X-ray diagnostics of plasma deposited thin layers

H.Wulff summer school „complex plasmas“, Hoboken 2008

X-ray diagnostics of plasma deposited thin layers

Harm WulffUniversity of Greifswald Institute of Biochemistry


Introduction

The phenomena associated with plasma

surface interactions involve an interesting mix of plasma physics, ion-solid collision physics, surface chemistry, and materials science.The deposition of films by plasma techniques as well the plasma

treatment of solid walls are well known and widely used methods. Nevertheless, the fundamental mechanisms of

plasma wall interaction are not yet understood in detail. To understand the complexities involved in plasma wall interaction it is necessary to take a close look at the substrate area. The substrate area is subjected to plasma radiation and especially to fluxes of energetic and reactive charged and neutral particles in various excited states.Of

particular interest is the analysis of film properties in the nanometre range.Grazing incidence diffractometry

(GIXD) and X-ray reflectometry (XR), have been established as well-suiting tools for investigation of chemical, physical and crystallographic properties of thin films and surface layers. Besides the X-ray methods also other analytical techniques can be used to give

information on surfaces and deposited films. In our plasma-wall interaction studies we have used XPS for chemical analysis and AFM for surface morphology characterization. The properties of plasma deposited films and plasma treated surfaces decide finally whether and for which purposes films or surfaces can be used for special tasks. In this chapter the basic principles of the x-ray techniques will be presented, that are the grazing incidence

diffractometry

(GIXD) and the X-ray reflectometry (XR), which can be used successfully. By means

of examples the efficiency of the x-ray methods will be demonstrated. These involve the characterization of ITO films, deposited on Si(100), especially the influence of oxygen gas flow and substrate voltages and studies of the alumina (Al2

O3

) formation during microwave plasma treatment of aluminium-films.


Model of plasma-wall-interactions

desorption

defectsimplantation

sputtering

surface-reactions

neutral particle

ion

adsorption

isle

formationdiffusion

substrate

Plasma-surface-interactions


Plasma surface interaction

This model shows the interaction of plasma species with solid surfaces. Chemical reactions, diffusion processes, particle deposition, defect formation and sputter processes can take place and

show the very complex behavior during plasma wall interaction.


Film property

X-ray method Alternatives

phase composition GIXD:Bragg angle, intensity

TEM

chemical composition (concentration depth profile)

GIXD: Bragg angle EDX, XPS, RBS, ERD

macrostress GIXD: Bragg angle substrate curvature, laser optics

grain size GIXD: line profile, line width

TEM, SEM

microstrain GIXD: line profile

preferred orientation GIXD: intensity, polfigure

crystal structure GIXD:Rietveld analysis, structure refinement

thickness GIXD: intensity XR: Kiessig fringes

interferometry, ellipsometry, TEM

density XR: critical angle of total reflection

ellipsometry

surface roughness interface roughness

XR: amplitude of Kiessig fringes

SEM, ellipsometry, AFM

diffusion behavior in situ GIXD, thermal and time resolved: intensity

SIMS, AES, combined with sputtering

crystallization rate melting point

in situ GIXD, thermal and time resolved: intensity

Survey of X-ray and alternative methods


Table Survey of x-ray and alternative methods

This table shows important and fundamental film properties and the x-ray methods which can detect these characteristics.


Schematic diagram of grazing incidence X-ray diffractometry (GIXD), 2θ

Bragg angle, ω

angle of incidence

substrate

layerdiffraction angle 2Θ

divergence slitx-ray source

incidence angle ω

monochromator

detectorwindow

detector

SOLLER slit

GIXD, assymmetric

Bragg case


GIXD

Thin polycrystalline films can be studied with advantage in a highly asymmetric Bragg case.In this technique the diffraction volume can be increased by decreasing the angle of incidence.In the schematic diagram of grazing incidence x-ray diffractometry

the optical path in GIXDcan be seen.The very asymmetric Bragg case reflection with a small incidence

angle ω

as well as thetechnique with specularly

reflected x-rays are labeled “grazing incidence x-ray diffraction”. There is no clear conceptual separation between these two techniques in the literature.


Advantage of GIXD-geometry for thin film investigations compared to conventional Bragg-Brentano measurement method

2 (degrees)Θ

int e

n sit y

(cp

s)

20 30 40 50 60 70 800

500

1000

1500

Ti TiTi

Si substrate(I>4000)

GIXRD

Bragg-Brentano

Ti

Ti

GIXD and BB: x-ray pattern


X-ray pattern

The x-ray patterns of a 50 nm thick Ti layer on a Si(100) wafer measured in normal Bragg-Brentano geometry (BB) and in the asymmetric case (GIXD) are displayed in this figure. The reflection positions are equal in both techniques for polycrystalline films, but in the GIXD technique the substrate reflections are suppressed and the intensities of the Ti reflections strongly increase.


