Radio frequency superconductivity for particle accelerators · 2018. 4. 12. · Radio frequency...

Sergey Belomestnykh (Fermilab)

Seminar at John Adams Institute for Accelerator Science

Oxford, UK, February 12, 2018

Radio frequency superconductivity for

particle accelerators:

Recent trends in physics and technology

2/12/2018 S. Belomestnykh | SRF: Recent trends in science and technology | Seminar at JAI 2

Outline

Introduction

What is RF superconductivity for particle accelerators?

SRF basics: surface resistance, Q vs. E

Recent SRF science breakthroughs & active areas of research:

o Nitrogen doping

o Nitrogen infusion

o Frequency dependence of Rs

o SRF in quantum regime

SRF over the world

Summary


Introduction

Over the past several years, the field of radio frequency superconductivity (SRF)

for particle accelerators is going through a period of Renaissance.

5 years ago, most of the community thought that the science and technology

reached maturity (even though we lacked understanding of some basic physics)

and one can achieve only incremental gains in the niobium cavity performance.

The field tended to be mostly technological with only few researchers trying to

study fundamental issues of SRF in niobium. Big improvement steps were thought

to be possible only with developing alternative materials (e.g. Nb3Sn).

Recent discoveries of nitrogen doping and infusion, magnetic flux expulsion,

opened new horizons and revived interest to studies of SRF basics, both

experimental and theoretical. More unexpected and intriguing results have been

obtained.

In this talk I will try to shed light upon some exciting recent results, show new

trends (Fermilab-centric view) and hopefully inspire young generation to turn their

attention to this field of research.


What is RF superconductivity

for particle accelerators?


Discovery of superconductivity: April 8th of 1911 Discovered in 1911 by Heike Kamerlingh Onnes and Gilles Holst after Onnes was

able to liquefy helium in 1908 (Nobel Prize in 1913).


Superconducting elements


Superconducting state The superconducting state is characterized by the critical temperature Tc and field Hc

The external field is expelled from a superconductor if Hext < Hc for Type I superconductors.

For Type II superconductors the external field can partially penetrate for Hext < Hc1 and will

completely penetrate at Hc2.

2

10c

ccT

THTH


Theories explaining superconductivity

Early developments: two-fluid model and London equations.

Phenomenological Ginzburg-Landau (GL) theory (1950, Nobel Prize in 2003) generalized

London equation to nonlinear problems.

Microscopic theory of superconductivity was developed by Bardeen, Cooper and Schrieffer

(BCS) in 1957 (Nobel Prize in 1972).

What do we need to recollect?

Magnetic field does not stop abruptly, but penetrates into the material with exponential

attenuation. The (London) penetration depth l is quite small, 20 – 50 nm.

According to BCS theory not single electrons, but (Cooper) pairs are carriers of the

supercurrent. However, the penetration depth remains unchanged.

The BCS ground state is characterized by the macroscopic wave function and a ground

state energy that is separated from the energy levels of unpaired electrons by an energy

gap. In order to break a pair an energy of 2D is needed:


Theories explaining superconductivity (2)

GL theory introduced coherence length x – a new scale of special variation of the superfluid

density and superconducting gap.

Also introduced is a dimensionless GL parameter k l / x, which is independent of

temperature.


What happens if AC field is applied?

At 0 < T < Tc not all electrons are bonded into Cooper pairs. The density of unpaired,

“normal” electrons is given by the Boltzman factor

Cooper pairs move without resistance, and thus dissipate no power. In DC case the lossless

Cooper pairs short out the field, hence the normal electrons are not accelerated and the SC

is lossless even for T > 0 K.

The Cooper pairs do nonetheless have an inertial mass, and thus they cannot follow an AC

electromagnetic fields instantly and do not shield it perfectly. A residual EM field remains

and acts on the unpaired electrons as well, therefore causing power dissipation.

