Field emission measurements on flat Cu samples relevant for CLIC accelerating - PowerPoint Presentation

Slide1

Field emission measurements on flat Cu samples relevant for CLIC accelerating

strucutres

S. Lagotzky, G. Müller

University

of Wuppertal, FB C – Physics Department, Wuppertal,

Germany30-09-2014

Motivation and theoryMeasurement techniquesSamplesField emission resultsConclusions and Outlook

Acknowledgements

:

Funding

by

BMBF

project

05H12PX6

Slide2

Dark current / electric

breakdown is the main field limitation of accelerating structures for CLIC (E

acc=100 MV/m, Epk=243 MV/m)Deep and quantitative understanding of the origin of breakdown processes is important!

Goal: Suppression of breakdowns by using proper surface treatmentsInvestigation of the enhanced field emission (EFE) from Cu surfaces as

precursorof breakdownWhat causes EFE from (relevant) Cu samples?

How to reduce/avoid EFE?MOTIVATION

Typical result of a dcdischarge on Nb

Slide3

Field emission

Field Emission (FE): „cold“ electron emission induced by high electric fields

Tunneling current

through the resulting barrierRound shape of the barrier caused by the mirror charge (100MV/m: 0.38eV)vs.

dc FE current given by Fowler-Nordheim (FN)

results in a straight line

(FN-Plot)

For Nb: 1nA/μm

2

@ 2000

MV/m

Slide4

enhanced field Emission (EFE)

Electric

field enhancement factor:

h = height of defect; r

= curvature radiusEL = local electric field on

defectES = electric field on flat surface

particles

/

protrusions

:

scratch:

metal-insulator-metal

(MIM)

:

→

emission

area

→ activation by burning of conducting channels:

MIV

Ambient

oxide layer

Nb

Insulator

MIM

Nb

(1) Nb surface oxide:

ad- or desorption lead to enhancement or reduction of field

resonance tunneling can occur:

(2) Adsorbates:

Slide5

Activation of emitters

In

sulating oxide layer (IOL, thickness dox ~ few nm) on metallic surfaces

Question: How are emitters activated?

Surface

defects (MIV)

Particulates (MIM)

 

 

 

 

Conducting channel (CC) is burned into oxide by activated emission current

Stronger if

 

always

stronger

Slide6

Emitter number density N

Question

: How many emitters become active at a certain field level?Single emitter is activated if

(1)Elim depends on dielectric breakdown of the IOL (dc) and

probably on microwave losses in rf cavitiesInclusion of all emitters at given Eact that

fulfil (1):Number of activated emitters:N(β

act) not known yet, reasonable approach*  

*H

.

Padamsee et al., p. 998 – 1000, Proc. PAC1993.

N

0

:

normalization

factor

c

s

:

surface condition factor

N

tot

: total number of emitters

Slide7

Resulting

N(Eact) dependence

N(Eact=0) = 0Exponential-like increase for low EactSee A. Navitski et al., PRSTAB 16, 112001 (2013)Saturation towards N

tot for high Eact Ntot,1

>Ntot,2 and cs,1>cs,2

Overall measured data should be fitted by using with

and

s

ee

S. Lagotzky

and

G. Müller, IPAC14, WEPME004

asked for submission to PRSTAB

 

Slide8

DC Field Emission Scanning Microscope (FESM)

sample

anode

piezotrans-latorselectron gunion gun

Regulated voltage V(

x,y

) scans at fixed FE current (typ. I = 1nA) and gap ∆z (Øanode = 300 µm, scan range ≤ 25ˣ25

mm

2

,

tilt

correct.

±

1 µm within

±5 mm

)

emitter

position, number

density N and localization of emitters

Local U(z) & I(V)

measurements of single emitters

→ E

on

(1 nA)

,

β

FN

,

S

FN

Ion

gun

(

E

ion

= 0 – 5

keV), SEM (low res.), AES, heat treatments

(< 1200°C)Clean laminar

air flow around

load-lockEx-situ SEM & EDX: Identification of emitting defects (positioning accuracy

~100 µm)

+AES

Slide9

Surface quality controll

Optical Profilometer (OP)

white

light irradiation and spectral reflection (chromatic aberration)20x20 cm² scanning range in 2 cm distanceCurved surface up to 5 cm height difference2 µm (3 nm) lateral (height) resolution Further zooming

by AFM:±2 µm positioning relative

to OP results98x98 µm2 scanning range3 (1) nm lateral (height) resolutioncontact or non-contact modes.

