Optimization of nanostructured oxide

Transcription

N° d’ordre
Année 2010
Thèse
Optimization of nanostructured
oxide-based powders
by
surface modification
Présentée devant
L’Institut National des Sciences Appliqués de Lyon
et
il Politecnico di Torino
pour obtenir
le grade de docteur
École doctorale: Matériaux de Lyon
Spécialité : Génie des Matériaux
par
Fernando Lomello
Jury
BONELLI Barbara
FANTOZZI Gilbert
LERICHE Anne
PALMERO Paola
REVERON Helen
Chercheur (Politecnico di Torino)
Professeur Emérite (INSA de Lyon)
Professeur (Université des Valenciennes et du Hainaut Cambrésis)
Chercheur (Politecnico di Torino)
Chargée de Recherche EPST - CNRS (INSA de Lyon)
Laboratoires de recherche
Matériaux: Ingénierie et Science (MATEIS) de l’INSA de Lyon
Dipartimento di Scienza dei Materiali e Ingegneria Chimica (DISMIC) del Politecnico di Torino
INSA Direction de la Recherche
2007-2010
SIGLE
ECOLE DOCTORALE
- Ecoles Doctorales – Quadriennal
NOM ET COORDONNEES DU RESPONSABLE
M. Jean Marc LANCELIN
Université Claude Bernard Lyon 1
Bât CPE
43 bd du 11 novembre 1918
M. Jean Marc LANCELIN
69622 VILLEURBANNE Cedex
Tél : 04.72.43 13 95 Fax :
[email protected]
Insa : R. GOURDON
ELECTRONIQUE,
M. Alain NICOLAS
ELECTROTECHNIQUE, AUTOMATIQUE Ecole Centrale de Lyon
E.E.A.
http://www.insa-lyon.fr/eea
Bâtiment H9
M. Alain NICOLAS
36 avenue Guy de Collongue
69134 ECULLY
Insa : C. PLOSSU
Tél : 04.72.18 60 97 Fax : 04 78 43 37 17
[email protected]
[email protected]
Secrétariat : M. LABOUNE
Secrétariat : M.C. HAVGOUDOUKIAN
AM. 64.43 – Fax : 64.54
M. Jean-Pierre FLANDROIS
EVOLUTION, ECOSYSTEME,
MICROBIOLOGIE, MODELISATION
E2M2
CNRS UMR 5558
http://biomserv.univ-lyon1.fr/E2M2 Université Claude Bernard Lyon 1
Bât G. Mendel
M. Jean-Pierre FLANDROIS
69622 VILLEURBANNE Cédex
Insa : H. CHARLES
Tél : 04.26 23 59 50 Fax 04 26 23 59 49
06 07 53 89 13
[email protected]
M. Didier REVEL
INTERDISCIPLINAIRE SCIENCESHôpital Cardiologique de Lyon
SANTE
EDISS
Bâtiment Central
28 Avenue Doyen Lépine
Sec : Safia Boudjema
M. Didier REVEL
69500 BRON
Tél : 04.72.68 49 09 Fax :04 72 35 49 16
Insa : M. LAGARDE
[email protected]
M. Alain MILLE
INFORMATIQUE ET
INFOMATHS MATHEMATIQUES
Université Claude Bernard Lyon 1
http://infomaths.univ-lyon1.fr
LIRIS - INFOMATHS
M. Alain MILLE
Bâtiment Nautibus
Secrétariat : C. DAYEYAN
Tél : 04.72. 44 82 94 Fax 04 72 43 13 10
[email protected] - [email protected]
MATERIAUX DE LYON
M. Jean Marc PELLETIER
INSA de Lyon
Matériaux
MATEIS
M. Jean Marc PELLETIER
Bâtiment Blaise Pascal
7 avenue Jean Capelle
69621 VILLEURBANNE Cédex
Secrétariat : C. BERNAVON
Tél : 04.72.43 83 18 Fax 04 72 43 85 28
83.85
[email protected]
MECANIQUE, ENERGETIQUE, GENIE
M. Jean Louis GUYADER
CIVIL, ACOUSTIQUE
MEGA
INSA de Lyon
Laboratoire de Vibrations et Acoustique
M. Jean Louis GUYADER
Bâtiment Antoine de Saint Exupéry
25 bis avenue Jean Capelle
Secrétariat : M. LABOUNE
Tél :04.72.18.71.70 Fax : 04 72 43 72 37
PM : 71.70 –Fax : 87.12
[email protected]
ScSo*
M. OBADIA Lionel
ScSo
Université Lyon 2
M. OBADIA Lionel
86 rue Pasteur
69365 LYON Cedex 07
Tél : 04.78.69.72.76 Fax : 04.37.28.04.48
Insa : J.Y. TOUSSAINT
[email protected]
*ScSo : Histoire, Geographie, Aménagement, Urbanisme, Archéologie, Science politique, Sociologie, Anthropologie
CHIMIE
CHIMIE DE LYON
http://sakura.cpe.fr/ED206
2
Acknowledgements
I would like thank specially to Dr. Paola Palmero, Dr. Barbara Bonelli, Prof. Gilbert Fantozzi,
Prof. Edoardo Garrone and Prof. Laura Montanaro for giving me the opportunity to work with them
and for taking care of my personal development. During the past three years, in both groups I grew
up scientifically and as a person, as well. Many people in the three laboratories encouraged me
directly and indirectly, as Prof. Jean-Marc Tulliani, with whom I share the same special passion for
Ferrari and motorsports.
I also want to thank to my colleagues from the laboratories who supported me and helped me in
many different situations. I am grateful to Mr. Yann Aman, Mrs. Mira Jaafar, Mrs. Mirella Azar &
Mr. Aurélien Pelletant, who made me feel at home while I stayed in Lyon. Also, I do not want forget
about Mrs. Liliane Quillot from INSA-Lyon who always found accommodation for me there.
Finally, I want to thank my family and my friends for their unconditional support during these three
years.
3
INDEX
INTRODUCTION
SECTION 1: Surface modification and characterization of
nanostructured ceramic powders in the attempt of improving
sinterability and microstructural tailoring
1. An introduction to ceramic nanopowders
1.1.
Definition and properties of nanopowders
1.2.
Main issues connected with the use of ceramic nanopowders
1.2.1.
Agglomeration and Dispersion
1.3.
Green Density of Nanopowders
1.4.
Nanosintering
1.4.1.
Pore size and its effects during densification behaviour
1.4.2.
Grain Growth
1.5.
Nanoaluminas
1.5.1.
Metastability of Transition Aluminas
1.5.2.
Vermicular Growth & Seeding
1.6.
Models developed so far to study the transformation kinetics of aluminas
1.7.
Sintering kinetics studied by stepwise isothermal dilatometry (SID method)
1.8.
Spectroscopic study of the surface of aluminas
1.8.1.
Vibrations of the solid
1.8.2.
Surface Vibrations
1.8.3.
Surface Hydroxyls Groups
1.8.4.
The possible role of defective crystal configurations
1.8.5.
The basicity of aluminas
1.8.6.
The acidity of aluminas
1.8.7.
Ammonia adsorptions
1.8.8.
The adsorption of carbon monoxide
2. Surface modification of a transition alumina
2.1.
Textural characterization of the starting material
2.2.
Effect of dispersion on powder granulometry
4
2.2.1.
Effect of the pH of the aqueous suspension on the agglomerate size
2.2.2.
Comparison among the other dispersion routes
2.2.3.
Effect of the dispersion on the specific area
2.2.4.
Effect of dispersion on powder composition and evolution
2.2.5.
Influence of the dispersion route on the phase evolution
2.2.6.
Effect of the dispersion on the kinetics of transformation
2.2.7.
Influence of the dispersion route on forming and sintering
2.2.8.
Investigation on Sintering Kinetics of Nanotek powders
2.2.9.
Influence of the dispersion route on the final microstructure
2.3.
Study of the effect of powder dispersion on its surface properties by means of IR
2.3.1.
IR spectra of samples outgassed at increasing temperatures
2.3.2.
Adsorption of CO at nominal 77 K
2.3.3.
Adsorption of CO 2
2.3.4.
Adsorption of NH 3
2.4.
Conclusions of the first part and perspectives
SECTION 2: Nanostructural materials for extreme applications
3. Nanostructured composite materials: elaboration and properties
3.1.
Nanocomposites materials: main classification
3.2.
Synthesis of composites powders
3.2.1.
Physical Methods
3.2.1.1.
Vapour condensation methods
3.2.1.2.
Spray Pyrolysis
3.2.1.3.
Thermochemical/flame decomposition of metalorganic precursors
3.2.2.
Chemical Methods
3.2.2.1.
Sol-Gel technique
3.2.2.2.
Reverse microemulsions/micelles method
3.2.2.3.
Precipitation from solutions
3.2.2.4.
Chemical synthesis of pre-ceramic polymers coupled with physical processing
techniques
3.2.2.5.
Colloidal processing route
3.2.2.6.
Mechanochemical Synthesis
3.3.
Forming and sintering
3.3.1.
Dry-Pressing
3.3.2.
Slip-Casting
3.3.3.
Sintering
5
3.3.3.1.
Hot-Pressing
3.3.3.2.
Hot Isostatic Pressing
3.3.3.3.
Spark-Plasma Sintering
3.3.3.4.
Microwave Sintering
3.4.
The interest nanocomposites ceramics: the mechanical properties
3.4.1.
Role of the second phase on the retention of the matrix grain growth
3.4.2.
Hardening
3.4.3.
Change in fracture strength
3.4.4.
The strength
3.4.4.1.
Reduction in processing flaw size
3.4.4.2.
Dislocation networks
3.4.4.3.
Crack healing
3.4.5.
Toughening mechanism
3.4.5.1.
Intrinsic Fracture Energy
3.4.5.2.
Crack Bowing
3.4.5.3.
Average Internal Stresses
3.4.5.4.
Toughening by transformation
3.4.6.
3.5.
Crack Deflection
Creep of particle reinforced composites
3.5.1.
Diffusion Creep
3.5.2.
Particle reinforced composites
3.6.
The Alumina-YAG system: elaboration and properties
3.6.1.
Alumina-Yttria phase diagram
3.6.2.
Mechanical properties at room temperature and high temperature
4. Elaboration and Characterization of Al 2 O 3 -5 vol.% YAG Nanocomposites
4.1.
Characterization of the as-received powders
4.2.
Elaboration of Al 2 O 3 -YAG composites
4.2.1.
Dispersion and Spray-drying
4.2.2.
Thermal Evolution
4.3.
Forming and Sintering
4.3.1.
Sintering behaviour followed by dilatometric analysis
4.4.
Microstructural Characterization
4.5.
Mechanical Characterization
4.5.1.
Mechanical properties at room temperature
4.5.2.
Mechanical properties at high temperature
4.5.2.1.
Creep of TM and CR
4.5.2.2.
Creep of Y-TM and Y-CR nanocomposites
4.5.3.
Activation energy
4.5.4.
Microstructural observation after creep tests
6
4.6.
Conclusions of the second part and perspectives
References
Appendix 1: Characterization techniques
1. Laser Granulometry
2. X-ray Diffraction (XRD)
3. Thermal Analysis: DTA-TG
4. Dilatometry
5. Scanning Electron Microscopy/Environmental Scanning Electron Microscopy
6. High Resolution Transmission Electron Microscopy (HR-TEM)
7. Specific Surface Area
8. Young’s Modulus Resonant Frequency Meter: Grindosonic
9. Hardness tests
10. Creep tests
7
Introduction
Conventional ceramic materials are widely used today in different fields: from structural to
biological applications and in devices such as lasers, semiconductors and piezoelectric
components. Such materials include oxides, carbides, nitrides, mixed oxides and composites, as
well.
Nanoceramics is one of the fields in which nanoscience and nanotechnology have shown a
remarkable progress in producing a variety of advanced materials with unique properties and
performance, in terms of chemical inertness, strength, hardness and high-temperature stability.
Nanoceramics is a term used to refer to ceramic materials fabricated from ultrafine particles, i.e.
with grain sizes less than 100 nm in diameter. In this field, a great deal of research has been
accomplished in the last 20 years and has resulted in significant outcomes that are of great impact
academically as well as industrially.
Nowadays, certain types of nanopowders are produced on an industrial scale. However, it is
difficult to produce fully dense parts that retain a nanocrystalline grain size. For this reason, most of
the research on nanoceramics has been performed on composites with microcrystalline matrices
and nanocrystalline second phases. These type of composites have been extensively studied
during the last years.
The aim of this thesis is to improve the physico-chemical properties of commercial alumina
nanopowders (NanoTek® transition alumina and Taimei α-alumina) in order to develop fully dense
materials with tailored microstructure.
This thesis is divided into four chapters. The first two chapters deal with the study of surface
modification of a transition alumina nanopowder and the last two chapters are devoted to the
production
and
mechanical
characterization
of
Al 2 O 3 /Y 3 Al 5 O 12
(Al 2 O 3 /
YAG)
micro/nanocomposites for extreme applications.
In the first chapter, the state-of-art is reported of the main techniques of powders modification by
mechanical means, with the aims to lower powder agglomeration and improve sinterability without
inducing a relevant grain growth during firing. Some of the most important issues are reviewed
such as agglomeration, nanosintering, principal issues regarding transition aluminas, models for
the study of the transformation kinetics, the description of the stepwise isothermal dilatometry and
finally study of the surface of aluminas by means of IR spectroscopy.
The second chapter shows the experimental results of the study of the effect of dispersion on
transition aluminas. The following dispersion routes were adopted: ball-milling (in both alumina and
zirconia media); magnetic stirring and attrition milling. The role of the dispersion route on
agglomeration, phase composition, phase development and surface functionalities was shown. In
addition, the best dispersion method was selected in order to develop dense materials with tailored
microstructures. This part of the thesis was carried out by means of several physico-chemical
characterization techniques (DTA-TG analysis, XRD, BET Specific Surface Area, HR-TEM, SEM,
FT-IR spectroscopy) from the LINCE (Laboratorio INgegnerizzazione materiali CEramici) and
SCREAM (Surface Chemistry and REActivity of Materials) laboratories.
The third chapter summarizes the state-of-art of the existing alumina-based nanocomposites for
extreme applications, with particular reference to the Al 2 O 3 / YAG systems. This chapter contains
literature reports concerning the synthesis of composite powders, the principal techniques
employed for consolidating the nanocomposites and the mechanical characterization at low- and
high-temperature.
The fourth chapter presents the development of Al 2 O 3 /YAG nanocomposites from production of
the composite powder, by means of surface modification of a Taimei α-alumina powder with yttrium
chloride, to its consolidation, by means of conventional (natural sintering) and non conventional
8
(Spark Plasma Sintering and Hot Pressing) techniques, up to its microstructural and mechanical
characterization at low- and high-temperature, with emphasis on the creep behaviour. This study
stems from the co-tutoring between the LINCE (Politecnico di Torino) and MATEIS (INSA-Lyon)
laboratories, by exploiting MATEIS expertise in the field of mechanical characterization of
materials.
9
“All truths are easy to understand once they are discovered; the point is to discover them”
Galileo Galilei
10
CHAPTER
1
An introduction to ceramic nanopowders
11
1. Introduction
In order to take advantage of the properties of bulk nanocrystalline materials, the nanometer range
powders have to be densified into parts of certain properties, geometry and size. The key to the
nanopowders consolidation process is to achieve densification with minimal microstructural
coarsening and/or undesirable microstructural grain growth.
Attempts to produce and densify nanopowders started as early as in 19681. These efforts were
related to MgO to achieve superplastic behaviour. During the 80s, the nanopowders production
was initiated on large scale and the attention was focused on nanopowders processing.
At the following decade the effort was emphasized to develop reproducible processing methods for
manufacturing nanopowders into useful components which retain nanometer properties.
During the second part of the 90s many significant advances in the theory of nanosintering
permitted to achieve fully dense parts with nanometer grain size (<100 nm). However, the
densification of nanopowders created new challenges. Powders agglomeration, contamination and
grain coarsening, complicated the production of large and dense parts. Lower temperatures were
employed in order to reduce the grain growth hindering the intergranular bonding, thus
compromising the expected mechanical strength. The most recent efforts were successful in
overcoming these problems, i.e. agglomeration and grain size control.
A number of reviews on nanopowders processing and general reviews in sintering issues have
been published2.
1.1. Definition and properties of nanopowders
Nanopowders are defined as powdered materials with individual particles having sizes under
100 nm. The particles in nanopowders are smaller than the wavelength of the visible light. The tiny
size of nanopowders gives them an extremely high surface area to volume ratio2,3.
Specific surface area of the particle is related to the average particle size, based on a geometrical
consideration. In case of spherical particles, the formula4 is given below.
SSAsphere 
SAsphere
msphere

4 r 2
3

 .Vsphere  .r
(1)
where SSA is the specific surface area (generally measured in m2/g), SA is the surface area, m is
the mass and V is the volume. Generally, the rougher the particle surface, the greater its surface
area regardless of the particle size. The equation showed above underestimates the surface
associated with the surface texture of the particles. For that reason, it is better to evaluate the SSA
by the BET isotherm (the method is explained in the Appendix I).
Another characteristic of nanopowders is their incoherent interfaces which introduces misfit
between the crystallites. This is the responsible for modifying locally the atomic microstructure by
reducing the atomic density and the coordination between the atoms compared with the perfect
crystal3. As it is shown in Figure 1.1 a, the reduced density is visible by high resolution electron
micrographs. However, the modified coordination it has been studied by means of X-ray diffraction
measurements as it is illustrated in Figure 1.1 b.
12
Figure 1.1- (a) Atomic structure in the core of grain boundary between two NiO tilted relative to one
another by 36.9° about a common [1 0 0] and (b) Coordination number for nanocrystalline Pd (12
nm crystal size) relative to the single crystalline as a function of the interatomic spacing5.
The relative coordination number for interatomic spacing are less than one. For instance, the
intercept of the curve at 0.95 indicated that about 5% of atoms of Pd occupy non-lattice sites.
This fact is though to be the responsible of changing the properties of nanostructured materials
compared with single crystals.
The third aspect, which differs the nanosized powders from the classical, micronic ones is the
morphological metastability related to fine grain size. This metastability in nanocrystalline
materials could be divided into three categories: compositional (extended solution ranges),
structural or topological (alternate crystal structure or amorphous phases) and morphological
(finely divided structures)6.
This is particularly true in the case of nanopowders, the retention of the initial metastability is the
challenge. However, the influence in some fields are not clear, for example in the mechanical field
it has not been proved yet the benefits in keeping the nanoregime11. Although, an arbitrary limit
of < 100 nm is often considered in order to keep the properties as it was mentioned before.
Some examples of materials are shown in Table 1.I, in which it is provided some information
regarding the critical grain size (maximum particle size in which nanoproperties are kept), the
structure and their corresponding equilibrium properties.
Table 1.I – Physical and structural Metastability Ranges6.
13
1.2. Main issues connected with the use of ceramic nanopowders
The potentialities of nanostructured ceramic powders can be affected by some drawbacks, first of
all by their inherent agglomeration, due to Van der Waals forces8. Therefore, in order to produce
dense materials, with tailored microstructure, it is important to control all the elaboration steps,
beginning with the dispersion.
1.2.1. Agglomeration and Dispersion
Most nanocrystalline powders are not composed by single nanometer sized particles. Such
particles known as crystallites are bonded together to form agglomerates or aggregates as it is
illustrated in Figure 1.2. The distinction between the agglomerate and the aggregate is usually
based on the degree of bonding.
Soft agglomerates in which the particles are bond by weak van der Waals and hard agglomerates
or aggregates in which particles show necks between adjacent particles11.
Figure 1.2 – Schematic illustration of nanocrystalline and agglomerates/aggregates9.
Those agglomerates and/or aggregates, could present two different types of pores; interagglomerate pores (micronic) which coexist with smaller inter-crystallite pores (nanometric).
In case of an agglomerated powder, the most common inconvenient is the presence of both intercrystalline and inter-agglomerate pores. If reference to Figure 1.3 is made, we can observe that the
pore size distribution of a non-agglomerated powder (curve c) is characterized by a single peak,
related to inter-crystalline pores, whose maximum is centred at a size corresponding to few
nanometres.
On the other hand, the agglomerated powder (curve a) will have two peaks; the first one
corresponding to the small inter-crystalline pores and the second due to larger inter-agglomerate
pores.
14
Figure 1.3 – Examples of agglomeration distribution: A) heavily agglomerated, B) more simply
agglomerated and C) the ideal case of agglomeration2.
Nowadays, some improvements in the powder de-agglomeration. For example, when TiO 2
powders are produced by the gas phase condensation route, it was recognized the crucial role of
the post-synthesis oxidation step of Ti for avoiding the solid state bonding that it is usually formed
within the particles. Similarly, wet chemical processes such as co-precipitation were shown to
display significantly less agglomeration of crystallites, thus indicating high sinterability12.
However, it is preferable to have soft agglomerates rather than hard agglomerates. Soft
agglomerates could be broken by low-energy mechanical methods (i.e. ball milling) or by
dispersion in a liquid. Hard agglomerates cannot be easily broken and must be removed from the
powder13.
In ceramic processing, colloidal suspensions have been used in consolidation for improving the
green body homogeneity and the microstructural control during sintering.
Each colloidal suspension or slurry contain particles with average sizes between 10-3 μm and 1 μm
within a continuous fluid matrix, called colloids14. They are being used increasingly in the
consolidation of ceramic powders to produce the green body. The ultrafine particles approach and
then separate from each other by Brownian motion, and as a result, settling of particles out of
solution does not occur.
A basic problem is the stability of colloidal suspension. It is also important to define the attractive
forces between forces. Attractive van der Waals forces V vdw exist between particles regardless of
whether other forces may be involved. If the attractive force is large enough, the particles will
collide and stick together, leading to rapid sedimentation of particle clusters, i.e. flocculation or
coagulation. Although in principle, the reduction in the attractive force can be also used, the
techniques employed to prevent flocculation rely on the introduction of repulsive forces. There are
three types of repulsive forces between particles: the repulsion between electrostatic charges V elec
(electrostatic stabilization), the repulsion between polymer molecules V ster (steric stabilization), or
the combination of the two forces described before (electrosteric stabilization).
In fact, as the colloidal stability is ruled by the total potential energy of interaction between particles
(V t ) it could be expressed as:
Vt  Vvdw  Velec  Vster (2)
15
Figure 1.4- Additivity of the van der Waals force between macroscopic bodies15.
To determine the van der Waals forces between macroscopic bodies, it is possible to assume that
the interaction between one molecule and a macroscopic body is simply the sum of the interactions
with all the molecules in the body (Fig. 1.4). For a sphere of radius a at a distance h, V vdw
proposed by Hamaker is given by:
Vvdw  
A.a 
h
h
h 
. 1 
 ln

6.h  2a  h a 2a  h 
(3)
where A is the constant of Hamaker16, which can be determined experimentally by direct
measurement of the surface forces for bodies separated by vacuum, air or liquids.
In order to obtain a stable colloidal dispersion, it is necessary to introduce repulsive forces to offset
the attractive potential. There are several ways for achieving this, but the most commonly used
are:
 Electrostatic stabilization [Fig. 1.5 a] in which the repulsion between the particles is based
on electrostatic charges on the particles.
 Steric stabilization [Fig. 1.5 b] in which the repulsion is produced by uncharged polymer
chains adsorbed onto to the particle surfaces.
 Electrosteric stabilization [Fig. 1.5 c], consisting of a combination of electrostatic and steric
repulsion, achieved by the adsorption of charged polymers (polyelectrolytes) onto the
particle surfaces.
Figure 1.5 – Schematic representation of (a) electrostatic stabilization for negatively charged
particles, (b) steric stabilization, and (c) electrosteric stabilization17.
16
1.3. Green Bodies Forming
The initial step in most densification processes is to compact the powders at room temperature, or
cold compactation, to form a green body. Final sintering results are largely dictated by the green
compact microstructure. A large number of initial point contact, smaller pores in a green density
compact and a uniform pore distribution favour higher final density. The agglomeration, as it was
mentioned, introduce an inhomogeneous particle packing structure and poor green density.
So, optimized and homogeneous green bodies can be potentially densified, since it is possible to
sinter at lower temperatures. On the contrary, inhomogeneities such as particle packing limits the
achievable final density.
Green bodies are frequently yielded by compaction processes; however, it is more critical for
nanocrystalline powders than for conventional ones.
Compaction involves sliding and rearrangement of particles, elastic compression at particle contact
points, plastic yielding form metals or fragmentation for brittle materials. Sliding and rearrangement
in nanopowders are severely restricted owing to large frictional forces among powder particles2,18.
These forces are a result of mechanical, electrostatic, van der Waals, and surface adsorption
phenomena which increase as a particle size decreases.
Particles bonded by weak agglomerates are thought to be easily slided into voids of the green
body. Besides, hard agglomerates generate large shearing forces necessary to produce neck
cleavage before sliding of these particles can take place19.
Figure 1.6 – Compactation behaviour of ceramic powders: hard agglomerated (curve 1) and
weakly ZrO 2 -17% Y 2 O 3 and agglomerated (curve 2) 5%Y-TZP 24,25 powders.
At lower pressures, hard agglomerates formed in ceramics cannot be fractured. To break these
agglomerates it is necessary to reach a critical pressure (Py). On a density-pressure plot (see
Figure 1.6 for yttria-stabilized zirconia powders25), the transition point Py where the slope change
occurs, is interpreted as the strength of the agglomerates. Above, Py the weak agglomerates break
and no transition is seen. When the powder is hardly agglomerated, there is no transition point, as
shown in Figure 1.6 - curve 2.
As a result, particle rearrangement is hindered and lower green densities are likely to be achieved
compared to conventional micron size powders (Figure 1.7). In addition, Chen and co-workers18
found an significant improvement in particle rearrangement by cryogenic compaction through liquid
nitrogen.
17
Figure 1.7– The effect of particle size on green density upon compactation (uniaxial pressing) for
50 nm Si 3 N 4 (squares), 8 nm ZrO 2 (diamond)20, 20 nm γ-Al 2 O 3 (little triangles) and 20 nm γ-Al 2 O 3
(little squares)18,21.
Moreover, Mayo2 found that wet processing provides significantly improved packing uniformity of
ceramic nanopowders. For instance, a green density of 74% of the theoretical density (TD) was
obtained by Rhodes22 for 12.5 nm Y 2 O 3 -doped ZrO 2 by centrifugation from a slip suspension.
Techniques that take advantage of wet processing are the centrifugation, the slip casting and the
pressure filtration.
Homogeneous and closed-packed green bodies yielded by centrifugation22 give rise to higher fired
density, at significantly lower temperatures, if compared to dry-pressed bodies, as shown in
Figure 1.8 for nano-crystalline zirconia24.
Figure 1.8 – Evolution of fired density as a function of the temperature for centrifuged and dry
pressed nanocrystalline ZrO 2 24.
Chen and co-workers21 developed an equation for calculating the pressure necessary to achieve a
certain density on the ground of volume changes. Literature reports that, at room temperature, it is
possible to reach a density as high as 75-90% for nanoceramics19.
18
Gallas et al.23 obtained very high densities (90%TD) for nanocrystalline Al 2 O 3 by cold isostatic
pressing applying a pressure of 5.6 GPa. The major inconvenient in applying large pressures,
apart from the equipment limitations, are the residual stresses that caused the fracture during
handling2.
However, not all the problems are caused by agglomeration. A large number of particle-particle
point contacts per unit volume in a nanopowder, represents a source of frictional resistance to
compaction of the powder, and thus the total frictional resistance to compaction could be higher.
Consequently, for a given applied stress during compaction, particles tend to re-arrange in a
in-homogeneous way compared to conventional powders26. The higher pressures, as it was
mentioned previously, can cause fracture during handling23. The figure below illustrated how
density and residual stress evolved as compaction pressure is increased.
Figures 1.9– Evolution of density and stress gradients in powder compact with increasing applied
stress2.
Cold isostatic pressing (CIP) could be used to decrease the problems related to stress gradients or
density gradients, introduced by uniaxial pressing as it is illustrated in Figures 1.9. Nevertheless,
there is one limitation as the maximum pressure could not be <550 MPa.
1.4. Nanosintering
Ceramics nanopowders have been fully densified by pressureless sintering: some examples are
contained in Table 1.II taken from several references6.
19
Table 1.II- Densities and Grain Sizes of some nanopowders densified by pressureless sintering6.
In case of nanopowders, the driving force is increased thus to reducing the sintering temperature.
In fact, the driving force of sintering is related to the reduction of the total interfacial energy. The
interfacial energy expressed as γ.A, where γ is the specific surface energy and A the total surface
area of the compact.
Figures 1.10– Schematic diagram illustrating the three stages of sintering2
As it is illustrated in Figures 1.10, the sintering curve follows three steps. These steps are
described in the following lines.



First stage: necks are formed by diffusion between adjacent particles. This neck formation
provides an increase of the mechanical strength. There is some speculation in case of
nanocrystalline ceramics, as fast surface diffusion combined with applied or residual
stresses could occur. This phenomena allows small grains to slip along their boundaries
and rearrange in a more packing efficient manner.
Second stage: is characterized by the formation of a kind of sponge containing
interpenetrating tubular pores to the external surface. In this stage of sintering, the density
increases from 6 % to around 90% of theoretical density. In the better situation, this tubular
pores shrink in radius29.
Third stage: the open pores finally shrink.
20
The reduction in size of pores during pressureless sintering occurs by diffusion of the internal
surface area. The instantaneous driving force for diffusion is assumed to be inversely proportional
to pore’s radius of curvature. The expression given for such phenomena is shown below:
 A  
 A 
c  co exp 
  co 1 

rkT 
 r kT 

(4)
where, c o is the equilibrium vacancy concentration in the bulk material, away from the pore, k is the
Boltzmann’s constant, Ω is the atomic volume, γ is the surface tension associated with pore/solid
interface and A is a geometric constant with a value of 1 for cylindrical pores and 2 for spherical
pores, respectively.
The densification is favorized in pores with tight curvature as they have vacancy concentration
near the pore surface; consequently, they support a higher vacancy concentration gradient
between the pore surface and the nearest grain boundary or external surface, and lately the small
pores sustain a greater vacancy flux out of the pore by Fick’s first law.
Sintering of nanopowders at low temperature was achieved for the first time during the 80s30,31,24.
Andrievski32 was one of the first who studied the effect of additives as Y 2 O 3 and in ZrO 2 . By using
Y 2 O 3 , it was achieved 98%TD (theoretical density) sintered at 1220°C. In case of the Bi 2 O 3 , it was
achieved a density of 93%TD, by sintering at 935°C.
In both cases the final grain size was 600-700 nm. The same concept was extended for different
nanograined ceramic powders such as Al 2 O 3 , CdO and TiO 2 by pressureless sintering obtaining
fully dense samples.
In certain studies, it has been published that fast heating rate in a ramp-and-hold sintering protocol
may contribute in obtaining better densities than with low-heating rate36. At low temperature, the
mechanism is controlled by surface diffusion which consists in neck formation and little
densification. Surface diffusion induce prevalently grain growth. However, at high temperatures the
lattice diffusion mechanism helps the material densification avoiding the excessive grain growth. In
the following Figure 1.11 it is described how the densification rate changes in function of the
temperature.
Figure 1.11– Diagram of change in densification rate with the temperature30.
The rapid rate will avoid the surface diffusion regime limiting the total grain growth. This method
could be suitable for nanopowders due to the high specific surface which generally conducts to
surface diffusion. This method was successfully applied in thin films. The ZrO 2 -3 mol% Y 2 O 3 was
sintered at different heating rates varying from 2 to 200°C/min; the authors found that by increasing
the heating rate, the densification is delayed, as shown in Figures 1.12.
21
Figure 1.12– Effect of heating rate on densification of nanocrystalline ZrO 2 -3 mol% Y 2 O 3 13.
A second drawback of employing fast heating rates is the generation of density gradients inside the
sample. In the following Figure 1.13 it is shown an example of the ZrO 2 -3 mol.% Y 2 O 3 sintered at
20°C/min up to 1200°C and the relative microstructure in different positions over the sample. As
shown by the images at higher magnification of Figure 1.13, the density slightly increases from the
center (top image) to the border (bottom image), indicating the presence of a density gradient.
Figure 1.13– SEM micrographs of ZrO 2 -3 mol.% Y 2 O 3 sintered up to 1200°C (20°C/min) for 2 h13.
Chu et al.36 studied the effect of the heating rate on coarsening and sintering of pure ZnO. The
authors found that the densification strain rate with a coarsening process is characterized by two
types of activation energies, one attributed to densifying processes at higher temperatures and the
other to non-densifying process at low temperature. Densification strain rate was defined by the
authors as
1 d
d
, where ρ is the density and
is the variation of density as a function of time.
dt
 dt
22
The authors studied the volumetric strain rate (the linear densification rate is one third of the
volumetric rate) as a function of the heating rate and the sintering temperature. The maximum
heating rate shifts to higher temperature with increasing the heating rate. For temperatures above
700°C the densification increases approximately linearly with heating rate. However, below this
temperature the densification rate increases slowly (Figures 1.14). The authors also confirmed by
SEM observations the hypothesis of the inverse proportional relationship between the grain size
and heating rate.
Figures 1.14– Sintering at different heating rates of ZnO: (A) densification strain rate versus
sintering temperature (B) change on densification strain as a function of the relative density36.
Similar data was obtained by Wang et al.31 for α-alumina and α alumina doped with zirconia and
titania. The result of the study was that sintering behaviour depends on the dopant, as it is shown
in Figures 1.15. By doping with TiO 2 there is an anticipation in densification, while in case of
adding ZrO 2 the densification is delayed. In the three samples the densification followed grain
boundary diffusion for the heating rates involved.
This study also permitted to corroborate the effect of heating rate in the densification, as lower
heating rates allows to reach higher fired densities. This fact is well shown by the Figures 1.15 for
the pure Al 2 O 3 , as well as, the Al 2 O 3 doped with 5 vol.% TiO 2 and ZrO 2 .
23
Figures 1.15– Density and densification as a function fo temperature for Al 2 O 3 , Al 2 O 3 -5 vol.%TiO 2
and Al 2 O 3 -5 vol.% ZrO 2 31.
Other hypothesis of the sintering mechanism was proposed by Zeng et al.48 The authors indicated
that the initial sintering of α-Al 2 O 3 up to a density of 77%TD is dominated by grain boundary
diffusion.
1.4.1. Pore Size and its effects during densification behaviour
As in conventional powders, full density or rapid sintering of nanosized powders is achieved when
the green sample contains a narrow peak in the pore size distribution (explained previously in
Section 1.2.1). It is well known that densification is retarded when the pore distribution is wide. The
removal of large pores requires high temperatures.
Kingery et al.37 developed a thermodynamic model of the pore shrinkage based on curvature
considerations. The authors predicted the pore size for the transition between pore shrinkage and
pore growth. A complicated event is the separation of large pores from the grain boundaries that
occurs in final sintering stages.
Brook et al.38 theorized pore-size-grain size map models to designate the regions where such pore
separation from grain boundary occurs. When pore-grain-boundary breakaway takes place, the
detached pores will no longer benefit from easy transport paths such as grain boundaries. Instead,
the transport is produced by lower volume diffusion and, consequently, densification occurs at a
slower rate. This is the stage when open porosity breaks down and pores become closed. Since
grain boundary migration is no longer restricted by pores, grain coarsening takes place, or pinning
of grain boundary migration due to residual pores is no longer effective.
The result is a hindered densification at the final sintering stage which avoids the significant grain
growth. This prevents high fired density being achieved. If coarsening occurs in late sintering
stages, the overall notion of nanograined size is compromised. Experimentally, a strong correlation
24
between the closure of open porosity and the onset of exaggerated grain growth was noted in
ceramic nanopowders.
Figure 1.16 – Densification behaviour: exaggerated grain growth in agglomerated TiO 2 and Al 2 O 3
non-agglomerated powder37,39,40.
As it is shown in Figure 1.16 and 1.17, accelerated grain growth occurs in agglomerated titania
powders at densities above 90% or when pores become closed41. However, for non-agglomerated
powders (case of alumina in the same Figure), no exaggerated grain growth takes places upon
densification.
Figures 1.17 – Densification behaviour: Effects of dopants on grain growth of TiO 2 41.
Mayo et al.42 studied the influence of agglomerate/pore size on the sintering temperature. In Figure
1.18 are shown the results for a nano-TiO 2 . Non-agglomerated powders (curve 1) achieve higher
densities at lower temperature (1073 K), as compared to the agglomerated ones (curves 2 and 3).
The second titania powder has a lower crystallite size (<16 nm) as compared to the previous one,
but it is agglomerated (curve 2): as a consequence, its densification occurs at higher temperature
(1173 K). Finally, the third powder is even more agglomerated (curve 3): as expected, high final
25
density is reached at a even higher temperature (1373 K). As declared by the Authors, high
sintering temperatures for agglomerated powders are the result of having large inter-agglomerate
pores.
Figure 1.18 – Effect of agglomerate/particle size on the sintering temperature of nano-TiO 2 :
agglomerate size (particle size): 1) non-agglomerated (<40 nm), 80 nm (16 nm), 3) 340 nm (8-10
nm)43.
Mayo2 developed a modified sintering law that directly accounts for pore size effects on the
densification rate:
1
d
1 1
 Q 
 n exp 

 1    dt d r
 RT 
(5)
where ρ is the density, d is the particle size, n is a constant dependent on the sintering
mechanism, r is the pore radius, Q is the activation energy, R is the gas constant, and T is the
absolute sintering temperature. This equation predicts that the highest densification rate occurs for
the finest pore size. It was found to hold throughout the sintering process and for large pore sizes,
at least when pore size distribution is uniform. This relationship has two implications. First, the pore
size in addition to grain size should be controlled during sintering. Fast sintering kinetics result
when pore sizes are small. Second, the densification rate is dictated by the instantaneous pore
size, not only the initial pore size. Therefore, to maintain a fast sintering rate in late sintering
stages, the pore should remain small at late sintering stages.
Small pore size is critical throughout the control of the final grain size. For this purpose, a small
and uniform pore distribution is desired in the green compact. Most often, pore distribution is
dictated by the green density value. A high green density with a small pore population is easily
reached in non- or weakly agglomerated powders2,25,44. Rhodes22 was one of the first researchers
who demonstrated that de-agglomeration powders may reduce sintering temperature from 1500°C
to 1100°C. In Mayo’s review2 it is well summarized the progress of producing non-agglomerated
ceramic powders.
The advantages of a small pore size and narrow size distribution, in reaching high densities, have
been studied by many researchers for non-agglomerated γ-Al 2 O 3 10,33.
26
1.4.2. Grain Growth
In nanomaterials, grain coarsening is described by an equation similar to that of conventional
materials:
d n  d o n  kt (6)
where d is the average grain diameter, d o is the initial grain diameter, n is the grain growth
exponent, k is an Arrhenius dependence (k=k o exp(-Q/RT)), and t is the time. From theory, n should
take on a value of 2 for normal grain growth in pure, single phase system, 3 for grain growth in the
solutes, and 4 in the presence of pores. Some researchers as Harmer et al. have determined
experimentally K pure materials and composites45. At low temperatures, n values are high or grain,
similar to regular materials. Such increased n values are rationalized by restricted grain boundary
mobility which is most commonly due to pore or solute effects.
The influence of pores on grain size has been well documented theoretically and experimentally in
ceramics2,11. Open pores, which are present throughout the second stage of sintering are effective
in limiting grain coarsening (Figure 1.10)32. The pinning action of the pores is difficult to predict7.
Liu et al.46 found a linear relationship between the inverse of grain size and the pore surface area
per unit volume. The experimental verification of this model was performed on porous materials by
adding dopants such as MgO in Al 2 O 3 achieving the total pore closure46,47.
Another strategy developed to control grain growth are by kinetic and thermodynamic approaches.
Restricting grain boundary mobility using particle pinning effect (see on page 15, 3.4.1 Role of the
second phase on the retention of the matrix grain growth) is an example of controlled coarsening
kinetics.
Zeng et al.48 proposed that by avoiding the powder agglomeration on α-Al 2 O 3 and having an
average particle size <20 nm, nanoceramics with grain size <100 nm may be produced. The
sintering temperature could be done by using i.e. the Herring’s scaling law.
The Herring’s scaling law20, made several assumptions based on the particle size and sintering
time based on the operational sintering mechanism. The Herring’s scaling law assumes that the
time t 1 required to sinter a particle of diameter D 1 to achieve a sintered neck size of X 1 is known.
Then the effect of a change in particle size can be predicted. The sintering time t 2 for a particle of
size D 2 to reach the same neck size ratio (X 1 /D 1 =X 2 /D 2 ) is given as,
t2  D2 


t1  D1 
m
(7)
where D 1 and D 2 are the particle sizes, t 1 and t 2 are the sintering time for different particles sizes,
and m the dimensionless scaling-law exponent. The m values were determined for different
mechanisms as follows: m=1 for viscous flows and plastic flow, m=3 for volume diffusion and m=4
for surface diffusion and grain boundary diffusion. There are still some discrepancies in using the
scale law6.
By applying the Arrehenius expression for temperature, the sintering temperature on the particle
size could be calculated by using the following formula:
 d  Q  1   1  
n ln  1         (8)
 d 2  R  T1   T2  
27
where Q is the activation energy, R is the gas constant, d 1 and d 2 are the different particles sizes
and T 1 and T 2 are their respective sintering temperatures.
Messing et al.33 calculated the activation energy for volume diffusion in alumina, obtaining
543 kJ/mol close to volume diffusion.
1.5. Nanoaluminas
1.5.1. Metastability of Aluminas
The commercial zirconia powder with nanometric dimension is easy available55. However, in case
of commercial α-aluminas, it is almost impossible to produce it with crystallites size lower than 80100 nm. This size makes difficult obtaining nanosized grains in the fired materials.
The alternative is to use transition alumina (δ-, γ-, θ-) which usually has a crystallite size of around
50 nm. The higher crystallite size of α-aluminas is caused by the calcination temperature required
to yield the α-phase (Figure 1.19).
Figure 1.19 – Flowchart showing transformation among Al-O compounds49.
1.5.2. Vermicular Growth & Seeding
In case of transition aluminas in the form of δ-, γ-, θ-, the transformation in α-phase is generally
accompanied by the formation of a vermicular microstructure, consisting of a network of large
pores36 (Figure 1.20 (A)). This morphology is induced by relative density change (Δρ/ρ) of about
17% which accompanies γ→α transformation (theoretical densities 3.41 g/cm3 and 3.897 g/cm3 for
gamma and alpha aluminas, respectively).
As a consequence of the development of such microstructure, the final stage of sintering requires
very high firing temperatures to consolidate the material up to the theoretical density, thus inducing
a significant grain growth32.
28
Figure 1.20– SEM micrograph of γ-Al 2 O 3 sintered for 100 min at 1400°C: (A) un-seeded and
1.25 wt.% seeded sample51.
Nordahl and Messing51 proposed to seed γ-Al 2 O 3 with α-alumina nuclei: they reported that
nucleation of -Al 2 O 3 is favoured55, the vermicular microstrusture is hindered and thus, sintering
can occur at lower temperature. Figure 1.20 compares the microstructure of fired alumina samples,
starting from un-seeded (A) and -alumina seeded (B) γ-Al 2 O 3 powders. If fired to the same
temperature (1400°C), a highly dense microstructure can be obtained only in the case of the
seeded sample. On the contrary, the un-seeded powder reached full densification at 1600°C, thus
resulting in a coarsened microstructure.
The seeding stimulate the transition alumina transformation into α-phase upon heating50. An
investigation has been done by Yen et al., who evaluated the activation energy of nucleation by
DTA profiles (see Section 1.6). The seeded sample showed a lower activation energy of
nucleation. However, the activation energy for the growth stage for both seeded and un-seeded
remain inalterable.
Palkar et al.53 produced nanostructured alumina by rapid nucleation of Al(OH) 3 derived from the
sol-gel process. Since the nucleation rate is higher that the growth rate, the rapid nucleation was
successful in avoiding the formation of the vermicular microstructure. As a consequence, full
densification was achieved at a very low temperature (1250°C).
Legros et al.54 claimed that in transition aluminas the formation of α-alumina occurs by nucleation
and growth process during heating. The nucleation steps depends on the density of the nuclei
which can result from the number per unit of α-grains present in the green bodies seeded.
The authors demonstrated that by increasing the mechanical contacts between transition particles,
which are potential nucleation sites, it is possible to promote nucleation, as it is shown in
Figure 1.21. Heating rate was also considered by the authors to be important as the nucleation
sites increase by increasing the heating rate. So the capacity of the particles rearrangement during
the phase-transformation seems to be the key. This process of re-arrangement depends on the
compact density, heating rate and amount of α-alumina present in the initial powder.
29
Figure 1.21- Schematic representation of rearrangement and coalescence mechanisms
responsible for high relative density changes during gamma to alpha transition53.
Bowen et al.55 concluded that no discernable densification is observable in γ-Al 2 O 3 before α-Al 2 O 3
transformation. This suggests that once diffusion processes become active enough for
densification of γ-Al 2 O 3 , the α-Al 2 O 3 transformation is already energetically favoured.
Despite all the efforts, the final grain size in aluminas produced from transition powders range from
600 nm to 1 μm9,10.
In the same direction, Wu et al.10 have shown that a rapid heat treatment of a transition alumina
leads to achieve a higher final density near the theoretical (Figure 1.22). The reason is the
formation of α-alumina seeds during the rapid heat treatment which is the responsible of increasing
the initial density of the power.
Also, the authors evaluated the effect of dopants as the MgO. As it is illustrated in Figure 1.22, the
doped material reached a final density near 80%TD. However, in this particular case the powder
preparation was achieved by mechanical milling which, as the authors claimed, was the
responsible of introducing some seeding which avoided the grain growth.
Figure 1.22 – Relative density vs. temperature for compact formed from (I) As-received Powder,
(II) Partially Transformed into α-alumina and (III) MgO-doped 250 ppm powder10.
30
As with γ-Al 2 O 3 , Messing et al.33,58-61 attempted to obtain microstructures with retained size using
Boehmite (γ-AlOOH) as starting material. For this reason the authors proposed to study the
influence of α-alumina (0.1/0.4 μm seeds) in Boehmite (γ-AlOOH) gels. They found an optimal ratio
of 5.1013 of seeds per cm3 of bohemite gel, which could be reached with a loading of 1.5 wt.% of αalumina seeds. In this particular case the samples reached a final density of >98%TD after 5 min
at 1150°C. It was revealed that this technique was efficient in reducing the α-phase transformation
temperature and in refining the final microstructure, as it is shown in Figure 1.23.
Figure 1.23– SEM micrograph of (A) un-seeded and (B) γ-AlOOH seeded gels after sintering at
1250°C57.
Similar results were found by other researchers60-61. For instance in case of Xie et al.61, compared
the α-alumina seeding introduced by doping and by the milling medium. In the particular case of
the second method, the authors found a better distribution of seeding in the sample.
The same procedure was extended to γ-Al 2 O 3 52,62, obtaining similar results such as the promotion
of a higher density. Wang et al.52 studied the benefits of introducing α-seeding on γ-Al 2 O 3 , as the
barrier of nucleation is removed so the only remaining barrier is diffusion with limited grain growth.
1.6 Models developed so far to study the transformation kinetics of aluminas
Avrami equation describes how solids transform from one phase to another at constant
temperature. It can describe the kinetics of crystallization, generally the transformation phase in
materials.
The equation is known as Johnson-Mehl-Avrami (JMA) equation63, and it is written in the following
way:
x  1  exp  kt 
n
(9)
where x is the amount of material transformed at the time t; k and n are constants with respect to t
the firing time. This equation could be only used to describe the transformation kinetics of many
solid state processes under isothermal conditions.
The rate constant k, could be expressed as:
k (T )   .exp   E / RT 
(10)
where ν is the frequency factor, R is the gas constant, and E is the activation energy associated
with the process which is often interpreted as the energy barrier opposing the reaction. The
constant k o , most called frequency factor, is a measure of the probability that a molecule having
energy E will participate in a reaction.
31
Isothermal process has been used for the determination of the activation energies of chemicals
reactions which usually takes a long time65. The majority of the published works have been
devoted to studying the kinetic parameters of crystal growth and glass devitrification by means of
differential thermal analysis (DTA)65-66. In these investigations a continuous isothermal method was
employed.
The experimental procedure is based in evaluating the x, for instance by means of XRD, in function
of temperature and time. The results can be plotted as it is illustrated in Figure 1.24.
Figure 1.24 – A series of isothermal α-time curves at different temperatures67.
By applying the natural logarithm in Eq.7, it is possible to build a second plot ln[ln(1/(1-x)] versus
ln(t) in order to obtain the n and k for the different temperatures. Finally, by constructing a third plot
using Eq.8 expressed in logarithm form ln k=ln ν-(E/RT) it is possible to obtain the constant ν and
the activation energy E.
A second method proposed by Kissinger, has already shown how activation energy and frequency
factor could be calculated from DTA experiments by making a number of differential thermal
patterns at different heating rates. This method has only value for homogenous reactions68-70.
Kissinger compared his results with several minerals of the kaolin group data obtained by
conventional isothermal techniques obtaining comparable results. The Kissinger’s method is based
in the assumption of Eq.7 and Eq.8.
When the temperature is changing with time, then the reaction rate is changing with time. The
reaction rate is:
dx  x   x  dT
   

dt  t T  T t dt
(11)
The rate of change of x with temperature, with the time coordinate fixed (∂x/∂T), is zero, because
fixing time also fixes the number and position of the particles constituting the system. The only
effect, of an instantaneous change in temperature is in the velocity of thermal motion of particles.
The total rate reaction may be expressed as,
E

dx
 A 1  x  e RT
dt
(12)
32
This expression holds for any value of T, whether constant or variable, so long as x and t are
measured at the same instant. When the reaction rate is maximum, its derivative with respect of
time is equal to zero, thus the reaction rate is
E


d  dx  dx  E dT
RT


Ae


 
2
dt  dt  dt  RT dt

(13)
The maximum value of dx/dt occurs at temperature T m defined by

Ae
E
RTm

E dT
RTm 2 dt
(14)
Replacing in the previous equation
  
d  ln 2 
 Tm    E
R
 1 
d 
 Tm 
(15)
where α is the heating rate dT/dt. If the reaction rate order is zero, the peak occurs when the
material is exhausted at T m , the temperature decomposition. Evaluating T m at different heating
rates, it is possible to build a plot ln (α/T m 2) versus 1/T m , and directly obtain the transformation
energy from the slope.
The are some requirements for applying Kissinger method taken from the Nordahl et al.72 Those
conditions are described in the following lines.

