MATERIALS FOR SUPERCONDUCTING ACCELERATORS: BEYOND BULK Nb - C.Z. Antoine - Indico

Page created by Cheryl Williams
 
CONTINUE READING
MATERIALS FOR SUPERCONDUCTING ACCELERATORS: BEYOND BULK Nb - C.Z. Antoine - Indico
MATERIALS FOR
 SUPERCONDUCTING
 ACCELERATORS:
 BEYOND BULK Nb

SRF 2021 Tutorials

C.Z. Antoine
+ some material gathered by A.M. Valante-Feliciano
MATERIALS FOR SUPERCONDUCTING ACCELERATORS: BEYOND BULK Nb - C.Z. Antoine - Indico
Preamble:
 - many slides will not be presented but will
 be available if you download the
 presentation.

 - if not mentioned in the slide, references
 are available in the comment page

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 2
MATERIALS FOR SUPERCONDUCTING ACCELERATORS: BEYOND BULK Nb - C.Z. Antoine - Indico
RF SUPERCONDUCTOR:
 CHOICE CRITERIA ?

C.Z. Antoine -SRF 2021 tutorial - Beyond Nb

| PAGE 3
MATERIALS FOR SUPERCONDUCTING ACCELERATORS: BEYOND BULK Nb - C.Z. Antoine - Indico
THOUSANDS OF SUPERCONDUCTORS …

 Thousands of SC exist, ~10 are currently used for applications, only
 bulk Nb works well for SRF !!!

 Source: © 2018 Macmillan Publishers
 Limited, part of Springer Nature

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 4
MATERIALS FOR SUPERCONDUCTING ACCELERATORS: BEYOND BULK Nb - C.Z. Antoine - Indico
IDEAL SRF MATERIAL: TAILORED FOR APPS

 High RRR not required for superconductivity
 but for thermal stabilization in case of defects

 Today RF side Future ? Protective Layer
 e.g. oxide layer = diffusion barrier, or
Natural oxide, not low secondary emission layer (↓ MP)
stable @ T>150°C
 “RF” layer
 Optimized for superconductivity:
 Low surface resistance (medium m,f,p,)
High RRR Nb, and/or High HP (high HC1 high HSH)
optimized for thermal
transfer, in case of
dissipation “thermal conducting” layer
 Optimized for thermal transfer (high RRR,
 high m,f,p., not necessarily Nb)

 “External” layer
Natural oxide, not
optimized to reduce Optimized for thermal transfer (low
Kapitza resistance Kapitza resistance, e,g, anodization)

 He side
 1000
 RBCS
 ( nW ) 4,5 K
 “RF” layer optimized for superconductivity: 900
 800
 Low surface resistance (medium m.f.p.) 700

 and/or* 600

 High HP (high HC1 or high HSH) 500
 400

 *Depends on the application 1 10 100
 l ( nm )
 1000
 (l  RRR)
 10000

 2021/06/24 C.Z. Antoine -SRF 2021 |tutorial
 5 - Beyond Nb
MATERIALS FOR SUPERCONDUCTING ACCELERATORS: BEYOND BULK Nb - C.Z. Antoine - Indico
ULTIMATE LIMITS IN SRF-1

 Niobium superconducting radiofrequency cavities
 Performances
 Eacc  HRF
 HPeak
 Q0 ( 1/RS)  TC => Nb3Sn, MgB2, NbN… (but not YBCO)
 Limit = magnetic transition of the SC material @ Hpeak
 HMax.
 Superconductivity only needed inside :
 Thickness ~10 l, ~ < 1 µm => thin films (onto a thermally
 conductive, mechanically resistant material, e.g. Cu)
 Today :
 Thin films exhibit too many defects
 Only Bulk Nb has high SRF performances (high Q0 and high Eacc)
 0
 Issues : getting “defect free” superconductors
 H field mapping in
 an elliptical cavity

 (Yes but not all defects are detrimental... See doping !)

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 6
MATERIALS FOR SUPERCONDUCTING ACCELERATORS: BEYOND BULK Nb - C.Z. Antoine - Indico
HIGH Q0, EACC IN SRF => MEISSNER STATE !

 Mixed state w.Vortex
 SC phase diagram (i.e. N. cond. flux line +
 screening currents)
 All SC applications except SRF: mixed state w. vortex
 ● Vortices dissipate in RF !
 H
 SRF => Meissner state mandatory ! HC2
 Limit ?
 HC1 = limit Meissner/mixed state HC
 ● Nb: highest HC1 (180 mT)
 Or HC1
 HSH “Superheating field”: Metastable state favored by H
 // to surface
 ● Difficult to get in real life !