D

Tkn

B

expnormal


What is RF superconductivity for accelerators? Radio frequency (RF) superconductivity for particle accelerators is a branch of accelerator

physics and engineering dealing with application of superconducting materials to

acceleration of charged particles in resonant RF cavities.

The science part of this field deals with investigating limitations of and developing methods

to improve the SRF cavity performance. In particular, how to reduce power dissipation in

SRF cavities and improve accelerating gradients.

Slowed down by factor of approximately 4x109 Input RF power at 1.3 GHz

~1 m

Niobium


RF superconductivity as a branch of accelerator

physics was born 57 years ago

“In a seminal paper published in June 1961 A. P. Banford and G. H. Stafford described how a future superconducting proton linear accelerator could run continuously, instead of at the 1% duty cycle of the 50 MeV proton accelerator that was operating at the time at the Rutherford High Energy Laboratory in the UK. The basic argument was that, because ohmic losses in the accelerating cavity walls increase as the square of the accelerating voltage, copper cavities become uneconomical when the demand for high continuous-wave (CW) voltage grows with particle energy. It is here that superconductivity comes to the rescue.”

(from Hasan Padamsee’s article “Advances in acceleration: the superconducting way,”

CERN Courier, November 2011)

Plasma Physics (Journal of Nuclear Energy Part C), 1961, Vol. 3, pp. 287 to 290.


Benefits of RF superconductivity The development of superconducting (SC) cavities for accelerators has enabled new

applications not previously possible with normal conducting (NC) structures.

SC cavities excel in applications requiring continuous wave (CW) or long-pulse accelerating

fields above a few MV/m (up to ~35 MV/m).

For NC cavities (usually made of copper) power dissipation in cavity walls is a huge constrain

in these cases cavity design is driven by this fact, optimized for lowest possible wall

dissipation small beam aperture.

The surface resistivity of SC cavities is 5-6 orders of

magnitude less than that of copper SC accelerating

system is more economical: less wall plug power, fewer

cavities required, …

Additional benefit: the cavity design decouples from the

dynamic losses (wall losses associated with RF fields)

free to adapt design to a specific application.

The presence of accelerating structures has a disruptive

effect on the beam and may cause various instabilities,

dilute beam emittance and produce other undesirable

effects. Fewer SC cavities less disruption. SC cavities

can trade off some of wall losses to a larger beam pipe

reduce disruption more.


SRF basics:

Rs, Q vs. Eacc


Surface resistance A convenient way to characterize power losses at radio frequency resonant cavities is to use

a so-called surface resistance. [ For normal conducting cavities Rs = 1/(sd), where s is the

specific conductivity and d is the skin depth. ] Then the power dissipation per unit area is

And the total power dissipation is obtained by integration

over the whole inner surface of the cavity.

Calculation of surface resistance must take into account numerous parameters. Mattis and

Bardeen developed theory based on BCS, which predicts

where A is the material constant

depends on the electron mean free path

20

2

1HRP sdiss

,2

T

T

TkBCS

c

cBeT

AR

D


Surface resistance (2)

While for low frequencies (≤ 500 MHz) it may be efficient to operate at 4.2 K (liquid

helium at atmospheric pressure), higher frequency structures favor lower

operating temperatures (typically superfluid LHe at 2 K, below the lambda point,

2.172 K).

Approximate expression for Nb:

]Ohm[1

1500

]MHz[102

67.172

4

T

BCS eT

fR


Surface resistance of cavities

The BCS surface resistance is described

by Mattis-Bardeen theory and comes from

thermally excited quasi-particles

The residual resistance can come from

different extrinsic contributions :

o Impurities/defects in the surface

o Hydrides precipitates

o Trapped magnetic flux

o ...

Residual resistance is significant for cavities

operating at 2 K.

𝑅𝑠 𝑇 = 𝑅𝐵𝐶𝑆 𝑇 + 𝑅𝑟𝑒𝑠

𝑅𝐵𝐶𝑆 𝑇 =𝐴𝜔2

𝑇𝑒−

∆𝑘𝐵𝑇

𝑅𝑟𝑒𝑠


Why Niobium?

