Reece - Squeezing the Mystery Out of Electropolishing Niobium

Squeezing the Mystery Out of Electropolishing Niobium

Charles Reece

Oct 6, 2010

Electropolishing of Nb cavities has been fruitfully employed for over 30 years. Most recently it has been procedurally applied to the 9-cell cavities for XFEL and ILC R&D.

Though no doubt EP has been demonstrated effective, basic understanding of its mechanisms has lagged behind. We continue to deepen our understanding of what “EP” does to niobium surfaces and apply that knowledge to optimize the process.

We want to understand the scale-specific details of surface leveling, believing that such understanding will yield confident process control and optimization.

SRF Thin Films & more Workshop 4-6 Oct 2010 2

The Basics – Understanding Potentials in Electrochemical Cell

• Provide an improved technical basis for specific process design, e.g. 9-cell cavities.

-8 -6 -4 -2 0 2 4 6 8 10 12 14 160.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Curr

ent (

A )

Voltage ( V )

cathode potential

anode potential

Measure potential of each electrode relative to a reference electrode (MSE) in the bath.

Power supply voltage

4

The Effect of Temperature on Cathode I-V Curve

Our studies show that the polarization resistance on the cathode decreases exponentially with temperature and is well behaved .

-8.0 -7.5 -7.0 -6.5 -6.0 -5.5 -5.0 -4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.00.0

0.5

1.0

1.5

2.0

2.5

Curr

ent (

A )

Voltage ( V )

Area Ratio of Nb/Al = 10 : 1 (Nb : 26.035 cm 2; Al : 2.6035 cm 2)Ref electrode & Thermal Couple nearby Nb ( < 5 mm ) 54.6 o C ( rise up to 60.9 o C)45.6 ( rise up to 47.7 o C)33.5 ( rise up to 34.1 o C)26.3 ( rise up to 26.6 o C)21.3 ( rise up to 21.7 o C)

-4.01 -1.38

0.4530.508

0.647 0.8110.981

0 10 20 30 40 50 60 70 801

10

100Equation: y = A1*exp(-x/t1) + y0Chi 2/DoF = 0.10763R 2 = 0.99531

y0 4.23282±1.66343A1 26.3770 ±3.27746t1 24.3274±6.86868

experimental result expontial fit interpolation

ohm

.cm

2

Temperature ( o C )

SRF Thin Films & more Workshop 4-6 Oct 2010

5

Observations• After separating the anode and cathode potentials, and

stabilizing temperature, one finds that the I-V plateau extends out past 20 V.

• The current density on the anode is directly proportional to the bulk F concentration and independent of amount of dissolved Nb, all other factors constant.

• Neglecting evaporative loss, HF concentration falls with consumption of Nb by

Nb2O5 + 10HF ⇒ 2NbF5+ 5H2O

Rs: electrolyte, temperature.

What is Electrochemical Impedance Spectroscopy?

Rp: polarization/charger transfer resistor

Cdl : double layer capacitor of electrode surface

Rwarburg: diffusion resistor

Rs: solution resistor

Nb Ref. Elec.Possible equivalent circuit of Nb-acid interface during EP

High Frequency Nyquist Plot

222

2

222 11 pdl

pdls

pdl

p

RCRC

jRRC

RZ

ωω

ω +−+

+=

ReZ

-ImZReZ -ImZ

Rp: temperature, concentration of reaction products, potential.

Rw: the frequency of potential perturbation

Cdl: electrolyte, temperature, potential, oxide layer

electrode roughness


Zreal (Ω)Rs remains constant

-Zim

ag(Ω

)EIS Study of Constant Current Density

0 2 4 6 8 10 12 14 16 180

5

10

15

20

25Area Ratio of Nb/Al = 10 : 1 (Nb : 26.035 cm 2; Al : 2.6035 cm 2)Ref electrode & Thermal Couple nearby Nb ( < 5 mm ) T = 21.5 o C

Anod

e Cu

rren

t Den

sity

( m

A/cm

2 )

Anode Potential ( V )

Rp increases with the potential

0.2 Hz

200 KHz

1.01 KHz =ωmax=1/RpCdl

0.2 Hz


NbBulk Electrolyte

Diffusion Layer(~ um)

Compact Salt Film(~ nm)

F -

%

F -

%

Distance

Distance

Diffusion-limited access of F- to the surface produces “best” polishing

Local temperature, flow and electrolyte composition affect the local F- gradient

• Anodization of Nb in H2SO4 forces growth of Nb2O5.• F- aggressively dissolves Nb2O5.• These competing processes result in current flow

and material removal.

