Any correspondence concerning this service should be sent to the repository administrator:

staff-oatao@inp-toulouse.fr

Open Archive Toulouse Archive Ouverte (OATAO) OATAO is an open access repository that collects the work of Toulouse researchers and makes it freely available over the web where possible.

This is an author -deposited version published in: http://oatao.univ-toulouse.fr/ Eprints ID: 3802

To link to this article: DOI:10.1016/j.tsf.2005.08.335

URL: http://dx.doi.org/10.1016/j.tsf.2005.08.335

To cite this document : Lenormand, Pascal and Lecomte, A. and Babonneau, D. and Dauger, A. ( 2006) X-ray reflectivity, diffraction and grazing incidence small angle X-ray scattering as complementary methods in the microstructural study of sol–gel zirconia thin films. Thin Solid Films, vol. 495 (n° 1-2). pp. 224-231. ISSN 0040-6090

doi:10.1016/j.tsf.2005.08.335

X-ray reflectivity, diffraction and grazing incidence small angle X-ray

scattering as complementary methods in the microstructural study of

sol–gel zirconia thin films

P. Lenormand a,*, A. Lecomte b, D. Babonneau c, A. Dauger b

a Centre Interuniversitaire de Recherche et d_Ingenierie des MATeriaux. Laboratoire de Chimie des Materiaux Inorganiques et Energetiques,

UMR 5085-CNRS Universite Paul Sabatier- Bat. 2R1-118 route de Narbonne – 31062 Toulouse Cedex 4, Franceb Science des Procedes Ceramiques et de Traitements de Surface, UMR 6638-CNRS ENSCI, 47-73 av. A. Thomas, 87065, Limoges Cedex, France

c Laboratoire de Metallurgie Physique, UMR 6630-CNRS Universite de Poitiers- UFR Sciences- SP2MI,

Boulevard Marie et Pierre Curie BP 30179- 86 962 Futuroscope Chasseneuil Cedex, France

Abstract

X-ray reflectometry, X-ray diffraction and grazing incidence small angle X-ray scattering have been complementary used to fully characterize

zirconia (ZrO2) thin films obtained by the sol–gel route. The films were synthesized on various sapphire (Al2O3), silicon (Si) and glass mirror-

polished wafers by a dip-coating process in a zirconia precursor sol. Versus the synthesis parameters as alkoxide sol concentration, withdrawal

speed and annealing temperature, the microstructure of the layer is managed and its different microstructural parameters such as thickness, mass

density, crystalline phase, grain size and spatial arrangement have been determined. The as prepared layers are amorphous. During a thermal

treatment at low temperature (<1000 -C), the layers thickness decreases while their mass density increases. Simultaneously the zirconia precursor

crystallises in the zirconia tetragonal form and the coating is made of randomly oriented nanocrystals which self organise in a dense close-packed

microstructure. At low temperature, this microstructural evolution is similar whatever the substrate. Moreover, the layer evolves as the

corresponding bulk xerogel showing that the presence of the interface does not modify the thermal microstructure evolution of the layer which is

controlled by a normal grain growth leading to relatively dense nanocrystalline thin films.

Keywords: Sol–gel coating; Zirconium oxide; X-ray scattering; Nanostructure

1. Introduction

The elaboration of inorganic thin layers presents a growing

interest because a new thermal, mechanical, optical, chemical

or electrical property is easily added to the substrates on

which they are deposited [1]. Among all the methods of thin

film synthesis, chemical or physical vapour deposition,

pulverization, molecular beam epitaxy or laser ablation, the

interest of sol–gel route is unquestionable mainly because of

its low cost, its easiness to control the synthesis parameters

and the high precision in mixing different precursors for the

fabrication of multicomponents oxides layers. Moreover, the

sol gel route suits quite well to very large or complex form

substrates [2].

* Corresponding author.

E-mail address: lenorman@chimie.ups-tlse.fr (P. Lenormand).

In this processing method, the constitutive matter of the film

is globally deposited by either a dipping, spinning or spraying

coating process. Condensation and hydrolysis reactions occur

simultaneously to the excess solvent evaporation to form the

film. Then, an adapted thermal treatment is essential to decom-

pose the precursor film to the desired inorganic compound, re-

sulting in either a glass or a polycrystalline film with nm-sized

grains and pores [2]. In some cases, higher temperature heat

treatment leads the coating to break into more or less connected

islands and epitaxy phenomena could appear through abnormal

grain growth [3]. In any cases, the thermal evolution, the final

structure and microstructure of the layer are strongly related to

its initial thickness and to the nanostructural feature of the

precursor.

