a review of niobium (on copper) sputtering technology s. calatroni

A Review of Niobium (on Copper) Sputtering Technology

S. Calatroni

9 October 2006 Sergio Calatroni - CERN - Sputtering Technology

2

• Why films: advantages / disadvantages• Residual resistance: topics for further R&D :

– Effect of roughness– Effect of film structure– Effect of hydrogen– Effect of surface oxidation

• Conclusions and perspectives

Introduction


3

Why films

• Advantages (primary objectives)– Thermal stability– Cost– Innovative materials

• Advantages (learned from experience)– Optimisation of RBCS at 4.2 K (sputtered niobium films)

– Reduced sensitivity to earth magnetic field

• Disadvantages (known from the beginning)– Fabrication and surface preparation (at least) as difficult as for bulk

• Disadvantages (learned from experience)– Deposition of innovative materials is very difficult

– Steep Rres increase with RF field (sputtered niobium films)


4

The surface resistance can be written in the form:

RRs s (H(HRFRF, H, Hextext, T) = R, T) = RBCS BCS (H(HRFRF, T) + R, T) + Rfl fl (H(HRFRF, H, Hextext, T) + , T) + RRres res (H(HRFRF))

The dependence of RBCS (0,T) on has been verified by changing the sputter gas

RBCS (HRF, T) has an intrinsic dependence of HRF

Rfl (HRF, Hext, T) has a dependence on similar to RBCS(0,T)

The surface resistance


5

RBCS at 4.2 K Nb bulk: ~900

nNb films: ~400

n

RBCS at 1.7 K Nb bulk: ~2.5

nNb films: ~1.5

n

1 10

350

400

450

500

550

600

650

700

750

800

850

900

2 5 20

RB

CS

(4.2

K)

[n

]

1+0/2

Theoretical and experimental BCS resistance at zero RF field

Nb bulk


6

Fluxon-induced losses at 1.7 K are characterized as RRflfl = (R = (Rflfl00 + R + Rflfl

11 H HRFRF) H) Hextext The minimum values are obtained using krypton as sputter gas:

RRflfl00 = 3n = 3n/G R/G Rflfl

11 = 0.4 n = 0.4 n/G/mT/G/mT

Triangles: bulk NbSquares: coatings on oxide-free copperCircles: coatings on oxidized copper

1 10

10

100

Rfl

0 [n

/G]

(a)

1 10

1

10

Rfl

1 [n

/G/m

T]

(b)

Fluxon-induced losses I


7

Nb/Cu films – categories

• There are two categories of films

– Films which are intrinsically films• Thin, small grains, microstrained, under stress• Problems: defects & microstructure, (impurities), surface state• Examples: magnetron sputtered films on oxidised copper

– The trend among workers is to aim for films which are bulk-like• Thick, large grains• Problems: hydrogen, surface quality• Examples: high-energy deposition techniques, annealed films,

(Nb Cu-clad)

Of course a film from one family may as well present all the problems typical of the other family…


8

Research lines around the world

• Effect of substrate: roughness, thermal impedance– Non uniform coating, H enhancement (demagnetization), increased

granularity. Thermal feedback Optimisation of substrate preparation (electropolishing), study of

angle-of-incidence effects, conformal coatings. Measurements of K

• Film structure – defects– Hc1 reduction, hysteretic losses

Towards a bulk-like film: bias sputter deposition, high-energy deposition techniques, high-temperature annealing of films

• Effect of hydrogen– Hydrides formation Measurements of H2 contents, outgassings

• Oxidation– Localized states, corrosion of grain boundaries Al2O3 cap layers


9

Electropolishing – Polarization curve

0

500

1000

1500

2000

2500

0

0.05

0.1

0.15

0.2

0.25

0 1 2 3 4 5 6

Cur

rent

den

sity

[A

/m2 ] P

olishing rate [µm

/s]

Cathode potential [-V]

Average roughness R

a [µm

]

Production of Cu(OH)2 on the surface!

Production of O2 bubbles!