• Above a certain anodization potential, the reaction rate plateaus, limited by how fast fresh F can arrive at the surface. (diffusion-limited)

• In this steady-state case, this Nb2O5 layer is a “compact salt film” with resistivity ∝ potential.

• The thickness of the salt film increases with applied potential, although the steady-state current does not change (plateau).

• In the diffusion-limited circumstance, material removal rate is blind to crystallography (avoids crystallographic etching).

• The diffusion coefficient determines the scale over which the F concentration varies from 0 to bulk.

We have successfully characterized the • temperature-dependent viscosity of the EP fluid• diffusion constant of F- in the fluid

This allows us to calculate the scale of optimum leveling.

We have also identified that a parallel etching process is present at higher temperatures – this works against obtaining smoothest surfaces efficiently with material removal.


Mass transport may occur by three mechanisms in an electrochemical cell. It is described by the Nernst-Planck equation, written for one–dimensional mass transfer along the x-axis as:

)()()()( xCxxCD

RTFz

xxCDxJ iii

iiii υφ

+∂

∂−

∂∂

−=

Migration - movement of ions driven by a gradient of electrical potential

Convection - natural convection driven by density gradientand forced convection (stirring, vibration, circulation)

Diffusion - movement of species ( F- )driven by a concentration gradient

Current-limited plateau is the result of a mass-transport limitation


)()()()( xCxxCD

RTFz

xxCDxJ iii

iiii υφ

+∂

∂−

∂∂

−=





If “really” in I-V plateau, gradient at surface must be negligible


)()()()( xCxxCD

RTFz

xxCDxJ iii

iiii υφ

+∂

∂−

∂∂

−=






In ideally static case, only diffusion matters.


)()()()( xCxxCD

RTFz

xxCDxJ iii

iiii υφ

+∂

∂−

∂∂

−=






One may learn about Di by deliberately controlling convection.

RDE : creates a defined solution flow pattern in which the mass transport of species is almost completely due to convection. By solving the convection equation with the boundary condition, the Levichequation can be used to describe the relationship of limiting current to the physical properties of electrolyte bath - diffusion coefficient (D) and kinematic viscosity (ν).

Levich equation

0.67 0.166 0.50.62J nFD cυ ω−=0.5 0.67 0.166( . ) 0.62slope J vs nFD cω υ−=

: coskinematic vis ityυ: rotation speed of the electrodeω

Determination of the limited species diffusion coefficient by rotating disk electrode (RDE) technique

c: concentration of active species

I-V curves of Nb electropolishing at different temperatures with RDE

0 2 4 6 8 10 12 14 16 18 20 220.0000.0010.0020.0030.0040.0050.0060.0070.0080.0090.0100.0110.0120.0130.0140.015

4000 RPM3600 RPM3200 RPM2800 RPM2400 RPM2000 RPM

1600 RPM

1200 RPM

800 RPM

400 RPM

Curr

ent (

mA)

Voltage (vs. MSE)

HF(49%):H2SO4 (96%)=1:10T=10 oC

0 RPM

0 2 4 6 8 10 12 14 16 18 20

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

4000 RPM3600 RPM3200 RPM

2800 RPM2400 RPM2000 RPM

1600 RPM1200 RPM

0 RPM

800 RPM

Curr

ent (

mA)

Voltage( vs.MSE)