X-ray Diffraction (XRD) and Small Angle X-ray Scattering

(SAXS) are very widespread and efficient non destructive tools

to investigate the structure and microstructure of bulk as well

as powder materials. From the standpoint of thin film X-ray

characterisation, Grazing Incidence Small Angle X-ray Scat-

tering (GISAXS) is a more recent technique [4] which

combines X-ray reflectometry (XRR) and transmission mode

SAXS theory. This method is a powerful tool to analyse

nanoparticles supported on the substrate surface as well as

buried in the surface region or present in a thin coating and

provides valuable microstructural information on the nano-

metric scale such as particle shape and size, inter-particle

distance or correlation length. . .[4,5]. XXR and GISAXS

studies of sol gel oxide thin films, either alone or in

combination, have recently attracted considerable research

interest leading to a very rapid increase of the number of

published papers and more particularly in the field of sol gel

mesoporous thin films [6–8].

In the present work, we aim to show that the combination of

these three X-ray techniques, GISAXS, XRR and XRD allows

to fully characterize a sol gel oxide thin layer and to follow its

thermal evolution. They will be applied to the case of sol–gel

derived zirconia (ZrO2) thin films deposited on various

substrates by a dip-coating process, from the initial stage and

during low temperature treatments, pure or stabilized zirconia

being a widely used oxide in a number of electrical, optical or

superconductivity applications. Nevertheless, to date, a few

XRR and GISAXS studies of sol gel zirconia based thin films

have been reported [9]. Corresponding gels and xerogels are

also considered in a complementary way.

2. Experimental details

2.1. Sample preparation

Zirconia precursor sols are prepared in the zirconium n-

propoxide, n-propanol, acetylacetone (acacH) and water

system. In an air dried glove-box, the zirconium n-propoxide

is first diluted with n-propanol and chelated by acetylacetone

which allows to control the hydrolysis and condensation

reactions avoiding precipitation [10,11]. Then, a water-n-

propanol mixture is added under vigorous mechanical stirring.

The zirconium n-propoxide concentration C =[Zr(OC3H7)4] is

varied from 0.01 to 0.5 mol/L, the hydrolysis and complexing

ratios being kept constant, W=[H2O]/[Zr(OC3H7)4]=10 and

R = [acacH]/[Zr(OC3H7)4] =0.7, respectively. The sols are

homogeneous, clear and transparent.

The coating of zirconia precursor films was carried out just

after the sol synthesis using a dipping–withdrawing method

under ambient conditions. Silica based glass, silicon and

sapphire polish mirror single crystals were used as coating

substrates. The withdrawal speed was ranging from 1 to 200

mm/mn. After a drying at 60 -C during 20 min, the films are

continuous, homogeneous and crack free as observed by

optical microscopy.

The dried films were annealed in an electric furnace at a rate

of 5 -C/mn up to the desired temperature and for the chosen

annealing time. The first steps of the zirconia grain growth

were analysed from a single thin film undergoing a ‘‘cumula-

tive heat treatment’’, i.e., the firing of the sample was all

stopped at 100 -C from 60 up to 1000 -C, the sample was air

quenched then successively characterized by XRR, GISAXS

and XRD before being reintroduced in the furnace at the same

temperature for the continuation of the annealing.

2.2. X-ray experimental setups

The thickness, mass density and both surface and interface

roughness (r.m.s.), as defined by Nevot et al. [12], of zirconia

thin films were measured by X-ray reflectivity. The data were

collected by using an original angular dispersive reflectometer

[13] initially developed by Naudon et al. [14]. All reflected

beams were simultaneously recorded by using a linear position

sensitive detector (INEL LPS 50). As no movement of both the

sample and the detector was necessary during the measure-

ment, the exposure time to data recording was obviously

considerably diminished, typically about 3 h. The microstruc-

tural parameters were obtained by fitting the experimental data

using the well-known Parratt formalism [15]. For this purpose,

the layer is assumed to exhibit a constant mass density in its

whole thickness. After heat treatment, the coating being only

constituted of tetragonal zirconia nanocrystals and porosity,

this mass density of the layer is then derived from the XXR

curves by comparing the real and imaginary parts of the fitted

layer refractive index to the theoretical values for dense

tetragonal zirconia phase. Both nm-scale surface and interface

roughnesses are described by a Debye–Waller like damping

term as introduced by Nevot et al. [12] in the calculation of the

Fresnel coefficients.

