www.DRAFT.ugent.be

Target effective electron emission coefficients during reactive sputtering of Oxides and

Nitrides

R. De Gryse, D. Depla, S. MahieuDepartment for Solid State Sciences

Ghent University (S1) B-9000 Ghent, BelgiumX.Y. Li

Department of Physics, Taiyuan University of Technology

030024 Taiyuan, China

ww

w.d

raft.

ugen

t.be

2

Lay Out

LAY OUT

- Problem we want to tackle

- Measuring procedure

- Results for nitrides

- Results for oxides

- Discussion

- Conclusion

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

3

Thornton : revised

0discharge

0 i i

WVmf

=ε ε γ

W0 : effective ionisation energy

εi : ion collection efficiency (for magnetron : almost 1)

ε0 : fraction of maximum possible number of ions (almost 1)

m : multiplication factor : accounts for ionisation in the sheath

f : effective ionisation probability : influenced by electron recapture

γi : ion induced secondary electron emission coefficient

* G. Buyle, “Simplified model for the DC planar magnetron discharge (PhD, UGENT,2005)

original : γeffective

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

4

Emission mechanisms

Electron emission mechanisms under ion bombardment

Kinetic emission Potential emission

- Coulomb interation - Auger neutralisation

- Electron promotion - Auger deexcitation

Typical for oxides Typical for metals

γiK > γiP

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

5

Hysteresis behaviour Ti/O2 and Al/O2

480

460

440

420

400

380

440

400

360

320

2.52.01.51.00.50.0oxygen flow (sccm)

disc

harg

e vo

ltage

(V) oxygen addition

oxygen removal

Ti

Al

Ti/O2

Discharge voltage INCREASES on addition of oxygen

Al/O2

Discharge voltage DECREASES on addition of oxygen

Measurements by S. Heirwegh

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

6

Measuring scheme

time

stat

usst

atus

volta

gest

atus

argon

oxygen

magnetron

VAr

VO2VoxAr

Δt480

460

440

420

400

380

oxygen addition oxygen removal

Ti

Oxygen flow

Dis

char

ge v

olta

ge

VAr

VO2

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

7

Discharge voltages: nitrides

Comparison between the discharge voltage measured in pure Ar (VAr),pure nitrogen (VN2)and the discharge voltage of an nitrided target in pure Ar (Vnitr,Ar).

700

600

500

400

300

200

100

0

disc

harg

e vo

ltage

(V)

Ag Al Au Ce Cr Cu In Mg Mo Nb Pb Pd Pt Re Ta Ti Y Zr

VN2 VNitr,Ar VAr

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

8

Inverse of discharge voltage versus γi4.0

3.5

3.0

2.5

2.0

1.50.250.200.150.100.050.00

4.0

3.5

3.0

2.5

2.0

1.50.250.200.150.100.050.00

4.0

3.5

3.0

2.5

2.0

1.50.250.200.150.100.050.00

4.0

3.5

3.0

2.5

2.0

1.50.250.200.150.100.050.00

AgAlAuCeCuCrInMgMoNbPbPdPtReTaTiYZr

ion induced electron emission yield (γ)

ion induced electron emission yield (γ) ion induced electron emission yield (γ)

ion induced electron emission yield (γ)

inve

rse

of th

e di

scha

rge

volta

ge (x

10-31/

V)

inve

rse

of th

e di

scha

rge

volta

ge (x

10-31/

V)

inve

rse

of th

e di

scha

rge

volta

ge (x

10-31/

V)

inve

rse

of th

e di

scha

rge

volta

ge (x

10-31/

V)

slope : 1/85.5 slope : 1/77.3

slope : 1/76.9 slope : 1/71.6

current : 0.4 Apressure : 0.4 Pa

current : 0.6 Apressure : 0.4 Pa

current : 0.6 Apressure : 0.6 Pa

current : 0.4 Apressure : 0.6 Pa

The inverse of the discharge voltage as a function of the ion induced γi fordifferent target materials under several experimental conditions. Themeasurements were performed with a conventional two inch magnetron in apure argon atmosphere. All targets had a purity of 99.99%. The dotted line is alinear fit. For all conditions the correlation coefficient has a value in the interval0.87-0.89.

