Superconducting Radiofrequency

Technology for Accelerators : State of

the Art and Emerging Trends 1st
Edition Hasan Padamsee
Visit to download the full and correct content document:
https://ebookmeta.com/product/superconducting-radiofrequency-technology-for-accel
erators-state-of-the-art-and-emerging-trends-1st-edition-hasan-padamsee/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Emerging Perspectives in Art Therapy Trends Movements

and Developments 1st Edition Richard Carolan (Editor)

https://ebookmeta.com/product/emerging-perspectives-in-art-
therapy-trends-movements-and-developments-1st-edition-richard-
carolan-editor/

Proceedings of the 7th Brazilian Technology Symposium

BTSym 21 Emerging Trends in Human Smart and Sustainable
Future of Cities Volume 1 1st Edition Yuzo Iano

https://ebookmeta.com/product/proceedings-of-the-7th-brazilian-
technology-symposium-btsym-21-emerging-trends-in-human-smart-and-
sustainable-future-of-cities-volume-1-1st-edition-yuzo-iano/

Shock of the New The Challenge and Promise of Emerging

Technology 1st Edition Udell

https://ebookmeta.com/product/shock-of-the-new-the-challenge-and-
promise-of-emerging-technology-1st-edition-udell/

The Future of Forensic Psychology : Core Topics and

Emerging Trends 1st Edition Sara Landström

https://ebookmeta.com/product/the-future-of-forensic-psychology-
core-topics-and-emerging-trends-1st-edition-sara-landstrom/
The Regulation and Management of Workplace Health and
Safety Historical and Emerging Trends 1st Edition Peter
Sheldon

https://ebookmeta.com/product/the-regulation-and-management-of-
workplace-health-and-safety-historical-and-emerging-trends-1st-
edition-peter-sheldon/

Art Schools and Place Geographies of Emerging Artists

and Art Scenes 1st Edition Silvie Jacobi

https://ebookmeta.com/product/art-schools-and-place-geographies-
of-emerging-artists-and-art-scenes-1st-edition-silvie-jacobi/

Opportunistic Networks: Fundamentals, Applications and

Emerging Trends 1st Edition Anshul Verma

https://ebookmeta.com/product/opportunistic-networks-
fundamentals-applications-and-emerging-trends-1st-edition-anshul-
verma/

Techniques for Food Authentication Trends and Emerging

Approaches 2nd Ed Raúl González-Domínguez

https://ebookmeta.com/product/techniques-for-food-authentication-
trends-and-emerging-approaches-2nd-ed-raul-gonzalez-dominguez/

Safety for Particle Accelerators 1st Edition Thomas

Otto

https://ebookmeta.com/product/safety-for-particle-
accelerators-1st-edition-thomas-otto/
Superconducting Radiofrequency Technology for Accelerators
Superconducting Radiofrequency Technology
for Accelerators

State of the Art and Emerging Trends

Hasan Padamsee
Author All books published by WILEY-VCH are carefully
produced. Nevertheless, authors, editors, and
Dr. Hasan Padamsee publisher do not warrant the information
Cornell University contained in these books, including this book,
Newman Laboratory to be free of errors. Readers are advised to keep
Cornell University in mind that statements, data, illustrations,
NY procedural details or other items may
United States inadvertently be inaccurate.

Cover Image: Courtesy of Fermi National Library of Congress Card No.: applied for
Accelerator Laboratory
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available
from the British Library.

Bibliographic information published by

the Deutsche Nationalbibliothek
The Deutsche Nationalbibliothek lists this
publication in the Deutsche Nationalbiblio-
grafie; detailed bibliographic data are available
on the Internet at <http://dnb.d-nb.de>.

© 2023 WILEY-VCH GmbH, Boschstraße 12,

69469 Weinheim, Germany

All rights reserved (including those of

translation into other languages). No part of
this book may be reproduced in any form – by
photoprinting, microfilm, or any other means –
nor transmitted or translated into a machine
language without written permission from the
publishers. Registered names, trademarks, etc.
used in this book, even when not specifically
marked as such, are not to be considered
unprotected by law.

Print ISBN: 978-3-527-41409-3

ePDF ISBN: 978-3-527-83629-1
ePub ISBN: 978-3-527-83630-7
oBook ISBN: 978-3-527-83631-4

Typesetting Straive, Chennai, India

 v

Contents

Preface xi

Part I Update of SRF Fundamentals 1

1 Introduction 3

2 SRF Fundamentals Review 7

2.1 SRF Basics 7
2.2 Fabrication and Processing on Nb-Based SRF Structures 11
2.2.1 Cavity Fabrication 12
2.2.2 Preparation 12
2.2.3 A Decade of Progress 15
2.3 SRF Physics 15
2.3.1 Zero DC Resistance 15
2.3.2 Meissner Effect 17
2.3.3 Surface Resistance and Surface Impedance in RF Fields 19
2.3.4 Nonlocal Response of Supercurrent 22
2.3.5 BCS 24
2.3.6 Residual Resistance 30
2.3.7 Smearing of Density of States 31
2.3.8 Ginzburg–Landau (GL) Theory 31
2.3.9 Critical Fields 34
2.3.10 Comparison Between Ginzburg–Landau and BCS 39
2.3.11 Derivation of Rs and X s 39

Part II High Q Frontier: Performance Advances and

Understanding 43

3 Nitrogen-Doping 45
3.1 Introduction 45
3.2 N-Doping Discovery 46
3.3 Surface Nitride 48
vi Contents

3.4 Interstitial N 49
3.5 Electron Mean Free Path Dependence 52
3.5.1 LE-μSR Measurements of Mean Free path 53
3.6 Anti-Q-Slope Origins from BCS Resistance 55
3.7 N-Doping and Residual Resistance 58
3.7.1 Trapped DC Flux Losses 58
3.7.2 Residual Resistance from Hydride Losses 58
3.7.3 Tunneling Measurements 59
3.8 RF Field Dependence of the Energy Gap 61
3.9 Frequency dependence of Anti-Q-Slope 63
3.10 Theories for Anti-Q-Slope 63
3.10.1 Xiao Theory 63
3.10.2 Gurevich Theory 68
3.10.3 Nonequilibrium Superconductivity 72
3.10.4 Two-Fluid Model-Based on Weak Defects 75
3.11 Quench Field of N-Doped Cavities 77
3.12 Evolution and Comparison of N-doping Recipes 83
3.13 High Q and Gradient R&D Program for LCLS-HE 83
3.14 N-Doping at Other Labs 88
3.15 Summary of N-doping 88

4 High Q via 300 ∘ C Bake (Mid-T-Bake) 91

4.1 A Surprise Discovery 91
4.2 Similarities to N-Doping 91
4.3 Mid-T Baking at Other Labs 95
4.4 The Low-Field Q-Slope (LFQS) and 340 ∘ C Baking Cures 97
4.5 Losses at Very Low Fields 100
4.6 Losses from Two-Level Systems (TLS) 100
4.7 Eliminating TLS Losses 101

5 High Q’s from DC Magnetic Flux Expulsion 105

5.1 Trapped Flux Losses, Sensitivity 105
5.2 Trapped Flux Sensitivity Models 106
5.3 Vortex Physics 108
5.4 Calculation of Sensitivity to Trapped Flux 110
5.5 Dependence of Sensitivity on RF Field Amplitude 112
5.6 DC Magnetic Flux Expulsion 114
5.6.1 Fast versus Slow-Cooling Discovery 114
5.6.2 Thermoelectric Currents 118
5.7 Cooling Rates for Flux Expulsion 122
5.8 Flux Expulsion Patterns 123
5.9 Geometric Effects – Flux Hole 127
5.10 Flux Trapping With Quench 127
5.11 Material Quality Variations 129
5.12 Modeling Flux Trapping From Pinning Variations 135
 Contents vii

Part III High Gradient Frontier: Performance Advances and

Understanding 139

6 High-Field Q Slope (HFQS) – Understanding and Cures 141

6.1 HFQS Summary 141
6.2 HFQS in Low-𝛽 Cavities 142
6.3 Deconvolution of RBCS and Rres 143
6.4 Depth of Baking Effect 145
6.4.1 From Anodization 145
6.4.2 From HF Rinsing 145
6.4.3 Depth of Magnetic Field Penetration by LE-μSR 145
6.5 Role of the Oxide Layer and Role of N-Infusion 148
6.6 SIMS Studies of O, H, and OH Profiles 151
6.7 Hydrogen Presence in HFQS 156
6.8 TEM Studies on Hydrides 158
6.9 Niobium–hydrogen Phase Diagram 160
6.10 H Enrichment at Surface 161
6.11 Q-disease Review 163
6.12 Visualizing Niobium Hydrides 165
6.12.1 Cold-stage Confocal Microscopy 165
6.12.2 Cold-stage Atomic Force Microscopy (AFM) 166
6.13 Model for HFQS – Proximity Effect Breakdown of Nano-hydrides 168
6.13.1 Baking Benefit and Proximity Effect Model 170
6.14 Positron Annihilation Studies of HFQS and Baking Effect 172
6.15 Point Contact Tunneling Studies of HFQS and Baking Effect 173

7 Quest for Higher Gradients: Two-Step Baking and

N-Infusion 175
7.1 Two-Step Baking 175
7.2 Subtle Effects of Two-Step Baking – Bifurcation 175
7.2.1 Bifurcation Reduction 177
7.3 N-Infusion at 120 ∘ C 181
7.4 N-Infusion at Medium Temperatures 184
7.5 Unifying Quench Fields 188
7.6 Quench Detection by Second Sound in Superfluid Helium 190

8 Improvements in Cavity Preparation 193

8.1 Comparisons of Cold and Warm Electropolishing Methods 193
8.2 Chemical Soaking 197
8.3 Optical Inspection System and Defects Found 199
8.4 Robotics in Cavity Preparation 200
8.5 Plasma Processing to Reduce Field Emission 201

9 Pursuit of Higher Performance with Alternate Materials 207

9.1 Nb Films on Cu Substrates 207
9.1.1 Direct Current Magnetron Sputtering 209
viii Contents

9.1.2 DC-bias Diode Sputtering at High Temperature (400–600 ∘ C) 209

9.1.3 Seamless Cavity Coating 210
9.1.4 Nb–Cu Films by ECR 211
9.1.5 Nb–Cu Films via High-Power Impulse Magnetron Sputtering
(HIPIMS) 212
9.2 Alternatives to Nb 214
9.2.1 Nb3 Sn 215
9.2.2 MgB2 217
9.2.3 NbN and NbTiN 221
9.3 Multilayers 222
9.3.1 SIS’ Structures 222
9.3.2 Theoretical Estimates 223
9.3.3 Results 223
9.3.4 SS’ Structures 225
9.4 Summary 227

Part IV Applications 229

10 New Cavity Developments 231

10.1 Crab Cavities for LHC High Luminosity 231
10.2 Short-Pulse X-Rays (SPX) System for the APS Upgrade 238
10.3 QWR Cavity for Acceleration 239
10.4 Traveling Wave Structure Development 241

11 Ongoing Applications 245

11.1 Overview 245
11.2 Low-Beta Accelerators for Nuclear Science and Nuclear
Astrophysics 246
11.2.1 ATLAS at Argonne 246
11.2.2 ISAC and ISAC-II at TRIUMF 247
11.2.3 SPIRAL II at GANIL 247
11.2.4 HIE ISOLDE 248
11.2.5 RILAC at RIKEN 249
11.2.6 SPES Upgrade of ALPI at INFN 249
11.2.7 FRIB at MSU 250
11.2.8 RAON 253
11.2.9 Spoke Resonator Structure Developments to Avoid Multipacting 254
11.2.10 JAEA Upgrade 255
11.2.11 HELIAC 256
11.2.12 SARAF 259
11.2.13 HIAF at IMP 259
11.2.14 IFMIF 260
11.3 High-Intensity Proton Accelerators 260
11.3.1 SNS 260
 Contents ix