Pure niobium has the highest critical temperature Tc among single elements, and Hc1 and Hc are both high.

Low Rs is needed for operation in superfluid helium at 2 to 4 K (typical accelerator operation domain).

High theoretical Meissner state breakup field (Hsh~240 mT) for an ideal surface, which scales with Hc.

Good formability is desirable for ease of cavity fabrication.

Pure intermetallic compounds, like Nb3Sn with a critical temperature of 18.1 K, look attractive for possible

4.2 K operation at first sight as they are “clean” superconductors. However, so far the gradients achieved in

Nb3Sn coated niobium cavities have been limited to below 19 MV/m, probably due to grain boundary effects

in the Nb3Sn layer. Residual resistance is also high and magnetic flux management is an issue.

Alloys are “dirty” superconductors due to their small mean free path and consequently have large BCS

surface resistivity and poor thermal conductivity.

High temperature superconductors have been tried in the past and showed very high surface resistances,

problems arise from very low coherence length = sensitivity to defects, gap anisotropy etc.

Type Tc Hc1 Hc Hc2 Fabrication

- K Oe Oe Oe -

Nb II 9.25 1700 2060 4000 bulk, film

Pb I 7.20 - 803 - electroplating

Nb3Sn* II 18.1 380 5200 240000 film

MgB2 II 39.0 300 4290 film

Hg I 4.15 - 411/339 - -

Ta I 4.47 - 829 - -

In I 3.41 - 281.5 - -

*) Other compounds with the same b-tungsten or A15 structure are under investigation as well.


Q(E) curve It is conventional to evaluate an SRF cavity performance using a Q(E) curve

L is the cavity length, Rs is the average surface resistance

R/Q is the cavity impedance (determined only by cavity shape)

G is the cavity geometry constant

Intr

insi

c q

ual

ity

fact

or

Q =

G/R

s

Increase max Eacc decrease accelerator length

Increase Q decrease required power

Accelerating gradient Eacc = Energy gain/cavity length

𝑃𝑑𝑖𝑠𝑠 =𝐸𝑎𝑐𝑐𝐿

2𝑅𝑠

𝐺∙𝑅 𝑄 =

𝐸𝑎𝑐𝑐𝐿2

𝑄∙𝑅 𝑄

Typical ILC-prepared cavity at T = 2 K

“Ideal” performance?


Q(E) evolution driven by science

0 5 10 15 20 25 30 35 4010

9

1010

1011

Q0

Eacc

(MV/m)

Elliptical Shape

1.3 GHz, 2 K



1.3 GHz, 2 K

0 5 10 15 20 25 30 35 4010

9

1010

1011

Q0

Eacc

(MV/m)

Bulk RRR > 300



1.3 GHz, 2 K

0 5 10 15 20 25 30 35 4010

9

1010

1011

Q0

Eacc

(MV/m)

High Pressure Water Rinse



1.3 GHz, 2 K

0 5 10 15 20 25 30 35 4010

9

1010

1011

Q0

Eacc

(MV/m)

120◦C bake


Q(E) evolution driven by science 1.3 GHz, 2 K

0 5 10 15 20 25 30 35 4010

9

1010

1011

Q0

Eacc

(MV/m)

EXFEL

Typical ILC-recipe prepared cavity


Only thin surface layer matters Inner surface nanostructure within ~100 nm completely determines RF losses in the cavity

RF fields

Helium cooling

RF currents <100 nm

Niobium ~3 mm

RF fields <0.1% of thickness

Final treatment is crucial to performance

Nb2O5

Image from linearcollider.org


State of the art Q(E) curve ~5 years ago

Increase max Eacc decrease accelerator length

Increase Q decrease required power

Typical ILC-prepared cavity at T = 2 K State of the art until ~5 years ago

LFQS MFQS

HFQS


Recent SRF science breakthroughs

& active areas of research


Nitrogen doping


Nitrogen doping: a breakthrough in Q

Discovered while trying to investigate niobium nitride thin films on cavities.