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

9

Relative variation of γi: nitrides

Relative change of the effective emission coefficient by nitridation of the metal target. γM and γC stands for respectively the γi of the metal and the compound.

-100 0 100 200

AlMg

YCeIn

PbTaMoReNbCrCuTiZrAuPtAgPd

γC,nitride

0.084

0.22 2.8 eV

0.031

0.22 1.76 eV

0.096

0.055 0.23 eV

0.27 6.2 eV

0.079

0.097

0.059 0.67 eV

0.0490.071

0.094

0.11 0.071 eV

0.18 1~2 eV

0.12

0.071

(γC-γM)/γM (%)

0.088

band gap

wide band gapmedium band gapsmall band gapconductorno datareacted ?

-100 0 100 200

AlMg

YCeIn

PbTaMoReNbCrCuTiZrAuPtAgPd

γC,nitride

0.084

0.22 2.8 eV

0.031

0.22 1.76 eV

0.096

0.055 0.23 eV

0.27 6.2 eV

0.079

0.097

0.059 0.67 eV

0.0490.071

0.094

0.11 0.071 eV

0.18 1~2 eV

0.12

0.071

(γC-γM)/γM (%)

0.088

band gap

-100 0 100 200

AlMg

YCeIn

PbTaMoReNbCrCuTiZrAuPtAgPd

γC,nitride

0.084

0.22 2.8 eV

0.031

0.22 1.76 eV

0.096

0.055 0.23 eV

0.27 6.2 eV

0.079

0.097

0.059 0.67 eV

0.0490.071

0.094

0.11 0.071 eV

0.18 1~2 eV

0.12

0.071

(γC-γM)/γM (%)

0.088

band gap

wide band gapmedium band gapsmall band gapconductorno datareacted ?

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

10

Relative variation of γi: oxides

Relative change of the effective emission coefficient by oxidation of the metal target.

-100 0 100 200

MgLiY

PbAl

CeNbMoReTaPtTi

CuCrAgAuInZr

γC,oxide

0.067

0.19

0.036

0.22

0.086

0.092

0.40

0.0360.044

0.27

0.022

0.038

0.057

0.078

0.37

0.13

0.067

(γC-γM)/γM (%)

0.38

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

11

Problem

PROBLEM !

Oxides are wide gap materials

and

γi (oxide) >> γi (metal)

is expected

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

12

Phelps compilation

0.001

2

4

68

0.01

2

4

68

0.1

2

4

68

1

seco

ndar

y el

ectro

n em

issi

on y

ield

1012 3 4 5 6 7

1022 3 4 5 6 7

1032 3 4 5 6 7

104

energy (eV)

Ar+

Ar

clean metals dirty metals

γi for argon ions and fast argon atoms bombardment (after[Phelps1999]) of clean metals and dirty metals. Above 250 eV ionsthe γi is substantially higher for a dirty, oxidized surfaces as comparedto the clean metal surface.

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

13

Wittmaack 19991. Implantation of low energy oxygen ions in Si

1.1: γi proportional to the surface coverage of SiO22. Sputter etching of SiO2 with Ne+

2.1: rapid decrease of γi2.2: preferential loss of oxygen2.3: γi of suboxide SiOx (x < 2) shows a negligible

variation as compared with that of the parent metal

Wittmaack’s conclusionsLay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

14

Suboxides

Powder composition for the production of the titania targets

TiO2 (wt. %) Ti (wt. %) Target phase composition x in TiO2-x

__________________________________________________________

100 0 TiO2 (rutile) 0.25

80 20 Ti2O3 (50%), TiO (50%) 0.6

65 35 TiO 1

50 50 TiO (small amount Ti2O) 1.25

30 70 Ti2O 1.6

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

15

γi of Ti-suboxides

0.13

0.12

0.11

0.10

0.09

0.08

0.07

effe

ctiv

e SE

EY

2.01.51.00.5x in TiO2-x

TiTiO1.75

Calculated effective γi as a function of the target stoichiometry fortitania suboxide targets.

