Breakthrough Technology for Very

High Quality Factors in SRF Cavities

Alexander Romanenko

27th Linear Accelerator Conference (LINAC’2014)

2 Sep 2014

Outline

• Two major discoveries at Fermilab

– Nitrogen doping to lower the BCS surface resistance

• Lower than the previously perceived “theoretical limit”

• Example at 16 MV/m: Q >3x1010 by doping vs. Q~1.5x1010 with the

standard ILC/XFEL cavity processing

• Also lowers residual

– Effective magnetic flux expulsion by fast/high thermal gradient

cooldown to achieve record low residual resistances

• Example: Q = 2.7x1011 in 27 mG ambient field

• Immediate implications

– LCLS-II

• Future perspectives

2 Sep 2014 Alexander Romanenko | LINAC'2014 2

• First breakthrough at Fermilab: Nitrogen doping can

drastically lower “fundamental” BCS losses in cavity walls

and increase Q several times

2 Sep 2014 Alexander Romanenko | LINAC'2014 3

A. Grassellino et al, 2013 Supercond. Sci. Technol. 26 102001 (Rapid Communication) – selected for highlights of 2013

0 5 10 15 20 25 30 35 4010

9

1010

1011

Q0

Eacc

(MV/m)

T= 2K

Nitrogen doping: a breakthrough in Q

2 Sep 2014 Alexander Romanenko | LINAC'2014 4

Standard state-of-the art preparation

This was the highest Q possible up to last year

Record after nitrogen doping – up to 4 times higher Q!

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

1.3 GHz

Breakthrough in quality factor: nitrogen doping

2 Sep 2014 Alexander Romanenko | LINAC'2014 5

• Injection of small nitrogen partial pressure at the end of 800C degassing followed by several ums of EP-> drastic increase in Q

• Reproduced on tens of 1- and 9-cell cavities at FNAL

T=2K

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

Characteristic anti-Q-slope

Doping is easily reproducible process

2 Sep 2014 Alexander Romanenko | LINAC'2014 6

2.00E+010 4.00E+010 6.00E+010 8.00E+0100

1

2

3

4

Count

Q0 (Eacc=16 MV/m, T=2K)

LCLS-II spec

EP+120C

bake

Fermilab nitrogen/argon

doping technology

N total Mean Standard

Deviation

Minimum Median Maximum

Q0 20 4.3e10 1.2e10 3.2e10 4.0e10 7.4e10

Obtained Q several times above state-of-the-art

Reproduced at other labs in 1- and 9-cells

Cornell

2 Sep 2014 Alexander Romanenko | LINAC'2014 7

JLab

Cutouts from N doped cavities - SIMS

2 Sep 2014 Alexander Romanenko | LINAC'2014 8

Only about 40 ppm of N is enough

What does N treatment do? N depth profiles by SIMS

2 Sep 2014 Alexander Romanenko | LINAC'2014 9

Nitrides Interstitial nitrogen in Nb

Non-doped

Doped

Depth (um)

2 4 6 8 10 12 14 16 184

6

8

10

standard treatment

standard treatment

nitrogen treatment

nitrogen treatment

R2K

BC

S (

n

)

Eacc

(MV/m)

Physics – origin of the effect

2 Sep 2014 Alexander Romanenko | LINAC'2014 10

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

Anti-Q-slope emerges from the BCS surface resistance decreasing with field

This is what Mattis-Bardeen theory predicted to be the lowest possible surface resistance for Nb -> we breached it!

Rs(T) = RBCS(T) + Rresidual

Nanostructural studies provide first clues

2 Sep 2014 11

• Hydrides may be the cause of the medium and high field Q slopes [see A. Romanenko, F. Barkov, L. D. Cooley, A. Grassellino, 2013 Supercond. Sci. Technol. 26 035003]

• Nitrogen doping may fully trap hydrogen => only intrinsic Nb behavior is then manifested?

Y. Trenikhina (IIT/FNAL), A. Romanenko – to be published

Nb

Electron diffraction patterns from the penetration depth taken at 94K reveal the difference

Non-doped Nb

Doped Nb

TEM on FIB-prepared cutouts

Secondary diffraction peaks appear signalling the formation of lossy niobium hydrides

Nb lattice

Nb lattice

Alexander Romanenko | LINAC'2014

Some possible mechanisms for intrinsic Nb behavior leading to

increasing Q with field

• Momentum of Cooper pairs changes the DoS

– B. P. Xiao et al, Physica C 490 (2013) 26-31

• Quasiparticle energy distribution deviates from thermal

equilibrium

– P. J. de Visser et al, Phys. Rev. Lett. 112, 047004 (2014)

• Time-dependent DoS

– A. Gurevich, Phys. Rev. Lett. 113, 087001 (2014)

2 Sep 2014 Alexander Romanenko | LINAC'2014 12

• Second breakthrough at Fermilab: cooldown rate and thermal

gradients around Tc drastically affect the Meissner effect and

can be used to achieve ultra-low residual resistances even in

high ambient fields

2 Sep 2014 Alexander Romanenko | LINAC'2014 13

A. Romanenko, A. Grassellino, O. Melnychuk, D. A. Sergatskov, J. Appl. Phys. 115, 184903 (2014)

Magnetic probes reveal the new physics

Full expulsion of the magnetic field should increase the field at equator ~2 times when going superconducting