11.3.2 ESS 262

11.3.3 Accelerator Driven Systems (CADS) 262
11.3.4 CiADS (China Initiative Accelerator Driven System) 265
11.3.5 Japan Atomic Energy Agency (JAEA) – ADS 266
11.3.6 High-Intensity Proton Accelerator Development in India 266
11.3.7 PIP-II and Beyond 267
11.4 Electrons for Light Sources – Linacs 269
11.4.1 European X-ray Free Electron Laser (EXFEL) 269
11.4.2 Linac Coherent Light Source LCLS-II and LCLS-HE (LCLS-High
Energy) 273
11.4.3 Shanghai Coherent Light Facility (SCLF) SHINE 278
11.4.4 Institute of Advanced Science Facilities (IASF) 279
11.4.5 Polish Free-Electron Laser POLFEL 279
11.5 Electrons for Storage Ring Light Sources 281
11.5.1 High-Energy Photon Source (HEPS) 281
11.5.2 Taiwan Photon Source (TPS) 281
11.5.3 Higher Harmonic Cavities for Storage Rings Chaoen WANG, NSRRC,
Taiwan 282
11.5.4 BNL 284
11.6 Electrons in Energy Recovery Linacs (ERL) for Light Sources &
Electron–Ion Colliders 285
11.6.1 Prototyping ERL Technology at Cornell 285
11.6.2 KEK ERLs 287
11.6.3 Light-House Project for Radiopharmaceuticals 288
11.6.4 Peking ERL 288
11.6.5 Berlin ERL 288
11.6.6 MESA ERL 289
11.6.7 SRF Photo-injectors for ERLs 289
11.7 Electrons for Nuclear Physics, Nuclear Astrophysics, Radio-Isotope
Production 290
11.7.1 CEBAF at Jefferson Lab 290
11.7.2 ARIEL at TRIUMF 291
11.7.3 ERL for LHeC at CERN 291
11.8 Crab Cavities for LHC High Luminosity 292
11.9 Ongoing and Near-Future Projects Summary 293

12 Future Prospects for Large-Scale SRF Applications 295

12.1 The International Linear Collider (ILC) for High-Energy Physics 295
12.2 Future Circular Collider FCCee 298
12.3 China Electron–Positron Collider, CEPC 300

13 Quantum Computing with SRF Cavities 303

13.1 Introduction to Quantum Computing 303
13.2 Qubits 303
13.3 Superposition and Coherence 304
x Contents

13.4 Entanglement 304

13.5 2D SRF Qubits 306
13.6 Josephson Junctions 307
13.7 Dilution Refrigerator for Milli-Kelvin Temperatures 308
13.8 Quantum Computing Examples 310
13.9 3D SRF Qubits 310
13.10 Cavity QED Quantum Processors and Memories 312

References 315
List of Symbols 365
List of Acronyms 369
Index 375
 xi

Preface

It has been more than 20 years since Wiley's 1998 publication and enthusiastic
reception of RF Superconductivity for Accelerators [1] and now more than 10 years
since the sequel in 2009: RF Superconductivity – Science, Technology, and Applica-
tions [2]. Many aspects of superconducting RF (SRF) development are thoroughly
covered in these two books, plus many review papers [3–6], and most completely
in the proceedings of international SRF conferences (1980–2021) published on
JACoW.org [7].
Over the period 2010–2022 there has been spectacular progress in terms of the
performance of SRF structures, scientific understanding of the improvements,
innovative cavity designs for new applications, and wide exploration of new mate-
rial avenues to take us beyond the capabilities of the popular standard of niobium,
as well as the large scale, worldwide implementation of the mature technology to
many new accelerators. Exciting new prospects are on the horizon.
It is time for a new volume on RF Superconductivity to provide a comprehensive
update for more than a decade of advances carried out by enthusiastic researchers all
over the world. A large fraction of the progress in SRF performance reported here is a
testament to the creativity and success of imaginative researchers who have worked
on innovative treatments, pursued efforts to gain understanding, and opened the
door to new applications. Our review of the field covers progress till January 2022.
No doubt there will be much additional progress reported in upcoming coming meet-
ings such as Tesla Technology Collaboration (TTC) Meetings, as well as Thin Film
SRF Conferences. We look forward to many new results by the time of the next SRF
Conference in 2023.
Experts as well as newcomers to the field, including students, will benefit from the
discussion of progress, as well as recent and forthcoming applications. Researchers
in accelerator physics may also find much that is relevant to their discipline. There
are now more than a thousand practitioners of the SRF field at more than 150 insti-
tutions and industries worldwide.
The book has four parts. Part I is the introduction and update of SRF funda-
mentals. Many of the SRF basics covered in the first two books will only be briefly
touched, although essentials will be summarized for the sake of completeness.
Part II covers performance advances and understanding at the high Q frontier. Part
III covers performance advances and understanding at the high gradient frontier.
xii Preface

Part IV covers new cavity and new treatment developments, as well as ongoing
applications and future prospects.
An exciting new development discussed briefly in Part IV is the use of SRF cavities
for quantum computing. Nb cavities offer a transformative vehicle for increasing the
coherence times of qubits from sub-milliseconds to seconds, promising to bring the
quantum computing field to quantum advantage over classical computers.

October 26, 2022 Hasan Padamsee

References

1 Padamsee, H., Knobloch, J., and Hays, T. (1998). RF Superconductivity for

Accelerators, 2e, 2008 Wiley.
2 Padamsee, H. (2009). RF Superconductivity: Science, Technology, and
Applications, 2e 2008. Wiley.
3 Kelly, M. (2012). Superconducting radio-frequency cavities for low-beta particle
accelerators. Rev. Accel. Sci. Technol. 5: 185–203. https://doi.org/10.1142/
9789814449953_0007.
4 Belomestnykh, S. (2012). Superconducting radio-frequency systems for high-ß
particle accelerators. In: Reviews of Accelerator Science and Technology:
Applications of Superconducting Technology to Accelerators (eds. A. Choa
and W. Chou) vol. 5, 147–184. World Scientific. https://doi.org/10.1142/
S179362681230006X.
5 Reece, C.E. and Ciovati, G. (2012). Superconducting radio-frequency technology
R&D for future accelerator applications. In: Reviews of Accelerator Science and
Technology: Applications of Superconducting Technology to Accelerators, vol. 5
(ed. A. Chao and W. Chou), 285–312. World Scientific. https://doi.org/10.1142/
S1793626812300113.
6 Gurevich, A. (2012). Superconducting radio-frequency fundamentals for par-
ticle accelerators. Rev. Accel. Sci. Technol. 5: 119–146. https://doi.org/10.1142/
S1793626812300058.
7 SRF Conference Proceedings.
 1

Part I

Update of SRF Fundamentals

 3

Introduction

Discovered in 1911, superconductivity is a fascinating phenomena of modern

physics with marvelous scientific and technological applications, such as powerful
magnets for medical imaging (magnetic resonance imaging [MRI]), for high energy
physics, in particular, the large hadron collider (LHC), for nuclear fusion, and a
wide range of modern applications.
The first major milestone in the history of superconductivity was the discovery
by Kamerlingh Onnes [1, 2] that the electrical resistance of various metals, such
as mercury, lead, and tin disappears when the temperature is lowered below some
critical temperature value, T c . Zero electrical resistance allows persistent currents in
superconducting rings. These currents flow without any measurable decrease up to
one year, allowing a lower bound of 105 years on their decay time. Compared to good
conductors, such as copper, which have a residual resistivity at low temperature of
the order of 10−6 Ω-cm, the resistivity of a superconductor is lower than 10−23 Ω-cm.
Subsequently, Meissner and Ochsenfeld [3] discovered perfect diamagnetism in
superconductors. Magnetic fields are excluded from superconductors. Any field orig-
inally present in the metal is expelled from the metal when lowering the temperature
below its critical value. Expulsion of magnetic field from walls of superconducting
cavities via the Meissner effect will be an important topic in Chapter 4.
Starting with pioneering efforts in the 1960's, RF superconductivity (SRF) finally
catapulted to an enabling technology since the 1980's. SRF has since equipped fron-
tier accelerators in high energy physics, nuclear astrophysics, nuclear physics, as
well as light sources and neutron sources for materials and life sciences. New appli-
cations are coming on line to intense proton sources for neutrino beams, and trans-
mutation of nuclear waste, as well as for deflecting cavities for beam tilts for higher
luminosity at LHC.
The primary advantages of the SRF technology have been discussed in the two
previous books [4, 5]. The most attractive features of applying SRF to particle acceler-
ators lie in the high accelerating gradient, Eacc , possible in continuous wave (cw) and
long-pulse operating modes, along with extremely low RF losses in the cavity walls
at cryogenic temperatures. There is another important advantage. The presence of
accelerating structures has a disruptive effect on the beam, limiting the quality of the
beam in aspects such as energy spread, beam halo, or the maximum current. SRF
systems can be shorter, and thereby impose less disruption to the beam. By virtue of
Superconducting Radiofrequency Technology for Accelerators: State of the Art and Emerging Trends,
First Edition. Hasan Padamsee.
© 2023 WILEY-VCH GmbH. Published 2023 by WILEY-VCH GmbH.
4 1 Introduction

1000
Spoke Elliptical
Frequency (MHz)

100 Half-wave

Quarter-wave

10
β = 0.1 β=1
Particle velocity [β = v/c]

Figure 1.1 Superconducting cavities spanning the full range of particle velocities.
Source: [6]/M. Kelly, Argonne National Lab/with permission from World Scientific Publishing.

low wall losses, SRF cavities can be designed with large beam holes (apertures) to
further reduce beam disruption and allow higher beam currents desirable.
There are two distinct types of superconducting cavities. The first type, TM-mode
cavities, is for accelerating charged particles that move at nearly the speed of light,
such as electrons in a high-energy linear accelerator (linac) or a storage ring. The
second type, TEM-mode cavities, is for particles that move at a small fraction
(e.g. 0.01–0.5) of the speed of light, such as the heavy ions. Structures for these
applications are the quarter wave resonator (QWR), the half wave resonator (HWR)
and the single spoke resonator (SSR), or one with multiple spokes. At intermediate
velocities, both TM and TEM types could be used, depending on the application.
Figure 1.1 [6] shows practical geometry sketches, and typical RF frequencies for
each cavity type, depending on the velocity of the particles spanning the full velocity
range of particles.
The QWR is the compact choice for low-𝛽 applications (𝛽 < 0.15) requiring ∼50%
less structure with less overall RF dissipation compared to the HWR for the same fre-
quency and 𝛽. (Here 𝛽 = v/c, where v is the speed of the particle under acceleration,
and c is the speed of light.) But the asymmetric field pattern in the accelerating gaps
produces vertical steering that increases with velocity. The QWR is less mechani-
cally stable than the HWR due to the unsupported end at the bottom in Figure 1.1.
Hence the HWR is more suitable in the mid-velocity range (𝛽 > 0.15) or where steer-
ing must be eliminated (i.e. for high intensity). It has a symmetric field pattern and
provides higher mechanical rigidity. But the HWR is larger, requires a larger cry-
omodule (CM), and has roughly twice the dissipation for the same 𝛽 and frequency.
The SSR is a more compact variant of the HWR. It opens a path to extension to sev-
eral accelerating gaps along the beam in a single resonator, using multiple spokes.
It provides a higher effective voltage, but with a narrower transit time acceptance.
 1 Introduction 5

This book will mostly focus on a review for the near velocity-of-light, or high-𝛽
accelerating cavities, and to particle accelerators that use these structures. We only
briefly cover some of the latest applications of low-𝛽 structures to major facilities.
For in-depth coverage of low-𝛽 cavities, we refer the reader to excellent articles [6],
and tutorials at International SRF conferences [7, 8].
This book will not cover many important topics in SRF, such as input couplers,
higher-order-mode couplers, tuners, and cryomodules. For latest developments in
these areas, we refer the reader to many papers published in the Proceedings of
the International SRF Conferences. The proceedings are available on the JACoW
website [9].
 7

SRF Fundamentals Review

2.1 SRF Basics

We briefly review the key figures of merit that characterize the performance of an
SRF cavity or structure, referring the reader to [4, 5] for in-depth coverage. The
first important parameter – the accelerating voltage V c – is the ratio of the maxi-
mum energy gain that a particle moving along the cavity axis can achieve, to the
charge of that particle. As all existing high-𝛽 multicell SRF structures operate in a
𝜋 standing-wave mode, the optimal length (active length) of the cavity cells is 𝛽𝜆/2.
Here 𝜆 is the rf wavelength. Next, the accelerating gradient is the ratio of the accel-
erating voltage per cell to the cell length, or Eacc = V c /(𝛽𝜆/2). The cavity quality
factor Q0 determines the number of rf cycles (multiplied by 2𝜋) required to dissipate
the energy stored in the cavity. The key performance factor of an SRF cavity is typ-
ically given by the Q0 versus E curve, showing how rf losses change as the gradient
(Eacc ) rises. The quality factor (Q0 ) is derived as a ratio of two values via Rs = G/Q0 ,
where G is the geometry factor, and Rs is the surface resistivity. As the name suggests,
the geometry factor is determined only by the shape of the cavity. Surface resistiv-
ity (often referred to as surface resistance, Rs ) depends only on material properties
and the rf frequency. The physics of surface resistance is dominated by the physics
of superconductors, and so will be a major topic of the book. The cavity’s shunt
impedance, Rsh , determines how much acceleration a particle can derive from a cav-
ity for a given power dissipation, Pc in the cavity walls. Hence Rsh = V c 2 /Pc . A related
quantity is the geometric shunt impedance Rsh /Q0 , or simply R/Q, which depends
only on the cavity shape. Two other important figures of merit are the ratios Epk /Eacc
and Bpk /Eacc of the peak surface electric field Epk and magnetic field Bpk to the accel-
erating gradient Eacc . The typical distributions of the electric and magnetic field
in a single cell 𝛽 = 1 cavity are shown in Figure 2.1a,b, as well as for a low-𝛽 QWR
in Figure 2.1c. Note that for the single cell 𝛽 = 1 cavity, the magnetic field is maxi-
mum near the equator, whereas the electric field is at a peak near the iris. Maximum
electric field locations for the QWR are shown in red.
For a given accelerating field, both Epk and H pk need to be minimized for a good
design. A high surface electric field can cause field emission of electrons, which

Note: Q0 and Q will be used interchangeably throughout the book.

