the earlier period
had neglected as unanswerable. It was above all the Neo-
Pythagoreans who gave up pure rationalism, and thus brought
Roman thought to admit new forms of immortality.
When the scientific school of the old Pythagorism came to an end
in Italy in the fourth century, the sect perpetuated itself obscurely in
mysterious conventicles, a sort of freemasonry of which the influence
in the Hellenistic period is difficult to measure or circumscribe. It
again took on new power in Alexandria under the Ptolemies. In this
metropolis, in which all the currents of Europe and Asia were
mingled, Pythagorism admitted at this time many ideas foreign to
the teaching of the old master of Samos. This teaching seems not to
have set forth a rigid, logically constructed theology, and the points
of contact with the beliefs of the East, which its ideas supplied,
favoured an accommodating syncretism. Pythagoras was said to have
had Plato as a disciple, and Plato was venerated almost as much as
the teacher he followed. The powerful structure of Stoic pantheism
did not fail to exercise an ascendancy over the theorists of the school.
This school had been, from its origin, in touch with the Orphic
mysteries and those of Dionysos and it remained so, but it was also
subject to the more remote influence of Babylonian and Egyptian
religions, and particularly of those Chaldean doctrines which the
Greeks had learnt to know after Alexander’s conquest.
This vast eclecticism, open to all novelties, did not bring about a
break with the past. Theology succeeded in effecting a reconciliation
with all, even the rudest and most absurd traditions of fable, by an
ingenious system of moral allegories. “Divine” Homer thus became a
master of piety and wisdom, and mythology a collection of edifying
stories. Demonology made it possible to justify all the traditional
practices of the cult, as well as magic and divination: everything
which seemed incompatible with the new idea of the divinity was
ascribed to lower powers. Thus the Pythagoreans could take up the
position not of adversaries or reformers but of interpreters of the
ancestral religion. They claimed that they remained faithful to the
wisdom of the sages who, at the dawn of civilisation, had received a
divine revelation, which had been transmitted first to Pythagoras and
then to Plato. They felt so sure that they were expressing the thought
of these masters, whose authority made law, that they did not
hesitate to subscribe the venerated names, by a pious fraud, to their
own writings. Nowhere did apocryphal literature have a more
luxuriant efflorescence than in these circles.
When the sect was introduced into Rome it sought, according to its
wont, to connect itself with old local traditions, and without much
difficulty it succeeded. The national pride of the conquerors of
Greece could, with some complacency, regard it as Italic. Pythagoras
passed for the teacher of King Numa, the religious legislator of the
city. Ennius had expressed this philosopher’s doctrine in his poems,
and altogether, from the time of the ancient republic onwards, the
half mythical moralist of Greater Greece enjoyed singular
consideration in Rome.[58]
But the first to give new life to the Pythagorean school, which had
died in Italy centuries before, was, according to Cicero, his friend, the
senator Nigidius Figulus, a curious representative of the scientific
religiosity which characterised the sect. This Roman magistrate, a
man of singular erudition, was bitten with all the occult sciences. A
grammarian, a naturalist and a theologian, he was also an astrologer
and magician and, on occasion, a wonder-worker. He did not confine
himself to theory but gathered about him a club of the initiate, of
whom we cannot say whether they were most attracted by scientific
curiosity, by austere morals or by mystic practices. Vatinius, the
relative and friend of Caesar, and, probably, the spiritualist Appius
Claudius Pulcher were the most prominent of this circle of converts.
It is significant that at much the same time the historian Castor of
Rhodes claimed to interpret Roman usages by Pythagorism,[59] and
the stories establishing a connection between the Roman state and
the reformers of Greater Greece were multiplied. In the Augustan age
a worldly poet like Ovid[60] thought it permissible to introduce into
his Metamorphoses, where no digression of the sort was to be looked
for, a long speech of Pythagoras explaining vegetarianism and
transmigration. A little later Antonius Diogenes, the romancer, found
in the same philosophy inspiration for his fantastic pictures of the lot
of souls.[61] All this goes to show how powerfully seductive the new
sect proved to be as soon as it was revived in Rome.
But it did not lack enemies. Public malignity did not spare these
mysterious theosophists who met in subterranean crypts. They were
blamed for neglecting the national cult, which had ensured the
greatness of the city, in order to indulge in condemned practices or
even to commit abominable crimes. It was a more serious matter
that their secret gatherings also excited the suspicion of the
authorities, and that the partakers were prosecuted as persons who
dealt in magic, which was punishable by law. The little Pythagorean
church seems not to have been able to maintain itself in the capital
for long. In Seneca’s time it was dead.[62]
But Pythagorism continued to find adepts in the empire and soon
returned to Rome. Under Domitian, Apollonius of Tyana made the
East resound with his preaching and miracles, and although thrown
into prison by this emperor he was in favour with his successors.
Under the Antonines, the false prophet Alexander of Abonotichos,
unmasked by Lucian, claimed to be a new incarnation of the sage of
Samos, whose wisdom he pretended to reveal in his mysteries. The
literary tradition of the sect maintained itself until the third century,
when it was absorbed by Neo-Platonism. In a period of syncretism,
the originality of this philosophy resided less in its doctrine than in
its observances, and when its conventicles were dissolved, it easily
merged itself in the school which professed to continue it. During its
long life Pythagorism had indeed had a powerful influence, not only
on the system of Plato and Plotinus, but also on the Oriental cults
which spread under the empire. It had supplied the first type of the
learned mysteries in which knowledge (γνῶσις) is at once the
condition and the end of sanctification.[63] Possibly it even penetrated
into Gaul at an early date, by way of Marseilles, and was thus known
to the Druids.
It would certainly be a mistake to look upon Pythagorism as a pure
philosophy, like Epicureanism and Stoicism. Its sectaries formed a
church rather than a school, a religious order, not an academy of
sciences. From a recent discovery in Rome, if my interpretation of
the monument is right,[64] we have learnt that the Pythagoreans met
in underground basilicas, constructed on the model of Plato’s cavern,
in which, according to the great idealist, the chained men see only
the shades of the higher realities.[65] A foundation sacrifice, that of a
dog and a young pig, was made before this basilica was constructed.
Its stucco decoration is borrowed almost entirely from Greek
mythology or the ceremonies of the mysteries. Secret rites and varied
purifications had to be accomplished in it; hymns accompanied by
sacred music were sung; and from a chair within the apse the doctors
gave esoteric teaching to the faithful. They taught them the symbols
in which the truths of faith and the precepts of conduct, formerly
revealed by Pythagoras and the other sages, were handed down in
enigmatic form. These remote disciples interpreted all the myths of
the past, and especially the Homeric poems, by psychological or
eschatological allegories. They laid down, as definite
commandments, a rule of strict observance which included all the
acts of daily life. At dawn, after he had offered a sacrifice to the rising
Sun, the pious man must decide on the way in which his day was to
be employed. Every evening he must make a threefold examination
of conscience, and, if he had been guilty of any sin of omission or
commission, must make an act of contrition. He was obliged to
follow a purely vegetarian diet and to practise many abstinences, to
make repeated prayers, to meditate lengthily. This austere and
circumstantial system of morals would ensure happiness and
wisdom on earth and salvation in the Beyond.
All the Neo-Pythagoreans agree in stating that the human soul is
related to God and therefore immortal. Many, like the Stoics, look
upon it as a parcel of the ether, an effluvium of burning and
luminous fluid which fills the celestial spaces and shines in the divine
stars. Others, who are nearer to Plato, believe it to be immaterial and
define it as a number in movement. Always, generation is regarded
as a fall and a danger for the soul. Enclosed in the body as in a tomb,
it runs the risk of corruption, even of perishing. Earthly existence is a
hard voyage on the stormy waters of matter, which are perpetually
rolling and surging. Thus a fundamental pessimism looked upon life
here below as a trial and a chastisement; a radical dualism placed the
body in opposition to the divine essence residing in it. The constant
care of the sage was to keep his soul from pollution by its contact
with the flesh. He abstained from meat and other foods which might
corrupt it; a series of tabus protected it against all contagion. Ritual
purifications restored to it its purity (ἁγνεία) which was continually
threatened. The unwearying exercise of virtue, the scrupulous
practice of piety preserved its original nature. Music, which caused it
to vibrate in harmony with the universe, and science, which lifted it
towards divine things, prepared its ascension to heaven. Meditation
was a silent prayer, which placed reason in communication with the
powers on high. Seized by love for the eternal beauties, it rose in its
transports even in this life to God, identified itself with Him and so
rendered itself worthy of a blessed immortality.[66]
When the death determined by destiny occurred, the soul escaped
from the body in which it was captive but kept its bodily form and
appearance, and this simulacrum (εἴδωλον) appeared to men in
dreams and after evocations. According to some Pythagoreans, this
subtle form was distinct from the soul (ψυχή), which ascended
immediately to the higher spheres. Others believed that, like a light
garment, it wrapped the soul, which was for some time constrained
to dwell here below.[67] After this shade had remained beside the
body or somewhere near the tomb for a certain number of days, it
rose in the atmosphere in which contended winds, water and fire,
and was purified by the elements. This zone, the lowest circle of the
world, was what fable had called hell (Inferi), and it was of this
passage from one circle to that next it that poets spoke when they
told of the Styx and Charon’s boat.[68] When the soul had been
purified it was borne, uplifted by the winds, to the sphere of the
moon. Here lay the boundary of life and death, the limit which
divided the residence of the immortals, where all was harmony and
purity, from the corrupt and troubled empire of generation. Thus the
luminary of the night was the first dwelling of the Blessed, and there
lay the Elysian Fields of the poets, Proserpina’s kingdom where rest
the shades. And the Fortunate Islands, of which the ancients sung,
were no other than the sun and the moon, celestial lands bathed by
the waters of the ether.[69]
The shade remained in the moon or was dissolved there, and pure
reason rose to the sun whence it came forth, or even reached the
summit of the heavens where reigned the Most High. A helpful
escort, called by mythology Hermes the Soul-Guide, or
psychopompos, led the elect to these Olympian peaks. There they
regained their true country, and as birth had been to them a death,
so their death was their rebirth. They enjoyed the contemplation of
the luminous gods. They were rapt by the ravishing tune of the
harmony of the spheres, that divine melody of which earthly music is
but a feeble echo.[70]
Some souls were kept on the banks of the Styx and could not cross
it: in other words, they were constrained to remain on the earth. The
dead who had not had religious burial must linger beside their
neglected bodies for a hundred years, the normal span of a human
life, before they were admitted to the place of purgation, where they
would sojourn for ten times that period.[71] In the same way those
who had died young or whose days had been cut short by violence
would not enter the purgatory before the due term of their life.[72] But
especially the souls of the criminal and the impious were thus
condemned to wander, restless and in pain, through the lower air,
which they filled with their multitude. It was these demoniac spirits
who returned as dismal phantoms to frighten the living, who were
evoked by wizards and who revealed the future in oracles.
Demonology accounted for all the aberrations of magic and
divination. These spirits rose to the aerial purgatory after they had
for long years tormented and been tormented, but they could not
reach the moon, which repelled them; they were condemned to
reincarnation in new bodies, either of men or of beasts, and were
once again delivered to the fury of the passions. These passions are
the Erinyes, of whom poets sung, that in Tartarus they burnt
criminals with their torches and scourged them with their whips. For
there was no subterranean hell: Hades was in the air or on our earth,
and the infernal sufferings described by mythology were the various
tortures inflicted on the souls condemned to transmigration.[73]
This religious philosophy, which, by a symbolism transforming the
meaning of the traditional beliefs, reconciled these with men’s
intelligence, did more than any other to revive faith in immortality.
Many enlightened men, like Cicero and Cato, had sought consolation
for the misfortunes of this world and a hope for the Beyond in
reading Plato, but Plato’s proof of immortality could convince only
those already convinced.[74] Pythagorism, on the other hand, offered
to restless souls a certainty founded on a revelation made to ancient
sages, and it satisfied at once the Roman love for order and rule, and
the human love for the marvellous and the mysterious. The evidence
of the effect of this philosophy is still recognisable, although it often
has not been recognised, in the compositions decorating many
sepulchral monuments and in the wording of several epitaphs. A
tombstone found at Philadelphia in Lydia is particularly curious.[75] It
bears a representation of the Y symbol, that is, of the diverging roads
between which man must choose when he leaves childhood behind
him. On the one side earthly travail leads the virtuous man to eternal
rest; on the other softness and debauchery bring the vicious man to a
gulf into which he falls. A metrical epitaph, found at Pisaurum
(Pesaro), hints covertly at the ideas of the school. This
commemorates a child who, in spite of his youth, had learnt the
dogmas of Pythagoras and read “the pious verses of Homer” as well
as the philosophers, and had studied in Euclid the sacred science of
numbers. His soul, runs the inscription,[76] “goes forward through the
gloomy stars of deep Tartarus towards the waters of Acheron,” a
sentence which can be understood only on the supposition that
Tartarus and Acheron had for the author a figurative meaning and
lay in the depths not of the earth but of the sky.
The belief in a celestial immortality which was thus propagated by