 Surface resistance:
 2
 RBCS  A(l ,  F , l,  n )
 4
 L e   / kT Meissner state TC T
 T Screening current over l,
 High Tc is better no magnetic field deeper
 T
MATERIALS FOR SUPERCONDUCTING ACCELERATORS: BEYOND BULK Nb - C.Z. Antoine - Indico
VORTEX PENETRATION WITH B //

 Surface barrier J
 (Bean & Livingston, 1964)

 Boundary condition. (J = 0) ≡ “image” vortices
 H0
 Supercurrent tends to push Vx inside
 Image antivortex tends to pull it out

 Before entering the material Vx have to cross a surface
 barrier:
 Image Vortex
 Vx thermodynamic Potential : 
 − J = 0 
 = 0 0 ൗ − 2 + 1 − 0
 l
 Image
G

 Meissner
 H0 < Hc1 “Ideal surface”
 Barrier disappears only at HSH~HC >HC1
 H0 = Hc1 Rationale used to predict SRF limits
 
 BUT
 H0 > Hc1
 If  localized defect w.: H 
 c ≪ Hc (or Tc ≪
 
 H0 = Hc T 
 c ) => early penetration of 1 or several Vx there
 ↗

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 8
MATERIALS FOR SUPERCONDUCTING ACCELERATORS: BEYOND BULK Nb - C.Z. Antoine - Indico
SUPERCONDUCTORS FOR SRF ?

 n µ0HC1 µ0HC2 µ0HC µ0HSH l  
 Material TC (K) Type
 (µWcm) (mT)* (mT)* (mT)* (mT)* (nm)* (nm)* (meV)

 Pb 7,1 n.a. n.a. 80 48 I

 Nb 9,22 2 170 400 200 219 40 28 1.5 II

 NbN 17,1 70 20 15 000 230 214 200-350 11 000 100-200 80-160 210-420 5,4

 NbTiN 17,3 35 30 150-200
MATERIALS FOR SUPERCONDUCTING ACCELERATORS: BEYOND BULK Nb - C.Z. Antoine - Indico
WHAT IS THE ACTUAL LIMIT (HP/HC1/HSH) ?

 Avalanche penetration/flux jumps

 ~100 µm in 1 ns (~RF period)
 
 Compare with l (field penetration depth)
 Nb : ~ 40 nm
 MgB2 ~ 200 nm
 Avalanche : high RF dissipation
 MgB2: http://www.nature.com/srep/2012/121126/srep00886/full/srep00886.html?message-global=remove&WT.ec_id=SREP-20121127

 In real world, cavities behavior is dominated by a few number of defects
 It is very important to measure the penetration field of samples in realistic conditions

 H H
 H//
 HSH HSH H//

 HC1 HC1

 H H
 T2C T2 T2C T2
 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 11
HSHNb3Sn
 (~ 400 mT @ 0 K) EFFECTS OF LOCAL DEFECTS

 Hays. "Measuring the RF critical field of Pb,
 Nb, NbSn". in SRF 97. 1997.
 Nb HC1
 (170 to 200 mT)

 Nb3Sn
 Cornell, 1997
 pulsed
 vs
 Nb

Recent results from
Cornell & FNAL CW
Posen,
2019 (23 MV/m)
2015 (17 MV/m)) 
 = 0 ቈ − ( ) ቉
 
 HC1Nb3Sn (~27mT)
 (50 mT pour le bulk) Dissipations :

 ~ 2K for Nb
 3,5 K for Nb3Sn

 Vortices enter more easily at lower temperature (counter intuitive !)?
 @ T~TC : H is low => low dissipations => easy to thermally stabilize
 @ T high dissipations => Favors flux jumps

 => We have to reduce defect density (yes but which ones?)

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 13
CHALLENGES TO FACE
ON THE ROUTE TOWARD OTHER
SUPERCONDUCTORS:
GENERALITIES

 C.Z. Antoine -SRF 2021 |tutorial
 PAGE- 14
 Beyond Nb
GENERAL ISSUES WITH SCs

 Needed: high TC, high HSH EX. : Grain boundaries
 (by defect high HC1)
 Advantages of niobium: pure metal.
 Highest TC of metallic SC, HC1
 Easy to form
 Uniform composition, no phase transition in the domain of interest
 Very large : makes it less sensitive to small crystalline defects
 (e.g. GB)
 Issues with alloyed, metallic SC compounds (e.g. NbTi) Some nm Compare with 
 Higher TCs, but smaller HC1, 
 Still relatively easy to form (harder)
 Usually several phases, not all of them SC
 Risk of non homogeneity

 Issues with non metallic SC compounds
 Higher TCs, but smaller HC1, 
 Brittle, no forming is possible, only films (OK for SRF, but a more
 complex fabrication route is needed)
 Usually several phases, not all of them SC
 Risk of non homogeneity
 http://www.phys.ufl.edu/~pjh/grain-boundry.html
 Sometimes local disorder =>
 ≠ local composition, possibly non SC
 Weak links e.g. NC grain boundaries
 = main reason why HTC do not apply in SRF .