Q-factor improvement after N-doping – up to 4 times higher Q than standard Nb cavities.

Typical Q vs Eacc curve obtained with

120 C bake (standard ILC treatment);

Avg Q with doping is 2-4 times state

of the art;

Example, for 1.3 GHz, 2 K, mid-field

Q ~ 1.5e10 versus > 3e10;

Systematically above Q obtained with

any other surface treatment.

Injection of small

nitrogen partial

pressure at the

end of 800 C

degassing

drastic increase

in Q.

0 5 10 15 20 25 30 35 4010

9

1010

1011

Q0

Eacc

(MV/m)

T= 2K

Anti-Q-slope

A. Grassellino et al., Supercond. Sci. Technol. 26, 102001 (2013) – Rapid Communications

MFQS


Doping Treatment: small variation from standard

protocol, large difference in performance

FNAL doping for LCLS-II (major steps):

o Bulk EP

o 800 C anneal for 3 hours in vacuum

o 2 minutes @ 800 C nitrogen diffusion

o 800 C for 6 minutes in vacuum

o Vacuum cooling

o 5 microns EP

Cavity after Equator Welding

EP 140 um

Ethanol Rinse

External 20 um BCP

Short HPR

800C HT Bake

RF Tuning

EP 40 um

Ethanol Rinse

Long HPR

Final Assembly

Long HPR

Helium Tank Welding

Procedure

VT Assembly

HPR

HOM Tuning

Ship to DESY

Leak Check

120C bake

XF

EL

X


Surface post nitrogen bake, pre-EP: poorly SC

nitride phases

Few Nb-nitride features

(Nb2N reflections) in Nb

near-surface.

Nitride “teeth” go ~0.2 μm

deep.

Flat Nb sample baked at 800C˚ for 2 min with N2 + 6 min annealing

Flat Nb sample baked at 800C˚ for 20 min with N2 + 30 min annealing

Bad (poorly SC) nitride phases that need to be removed

via EP correlate with poor performance (pre-EP) Q~1e7

Nb [113]+Nb2N [210]+? Pt layer

Y. Trenikhina, MOPB055, SRF15


Origin: “reversed” field dependence of RBCS

A. Grassellino et al, 2013 Supercond. Sci. Technol. 26 102001 (Rapid Communication) A. Romanenko and A. Grassellino, Appl. Phys. Lett. 102, 252603 (2013)

𝑅𝑠 𝑇 = 𝑅𝐵𝐶𝑆 𝑇 + 𝑅𝑟𝑒𝑠

Reverse field dependence of the BCS surface resistance component lowest RBCS .

Lower than typical residual resistances (seems to zero all contributions but trapped flux).


Physics – perceived BCS limit has been overcome

Anti-Q-slope emerges from the BCS

surface resistance decreasing with

field.

This was thought to be the lowest

possible BCS resistance.

N doping brings also lower than

typical residual resistance

< 2 nanoOhms (non trapped flux

related).

A. Grassellino et al, 2013 Supercond. Sci. Technol. 26 102001 (Rapid Communication) A. Romanenko and A. Grassellino, Appl. Phys. Lett. 102, 252603 (2013)


Nitrogen doping – from research to production Shortly after its discovery, the nitrogen doping was adopted by LCLS-II project.

After a short R&D period (Fermilab, JLab and Cornell), the recipe was successfully

transferred to industry.

Plot shows performance of SRF cavities from

two prototype LCLS-II cryomodules (Fermilab

and JLab): avg. Q = 3.6e10, avg. Eacc =

22.2 MV/m highest average Q ever

demonstrated in vertical tests of 1.3 GHz

nine-cell cavities at 2 K, 16 MV/m. Cavities

from vendors demonstrate similar

performance.