The temperature of the thermocouple is the temperature of the entire sample.

The peak temperature of the DTA peak represents the temperature of the maximum
reaction rate.

The heat capacity remains constant during the test.

The reaction is first order. Due to the consistency of reaction order obtained by the various
analysis techniques given in Table 1.III, it is assumed that the phase transformation is of
the first order.
This last point is based on the last densification stage, as for example in Boehmite the final
microstructure is supported by a denditric or vermicular growth process.
33
Table 1.III- Differential thermal analysis methods for determining kinetic parameters of the θ to
α-Al 2 O 3 phase transformation in γ-Al 2 O 3 72.
The application of Kissinger’s method was employed in evaluating the effect α-alumina seeding on
Boehmite and γ-Al 2 O 3 . The authors found that the addition of α-alumina seeds have a significant
effect on the transformation techniques in increasing the final density and reducing the activation
energy for the θ to α-Al 2 O 3 phase transformation (Figure 1.25).
Figure 1.25 – Determination of the activation energy for θ to α-Al 2 O 3 phase transformation in: (A)
seeded Boehmite and (B) γ-Al 2 O 3 72.
Kao et al.73 evaluated the effect of seeding in θ-aluminas, studying the phase transformation into αAl 2 O 3 . The authors obtained higher values (650 +/- 50 kJ/mol) by a modified Arrhenius method.
They proposed an explanation for diffusion mechanism which happens during θ to α
transformation. The oxygen atoms diffuse from the matrix the θ across the boundary during
transformation to α phase, and they generate vacancy clusters on the θ→α interface. The vacancy
clusters are eliminated by the interphase boundary diffusion (D i ) and by the concentration gradient
near the interface.
The activation energy of lattice diffusion (D l ) and grain boundary diffusion of (D b ) of oxygen in
Al 2 O 3 are 700 +/- 30 and 500 +/- 30 kJ/mol, respectively. Due to the similar activation energies of
D i and D b which are lower than D l , the diffusion of oxygen should be the mechanism which
controls the transformation into α-phase (Figure 1.28). Thus, the transformation mechanism is
based on the structural rearrangement by diffusion of Al 2 O 3 in the lattice.
34
They sustain that activation energies given by other researchers might imply other transformation
mechanisms.
Figure 1.26- Schematic diagram of the possible diffusion route and vacancy concentration profile of
the θ→α transformation73.
Other researchers as Yang et.al76 claimed that the discrepancy in total activation energies exist
ranging from ≈200 to ≈650 kJ/mol. The reason is that there are three coexisting barriers, and
consequently is difficult to determinate each of them. Yen et al.84 defined three steps in the θ→α
transformation (Figure 1.27) which are explained in the following lines:

Θ-crystallites grow to the critical size of phase transformation (d θ →d cθ ).

Θ-crystallites increase to 22 nm and d cθ transform to α-nuclei (d cθ →d cα ).

α-nuclei of d cα (17 nm) coarsen, exceeding the primary size (d p ) and with crystallite size
≈45 nm then the phase transformation is completed.
Figure 1.27 – Schematic description of the growth phenomena of θ→α transformation74-75.
It is believed that the activation energy of θ-crystallite growth can be the most significant factor that
dominates the discrepancy.
The θ→α transformation is mainly composed by nucleation and nucleation + growth processes:
related activation energy are 85 and 580 kJ/mol74-75 respectively. The last value is closer to the
previously reported values.
35
Figure 1.28 – Free energy changes during θ→α transformation as a function of radius r of an
alumina crystallite74.
Yang et al.76 studied the θ→α transformation of three different samples: the first was the unmodified, raw powder; the second was dispersed by mechanical stirring; the third was dispersed
and than uniaxially-pressed. The authors observed that the dispersion was effective in
disintegrating the agglomerates; the subsequent application of uniaxial-pressing decreased the
inter-crystallite distance and promoted the θ→α transformation, by lowering the related
temperature.
If the θ→α transformation temperature is lowered, the vermicular growth is also hindered because
of retaining the crystallite growth (Figure 1.28)
1.7 Sintering Kinetics studied by Stepwise isothermal Dilatometry
Stepwise Isothermal Dilatometry (SID) is a technique which has been proven to be very useful in
sintering studies of ceramic powders taken from the original publication of El Sayed et al.77.
Compared to conventional dilatometry, where the samples is heated at a constant rate, SID has
the advantage that the controlling mechanism and the activation energy can be determined.
The characteristic of SID is that the heating of the sample is controlled by the magnitude of the
derived signal, i.e. the dimensional change signal dl/dt. Sintering takes place in isothermal steps
where the time span for each step depends upon the sintering rates and it can be controlled by the
setting of the two threshold values. Generally the setting is chosen according to the
experimentation. In Figure 1.29 , it is shown a schematic representation of this technique.
36
Figure 1.29- Example of shrinkage and temperature curve recorded during stepwise isothermal
dilatometry77.
Each isothermal steps in the initial sintering stage the densification can be expressed in the
following way.
y
l
n
  K (T ).t 
lo
K (T )  A. ..D / k .T .r P
(16)
(17)
where lo the initial sample length at the start of sintering, K(T) the Arrhenius constant, D the
diffusion coefficient, r the particle radius, γ the surface tension, Ω the surface tension, k the
Boltzmann constant and A, n, p constants whose values which depend on the sintering
mechanisms.
In order to apply this equation it is necessary that shrinkage and time are recorded simultaneously.
For initial sintering stage it is impossible to determine the absolute time involved in the single
steps, thus obliging to remove the variable time. In this way differentiation of y could be done as it
is expressed in the expression below.
.
y   K T  / N  . y  N 1 (18)
.
where y is the shrinkage rate and N  1/ n .
.
Plotting ln y in function of ln  y  as in Figure 1.30, determined for all the points in the curve. For
each isothermal step, a slope could be obtained. For each steps could be obtained N and n from
.
the slope and K(T) from the intercept with the ln y axis.
37
Figure 1.30 - Shrinkage rate versus shrinkage for isothermal steps during sintering of CeO 2 77.
By using the Arrhenius expression and applying logarithms on both sites.
ln  K T    ln A  Q / RT
(19)
where A is the pre-exponential factor and Q the activation energy. This values can be calculated by
plotting the ln  K (T )  in function of 1/T, as it is shown in Figure 1.31.
Figure 1.31 – Arrhenius constant (K(T)) versus reciprocal temperature 1/T for sintering of CeO 2 77.
1.8 Spectroscopic study of the surface of aluminas
In the practical language of catalysis and surface chemistry, the term alumina is normally referred
to the so-called “transition aluminas”.
As this paragraph it is mainly devoted to the surface and spectroscopic features of aluminas, the
attention is focused in most commonly used crystalline cubic transition phase Al 2 O 3 systems and
other less common systems like, for instance, amorphous aluminas and hexagonal-type aluminas
(χ-Al 2 O 3 and κ-AI 2 O 3 ) are neglected. Besides transition aluminas, also some of the features of Al
hydrates and of the corundum phase will be necessarily mentioned every now and then. In fact, the
structural and coordinative evolution of the AI oxide system from one extreme to the other will turn
38
out to be quite important for the definition and comprehension of some of the surface properties of
the transition phases.
For these reasons, and to avoid confusion, for all the systems considered here either the complete
phase designation (e.g., γ-AlOOH, θ-Al 2 O 3 and α-Al 2 O 3 ) or the current phase name (as instance,
boehmite, transition aluminas and corundum) will be used.
The complex thermal evolution of the Al hydroxides (Al(OH) 3 ; gibbsite, bayerite and nordstrandite)
towards corundum, passing through the monohydrates (boehmite and pseudoboehmite) and the
various transition alumina phases, has been thoroughly studied in the sixties by Lippens78. His
detailed phase description is considered to be still valid and has been constantly referred to in all
more recent studies carried out in the field.
As mentioned, catalytic aluminas belong to the group of transitional Al 2 O 3 phases, (meta-) stable
in the ca. 750-1370 K range. They are usually further divided in two families79: the so-called lowtemperature transition phases (γ- and η-Al 2 O 3 ) and the high-temperature transition phases (δ- and
θ-Al 2 O 3 ). On a crystallographic ground, the irreversible thermal transition from low- to hightemperature transition aluminas has been reported to be a continuous process79, i.e., a mere
transition of the order disorder type.
In fact, the structural differences between the two families of aluminas are relatively small as all
transition aluminas belong to the cubic system and have the nature of defective spinels. On a
catalytic ground, the passage from low-temperature to high-temperature transition aluminas is
more critical. In fact, the high-temperature transition phases are definitely less active than the low
temperature ones. This is not merely due to the lower surface area of the former ones (brought
about by the higher order and larger particle size), but must reflect a different population of surface
active sites.
In normal spinels, in which no inversion phenomena occur (inversion as observed, for instance, in
NiAI 2 O 4 ), M2+ cations occupy only one eighth of the tetrahedral sites in the cubic close packed
array of oxide ions, whereas Al3+ (or other possible trivalent cations) occupy half of the octahedral
sites, so that the general formula is often written as MIV[AI 2 VIO 4 ] or, with reference to the unit cell,
M 8 IV [Al 16 IV O 32 ].
The defective nature of the transition alumina spinels derives from the presence in aluminas of only
trivalent cations, so that some of the lattice positions occupied by cations in mixed-oxide spinels
must remain empty to guarantee electrical neutrality. The overall formula referred to the unit cell is
then Al 8 IV [Al 13 VI 1/3 □ 2 2/3 O 32 ]. The square symbol denotes the presence of cationic vacancies with
respect to the ideal spinel structure: this notation implies, as suggested by Wilson and McConnell
for δ-Al 2 O 3 , that cation vacancies are essentially located in octahedral sites, but vacancies should
be more realistically imagined as randomly distributed between tetrahedral and octahedral
cavities80.
It is also logical to expect some of the vacant cationic positions, imposed by the Al 2 O 3
stoichiometry, to be present also in the surface layer of transition aluminas. This is certainly one
more factor, besides the double coordination presented by AI ions, that is bound to be somehow
responsible for the complex and variable surface situation typical of the transition alumina systems.
1.8.1
Vibrations of the solid
In Figure 1.32 reports the IR spectra of the fundamental modes of several Al oxidic systems;
Figure 1.32 (A) is taken from a recent work by Busca et al.81 and concerns KBr-pellets of the
phases γ-Al 2 O 3 (a), θ-Al 2 O 3 (b) and α-Al 2 O 3 (c), whereas Figure 1.32 (B) shows, for comparison,
original KBr-pellet spectra of δ-AI 2 O 3 (d), η-Al 2 O 3 (e) and γ- AIOOH (f).
39
The spectra of Figure 1.32 confirm that: (i) when the sole octahedral coordination of AI ions is
present, like in corundum and boehmite (c, f), the spectrum is dominated by the strong stretching
mode of AIO 6 octahedra82, centered in the 750-600 cm-1 region. The absorption may be split into
more components (two in the case of α-Al 2 O 3 and three in the case of γ-AIOOH), due to lowering
of the local symmetry and resolution of degenerate modes, but no strong absorptions are ever
observed at  > 800 cm-1; (ii) when also the tetrahedral coordination of Al ions is present, like in
transition aluminas (a, b, d and e), there is also a strong and broad absorption in the region 900750 cm-1, due to stretching vibrations of a lattice of interlinked tetrahedra82. In particular, in the
case of the high temperature spinel transition phases the band of AlO 4 tetrahedra becomes
broader, stronger and partly resolved into several sharp components, due to the higher crystalline
order achieved. This can be observed with δ-Al 2 O 3 preparation (spectrum d in Figure 1.32) and,
even more, with θ-Al 2 O 3 (spectrum b in Figure 1.32 and the spectrum of a fairly pure θ-Al 2 O 3
preparation reported by Tarte 82).
Figure 1.32 – IR spectra in the skeletal region of some Al oxide preparations. Section A: γ-Al 2 O 3
(a), θ-Al 2 O 3 (b), α- Al 2 O 3 (c), δ-Al 2 O 3 (d), η- Al 2 O 3 (e) and γ-AlOOH (f)79.
1.8.2
Surface Vibrations
Surface-localized vibrational modes have been reported for high-area transitions between different
lattices of aluminas79,83,84. When a strong base such as pyridine is adsorbed on highly dehydrated
transition aluminas, an appreciable increase of the IR transparency on the high-frequency side of
the alumina cutoff is observed. This effect was interpreted on the basis of in situ experiments as
due to the relaxation of some Al–O vibrations localized in the surface layer and absorbing in the
spectral region around 1000 cm−1 79,83,84.
Differential absorbance spectra showed that a discrete weak and broad band centred at about
1050 cm−1 appears it is shown the spectra of dehydrated aluminas at increasingly higher
temperatures (Figure 1.34). This band was ascribed to Al–O vibration modes localized in surface
defective structures of the following type:
40
Figure 1.33 – Surface defective structure in highly dehydrated aluminas93.
The surface defect would be created during high-temperature surface dehydroxylation and readily
destroyed during hydration of the alumina (the band at about 1050 cm−1 disappeared upon water
adsorption84. Later, both Marchese et al.84 and Morterra et al.79 confirmed the identification of a
surface-localized Al–O vibration mode on both low- and high-temperature transition aluminas. They
showed that no surface chemical changes (like, for instance, the rupture of a straightened Al–O
bond) are actually needed to have the Al–O vibration shifted downwards to a lower value and that
upon the reversible adsorption of a soft base, like CO, a complex band localized in the
1100–1000 cm−1 region is gradually and reversibly eliminated. Morterra et al.86 inferred that on
highly dehydrated aluminas the adsorption of CO does not cause the rupture of bonds, but just
polarization by weak σ-coordination onto unsaturated surface cations.
Figure 1.34 – Differential absorbance spectra showing the elimination of surface-localized
vibrational states upon adsorption of CO at 77 K onto γ-Al 2 O 3 activated at 1023 K79.
The complex band at about 1050–1100 cm−1 has been interpreted as being due to spinel AlIV–O
stretching modes localized at the surface. Such modes would be shifted upwards with respect to
the regular bulk AlIV –O stretching mode(s), which absorb at ca. 850 cm−1, and the shift caused by
the surface increase of covalence, e.g. of the surface decrease of the Madelung energy, brought
about by crystal truncation and surface dehydration. Upon gas–solid adsorption, ligands are added
to the coordination sphere of the surface cations, the overall coordination of surface ions
increases, and the covalence of the surface then decreases. As a consequence, the surface
localized AlIV –O vibrational states shift downward toward the regular spectral positions of the bulk
AlIV –O vibrations and outside of the spectral range observable in the in situ transmission mode.
1.8.3
Surface Hydroxyls Groups
Broken bonds at crystal surface become saturated as a result of dissociation of the H 2 O
molecules: hydrogen is bonded to an oxygen atom, whereas the hydroxyl group bonds to the metal
41
atom. The appearance in the spectra of several bands characteristic of hydroxyl groups, as well as
their
different spectral features and chemical properties (acidity, reactivity), are due to the exposure of
several crystal faces and different kinds of defects on the oxide surface.
Oxides which had been heated at only moderate temperatures exhibit several hydroxyls bands, the
appearance of which depends on the preparative conditions and pre-treatment temperature. In
general, the IR band due to the OH stretching vibration of an hydroxyl group appears in the 38003000 cm-1 range and is strengthened in integrated intensity and broadened if the group is involved
in hydrogen-bonding interactions.
It is now clear that the spectral features of isolated surface hydroxyl groups depend on the
chemical structures of the oxides, and their detailed interpretation conversely enables conclusions
to be drawn about the structures of the oxide surfaces and about their active sites. The most
generalized treatment of the influence of crystalline structure on the IR spectra of surface OH
groups has been reported by Tsyganenko and Filimonov88.
A reason for the appearance of several bands due to free surface hydroxyl groups in the IR spectra
is the capability of the oxygen atoms of the OH groups to be in contact with several immediate
neighboring metal atoms. Hence, the number of the latter should exert a decisive influence on the
vibrational frequency, νOH, which is observed. For isolated (free) hydroxyl groups, the oxygen can
be bound to one, two, three, etc., metal atoms (types I, II, III, etc) – Figures 1.35.
Figures 1.35 – Three types of hydroxyls groups are possible at the surface of transition aluminas93.
The formation of a coordination bond is found to diminish the frequency of the OH stretching
vibration and hence the bands ascribed to the OH groups I, II and III will be positioned accordingly.
In order to establish a model for hydroxyl coverage, an analysis of the structure of specific crystal
planes has been carried out88, with the expectation being that the planes appearing during crystal
growth from the gas phase would be predominant. The oxygen of the OH group can form bonds
with several metal atoms if this is allowed by their mutual arrangements. The oxygen atoms of the
surface OH groups always occupy those positions where the O atoms would have been present in
the infinite lattice. The number of metal atoms around the oxygen of an OH group is always smaller
than the coordination number of oxygen in the lattice, i.e. the maximum number of OH groups for
the oxide in question equals the oxygen coordination number minus one.
The stretching frequencies of non-associated OH groups are thus determined mainly by their local
surroundings, that is, by the number of bound metal atoms and their chemical nature, and to a
lesser extent, by their coordination numbers. In accordance with data reported by Tsyganenko89,
the dependence of νOH for type I hydroxyls bound to atoms of different metallic elements is not a
smooth function of these atoms’ electronegativities or their positions in the Periodic Table. For
elements in the Second Period, the maximum occurs for aluminum, while for types II and III
hydroxyls, the maximum values shift towards the less acidic metals. The value of νOH increases in
going from type III to type I hydroxyls.
42
Table 1.IV-Positions of absorption bands of isolated OH groups on alumina and their possible
assignments93.
The conceptions which describe the surface properties of different aluminas are quite interesting.
Peri91 in 1965 first proposed a model of the surface of alumina which was founded on the
43
hypothesis that the planes exposed preferentially are those of index (100), thus explaining the
absorption bands observed in the Al 2 O 3 spectra of five types of surface hydroxyl groups
(Table 1.IV). Although this model does not adequately describe all of the surface properties, it is
still of considerable interest. In 1978, Knözinger and Ratnasamy89 proposed a very detailed OH
model as an extension of the Peri model91. The basic assumption of this model is that a mixture of
low-index crystal planes ((111), (110) and (100)) are exposed on the surface of the crystallites. The
relative abundance of different faces is assumed to vary for different aluminas. Five types of OH
groups were considered, corresponding to the coordination of the hydroxyl groups, either to
tetrahedral or octahedral aluminum anions, a combination of each, or to both (Figure 1.36). Five
absorption bands in the region 3800–3700 cm−1 have been assigned to these different types of OH
groups. The very important concept to arise from this analysis is that surface OH groups have
different net charges as a function of their environments.
Figure 1.36- Different surface OH groups on alumina according to the model of Knözinger and
Ratnasamy90.
For example, a type-III OH with a net positive charge of +0.5 is expected to be the most acidic one.
According to this model, it is possible to foresee an increase in the acidity of the Al3+ sites, and in
the basicity of the O2− sites, with the temperature of activation of the alumina since the lability of
the OH groups is a function of the basicity of the oxide.
Quantum chemical calculations are needed in order to analyze the nature of the anion hydroxyl
groups. In the literature, there are different viewpoints concerning assignment of the highest
frequency band at 3800 cm−1 according to Morterra and Magnacca79, this band belongs to an OH
group bound to tetrahedral aluminum, whereas Knözinger and Ratnasamy90 have attributed it to
the most basic OH group bound to an octahedral aluminum atom. Calculations of both the charge
on the hydrogen atom and the νOH of hydroxyl groups bound with different numbers of aluminum
atoms in different coordinations showed that the highest-frequency absorption band in the
spectrum of aluminum oxides is due to the OH groups bound to four-coordinated aluminum. These
results confirm the conclusion made by Knözinger and Ratnasamy90 that the OH groups bound to
octahedral aluminum have less acidity.
Qualitative quantum chemical models of the dependence of the acidity and frequency
characteristics of surface OH groups on both the number and electronegativity of the metal atom
connected with them have been examined by Pelmenshikov et al.91. This model has been used to
44
explain the lack of a smooth correlation between the νOH of mono-coordinated OH groups and the
electronegativity of the metal atoms in numerous oxides. The dependence of νOH on the
electronegativity of the metal atom (X M ) is found to be represented by a curve with a maximum
reached when q = 0 in the point of change in the polarity of the OH bond.
It is not possible to describe all of the properties of the active sites, in particular, for the OH groups,
because of the use of various approximations in the supposed models. For alumina oxides, Busca
et al.81,94 and Della Gatta et al.95 have proposed models based on the presence of cation vacancies
and the corresponding OH structures are reported in Figures 1.37.
The investigation of transition aluminas showed that the IR spectra of Al 2 O 3 are also more complex
and contain at least nine quite clearly resolved absorption bands of hydroxyls which are not
hydrogen-bonded. The question of the assignment of the bands to free or H-bound OH groups of
alumina has been analyzed in detail by Chukin and co-workers93.
Trokhimets et al.93 interpreted the observed bands on the assumption that the vibration frequency
of the OH groups depends on the coordination number of both aluminum and oxygen.
It was suggested that the aluminum coordination number on Al 2 O 3 surfaces can be equal to five,
as well as four and six. The different types of hydroxyl group absorptions, their effective charges
and suggested assignments are listed in Table 1.IV above. As follows from this table, the values of
the effective charges on OHs in groups I, II, and III form three almost nonoverlapping regions.
Figures 1.37- Possible OH structures and corresponding νOH frequencies, at the surface of defectcontaining spinel transition aluminas. (Symbols: □, cation vacancy; ν, average value of νOH
frequency)79.
Detailed data on such spectral manifestations of hydroxyl groups on aluminas are reported
elsewhere79,90. Chukin and co-workers93 have presented a model of the surface of γ-Al 2 O 3 , which
easily explains both hydration and dehydration of the Al 2 O 3 surface, as well as the reversibility of
dehydroxylation and rehydroxylation. On the basis of an examination of the decomposition
mechanisms on bohemite–corundum faces, a model for the primary crystal lattice was developed,
based on IR spectroscopic data and crystal structure analysis of the oxides and hydroxides of
aluminum. A joining of dehydroxylated boemite ‘packets’ during heat treatment is assumed to be
accompanied by the formation of Al–O–Al bonds and migration of a portion of the Al3+ cations into
the freshly formed octahedral cationic vacancies. Three types of Lewis acid sites and six types of
OH groups have been identified on the fully hydroxylated γ-Al 2 O 3 . According to Chukin and
Seleznev93, the partially dehydroxylated surface of γ-Al 2 O 3 consists of seven types of electron-
45
accepting centers, three types of extra lattice Al3+ cations, two types of Al cus 3+ (coordinatively
unsaturated site), and two types of electron-deficient O atoms.
The values of proton affinity (PA) for different surface hydroxyls of Al 2 O 3 (Table 1.V), calculated
from the shifts of the OH frequencies due to formation of H-bonds with bases9can be used to
compare the chemical properties of this oxide surface hydroxyls.
Table 1.V- values of Proton affinity for different surface hydroxyls of Al 2 O 3 93.
In order to determine the number of proton-containing sites, two approaches have been used. One
of these is based on measurements of the number of OH groups directly from the intensities of the
νOH bands; the second involves measurement of the number of protonated bases generated.
If OH groups are not isolated, the concentration of protonic centers is determined by the intensity
of the bands in spectra of protonated probes (ammonia, pyridine, etc.), as, for example, has been
done in the case of other materials like supported heteropoly acids (HPAs), sulfates, etc.
1.8.4
The possible role of defective crystal configurations
Few last considerations remain to be done on the surface hydroxyls of aluminas and in particular
on the OH species absorbing at ca. 3775 cm -1. Its unique behavior deserves here some further
comment.
(I) The band is present only on (all) transition aluminas, whereas it is totally absent on boehmite
(Figure 1.38) and on well crystallized α-Al 2 O 3 79,81. It is thus certainly ascribable to OH groups
involving in their coordination sphere AIIV ions and this is either in agreement or compatible with
most of the OH models proposed. What the various models proposed do not explain is why the OH
band at ca. 3775 cm-1 is by far the sharpest and the most reactive OH species at the surface of
aluminas.
Figure 1.38- The high wavenumbers region of the OH spectrum of some Al oxides. (a) γ-AlOOH
activated at 300 K, (b-e) η-Al 2 O 3 , γ-Al 2 O 3 , θ-Al 2 O 3 and α-Al 2 O 3 activated at 773 K79.
46
Busca's model does predict a different activity for this OH species (□-AllV-OH) with respect to its
high frequency partner (AIIV- OH; v = 3800 cm-1), but the different activity should be expected to be
a higher basicity. What is found in practice is the 3775 cm-1 OH species is more active in respect to
all types of molecules and participates more actively in all catalytic reactions involving OH groups.
The higher activity of the 3775 cm-1 species reflects mainly a higher accessibility of the OH species
to all types of surface probes and this should reflect the possible presence of the OH group in
particularly exposed zones of the surface. For this reason, Morterra et al.86 have attributed the OH
band at 3775 cm-1 to AIIV-OH groups present in portions of the surface belonging to
crystallographically defective configurations.
The latter are expected to be quite frequent in porous systems of high surface area and poor
crystallinity; it is here recalled that Soled92 showed how the surface area of γ- and η-Al 2 O 3 must be
high for structural reasons (and, in fact, it is normally as high as 200-250 m2 g-1).
None of the models discussed so far indicates a special accessibility for the OH species
responsible for the 3775 cm-1 band. This is possibly so because all models discuss the surface of
spinel aluminas only in terms of regular crystal plane terminations (the 'top' termination of particles)
and do not consider the large incidence of structural defects in high area porous materials (the
'side' terminations of particles).
The 3775 cm-1 OH band has been considered, in all models proposed, as due to a 'regular' OH
species, whereas it is probably not. The unique nature of the OH band at 3775 cm-1 is
demonstrated also by its lability.
It was previously reported by Zecchina et al.98 and it is here confirmed by the spectra in
Figure 1.39, that on low-temperature transition aluminas (and especially on η-Al 2 O 3 ) the thermal
elimination of the OH band at 3775 cm-1 (at temperatures as high as ca. 1100 K) occurs
irreversibly.
Figure 1.39- The OH spectral pattern of an η-Al 2 O 3 sample treated in various conditions. (a-c) the
starting sample, activated at 673, 773 and 1023 K respectively, (d-f) after activation at 1123 K
(almost complete dehydroxylation), the samples was retreated at 300 K, and further activated 673,
773 and 1023 K respectively79.
47
Rehydration at ambient temperature of the virtually fully dehydrated alumina, followed by a second
dehydration run, yields an OH pattern in which the relative intensities of the various OH species
are altered and, in particular, the band at 3775 cm-1 is very scarce and, sometimes, virtually
absent. Only if the rehydration process is carried out with water vapor at temperatures as high as
670 K, in a sort of hydrothermal process, does the band at 3775 cm-1 recover most of its original
intensity87.
This indicates that, during the high temperature dehydration of aluminas, reconstruction effects are
operative and that the most defective configurations (i.e., those yielding the most reactive surface
sites) tend to be annihilated.
The interpretation here proposed for the OH band at 3775 cm-1 is confirmed by the behavior of the
high temperature transition aluminas. As mentioned in the introduction section, structurally these
aluminas are still defective spinel systems, but characterized by higher crystalline order, larger
crystallites and more regular (i.e., sharper) terminations of the particles. In the OH spectral region
(curve d of Figure 1.38), the higher order of the high-temperature transition aluminas is reflected by
a clear splitting of the OH bands of type II (two components are resolved at ca. 3745 and ca. 3730
cm-1) and sometimes also of the OH bands of type III (two components are often resolved at ca.
3710 and ca. 3680 cm-1 93). The OH species at 3775 cm-1, that was tentatively ascribed to AllV-OH
groups in exposed and/or defective crystallographic configurations, is either very weak (curve d of
Figure 1.38) or totally missing99.
Also the adsorption of CO confirms this: on δ-Al 2 O 3 94 and on θ-Al 2 O 3 96 a band of strongly
adsorbed CO, located at ν > 2230 cm-1, is either missing or very weak.
1.8.5
The basicity of aluminas
The surface basicity of transition aluminas is quite low: in fact, Al 2 O 3 -based catalysts are fairly
important, on a catalytic and on a conventional chemical ground, for their acidity rather than for
their basicity. As a consequence, whenever a basic catalyst or catalyst support is needed, either
other oxides are resorted to or aluminas are doped with variable amounts of basic elements.
For a long time the surface basicity of oxides has been tested by the gas-solid adsorption of CO 2
that seems to be still the routine way to evaluate the basicity of aluminas and of its changes with
surface chemical modifications.
Detailed results resulting from IR spectroscopic investigations of CO 2 adsorption over different
oxides have been presented by Busca and Lorenzelli80. It should be taken into consideration that
there are differences in the spectral images (structures of surface carbonates) of CO 2 adsorption
on oxidized and reduced samples of the transition-metal oxides. In addition, CO 2 can be
decomposed on reduced surfaces of transition-metal oxides, oxidizing the surface and producing
carbon monoxide.
48
Figure 1.40- Spectra of end-on surface complexes of CO 2 on various oxides (PCO 2 =12 Torr), (a)
γ-ALOOH activated at 300 K, (b-c) γ-Al 2 O 3 activated at 373 and 773 K, (d) θ- Al 2 O 3 activated at
1023 K, (e) α-Al 2 O 3 activated at 1023 K79.
Because CO 2 molecules interact specifically with the cations, and their spectral characteristics
depend on the cation properties, the spectra of adsorbed CO 2 molecules have been used to study
the properties of aprotic sites on oxides. The spectrum of CO 2 molecules adsorbed on the highly
dehydroxylated surface of alumina–silica gel96 has an absorption band at about 2375 cm−1, which
gradually shifts toward higher frequencies as CO 2 molecules are removed by evacuation at room
temperature. It is assumed that the adsorbed CO 2 molecule preserves its linearity, and is adsorbed
on centers through ion-quadruple interactions. In addition to the main intense absorption band at
2375 cm−1 (Figure 1.40), there are two absorption bands at 2405 and 2355 cm-1 in the spectra.
Peri96, when studying the CO 2 adsorption on Al 2 O 3 , these to rotation branches of the absorption
bands of the adsorbed molecules. It is assumed, moreover, that the adsorption of CO 2 proceeds
selectively on a small fraction of the aprotic Lewis centers, namely, on the so-called α-centers.
According to Peri97, the centers which selectively adsorb CO 2 molecules also selectively absorb
the molecules of butene, acetylene and HCI. Molecules of H 2 O and NH 3 are adsorbed fairly well
on all aprotic centers, including the α-centers.
The carbonate–carboxylate structures are characterized by absorption bands in the region 1300–
1900 cm-1 which originated on the surface, not only during the CO 2 adsorption, but also due to
adsorption with decomposition of several other molecules85,86. These compounds can be produced
during the pre-treatment of the sample in vacuo at high temperatures due to the oxidation of
organic contaminants, probably due to vapors from greases and oil of vacuum pumps. Moreover,
they are produced during the adsorption of the simple molecules like CO 2 and CO96. Thus, the
spectrum of CO 2 adsorbed on alumina has absorption bands at ca. 1750, 1635, 1500 and 1235
cm-1 attributed to carbonate surface structures. The absorption band of molecularly adsorbed CO 2
was observed at ca. 2350 cm-1. Carbon dioxide is strongly chemisorbed by alumina at 298 K: in
this case, the spectrum has absorption bands at ca. 1770, 1640, 1480 and 1320 cm-1 93.
49
Peri 96,97 established the presence of at least three different types of compounds formed during the
adsorption of CO 2 on an alumina surface evacuated at 873 K. The absorption bands in the 1800–
1870 cm-1 region were assigned to molecularly adsorbed CO 2 . An interpretation of the spectra of
CO 2 adsorbed on Al 2 O 3 has also been made by Fink93 (Table 1.VI). Surface compounds of the
types I–III are assumed to be in equilibrium. A type-II compound predominates on the surface at
233 K, while a type-III compound predominates at 298 K. A compound of type- III remains on the
surface evacuated at temperatures of up to 473 K. Adsorption at temperatures above 473 K leads
to the formation of a Type-IV compound.
Table 1.VI- Types of surface compounds formed during the chemisorption of carbon dioxide on the
surface of γ-Al 2 O 3 93.
Figure 1.41- IR absorbance spectra of CO 2 adsorbed, in a bent form, on various Al oxidic systems.
(a) γ-AlOOH activated at 300 K, (b-d) γ-Al 2 O 3 activated at 300, 773 and 1023 K, (c) θ- Al 2 O 3
activated at 773 K, (f) α-Al 2 O 3 activated at 1023 K79.
50
The difference in the behavior of the absorption bands at 1480 and 1400 cm-1 during the
adsorption of CO 2 on alumina at temperatures ranging from 293 to 523 K indicates that
monodentate carbonate structures are not formed at the surface. The absorption bands at 1650
and 1230 cm−1 were attributed to bidentate carbonate structures. There were also differences in
the interaction of adsorbed CO 2 molecules with different types of surface hydroxyl groups of
alumina.
Figure 1.42 – Schematic representation of the band positions of carbonate-like species at the
surface of metal oxides79.
The interaction of CO 2 with the surface oxygens of different oxide produces spectra implying the
formation of several different types of carbonates in the region 1200–1800 cm-1, which is typical for
the stretching vibrations of CO bonds in individual carbonates, frequently, the spectra of the
surface compounds are identical to the spectra of bulk carbonates, thus implying the presence of
free CO 3 2− ions. IR spectra of carbonates can be used to establish the type (and sometimes the
structure) of the carbonates formed on the oxide surfaces91,99.
1.8.6
The acidity of aluminas
In the case of transition aluminas, whose surface acidity is by far the most important feature, the IR
technique has been used quite extensively with great variety of adsorbing probe molecules. Merits
and limits of the various IR-adsorption procedures adopted have been reviewed by several
authors.
The study of ammonia and pyridine adsorption on oxides by means of infrared spectroscopy is a
classical method for identifying both Brønsted and Lewis acid sites. The formation of NH 4 + (PyH+)
ions is a criterion of the presence of Brønsted acid centers, while the presence of coordinated
ammonia (pyridine) shows that Lewis acid centers are present at the surface. The spectral features
of the coordinated ammonia (pyridine) molecule are significantly different from those of ammonia
(pyridinium) ions (Table 1.VII). This allows the identification of the formation of such complexes by
infrared spectroscopy. The greatest differences are observed in the spectral region of the NH
deformation vibration frequencies.
51
Table 1.VII- Spectral characteristics of NH 3 molecule93.
Figure 1.43– Schematic representation of the spectral range of the 8a-8b, 19a-19b modes for
some py-containing systems93.
Due to their strong basic properties and proton affinities, both ammonia and pyridine are good
probe molecules for establishing the presence of even weak Lewis and Brønsted acid sites on the
surface. However, the distinctions that can be made by these two probe molecules are not
identical. The first difference between them is their relative sizes, i.e. NH 3 < pyridine. Another
difference is their relative basicity: in an aqueous solution, ammonia is a stronger base than
pyridine99, the pKa value for ammonia is around 9, while that for pyridine is about 5. However, in
the gas phase the basicity of pyridine is significantly greater than that of NH 3 103.
52
It could be argued that for metal surfaces 103,104, gas-phase basicity is a more appropriate measure
of the basic strength of the adsorbate, as no solvation effects occur during adsorption. For metal
oxides, the situation is more complicated. When the adsorption proceeds via coordination to the
metal cation, the gas-phase basicity is probably the appropriate one to apply. However, when
hydrogen-bonding or adsorption on a Brønsted acid site takes place, neighbouring anions or
hydroxyl groups may be involved in the interaction100.
1.8.7
Ammonia adsorption
Ammonia can be bound to the oxide surface by (i) a hydrogen-bond, NH· · ·O2−, with a surface
oxygen or with the oxygen of a surface hydroxyl group, (ii) a bond between the nitrogen atom and
a surface hydroxyl atom, or (iii) a coordination bond with a surface cation (Lewis acid site):
Figures 1.44 – Schematic representation of the ammonia molecule bounds on a oxide surface93.
Complete proton transfer can occur with the formation of an NH 4 + ion (Figures 1.44) or
alternatively ammonia dissociative adsorption with the formation of surface NH 2 and OH groups
can also take place (Figures 1.45):
Figures 1.45 – Schematic representation of the formation of surface NH 2 93.
Each type of surface complex gives information about the types of surface centers through
qualitative differences in its spectral manifestations and can also reflect qualitative differences in
the properties of the same type of center.
The spectral manifestations of adsorbed ammonia have been analyzed in detail by Filimonov and
co-workers and Davydov 93. The mechanism of ammonia adsorption depends significantly on the
state of the hydroxyl coverage of the oxide surface. In the case of strongly hydroxylated surfaces,
the adsorption predominantly occurs on the surface hydroxyl groups via the formation of hydrogenbonds between these OH groups and nitrogen atoms (the band δ s NH 3 in such a form of adsorption
appears in the region 1150–1100 cm−1). This form of adsorption is thermally unstable and is also
reversed by evacuation at room temperature.
The number of coordinatively unsaturated surface cations capable of forming a coordinative bond
with ammonia molecules(δ s NH 3 at 1280–1150 cm-1) grows with the dehydroxylation of the oxide
surface. The detailed position of the δ s NH 3 band is, in this case, sensitive to the electron-acceptor
ability of the cation and increases when the cation electronegativity increases93. The δ s -NH 3 bands
can be also observed at lower frequencies (<1150 cm-1). In such cases, it is difficult to distinguish
these bands from those of hydrogen-bonded ammonia. It is then necessary to use additional
procedures to prove that the relevant species is formed and hence that sites of a corresponding
53
nature are present on the surface. The frequencies of different ammonia complexes observed in
the spectra of ammonia adsorbed on a series of simple and complex oxides are collected in
Table 1.VIII. In the spectra of several oxides, especially those pre-evacuated at high temperatures,
bands attributable to surface hydroxyl groups appear and grow in intensity upon the adsorption of
ammonia. This shows that some of the ammonia molecules dissociate on the surface of these
oxides, forming OH and NH 2 groups, thus revealing the presence of surface acid–base pairs such
as Mn+O−2. It should be noted that spectral identification of NH 2 groups by using only the
absorption bands of the νNH and δNH types, without taking into account the appearance of the
surface hydroxyl groups, is not reliable.
Table 1.VIII – Special features of coordinated ammonia and ammonium ion on alumina93.
The δNH 2 and νNH 2 values for both metal amides M(NH 2 ) n and surface NH 2 groups occur over
wide ranges of 1490–1630 and 3200–3580 cm−1 (Table 1.IX). As shown, the position of δNH 2 of
such groups is overlapped by the δ as of coordinated NH 3 .
Table 1.IX- Vibrational frequencies of NH 2 groups93.
This is why there are difficulties in assigning bands to the vibrations of NH 2 species adsorbed on
oxides, and especially on those oxides with strong electron-acceptor sites, because both NH 2
groups and coordinatively bound ammonia can be present on the surface simultaneously. The
covalence of the M–N bond increases with increasing metal electronegativity, as is shown by the
spectra of the alkali-earth metals. This is accompanied by a simultaneous increase in all
frequencies of the NH 2 vibrations.
1.8.8
The adsorption of carbon monoxide
The adsorption of CO on numerous oxides has been investigated in order to identify both Lewis
acid sites and surface oxides. Some of the data obtained are presented in Table 1.X. To describe
the detailed structure of the surface and active surface sites, examination of such systems has
been made on the basis of both the crystallographic structure of the oxides and data obtained for
the number and properties of hydroxyl groups.
Table 1.X- Spectral features of CO adsorbed on the surfaces of the various oxide systems32.
54
There are a number of studies in the literature of the interaction of CO with aluminas at different
temperatures 79-93. There are no doubts that the strongest centers of the alumina surface indicated
by the CO adsorption (the band at 2235 cm−1) are due to Al3+ cus ions 79,90,93. Detailed investigations
showed that these centers are involved in the so-called ‘X-centers’, the concentrations of which are
small (about 1016 centers m−2). X-centers are acid–base pairs, e.g. Al3+ cut O2−. Support for this
assignment comes from the absence of dissociative adsorption of both ammonia and propene
upon their adsorption on Al 2 O 3 samples treated with NaOH. The band at 2235 cm-1, characteristic
of the strongest centers of the alumina surface, is not obtained for the latter types of sample. The
nature of the complexes characterized by the absorption band at 2215 cm−1 which is also observed
upon CO adsorption at room temperature, is not so clear. However, they are probably also due to
the presence of Al3+ cut ions79.
A comparison of the spectra of CO adsorbed on Al 2 O 3 at different temperature is shown below in
Figure 1.46. At room temperature, two bands at 2210 and 2235 cm-1 are identified, while as the CO
pressure increases, the intense absorption of CO in the gas phase appears in the 2100–2200 cm-1
region so that it is difficult to identify the weakly bound complexes on the surface with the lowerfrequency bands91. On saturation of the surface at low temperature, three new bands appear in the
spectrum. One of these belongs to physically adsorbed CO because its position corresponds to
that in the gas and liquid states of the molecule: νCO = 2140–2150 cm-1.
The 2160 cm-1 band corresponds to CO adsorbed on OH groups through hydrogen-bonding, as
confirmed a decrease in the intensity of the original OH groups with the simultaneous appearance
of lower-frequency νOH bands. Similar results have been obtained by Zaki and Knözinger 90.
Bands in the region above 2180 cm-1 belong to the complexes of CO with Lewis acid sites. Thus,
the bands in the 2195–2210 cm-1 (species (CO) A ), 2215–2220 cm−1 (species (CO) B ) and 2235–
2240 cm-1 (species (CO) C ) regions are assigned to the presence of coordinatively unsaturated
tetrahedral Al3+ ions. The higher-frequency (CO) C and (CO) B bands have been assigned to two
families of sites involving Al3+ cut ions located in crystallographically defective configurations, and
the lower-frequency adsorbed species (CO) A to sites involving particularly exposed Al3+ cut ions
located in extended areas of regular low-index crystal planes79.
Fig. 1.46 – IR spectra of CO adsorbed at 293 K (1) and 173 K (2) after dehydration at 773 K: (1,2)
γ-Al 2 O 3 93.
It should be pointed out that Morterra and Magnacca107 have suggested that Al3+ cuo (coordinatively
unsaturated octahedral site) ions located in octahedral positions cannot be coordinatively
unsaturated and adsorb CO. The absence of absorption bands in the IR spectra of CO adsorbed
on oxidized α-Fe 2 O 3 91 and α-Al 2 O 3 106,107 provides evidence for this suggestion. However,
according to calorimetric data, there is a form of adsorption of CO with a heat of adsorption of
around 36 kJ mol−1 in the case of α-Al 2 O 3 . Such a value corresponds to the absorption band at
55
about 2170–2180 cm-1 which was not observed for these samples, probably due to their low
specific areas (2 m2 g−1) and small number of such active centers.
In several studies the existence of a great number of Lewis acid sites, including three-, four- and
five-coordinated Al3+ cations, as well as pairs of Lewis acid sites which form upon the removal of a
bridge oxygen atom, has been analyzed. Borovkov et al.93 observed two bands in the
2210–2215 cm-1 region and at 2245 cm-1 in the infrared spectra of CO adsorbed on different Al 2 O 3
samples. The high-frequency band is attributed to CO molecules adsorbed on α- or X-centers,
which include the Al+ O− acid–base pairs.
The concentrations of the different surface centers are important characteristics, and many authors
have tried to estimate these by different ways, including IR spectroscopy of adsorbed CO. The
Lewis acid site concentration on transition aluminas is quite high, depending on the degree of
dehydration, and amounts to 0.015–0.03 sites nm−2. Knözinger and Ratnasamy90 have given value
of about 0.1 sites nm−2. According to Morterra and Magnacca79, the concentration of Lewis acid
sites on γ-Al 2 O 3 activated at 773 K equals 0.1 molecules nm−2 and 0.23 molecules nm−2 on
alumina activated at 1023 K. The distribution among the different sites is 0.14 molecules nm−2 of
strong acid Lewis sites on the crystal planes, about 0.06 molecules nm−2 for strong Lewis acid sites
in crystal defects, and about 0.03 molecules nm−2 for the strongest and more defective sites. The
maximum concentration of pyridine strongly adsorbed onto the Al3+ cut (coordinatively unsaturated
tetrahedral site) ions (derived from the band at 1625 cm-1) is three to four times larger than the
maximum overall CO uptake obtainable at 300 K, due to the fact that pyridine is a stronger base
than CO and can detect also weaker sites.
56
CHAPTER
2
Surface modification
of a
transition alumina
57
Surface modification of a transition alumina
2.
Introduction
The request to produce fully dense nanostructured ceramics has received much attention over
the past 10 years, focusing the interest on the processing of ultra-fine and low-agglomerated
primary particles, required for obtaining highly dense ceramics with tailored microstructures. As
it was reviewed in the previous chapter, this requirement becomes particularly strict in the case of
transition aluminas, whose metastability has a critical influence on their sintering behaviour.
The main purpose of the first part of this thesis is to correlate the chemical-physical surface
properties of transition alumina to its sintering behaviour. The final goal is pursued by inducing
surface modification by dispersion steps, namely magnetic stirring, ball-milling and attrition
milling, for improving the densification during sintering obtaining fully dense ceramics with tailored
microstructures.
For this aim a transition alumina was employed. The optimization of powder thermal reactivity
was studied by means of a synergic action of several chemical-physical characterization
techniques and compared to the as-received material.
2.1
Textural characterization of the starting material
The development of this work has firstly carried out by using a commercial transition alumina
powder (Nanotek® by Nanophase Technologies Corporation, Darien, IL, U.S.A.).
This alumina powder is prepared by Physical Vapor Synthesis (PVS) as it was published in
literature1. This powder, labeled now as A, is characterized by an average particle size of 47 nm,
a specific surface area of 35 m2/g and it is composed by two phases, δ and γ alumina, as it is
claimed by the producer2.
Intensity (a.u.)
The XRD pattern of the as-received powder is presented in Figure 2.1. The XRD pattern reveal
that powder is a mixture of transition alumina phases, precisely δ-Al 2 O 3 (ICDD file n°. 040877)
and γ-Al 2 O 3 (ICDD file n°. 48-0367).