the half philosophical, half religious sect of the Pythagoreans was to
find a powerful interpreter in a thinker who had a predominant
influence over his contemporaries and the succeeding generation—in
Posidonius. We know little of his life. Born at Apamea in Syria, about
the year 135, he early left his native country, of which he seems to
have kept a poor opinion, and as a young student in Athens he
attended the lectures of the older Stoic Panaetius. The universal
curiosity which was to make him a scholar of encyclopaedic
knowledge soon impelled him to take long journeys, in which he even
reached the shores of the Atlantic and studied the tides of the ocean.
Upon his return he opened a school in the free city of Rhodes and
there numbered Cicero among his hearers. When he died at the age
of eighty-four the prestige he enjoyed both in the Roman world and
among the Greeks was immense. He owed his intellectual
ascendancy as much to the marvellous variety of the knowledge
which he displayed, as philosopher, astronomer, historian,
geographer and naturalist, as to his copious, harmonious and highly
coloured style.
A theologian rather than a logician, a scholar rather than a critic,
he did not construct an original metaphysical system comparable to
those of the great founders of schools. But Posidonius was the most
prominent representative of that syncretism which, as we have seen,
showed itself in the Pythagoreans before his day and which reigned
in the world about him, because men were weary of the sterile
discussions of opposing thinkers. He gave the support of his
authority and his eloquence to the eclecticism which reconciled the
principles of the ancient Greek schools. Moreover, his Syrian origin
led him to combine these doctrines with the religious ideas of the
East, which had with astrology given the Hellenes a new conception
of man and of the gods.[77]
It is exactly here that Posidonius is important from the point of
view of our subject: his tendencies represent a direct reaction against
the scepticism of his master Panaetius, who denied both the survival
of the soul[78] and the possibility of divination. Posidonius introduced
into Stoicism momentous ideas derived at once from Pythagorism
and from Eastern cults, and sought to establish them firmly by
connecting them with a system of the world, which his vast
intelligence had sought to understand in all its aspects. His faith in
immortality is strictly related to his cosmography and receives
support from his physics.
It was this system of the world which was, thanks to Ptolemy’s
authority, to perpetuate itself on the whole until the time of
Copernicus. We will here give a broad outline of its essential features,
because the eschatological doctrines were to remain for centuries
connected with it. The terrestrial globe was held to be suspended,
motionless, in the centre of the universe, surrounded by an
atmosphere formed of the three other elements and reaching to the
moon. That part of the atmosphere which was near the earth was
thickened and darkened by heavy vapours rising from the soil and
the waters. Above, there moved a purer and lighter air which, as it
neared the sky, was warmed by contact with the higher fires. Still
higher were ranged the concentric spheres of the seven planets,
wrapped in ether, a subtle and ardent fluid—first the moon, which
still received and gave back the exhalations of the earth,[79] then
Mercury and Venus, the two companions of the sun in his daily
course. The fourth place, that is, the middle point of the
superimposed heavens, was occupied by the luminary of the day,—
here Posidonius forsakes Plato and follows the Chaldeans,—the
burning heart of the world, the intelligent light which is the source of
our minds.[80] Above the sun moved the three higher planets—Mars,
Jupiter and Saturn. And these seven wandering stars were
surrounded by the sphere of the fixed stars, which were animated by
constant and uniform movement. That sphere marked the world’s
boundary: beyond it there was only void or the ether.
The universe, as this philosophy imagined it, had therefore well-
defined limits: when men raised their eyes to the constellations of
the firmament, they thought they perceived its end. The depths of the
sky were not then unfathomable; he who sank his gaze in them was
not seized with giddiness at the abysses nor bewildered by
inconceivable magnitudes, and was not tempted to cry with Pascal:
[81]
“The eternal silence of these boundless spaces affrights me.” Nor
was the universe then a multiplicity of heavenly bodies moving to an
unknown goal and perpetually transformed, transitory
manifestations of an energy developed for undiscoverable ends. The
conception formed of the world was static, not dynamic. It was a
machine of which the wheels turned according to immutable laws, an
organism of which all the parts were united by reciprocal sympathy
as they acted and reacted on each other.
This organism was alive, penetrated throughout by the same
essence as the soul which maintains our life and thought. This soul
was an igneous breath of which the moral corruption was conceived
quite materially. When it gave itself up to the desires of the senses, to
corporeal passions, its substance thickened and was troubled, and
the mud of this pollution adhered to it like a crust. When the soul left
the body at the time of death, it became a spirit like the multitude of
demons who peopled the atmosphere. But its lot varied in
accordance with its condition. If it were laden with matter, its weight
condemned it to float in the densest air, the damp-charged gas which
immediately surrounded the earth, and its very composition then
caused it to reincarnate itself in new bodies.[82] But if it had remained
free from all alloy its lightness caused it to pass immediately through
this heavier layer of air and bore it to the higher spaces. It stopped in
this ascension when, within the ether which was about the moon, it
found itself in surroundings like its own substance. Some elect
beings, the divine spirits of the sages, kept such purity that they rose
through the ether as far as the highest astral spheres. In this system
the doctrine of immortality is seen to be closely knit up with
cosmography.
If Posidonius has largely borrowed these ideas from the
platonising Pythagorism of his period, he forsook this philosophy on
an essential point. As a Stoic he did not admit the transcendency of
God. For him, God was immanent in the universe; the seat of the
directing reason of the world (ἡγεμονικόν) was the sphere of the
fixed stars, which embraced all other spheres and determined their
revolutions. There too, at the summit of the world but not outside it,
the spirits of the blessed gathered; from these high peaks they
delighted to observe earthly happenings; and when a pious soul tried
to rise to them, these succouring heroes, like our saints, could lend
their aid and protection.
This philosophy did not draw its power of persuasion only from its
logical consistency, which satisfied reason, but it also made a strong
appeal to feeling. Posidonius caused a broad stream of mystical
ideas, undoubtedly derived from the beliefs developed by the astral
religions of the East, to flow into the arid bed of a Stoicism which had
become scholastic. For him, reason was not enclosed in the body,
even when it sojourned here below; it escaped from it to pass with
marvellous swiftness from the depths of the sea to the ends of the
earth and the top of the heavens; it flew through all nature, learning
to know physical laws and to admire the divine order ever more and
more. Above all it could never weary of the sight of the glowing
constellations and their harmonious movements. It felt with
emotion, as it gave itself up to contemplating them, its kinship with
the celestial fires; it entered into communion with the higher gods.
In enthusiastic terms, echoed by his imitators, Posidonius described
the ecstasy which seized him who left the earth, who felt himself
transported to the midst of the sacred chorus of the stars and who
followed their rhythmic evolutions. In these transports, the soul did
not only win to infinite power, but also received from heaven the
revelation of the nature and cause of the celestial revolutions. Thus
even in this life it had a foretaste of the beatitude which would
belong to it after death when reason, rid of the weak organs of the
senses, would directly perceive all the splendours of the divine world
and would know its mysteries completely.[83]
This theology attributed to man a power such as to satisfy his
proudest feelings. It did not regard him as a tiny animalcule who had
appeared on a small planet lost in immensity, nor did it, when he
scrutinised the heavens, crush him with a sense of his own pettiness
as compared with bodies whose greatness surpassed the limits of his
imagination. It made man king of creation, placed him in the centre
of a still limited world of which the proportions were not so vast that
he could not travel all over his domain. If he could tear himself from
the domination of his body, he became capable of communicating
with the visible gods who were almost within his reach and whom he
might hope to equal after his passage here below. He knew himself to
be united to them by an identity of nature which alone explained how
he understood them.
“Quis caelum possit, nisi caeli munere, nosse
Et reperire deum, nisi qui pars ipse Dei est?”[84]
“Who could know heaven save by heavenly grace, or find God if he