Information depth, asymmetric Bragg case

Information depth of Cu Kα

radiation depending on incidence angles, calculated for 0.05, 0.1 and 0.5µm thick In films

0 1 2 3 4 5

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.5 µm

0.1 µm

0.05 µm

incidence angle ω (degrees)

info

rmat

ion

dept

h T

(µm

)

The film information depth T

depends on the thickness x0

, the mean absorption coefficient µ

of the film material and the incidence angle ω

of the

X-ray

beam:

( ))exp(110 Zxµ

ZµT ⋅⋅−−

⋅=

with

)2sin(1

sin1

ωθω −+=Z


Information depth, asymmetric Bragg case:

In the analysis of layers, the information depth T

of x-rays is an important factor, in particular, if gradients of structure parameters occur in the films.

For indium films with a thickness of about 50 nm to 100 nm the influence of “omega”

is only small, but in 500 nm thick films the information depth can be ruled. In this case the prove of “gradients”

is possible.The film information depth T

strongly depends on the thickness x, the mean absorption coefficient µ

and of course on the incidence angle “omega”.

A further advantage of GIXD compared to the normally used BB geometry is, that the information depth is independent of the Bragg angle 2θ.


ωοE0

Bragg

planes

ΘB

diffracted

wave

EsEhωh ωο

transmittedwave

Specular

wave Specular

wave

GIXD: Bragg case, specularly

reflected

The wave Eh

is generated by the Bragg diffraction of the specularly

reflected wave Es

.The most important fact is that Eh

principally contains information on the structure of verythin surface layer.


GIXD, Bragg case, specularly reflected

For angles of incidence below the critical angle

ωc

the penetration depth perpendicular to the surface is in the order of nanometers, determined by the evanescence of the electrical field.

This GIXD is a scattering geometry combining the Bragg condition

with the conditions of x-ray total external reflection from crystal surfaces.


Penetration depth and reflectivity at grazing angles

0.0 0.5 1.0 1.5

103

GaAs / CuKα1

criti

cala

ngle

ωo

=|χ0

|1/2

pene

tratio

n de

pth

[A]

specularreflectivity

incidence angle [degree]

100

1

0.1

0.01

10-3

10-4

10-510

104

105


Penetration depth and reflectivity of x-rays at grazing incidence

This provides superior characteristics of GIXD as compared to the other diffraction schemes in the studies of thin surface layers, since penetration

depth of x-rays inside the slab is reduced by three orders of magnitude, typically from 1-10 µm (normal BB geometry) to 50-500 nm (asymmetric Bragg case) to 1-10 nm in the Bragg case, specularly

reflected.

λπ

ρσχω ⋅Σ

+Σ===

i

iA

AfZrN

2)(2 02/1

00


X-ray reflectometry

X-ray reflectometry around the critical angle of total reflection allows:•

determination of film thickness•

mass density•

surface and interface roughness•

irrespective of the crystalline structure.

X-ray reflectometry is equally well

applicable to crystalline, polycrystalline andamorphous materials

It only requires a sufficient flat sample


Schematic diagram of X-ray reflectometry

substrate

layerreflection

angle Θincidence

angle ω

divergence

slit

reflectedincident x-ray

x-raysource

detector

Geometry of XR


The basic principle is shown in this figure. In case of thin films on a substrate constructive interference occurs between the beam reflected at the surface and the beams reflected at the interfaces.

Constructive interference results in intensity maxima called “Kiessing

fringes”, whose angular spacing is characteristic for the thickness of a layer.

Geometry of X-ray reflectometry


XR: simulation silicon/aluminum/air

Simulation of the reflectivity of a 30 nm Al-layer on Si with different roughness σ1

(air-Al) and σ2

(Al-Si) : curve a: σ1 = σ2

= 0 nm;

curve b: σ1

=2 nm, σ2

= 0 nm;

curve c:

σ1

= 0 nm, σ2

= 2 nm; curve d: σ1

= σ2

= 2 nm.

0.5 1.0 1.5 2.0 2.5 3.0

10-7

10-6

10-5

10-4

10-3

10-2

10-1

100

2ΘC

d

b

ca

inte

nsity

2Θ (degrees)

distance between the interference fringes

film thickness x

angle of total reflection density of the layer ρ

decrease of intensitysurface roughness σ

attenuationinterface roughness σ


X-ray reflectometry simulation

Reflectometry simulations for a 30 nm thick aluminum layer on a silicon

substrate demonstrate the influence of various surface and interface roughnesses.

These simulations were made with the program LEPTOS from Bruker

AXS.

The value of the critical angle of total reflection 2Θc

can be used to determine the mass density ρ

of a deposited film.


Characterization of ITO films, deposited on Si

(100) substrates by means of dc-planar

magnetron sputtering.

Influence of oxygen gas flow and negative substrate voltages


dc-planar magnetron sputtering

bias-voltagesubstrate

Target

Experimental

sputtering-parameter

0 W, 350 V, 85 mA

Target metallic In/Sn (90/10) Substrate Si(100)-wafer

base

pressure

10-8

mbar

sputter

pressure

5,6•10-3 mbar

sputtering

gas 15 sccm argon

reactive gas 0 ... 2 sccm oxygensubstrate voltage 0, - 50, -100 VDeposition time 120 s


Equipment dc magnetron sputtering

The films were deposited by reactive dc magnetron sputtering.The target material was metallic In/Sn.The reactive gas was oxygen. The gas flux was varied between 0 and 2 sccm. Substrate voltages of 0 V, –50 V and –100 V were applied.


