Progress in high Q and high gradient R&D Anna Grassellino - - PowerPoint PPT Presentation

Progress in high Q and high gradient R&D

Anna Grassellino Tesla Technology Collaboration Meeting Vancouver, TRIUMF, February 2019

Outline

• Two main research directions to push high Q at high

gradients:

– High temperature (> 800C) nitrogen doping – Low temperature treatments ( ~ 50C-200C) with or without nitrogen

• With increasing importance of cooldown studies/details in the

whole temperature range ~300->2K

• A possible common matrix that ties it all together: nano-

hydrides?

• Theoretical advancements and path forward

Grassellino - Progress in High Q/high G 2

State of the art in high Q and high G (1.3 GHz, 2K)

• Q>3e10 @35 MV/m with N doping
• Q >1e10 at 49 MV/m (Bpk = 210 mT) with 75/120C bake

Grassellino - Progress in High Q/high G 3

EP 120C bake 75/120C N infusion N doping

Important: insufficient bulk removal or high defect density material (insufficiently annealed) will cause extra residual resistance

Breakdown in Surface Resistance: RBCS and R0 field dependence for state of the art treatments

• Largest advantage in Q from high T doping comes from reversal of RBCS (factor of 3-5

lower at mid field than 120C/EP) PLUS lowest residual resistance (EP grown oxide,

• n nitrogen enriched layer)
• 120C Infusion reduces RBCS compared to regular 120C bake, PLUS has lowest

residual (oxide grown in furnace, on oxide enriched layer)

• 75/120C gives same RBCS as N infusion…why? But higher residual (as regular 120C)
• What is behind these field dependencies? What is this knee between increase and

decrease of the BCS resistance?

Grassellino - Progress in High Q/high G 4

High T Doping

Grassellino - Progress in High Q/high G 5

• Record Q values achieved at all frequencies (see breakout

talks by Martinello for 650MHz and Bafia for 1.3GHz)

Grassellino - Progress in High Q/high G 6

High Temperature Doping is key for Highest Q T=2K

See also Grassellino et al, Superconductor Science and Technology, Volume 26, Number 10 Martinello et al Phys. Rev.

• Lett. 121, 224801

650 and 2.6 data data to be published (FNAL)

Where do we stand with high T N doping quench fields?

• High T N doping is key to record Q values: produces systematically

lowest BCS (and residual) surface resistance values: eg Q~6e10 @ 20 MV/m, 1.3 GHz, 2K! Or Q ~ 7-8e10 @ 650 MHz, mid field, 2K

• Achievable quench field has evolved – from being limited to ~20 MV/m

in earlier days to up to 35 MV/m today, in single and nine cells

• What are the important steps that have led to such performance

improvement?

Grassellino - Progress in High Q/high G 7

?

Grassellino et al, Superconductor Science and Technology, Volume 26, Number 10 Bafia et al, TTC ARIES @ CERN

“Recipe” changes yielding gradient advancements

• A first big step in ~2014 @ FNAL was achieved moving from longer to

shorter duration of doping (example 20/30à 2/6)

• Recently, further improvements have been reached with some new

doping “recipes” (see Gonnella, Bafia and Palczewski talks) from simple tweaks (2/0) to more dramatic changes (3/60)

Grassellino - Progress in High Q/high G 8 23 MV/m 23 MV/m 27 MV/m 33 MV/m

High Q R&D for LCLS-2, 2015 High Q R&D for LCLS-2 HE, 2018

Sequential Doping Study of same cavity

9

2/6 + 5um EP (baseline): +40um EP reset 2/0 + 5um EP:

• Higher Q and quench

increases by +6MV/m +40um EP reset 3/60 + 5um EP:

• Quench improves by

additional +2MV/m, Q0=6E10 @ 20MV/m!

Grassellino - Progress in High Q/high G

f=1.3GHz T=2K

• One of the leading thoughts on quench in N doped cavities has been that

higher concentration/lower mfp could reduce the quench field (corroborated by the fact that lighter doping or deeper EP typically yield higher gradients)

• In reality, data does not show a clear correlation with mean free path
• More detailed SIMS studies ongoing to systematically relate surface N

concentration to achievable field

Role of mean free path/nitrogen concentration?