The GISAXS measurements were performed by using a

conventional laboratory SAXS experiment, modified and

adapted to the case of grazing incidence [13]. The sample

holder was equipped with both two translations and one step by

step rotation with optical encoder which allowed a very precise

surface sample adjustment in the direct beam and the control of

the incidence angle. The scattered intensity was recorded using

a linear position sensitive detector (Elphyse) positioned parallel

to the surface sample. Its height position was precisely

controlled and adjusted around the reflected beam which is

masked by a lead beam stop set along the vertical axis.

The apparatus setting both to position the sample accord-

ingly to the X-ray beam and to determine the sample angular

origin was classically made [16]. To totally penetrate the

coating, the grazing angle of the incident beam, typically about

0.3-–0.4-, might be slightly larger than the layer critical angle

ac, ac being previously measured by X-ray reflectivity. So the

transmitted beam gone through the layer and acted as a primary

beam to give rise to a SAXS signal [4,17,18].

The scattered intensity I(q) is presented as a function of the

modulus of the scattering vector q, q =4ksin(h) /k where h is

half the scattering angle and k the wavelength of the incident

radiation (Cuka1 radiation). A sample-detector distance of 0.5

m, was used in order to cover a q-range from about 0.3 to 4

nm�1. The origin of the reciprocal space was calculated from

the refraction index of the coating as determined from the XRR

curves and then the spectra were corrected for refraction and

absorption effects before analysis [19]. Complementary bi-

dimensional GISAXS patterns have been collected with an

imaging plate detector by using the synchrotron radiation

facilities of LURE-DCI on D22 station [20].

X-ray diffraction was used to determine the crystalline

phases after heat treatment. The diffractometer consists in a

Debye–Scherrer like set-up, operating on flat samples, fitted

with a forward quartz monochromator (Cuka1 radiation) and a

Curve Position Sensitive Detector (CPSD INEL, 120- angularaperture) allowing the diffracted peaks to be simultaneously

recorded [21]. This asymmetric geometry under fixed inci-

dence is particularly suitable for surface characterization and

thin film structure analysis. The incidence angle was fixed to

about 4 to 7- in order to favour the layer irradiated volume

without impairing the resolution, i.e., without an excessive

broadening of the X-ray diffraction line profile.

3. Results

3.1. Thin film synthesis

Coatings deposited on glass, Si and Al2O3 single crystal

wafers have been synthesised from various alkoxide concen-

tration sols C, ranging from 0.01 to 0.25 mol/L, and

investigated by X-ray reflectivity, as prepared and after a heat

treatment at 600 -C during 30 min which allows the removal of

all the organics. In all cases, the reflected intensity curves

present oscillations (Kiessig fringes) which are significant of

continuous, homogeneous and uniform layers [22]. For

example, the experimental X-ray reflectivity curves obtained

from the annealed ZrO2 layers coated on Si substrate are

plotted in Fig. 1, the solid line corresponding to the best fit of

Inte

nsity

(a.

u.)

0.05 0.1C (mol/L)

0

0

20

40

60

80

100

120

140

0

100

10-1

10-2

10-3

10-4

10-5

10-6

10-70.01

30 mn – 600°C

0.5

0

20

40

60

80

100

120

140

0

Thi

ckne

ss (

nm)

0

20

40

60

80

100

120

140

0

Incidenc

Fig. 1. Zirconium alkoxide sol concentration dependence of the X-ray reflectivity cu

0.5 h. Curves are arbitrary vertically shifted for clarity. Experimental points (symbo

evolution of the thickness of the as prepared and annealed at 600 -C for 30 mn th

the Parratt model function. From these simulations, the layer

microstructural parameters, its thickness, mass density and

roughness have been deduced.