Electrical resistance seen by the polishing current:

Diffusion layer ~ 0.1 Bath volume ~ 0.1

(Nb EP bath: 10 times less)Polishing

Standard CuStandard CuElectropolishing:Electropolishing:55% vol. H3PO4

45% vol. butanol


10

Numerical modelling of the cathode by simulation of the entire EP process with the Elsy 2D/3D computer code(www.elsyca.be)

Electropolishing – Cathode design

Cathode activeregion

Current density is uniform over all the cell surface

http://www.elsyca.be/


11

Measurement of pinholes

Irregularities on the substrate surface shadowing effect film inhomogeneities

2pSQ

QQ

C

film

S

p1 p2

Substrate disk

Machining and

cleaning

Film deposition

Substrate removal

He leak rate experiment ‹inc› fraction of leaky

film surface

- equator 9° 4.4 ppm- (~iris 50° 25 ppm)

- equator 9° 0.1 ppm

CP

EP


12

Defects in Cu substrate

Electropolished copper surface

• average roughness: 0.02 µm

• A few defects still appearing

Cross section of a copper cavity

• Defects are present inside !

• Not an artifact of the preparation

Thanks to: G. Arnau-Izquierdo


13

Variation of properties with incidence angle

Std.Dev. of grey levels of SEM images (CERN 1999)

XRD spectraAFM roughness

From: V. Palmieri, D. Tonini – INFN-LNL


14

Incidence angle and residual resistance in low- cavities

Correlation between the incidence angle of the film and the residual resistance, measured on 352 MHz Nb/Cu cavities

0

10

20

30

40

50

60

70

0

50

100

150

200

250

300

35 40 45 50 55 60 65 70 75

Rs

0(4.2K)

Rs

0(2.5K)

Rs

1(4.2K)

Rs

1(2.5K)

Rs0 [

n]

Rs 1 [n

.m/M

V]

Incidence angle [º]It seems there is a “threshold” effect


15

Angle of incidence – post magnetron – conformal coating

Niobium cathode

Cavity

B

Magnetic field lines follow the cavity shape

V. Palmieri, R. Preciso, V.L. Ruzinov, S. Yu. Stark, “A DC Post-magnetron configuration for niobium sputtering into 1.5 GHz Copper monocells”, Presented at the 7th Workshop on RF Superconductivity


16

Angle of incidence in spherical cavity

20 40 60 80

10

20

30

40

50

60

70

80

PACO cavity INFN Genova

10 20 30 40 50

10

20

30

40

50

60

70

80

Standard 1.5 GHz cavity

Ave

rag

e in

cide

nce

angl

e of

th

eN

b co

atin

g [d

egre

es]

Length along the cavity axis [mm]

Length along the cavity axis [mm]

Ave

rag

e in

cide

nce

angl

e of

th

eN

b co

atin

g [d

egre

es]

The cathode is not point-like

The incidence angleis always >0


17

Surface resistance of spherical cavityR

s [n

Ohm

]

Rs

[Ohm

]

1.7 K

Standard 1.5 GHz Nb/Cu cavity

Reducing the angle of incidence does not change Rs(E). However, the angle is always greater than zero, and whether this is creating any effect is only matter of speculation – for the time being


18

Thermal impedance film-substrate

Cu (RRR=100) EP 4300 ± 200 Wm-2K-1

Cu EP + 1.5 µm Nb 4100 ± 200 Wm-2K-1

Nb (RRR=180) EP 1200 ± 200 Wm-2K-1

Nb EP + 1.5 µm Nb 1000 ± 200 Wm-2K-1

The overall thermal impedance has been measured for pure Nb and Cu, and for Nb/Cu and Nb/Nb films, on 2-mm thick disks.

Nb (RRR=670)(extrap.) 2500 ± 200 Wm-2K-1

(Still lower than Nb/Cu, but Nb cavities performs better at high field !!)