Superconducting Radiofrequency Technology for Accelerators: State of the Art and Emerging Trends,
First Edition. Hasan Padamsee.
© 2023 WILEY-VCH GmbH. Published 2023 by WILEY-VCH GmbH.
8 2 SRF Fundamentals Review

Beam tube Cell Beam tube

Magnetic
field

Symmetry axis
Electric field

Iris Iris

Magnetic field
(a) Equator

(b)

(c)

Figure 2.1 (a) Electric and magnetic field distributions for a single-cell TM010 cavity.
Source: [10] Courtesy of J. Knobloch, Cornell University. (b) Microwave Studio® [11]
simulations of the electric field (left) and magnetic field (right) in a TM010 mode [12]
Courtesy of D. Bafia, Illinois Institute of Technology. The phase of the magnetic field is 90∘
shifted relative to the phase of the electric field. (c) Electric field (left) and magnetic field
(right) simulation for the QWR [13]. Zhang and Venturini Delsolaro/JACoW/CC BY 3.0.
 2.1 SRF Basics 9

Ideal performance

Thermal instability

Q0
Multipacting
Q-slope
Quench
Field emission

Hydrogen Q-disease
Not to scale
3 6 10 20 40 MV/m
Eacc (MV/m)

Figure 2.2 Typical Q versus E curves obtained for cavities exhibiting various performance
limiting phenomena such as: hydrogen-Q-disease, multipacting, thermal instability (or
quench), field emission, or high field Q-slope (HFQS). The flat curve depicting ideal
performance is rarely (or never) achieved. The X-axis for gradient is not to scale [14].

impact and heat the cavity wall, often leading to a premature breakdown of super-
conductivity (called “quench”). Field emission electrons also generate undesirable
“dark current” in the accelerator. A high surface magnetic field may limit the cav-
ity’s performance at high gradients if rf heating from a high resistance region (such
as a defect) triggers a quench of superconductivity, or if the local field approaches
the critical rf magnetic field, discussed in more detail in later chapters.
The key performance of an SRF cavity is expressed by measuring the Q0 versus
Eacc curve. As shown in Figure 2.2, the Q0 departs from the ideal flat curve due
to limitations arising from various phenomena such as the hydrogen-related
Q-disease, multipacting, breakdown from a defect, field emission, high field Q-slope
(HFQS), and medium field Q-slope (MFQS). Each of these phenomena has been
extensively studied with great progress in understanding the fundamental causes.
Remedies have been developed to overcome the limitations and to return cavity
behavior toward the ideal, flat Q0 versus Eacc curve.
Temperature mapping of the outer wall of the cavity has played a crucial role
in understanding and curing many of these limitations. Figures 2.3 and 2.4 show
the earliest system [15] for rapid mapping the outer-wall temperature below
the lambda point of liquid He (2.2 K). Figure 2.4 also shows a temperature map
when there is heating at a defect that eventually leads to a quench at a higher field.
The thermometry system shown here has been improved [16] and adopted by many
labs [17–19].
The performance of an SRF cavity depends on the maximum values of the peak
surface fields that can be tolerated without increasing the microwave surface resis-
tance substantially, or without causing a breakdown of superconductivity. A high
surface electric field can cause field emission of electrons, degrading the Q0 . A high
surface magnetic field may limit the gradient of the cavity through heating at a
defect followed by thermal runaway (Figure 2.4), or through a magnetic transition
to the normal state at the local critical magnetic field. The ultimate accelerating field
achievable for an ideal Nb cavity is set by the rf critical magnetic field, theoretically
10 2 SRF Fundamentals Review

(a) (b)

Pogostick Mangamin wires

G10 housing Stycast epoxy

GE varnish

Allen-Bradley resistor (100 Ω)

(c) 1 cm 0.4 cm

Figure 2.3 (a) A single thermometer board holding 21 carbon-resistor thermometers. The
shape of the board matches the contour of a 1-cell cavity [10] Courtesy of J. Knobloch,
Cornell University. (b) A single thermometer encased in epoxy. The sensing element is a
100 Ω Allen–Bradley carbon resistor the surface of which is ground down to just expose
the carbon element for higher sensitivity. Source: Courtesy of J. Knobloch, Cornell
University. (c) Schematic of the thermometer housing showing the spring-loaded pogo stick
that helps to keep contact with the cavity wall, and the leads of manganin wire to limit the
stray heat input. The face of the thermometer is painted with insulation.
Source: [10, 16]/with permission of AIP Publishing LLC.

equal to the superheating critical magnetic field [21], H sh . For ideal niobium, H sh
at 2 K is about 0.22 T, which translates to a maximum accelerating field of about
52 MV/m for a typical shape 𝛽 = 1 niobium structure, and roughly 30 MV/m for a
typical 𝛽 < 1 Nb structure.
Other important design features for an SRF structure discussed further in [22] are
cell-to-cell coupling for multicell structures, Lorentz-force (LF) detuning coefficient,
 2.2 Fabrication and Processing on Nb-Based SRF Structures 11

(a)

(b) 0 mK ΔT 120 mK (c) 1 og(ΔT [mK]) 4

Figure 2.4 (a) Thermometers positioned on a cavity wall. Apiezon-N grease promotes
thermal contact between the thermometer and the cavity wall. Some boards are
removed to expose the cavity. [10, 16]. Courtesy of J. Knobloch, Cornell University &
with permission of AIP through CCC. (b) Sample temperature map showing heating at a
sub-mm defect site that leads to quench at higher fields. (c) At higher RF field, the
defect heating grows to cause a quench of superconductivity, and a large region of the
cavity surface around the defect shows high temperatures. Source: [20]/H. Padamsee,
Cornell University.

input power required for beam power (Pb ), coupling strength of input coupler (Qext ),
higher order mode (HOM) frequencies, HOM shunt impedances and HOM Q values.
Mechanical properties also play a role in ensuring stability under atmospheric load-
ing and temperature differentials, to minimize Lorentz-force detuning, and to keep
microphonics detuning under control.

2.2 Fabrication and Processing on Nb-Based SRF

Structures
To appreciate the latest progress in the performance and applications of SRF cav-
ities it is helpful to briefly review the main features of customary fabrication and
processing methods. The short review will help understand how the evolution of
12 2 SRF Fundamentals Review

fabrication and surface treatment practices couple to the solution of the performance
difficulties mentioned above, such as the hydrogen Q-disease, field emission and
quench. More detail information about the fabrication and processing is available
in [4, 5, 22].

2.2.1 Cavity Fabrication

Several industries provide niobium sheets with well-defined cavity specifications
[23]. The sheets are inspected for flatness, uniform grain size (typically 50 μm),
near-complete and uniform recrystallization, RRR value (>300), and good surface
quality (absence of scratches). Here RRR stands for Residual Resistivity Ratio, and
is a measure of the purity of Nb. Since the many fabrication stages can embed
“defects,” such as impurity inclusions, pits, bumps, or scratches, each sheet from
industry is scanned with eddy-current scanning [24, 25] to weed out defective
sheets. Defects can lead to breakdown of superconductivity (quench) either by over-
heating, or by lowering the local critical field, resulting in a magnetic quench. The
high RRR Nb helps to stabilize defect heating due to the high thermal conductivity
that accompanies the high RRR.
For a 𝛽 ∼ 1 structures, half-cells are stamped, spun, or hydro-formed, checked with
the coordinate measuring machine (CMM) for the correct shape, then trimmed for
weld preparations. Cavity parts are given a light (20 μm) (Buffered Chemical Polish)
BCP etch to prepare for electron beam welding. Electron beam welding is a critical
process with carefully developed parameters. A smooth weld under bead with com-
plete absence of spatter is essential for high field performance. This can be achieved
with defocused electron-beam welding [26], or by using a raster with a rhombic or
circular pattern as described in [27]. To avoid RRR degradation, the vacuum in the
electron-beam welder should be better than 2 × 10−5 Torr. All welds are inspected
for complete, smooth under bead, flat on the inside, and no weld spatter. After com-
pleting a single-cell or multicell structure, the inside surface is inspected optically.
A special optical inspection apparatus has been developed and widely adopted [28].
Mechanical measurements ensure straightness and correct dimensions. The elec-
tric field profile along the beam axis is checked and adjusted. The goal is usually
98% field flatness. A “flat” field profile is achieved by tuning the cells relative to
each other by squeezing or stretching the cells mechanically to adjust and properly
match the frequency of each cell.
Most low-𝛽 resonators are made from bulk niobium with high RRR (150–300).
Fabrication of parts include machining, forming, rolling, and welding. Recently, wire
electric discharge machining (EDM) has been developed together with industry [6]
which has little possibility for foreign material inclusions as compared to traditional
machining. Parts are joined together by electron beam welding in high vacuum.

2.2.2 Preparation
Niobium cavities undergo a first stage etching (100–150 μm) to remove the “surface
damage” layer. Methods used for material removal are standard buffered chemical
 2.2 Fabrication and Processing on Nb-Based SRF Structures 13

polishing (BCP), electropolishing (EP) [29], and centrifugal barrel polishing (CBP)
[30]. By far the best method proven is EP [31], giving the smoothest surface
(roughness <0.3 μm) and leading eventually to highest gradients.
BCP is a technically simpler process and used for pre-etching parts for cleaning
and welding. BCP is chemical etching with a mixture of HF (40% concentration),
HNO3 (65%), and H3 PO4 (85%) acids in a volumetric ratio of 1 : 1 : 2. The process
is exothermic so that good heat exchange and stirring are necessary for uniform
material removal, and to keep the acid temperature below 15 ∘ C to prevent excess
hydrogen take-up (Section 6.11). BCP yields sharp grain boundary steps of 1–2 μm,
and several μm at the weld because the etch rate depends on the crystal grain
orientation. Such sharp steps are undesirable due to local field enhancement and
lower quench fields.
EP is carried out with an acid mix of HF (40%) and H2 SO4 (98%) in a ratio of 1 : 9
as the electrolyte. The niobium cavity serves as the anode and a high-purity alu-
minum cathode is inserted into the cavity. A typical arrangement has the cavity and
the cathode in horizontal orientation with electrolyte filling about 60% of the cavity.
The assembly is slowly rotated to allow uniform etching and polishing of the surface.
The acid is circulated to an external reservoir where it is also cooled. Nitrogen gas is
circulated via the cathode structure to expel hydrogen produced. Steps at the grain
boundaries are reduced to below 0.2 μm (Figure 2.5). After EP, removal of sulfur

BCP
R1 = 1.45 µm

Mag. 5.1 ×
Nb Sample
BCP Etch
~ 60 µm
Ra = 1.45 µm

(a)