HF(49%):H2SO4(96%)=1:10T=19oC

400 RPM

0 2 4 6 8 10 12 14 16 18 200

25

50

75

100

125

150

175

200

225

250

4000 RPM3600 RPM3200 RPM

2400 RPM2800 RPM2000 RPM

1600 RPM1200 RPM800 RPM400 RPM

HF(49%):H2SO4 (96%)=1:10T=30 oC

Curre

nt D

ensit

y (m

A/cm

2 )

Voltage (V) (vs. MSE)

0 RPM

0 2 4 6 8 10 12 14 16 18 200

50

100

150

200

250

300

350

400

2400 RPM

3600 RPM3200 RPM

4000 RPM

2000 RPM1600 RPM

1200 RPM800 RPM 400 RPM

Curre

nt D

ensit

y (m

A/cm

2 )

Voltage(V) (vs. MSE)

HF(49%):H2SO4(96%)=1:10T=41oC

0 RPM

10 °C 19 °C

30 °C 41 °C

0 – 4000 rpm

Note only ~20% current rise with 400 rpm, at constant temperature

Strong evidence for temperature-dependent etching in parallel with the diffusion-limited process. The second process is neither dependent on potential or diffusion.

⇒ For best leveling, stay below 20°C

0 2 4 6 8 10 12 14 16 18 200

50

100

150

200

250

300

350

slope=10.14

slope=9.82

slope=5.68

slope=4.23

slope=2.87

slope=1.79

Curr

ent d

ensi

ty(m

A/cm

2 )

ω1/2 (rad.s-1)1/2

T= 1 +/- 0.5 oCT= 9.0 +/- 0.5oCT= 19.0 +/- 0.5oCT= 30.0 +/- 0.5oCT= 41.0 +/- 0.5oCT= 50.0 +/- 0.5oC

0.5 0.67 0.166( . ) 0.62slope J vs nFD cω υ−=

Excellent linear fit provides definitive evidence of a diffusion-limited process. Knowing ν and c yields D.

RDE measurements

cF = 2.67×10-3 mol/cm3

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0 10 20 30 40 50

ν(c

m2/

s)

T (°C)

Kinematic Viscosity of 1:10 HF/H2SO4 Electrolyte

H. Tian, JLab

Measured using a Brookfield DV-II pro viscometer

0.0E+005.0E-081.0E-071.5E-072.0E-072.5E-073.0E-073.5E-07

0 10 20 30 40 50

D(c

m2/

s)

T (°C)

Diffusion Coefficient of 1:10 HF/H2SO4 Electrolyte

RDE measurements + viscosity measurements + concentration determine the Diffusion coefficient

cF = 2.67×10-3 mol/cm3

Determining Electrolyte Physical Properties

0 10 20 30 40 5068

10121416182022242628303234

Temperature (oC)

Diffu

sion

laye

r thi

ckne

ss (µ

m)

1δ 2δ

(0, ) 0F

C t− ≈

BulkF

C −

cJ n F Dδ

= × × ×

Estimation of diffusion layer thickness in 1:10 HF/H2SO4 Electrolyte at different temperatures

cF = 2.67×10-3 mol/cm3

There exists a F- concentration gradient within the 10-20 µm away from the surface.

On this scale and below, peaks are dissolved much faster than valleys.

10-2 10-1 100 10110-2

10-1

100

101

102

103

104

105

106

107

Spatial frequency (µm-1)

PS

D(n

m2 )

PSD of Fine CBP Nb Surface Before/ After EP

KEK fine CBP fine grain sample 2KEK fine CBP large grain sample 9KEK fine CBP single crystal sample 13KEK fine CBP large grain sample 9 after EPKEK fine CBP single crystal sample 13 after EPKEK fine CBP fine grain sample 2 after EP

RMS~200nm

AFM Measurement ( 50µm*50µm)

RMS~7nm

RMS~40nm

With “standard” 1:10 HF/H2SO4 Electrolyte at 30°CNb crystallography affects the polishing effectiveness.

With identical starting topography from CBP, given identical 100 min “EP” at 30°C, single-crystal material was significantly smoother.