were not himself a part of God?”—words of the Roman poet who
echoes Posidonius’ teaching.
It is easy to understand that such ideas were readily adopted at a

time when human minds, tired of inconclusive disputes, despaired of
ever reaching truth by their own strength. The astral mysticism
eloquently preached by Posidonius was to influence all the later
Stoicism. Seneca in particular, in the numerous passages in which he
speaks of the misery and baseness of life in the body and celebrates
the felicity of the pure souls who live among the stars, shows the
imprint of the philosopher of Apamea. And this philosopher also
exerted a far-reaching action beyond the narrow circle of the school.
The erudition of the antiquarian Varro, the poems of Virgil and
Manilius and the biblical exegesis of Philo the Jew, all drew on him
for inspiration. But the author in whom we can best discern his
influence is his pupil Cicero, the abundance of whose writings allows
us to follow the evolution of his thought, which is characteristic of
the whole society of his time.
It is beyond doubt that Cicero was an agnostic for the greater part
of his life. His mind found satisfaction in the scepticism of the New
Academy, or rather he adopted towards the future life the received
attitude of the world in which he lived, where the problem of the
soul’s origin and destiny was regarded as not only insoluble but also
idle, as unworthy to absorb the minds of men who should devote
their energies to the service of the state. The question of the cult to
be rendered to the Manes had been settled once for all by the ancient
pontifical law. Old Rome distrusted speculations as to the Beyond
because they dangerously diverted thought from actual realities. But
Cicero, by his study of the writings of his master Posidonius and by
his intercourse with the senator Nigidius Figulus, a fervent adept of
Pythagorism,[85] had been brought into contact with the stream of
mystical ideas which was beginning to flow through the West.
Gradually, as he grew older and life brought him disappointments,
his thought was more attracted to religious ideas.[86] In 54, when he
had given up political life, he composed the Republic, an imitation of
Plato’s work on the same subject. As Plato had introduced the myth
of Er the Armenian at the end of his work, so his Roman imitator
concludes with the puzzling picture of “Scipio’s Dream,” where the
destroyer of Carthage receives the revelations of the conqueror of
Zama. The hero, from the height of the celestial spheres, expounds
that doctrine of astral immortality which was common to the
Pythagoreans and to Posidonius. It is given as yet only as a dream, a
vision the truth of which is in no way guaranteed. But in 45 B. C.
Cicero suffered a cruel loss in the death of his only daughter Tullia.
His grief persuaded him that this beloved being still lived among the
gods. Even while he accused himself of unreasonable weakness, he
ordered that not a tomb but a chapel (fanum), consecrating her
apotheosis, should be raised to this young woman. The letters he
wrote at this time to Atticus, from the shores of the Pomptine
Marshes, in the solitude of Astura, apprise us of his most intimate
feelings. He gave vent to his sorrow in writing a Consolatio, and in its
preserved fragments we see him strangely impressed by the
Pythagorean doctrines: he speaks of the soul, exempt from all
matter, as celestial and divine and therefore eternal, of the soul’s life
here below as a penalty inflicted on it because it is born to expiate
anterior crimes (scelerum luendorum causa).
Cicero’s sensitive spirit, troubled by the perplexing problem of our
destiny, did not turn to the old discredited beliefs but to the new
conceptions which a mystical philosophy had brought from the East.
Hortensius and the Tusculans, written in this period of his life, show
us the empire which the Neo-Stoicism of his master and the Neo-
Pythagorism of his fellow-senator then exercised over his mind,
saddened and disillusioned as he was, and show us too how he
sought consolation for the private and public ills which were
overwhelming him in the luminous doctrine of a blissful survival.
This spiritual evolution is an image of the great change which was
about to take place in the Roman world.
Stoic philosophy, although its maxims had been popularised by