Clean laminar air flow (LAF) from the back Granite plate with active damping system

CCD camera for fast positioning

interferometric

film

thickness

sensor

(IF)

Slide10

SAMPLES

Protection

capSample

HolderFESM adapterInvestigatedsurface

Flat

Cu

samples Diameter: ~11 mmHole as mark to relocate the emitter position in different systems (accuracy ~ 500 µm)Diamond turned (DT) and partially chemically etched (0.6 µm, SLAC treatment at RT for 5 s)

using H

3

PO

4

(70.0%), HNO

3

(23.3%), acetic glacial acid (6.6%), and

HCl

(0.49

%)

Glued with SEM button on a holder and mounted to an adapter for the FESM at BUW

Final cleaning at BUW

Teflon protection

cap to avoid damage and contaminations after polishing

and

cleaning

Slide11

SURFACE QUALITY

Sample surface very flat (±0.5 µm)

Many pits (N < 18 mm-2) due to

etchingGrain size: 1300 µm² - 5.3 mm²Average roughness: Ra/Rq = 150/230 nm

Samples measured with OP

before

DIC in the FESM relevant area

DT & SLAC

Slightly waved surface (λ~ 0,5 – 1 mm)

Many ridges from DT

Damage layer?

Average roughness: R

a

/R

q

= 126/145 nm

DT

Slide12

Field emission results DT + SLAC treatment

0

5 mm

5 mm03015

05 mm

5 mm06030

05 mm5 mm0

100

50

Field [MV/m]

Field [MV/m]

Field [MV/m]

Sample got DT and SLAC treatment, but no final cleaning

First

emission

at 30 MV/m

N = 28 ± 11 cm

-2

N = 152 ± 25 cm

-2

38 emission sites at E

act

= 100 MV/m

Emitter number density : 152 cm

-2

Emitter uniformly distributed in the scanned area

Activation field E

act

> onset field E

on

E

act

= 80 MV/m, E

on

=47 MV/m

Similar results on a second sample

Slide13

Field emission results DT + ionized N2

cleaningSample got only DT and final cleaning by ionized N2

(p ≈ 5 bar)

First emissionat 130 MV/mN = 20 ± 9 cm-21

2

35N = 52 ± 14 cm-2

13 emission sites at Eact = 190 MV/mEmitter number density : 52 cm-2Emitter uniformly distributed in the scanned area

Activation field E

act

> onset field E

on

E

act

= 180 MV/m, E

on

=114 MV/m

Slide14

EFE activation statistics

Emission from surfaces without any cleaning starts at 30 MV/mlg(N) increases linear with inverse field as expected

229/372 emitters/cm2 on SLAC-etched samples without N

2 at E = 243 MV/mCleaning with N2 shows a reduction of N down to 124 emitters/cm2 most probably due to

removal of large particulates

Linear Fits A + B×E-1 :A

= 2.67119, B = -75.79643A = 2.87218, B = -73.42952A = 3.77774, B = -408.55786

Slide15

Single emitter characteristics on SLAC-etched samples

without cleaning

EDX

shows S, Cl, KEDX shows S, Cl, SiEFE is

dominated by

particulates (d ~ 10 – 30 µm) → Removing particulates

to reduce EFE

20

emitters

investigated

with

SEM/EDX:

12

particulates

(Al, Cl, S, Si, K)

2

surface

defect

6

emission

sites

:

unknown

origin

0

100

50

Field [MV/m]

Slide16

DRY ICE CLEANING SYSTEM

Cleaning the surface by

Pressure

and shearing forces due to high velocity of snow crystalsBrittling of contaminations by rapid coolingPowerful rinsing due to the 500 times increased volume after sublimationSolvent cleaning by melted CO2 snow particles