GIXD: phase

analysis

Diffraction patterns of ITO-films deposited at different oxygen flows, Usub

= 0 V

26 28 30 32 34 36 38 40

100

200

300

400

500

In(1

10)

In(0

02)

In(1

01)

USub= 0 V

2.0 sccm O2

1.5 sccm O2

1.0 sccm O2

0.5 sccm O2

0 sccm O2

2Θ (degrees)

inte

nsity

(a.u

.)


GIXD, phase analysis

During deposition, the oxygen flow influences chemical and phase

composition.Deposition without oxygen forms crystalline metallic In/Sn

films. No preferred orientation was observed. The intensity ratios are similar to those of the x-ray pattern of polycrystalline bulk In. With increasing oxygen flow only small amounts of crystalline metallic In/Sn

can be detected in the layers. Oxygen flows higher than 0.5 sccm prevent the formation of a crystalline phase.

A disadvantage of x-ray diffraction methods is the fact that only crystalline materials can be investigated. X-ray amorphous or amorphous materials do not give information on chemical or crystallographic properties.


GIXD: In situ phase analysis

26 28 30 32 34 36 38 40

20

40

60

80

100

120

0 sccm O2USub= 0 V

In (1

10) 3

In (0

02) 2

In2O

3 (40

0)3

In2O

3 (22

2)x

In (1

01) x

as-deposited

after 1h at 200°Cafter 4h at 200°C

after 7h at 200°C

after 12h at 200°C

2Θ (degrees)

intensity (a.u.)

Annealing

process

-

in situ

high temperature

diffractometryPhase transformation of metallic In/Sn

to crystalline ITO at 200°C


GIXD in situ phase analysis (annealing)

The films were deposited without substrate heating. Due to a post deposition heat treatment the phase transformation of metallic In/Sn

to crystalline Indium-tin-oxide can be observed.After an annealing time of 1h at 200°C no metallic In/Sn

is detectable with GIXD. The new peaks can be identified as reflections from the In2

O3

in the bixbiyte

structure type.

Post deposition treatment is often used to improve the crystallinity

and therefore the desired properties: high conductivity and high transparency in TCO films.


X-ray reflectivity measurements of films deposited at different oxygen flows, Usub

= 0 V: 0 sccm O2

, thickness 15.1 nm, roughness 1.52 nm;

1 sccm O2

, thickness 11.5 nm, roughness 0.98 nm; 2 sccm O2

, thickness 7.1 nm, roughness 0.75 nm

0,5 1,0 1,5 2,0 2,5 3,0103

104

105

106

107

108

0 sccm O2

1 sccm O2 2 sccm O2

Inte

nsity

(cps

)

2Θ (degrees)

XR: film thickness, density, roughness


XR: film thickness, density, roughness

Film thickness and roughness in dependence of the oxygen flow in

the deposition chamber are presented in the figure.The results are from x-ray reflectivity measurements. For these x-ray reflectometry measurements the deposition time was kept constant (30 s). Without oxygen (the black curve) the film is 15.1 nm, with 1 sccm 11.5 nm and with 2 sccm oxygen the thickness is 7.1 nm.

The thickness decreases with increasing oxygen flow and this also holds for the surface roughness of the films.


Gas composition influence film properties and film growth

0,0 0,5 1,0 1,5 2,0

4

5

6

7

8

0 V

dens

ity (g

cm

-3)

oxygen flow (sccm)

0,2

0,3

0,4

0,5

0,6

0,7

deposition rate (nm s

-1)

Density ( ) and deposition rate (•) of samples deposited at0 V substrate voltage vs. oxygen flow


Gas composition influence film properties and film growthdensity and deposition rate

From the x-ray reflectometry results film density and the deposition rate can be calculated. The dependence of film density and growth rate on oxygen flow in

the deposition chamber are presented in the next figure. Without oxygen the growth rate is 0.6 nm/s and decreases to 0.2 nm/s for an oxygen flow of 2 sccm. Simultaneously the layer density increases from 4.5 g/cm3

to 7.2 g/cm3. An oxygen flow higher than 1 sccm leads to film densities similar to bulk values of indium or indium oxide (7.28 und 7.12 g/cm3,

respectively)The small densities in the more metallic films suggest a high amount of voids in the these films. Increasing oxygen flows yield more compact layers. The drop of the roughness confirms this assumption. The assumption of voids in the more metallic films is also supposed by AFM images.


AFM images support the x-ray results

AFM micrographs, samples deposited (a) without O2

and (b) at 1.5 sccm O2

; Usub

= 0 V, deposition time 30 s

2000 nm

0 nm

1000 nm

2000 nm0 nm 1000 nm

2000 nm

0 nm

1000 nm

2000 nm0 nm 1000 nm

a) b)


AFM images support the x-ray results

AFM micrographs of these films clearly demonstrate that grains become smaller with increasing oxygen flow. The metallic film shows large grain sizes forming a rough film surface. Deposition with higher oxygen partial pressure causes a

smooth surface, grains are not clearly observable.