The thickness of the dried layers versus the alkoxide

concentration C varies linearly from about 7 to 120 nm for

C =0.01 and 0.25 mol/L, respectively (insert of Fig. 1). During

calcination, the films thickness decreases and evolves also

linearly from 3 to 45 nm for the previously quoted concentra-

tions, respectively. The XRD investigations showing the

annealed layers are crystallised in the tetragonal form of

zirconia (see below), the mass density of the film can be

deduced from the XRR curves. At 600 -C, the measured mass

density of the film is around 4I3I103 Kg/m3 and seems to be

independent of the alkoxide concentration. The root means

square roughness parameters, measured for each interface, are

lower than 1 nm and in the same range that the roughness of

the Si single crystal. So low r.m.s. values attest of the quality of

the sol–gel layers. Similar results are obtained on glass and

sapphire single crystal substrates. Finally, the thickness of

layers coated on glass substrates has been also varied by

managing the withdrawal speed in a 1 to 200 mm/mn range.

The film thickness versus the withdrawal speed evolves

according to a power law with an exponent equal to 0.64, a

value very close to the expected one, 2 /3, as predicted from the

Landau–Levich law [2]. We turn now to the thermal evolution

of these layers.

3.2. Thermal evolution of the thin films microstructure

The thermal evolution of the thin films microstructure has

been carefully investigated on a single thin film coated on the

0.25

1

0.05 0.1 0.15 0.2 0.25 0.3

as-prepared

30 mn ñ 600∞C

1.5

0.05 0.1 0.15 0.2 0.25 0.30.05 0.1 0.15 0.2 0.25 0.3

as-prepared

e angle (°)α

C (mol/L)

30 mn – 600°C

rves for thin films coated on Si wafer after a thermal treatment at 600 -C during

ls) are well fitted by using the recursive Parratt formalism (lines). In the insert,

in films versus the zirconium alkoxide sol concentration.

Inte

nsity

(a.

u.)

As prepared300°C

400°C

500°C

600°C

700°C

800°C

900°C

1000°C

100

10-1

10-2

10-3

10-4

10-5

10-6

10-7

10-8

10-10

Al2O3 wafer

10-9

10-11

0 0.5 1

Incidence angle (°)α

Fig. 3. X-ray reflectivity curves for a zirconia thin film coated on sapphire

wafer and annealed using a cumulative thermal heat treatment up to 1000 -C.

Curves are arbitrary vertically shifted for clarity.

most refractory and thermally stable substrate, i.e., Al2O3

substrate. The sol composition was C=0.25 mol/L, R =0.7 and

W=10 and the withdrawal speed was 200 mm/mn. Then, the

resulting sample has been annealed with the ‘‘cumulative heat

treatment’’ up to 1000-C, as described in the experimental part.

X-ray diffraction measurements show that the thermal

treatment induces, since 500 -C, the crystallisation of the

zirconia precursor in the tetragonal zirconia form. The relative

intensities of the peaks being in good agreement with JCPDS

files, the tetragonal zirconia grains are randomly oriented (Fig.

2). Increasing the temperature results in a progressive

apparition of the monoclinic zirconia phase.

The experimental X-ray reflectivity curves recorded at the

different temperatures and for the uncovered Al2O3 substrate

are plotted in Fig. 3. The intensity of the reflected beam by the

Al2O3 single crystal substrate shows a classical evolution of the

reflected intensity versus the incidence angle a for a dioptre

separating two media of different electronic densities. The

measured critical angle corresponding to total reflection, ac=0.291-, is in very good agreement with the theoretical value of

0.289- and the r.m.s. is about 0.6 nm. The reflectivity curve of

the dried layer exhibits a critical angle of about 0.22- and a

shoulder on a weak angular domain between 0.24- and 0.28-where kiessig fringes are also observed. The electronic density

of the layer is smaller than the one of the Al2O3 substrate, so

that, in this angular domain, the refracted X-ray beams being

propagated into the coating are in total reflection on the

substrate [13,23].

As the temperature increases, both the position of the critical

angle and the period of the interference fringes increase. The

corresponding thicknesses, mass densities of the films and

surface and interface roughness deduced from the simulations

with the Parrat model are displayed in Fig. 4.

The layer thickness brutally decreases from 130 to 60 nm

for the dried film and for the 500 -C point, respectively. This

behaviour is due to the removal of residual organics initially

contained in the coating. After 500 -C, the thickness continuesto slowly diminish while the mass density of the film increases.