Higher Q would allow SLAC to use only one

cryoplant (of purchased two) to run the

machine and use the second cryoplant to

support the energy upgrade of LCLS-II

from 4.2 GeV to 8 GeV.

Two drawbacks of N-doping:

1. Achievable accelerating gradient is lower than that

of 120 C baked cavities (35-40 MV/m)

2. Nitrogen-doped cavities are more sensitive to

trapped flux losses.


Next step in the cavity performance 1.3 GHz, 2 K

0 5 10 15 20 25 30 35 4010

9

1010

1011

Q0

Eacc

(MV/m)

LCLS-II, PIP-II, PIP-III

Nitrogen doping


Do we understand how to choose best cavity treatment?

N-doping modify the mean free path → close to theoretical minimum of RBCS

N-doping seems to increase the reduced energy gap D/kBTc

Adding together all the 𝑅𝑆 contributions, it is possible to predict which treatments lowers Rs, taking into account also trapped flux

Best compromise is given by light N-doping treatments

𝑅𝑆 2 𝐾 = 𝑅𝐵𝐶𝑆 2 𝐾 + 𝑅𝑓𝑙 + 𝑅0

Residual resistance:

4 nW: 120 C baked cavities

2 nW: EP and optimally N-doped cavities

𝑅𝐹𝑙 = 𝐵𝑒𝑥𝑡 ∙ 𝜂 ∙ 𝑆 𝐵𝑒𝑥𝑡: external magnetic field 𝜂: flux trapping efficiency

M. Martinello et al, Appl. Phys. Lett. 109, 062601 (2016)


Nitrogen infusion


Nitrogen infusion: higher Q at higher gradients Composition and mean free path in first nanometers

of cavity surface have been shown to be crucial for

both Q and gradient performance.

N-doping at T > 800 C proven to manipulate mean

free path, but constantly throughout several

microns, giving high Q.

120 C bake known to manipulate mean free path at

very near surface on clean bulk, and produce the

highest gradients.

Therefore, it was decided to study how to better

“engineer” a dirty layer on top a clean bulk Nb,

using low T nitrogen treatments aim to create a

few to several nanometers of nitrogen-enriched

layer on top of clean EP bulk, to attempt to bring

together the benefit of the Q and gradient

Nitrogen enriched nanometric layer to be created in

the furnace post 800 C treatment – when no oxide

is present at the moment of injection of nitrogen at

low T.

Studies aim also at fundamental understanding of

HFQS and 120 C cure of high field Q-slope.

0 20 40 60 80 100 120 140 16010-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Chamber pressure

Cavity temperature

Elapsed time (h)

Pre

ssu

re (

To

rr)

TE1AES015 & TE1PAV007 20161116 SKC

0

100

200

300

400

500

600

700

800

900

Tem

per

atu

re (°C

)

Heat treatment:

800 °C, 3 h in UHV

160 °C, 48 h with N2 at 25´10-3 Torr

160 °C, 96 h in UHV

0 10 20 30 40 50 60 7010-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Chamber pressure

Cavity temperature

Elapsed time (h)P

ress

ure

(T

orr

)

TE1PAV010 20160106 SKC

0

100

200

300

400

500

600

700

800

900

Tem

per

atu

re (°C

)

Heat treatment:

800 °C, 3 h in UHV

120 °C, 48 h in UHV

0 10 20 30 40 50 60 7010-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Chamber pressure

Cavity temperature

Elapsed time (h)

Pre

ssu

re (

To

rr)

TE1AES015 20160519 SKC

0

100

200

300

400

500

600

700

800

900

Tem

per

atu

re (°C

)

Heat treatment:

800 °C, 3 h in UHV

160 °C, 48 h with N2 at 25´10-3 Torr

0 10 20 30 40 50 60 7010-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Chamber pressure

Cavity temperature

Elapsed time (h)

Pre

ssu

re (

To

rr)

TB9AES017 20160613 SKC

0

100

200

300

400

500

600

700

800

900

Tem

per

atu

re (°C

)

Heat treatment:

800 °C, 3 h in UHV

120 °C, 48 h with N2 at 25´10-3 Torr

o Bulk electro-polishing

o High T furnace:

• 800 C, 3 hours, high vacuum

• 120 C, 48 hours with N2

(25 mTorr)

o No chemistry post-furnace

o HPR, VT assembly

Slides on N-infusion are courtesy of A. Grasselino


Results: ILC recipe vs. nitrogen infusion

Same cavity, sequentially processed,

no EP in between

Achieved:

45.6 MV/m 194 mT, with

Q ~ 2e10!