XPS measurements confirm the existence of oxygen besides In and Sn.

These experimental results show that the coatings become x-ray amorphous with increasing oxygen flow and that these amorphous layers contain the ITO (indium tin oxide) phase.


Diffraction patterns of films deposited at different oxygen flows, Usub

= -50 V

26 28 30 32 34 36 38 40

50

100

150

200

250

USub = -50 V

In(0

02)

In(1

01)

2.0 sccm O2

1.5 sccm O2

1.0 sccm O2

0.5 sccm O2

0 sccm O2

2Θ (degrees)

inte

nsity

(a.

u.)

Negative substrate voltage works like a reduced oxygen flow


Negative substrate voltage works like a reduced oxygen flow

To obtain further information on the effect of energy flux due to ion energy the substrate voltage U

was changed from 0 to –50 V or –100 V at the same oxygen flows. The additional energy flux to the growing films (due to increased ion energy) can be responsible for changes observed in diffraction patterns and also in film properties.

There are noticable

differences in the profiles and peak positions and therefore in the microstructure between the 0 V samples and films

deposited at –50 V.An increased amount of metallic In/Sn

in the films was detected in films deposited at –50 V substrate voltage although the supply of oxygen in the discharges remains the same as in 0 V experiments.


Substrate bias influences film properties and film growth

Density ( ) and deposition rate (•) of samples deposited at 0V, -50V and –100V substrate voltage vs. oxygen flow

0,0 0,5 1,0 1,5 2,0

4

5

6

7

8

-100 V-50 V0 V

dens

ity (g

cm

-3)

oxygen flow (sccm)

0,2

0,3

0,4

0,5

0,6

0,7

dep. rate (nm s

-1)


Substrate bias influences film properties and film growth

Film density and growth rate show a similar dependence on oxygen

flow as it was observed for films deposited without negative substrate voltage.However, the measured densities have smaller values, particularly in the middle field from 0.5 to 1.5 sccm and the films grown at 0.5 sccm to 2 sccm oxygen flow exhibit growth rate higher for –50 V and –100 V substrate voltage. From these results one can draw the conclusion that the negative

substrate voltage works like a diminished or reduced oxygen flow. This means that in the plasma are appreciable amounts of negative oxygen ions.


-50V

30 32 34 36 38

50

100

150

200

250

300

350

400

2Θ (degrees)

in

tens

ity (a

.u.)

0V

Lattice defects in thin surface layers influence x-ray data

X-ray pattern In/Sn:line shape and line position wereinfluenced by negative subtrate

voltage


Lattice defects in thin surface layers influence x-ray data

If we compare the indium x-ray profiles at 0 V substrate voltage and the x-ray profiles of a film deposited at –50 V substrate voltage we can observe that the peak profiles are broadened and the peak position is shifted to

higher angles due to the additional ion energy flux to the growing films.


Characterization of defect structures

by x-ray investigations

The line shift of broadened profiles is defined by the centre of gravity.

The shape analysis of diffraction peaks essentially comprises three problems: (I)

extraction of the pure physical line profile from the experimental profile,(II)

unraveling of the contributions of various types of lattice imperfections to the physical line profile and

(III)

quantitative estimation of substructure parameters.

Lattice defects in thin films can influence x-ray data. Defect structures are partially or wholly manifested by diffraction intensity, line shape, or the change in line shape with respect to the diffraction angle 2θ. Imperfections of the first type, such as point defects, displacement disorders or substitution disorders, shift the line position; imperfections of the second type such as domain sizes or dislocations act on

the

diffraction line shape.


X-ray profiles:

g(x) ideal, h(x)

with defects


X-ray profiles: g(x) ideal, h(x) with defects

The experimentally observable diffraction line profile of an x-ray reflection h(x)

is the convolution of a physical line profile f(y)

caused by lattice disorder of the second type and an instrumental line profile g(x).

X-ray profile analysisThe profile g(x)

was determined with a standard material that contains no defects, strain or particle size broadening. LaB6

, a SRM (standard reference material) from NIST was used. The apparatus function g(x)

is separated from the experimental measured profile h(x).

The applied method was the Stokes Fourier series.

The Fourier coefficients F(L)

of the resulting physical line profile f(y)

contain information on particle size P, mean strains due to internal stresses, which are constant within a crystallite or a subgrain

S, and restricted randomly distributed dislocations D.


