Post on 20-Jul-2020
Overview of Flux Trapping at Cornell
J. T. ManiscalcoCLASSE, Cornell University
11/9/18 James Maniscalco 1
Introduction / Teaser Slide
11/9/18 James Maniscalco 2
Mean Free Path [nm]100 101 102 103
Rre
s,B/B
trapp
ed
[n Ω
/mG
]
0
1
2
3
4
5Experimental Data
Gurevich Theory, ℓp = 75ℓPower Fit: y = 16.2/
√
ℓ
• In general linear sensitivity with an “offset” (y-intercept) parameter and a “slope” parameter. Linear slope predicted by “collective weak pinning” (see D. Liarte’s talk later today!)
• Parameters dependent on cavity treatment and material parameters, especially the electron mean free path, doping depth, and frequency.
• All Cornell cavities made from the same Nb stock
Outline• Overview of flux sensitivity measurement procedure @ Cornell• Results for doped niobium
• 1.3 GHz high-T (800 ºC) N-doping (including “2/6” recipe)• 1.3 GHz 160 ºC impurity doping• 2.6 GHz “2/6” N-doping
• Results for Nb3Sn• 1.3 GHz cavities
11/9/18 James Maniscalco 3
Flux Sensitivity Test Procedure
11/9/18 James Maniscalco 4
Helmholz coil
I
I
T3
T2
T1B1
B2
B3
• Fast cool to expel mag. flux; slow cool to trap it
• Temperature sensors (Cernox) to measure ∆T at Tc
• Flux gate magnetometers to measure ambient field, applied field, expulsion, trapping
• Note: dT/dx across cellmatters, not dT/dt
• Though cooling quickly is aneasy way to get a large dT/dx.
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Bpk
(T)
0
0.5
1
1.5
2
2.5
3
3.5
4
RB
CS
()
10-8
1.6
1.7
1.8
1.9
2
2.1
2.2
Bath
tem
pera
ture
(K
)
RBCS
R0
Flux Sensitivity Test Procedure• One test with fast cool• One or more tests with slow
cools, varying trapped flux• Collect RF test data:
Rs(BRF , T) = G / Q0
• Exponential fitting (SRIMP algorithm) to separate Rs:
11/9/18 James Maniscalco 5
𝑅" = 𝑅$%& 𝐵(), 𝑇 + 𝑅-(𝐵())
Flux Sensitivity Test Procedure• Compare R0 from different
cooldowns to determine sensitivity:
11/9/18 James Maniscalco 6
𝑅- 𝐵trapped =𝑅- slowcool − 𝑅-(fastcool)
𝑆 = 𝑅- 𝐵trapped /𝐵trapped= 𝑎 @ 𝐵() + 𝑏
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
BRF
(T)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
R0/B
tra
pp
ed (
/mG
)
10-8
R0 = a*B
RF*B
trapped + b*B
trapped
a = 0.021 0.003 n /mTRF
/mGtrapped
b = 4.8 0.6 n /mGtrapped
a = 0.018 0.003 n /mTRF
/mGtrapped
b = 4.6 0.4 n /mGtrapped
16 2 mG trapped
36 3 mG trapped
Multiple cooldowns, same sensitivity!
Peak surface magnetic field (mT)
0 20 40 60 80 100 120 140
Qualit
y fa
ctor
Q0
109
1010
1011
C3(P1)
VEP niobium
1.3 GHz high-T N-doping• Treatment protocol:
• 800 ºC UHV degas bake (5-8 hr)• 800 ºC 40 mTorr (6 Pa) N2 doping bake (2-60 minutes)• 800 ºC UHV anneal bake (0-60 minutes)• Final VEP (2-20 µm)
11/9/18 James Maniscalco 7
Depth [ µm]0 20 40 60 80
Nitr
og
en
Co
nce
ntra
tion
[Ato
ms/
cm
3 ]
1018
1019
1020
1021
800°C: 20 min, 30 min anneal
990°C: 5 min
1.3 GHz high-T N-doping
11/9/18 James Maniscalco 8
Strong “offset”,no slope in BRF
“offset” highly sensitive to ℓ𝒆
See D. Gonnella et al. (J. Appl. Phys. 2016)and J. T. Maniscalco et al. (J. Appl. Phys. 2017)
E acc [MV/m]0 5 10 15 20
Rre
s [Ω]
10-9
10-8
10-7 Btrapped = 0 mG
Btrapped = 3 mG
Btrapped = 7 mG
Mean Free Path [nm]100 101 102 103
Rre
s,B/B
trapp
ed
[n Ω
/mG
]
0
1
2
3
4
5Experimental Data
Gurevich Theory, ℓp = 75ℓPower Fit: y = 16.2/
√
ℓ
Mean Free Path [nm]100 101 102 103
Rre
s,B/B
trapp
ed
[n Ω
/mG
]
0
1
2
3
4
5Experimental Data
Gurevich Theory, ℓp = 75ℓPower Fit: y = 16.2/
√
ℓ
1.3 GHz high-T N-doping
11/9/18 James Maniscalco 9
Sensitivity also likely dependent on properties of bulk Nb!(different cavity material at different labs)
M. Martinello et al.(J. Appl. Phys. 2016)
Flux expulsion also linked to material stock in doped cavities!see D. Gonnella et al. (NIM-A 2018), S. Posen et al. (J. Appl. Phys. 2016)
1.3 GHz high-T N-doping
11/9/18 James Maniscalco 10
Takeaway message:
Doping strongly impacts trapped flux losses:1) lowers the mean free path, which increases losses2) increases pinning effects which decrease losses
Nitrogen’s link to pinning is still not well understood.Other material parameters might be relevant too!