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 15
THIN FILMS DEPOSITION

Nb : l~50 nm => only a few 100s nm of SC necessary (the remaining thickness= mechanical
support) => Make thin films !
 Advantages

 Thermal stability (substrate cavity = copper, Aluminum,… W)
 Cost
 Opens route to innovative materials
 Optimization of RBCS possible (e.g. by playing with m.f.p)

 Disadvantages
 Fabrication and surface preparation of substrate (at least) as difficult as for bulk Nb
 Steep Q0 decrease often observed by increase of RF field (sputtered niobium films, improved lately)
 Deposition of innovative materials is very difficult (large parameters space to be explored)
 Most of the known SC have been optimized for wire applications (low HC1, defects, pinning
 centers… ) => most of the literature recipes are not fitted for SRF application   

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 16
DEPOSITION TECHNIQUES:
 3 MAJORS FAMILIES

Physical deposition techniques (PVD, MS, DS…)
 line of sight techniques
 issues: getting uniform thickness/structure
 internal stress and adhesion
 limited for complex geometry
Thermal diffusion films
 limited compositions available
 non uniform composition issues (S shaped
 diffusion front, differential diffusion rate with substrate grain orientation)

Chemical techniques CVD, ALD
 conformational even in complex shape
 very quick for large surfaces
 issues: get the proper crystalline structure

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 17
SEARCH FOR BETTER STRUCTURE
 Generalized Structure Zone Diagram
Structure zone model (from A. Anders)
derived from Thornton’s diagram, 1974

 Energetic deposition
 (HPIMS, CED, VAD…)
 => Bulk like films

 Magnetron sputtering
 => A lot of defects
 Cu limits annealing
 temperature
 © Andre Anders, 2010
 11 recrystallization
 A. Anders, Thin Solid Films 518, 4087 (2010).

 Unfortunately, more “bulk-like” Nb Films gave disappointing
 RF results at the beginning . Not fully understood yet
 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb 19
Nb/ Cu:
 example of the issues
 when dealing with thin fims

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 21
SPUTTERED Nb FILMS

 The only Nb films deployed in accelerators were made by
 magnetron or diode sputtering (CERN).
 Reached relatively low surface fields => Eacc ~ 5 MV/m.
 Exponential slope in Rs and Q0

1.5 GHz Nb/Cu cavities, sputtered w/ Kr @ 1.7 K (Q0=295/Rs)

 Bulk Nb

 Nb films – sputtered

 1.5 GHz Nb/Cu

 LEP II
 350MHz Nb/Cu (4.2K)

 Possible origin of the slope
 Depinning of trapped flux
 Low HC1
 Early vortex penetration due to roughness
 CERN 2000 Current concentration due to porosities
 (generating local electrical field)

 2021/06/24 C.Z. Antoine -SRF 2021 22
 tutorial - Beyond Nb
EXAMPLE OF QUALITY ISSUES OF FILMS

 Magn. Sput. Nb 0° 45° 90°

 Line of sight issues 400 µm

 => porosities
 [G. Rozas]

 Sensitivity to Cu roughness (the smoother, the better)

Inverted pyramid crystalline growth Ra = 0.1 µm

 Internal stress
 Advantage: higher Tc (up to a certain Ra = 0.15 µm
 impurity concentration)
 Disadvantage: adhesion issues Ra = 0.025 µm
 (peeling)
 Ra = 0.01 µm
 High impurities content Ra = 0.015 µm
 Nb = getter material (nearly as good as
 Ti => high interstitial content)
 Carrier gas incorporation (Ar) [M. Ribeaudeau, PhD]

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial -| Beyond
 23 Nb
SUBSTRATE ISSUES

 Cu and Nb not miscible (especially in presence of O)
 Advantage: low interdiffusion
 Disadvantage: adhesion issues (peeling)
 Issues on Cu welding arreas
 Best results are not always [data from CERN + where
 AM Valente-Feliciano]
 expected:

 Standard Oxide-free
 films films

 RRRmax =28 RRRmax= 40

 Bulk like films did not perform better !
(but recent changes !!!)
 New approaches:
 Bulk machining
 Electroplating
 Laser surfacing …
 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 24
SEAMLESS ISOLDE CAVITY

 Local magnetometry on a sample deposited in
 the same condition
 First evidence of a bulk-like behavior for a thin
 film !!!!