  


  
10
20



30
40
50


60
70
Diffraction Angle (2.)
Figure 2.1- XRD pattern of A powder.
58
In order to evaluate the particle average size, the as-received powder was submitted to HR-TEM
observation as it is shown in Figure 2.2.
50 nm
Figure 2.2- XRD pattern of A powder.
The observation allowed to define the primary particle size distribution, reported in the following
image Figure 2.3. A mean size of about 37 nm was determined, with a geometrical standard
deviation of 20.5 nm after analysing some micrographs with a number of particles ranging from
120-150. The result of this analysis is in good agreement with the result published by Azar et al.3.
Percentual Frequency [%]
18
16
14
12
10
8
6
4
2
0
0
20
40
60
80
100
120
Particle Size [nm]
Figure 2.3 – Particle size distribution of A powder.
Similarly, the powder placed on a graphite disk was submitted to SEM observations, Figure 2.4.
The powder is characterized by large agglomerates/aggregates; so, the powders was submitted
to several dispersion process, as will be discussed in the following.
100 m
Figure 2.4– SEM micrograph of the A powder.
59
The initial granulometry of A powder has been performed by using laser granulometry after
dispersing the powder in distilled water in order to obtain a suspension having a dilution factor of
7 vol.%. The instrumental detection range is included between 300 nm and 300 m. The
cumulative size distribution by both volume and number, as a function of agglomerate size, it is
presented in Figure 2.5.
Cumulative frequency (%)
100
80
60
40
20
0
0
5
10
15
20
25
Agglomerate Size [m]
Figure 2.5 - Cumulative size distribution by volume (solid line) and by number (dashed line) as a
function of agglomerate size of A powder
As it is shown, in nanometric powders the presence of a low percentage of agglomerates could
strongly modify the volume distribution. In contrast, in the case of number distribution it is not
possible to evidence the presence of these agglomerates. Agglomerate sizes, corresponding to
(d 10 ) 10, (d 50 ) 50 and (d 90 ) 90% of the cumulative volume distribution, were 1.7, 5.5 and 10.4 μm,
respectively. In the case of the number distribution, the agglomerates sizes are <0.3, 0.36,
0.89 μm corresponding to (d 10 ) 10, (d 50 ) 50 and (d 90 ) 90%, respectively.
Herewith, the crystallization of the A powder into the stable α-phase was evaluated by means of
simultaneous differential thermal analysis up and thermogravimetry. The test was carried out up
to 1450°C in static air (heating rate of 10°C/min). In following Figure 2.6 are shown the results
from DTA-TG tests.
-40
0
-30
-20
DTA [V]
0
-4
10
20
TG [%]
-2
-10
1325.3°C
-6
30
40
Exo
50
-8
60
200
400
600
800
1000
1200
1400
Temperature [°C]
Figure 2.6 - DTA-TG Curves of the A powder: dotted line corresponds to DTA and the solid to TG,
respectively.
60
DTA curve exhibited an exothermal signal at about 1325°C, imputable to the phase
transformation θ→α, as reported in literature4. TG analysis revealed a limited weight loss of about
2.05 %.
BET (Brunauer, Emmer, Teller) test was performed by outgassing at 150°C for 2 h. The test was
conducted by N 2 adsorption/desorption isotherm at 77K, i.e. after removal of water and surface
contaminant. In Figure 2.7 are shown both curves correspond of Type II isotherms, typical of
either non-porous adsorbents or relatively large pores. The BET SSA measured on the A sample
was 34.5 m2g-1 in good agreement with the information stated by the producer.
35
A
30
3
-1
Volume (cm g )
25
20
15
10
5
0
0,0
0,2
0,4
0,6
0,8
1,0
p/p0
Figure 2.7- N 2 adsorption/desorption isotherms at 77 K of A powder
out-gassed at 150°C: full and white refer to adsorption and desorption.
Sintering behaviour of A was investigated by dilatometric analysis performed on uniaxially
pressed bars (at 350 MPa), by heating up to 1500°C (heating rate of 10°C/min), with a soaking
time of 3 h at the maximum temperature and cooling down up to 20°C (cooling rate of 20°C/min)
Zone II
-0,02
-0,0002
-0,04
-0,0004
-0,06
-0,0006
L/L
o
0,0000
-0,08
-0,0008
-0,10
-0,0010
-0,12
-0,0012
-0,14
-0,0014
200
400
600
800
1000
1200
1400
derivative signal
Zone I
0,00
1600
Temperature [°C]
Figure 2.8– Dilatometric (solid line) and derivative (dashed line) curves of A.
Finally, in order to complete the characterization of the as-received powder a dilatometric test
was performed on uniaxially pressed bars (350 MPa) by heating up to 1500°C (heating rate of
10°C/min) and soaking time of 3 h at maximum temperature.
61
The dilatometric curve (Figure 2.8) shows a typical example of a transition alumina, characterized
by two regimes of densification during constant heating, named zone I and II. The zone I is
imputable to the transformation into α-phase, associated to a volume decrease and
rearrangement phenomena in good agreement with literature data4. The peak on the derivative
curve located at about 1147°C indicates the temperature associated to δ→α transformation. The
densification step (zone II) is located from 1170°C to the maximum sintering temperature which is
associated to the α-alumina densification. A certain disagreement between the derivative peak
from dilatometric analysis and the value determined by DTA can be imputed to the nature of the
samples, powder compact for the former and powders for the latter. In the derivative curve, the
peak related to the temperature of maximum densification rate is not observable.
The green density evaluated by weight and geometrical measurements of pressed bar is
1.86 g/cm3 which corresponds to 53.3 % of the theoretical density (taking in consideration that
Nanotek3 has a reference value given by the producer of 3.49 g/cm3). The fired density was
2.79 g/cm3 which implies a 70.4 %TD (reference value for α-alumina 3.96 g/cm3).
2.2
Effect of dispersion on powder granulometry
In order to reduce the starting agglomerate size, A powder was dispersed by magnetic stirring in
pure distilled water, by preparing aqueous suspensions with a solid content of 50%, and
maintained under magnetic stirring for 170 h. This powder is labeled as A MS .
The pH was measured on the suspension as a function of time. The suspension presented a pH
of about 5.1 for the as-dispersed powder and it stabilized to the value of 5.6 after 24 hours of
magnetic stirring, remaining almost stable. It is possible to assume that 5.6 is the natural pH of
the suspension.
After dispersion, the cumulative size distribution by volume of the dispersed samples are
collected in Figure 2.9, showing a significant reduction of soft agglomerate size.
Cumulative Distribution [%]
100
80
60
40
Samples
d50 [µm]
A
AMS
8,08
0,44
20
0
0
4
8
12
16
20
24
Figure 2.9 - Cumulative distribution by volume of: A (solid line, no symbols) and A MS (circles).
Insert: d 50 values of the overall distributions
After dispersion, agglomerates size corresponding to (d 10 ) 10, (d 50 ) 50, and (d 90 ) 90% of the
cumulative volume distribution, were <0.3, 0.44 and 0.75 μm, respectively. In Figure 2.10 is
illustrated as a comparison the evolution of d 50 during dispersion up to 170 h. As it is shown d 50
was successfully reduced in 1 order of magnitude from 8.08 μm to 0.44 μm.
62
9
A
8
7
d50 [m]
6
5
4
3
2
AMS
1
0
0
20
40
60
80 100 120 140 160 180
stirring time (h)
Figure 2.10- Evolution of d 50 values as a function of the dispersion time under magnetic stirring
2.2.1
Effect of the pH of the aqueous suspension on the agglomerate size
A parallel study has been done in order to study the effect of the pH on de-agglomeration. On the
ground of the results reported in literature (Figure 2.11), the Nanotek powder has the isoelectric
point at about 9.3. At the isoelectric point, the surface charge of the particles is near zero and the
absence of a surface charge implies agglomeration.
Figure 2.11 – Zeta potential of the Nanotek powder as a function of pH5.
In this context, a stable suspension can be prepared at pH higher that 10 or lower than 8. Taking
in consideration the zeta potential curve, pH equal to 4 should be considered as the best
condition for the dispersion.
Nanotek was dispersed by magnetic stirring, as in the previous case having a solid content of
50 wt%. Diluted HCl was added to decrease the pH up to 4. The pH was monitored and corrected
every 24 hours. As it is shown in Figures 2.12 / 2.13 , after 170 h of dispersion agglomerates size
63
corresponding to (d 10 ) 10, (d 50 ) 50 and (d 50 ) 90% in volume, were <0.3, 0.44 and 1.11. If
compared to the natural, unmodified pH suspension, no significant improvement was found.
100
80
60
Samples
A
AMS pH=4
40
d50 [m]
8.08
0.64
20
0
0
4
8
12
16
20
24
Figure 2.12 – Cumulative distribution by volume of: A (solid line, no symbols) and A MS pH=4
(circles). Insert: d 50 values of the overall distributions
9
A
8
7
d50 [m]
6
5
4
3
2
AMS pH=4
1
0
0
20
40
60
80
100
120
140
160
180
stirring time (h)
Figure 2.13 – Evolution in time of the Agglomerate Size (d 50 ) in function of the dispersion time of
A MS pH=4.
A second test has been conducted at pH equal to 3, with the aim of evaluating the effect of a
lower pH on the powder dispersability. As it is shown in figures 2.14/2.15 , the agglomerate sizes
corresponding to (d 10 ) 10, (d 50 ) 50 and (d 90 ) 90 % in volume were <0.3, 0.53 and 0.9 μm. As a
conclusion, the modification of the slurry pH from natural value (5.6) to 4 and 3 did not produce
any significant effect neither on the final particle size, nor on the dispersion time. On the ground
of such results, the following dispersions were carried out on unmodified suspensions, left to their
natural pH.
64
100
80
60
d50 [m]
Samples
A
AMS pH=3
40
8.08
0.53
20
0
0
4
8
12
16
20
24
Figure 2.14 – Cumulative distribution by volume of: A (solid line, no symbols) and A MS pH=3
(circles). Insert: d 50 values of the overall distributions
9
A
8
7
d50 [m]
6
5
4
3
2
AMS pH=3
1
0
0
20
40
60
80
100
120
140
160
180
stirring time (h)
Figure 2.15 – Evolution in time of the Agglomerate Size (d 50 ) in function of the dispersion time of
A MS pH=3.
2.2.2
Comparison among other dispersion routes
As in the previous case, powder A was submitted to ball milling with the aim of comparing the
effectiveness of such method with the lower energetical magnetic stirring. The angular velocity
was optimized by following the literature data6,7. The test were conducted under a nominal
angular velocity ω n , using the following formula (1) taken from literature:
65
n  0.6c 
25.4
Di
where
c 
42.3
Di
(1)
Figure 2.16 – Scheme of Ball Mill used for this study6.
A PE container of 50 ml. volume was employed to determinate the nominal rotational velocity
(see Figure 2.16). The test was followed up by testing both α-alumina (Bitossi-Φ= 2 mm, A BMα )
and zirconia (TSZ TOSOH- Φ= 1,75 mm, A BMz ) spheres (powder to spheres weight ratio of 1:10).
The solid content of 50wt.% was selected to follow these tests. After 3 hours of dispersion,
similar agglomerates sizes were found for A BMα and A BMz and they were similar to A MS . The
results of the cumulative size distribution by volume for both spheres nature are collected in
Figure 2.17.
100
80
60
40
20
0
0
4
8
12
Samples
d50 [µm]
A
ABM
8,08
0,50
ABMz
0,58
16
20
24
Figure 2.17 - Cumulative distribution by volume of: A (solid line, no symbols), A BMα (triangles) and
A BMz (squares). Insert: d 50 values of the overall distributions.
A BMα and A BMz powders were submitted to XRD analysis, in order to evidence if powders were
contaminated by the milling media. No -alumina or zirconia XRD reflections were found in the
dispersed powders, as shown in Figure 2.18: if contaminated, the pollutants amount was so low,
not to be detectable by XRD.
66
Intensity (a.u)



  
 

 
 
 



ABM



  
10
20
 
 
 
 


30
40
50


ABMz
60
70
Diffraction Angle (2.
Figure 2.18 – XRD patterns of Nanotek after ball milling.
The last technique employed to disperse the powder was attrition milling. This dispersion route
was selected in order to compare the performance of this higher-energy method among the
others. In this particular case, the sample was labelled as A AM . The test was conducted using αalumina (Bitossi-Φ= 2 mm) spheres (powder to spheres weight ratio of 1:10). As in the previous
case, the angular velocity was set up following literature data9-10.
After 6 h of attrition milling, the cumulative size distribution by volume of the dispersed samples
corresponding to (d 10 ) 10, (d 50 ) 50 and (d 90 ) 90% of the cumulative volume distribution, were
0.46, 0.88 and 1.54 μm as it is shown in the figure below (Figure 2.19).
100
80
60
40
Samples
d50 [µm]
A
AAM
8.08
0.88
20
0
0
4
8
12
16
20
24
Figure 2.19 - Cumulative distribution by volume of: A (solid line, no symbols) and A AM (squares)
The results obtained denoted that the adopted milling conditions have to be optimized. In fact, a
lower powders dispersability was yielded if compared to the previous routes. Additionally, some
α-alumina seeds contaminated the powder, as determined after analyzing the attrition milled
material by XRD. In fact, near transition aluminas, some peaks were ascribed to α-alumina phase
(ICDD file n°. 81-2266) as it is shown in Figure 2.20.
67
Figure 2.20 – XRD pattern of Nanotek after attrition milling.
As a conclusion, both magnetic stirring and ball-milling were effective methods for powders deagglomeration; moreover, by ball-milling, dispersed suspensions can be prepared in a very
limited time. On the contrary, attrition milling carried out in the previously reported conditions was
less effective and also produced a significant pollution by the milling media into the ground
material. As a consequence, in the following experimentation, this dispersion route was no more
considered.
2.2.3
Effect of the dispersion on the specific surface area
The samples A MS , A BMα and A BMz were submitted to N 2 adsorption/desorption isotherm and were
performed on the samples outgassed at 150°C as in the case of A.
35
A
AMS
30
3
-1
Volume (cm g )
25
20
15
10
5
0
0,0
0,2
0,4
0,6
0,8
1,0
p/p0
Figure 2.21- N 2 adsorption/desorption isotherms at 77 K out-gassed at 150°C, comparison
between A and A MS : full and white refer to adsorption and desorption.
68
The Figure 2.21 reports curves with similar shape for both samples, corresponding to
Type II isotherms. The measured BET SSA value for A MS were 37.3 m2 g-1, showing a limited
effect of magnetic stirring on the specific surface area (SSA of raw powder A was 34.5 m2g-1, as
previously reported). The A MS exhibited a small hysteresis loop and may although indicate a small
change of inter-particles porosity, probably as a consequence of the prolonged dispersion under
magnetic stirring, which should also be responsible for the increase in surface area. These results
showed the presence of inter-particle porosity may be excluded.
For the ball-milled samples, it was not observed a significant SSA increase as respect to A
(Figures 2.22), probably because of the shorter dispersion time in comparison with the A MS . The
comparison (Figures 2.22) between A and the ball milled samples of N 2 adsorption/desorption
isotherms at 77 K is shown.
35
A
ABM
25
3
-1
Volume (cm g )
30
20
15
10
5
0
0,0
0,2
0,4
0,6
0,8
1,0
0,6
0,8
1,0
p/p0
a)
35
A
ABMz
25
3
-1
Volume (cm g )
30
20
15
10
5
0
0,0
b)
0,2
0,4
p/p0
Figures 2.22- N 2 adsorption/desorption isotherms at 77 K out-gassed at 150°C, comparison
between A with (a) A BMα and (b) A BMz : full and white refer to adsorption and desorption.
69
2.2.4 Effect of dispersion on powder composition and evolution
DTA analysis was performed on the three dispersed samples and their thermal behaviour
compared with that of sample A.The dispersion process significantly affect the thermal behaviour
of the raw powders, at both low (i) and high (ii) temperature regime, as shown in Figure 2.23.
DTA signal [20 V]
i) At low temperature regime, the dispersed samples show a endothermic signal at about 280°C,
which was not present into the raw, A sample. If the TG curves are concerned, it can be observed
that the above endothermic signals are accompanied by an abrupt mass loss of about 0.5-1%.
The insert of Figure 2.23 shows, for instance, the TG curve of A MS in the 150°-500°C temperature
range, in which a mass loss of 0.7% is recorded.
279°C
1325°C
1203°C
282°C
1210°C
286°C
1180°C
A
AMS
ABMz
ABM
exo
200
400
600
800 1000 1200 1400
Temperature [°C]
Figure 2.23- DTA curves of the samples. In the insert, A MS TG curve.
Intensity (a.u.)
In order to better understand such difference in thermal behavior between A and the dispersed
materials, the XRD patterns of A and A MS were compared (Figure 2.24).

 



15
AMS
20
25
 



 
10

  
20


30







A
 

 


40

50



60
70
Diffraction Angle (2
Figure 2.24- XRD pattern of A and A MS . In the insert, XRD pattern of A MS in the 15-25° (2θ)
range.
As already reported, NanoTek® is a mixture of transition alumina phases, precisely -Al 2 O 3 and
- Al 2 O 3 . If A and A MS are compared, diffraction patterns of - and -phases are seen in both
spectra; the only differences among them are ascribable to peaks at 2 = 18.2° and 2 = 20.3° in
70
the A MS spectrum (see arrows in Figure 2.24). For a sake of clarity, inset to Figure 2.24 reports a
detail of the 15° – 25° 2range, in which their spectra are superimposed: it is shown that the two
extra-peaks actually appear after magnetic stirring.
The two peaks are very close, in both position and relative intensity, to the (002) and (200)
reflections of gibbsite, respectively (ICDD 76-1782). The two polymorphs of aluminium hydroxide
Al(OH) 3 , gibbsite and bayerite, can be however discriminated by the diffraction peak at about 18°,
occurring at 2 = 18.2° in gibbsite and 18.8° in bayerite. The formation of a crystalline gibbsite
Al(OH) 3 phase may thus be inferred. Its formation was imputed to the prolonged magnetic stirring
of A in water to produce A MS , as will be better explained in the following.
The formation of γ-Al(OH) 3 during water dispersion of transition aluminas was already reported in
literature13. This process is described by the following reaction:
1

 Al 3  3 H 2O
Al2O3     3H  

2
2
(1)
In this case, the hydrated Al3+ ion is only stable under acidic conditions, so its hydrolysis has to be
considered:

 Al  OH   3H 
Al 3  3H 2O 
(2)

3
Equations (1) and (2) can be summarized in the following one:
1
3

 Al  OH 
Al2O3     H 2O 

3
2
2
(3)
According to literature12,13, the pH of the alumina suspension plays a key role into the Al(OH) 3
formation. In fact, bayerite precipitation from -aluminas occurred only at pH higher than 4.5-5, as
occurs in our investigation.
This result is in good agreement with the publication of Pourbaix et al.14, which reports the
solubility diagram of alumina and five polymorphs of aluminium hydrates - empirically obtained by
knowing the free formation enthalpies - as a function of pH (Figure 2.25). The minimal solubility of
many aluminium hydrates and hydroxides was found at pH 5.1, while their solubility significantly
increased at both lower and higher pH.
71
Figure 2.25- Influence of pH on the solubility of Al 2 O 3 and the aluminium hydrates at 25°C14.
In fact, in samples dispersed for 170 h at acidic conditions (namely at pH 4 and 3) no gibbsite
peaks were detected, as it is shown in Figure 2.26.
Intensity (a.u.)






 

  




AMS pH=4



  
10
20



 


30
40

50


AMS pH=3
60
70
Figure 2.26- XRD patterns of A MS dispersed at pH 3 and 4 for 170 h
On the ground of the above considerations, the endothermic signals of the DTA curves collected
in Figure 2.23 can be reasonably associated to the gibbsite thermal decomposition.
According to literature12, the dehydration of Al(OH) 3 proceeds with the intermediate formation of
boehmite (AlOOH), according to the following reactions:
Al (OH )3  AlOOH  H 2O
AlOOH 
1
1
Al2O3  H 2O
2
2
(4)
In fact, X. Carrier et al.12 carried out TGA/DTA curves on a transition alumina sample suspended
for 168 h at pH 11, under continuous stirring. They found a rather sharp endothermic DTA peak at
250 ◦C concomitant with a weight loss; a broad hump centred at about 550°C was also seen by
the Authors on the DTA curve, accompanying a progressive weight loss. It was associated to the
dehydration of boehmite (AlOOH) formed from the dehydration of Al(OH) 3 .
They same Authors strengthened their founding with TEM analysis, which allowed the
observation of large particles (about 200 nm long) besides the small γ–alumina nanometric ones.
Electron diffraction patterns allowed the assignation of the gibbsite structure to the large grains,
according to their XRD data.
In our experimentation, any further DTA endothermic signals, ascribable to the boehmite
dehydration, was observed, probably due to the low fraction of gibbsite formed upon stirring. In
order to confirm such hypothesis, equation (4) was exploited to determine the gibbsite amount; it
was calculated from the weight of water removed at 280°C and normalized to the mass of the
sample after completing the dehydration. It was estimated that about 2 wt.% of gibbsite was
formed upon hydration of transition alumina.
72
Intensity (a.u)
It is interesting to note that magnetic stirring actually plays a role in the formation of the hydroxide
phase, since a blank sample prepared by suspending the powder in water for 170 h without
stirring did not show any change in the XRD patterns (see Figure 2.27)











 
10
20

30
40
50


60
70
Figure 2.27: XRD pattern of A powder suspended for 170 h in distilled water, without stirring
Furthermore, XRD measurements on powder stirred for increasing times showed that the two
peaks appear after 36 h and then their intensities remain almost constant up to 170 h. In the
following Figure 2.28, XRD patterns evolution of A MS as a function of the stirring time is reported.
Figure 2.28- XRD pattern evolution of A MS as a function of the stirring time
Finally, HRTEM was conducted on A and A MS samples (Figure 2.29 for A sample). Both powders
are composed by spherical particles of heterogeneous diameters, in the 5 – 100 nm range,
forming agglomerates with variable size, while “single particles” were rarely observed in both
materials. In both cases, it can be observed that particles are covered by a poorly ordered layer
of about 2.5 nm thickness (see insert in Figure 2.29), contrasting with the well crystallizer inner
part in which the crystalline lattice fringes could be easily observed. The presence of such
73
amorphous layer, most probably induced by the physical vapour synthesis15 of the nanostructured
powders, seems to cover the majority of particles and will be directly in contact with water during
dispersion under magnetic stirring and therefore affecting the surface properties of the final
material.
In addition, From HRTEM analysis it was not possible to detect any gibbsite Al(OH) 3 particles,
unlike other works12, reporting the formation of well-shaped and large gibbsite crystals separated
from the alumina surface after suspension in water.
Figure 2.29- Selected HRTEM picture of sample A; the insert reports a micrograph, taken at a
higher magnification, showing the presence of an outer amorphous layer.
ii) Figure 2.23 shows a different thermal behaviour also in the high temperature range. It is
evident, in fact, the role of the dispersion in lowering the crystallization temperature of the αalumina phase.
The sharp exothermal signals were, in fact, associated to the  to  transformation temperature
and they were detected at significantly lower temperatures for the dispersed materials as
compared to sample A. Precisely, such transformation signal was detected at 1325°, 1203°,
1180° and 1210°C for A, A MS , A BMα and A BMz , respectively.
In addition, if the transformation temperatures of the three dispersed powders are compared, a
further difference can be observed. The lowest crystallization temperature was determined for
A BMα , the highest to A BMz . As already observed in literature10, such difference can be imputed to
a seeding effect due to the milling media, even if undetectable by XRD analysis (see Figure 2.18).
In fact, alpha-alumina seeds are able to promote the above transformation11, while zirconia ones
have the opposite effect, and delay the α-phase crystallization16.
This aspect will be deepened in the following, since a systematic investigation of the phases
evolution, as a function of the calcinations temperature, in the as-received and dispersed samples
was performed.
2.2.5 Influence of the dispersion route on the phase evolution
Sample A, as already reported, is composed by δ-Al 2 O 3 and γ-Al 2 O 3 phases. After calcination at
1100°C for 0.5 h, some traces of θ-Al 2 O 3 (ICDD file n°11-0517) appear. Subsequently, raising the
calcination temperature up to 1150°C for 0.5 h, θ-Al 2 O 3 tends to increase its relative abundance.
Finally, by increasing the calcination temperature up to 1200°C for 0.5 h it is possible to denote
74
Intensity (a.u)
the first traces of α-Al 2 O 3 (ICDD file n°10-0173). In the Figure 2.30 it is shown the thermal
evolution of sample A as a function of the calcination temperature.


 
  


 

 


 























 

  


 


1200°C



1150°C



10


20
    
30


40
1100°C


 



50
60
non-treated
70
Figure 2.30– Thermal evolution of sample A
The dispersed samples were then submitted to the same calcination temperatures. In the case of
A MS (Figure 2.31), the powder exhibited a different thermal evolution if compared with A. By
calcining at 1100°C for 0.5 h, the powder is composed by a mixture of δ-Al 2 O 3 , γ-Al 2 O 3 and θAl 2 O 3 phases. At 1150°C, a significant amount of the alpha-phase was detected in the material
near θ-Al 2 O 3 ; finally, at 1200°C the powder is mainly composed by well-crystallized α-Al 2 O 3
phase.
Intensity (a.u)









1200°C



  
  
10
20

 
  



 
30





   



40
50



1150°C
  






60
1100°C
As-received
70
Figure 2.31 – Thermal evolution of A MS
The XRD pattern of sample A BMα (Figure 2.32) calcined at 1100°C presents the three phases:
δ-Al 2 O 3 , γ-Al 2 O 3 and θ-Al 2 O 3 . By rising the calcinations temperature up to 1150°C the
transformation into α-phase is completed.
75
Intensity (a.u)







  
  
10


1200°C









1150°C

 
 
  




 
1100°C



  


  


As-received
20
30
40
50
60
70
Figure 2.32 – Thermal evolution of A BMα .
Intensity (a.u.)
A BMz (Figure 2.33) calcined at 1100°C is prevalently composed by δ and θ phases. At 1150°C
the powder is a mixture of θ and α phases and finally at 1200°C only α-phase appears.












  
  
10
20
 



 




  
  
30




1200°C




 
  


1150°C

1100°C



40

50
 

non-treated
60
70
Figure 2.33 – Thermal evolution of A BMz .
For a better comparison between samples, Figure 2.34 collects the XRD patterns of the four
materials after calcining at 1150°C for 0.5 h. As it is shown in the figure, the dispersion route has
been effective in modifying the high-temperature crystallization path of the Nanotek powder. After
calcination at 1150°C for 0.5 h (Figure 2.34), A was composed only by transition phases (namely
δ and θ-phases); in A MS , α-Al 2 O 3 was the prevalent phase near θ-Al 2 O 3 ; a pure, well-crystallized
α-Al 2 O 3 was detected in A BMα and finally, A BMz was a mixture of transition and α-Al 2 O 3 phases.
These data state the role of dispersion in lowering the transformation temperature, according to
DTA curves (see Figure 2.23).
76
Intensity (a.u.)




 












  
20
 

 
30
40




10



 
ABMz







50
ABM



 
  
60
AMS
A
70
Diffraction Angle (2)
Figure 2.34 - XRD pattern of the samples calcined at 1150°C for 0.5 h.
In addition, if the XRD patterns of the three dispersed samples are compared, a further difference
can be ascribed. If we consider the un-polluted A MS sample as a reference, -alumina phase is
contained in a higher and lower fraction in A BM and A BMz , respectively. So, the anticipation of the
temperature transformation occurring in A BMα is an probably related to some seeding (not visible
in the XRD pattern) introduced during the milling process which promotes the α-alumina
crystallization. On the contrary, a delay in crystallization in A BMz compared with A MS and A BMα is
reasonably attributable to zirconia impurities introduced during milling, in agreement with
literature16.
2.2.6 Effect of the dispersion on the kinetics of transformation
The kinetics of  to -alumina phase transformation of A and of the dispersed samples was
studied by employing the Kissinger’s method. The Kissinger’s plots are collected in Figure 2.35,
while the related activation energy are reported in Table I.
The obtained data are discussed in the following, by separately considering the transformation
temperature (i) and the related activation energy (ii).
i)
The comparison among the Kissinger’s plots allows to evidence again the significant role
of the dispersion process in lowering the transformation temperature. In addition, if the unpolluted stirred sample, A MS , is kept as a reference, the lowering of the transformation
temperature in A BM was more relevant as compared to the delaying effect presented by
A BMz . Such behaviour was, once again, imputed to a different effect of alumina and
zirconia seeds form the milling media which affect in a different way the transformation
process.
77
A10ABMzAMS ABM
-ln(dT/dt / Tm2)
A A3
0.2
0,00060
0,00065
0,00070
1/Tm
Figure 2.35- Determination of the activation energy for transition to α-Al 2 O 3 phase transformation.
In insert are the calculated activation energies are reported.
Table I - Activation energy (E a ) for θ to α-Al 2 O 3 phase transformation (kJ/mol)
Sample
A
A MS
A BMα
A BMz
A α3
A α10
E a (kJ/mol)
486
498
480
496
502
438
R2
0.9959
0.9946
0.9969
0.9993
0.9892
0.9990
So, in order to discriminate the role of dispersion form that of seeding in affecting the
transformation temperature, two new materials were prepared. Samples A α3 and A α10 were
produced by flash plunging the sample A into in a tubular furnace kept at 1290°C for 3 and 10
minutes, respectively. The aim of this thermal treatment was to induce the crystallization of
certain amounts of α-Al 2 O 3 phase in the powders, but trying to minimize the crystallite growth
during the heat treatment. In Figure 2.36, XRD patterns of the above materials are presented: in
A α3 the alpha phase was almost undetectable (as in the case of A BM ) while an appreciable Al 2 O 3 XRD peak was indeed observed in A α10 .
Figure 2.36- XRD patterns of A α3 and A α10 in the 38-50° range.
78
The comparison among the Kissinger plots of sample A with those of A BM, A 3 and A 10 shows
that also seeding lowers the transformation temperature: the higher the -seeds amount, the
lower the transformation temperature. So, the more pronounced anticipation occurring in A BM
should be reasonably imputed to a synergic effect induced by coupling dispersion and seeding
with -phase. On the ground of this hypothesis, also the “delaying” effect of A BMz should be
imputed to the seeding effect induced by zirconia milling media, which acts in a negative way on
the transformation temperature, as expected on the ground of literature16.
ii) Concerning activation energy, also in this case the role of dispersion and seeding should
be separately discussed.
The activation energy of A was 486 kJ/mol, in a good agreement with literature data17, but it was
almost unaffected by the dispersion routes. This result is only in partial agreement with literature
data, in which a lower  to -Al 2 O 3 transformation temperature due to an effective dispersion
process was also reflected into a lower related activation energy17-18. On the contrary, seeding is
able to lower the above energy, as shown by A 10 , for which a decrease of about 10%, as
compared to A, was determined. However, a similar decrease was not presented by A 3 and
A BM , thus suggesting that detectable amounts of -seeds are required for affecting the activation
energy of the transformation.
2.2.7 Influence of the dispersion route on forming and sintering
As in the case of A, the dispersed samples were submitted to dilatometric analysis performed on
uniaxially pressed bars (at 350 MPa) by heating up to 1500°C (heating rate of 10°C/min) and a
soaking time of 3 h at maximum temperature.
Firstly, the effect of dispersion of the compactability, under dry pressing, of the powders was
investigated. In Table II, their green density, calculated from weight and geometrical
measurement of the pressed bars is reported. A slight increment of the green density from the
raw powder to the dispersed ones was observed. In addition, A BMz , presented the higher value: if
the density of zirconia is concerned (reference value for TSZ is 6.09 g/cm3), such datum can be
explained on the ground of a possible seeding from the milling media.
Table II- Green density of differently sintered samples
Sample Green Density (g/cm3) Green Density (%TD 1 )
A
1.85
46.9
A MS
1.89
47.8
A BMα
1.93
48.9
A BMz
2.04
51.7
A α10
1.91
48.2
As evidenced by the following Figures 2.37, all the dispersed materials present a two-step
sintering behaviour, as in the case of sample A. In spite of this, the samples differ into onset
sintering temperature and θ→α transformation temperature, (evidenced by the derivative curve:
see dotted lines in the following graphics).
In order to better evidence their different behaviour, Table III collects the onset sintering (T onset ,
°C) and the δα transformation (T δα , °C) temperatures. In addition, the total linear shrinkage
(ΔL/L 0 ) tot (%), as well as shrinkage percentages recovered during the heating step (ΔL/L 0 ) heating ,
1
Referred to the α-Al 2 O 3 theoretical density 3.96 g/cm3
79
shared into two contributions, one associated to Zone I (ΔL/L 0 zone I ), the other related to Zone II
(ΔL/L 0 zone II ), and finally the shrinkage due to the isothermal step (ΔL/L 0 ) isothermal , are collected.
In some cases, the derivative curve allows to detect a second significant sintering temperature,
i.e. the temperature of maximum sintering rate of the α-phase (T α max ). If detectable, this value is
again collected in Table IV.
0,0000
0,00
-0,02
-0,02
-0,0002
1420°C
-0,0002
-0,04
1420°C
-0,0004
-0,0008
-0,12
1135°C
-0,0010
-0,14
-0,10
-0,18
-0,0008
-0,12
-0,0010
-0,14
1121°C
-0,0012
-0,16
-0,0006
-0,08
o
o
L/L
-0,10
L/L
-0,0006
-0,08
-0,0004
-0,06
derivative signal
-0,06
derivative signal
-0,04
0,0000
0,00
-0,0012
-0,16
-0,18
-0,0014
-0,0014
-0,20
-0,20
200
400
600
a)
800
1000
1200
1400
200
1600
400
600
b)
Temperature [°C]
0,0000
0,00
-0,02
800
1000
1200
1400
1600
Temperature [°C]
0,0000
0,00
-0,02
-0,0002
-0,0002
-0,04
-0,0004
-0,0008
-0,12
-0,0010
1145°C
-0,14
-0,18
-0,10
-0,0008
1145°C
-0,12
-0,0010
-0,14
-0,0012
-0,16
-0,0012
-0,16
o
o
L/L
-0,10
-0,0006
-0,08
L/L
-0,0006
-0,08
-0,0004
-0,06
derivative signal
-0,06
-0,18
-0,0014
derivative signal
-0,04
-0,0014
-0,20
-0,20
200
400
600
800
1000
1200
1400
200
1600
400
600
800
1000
1200
1400
1600
Temperature [°C]
Temperature [°C]
d)
c)
Figures 2.37 – Dilatometric (solid line) and derivative (dashed line) curves: (a) A MS , (b) A BMα , (c)
A BMz and (d) A α10 .
Table III- Total linear shrinkage (ΔL/L 0 ) tot , shrinkage percentages recovered during the heating
step (ΔL/L 0 ) heating , associated to Region I (ΔL/L 0 Zone I ), and to Region II (ΔL/L 0 Zone II ), shrinkage
during isothermal step (ΔL/L 0 ) isothermal , onset temperature (T onset ), temperature of
δα
transformation (T δα , °C)
Sample
ΔL/L 0total
(%)
ΔL/L 0heating
(%)
A
A MS
A BMα
A BMz
A α10
14.6
17.7
19.4
19.2
17.1
10.9
13.9
17.4
14.4
12.4
ΔL/L 0
ΔL/L 0
ΔL/L 0
Zone I
Zone II
isothermal
(%)
6.5
7.1
7.3
6.5
6.0
(%)
3.1
5.9
9.2
7.0
5.8
(%)
3.0
3.8
2.0
4.8
4.7
T onset
(°C)
T δα 
(°C)
T α  max
(°C)
1021
996
996
1030
996
1147
1135
1121
1145
1145
Not detected
1420
1420
Not detected
Not detected
As it is shown in Table III, the dispersion slightly lowered the T δα , as seen in samples A MS and
A BMα . Once again, the lowest transformation temperature detected in A BM was imputed to
synergic effects of dispersion and seeding. The effect of the dispersion on T δα in sample A BMz is
not observable as seeding introduced by the zirconia medium countered this phenomenon16.
80
Samples present almost the same onset temperature: however, slightly higher values were
presented by samples A and A BMz .
Similar shrinkages were found in samples during the first sintering step, specially in the case of
A BMα and A MS (≈7%). During the second step, all the samples shrank considerably more than
sample A which exhibited a total linear shrinkage of about 3 %.
However, among them A BMα achieved a larger linear shrinkage during the second step. As a
consequence, its densification during the isothermal step was limited - if compared with the other
samples (≈2 %). This fact is in good agreement with the position of the second peak on the
derivative curve, which represents the maximum densification rate. This peak was only recorded
at 1420°C in samples A BMα and A MS .
In sample A BMz it was observed a delay in densification during the beginning of the second step,
as reported in literature19. Nevertheless, it recovered before isothermal step. During the dwell
period, samples A BMz and A α10 presented similar linear shrinkage of about 4.7 %. Conversely,
sample A MS in the last step recorded 3.8%.
As a conclusion, the sintering behaviour of A BMα shows the combination of the single effects of
A MS and A α10 , i.e. dispersion and seeding.
Sample A presented the lowest total shrinkage value: this datum coupled with its lowest green
density, allowed to explain the lowest fired density (75.3%) of such sample (values are collected
in Table IV). A significant increment (of 10-15%) of the final density was achieved by the
dispersed materials. Moreover, a similar increment was reached by the un-dispersed, seeded
A α10 sample. This datum was of interest, since it allows to explain the highest fired density of A BMα
as compared to all the other samples, since it combines the positive effects of both dispersion
and seeding.
Table IV- Fired density of differently sintered samples
Sample
Fired Density (g/cm3)
Fired Density (%TD1)
A
2.98
75.3
A MS
3.25
82.1
A BMα
3.42
86.4
A BMz
3.39
85.6
A 10
3.36
85.8
On the ground of such results, it can be reasonably assumed that a further increment in the final
density can be achieved by dispersing A α10 sample. Some attempts were performed but even
after several hours of ball-milling (3 h), the powders were still agglomerated. As a future activity,
optimized processed powders should be produced by coupling the flash heating step to an
effective dispersion process, thus to induce both effects.
2.2.8 Investigation on Sintering Kinetics of Nanotek powders
Stepwise isothermal dilatometry (SID) was applied for investigating the sintering kinetic of
nanopowders. Shrinkage curves recorded for each isothermal step were fitted polynomially and
the shrinkage rate was calculated from shrinkage value. The method is accurately described in
chapter one.
81
Three samples were chosen for this study, namely samples A, A MS and A BMα . The selection of
these samples was based on the transformation kinetic data, putting emphasis on A BMα . For this
aim, dilatometric tests were performed employing bars produced by cold uniaxial pressing at
350 MPa. In this study, the 1100°C-1400°C was chosen as temperature range of study. This first
material submitted to the study was A sample. The apparent activation energy (data collected in
Table V) for A sample was calculated fitting the values with a linear slope obtaining a value of 690
kJ/mol as it is illustrated in the following Figure 2.38.
Q=689.9 kJ/mol
2
R =0.9943
-2
-4
-6
-ln(nK)
-8
-10
-12
-14
-16
-18
-4
5,7x10
-4
6,0x10
-4
6,3x10
-4
6,6x10
-4
6,9x10
-4
7,2x10
-4
7,5x10
-1
1/T (K )
Figure 2.38 – Arrhenius constant (K(T)) versus reciprocal temperature 1/T of A sample.
Table V - Activation energy of the different samples.
SAMPLE
E a (kJ/mol)
R2
n
A
A MS
A BMα
690
606
450
0.9943
0.9342
0.9899
0.15
0.10
0.18
The second sample analyzed was the A MS . By applying the same procedure, the apparent
activation energy was 606 kJ/mol, which is in good agreement with the data obtained regarding
the kinetics transformation data20-21.
Similarly, the ball-milled sample named A BMα was submitted to the same analysis. In this
particular case, as it was expected, A BMα exhibited a lower activation energy (514.5 kJ/mol)
referable to the some seeding introduced by the milling media.
As reported in literature21, a sharp change of parameter n (from 0.24-0.07 in case of A) in the
temperature range (1100-1200°C) was observed in all samples correlated with phase
transformations.
In literature it was not found relevant information about this method applied on transition
aluminas. For this reason many studies attempted to understand the activation energies of single
phase constituents.
For instance, Wang et al.20 applied this technique to evaluate the kinetics parameters of a
macroporous α-Al 2 O 3 . In this particular case, values were calculated in the range 1200-1400°C
obtaining an apparent activation energy of 415 kJ/mol and an average exponent n equal to 0.232.
82
A second study made by Holkova et al.21 attempted to study and establish a comparison between
the stepwise isothermal dilatometry. For boehmite and α-Al 2 O 3 evaluated by SID, the authors
found apparent activation energies of 950.1 kJ/mol and 541.1 kJ/mol, respectively.
A similar behaviour was found by the same authors21, by adding boehmite into α-Al 2 O 3 which
increased the apparent activation energy to 597.7 kJ/mol. In this context, the effect of introducing
some α-Al 2 O 3 milling contamination into the powder may be the principal responsible of reducing
the apparent activation energy in sample A BMα .
2.2.9 Influence of the dispersion route on the final microstructure
SEM observations were performed in SE mode on the fracture surface of the fired bodies
obtained by uniaxial pressing. The corresponding micrographs are reported in Figures 2.39.
The SEM analysis revealed a different microstructural development. Sample A (Figure 2.39 a),
presented a vermicular microstructure accompanied by inhomogeneous morphologies. It is made
of fine grained alumina particles of about ≈0.9 μm entrapping diffuse residual porosity.
The role of the dispersion route on the microstructural evolution was evidenced by comparing A MS
and the ball-milled samples. Sample A MS (Figure 2.39 b) is less porous compared with A, in spite
of its better densification, the ultrafine primary particle size was retained, yielding round-shaped
grains of ≈0.77 μm, entrapping mostly intragranular porosity.
The microstructure of A BMα (Figure 2.39 c) was highly dense, consisting of well-facetted alumina
grains of about 2 μm with a limited residual porosity mostly located, as in the previous case, in
intragranular position. A lower porosity was observed in A BMz (Figure 2.39 d), characterized, as
reported in literature, by a finer microstructure19. The average size of alumina grain was 1.5 μm.
The fracture mode was radically different: in the case of A it is mostly transgranular, while in A MS
it is prevalently intergranular; finally A BMα and A BMz presented both inter- and transgranular
modes.
a)
b)
83
c)
d)
Figures 2.39 – SEM micrographs of the sintered materials: (a) A, (b) A MS , (c) A BMα and (d) A BMz .
(SEM observation performed on the fracture surface)
2.3 Study of the effect of powder dispersion on its surface properties by means of IR
spectroscopy
2.3.1 IR spectra of samples outgassed at increasing temperatures
The species at the surface of A and A MS powders were studied by means of FT-IR spectroscopy.
Samples were outgassed at increasing temperatures, namely 150, 350 and 500°C, since the
presence of surface gibbsite should give rise not only to peculiar IR bands in the hydroxyls range
(3900-3000 cm-1), but also to different adsorbed species and different behaviour towards thermal
treatments.
In Figures 2.40, FTIR spectra of A outgassed at 150, 350 and 500°C are reported the spectrum of
the sample outgassed at 150°C shows abroad absorption band in the stretch range 3900-3000
cm-1 range.
Absorbance
4
a
3
150°C
2
350°C
1
500°C
0
3600
3200
2800
2400
2000
1600
1200
Wavenumbers (cm-1)
84
3675
b
3725
Absorbance
3600
3775
3790
350°C
3680
500°C
3800
3600
3400
3200
3000
-1
Wavenumbers (cm )
Figures 2.40- FT-IR spectra recorded on sample A: (a) outgassed at 150°C, 350°C and 500°C
and (b) detail of the hydroxyls spectra outgassed at 350 and 500°C.
As it was reviewed in the first chapter, this is typical for a highly hydrated surface caused by the
atmospheric moisture. Bands in the 1650-1200 cm-1 range are related to several carbonate-like
species, usually observed in transition aluminas23,24 and definitely removed by outgassing at
500°C.
In Figure 2.40 b, a detail of hydroxyls spectra of sample A outgassed at 350 and 500°C is
reported, showing the presence of different OH groups.
According to the model proposed by Knözinger and Ratnasamy, Bands at 3790 and 3775 cm-1
are assigned to I b and I a hydroxyls, i.e. free terminal hydroxyls bonded to octahedral (AlVI) and
tetrahedral (AlIV) aluminium ions. The 3725 cm-1 band is assigned to di-bridged free OH group
(type II a ) and 3675 cm-1 band to hydroxyls (type III a ), which are tri-bridged among two octahedral
(AlVI) and one tetrahedral (AlIV) aluminium ions; band at ≈3600 cm-1, with a tail on the lower
wavenumbers side, is assigned H-bonded hydroxyls, which should be eliminated after outgassing
at 500°C.
Figures 2.41 reports IR spectra of A MS outgassed at 150, 350 and 500°C. Carbonate-like species
bands at 1650-1200 cm-1 are stable to thermal treatment at 500°C, indicating that after magnetic
stirring the surface present stronger basic sites to which CO 2 may coordinate.
85
Absorbance
4
3
2
150°C
350°C
1
500°C
0
3600
3200
2800
2400
2000
1600
1200
-1
Wavenumbers (cm )
Figure 2.41- FT-IR spectra recorded on sample A MS outgassed at 150°C, 350°C and 500°C.
The difference of the magnetic stirring on the surface properties is shown Figures 2.42, in which
hydroxyls spectra of two samples outgassed at 150°C (a), 350°C (b) and 500°C (c) are
compared.
A
AMS
a
Absorbance
3.0
3725
2.5
2.0
3780
1.5
3900
3600
3300
3000
-1
Wavenumbers (cm )
86
2.5
3675
b
3725
A
AMS
3580
Absorbance
3775
2.0
3790
1.5
3900
3600
3300
3000
-1
Wavenumbers (cm )
2.5
c
A
AMS
3725
Absorbance
3775
3680
2.0 3790
1.5
3800
3600
3400
3200
3000
-1
Wavenumbers (cm )
Figures 2.42- Normalized FT-IR spectra recorded on samples A and A MS outgassed at (a) 150°C,
(b) 350°C and (c) 500°C.
87
Spectra corresponding to the samples outgassed at 150°C (Figure 2.42 a) are dominated by the
broad absorption of H-bonded hydroxyls (below 3600 cm-1), whereas at higher wavenumbers,
bands are seen at 3780 and 3725 cm-1 due to free AlVI-OH and AlIV-OH (I a ). In sample A MS an
additional band appears at 3330 cm-1 which cannot be imputed to residual water molecules are
removed at room temperature. This band is reasonably attributed to OH stretch mode of gibbsite
hydroxyls. Gibbsite structural OH groups have been carefully studied by means single-crystal
Raman and FTIR methods24, which allow to single our several distinct types of structural OH
groups, basically inter-layer and intra-layer hydrogen bonded hydroxyls. With single crystals,
authors were able to single out six νOH in both IR and Raman spectra. In this case, only the
presence of a new broad absorption was observed at 3330 cm-1. The observed frequency was in
this case 3376 cm-1, due to OH-species interacting by H bonding.
The stability of OH species was studied by increasing the outgassing temperature assuming the
decomposition of Gibbsite below 300°C: after outgassing at 350°C, hydroxyls bands decrease in
intensity, due to surface dehydroxylation and the main difference between the two samples is the
higher intensity of the band of H-bonded hydroxyls with sample A MS at 3590 cm-1 indicating that
more hydroxylated surface formed during magnetic stirring.
Figure 2.42 (c) reports hydroxyls spectra of samples outgassed 500°C, the typical surface
features of de-hydroxylated transition alumina are expected, with bands at 3790 cm-1 (type I b
hydroxyls), 3775 cm-1 (type I a hydroxyls), 3725 cm-1 (type II a hydroxyls) and 3680 cm-1 (type III a
hydroxyls), whereas 3580 cm-1 band (H-bonded hydroxyls) is removed.
It was confirmed that surface physico-chemical properties were modified by the dispersion route,
the main difference being the OH species population due to the presence of Gibbsite on sample
A MS .
2.3.2 Adsorption of CO at nominal 77 K
Carbon monoxide is widely used as probe molecule to study both Lewis and Brønsted acidic
sites at the surface of oxides and zeolites24,27-31. When electrostatic interaction takes place
between CO and the adsorbing site, like in this case, a hypsochromic shift occurs, with respect to
free CO molecule (2143 cm-1), and characteristic bands are seen in the C≡O stretch region
(2250-2050 cm-1)27. Being the interaction very weak, low temperatures are needed and
experiments are performed at the nominal temperature of liquid nitrogen.
Increasing pressures of CO (in the 0.05–30 mbar range) were dosed, at the nominal temperature
of N 2(l) , on the two samples outgassed at 150, 350 and 500°C: Figures 2.43 and 2.44 report
normalized difference spectra, obtained by subtraction of bare samples spectra reported in
Figures 2.42.
CO dosage on sample A outgassed at 150°C (Figure 2.43 a) gives rise to the formation of i) a
main band at 2152 cm-1; ii) a weaker band in the 2189 – 2178 cm-1 range and iii) a minor
absorption at about 2107 cm-1.
The 2152 cm-1 band is due to CO molecules interacting via H-bonding with AlIV-OH species
originally absorbing at 3725 cm-1 (in Figure 2.42 b), whereas 3780 cm-1 hydroxyls are very weak
acid and do not interact with CO. The weak absorption at 2107 cm-1 is probably related to that at
2152 cm-1 and is assigned to CO molecules adsorbed through the O atom (CO---HO adducts),
according to previous work32.
The band at 2189 cm-1, shifting to 2178 cm-1 with coverage, is assigned to CO molecules
interacting with weak Lewis acidic sites, like 5-coordinate Al3+ or, most probably, coordinatively
unsaturated tetrahedral Al3+ of low index crystal planes33-34.
88
Figure 2.43 b reports difference spectra recorded after CO dosage on sample A outgassed at
350°C: two bands are seen at 2197 cm-1, shifting with coverage to 2183 cm-1, and at 2160 cm-1,
shifting to 2152 cm-1. The former is assigned to CO adsorbed on Al3+ sites forming an extended
phase, the latter to CO molecules H-bonded to hydroxyls with different acidity. With respect to
sample A pre-treated at 150°C, the band of CO on Al3+ sites appears more intense and shifted to
higher wavenumbers: this is ascribed to surface de-hydroxylation with formation of new stronger
Lewis sites (coordinatively unsaturated Al3+ ions). The band of CO interacting with OH species is
seen to shift with coverage from 2160 to 2152 cm-1 due to the presence of several hydroxyls with
different acidity, as shown in Figure 2.42 b.
With sample A outgassed at 500°C (Figure 2.43 c), bands are seen of CO adsorbed on an Al3+
sites forming an extended phase (band at 2198 cm-1 shifting with coverage to 2183 cm-1) and on
residual hydroxyls (band at 2160 cm-1 shifting with coverage to 2156 cm-1).
The relative intensities of bands due to CO adducts with Al3+ ions and OH species changed with
the hydration degree of the surface, since new coordinatively unsaturated sites become available
at the expenses of hydroxyls removed at higher temperature.
a
OH
IOH /IAl3+ = 11.5
Absorbance
1,5
1,0
0,5
Al
3+
0,0
2250
2200
2150
2100
2050
-1
Wavenumbers (cm )
89
b
I /I
= 1.6
OH Al3+
Absorbance
1,5
1,0
0,5
0,0
2250
2200
2150
2100
2050
-1
Wavenumbers (cm )
c
IOH/IAl3+ = 1.1
Absorbance
1,5
1,0
0,5
0,0
2250
2200
2150
2100
2050
-1
Wavenumbers (cm )
Figures 2.43- FT-IR difference spectra, in the CO stretch region 2250 – 2050 cm-1 recorded after
dosing CO on sample A outgassed at 150°C (a); 350°C (b) and 500°C (c). CO equilibrium
pressures range: 0.5 - 20 mbar;
90
Figures 2.44 report corresponding spectra recorded in same conditions, i.e. under the same
equilibrium CO pressures, on sample A MS pre-treated in the same way. As a whole, the same
surface species were observed, i.e. coordinatively unsaturated Al3+ ions and surface hydroxyls,
but with the following relevant differences:

with A MS outgassed at 150°C (Figure 2.44 a), at low coverage the band of CO H-bonded
to hydroxyls is seen at 2154 cm-1, whereas at higher CO equilibrium pressures another
component is seen at 2149 cm-1. The difference with respect to the corresponding band
on sample A outgassed at the same temperature is better shown in Figure 2.44 b,
reporting spectra recorded on samples A and A MS under the same CO pressure
(10 mbar). The smaller shift with respect to the free molecule mode (2143 cm-1) indicates
that the component at 2149 cm-1 should be related to CO interacting with weaker acidic
hydroxyls, like those originally absorbing at 3330 cm-1 (Figure 2.42 a), stemming from the
hydroxide phase (gibbsite);

normalized spectra recorded under the same CO equilibrium pressures allow to draw
some semi-quantitative observation: with A MS , intensities of CO bands are always smaller
than with A, indicating a smaller amount of sites actually accessible at the surface. The
ratio I OH /I Al3+, reported for each experiment, between intensities of the bands due to
CO---HO and CO---Al3+ adducts, respectively, may be used to evaluate the relative
abundance of Lewis and Brønsted sites. After treatment at 150°C, I OH /I Al3+ is 11.5 and
10.1 for A and A MS , respectively: this can be explained by the fact that 3330 cm-1
hydroxyls belonging to gibbsite, though abundant, are less acidic and less prone to
interact with CO; after treatment at 350°C, I OH /I Al3+ is 1.6 and 2.0 for A and A MS ,
respectively, since A MS surface is more hydrated, due to prolonged stirring in water. After
outgassing at 500°C, I OH /I Al3+ is the same for both materials, due to the formation of the
same.
IOH/IAl3+ = 10.1
a
Absorbance
1,5
1,0
0,5
0,0
2250
2200
2150
2100
2050
-1
Wavenumbers (cm )
91
1,5
A
A MS
b
2149
Absorbance
1,0
0,5
0,0
2200
2150
2100
-1
W avenumbers (cm )
c
I /I
= 2.0
OH Al3+
Absorbance
1,5
1,0
0,5
0,0
2250
2200
2150
2100
2050
-1
Wavenumbers (cm )
92
d
IOH/IAl3+ = 1.0
Absorbance
1,5
1,0
0,5
0,0
2250
2200
2150
2100
2050
-1
Wavenumbers (cm )
Figures 2.44- FT-IR difference spectra, in the CO stretch region 2250 – 2050 cm-1, recorded after
dosing CO (equilibrium pressures in the 0.5 - 20 mbar range) on sample A MS out-gassed at
150°C (a); 350°C (c) and 500°C (d). CO equilibrium pressures range: 0.5 - 20 mbar; difference
spectra obtained by subtracting spectra of bare sample reported in Figure 2.42. Section b:
comparison of difference spectra recorded at the same CO equilibrium pressure (10 mbar) on
samples A and A MS outgassed at 150°C.
2.3.3 Adsorption of CO 2
CO 2 interaction with the surface uncoordinated cations (Lewis acid sites) occurs by σ-change
release from one of O lone pair orbitals, so the molecule loses the center of symmetry preserving
its linear shape. By increasing the outgassing temperature, the band due to CO 2 adsorbed on Al3+
sites increases in intensity, whereas that due to CO 2 interacting with OH groups decreases. This
result is due to progressive surface dehydroxylation and is in agreement with CO adsorption
spectra.
CO 2 was dosed by increasing the pressure (in the 0.05-40 mbar range), on both samples
outgassed at 150, 350 and 500°C. Figure 2.45 reports the comparison between A and A MS
outgassed at 150°C after dosing ≈20 mbar CO 2 . Only one band is seen at 2344 cm-1, readily
assigned to CO 2 molecules interacting with surface OH groups.
93
2344
Absorbance
1,5
A
AMS
1,0
0,5
0,0
2400
2380
2360
2340
2320
2300
2280
-1
Wavenumber [cm ]
Figure 2.45– Comparison of difference spectra recorded at the same CO 2 equilibrium pressure
(20 mbar) on samples A and A MS outgassed at 150°C.
In Figures 2.46 reports difference spectra recorded after CO 2 dosage on sample A degassed at
350°C. Two bands are seen at 2360 cm-1 and 2345 cm-1. The former is assigned to CO 2
molecules interacting with weak Lewis acidic sites like coordinatively unsaturated tetrahedral Al3+
of low index planes, the latter is assigned to CO 2 molecules interacting via H-bonding with
OH species 35-39.
1,2
ab
OH
Absorbance
1,0
0,8
0,6
3+
Al
0,4
0,2
0,0
2400
2380
2360
2340
2320
2300
2280
-1
Wavenumber (cm )
94
0,6
bb
OH
Absorbance
0,5
3+
Al
0,4
0,3
0,2
0,1
0,0
2400
2380
2360
2340
2320
2300
2280
-1
Wavenumber (cm )
Figures 2.46- FT-IR difference spectra, in the stretch region 2400-2280 cm-1 recorded after dosing
CO 2 in sample at A outgassed at: (a) 350°C and (b) 500°C.
Figures 2.47 reports spectra recorded after dosing the same CO 2 equilibrium pressures on A MS .
2,0
1,8
ab
OH
Absorbance
1,6
1,4
1,2
1,0
0,8
3+
Al
0,6
0,4
0,2
0,0
2400
2380
2360
2340
2320
2300
2280
-1
Wavenumber (cm )
95
0,25
bb
Al
Absorbance
0,20
3+
OH
0,15
0,10
0,05
0,00
2400
2380
2360
2340
2320
2300
2280
Wavenumber (cm-1)
CO 2 in sample at A MS outgassed at: (a) 350°C and (b) 500°C.
Figures 2.48 reports normalised difference spectra in the 1900-1400 cm-1 recorded after dosing
20 mbar CO 2 on samples outgassed at 350 and 500°C. The same carbonate species exist in A
and A MS . No significant differences were found on strength of the basic sites evaluated by
[Δν=III ν as (COO-)- III ν s (COO-)], according to Peri36. This calculation confirms the similarities of
dehydrated surfaces in terms of strength of basic sites.
0,35
ab
A
AMS
0,30
Absorbance
0,25
0,20
0,15
0,10
0,05
0,00
-0,05
-0,10
1900
1800
1700
1600
1500
1400
-1
Wavenumber (cm )
96
0,5
bb
A
AMS
Absorbance
0,4
0,3
0,2
0,1
0,0
-0,1
1900
1800
1700
1600
1500
1400
-1
Wavenumber (cm )
CO 2 in sample at A MS outgassed at: (a) 350°C and (b) 500°C.
The slight more intense carbonate bands observed in sample A MS outgassed at 500°C indicate a
slightly higher abundance of basic sites.
2.3.4 Adsorption of NH 3
As reported in Chapter I, NH 3 is very often used as a probe to determine the number and nature
of acidic surface sites.
NH 3 was dosed by increasing the pressure (in the 0.05-40 mbar range) at room temperature. In
the following Figures 2.49/2.50 are shown normalized FT-IR spectra recorded on samples A and
A MS recorded after dosing NH 3 outgassed at 150, 350 and 500°C.
3,5
ab
NH

as
Absorbance
3,0
a
+
4
NH
3
2,5
2,0
1,5
4000
3500
3000
2500
2000
1500
-1
Wavenumber (cm )
97
bb

as
a
NH
3
Absorbance
2,0
1,5
1,0
4000
3500
3000
2500
2000
1500
-1
Wavenumber (cm )
2,0
cb

Absorbance
as
a
NH
3
1,5
1,0
4000
3500
3000
2500
2000
1500
-1
Wavenumber (cm )
Figures 2.49- Normalized FT-IR spectra recorded on samples A recorded after dosing NH 3
outgassed at (a) 150°C, (b) 350°C and (c) 500°C (dashed line indicates the final outgassing at
room temperature)
Figures 2.49 reports the IR spectra after NH 3 dosage on sample A outgassed at 150, 350 and
500°C. As it is shown NH 3 molecules interacts via H-bonding (indicate with an arrow) and leads
the formation of δ as NH 3 and the δNH 4 + ion as a result of the interaction with the Lewis and
Brønsted sites.
As reported in Figures 2.50, a similar behaviour was found in A MS sample leading the formation of
the same species.
98
ab
4,0
NH
+
4
Absorbance
3,5
a

3,0
as
NH
3
2,5
2,0
1,5
4000
3500
3000
2500
2000
1500
-1
Wavenumber (cm )
3,15
bb
3,10

Absorbance
as
a
NH
3
3,05
3,00
2,95
4000
3500
3000
2500
2000
1500
-1
Wavenumber (cm )
Absorbance
2,4
cb

as
2,2
a
NH
3
2,0
1,8
1,6
4000
3500
3000
2500
2000
1500
-1
Wavenumber (cm )
Figures 2.50- Normalized FT-IR spectra recorded on samples A MS recorded after dosing NH 3
outgassed at (a) 150°C, (b) 350°C and (c) 500°C (dashed line indicates the final outgassing at
room temperature)
99
Table VI reports the values corresponding to the position of Lewis and Brønsted bands with the
relative position.
Table VI- Position of δ as NH 3 and δNH 4 + in samples A and A MS .
Sample
A - Degassed 150°C
A - Degassed 350°C
A - Degassed 500°C
A MS - Degassed 150°C
Lewis δ as NH 3 (cm-1)
1623
1626
1626
1628
1626
1626
Brønsted δNH 4 + (cm-1)
1460
Not detected
Not detected
1404
Not detected
Not detected
As a comparison, δNH 4 + is slightly shifted in the case of A MS but δ as NH 3 mostly preserves its
position in both samples.
100
Conclusion
Modification of a nano-crystalline commercial transition alumina was produced by different
dispersion routes, as magnetic stirring or ball-milling with the aim of improving the sinterability
and optimizing the fired microstructures. The study of the effect has been carried out by
complementary techniques in the field of the materials science and surface physico-chemistry.
The work gave rise that dispersion is effective in reducing the agglomerate size and modifying the
crystallization path at low and high temperature regimes. Moreover, the dispersion induced some
physico-chemical surface modification.
Stirred sample has a surface more hydrophobic as well as more basic, compared with asreceived sample. Spectroscopic techniques permitted to evidence that stronger basic sites exists
on the surface of the stirred sample. In fact, gibbsite phase was determined near transition
aluminas in the stirred sample, which decomposes at about 280°C. Such hydroxide did not form
on a blank sample suspended in water without stirring.
The second conclusion is that dispersion do not modify the activation energy. However, this value
was only decreased in the sample in which certain amount of α-phase was introduced by flash
heating cycle. Furthermore, dispersed samples showed a better sinterability as higher densities
were achieved, as well as, a different microstructural development. This fact permits to avoid the
formation of the vermicular microstructure usually found in non-dispersed transition aluminas.
In future, optimized powders should be produced by coupling flash heating followed by an
effective dispersion process with the purpose of improving the powder sinterability.
101
CHAPTER
3
Nanostructured composite
materials: elaboration and properties
102
Nanostructured composite materials: elaboration and properties
3.1. Nanocomposites materials: main classification
Ceramic composites classification has been proposed for the first time by Niihara1. They can be
divided into two types: microcomposites and nanocomposites. In the microcomposites, micro-size
second phases such as particulates, platelets, whiskers and fibers are dispersed at the grain
boundaries of the matrix. Nanocomposites can be defined as multiphase solid materials where one
of the phases has one, two or three dimensions less than 100 nanometers. Niihara proposed four
types, as it is shown in Figure 3.1
Figure 3.1- Classification of ceramics nanocomposites 1,2
As it is drawn in the figures above, the nano-sized particles can be dispersed mainly within the
matrix grains (intra-type), at the grain boundaries of the matrix (inter-type) or can occupy both
positions (intra/inter-type). The last classification is composed of dispersoid and matrix grains with
nanometer-size.
The most remarkable advantages of the intra- and intergranular nanocomposites are the improved
fracture strength and toughness and reliability at room temperature; in addition, increased hightemperature mechanical properties such as hardness, strength, and creep resistance can be
found. On the other hand, the nano/nano composites have the purpose of adding new functions
such as machinability and the superplasticity like metals to ceramics.
Many different nanocomposites were reported in the literature. The most studied ones were Al 2 O 3
and Si 3 N 4 matrix reinforced by SiC particles. Some properties of alumina-based nanocomposites
were reviewed by Kuntz 3, as presented in the following figure.
The interest of using these materials occurred after the publication of Niihara1 which showed an
important increase of the fracture strength from 350 MPa in Al 2 O 3 up to 1000 MPa in Al 2 O 3 - 5
vol.% SiC.
103
Table 3.I : Different alumina-based composites reviewed by Kuntz3.
3.2. Synthesis of composites powders
In principle, any method capable of producing very fine grain size materials can be used to process
nanocomposites. Essentially, all of the successful processes have attributes that enable the
crystalline phases to nucleate but suppress the growth of nuclei.
Table 3.II summarizes the main used processing routes involved in the synthesis of ceramic-based
nanocomposites. In addition, the main physical, chemical and mechanical/mechanicochemical
routes are reviewed in the following items.
3.2.1. Physical Methods
3.2.1.1.
Vapour condensation methods
The inert gas condensation technique (Figure 3.2), conceived by Gleiter4, consists of evaporating a
metal (by resistive heating, radio-frequency, heating, sputtering, electron beam heating,
laser/plasma heating, or ion sputtering) inside a chamber that is evacuated to a very high vacuum
of about 10-7 Torr and then backfilled with a low-pressure inert gas, like helium. They generally
involve two steps: in the former, a metallic nanophase powder is condensed under an inert
104
convection gas. As a consequence, a supersaturated metal vapour is obtained in the chamber. In
the latter, the powder is oxidized by allowing oxygen into the chamber (to produce metal oxide
powder).
Table 3.II- Main synthesis Methods of Ceramic based Nanocomposites5.
A subsequent annealing process at high temperatures is often required to complete the oxidation.
The system consists of a vapour source inside a vacuum chamber containing a mixture of an inert
gas, usually argon or helium, mixed with another gas, which is selected on the ground depending
on the material to prepare. Oxygen is mixed with the inert gas to produce metal oxides.
NH 3 is usually used to prepare metal nitrides and an appropriate alkane or alkene, as a source of
carbon and it is usually used to prepare metal carbides.
Nanoparticles are formed when supersaturation is achieved above the vapor source. A collection
surface, usually cooled by liquid nitrogen, is placed above the source. The particles are transported
to the surface by a convection current or by a combination of a forced gas flow and a convection
current, which is set up by the difference in the temperature between the source and the cold
surface. Some improved systems involve a way to scrap the nanoparticles from the cold collection
surface so that the particles would fall into a die and a unit where they can be consolidated into
pellets. Supersaturated vapor can be achieved by many different vaporization methods. The most
common techniques include thermal evaporation, sputtering, and laser methods. A variety of
nanoscale metal oxides and metal carbides have been prepared using laser-vaporization
techniques. The peak densities of the as-compacted metal samples have been measured with
105
values of about 98.5% of bulk density. However, it has been established that porosity has a
important effect on the mechanical strength.
Figure 3.2- Schematic drawing of the inert gas condensation technique for production of nanoscale
powder 6,7.
The advantages of vapor condensation methods include versatility, good performance and highpurity products. On the other hand, they can be employed to produce films and coatings.
Furthermore, laser-vaporization techniques allow the production of high-density, directional, and
high-speed vapor of any metal within an extremely short time. Despite the success of these
methods, they have the disadvantage that the production cost is still high because of low yields.
Heating techniques have other disadvantages that include the possibility of reactions between the
metal vapors and the heating source materials.
3.2.1.2.
Spray pyrolysis
This technique is known by several other names including solution aerosol thermolysis (Figure
3.3), evaporative decomposition of solutions, plasma vaporization of solutions, and aerosol
decomposition. The starting materials in this process are chemical precursors, usually appropriate
salts, in solution, sol, or suspension. The process involves the generation of aerosol droplets by
nebulizing or ‘‘atomization’’ of the starting solution, sol, or suspension. The generated droplets
undergo evaporation and solute condensation within the droplet and the drying stage.
Subsequently, it is followed by a thermolysis stage of the precipitate at higher temperature in order
to form a microporous particle.
Different techniques for atomization are employed including pressure, two-fluid, electrostatic, and
ultrasonic atomizers. These atomizers differ in droplet size (2–15 mm), rate of atomization, and
droplet velocity (1–20 m/sec). These factors affect the heating rate and residence time of the
droplet during spray pyrolysis which, in turn, affect some of the particle characteristics including
particle size.
106
For a specific atomizer, particle characteristics, including particle size distribution, homogeneity,
and phase composition depend on the type of precursor, solution concentration, pH, viscosity, and
the surface tension.
Figure 3.3- Schematic drawing of thermal spray process, showing the different variables involved8.
Aqueous solutions are usually used because of their low cost, safety, and the availability of a wide
range of water-soluble salts. Metal chloride and nitrate salts are commonly used as precursors
because of their high solubility. Precursors that have low solubility or those that may induce
impurities, such as acetates that lead to carbon in the products, are not preferred.
The advantages of this method include the production of high-purity nanosized particles,
homogeneity of the particles as a result of the homogeneity of the original solution, and the fact
that each droplet/particle goes through the same reaction conditions. The disadvantages of spray
pyrolysis include the need for large amounts of solvents and the difficulty to scale-up the
production. The use of large amounts of nonaqueous solvents increases the production expenses
because of the high cost of pure solvents and the need for proper disposal.
3.2.1.3.
Thermochemical/flame decomposition of metalorganic precursors
Flame processes have been widely used to synthesize nanometer-sized particles of ceramic
materials. This is another type of gas-condensation technique with the starting material being a
liquid chemical precursor. The process is referred to as chemical vapour condensation (CVC). In
this process, chemical precursors are vaporized and then oxidized in a combustion process using
a fuel-oxidant mixture such as propane–oxygen or methane–air (Figure 3.4).
It combines the rapid thermal decomposition of a precursor–carrier gas stream in a reduced
pressure environment with thermophoretically driven deposition of the rapidly condensed product
particles on a cold substrate. The flame usually provides a high temperature (1200–3000 K), which
promotes rapid gas-phase chemical reactions.
.
107
Figure 3.4 – (A) Silicon nanoparticles by CVC methods; (B) Pieces of nanocomposites prepared by
adding metal salts to a sol before gelation8.
A variety of chemical precursors can be used including metal chlorides, such as TiCl 4 to prepare
TiO 2 and SiCl 4 to prepare SiO 2 10, metal-alkyl precursors, metal alkoxides, and gaseous metal
hydrides, such as silane as a source of silicon for the preparation silica. Chlorides have been the
most widely used precursors in the industry and the process is sometimes referred to as the
‘‘chloride process.’’ The high vapor pressure of chlorides and the fact that they can be safely stored
and handled make them excellent potential precursors. The disadvantages of using chloride
precursors are the formation of acidic gases and contamination of the products with halide
residues. Flame processes are used industrially to produce commercial quantities of ceramic
particulates, such as silica and titania. This is because of the low cost of production as compared
to all other methods. The disadvantage of flame synthesis is that the control of particle size (both
primary particle and aggregates size), morphology, and phase composition is difficult and limited.
3.2.2. Chemical Methods
3.2.2.1.
Sol–gel technique
The sol–gel process is typically used to prepare nanometer-sized particles of metal oxides
(Figure 3.5)12. This process is based on the hydrolysis of metal reactive precursors, usually
alkoxides in an alcoholic solution, resulting in the corresponding hydroxide. Condensation of the
hydroxide by giving off water leads to the formation of a network-like structure. When all hydroxide
species are linked, gelation is achieved and a dense porous gel is obtained. The gel is a polymer
of a three-dimensional skeleton surrounding interconnected pores. Removal of the solvents and
appropriate drying of the gel result in an ultrafine powder of the metal hydroxide. Further heat
treatment of the hydroxide leads to the corresponding powder of the metal oxide. As the process
starts with a nanosized unit and undergoes reactions on the nanometer scale, it results in
nanometer-sized powders. For alkoxides that have low rates of hydrolysis, acid or base catalysts
can be used to enhance the process
When drying is achieved by evaporation under normal conditions, the gel network shrinks as a
result of capillary pressure that occurs and the hydroxide product obtained is referred to as
xerogel. However, if supercritical drying is applied using a high-pressure autoclave reactor at
temperatures higher than the critical temperatures of solvents, less shrinkage of the gel network
occurs as there is no capillary pressure and no liquid–vapor interface, which better protects the
porous structure. The hydroxide product obtained is referred to as an aerogel. Aerogel powders
usually demonstrate higher porosities and larger specific surface areas than analogous xerogel
powders.
108
Sol–gel processes have several advantages over other techniques to synthesize nanopowders of
metal oxide ceramics. These include the production of ultrafine porous powders and the
homogeneity of the product as a result of homogeneous mixing of the starting materials on the
molecular level (Figure 3.5).
Figure 3.5- An example of sol-gel processing conditions on film formation8.
3.2.2.2.
Reverse microemulsions/micelles method
The reverse micelle approach is one of the recent promising routes to nanocrystalline materials
including ceramics. Surfactants dissolved in organic solvents form spheroidal aggregates called
reverse (or inverse) micelles9. In the presence of water, the polar head groups of the surfactant
molecules organize themselves around small water pools (≈100 Å), leading to dispersion of the
aqueous phase in the continuous oil phase.
Reverse micelles are used to prepare nanoparticles by using a water solution of reactive
precursors that can be converted to insoluble nanoparticles. Nanoparticles synthesis inside the
micelles can be achieved by different methods including hydrolysis of reactive precursors, such as
alkoxides, and precipitation reactions of metal salts.
Solvent removal and subsequent calcination lead to the final product. Several parameters, such as
the concentration of the reactive precursor in the micelle and the weight percentage of the aqueous
phase in the microemulsion, affect the properties, including particle size, particle-size distribution,
agglomerate size, and the phases of the final ceramic powders.
There are several advantages when this method is applied including the ability to prepare very
small particles and the ability to control the particle size. Disadvantages include low production
yields and the need to use large amounts of liquids.
3.2.2.3.
Precipitation from solutions
Precipitation is one of the conventional methods to prepare nanoparticles of metal oxide ceramics.
This process involves dissolving a salt precursor, usually chloride, oxychloride or nitrate, such as
AlCl 3 to make Al 2 O 3 , Y(NO 3 ) 3 to make Y 2 O 3 , and ZrCl 4 to make ZrO 2 11,12 . The corresponding
metal hydroxides are usually obtained as precipitates in water by adding a base solution such as
sodium hydroxide or ammonium hydroxide solution.
109
The remaining counter-ions are then washed away and the hydroxide is calcined after filtration and
washing to obtain the final oxide powder. This method is useful in preparing ceramic composites of
different oxides by co-precipitation of the corresponding hydroxides in the same solution. Solution
chemistry is also used to prepare non-oxide ceramics or pre-ceramic precursors that can be
converted to ceramics upon pyrolysis.
One of the disadvantages of this method is the difficulty in controlling the particle size and size
distribution. Very often, fast and uncontrolled precipitation takes place resulting in large particles.
3.2.2.4.
Chemical synthesis of pre-ceramic polymers coupled with physical
processing techniques
This method is based on the use of molecular precursors, which facilitates the synthesis of
nanomaterials containing phases of desired compositions. It involves a chemical reaction to
prepare an appropriate polymer, which is then converted into ceramic material upon pyrolysis.
Figure 3.6- General structural formulas of polycarbosilanes and polysilazanes13.
Using chemical reactions to prepare the pre-ceramic polymer not only allows control of phase
compositions but also overcomes the limitation of low production yields of the physical methods.
This method has been very useful in preparing nonoxide ceramics such as silicon carbide and
silicon nitride. The conversion of an organometallic precursor into a ceramic depends on different
parameters such as the molecular structure of the precursor and the pyrolysis conditions
(temperature, duration, and atmosphere). Metal carbides and metal nitrides have been obtained by
pyrolysis of polymers containing the appropriate elements such as Si or Al and C or N. These
polymers are called pre-ceramic polymers and are prepared from simpler chemical precursors. A
considerable amount of free carbon from the thermolysis process is very often a problem. Silicon
carbide (SiC) and silicon nitride (Si 3 N 4 ) are the most studied ceramic materials prepared via this
route. They are usually synthesized by the pyrolysis of polycarbosilanes and polysilazanes, for
which general structural formulas are shown in Figure 3.6, at temperatures between 1000°C and
1200°C.
This method could be employed to prepare Al 2 O 3 /SiC with ultrafine SiC particles using a
Si-containing precursor15. The polycarbosilane (Figure 3.6) is coated into a surface-modified
alumina powder and pyrolysed at 1500°C to produce ultrafine SiC particles of less than 20 nm.
The following flow chart shows the different elaboration methods generally employed for producing
Al 2 O 3 /SiC nanocomposites.
110
Figure 3.7 – Flow chart representing the processing of Al 2 O 3 /SiC nanocomposites by (A) classical
powder processing, (B) sol-gel processing and (C) the polymer coating route15.
3.2.2.5.
Colloidal processing route
Schehl et al.16 developed a new elaboration process for alumina-based nanocomposites (Figure
3.8), known as colloidal processing route. It consists in doping of a commercial, high-purity alumina
powder with different alkoxides. Precisely, alumina slurries were dispersed in absolute ethanol.
Suitable precursors were then drop-wise added to the alumina slurries. The slurries were first dried
under magnetic stirring at 70°C and subsequently in air at 120°C for 24 h in order to eliminate the
traces of alcohol. The dried powders were subsequently crushed in a mortar, to remove
agglomerates resulting from the drying process. The final phases are yielded by subsequent
calcinations processes.
When alumina particles are dispersed in ethanol, protons or hydroxyls are adsorbed on the surface
of the alumina particles, as it is shown in Figure 3.8. The addition of metal alkoxides to this
dispersion causes a substitution reaction between the metal alkoxides and the OH groups on the
surface of the alumina, as a result, the surface of the oxide particle is coated with a metal alkoxide.
Figure 3.8 – Substitutional reaction between the metal alkoxides and the OH groups on the surface
of the alumina16.
111
Palmero et al.17 showed an alternative route which consists in doping commercial alumina with
aqueous solution of inorganic salts.
3.2.2.6.
Mechanochemical synthesis
Mechanochemical synthesis involves mechanical activation of a solid-state reaction. This process
has been successfully used to make nanopowders such as Al 2 O 3 and ZrO 2 . Selected precursors
(usually a salt and a metal oxide) react under milling and subsequent heating, to form a mixture of
dispersed nanoparticles of the desired oxide within a salt. Nanoparticles of Al 2 O 3 (10–20 nm), for
example, can be prepared by milling AlCl 3 with CaO14.
2AlCl3 + 3CaO   -Al2O3 + 3CaCl2
(1)
3.3. Forming and Sintering
3.3.1. Dry Pressing
Dry pressing is a suitable method to produce simple solid shapes and consists of three basic
steps: filling the die, compacting the content, and ejecting the pressed solid.
Figure 3.9 - Stages in Dry Pressing18.
Figure 3.9 shows a schematic diagram of the double action dry-pressing process, in which both the
top and bottom punches are movable. When the bottom punch is in the low position, a cavity is
formed in the die and this cavity is filled with free flowing powder.
Once the cavity has been filled, the powder is pressed in the die. The top punch descends and
compresses the powder either to a predetermined volume or to a set pressure. During pressing,
the powder particles must flow between the closing punches so that the space between them is
uniformly filled.
A particle size distribution between 20 and 200 μm is often preferred for dry pressing: a high
volume fraction of small particles causes problems with particle flow and also results in the sticking
of the punches. The pressures used in dry pressing may be as high as 300 MPa, depending upon
material and press type. After pressing, both punches move upward until the bottom punch is level
with the top of the die and the top punch is clear of the powder-feeding mechanism. The compact
is then ejected, the bottom punch is lowered, and the cycle is repeated.
Because the dry-pressing process is so simple and involves low capital equipment costs it is the
most widely used high-volume forming process for ceramics. Production rates depend on the size
and shape of the part and on the type of press used.
112
3.3.2. Slip Casting
The slip is poured into a mold with the required shape. The porous nature of the mold provides a
capillary suction pressure, estimated to be of the order of ≈0.1–0.2 MPa, which draws the liquid
from the slurry into the mold. A consolidated layer of solids, referred to as cast, forms on the walls
of the mold (Figure 3.10). After a sufficient thickness of the cast is formed, the surplus slip is
poured out and the mold and cast are allowed to dry. Normally, the cast shrinks away from the
mold during drying and can be easily removed. Once fully dried, the cast is heated to burn out the
binder and sintered to produce the final sample.
Figure 3.10 - Schematic illustration of the drain-casting process: (a) the mold is filled with slip; the
liquid is drained from the mold, forming a compact along the mold walls, (b) after it is possible to
remove the sample once the sample is dried, subsequently green ceramic is removed18.
Among the various forming procedures, slip casting has received much attention, since it has the
main advantage of eliminating the drying step of slurries, which can lead to the formation of hard
agglomerates and, consequently affecting achievable green density13.
3.3.3. Sintering
A major challenge encountered in the consolidation of ceramic nanopowders is mitigation of grain
growth during sintering. Typically, a nanopowder compact experiences rapid densification during
the early stages of sintering, driven by the large surface to volume ratio. At this stage, the grain
size of the partially sintered material remains small due to the presence of a uniform distribution of
nanopores, which act as barriers to grain-boundary migration. However, in the final stages of
sintering (>90% of the theoretical density, TD), when the nanopores disappear, it is followed by
undesiderable grain growth that occurs, often leading to a micrograined sintered product.
In the following, the most exploited sintering methods to consolidate nanopowders into dense, fine
composite ceramics are briefly described.
3.3.3.1.
Hot Pressing
Gao et al.22 produced YAG-Al 2 O 3 composites from commercial alumina powders sintered by HP.
In order to obtain fully dense samples, sintering temperature was set to 1600°C for 1 h. The
applied pressure, during sintering, was 25 MPa. In Figure 3.11 are shown the microstructures of
two composites with different compositions.
113
Figure 3.11 – SEM micrographs of (A) YAG- 10 vol.% Al 2 O 3 and (B) YAG- 45 vol.% Al 2 O 3 22.
3.3.3.2.
Hot Isostatic Pressing
The hot isostatic pressing apparatus (Figure 3.12) consists of a high-temperature furnace enclosed
in a water-cooled autoclave capable of withstanding high internal gas pressures. As pressurization
gas, it is widely used argon and/or helium17.
HIP has the potential of solving some of the major limitations of hot-pressing. It makes possible
net shape forming because the pressure is equally applied from all directions. As a consequence,
the material has greater uniformity eliminating the preferred orientation, resulting high-strength and
Weibull modulus.
Figure 3.12 – Schematic diagram of a pressure vessel with a sample for HIP21.
3.3.3.3.
Spark-Plasma Sintering
Spark-plasma sintering is a low-pressure sintering method that makes use of a plasma discharge
through a powder compact to achieve rapid densification. The discharge is most effective when a
DC current is applied in an on-off pulsing mode. It has been suggested that DC pulsing generates:
(1) spark plasma,(2) spark impact pressure, (3) Joule heating, and (4) an electrical field diffusion
effect.
114
The inventors claimed that the pulsed direct current generates a spark discharge and/or plasma is
the responsible of cleaning the surfaces of the samples of CO 2 , H 2 O and OH-. These processes
promote the material transfer and make rapid densification of the powder compact possible at low
temperature and short holding time.
In a typical operation, powders are loaded into a graphite die and heated by passing an electric
current through the assembly. The experimental setup and electric field effects are illustrated in
Figure 3.13.
Figure 3.13 – Scheme of SPS sintering apparatus and mechanisms involved15.
The low heat capacity of the graphite die allows rapid heating, thus promoting heat and mass
transfer. Hence, SPS rapidly consolidates powders to near TD through the combined actions of
rapid heating rate, pressure application, and possibly powder surface cleaning. In most
investigations, SPS is carried out under vacuum. Starting with a cold-pressed (≈ 200MPa) powder
compact, typical processing parameters are: (1) an applied pressure of <100 MPa, (2) pulse
duration of 12 ms and pulse interval 2 ms, and (3) pulse current of ≈2000 A at a maximum voltage
of 10 V. Typical heating rates range from 150 to 500°C/min.
As Shen et al.22 summarised three reasons that contributes to the rapid densification:



The application of a mechanical pressure (as in HP and HIP).
The use of rapid heating rates.
The use of pulsed direct current, implying that the samples are exposed to an electric field.
SPS has been used to consolidate a wide variety of materials, including metals, intermetallics,
ceramics, composites, and polymers. As for functionally graded materials and nanocrystalline
materials, which are difficult to sinter by conventional methods, the advantage of SPS is more
evident. For example, Zhan et al.23 successfully sintered nanocrystalline α-Al 2 O 3 with 5 vol.%
carbon nanotubes at 1150°C in 3 min, while conventional hot pressing of α-Al 2 O 3 requires 1500 to
1600°C for 3 to 4 h. Hence, with a lower sintering temperature and shorter sintering time, SPS
enables better control of structure and properties of the consolidated material.
Gao et al.24 compared the microstructures (Figure 3.14) of YAG-Alumina composites obtained by
HP (A) and SPS (B). HP was performed at 1200 K, with a heating rate of 600°C/min and an
applied pressure of 40 MPa; SPS indeed was carried up at 1573 K (with a heating rate of
600 K/min) under an applied pressure of 40 MPa.
115
Figure 3.14 – SEM micrographs of YAG- 10 vol.% Al 2 O 3 (A) densified by SPS and (B) HP sintering
routes24.
As a result, the SPS specimens were fully dense with an average grain size 3 times smaller than
by HP as it is illustrated in Figure 3.14.
3.3.3.4.
Microwave Sintering
Microwave sintering allows the consolidation of the materials with shorter processing time than
conventional sintering. The heating flow is generated inside the samples (see Figure 3.15) as a
consequence of absorption of electromagnetic waves by electric dipoles, and not derived from
external sources as in conventional heating processes.
Temperature gradients are also reduced and an overall short sintering time minimizes grain
growth. In fact, by rapid heating rate, the low-temperature region in which grain growth occurs
during densification can be bypassed. In addition, binder is easily removed and thermal stresses
inside the samples are significantly reduced.
Since grain boundaries are the primary sites for electric dipoles, microwave sintering appears
particularly attractive for the densification of nanocrystalline powders. The method has been
applied to ceramic nanoparticles such as TiO 2 25-26 and Al 2 O 3 27-29.
Figure 3.15 – Comparison between conventional and microwave sintering techniques26.
116
In these materials, only densities lower than 95% have been achieved, with grain size smaller than
100 nm. For instance, microwave sintering had to be restricted to 1425 K to maintain a nanometer
grain size in γ-Al 2 O 3 which fully transformed to α-Al 2 O 3 , reaching a final density of 93%.
Tian et al.30 studied microwave sintering of Al 2 O 3 -TiC composites for potential cutting tool
applications. Although high densities were achieved, high power levels permit to reduce the
sintering temperature up to 1750°C.
3.4. The interest in nanocomposites ceramics: the mechanical properties
3.4.1. Role of the second phase on the retention of the matrix grain growth
Second phase fine particles, if uniformly distributed into a micronic matrix, can limit grain growth,
by the well known pinning effect32. In 1948, Zener32-33 explained how inclusions retard grain growth
and described the dependence of maximum matrix grain size (d max ) as a function on the particle
size (r p ) and second phases volume fraction (f):
d max 
4rp
3f
(2)
In Figure 3.16 (A) it is schematically shown how shrinking interacts with spherical inclusions.
During shrinking, the grains encounter the inclusion and an increasing proportion of the boundary
is removed. When N inclusions are simultaneously intersected, the maximum amount of grainboundary is πR2, which corresponds to a decreasing in free energy of πR2γ. If the grain continues
shrinking, the grain boundary area adjacent to the inclusions must withdraw up to the third position
see Figure 3.16 (A) up to the breaking, regaining its area occupied by the inclusion during bowing.
The relative variation of free energy versus the volume fraction is reported in Figure 3.16 (B).
Figure 3.16 – (A) Schematic representation of grain boundaries of a tetrahedral grain interacting
with spherical inclusions. (B) Energy of the tetrahedral grain versus its grain volume variation33.
In such a way, Zener provided the concept for the use of inclusions to hinder the grain growth. For
instance, the effect of a SiC inclusion is reported in Figure 3.17, in which it is represented the
alumina grain size in function of the sintering time at 1700°C. Retaining grain size effect by the
inclusions result more efficient when the second phase content increases.
117
Figure 3.17 – Alumina grain size evolution in function of sintering time and SiC content34.
A major grain growth restraint can be achieved by the elaboration of the so-called duplex
microstructures. Two immiscible phases, contained in similar volume fractions, can give rise to
interconnected structures, so that the growth of each phase is hindered by the other one and longorder interdiffusion is strongly limited31. For instance, in Figure 3.18, the microstructure of
Al 2 O 3 /50 vol.% ZrO 2 composite (Figure 2) is compared to those of the single-phase constituents,
sintered under the same conditions. A significantly finer microstructure was produced in the
composite, compared to the pure-phase materials.
Figure 3.18 – SEM micrograph of pure alumina (1), pure Zirconia (3) and of the composite
Al 2 O 3 /50 vol.% ZrO 2 (2)31.
3.4.2. Hardening
Hardness of materials is one of the most important mechanical properties from an engineering
point of view. Hardening in metals is performed by the inhibition of dislocation glide, which can be
done by microstructural tailoring. The Hall-Petch35-36 hardening is described by the following
relationship
H V = H 0 + k H d -1/2
(3)
where H is the measured hardness, Ho and k H specific constants - depending on the material and
d is the grain size - .
The similar concept will be used to explain the strengthening mechanism in the next section. In
ceramics, the correlation between hardness and microstructure has not been cleared, although
many papers reported this subject37. In addition, hardness in nanoceramics with grain sizes lower
than 1 μm has not been examined in detail.
118
Figure 3.19 – Variation of hardness vs. grain size for monolithic alumina50
From both the experimental and the theoretical points of view, it is difficult to estimate the plasticity
of hard elastic-plastic material which complicates the understanding of the hardness in ceramics, if
compared to metals.
Many works report hardness vs. grain size. Data reported by Shen et al.38 (Figure 3.19) for
monolithic alumina also showed the effect of the processing parameters on the hardness values.
By varying these parameters, it is possible to reduce the grain size and consequently reduce
mobility of dislocations in small grains.
3.4.3. Change in fracture mode
The addition of nano-size dispersed particles change the fracture mode from intergranular to
transgranular, resulting in a strengthening mechanism.
The first explanation was given by Awaji et al.39, who attributed the strengthening mechanism in
nanocomposites to dislocation activities. In Figure 3.20, the crack paths occurring in monolithic
alumina (a) and into an alumina-based nanocomposite (b) are shown.
In pure alumina, the residual stresses are formed at the interfaces as a result of the thermal
expansion anisotropy. When a crack propagates, it selects the interfaces under the residual
traction: the predominant fracture mode is intergranular (Figure 3.20 A).
On the contrary, in case of the composite (Figure 3.20 B), the dislocations generated around the
dispersed particles relieve the tensile residual stresses in the matrix, and consequently reduce the
defect size at the grain boundaries. The dislocations are difficult to move in ceramics at room
temperature: they act as stress concentrations sites and create small nanocracks around the main
crack tip. So, the change in the fracture mode from intergranular to transgranular was imputed to
the reduction of both the defect size along the grain boundaries and the transgranular strength in
the matrix.
119
Figure 3.20 – Schematic description of the strengthening mechanism as a function of crack path
for monolithic alumina (a) and nanocomposites (b)39.
A second explanation, given by Ohji et al.40, lies in the differences in relaxation of the tensile
residual stresses around the intergranular and intragranular particles. The internal stresses are
relaxed by lattice and grain-boundary diffusion around the intragranular and intergranular particles,
respectively.
As a consequence, the tangential tension around the intragranular particle of a sintered body is
always greater than around the interfacial particles, and in the case of crack, it will propagate
intergranularly (Figure 3.21).
Figure 3.21 – Schematic representation of internal stresses around the particles and crack
propagations40.
120
3.4.4. The strength
The matrix grain size in nanocomposites not only decreases with the decreasing of the sintering
temperature but it also decreases with increasing the second-phase content, as reported in section
3.4.1. This feature has to be taken into account for a true comparison of mechanical properties of
different nanocomposites. The strength of ceramics (σ), as already reported for hardness (Section
3.4.2), usually follows a Hall-Petch relation35-36:
   o  k .d 1/ 2
(4)
where σ is the strength for materials with grain size d, σ o is the friction stress and k is a constant.
Chantikul et al.46 carried out an investigation of the correlation between strength and grain size by
indentation tests on monolithic alumina with a grain size range from 2 to 80 μm, (Fig 3.22). The
authors found that strength follows the Hall-Petch relationship.
Figure 3.22 – Variation of fracture strength vs. grain size for monolithic alumina46.
Figure 3.23 collects data from different authors of strength of alumina-based nanocomposites as a
function of SiC content. The best results were obtained by Niihara1, who proved that only 5 vol.%
of nanosized SiC increased the fracture strength of pure alumina from about 300 to 1050 MPa.
Subsequent SiC addition lowers the strength at a constant value of 800 MPa, as a consequence of
SiC particles agglomeration.
Borsa et al.41 observed a slight increase in strength after de addition of SiC in one alumina matrix,
in contrast with the literature data 1,2 with show gains of 40%.
121
Figure 3.23 – Strength of Al 2 O 3 /SiC nanocomposites as a function of SiC content: (●) Niihara et
al.1 by three-point bend test and Vickers indentation; (□) Borsa et al.41 by four- point bend test; (▲)
Zhao et al.2 by four- point bend test and indentation-strength method; (▼) Davidge et al.42 by threepoint bend test and notched beams.
Niihara et al. 1 found a further improvement of strength up to 1550 Mpa, by post-annealing at
1300°C for 1 h in air or inert atmosphere (Figure 3.24). Chou et al.43 confirmed this behaviour: they
recorded an increase of about 50% of fracture strength of annealed composites as compared to
un-annealed ones.
Figure 3.24 – Improvement on fracture strength of alumina as a function of SiC content and postannealing treatment1
SPS was employed as a densification technique for Al 2 O 3 /SiC composites by Gao et al.44. The
authors studied the influence of the sintering temperature on the fracture strength. Figure 3.25
shows the strength dependence as a function of the sintering temperature. The maximum strength
of 1000 MPa is obtained at 1450°C. By increasing the sintering temperature over 1450°C there is
a decay of strength due to excessive grain growth.
Zhan et al.45 synthesized Al 2 O 3 /SiC composites by an innovative process named RHP (reactive
hot pressing). Mullite, aluminium and carbon were used as starting materials; they were heated at
1800°C, increasing the pressure up to 30 MPa. The final product was characterized by a SiC
content of 18.25 vol.% calculated from the formula below:
3  3 Al2O3 .2 SiO2   8 Al  6C 
13 Al2O3  6 SiC (5)
122
The produced composite material showed a very high fracture strength of 795 (+/- 160) MPa. The
Authors suggested that such high value was a consequence of the existence of a non amorphous
phase between the two phases, as evidenced by HR-TEM observation (Figure 3.26).
Figure 3.25 – Bending Strength versus sintering temperature for Al 2 O 3 /5 vol.% SiC sintered by
SPS44.
Figure 3.26– High-resolution TEM image of the grain boundary of Al 2 O 3 /SiC nanocomposite45.
3.4.4.1.
Reduction in processing flaw size
Niihara and Zhao et al.1,3 proposed that strengthening arise due to the refinement of the
microstructural scale and/or the compressive stresses around SiC particles which inhibit the grain
boundary fracture of that matrix during cooling from the fabrications. This phenomena is the
responsible for reducing the critical flaw size. The author found an increase from 6.5 to 13.6 by the
dispersion of only 5 vol.% SiC. The Weibull plot is reported in Figure 3.27. This conclusion
suggests that the reliability of Al 2 O 3 is improved by the nano-size SiC dispersions into the matrix
Al 2 O 3 .
123
Figure 3.27 – Weibull plot for Al 2 O 3 /5 vol.% SiC nanocomposites compared with the monolithic
alumina1
3.4.4.2.
Dislocation networks
Another strengthening mechanism for Al 2 O 3 /SiC nanocomposites based on flaw size reduction has
been presented by Niihara. He proposes that strengthening arises due to the refinement of the
microstructural scale from the order of the alumina grain size to the order of the interparticle
spacing, thus reducing the critical flaw size.
The occurrence of subgrain or low angle grain boundaries is widely acknowledged. During cooling
down, SiC particles can generate dislocations owing to internal stresses. At high temperatures
these dislocations can propagate and form dislocation networks.
However, Niihara’s strength and toughness values for the Al 2 O 3 /SiC system of 1 GPa and
4.8 MPa1/2, respectively, yielded a Griffith critical flaw size of 18 μm. A refinement of the
microstructure from the average alumina matrix grain size of 1.5 μm to the average interparticle
spacing of 200 nm plays a minor role.
3.4.4.3.
Crack healing
Zhao et al.47 suggest that SiC particles only indirectly influence the strength by enabling the
compressive stresses induced by the grinding process. These retained compressive stresses are
located on the surface region of the specimens. Another theory is that cracks in nanocomposites
can heal during annealing.
After annealing at 1300°C in Ar for 2 h the materials behave completely differently. Whereas
cracks in alumina grow, cracks in nanocomposites close, thus explaining the strength increase of
annealed nanocomposites.
3.4.5. Toughening mechanisms
3.4.5.1.
Intrinsic Fracture Energy
Toughening and strengthening mechanisms in nanocomposites could be explained by the Griffith’s
energy equilibrium and the residual stresses around second-phase particles dispersed in matrix
grains.
It is well known that many polycrystalline ceramic and ceramic-based composites exhibit R-curve
behaviour due to the crack bridging mechanism. The mechanism could be expressed as:
124
K R  a   K i  K R  a 
(6)
where K R (∆a) is the fracture toughness of the material exhibiting R-curve behaviour, K i is the
intrinsic fracture toughness, ∆K R (∆a) is the intrinsic increase in fracture toughness after certain
extension from the initial crack tip, ∆a.
An example of cracked surface in polycrystalline ceramics with rising R-curve behaviour is shown
in Figure 3.28.
Figure 3.28– Schematic drawing of a critical FPZ and the bridging in the wake of a polycrystalline
ceramics with the R-curve behaviour39.
The comparison between the previous formula and the Figure 3.28, indicates that the intrinsic
fracture toughness, K I , is related to the energy required to create the damaged FPZ at the crack tip
and that ∆K R is caused by the shielding effects of bridging in a process zone wake. The FPZ
(fracture process zone) is the place in which micro-failure mechanisms take place. Such processes
include micro cracking, crack deviation and crack branching which all contribute to the fracture
energy. So, the first term in equation (6) is a fracture energy for formation of the FPZ at the crack
tip, while the second one is the mechanical energy consumed at the bridging.
Therefore, it is possible to modify the Griffith-Irwin formula, which can be expressed for mode I
extension as:
KR2
 2.   i   R 
E'
(7)
The equation describes the critical energy release rate after certain crack extension in materials
with an R-curve, and γ i (mode I fracture per unit area of the cracked surface to create the critical
FPZ size) and γ R (fracture energy per unit area to be consumed at the bridging) are the intrinsic
and extrinsic fracture energy per unit area of the cracked surface.
Much effort has been done towards increasing the intrinsic energy consumed in the FPZ and the
extrinsic energy consumed at the bridging by means of whisker, fiber, and platelet reinforcement in
the matrix. The toughening mechanism of nanocomposites is related to the intrinsic fractured
energy, γ i , which is formed around the second-phases, and is expected to make many nanocracks
from the origins of stress concentrations, such as dislocations in a vicinity of a main crack tip. The
nanocracks play a role of expanding the size of FPZ and consequently enhancing the material
toughness39.
125
Figure 3.29 – Schematic description of the toughening mechanism for nanocomposites39.
Figure 3.29 reports an illustration of the FPZ at a main crack tip in an annealed Al 2 O 3 /SiC
nanocomposites. In a matrix grain, sub-grain boundaries with dislocation or dislocations networks
are generated around the dispersed SiC particles by means of annealing. When a main crack tip
reaches this area, the sessile dislocations in the ceramic matrix at room temperature will operate
as nano-crack nuclei in the highly stressed area. The FPZ consequently is expanded obliged by
nano-crack formation.
Ohji et al.39 proposed a comparison between the behaviour between the monolithic alumina and
the composite, for the same initial crack size. The strength of the materials can be determined by
the slope of the tangent line from the initial crack length to the respective R-curves (Figure 3.30).
As it was reviewed by the authors, the toughening of the monolith, due to grain bridging and
pullout, microcracking, etc. requires a crack extension of approximately 300 μm. However, for the
nanocomposites, the particle bridging operates inside a grain in the composite and consequently,
the fracture resistance increases in the 200-300 μm crack size.
Figure 3.30 – Variation of the fracture-resistance in function of the square-root of crack extension
for the Al 2 O 3 /SiC and the monolithic alumina40.
126
In Al 2 O 3 /SiC nanocomposites, Niihara 1 was the first in demonstrating the improvement in fracture
toughness up to 4.2 MPa.m1/2 corresponding to a 5% of SiC content. Other authors as Davidge 42
found a little improvement, as it is shown in Figure 3.31.
Figure 3.31 - Toughness of Al 2 O 3 /SiC nanocomposites as a function of SiC content: (●) Niihara et
al.1 by three-point bend test and Vickers indentation; (▼) Davidge et al.42 by three-point bend test
and notched beams.
3.4.5.2.
Crack Bowing
Particles in nanocomposites can also cause local changes in crack velocity. This effect can be
described as crack bowing and has been proposed as a mechanism for increasing the fracture
toughness of brittle materials. Green49 has developed an analytical expression to characterise
numerically the fracture toughness associated with the crack bowing effect. This expression
depends on the free interparticle spacing, A, but is independent of particle size. In nanocomposites
a crack can rest at SIC particles but it is not clear if crack bowing occurs. Again, a whole crack
front has to be investigated in order to study the effect of crack bowing.
Pezzotti et al.50 have presented a theoretical approach for modelling toughness and strength in
ceramic/ceramic and especially in ceramic/metal nanocomposites. Their model is based on the
effect of the crack bowing effect. The authors conclude that, in contrast to metallic inclusions,
ceramic nanosized dispersoids are completely ineffective on the material strength.
3.4.5.3.
Average Internal Stress
Since the microstructures of nanocomposite ceramics are formed during sintering at high
temperatures, differences in the thermal expansion coefficients of the matrix (α matrix ) and of the
nano-particles (α particles ) cause strains during cooling. These differences in thermal expansion <α*>
can be calculated by an integration over temperature. The upper limit is taken as the temperature
below which plastic deformation is insignificant (T plastic ) and the lower limit is the room
temperature15.
  * 
Tplastic
 
particle
  matrix dT
(8)
To
The thermal expansion misfit stress, σ T , inside a single spherical inclusion in an infinite matrix can
be described by the following expression:
T 
 * 
1   m 1  2 m

2 Em
2E p
(9)
127
where E and v are Young’s modulus and Poisson’s ratio of the matrix (m) and the particles (p). The
tangential, σ Tt , and the radial, σ Tr , stress distributions in the matrix around the particle are given by:
 Tt  
T  r 
 
2 x
3
3
r
 Tr   T   (10)
x
where r denotes the radius of the inclusion and x is the radial distance from the inclusion surface.
The residual stress around the dispersed particles could be represented using a simplified model
consisting of spherical particles within a concentric matrix sphere as it is shown in Figure 3.32 to
clarify the dislocation idea.
Figure 3.32 – A spherical particle with a glassy phase within a concentric sphere of a matrix
grain39.
For Al 2 O 3 /SiC the existence of a surrounding layer (glass-phase) on the second-phase particle is
considered for analysing the effects of the interlayer on the stress distributions. In Figure 3.33 he
residual stresses have the highest values at the particle/matrix boundary and reduce drastically as
the distance from the boundary increases.
Figure 3.33 – Comparison of stresses around a dispersed particle without an interlayer in matrices
of finite and infinite spheres39.
128
3.4.5.4.
Toughening by Transformation
Dispersions of zirconia particles can be used to toughen other ceramics. Zirconia-toughened
alumina (ZTA) is the most common example and allows the strength and toughness of alumina to
be improved while retaining some of its advantages over zirconia (i.e., it almost doubles the
stiffness)51.
ZTA is usually made by sintering mixtures of powders of alumina and zirconia (10 to 30 vol%)
containing some stabilizing oxide. The typical size of the zirconia particles in ZTA is similar to the
grain size of tetragonal zirconia polycrystals (TZP), and often the same amounts of stabilizer are
used as in TZP.
It should be noted, however, that the high stiffness of alumina will constrain the t → m
transformation more than in TZP, while the thermal expansion mismatch (≈5 × 10−6 K−1) between
the two ceramics, during cooling after sintering, causes tensile residual stresses in the zirconia
which will help the transformation of Zirconia.
To optimize the processing of transformation toughened material, it is essential to understand the
toughening mechanism and how it relates to microstructure. In understanding the origin of
transformation toughening, it is instructive to consider the propagation of a preexisting straight
crack as it is gradually loaded in fracture mode I crack by the application of an external force.
Initially, the crack tip is surrounded by untransformed material, which consists wholly or partly of
metastable particles of t-ZrO 2 (Figure 3.34 a). When a tensile load is applied, large stresses arise
close to the crack tip and the t-ZrO 2 in the region over which a critical stress (assumed
hydrostatic), σT, it is exceeded which cause the expansion of frontal zone of transformation as the
stress intensity at the crack tip is increased (Figure 3.34 b).
This frontal zone is dilated compared with the surrounding, untransformed material resulting in
residual stresses both within the frontal zone and in the untransformed material around it. The
residual stress, σ R , in the transformation zone is broadly compressive while the stress outside has
both tensile and compressive components and decays rapidly with distance from the particle. As
soon as the crack begins to grow the situation changes. The crack tip enters the frontal zone and
the compressive stresses within it, act to close the crack faces just behind the tip, producing a
strongly negative internal contribution, K T , to the total stress intensity (Figure 3.34 c).
129
Figures 3.34 – Schematic representation of events leading to transformation toughening: (a)
unstressed crack in matrix containing t-ZrO 2 particles, (b) stress-induced t→m transformation
occurs ahead the crack to form frontal zone of transformation and (c) crack grows into compressed
transformation zone which extends ahead of the crack51.
The condition for further crack growth remains so that the total stress intensity is equal to the local
toughness of the material and thus the externally applied stress intensity required for crack growth
increases and is given by:
K c  TL  KT
(11)
As K T is negative, the macroscopically measured toughness, K ∞c , is greater than TL and the
apparent toughness increases. This is the essence of the transformation toughening effect.
Many authors have shown how to increase the fracture toughness of alumina. Three decades ago
Claussen et al.51 published the first works related to the high fracture toughness of composites in
Al 2 O 3 -ZrO 2 system by transformation toughening. This type of material is generally characterized
by a micrometric matrix and a second, nanometric phase, excepting the works of Bhaduri et al.52,
Kim et al. 53 and Vasylkiv et al.56, which concern nano/nano-composites.
Bhaduri et al.52 measured the toughness in order of 8.4 MPa.m1/2 in nano/nano Al 2 O 3 /ZrO 2
composites which higher compared with conventional micro/nano nanocomposites. This result is
comparable with the micro/nano composites developed by De Aza et al.51 who measured 6
MPa.m1/2.
Schehl et al.16 produced Al 2 O 3 –5 wt. % ZrO 2 composites by the colloidal processing route (see
Section 3.2.2.5). They obtained a toughness 7.5 MPa.m1/2, significantly higher than that of
monolithic alumina (3.5 MPa.m1/2 1).
Zhan et al.54 produced Al 2 O 3 – 20 vol. % ZrO 2 nanocomposites by SPS with grain sizes of 265 nm
for alumina and 96 nm for zirconia. The full dense composite showed a fracture toughness of 8.9
MPa.m1/2, significantly higher than that of pure nanocrystalline alumina (3.3 MPa.m1/2), densified by
the same route.
Vasylkiv et al.56 produced yttrium stabilized tetragonal ZrO 2 and Y-TZP/Al 2 O 3 , with an addition of
0.2 to 0.7 wt.% of alumina. The Y-TZP with 0.35 wt.% Al 2 O 3 content exhibited the maximum
toughness of 15.7 MPa.m1/2. In the following Figure 3.35 the results obtained by the authors are
reported.
130
Figure 3.35 – Fracture toughness of ZrO 2 /Al 2 O 3 as a function of the Yttria-stabilizer
content56 and alumina weight percentage.
As in Al 2 O 3 /SiC, Niihara et al.58 have developed a nanocomposite Al 2 O 3 /SiC/Y-TZP. The authors
have observed that the incorporation of SiC limits the phase transformation of zirconia. The
authors found an improvement in fracture toughness due to the thermal coefficient mismatch
between the grains of SiC, Al 2 O 3 and ZrO 2 .
Nawa et al.59 have developed Ce-TZP/Al 2 O 3 which possesses a high strength, preserving
significant high toughness. They found an optimum result by adding 0.05 mol.% TiO 2 in
10Ce-TZP/30 vol% Al 2 O 3 composite. The researchers obtained a strength of 950 MPa and a
toughness of 18.3 MPa.m1/2, which is a good compromise in terms of strength and toughness.
3.4.6. Crack Deflection
The strong cohesion between the nano-particle and the matrix enables the crack to make a turn
within the nano-particle and to enter the matrix grain. The subsequent transgranular fracture
proceeds along the direction to maximize the mode I stress intensity factor.
The toughening factor  is defined by the following formula, as it was reviewed by Tan et al.60:
GC nm
   m  1 (12)
GC
where G c nm and G c μm are the critical energy release rates of the nanocomposite ceramics. The
comparison between the model and the experimental points, in which the volume fraction (Vf) of
the second phase is varied, as it is shown in Figure 3.36.
131
Figure 3.36 – Toughening factor of the nanocomposite ceramics with respect to the volume
fraction of the dispersed nanoparticles60.
The energy release rate for the tilt crack to advance with an θ angle, can be expressed with the
formula below:
G 
2
2
1 2
K I   K II 
E


(13)
The occurrence of intergranular or transgranular relies on the relative values of Gθ/G C gb and
Gθ/G C la, where G C gb and G C la are the fracture energy of the grain boundary and the fracture energy
of the lattice with the absence of the nanoparticles, respectively.
In case of mode I loading, it can be defined as a trajectory characteristic angle in the following way:
1/ 4
 G gb 
O  2 arccos  C la 
 GC 
(14)
Intergranular fracture occurs when  O   0,  O  while transgranular fracture occurs when
O   0,  / 2 . The probability of occurring an intergranular fracture f is defined by:
f 
2

(15)
Nano-particles along the grain boundaries may steer the crack into the matrix grains. In
Figure 3.37 (A) it is shown the case when the second phase is absent.
132
Figure 3.37– Effect of nanoparticles on grain boundary60
The main crack extends intergranularly along the grain boundary, since the fracture resistance of
the grain boundary is lower than that of the grain lattice. In Figure 3.37 (B), nano-particles scatter
along the grain boundary.
Figure 3.38 – Slice model of crack extension in nanocomposite ceramics60.
In which G c ngb is the critical energy release rate for the debonding of a nano-particle. Supposing
that the crack tip reaches the nano-particle, the angle formed between the main crack (along x 1
direction) and the matrix/nano-particle interface, is defined as θ. The energy release rate is G A for
the crack to advance along x 1 direction, and G B for the crack to advance along the matrix/nanoparticle interface.
The following expression describes the relationship between the energy release rate G A for the
crack to advance along the x 1 direction and G B which is the energy required to advance along the
matrix/nano particle interface.
GB  GA cos 4  / 2  (16)
In this context, transgranular fracture could only occurs when:
GA
GB
(17)
 nbc
la
GC
GC
requiring:
GC ngb  GC la cos 4  / 2 
(18)
where G C la is the lattice toughness of the second phase particles in case of crack entering the
nanoparticles, and the lattice toughness of Al 2 O 3 in case of crack entering the matrix.
133
A crack encountering a particle-free grain boundary can extent transgranularly by a probability of
1  f  . With the presence of a second phase, the probability raises to 1  f  f .Vf  .
The toughening by mechanism induced by the second nanophase could be expressed as:
GC la  GC gb
 I  fV f
GC  m
(19)
which indicates the direct proportionality with V f .
Microstructural data showed that cracks within the matrix grains exhibit a wavy path, as it is shown
in Figure 3.38. A transgranular crack with the macroscopic direction along x 1 is influenced by
nanoparticles. In Figure 3.38 nanoparticles A, B, C and D perturbed the crack path χ (x 1 ) formed
in the section π.
In this case the lattice resistance to a wavy crack during the transgranular fracture is G C la’.
Assuming that the surface energy is the same for a wavy or a flat crack, it is possible to write the
following equation:
GC la '/ GC la  l '/ l
(20)
where l’ is the arc length of the zigzag crack χ(x 1 ), and l is the projected length of χ(x 1 ) on the x 1
direction. The ratio between the arc length and the projected length of the zigzag crack determines
the toughening due to wavy fracture surface.
3.5. Creep
Creep is a time-dependent permanent deformation that is often due to diffusion processes rather
than dislocation motion. Under high temperature, the magnitude of the creep strain ε vs. time, t,
critically depends on the applied stress, σ, and on the temperature, T, according to the general
formula:
  f  , T , t 
(21)
Creep curves consist in a plot of strain vs. time; they can be usually separated into three stages:

Transient or primary creep: following a spontaneous elastic strain, the creep rate, έ
decreases with time;

Steady-state or secondary stage: strain increases with time, the creep rate is constant and
deformation may continue for a long time.

Tertiary stage: a rapid increase in creep rate just before failure.
It may be noted that the three stages are not independent because creep is a continuous
phenomenon. Depending on the applied stress or the temperature, one stage of the creep may be
missing.
Under low stresses, the third stage may disappear. In contrast, under high stresses, the secondary
stage may be replaced by an inflexion point. The typical creep examples are shown in Figure 3.39.
134
Figure 3.39 – Schematic illustration of creep curves expresses as strain vs. time at constant stress:
(a) presence of two stages (1 to 2) at low temperature or low stress, (b) presence of the three
creep stages (1 to 3), (c) at high temperature or high stress where the second stage is replaced by
and inflexion point62.
Generally, the analysis of creep results is only related to the “steady-state” stage. This avoids the
problems in defining equation that quantify creep-curves. Most creep curves models predict a
dependence of creep rate on the temperature. Most of creep models are based on the general
Norton relationship61:
.

A1.D  n
.
k .T d m
(22)
where A 1 is a material dependent constant, D the diffusion coefficient, k is Boltztman’s constant, T
the absolute temperature, d the grain size, σ is the applied stress, and m and n are the grain size
and stress exponents, respectively.
The diffusion coefficient is given by:
 Q 
D  Do .exp  

 RT 
(23)
where D O is a constant, R is the gas constant, and Q is the activation energy, m, n, and Q depend
on the creep mechanisms acting during creep. In creep analysis, Q is determined by a plot of ln(έ)
vs. 1/T.
3.5.1. Diffusion Creep
Diffusion creep refers to the deformation of crystalline solids by the diffusion of vacancies through
their crystal lattice. There is no dislocation motion; vacancies diffuse from the grain boundaries
located nearly perpendicular to the tensile axis to those located parallel to the tensile axis.
This process is named Nabarro-Herring creep63,64 (Figure 3.40 (A)) in which the steady state creep
is given by:
.
D .Gb  b 
. 
  9.3 l
kT  d 
2
 
. 
G
(24)
135
A second mechanism, proposed by Coble65, is due to vacancy flow occurring through the grain or
the grain boundary (Figure 3.40 (B)). The Coble creep is given by the following expression:
Gb     b 
  33.4.Dgb . .   .  
kT  b   d 
.
3
 
. 
G
(25)
In both equations, D l and D gb represent the diffusion coefficients in the lattice and at the grain
boundary, b is the Burgers’ vector, G is the shear modulus, and δ represents the effective grain
boundary thickness.
Both models lead to a stress exponent n=1, although the grain size exponent differ (m=2 or 3,
respectively). Cannon and Langdon66 found that Coble creep is prevalent when the grain size is
small.
Furthermore, Coble creep is favoured at low temperatures, because the activation energy for grain
boundary is lower. Nabarro-Herring and Coble creep can take place in parallel so that actual creep
rates will involve both components and both diffusion coefficients. Generally in ceramics, in the
situation in which anions and cations are diffusing creates further complications 62,64. Both
mechanisms are presented in the Figure 3.40.
Figure 3.40 – Diffusion creep and GBS (Grain Boundary Sliding) models62.
3.5.2. Creep of particle reinforced composites
Creep behaviour of nanocomposites, made by a micrometer-sized alumina matrix with nanosized
reinforcing particles, has been studied by many researchers. The most remarkable advantage of
ceramic nanocomposites lies in their high-temperature strength and the creep resistance.
However, mechanical properties at room temperature -as it was reviewed- are also significantly
improved. As Alumina/Silicon Carbide is extensively studied, it will be considered at first.
Ohji et al.68 have investigated the tensile bending creep of and alumina/17 vol.% SiC at 1200°C
and 1100°C in air. Nanocomposites and polycrystalline alumina were prepared by hot pressing of
mixed powder with average particle sizes less than 100 nm. In both cases, the chosen sintering
temperatures were similar, 1800°C for the nanocomposite and 1300°C for the monolith with the
aim of obtaining a similar matrix grain size ≈ 2 μm. In the particular case of nanocomposites, SiC
particles were located within and along the alumina grains. The typical tensile creep curves of
samples are shown in Figure 3.41.
136
Figure 3.41 – Tensile creep behaviour of monolithic Al 2 O 3 (curve 1) and Al 2 O 3 /SiC nanocomposite
(curve 2) at 50 MPa and 1473 K68.
As it is shown in Figure 3.42, the curve of monolithic alumina (1) exhibited a lifetime around 150 h
and creep stain of about 4 % at failure. On the contrary, the nanocomposite (2) had a lifetime of
1120 h (10 times longer) and a failure strain of 0.5 % (eight times longer). Under microstructural
observation, the nanocomposite presented less microcracks and cavities. The creep rate of the
nanocomposite was reduced in three to four order of magnitude (Figure 3.48). The stress exponent
for the monolithic alumina was 2.2 in tension and 2.9 in bending, in good agreement with literature
data (Canon and Langdon66). However, in case of the nanocomposites, it was 3.1 in tension and
2.2 in bending, respectively.
Figure 3.42 – Stress dependence of the strain rate in tension (curves 1 and 1’) or flexure (curves 2
and 2’) of Al 2 O 3 and Al 2 O 3 /SiC67.
A similar study was performed by Descamps et al.69, who investigated the differences of n values
between a monolithic alumina and 5 vol.% SiC. The authors found a contrast in n value, being 1.68
for monolithic alumina and 3.68 for the nanocomposites.
137
Figure 3.43 – Creep rate vs. inverse temperature of a monolithic alumina (curve 1) and a
nanocomposite (curve 2) at stress of 75 MPa69.
The incorporation of SiC dispersed particles leads to an important increase in activation energy
under bending stress of 75 MPa. In fact, the activation energy of the nanocomposite was 917 kJ
mol-1 in contrast with 466 kJ mol-1 for monolithic alumina (Figure 3.43).
A slightly different result was obtained by Thompson et al.70, for Al 2 O 3 /SiC nanocomposites
characterized by a matrix average grain size of 2.2 μm and dispersed submicrometer SiC particles
of 0.15 μm. During creep test, the nanocomposite exhibited an absence of both the primary and
steady-state; only the tertiary stage was recorded (Figure 3.44).
Figure 3.44– Strain rate vs. strain for an alumina/SiC nanocomposite at 1473 K at a tensile stress
of 100 MPa70.
Creep rate was decreased by two/three orders of magnitude, as compared with monolithic alumina
with a similar grain size (Figure 3.45). In addition, the effect of SiC morphology was investigated:
the particle and whisker-reinforced alumina composites showed a similar activation energy at 100
Mpa, of about ≈840 kJ mol-1.
138
Figure 3.45 –Creep rate vs. inverse temperature for monolithic Al 2 O 3 (curve 1) and Al 2 O 3 – 5 vol.%
SiC/ Al 2 O 3 – SiC w nanocomposites (curve 2)70.
In spite of these results, the role of the SiC particles is difficult to describe. Ohji et al.68 have
noticed a rotation of intergranular particles that “plunge” into alumina matrix in association with
grain boundaries sliding and the formation of small cavities around the SiC particles (Fig 3.46). The
plunging indicates an increase of particle pinning at grain boundaries which is the responsible for
obstructing the grain boundaries sliding phenomena.
Figure 3.46 – Microstructure of Al 2 O 3 /17 vol.% silicon carbide nanocomposite after tensile creep at
1573 K and 50 MPa68.
Ohji et al. in a second publication71, explained how SiC particles could alter the grain boundary
chemistry. The silicon carbide interface is much stronger than the alumina-alumina interface, which
contains a glassy phase. Thus, the improvement of the creep resistance may be attributed to the
inhibition of vacancy creation and annihilation at the SiC-Al 2 O 3 interface due to the decreasing of
the boundary diffusion rate.
139
The reduction of the boundary diffusion rate may promote a growth of isolated spherical cavities
leading to a decay in cavity connection and resulting failure.
Figure 3.47 – TEM micrograph of a nanocomposite after creep test, showing the presence of
plastically deformed grains71.
The augmentation of grain boundary cohesion by SiC particles inhibits grain boundary sliding and
some grains are forced by the neighbours. Thus, a plastic deformation based on the dislocation
motion is illustrated in Figure 3.47.
As it was reviewed, the increase in n values, coincides with the creep mechanism involving
dislocation glide or climb, thus supporting the assumption that dislocation motion plays a role in
creep behaviour. Arzt and Grahle72 proposed that climbing grain-boundary acts as a source for
vacancy diffusion processes and they presented a model based on the pinning of these
dislocations by boundary-grain particles.
The best results for creep resistance for Al 2 O 3 /SiC were obtained by Descamps et al.70 for 10
vol.% SiC. It seems that the addition of nanosized particles is the only condition that improves
creep resistance, as it is also important to tailor the microstructure, by locating the particles at grain
boundaries.
3.6. The Alumina-YAG system: elaboration and properties
3.6.1. Alumina-Yttria phase diagram
The early attempts of Rhodes and Reid73 to sinter transparent lamp envelopes began with an
attempt to sinter pure Y 2 O 3 . One deagglomeration procedure was achieved by ball-milling using
the highest purity AI 2 O 3 spheres. The researchers found that samples reached higher densities.
140
Figure 3.48 – Y 2 O 3 -Al 2 O 3 phase diagram proposed by Noguchi and Mizuno74.
Microprobe results confirmed that they had unintentionally "contaminated" the powder with an
excellent sintering aid. An extensive program was launched to take advantage of this accidental
finding.
The most recent Y 2 O 3 -Al 2 O 3 phase diagram by Noguchi and Mizuno74 (Figure 3.48), shows a solid
solution region up to 4 mole % Al 2 O 3 in Y 2 O 3 at 1940°C. This portion of the diagram is dashed to
indicate some uncertainty. The initial work to use Al 2 O 3 deliberately as a sintering aid was based
on the theory that it should be possible to sinter in the liquid plus Y 2 O 3 field and then drive the
Al 2 O 3 back into solid solution by annealing at a lower temperature (1940°C). If this is successful, it
would qualify as a transient liquid-phase sintering technique.
An extensive research program was carried out with Al 2 O 3 additions from 0.02 to 10.0 mole % and
sintering temperatures from 1800 to 2200°C in a H 2 atmosphere. Numerous attempts were made
to incorporate three temperature holds. Samples were held at temperatures ranging from
1700 to 1900°C to allow diffusion enough time to form the solid solution; then sintered above the
1940°C eutectic over a range from 2000 to 2200°C and annealed for 1 to 8 h in the range of 1800
to 1940°C to drive the eutectic phase back into solution (Figure 3.48).
The researchers did not corroborate after the sintering cycle, the formation of a solid solution. The
only effect of the lower temperature held at the end of the cycle was to crystallize the grainboundary liquid phase.
141
Figure 3.49 – Translucent microstructure achieved with 0.5 mole % Al 2 O 3 74.
In Figure 3.49 it is shown the microstructure after three-temperature sintering cycles in an attempt
to minimize the second phase. The residual grain boundary phase is evident even though the
sample's starting composition contained 0.5 mole % Al 2 O 3 . The effectiveness of liquid-phase
sintering in pore elimination is also shown. Extensive studies on the effect of powder properties on
sintered optical properties were discussed by Palilla et al.73 in conjunction with the development of
this material as a lamp envelope.
The lamp envelope development work demonstrated that the solubility limit may be on the order of
100 ppm, supporting the earlier Al 2 O 3 -Y 2 O 3 phase diagram of Toropov et al.75, which is shown in
Figure 3.50. The Y 2 O 3 -rich end of the diagram shows no solubility for AI 2 0 3 . The cubic to
hexagonal transition at ~ 2200°C for pure Y 2 O 3 is not accounted (Figure 3.50).
Figure 3.50 – Y 2 O 3 -Al 2 O 3 phase diagram proposed by Toropov et al75.
142
It is still not clear which phases appear during cooling of Al 2 O 3 -Y 2 O 3 mixtures with compositions
corresponding to YAG, YAlO 3 (YAP) or Y 4 Al 2 O 9 compounds. The pseudo-binary Al 2 O 3 -Y 2 O 3
system is shown in Figure 3.50 76. Caslavsky et al.76 performed optical differential thermal analysis
and reported that it melts if heated to temperatures below 2263 K following the equilibrium phase
diagram (Figure 3.51), while it melts cooled down from temperatures above 2263 K followed the
metastable phase diagrams. However, in general, the selection of the solidified structure is
determined not by the melting temperature before cooling but by a combination of nucleation and
the growth velocity of the interface.
Figure 3.51 –Al 2 O 3 -rich portion of the phase diagram of the Al 2 O 3 -Y 2 O 3 system76.
The maximum melting temperature before cooling significantly affects nucleation of YAG. The melt
heated up to temperatures above 2273 K never nucleates above 1973 K. Melts heated above
2273 K are difficult to solidify. Even when the melt is kept at 1993 K for 1.8 ks. solidification does
not occur. Caslavsky et al. 76 reported a possibility of two immiscible liquids above 2273 K,
because the melt was opaque for the YAG composition.
If a miscibility gap exists above 2273 K and the melt is kept above this temperature, coarsening
should occur, but this was not the case in Caslavsky’s experiments. Mituzani et al.68 have
developed equipment for optical differential thermal analysis which has been used to measure the
differential thermal analysis curves of Al 2 O 3 -18.5mol% Y 2 O 3 . However, no exothermic or
endothermic heat was detected during heating up to 2473 K. Coordination of oxygen around
aluminium may have an important role in the melt68. Fourfold coordination exists in the YAG
structure, while only Al-O octahedra exists in the α-Al 2 O 3 and YAP structures. Coordination in the
melt may change during heating and affect the nucleation behaviour. For example, the Al 2 O 3 /YAG
system appears even when specimens solidified as the metastable Al 2 O 3 /YAP system are
remelted and slowly cooled.
As it was described above, alkoxide and metalorganic derived precursors have been used to
produce oxide phases in the Al 2 O 3 –Y 2 O 3 system. Compositions aimed at garnet frequently
produce YAG as the first product of crystallisation 77-78, though occasionally an intermediate
hexagonal YAlO 3 served as the crystalline precursor to YAG81-83. These results point out the need
to preserve molecular homogeneity during hydrolysis (gelation) or during decomposition of the gel
or metalorganic precursor.
143
In situations where segregated phases appear (a.i., Y 2 O 3 , Y 4 Al 2 O 9 , YAlO 3 ) a subsequent reaction
is needed at higher temperatures to produce garnet73, except in one instance where in YAG was
formed using separate Y and Al precursors in an autoclave at only 300°C80. In Figure 3.52, it is
presented the most complete phase diagram found in literature, published by Roth et al85.
Figure 3.52 –Al 2 O 3 -Y 2 O 3 system phase diagram with the investigated positions85.
3.6.2. Mechanical properties at room temperature and high temperature
Al 2 O 3 -YAG composites are deeply investigated, thanks to the high creep resistance of YAG, that
makes such composites suitable for high temperature applications. In addition, the two phases
have mutual insolubility, similar thermal expansion coefficients and chemical stability86.
Figure 3.53 – Schematic procedure for producing directionally solidified Al 2 O 3 -YAG eutectic
composites88.
144
Mah et al.87 studied the processing and the mechanical properties of Al 2 O 3 /YAG eutectic
composites prepared by the directionally solidified method, as it is shown in Figure 3.53. The
authors found that composites had flexural strength of 373 MPa and a fracture toughness of 4.3
MPa.m1/2 at room temperature. In comparison with monolithic Al 2 O 3 or YAG, the composite has a
significantly higher fracture toughness at elevated temperature. Waku et al.88 prepared Al 2 O 3 -YAG
nanocomposites by hot-pressing at 1973°C under a pressure of 50 MPa. The nanocomposite has
a important flexural strength of 450 MPa, but its strength felt drastically above 1073°C.
Vrolijk et al.89 reported a comparison between a monocrystalline and polycrystalline Al 2 O 3 /YAG
samples prepared by hot-pressing (see Figure 3.54) showing their mechanical behaviour at a lowand high-temperature regime. It is clearly shown the advantages of producing Al 2 O 3 /YAG eutectic
composites, as the flexural strength is maintained almost constant up to the melting point.
Figure 3.54 – Flexural Strength of Al 2 O 3 /YAG eutectic composites: comparison between the
unidirectionally solidified and the polycrystalline samples89.
The production of eutectic composites is quite complex; so, the elaboration of fine-grained
polycrystalline Al 2 O 3 -YAG may be a useful solution.
The discrepancy regarding the flexural strength was explained by the authors, as the difference of
interface between both phases. The polycrystalline exhibits an amorphous layer detected by
means of TEM (Figure 3.55 a), while the monocrystalline does not have it (Figure 3.55 b).
Although, the reports concerning the mechanical properties of polycrystalline Al 2 O 3 /YAG
composites are still very few.
Figure 3.55 – High resolution TEM image of the grain boundary between the Al 2 O 3 and YAG: (a)
polycrystalline and (b) monocrystalline samples89.
Li et al.90 prepared Al 2 O 3 /25 vol.% YAG euthectic composite using the co-precipitation method.
The samples were sintered by hot pressing up to 1400°C for 1 h and at 30 MPa. The room
145
temperature fracture strength and fracture toughness of the composites were 611 MPa and
4.53 MPa.m1/2, respectively, thanks to the reduction of the grain size of the matrix and the good
dispersion of the YAG phase in both inter and intra-granular positions.
Figure 3.56 – Temperature dependence of fracture strength for monolithic alumina and
YAG systems91.
French et al.91 described the variation of fracture toughness by increasing the temperature, as it is
shown in Figure 3.56. The authors mainly compared an Al 2 O 3 /50 vol.% YAG with the single-phase
constituents. The nanocomposite exhibited a lower decay of toughness among the others,
confirming the benefit of producing this type of materials.
Schehl et al.16 prepared Al 2 O 3 / YAG polycrystalline composites by colloidal processing (see
Section 3.2.2.5). Materials containing YAG showed an important K IC value of 5.8 MPa.m1/2
compared with the monolithic alumina (4.5 MPa.m1/2). The increase in K IC was attributed to the
reaction between YAP and Al 2 O 3 to give YAG which consequently gave rise to a volume increase.
This phenomena creates a homogeneous stress field at alumina grain boundaries which is the
main responsible of blocking the flaw formation during sintering.
The behaviour of yttrium-doped alumina in the high-temperature regime has been a subject of
several studies. From these results, it was clarified the beneficial effect of YAG on high
temperature deformation, as the creep rate could be lower than two order of magnitude compared
to monolithic alumina. The phenomenon was attributed to the segregation of Y3+ at Al 2 O 3 grain
boundaries, which blocks the diffusion of ions along the grains boundaries, with a consequent
reduction in grain-boundary diffusion and thus decreasing the creep rate. If creep rate is controlled
by point defects on their transport along grain boundaries, the strong segregation of Y3+ at grain
boundaries is likely to hinder this process.
Duong et al.92 investigated the creep behaviour of fine-grained two-phase Al 2 O 3 -YAG over a
temperature range 1400°C-1500°C under applied stress between 3 and 20 MPa. The authors
prepared two composites. Al 2 O 3 /50% vol.% YAG and Al 2 O 3 /75% vol.% YAG nanocomposites,
labelled as AY50 and AY75 respectively, which were sintered in air at 1600°C for up to 120 h.
In both cases, microstructure was composed by an average matrix grain size of about 8-10 μm,
while YAG had an average grain size of 3 μm in both samples. Figure 3.57 shows the creep rate at
1400°C for composites compared with the single constituents. The Al 2 O 3 /50% vol.% YAG has the
lowest strain rate among all the samples. The authors attributed the difference in creep behaviour
since AY75 has the finer microstructure among the other materials involved.
146
Figure 3.57 - Stress dependence of the strain rate in tension for Al 2 O 3 and YAG composites at
1400°C92.
Other authors, such as French et al.93, have studied the creep behaviour of duplex microstructures.
In this particular work, the nanocomposites were produced by mechanical mixture (ball milling) and
solid state reaction of the single constituents, finally sintering in air at 1650°C for 2 h. The final
microstructure was characterized by an average grain size of ≈ 2 μm. Three extra samples were
prepared to be compared with the composite; the Al 2 O 3 , YAG and Al 2 O 3 doped with 100 ppm of
yttrium.
In Figure 3.58, the creep rate plots for different systems are exposed for an applied stress of
75 MPa. French and co-workers arrived to the conclusion that with only 1000 ppm of Y3+ it is
possible to decrease the creep rate by two orders of magnitude. The activation energy for this
1000 ppm yttrium-doped Al 2 O 3 was 698 kJ/mol.
Figure 3.58 - Creep rate vs. (a) applied stress and (b) temperature for the single constituents
Al 2 O 3 /YAG, Al 2 O 3 doped with Y+3 (1000 ppm) and AY50 nanocomposite91.
The most relevant found in literature values regarding to Al 2 O 3 /YAG nanocomposites are
summarized in Table I.
147
% vol. YAG
σ (MPa)
T (°C)
n
Reference
1500-1600
1100-1400
1400-1500
Grain Size
(μm)
2
2
8-10
0
50
50
75
5
≈5
100
10-152
35-75
10
1
2.6
1.1
94
40-80
30-250
20
1400
1200-1400
1400-1600
≈1.6
3.6
3-6.3
1.2
1-1.46
1.2
95
93
92
96
97
Table I – Summary of creep tests data for Al 2 O 3 and YAG systems.
Schehl et al.16 as it was mentioned before, sintered an alumina-YAG composite in air at 1600°C for
2 h. For this particular nanocomposite, the final microstructure was formed by alumina matrix
grains with an average grain size of ≈4 μm, and YAG grain in both inter-intragranular positions of
350 nm.
The results obtained by Schehl et al.16 confirmed the positive effect of the dopant. The values
obtained by the authors for the nanocomposite and monolithic alumina are summarized in Table II.
In order to compare the strain rates, the creep rates measured were normalised with the smallest
grain size employing the following formula6.
p
 d  .
 n   A  . S
 d AY 
.
(26)
.
where  n is the measured creep rate for Y-TM and Y-CR, d A is the mean grain size of the
monolithic alumina and d AY is the mean grain size of the composite.
For the calculation of the normalised creep rate, the authors used grain size exponent (p) equal to
3 for grain boundary diffusion controlled creep and 2 for lattice diffusion controlled creep.
Sample L (μm)
A
5.9
AY
3.56
Strain Rate (s-1)
5.7.10-9
2.0.10-9
Strain Rate for p=2 (s-1)
7.5.10-8
9.9.10-9
Strain Rate for p=3 (s-1)
2.7.10-7
2.2.10-8
Table II – Creep test data obtained by Schehl et al.12
By using p=2 the creep rates have approximately one order of magnitude lower than monolithic
alumina. However, employing the p=3 the value is lowered 13 times.
Other authors as Satapathy and Chokshi95 produced Al 2 O 3 /YAG nanocomposites by solid-state
reaction of alumina/yttria powders during sintering at 1500°C for 6 h. The authors found a similar
behaviour as Schehl et al.16 concluding that creep mechanism is controlled the grain boundary
diffusion65.
However, for normalised creep rates using p=3, the achieved values differ from the previous work,
obtaining a decrease of only a 3 times at 1400°C. The authors concluded that over the chosen
temperature, the YAG phase does not influence significantly the creep mechanism. The results
obtained by the authors are summarized in Figure 3.59.
148
Figure 3.59 - Creep rates
and Al 2 O 3 -5 vol.% YAG 95.
vs.
(a)
applied
stress
and
(b)
temperature
Al 2 O 3
The main goal of a second work, from Schehl and co-workers96 was to extend the information
regarding alumina-YAG nanocomposite. The authors arrived to the conclusion that the creep
mechanism is lattice diffusion (Nabarro-Herring) by measuring a stress component of nearly 1 at
1200-1300°C. In Figure 3.60 it is shown the comparison of creep rates for different applied
stresses between the alumina-YAG composite and the monolithic alumina.
Figure 3.60 - Creep rate vs. applied stress of monolithic Al 2 O 3 and Al 2 O 3 ≈5 vol.% YAG96.
Activation energies calculated for monolithic alumina were 517 kJ/mol for an applied stress of
50 MPa and 527 kJ/mol 70 MPa, in good agreement with the literature93. However, in the case of
the AY nanocomposites the activation energies were 554 kJ/mol for an applied stress of 70 MPa
and 595 kJ/mol for an applied stress of 100 MPa, respectively. The normalised creep rates with
p=2 are shown in Figure 3.61.
The authors explained that YAG particles at grain boundaries avoid the grain boundary sliding
during deformation. Due to the high activation energy required to deform YAG nanosized particles,
149
the activation energy associated with the grain boundary diffusion of Al3+ rises to values close to
polycrystalline YAG.
Figure 3.61 – Normalised creep rates of Al 2 O 3 and Al 2 O 3 -5 vol.% YAG96.
A significant improvement of strain rate of about 1 order of magnitude was found by the authors in
1200-1400°C range for AY nanocomposites.
150
CHAPTER
4
Elaboration and Characterization
of
Al 2 O3 -5 vol.% YAG Nanocomposites
151
Elaboration and Characterization of Al2O3-5 vol.% YAG Nanocomposites
4. Introduction
Two commercial α-alumina nanopowders were employed to prepare Al 2 O 3 -5 vol. % YAG
nanocomposites. Powders have been doped by using aqueous solutions of yttrium salts and
subsequently by a solid state reaction is yielded the YAG phase.
The aim of this work is to investigate the effect of different densification routes and forming
methods on the microstructural and mechanical properties of Al 2 O 3 -5 vol. % YAG
nanocomposites.
As it was reviewed in chapter III, a second phase is believed to reinforce the Al 2 O 3 by introducing
nano-size dispersoids within matrix grains and the grain boundaries. Nanoparticles are the
responsible of modifying the crack propagation behaviour due to the difference in thermal
expansion coefficients1,2.
YAG is believed to be a suitable candidate as its creep resistance, and its thermal expansion and
stability in contact up to 1700°C3. Many researchers explained this improvement as cation ions
(Y3+) segregated to alumina grains boundaries reduce the grain boundary diffusivity, thus
decreasing the creep rate5-7.
Torrecillas et al.7 produced Alumina-YAG composites by the colloidal processing route, obtaining a
significant improvement of nearly 2.5 times in creep rate in a range of 1200-1400°C. Recently,
Palmero et al.4 successfully exploited the post doping of commercial power by using YCl 3 to
prepare Al 2 O 3 -5vol.% YAG.
The materials developed were submitted to mechanical characterization correlating the
microstructural features to the mechanical properties.
4.1. Characterization of the as-received powders
Two commercial α-alumina powders were employed for the production of Al 2 O 3 -5 vol.% YAG
composites.
TM-DAR TAIMICRON, supplied by Taimei Chemical Co., Japan, is made of pure alpha alumina.
It is characterized by a theoretical density of 3.96 g/cm3, a specific surface of 4.5 m2/g and a
particle mean size of 350 nm. TM is produced by thermal decomposition of aluminum salt and
ammonium bicarbonate as it is claimed by the producer8. XRD analysis performed on the starting
material confirmed that only α-alumina (ICDD 81-2266) is present, as it is presented in Figure 4.1.