Particulates with d > 100 nm are removedCommercial DIC system (SJ-10, CryoSnow) installed in cleanroom (class iso 5)

hand gunnozzle

Clean roomenvironmentcontrolpanelinlet CO2

i

nlet N

2

Slide17

Cleaning process

Cleaning of (grounded) samples with handgun (d ~ 5 cm) typically for 5 min

Liquid CO

2 (10 bar) and N2 (8 - 10 bar, propellant gas)Flat (12x3 mm) or round (Ø = 5 - 10 mm) jet of CO2 snow particlesSamples are treated 2.5 min under 90°/ 45°and 3 x rotated in 90°steps

Teflon protection caps are cleaned as well

Slide18

Avoiding particulate contaminations

A cleanroom environment (class ISO 3) was installed around the load-lock of the FESM to avoid particulate contaminations during installation of samples

Protection cap mechanically fixed until sample reaches cleanroom environment

Protection cap loosened under laminar air flowFinal removement of cap in preparation chamber at p ~ 10-7 mbar

Slide19

EFE results after DIC on DT + SLAC sample (17E)

Field maps between 120 - 300 MV/m, 20 (10) MV/m steps for E < (>) 200 MV/mScanned area: 5x5 mm², truncated cone anode

(W, Ø= 300 µm), step size = 150 µm, Δz = 25 µm (E ≥ 240 MV/m), 40 µm (200 – 240 MV/m)

or 50 µm (E < 200 MV/m)

No EFE at

120 MV/mDischarge at 140 MV/m

First stable EFE at 240 MV/m14

emission

sites

(

including

discharge

)

at

E

act

= 300 MV/m

Emitter

number

density

: 56 cm

-2

EFE

free

region

in

the

scanned

area

at

E = 300 MV/m

Activation

field Eact > onset

field Eon

Eact = 260 MV/m, E

on =128 MV/m

Slide20

EFE results after DIC on DT + SLAC sample (18E)

Field maps between 140 - 260 MV/m, 10 (20) MV/m steps for E > (<) 200 MV/m

Scanned area: 5x5 mm², truncated cone anode (W, Ø= 300 µm), step size = 150 µm, Δz = 25 µm (E ≥ 240 MV/m), 40 µm

(180 – 240 MV/m) or 50 µm (E < 180 MV/m)

23

emission sites at E

act = 260 MV/mEmitter number density : 92 cm-2EFE free

region

in

the

scanned

area

at

E = 260 MV/m

Activation

field

E

act

>

onset

field

E

on

E

act

= 260 MV/m,

E

on

=168 MV/m

Slide21

Efe activation statistics after dic

DIC reduces N significantly from

N= 229 cm-2

and N = 372 cm-2 (N = 124 cm-2) without (with) N2 cleaning

to N=29 cm-2

@ E = 243 MV/m Chemical etching did not further reduce N

log(N) increases nearly exponentially with E-1 as expectedBut still at least ~ 30 emitters in the iris area of a CLIC accelerating structure (~1 cm2)

Averaged over

4

samples

Linear

Fits

A + B×E

-1

:

A = 3.77774, B = -

408.5579

A

=

3.62966,

B =

-501.8538

A =

2.95432,

B =

-362.2702

Slide22

SINGLE EMITTER CHARACTERISTICS on DIC samples

Local I(V) curves of 49 emission sites selected from E(

x,y) maps; SEM/EDX analysis of the Cu surface revealed: 57% surface defects; 12% small (< 2 µm) particulates (Al, Si, W); 31% unidentifed;

Rather stable FN-like EFE

Slight jumps, probably due to melting of micro-tips

More unstable EFE

Changed slope at high fields due to bad electrical contact to bulk

Slide23

Single emitter statistics

More examples for DT + SLAC + DIC samples:

Ca

Si, Al

Al, Si

Slide24

Reproduction of dic-effect after slac etching

Applying DIC on DT sample has led to a significantly reduced N = 29/cm2

Field maps of two more DT+SLAC+DIC samples:

05 mm08040Field [MV/m]

First emission(old results: 140 MV/m)

0

5 mm06040Field [MV/m]#1

#2

0

5 mm

0

160

80

Field [MV/m]

0

5 mm

0

160

80

Field [MV/m]

0

5 mm

0

180

9

0

Field [MV/m]

0

5 mm

0

210

80

Field [MV/m]

Slide25

Comparison to old results

New

results on etched samples

are worseA = 2.67119, B = -75.79643A = 2.87218, B = -73.42952

N = 229 / 372 emitter

/cm2 at E = 243 MV/mWhat is

the reason for that?Only 1 particulate on two samples but 7 surface defects (etched pits + stains) most-likely caused by SLAC treatment!