EP R1 = 0.31 µm

Mag. 5.1 ×
Nb Sample
BCP Etch
~ 60 µm
+ EP 144 µm
Ra = 0.31 µm

(b) 1.2 mm

Figure 2.5 Comparison of surface roughness Nb surface treated with (a) BCP and (b) EP.
Source: [35]/Courtesy of L. Popielarski, FRIB, MSU/U.S. Department of Energy, Office of
Science.
14 2 SRF Fundamentals Review

residues [32] must be carried out with ultrasonic degreasing for a couple of hours in
ethanol [33] or in detergent and water [34].
Tumbling or CBP is used to remove irregularities at welds, as well as pits, and
scratches. The slow rate of material removal is highly dependent on the tum-
bling medium and rotation speeds. The final finish can be mirror-like, with 10 nm
roughness, but the mirror quality by itself has not been shown to yield higher
performance than EP. A final stage of light (5–20 μm) chemical etching via EP is still
necessary to remove the tumbling abrasive embedded in the surface [36]. CBP has
successfully been used to repair cavities with mechanical defects, such as pits, and
bring them to high field performance [36, 37]. All chemical treatment or mechanical
abrasion admits hydrogen into the bulk, leading to a Q-disease (Section 6.11).
After chemical etching and rinsing, high pressure water rinsing (HPR) is carried
out at 100 atm water pressure for several hours to remove chemical residues and
attached particles to avoid field emission or thermal breakdown. HPR is effective
in scrubbing the surface free of impurities from chemical processing. The water
must be particulate-free (using 0.1 μm or better filters) deionized water at a nozzle
pressure of ∼100 bar to remove chemical residues [38, 39]. HPR is performed in
an ISO5 (Class 100) or better clean room to prevent dust contamination during
the process. Dust particles and chemical surface contamination results is heavy
field emission, so both HPR and clean room environments are essential for good
performance.
All preparation procedures for bulk material removal (BCP, EP, CBP) carry a risk
of H evolution and absorption so that a furnace treatment at 600–800 ∘ C is necessary
to remove the H and to avoid the Q-disease. As shown in Figure 2.2, Q-disease will
cause the Q to fall at low fields due to the formation of niobium hydrides during
cool-down (Section 6.11). After furnace treatment, the cell-to-cell field profile is
remeasured and readjusted. The final chemical treatment is a light etch (about 20 μm
material removal) by EP to reach the highest field levels.
After final EP, the cavity is transported to the clean room where the inside surface
is once again given a high-pressure rinse with high purity water jets for many hours.
The cavity dries in the clean room.
A final HPR takes place after assembly of the necessary flanges, and the field mon-
itor probe. The cavity is then ready for evacuation for rf tests. Great care must be
exercised to avoid recontamination during the subsequent cavity handling, compo-
nent assembly, and installation [40].
To reach the highest fields, an electropolished cavity needs to be baked at 120 ∘ C for
48 hours [17, 41]. The mild baking step provides several benefits: first and foremost
to remove the HFQS (Section 6.4). Chapter 6 discusses new research to understand
the origins of HFQS and physics of the 120 ∘ C bake benefits. Additional benefits of
baking are to reduce the processing time for multipacting by degassing water from
the surface, and lowering the secondary electron emission coefficient, as well as to
reduce the BCS surface resistance for higher Q values by lowering the electron mean
free path (Section 2.3.5). The bake is normally carried out with the inside of the cavity
in a good vacuum (∼10−8 Torr).
 2.3 SRF Physics 15

2.2.3 A Decade of Progress

The decade 2010–2020 has brought enormous progress to the physics, technology,
and applications of SRF cavities, the major rationale for this book. Here we summa-
rize some of the highlights. Typical Q versus E curves for niobium cavities shows
three distinct regions of Q-slope changes (Figure 2.6 [a]): low, medium, and high
field Q-slopes [5], abbreviated as low-field Q-slope (LFQS), MFQS, and HFQS. The
book reviews progress in understanding these Q-slope regimes.
Unprecedented Q values (>2 × 1011 at 1.3 GHz and 1.5 K) have been attained up
to Eacc = 20–30 MV/m. These advances were achieved by novel surface preparation
techniques, such as nitrogen doping (Chapter 3), and 300 ∘ C (mid-T) baking
(Chapter 4), along with special cavity cool-down procedures to eliminate the
residual resistance contribution from trapped dc magnetic flux (Chapter 5). These
recent accomplishments have translated into significant increases (factors >2–3) in
the efficiency of cw particle accelerators (e.g. LCLS-II at SLAC) operated at medium
accelerating fields of about 20 MV/m.
On the high gradient frontier, new treatments of nitrogen infusion (Section 7.3)
and two-step baking (75/120C) (Section 7.1) have paved the way for gradients
near 50 MV/m. The benefits of these new discoveries are collected in a Q ver-
sus E collage shown in Figure 2.6b. The proximity effect, nano-hydride model
(Section 6.13) for the HFQS, and the 120 ∘ C mild baking cure, which inhibits the
formation of the harmful hydrides, have gained much experimental support. The
model provides a platform to account for the gradient improvements with new
techniques.
Another mystery from the past, the LFQS, is now fully understood as originat-
ing from two-level states (TLS) in the niobium pentoxide (Section 4.6). A technique
(340 ∘ C baking) has been found to substantially remove the oxide and reduce the
TLS losses to achieve record low residual resistance values under one nano-Ohm.
A very important benefit of the advances in the low field arena is to improve the
lifetime of qubits (Section 13.9) opening the door to higher coherence times for quan-
tum computing (Chapter 13).

2.3 SRF Physics

2.3.1 Zero DC Resistance
The two-hallmark features of superconductivity are zero resistance for dc currents
and the Meissner effect. An early treatment for superconductivity comes from the
London equations [43, 44] that account for these two salient properties. The London
brothers provided a phenomenological description of superconductivity based on a
two-fluid type concept proposed by Gorter and Casimir [45]. Here the superfluid
and normal fluid densities ns and nn are associated with velocities vs and vn . The
densities satisfy

ns + nn = n (2.1)
16 2 SRF Fundamentals Review

2.0×1010

1.5×1010
Low field Effect of 120°C bake
Q slope Medium field
Q slope
Q0

1.0×1010

Quench

5.0×109
High field
Q slope
0.0
0 20 40 60 80 100 120 140 160 180
(a) Bpeak (mT)

1010
Q0

75/120°C Mode Bake

N-Doped
LTB-120Cx48hr
N-Infused
EP
109
0 10 20 30 40 50
(b) Eacc (MV/m)

Figure 2.6 (a) Low field, medium field, and high field Q-slopes observed after standard
treatment of EP. 120 C bake removes the HFQS, leaving an extended region of MFQS [42]
Courtesy of A. Grassellino, Fermilab. (b) Q0 versus E acc at 2 K of 1.3 GHz Nb SRF cavities
treated at FNAL with state-of-the-art surface treatments, such as nitrogen-doping, two-step
baking, and nitrogen infusion, compared to standard treatments of EP or EP followed by
120 C baking [12] Courtesy of D. Bafia, Fermilab.
 2.3 SRF Physics 17

where n is the average electron number per unit volume. The two current densities
satisfy
Js = −ens vs, Jn = −enn vn (2.2)
We can think of the first and second London equation pair below (2.3) as describ-
ing the electric and magnetic fields inside a superconductor
( )
⃗= 𝜕
me
E ⃗s
J
𝜕t ns e2
m
⃗ = − e ∇ × J⃗s
B (2.3)
ns e2
The electric field equation of the pair accounts for zero resistivity in superconduc-
tors that carry a direct current (dc) because using the supercurrent component of
Eq. (2.2), it can be re-expressed as:
𝜕v
m = −eE (2.4)
𝜕t
Equation (2.4) implies that in the presence of an electric field, E, the electrons
in an ideally perfect conductor are accelerated freely, with zero resistance. The first
London equation also implies that when the current is constant in time, as in the
case of dc, there is no electric field inside the superconductor, and conduction takes
place without losses.
𝜕js
= 0, E = 0 (2.5)
𝜕t
Equation (2.5) should be contrasted with the conductivity equation for a normal
metal or Ohm’s Law. In the Drude formulation [46, 47]
nn e2 𝜏
jn = E = 𝜎n E (2.6)
m
n e2 𝜏 n e2 𝓁
𝜎n = n = n (2.7)
m mvF
where 𝜏 is the average time between collisions or the electron relaxation time, m the
electron mass, 𝓁 is the electron mean free path, and vF is the Fermi velocity.

2.3.2 Meissner Effect

Using Maxwell’s equations with the London equations, we derive the Meissner
effect. Starting with Maxwell’s equation
⃗ B = 𝜇0 J⃗s
∇× (2.8)
Apply the curl
ns e2
⃗ × B = 𝜇0 ∇ × J⃗s = −𝜇0
∇×∇ B (2.9)
me
Using the second London equation for B,
ns e2
⃗ = −𝜇0
∇(∇ ⋅ B)−∇2 B B (2.10)
me
18 2 SRF Fundamentals Review

Since ∇ ⋅ B = 0, this leads to a wave – type differential equation

ns e2 1 ⃗
⃗ = 𝜇0
∇2 B B= 2B (2.11)
me 𝜆L
where

𝜆L = (m∕ns e2 𝜇0 )1∕2 (2.12)

𝜆L is defined as the London penetration depth. (The value is 39 nm for high-purity

niobium.) In the case of a semi-infinite superconducting slab that occupies the space
z > 0, the solution to Eq. (2.11) is
⃗ (z) = B0 e−z∕𝜆L x̂
B (2.13)

which states that the parallel component of the B field decays exponentially with
distance from the surface, to satisfy B = 0 inside the superconductor. This is the
Meissner effect.
There are several consequences of the Meissner effect – important for Chapter 4.
When a superconducting material is cooled through its transition temperature,
T c , in the presence of an external dc magnetic field, the magnetic flux is abruptly
expelled from the volume of the superconductor when the ideal conditions are
fulfilled. (As we will see for Nb cavities, this condition is met only in certain
circumstances.) The second aspect of the Meissner effect: An externally applied
magnetic field (less than the critical field) cannot penetrate into the interior of
a superconductor when it is below T c . The field is excluded from the volume of
the superconductor. A metal in the superconducting state therefore never allows
magnetic flux density to exist in the interior. In a superconductor,

𝜕B∕𝜕t = 0 inside and B = 0 inside (2.14)

The external magnetic field is confined to a thin layer characterized by the pene-
tration depth, 𝜆L , near the surface of the superconductor. The London penetration
depth sets the characteristic decay length for the magnetic field. Any supercurrents
in the superconductor are also restricted to flow in this near-surface region.
Physically, the exponentially decaying magnetic field can be explained by
“screening currents” that are excited near the surface and flow with zero resistance,
generating their own magnetic fields that perfectly cancel out the applied field far
from the surface.
The London equations can also be motivated from a quantum-mechanical argu-
ment. In the absence of a magnetic field, the ground state has zero net momentum,
with the canonical momentum:
−eA
p = mv + =0 (2.15)
mc
with m the electron mass and A is the vector potential. Thus the local average veloc-
ity becomes
−eA
⟨vs ⟩ = (2.16)
mc
 2.3 SRF Physics 19

As before (Eq. [2.2]), the current density Js can be expressed as

−ns e2 A −A
Js = −ns e⟨vs ⟩ = = (2.17)
mc 𝛬c
m
𝛬= (2.18)
ns e2
Note that in this case, the relationship between the supercurrent and the vector
potential is “local,” i.e. the supercurrent depends on the local value of the vector
potential only. Taking the time derivative of both sides, leads to the first London
equation, as before:
𝜕
E= (𝛬Js ) (2.19)
𝜕t
By taking the curl of Eq. (2.19), one obtains the second London equation, as
follows:
𝜕 𝜕
∇ × E = − (B) = ∇ × (ΛJs ) (2.20)
𝜕t 𝜕t
B = −∇ × (ΛJs ) (2.21)

2.3.3 Surface Resistance and Surface Impedance in RF Fields

In the earliest and simplest picture of the two-fluid model [45], the current flow-
ing through a superconductor under the influence of an rf field can be separated
into two noninteracting fluids: a “superfluid” that moves with zero resistance, and
a “normal fluid” that experiences dissipation. As the temperature of the supercon-
ductor decreases below the transition temperature, Tc , the density of the normal
fluid carriers (nn ) decreases, and the density of the superfluid (ns ) carriers increases.
The total density remains constant so that ns + nn = n0 . In the two-fluid model,
the temperature dependence of the superfluid density is postulated to be
( ( )4 )
T
ns = n0 1 − (2.22)
Tc

The penetration depth has been empirically shown [48] to have the same temper-
ature dependence observed in the density of superfluid electrons in the two-fluid
model:
𝜆L
𝜆= √ (2.23)
( )4
T
1− T
c

As seen in Figure 2.7a, the penetration depth is approximately independent of

temperature until T/Tc ≈ 0.6. Above this temperature, it quickly diverges toward
infinity at Tc .
In Section 2.3.1 we discussed dc resistance for dc currents. For the case of rf cur-
rents, let
js = js0 ei𝜔t (2.24)
20 2 SRF Fundamentals Review