This diminution of the thickness accounts for the increase of

the mass density of the coating. Indeed, this mass density

varies from 3.8.103 to 5.15.103 Kg/m3 corresponding to an

6030 40 50 8070

600°C

θ θ 60 8070

Inte

nsity

(a.

u.) 5

2

4

1

3

500°C

800°C

900°C

1000°C

2 (°)

103×

Fig. 2. X-ray diffraction patterns of a thin film coated on sapphire wafer and

annealed at various temperatures from 500 up to 1000 -C.

augmentation of 35% while the thickness decreases from 63 to

47 nm, i.e., a diminution of around 25%. So, the densification

of the layer depends only on the diminution of the thickness. At

1000 -C, the measured mass density of the film is about

5I15I103 Kg/m3, i.e., about 86% of the theoretical mass density

of tetragonal zirconia. Once again, the root means square

roughness parameters of the as prepared coating as well as

those of the annealed sample attest of the quality of the sol–gel

layers. Until 800 -C, the r.m.s. parameters, less than 1 nm for

each interface, are in the same range than the one of the Al2O3

single crystal. After 800 -C, both the surface and interface

roughness increase.

The Fig. 5 shows the one dimensional GISAXS patterns

recorded after heat treatments at the different temperatures. The

incidence angle of the X-ray beam a was always set to 0.4-with respect to the sample surface and reflectivity results so

that the whole layer was fully irradiated whatever its heat

treatment temperature, the detector being positioned around the

reflected beam.

Above 300 -C, the collected scattered intensity patterns

exhibit a peak at a non zero wave vector qm pointing out the

existence of nanometric heterogeneities in electron density of

the layer. When the temperature increases, the scattered

intensity peak height Im increases while its position qm moves

towards smaller and smaller q values. Such pattern feature

indicates the existence of a correlation length n, in the plane of

0

0,5

1

1,5

2

2,5

3

3,5

0 200 400 600 800 1000

Rou

ghne

ss (

nm)

Temperature (°C)

Mass density (K

g/m3)

Thi

ckne

ss (

nm)

Surface

Interface

0

20

40

60

80

100

120

140

0 200 400 600 800 10003

3.5

4

4.5

5

x 103

Fig. 4. Firing temperature dependence of the thickness, mass density, both surface and interface r.m.s. roughness as deduced from the X-ray reflectivity curves for a

zirconia thin film coated on sapphire wafer.

the coating, which can be deduced from the position of the

intensity maximum qm by n =2k/qm [24]. With the increase of

the temperature, this correlation length grows from 4 to about

15 nm for 300 and 800 -C, respectively.Since 500 -C, a temperature corresponding to the end of the

organics removal, the scattered intensity I(q) decreases, in the

large q range, according to a power law with an exponent near

to �4 (insert Fig. 5). This behaviour shows that the Porod_slaw is obeyed and points out that the scattering entities are

homogeneous with a well-defined smooth surface [24].

4. Discussion

The discrepancy between the experimental and calculated

XXR curves could be explained, on one hand, by the use of a

too simple model. Some attempts by using more complex layer

models to fit the XRR curves have been unsuccessfully done.

Moreover, we have no relevant indications from layer TEM

micrographs or other experimental techniques to justify such

complex models [13]. On the other hand, the observed

additional intensity on experimental XXR curves probably

rises from the X-ray scattering by the tetragonal zirconia

crystals inside the layers in agreement with the GISAXS results

(see below). Moreover, for the as–prepared layer which is only

composed of very small colloidal and polymeric species

without any porosity, we do not observe such additional

intensity [13]. Due to the reflectometer geometry, it is rather

difficult to take into account this additional X-ray scattering for

modelling the XXR curves. Nevertheless, in regards with the

angular range where it appears, we think that it has no major

impact on the measured mass density of the layer as well as its

thickness.

The grazing incidence X-ray scattering signal is assigned to

the appearance and growth of the zirconia nanocrystals. The

sharp interference peak is significant, on one hand, of a very

homogeneous and narrow zirconia nanocrystals sizes distribu-

tion and, on the other hand, that the zirconia nanocrystals self

organise in a dense close-packed microstructure in agreement

with the high mass density of the layer as measured by X-ray

reflectivity and cross-sectional TEM images [13]. In such

paracrystalline lattice model [25], the correlation length ncorresponds to the nearest-neighbour distance. At the end of

the organics removal, at about 500 -C, the zirconia nanocrys-

tals are obviously in contact and the nearest-neighbour distance

is assimilated to zirconia nanocrystals size.