Fluxgate magnetometers

H

A. Romanenko, A. Grassellino, O. Melnychuk, D. A. Sergatskov, J. Appl. Phys. 115, 184903 (2014)

2 Sep 2014 Alexander Romanenko | LINAC'2014 14

Efficient flux expulsion

Poor flux expulsion

It turns out the expulsion efficiency can be controlled by the cooldown procedure through Tc=9.2K (fast/slow, uniform or not)

Same Meissner behavior for EP, EP+120C, N doping, fine/single grain, cooling is what matters

Bare N doped 9-cell in vertical test

Oct, 2013

Efficient flux expulsion

Poor flux expulsion

2 Sep 2014 Alexander Romanenko | LINAC'2014 15

0 5 10 15 20 251.0x10

10

1.2x1010

1.4x1010

1.6x1010

1.8x1010

2.0x1010

2.2x1010

2.4x1010

2.6x1010

2.8x1010

3.0x1010

3.2x1010

3.4x1010

3.6x1010

3.8x1010

4.0x1010

#1: First fast from 300K

#2: Slow from 15K

#3: Fast from 15K

Q0

Eacc

(MV/m)

Bare/dressed cavities behave identical

2 Sep 2014 Alexander Romanenko | LINAC'2014 16

Flux expelled efficiently

Flux mostly trapped

Thermal currents are present but have no effect

Dressed 9-cell – VTS measurements

17

No effect of thermal currents in VTS of dressed cavities

Large (125 mG) magnetic fields generated by thermal currents, no effect on Q

Cooldown from 300K

2 Sep 2014 Alexander Romanenko | LINAC'2014

18

Dressed cavities behave exactly the same as bare

-15

-10

-5

0

5

65000 70000 75000

0

5

10

15

20

25

Mag

ne

tic fie

ld (

mG

) Cell 5 axial

Cell 1 axial

Cell 1 tangential

Cell 9 axial

Tem

pe

ratu

re (

K)

Time (sec)

Cell 1

Cell 1

Cell 9

Cell 9 - Cell 1

Warmup after slow

Fast from 15K

Everything the same as in our J. Appl. Phys. 115, 184903 (2014)

2 Sep 2014 Alexander Romanenko | LINAC'2014

Confirmed at Cornell and Jlab – VTS and HTS

2 Sep 2014 Alexander Romanenko | LINAC'2014 19

Cornell HTS measurements of the Fermilab N-doped 9-cell [See MOPP018 ]

JLab VTS data on a 9-cell [See TUPP138 ]

Possible mechanism #1

• Thermal gradient at the superconducting/normal conducting

boundary is aiding the flux expulsion => the higher dT/dx the

better (fast and from higher temperature preferred)

• See [J. Appl. Phys. 115, 184903 (2014)] for details

2 Sep 2014 Alexander Romanenko | LINAC'2014 20

Example of thermal difference across the 1-cell cavity

Possible mechanism #2

• See [J. Appl. Phys. 115, 184903 (2014)] for details

2 Sep 2014 Alexander Romanenko | LINAC'2014 21

Fast Slow

For this mechanism uniformity is “bad” -> leads to islands

22

Optimal cooling conditions produce record high Qs even when

ambient fields are large

Q=2.7e11 in 23mG!

2 Sep 2014 Alexander Romanenko | LINAC'2014

Also means that magnetic shielding requirements may be relaxed

Implications

• CW operation becomes possible for short-pulse machines

• LCLS-II at SLAC

– Adopted N doping as a baseline

• 2x higher Q leads to a factor of 2 decrease in required refrigeration => 1

cryoplant less = savings of ~50-100M$ in capital costs + 10s of M$ in

operational costs

• PIP-II at FNAL

– Increase of duty factor from 5% to 30% desired by Mu2e becomes

possible with doubling the Q

• Q can be preserved/improved and the magnetic shielding requirements

relaxed if the optimal cooling is mastered in the cryomodule

– Looks straightforward, no show stoppers here

• High gradient/high Q becomes possible => future e+e- machines at

drastically reduced costs

2 Sep 2014 Alexander Romanenko | LINAC'2014 23

Impact of Increasing Q on optimal Gradient

• ILC requirements: Eacc > 35 MV/m/ Q0 > 8x109

• Improving Q0 from 8x109 to 3x1010 has a big impact on machine

cost

• Optimal gradient moves to ~ (70 - 80) MV/m

2 Sep 2014 Alexander Romanenko | LINAC'2014 24

Conclusions

• We have two new breakthroughs increasing the Q

– Nitrogen doping

• Doped cavities become even more efficient at higher fields

– Efficient flux expulsion

• Opens up the route to minimize the residual resistance even in

poorly shielded realistic environment

– May allow to relax the specs on magnetic shielding

• Exciting time in SRF

– We will follow the science of these discoveries further and see

where it leads us

2 Sep 2014 Alexander Romanenko | LINAC'2014 25

THANK YOU

2 Sep 2014 Alexander Romanenko | LINAC'2014 26

27

BACKUP: Thermal currents easily minimized/removed by

cooling fast from lower T – for example ~20K

Thermal gradient

No thermal currents

2 Sep 2014 Alexander Romanenko | LINAC'2014