λ L(T) 4 Figure 2.7 (a) Variation of

–––– penetration depth versus
λ L(0) 3.5
reduced temperature T/T c [12]
3
Courtesy of D. Bafia, Illinois
2.5 Institute of Technology. (b)
2 Variation of the penetration
1.5 depth with the mean free path.
Source: [48, 49]/with
1
permission of the Royal
0.5 Society.
T
––
(a) 0.2 0.4 0.6 0.8 1 1.2 Tc

10
106λ0 (cm)

0 20 40 60 80
(b) 106l (cm)

From the first London equation, we obtain

𝜕js 1 1
= E, i𝜔js = E (2.25)
𝜕t 𝜇0 𝜆2L 𝜇0 𝜆2L
giving an expression for the supercurrent component:
−i
js = E = −i𝜎s E (2.26)
𝜔𝜇0 𝜆2L
Using the definition of 𝜆2L from Eq. (2.12), we obtain:
1 ns e2
𝜎s = = (2.27)
𝜔𝜇0 𝜆2L m𝜔
to give the total two-fluid current as:
J = (𝜎n − i𝜎s )E (2.28)
For alternating fields,
E = E0 ei𝜔t (2.29)
∇ E = 𝜇0 𝜔i(𝜎n − i𝜎s )E
2
(2.30)
Consequently, the surface impedance of a superconductor is given by
√
i𝜔𝜇0
Zs = (2.31)
𝜎n − i𝜎s
To derive a simpler expression for Zs , a useful approximation is:
𝜎n ≪ 𝜎s (2.32)
 2.3 SRF Physics 21

i.e. the conductivity of the normal fluid is very much smaller than the conductivity
of the superfluid, for the two reasons discussed below. From
nn e2 𝜏 n e2
𝜎n = and 𝜎s = s (2.33)
m m𝜔
We note first that at T ≪ T c the number of unpaired electrons, nn , is very much
smaller than the number of paired electrons, ns . Secondly, for the normal conducting
electrons, the relaxation time (10−14 s) between collisions is very much smaller than
the rf period (10−9 s), i.e.
𝜏 ≪ 1∕𝜔 (2.34)
In analogy with normal conductors,
∇2 E = 𝜏 2 E (2.35)
where
√
𝜏= 𝜇0 𝜔i(𝜎n − i𝜎s ) (2.36)
We define the real and imaginary parts of the impedance
Zs = Rs + iXs (2.37)
Using the above approximations, and with some complex number algebra manip-
ulations (given in a separate Section 2.3.11 to avoid interrupting the flow of con-
cepts), the real and imaginary parts of the impedance become
1
Rs = 𝜎n 𝜔2 𝜇02 𝜆3L and
2
Xs = 𝜔𝜇0 𝜆L (2.38)
We can rewrite the surface resistance as:
𝜎
Rs ≈ n 2 (2.39)
𝜆L 𝜎 s
Note the important consequences of Eq. (2.38) are the two salient features of the
surface resistance of a superconductor:
Rs ∝ 𝜔2 and Rs ∝ 𝜎n (2.40)
Thus the superconducting state surface resistance (i) increases as the square of
the rf frequency and (ii) is proportional to the dc normal state conductivity, an unex-
pected trend.
According to Eq. (2.7), 𝜎 n is directly proportional to the electronic mean free
path, 𝓁, so Rs increases with 𝓁; hence dirty superconductors (small 𝓁) show lower
SRF dissipation. For example, a niobium cavity made with high purity material and
longer mfp will have a lower Q than a cavity made from niobium with impurities
that lower the mfp. A simple way to interpret this unexpected trend is as follows:
when the normal electrons have a higher conductivity, they draw a relatively
higher fraction of the current, contributing more to surface resistance, and thereby
increasing the SRF surface resistance.
The electron mean free path 𝓁 (mfp) represents the average distance traveled
by an electron moving through the material at the Fermi velocity vF between
22 2 SRF Fundamentals Review

subsequent scattering events. The mfp is dependent on material properties, such

as the interstitial impurity content. The mfp will feature prominently in Chapter 3
where N-doping introduces N as an interstitial impurity to have a substantial effect
on cavity performance.

2.3.4 Nonlocal Response of Supercurrent

From microwave wave absorption measurements of the penetration depth in
samples of Sn doped with In, Pippard and Bragg [48] observed that impurities
increase the penetration depth of superconductors, as shown in Figure 2.7b above.
But the London penetration depth picture contains no mfp dependence. In the
London model, Eq. (2.17) shows that the current density depends on the local value
of vector potential. Therefore, the London model is called “local.”
To develop an explanation for the mfp dependence of the penetration depth, Pip-
pard proposed a new “nonlocal” description to relate the supercurrents to an average
of the vector potential calculated over a region. He defined the size of this region by
the “coherence length,” 𝜉, which also depends on the mean free path 𝓁. Pippard
generalized the London theory to include these nonlocal effects in analogy with the
anomalous skin effect in normal metals [50, 51]. In this effect, the microwave surface
resistance of clean normal metals reaches an anomalous limit instead of continu-
ing to fall steadily with temperature, as does the dc resistivity. Pippard realized that
the saturation occurs when the electron mean free path becomes larger than the
skin depth so that only those electrons moving parallel to the sample surface remain
effective in responding to the electric field. The local model is applicable when the
mfp is short.
Pippard introduced a similar nonlocal effect for superconductors. The current in
a superconductor at a given point depends on the electric field at neighboring points
through which the electrons have traveled since last being scattered. Thus the super-
conducting current depends on the electric field at distant points – the range being
limited by an electromagnetic coherence length, 𝜉. The response of the supercurrent
to the vector potential is smeared out over a volume.
With such a description, the local form of the proportionality between j(r) and
A(r) in Eq. (2.17), repeated below
1 e2 ns (T)
Js (r) = − A(r), 𝛬(T) = (2.41)
c𝛬(T) m
is changed to the nonlocal relation based on a modification to Ohm’s law similar to
that adopted for the anomalous skin effect. In the nonlocal description,
′
3 R[R ⋅ A(r )]e−R∕𝜉 ′
J(r) = − dr (2.42)
4𝜋𝜇0 𝜆 𝜉0 ∫
2 R4
′
where R = r − r , takes into account that the current at a point r depends on the
′
vector potential at a point r within a distance 𝜉 0 , called the coherence length. Thus
⇀ ⇀
⃑ over a range 𝜉. Figure 2.8 contrasts the
J ( r ) depends on a weighted average of A
nonlocal and local behaviors for one dimension.
 2.3 SRF Physics 23

A A

ξ
ξ

x x
λ λ
(a) Local (London) (b) Non-local

Figure 2.8 (a) In the London (local) model, the supercurrent density depends on the local
value of the vector potential, A (b) In Pippard’s nonlocal model, the supercurrent density
depends on the vector potential over a region with size defined by the larger coherence
length, 𝜉. Source: [52]/with permission of Dale J. Van Harlingen.

Because of the nonlocality, the new length scale, 𝜉 can change due to scattering
with impurities. Assuming that the average distance it takes to scatter electrons is
the electronic mean free path (𝓁), the coherence length is modified by
1 1 1
= + (2.43)
𝜉 𝜉0 l

Using the Maxwell’s equation ∇2 A = −𝜇0 J, the nonlocal equation between J and
A can be solved for the penetration depth 𝜆, which is defined as:
∞
1
𝜆= Hy dz (2.44)
Hy (0) ∫0
The nonlocal result is simplified in the two limits:
√ √
𝜉0 𝜉
Short mfp∶ 𝜉 ≪ 𝜆 ∶ 𝜆 = 𝜆L = 𝜆L 1 + 0 (2.45)
𝜉 𝓁
( )1∕3
Long mfp∶ ξ ≫ 𝜆 ∶ 𝜆 = 0.62𝜆2L 𝜉∘ (2.46)

Nonlocality over the coherence length modifies the penetration depth depend-
ing on the impurity concentration, and the mfp. Thus, material purity varies the
effective size of superconducting carriers and the effective distance into which the
penetration depth extends. For 𝜉 ≪ 𝜆, the penetration depth increases as the mean
free path decreases, while for 𝜉 ≫ 𝜆, 𝜆 becomes independent of mfp, in agreement
with the experimental result shown in Figure 2.7b. Pippard’s 𝜆 is sometimes called
the effective penetration depth to distinguish it from the London’s definition. The
London local theory can be considered as the limit of the Pippard theory for dirty
superconductors (𝜉 0 ≫ 𝓁).
After the BCS theory (Section 2.3.5) it was realized that the nonlocality originates
from the fact that superconductivity is carried by Cooper pairs formed by two elec-
trons that can be hundreds of nanometers apart. If they are far apart, two electrons
of the same pair “feel” very different values of the vector potential.
Pippard estimated the scale of 𝜉 from the uncertainty principle. By assuming that
only electrons within a very thin shell of thickness kb Tc around the Fermi surface
24 2 SRF Fundamentals Review

partake in superconductivity, they should have momentum Δp ≈ kB T c /vF . By the

uncertainty principle
ℏ ℏvF
Δx ⩾ ≈ (2.47)
Δp kB Tc
The coherence length becomes
ℏv
𝜉0 ≈ a F (2.48)
kb Tc
Pippard found that he could fit the experimental data for tin and aluminum by
using the same value of a = 0.15. Later BCS confirmed this result by deriving the
value a = 0.18.

2.3.5 BCS
Many brilliant physicists – Bohr, Einstein, Feynman, Born, and Heisenberg – tried to
develop a microscopic theory of superconductivity; but only Bardeen, Cooper, and
Schrieffer (BCS) [53, 54] were successful in 1957 with their Nobel prize-winning
theory. This was almost 50 years after the experimental discovery of superconduc-
tivity by Kamerlingh–Onnes!
One of the basis of the BCS theory was the fundamental theorem [55] that for
a system of many electrons at small T where thermal excitations are feeble, any
weak attraction, no matter how small, can bind two electrons together, forming the
so-called “Cooper pair”.
There were two key experimental discoveries that made some of the main
properties of superconductivity transparent before the BCS theory came along. The
exponential decay of the specific heat at low temperatures showed that the energy
spectrum of a superconductor must have a gap. The first experimental evidence for
a gap is due to Corak et al. [56] who measured the specific heat of a superconductor.
Below T c the specific heat has an exponential behavior
cs ≈ a𝛾Tc e−bTc ∕T (2.49)
( )
with b ≈ 1.5, whereas in the normal state cn ≈ 𝛾T where 𝛾 = (𝜋 2 ∕3)kB2 g(𝜀F ) is the
coefficient of the normal state electronic specific heat.
The second key was the isotope effect. The transition temperature of materials con-
taining a different element isotope depends on the mass M of the isotope, as M −1/2 .
This makes the critical temperature and critical field larger for lighter isotopes. Since
M is related only to the ions in the lattice, the isotope effect indicates that the lat-
tice must play a key role in the formation of the superconducting state via lattice
vibrations (phonons). Frölich suggested that the mechanism behind the weak attrac-
tive force binding the Cooper pairs is the same mechanism responsible for electrical
resistivity in metals, i.e. the interaction of conduction electrons with lattice vibra-
tions. These observations led to the role of the phonons in superconductivity [57].
An illustrative picture for the BCS theory follows. An electron interacts with the
lattice by virtue of the Coulomb attraction with the metallic ions. The lattice deforms,
generating phonons. A second electron in the vicinity of the deformed lattice lowers
its energy, resulting in an overall electron–electron attraction via a phonon. The
attractive interaction results in a negative potential energy contribution, lowering
 2.3 SRF Physics 25

the total energy of the electron system. The negative potential energy associated
with a Cooper pair is the binding energy of that pair. In the quantum-mechanical
picture, the attractive interaction is due to exchange of lattice phonons between
electrons, in analogy with exchange of photons that mediate the electromagnetic
force.
The main element of the BCS theory is that the Cooper pairs are bound states
formed by two electrons of opposite spins and opposite momenta. The Cooper pairs
form a coherent macroscopic ground state with a gapped spectrum. The energy gap
between the BCS ground state and the first excited state is one of the most remark-
able features of the BCS theory. The gap energy is the minimum required to create an
excitation from the superconducting ground state. The binding energy of a Cooper
pair is two times the energy gap.
The size of the gap depends on the strength of the electron–phonon interaction;
hence it is different for each material. The energy gap also depends on the tempera-
ture: as the temperature increases from 0 to T c , the gap slowly decreases, dropping
rapidly to zero near T c . A useful approximate formula [58] describing the tempera-
ture dependence of the gap is:
√ ( )
Δ(T) 𝜋t2
= cos (2.50)
Δ0 2
Cooper pairs are Bosons and therefore condense according to Bose–Einstein statis-
tics when the temperature decreases below the critical temperature T c . With charge
q = 2e, Cooper pairs propagate through the lattice carrying current, but with zero
resistance. They are analogous to the superfluid carriers in the two-fluid model.
When current flows in the superconductor, each Cooper pair acquires the same
momentum. The unpaired electrons behave as normal electrons in the two-fluid
model, and are called quasiparticles. The number density of quasiparticles follows:
( )
Δ(T)
nn ∝ exp − (2.51)
kB T
Cooper pairs form when electrons interact with each other within a certain dimen-
sion, the coherence length 𝜉 0 . Therefore, the coherence length may be thought as
the dimension of a single pair. BCS estimated its magnitude as:
ℏvF ℏvF
𝜉0 = ≈ (2.52)
𝜋Δ0 𝜅B Tc
in agreement with Pippard (Eq. [2.47]). The separation between electrons in a
Cooper pair is large enough that millions of other pairs have their centers of mass
positioned between them so that Cooper pairs substantially overlap with each other.
In the normal ground state of a metal, the kinetic energy (and hence the
total energy) of the system is minimum when the momenta of the plane wave
states of the electrons fill up a sphere of radius pF = ℏkF (i.e. a Fermi sphere) in
three-dimensional momentum space. Figure 2.9a illustrates a Fermi sphere of
radius kF in three-dimensional k-space for a free electron gas at T = 0 ∘ K. Electrons
well inside the Fermi sphere cannot scatter to higher levels inside the sphere
because of the Pauli exclusion principle. The corresponding normal density of
states, N(E), is shown in Figure 2.9b. Here N(E)dE is the number of electron states
26 2 SRF Fundamentals Review