(see Danilo’s work on collective weak pinning – how material parameters could affect pinning strength and vortex loss magnitude)
1.3 GHz 160 ºC N-doping
11/9/18 James Maniscalco 11
• Treatment protocol:• 800 ºC UHV degas bake (5 hr)• UHV ramp down to 160 ºC• 160 ºC UHV rest (3 hr)• 160 ºC, 40 mTorr (6 Pa) N2 doping
bake (1 day for test shown here)• 160 ºC optional UHV anneal bake
(not used for tests shown here)
• Little/no post chemistry! Maybe HF rinse or oxypolish for light surface removal ≪ 1 µm.
Thinner doped layer, different impurity content,
different impurities
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
BRF
(T)
0
0.2
0.4
0.6
0.8
1
1.2
R0/B
trapped (
/mG
)
10-9
R0 = a*B
RF*B
trapped + b*B
trapped
a = 0.0034 0.0006 n /mTRF
/mGtrapped
b = 0.83 0.05 n /mGtrapped
42 2 mG trapped
Mean Free Path [nm]100 101 102 103
Rre
s,B/B
trapp
ed
[n Ω
/mG
]
0
1
2
3
4
5Experimental Data
Gurevich Theory, ℓp = 75ℓPower Fit: y = 16.2/
√
ℓ
1.3 GHz 160 ºC N-doping
11/9/18 James Maniscalco 12
Lower “offset” sensitivity, higher “slope” sensitivitycompared to 800 ºC-doped cavities
(preliminary)
1.3 GHz 160 ºC N-doping
11/9/18 James Maniscalco 13
Takeaway message:
Shallow 160 ºC doping gives lower sensitivity than 800 ºC dopingNote: dopant may be different…
Results indicate linear field dependencemore prominent for low-temperature dopings
(see again Danilo’s work/talk)
2.6 GHz “2/6” N-doping• Treatment protocol:
• 800 ºC UHV degas bake (3 hr)• 800 ºC 40 mTorr (6 Pa) N2 doping bake (2 minutes)• 800 ºC UHV anneal bake (6 minutes)• Final VEP (6 µm)
11/9/18 James Maniscalco 14
Mean Free Path [nm]100 101 102 103
Rre
s,B/B
trapp
ed
[n Ω
/mG
]
0
1
2
3
4
5Experimental Data
Gurevich Theory, ℓp = 75ℓPower Fit: y = 16.2/
√
ℓ
2.6 GHz “2/6” N-doping
11/9/18 James Maniscalco 15
“offset” sensitivity 2x higher than 1.3 GHz cavities with similar ℓDR0/Btrapped∝ ω?
see J. T. Maniscalco et al. (LINAC 2018)
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
BRF
(T)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
R0/B
trapped (
/mG
)
10-8
R0 = a*B
RF*B
trapped + b*B
trapped
a = 0.021 0.003 n /mTRF
/mGtrapped
b = 4.8 0.6 n /mGtrapped
a = 0.018 0.003 n /mTRF
/mGtrapped
b = 4.6 0.4 n /mGtrapped
16 2 mG trapped
36 3 mG trapped
2.6 GHz “2/6” N-doping
11/9/18 James Maniscalco 16
Takeaway message:
High frequency → stronger sensitivity
Initial results indicate linear behavior with frequency, i.e.R0/Btrapped∝ ω
1.3 GHz Nb3Sn• Treatment protocol:
• Nb3Sn grown on Nb cavity substrate by vapor diffusion
11/9/18 James Maniscalco 17
20× more efficient than Nb at 4.2 K!
1.3 GHz Nb3Sn
11/9/18 James Maniscalco 18
Typical results: linear sensitivity
More sensitive slope than doped Nb, similar
“offset” sensitivity
see D. L. Hall et al. (IPAC 2017)𝑑𝑅
𝑑𝐵FGHI= 21 ± 1 pΩ mG⁄ mT⁄ ⋅ 𝐵() + 0.47 ± 0.02 nΩ ∕ mG
3 pΩ mG⁄ mT⁄for 160 ºC doping
≤ 0.5 nΩ/mGfor clean Nb
compare:
M1 M2
1.3 GHz Nb3Sn
11/9/18 James Maniscalco 19
Also sensitive to thermoelectric currents
(Seebeck effect)
Cavity must be cooled uniformly!Need to limit flux sensitivity.
hot
cold
V1 →
V2 →
II
1.3 GHz Nb3Sn
11/9/18 James Maniscalco 20
Takeaway message:
Clear linear behavior, strong slopeLow “offset” sensitivity – similar to undoped cavities
Review
11/9/18 James Maniscalco 21
• Flux sensitivity typically linear in RF field• Coefficients of sensitivity linked to material parameters
• Electron mean free path• Doping depth• Impurity species• Frequency• Niobium stock• Nb3Sn vs. doped Nb vs. clean Nb
(vs. Nb/Cu films not addressed here)• Nb3Sn: R0 from trapped flux important because RBCS is so small!
• Higher “slope” sensitivity than doped Nb• Bimetallic interface necessitates slow, uniform cooling
and good magnetic shielding/hygiene
11/9/18 James Maniscalco 22
Thanks for your attention!
See also:
D. B. Liarte’s talk later today (11h59):“Vortex dynamics and hysteretic flux losses due to pinning”