The thermal impedanceof the film (if existing) has no effect

on Rres at high RF field

Thanks to: G. Vandoni, J-M Rieubland, L. Dufay

T2

T1

H2

Conflat flange

Conflat ring

Sample


19

Nb/Nb

1

10

100

0 5 10 15 20 25 30 35 40

Q0 [

109 ]

Eacc

[MV/m]

1p5.1 Nb bulk cavity

1p5.2 Nb/Nb (quench limited)

Thanks to: V. Palmieri, D. Reschke, R. Losito


20

Al substrate (better thermal conductivity)

10

100

1000

104

0 1 2 3 4 5

Rs

[nO

hm]

Eacc

[MV/m]

4.2 K

1.7 K

Al 99.999% purity: 10X thermal conductivity of Cu at 4.2 KSpinning + chemical polishing + coating

Thanks to: V. Palmieri, G. Lanza


21

Film structure – FIB cross sections

Standard films Oxide-free films

0.5 µm0.5 µm 0.5 µm0.5 µm

Courtesy: P. Jacob - EMPA

Grain size with Focussed Ion Beam micrographs


22

Grain size ~ 100 nm

Fibre textureDiffraction

pattern:powder diagram

Grain size ~ 1-5 µmHeteroepitaxy

Diffraction pattern:zone axis [110]

500nm

500nm

TEM views I – plan view


23

(110) fibre texture

substrate plane

HeteroepitaxyNb (110) //Cu(010) Nb (110) //Cu(111)Nb (100) //Cu(110)

500nm

500nm

TEM views II – cross section


24

200K

100K

Crystallographic defects

TEM views III - defects

~100 nm


25

High-energy deposition techniques

• Crystalline defects, grains connectivity and grain size may be improved with an higher substrate temperature which provides higher surface mobility (important parameter is Tsubstrate/Tmelting_of_film)

• However the Cu substrate does not allow heating The missing energy may be supplied by ion bombardment

– In bias sputter deposition a third electron accelerates the noble gas ions, removing the most loosely bound atoms from the coating, while providing additional energy for higher surface mobility

“Structure Zone Model”

– Other techniques allow working without a noble gas, by ionising and accelerating directly the Nb that is going to make up the coating

– These techniques allow also to obtain“conformal” coatings that followthe surface profile better filling voids.


26

Nb/Cu bias deposition – First SEM images at CERN

No bias Bias -60 V

Bias -80 V Bias -100 V

5 µm


27

Biased sputtering at LNL


28

Evaporation + ECR (JLAB)

• Niobium is evaporated by e-beam, then the Nb vapours are ionized by an ECR process. The Nb ions can be accelerated to the substrate by an appropriate bias. Energies in excess of 100 eV can be obtained.

From: A.-M. Valente, G. Wu

Generation of plasma inside the cavity 3 essential components:Neutral Nb vaporRF power (@ 2.45GHz)Static B ERF with ECR condition

Why ECR?No working gas High vacuum means reduced impuritiesControllable deposition energy,90-degree deposition flux (Possible to help control the crystal

structure)Excellent bonding No macro particlesFaster rate (Conditional)


29

Evaporation + ECR: results on samples

• Obvious advantage: no noble gas for plasma creation

• Sample tests: good RRR and Tc, 100-nm grain size, lower defect density and smooth surfaces

60eV

90eV 4000X4000 µm2

3-D Profilometer ImagesTEM Images


30

Application to cavities (JLAB)


31

Plasma Arc (INFN)

• In the plasma arc an electric discharge is established directly onto the Nb target, producing a plasma plume from which ions are extracted and guided onto the substrate by a bias and/or magnetic guidance

• Magnetic filtering (and/or arc pulsing) is also necessary to remove droplets

• A trigger for the arc is necessary: either a third electrode, or a laser

• Arc spot moves on the Nb cathode

at about 10 m/s• Arc current is 100-200 A• Cathode voltage is ~ 35 V• Ion current is 100-500 mA on the

sample-holder (2-10 mA/cm2)• Base vacuum ~ 10-10 mbar• Main gas during discharge is

Hydrogen (~ 10-7 mbar)• Voltage bias on samples 20-100 V From: R. Russo, A. Cianchi, S. Tazzari


32

Plasma Arc – Need for filtering

Arc source

Nb droplets

5 µm

10 µm

Magnetic filter

Nb droplets


33

Planar arc – cavity deposition set-up

New ideas are put forward for using a planar arc for cavity coating

5 µm

10 µm


34

Planar arc – RF measurements on samples!

Cu samples with Nb ARC-coating. Used as a baseplate of 6 GHz cavity operating in the TE011 mode. At low field, the surface resistance is in the range 3÷6 µOhm, as compared to the BCS Rs of 0.22 µOhm at 2.2 K and small mean free path. The Q remained constant up to a field of 300 Oe.