Grassellino - Progress in High Q/high G 10

Hc1

GL(0K)

Quench above theoretical Hc1 Hc1

GL(0K)

Open squares data points from M. Checchin talk @TTC RIKEN 2017, new data from D. Bafia in solid colour, presented at TTC ARIES @CERN

Palczewski, Reece MOPB039 Proceedings of SRF2015, Whistler, BC, Canada

Nano-Hydrides in N doped cavity cutouts (Romanenko/Sung)

Grassellino - Progress in High Q/high G 11

300K 200K 100K

See Z. Sung (FNAL) breakout talk

Nanohydrides form in the range 200-100K, fewer than other treatments and size ~30-50 nm

300K 200K 100K Temp [K]

A proposed model to explain quench in N doped cavities

Grassellino - Progress in High Q/high G 12

• Field of first vortex entry will depend on size of superficial defects compared to coherence

length

• Doping recipe and final N level modifies the coherence length (mfp) but also size of hydrides
• Think of hydrides as surface ‘defects’ that will lower field of first entry
• Possible that N doping brings the coherence to unfavorable point compared to other

treatments, coherence length comparable to size of the hydrides (which is exactly the case)

• Possible pathway forward: decouple coherence from hydrides size (move to dirtier or cleaner
• r longer second step outgassing cycles e.g. 3/60min to reduce hydrides size)

NbH ~40 nm

N doping 120C bake EP z ~ 40-100 nm z ~ 2-10 nm z ~ 1000 nm

NbH ~40 nm NbH < 10 nm

Model under development, Grassellino and Sauls (Northwestern U, CAPST)

• The new 3/60 recipe (see Palczewski, Bafia, Gonnella, Martinello) leads to even

further reduction in Rbcs (B), leading to extraordinary Q > 6e10 at 2K

• To be studied and validated: is this related to smaller size of nanohydrides or fewer

due to the longer post doping anneal time (giving a larger avg gap)

• Mean free path/concentration of 3/60 seems not too distant from 2/6 recipe so

cannot explain by itself the reduction in BCS and especially the stronger reversal

• Interesting question: how much lower can we go in Rbcs?

Pushing the Q even further via high T doping

Grassellino - Progress in High Q/high G 13

T = 2K f=1.3GHz

Bafia et al, TTC ARIES @ CERN Bafia et al, TTC ARIES @ CERN Martinello et al, Appl. Phys. Lett. 109, 062601 (2016)

Low T treatments

Grassellino - Progress in High Q/high G 14

The new 75/120C findings

Grassellino - Progress in High Q/high G 15

• We have recently focused our attention to the unexpected finding that a

pre-120C bake step of ~4 hours at 75C seem to lead consistently to unprecedented accelerating gradients ~49 MV/m (210 mT, TESLA shape)

• However, under the ILC cost reduction effort, as we study more and more

cavities, and exchange cavities worldwide, some new interesting findings are emerging in terms of Q and achievable accelerating gradient cooldown dependence

75/120C bake cavities

See Grassellino et al arXiv:1806.09824

Finding 1: the strange ‘branching’ performance for 75/120C

• On dozens of tests and several cavities now, we see switch in performance for

same cavity with no retreatment in between (always under vacuum)

• Effects of magnetic fields, dewars, cables, top plates have been excluded
• Some correlation has been found with cooldown speed near room T and starting T

~320-340K

• See Daniel Bafia breakout talk for many details on this study

Grassellino - Progress in High Q/high G 16

Bafia, Grassellino, to be published

Two 75/120C cavities sent from FNAL to Jlab and Cornell

• Cornell gradient matches our 49 MV/m (see Maniscalco breakout talk)
• Jlab reproduced exactly the upper/lower branching behavior in two

separate cooldowns (see Palczewski breakout talk)

• Two more cavities on their way to DESY and KEK

Grassellino - Progress in High Q/high G 17

5 10 15 20 25 30 35 40 45 50

Eacc (MV/m)

109 1010 1011

Q0

Cornell FNAL

Courtesy of Liepe, Maniscalco, Cornell Courtesy of Palczewski, Jlab

More puzzling differences – infusion cavities at KEK and DESY

• FNAL sent infused cavities to KEK and DESY for retest (see

Umemori and Wenskat talks in breakouts)

• Substantial differences seen in Rbcs and for different cooldowns,

but similar residual and quench fields

Grassellino - Progress in High Q/high G 18

BCS nearly doubled! Why?