X-ray

profile

analysis

∫ −∗= dyyxgyfxh )()()(

)()()(

LGLHLF =

STOKES method

F(L), H(L) G(L) are the Fourier Transforms of f(x), h(x) and g(x)normalized to

1)0()0()0( === GHF

and

)(hkldnL ∗=)(hklTT =

)(hklBB =0L

domain of definition of experimental line profile

effective particle size

mean total dislocation density

length proportional to the core radius r0

of the strain field of dislocation

mean square microstrain

due to internal stress>>=<< )(22 hklεε

)()()()( LALALALF dsp ∗∗=

20

2 )/ln(/)(ln LLLBKTLLF ⟩+><⟨+=− ε

))/ln(exp()exp()/exp()( 20

22 LLLBLKTLLF −∗><−∗−= ε


WARREN-AVERBACH-plot

KRIVOGLAZ-WILKENS-plot

LLKTLLF ><+=− )(/1/)(ln 2ε

LBLBKTLLLF ln)ln(/1/)(ln 022 −+><+=− ε

X-ray profile analysis: single line


X-ray profile analysis: single line

On the base of the Warren Averbach

theory or the Krivoglaz-Wilkens

theory microstrains

and dislocation densities can be calculated.


0 5 10 15 20 25 30 35 40 45 50 55 60

0,0

0,2

0,4

0,6

0,8

1,0

Fourier coefficients of the physical line profile

F(L)

nL

0 V -50 V -100 V

In/Sn

films: results of the line profile analysis (i)

Cosine Fourier coefficients of the evaluated physical line profiles


In/Sn films: results of the line profile analysis (i)

From the linear part of a plot Fourier coefficients versus the domain ofdefinition of the experimental line profile the mean domain sizes can be estimated. The differences in the curves depending on the substrate voltage

can be clearly seen.


0 10 20 30 40 50

0,02

0,03

0,04

0,05

0,06

0,07

0,08

- ln

[F(L

)]/nL

nL / nm

L = 3.07nm

-100V

0 V

-50V

WARREN-AVERBACH-plot

of In(101)-Reflection

In/Sn

films: results of the line profile analysis (ii)


In/Sn films: results of the line profile analysis (ii)

This is a typical WA plot. From the slope information on the mean micro strain can be obtained.

The KW-plot gives information on the dislocation density.


Physical parameters of In/Sn

films deposited at various substrate voltages

Usub 0 V -50 V -100 V T / nm 74 43 71 <ε2>1/2 1,87·10-3 2,14·10-3 1,98·10-3 ρV / cm-2 0,56·10-11 1,10·10-11 0,64·10-11 d101 / Å 2,718 2,714 2,717 ΔV / V reference -0,00455 -0,00149

Strong influence of negative substrate voltage on film microstructureIn the -50 V samples the crystallite growth is strongly disturbed, domain sizes are small and microstrain

and dislocation density are high. The peak shift to larger 2θ

values can attribute to contraction of the unit cell due to an increasing concentration of vacancies in these films.The defect concentration of the films deposited without negative

substrate voltage and at Usub

= -100 V are similar, but significantly smaller than

that of the -50 V samples.

The development of the defects in the -100 V samples is not quite clear. We assume that the increase energy flux increases the mobility of the In-atoms. Thus more often regular lattice positions can be occupied.


Study of Al2

O3

formation during microwave plasma treatment of Al films

in Ar-O-gas mixtures

The next example concerns the kinetics of aluminum oxide formation during a microwave plasma treatment of Al-films in different Ar/O2

gas mixtures and different plasma powers.


Motivation

I

Which influence do chemical reactive (O) or chemical non-reactive plasma (Ar) exert onto thin Al-layers (wall)?

II

How do plasma activated gases affect the structure and the composition of the coatings ?

III

How do plasma-activated species influence the kinetics of the formation or modification of the layers ?


schematic diagram of the plasma chamber

substrate holder

thin

Al films

microwaveincoupling

mass

flow

controller

Working pressure 4*10-1 mbar gas Ar/O2(1), Ar/O2(2), O2, gas flow 20 sccm microwave power 200 W, 700 W, 1100 W process time 10 min to 1 h

process

parameters

Oxygen partial pressure at total gas flow of 20 sccm

Ar/O2

(1) pO2

= 1.5x10-6

barAr/O2

(2) pO2

= 2.6x10-6 barO2 pO2

= 6.1x10-6

bar


schematic figure of the plasma chamber:

The Al-coatings used in this experiments were prepared by thermal evaporation of Al on Si

(100) wafers. The typical thickness of the Al-layers was determined by x-ray reflectometry to 30 to 60 nm. The plasma treatment experiments were carried out in microwave plasma chamber SLAN. The Al-samples were laid onto a substrate holder, which is centered in the quartz tube of the plasma reactor.Plasma power P, gas composition and exposure time were varied. The used gases were Ar/O2

(1), Ar/O2

(2) and oxygen.


Characterization of plasma treated Al-

coatings

•

Grazing Incidence X-ray Diffractometry (GIXD)

•

X-ray Reflectometry (XR)

-

interference of the beam reflected at the surface and the interface layer-substrate

-

distance between the interference fringes film thickness x- angle of total reflection density of the layer ρ- decrease of intensity surface roughness σ- attenuation interface roughness σ

•

X-ray Photoelectron Spectroscopy (XPS)

•

Fourier Transform Infrared Spectroscopy (FTIR)


25 30 35 40 45 50 55 60 65 70 75 800

20

40

60

80

100

120

140

(311)

(220)

(200)

(111)

ω = 1.0

ω = 0.5

inte

nsity

[a.u

.]