education and literature, was almost as incapable of exercising a
wide influence on the deep masses of the people as the esoteric
theosophy revealed in the aristocratic conventicles of the
Pythagoreans. The urban “plebs,” to which slavery and trade had
given a strong admixture of Eastern blood, and the peasants of the
rural districts, where the gaps caused by depopulation were filled up
by a foreign labour supply, were beginning at the end of the
Republican period to hear new dogmas preached, dogmas which
were winning an ever increasing number of believers. The ancient
national cults of Greece and Rome aimed above all at ensuring civic
order and earthly welfare, and paid small regard to the spiritual
perfection of individuals and their eternal future. But now exotic
cults claimed to reveal the secret of immortality to their adepts.[87]
The Oriental mysteries, propagated in the West, united in the
promise of securing holiness in this life and felicity in the next, while
they imparted to their initiates the knowledge of certain rites and
required submission to certain precepts. Instead of the fluctuating
and disputable beliefs of philosophers as to destiny in the Beyond,
these religions gave certainty founded on divine revelation and on
the faith of countless generations attached to them. The truth, which
the mysticism of the thinkers looked to find in direct communication
with heaven, was here warranted by a venerable tradition and by the
daily manifestations of the gods adored. The belief in life beyond the
grave, which had in ancient paganism been so vague and melancholy,
was transformed into confident hope in a definite beatitude.
Participation in the occult ceremonies of the sect was an infallible
means of finding salvation. A society that was weary of doubt
received these promises eagerly, and the old beliefs of the East
combined with an eclectic philosophy to give a new eschatology to
the Roman Empire.
The salvation ensured by the mysteries was conceived as
identification with the god venerated in them. By virtue of this union
the initiate was reborn, like this god, to new life after he had
perished, or, like him, escaped from the fatal law of death which
weighs on humanity. He was “deified” or “immortalised,” after he
had taken part, as actor, in a liturgical drama reproducing the myth
of the god whose lot was thus assimilated to his own. Purifications,
lustrations and unctions, participation in a sacred banquet,
revelations, apparitions and ecstasies—a complicated series of
ceremonies and instructions helped to bring about this
metamorphosis of the faithful whom a higher power absorbed or
penetrated with its energy. We shall return to this sacramental
operation which made pious souls equal to the divinity.[88]
There is another point on which, in the course of this historical
introduction, we must dwell a little longer, namely, the evolution
undergone by the conception of the Beyond taught in the different
mysteries and the share of philosophy in the transformation. For if in
the various sects the liturgy was usually preserved with scrupulous
fidelity, its theological interpretation varied considerably as time
passed. In paganism much doctrinal liberty was always combined
with respect for rites.
Some of the mysteries often gave in their beginnings a rather
coarse idea of the future life, and the pleasures which might be
enjoyed therein were very material. The ancient Greek conception,
going back to Orphism, was, as we have seen, that of a subterranean
kingdom divided into two contrasted parts—Tartarus where the
wicked, plunged in a dark slough or subjected to other pains,
suffering the chastisement of their faults, on the one side; on the
other, the Elysian Fields, those flowered, luminous meadows, gay
with song and dance, in which the blessed pursued their favourite
occupations, whether they were allowed to dwell there for ever, or
whether they awaited there the hour fixed for their rebirth on earth.
[89]
This eschatology, which had become the common possession of
the Hellenes, was certainly that of the mysteries of Greece and in
particular of the mysteries of Eleusis. But these mysteries were never
more than local religions: however numerous were the initiates
attracted by their renown, they were bound to the soil where they
were born. Thus their influence was very limited in the Roman
period and cannot be compared with that of the universal cults which
were propagated throughout the Mediterranean world. As for
Orphism, which was never connected with any one temple, it is
doubtful whether it still constituted an actual sect, and if it did, it
certainly spread over a very narrow field. Its influence was
perpetuated chiefly because it was absorbed by Pythagorism.
Among the mysteries propagated in the West, the most ancient
were those of the Thraco-Phrygian gods, Dionysos and Sabazios, who
were indeed looked upon as identical. We know that in 186 B. C. a
senatus consultum forbade the celebration of the Bacchanalia in
Italy, and in 139 some sectaries of Jupiter Sabazius, who identified
this god with the Jahve-Sabaoth of the Jews, were expelled from
Rome by the praetor at the same time as the “Chaldeans.” The cult
practised by the votaries of Bacchus or Liber Pater, whose
confraternities were maintained until the end of paganism, differed
profoundly from the Dionysos worship of ancient Greece: a number
of Oriental elements had been introduced into it; in particular, the
relations between Dionysos and Osiris, which go back to a very
remote period, had become singularly close in Egypt. However, many
reliefs on tombstones and the celebrated paintings found in the
catacombs of Praetextatus prove that the cults of the Thraco-
Phrygian gods remained faithful to the old idea of a future life. The
shade went down into the bowels of the earth, never again to leave
them. If judged worthy, it took part in an eternal banquet, of which
the initiate received a foretaste on earth, in the feasts of the
mysteries. Sacred drunkenness, a divine exaltation, was the pledge of
the joyous intoxication which the god of wine would grant in Hades
to the faithful who had united themselves to him.[90]
In 205, towards the end of the second Punic war, the cult of
Cybele, the Great Mother of the Gods, and of Attis, her associate, was
transported from Pessinus in Phrygia and officially adopted by the
Roman people. The great feasts of this religion were celebrated in
March about the equinox and commemorated the death and
resurrection of Attis, the emblem of vegetation, which, after it has
withered, flowers again in the spring. The faithful associated their
own destiny with the lot of their god: like him they would be reborn
to a new life after they had died. Their doctrines on this point were
certainly transformed as time passed, for no Oriental cult which
spread in the West underwent more evolution, since none was more
fundamentally barbarous when it came from Asia. Originally, Cybele
was the goddess of the dead, because Mother Earth receives them
into her bosom. Every Phrygian tomb is a sanctuary and its epitaph a
dedication: often the graves are consecrated to the goddess and bear
her image or that of the lion, her substitute. Often too the tombstone
has the shape of a door, the door of the subterranean world whither
the dead descend. The belief seems to have been held that the
deceased were absorbed in the Great Mother who had given them
birth, and that they thus participated in her divinity. She brought
forth corn and grapes for men and thus sustained them day by day,
and the bread and wine, taken in the meal which was the essential
act of the initiation, would ensure immortality to those who were of
the mystery. “Thou givest us the food of life with unfailing
constancy,” says a prayer, “and when our soul departs we will take
refuge in thee. Thus all that thou givest, always falls to thee again.”[91]
Towards the end of the Republic the mysteries of Isis and Serapis,
which had come from Alexandria and had already spread through
the south of Italy, established themselves in Rome and maintained
themselves there in spite of opposition from the senate. Under the
Empire, the Egyptian religion displayed all the pomp of its liturgy in
magnificent temples and had a number of votaries in every province.
The cult of Osiris, of which that of Serapis was a form, was originally
a cult of the fields, like that of Attis, and the great feast which its
adherents celebrated in autumn recalls the Phrygian spring feasts.
The death of Osiris, whose body had been torn to pieces by Seth, was
mourned; and when Isis had found the scattered fragments of the
corpse, joined them together and reanimated it, noisy rejoicing
followed the lamentation. Like the initiates of Cybele and Attis, those
of Isis and Serapis were associated with the passion and resurrection