Grassellino et al, Superconductor Science and Technology, Volume 30, Number 9 Courtesy of Kensei Umemori, KEK

• Substantially lower Q and G from 350K/top cooldown
• BCS decreases, residual increases a lot, the “knees” move

at corresponding points with a ‘breakdown’ field compatible with the proximity effect model of nanohydrides as introduced by Romanenko Superconductor Science and Technology, Volume 26, Number 3

• Cooldown From 294K vs ~350K

Cooldown from bottom vs top

Grassellino - Progress in High Q/high G 19

Finding 2: unequivocal performance change for regular 120C bake

Grassellino, Bafia, to be published

Non equilibrium behavior of surface resistance shifting earlier

Cool Down Profiles ad fluxgates of AES010: zero B field

Grassellino - Progress in High Q/high G 20

Fully compensated ZERO B field (longitudinal), close to zero transverse

300K 10 um Heating from 300K Preliminary AFM studies of 120C bake sample warming up from 300 to 380K

Courtesy of Z. Sung, to be published

Grassellino - Progress in High Q/high G 21

320K 10 um Heating from 300K

Courtesy of Z. Sung, to be published

Grassellino - Progress in High Q/high G 22

340K 10 um Heating from 300K

Courtesy of Z. Sung, to be published

Grassellino - Progress in High Q/high G 23

360K 10 um Heating from 300K

Courtesy of Z. Sung, to be published

Grassellino - Progress in High Q/high G 24

380K 10 um Heating from 300K almost exactly what you would expect at 100C…”boiling” like behavior

Courtesy of Z. Sung, to be published

Grassellino - Progress in High Q/high G 25

360K 10 um Cooling from 380K

Courtesy of Z. Sung, to be published

Grassellino - Progress in High Q/high G 26

340K 10 um Cooling from 380K A new ‘chickenpox’ appears!

Courtesy of Z. Sung, to be published

Grassellino - Progress in High Q/high G 27

340K Cooling from 380K

Courtesy of Z. Sung, to be published

Grassellino - Progress in High Q/high G 28

• 0.50

4.50 9.50 14.50 19.50 24.50 29.50 34.50 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

Height [nanometer]

Diameter [micrometer]

Chart Title

Particle A Particle B

Particle A: as-reside particle from 300K Particle B: appeared on 380K scan

Courtesy of Z. Sung, to be published

Grassellino - Progress in High Q/high G 29

Explaining field dependences with proximity breakdown (aka Romanenko “I said so 6 years ago”)

Grassellino - Progress in High Q/high G 30

Romanenko et al Superconductor Science and Technology, Volume 26, Number 3

Explaining accelerating field limitations

Grassellino - Progress in High Q/high G 31

NbH ~40 nm

N doping 120C bake EP z ~ 40-100 nm z ~ 2-10 nm z ~ 1000 nm

NbH ~40 nm NbH < 10 nm Nanohydrides too small compared to coherence à late proximity breakdown à HFQ slope Nanohydrides very large compared to coherence à early proximity breakdown à MFQ slope Nanohydrides comparable to coherence à no MFQS, no HFQS, ‘violent’ proximity breakdown causing earlier quench

Conclusions

• Exciting time, steady progress in high Q and high G for

niobium SRF cavities

• Worldwide collaboration working well highlighting

differences

• Boundaries of niobium potential are not reached, as

extrinsic mechanisms (eg nanohydrides) are still affecting performance

• It is time to study and understand in detail nano-hydrides

formation in different temperature regions for different processing and learn optimal preparation AND cooling parameters for performance maximization