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 25
DENSIFICATION /MITIGATION OF THE Q-SLOPE

2021/06/24 C.Z. Antoine -SRF 2021 tutorial -| Beyond
 26 Nb
OTHER SUPERCONDUCTORS

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 28
MOST COMMON SUPERCONDUCTORS:
 - A2 (e.g. NbTi, Transition metal alloys, BCC structures)
 - B1 (e.g. NbN, NbTiN, Transition metal carbide or nitride, NaCl
 structures)

 - A15 (e.g. Nb3Sn, Compounds, NaCl structures)
 - 2-D SC (Compounds, anisotropic)
 - MgB2
 - Cuprates, Pnictides
 - (others TaS2, organic…)

 - SPECIAL SRF: METAMATERIALS (Multilayers)

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 29
A2 SC ALLOYS: e.g. NbTi

 Ti

BCC pure metal and solid solution alloy

 Ti precipitates ( ~0,4 µm)
 NbTi widely used in coils NC Metal => RF dissipation !!!!

 Available alloys range around 45-55 % Ti
 Ti is not fully miscible inside Nb (Ti
 precipitates ꓱ at low T when [Ti] > 5 W%)
 => no RF !!!

 Ti  = Nb 
 Ti precipitates in a niobium matrix (with a few Nb
 replaced by substitutional Ti) => ~ same TC, same HC
 as Nb, but not same l => high k
 http://www.dierk-raabe.com/titanium-
 alloys/biomedical-titanium-alloys/

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 30
A1 SC COMPOUNDS: e.g. NbN

 https://link.springer.com/content/pdf/10.1007%2F978-1-4757-0037-4.pdf

BCC pure metal + smaller atoms (N, C) in interstitial
location => NaCl structure

 NbN cubic phase : TC ~17-18 K
 NbTiN stabilization of cubic (SC) phase
 NbN not too sensitive to local variation
 of composition !
 Solid solution => relatively easy
 fabrication (thermal diffusion, reactive sputtering…)
 Good model SC
 Widely used for JJ and SC electronics

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 31
A15 COMPOUNDS : HIGH TC

 compound TC (K) compound TC (K) compound TC (K) compound TC (K)
 Ti3Ir 4.6 V3Os 5.15 Nb3Os 0.94 Cr3Ru 3.43
 Ti3Pt 0.49 V3Rh 0.38 Nb3Rh 2.5 Cr3Os 4.03
 Ti3Sb 5.8 V3Ir 1.39 Nb3Ir 1.76 Cr3Rh 0.07
 V3Ni 0.57 Nb3Pt 10 Cr3Ir 0.17
 Zr3Au 0.92 V3Pd 0.08 Nb3Au 11
 Zr3Pb 0.76 V3Pb 3.7 Nb3Al 20.3 Mo3Re 15
 V3Au 3.2 Nb3Ga 18.9 Mo3Os 11.68
 V3Al 9.6 Nb3In 8 Mo3Ir 8.1
 V3Ga 15.4 Nb3Ge 23 Mo3Pt 4.56
[after Due-Hugues] V3In 13.9 Nb3Sn 18.3 Mo3Al 0.58
 V3Si 17.1 Nb3Bi 2.25 Mo3Ga 0.76
 V3Ge 7 Mo3Si 1.3
 V3Sn 4.3 Ta3Ge 8 Mo3Ge 1.4
 V3Sb 0.8 Ta3Sn 6.4
 Ta3Sb 0.72

 Extreme brittleness !!!
 cannot be plastically formed Phases with proper stoichiometry
 thin/thick film route only ! (A3B) not stable in normal condition
 (RT to Cryogenic temp)
 nm µm => Quenching necessary

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 32
A15 COMPOUNDS : NARROW DOMAIN OF SC

 A3B
 Nb3Ga Nb3Al Nb3Ge

B atoms occupy corners and center of BCC structure
A atoms form orthogonal chains bisecting the faces of the BCC unit cell.
Linear Chain Integrity is crucial for Tc (long-range order required)

 Narrow range of concentration for the SC phase:
 Highest TC area is even narrower
 Difficult to get uniform SC phase everywhere*
 Special issues at grain boundaries: “intrinsic” local deviation of
 stoichiometry*
 In Nb3Sn wires : GB exhibit degraded SC => weak links, pinning
 centers Compare with 

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 33
Nb3Sn

 Special interest for SRF
 since the 1980’s

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 34
Nb3Sn ON Nb (thermal way)

 Pioneer work: Wuppertal, Cornell

 [Lieppe, SRF 2013]

 @ 4,2 K: Q0 x 20 compare to Nb, @ 2K ~ the same
 Limited in Eacc , best results today ~20 MV/m
 Important developments: FNAL, JLAB, CERN, PKU….