Q ~ 2.3e10 at Eacc ~ 35 MV/m

Repeatable increase of Q by a

factor of two, increase of gradient

~15%

Potential application – ILC,

significant cost reduction of the

machine.

New potential breakthrough: very high Q at very high

gradients with low temperature (120C) nitrogen treatment

4/12/16Alexander Romanenko | FCC Week 2016 - Rome34

- Record Q at fields > 30 MV/m

- Preliminary data indicates potential 15% boost in achievable quench fields

- Can be game changer for ILC!



Kubo and Checchin models on bi-layer potentially

increasing achievable accelerating gradients This idea is supported by Checchin (FNAL) and Kubo (KEK) models on bi-layer structure (e.g. dirty N-doped

layer on clean Nb) – claim that can enhance the achievable accelerating gradient.

Ideal Depth of this layer? Can this trick help push beyond the 200 mT or achieve 200 mT with higher yield?

We are investigating this empirically via low-T N-infusion (different T and durations)

TTC@Saclay 40

In addition to the BL barrier, we have the second barrier due to the S-S

boundary. The second barrier is also imperfect: easily weakened by defects.

However, we have a second chance to stop the vortex penetration.

The S-S bilayer

structure

defect

defect

defect

T. Kubo, TTC Meeting 2016


Exploring doping/infusion parameter space

This is still very active area

of research:

o Nitrogen infusion at

various temperatures /

exposure times

o Doping with other

materials

o Better understanding

surface properties to get

insight on how to ”nano-

engineer” niobium for

different applications

High Q0 (e.g. LCLS-II)

High Q0 & High Eacc (e.g. ILC)



Large parameter space – T and duration being explored

Q ~ 6e10 at 15 MV/m!

Q > 3e10 at 31.5 MV/m!


0 5 10 15 20 25 30 35 40 4510

9

1010

1011

Q0

Eacc

(MV/m)


Is the evolution nearly complete?

1.3 GHz, 2 K

Nitrogen infusion

ILC cost reduction


Frequency dependence of Rs


Frequency dependence of Rs and non-equilibrium SC

Mattis-Bardeen theory predicts quadratic frequency dependence of Rs. However, the theory is

valid only at “zero” fields. Does it need modifications when we consider field dependence?

Another active research area. The following cavities were studied so far:

0 5 10 15 20 25 30 35 40 45109

1010

1011

Q0

Eacc (MV/m)

N-doping

EP/BCP

120 C baking

1.3 GHz, 2 K

650 MHz 1.3 GHz 2.6 GHz 3.9 GHz

EP

BCP

120 C baking

2/6 N-doping Slides on frequency dependence are courtesy of M. Martinello


Normalized 𝑹𝑻 𝟐 𝑲 for 120 C Baking

*Some measurements were admin limited between 15-20 MV/m to avoid quench so, in order to compare the different curves, only data till ~20 MV/m are shown

Slides on frequency dependence are courtesy of M. Martinello


Normalized 𝑹𝑻 𝟐 𝑲 for 120 C Baking


At low field RT follows the 2 trend suggested by the Mattis-Bardeen theory


Q-factor of 2.6 GHz at high field tends to the one at 1.3 GHz

120 C baked cavities

Q-factor of 2.6 GHz cavity converge to the one at 1.3 GHz at high gradients

T=2 K



Normalized 𝑹𝑻 𝟐 𝑲 for N-doping

Higher frequency leads to stronger anti-Q-slope!