These microstructural information are only valuable and

accurate in the plane of the layer in respect with the chosen

geometry and setting of the laboratory experiment, with the

linear detector parallel to the sample surface, which only

provide a section of the reciprocal space map, i.e., a one

Inte

nsity

(a.

u.)

q (nm-1)

700°C 800°C

log q (nm-1)

500°C 600°C 400°C

0

500

1000

1500

2000

2500

0

0

5

10

15

-0.8 -0.4 0 0.4 0.8 1.2

log

Inte

nsity

(a.

u.)

-4

1 2 3

Fig. 5. Grazing incidence small angle X-ray scattering patterns versus modulus

of the scattering vector q for a zirconia thin film coated on sapphire wafer and

annealed using a cumulative thermal treatment up to 1000 -C. In the insert,

these curves are plotted on a log– log scale with arbitrary vertical shifts for

clarity.

2

Inte

nsity

(a.

u.)

q (nm-1)

Lure D22

Laboratory

30 mn – 600 °C

0

100

200

300

400

500

600

0 0.5 1 1.5

0.1 Å-1

Fig. 6. Bi-dimensional GISAXS pattern recorded at ID22 of LURE for a

zirconia thin film coated on sapphire wafer and annealed at 600 -C during 0.5

h. One-dimensional GISAXS pattern as derived from the bi-dimensional map

by integrating the scattered intensities on 30- angular domains centred around

the diagonals (full line) and the one-dimensional GISAXS pattern collected on

the laboratory experiment on the same sample (symbols) versus modulus of the

scattering vector q.

dimensional GISAXS pattern. To recover the out of plane film

microstructure information, the detector has to be vertically

shifted to build up step by step the bi-dimensional map which

is very time consuming. To overcome this drawback, a bi-

dimensional GISAXS pattern has been performed at the

synchrotron radiation facilities of LURE on an analogous

sample, elaborated with the same Al2O3 substrate, sol

composition (C =0.25 mol/L, R =0.7, W=10), withdrawal

speed (200 mm/mn) and only differs from its thermal treatment

which was a single heat treatment at 600 -C for 30 mn.

The bi-dimensional GISAXS pattern was collected by

setting the incidence angle of the X-ray beam a to 0.4- with

respect to the sample surface and reflectivity results (a >ac).

The scattering map presents a quasi-circular half ring centred

on the origin of the reciprocal space indicating that the dense

close-packed zirconia nanocrystals microstructure is isotropic

inside the whole thin film (Fig. 6). Moreover, a one-

dimensional scattering curve I(q) has been extracted by

integrating the scattered intensities on 30- angular domains

centred around the diagonals of the scattering map and

compared to the one-dimensional laboratory GISAXS pattern

carried out on the same sample. After an adjustment of

scattering intensities, the remarkable agreement between the

two curves validates the laboratory experimental set-up. In

respect with the isotropic spatial arrangement of the zirconia

nanocrystals inside the layer and the low temperature heat

treatments, the one dimensional GISAXS measurements are

appropriated to accurately recover the parameters of prime

importance we are interesting in which are the zirconia

nanocrystals size and spatial arrangement.

The comprehension of the thin film formation and its

thermal transformation is very closely related to the nature of

the sol precursor from which it rises. In this aim, the sol to gel

transition of a sol with the same composition as the one used

for the film synthesis, C =0.25 mol/L, R =0.7, W=10, has been

previously studied by small angle X-ray scattering [26]. The

gel structure is made up of connected fractal clusters of nearly

equal size, around 50 nm, with an apparent fractal dimension of

1.7, resulting from the aggregation of very small zirconium rich

primary particles about 2–3 nm readily formed just after the

water is added. When this gel is conventionally dried, the

clusters fractal structure collapses under very strong capillary

force leading to a very important shrinkage. Throughout this

microstructure collapse, the scattering intensity totally disap-

pears [13]. During subsequent heat treatment at low tempe-

rature, the bulk xerogel crystallises under the tetragonal

ZrO2 / Glass

ZrO2 / Al2O3

ZrO2 xerogel

0

200

400

600

800

1000

1200

0 0.5 1 1.5 2 2.5

1

2

3

4

5

6

7

8

9

10

-1 -0.5 0 0.5 1

-4

Inte

nsity

(a.

u.)

q (nm-1)

log q (nm-1)

30 mn – 600°C

log

Inte

nsity

(a.

u.)