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 35
HSHNb3Sn
 (~ 400 mT @ 0 K) EFFECTS OF LOCAL DEFECTS

 Hays. "Measuring the RF critical field of Pb,
 Nb, NbSn". in SRF 97. 1997.
 Nb HC1
 (170 to 200 mT)

 Nb3Sn
 Cornell, 1997
 pulsed
 Vs
 Nb

Recent results from
Cornell & FNAL CW
Posen,
2019 (23 MV/m)
2015 (17 MV/m))

 HC1Nb3Sn (~27mT)
 (50 mT pour le bulk)

 ~ 2K for Nb
 3,5 K for Nb3Sn

 => We have to reduce defect density (yes but which ones?)

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 36
https://indico.jlab.org/event/405/contributions/7865/attachments/6613/8979/Zeming%20Sun_Thin%20film%20workshop%20slide.pdf

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 37
OTHER APPROACHES

 Sputtered Nb3S films on copper
 Activities at Cern, STFC, Jlab, Cornell,,,
 RT deposited films : right composition but no A15
 structure
 Heating of substrate (CERN)
 And/or post annealing
 Other issues
 Cracks on the layer (due to differential dilatation coef)
 Diffusion of copper in the layer
 Carrier gas incorporation (Ar, Kr)
 Sn evaporation at higher temperature (> 1000°C)

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 41
OTHER APPROACHES…2

 Electrochemical deposition + diffusion through copper
 Proposed at FNAL
 Inspired from wire fabrication
 Not expensive !!!!

 Multilayer is heated => solid state diffusion
 Cu lowers the formation Tp° of A15 phase and suppresses the
 unwanted NbSn2 and Nb6Sn5 phases.

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 42
Nb3Sn: TENTATIVE DESCRIPTION OF THE LANDSCAPE

Method Sn vapor/ Sputtering Co-sputtering Sequential Bronze routes
 liquid ! target sputtering + (electrochemical
 deposition +
 diffusion annealing annealing)

substrate Nb only Nb or Cu Nb or Cu Nb or Cu Nb or Cu

Activities at Cornell Cornell STFC* IMP IMP
 FNAL Cern Peking U. Jlab Cornell
 JLAB Jlab Darmstadt ODU FNAL
 INFN ASC(FSU) Wisconsin U.* FNAL
 KEK

 * In Multilayer structures

 NB : work on V3Si also in progress

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 43
2-D SC (Compounds, anisotropic)
 - MgB2
 - Cuprates, Pnictides
 - Multilayers

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 44
MAGNESIUM DIBORIDE (MgB2)

 Graphite-type boron layers
 BCS type superconductor separated by hexagonal close-
 packed layers of magnesium
 TC ~40 K, two-gap nature
 Advantages:
 Very high TC (higher temp operation)
 Semimetal, cheap (fertilizer !)
 , l of high quality* MgB2 similar to Nb
 (~50 nm) (transparency of GB to current flow)
 Low n (lower RS)
 p=2.3meV, 2D, in-plane s-orbital
 Disadvantages: s=7.1meV 3D, out-of-plane p-orbitals
 Orientation issues (in polycrystalline materials !)
 RF dominated by lower gap  !
 Still better than Nb :
ΔNb= 1.5 meV < ΔπMgB2= 2.3 meV < ΔNb3Sn= 3.1 meV
MAGNESIUM DIBORIDE (MgB2)

 Phase diagram: at low Mg pressure only extremely low
 deposition temperatures can be used
 Optimal T for epitaxial growth ~ Tmelt/2
 For MgB2 Tmelt/2 = 540°C => PMg ~11 Torr
 Too high for UHV deposition techniques (PLD, MBE...)

 At PMg = 10-4-10-6 Torr, and Tsub ~ 400°C
 Compatible with MBE, and other deposition techniques
 MgB2 is stable, but no MgB2 formation:
 Mg atoms re-evaporate before reacting with B

 At PMg = 10-4-10-6 Torr, and lower T
 MgB2 is stable,
 If Tsub >250°C, free Mg is lost because the re-
 evaporation rate is higher than the impinging rate
 If Tsub < 250°C
 .Growth rate is very slow, Z.-K. Liu et al., APL 78(2001) 3678.
 Mg
 (kinetically limited by available Mg)

 MgB2
 evaporation pressure of Mg from
 MgB2 < decomposition curve of MgB2 M. Naito and K. Ueda,
 < Mg vapor pressure Kinetically limited Mg SUST 17 (2004) R1

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 46
MgB2 – HPCVD ON METAL SUBSTRATES
 HYBRID PHYSICAL CHEMICAL VAPOR DEPOSITION
 MgB2
[X. Xi- TempleUniversity]

 Polycrystalline MgB2 films deposited:
 On stainless steel, Nb, TiN, and other
 substrates.
 Flat samples and tubes (conformational) interface SiC

 Fitted for SRF apps:
 RRR>80
 low resistivity (
MGB2 – OTHER ROUTE S
 RF measurement @ MIT/Lincoln Lab
In-situ reactive evaporation @ 550˚C
 High quality flat samples
 Difficult to apply to complex geometries

 [T. Tajima, LANL]

 Superconducting
 Technologies Inc.