Higher frequency is

favorable for Q, and can

be also for higher

gradients.

Understanding the

reversal of RBCS with the

RF field:

o The non-equilibrium

quasiparticle distribution

driven by microwave

fields

o Need solid theoretical

basis and

measurements of some

Nb properties

M. Martinello et al, https://arxiv.org/abs/1707.07582


Comparison in terms of Q-factor at 2.0 K


T = 2 K


Comparison in terms of Q-factor at 2.0 K


T = 2 K

1.3 GHz wins over 650 MHz at ~10 MV/m

3.9 GHz wins over 2.6 GHz at ~13 MV/m


Unprecedented medium field Q0 at 3.9 GHz


Q-factor of N-doped 3.9 GHz comparable to 120 C baked 1.3 GHz cavity at ~ 20 MV/m

T = 2 K

𝑄0~1.5 ∙ 1010

Further improvement with 900 C bake

0 20 40 60 80 100 120109

1010

1011

Q0

Eacc (MV/m)


Enabling future efficient HEP accelerators

Q > 2e10

At high field

Eacc > 100 MV/m

Q > 3e10

non-equilibrium SC?

new materials?

Nitrogen Infusion

Nitrogen infusion showed that it is possible to achieve both high Q and high

gradient at the same time.

We hope that further progress in SRF experiment and theory will allow us to

achieve much better performance and enable new particle accelerators.


SRF at low T & low field

(toward mK / single-photon scale):

from accelerators to quantum computers


Renewed (now practical) importance of the LFQS and low T

A. Romanenko et al, Appl. Phys. Lett. 105, 234103 (2014)

How will the best cavities we have behave at ultralow fields for

various applications? Quantum computing /

quantum memory Dark sector photons searches Gravitational effects search ….

These applications are

interested in high Q at very low fields

LFQS is present after all treatments

What is the cause of the low field Q slope and what happens with Q as we decrease the field further

LFQS


LFQS measurements toward quantum regime & TLS model

Q measured using a single-shot method (decay from PLL state)

Good news: LFQS stops below 0.1 MV/m with Q ~ 3x1010

Previous models: Halbritter,

Palmieri, Weingarten

From 2D resonator world: non-linear dissipation in two-level systems of an amorphous dielectric layer

Qu

alit

y F

acto

r

Eacc (MV/m)

0.001 0.01 0.1 1 10

1x1010

3x1010

5x1010

7x1010

9x1010

Saturation of the

Q decrease

Fit to TLS model

Ec = 0.1 MV/m

b = 0.19

CWSS RBW=10 kHzSS RBW=30 Hz

A. Romanenko and D. I. Schuster, Phys Rev Lett. 119, 264801 (2017)

121

2 =TTR

w

1tan -µ Ed

tan

d

E

T = 1.5 K


Effect of the oxide layer

Thicker oxide (anodization) has a drastic effect at low fields

Similar to how we characterize losses due to surface currents via the geometry factor, one can introduce a similar term, the surface participation ratio, characterizing dissipation due to surface dielectric losses

A. Romanenko and D. I. Schuster, Phys Rev Lett. 119, 264801 (2017)

Nb

Nb2O55 nm

Nb

Nb2O5100 nm


Next step: toward quantum regime

Demonstration of T = 10 mK and <N> ~ 1 photon high Q in 2018. Large dilution refrigerator allows exploring fundamental physics of residual resistance of

SRF cavities at very low temperature.