Intensity (a.u.)






10
20
30
40
50

60
70
Figure 4.1 – XRD pattern of the as-received TM-DAR TAIMICRON
152
Similarly, as for Nanotek the initial granulometry of TM-DAR TAIMICRON was evaluated by laser
granulometry after dispersing the powder in distilled water in obtain a dilution factor of 7 vol. %.
The cumulative distribution by volume and number are presented in Figure 4.2.
100
80
60
40
20
0
0
5
10
15
20
25
Figure 4.2– Cumulative size distribution by volume (solid line) and by number (dashed line) as a
function of agglomerate size of as-received TM-DAR TAIMICRON
Laser granulometry lead to determine agglomerates size of 8.6, 29.5 and 66.7 μm, corresponding
to (d 10 ) 10, (d 50 ) 50 and (d 90 ) 90% of the cumulative volume distribution. In case of the number
distribution, the agglomerates sizes are <0.3, 0.36 and 0.61 μm corresponding to (d 10 ) 10, (d 50 ) 50
and (d 90 ) 90 %, respectively.
SEM micrograph of the as-received TAIMEI powder provided by the supplier is shown in Figure 4.3
which confirms that the powder is sub-micrometric.
Figure 4.3 – SEM micrograph of TM-DAR TAIMICRON8.
The second powder employed in this study the CR1 alumina powder, labelled in this study CR,
supplied by Baikowski, France9. Powder CR is characterized by a higher average particle size of
about 0.6 μm and a lower specific surface area of 3 m2/g. In the following Figure 4.4, is shown the
morphology of the CR1.
153
Figure 4.4- SEM micrograph of as-received CR19.
As supplied CR powder was characterized by laser granulometry, yielding agglomerate size of
0.77, 2.36 and 4.57 μm, corresponding to (d 10 ) 10, (d 50 ) 50 and (d 90 ) 90 % of the cumulative
volume distribution. In case of the number distribution, the agglomerates sizes are <0.3, 0.4 and
1 μm corresponding to (d 10 ) 10, (d 50 ) 50 and (d 90 ) 90 %, respectively. The results are shown in
Figure 4.5.
100
80
60
40
20
0
0
10
20
30
40
50
60
70
80
90
100
Agglomerate size [m]
Figure 4.5- Cumulative size distribution by volume (solid line) and by number (dashed line) as a
function of agglomerate size of as-received CR.
4.2. Elaboration of Al 2 O 3 - 5vol.% YAG composites
4.2.1. Dispersion and Spray Drying
The powders were dispersed in distilled water and ball milled for 3 h and 113 h for TM-DAR and
CR1 powders, respectively.
154
After dispersion, a significant reduction of the starting agglomerate size was achieved. In the
particular case of TM, the agglomerate sizes were <0.3, 0.53 and 1.12 μm corresponding to (d 10 )
10, (d 50 ) 50 and (d 90 ) 90 % of the cumulative volume distribution. For the second powder CR, after
ball milling due to the agglomerates in a larger extent, the values decreased to 0.44, 0.81 and
1.37 μm. In Figures 4.6, the cumulative size distribution by volume of the as-received and
dispersed powders are collected.
100
80
60
40
20
0
0
10
20
30
40
50
a)
100
80
60
40
20
0
b)
0
10
20
30
40
50
60
70
80
90
100
Figures 4.6- Cumulative size distribution by volume of the as-received (solid line) and dispersed
(squares) TM-DAR (a) and CR1 (b) powders.
Thereafter, an aqueous solution of YCl 3 .6H 2 O (Sigma-Aldrich, 99.99%) was added and dispersed
in both slurries. Suspensions were maintained under magnetic stirring for 2 h. The slurries were
diluted to 4 wt. % and spray-dried.
4.2.2. Thermal Evolution of the powders
Doped powders were calcined at different temperatures and submitted to X-ray diffraction. The
tests were performed by plunging the powders into a tubular furnace kept at fixed temperature for
three minutes. This technique was selected in order to avoid the crystallite growth and the
aggregates formation. By this method, it was also prevent the possible elution of yttrium dopant
during the slip casting process.
155
The first sample analyzed was doped-TM. As it is presented in Figure 4.7, Yttrium aluminates start
to crystallize at 1050°C, with the formation of YAP phase (as a trace), whose intensity slightly
increased by increasing the calcination temperature. Starting from 1220°C, also YAG phase, near
YAP, was detected. Consequently, in case of doped-TM by increasing the calcination temperature
up to 1250°C it is possible to assign the first peaks corresponding to the crystallization of YAG at
1220°C. In this particular case, the sample was labelled as Y-TM.

P
Y
Y
Y
Y

Y

P
P

P

P

P
1250°C



counts (a.u.)



1220°C



1200°C




1150°C


1100°C



1050°C
25
30
35
40
45
50
2 Theta (degrees)
Figure 4.7- XRD pattern evolution of Y-TM in the range 22-50° 2θ
After high-temperature calcination of Y-TM at 1500°C for 3h, it is possible to obtain a composite
composed only by well-crystallized YAG and Al 2 O 3 as it is shown in Figure 4.8.


12000

Intensity (a.u.)
10000

8000

6000
4000


 
Y
2000
Y
Y
YY
20
30
Y 
Y Y
Y Y 
0
10
40
50
60
70
2 Theta (degrees)
Figure 4.8- XRD pattern of Y-TM calcined at 1500°C for 3 h (α= α-Al 2 O 3 , Y= YAG)
156
By “flash” calcining the doped-CR at 1050°C for 3 min, it is possible to observe pure YAG
(JCPDF N°33-0040) crystallized near α-Al 2 O 3 . As in the previous case, the treated doped-CR was
labelled as Y-CR.
In Figure 4.9, the XRD comparison between Y-CR 1050°C and Y-TM 1150°C is presented in the
22-50° (2θ) range.


counts (a.u.)


Y Y
Y
Y
Y
Y
Y-CR 1050°C
P
Y-TM 1150°C
25
30
35
40
45
50
2 Theta (degrees)
Figure 4.9- XRD patterns in the 22-50° 2θ of Y-CR “flash” heated at 1050°C for 3 min and Y-TM
“flash” heated at 1150°C for 3 min (α= α-Al 2 O 3 , Y= YAG, P = orthorhombic YAlO 3 )
As a conclusion, Y-TM and Y-CR powders differ in crystallization behaviour. In fact, Y-TM “flash”
heated at 1150°C for 3 min shows traces of orthorhombic perovskite YAlO 3 (JCPDF N° 70-1677,
YAP) and α-Al 2 O 3 (JCPDF N° 46-1212).
On the ground of the results obtained, “flash” calcination temperatures were 1150°C for TM-DAR
and 1050°C for CR1.
4.3. Forming and Sintering
Powder suspension was prepared with a solid loading of 50 wt. % and ball-milled for 24 h.
Subsequently, suspensions were cast into slip-casting moulds (SC) - made of Plexiglas which lay
on a porous alumina plate - (Figure 4.10) or dry in oven before uniaxial pressing (P). Subsequently,
Y-TM and Y-CR were sintered by natural sintering (NS), hot-pressing (HP) and spark plasma
sintering (SPS).
Figure 4.10 - Slip-casting moulds used to produce the samples.
157
For a sake of clarity, the complete elaboration process is schematically shown in the flow-chart of
Figure 4.11.
TM-DAR / CR1
dispersed slurries
YCl 3 6H 2 0
aqueous solution
Doped
suspensions
Homogenization
for 2 h
Spray drying
Fast calcination at
1050°C (TM-DAR) and 1150°C (CR1)
Y-TM and Y-CR powders
Uniaxial pressing.
Y-TM-P
Y-CR-P
Slip casting
Y-TM-SC
Y-CR-SC
Natural sintering: NS
Hot-Pressing: HP
Spark Plasma Sintering: SPS
Figure 4.11 - flow-chart of the elaboration process with samples designation as a function of the
raw alumina powders and forming methods. Designation for the sintering routes are also reported.
In the particular case of NS, it was performed up to 1500°C for 3 h (heating rate of 10°C/min to
1100°C and then 2°C/min up to 1500°C). Besides, HP samples were sintered in pellets of 20 or 40
mm, by heating up to 1450°C for 1 h (heating rate of 10°C/min up to 1100°C and 2°C/min up to
1450°C with a holding time of 1 h) with an applied pressure of 80 MPa.
In parallel, SPS samples were sintered by SPS heating up to 1350°C for Y-TM and 1450°C for YCR under an applied pressure of 75 MPa. Heating rates and soaking time in both cases were
154°C/min and 3 min, respectively. This process was followed by a second dwell time at 1134°C
(cooling rate of 617°C/min) for 5 min.
4.3.1. Sintering behaviour followed by dilatometric analysis
Samples Y-TM and Y-CR were uniaxially pressed in bars at 350 MPa and subsequently submitted
to dilatometric analysis, performed up to 1500°C for 3 h (heating rate of 10°C/min to 1100°C and
then 2°C/min up to 1500°C). The green bodies reached a density of 2.29 g/cm3 for the pressed
Y-TM and 2.25 g/cm3 for the pressed Y-CR, respectively. The green density was calculated from
weight and geometrical measurements.
158
In Figures 4.12 the comparison of dilatometric curves of Y-TM and Y-CR are presented.
Derivative signal
A
1335°C
LL
0.05
0
200
400
600
800
1000
1200
1400
1600
Temperature [°C]
Derivative signal
B
1465°C
LL
0.05
0
200
400
600
800
1000
1200
1400
1600
Temperature [°C]
Figures 4.12- Dilatometric (solid line) and derivative (dashed line) curves: (A) Y-TM-P and
(B) Y-CR-P.
For the samples linear shrinkages of 16.9% and 16.4% were recorded for Y-TM and Y-CR.
Derivatives curves shows the maximum sintering rate at 1335°C and 1465°C for Y-TM and Y-CR,
respectively. After sintering, Y-TM reached almost theoretical density (99.7% TD); while a poor
value was still achieved by Y-CR (94.4% TD).
4.4. Microstructural Characterization
Sintered samples were submitted to microstructural characterization, by using SEM (Hitachi
S2300) and ESEM (FEI XL30 ESEM FEG) microscopy. Microstructural observation was performed
on polished surfaces.
In order to compare the microstructural features of the composite materials to those of pure
alumina samples, TM and CR slip cast bodies were also prepared and pressureless sintered. Their
159
fired microstructures are presented in Figure 4.13: as shown, the two materials presented
important differences regarding the grain size. Sample TM, as expected, shows a dense and
homogeneous microstructure -with some intergranular porosity- composed by a homogeneous
microstructure with and average size of 1.43 μm.
On the other hand, CR is characterized by a equiaxed microstructure with a average grain size of
3.05 μm and some intergranular porosity, as well as, intra-granular porosity. This fact was
expected as the final density was lower (97.4 %) compared with TM (99.4 %).
b)
Figures 4.13- Micrographs at different amplifications of polished surfaces: (a) TM-SC and
(b) CR-SC sintered by pressureless sintering at 1500°C for 3 h.
The microstructural observation performed on Y-TM-P and Y-TM-SC samples (Figures 4.14)
revealed that the matrix was composed by micronic alumina grains of about ≈0.8 μm, having both
equiaxed and elongated shape. The second phase was mainly located at grain boundaries or
triples joints with few intra-granular particles distributed throughout the alumina matrix. These
composites exhibited a YAG grain size of about 350 nm and 450 nm for Y-TM-SC and Y-TM-P,
respectively.
As it is observed in Figures 4.14, the second phase (YAG) is the responsible of avoiding the
significant matrix grain growth by the grain boundary pinning effect (see section 3.4.1 –
Chapter III).
a)
b)
Figures 4.14- ESEM micrographs of polished surfaces: (a) Y-TM-SC (SE-SEM image) and (b) YTM-P (BSE-ESEM images) pressureless sintered at 1500°C for 3 h.
160
In the following Figures 4.15 are presented the microstructures corresponding to Y-CR-SC and
Y-CR-P. As in the previous case, the microstructure of naturally sintered Y-CR samples are formed
by equiaxial alumina grains with an average grain size of 0.98 μm and YAG particles within the αalumina grains.
In these particular cases, YAG mean size were about 650 nm and 400 nm in case of the slip
casted and the pressed samples, respectively. The second phase occupy – as in the case of Y-TM
– the grain boundaries or triple points positions. The analysis on polished surfaces revealed the
difficulty to corroborate the existence of intra-granular grains. As a comparison, between forming
methods, it was only found a slight variation regarding the YAG average grain size between the
samples.
a)
b)
Figures 4.15- ESEM micrographs of polished surfaces: (a) Y-CR-SC and (b) Y-CR-P pressureless
sintered at 1500°C for 3 h (GSE-ESEM images).
Samples Y-CR (Figure 4.15) show higher residual porosity compared with the Y-TM samples.
Specially, in sample Y-CR-P it was observed an important inter-granular porosity, as it is shown in
Figure 4.15 b.
Samples Y-TM and Y-CR pressed/slip-casted were sintered by non-conventional techniques,
namely hot-pressing (HP) and Spark Plasma Sintering (SPS) as mentioned before. By using these
techniques it was possible to decrease the sintering temperature in the range 1350-1450°C
yielding almost theoretical densities.
a)
b)
Figures 4.16- ESEM micrographs of polished surfaces: (a) Y-TM sintered by HP and (b) Y-TM
sintered by SPS (GSE-ESEM images).
161
Hot pressing was carried out at the same temperature 1450°C for both Y-TM and Y-CR. However,
SPS was performed at different temperatures: 1350°C for Y-TM and 1450°C Y-CR.
No significant differences were found between the forming methods in terms of final density and
microstructural features as seen in Y-TM NS samples. Only in the slip-casted Y-TM HP a slight
improvement in final density of ≈1% was found, with respect to the pressed sample.
Figures 4.16 show the micrographs corresponding to Y-TM-P sintered by HP and SPS.
Microstructures were finer in both cases compared with pressureless sintered samples with low
inter-granular porosity. In this particular case, SPS leads to achieve almost theoretical density
(99.4 +/- 0.1%) with a lower sintering temperature and a shorter soaking time. The measured
average grain sizes were 650 nm and 430 nm for the alumina and YAG in HP. Whereas, in case of
SPS the average grain sizes were 510 nm and 265 nm for the alumina and YAG, respectively.
It was difficult to determine accurately by the microstructural observation the position of the YAG
particles. Although, YAG particles seem to be located at the alumina matrix, mostly at
inter-granular positions (Figure 4.16).
a)
b)
Figures 4.17- ESEM micrographs of polished surfaces: (a) Y-CR sintered by HP and (b) Y-CR
sintered by SPS (GSE-ESEM images).
Y-CR samples sintered by HP (Figure 4.17 a) show a matrix formed by alumina grains with an
average grain size of 0.8 μm and a better distribution of the YAG phase – if compared with
Y-CR SPS (Figure 4.17 b) - of about 610 μm located at inter-granular positions.
As a comparison, Y-CR sintered by SPS (Figure 4.17 b) exhibits a less fine alumina since the
average grain size 1.13 μm. YAG grains present an average grain size of 570 nm. As in the
previous case, the second phase was located at inter-granular position.
The higher matrix grain size in Y-CR samples sintered by SPS and HP –as expected- is ascribable
to the higher mean particle size of CR compared with TM powder. Moreover, YAG average grain
size also was slightly higher in Y-CR samples specially among SPS samples.
The obtained Y-CR samples have a microstructure comparable to the nanocomposites produced
by Gao et al.10. The researchers produced 90 vol.% alumina-10 vol.% YAG sintered by SPS at
1300°C in which YAG phase (500 nm) was homogeneously distributed in the micronic alumina
matrix.
162
4.5. Mechanical characterization
4.5.1. Mechanical properties at room temperature
Fired samples were polished up to 0.1 μm–finish. Vickers hardness measurements (HV) was
determined by Vickers Testwell FV-700 with a load of 98.1 N (10 kg) held for 10 sec. Hardness
was calculated over an average of five measurements. Fired densities were determined by the
Archimedes’s method. As theoretical density it was used a value of 3.96 g/cm3 for α-alumina and
4.55 g/cm3 for YAG, resulting a theoretical density for the composite alumina-5 vol. % YAG
calculated by the rule of mixtures equal to 3.99 g/cm3.
In Table I are exposed the values of hardness and toughness of all the fired specimens.
Table I: Fired density (%TD) , Vickers Hardness (GPa) and K IC (MPam1/2) values for slip cast
(SC) and pressed (P) Y-TM bodies densified by pressureless sintering (NS), hot-pressing (HP) and
spark plasma sintering (SPS)
Y-TM
Sintering
Sample
route
Fired
Density
(%TD)
Hardness
(GPa)
K IC
(MPa.m1/2)
NS-1500
SC
99.7
18.8 +/- 1.13
4.42
NS-1500
P
98.2
16.3 +/- 1.84
5.11
HP-1450
SC
99.8
19.8 +/- 2.94
6.95
HP-1450
P
98.5
18.0 +/- 0.88
6.21
SPS-1350
SC
99.5
19.1 +/- 0.52
5.57
SPS-1350
P
99.4
19.9 +/- 0.5
5.82
Fired
Density
(%TD)
Hardness
(GPa)
K IC
(MPa.m1/2)
Y-CR
Sintering
Sample
route
NS-1500
SC
94.4
18.7 +/- 0.49
4.7
NS-1500
P
95.6
19.1 +/- 1.73
7.12
HP-1450
SC
99.1
21.4 +/- 1.71
6.03
HP-1450
P
98.9
19.7 +/- 1.10
6.37
SPS-1450
SC
99.7
18.8 +/- 0.81
7.12
SPS-1450
P
99.7
19 +/- 0.64
7.32
As it is shown in Table I, hardness values vary depending the starting material, the forming method
and the consolidation method.
The Vickers hardness values measured in Y-TM samples range from 16.3 to 19.9 GPa. These
values are a consequence of the combination of the effect of average grain size and the final
163
densities on the hardness. The influence of the grain size on the hardness has been discussed in
the Chapter III – Section 3.4.211,12.
For instance, naturally sintered Y-TM samples were strongly influenced by the final density as no
significant differences were found concerning the grain sizes. The same phenomenon was
observed in Y-TM HP. Samples Y-TM sintered by SPS presented similar hardness values as
microstructure and density were similar in both samples.
In Y-CR samples were measured hardness values in an interval of 18.7-21.4 GPa. Similar values
were measured in NS samples taking in consideration the standard deviation (+/- 1.73 GPa) in the
case of Y-CR-P NS-1500. In Y-CR HP samples were measured the highest hardness value since
they present the smallest grain size (≈0.8 μm) among all the samples. In particular, in HP slipcasted sample it was measured the highest value 21.4 GPa since its density was slightly higher if
compared with the pressed sample.
The lowest value of Y-TM-P NS sample is ascribable to some defects produced during the forming
process not visible in the previous microstructural observation. Finally, Y-CR samples sintered by
SPS presented as in the previous case similar characteristics with respect to density and final
microstructure.
The values are satisfactory if compared with the values published by Wang et al.13, who produce
high density alumina-25 vol.% YAG composite, made by sub-micronic alumina matrix with a
homogeneous distribution of fine YAG grains (100-600 nm). For this study the authors obtained a
hardness of 16.14 GPa.
Young modulus was determined by the impulse excitation technique. The impulse excitation
technique (Grindosonic method) allows to assess the elastic modulus of a bar-shaped sample at
ambient temperature, necessary to obtain the fracture toughness values. For the different modes
of resonance, the specimen is supported as it is illustrated in the Figure 4.18. The vibration of the
bar is recorded by a microphone or a transducer. The fundamental frequency of the bar is
determinate by a frequency analyzer. With the aim to corroborate the Young modulus, Grindosonic
method was conducted based on the rules C 1259-01 from ASTM15, employing bar-shape
samples of 40 mm diameter pellets.
Figure 4.18- Rectangular specimen for Out-of-Plane Flexure7.
 mf 2   L3 
E  0.9465  f   3  .T1
 b  t 


where
T1  1  6.585 1  0.0752   0.8109 2   t / L   0.868  t / L  
2
(1)
4
4


8.340 1  0.2023  2.173 2   t / L 


2
1.000  6.338 1  0.1408  1.536  2   t / L  


where:
E= Young’s modulus, Pa ;
164
M=mass of the bar, g ;
b= width of the bar, mm ;
L= length of the bar, mm ;
t= thickness of the bar, mm;
f f = fundamental resonant frequency of bar in flexure, Hz;
T 1 = correction factor for fundamental flexural mode to account for finite thickness of bar;
μ= Poisson’s ration.
Young modulus evaluated in both samples yielded a value of 419.2 GPa for the slip-casted Y-TM
and 415.5 GPa for the slip-casted Y-CR. As a comparison, the elastic modulus, given by the rule of
mixtures was also calculated being 409.2 GPa, in good agreement with the experimental value.
Fracture toughness was estimated using the indentation method using the Anstis’ formula14, as
shown below.
K IC  A.  E / H  .  P / c 2 / 3 
1/ 2
(2)
where
c is the length of the crack from the center of the impression;
A is a geometric constant equal to 0.016;
P the change expressed in Newton;
H the hardness value;
E the Young’s modulus.
In the following Figures 4.19 , the Y-TM-NS fracture surfaces are presented. As it is shown Y-TM-P
fracture surface exhibits inter-granular fracture, whereas in the case the of Y-TM-SC is mostly
inter-granular with some trans-granular areas (indicated with arrows).
a)
b)
Figures 4.19 - ESEM micrographs on fracture surface of: (a) Y-TM-SC and (b) Y-TM-P by
pressureless sintering at 1500°C for 3h.
As in the previous case, Figures 4.20 present the microstructures of Y-CR NS samples. Y-CR-SC
fracture surface is mainly inter-granular. Fracture surface of sample Y-CR-P shows a mixture of
trans-granular (indicated with arrows) and inter-granular modes.
165
a)
b)
Figures 4.20 - ESEM micrographs on fracture surface of: (a) Y-CR-SC (b) Y-CR-P by pressureless
sintering at 1500°C for 3 h.
ESEM observation on Y-TM samples sintered by non-conventional routes revealed that fracture
was -as in the previous case,- a mixture of trans-granular (indicated with arrows) and inter-granular
as shown in Figures 4.21 .
a)
b)
Figures 4.21- ESEM micrographs on fracture surface of: (a) Y-TM-P by HP 1450°C, (b) Y-TM-P by
SPS 1350°C (GSE images).
Finally, fracture surfaces of samples Y-CR sintered by HP and SPS are shown in Figures 4.22.
The fracture surfaces reveal that both samples prevalently presented trans-granular fracture
(indicated with arrows) with some inter-granular zones.
166
a)
b)
Figures 4.22- ESEM micrographs on fracture surface of: (a) Y-CR-P by HP 1450°C, (b) Y-CR-P by
SPS 1450°C (GSE images).
Y-TM samples densified by natural sintering presented different fracture modes. As seen in
Y-TM-P (Figure 4.19 (b)) some transgranular areas were detected. This fact may be the
responsible of increasing the fracture toughness from 4.42 MPa.m1/2 of the SC sample to
5.11 MPa.m1/2 in case of P material.
Nanocomposites, as reported in literature, show a clear change in the fracture mode from
inter-granular to trans-granular. In the case of a monolithic material, since the fracture resistance of
the grain boundary is lower compared with the one at the grain lattice, the crack will propagate
inter-granularly along the grain boundaries.
However, the toughening effect of the nanoparticles at the grain boundaries may steer the tip crack
into the grain. As a result, the fracture becomes then trans-granular increasing the toughness of
the composite. Once the tip crack is inside the grain, the propagation can be pinned by
intragranular particles, producing the zig-zag path (see section 3.4.6 – Crack deflection).
The same phenomenon was observed in naturally sintered Y-CR samples in which an important
increase of toughness occurs reaching 7.12 MPa.m1/2 in the case of sample Y-CR-P.
Y-TM samples sintered by non-conventional sintering HP and SPS exhibit similar fractures path.
The toughness differences observed among the samples are reasonably imputable to the
difference in matrix average grain size: 0.65 μm versus 0.51 μm in the case of HP and SPS,
respectively.
It is well known that toughness increases linearly in function of d1/2 as reported by Swain17. This
fact may explain the disparities in non-conventional sintered samples regarding the toughness
values. In order to clarify this fact, in Figure 4.23 it is shown the influence of the grain size on
toughness of monolithic alumina and alumina nanocomposites reinforced with SiC whiskers.
167
Figure 4.23 – Influence of the grain size on the fracture toughness of an alumina and alumina
nanocomposites reinforced with 10 and 20 % of SiC whiskers17.
The same phenomenon was observed in Y-CR non-conventional sintered samples as fracture
modes in HP and SPS were similar. In this particular case, Y-CR SPS exhibited a higher
toughness value 7.32 MPa.m1/2 compared with 6.37 MPa.m1/2 (HP) again attributable to the higher
matrix average grain size 1.13 μm (SPS) against 0.8 μm (HP). This statement is based, as fracture
modes in both samples were similar.
As a comparison with literature, it was found a range of toughness between 3 to 5.8 MPa.m1/2 for
YAG reinforced alumina nanocomposites for 5-50 vol.% 5,13,18.
4.5.2. Mechanical properties at high temperature
Creep tests for alumina and doped aluminas were performed in a 4-point-bending fixture at the
temperature of 1200°C and an applied stress of 100 MPa. Samples were tested under an applied
stress of 100 MPa due to the limited number of samples. The heating rate of the furnace was
300°C/h in order to avoid significant thermal gradients.
The specimens were parallelepipeds with dimensions of about 3 x 4 x 32 mm3, in which the tensile
face of all the specimens was polished with diamond paste down to 3 μm and the edges were
chamfered (about 45°) in order to avoid the influence of microcracks during creep.
Natural sintered bar-shaped samples were produced by slip-casting, whereas less-conventional
sintered samples were produced by uniaxially pressing, as the forming method. In this particular
case it did not influence the final properties in terms of microstructure and final density.
A four-point bending fixture was used with an inner/outer span of 30 and 15 mm respectively. The
flexural stress on the tensile face of the specimen was calculated employing the following
expression:

3P  L  L' 
2bw2
(3)
where P is the applied load, L the outer span, L’ the inner span, b the sample width and w the
sample height.
To determine the creep strain, it was used the method described in literature19. This method is
based on the assumption that creep strain ε can be calculated from the deflection y at the center of
the specimen and the deflection, whereas there is no major cracking in the specimen and the
deflection y is small compared with the inner span L’.
168
  K (n). y
(4)
where
K ( n) 
2w  n  2 
 L  L '  L  L '  n  1
The constant K(n), in addition to its dependence on n, is also a function of L and L’. Hollemberg et
al.7 have shown that for (L/L’) values close to 2, K(n) are almost insensitive to n.
.
The steady-state creep rate  is defined by the following equation:
.
A
n
 Q 
exp  

d
 RT 
(5)
p
where
A: a material constant;
σ: the applied stress;
n: the stress exponent;
d: the grain size;
p: the grain size exponent;
Q: the activation energy for creep;
R: the gas constant;
T: the absolute temperature.
4.5.2.1.
Creep of TM and CR
The creep curves were obtained for TM and CR and they are shown in Figure 4.24 - sintered at
1500°C with a soaking time of 3 h – in order to compare the effect of the second phase on the
creep behaviour. Both curves exhibited two stages: transient and steady-state. The strain rate has
been determined by the evaluation of the slope of the creep curve on the steady-state stage.
0,9
0,8
0,7
Strain (%)
0,6
0,5
0,4
TM
CR
0,3
0,2
0,1
0,0
0
2
4
6
8
10
12
14
16
Time (h)
Figure 4.24- Creep curves of TM and CR at 100 MPa and 1200°C.
The instantaneous strain rate in function of the time is shown in the Figure 4.25. In these curves it
is possible to corroborate that an almost constant strain rate as observed in the second state. A
169
deformation (ε) under 1% was found in both samples being 0.79% in TM and 0.69% for CR,
respectively.
Strain Rate (1/s)
1E-6
1E-7
1E-8
TM
CR
1E-9
0
2
4
6
8
10
12
14
Time (h)
Figure 4.25- Strain rate of TM and CR at 100 MPa and 1200°C.
The strain rate calculated from the slope of the curve deformation vs. time has been 1.04.10-8 s-1 in
TM and 1.61.10-8 s-1 in CR, respectively. The discrepancy observed in terms of strain rate has
been imputed to the effect of the porosity on the creep behaviour as reported by Langdon21. The
author has analyzed the effect of the porosity on the creep rate in aluminas. The author proposed
the following theoretical formula to explain the effect of the porosity.