A baseline of 2.2 µOhm is measured with this cavity with a solid Nb plate


35

Liner arc for cavity deposition (Soltan Institute)


36

HPPMS: advantages

From W. Sproul, AEI


37

Principle: high power pulses of short duration– Peak value typically 100 times greater than conventional

magnetron sputtering– Pulse width of 100 - 150 µsec, repetition rate ~50 Hz– Peak power densities of 1-3 kW/cm2

– Discharge voltages of 500-1000 V– Peak current densities ~ A/cm2

Consequences– High degree of target material ionization– High secondary electron current– Promotes ionization of sputtered species– Can approach 100%, vs. ~1% for conventional sputtering– An applied bias allows attracting ions to the substrate

Operating principle


38

Ion density

From W. Sproul, AEI


39

HPPMS coatings in trenches (valid for any ion coating technique)

J. Alami et Al, J.Vac.Sci.Technol. A23(2005)278


40

First CERN results

Pulse voltage (-500V/div)

Pulse current (100A/div)

Current on sample (-100 V bias, 1A/div)


41

First SEM pictures (all films ~100 nm thick)

Substrate floating

Substrate grounded

Substrate biased -100V

Note: poor substrate preparationUnfortunately no new experiments were possible since last year


42

Films as a bulk – Hydrogen becomes a problem!

~1 µm grain size, RRR 28

~5 µm grain size, RRR 55

0.1 µm grain size, RRR 11

H2 content is ~ 0.1 at. % for sputtered Nb/Cu films (in niobium bulk it is0.02 at.%) and it is picked up from vacuum system during deposition.

A possible solution: high-temperature annealing, but it does not work with copper cavities. Proposal (L. Hand, W. Frisken): molybdenum cavities.

There are more differences between these Nb/Cu films than those listed,

this is just a basis for reflection


43

Film as a bulk – H2 measurements

From: L. Hand, Cornell U. – W. Frisken, York U.

There are new results on measurements of H2 content by measuring the lattice parameter and the total impurity content


44

Grain boundaries and surface oxidation

Famous drawings by Halbritter. Several effects might take place: ITE, flux penetration, Hc1 depression, lower Tc, etc.


45

Possible solution – Al2O3 cap layers

• Technique routinely used for S-N-I-S Josephson junctions: a 5-nm thick Al layer is deposited onto the Nb base electrode, and let oxidize in air. Most of it is transformed to Al2O3 but some remains metallic.

• It is important to prevent any surface contamination of Nb prior to Al coating, to reduce the coalescence of the Al atoms.

• Other possible solution: NbN overlayer (J. Halbritter)

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50

Thickness (nm)

Elem

ent "

conc

entra

tion"

(at.

%)

Al 2p

Nb 3d5

O 1s

EAl = 8.7 nm

EO = 2.4 nm

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50

Thickness (nm)

Elem

ent "

conc

entra

tion"

(at.

%)

Al 2p

Nb 3d5

O 1s

EAl = 12.8 nm

EO = 2.3 nm

5 nm Al (nominal) 10 nm Al (nominal)

XPS depth profile


46

State-of-the-art 20 years ago


47

State-of-the-art at 1500 MHz – 1.7 K – single cell

1

10

100

1000

0 5 10 15 20 25

Rs

[nO

hm]

Eacc

[MV/m]

Q 1x1010 @ 15 MV/m

Q 3x109 @ 20 MV/m

28 MV/m reached in a large LEP cryostat (He evaporation at large power)


48

Conclusion

• In an ideal world niobium films would be the best solution for accelerating cavities at any beta.

• However they are presently not competitive for reaching the highest fields because of the increase of Rres: WE MUST STRUGGLE AND UNDERSTAND WHY!

• Films are still a valuable option for lower fields and operation at 4.2 K

• The technique of choice is at present sputter deposition: a prerequisite for it is substrate design and its preparation

• Four proposed research lines:– Substrate effects– New deposition techniques (“energetic”)– Hydrogen effects– Cap layers

• Several of the above research lines are interdependent

a review of niobium (on copper) sputtering technology s. calatroni

Documents

tsergio calatroni cern

familysergio calatroni

nnb films

t rfl hrf

surface preparation

surface qualityexamples

surface stateexamples

surface oxidationconclusions