First SRF cavity is being mounted inside the dilution refrigerator at Fermilab


SRF accelerators around the world

RAONSoleil

Circular

Lightsrc

<10cavities

Linear

NPHEP

Produc’nOper’n10-100cavities

100-1000cavities >1000cavities

CEBAF

SNS

FRIBEICCLSISAC

LHCXFE

L

ESS

TLS

BESSY

ATLAS

FLASHCESR

C-ADS

LIPAcBEPC-IIHIE-ISOLDE cERL

SPIRAL2

ALPI

ALICEELBE

ANURIB

J-PARCSC-LINAC

SARAF

ANUUSP

PIP-II/IIIFCC

ILC

ISNSADS

MaRIELCLS-II CepC-SppC

SamPosen


SRF projects in progress / planned

EIC

SC-LINAC

SARAF

HIE-ISOLDE cERLJ-PARC

LIPAc

ANURIB

BEPC-II

TLS

ANUUSP

ALPI

BESSYELBE

FLASHSoleil

ALICE

LHC

ATLAS

CEBAF

SNS

CLS

CESR

ISACSPIRAL2

RAON

Circular

Planning

Light src

<10 cavities

Linear

NPHEP

Produc’nOper’n10-100 cavities

100-1000 cavities >1000 cavities

ESS

C-ADS

ISNSADS

MaRIE

PIP-II/IIIFCC

CepC-SppC

FRIBXFE

L

ILC

LCLS-II

Sam Posen


Major SRF Projects in Progress

Project No. of

Cryomodules

No. of

Cavities

Voltage

(MV)

LCLS-II 1.3 GHz 35 280 4000

LCLS-II 3.9 GHz 2 12 55

SCLF 75 600 8000

FRIB 46 328 200/nucleon

RISP 45 320 200/nucleon

SNS-upgrade 7 28 300 - 400

ESS 43 150 2000

LHC-HL 8 16

Total 250 1700 15 GV


Planned SRF Projects, Some ?

Project No. of

Cryomodules

No. of

Cavities

Voltage

(MV) LCLS-II-upgrade 20 160 4000

KEK-ERL 22 200 3000

euV ERL 9 72 800

FRIB-upgrade 46 328 200/nucleon

eRHIC ERL 18 72 1300

PIP-II 25 116 800

PIP-IIIa 18 110 2200

PIP-IIIb 22 176 5000

India SNS 27 126 1000

C-ADS 55 220 1500

Total 350 1900 25 GV


International Linear Collider

Overview of Future Colliders, H. ZhuInstitute of High Energy Physics

International Linear Collider (ILC)

• e+e- linear collider with Superconducting RF linac

• Baseline: √s = 500 GeV (31 km) → upgrade later to ~ √s= 1 TeV (50 km),

luminosity of 1.8 × 1034 cm-2 s-1 with optional upgrade, one interaction point

(IP) with two detectors: ILD and SiD with push-pull

4

Japanese Association of High Energy Physics (JAHEP) proposed the prompt construction in Japan of ILC as a Higgs Factory at 250 GeV. ICFA expressed its support for this ILC option.

The Linear Collider Board estimates cost of 250 GeV starting point will be 40% less than the cost of 500 GeV TDR cost.

A decision from the Japanese government is expected soon, in 2018? Continued SRF performance improvement R&D, cost reduction R&D and

optimization R&D will be very beneficial.

This will be the largest SRF project , >10 times E-XFEL


Summary SRF science and technology is going through a period of Renaissance.

There are several new research directions opened up in the last ~5 years, four

of each I reviewed in this presentation: nitrogen doping, nitrogen infusion,

frequency dependence of surface resistance, and SRF in quantum regime.

Other research areas, which I did not have time to discuss, include: nature of

losses due to trapped magnetic flux; study of flux expulsion; non-

equilibrium superconductivity and ultimate gradient limit; Nb3Sn SRF cavities

and other materials.

Nb-based SRF accelerator technology is mature and became the technology of

choice for many new SRF accelerators.

BUT: there are still many problems that need attention and careful

investigation.

These are exciting times and the field needs more young and energetic

researchers! There will always be ample opportunities for imagination, originality,

and common sense.


Thank you!

Radio frequency superconductivity for particle accelerators · 2018. 4. 12. · Radio frequency...

Documents

Transcript of Radio frequency superconductivity for particle accelerators · 2018. 4. 12. · Radio frequency...