2 theta [°]

diffractogram

of an untreated Al-layer at different incident angles

GIXD: Phase analysis


GIXD: Phase analysis

The x-ray patterns of a 50 nm thick untreated Al-

layer at different incident angles confirm the existence polycrystalline al films. These patterns are typical for polycrystalline fcc

Al. There is no preferred orientation. In our first test with Ar/O2

-gas and different plasma power we could observe, that also in Ar/O2

-gas with only small amounts of oxygen the intensity of the Al reflection decreases, but the whole film thickness increases to more than 50 nm. Therefore we drew the conclusion, that also in Ar/O2

-gas mixtures with small oxygen concentrations chemical reactions take place.


Determination of the

partial pressure

of oxygen

with

a potentiometric

O2

-sensor

(ZIROX®)

pO2 decreases

withincreasing

microwave

power

reduced fraction of molecular O2 in plasma

Fraction

of activated

oxygen

ϕactive

at different powers

and gases

ϕactive (200 W) ϕactive (700 W) ϕactive (1100 W) Ar/O2-plasma (1)

0.00274 0.00313 0.00342

Ar/O2-plasma (2)

0.0364 0.0677 0.087

O2-plasma 0.0457 0.1501 0.2104

ϕactiveO without plasma O with plasma

total

p pp

=−

2 2, ,

0 200 400 600 800 1000 1200

0,0

1,0x10-6

2,0x10-6

3,0x10-4

4,0x10-4

5,0x10-4

6,0x10-4

O2-plasma Ar/O(2)-plasma Ar/O(1)-plasma

p O2 [b

ar]

microwave power [W]


By means of a potentiometric

oxygen sensor we could determine the partial pressure of oxygen for Ar/O2

(1), Ar/O2

(2) and oxygen gas. This potentiometric

oxygen sensor was now used to determine the fraction of activated oxygen species in the plasma for the different plasma gases and in dependence on plasma power.

Plasma off:The measured values are the normal oxygen partial pressures in the used gas (neutral O2

).

Plasma on:The values are the partial pressures of only molecular oxygen. The molecular oxygen partial pressure pO2

decreases with increasing microwave power. That means we have a reduced fraction of molecular oxygen

in the plasma. From the difference between partial pressure of molecular oxygen

without plasma and with plasma we can calculate the portion of activated oxygen species (atomic O, ions …) in the recipient.


35 36 37 38 39 40 41 42

20

40

60

80

100

140

9 h8 h

7 h6 h

5 h4 h

3 h2 h

1 horiginal

2 theta / °

int /

cps

formed

Al2

O3

is

x-ray amorphous quantitative

description was madeindirectly by determinationof the decrease of Al(111) integral intensity in combination with the total thickness of the film

Decrease

of the

integral intensity

of the

Al-(111)-reflection after

each

1h plasma

oxidation

(O2, 700 W)

Decrease of the Al integral intensity confirms the chemical reaction


Decrease of the Al integral intensity confirms the chemical reaction

The figure shows the decrease of the Al (111) peak intensity after each 1 hour of plasma exposure.

These values correlate very well with the results of the reflectometry measurements.


10-1

100

101

102

103

104

105

106

107

108

0.5 1.0 1.5layer thickness roughness dens. [nm] [nm] [g/cm3]Al2O3 2 1.7 2.9Al 49 0.7 2.7Al2O3 1 0 3.4 Si 0.3 2.3

as-deposited fit

inte

nsity

0.5 1.0 1.5 2.0

10-1

100

101

102

103

104

105

106

107

108

1h plasma oxidation fit

layer thickness roughness dens. [nm] [nm] [g/cm3]Al2O3 22 1.9 2.9Al 34 2 2.7Al2O3 1 1 3.0 Si 0.6 2.3

0.5 1.0 1.510-1

100

101

102

103

104

105

106

107

108


layer thickness roughness dens. [nm] [nm] [g/cm3]Al2O3 39 2.7 2.9Al 23 1 2.7Al2O3 1 0.6 3.0 Si 0 2.3

2 theta/°

0.5 1.0 1.5 2.010-1

100

101

102

103

104

105

106

107

108


layer thickness roughness dens. [nm] [nm] [g/cm3]Al2O3 41 2.4 2.9Al 18 3 2.7Al2O3 1 0.4 3.0 Si 0 2.3

X-ray reflectometry measurements

Overview of the Al samples, oxidized in plasma for 1h, 4h and 7h in comparison to the as-deposited layer. The values in the legend present the film parameters.


X-ray reflectometry measurements Reflectivity curves

Four reflectivity curves are shown in figure. The values in the legend demonstrate the film parameters obtained from the simulation with the simulation

program REFSIM. The best fit can be obtained with the assumption of four layers.

The two top layers describe the chemical reaction of Al (aluminum) to Al2

O3 (alumina). The whole thickness in the as deposited film is 51 nm, the whole thickness

after 1h plasma oxidation is 57 nm.

