Point contact tunneling spectroscopy and Atomic layer deposition for superconducting rf cavities Thomas Prolier, Mike Pellin, Jim Norem, Jeff Elam. Collaboration:

Point contact tunneling spectroscopy and Atomic layer deposition for superconducting rf cavities

Thomas Prolier, Mike Pellin, Jim Norem, Jeff Elam.

Collaboration:-Jlab: P. Kneisel, G. Ciovati-Fermilab: L.Cooley, G. Wu, C. Cooper

- IIT: J. Zasadzinki

Superconducting Radio-Frequency (SRF)

Department of Energy – Office of Science• DOE-OS is in the particle accelerator business (ILC ($19B), RIA($0.4 B),

NSLS-2($0.5B) , SNS($1.8B), APS, APS-ERL, etc.)

• Orbach to HEPAP 2/22/07 “DOE is committed to continuing a vigorous R&D program of accelerator technology SCRF is a core capability having broad applicability, both to the ILC and to other future accelerator-based facilities as well. Out FY2008 request for ILC R&D and SCRF technology confirms this commitment

1 m15 km • 30 km of ultra pure Nb bellows

• 2 K• very high electrical and magnetic fields

3

Outline

Performance limitations: Point contact spectroscopy: a probe of the surface superconductivity.

Atomic Layer Deposition: synthesizing new materials and application to RF cavities.

Niobium surfaces are complex, important, and currently poorly controlled at the nm level

4

45

nm

RF

dep

thInclusions,

Hydride precipitatesSurface oxide Nb2O5 5-10 nmMagnetic!

Interface: sub oxides NbO, NbO2

often not crystalline (niobium-oxygen

“slush”)

Interstitials dissolved in

niobium (mainly O, some C, N, H)

Grain boundaries

Residue from chemical

processing

Clean niobiume- flow only in the top 45 nm

Probe the surface superconductivity

Point Contact Tunneling (PCT) Spectroscopy – a Surface Probe of Nb superconductivity

5

6 Tesla magnet 1.6-300 K

2-

Ideal BCS superconductor

• Measure of the superconducting gap Δ

• The ZBC value -> Number of normal electron

Normal electrons in gap => dissipation and lower Q

PCT: 1- insight into Mild Baking Procedure Improvement: Small changes in O stoichiometry -> Magnetic Oxide reduction

Unbaked Niobium

T.Proslier, J.Zasadzinski, M.Pellin et al. APL 92, 212505 (2008)

Cavity-grade niobium single crystal (110)-electropolished

ILC-Single crystal cavities P.Kneisel

Qo improvement - 1.6

Average ZBC ratio = 1.6

2-

Ideal BCS, T=1.7KBaked Niobium 120C-24h

Cavities have dissipative losses due to Cooper pair breaking!-> Nb2O5--, NbO2--

7

PCT: 2- Hot and cold spots in SRF cavity (from J-lab)

Anomalous spectrumOnly on hot spots

“Normal” spectrum

Hot spots: show dissipative behaviorHigher ZBC and anomalous spec.lower gap values (1.3<∆<1.55)

Cold spots: “normal” dissipationLow ZBC values Normal gap values (1.5<∆<1.55)

Origin of peculiar spectrum and dissipation?

Correlates with cavities results! (once again)

8

PCT: 2 Hot and cold spots in SRF cavity, Origin

Temp. dep: peak at 0 mV bias increases Killing superconductivity by applying a mag. Field

High bias peak:LOSSES!!

fits with Appelbaum theory -> Magnetic impurities in the oxides !!J>0 -> antiferromagnetic coupling

-First time measured on Nb oxides

-Same behavior observed on unbaked Nb coupons !

IIT: EPR revealed magnetic moments tooFSU: theory, dissipation and magnetism

To be published

How to make better cavities

Add a better dielectric (thanks to Intel) and bake

O2

O

Atomic layer deposition (ALD)

Al2O3(2nm)NbOx

Nb

Heating -> Reduction + diffusion of the oxides

Baking, but now protected form O (Al2O3)

- ∆ (1.55meV = Nb).

- Γ (dissipation)

- 500 oC bake should significantly reduce dissipation

Th.Proslier, J. Zasadzinski, M.Pellin et al. APL 93, 120958 (2008)

11

Cavities used for ALD

Jlab (P. Kneisel) provided four different niobium cavities to ANL for atomic layer deposition: Cavity 1:

Material: RRR > 300 poly-crystalline Nb from Tokyo-DenkaiShape/frequency: Earlier KEK shape, 1300 MHzBaseline: electropolished, in-situ baked

Cavity 2 :Material: RRR > 300 large grain Nb from Tokyo-DenkaiShape/frequency: TESLA/ILC shape, 1300 MHzBaseline: BCP, in – situ baked

Cavity 3:Material: RRR > 300 poly-crystalline Nb from FansteelShape/Frequency: CEBAF shape, 1497 MHzBaseline: BCP only

Cavity 4: Shape/Frequency: CEBAF Single cell cavityBaseline: BCP + 600oC UHV bake.

