Superconducting Radiofrequency Technology For Accelerators: State of The Art and Emerging Trends 1st Edition Hasan Padamsee
Superconducting Radiofrequency Technology For Accelerators: State of The Art and Emerging Trends 1st Edition Hasan Padamsee
Superconducting Radiofrequency Technology For Accelerators: State of The Art and Emerging Trends 1st Edition Hasan Padamsee
https://ebookmeta.com/product/emerging-perspectives-in-art-
therapy-trends-movements-and-developments-1st-edition-richard-
carolan-editor/
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/
https://ebookmeta.com/product/shock-of-the-new-the-challenge-and-
promise-of-emerging-technology-1st-edition-udell/
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/
https://ebookmeta.com/product/art-schools-and-place-geographies-
of-emerging-artists-and-art-scenes-1st-edition-silvie-jacobi/
https://ebookmeta.com/product/opportunistic-networks-
fundamentals-applications-and-emerging-trends-1st-edition-anshul-
verma/
https://ebookmeta.com/product/techniques-for-food-authentication-
trends-and-emerging-approaches-2nd-ed-raul-gonzalez-dominguez/
https://ebookmeta.com/product/safety-for-particle-
accelerators-1st-edition-thomas-otto/
Superconducting Radiofrequency Technology for Accelerators
Superconducting Radiofrequency Technology
for Accelerators
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.
Contents
Preface xi
1 Introduction 3
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
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.
References
Part I
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
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
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
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)
GE varnish
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)
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.
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].
2.2.2 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
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
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.
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,
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
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
10
106λ0 (cm)
0 20 40 60 80
(b) 106l (cm)
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
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
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.
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
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)
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
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.
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.
Title: Dante
Language: English
NEW YORK
E. P. DUTTON & COMPANY
681 Fifth Avenue
Copyright, 1923
By E. P. Dutton & Company
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