From the decrease of the Al layer thickness the converted part of Al film can be determined. From these values the maximum amount of stoichiometric

Al2

O3

which could have formed is easy to calculate. We only need the densities of Al and Al2

O3

from the x-ray reflectivity simulation.

x

is the film thickness, ρ are the densities, M

the Molecular weight.

The film thickness of Al follows for the supposed densities, and

with the molecular weight of Al and Al2

O3

, that 1 nm Al metal gives 1.75 nm Al2

O3

.

3232

32OAl

OAl

OAlal

Al

Al xM

xM

⋅=⋅ρρ


0 200 400 600 800 1000 12000

500

1000

1500

2000

2500

O KVV Auger

C 1s

O 1s

Al 2sAl 2p

inte

nsity

[a.u

.]

energy [eV]

4000 3500 3000 2500 2000 1500 1000 50093

94

95

96

97

98

99

100

original Al-layer

after 9x1h microwave plasma (O2)

trans

mis

sion

[a.u

.]

wave number [cm-1]

XPS spectrum

of an oxidized Al- layer

(9x1h, Ar-plasma)

FTIR-spectra,

band at 950 cm-1

corresponds to LO-vibration

of Al-O bonds

in case

of microwave induced

plasma

oxi-

dation, stoichiometric Al2

O3

will be

formed,

independent of plasma

gas composition

Characterization of amorphous films: XPS, FTIR


Characterization of amorphous films: XPS, FTIR

XPS and FTIR measurements confirm the formation of Al2

O3

. The composition of the oxide film, expressed as the O/Al atomic ratio, was found to be 1.5 from the total intensity of the O 1s main peak and the oxidic

Al 2p main peak of the XPS spectrum.

FTIR spectroscopy of plasma oxidized films also confirm the stoichiometry

to be Al2

O3

.


2000 nm

0 nm

1000 nm

2000 nm0 nm 1000 nm

2000 nm

0 nm

1000 nm

2000 nm0 nm 1000 nm

AFM -

images of Al2

O3 formed by plasma oxidation in Ar-microwave plasma

Al, as deposited P = 700 W

AFM images reveal changes in surface layers


AFM images reveal changes in surface layers

This figure shows the AFM images of Al2

O3

formed by plasma oxidation in Ar/O2

microwave plasma.The increase in the grain size of the upper layer after treatment in an Ar/O2

plasma for 9 hours at 700 Watt is obvious. Potentially the upper

atomic layers are still crystalline, but GIXD is not sufficiently sensitive to detect these small crystallites.


0 1 2 3 4 5 6 7 8 90

5

10

15

20

25

30

35

9x1h Ar-plasma, 1100 W, ω:1.0

fit functionx = 10.5 * √t - 1.5 * t +x0, Al2O3

calculated data-set fit experimental fit

oxid

e th

ickn

ess

x [n

m]

time t [h]

-

oxide growth follows -law diffusion determined process

-

simultaneously going sputter process limits the oxide formation

Calculated thickness of

formed alumina in comparison to the experimental determined ones.The difference is caused by a sputter process.t

t

formation of alumina

theoretical function

tbx =


Formation of alumina

In the x-ray pattern only reflections of the Al-film could be observed, indicating that the Al oxide formed was x-ray amorphous. From the measured thickness of the untreated Al-film (from x-ray reflectometry) and from the integral intensity of the Al (111) peak (GIXD) the thickness of residual Al was calculated after each oxidation step. Together with total thickness of the Al/Al2

O3

coating, this enabled us to estimate the produced amount of aluminum oxide (red curve). From the balance of the number of moles, with consideration of measured oxide densities (from x-ray reflectometry) it is possible to calculate the amount of oxide which is theoretically expected (black curve). The oxide film growth is clearly shown. Dry thermal oxidation by

heating a pure Al-metal surface to a maximum temperature of 573 K (300°C) without plasma effects solely the formation of the well known 2 nm native oxide layer. The fit-function for theoretical values of oxide thickness x

is x = b*sqrt

referring to a diffusion limited process. The deviation of the experimental results from the theoretical data-set is attributed to a linear sputter process. The experimental data follow the function x = b*sqrt

t-at, with b

the diffusion rate constant and a

the sputter rate.


mm xztxmA ),(e μ−=

( ) ( )( ) . 110

10

1

n

x

mnmnm

m

dxCC

xff μμμμμμ +−−+−= ∫

∫=d

mm dxAtxCAIKBtI0

m000 C),( )( ∂∂

∂∂

C x tt

DC x t

x( , ) ( , )

=2

2

C x tC n

n d xd

en

n

n Dtd

'

'( , ) ( )

cos0

1

1

2 121

4 12 1

2 12

2

2= −

−−

− −⎛⎝⎜

⎞⎠⎟

−

−

∞ −−⎛

⎝⎜

⎞⎠⎟

∑ππ

π

Fick´s

second law

0.0 0.5 1.0

dx

d

xm

0

C/C0

x

0.0 0.5 1.0

xm

0

d

Am

x

dxm

dxdV=Adx

substrate

w 2Θm

I0 dIm

layer

Determination of diffusion coefficients from time dependence of x-ray reflections


Determination of diffusion coefficients from time dependence of x-ray reflections

The basic idea is that the x-ray integral intensity of a volume element depends on the concentration ratio, which is a function of time. The mean absorption coefficient changes with plasma operation time due to chemical reaction.On the base of Fick´s

second law and a linear sputter process we have developed a mathematical model for calculation of the diffusion coefficient D

and the sputter coefficient S.