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion γi

ww

w.d

raft.

ugen

t.be

16

Effect of electronic properties

What can we conclude ?

Electronic properties of (Ti) suboxide surfaces are

completely different as compared to the surface

properties of the basic oxides.

Remember DC-sputtering of TiOx targets!

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

17

Model

- Reactive magnetron sputtering with oxygen:⇒ Formation of oxide layer on target by shallow oxygen implantation⇒ Due to preferential sputtering, the target oxides can be reduced to suboxides with low γi values⇒ Discharge voltage will increase under poisoning condition

- Which target materials are sensitive to reduction and the formation of suboxides ??

- Are there target materials which do not form suboxides under ion bombardment ??

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

18

Surprise

1. The oxides of Al, Mg, Ce, Y sputter congruently→ no preferential sputtering→ no formation of suboxides→ surface composition = bulk composition

= fully oxidized materialThese oxides are exactly the high γi oxides !

2. Nitrides show little or no preferential sputtering→ surface composition = bulk composition→ wide gap nitrides remain wide gap materials under ion

bombardment and vice versa

- Electronic properties of both classes of target materials are not affected by ion bombardment !!

- Can we predict this behaviour ?

From literature:

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

19

Reduction factor

Surface reduction factor R can be estimated:

s b

OM M

O M O

YX XX Y X

⎛ ⎞ ⎛ ⎞=⎜ ⎟ ⎜ ⎟

⎝ ⎠ ⎝ ⎠

2m 1 2m

O M M

M O O

Y AM UY AM U

−⎛ ⎞ ⎛ ⎞

= ⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠

( )( )

sM O

bM O

X XR

X X=1.

2.

3.

preferential sputtering (Malherbe)

collision cascade (Sigmund)

AMM, AMO : atomic masses

UM, UO : binding energies

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

20

γi versus R

The calculated γi for oxides as a function of the reductionfactor R calculated with the model of Malherbe et al. (withm=0.05). Nitrides show no preferential sputtering

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

0.5

0.4

0.3

0.2

0.1

0.01.81.61.41.21.00.80.6

AgAlAuCeCrCuInLiMgMoNbPbPtReTaTiYZr

reduction R

γ i

γi

ww

w.d

raft.

ugen

t.be

21

Conclusion1. Electronic properties of the target surface under ion bombardment

(preferential sputtering or not) are the dominant factors in the understanding of the ion induced secondary electron emission.

2. Small gap materials and metals have a low γi

- oxides of Nb, Mo, Re, Ta, Ti, Cu, Cr, In, Zr which under ion bombardment reduce to suboxides

- nitrides of In, Pb, Ta, Mo, Re, Nb, Cr, Cu, Ti, Zr, Pd which are either semiconductors or conducting nitrides. These materials show no or very small preferential sputtering

Potential emission is the dominant mechanism for production of secondary

electrons (low γi )

→ Discharge voltage increases upon poisoning !

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

22

Conclusion

3. Wide gap materials or insulators have a high γi

- oxides of Mg, Li, Y, Pb, Al, Ce which are not subjected to preferential sputtering under ion impact

- nitrides of Al, Mg, Y, Ce which are also not sensitive to preferential sputtering (as all nitrides under study)

Kinetic emission is the dominant mechanism of secondary electron emission (high γi)

→ Discharge voltage decreases upon poisoning !

4. The hysteresis behaviour under reactive sputtering is understood and predictable !

Lay Out

The problem

Measuring procedure

Nitrides

Oxides

Discussion

Conclusion

ww

w.d

raft.

ugen

t.be

23

Acknowledgements

The authors are indebted to the

for financial support

company