of their god. And, in the same way, the oldest conception of
immortality in these mysteries was that the departed went down into
the infernal regions, where a man became another Serapis, a woman
another Isis, which is to say that they were assimilated to the gods
who had granted them salvation.[92] This is why on numerous funeral
reliefs the dead man, who has become a hero and is shown lying on a
couch, bears on his head the bushel (modius) which is the attribute
of Serapis. In consequence, however, of the identification of this god
with Dionysos, the joys beyond the grave are also represented as a
feast in the Elysian Fields at which the great master of banquets
presides.[93]
All these mysteries conceive immortality as a descent of the dead
into Hades. For them, the kingdom of the dead lies in the bosom of
the earth. Those who have been initiated will there enjoy a felicity
made up of purely material pleasures, or they will be identified with
the powers who reign over the nether world and will have part in
their divine life. It will be noticed how closely this last conception
approached to that of ancient Stoicism, according to which the
various parts of the human organism, dissociated by death, were to
regain their integrity in the divine elements of the universe.
Quite another doctrine was propagated by the Syrian cults and the
Persian mysteries of Mithras, which spread in the West in the first
century of our era. These religions taught that the soul of the just
man does not go below the ground but rises to the sky, there to enjoy
divine bliss in the midst of the stars in the eternal light. Only the
wicked were condemned to roam the earth’s surface, or were dragged
by the demons into the dusky depths in which the spirit of evil
reigned. Opinions differed as to the region of heaven in which the
souls of the elect dwelt. The “Chaldeans,” who looked upon the sun
as the master and the intelligence of the universe, made him the
author of human reason, which returned to him after it had left the
body, while for the priests of Mithras the spirit rose, by way of the
planetary spheres, to the summit of the heavens. We will have to
examine later the different forms of astral immortality.[94] But you
will already have noticed how nearly this immortality, as formulated
by the Iranian and Semitic sects, approximated to the doctrine
taught by Pythagorism and adopted by Neo-Stoicism.
This meeting of the two doctrines was not an effect of chance. The
idea that souls are related to the celestial fires, whence they descend
at birth and whither they reascend at death, had probably been
borrowed by the ancient Pythagoreans from the astral religions of
the East. Recent research seems to have established the fact of its
Chaldeo-Persian origin. But the Greek philosophers, according to
their wont, defined and developed this idea in an original way. In the
Hellenistic period, when they adopted astrology, they were subject
for the second time to the ascendancy of the scientific religion of the
“Chaldeans”; and, in their turn, they reacted on the Oriental cults
when these spread in the Graeco-Roman world. We have sure
evidence that the mysteries of Mithras were, in particular, strongly
affected by the influence of the Pythagorean sect, which was itself
organised like a kind of mystery. In a more general way, philosophy
introduced into the mysteries ethical ideas and, instead of the purely
ritualistic or rather magical means of salvation, some moral
requirements became necessary to earn immortality.
There is here a mass of actions and reactions of which the details
escape us; but we can form some idea of such a syncretism from the
remains of the theological writings attributed to Hermes
Trismegistus, from the writings, that is, which are supposed to
contain the revelations of the Egyptian god Thot. This professedly
Egyptian wisdom includes a number of ideas and definitions which
are characteristic of Posidonius and Neo-Pythagorism. The Greek
and the Egyptian elements are so closely associated in it that it is
very difficult to separate the one from the other. We find another
example of the same mixture in the “Chaldaic Oracles,” which were
probably composed about the year 200 of our era and which became
one of the sacred books of Neo-Platonism. Unlike the Hermetic
writings, this collection of verses does indeed seem to have belonged
to a sect practising an actual cult: its greater part is taken up with
mythology, and the fantastic mysticism of the East is more
prominent here than in the Hermetic lore, but the mind of the
compiler of these revelations was also penetrated by the ideas which
the Greek masters had widely circulated.
The tenet of astral immortality, which philosophy shared with the
cults emanating from Syria and Persia, imposed itself on the ancient
world. It is curious to notice how it was introduced into the theology
of the very mysteries to which it was at first foreign: Attis ended by
becoming a solar god, and thenceforward it was in the heights of
heaven that Cybele was united to the souls she had prevented from
wandering in darkness and had saved from hell. The priests of the
Alexandrian divinities were similarly to explain that the dead had not
their dwelling in the interior of our globe, but that the
“subterranean” (ὑπόγειος) kingdom of Serapis was situated beneath
the earth, that is, in the lower hemisphere of heaven, bounded by the
line of the horizon.[95]
According as the Oriental religions were more largely propagated,
faith in a new eschatology spread gradually among the people; and
although memories and survivals of the old belief in the life of the
dead in the grave and the shade’s descent into the infernal depths
may have lingered, the doctrine which predominated henceforward
was that of celestial immortality.
The distance separating the age of Augustus from that of the
Flavians on this point can be measured by reading Plutarch’s moral
works (about 120 A. D.). A constant preoccupation with religious
matters, and in particular a learned curiosity as to the cults of the
East, shows itself in this Greek of Chaeronea, living in a country
which, in its pride in its own past, had more than any other resisted
the invasion of exoticism. Further, the eclectic philosopher likes to
insert in his dissertations myths in which, after the fashion of Plato,
he expounds the lot of souls in the Beyond and their struggle to rise
heavenwards. An attempt has been made to prove—wrongly, I think
—that he is here inspired by Posidonius. These apocalyptic visions,
which claim to reveal truths previously ignored, are not taken from
that well-known writer; they have a religious imprint which betrays
sacerdotal influence, and the philosophic ideas they contain are
those which were part of the common wisdom of the Pythagoreans
and the mysteries.
There doubtless still were in the second century Stoics, like the
Emperor Marcus Aurelius, for whom the future life was a mere
hypothesis, or at most a hope (p. 14), as well as sceptics, like Lucian
of Samosata, whose irony mocked all beliefs. But gradually their
number diminished and the echo of their voices grew feebler. Faith
in survival deepened as present life came to seem a burden harder
and harder to bear. The pessimistic idea that birth is a chastisement
and that the true life is not that passed on earth, imposed itself in
proportion to the growth of public and private ills and to the
aggravation of the empire’s social and moral decline. In the period of
violence and devastation which occurred in the third century, there
was so much undeserved suffering, there were so many unjust
failures and unpunished crimes, that men took refuge in the
expectation of a better life in which all the iniquity of this world
would be retrieved. No earthly hope then brightened life. The
tyranny of a corrupt bureaucracy stifled every attempt at political
progress. Science seemed exhausted and no longer discovered
unknown truths; art was struck with sterility of invention and
reproduced heavily the creations of the past. An increasing
impoverishment and a general insecurity constantly discouraged the
spirit of enterprise. The idea spread that humanity was smitten by
incurable decay, that society was on the road to dissolution and the
end of the world was impending. All these causes of discouragement
and pessimism must be remembered in order to understand the
dominance of the old idea, then so often repeated, that a bitter
necessity constrains the spirit of man to enclose itself in matter, and
that death is a liberation which delivers it from its carnal prison. In
the heavy atmosphere of a period of oppression and powerlessness,
the despondent souls of men aspired with ineffable ardour to the
radiant spaces of heaven.
The mental evolution of Roman society was complete when Neo-