J Lab Cavity 1:After ALD Synthesis (10nm Al2O3 + 3nm Nb2O5), 250oC

Only last point shows detectable field emission. 2nd test after 2nd high pressure rinse. (1st test showed field emission

consistent with particulate contamination)

108

109

1010

1011

Quench @Eacc = 32.9 MV/m

Q0

Eacc [MV/m]0 5 15 20 25 30 3510

Atomic Layer Deposition (10 nm Al2O3 + 3 nm Nb2O5)

Previous Best Cavity Performance (Initial Electro-Polish and Bake)

Cavity As Received For Coating

Single Cell Cavity Test (J Lab 6/27/08)Argonne Cavity Coating Procedure

J lab Cavity 2: Large grain,10 nm Al2O3 + 3 nm Nb2O5 (250oC)

13

Second coating: 5 nm Al2O3 + 15 nm Nb2O5

First coating: 10 nm Al2O3 + 3 nm Nb2O5

Baseline Test 2 Test 1

J Lab Cavity 3: Annealing 450C/20hrs + Coating: 5nm Al2O3+15 nm Nb2O5

14

High Temp. baking: T maps and Rs(T)

T-map at the highest field measured during the test after 120 °C, 23 h UHV bake.

T-map at the highest field measured during the test after 450 °C, 20 h heat treatment

10

100

1000

0.22 0.27 0.32 0.37 0.42 0.47

1/T [1/K]

Rs [

nW]

Add. HPR

120C/23h UHV bake

450C/20h HT

Treatment D/kTc ℓ (nm) Rres (nW)

Add. HPR 1.866 ± 0.018 19 ± 44 16.0 ± 0.8

120 °C/23 h bake 1.879 ± 0.005 18 ± 55 16.3 ± 0.5

450 °C/20 h HT 1.911 ± 0.026 58 ± 17 93.8 ± 0.2

Ohmic losses

But HT baking: Improve the super. properties

Preliminary Conclusion and High temp annealing

The ALD process is compatible with SRF cavity processing

Promising if one thinks about multi-layer coatings ( A. Gurevich).

development of the process is necessary

The appearance of multipacting in cavity 1 and 2 is concerning, but can

be overcome by additional coating.

Baking doesn’t improve cavity performance: cracks can appear due to

strong Nb oxide reduction -> path for oxygen injection -> Ohmic losses

need a in-situ baking + ALD coating set up.

16

17

ALD Can Produce Layered SRF Structures with significantly higher Hc1

than Nb

Build “nanolaminates” of superconducting materials

~ 10- 100 nm layer thicknesses with 10 nm Alumina Between.

Hc1 Enhancement Scales with:

~Tc,laminate/Tc,base

For NbN laminate layers -> ~1.5 HC1 enhancement

50 MV/m -> 75 MV/m

Nb, Pb

Insulating layers

Higher-TcSC: NbN, Nb3Sn, etc

SEM XPS XRR-RBS

SQUIDRRR

New materials by Atomic Layer Deposition:

NbF5 + Si2H6 = NbSi + reaction product, copy on : WF6 + Si2H6 = W + RP

On Si (100): NbSi superconductor 3.1KOn Quartz: Nb3Si5On MgO: NbSi2Elastic stress in the film-mismatch ?

Nb3Si superconductor at 18KVary substrate and growth conditions Post-annealing studies

Model for A15 compounds: Nb3Ge (20K): NbF5 + Ge2H6 = NbGe + reaction products

Etc… To be published

Fast growth rate:2.5 Å/Cy

Grows only WNot on oxides

New material by Atomic Layer deposition

New precursor NbF5 for NbN, Nb2O5 grows much faster!

19

Zinc pulse growth for NbN and TiN:

NbCl5 + NH3 + Zn = NbN + ZnCl2 + HCl

TiN films: resistivity ρ=50 µΩ.cm for 10 nm films! (350 without Zn)

NbN films: resistivity ρ=200 µΩ.cm (450 without Zn -> Tc= 5.5 K), same ρ for sputtered film with Tc=16K!

To be measured:

-Superconducting properties with Zn pulse -> Multilayers-Vary the substrate (Sapphire’s) to match lattice parameter (epitaxial growth?)-Post annealing in controlled atmosphere

No studies of superconductivity by ALD and interactions substrate-films Phase space of parameter to study is large

Magnetic impurities as a possible explanation for RF dissipation: Mild baking effect Hot spots Origin = Oxides, vacancies?