Grassellino - Progress in High Q/high G 32

Backup slides

Grassellino - Progress in High Q/high G 33

Anomaly found during the low T bake

• A thermocouple went faulty and oven went to standby
• Cavity lingered around 75C for about 2 hours, then resumed

the 120C 48 hours

Grassellino - Progress in High Q/high G 34

Repeated on second cavity TE1AES009 (fine grain, AES, WC)

Grassellino - Progress in High Q/high G 35

5 10 15 20 25 30 35 40 45 50 55 109 1010 1011

ACC003: EP+120C - regular 1DE3: Modified 120C bake 1DE3: Re-calibrate/check AES009: Modified 120C bake AES009: cooldown #2 AES009: cooldown #3

Q0 Eacc (MV/m)

EP+ 75C 4 hrs+ 120C 48 hours Regular 120C

• A. Grassellino et al, https://arxiv.org/abs/1806.09824

Next surprise: the BCS resistance is ~ as 120C N infused

Grassellino - Progress in High Q/high G 36

4 8 12 16 20 24 28 32 36 40 44 48 2 4 6 8 10 12 14 16

1DE3 - 70C/120C bake 1DE20 - 70C/120C bake TE1AES015 - 120C infused TE1AES015 - regular 120C bake

Residual resistance (nOhm) Eacc (MV/m)

• A. Grassellino et al, https://arxiv.org/abs/1806.09824

Grassellino - Progress in High Q/high G 37

5 10 15 20 25 30 35 40 45 50 5 10 15 20 25 30 35 40 45 50 4 6 8 10 12

Post oxidation AES015C: 120C 48 hrs Before oxidation AES011: 120C 48 hrs / no N2 AES015D: 120C 48 hrs / with N2 PAV010F: 120C 48 hrs / with N2 Infusion restored AES018 - 120C 24 hrs with regular caps AES025 - 120C 60 hrs BCP AES021 - 120C 60 hrs EP 1DE3 70C/120C 1DE20 70C/120C

T=2K RBCS (nOhm) Eacc (MV/m)

Next surprise: the BCS resistance is ~ as 120C N infused

So, what is happening?

• 70C seems to be another magic temperature in niobium
• 1960-70s literature studies suggest that 70C is associated

with changes in vacancies, while 120C changes in dislocations (Bordoni or Hasiguti type process)

Grassellino - Progress in High Q/high G 38

JOURNAL OF MATERIALS SCIENCE 2 (1967) 559-566

The oc and 13 Peaks in Cold-Worked Niobium

• M. W. STANLEY*, Z. C. SZKOPIAK

Department of

Metallurgy and Mater&Is Technology, University of Surrey, London, SW11, UK Received 5 June 1967

Internal-friction measurements (at 1 c/sec) have been carried out over the temperature range from 90 to 290 ~ K, on niobium specimens deformed at room temperature. In the as-cold-worked material, a broad peak at about 115 ~ K (o~ peak) is observed. The ~ peak increases with the amount of deformation and decreases with increasing interstitial impurity content. On subsequent annealing, the height of the peak decreases by about 50~ over the temperature range from 90 to 140 ~ C, and to negligible values from 250 to 340 ~ C. As a result of annealing for 2 h at 70 ~ C, a group of peaks (fl peaks) occurred at about 200 ~

• K. The fl peaks are independent of the amount of deformation prior to annealing and

the interstitial impurity content. On further annealing, the relaxation strength of the peaks increases with temperature up to about 100 ~ C, remains constant between 100 and 240 ~ C, and subsequently gradually decreases to negligibly low values at about 340 ~ C. The a peak, and its variation with deformation, impurity content, and annealing, can be accounted for in terms of relaxation mechanisms involving dislocations (i.e. a Bordoni-

• r Hasiguti-type process observed in fcc metals). This is a generally accepted concept at
• present. The fi peaks, on the other hand, could only be adequately accounted for by

relaxation processes involving complexes of deformation-created point defects and inter- stitial impurities.