Physical evidence for a significant etching activity being present at 30°C

Not all Nb “EPs” are the same

Avoid sulfur production at the cathode• Most electropolishing applications attempt to maximize the surface area of

the cathode to avoid process complications (cost).• In contrast to this, typical horizontal cavity EP circumstances have a

cathode:anode active area ratio of 1:10.• Result is high current density on the cathode and necessary high

overpotential to drive the current. This risks driving other chemistry.

SO42- + 8 H+ + 6 e- → S + 4 H2O

may proceed if cathode overpotential is too high

-8 -6 -4 -2 0 2 4 6 8 10 12 14 16 180.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Curr

ent (

A )

Voltage ( V )

Area ratio of Nb/ AlT = 20.5 +/- 1.3 o C Al area kept unchanged ( 2.6 cm 2 )10 : 1 ; 8 : 1 ; 6 : 14 : 1 ; 2 : 1 ; 1 : 1

-8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 200

20406080

100120140160180200220240260280300320340

-8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20

020406080100120140160180200220240260280300320340

Anod

e Cu

rrent

Den

sity

( mA/

cm2 )Area ratio of Nb/ Al

T = 20.5 +/- 1.3 o C Al area kept unchanged ( 2.6 cm 2 )10 : 1 ; 8 : 1 ; 6 : 14 : 1 ; 2 : 1 ; 1 : 1

Cath

ode

Curre

nt D

ensit

y ( m

A/cm

2 )

Voltage ( V )



21

External Temperature control of Cavity EPWater spray on exterior during 1 rpm rotation

22

Nb 7-cell Cavity EP with Temperature Control33µm removal

23

Consistent positive results from controlled 30 µm EP after BCPJLab 7-cell “LL” cavities

Implications:• We should expect the best micropolishing for topographic features

smaller than ~ 15 µm, so start with surfaces that are consistently smooth to this scale: CBP or CMP.

• This process we call “EP” also has a temperature-dependent etching process present. So, control and minimize the temperature as much as is practical.

• Reduce sulfur production at the cathode by minimizing cathode current density and improving the reaction kinetics for hydrolysis at the cathode.

• Noted cavity performance difference with EP current density 17-36 mA/cm2 (> 35 MV/m)

1:10 HF/H2SO4 Electrolyte with NbIf the objective is maximally smooth surfaces:

Next:• Characterize the topographic transformation function sufficient to

gain predictive power over topography as a function of EP time.

25

Conclusion

Basic process studies are yielding improved understanding of theelectropolishing of niobium. This understanding leads to betterprocess control and product reproducibility. It may also lead tonew efficient process designs from first-principles.

Application of light, controlled EP is expected to significantlyimprove the technical contingency of the 12 GeV Upgrade projectand lower future cryogenic operating costs.

ContributorsH. Tian JLab, (W&M)O. Trofimova JLabM. J. Kelley JLab, W&M, VTL. Zhao JLab, W&MS. Corcoran VTS. Brankovic U HoustonG. Ribeill JLab (DOE SULI)

Key References:

A. J. Bard and L. R. Faulkner, Electrochemical Methods, Wiley: New York, 1980.H. Tian, S. G. Corcoran, C. E. Reece and M. J. Kelley, J. Electrochem. Soc. 155(2008), p. D563.V.G. Levich, Physicochemical Hydrodynamics, Prentice-Hall, New York, 1962.F. Eozénou, S. Berry, C. Antoine, Y. Gasser, J.-P. Charrier and B. Malk., PhysRev-STAB 13, 083501 (2010).H. Tian and C. E. Reece, PhysRev-STAB 13, 083502 (2010).H. Tian, Ph.D. Dissertation, Dept. of Applied Science, College of William and Mary, (2008).C. Reece and H. Tian, Contribution to LINAC10 THP010 (Sept 2010)

“A Novel Approach to Characterizing the Surface Topography of Niobium Superconducting Radio Frequency (SRF) Accelerator Cavities,” H. Tian, C. E. Reece, and M. J. Kelley, Appli. Surf. Sci., (submitted) (2010).

Authored by Jefferson Science Associates, LLC under U.S. DOE Contract No. DE-AC05-06OR23177.

Technology

Reece - Squeezing the Mystery Out of Electropolishing Niobium