Diffusion coefficient

D

and sputter

rate S

in dependence

on the

incoupled

microwave

power

D

and S

increase

with

incoupled

microwave

power

as well as with

increasing

oxygen

in the

plasma.

200 400 600 800 1000 12000,0

1,0x10-16

2,0x10-16

3,0x10-16

4,0x10-16

5,0x10-16

6,0x10-16

7,0x10-16

O2-plasma Ar/O (2)-plasma Ar/O(1)-plasma

diffu

sion

s co

effic

ient

D [c

m2 /s

]

power [W]200 400 600 800 1000 1200

1,0x10-4

1,5x10-4

2,0x10-4

2,5x10-4

3,0x10-4

3,5x10-4

4,0x10-4

4,5x10-4

5,0x10-4

sput

ter r

ate

S [n

m/s

]

O2-plasma Ar/O (2) plasma Ar/O (1)-plasma

power [W]


D and S increase with microwave power as well as with increasing oxygen in the plasma

This figure shows the diffusions coefficients and the sputter rate S in dependence on the microwave plasma.At constant power D

increases as the amount of activated oxygen in the plasma is increased. D

also increases with increasing microwave power.Because of the effect of microwave power on the kinetic energy of the plasma particles an increasing number of defects is incorporated in the aluminum films and the oxide layers. This causes a subsequent penetration of activated oxygen

trough the oxide film to the Al-film coating and oxide can be produced more quickly.

The sputter rate S

depends on plasma power and plasma gas mixture. Both, the kinetic energy of plasma particles and the amount of activated oxygen, obviously influence the sputter process.


Arrhenius-plot: Determination of activation energy EA

Activation energies EA

point to a similar diffusion mechanism

EA

for thermally induced oxidation fcc

Al:

EA

= 131.2 kJ/mol

(lattice diffusion)

EA

= 69.1 kJ/mol

(grain boundary diffusion)

1,7 1,8 1,9 2,0 2,1 2,2 2,3 2,4 2,5-38,5

-38,0

-37,5

-37,0

-36,5

-36,0

-35,5

-35,0

EA =19.0 kJ/mol

EA =23,4 kJ/mol

EA=23,2 kJ/mol

O2-Plasma (700W) fit O2-Plasma (200W) fit Ar/O-Plasma (700W) fit

ln D

1/T [10-3K-1]

320 300 280 260 240 220 200 180 160 140

Θ [°C]


Activation energies EA point to a similar diffusion mechanism

The activation energy EA

for the diffusion process in microwave plasma was calculated in usual way from the Arrhenius

plot. The activation energies point to a similar diffusion mechanism irrespective of the microwave power and the used plasma gas mixtures. The values are

clearly smaller than the activation energy for thermal induced oxidation

of aluminum. For thermal induced diffusion in fcc

aluminum the activation energies are -131 kJ/mol (lattice diffusion) and -69 kJ/mol (grain boundary diffusion).The different reaction path compared to a thermal stimulated process can be assigned to atomic or ionic species even at comparatively low temperatures.

As expected, the sputter rates are independent of temperature in

the temperature range under investigation.


Summary•

thin Al-coatings were oxidized in a microwave plasma

•

kinetic description of the process was realized using grazing incidence x-ray techniques, additional diagnostics were XPS + FTIR

•

for the first time a different fraction of activated oxygen species in dependence on gas composition and power was measured using a potentoimetric oxygen-sensor

•

formed Al2

O3

is x-ray amorphous, quantitative description of oxide growth was made indirectly by the decrease of the reacted Al metal

•

the oxidation is a diffusion determined process, a sputter process limits the growth of Al2

O3

•

diffusion coefficients D

and sputter rates S

were simulated using a mathematical model considering the time-dependent decrease of integral intensity of Al-reflections

•

D

and S

depend on the fraction of activated oxygen species in the plasma

•

a model for the oxidation process of aluminium in a microwave discharge was deduced


Summary

•

Grazing incidence x-ray diffractometry

and x-ray reflectometry (in combination with XPS and AFM) were used to study non destructive the microstructure of thin films as well as the influence of plasma parameters on microstructure and film formation processes.

•

The examples were aimed to demonstrate the effectiveness of x-ray methods to investigate plasma treated surfaces

•

Four types of diagnostic methods have been discussed, which are

of interest to investigations of plasma treated surfaces

-

Phase analysis-

Defect structure analysis

-

Film formation analysis-

Study of kinetic processes

X-ray diagnostics of plasma deposited thin layers

Documents

Transcript of X-ray diagnostics of plasma deposited thin layers