• 1. Introduction

Internal-friction relaxation peaks at sub-zero temperatures in cold-worked niobium have been denoted a and fl peaks by Chambers and Schultz [1, 2]. At a frequency of oscillation of 1 c/sec, the ~ peak occurs at about 100 ~ K and the fl peak at about 200 ~ K. The spectrum of internal friction at these temperatures indicates that, in each case, there is more than one relaxation process in operation [2, 3]. In fcc metals, relaxation peaks occurring at sub-zero temperatures are caused by mechanisms involving either pure dislocations (Bordoni peak) or dis- location-point-defect relaxations (Hasiguti peak). The relaxation mechanism of the Bordoni peak is based on the thermally activated motion of dislocation kinks over energy barriers along the Peierls barrier [4]. The Hasiguti mechanism, on *Now at the Department of Physical Metallurgy~ University

• f Birmingham, UK.

the other hand, involves the breaking away, under cyclic stressing, of dislocation lines from their atmospheres of point defects. Before these dislocations have had the opportunity to return to their original positions, the point defects migrate to them and re-pin them. The repeated breaking and re-forming of the pinning points results in a relaxation peak [5]. It is now clear that the a and /3 peaks in bcc metals cannot be accounted for by the models used to explain the Bordoni and Hasiguti peaks in fcc metals [2]. This is primarily due to the basically different annealing behaviour of the ~ and/3 peaks from the Bordoni and Hasiguti peaks respectively. It is, however, currently agreed that the motion of dislocations is directly involved in the relaxation processes causing the a peaks. With regard to the/3 peaks, Cham- 559

So, what is happening?

• These finding may be suggesting that quench in Nb is

currently of extrinsic nature, likely nano-hydrides, and that changes in vacancies or dislocations happening at magic temperatures are helping suppressing their formation, or changing their phase and size (see Z. Sung studies, TTC Milan)

Grassellino - Progress in High Q/high G 39

At 4 K 800°C + BCP on hot spot cut-out 120°C baked cavity cut-out

Positron Annihilation Studies on Nb sample

Grassellino - Progress in High Q/high G 40

Figure A.2: S-parameter at a fixed positron energy E = 2 keV vs. baking temperature for a single grain BCP sample. Note the onset of the S-parameter decrease at T = 80±C.

• A. Romanenko Ph.D. Thesis,

Cornell University (2009)

Combined with N infusion results, data so far suggest that the key to high gradients requires hitting magic temperatures

Grassellino - Progress in High Q/high G 41 Temperature Duration of N2 injection (@25 mTorr) Eacc max, average Q @ 21 MV/m -2K Q@ 35 MV/m - 2K Limitation Average on # single cell cavities 120°C 24 hrs 38 MV/m 2.5e10 1.7e10 Q slope @30 1 120°C 48 hrs 43 MV/m (max 45.6) 2.5e10 2.3e10 Quench 6 75/120°C 4hrs/48 hrs >46 MV/m (max 49 MV/m) 2.3e10 2e10 Quench 4 120°C 48 hrs w/o N2 36 MV/m 2.5e10 1e9 Q slope @30 2 120°C 60 hrs 43 MV/m (max 44.5) 3e10 2.5e10 Quench 3 120°C 60 hours (BCP) 33 MV/m 2.7e10 ~2e10 @30 Q slope @28 1 120°C ** 90 hrs 42 MV/m 2.3e10 2e10 Quench/slope 2 ** (non well annealed NX) 140°C 48 hrs 35 MV/m 2.5e10 Quench 2 160°C 48 hrs 36 MV/m 3e10 1e10 Q slope@30 1 160°C 48 hrs with N2/48 wo 35 MV/m 4e10 2.5e10 Q slope@25 1 160°C 48 hrs with N2/96 wo 34 MV/m 3e10 8e9 Q slope@25 1 170°C 48 hrs with N2/48 wo 27 MV/m 4e10

• Quench/Q

slope @25 2 200°C 48 hrs 28 MV/m 3.5e10

• Q slope @15

1