Platonism took upon itself the office of directing minds. The
powerful mysticism of Plotinus (205–262 A. D.) opened up the path
which Greek philosophy was to follow until the world of antiquity
reached its end. We shall not undertake to notice in this place the
discrepancies of the latest teachers who theorised about the destiny
of souls. In the course of these lectures we shall have occasion to
quote some of the opinions of Porphyry, the chief disciple of
Plotinus, and of his successor Jamblichus, who was, like himself, a
Syrian. We will here do no more than indicate broadly what
distinguished the theories of this school from those which had
hitherto been dominant.
The system generally accepted, by the mysteries as by philosophy,
was a pantheism according to which divine energy was immanent in
the universe and had its home in the celestial spheres. The souls,
conceived as material, could in consequence rise to the stars but did
not leave the world. The Neo-Pythagoreans themselves had not had a
very firmly established doctrine on this point: while some of them
stated that reason was incorporeal, others, as we have seen (p. 24),
admitted with the Stoics that it was an igneous substance. It is true
that even in paganism the appearance can be discerned of the belief
in a Most High (Ὕψιστος) or an unknown god Ἄγνωστος, whom
some people supposed to dwell above the starry heavens, beyond the
limits of the world, and towards whom pious spirits could rise. The
revivers of Platonic idealism asserted the transcendence of God and
the spirituality of the soul more strongly and clearly. A whole chapter
of the Enneades of Plotinus is taken up with refuting those who held
the soul to be material.[96] As a principle of life and movement, it is
stated to be immortal by its very essence, so that if it kept its purity
perfect, it would find after its passage here below eternal felicity in
the intelligible world.
The Neo-Platonists preserved the idea, which had previously been
admitted, that this intellectual essence comes down to earth through
the planetary spheres and the atmosphere, and that as it sinks in the
luminous ether and the damp air, it becomes laden with particles of
the elements through which it passes. It surrounds itself with a
garment or, as it is sometimes called, with a vehicle (ὄχημα) which
thickens as it gradually draws near us.[97] This subtle body, the seat of
the passions and of feeling, is intermediary between the spiritual
principle which has issued from God and the flesh in which it is to
enclose itself, and for certain philosophers it survives death and
accompanies the soul to the Beyond, at least if the soul, not being
free from earthly admixture, cannot wholly leave the world of sense,
and therefore rises only to the planetary circle or to that of the fixed
stars.
When the soul has suffered even more from the taint to which its
contact with matter, the source of evil, exposes it, it is doomed to
reincarnate itself in a new body and again to undergo the trial of this
life. When it has become incurably corrupt and burdened with evil, it
goes down into the depths of Hades.
Following Plato, Plotinus and his successors have adopted the
Pythagorean doctrine of metempsychosis. They have even developed
it, as we shall see,[98] together with the whole pessimistic and ascetic
conception of life, the conception which looks at birth as a pain and a
fall, a temporary subjection to a body from which emancipation must
be sought. It is only after this liberation that the soul can reach
perfect wisdom; it must no longer be troubled by the senses if it is to
attain to the end of existence, to union with God.
This union can be realised even during this life in moments of
ecstasy, in which the soul rises above thought and gives itself up
entirely to love for the ineffable Unity in which it is absorbed. Like
many other mystics, Plotinus disdains the ceremonies of positive
cults: they were superfluous to the sage who could of himself enter
into communion with the supreme Being. But even his disciple
Porphyry conceded a greater value to rites and initiations. If they
were powerless to lead the partakers of mysteries to the highest
degree of perfection, their effect yet was to render men worthy to live
among the visible gods who people heaven.[99] But only philosophical
wisdom could rise to the intelligible world and the Unknowable.
The principle of a mystical relation between man and the divinity
was to lead Neo-Platonism to more and more reverence for religious
traditions. For it was held that in the past the revelation of truth had
been granted by Heaven not only to divine Plato and the sages of
Greece, but to all the founders of barbarous cults and authors of
sacred writings. They all communicated profound teaching, which
they sometimes hid beneath the veil of allegory. Inspired by the
symbolism of the Pythagoreans, the last representatives of Greek
philosophy claimed to rediscover the whole of Platonic metaphysics
and the Platonic doctrine of immortality in the myths and rites of
paganism. The speeches of Julian the Apostate on the Sun-King and
the Mother of the Gods are characteristic examples of this bold
exegesis, destitute of all critical and even all common sense, which
was adopted by the last champions of the old beliefs.
These aberrations of Neo-Platonic thought must not hide the