Fig. 7. Transmission mode small-angle X-ray scattering curve of bulk zirconia

xerogel annealed at 600 -C during 0.5 h and GISAXS curves of ZrO2 layers

coated on both Al2O3 and glass substrates after thermal treatment at 600 -Cduring 0.5 h versus modulus of the scattering vector q. In the insert, these

curves are plotted on a log– log scale with arbitrary vertical shifts for clarity.

zirconia phase and the SAXS signal in the transmission mode

is recovered and exhibits a well defined scattered intensity peak

at a non zero scattering vector.

During the dipping, the liquid film runs out on the substrate,

adheres to its surface and the evaporation of solvents leads to

its rapid solidification. The as prepared layer could be

considered as a xerogel obtained by a concentration effect.

As for the bulk xerogel, we do not observe GISAXS signal for

this raw layer even by using the synchrotron radiation facilities

of LURE. At the beginning of the annealing, the correlation

length corresponding to the appearing zirconia nanocrystals,

about 4 nm at 300 -C, is in good agreement with the zirconium

rich primary particle size present in the sol. During the

temperature increase, the layer microstructure evolves by

normal grain growth of the zirconia nanocrystals in agreement

with the theoretical approach developed by Thompson [27].

For a given low temperature thermal treatment and whatever

the substrate, the thin film GISAXS as well as the bulk xerogel

SAXS patterns display quite similar feature in spite of very

different experimental data recording (Fig. 7). At this stage, the

layer evolves as bulk xerogel and the film-substrate interface

does not seem to influence its microstructural evolution. These

results are in good agreement with the observations carried out

on similar systems [28].

The homogeneous in situ crystallisation of the zirconia

precursor leads to relatively dense layer, up to 86% of the

theoretical mass density of tetragonal zirconia at 1000 -C, inspite of the low temperature applied heat treatments with

respect to the melting temperature of ZrO2 (¨2680 -C). Thelayer densification occurs through a rapid decrease of its

thickness in a similar way as mentioned by Rizzato et al. about

the zirconia film evolution obtained from a zirconium chloride

[29]. After 800 -C, the slight increase of both the surface and

interface roughness is due to the zirconia grains size which

reaches the same order than the film thickness. With regard to

the thin film quality and more particularly to interface and

surface states, the sol–gel thin film synthesis appears to be a

valuable alternative route to the physical processed layers

[30,31].

5. Conclusion

The complementarity of non-destructive experimental tech-

niques such as XRR, XRD and GISAXS, has been successfully

used to fully characterise the microstructural evolution of ZrO2

based sol–gel coatings during thermal treatments at low

temperature.

Whatever the substrate (glass, Si and Al2O3 single crystals),

the layers, synthesized by a dip-coating process, are continu-

ous, homogeneous, crack free and exhibit very good quality of

both surface and interface states. Their nanometric thicknesses

are easily managed through the alkoxide sol concentration as

well as the withdrawal speed.

During the thermal treatment at low temperature, the removal

of the organic residues is concomitant to the crystallisation of

the amorphous film which is transformed in a polycrystalline

layer with a random orientation of tetragonal zirconia nano-

crystals. The densification of the layer, up to 86% of the

theoretical tetragonal zirconia mass density at 1000 -C, occursby a one dimensional shrinkage, i.e., by a strong reduction of

their thickness. Due to their narrow size distribution, the

tetragonal zirconia nanocrystals are self organised in a dense

close-packed microstructure which is conserved and evolves by

a normal grain growth mechanism up to 1000 -C.In this low temperature range and whatever the substrate,

the microstructure of zirconia thin film evolves as

corresponding zirconia bulk xerogel. At this stage, the layer–

substrate interface does not seem to influence the microstruc-

ture of the thin film.

References

[1] S.A. Chambers, Surf. Sci. Rep. 39 (2000) 105.

[2] C.J. Brinker, G.W. Scherer, Sol–Gel Science: The Physics and Chemistry

of Sol–Gel Processing, Academic Press, New York NY, 1990.