N(E)
kz N(E) ∞ E½

f(E)
N(0)
kz
ky

kx
E
EF EF = µ E
(a) (b) (c)

Figure 2.9 (a) Fermi sphere for the free electron gas for a normal metal at T = 0 K.
(b) Electron density of states. (c) The Fermi distribution function showing the occupation
of states. The shaded region represents occupied states at T = 0 K.

with energy between E and E + dE. N(0) denotes the density of states at the Fermi
surface at absolute zero. The density of states (DOS) is essentially the number of
different states at a particular energy level that electrons are allowed to occupy.
The Fermi distribution function for T → 0 is the 𝜃-function, meaning that all the
states are occupied up to the Fermi energy, EF = 𝜇, where 𝜇 is the chemical potential,
as shown in Figure 2.9c.
1
f (E, T) = lim f (E, T) = 𝜃(𝜇 − E) (2.53)
e(E−𝜇)∕T + 1 T→0
The phonon-induced attractive interaction affects only those electrons in the
vicinity of the Fermi surface. As already pointed out by Pippard, the fraction of
electrons that cooperate in superconductivity (i.e. in the formation of Cooper
pairs) are those with energy 𝜅 B Tc ≈ ℏ𝜔D around the Fermi level (EF ), where 𝜔D
is the Debye frequency (when the wavelength of the phonon frequency reaches
the smallest length, i.e. of the unit cell.). In BCS, electrons that are inside such an
energy shell are pushed toward lower energies, and condense into Cooper pairs
at the ground state with energy (EF − Δ0 ). The energy gap has the same symmetry
as the Fermi surface of the normal state (Figure 2.10 a). To scatter electrons well
below the Fermi level into unoccupied states requires phonon frequencies much
larger than that generated from the electron interaction.
Formation of the gap substantially modifies the density of states N(E) to intro-
duce singularities at the gap edges (Figure 2.10b). Since Cooper pairs follow the
Bose–Einstein distribution, a virtually infinite number of them can occupy the same
energy level, hence the DOS of a superconductor takes the form:
E
N(E) = N(0) √ (2.54)
E2− Δ2
where singularities arises when |E| = Δ.
The DOS can be measured by point contact tunneling spectroscopy (PCTS)
as presented in Sections 3.7.3 and 6.15. During the measurement, the tunneling
current depends only on the charge density within eV below the Fermi surface,
where (−V) is the applied bias voltage. Thus PCTS can measure the integrated
density of full states below the Fermi level. By modulating the bias voltage by dV
(typically a few mV) around the dc voltage V of interest, a current modulation dI
 2.3 SRF Physics 27

kz N(E)

Δ
kz
ky
N(0)
kx
E
(a) (b) EF –Δ EF EF +Δ

N(E)

N(0)
Thermally excited
quasiparticles
E
(c) EF –Δ EF EF +Δ 1
e(E–EF)/KBT + 1

Figure 2.10 (a) Fermi surface (dotted line) for the ground state of a BCS superconductor at
T = 0 K. Source: Adapted from Sonier [59]. (b) The density of states with singularities at the
edges due to formation of the gap. The shaded region represents occupied states of the
superconducting electrons [59]. (c) The density of states at T > 0 K. When Cooper pairs are
broken the resulting normal electrons occupy states above E F + Δ. [59] following the
Fermi–Dirac distribution as a function of the energy relative to the Fermi level.

can be measured to give the conductance g(V) = dI/dV, which is proportional to

the DOS.
As shown in Figure 2.10, a superconductor is characterized by the density of states,
and the distribution function of the electrons f (E). The DOS exhibits an energy
gap due to Cooper pair formation, and the distribution function is the Fermi–Dirac
Distribution in thermal equilibrium.
With 𝜔c as the cutoff frequency for pair attraction, and N(0) as the density of states
at the Fermi level for electrons of one spin orientation, and V as the potential of the
attractive interaction, BCS solved the self-consistent gap equation at zero Kelvin
in the weak coupling limit, kb T c ≪ ℏ𝜔c ⟩, N(0)V ≪ 1, to obtain the relationships
between the transition temperature, phonon frequency, and superconducting gap:
− N 1V
2Δ0 = 3.53kb Tc = 4ℏ𝜔c e 0 (2.55)
Since N(0) and V are material-dependent parameters, the energy gap varies with
materials. Hence the ratio Δ0 /kb T c varies from superconductor to superconductor
and is usually between 1.5 and 2.2 [21].
BCS addressed the electrodynamic properties of superconductors using the
Pippard nonlocal approach. Mattis and Bardeen [60] derived expressions for the
28 2 SRF Fundamentals Review

conductivity and from there the rf surface resistance and reactance. (The SRF
community refers to the Mattis–Bardeen surface resistance as the BCS surface
resistance.) Their treatment determines dissipation from two mechanisms: thermal
excitation of quasiparticles, and absorption of photons with energy greater than the
energy gap. The second term does not appear unless ℏ𝜔 > 2Δ and is therefore not
relevant for SRF cavities which generally have frequencies ℏ𝜔 ≪ 2Δ.
The Mattis–Bardeen theory does not consider the field dependence of surface
impedance which is further explored in Section 3.10. In particular, the decrease
of surface resistance at high fields after N-doping (Chapter 3) is of great interest
in SRF applications, for example, to explain the unexpected increasing Q versus E
(anti-Q-slope).
In the BCS picture of the microwave surface resistance, the Mattis–Bardeen theory
shows that Rs is proportional to the following integral:
∞
Rs ∝ |M 2 |N(𝜀)N(𝜀 + ℏ𝜔)[f (𝜀) − f (𝜀 + ℏ𝜔)]d𝜀 (2.56)
∫Δ
Here N(E) is the quasiparticle density of states, evaluated at quasiparticle
energies 𝜀 and 𝜀 + ℏ𝜔 representing the absorption of one rf photon of energy
ℏ𝜔. The function f (E) is the quasiparticle distribution function. M is the matrix
element for the interaction. The BCS resistance can be accurately calculated from
Mattis–Bardeen/BCS theory using a computer code developed by J. Halbritter,
which solves the BCS equations [61]. An online version of this code is available as
SRIMP [62]
For frequencies smaller than the gap frequency, energy from the electromag-
netic field is absorbed only by thermally excited quasiparticles. The dissipated
power is given by the net number of absorbed photons multiplied by the photon
energy and by a matrix element which describes the absorption process and
T
then integrated over available states and photons. For T < 2c , ℏ𝜔 < kT, ℏ𝜔 < Δ,
the matrix element can be considered constant, and the following approximate
expression for the surface resistance is obtained
( )
𝜇02 𝜔2 𝜆3 𝜎n Δ 2.246kT − kTΔ
RBCS = ln e (2.57)
kT ℏ𝜔
which can be further approximated by
A
RBCS = 𝜔2 e−Δ∕kT (2.58)
T
where A is a constant which depends on the following parameters: London penetra-
tion depth 𝜆L, coherence length 𝜉 0 , Fermi velocity vF , and mean free path 𝓁. This
relation is valid for T < T c /2 [4].
These equations confirm (as before) that the surface resistance decreases expo-
nentially with temperature, depends on the square of the rf frequency 𝜔, and is
directly proportional to the conductivity of the normal-conducting state 𝜎 n . The
surface resistance also depends on the mean free path via 𝜆3 𝜎 n , in agreement with
Eq. (2.38).
To extract the dependence of RBCS with the mean free path, as it gets shorter
with impurities, it is convenient to take into account the Pippard’s definition of 𝜆
 2.3 SRF Physics 29

10–6
Before baking

Surface resistance (Ω)

After 120 °C 48 h bake
10–7
RBCS (Ω)

10–8
10–8
T=2K
Δ/kTc = 1.85 10–9
Δ/kTc = 1.95
Δ/kTc = 2.05
10–9 10–10
1 10 100 1000 0.2 0.3 0.4 0.5 0.6 0.7 0.8
ℓ (nm) 1/Temperature (1/K)
(a) (b)

Figure 2.11 (a) BCS surface resistance as a function of the mean free path for different
values of reduced energy gap Δ/kTc , [49] Courtesy of M. Martinello, Fermilab. (b) Surface
resistance versus 1/T showing how the BCS resistance changes for two different mfp
values. Source: [63]/G. Ciovati, Jefferson Lab with permission of AIP Publishing LLC.

shown in Eq. (2.45). The conductivity 𝜎 n is defined by the Drude formula (Eq. (2.7)).
Therefore:
( )
𝜉0 3∕2
RBCS ∝ 𝜆 𝜎n ∝ 𝓁 1 +
3
(2.59)
𝓁
This relation has a minimum at 𝓁 = 𝜉/2, meaning that the surface resistance is
minimized when the mean free path is around 𝜉/2, as seen in Figure 2.11a.
In Figure 2.11a, curves of RBCS as a function of 𝓁 calculated from SRIMP are
shown for different values of the reduced energy gap Δ/kTc , keeping all the other
parameters fixed [49]. It is clear that with increasing Δ/kTc the resistance decreases.
Figure 2.11b shows the surface resistance as a function of temperature for two
different mean free paths achieved for Nb by 120 ∘ C baking (Section 6.1). Measured
values are compared with calculations from the BCS/Mattis and Bardeen theory
using the SRIMP code [61, 64].
From experiments, the cavity mean free path can be determined with the help of
the SRIMP code by fitting the measured Rs versus T data, or by fitting the resonance
frequency change as a function of temperature during the cavity warm-up to T c .
During warm-up of a cavity to near T c , the penetration depth increases along with
the volume occupied by the rf magnetic field. This leads to a decrease of the reso-
nance frequency, in agreement with the Slater’s theorem [65].
GΔf
Δ𝜆 = (2.60)
𝜇0 𝜋f 2 (T0 )
where Δf = f (T) − f (T 0 ), Δ𝜆 = 𝜆(T) − 𝜆(T 0 ) and T 0 is the temperature at which the
frequency becomes almost constant, usually 7.5 ∘ K for Nb. The frequency variation
can be approximated (according to Eq. [2.23]) as:
[ ( )4 ]
1 1 T
= 1− (2.61)
𝜆2 (T) 𝜆20 Tc
30 2 SRF Fundamentals Review

Experimental data
400 SRIMP fit
Model DeltaLambda_T (user)

Reduced Chi-Sqr 516.04442

300 Adj. R-Square 0.99954
Value Standard error
f0 1.369 0
Δλ (nm)

Tc 8.97 0
200 gap 2.08894 0.02866
DLambda lambda 390 0
xi 600 0
mfp 1477.30875 46.46082
T0 7.41 0
100

7.5 8.0 8.5 9.0

Teq (K)

Figure 2.12 SRIMP interpolation of the penetration depth variation as a function of

the temperature of a cavity to give the mfp and energy gap. Source: [49] Courtesy of
M. Martinello, Fermilab.