 Plasma electrolytic oxidation (PEO)
 MgB2 particles in suspension in an electrolyte
 MgB2 Islands deposited on the surface
 Issues : homogeneity, purity
 To be further explored

 [R. Valizadeh, STFC]

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 48
HPCVD MGB2 – RF MEASUREMENTS

11.4 GHz TE013 cavity @ SLAC 7.5 GHz sapphire-loaded TE011 cavity at JLab
 The MgB2 coatings were also characterized at 11.4
 GHz at SLAC using a cryogenic RF system. B.P.Xiao et al., SUST 25
 (2012) 095006.
 The samples showed a Q factor comparable to a
 Nb reference sample and higher than the Cu
 reference sample.
 The films showed a Tc of 37 K.

 Stripline resonator
 Scaled to 1,5 Ghz
 Lower surface resistance comparable to Nb film.

 [P. Welander, SLAC]

 Nb on
 sapphire
 MgB2 on
 LAO

 MgB2 on
 sapphire
 Oates et al., SUST
 23, 034011 (2010)
THE IMPORTANCE OF MEAN FREE PATH, MgB2
 EXAMPLE

Most developments of HTc are done in view of magnet applications : small  high l
by playing with ℓ
Small x makes the superconductors very sensitive to (usual) crystalline defects
(wanted for magnets, not for SRF)
Typical values found in literature for MgB2 : l ~100-150 nm  ~ 5-10 nm

 ℓ mean free path
if ℓ ↗ then: 1 1 1
 ↗ = +
  0 ℓ
 l↘
 k↘↘ l~50 nm
 ~
 some 10 nm ?

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 52
YBCO FAMILY… NOT FOR SRF !

 MonoXtal : Jc maximum for (a,b) planes and minimum when // c axis
 c (~0,03 nm) try to introduce preferential orientation (epitaxy): difficult to get
 on a cavity (but is applied to fabricate tapes for magnets)

 D-symmetry of the gap
 superconducting gap is also anisotropic
 = zero at 4 line nodes located at the diagonals of the Brillouin zone
 = 0 => power law for RS: RS α to T2-T3
 For the recall: gaps of conventional SC have s symmetry: http://mason.gmu.edu/~grobert1/2014syl641.htm

 isotropic and RS α e-/T (BCS resistance)

 Gap ()
 Fermi Fermi
 surface surface

 = 0
 3.6 nm
d-wave s-wave
 https://areeweb.polito.it/ricerca/
 Twin boundary in YBCO
 superconductivity/melt.htm

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 53
PNICTIDE FAMILY… MAYBE YES ?

Oxypnictide base: ReOMPn
 M = Fe, Co, Ni
 Pn = As or P
 Re = La, Nd, Sm, Pr
A lot of common with YBCO
 High TC (10-55 K up today)
 Layered structure
 Brittle material
 d-wave symmetry observed for some member of the family

But most compounds exhibit s-wave gaps…?
 Opening for SRF application ?

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 54
PNICTIDE FAMILY… MAYBE YES ?

A lot of common with YBCO NaFe1−xCoxAs (x = 0.0175)
 High TC (10-55 K up today) = ferromagnetic
 Layered structure NaFe1−xCoxAs (x=0.045)
 Brittle material = SC
 but
 Most compounds exhibit s-wave gaps
 Very sensitive to impurities content (either magnetic or not)

~ 5-6 meV
>>  Nb3Sn
(3,1meV)

 Ba0.6K0.4Fe2As2

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 55
MULTILAYERS

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 57
AFTER NIOBIUM : NANOCOMPOSITES
 MULTILAYERS

Structures proposed by A. Gurevich in 2006, SRF tailored

 Dielectric layer
 Small  vortex (short -> low dissipation)
 Quickly coalesce (w. RF)
 Blocks avalanche penetration
 => Multilayer concept for RF application
 Nanometric I/S/I/ layers deposited on Nb
 SC nanometric layers ( 100 nm) => HC1 =>
 Vortex enter at higher field HNb
 Nb surface screening => allows high magnetic Happlied //
 field inside the cavity => higher Eacc Cavity's interior
 →
 SC w. high TC than Nb (e.g. NbN): R SNbN  1 R SNb dsc
 10
 => Q0multi >> Q0Nb S-I-S

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 58
FIRST APPROACH: TRILAYERS

 Meissner state stable if:
 d
 Screening current @ both SC surface is < depairing current
 J(0) < Jd = Hs/λ and J(d) < Jd0 = Hs0/λ0
 If d is small, HSHS is high, but most of the field reach S0
 If d is thicker, HSHS is lower, but screening is more effective S I S0
 ꓱ an optimum thickness and a maximum screening field!!!