school’s historical importance from us, any more than the excesses of
the superstitious theurgy which invaded it. When it revived Plato’s
idealism, it produced a lasting change in the eschatological ideas
which prevailed in paganism, and it deeply influenced even the
Christian doctrines of immortality held since the fourth century. This
will be better seen, we hope, in the course of these lectures.[100] It
may be said that the conception of the lot of souls which reigned at
the end of antiquity persisted on the whole through the Middle Ages
—the immaterial spirits of the just rising through the planetary
spheres to the Supreme Being enthroned above the zone of the fixed
stars; the posthumous purification of those whom life has sullied in a
purgatory intermediary between heaven and hell; the descent of the
wicked into the depths of the earth where they suffered eternal
chastisement. This threefold division of the universe and of souls was
largely accepted at the time of the Empire’s decline by pagans and by
Christians, and after long centuries it was again to find magnificent
expression in Dante’s “Divine Comedy.” Before it could be destroyed
astronomy had to destroy the whole cosmography of Posidonius and
Ptolemy on which it was based. When the earth ceased to be the
centre of the universe, the one fixed point in the midst of the moving
circles of the skies, and became a tiny planet turning round another
heavenly body, which itself moved in the immensity of space, among
an infinity of similar stars, the naïve conception formed by the
ancients of the journey of souls in a well-enclosed world could no
longer be maintained. The progress of science discredited the
convenient solution bequeathed to scholasticism by antiquity, and
left us in the presence of a mystery of which the pagan mysteries
never had even a suspicion.
I
AFTER LIFE IN THE TOMB
W hen Cicero in his Tusculans[101] first touches on the question of