P
1

m
n 
.
Gb  b      1     1 P 
  A.D. .   .   .
n
kT  d   G 
1  P 2 / 3 
1 n
(6)
where A is a dimensionless constant, D is the coefficient of diffusion, G is the shear modulus, b is
the Burgers vector, k is the Boltzmann’s constant, T is the temperature in Kelvin, d is the grain
size, m is the grain size exponent, n is the stress exponent, β is a constant equal to -4 +/- 0.07 for
alumina when porosity varies between 0.3 and 16 % and P is the porosity expressed in %. By
applying the previous equation it is possible to plot a graph, in which it is calculated the creep rate
as a function of the porosity and the stress exponent, as it is shown in Figure 4.26.
Figure 4.26 – Creep rate as a function of porosity for n values from 1 to 521.
170
Considering that the differences regarding the porosity were ≈5% TD and n values reported in
literature varied from 1 to 1.5 (see Section 3.6.2 - Chapter III). In theory, the strain rate would
increase from 4 to 50 %.
The difference between the strain rates of TM and CR was 53.84 %. The empirical value is higher
than the expected. Theoretically, the increment caused by the porosity, as expected, should be
lower as CR presented a higher average grain size (3.05 μm in CR versus 1.43 μm in TM).
4.5.2.2.
Creep of Y-TM and Y-CR nanocomposites
As for TM and CR, the same procedure was extended to the composites carrying test until the
rupture -as for both aluminas-. In Figure 4.27 (a) it is shown the evolution of the strain for Y-TM as
a function of time for a given stress of 100 MPa at 1200°C. The test gave rise to the contained
deformation under creep of about ≈ 0.26% (in the case of Y-TM HP), which was significantly lower
if compared to the monolithic TM. In fact, especially in case of Y-TM SPS after 55 h, deformation is
reduced in ≈7 times in comparison with TM.
0,30
0,25
Strain (%)
0,20
0,15
0,10
0,05
Y-TM NS-1500
Y-TM HP-1450
Y-TM SPS-1350
0,00
0
5
10
15
20
25
30
35
40
45
50
55
60
Time (h)
a)
1E-6
Strain rate (1/s)
1E-7
1E-8
1E-9
1E-10
Y-TM NS-1500
Y-TM HP-1450
Y-TM SPS-1350
1E-11
0
b)
5
10
15
20
25
30
35
40
45
50
Time (h)
Figures 4.27- (a) Creep curves and (b) Strain Rate of Y-TM.
171
In Figure 4.27 (b) the strain rates calculated in the steady-state corresponding to Y-TM samples
are shown. The strain rates έ n were 1.8.10-7 s-1 for Y-TM NS, 4.2.10-10 s-1 for Y-TM HP and
1.35.10-9 s-1 for Y-TM SPS, respectively.
As it can be seen, in the Figure 4.27 (b) Y-TM sintered by HP has the lowest strain rate among the
samples of about έ=4.2.10-10 s-1. No significant differences were found to explain the disparity
among the results, since grain sizes and densities in Y-TM samples are comparable.
Not normalized strain rates were successfully reduced ≈25 and ≈8 times in samples: Y-TM HP and
Y-TM SPS in contrast to TM.
0,7
0,6
Strain (%)
0,5
0,4
0,3
0,2
0,1
Y-CR NS-1500
Y-CR HP-1450
0,0
Y-CR SPS-1450
0
5
10
15
20
25
30
35
40
45
50
55
Time (h)
a)
Strain rate (1/s)
1E-7
1E-8
Y-CR NS-1500
Y-CR HP-1450
Y-CR SPS-1450
1E-9
0
b)
5
10
15
20
25
30
35
40
45
50
55
Time (h)
Figures 4.28- (a) Creep curves and (b) Strain Rate of Y-CR samples.
172
Subsequently, the procedure was extended to Y-CR samples. In the case of Y-CR, the strain in all
the cases is substantially reduced ≈5 times up to 0.134% in the case of Y-CR HP in comparison
with CR sample. Only in sample Y-CR SPS, it was measured a little improvement of ε=0.63%.
As it is shown in Figure 4.28 b, an important strain rate reduction was found in the natural sintered
sample, namely Y-CR NS. This phenomenon is ascribable to the average grain size (0.98 μm) and
a good distribution of the second phase within the alumina grains.
Strain rates measured on the samples showed in Figure 4.28 (b) were 2.12.10-9 s-1 for Y-CR NS,
4.47.10-9 s-1 for Y-CR HP and 5.44.10-8 s-1 for Y-CR SPS, respectively. As in the previous case,
non-normalized strain rates denoted a reduction of ≈8 and ≈4 times in samples Y-CR NS and
Y-CR HP with respect to CR.
For the sake of clarity, all the values regarding the grain size and strain rates of the different
samples are contained in Tables II/III. In order to establish a comparison among the samples, the
normalized strain rate was calculated using the criterion taken from literature5,7,18.
The normalized strain rate έ n can be calculated by the following equation, taking into account the
differences in grain sizes. The formula used for calculation is shown below.
P
 1  .
n 
 . B
 d A / dB 
.
(7)
Where έ n is the measured creep rate for Y-TM and Y-CR, d A is the mean size of sample A and d b
is the mean grain size of sample B. Two comparisons have been done: the former, considering the
average matrix grain size of the pure alumina, labelled as d A (έ nA ) and the latter employing for the
calculation, the smallest matrix alumina grain size in the case of the composite (έ nY ). The second
criterion has been chosen with the aim of establishing the advantage of introducing a second
phase into the alumina matrix.
The grain size exponents p are 2 and 3 for lattice diffusion controlled creep and for grain boundary
diffusion controlled creep, as reported in literature. It was assumed that creep is controlled by
lattice diffusion controlled creep, for this reason creep rates were normalized using an exponent of
a grain size equal to 2 18,23-24.
In Table II it is shown the comparison in terms of normalized strain rates among the TM
specimens.
Table II- Strain rates of TM samples and normalized strain rates of the sintered samples at 1200°C
with an applied stress of 100 MPa.
Sample
TM NS-1500
Y-TM NS-1500
Y-TM HP-1450
Y-TM SPS-1350
Strain rate (s-1)
-8
1.04.10
1.8.10-7
4.2.10-10
1.35.10-9
Fired
Density
99.4
99.8
98.5
99.4
Grain Size
Al 2 O 3 (μm)
1.43
0.77
0.65
0.51
Grain Size
YAG (μm)
0.45
0.43
0.26
.
 nA
(s-1)
1.04.10-8
5.22.10-8
8.68.10-11
1.72.10-10
.
 nY
(s-1)
8.18.10-8
4.10.10-7
6.82.10-10
1.35.10-09
For a better understanding, values corresponding to the different strain rates were plotted in
Figure 4.29. Non important differences were found among the samples, since grain sizes are
comparable -as it was mentioned before-.
173
Strain Rate (1/s)
1E-7
1E-8
1E-9
TM NS-1500
Y-TM NS-1500
Y-TM HP-1450
Y-TM SPS-1350
Samples
a)
Strain Rate (1/s)
1E-8
1E-9
1E-10
TM NS-1500
Y-TM NS-1500
Y-TM HP-1450
Y-TM SPS-1350
Samples
b)
Strain rate (1/s)
1E-7
1E-8
1E-9
TM NS-1500
Y-TM NS-1500
Y-TM HP-1450
Y-TM SPS-1350
Samples
c)
Figures 4.29- (a) Creep rates of TM at 1200°C under a stress of 100 MPa, (b) normalized creep
rates with έ nA and (c) normalized creep rates with έ nY of different samples.
On one hand, taking the έ nA criterion for Y-TM samples, creep rate is increased by a factor of 5 in
Y-TM NS with respect to TM. This fact is difficult to explain, if the homogeneous microstructure and
the very high final density is considered. Probably, it could be imputed to the presence of nondetected defects during the samples elaboration by slip-casting. Finally, in samples sintered by
174
non-conventional sintering routes it was observed a clear improvement, since they present the
lowest creep rate, reduced by a factor of 119 and 60 in Y-TM HP and Y-TM SPS, respectively.
On the other hand, employing the second criterion έ nY –i.e. using the matrix grain size of Y-TM
SPS -, the creep rate is doubled compared with HP in comparison with TM. Finally in comparison
with Y-TM NS, the creep strain is increased by a factor of 303 and in the case of TM is lowered 60
times.
Table III-Strain rates of CR samples and normalized strain rates of the sintered samples at 1200°C
with an applied stress of 100 MPa.
Sample
Strain rate (s-1)
CR NS-1500
Y-CR NS-1500
Y-CR HP-1450
Y-CR SPS-1450
1.61.10-8
2.12.10-9
4.47.10-9
5.44.10-8
Fired
Density
97.4
95.6
98.9
99.7
Grain Size
Al 2 O 3 (μm)
3.05
0.98
0.8
1.13
Grain Size
YAG (μm)
0.65
0.61
0.57
.
 nA
(s-1)
1.61.10-8
2.19.10-10
3.08.10-10
7.47.10-9
.
 nY
(s-1)
2.34.10-7
3.18.10-9
4.47.10-9
1.09.10-7
Strain rate (1/s)
1E-6
1E-7
1E-8
CR NS-1500
Y-CR HP-1450
Y-CR NS-1500
Y-CR SPS-1450
Samples
a)
Strain rate (1/s)
1E-8
1E-9
1E-10
CR NS-1500
b)
Y-CR NS-1500
Y-CR HP-1450
Y-CR SPS-1450
Samples
175
Strain rate (1/s)
1E-7
1E-8
1E-9
CR NS-1500
c)
Y-CR HP-1450
Y-CR NS-1500
Y-CR SPS-1450
Samples
Figures 4.30- (a) Creep rates of CR at 1200°C under a stress of 100 MPa, (b) normalized creep
.
.
rates with  nA and (c) normalized creep rates with  nY of different samples.
The έ nA criterion in CR samples, showed that creep rate is reduced by a factor of 74 and 53 in
Y-CR NS and Y-CR HP in comparison with CR. However, in the Y-CR SPS it is only lowered by a
factor of 2. The second criterion έ nY (employing Y-CR HP for the calculation) resulted a creep
strain increased by a factor of 1.4 in case of the Y-CR NS sample. Although, creep strain is
reduced by a factor of 52 and 24 with respect to and Y-CR SPS, respectively.
Sample Y-CR NS present a higher matrix average grain size compared with Y-TM which is the
responsible for conferring a better creep behaviour (see equation 5). In this context, the second
phase distribution, as well as the matrix grain size, seem to be the key for improving the creep
resistance in composites.
These results are particularly in good agreement with the work published by Yoshida et al.6, who
doped high-purity alumina powder (TM-DAR, Taimei) with 0.045 mol% Y 2 O 3 . The authors
attributed the reduction in creep rate, 200 times lower than the un-doped alumina for an applied
stress of 50 MPa, to the segregation of Y3+ cations to alumina grain boundaries. This segregation
is the responsible of restricting the grain boundary diffusion of Al3+ ions.
In comparison with the literature, Schehl et al.18 found a non-normalized strain rate of 2.0.10-9 s-1
for alumina-YAG composites in the same experimental condition, giving importance to the values
obtained.
A comparison among samples using the criterion of normalization with έ nA is shown in Figure 4.31.
Monolithic aluminas showed a similar creep behaviour, taking into account the differences in terms
of grain sizes and porosity in TM and CR. The importance of a higher matrix grain size has been
evident in Y-CR NS sample, as it exhibited a better creep behaviour compared with Y-TM NS.
Non-conventional sintered samples exhibited different creep behaviour. This fact is imputed to
some heterogeneous distribution of the second phase in some areas of Y-CR samples.
176
Strain Rate (1/s)
1E-8
Y-TM
Y-CR
1E-9
1E-10
NS
Y HP
Y NS
Y SPS
Figure 4.31- Comparison among samples.
For the sake of clarity, in Figures 4.31/4.32 are shown the comparison between HP and SPS
samples.
From the analysis of the Figure 4.31 (at different magnification), it is possible to observe a slightly
different distribution of the second phase in HP samples. Samples Y-CR sintered by HP exhibited
an irregular YAG distribution compared with SPS.
a)
b)
Figures 4.31- ESEM micrographs on polished surfaces of: (a) Y-TM HP-1450 and
(b) of Y-CR HP-1450 (GSE-ESEM images)
This phenomenon is more evident in samples sintered by SPS. As it is shown in Figure 4.32 (b), in
this particular Y-CR SPS sample’s region, only few points of YAG phase were found (indicated with
arrows). This fact may explain the differences in terms of έ nA normalized strain rates (being ≈44
times higher in sample Y-CR- see Figure 4.30), in spite of the higher matrix grain size of Y-CR
sample.
177
a)
b)
Figures 4.32- ESEM micrographs on polished surfaces of: (a) Y-TM SPS-1350 and
(b) of Y-CR SPS-1450 (GSE-ESEM images)
4.5.3. Activation energy
The activation energy of the nanocomposites was determined in the temperature range of
1200-1400°C with an applied stress of 100 MPa for the most promising material in terms of creep
resistance, operational time, second phase distribution and elaboration process, the
Y-CR NS-1500.
The activation energy has been determined by the slope of ln(ε)=f(1/T) obtained at constant
applied stress of 100 MPa. In the following Figure 4.33 it is shown ln έ-1/T graph.
-6
Strain Rate (1/s)
1x10
Q=644 kJ/mol
-7
1x10
-8
1x10
-9
1x10
6,0
6,2
6,4
6,6
6,8
-4
1/RT x 10 (1/K)
Figure 4.33 – Steady-state creep rate vs. the reciprocal of the absolute temperature of the
Y-CR NS-1500 composite.
178
The activation energy for the Y-CR NS-1500 was 644.4 kJ/mol in good agreement with the results
found literature, i.e. it was measured for alumina 630 kJ/mol23. A comparable result was found in
literature, French et al.5 obtained 698 kJ/mol for a 1000 ppm yttrium-doped Al 2 O 3 with an applied
stress of 75 MPa in the range of 1125-1350°C.
Yoshida et al.6 obtained 830 kJ/mol for Al 2 O 3 doped with 0.045% Y 2 O 3 for an applied stress of
100 MPa in the range of 1250-1350°C. Similarly, Torrecillas et al.7 produced alumina-YAG
nanocomposites by the colloidal processing route (the process is shown in detail in Chapter III)
found an activation energy of 595 kJ/mol for an applied stress of 100 MPa.
It is known that activation energy for Al3+ grain boundary diffusion is 420 kJ/mol, whereas for Al3+
lattice diffusion the activation energy is 578 kJ/mol, measured in the 1200-1400°C range19-21.
In creep of monolithic aluminas a combination of both mechanisms is supposed to coexist, in which
lattice diffusion is supposed to be the dominant (Nabarro-Herring)19-21. In composites, as it was
reviewed by some researchers, YAG particles formed at grain boundaries may inhibit the grain
boundary sliding7. This phenomenon is caused by a combination of high activation energy
necessary to deform these nanoparticles and the absence of defects in YAG nanoparticles raise
the total activation energy closer to lattice diffusion of Al3+ in YAG nanoparticles7.
4.5.4. Microstructural observation after creep tests
In Figures 4.34 it is shown the fracture surface after creep measurements. The micrograph gives
rise to a non-significant grain growth occurred during creep and mainly inter-granular fracture
occurred on samples. The second conclusion was that no cavitation was observed in the samples.
a)
b)
Figures 4.34- SEM micrographs of fracture surfaces: (a) Y-TM-SC and (b) Y-CR-SC
(GSE-ESEM images).
179
Conclusion of the second part and perspectives
The developed Al 2 O 3 -YAG composites showed good mechanical properties (hardness, toughness,
and Young modulus) at room temperature. In both Y-CR and Y-TM, samples sintered by HP and
SPS presented the highest hardness values (≈20 GPa) ascribable to the higher density measured.
Toughness values were strongly influenced by the matrix average grain size, since higher values
were found in samples having higher grain size reaching ≈7 MPa.m1/2. This phenomenon was
corroborated specially in Y-CR sintered by NS and by SPS which have higher average grain size
0.98 and 1.13 μm, respectively.
The creep behaviour in monolithic TM and CR was ruled by the differences in terms of porosity, as
it was reviewed in the literature. The influence of grain size was evidenced in the naturally sintered
materials, since differences were measured involving the porosity (higher in CR).
Samples sintered by non-conventional routes presented differences caused by an inhomogeneous
distribution of the second phase, as revealed by SEM observation. This phenomenon was
confirmed by Archimedes’ measurements, as not substantial differences were found concerning
the fired densities.
These results in non-conventional sintered samples showed that it is necessary to optimize the
second phase distribution, as some discrepancies exist in terms of strain rates. This fact could
permit to clarify this phenomenon.
However, good results were obtained in the composites with respect to creep behaviour which
proved that YAG formation along/inside grain boundaries improved the creep resistance, as it is
believed that Y3+ reduces the grain boundary diffusivity.
Further studies have to be performed on samples with the aim of completing the mechanical
characterization at high temperature. It is necessary to obtain stress exponent values by extending
the same approach to more stresses and temperatures and consequently clarify the creep
mechanism for a correct normalization.
180
References
181
References
Chapter I
[1]
MORGAN, P.E.D., Superplasticity in Ceramics, in: Ultrafine-Grained Ceramics. J.J. Burke, L.R.
Norman, and V. Weiss, editors., Syracuse University Press, Syracuse, 1968.
[2]
MAYO, M., Processing of Nanocrystalline Ceramics from Ultrafine Particles. International
Material Reviews, 1996, vol. 41, pp. 85-115.
[3]
Azo Nanotechnology [on-line]. Available at: http://www.azo.com/ (consulted 20.11.2009)
[4]
CHOE, G.M., KURIHARA, L.K., Chemical Synthesis of Processing of Nanostructured Powders
and Films. In: KOCH, C.C., Nanostructured Materials: processing, Properties and Applications.
William Andrew Publishing, New York, 2002.
[5]
GLEITER, H., Nanostructured Materials: State of Art and Perspectives. NanoStructured
Materials, 1995, vol. 6, pp. 3-14.
[6]
GROZA, J.R., Nanosintering. Nanostructured Materials, 1999, Vol. 12, pp. 987-992.
[7]
LANGE, F.F. Powder processing science and technology for increased reliability. Journal of the
American Ceramic Society, 1989, vol. 72, pp. 3-15.
[8]
NIIHARA, L., New Design Concept of Structural “Ceramic Nanocomposites”. The Centennial
Memorial Issue of the Ceramic Society of Japan, 1991, vol. 99, n°10, pp. 974-982.
[9]
BOWEN, P., CARRY, C., LUXEMBOURG, D., HOFMANN, H., Colloidal processing and
sintering of nanosized transition aluminas. Powder Technology, 2005, vol. 157, pp. 100-107.
[10] WU, S.J., DE JONGHE, L.C., Sintering of Nanophase γ-Al2O3 powder. Journal of the
[11] ZHAN, G.D, KUNTZ, J., WAN, J., GARAY, J., MUKHERJEE, A.K., A novel processing to
developed a Dense Nanocrystalline Alumina Matrix (<100 nm) Nanocomposite Material.
Journal of the American Ceramic Society, 2002, vol. 86, vol. 1, pp. 200-202.
[12] MEYERS, M.A., MISHRA, A., BENSON, D.J., Mechanical properties of nanocrystalline
materials. Progress in Materials Science, 2006, vol. 51, pp. 427-556.
[13] CHEN, D.J., MAYO, M.J., Densification and grain growth of ultrafine 3mol% Y 2 O 3 -ZrO 2
ceramics. Nanostructured materials, 1993, vol. 2, n°5, pp. 469-478.
[14] LEWIS, J.A., Colloidal processing of ceramics. Journal of American Ceramic Society, 2000,
vol. 83, pp. 2341-2359.
[15] RAHAMAN, M.N., Ceramic Processing and Sintering. Marcel Dekker, New York, 2003.
[16] HAMACKER, H.C., The London-van der Waals attraction between spherical particles. Physica,
1937, vol. 4, pp. 1058-1072.
[17] NAPPER, D.H., Polymeric Stabilization of Colloidal Dispersions. Academic Pres, New Yok,
1983.
182
References
[18] CHEN, W., PECHENIK, A., DAPKUNAS, S.J., PIERMARINI, G.J., MALGHAN, S.G., Novel
Equipment for the Study of Compactation of Fine Powders. Journal of the American Ceramic
Society, 1995, Vol. 77, pp. 1005-1010.
[19] FERKEL, H, HELLMIG, R.J., Effect of nanopowders deagglomeration on densities of
nanocrystalline ceramic green bodies and their sintering behavior. Nanostructured Materials,
1999, vol. 11, n° 5, pp. 617-622.
[20] HERRING, C., Effect of Change of scale on Sintering Phenomena. Journal of Applied Physics,
1950, vol. 21, pp. 301-303.
[21] CHEN, W., MALGHAN, S.G., Investigations of Compactation Equations for Powders. Powder
technology. Powder technology, 1994, vol. 81, pp. 75-81.
[22] RHODES, W.H., Agglomerate and Particle Size Effects on Sintering Yttria-Stabilized Zirconia.
Journal American Ceramic Society, 1981, vol. 64, pp. 19-22.
[23] GALLAS, M.R., ROSA, A.R., COSTA, T.H., DA JORNADA, J.A.H., High Pressure Compaction
of Nanosized Ceramic Powder. Journal Material Research, 1997, vol. 12, pp. 764-768.
[24] SAGEL-RANSIJN, C.D., WINNBUST, A.J.A., KERKWIJK, B., BURGGRAF, A.J., VEWWEIJ,
H., Production of Defect-Poor Nanostructured Ceramics of Yttria-Zirconia. Journal of the
European ceramic Society, 1997, vol. 17, pp. 831-841.
[25] VAN DE GRAAF, M.A.C.G., TERMAAT, J.H.H., BURGGRAAF, A.J., Microstructure and
Sintering Kinetics of Highly Reactive ZrO 2 -Y 2 O 3 Ceramics. Journal of Materials Science, 1985,
vol. 20, pp. 1407-1418.
[26] BOSCHI, A.O., GILBERT, E., in: Advanced Ceramic Processing and Technology vol.I, ed.
J. G.P. Binner, 71-93, New Jersey, 1990.
[27] RHODES, W.H., Agglomerate and Particle Size Effects on Sintering Yttria-Stabilized Zirconia.
Journal of the American Ceramic Society, 1981, vol. 64, pp. 19-22.
[28] CHEN, P.L., CHEN, I.W., Sintering of fine oxide powders: II Sintering Mechanisms. Journal of
the American Ceramic Society, 1997, vol. 80, pp. 637-645.
[29] COBLE, R.L., Sintering Crystalline Solids: I. Intermediate and Final State Diffusion Models.
Journal of Applied Physics, 1961, Vol. 32, n° 5, pp. 787-792.
[30] BAUMGARTNER, C.E., Fast Firing and Conventional Sintering of Lead Zirconate Titanate
Ceramics. Journal of the American Ceramic Society, vol, 71, n°7, pp. 350-353.
[31] WANG, J., RAJ, R., Estimate of the Activation Energies for Boundary Diffusion from
rate-controlled sintering of pure alumina, and alumina doped with Zirconia and Titania. Journal
of the American Ceramic Society, 1990, vol. 73, n°5, pp. 1172-1175.
[32] ANDRIEVSKI, R.A., Compactation and Sintering of Ultrafine Powders. International Powder
Metallurgy, 1994, vol. 30, pp. 59-66
[33] MESSING, G.L., KUMAGAI, M., Low-Temperature Sintering of α-Alumina Seeded Bohemite
Gels. American Ceramic Bulletin, 1994, vol. 73, pp. 88-91.
183
References
[34] YEH, T.S., SACKS, M.D., Low-Temperature Sintering of Aluminium Oxide. Journal of the
American Ceramic Society, 1998, vol. 71, n°10, pp. 841-844.
[35] KIM, B.C., LEE, J.H., KIM, J.J., IKEGAMI, T., Rapid rate of nanocrystalline indium tin oxide
ceramics. Materials Letters, 2002, vol. 52, pp. 114-119.
[36] CHU, M.Y., RAHAMAN, M.N., DE JONGHE, L.C., Effect of Heating Rate on Sintering and
Coarsening. Journal of the American Ceramic Society, 1991, vol. 74, n°6, pp. 1217-1725.
[37] KINGERY, W.D., BOWEN, H.K., UHLMANN, D.R., Introduction to Ceramics, 2nd ed., Wiley,
New York, 1976.
[38] BROOK, R.J., Pore-Grain Boundary Interactions and Grain Growth. Journal of the American
Ceramic Society, 1969, vol. 52, pp. 56-57.
[39] AVERBACK, R.S., HOFLER, H.J., TAO, R., Processing of Nano-grained Materials. Materials
Science Engineering A, 1993, vol. 166, pp. 169-177.
[40] CAMERON, C.P., RAJ, R., Grain-Growth Transition during Sintering of Colloidally prepared
Alumina Powder Compacts. Journal of the American Ceramic Society, vol. 71, n° 12, 1988.
[41] HAHN, H., AVERBACK, R.S., HOFLER, H.J., LOGAS, J., Sintering and Deformation of
Nanocrystalline Ceramics, in: Cluster and Cluster-Assembled Materials. Materials Research
Society, Pittsburgh,1991.
[42] MAYO, M., HAGUE, D.C., CHEN, D.J., Processing nanocrystalline Ceramics for Applications
in: Superplasticity. Materials Science Engineering A, 1993, vol. 166, pp. 145-159.
[43] GROZA, J.R., DOWDING, R.J., Nanoparticles materials densification. NanoStructured
Materials, 1996, vol. 7, n° 7, pp. 749-768.
[44] AHN, J.P., HUH, M.Y., PARK, J.K., Effect of Green density on Subsequent densification and
Grain Growth of Nanophase SnO 2 Powder During Sintering. Nanostructured Materials, 1997,
vol. 8, pp. 637-643.
[45] HARMER, M.P., CHAN, H.M., MILLER, G.A., Unique opportunities for microstructural
engineering with duplex and laminar ceramic composites. Journal of the American Ceramic
Society, 1992, vol. 75, n° 7, pp. 1715-1728.
[46] LIU, Y., PATTERSON, B.R., Grain Growth Inhibition by Porosity. Acta Metallurgica et
Materialia, 1993, vol. 41, pp. 2651-2656.
[47] CHEN, P.L., CHEN, I.W., Sintering of fine oxide powders: I Microstructural evolution. Journal of
the American Ceramic Society, 1996, vol. 79, pp. 3129-3141.
[48] ZENG, W., GAO, L., GUI, L., GUO, J., Sintering kinetics of α-Al 2 O 3 powder. Ceramics
International, 1999, vol. 25, pp. 723-726.
[49] RODRIGUEZ, J.A., FERNANDEZ-GARCIA, M., Synthesis, Properties and Applications of
Oxide Materials. Ed. John Wiley & Sons, New Jersey, 2007.
[50] YAN, M.F., RHODES, W.W., Low Temperature Sintering of TiO 2 . Material Science and
Engineering, 1983, vol. 61, pp. 59-66.
184
References
[51] NORDAHL, C.S., MESSING, G.L., Sintering of α-Al 2 O 3 -seeded nanocrystalline γ-Al 2 O 3 .
Journal of the European ceramic Society, 2002, vol. 22, pp. 415-422.
[52] WANG, Y., SURYANARAYANA, C., AN, L., Phase transformation in nanometer-sized
γ-Alumina by Mechanical Milling. Journal of the American Ceramic Society, 2005, vol. 88, n°3,
pp. 780-783.
[53] PALKAR, V.R., THAPA, D., MULTANI, M.S., MALGHAN, S.G., Densification of nanostructured
alumina assisted by rapid nucleation of α-alumina. Materials Letters, 1998, vol. 36,
pp. 235-239.
[54] LEGROS, C., CARRY, C., BOWEN, P., HOFMANN, H., Sintering of a Transition Alumina:
Effects of Phase Transformation, Powder Characteristics and Thermal Cycle. Journal of the
[55] BOWEN, P, CARRY, C., From powders to sintered pieces: forming, transformations and
sintering of nanostructured ceramic oxides. Powder Technology, 2002, vol. 128, pp. 248-255.
[56] YEN, F.S., LO, H.S., WEN, H.L., YANG, R.J., θ- TO α-phase transformation subsystem
induced by α-Al 2 O 3 –seeding in boehmite-derived nano-sized alumina powders. Journal of
Crystal Growth, 2003, vol. 249, pp. 293-293.
[57] MESSING, G.L., KUMAGAI, M., SHELLEMAN, R.A., MC ARDLE, J.L., Seeded transformation
microstructural control in ceramics. In: HENCH, L.L., ULRICH, D.D. (Eds.), Science of Ceramic
Chemical Processing, John Wiley & Sons, New York, 1985, pp. 259-271.
[58] KUMAGAI, M., MESSING, G.L., Enhanced densification of Boehmite Sol-Gel by Alpha Alumina
Seeded. Journal of the American Ceramic Society, 1984, vol. 67, pp. 230-231.
[59] KUMAGAI, M., MESSING, G.L., Control transformation and sintering of Bohemite Sol-Gel by
alpha-alumina seeded. Journal of the American Ceramic Society, 1985, vol. 68, pp.500-505.
[60] KANO, J., SAEKI, S., TANJO, M., SAITO, F., Reduction in α-Transformation Temperature of
Gibbsite and Boehmite by Powder Mixing with Seed. Journal of Chemical Engineering of
Japan, 2001, vol. 34, n°5, pp. 595-599.
[61] XIE, Z.P., LU, J.W., GAO, L.C., LI, W.C., XU, L.H., WANG, X.D., Influence of different seeds
on transformation of aluminium hydroxides and morphology of alumina grains by hot-pressing.
Materials and Design, 2003, vol. 24, pp. 209-214.
[62] PACHULA, M.L.,YING, J.Y., Mechanical synthesis of nanocrystalline α-Al 2 O 3 seed for
enhanced transformation kinetics. Nanostructured Materials, 1997, vol. 9, pp. 161-164.
[63] LO, J.R., TSENG, T.Y., Phase development and activation energy of the Y 2 O 3 -Al 2 O 3 system
by a modified sol-gel process. Materials Chemistry and Physics, 1998, vol. 58, pp. 56-62.
[64] KAO, H.C., WEI, W.C., Kinetics and Microstructural Evolution of Heterogeneous
Transformation of θ-Alumina to α-Alumina. Journal of the American Ceramic society, 2000, vol.
83, n°2, pp. 362-368.
[65] MALEK, J., The applicability of Johnson-Mehl-Avrami model in the thermal analysis of the
crystallization kinetics of glasses. Themochimica Acta, 1995, vol. 267, pp. 61-73.
185
References
[66] YAN, Z.J., DANG, S.E., WANG, X.H., LIAN, P.X., Applicability of Johnson-Mehl-Avrami model
to crystallization of Zr 60 Al 15 Ni 25 bulk amorphous alloy. Transaction of Nonferreous Metallurgy
Society of China, 2008, vol. 18, pp. 138-144.
[67] BROWN, M.E., Introduction to Thermal Analysis: Techniques and Applications. Chapman and
Hall ed., London, 1988.
[68] KISSINGER, H.E., Variation of Peak Temperature with Heating Rate in Differential Thermal
Analysis. Journal of Research of the National Bureau of Standards, 1956, vol. 57, n°4, 217221.
[69] KISSINGER, H.E., Reaction Kinetics in Differential Thermal Analysis. Analytical Chemistry,
1957, vol. 29, N°11, pp. 1702-1706.
[70] AUGIS, J.A., BENNETT, J.E., Calculation of the Avrami parameters for heterogeneous solid
state reactions using a modification of the Kissinger Method. Journal of the Thermal Analysis,
1978, vol. 13, pp. 283-292.
[71] KERR, M.C.., REED, J.S., Comparative grinding kinetics and grinding energy during ball milling
and attrition milling. American Ceramic Society Bulletin, 1992, vol. 71, n°12, pp. 1809-1815.
[72] NORDAHL, C.S., MESSING, G.L., Thermal analysis of phase transformation kinetics in
α-Al 2 O 3 seeded Bohemite and γ-Al 2 O 3 . Themochimica Acta, 1998, vol. 318, pp. 187-199.
[73] KAO, H.C., WEI, W.C., Kinetics and Microstructural Evolution of Heterogeneous
Transformation of θ-Alumina to α-Alumina. Journal of the American Ceramic Society, 2000, vol.
83, n°2, pp. 362-368.
[74] YEN, F.S., CHANG, J.L., YU, P.C., Growth Thermodynamics and Kinetics of Nano-sized
Alpha-Al2O3 Powders. Key Engineering Materials, 2003, vol. 249, pp. 179-182.
[75] YU, P.C., YEN, F.S., Preparation of Nano-sized Alpha-Alumina Crystallites Using Theta- Theta
Alumina Composite Powders. Key Engineering Materials, 2003, vol. 249, pp. 175-178.
[76] YANG, R.J., YEN, F.S., LIN, S.M., CHEN, C.C., Microstructure-controlled effects on
temperature reduction of crystallite formation. Journal of Crystal Growth, 2007, vol. 299, pp.
429-435.
[77] EL SAYED, M., TOFT SORENSEN, O., Initial sintering stage kinetics of CeO 2 studied by
Stepwise Isothermal Dilatometry, Report Risø R-518, February 1985.
[78] LIPPENS, B.C., STEGGERDA, J.J., In: LINSEN, B.G. (Editor), Physical and Chemical Aspects
of adsorbents and Catalysts, Academic Press, New York, 1970, p. 171.
[79] MORTERRA, C., MAGNACCA, G., A case study: surface chemistry and surface structure of
catalytic aluminas, as studied by vibrational spectroscopy of adsorbed species. Catalysis
Today, 1997, vol. 27, pp. 497-532.
[80] WILSON, S.J., MCDONNELL, J.D.C., A kinetic study of the system γ-AlOOH/ Al 2 O 3 . Journal of
the Solid State Chemistry, 1980, vol. 34, pp. 315-322.
[81] BUSCA, G., LORENZELLI, V., RAMIS, G., WILLEY, R.J., Surface sites on the spinel-type and
corundum-type metal oxide powders. Langmuir, 1993, vol. 9, pp. 1492-1499.
186
References
[82] TARTE, P., Infra-red spectra of inorganic aluminates and characteristic vibrational frequencies
of AlO 4 Tetrahedra and AlO 6 Octahedra. Spectrochimica Acta Part A, 1967, vol. 23,
pp. 2127-2143.
[83] KNOZINGER, H., KRIETENBRINK, H., MULLER, H.D., SCHULTZ, M., Cooperative effects in
surface chemical and heterogeneously catalysed reaction on metal oxides, in Proceedings of
the 6th International Congress on Catalysis, London, 1976, BOND, G.C., WELLS, P.B.,
TOMPKINS, F.S. (Eds.), The Royal Society of Chemistry, London, 1977, vol. 1, pp. 183-194.
[84] LAVALLEY, J.C., BENAISSA, M., IR spectroscopic evidence for surface vibrational modes
formed upon dehydroxylation of alumina. Journal of the Chemical Society, Chemical
Communications, 1984, pp. 908-909.
[85] MARCHESE, L., BORDIGA, S., COLUCCIA, S., MARTRA, G., ZECCHINA, A., Structure of the
surface sites of δ-Al 2 O 3 as determined by high-resolution transmission electron microscopy,
computer modelling and IR spectroscopy of adsorbed CO. Journal of the Chemical Society:
Faraday Transactions I, 1993, vol. 89, pp. 3483-3489.
[86] MORTERRA, C., BOLIS, V., MAGNACCA, G., IR spectroscopic and microcalorimetric
characterization of Lewis acid sites on (transition-phase) Al 2 O 3 using adsorbed CO. Langmuir,
1994, vol. 10, pp. 1812-1824.
[87] FIERRO, J.L.G., Spectroscopic Characterisation of Heterogeneous Catalysts, Part A: Method
of surface Analysis, Part B: Chemisorption of Probe-Molecules, Elsevier, Amsterdam, 1990.
[88] TSYGANENKO, A.A., FILIMONOV, V.N., Infrared spectra of surface hydroxyls groups and
crystalline structures of oxides. Journal of Molecular Structure, 1973, vol. 19, pp. 579-589.
[89] TSYGANENKO, A.A., Structure and properties of hydroxylated surfaces of oxides. Reaction
Kinetics and Catalysis Letters, 1993, vol. 50, pp. 33-38.
[90] KNOZINGER, H., RATNASAMY, C., Catalytic Aluminas: Surface Models and Characterization
of Surface Sites. Catalysis Reviews: Science and Engineering, 1978, vol. 17, pp. 31-70.
[91] PERI, J.B., IR and gravimetric study of the surface hydration of γ-alumina. A model for the
surface of γ-alumina. IR study of adsorption of Ammonia on dry γ-alumina. Journal of Physical
Chemistry, 1965, vol. 69, pp. 220-221.
[92] KAWAKASHI, H., YOSHIDA, S., Theoretical approach to the basicity of alkaline-earth metal
oxide catalysis. Journal of the Chemical Society. Faraday Transactions 2, 1984, vol. 80, pp.
921-932.
[93] DAVYDOV, A., Molecular Spectroscopy of Oxide Catalyst Surfaces. John Wiley & Sons Ltd.
The Atrium, England, 2003.
[94] BUSCA, G., LORENZELLI, V., ESCRIBANO, V. S., GUIDETTI, R., FT-IR study of the surface
properties of the spinels NiAl 2 O 4 in relation to those of the transition aluminas. Journal of
Catalysis, 1991, vol. 131, pp. 167-177.
[95] DELLA GATTA, G., FUBINI, B., GHIOTTI, G., MORTERRA, C., Chemisorption of CO on
various transition aluminas. Journal of Catalysis, 1976, vol. 43, pp. 90-98.
[96] PERI, J.B., IR study of the adsorption of CO 2 , HCl and other molecules on acid sites on silicaalumina and γ-alumina. Journal of Physical Chemistry, vol. 70, pp. 3168-3179.
187
References
[97] PERI, J.B., IR study of OH and NH 2 groups on the surface of a dry silica aerogel. Journal of
Physical Chemistry, vol. 70, pp. 2937-2945.
[98] ZECCHINA, A., COLUCCIA, S., MORTERRA, C., Infrared spectra of molecules adsorbed on
oxide surfaces. Applied Spectroscopy Rewiew, 1985, vol. 21, pp. 259-310.
[99] LITTLE, L.H., Infrared Spectra of Adsorbed Molecules. Academic Press, London, 1966.
[100] DIXIT, L., RAO, P.T., Spectroscopy in the measurements of acidic-basic properties solids.
Applied Spectroscopy Review, 1996, vol. 31, pp. 369-472.
[101] PARRY, E.P, An infrared study of pyridine adsorbed on acidic solids. Characterization of
surface acidity. Journal of Catalysis, 1963, vol. 2, pp. 371-379.
[102] TAAGELPERA, M., HENDERSON, W.G., BROWNLEE, R.T.C., BEAUCHAMP, J.L., HOLTZ,
D., TAFT, R.W. Gas phase basicities and pyridine subtituent effects. Journal of the American
Chemical Society, 1972, vol. 94, pp. 1369-1374.
[103] JENSEN, W.B., the Lewis Acid-Base Concepts. An Overview. Wiley, New York, 1980.
[104] STAIR, P.C., The concept of Lewis acids and bases applied to the surfaces. Journal of the
Chemical Society, 1972, vol. 104, pp. 4044-4052.
[105] ZAKI, M.I., KNOZINGER, H., CO- a low temperature IR probe for the characterization of
hydroxyls group properties on metal oxide surfaces. Material Chemistry and Physics, 1987, vol.
17, pp. 201-215.
[106] ZECCHINA, A., SCARANO, D., BORDIGA, S., RICCHIARDI, G., SPOTO, G., GEOBALDO, F.,
IR studies of CO and NO adsorbed on well characterized oxide single microcrystals. Catalysis
Today, 1996, vol. 27, pp. 403-435.
[107] MORTERRA, C., MAGNACCA, G., DEL FAVERO, N., IR study of CO adsorption at 77 K on
α-Al 2 O 3 . Langmuir, 1993, vol. 9, pp. 642-645.
Chapter II
[1] FERKEL, H., HELLMIG, R.J., Effect on the nanopowders deagglomeration on the densities of
nanocrystalline ceramic green bodies and their sintering behaviour. Nanostructural materials,
1999, vol. 11, n°5, pp. 617-622.
[2] Nanophase Technologies Corporation [on-line]. Available at: http://www.nanophase.com/
(consulted 24.11.2009)
[3] AZAR, M., PALMERO, P., LOMBARDI, M., GARNIER, V., MONTANARO, L., FANTOZZI, G.,
CHEVALIER, J., Effect of initial particle packing on the sintering of nanostructured transition
alumina. Journal of the European Ceramic Society, 2008, vol. 28, pp. 1121-1128.
[4] LEGROS, C., CARRY, C., BOWEN, P., HOFMANN, H., Sintering of a Transition Alumina:
Effects of phase Transformation, Powder Characteristics and Thermal Cycle. Journal of the
European Ceramic Society, 1999, vol. 19, pp. 1967-1978.
188
References
[5] CHEN, T., SOMASUNDARAN, P., Preparation of novel core-shell nanocomposite particle by
controlled polymer bridging. Journal of the American Ceramic Society, 1998, vol. 81, pp. 140144.
[6] HAUSSONNE, J.M., CARRY, C., BOWEN, P., BARTON,, J., Traité des matériaux: céramiques
et verres – Principes et techniques d’elaboration, vol. 16, Lausanne: Polytechniques et
Universitaires Romandes, 2005, pp. 815; Steitz B., Broyage et atomisation,
http://mxsg3.epfl.ch/ltp/cours/LTP_TPCeramiques TP1 Broyage.pdf
[7] DOCKTOR, T., WEYMOUTH, P., BARSON, T., Proper Techniques for the Ball-Milling of
Ceramic Glazes. American Ceramic Society Bulletin, 1994, Vol. 73, n°1, pp. 54-58.
[8] STANLEY, D.A., SADLER, L.Y., BROOKS, D.R., SCHWARTZ, M.A., Attrition Milling of
Ceramic Oxides. Ceramic Bulletin, 1974, Vol. 53, n°11, pp. 813-829.
[9] SADLER, L.Y., STANLEY, D.A., BROOKS, D.R., Attrition Mill Operating Characteristics.
Powder Technology, 1975, Vol. 12, pp. 19-28.
[10] RABE, T., LINKE, D., Attrition Milling of Silicon Nitride Powder under conditions for Minimal
Impurity Pickup. Ceramics International, 1992, Vol. 12, pp. 161-166.
α-Al 2 O 3 seeded Bohemite and γ-Al 2 O 3 . Themochimica Acta, 1998, vol. 318, pp. 187-199.
[12] CARRIER, X., MARCEAU, E., LAMBERT, J.F., CHE, M., Transformation of γ-alumina in
aqueous suspensions 1. Alumina chemical weathering studied as a function of pH. Journal of
Colloid and Interface Science, 2007, vol. 308, pp. 429-437.
[13] ROELOFS, F., VOGELBERGER, W., Dissolution kinetics of nanodispersed γ-alumina in
aqueous solution at different pH: Unusual kinetic size effect and formation of a new phase.
Journal of Colloid and Interface Science, 2006, vol. 303, pp. 450-459.
[14] POURBAIX, M., ZOUBOUV, N., VAN MUYLDER, J., Atlas d’équilibres électrochimiques.
Centre Belge d’étude de la Corrosion, Gauthier-Villars & Editeurs-Imprimeur-Libraire, Paris,
1963.
[15] HOFMEISTER, H., KODDERITZSCH, P., DUTTA, J., Structure of nanometersized silicon
particles prepared by various gas phase processes. Journal of Non-Crystalline Solids, 1998,
vol. 232, pp. 182-187.
[16] DJURICIC, B., PICKERING, P., GLAUDE, P., MCGARRY, D., TAMBUYSER, P., Thermal
stability of transition phases in zirconia-doped alumina. Journal of Materials Science, 1997, vol.
32, pp. 589-601.
[17] YANG, R-J, YEN, F.S., LIN, S.M., CHEN, C.C., Microstructure-controlled effects on
temperature reduction of α-Al 2 O 3 crystallite formation. Journal of Crystal Growth, 2007,
pp.
429-435.
Al 2 O 3 seeded boehmite and γ-Al 2 O 3 . Thermochimica Acta, 1998, vol. 318, pp. 187-199.
α-
[19] BOWEN, P., CARRY, C., LUXEMBOURG, D., HOFMANN, H., Colloidal processing and
sintering of nanosized transition aluminas. Powder Technology, 2005, vol. 157, pp. 100-107.
189
References
[20] WANG, H., LIU, X., CHEN, F., HENG, G., SORENSEN, T.O., Kinetic and Mechanism of a
Sintering Process for Macroporous Alumina Ceramics by Extrusion. Journal of American
Ceramic Society, 1998, vol. 81, n°3, pp. 781-784.
[21] HOLKOVA, Z., PACH, L., KOVAR, V., SVETIK, S., Kinetic Study of Al2O3 Sintering by
Dilatometry Ceramics. Ceramics-Silikaty, 2003, vol. 47, n°1, pp. 13-19.
[22] WANG, Y., SURYANARAYANA, C., AN, L., Phase transformation in nanometer-sized
γ-Alumina by Mechanical Milling. Journal of the American Ceramic Society, 2005, vol. 88, n°3,
pp. 780-783.
[23] PACHULA, M.L., YING, J.Y., Mechanical synthesis of nanocrystalline α-Al 2 O 3 seed for
enhanced transformation kinetics. Nanostructured Materials, 1997, vol. 9, pp. 161-164.
Today, 1997, vol. 27, pp. 497-532.
[25] IORDAN, A., ZAKI, M.I., KAPPENSTEIN, C., Interfacial chemistry in the preparation of catalytic
potassium-modified aluminas. Journal of the Chemical Society.: Faraday Transactions, 1993,
vol. 89, n°14, pp. 2527-2536.
[26] WANG, S.L., JOHNSON, C.T., Assignment of the structural OH stretching bands of gibbsite.
American Mineralogist, 2000, vol. 85, pp. 739.
[27] ZECCHINA, A., OTERO AREAN, C., Diatomic molecular probes for mid-IR studies of zeolites.
Chemical Society Reviews, 1996, vol. 25, pp. 187-197.
[28] BONELLI, B., ONIDA, B., CHEN, J.D., GALARNEAU, A., DI RENZO, F., FAJULA, F.,
GARRONE, E., Spectroscopic chracterization of the strength and stability of the acidic sites of
Al-rich microporous micelle-templated silicates. Microporous and Mesoporous Materials, 2004,
vol. 67, pp. 95-106.
[29] LERCHER, J.A., GRUNDLING, C., EDER-MIRTH, G., Infrared studies of the surface acidity of
oxides and zeolites using adsorbed molecules. Catalysis Today, 1996, vol. 27, pp. 353-376.
[30] CAIRON, O., CHEVREAU, T., LAVALLEY, J.C., Brønsted acidity of extraframework debris
in steamed Y zeolites from FTIR study of CO adsorption. Journal of the Chemical
Society Faraday Transactions, 1998, vol. 94, pp. 3039.
[31] HADJIIVANOV, K.I., VAYSSILOV, G.N., Characterization of oxide surfaces and zeolites by
carbon monoxide as an IR probe molecules. Advanced catalysis, 2002, vol. 47, pp. 307-511.
[32] AREAN, C.O., MANOILOVA, O.V., TSYGANENKO, A.A., PALOMINO, G.T., MENTRUIT, M.P.,
GEOBALDO, F., GARRONE, E., Thermodynamics of hydrogen bonding between CO and the
supercage Brønsted acid sites of H-Y zeolite – Studies from variable temperature IR
spectrometry. European Journal of Inorganic Chemistry, 2001, vol. 7, pp. 1739-1743.
[33] MORTERRA, C., BOLIS, V., MAGNACCA, G., IR spectroscopic and microcalorimetric
characterization of Lewis acid sites on (transition-phase) Al 2 O 3 using adsorbed CO. Langmuir,
1994, vol. 10, pp. 1812-1824.
[34] BOLIS, V., CERRATI, G., MAGNACCA, G., MORTERRA, C., Surface acidity of metal oxides.
Combined microcalirometric and IR-spectroscopic studies of variously dehydrated systems.
Thermochimica Acta, 1998, vol. 321, pp. 63.
190
References
[35] DAVYDOV, A., Molecular Spectroscopy of Oxide Catalyst Surfaces. John Wiley & Sons Ltd.
The Atrium, England, 2003.
[36] PERI, J.B., Infrared Study of Adsorption of Carbon Dioxide, Hydrogen Chloride, and other
molecules on Acid Sites on Dry Silica-Alumina and γ-Alumina. Journal of physical Chemistry,
1966, vol. 70, pp. 3168-3179.
Today, 1997, vol. 27, pp. 497-532.
[38] DAVYDOV, A.A., SHEPIT’KO, M.L., BUDNEVA, A.A., IR-Spectroscopic Methods for Studying
the Basic Centers of the Surface of Oxides. Analysis of IR Spectra of Adsorbed ProbeMolecule CO 2 as a Method for Studying Basic center at the Surface of Oxides. Kinetics and
catalysis, 1994, vol. 35, n°2, pp. 272-278.
[39] LAVALLEY, J.C., Infrared spectrometric studies of the surface basicity of metal oxide and
zeolites using adsorbed probe molecules. Catalysis Today, 1996, vol. 27, pp. 377-401.
[40] ZECCHINA, A., MARCHESE, L., BORDIGA, S., PAZE, C., GIANOTTI, E., Vibrational
Spectroscopy of NH4+ Ions in Zeolitic Materials: An IR Study. Journal of the Physical Chemistry
B, 1997, vol. 101, pp. 10128-10135.
Chapter III
[1]
NIIHARA, K., New Design Concept of Structural “Ceramic Nanocomposites”. The Centennial
[2]
ZHAO, J., STEARNS, L., HARMER, M.P., CHAN, H.M., MILLER, G.A., Mechanical Behavior of
Alumina-Silicon Carbide “Nanocomposites”. Journal American Ceramic Society, 1993, vol. 76,
n°2, pp. 503-510.
[3]
KUNTZ, J.D., ZHAN, G.D., MUKHERJEE, A.K., Nanocrystalline-Matrix Ceramic Composites
for Improved Fracture Toughness. MRS Bulletin, 2004, vol. 1, pp. 22-27.
[4]
GLEITER, M., Nanocrystalline Materials. Advanced Structural and Functional Materials, 1991,
pp. 1-37.
[5]
BHADURI, S., BHADURI, S.B., Recent Developments in Ceramic Nanocomposites . Journal of
the Mineral, Metals and Materials Society, 2007, vol. 50, pp. 1543-1551.
[6]
MORRIS, D.G., Mechanical behavior of nanostructured materials. Transaction Technical
Publication, 1998.
[7]
LEE, K.M., LEE, D.J., AHN, H., XRD and TEM studied on tin oxide (II) nanoparticles prepared
by inert gas condensation. Material Letters, 2004, vol. 58, pp. 3122-3125.
[8]
AJAYAN, P.M., SCHADLER, L.S., BRAUN, P.V., Nanocomposite Science and Technology.
WILEY, Weinheim, 2003.
[9]
HUANG, K.L., YIN, L.G., LIU, S.Q., LI, C.J., Preparation and formation mechanism of Al 2 O 3
nanoparticles by reverse microemulsion. Transaction of Nonferreous Metallurgy Society
China, 2007, vol. 17, pp. 633-637.
191
References
[10] KIM, J., LEE, J.W., PARK, B.Y., CHOI, C.J., Characteristics of Fe/SiO 2 nanocomposite
powders by the chemical vapor condensation process. Journal of Alloys and Compounds,
2008, vol. 449, pp. 258-260.
[11] HASSANZADEH-TABRIZI, S.A., TAHERI-NASSAJ, E., Economical synthesis of Al 2 O 3
nanopowder using a precipitation method. Materials Letter, 2009, vol. 63, pp. 2274-2276.
[12] KOCH, C.C., Nanostructured Materials: Processing, Properties and Potential Applications.
William Andrew Publishing, New York, 2002.
[13] SCHWARZ, J.A., CONTESCU, C.L., PUTYERA, K., Dekker Encyclopedia of Nanoscience and
Nanotechnology. Taylor & Francis, New York, 2004.
[14] DING, J., MSUZUKI, T., MCCORNICK, P.G., Ultrafine alumina particles prepared by
mechanochemical/thermal processing. Journal of the American Society, 1996, vol. 79, n°11,
pp. 2956-2958.
[15] STERNITZKE, M., Review: Structural Ceramic Nanocomposites. Journal of the European
[16] SCHEHL, M., DIAZ, L.A., TORRECILLAS, R., Alumina nanocomposites from powder-alkoxide
mixtures. Acta Materialia, 2002, vol. 50, pp. 1125-1139.
[17] PALMERO, P., NAGLIERI, V., CHEVALIER, J., FANTOZZI, G., MONTANARO, L., Aluminabased nanocomposites obtained by doping with inorganic salt solutions: Applications to
immiscible and reactive systems. Journal of the European Ceramic Society, 2009, Vol. 29, pp.
59-66.
[18] CARTER, B.C., NORTON, M.G., Ceramic Materials: Science and Engineering. Springer,
Berlin, 2007.
[19] COBLE, R.L., Sintering Crystalline Solids: I. Intermediate and Final State Diffusion Models.
Journal of Applied Physics, 1961, Vol. 32, n° 5, pp. 787-792.
[20] GROZA, J.R., Materials Processing Handbook. CRC Press, London, 2007.
[21] RAHAMAN, M.N., Ceramic Processing and Sintering. Marcel Dekker, New York, 2003.
[22] SHEN, Z., JOHNSSON, M., ZHAO, Z., NYGREN, M., Spark Plasma Sintering of Alumina.
Journal of the American Ceramic Society, 2002, Vol. 85, n°8, pp. 1921-1927.
[23] ZHAN, G.D., KUNTZ, J.D., WAN, J., MUKHERJEE, A.K., Single-wall carbon nanotubes as
attractive toughening agents in alumina-based nanocomposites. Nature Materials, 2003, vol. 2,
pp. 38-42.
[24] GAO, L., SHEN, Z., MIYAMOTO, H., NYGREN, M., Superfast Densification of Oxide/Oxide
Ceramic Composites. Journal of the American Ceramic Society, 1999, vol. 82, n°4, pp. 10611063.
[25] BENGISU, M., Engineering Ceramics, Springer, Berlin, 2001.
[26] LEWIS, D., RAYNE, R.J., BENDER, B.A., KURIHARA, L.K., CHOW, G.M., FLITFLET, A.,
KINCAID, A., BRUCE, R. Conventional and High frequency Microwave Processing of
Nanophase Ceramic Materials, Nanostructured Materials, 1997, vol. 9, pp. 97-100.
192
References
[27] SUTTON, W.H., Microwave processing of ceramic materials. American Ceramic Society
Bulletin, 1989, vol. 68, n°2, pp. 376-389.
[28] BYKOV, Y., GUSEV, S., EREMEEY, A., HOLOPSTEV, V., MALYGIN, N., PIRARUNAS, S.,
SOROKIN, A., SHUROV, A., Sintering of Nanophase Oxide Ceramics by using Millimeter
Wave Radiation. Nanostructured Materials, 1995, vol. 6, pp. 855-858.
[29] FREIM, J., MCKITTRICK, J., KATZ, J., SICKAFUS, K., Microwave Sintering of Nanocrystalline
α-Al 2 O 3 . Nanostructured Materials, 1994, vol. 4, pp. 371-385.
[30] TIAN, Y.L., JOHNSON, D.L., BRODWIN, M.E., Microwave Sintering of Alumina-TiC
Composites. In: Ceramic Transaction, Vol. 1B, ceramic Powder Science II. Ed. By Messing,
G.L., FULLER, E.R., HAUSNER, H., American Ceramic Society, Westerville, OH, 1988.
[31] FRENCH, J.D., HARMER, M.P., CHAN, H.M. et al., Coarsening-resistant dual-phase
interpenetrating microstructures. Journal of the American Ceramic Society, 1990, vol. 73, n°8,
pp. 2508-2510.
[32] ZENER, C., Kindly quoted by C.S. Smith. Transaction of the Metallurgical Society of AIME,
1949, vol. 175, n° 15.
[33] LANGE, F.F., Powder Processing Science and Technology for Increased Reliability. Journal of
the American Ceramic Society, 1989, vol. 72, n°1, pp. 3-15.
[34] STEARNS, L., HARMER, M.P., Particle-inhibited grain growth in Al 2 O 3 -SiC: I Experimental
results. Journal of the American Ceramic Society, 1996, vol. 79, n°12, pp. 3013-3019.
[35] HALL, E.O.H., The deformation and ageing of mild steel: III Discussion of Results Proc. Phys.
Soc. B, B64, 1951, pp. 747-753.
[36] PETCH, N.J., The cleavage strength of polycrystals. J. Iron Steel Inst., 174, 1953, pp. 25-28.
[37] RICE, R.W., WU, C.C., BORCHANLT, F., Hardness-Grain Size Relation in Ceramics. Journal
of the American Ceramic Society, 1994, vol. 77, n°10, pp. 2539-2553.
[38] SHEN, Z., JOHNSSON, M., ZHAO, Z., NYGREN, M., Spark Plasma Sintering of Alumina.
Journal of the American Ceramic Society, 2002, vol. 85, n°8, pp. 1921-1927.
[39] AWAJI, H., CHOI, S.M., Review: Ceramic-based nanocomposites. Res. Devel. Mat. Sci. Eng.,
2002, vol. 1, pp. 585-597.
[40] OHJI, T., KUSUNOSE, T., NIIHARA, K., Threshold Stress in Creep of Alumina-Silicon Carbide
Nanocomposites. Journal of the American Ceramic Society, 1998, vol. 81, n°10, pp. 27132716.
[41] BORSA, C.E., JONES, N.M., BROOK, R.J., TODD, R.I., Influence of Processing on the
Microstructural development and flexure strength of Al2O3/SiC nanocomposites. Journal of the
European Ceramic Society, 1997, Vol. 17, pp. 865-872.
[42] DAVIDGE, R.W., BROOK, R.J., CAMBIER, F., Fabrication, properties, and modelling of
engineering ceramics reinforced with nanoparticles of silicon carbide. British Ceram. Trans,
1997, Vol. 96, pp. 121-127.
193
References
[43] CHOU, I.A., CHAN, H.M., HARMER, M.P., Effect of annealing environment on the crack
healing and mechanical behaviour of silicon carbide-reinforced alumina nanocomposites.
Journal of the American Ceramic Society, 1998, Vol. 81, n°5, pp. 1203-1208.
[44] GAO, L., WANG, H.Z., HONG, J.S., MIYAMOTO, H., MIYAMOTO, K., NISHIKAWA, Y.,
TORRE, S.D.D.L., Mechanical Properties and Microstructure of Nano-SiC-Al 2 O 3 Composites
Densified by Spark Plasma Sintering. Journal of the European Ceramic Society, 1999, vol. 19,
pp. 609-613.
[45] ZHAN, G.J., YANG, J.F., ANDO, M., OHJI, T. Reactive Hot Pressing of Alumina-Silicon
Carbide Nanocomposites. Journal of the American Ceramic Society, 2004, vol. 87, n°2, pp.
299-301.
[46] CHANTIKUL, P., BENNISON, S.J., LAWN, B.R., Role of Grain Size in the Strength and RCurve Properties of Alumina. Journal of the American Ceramic Society, 1990, vol. 73, n°8, pp.
2419-2427.
[47] ZHAO, J., STERNS, L.C., HARMER, M.P., CHAN, H.M., MILLER, G.A., COOK, R.E.,
Mechanical behavior of aluminas-silicon carbide nanocomposites. Journal of the American
[48] DERBY, B., Ceramic nanocomposites: mechanical properties. Current Opinion in solid State &
Materials Science, 1998, vol. 3, pp. 490-495.
[49] GREEN, D.J., Fracture toughness predictions for crack bowing in brittle particulate composites.
Journal American Ceramic Society, 1983, vol. 66, C4-C5.
[50] PEZZOTTI, G., NISHIDA, T., SAKAI, M., Physical limitations of the inherent toughness and
strength in ceramic-ceramic and ceramic-metal nanocomposites. Journal of the Japanese
[51] CLAUSSEN, N., Fracture Toughness of AI 2 O 3 with an Unstabilized ZrO 2 Dispersed Phase.
Journal of the American Ceramic Society, 1976, Vol. 59, N°1, pp. 49-51.
[52] BHADURI, S., BHADURI; S.B., Enhanced Low Temperature Toughness of Al 2 O 3 -ZrO 2
Nano/Nano Composites. NanoStructured Materials, 1997, vol. 8, n°6, pp. 755-763.
[53] KIM, B.N., HIRAGA, K., SAKKA, Y., JANG, B.K, Effect of cavitations on superplastic flow of
10% Zirconia-dispersed alumina. Scripta Materialia, 2001, Vol. 45, pp. 61-67.
[54] ZHAN, G.D., KUNTZ, J., GARAY, J., MUKHERJEE, A.K., A Novel Processing Route to
Develop a Dense Nanocrystalline Alumina Matrix (<100 nm) Nanocomposite Material. Journal
of the European Ceramic Society, 2002, vol. 86, n°1, pp. 200-202.
[55] SADANGI, R.K., SHUKLA, V., KEAR, B.H., Processing and properties of ZrO 2 (3Y 2 O 3 )-Al 2 O 3
Nanocomposites. International Journal of Refractory Metals & Hard Materials, 2005, vol. 23,
pp. 363-368.
[56] VASYLKIV, O., SAKKA, Y., SKIROKHOD, V. V., High-Toughness Tetragonal Zirconia/AluminaCeramics. Key Engineering Materials, 2006, vol. 317-318, pp. 615-618.
[57] De AZA, A.H., CHEVALIER, J., FANTOZZI, G., SCHEHL, M., TORECILLAS, R., Crack growth
resistance of alumina, Zirconia and Zirconia toughened alumina ceramics for joint prostheses.
Biomaterials, 2002, vol. 23, pp. 937-945.
194
References
[58] NIIHARA; K:; UNAL, N., NAKAHIRA; A:, Mechanical properties of (Y-TZP)-alumina-silicon
carbide nanocomposites and the phase stability of y-TZP particles in it. Ceram. Soc. Japan,
1991, pp. 404-417.
[59] NAWA, M., BAMBA, N., SEKINO, T., NIIHARA, K., The effect of TiO 2 Addition on
Strengthening and Toughening in intragranular Type of 12Ce-TZP/Al 2 O 3 Nanocomposites.
[60] TAN, H., YANG, W., Toughening mechanisms of nano-composite ceramics. Mechanics of
Materials, 1998, vol. 30, pp. 111-123.
[61] NORTON, F.H., Models for Creep of Fibrous Composite Materials. Materials Forum, 1929, vol.
11, pp. 133-139.
[62] FANTOZZI, G., CHEVALIER, J., OLAGNON, C., CHERMANT, J.L, Creep of ceramic matrix
composites. In: Warren R (eds) Comprehensive Composite Materials, 2000, Vol. 4
Carbon/Carbon. Cement and Ceramic Matrix Composites. Elsevier, Oxford, UK, Chapter 7.
[63] NABARRO, F.R.N., Deformation of Crystals by Motion of Single ions. Proceedings of the
Bristol Conference in Strength of Solids. The Physical Society, 1948, pp. 75-90.
[64] HERRING, C., Diffusional Viscosity of a Polycrystalline Solid. Journal of Applied Physics, 1950,
vol. 21., pp. 437-445.
[65] COBLE, R.L., A Model of Boundary Diffusion Controlled Creep in Polycrystalline Materials.
Journal of Applied Physics, 1963, vol. 34, pp. 1679-1682.
[66] CANNON, W.R., LANGDON, T.G., Review, Creep of Ceramics, Part 1 Mechanical
Characteristics. Journal of Materials Science, 1983, vol. 18, pp. 1-50.
[67] CARTER, C.B., NORTON, M.G., Ceramic Materials: Science and Engineering. Springer, New
York, 2007.
[68] OHJI, T., NAKAHIRA, A., HIRANO, T., NIIHARA, K., Tensile Creep Behavior of Alumina/Silicon
Carbide Nanocomposite. Journal of the American Ceramic Society, 1994, vol. 77, n°12, pp.
3259-3262.
[69] DESCAMPS, P., O’SULLIVAN, D., POORTEMAN, M., DESCAMPS, J.C., LERICHE, A.,
CAMBIER, F., Creep Behaviour of Al 2 O 3 -SiC Nanocomposites. Journal of the European
[70] THOMPSON, A.M., CHAN, H.M., HARMER, M.P., Tensile Creep of Alumina-Silicon Carbide
“Nanocomposites”. Journal of the American Ceramic Society, 1997, vol. 80, n°9, pp. 22212228.
[71] OHJI, T., HIRANO, T., NAKAHIRA, A., NIIHARA, K., Particle/Matrix Interface and Its Role in
Creep Inhibition in Alumina/Silicon Carbide. Journal of the American Ceramic Society, 1996,
vol. 79, n°1, pp. 33-45.
[72] ARZT, E., GRAHLE, P., High temperature creep behaviour of oxide dispersion strengthened
NiAl intermetallics. Acta materialia, 1998, vol. 46, n°8, pp. 2717-2727.
[73] ALPER, A.M., Phase Diagrams in Advanced Ceramics. Academic Press, London, 1995.
195
References
[74] NOGUCHI, T., MIZUNO, M., Phase diagrams for Ceramics 1975 Supplement. Levin and Mc
Murdie Eds., pp. 132, American Ceramic Society, Westerville OH, 1975.
[75] TOROPOV, N.A., BONDAR, I.A., GALAKHOV, F.Y. VINOGRADOVA, Phase diagrams for
Ceramics 1969 Supplement, Levin et al. Eds., Westerville OH, 1969.
[76] CASLAVSKY, R.L., VIECHNICKI, D.J., Melting behaviour and metastability of yttrium
aluminium (YAG) and Al 2 O 3 determined by optical differential thermal analysis. Journal
Material Science, 1980, vol. 15, pp. 1709-1719.
[77] LIU, Y., ZHANG, Z.F., KING, B., HALLORAN, J., LAINE, R.M., Synthesis of Yttrium Aluminium
Garnet from Yttrium and Aluminium Isobutyrate Precursors. Journal of the American Ceramic
Society, 1996, vol. 79, n°2, pp. 385-394.
[78] LIU, Y., ZHANG, Z.F., HALLORAN, J., LAINE, R.M., Yttrium Aluminium Garnet Fibers from
Metalorganic Precursors. Journal of the American Ceramic Society, 1998, vol. 81, n°3, pp.
1029-1032.
[79] GOWDA, G., Synthesis of yttrium aluminates by the sol-gel process. Journal of Material
Science Letters, 1986, vol. 5, n°10, pp. 1029-1032.
[80] KAMAT, R.V., PILLAI, K.T., VAIDYA, V.N., SOOD, D.D., Synthesis of yttrium aluminium garnet
by the gel entrapment technique using hexamine. Materials Chemistry and Physics, 1996, vol.
46, n°1, pp. 67-71.
[81] HAY, R.S., Phase transformation and microstructure evolution in sol-gel derived yttrium
aluminium garnet films. Journal of Materials Research, 1993, vol. 8, pp. 578-604.
[82] YAMAGUCHI, K., TAKEOKA, K., HIROTA, K., TAKANO, H., HAYASHIDA, A., Formation of
Alkoxy-Derived Yttrium Aluminium Oxides. Journal Materials Science, 1992, vol. 27, pp. 12611264.
[83] YAMAGUCHI, O., TAKEOKA, K., HAYASHIDA, A., Formation of alkoxyde-derived Y 3 Al 5 O 12 .
Journal of Materials Science Letters, 1991, vol. 10, pp. 101.
[84] INOUE, M., OTSU, H., KONINAMI, H., INUI, T., Synthesis of yttrium aluminium garnet by
glycothermal method. Journal of the American Ceramic Society, 1991, vol. 74, pp. 1452-1454.
[85] ROTH, R.S., Phase Equilibrium Diagrams: Phase Diagrams for Ceramists, Vol. XI. The
American Ceramic Society, Westville, OH, 1995, pp.107.
[86] KELLER, K., MAH, T., PARTHASARATHY, T., Processing and mechanical properties of
polycrystalline Y3Al5O12. Ceram. Eng. Sci. Proc., 1990, Vol. 1, pp. 1122-1133.
[87] MAH, T., PARTHASARATHY, T., MATSON, L., Processing and mechanical properties of
Al 2 O 3 /Al 3 Al 5 O 12 (YAG) eutectic composite. Ceram. Eng. Sci. Proc., 1990, Vol. 11, pp. 16171633.
[88] WAKU, Y., NAKAGAWA, N., WAKAMOTO, T., OHTSUBO, H., SHIMIZU, K.,
KOHTOKU, Y., The Creep and Thermal Stability Characteristics of Unidirectionally
Solidified Al 2 O 3 /YAG Eutectic Composite. J. Mater. Sci., 1998, Vol. 33, pp. 1217.
[89] VROLIJK, J.W., WILLEMS, J.W., METSELAAR, R., Coprecipitation of yttrium and aluminum
hydroxides for preparation of yttrium aluminum garnet. Journal of the European Ceramic
Society., 1990, Vol. 6, pp. 47–51.
196
References
[90] LI, W.Q., GAO, L., Processing, Microstructure and Mechanical Properties of 25 vol.% YAGAl 2 O 3 nanocomposites. Nanostructured Materials, 1999, Vol. 11, N°8, pp. 1073-1080.
[91] FRENCH, J.D., CHAN, H.M., HARMER, M.P., MILLER, G.A., High-Temperature Fracture
Toughness of Duplex Microstructures. Journal of the American Ceramic Society, 1996, Vol. 79,
n°1, pp. 58-64.
[92] DUONG, H., WOLFENSTINE, J., Creep Behavior of Fine Grained Two-Phase Al 2 O 3 -YAG
Materials, Mater. Sci. Eng., 1993, A 173, pp. 173-179.
[93] FRENCH, J.D., ZHAO, ,J., HARMER, M.P., CHAN, H.M., MILLER, G.A., Creep of duplex
microstructures. Journal of the American Ceramic Society, 1994, vol. 77, pp. 2857-2865.
[94] KOTTADA, R.S., CHOKSHI, A.H., The High-Temperature Tensile and Compressive
Deformation of Magnesia Doped Alumina. Acta Materalia, 2000, vol. 48, pp. 3905-3915.
[95] SATAPATHY, L.N., CHOKSHI, A.H., Microstructural Development and Creep Deformation in
an Alumina- 5 % Yttrium Aluminium Garnet Composite. Journal of the American Ceramic
Society, 2005, vol. 88, n°10, pp. 2848-2854.
[96] TORRECILLAS, R., SCHEHL, M, DIAZ, L.A., Creep behaviour of alumina-mullite-zirconia
nanocomposites obtained by a colloidal processing route. Journal of the European Ceramic
Society, 2007, vol. 27, pp. 143-150.
[97] PARTHASARATHY, T.A., MAH, T., KELLER, K., Creep Mechanisms of Polycrystalline Yttrium
Aluminium Garnet. Journal of the American Ceramic Society, 1992, vol. 75, pp. 1756-175.
Chapter IV
[1] NIIHARA, K., New Design Concept of Structural “Ceramic Nanocomposites”. The Centennial
[2] AWAJI, H., CHOI, S.M., Review: Ceramic-based nanocomposites. Res. Devel. Mat. Sci. Eng.,
2002, vol. 1, pp. 585-597.
[3] KELLER, K., MAH, T., PARTHASARATHY, T.: Processing and mechanical properties of
polycrystalline properties of polycrystalline Y 3 Al 5 O 12 . Ceramic Engineering and Science
Proceedings, 1990, 11, pp. 1122-1133.
[4] PALMERO, P., NAGLIERI, V., CHEVALIER, J., FANTOZZI, G., MONTANARO, L., Aluminabased nanocomposites obtained by doping with inorganic salt solutions: Applications to
immiscible and reactive systems. Journal of the European Ceramic Society, 2009, Vol. 29, pp.
59-66.
[5] FRENCH, J.D., ZHAO, J., HARMER, M.P., CHAN, H.M., MILLER, G.A., Creep duplex
microstructures. Journal of the American Ceramic Society, 1994, vol. 77, pp. 2857-2865.
[6] YOSHIDA, H., IKUHARA, Y., SAKUMA, T., High-temperature creep resistance in rare-earthdoped fine-grained Al2O3. Journal of Materials research, 1998, vol. 13, pp. 2597-2601.
[7] TORRECILLAS, R., SCHEHL, M., DIAZ, L.A., MENENDEZ, J.L., MOYA, J.S., Creep behaviour
of alumina/YAG nanocomposites obtained by a colloidal processing route. Journal of the
European Ceramic Society, 2007, vol. 27, pp. 143-150.
197
References
[8] Taimei Chemical Co., LTD. [on-line]. Available at: http://taimei-chem.co.jp/ (consulted
24.11.2009)
[9] Baikowski Chimie
24.11.2009)
[on-line].
Available
at:
http://www.baikowskichimie.com/
(consulted
[10] GAO, L., SHEN, Z., MIYAMOTO, H., NYGREN, M., Superfast Densification of Oxide-Oxide
Ceramic Composites. Journal of the American Ceramic Society, 1999, vol. 82, n°4, pp. 10611061.
[11] HALL, E.O.H., The deformation and ageing of mild steel: III Discussion of Results Proc. Phys.
Soc. B, B64, 1951, pp. 747-753.
[12] PETCH, N.J., The cleavage strength of polycrystals. J. Iron Steel Inst., 174, 1953, pp. 25-28.
[13] WANG, H., GAO, L., Preparation and microstructure of polycrystalline Al 2 O 3 -YAG composites.
Ceramics International 27 (2001), 721-723.
[14] ANSTIS, G.R., CHANTIKUL, P., LAWN, B.R., MARSHAL, D.B., A critical evaluation of
indentation techniques for measuring fracture toughness. I. Direct crack measurements.
[15] ASTM Designation: C 1259-01. Standard Test Method for Dynamic Young’s Modulus, Sher
Modulus, and Poisson’s Ration for Advanced Ceramics by Impulse Excitation of Vibration.
ASTM Committee on Advance Ceramics, 2001.
[16] RICE, R.W., Mechanical Properties of ceramics and Composites: grain and Particle effects.
Marcel Dekker, Inc., New York. Basel, 2000.
[17] SWAIN, M.V., Structure and properties of Ceramics, Vol. 11, Materials Science and
Technology, Cahn R.W. Haasen P., Kramer, E.J, VCH Weinheim, 1994.
[18] SCHEHL, M., DIAZ, L.A., TORRECILLAS, R., Alumina nanocomposites from powder–alkoxide
mixtures. Acta Materialia 50 (2002) 1125–1139.
[19] HOLLEMBERG, G.W., TERWINLLINGER, G.R., GORDON, R.S., “Calculation of stresses and
strains in four point bending creep tests”, Journal of the American Ceramic Society, 1971, vol.
54, pp. 196-199.
[20] SATAPATHY, L.N., CHOKSHI, A.H., Microstructural development and creep deformation in an
alumina-5% yttrium aluminium garnet composite. Journal of the American Ceramic Society,
2005, vol. 88, pp. 2848-2854.
[21] LANGDON, T.G., Dependence of Creep Rate on porosity. Journal of the American Ceramic
Society, 1962, vol. 45, pp. 630-631.
[22] NABARRO, F.R.N., Deformation of Crystals by Motion of Single ions. Proceedings of the
Bristol Conference in Strength of Solids. The Physical Society, 1948, pp. 75-90.
[23] HERRING, C., Diffusional Viscosity of a Polycrystalline Solid. Journal of Applied Physics, 1950,
vol. 21., pp. 437-445.
[24] COBLE, R.L., A Model of Boundary Diffusion Controlled Creep in Polycrystalline Materials.
Journal of Applied Physics, 1963, vol. 34, pp. 1679-1682.
198
References
[25] CHEVALIER, J., OLAGNON, C., FANTOZZI, G., Creep behaviour of alumina, zirconia and
zirconia toughened alumina. Journal of the European Ceramic Society, 1997, vol. 17,
pp. 859-864.
199
Appendix A
200
Appendix A1
1.
Laser Granulometry
Figure A1 – Granulometer Fritsch model Analysette 22 Compact.
For the development of this PhD thesis, a laser particle size analyser Fritsch model Analysette 22
Compact (Figure A1) was employed to evaluate the particle size distribution. The granulometer is
suitable for wet measurements of granulometric powder with particle size ranging from 0.31 to
300.74 μm range. Laser granulometry is a technique based on the light diffraction.
It is based on the Fraunhofer theory, using the following hypotheses:



Spherical particles are considered to be non porous and non opaque.
The diameter of the particles is higher compared with laser wavelength.
The particles are in constant motion.
The particles diffract light efficiently, regardless of their size. When a laser beam sheds light on a
particle, diffraction patterns can be observed. The intensity of the diffracted radiation and the
deviation differ according to the size of the particles. The larger the particle, more light will deviate
and the weaker its deviation angle, in relation to the propagation will be.
The first step is to dilute the sample. The apparatus measure the background in order to record
the diffraction phenomena caused by the solvent. Subsequently, the sample solution is injected
into the measuring cell, each particle that passes through the radiation beam deviates the light
which is analysed by detectors.
2.
X-ray diffraction (XRD)
The most widespread use of X-ray powder diffraction is for the identification of crystalline
compounds by their diffraction pattern. This experimental method is employed to determine the
structure of the crystalline materials and to measure the crystalline size by the Scherrer method
(explained in chapter I).
XRD diffraction analysis were performed by using a diffractometer Philips PW 1710, with a Cu K α
radiation (λ=1.5405600 Å). In Figure A2 it is illustrated the instrument in detail.
201
Appendix A1
Figure A2 – Schematic representation of the diffractometer Philips PW 1710.
In X-ray powder diffractometry, X-rays are generated within a sealed tube under vacuum. A current
is applied which heats a filament within the tube, the higher the current, the greater the number of
electrons emitted from the filament.
A high voltage, typically 15-60 kilovolts, is applied within the tube. This high voltage accelerates the
electrons, which then hit a target, commonly made of copper. When these electrons hit the target,
X-rays are produced. The wavelength of these x-rays, is characteristic of that target. These X-rays
are collimated and directed onto the sample, which has been ground to a fine powder (typically to
produce particle sizes of less than 10 microns).
A detector detects the X-ray signal; the signal is then processed either by a microprocessor or
electronically, converting the signal to a count rate, changing the angle between the X-ray source,
the sample and the detector at a controlled rate by means of a goniometer. When an X-ray beam
hits a sample and is diffracted, we can measure the distances between the planes of the atoms by
applying Bragg's Law. Bragg's Law is n.  2.d .sin   , where the integer n is the order of the
diffracted beam, 1 is the wavelength of the incident X-ray beam, d is the distance between
adjacent planes of atoms (the d-spacings), and θ is the angle of incidence of the X-ray beam.
202
Appendix A1
Figure A3 – Illustration of the goniometer and angles involved.
Since we know λ and we can measure θ, we can calculate the d-spacings. The geometry of an
XRD unit is designed to accommodate this measurement (Figure A3).The characteristic set of
d-spacing generated in a typical X-ray scan provides a unique pattern of the material. When
properly interpreted, by comparison with standard reference patterns from ICDD (International
Centre for Diffraction Data) files, this "fingerprint" allows for identification of the material.
3.
Thermal analysis: DTA-TG
Differential thermal analysis involves heating or cooling a test sample and an inert reference under
identical conditions, while recording any temperature difference between the sample and
reference. This differential temperature is then plotted against time, or against temperature.
Changes in the sample which lead to the absorption or evolution of heat can be detected
compared to the inert reference.
Differential temperatures can also arise between two inert samples when their response to the
applied heat treatment is not identical (Figure A4). DTA can therefore be used to study thermal
properties and phase changes which do not lead to a change in enthalpy. The baseline of the DTA
curve should then exhibit discontinuities at the transition temperatures and the slope of the curve at
any point will depend on the microstructural constitution at that temperature.
Figure A4- Schematic illustration of DTA-TG cell.
The area under a DTA peak can be imputed to the enthalpy change and it is not affected by the
heat capacity of the sample. DTA may be defined formally as a technique for recording the
203
Appendix A1
difference in temperature between a substance and a reference material against either time or
temperature as the two specimens are subjected to identical temperature regimes in an
environment heated or cooled at a controlled rate.
Simultaneously, thermogravimetric (TG) is also carried out, which follows the mass loss of the
sample during heating as a result of decomposition, desorption and/or dehydration phenomena. In
our particular case, the tests were carried out by using a system of simultaneous DTA-TG Netzsch
STA 409C which also provides the possibility of employing the machine as DSC-TG. Pictures of
the device are shown below in Figure A5.
Figure A5- Thermal analysis device (DTA-TG)
4.
Dilatometry
A dilatometer Netzsch 402E was employed with the aim to study the sintering behaviour and
improving the sintering conditions in order to achieve the highest density. The temperature limit of
the machine is 1550°C. Ceramic powders were pressed into bars, introduced into the sample
holder and submitted to a controlled thermal cycle.
During the test, the length variation is measured as a function of the time. In this particular case,
the displacement is measured by a transducer as it is shown in Figure A6. This data permits to
calculate the coefficient of thermal expansion by the following equation.

L f  Li
Lo .T

L
(1)
Lo .T
204
Appendix A1
where ΔT is the variation of temperature, L o and L f are the initial and final length of the sample.
Figure A6- Netzsch 402E dilatometer.
5.
Scanning Electron Microscopy/Environmental Scanning Electron Microscopy
The SEM is an instrument that produces a largely magnified image by using electrons instead of
light to form an image. A beam of electrons is produced at the top of the microscope by an
electron gun. The electron beam follows a vertical path through the microscope, which is held
within a vacuum. The beam travels through electromagnetic fields and lenses, which focus the
beam down toward the sample. Once the beam hits the sample, electrons and X-rays are ejected
from the sample.
Detectors collect these X-rays, backscattered electrons, and secondary electrons and convert
them into a signal that is sent to a screen similar to a television screen. This produces the final
image. In our case, a microscope Hitachi S2300 has been used. Samples are coated by gold
sputtering in order to increase the electrical conductivity. The second technique employed to
observe the surface of the samples, was the environmental scanning electron microscope (ESEM).
An ESEM microscope used was a FEI XL30 ESEM FEG. This technique permits the image of wet
systems with no prior specimen preparation. Additionally, sample environment can be dynamically
205
Appendix A1
altered, hydration and dehydration processes can be followed inside the sample chamber. A
schematic representation of the microscope is shown in Figure A7.
Fig A7- ESEM microscopy schematic diagram.
6.
High Resolution Transmission Electron Microscopy (HR-TEM)
Transmission Electron Microscopy (TEM) is a technique where an electron beam interacts and
passes through a specimen. The electrons are emitted by a source and are focused and magnified
by a system of magnetic lenses. The geometry of TEM is shown in Figure A8. The electron beam
is confined by the two condenser lenses which also control the brightness of the beam, passes the
condenser aperture and hits the sample surface. The electrons that are elastically scattered
consist of transmitted beams, which pass through the objective lens.
In this PhD thesis, HRTEM (High Resolution Transmission Electron Microscopy) pictures were
collected on a JEOL 3010-UHR instrument, operated at 300kV and equipped with a 2kx2k pixel
Ultrascan 100 camera.
206
Appendix A1
Figure A8- TEM microscopy schematic diagram.
The objective lens forms the image display and the following apertures, the objective and the
selected area aperture are used to choose the elastically scattered electrons that will form the
image of the microscope. Finally, the beam goes to the magnifying system that consisted of three
lenses, the first and second intermediate lenses control the magnification of the image and the
projector lens.
The operation of TEM requires an ultra high vacuum and a high voltage. The first step is to find the
electron beam, so the lights of the room must be turned off. Through a sequence of buttons and
adjustments of focus and brightness of the beam, we can adjust the settings of the microscope so
that by shifting the sample holder we find the thin area of the sample. Then tilting of the sample
begins by rotating the holder.
Different types of images are obtained in TEM, using the apertures properly and the different types
of electrons. As a result, diffraction patterns are shown because of the scattered electrons. If the
unscattered beam is selected, we obtain the Bright Field Image. Dark Field Images are attained if
diffracted beams are selected by the objective aperture. Also in TEM, analysis is done with EDX
(Energy Dispersive X-ray), EELS (Electron Energy Loss Spectrum), EFTEM (Energy Filtered
Transmission Electron Microscopy),etc.
In transmission microscopy, we can actually see the specimen’s structure and its atomic columns,
thus compositional and crystallographic information is achieved. However, it is a very expensive
technique, expertise is needed and the sample preparation phase is too difficult so that very thin
samples are achieved.
The first step is to decide whether the sample is useful to be observed and in which view, plan or
cross-section. Due to the strong interaction between electrons and matter, the specimens have to
be rather thin, less than 100nm. This is achieved with several methods, depending on the material.
In general, mechanical thinning is used to thin and polish the sample. Then it is glued with epoxy
glue on a really small and round holder. Whereas TEM data come from the edges of a hole in the
centre of the specimen, in sample preparation, the hole is created by the method of ion thinning.
207
Appendix A1
Ion thinning is a method where a specimen is usually irradiated with beams of Ar ions, and after a
period of time, a hole is created. To minimize the damage created during focus ion beam milling,
the embedded sample can first be coated with a metal deposition layer. Consequently, sample
preparation is a precise and a severe procedure, which may affect the results of the microscopic
analysis and study.
TEM provides accurate measurements and studies in different types of materials, since
observations are in atomic scale in HRTEM. This is due to technology that reduces the errors and
corrects more and more the interferences in formed images.
In order to improve the results of TEM, ultra high vacuum with no vibrations is needed and this fact
results in the production of different types of pumps such as mechanical pumps, oil diffusion
pumps, ion getter pumps and cooled stage. Higher voltage up to 3MV and small probe size were
developed, and methods to assure monochromaticity and coherency of the electrons. This is a way
to avoid chromatic aberration and «spherical aberration», the most usual errors in electron
microscopy.
Nowadays, high resolution transmission electron microscopes offer resolutions up to 0.1 nm at
300kV and probe diameters up to 0.34nm. Thus, future trends include the use of ultrahigh vacuum
TEM instruments for surface studies and computerized data acquisition for quantitative image
analysis.
7.
Fourier Transformed Infrared Device (FT-IR)
For FT-IR (Fourier Transform Infra-Red) measurements, powder samples were pressed into selfsupporting wafers. Spectra were collected at a resolution of 2 cm-1, on a Bruker FTIR Equinox 55
spectrophotometer equipped with a MCT detector. The schematic representation of the apparatus
is shown in Figure A9.
Figure A9- Bruker FTIR Equinox 55 spectrophotometer.
Fourier Transform Infrared (FT-IR) spectrometry was developed, in order to overcome the
limitations encountered with dispersive instruments. The main difficulty was the slow scanning
process. A method for measuring all of the infrared frequencies simultaneously, rather than
individually, was needed.
208
Appendix A1
A solution was developed and it exploited a very simple optical device called an interferometer.
The interferometer produces a unique type of signal which has all of the infrared frequencies
“encoded” into it. The signal can be measured very quickly, usually in the order of one second or
so. Thus, the time element per sample is reduced to a matter of a few seconds rather than several
minutes. The schematic diagram is shown in Figure A10.
Figure A10- The normal instrumental process.
Most interferometers employ a beamsplitter which takes the incoming infrared beam and divides it
into two optical beams. One beam reflects off of a flat mirror which is fixed in place. The other
beam reflects off from a flat mirror which is on a mechanism that allows this mirror to move a very
short distance (typically a few millimeters) away from the beamsplitter. The two beams reflect off
from their respective mirrors and are recombined when they meet back at the beamsplitter. Since
the path that one beam travels is a fixed length and the other is constantly changing as its mirror
moves, the signal which exits the interferometer is the result of these two beams “interfering” with
each other. The resulting signal is called an interferogram which has the unique property that every
data point (a function of the moving mirror position) which makes up the signal that has information
about every infrared frequency which comes from the source.
This means that as the interferogram is measured, all frequencies are being measured
simultaneously. Thus, the use of the interferometer results in extremely fast measurements. Since
the analyst requires a frequency spectrum (a plot of the intensity at each individual frequency) in
order to make an identification, the measured interferogram signal can not be interpreted directly.
A means of “decoding” the individual frequencies is required. This can be accomplished via a wellknown mathematical technique called the Fourier transformation. This transformation is performed
by the computer which then presents the user with the desired spectral information for analysis.
8.
Specific Surface Area
BET (Brunauer, Emmett, Teller) specific surface areas (SSA) were measured by means of N 2
adsorption/desorption isotherms at 77 K performed by using a Quantachrome Autosorb 1C
instrument (Figure A11).
The gas adsorption method permits to measure the amount of gas adsorbed on the surface of a
powder sample as a function of the pressure of the adsorbate gas, and it is used to determine the
specific surface area of a powder sample.
The concept theory is an extension of the Langmuir theory, which is a theory for monolayer
molecular adsorption, to multilayer adsorption with the following hypotheses:

Gas molecules physically adsorb on a solid in layers infinitely.
209
Appendix A1

There is no interaction between each adsorption layer.

Langmuir theory can be applied to each layer.
Taking into account these conditions, the resulting BET equation is shown below.
1
C 1 P
1


P
 VmC Po VmC
Va  o  1
P 
(2)
where P is the partial vapor pressure of adsorbate gas in equilibrium, P o is the saturated pressure
of the adsorbate gas, V a is the volume of the gas adsorbed at equilibrium, V m is the volume of the
gas adsorbed in a monolayer and C is a dimensionless constant related to the enthalpy of
adsorption and condensation of the adsorbate gas.
Figure A11 - Quantachrome Autosorb 1C Instrument.
The specific surface area, SSA, is determined from V m , the volume of gas adsorbed in a
monolayer on the sample.
S
Vm .N .a
m
(3)
where S is the specific surface area, N is the Avogadro constant, a is the effective cross-sectional
area of one adsorbate molecule and m is the mass of the test powder.
210
Appendix A1
Figure A11 – Schematic Diagram of the device.
Test powder is placed in a sample contained with a known volume (Figure A11) and the volume of
gas adsorbed is determined from the change in the pressure associated with the adsorption of gas
on the surface of the sample powder.
Initially, a pre-treatment is performed in order to remove gases or vapours, that have been
physically adsorbed onto the sample surface, by outgassing the sample under reduced pressure.
After pre-treatment is completed, the sample container should be precisely weighed with the
sample and the mass of the tared contained measured previously should be subtracted, to obtain
the true mass of the powder.
A fixed quantity of the adsorbate gas is introduced intro the sample container surrounded by liquid
nitrogen. The pressure decreases until gas/solid adsorption reaches to a new equilibrium. The
volume of gas adsorbed is calculated from the difference between the volume of the un-adsorbed
gas and the volume remaining in the void volume.
For a multiple point BET analysis, it is possible to calculate the specific surface area by repeating
the measurements under equilibrium pressure of the adsorbate gas in the range of 0.05 to 0.30.
9.
Young's Modulus Resonant Frequency Meter: Grindosonic
The Young’s modulus was determined the Grindosonic method, which allows measurement of the
resonant frequency of a ceramic sample at room temperature from which it can be obtained the
Young's modulus of the sample as described in ASTM C885. The machine is shown in Figure
A12.
The method consists on inducing a vibration on a parallelepiped cylinder along its longest
dimension, a detector transforms these mechanical vibrations into an electrical signal that can be
registered. Then, assuming a value of Poisson’s constant it can be obtained Young's modulus of
the material.
The Young's modulus is an intrinsic property of materials of extreme importance in the design of
materials and facilities and that affects other material properties such as resistance to thermal
shock.
211
Appendix A1
Figure A12 – Grindosonic machine.
10.
Hardness Tests
Figure A13 - Vickers Testwell FV-700
The indentation tests were performed using the Vickers Testwell FV-700 device (Figure A13),
which has the possibility of varying the charge from 0.3 to 30 kg. The charge was fixed in 98.1N
(10 kg). Five indentations were performed in each samples.
The toughness was determined by indentation tests by using Anstis’ formula, as it is shown below.
1/ 2
K IC
E  P 
 A.   .  2 / 3 
H  c 
c : length of the crack from the center of the impression
A : geometric constant equal to 0.016.
P : change expressed in Newton - in our case equal to 98.1N.
H : Hardness value obtained directly from the machine.
E : Young’s Modulus. In our case it was determined by the volume fraction law shown below.
E  f Al2O3 .E Al2O3  fYAG .EYAG
212
Appendix A1
d
2c
Figure A14 - Optical microscope’s observation of an impression.
11.
Creep Tests
Creep resistance tests were carried out in a 4-point bending fixture in a lever arm machine
(Figure A15 (A)). Samples were fixed in an alumina holder with distances between the different
points of 28 and 14 mm (Figure A15 (B)).
Bar-shaped samples were parallelepipeds of 32 x 4 x 3 mm3. In order to avoid surface flaws the
edges were chamfered at about 45°. Additionally, samples were polished with diamond paste up to
3 μm. Inside the sample container the samples were centred with a bar-shaped block with the aim
of assuring the parallelism of the faces.
The maximum load allowed for the test is 500 N. Whereas, the maximum temperature of the
machine is 1400°C. Thermocouples inside the chamber have a precision of +/- 3 degrees.
Deformation was recorded by a LVDT (Linear variable differential transformer) sensor with a
resolution of 1 μm.
The heating rate for all the test was 300°C/min in order to homogenise the temperature of the
chamber. During tests two variables were simultaneously recorded: the displacement and the
temperature as a function of the time.
The machine is presented in the Figure A15.
213
Appendix A1
Figure A15- Creep machine (A) and sample holder (B).
214
FOLIO ADMINISTRATIF
THESE SOUTENUE DEVANT L'INSTITUT NATIONAL DES SCIENCES APPLIQUEES DE LYON
NOM : LOMELLO
DATE de SOUTENANCE : 20 mai 2010
(avec précision du nom de jeune fille, le cas échéant)
Prénoms : FERNANDO
TITRE :
Optimization of nanostructured
oxide-based powders
by
Surface modification
NATURE : Doctorat
Numéro d'ordre : 2010-ISAL
Ecole doctorale : École Doctorale Matériaux de Lyon
Spécialité : Génie des Matériaux
Cote B.I.U. - Lyon : T 50/210/19
/
et
bis
CLASSE :
RESUME:
Ce travail a été consacré d’une part à l’étude de l'effet de la dispersion dans une alumine de transition commerciale et d’autre
part à l’étude des propriétés mécaniques à basse et haute température des nanocomposites Al 2 O 3 -5 vol. % YAG.
Pour étudier l’effet de la dispersion dans l’alumine de transition, différentes techniques de caractérisation appartenant à la
physico-chimie des surfaces et à la science des matériaux ont été utilisées comme l’analyse thermogravimétrique et l’analyse
thermique différentielle (ATD-TG), la diffraction des rayons X (DRX), l’adsorption d’azote (B.E.T.), la microscopie
électronique en transmission (MET) et l’infrarouge à transformée de Fourier (IR-TF). En particulier, les alumines de transition
présentent des phases métastables qui subissent des transformations pendant le frittage et provoquent la formation d’une
structure vermiculaire avec de larges porosités. La densité finale et la microstructure ont été améliorées grâce à une dispersion
efficace de la poudre initialement agglomérée qui permet un réarrangement des particules et facilite la transformation vers la
phase alpha. L’étude de l’influence de la dispersion sur la cinétique de transformation (Méthode de Kissinger) et la cinétique
de frittage (Méthode SID) a été développée.
Dans la deuxième partie de la thèse, le travail a été centré sur la production des nanocomposites Al 2 O 3 -5 vol. % YAG à partir
de deux alumines commerciales frittées naturellement et par des méthodes non-conventionnelles comme le pressage à chaud
(HP) et le spark plasma sintering (SPS). La caractérisation mécanique à température ambiante (dureté, ténacité, module
d’élasticité) a été corrélée à une étude microstructurale (ESEM). Des valeurs intéressantes de dureté et de ténacité ont été
mesurées dans les échantillons frittés par SPS et HP, environ 20 GPa et 7 MPa.m1/2, respectivement.
Pour la caractérisation à haute température, les essais de fluage ont été conduits sous air en flexion 4 points à 1200°C sous une
contrainte de 100 MPa. Les résultats montrent que les propriétés mécaniques à haute température dépendent fortement de la
répartition de la deuxième phase dans la matrice d’alumine. Dans tous les cas, les résultats obtenus sont intéressants.
MOTS-CLES :
Alumine, Dispersion, Transformation de phase, Frittage, Microstructure, YAG, Nanocomposites, Fluage
Laboratoire (s) de recherche : Dipartimento di Scienza dei Materiali e Ingegneria Chimica (DISMIC)
Matériaux: Ingénierie et Science (MATEIS)
Directeur de thèse: FANTOZZI Gilbert, PALMERO Paola
Président de jury:
Composition du jury : BONELLI Barbara, FANTOZZI Gilbert, LERICHE Anne, PALMERO Paola, REVERON Helen

Optimization of nanostructured oxide

Transcription

Similar documents

Franz Schubert Impromptus, Moments Musicaux, Klavierstücke for

Berthold Farmers Elevator, LLC Pre

reactive sintering

Ceramic Ventilated Cladding

Pimp my memory - Temple University

[email protected] [email protected]

ibostar plus. - Daetwyler USA

Studs Terkel, with Tony Parker

ce1"f[,Caf~ w~re. ra.Y8I~ 1336.

Study the evolution of British ceramics, from rudimentary Stone Age