Maximum screening field Hm at the optimum S thickness dS = dm

 d/λ

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 59
SIS OPTIMIZATION: IMPORTANCE OF MODELS

 A. Gurevich, T. Kubo
 First approach: trilayers Exp decay
 Boundary conditions implemented (including effect of an
 insulating layer. finite thickness)) w. boundary
 HSH determined initially in London approx., further cond.
 improved w. quasiclassical theory (valid @T
TRILAYER OPTIMIZATION (…)

 A. Gurevich, T. Kubo
 Go for realistic condition
 Layers present defects, non-negligible surface roughness, non-uniform thickness.
 → HSHS suppressed due to of the local screening current enhancement.
 Introducing material suppression factor η =f(defect size and aspect ratio, ξS )
 η ~ 0.85 for typical electropolished Nb surface)
 HSHSIS and optimal S layer thickness dmS can be determined w. surface topographical data

Ideal Nb substrate Nb with defects*,
with BC1=170 mT with BC1=50 mT

 * e.g. morphologic
 defects that allow earlier
 vortex penetration

 See exp proof later on

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 62
WHAT IS THE LIMIT (HP/HC1/HSH) ?

 Real world cavities behavior is dominated by a few number of defects
 It is very important to measure the penetration field of samples in realistic conditions

 H H
 H//
HSH HSH H//

HC1 HC1

 H H
 T2C T2 T2C T2

 Local magnetometry
 ~ Same geometry as cavities
 No shape/edge effect (vs DC/
 Squid magnetometry)
 No demagnetization effect
 Measures actual penetration field
 wherever it is HP/HC1/HSH

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 63
EXPERIMENTAL DETAILS

 H
 maximum field without Vx (max Eacc)
 HC2

 maximum pinning
 Vx ~ liquid

 HMelting

 Vx lattice
 HC1

 Ia/Ha

 Meissner state

 TC T
Low frequency ≡ DC :
 flux line moving in AC field
 0 < Ha< HC1 => R=O, Meissner state
 Vx pinned in
 HC1< Ha< HM =>Vx are trapped, R=O, Campbell regime the SC
 HM< Ha< HC2 =>Vx are moving liquid like, R≠O, Flux flow regime
 Third harmonic signal arise from flux line tension (affects the e-
 inside the Cu coil),
 It does not depend on dissipation inside Nb, BUT depends on #
 of Vx trapped there (and length).

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 64
Nb – INSULATOR – NbN MODEL

 NbN coating by Magnetron Sputtering
 NbN single layers series
 NbN SL / “thick” Nb layer
 Magnetron sputtered
 MgO as dielectric layer
 Far from perfect…

 MgO (nm) NbN (nm)
 Nb (nm) TC (K)
 Calc(actual) Calc(actual)
 250† 14 0 8.9
 250† 14 25 15.5
 500 10 (10.3) 50 (65) 15*
 500 10 (8.4) 75 (72) 14.1*
 500 10 (9.8) 100 (94) 14*
 500 10 125 14.3*
 500 10 (6.7) 150 (132) 15.9*
 500 10 (10.4) 200 (164) 15*
 † Not same batch, deposited on the same conditions, but substrate = sapphire
 *As determined with magnetometry, see below.

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 65
SPUTTERED (DEFECTIVE) MATERIALS…

 Typical defects…
 NbN 100nm
 Low HC1 NbN Thickness ~ 105nm
 Thickness ≠ uniform
 …

 250 nm Nb on saphire
 DC Squid magnetometry @ 4.5
 K Nb Thickness ~ 425nm

 Rippled interfaces

 NbN 50nm
 ~ 18 mT
As expected for MS
 samples

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 66
COMPARAISON WITH THEORY

 Theoritical predictions from T. Kubo (KEK)
 x
 x
 x
 x
Ideal Nb substrate x
 Nb with defects*,
 x
with BC1=170 mT with BC1=50 mT
 x
 x
 x
 * e.g. morphologic
 x defects that allow earlier
 x
 x
 vortex penetration See
 SST paper cited earlier

 T (K)
 02 4 6 8 10 12 14 16

 80

 70 Nb
 NbN 25nm
 The enhancement of the field
 60 NbN 50nm
 NbN 75nm
 penetration increases with thickness of
 Hc1 (mT)

 50 NbN 100nm
 NbN 125nm NbN
 40
 NbN 150nm
 30 NbN 200nm It reaches a saturation at thicknesses >
 20 100 nm
 10

 0
 0 40 80 120 160 200 240 280

 T2 (K2)