the immortality of the soul, he begins by citing in its support
the fact that belief in it has existed since earliest antiquity. He states
that unless the first Romans were convinced that man was not
reduced to nought, when he left this life, and that all feeling was not
extinguished in death, there could be no explanation of the rules of
the old pontifical law as to funerals and burials, rules the violation of
which was regarded as an inexpiable crime. This remark is that of a
very judicious observer. There subsist in the funeral rites of all
peoples, in the ceremony of mourning established by the religious
law or by tradition, customs which derive from archaic conceptions
of life beyond the tomb and which are still followed although their
original meaning is no longer understood. Modern learning has
sometimes successfully sought to elucidate them, borrowing light
from the practices of savage peoples and from European folk-lore.
We will not enter the domain of these researches, for since our
special purpose here is to expound the ideas as to immortality held in
later times, we have to consider only the beliefs which were still alive
in that period. A false interpretation supplied by a philosopher may
have more historical value for us than the true explanation of an
institution which had lost its meaning.
But even among the ideas which were neither obliterated nor
discredited, conceptions which originated at very different dates
have to be distinguished.
The doctrines of paganism, like the soil of our planet, are formed
of superimposed strata. When we dig into them we discover
successive layers under the upper deposits of recent alluvia. Nothing
was suddenly destroyed in ancient religions; their transformations
were never revolutionary. Faith in the past was not entirely abolished
when new ways of believing were formed. Contradictory opinions
could exist side by side for a long time without any shock being
caused by their disagreement; and it was only little by little and
slowly that argument excluded one way of thinking to give place to
the other, while there were always hardy survivals left, both in
thought and in customs. Thus the beliefs as to the future life which
were current under the Roman Empire present a singular mixture,
coarse ideas going back to the prehistoric period mingling with
theories imported into Italy at a late date.
We will today examine the oldest of all the ways of considering
survival in the Beyond: life in the tomb.
Ethnology has proved that among all peoples the belief that the
dead continue to live in the tomb has reigned, and sometimes still
reigns. The primitive man, disconcerted by death, cannot persuade
himself that the being who moved, felt, willed, as he does, can be
suddenly deprived of all his faculties. The most ancient and the
crudest idea is that the corpse itself keeps some obscure
sensitiveness which it cannot manifest. It is imagined to be in a state
like sleep. The vital energy which animated the body is still attached
to it and cannot exist without it. This belief was so powerful in Egypt
that it inspired a whole section of the funeral ritual and called forth
the infinite care that was taken to preserve mummies. Even in the
West it survived vaguely, and traces of it might still be discovered
today. Lucretius combats this invincible illusion of men who, even
while they affirm that death extinguishes all feeling, keep a secret
uneasiness as to the suffering which their mortal remains may
undergo and are frightened by the idea that their bodies may be
eaten by worms or carnivorous animals. They cannot separate
themselves from this prone body, which they believe is still their self.
Why, continues the poet, would it be more painful to be the prey of
wild beasts than to be burnt by the flame of the pyre, to freeze lying
on the icy slab of the grave or to be crushed by the weight of heaped-
up earth?[102] This very fear that the earth may weigh heavily on those
who are deposited in the grave shows itself among many peoples who

Fundamentals of Modern VLSI Devices 3rd Edition with Bookmarks Yuan Taur And Tak H. Ning 2024 Scribd Download

Uploaded by

Copyright:

Available Formats

Fundamentals of Modern VLSI Devices 3rd Edition with Bookmarks Yuan Taur And Tak H. Ning 2024 Scribd Download

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fundamentals of Modern VLSI Devices 3rd Edition with Bookmarks Yuan Taur And Tak H. Ning 2024 Scribd Download

Uploaded by

Copyright:

Available Formats

Download the full version of the ebook at ebookfinal.

Fundamentals of Modern VLSI Devices 3rd Edition

Download more ebook instantly today at https://ebookfinal.com

Fundamentals of Multinational Finance 3rd Edition Michael

Fundamentals of Liquid Crystal Devices 2nd Edition Deng-Ke

Modern VLSI Design IP Based Design 4th Edition Wayne Wolf

ESD Circuits and Devices 1st Edition Steven H. Voldman

Chemical Process Safety Fundamentals with Applications 3rd

Twentieth century Chinese Translation Theory Tak

Fundamentals of Geometric Dimensioning and Tolerancing 3rd

Fundamentals of Retiree Group Benefits 1st Edition Dale H.

Yuan Taur is Distinguished Professor of Electrical and Computer Engineering at the

Cambridge University Press is part of the University of Cambridge.

Preface to the Third Edition page xiii

2 Basic Device Physics 9

3 pn Junctions and MetalSilicon Contacts 43

4.4.3 Quantum-Mechanical Solution in Weak Inversion 131

5 MOSFETs: Long Channel 171

6 MOSFETs: Short Channel 206

6.4 MOSFET Degradation and Breakdown at High Fields 256

7 Silicon-on-Insulator and Double-Gate MOSFETs 264

8 CMOS Performance Factors 295

9 Bipolar Devices 361

10 Bipolar Device Design 425

10.4 SiGe-Base Vertical Bipolar Transistors 437

11 Bipolar Performance Factors 485

11.5.3 Power-Dissipation Issues with Resister-Load Bipolar Logic

12 Memory Devices 521

Description Symbol Value and unit

Electronic charge q 1.6 1019 C

Energy ¼ charge voltage E ¼ qV Joule ¼ Coulomb Volt

Symbol Description Unit

Symbol Description Unit

Cit Interface trap capacitance per unit area F/cm2

Symbol Description Unit

Ex Vertical ﬁeld in silicon V/cm

Symbol Description Unit

Ioff MOSFET off current A

Symbol Description Unit

np Density of electrons in p-region cm3

Symbol Description Unit

r0 Output resistance in small-signal hybrid-π equivalent-circuit model Ω

Symbol Description Unit

τR Reverse transit time of bipolar transistor s

Symbol Description Unit

Wdm Maximum depletion-layer width in MOS cm

1.1 Evolution of VLSI Device Technology 1

1.1 Evolution of VLSI Device Technology

1.1.1 Historical Perspective

1940 1950 1960 1970 1980 1990 2000 2010 2020

dissipation, but by chip fabrication technology. Another advantage of CMOS circuits

1.1.2 Recent Developments

Minimum feature size (µm)

1E+7 4Mb 0.1

1.2 Scope and Brief Description of the Book

Chapter 2: Basic Device Physics

Chapter 3: p–n Junctions and Metal–Silicon Contacts

Chapter 4: MOS Capacitors

Chapter 5: MOSFETs: Long Channel

Chapter 6: MOSFETs: Short Channel