For example, the frequency decreases about 5 kHz between liquid helium
temperatures and T c , for a 1.3 GHz Nb cavity. For dirty materials, the variation
of the penetration depth is larger than for cleaner materials, as expected from
Pippard’s definition of the penetration depth:
( )
𝜉 1∕2
𝜆0 = 𝜆L 1 + 0 (2.62)
𝓁
The SRIMP code is used to fit 𝜆 versus T data, with the following fixed parameters:
critical temperature, T c , coherence length (𝜉 0 = 38 nm), and the London penetration
depth (𝜆L = 39 nm). The variable parameters used to fit the data are: mean free path
𝓁 and reduced energy gap (Δ/kTc ). Figure 2.12 shows an example of the SRIMP fit
for frequency variation near T c .
This procedure yields the variation of the penetration depth, as a function of the
temperature close to T c [49], and from there the mean free path.

2.3.6 Residual Resistance

When compared with experimental data, the theoretical behavior of the BCS
surface resistance is generally good up to a certain temperature (see Figure 2.11b).
Below this temperature, rather than continuing to decrease, the surface resistance
approaches a constant value, called the residual surface resistance R0 or Rres The
complete expression of the surface resistance must therefore take into account not
only the temperature-dependent term derived from the Mattis and Bardeen theory,
RBCS , but also a temperature-independent term, R0 :
Rs = RBCS + R0 (2.63)
 2.3 SRF Physics 31

N/N(0)
4

Г/Δ0 = 0

Г/Δ0 = 0.05
3
Г/Δ0 = 0.5

E/Δ0
–3 –2 –1 0 1 2 3

Figure 2.13 Superconducting density of states with various Dynes smearing parameter
values. Source: [12] Courtesy of D. Bafia, Illinois Institute of Technology.

Possible causes for residual resistance may be the presence of defects, magnetic
flux trapped in the superconductor, the presence of hydrides, the oxidized layer on
the surface, residuals from chemical or mechanical surface treatment, and weld-
ing imperfections, and so on [4]. A fundamental contribution may also arise from
quasiparticle states that exist within the gap as discussed below in Section 2.3.7.

2.3.7 Smearing of Density of States

Dynes introduced a phenomenological smearing parameter 𝛤 [66]. This parame-
ter describes a finite population of quasiparticle states that exist within the gap and
modifies the density of states Eq. (2.54) to smear out the singularity. The height of
the smeared coherence peak gives a phenomenological measure of the level of inelas-
tic scattering. Figure 2.13 shows the density of states for different values of 𝛤 . The
Dynes smearing parameter causes the peak to decrease in height and the captured
states in the sub-gap region.
N E + i𝛤
= √ (2.64)
N0 (E + i𝛤 )2 − Δ2

2.3.8 Ginzburg–Landau (GL) Theory

Note: Though formally H is the symbol for magnetic field strength, B (the magnetic
flux density) is often used by a variety of authors, especially in Figures adopted from
their work.
The BCS theory provides a thorough understanding of a large portion of the
physics of superconductors. However, its quantum-mechanical formalism makes
it difficult to use when addressing the macroscopic behavior of superconductors.
Another random document with
no related content on Scribd:
The Project Gutenberg eBook of Dante
 This ebook is for the use of anyone anywhere in the United
States and most other parts of the world at no cost and with
almost no restrictions whatsoever. You may copy it, give it away
or re-use it under the terms of the Project Gutenberg License
included with this ebook or online at www.gutenberg.org. If you
are not located in the United States, you will have to check the
laws of the country where you are located before using this
eBook.

Title: Dante

Author: Edmund G. Gardner

Release date: October 31, 2023 [eBook #71990]

Language: English

Original publication: New York: E. P. Dutton and Company, 1923

Credits: Tim Lindell and the Online Distributed Proofreading

Team at https://www.pgdp.net (This file was produced
from images generously made available by The
Internet Archive/American Libraries.)

*** START OF THE PROJECT GUTENBERG EBOOK DANTE ***

 DANTE
BY
EDMUND G. GARDNER, M.A.

NEW YORK
E. P. DUTTON & COMPANY
681 Fifth Avenue
 Copyright, 1923
By E. P. Dutton & Company

All Rights Reserved

PRINTED IN THE UNITED STATES OF AMERICA

 TO
PHILIP H. WICKSTEED
A SMALL TRIBUTE
OF
DEEP AFFECTION AND HIGH ESTEEM
 AUTHOR’S NOTE
I would ask the reader to take the present volume, not as a new
book on Dante, but merely as a revision of the Primer which was first
published in 1900. It has been as far as possible brought up to date,
the chief modifications being naturally in the sections devoted to the
poet’s life and Opere minori, and in the bibliographical appendix; but
the work remains substantially the same. Were I now to write a new
Dante Primer, after the interval of nearly a quarter of a century, I
should be disposed to attach considerably less importance to the
allegorical meaning of the Divina Commedia, and to emphasise,
more than I have here done, the aspect of Dante as the symbol and
national hero of Italy.
E. G. G.
London, July, 1923.
 N.B.—The “Sexcentenary Dante” (the testo critico published under
the auspices of the Società Dantesca Italiana) adopts a slightly
different numbering of the chapters, or paragraphs, of the Vita Nuova
and the second treatise of the Convivio from that presented by the
“Oxford Dante” and the “Temple Classics.” I have kept to the latter
(which is indicated in brackets in the testo critico). Similarly, I have
followed the numbering of the Epistolae in Dr. Toynbee’s edition and
the “Oxford Dante” (also given in brackets in the testo critico). In the
section on the lyrical poetry, Rime refers to the testo critico as edited
by Professor Barbi, O. to the new Oxford edition revised by Dr.
Toynbee. In the closing passage of the Letter to a Florentine friend, I
have followed the reading retained by Dr. Toynbee. I have frequently
availed myself of Dr. Wicksteed’s translation of the Letters and
Monarchia, of Mr. A. G. F. Howell’s version of the De Vulgari
Eloquentia, and occasionally of Carlyle’s rendering of the Inferno.
Every student of Dante must inevitably owe much to others; but, in
this new edition of my Primer, I would express my indebtedness in
particular to the writings of Dr. Paget Toynbee, Dr. Philip H.
Wicksteed, the late Ernesto Giacomo Parodi, and Prof. Michele
Barbi.
⁂ To the Bibliographical Appendix should be added: A.
Fiammazzo, Il commento dantesco di Graziolo de’ Bambaglioli
(Savona, 1915), and P. Revelli, L’Italia nella Divina Commedia
(Milan, 1923).
 CONTENTS
CHAPTER I
Dante in his Times—
i. The End of the Middle Ages.—ii. Dante’s 1
Childhood and Adolescence.—iii. After the
Death of Beatrice.—iv. Dante’s Political Life.
—v. First Period of Exile.—vi. The Invasion of
Henry VII.—vii. Last Period of Exile.—viii.
Dante’s Works and First Interpreters
CHAPTER II
Dante’s Minor Italian Works—
i. The Vita Nuova.—ii. The Rime.—iii. The 67
Convivio
CHAPTER III
Dante’s Latin Works—
i. The De Vulgari Eloquentia.—ii. The 102
Monarchia.—iii. The Epistolae.—iv. The
Eclogae.—v. The Quaestio de Aqua et Terra
CHAPTER IV
The “Divina Commedia”—
i. Introductory.—ii. The Inferno.—iii. The 136
Purgatorio.—iv. The Paradiso
Bibliographical Appendix 223
Diagrams and Tables 233
Index 249
DANTE
 CHAPTER I
DANTE IN HIS TIMES
1. The End of the Middle Ages
From Gregory VII. to Frederick II.—The twelfth and thirteenth
centuries cover the last and more familiar portion of the Middle Ages.
They are the period of chivalry, of the crusades and of romance,
when the Neo-Latin languages bore fruit in the prose and poetry of
France, the lyrics of the Provençal troubadours, and the earliest
vernacular literature of Italy; the period which saw the development
of Gothic architecture, the rise of scholastic philosophy, the
institution of the Franciscan and Dominican orders, the recovery by
western Europe of the works of Aristotle, the elevation of Catholic
theology into a systematic harmony of reason and revelation under
the influence of the christianised Aristotelianism of Albertus Magnus
and Thomas Aquinas. The vernacular literature of Italy developed
comparatively late, and (until the time of Aquinas and Bonaventura)
her part in the scholastic movement was secondary to that of
France, but she had led the way in the revival of the study of Roman
law and jurisprudence, which centred at Bologna, where the great
Irnerius taught at the beginning of the twelfth century. It was thus that
the first European university, studium generale, came into being, and
Bologna boasts the proud title alma mater studiorum.
There are two predominant political factors in Italy which appear at
the end of the twelfth century, and hold the field up to the time of
Dante’s birth. Out of the war of investitures between Pope and
Emperor, the struggle which we associate mainly with the name of
Hildebrand (Gregory VII.), emerged the Italian city-states, the free
communes of northern and central Italy, whose development
culminated in the heroic resistance offered by the first Lombard
League to the mightiest of mediaeval German Caesars, Frederick I.
(Barbarossa), which won the battle of Legnano (1176) and obtained
the peace of Constance (1183). In the south, the Normans—
conquering Apulia and Calabria, delivering Sicily from the Saracens
—consolidated their rule into a feudal monarchy, making their capital
Palermo one of the most splendid cities of the mediaeval world. The
third and last of these Norman kings of Sicily, William II. (Par. xx. 61-
66), died in 1189. The son of Barbarossa, Henry VI., claimed the
kingdom in the right of his wife Constance (Par. iii. 115-120), and
established the Suabian dynasty on the throne. His son, Frederick
II., continued the cultured traditions of the Norman kings; but the
union in his person of the kingdom of Sicily with the Empire led to a
continuous struggle with the Italian communes and the Papacy,
which embittered his closing years until his death in 1250. The reign
of Frederick II. is the period of the Guelf and Ghibelline factions, and
the beginning of the rise of tyrants in the Italian cities, tyrants of
whom the most terrible example was Ezzelino da Romano (Inf. xii.
110).
The Battle of Benevento.—The policy of Frederick II. was
continued by his son Manfred (crowned King of Sicily in 1258),
against whom Pope Clement IV., claiming the right to dispose of the
kingdom as a fief of the Church, summoned Charles of Anjou, the
brother of St. Louis of France. Charles entered Italy (Purg. xx. 67),
encountered and defeated Manfred on the plains of Grandella near
Benevento, in February 1266, and the papal legate refused the rites
of Christian burial to the fallen king (Purg. iii. 124-132). This battle of
Benevento marks an epoch in Italian history. It ended for the time the
struggle between the Roman Pontiffs and the German Caesars; it
initiated the new strife between the Papacy and the royal house of
France. Henceforth the old ideal significance of “Guelf” and
“Ghibelline,” as denoting adherents of Church and Empire
respectively, becomes lost in the local conflicts of each Italian district
and city. The imperial power was at an end in Italy; but the Popes, by
calling in this new foreign aid, had prepared the way for the
humiliation of Pope Boniface at Anagni and the corruption of
Avignon. The fall of the silver eagle from Manfred’s helmet before the
golden lilies on Charles’s standard may be taken as symbolical. The
preponderance in Italian politics had passed back from Germany to
France; the influence of the house of Capet was substituted for the
overthrown authority of the Emperor (Purg. xx. 43, 44). Three weeks
after the battle Charles entered Naples in triumph, King of Apulia and
Sicily; an Angevin dynasty was established upon the throne of the
most potent state of Italy.
Art and Letters.—This political transformation was profoundly
felt in Italian literature. A new courtly poetry, that of the so-called
“Sicilian School,” had come into being in the south, partly based on
Provençal models, in the first quarter of the thirteenth century. Its
poets—mainly Sicilians and Apulians, but with recruits from other
parts of the peninsula—had almost given to Italy a literary language.
“The Sicilian vernacular,” writes Dante in his De Vulgari Eloquentia,
“seems to have gained for itself a renown beyond the others; for
whatever Italians produce in poetry is called Sicilian, and we find that
many native poets have sung weightily.” This he ascribes to the
fostering influence of the imperial rule of the house of Suabia:
“Those illustrious heroes, Frederick Caesar and his well-begotten
son Manfred, showing their nobility and rectitude of soul, as long as
fortune lasted, followed human things, disdaining the bestial;
wherefore the noble in heart and endowed with graces strove to
cleave to the majesty of such great princes; so that, in their time,
whatever the excellent among Italians attempted first appeared at
the court of these great sovereigns. And, because the royal throne
was Sicily, it came about that whatever our predecessors produced
in the vernacular is called Sicilian” (V. E. i. 12). The house of Anjou
made Naples their capital, and treated Sicily as a conquered
province. After Benevento the literary centre of Italy shifted from
Palermo and the royal court of the south to Bologna and the
republican cities of Tuscany. Guittone d’Arezzo (Purg. xxvi. 124-126)
founded a school of Tuscan poets, extending the field of Italian
lyrical poetry to political and ethical themes as well as love (which
had been the sole subject of the Sicilian School). The beginnings of
Italian literary prose had already appeared at Bologna, with the first
vernacular models for composition of the rhetoricians, the masters of
the ars dictandi. Here, within the next eight years, St. Thomas
Aquinas published the first and second parts of the Summa
Theologica; and the poetry of the first great singer of modern Italy,
Guido Guinizelli (Purg. xxvi. 91-114), rose to spiritual heights
undreamed of in the older schools, in his canzone on Love and true
nobility: Al cor gentil ripara sempre Amore; “To the gentle heart doth
Love ever repair.” And, in the sphere of the plastic arts, these were
the years that saw the last triumphs of Niccolò Pisano, “the Father of
Sculpture to Italy,” and the earliest masterpieces of Cimabue, the
teacher of Giotto (Purg. xi. 94-96), the shepherd boy who came from
the fields to free Italian painting from Byzantine fetters, and who
“developed an artistic language which was the true expression of the
Italian national character.”