High temperature baking works on samples but not yet on cavities ALD a tool for building new materials

Compatible with RF cavities NbN, NbSi, TiN etc… Plasma ALD

Summary

New task force:- Postdocs and students -> Accelerate the process

outlook

21

(1)Nb deposition on Nba) New Cavity Designs b) Enable Continuity of Superconducting Surface (fewer perfect welds)

(2)Other layered structuresa) Reward : Performance far beyond NbN b) Risk: New ALD Synthesis Methods Need to be developed with

semiconductor impurity levels.

(3) Nb deposition on alumina coated Cua) Reward : Significant Cost Reductions for Materials, Fabrication, and

Coolingb) Risk: Dissimilar materials require stress management (Cu is bad,

alumina is better)

(4) Field emission for warm and cold cavitiesParticulate tolerant?

ALD Reaction Scheme

• ALD involves the use of a pair of reagents.• each reacts with the surface completely• each will not react with itself

• This setup eliminates line of site requirments

• Application of this AB Scheme• Reforms the surface• Adds precisely 1 monolayer

• Pulsed Valves allow atomic layer precision in growth

• Viscous flow (~1 torr) allows rapid growth• ~1 mm / 1-4 hours

0

500

1000

1500

2000

2500

3000

3500

4000

0 500 1000 1500 2000 2500 3000AB Cycles

Th

ick

nes

s (Å

)

Ellipsometry Atomic Force Microscopy

• Film growth is linear with AB Cycles• RMS Roughness = 4 Å (3000 Cycles)• ALD Films Flat, Pinhole free

Mixed Oxide Deposition: Layer by Layer

Mixed Layer Growth• Layer by Layer• note “steps”• atomic layer sequence

“digitally” controlled

• Films Have Tunable Resistivity, Refractive Index, Surface Roughness, etc.

[(CH3)3Al // H2O]

100 nm

ZnO

ZnO

Al2O3

Al2O3

[(CH3CH2)2Zn// H2O]

• Mixed Layers w/ atomic precision• Low Temperature Growth• Transparent• Uniform• Even particles in pores can be

coated.

ALD: The Only Viable Method for SRF Surface Control!

25

Niobium is from a surface scientists point of view a difficult material to deal with.– Extremely reactive.– Native Oxide is complex and passivates poorly

Semiconductor Industry – a clue– Silicon is reactive but oxide is simple and

passivates well (but has a low dielectric constant)– Gate dielectric oxides are now being used on Si

metal (and being produced by ALD

20 m2 / batch) Grow a dielectric oxide with superior properties to the

Niobium Oxides– Simple - non-interactive with the sc layer– Passivating (stable surface, protective of the Nb

metal underneath)– Parallel Growth Method Entirely adaptable to SRF

Si

HfO2

Epoxy

ALD Thin Film Materials

A Solution? Atomic Layer Deposition -> non-dissipative dielectric layer

27Mike Pellin

1. Use Atomic Layer Deposition (ALD) to synthesize a dielectric diffusion barrier on the Nb surface

2. Bake cavity to “dissolve” the O associated with the Nb layer into the bulk

Nb NbO

Nb2O5--

NbO2

Al2O3

Nb

Al2O3

ALD coated + Baking > 450°C

Mild baked before ALD

Test

Cavity 4: to be coated by SiN + NbSi (below 200oC)

28SRF 2009

Understanding Cavity Eacc and Q

29

Q-slope problem

Rs = RBCS + Rres

RBCS = C-4-2l exp(--/kT)

Experimental Goals:• Measure - at the surface• Tunneling Spectroscopy is ideal

P.Kneisel et al. 12 th SRF workshop Cornell 2005B.Visentin SRF workshop 2003G.Ciovati, P.Kneisel, A.Gurevich PRST, 10 2007

C.Antoine SRF workshop 2004

H(-)

Nb NbOx NbO Nb2O5-δNbO2

-

B(r)

Surface

Q-slope disappears, Q0 increased

SRF Impedance is a surface effect (-~45 nm, Nb) depends on the energy gap at the surface altered by proximity effects, magnetic scattering.

SRF 2009

0 5 10 15 20 25 30 351.0E+08

1.0E+09

1.0E+10

1.0E+11

1.3 GHz Cavity, KEK Shape

450C for 20 hrs ALD + HPR

Eacc [MV/m]

Q0

Quench or discharge?

Quench @ Eacc = 32.9 MV/m

J-lab cavity 1 + HT annealing (450oC for 20 hrs).

30SRF 2009

J Lab Cavity 3: Small grain 2 steps Coating, first: 15 nm Al2O3 at 90oC

31SRF 2009