Etching damages the surface and is too bad for high-gradient cavities

Slide26

FN-PARAMETERS STATISTICS

Fitting the FN-equation to the FN-plots of every measured emitter using the assumption φ

= 4.65 eV:Characterizing every emitter by the resulting

FN-parameters βFN and SFNA = 154, B = 6830

[E] = MV/m, [I] = ASurface defects & particulates reveal mainly β

FN = 10 - 70 15% with βFN

< 150, mainly particulatesNo correlation with geometric field enhancement (SFN ~ βFN-2), especially at low βFN

-values

Clear hint for other EFE mechanisms like MIV- and MIM-emission

Surface

defects

Particulates

U

nknown

Slide27

Origin of breakdowns in rf structures

After activation: Field level of IFN = 1 nA is reduced → E

on(1 nA) < EactDetermination of Eon

(1 nA) by local emitter field calibration (U(z)-measurement)Eon for surface defects & particulates similarFew emitters have very low Eon<100 MV/m!

Measure for the activation strength:field reduction factor ρ

= Eact/EonDetermination

of ρ for 49 emission sites:61% (20%) of

emitters

show

ρ

< 2 (> 3)

90%

of

emitters

with

ρ

> 3

are

caused

by

surface

defects

after DIC

High-

ρ

emitters are most likely candidates for triggering BDs in accelerating structures due to the exponential current increase after their activation!

Slide28

Examples: Two candidates for breakdowns

Activated

between 240 – 250 MV/m, E

on = 54 MV/m → ρ = 4.62 ± 0.19Calculated current at 243 MV/m with βFN

= 17 and SFN

= 1.1×1025 m²! : IFN ~ 10

22 A!Activated between 130 – 140 MV/m, Eon

= 80 MV/m →

ρ

= 1.75 ± 0.12

Calculated

c

urrent

at 243 MV/m

with

β

FN

= 43

and

S

FN

= 2.4×10

2

µm² : I

FN

~ 3 mA

Nearly

indipendent

of

φ

Slide29

Observing discharges in the fesm

Sometimes discharges happen also during measurements in the FESM by accidentDuring scans if the current jump is faster than the voltage regulation (~2

ms)During local measurement because of activation effects

Discharges (most-likely caused by high-

ρ emitters

) destroy the surface and lead to the formation of new stable and strong emitters, similar to BDs in cavitiesIn accelerating structures this new emitter triggers the next BD, which forms another emitter, that ignite a BD etc.Avoiding the original emitter potentially avoids the following BDs and reduces the BDR

Slide30

conclusions

Scaling law

for field dependence of emitter

number density on Cu samples foundActual surface quality of the Cu samples is not sufficient for high-gradient CLICstructures and shows up to N = 370 cm-2 at E = 243 MV/m

DIC decreases N significantly by a factor of ~ 10 down to N = 29 cm-2 (E = 243 MV/m)

Still not good enough for accelerating structuresMight reduce the BDR and/or the conditioning time of the accelerating structuresEtching the surfaces by SLAC treatment damages the surface and produces surface defects that emit at 243 MV/mReplacing SLAC treatment by electropolishing might reduce the BDR even more

Geometrical field enhancement is not sufficient to explain the observed EFE from relevant Cu surfaces Alternative emission processes like the MIV/MIM-model or voidsEmitters with

ρ

>

2

are one candidate for causing BD in the accelerating structures

and are mainly caused by remaining surface defects after DIC

Slide31

OUTLOOK

Improving DIC by optimizing the parameters (pressure, cleaning time, cleaning procedure, CO

2/N2 ratio, …) to remove even more particulatesMeasuring two samples with only DT after DICEmitter processing by current or by ion bombardment and SEM investigations before and after this conditioning