2. Dante’s Childhood and Adolescence

Birth and Family.—Dante Alighieri, in its Latin form Alagherii, was
born at Florence in 1265, probably in the latter part of May, some
nine months before the battle of Benevento. His father, Alighiero di
Bellincione di Alighiero, came of an ancient and honourable family of
that section of the city named from the Porta San Piero. Although
Guelfs, the Alighieri were probably of the same stock as the Elisei,
decadent nobles of supposed Roman descent, who took the
Ghibelline side in the days of Frederick II., when the city was first
involved in these factions after the murder of young Buondelmonte in
her “last peace” in 1215 (Par. xvi. 136-147). Among the warriors of
the Cross, in the Heaven of Mars, Dante meets his great-great-
grandfather Cacciaguida. Born probably in 1091, Cacciaguida
married a wife from the valley of the Po, a member of one or other of
the families afterwards known as the Aldighieri or Alighieri at Ferrara,
Parma, and Bologna, was knighted by Conrad III., and died in battle
against the infidels in the disastrous second crusade (Par. xv. 137-
148). None of Cacciaguida’s descendants had attained to any
distinction in the Republic. Brunetto di Bellincione, Dante’s uncle,
probably fought for the Guelfs at Montaperti in 1260, where he may
have been one of those in charge of the carroccio, the battle-car
which accompanied the army. Besides Cacciaguida and his son
Alaghiero, or Alighiero, the first to bear the name, who is said by his
father to be still in the purgatorial terrace of the proud (Par. xv. 91-
96), the only other member of the family introduced into the Divina
Commedia is Geri del Bello, a grandson of the elder Alighiero and
cousin of Dante’s father, a sower of discord and a murderer (Inf. xxix.
13-36), whose violent and well-deserved death had not yet been
avenged.
The Florentine Republic.—As far as Florence was concerned,
the real strife of Guelfs and Ghibellines was a struggle for
supremacy, first without and then within the city, of a democracy of
merchants and traders, with a military aristocracy of partly Teutonic
descent, who were gradually being deprived of their territorial and
feudal sway, which they had held nominally from the Emperor in the
contado, the country districts of Tuscany included in the continually
extending Florentine commune. Although the party names were first
introduced into Florence in 1215, the struggle had virtually begun
after the death of the great Countess Matilda in 1115; and had
resulted in a regular and constitutional advance of the power of the
people, interrupted by a few intervals. It was in one of these intervals
that Dante was born. The popular government (Primo Popolo), which
had been established shortly before the death of Frederick II. in
1250, and worked victoriously for ten years, had been overthrown in
1260 at the disastrous battle of Montaperti, “the havoc and the great
slaughter, which dyed the Arbia red” (Inf. x. 85, 86). The patriotism of
Farinata degli Uberti saved Florence from total destruction, but all
the leading Guelf families were driven out, and the government
remained in the power of a despotic Ghibelline aristocracy, under
Manfred’s vicar, Count Guido Novello, supported by German
mercenaries. After the fall of Manfred, an attempt was made to effect
a peace between the Ghibellines and the people; but a revolution on
St. Martin’s Day, November 11th, 1266, led to the expulsion of Guido
Novello and his forces, and the formation of a provisional democratic
government. In January 1267 the banished Guelfs—many of whom
had fought under the papal banner at Benevento—returned; on
Easter Day French troops entered Florence, the Ghibellines fled, the
Guelfs made Charles of Anjou suzerain of the city, and accepted his
vicar as podestà. The government was reorganised, with a new
institution, the Parte Guelfa, to secure the Guelf predominance in the
Republic.
 The defeat of young Conradin, grandson of Frederick II., at
Tagliacozzo in 1268, followed by his judicial murder (Purg. xx. 68),
confirmed the triumph of the Guelfs and the power of Charles in Italy.
In Florence the future conflict lay between the new Guelf aristocracy
and the burghers and people, between the Grandi and the Popolani;
the magnates in their palaces and towers, associated into societies
and groups of families, surrounding themselves with retainers and
swordsmen, but always divided among themselves; and the people,
soon to become “very fierce and hot in lordship,” as Villani says,
artisans and traders ready to rush out from stalls and workshops to
follow the standards of their Arts or Guilds in defence of liberty. In the
year after the House of Suabia ended with Conradin upon the
scaffold, the Florentines took partial vengeance for Montaperti at the
battle of Colle di Valdelsa (Purg. xiii. 115-120), where the Sienese
were routed and Provenzano Salvani (Purg. xi. 109-114) killed. It is
said to have been Provenzano Salvani who, in the great Ghibelline
council at Empoli, had proposed that Florence should be destroyed.
Dante’s Boyhood.—It is not clear how Dante came to be born in
Florence, since he gives us to understand (Inf. x. 46-50) that his
family were fiercely adverse to the Ghibellines and would naturally
have been in exile until the close of 1266. Probably his father, of
whom scarcely anything is known, took no prominent part in politics
and had been allowed to remain in the city. Besides the houses in
the Piazza San Martino, he possessed two farms and some land in
the country. Dante’s mother, Bella (perhaps an abbreviation of
Gabriella), is believed to have been Alighiero’s first wife, and to have
died soon after the poet’s birth. Her family is not known, though it
has been suggested that she may have been the daughter of
Durante di Scolaio degli Abati, a Guelf noble. Alighiero married
again, Lapa di Chiarissimo Cialuffi, the daughter of a prominent
Guelf popolano; by this second marriage he had a son, Francesco,
and a daughter, Tana (Gaetana), who married Lapo Riccomanni.
Another daughter, whose name is not known, married Leone Poggi;
it is not quite certain whether her mother was Bella or Lapa. Dante
never mentions his mother nor his father, whom he also lost in
boyhood, in any of his works (excepting such indirect references as
Inf. viii. 45, and Conv. i. 13); but, in the Vita Nuova, a “young and
gentle lady, who was united to me by very near kindred,” appears
watching by the poet in his illness. In the loveliest of his early lyrics
she is described as

Adorna assai di gentilezze umane,

which Rossetti renders:

Exceeding rich in human sympathies.

This lady was, perhaps, one of these two sisters; and it is tempting to
infer from Dante’s words that a tender affection existed between him
and her. It was from Dante’s nephew, Andrea Poggi, that Boccaccio
obtained some of his information concerning the poet, and it would
be pleasant to think that Andrea’s mother is the heroine of this
canzone (V. N. xxiii.); but there are chronological difficulties in the
identification.
Sources.—Our sources for Dante’s biography, in addition to his
own works, are primarily a short chapter in the Chronicle of his
neighbour Giovanni Villani, the epoch-making work of Boccaccio,
Filippo Villani’s unimportant sketch at the end of the fourteenth and
the brief but reliable life by Leonardo Bruni at the beginning of the
fifteenth century. In addition we have some scanty hints given by the
early commentators on the Divina Commedia, and a few documents,
including the consulte or reports of the deliberations of the various
councils of the Florentine Republic. Boccaccio’s work has come
down to us in two forms: the Vita di Dante (or Trattatello in laude di
Dante) and the so-called Compendio (itself in two redactions, the
Primo and Secondo Compendio); the researches of Michele Barbi
have finally established that both are authentic, the Compendio
being the author’s own later revision. The tendency of recent
scholarship has in a considerable measure rehabilitated the once
discredited authority of Boccaccio, and rejected the excessive
scepticism represented in the nineteenth century by Bartoli and
Scartazzini.
Beatrice.—Although Leonardo Bruni rebukes Boccaccio, “our
Boccaccio that most sweet and pleasant man,” for having lingered so
long over Dante’s love affairs, still the story of the poet’s first love
remains the one salient fact of his youth and early manhood. We
may surmise from the Vita Nuova that at the end of his eighteenth
year, presumably in May 1283, Dante became enamoured of the
glorious lady of his mind, Beatrice, who had first appeared to him as
a child in her ninth year, nine years before. It is not quite certain
whether Beatrice was her real name or one beneath which Dante
conceals her identity; assuredly she was “Beatrice,” the giver of
blessing, to him and through him to all lovers of the noblest and
fairest things in literature. Tradition, following Boccaccio, has
identified her with Bice, the daughter of Folco Portinari, a wealthy
Florentine who founded the hospital of S. Maria Nuova, and died in
1289 (cf. V. N. xxii.). Folco’s daughter is shown by her father’s will to
have been the wife of Simone dei Bardi, a rich and noble banker.
This has been confirmed by the discovery that, while the printed
commentary of Dante’s son Pietro upon the Commedia hardly
suggests that Beatrice was a real woman at all, there exists a fuller
and later recension by Pietro of his own work which contains a
distinct statement that the lady raised to fame in his father’s poem
was in very fact Bice Portinari. Nevertheless, there are still found
critics who see in Beatrice not a real woman, but a mystically exalted
ideal of womanhood or a merely allegorical figure; while Scartazzini
at one time maintained that the woman Dante loved was an
unknown Florentine maiden, who would have been his wife but for
her untimely death. This can hardly be deduced from the Vita Nuova;
in its noblest passages the woman of Dante’s worship is scarcely
regarded as an object that can be possessed; death has not robbed
him of an expected beatitude, but all the world of an earthly miracle.
But, although it was in the fullest correspondence with mediaeval
ideals and fashions that chivalrous love and devotion should be
directed by preference to a married woman, the love of Dante for
Beatrice was something at once more real and more exalted than
the artificial passion of the troubadours; a true romantic love that
linked heaven to earth, and was a revelation for the whole course of
life.
Poetry, Friendship, Study.—Already, at the age of eighteen,
Dante was a poet: “I had already seen for myself the art of saying
words in rhyme” (V. N. iii.). It was on the occasion of what we take as
the real beginning of his love that he wrote the opening sonnet of the
Vita Nuova, in which he demands an explanation of a dream from
“all the faithful of Love.” The new poet was at once recognised.
Among the many answers came a sonnet from the most famous
Italian lyrist then living, Guido Cavalcanti, henceforth to be the first of
Dante’s friends: “And this was, as it were, the beginning of the
friendship between him and me, when he knew that I was he who
had sent that sonnet to him” (cf. Inf. x. 60). In the same year, 1283,
Dante’s name first occurs in a document concerning some business
transactions as his late father’s heir.
There are no external events recorded in Dante’s life between
1283 and 1289. Boccaccio represents him as devoted to study. He
certainly owed much to the paternal advice of the old rhetorician and
statesman, Brunetto Latini, who had been secretary of the commune
and, until his death in 1294, was one of the most influential citizens
in the state: “For in my memory is fixed, and now goes to my heart,
the dear, kind, paternal image of you, when in the world, from time to
time, you taught me how man makes himself eternal” (Inf. xv. 82). Of
his growing maturity in art, the lyrics of the Vita Nuova bear witness;
the prose narrative shows that he had studied the Latin poets as well
as the new singers of Provence and Italy, had already dipped into
scholastic philosophy, and was not unacquainted with Aristotle. At
the same time, Leonardo Bruni was obviously right in describing
Dante as not severing himself from the world, but excelling in every
youthful exercise; and it would seem from the Vita Nuova that, in
spite of his supreme devotion for Beatrice, there were other
Florentine damsels who moved his heart for a time. Dante speaks of
“one who, according to the degrees of friendship, is my friend
immediately after the first,” and than whom there was no one nearer
in kinship to Beatrice (V. N. xxxiii.). Those who identify Dante’s
Beatrice with the daughter of Messer Folco suppose that this second