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 68
COMPARAISON WITH THEORY

 Theoritical predictions from T. Kubo (KEK)
 x
 x
 x
 x
Ideal Nb substrate x
 Nb with defects*,
 x
with BC1=170 mT with BC1=50 mT
 x
 x
 x
 * e.g. morphologic
 x defects that allow earlier
 x
 x
 vortex penetration See
 SST paper cited earlier

 According to
 2nd transition of V3

 150
 140
 130
@ 4.5 K Nb
~ + 110 mT?
 120
 110
 NbN 25nm The enhancement of the field
~25-30 MV/m NbN 50nm
 100
 NbN 75nm penetration increases with thickness of
ILC shape 90
 Hc1 (mT)

 NbN 100nm
 80
 NbN 125nm NbN
 70
 NbN 150nm
 60
 NbN 200nm
 It reaches a saturation at thicknesses >
 50
 40
 100 nm
 30
 20
 10
 0
 0 50 100 150 200 250 300
 2 2
 T (K )
 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 69
CLOSEUP OF 3rd HARMONIC SIGNAL

 V3(V)
 For a given Happl, we observe 3 0,00012
 Phase(deg)

 ≠ transition temperatures 0,00010
 115

 Phase signal 110
 0,00008 1

 Phase(deg)
 V3(V) 105

 V3(V)
 Phase(deg) 0,00006
 0,00012 150
 100
 0,00004
 0,00010
 100 95
 0,00002

 0,00008
 90
 0,00000

 Phase(deg)
 50
V3(V)

 0,00006 6 8

 0
 T~T Temperature(K)
 0,00004
 T >>T 3 V3(V)
 Phase(deg)
 150
 0,00002 0,00004
 -50

 0,00000 100

 -100
 4 6 8 10 12 14 16 18

 Phase(deg)
 50
 Temperature(K)

 V3(V)
 0
 2
 T~T: within noise level Voltage signal -50
 T >>T: dramatic transition
 -100
 6 8
 Temperature(K)

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 70
ROLE OF THE DIELECTRIC LAYER !

 Why do we have two transitions ?
 H app

 Defect  Thin SC layer NbN
  Insulator MgO
  Thick SC layer Nb

 Field lines

 First transition H // surface => surface
 barrier†
 A defect locally weakens
 the surface barrier
 1st transition, vortex
 blocked by the insulator
 ~100 nm => low dissipation.
 2nd transition, propagation
 of vortex avalanches (~100
 µm) => high dissipation.
 Dielectric layer = efficient
 protection !!!

 Second transition

2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | PAGE 71
SIS : IRREVERSIBILITY => NO VORTEX
 PINNING
1st TEST RF @ 3,88 GHZ (4 25nm NbnN LAYERS)
 Accelerator cavities'
 Rs = f(T) operating range

 8

 7 Rs(ML4)

 6 Rs(Nb) (scaling
 in w2)
 5

 Rs (µOhm)
 4
 3.88 GHz =>
 3 RBCS~ 9 x R (1.3 GHz)

 2

 1

 0
 0 1 2 3 4 5
 T(K)

Comparison is done with a high performance 1.3 GHz Nb cavity (scaling in 2)
Indium gasket presents some defects measured with thermometric map => extra RF losses
Residual resistance comes from NbN + bulk Nb substrate + indium gasket. Further investigations
needed.

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 77
[A.M.Valente-Feliciano]

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 78
ML WITHOUT DIELECTRIC INTERLAYER

 µ-SR

[Junginger, SRF 2017]

 The SS boundary provides an additional barrier to prevent penetration of
 vortices. It would not be as robust as the I layer of the SIS structure, but it also
 contributes to pushing up the onset of vortex penetration.
 [Kubo, SST]

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 79
CONCLUSIONS AND PERSPECTIVES

Superconducting cavities are dominated by their surface quality (Niobium
AND other SC !)

Niobium is close to its ultimate limits, but can be surface tailored (doping)

HSH difficult to reach in real “accelerator cavities” (low T, large scale cavity
fabrication, surface defects,…)

ML structures seem to be a promising way to go toward realistic complex
materials (+ Nb cavity upgrade)

Renewed activity on bulk-like Nb films (cost issues) and high HSH SC e.g.
Nb3Sn or NbN (higher performances)

Look for higher Q0, not only Eacc !

WE ARE ON THE EVE OF A TECHNOLOGICAL REVOLUTION FOR SRF
CAVITIES !

 2021/06/24 C.Z. Antoine -SRF 2021 tutorial - Beyond Nb | 80
You can also read