[3] A. Boulle, O. Masson, R. Guinebretiere, A. Dauger, Thin Solid Films 434

(2003) 1.

[4] J.R. Levine, J.B. Cohen, Y.W. Chung, P. Georgopoulos, J. Appl. Cryst. 22

(1989) 528.

[5] J.R. Levine, J.B. Cohen, Y.W. Chung, Surf. Sci. 248 (1991) 215.

[6] S. Dourdain, J.F. Bardeau, M. Colas, A. Mehdi, B. Smarsly, B. Ocko, A.

Gibaud, Appl. Phys. Lett. 86 (2005) 113108.

[7] A. Gibaud, S. Dourdain, O. Gang, B.M. Ocko, Phys. Rev. B. 70 (2004)

161403 (R).

[8] D. Grosso, F. Cagnol, J. Galo, de A.A. Soler-Illia, E.L. Crepaldi, H.

Amenitsch, A. Brunet-Bruneau, A. Bourgeois, C. Sanchez, Adv. Funct.

Mater. 14–4 (2004) 309.

[9] A.P. Rizzato, C.V. Santilli, S.H. Pulcinelli, A.F. Craiecich, J. Appl. Cryst.

36 (2003) 736.

[10] R. Guinebretiere, A. Dauger, A. Lecomte, H. Vesteghem, J. Non-Cryst.

Solids 147–148 (1992) 542.

[11] M. Chatry, M. Henry, J. Livage, Mater. Res. Bull. 29–5 (1994) 517.

[12] L. Nevot, B. Pardo, J. Corno, Rev. Phys. Appl. 23 (1988) 1675.

[13] P. Lenormand, Thesis, Etude de l_evolution microstructurale de

precurseurs d_oxyde de zirconium a l’etat de gel, xerogel, couche

mince et aerogel par diffusion de rayons X, University of Limoges,

France, 2001.

[14] A. Naudon, J. Chihab, P. Goudeau, J. Mimault, J. Appl. Cryst. 22 (1989)

460.

[15] L.G. Parratt, Phys. Rev. 95–2 (1954) 359.

[16] A. Naudon, D. Thiaudiere, J. Appl. Cryst. 30 (1997) 822.

[17] A. Naudon, D. Babonneau, Z. Metallkd. 88 (1997) 596.

[18] D. Babonneau, A. Naudon, T. Cabioc’h, O. Lyon, J. Appl. Cryst. 33

(2000) 437.

[19] B. Kutsch, O. Lyon, M. Schmitt, M. Mennig, H. Schmidt, J. Appl. Cryst.

30 (1997) 948.

[20] J.M. Dubuisson, J.M. Dauvergne, C. Depautex, P. Vachette, C.E.

Williams, Nucl. Instrum. Methods A 246 (1995) 636.

[21] O. Masson, R. Guinebretiere, A. Dauger, J. Appl. Cryst. 29 (1996) 540.

[22] H. Kiessig, Ann. Physik. 10 (1931) 769.

[23] Y. Ijdiyaou, K. Hafidi, M. Azizan, E.L. Ameziane, A. Outzourhit, L.

Dreesen, K. Benai, Sol. Energy Mater. Sol. Cells 52 (1998) 27.

[24] O. Glatter, O. Kratky, Small Angle X-ray Scattering, Academic Press,

New York NY, 1982.

[25] H. Matsuoka, H. Tanaka, T. Hashimoto, N. ISE, Phys. Rev. B 36–3

(1987) 1754.

[26] A. Lecomte, P. Lenormand, A. Dauger, J. Appl. Cryst. 33 (2000) 496.

[27] C.V. Thompson, Annu. Rev. Mater. Sci. 20 (1990) 245.

[28] M.G. Cain, F.F. Lange, J. Mater. Res. 9–3 (1994) 674.

[29] A.P. Rizzato, C.V. Santilli, S.H. Pulcinelli, J. Non–Cryst. Solids 247

(1999) 158.

[30] V. Holy, J. Kubena, I. Ohlidal, K. Lischka, W. Plotz, Phys. Rev. B 47–23

(1993) 15896.

[31] M. Jergel, V. Holy, E. Majkova, S. Luby, R. Senderak, J. Appl. Cryst. 30

(1997) 642.